Heme Nitrosylation of Deoxyhemoglobin by S -Nitrosoglutathione Requires Copper*

NO reactions with hemoglobin (Hb) likely play a role in blood pressure regulation. For example, NO exchange between Hb and S -nitrosoglutathione (GSNO) has been reported in vitro . Here we examine the reaction between GSNO and deoxyHb (HbFe II ) in the presence of both Cu(I) (2,9-dimethyl-1, 10-phenanthroline (neocuproine)) and Cu(II) (diethylenetriamine- N , N , N (cid:1) , N (cid:2) , N (cid:2) -pentaace-tic acid) chelators using a copper-depleted Hb solution. Spectroscopic analysis of deoxyHb (HbFe II )/GSNO incubates shows prompt formation ( < 5 min) of (cid:1) 100% heme-nitrosylated Hb (HbFe II NO) in the absence of chelators, 46% in the presence of diethylenetriamine- N , N , N (cid:1) , N (cid:2) , N (cid:2) pentaacetic acid, and 25% in the presence of neocuproine. Negligible ( < 2%) HbFe II NO was detected when neocuproine was added to copper-depleted HbFe II / GSNO incubates. Thus, HbFe II NO formation via a mech-anism involving free NO generated by Cu(I) catalysis of GSNO breakdown is proposed. GSH is a source of reducing equivalents because extensive GSSG

Possible exchange of NO between thiols and hemoglobin (Hb) 1 in red blood cells (RBCs) has been the focus of intense interest recently (1,2). It has been suggested that GSNO or S-nitroso-L-cysteinyl could act as an NO ϩ donor to Cys␤ 93 of oxyhemoglobin (HbFe II O 2 ) in a trans-S-nitrosation reaction (3). However, we have shown that free NO must first be released from GSNO (4), the S-nitroso form of the dominant thiol in the RBC (5), or S-nitroso-L-cysteinyl 2 in a Cu(I)-catalyzed reaction. S-Nitrosation of Cys␤ 93 then occurs in a Cu(II)-catalyzed reaction (4) as reported also for Cys 34 of bovine serum albumin (6). For Hb to function as a blood pressure regulator in an O 2sensitive manner, release of NO from Cys␤ 93 of HbFe II is necessary (7). One possibility is that NO is delivered to tissues via S-nitrosation from Cys␤ 93 to GSH or another thiol (8) that promotes NO transport across the RBC.
HbSNO ϩ RSH º HbSH ϩ RSNO REACTION 1 Recently it has been suggested that delivery to tissues of all of the NO bound to Cys␤ 93 of Hb would result in extensive vasodilation, which would be fatal (9). Hence, it was proposed that most of the NO released from Cys␤ 93 is actually captured by HbFe II . Capture of NO released from GSNO is also possible and would compete with the exit of NO from the RBC. In fact, Spencer et al. (10) reported direct reductive cleavage of GSNO by HbFe II and capture of the released NO by another Fe II center (HbЈFe II ).
HbFe II ϩ GSNO ϩ H ϩ º HbFe III ϩ GSH ϩ NO HbЈFe II ϩ NO º HbЈFe II NO REACTIONS 2 AND 3 A trace amount of Cu(I) serves as a highly efficient catalyst of S-nitrosothiol breakdown (11). We have suggested that neocuproine, a Cu(I)-specific chelator, inhibits NO release from GSNO in solutions of HbFe II O 2 (4). Therefore, we considered it likely that neocuproine would also inhibit NO release from GSNO in solutions containing HbFe II . To distinguish between direct reductive cleavage of GSNO by HbFe II (Reaction 2) and Cu(I)-catalyzed release (Reaction 4), it is necessary to remove all trace copper or prevent its turnover via redox cycling using GSH (Reaction 5) or another donor in the Hb-containing solutions.
Cu(I) ϩ GSNO ϩ H ϩ º Cu(II) ϩ GSH ϩ NO Cu(II) ϩ GSH º Cu(I) ϩ 1 ⁄2 GSSG ϩ H ϩ REACTIONS 4 AND 5 Here we report the results of a detailed examination of HbFe II /GSNO incubates after 5 min in the presence of preferential chelators of Cu(I) (neocuproine) and Cu(II) (DTPA). Solutions of HbFe II that were not dialyzed and solutions that underwent exhaustive dialysis versus EDTA were used. Our direct spectroscopic and ESI-MS analyses reveal that NO re-lease from GSNO is Ͻ2% in the presence of neocuproine in HbFe II /GSNO solutions containing dialyzed Hb. HbFe II NO formation, and hence GSNO breakdown, is ϳ100% within 5 min in the absence of chelators. HbFe II NO formation is decreased by ϳ50 -75% in the presence of DTPA and neocuproine and in solutions containing dialyzed Hb without neocuproine. Because trace copper was found in all reagents by ICP-MS, these observations are consistent with Cu(I)-catalyzed release of NO (Reaction 4).
The source of the reducing equivalents for the prompt Cu(I)catalyzed reductive cleavage of GSNO is also of interest. Extensive GSSG was formed in HbFe II /GSNO incubates in the absence of metal chelators, indicating that GSH is the main source of reducing equivalents under these conditions (Reaction 5). We detected less GSSG than expected by ESI-MS in HbFe II /GSNO incubates containing DTPA, although we observed ϳ50% GSNO breakdown. This suggested HbFe II as a possible additional source of reducing equivalents to [Cu(II)-(DTPA)] 2Ϫ because related EDTA complexes were shown to be redox-active with Hb (12). Careful examination of the absorption (optical and FTIR) spectra revealed the presence of HbFe III . Thus, our results indicate that the direct reduction of GSNO by HbFe II is unlikely to play a role in NO transport in RBCs. Nonetheless, HbFe II may indirectly provide a source of electrons for the reductive cleavage of GSNO or other S-nitrosothiols in vivo.
The realization that commonly used Cu(I) (neocuproine) and Cu(II) chelators (EDTA and DTPA) may not always prevent copper turnover is an important consideration in deciphering mechanisms of S-nitrosothiol signaling and NO biochemistry in general. Although neocuproine, a tight binding Cu(I) chelator (K d ϭ 1.2 ϫ 10 Ϫ19 M 2 ) (13,14), is a better inhibitor than DTPA of GSNO breakdown (Reaction 4) in Hb-containing solutions, dialysis of the Hb samples and neocuproine addition were necessary to obtain negligible GSNO breakdown.
Finally, the prompt changes in HbFe II O 2 on incubation with GSNO are compared with those observed for the HbFe II incubates. The key results of this comparison are that in the absence of chelators, S-nitrosation of Cys␤ 93 of the oxy protein is extensive, and this competes with NO capture by the Fe II O 2 heme to form HbFe III and NO 3 Ϫ . In contrast, no S-nitrosation of the deoxy protein is detected, suggesting that all of the NO released from GSNO is captured by the Fe II heme.

Materials
Human hemoglobin A was obtained from Sigma and used without further purification. Nanopure water (specific resistance, 18 M⍀-cm) obtained from a Millipore Simplicity water purification system and treated with Chelex-100 (Sigma) to remove trace metal ions was used to prepare all H 2 O solutions. The reactions were carried out in 200 mM sodium phosphate buffer, pH 7.2 (NaPi), prepared from sodium phosphate salts (Fisher) in nanopure H 2 O. Stock solutions of 15 mM diethylenetriamine-N,N,NЈ,NЉ,NЉ-pentaacetic acid (DTPA; ICN) and 650 M 2,9-dimethyl-1,10-phenanthroline (neocuproine; Sigma) were prepared in NaPi. Stock solutions of 250 mM GSNO (Cayman) in NaPi were prepared just before use in a glove bag (Aldrich) under nitrogen, and the GSNO concentrations were determined spectrophotometrically (⑀ 333.5 nm ϭ 774 M Ϫ1 cm Ϫ1 ) (15). GSH and GSSG were obtained from Sigma, and N 2 and NO gases were from Praxair. NO was purged into a 10% KOH water solution before use (16).

Methods
Preparation and Optical Spectroscopy of Hb Samples-Typically 1 g of lyophilized metHb (HbFe III ) from the bottle was dissolved in 2 ml of NaPi. After 2 min of centrifugation at 12,000 rpm, the precipitate was discarded, and the dark red solution of HbFe III was stored at 4°C prior to use. An aliquot (10 l) of the Hb solution was pipetted onto a 13-mm CaF 2 window of a dismountable FTIR type cell (Harrick). The cell was immediately assembled using a 6-m Teflon spacer (Harrick) and placed in a custom-made bracket in a Beckman DU 650 UV-visible spectrophotometer. HbFe III concentrations were found to be 32 mM in heme assuming ⑀ 500 nm ϭ 10 mM Ϫ1 cm Ϫ1 /heme and ⑀ 630 nm ϭ 4.4 mM Ϫ1 cm Ϫ1 /heme (17,18). This was confirmed by diluting the samples 10 3fold, adding potassium ferricyanide and excess KCN (BDH chemicals), and reading the absorbance of the CN Ϫ adduct at 540 nm (⑀ 540 nm ϭ 11.0 mM Ϫ1 cm Ϫ1 /heme) in a 1-cm cuvette on a Beckman spectrophotometer.
HbFe II was prepared in the glove bag under N 2 by treating HbFe III with equimolar sodium dithionite (Fisher) (19) followed by desalting on a 1.6 ϫ 2.5-cm HiTrap Sephadex G-25 column (Amersham Biosciences), and the HbFe II concentration was calculated using ⑀ 555 nm ϭ 12.5 mM Ϫ1 cm Ϫ1 /heme (20). HbFe II O 2 was prepared from HbFe II by introducing a small volume of air into the sample using a syringe; for example, 720 l of air was added to 200 l of a 32 mM heme sample. A single addition of O 2 in slight excess (8.5 mol of O 2 /8 mol of heme) yielded fully oxygenated Hb as indicated by the Soret spectrum recorded following a 5-min equilibration. HbFe II NO and HbFe III NO were prepared from HbFe II and HbFe III on exposure to NO gas. The optical spectra were recorded using a scan time of 1200 nm/min. Use of the FTIR type cell to record the optical spectra of the products formed in the HbFe II /GSNO incubates allowed measurements to be made at close to physiological concentrations of Hb (2-4 mM) (21).
Preparation of Dialyzed Hb Samples-Approximately 500 mg of HbFe III was dissolved in 5 ml of 100 mM Na 2 EDTA (Sigma), pH 7.0, and allowed to stand for 30 min. This solution was dialyzed at 4°C versus 500 ml of 10 mM Na 2 EDTA, pH 7.0, which was replaced with fresh solution six times in 24 h. The dialysis was continued versus EDTA-free H 2 O, which was replaced with fresh solution 12 times in 48 h. After dialysis, HbFe III was lyophilized and dissolved in NaPi to a concentration of 30 mM heme.
ICP-MS Analysis-A PE Sciex Elan 6000 ICP-MS with a cross-flow nebulizer and a Scott type spray chamber was used to determine the amount of copper in the Hb and GSNO samples and in the buffers (see Table I). The RF power was 1000 W, and the argon flow was 0.85 liter/min, which gave the best sensitivity as determined by the recommended optimization procedure. The optimum lens voltage was determined by maximizing rhodium sensitivity, and the data were acquired in the pulse count mode (22). Stock Hb solutions in NaPi were added to 50 l of 30% H 2 O 2 (ACP Chemicals, Inc.) and 500 l of concentrated HNO 3 (OmniTrace Ultra high purity, EM Science) to give a final heme (iron) concentration of 4.3 mM, and the samples were ashed using a Bunsen burner. The residue was dissolved in 10 ml of 5% (v/v) HNO 3 , and ICP-MS analyses for copper and iron were performed. An internal standard of 9 nM (0.500 ppb) manganese prepared from a 1000 ppm manganese standard solution (ACP Chemicals, Inc.) was added to all of the solutions. The standard curves were prepared by diluting 1000 ppm copper and iron standard solutions (ACP Chemicals, Inc.) in 5% (v/v) HNO 3 and nanopure water to give 0 -8 M (0 -0.500 ppm) copper and 0 -9 M (0 -0.500 ppm) iron. All of the reported ICP-MS data are the results of at least triplicate experiments, and in all cases the standard deviations were Ͻ5%.
FTIR Analysis-Approximately 20 l of Hb (28 mM heme) in NaPi was added by syringe onto a 13-mm CaF 2 window in the glove bag under N 2 where necessary. The FTIR cell was immediately assembled with a 250-m Teflon spacer (Harrick), and the spectra were recorded at 25°C on a Nicolet Magna-IR 550 spectrometer with a MCT detector cooled to 77 K and purged with dry air from a Whatman FTIR Purge (model 75-52). All of the reported spectra are averages of 500 scans recorded in 5.52 min at a resolution of 2 cm Ϫ1 using a Happ-Genzel apodization with a velocity and aperture of 4.4303 cm/s and 2, respectively. Omnic (Nicolet) software was used for subtraction, base-line correction, smoothing, and Fourier transform self-deconvolution employing a half-width-at-half-height of 0.6 cm Ϫ1 and an enhancement (K factor) of 1. Subtraction of water vapor absorption from the spectra was performed by the method of Dong et al. (23,24).
ESI-MS Analysis-Stock Hb solutions (28 mM heme) in NaPi were diluted 10 Ϫ3 -fold with H 2 O to give ϳ0.5 g/l protein. The aliquots were infused into the ESI source of the mass spectrometer (ThermoFinnigan SSQ 7000) by flow injection from the high performance liquid chromatography (Agilent 1090) using a 100-l loop (but no column) at 50 l/min with 75% CH 3 CN (0.05% trifluoroacetic acid) as a mobile phase. Stock (250 mM) GSH, GSNO, and GSSG solutions in 200 mM NaPi, pH 7.2, were diluted 500-fold with H 2 O and 10-fold to 50 M with 75% CH 3 CN (0.05% trifluoroacetic acid), and their mass spectra were obtained as for the Hb solutions.
Multi-component Analysis of the Optical Spectra of Incubates-The absorbance at a given wavelength is the sum of the absorbances (A) of all species at that wavelength: where ⑀ x is the molar extinction of X at the selected wavelength, b is the pathlength (6 m), and [X] is the molar concentration of X. Assuming a limited number of oxidation and coordination states for iron of Hb, the spectra recorded for the incubates were mathematically disassembled into those of their components. Absorbances at the Soret maxima of the Hb species were used to generate n Beer's law expressions, from which the concentrations (

RESULTS
ICP-MS Analysis-Because the stability of S-nitrosothiols is highly dependent on the copper content, all of the solutions were examined by ICP-MS for trace copper. The results are summarized in Table I and reveal that 19 M copper was found in 5 mM Hb (20 mM heme) solutions. This decreased to 2 M copper following dialysis versus EDTA. NaPi buffer (200 mM) and the 250 mM stock solutions of GSH and its derivatives were found to contain ϳ1 M copper (Table I). These values differ from those reported previously using atomic absorption spectroscopy, where ϳ50 M copper was found in solutions containing 5 mM Hb, but Ͻ1 M copper was found in 5 mM dialyzed Hb as well as in NaPi (4). Because ICP-MS is more sensitive, more accurate, and did not show matrix effects (as verified using a manganese internal standard) compared with atomic absorption spectroscopy (25), the present copper analyses are considered more reliable.
Optical Absorption Spectra- Fig. 1a compares the spectra of HbFe II and HbFe III and their NO adducts in the Soret and visible regions. As expected, high spin Fe II and Fe III hemes exhibit Soret maxima at 430 and 405 nm, respectively, and visible bands at 556 nm (Fe II ) and 500, 540, and 580 nm (Fe III ). Soret maxima are observed at 418 nm (⑀ ϭ 130 mM Ϫ1 cm Ϫ1 ; Fe II NO) and 416 nm (⑀ ϭ 137 mM Ϫ1 cm Ϫ1 ; Fe III NO) for the heme-NO adducts of Hb. The corresponding visible bands are at 545 and 575 nm (Fe II NO) and 540 and 565 nm (Fe III NO). These values agree with those reported previously (19,26).
The products formed on mixing HbFe II and GSNO under anaerobic conditions were directly probed by comparing their spectra with those in Fig. 1a. Evidence for heme-iron nitrosylation is clearly seen in the Soret and visible bands of the HbFe II /GSNO incubates without chelators (Fig. 1b). The Soret maximum blue-shifted from 430 to ϳ418 nm within 5 min of mixing HbFe II and GSNO, and the visible region resembles that of HbFe II NO in Fig. 1a. Multi-component analysis of the 5-min spectrum reveals almost complete conversion of HbFe II to HbFe II NO (Table II). Thus, rapid release of NO from GSNO occurred in the HbFe II /GSNO sample.
Heme nitrosylation is also evident 5 min after mixing HbFe II and GSNO in the presence of 200 M DTPA (Fig. 1c). The extent of nitrosylation is less because the blue shifting of the Soret maximum is less, and the visible maximum of HbFe II at 556 nm is still evident, indicating that DTPA decreases the amount of NO released from GSNO. Multi-component analysis of the spectrum of the 5-min HbFe II /GSNO/DTPA incubate reveals the presence of 46% HbFe II NO, 44% HbFe II , 8.5% HbFe III , and 1% HbFe III NO ( Fig. 2a and Table II). When HbFe II and GSNO were mixed in the presence of neocuproine, the spectral changes revealed that significantly less heme nitrosylation occurred (Fig. 1d) than in the presence of DTPA, which was confirmed by multi-component analysis (Table II). Approximately 33% heme nitrosylation was found in the incubate of dialyzed HbFe II in the absence of chelators (Fig. 1e and Table II), but the addition of 150 M neocuproine essentially shut down NO release from GSNO (Fig. 1f) because 98% HbFe II was found in the 5-min incubate of dialyzed HbFe II /GSNO/ neocuproine (Table II).
To compare the heme species formed in HbFe II /GSNO incubates with those formed in HbFe II O 2 /GSNO incubates, the spectra of the latter were recorded. The Soret spectra of HbFe II O 2 /GSNO (Fig. 3, a and c) reveal that HbFe III is formed within 5 min, and multi-component analysis uncovered the   (Table II). Significantly less HbFe III formation is detected following mixing of HbFe II O 2 with GSNO in the presence of DTPA (Fig. 3b). This is confirmed by multi-component analysis of the 5-min spectrum, which reveals only 23 and 12% HbFe III formation in HbFe II O 2 / GSNO incubates containing DTPA and neocuproine, respectively. The formation of HbFe III likely occurs from the well characterized reaction of free NO and HbFe II O 2 (27,28):

REACTION 6
FTIR Spectra-FTIR spectroscopy is a valuable probe of protein thiols because the SH stretching vibration (SH) falls in a spectral window (ϳ2500 cm Ϫ1 ) with minimum H 2 O and protein absorption (29 -31). Although ϳ5 mM Hb is necessary to observe the weak IR (SH) absorption (29 -31), comparable Hb concentrations are found in RBCs (ϳ3 mM) (21). The FTIR spectrum of HbFe II in the (SH) region recorded in the absence and presence of GSNO and metal chelators (Fig. 4a, spectrum   1) exhibits the three (SH) peaks at 2576, 2563, and 2558 cm Ϫ1 assigned previously to Cys␤ 93 , Cys␤ 112 , and Cys␣ 104 , respectively (29,30).
The spectrum of HbFe II plus GSNO (Fig. 4a, spectrum 4) exhibits a peak for Cys␤ 93 (SH) at 2584 cm Ϫ1 , which is close to that of HbFe II NO (2585 cm Ϫ1 ). The (SH) band for Cys␤ 93 is also at 2584 cm Ϫ1 in the HbFe II /GSNO/DTPA spectrum (Fig.  4a, spectrum 5), but in the presence of both DTPA and neocuproine (Fig. 4a, spectrum 2) Cys␤ 93 possesses the same (SH) (2576 cm Ϫ1 ) as HbFe II alone (Fig. 4a, spectrum 1). These results corroborate those from the analysis of the visible spectra, which revealed that the major Hb species are HbFe II NO and HbFe II in the 5-min HbFe II /GSNO incubates in the presence of DTPA and neocuproine, respectively. Furthermore, no evidence for HbSNO formation is seen in Fig. 4a, because the (SH) absorption of Cys␤ 93 does not appear to lose intensity on exposure to GSNO (Fig. 4a, spectrum 1 versus spectra 2 and 3 and  spectrum 6 versus spectra 4 and 5).
Loss of Cys␤ 93 (SH) absorption is clearly evident in Fig. 4b  (spectrum 1). Thus, HbSNO formation does occur in the HbFe II O 2 /GSNO incubates in the absence of metal chelators as we reported previously (4). The addition of DTPA prevents loss of Cys␤ 93 (SH) intensity, but the spectrum is not identical to  that of HbFe II O 2 alone (Fig. 4b, spectrum 2 versus spectrum 4).
Direct monitoring of the heme shows 23% conversion of HbFe II O 2 to HbFe III within 5 min in the HbFe II O 2 /GSNO/ DTPA incubate (Fig. 3 and Table II). Thus, the FTIR data confirm that DTPA does not prevent release of NO from GSNO (Reaction 4), but it does inhibit S-nitrosation of Cys␤ 93 (Fig. 4b,  spectrum 2 versus spectrum 1). The FTIR spectrum of the HbFe II O 2 /GSNO/neocuproine incubate is essentially identical to that of HbFe II O 2 alone (Fig. 4b, spectrum 3 versus spectrum  4), which is consistent with the optical results where 88% HbFe II O 2 was found in the neocuproine incubate (Table II).
Mass Spectral Analysis-ESI-MS was used to probe HbSNO formation in the HbFe II /GSNO incubates (4). No peak corresponding to S-nitrosation of the ␤-subunit of HbFe II was observed under any conditions with or without metal chelators (Fig. 5a) or in the dialyzed HbFe II samples. On the other hand, S-nitrosation of the ␤-subunit was detected following incubation of HbFe II O 2 with GSNO in the absence of metal chelators (Fig. 5b) but not in their presence (data not shown). These MS results support those from FTIR spectroscopy in that HbSNO is formed only in the HbFe II O 2 /GSNO incubates in the absence of metal chelators (Fig. 4b, spectrum 1).
In the low m/z region of the ESI mass spectra, the low molecular weight products formed in the HbFe II /GSNO incubates can be monitored. In the absence of metal chelators, a weak GSNO (m/z 337) and an intense GSSG peak (m/z 613) are observed (Fig. 6b) that are not present in the spectrum of Hb alone (Fig. 6a). In mass spectrum of the HbFe II /GSNO/neocuproine incubate (Fig. 6d), GSNO is the dominant glutathione species, and little GSH is present, consistent with decreased GSNO breakdown in the presence of the Cu(I)-specific chelator. In Fig. 6c, the GSNO intensity is ϳ50% of that in the presence of neocuproine, indicating that DTPA is less effective in inhibiting the release of NO from GSNO in Hb solutions. However, a well defined GSSG peak (m/z 613) is not observed in Fig. 6c as expected if GSH is the major source of electrons for the reduc-tive cleavage of GSNO (Reactions 4 and 5). A relatively intense GSH peak persists in Fig. 6c, and also the optical spectra reveal that HbFe III is formed in the HbFe II /GSNO/DTPA incubate (Table II) Fig. 5. The peaks labeled ␣ 24 and ␣ 25 were caused by ␣-globin ϩ 24 H ϩ and ␣-globin ϩ 25 H ϩ , respectively. All of the peak intensities are relative to the (heme ϩ Na) ϩ peak at m/z 638 (100%). 21% HbFe III (Fig. 7). This can be compared with 19% HbFe III formation in the HbFe II /GSNO/DTPA incubates (Table II). DISCUSSION Spencer et al. (10) have reported that HbFe II can directly reduce GSNO and that the released NO is captured by additional Fe II heme. Because the release of NO from GSNO is known to be Cu(I)-catalyzed (32), we monitored HbFe II /GSNO incubates by optical and FTIR spectroscopies to probe changes occurring at the heme and Cys␤ 93 centers, respectively. In addition, the incubates were analyzed by ESI-MS to examine changes in the mass of the protein and to determine the low molecular weight species produced.
The optical spectra shown in Fig. 1 clearly reveal that trace Cu(I) is required for the release of NO from GSNO (Reaction 4). Essentially no HbFe II NO is formed in the HbFe II /GSNO incubate containing both copper-depleted Hb (i.e. dialyzed Hb in Table I) and neocuproine. The combined data in Tables I and II indicate that total inhibition of Cu(I) catalysis of GSNO breakdown requires removal of most of the copper as well as neocuproine addition to Hb-containing solutions. Thus, the results summarized in Table II are not consistent with direct reduction of GSNO by HbFe II (10) (Reaction 2) when equimolar heme and GSNO are present. Given the rapid (Ͻ5 min) Cu(I)-catalyzed release of NO from GSNO (Fig. 1b), it is likely that coppercatalyzed reductive cleavage of GSNO also occurs in vivo.
The next question that arises is, of course, what is the source of reducing equivalents to regenerate Cu(I) following electron transfer to GSNO to release NO (Reaction 4)? Possible electron donors to Cu(II) are GSH (Reaction 5) and HbFe II . With GSH as a donor, two molecules of NO would be released or two heme Fe II NO adducts would be formed per molecule of GSSG produced. The mass spectrum reveals extensive formation of GSSG in the HbFe II /GSNO incubates in the absence of chelators (Fig. 6b). Although a peak corresponding to its Na ϩ adduct occurs at m/z 635, the expected GSSG peak at m/z 613 is not clearly evident in HbFe II /GSNO incubates containing DTPA. Nonetheless, GSNO does release NO, as evidenced by the loss of GSNO peak intensity relative to the base peak at m/z 638 (Fig. 6, c versus d) and the formation of HbFe II NO ( Fig. 1c NO), which would decrease the concentration of GSSG produced to 2.84 mM. This is approximately one-third of the GSSG produced in HbFe II /GSNO incubates where a strong GSSG peak is observed at m/z 613. The lower than expected intensity of the GSSG (m/z 613) peak in Fig. 6c is attributed to ion suppression. The production of HbFe III in the HbFe II incubates in the presence but not the absence of DTPA (Table II) suggested that [Cu(II)(DTPA)] 2Ϫ may accept an electron from HbFe II , which was confirmed. The addition of authentic [Cu(II)(DTPA)] 2Ϫ to HbFe II led to oxidation of 21% of the heme (Fig. 7). Electron transfer between [Fe(II)(EDTA)] 2Ϫ and HbFe III has been reported previously (12).
No loss of (SH) intensity was detected by FTIR when HbFe II was treated with GSNO with or without metal chelators (Fig.  4a). This is consistent with the proposal that the allosteric transition of Hb controls Cys␤ 93 S-nitrosation (33). The FTIR results also show that Cu(I) is required for HbFe II NO formation because neocuproine inhibited blue shifting of the Cys␤ 93 (SH) peak (Fig. 4a, spectra 2 and 3) that accompanies heme nitrosylation (29). The reactions in the HbFe II /GSNO incubates reported here are shown in Scheme 1. Clearly the oxidation of HbFe II by [Cu(II)(DTPA)] 2ϩ is unlikely to be of importance in vivo, but the requirement of copper for NO release from GSNO shown in Scheme 1 could well have physiological relevance.
The key result that copper is also required for S-nitrosation of Cys␤ 93 of HbFe II O 2 (4) can be deduced from Figs. 4b and 5b. S-Nitrosation of Cys␤ 93 is observed only in the absence of metal chelators (Fig. 4b, spectrum 1, and Fig. 5b). Thus, we propose Scheme 2 for the reactions of NO with HbFe II O 2 . NO is targeted to Cys␤ 93 in the absence of chelators, but in the presence of DTPA, all of the NO released from GSNO is targeted to the Fe II O 2 heme (Figs. 3 and 4b) and converted to NO 3 Ϫ (Reaction 6). Scheme 2 predicts that efficient Cu(II)-catalyzed S-nitrosation of Cys␤ 93 will preserve the biological activity of NO by preventing its conversion to nonvasoactive NO 3 Ϫ . The capture of NO by HbFe II (Scheme 1) should not lead to loss of its vasoactive power as long as Fe II -NO adduct formation is reversible on a physiologically relevant time scale. It has been proposed that the Fe II heme centers of partially oxygenated Hb, as found in the RBC (1), have a much lower affinity for NO than fully deoxygenated HbFe II (9). Thus, copper control of NO reactivity with Hb, plus weaker binding of NO to HbFe II in vivo than in vitro, could explain why RBC Hb does not act a sink for most of the NO produced in the vascular system. A possible copper catalyst in vivo is copper-zinc superoxide dismutase, which is abundant in the RBC (34).
It is of interest to compare the extent of the prompt changes in HbFe II and HbFe II O 2 on incubation with GSNO. When 15 mM GSNO is incubated with 15 mM (heme) HbFe II in the absence of metal chelators, 99% is converted to HbFe II NO within 5 min, whereas only 46% of HbFe II O 2 is converted to HbFe III under the same conditions. This is easy to understand when we consider that ϳ7.5 mM NO released from GSNO was targeted to Cys␤ 93 in HbFe II O 2 but not in HbFe II (Figs. 4 and 5). In the DTPA incubates 44% HbFe II versus 77% HbFe II O 2 remain, and in neocuproine incubates 75% HbFe II versus 88% HbFe II O 2 remain after the same 5-min period. Clearly, the increased stability of the Hb reactants in the presence of metal chelators is due to inhibition of GSNO breakdown, but the lower consumption of HbFe II O 2 relative to HbFe II cannot be attributed to NO trapping by Cys␤ 93 in the former because this does not occur in the presence of chelators (Fig. 4b). More efficient NO trapping by HbFe II (35) than by HbFe II O 2 (k ϭ 3-5 ϫ 10 7 M Ϫ1 s Ϫ1 for Reaction 6) (36) would give rise to the higher HbFe II consumption in the incubates. Also, the reaction of NO with any free O 2 in the HbFe II O 2 incubate would decrease the amount of HbFe II O 2 consumed.
In this work we also examined the spectra of the HbFe II / GSNO incubates after 1 h. The results obtained indicate that the prompt products undergo further reaction over longer times. For example, the Soret maximum after 1 h of HbFe II / GSNO incubate still shows a maximum at 418 nm but with decreased intensity (data not shown). Because the spectrum of authentic HbFe II NO in the absence of any glutathione-derived species is stable for Ͼ1 h, the HbFe II NO initially formed must react with some reagent in the incubate. 2 The Soret maximum of the HbFe II /GSNO/DTPA incubate after 1 h (data not shown) is blue-shifted (409 nm) from that at 5 min (411 nm) (Fig. 1c). We initially attributed this blue shift to increased HbFe II NO formation caused by NO-driven reduction of HbFe III using reducing equivalents from GSH. However, the addition of GSH to authentic HbFe III NO also gave rise to a Soret maximum at 411 nm and not the expected HbFe II NO peak at 418 nm. The biological relevance of these slow reactions is questionable because, for example, any HbFe III formed in the RBC would be rapidly converted to HbFe II by methemoglobin reductase (37). CONCLUSIONS The results reported here provide insight into the mechanism of heme nitrosylation following mixing of HbFe II with GSNO. The data presented are inconsistent with direct reduction of GSNO by HbFe II but are consistent with catalysis of GSNO breakdown by trace copper and free NO generation in the reactions. Because HbFe II is stable in the presence of GSNO when free copper is rigorously excluded from the system, direct reduction of GSNO by HbFe II is unlikely to be of biological significance. Trace copper also controls targeting of NO released from GSNO to the Cys␤ 93 and Fe II O 2 centers of HbFe II O 2 . Thus, we conclude that copper catalysis of S-nitrosation and S-denitrosation plays a key role in preserving the vasoactivity of NO in the RBC.