Mitochondrial outer membrane permeabilization by proapoptotic Bcl-2 family proteins, such as Bax, plays a crucial role in apoptosis induction. However, whether this only causes the intracytosolic release of inducers of caspase-dependent death, such as cytochrome c, or also of caspase-independent death, such as apoptosis-inducing factor (AIF) remains unknown. Here, we show that on isolated mitochondria, Bax causes the release of cytochrome c, but not of AIF, and the association of AIF with the mitochondrial inner membrane provides a simple explanation for its lack of release upon Bax-mediated outer membrane permeabilization. In cells overexpressing Bax or treated either with the Bax- or Bak-dependent proapoptotic drugs staurosporine or actinomycin D, or with hydrogen peroxide, caspase inhibitors did not affect the intracytosolic translocation of cytochrome c, but prevented that of AIF. These results provide a paradigm for mitochondria-dependent death pathways in which AIF cannot substitute for caspase executioners because its intracytosolic release occurs downstream of that of cytochrome c.
Most proapoptotic stimuli require a mitochondria-dependent step, involving outer membrane permeabilization controlled by pro- and antiapoptotic members of the Bcl-2 family and leading to intracytosolic release of mitochondria intermembrane space proteins that can either trigger caspase activation, such as cytochrome c, or caspase-independent death pathways, such as apoptosis-inducing factor (AIF;*Martinou and Green, 2001; Zamzami and Kroemer, 2001). Bax is one of the main proapoptotic Bcl-2 family proteins; the presence of either Bax or Bak being required for most mitochondria-dependent cell death processes, including those induced by tBid, the caspase-activated form of the “BH3-domain only” proapoptotic Bcl-2 family protein Bid and by proapoptotic drugs such as staurosporine and actinomycin D (Wei et al., 2001). Bax has been reported to induce cytochrome c release both in Bax-expressing cells and in isolated mitochondria incubated with recombinant oligomerized Bax (Eskes et al., 1998; Jurgenmeier et al., 1998; Finucane et al., 1999; Antonsson et al., 2000). However, whether Bax-mediated mitochondria outer membrane permeabilization also induces the release of AIF (Susin et al., 1999) is so far unknown.
Here, we show that the mitochondrial outermembrane permeabilization induced either by Bax, by Bax- or Bak-dependent proapoptotic drugs, or by hydrogen peroxide (H2O2) results in the intracytosolic release of cytochrome c, but that subsequent caspase activation is required to induce the translocation of AIF into the cytosol. These findings identify the mitochondrial response to several proapoptotic stimuli as a selective process leading to a hierarchical ordering of the effectors involved in cell death induction.
Results And Discussion
To investigate whether Bax may induce the mitochondrial release of AIF, we incubated freshly purified mitochondria from human HeLa cells with recombinant oligomeric Bax, and performed Western Blot analysis of cytochrome c and AIF in both the mitochondria supernatants and pellets. For the detection of AIF, we used a pAb specific for human AIF (Fig. S1), obtained as indicated in Materials and methods. Mitochondrial release of cytochrome c was maximal for a Bax concentration of 100 nM (Fig. 1 A), and a kinetic analysis indicated that cytochrome c release was rapid; almost completed in 15 min (Fig. 1 B). In contrast, Bax did not induce any detectable mitochondrial release of AIF after incubations for 30 min with Bax concentrations up to 200 nM (Fig. 1, A and B). Because the isolated mitochondria we used were purified from HeLa cancer cells, and because cancer cells accumulate various alterations that favor cell death repression (Evan and Littlewood, 1998), we investigated the response of mitochondria from primary cells. Using another commercially available pAb specific for AIF, we explored the response of mitochondria from rat primary liver cells to either oligomeric Bax or to a form of oligomerized Bax obtained by mixing recombinant monomeric Bax with low concentrations of recombinant tBid. Although neither 100 nM monomeric Bax nor 10 nM tBid alone had any effect, together they induced the release of cytochrome c, but not of AIF (Fig. 1 C), as oligomeric Bax (unpublished data). At higher concentrations (50 or 100 nM), tBid by itself also induced the release of cytochrome c, but not of AIF (Fig. 1 D). Because tBid requires the presence of endogenous Bak on mitochondria to induce cytochrome c release from isolated murine liver cell mitochondria (Wei et al., 2000), tBid alone probably acted through the formation of tBid/Bak oligomers. Together, our data indicated that the lack of AIF release did not depend on the cell (primary or cancer) or species (human or rat) origin of the mitochondria.
AIF was originally described as a soluble protein localized in the mitochondrial intermembrane space (Susin et al., 1999). Therefore, a possible explanation for our findings was that Bax and tBid induce the formation of selective outer membrane pores (Martinou and Green, 2001), allowing the passage of cytochrome c but not of AIF. Alternately, Bax and tBid cause a nonselective process of outer membrane permeabilization (Zamzami and Kroemer, 2001), but AIF is not an intermembrane space–soluble protein. To discriminate between these two possibilities, we explored the localization of AIF and cytochrome c in subfractions of purified mitochondria from rat primary liver cells. When compared with its total amount in whole mitochondria, only a part of cytochrome c colocalized both with the mitoplasts (MP), obtained after outer membrane removal and consisting of the mitochondrial inner membranes (MIMs) and the matrix, and with the enriched, purified inner membranes (Fig. 2 A). In contrast, around the same amount of AIF colocalized with whole mitochondria, MP, and MIM, suggesting an absence of soluble AIF in the intermembrane space (Fig. 2 A). To investigate whether AIF is an integral component of the inner membrane or is rather peripherally associated with it, we subjected the MPs to sodium carbonate treatment. The alkali treatment caused a complete loss of AIF colocalization with the MPs, whereas the cytochrome c oxidase subunit IV (Cox IV) protein, an integral component of the inner membrane, was unaffected (Fig. 2 B). Together, these findings indicated that most AIF is peripherally associated with the mitochondrial inner membrane, presumably with its external side, and suggested that the outer membrane permeabilization induced by Bax or tBid is not sufficient to allow the detachment of AIF from the inner membrane.
Next, we investigated whether the mitochondrial release of cytochrome c may also be dissociated from that of AIF in cells overexpressing Bax. Bax-mediated mitochondrial release of cytochrome c does not require caspase activation and occurs in the presence of caspase inhibitors (Eskes et al., 1998; Finucane et al., 1999; Martinou and Green, 2001). We transiently transfected human 293T cells with an expression vector encoding Bax, in the absence or presence of the broad caspase inhibitor peptide z-VAD-fmk (100 μM), and performed a kinetic Western blot analysis. Bax expression led to a time-dependent increase in intracytosolic release of cytochrome c and AIF, caspase-9, and caspase-3 processing, PARP cleavage, and cell death (Fig. 3 A). z-VAD-fmk did neither prevent Bax expression nor Bax-induced intracytosolic release of cytochrome c, but prevented cell death induction (Fig. 3 A), nuclear DNA degradation (hypodiploidy) and mitochondrial transmembrane potential (ΔΨm) loss (Fig. S2), caspase-9, and caspase-3 processing, PARP cleavage, and the intracytosolic release of AIF (Fig. 3 A). 100μM BAF, another broad caspase inhibitor, showed the same effect as z-VAD-fmk, preventing both cell death and the release of AIF, but not the release of cytochrome c (Fig. 3 B). To confirm these findings, we used immunofluorescence microscopy analysis to compare the localization of Bax, cytochrome c, AIF, and Hsp60 (a protein of the mitochondria matrix) in HeLa cells 18 h after transfection with a vector expressing GFP-Bax, in the absence or presence of z-VAD-fmk. Some nonspecific nuclear staining was induced by the polyclonal anti-AIF antibody (Fig. S1), and also by the control sera (unpublished data) in control HeLa cells. z-VAD-fmk prevented the induction of nuclear features of apoptosis in GFP-Bax–expressing cells (Fig. 3 C), but did neither prevent the induction of a punctuated cytosolic staining of Bax, that colocalized with Hsp 60, indicating an insertion of Bax into mitochondria, nor the induction of a diffuse cytosolic staining of cytochrome c (Fig. 3 C and D). In contrast, z-VAD-fmk treatment prevented the induction of a diffuse staining of AIF (Fig. 3, C and D), confirming that Bax did not induce the mitochondrial release of AIF in the presence of caspase inhibitors.
The proapoptotic drug staurosporine does not require the presence of AIF to cause cell death (Joza et al., 2001), but requires the presence of either Bax or Bak to induce mitochondrial release of cytochrome c and cell death (Wei et al., 2001). We treated HeLa cells with 2 μM staurosporine for 9 h in the absence or presence of 100 μM z-VAD-fmk. z-VAD-fmk prevented staurosporine-mediated apoptosis induction (Fig. 4 A), as well as caspase-9 and caspase-3 processing, PARP cleavage (Fig. 4 B), and the mitochondrial release of AIF (Fig. 4 C), but had no effect, as previously reported (Goldstein et al., 2000), on the mitochondrial release of cytochrome c (Fig. 4 C). Actinomycin D, as staurosporine, requires the presence of Bax or Bak to induce cytochrome c release and cell death (Wei et al., 2001). z-VAD-fmk also prevented apoptosis induction by 10 μM actinomycin D (16 ± 5 versus 68 ± 7), as well as mitochondrial release of AIF, but not of cytochrome c (Fig. 4 D).
Because staurosporine induced a rapid and marked cell shrinkage even in the presence of z-VAD-fmk, immunofluorescence studies were difficult to interpret (unpublished data). In contrast, cells did not undergo marked shrinkage when treated with actinomycin D and z-VAD-fmk. Therefore, we performed immunofluorescence analysis of cytochrome c and AIF in HeLa cells treated with actinomycin D, and observed that z-VAD-fmk prevented the mitochondrial release of AIF, though having no effect on the release of cytochrome c (Fig. 4, E and F). For further confirmation, we increased the detectability of AIF by overexpressing full-length human AIF in HeLa cells. 18 h after AIF transfection, we treated the HeLa cells for 9 h with actinomycin D in the absence or presence of z-VAD-fmk, and observed the same pattern of cytochrome c release and lack of AIF release in the cells treated with both actinomycin D and z-VAD-fmk (Fig. 4 G).
Cell death due to the mitochondrial release of AIF has been reported to occur in the presence of caspase inhibitors in response to certain stimuli (Joza et al., 2001; Pardo et al., 2001), including H2O2 (Yu et al., 2002). Using an AIF-specific mAb that did not induce background nuclear staining, we performed immunofluorescence analysis of HeLa cells treated with 400 μM H2O2 in the absence or presence of z-VAD-fmk. As in Bax-expressing and in actinomycin D–treated cells, z-VAD-fmk prevented both apoptosis and AIF release in H2O2-treated cells, while not affecting cytochrome c release (Fig. 5, A–C). Higher H2O2 concentrations caused necrotic cell death but did not induce AIF release in the presence of z-VAD-fmk (unpublished data). The very recent paper that intramitochondrial AIF acts as a free radical scavenger decreasing H2O2-mediated cell death (Klein et al., 2002) is consistent with the notion that AIF is not a selective effector of H2O2-induced death. Whether particular stimuli may allow a selective caspase-independent process of AIF release remains to be confirmed. Our observation that AIF is not a soluble mitochondrial intermembrane space protein, but rather is attached to the inner membrane, provides a simple potential explanation for our findings, and for recent findings by others, indicating that AIF is not among the proteins released by isolated mitochondria on incubation with tBid (Van Loo et al., 2002). Our observation that caspase inhibitors prevent AIF release in cells exposed to several proapoptotic stimuli, including Bax, Bax- or Bak-dependent proapoptotic drugs, and H2O2 strongly suggests that activation of given caspases after mitochondrial outer membrane permeabilization may be the limiting step for AIF detachment from the inner membrane and intracytosolic release. However, because caspase inhibitors may also prevent the activity of other proteases (Wolf et al., 1999), the precise pathway leading to AIF release downstream of cytochrome c release remains to be investigated.
Here, we have identified the mitochondrial response to Bax and other proapoptotic stimuli as a very selective process leading to a hierarchical ordering of the effectors involved in cell death induction. Our findings provide a paradigm for mitochondria-dependent cell death pathways involving a postmitochondrial level of pharmacological and possibly endogenous regulation that precedes the intracytosolic release of the caspase-independent death effector AIF.
Materials And Methods
Mitochondria isolation and in vitro cytochrome c and AIF release
Mitochondria were isolated from HeLa cells and rat liver by sucrose density gradient centrifugation, and in vitro cytochrome c and AIF release was assessed as described previously (Eskes et al., 1998; Desagher et al., 1999). Production of full-length recombinant oligomeric, monomeric Bax and monomeric tBid (Antonsson et al., 2000; Kudla et al., 2000), and preparation of mitochondrial membrane vesicles from rat liver mitochondria (Mayer et al., 1993) were performed as described previously.
Cell culture and transfection
HEK 293T and HeLa cells were cultured in DME (GIBCO BRL) supplemented with 10% FCS, 2 mM l-glutamine, 50 IU penicillin, and 50 μg·ml−1 streptomycin under standard conditions. Transient transfections of 293T and of HeLa cells were performed using calcium phosphate and LipofectAMINE™ (GIBCO BRL), respectively. pcDNA-AIF (Susin et al., 1999) was provided by G. Kroemer and S. Susin (CNRS-UMR 8125, Institut Gustave Roussy, Villejuif, France).
Cell death measurement
At different time points after transfection or drug treatment, cells were recovered after incubation in PBS containing 1 mM EDTA, washed in PBS, and fixed with 4% PFA. Nuclei were then colored with Hoechst 33342 (Sigma-Aldrich). Green cells with blebbing, shrinkage, and apoptotic (condensed and fragmented) nuclei were considered as dying.
Cells were harvested in isotonic mitochondrial buffer (210 mM mannitol, 70 mM sucrose, 1 mM EDTA, and 10 mM Hepes, pH 7.5) supplemented with protease inhibitor cocktail Complete (Boehringer), and homogenized for 30–40 strokes with a Dounce homogenizer. Samples were transferred to Eppendorf centrifuge tubes and centrifuged at 500 g for 5 min at 4°C to eliminate nuclei and unbroken cells. Supernatant was then centrifuged at 10,000 g for 30 min at 4°C to obtain the heavy membrane pellet enriched for mitochondria, and the resulting supernatant was stored as the cytosolic fraction.
Preparation of cellular lysates, immunoblotting, and immunofluorescence were performed as described previously (Desagher et al., 1999; Finucane et al., 1999). Antibodies used in immunoblotting were as follows: mouse mAb against cytochrome c (clone 7H8.2C12; PharMingen), PARP (BD Biosciences), Bax (N20; Santa Cruz Biotechnology, Inc.), VDAC (Calbiochem), Cox IV (Molecular Probes, Inc.), or actin (Sigma-Aldrich), and rabbit pAbs specific for caspase-9 (Cayman Chemical). For human AIF detection, we used a rabbit polyclonal anti-AIF antibody that we obtained from rabbits immunized against a mixture of three different human AIF peptides (amino acids 106–120, 512–526, and 588–602). Each peptide was detected by the anti-AIF antibodies up to the dilution of 1/10,000 by ELISA. The anti-AIF antibody also detected recombinant human AIF protein (Susin et al., 1999), provided by G. Kroemer and S. Susin. For rat AIF detection, we used a goat polyclonal anti-AIF antibody (D-20; Santa Cruz Biotechnology, Inc.). All antibodies were used at dilution 1/1,000, then visualized using HRP-conjugated secondary antibodies (Amersham Biosciences), followed by enhanced chemiluminescence (Amersham Biosciences). Antibodies used in immunofluorescence were as follows: our rabbit polyclonal antibody against human AIF (1/200) or in Fig. 5, a commercially available mouse IgG2b mAb against human AIF (AIF [E-1]; Santa Cruz Biotechnology, Inc.) (1/50), that was raised against a recombinant protein corresponding to amino acids 1–300 of human AIF; an anti–cytochrome c mAb (6H2.B4, 1/200; BD Biosciences) or a sheep polyclonal anti–cytochrome c (1/600; Sigma-Aldrich), and an anti-Hsp60 mAb (1/200; StressGen Biotechnologies). Antibodies were then detected using TRITC-labeled or FITC-labeled secondary antibody (1/200), and cells examined under a UV Fluorescence Microscope (3CCD; Leica) or a Confocal Microscope (LSM 510; Carl Zeiss MicroImaging, Inc.).
Online supplemental material
The specificity of our rabbit polyclonal antibody against human AIF was assessed by Western blotting and immunofluorescence assays (Fig. S1). The preventive effect of the caspase inhibitor z-VAD-fmk on Bax-induced cell death was investigated by flow cytometry analysis of DNA degradation and mitochondrial permeability (ΔΥm) loss in Bax-transfected 293 T cells (Fig. S2).
We thank G. Kroemer and S. Susin for recombinant human AIF protein and AIF expression vector pcDNA-AIF (Susin et al., 1999), and S. Desagher (Montpellier, France) for helpful discussions.
This work was supported by INSERM, Université Paris 7, AP-HP, and grants from Agence Nationale de Recherches sur le Sida, Ensemble contre le Sida, Paris 7 Valorisation, Etablissement Francais des Greffes and Fondation pour la Recherche Medicale (FRM) (to J.C. Ameisen), Fonds National Suisse (subside no. 3100-061380; to J.C. Martinou), and doctoral fellowships from Delegation Generale de l'Armement and from FRM (to D. Arnoult).
Note added in proof. While our paper was in proof, it was reported that the mitochondrial release of the AIF homologue WAH-1 is also caspase (CED-3) dependent in the nematode Caenorhabditis elegans (Wang, X., C. Yang, J. Chai, Y. Shi, D. Xue. 2002. Science. 298:1587–1592). Thus, the process that we describe here, in mammalian cells, may be evolutionarily conserved.
The online version of this article contains supplemental material.
Abbreviations used in this paper: AIF, apoptosis-inducing factor; Cox IV, cytochrome c oxidase subunit IV; H2O2, hydrogen peroxide; MIM, mitochondrial inner membrane; MOM, mitochondrial outer membrane; MP, mitoplast.