Introduction

The permeability transition pore complex (PTPC), also called the mitochondrial megachannel (Zoratti and Szabo, 1994), is a polyprotein complex, that has been implicated in the regulation of mitochondrial matrix homeostasis (Bernardi et al., 1998), and more recently, in mitochondrial membrane permeabilization (MMP), a process leading either to apoptosis, necrosis or autophagy (Reed et al., 1998; Kroemer and Reed, 2000; Elmore et al., 2001). Its opening as a large unselective channel leads to the mitochondrial permeability transition (MPT) and results in an increase in the permeability of the inner membrane (IM), permitting the influx or efflux of any molecule of MW 1500 kDa. Consequently, the mitochondrial matrix swells and the mitochondrial transmembrane potential (m) dissipates (Gunter and Pfeiffer, 1990). In physiological conditions, PTPC pore function is regulated by a variety of signals, including the matrix pH, the concentration of divalent cations (Ca²⁺, Mg²⁺), the intracellular redox equilibrium and the matrix volume (Halestrap et al., 1997; and for reviews, Bernardi et al., 1998; Crompton, 1999). As measured by electrophysiology, PTPC exhibits multiple conductance states, suggesting that the channel is composed of several cooperating subunits (Szabo et al., 1992; Szabo and Zoratti, 1992). In contrast, in lethal conditions, PTPC opening can lead to an irreversible MMP, and the release of apoptogenic proteins into the cytosol triggering the ultimate degradation phase of apoptosis and cell death (Zamzami et al., 1995; Kroemer et al., 1997). Depending on conditions of apoptosis induction, PTPC opening has been suggested to precede Bax translocation from cytosol to the mitochondria (De Giorgi et al., 2002) or to occur concomitantly with Bax translocation, but before the release of cytochrome c (Cyt c) and apoptosis-inducing factor (AIF) from the mitochondria (see for a review Kroemer and Reed, 2000). Of note, Ott et al (2002) found that the most efficient method to release up to 85% of Cyt c was the induction of MPT in association with an oxidative stress (Ott et al., 2002). Alterations of PTPC function may lead to an excess of or a lack in apoptosis, and thus may contribute to numerous pathologies, such as ischemia–reperfusion (Lemasters et al., 1997), cardiomyopathy (Dorner et al., 1997), neurodegeneration (Friberg and Wieloch, 2002), lupus erythematosus (Bribes et al., 2003) or malignant transformation (Brenner et al., 2003).

The precise composition of PTPC has been a matter of controversy. Based on copurification experiments, several nonexclusive models have been proposed: (i) a model in which the adenine nucleotide translocase (ANT, the major protein of the IM), the voltage-dependent anion channel (VDAC, the most abundant protein of the outer membrane (OM)) and the peripheral benzodiazepine receptor (PBR, in the OM), form a tricomposite structure (McEnery et al., 1992); (ii) a model in which PTPC is composed of three proteins, ANT, VDAC and cyclophilin D (CypD, in the matrix) (Crompton et al., 2002); (iii) a minimal model in which PTPC would be composed of ANT and VDAC, the channel unit formed by these two proteins being regulated by CypD (Halestrap et al., 2002); and (iv) a model in which PTPC contains, at least, hexokinase (HK, in the cytosol), PBR, VDAC, ANT, creatine kinase (CK, in the intermembrane space) and CypD (in the matrix) (Figure 1) (Beutner et al., 1996; Marzo et al., 1998a). Only the three last models have been substantiated by functional studies using native or recombinant proteins reconstituted in artificial biomembrane (e.g. liposomes, black lipid membranes), showing that these proteins function similarly to the permeability transition pore in the mitochondrion. Moreover, we and others demonstrated that PTPC function could be modulated by physical interaction of some of its members with onco- and anti-oncoproteins from the Bax/Bcl-2 family. Indeed, ANT, VDAC and HK (also called glucokinase) were identified as ligands of Bax, Bcl-2, Bid, Bcl-xL, Bad and Bak (Marzo et al., 1998b; Shimizu et al., 2000a; Cao et al., 2001; Capano and Crompton, 2002; Cheng et al., 2003; Danial et al., 2003). Among these proteins, some members of Bax/Bcl-2 family were shown to be regulators of channel activity of ANT and VDAC (Brenner et al., 2000a; Shimizu et al., 2000a), as well as modulators of the ADP/ATP translocase activity of ANT (Belzacq et al., 2003). In summary, the occurrence of several distinct 'supercomplexes' involved in the regulation of MMP is currently admitted, but their dynamic evolution during cell death is not yet known.

Figure 1.

Expression of PTPC constituents in normal tissues and tumor cell lines. (a) Schematic presentation of PTPC. PTPC is a polyprotein complex built up on the two mitochondrial membranes (OM and IM). It contains, at least, HK (cytosol), the PBR (OM), the VDAC (OM), CK (inter membrane space), the ANT (IM) and CypD (matrix). (b and c) Co-immunoprecipitation of PTPC constituents in normal tissue (b) or in human tumoral cell lines (c). Mitochondria were isolated from rat brain, liver and heart, and from HT29 (colorectal adenocarcinoma), HeLa (cervical adenocarcinoma) and MCF7 (breast adenocarcinoma) cells, and directly analysed by immunoblot after membrane solubilization (Mito) or subjected to immunoprecipitation with an ANT-specific antiserum before immunoblot (IP). Detection of Cyt c and Cox II are controls of mitochondrial IM solubilization and of specificity of co-immunoprecipitation. Each immunoblot has been reproduced at least three times

Full figure and legend (79K)

To characterize the composition of PTPC during chemotherapy, we report that ANT can be associated with VDAC, HK, CypD and Bcl-2. Some protein–protein interactions (HK, Bax, Bcl-2, etc.) changed during apoptosis induction by treatment with etoposide (Etop), an inhibitor of topoisomerase II (Hande, 1998). Major changes coincided with the induction of MMP and preceded nucleus alteration. Other interactions (ANT, VDAC, etc.) were maintained until chromatin condensation and nuclear degradation occurred, suggesting that they are critical for proapoptotic PTPC function. Proteomic analysis of proteins co-immunoprecipitating with ANT allowed us to identify several novel PTPC regulators and to characterize a glutathione (GSH)-S-transferase (GST) as a repressor of proapoptotic PTPC pore function.

Results

PTPC composition in normal tissues and various human cancer cell lines

In many models of chemotherapy-induced apoptosis (Brenner and Kroemer, 2000) permeabilization of mitochondrial membrane involves the opening of PTPC, a polyprotein complex that would be composed of soluble proteins as well as of membranous proteins. As proposed previously by several authors (McEnery et al., 1992; Beutner et al., 1996; Marzo et al., 1998a; Crompton et al., 2002; Halestrap et al., 2002), PTPC may contain proteins from both mitochondrial membranes (VDAC, PBR, ANT), from cytosol (HK), from the matrix (CypD) and from intermembrane space (CK) (Figure 1a). To define the composition of PTPC in physiological conditions, mitochondria of three normal rat tissues (brain, heart and liver) and three human tumor cell lines (HT29, HeLa, MCF7 cells) were extracted and enriched by centrifugation to preserve crossorganelles contact sites (e.g. endoplasmic reticulum–mitochondrion). The protein content was normalized, analysed by SDS–PAGE and immunoblotted with various antibodies specific to PTPC putative members (Figure 1b, Mito and 1c, Mito). Simultaneously, proteins were immunoprecipitated with a specific polyclonal anti-ANT serum in the presence of 0.5% Triton X-100, a concentration that preserves PTPC organization (Beutner et al., 1996; Marzo et al., 1998a; Belzacq et al., 2003). The choice of Triton X-100, at the same concentration as that used to purify PTPC was also dictated by the fact that it preserves the maximal PTPC activity during its purification procedure in comparison with various structurally unrelated detergents such as 3-[(3-cholamidopropyl)dimethylammonio]-1-propane-sulfonate (CHAPS), octylglucoside, hecameg, sodium deoxycholate as well as sulfobetaine derivatives. However, we cannot exclude that some protein–protein interactions have been lost during the Triton X-100 solubilization procedure (Brenner et al., 2000b; C Brenner, unpublished results). The composition of the extracted complex, called the ANT interactome, was analysed by SDS–PAGE and immunoblotting with the same antibodies (Figure 1b, ANT-IP and 1C, ANT-IP). The ratio of ANT/VDAC appeared similar in all mitochondria independent of the origin of the tissue (Figure 1b, Mito) or of the cell line (Figure1c, Mito), indicating the constitutive expression of these two proteins. In contrast, the expression level of Bax, Bcl-2 and HK varied according to the tissue (Figure 1b, Mito) and the cell type (Figure 1c, Mito). Bax was not present in the mitochondria of healthy tissues (Figure 1b, Mito), was present in the mitochondria of HeLa cells and was detected as a trace in the mitochondria of HT29 and of MCF7 cells (Figure 1c, Mito), in a peripherally associated and inactive form (Crompton, 2000). Moreover, in the three cell lines, mitochondrial Bax was weakly associated with ANT (Figure 1c, ANT-IP) in accordance with previous observations. Bcl-2 was mainly present in the mitochondria of brain and heart tissues in comparison to the liver, in which it is not expressed (Figure 1b, Mito). In the same tissues, Bcl-2 was found within the ANT interactome (Figure 1b, ANT-IP). It should be noted that the amount of mitochondrial Bcl-2 associated with ANT in the brain was significantly higher than in the other tissues (Figure 1b, ANT-IP). Bcl-2 was also detected in the mitochondria of the three cell lines, mainly in MCF7 cell line (Figure 1c, Mito) and in association with ANT (Figure 1c, ANT-IP). Finally, while HK (detected by an anti-HKI antibody) was present in mitochondria as well as in the ANT interactome of the three tumoral cell lines (Figure 1c, Mito, ANT-IP), the brain and heart tissues (Figure 1b, Mito, ANT-IP), it was not detected in the mitochondria of liver tissue as expected, given that liver parenchymal cells contain glucokinase instead of HK (Figure 1b, Mito). CypD was determined using an approach combining 1D SDS–PAGE and MALDI–TOF MS method and its association with ANT was detected in the three tumoral cell lines (data not shown).

As controls of full solubilization of mitochondrial membranes and of specificity of co-immunoprecipitation, Cyt c (intermembrane space) and Cyt c oxidase II (Cox II) (IM) were, respectively, detected in the mitochondria, yet did not co-immunoprecipitate with ANT (Figure 1b and c).

Kinetics of Etop-induced cell death

Etop is a topoisomerase II inhibitor that is currently used for the treatment of wide range of neoplasias including lymphomas, breast and lung cancers (Hande, 1998). Induction of apoptosis by this DNA-damaging agent is known to involve Bax translocation to the mitochondria, MPT and release of mitochondrial Cyt c (Robertson et al., 2000; Karpinich et al., 2002). To characterize the kinetics of apoptotic induction by Etop in the HT29 cell line, we assessed three critical apoptosis-related parameters: the dissipation of m measured with m-sensitive dyes JC-1 and 3,3' dihexyloxacarbocyanine iodide (DiOC₍₆₎3), nuclear alterations examined with the DNA-intercalating agent Hoechst 33324 and the loss of DNA content determined by propidium iodide (PI) staining. JC-1 staining of Etop-treated HT29 cells revealed a qualitative progressive m loss (Figure 2a). Thus, mitochondria of untreated HT29 cells presented a high m, as shown by a punctuated cytoplasmic red fluorescence (Figure 2a, Co). At 12 h after treatment with 100 M Etop, mitochondria of some cells exhibited a green fluorescence indicative of a m loss (Figure 2a, 12 h). Subsequently, the fluorescence became green, weak and diffuse (Figure 2a, 16 and 36 h). This kinetics of m dissipation in the course of Etop treatment was further quantified by flow cytometry in Figure 2b, as evidenced by a progressive increase in the percentage of cells harboring a low m from 61 to 541% over a 36 h incubation period. Simultaneously, HT29 cells were labeled with the Hoechst 33324 dye and analysed for chromatin condensation (Figure 2a). As the nuclei of untreated cells remained intact, nuclear chromatin condensation was detected 20 h after treatment with etoposide, and apoptotic body formation appeared at 36 h. These results exclude the possibility of necrotic death and correlate with the percentage of cells showing hypoploidy indicative of a loss of nucleic acids that increased gradually with the time of the treatment from 122% at 12 h to 521% at 36 h (Figure 2b).

Figure 2.

Time course of Etop effect on the HT29 colon carcinoma cell line. HT29 cells were treated with100 M Etop or not (Co) for 12, 16, 20 or 36 h. (a) Cells were stained with Hoechst 33324 and JC-1. The loss of mitochondrial membrane potential is indicated by the progressive loss of red-JC-1-aggregate fluorescence and cytoplasmic diffusion of green monomer fluorescence following exposure to Etop. The nuclei showing chromatin condensation are indicated with an arrow and those showing apoptotic bodies with a star. (b) Cells were stained with DiOC6(3) for the measurement of m loss and with PI for the measurement of hypoploidy, followed by cytofluorometric analysis. Results are representative of four independent determinations. (c) Conformational activation and subcellular localization of Bax after Etop treatment. Cells were immunostained with a monoclonal antibody that specifically recognizes the conformationally active Bax (mouse monoclonal antibody 6A7) as well as a rabbit polyclonal antibody raised against cyt c. The two channels are shown separately for anti-Bax, for anti-Cyt c and merge. (d) Flow cytometric analysis of Bax immunofluorescence in cells treated with Etop (bold curve) and untreated cells (Co; normal curve), fixed and immunostained with the monoclonal antibody anti-Bax 6A7. In control cells, no label was detected. The percentage indicates the total number of fluorescent cells on the right of the dash line, after Etop treatment for 12 h (left panel), 16, 20 to 36 h (right panel)

Full figure and legend (285K)

Activation and translocation of Bax from the cytosol to mitochondria is a central event of the apoptotic program. Since one of the mechanisms proposed for Bax activation is its conversion to a peculiar conformation favoring the exposure of a N-terminus epitope (Desagher et al., 1999), we explored the kinetics by which apoptosis induced by Etop could induce Bax translocation to the mitochondria and its conformational change in comparison with the m loss. Bax immunostaining was then assessed by employing a monoclonal antibody that specifically recognizes the N-terminal epitope 6A7 only when it has been exposed in some cell lines (Desagher et al., 1999), such as HT29 cells. Thus, the treatment of HT29 cells with Etop lead to a punctuate 6A7 immunodistribution of fluorescence within the cytoplasm, indicating the activation of mitochondrial Bax (Figure 2c, 12–36 h), whereas untreated cells did not display any positive immunoreactivity (Figure 2c, Co). This effect was confirmed and quantified by flow cytometric analysis of the same samples (Figure 2d). Shortly after treatment with Etop (12 h), 91% of HT29 cells displayed a high red fluorescence. Furthermore, at this time, a fraction of activated Bax was localized to the mitochondrial membrane (Figure 2c), as shown by the yellow fluorescence, that corresponds to the merge of the red and the green fluorescence, indicating a colocalization of Bax and Cyt c in treated cells. In addition, during the Etop treatment, Cyt c remained mitochondrial up to 20 h as indicated by immunofluorescence (Figure 2c). At 36 h, Cyt c diffused into the cytosol of treated HT29 cells, indicating its release from mitochondria subsequently to the early m loss (Figure 2c, 36 h).

In conclusion, in agreement with previous reports (Karpinich et al., 2002; Ott et al., 2002), 12 h after treatment, Etop induces the translocation of Bax to mitochondria, as well as its conformational activation in nearly all cells (90%). At this time, only 15% of cells exhibit a m dissipation, suggesting that mitochondrial localization of Bax and its activation occur upstream of a significant depolarization of the mitochondrial membrane (>35%). Similarly, the activation of Bax occurs upstream of Cyt c release, nuclear chromatin condensation and apoptotic body formation. Thus, upon a 36 h treatment of HT29 cells with Etop, in 681% of the cells, Bax has translocated in the mitochondria (Figure 2d), and 521% of HT29 cells died by apoptosis, since their nuclei are fragmented and their viability is irreversibly compromised

Dynamics of intra-PTPC protein–protein interactions

Next, we investigated intra-PTPC interaction changes in HT29 cells by immunoprecipitation with the polyclonal anti-ANT serum (as in Figure 1), SDS–PAGE and immunoblotting after various periods of Etop treatment. Protein–protein interactions with ANT were resistant to an increase in ionic strength from 150 to 500 mM NaCl during the immunoprecipitation (not shown).

The amount of Bax contained within the ANT interactome increased during Etop treatment, whereas the quantity of Bcl-2 immunoprecipitated with ANT decreased at 16 and 20 h (Figure 3a, ANT-IP). This variation in the ANT-Bax interaction correlated with an increased mitochondrial content of Bax and was detected early around 12 h after the addition of Etop (Figure 3a). In contrast, after 36 h of treatment, the fraction of VDAC, HK1 and Bax interacting with ANT increased significantly, yet the amount of immunoprecipitated Bcl-2 did not increase (Figure 3a). As a possibility, the stoichiometry of this complex may reflect the composition of the proapoptotic PTPC at a late phase of Etop-induced apoptosis.

Figure 3.

Interaction changes between proteins associated with ANT upon the early phase of apoptosis. (a) Mitochondria from HT29 cells treated with 100 M Etop or from untreated control cells were isolated at various times after treatment. Co-immunoprecipitation was then performed as described in Figure 1, and the proteins were analysed by SDS-PAGE followed by immunoblot with anti-rabbit heart ANT polyclonal serum, mAbs specific for Bax (B9), VDAC, Bcl-2 (C2) or hexokinase (HK-1). (b) HT29 cells were treated with 100 M Etop for 20 h in the presence or absence of various doses of inhibitors, z-VAD.fmk (z-VAD), Furo or CsA. After treatments, mitochondria were purified and co-immunoprecipitated with the anti-rabbit heart ANT polyclonal serum, and proteins were analysed by immunoblot with mAbs specific for Bax (B9) or Bcl-2 (C2). Each experiment has been reproduced at least three times

Full figure and legend (80K)

To investigate the identity of endogenous signals and the role of other factors/proteins influencing the association between ANT, Bax and Bcl-2 changes during apoptosis, cells were treated with Etop in the absence or presence of various pharmacological inhibitors (z-VAD.fmk, cyclosporin A (CsA) and furosemide (Furo)). When used alone, these agents were devoid of any detectable effects on the ANT interactome, indicating that the concentrations of agents were compatible with our cellular experiments. As shown in Figure 3b, pretreatment with 10 M of z-VAD.fmk blocked the association of Bax with ANT in response to Etop, whereas the association of Bcl-2 with ANT was not modulated. In addition, pretreatment of HT29 cells with 5 M of the chloride channel inhibitor Furo, which inhibits Bax translocation (Karpinich et al., 2002), or 10 M of the MPT inhibitor CsA, which prevents CypD binding to ANT and pore opening (Woodfield et al., 1998), reduced Bax association with ANT and restored the association of Bcl-2 within the ANT-containing protein complex (Figure 3b).

Proteomic analysis identifies new putative regulators of PTPC

Several nonmitochondrial proteins such as protein kinases (Majumder et al., 2001), p53 (Marchenko et al., 2000), TR3/NUR77 (Li et al., 2000) and IB- (Bottero et al., 2001) have been found to translocate to the mitochondria and to modulate (directly or not) MMP during cell death in a variety of apoptosis induction models. To identify new protein interactions occurring within the PTPC of apoptotic cells, we designed a co-immunoprecipitation approach followed by a proteomic analysis of individual spots. Mitochondrial proteins from control and Etop-treated HT29 cells were immunoprecipitated with the anti-ANT polyclonal serum and analysed by 2D SDS–PAGE (Figure 4a), followed by mass spectroscopy identification of individual proteins found within the ANT complex, normalization in protein content of all gels and quantification of each spot as described in Material and methods. The identity of 18 proteins and their relative quantities were established by MALDI–TOF MS (Tables 1 and 2). New partners of PTPC proteins interacting with ANT in non-apoptotic cells were identified. Such proteins include the mitochondrial carrier UCP-1, some phosphatases, the phospholipase PLC4, two subunits of NADH ubiquinone oxidoreductase from the respiratory chain, three SH-modifying enzymes (GST A3-3, thiosulfate sulfurtransferase, CIC-1, chloride channel), seven proteins from intracellular signaling pathways, annexin A2 and two oncoprotein-associated proteins, p53-regulated sestrin 1 and BRCA1-associated RING domain protein 1. To assess changes in the proteome that occurred with exposure of the cells to chemotherapy, we compared the 2D SDS–PAGE patterns of proteins of ANT complex from untreated and treated cells by Etop at various times (Figure 4a). Until 20 h after Etop treatment, the ANT complex manifested a decrease in annexin A2 (-4.6-fold) and IP3 receptor content (-2.7-fold), yet an increase in BARD-1 content (+3.6-fold) (Figure 4a, Table 2a). Five spots harbored an increased intensity early after treatment (12 and 16 h): UCP-1 (+1.6; +4.6-fold), a GSH transferase, GST A3-3 (+2.8; +1.6-fold), a JNK- or MAP-stimulated phosphatase (+4; +7-fold), ARNO3/Grp1 general receptor of phosphoinositide (+1.4; +2.2-fold) as well as the 20 kDa subunit of NADH ubiquinone oxidoreductase (+1.4; 1.9-fold). In contrast, three spots decreased in intensity early after Etop treatment and increased at 36 h, suggesting an enhanced interaction with the ANT complex. These spots were identified as proteins of the intermediate signaling pathways, RYR3 (-4.9 vs +1.3-fold), RYR2 (-4.6 vs +1.4-fold) and PTPRF (-3.7 vs +2.4-fold). In addition, we found that the interaction of the other proteins with ANT complex was variable during Etop treatment (Table 2a). These proteins were identified as the 17 kDa subunit of respiratory chain proteins (from 0 to +1-fold), UCP-1 (from 0 to +4.6-fold), the -subunit of PP2A (from -2.9 to 3-fold), PLC4 (from -2.5 to +1.7-fold), the CIC-1 chloride channel (from -3.7 to +1.8-fold), GAP1 (from -2.7 to +4.9-fold), thiosulfate sulfurtransferase (from -1.7 to +3.9-fold) and sestrin 1 (from -1.3 to +2-fold). At 36 h after Etop treatment, the ANT complex exhibited a minimal proapoptotic composition, including IP3R, GAP1, the -subunit of PP2A, RYR3, RYR2, PTPRF, sestrin 1 and BARD-1.

Figure 4.

2D analysis of proteins of the ANT interactome. (a) Mitochondria from HT29 cells treated with 100 M Etop (T12-36) or untreated cells (Co) were isolated at various times after treatment. Co-immunoprecipitation was then performed as described in Figure 1, and the proteins were analysed by 2D SDS–PAGE and identified by Coomassie blue staining. Circles represent proteins identified by mass spectroscopy and indicated in Table 1. (b) HT29 cells were treated with 100 M Etop or untreated cells (Co) for 20 h in the presence or absence (Etop) of various inhibitors, z-VAD.fmk (z-VAD), furo or CsA. Cells were pretreated with the inhibitor 30 min at 37°C before the addition of Etop. Mitochondria were then purified and co-immunoprecipitated with the anti-rat heart ANT polyclonal serum, and proteins were analysed by 2D SDS–PAGE and identified by Coomassie blue staining

Full figure and legend (145K)

Table 1 - Peptide mass fingerprinting identification of proteins.

Full table

Table 2 - Proteomic analysis of the ANT interactome during Etop-induced apoptosis.

Full table

Pretreatment with z-VAD.fmk or CsA reverses the effect of Etop on all protein associations with ANT and restored the proteomic profile of immunoprecipitated proteins from untreated cells (Figure 4b). In addition, pretreatment of HT29 cells with Furo, which inhibits Bax translocation (Karpinich et al., 2002) and reduced Bax association with ANT complex (Figure 3b), decreased the ANT-complex association of eight proteins, UCP-1, the two oxidoreductases, GST-A3-3, IP3 receptor, PTPRF and ARNO3/Grp1 (Figure 4b, Table 2b), improved the association of sestrin 1 and GAP1 (IP4BP) and had no effect on the association of eight other proteins (Table 2).

GSH depletion increases the apoptotic effect of chemotherapeutic drugs

GST, a protein catalysing the conjugation of reduced GSH with a variety of electrophile (Hayes and Pulford, 1995), is one of the main antioxidant systems in mammalian cells. As shown above, association of GST with the ANT interactome decreased under apoptosis induction by Etop (-11-fold), suggesting that GST and/or GSH could have a role in this process. To address this hypothesis, we comparatively examined the capacity of a GSH synthesis inhibitor L-buthionine-(S,R)-sulfoximine (BSO), to modulate cell death and, notably, reactive oxygen species (ROS) production and m loss in HT29 cells in response to different anti-cancer drugs. Indeed, BSO reportedly induces a depletion of 95% intracellular GSH, an enhanced content in ROS and mitochondrial permeability induced by cytotoxic compounds in HL60 cells (Armstrong and Jones, 2002). Drugs were selected on the basis of their capacity to induce apoptosis via a direct effect on ANT (lonidamine (LND), arsenite trioxide (ARS), melphalan (MLP)) or via an indirect effect such as staurosporine (STS) (Larochette et al., 1999; Ravagnan et al., 1999; Belzacq et al., 2001; Belzacq and Brenner, 2004). HT29 cells pretreated with BSO, were exposed to STS, LND, ARS, MLP or Etop, and two apoptosis-related parameters were measured: the dissipation of m (DiOC₍₆₎3) and the ROS cell content as a measure of the oxidation of hydroethidine (HE) into ethidium. BSO-mediated GSH depletion potentiated the m loss induced by LND (+252%), ARS (+212%) and Etop (+173%), but not significantly the ROS generation (+31, 51 and 81%, respectively) (Figure 5). BSO treatment enhanced dramatically the ROS accumulation (+272%), as well as m loss (+403%), induced by MLP (Figure 5). In contrast, neither m (+1%) nor ROS (+81%) induced by STS were modulated by the depletion of GSH. This suggests that the redox status of cells, and notably their content in GSH, may modulate ANT-mediated cell death and, therefore, the outcome of chemotherapy-treated cells.

Figure 5.

GSH depletion increases the apoptotic effect of chemotherapeutic drugs. HT29 cells were exposed for 20 h to100 M Etop (Etop) or for 36 h to 750 M LND (LND), 100 M arsenite (Ars), 50 M MLP (MLP) or 50 nM STS (STS), preceded by a treatment of GSH-depleting agent (+BSO) for 24 h or not (-BSO). Cells were then stained with DiOC6(3) and HE and analysed by followed by flow cytometry. Numbers indicate the percentage of total cells found in each quadrant. Data are representative of three experiments

Full figure and legend (195K)

GST is a PTPC repressor

The GST A3 isoenzyme has been identified by MALDI–TOF analysis on the basis of four internal peptides generated by trypsinolysis (positions 14–33, 121–127, 121–129 and 205–218) covering 19% of the whole protein (43aa/222aa). Within the GST A class, sequence homology between GST A3-3 is 90% with GST A1-1; 88% with GST A2-2; 85% with GST A5-5% and 54% with GSTA 4-4. However, GST homologies fall less than 25% between classes. Among the class A, these isoforms possess distinct tissue and substrate specificities, and one of them (GSTA4-4) has been reported to be mitochondrial (Gardner and Gallagher, 2001). Thus, based on our proteomic data, we propose that ANT can be associated with GSTA3-3 in physiological conditions, and that a fraction of GSTA3 could be mitochondrial. However, we cannot exclude that the ANT-associated GST is actually another GST isoenzyme.

To confirm the association of GST with PTPC in normal tissue and to determine its role in MMP, we purified rat brain PTPC from isolated mitochondria (Marzo et al., 1998a; Brenner et al., 2000b). First, we determined the presence of GST in protein samples collected from the subsequent steps of the PTPC and ANT purification procedures: brain homogenate, Triton X-100-extracted mitochondrial proteins, prepurified PTPC-containing fraction and purified ANT by immunoblotting, and showed that GST could be detected in all samples collected from the PTPC purification protocol (Figure 6b), but not in the purified ANT-containing fraction. Moreover, the copurified mitochondrial GST was functional as evidenced by the catalysis of the conjugation of GSH to 1-chloro-2,4-dinitrobenzene (Figure 6a). Then, the prepurified PTPC was reconstituted into liposome (Brenner et al., 2000b). Upon treatment with Ca²⁺, a classical MPT inducer, or with atractyloside (Atr), an ANT ligand, PTPC proteoliposomes release liposome-encapsulated 4-methyl-umbelliferylphosphate (4-MUP) in a dose-dependent manner (Figure 6c and d). This release was inhibited by the addition of exogenous GSH (Figure 6c and d, +Atr or +Ca²⁺+GSH), but was not prevent by the presence of GST alone (Figure 6c and d, +Atr or +Ca²⁺-GSH). A similar experiment was repeated using liposomes containing purified ANT instead of the entire PTPC (Figure 6e). When GSH and recombinant GST were added to the ANT proteoliposomes, we observed that the combination of the two compounds (but none of these agents alone) inhibited the permeabilizing effect of Atr and Ca²⁺. This indicates that GST alone is not able to modulate the pore function of PTPC and ANT and requires the presence of GSH to repress the pore opening function.

Figure 6.

Functional interaction of PTPC with mitochondrial GST. (a) A mitochondrial GST activity copurifies with PTPC and is functional. GST specific activity (M/min/mg prot), HK specific activity (M/min/mg prot) and protein content (mg/ml) was determined in elution fractions of PTPC chromatographic purification. (b) Identification of a mitochondrial GST by Western blot. Proteins collected from several steps of the PTPC and ANT purification procedures were separated by SDS–PAGE, and GST presence was detected by immunoblot with a goat monoclonal anti-GST antibody. H, 50 g of rat brain homogenate proteins; T, 20 g of rat brain proteins solubilized with Triton X-100; 5 g of rat heart ANT, PTPC, 5 g of proteins from the permeability transition pore purified by ion-exchange chromatography. (c) GSH modification of PTPC proteins inhibits Atr-induced pore formation of PTPC in proteoliposomes. The percentage of release of 4-MUP was measured as described in Materials and methods. (d) GSH modification of PTPC inhibits Ca²⁺-induced pore formation of PTPC. (e) GSH modification prevents Atr- and Ca²⁺-induced pore formation of ANT. PTPC or ANT proteoliposomes were treated for 30 min with GSH alone or +GST and for 1 h with Atr or Ca²⁺. Then, the percentage of release of 4-MUP was measured. Results are representative of three experiments

Full figure and legend (116K)

Discussion

PTPC composition in normal conditions

Using a combination of co-immunoprecipitation from normal tissues or cancer cell lines and proteomic analysis, we identified some of the protein–protein interactions occurring within the PTPC, namely the ANT interactome. This methodology presents the advantage to respect endogenous ratios of proteins as well as the physicochemical environment of these proteins, in comparison with in vitro screening of protein–protein interactions, such as pull down of recombinant proteins or yeast two-hybrid system. Independent of the tissue or the cell line considered, VDAC, ANT and CypD are constitutively expressed in mitochondrial membranes and were always found associated with each other at similar levels (Figure 1). In contrast, a set of other proteins such as HK, Bax and Bcl-2 can be also found within the ANT complex, their respective ratios depending on the tissue specificity of their expression or on their mitochondrial association (Figure 1). Finally, in agreement with previous models (Woodfield et al., 1998; Crompton, 1999), the minimal interaction found whatever the origin of mitochondria in cell culture or ex vivo is the interaction ANT-VDAC-CypD, underscoring the importance of this interaction for the physiological role of PTPC.

How can PTPC mediate the Bax/Bcl-2-dependent MMP?

The mechanism through which proapoptotic members of Bcl-2 family (Bax, Bak, Bad, Bid, etc.) induces the release of Cyt c, AIF and other mitochondrial intermembrane space proteins members are still elusive. Among various models, it has been proposed that they permeabilize the OM by forming oligomer (Martinou et al., 2000). In contrast, we and other groups proposed that the proapoptotic molecule Bax can cooperate with PTPC members (ANT and/or VDAC) to induce MMP (and subsequent cell death) by the formation of composite channels (Marzo et al., 1998a; 1998b; Shimizu et al., 1999; Brenner et al., 2000a). As a possibility, Bcl-2 can inhibit ANT pores (Marzo et al., 1998a; 1998b; Shimizu et al., 2000b) and stimulates the exchange of ADP and ATP by ANT (Belzacq et al., 2003).

In the present study, we correlated the cascade of apoptosis induction to changes affecting the composition of the ANT interactome (Figures 2, 3 and 4). Indeed, during Etop treatment, some protein–protein interactions were maintained (e.g. ANT-VDAC), some increased (e.g. ANT-Bax, -HK), whereas others decreased (e.g. ANT-Bcl-2) (Figure 3). Focusing on the kinetic evolution of the ANT-Bax-Bcl-2 interaction, we observed that such changes coincided with the m dissipation (Figure 2a and b), mitochondrial Bax translocation (Figure 2c) and preceded the Cyt c release (Figure 2c) and nuclear chromatin condensation (Figure 2a). Thus, the proapoptotic ANT-Bax interaction increases at the same time as Bax translocates to the mitochondria, exposes its N-terminus and dissociates the ANT-Bcl-2 interaction. Using three pharmacological inhibitors of apoptosis (Figures 3 and 4), at least five critical events accompanying the process of ANT-Bax cooperation were identified. Indeed, as shown in Figure 3b, Bax–ANT proapoptotic interaction requires a relocalization of Bax in mitochondrial membrane (inhibited by Furo), the association of ANT with CypD (inhibited by CsA) and an (unknown) caspase-dependent proteolytic event (inhibited by z-VAD.fmk). In contrast, the concomitant dissociation of Bcl-2 from ANT was not modulated by z-VAD.fmk, indicating that the process is independent of caspase. Then, it remains elusive whether the signal(s) triggering the dissociation of the ANT-Bcl-2 could be an alteration in intracellular pH or ionic strength, generation of ROS or yet another biochemical event.

GST repression of PTPC and ANT pore opening

To take advantage of our proteomic approach, we characterized the regulation of PTPC by GST. The main function of GST is the detoxification of xenobiotics or products of oxidative stress. GST expression levels determine the propensity of developing bladder, colon, skin and lung cancer (Hayes and Pulford, 1995). Here, we show that, in a normal tissue (Figure 6) and cancer cell lines (Figure 4), PTPC is associated with one of the GST isoenzymes and that the interaction is progressively lost after apoptosis induction. Three arguments support the hypothesis that GST could function as an endogenous repressor of in vivo and in vitro opening of PTPC. First, apoptosis induction by chemotherapeutic agents targeting ANT such as LND (Ravagnan et al., 1999; Belzacq et al., 2001), MLP (Belzacq and Brenner, 2004) and arsenite (Larochette et al., 1999; Belzacq et al., 2001) is stimulated by GSH depletion, whereas apoptosis induced by an agent acting not directly on PTPC, such as STS, is not influenced by GSH intracellular concentration (Figure 5). Second, direct addition of GSH on PTPC liposomes (that contain a GST in an active form) or a combined addition of GSH plus recombinant GST on ANT liposomes inhibit pore opening induced by Ca²⁺ and Atr (Figure 6). Three putative cysteines of ANT, Cys(56), Cys(159) and Cys(256), could be the targets of GSH-mediated crosslinking on matrix-facing loops of the ANT. These residues face the mitochondrial matrix and influence CypD binding, at least in conditions of oxidative stress (Halestrap and Brenner, 2003). This would be in accord with the local topology of GST A3.3, which is likely to be located in the mitochondrial matrix, yet requires further experimental verification. Third, GST have also been shown to modulate protein functions such as stress kinase activity (Adler et al., 1999; Yin et al., 2000; Gilot et al., 2002) and RYR channel activity (Dulhunty et al., 2001) by protein–protein association, independent of any GSH-conjugation activity. Thus, as a GSH modification of ANT has not been evidenced, GST could also repress PTPC and ANT pore function by physical interactions without enzymatic modification, notably by interaction with electrophilic residues. Additional investigations are now required to decipher whether the mechanism of action GSH is direct or not and dependent or not of GST.

How could PTPC integrate multiple death signals?

In physiological conditions, PTPC contributes to maintain mitochondrial homeostasis by allowing the free diffusion of small solutes (MM <1.5 kDa) between the matrix and the cytosol. PTPC participates also to the general metabolism, allowing the ADP/ATP exchange (e.g. channeling of ADP and ATP through ANT and VDAC for CK and HK activity) and the lipidic metabolism (e.g. PBR). This role is facilitated by Bcl-2 that favors the vital function of ANT by preventing its transition as a lethal pore (Belzacq et al., 2003). Moreover, even if the whole relevance of ANT–GST interaction remains to be explored, GST that interacts in normal conditions with PTPC may favor the stabilization of ANT as an ADP/ATP carrier and, in turn, the metabolic function of PTPC. Nevertheless, accumulating evidence also suggests that the PTPC constitutes a crossroad of apoptosis regulation via its capacity of integrating multiple pathways of cell death (Dorner et al., 1997). Indeed, PTPC pore opening has been implicated in physiological or pathological cell death such as ischemia–reperfusion, neuronal cell death, toxins and ROS-induced death and cancer. Our approach led us to identify three classes of proteins that may account for PTPC ability to switch from a physiological role to a lethal function (Figure 7). First, in normal conditions, PTPC, and more precisely ANT, is connected to an array of proteins involved in the general mitochondrial metabolism (e.g. UCP1, the dicarboxylate carrier, 17 and 20 kDa subunits of the NADH ubiquinone oxidoreductase, SH-modifying enzymes, Table 1), in intracellular signaling pathways (e.g. IP3/calcium, phosphatase and phospholipase-mediated pathways, Table 1) and in stress-activated pathways (e.g. DNA damage response and p53-regulated stress pathway, Table 1). No evidence of interaction of ANT with UCP2, which has been implicated in the control of cellular redox signaling, was found. After Etop treatment, these interactions evolved strongly to favor apoptosis, suggesting the hypothesis that antiapopoptotic interactions disappeared to the profit of proapoptotic interactions. For example, the decrease of interaction of ANT with Bcl-2 and GAP1, an inhibitory protein of the Ras pathway, is consistent with the capacity of mitochondrial Ras to induce a Bcl-2-dependent apoptosis (Denis et al., 2003).

Figure 7.

Scheme of the regulation of the ANT interactome during apoptosis. The dynamic changes in the ANT interactome in response to apoptosis induction reflect how PTPC proteins sense intracellular alterations, adopt a lethal conformation via changes in protein–protein interactions and trigger cell death. In physiological conditions, PTPC proteins are linked to proteins from the general metabolism and to proteins from a variety of intracellular pathways to allow crossorganelles communication. After stimulation by a death signal, PTPC would lose these interactions and build up new links with stress response pathways and specific intracellular signaling pathways. Thus, PTPC would integrate multiple injury signals and coordinate its response by modulating specifically its protein–protein interactions

Full figure and legend (86K)

Moreover, a loss of interaction of ANT with key proteins involved in the metabolism such as mitochondrial carriers or proteins involved in the respiratory chain (e.g. 1; dicarboxylate carriers, two subunits of the NADH oxidoreductase) could be linked to the bioenergetic catastrophe that accompanied the execution of MMP and the release of soluble intermembrane proteins into the cytosol (Kroemer and Reed, 2000). Indeed, the lethal opening of PTPC is associated with the collapse of the mitochondrial m, uncoupling of the respiratory chain, hyperproduction of superoxide anions, disruption of mitochondrial biogenesis and outflow of matrix calcium and GSH, which reflect a profound remodeling of mitochondria during apoptosis (Kroemer and Reed, 2000; Reed and Green, 2002; Gottlieb et al., 2003).

Conversely, the increase in certain specific interactions of proteins with ANT can be explained on the basis that they promote the apoptosis program. Thus, the increase of proteins from the IP3/calcium pathway within the ANT interactome could be explained (i) by the sensitivity of PTPC opening to respond to calcium signaling (Ichas et al., 1997; Ichas and Mazat, 1998; Jouaville et al., 1998; Haouzi et al., 2001; 2002), notably to IP3R-mediated calcium waves induced by release from the endoplasmic reticulum (Hajnoczky et al., 2000) and (ii) by the activation of proteins from the endoplasmic reticulum during Etop-induced apoptosis, such as GRP78 and caspase-7 (Reddy et al., 2003). Similarly, the increase in ANT interaction with PTPRF protein, a protein that activates the GDP/GTP exchange by Rac1, is coherent with the capacity of Rac1 to modulate apoptosis (Murga et al., 2002).

In conclusion, these dynamic changes in the ANT interactome, occurring concomitantly with the loss of m, may reflect how PTPC proteins sense intracellular alterations, adopt a lethal conformation and trigger cell death (see model, Figure 7). Therefore, we propose that PTPC could integrate multiple injury signals and that changing its intraprotein–protein interactions orchestrate the lethal response.

Materials and methods

Isolation of tissue and cell line mitochondria

Rat liver mitochondria were purified as described previously (Susin et al., 2000). Briefly, rat liver was homogenized in 0.3 M saccharose, 5 mM TES, 0.2 mM EGTA (pH 7.2) and mitochondria were isolated by differential centrifugation (2 10 min, 500 g and 10 min, 10 000 g at 4°C). The resulting pellet was resuspended and mitochondria were purified on a Percoll gradient (18, 30 and 60% (v/v) Percoll in 0.3 M saccharose, 10 mM TES, 0.2 mM EGTA (pH 6.9)) and spun down at 10 000 g for 10 min. Mitochondria were collected between the 30 and 60% layers, diluted 10 times in 0.3 M saccharose, 5 mM TES, 0.2 mM EGTA (pH 7.2) and washed by centrifugation at 8000 g for 10 min. Rat heart mitochondria were isolated following the published protocol (Marzo et al., 1998b). After tissue disruption, mitochondria were suspended in 220 mM mannitol, 70 mM saccharose, 10 mM HEPES, 0.2 mM EDTA and 0.5 mg/ml subtilisin (Sigma) (pH 7.4), kept on ice for 8 min and sedimented twice by differential centrifugation (5 min, 500 g and 10 min, 10 000 g at 4°C). Rat brain mitochondria were prepared as described previously (Beutner et al., 1996). Rat brains were homogenized in 1 mM -monothioglycerol, 10 mM glucose (pH 8.0) and centrifuged for 10 min at 2000 g and for 10 min at 12 000 g at 4°C. HT29, HeLa and MCF7 cells were recovered by trypsinization and washed in PBS and resuspended in 10 mM Tris-HCl, 0.15 mM MgCl2 and 10 mM KCl (pH 7.6) containing 0.4 mM phenylmethylsulfonyl fluoride. Cells were broken with a Dounce homogenizer at 4°C, sedimented twice by differential centrifugation (5 min, 500 g and 10 min, 6800 g at 4°C) and resuspended at 0.4 mg protein/ml (Belzacq et al., 2003).

Cell culture and treatments

HT29 cells, HeLa and MCF-7 cells were cultured in DMEM medium supplemented with 10% heat-inactivated FCS and antibiotics at 37°C under 5% CO2. To induce apoptosis, 3 10⁵ of HT29 cells/ml were treated with Etop (Sigma, St Louis, MO, USA), STS (Sigma), ARS (Sigma), MLP (Sigma) and LND (generous gift from Dr MF Poupon, Institut Curie, Paris, France) for various periods of time at 37°C. When indicated, cells were also pretreated 30 min with the following compounds: Furo (Sigma), CsA ( BIOMOL, Research Labs, Plymouth Meeting, PA, USA) and z-VAD.fmk (BIOMOL, Research Labs). For GSH depletion, BSO (Sigma) was dissolved and added to cells at a final concentration of 0.2 mM for 24 h before adding the apoptotic inducer (Haouzi et al., 2002). In order to check the efficiency of GSH depletion, GSH level was measured in untreated cells as well as cells treated with BSO. Thus, cells were incubated for 30 min with monochlorobimane (MCB; Molecular Probes, Leiden, The Netherlands), which forms a fluorescent GSH–MCB adduct measurable by flow cytometry. A decrease in fluorescence was detected in 90% BSO-exposed cells.

Cytofluorometry

For flow cytometric analysis, HT29 cells were harvested, resuspended in fresh complete DMEM medium supplemented with the appropriate fluorescent probes at 20 nM DiOC(6)3 (Molecular Probes) and 2 M dihydroethidine (HE; Molecular Probes) and incubated 15 min at 37°C (Jan et al., 2002). For the detection of apoptotic DNA loss, cells were harvested, fixed with cold 70% ethanol, washed three times with PBS and stained with 5 g/ml PI (Jan et al., 2002; Molecular Probes). For immunostaining, cells were harvested, fixed with paraformaldehyde (0.25% (wt/vol)) and incubated for 1 h at room temperature with the mouse monoclonal antibody anti-Bax 6A7 (Desagher et al., 1999; BD Biosciences; dilution 1 : 100) in PBS, 1 mg/ml BSA and 100 g/ml digitonine. After three washes in PBS 1 mg/ml BSA, cells were incubated for 1 h at room temperature with the fluorescein-labeled goat anti-mouse IgG (Pharmingen BD Biosciences; dilution 1 : 300), washed three time with PBS before being analysed by flow cytometry (FACSVantage, Becton Dickinson, San Jose, CA, USA).

Fluorescence microscopy

For assessment of mitochondrial and nuclear features of apoptosis, cells cultured on a coverslip were stained with the m-sensitive fluorochrome, 5,5',6,6'-tetrachloro-1,1', 3,3'-tetraethylbenzimidazolylcarbocyanine iodide (2 M, JC-1; Molecular Probes) and with the DNA-intercalating Hoechst 33342 (2 M, Sigma) diluted in a complete medium for 15 min incubation at 37°C as previously described (Jan et al., 2002). After 15 min incubation at 37°C, cells were examined with a fluorescence microscope (Leica; DMRB type). For immunostaining, HT29 cells cultured on coverslips were fixed with paraformaldehyde (4% (wt/vol)) and picric acid fixed (0.19% (vol/vol)) and permeabilized as described previously (Castedo et al., 2002) and stained with a mouse monoclonal antibody anti-Bax 6A7 (Desagher et al., 1999) and a rabbit polyclonal antibody raised against Cyt c (H-104; Santa Cruz Biotechnology, Santa Cruz, CA, USA) (dilution 1 : 100, each) and revealed with Alexa fluor 568 goat anti-mouse IgG1 (dilution 1 : 100; Molecular Probes) and with the fluorescein-labeled goat anti-rabbit IgG (dilution 1 : 300). Cells were then examined with a fluorescence microscope (Leica; DMRB type).

PTPC and ANT proteoliposomes preparation and pore opening analysis

PTPC was purified from rat brains and ANT from rat hearts as described previously and reconstituted into proteoliposomes by the surfactant dialysis method (Marzo et al., 1998a; 1998b; Brenner et al., 2000b). Proteoliposomes were loaded with 4-MUP (Brenner et al., 2000b) in 10 mM KCl, 10 mM HEPES, 125 mM saccharose (pH 7.4), by sonication (20% of 250 W, 22 s on ice, Branson sonifier 250), washed on Sephadex PD-10 columns (Pharmacia, Uppsala, Sweden), dispended in 96-well microtiter plates and incubated with various doses of Ca²⁺ at RT. The release of 4-MUP was quantified by the addition of alkaline phosphatase (which converts 4-MUP into the fluorochrome 4-methylumbelliferone) and fluorescence recording at exc 360 nm and em 450 nm with a spectrofluorimeter (TECAN, Austria GmbH, Salzburg, Austria). The amount of encapsulated 4-MUP was determined by adding 0.5% Triton X-100 to proteoliposomes. The maximal fluorescence induced by Ca²⁺ was then identified as 100% 4-MUP release and the fluorescence induced by the pretreatment of liposomes by GSH was calculated as a percentage of Ca²⁺- or Atr-induced 4-MUP release.

Enzymatic assays

During PTPC purification, the presence of HK activity was detected in rat brain homogenate, Triton X-100 supernatant or chromatography elution fraction as described elsewhere (Brenner et al., 2000b). A measure of 2, 5 or 10 l of sample were mixed with 100 l of reaction buffer (100 mM thriethanolamine, 10 mM EDTA, 16 mM MgSO47H2O, 150 mM KCl (pH 7.6)), 5 l glucose 0.1 M, 5 l NADP 20 mg/ml, 5 l glucose-6-phosphate deshydrogenase (Roche) and 5 l ATP 0.1 M and qsp 200 l in a microtiter plate well. Similarly, the presence of GST activity was determined by mixing 2–10 l of sample with 6 l of 20 mM GSH and 3 l of 40 mM CDNB, qsp 200 l PBS buffer (pH7.4) for purification fractions or qsp 200 l 10 mM HEPES and 125 mM saccharose (pH 7.4) for proteoliposomes. Absorbance was then monitored at 340 nm for 20 min with a microtiter plate reader (TECAN).

Co-immunoprecipitation assays

HT29 cells (5 10⁶ cells/75 cm² flask) were treated with 100 M Etop for different periods of time to induce apoptosis (Belzacq et al., 2003) and mitochondria were isolated and resuspended in 10 mM Tris-HCl, 0.15 mM MgCl₂, 10 mM KCl (pH 7.6) containing 0.5% Triton X-100 and 0.4 mM phenylmethylsulfonyl fluoride at a final concentration of 0.4 mg protein/ml. Subsequently, 20 l of a rabbit polyclonal anti-rat heart ANT serum (Belzacq et al., 2003) was added to 100 l of mitochondrial suspension (90 min, 37°C) to co-immunoprecipitate ANT and ANT-interacting proteins. Then, 40 l of proteinA/proteinG agarose beads (Santa-Cruz Biotechnology) were added (30 min, 37°C). The beads were washed three times with 1 ml PBS and resuspended in SDS–PAGE sample buffer prior to electrophoresis.

Protein one-dimensional electrophoresis and Western blot

Proteins were analysed by SDS–PAGE (12.5%, 25 g protein/lane) according to Laemmli (Laemmli, 1970) and immunoblotting with anti-rat heart ANT polyclonal serum (Belzacq et al., 2001) or antibodies specific for Bax (B9; Santa-Cruz Biotechnology, dilution 1 : 250), Cyt c (7H8.2C12, BD Pharmingen, San Diego, CA, USA, dilution 1 : 500), Bcl-2 (C 21 or C2; Santa-Cruz Biotechnology, dilution 1 : 250), VDAC (31HL, ab-4; Calbiochem Corp., La Jolla, CA, USa, dilution 1 : 1000), HK I (Chemicon International, Temecula, CA, USA, dilution 1 : 1000), CypD (42–54; Calbiochem, dilution 1 : 100), PBR (Trevigen, dilution 1 : 100) and Cox II (Molecular Probes, dilution 1 : 100). Proteins were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech, Rockford, IL, USA). To avoid strong signals due to the crossrevelation of heavy and light chains of immunoglobulin, membranes were cut around 55 and 25 kDa before ECL revelation.

Two-dimensional electrophoresis

Linear gradient pH 3–10 Dry Strips (Amersham Pharmacia Biotech, Upsalla, Sweden) were used in this study. Air-dried protein pellets were solubilized in sample solution containing 7 M urea, 2 M thiourea, 25 mM DTT, 4% CHAPS and 2% IPG buffer (Amersham Pharmacia Biotech). The surfactant CHAPS and the chaotropic thiourea were used throughout the isoelectric focusing to improve protein solubility and transfer to the second dimension. Equal amounts of proteins, 200 g, were loaded onto the gels in the first dimension. Isoelectric focusing was carried out using Immobiline Dry Strips on a Multiphor II electrophoresis system (Amersham Pharmacia Biotech). The strips were equilibrated in the presence of DTT followed by iodoacetamide before being placed on the top of the second dimension gel. SDS–PAGE (12.5%) was performed according to Laemmli (Zoratti and Szabo, 1994) in 16 16 0.1 cm³ slab gels (Protean II, Bio-Rad, Hercules, USA). For protein quantitation, we used a colloidal Coomassie G250 staining method. All the gels were then scanned using a calibrated imaging densitometer (GS-710, Bio-Rad) with a dynamic range from 0 to 3 U of OD and a resolution of 42 m. Image analysis, gel matching and quantification of the protein amounts in individual spots were performed using the Melanie II software (Bio-Rad). Individual spots were automatically detected. A volume was attributed to each spot, which takes into account both the size and the intensity of the spot. This was carried out for all the proteins detected on the gel, including the heavy chain of antibodies. The Melanie software then calculated, for each protein spot, the relative volume (%Vol), which is defined as the ratio between the volume of the particular spot and the total sum of all the volume on the gel. This normalized value is thus independent of variations between gels, particularly due to protein loading and staining. The ratio of %Vol, for a given spot, between two similar gels, was always below 1.1. In our differential studies, proteins were considered to be significantly changed if the mean %Vol for an individual protein was at least 1.2-fold greater than the control one. Molecular weights were calibrated by comigrating LMW markers (94.0, 67.0, 43.0, 30.0, 20.1 and 14.4 kDa) and peptide markers (16.9, 14.4, 10.700, 8.1, 6.2 and 2.5 kDa) with bacterial proteins. Similarly, isoelectric points were calibrated using the broad pI kit (isoelectric points 4.55, 5.20, 5.85, 6.55 and 6.85 were resolved on such gels). All markers were from Amersham Pharmacia Biotech. Some members of PTPC harboring a basic pI (>9) were not resolved in 2D gels and were identified in 1D PAGE, by immunoblotting and MALDI-TOF MS analysis. Results are the mean of at least three independent experiments.

Peptide mass fingerprinting using MALDI mass spectrometry

In-gel tryptic digestion of 2D protein spots was as follows: gel pieces were washed twice in 25 mM ammonium bicarbonate buffer (pH 8.0) containing 50% (v/v) acetonitrile prior to vacuum-drying. Rehydration was performed in 50 mM ammonium bicarbonate buffer (pH 8.0) containing 0.02 g/l sequencing grade modified porcine trypsin (Promega, Madison, WI, USA). Trypsin digestion was performed for 18 h in a thermomixer (Eppendorf, Hamburg, Germany) at 37°C and 500 r.p.m. and stopped by the addition of 0.4% (v/v) trifluoroacetic acid. The resulting peptides were recovered in the supernatant and spotted directly onto a MALDI plate. A 0.5 l aliquot was allowed to dry at room temperature before the addition of a 0.5 l aliquot of the matrix solution. This dried-droplet sampling method was employed using a solution at 5 g/l, prepared fresh daily, of matrix -cyano-4-hydroxycinnamic acid in 50% (v/v) acetonitrile containing 0.1% (v/v) trifluororacetic acid. Mass spectra were acquired on a Voyager-DE-STR time-of-flight mass spectrometer (Applied Biosystems, Framingham, MA, USA) equipped with a nitrogen laser (Laser Science, Franklin, MA, USA; emitting at =337 nm). The accelerating voltage used was 20 kV. All spectra were recorded in positive reflector mode with a delayed extraction of 130 ns and a 62% grid voltage. The spectra were calibrated using an external calibration which was composed of: Des-Arg Bradykinin (M+H)+=904.4 kDa, angiotensin (M+H)+=1296.6 kDa, neurotensin (M+H)+=1672.9 kDa, melittin (M+H)+=2845.7 kDa and insulin B chain bovin (M+H)+=3494.6 kDa and an internal calibration with tryptic autodigestion ion peaks. Peptide masses were queried against entries for mammals in the NCBI nr database using the Mascot peptide mass finger-printing program from Matrix Science (http://www.matrixscience.com). For a match to be considered, a minimum of three matching peptides was required. Modification of cysteines by carbamidomethylation in addition to possible modification by acrylamide were considered during the searches. For protein quantitation, we used a colloidal Coomassie G250 staining method. All the gels were then scanned using a calibrated imaging densitometer (GS-710, Bio-Rad) with a dynamic range from 0 to 3 U of OD and a resolution of 42 m. Individual spots were automatically detected using the Melanie software. A volume was attributed to each spot, which takes into account both the size and the intensity of the spot. This was carried out for all the proteins detected on the gel, including the heavy chain of antibodies. The Melanie software then calculated, for each protein spot, the relative volume (%Vol), which is defined as the ratio between the %Vol of the particular spot and the total sum of all the %Vol on the gel. This normalized value is thus independent of variations between gels, particularly due to protein loading and staining. The ratio of %Vol, for a given spot, between two similar gels, was always below 1.1. In our differential studies, proteins were considered to be significantly changed if the mean %Vol for an individual protein was at least 1.2-fold greater than the control one.

References

1. Adler V, Yin Z, Fuchs S, Benezra M, Rosario L, Tew K, Pincus M, Sardana M, Henderson C, Wolf C, Davis R and Ronai Z. (1999). EMBO J., 18, 1321–1334. | Article | PubMed | ISI | ChemPort |
2. Armstrong JS and Jones DP. (2002). FASEB J., 7, 7–12.
3. Belzacq AS and Brenner C. (2004). Curr. Drug Target, 4, 517–524.
4. Belzacq AS, El Hamel C, Vieira HL, Cohen I, Haouzi D, Metivier D, Marchetti P, Brenner C and Kroemer G. (2001). Oncogene, 20, 7579–7587. | Article | PubMed | ISI | ChemPort |
5. Belzacq AS, Vieira HL, Verrier F, Vandecasteele G, Cohen I, Prevost MC, Larquet E, Pariselli F, Petit PX, Kahn A, Rizzuto R, Brenner C and Kroemer G. (2003). Cancer Res., 63, 541–546. | PubMed | ISI | ChemPort |
6. Bernardi P, Colonna R, Costantini P, Eriksson O, Fontaine E, Ichas F, Massari S, Nicolli A, Petronilli V and Scorrano L. (1998). Biofactors, 8, 273–281. | PubMed | ISI | ChemPort |
7. Beutner G, Rück A, Riede B, Welte W and Brdiczka D. (1996). FEBS Lett., 396, 189–195. | Article | PubMed | ISI | ChemPort |
8. Bottero V, Rossi F, Samson M, Mari M, Hofman P and Peyron J. (2001). J. Biol. Chem., 276, 21317–21324. | Article | PubMed |
9. Brenner C, Cadiou H, Vieira HL, Zamzami N, Marzo I, Xie Z, Leber B, Andrews D, Duclohier H, Reed JC and Kroemer G. (2000a). Oncogene, 19, 329–336. | Article | PubMed | ISI | ChemPort |
10. Brenner C and Kroemer G. (2000). Science, 289, 1150–1151. | Article | PubMed | ISI | ChemPort |
11. Brenner C, Le Bras M and Kroemer G. (2003). J. Clin. Immunol., 23, 73–80. | Article | PubMed | ChemPort |
12. Brenner C, Marzo I, de Araujo Vieira HL and Kroemer G. (2000b). Methods Enzymol., 322, 243–252. | Article | PubMed | ChemPort |
13. Bribes E, Galiegue S, Bourrie B and Casellas P. (2003). Immunol. Lett., 85, 13–18. | Article | PubMed |
14. Cao G, Minami M, Pei W, Yan C, Chen D, O'Horo C, Graham SH and Chen J. (2001). J. Cereb. Blood Flow Metab., 21, 321–333. | Article | PubMed | ISI | ChemPort |
15. Capano M and Crompton M. (2002). Biochem. J., 367, 169–178. | Article | PubMed | ISI | ChemPort |
16. Castedo M, Roumier T, Blanco J, Ferri K, Barretina J, Tintignac L, Andreau K, Perfettini J, Amendola A, Nardacci R, Leduc P, Ingber D, Druillennec S, Roques B, Leibovitch S, Vilella-Bach M, Chen J, Este J, Modjtahedi N, Piacentini M and Kroemer G. (2002). EMBO J., 21, 4070–4080. | Article | PubMed | ISI | ChemPort |
17. Cheng E, Sheiko T, Fisher J, Craigen W and Korsmeyer S. (2003). Science, 301, 513–517. | Article | PubMed | ISI | ChemPort |
18. Crompton M. (1999). Biochem. J., 341, 233–249. | Article | PubMed | ISI | ChemPort |
19. Crompton M. (2000). Curr. Opin. Cell Biol., 12, 414–419. | Article | PubMed | ChemPort |
20. Crompton M, Barksby E, Johnson N and Capano M. (2002). Biochimie., 84, 143–152. | Article | PubMed | ISI | ChemPort |
21. Danial N, Gramm C, Scorrano L, Zhang C, Krauss S, Ranger A, Datta S, Greenberg M, Licklide L, Lowell BB, Gygi S and Korsmeyer S. (2003). Nature, 424, 952–956. | Article | PubMed | ISI | ChemPort |
22. De Giorgi F, Lartigue L, Bauer M, Schubert A, Grimm S, Hanson G, Remington S, Youle R and Ichas F. (2002). FASEB J., 16, 607–609. | PubMed | ChemPort |
23. Denis G, Yu Q, Ma P, Deeds L, Faller D and Chen C. (2003). J. Biol. Chem., 278, 5775–5785. | Article | PubMed | ISI | ChemPort |
24. Desagher S, Osen-Sand A, Nichols A, Eskes R, Montessuit S, Lauper S, Maundrell K, Antonsson B and Martinou JC. (1999). J. Cell Biol., 144, 891–901. | Article | PubMed | ISI | ChemPort |
25. Dorner A, Schulze K, Rauch U and Schultheiss HP. (1997). Mol. Cell Biochem., 174, 261–269. | Article | PubMed |
26. Dulhunty A, Gage P, Curtis S, Chelvanayagam G and Board P. (2001). J. Biol. Chem., 276, 3319–3323. | Article | PubMed |
27. Elmore S, Ian T, Grissom S and Lemasters J. (2001). FASEB J., 15, 2286–2287. | PubMed | ISI | ChemPort |
28. Friberg H and Wieloch T. (2002). Biochimie, 84, 241–250. | Article | PubMed | ISI | ChemPort |
29. Gardner J and Gallagher E. (2001). Arch. Biochem. Biophys., 390, 19–27. | Article | PubMed |
30. Gilot D, Loyer P, Corlu A, Glaise D, Lagadic-Gossmann D, Atfi A, Morel F, Ichijo H and Guguen-Guillouzo C. (2002). J. Biol. Chem., 277, 49220–49229. | Article | PubMed | ISI | ChemPort |
31. Gottlieb E, Armour S, Harris M and Thompson C. (2003). Cell Death Differ., 10, 709–717. | Article | PubMed | ISI | ChemPort |
32. Gunter T and Pfeiffer D. (1990). Am. J. Physiol., 258, C755–C786. | PubMed | ISI | ChemPort |
33. Hajnoczky G, Csordas G, Krishnamurthy R and Szalai G. (2000). J. Bioenerg. Biomembr., 32, 15–25. | Article | PubMed | ISI | ChemPort |
34. Halestrap A and Brenner C. (2003). Curr. Med. Chem., 10, 1507–1525. | Article | PubMed | ISI | ChemPort |
35. Halestrap AP, McStay GP and Clarke SJ. (2002). Biochimie, 84, 153–166. | Article | PubMed | ISI | ChemPort |
36. Halestrap AP, Woodfield KY and Connern CP. (1997). J. Biol. Chem., 272, 3346–3354. | Article | PubMed | ISI | ChemPort |
37. Hande K. (1998). Eur. J. Cancer, 34, 1514–1521. | Article | PubMed | ISI | ChemPort |
38. Haouzi D, Cohen I, Vieira H, Poncet D, Boya P, Castedo M, Vadrot N, Belzacq A, Fau D, Brenner C, Feldmann G and Kroemer G. (2002). Apoptosis, 7, 395–405. | Article | PubMed | ISI | ChemPort |
39. Haouzi D, Lekehal M, Tinel M, Vadrot N, Caussanel L, Letteron P, Moreau A, Feldmann G and Fau DPessayre D. (2001). Hepatology, 33, 1181–1188. | Article | PubMed | ISI | ChemPort |
40. Hayes J and Pulford D. (1995). Crit. Rev. Biochem. Mol. Biol., 30, 445–600. | PubMed | ISI | ChemPort |
41. Ichas F, Jouaville L and Mazat J. (1997). Cell, 89, 1145–1153. | Article | PubMed | ISI | ChemPort |
42. Ichas F and Mazat JP. (1998). Biochim. Biophys. Acta, 1366, 33–50. | Article | PubMed | ISI | ChemPort |
43. Jan G, Belzacq AS, Haouzi D, Rouault A, Metivier D, Kroemer G and Brenner C. (2002). Cell Death Differ., 9, 179–188. | Article | PubMed | ChemPort |
44. Jouaville LS, Ichas F and Mazat JP. (1998). Mol. Cell. Biochem., 184, 371–376. | Article | PubMed |
45. Karpinich NO, Tafani M, Rothman RJ, Russo MA and Farber JL. (2002). J. Biol. Chem., 277, 16547–16552. | Article | PubMed | ISI | ChemPort |
46. Kroemer G and Reed J. (2000). Nat. Med., 6, 513–519. | Article | PubMed | ISI | ChemPort |
47. Kroemer G, Zamzami N and Susin SA. (1997). Immunol. Today, 18, 44–51. | Article | PubMed | ISI | ChemPort |
48. Laemmli UK. (1970). Nature, 227, 680–685. | Article | PubMed | ISI | ChemPort |
49. Larochette N, Decaudin D, Jacotot E, Brenner C, Marzo I, Susin SA, Zamzami N, Xie Z, Reed J and Kroemer G. (1999). Exp. Cell Res., 249, 413–421. | Article | PubMed | ISI | ChemPort |
50. Lemasters J, Nieminen A, Qian T, Trost L and Herman B. (1997). Mol. Cell. Biochem., 174, 159–165. | Article | PubMed | ChemPort |
51. Li H, Kolluri SK, Gu J, Dawson MI, Cao X, Hobbs PD, Lin B, Chen G, Lu J, Lin F, Xie Z, Fontana JA, Reed JC and Zhang X. (2000). Science, 289, 1159–1164. | Article | PubMed | ISI | ChemPort |
52. Majumder PK, Pandey P, Sun X, Cheng K, Datta R, Saxena S, Kharbanda S and Kufe D. (2001). J. Biol. Chem., 12, 465–470.
53. Marchenko ND, Zaika A and Moll UM. (2000). J. Biol. Chem., 275, 16202–16212. | Article | PubMed | ISI | ChemPort |
54. Martinou JC, Desagher S and Antonsson B. (2000). Nat. Cell Biol., 2, E41–43. | Article | PubMed | ISI | ChemPort |
55. Marzo I, Brenner C, Zamzami N, Jurgensmeier JM, Susin SA, Vieira HL, Prevost MC, Xie Z, Matsuyama S, Reed JC and Kroemer G. (1998b). Science, 281, 2027–2031. | Article | PubMed | ISI | ChemPort |
56. Marzo I, Brenner C, Zamzami N, Susin SA, Beutner G, Brdiczka D, Remy R, Xie ZH, Reed JC and Kroemer G. (1998a). J. Exp. Med., 187, 1261–1271. | Article | PubMed | ISI | ChemPort |
57. McEnery M, Snowman A, Trifiletti R and Snyder SH. (1992). Proc. Natl. Acad. Sci. USA, 89, 3170–3174. | Article | PubMed | ChemPort |
58. Murga C, Zohar M, Teramoto H and Gutkind J. (2002). Oncogene, 21, 207–216. | Article | PubMed | ISI | ChemPort |
59. Ott M, Robertson JD, Gogvadze V, Zhivotovsky B and Orrenius S. (2002). Proc. Natl. Acad. Sci. USA, 99, 1259–1263. | Article | PubMed | ChemPort |
60. Ravagnan L, Marzo I, Costantini P, Susin SA, Zamzami N, Petit PX, Hirsch F, Goulbern M, Poupon MF, Miccoli L, Xie Z, Reed JC and Kroemer G. (1999). Oncogene, 18, 2537–2546. | Article | PubMed | ISI | ChemPort |
61. Reddy R, Mao C, Baumeister P, Austin R, Kaufman R and Lee A. (2003). J. Biol. Chem., 278, 20915–20924. | Article | PubMed | ISI | ChemPort |
62. Reed J and Green D. (2002). Mol. Cell., 9, 1–3. | Article | PubMed | ISI | ChemPort |
63. Reed JC, Jurgensmeier JM and Matsuyama S. (1998). Biochim. Biophys. Acta, 1366, 127–137. | Article | PubMed | ISI | ChemPort |
64. Robertson J, Gogvadze V, Zhivotovsky B and Orrenius S. (2000). J. Biol. Chem., 275, 32438–32443. | Article | PubMed | ISI | ChemPort |
65. Shimizu S, Ide T, Yanagida T and Tsujimoto Y. (2000a). J. Biol. Chem., 275, 12321–12325. | Article | PubMed | ISI | ChemPort |
66. Shimizu S, Konishi A, Kodama T and Tsujimoto Y. (2000b). Proc. Natl. Acad. Sci. USA, 97, 3100–3105. | Article | PubMed | ChemPort |
67. Shimizu S, Narita M and Tsujimoto Y. (1999). Nature, 399, 483–487. | Article | PubMed | ISI | ChemPort |
68. Susin SA, Larochette N, Geuskens M and Kroemer G. (2000). Methods Enzymol., 322, 205–208. | Article | PubMed |
69. Szabo I, Bernardi P and Zoratti M. (1992). J. Biol. Chem., 267, 2940–2946. | PubMed | ISI | ChemPort |
70. Szabo I and Zoratti M. (1992). J. Bioenerg. Biomembr., 24, 111–117. | Article | PubMed | ChemPort |
71. Woodfield K, Ruck A, Brdiczka D and Halestrap AP. (1998). Biochem. J., 336, 287–290. | PubMed | ISI | ChemPort |
72. Yin Z, Ivanov V, Habelhah H, Tew K and Ronai Z. (2000). Cancer Res., 60, 4053–4057. | PubMed | ISI | ChemPort |
73. Zamzami N, Marchetti P, Castedo M, Zanin C, Vayssière J-L, Petit PX and Kroemer G. (1995). J. Exp. Med., 181, 1661–1672. | Article | PubMed | ISI | ChemPort |
74. Zoratti M and Szabo I. (1994). J. Bioenerg. Biomembr., 26, 543–553. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We thank Dr D Haouzi for helpful discussion and C Henry for her help in mass spectrometry analysis. This work is supported by grants from l'Association pour la Recherche sur le Cancer (ARC), la Fondation pour la Recherche Médicale (FRM), the Ministère délégué à la Recherche et aux Nouvelles Technologies (MRNT) to CB, a special grant by the Ligue contre le Cancer to GK and from INRA, Institut National de la Recherche Agronomique to GJ. FV and AD were supported by fellowships from the MRNT. MLB receives a postdoctoral fellowship from the Centre National de la Recherche Scientifique (CNRS).