Caffeine Sensitizes Human H358 Cell Line to p53-mediated Apoptosis by Inducing Mitochondrial Translocation and Conformational Change of BAX Protein*

The mechanisms involved in p53-mediated cell death remain controversial. In the present study, we investigated this cell death pathway by stably transfecting the p53-null H358 cell line with a tetracycline-dependent wild type p53-expressing vector. Restoration of p53 triggered a G2/M cell cycle arrest and enhanced BAX protein expression, without inducing apoptosis or potentiating the cytotoxic effect of etoposide, vincristine, and cis-platinum. Accordingly, overexpression of BAX in H358 cells, through stable transfection of a tetracycline-regulated expression vector, did not induce cell death. Interestingly, the methylxanthine caffeine (4 mm) promoted the translocation of BAX from the cytosol to the mitochondria. In the setting of an overexpression of BAX, caffeine induced a conformational change of the protein and apoptosis. The consequences of caffeine were independent of its cell cycle-related activities. All together, caffeine synergizes with p53 for inducing cell death through a cell cycle-independent mechanism, involving mitochondrial translocation and conformational change of BAX protein.

DNA damage can induce accumulation of the p53 transcription factor. In turn, p53 regulates the cellular response to DNA damage by mediating cell cycle arrest, DNA repair, and/or cell death, which is essential for preventing the accumulation of genetic alterations (1)(2)(3). p53 is probably the most commonly mutated gene in human tumors. Whether p53 inactivation is a resistance factor to DNA damaging agents remains a controversial issue. However, restoration of the wild type protein expression through gene transfer was proposed as a treatment strategy for malignant tumors, either by inducing cell death or by sensitizing tumor cells to chemotherapy (4).
Several p53-activated apoptotic pathways, which depend or not on the transcriptional activity of the protein, were identi-fied (5)(6)(7). This transcription factor can transactivate several mediators of apoptosis, including the death receptors Fas/CD95 (7)(8)(9) and KILLER/DR5 (10), several genes involved in the generation of reactive oxygen species known as PIGs (for p53induced genes) (11) and the pro-apoptotic protein BAX that belongs to the Bcl-2 family of proteins (12)(13)(14)(15). Studies on transgenic mice indicated that BAX contributes, at least in part, to the p53-dependent apoptotic pathway in fibroblasts (13) and brain tumor cells (14). p53 mutants that fail to induce apoptosis also fail to transactivate BAX (16 -18). However, BAX accumulation has been observed in cells that do not undergo apoptosis, suggesting that an increase in BAX expression level is not sufficient to trigger apoptosis in response to p53 induction (19 -21).
BAX protein controls cell death through its participation in mitochondria disruption and cytochrome c release from this organelle (22)(23)(24). Once cytosolic, cytochrome c induces oligomerization of APAF-1, which in turn recruits procaspase-9 and initiates the caspase cascade leading to the cell dismantling (25). This mitochondrial pathway to cell death has been shown to be activated during p53-dependent apoptosis in several cell systems (20,26,27).
To further investigate the p53-mediated apoptotic pathway, we used the p53-null non-small cell lung carcinoma H358 cell line transfected with a tetracycline-inducible wild type p53expressing vector. We show that wild type p53 expression triggers G 2 /M cell cycle arrest without inducing apoptosis, although a marked increase in BAX expression is observed. Interestingly, the methylxanthine caffeine is able to promote the translocation of BAX from the cytosol to the mitochondria and to trigger apoptosis in a cell cycle-independent manner.

EXPERIMENTAL PROCEDURES
Plasmids and DNA Transfection-pTRE-p53 and pTRE-BAX plasmids were constructed by subcloning, respectively, the wild type p53 cDNA or influenza hemagglutinin (HA) 1 protein epitope-tagged BAX cDNA downstream of the Tet-regulated promoter into pTRE plasmid. Construction was carried out according to the manufacturer's instructions (CLONTECH, Ozyme, Saint Quentin en Yvelines, France).
Cell Culture, Drug Treatment, and Serum Withdrawal-H358 cells were cultured as adherent monolayers in RPMI 1640 medium supplemented with 10% (v/v) heat-inactivated fetal calf serum and glutamine (2 mM) at 37°C in a humidified atmosphere of 5% CO 2 . All products were purchased from Life Technologies, Inc. (Cergy Pontoise, France). H358/TetON/p53, H358/TetON/BAX, and H358/TetON control cells were maintained in medium as described above, supplemented with hygromycin (50 g/ml) and Geneticin (G418: 100 g/ml) (both from Life Technologies, Inc.). To induce p53 or HA-BAX expression, doxycycline (Sigma-Aldrich) was added to the medium to a final concentration of 1 g/ml, which is non-toxic for the cells.
Caffeine, etoposide, vincristin, and cis-platinum were purchased from Sigma-Aldrich. Stock solutions of etoposide, vincristin and cisplatinum in Me 2 SO were stored at Ϫ20°C. Control cells received Me 2 SO solvent alone. The final concentration of Me 2 SO solvent in the culture medium never exceeded 1%, which is non-toxic for the cells. Caffeine was dissolved in fresh medium just before use. For serum deprivation, cells were rinsed once in PBS and cultured for at least 8 h in fetal calf serum-free medium. Medium was then changed and culture continued in serum-free medium for the indicated time. Drug treatments and serum deprivation were performed 24 h after p53 or BAX induction by diluting drug in fresh medium in the presence or absence of doxycycline (1 g/ml).
Analysis of Apoptotic Cell Death and Cytotoxic Assay-The morphological changes related to apoptosis were assessed by fluorescence microscopy after staining of cells with Hoechst 33342 (5 g/ml, Sigma-Aldrich), and the percentage of apoptosis was scored after counting at least 200 cells.
For cytotoxicity assay, cells were plated at the initial density of 10 4 cells/well in 96 wells plate in RPMI 1640 medium supplemented with 10% fetal calf serum and glutamine (2 mM) at 37°C. After 96 h of drug treatment, cells were washed in PBS, fixed in 100% ethanol for 15 min, and stained with methylene blue (1% in borate buffer 0.01 M, pH ϭ 8.5). Plates were washed with water, and methylene blue was eluted by 0.1 N HCl. The absorbance measurements were performed at 630 nm. Each determination was performed in triplicate. Cell survival was calculated as follows: Mean absorbance in drug-treated wells Mean absorbance in untreated wells ϫ 100 (Eq. 1)

Analysis of Protein Expression by Western
Blotting-Cells were washed twice in PBS and incubated in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM NaF, 1 mM Na 3 VO 4 , 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin, and 10 g/ml pepstatin) for 30 min on ice. Protein content was assessed by the Bio-Rad D C Protein Assay kit (Bio-Rad S.A., Ivry sur Seine, France). Cell lysates (20 -40 g of proteins) were subjected to polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane (Millipore, Saint Quentin Yvelines, France). After blocking nonspecific binding sites for 30 min with 5% nonfat milk in TPBS (PBS, 0.1% Tween 20) at room temperature, the membrane was incubated overnight at ϩ4°C with primary antibody diluted in TPBS. The used primary antibodies were anti- (PharMingen). To ensure equal loading and transfer, membranes were reprobed for actin using anti-actin rabbit polyclonal antibody (1/3000) (Sigma-Aldrich). The immunoreactive proteins were visualized using horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibodies (both from Amersham Pharmacia Biotech, Orsay, France) diluted 1/5000 in TPBS and enhanced chemoluminescence (ECL; Amersham Pharmacia Biotech).
Analysis of Cell Cycle and Protein Immunostaining by Flow Cytometry-For cell cycle analysis, cells were washed in ice-cold PBS, fixed in 70% ethanol for 30 min, and incubated for 15 min in the presence of RNase A (20 g/ml). Cellular DNA was stained with propidium iodide (10 g/ml in PBS) (Roche Molecular Biochemicals, Meylan, France).
After three washes in PBS, cells were incubated for 30 min with fluorescein isothiocyanate-conjugated mouse or rabbit secondary antibodies (Jackson Immunoresearch Laboratories, Interchim) diluted 1/100 in PBS plus 1% BSA, washed again in PBS, and resuspended in 1 ml of PBS. Analysis was performed on a FACScan flow cytometer (Becton-Dickinson, Le Pont de Claix, France) using Cellquest software. Green fluorescence (fluorescein isothiocyanate, FL-1) was detected at 530 nm and red fluorescence (propidium iodide, Fl-2) at 575 nm.
Preparation of Heavy Membrane and Cytosolic Enriched Fractions-Cells were washed twice in ice-cold PBS and then resuspended at 5 ϫ 10 7 /ml in ice-cold lysis buffer (20 mM Hepes-KOH, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 250 mM sucrose, 1 mM dithiothreitol, 1 mM NaF, 1 mM NaVO 4 , 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin, and 10 g/ml pepstatin). Cells were broken down by passing through a 25-gauge needle fitted on a 2-ml syringe, and the suspension was centrifuged at 700 ϫ g for 7 min at ϩ4°C in an Eppendorf centrifuge. Supernatant was centrifuged at 10,000 ϫ g for 20 min. The resulting pellet containing heavy membrane (designated as heavy membrane fraction (HMF)) was washed once in lysis buffer and resuspended in lysis buffer containing 1% Triton X-100. The supernatant was subjected to ultracentrifugation at 100,000 ϫ g for 30 min to pellet remaining membranes, and the resulting supernatant was termed cytosolic fraction. HMF and cytosolic fraction were assessed for protein content using the Bio-Rad D C Protein Assay kit, and 20 g of proteins were subjected to polyacrylamide gel electrophoresis and analyzed by Western blotting for BAX content. The relative purity of fractions was ascertained by Western blotting using the mouse monoclonal anti-mitochondrial-HSP70 (Affinity Bioreagent, Interchim) as a marker of mitochondria. The equal proteins loading and transfer were ensured by reprobing the membranes using anti-actin rabbit polyclonal antibody (Sigma-Aldrich).
Immunofluorescence Staining and Microscopic Analysis-At different times after p53 or BAX induction, cells were fixed in paraformaldehyde (0.5%) for 5 min at room temperature, washed twice in PBS, and incubated for 2 h at room temperature in the presence of anti-BAX rabbit polyclonal antibody (1/100) (N-20; Santa Cruz) or anti-BAX rabbit polyclonal antibody (1/100) (PharMingen) both raised against the peptide sequence amino acids of N terminus BAX protein or anti-HA mouse monoclonal antibody (1/1000) (Babco, Eurogentec, Angers, France) and/or anti-mitochondrial-HSP70 mouse monoclonal antibody (1/100) (Affinity Bioreagent, Interchim, Montluçon, France) or mouse or rabbit irrelevant immunoglobulins (PharMingen). Antibobies were diluted in PBS containing 0.1% saponin and 1% BSA. For cytochrome c staining, cells were fixed in 4% paraformaldehyde for 5 min, then washed twice in PBS and permeabilized with 1% Triton X-100. After saturation of nonspecific binding with a 5% milk solution in PBS, cells were incubated for 1 h with anti-cytochrome c monoclonal antibody   control of a tetracycline-dependent promoter. Three H358/ tetON/p53 clones were selected for subsequent studies. The tetracycline analog doxycycline (1 g/ml) induced the expression of p53 in H358/tetON/p53 cells within 24 h (Fig. 1A). The biological activity of transfected p53 was assessed by its ability to up-regulate its transcriptional targets p21 and BAX (Fig.  1A). As we observed previously (28), we did not detect any expression of Bcl-2 in these cells (data not shown) and the level of the Bcl-2-related pro-apoptotic protein BAK was not modified after p53 expression (Fig. 1A). Despite the increase in BAX protein level, expression of p53 failed to induce characteristic morphological changes of apoptosis, as observed after staining of cells with the dye Hoechst 33342 (Fig. 1B). In addition, wild type p53 expression did not sensitize H358 cells to the toxic activity of cis-platinum, vincristine, or etoposide (Fig. 1C).

Expression of Wild
The Methylxanthine Caffeine Induces Apoptosis in p53-expressing Cells-Expression of p53 in H358/tetON/p53 cells increased the number of cells in the G 2 /M phase of the cell cycle within 48 h of doxycycline treatment (Table I). In order to bypass this G 2 /M arrest, cells were either serum-deprived or treated with the methylxanthine caffeine (4 mM). As expected, serum removal from culture medium or exposure to caffeine during 48 h significantly decreased the number of cells in G 2 /M phase and increased the number of cells in G 0 /G 1 phase, both in p53-null and in p53-expressing cells (Table I). Whereas serum deprivation did not induce a significant apoptosis in any condition (Table I), caffeine induced a dose-and time-dependent apoptosis in p53-expressing cells that was not observed in p53-null cells (Table I and Fig. 2A). Since caffeine-mediated apoptosis was not observed in H358/tetON control cells, we ruled out the possibility of a nonspecific effect of caffeine in the presence of doxycycline. In accordance with these observations, p53 expression was not sufficient to induce the proteolytic activation of procaspase-3 (Fig. 2B). Similarly, caffeine treatment of p53-null cells failed to induce caspase-3 proteolysis. By contrast, caffeine treatment of p53-expressing cells was associated with the proteolytic cleavage of procaspase-3 in a p12 active fragment (Fig. 2B). Caffeine did not influence the level of expression of p53, BAX, and p21 in either p53-null or p53expressing cells (Fig. 2B). Neither wild type p53 nor caffeine had any influence on the expression of the apoptosis inhibitor protein survivin (Fig. 2B) (29).
Caffeine Induces Apoptosis in BAX-overexpressing Cells-In order to determine whether caffeine acts upstream or downstream of BAX in the p53-dependent apoptotic pathway, we transfected BAX cDNA in H358 cells, using the same tetracycline-inducible expression system as for wild type p53. To discriminate between the endogenous and induced protein, an influenza hemagglutinin protein epitope (HA) was added to BAX. Doxycycline induced a dose-dependent expression of HA-BAX (Fig. 3A). In agreement with the results obtained in H358/ tetON/p53 clones after p53 induction, overexpression of BAX did not affect cell survival (Fig. 3A), nor did it sensitize the cells to drug-induced apoptosis (data not shown). Even though BAX overexpression did not induce any accumulation of cells in the G 2 /M phase of the cell cycle (Fig. 3A), caffeine could induce a dose-dependent apoptosis in BAX-overexpressing cells (Fig.  3B). This was associated with the procaspase-3 proteolytic cleavage in its p12 active fragment (Fig. 3B). Caffeine did not influence the level of endogenous BAX or HA-BAX when studied by Western blotting (Fig. 3B). However, it modified the pattern of expression of the HA-tagged protein when studied by immunofluorescence using an anti-HA antibody (Fig. 3C). In the absence of caffeine, HA-BAX protein exhibited a punctuated as well as diffused staining pattern, whereas, in the presence of caffeine, HA-BAX protein showed a predominantly punctuated staining pattern, suggesting that caffeine induced a change in HA-BAX intracellular distribution.
Caffeine Induces a Subcellular Redistribution of BAX-To further analyze the influence of p53 and caffeine on BAX subcellular localization, H358/TetON/p53 cells were fractionated into cytosolic fractions and HMF (Fig. 4). In control, p53-null cells, BAX protein was expressed mainly in the cytosolic fraction and more weakly in the HMF. Induction of p53 expression by doxycycline increased the expression of BAX in both the cytosolic fraction and the HMF, without modifying its subcellular distribution. Caffeine induced a strong decrease in cytosolic BAX expression within 72 h, whereas the protein accumulated in the HMF in both p53-null and p53-expressing cells (Fig. 4).
Caffeine Induces a Conformational Change of BAX Protein-Previous studies have shown that BAX protein underwent a conformational change exposing its N terminus during apoptosis (30,31). Using epitope-specific anti-BAX antibodies that specifically recognize BAX protein with an exposed N-terminal extremity but not the protein in its native conformation, we analyzed the conformational state of BAX protein. A flow cytometry analysis showed that induction of p53 by doxycycline in H358/tetON/p53 had a limited influence on the exposed N terminus BAXassociated immunofluorescence (Fig. 5A, 3). Thus, induction of p53 increased the expression of BAX protein (as detected by Western blotting) in its native conformation with an unexposed N terminus. Caffeine did not modify the level of BAX N terminus staining in p53-null cells (Fig. 5A, 2). By contrast, it induced an exposure of the N terminus of BAX in p53-expressing H358/ TetON/p53 cells (Fig. 5A, 4), as revealed by an increase in exposed N terminus BAX-associated fluorescence. Similarly, when examined by microscopy after immunostaining using the same epitope-specific BAX antibody, N terminus BAX was not detected in p53-null cells (Fig. 5B, 1), in H358 cells expressing p53 upon doxycycline treatment (Fig. 5B, 3), or in p53-null cells exposed to caffeine (Fig. 5B, 2). In p53-expressing cells, caffeine induced an intense punctuated staining of N terminus BAX (Fig. 5B, 4) that co-localized with the mitochondrial HSP70 protein (Fig. 5C). In addition, apoptotic cells identified by their fragmented nuclei were positive for N terminus BAX staining and demonstrated a diffuse cytochrome c staining, which contrasted with the punctuated staining observed in non-apoptotic cells (Fig. 5D). This later observation suggested that the exposure of the N terminus of BAX was associated with the cytochrome c release from mitochondria. DISCUSSION Several pathways have been described to mediate p53-induced apoptosis (5). One of these involves BAX, a pro-apoptotic member of the Bcl-2 family of protein (13)(14)(15)(16)(17)(18). We show here that expression of wild type p53 in H358 human p53-null cancer cell line induces BAX accumulation in the cytosol without inducing cell death. At least one additional signal is required to modify BAX localization and conformation, thereby inducing cytochrome c release and caspase activation. We demonstrate that the methylxanthine caffeine is able to provide such an additional signal.
p53 expression in p53-null cells was shown to lead either to cell cycle arrest or apoptosis. Several studies brought evidence that the level of p53 expression was critical for its pro-apoptotic activity (32)(33)(34). p53-induced apoptosis was usually observed when p53 is strongly expressed, e.g. through transient transfection or viral infection, whereas the protein expressed at a lower level, e.g. by using a temperature-sensitive mutant or an inducible system, did not trigger cell death (19,(35)(36)(37)(38). Accordingly, in our cellular model, we have shown previously that transient transfection of p53 could induce apoptosis. 2 This was confirmed in the H358/tetON/p53 clone cultured in the absence of doxycycline (data not shown), indicating that the p53-mediated apoptotic pathway was functional in these cells. Thus, the lack of cell death in H358/tetON/p53 cells expressing p53 upon doxycycline exposure may be due to a low level of p53 expression. Another possibility is that cellular stress mediated by transient gene transfer synergizes with p53 to mediate apoptosis.
Whatever the mechanism, doxycycline-mediated p53 expression does not trigger H358 cell death while it increases BAX expression. The subcellular localization of BAX plays a central role in its pro-apoptotic activity (20,40,41). When associated with mitochondria, BAX promotes the cytosolic release of cytochrome c, which in turn, activates caspase-9 and -3 (22)(23)(24)(25). However, this translocation to the mitochondria remains insufficient to trigger cell death since caffeine induces relocalization of the protein in both p53-null cells, which do not die upon caffeine exposure, and p53-expressing cells, in which caffeine triggers cell death. These results are consistent with the study by Deng and Wu (20) in mouse embryo fibroblast cells, showing that Peg3/Pw1 protein, which induced BAX translocation from cytosol to mitochondria, did not lead to apoptosis when expressed alone, but greatly enhanced apoptosis induced by BAX overexpression and cooperated with p53 to induce cell death. Two additional events are observed in cells that die upon caffeine treatment; BAX protein is overexpressed, and the protein undergoes conformational changes. In p53-expressing cells, caffeine-induced cell death correlates with exposure of BAX N terminus and cytochrome c release. Exposure of N terminus of BAX and related proteins such as BAK has been shown to be associated with apoptosis in several cell types (30,31,42). This exposure could reflect a conformational change of the proteins that allows a more efficient insertion into mitochondrial membrane to promote the release of pro-apoptotic proteins such as cytochrome c (43)(44)(45). Caffeine appears to promote BAX translocation, whereas increased expression of the protein may facilitate the conformational change that is required for BAX membrane insertion.
The methylxanthine caffeine has been the first reported drug that overrides the G 2 /M checkpoint. How this abrogation of G 2 /M cell cycle arrest increases tumor cell sensitivity to ionizing radiations (46 -49), DNA-damaging agents (50), or overexpression of p53 (28) remains poorly understood. Serum deprivation, which induces an accumulation of cells in G 0 /G 1 similar to that observed in caffeine-treated cells, does not induce apoptosis of p53-expressing cells, indicating that caffeine-induced cell death may not be related to G 0 /G 1 accumulation. In addition, caffeine induced apoptosis both in p53-expressing cells that were arrested in G 2 /M and in cycling BAX-overexpressing cells. Thus, caffeine appears to sensitize tumor cells to apoptosis in a cell cycle-independent manner. Similarly, caffeinemediated sensitization to ionizing radiations was shown not to depend on the cell cycle effects of the compound (52).
At the concentration used in our study (4 mM), caffeine was shown to inhibit the activity of several related enzymes that include ATM, ATR, and mammalian target to rapamycin kinase (47,49,53). ATM was shown to modulate p53 activity toward cell cycle arrest rather than apoptosis (54,55). We may speculate that caffeine-mediated inhibition of ATM may contribute to induce apoptosis in p53-expressing cells. However, whether inhibition of ATM or related proteins is connected to the observed changes in BAX N terminus expression in p53expressing cells will require further investigations. Preliminary studies indicate that high concentrations (40 M) of wortmannin that inhibit ATM (47) sensitize H358 cells to apoptosis induced by overexpression of BAX (data not shown). Wortmannin also inhibits the phosphatidylinositol 3-kinase, another enzyme related to ATM. This kinase has been shown to prevent conformational changes in BAX protein in cells detached from their extracellular matrix (39). However, at concentrations that inhibit phosphatidylinositol 3-kinase (100 nM), wortmannin does not trigger apoptosis in BAX-overexpressing H358 cells (data not shown).
In conclusion, we demonstrate that both increased expression of BAX and its translocation to the mitochondria are required to trigger apoptosis in H358 cells. Caffeine relocalizes BAX from the cytosol to the mitochondria. When BAX is over-expressed, e.g. as a consequence of p53 expression, this relocalization is associated with conformational changes of the protein and apoptosis. These results strengthen the interest of caffeine and related molecules as sensitizers in treating human tumors.