Neuroglobin, nitric oxide, and oxygen: Functional pathways and conformational changes

Abstract

Neuroglobin (Ngb) is a globin expressed in the nervous system of humans and other organisms that is involved in the protection of the brain from ischemic damage. Despite considerable interest, however, the in vivo function of Ngb is still a conundrum. In this paper we report a number of kinetic experiments with O₂ and NO that we have interpreted on the basis of the 3D structure of Ngb, now available for human and murine metNgb and murine NgbCO. The reaction of reduced deoxyNgb with O₂ and NO is slow (t_1/2 ≈ 2 s) and ligand concentration-independent, because exogenous ligand binding can only occur upon dissociation of the distal His-64, which is coordinated to the ferrous heme iron. By contrast, NgbO₂ reacts very rapidly with NO, yielding metNgb and by means of a heme-bound peroxynitrite intermediate. Steady-state amperometric experiments show that Ngb is devoid of O₂ reductase and NO reductase activities. To achieve this result, we have set up a protocol for efficient reduction of metNgb using a mixture of FMN and NADH under bright illumination. The results are discussed with reference to a global scheme inspired by the 3D structures of metNgb and NgbCO. Based on the ligand-linked conformational changes discovered by crystallography, the pathways of the reactions with O₂ and NO provide a framework that may account for the involvement of Ngb in controlling the activation of a protective signaling mechanism.

Neuroglobin (Ngb) is a heme protein belonging to the extended globin family, which was discovered by Burmester et al. (1) in the brain of humans and mice. Interestingly, Ngb is predominantly expressed in specific areas of the brain, notably the frontal lobe, the subthalamic nucleus, and the thalamus (1, 2), although at a fairly low concentration (approximately micromolar) compared with, for example, myoglobin (Mb) in the heart (≈0.2 mM) (3, 4). The unexpected discovery of Ngb raised considerable curiosity from the standpoints of evolutionary biology and pathophysiology alike. The interest in the structure and function of Ngb intensified considerably after the recent findings by Greenberg and collaborators (5, 6) that (i) its expression in neurons is up-regulated under conditions of hypoxia and (ii) the extent of ischemic damage after experimental stroke in the rat is reduced when the amount of Ngb expressed in the brain is enhanced and vice versa.

Ngb was found to bind O₂ reversibly at the ferrous heme iron, with an affinity [P₅₀ ≈ 2 torr (1 torr = 133 Pa)] (1) comparable with that of Mb (P₅₀ ≈ 1 torr) (7). This observation suggested that its function in the brain would mimic that of Mb in red muscles, i.e., storage, transport, and facilitated intracellular diffusion of O₂. This hypothesis, however, is hardly tenable given the low average concentration of Ngb (8). Like other heme proteins, ferrous Ngb also binds CO and NO (9–13).

The 3D structures of metNgb from humans (14) and mice (15) show the typical three-over-three α-helical globin fold. Over and above a number of interesting features that emerged from examination of the high resolution 3D structures, two findings stand out: namely (i) the ferric heme iron is hexacoordinated, with the distal His(E7)64 and the proximal His(F8)96 directly bound to the metal ion, and (ii) the protein contains a very large (≈290 A³) internal cavity that is branched and connects the solvent with the proximal and distal heme sides. Although the latter feature was totally unexpected (and rather intriguing), the former confirmed extensive spectroscopic data (1, 9, 10, 13). Endogenous hexacoordination implies that binding of O₂ and other external ligands to the ferrous and ferric heme iron can only occur upon rupture of the 6th coordination bond with His-64, as shown by extensive kinetic data obtained largely by laser photolysis of the adducts of reduced Ngb with O₂, CO, and NO (8, 9, 11–13) and also by stopped-flow experiments with CO (8).

The physiological role of Ngb is still somewhat mysterious. As outlined above, Ngb is unlikely to be involved primarily in O₂ storage and facilitated diffusion because of its low average concentration. Other possible functions related to signal transduction and O₂ sensing, enzymatic activity similar/analogous to NADH reductases, or NO detoxification have been postulated (8, 16). In this paper, we have addressed some of these possibilities. We present transient and steady-state kinetic experiments that are interpreted in the light of the recently solved 3D structures of Ngb (14, 15, 17).

Experimental Procedures

Murine recombinant Ngb was purified as reported in ref. 11 and 13. Most experiments were carried out with the wild-type protein, which contained Cys-55 and Cys-120. However, some critical experiments were repeated with the double mutant used for crystallography (15), which contained Ser at these two positions, yielding identical results.

Kinetic experiments were carried out in 50 mM phosphate buffer, pH 7.0/20 μM EDTA. Anaerobic conditions were obtained by extensive N₂-equilibration of the buffer. Contaminant O₂ was further scavenged by addition of 2 mM glucose, 8 units/ml glucose oxidase, and 260 units/ml catalase. NO stock solutions were prepared by equilibrating degassed water with 1 atm (1 atm = 101.3 kPa) NO gas at room temperature.

Stopped-flow experiments were carried out at 5°C or 20°C with an instrument equipped with a diode array and a light path of 1 cm (model no. DX.17MV, Applied Photophysics, Surrey, U.K.); absorption spectra were collected with an acquisition time of 2.56 ms. Data were analyzed with matlab (MathWorks, Natick, MA).

Amperometric detection of O₂ and NO in solution was achieved at room temperature with selective Clark electrodes (ISO-NO Mark II, World Precision Instruments, Stevenage, U.K.). The volume of the measurement chamber was 1 ml. In the amperometric experiments, FMN was photoexcited in the presence of excess NADH by illumination of the reaction chamber with a 150-W tungsten lamp.

Results and Discussion

Ligand Binding by Reduced Ngb. The crystallographic structures of metNgb from humans (14) and mice (15) show that the distal His-64 is directly coordinated to the ferric heme iron. Mutagenesis of His-64 to Leu (8, 13) confirmed that the imidazole side chain is indeed providing coordination to the heme iron. Although the structure of reduced Ngb is not available, spectroscopy proved beyond a doubt that this species is also hexacoordinate (1, 9–11). Therefore, binding of exogenous ligands demands dissociation of the His-64–Fe²⁺ coordination bond according to the scheme shown in Fig. 1, where H₉₆ denotes the proximal His(F8). Accordingly, binding of O₂, NO, or CO should be rate-limited by formation of the pentacoordinate intermediate, which is hardly populated in reduced Ngb [the equilibrium value for murine Ngb being <0.1% (8)].

Fig. 1.

Reactions with O₂ and NO. (A) Overall scheme of some reactions of Ngb. The species on the left (reduced) and those on the right (met or ferric) are hexacoordinated by His-96 and His-64 (highlighted in blue); the other three species (highlighted in red) are supposed to have the 3D structure determined for NgbCO (17), lacking the internal coordination bond with His-64. The reduced pentacoordinate form is a transient species not significantly populated at equilibrium. (B) Overall time course of the reaction of reduced Ngb with O₂ (red) and NO (black) at pH 7.0 and 20°C. The reaction is slightly heterogeneous, the overall rate coefficient being k′ ≈0.4 s^-1; it may be appreciated that the rate is NO concentration-independent. (C) View of the active site of murine Ngb in the met (blue) and CO-bound (red) state. Upon CO ligation, the coordination bond with His-64 (above the heme plane) is broken. The imidazole ring moves only slightly, whereas the heme tilts and slides toward a preexisting space (17). This shift is associated with (i) a substantial change of the position of the proximal His-96 (seen at the bottom) and (ii) the disappearance of the proximal branch of the large cavity (contour highlighted in blue) and its extension on the distal side (indicated with orange contour toward the left).

As shown in Fig. 1B, combination of O₂ and NO to ferrous Ngb is indeed a slow reaction (t_1/2 ≈ 2 s for murine Ngb), in agreement with stopped-flow data on CO binding reported by Dewilde et al. (8). Moreover, the rates are independent of the nature and the concentration of the ligand. The overall time course is slightly heterogeneous, possibly because of the presence of two conformers of the heme, as shown by NMR (18) and x-ray crystallography (15) of murine metNgb.

CO binding to murine Ngb is associated with extensive conformational changes (Fig. 1C), involving not only breakage of the Fe²⁺–His-64 bond but also a large repositioning of the heme and a concomitant reorganization of the large internal cavity (17). Available kinetic data (8, 11) afford additional considerations in the light of the new structural information. In particular, it can be inferred that the rate coefficient of these ligand-linked structural changes must be ≈0.1 s^-1, as estimated from the reformation of the Fe²⁺–His-64 coordination bond in kinetic competition with ligand rebinding after laser photolysis. Conversely, the rate of ≈0.1 s^-1 limiting the binding of external ligands must be largely contributed by the barrier for breaking the sixth coordination bond. Incidentally, this rate is considerably faster than expected a priori, possibly because of the tendency of the heme to slide into a more stable configuration upon CO binding (17). In essence, relocation of the heme seems to counterbalance the loss of the distal coordination bond stability.

CO binding also induces a closure of the gate connecting the large cavity to the bulk (17), thereby segregating momentarily small molecules eventually trapped. Because the tunnel is largely coated by hydrophobic side chains (14, 15) and given the higher solubility of diatomic gases in nonpolar media (NO solubility being at least 3-fold increased as compared with H₂O), the internal volume may be enriched in NO and O₂ relative to the solvent. This putative increase in local concentration, although unnecessary to account for the high association rate coefficients to pentacoordinated reduced Ngb (k_on ≥ 10⁸ M^-1·s^-1) (8, 9, 12, 13), is intriguing. Therefore, we carried out experiments related to possible functions of Ngb for NO scavenging and detoxification.

The Reaction of NgbO₂ with NO. It is known that HbO₂ and MbO₂ react rapidly with NO, yielding nitrate and metHb or metMb (7, 19). Only fairly recently, however, it was emphasized that the role of MbO₂ as an efficient scavenger of NO in the heart and skeletal muscle, may protect cytochrome c oxidase from being inhibited by NO (20). This scavenging role was shown by elegant in vivo experiments carried out with Mb knockout mice (21). Because NO detoxification has been proposed as a possible function of Ngb, we have investigated this reaction by stopped-flow experiments.

Because NgbO₂ in vitro was found to autoxidize more rapidly than MbO₂ (1, 8), we used a double mixing protocol to ensure absence of metNgb. Anaerobically reduced Ngb (prepared as detailed in the legend to Fig. 2) was first mixed with air-equilibrated buffer to obtain the oxygenated adduct (NgbO₂), and, after a preset delay time (Δt = 1–100 s), it was mixed with NO at final concentrations from 10 to 250 μM. The results of these experiments are reported in Fig. 2. On the basis of information available for Mb and Hb (19, 22), a minimum scheme is expected (Scheme 1) in which the intermediate is a peroxynitrite adduct of metNgb (≡Ngb⁺), which isomerizes and dissociates, yielding nitrate and metNgb. For HbO₂ and MbO₂, the second step is the fastest under most conditions, and the overall reaction follows bimolecular behavior (19). For MbO₂, the peroxynitrite intermediate was observed only at very high pH (22), and it was reported that its maximum absorption in the Soret is blue-shifted (λ = 410 nm) compared with MbO₂ (λ = 417 nm).

Fig. 2.

The reaction of ferrous oxygenated Ngb with NO. deoxyNgb was obtained by incubating degassed metNgb for a few minutes with 400 μM NADH plus 1 μM FMN under bright white light illumination. Anaerobic conditions were ensured by addition of 2 mM glucose, 8 units/ml glucose oxidase, and 260 units/ml catalase. In the stopped-flow apparatus operating in the sequential mixing mode, deoxyNgb was first mixed with air-equilibrated buffer (50 mM phosphate buffer/20 μM EDTA, pH = 7) to obtain NgbO₂ after a suitable delay time (30 s or 100 s at 20°C or 5°C, respectively); thereafter, this solution was mixed with an anaerobic solution containing NO. NgbO₂ reacted with NO after the second mixing, yielding in the dead-time an intermediate (spectroscopically different from NgbO₂) that decays to metNgb. (A) Time course of decay of the intermediate to metNgb at 5°C and 20°C, as obtained from the absorption changes at 399–419 nm after normalization. Fitted rate constants: k ≈ 300 s^-1 at 20°C and [NO] = 10 μM(▪) or 250 μM(□); k ≈ 85–95 s^-1 at 5°C and [NO] = 10 μM (○), 40 μM (•), or 200 μM (▵). Experiments were carried out at [Ngb] ≈ 4.5 μM or [Ngb] ≈ 16 μM (final concentration). (Inset) Difference absorption spectra collected at 20°C over the first 25 ms, with reference to the spectrum of metNgb. [NO] = 250 μM (final concentration). The arrows indicate the progress of the reaction. (B and C) Absolute absorption spectra in the Soret (B) and in the visible region (C). The thick line represents the intermediate formed in the dead-time of the stopped flow (2.5 ms) calculated at t = 0. The thin line represents the endpoint species of the decay (see A), corresponding to metNgb. The dotted line represents NgbO₂. The peak wavelengths, λ_max, for the three species are indicated in B.

Scheme 1.

The spectra in Fig. 2 B and C show that the first observable species after mixing with NO is different from NgbO₂, suggesting that the intermediate is completely populated in the dead-time of the instrument. Moreover, this initial species has a Soret maximum at 407–408 nm, which in blue-shifted with respect to NgbO₂, in agreement with the expectation that this is indeed a Fe(III)-peroxynitrite intermediate. This species decays to metNgb monoexponentially, with a first-order rate coefficient of k₂ ≈ 90 s^-1 at 5°C and ≈300 s^-1 at 20°C; this decay is independent of NO concentration (Fig. 2 A), consistent with Scheme I. A minimum estimate of the second-order rate coefficient in Scheme I yields k₁ > 7 × 10⁷ M^-1·s^-1 at 5°C, a value not too different from that reported at 20°C for MbO₂ (≈4 × 10⁷ M^-1·s^-1). As a control, we mixed metNgb with NO under identical conditions and observed no absorption changes within seconds (data not shown).

In conclusion, the reaction of NgbO₂ with NO is very rapid and proceeds by means of a peroxynitrite intermediate transiently bound to metNgb. We propose that the observed decay process (Fig. 2A) corresponds to dissociation of the peroxynitrite intermediate, yielding as the final product. The peculiar structure of liganded Ngb (17) involving sliding of the heme may or may not be relevant to the relatively extended lifetime of the peroxynitrite intermediate. It is obvious that the significance of this reaction in scavenging NO in the neuron (by analogy to MbO₂ in the heart) depends on the concentration of Ngb, which is fairly low (1) except in the neurons of the retina (23), where it is ≈100 μM. More to the point, this mechanism can only be of physiological value if the brain contains an efficient metNgb reductase system to restore reduced Ngb. This system has yet to be identified.

Enzymatic Functions of Ngb? To test possible redox functions of Ngb, we searched for an efficient electron donor system that quickly reduces metNgb in a steady-state assay without efficiently scavenging O₂ or NO; thus, dithionite was a priori excluded. Under anaerobic conditions, we observed that 10 mM ascorbate plus either 0.5 mM Ru(II)-hexamine or 1 mM tetramethyl-p-phenylene diamine as well as 10 mM ferrocyanide were ineffective electron donors. However, a mixture of 1 mM NADH and micromolar concentrations of FMN under illumination proved to be an efficient reducing system without significant interference with O₂ or NO.

Mixing anaerobically prepared metNgb with NADH alone in the diode array stopped-flow apparatus (white light illumination) is associated with a slow (t_1/2 ≈ 10 s) reduction of Ngb (Fig. 3A); the overall time course is heterogeneous and can be fitted to a biexponential function (k₁ = 0.9 s^-1 and k₂ = 0.1 s^-1, with relative amplitudes of 0.2 and 0.8, respectively). Addition of micromolar quantities of FMN under otherwise identical conditions leads to a considerable increase in the reduction rate of metNgb. The time course now corresponds to a homogeneous reduction process, and the apparent rate coefficient is a linear function of FMN concentration (Fig. 3B). At 10 μM FMN, reduction is complete in <1 s. The formation of a complex between metNgb and FMN was excluded because addition of up to 0.8 μM protein does not affect the fluorescence emission of 0.2 μM FMN in the 460–660 nm range (data not shown).

Fig. 3.

Ngb reduction by photoactivated NADH/FMN. (A) Reduction of metNgb by NADH alone. Shown are absorption spectra collected from 2.5 ms to 50 s after mixing metNgb (≈8 μM) with NADH (2 mM) in the diode array stopped-flow apparatus. (Inset) Best fit of the reaction time course as obtained by singular value decomposition analysis; k ≈ 0.1 s^-1 for the main kinetic component. (B) Dependence of the rate coefficient for Ngb reduction on the concentration of FMN in the presence of 1 mM NADH and under illumination by white light. Conditions were as described for Fig. 2 (t = 20°C).

The reduction of metNgb by NADH/FMN is a photoactivated process, as demonstrated by static spectrophotometry. Addition of NADH or FMN or a mixture of the two to a cuvette containing degassed metNgb does not lead to reduction of the protein, even after several minutes in the dark. However, if the cuvette is illuminated (by a 150-W tungsten lamp), complete reduction occurs within a few seconds with 500 μM NADH and 0.5 μM FMN. This finding proved useful to test the possible enzymatic functions of Ngb.

By using NADH/FMN, we tested for O₂ reductase activity of Ngb by amperometric experiments under white light illumination. As shown in Fig. 4A, the addition of NADH/FMN to air-equilibrated buffer is followed by a small but detectable increase in the rate of O₂ consumption. More importantly, the subsequent addition of up to 0.9 μM metNgb (final concentration) does not significantly enhance O₂ consumption. This experiment shows that Ngb reduced by photoactivated NADH/FMN, although binding O₂ with high affinity, is unable to catalyze its reduction.

Fig. 4.

Testing the O₂ reductase and NO reductase activity of Ngb. Measurements were carried out at room temperature either in the dark or under bright white light illumination of the reaction chamber to ensure an efficient, steady reduction of Ngb by the NADH/FMN mixture. (A) O₂ consumption measurement under illumination. After addition of NADH and FMN, the addition of several aliquots of metNgb does not significantly affect the O₂ consumption rate. Fast response of the electrode is demonstrated by addition of dithionite. In comparison, addition of 0.2 μM beef heart cytochrome c oxidase in the presence of excess reductant (ascorbate, tetramethyl-p-phenylene diamine, and cytochrome c) would cause O₂ exhaustion in <30 s under similar experimental conditions. (B) NO measurement. After addition of three aliquots of NO (final concentration ≈ 30 μM) to degassed buffer, NO consumption was monitored either in the dark or under illumination. The addition of degassed NADH and FMN accelerates NO consumption in both cases, with a larger effect under illumination; however, the subsequent addition of anaerobic metNgb has essentially no effect on the rate of NO consumption. In the dark, residual NO in solution is promptly scavenged by addition of dithionite. In comparison, addition of 2 μM bacterial NO reductase in the presence of excess ascorbate and tetramethyl-p-phenylene diamine would cause NO to be completely consumed in ≈1 s under similar experimental conditions.

We then proceeded to assess a possible NO reductase activity of Ngb. Under anaerobic conditions, prokaryotic NO reductases catalyze the reduction of NO according to Scheme 2. Bona fide NO reductase has been purified from several bacteria (24). It was later observed that the same reaction can be catalyzed by some bacterial cytochrome c oxidases (25). Moreover, a class of flavodiiron enzymes shown to catalyze the same reaction in Escherichia coli (26, 27) are also expressed in some eukaryotes, notably in pathogenic protists (28).

Scheme 2.

We used a NO-selective electrode to follow the rate of NO decomposition under anaerobic conditions, using the NADH/FMN system as electron donor in the dark or in the light. The results depicted in Fig. 4B show that NO decomposition is accelerated after addition of NADH and FMN, this effect being somewhat more pronounced under illumination. The important point, however, is that addition of metNgb (up to 2 μM) has no significant effect on the rate of NO consumption, both in the dark (when reduction is known to be very slow) and in the light (when it is quickly reduced). Likewise, Ngb proved unable to reductively degrade NO at lower NO concentrations (2–5 μM) (data not shown); although, in this case, a lower concentration of FMN (0.2 μM) had to be added to prevent a fast decomposition of NO within the time scale of the assay, i.e., several minutes.

These experiments exclude the possibility that Ngb is able to catalyze the reduction of O₂ or NO, at least under the conditions explored. Although it cannot be proven, it is unlikely that Ngb can acquire these enzymatic functions in vivo.

Remarks on Physiological Implications

The discovery that Ngb binds O₂ reversibly (P₅₀ ≈ 2 torr) (1) suggested that its role in the brain may be similar to that of Mb in red muscles, i.e., to store O₂ and facilitate intracellular O₂ diffusion from the periphery of the cell to the mitochondria. These functions of Mb, however, are only possible because of its very high concentration (≈0.2 mM) in the muscle (3, 4). A role of Mb in facilitating intracellular O₂ diffusion is only significant when its concentration counterbalances the difference in translational diffusion coefficient between O₂ and the protein. Given that the average concentration of Ngb in the brain is rather low (approximately micromolar), except in the retina (23), it is very unlikely that O₂ storage and transport is physiologically relevant (8).

To outline the relevant pathophysiological scenario for a discussion on the role of Ngb in the protection against neuronal death, it should be recalled that Sun et al. (6) demonstrated a neuroprotective effect of Ngb in vivo. Their results show that an increase in the expression level of Ngb is associated to a reduced severity in the histological and functional deficits after focal cerebral ischemia in rats, particularly evident in the so-called penumbral cortical tissue around the occluded vessel. This area is presumed to be fairly hypoxic ([O₂] << 10 μM) but experiencing an overproduction of NO because of induced expression of NO synthetases (29). This finding justifies our interest in exploring the reactions of Ngb with O₂ and NO, summarized in Fig. 1 A.

Examination of the scheme outlines the interplay between the control of reaction rates and the ligand-linked conformational changes observed in Ngb (17). The hexacoordinate-reduced Ngb²⁺ is in equilibrium with the pentacoordinate species that (although hardly populated, < 0.1%) is the only state that can bind O₂, NO, or CO. Because the bimolecular rate coefficients for O₂ and NO binding to pentacoordinate Ngb²⁺ are very high and almost identical (k_on ≈ 1 × 10⁸ to 3 × 10⁸ M^-1·s^-1 at 20°C) (8, 9, 12, 13), the branching ratio will be determined solely by the relative concentrations of the two gases in the tissue. The fairly slow rate of this molecular device is due to the rupture of the distal coordination bond with His-64, with a half-life of ≈2 s. The scheme in Fig. 1 also indicates (with color code) the large conformational change coupled to exogenous ligand binding (17), which is rather unusual for noncooperative O₂ carriers (as may be seen by comparison with Mbs) (30) and is therefore intriguing.

The fate of the two adducts (NgbO₂ and NgbNO) is quite different in an environment containing both O₂ and NO at comparable concentrations (as assumed to be the case in the penumbral area surrounding the core region affected by a stroke). The formation of NgbNO is almost a dead-end pathway because (i) thermal dissociation of NO is extremely slow [t_1/2 > 30 min according to Van Doorslaer et al., (12)] and (ii) reaction with O₂ leading to metNgb, although possible, is also very slow (at [O₂] < 10 μM, t_1/2 > 10 min) (31). Because we have shown that Ngb under anaerobic and reducing conditions does not function as a NO reductase (see Fig. 4), the adduct NgbNO is presumably “frozen” in the liganded conformation typical of NgbCO and thus cannot exert any signaling effect, such as that discovered by Wakasugi et al. (32) for metNgb (see below).

The alternative pathway through NgbO₂ is more dynamic not only because oxygenation is quickly reversible but because the oxygenated derivative reacts very rapidly with free NO, as shown by the results reported in Fig. 2. The half-life of this reaction (which is essentially irreversible) is milliseconds even at NO concentrations in the micromolar range, yielding as products and metNgb. With reference to a possible signaling function of Ngb under conditions of hypoxia and reperfusion, Wakasugi et al. (32) demonstrated that metNgb (but not NgbCO) interacts specifically with a component of the GDP/GTP signal transduction pathway. These authors observed that metNgb is endowed with guanine nucleotide dissociation inhibitor activity because it binds to the GDP-bound formofGα, inhibiting GDP/GTP exchange and thereby liberating Gβγ with beneficial effects for cell survival. On the contrary, NgbCO (and by analogy NgbO₂ and NgbNO) is inactive, a finding that is likely to be understood based on the large ligand-linked conformational change recently demonstrated by Vallone et al. (17). Within this perspective, the O₂-binding pathway with subsequent oxidation by NO (see Fig. 1A) would have the role of competing effectively with direct formation of NgbNO, which excludes Ngb from the protective signaling function outlined above. In addition, this pathway would dispose of NO by means of a rapid reaction with NgbO₂, which may in turn protect cellular respiration jeopardized by the inhibitory effect of NO on cytochrome c oxidase activity (33–35). It is clear, however, that, in order to be physiologically relevant, the cycle outlined in Fig. 1 demands the presence of an efficient metNgb reductase in the brain, which is yet to be discovered.