A new pathway of nitric oxide/cyclic GMP signaling involving S-nitrosoglutathione.

Nitric oxide (NO), a physiologically important activator of soluble guanylyl cyclase (sGC), is synthesized from L-arginine and O2 in a reaction catalyzed by NO synthases (NOS). Previous studies with purified NOS failed to detect formation of free NO, presumably due to a fast inactivation of NO by simultaneously produced superoxide (O-2). To characterize the products involved in NOS-induced sGC activation, we measured the formation of cyclic 3',5'-guanosine monophosphate (cGMP) by purified sGC incubated in the absence and presence of GSH (1 mM) with drugs releasing different NO-related species or with purified neuronal NOS. Basal sGC activity was 0.04 +/- 0.01 and 0.19 +/- 0.06 micromol of cGMP x mg-1 x min-1 without and with 1 mM GSH, respectively. The NO donor DEA/NO activated sGC in a GSH-independent manner. Peroxynitrite had no effect in the absence of GSH but significantly stimulated the enzyme in the presence of the thiol (3.45 +/- 0.60 micromol of cGMP x mg-1 x min-1). The NO/O-2 donor SIN-1 caused only a slight accumulation of cGMP in the absence of GSH but was almost as effective as DEA/NO in the presence of the thiol. The profile of sGC activation by Ca2+/calmodulin-activated NOS resembled that of SIN-1; at a maximally active concentration of 200 ng/0.1 ml, NOS increased sGC activity to 1.22 +/- 0.12 and 8.51 +/- 0.88 micromol of cGMP x mg-1 x min-1 in the absence and presence of GSH, respectively. The product of NOS and GSH was identified as the thionitrite GSNO, which activated sGC through Cu+-catalyzed release of free NO. In contrast to S-nitrosation by peroxynitrite, the novel NO/O-2-triggered pathway was very efficient (25-45% GSNO) and insensitive to CO2. Cu+-specific chelators inhibited bradykinin-induced cGMP release from rat isolated hearts but did not interfere with the direct activation of cardiac sGC, suggesting that thionitrites may occur as intermediates of NO/cGMP signaling in mammalian tissues.

(25-45% GSNO) and insensitive to CO 2 . Cu ؉ -specific chelators inhibited bradykinin-induced cGMP release from rat isolated hearts but did not interfere with the direct activation of cardiac sGC, suggesting that thionitrites may occur as intermediates of NO/cGMP signaling in mammalian tissues.
The NO/cGMP pathway involving NO-mediated activation of soluble guanylyl cyclase (GTP pyrophosphate-lyase (cyclizing), EC 4.6.1.2.; sGC 1 ) is essential to signal transduction in several biological systems (1). In the vasculature, NO/cGMP signaling is important for the regulation of blood pressure and platelet function (2); in the brain, this pathway controls the release of neurotransmitters such as glutamate and acetylcholine (3). Biosynthesis of NO is triggered by autacoids increasing the intracellular concentration of free Ca 2ϩ , resulting in activation of Ca 2ϩ /calmodulin-dependent NOS (EC 1.14.13.39), complex homodimeric enzymes that catalyze the synthesis of NO from the guanidino moiety of the amino acid L-arginine (4 -7). The oxidation of L-arginine is catalyzed by a cytochrome P 450 -type heme iron in the oxygenase domain of NOS with O 2 serving as a cosubstrate. The electrons required for reduction of O 2 are shuttled from the cofactor NADPH to the heme via a flavincontaining cytochrome P 450 reductase that forms the C-terminal half of the NOS protein. This electron transport chain only operates when Ca 2ϩ /calmodulin is bound to the enzyme, which then effects the Ca 2ϩ regulation of endothelial and neuronal NO synthesis.
At low concentrations of L-arginine or in its absence, the enzymatic reduction of O 2 uncouples from substrate oxidation and results in the generation of superoxide anions and H 2 O 2 (8 -12). The effective coupling of the reaction requires not only saturation with L-arginine but also the pteridine cofactor H 4 biopterin (13). Since the two subunits of neuronal NOS bind H 4 biopterin in a highly anticooperative manner, the purified enzyme always contains Յ1 molecule of H 4 biopterin/dimer, i.e. it consists of a H 4 biopterin-containing and a H 4 biopterin-free subunit (14). In this state, the enzyme can form L-citrulline and is stimulated about 2-fold upon binding of H 4 biopterin to the low affinity site of the pteridine-free subunit. Together with our recent findings that the two NOS subunits function independently (15), this very unusual binding behavior of H 4 biopterin appears to have interesting functional consequences; at low concentrations of free H 4 biopterin, the H 4 biopterin-containing subunit of NOS is expected to form NO, whereas the uncoupled NADPH oxidation catalyzed by the pteridine-free subunit should generate O 2 . , which reacts at a nearly diffusion-controlled rate (k ϭ 4.3-6.7 ϫ 10 9 M Ϫ1 s Ϫ1 ) with NO to form peroxynitrite, a powerful oxidant and cytotoxic species (16) (17,20). As an alternative explanation of these puzzling results, it has been suggested that the true product of NOS is NO Ϫ (18), which could be converted to NO by stoichiometric amounts of SOD (21). NO was shown to be the only redox form of nitrogen monoxide capable of activating sGC (22), raising the question of how the NO signal is carried from NOS to sGC. A possible mediator is peroxynitrite acting via the nitrosation of GSH to GSNO. Although this reaction is inefficient and yields maximally 1% GSNO (23,24), there could be as yet unrecognized more efficient nitrosative pathways accounting for GSNO biosynthesis in vivo. Alternatively, cells may contain sufficient SOD to prevent inactivation of NO by O 2 . . As it is difficult to predict which of these pathways is involved in the true chemical route of NOS/sGC signaling, we have designed an in vitro study using a reconstituted system consisting of purified neuronal NOS and purified sGC. This system allowed us to compare the action profile of NOS with that of donor compounds of several nitrogen oxides and to study the activation of sGC by active NOS in the presence of GSH and SOD.

EXPERIMENTAL PROCEDURES
Materials-Purified recombinant rat neuronal NOS was obtained from baculovirus-infected insect cells as described (25). sGC was purified from bovine lung by immunoaffinity chromatography as described previously (26). Alkaline stock solutions of peroxynitrite (80 -100 mM) were prepared and quantified as described (27 Isolated Heart Perfusion-Rat hearts were isolated and retrogradely perfused with Krebs-Henseleit bicarbonate buffer at 9.0 ml ϫ min Ϫ1 ϫ g Ϫ1 wet weight as described previously (28). After equilibration (45 min), hearts were perfused via side line for 15 min with bradykinin (0.1 M) or SPER/NO (0.1 mM), followed by perfusion (15 min) with L-NNA (0.2 mM), the copper ion-selective chelators neocuproine or BCS (both Cu ϩ -selective), or cuprizone (Cu 2ϩ -selective) (all chelators at 0.2 mM) in the absence of agonists. Finally, agonists were perfused together with copper chelators or L-NNA, again over 15 min. Coronary effluent samples were collected in 3-min intervals during the first and last period of 15 min, i.e. the agonist infusion periods, and stored at Ϫ20°C pending analysis by radioimmunoassay.
Determination of GSNO-HPLC analysis of GSNO was performed on a C 18 reversed phase column with 20 mM phosphate buffer (pH 3.0), containing 20 M neocuproine to prevent GSNO decomposition, at 1 ml/min and detection at 338 nm as described (24). To remove nitrite, samples were treated with 10 l of ammonium sulfamate (10 mM) prior to acidification (pH ϳ3). Calibration curves were established with GSNO under enzyme assay conditions. GSNO was not detectable in samples containing up to 0.1 mM NaNO 2 .
For a more sensitive determination of GSNO, NO was released by addition of Cu(NO 3 ) 2 (29) and quantified with an NO electrode connected to an Apple Macintosh computer via an analog-to-digital (A/D) converter (World Precision Instruments, Mauer, Germany) (17). The output current was recorded at 0.6 Hz under constant stirring at 37°C in open glass vials. For calibration, Cu(NO 3 ) 2 (10 mM final) was added to GSNO under NOS/sGC assay conditions (see below). Linear calibration curves were obtained from plots of the NO versus the GSNO (0.05-5 M) concentration. GSNO was not detectable in samples containing up to 50 M NaNO 2 . As described in detail elsewhere, 2 this method is selective for thionitrites but does not allow the specific detection of GSNO in biological samples where sulfhydryls other than GSH may be present.
In some experiments, the total concentration of nitrosothiols was determined spectrophotometrically according to the method of Saville (30).

Effect of GSH on the Activation of sGC by NO Donors and
Neuronal NOS-To allow a reliable comparison of the effects of NO donors with that of the enzymatic NOS product(s), determination of sGC activity was performed in 50 mM phosphate buffer (pH 7.4) containing 0.5 mM GTP, 3 mM MgCl 2 , 1 mM cGMP, 0.1 mM L-arginine, 50 M NADPH, 10 M CaCl 2 , and 10 g/ml calmodulin. Under these conditions, NOS activity was 0.08 -0.1 mol of L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 , whether or not GSH (1 mM) was present.
In the absence of added NO, purified sGC exhibited basal activities of 0.04 Ϯ 0.01 and 0.19 Ϯ 0.06 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 without and with 1 mM GSH, respectively. As shown in Fig. 1, cGMP formation was increased about 300-fold to 12.32 Ϯ 0.55 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 by the NO donor DEA/NO (1 M); the identical maximal enzyme activity was obtained in the presence of GSH. Peroxynitrite (0.1 mM) had no effect in the absence of the thiol but led to a moderate enzyme stimulation (about 20-fold) when incubated with 1 mM GSH. GSH (1.01 Ϯ 0.19 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 ) but was nearly as effective as DEA/NO in the presence of the thiol (9.50 Ϯ 1.01 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 ). The nitroxyl donor Angeli's salt decomposes with a half-time close to that of DEA/NO (t1 ⁄2 at 37°C and pH 7.4 ϭ 2.1 and 2.3 min, respectively) (34). At 10 M, i.e. a concentration that yields 10-fold higher rates of product release than the maximally active concentration of DEA/NO (1 M), Angeli's salt produced only little effects in the absence and presence of GSH (0.64 Ϯ 0.09 and 1.46 Ϯ 0.29 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 , respectively). The effects of the nitrogen oxide donors were compared with that of Ca 2ϩ /calmodulin-activated neuronal NOS (containing ϳ0.4 eq of H 4 biopterin/monomer) coincubated with sGC at saturating concentrations of L-arginine but in the absence of exogenous H 4 biopterin. At a maximally effective concentration of 200 ng/ 0.1 ml, NOS stimulated sGC to 1.22 Ϯ 0.12 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 in the absence of GSH but led to a pronounced accumulation of cGMP in the presence of the thiol (8.51 Ϯ 0.88 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 ). GSH triggered sGC activation by NOS with an EC 50 value of 0. 21 Ϯ 0.06 mM (data not shown).
Effects of SOD on sGC Activation by NOS-As shown in Fig.  2A, presence of SOD led to an apparently GSH-independent activation of sGC by NOS (200 ng/0.1 ml). The effective concentrations of SOD (EC 50 ϭ 0.66 Ϯ 0.13 unit/ml) were about 200-fold lower than those required to detect free NO under the same conditions (EC 50 ϭ 135 Ϯ 7 units/ml; data not shown). Identical data were obtained with MnSOD from Escherichia coli (data not shown). In the presence of GSH, SOD increased NOS-stimulated cGMP formation to 13.74 Ϯ 1.28 mol ϫ mg Ϫ1 ϫ min Ϫ1 , a value close to that obtained with DEA/NO (cf. Fig.  1). As evident from the concentration-response curves shown in Fig. 2B, the system was highly sensitive to NOS. In the presence of GSH, the Ca 2ϩ /calmodulin-activated enzyme stimulated sGC with an EC 50 value of 22.5 Ϯ 7.5 ng/0.1 ml (0.7 nM) (filled circles). SOD (unfilled squares) decreased the EC 50 of NOS to 2.2 Ϯ 0.6 ng/0.1 ml. The slight increase of basal sGC activity (2.56 Ϯ 0.74 mol of cGMP ϫ mg Ϫ1 ϫ min Ϫ1 ) was inhibited by hemoglobin and the heme-site inhibitor ODQ (35) (data not shown), indicating that it was due to stabilization of ambient NO (36). If SOD and GSH were present together (filled squares), the resulting EC 50 (8.2 Ϯ 1.2 ng of NOS/0.1 ml) was between the values obtained with either SOD or GSH alone.
Characterization of NOS-induced sGC Activation in the Presence of GSH-The surprising difference between sGC activation by NO/O 2 . (generated by NOS or SIN-1) and peroxynitrite (cf. Fig. 1) prompted us to perform further experiments to better define their respective modes of action and to characterize possible intermediates of these pathways. Very similar data to those described below were obtained when sGC was activated with the NO/O 2 . donor SIN-1 instead of NOS. 3 Under physiological conditions, the reactivity of peroxynitrite is dramatically modified by its rapid reaction with CO 2 (37). As shown in Fig. 3, CO 2 (delivered as NaHCO 3 ) almost completely blocked GSH-dependent activation of sGC by peroxynitrite (A) but only slightly reduced cGMP accumulation triggered by NOS (B). Since latter effect was also observed when DEA/NO was used to activate sGC (data not shown), NaHCO 3 and/or CO 2 apparently slightly interfere with NO stimulation of sGC.
These data showed that the GSH-dependent activation of sGC by the NOS products NO  intermediate with NO-like biological activity. The thionitrite GSNO appeared to be a likely candidate. This compound is not a direct activator of sGC but releases free NO in the presence of trace amounts of Cu 2ϩ and reducing agents such as thiols or ascorbate (24,29). We analyzed the product of NOS and GSH by HPLC and obtained a peak with an absorbance maximum at 338 nm, which co-eluted with authentic GSNO (Fig. 4). GSNO formation triggered by NO/O 2 . was unlikely to be catalyzed by trace metals because neocuproine (20 M) was present throughout the experiments and neither EDTA (10 mM), nor diethylenetriamine pentaacetic acid (50 M), nor cuprizone (50 M) had any effect on GSNO formation from SIN-1 and GSH (1 mM each) (data not shown). Because EDTA and diethylenetriamine pentaacetic acid inhibited NOS activity measured as formation of L-citrulline (data not shown), these chelators could not be tested in the NOS/GSH system. Enzymatic formation of an S-nitrosothiol was further confirmed by incubation of NOS in the presence of 1 mM GSH and electrochemical detection of NO release by treating the samples with Cu(NO 3 ) 2 (Fig. 5A). In agreement with previous studies (17,18), Ca 2ϩ /calmodulin-activated NOS did not produce detectable amounts of free NO. However, a pronounced NO signal was obtained when Cu(NO 3 ) 2 was added to the incubation mixtures 6 min later. The shape of the signal was identical to that obtained by addition of Cu(NO 3 ) 2 to authentic GSNO, and Cu(NO 3 ) 2 was without effect when NOS was incubated in the absence of a thiol (data not shown). Together with the HPLC data, these results strongly suggested that incubation of NOS with GSH led to formation of GSNO, which released NO in a copper-dependent reaction. The enzymatic formation of GSNO was insensitive to NaHCO 3 (Fig. 5A), in marked contrast to the nitrosation reaction triggered by ONOO Ϫ (Fig. 5B). The same differential effects of NaHCO 3 were observed when GSNO was analyzed by HPLC (data not shown).
The electrochemical method was used to quantify GSNO formation by NOS incubated in the presence of 1 mM GSH. To compare the values with NOS activity, formation of L-citrulline was determined in the same samples. As shown in Fig. 6, formation of GSNO depended on the enzyme concentration and accounted for 25-45% of the L-citrulline production. For instance, at a NOS concentration of 100 ng/0.1 ml, which produced ϳ80% of maximal sGC stimulation (cf. Fig. 2B), NOS generated 0.61 Ϯ 0.12 M L-citrulline and 0.23 Ϯ 0.03 M GSNO. Since authentic GSNO stimulated sGC under identical conditions with an EC 50 of 0.35 Ϯ 0.06 M (data not shown), these data suggest that sGC activation by NOS is explained by enzymatic formation of GSNO.
Incubation of NOS at a high concentration (2,000 ng/0.1 ml) but under otherwise identical conditions allowed the quantitative determination of GSNO with additional, less sensitive  ions affects agonist-induced cGMP formation in a perfused isolated rat heart system. As shown in Fig. 7A, the endotheliumdependent vasodilator bradykinin stimulated a N G -nitro-L-arginine (L-NNA)-sensitive release of cGMP into the coronary effluent, which was strongly inhibited by the Cu ϩ -selective chelators neocuproine and BCS but not by the Cu 2ϩ chelator cuprizone (38). To exclude that these effects were due to an interference of the drugs with sGC stimulation or cGMP outward transport, similar experiments were performed with the NONOate SPER/NO, which is expected to activate cardiac sGC directly. Release of cGMP triggered by the NO donor was not affected by neocuproine or by cuprizone (Fig. 7B), demonstrating that Cu ϩ -specific chelators are specific inhibitors of agonist-induced cGMP release from rat heart.

DISCUSSION
The present study was designed to elucidate the molecular mechanisms involved in cGMP accumulation triggered by Ca 2ϩ -activated NOS. Purified sGC was functionally reconstituted with purified neuronal NOS or incubated with donors of nitrogen oxides to learn about the conditions required for the effective transduction of the NOS signal and the identity of the intermediates involved. Special emphasis was given to the role of the sulfhydryl compound GSH, which occurs in concentrations of 1-10 mM in mammalian cells (39). The NO donor DEA/NO stimulated sGC in a GSH-independent manner, demonstrating that the direct activation of the enzyme by NO does not require the presence of a thiol. However, earlier reports appear to show the opposite (40), but sodium nitroprusside or organic nitrates, which were used as NO donors before better drugs became available in the early 1990s, do not release free NO in well defined first order reactions but require chemical transformations to become biologically active.
NOS has never been observed to produce NO unless fairly high amounts of SOD were present (17,18 . donor SIN-1 (33), with a pronounced GSH dependence of sGC activation that was overcome by addition of SOD. According to an alternative proposal, NOS produces nitroxyl (NO Ϫ ), which is converted to NO by stoichiometric amounts of SOD (18). However, we observed only minor activation of sGC with Angeli's salt at a concentration yielding a 10-fold higher rate of product release than achieved with a maximally active concentration of DEA/NO (1 M). The small effect that we observed with Angeli's salt may be due to release of minor amounts of NO and/or aerobic conversion of NO Ϫ to NO (41). Together with the present knowledge about NOS enzymology, our results argue against NO Ϫ as the biologically active species produced by NOS, although we cannot exclude that the enzyme produces an NO Ϫ species with different chemical properties than the NO Ϫ released from Angeli's salt. Activation of sGC by NOS was rendered GSH-independent by remarkably low concentrations of SOD, which are clearly below the average concentration of SOD in tissues (10 M corresponding to about 1,000 units/ml) (42). A maximal effect was obtained with 5 units of SOD/ml, conditions under which enzymatic NO formation is below the detection limit of the NO electrode. 4 These surprising results are difficult to explain but may be of utmost physiological importance because they suggest that low levels of SOD, which are far below the levels required to outcompete the peroxynitrite reaction, may be sufficient for the effective functional coupling of NOS activity to cGMP accumulation. At a first glance, the data also seem to suggest that the reaction of NO/O 2 . with GSH is less important because of the ubiquitous occurrence of SOD in tissues. However, GSH largely antagonized the pronounced leftward shift of the NOS concentration-response curve caused by SOD (cf. these major components of the cellular defense machinery against oxidative stress (43) (44), GSNO formation must involve a comparably rapid reaction. Free NO does not nitrosate thiols at significant rates (45), but the reaction of NO with O 2 results in formation of a potent nitrosating intermediate (46,47). The autoxidation reaction is second order with respect to NO and thus too slow to compete with peroxynitrite formation at submicromolar concentrations of NO (48 -50). According to a recent proposal, however, S-nitrosation does occur at low NO concentrations with O 2 or other oxidants serving as electron acceptors (51). Unfortunately, there is no rate constant available for this reaction, making it difficult to judge whether it could account for our observations. The reaction of NO with peroxynitrite, which also could yield a nitrosating species is too slow to be involved in GSNO formation (27 It is conceivable that GSH oxidation by initially formed peroxynitrite is a source of thiyl radicals. Such a mechanism would predict that maximally 50% of product flow were recovered as GSNO, a value that agrees well with our observations (cf. Fig. 6). Based on the finding that GSNO formation was insensitive to NaHCO 3 , this proposal has to be dismissed if the 1-electron oxidation of GSH by peroxynitrite (52,53) is inhibited by CO 2 . Current evidence points to a partial inhibition (54,55), but the effect of CO 2 on thiyl radical formation has not been reported. Although the precise mechanism of NO/O 2 . -triggered S-nitrosation remains to be clarified, this pathway may at least partially explain the occurrence of S-nitrosothiols in mammalian tissues (56)(57)(58). These compounds are relatively insensitive to O 2 and O 2 . and may serve as a storage or transport form of NO (56,59,60). GSNO can nitrosate thiol groups of proteins and so regulate important physiological variables, e.g. the activity of tissue-type plasminogen activator (61), the down-regulation of the N-methyl-D-aspartate receptor (62), and the O 2 affinity of hemoglobin (63). Our data show that NO and O 2 . take part in reactions clearly more diverse than previously known (see scheme in Fig. 8).
The key determinants of the link between NOS and sGC are the levels of SOD and GSH. With less GSH or SOD, as inferred in certain pathologies (64,65), NO/O 2 . will exacerbate peroxynitrite-mediated tissue damage (16). Low levels of SOD would not prevent formation of peroxynitrite but appear to be sufficient for the coupling of NOS activity to cGMP accumulation. Whether SOD is present or not, the NO/O 2 . pathway appears to be significantly shifted toward GSNO in the presence of physiologically occurring concentrations of GSH.
Since the biological activities of GSNO differ considerably from those of NO (66), the homolytic cleavage of the thionitrite may represent an additional regulatory site of NO/cGMP signaling by which the outcome of NOS activity is switched toward cGMP accumulation, vascular relaxation, inhibition of platelet function, and synaptic transmission in the brain. It is well established that GSNO decomposition is catalyzed by Cu ϩ ions (29,38,67,68). Although a marked mobilization of redoxactive copper occurs in myocardial ischemia/reperfusion injury (69), the cellular availability of Cu ϩ is probably limited under physiological conditions, suggesting that enzymatic pathways of GSNO decomposition may be more relevant than the nonenzymatic Cu ϩ mechanism. Several enzymes including GSH peroxidase (70,71), thioredoxin reductase (72), and ␥-glutamyl transpeptidase (73) appear to catalyze reactions leading to NO release from GSNO. Of note, a Cu ϩ -dependent enzymatic activity was reported to catalyze GSNO decomposition in platelets (74). In the present study, we found that Cu ϩ -selective chelators led to a pronounced inhibition of bradykinin-induced release of cGMP into the coronary effluent of isolated perfused rat hearts, whereas cGMP release upon direct activation of cardiac sGC was not affected by the chelators. Although these data are good circumstantial evidence that Cu ϩ -dependent release of NO from endogenous thionitrites may be essentially involved in cardiac NO/cGMP signaling, further studies are needed to unequivocally demonstrate the role of GSNO as a product of the NOS pathway in mammalian tissues.  (16). Nitrosation of GSH by peroxynitrite (path a1) occurs at low efficiency (Յ1%) (23,24) and is inhibited by CO 2 (delivered as NaHCO 3 ). Presence of GSH (path b) leads to conversion of 25-45% of total NO/O 2 . to GSNO, which is neuroprotective and may serve as storage form of NO (66). Release of NO from GSNO (path b1) may involve enzymatic and non-enzymatic mechanisms. In the absence of GSH, the pathway is also shifted toward free NO and cGMP by relatively low concentrations (ϳ5 units/ml) of the O 2 . scavenger SOD (path c).