Disruption of the Intracellular Sulfhydryl Homeostasis by Cadmium-induced Oxidative Stress Leads to Protein Thiolation and Ubiquitination in Neuronal Cells*

Cadmium is a potent cell poison known to cause oxidative stress by increasing lipid peroxidation and/or by changing intracellular glutathione levels and to affect the ubiquitin/ATP-dependent proteolytic pathway. However, the cellular mechanisms involved in cadmium toxicity are still not well understood, especially in neuronal cells. To investigate the relationship between cadmium-induced oxidative stress and the ubiquitin/ATP-dependent pathway, we treated cultures of neuronal cells with different concentrations of the metal ion. In addition to decreases in glutathione levels, we observed marked increases in protein-mixed disulfides (Pr-SSGs) after exposure of HT4 cells (a mouse neuronal cell line) or rat primary mesencephalic cultures to Cd2+. The increases in intracellular levels of Pr-SSGs were concurrent with increases in the levels of ubiquitinated proteins (Ub proteins) when the HT4 cells were subjected to lower (25 μm or less) concentrations of cadmium. However, higher concentrations of cadmium (50 μm), which were toxic, led to increases in Pr-SSGs but inhibited ubiquitination, probably reflecting inhibition of ubiquitinating enzymes. The cadmium-induced changes in Pr-SSGs and Ub proteins were not affected when more than 85% of intracellular glutathione was removed from the cells by the glutathione synthetase inhibitorl-buthionine-(S,R)-sulfoximine. However, the reducing agent dithiothreitol, which prevented the build up of Pr-SSGs in the cell, also blocked the accumulation of Ub proteins induced by cadmium. In addition, dithiothreitol blocked the effects of the higher toxic (50 μm) concentrations of cadmium on cytotoxicity and on glutathione, Pr-SSGs, and Ub proteins. Together, these results strongly suggest that changes in the levels of intracellular Pr-SSGs and ubiquitin-protein conjugates in neuronal cells are responses closely associated with the disruption of intracellular sulfhydryl homeostasis caused by cadmium-mediated oxidative stress.

One of the hallmarks of neurodegeneration is the appearance of intraneuronal inclusions consisting of ubiquitin-protein conjugates (1,2). The mechanisms generating such abnormal inclusions remain unknown. Ubiquitination of proteins occurs posttranslationally and is a complex ATP-dependent process in which ubiquitin is sequentially activated, transferred to ubiquitin-conjugating enzymes, and then ligated to protein substrates (reviewed in Ref. 3). Very often, more than one ubiquitin is attached to the target proteins, forming polyubiquitin chains (4). Ubiquitin can be removed from the ubiquitin-protein conjugates by deubiquitinating enzymes (5).
Covalent binding of ubiquitin to proteins in the cytosol and in the nucleus is frequently viewed as a means by which proteins are marked for subsequent degradation by the ubiquitin/ATPdependent proteinase, commonly known as the 26 S proteasome (reviewed in Refs. 6 and 7). In general, ubiquitinated proteins (Ub proteins) 1 do not accumulate in healthy cells. They are rapidly degraded by the 26 S proteasome (reviewed in Ref. 7). The failure to eliminate the ubiquitin-protein deposits in the degenerating neurons may result either from a malfunction of the ubiquitin/ATP-dependent proteolytic pathway or from structural changes in the protein substrates rendering them inaccessible to the proteolytic machinery. The accumulation of Ub proteins can then lead to proteotoxicity.
Oxidative stress is one of the mechanisms that contributes to structural changes or misfolding of proteins. Substantial evidence has accumulated showing that oxidative stress may play an important role in neurodegeneration (reviewed in Refs. 8 -10). The reactive oxygen species resulting from episodes of oxidative stress promote the modification of cellular proteins (11). Cells possess a protective mechanism to overcome the potentially toxic accumulation of oxidatively modified proteins, namely an increase in proteolysis (12,13). More recently, Davies and co-workers (14) demonstrated that oxidative stress in cultured liver epithelial cells led to measurable changes in intracellular proteolysis. The degradation of oxidatively modified proteins was postulated to occur via ubiquitin-independent and ATP-independent mechanisms (15,16).
On the other hand, studies with yeast showed that overexpression of the polyubiquitin gene conferred resistance to oxidative stress in cells grown by respiration (17). In addition, Taylor and co-workers (18 -20) detected significant increases in ubiquitin-protein conjugates, ubiquitin-activating and ubiquitin-conjugating enzyme activity, and intracellular proteolysis in lens epithelial cells recovering from episodes of oxidative stress induced by H 2 O 2 . Together, these studies suggest that the ubiquitin/ATP-dependent proteolytic system may play a role in the removal of oxidatively modified proteins.
To further investigate the mechanisms underlying the removal of oxidatively modified proteins in mammalian cells, we chose to induce oxidative stress with cadmium in a neuronal cell line (HT4) and in rat mesencephalic primary cultures. Cadmium is a potent cell poison known to cause oxidative stress (reviewed in Ref. 21) and to affect the ubiquitin/ATP-dependent proteolytic pathway (22,23). Our results show that the heavy metal decreased intracellular glutathione concentrations and increased the levels of proteinmixed disulfides (Pr-SSGs) and of ubiquitin-protein conjugates in a time-and concentration-dependent manner. In addition, we demonstrate that only a small pool of glutathione (less than 15% of the total) is sufficient to produce significant increases in Pr-SSGs levels in response to cadmium. Most importantly, we show that the thiolreducing agent dithiothreitol blocks the increases in Pr-SSGs and Ub protein levels produced by cadmium, indicating that one of the mechanisms responsible for cadmium toxicity is the perturbation of intracellular sulfhydryl homeostasis.

EXPERIMENTAL PROCEDURES
Cell Cultures-HT4 cells were derived from a mouse neuroblastoma cell line containing a recombinant temperature-sensitive mutant of SV40 large T antigen. When grown at 39°C (nonpermissive temperature), HT4 cells differentiate with neuronal morphology, express neuronal antigens, synthesize and secrete nerve growth factor, and express receptors for nerve growth factor (24) and for glutamate (25). The cells were maintained at 33°C in Dulbecco's modified Eagle's medium containing 5% fetal bovine serum and 100 units/ml penicillin, 100 g/ml streptomycin in 5% CO 2 . To induce differentiation, the temperature was changed to 39°C, at which the cells were kept for 3 days. Following the period of differentiation, the cells were maintained at 37°C, at which they were kept for at least 7 h prior to treatment with the heavy metal.
Cultures of embryonic rat mesencephalon were prepared as described in Mytilineou et al. (26). Briefly, on day 14 of gestation, the mesencephalon was surgically removed, and the cells were dissociated mechanically and plated at a density of ϳ100,000 cells/cm 2 on polyornithine-coated 35-mm tissue culture dishes. The feeding medium was minimum Eagle's medium supplemented with glucose (30 mM), sodium bicarbonate (44.6 mM), 10% fetal calf serum, and 10% horse serum. Experiments were run after 4 or 5 days in culture.
Glutathione Assay-Total glutathione was quantified as described previously (26) by a modification of the standard recycling assay based on the reduction of 5,5-dithiobis-(2-nitrobenzoic acid) in the presence of glutathione reductase and NADPH. The assay measures both GSH and GSSG; normally, GSSG constitutes less than 5% of the total glutathione in control cell cultures. The medium was first aspirated, and then the cells were rinsed twice with PBS and harvested in 500 l of 0.4 N perchloric acid, followed by sonication for 10 s (setting of 4; Vibra-Cell model V1A; Sonics and Materials, Danbury, CT) and centrifugation at 16,000 ϫ g for 15 min. Glutathione and protein assays were performed on the supernatant and pellet, respectively. In several experiments, GSSG was separately measured after removal of GSH with N-ethylmaleimide (26).
Assay for Protein-Mixed Disulfides-Pr-SSGs were quantified from pellets of the same samples prepared for the glutathione assays, following a modification of the procedure of Akerboom and Sies (27). The protein pellet was rinsed and sonicated twice with 0.4 N perchloric acid and resuspended in 0.01 M Tris-HCl, pH 7.5. The solution was then treated for 45 min at 41°C with 0.25% sodium borohydride at neutral pH to reduce the disulfide linkage. Excess borohydride was subsequently decomposed by acidification. Liberated GSH was then measured as described above. The liberated GSH reflected that which was previously linked to Pr-SSGs. GSH standards were assayed in a similar medium.
Cytotoxicity Studies-Cell viability was assessed by a modification of the method described in Mosmann (28). 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide at a final concentration of 0.5 mg/ml was dissolved in Dulbecco's modified Eagle's medium (with 1 g/liter of D-glucose) containing 5% fetal bovine serum and 100 units/ml penicillin, 100 g/ml streptomycin. The incubation medium was removed, and then each well of a 24-well plate was incubated with 0.5 ml of the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide solution for 1 h at 37°C in 5% C0 2 . The supernatant was then aspirated, and 1 ml of a solution of 0.04 M HCl in isopropanol was added and gently shaken to dissolve the precipitated dye. The solution was transferred into 1.5-ml microcentrifuge tubes and centrifuged at 16,000 ϫ g for 5 min, and the absorbance of the supernatant was read at 550 and 620 nm with a plate reader (ATCC model 340; SLT Laboratory Instruments, Hillsborough, NC). The results were expressed as the difference between the values obtained at the two wavelengths.
Preparation of Cell Extracts for Western Blotting-Following the indicated treatments, cell extracts were prepared as described previously (29). Proteins were separated by SDS-polyacrylamide gel electrophoresis (following the method of Laemmli (30)) on 8% gels for ubiquitin-protein conjugate detection. Identification of the Ub proteins was by Western blotting, and the antigens were visualized by a horseradish peroxidase method (Bio-Rad) utilizing the substrate 3,3Ј,5,5Ј-tetramethylbenzidine (Kirkegaard & Perry Laboratories, Gaithersburg, MD). Quantitative analysis of the immunostaining was by image analysis as described previously (31).
Protein Determination-Protein determination was by a bicinchoninic acid assay kit (Pierce) and by the method of Lowry et al. (32) using bovine serum albumin as a standard.
Statistical Analysis-Statistical comparisons were performed with the Tukey-Kramer multiple comparison test (Instat 2.0, Graphpad Software, San Diego, CA). The oxidation of GSH to GSSG may lead concomitantly to a thiolation of protein-SH groups to form protein-mixed disulfides (33). To verify this paradigm, we evaluated the effect of Cd 2ϩ on intracellular levels of protein-mixed disulfides by determining the reductive release of GSH from perchloric acidprecipitated proteins. Fig. 2 shows that Pr-SSGs is elevated by treatment with cadmium. The increase was significant even after 1 h of treatment with the highest concentration of Cd 2ϩ (50 M) and reached 3.2-, 4.0-, and 5.8-fold of control levels after 1, 2, and 4 h, respectively ( Fig. 2A). Maximum levels (6.8-fold of control) were observed after 6 h of exposure to 50 M Cd 2ϩ ( Fig. 2A). Increases in Pr-SSGs levels were also observed when the cells were treated with 25 or 10 M heavy metal up to 6 h. Significant changes in Pr-SSG levels were detected with 10 M metal ion after 4 h of treatment ( Fig. 2A). Dose-dependent studies indicated that when cells were exposed to increasing concentrations of CdSO 4 , the greatest elevation in Pr-SSGs (7.6-fold) was seen after treatment with 100 M (Fig. 2B). Due to its cytotoxic effect, higher concentrations of the divalent metal were not tested. Exposures of 1 h elicited significant increases in the levels of Pr-SSGs at concentrations of 50 M Cd 2ϩ or higher (Fig. 2B).

Cadmium Induces a Time-and Dose-dependent Decrease in
Cadmium Induces Changes in Glutathione and Protein-Mixed Disulfides in Primary Cultures of Fetal Rat Mesencephalon-To establish that the Cd 2ϩ effect was not restricted to transformed neuronal cells, such as the HT4 cell line, we studied the effect of the heavy metal on primary cultures of embryonic rat mesencephalon. As with HT4 cells, incubations for 1 h with all Cd 2ϩ concentrations tested led to no statistically significant changes in the cellular levels of glutathione in the mesencephalic cultures (Fig. 3A). However, longer (4 h) incubations with the divalent metal induced significant changes in the cellular glutathione levels. The lowest concentration of CdSO 4 tested (5 M) produced a transient increase in intracellular glutathione, but the highest concentration tested (100 M) caused a drop to 45% of control levels (Fig. 3A).
The dose-dependent elevation in Pr-SSGs induced by the heavy metal in the mesencephalic cultures (Fig. 3B) was parallel to but not as great as that detected in HT4 cells. After 4 h of treatment, the highest concentration tested (100 M) induced an 8-fold increase in Pr-SSGs in the HT4 cells as compared with controls, but only a 4-fold increase was observed in the treated mesencephalic cultures (compare Figs. 2B and 3B).
Cadmium Decreases HT4 Cell Viability-The time-dependent cytotoxicity of 10, 25, and 50 M cadmium was assessed with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay (Fig. 4). Treatment with 10 and 25 M Cd 2ϩ was not cytotoxic up to 8 and 6 h of incubation, respectively. However, there was a sharp drop in viability in cells treated for 8 h with 25 M heavy metal. Treatment with 50 M Cd 2ϩ significantly reduced cell viability even after 1 h of incubation (Fig. 4).
Cadmium Induces a Time-and Dose-dependent Change in the Levels of Ubiquitin-Protein Conjugates in HT4 Cells-The cadmium-induced thiolation of protein-SH groups leading to the formation of Pr-SSGs may provoke the misfolding of proteins, which may then be targeted for degradation by the ubiquitin/ATPdependent proteolytic pathway. Therefore, we examined whether cadmium ions led to an accumulation of Ub proteins in the treated cells. As shown in Figs. 5 and 6, the levels of ubiquitin conjugates were significantly increased after 1 and 2 h of treatment with 25 and 10 M CdSO 4 , respectively. The greatest levels (4.3-fold of control) were detected after 8 h of exposure to 10 M heavy metal. The highest dose of cadmium tested (50 M) caused no detectable increases in the accumulation of Ub proteins. In fact, the levels of ubiquitin-protein conjugates in cells treated with 50 M Cd 2ϩ for 6 and 8 h were significantly below the control levels (Figs. 5 and 6). The latter immunoblot was overstained to show this decrease (Fig. 5,  right panel).
Comparison of the Effects of Glutathione Depletion and Protein Thiol Reduction on the Intracellular Changes Induced by Cadmium-Glutathione is considered the most important intracellular thiol involved in the formation of protein-mixed disulfides. Therefore, we determined whether depletion of intracellular GSH would interfere with the cadmium-induced  formation of protein-mixed disulfides. HT4 cells were incubated with 5 M glutathione synthetase inhibitor L-buthionine-(S,R)-sulfoximine (L-BSO) for 24 h followed by an additional 4 h of incubation with fresh reagent, preceding the cadmium treatment. Under these conditions, the concentrations of glutathione (20 -30 nmol/mg of protein), as well as those of Pr-SSGs (0.1-0.2 nmol of GSH equivalent/mg of protein), were reduced to 14 and 30% of control levels, respectively (Fig. 7A). In contrast, the increase in the levels of Pr-SSG resulting from treatment with cadmium was not blocked by depletion of more than 85% of glutathione (Fig. 7A). Total depletion of intracellular glutathione was never accomplished, even when the cells were treated with higher concentrations of L-BSO for longer periods of time (results not shown). Therefore, the data demonstrate that the residual 3-4 nmol of glutathione/mg of protein left in the cell are sufficient for the formation of Pr-SSGs in response to the cadmium treatment. The cadmium-induced accumulation of Ub proteins was also not affected by the glutathione deficiency (Fig. 7A).

FIG. 2. Cadmium increases Pr-SSG levels in a time-dependent (A) and dose-dependent (B) fashion in HT4 cells. Intracellular
When, in separate experiments, all of the intracellular glutathione was removed by subsequent addition of N-ethylmaleimide (60 M for 1 h), no Pr-SSGs were formed in response to cadmium (not shown).
One of the mechanisms involved in protein S-thiolation is the oxidation of sulfhydryl groups of cysteine residues in proteins by glutathione disulfide (GSSG). The protein-bound glutathione (Pr-SSG) should be reduced by dithiothreitol (DTT), an effective reducing agent useful in the study of thiol-disulfide exchange reactions. To determine whether there was an association between Pr-SSG formation and the accumulation of ubiquitinated proteins, we incubated cells with DTT (1 mM). Exposure of the neuronal cells to the reducing agent alone significantly (p Ͻ 0.05) decreased the control levels of Pr-SSGs, without affecting the control levels of glutathione and Ub proteins. However, addition of the reducing agent to cadmium- treated cells concomitantly decreased the levels of Pr-SSGs and of Ub proteins (Fig. 7B).
Prevention of the Cytotoxic Effect of Cadmium by the Thiolreducing Agent DTT-As noted earlier, incubations with 50 M cadmium significantly diminished cell viability (Fig. 4) and suppressed Ub proteins (Figs. 5 and 6). To determine whether one of the mechanisms involved in cadmium toxicity is the oxidation of protein thiols, we attempted to block the toxic effect of 50 M cadmium by treating cells with increasing concentrations of DTT.
As shown in Fig. 8, concentrations of the reducing agent up to 1 mM prevented the decreases in glutathione (Fig. 8A) and Ub proteins (Fig. 8C) and the increases in Pr-SSGs (Fig. 8B) and cytotoxicity (Fig. 8D) observed in the presence of 50 M Cd 2ϩ . Nevertheless, Pr-SSG was still increased by 37% over control under these experimental conditions.
The highest DTT concentration tested (10 mM) was as effective as 1 mM in reversing the decrement in glutathione (Fig.  8A). However, it was less effective than the lower concentration in blocking the changes in the other three parameters tested, namely Pr-SSGs, Ub proteins, and cytotoxicity, perhaps due to its inhibition of protein synthesis (34). DISCUSSION Cadmium accumulates in humans throughout their lives because of its very long half-life (35). The heavy metal is a substantial industrial and environmental pollutant that seriously injures a variety of organs, such as the brain, liver, testis, and kidneys (for a review, see Ref. 36). Recent studies demonstrated that cadmium toxicity was mediated by the oxidative damage of essential cellular macromolecules (reviewed in Ref. 21). For example, the heavy metal was shown to increase lipid peroxidation in the brain, an organ particularly sensitive to cadmium toxicity (37), and in hepatocytes and testicular Leydig cells (38, 39). In addition, Cd 2ϩ increased cellular levels of hydrogen peroxide in Leydig cells (39) and inhibited SOD in the liver and kidneys (40). Cadmium ions were also shown to cause changes in intracellular glutathione concentrations and to induce the synthesis of metallothioneins, cysteine-rich proteins that avidly bind the metal ion (reviewed in Ref. 21).
The cellular mechanisms involved in cadmium toxicity are still not well understood. The heavy metal interacts with thiol groups of proteins with a greater affinity than Zn 2ϩ and may therefore disrupt the structure of certain cellular proteins (41). In addition, Cd 2ϩ forms complexes with reduced glutathione (GSH), binding mostly to the sulfhydryl group of the cysteinyl moiety (42). The heavy metal may therefore contribute to an imbalance of the sulfhydryl homeostasis in the cell.
To test this hypothesis, we investigated the effect of cadmium on a mouse neuronal cell line (HT4 cells) and on rat mesencephalic primary cultures. Our studies are the first to show that cadmium induced a time-(up to 8 h) and dose-dependent increase in protein-mixed disulfides reflecting decreases in glutathione. Many other studies with nonneuronal cells report similar decreases in glutathione (reviewed in Ref. 21) but did not measure protein S-thiolation.
To explain our findings we propose the following mechanism (Scheme 1) mediating cadmium action.
Although Cd 2ϩ does not by itself facilitate the aerobic oxidation of GSH in solution at neutral pH, 2 it induced oxidation of GSH to GSSG within cells by a yet unidentified mechanism. Cadmium may therefore stimulate other intracellular events leading to the oxidation of GSH to GSSG, which in turn promotes the oxidation of protein thiol groups. A second possibility is that cadmium induces intracellular oxidation of GSH or protein thiols to thiyl radicals (43), which may in turn generate mixed disulfides.
The cadmium-induced elevations in Pr-SSGs were not blocked by depletion of more than 85% of glutathione by the glutathione synthetase inhibitor (L-BSO), suggesting that less than 15% of intracellular GSH can sustain significant increases in Pr-SSGs. This small cellular pool of GSH consistently failed to be depleted by the L-BSO and cadmium treatment.
The effect of cadmium could also be mediated by a third mechanism, described in Scheme 3. The divalent metal may form complexes directly with the thiol groups of proteins, leading to oxidized aggregates, such as Pr-SS-Pr (disulfide-linked proteins).

Pr-SH O ¡
Cd 2ϩ Pr-SS-Pr SCHEME 3 Both Pr-SSG and Pr-SS-Pr products were identified in cells treated with iodoacetamide, an alkylating reagent known to decrease glutathione levels and to induce oxidative stress (34).
Our study also shows that decreases in glutathione and accumulation of Pr-SSGs were detected in cadmium-treated mesencephalic cultures. These primary cultures contain neurons and glial cells, possibly explaining why the changes may not be as great as those observed in the pure neuronal HT4 cell cultures. For example, astroglia cells were shown to tolerate low levels of lead exposure, which could be toxic to neuronal cells (44).
In addition to producing increments in Pr-SSGs, we found that cadmium has a biphasic effect on the ubiquitin/ATP-dependent proteolytic pathway. Although low concentrations of Cd 2ϩ (25 M or less) increase the intracellular levels of Ub proteins, higher concentrations (50 M or more) have the opposite effect. The accumulation of Ub proteins observed in the presence of low cadmium concentrations could result from: (i) a direct inhibition of the activity of the 26 S proteasome; (ii) an overload of the ubiquitin/ATP-dependent pathway due to an increase production of structurally damaged proteins, such as the Pr-SSGs or Pr-SS-Pr; and/or (iii) inhibition of the activity of deubiquitinating enzymes. The latter are thiol isopeptidases (45), and the essential active site sulfhydryl groups may be-come oxidized in the presence of Cd 2ϩ .
Higher concentrations of cadmium (50 M) decreased the intracellular levels of Ub proteins. This result may be explained by a decline in the catalytic activities of the ubiquitinating enzymes, including ubiquitin-activating (E1) and ubiquitin-conjugating (E2) enzymes. These enzymes contain sulfhydryl groups in the active sites (reviewed in Ref. 46) and may be oxidized in the presence of high cadmium concentrations. This hypothesis is supported by recent studies demonstrating that ubiquitin-activating and ubiquitin-conjugating enzymes are inactivated by S-thiolation under conditions of oxidative stress induced by exposure to hydrogen peroxide (19,20). In addition, others have shown that in yeast, expression of the ubiquitin-conjugating enzymes and of the poly ubiquitin gene is highly increased after exposure to 100 M cadmium for 30 min and that strains defective in proteasome activity are more susceptible to cadmium toxicity (22). However, ubiquitin overexpression did not increase yeast tolerance to cadmium toxicity (47).
Two effects of cadmium discussed above, namely increases in Pr-SSGs and Ub proteins, could be reversed by the thiol-reducing agent DTT (Fig. 7). These results conclusively show that the heavy metal perturbs the thiol-disulfide redox status of intracellular proteins.
In summary, cadmium induces the loss of glutathione, the oxidation of protein thiols to Pr-SSGs, and the accumulation of ubiquitinated proteins in the neuronal cells. These effects can be reversed by a thiol-reducing agent, indicating that they result from perturbations of the thiol-disulfide redox status of intracellular proteins. The formation of Pr-SSGs appears to be linked to inactivation of the Ub/ATP-dependent pathway, which leads to the accumulation of ubiquitinated proteins. This general mechanism may reflect cellular responses to other agents that promote the modification of the structure of intracellular proteins and inhibit the Ub/ATP-dependent pathway. Failure to overcome this inhibition may result in the proteotoxic accumulation of ubiquitinated proteins in intracellular inclusions and may lead to cell degeneration.