Diphenyleneiodonium triggers the efflux of glutathione from cultured cells.

Diphenyleneiodonium (DPI) is a broad-spectrum flavoprotein inhibitor commonly used to inhibit oxidant production by the NADPH oxidase of phagocytic and nonphagocytic cells. A previous study has shown that DPI can sensitize T24 bladder carcinoma cells to Fas-mediated apoptosis. We observed DPI to deplete intracellular reduced glutathione (GSH) in T24 cells and a range of other primary and transformed cell types. The effect was immediate, with 50% loss of intracellular GSH within 2 h of treatment with DPI. The glutathione was quantitatively recovered in the extracellular medium, indicating that efflux was occurring. The loss of GSH was blocked with bromosulfophthalein, an inhibitor of the canalicular GSH transporters. We conclude that DPI induces a dramatic efflux of cellular GSH from T24 cells via a specific transport channel. This provides a potential mechanism for its proapoptotic effect, and it also has important implications for the regulation of glutathione homeostasis in cells.


Diphenyleneiodonium (DPI) is a broad-spectrum flavoprotein inhibitor commonly used to inhibit oxidant production by the NADPH oxidase of phagocytic and nonphagocytic cells.
A previous study has shown that DPI can sensitize T24 bladder carcinoma cells to Fasmediated apoptosis. We observed DPI to deplete intracellular reduced glutathione (GSH) in T24 cells and a range of other primary and transformed cell types. The effect was immediate, with 50% loss of intracellular GSH within 2 h of treatment with DPI. The glutathione was quantitatively recovered in the extracellular medium, indicating that efflux was occurring. The loss of GSH was blocked with bromosulfophthalein, an inhibitor of the canalicular GSH transporters. We conclude that DPI induces a dramatic efflux of cellular GSH from T24 cells via a specific transport channel. This provides a potential mechanism for its proapoptotic effect, and it also has important implications for the regulation of glutathione homeostasis in cells.
Diphenyleneiodonium (DPI) 1 is a flavoprotein inhibitor whose targets include the phagocyte NADPH oxidase, nitric oxide synthase, xanthine oxidase, cytochrome P450 reductase, and NADH:ubiquinone oxidoreductase (1)(2)(3)(4)(5). Electron transport through the flavin moieties of these complexes causes reduction of DPI to its radical form, followed by irreversible phenylation of either the flavin or adjacent amino acid and heme groups (6,7). Despite its nonspecific mode of action, DPI has been used extensively in recent years to block NADPH oxidase activity in nonphagocytic cells, where the resultant oxidants are proposed to play a role in cell signaling (8).
DPI has also been shown to sensitize T24 bladder carcinoma cells to Fas-mediated apoptosis (9). Fas (CD95/Apo-1) is a member of the tumor necrosis factor receptor superfamily and is widely expressed on normal and transformed cells. Binding of the receptor with the Fas ligand or an agonistic antibody triggers activation of the caspase cascade and the induction of apoptosis (reviewed in Refs. 10 and 11). Resistance to Fasmediated apoptosis has been observed during normal cell development and also in many types of tumor cells (12)(13)(14)(15)(16)(17)(18). Naïve T cells are resistant to Fas ligand but become sensitive several days after antigen activation (12). In tumor cells, down-regulation of signaling through Fas has been implicated in tumorigenesis by promoting cell survival and enabling evasion of cytotoxic T cells and in resistance to chemotherapeutic treatment (17,19,20).
Potential mechanisms of resistance explored thus far include decreased receptor levels, increased expression of decoy receptors, failure to form a functional signaling complex upon receptor oligomerization, inactive caspase mimics, and modulation of cell death by the antiapoptotic proteins of the Bcl-2 family (12,(21)(22)(23)(24)(25). The ability of DPI to sensitize cells to Fas-mediated apoptosis argues for some form of redox regulation. The observations of persistent oxidative stress in tumor cells (26 -28), combined with the sensitivity of caspases to oxidative inactivation (29), raised the possibility of a DPI-sensitive NADPH oxidase blocking apoptosis in resistant cells.
Reduced glutathione (GSH) is an important intracellular redox buffer that can be oxidized to glutathione disulfide (GSSG) and protein mixed disulfides and is in part responsible for the maintenance of protein thiols in a reduced state (30,31). Total intracellular glutathione levels are under dynamic control. There is a continual efflux of GSH from cells, followed by its extracellular degradation, uptake of cysteine-containing precursors, and subsequent resynthesis (32). Inhibition of GSH synthesis with buthionine sulfoximine depletes cells of GSH over several hours (31).
We measured GSH and GSSG levels before and after treatment with DPI and anti-Fas antibody to determine if the switch in sensitivity to Fas is associated with changes in the glutathione redox couple in T24 cells. To our surprise, we found that DPI triggered rapid GSH efflux, with Ͼ50% loss over 2 h. This provides a potential mechanism for its proapoptotic effect, and it also has important implications for the regulation of glutathione homeostasis in cells.
Cell Culture-The T24 bladder carcinoma and Jurkat T lymphocyte cell lines were obtained from American Type Culture Collection (Manassas, VA). T24 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine. Jur-* This work was supported in part by an Otago University Research Grant and by the Cancer Society of New Zealand. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
kat cells were maintained in RPMI 1640 medium containing 10% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Human umbilical vein endothelial cells were isolated from umbilical cords by collagenase digestion (33) and cultured in Medium 199 supplemented with 20% fetal bovine serum, 100 g/ml heparin, 30 g/ml endothelial cell growth factor, 25 units/ml penicillin, and 25 g/ml streptomycin on fibronectin-cultured plates (34). Fibroblasts were isolated from human infant foreskins and cultured in minimum essential medium containing 15% fetal bovine serum, 2 mM glutamine, 25 units/ml penicillin, and 25 g/ml streptomycin as described previously (35). All cell types were maintained at 37°C in a humidified atmosphere with 5% CO 2 .
Assessment of Membrane Integrity-Plasma membrane integrity was monitored using trypan blue staining. T24 cells were harvested from culture dishes using a trypsin/EDTA solution, and cells were resuspended in 0.2% trypan blue in phosphate-buffered saline and analyzed by phase-contrast microscopy. Trypan blue-positive cells were expressed as a percentage of the total number of counted cells.
Determination of Caspase Activity-The measurement of DEVD-AMC cleavage was modified from Nicholson et al. (36). After treatment, cells were harvested with trypsin/EDTA and stored as cell pellets at -80°C. Immediately before assaying, the pellets were thawed by the addition of 100 l of buffer (100 mM HEPES, 10% sucrose, 5 mM dithiothreitol (DTT), 10 Ϫ4 % Nonidet P-40, and 0.1% CHAPS at pH 7.25) containing 50 M DEVD-AMC. The rate of release of fluorescent AMC was monitored at 37°C (excitation, 370 nm; emission, 445 nm), and the amount of AMC liberated was calculated from a standard curve generated with the free compound.
Glutathione Determination-Cellular GSH and GSSG levels were measured by HPLC after derivatization with dansyl chloride as modified from Martin and White (37). In brief, the supernatant was removed, adherent cells were washed twice in phosphate-buffered saline, and lysis buffer and iodoacetate were added. The cell samples were then transferred to microcentrifuge tubes, the pH was corrected to pH 8 -8.5 with lithium hydroxide, and the samples were incubated for 30 min. An equal volume of 4 mg/ml dansyl chloride was added, and the extracts were left in the dark for 60 min. Samples were then extracted in chloroform and kept at 4°C until HPLC analysis. Cell supernatants were derivatized as described above, except that a 2ϫ concentration of lysis buffer was used. To quantitatively measure GSH in the medium, the cell supernatant was incubated with 500 M DTT for 10 min before the derivatization procedure. HPLC analysis was performed as described previously (37).

DPI Sensitizes T24 Cells to Fas-mediated Apoptosis-T24
cells are one of the many cell lines known to be resistant to Fas-induced apoptosis, despite surface expression of the receptor. As shown previously (9), pretreatment of T24 cells with the flavoprotein inhibitor DPI sensitized these cells to anti-Fas antibody (Fig. 1). Whereas the antibody or DPI alone had little effect on the cells, characteristic apoptotic changes were detectable within 6 h of a combination treatment. Cells were detaching from the plate, and extensive membrane blebbing was evident. Caspase-3-like activity was increased 6-fold above control levels, indicative of apoptosis (Fig. 1). By 24 h, Ͼ60% of these cells were unable to exclude trypan blue (Fig. 1). DPI was equally effective at causing cell death when added concurrently or 2 h after the addition of anti-Fas antibody and also when it was present for 1 h and then washed away before antibody addition (Fig. 1).
Effect of DPI on Intracellular GSH Levels-To determine whether DPI affected the redox balance of the cells, we measured intracellular GSH and GSSG levels using the dansyl chloride HPLC assay. To our surprise, we observed a rapid depletion of intracellular GSH in the DPI-treated T24 cells (Fig. 2). The loss was apparent within 30 min of adding 10 M DPI. After 2 h of treatment, GSH levels had dropped to 47 Ϯ 10% (S.D.; n ϭ 9) of the control value, and after 4 h only 29 Ϯ 6% (S.D.; n ϭ 5) of the GSH remained in the cells (Fig. 3). Concurrent with the loss in cellular GSH, GSSG levels also dropped after incubation with DPI ( Fig. 2, inset). Two h of treatment decreased T24 GSSG levels to 42 Ϯ 12% (S.D.; n ϭ 11) of control, and 4 h of treatment with DPI decreased T24 GSSG levels to 27 Ϯ 12% (S.D.; n ϭ 5). Whereas the direct GSSG/GSH ratio remained constant, GSH levels are squared when calculating the glutathione redox couple; therefore, we would predict a more oxidizing intracellular environment. Determination of the absolute reduction potential is dependent on the cell volume, but by substituting the percentage losses in GSH and GSSG into the Nernst equation (38), we calculated that DPI would cause a loss in reduction potential of 9 mV at 2 h and 16 mV at 4 h. 2 The anti-Fas antibody itself had a minimal effect on the GSH levels (Fig. 3). When DPI was present for 1 h and then washed away, and the cells were incubated for an additional hour in fresh media, GSH levels were almost identical to those in cells that had been exposed to DPI for the entire 2-h period (Fig. 3), indicating that GSH continued to decrease after DPI removal. This correlates with the ability of DPI pretreatment and removal to sensitize the cells to apoptosis (Fig. 1). Treatment with increasing concentrations of DPI caused a dose-dependent loss in cellular GSH, giving an IC 50 of ϳ3 M (Fig. 4). This compares with the 0.5 M DPI observed for inhibition of super- oxide production by the neutrophil NADPH oxidase (39).
To investigate whether the DPI-induced GSH loss is specific to T24 cells, we measured GSH levels in Jurkat T lymphocytes and in primary cultures of endothelial cells and fibroblasts after exposure to DPI. All cell types suffered a progressive loss in intracellular GSH after DPI treatment, with untreated cells either maintaining or increasing their GSH levels over the 4-h period (Fig. 5). Thus, DPI induces a rapid depletion of the glutathione pool in both primary and transformed cells, suggesting a common target in these cell types.
Mechanism of GSH Loss in DPI-treated Cells-We collected the extracellular medium of DPI-treated T24 cells to determine whether glutathione had been exported from the cells. A marked increase in the area of a new peak with a retention time of ϳ7 min was observed (Fig. 6A). Due to a large excess of cystine in the medium, any GSH present extracellularly is likely to react via thiol-disulfide exchange to form a GSHcysteine mixed disulfide (40). To confirm that the unknown peak was the mixed disulfide, the products of the reaction between GSH and cystine were also analyzed by HPLC. The major peak of this reaction eluted at the same time as the unknown peak, and when the extracellular medium was incubated with DTT before the derivatization procedure, the disulfide peak disappeared, and GSH was detected (data not shown). Quantification of extracellular levels after reduction with DTT showed that almost all of the GSH lost from the DPI-treated cells was recovered in the supernatant (Fig. 7).
It is possible that GSSG could be exported from the cells and react with any trace cysteine present to give the mixed disulfide. Addition of GSSG to DMEM produced some of the mixed disulfide, but a significant amount of the GSSG remained after 3 h (Fig. 6B). The absence of GSSG in the extracellular medium of DPI-treated cells (Fig. 6A) argues against the export of the oxidized species. The experiments were also repeated in cystine-free media to prevent conversion to the mixed disulfide, therefore trapping the exported glutathione in its original form. After 3 h of treatment with DPI, both GSH and GSSG were detected in the extracellular medium (Fig. 6C). However, the majority of the GSSG is likely to have come from oxidation of exported GSH because the addition of GSH to cystine-free medium resulted in a substantial conversion to GSSG over a similar incubation period (Fig. 6C). Therefore, although we cannot exclude some of the glutathione being exported from the cells as GSSG, the principal exported species appears to be GSH.
Because cells export and recycle GSH, extracellular accumulation could occur as a result of impaired uptake and resynthesis rather than increased efflux. To differentiate between these possibilities, we measured both intracellular and extracellular GSH in the presence of various inhibitors of GSH metabolism and compared their effects with that of DPI. Buthionine sulfoximine, an inhibitor of ␥-glutamylcysteine synthetase that catalyzes the rate-limiting step of GSH synthesis, and acivicin, an inhibitor of the enzyme ␥-glutamyl transpeptidase involved in recycling extracellular GSH, both decreased T24 GSH levels by ϳ30% over a 4-h period (Fig. 7). This was considerably slower than DPI, whereas adding both together lowered GSH to a level approaching that of DPI. However, neither caused the accumulation of extracellular GSH. Incubation of T24 cells in cystine-free medium, which will impair both new synthesis and the recycling pathways, induced a rapid drop in intracellular GSH similar to that of DPI. However, there was again no corresponding increase in GSH in the cell supernatant, indicating a different mode of action than DPI (Fig. 7). Unlike DPI, incubation with buthionine sulfoximine, acivicin, and cystinefree medium all resulted in a loss in the total GSH levels (Fig. 7). We conclude that DPI decreases intracellular GSH levels by enhancing the rate of GSH efflux rather than by inhibiting precursor uptake or biosynthesis.
GSH transport has been studied almost exclusively in hepa-tocytes, with GSH carrier systems described in both the sinusoidal and the canalicular membranes, mediating release of GSH into the blood and the bile, respectively (41)(42)(43)(44). There is some evidence that these same transporters exist in the plasma membrane of other mammalian cells (45,46). When T24 cells were incubated with DPI and L-methionine, an inhibitor of the low affinity sinusoidal GSH transporter (47,48), the extent of intracellular GSH loss was the same as that with DPI alone, suggesting that this carrier was not responsible for the observed efflux (Fig. 8). BSP prevents GSH efflux from the canalicular membrane of hepatocytes, inhibiting both the high and low affinity transport components (42,49). Addition of BSP alone to T24 cells resulted in some loss of intracellular GSH, probably because BSP is a substrate for the glutathione Stransferases (50). When cells were treated with BSP together with DPI, a concentration-dependent reduction in the DPIinduced GSH loss was observed (Fig. 8). Notably, 1 mM BSP prevented the DPI-induced GSH loss completely, maintaining intracellular GSH at the same levels as BSP alone. An analogue, DBSP, also dose-dependently reduced the DPI-mediated GSH loss, with complete inhibition occurring at 4 mM. However, it also lowered T24 GSH levels (Fig. 8). These results strongly suggest that a BSP-and DBSP-sensitive carrier, probably one of the canalicular-like transporters, is mediating the DPI-induced efflux of GSH out of T24 cells. Because BSP and DBSP lower intracellular GSH levels themselves, and they also enhance Fas-mediated apoptosis (data not shown), we were unable to use them to test whether DPI still sensitizes T24 cells to apoptosis when GSH efflux is prevented. DISCUSSION We have discovered that the flavoprotein inhibitor DPI causes a rapid loss of intracellular GSH from cultured cells, with the missing GSH recovered outside of the viable cell. This phenomenon was not cell type-specific because it was observed in all the primary and transformed cell types tested. The turnover of cellular glutathione is well characterized, and GSH efflux is an integral component of this pathway (Fig. 9) (32).
Our results indicate that the rate of GSH efflux is susceptible to modulation. Indeed, efflux can be increased to such an extent that it outpaces the capacity of the cell for resynthesis and leads to severe depletion of intracellular GSH within a few hours.
The increased GSH efflux was prevented by inhibitors of the canalicular GSH transporters, indicating a specific efflux mechanism. GSH transporters involved in extrusion of the tripeptide are poorly characterized and remain difficult to clone, and there is a general reliance on inhibitor studies to distinguish their activity (43,44). The concentration at which BSP inhibited efflux favors the low affinity canalicular transport component as being the transporter responsible for DPIinduced GSH loss. Higher concentrations are required to inhibit it compared with the high affinity form (42), although it is not even definitive whether the two are different proteins.
We do not know how DPI is triggering GSH efflux. A DPIsensitive flavoprotein may directly regulate the activity of a GSH transporter or associated regulatory proteins. Alternatively, a cell might respond to general metabolic changes upon treatment with DPI, such as an accumulation of NAD(P)H, by increasing the rate of GSH efflux. We are able to exclude two alternate mechanisms. Firstly, increased amounts of DPI did not further increase the rate of GSH loss from the cells, indicating that a direct conjugation reaction was not responsible for GSH loss. Also, DPI could have acted by inhibiting glutathione reductase, leading to the formation of GSSG and its export. We could not detect evidence of significant GSSG formation inside or outside of the cells, and the glutathione reductase inhibitor bis-chloroethyl-nitrosourea had no effect on GSH levels (data not shown).
The ability of DPI to lower intracellular GSH levels provides a feasible but as yet unproven mechanism for its ability to sensitize T24 cells to Fas-mediated apoptosis. A depletion of intracellular GSH has been described in a number of different apoptotic systems (51)(52)(53), with several studies showing that GSH loss in cells undergoing apoptosis is the result of accelerated efflux rather than depletion by oxidation (54,55). In Jurkat cells treated with anti-Fas antibody, GSH efflux occurred via a specific membrane channel, the low affinity canalicular-like transporter, and the efflux was a downstream event dependent on caspase activation (54). In contrast, others have shown that U937 cells and HepG2 cells induced to undergo apoptosis with puromycin efflux intracellular GSH, that the extrusion precedes other apoptotic changes, and that maintenance of GSH was able to delay apoptosis (55). The results are not necessarily in direct conflict. It may be that efflux of GSH is a common process in apoptotic cells but that in some cell types, the loss of GSH is required early to allow successful induction of the apoptotic program. It is also of interest that the antiapoptotic protein Bcl-2 is linked to changes in GSH metabolism in cells, in particular, a shift in the cellular redox state toward a more reduced environment (56), increased intracellular GSH, altered compartmentalization, and even modulation of GSH efflux pathways (57)(58)(59)(60).
In our model, the time course of GSH efflux is consistent with sensitization, as is the ability to remove the DPI before activation of the Fas pathway and still see GSH efflux and sensitization. At this point, however, GSH efflux is associated with rather than responsible for the sensitization of T24 cells to Fas-mediated apoptosis. A crucial test of this hypothesis is to prevent GSH efflux and see if DPI still sensitizes the cells to apoptosis. Unfortunately, the transport inhibitors also depleted intracellular GSH, and they enhanced Fas-mediated apoptosis themselves. It will be important to identify the target of DPI and elucidate the pathways involved in controlling GSH efflux, so that more selective compounds and/or molecular approaches can be used to determine the role of DPI in sensitizing the tumor cells to apoptosis. FIG. 9. Pathway of GSH metabolism. The majority of GSH in cultured cells is derived from cystine present in the culture medium. Intracellular GSH is also effluxed from the cell, where it reacts with cystine to form GSH-cysteine mixed disulfide. Exported GSH may also react with ␥-glutamyl transpeptidase and cystine to form ␥-glutamylcystine, which is transported into the cell and reduced intracellularly to form ␥-glutamyl cysteine and cysteine.