Acute cadmium exposure inactivates thioltransferase (Glutaredoxin), inhibits intracellular reduction of protein-glutathionyl-mixed disulfides, and initiates apoptosis.

Oxidative stress broadly impacts cells, initiating regulatory pathways as well as apoptosis and necrosis. A key molecular event is protein S-glutathionylation, and thioltransferase (glutaredoxin) is a specific and efficient catalyst of protein-SSG reduction. In this study 30-min exposure of H9 and Jurkat cells to cadmium inhibited intracellular protein-SSG reduction, and this correlated with inhibition of the thioltransferase system, consistent with thioltransferase being the primary intracellular catalyst of deglutathionylation. The thioredoxin system contributed very little to total deglutathionylase activity. Thioltransferase and GSSG reductase in situ displayed similar dose-response curves (50% inhibition near 10 micrometer cadmium in extracellular buffer). Acute cadmium exposure also initiated apoptosis, with H9 cells being more sensitive than Jurkat. Moreover, transfection with antisense thioltransferase cDNA was incompatible with cell survival. Collectively, these data suggest that thioltransferase has a vital role in sulfhydryl homeostasis and cell survival. In separate experiments, cadmium inhibited the isolated component enzymes of the thioltransferase and thioredoxin systems, consistent with the vicinal dithiol nature of their active sites: thioltransferase (IC(50) approximately 1 micrometer), GSSG reductase (IC(50) approximately 1 micrometer), thioredoxin (IC(50) approximately 8 micrometer), thioredoxin reductase (IC(50) approximately 0.2 micrometer). Disruption of the vicinal dithiol on thioltransferase (via oxidation to C22-SS-C25; or C25S mutation) protected against cadmium, consistent with a dithiol chelation mechanism of inactivation.

Thioltransferase (glutaredoxin) is a member of the TDOR 1 enzyme family, which also includes thioredoxin and their corresponding reductase enzymes GSSG reductase and thioredoxin reductase, respectively. Thioltransferase is a 12-kDa cytosolic protein that has been characterized in vitro as a specific catalyst for the reduction of protein-glutathionyl-mixed disulfides (protein-SSG) (1)(2)(3)(4)(5). The reaction catalyzed by thioltransferase is also selective for GSH as the reducing substrate (3). This thiol-disulfide interchange reaction is likely crucial for maintaining intracellular thiol status (2,6). Under normal conditions the intracellular milieu is predominately reducing (typical GSH/GSSG ratio Ն 100), but physiological and pathophysiological processes like aging, cardiovascular and neurodegenerative diseases, AIDS, and cancer chemotherapy can shift the redox balance toward an oxidizing milieu. A crucial component of the cellular redox balance is modulation of the thioldisulfide status of critical cysteine residues on proteins, and a prevalent modification of cysteine residues is reversible S-glutathionylation of proteins. Accumulation of protein-SSG has been reported in different cell types under a variety of oxidative conditions (7)(8)(9)(10)(11). With many sulfhydryl proteins, glutathionylation is inactivating. Examples include the transacting factor NF-1 (12), protein tyrosine phosphatase 1B (13), and phosphofructokinase (14,15). With other proteins glutathionylation can be an activation event. Examples include HIV-1 protease (16,17) and microsomal glutathione S-transferase (18). Contrasting effects of reversible glutathionylation for different proteins studied in vitro suggest that protein-SSG formation is a mechanism of regulation and/or intracellular signal transduction (2,6,11,13,17,19) and implicate thioltransferase as a key player in these cellular processes.
Cysteine residues can be modified reversibly in other ways that also can affect the activities of proteins. Examples include sulfenic acids and intra-or intermolecular disulfides. Thioltransferase does not catalyze reduction of these oxidized forms unless they react first with GSH to form protein-SSG (2). However, thioredoxin has been reported to catalyze reduction of these sulfhydryl modifications (4, 15, 20 -22), displaying preferential reduction of intramolecular disulfides and proteinsulfenic acids compared with intermolecular disulfides and protein-SSG (15,20). In contrast to thioltransferase, thioredoxin is neither an efficient nor selective catalyst for the reduction of protein-SSG (1,4,15). The different substrate selectivities of thioltransferase and thioredoxin suggest that the two TDOR enzyme systems may contribute synergistically to sulfhydryl homeostasis.
Cadmium is a heavy metal that exhibits both acute and chronic toxicity. In cases of severe acute exposure, death can result in a few days and long term exposure to lesser amounts can result in damage to kidney, lung, and bone tissue (23). Cadmium also triggers stress-like responses in various signaling cascades (24), including modulation of the regulation of the tumor suppressor gene p53, various cytokines, and proteins like metallothionine and glutathione synthetase. Cadmium has no known beneficial effects in the human body, but it is used extensively in the production of human consumables like batteries, plastics, various pigments, and metal coatings. Although Cd ϩ2 has been implicated as a pro-oxidant, it differs from other pro-oxidant metal ions like Cu ϩ1 and Fe ϩ2 , because it does not participate in redox cycling (23). This implicates direct binding of cadmium to critical cellular components as the mechanism of toxicity, and proteins with vicinal disulfides are expected to be particularly sensitive to inactivation by cadmium (25).
Recently it was reported (26) that short term incubation of HT4 neuronal cells with cadmium led to an increase in the level of cellular protein-SSG. Because cadmium is not a direct oxidant, we hypothesized that protein-SSG accumulation was likely due to inhibition of thioltransferase, the expected catalyst of protein-SSG deglutathionylation in cells (2,4,15,19). To test this hypothesis we examined the effects of cadmium exposure on H9 and Jurkat cells, two T-cell-derived lines that have been extensively studied for their responses to oxidative stress.
Here we report that cadmium treatment of cells is associated with specific coordination to the TDOR proteins, all of which contain vicinal cysteine residues. This coordination leads to potent inhibition of the thioltransferase system in situ, and this effect corresponds to a concomitant decrease in the rate of cellular protein-SSG dethiolation. Although the enzymes of the thioredoxin system are also sensitive to cadmium inhibition in vitro, the thioredoxin system does not contribute significantly to cellular protein-SSG reduction. Together these data indicate that the thioltransferase system is the primary mechanism for homeostatic deglutathionylation of protein-SSG in cells. In addition, acute treatment of cells with cadmium in the concentration range that inhibits thioltransferase leads to apoptosis, and cells transfected with the antisense cDNA of thioltransferase do not survive selection. These data suggest that thioltransferase is necessary for cell survival and that it plays a role in initiation of apoptosis.

EXPERIMENTAL PROCEDURES
Materials-GSSG reductase (bovine and yeast) was obtained from Sigma. Thioredoxin reductase (bovine and Escherichia coli) was purchased from Sigma and American Diagonstica (Greenwich, CT). E. coli thioredoxin was obtained from Calbiochem. NADPH was obtained from Roche Molecular Biochemicals. L-Cysteinyl-glutathione disulfide was purchased from Toronto Research Chemicals, Inc. Lipoamide disulfide (DL-6,8-thioctic acid amide) was obtained from Sigma. RPMI 1640 medium, fetal bovine serum, penicillin-streptomycin mix, glutamine, trypan blue, Lipofectin, and Lipofectamine were from Life Technologies, Inc. All other chemicals were reagent grade from standard sources. Dr. David Davis (National Institutes of Health) generously donated the Jurkat cells (clone E6 -1), and the H9 cells and K562 cells were obtained from the American Type Culture Collection. pRep4 and pRep10 expression vectors were from Invitrogen. Sephadex G-75 gel filtration resin was obtained from Amersham Pharmacia Biotech. The micro BCA protein assay and Slide-A-Lyzer dialysis cassettes were obtained from Pierce. YM 30 microconcentrator membranes were from Millipore. Tran 35  Enzymes-Recombinant human thioltransferases (wild type and mutant) were expressed and purified from E. coli as described previously (5,27). The thioltransferase enzymes and bovine GSSG reductase were diluted in PGN buffer (100 mM sodium/potassium phosphate, pH 7.5, 0.5 mM GSH, and 0.2 mM NADPH) for inhibition experiments. Purified bovine thioredoxin was obtained via stepwise chromatography as described previously (28); frozen aliquots of the partially purified enzyme were thawed and eluted from the final DEAE column with PGN buffer containing 0.5 M NaCl. To remove EDTA contained in the manufacturer's buffer system, aliquots of the stock solution of bovine thioredoxin reductase were diluted in PGN buffer to 0.09 mg/ml and reconcentrated with a microconcentrator with a YM 30 membrane (Millipore). This procedure was repeated twice. The final concentrate was diluted to 0.9 mg/ml in PGN buffer.
Preparation of BSA-SSG[ 35 S]-BSA-SSG[ 35 S] was prepared as described previously (3), with the following modifications. After the reaction of S-carboxymethyl-BSA with N-succinimidyl pyridyl bis(3,3Ј-dithiopropionate) was quenched with glycine, BSA was separated from small molecules by dialysis against 100 mM sodium phosphate, pH 7.0, overnight with two changes of buffer. The modified BSA was then treated with 4 mM [ 35 S]GSH for 1 h at room temperature. The resulting BSA-SSG[ 35 S] product was separated from [ 35 S]GSH as described previously, and it typically had Ն0.9 GS-equivalent/mol of BSA.
Cell Exposure to Cadmium-Cells in log phase growth (6 ϫ 10 5 to 1 ϫ 10 6 , within 48 h of passage) were centrifuged and resuspended in phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 4.3 mM sodium phosphate, 1.4 mM potassium phosphate, pH 7.4) to a density of 7 ϫ 10 5 cells/ml. These cells were treated by adding 100-l aliquots of concentrated cadmium solutions in deionized water to 5 ml of the cells in PBS, so that final cadmium concentrations were between 0 and 100 M. Controls (100 l of H 2 O only) were performed in parallel. The cells were incubated at 37°C at 5% CO 2 for 30 min. Cadmium chloride and cadmium acetate gave equivalent results.
Cell Lysates-Cells (which settled during treatment) were resuspended in a minimal amount (usually 1 ml) of one of the various treatment solutions and centrifuged, and the resulting supernatant was removed by suction. Cell pellets were resuspended in lysis buffer, pH 7.4 (adapted from Ref. 29) (50 mM Tris-HCl, 2.2 mM sodium phosphate, 0.7 mM potassium phosphate, 0.75% Triton X-100 (v/v), 369 mM NaCl, 1.35 mM KCl, 2 l/ml dimethyl sulfoxide, 10 g/ml leupeptin, 20 g/ml aprotinin, and 25 M p-nitrophenyl, pЈ-guanidinobenzoate), in the absence or presence of 50 mM NEM. NEM was added to alkylate remaining free sulfhydryls when the extracts were being analyzed for protein-SSG. Protein content was assessed with the BCA assay using BSA as standard.
Assay of the Thiol-disulfide Oxidoreductase Activities in Cell Lysates-Lysates of control cells and cadmium-treated cells were analyzed for thioltransferase activity with the standard radiolabel assay (3), which monitors time-dependent release of radioactivity from BSA-SSG[ 35 S]. Aliquots of lysate and PGN buffer with or without yeast GSSG reductase (2 units/ml, final) were warmed separately to 30°C, mixed, and then an aliquot of BSA-SSG[ 35 S] (0.1 mM, final) was added to initiate the reaction (total volume 0.5 ml). Aliquots of the resultant reaction mixtures (30°C) were precipitated with ice-cold trichloroacetic acid (10%, final) at 0.5, 1, 2, and 3 min. After centrifugation, the supernatants were analyzed for 35 S by scintillation counting with Ն97% efficiency (cpm). The total rates of deglutathionylation (slopes of [ 35 S]GSH released versus time) were corrected for nonenzymatic deglutathionylation by subtracting the rate of [ 35 S]GSH released by GSH in the absence of cell lysate. The resultant enzymatic rates were expressed as nanomoles of product/min/mg of cellular protein.
Cellular thioredoxin activity was assayed using the radiolabel assay described for thioltransferase with the following modifications: GSH was omitted, GSSG reductase was either omitted or replaced with thioredoxin reductase (3.6 g/ml, final), and time points were taken at 0.25, 15, 30, and 60 min. Addition of 5 mM EDTA to the reactions did not alter the results.
Standard spectrophotometric assays were used to determine the cellular activities of GSSG reductase (30) and thioredoxin reductase (31). Lysates (30 l) and PGN buffer (for GSSG reductase) or PGN buffer without GSH (for thioredoxin reductase) were warmed separately to 30°C (5 min). The two components were mixed, and the reactions were initiated with substrate (1 mM final for GSSG or 5,5Јdithiobis(nitrobenzoic acid), respectively). GSSG reductase activity was monitored for 5 min at 340 nm, and thioredoxin reductase activity was followed for 8 min at 405 nm in separate assays. The rates are presented as nanomoles of product/min/mg of cell protein. Although typical procedures include EDTA in the assay mixtures, we purposely omitted EDTA to avoid competitive chelation of cadmium. Instead buffers were routinely pretreated with Chelex (0.1 g/ml) that had been washed thoroughly with deionized water (18-megohm resistance).
In Vitro Inhibition of the Thiol-disulfide Oxidoreductase Enzymes by Cadmium-Cadmium-mediated inhibition of the TDOR enzymes was tested in PGN buffer treated with Chelex (as above). GSH was included in all of the inhibition reactions, because it is naturally present in the intracellular environment. Because 5,5Ј-dithiobis(nitrobenzoic acid) reacts with GSH immediately, thioredoxin reductase activity was assayed with lipoamide, a known substrate for thioredoxin reductase (31) that is not rapidly reduced by GSH.
All four TDOR enzymes were tested similarly for inhibition by cadmium. Concentrated PGN buffer, containing the respective enzymes, and stock solutions of cadmium in water (or water alone) were prewarmed (5 min) separately to 30°C. Then 30 l of the cadmium solution (or water) was added to the enzyme solutions to achieve final cadmium concentrations of 0 -200 M. The reactions were immediately initiated with the appropriate substrate (final volume 200 l). Each enzyme was tested with seven concentrations of cadmium (0.1-200 M) encompassing the full range of inhibition for each enzyme.
Activity of thioltransferase (wild type or mutant) in the absence or presence of cadmium was assayed using both spectrophotometric and radiolabel assays (3). The spectrophotometric assay was adapted for use with a 96-well plate reader (Thermomax, Molecular Devices), and yeast GSSG reductase was used as coupling enzyme. Because the yeast GSSG reductase was inhibited at higher cadmium concentrations, separate controls (minus thioltransferase) were performed for each concentration of cadmium. In the radiolabel assay GSSG reductase was omitted to obviate this complication, and time points were taken at 0.25, 0.5, 1.0, and 1.5 min. The two assays yielded identical results.
Mammalian (bovine) GSSG reductase was analyzed for activity in the absence or presence of cadmium as described above for analysis of GSSG reductase in cell lysates.
Activity of mammalian (bovine) thioredoxin reductase in the absence or presence of cadmium was assayed using 1 mM lipoamide (stock solution 10 mM in 20% ethanol) as the substrate.
Activity of mammalian (bovine) thioredoxin in the absence or presence of cadmium was analyzed as previously (32) using insulin as the substrate, except that DTT was diminished from 2 to 0.5 mM to decrease potential cadmium chelation by DTT. To estimate the magnitude of this DTT effect, inhibition of thioredoxin reductase by cadmium was tested (as above) in the absence and presence of 0.5 mM DTT.
Cadmium-mediated Inactivation of Oxidized Thioltransferase-Thioltransferase (wild type) was pretreated with or without cysteine-SSG (50 mM) in 100 mM potassium phosphate buffer (pH 7.5) for 10 min at 30°C. Thioltransferase samples so treated were tested for sensitivity to IAA (300 M) and cadmium (20 M) in comparison to control samples run in parallel. The reaction mixtures were incubated for 10 min at 30°C, then tested for activity (see above) by dilution into PGN buffer containing yeast GSSG reductase so that the final concentrations of IAA and cadmium in the assay mixture did not exceed 9 and 0.3 M, respectively. These concentrations of IAA and cadmium were confirmed separately to have no substantial effect on the coupling enzyme yeast GSSG reductase (Ref. 33

and the present study).
Radiolabeling of Cellular Glutathione-The glutathione pool in H9 or Jurkat cells was radiolabeled by adapting a previously described procedure (4,34). Cells (3 ϫ 10 7 ) were incubated in serum free medium (10 ml) with cycloheximide (50 g/ml) at 37°C for 30 min to inhibit protein synthesis, allowing the radiolabel to be incorporated predominately into cellular GSH. The extent of inhibition of protein synthesis was calculated by comparing the amount of radiolabel incorporated in cells treated with and without cycloheximide. The incorporation of 35 S radiolabel into proteins from cycloheximide-treated cells was Ͻ2% of the control treated cells, i.e. Ͼ98% inhibition.
After treatment with cycloheximide, all but 1.4 ml of the serum free medium was removed and 1.4 ml of PBS containing cycloheximide (50 g/ml) was added, diluting the cystine and methionine contained in RPMI medium 2-fold to 100 and 50 M, respectively. Approximately 200 Ci (Ϸ20 l) of Tran 35 s-Label ([ 35 s]methionine plus [ 35 s]cysteine) was added to the cell mixture. The cells were incubated at 37°C for 4 h, centrifuged, washed three times, and then diluted to 7 ϫ 10 5 cells/ml in PBS for cadmium or control treatment (30 min), as described above. The cells were sometimes treated with H 2 O 2 and/or replete medium before cell lysates were prepared, as described above. Lysates were dialyzed repeatedly against PBS in Slide-A-Lyzer cassettes with a molecular mass cutoff of 10,000 Da until the radioactivity in the dialysate was less than 200 cpm/ml (typically accomplished with three buffer changes over an 8-to 12-h period).
The use of 100 Ci of [ 35 S]cysteine (instead of Tran 35 S-Label) resulted in a higher percentage of DTT-releasable radiolabel (see definition below) associated with cellular proteins. This improved the ratio of disulfide-associated [ 35 S]GSH to total protein radioactivity without al-tering any of the experimental results (i.e. improved signal-to-background ratio).
Analysis of Protein-SSG Content-After dialysis to remove excess [ 35 S]GSH, equal amounts of cell protein were added to two reaction mixtures, one with 10 mM DTT in water and the other with water alone. These reaction mixtures were incubated at 37°C for 30 min and then precipitated with ice-cold 10% trichloroacetic acid. The supernatants were analyzed for released radioactivity. The percentage of DTT-releasable radiolabel (which corresponds to [ 35 S]GSH) was calculated by subtracting the radioactivity in the supernatants of samples treated without DTT from the radioactivity in the supernatants of corresponding samples that were treated with DTT. This number was divided by the total amount of protein-associated radioactivity. DTT-releasable radiolabel is equated with [ 35 S]GSH based on analogy to many previous studies. Thus, under comparable conditions it was demonstrated (8,10,11,19) that protein-SSG[ 35 S] is the predominant (Ն90%) radiolabeled protein adduct formed under conditions of oxidative stress.
Cell Growth, Death, and Apoptosis-Cadmium treatment (in 5 ml of PBS) was carried out for 30 min as described above. However, prior to harvesting the cells, 4 ml of PBS was removed from the cells and 4 ml of fresh medium was added back. The cells were returned to the incubator, and cell samples (0.5 ml) were taken at 4, 8, 12, and 24 h, treated with trypan blue, and counted with a hemocytometer. Results are presented as total cell counts and live cells (percentage which excluded trypan blue). To test for apoptotic cells at 12 and 24 h, 2 ϫ 10 5 cells (in 25 l of medium) were either stained directly with acridine orange and ethidium bromide (each 100 g/ml) or with propidium iodide as described previously (35). Approximately 1.5 ϫ 10 7 cells were centrifuged, fixed with ethanol, treated with citric acid buffer (0.1 M, pH 7.8), and stained with propidium iodide. The cells were then analyzed with an EPICS XL-MCL Flow Cytometer (Coulter Corp., Miami, FL) in the Facility at Case Western Reserve University. Propidium iodide was excited at 488 nm, with an air-cooled argon ion laser operated at 15 milliwatts, and emission was detected with a 630-nm band pass optical filter. Data acquisition and determination of the apoptotic fraction were performed with XL System 2 (version 3.0) software (Coulter Corp.). DNA content signals were linearly amplified. The integrated and peak DNA signals were used for aggregate discrimination. All samples were filtered with a 53-m mesh prior to analysis. Regular cell cycle distribution was observed in control cells, but in the treated samples a sub G 0 /G 1 peak was observed, indicative of apoptosis (36).
Transfection-The hygromycin-resistant, constitutively active expression vectors pRep4 and pRep10 were modified by insertion of thioltransferase cDNA (27) into the multicloning region (identical but reversed in the two plasmids) between the BamHI and HindIII sites to construct sense (pRep 10) and antisense (pRep 4) plasmids. Transfection was accomplished with Lipofectamine using manufacturer's protocols, and cells were photographed at 6 weeks. K562 cells transfected with sense cDNA were selected for 3 weeks with hygromycin, allowed to expand for 1 week, and then frozen in liquid nitrogen at 2 ϫ 10 6 cells/ml. Clones were obtained from these frozen stocks by growing them to 5 ϫ 10 4 cells/ml and then serially diluting back to 2.5 cells/ml in medium with 300 units/ml hygromycin. This cell mixture was distributed into a 96-well plate (300 l per well). Two days later, wells with a single cell (observed by light microscopy) were monitored daily and maintained in the presence of medium until they were passed into larger flasks. A number of these clones were analyzed for thioltransferase activity and content (Western blot analysis).
Statistical Analysis-All experiments are presented as mean Ϯ S.E. Where error bars are not visible on the figures, they are contained within the size of the data symbol. Where applicable, Student's twotailed t test was used to determine statistically significant differences (WINKS 4.5 program, TexaSoft, Cedar Hill, TX).

Cellular Activity of Thioltransferase and Thioredoxin-Ly-
sates of H9 and Jurkat cells were analyzed for deglutathionylase activity of endogenous thioltransferase and thioredoxin ( Table I). The thioltransferase activity (GSH-dependent dethiolase activity) in both cell lines was substantial (Table I, left column). In contrast, little if any thioredoxin-mediated protein-SSG dethiolase activity (i.e. GSH-independent) was detectable even with extended 60-min reaction times, addition of 5 mM EDTA, or addition of excess thioredoxin reductase (Table I, right column).
The lack of thioredoxin-mediated deglutathionylase activity in lysates is consistent with previous reports of much lower catalytic activity of thioredoxin for protein-SSG substrates when compared to thioltransferase (1,4,15). A full comparative kinetic analysis of the two enzyme systems has not been reported, so we examined the relative catalytic efficiencies (k cat /K m ) of these enzymes. The best prototype substrate for this comparison is cysteine-SSG, because it represents the commonality of all protein-SSG substrates without the steric constraints imparted by location of the cysteine-SSG moiety in the respective proteins. Accordingly, thioltransferase was 5000 times more efficient (k cat /K m ) than thioredoxin in catalyzing deglutathionylation of cysteine-SSG (Table II). Inactivation of Cellular Thioltransferase by Cadmium-To test whether cellular thioltransferase was inactivated, H9 and Jurkat cells were treated with cadmium (0 -100 M) for 30 min in PBS. Cell-derived thioltransferase activity was lost in a dose-dependent manner as reflected by the diminution in the rate of reduction of BSA-SSG[ 35 S] as a function of cadmium concentration (Fig. 1A). This inactivation was irreversible, because removal of cadmium from the cell lysates by overnight dialysis at 4°C did not change the percentage activity remaining in the cadmium-treated samples relative to control samples dialyzed in parallel (data not shown). Cellular GSSG reductase was also inhibited by cadmium over the same concentration range as thioltransferase (Fig. 1B). The lack of observable cellular thioredoxin-mediated reduction of protein-SSG (Table  I) indicates that the effect of cadmium on deglutathionylation can be ascribed exclusively to the thioltransferase system.
To investigate the effect of cadmium on thioltransferasemediated catalysis of reduction of endogenous protein-SSG in situ, cells were pretreated with 0.5 mM H 2 O 2 in PBS to maximize intracellular protein-SSG formation. 2 After replacement of the H 2 O 2 -containing PBS with complete medium, the initial rates (Յ25% substrate depletion) of deglutathionylation of endogenous protein-SSG in control (0 -0.5 min) and cadmiumtreated cells (0 -2 min) were found to be Ն0.47/min and 0.12/ min, respectively (Fig. 2). This difference corresponds to a 75% net loss of protein-SSG deglutathionylation capacity after exposure of cells to cadmium plus H 2 O 2 compared with H 2 O 2 alone, however, the control rate may be an underestimate, because it is defined only by the earliest point that could be measured.
The loss of in situ protein-SSG deglutathionylation activity (Fig. 2) is consistent with cadmium-mediated inactivation of cellular thioltransferase activity measured in vitro (Fig. 1A). The 75% inactivation (intact cells) was somewhat different from the 90% inactivation (measured in vitro), but this apparent discrepancy can be reconciled (see "Discussion").
Cadmium Affects the Accumulation of Protein-SSG-We tested whether cadmium treatment of H9 and Jurkat cells would result in accumulation of cellular protein-SSG as observed in mouse neuronal cells (HT4) (26,37). Cadmium treatment alone did not lead to a substantial change in detectable protein-SSG in H9 or Jurkat cells (data not shown), suggesting less basal oxidative stress in H9 and Jurkat cells compared with HT4 cells. Nevertheless, cadmium treatment did inacti-  35 S] deglutathionylation by cell lysates Cell lysates (prepared as described) were added to assay mixtures containing 100 mM K phosphate, 0.2 mM NADPH and the additives necessary to test for activity of thioltransferase and thioredoxin (see "Experimental Procedures"). Rates are presented as nmol/min/mg cellular protein (mean Ϯ SE). The left column shows thioltransferase activity (GSH-dependent) measured for 3 min, while the right column presents thioredoxin activity (GSH-independent) measured for 60 min. The background rates that occurred with complete reaction mixtures but no cell lysate were subtracted (background rates comprised approximately 45% and 7% of the raw rates for the GSH-dependent and GSH-independent reactions, respectively).
b Assayed in K phosphate buffer with NADPH and thioredoxin reductase.

TABLE II
Relative catalytic efficiencies of thioltransferase and thioredoxin with the prototype substrate cysteine-SSG The two enzymes were assayed in K phosphate buffer pH ϭ 7.5 with NADPH and GSH and 2 units/ml GSSG reductase (human thioltransferase), or with NADPH and 90 nM thioredoxin reductase (E. Coli thioredoxin). The activities were measured with various concentrations of cysteine-SSG (2-1000 M), and the non-enzymatic rates were subtracted. The substrate concentration dependence of the initial rates were plotted as velocity vs. [cysteine-SSG], and the data were fit to a rectangular hyperbolic relationship by non-linear least squares analysis to obtain the respective values for k cat and K M . At least three determinations of initial rates were measured for each concentration of cysteine-SSG. Lysates were prepared and assayed for protein content and thioltransferase activity as described under "Experimental Procedures." Each experiment was repeated at least seven times, and the activities are presented as nanomoles/min/mg (mean Ϯ S.E.). B, inactivation of cellular GSSG reductase activity by cadmium treatment: H9 cells (dotted bars) and Jurkat cells (solid bars) in PBS were treated with various concentrations of cadmium, and the resulting lysates were analyzed for protein content and GSSG reductase activity as described under "Experimental Procedures." Each experiment was repeated at least three times, and the results are presented as nanomoles/min/mg (mean Ϯ S.E.). vate thioltransferase (Fig. 1A), which is responsible for essentially all of the cellular deglutathionylase activity (Table I). Therefore, under oxidative stress imposed by H 2 O 2 treatment, cadmium pretreatment would be expected to augment protein-SSG accumulation in H9 or Jurkat cells. Indeed, H 2 O 2 -induced protein-SSG accumulation was greater in cadmium-pretreated cells (Fig. 3A).
We then tested whether cadmium pretreatment altered the pattern of protein-SSG formation induced by H 2 O 2 . In fact, cells treated with cadmium and H 2 O 2 gave a very similar array of protein-SSG[ 35 S] bands on SDS-polyacrylamide gel electrophoresis as did the cells treated with H 2 O 2 alone (Fig. 3B). This result indicates that cadmium was not an effective competitor for the majority of glutathionylation sites on thiol-containing proteins. This observation was confirmed by densitometric measurements of the entire lanes (see Fig. 3 legend), and it is consistent with the affinity of cadmium for vicinal thiols versus isolated thiols that serve as glutathionylation sites (see below). The apparent molecular masses of the predominant glutathionylated proteins in H9 cells after maximal H 2 O 2 treatment (Ϯ cadmium) were 97, 64, and 45 kDa, and a 23-kDa doublet.
Relative Sensitivity of the Thiol-disulfide Oxidoreductase Enzymes to Cadmium Inhibition in Vitro-Because all of the TDOR enzymes contain vicinal thiols at their active sites, we determined their relative sensitivities to inhibition by cadmium in vitro. GSH was included in all of the reaction mixtures to mimic intracellular thiol status. Among the four TDOR enzymes tested, thioredoxin reductase was the most sensitive (Fig. 4, IC 50 Ϸ 0.2 M). Thioltransferase and GSSG reductase both displayed IC 50 values near 1 M. The apparent IC 50 value for thioredoxin was 40 M in the presence of 0.5 mM DTT (which is necessary for turnover of thioredoxin in the absence of thioredoxin reductase). To test the potential effect of DTT on cadmium inhibitory potency, we retested cadmium inhibition of thioredoxin reductase in the presence of 0.5 mM DTT, and found the cadmium concentration-response relationship shifted to the right by about a factor of 5 (data not shown). Accordingly, the IC 50 for inhibition of thioredoxin in the absence of DTT can be estimated at 8 M, as depicted by the solid line in Fig. 4.
To test further the supposed vicinal thiol chelation mechanism of cadmium inactivation/inhibition, the cadmium sensitivity of unaltered thioltransferase was compared with two altered forms; namely, (a) oxidized thioltransferase, where the active site vicinal thiols are converted to an intramolecular disulfide (C22-SS-C25), and (b) mutant thioltransferase (C7S, C25S, C78S, C82S) where the active site vicinal thiols are now a thiol (Cys 22 ) and a hydroxyl (Ser 25 ) (5). Pretreatment of wild type thioltransferase with disulfides converts the active site to C22-SS-C25 and protects the enzyme from alkylative inactivation by iodoacetic acid (1,38). Therefore, thioltransferase was preincubated in the absence or presence of cysteine-SSG, followed by treatment with no agent (control), IAA, or cadmium. The oxidized thioltransferase retained full activity, whereas the reduced (dithiol) thioltransferase was inactivated by IAA and cadmium consistent with the mechanism of vicinal thiol chelation for cadmium-mediated inactivation/inhibition (Table  III). The mutant thioltransferase was 10-fold less sensitive to cadmium-mediated inhibition (IC 50 Ϸ 10 M) compared with wild type (1 M). The residual inhibitory effect of cadmium on the mutant thioltransferase is likely due to the conservative substitution of cysteine with serine, which retains some ability to coordinate with cadmium to form a chelate complex. In fact, this inhibition was fully reversed upon dilution, in contrast to the wild type enzyme (Table III), i.e. the mutant without vicinal thiols was insensitive to irreversible inactivation.
Effects of Cadmium Treatment on Cell Viability-Cellular toxicity, apoptosis, and necrosis have previously been reported as consequences of cadmium treatment, but these responses have usually been determined after exposures much longer than those used in the present study. Therefore, we measured the effect of acute cadmium exposure on cell viability. A 30-min cadmium treatment followed by replacement of the medium resulted in inhibition of cell growth followed by cell death as measured by trypan blue exclusion (Fig. 5). Death of cadmiumtreated cells compared with control cells was evident at 12 h for both cell lines. However, H9 cells (Fig. 5, left panels) were more sensitive to cadmium than were Jurkat cells (Fig. 5, right  panels).
Whether apoptosis accounted for cadmium-mediated cell death was determined using three different methods. Nuclear staining with acridine orange and ethidium bromide showed characteristic apoptotic morphology in both cell types, but H9 cells were more sensitive than Jurkat cells. H9 cells also showed clear DNA laddering, whereas Jurkat cells showed none (data not shown). The apoptotic response to cadmium was quantified by flow cytometry of propidium iodide-stained cells (Table IV). By all measures H9 cells were more sensitive to cadmium-mediated apoptosis and cell death than were the Jurkat cells.
Transfection of Cells with Antisense Thioltransferase cDNA Impacts Cell Survival-To manipulate the level of thioltransferase directly, cells were transfected with sense or antisense thioltransferase cDNA (Fig. 6). It was expected that derivative cell lines with various amounts of thioltransferase would be established by manipulating K562 cells (relatively low endogenous thioltransferase) and Jurkat cells (relatively high endogenous thioltransferase). Remarkably, in three separate experiments with both K562 and Jurkat cells, transfection with antisense thioltransferase cDNA was not compatible with cell survival. Cells transfected with antisense cDNA for thioltransferase did not survive selection (Fig. 6B), whereas cells transfected with vector alone (Fig. 6A) or cells transfected with sense The cells were resuspended (an aliquot was taken as the t ϭ 0) and centrifuged, and the cell pellets were resuspended in replete medium at 37°C. Then 1-ml aliquots were taken at the time points indicated and centrifuged, and the cell pellets were lysed in the presence of 50 mM NEM. These lysates were dialyzed to remove free radiolabel and then analyzed for DTTreleasable radioactivity ([ 35 S]GSH). The amount of [ 35 S]GSH released by DTT at t ϭ 0 corresponded to 50 Ϯ 2.5% and 50 Ϯ 2.3% of total protein-associated radioactivity for control and cadmium-treated cells, respectively. The data are presented as fraction of total protein-associated [ 35 S]GSH (mean Ϯ S.E., n ϭ 3) compared with what was released at t ϭ 0 (maximum release). The initial rates were calculated from the data for the 0-to 0.5-min time period for control cells and the 0-to 2-min time period for cadmium-treated cells (i.e. where substrate depletion was not greater than 25% in either case).
cDNA for thioltransferase all survived selection (Fig. 6C and data not shown). K562 cells transfected with sense cDNA for thioltransferase displayed expected increases in deglutathionylase activity (Fig. 6C), whereas Jurkat cells transfected with sense cDNA did not display the expected increase (data not shown). The collective data of Figs. 5 and 6 suggest that thioltransferase is required for cell viability (see "Discussion"). DISCUSSION Thioltransferase and thioredoxin are relatively abundant cytosolic enzymes distributed ubiquitously in nature and considered to play important roles in sulfhydryl homeostasis. The two enzymes are distinguished by different cofactors and by different substrate specificities. Thioltransferase specifically catalyzes the reduction of glutathione-containing mixed disulfides, in particular protein-SSG. Although thioredoxin can catalyze reduction of protein-SSG substrates, it is far less efficient than thioltransferase (Table II) (1,4,15,21) and would not be expected to contribute to physiological deglutathionylation. Reversal of oxidative stress-induced protein-SSG formation has been observed in numerous studies (8 -11, 19, 39, 40), but the enzymatic basis for the deglutathionylation was not addressed. In this report the thioltransferase-mediated (GSH-dependent) rate of BSA-SSG[ 35  The cells were then centrifuged and resuspended in replete medium, and 1-ml aliquots were treated with the indicated concentrations of H 2 O 2 for 10 min, and then centrifuged. The cell pellets were analyzed as described under Fig. 2. The data are presented as percentage of total protein radiolabel released by DTT (mean Ϯ S. E., n ϭ 3). B, lysates of radiolabeled H9 cells treated with 0.5 mM H 2 O 2 Ϯ cadmium, were combined and treated Ϯ DTT at equivalent protein concentrations. Equal amounts of protein (1.7 mg) were run on a 10 -20% SDS-polyacrylamide gel electrophoresis gel, transferred to a polyvinylidene difluoride membrane, exposed to film, and finally treated with Ponceau S. Left panel, densitometric scan of membrane after staining with Ponceau S. Right panel, densitometric scan of autoradiograph. Densitometry measurements of the lanes on the autoradiograph were: control ϭ 180, control with DTT ϭ 45, cadmium ϭ 161, and cadmium with DTT ϭ 34.

TABLE III
Effects of IAA and cadmium pretreatment on various forms of thioltransferase Wild type thioltransferase was pretreated with or without cysteine-SSG and then inactivation reactions were performed with IAA (300 M) or cadmium (20 M) in parallel with control reactions in K phosphate alone. Aliquots of each reaction mixture were diluted into PGN buffer and analyzed for residual activity (i.e. irreversible inactivation). Mutant thioltransferase (C7S, C25S, C78S, C82S) was not pretreated but it was inactivated in the same manner as the treated samples of wild type thioltransferase. The results are presented as nmol/min/ml Ϯ SE (n ϭ 4 for wild type, and n ϭ 3 for mutant thioltransferase). entirely attributable to cellular thioltransferase activity (Table  I). Previously we found that thioltransferase accounted for all of the hemoglobin-SSG deglutathionylase activity of human hemolysates (2,41), and Jung and Thomas (4) ascribed the major deglutathionylase activity of rat hepatocytes to thioltransferase. Thus, it seems appropriate to generalize that thioltransferase is the primary intracellular catalyst of protein-SSG deglutathionylation.
The Effect of Cadmium on the Thiol-disulfide Oxidoreductase Enzymes-In the current study, cadmium treatment of cells in a range shown previously to alter cellular functions (24) inac-tivated intracellular thioltransferase and its coupling enzyme GSSG reductase (Fig. 1), consistent with in vitro effects of cadmium on these two enzymes (Fig. 4). Our results for mammalian GSSG reductase agree with those of Muller (42) but differ from another study that reported no inhibition (43). However, the latter study included EDTA in the assay of GSSG reductase, likely abolishing the cadmium effect. Mammalian thioredoxin reductase is quite sensitive to cadmium according to our in vitro data (IC 50 Ϸ 0.2 M). This result compares well with cadmium inhibition of E. coli thioredoxin reductase (25), and it is consistent with the reported protective effect of EDTA on mammalian thioredoxin reductase (44). Our in vitro results suggest that the thioredoxin system would be inhibited at lower concentrations of cadmium in situ than those that inhibit the thioltransferase system and intracellular protein-SSG dethiolation. This analysis further supports the conclusion that only the thioltransferase system is pertinent to intracellular protein-SSG dethiolation.
The Mechanism of Cadmium-Enzyme Interaction-As suggested previously (25), the likely mechanism of cadmium-mediated inactivation/inhibition of the TDOR enzymes is coordination with the vicinal thiol groups contained in all four of these enzymes. Our current data support this interpretation in several ways: 1) oxidized thioltransferase (C22-SS-C25) was insensitive to cadmium-mediated inactivation (Table III) (Table III), and it was 10-fold less sensitive to inhibition by cadmium compared with wild type enzyme; and 3) cadmium did not alter the profile of proteins that are glutathionylated in response to H 2 O 2 treatment (Fig. 3B), consistent with a monothiol nature of these sites. Because cadmium pretreatment did not alter the profile of protein-SSGs induced by H 2 O 2 , this means that the endogenous substrates for thioltransferase were the same in control and cadmium pretreated cells. Hence the diminution of protein-SSG reduction by cadmium can be attributed unambiguously to inhibition of the thioltransferase enzyme system rather than to the alteration of substrate.
As hypothesized, the cadmium-mediated inhibition of thioltransferase activity in situ and in vitro (Figs. 1 and 4) correlated with cadmium-mediated inhibition of intracellular protein-SSG dethiolation (Fig. 2). When cells in PBS were treated with 100 M cadmium, cellular thioltransferase activity measured in vitro was decreased by 90%. When cells in PBS were treated with 100 M cadmium and 0.5 mM H 2 O 2 and then resuspended in medium, the initial rate of intracellular deglutathionylation measured in situ was decreased by 75%. This apparent difference in inhibitory effect can reasonably be attributed to an underestimate of the initial rate for the untreated cells and/or to differences in conditions of the two experiments. In particular, addition of replete medium can recruit a portion of the endogenous thioltransferase protected from cadmium inactivation by formation of the intramolecular disulfide (C22-SS-C25) in response to the oxidative H 2 O 2 treatment, a modification readily reversed by GSH. In addition, a small portion of the net deglutathionylation in situ may be due to direct action of GSH that is not inhibited by cadmium and cannot be assessed separately as it can be in the in vitro measurements. Thus the in situ experiment (Fig. 2) probably reflects an underestimation of the actual sensitivity of thioltransferase to inactivation by cadmium.
The Involvement of Cellular Cofactors in Thiol-disulfide Homeostasis-Cellular thioltransferase activity can be affected both directly (inactivation/inhibition) and indirectly (depletion of the second substrate, GSH). Others have reported that GSH is a necessary factor for cellular reduction of protein-SSG, and these results are consistent with thioltransferase being responsible for enzymatic deglutathionylation. In particular, regeneration of GSH by NADPH and GSSG reductase (40) was reported to be the rate-limiting step in erythrocyte reduction of protein-SSG, and this was linked to erythrocyte glucose 6-phosphate dehydrogenase activity. Glucose 6-phosphate dehydrogenase was reported to be inhibited partially by cadmium in hepatocytes, but it was less sensitive than GSSG reductase (42). In another context we found that, in genetically altered CHO cells where glucose-6-phosphate dehydrogenase was knocked out and NADPH generation was compromised, endogenous reduction of disulfides was also impaired (45). These altered CHO cells appeared to be devoid of reduction activity attributable to the thioredoxin system; however, activity attributable to the thioltransferase system (GSH-dependent) was diminished but not abolished (45). This distinction is likely due to the direct dependence of the thioredoxin system on NADPH for turnover, whereas the thioltransferase system can operate without NADPH so long as the supply of GSH is sufficient (both for reduction of thioltransferase and formation of glutathionyldisulfide substrates). Two mechanisms, de novo synthesis and NADPH-dependent turnover of GSSG, maintain cellular GSH. In the current study H 2 O 2 -treated cells replenished with medium ( Fig. 2) reduced accumulated protein-SSG rapidly (Ն0.47/ min, initial rate), whereas cells maintained in PBS displayed very little reduction of protein-SSG (Ͻ0.005/min, i.e. Ͻ1% of the rate in replete medium), indicative of GSH deprivation. These results reflect the rate-determining step of the thioltransferase catalytic cycle where regeneration of thioltransferase-S Ϫ from the thioltransferase-SSG intermediate requires GSH (3). Thus, cellular protein-SSG status could be altered by several different perturbations that impact on the activity of the thioltransferase system, including changes in GSH concentration, concentration of NADPH, and thioltransferase activity.
Cadmium and Cell Survival-Cadmium has been reported to cause apoptosis and necrosis depending upon the concentration and period of exposure (usually 8 h or more) (46 -48). Cadmium treatment (4 -50 M) of CEM-C12 cells, a T-cell line, in medium with 2% serum for 18 h was associated with apoptosis (46). In another report (47) T-cells, monocytes, and B-cells were all shown to undergo cell death in response to cadmium treatment (0.1 M to 10 mM) but only after 7 h or more of incubation with cadmium. The one exception was T-lymphocytes treated for only 1 h but with 10 mM cadmium, a very high concentration (47). Only 20% of the T-lymphocytes were viable 24 h later; however, apoptosis was not addressed as a mechanism of cell death. In another study CCRF-CEM, Raji, and Molt-3 cells were treated with cadmium in media containing 5% serum for 18 h, and all three cell lines underwent apoptosis at low concentrations of cadmium but underwent necrosis at higher concentrations (above 50 M) (48).
In contrast, we examined whether a more limited cadmium exposure (30 min), which inactivates cellular thioltransferase activity (Fig. 1A), would lead to cell death or apoptosis. In H9 and Jurkat cells, 30-min cadmium treatment followed by incubation in fresh medium for 12 or more additional hours resulted in decreased cell viability according to trypan blue exclusion (Fig. 5). Cadmium treatment also caused growth inhibition and inflicted irreversible damage that led to death by apoptosis in a concentration-dependent manner (Figs. 5 and 6 and Table IV). Thus, loss of thioltransferase correlates with initiation of cellular apoptosis. However, Jurkat cells were less sensitive than H9 cells to cadmium-mediated apoptosis and cell death. This difference cannot be ascribed solely to loss of thioltransferase activity, because Jurkat and H9 cells displayed similar cadmium-concentration dependent losses (Fig. 1A). The basis for different extents of apoptotic cell death in these two cell types is unknown, possibly reflecting differences in regulation of the cellular apoptotic machinery among different cell types. A report (49) published after completion of the current study also showed relatively short term (2 h) exposure to cadmium (200 M) led to apoptosis in U-937 cells. The authors suggested p38 mitogen-activated protein kinase activation as a specific event in initiation of apoptosis. Because p38 mitogenactivated protein kinase is activated by oxidative stress (50), and cadmium inhibition of thioltransferase exacerbates oxidative stress via interference with sulfhydryl homeostasis, the two studies are complementary and identify mechanisms of cadmium action that may be related sequentially and/or synergistically.
The concept of initiation of apoptosis by modulation of thiol status has been addressed previously (51)(52)(53)(54)(55). Increasing the level of GSH is associated with inhibition of apoptosis mediated by a variety of known stimulators (56 -59). Loss of GSH has also been shown to sensitize cells to apoptotic induction, but loss of GSH was not sufficient in itself to initiate apoptosis (52). These findings suggest that GSH acts to protect against the modification of sensitive protein-SH groups and further that a sulfhydryl modification is necessary to initiate apoptosis by this pathway. Caspases, cysteine proteinases implicated as mediators of apoptosis, were shown to be regulated by changes in redox status (i.e. changes in GSH or thioredoxin) (54,60), and elevated levels of cellular thioredoxin have been correlated with inhibition of apoptosis (61). We hypothesize that cellular apoptosis could be initiated by adduction of sensitive sulfhydryl groups on one or more specific proteins and that when the repair mechanisms for these proteins (thioltransferase and thioredoxin systems) are overwhelmed an apoptotic response occurs.
Thioltransferase and Cell Survival-To enable studies of the cellular function(s) of thioltransferase more directly we set out to alter the levels of thioltransferase by transfecting cells with sense and antisense thioltransferase cDNA. K562 cells transfected in the sense direction survived selection and showed increased levels of thioltransferase activity (Fig. 6C) and corresponding increases in thioltransferase protein content (data not shown). Clones derived from such transfected K562 cells maintained high levels of thioltransferase. However, after we froze the cells for storage and re-thawed them for analysis, they no longer retained high thioltransferase activity, and this led us to explore other cell types. Because the H9 and Jurkat lines had higher levels of endogenous thioltransferase, we attempted to diminish its content by transfection with antisense cDNA.
Jurkat cells (and K562 cells) transfected with antisense cDNA for thioltransferase did not survive selection, whereas the cells transfected with empty vector did survive (Fig. 6, A and B). This suggests that loss of cellular thioltransferase was incompatible with cell survival.
In a study by another research group (62) the levels of thioltransferase were increased by transfection in some MCF-7 cells, but further increases were not observed in a cell line that already had elevated amounts of endogenous thioltransferase through chemical induction. They offered the interpretation that high levels of thioltransferase might be cytotoxic. In our K562 cells increased thioltransferase activity was not harmful (Fig. 6C), and Jurkat cells transfected with thioltransferase cDNA in the sense direction survived selection but did not show increased amounts of thioltransferase (data not shown). Thus, our results parallel the findings of Meyer and Wells (62). Although high levels of thioltransferase might be toxic in some situations, an alternative explanation is that cells may have mechanisms of feedback regulation at either the protein or mRNA level that limit the steady-state amount of thioltransferase. MCF-7 cells with higher levels of thioltransferase (62,63) displayed resistance to adriamycin, a mediator of oxidative stress. We interpret the correlation between increased thioltransferase and resistance to adriamycin-induced oxidative stress as indicative of prevention of protein-SSG accumulation that could trigger apoptosis.
The process of reversible S-glutathionylation of key proteins has many potential cellular functions (2,15,18,40,64). The transmission of cellular signals is crucial to many cellular processes, including differentiation, proliferation, and apoptosis. The implication that various types of modification of sensitive sulfhydryl groups are involved in these processes suggests that both thioltransferase and thioredoxin contribute to cellular regulation. Because they have different substrate preferences, they are predicted to have different but synergistic roles. Thus, loss of both TDOR systems (as in cadmium treatment) would make a cell especially prone to apoptosis.
Conclusions-Cadmium is a toxic agent that inhibits all of the enzymes of the thioltransferase and thioredoxin systems via interaction with vicinal thiols. This allowed us to show that thioltransferase is the preponderant mechanism of protein-SSG reduction in H9 and Jurkat cells. We also showed that acute treatment of cells with cadmium initiates cell death by apoptosis. Loss of thioltransferase activity is a common factor in cadmium-and antisense-mediated cell death, suggesting that loss of thioltransferase is an event that may initiate apoptosis. Together, these results reinforce the concept that thioltransferase is a key intracellular player in the enzymatic regulation of redox-sensitive proteins.