GAPDH, a novel regulator of the pro-apoptotic mitochondrial membrane permeabilization


Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is a pleiotropic enzyme that is overexpressed in apoptosis and in several human chronic pathologies. Here, we report that the protein accumulates in mitochondria during apoptosis, and induces the pro-apoptotic mitochondrial membrane permeabilization, a decisive event of the intrinsic pathway of apoptosis. GAPDH was localized by immunogold labeling and identified by matrix-assisted laser desorption/ionization-time of flight and nano liquid chromatography mass spectroscopy/mass spectroscopy in the mitochondrion of various tissues and origins. In isolated mitochondria, GAPDH can be imported and interact with the voltage-dependent anion channel (VDAC1), but not the adenine nucleotide translocase (ANT). The protein mediates a cyclosporin A-inhibitable permeability transition, characterized by a loss of the inner transmembrane potential, matrix swelling, permeabilization of the inner mitochondrial membrane and the release of two pro-apoptotic proteins, cytochrome c and apoptosis-inducing factor (AIF). This novel function of GAPDH might have implications for the understanding of mitochondrial biology, oncogenesis and apoptosis.


Glyceraldehyde-3-phosphate-dehydrogenase (GAPDH, EC. is a soluble protein that harbors multiple intracellular functions. First discovered as a glycolytic dehydrogenase, GAPDH might exert several functions as diverse as apoptosis induction, receptor-associated kinase, tRNA export or DNA repair (Sawa et al., 1997; Sirover, 1997; Laschet et al., 2004). These functions have been linked to the various intracellular localizations of the enzyme, which has been found in the cytosol, the nucleus, ER-golgi-vesiculae, mitochondria, as well as associated with the plasma membrane (Ryzlak and Pietruszko, 1988; Tisdale et al., 2004). Moreover, GAPDH binds to numerous endogenous partners (e.g. inositol-1,4,5 triphosphate (IP3R)), (Ca2+)/calmodulin-dependent protein kinase phosphatase, Rab2, tRNA, mRNA and DNA) (Tisdale et al., 2004; Ishida et al., 2005; Patterson et al., 2005) and virus-encoded proteins, mRNA and tRNA from Hepatitis B, Hepatitis A and parainfluenza viruses (Sirover, 1997).

Reportedly, GAPDH expression can be modulated by the cell proliferative state, and GAPDH can be overexpressed in some cancers (Gong et al., 1996; Cuezva et al., 2004) and in neurodegenerative disorders such as Hungtington, Parkinson and Alzheimer diseases (Mazzola and Sirover, 2002). Moreover, GAPDH antisense rescues cerebellar and mesencephalic dopaminergic neurons from cell death induction in vitro (Ishitani et al., 1996; Fukuhara et al., 2001). Recently, GAPDH has been shown to undergo S-nitrosylation in the nucleus, an event that may be critical for NO-induced cell death (Hara et al., 2005).

However, the potential lethal role of GAPDH located in mitochondria is still elusive. The mitochondrion is the organelle that controls the intrinsic pathway of cell death in thus far that it can release a whole panel of apoptogenic effectors into the cytosol such as cytochrome c (Cyt c), apoptosis-inducing factor (AIF), EndoG, caspases and Smac/Diablo (Brenner and Kroemer, 2000; Green and Kroemer, 2004). In many models of cell death, the outer mitochondrial membrane (OM) becomes permeabilized, through a variety of distinct mechanisms (for a review, see Green and Kroemer, 2004). One particular modality of cell death-associated mitochondrial membrane permeabilization (MMP) is the permeability transition, which manifests by a loss of the inner transmembrane potential (ΔΨm) (Zamzami et al., 1995a, 1995b), matrix swelling and the long-lasting opening of the permeability transition pore complex (PTPC) at the contact sites. The PTPC is composed by a minimum unit of three proteins, namely the voltage-dependent anion channel (VDAC) in the outer membrane (OM), the adenine nucleotide translocase (ANT) in the inner membrane (IM) and cyclophilin D in the matrix (CypD). These three proteins can interact within the contact site between the inner and the outer mitochondrial membranes (IM and OM, respectively). The minimum unit is regulated in its opening probability by the metabolic state of the cells, at the cytosol/intermembrane/matrix interphases, as well as by a panoply of additional proteins whose expression can be cell type-specific (Beutner et al., 1996; Crompton et al., 1998; Halestrap and Brenner, 2003; Basso et al., 2005). At the level of the OM, homo or hetero-oligomers composed of cytosolic proteins (e.g. Bax, Bid) (Antonsson et al., 2001) and/or constitutive mitochondrial proteins (e.g. Bak, VDAC) (Cheng et al., 2003) or local ruptures (Vander Heiden et al., 2000) have been proposed to mediate the release of intermembrane space proteins. Depending on the model, caspases and Bax/Bcl-2 family members can initiate, participate and/or regulate the mitochondrial phase of apoptosis (Brenner and Kroemer, 2000; Green and Kroemer, 2004). In addition, it appears that a large number of pro-apoptotic effectors, proteins or non-proteaceous in nature, can translocate to mitochondria and exert MMP-modulatory functions, thus influencing cell-fate at the level of this organelle. However, the fine molecular mechanisms of apoptotic MMP regulation remain to be solved.

To investigate the putative pro-apoptotic role of GAPDH at the mitochondrial level, we hypothesized that the protein might contribute to the mitochondrial phase of apoptosis and influence pro-apoptotic mitochondrial membrane permeabilization. Thus, using a combined biochemical and pharmacological approach, we showed that a significant fraction of GAPDH localizes to mitochondria, that GAPDH can induce MMP and the release of pro-apoptotic proteins such as Cyt c and AIF from isolated mitochondria, and that GAPDH can interact with one of the components of the PTPC, VDAC1.


Analysis of subcellular distribution of GAPDH and mass fingerprint analysis

Previous studies suggested that GAPDH might locate to multiple subcellular compartments (Ryzlak and Pietruszko, 1988; Mazzola and Sirover, 2003). Thus, to define the intracellular distribution of GAPDH in normal tissue, subcellular localization of GAPDH was analyzed by immunogold labeling and transmission electron microscopy of normal rat liver tissues. Unambiguously, GAPDH was detected in mitochondria (Figure 1Aa and b), the cytosol (Figure 1Aa) and the nucleus (Figure 1Aa and c), associated with the plasma membrane (Figure 1Aa), as well as in the endoplasmic reticulum (Figure 1Aa and d). More precisely, GAPDH localized at the proximity of the mitochondrial membrane, as the GAPDH label resembles to that obtained for ANT (IM) (Figure 1Ae) and for the VDAC (OM) (Figure 1Af). The mitochondrial immunogold staining of GAPDH was conserved after organelle isolation (Figure 1Ag). We then subjected human tumoral HeLa cells (cervical carcinoma) to subcellular fractionation, followed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot detection of the 38 kDa GAPDH. Using this technique, we found GAPDH in the nucleus, as well as in mitochondria, the endoplasmic reticulum and the cytosol (Figure 1B). The relative proportions were analysed by densitometry of immunoblot and estimated to be 27.9, 24.3, 20.2 and 27.5%, respectively. Mitochondrial proteins from HeLa cells were then separated by two-dimensional electrophoresis as described in the Materials and methods section, and spots were excised from the gel and digested with trypsin (Figure 1C). The resulting peptides were subjected to matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) analysis and confirmed by nano high-performance liquid chromatography and mass spectrometry (nanoLC MS/MS). The Mascot peptide mass-fingerprinting program was used to match peptide masses with proteins in the NCBI database. Two independent analyses revealed the presence of the isoform of GAPDH within five spots corresponding to the expected molecular mass (MM) of 38 kDa and isolectric point (8.6<Pi<9.8) of GAPDH (Figures 1C and 2). Importantly, using immunoblot (Figure 1D), MALDI-TOF (Figure 2c) and spectrophotometric measure of enzymatic activity (0.5 nmol/min/mg), GAPDH was also identified in a pre-purified fraction of brain PTPC (Brenner et al., 2000), which contains VDAC, ANT and CypD, suggesting the existence of a mitochondrial membrane sub-environment containing these proteins.

Figure 1

Identification of GAPDH in mitochondria. (A) Subcellular localization of GAPDH in rat liver. Electron microscopy showing the ubiquitous distribution of GAPDH within hepatocytes ( × 6300, a, arrows), mitochondria ( × 35 000, b, g), nucleus ( × 35 000, c), endoplasmic reticulum ( × 35 000, d) and the mitochondrial distribution of ANT ( × 35 000, e) and VDAC ( × 35 000, f). (B) Subcellular fractionation of HeLa cells. Cells were broken and submitted to differential centrifugation to fractionate the nucleus (n), the mitochondria (m), the endoplasmic reticulum (er) and the cytosol (cyt). In total, 25 μg of proteins were separated in SDS–PAGE and immunoblotted for GAPDH and specific subcellular markers, pCNA for (n), Cox2 for (m), GRP78/BIP for (er), tubulin for (cyt). (C) Two-dimensional electrophoresis analysis of mitochondrial proteins. Mitochondria from HeLa cells were isolated and the proteins were analyzed by 2D SDS–PAGE and revealed by Coomassie blue staining. Circles represent five spots submitted to MALDI-TOF and nanoLC MS/MS analysis and identified in Figure 2 as GAPDH. (D) Association of GAPDH with PTPC proteins. PTPC was purified from rat brain and proteins were precipitated with acetone, solubilized by Laemmli buffer and 120 μg of proteins were separated by SDS–PAGE and revealed by immunoblotting for GAPDH, VDAC and ANT as described in Material and methods. Note that the weak intensity of the ANT band, as compared to VDAC band, is due to the immunoblot and does not reflect the real quantitative ratio ANT/VDAC within PTPC fractions. Cyt, cytosol; m, mitochondria; pm; plasma membrane; n, nucleus; er, endoplasmic reticulum.

Figure 2

GADPH identification by MALDI-TOF and nanoLC MS/MS analysis. (a) Sequence of the human liver GADPH isoform. Underlined peptides have been identified by nanoLC MS/MS analysis of 2D separated mitochondrial proteins from HeLa cells. (b) Details of the MS/MS analysis Mascot scores above 38 (bold) indicate identity or extensive homology with a P<0.05. The percent coverage is 48%. (c) Sequence of the mouse liver GADPH isoform. Underlined peptides have been identified by MALDI-TOF analysis from PTPC proteins purified from mouse brain.

GAPDH expression level and mitochondrial enrichment during apoptosis

Using immortalized human embryonic kidney cell line (HEK293) and five cancer cell lines, HeLa, HeLa-Bcl-2, HepG2; HT29 (colon carcinoma) and MCF-7 (breast carcinoma), we observed that GAPDH expression is modulated within cancer cell lines compared to normal cells by reverse transcription–polymerase chain reaction (RT–PCR) analysis (Figure 3A). Intrigued by this finding, we measured the expression level of GAPDH in whole-cell lysates and in the mitochondria of cells treated to undergo apoptosis. In the tumoral HeLa cells, the total GAPDH intracellular level remained unchanged after several treatments such as serum deprivation, protein kinase inhibitor staurosporine, DNA-damaging agent etoposide or mitochondriotoxic drug, lonidamine (Figure 3Ba). However, the treatments induced a significant GAPDH increase in the mitochondrial fraction, which was not observed for the constitutive mitochondrial protein, VDAC (Figure 3Ba). Similar results were found with the immortalized human embryonic kidney cell line (HEK293), for which a nuclear translocation of GAPDH induced by serum deprivation has been previously shown (Sawa et al., 1997) (Figure 3Bb). This indicated that the mitochondrial GAPDH corresponds to an imported protein and not a newly synthesized fraction. Moreover, in a similar experiment in HeLa cells stably overexpressing the oncoprotein Bcl-2, the GAPDH mitochondrial import was totally inhibited, independently of the stimulus used (Figure 3Bc), thus preventing any effect of the protein at the mitochondrial level and suggesting a Bcl-2-dependent mechanism of GAPDH translocation.

Figure 3

GAPDH is significatively induced in cancer cell lines and enriched in mitochondria fraction during apoptosis induction. (A) RT–PCR GAPDH quantification. Endogenous GAPDH messengers were detected in the indicated human cell lines. (B) Mitochondrial GAPDH level during apoptosis induction. After treatments for 6 h by various apoptotic inducers (STS: Staurosporine, 250 nM; ETO: Etoposide, 250 μ M; LND: Lonidamine, 500 μ M) and after 20 h of serum deprivation (SD) in HeLa (a), HEK293 (b), HeLa-Bcl-2 (c) cell lines, endogenous GAPDH protein appears to be accumulated in mitochondrial fraction in a Bcl-2-dependent mechanism, whereas its total cellular level is not modulated. A total of 50 μg of proteins have been loaded, separated in SDS–PAGE and analysed by Western blotting as described in the Materials and methods section. The experiment has been reproduced three times.

Permeabilizing effects of GAPDH on isolated liver mitochondria

As an initial step of the characterization of mitochondrion-associated GAPDH function, we determined whether the protein was enzymatically active. The canonical enzymatic function of GAPDH is to catalyse the oxidation and phosphorylation of glyceraldehyde-3-phosphate (G3P) to the 1,3-bisphosphoglycerate (1,3 BPG), using NAD+ as a co-substrate. The addition of GAPDH substrates to isolated mitochondria allows for the detection of a mitochondrial GAPDH activity of 21 nmol NADH/min/mg of mitochondrial proteins, in the presence of Triton X-100. Next, we added exogenous purified GAPDH to mouse mitochondria to determine its putative association with the organelle. After incubation of mitochondria together with GAPDH in a hypo-osmotic buffer and extensive washing in 0.1 M Na2CO3, pH 12, we pelleted mitochondria by centrifugation and performed immunoblot detection of mitochondrion-associated GAPDH using an antibody that recognizes exogenous (human) but not the endogenous mouse GAPDH (Figure 4a). This approach revealed that the association of GAPDH and mitochondria was tight and proportional to the incubation time (Figure 4a). Moreover, carbonylcyanide m-chlorophenylhydrazone (mClCCP), which depolarizes mitochondria, prevents the detection of GAPDH, and proteinase K did not digest mitochondria-associated GAPDH (Figure 4a), indicating that the association of GAPDH with mitochondria requires an inner transmembrane potential and leads to the import of GAPDH. When mitochondria were treated for a period of 90 min with purified exogenous GAPDH, followed by washes of the organelles, it was found that the mitochondrial supernatant contained two apoptogenic factors, AIF and Cyt c, both of which are normally contained in the intermembrane space (Figure 4b). This GAPDH-induced release of AIF and Cyt c was similar to that observed after the treatment of mitochondria with calcium (Ca2+), the prototypic inducer of mitochondrial permeability transition (Figure 4b). Then, we evaluated the hypothesis that GAPDH could mediate this permeabilization of mitochondrial OM by an action on the PTPC (Figure 5). GAPDH induced swelling of the matrix of isolated rat liver mitochondria, as could be assessed by a decrease in the absorbance at 540 nm (Bernardi et al., 1992). The kinetics of the mitochondrial swelling was dose dependent, as shown at 15 and 45 min (Figure 5A), proportional to the amount of imported GAPDH (see also Figure 4A), maximal and of similar amplitude at 90 min for 2–15 μg of GAPDH (Figure 5A).

Figure 4

GAPDH induces the mitochondrial membrane permeabilization. (a) Association of purified GAPDH. In all, 200 μl of isolated mitochondria (0.2 mg/ml) have been incubated with 5 μg of GAPDH in a hypo-osmotic buffer or the same volume of buffer at 25°C for various periods of time. Mitochondria were then washed in 0.1 M Na2CO3 pH 12, pelletted and mitochondria-associated proteins were analysed by SDS–PAGE and immunoblot for GAPDH. m-ClCCP (1 μ M) was added as a pretreatment to mitochondria to abolish the transmembrane inner potential, and 25 μg/ml proteinase K was added for 30 min to digest GAPDH and to check its accessibility. (b) GAPDH mediates the release of AIF and Cyt c. Following the addition of 5 μg GAPDH, 100 μ M Ca2+ or the same volume of buffer for 90 min at 25°C, 200 μl of isolated mitochondria (0.2 mg/ml) were spun down and the proteins in the supernatant were precipitated with acetone and analysed by SDS–PAGE and immunoblot for Cyt c and AIF. Experiments have been repeated three times.

Figure 5

Mechanisms of GAPDH mitochondrial membrane permeabilization. (A) GAPDH activates dose-dependent mitochondrial swelling. In total, 200 μl of rat liver mitochondria (0.2 mg/ml) were treated with 100 μ M calcium (Ca2+) and the indicated doses of GAPDH, followed by recording of the changes in absorbance at 540 nm. Ca2+ effect was normalized at 100%. (B) GAPDH induces a loss of inner transmembrane potential (ΔΨm). The ΔΨm loss was measured as an increase in the fluorescence of Rhodamine 123 and Ca2+ effect was normalized to 100%. (C) Transmission electronic microscopy analysis of the morphology of intact rat liver mitochondria (a) of 200 μl of treated mitochondria (0.2 mg/ml) by 100 μ M calcium (b), and 5 μg of GAPDH (c) for 90 min at 25°C. × 35 000. (D) GAPDH induces inner mitochondrial membrane permeabilization. In total, 200 μl of mitochondria (0.2 mg/ml) were treated with 100 μ M Ca2+ or 5 μg GAPDH for 60 min, followed by assessment of matri × citrate synthase activity after sequential addition of 0.1 mM DTNB; 0.3 mM acetyl CoA, 0.5 mM oxaloacetate. (E) GAPDH permeabilizes PTPC-containing proteoliposomes. The indicated doses of GAPDH or 600 μ M Ca2+were incubated with 50 μl 4-MUP pre-loaded plain liposomes and PTPC-proteoliposomes for 1 h at 20°C. Phosphatase alkaline was then added for 15 min at 37°C to convert released 4-MUP into MU. The 4-MUP release induced by GAPDH was evaluated as the percentage of 4-MUP release induced by 600 μ M calcium. Co., control. Experiments have been repeated three times and a representative experiment is shown in (A, B, C). Bars represent standard deviation.

The swelling was preceded by a loss of inner transmembrane potential (ΔΨm) (Figure 5B), as quantified using the rhodamine dequenching method (Narita et al., 1998). Intrigued by the fact that GAPDH did not induce a complete swelling but a complete depolarization at 90 min (Figure 5A and B), we performed a transmission electronic microscopy analysis of mitochondria treated by GAPDH or Ca2+ (Figure 5Ca, b and c). Interestingly, the morphology of GAPDH-treated mitochondria differ drastically from Ca2+-treated organelles. GAPDH-treated mitochondria swelled and exhibited a partial loss of internal material, whereas cristae organization was profoundly remodelled. The main difference was in the limited loss of the round shape of GAPDH-treated mitochondria compared to the calcium-treated mitochondria. This constitutes a plausible explanation of the difference in the turbidity observed at the end of the treatment.

In addition, GAPDH induced the parallel permeabilization of the IM, as shown by an enzymatic assay (Korge and Weiss, 1999) that measures the accessibility of acetyl CoA (MM of 809 Da) to a matricial enzyme, citrate synthase (Figure 5C). When the IM is impermeable, acetyl CoA cannot be metabolized by citrate synthase. However, upon addition of GAPDH, the IM became substrate-permeable. Next, we purified the PTPC from rat brain and incorporated it into proteoliposomes that were loaded with 4-methyl-umbelliferylphosphate (4-MUP) (Belzacq et al., 2003). Such PTPC-containing proteoliposomes were treated with Ca2+ or increasing doses of GAPDH. Upon treatment with purified GAPDH, PTPC proteoliposomes released encapsulated 4-MUP in a dose-dependent manner to a level of about 20% of that considered to be the maximum level induced by Ca2+ (Figure 5D). This corresponds to the extent of permeabilization induced by atractyloside, an ANT ligand. In contrast, GAPDH was unable to permeabilize plain liposomes (Figure 5D), supporting the idea that the effect of GAPDH on proteoliposomes must involve the direct cooperation between GAPDH and one or several PTPC protein subunits.

Biochemical and pharmacological characterization of GAPDH-induced mitochondrial swelling

To investigate the mechanisms of GAPDH-induced mitochondrial permeabilization, we used two pharmacological inhibitors targeting PTPC proteins, cyclosporin A (CsA) and 4,4′-diisothiocyanatostilbene-2,2′-disulfonate (DIDS). CsA is known to inhibit the binding of cyclophilin D (CypD) to ANT and, in turn, to inhibit the mitochondrial permeability transition induced by diverse agents such as Ca2+, oxidative stress as well as glutathione depletion (reviewed in Halestrap and Brenner, 2003). DIDS inhibits the VDAC protein by a direct interaction (Thinnes et al., 1994) and by channel activity interference (Han et al., 2003). Thus, Ca2+-induced swelling and ΔΨm loss were prevented by CsA and delayed by DIDS, as shown in Figure 6a and b, respectively. In contrast, the swelling and the ΔΨm loss induced by GAPDH were inhibited either by CsA or by DIDS. This result suggested that, like Ca2+, GAPDH did require an ANT-CypD interaction (which is blocked by CsA) (Halestrap and Brenner, 2003) and would also cooperate with VDAC (which is inhibited by DIDS) (Figure 6b).

Figure 6

Characterization of GAPDH-induced swelling. (a) Regulation of GAPDH-induced swelling. Rat liver isolated mitochondria were suspended in a hypo-osmotic buffer and their absorbance at 540 nm was recorded for 90 min at 20°C. Various inhibitors (1 μ M CsA, 5 μ M DIDS) were added 2 min before the MMP inducers (100 μ M Ca2+ and 5 μg GAPDH). (b) Regulation of GAPDH-induced an ΔΨm loss. The ΔΨm loss was measured by Rhodamine 123 dequenching. (c) Regulation of GAPDH-induced permeabilization of the inner mitochondrial membrane. Mitochondria pre-treated (or not) by 5 μ M DIDS were analysed for IM permeabilization by measuring the citrate synthase activity. Ca2+ effect was normalized to 100%. Experiments have been repeated three times and one representative experiment is shown in (a, b). In (c), bars represent standard deviation.

Although the swelling and the depolarization induced by Ca2+ were delayed by DIDS, the agent inhibited totally the Ca2+-induced inner mitochondrial membrane permeabilization, suggesting several effector mechanisms for this ion (Figure 6c). In addition, DIDS prevents fully the permeabilizing effect of GAPDH (Figure 6c). It is noteworthy that DIDS had no inhibitory effect on citrate synthase (not shown). In synthesis, these results suggest that the induction of mitochondrial permeability transition by GAPDH would involve a functional cooperation with the PTPC.

To gain further insights into the mechanism of the GAPDH-induced MMP, we selected a set of mitochondrial substrates and defined their inhibitory spectrum on GAPDH-induced matrix swelling. Thus, ATP (as an ANT ligand (Pfaff and Klingenberg, 1968)) and NADH2 (as an inhibitor of VDAC (Lee et al., 1994)), heat inactivation (as a denaturation control) and iodo-acetic acid (IOA, an inhibitor of the enzymatic function of GAPDH) inhibited GAPDH-induced swelling (Figure 7A and B), whereas ADP and nelfinavir (NFV), two ANT pore function inhibitors (Belzacq et al., 2003; Weaver et al., 2005) and the GAPDH glycolytic substrates G3P plus NAD+ had no inhibitory effect (Figure 7A and B). In contrast to heat and IOA, CsA, DIDS, ATP did not affect the enzymatic activity of purified GAPDH in solution (Figure 7Ca). These inhibitors or treatments did not affect the import of the protein in mitochondria (Figure 7Cb).

Figure 7

Regulation of GAPDH-induced swelling and GAPDH glycolytic activity. (A) Inhibition of GAPDH effects by a panel of compounds. 10 μ M ATP, 10 μ M ADP, 10 μ M Nelfinavir (NFV) and 100 μ M NADH2 were added 2 min before 5 μg GAPDH addition on 200 μl of isolated mitochondria (0.2 mg/ml) and swelling measurement. (B) Structural regulation. GAPDH was heated to 100°C for 10 min for complete denaturation or treated with 1 mM IOA inhibitor or its substrates 500 μ M G3P and 500 μ M NAD+. (C) Regulation of GAPDH glycolytic activity. (a) The purified enzyme was treated by the same agents or procedures as in (A and B), and the enzymatic activity of the protein was evaluated. (b) The import of GAPDH was measured by Western blot on mitochondria as described in Materials and methods section. Co., control. Experiments have been repeated three times and one representative experiment is shown in (A and B). In (C), bars represent standard deviation.

GAPDH interacts physically with VDAC1, but not ANT

As GAPDH possesses no permeabilizing properties on plain liposomes and appears to require PTPC proteins to exert its effects (Figure 5D), we decided to identify the putative mitochondrial interactor of GAPDH. Hence, we attempted co-immunoprecipitation of VDAC, ANT and GAPDH using isolated mitochondria as the starting material. However, the immunoprecipitation results were not interpretable as GAPDH and VDAC bound non-specifically to many types of beads, despite the extensive pre-saturation of these beads, suggesting that the proteins are ‘sticky’. We therefore changed the experimental strategy and performed blue native electrophoresis (BN-PAGE) and dot-blot experiments. Thus, using BN-PAGE for the separation of mouse isolated mitochondria proteins, we found a molecular complex containing (at least) VDAC and GAPDH at 266 kDa (Figure 8A). Using pure proteins, GAPDH was detected as dimers (72 kDa) and tetramers (160 kDa; Figure 8B), and in high MM oligomers in the presence of rhVDAC1 (Figure 8B) by the same methodology. These oligomers migrated in gels as an estimated mass of 720, 1135 and >2000 kDa and were disrupted by dithiotheritol (DTT) (Figure 8B). Purified human GAPDH was then dotted onto nitrocellulose membranes and incubated with purified rhVDAC1 or ANT. After extensive washing, the membranes were subjected to the immunochemical detection of VDAC and ANT (Figure 8Ca and b). On the basis of this in vitro assay, GAPDH was found to interact directly with rhVDAC1, but not with ANT (Figure 8C). The interaction between GAPDH and rhVDAC1 resisted treatment up to 0.250 M NaCl (with a higher signal at 0.15 M NaCl, physiological ionic strength), and to SDS, heat denaturation, IOA, DIDS, ATP, NADH2 and CsA. However, 0.1 and 1 M NaCl decreased the interaction, and DTT, a thiol reagent, abolished the GAPDH–VDAC1 interaction, suggesting the involvement of ionic bonds and thiols within VDAC and/or of GAPDH (Figure 8Ca). Next, we assessed the ability of membrane-imbedded VDAC to act as an anchor/receptor of GAPDH. Clearly, the presence of VDAC1 increased GAPDH binding to the proteoliposomal membranes, suggesting that VDAC alone, when inserted in a membrane environment, might be sufficient to serve as an anchor for GAPDH (Figure 8D).

Figure 8

Physical interaction between VDAC and GAPDH. (A) BN-PAGE-detectable heteroligomerization of GAPDH with VDAC. Mitochondrial proteins were first separated by a 6–20% non-denaturing gel and in the second dimension by a 10% denaturing gel. The gel was stained by Coomassie blue and transferred to a nitrocellulose membrane for immunoblot detection of GAPDH and VDAC. (B) BN-PAGE-detectable heteroligomerization of purified GAPDH with VDAC. GAPDH or GAPDH–VDAC complexes were first separated by a 6–20% non-denaturing gel and in the second dimension by a 10% denaturing gel. When indicated, GAPDH–VDAC complexes were treated with 10 mM DTT before BN-PAGE and Western-blot anti-GAPDH. (C) Dot-blot identification of a direct interactor of GAPDH. GAPDH was dotted onto a nitrocellulose membrane treated with various concentrations of NaCl, 1% SDS, heat, 5 μ M DIDS, 1 mM IOA, 5 μ M ATP, 100 μ M NADH2, 1 μ M CsA or 10 mM DTT, before incubation with purified rhVDAC1 (0.1 mg/ml) or purified ANT (0.1 mg/ml), followed by washing and immunodetection of retained VDAC and ANT. (D) Anchorage of GAPDH on VDAC. Purified GAPDH was added to plain liposomes or VDAC liposomes. After incubation, liposomes were washed and analysed for GAPDH retention by measuring GAPDH activity. (E) Implication of thiols in the mitochondrial swelling activity of GAPDH. Purified GAPDH (5 μg) was either pretreated or not pretreated with 10 mM DTT and added to 200 μl of isolated rat liver mitochondria (0.2 mg/ml). Mitochondrial swelling was evaluated by measuring the absorption at 540 nm, while considering the effect of 100 μ M Ca2+ as the maximal effect (100%). These experiments have been repeated three times and a representative experiment is shown in (A, B and C). In (D and E), bars represent standard deviation.

Finally, if the DTT-inhibitable interaction between VDAC1 and GAPDH is important for GAPDH-induced MMP, then DTT should suppress GAPDH-induced matrix swelling. Although DTT had no inhibitory effect on the enzymatic activity of GAPDH (not shown), it did inhibit the mitochondrial swelling induced by GAPDH but not that induced by Ca2+ (Figure 8E). Although we cannot exclude the fact that DTT acts on other mitochondrial targets or impairs the mitochondrial import of GAPDH, these data are compatible with the hypothesis that the VDAC1–GAPDH interaction mediates the capacity of GAPDH to induce the mitochondrial permeability transition.


In this report, we describe a novel pro-apoptotic function of GAPDH in the mitochondrion (Figures 1 and 2). During apoptosis induction by various stimuli, a fraction of the total protein accumulates in a Bcl-2-dependent mechanism to the mitochondria of several tumoral cell lines (Figure 3). Moreover, we show for the first time that exogenous GAPDH associates with mitochondria and induces PTPC-dependent pro-apoptotic MMP via an association with VDAC1 (Figure 4). The protein triggers the loss of ΔΨm, matrix swelling, inner mitochondrial membrane permeabilization and the release of Cyt c and AIF through the OM (Figures 4, 5, 6, 7 and 8). Once into the cytosol and the nucleus, respectively, these two factors can trigger the caspase-dependent and caspase-independent degradation that characterize the late phase of cell death (Green and Kroemer, 2004). It is accepted that GAPDH is found in multiple subcellular sites. A possible association with mitochondria from the human brain has been previously proposed on the basis of glycolytic activity (Ryzlak and Pietruszko, 1988). Here, we identified GAPDH in mitochondria of human epithelial cells, rat liver and mouse/rat brain by MALDI-TOF and nanoLC MS/MS (Figures 1 and 2). Although GAPDH possesses a putative mitochondrial import presequence (MGKVKVGVNGFRIGRL, 0.9334 of probability of export, MitoProtII1.0a4,, it is not clear whether the protein is actively imported into mitochondria or whether it rather associates with mitochondrial membrane proteins (such as VDAC). Indeed, the experiments performed on isolated mitochondria were carried out in the absence of exogenous ATP and cytosolic extracts that would be required for canonical, active import of GAPDH into mitochondria using the TOM/TIM machinery. In all our experiments, mitochondria were suspended in a hypo-osmotic buffer (Susin et al., 1999), allowing the generation of an ΔΨm, matrix volume changes and the opening of PTPC (Figures 4, 5, 6, 7 and 8). Thus, when exogenous GAPDH was added to liver mitochondria in this buffer, the protein associated tightly with mitochondria in a ΔΨm-dependent manner (Figure 4). However, the immunogold labeling of GAPDH (Figure 1) and its inaccessibility to proteinase K digestion (Figure 4) support the association of GAPDH with mitochondrial membranes. GAPDH has been involved in apoptosis in several physiopathological models, such as neurodegenerative diseases (Chuang and Ishitani, 1996; Chuang et al., 2005). For instance, during apoptosis of a neuroblastoma cell line and fibroblasts, its expression increased correlatively in the nucleus (AbdelRahman et al., 1999), suggesting that GAPDH initiates apoptosis from this organelle. However, intrigued by the presence of GAPDH within the mitochondrion during apoptosis induction (Figure 3B), and more precisely within the PTPC (Figure 1), we explored the capability of GAPDH to modulate MMP and found that the protein can bind to VDAC1 in vitro, as shown by three independent techniques: (i) BN-PAGE (Figure 8A and B), (ii) dot blot (Figure 8C) and (iii) proteoliposomes (Figure 8D). Although we do not exclude the possibility that the protein might have alternative intra-mitochondrial targets, the effects of GAPDH could be mediated by a direct interaction with VDAC1. In accord with this speculation, we found that pharmacological inhibitors specific of VDAC, DIDS and NADH2, and of ANT, ATP and one inhibitor of the ANT-CypD interaction, CsA, inhibited the permeabilizing effects of GAPDH (Figure 7). As a caveat, it should be noted that most of our observations are based on the use of exogenous GAPDH added to mitochondria and therefore, the role of endogenous GAPDH remains elusive. However, we speculate that endogenous GAPDH associated with mitochondria could have a pro-apoptotic function, at least in some circumstances. First, a signal could convert it into a pro-apoptotic protein by modification of its conformation/function. Second, during apoptosis, its association with mitochondria could increase to a threshold level to facilitate MMP (see Figure 3). The second hypothesis is supported by the fact that the amount of endogenous protein present in the mitochondrion and in PTPC proteoliposomes is not sufficient per se to promote mitochondrial alterations (Figure 1) and that the addition of exogenous GAPDH triggers the permeabilization without the need for an additional pro-apoptotic signal. This hypothesis is also supported by the fact that the overexpression of GAPDH triggers cell death (Sunaga et al., 1995; Ishitani et al., 1998; Tajima et al., 1999).

What are the underlying mechanisms of GAPDH-induced MMP? On the basis of pharmacological and binding experiments (Figures 5, 6, 7 and 8), GAPDH might associate with VDAC1 in an environment close to the PTPC members, presumably the contact sites (Beutner et al., 1998). Some PTPC inhibitors, which limit the diffusion through the pore (e.g. ATP, NADH2, CsA), but not ANT pore inhibitors (e.g. ADP, NFV), directly affect the capacity of GAPDH to induce matrix swelling, perhaps by blocking the PTPC in its closed conformation. In addition, the inhibition of swelling, depolarization and the IM induced by GAPDH was more sensitive to DIDS inhibition than calcium, suggesting a different effector mechanism based on a primary interaction with an OM protein, VDAC. The interaction of GAPDH with VDAC1 may be based on the formation of inter- or intra-molecular disulfide bridges (as DTT abolished the binding of the two proteins), yet is unlikely to involve secondary and tertiary structures (because denaturation by heat and SDS do not affect binding) (Figure 7). Moreover, as the reducing agent DTT can inhibit the interaction of GAPDH and VDAC (as well as GAPDH-induced swelling of isolated mitochondria), it is tempting to speculate that a pro-oxidant intracellular milieu, which is usually observed in apoptosis, would favor the binding of the two proteins and their functional cooperation (Le Bras et al., 2005). An association with ANT/CypD was required for the GAPDH-induced loss of the ΔΨm and matrix swelling. This suggests a process in which GAPDH affects PTPC, which causes OM and IM permeabilization, dissipation of the ΔΨm and matrix swelling. A similar process has been described for the pro-apoptotic action of three pathogen-encoded proteins, Vpr from HIV-1 (Jacotot et al., 2000), HBx from Hepatitis B virus (Shirakata and Koike, 2003) and porB from Neisseria (Muller et al., 2000). In binding experiments, GAPDH interacts with the VDAC1 isoform, although binding to other VDAC isoforms has not been measured. The three isoforms of human VDAC are very similar in their primary sequence and each of them is dispensable for cellular respiration (Wu et al., 1999), suggesting that they can assume redundant vital functions. However, the VDAC isoforms may be distinct in their pro-apoptotic activity. Thus, VDAC1, the most abundant isoform, is likely to be the one that engages in pro-apoptotic interaction with Bax (Zaid et al., 2005), while VDAC2 can bind and block Bak in an inactive conformation (Cheng et al., 2003). Moreover, as shown here, VDAC1 may be the partner of GAPDH in its pro-apoptotic facet. In the future, it will be interesting to determine the consequences of the interaction between GAPDH and VDAC on the various functions of VDAC. Indeed, VDAC is mainly known as a voltage-gated OM pore, allowing the diffusion of solutes <5 kDa (Colombini, 1983; Zizi et al., 1998). However, other functions such as apoptosis regulation via an interaction with Bcl-2 family members (Shimizu et al., 1999; Vander Heiden et al., 2000; Cheng et al., 2003; Rostovtseva et al., 2004; Zaid et al., 2005), kinase binding (Azoulay-Zohar et al., 2004), NADH ferricyanide reductase activity (Baker et al., 2004) as well as calcium transport (Gincel et al., 2001) have been reported. Thus, preliminarily, we observed that GAPDH increases the Ca2+ uptake through VDAC into proteoliposomes and not into plain liposomes (not shown). The modulation of these numerous VDAC functions by GAPDH might have a major impact on mitochondrial biology, oncogenesis and apoptosis.

Materials and methods

Nelfinavir (NFV) was a generous gift from A Badley (Mayo Clinic, Rochester, MN, USA). When not indicated, chemicals are purchased from Sigma (St Louis, MO, USA).

Cell culture, treatments and subcellular fractionation

HeLa Neo cells (generous gift from Victor Goldmacher), HT29 (generous gift from A Zweibaum), HEK293 (generous gift from JM Rossignol) and MCF7 cells (generous gift from Theraptosis) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum and antibiotics at 37°C under 5% CO2. To induce apoptosis, cells were treated with Lonidamine (LND), Etoposide (ETO) and Staurosporine (STS) for 6 h at 37°C or without serum for 24 h at 37°C. Total cell lysates prepared from freshly collected cells were washed three times in ice-cold PBS and incubated in buffer (10 mM Tris-HCl, pH 7.6, 1.5 mM MgCl2, 10 mM KCl, 250 mM saccharose, 10 μ M cytochalasin B, 0.4 mM PMSF) for 30 min on ice. Cells were lysed in a Dounce homogenizer (200 strokes, 300 r.p.m.). The nuclei were pelleted by centrifugation at 600 g for 10 min and washed three times in buffer. The nuclei were purificated by centrifugation on a 30% saccharose solution at 800 g for 10 min. Mitochondria were isolated from the supernatant by centrifugation at 6800 g for 10 min and the pellet was washed three times in buffer. The supernatant of mitochondria fraction was centrifuged at 100 000 g for 60 min. The pellet contained the microsomal fraction and the supernatant corresponded to the cytosolic fraction. Protein concentration was determined by the micro bicinchoninic acid assay. The relative proportion of GAPDH within each compartment was determined by the densitometry of immunoblot by ImageJ software.

Protein one-dimensional electrophoresis and Western blot

Proteins were analysed by SDS-PAGE (10%, 25 μg protein/lane) according to Laemmli (1970) and revealed by immunoblotting with anti-GAPDH serum (ab9485, Abcam, Cambridge, UK), anti-AIF (Chemicon, Hampshire, UK) and anti-cytochrome c (BD Biosciences, Le Pont de Clay, France), a polyclonal serum anti-ANT (Eurogentec, Seraing, Belgium), a polyclonal serum anti-VDAC (Eurogentec, Seraing, Belgium), anti-PDI (Stressgen, Victoria, Canada), anti-tubulin (Serotec, Cergy Saint Christophe, France), anti-pCNA (Santa-Cruz, CA, USA) and anti-Cox 2 (Santa-Cruz, Heidelberg, Germany). Proteins were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech, Rockford, IL, USA).

Protein two-dimensional electrophoresis and protein identification

First dimension: BN-PAGE and IEF

Blue native electrophoresis (BN-PAGE) was used as described (Schagger et al., 1994). Linear 6–20% polyacrylamide gradient gels were used for BN-PAGE. Native electrophoresis was performed at 4°C using a vertical electrophoresis system (mini Protean 3 cell, Bio-Rad, CA, USA). For isoelectric focusing (IEF), linear gradient pH 3–10 ReadyR Strip IPG Strip (Bio-Rad) were used. Protein pellets were solubilized in 7 M urea, 2 M thiourea, 20 mM DTT, 4% Nonidet P-40 and 0.4% Bio-Lyte 3-10 (Bio-Rad). IEF was carried out using immobilized pH gradient on a PROTEAN IEF cell system (Bio-Rad).

Second dimension

The strips were equilibrated in the presence of DTT followed by iodoacetamide. The strips were then placed at the top of the second dimension gel and submitted to SDS–PAGE, as described in the instruction Manual of ReadyStrip IPG strips (Bio-Rad).

Peptide mass fingerprinting using MALDI-TOF mass spectrometry

In-gel tryptic digestion of 2D protein spots was performed and resulting peptide were submitted to MALDI as described (Verrier et al., 2004). 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). Peptide masses were queried against entries for mammals in the NCBI database using the Mascot peptide mass finger-printing program from Matrix Science ( For a match to be considered, a minimum of three matching peptides was required.

NanoLC and MS/MS

Spots of protein were excised from a 2-D gel and subjected to in-gel tryptic digestion as above. The resulting peptides were extracted and subjected to nanoscale reverse phase liquid chromatography on a modular LC Packings Ultimate HPLC system equipped with a Famos autosampler and a Switchos micro column switching device (LC Packings – a Dionex company, Amsterdam, The Netherlands). The tryptic-digested samples were diluted in an aqueous solution containing 0.1% trifluoroacetic acid and preconcentrated and de-salted at a flow rate of 20 μl/min on a 5 mm × 300 μm PepMap C18 precolumn (100 Å, 5 μm, LC Packings). The mobile phase flow from pump C was used to load and wash the sample for 5 min with an aqueous solution containing 0.1% trifluoroacetic acid and 2% acetonitrile. The peptides were then eluted onto a 150 mm × 75 μm analytical PepMap C18 column (100 Å, 3 μm, LC Packings). Chromatographic separation used a gradient elution of 95% solution A (acetonitrile/water 2:98, v/v) to 50% solution B (acetonitrile/water 95:5, v/v), both containing 0.08% formic acid and 0.01% trifluoroacetic acid, over 40 min at a flow rate of 200 nl/min. The nanoscale LC eluent from the analytical column was directed to the nanoelectrospray ionization source of a QSTAR®XL global hybrid quadrupole/time-of-flight mass spectrometer (Applied Biosystem) operated in positive ion mode. A voltage of approximately 2 kV was applied to the spray needle (Picotip Emitter, 360/10 μm, New Objective, MA, USA). Mass spectra were acquired with the Analyst 1.1 software using MS survey during 1 s followed by MS/MS during 3 s. The instrument was calibrated with a multi-point calibration using selected fragment ions that resulted from the collision-induced decomposition (CID) of Glu-fibrinopeptide B. A data directed analysis was employed to perform MS/MS analysis on doubly and triply charged precursor ions. Product (fragmentation) ion MS/MS spectra were collected from m/z 60 to m/z 2000. Raw data were automatically analysed on a local server harboring Mascot and an updated Swissprot library database.

Subcellular localization by transmission electron microscopy and immunogold labeling

Wistar rat (males, 12–13 weeks) liver fragment and isolated mitochondria were fixed in 4% paraformaldehyde and 0.5% glutaraldehyde in 0.1 M sodium phosphate buffer pH 7.4 during 2 h, followed by 4% paraformaldehyde during 3 h. Samples were gradually dehydrated in ethanol (30–100%) on ice and embedded in LRW resin, polymerized for 24 h at 60°C. Thin sections (80 nm) were collected onto 300 mesh nickel grids, and processed for immunochemistry with anti-GAPDH serum (ab9485, Abcam, Cambridge, UK), anti-VDAC polyclonal serum (Eurogentec), anti-ANT polyclonal serum (Eurogentec) and goat anti-rabbit serum (Tebu, Le Perray, en Yvelines, France). Useful dilution of antibodies has been determined as the highest dilution, giving no background on the section of isolated mitochondria and an intra-mitochondrial labeling. The observation was processed with a Philips CM12 transmission electron microscope at 80KV.

RNA extraction and RT–PCR

Total cellular RNAs were extracted using an Rneasy Mini Kit (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. The first cDNA strand was synthesized from 1 μg of Dnase-treated total RNA, using 200 U of Mu-MLV Reverse Transcriptase (Qbiogene, Evry, France), for 1 h at 37°C, followed by 10 min at 70°C to denature RTase. PCR was then performed with 1 U of Taq DNA polymerase and 50 pmol of oligonucleotides specific for GAPDH (GAPDH forward 5′-IndexTermGGTCGGAGTCAACGGATTTGGTCG-3′; GAPDH reverse 5′-IndexTermCCTCCGACGCCTGCTTCACCAC-3′) or 18S (18S forward 5′-IndexTermGTAACCCGTTGAACCCCATT-3′; 18S reverse 5′-IndexTermCCATCCAATCGGTAGTAGCG-3′). Amplicons were analysed by electrophoresis on a 2% agarose gel.

Association of purified GAPDH and mitochondria

Mitochondria (0.5 mg/ml) and human purified GAPDH (EC., 48 μmol NADH/min/mg, Sigma) were incubated in a hypo-osmotic buffer (10 mM Tris-Mops, pH 7.4, 5 mM succinate, 200 mM sucrose, 1 mM Pi, 10 μ M EGTA, 2 μ M rotenone) for various periods of time at 20°C. Mitochondria were then pelleted by centrifugation at 6.800 g for at least 10 min at 4°C and washed in 0.1 M Na2CO3, pH 12, for 10 min to remove loosely associated proteins and to preserve tightly bound proteins (Poncet et al., 2004). Mitochondrial proteins were dissolved in Laemmli buffer, and analysed by SDS–PAGE and immunoblotting to determine the extent of association of GAPDH within the mitochondria.

Isolation of rat liver mitochondria and measurements

Mitochondria were isolated from rat liver (Wistar males, 12–14 weeks) by differential centrifugations and purified on Percoll gradient according to Belzacq et al. (2003). For swelling, depolarization and citrate synthase activity measurements, the mitochondria were diluted at 0.2 mg/ml in the hypo-osmotic buffer (10 mM Tris-Mops, pH 7.4, 5 mM succinate, 200 mM sucrose, 1 mM Pi, 10 μ M EGTA, 2 μ M rotenone). The mitochondrial swelling was measured by the decrease in absorbance at 540 nm. The depolarization of the mitochondria was measured by the rhodamine 123 (1 μ M) fluorescence dequenching (excitation: 485 nm, emission: 535 nm, Molecular Probes, Cergy Pontoise, France). The citrate synthase activity was measured by following the coupling between coenzyme A (CoASH) and 5,5′-dithiobis-(2-nitrobenzoate) (DTNB), essentially as described by Korge et al. (Korge and Weiss, 1999). In brief, the DTNB was added first to the mitochondrial suspension, just before the inducer (Ca2+ or GAPDH). The suspension was incubated for 1 h 30 min at 20°C to induce a complete swelling. Oxaloacetate and acetylCoA were then added, and the increase in absorbance at 412 nm was measured.

ANT and VDAC purification and reconstitution into liposomes

The rhVDAC1 plasmid that harbors the human gene for producing recombinant human VDAC1 (rhVDAC1) was a gift from Pr H Tang. rhVDAC1 was prepared as previously described by Shi et al. (2003) with slight modifications. During purification, the protein was bound onto Ni-NTA beads and washed with 20 ml of buffer B (4 M guanidine, Tris 20 mM pH 7.9, NaCl 500 mM) with 20 mM imidazole. The protein was then refolded with 3 × 2 ml of buffer B mixed with buffer C (2% octyl-D-glucoside (OG), 20 mM Tris pH 7.9, 200 mM NaCl and 10% glycerol) plus 20 mM imidazole at the following ratios: B:C=1:3, 1:7, 1:15, respectively. The resin was washed with 20 ml of buffer C plus 20 mM imidazole and then the protein was eluted with 6 ml of buffer C supplemented with 500 mM imidazole. To reconstitute VDAC into proteoliposomes, palmitoyloleoylphosphocholine (POPC) was dissolved in chloroform at 10 mg/ml and dried under N2. The lipid film was resuspended in liposome buffer (10 mM sodium phosphate buffer pH 7.5, 100 mM NaCl) and solubilized with Triton-X 100 at a ratio of Triton:lipid; 2w:1w. After 1 h incubation, rhVDAC1 was added at a lipid to protein ratio of 10, and the mixture was incubated for 1 h and dialysed against 1000 vol of liposome buffer at 4°C during 48 h. The liposomes were finally recovered by ultracentrifugation at 100 000 g during 1 h, and resuspended in liposome buffer and stored at 4°C until use. The functionality of the protein into proteoliposomes was tested by the measure of calcium uptake and of polyethylene glycol exclusion. ANT was purified from rat heart as described previously. The permeability transition pore complex was purified from rat brain and reconstituted into cholesterol:phosphatidylcholine; 1w:50w liposomes (Brenner et al., 2000). Mitochondria were briefly isolated from four rat brains by homogenization in 1 mM aminothioglycerol–10 mM glucose (pH 8.0) and centrifugation (15 min, 12 000 g, 4°C). Following solubilization (0.5% (v:v) Triton X-100 (Roche, Meylan, France, membrane research grade), 30 min, RT) and ultracentrifugation (40 min, 50 000 g, 4°C), soluble proteins contained within the supernatant were mixed with 17 g of DE52 resin (Whatman), previously washed and equilibrated in buffer A (1.5 mM Na2HPO4, 1.5 mM K2HPO4, 0.1 mM glucose and 1mM DTT, (pH 8.0)). A flow pressure limited chromatography (FPLC) column (GE-Healthcare, Velizy, France) was then filled with the resin and, after equilibration with 6 ml of buffer A, proteins were eluted with a linear gradient of KCl from 50 to 400 mM KCl at a flow rate of 0.8 ml/min. Hexokinase activity-containing fractions were mixed (v:v) with pre-formed phosphatidylcholine-cholesterol (5:1) lipid vesicles in the presence of octyl-D-glucoside and dialyzed overnight at 4°C to eliminate surfactants traces. The ability of GADH to activate the pore function of PTPC was quantified as the percentage of entrapped 4-methyl umbelliferyl phosphate (4-MUP) release compared to calcium induced 4-MUP release (Belzacq et al., 2003).

Protein–protein interaction analysis by Dot-blot in GAPDH/VDAC liposomes

GAPDH (2 μg) was spotted onto a dry nitrocellulose membrane (Trans-BlotR pure nitrocellulose Membrane (0.2 μm), Bio-Rad) and hybridized with rhVDAC1 or purified ANT (0.1 mg/ml) in a buffer solution (150mM sodium phosphate buffer, pH 7.0, 0.5% Triton X-100). After 1 h, the membrane was washed three times with the hybridization buffer and probed with an antibody anti-VDAC or anti-ANT. Immunodetection was carried out in hybridization buffer. Liposomes with or without rhVDAC1 were mixed with 0.1 or 1 μg of purified GAPDH and incubated during 1 h at 20°C under constant stirring. The mixture was then ultracentrifuged for 1 h at 100 000 g twice, to eliminate non-specific binding of GAPDH to liposomes. The last pellet was resuspended in the GAPDH enzymatic activity buffer, the substrate and co-substrate were added, and the GAPDH activity associated to liposomes was quantified.

GAPDH enzymatic assay

Glycolytic activity of GAPDH was quantified by monitoring the increase in absorbance at 340 nm as described (Brassard et al., 2004).



apoptosis-inducing factor


adenine nucleotide translocator

Cyt c:

cytochrome c


cyclosporin A






mitochondrial transmembrane potential


glyceraldehyde-3-phosphate dehydrogenase


carbonylcyanide m-chlorophenylhydrazone


mitochondrial membrane permeabilization


4-methylumbelliferyl phosphate

nanoLC MS/MS:

nano liquid chromatography mass spectroscopy/mass spectroscopy


inner membrane


outer membrane


propidium iodide


reactive oxygen species


permeability transition pore complex


peripheral benzodiazepine receptor


voltage-dependent anion channel


  1. AbdelRahman WM, Georgiades IB, Curtis LJ, Arends MJ, Wyllie AH . (1999). Role of BAX mutations in mismatch repair-deficient colorectal carcinogenesis. Oncogene 18: 2139–2142.

    CAS  Article  Google Scholar 

  2. Antonsson B, Montessuit S, Sanchez B, Martinou JC . (2001). Bax is present as a high molecular weight oligomer/complex in the mitochondrial membrane of apoptotic cells. J Biol Chem 276: 11615–11623.

    CAS  Article  Google Scholar 

  3. Azoulay-Zohar H, Israelson A, Abu-Hamad S, Shoshan-Barmatz V . (2004). In self-defence: hexokinase promotes voltage-dependent anion channel closure and prevents mitochondria-mediated apoptotic cell death. Biochem J 377: 347–355.

    CAS  Article  Google Scholar 

  4. Baker MA, Lane DJ, Ly JD, De Pinto V, Lawen A . (2004). VDAC1 is a transplasma membrane NADH-ferricyanide reductase. J Biol Chem 279: 4811–4819.

    CAS  Article  Google Scholar 

  5. Basso E, Fante L, Fowlkes J, Petronilli V, Forte M, Bernardi P . (2005). Properties of the permeability transition pore in mitochondria devoid of cyclophilin D. J Biol Chem 280: 18558–18561.

    CAS  Article  Google Scholar 

  6. Belzacq A, Vieira H, Verrier F, Vandecasteele G, Cohen I, Prevost M et al. (2003). Bcl-2 and bax modulate adenine nucleotide translocase activity. Cancer Res 63: 541–546.

    CAS  PubMed  Google Scholar 

  7. Bernardi P, Vassanelli S, Veronese P, Colonna R, Szabo I, Zoratti M . (1992). Modulation of the mitochondrial cyclosporin A-sensitive permeability transition pore by the proton electrochemical gradient. Evidence that the pore can be opened by membrane depolarization. J Biol Chem 267: 2934–2939.

    CAS  PubMed  Google Scholar 

  8. Beutner G, Ruck A, Riede B, Brdiczka D . (1998). Complexes between porin, hexokinase, mitochondrial creatine kinase and adenylate translocator display properties of the permeability transition pore. Implication for regulation of permeability transition by the kinases. Biochim Biophys Acta 1368: 7–18.

    CAS  Article  Google Scholar 

  9. Beutner G, Ruck A, Riede B, Welte W, Brdiczka D . (1996). Complexes between kinases, mitochondrial porin and adenylate translocator in rat brain resemble the permeability transition pore. FEBS Lett 396: 189–195.

    CAS  Article  Google Scholar 

  10. Brassard J, Gottschalk M, Quessy S . (2004). Cloning and purification of the Streptococcus suis serotype 2 glyceraldehyde-3-phosphate dehydrogenase and its involvement as an adhesin. Vet Microbiol 102: 87–94.

    CAS  Article  Google Scholar 

  11. Brenner C, Kroemer G . (2000). Apoptosis. Mitochondria – the death signal integrators. Science 289: 1150–1151.

    CAS  Article  Google Scholar 

  12. Brenner C, Marzo I, de Araujo Vieira HL, Kroemer G . (2000). Purification and liposomal reconstitution of permeability transition pore complex. Methods Enzymol 322: 243–252.

    CAS  Article  Google Scholar 

  13. Cheng E, Sheiko T, Fisher J, Craigen W, Korsmeyer S . (2003). VDAC2 inhibits BAK activation and mitochondrial apoptosis. Science 301: 513–517.

    CAS  Article  Google Scholar 

  14. Chuang D, Hough C, Senatorov V . (2005). Glyceraldehyde-3-phosphate dehydrogenase, apoptosis, and neurodegenerative diseases. Annu Rev Pharmacol Toxicol 45: 269–290.

    CAS  Article  Google Scholar 

  15. Chuang DM, Ishitani R . (1996). A role for GAPDH in apoptosis and neurodegeneration. Nat Med 2: 609–610.

    CAS  Article  Google Scholar 

  16. Colombini M . (1983). Purification of VDAC (voltage-dependent anion-selective channel) from rat liver mitochondria. J Membr Biol 74: 115–121.

    CAS  Article  Google Scholar 

  17. Crompton M, Virji S, Ward JM . (1998). Cyclophilin-D binds strongly to complexes of the voltage-dependent anion channel and the adenine nucleotide translocase to form the permeability transition pore. Eur J Biochem 258: 729–735.

    CAS  Article  Google Scholar 

  18. Cuezva JM, Chen G, Alonso AM, Isidoro A, Misek DE, Hanash SM et al. (2004). The bioenergetic signature of lung adenocarcinomas is a molecular marker of cancer diagnosis and prognosis. Carcinogenesis 25: 1157–1163.

    CAS  Article  Google Scholar 

  19. Fukuhara Y, Takeshima T, Kashiwaya Y, Shimoda K, Ishitani R, Nakashima K . (2001). GAPDH knockdown rescues mesencephalic dopaminergic neurons from MPP+-induced apoptosis. NeuroReport 12: 2049–2052.

    CAS  Article  Google Scholar 

  20. Gincel D, Zaid H, Shoshan-Barmatz V . (2001). Calcium binding and translocation by the voltage-dependent anion channel: a possible regulatory mechanism in mitochondrial function. Biochem J 358: 147–155.

    CAS  Article  Google Scholar 

  21. Gong Y, Cui L, Minuk GY . (1996). Comparison of glyceraldehyde-3-phosphate dehydrogenase and 28s-ribosomal RNA gene expression in human hepatocellular carcinoma. Hepatology 23: 734–737.

    CAS  Article  Google Scholar 

  22. Green DR, Kroemer G . (2004). The pathophysiology of mitochondrial cell death. Science 305: 626–629.

    CAS  Article  Google Scholar 

  23. Halestrap AP, Brenner C . (2003). The adenine nucleotide translocase: a central component of the mitochondrial permeability transition pore and key player in cell death. Curr Med Chem 10: 1507–1525.

    CAS  Article  Google Scholar 

  24. Han D, Antunes F, Canali R, Rettori D, Cadenas E . (2003). Voltage-dependent anion channels control the release of the superoxide anion from mitochondria to cytosol. J Biol Chem 278: 5557–5563.

    CAS  Article  Google Scholar 

  25. Hara MR, Agrawal N, Kim SF, Cascio MB, Fujimuro M, Ozeki Y et al. (2005). S-nitrosylated GAPDH initiates apoptotic cell death by nuclear translocation following Siah1 binding. Nat Cell Biol 7: 665–674.

    CAS  Article  Google Scholar 

  26. Ishida A, Tada Y, Nimura T, Sueyoshi N, Katoh T, Takeuchi M et al. (2005). Identification of major Ca(2+)/calmodulin-dependent protein kinase phosphatase-binding proteins in brain: biochemical analysis of the interaction. Arch Biochem Biophys 435: 134–146.

    CAS  Article  Google Scholar 

  27. Ishitani R, Kimura M, Sunaga K, Katsube N, Tanaka M, Chuang DM . (1996). An antisense oligodeoxynucleotide to glyceraldehyde-3-phosphate dehydrogenase blocks age-induced apoptosis of mature cerebrocortical neurons in culture. J Pharmacol Exp Ther 278: 447–454.

    CAS  PubMed  Google Scholar 

  28. Ishitani R, Tanaka M, Sunaga K, Katsube N, Chuang DM . (1998). Nuclear localization of overexpressed glyceraldehyde-3-phosphate dehydrogenase in cultured cerebellar neurons undergoing apoptosis. Mol Pharmacol 53: 701–707.

    CAS  Article  Google Scholar 

  29. Jacotot E, Ravagnan L, Loeffler M, Ferri KF, Vieira HL, Zamzami N et al. (2000). The HIV-1 viral protein R induces apoptosis via a direct effect on the mitochondrial permeability transition pore. J Exp Med 191: 33–46.

    CAS  Article  Google Scholar 

  30. Korge P, Weiss JN . (1999). Thapsigargin directly induces the mitochondrial permeability transition. Eur J Biochem 265: 273–280.

    CAS  Article  Google Scholar 

  31. Laemmli UK . (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227: 680–685.

    CAS  Article  Google Scholar 

  32. Laschet JJ, Minier F, Kurcewicz I, Bureau MH, Trottier S, Jeanneteau F et al. (2004). Glyceraldehyde-3-phosphate dehydrogenase is a GABAA receptor kinase linking glycolysis to neuronal inhibition. J Neurosci 24: 7614–7622.

    CAS  Article  Google Scholar 

  33. Le Bras M, Clement MV, Pervaiz S, Brenner C . (2005). Reactive oxygen species and the mitochondrial signaling pathway of cell death. Histol Histopathol 20: 205–220.

    CAS  PubMed  Google Scholar 

  34. Lee A, Zizi M, Colombini M . (1994). Beta-NADH decreases the permeability of the mitochondrial outer membrane to ADP by a factor of 6. J Biol Chem 269: 30974–30980.

    CAS  PubMed  Google Scholar 

  35. Mazzola JL, Sirover MA . (2002). Alteration of intracellular structure and function of glyceraldehyde-3-phosphate dehydrogenase: a common phenotype of neurodegenerative disorders? Neurotoxicology 23: 603–609.

    CAS  Article  Google Scholar 

  36. Mazzola JL, Sirover MA . (2003). Subcellular localization of human glyceraldehyde-3-phosphate dehydrogenase is independent of its glycolytic function. Biochim Biophys Acta 1622: 50–56.

    CAS  Article  Google Scholar 

  37. Muller A, Gunther D, Brinkmann V, Hurwitz R, Meyer TF, Rudel T . (2000). Targeting of the pro-apoptotic VDAC-like porin (PorB) of Neisseria gonorrhoeae to mitochondria of infected cells. EMBO J 19: 5332–5343.

    CAS  Article  Google Scholar 

  38. Narita M, Shimizu S, Ito T, Chittenden T, Lutz RJ, Matsuda H et al. (1998). Bax interacts with the permeability transition pore to induce permeability transition and cytochrome c release in isolated mitochondria. Proc Natl Acad Sci USA 95: 14681–14686.

    CAS  Article  Google Scholar 

  39. Patterson RL, van Rossum DB, Kaplin AI, Barrow RK, Snyder SH . (2005). Inositol 1,4,5-trisphosphate receptor/GAPDH complex augments Ca2+ release via locally derived NADH. Proc Natl Acad Sci USA 102: 1357–1359.

    CAS  Article  Google Scholar 

  40. Pfaff E, Klingenberg M . (1968). Adenine nucleotide translocation of mitochondria. 1. Specificity and control. Eur J Biochem 6: 66–79.

    CAS  Article  Google Scholar 

  41. Poncet D, Larochette N, Pauleau AL, Boya P, Jalil AA, Cartron PF et al. (2004). An anti-apoptotic viral protein that recruits Bax to mitochondria. J Biol Chem 279: 22605–22614.

    CAS  Article  Google Scholar 

  42. Rostovtseva T, Antonsson B, Suzuki M, Youle R, Colombini M, Bezrukov S . (2004). Bid but not Bax regulates VDAC channels. J Biol Chem 279: 13575–13583.

    CAS  Article  Google Scholar 

  43. Ryzlak M, Pietruszko R . (1988). Heterogeneity of glyceraldehyde-3-phosphate dehydrogenase from human brain. Biochim Biophys Acta 954: 309–324.

    CAS  Article  Google Scholar 

  44. Sawa A, Khan AA, Hester LD, Snyder SH . (1997). Glyceraldehyde-3-phosphate dehydrogenase: nuclear translocation participates in neuronal and nonneuronal cell death. Proc Natl Acad Sci USA 94: 11669–11674.

    CAS  Article  Google Scholar 

  45. Schagger H, Cramer W, von Jagow G . (1994). Analysis of molecular masses and oligomeric states of protein complexes by blue native electrophoresis and isolation of membrane protein complexes by two-dimensional native electrophoresis. Anal Biochem 217: 220–230.

    CAS  Article  Google Scholar 

  46. Shi Y, Jiang C, Chen Q, Tang H . (2003). One-step on-column affinity refolding purification and functional analysis of recombinant human VDAC1. Biochem Biophys Res Commun 303: 475–482.

    CAS  Article  Google Scholar 

  47. Shimizu S, Narita M, Tsujimoto Y . (1999). Bcl-2 family proteins regulate the release of apoptogenic cytochrome c by the mitochondrial channel VDAC. Nature 399: 483–487.

    CAS  Article  Google Scholar 

  48. Shirakata Y, Koike K . (2003). Hepatitis B virus X protein induces cell death by causing loss of mitochondrial membrane potential. J Biol Chem 278: 22071–22078.

    CAS  Article  Google Scholar 

  49. Sirover MA . (1997). Role of the glycolytic protein, glyceraldehyde-3-phosphate dehydrogenase, in normal cell function and in cell pathology. J Cell Biochem 66: 133–140.

    CAS  Article  Google Scholar 

  50. Sunaga K, Takahashi H, Chuang DM, Ishitani R . (1995). Glyceraldehyde-3-phosphate dehydrogenase is over-expressed during apoptotic death of neuronal cultures and is recognized by a monoclonal antibody against amyloid plaques from Alzheimer's brain. Neurosci Lett 200: 133–136.

    CAS  Article  Google Scholar 

  51. Susin SA, Lorenzo HK, Zamzami N, Marzo I, Brenner C, Larochette N et al. (1999). Molecular characterization of mitochondrial apoptosis-inducing factor. J Exp Med 189: 381–393.

    CAS  Article  Google Scholar 

  52. Tajima H, Tsuchiya K, Yamada M, Kondo K, Katsube N, Ishitani R . (1999). Over-expression of GAPDH induces apoptosis in COS-7 cells transfected with cloned GAPDH cDNAs. Neuroreport 10: 2029–2033.

    CAS  Article  Google Scholar 

  53. Thinnes FP, Florke H, Winkelbach H, Stadtmuller U, Heiden M, Karabinos A et al. (1994). Channel active mammalian porin, purified from crude membrane fractions of human B lymphocytes or bovine skeletal muscle, reversibly binds the stilbene-disulfonate group of the chloride channel blocker DIDS. Biol Chem Hoppe Seyler 375: 315–322.

    CAS  Article  Google Scholar 

  54. Tisdale EJ, Kelly C, Artalejo CR . (2004). Glyceraldehyde-3-phosphate dehydrogenase interacts with Rab2 and plays an essential role in endoplasmic reticulum to Golgi transport exclusive of its glycolytic activity. J Biol Chem 279: 54046–54052.

    CAS  Article  Google Scholar 

  55. Vander Heiden MG, Chandel NS, Li XX, Schumacker PT, Colombini M, Thompson CB . (2000). Outer mitochondrial membrane permeability can regulate coupled respiration and cell survival. Proc Natl Acad Sci USA 97: 4666–4671.

    CAS  Article  Google Scholar 

  56. Verrier F, Deniaud A, LeBras M, Metivier D, Kroemer G, Mignotte B et al. (2004). Dynamic evolution of the adenine nucleotide translocase interactome during chemotherapy-induced apoptosis. Oncogene 23: 8049–8064.

    CAS  Article  Google Scholar 

  57. Weaver JG, Tarze A, Moffat TC, Lebras M, Deniaud A, Brenner C et al. (2005). Inhibition of adenine nucleotide translocator pore function and protection against apoptosis in vivo by an HIV protease inhibitor. J Clin Invest 115: 1828–1838.

    CAS  Article  Google Scholar 

  58. Wu SL, Sampson MJ, Decker WK, Craigen WJ . (1999). Each mammalian mitochondrial outer membrane porin protein is dispensable: effects on cellular respiration. Biochimica Et Biophysica Acta – Mol Cell Res 1452: 68–78.

    CAS  Article  Google Scholar 

  59. Zaid H, Abu-Hamad S, Israelson A, Nathan I, Shoshan-Barmatz V . (2005). The voltage-dependent anion channel-1 modulates apoptotic cell death. Cell Death Differ 12: 751–760.

    CAS  Article  Google Scholar 

  60. Zamzami N, Marchetti P, Castedo M, Decaudin D, Macho A, Hirsch T et al. (1995a). Reduction in mitochondrial potential constitutes an early irreversible step of programmed lymphocyte death in vivo. J Exp Med 182: 367–377.

    CAS  Article  Google Scholar 

  61. Zamzami N, Marchetti P, Castedo M, Zanin C, Vayssière J-L, Petit PX et al. (1995b). Sequential reduction of mitochondrial transmembrane potential and generation of reactive oxygen species in early programmed cell death. J Exp Med 181: 1661–1672.

    CAS  Article  Google Scholar 

  62. Zizi M, Byrd C, Boxus R, Colombini M . (1998). The voltage-gating process of the voltage-dependent anion channel is sensitive to ion flow. Biophys J 75: 704–713.

    CAS  Article  Google Scholar 

Download references


We thank C Henry (platform PAPSS, INRA, and C Longin (platform MIMA2, for their professional assistance in mass spectrometry and electron microsocopy, respectively. This work was supported by grants funded by ARC, CNRS and French Ministry of Research. (to CB), by a special grant of the League against Cancer and by the European Union (Trans-Death, RIGHT) (to GK). MLB received a postdoctoral fellowship of CNRS. AT and AD received a doctoral and post doctoral fellowships from the Ligue Contre le Cancer.

Author information



Corresponding author

Correspondence to C Brenner.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Tarze, A., Deniaud, A., Le Bras, M. et al. GAPDH, a novel regulator of the pro-apoptotic mitochondrial membrane permeabilization. Oncogene 26, 2606–2620 (2007).

Download citation


  • apoptosis
  • mitochondria
  • VDAC
  • ANT
  • permeability transition

Further reading


Quick links