Human glutathione transferase P1-1 and nitric oxide carriers; a new role for an old enzyme.

S-Nitrosoglutathione and the dinitrosyl-diglutathionyl iron complex are involved in the storage and transport of NO in biological systems. Their interactions with the human glutathione transferase P1-1 may reveal an additional physiological role for this enzyme. In the absence of GSH, S-nitrosoglutathione causes rapid and stable S-nitrosylation of both the Cys(47) and Cys(101) residues. Ion spray ionization-mass spectrometry ruled out the possibility of S-glutathionylation and confirms the occurrence of a poly-S-nitrosylation in GST P1-1. S-Nitrosylation of Cys(47) lowers the affinity 10-fold for GSH, but this negative effect is minimized by a half-site reactivity mechanism that protects one Cys(47)/dimer from nitrosylation. Thus, glutathione transferase P1-1, retaining most of its original activity, may act as a NO carrier protein when GSH depletion occurs in the cell. The dinitrosyl-diglutathionyl iron complex, which is formed by S-nitrosoglutathione decomposition in the presence of physiological concentrations of GSH and traces of ferrous ions, binds with extraordinary affinity to one active site of this dimeric enzyme (K(i) < 10(-12) m) and triggers negative cooperativity in the vacant subunit (K(i) = 10(-9) m). The complex bound to the enzyme is stable for hours, whereas in the free form and at low concentrations, its life time is only a few minutes. ESR and molecular modeling studies provide a reasonable explanation of this strong interaction, suggesting that Tyr(7) and enzyme-bound GSH could be involved in the coordination of the iron atom. All of the observed findings suggest that glutathione transferase P1-1, by means of an intersubunit communication, may act as a NO carrier under different cellular conditions while maintaining its well known detoxificating activity toward dangerous compounds.

Glutathione transferases (EC 2.5.1.18) (GSTs) 1 are a superfamily of enzymes involved in the detoxication of the cell against toxic and carcinogenic compounds (1). The human cytosolic GSTs are dimeric proteins grouped into at least eight gene-independent classes (Alpha, Kappa, Mu, Omega, Pi, Sigma, Theta, and Zeta) on the basis of their amino acid sequence, substrate specificity, and immunological properties (2)(3)(4)(5)(6)(7)(8). Their three-dimensional structures do not differ significantly despite low sequence homology (9 -12). Each subunit contains a very similar binding site for GSH (G-site) and a second one for the hydrophobic co-substrate (H-site). Slight structural differences at the H-site confers a certain degree of substrate selectivity. For a more detailed review on the molecular properties of GSTs see Ref. 13.
The physiological role of GSTs does not appear unequivocal. A well known function of GSTs is to promote the conjugation of the sulfur atom of glutathione to an electrophilic center of endogenous and exogenous toxic compounds, thereby increasing their solubility and excretion (1). The Alpha class also displays a peroxidase activity with organic peroxides (14). Because of the considerable level of expression of GSTs in many tissues and their property to bind a number of large hydrophobic compounds, a possible role of "ligandins" (ligand carriers) has been also proposed (15). Recent advances suggest a role for GST P1-1 in the regulation of Jun kinase protein (a stressactivated protein that phosphorylates c-Jun) (16) and as a "tissue" transglutaminase-specific substrate in neural cells committed to apoptosis (17). It has also been reported that certain GSTs (class Theta) may inhibit the proapoptotic action of Bax (18), and the most recently discovered class Omega, GST O1-1, modulates calcium channels, thus protecting mammalian cells from apoptosis induced by Ca 2ϩ mobilization (19). Thus, GSTs can be considered as multi-functional enzymes devoted to various aspects of cell defense.
Notwithstanding this versatility, GSTs have never been considered to be involved in the complex mechanisms of detoxification or storage of nitric oxide. NO is formed in organisms from endogenous or exogenous sources and triggers a number of cellular responses, including the regulation of blood pressure, the relaxation of smooth muscle, and the modulation of immunity (20,21). In the central nervous system, NO is a neuronal messenger and is possibly responsible for the devel-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (22)(23)(24)(25). NO has a very short life time in the cell, and its stabilization and transport are promoted by specific NO carriers of low molecular mass, which are formed by the interaction of nitric oxide with ferrous ions (dinitrosyl-iron complexes (DNIC)) (26 -28) or with GSH (Snitrosoglutathione (GSNO)) (29 -32). DNIC bind to proteins like albumin forming paramagnetic NO-Fe-protein complexes that can be detected by ESR spectroscopy (33). Similar complexes, known as 2.03 complexes, because of the g value of their characteristic ESR spectra, have been observed in cells and tissues exposed to NO, but nothing is known about the proteins involved (34 -37). Recently, kinetic studies showed that both GSNO and dinitrosyl-diglutathionyl-iron complex (DNDGIC) are competitive inhibitors of GSTs, and they have been proposed as activity modulators of these enzymes (38,39). Starting from these preliminary findings, we have now investigated the interaction of these compounds with GST P1-1, using ESR spectroscopy, site-directed mutagenesis, mass spectrometry, and molecular modeling. Our results reveal a surprising and sophisticated mode of interaction suggestive of a new role for this enzyme in the cell as a NO reservoir or NO scavenger protein.

EXPERIMENTAL PROCEDURES
Expression Plasmids and Site-directed Mutagenesis-The plasmid pGST-1, producing large amounts of recombinant wild-type GST P1-1 in the cytoplasm of Escherichia coli, has been described previously (40). The expression plasmid p18Seq-1, reported previously (41), was used to generate the single-stranded DNA template to be used for site-directed mutagenesis of Cys 47 and Cys 101 residues, according to the method described by Kunkel et al. (42) with minor modifications.
Protein Expression and Purification-Native and mutant GST P1-1 enzymes were produced as described previously (40,41). Briefly, TOP 10 E. coli cells harboring plasmid pGST-1 or plasmid expressing Cys 47 and Cys 101 mutant enzymes (pGST-A47 and pGST-A101) were grown in LB medium containing 100 g/ml ampicillin and 50 g/ml streptomycin. The synthesis of GST was induced by the addition of 0.2 mM isopropyl-1-thio-␤-galactopyranoside when the absorbance at 600 nm was 0.5. Eighteen hours after induction, the cells were harvested by centrifugation and lysed as described previously (40). Native and GST mutant enzymes were purified by affinity chromatography on immobilized glutathione (43). After affinity purification, the native and the mutant enzymes (C47A and C101A) were homogeneous as judged by SDS-polyacrylamide gel electrophoresis (44). Protein concentration was determined by the method of Lowry et al. (45).
Kinetic Studies-The enzymatic activities were determined spectrophotometrically at 25°C with 1-chloro-2,4-dinitrobenzene (CDNB), as co-substrate, following the product formation at 340 nm, ⑀ ϭ 9600 M Ϫ1 cm Ϫ1 (46). Spectrophotometric measurements were performed in a double beam Uvicon 940 spectrophotometer (Kontron Instruments) equipped with a thermostatted cuvette compartment. Initial rates were measured at 0.1-s intervals for a total period of 12 s after a lag time of 5 s. Enzymatic rates were corrected for the spontaneous reaction.
Apparent kinetic parameters k cat and K m CDNB were determined in 0.1 M potassium phosphate buffer, pH 6.5, and 0.1 mM EDTA, containing a fixed concentration of GSH (10 mM) and variable concentrations of CDNB (0.1-2 mM). The collected data were fitted to the Michaelis-Menten equation by nonlinear regression analysis using the GraphPad Prism (GraphPad Software, San Diego, CA). In a more accurate analysis, kinetic data were also fitted to Equation 1, which considers two distinct enzyme populations characterized by different K m values for GSH.
v ϭ V max 1 ͓S͔ The apparent K m GSH was also determined at fixed CDNB concentration (1 mM) and variable GSH concentrations (from 0.02 to 20 mM). Kinetic parameters reported in this paper represent the mean of at least three different experimental data sets.
Preparation of GSNO-GSNO was prepared as described previously (47). Briefly, a few drops of HCl were added to a solution containing equimolar amounts of GSH and sodium nitrite until pH 1.5 was reached. After standing for 5 min at room temperature, the red GSNO was neutralized with NaOH. GSNO displays an absorption maximum of 750 M Ϫ1 cm Ϫ1 at 332 nm and appears to be stable for a few days at room temperature. Appropriate aliquots of freshly synthesized compound were stored at Ϫ80°C and used when necessary, after checking their absorbance at 332 nm.
Synthesis of DNDGIC-Dinitrosyl-diglutathionyl-iron complex was synthesized according to the following procedure. Suitable amounts of ferrous ions (FeSO 4 , ranging from 10 to 50 M) were added to a mixture containing 20 mM GSH and 2 mM GS-NO in 0.1 M phosphate buffer, pH 7.4, and 25°C. The synthesis of the complex was completed in the first 15-20 min and gives an extinction coefficient of 3000 M Ϫ1 cm Ϫ1 at 403 nm. Dinitrosyl-dicysteinyl-iron complex, under slightly different conditions, displays an identical extinction coefficient at 395 nm (48).
Interaction of GSNO with GST P1-1-GST P1-1 or Cys mutant enzymes (ranging from 0.03 to 1 mg/ml) were incubated with 2 mM GSNO in 0.1 M phosphate buffer, pH 7.4, at 25°C. At various times aliquots were taken and assayed for enzymatic activity with CDNB as co-substrate. In a different set of experiments, 20 mM GSH was added to the mixture containing the above concentrations of GSNO and protein with or without 1 mM EDTA.
UV Difference Spectra-Difference spectra were performed with a double beam Uvicon 940 spectrophotometer (Kontron Instruments) equipped with a thermostatted cuvette compartment. Spectral data of free and bound DNDGIC complex were recorded in the range 290 -500 nm at 25°C at pH 7.4 using quartz cuvettes (1-cm light path).
Interaction of DNDGIC with GST P1-1-Inhibition of GST P1-1 by DNDGIC was measured by incubating variable amounts of DNDGIC (from 0.5 to 9 M) with 0.1 mg/ml (4.4 M of active sites) of GST P1-1. After 2 min 10-l aliquots (final concentration of GST P1-1 is 44 nM active sites) were assayed for activity at pH 6.5 (final volume, 1 ml) in the presence of 10 mM GSH and 1 mM CDNB. Thus, the final concentration of DNDGIC in the activity sample ranged from 5 to 90 nM. In a second experimental set, GST P1-1 (44 nM of active sites) was incubated at pH 7.4 with variable amounts of DNDGIC (from 0.1 to 22 M). After 2 min the pH was adjusted to 6.5, the GSH concentration was brought to a final concentration of 10 mM, and 1 mM CDNB was then added for activity measurement.
Electron Spin Resonance-ESR spectra were recorded with an ESP300 instrument (Bruker, Karlsruhe, Germany) operating at X-band frequency. Low temperature measurements (100 K) were made using a standard TE 102 -type cavity with samples in 3-mm-inner diameter quartz tubes, whereas experiments performed at room temperature were made using a high sensitivity TM 110 -type cavity with samples contained in 1.10-mm-inner diameter capillary tubes. ESR spectra were recorded at 20 mW microwave power and 1 G modulation; for resolution of the hyperfine structure some spectra were recorded using 0.2 G modulation.
Mass Spectrometry-GST P1-1 (final concentration, 2.0 mg/ml) was incubated with 2 mM GSNO, under the conditions reported above, for 15 min. Thereafter, it was passed through a Sephadex G-25 column, equilibrated with distilled water, containing 0.1 mM EDTA, to remove any excess of reagent and used soon for mass spectrometry studies. A sample of unmodified human GST P1-1 was equilibrated in the same manner, before mass spectroscopic analysis. The samples were diluted with 99% acetonitrile to give the final solution of acetonitrile/water 50% ϩ formic acid 0.2% (v/v). Mass spectrometry spectra were collected in continuous flow mode by connecting the built-in infusion pump directly to the ion spray probe. Molecular masses of both unmodified and modified GST P1-1 enzymes were determined using a Triple Quadrupole liquid chromatography/mass spectrometry/mass spectrometry mass spectrometer Applied Biosystem Sciex API 365 (Concord, Canada). The data were acquired and processed using Mass Chrom 1.2 (Applied Biosystem Sciex, Concord, Ontario, Canada), including BioMultiView 1.3.1 for mathematical transformation of ion spray spectra to true mass scale.
Molecular Modeling-Molecular modeling was performed on a Silicon Graphics O 2 work station using the software package O (49). The model was based on the 1.9 Å resolution crystal structure of human class Pi GST P1-1 in complex with GSH (50). A model of DNGIC was constructed assuming tetrahedral geometry and metal-ligand distances based on measurements from small molecule crystal structures (51), with the topology and parameters files for the DNGIC complex obtained using XPLO2D and refined using crystallography and NMR system (52). It is not clear whether the complex adopts a tetrahedral or octahedral (with two water ligands) configuration when bound to the enzyme. However, either configuration will fit into the active site, and the same conclusions can be drawn from either model. The DNGIC model was docked into the active site of the enzyme assuming that the GSH ligand of DNGIC occupied the same site as the GSH ligand in the crystal structure of the enzyme-GSH complex. It was immediately noted that the hydroxyl moiety of Tyr 7 was in covalent bonding distance of the iron atom. A model of the complex covalently bound via Tyr 7 was constructed. The resultant model was energy minimized with CNS (52). Initially harmonic restraints of 10 kcal/mol were applied to the protein atoms, but subsequently the restraints were removed in the final stages of minimization. The final energy minimized model exhibited no significant structural differences with the crystal structure of the enzyme. Graphic representation was produced by the computer program MOL-SCRIPT (53).

RESULTS AND DISCUSSION
Interaction of GSNO with GST P1-1 in the Absence of GSH-GSNO is a competitive inhibitor for GST P1-1 (38,39), but in the absence of GSH this compound also causes a time-dependent partial inactivation of the enzyme. After 10 min of incubation, the activity is reduced to about 70% of its initial value and remains unchanged even after 1 h of incubation (Fig. 1). The presence of 1 mM EDTA in the incubation mixture does not affect significantly the extent and the rate of inactivation. Treatment of the inactivated enzyme with 100 mM 1,4-dithiothreitol restores completely the original activity. The modified enzyme does not follow strictly Michaelian kinetics and shows an apparent K m value for GSH of 0.45 mM, which is three times higher than in the native enzyme and accounts completely for the observed loss of activity, using the standard GSH concentration of 1 mM (46). Cys 47 and Cys 101 , the most reactive among the four cysteines of each subunit (41,54), are possible targets for the reversible modification by GSNO. Titration of the protein sulfhydryls after 1 h of GSNO (2 mM, pH 7.4) treatment shows that three sulfhydryl groups/dimer have been lost. The UV spectrum of the modified enzyme (Fig. 1, inset) shows a maximum at 332 nm, which corresponds to that found for the nitroso-thiol species (47). The amount of the protein CysNO, calculated on the basis of an extinction value of 750 M Ϫ1 cm Ϫ1 at 332 nm (47), is 1.6 mol/subunit. When GSNO is reacted for 1 h with the C47S mutant enzyme, no inactivation was observed, and two cysteines/dimer are nitrosylated. Conversely, the C101A mutant only yields 1 mol of CysNO/dimer. Thus, it appears that Cys 47 and Cys 101 act as primary targets for a specific S-nitrosylation by GSNO, but only the modification of Cys 47 decreases the enzyme activity. Furthermore, when this residue is nitrosylated in one subunit, the second in the adjacent subunit becomes less accessible. In fact, the kinetic data of the modified enzyme can be fitted more convincingly to a kinetic scheme involving two enzyme populations with different K m values for GSH (see Equation 1). The best fit gives V max 1 ϭ V max 2 and K m values for GSH of 0.2 mM and 1.2 mM, respectively. This result is also consistent with two subunits with different affinities in the same dimeric population. The observed half-site reactivity of Cys 47 compensates for the lowering of the affinity for GSH (like that observed in the modified subunit) and could reflect a self-preservation role of enzyme activity. This behavior is not unusual for GST P1-1 and is reminiscent of the nonequivalent reactivity of the two subunits of the horse GST P1-1 in their reaction with sulfhydryl reagents (55).
Mass Spectrometric Analysis of NO-modified GST P1-1-To gain direct evidence of S-nitrosylation of human GST P1-1 we have carried out ion spray ionization-mass spectrometry of the modified GST P1-1, and the results are shown in Fig. 2. Ion spray mass spectra of the unmodified protein (Fig. 2a) showed two main peaks corresponding to the molecular masses of 23,226 and 23,356 Da, respectively. The former molecular mass corresponds to that calculated from its cDNA, whereas the latter is due to the failed removal of N-terminal methionine, and both are consistent with data reported previously (40,56). In the Fig. 2b, we observe two series of peaks with mass increases of 30 Ϯ 1 Da, in addition to the unmodified proteins (23,226 and 23,356 Da, respectively), suggesting nitrosylation of GST P1-1. In particular, it appears that the first species (23,226 Da) exhibits molecular masses of M ϩ ϩ30 and M ϩ ϩ 60, respectively; the second species (23,356 Da) shows the same mass increases of ϩ30 Ϯ 1 and ϩ60 Ϯ 1 Da and, to lesser extent, a third peak with a mass increase of ϩ 90 Ϯ 1 Da. Thus, in accordance with the data reported above, these spectra demonstrate the simultaneous presence of, at least, the mono-and di-nitrosylated subunits, whereas the expected mass of about M ϩ ϩ305, for glutathionylated GST P1-1, was not found. The presence of a third peak with a mass increase of ϩ 90 Ϯ 1 Da in the species with N-terminal methionine requires further investigation because it may be due also to a methionine oxidation. Further mass spectrometry studies on Cys mutant enzymes will be made to unravel the mechanism of GST P1-1 modification.
Interaction of GSNO with GST P1-1 in the Presence of GSH-It is well known that saturating amounts of GSH protect from the chemical modifications caused by thiol reagents. Cys 47 , located at the end of the mobile helix-2, is exposed to the solvent only in the apoenzyme, whereas it is buried in the holoenzyme (55,57,58). When GSNO (2 mM) is incubated with 0.1 mg/ml GST P1-1 in the presence of 20 mM GSH and 1 mM EDTA, no inactivation can be observed within 1 h of incubation (Fig. 3). However, in the absence of EDTA, a fast inactivation occurs that leaves 50% of the original activity after 10 min of incubation (Fig. 3). Similar half-site inhibition is observed at lower GST concentrations, but the extent of this phenomenon is reduced by increasing the enzyme concentration. For example, at 1 mg/ml, it does not exceed 5% (Fig. 3). The cause of this inhibition will be clarified below and likely depends on metal traces present in the buffers.

Dinitrosyl-Diglutathionyl Iron Complex Is Formed in the GSH-GSNO System with Ferrous Ions-When GSNO (2 mM)
and GSH (20 mM) are incubated at 25°C in the presence of substoichiometric amounts of ferrous ions (10 -50 M), a fast reaction occurs, signaled by the appearance of an UV absorption peak, centered at about 400 nm, which is suggestive of the formation of the dinitrosyl-diglutathionyl iron complex (48) (Fig. 4). Kinetics of this process cannot be described satisfactorily by a simple mono-exponential function but fit better to a two phase exponential function (Fig. 5). ESR spectra confirm the presence of a paramagnetic species with ESR signals at Ϫ70°C and 25°C, similar to those reported for other DNICs (26 -33) (Fig. 6A). The appearance of the paramagnetic complex parallels the UV spectral increase at 403 nm (Fig. 5). After 20 min, 90% of the added ferrous ions is present as DNDGIC, and this complex is stabilized up to 180 min. After this time, a slow decomposition occurs, characterized by a t1 ⁄2 of about 90 min. Diluted DNDGIC solutions, synthesized by the conventional procedures and without a NO-generating system, exhibit very short half-lives (about 1 min at micromolar concentrations) (33). The protocol described here provides a convenient proce-dure for a quantitative and stabilized preparation of DNDGIC, based on slow release of NO by GSNO in the presence of physiological concentrations of GSH and substoichiometric iron. The apparent stability of DNDGIC is likely due to a steady-state equilibrium coming from simultaneous decomposition and resynthesis of the complex.
Binding of DNDGIC to GST P1-1-DNDGIC has been described as a competitive inhibitor of GST P1-1 toward GSH (39), but a more careful investigation reveals different and unexpected properties of interaction with this enzyme. Substoichiometric amounts of DNDGIC yield a degree of inhibition that corresponds exactly to an interaction of 1:1 active siteinhibitor. 50% of the original activity is reached when the molar amount of DNDGIC corresponds to 50% of the active sites (Fig. 7a). Increasing concentrations of DNDGIC (up to 4-fold excess over free active sites) does not produce any further  Fig. 3. OE, the same experiment in the presence of 1 mg/ml of GST P1-1, followed at 408 nm. b, the same experiments as in a followed by means of ESR measurements. significant inhibition. This behavior can be explained by a very strong interaction of the complex with one of the two binding sites of the dimeric enzyme. By assuming that DNDGIC causes competitive inhibition (39), a quantitative evaluation of the dissociation constant for the DNDGIC-GST complex (K i ) can be made observing that the 50% inactivated enzyme maintains its half activity even at the high dilution of 10 Ϫ9 M and in the presence of 10 mM GSH. This means that K i must be Ͻ1.5 ϫ 10 Ϫ12 M, which represents the strongest interaction of a small compound with this enzyme. The second G-site binds DNDGIC only at very much higher DNDGIC concentrations (Fig. 7b). This second weaker interaction is characterized by a K i of 2.6 ϫ 10 Ϫ9 M. Considering the homodimeric structure of this enzyme, this behavior is likely due to a strong negative cooperativity triggered by the interaction of DNDGIC with the first subunit that dramatically lowers the affinity of the vacant subunit. Structural communication between subunits has been previously reported for GST P1-1. It is well known that this enzyme has a latent cooperativity that becomes evident at extreme temperatures or by point mutation of crucial residues (41, 59 -61). The observed negative cooperativity mainly involves the affinity for DNDGIC. In fact, the K m value for GSH of the vacant subunit (calculated in the half-inactivated enzyme) appears only slightly higher (0.35 mM) than that shown by the native enzyme (0.15 mM), whereas the K m value for CDNB is unchanged. Even in this case, as observed for Cys 47 nitrosylation, this behavior seems to be a mechanism by which the enzyme protects itself against complete inactivation because of DNDGIC binding.
The preliminary finding of a particular inhibition obtained by reacting the enzyme with GSNO and GSH without the addition of exogenous iron (Fig. 3) can be now rationalized. In fact, about 2 M of DNDGIC are formed using the spurious contaminating iron present in the buffer. This low DNDGIC concentration (which will be confirmed by ESR analysis) is enough to cause 50% inhibition when reacted with 0.1 mg/ml GST P1-1 (4.4 nmol/ml active site) or with lower enzyme concentrations but not with 1 mg/ml enzyme concentration.
Competition with Serum Albumin-DNDGIC binds to serum albumin with high affinity, and this protein could be a reservoir of S-nitrosylating species (33). Experiments of competition of this protein with GST P1-1 may give an estimation of the relative affinity of the two proteins against DNDGIC. We found that even 50 mg/ml of serum albumin does not affect at all the inhibition of 0.1 mg/ml of GST P1-1 because of the interaction with 2 M of DNDGIC, thus indicating that GST P1-1 has at least 500 times more affinity than albumin for DNDGIC.
UV Spectral and Chemical Properties of DNIC Bound to GST P1-1-The UV spectrum of DNDGIC bound to the enzyme in the presence of 20 mM GSH is similar but not identical to the free species showing a slight red shift of the maximum at 408 nm and with a less pronounced descending limb below 400 nm (Fig. 4). The kinetics of DNDGIC formation is identical to that obtained in the absence of enzyme, so GST P1-1 does not appear directly involved in the complex formation (Fig. 5). The complex is tightly bound to the enzyme and remains associated to the enzyme even after a Sephadex G25 chromatography. The UV spectrum, observed after removal of the excess of GSH, is . After 2 min the pH was adjusted to 6.5, the GSH concentration was brought to a final concentration of 10 mM, and 1 mM CDNB was then added for activity measurement. broad (Fig. 4), but the subsequent addition of 20 mM GSH restores the descending limb below 400 nm (spectrum not shown). Thus, at least three similar but not identical DNDGIC forms can be distinguished: the free mononuclear complex, the dinitrosyl-iron complex bound to the enzyme in the presence of an excess of GSH, and a third species obtained after removal of the excess of GSH. Notably, the bound complex is stable for hours, and only 24 h after Sephadex chromatography it spontaneously decomposes with the full recovery of the enzyme activity. This represents a remarkable stabilization of the bound complex when compared with its fast decomposition in the free form (33).
It has been observed that the DNDGIC analogue dinitrosyldicysteinyl iron complex binds tightly to bovine serum albumin, and one or both cysteines involved in the iron complex may be replaced by protein sulfhydryls or imidazoles (33). Titration of the amount of GSH involved in the DNDGIC-GST P1-1 complex may be informative if a similar event also occurs in GST P1-1. After binding of 44 M DNDGIC to 88 M GST P1-1 (final volume, 1 ml), the 50% inactivated enzyme was passed through a G-25 Sephadex column to remove the excess of GSH. The enzyme displays the same half-activity, and the absorbance at 403 nm confirms that about 40 nmol of the complex are still bound to the enzyme. By adding 10 mM potassium cyanide, the spectral band of DNDGIC at 403 nm disappears within a few minutes, and the enzyme recovers the original full activity. In fact, CN Ϫ ions bind tightly to ferrous ions to give the extremely stable [Fe(CN) 6 ] Ϫ4 complex (K inst ϭ 10 Ϫ35 at 291 K). After cyanide treatment, the amount of free GSH has been titrated by adding 1 mM chloro-dinitrobenzene, a good co-substrate for GST, and by measuring the amount of GS-DNB product formed (⑀ ϭ 9,600 M Ϫ1 cm Ϫ1 at 340 nm). The result is 40 nmol of GSH detected, which corresponds exactly to 50% of that expected if the bound DNDGIC would retain the same original composition, thus indicating that one protein group replaced one GSH molecule as iron ligand. Note that no GSH can be titrated without previous potassium cyanide treatment, indicating complete removal of the free GSH by gel chromatography. Both C47A and C101A mutant enzymes give a very similar affinity and inhibition pattern by DNDGIC as the native enzyme, so the Cys 47 and Cys 101 sulfhydryls are not involved in the stabilization of the iron complex in the G-site.
ESR Studies-At 77 K, DNDGIC shows an anisotropic ESR signal of axial symmetry with g factors g ϭ 2.04 and g ϭ 2.01 (Fig. 6A, trace a). When 20 M DNDGIC are reacted with 44 M GST P1-1 in the presence of 20 mM GSH, the spectrum at 77 K changes slightly, indicating a shift toward a more defined axial symmetry (Fig. 6A, trace b). This spectrum is probably due to the bound complex that still retains two GS Ϫ ions as iron ligands. When the excess of GSH is removed by G25 Sephadex chromatography, the ESR spectrum changes, with a g ϭ 2.03 and g ϭ 2.01, which corresponds to the monoglutathionyl complex (Fig. 6A, trace c), as confirmed by titration data (see above). Readdition of 10 mM GSH to this species, shifts again the ESR spectrum toward that shown by the authentic diglutathionyl species (data not shown).
As expected, the anisotropy of the ESR signal of the bound complex does not change significantly at 298 K, because of the slow tumbling rate of the enzyme, whereas the free complex exhibits a quite isotropic signal at g ϭ 2.03 with multiple lines hyperfine structure (Fig. 6B). Thus, as also indicated by UV analysis, three slightly different paramagnetic forms can be identified: the DNDGIC in a free form, a DNDGIC bound to the GST P1-1 observed with GSH in excess, and a third species in which one GSH molecule has probably been replaced by a ligand provided by the protein or by a solvent molecule.
Molecular Modeling-A view of the active site of the covalent DNGIC (as obtained after removal of the excess GSH) is shown in Fig. 8. The final model shows that: (i) one of the GSH ligands of DNDGIC can dock into the G-site and adopt the canonical extended conformation seen in crystal structures of GST-GSH complexes, (ii) Tyr 7 is close enough to displace the other GSH ligand to generate a stable enzyme-inhibitor complex, and (iii) the NO moieties of the complex form van der Waal's interactions with Ile 104 and Tyr 108 . In addition there are possible polar interactions with Tyr 108 and the main chain nitrogen of Gly 205 .
Concluding Remarks-The findings reported above propose a novel role for GST P1-1 in its interaction with physiological NO carriers like GSNO and DNDGIC. In the absence of GSH, Cys 47 and Cys 101 represent the target residues for GSNO interaction. The modification that occurs on these residues may be restricted only to S-nitrosylation reaction. Different modifications like S-glutathionylation of cysteines residues, as observed in sarcoplasmic reticulum Ca-ATPase (62) and in neurogranina/RC3 and neuromodulin/GAP-43 (63), can be ruled out on the basis of mass spectrometry data. The nitrosylation of Cys 101 does not cause any activity perturbation, whereas the modification of Cys 47 decreases the affinity for GSH to 1.2 mM. However, when one Cys 47 has been nitrosylated, negative cooperativity masks the Cys 47 residue of the second subunit, which remains unmodified even after prolonged incubation times with GSNO. The nitrosylated enzyme displays only slightly increased apparent K m value for GSH, and then it retains a large fraction of its detoxicating potential under physiological GSH concentrations. Thus, GST P1-1, in case of cellular GSH depletion, may act directly as NO protein carrier without detriment to its detoxicating activity. This corresponds to certain cellular conditions such as strong oxidative stress, etc. Under normal physiological conditions (millimolar GSH concentrations and metal traces), GSNO does not react directly with the enzyme but through the dinitrosyl-diglutathionyl iron complex. In fact, DNDGIC is rapidly formed by GSNO decomposition at millimolar GSH concentrations and micromolar ferrous ions. The interaction of this complex with GST P1-1 proceeds through a selected and sophisticated binding mechanism that can be summarized as follows, on the basis of UV and ESR data and molecular modeling: (i) In the presence of excess of GSH, DNDGIC binds to the G-site of the enzyme. One of the two GSH molecules is stabilized by the classical interactions of the G-site with its natural substrate; the NO moieties are stabilized by van der Waal's and possible polar interactions with protein atoms. All of these interactions account for the very high affinity of the enzyme for this complex (K i Ͻ 10 Ϫ12 ). When the excess of GSH is removed, one of the two GSH molecules of the bound DNDGIC is lost from the complex, and one protein residue, possibly the hydroxyl group of Tyr 7 , is probably involved in the coordination of the iron atom. (ii) Binding of DNDGIC triggers a structural modification in the vacant subunit, which lowers its affinity for the complex by about 3 orders of magnitude (K i ϭ 2 ϫ10 Ϫ9 ).
In conclusion, GST P1-1 appears to be a protein precisely designed for a specific and sophisticated interaction with the natural compound DNDGIC. Interestingly, serum albumin, previously suggested as a physiological DNDGIC carrier, displays more than 500 times lower affinity for this complex when compared with GST P1-1.
Because of the wide distribution of this cytosolic enzyme in several tissues, this property suggests a new putative role for this enzyme in the cell as a DNDGIC carrier protein. Interestingly, ESR studies revealed that, in animal tissues, DNDGIC is bound to unknown proteins with apparent molecular mass in the range 50 -120 kDa (64). In some cases, i.e. in the rat aorta (65), such complexes give ESR signals different from DNDGIC in the free form, but very similar to those seen after reaction with GST P1-1.
Our preliminary data suggest that other recently evolved GSTs, i.e. Alpha and Mu isoenzymes, interact with DNDGIC like the Pi enzyme. However, the older bacterial enzyme, which is closer to the ancestral precursor of GSTs and lacks the crucial Tyr residue in the active site, does not interact with DNDGIC. Thus, we believe that the present data, obtained for the human GST P1-1, may represent only a first piece of a more complex scenario that can redefine and enlarge the physiological role of the recently evolved GSTs.