Human S-Nitroso Oxymyoglobin Is a Store of Vasoactive Nitric Oxide*

Nitric oxide (·NO) regulates vascular function, and myoglobin (Mb) is a heme protein present in skeletal, cardiac, and smooth muscle, where it facilitates O2 transfer. Human ferric Mb binds ·NO to yield nitrosylheme and S-nitroso (S-NO) Mb (Witting, P. K., Douglas, D. J., and Mauk, A. G. (2001) J. Biol. Chem. 276, 3991–3998). Here we show that human ferrous oxy-myoglobin (oxyMb) oxidizes ·NO, with a second order rate constant k = 2.8 ± 0.1 × 107 m–1·s–1 as determined by stopped-flow spectroscopy. Mixtures containing oxyMb and S-nitrosoglutathione or S-nitrosocysteine added at 1.5–2 moles of S-nitrosothiol/mol oxyMb yielded S-NO oxyMb through trans-nitrosation equilibria as confirmed with mass spectrometry. Rate constants for the equilibrium reactions were kforward = 110 ± 3 and kreverse = 16 ± 3 m–1·s–1 for S-nitrosoglutathione and kforward = 293 ± 5 and kreverse = 20 ± 2 m–1·s–1 for S-nitrosocysteine. Incubation of S-NO oxyMb with Cu2+ ions stimulated ·NO release as measured with a ·NO electrode. Similarly, Cu2+ released ·NO from Mb immunoprecipitated from cultured human vascular smooth muscle cells (VSMCs) that were pre-treated with diethylaminenonoate. No ·NO release was observed from VSMCs treated with vehicle alone or immunoprecipitates obtained from porcine aortic endothelial cells with and without diethylaminenonoate treatment. Importantly, pre-constricted aortic rings relaxed in the presence of S-NO oxyMb in a cyclic GMP-dependent process. These data indicate that human oxyMb rapidly oxidizes ·NO and that biologically relevant S-nitrosothiols can trans-(S)nitrosate human oxyMb. Furthermore, S-NO oxyMb can be isolated from cultured human VSMCs exposed to an exogenous ·NO donor at physiologic concentration. The potential biologic implications of S-NO oxyMb acting as a source of ·NO are discussed.

M ؊1 ⅐s ؊1 for S-nitrosoglutathione and k forward ‫؍‬ 293 ؎ 5 and k reverse ‫؍‬ 20 ؎ 2 M ؊1 ⅐s ؊1 for S-nitrosocysteine. Incubation of S-NO oxyMb with Cu 2؉ ions stimulated ⅐ NO release as measured with a ⅐ NO electrode. Similarly, Cu 2؉ released ⅐ NO from Mb immunoprecipitated from cultured human vascular smooth muscle cells (VSMCs) that were pre-treated with diethylaminenonoate. No ⅐ NO release was observed from VSMCs treated with vehicle alone or immunoprecipitates obtained from porcine aortic endothelial cells with and without diethylaminenonoate treatment. Importantly, pre-constricted aortic rings relaxed in the presence of S-NO oxyMb in a cyclic GMP-dependent process. These data indicate that human oxyMb rapidly oxidizes ⅐ NO and that biologically relevant S-nitrosothiols can trans-(S)nitrosate human oxyMb. Furthermore, S-NO oxyMb can be isolated from cultured human VSMCs exposed to an exogenous ⅐ NO donor at physiologic concentration. The potential biologic implications of S-NO oxyMb acting as a source of ⅐ NO are discussed.
Endothelium-derived ⅐ NO, generated through the action of nitric-oxide synthase(s) on L-arginine, plays a vital role in blood vessel dilation and thereby in the regulation of peripheral vascular resistance and ultimately circulating blood pressure (1,2). To elicit vessel dilation, ⅐ NO binds to and activates its molecular target, soluble guanylyl cyclase (3), within vascular smooth muscle cells (VSMCs), 1 which in turn catalyzes the conversion of guanosine-5Ј-(3-thiotriphosphate) to cyclic GMP (cGMP) (4). Synthesized cGMP activates a cascade of effector proteins that initiates VSMC relaxation and thereby promotes vessel dilation.
The heme protein myoglobin (Mb) is present in cardiac, skeletal, and human smooth muscle (5,6). In cardiac muscle, the concentration of Mb ranges from 0.3 to 0.5 mM (5), whereas the precise concentration of Mb in smooth muscle is not known. The role of intracellular Mb is generally accepted as that of a passive di-oxygen storage protein that facilitates di-oxygen transfer from the extra-to intracellular space. However, in vitro studies have shown that oxygenated ferrous Mb (oxyMb) also rapidly reacts with dissolved ⅐ NO gas (k ϭ 10 7 M Ϫ1 ⅐s Ϫ1 ) to yield higher order N-oxides such as nitrate (7). In addition, both ferrous deoxy and ferric Mb form stable heme-NO complexes (Mb⅐NO) with dissolved ⅐ NO gas (dissociation constant K d ϭ 10 Ϫ5 M) (8). Together, these chemical reactions have the potential to effectively eliminate ⅐ NO within its expected lifetime in biological systems, suggesting that Mb could play an active role in maintaining ⅐ NO homeostasis. Indeed, there is support for this notion. For example, Mb limits the extent of ⅐ NO-induced inactivation of cytochrome c oxidase (9). Also, conversion of ferrous to ferric Mb regulates the myocardial concentration of ⅐ NO (10), which is crucial for maintaining overall heart function, coronary blood flow, and contractility, processes that are more severely affected by ⅐ NO in mice lacking Mb compared with wild-type animals (10). Thus, the focus for Mb has shifted from O 2 transport to a central regulatory role in ⅐ NO homeostasis (11).
An intriguing feature of human Mb is that it possesses a reactive cysteine residue (Cys 110 ) (12,13). Under aerobic conditions, Cys 110 reacts with ⅐ NO to yield S-NO Mb (14), similar to S-NO Hb (15). Unlike Hb, however, for which nitrosation is dependent on the allosteric (R-T) transition (16), the degree of Mb oxygen saturation does not affect accessibility of the Cys 110 for S-nitrosation, as judged by comparable x-ray crystal structures of ferric and ferrous Mb bound to a diatomic ligand (see Fig. 5  tains Mb in the reduced state for oxygenation to yield oxyMb (18). It is not clear at present whether S-nitrosation of ferrous oxyMb is feasible under physiologic conditions, although one would expect oxyMb to rapidly oxidize ⅐ NO based on the high rate constant for the reaction of oxyMb and dissolved ⅐ NO gas (19). Herein we demonstrate that Cys 110 S-nitrosation of human ferrous oxyMb (the predominant physiologic form of the protein) occurs through trans-nitrosation equilibria reactions with low molecular mass S-nitrosothiols (RS-NO), that ⅐ NO released from S-NO oxyMb can dilate pre-constricted vessels similar to authentic endothelium-derived relaxant factor, and that Mb immunoprecipitated from VSMCs pre-treated with an ⅐ NO donor can release ⅐ NO.
Animals-New Zealand White rabbits (2.5-3 kg) were obtained from a commercial farm (Wauchope, New South Wales, Australia) and housed individually for the entire study period. Rabbits received normal chow with feed and water provided ad libitum for an acclimation period of 2 weeks. Local ethics committee approval was obtained before commencing the study.
Cell Culture-Cultured primary human VSMCs (American Type Culture Collection) and porcine aortic endothelial cells (PAECs; Cell Applications) were maintained in Dulbecco's modified Eagle's medium/ Ham's F-12 (JRH Biosciences) and M199 media, respectively. All media preparations were supplemented with 10% fetal bovine serum (Sigma), 2 mm L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a humidified atmosphere of 5% CO 2(g) . Culture media for PAECs also contained 50 g/ml heparin sulfate. For experiments, VSMCs and PAECs were cultured to 90 -100% confluence and used between passage 5 and 8.
Preparation of Recombinant Wild-type and C110A Variant Human Mb-DNA manipulations were performed as described previously (22). Point mutations in the Mb sequence were confirmed by DNA sequence analysis before protein expression in bacteria. Once the sequence was confirmed, the BamHI-HindIII fragment from the amplified DNA, which also contained the mutant Mb coding, was ligated to the BamHI-HindIII fragment from the pMb3 vector to yield the expression vector (12) that was later transformed to the appropriate cell line for protein expression as described previously (23). Preparations of purified wildtype and C110A variant of human Mb were snap-frozen in liquid nitrogen and stored at Ϫ80°C before use. All preparations exhibited a A 409 /A 280 ratio of peak absorbance of 5 (data not shown), indicative of the purity of the protein preparations.
Preparation of S-NO oxyMb-Recombinant human oxyMb was prepared by chemical reduction of the recombinant human ferric Mb with a 2-fold excess of DTT and stirring under an atmosphere of air for 10 min. Residual DTT was removed from the preparation by three successive gel filtration columns (PD-10 pre-packed column; Amersham Biosciences). Formation of oxyMb was confirmed by electronic absorbance spectroscopy with characteristic absorptions at 543 and 580 nm. Samples of recombinant human oxyMb were then treated with either Snitrosoglutathione (GS-NO) or S-nitrosocysteine (CYS-NO) (final concentration of added RS-NO Ͻ 2 mol/mol oxyMb) dispersed in phosphate buffer (50 mM, pH 7) and then left to equilibrate in the dark at 20°C. Stock solutions of GS-NO or CYS-NO were prepared immediately before use as described previously (24). After 60 min of equilibration, excess low molecular mass RS-NO was removed by repeated size exclusion chromatography. For samples designated for mass analyses, S-NO oxyMb was purified with simultaneous change of buffer to Nanopure water containing 100 M DTPA (high salt concentrations yield protein adducts that affect mass determinations). Finally, S-nitrosation of oxyMb was confirmed by an increased absorbance at ϳA 330 (⑀ 333 nm ϭ 3667 M Ϫ1 ⅐cm Ϫ1 (14)) and verified unambiguously by electrospray ionization mass spectrometry (ESI-MS) (see below). Samples of S-NO oxyMb were maintained in the dark and used within 5 min of purification.
Stopped-flow Kinetic Measurements-Where required, saturated solutions of authentic ⅐ NO gas were prepared, and the concentration of dissolved gas was standardized as described previously (14). Kinetic determinations for the reaction of oxyMb with dissolved ⅐ NO gas were performed with an Applied Photophysics SX-17 MV stopped-flow spectrophotometer as described previously (23). Typically, 250 time-dependent spectra (logarithmic time base; integration, 2.56 ms; dead time, ϳ2 ms; ϭ 350 -750 nm; resolution, 1 nm) were collected at 25°C. Kinetic data were processed using Pro-Kineticist global analysis software (Pro-Kineticist version 4.1; Applied Photophysics, Leatherhead, UK), as described previously (25). Apparent rate constants (k obs ) were then determined by linear regression.
Mass Analyses-Where required, molecular mass was measured routinely for native and modified Mb by ESI-MS as described in detail elsewhere (14). Briefly, mass spectra were acquired using a hybrid tandem mass spectrometer (Applied Biosystems, Foster City, CA). Samples (ϳ10 pmol, 1 l) were dissolved in water:acetonitrile (ϳ20:80) and loaded into nanospray needles (Proxeon), and the tip was positioned ϳ10 mm from the orifice. Nitrogen was used as curtain gas, and a potential of ϩ800 V was applied to the needle. Next, time-of-flight scan was acquired (m/z 50 -2000, 1 s) and accumulated for ϳ1 min into a single file. These conditions favor the detection of Mb apoprotein due to unfolding of the tertiary structure and loss of the heme prosthetic group (26). Mass accuracy of the system was tested routinely before use. Mass values were obtained by standard fitting analyses of the various m/z distributions.
Measurement of ⅐ NO from RS-NO-Release of ⅐ NO from a range of low molecular mass or proteinaceous RS-NO was monitored by a ⅐ NOselective electrode (ISO-NO MII; World Precision Instruments Inc.) coupled with a DUO-18 TM data recorder (v1.55; World Precision Instruments Inc.). Briefly, the electrode was pre-equilibrated in phosphate buffer (50 mM, pH 7.4) containing 100 M copper(II)sulfate (Cu 2ϩ ) under an atmosphere of argon gas. Authentic RS-NO (or immunoprecipitate obtained from VSMCs or PAECs) was added to the solution, and the time-dependent increase in current was monitored until the current stabilized. Area under the peak response curve was estimated with integration software supplied with the data recording system. The amount of ⅐ NO liberated from the various RS-NO was then compared with a standard curve generated using authentic GS-NO (solution concentration standardized with ⑀ 366 nm ϭ 770 M Ϫ1 ⅐cm Ϫ1 ) prepared as described previously (24). Standard curves were generated before commencement of individual experiments to verify electrode function and to account for the day-to-day variation in electrode response factor.
Assessment of Vessel Relaxation-Vessel bioassays were performed with a vessel myobath (WPI, Coherent, Australia) as described in detail elsewhere (27). Briefly, rabbit aortic rings were suspended in organ chambers and incubated in Krebs buffer with constant degassing (Carbogen gas mixture, 5% CO 2 , 95% O 2 ) for 30 min. After equilibration, rings were pre-constricted with an increasing dose of phenylephrine. Next, the rings were washed thoroughly to allow complete relaxation and then pre-constricted with a dose of phenylephrine to yield half the maximal constriction force value. Vessel relaxation was assayed in response to SNP (positive control), S-NO oxyMb (prepared by transnitrosation with CYS-NO), and the oxyC110A variant of human Mb pre-treated with CYS-NO in identical fashion to wild-type oxyMb (negative control), and dilation was expressed as a percentage of the preconstriction force value. In some studies, vessels were denuded of the endothelium by gently applying the blunt end of a surgical tweezer to the inner surface of the aortic ring. This procedure had no effect on constriction to phenylepherine, although it eliminated the vessel response to the endothelium-dependent agonist acetylcholine (data not shown).
Preparation of Aortic Homogenates for cGMP Assessment-To determine tissue cGMP, aortic rings were first incubated at 37°C in Krebs buffer supplemented with 200 M 3-isobutyl-1-methylxanthine. Next, the rings were exposed to the various vasoactive agents for 15 min, removed from the myobath and immediately cut into small pieces, frozen in liquid nitrogen, and finally pulverized into a powder with a mortar and pestle. The powdered tissue was then transferred to a glass tube and diluted with 2 ml of DPBS containing 5 M butylated hydroxytoluene, 2 mM EDTA, 200 M 3-isobutyl-1-methylxanthine, and Com-plete® protease inhibitors mixture added as per the manufacturer's instruction (Roche Applied Science). The tissue was then homogenized with a rotating piston and matching Teflon-lined tube as described previously (28). Samples of homogenate (50 l) were removed for protein determination (BCA assay; Sigma), and the remainder was employed for tissue cGMP determinations using a commercial kit (Cayman Chemical).
Monitoring Accumulation of the Native Human Mb and Its Homodimer-Where required, reaction mixtures employed for ⅐ NO evolution studies were subsequently analyzed for the accumulation of a Mb disulfide dimer (expected mass, 34,107 atomic mass units) by SDS-PAGE and after staining with Coomassie Blue as described previously (13). Where required, selected samples were also pre-incubated with DTT to reduce disulfide cross-links.
In some experiments, confluent cultured human VSMCs or PAECs (ϳ3-5 ϫ 10 6 cells) were washed thoroughly with HPSS prepared as described previously (29), overlaid with HPSS, and treated without (vehicle alone, control) or with DeaNO administered at final concentration of 100 nM or 10 M (corresponding to a rate of ⅐ NO release of 0.6 or 60 nM⅐s Ϫ1 , respectively, as determined from the half-life (t1 ⁄2 ) ϭ 2 min at 37°C indicated by the manufacturer). Control and ⅐ NO-treated cells were harvested after a 5-or 60-min incubation, washed thoroughly with phosphate-buffered (50 mM, pH 6.5) lysis solution containing 1% (v/v) Triton X-100, a Complete® mixture of protease inhibitors (Roche Applied Science), and 100 M DTPA. Next, cells were lysed by repeated (3ϫ) freeze-thaw and centrifuged (15,000 rpm), and the supernatant was treated with monoclonal anti-human Mb antibody (final dilution, 1:500 (v/v); Sigma) followed by addition of G-protein-linked Sepharose (Sigma) to yield immunoprecipitates of cytosolic Mb. The presence of Mb in isolated immunoprecipitates was confirmed by SDS-PAGE and Western blotting and tested for stored ⅐ NO using the ⅐ NO-selective electrode in the presence of 100 M Cu 2ϩ . Suitable controls included generation of immunoprecipitate samples with PAECs and G-proteinlinked Sepharose in the presence and absence of Mb antibody.
Kinetics of Human oxyMb Trans-nitrosation by Added RS-NO-The consumption of GS-NO or CYS-NO and concomitant accumulation of their corresponding reduced thiol forms were monitored time-dependently in the presence of human oxyMb using high performance liquid chromatography (HPLC) as described previously (30), with minor modification. Briefly, 25 M oxyMb was treated with 2-fold molar excess of GS-NO or CYS-NO in the presence of 100 M DTPA, and the reaction mixture was equilibrated in the dark at 20°C. Reaction mixtures were sampled at regular intervals over 60 min as indicated in the figures. Samples were then treated with trichloroacetic acid (final concentration, 4% (v/v)) to precipitate the protein and centrifuged, and the supernatant was analyzed using a LC-18 column (5 m, 25 ϫ 0.46 cm) eluted at 0.6 ml/min with a 10 mM sodium acetate buffer (pH 5.5) containing 50 M DTPA and monitored at 214 nm for RS-NO and reduced thiols. Retention times for the various analytes were as follows: GSH, 5.1 min; GS-NO, 7.4 min; cysteine, 4.9 min; and CYS-NO, 5.8 min. The peak absorbance with retention time 7.8 min was present in all preparations of GS-NO and showed no increase in area in the presence of human oxyMb (data not shown). Reduced thiols were standardized using authentic commercial samples (GSH and cysteine), whereas corresponding RS-NO was generated and standardized as described previously (24). Where required, rate constants were determined by data fitting with a curve generated as described previously (31) and using Prism software version 3.0 (GraphPad Software).
Statistical Analyses-Statistical differences in relaxation response obtained from the vessel studies were determined with one-way analysis of variance analyses. Student's t tests were performed to determine significant changes between data sets, with Welch's correction employed for unequal variances where appropriate. In all cases, statistical significance was accepted at the 95% confidence interval (p Ͻ 0.05).

RESULTS
The second order rate constant for the reaction of oxyMb with dissolved ⅐ NO gas has been determined for a various mammalian Mb, and ranges between ϳ0.3 and 4.4 ϫ 10 7 M Ϫ1 ⅐s Ϫ1 (7, 19, 34). Consistent with these studies, mixing re-combinant human oxyMb with increasing concentrations of ⅐ NO resulted in the rapid, dose-dependent conversion of ferrous oxyMb to ferric Mb (Fig. 1, A and B). Global simulation of the data afforded estimates for the observed rate constant of 2.8 Ϯ 0.1 ϫ 10 7 M Ϫ1 ⅐s Ϫ1 with a linear fit of R 2 ϭ 0.98 (Fig. 1C). Parallel steady-state product analyses indicated a rapid and near stoichiometric conversion of oxyMb to ferric Mb with increasing dose of the chemical ⅐ NO donor (DeaNO), as estimated by electronic absorbance spectroscopy (Fig. 2). Interestingly, ferric nitrosyl-Mb (Mb⅐NO) formed in relatively minor yields at Ն1 mol of the chemical ⅐ NO donor per mole of oxyMb (Fig. 2). In contrast to this rapid conversion of oxyMb to ferric Mb by DeaNO, addition in the dark of GS-NO or CYS-NO (data not shown) to human oxyMb at Ͻ2 mol RS-NO/mol oxyMb did not promote significant oxyMb oxidation (Fig. 2). Notably, Mb⅐NO was not detected (limit of detection, ϳ0.1 M) over the dose range of RS-NO employed. Increasing the ratio of added RS-NO to Ͼ2 mol RS-NO/mol oxyMb significantly increased the extent of oxyMb oxidation, indicating that free ⅐ NO was released from RS-NO despite the presence of the metal chelating agents EDTA and DTPA. Alternatively, RS-NO may be oxidized directly by ferrous deoxyMb, present in preparations of oxyMb, in analogy to the oxidation of GS-NO by ferrous deoxyhemoglobin (35). In contrast, addition of the corresponding concentration of GSH or cysteine alone did not affect rates of oxyMb autoxidation (data not shown). Subsequently, all trans-nitrosation reactions were performed with RS-NO in low Parallel mass analyses of the apoprotein in reaction mixtures containing oxyMb and ⅐ NO (derived from DeaNO) at molar ratios of 2 indicated that protein mass remained unchanged, suggesting that S-nitrosation had not occurred (data not shown). Electrospray ionization mass analyses were restricted to Mb apoprotein to avoid the possibility of any ambiguities derived from nitrosylation of the heme prosthetic group. Notably, S-NO oxyMb was formed through trans-nitrosation equilibria by reaction of oxyMb with physiologic low molecular mass RS-NO. Thus, reaction of recombinant human Mb (17,054 Ϯ 2 atomic mass units, mean Ϯ S.D.; n ϭ 3) with GS-NO or CYS-NO gave S-NO oxyMb (17,084 Ϯ 4 atomic mass units, mean Ϯ S.D.; n ϭ 3) as verified unambiguously by ESI-MS spectrometry (Fig. 3). Table I shows the percentage of conversion of human oxyMb to the corresponding S-NO oxyMb as determined by comparing the extent of S-NO oxyMb accumulation relative to the native unmodified protein in the same sample (Fig. 3, inset). It was noted that all preparations of S-NO oxyMb contained unmodified parent protein that was not separated from the S-nitrosated protein. Assuming that the distribution of native and modified Mb accounts for all the human Mb in the reaction, trans-nitrosation reactions with CYS-NO consistently afforded higher yields of S-NO oxyMb than that obtained with GS-NO (Table I). This difference in yield (CYS-NO versus GS-NO) reflects the relative steric bulk of glutathione relative to cysteine that imparts a greater stability to the corresponding low molecular mass S-nitroso-adduct (36) and results in a lower yield of S-NO oxyMb.
Trans-nitrosation is an equilibrium reaction occurring spontaneously under physiological conditions (31). To determine the respective second order rate constants, the time-dependent consumption of GS-NO (Fig. 4A) and CYS-NO (Fig. 4B) in the presence of human oxyMb was monitored together with the accumulation of GSH and cysteine. Using conditions identical to those employed to produce samples for mass analyses, the reaction profiles for trans-nitrosation of human oxyMb with a 2-fold molar excess of RS-NO were established (see the insets in Fig. 4, A and B). Consumption of RS-NO resulted in near stoichiometric accumulation of the corresponding reduced thiol that is a surrogate for S-NO oxyMb accumulation and at equi-librium closely matched the yield of S-NO Mb determined by mass spectrometry. This observation strongly supports the idea that native and S-nitrosylated Mb detected by mass spectrometry accounted for all Mb in each trans-nitrosation reaction, with no other significant protein modifications evident. Fitting the data shown in Fig. 4 to a second order process (31) afforded estimates of the values for the forward (k forward ) and reverse (k reverse ) rate constants for the various equilibria (Table I). Equilibrium constants (K) were then determined from the relationship K ϭ k forward /k reverse (Table I). Notably, the ϳ2-fold difference in the K value determined for trans-nitrosation with CYS-NO and GS-NO also reflected the relative yields of S-NO oxyMb estimated by mass spectrometry and quantitative HPLC analyses (Table I), indicating that S-nitrosylation of oxyMb is favored in the presence of low molecular mass RS-NO.
Both capture and the subsequent release of ⅐ NO from endothelium-derived relaxant factor (e.g. RS-NO) are important to the preservation of ⅐ NO bioactivity in vivo (21). At present, it is generally accepted that Cu 2ϩ catalyzes the decomposition of RS-NO to ⅐ NO and the corresponding disulfide dimer (37-39). To assess whether S-NO oxyMb was capable of releasing ⅐ NO, we therefore exposed various RS-NO to Cu 2ϩ . In the case of GS-NO, this yielded a time-dependent release of

TABLE I Trans-nitrosation of human oxyMb with added GS-NO or CYS-NO
Wild-type recombinant human oxyMb (25 M) was treated with 50 M RS-NO for 60 min at 20°C in the dark. Excess RS-NO was removed by gel filtration, and the yield of S-NO oxyMb was determined by ESI-MS. Rate constants k forward and k reverse were determined as described under "Experimental Procedures" (see also Fig. 4). The equilibrium constant (K) was determined from K ϭ k forward /k reverse (31). Data represent the mean Ϯ S.D. of four independent preparations of S-NO oxyMb. ⅐ NO (Fig. 5A) that corresponded to a linear dose-response curve (R 2 ϭ 0.99) over the concentration range tested (Fig.  5A, inset). Incubation of S-NO oxyMb (generated by transnitrosation with GS-NO or CYS-NO) also yielded ⅐ NO (Fig.  5B) to an extent proportional to that of protein S-nitrosation determined by ESI-MS (cf. Table I and Fig. 5C). In contrast, incubation of the C110A variant of human Mb that lacks the reactive thiol residue and cannot be S-nitrosated (14) or of unmodified wild-type Mb gave no significant response (Fig.  5C). Interestingly, the molar yield of ⅐ NO detected by the electrode was relatively low. For example, authentic GS-NO yielded ϳ0.6 mol ⅐ NO/mol substrate. This may reflect the competition for ⅐ NO between the ⅐ NO-selective electrode membrane and residual dissolved oxygen. Reaction between ⅐ NO and dissolved oxygen yields higher order N-oxides in a rapid reaction (40), whereas, at least for S-NO oxyMb, ⅐ NO oxidation may be facilitated by protein-bound di-oxygen. Analyses of the reaction mixtures containing S-NO oxyMb and Cu 2ϩ ions with SDS-PAGE indicated the presence of a Mb dimer sensitive to DTT (Fig. 5B, inset). Significantly, the extent of this homodimer formation reflected the yield of both S-nitrosated protein (Table I) and ⅐ NO (Fig. 5C). Next, we determined whether S-NO oxyMb is a potential source of bioactive ⅐ NO using a biological model system that assesses vascular function (Fig. 6). Addition of S-NO oxyMb to pre-constricted rabbit aortic vessels caused an immediate and dose-dependent relaxation of magnitude comparable with that observed with the corresponding concentration of SNP (Fig. 6A). Vessel relaxation determined in response to S-NO oxyMb was independent of the presence of an intact endothelium, inhibited by 1H-(1,2,4)oxa-diazole(4,3-a)quinoxalin-1-one (Fig. 6A), and caused an increase in the tissue concentration of cGMP (Fig. 6B). In contrast, the C110A variant of oxyMb pre-treated with CYS-NO failed to both elicit vessel relaxation (Fig. 6A) and increase tissue concentrations of cGMP (Fig. 6B). Together, these findings suggest that S-NO oxyMb elicits vessel relaxation through the activation of soluble guanylyl cyclase.
Finally, we assessed whether intracellular S-NO oxyMb is formed in human VSMCs exposed to ⅐ NO (Fig. 7). The physiologic concentration of ⅐ NO ranges from 0.01 to 1 M in vascular (41,42) and myocardial tissues (43), with higher concentrations (ϳ13 Ϯ 4.3 M) detected in an animal model of allograft rejection (44) using an ⅐ NO-selective electrode. In these studies, the corresponding rates of ⅐ NO release, estimated from the time to reach maximal ⅐ NO concentration, were 2 and 140 nM⅐s Ϫ1 for coronary (42) and internal mammary arteries (43) stimulated with bradykinin and acetylcholine, respectively, and 325 nM⅐s Ϫ1 for cardiac allografts. We therefore exposed VSMCs to ⅐ NO doses in this pathophysiologic range. Thus, confluent VSMCs were exposed to ⅐ NO generated at ϳ1 and 60 nM⅐s Ϫ1 derived from the decomposition of added DeaNO (corresponding to final ⅐ NO concentrations of 0.15 and 15 M in the media) and cultured further for 5 or 60 min. These times were chosen to correspond to a time required for the complete decomposition of DeaNO (t1 ⁄2 ϭ 2 min, 37°C) and a time at which trans-nitrosation of oxyMb by intracellular RS-NO would be expected to reach equilibrium (see Fig. 4), respectively. Next, Mb was immunoprecipitated and assessed for ⅐ NO release induced by Cu 2ϩ (Fig. 7) and SDS-PAGE with Western blotting (Fig. 7C, inset). As expected, and independent of ⅐ NO pre-treatment, Mb was detected in VSMCs but not in PAECs (Fig. 7C, inset). Isolated Mb immunoprecipitates obtained from VSMCs exposed to 10 M DeaNO and incubated for 5 or 60 min consistently yielded ⅐ NO in the presence of Cu 2ϩ (Fig. 7, A and B); increased ⅐ NO release was detected in samples incubated for 1 h after DeaNO treatment. By contrast, immunoprecipitates from VSMCs exposed to vehicle (control) or PAECs pretreated with ⅐ NO did not yield measurable ⅐ NO (Fig. 7, C and  D). Also, Mb immunoprecipitates obtained from VSMCs treated with 100 nM DeaNO failed to yield measurable ⅐ NO independent of the incubation time (data not shown). Together, these data support the notion that human Mb can yield a stable protein RS-NO in VSMCs, at least when exposed to ⅐ NO produced at relatively high concentrations and rate of release from DeaNO. DISCUSSION There is growing evidence to support the idea that proteinbound forms of ⅐ NO act as stores of relaxing factor for VSMCs (36,45). For example, isolated rat aortic vessels incubated with ⅐ NO donors release a labile, relaxing, and soluble guanylyl cyclase-activating factor that is associated with protein thiols (46). However, identification of the specific protein(s) responsible for the ⅐ NO bioactivity-enhancing factor in the vessel wall has proven elusive. Here we demonstrate for the first time that recombinant human oxyMb yields an S-nitroso adduct. This is achieved by a relatively small (ϳ2-fold) molar excess of low molecular mass RS-NO and occurs at Cys 110 on human oxyMb via trans-nitrosation equilibria. In contrast, direct S-nitrosation by ⅐ NO under aerobic conditions is excluded as a mechanism due to the rapid rate of oxyMb-mediated oxidation of ⅐ NO. The yield of S-NO Mb observed via trans-nitrosation is dependent on the steric constraints of the donor RS-NO. Similar to S-NO ferric Mb (47), the ⅐ NO stored in the form of S-NO oxyMb can be released by Cu 2ϩ , as demonstrated directly using a ⅐ NO-sensitive electrode, and in the absence of added Cu 2ϩ , S-NO oxyMb can relax constricted blood vessels in vitro. In the vascular function studies described here, S-NO oxyMb was added to isolated vessels, whereas in vivo, any S-NO oxyMb formed would be expected to be present in VSMCs. Therefore, whether the vessel relaxing activity of S-NO oxyMb has biological significance and precisely what intracellular concentration of S-NO oxyMb accumulates in VSMCs under different conditions remain to be established.
Myoglobin is present in skeletal and cardiac muscle at relatively high concentration. For example, in the cytoplasm of cardiac myocytes, the Mb concentration is estimated to be ϳ350 M (5,48). More recently, Mb has been localized to human smooth muscle (6), although the precise concentration in this tissue is not known. Within the sarcoplasm of smooth or skeletal muscle, translational diffusion of oxyMb, balanced by a reverse flow of ferrous deoxyMb, is believed to support a flux of oxygen from the sarcolemma (closest to the capillary) to the mitochondria (48). The proportion of cardiac ferrous deoxyMb to oxyMb determined in situ is at least 10% under resting conditions (49). The high translational (50,51) and virtually unimpeded rotational diffusion (51) of Mb suggests that the protein is capable of moving rapidly from the sarcolemmal boundary through the cytoplasm and on to the mitochondrial target to support oxidative phosphorylation. In this process, Mb is responsible for local dissipation of oxygen near the capillary and establishing a shallow oxygen gradient within the sarcoplasm (52). Indeed, the high motility of Mb within muscle cells coupled with the high rate constant for oxidation of ⅐ NO by oxyMb is taken as evidence to support the notion that oxyMb regulates intracellular concentrations of ⅐ NO (48). For example, ⅐ NO partitioning in the mitochondrial membrane impacts upon mitochondrial respiration (9,53) through binding to the binuclear heme center of cytochrome c oxidase (54). The observation that vascular ⅐ NO catabolism decreases with a decreasing oxygen gradient extending away from the capillary (55) and subsequently increases again at the sites of mitochondria (56) nitrosation with (1) CYS-NO and (2) GS-NO that yield ϳ85% and 20% conversion to the S-nitrosated protein, respectively (Table I) (where oxyMb is concentrated) is taken as strong evidence to support the idea that oxyMb contributes significantly to intracellular ⅐ NO homeostasis.
Having established the potential for Mb to play an active role in regulating ⅐ NO bioavailability in smooth, cardiac, and skeletal muscle, it is pertinent to address the mechanisms of this process. Catabolism via oxyMb-mediated oxidation is a primary pathway that decreases ⅐ NO bioavailability, as evident from the modulated myocardial responses to ⅐ NO in mice lacking Mb as compared with those in wild-type animals (10). This pathway appears primarily important for processes taking place within the physiological lifetime of ⅐ NO. However, it is also worth considering the potential for Mb to prolong the lifetime (and hence, the bioavailability) of vaso-dilating ⅐ NO through formation of S-NO oxyMb. For example, the high degree of diffusion of intracellular Mb allows for the possibility of ⅐ NO transfer from the extracellular space to the cytosol within cardiac muscle cells or VSMCs through transnitrosation reactions at the cell membrane, in analogy to that reported for the cell surface protein disulfide isomerase (57). Furthermore, low molecular mass thiols are present in cells and thought to store bioactive ⅐ NO (21), although for steric reasons they are less stable than S-nitroso proteins (33,58,59). Protein S-nitrosation is emerging as a fundamental posttranslational protein modification that plays a key role in modulating ⅐ NO bioavailability. It is possible that proteins represent a biologic sink for bioactive ⅐ NO and that S-nitroso proteins represent a subsequent source of bioactive ⅐ NO that ultimately contributes to ⅐ NO bioavailability in human vessels. Our data support both ideas because Mb immunoprecipitates obtained from ⅐ NO pre-exposed VSMCs released ⅐ NO. These findings are consistent with a recent study by Janero et al. (60) indicating the presence of S-and N-nitroso adducts together with metal nitrosyls in organs of from rats exposed to glyceryl trinitrate. Notably, although not a major target for nitros(yl)ation, the vasculature contained S-and N-nitros(yl)ated products that were differentially localized in venous (90% RS-NO) and aortic (30% RS-NO) vessels (60).
The estimated values for k forward and k reverse for transnitrosation of oxyMb are similar to those of S-nitrosylation of bovine serum albumin (31), and S-NO serum albumin is formed in vivo (61). They are ϳ1000-fold greater than the k forward value for S-nitrosylation of ferrous deoxy-or oxyhemoglobin (7, 16, 62), yet S-NO Hb is also formed in low nanomolar concentrations in vivo (63). Therefore, it appears likely that S-NO oxyMb is formed at least in the myocardium, FIG. 6. Pre-constricted aortic segments relax upon treatment with increasing doses of S-NO oxyMb through a cGMP-dependent process. Excised rabbit aortic ring segments (A) with (filled symbols) and without (open symbols) an intact endothelium were suspended in a vessel bio-assay system, pre-constricted with phenylepherine, and treated with increasing doses of SNP (circles), S-NO oxyMb (squares), and C110A variant of human Mb (triangles) that had been pre-treated with CYS-NO in identical fashion to the wild-type protein. In some studies, 1H-(1,2,4)oxa-diazole(4,3-a)quinoxalin-1-one (final concentration, 10 M) was added before the vaso-stimulus (inverted triangles). Vessel dilation was monitored and expressed as a percentage of initial constriction force elicited at half the phenylepherine concentration to achieve maximal constriction. Also, rabbit aortic segments (B) were treated as described in A, except that all vessel segments were incubated with 200 M 3-isobutyl-1-methylxanthine before exposure to the various agents. Vessels were exposed to control (vehicle alone) or 10 M of either SNP, human S-NO oxyMb, or oxyC110A variant of human Mb (equilibrated with 2 mol of excess of CYS-NO) for 15 min and then immediately frozen in liquid nitrogen, homogenized, and assessed for cGMP content with a commercial kit. Data represent the mean Ϯ S.D. from six independent vessel segments. *, significantly different (p Ͻ 0.001) than that obtained with control aortic segments or aorta exposed to oxyC110A variant of human Mb that had been pre-treated with CYS-NO.

FIG. 7. Myoglobin immunoprecipitates obtained from human
VSMCs pre-treated with DeaNO contain Cu 2؉ -releasable ⅐ NO. Cultured human VSMCs or PAECs (3 ϫ 10 6 cells) were pre-equilibrated in HPSS and then treated with 10 M DeaNO or vehicle (control), harvested, and lysed, and where relevant, Mb was immunoprecipitated as described under "Experimental Procedures." Immunoprecipitates obtained from human VSMCs exposed to 10 M DeaNO and incubated for an additional (A) 5 or (B) 60 min, (C) VSMCs exposed to vehicle alone, or (D) immunoprecipitates obtained from PAECs exposed to 10 M DeaNO and incubated for an additional 60 min were subsequently tested for the presence of stored ⅐ NO by incubation with 100 M Cu 2ϩ , and released ⅐ NO was detected electrochemically. Inset in C shows a representative Western blot for Mb obtained from recombinant human Mb (lane 1), PAECs exposed to DeaNO (lane 2), human VSMCs exposed to DeaNO (lane 3), and human VSMCs exposed to vehicle alone (lane 4). Arrow indicates the region of the 20-kDa molecular mass marker. Data are representative of four independent experiments with different cell preparations.
where Mb concentrations approach millimolar levels. Our data indicate that it may also be formed in VSMCs, at least under conditions in which ⅐ NO is produced at a moderate rate (60 nM⅐s Ϫ1 ), keeping in mind that the physiologic rate of ⅐ NO production ranges from 2 to 140 nM⅐s Ϫ1 (42)(43)(44) and is dependent on both the type of vascular bed and the vasostimulus applied. The facile reaction between oxyhemoglobin and ⅐ NO is increasingly viewed as a competitive reaction that severely limits the physiological relevance of S-nitroation of Hb (64). By analogy, the high rate of ⅐ NO oxidation by oxyMb may also limit the physiologic relevance of S-NO Mb.
Overall, available data suggest that Mb plays a multifaceted role in ⅐ NO homeostasis in skeletal, cardiac, and possibly human smooth muscle (summarized in Scheme 1) via elimination of ⅐ NO via both oxidation and formation of Mb⅐NO complexes, as well as the potential for ⅐ NO preservation through Mb Snitrosation. Thus, the sarcoplasmic concentration of endothelium-derived relaxant factor in humans may well depend on the balance between the rates of oxyMb reaction with ⅐ NO, the recycling of ferric Mb to oxyMb, and the formation and subsequent decomposition of S-NO oxyMb. Importantly, the presence of a reactive cysteine sulfhydryl group in Mb isoforms obtained from rat heart (65,66) and tuna (67) indicates that regulation of vascular ⅐ NO through formation of S-NO oxyMb may be important in species other than humans. A better understanding of the mechanisms of ⅐ NO regulation by intracellular Mb may help elucidate the processes by which ⅐ NO chemistry interfaces with its biological function and warrants additional studies on elucidating the precise physiologic role of S-NO oxyMb in both the myocardium and vasculature in humans.