Bcl-2 is a prosurvival factor that reportedly prevents the nonspecific permeabilization of mitochondrial membranes, yet enhances specific ADP/ATP exchange by these organelles. Here, we show that Bcl-2 enhances the ADP/ATP exchange in proteoliposomes containing the purified adenine nucleotide translocase (ANT) in isolated mitochondria and mitoplasts, as well as in intact cells in which mitochondrial matrix ATP was monitored continuously using a specific luciferase-based assay system. Conversely, Bax, which displaces Bcl-2 from ANT in apoptotic cells, inhibits ADP/ATP exchange through a direct action on ANT. The Bax-mediated inhibition of ADP/ATP exchange can be separated from Bax-stimulated formation of nonspecific pores by ANT. Chemotherapy-induced apoptosis caused an inhibition of ANT activity, which preceded the loss of the mitochondrial transmembrane potential and could be prevented by overexpression of Bcl-2. These data are compatible with a model of mitochondrial apoptosis regulation in which ANT interacts with either Bax or Bcl-2, which both influence ANT function in opposing manners. Bcl-2 would maintain the translocase activity at high levels, whereas Bax would inhibit the translocase function of ANT.
Bcl-2-like proteins are assumed to exert most of their apoptosis-regulatory function at the level of mitochondrial membranes (1, 2, 3, 4, 5). Bcl-2-like antiapoptotic proteins are constitutively present in mitochondria and would favor the maintenance of normal mitochondrial function, namely by preventing nonspecific permeabilization of mitochondrial membranes while simultaneously maintaining the exchange of small molecules (e.g., ADP, ATP, NADH) on mitochondrial membranes, which is essential for oxidative phosphorylation. Conversely, Bax-like proapoptotic proteins, which translocate to mitochondria only in conditions of apoptosis induction, would favor MMP4 and irreversible loss of mitochondrial function. Although Bcl-2 and Bax can form heterodimers and neutralize each other through direct physical interactions, it appears that both proteins can modulate mitochondrial function and apoptosis independently from each other (1, 2, 3, 4, 5). The exact molecular mechanisms through which these effects are achieved are a matter of intense debate exacerbated by contradictory findings. For instance, one group reported that Bcl-2 (or its close homologue Bcl-XL) would prevent the VDAC to form nonspecific, large channels leading to the MMP-associated release of proteins from mitochondria (6, 7, 8). In strict contrast, another group reported that Bcl-2 would prevent VDAC to close and that this effect would maintain specific nutrient exchange on mitochondrial membranes (9, 10).
Intrigued by these contradictions, we decided to reevaluate the effect of Bcl-2 and Bax on another mitochondrial protein, namely the ANT, which is known to interact with VDAC within the so-called PTPC (5, 11, 12). ANT is a bifunctional protein that, in physiological conditions, exchanges ATP and ADP on the inner mitochondrial membrane and, in apoptotic conditions, can form a nonspecific pore. According to our published observations (13, 14, 15), pore formation by ANT would require a physical interaction with Bax as well as specific interactions with proapoptotic ANT ligands such as Atr and the protein Vpr from human immunodeficiency virus-1. In contrast, Bcl-2 would suppress pore formation by ANT (14). Here, we report that Bcl-2 and Bax do not only modulate pore formation by ANT but, importantly, also influence the enzymatic function of ANT as an ADP/ATP antiporter. These results have been obtained in a variety of different systems (proteoliposomes containing purified ANT, Bcl-2, and/or Bax; isolated mitochondria and mitoplasts; intact cells) and shed new lights on the apoptosis-regulatory functions of Bcl-2/Bax-like proteins.
MATERIALS AND METHODS
Immunogold and Coimmunoprecipitation Assays.
HeLa-Bcl-2 cells (16) were cryofixed and subjected to immunoelectron microscopy for the detection of ANT (16) and Bcl-2 (anti-Bcl-2 ΔC 21 mAb; Santa Cruz Biotechnology, Santa Cruz, CA). HT29 cells (5 × 106 cells/75 cm2 flask) were treated with 100 μm etoposide for 20 h to induce apoptosis, and mitochondria were isolated (17) 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 concentration of 0.4 mg protein/ml. The immunoprecipitation was performed by adding 20 μl of a rabbit polyclonal anti-rat heart ANT serum (18) to 100 μl of mitochondrial suspension (90 min, 37°C). Then, 40 μl of protein A/protein G agarose beads (Santa Cruz Biotechnology) were added (30 min, 37°). The beads were washed three times with 1 ml of PBS and resuspended in electrophoresis sample buffer. Proteins were analyzed by SDS-PAGE (12.5%, 25 μg of protein/lane) and immunoblotting with anti-rat heart ANT polyclonal serum (17) or mAbs specific for Bax (P-19; Santa Cruz Biotechnology), cytochrome c (7H8.2C12; PharMingen), or Bcl-2 (anti-Bcl-2 ΔC 21 mAb; Santa Cruz Biotechnology).
ANT was purified from rat heart mitochondria (19), routinely checked to be VDAC-free (14) and reconstituted into proteoliposomes (phosphatidylcholin/cardiolipin [45:1; w:w]) either alone or in the presence of recombinant Bax, BaxΔIGDE, Bcl-2, or Bcl-2Δ145 (20) at the indicated molar ratios. Proteoliposomes were loaded either with 4-MUP (17) or with ATP (1 mm) in 10 mm KCl, 10 mm HEPES, 125 saccharose (pH 7.4), by sonication (25% 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 the indicated agents (AMP, ADP, Atr) at room temperature for 30 min, as described previously (13, 21, 22). The dose of ANT (quantified by the Bradford method) in liposomes was 0.1 mg/ml. To quantify ATP release from liposomes, a luciferase/luciferin-based luminescence method (kit HS II; Boerhinger Mannheim, Germany) was used. The release of 4-MUP was quantified by addition of alkaline phosphatase (which converts 4-MUP into the fluorochrome 4-methylumbelliferone; Ref. 17). The maximum 4-MUP release was determined by adding 5% Triton X-100 to proteoliposomes. The percentage of 4-MUP release induced by treatment of liposomes by Atr was determined as [(Atr − treated liposomes fluorescence − untreated liposomes fluorescence)/(TX-100-treated liposomes fluorescence − untreated liposomes fluorescence)] × 100. The maximal fluorescence induced by 200 μm Atr (which induces a specific, fully ADP-inhibitable 4-MUP release from ANT proteoliososmes) was then identified as 100% 4-MUP release, and the fluorescence induced by the treatment of liposomes by another product was calculated as a percentage of Atr-induced 4-MUP release.
ADP/ATP Translocase Activity in Isolated Mitochondria and Mitoplasts.
Liver mitochondria (from female, 8–12-month-old C57/Bl6 mice and age- and sex-matched congenic mice expressing a Bcl-2 transgene under the l-type pyruvate kinase gene promoter) (23) were purified (18) and suspended (7 mg protein/500 μl) in 0.6 m mannitol, 0.2% BSA, 10 mm MOPS, and 0.1 mm EDTA (pH 6.8). The respiratory rate, respiratory control, and membrane potential of mitochondria purified from control livers and Bcl-2-overexpressing livers were undistinguishable (data not shown). For kinetic analyses of ADP/ATP exchange, we modified a method developed by Klingenberg et al. (24). Isolated mitochondria were incubated with 15 μl of [2,8-3H] ATP (40 Ci/mmol) for 45 min at 4°C in the presence or absence of recombinant Bax protein (2 μg/ml), and washed twice to eliminate free [2,8-3H] ATP. The exchange was initiated by addition of cold ADP (standard dose: 400 μm) and stopped by addition of 100 μm Atr after 10 s and centrifugation (6800 × g, 10 min, 4°C). Mitoplasts were generated by incubation of fresh mitochondria (7 mg protein/500 μl) [mitochondria:hypotonic buffer, 1:10, v:v] in 20 mm HEPES, 1 mg/ml BSA (pH 7.4) for 20 min at 4°C. KPA and NADH were used at 2 and 300 μm, respectively.
Dynamic in Vivo Measurement of Mitochondrial ATP.
HeLa cells stably transfected with the Neomycin resistance gene (pCDNA3.1) or human Bcl-2 (16) were seeded onto glass coverslips and grown to 50% confluence before transfection with 4 μg of DNA encoding mtLuc (luciferase fused to the mitochondrial presequence of COX VIII; Ref. 25). Thirty-six h later, coverslips (with 2–3 × 105 cells) were transferred to the perfusion chamber, treated with digitonin 100 μm for 1 min, and washed for 2 min before addition of luciferin (20 μm), ATP, and/or ADP. All measurements were carried out in a buffer mimicking the cytosolic ionic composition [130 mm KCl, 10 mm NaCl, 1 mm MgSO4, 5 mm succinate, 0.5 mm K2HPO4, and 20 mm HEPES (pH 7.0) at 37°C). Drugs were added in the same medium. Luminescence was recorded continuously in a low-noise luminometer (26), and results were normalized on the maximum luminescence, obtained with an equimolar mixture of ADP and ATP (e.g., 250 μm ATP + 250 μm ADP), which depends on the total number of luciferase-transfected cells. For apoptosis induction, STS was used at 2 μm. Cells were incubated with 3,3′ dihexyloxacarbocyanine iodide, 20 nm, 15 min, 37°C, followed by cytofluorometry on a FACS Vantage (Becton Dickinson) while gating the forward and the side scatters on viable cells to obtain information on the Δψm (27, 28).
RESULTS AND DISCUSSION
ANT Physically Interacts with Bcl-2 in Normal Cells and with Bax during Apoptosis.
Bcl-2 reportedly is confined to the outer mitochondrial membrane, which would be in conflict with a possible Bcl-2 effect on ANT (see Refs. 1, 2, 3, 4, 6, 7, 8, 9, 10, 29, 30 but Refs. 31, 32). Immune electron microscopic examination of mitochondria in intact cells revealed, however, that Bcl-2 is also found within the inner mitochondrial membrane, mainly organized in clusters (Fig. 1,A). Coimmunoprecipitation assays confirmed an interaction between Bcl-2 and ANT (Fig. 1,B, IP) previously suggested by yeast-two-hybrid studies (13). Upon induction of apoptosis, the interaction between ANT and Bax increased while that with Bcl-2 decreased (Fig. 1,B), concomitantly with the translocation of Bax to mitochondria (Fig. 1,B, Mito), and the release of cytochrome c (Fig. 1 B, Mito). These observations indicate dynamic changes in the intramolecular interactions operating within the PTPC. Moreover, they underscore the possibility that ANT switches from a preponderant Bcl-2-associated state (in physiological conditions) to a mostly Bax-associated state (in apoptosis).
Bcl-2 and Bax Modulate ANT-mediated ADP/ATP Exchange in Proteoliposomes.
ANT proteoliposomes (Fig. 2,A) can be used as a model system for the simultaneous detection of membrane permeabilization (detected as the release of encapsulated 4-MUP) and of specific ADP/ATP antiport (detected as the release of encapsulated ATP, stimulated by exogenous ADP; Fig. 2, B and C). Using this system for separate assessment of the two ANT functions (pore and antiporter), we found that a number of proapoptotic agents such as Atr (Fig. 2,B), Ca2+, diazenedicarboxylic acid bis 5N,N-dimethylamide, and t-butylhydroperoxide (data not shown) did induce ANT pore opening. The dose response (Fig. 2,B) and kinetics (data not shown) of the Atr-induced ANT pore opening was similar to that reported in the literature (21, 33, 34), and the Atr-induced pore opening was inhibited by 500 μm ADP, as to be expected (33, 34). Atr failed to permeabilize protein-free liposomes or liposomes containing monomeric Bax (data not shown), as to be expected from electrophysiological studies on recombinant Bax incorporated into planar bilayers (14). Monomeric recombinant Bax failed to cause 4-MUP release when incorporated into ANT liposomes alone. Bax increased the response of ANT proteoliposomes to the Atr-mediated membrane permeabilization (Fig. 2,D). However, Bax (but not the inactive Bax mutant ΔIGDE) decreased, in a dose-dependent fashion, the translocase activity of ANT (Fig. 2,E). Bcl-2 (but not its inactive truncation mutant Bcl-2Δ145) counteracted the Atr-stimulated ANT-mediated 4-MUP release (Fig. 2,F). In addition, full-length Bcl-2 did cause a dose-dependent increase in ANT-mediated ADP/ATP antiport (Fig. 2 G). These results indicate that, in addition to regulating pore formation by ANT, Bcl-2 and Bax influence the ADP/ATP translocase activity of ANT.
Bcl-2 and Bax Modulate ADP/ATP Exchange in Mitochondria and Mitoplasts.
When added to purified mouse liver mitochondria, recombinant Bax incorporated into mitochondrial membranes (35, 36). We used nonoligomeric Bax at a dose that does not induce mitochondrial swelling (35, 36). Preincubation with Bax caused a significant reduction in ADP/ATP exchange, as compared with sham-preincubated control mitochondria, with a decrease in the Km and Vmax of ADP/ATP exchange (determined after 10 s of incubation with ADP), suggesting (but by no mean proving) that Bax competes for ADP binding (Fig. 3,A), a notion that would be in accord with the vicinity of the ADP and Bax-binding domains of ANT (13, 37). Alternatively, a conformational transition could explain the Bax effect. In addition, we found that purified mitochondria from mice overexpressing a liver-targeted Bcl-2 transgene (23) exhibited an increased ADP/ATP exchange, with a doubling of Vmax, yet no major change in the Km (Fig. 3,B). Similar results were found when mitoplasts (i.e., mitochondria with disrupted outer membrane) were used to assess the Bax and Bcl-2 effects on ADP/ATP antiport. Bax inhibited and Bcl-2 stimulated the ADP-induced ATP release (Fig. 3,C). In addition, König’s polyanion and NADH, two VDAC channel blockers, failed to interfere with ADP/ATP exchange in mitoplasts, whereas Bax and Bcl-2 continued to affect the ADP/ATP antiport. This suggests that VDAC does not mediate the Bax and Bcl-2 effects on ANT (Fig. 3 C). Altogether, these data confirm the results obtained with ANT proteoliposomes in a more physiological setting and underscore that the effect of Bcl-2 and Bax are not mediated via VDAC (which is irrelevant for ADP/ATP exchange in mitoplasts).
Bcl-2 Enhances the ANT-dependent ADP/ATP Exchange in Intact Cells.
To investigate mitochondrial ADP/ATP translocation in intact cells, we engineered HeLa cells to express an ATP sensor in the mitochondrial matrix. This was achieved by transfecting cells with a luciferase cDNA construct that directs luciferase expression to the mitochondrial matrix (mtLuc) because of fusion with the NH2-terminal mitochondrial import sequence of cytochrome c oxidase subunit VIII (25, 38). In the presence of luciferin, the luciferase emits a luminescence signal, the amplitude of which amplitude depends on the local ATP concentration. When HeLa control cells (Neo) were permeabilized with digitonin (which does not affect the inner mitochondrial membrane) and the mitochondrial F1F0-ATPase was inhibited by oligomycin, addition of exogenous ATP plus ADP led to a sudden increase in the luminescence signal, followed by its decrease, the slope of which reflects ADP/ATP exchange on mitochondrial membrane (Fig. 4,A). HeLa cells overexpressing Bcl-2 reproducibly exhibited an accelerated matrix ATP decline, indicative of an increased ATP efflux from mitochondria (Fig. 4,A). Perfusion of oligomycin during the experiment reduced the contribution of the F1F0-ATPase and allowed to reveal a specific effect of Bcl-2 effect on ANT translocase activity (Fig. 4, A versus C). This difference between NEO and Bcl-2-overexpressing cells persisted in the presence of CsA (Fig. 4,B), excluding the contribution of CsA-inhibitable PTPC pore opening. However, the difference disappeared in the presence of the ADP/ATP translocation inhibitor Atr, underlining a new functional interaction between Bcl-2 and ANT (Fig. 4,E), not mediated by a pore opening effect of Atr (Fig. 4,F). When Neo cells were treated with STS to induce apoptosis, the ADP/ATP exchange was reduced after as little as 1 h of incubation (Fig. 5,A), which is well before the loss of the mitochondrial transmembrane potential (Fig. 5,B). In contrast, STS failed to cause a reduction in mitochondrial ADP/ATP exchange in Bcl-2-overexpressing cells (Fig. 5 A). Altogether, these data confirm that Bcl-2 can modulate the ADP/ATP translocase activity in vivo during chemotherapy-induced apoptosis.
Apoptosis induced by growth factor withdrawal reportedly reduces ADP/ATP exchange on mitochondrial membranes. This effect has been attributed to cytosolic alkalinization (39), down-regulation of mitochondrion-associated hexokinase II, and other glycolytic enzymes (40, 41, 42), as well as a closure of the VDAC (42), perhaps as a result of a hyperpolarization of the inner mitochondrial membrane with charge transfer to the outer membrane (9, 43). Bcl-2 would maintain the flux of metabolites on mitochondrial membranes, either through its effect to maintain a normal proton flow on the inner mitochondrial membrane (44) or through a functional interaction with VDAC preventing it from closing and from shutting down metabolite transport on the outer membrane (42).
The data contained in this paper unravel a long-sought-for link between metabolism and apoptosis. As shown here, Bax can interact with ANT in apoptotic conditions (Fig. 1), secondary to its translocation from the cytosol to mitochondria (i.e., stimulated by cytosolic alkalinization and inhibited by hexokinase II) (40, 45). Within mitochondria, Bax can then reduce ADP/ATP exchange (Fig. 3), presumably through a direct effect on ANT (Fig. 2). Thus, Bax can inhibit ANT translocase activity without inducing pore formation (Fig. 3,C), and pore formation by the ANT/Bax complex requires an additional signal such as Atr (Fig. 2,D). Accordingly, early apoptosis is accompanied by an inhibition of ANT activity, which precedes the loss of the Δψm (Fig. 5). On theoretical grounds, the Bax-mediated inhibition of ADP/ATP flux may be expected to cause a hyperpolarization of the Δψm, given that translocase activity (which exchanges ATP4− against ADP3−) normally dissipates part of the electrochemical gradient. This should cause a Δψm hyperpolarization and matrix alkalinization, two phenomena that are frequently associated with early apoptosis (43, 46). It appears plausible that when matrix alkalinization and local alterations in ADP/ATP concentrations persist, nonspecific pore formation by the ANT/Bax complex occurs, resulting in permanent Δψm loss. The data reported here, namely ANT-Bax coimmunoprecipitation (Fig. 1,B), ANT-Bax pore formation (Fig. 2,D), Bax inhibition of ADP/ATP exchange (Figs. 2,G and 3,A), as well as previous studies (43, 46, 47), are compatible with this hypothesis. Moreover, Bcl-2-like proteins can enhance the ANT activity, both in vitro (Figs. 2 and 3) and in vivo (Figs. 4 and 5), an observation that can be correlated with Bcl-2’s capacity to prevent the apoptosis-associated transient Δψm increase (43) and matrix alkalinization (46). Bcl-2 has previously been shown to maintain the ATP/ADP exchange on mitochondrial membranes, and this effect has been explained by an action on the outer membrane, presumably on VDAC (47). Although our data certainly do not exclude a functional interaction between Bcl-2 and VDAC, they suggest that part of the Bcl-2 effect can be attributed to an interaction with ANT. Thus, Bcl-2 enhances the translocase activity of purified (VDAC-free) ANT in proteoliposomes (Fig. 2), and it does stimulate ADP/ATP exchange on mitochondria on which the outer membrane has been permeabilized (Fig. 3). Moreover, our scenario of Bax and Bcl-2-mediated metabolic and apoptotic regulation is compatible with the previously reported finding that both proteins can function independently of each other (48, 49, 50).
In summary, the reported effects of Bax and Bcl-2 on ANT activity are of heuristic value for the comprehension of apoptosis regulation. Future studies will have to elucidate the complex interplay between apoptosis-relevant components of the PTPC (e.g., ANT, VDAC, Bcl-2, Bax, hexokinase II …). Such studies appear particularly important in view of the fact that intra-PTPC protein-protein interactions strongly change during the switch from normal physiology to incipient cellular demise.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by grants from ANRS, ARC, FRM, EC (QLG1-CT-1999-00739), LNC, and Ministry of Science.
The abbreviations used are: MMP, mitochondrial membrane permeabilization; ANT, adenine nucleotide translocator; Atr, atractyloside; CsA, cyclosporin A; Δψm, mitochondrial transmembrane potential; mAb, monoclonal antibody; KPA, König’s polyanion; 4-MUP; 4-methylumbelliferylphosphate; PTPC, permeability transition pore complex; STS, staurosporine, VDAC, voltage-dependent anion channel.
We thank Professor G. Rigoulet, Dr. V. Trézéguet, Professor Lauquin (Bordeaux, France), Dr. J. C. Reed (The Burnham Institute, La Jolla, CA), and Dr. V. Goldmacher (ImmunoGen, Cambridge, MA).