Glutathione-dependent reductive stress triggers mitochondrial oxidation and cytotoxicity

Abstract

To investigate the effects of the predominant nonprotein thiol, glutathione (GSH), on redox homeostasis, we employed complementary pharmacological and genetic strategies to determine the consequences of both loss- and gain-of-function GSH content in vitro. We monitored the redox events in the cytosol and mitochondria using reduction-oxidation sensitive green fluorescent protein (roGFP) probes and the level of reduced/oxidized thioredoxins (Trxs). Either H₂O₂ or the Trx reductase inhibitor 1-chloro-2,4-dinitrobenzene (DNCB), in embryonic rat heart (H9c2) cells, evoked 8 or 50 mV more oxidizing glutathione redox potential, E_hc (GSSG/2GSH), respectively. In contrast, N-acetyl-l-cysteine (NAC) treatment in H9c2 cells, or overexpression of either the glutamate cysteine ligase (GCL) catalytic subunit (GCLC) or GCL modifier subunit (GCLM) in human embryonic kidney 293 T (HEK293T) cells, led to 3- to 4-fold increase of GSH and caused 7 or 12 mV more reducing E_hc, respectively. This condition paradoxically increased the level of mitochondrial oxidation, as demonstrated by redox shifts in mitochondrial roGFP and Trx2. Lastly, either NAC treatment (EC₅₀ 4 mM) or either GCLC or GCLM overexpression exhibited increased cytotoxicity and the susceptibility to the more reducing milieu was achieved at decreased levels of ROS. Taken together, our findings reveal a novel mechanism by which GSH-dependent reductive stress triggers mitochondrial oxidation and cytotoxicity.—Zhang, H., Limphong, P., Pieper, J., Liu, Q., Rodesch, C. K., Christians, E., Benjamin, I. J. Glutathione-dependent reductive stress triggers mitochondrial oxidation and cytotoxicity.

Intracellular redox homeostasis is essential for energy production through coordinately regulated mechanisms linked to key signal transduction networks and mitochondrial oxidative phosphorylation. Studies in subcellular compartments demonstrate that individual organelles have different redox requirements, principally driven by the reduced (GSH) and oxidized glutathione (GSSG) redox couple (1). The redox milieu for cytosolic proteins, comprising the enzymatic and nonenzymatic regulatory systems, is generally more reducing for reversible oxidation and reduction of protein thiols (2, 3). In the cytosol, the GSH/GSSG ratio ranges from 30:1 to 100:1, with a redox potential of −290 mV (4). In contrast, a more oxidizing environment of the endoplasmic reticulum facilitates disulfide bond formation and maturation of secretory proteins, a process termed oxidative protein folding. This compartment contains millimolar concentrations of both GSH and GSSG in which the GSH/GSSG ratio ranges between 1:1 to 3:1 to achieve its redox potential (−170 to −185 mV; ref. 4). The GSH/GSSG ratios of 20:1 to 40:1 and redox potentials approaching −250 mV to −280 mV have been reported in the matrix of mitochondria, whereas its inner membrane space is more oxidizing (1).

To control the cellular redox environment, cells contain two primary redox regulatory systems: the GSH/glutathione peroxidase/glutathione-S-transferase/glutaredoxin system and the thioredoxin (Trx)/perioxidredoxin/methionine sulfoxide reductase pathways (5, 6). Such complementary roles were shown recently from the loss of both cytosolic Trx1 and mitochondrial Trx2 in Saccharomyces cerevisiae, which resulted in elevated concentrations of GSH and GSSG and indicated a functional link between the Trx and GSH systems for handling the detoxification of reactive oxygen species (ROS; refs. 7, 8). Likewise, deletion of Trx reductase in yeast also caused elevated concentrations of GSH and GSSG (9). Although the intracellular GSH/GSSG ratio can regulate Trx function reversibly through glutathionylation (10), the first direct evidence for interdependence between redox systems came from studies of the yeast mutant for glutathione reductase (Glr1), which increases GSSG but does not affect the redox state of Trxs (11). The mechanisms for the crosstalk between the glutathione-glutaredoxin and Trx systems remain obscure.

Despite decades of studies on redox biology, the molecular and cellular mechanisms underlying reductive stress remain obscure (12). A few earlier reports have drawn attention to the deleterious effects of reductive stress in both unicellular eukaryotic and mammalian cells (9, 12–14). The most compelling recent findings for reductive stress were described in our previous study associated with dysregulation of glutathione homeostasis (increased GSH level and ratio of GSH/GSSG) and protein aggregation cardiomyopathy in experimental mice (12, 13). These studies have stimulated our interest to develop complementary model systems, mimicking experimental reductive stress, in which key molecular pathways could be identified and more readily deciphered in vitro. This study was designed primarily to determine the biochemical and molecular consequences—at the opposing ends of the redox spectrum mimicking oxidative and reductive stress—mediated by key effector pathways of glutathione homeostasis in cultured cells. Here, we have used the redox biosensors, such as reduction-oxidation sensitive green fluorescent proteins (roGFPs), Trx1, and Trx2 to validate independently that GSH depletion, after either H₂O₂ or 1-chloro-2,4-dinitrobenzene (DNCB) treatments, induced more oxidizing GSH redox potential and intracellular oxidation in embryonic rat heart H9c2 cells. For the first time, we report that either pharmacologic [i.e., N-acetyl-l-cysteine (NAC)] or genetic [i.e., glutamate cysteine ligase catalytic subunit (GCLC)/glutamate cysteine ligase modifier subunit (GCLM) overexpression] maneuvers initially triggered a more reducing GSH redox potential or reductive stress, which is followed by pathogenic mitochondrial oxidation at decreased levels of ROS in vitro. Together, we have established that alterations of glutathione homeostasis affect the redox status such that adaptations of subcellular compartments might critically determine survival and cell death fates.

MATERIALS AND METHODS

Reagents and antibodies

DNCB (C6396), NAC (A9165), N-ethylmaleimide (NEM; E3876), and trichloroacetic acid (T0699) were obtained from Sigma-Aldrich (St. Louis, MO, USA). 4-Acetoamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS; A485), dichlorofluorescein diacetate (CM-H2DCFDA; C6827) and AlamarBlue (DAL1025) were from Invitrogen (Grand Island, NY, USA). The anti-Trx1 and anti-Trx2 antibodies were obtained from Cell Signaling Technology (Danvers, MA, USA) and Ab Frontier (Seoul, Korea), respectively. Likewise, the anti-mouse GFP and anti-GSH antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Virogen (Watertown, MA, USA), respectively. Pcmv6-XL4-GCLC and Pcmv6-XL4-GCLM plasmids were purchased from Origene (Rockville, MD, USA). Cytosolic roGFP (cyto-roGFP) and mitochondrial roGFP (mito-roGFP) were gifts from Dr. S. James Remington (University of Oregon, Eugene, OR, USA).

Cell culture and transfection

H9c2 cells and human embryonic kidney 293 T (HEK293T) cells were grown in DMEM (Invitrogen) supplemented with 10% fetal calf serum (FCS; Invitrogen) and penicillin/streptomycin (Invitrogen) in a 5% CO₂ humidified atmosphere. At 1 d before transfection, cells were trypsinized and plated (1×10⁴cells/mm²) in growth medium without antibiotics to achieve 80% confluence. All transfections were carried out using Lipofectamine 2000 (Invitrogen) with 2 or 4 μg of plasmids in 6-cm dishes. To test for transgene expression, the transfected cells were incubated for 48 h in a CO₂ incubator before harvesting.

GSH and GSSG measurement

The GSH and GSSG levels were measured with a kit from Cayman, (Ann Arbor, MI, USA), which employs a spectrophotometeric glutathione reductase recycling assay. Briefly, cells were washed twice with chilled PBS and scraped into a cold buffer containing 0.2 M 2-(N-morpholino) ethanesulphonic acid, 50 mM phosphate, and 1 mM EDTA (pH 6.0), then sonicated on ice for 20 s and centrifuged at 10,000 g for 15 min at 4°C. The supernatants were removed for analysis according to the manufacturer's instruction. All the determinations were normalized to protein content determined by BCA protein assay kit. The absorbance was recorded at 405 nm using a plate reader at 5-min intervals for 30 min.

Redox potential calculation

GSH and GSSG levels have been used previously to estimate the redox potential of a cell (15). Using the measured amounts of GSH and GSSG while assuming an H9c2 cell volume of 2 pl (16), intracellular absolute concentrations ([GSH] and [GSSG]) were estimated and entered into the Nernst potential equation for the glutathione redox potential half-reaction:

Investigation of redox status in subcellular compartments using roGFP

H9c2 cells were transiently transfected with cyto-roGFP and mito-roGFP using Lipofectamine 2000. After 48 h of incubation in culture medium, cells were imaged on an Olympus automated microscope controlled by Metamorph 6 software (Olympus, Tokyo, Japan). Dual-excitation ratio imaging used excitation filters 405 and 488 nm, with emission at 515 nm. Exposure time was 200–1000 ms. Raw data were exported to ImageJ software (U.S. National Institutes of Health, Bethesda, MD, USA) for analysis. Fluorescence images were background-corrected by manual selection of background regions.

AMS-alkylated redox Western blot

Cells were treated with reagents for 0–30 min and were washed twice with ice-cold PBS. Cells were precipitated with ice-cold trichloroacetic acid (10%) for 30 min at 4°C. Samples were centrifuged at 12,000 g for 10 min and washed twice with 100% acetone. The protein pellets were dissolved in nonreducing buffer (100 mM Tris-HCl, pH 6.8; 2% SDS; and 25 mM AMS). Cyto-roGFP redox forms were separated using 15% nonreducing SDS- PAGE. Trx2 redox forms were separated using 17% nonreducing SDS- PAGE. The gel was electroblotted onto a polyvinylidene fluoride membrane and probed with anti-GFP or anti-Trx2 antibody. Bands corresponding to cyto-roGFP or Trx2 were visualized using ECL. Bands were quantified using ImageJ software.

NEM-alkylated redox Western blot

For mito-roGFP and Trx1 redox analysis, cells were washed twice with ice-cold PBS immediately after treatment. Cells were precipitated with chilled trichloroacetic acid (10%) for 30 min at 4°C. Samples were centrifuged at 12,000 g for 10 min and washed twice with 100% acetone. The protein pellets were dissolved in nonreducing buffer (100 mM Tris-HCl, pH 6.8; 2% SDS; and 40 mM NEM). Mito-roGFP redox forms were separated using 15% nonreducing SDS-PAGE. Trx1 redox forms were separated using 17% nonreducing SDS- PAGE.

ROS measurement

ROS production was measured with the cell-permeable probe CM-H2DCFDA (C6827; Molecular Probes; Invitrogen) as described previously (17). Cells were plated 24 h before assay in a 96-well plate (8×10³/well). The CM-H2DCFDA dye was loaded by incubation at a concentration of 10 μM for 30 min at 37°C. After the incubation, cells were washed twice with PBS and treated with different reagents. Fluorescence was quantified using a fluorescence microplate reader (Packard Fluorocount; GMI, Albertville, MN, USA) with excitation at 485 nm and emission at 530 nm.

Cell viability

Cell viability was determined using the nontoxic AlamarBlue (Invitrogen, DAL1025). At 1 d before treatment, 8 × 10³ H9c2 cells in 100 μl growth medium per well were plated into 96-well plate. After subconfluent exponentially growing H9c2 cells were incubated with H₂O₂, BSO, DNCB, or NAC for the indicated times, 0.1 vol of AlamarBlue reagent was added directly to cells in culture medium 2 h before reading fluorescence with excitation at 540 ± 35 nm and emission at 600 ± 40 nm by using an Flx 800 plate reader (BioTek, Winooski, VT, USA). Percentage survival of control was quantified by using the manufacturer's protocol.

Data analysis

All the experiments were repeated 3 times. For Western blots, one representative image is shown in figures. Values are represented as means ± sd. Student's t test was used for statistical analysis. Statistical significance was set at P < 0.05.

RESULTS

H₂O₂-dependent ROS induces oxidizing GSH redox potential and compartmental oxidation

To determine how acute oxidant treatment affects GSH redox couple and redox homeostasis, H9c2 cells were treated with 0.4 mM H₂O_2. This treatment modified neither the total amount of GSH nor the reduced GSH level (Fig. 1A) but decreased the ratio of GSH/GSSG (30–44%) due to significantly increased GSSG levels (Fig. 1B, C). From these values, we found by the Nernst equation that the glutathione redox potential E_hc (GSSG/2GSH; −211±1.4 mV) was 8 mV higher or more oxidized compared with the control (−219±0.9 mV; P<0.05; Fig. 1D).

Figure 1.

Effects of hydrogen peroxide (H₂O₂) treatment cause a more oxidizing GSH redox potential in H9c2 cells. A, B) H9c2 cells were treated with 0.4 mM H₂O₂ at indicated time points, and intracellular total GSH (A), reduced GSH (A), and oxidized GSH (B) levels were determined by glutathione reductase recycling assay, as described in Materials and Methods. C) GSH/GSSG ratio was calculated using reduced and oxidized GSH concentration. D) Glutathione redox potential was calculated using the Nernst equation. Values represent means ± sd (n=3). *P < 0.01, ^#P < 0.05 vs. control.

To assess the redox state in cytoplasm and mitochondria directly, we used roGFP, whose formation of an engineered disulfide bond alters its fluorescent properties at two different excitation wavelengths, to dynamically monitor subcellular redox status of GSH/GSSG within living cells (18, 19). In H9c2 cells, we transiently expressed either cyto-roGFP or mito-roGFP and periodically monitored roGFP using fluorescence microscopy (Supplemental Figs. S1 and S2). Excitation ratio (405 and 488 nm) for cyto-roGFP and mito-roGFP was markedly increased within 5 min after treatment with 0.4 mM H₂O₂ and reached a plateau by 20 min (Fig. 2A), which are similar to published reports by Dooley et al. (18) and Hanson et al. (19).

Figure 2.

Effects of H₂O₂ treatment shift the cytosol and mitochondria to a more oxidizing state in H9c2 cells. A) Cytosolic roGFP and mitochondrial roGFP were transfected transiently into H9c2 cells using Lipofectamine 2000. After 48 h, cells were imaged dynamically in the presence of 0.4 mM H₂O₂. Dual-excitation ratio imaging used excitation filters (405 and 488 nm). Ratiometric response was measured from individual H9c2 cells. Values represent means ± sd (n=50). B) Redox Western blot of cytosolic and mitochondrial proteins were performed as described in Materials and Methods using anti-GFP, anti-Trx1, and anti-Trx2 antibodies. For reducing SDS-PAGE, 5% dithiothreitol was added to the samples. C) Reduced and oxidized forms of cytosolic roGFP, Trx1, mitochondrial roGFP, and Trx2 were quantified using ImageJ software. Values represent means ± sd (n=3). *P < 0.05 vs. 0 min.

Redox Western blots of compartmentalized proteins, such as cyto-roGFP and mito-roGFP, or Trx1 and Trx2, are powerful tools for studies of thiol/disulfide redox state in subcellular compartments when combined with steps to circumvent experimental artifacts introduced during sample extraction (1, 20). The derivatization of the reduced cyto-roGFP with AMS increased its relative mass and slowed its migration, whereas NEM alkylation did not lead to any differential migration of reduced/oxidized cyto-roGFP (Fig. 2B and Supplemental Fig. S3A). Similar to previously reported data (14), NEM derivatization allowed the identification of reduced Trx1, which migrates faster than the oxidized form (Fig. 2B). After exogenous oxidation with H₂O_2, both the cyto-roGFP sensor and cytosolic Trx1 expectedly displayed shifts to their oxidized isoforms (Fig. 2B, C), whereas their migratory isoforms were indistinguishable after H₂O₂ treatment under reducing conditions (Fig. 2B). Regarding mitochondrial redox proteins, mito-roGFP could not be probed after AMS derivatization (Supplemental Fig. S3B). However, NEM-derivatized reduced mito-roGFP migrated as a band of the expected molecular weight under nonreducing conditions and reached barely detected levels (∼2% reduced mito-roGFP) in response to H₂O₂ (Fig. 2B, C). When run on reducing SDS-PAGE, no difference was found in the pattern of mito-roGFP before or after H₂O₂ treatment (Fig. 2B). Marked loss of mito-roGFP in response to H₂O₂ was either due to low concentration or limitations for detecting oxidized species under these assay conditions. Under nonreducing conditions, AMS-derivatized reduced Trx2 migrated more slowly than oxidized Trx2. H₂O₂ exposure resulted in a rapid decrease of reduced Trx2 (50% reduced) compared with control (98% reduced; Fig. 2B, C). Together, our results support the feasibility, sensitivity, and specificity of roGFPs and Trx1/2 as endogenous biosensors for studies of compartmental redox state in living cells.

DNCB-dependent depletion of glutathione modulates redox homeostasis

The GSH and Trx systems serve complementary roles for maintaining intracellular redox homeostasis. Trxs are effectively reduced by Trx reductases, which are enzymatic targets of the specific inhibitor, DNCB (21), a GSH-depleting agent (22). We observed that a 15-min DNCB treatment (100 μM) was sufficient to cause a dramatic 4-fold decrease in total and reduced GSH levels (Fig. 3A). This condition occurred without changes of oxidized GSSG levels (Fig. 3B), which suggests that the thiol depletion by Trx reductase inhibition is compensated by GSH reduction without cumulative effects on GSH oxidation. Accordingly, a dramatic diminution of the GSH/GSSG ratio (80–84%) results in a 50-mV greater oxidizing GSH redox potential (−167±1.2 mV) compared with the control (−217±2.1 mV; P<0.01; Fig. 3C, D). By fluorescence microscopy and redox Western blot, we similarly monitored the redox state of cyto-roGFP and mito-GFP, both of which became significantly more oxidized following DNCB treatment compared with controls (P<0.05; Fig. 4C). The redox state of Trx1 and Trx2 shifted completely to oxidized isoforms, reflecting the absence of recycling by Trx reductase after DNCB inhibition.

Figure 3.

GSH depletion by DNCB causes a more oxidizing GSH redox potential in H9c2 cells. A, B) Measurement of GSH (A) and GSSG (B) levels was performed at the indicated period after 100 μM DNCB treatment. C, D) GSH/GSSG ratio (C) and redox potential (D) were determined using reduced GSH and GSSG concentration. Values represent means ± sd (n=4). *P < 0.01 vs. control.

Figure 4.

Effects of GSH depletion by DNCB shift the cytosol and mitochondria to a more oxidizing state in H9c2 cells. A) After H9c2 cells were transfected transiently with cytosolic roGFP and mitochondrial roGFP for 48 h, the ratio of fluorescence obtained at 405 nm and 488 nm excitation wavelengths was observed in the presence of DNCB. Values represent means ± sd (n=50). B) Redox Western blot was performed to investigate redox state of cytosolic roGFP, Trx1, mitochondrial roGFP, and Trx2, as mentioned in Materials and Methods. C) Quantification of reduced form was conducted by ImageJ software. Reported values are means of 3 independent experiments. *P < 0.05 vs. 0 min.

NAC treatment increases reduced GSH but promotes oxidation in the mitochondria

Biosynthesis of glutathione in the cytosol requires the complementary and coordinated sequential actions of γ-glutamylcysteine synthetase (γ-GCS) and glutathione synthase. NAC, a precursor of glutathione, is a widely used thiol-containing antioxidant and modulator of the intracellular redox state. To test this hypothesis, we treated H9c2 cells with 4 mM NAC and determined the total intracellular pools for both oxidized and reduced glutathione. As expected, NAC treatment led to significant ∼3-fold increases in total levels of GSH, and reduced GSH and GSSG (Fig. 5A, B). Although the GSH/GSSG ratio decreased due to correspondingly more oxidized glutathione (Fig. 5C), the Nernst equation gave a 7-mV more reducing E_hc after NAC treatment (−197±1.6 vs. −190±2.6 mV; P<0.05; Fig. 5D).

Figure 5.

GSH repletion by NAC causes a GSH/GSSG shift and reduced GSH redox potential in H9c2 cells. A, B) Total GSH, reduced GSH (A), and oxidized GSH (B) levels in response to 4 mM NAC were analyzed at the indicated time. C, D) Reduced and oxidized GSH concentration was used for GSH/GSSG ratio (C) and glutathione redox potential (D). Values represent means ± sd (n=4). *P < 0.01, ^#P < 0.05 vs. control.

We then examined whether NAC-induced decrease in E_hc could modify the redox state of cytoplasm and mitochondria. We observed little change in the percentage of reduced isoform of either cyto-roGFP or Trx1 on exposure to NAC over 2 h (Fig. 6A, B). Unexpectedly, the fluorescence response of cells expressing mito-roGFP undergoes dramatic oxidizing redox changes following NAC treatment (Fig. 6A). Following NAC treatment for 60 min, the redox state of mito-roGFP is predominantly oxidized (i.e., 28% reduced) and Trx 2 correspondingly shifts from 56% reduced form to 36% reduced form after NAC exposure compared with controls (Fig. 6B, C). These findings revealed that NAC treatment results in the oxidization of two complementary redox biosensors representative of mitochondrial proteins.

Figure 6.

Effects of GSH repletion by NAC result in oxidation of mitochondrial roGFP and Trx2 in H9c2 cells. A) Cytosolic roGFP and mitochondrial roGFP were transiently transfected into H9c2 cells. After 48 h, fluorescence at excitation wavelengths 405 and 488 nm was dynamically recorded in the presence of 4 mM NAC. Ratiometric responses were plotted as shown (n=40). B) Nonreducing SDS-PAGE and immunoblot of AMS-treated or NEM-treated samples after NAC treatment were probed using anti-roGFP, anti-Trx1, and anti-Trx2. C) Quantification of reduced forms of mitochondrial roGFP and Trx2. Reported values are means of 3 independent experiments. *P < 0.05 vs. 0 min.

Overexpression of GCLC or GCLM exhibits prooxidative properties

To investigate the effects of increased GSH level further, we chose to overexpress the genes involved in GSH de novo synthesis. γ-GCS catalyzes the rate-limiting step in glutathione biosynthesis, and is a heterodimer composed of a catalytic subunit, GCLC, and its regulatory subunit, GCLM (23–26). In transfected H9c2 cells, we found that protein levels of GCLC and GCLM were increased ∼4-fold with pCMV-GCLC and pCMV-GCLM, respectively (Supplemental Fig. S4), but had negligible effects on the glutathione levels compared with cells transfected with vector alone (data not shown). In contrast, using similar approaches in HEK293T cells, we observed a ≥20-fold increase of either GCLC or GCLM proteins and significantly ∼4-fold higher GSH levels (Fig. 7A–C). This generated a reductive GSH redox potential of −224 ± 2 mV, a significantly (12 mV) lower value compared with vector-transfected control (−212±3.6 mV; P<0.05; Fig. 7E). Nonreducing Western blots for the redox biosensors Trx1 and Trx2 demonstrated that, unlike Trx1, mitochondrial Trx2 was significantly more oxidized after either GCLC or GCLM overexpression in HEK293T cells (Fig. 7F). These findings independently validate the compartment-specific prooxidative effects of reducing GSH redox potential.

Figure 7.

Effects of GCLC or GCLM overexpression cause GSH repletion and oxidation of Trx2 in HEK293T cells. PCMV-GCLC or pCMV-GCLC plasmids were transfected transiently into HEK293T cells for 48 h. A) Protein levels were determined by Western blot using anti-GCLC or anti-GCLM antibody. B–E) Total GSH (B), reduced GSH (B), oxidized GSH (C), GSH/GSSG ratio (D), and redox potential of glutathione (E) were measured as Fig. 1. F) Redox states of Trx1 and Trx2 were analyzed using redox Western blot. Values represent means ± sd (n=4). *P < 0.01 vs. vector; ^#P < 0.05 vs. control.

Increased reductive GSH redox potential decreases cell viability

To determine the molecular consequences from either pharmacologic (e.g., NAC) or genetic (i.e., GCLC, GCLM) approaches that induce a reducing GSH redox potential and mitochondrial oxidation, we assessed ROS production and cell viability of H9c2 and HEK293T cells using H₂O₂ or DNCB treatments as controls. Maneuvers such as H₂O₂ and DNCB were associated with increases in ROS production measured by CM-H2DCFDA fluorescence and decreases in cell viability, indicating a central requirement for GSH for detoxifying OH^· or H₂O₂ and cytoprotection of organelles (Fig. 8A–D). When exposed to NAC for 12 or 24 h, the viability of H9c2 cells decreased in dose-dependent manner (EC₅₀ 4 mM; Supplemental Fig. S6), despite paradoxically decreased ROS levels assessed with CM-H2DCFDA fluorescence (Fig. 8E, F). Similarly, the viability of HEK293T cells was significantly decreased with GCLC or GCLM overexpression for 24 h or 48 h, without corresponding increases in ROS production (Fig. 8G, H). To our knowledge, these complementary results from either pharmacologic or genetic studies are the first to establish causal relationship between the reductive GSH redox potential, mitochondrial oxidation, and cell viability.

Figure 8.

Monitoring of ROS production and analysis of cell viability are shown after GSH depletion or GSH repletion. A–F) H9c2 cells were loaded with 10 μM CM-H2DCFDA to measure ROS generation in the presence of H₂O₂ (A), DNCB (C), and NAC (E) for the indicated concentration and time course. Percentage of fluorescence increase compared with pretreatment was graphed as means ± sd (n=8). Cell viability of H9C2 cells was assessed by Alarm Blue after exposure to H₂O₂ (B), DNCB (D), and NAC (F) for the period indicated. G) HEK293T cells were loaded with CM-H2DCFDA and transfected with pCMV-GCLC or pCMV-GCLM plasmids for 24 or 48 h. H) pCMV-GCLC or pCMV-GCLM plasmids were transiently transfected into HEK293T cells for 24 or 48 h. Cell viability was determined by Alarm Blue. Values represent means ± sd (n=8). *P < 0.01 compared to control or vector.

DISCUSSION

In this study, we provide evidence for a causal mechanism for GSH-induced reductive stress triggering mitochondrial oxidation and cytotoxicity in cultured cells. We demonstrate that either NAC or GCLC or GCLM overexpression causes mitochondrial oxidation and cytotoxicity, such that the increased sensitivity from the more negative redox milieu can be achieved at lower ROS levels in cultured cells. These findings are significant since they provide compelling evidence for reductive stress, in the form of increased reducing GSH, which is dissociated from excess ROS per se, linked to cell death.

Although approaches for increasing GSH levels through NAC have been shown to protect against oxidative stress-induced cell death (27, 28), recent studies document that NAC can induce cell death in vascular smooth muscle cells and enhance 5 fluorouracil-induced cell death in colorectal carcinoma cell lines or hypoxia-induced apoptosis in murine embryonic fibroblasts (29–31). A previous study has reported that NAC exposure led to G1 arrest in mouse fibroblasts through the immediate response of increased levels of superoxide (32). Our data suggest that excess GSH, from either NAC treatment or overexpression of GCLC or GCLM, results in a more reductive glutathione redox potential without increased shifts in the amount of reduced redox biosensors in either the cytoplasm or mitochondria. In contrast, mito-roGFP or Trx2 was present in a more oxidizing form, indicating that the GSH mediated-reductive stress culminated with prooxidative consequences in mitochondria. Contrary to the common belief that NAC functions solely as an antioxidant, our findings underscore the importance of considering the dynamic events related to oxido-reductive signals and redox-dependent pathways.

Mounting evidence suggests that interpretation of cellular redox environment might be more challenging than previously recognized (33). Recent advances provide methods to evaluate compartmental redox state for nuclei, mitochondria, and the extracellular compartment. The redox Western blot analysis as well as organelle-targeted roGFPs allows quantification of organelle-specific redox changes in cells (1). First, we validated our roGFP probes using H₂O₂ and dithiothreitol as surrogate oxidant and reductant stimulants, as shown in Supplemental Fig. S2. Our data also showed that GSH repletion causes reduction of the glutathione redox potential, which shifts mitochondrial Trx2 to a more oxidized form. We acknowledge the possibility that our calculation of the redox potentials could be influenced by the permeability of GSH and GSSG across these subcellular membranes. However, GSH 2-oxoglutarate and dicarboxylate carriers allow the transport of GSH from cytosol into mitochondria, suggesting that the respective GSH and GSSG pools are not interchangeable (34). Such events appear to elicit dynamic and concordant responses in GSH redox potential as shown by the effects of either pharmacologic or genetic maneuvers.

Trxs are a class of small multifunctional redox-active proteins that complement the GSH system in protection against oxidative stress. Trx2 has been shown to interact with specific components of the mitochondrial respiratory chain. Trx2 also plays an important role in the control of the mitochondrial membrane potential and the protection against mitochondrial apoptosis signaling pathway (35, 36). Cells deficient in Trx2 display increased ROS and mitochondria-dependent apoptosis (37, 38). In addition, Trx2 is involved in the inhibition of apoptosis signal-regulating kinase 1 (ASK1)-mediated apoptosis through their physical interaction (39). Oxidation of Trx2 releases ASK1 and allows for the initiation of apoptosis. Oxidation of Trx2 following a high dose of NAC or overexpression of GCLC or GCLM in the present work might affect mitochondrial membrane potential or ASK1 activity and cause cytotoxicity. Our ongoing investigations are examining whether Trx2 oxidation plays a role in the NAC-induced cytotoxicity of H9c2 cells.

The precise mechanisms for mitochondrial oxidation and cytotoxicity by the reductive E_hc is presently unknown, but several factors are possible. Mitochondrial import of GSH via the 2-oxoglutarate and dicarboxylate carriers will generate the availability of GSH for mitochondrial peroxidases, thereby affecting H₂O₂ levels. It is also conceivable that the increased mitochondrial pool from GSH biosynthesis affects the electron transport chain leading to superoxide generation. Therefore, it is predicted not only that the catalysis of O₂ by mitochondrial MnSOD should generate H₂O₂ but that such types of ROS should oxidize the cell-permeable fluorescent DCFH-DA, permitting its intracellular detection. Because our studies demonstrate that DCFH-DA was decreased significantly instead of increased compared with controls, these findings indicate that levels of either superoxide or H₂O₂ production were not elevated by the reductive GSH potential. These results are significant since the lower DCFH-DA, indicating decreased pools of ROS, refutes the argument that DCFH-DA might undergo autooxidation, and thereby confound its reliability for ROS measurement under such experimental conditions (40).

A more plausible hypothesis under investigation is that glutathionylation of presently unspecified targets contributes to mitochondrial oxidation and, perhaps, cytotoxicity (41). Protein GSH-mixed disulfides are increasingly recognized as a potential mechanism of intracellular signaling, and protein glutathionylation, a major post-translation modification, is influenced by both oxidative stress and the GSH/GSSG ratio. As the GSH/GSSG ratio declined after NAC treatment, we found that NAC also induces the glutathionylation of presently uncharacterized proteins but not to the same extent as H₂O₂ treatment (Supplemental Fig. S5). Casagrande et al. (10) reported that Trx activity was abolished through the glutathionylation of Trx linked to decreased GSH/GSSG ratio. It is conceivable that the NAC-induced oxidation of Trx2 is mediated by decreased GSH/GSSG ratio and glutathionylation of Trx2.

Note added in proof:

Albrecht et al. (42) used roGFP to confirm that NAC induced mitochondrial H₂O₂ in flies.