Regulation of Annexin A2 by Reversible Glutathionylation*

The annexin A2-S100A10 heterotetramer (AIIt) is a multifunctional Ca2+-dependent, phospholipid-binding, and F-actin-binding phosphoprotein composed of two annexin A2 subunits and two S100A10 subunits. It was reported previously that oxidative stress from exogenous hydrogen peroxide or generated in response to tumor necrosis factor-α results in the glutathionylation of Cys8 of annexin A2. In this study, we demonstrate that AIIt is an oxidatively labile protein whose level of activity is regulated by the redox status of its sulfhydryl groups. Oxidation of AIIt by diamide resulted in a time- and concentration-dependent loss of the ability of AIIt to interact with phospholipid liposomes and F-actin. The inhibitory effect of diamide on the activity of AIIt was partially reversed by dithiothreitol. In addition, incubation of AIIt with diamide and GSH resulted in the glutathionylation of AIIt in vitro. Mass spectrometry established the incorporation of 2 mol of GSH/mol of annexin A2 subunit at Cys8 and Cys132. Glutathionylation potentiated the inhibitory effects of diamide on the activity of AIIt. Furthermore, AIIt could be deglutathionylated by glutaredoxin (thiol transferase). Thus, we show for the first time that AIIt can undergo functional reactivation by glutaredoxin, therefore establishing that AIIt is regulated by reversible glutathionylation.

The molecular mechanisms by which the cell alleviates oxidative stress and achieves redox homeostasis are still a matter of considerable debate. However, the modulation of the thiol disulfide status of critical cysteine residues on proteins is being recognized as a critical mechanism of oxidative signal transduction as well as a cellular response to protect key regulatory molecules from oxidative insult (1)(2)(3). Recent evidence suggests that the reversible covalent modification of cysteine residues by the tripeptide glutathione (␥-Glu-Cys-Gly) plays a significant role in the antioxidant network of the cells and is involved in regulating individual aspects of cellular function (3,4). Although proteins can bind cysteine, GSH, and homocysteine to generate mixed disulfides, GSH is the dominant ligand, as it occurs in the cell at concentrations between 1 and 10 mM (5,6). S-Glutathionylation has been shown to alter the function of a number of discrete proteins under oxidant stress (7,8). Fur-thermore, the formation of a mixed disulfide with glutathione precludes the irreversible oxidation of the cysteine thiol to a sulfinic or sulfonic acid and enables reactivation of the protein by cellular thioreductases.
Annexins compose a large multigene family of water-soluble proteins that can bind to negatively charged phospholipids and cellular membranes in a Ca 2ϩ -dependent fashion (9 -11). The annexin family is structurally characterized by two domains: a highly conserved ␣-helical protein core consisting of four 70amino acid repeats (eight repeats in the case of annexin VI) and a variable N-terminal segment (12). Annexin A2 is unique among the annexins, for its N-terminal tail possesses a high affinity binding site for a dimeric protein (monomeric M r 11,000), S100A10 (13). Annexin A2 can thus exist as a monomer or as its heterotetrameric complex ((annexin A2) 2 -(S100A10) 2 ), known as the annexin A2-S100A10 heterotetramer (AIIt). 1 Annexin A2 has been shown previously to aggregate and fuse biological membranes (14 -16), and AIIt has been identified as a cross-linker between the plasma membrane and secretory granules (17,18). These observations have led to the suggestion that annexin A2 and AIIt are involved in regulating membrane trafficking events such as exocytosis and endocytosis (reviewed in Ref. 13).
In addition to phospholipid binding activity, a number of annexins have been identified as Ca 2ϩ -dependent F-actin-binding proteins (12). The cytoskeleton of eukaryotic cells has a predominant role in numerous processes such as cell-cell interaction and cellular morphology (19,20). Furthermore, the cytoskeleton participates in signal transduction events that direct cell migration and proliferation and membrane trafficking (21)(22)(23). Annexin A2 can bind F-actin and possesses a pronounced Ca 2ϩ -dependent filament bundling activity when expressed in its heterotetrameric form (9,13). Our laboratory has shown previously that the F-actin-binding site can be mapped directly to the C terminus of annexin A2, as truncation of the last 9 amino acid residues of the C terminus inhibits F-actin binding activity of annexin A2 and AIIt (24).
Annexin A2 was recently identified as an oxidant-sensitive protein in HeLa cells. It was shown that oxidant stress either from exogenous H 2 O 2 or generated in response to tumor necrosis factor-␣ increases incorporation of a glutathione analog into the reactive cysteine of the annexin A2 N-terminal domain (25). In addition, Singh and Liu (26) and Liu and co-workers (27) have demonstrated that chemical modification of AIIt by the sulfhydryl reagent N-ethylmaleimide or by treatment with peroxynitrite results in a loss of liposome aggregation activity in vitro. Although annexin A2 has been implicated in numerous pathophysiological processes, very few studies have been di-rected toward identification of the role these reactive cysteine residues may play in the biological activity of the protein.
In this study, we show for the first time that AIIt is an oxidatively labile protein whose level of activity is regulated by the redox status of its sulfhydryl groups. We also report that glutathionylation of AIIt in vitro results in the modification of Cys 8 and Cys 132 of the annexin A2 subunit and results in the inhibition of the phospholipid and F-actin binding activity of AIIt. However, deglutathionylation of AIIt by glutaredoxin restores AIIt activity. These results suggest that the cysteine(s) of annexin A2 may play a role in the regulation of the biological activities of the protein and can be regulated by reversible glutathionylation.
Mutagenesis of Annexin A2-Bacterial expression vectors containing the wild-type sequence for annexin A2 (pAED4.91-annexin A2) were mutated using the QuikChange TM site-directed mutagenesis kit (Stratagene). Primers that introduced Cys-to-Ser mutations at positions 8 and 132 of annexin A2 were synthesized. All of the mutations introduced were verified by DNA sequence analysis. These various plasmids were then transformed into Escherichia coli BL21(DE3) cells and grown as described previously (28). Recombinant proteins were purified as described previously (28).
Purification of Recombinant AIIt-Equimolar amounts of human recombinant annexin A2 and human recombinant S100A10 were incubated at 25°C for 15 min and then at 4°C for 1 h, and the peak of the heterotetramer was isolated from the individual subunits by gel permeation chromatography. Proteins were subsequently stored in buffer A (40 mM Tris-HCl (pH 7.4), 140 mM NaCl, 0.1 mM EGTA, and 0.1 mM dithiothreitol (DTT)) at Ϫ80°C until used.
Preparation of Biotinylated Glutathione Ethyl Ester (BioGEE)-Bio-GEE was prepared according to a modified procedure of Finkel and co-workers (25). To a solution of 40 mM GSSG in 50 mM Na 2 HPO 4 (pH 8.0) was added a 0.15 M solution of N-hydroxysuccinimidobiotin in Me 2 SO to a final concentration of 18 mM. The mixture was shaken gently overnight at room temperature. DTT was subsequently added to a final concentration of 80 mM, and the solution was incubated at 60°C for 1 h. Unreacted glutathione was removed by incubating the reaction mixture with AG 50W-X4 resin (12 ml) for 1 h at room temperature with shaking. The resultant slurry was centrifuged briefly, and the pH of the extracted supernatant was adjusted to 5.0 and loaded onto an AG 1-X4 column equilibrated in 50 mM sodium acetate (pH 5.2). The column was washed with acetate buffer (6 bed volumes) and then with 90% EtOH (6 bed volumes). The glutathione-containing fractions were eluted with 0.5 M HCl in 90% EtOH. Biotinylated glutathione elutes at pH Յ3. Thiolcontaining fractions were determined using Ellman's reagent, and the solvent was subsequently removed in vacuo using a rotary evaporator. The resultant residue was treated with 1.25 M HCl in anhydrous EtOH at room temperature overnight; the solvent was evaporated using a Speed-Vac; and the concentration of free thiol was determined before use.
Mass Spectrometry-The molecular mass of control or glutathionylated AIIt was initially analyzed by liquid chromatography/mass spectrometry on a Micromass Q-TOF 2 mass spectrometer. Prior to tryptic digestion, AIIt was separated into annexin A2 and S100A10 subunits and purified using reversed-phase chromatography on a Zorbax 300 SB-C8 column (1 mm ϫ 15 cm). The annexin A2 subunit was dried and rehydrated into 100 mM ammonium bicarbonate, and trypsin was added. Digestion was allowed to proceed for 10 min (limited digestion) or overnight (complete digestion) at 37°C. An aliquot of the reaction mixture was also reduced with 10 mM tris (2-carboxyethyl)phosphine hydrochloride (TCEP). The Micromass Q-TOF 2 mass spectrometer with a CapLC high pressure liquid chromatography system was used to analyze samples. Data obtained by Q-TOF liquid chromatography-tandem mass spectrometry (LC-MS/MS) was searched using the software search engine Mascot from Matrix Science.
Method for Free Sulfhydryl Determination Using Ellman's Reagent-The Ellman method for assaying thiols is based on the reaction of thiols with chromogenic 5,5Ј-dithiobis(2-nitrobenzoic acid) whereby formation of the yellow dianion of 5-thio-2-nitrobenzoic acid is measured. Samples were incubated with 200 M 5,5Ј-dithiobis(2-nitrobenzoic acid) in 100 mM Na 2 PO 4 (pH 8.0) in a total volume of 200 l. Each sample was then analyzed spectrophotometrically at 405 nm using an EL x -808 spectrophotometric plate reader (Bio-Tek Instruments). The results were related to a standard curve generated using reduced glutathione as the standard. Argon Dialysis of Proteins-After purification and reconstitution, wild-type and mutant AIIt, annexin A2, and S100A10 were dialyzed against pre-degassed buffer A under argon to prevent possible oxidation of cysteine residues.
Modification of AIIt by Glutathionylation-Dialyzed AIIt (8 -10 M) was treated with GSH (2 mM) and diamide (1 mM) in degassed buffer A at room temperature for 2 h. Excess diamide and GSH were then removed by performing three buffer exchanges using an Amicon Centricon-10 microconcentrator. Glutaredoxin-mediated deglutathionylation of AIIt was examined by incubation of glutathionylated AIIt (18 M) with 0.5 mM GSH and 1.0 M glutaredoxin in buffer A at 25°C.
Electrophoresis and Western Blotting-Samples were treated with SDS-PAGE sample buffer, subjected to SDS-PAGE, and electrophoretically transferred to a nitrocellulose membrane (0.45-m pore size) at 4°C for 1 h. The membrane was blocked with 5% skim milk in Trisbuffered saline (TBS; 10 mM Tris-HCl and 154 mM NaCl (pH 7.4)) for 1 h at room temperature and then treated for 1 h at room temperature with 1.0 g/ml monoclonal anti-annexin A2 or anti-S100A10 antibody or 2 g/ml anti-glutathione antibody in TBS with 5% skim milk. The blot was washed thoroughly with TBS containing 0.1% Tween 20 and then incubated at room temperature with a 1:3000 dilution of horseradish peroxidase-conjugated goat anti-mouse secondary antibody in TBS with 5% skim milk. In the case of biotinylated samples, the blocked membrane was incubated with 0.4 g/ml horseradish peroxidase-conjugated streptavidin in TBS with 5% bovine serum albumin. After the respective antibody incubations, the membrane was washed extensively with TBS containing 0.1% Tween 20 and visualized using enhanced chemiluminescence (Pierce).
Phospholipid Vesicle Aggregation Assay-In 12 ϫ 75-mm culture tubes, 50 l each of 20 mg/ml phosphatidylserine, phosphatidylethanolamine, and cholesterol (dissolved in chloroform) were shelled by N 2 gas. The resultant lipid residue was resuspended in 1 ml of phospholipid aggregation buffer (50 mM HEPES (pH 7.5) and 2 mM MgCl 2 ) and sonicated at 75 watts for three 15-s bursts with a Braun probe sonicator to generate phospholipid vesicles at a concentration of 1 mg/ml. Phospholipid vesicle aggregation was assayed in a Varian UV-visible spectrophotometer at 450 nm in a final volume of 600 l. Initially, phospholipid vesicles (35 g), Ca 2ϩ (8 M), and phospholipid aggregation buffer were combined, resulting in a final phospholipid concentration of 0.055 mg/ml. To initiate the reaction, 20 g of AIIt or annexin A2 were added to the reaction mixture. Absorbance readings were taken continuously for 10 min. Results are expressed as the percentage of starting A 450 nm (no AIIt added) and are typically presented as mean Ϯ S.D. (n ϭ 3) as outlined previously (29). For the calculation of percent AIIt activity, the extent of phospholipid aggregation at 10 min by untreated AIIt was considered as 100% activity.
Phospholipid Vesicle Binding Assay-Following phospholipid liposome aggregation, the reaction mixture (0.600 ml) was centrifuged at 14,000 ϫ g in a desktop centrifuge. The protein/phospholipid ratios used in the analysis ranged from 0.001 to 0.02 AIIt/phospholipid. The pellet was resuspended in 30 l of SDS disruption buffer and boiled for 3-5 min. The pellet fraction was then resolved by SDS-PAGE and stained with Coomassie Blue.
F-actin Bundling Assay-F-actin was prepared as described previously (30). F-actin bundling was measured by the light scattering intensity perpendicular to the incident light in a PerkinElmer Life Sciences LS-50B luminescence spectrophotometer as described previously (31). The excitation and emission wavelengths were both set at 400 nm with 10-and 5-nm slit widths, respectively. F-actin (0.60 M) was incubated in bundling buffer (25 mM MOPS (pH 7.0) and 0.33 mM ATP), and CaCl 2 (500 M) was added to the bundling assay mixture to a final volume of 0.600 ml. Bundling was initiated by the addition of AIIt to a final concentration of 0.85 M. After a 15-min incubation at room temperature, light scattering intensity was recorded in at least triplicate. Following the F-actin bundling assay, the reaction mixture (0.600 ml) was centrifuged at 14,000 ϫ g for 10 min in a desktop centrifuge. The pellet (which contained F-actin bundles) was resuspended in 30 l of SDS disruption buffer and boiled for 3-5 min. The pellet fraction was then resolved by SDS-PAGE and stained with Coomassie Blue.
F-actin Binding Assay-F-actin binding was performed as described above, except that the reaction volume was 0.200 ml. After the 15-min incubation period, the sample was centrifuged at 400,000 ϫ g for 30 min in a Beckman Optima TLX ultracentrifuge. The pellet was first solubilized in 25 l of 100 mM KCl for 30 min, and then 25 l of SDS disruption buffer were added. Subsequently, the mixture was boiled for 3-5 min. The pellet fraction was resolved by SDS-PAGE.
Miscellaneous Techniques-Protein concentrations were determined using Coomassie Brilliant Blue and standardized the concentrations to bovine serum albumin as described by Bradford (32). All reagents used were analytical grade or better. Data were analyzed using SigmaPlot (Jandel Scientific).

Oxidant-dependent Incorporation of Glutathione into
Annexin A2-BioGEE is a biotinylated membrane-permeable analog of glutathione that was developed to detect incorporation of glutathione into proteins exposed to oxidative stress. It has been demonstrated previously using BioGEE that two prominent S-glutathionylated proteins of HeLa cells include thioredoxin peroxidase and annexin A2 (25). We have investigated the generality of this observation using HT1080 fibrosarcoma cells as a model system. As shown in Fig. 1A, oxidant stress from exogenous hydrogen peroxide caused an increase in the incorporation of BioGEE into a large number of cellular proteins. Furthermore, reduction of these proteins with DTT resulted in the loss of the label, establishing that these proteins were indeed S-glutathionylated. However, treatment of cells with hydrogen peroxide did not alter the cellular protein levels of annexin A2 (Fig. 1B). Next, we exposed the HT1080 cells to hydrogen peroxide and purified the glutathionylated proteins from the crude cellular extract by pull-down with streptavidinagarose, followed by elution with DTT. The glutathionylated proteins were resolved by SDS-PAGE, and annexin A2 was identified by Western blotting. As shown in Fig. 1C, treatment of cells with hydrogen peroxide resulted in the enrichment of annexin A2 in the glutathionylated protein fraction. This establishes that prior exposure of the HT1080 cells to hydrogen peroxide enhanced the incorporation of the biotinylated glutathione analog into annexin A2, therefore confirming the suggestion that annexin A2 becomes S-glutathionylated upon oxidative stress (25). and treated AIIt was used as a measure of the amount of phospholipid aggregation activity remaining. Inset, AIIt was treated with 1 mM diamide at room temperature for the indicated times. Immediately after taking the first absorbance reading, 20 g of AIIt were diluted 12-fold into the reaction mixture, and A 450 nm was continuously recorded for 10 min. Results are shown as a percentage of starting A 450 nm (no AIIt added). B, AIIt was incubated with various concentrations of diamide for 2 h at room temperature, and the amount of phospholipid aggregation activity remaining after treatment was determined. Results are expressed as mean Ϯ S.D. (n ϭ 3). C, AIIt (20 g) was incubated in the absence (lane 5) or presence of 2 mM (lanes 1 and 2) or 5 mM (lanes 3 and 4) diamide. After 15 min, the reaction mixture was either left untreated (lanes 1 and 3) or adjusted to 9 mM DTT and incubated for an additional hour (lanes 2 and 4). The reaction mixture was then analyzed by SDS-PAGE under nonreducing conditions. D, in some experiments, this reaction mixture was also assayed for liposome aggregation activity. At the end of the assay (10 min), the mixture was centrifuged, and the pellet was analyzed by SDS-PAGE under reducing conditions. AIIt (lane 1), AIIt treated with diamide (lanes 3 and 6), or AIIt treated with diamide followed by DTT (lanes 4 and 7) was incubated with phospholipid liposomes (Pl); and after completion of the aggregation reaction (10 min), the mixture was centrifuged, and pellets (phospholipid-bound AIIt) were analyzed by SDS-PAGE. Controls for the nonspecific pelleting of AIIt in the absence of phospholipid are also included (lanes 2 and 5). E, the interaction of diamide-treated AIIt with F-actin was also examined. F-actin bundling activity was quantitated by measurement of light scattering after 15 min at room temperature. Results are expressed as mean Ϯ S.D. (n ϭ 3). Inset, after 15 min at room temperature, the reaction mixture was centrifuged at 14,000 ϫ g for 10 min (upper panel) or at 400,000 ϫ g for 30 min (lower panel). The pellets were resolved by SDS-PAGE and stained with Coomassie Blue. Lanes a and aЈ, F-actin alone; lanes b and bЈ, AIIt ϩ F-actin; lanes c and cЈ, 1 mM diamide-treated AIIt ϩ F-actin; lanes d and dЈ, 5 mM diamide-treated AIIt ϩ F-actin; lanes e and eЈ, 1 mM diamide-treated AIIt incubated with 10 mM DTT for 1 h ϩ F-actin; lanes f and fЈ, 5 mM diamide-treated AIIt incubated with 10 mM DTT for 1 h ϩ F-actin. Anx2, annexin A2.
AIIt Is Sensitive to Sulfhydryl Oxidation-Diamide is a highly specific thiol oxidant that preferentially oxidizes small thiol-containing molecules such as glutathione (33,34). However, diamide can also oxidize exposed cysteines of proteins. The observation that AIIt is glutathionylated in vivo suggests that AIIt might be susceptible to sulfhydryl oxidation. We tested the susceptibility of AIIt to thiol oxidation by measuring the activity of AIIt after treatment with diamide. As a measure of the biological activity of AIIt, we chose the well characterized ability of AIIt to aggregate phospholipid vesicles in a Ca 2ϩ -dependent manner (reviewed in Refs. 12 and 35). As shown in Fig.  2 (A and B), diamide dramatically inhibited the activity of AIIt in a time-and concentration-dependent manner. Half-maximal inhibition of AIIt activity occurred at 0.24 mM diamide; and at 2 mM diamide, the activity was maximally inhibited by ϳ90%.
We then determined whether the oxidation of AIIt by diamide was reversible. As shown in Fig. 2B, incubation of AIIt with 2 mM diamide resulted in a significant loss of protein activity, and this loss could be partially reversed by DTT. In contrast, more rigorous treatment of AIIt with 5 mM diamide resulted in a loss of the ability of DTT to restore the activity of the protein. This suggests that the inhibition of the activity of AIIt by diamide involves oxidation of sulfhydryl groups because it was partially reversed by DTT (Fig. 2B). The inability of DTT to stimulate recovery of the activity of AIIt at higher diamide concentrations suggests that the protein could also be irreversibly oxidized by diamide.
Our results demonstrate that oxidation of AIIt by diamide causes a dramatic loss of AIIt activity and that DTT can partially reverse this diamide effect. Diamide may therefore be blocking the activity of AIIt by inducing disulfide bond formation between the subunits of AIIt or between the two different molecules of AIIt. To determine the mode of action whereby diamide may be inducing structural changes to the protein, nonreducing SDS-PAGE analysis of diamide-treated and untreated AIIt was performed. As shown in Fig. 2C, diamide treatment of AIIt resulted in a small shift in the mobility of the protein as well as the formation of aggregates. The change in the appearance and mobility of AIIt upon SDS-PAGE was reversed if the diamide-treated protein was incubated with DTT. These observations suggest that diamide causes the oxidation of AIIt thiols and the formation of disulfide linkages between AIIt molecules.
To examine whether diamide treatment affected the binding of AIIt to the phospholipid vesicles, the aggregation mixture was centrifuged, and the AIIt pellet was analyzed by SDS-PAGE. Analysis of AIIt bound to the phospholipid vesicles revealed that oxidation of AIIt by diamide resulted in a loss of the ability of AIIt to bind to the phospholipid vesicles (Fig. 2D). Furthermore, incubation of the diamide-oxidized protein with DTT afforded partial restoration of AIIt activity.
It has been shown previously that annexin A2 binds F-actin in a Ca 2ϩ -dependent manner in vitro (31,36). Recent studies have established that the C terminus of the annexin A2 subunit of AIIt comprises an F-actin-binding domain. Interestingly, the last 9 amino acid residues of the C terminus of the annexin A2 subunit ( 330 LLYLCGGDD 338 ) appear to be entirely responsible for the F-actin binding activity of AIIt (24). Upon Ca 2ϩ -dependent binding to F-actin, AIIt (but not annexin A2) rapidly and reversibly forms anisotropic F-actin bundles (31). We used light scattering to examine the ability of diamide-oxidized AIIt to bundle actin filaments. As shown in Fig. 2E, oxidation of AIIt by diamide resulted in a loss of F-actin bundling. Addition of DTT partially reversed the oxidative effects of 1 mM diamide, but failed to reverse the oxidative effects of 5 mM diamide. To confirm that the decrease in light scattering was due to a loss of formation of supramolecular structures (bundles) of F-actin, we took advantage of the fact that F-actin does not sediment at 14,000 ϫ g, but that the F-actin bundles, which consist of both F-actin and AIIt, can be harvested at this centrifugal force. As shown in Fig. 2E (inset, upper panel), the formation of the F-actin bundles by diamide-oxidized AIIt was barely detectable, and DTT treatment partially restored F-actin bundle formation.
Next, we examined whether or not the lack of F-actin bundling activity displayed by diamide-oxidized AIIt was due to decreased binding of F-actin. We assessed the ability of diamide-oxidized AIIt to bind to F-actin using high speed centrifugation (400,000 ϫ g). Under these experimental conditions, F-actin-binding proteins co-sedimented with F-actin. As shown in Fig. 2E (inset, lower panel), oxidation of AIIt resulted in a loss of F-actin binding activity, which was partially reversed by DTT. However, 5 mM diamide irreversibly oxidized AIIt. Thus, higher concentrations of diamide cause the irreversible oxidation of AIIt and the subsequent loss of the phospholipid and F-actin binding activity of the protein.
AIIt Is Readily Modified by Glutathionylation-The observation that AIIt is inhibited by diamide is consistent with the demonstration that AIIt is susceptible to glutathionylation in vivo. However, before we could investigate the effect of glutathionylation on the activity of AIIt, we needed to generate sufficient quantities of the glutathionylated protein in a cellfree system. Therefore, we incubated AIIt with reduced glutathione and diamide and analyzed the reaction product by Western blotting using an antibody to glutathione. As shown in Fig.  3A, Western blot analysis confirmed the formation of a mixed disulfide between glutathione and the annexin A2 subunit. Because incubation of glutathionylated AIIt with the sulfhydryl-reducing agent DTT prior to electrophoresis resulted in the removal of glutathione immunoreactivity, we concluded that AIIt was S-glutathionylated.
Glutathionylated AIIt was then subjected to mass spectrometric analysis (Fig. 3). The molecular masses of the S100A10 subunit were determined to be 11,071 and 11,087 Da for the diamide/glutathione-treated and untreated proteins, respectively. From these results, it was apparent that S100A10 was not glutathionylated. The average molecular mass of the diamide/glutathione-treated annexin A2 subunit was determined to be 39,131 Da compared with 38,521 Da for the untreated protein (Fig. 3, B and C). Assuming a molecular mass of 305 Da for glutathione, this difference in molecular mass of 610 Da suggested that each annexin A2 subunit had been covalently modified by 2 mol of glutathione/mol of annexin A2 subunit. This estimation was also confirmed by determining the number of free thiols in glutathionylated AIIt. As assessed by the Ellman method, an overall loss of ϳ4.5 thiols between untreated AIIt (6.90 Ϯ 0.22, mean Ϯ S.D., n ϭ 4) and glutathionylated AIIt (2.40 Ϯ 0.12, mean Ϯ S.D., n ϭ 4) was observed.
To determine the S-glutathionylation sites in annexin A2, both S-glutathionylated AIIt and annexin A2 were subjected to tryptic digestion, and the resultant peptides were analyzed by Q-TOF liquid chromatography-tandem mass spectrometry. Limited tryptic digestion of the annexin A2 subunit has been reported to result in the fragmentation of the protein into a 27-amino acid N-terminal domain peptide and a core protein (35). The N-terminal peptide contains only a single thiol, Cys 8 . As shown in Fig. 3E, the monoisotopic molecular mass of the N-terminal domain peptide produced by limited tryptic digestion of the glutathionylated protein was 3202.5 Da compared with an expected molecular mass of 2898.2 Da for the native peptide. The difference in molecular mass of 304.3 Da is consistent with the presence of glutathione on the peptide. Treat- ment of the peptide with the reducing agent tris(2-carboxyethyl)phosphine hydrochloride resulted in a decrease in the molecular mass from 3202.2 to 2897.3 Da (Fig. 3D), consistent with the S-glutathionylation of the peptide at Cys 8 .
Next, we performed a complete tryptic digestion of the glutathionylated annexin A2 subunit and analyzed the digest by Q-TOF liquid chromatography-tandem mass spectrometry. Two peptides were identified as containing glutathione. The first peptide (STVHEILCK) confirmed the limited tryptic digest data that identified Cys 8 as a glutathionylated residue. The second peptide (GLGTDEDSLIEIICSR) identified Cys 132 as a glutathionylated residue. Collectively, the mass spectral data established that Cys 8 and Cys 132 thiols, located in the N-terminal region and second annexin fold repeat, respectively, are readily glutathionylated in vitro.
Inhibition of AIIt Activity by S-Glutathionylation-We initially observed that thiol oxidation of AIIt by diamide inhibited the phospholipid aggregation activity of the protein and that 1 mM diamide produced an inhibition of AIIt activity by ϳ70% (Fig. 2). Upon incubation of AIIt with 1 mM diamide and 2 mM reduced glutathione, a glutathionylated protein with 2 mol of glutathione/mol of annexin A2 subunit was isolated (Fig. 3). As shown in Fig. 4A, at the concentration of AIIt normally used in our activity assays, 0.37 M (20 g), glutathionylated AIIt was inactive. This suggests that glutathione potentiates the inhibitory effect of diamide on the activity of the protein. When increased concentrations of glutathionylated AIIt were used in the assay, some activity could be observed. Because the possibility existed that glutathionylation of AIIt resulted in an increase in the Ca 2ϩ requirement for aggregation, we examined the activity of the protein with 50-fold more Ca 2ϩ . As shown in Fig. 4A, the increased Ca 2ϩ failed to increase the activity of the glutathionylated protein, suggesting that glutathionylation of AIIt does not simply alter the Ca 2ϩ requirement for liposome aggregation.
Although the glutathionylation of AIIt blocks its activity, it was unclear if this represented a reversible mechanism for the protection of thiols against irreversible oxidation. Therefore, we examined the effect of removal of glutathione on AIIt activity. Because reduction of glutathionylated proteins results in the removal of glutathione, we examined the effect of DTT on the activity of glutathionylated AIIt. As shown in Fig. 4B, incubation of glutathionylated AIIt with DTT reactivated the protein, with full recovery of AIIt activity being achieved in ϳ40 min.
Glutaredoxin (thiol transferase) is a well characterized enzyme that functions in vivo to reverse protein-glutathionemixed disulfides by utilizing reduced glutathione as an electron donor. Glutaredoxin is therefore a critical component of the cellular machinery that serves to maintain and reverse the glutathionylation of susceptible protein thiols in vivo. We therefore tested the possibility that glutathionylated AIIt is a substrate of glutaredoxin. As shown in Fig. 4C, incubation of glutathionylated AIIt with glutaredoxin and glutathione resulted in the complete restoration of AIIt activity within 20 min.
A previous study reported that incubation of purified AIIt with S-nitrosoglutathione results in the inhibition of AIIt-mediated liposome aggregation, but does not affect the binding of AIIt to phospholipid (37). However, as shown in Fig. 4D, glutathionylated AIIt failed to bind to phospholipid liposomes. Furthermore, incubation of glutathionylated AIIt with glutaredoxin resulted in the complete reversal of the inhibitory effects of glutathione. Collectively, these results demonstrate that AIIt can be reversibly S-glutathionylated and that glutathionylation of AIIt blocks the ability of the protein to bind to and aggregate phospholipid.
Glutathionylation Inhibits the Interaction of AIIt with Factin-We also examined the ability of glutathionylated AIIt to bundle actin filaments using light scattering. As shown in Fig.  5A, glutathionylated AIIt showed a dramatic 90% reduction in F-actin filament bundling activity as measured by light scattering. In contrast, incubation of AIIt with 1 mM diamide resulted in a loss of ϳ60% of the F-actin bundling activity of AIIt (Fig. 2E). This further suggests that glutathione enhances the inhibitory effect of diamide on the activity of AIIt. Furthermore, we also observed that incubation of glutathionylated AIIt with DTT or glutaredoxin afforded recovery of 50% of the bundling activity of the protein (Fig. 5A).
in the light scattering analysis were re-analyzed by SDS-PAGE. As shown in Fig. 5B, the formation of the F-actin bundles by glutathionylated AIIt was barely detectable, and glutaredoxin treatment partially restored this F-actin bundle formation. Next, we examined whether or not the lack of Factin bundling activity displayed by glutathionylated AIIt was due to a decreased binding affinity for F-actin. As shown in Fig.  5C, glutathionylated AIIt displayed only ϳ15-20% of the Factin binding activity of wild-type AIIt. Furthermore, incubation of glutathionylated AIIt with thioredoxin resulted in the partial recovery of the F-actin binding activity of AIIt.
Role of Cysteine Thiols in AIIt-mediated Phospholipid Aggregation and Binding-Our results show that glutathionylation of Cys 8 and Cys 132 of the annexin A2 subunit of AIIt results in a loss of phospholipid-and F-actin binding activity. Because it remained unclear whether one or both of these cysteines were responsible for the glutathione-induced inactivation of AIIt, we decided to investigate the protein activities of two serine AIIt mutants. C8S and C132S AIIt mutants were prepared by combining either of the mutant annexin A2 subunits with the wild-type S100A10 subunit and purifying the resultant heterotetramers. Studies indicated that the C8S AIIt mutant retained ϳ80% of its activity (Fig. 6). However, incubation of this mutant with diamide resulted in a dramatic loss of activity, which was not recoverable by incubation of the oxidized protein with DTT. This implies that, in addition to Cys 8 , other cysteines are involved in the diamide-induced loss of AIIt activity. The inability to reversibly oxidize the C8S AIIt mutant suggests that this cysteine plays a key role in the DTT-induced reactivation of the oxidized protein. Unfortunately, the C132S AIIt mutant retained only ϳ40% of its activity. The CD spectrum of this mutant was aberrant, showing a loss of helical structure (data not shown). Therefore, we were unable to determine the role that Cys 132 plays in the diamide-induced loss of activity of AIIt. DISCUSSION Formation of mixed disulfides between glutathione and cysteines in proteins (glutathionylation) has long been known to occur upon oxidative stress (reviewed in Refs. 3, 7, and 38). Recently, protein glutathionylation has gained attention as a possible means of redox regulation of protein function. Thus, glutathionylation can serve the dual purpose of redox signaling as well as protecting protein from irreversible oxidative modification. Glutathionylation may result in the inhibition of protein function, as has been reported for such proteins as enolase or 6-phosphogluconolactase, whereas for others such as cyclophilin, the activity does not change with glutathionylation (8).
Although the majority of studies have reported an inhibition of enzyme activity by glutathionylation, it appears that the position of the cysteine undergoing modification is important in determining the effect of glutathionylation. For instance, human immunodeficiency virus type 1 protease can be inhibited or stabilized by glutathionylation depending on which cysteine is involved (39), whereas matrix metalloproteinases are activated upon glutathionylation of the autoinhibitory domain (40).
In this study, we have confirmed the observation that oxidative stress of cells results in the glutathionylation of the annexin A2 subunit of AIIt. This observation further supports the possibility that AIIt might be regulated by an oxidative mechanism (25). We have now shown that AIIt is an oxidatively labile protein whose level of activity is regulated, in part, by the redox status of its sulfhydryl groups. Oxidation of AIIt by diamide results in a loss of the ability of AIIt to interact with phospholipid liposomes and F-actin. Furthermore, glutathionylation of AIIt mimics the inhibitory effects of diamide and inhibits the interaction of AIIt with both phospholipid liposomes and F-actin. Most important, AIIt can be deglutathionylated by glutaredoxin. Thus, we have shown for the first time that AIIt can undergo functional reactivation by glutaredoxinmediated deglutathionylation.
AIIt is a Ca 2ϩ -binding protein that is composed of two annexin A2 (p36) and two S100A10 (p11) subunits (reviewed in Refs. 12, 35, and 41-43). The predominant intracellular form of annexin A2 appears to be complexed with S100A10 as AIIt. The annexin A2 subunit is a planar curved molecule with opposing convex and concave sides. The convex side faces the biological membrane and contains the Ca 2ϩ -and phospholipid-binding sites. The concave side faces the cytosolic milieu and contains both the N-and C-terminal regions of the protein. The Cterminal region consists of multiple ligand-binding sites, including binding sites for F-actin (24), fibrin (44), and heparin (45). The N-terminal region contains the binding site for the S100A10 subunit, which is a member of the S100 family of Ca 2ϩ -binding proteins (46). Annexin A2 contains 4 cysteine residues: Cys 8 , Cys 132 , Cys 261 , and Cys 334 . Cys 8 and Cys 334 are located on the convex surface in the N-and C-terminal regions, respectively. Cys 334 is not exposed until the molecule binds plasmin, at which point this residue cleaves a disulfide bond in plasmin (28). The function(s) of Cys 132 and Cys 261 are unknown, although it has been suggested, based on crystallographic data, that these residues may form a disulfide in the annexin A2 monomer (47). In contrast, Cys 8 has been well characterized. This residue has been shown to be a reactive thiol that is glutathionylated in vivo and that reacts with homocysteine and N-ethylmaleimide in vitro (25,26,48).
Kosower et al. (49) examined the reaction of diamide with a variety of sulfhydryl reagents and concluded that the reaction of diamide with glutathione has the largest rate constant among all of the numerous sulfhydryl reagents tested. Glutathione is the most abundant non-protein thiol in the cell; thus, diamide oxidizes glutathione to GSSG in a rapid and specific FIG. 6. Role of cysteine residues in the redox regulation of AIIt activity. Recombinant wild-type or mutant AIIt was constructed by combining recombinant annexin A2 and S100A10 subunits and isolating the heterotetramer. AIIt was left untreated (None); treated with 2 mM diamide for 15 min (Diamide); or treated with 2 mM diamide for 15 min, subjected to ultrafiltration to remove the diamide, and incubated with 5 mM DTT for 1 h (DTT). 20 g of wild-type AIIt (bar 7), C8S AIIt (bars 1-3), or C132S AIIt (bars 4 -6) were added to the reaction mixture immediately after taking the first absorbance reading, and A 450 nm was continuously recorded for 10 min. Results are shown as a percentage of the activity of untreated wild-type protein and are expressed as mean Ϯ S.D. (n ϭ 5). Inset, after completion of the aggregation reactions, phospholipid-associated AIIt was visualized by centrifugation of the reaction mixtures at 14,000 ϫ g and subsequent resolution of the phospholipidassociated protein by SDS-PAGE. manner. Furthermore, diamide will react preferentially with a small thiol such as GSH rather than a protein thiol because GSH is less sterically hindered (50). Our results suggest that, in the absence of glutathione, diamide reacts with an accessible protein thiol. However, when glutathione and diamide are both present in the reaction with annexin A2, glutathione preferentially reacts with diamide, producing an activated sulfenylhydrazine intermediate. This activated glutathione species can then react with a protein sulfhydryl group to produce glutathionylated annexin A2. The treatment of AIIt with diamide and reduced glutathione resulted in the incorporation of 2 mol of glutathione/mol of annexin A2 subunit. Subsequent analysis by mass spectrometry revealed that Cys 8 and Cys 132 were oxidized to mixed disulfides with glutathione. The loss of AIIt activity was recovered by the addition of DTT, presumably by reducing the mixed disulfides. In addition, inactivated AIIt was reactivated enzymatically by the glutathione-specific dethiolase enzyme glutaredoxin (thiol transferase), further supporting the proposal that the inactivated form of AIIt is a glutathionyl-mixed disulfide. This is the first demonstration of the regulation of AIIt activity by reversible glutathionylation.
Glutathionylated proteins also can be formed by reaction with S-nitrosoglutathione, a pathway requiring the formation of this compound from nitric oxide. It was demonstrated recently that incubation of AIIt with S-nitrosoglutathione, a nitric oxide donor, results in a loss of liposome aggregation (but not liposome binding) by AIIt (37). However, it was not established in these studies if a glutathionyl-mixed disulfide is formed with AIIt. This report did present the possibility that the glutathionylation of AIIt may be also linked to the nitric acid cell signaling pathway.
The mechanism by which glutathionylation inhibits the activity of AIIt is unclear. It is possible that the glutathionylation of Cys 8 and Cys 132 simply blocks the sulfhydryl moiety and prevents its function as a reactive nucleophile in the cellular medium. Another possibility is that the glutathionyl peptide introduces negative charges in AIIt and causes a conformational change in the molecule. In support of this suggestion is our observation of the phosphorylation of Tyr 23 or Ser 25 of annexin A2, which results in the addition of a negative charge to these residues, causing a loss of AIIt activity (51,52). Alternatively, it is conceivable that introduction of the bulky glutathione group into annexin A2 causes a disruption of the conformation of the protein, resulting in a loss of activity.
The exact role that glutathionylation of AIIt plays in vivo is unclear. Because the redox status of the cysteines of AIIt appears to be an important regulator of phospholipid and F-actin binding, it is possible that glutathionylation serves to protect these residues from irreversible oxidation. During oxidative insult, reactive cysteine residues are rapidly oxidized initially to the sulfenic acid form. This sulfenic acid derivative can then react with glutathione to form the mixed disulfide. The cysteinesulfenic acid derivative can easily further oxidize to its irreversible sulfinic and sulfonic acid forms if it is not converted to a more stable and reversible end product such as the glutathionyl derivative. Our observation that diamide can both reversibly and irreversibly oxidize AIIt suggests that diamide can oxidize Cys 8 and Cys 132 to the sulfenic acid and then to the sulfinic acid derivatives. Glutathionylation of the cysteinesulfenic acid derivative will therefore prevent the protein from further oxidation to its irreversible forms and thus constitutes an efficient mechanism for the regulation of AIIt activity.
To further address the issue of whether or not Cys 132 glutathionylation is critical for regulation of AIIt activity, we prepared C8A and C132A mutants. Unfortunately, the C8A mutant failed to exhibit any phospholipid liposome aggregation activity. In addition, the CD spectrum of this mutant showed a loss of ␣-helical structure, suggesting that this mutant is unstable. Although stable, the C132A mutant failed to form the heterotetramer. Therefore, we compared the effect of diamide on the wild-type and C132A mutant annexin A2 proteins. Surprisingly, we observed that the phospholipid aggregation activity of the C132A mutant was not inhibited by diamide (data not shown). This suggests that either Cys 132 plays an important role in the sensitivity of annexin A2 to oxidation or that mutation of Cys 132 results in a conformational change in annexin A2 that results in loss of diamide sensitivity. It is therefore unclear whether or not this residue plays a role in the regulation of the activity of the heterotetramer.
The glutathionylation of AIIt may be a molecular mechanism through which oxidation can induce changes in protein conformation and therefore function, creating a control point for redox regulation of AIIt activity. Glutathione may serve as a second messenger of redox signaling because of its capacity to form a reversible mixed disulfide bond with protein thiols. We have demonstrated that glutaredoxin catalyzes reversible Sglutathionylation of AIIt. We therefore propose that the intracellular function(s) of AIIt may be regulated by redox signaling.