12-O-tetradecanoylphorbol-13-acetate may both potentiate and decrease the generation of apoptosis by the antileukemic agent arsenic trioxide in human promonocytic cells. Regulation by extracellular signal-regulated protein kinases and glutathione.

Arsenic trioxide (As(2)O(3)) caused apoptosis in U-937 human promonocytic cells. This effect was potentiated by the simultaneous addition of the glutathione (GSH) synthesis inhibitor DL-buthionine-(R,S)-sulfoximine or the protein kinase C activators 12-O-tetradecanoylphorbol-13-acetate (TPA) and bryostatin 1. In addition TPA decreased the intracellular GSH content, caused ERK activation, and potentiated the As(2)O(3)-provoked activation of p38 and JNK. The addition of N-acetyl-L-cysteine, the PKC inhibitor GF109203X, and the MEK/ERK inhibitors PD98059 and U0126 attenuated both apoptosis induction and GSH decrease, whereas the p38 inhibitor SB203580 and the JNK inhibitor SP600125 were ineffective. TPA also potentiated ERK activation and GSH depletion when added simultaneously to cadmium chloride (CdCl(2)) and doxorubicin. However, TPA only enhanced apoptosis in the case of CdCl(2), which is a GSH-sensitive agent, whereas it reduced the toxicity of doxorubicin and other DNA-specific drugs. Finally, preincubation for 14-24 h with TPA did not potentiate but, instead, attenuated the As(2)O(3)- and CdCl(2)-provoked apoptosis. The same result was obtained by preincubation with bryostatin 1 and other differentiation inducers. It is concluded that TPA increases the apoptotic action of As(2)O(3), an effect mediated by ERK activation and GSH depletion. However, the increase in apoptosis is only effective in non-differentiated cells.

Arsenic is a widespread environmental contaminant with mutagenic, teratogenic, and carcinogenic effects (1). Despite this, arsenicals have been used for many years as therapeutic agents. In particular, arsenic trioxide (As 2 O 3 ) has recently attracted great attention because of its capacity to cause complete remission of newly diagnosed and relapsed acute promyelocytic leukemia. In fact, at physiologically tolerable concentrations (Ͻ5 M in plasma), As 2 O 3 readily destroys acute promyelocytic leukemia cells by apoptosis, with a mechanism that involves the degradation of the promyelocytic leukemiaretinoic acid receptor␣ fusion oncoprotein, generally expressed in this type of leukemia (2). Moreover, albeit with lower efficacy, this agent causes apoptosis in other cell types, indicating that it may also be useful for the treatment of other malignancies (2)(3)(4). For this reason it is of great interest to analyze the molecular mechanisms responsible for cell death induction by arsenic along with the factors that may modulate (either potentiating or reducing) its toxicity.
One of the most relevant aspects in the regulation of cell death is the signaling of apoptosis by serine/threonine kinases, a broad category of kinases that includes among others the calcium-dependent protein kinases (PKCs) 1 and the mitogenactivated protein kinases (MAPKs) (5). 12-O-Tetradecanoylphorbol-13-acetate (TPA) is a powerful PKC activator that has been commonly reported to inhibit the generation of apoptosis by receptor activation (6 -8), growth factor deprivation (9), and different cytotoxic agents (10 -13). Nevertheless, this phorbol ester has occasionally been observed to potentiate apoptosis induction (14,15). Among the three main members that integrate the MAPK family in mammalian cells, the stress-activated protein kinase 1 (c-Jun NH 2 -terminal kinases (JNKs)) and stress-activated protein kinase 2 (p38) are generally associated to apoptosis induction. By contrast, the extracellular signal-regulated protein kinases (ERK1/2, p44/42) are generally associated to mitogenesis and as such inversely related to apoptosis (5). However, there are also some cases in which the ERKs may exert a pro-apoptotic action (16 -19). Of note, the PKC and MAPK pathways are not totally independent. For instance, PKC activates the MEK/ERK pathway via Raf-1 phosphorylation (20,21), which might account at least in part for the capacity of TPA and other PKC activators to modulate apoptosis.
Some authors have recently examined the capacity of TPA to modulate the toxicity of As 2 1 The abbreviations used are: PKC, calcium-dependent protein kinase; Ac-DEVD-pNA, N-acetyl-Asp-Glu-Val-Asp-p-nitroaniline; BSO, DL-buthionine-R,S-sulfoximine; ERK, extracellular signal-regulated protein kinase; GSH, reduced glutathione; MAPK, mitogen-activated protein kinase; NAC, N-acetyl-L-cysteine; TPA, 12-O-tetradecanoylphorbol-13-acetate; VD3, 1␣,25-dihydroxyvitamin D 3 ; Z-VAD-Fmk, benzyloxy-carbonyl-Val-Ala-Asp-fluoromethyl ketone; MEK, mitogenactivated protein kinase/extracellular signal-regulated kinase kinase; JNK, c-Jun NH 2 -terminal kinase. non-coincident results. For instance, Sordet et al. (22) indicate that TPA-differentiated cells exhibited an increased susceptibility to apoptosis induction by As 2 O 3 , which was associated to a decrease in reduced glutathione (GSH) intracellular content and to the accumulation of reactive oxygen species. However, other authors indicate that the phorbol ester did not affect (23) or even reduce (24) the generation of apoptosis by As 2 O 3 . These discrepancies led us to examine the capacity of TPA and some related compounds to modulate the apoptotic action of As 2 O 3 and other cytotoxic agents in U-937 promonocytic leukemia cells using different experimental conditions. The obtained results indicated that the generation of apoptosis by the GSHsensitive agents As 2 O 3 is potentiated by the simultaneous addition of TPA or bryostatin 1. The increase in apoptosis is mediated by ERK activation and seems to be a consequence, at least in part, of GSH depletion. However, after a prolonged preincubation TPA does not potentiate and instead decreases the As 2 O 3 toxicity, an effect also obtained by preincubation with other differentiation inducers.
Cells and Treatments-U-937 human promonocytic leukemia cells (25) and HL-60 human promyelocytic leukemia cells (26) were routinely grown in RPMI 1640 supplemented with 10% (v/v) heat-inactivated fetal calf serum and 0.2% sodium bicarbonate and antibiotics in a  (20 mM) was prepared in distilled water. A stock solution of 1␣,25-dihydroxyvitamin D 3 (VD3, 1 mM) was prepared in ethanol. All these solutions were stored at Ϫ20°C. Stock solutions of arsenic trioxide and cadmium chloride (100 mM) were prepared in distilled water, and stock solutions of 4,6-diamino-2-phenylindole (10 g/ml) and propidium iodide (1 mg/ ml) were prepared in phosphate-buffered saline. These solutions were stored at 4°C. DL-Buthionine-R,S-sulfoximine (BSO) was dissolved in distilled water at 50 mM, and sodium butyrate was dissolved in RPMI 1640 at 100 mM just before application.
Determination of Apoptosis-Distinctive characteristics of apoptotic cells were the presence of chromatin condensation/fragmentation and the acquisition of sub-G 1 DNA content. To examine chromatin structure, cells were fixed with ethanol, stained with 4,6-diamino-2-phenylindole, and examined by fluorescence microscopy. To measure DNA content, cells were permeabilized, stained with propidium iodide, and examined by flow cytometry. These procedures were described in detail elsewhere (27).
Measurement of Caspase-3 Activity-Samples of 4 ϫ 10 6 cells were collected by centrifugation, washed twice with ice-cold phosphate-buffered saline, resuspended in 50 l of ice-cold lysis buffer (1 mM dithiothreitol, 0.03% Nonidet P-40 (v/v), in 50 mM Tris, pH 7.5), kept on ice for 30 min, and finally centrifuged at 14,000 ϫ g for 15 min at 4°C. Samples containing aliquots of the supernatants (corresponding to 10 g of total protein), 8 l of Ac-DEVD-pNA, and phosphate-buffered saline to complete 200 l were prepared in triplicate in 96-well microtiter plates and incubated for 1 h at 37°C. The absorption was measured by spectrophotometry at 405 nm.
Determination of Cell Differentiation-Cell differentiation was assessed by measuring the surface expression of CD11b/CD18 and CD11c/ CD18 leukocyte integrins. With this aim indirect immunofluorescence assays were carried out using the Bear 1 (anti-CD11b) and HC1/1 (anti-CD11c) monoclonal antibodies in combination with flow cytometry following the previously described procedure (28).
Measurement of GSH Levels-The total cellular GSH content was determined by fluorometry after cell loading with monochlorobimane following the previously described procedure (29).
Immunoblot Assays-To obtain total cellular protein extracts cells were collected by centrifugation, washed with phosphate-buffered saline, and lysed by 5 min of heating at 100°C followed by sonication in Laemmli buffer containing a protease inhibitor mixture, 10 mM sodium fluoride, and 1 mM sodium orthovanadate. The extracts were analyzed by SDS-polyacrylamide gel electrophoresis, blotted onto membranes, and immunodetected as previously described (27). DNA content (E). In F the kinase inhibitors were added at the indicated times in relation to As 2 O 3 plus TPA (here considered as hour zero). In all other experiments the kinase inhibitors were added 1 h before As 2 O 3 plus TPA. All other conditions were as in Figs. 1 and 3. Fig. 1 shows the capacity of TPA and As 2 O 3 , alone and in combination to induce the expression of apoptotic markers in human myeloid cells. As 2 O 3 caused a concentration-dependent (Fig. 1A) and time-dependent (Fig.  1B) increase in the frequency of U-937 promonocytic cells with fragmented chromatin, which is characteristic of apoptosis. TPA (20 nM) was almost innocuous in itself (ϳ8% of apoptotic cells at 24 h of treatment versus 4% in untreated cells) but greatly potentiated the apoptotic action of As 2 O 3 when both agents were simultaneously applied. The potentiation by TPA of the As 2 O 3 -provoked cell death was confirmed by measuring the frequency of cells with a decreased (sub-G 1 ) DNA content, which is also an indicator of apoptosis (Fig. 1C), and was further corroborated by measuring the stimulation of caspase-3 activity ( Fig. 2A) and the capacity of the caspase inhibitor Z-VAD-Fmk to inhibit cell death (Fig. 2B). TPA also potentiated the As 2 O 3 -provoked apoptosis in HL-60 human promyelocytic cells (Fig. 1D), showing that the phenomenon is not restricted to the U-937 cell line. On the basis of the results here obtained, the concentration of 4 M As 2 O 3 was adopted for further experiments except when otherwise indicated.

Apoptosis Induction-
TPA is a potent PKC activator as well as a differentiation inducer of myeloid cells. For this reason we found it of interest to examine whether the generation of apoptosis by arsenic could be altered by the simultaneous administration of 10 nM VD3 or 0.75 mM sodium butyrate, which in our experiments induced myeloid cell differentiation (28,30), or 10 nM bryostatin 1, which is a PKC activator (albeit with different isoform specificity than TPA) (31) and differentiation inducer (32). The results in Fig. 3 indicate that bryostatin 1 potentiated the As 2 O 3 -provoked apoptosis, but VD3 and sodium butyrate were ineffective. This indicates that the stimulatory action of TPA on apoptosis is not related with differentiation induction.
MAPK Activation-As indicated above PKC activation may in turn activate the ERK pathway. For this reason, immunoblot assays were carried out to determine the activation of ERK1/2, as measured by their increased phosphorylation, in cells treated with As 2 O 3 and TPA either alone or in combination. It was found that arsenic alone did not significantly cause ERK activation (Fig. 4B), whereas this kinase was greatly activated by TPA alone (Fig. 4A) and by the combination of As 2 O 3 plus TPA (Fig. 4B). The activation was already observed at 1 h of treatment and still persisted at 24 h.
To analyze whether there is a cause-effect relationship between ERK activation and potentiation of apoptosis by TPA we made use of appropriate pharmacological inhibitors, namely the PKC inhibitor GF109203X (33) and the MEK/ERK inhibitors PD98059 and U0126 (34,35). Other more direct experimental approaches were not employed due to the poor efficacy of transfection of U-937 cells (result not shown). Control assays indicated that 1 M GF109203X, 2.5 M U0126, and 20 M PD98059 prevented the As 2 O 3 plus TPA-induced ERK phosphorylation (Fig. 4C), proving the efficacy of the inhibitors and confirming that PKC activation in fact precedes and regulates ERK activation. As shown in Fig. 4, D and E, the kinase inhibitors attenuated apoptosis in As 2 O 3 plus TPA-treated cells, indicating that under these experimental conditions ERK activation effectively mediates apoptosis induction. The same result was obtained using bryostatin 1 instead of TPA (Fig. 4D). As demonstrated by kinetic assays, the kinase inhibitors were only effective when applied at the same time as TPA plus As 2 O 3 or shortly thereafter (up to 4 h, approximately) (Fig. 4F).
For comparison we also measured the behavior of p38 and JNK since these kinases may be also activated by TPA in U-937 cells (36). As indicated in Fig. 5A, treatment with As 2 O 3 alone induced the phosphorylation/activation of p38 and JNK, which was potentiated by TPA. The administration of 10 M SB20358, specific for p38 (37), or 10 M SP600125, specific for JNK (38), the maximum concentrations that were non-toxic in long term treatments, reduced kinase activation (Fig. 5B) but did not attenuate apoptosis (Fig. 5C). This suggests that p38 and JNK are not primarily responsible for the potentiation by TPA of the As 2 O 3 -provoked apoptosis.
Changes in GSH Content-It has been described that As 2 O 3 is a GSH-sensitive agent, in the sense that its toxicity is enhanced after GSH depletion (39,40). This was corroborated by us using BSO, a specific inhibitor of ␥-glutamylcysteine synthetase activity, the rate-limiting enzyme for GSH biosynthesis (41). In fact BSO reduced the intracellular GSH content ( Fig  6A, left panel) and enhanced the generation of apoptosis by As 2 O 3 (Fig. 6A, right panel). In addition, it was reported that TPA-differentiated myeloid cells exhibited lower GSH content than non-differentiated cells (22). For these reasons, we wanted to measure the alteration of GSH levels in cells treated with As 2 O 3 and TPA alone and in combination and in the absence or the presence of the PKC and MAPK inhibitors. For comparison we also measured the GSH content in cells treated with the differentiation inducer VD3. The results, represented in Fig. 6, B-D, were as follows. (i) Treatment with 4 M As 2 O 3 alone (which as indicated above moderately induced apoptosis but failed to induce ERK activation) did not decrease the GSH content (Fig. 6B). (ii) Treatment with TPA alone (which as indicated above did not cause significant apoptosis but induced ERK activation) caused GSH depletion (Fig. 6B). Of note, no GSH decrease was observed in cells treated for 24 -72 h with VD3 (Fig. 6B) although at 72 h of treatment the cells expressed differentiation markers (results not shown). (iii) The combination of As 2 O 3 plus TPA (which as indicated above greatly induced apoptosis as well as ERK activation) caused a greater GSH decrease than TPA alone (Fig. 6B). (iv) The GSH depletion caused by TPA alone or by As 2 O 3 plus TPA was attenuated by the PKC and ERK inhibitors (which as indicated above attenuated apoptosis) but was not affected by the p38 and JNK inhibitors (which also failed to prevent apoptosis) (Fig. 6C). Similar results were obtained using bryostatin 1 instead of TPA (Fig. 6C). And (v), the administration of PD98059 7 h after As 2 O 3 plus TPA was unable to restore the GSH level (Fig. 6D), which is congruent with the above-indicated inability of the ERK inhibitor to prevent apoptosis at this time of treatment. Taken together these results indicate the existence of a strict correlation between ERK activation, GSH depletion, and potentiation of apoptosis in As 2 O 3 plus TPA-treated cells. Moreover, they suggest that GSH depletion is not a mere consequence of cell death or differentiation induction.
Finally, experiments were carried out using NAC, a GSHincreasing agent earlier employed by other authors in combination with As 2 O 3 (39,42). We found that the administration of 10 mM NAC did not prevent ERK activation (result not shown) but restored the GSH content (Fig. 6E, left panel) and reduced apoptosis (Fig 6E, right panel) in As 2 O 3 plus TPA-treated cells. Of note, the possibility that such attenuation of apoptosis could be due to arsenic scavenging by NAC may be ruled out since NAC did not reduce the toxicity of As 2 O 3 plus LY294002, an inhibitor of the phosphatidylinositol 3-kinase pathway that potentiated the As 2 O 3 -provoked apoptosis (Fig. 6E, right  panel), and NAC also failed to prevent the As 2 O 3 -provoked activation of HSP70 expression, a stress-inducible protein (Fig.   6F). These results, which are fully consistent with earlier observations (39,42), support the existence of a cause-effect relationship between GSH depletion and potentiation of apoptosis in our experimental model.
Effects of Other Cytotoxic Drugs-The preceding results strongly suggest that the potentiation of As 2 O 3 -provoked apoptosis by TPA could be the consequence, at least in part, of GSH depletion. If this was the case we could predict a similar potentiation of apoptosis when the phorbol ester is used in combination with other GSH-sensitive agents but probably not with GSH-insensitive drugs. Earlier observations indicated that GSH depletion enhanced the toxicity of the heavy metal cadmium (43, 44) but did not affect the toxicity of anti-DNA topoisomerase drugs in myeloid cells (29,45). Hence, we decided to compare the effect of BSO and TPA on U-937 cells treated with cadmium chloride (CdCl 2 , 40 M) and with the antitumor anti-DNA topoisomerase drugs etoposide (0.5 M), camptothecin (50 nM), and doxorubicin (0.5 M). As expected, BSO potentiated apoptosis in the case of CdCl 2 but not in the case of the anti-topoisomerase drugs (Fig. 7A), and according to our prediction, TPA only potentiated apoptosis in the case of CdCl 2 (Fig. 7B). Indeed, TPA reduced the toxicity of the antitopoisomerase drugs (Fig. 7B), which is consistent with the commonly reported protective action of the phorbol ester (6 -13).
Of note, the different effect of TPA on the toxicity of CdCl 2 and anti-topoisomerase drugs may not be attributed to a different behavior of ERKs or GSH. In fact, TPA potentiated ERK activation (Fig. 7C) and exacerbated GSH depletion (Fig. 7D) by both CdCl 2 and doxorubicin. Moreover, GF109203X and PD98059 restored the GSH level in both CdCl 2 plus TPA-and doxorubicin plus TPA-treated cells (Fig. 7D). However, although the kinase inhibitors attenuated apoptosis induction by CdCl 2 plus TPA (as in the case of As 2 O 3 plus TPA), they did not attenuate and even increased apoptosis in the case of doxoru- FIG. 6. Modulation of GSH levels. A, the left histogram shows the relative GSH level in U-937 cells treated for the indicated time-periods with 1 mM BSO, and the right histogram shows the frequency of apoptosis in cell cultures treated for 24 h with 1 mM BSO alone and for the indicated time periods with 4 M As 2 O 3 with or without BSO. B, relative GSH levels in cells treated for the indicated time periods with As 2 O 3 alone, TPA alone, VD3 alone, and the combination of As 2 O 3 plus TPA. C, relative GSH levels in cells treated for 24 h with TPA alone, with As 2 O 3 plus TPA, and with As 2 O 3 plus bryostatin 1 in the absence (-) or presence of GF109203X, PD98059, U0126, SB20358, and SP600125. D, similar experimental design as in C with the exception that PD-98059 was added 7 h after As 2 O 3 plus TPA. bicin plus TPA (Fig. 7E). Taken together, these results corroborate the conclusion that the unusual pro-apoptotic action of TPA and ERKs in the case of As 2 O 3 is due to the GSH sensitivity of this agent.
Effect of Preincubation with TPA-In all the preceding experiments, TPA was administered simultaneously to As 2 O 3 or the other cytotoxic drugs. Hence, new experiments were carried out in which the cells were preincubated for different times with the phorbol ester before treatment with As 2 O 3 . Some of the obtained results are represented in Fig. 8. Preincubation with TPA for up to 6 h was still compatible with a potentiation of apoptosis. However, preincubation for 14 -24 h, a treatment period that suffices to induce the expression of differentiation markers (46, and results not shown), did not potentiate and, instead, attenuated the As 2 O 3 -provoked apoptosis (Fig. 8, A  and B) and, accordingly, decreased caspase-3 activity (Fig. 8C). The decrease in apoptosis occurred despite the reduced GSH level, which remained below 40% of control value during the whole period of As 2 O 3 treatment (result not shown). Using this experimental design, similar results were obtained when the cells were treated with CdCl 2 or anti-topoisomerase drugs instead of As 2 O 3 (Fig. 8D) or when they were preincubated for 24 h with bryostatin 1 or for 48 h with the differentiation inducers VD3 and sodium butyrate instead of TPA (Fig. 8E). DISCUSSION The results in this work corroborate earlier observations indicating that As 2 O 3 and CdCl 2 cause death by apoptosis in myeloid leukemia cells and that the toxicity of these agents is exacerbated by GSH depletion (39,44). The increase in toxicity may be manifested as an increase in the frequency of apoptosis (Refs. 39 and 40 and results in this work) or even as a change in the mode of death from apoptosis to necrosis (44), 2 depending on the experimental conditions. By contrast, the toxicity of anti-DNA topoisomerase drugs was apparently insensitive to GSH depletion, a result also consistent with earlier observations in the myeloid cell model (29,45). The dependence of As 2 O 3 and CdCl 2 toxicity on GSH content may be explained by the capacity of arsenic and cadmium ions to directly react with GSH (43,47). Thus, GSH depletion may lead to a decrease in GSH-metal interactions, increasing the intracellular concentration of free arsenic and cadmium ions and, hence, their toxicity. In particular, the increase in metal ion concentration could exacerbate mitochondrial dysfunction, since it is known that arsenic and cadmium directly target the mitochondria (48,49). In addition, GSH depletion might potentiate cell death by facilitating the accumulation of reactive oxygen species due to the reduction in glutathione peroxidase activity (50). Nevertheless, the relevance of this mechanism in the present experiments is questionable, since treatment with low As 2 O 3 concentrations did not produce a detectable increase in peroxide levels in the U-937 cell system (51). 2 As indicated above, TPA was reported to potentiate (22) and reduce (24) the generation of apoptosis by As 2 O 3 in myeloid leukemia cells. Our present results indicate that both types of response are possible, depending on the conditions of treatment. Thus, TPA potentiated apoptosis when applied simultaneously to, or shortly before As 2 O 3 and CdCl 2 but decreased apoptosis after a prolonged (14 -24 h) incubation period. This contrasts with the response of anti-topoisomerase drugs, the toxicity of which was always reduced by TPA independently of 2 C. Ferná ndez, A. M. Ramos, P. Sancho, D. Amrá n, E. de Blas, and P. Aller, unpublished results. the conditions of treatment. The potentiation of As 2 O 3 toxicity by TPA was mediated by PKC activation, since the same result was obtained using bryostatin 1 instead of TPA, and apoptosis was attenuated by the PKC inhibitor GF109203X. The action of PKC was in turn mediated by ERK activation, since GF109203X prevented the TPA plus As 2 O 3 -provoked ERK phosphorylation, and the potentiation of apoptosis by TPA or bryostatin 1 was reduced by the ERK inhibitors U0126 and PD98059. This infrequent pro-apoptotic effect of ERKs, normally considered as a survival-inducing kinase, contrasts with the apparent irrelevance of p38 and JNK, normally considered as pro-apoptotic kinases. In fact, TPA potentiated the activation p38 and JNK, but SB203580 and SP600125 failed to reduce the TPA plus As 2 O 3 -provoked apoptosis. Concerning the reduction of As 2 O 3 and CdCl 2 toxicity after a prolonged preincubation with TPA, we may reasonably suppose that this is a consequence of differentiation. In fact, U-937 cells treated with TPA for 14 -24 h already express differentiation markers, and the toxicity of As 2 O 3 was also decreased by preincubation with other differentiation inducers, namely bryostatin 1, sodium butyrate, and VD3. The resistance of differentiated myeloid cells to apoptosis induction by diverse cytotoxic agents has been also documented by other authors (11,(52)(53)(54). Because differentiation and apoptosis are alternative, mutually excluding pathways (55), a restraint in apoptosis may be expected once the differentiation program has been executed.
Finally, the present results indicate that GSH depletion is a key factor linking ERK activation and potentiation of apoptosis in As 2 O 3 plus TPA-treated cells. This conclusion is supported by the concurrence of multiple evidences. In fact (i) TPA-mediated ERK activation always correlated with GSH decrease (Fig.  4, A and B, versus Fig. 6B and Fig. 7, C versus D) no matter the effect of TPA on apoptosis, and the administration of ERK inhibitors restored the GSH content (Figs. 6C and 7D). More-over, treatment with As 2 O 3 alone did not cause ERK activation nor GSH decrease (Fig. 4B versus Fig. 6B). (ii) The apoptotic action of As 2 O 3 was potentiated after GSH depletion by BSO and TPA, and this potentiation was abrogated when the GSH level was restored by NAC (Fig. 6). (iii) Treatment with TPA alone caused GSH decrease without significant apoptosis (Fig.  6B versus Fig. 1), indicating that GSH depletion is not a mere consequence of cell death in this type of experiment. And (iv) finally, TPA-mediated ERK activation and GSH depletion correlated with apoptosis increase only in the case of GSH-sensitive agents such as As 2 O 3 and CdCl 2 . By contrast, TPA and ERKs played their canonical, anti-apoptotic role when used with GSH-insensitive, anti-DNA topoisomerase drugs (Fig. 7). Hence, we may reasonably conclude that ERK activation leads to GSH depletion, which in turn accounts at least in part for the increased apoptosis in TPA plus As 2 O 3 -treated cells. Experiments are in course to determine the mechanism(s) responsible for GSH depletion, i.e. synthesis inhibition or accelerated loss. In addition, it remains to be investigated whether the conclusions here obtained may be extended to other antitumor drugs, e.g. cisplatin, a GSH-sensitive drug (29), the toxicity of which may be potentiated by ERK activation (18).