Mitochondrial control of apoptosis

Mitochondrial Control of Nuclear Apoptosis By Naoufal Zamzami,* Santos A. Susin,* Philippe Marchetti,* Tamara Hirsch,* Isabel G6mez-Monterrey,r Maria Castedo,* and Guido Kroemer* From * Centre National de la Recherche Scientifique (CNRS)-UPR420, F-94801 Villejuif, France; and r Institute for Medical Chemistry, Consejo Superior de Investigadtnes Cientificas, 28008 Madrid, Spain Summary Anucleate cells can be induced to undergo programmed cell death (PCD), indicating the existence of a cytoplasmic PCD pathway that functions independently from the nucleus. Cytoplasmic structures including mitochondria have been shown to participate in the control of apoptotic nuclear disintegration. Before cells exhibit common signs of nuclear apoptosis (chromatin condensation and endonuclease-mediated DNA fragmentation), they undergo a reduction of the mitochondrial transmembrane potential (A~m) that may be due to the opening of mitochondrial permeability transition (PT) pores. Here, we present direct evidence indicating that mitochondrial PT constitutes a critical early event of the apoptotic process. In a cell-free system combining purified mitochondria and nuclei, mitochondria undergoing PT suffice to induce chromatin condensation and DNA fragmentation. Induction of PT by pharmacological agents augments the apoptosis-inducing potential of mitochondria. In contrast, prevention of PT by pharmacological agents impedes nuclear apoptosis, both in vitro and in vivo. Mitochondria from hepatocytes or lymphoid cells undergoing apoptosis, but not those from normal cells, induce the disintegration of isolated Hela nuclei. A specific ligand of the mitochondrial adenine nucleotide translocator (ANT), bongkrekic acid, inhibits PT and reduces apoptosis induction by mitochondria in a cell-free system. Moreover, it inhibits the induction of apoptosis in intact cells. Several pieces of evidence suggest that the proto-oncogene product Bcl-2 inhibits apoptosis by preventing mitochondrial PT. First, to inhibit nuclear apoptosis, Bcl-2 must be localized in mitochondrial but not in nuclear membranes. Second, transfection-enforced hyperexpression of Bcl-2 directly abolishes the induction of mitochondrial PT in response to a protonophore, a pro-oxidant, as well as to the A N T ligand atractyloside, correlating with its apoptosis-inhibitory effect. In conclusion, mitochondrial PT appears to be a critical step of the apoptotic cascade. ince it has been shown that anucleate cells (cytoblasts) can be induced to undergo programmed cell death (PCD) 1 (1-3), it has become clear that a cytoplasmic PCD pathway must function independently from the nucleus. Both mitochondria (4) and specific ced-3-1ike proteases S 1Abbreviations used in this paper: ANT, adenine nucleotide translocator; Atr, atractyloside;BA, bongkrekicacid; CsA, cyclosporinA; DAPI, 4'-6diamidino-2-phenylindole dihydrochloride;DEX, dexamethasone;diamide, diazenedicarboxylicacid his 5N,N-dimethylamide; DiOC6(3), 3,3'dihexyloxacarbocyanineiodide; Aq~m,mitochondrialtransmembrane potential; GAIN,D-galactosamine;ICE, IL-113convertingenzyme;MCB, monochlorobimane; mCICCP, carbonylcyanidem-chlorophenylhydrazone; mtDNA, mitochondrialDNA; PCD, programmedcell death; PT, permeability transition; P,OS, reactive oxygen species; R.tL, ruthenium red; ter-BHP,ter-butylhydroperoxide. N. Zamzamiand S.A. Susincontributedequallyto this paper. This paper is dedicatedto Jos6 Uriel. 1533 (5-7) have been accused of participating in the cytoplasmic control of apoptotic nuclear disintegration. We and others (8-12) have recently demonstrated that cells undergo a reduction of the mitochondrial transmembrane potential (A~m) before they exhibit common signs of nuclear apoptosis (chromatin condensation and endonuclease-mediated DNA fragmentation). This applies to different cell types (neurons, fibroblasts, B and T lymphocytes, pre-B cells and thymocytes, myelomonocytic cells) and to different physiological apoptosis inducers (growth factor withdrawal, tumor necrosis factor, ceramide, glucocorticoids, activation-induced cell death, positive and negative selection, irradiation; 8-13). Moreover, these observations extend to pathogen-induced apoptosis, including irradiation-induced PCD (13) and HIV-l-triggered T lymphocyte PCD (14). When PCD is prevented either by genetic manipulations (e.g., p53 loss mutation, bcl-2 hyperexpression) or by pharmacological j. Exp. Med. 9The Rockefeller University Press 90022-1007/96/04/1533/12 $2.00 Volume 183 April 1996 1533-1544 agents (N-acetylcysteine, protease inhibitors, linomide), both mitochondrial and nuclear signs o f apoptosis are abolished (10, 12, 13). Moreover, cells that have lost their Axlr m appear to be irreversibly programmed to die (10). Although these observations suggest the involvement o f mitochondria in apoptosis, they do not clarify the cause-effect relationship between mitochondrial dysfunction and subsequent nuclear apoptosis. It appears clear that reactive oxygen species (ROS), which may be generated by uncoupled mitochondria (9, 12), are not essential for the apoptotic process (15-17). Thus, whenever a cause-effect relationship between mitochondrial disorders and nuclear apoptosis exists, it must be mediated by factors other than R O S . The aim o f this paper was to unravel the existence of such a pathway linking mitochondrial dysfunction to nuclear disintegration. As to the mechanism ofapoptotic Axltm disruption, pharmacological experiments suggest that it involves the opening o f so-called mitochondrial permeability transition (PT) pores (12, 18). Under normal conditions, the inner mitochondrial membrane is quasi-impermeable for small molecules, thus allowing for the creation o f the electrochenfical gradient which is indispensable for mitochondrial function. However, in determined circumstances, opening o f P T pores or "megachannels" allows for the free distribution o f solutes o f < 1 , 5 0 0 daltons and o f some proteins, thereby disrupting the A~tI'tm and associated mitochondrial functions (19, 20). In isolated mitochondria, P T is accompanied by colloidosmotic swelling and uncoupling o f oxidative phosphorylation, as well as by the loss o f low molecular weight matrix molecules such as calcium and glutathione (19-21). It may be important to note that P T is modulated by multiple different physiological and pharmacological inducers and inhibitors (for a review see reference 22) and that P T is both the cause and the consequence o f AW m dissipation, as well as o f reactive oxygen metabolite production (19-29). In other terms, P T results ipso facto in A ~ m disruption and later in R O S hyperproduction, but AXlt m reduction and R O S themselves can also provoke PT, as do many other factors (divalent cations, pH variations, peptides, etc; 22). The exact molecular composition o f the P T pore is not known. However, it appears that at least one inner mitochondrial transmembrane protein, namely the adenine nucleotide translocator (ANT), is involved in P T pore formation (for reviews see references 19, 20) and that A N T associates with several molecules o f the outer mitochondrial membrane such as the peripheral benzodiazepine receptor and the voltage-dependent anion channel (30). A N T ligands such as atractyloside (Atr) and bongkrekic acid (BA) enhance or reduce the probability o f PT, respectively (31-35). Based on these premises, we have tested the hypothesis that PT might be the critical event determining the apoptosis-inducing potential of mitochondria. Using a cell- and cytosol-free system in which purified mitochondria and nuclei are confronted, we show that induction of PT by the A N T ligand Atr or other less specific P T inducers causes isolated mitochondria to trigger nuclear apoptosis. In contrast, inhibition o f PT by the Atr antagonist BA, as well 1534 as by a variety o f additional P T inhibitors, abolishes mitochondria-mediated nuclear apoptosis. The apoptosis-inhibitory proto-oncogene product Bcl-2 functions as an endogenous inhibitor o f mitochondrial PT. These data establish mitochondrial P T as a critical event o f apoptosis. Materials and M e t h o d s Animals and In Vivo Treatments. Male 6--10-wk-old BALB/c mice were injected simultaneously with D-galactosamine (GAIN; 10 mg i.p.) and/or LPS from Escherichiacoli (Signa Chemical Co., St. Louis, MO; 50 btg i.v.), 5 h before removal of the liver (36). Alternatively, splenocytes were recovered from BALB/c mice 12 h after injection of 1 mg i.p. dexamethasone (DEX; Sigma Chemical Co.) in 200 btl PBS or PBS alone (10, 37). Cell Lines and In Vitro Culture Conditions. U937 cells were depleted from mitochondrial DNA (mtDNA) by continuous ethidium bromide selection for 4 mo (15). Control experiments revealed that such cells become resistant to antimycin A, which blocks the mtDNA-encoded complex III. Moreover, no mtDNA could be detected by PCP,. (not shown). 2B4.11 T cell hybridoma cell lines stably transfected with an SFFV.neo vector containing the human bcl-2 gene or the neomycin (Neo) resistance gene only (38, 39) were kindly provided by Jonathan Ashwell (National Institutes of Health, Bethesda, MD). Cells were cultured in RPMI-1640 medium containing 5% FCS. Apoptosis was induced by culturing cells in the presence of the indicated concentration ofdiazenedicarboxylic acid bis 5N, N-dimethylamide (diamide) or carbonyl cyanide m-chlorophenylhydrazone (mC1CCP; both from Sigma Chemical Co.). DNA fragmentation of non- or ",/-irradiated (10 Gy) thymocytes (106 cells/lane) was monitored after culturing cells for 4 h in the presence of DEX (1 I~M), etoposide (10 ~M; Sigma Chemical Co.), and/or BA (50 btM; purified as described in reference 40), kindly provided by Dr. J.A. Duine (Delft University, Delft, The Netherlands). Cell-free System of Apoptosis. Nuclei from HeLa or 2B4.11 cells were purified on a sucrose gradient, as described (41), and were resuspended in CFS buffer (220 nM mannitol, 68 mM sucrose, 2 mM NaC1, 2.5 mM PO4H2K, 0.5 mM EGTA, 2 mM C12Mg, 5 mM pyruvate, 0.1 mM PMSF, 2 mM ATP, 50 Ixg/nfl creatine phosphokinase, 10 mM phosphocreatine, 1 mM dithiothreitol, and 10 mM Hepes-NaOH, pH 7.4; reagents from Sigma Chemical Co.). Nuclei were conserved at -20~ in 50% glycerol for up to 8 d as described (41, 42). Mitochondria were purified from BALB/c mouse livers, splenocytes, or U937 cells on a Percoll gradient (43) and were stored on ice in B buffer (400 mM mannitol, 10 mM PO4H2K, 5 mg/ml BSA, and 50 mM TtisHC1, pH 7.2) for up to 4 h. For quantitation of nuclear apoptosis, both nuclei (5,000g, 5 rain) and mitochondria (2 • 104g, 3 min) were spun down and washed twice in CFS buffer before being mixed. In standard conditions, mitochondria (500 ng/~l protein final concentration) were cultured at 37~ for 90 min with 103 nuclei per ~1 CFS containing a number of different agents: AIF3, (20 p,M), Atr (5 raM; Sigma Chemical Co.), BA (50 ~M), CaC12 (500 I.zM), mC1CCP (10 p~M), diamide (100 p,M), cyclosporin A (CsA, 10 /a,M; Sandoz AG, Basel, Switzerland), N-methylVal-4CsA (SDZ 220-384, 10 btM; kindly provided by Dr. Roland Wenger, Sandoz), monochlorobimane (MCB; 30 p~M), phosphotyrosine (P; 10 raM), ruthenium red (RR, 100 ~M; Sigma Chemical Co.), ter-butylhydroperoxide (ter-BHP, 50 btM; Sigma Chelnical Co.), ZnC12 (1 raM), AcYVAD-CHO (IL-1]3 converting enzyme [ICE] inhibitor I), AcYVAD-chloromethylketone (ICE inhibitor II), and/or AcDEVD-CHO (inhibitor of CPP32/ Mitochondrial Regulation of Apoptosis Ced3/Yama; Bachem, Basel, Switzerland). Nuclei were stained with 4'-6-diamidino-2-phenylindoledihydrochloride (DAPI; 10 ~M) and examined by fluorescence microscopy (5), or were analyzed by agarose gel electrophoresis (106 nuclei/lane) (44). Cytofluorometric Analysis. For AxI/mdeterminations, isolated mitochondria were incubated for 15 rain at 37~ in the presence of DiOC6(3 ) (80 riM) (45), followed by addition of mC1CCP (50 p~M), BA (50 I~M) and/or Atr (5 raM), and recording of the fluorescence in an Elite cytofluorometer (Coulter Corp., Hialeah, FL) 5 rain later. Loss of nuclear DNA (hypoploidy) was determined by propidium iodine staining of ethanol-fixed cells, as described (46). Large Amplitude Swelling of Isolated Mitochondria. Large amplitude swelling is a colloidosmotic process that is observed among isolated mitochondria undergoing PT in solutions containing low protein concentrations (22). For determination of swelling, mitochondria were washed and resuspended in B buffer (100 ~g protein/10 ~1 buffer), followed by addition of 90 p~M CFS buffer and recording of adsorption at 540 nm in a spectrophotometer (model DU 7400; Beckman Instruments, Inc., Fullerton, CA), as described (26). The loss of absorption induced by 5 mM Atr within 5 rain was considered 100% the value of large amplitude swelling. Characterization of Factors Contained in the Supernatant of Mitochondria. Hepatic mitochondria (1 mg/ml in CFS buffer) were left untreated or were incubated with Atr (5 raM) for 10 rain at room temperature, followed by ultracentrifugation (1.5 • 10s g, 30 rain, 4~ Supernatants were either left untreated or centrifuged through a Centricon 10 membrane (Amicon Inc., Beverly, MA) to separate proteins with an approximate molecular mass of > and <10 kD, following the manufacturers' recommendation. The fraction >10 kD was reconstituted with CFS to the original volume. Supematants (50% volume) were mixed with nuclei (103/~1) in the presence or absence of various antioxidants (50 p~M N-t-butyl-ot-phenylnitrone,230 ~M trolox, 600 IxM L-ascorbate, or 1 mg/ml catalase; Sigma Chemical Co.), and nuclei were stained with DAPI after 90 min of culture at 37~ Results and D i s c u s s i o n Isolated Mitochondria Undergoing P T Induce Nuclear Apoptosis in a Cell-free System. O n e o f the P T pore constituents is the A N T . T h e A N T adopts different molecular conformations w h e n exposed to two specific ligands, Atr and BA (31, 47). Atr favors the opening of the P T pore, whereas BA reduces the probability of P T pore gating (31-35, 47). W e have tested the capacity o f purified mitochondria with open and closed P T pores to induce nuclear apoptosis in a modified cell-free system (4, 42). Mitochondria were purified from the murine liver and were mixed with Hela n u clei in an isotonic buffer containing an A T P regeneration system. O n their own, unmanipulated mitochondria derived from healthy cells are incapable of inducing signs of nuclear apoptosis such as chromatin condensation and endonuclease-mediated D N A fragmentation. However, mitochondria exposed to a dose of Atr that causes PT, determ i n e d either as large amplitude swelling (Fig. 1 A) or as disruption of the AXI~m (Fig 1 B), do induce nuclear apoptosis (Fig. 1 C) in a time- and dose-dependent fashion (Fig. 1 D). Atr does not induce apoptosis itsel~ it only favors the induction o f nuclear apoptosis w h e n mitochondria are present (Fig. 1 C). In the presence o f B A , mitochondria fail to undergo P T in response to Atr (Fig. 1, A and B) and lose the capacity to induce chromatin condensation (Fig. 1 C) and associated oligonucleosomal D N A fragmentation (Fig. Figure !. Regulation of mitochondrial PT and mitochondriamediated nuclear apoptosis by two adenine nucleotide translocator ligands. (A) Effect of atractyloside (Atr) and bongkrekic acid (BA) on mitochondriallarge amplitude swelling. (Arrows) Time points at which BA (50 p~M) and/or Atr (5 mM) were added to mitochondria. (B) A~,,, of Atr- and BA-treated mitochondria, as determined by incorporation of DiOC6(3) into mitochondria, 5 rain after addition of Atr and/or BA. Incubation with the protonophor mC1CCP, which completelydisrupts the A~ m, unravels DiOC6(3) background fluorescence. (C) Chromatin distribution of isolated HeLa nuclei cultured with mitochondria, Atr, and/or BA for 90 rain. Single nuclei stained with DAPI representing the dominant phenotype (I>80%) are shown. (D) Time and dose dependence of nuclear chromatin condensation.Nuclei were incubated during the indicated interval with the specified amount of mitochondria (doses in I~g protein) and/or Atr. (E) Fragmentationof nuclear DNA induced by mitochondria. Purified HeLa nuclei were coculturedwith mitochondria (500 ng/~tl), Atr, and/or BA for 90 min (same conditionsas in C), followedby agarose gel electrophoresis of ethidium bromide-stained DNA. Lane I: Atr only; lane 2: mitochondriaonly; lane 3: mitochondriaplus Atr; and lane 4: mitochondriaplus Atr Flus BA. 1535 Zamzami et al. 1 E). These findings suggest that PT controls the apoptosisinducing capacity ofmitochondria. Mitochondria Lacking mtDNA Can Undergo PT and Cause Nuclear Apoptosis in a Cell-free System. Although most proteins contained in mitochondria are encoded by nuclear genes, a number of proteins including some components of the respiratory chain complexes I, III, and IV, are encoded by the mitochondrial genome. Previously, it has been reported that cell lines lacking mtDNA can undergo fullblown nuclear apoptosis (15), and this finding could give rise to the interpretation that mitochondria are not important for the control ofapoptosis. To challenge this (over)interpretation, we purified mitochondria from cells lacking mtDNA (po cells), as well as from control p+ cells, and tested their apoptosis-inducing potential. As shown in Fig. 2 A, pO mitochondria can undergo large amplitude swelling in response to Atr, exactly as do p+ control organelles. Moreover, in the presence of Atr, pO mitochondria are as efficient inducers of nuclear disintegration, as are control p+ mitochondria (Fig. 2 B). This indicates that all mitochondrial functions critical for apoptosis induction are encoded by nuclear genes. Thus, in accord with the published data (15), the capacity of mitochondria to induce nuclear apoptosis does not depend on the presence of mtDNA. Furthermore, the fact that respiration-deficient po mitochondria (which may be expected to produce less ROS than p+ control organelles) conserve their proapoptotic activity, suggests that R O S do not mediate apoptosis in this cell-free system. Strict Correlation Between PT-associated Swelling and Nuclear Apoptosis Induction. The above results suggest that mitochondria are indeed efficient inducers of nuclear apoptosis, provided that they are undergoing PT. This conclusion is corroborated by the strict correlation between PT and the proapoptotic effect of mitochondria, when PT is induced by a variety of different molecules: Atr, the pro-oxidant terBHP, and calcium ions via the thiol-cross-linking agent diamide or the protonophore mC1CCP (Fig.3). All these re- A B .qA..I 100" iI~II0" pO ~l~ 20" II t li l i time (sec) ii# z2o II @d _Illl. LIts-iL .IL# LAtrl #+ po Figure 2. Mitochondrialswelling and nuclear apoptosis induction do not require mitochondrialDNA. (A) Large amplitude swelling of mitochondria lacking mtDNA. Mitochondria obtained from normal U937 cells (p§ or U937 cellslackingmtDNA (pC)were incubatedwhile assessing absorbanceat 540 nM. (Arrow)Addition of Atr (5 mM). (B) Induction of nuclear apoptosisby mitochondrialacking mtDNA. HeLa nuclei were cultured for 90 rain with Atr and/or mitochondriaobtained from p+ or poU937 cells,followedby determinationof the frequencyofapoptoticor damaged nuclei. 1536 agents are thought to act via distinct mechanisms: Atr by virtue of its capacity to interact with the ANT (31-35); calcium via conformational effects on proteins that are yet poorly understood (19, 20, 22); hydroperoxides via oxidation of mitochondral glutathione and pyridine nucleotides (48); diamide via its thiol-cross-linking action on the ANT(49, 50); and protonophores via dissipation of the proton gradient (A~m), then entailing PT as a secondary phenomenon (24, 27, 28). Control experiments indicate that none of these reagents induces nuclear apoptosis by itself, i.e., in the absence ofmitochondria (not shown). Atrinduced swelling and nuclear apoptosis are efficiently inhibited by BA and CsA, as well as by the thiol reagent MCB. CsA can be substituted for by the nonimmunosuppressive CsA analogue N-methylVal-4-CsA (51), indicating that its PT and apoptosis-inhibitory effect is not mediated via calcineurin. A series of substances previously reported to inhibit apoptosis in a cell-free system (4) can also inhibit both PT and mitochondria-mediated nuclear apoptosis: phosphotyrosine, ZnC12, and A1F 3 (Fig. 3). Other substances (4) have no or little (<20%) inhibitory effects on Atr-induced swelling and apoptosis: calpain inhibitors I and 1I, GTPTS, and ionomycin. Similarly, synthetic tetrapeptide inhibitors of ICE and of CPP32/Yama fail to interfere with PT and PT-dependent nuclear apoptosis (Table 1). Calcium-driven but not Atr-induced PT and apoptosis are selectively inhibited by R R , a specific inhibitor of the mitochondrial calcium uniport (52), underscoring the fact that proapoptotic calcium effects are indeed mediated by mitochondria. As expected (50), the syn-9,10-dioxa-bimanine halogen derivative MCB is particularly effective in inhibiting PT and apoptosis induced by diamide (Fig. 3). In accord with published data (for a review see reference 22), none of the inhibitors used in this study is capable of providing long-term (>30 rain) protection against mitochondrial swelling in response to the whole panel of PT inducers. This probably reflects the profound differences in the molecular mechanisms of PT caused by different inducers (19, 20, 22, 24, 27, 28, 48-50). In synthesis, the strict correlation existing between mitochondrial PT and mitochondria-mediated nuclear apoptosis suggests that PT is indeed a crucial event in the regulation ofapoptosis induction by mitochondria. Moreover, the fact that none of the inhibitory substances (BA, CsA, MCB, R R , phosphotyrosine, ZuC12, A1F3) suppresses PT and apoptosis in response to all PT inducers (Fig. 3) suggests that they do not directly affect nuclei but rather act via PT modulation. Finally, the data summarized in Fig. 3 underscore the complex pharmacology ofmitochondrial PT. Mitochondriafiom Cells UndergoingApoptosis Transfer Nuclear Apoptosis to a Cell-freeSystem. According to several studies (8-10, 12), mitochondrial function is perturbed early during the apoptotic process. Accordingly, mitochondria isolated from hepatocytes exposed in vivo to an apoptosisinducing combination of GaIN and LPS (36, 53), but not control cells treated with GaIN or LPS only, display a reduced uptake of the cationic lipophilic dye 3,3' dihexyloxacarbocyanine iodide (DiOC6[3]) (Fig. 4 A), indicating a Mitochondrial Regulation of Apoptosis mitochondrial sO BA CsASDZMCBRR P ZnCI2AIF3t sO BA CsASDZMCBRR P ZnCI2AIF3s 0 Atractyloside sO BA CsASDZMCBRR P ZnCI2AIF3s ter-BHP ~, .O BA CO, SDZMCBRR P ZnCUAIF3 .0 BA CsASDZMCBRR P ZnCL~AIF3 . ~ Ca ~ BA CO, SDZMCBRR P ZnClaAIF~ mCICCP diamide Figure 3. Correlation between mitochondrial swelling and mitochondrial induction of nuclear apoptosis. Mitochondria from hepatocytes were incubated with the PT inducers Atr (5 re_M), ter-BHP (50 FtM), CaC12 (500 p,M), mCICCP (10 ~M), or diamide (100 ~M) and/or the PT inhibitors BA (50 I~M), CsA (10p~M), N-methylVal-4-CsA (SDZ; 10 I~M), MCB (30 ~M), or R R (100 ~M). Mitochondria were also incubated with P (10 mM), ZnC12 (lmM), or AIF3 (20 ~M). The percentage of condensed nuclei cocultured with mitochondria was recorded after 90 min of incubation at 37~ (open columns, X + SEM of triplicates). Large amplitude swelling was recorded after 5-90 min of culture (blackcolumns). Data are shown for 60 rain, when the correlation between nuclear apoptosis and mitochondrial swelling is optimal. These results are representative of five independent experiments. decrease o f the A~klfm. Such m i t o c h o n d r i a f r o m apoptotic liver cells cause n u c l e a r apoptosis i n vitro (Fig. 4 B). S i m i larly a fraction o f m i t o c h o n d r i a from splenocytes treated i n vivo w i t h the glucocorticoid analogue D E X (10, 37) display a r e d u c e d A ~ m (Fig. 5 A) a n d cause apoptosis o f isolated Hela nuclei in vitro (Fig. 5 B). Thus, m i t o c h o n d r i a from different cell types u n d e r g o i n g apoptosis in vivo are T a b l e 1. e n d o w e d w i t h the capacity o f apoptosis i n d u c t i o n in a ceUfree system. Inhibition of P T Inhibits Nuclear Apoptosis both In Vitro and In Vivo. T h e A N T ligand B A is the agent w i t h the broadest P T - i n h i b i t o r y s p e c t r u m a m o n g all substances tested thus far (Figs. 1 and 3) and is the o n l y P T i n h i b i t o r that is truly specific for a m i t o c h o n d r i a l structure. W e Substances that Fail to Modulate P T and Mitochondria-dependent Nuclear Apoptosis Substance Dose range Calpain inhibitor I Ca]pain inhibitor II ICE inhibitor I ICE inhibitor II AcDEVD-CHO GTP3~S Ionomycin 100 100 100 100 100 100 10 heM-1 m M IxM-1 m M heM-500 p.M p~M-500 p.M p~M-500 p~M p~M-i m M ~ M - 1 0 0 I-~M Inhibition of PT Inhibition of nuclear apoptosis None* None None None None None None None None None None None None None *The indicated substances were employed to modulate the induction of PT in isolated mitochondria induced by the following reagents: Atr, terBHP, calcium, m-C1CCP, and diamide (same conditions and concentrations as in Fig. 3). Absence of inhibition indicates <20% suppression of either mitochondrial swelling (measured at 60 min as in Fig. 3) or nuclear condensation (measured at 90 rain in the same conditions as in Fig. 3), in response to all tested PT inducers. 1537 Zamzami et al. Control 9= GaIN Figure 6. BA inhibits apoptosis of thymocytes induced by various stimuli. Murine thymocytes were exposed to DEX (1 I,M), etoposide (10 IzM), or cultured after y-irradiation (t0 Gy) in the presence or absence of BA (50 }*M). DNA fragmentation of thymocytes (106 cells/ lane) was monitored after a 4-h culture period. Results are representative of three independent experiments. LPS LPS/GalN , iogDiOC6(3) ~ ~ ~ ,~ Figure 4. Mitochondria from apoptotic hepatocytescause nuclear apoptosis in a cell-free system. (A) Reduction of the A~ m in mitochondria from cellsundergoing apoptosisin vivo. Liver mitochondria were obtained from animals treated with an apoptosis-inducing combination of GaIN and/or LPS, followed by DiOCt(3) labeling for A~tIf m a s s e s s m e n t . (B) Mitochondria from apoptotic hepatocytes induce nuclear apoptosis. After the indicated in vivo treatment, liver mitochondria were purified and added to HeLa nuclei in the presence or absence of BA (50 ~M), followed by evaluation ofchromatin condensation as in Fig. 1. Data are representative of two independent experiments. therefore tested the effect o f BA on the m i t o c h o n d r i a mediated transfer o f apoptosis from w h o l e cells undergoing P C D to the cell-free system. As shown in Figs. 4 A and 5 A, B A partially reduces the proapoptotic effect o f m i t o chondria from G a l N / L P S - s t i m u l a t e d hepatocytes o f D E X p r i m e d splenocytes in vitro. This inhibition is sigxtificant: 48 + 10% for G a l N / L P S - t r e a t e d hepatocyte mitochondria and 44 + 9% for D E X - p r i m e d splenocyte mitochondria. In addition, BA is highly efficient ( > 9 0 % inhibition) in preventing the death o f intact thymocytes exposed to a series o f different apoptosis inducers: D E X , irradiation, and topoisomerase inhibition (Fig. 6). These results corroborate the notion that mitochondria are indeed involved in the |' I log DIOC6(3) Figure 5. Transferof apoptosis by mitochondria from apoptotic splenocytes. (A) A~ m disruption of mitochondria from DEX-primed splenocytes. Mitochondria were purified from control splenocytes (co.)or from splenocytes exposed to the glucocorticoid DEX in vivo, followed by staining with DiOCt(3). (B) Mitochondria from apoptotic splenic lymphocytes induce nuclear apoptosis. Mitochondria from control or DEXprimed splenocyteswere mixed with HeLa nuclei either in the absence or in the presence of BA to assess their apoptosis-inducing potential as in Fig. 4 B. Splenic mitochondria were tested at a concentration of 250 ng/ I*1.Typical results out of three independent experiments are shown. 1538 apoptotic cascade in vivo and that mitochondrial P T is both sufficient and necessary to induce nuclear apoptosis. Bcl-2 Inhibits Apoptosis by Preventing Mitochondrial PT. T h e p r o t o - o n c o g e n e product Bcl-2 inhibits apoptosis in response to a n u m b e r o f different stimuli (for a review see reference 54) and prevents b o t h the mitochondrial and the nuclear manifestations o f apoptosis (12). Bcl-2 is localized in the mitochondrial outer membrane and endoplasmatic reticulum, as well as in nuclear membranes (55-57). W i t h i n the mitochondrion, it is found at the i n n e r - o u t e r m e m brane contact site, where P T pores are expected to form (20). T o map the antiapoptotic function o f Bcl-2 either to mitochondria or to nuclei, w e purified these organelles from hBcl-2-transfected murine T cell hybridoma cells (39), as well as from mock-transfected controls. R.econstitution experiments indicate that Atr-treated mitochondria 8 o m hBcl-2-transfected cells fail to provoke nuclear apoptosis (Fig. 7 A) in conditions in which mitochondria from vector-transfected cells (Fig. 7 A) or from hepatocytes constitutively lacking Bcl-2 expression (Figs. 1 and 3) do induce nuclear apoptosis. In contrast, nuclei from hBcl-2-transfected cells readily condense and fragment in the presence o f Atr and control mitochondria (Fig. 7, A and B). Thus, in accord with previous genetic (55, 57) and functional (4) studies, the mitochondrial but not the nuclear localization o f Bcl-2 is critical for its antiapoptotic function. In control experiments, mixtures o f Bcl-2-transfected and control m i tochondria induce apoptosis (Fig. 7 A), indicating that the B c l - 2 - m e d i a t e d inhibition o f apoptosis acts in cis and cannot be attributed to cytosolic Bcl-2 contaminating the m i tochondrial preparation. In addition, isolated mitochondria from Bcl-2-transfected cells are protected against A t r induced PT, i.e., they fail to undergo large amplitude swelling and A ~ m disruption in response to Atr (Fig. 7, C and D). Bcl-2 is a potent inhibitor o f some death pathways, including pro-oxidants (58), but is comparatively inefficient in preventing calcium-induced and antigen r e c e p t o r mediated P C D (38, 39, 59). W e therefore tested whether Bcl-2 w o u l d be a universal inhibitor o f P T or rather, whether it w o u l d have a selective effect. As shown in Fig. 8 A, Bcl-2 prevents large amplitude swelling o f isolated m i tochondria in response to M - C 1 C C P and ter-BHP, but not in response to calcium or diamide. These data underscore that different P T inducers obey different mechanisms; this is also suggested by experiments involving P T inhibitors Mitochondrial Regulation of Apoptosis F i g u r e 7. Mechanism of the antiapoptotic effect of Bcl-2. (A) Functional mapping of the site at which Bcl-2 acts to prevent Atr-induced apoptosis. Nuclei and mitochondria from Bcl-2- or Neo-transfected cells were cocultured in the presence or absence of Atr (5 raM), as indicated by black squares. After 90 rain ofcoculture, nuclei were stained with DAPI and analyzed for apoptotic morphology. (B) Representative nuclei from Bcl-2-transfected cells incubated with the indicated type of mitochondria and/or Atr (same experiment as A). (C) Bcl-2 directly inhibits the Atr-induced large amplitude swelling ofmitochondria. Mitochondria from Neo- or Bcl-2-transfected cells were monitored for large amplitude swelling (as in Fig. 1 A). (Arrows)Repeated addition of 2.5 mM Atr (final concentration 5 raM). (D) Bcl-2 inhibits the Atr-induced disruption of the mitochondrial transmembrane potential. Mitochondria were labeled with DiOC6(3), cultured for 5 min in the presence or absence of 5 mM Atr, and were then analyzed by cytofluorometry. (Fig. 3). T h e pattern o f the bcl-2 effect corresponds most closely to that o f BA, i.e., it inhibits P T i n d u c e d b y Atr (Figs. 3 and 7), m - C 1 C C P , and ter-BHP, b u t n o t calcium or diamide (Figs. 3 and 8). Again, as i n the case o f B A , Bcl2 - m e d i a t e d i n h i b i t i o n o f P T results in the abolition o f the apoptotic potential o f isolated m i t o c h o n d r i a . M o r e i m p o r tantly, the B c l - 2 - d r i v e n i n h i b i t i o n o f m i t o c h o n d r i a l swelli n g (Fig. 8 A) correlates w i t h its apoptosis-inhibitory p o tential in cells. Bcl-2 protects against apoptosis o f T cell h y b r i d o m a cells i n d u c e d b y m - C 1 C C P (Fig. 8 B) and oxidants such as H 2 0 2 (58), yet fails to confer p r o t e c t i o n against diamide (Fig. 8 /3) and C D 3 cross-linking (12, 38, Ali mCICCP _ . ~ u ~ [ ~ . _ ~ 39). Thus, Bcl-2 does n o t p r e v e n t apoptosis w h e n death is i n d u c e d via such agents as diamide (Fig. 8 B) against whose P T - i n d u c i n g potential it does n o t protect (Fig. 8 A). Again, these data are in accord w i t h the hypothesis that Bcl-2 prevents apoptosis b y virtue o f its P T - i n h i b i t o r y potential. A Soluble Factor Released from Mitochondria Undergoing P T Mediates Nuclear Disintegration. As s h o w n above, m i t o c h o n dria u n d e r g o i n g P T i n d u c e apoptotic n u c l e a r disintegration in a cell-free system. W h e r e a s some authors have s h o w n that m i t o c h o n d r i a are necessary to i n d u c e apoptosis in cellfree systems (4, 60), others have f o u n d that cytosolic (organelle-free) extracts m a y be suflficient to i n d u c e n u c l e a r t-BHP o 3o ~ 9o~2otso~so o 3o ~ ~ t 2 o t s m s o o 3o ,o 9o~z.~so~so o 3 o . time ( s e c ) B ~eo ~12.~.~. mCICCP Figure 8. IIBd-2 dlamlde Correlation of the antiapoptotic and the PT-inhibitory effect ofBcl-2. (A) Effect of Bcl-2 on large amplitude swelling of isolated rnitochon- dria. (Arrows)The indicated reagents were added (same concentrations as in Fig. 3), while absorbance at 540 nm was monitored. (B) Spectrum of Bcl-2mediated inhibition of apoptosis in whole cells. Bcl-2, or Neo-transfected T cell hybridoma cells were cultured with the indicated dose ofmC1CCP pr diamide for 6 or 24 h, respectively. The percentage of cells with nuclear hypoploidy was determined after ethanol fixation and staining with propidium iodine. 1539 Zamzami et al. 10oi e~l U't ~Ir ~ 20- lr/A I rill + u I SN from Atr-treated mitochondria Figure 9. Partial characterization of a proapoptotic activity released from Atr-treated mitochondria. Liver mitochondria were incubated in the presence or absence of 5 mM Atr, followed by ultracentrifugation (150,000 g, 30 min). Isolated HeLa nuclei were incubated in the presence of this supernatant to determine the frequency of cells exhibiting apoptotic morphology (90 rain, 37~ Supernatants were heat treated (70~ 5 min) or centrifuged through membranes with a molecular mass exclusion of~10 kD before the test. Alternatively, the antioxidants N-t-butyla-phenylnitrone (50 p.M), trolox (230 I~M), L-ascorbate (600 I~M), or catalase (1 mg/ml) were added to the assay. apoptosis in vitro (5, 42). Prompted by this apparent contradiction, we tested whether mitochondria undergoing PT would release a soluble proapoptotic factor. As shown in Fig. 9, mitochondria treated with Atr release (a) soluble factor(s) into the supernatant (150,000 g, 30 rain) that can induce chromatin condensation in isolated HeLa nuclei. This activity is heat sensitive (70~ 5 rain), has a molecular mass > 1 0 kD, and is not neutralized by antioxidants such as N-t-butyl-cx-phenylnitrone or the water-soluble vitamine E analogue trolox (Fig. 9). In conclusion, at least part of the apoptotic activity of mitochondria is mediated by one or several proteins and does not involve R.OS. P T dependent release of proteins from mitochondria has been reported previously (61). Concluding Remarks As shown in this article, mitochondria from hepatic, m y elomonocytic, or lymphoid cells induce nuclear apoptosis, provided that they undergo PT. Modulation of P T determines the apoptosis-inducing effect of mitochondria in a cell-free system. Moreover, inhibition of P T by BA, a specific ligand of one P T pore constituent, reduces naturally occurring apoptosis, and Bcl-2 apparently functions as an endogenous P T inhibitor. Although these findings establish mitochondrial P T as a critical event in early apoptosis, they do not resolve a number of issues concerning the cellular biology of apoptosis. According to studies performed in Caenorhabditis elegans, at least two gene products, ced-3, which encodes a cysteine protease, and ced-4, whose function is unknown, are required for apoptosis to occur (62). At present, the sequence 1540 of events that eventually link ced-3-like proteases and ced-4 to mitochondria remains unknown. At present, it appears clear that both Bcl-2 (which controls PT; Figs. 7 and 8) and protease activation control two checkpoints of the apoptotic cascade (63). Tetrapeptide inhibitors of the ced-3 homologue C P P 3 2 / Y a m a and of ICE fail to interfere with the induction of P T in isolated mitochondria. Moreover, they fail to inhibit the mitochondria-mediated induction of nuclear apoptosis (Table 1). W h e n thymocyte apoptosis is induced by Fas/CD95 cross-linking, inhibition of ICE prevents both the nuclear manifestations of apoptosis and the A~r~ disruption (Marchetti, P., and G. Kroemer, unpublished results). This may indicate that at least some of the members of the family of ced-3-1ike proteases regulate events that are upstream of mitochondria. At present, h o w ever, our data cannot distinguish between two alternative possibilities. First, the PT and the protease-regulated checkpoints of the apoptotic effector phase could be placed in a serial (hierarchical) fashion. Second, both protease activation and P T could form part of parallel pathways culminating in nuclear apoptosis. It remains largely u n k n o w n h o w Bcl-2 regulates PT on the molecular level. Bcl-2 does not prevent P T as such; it prevents the induction of P T by determined stimuli such as Atr, mC1CCP, and ter-BHP, but not calcium or diamide (Figs. 7 and 8). Bcl-2 could act via direct molecular association with constituents of the P T pore, a possibility that is suggested by the localization of both Bcl-2 and P T pore constituents at inner-outer membrane contact sites (5557). Alternatively, Bcl-2 could affect P T indirectly. Thus, it enhances oxidative phosphorylation (64) and causes mitochondrial inner membrane hyperpolarization (65), which in turn would reduce the probability of P T (24). It has previously been reported that mitochondrial membrane localization is necessary to mediate Bcl-2 suppression o f a p o p t o sis, namely when apoptosis is induced by EIB-defective adenovirus (57) and when it is triggered by IL-3 starvation of IL-3-dependent 32D cells (55). In contrast, in some other systems of apoptosis induction, a mutated Bcl-2 molecule lacking the membrane localization domain (4, 58), as well as the naturally occurring apoptosis-inhibitory Bcl-2 analogue B c l - X A T M (a splice variant of Bcl-X that lacks the transmembrane domain; 66), maintain their antiapoptotic potential. However, the fact that soluble, ubiquitous Bcl-2 still maintains at least part of a its antiapoptotic function does not formally exclude that it acts on the external membrane of mitochondria. T h e present data suggest an intimate linkage between Bcl-2 and mitochondrial regulation. In this context it may be intriguing that the C. etegans bcl-2 homologue, ced-9, is an element of a polycistronic locus that also contains cyt-1, a gene that encodes a protein similar to cytochrome b560 of the mitochondrial respiratory chain complex II (67). Thus both functional and genetic evidence link Bcl-2 to mitochondrial regulation. Irrespective of the exact molecular mechanism by which Bcl-2 affects PT, the finding that Bcl-2 does inhibit PT, at least in response to certain stimuli (Figs. 7 and 8), provides an explanation for hitherto apparently contradictory reports. Mitochondrial Regulation of Apoptosis Bcl-2 hyperexpression has been reported to inhibit the production and/or adverse effects o f 1LOS (58, 68), that in turn, however, are not obligatory for apoptosis (16). In accord with these findings, Bcl-2 prevents oxidant-mediated P T (Fig. 8). Moreover, it prevents the mitochondrial R O S formation that is secondary to P T (12). Thus, Bcl-2 impedes P T as well as two dissociable consequences o f P T : (a) nuclear apoptosis, and (b) mitochondrial uncoupling and superoxide anion generation. A further issue that remains to be elucidated is the m o lecular mechanism by which isolated mitochondria undergoing PT cause nuclear chromatin condensation and endonuclease activation. It appears clear that this mechanism is neither cell type nor species specific, given that, for example, mouse liver mitochondria in P T can promote the apoptotic disintegration of nuclei purified from human fibroblast-like nuclei (Fig. 1). O u r data indicate that mitochondria contain or are associated with (a) pre-formed soluble mediator(s) > 1 0 k D that is/are released after P T and that alone is/are sufficient to cause nuclear apoptosis (Fig. 9). In accord with published experiments performed on intact cells (16, 17), antioxidants do not neutralize this apoptosis inducer (Fig. 9). Thus, R O S that are formed by mitochondria after P T do not participate in the induction o f nuclear apoptosis; this is also indicated by experiments involving p~ cells that lack a functional respiratory chain (15, and Fig. 2). Moreover, it appears improbable that Ced-3-like proteases would be responsible for this apoptosis-inducing activity, given that the mammalian Ced-3 analogue CPP32 per se is not sufficient to induce nuclear apoptosis in a cell-free system (6). Thus, the molecular events linking mitochondrial P T to nuclear apoptosis await further characterization. From the available data, it appears that AXI/m disruption, which presumably is mediated by PT, is a constant feature o f early apoptosis (8-14). Indirect biochemical evidence has previously accused P T to participate in the postischemic or toxin-mediated death o f myocardial cells and hepatocytes (69-72), thus again suggesting that P T is a general regulator of cell death. Indeed, the P T pore is an attractive candidate for a death switch that, once activated, marks a point of no return in PCD. At least six reasons support this concept. First, as shown here, P T is both necessary and sufficient to cause nuclear apoptosis. Second, opening o f P T pores entails multiple potentially lethal alterations of mitochondrial function (loss o f A ~ m, uncoupling of the respiratory chain, hypergeneration o f R O S , and loss of mitochondrial glu- tathione and calcium; 12, 19-21) and thus may initiate pleiotropic death pathways. Moreover, as shown here, P T triggers a nuclear apoptosis effector pathway whose biochemical components remain elusive. Third, the P T pore functions as a sensor for multiple physiological effectors (divalent cations, ATP, ADP, N A D , A@m, pH, thiols, and peptides), thereby integrating information on the electrophysiological, redox, and metabolic state of the cell (19, 20, 73, and Fig. 3). Thus, different death inducers can converge at this level. Fourth, given that a P T pore constituent such as the A N T is essential for energy metabolism, mutations in this apoptosis-regulatory device will be mostly lethal for the cell. In teleological terms, this would have the advantage o f precluding apoptosis-inhibitory (oncogenic) mutations at this level of the apoptotic cascade. Fifth, at least one of the P T constituents, the A N T , is encoded by several members of a gene family that are expressed in a strictly tissue-specific manner (74). Thus, P T pores may be regulated in each cell type in a slightly different fashion. Sixth, P T is endowed with self-amplificatory properties in the sense that loss of matrix Ca 2+ and glutathione, depolarization of the inner membrane, and increased oxidation o f thiols, that result from P T pore opening, all increase the P T pore-gating potential (19-21, 23-29). T h e self-amplificatory property of P T is also underscored by the data presented in this paper. Thus, induction of PT induces A ~ m disruption (Figs. 1 and 7) and, conversely, AxIfm depolarization by mC1CCP causes PT, measured as large amplitude swelling (Figs. 3 and 8). Similarly, oxidant treatment causes P T (Figs. 3 and 8), and P T will ultimately entail mitochondrial generation o f tLOS (12). The fact that some consequences o f P T (e.g., AW m dissipation, R O S generation) themselves may cause P T suggests that P T may engage in a positive feedback loop that contributes to apoptotic autodestruction. Thus, P T would have to respond in an allor-nothing fashion and, once activated, would seal the cell's fate in an irreversible fashion. Accordingly, cells exhibiting an immediate consequence of PT, that is AXI'?m r e d u c t i o n , are irreversibly committed to cell death (10). Apart from these theoretical considerations, the current data suggest that the P T pore occupies a central position in apoptosis regulation. It therefore becomes an attractive target for regulation by pharmacological agents, as well as by endogenous apoptosis regulators belonging to the everexpanding Bcl-2 gene family. We are indebted to Dr. J.A. Duine for the gift of BA; Dr. Javier Naval (University of Zaragoza, Zaragoza, Spain) for p~ cells; Dr. Jonathan Ashwell for Bcl-2-transfected cells; and Dr. Roland Wenger for SDZ 220384. This work was supported by Association pour la Kecherche sur le Cancer, Agence Nationale pour la Recherche sur le SIDA, CNRS, Fondation pour la Recherche M~dicale, Institut National de la Sant~ et de la Recherche M~dicale, NATO, and the Leo Foundation (G. Kroemer). N. Zamzami and S.A. Susin received fellowships from Institut Scientifique Roussel and the Spanish Government, respectively. 1541 Zamzami et al. Address correspondence to Dr. Guido Kroemer, CNRS-UPR420, 19, rue Guy M6quet, B.P.8, F-94801 Villejuif, France. Received for publication 21 September 1995 and in revised form 6 December 1995. References 1. Jacobson, M.D., J.F. Burne, and M.C. Raft. 1994. Programmed cell death and Bcl-2 protection in the absence of a nucleus. E M B O (Eur. Mol. Biol. Organ.)J. 13:1899-1910. 2. Schulze-Osthoff, K., H. Walczak, W. Droge, and P.H. Krammer. 1994. Cell nucleus and DNA fragmentation are not required for apoptosis.J. Cell Biol. 127:15-20. 3. Nakajima, H., P. Golstein, and P.A. Henkart. 1995. The target cell nucleus is not required for cell-mediated granzymeor Fas-based cytotoxicity.J. Exp. Med. 181:1905-1909. 4. Newmeyer, D.D., D.M. Farschon, and J.C. Reed. 1994. Cell-free apoptosis in xenopus egg extracts: inhibition by Bcl-2 and requirement for an organelle fraction enriched in mitochondria. Cell. 79:353-364. 5. Lazebnik, Y.A., S.H. Kaufmann, S. Desnoyers, G.G. Poirier, and W.C. Earnshaw. 1994. Cleavage of poly(ADP-ribose) polymerase by a proteinase with properties like ICE. Nature (Lond.). 371:346-347. 6. Nicholson, D.W., A. All, N.A. Thornberry, J.P. Vaillancourt, C.K. Ding, M. Gallant, Y. Gareau, P.R. Griffin, M. Labelle, Y.A. Lazebnik, et al. 1995. Identification and inhibition of the ICE/CED-3 protease necessary for mammalian apoptosis. Nature (Lond.). 376:37-43. 7. Martin, S.J., and D.R. Green. 1995. Protease activation during apoptosis: death by a thousand cuts? Cell. 82:349-352. 8. Deckwerth, T.L., and E.M. Johnson. 1993. Temporal analysis of events associated with programmed cell death (apoptosis) of sympathetic neurons deprived of nerve growth factor. J. Cell Biol. 123:1207-1222. 9. Vayssi~re, J.-L., P.X. Petit, Y. Risler, and B. Mignotte. 1994. Commitment to apoptosis is associated with changes in mitochondrial biogenesis and activity in cell lines conditionally immortalized with simian virus 40. Proc. Natl. Acad. Sci. USA. 91:11752-11756. 10. Zamzami, N., P. Marchetti, M. Castedo, C. Zanin, J.-L. Vayssi~re, P.X. Petit, and G. Kroemer. 1995. Reduction in mitochondrial potential constitutes an early irreversible step of programmed lymphocyte death in vivo. J. Exp. Med. 181: 1661-1672. 11. Petit, P.X., H. LeCoeur, E. Zorn, C. Duguet, B. Mignotte, and M.L. Gougeon. 1995. Alterations ofmitochondrial structure and function are early events of dexamethasone-induced thymocyte apoptosis.J. Cell Biol. 130:157-167. 12. Zamzami, N., P. Marchetti, M. Castedo, D. Decaudin, A. Macho. T. Hirsch, S.A. Susin, P.X. Petit, B. Mignotte, and G. Kroemer. 1995. Sequential reduction of mitochondrial transmembrane potential and generation of reactive oxygen species in early programmed cell death. J. Exp. Med. 182: 367-377. 13. Castedo, M., A. Macho, N. Zamzami, T. Hirsch, P. Marchetti, J. Uriel, and G. Kroemer. 1995. Mitochondrial perturbations define lymphocytes undergoing apoptotic depletion in vivo. Eur. J. Immunol. 25:3277-3284. 14. Macho, A., M. Castedo, P. Marchetti, J.J. Aguilar, D. Decaudin, N. Zamzami, P.M. Girard, J. Uriel, and G. Kroemer. 1995. Mitochondrial dysfunctions in circulating T lympho1542 cytes from human immunodeficiency virus-1 carriers. Blood. 86:2481-2487. 15. Jacobson, M.D., J.F. Burne, M.P. King, T. Miyashita, J.C. Reed, and M.C. Raft. 1993. Bcl-2 blocks apoptosis in cells lacking mitochondrial DNA. Nature (Lond.). 361:365-369. 16. Jacobson, M.D., and M.C. Raft. 1995. Programmed cell death and Bcl-2 protection in very low oxygen. Nature (Lond.). 374:814-816. 17. Shimizu, S., Y. Eguchi, H. Kosaka, W. Kamlike, H. Matsuda, and Y. Tsujimoto. 1995. Prevention of hypoxiainduced cell death by Bcl-2 and Bcl-xL. Nature (Lond.). 374: 811-813. 18. Kroemer, G., P.X. Petit, N. Zamzami, J.-L. Vayssi~re, and B. Mignotte. 1995. The biochemistry ofapoptosis. FASEB (Fed. Am. Soc. Exp. Biol.)J. 9:1277-1287. 19. Bernardi, P., K.M. Broekemeier, and D.R. Pfeiffer. 1994. Recent progress on regulation of the mitochondrial permeability transition pore; a cyclosporin-sensitive pore in the inner mitochondrial membrane. J. Bioenerg. Biomembr. 26:509517. 20. Zoratti, M., and I. Szab6. 1994. Electrophysiology of the inner mitochondrial membrane.J. Bioenerg. Biomembr. 26:543553. 21. Reed D.J., and M.K. Savage. 1995. Influence of metabolic inhibitors on mitochondrial permeability transition and glutathione status. Biochim. Biophys. Acta. 1271:43-50. 22. Zoratti, M., and I. Szab6, 1995. The mitochondrial permeability transition. Biochim. Biophys. Acta - Rev. Biomembranes. 1241:139-176. 23. Bemardi, P., S. Vassanelli, P. Veronese, R. Colonna, I. Szab6, and M. Zoratti. 1992. Modulation of the mitochondrial permeability transition pore. Effect of protons and divalent cations.J. Biol. Chem. 267:2934-2939. 24. Bemardi, P. 1992. Modulation of the mitochondrial cyclosporin A~ensitive permeability transition pore by the proton electrochemical gradient. J. Biol. Chem. 267:8834-8839. 25. Szab6, I., P. Bernardi, and M. Zoratti. 1992. Modulation of the mitochondrial megachannel by divalent cations and protons.J. Biol. Chem. 267:2940-2946. 26. Petronilli, V., C. Cola, S. Massari, R. Colonna, and P. Bernardi. 1993. Physiological effectors modify voltage-sensing by the cyclosporin A~sensitive mitochondrial permeability transition pore.J. Biol. Chem. 268:21939-21945. 27. Bernardi, P., P. Veronese, and V. Petronilli. 1993. Modulation of the mitochondrial cyclosporin A-sensitive permeability transition pore. I. Evidence for two separate Me 2+ binding sites with opposing effects on the pore open probability. J. Biol. Chem. 268:1005-1010. 28. Petronilli, V., C. Cola, and P. Bernardi. 1993. Modulation of the mitochondrial cyclosporin A-sensitive permeability transition pore. II. The minimal requirements for pore induction underscore a key role for transmembrane electric potential, matrix pH, and matrix Ca2+. J. Biol. Chem. 268:1011-1016. 29. Petronilli, V., A. Nicolli, P. Costantini, R. Colonna, and P. Bernardi. 1994. Regulation of the permeability transition Mitochondrial Regulation of Apoptosis pore, a voltage-dependent mitochondrial channel inhibited by cyclosporin A. Biochim. Biophys. Acta. 1187:255-259. 30. McEnery, M.W., A.M. Snowman, R.R. Trifiletti, and S.H. Snyder. 1992. Isolation of the mitochondrial benzodiazepine receptor: association with the voltage-dependent anion channel and the adenine nucleotide carrier. Proc. Natl. Acad. Sci. USA. 89:3170-3174. 31. Klingenberg, M. 1980. The ADP-ATP translocation in mitochondria, a membrane potential controlled transportJ. Membr. Biol. 56:97-105. 32. Toninello, A., D. Siliprandi, and N. Siliprandi. 1984. On the mechanism by which Mg2+ and adenine nucleotides restore membrane potential in rat liver mitochondria deenergized by Ca 2+ and phosphate. Biochem. Biophys. Res. Commun. 111: 792-797. 33. Halestrup, A.P., and A.M. Davidson. 1990. Inhibition of CaZ+-induced large-amplitude swelling of liver and heart rnitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial-matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenine nucleotide translocase. Biochem.J. 268:153-160. 34. Marry, l., G. Brandolin, J. Gagnon, R. Brasseur, and P.V. Vignais. 1992. Topography of the membrane-bound ADP/ ATP carrier assessed by enzymatic proteolysis. Biochemistry. 31:4058-4065. 35. Majima, E., Y. Shinohara, N. Yamaguchi, Y.M. Hong, and H. Terada. 1994. Importance of loops of mitochondrial ADP/ATP carrier for its transport activity deduced from reactivities of its cysteine residues with the sulflaydril reagent eosin-5-maleimide. Biochemistry. 33:9530-9536. 36. Leist, M., F. Gantner, I. Bohlinger, P.G. Germann, C. Tiegs, and A. Wendel. 1994. Murine hepatocyte apoptosis induced in vitro and in vivo by TNF-alpha requires transcriptional arrest.J. Immunol. 153:1778-1788. 37. Gonzalo, J.A., A. Gonzklez-Garcia, C. Martinez-A., and G. Kroemer. 1993. Glucocorticoid-mediated control of the clonal deletion and activation of peripheral T cells in vivo. J. Exp. Med. 177:1239-1246. 38. Green, D.R., A. Mahboubi, W. Nishioka, S. Oja, F. Echeverri, Y. Shi, J. Glynn, Y. Yang, J. Ashwell, and R. Bissonnette. 1994. Promotion and inhibition of activation-induced apoptosis in T-cell hybridomas by oncogenes and related signals. Immunol. Rev. 142:321-342. 39. Aten, J., P. Prigent, P. Poncet, C. Blanpied, N. Claessen, P. Druet, and F. Hirsch. 1995. Mercuric chloride-induced programmed cell death ofa murine T cell hybridoma: I. Effect of the proto-oncogene Bcl-2. Cell. Immunol. 161:98-106. 40. Lumbach, G.W.M., H.C. Cox, and W. Berends. 1970. Elucidation of the chemical structure of bongkrekic acid. I. Isolation, purification and properties of bongkrekic acid. Tetrahedron. 26:5993-5999. 41. Wood, E.R., and W.C. Earnshaw. 1990. Mitotic chromatin condensation in vitro using somatic cell extracts and nuclei with variable levels of endogenous topoisomerase II. J. Cell Biol. 111:2839-2850. 42. Lazebnik, Y.A., S. Cole, C.A. Cooke, W.G. Nelson, and W.C. Earnshaw. 1993. Nuclear events ofapoptosis in vitro in celt-free mitotic extracts: a model system for analysis of the active phase ofapoptosis.J. Cell Biol. 123:7-22. 43. Boutry, M., and M. Briquet. 1982. Mitochondrial modifications associated with the cytoplasmic male sterility in faba beans. Eur.J. Biochem. 127:129-135. 44. Kawabe, Y., and A. Ochi. 1991. Programmed cell death and 1543 Zamzami et al. extrathymic reduction of V[38 + CD4 + T cells in mice tolerant to Staphylococcus aureus enterotoxin B. Nature (Lond.). 349: 245-248. 45. Petit, P.X., J.E. O'Connor, D. Grunwald, and S.C. Brown. 1990. Analysis of the membrane potential of rat- and mouseliver mitochondria by flow cytometry and possible applications. Eur. J. Biochem. 389-397. 46. Nicoletti, I., G. Migliorati, M.C. Pagliacci, and C. Riccardi. 1991. A rapid simple method for measuring thymocyte apoptosis by propidium iodide staining and flow cytometry. J. Immunol. Meth. 139:271-280. 47. Brandolin, G., A. Le-Saux, V. Trezeguet, G.J. Lauquinn, and P.V. Vignais. 1993. Chemical, immunological, enzymatic, and genetic approaches to studying the arrangement of the peptide chain of the ADP/ATP carrier in the mitochondrial membrane.J. Bioenerg. Biomembr. 25:493-501. 48. Bellomo, G., A. Martino, P. Richelmi, G.A. Moore, S.A. Jewell, and S. Orrenius. 1984. Pyridine-nucleotide oxidation, Ca 2+ cycling and membrane damage during tert-butylhydroperoxide metabolism by rat liver mitochondria. Eur. J. Biochem. 140:1~. 49. Petronilli, V., P. Costantini, L. Scorrano, R. Colonna, S. Passamonti, and P. Bernardi. 1994. The voltage sensor of the mitochondrial permeability transition pore is tuned by the oxidation-reduction state ofvicinal thiols. Increase of the gating potential by oxidants and its reversal by reducing agents. .]. Biol. Chem. 269:16638-16642. 50. Costantini, P., B.V. Chernyak, V. Petronilli, and P. Bernardi. 1995. Selective inhibition of the mitochondrial permeability transition pore at the oxidation-reduction sensitive dithiol by monobromobimane. FEBS (Fed. Eur. Biochem. Sot.) Lett. 362:239-242. 51. Zenke, G., G. Baumann, R. Wenger, O. Hiestand, V. Quesniaux, E. Andersen, and M.H. Schreier. 1993. Molecular mechanisms of immunosuppression by cyclosporins. Ann. N.Y. Acad. ScL 23:330-335. 52. Broekemeier, K.M., R.J. Krebsbach, and D.R. Pfeiffer. 1994. Inhibition of the mitochondrial Ca 2+ uniporter by pure and impure ruthenium red. Mol. Cell. Biochem. 139:33-40. 53. Leist, M., F. Gantner, I. Bohlinger, G. Tiegs, P.G. Germann, and A. Wendel. 1995. Tumor necrosis factor-induced hepatocyte apoptosis precedes liver failure in experimental murine shock models. Am]. Pathol. 146:1220-1234. 54. Gory, S. 1995. Regulation of lymphocyte survival by the Bcl-2 gene family. Ann. Rev. Immunol. 13:513-543. 55. Tanaka, S., K. Saito, andJ.C. Reed. 1993. Structure-function analysis of the Bcl-2 oncoprotein. Addition ofa heterologous transmembrane domain to portions of the Bcl-2[3 protein restores function as a regulator of cell survival. J. Biol. Chem. 268:10920-10926. 56. Krajewski, S., S. Tanaka, S. Takayama, M.J. Schibler, W. Fenton, andJ.C. Reed. 1993. Investigation of the subcellular distribution of the bcl-2 oncoprotein: residence in the nuclear envelope, endoplasmic reticulum, and outer mitochondrial membranes. Cancer Res. 53:4701-4714. 57. Nguyen, M., P.E. Branton, P.A. Walton, Z.N. Oltvai, S.J. Korsmeyer, and G.C. Shore. 1994. Role of membrane anchor domain of Bcl-2 in suppression of apoptosis caused by E1B-defective adenovirus.-]. Biol Chem. 269:16521-16524. 58. Hockenbery, D.M., Z.N. Oltavai, X.-M. Yin, C.L. Milliman, and S.J. Korsmeyer. 1993. Bcl-2 functions in an antioxidant pathway to prevent apoptosis. Cell. 75:241-251. 59. Caron-Leslie, L.A.M., R.B. Evans, andJ.A. Cidlowski. 1994. Bcl-2 inhibits glucocorticoid-induced apoptosis but only partially blocks calcium ionophore or cycloheximide-regulated apoptosis in $49 cells. FASEB (Fed. Am. Soc. Exp. Biol.)J. 8: 639-645. 60. Martin, S.J., D.D. Newmeyer, S. Mathisa, D.M. Farschon, H.G. Wang, J.C. Reed, R.N. Kolesnick, and D.R. Green. 1995. Cell-free reconstitution ofFas-, UV radiation- and ceramide-induced apoptosis. E M B O (Eur. Mol. Biol. Organ.) J. 14:5191-5200. 61. Igbavboa, U., C.W. Zwizinski, and D.R. Pfeiffer. 1989. Release of mitochondrial matrix proteins through a Ca2+requiring, cyclosporin-sensitive pathway. Biochem. Biophys. Res. Commun. 161:619-625. 62. Ellis, R.E., Y. Yuan, and H.R. Horvitz. 1991. Mechanisms and functions of cell death. Annu. Rev. Cell. Biol. 7:663--698. 63. Oltvai, Z.N., and S.J. Korsmeyer. 1994. Checkpoints of dueling dimers foil death wishes. Cell. 79:189-192. 64. Smets, L.A., J. Van den Berg, D. Acton, B. Top, H. van Rooij, and M. Verwijs-Janssen. 1994. BGL-2 expression and mitochondrial activity in leukemic cells with different sensitivity to glucocorticoid-induced apoptosis. Blood. 5:16131619. 65. Hennet, T., C. Richter, and E. Peterhans. 1993. Tumour necrosis factor-alpha induces superoxide anion production in mitochondria ofL929 cells. Biochem.J. 289:587-592. 66. Fang, W., J.J. Rivard, D.L. Mueller, and T.W. Behrens. 1994. Cloning and molecular characterization of mouse bcl-x in B and T lymphocytes.J. Immunol. 153:4388-4398. 67. Hengartner, M.O., and H.R. Horvitz. 1994. C. elegans cell survival gene ced-9 encodes a functional homolog of the 1544 mammalian proto-oncogene bcl-2. Cell. 76:665-676. 68. Kane, D.J., T.A. Sarafian, R. Anton, H. Hahn, E.B. Gralla, J.S. Valentine, T. Ord, and D.E. Bredesen. 1993. Bcl-2 inhibition of neural death: decreased generation of reactive oxygen species. Science (Wash. DC). 262:1274-1277. 69. Crompton, M., H. Ellinger, and A. Costi. 1988. Inhibition by cyclosporin A of a CaZ+-dependent pore in heart mitochondria activated by inorganic phosphate and oxidative stress. Biochem. J. 255:357-360. 70. Andreeva, L., A. Tanveer, and M. Crompton. 1995. Evidence for the involvement of a membrane-associated cyclosporin-A-binding protein in the Ca2+-activated inner membrane pore of heart mitochondria. Eur. J. Biochem. 230: 1125-1132. 71. Mehendale, H.M., R.A. Roth, A.J. Gandolfi, J.E. Klaunig, J.J. Lemasters, and L.R. Curtis. 1994. Novel mechanisms in chemically induced hepatotoxicity. FASEB (Fed. Am. Soc. Exp. Biol.)J. 8:1285-1295. 72. Pastorino, J,G., J.W. Snyder, J.B. Hoek, and J.L. Farber. 1995. Ca 2+ depletion prevents anoxic death of hepatocytes by inhibiting mitochondrial permeability transition. Am J. Physiol. 268:C676-685. 73. Pfeiffer, D.R., T.I. Gudz, S.A. Novgorodov, and W.L. Erdahl. 1995. The peptide mastoparan is a potent facilitator of the mitochondrial permeability transition. J. Biol. Chem, 270: 4923-4932. 74. Walker, J.E., and M.J. Runswick. 1993. The mitochondrial transport protein superfamily. J. Bioenerg. Biomembr. 5:435446. Mitochondrial Regulation of Apoptosis 

Mitochondrial Control of Nuclear Apoptosis By Naoufal Zamzami,* Santos A. Susin,* Philippe Marchetti,* Tamara Hirsch,* Isabel G6mez-Monterrey,r Maria Castedo,* and Guido Kroemer* From * Centre National de la Recherche Scientifique (CNRS)-UPR420, F-94801 Villejuif, France; and r Institute for Medical Chemistry, Consejo Superior de Investigadtnes Cientificas, 28008 Madrid, Spain Summary Anucleate cells can be induced to undergo programmed cell death (PCD), indicating the existence of a cytoplasmic PCD pathway that functions independently from the nucleus. Cytoplasmic structures including mitochondria have been shown to participate in the control of apoptotic nuclear disintegration. Before cells exhibit common signs of nuclear apoptosis (chromatin condensation and endonuclease-mediated DNA fragmentation), they undergo a reduction of the mitochondrial transmembrane potential (A~m) that may be due to the opening of mitochondrial permeability transition (PT) pores. Here, we present direct evidence indicating that mitochondrial PT constitutes a critical early event of the apoptotic process. In a cell-free system combining purified mitochondria and nuclei, mitochondria undergoing PT suffice to induce chromatin condensation and DNA fragmentation. Induction of PT by pharmacological agents augments the apoptosis-inducing potential of mitochondria. In contrast, prevention of PT by pharmacological agents impedes nuclear apoptosis, both in vitro and in vivo. Mitochondria from hepatocytes or lymphoid cells undergoing apoptosis, but not those from normal cells, induce the disintegration of isolated Hela nuclei. A specific ligand of the mitochondrial adenine nucleotide translocator (ANT), bongkrekic acid, inhibits PT and reduces apoptosis induction by mitochondria in a cell-free system. Moreover, it inhibits the induction of apoptosis in intact cells. Several pieces of evidence suggest that the proto-oncogene product Bcl-2 inhibits apoptosis by preventing mitochondrial PT. First, to inhibit nuclear apoptosis, Bcl-2 must be localized in mitochondrial but not in nuclear membranes. Second, transfection-enforced hyperexpression of Bcl-2 directly abolishes the induction of mitochondrial PT in response to a protonophore, a pro-oxidant, as well as to the A N T ligand atractyloside, correlating with its apoptosis-inhibitory effect. In conclusion, mitochondrial PT appears to be a critical step of the apoptotic cascade. ince it has been shown that anucleate cells (cytoblasts) can be induced to undergo programmed cell death (PCD) 1 (1-3), it has become clear that a cytoplasmic PCD pathway must function independently from the nucleus. Both mitochondria (4) and specific ced-3-1ike proteases S 1Abbreviations used in this paper: ANT, adenine nucleotide translocator; Atr, atractyloside;BA, bongkrekicacid; CsA, cyclosporinA; DAPI, 4'-6diamidino-2-phenylindole dihydrochloride;DEX, dexamethasone;diamide, diazenedicarboxylicacid his 5N,N-dimethylamide; DiOC6(3), 3,3'dihexyloxacarbocyanineiodide; Aq~m,mitochondrialtransmembrane potential; GAIN,D-galactosamine;ICE, IL-113convertingenzyme;MCB, monochlorobimane; mCICCP, carbonylcyanidem-chlorophenylhydrazone; mtDNA, mitochondrialDNA; PCD, programmedcell death; PT, permeability transition; P,OS, reactive oxygen species; R.tL, ruthenium red; ter-BHP,ter-butylhydroperoxide. N. Zamzamiand S.A. Susincontributedequallyto this paper. This paper is dedicatedto Jos6 Uriel. 1533 (5-7) have been accused of participating in the cytoplasmic control of apoptotic nuclear disintegration. We and others (8-12) have recently demonstrated that cells undergo a reduction of the mitochondrial transmembrane potential (A~m) before they exhibit common signs of nuclear apoptosis (chromatin condensation and endonuclease-mediated DNA fragmentation). This applies to different cell types (neurons, fibroblasts, B and T lymphocytes, pre-B cells and thymocytes, myelomonocytic cells) and to different physiological apoptosis inducers (growth factor withdrawal, tumor necrosis factor, ceramide, glucocorticoids, activation-induced cell death, positive and negative selection, irradiation; 8-13). Moreover, these observations extend to pathogen-induced apoptosis, including irradiation-induced PCD (13) and HIV-l-triggered T lymphocyte PCD (14). When PCD is prevented either by genetic manipulations (e.g., p53 loss mutation, bcl-2 hyperexpression) or by pharmacological j. Exp. Med. 9The Rockefeller University Press 90022-1007/96/04/1533/12 $2.00 Volume 183 April 1996 1533-1544 agents (N-acetylcysteine, protease inhibitors, linomide), both mitochondrial and nuclear signs o f apoptosis are abolished (10, 12, 13). Moreover, cells that have lost their Axlr m appear to be irreversibly programmed to die (10). Although these observations suggest the involvement o f mitochondria in apoptosis, they do not clarify the cause-effect relationship between mitochondrial dysfunction and subsequent nuclear apoptosis. It appears clear that reactive oxygen species (ROS), which may be generated by uncoupled mitochondria (9, 12), are not essential for the apoptotic process (15-17). Thus, whenever a cause-effect relationship between mitochondrial disorders and nuclear apoptosis exists, it must be mediated by factors other than R O S . The aim o f this paper was to unravel the existence of such a pathway linking mitochondrial dysfunction to nuclear disintegration. As to the mechanism ofapoptotic Axltm disruption, pharmacological experiments suggest that it involves the opening o f so-called mitochondrial permeability transition (PT) pores (12, 18). Under normal conditions, the inner mitochondrial membrane is quasi-impermeable for small molecules, thus allowing for the creation o f the electrochenfical gradient which is indispensable for mitochondrial function. However, in determined circumstances, opening o f P T pores or "megachannels" allows for the free distribution o f solutes o f < 1 , 5 0 0 daltons and o f some proteins, thereby disrupting the A~tI'tm and associated mitochondrial functions (19, 20). In isolated mitochondria, P T is accompanied by colloidosmotic swelling and uncoupling o f oxidative phosphorylation, as well as by the loss o f low molecular weight matrix molecules such as calcium and glutathione (19-21). It may be important to note that P T is modulated by multiple different physiological and pharmacological inducers and inhibitors (for a review see reference 22) and that P T is both the cause and the consequence o f AW m dissipation, as well as o f reactive oxygen metabolite production (19-29). In other terms, P T results ipso facto in A ~ m disruption and later in R O S hyperproduction, but AXlt m reduction and R O S themselves can also provoke PT, as do many other factors (divalent cations, pH variations, peptides, etc; 22). The exact molecular composition o f the P T pore is not known. However, it appears that at least one inner mitochondrial transmembrane protein, namely the adenine nucleotide translocator (ANT), is involved in P T pore formation (for reviews see references 19, 20) and that A N T associates with several molecules o f the outer mitochondrial membrane such as the peripheral benzodiazepine receptor and the voltage-dependent anion channel (30). A N T ligands such as atractyloside (Atr) and bongkrekic acid (BA) enhance or reduce the probability o f PT, respectively (31-35). Based on these premises, we have tested the hypothesis that PT might be the critical event determining the apoptosis-inducing potential of mitochondria. Using a cell- and cytosol-free system in which purified mitochondria and nuclei are confronted, we show that induction of PT by the A N T ligand Atr or other less specific P T inducers causes isolated mitochondria to trigger nuclear apoptosis. In contrast, inhibition o f PT by the Atr antagonist BA, as well 1534 as by a variety o f additional P T inhibitors, abolishes mitochondria-mediated nuclear apoptosis. The apoptosis-inhibitory proto-oncogene product Bcl-2 functions as an endogenous inhibitor o f mitochondrial PT. These data establish mitochondrial P T as a critical event o f apoptosis. Materials and M e t h o d s Animals and In Vivo Treatments. Male 6--10-wk-old BALB/c mice were injected simultaneously with D-galactosamine (GAIN; 10 mg i.p.) and/or LPS from Escherichiacoli (Signa Chemical Co., St. Louis, MO; 50 btg i.v.), 5 h before removal of the liver (36). Alternatively, splenocytes were recovered from BALB/c mice 12 h after injection of 1 mg i.p. dexamethasone (DEX; Sigma Chemical Co.) in 200 btl PBS or PBS alone (10, 37). Cell Lines and In Vitro Culture Conditions. U937 cells were depleted from mitochondrial DNA (mtDNA) by continuous ethidium bromide selection for 4 mo (15). Control experiments revealed that such cells become resistant to antimycin A, which blocks the mtDNA-encoded complex III. Moreover, no mtDNA could be detected by PCP,. (not shown). 2B4.11 T cell hybridoma cell lines stably transfected with an SFFV.neo vector containing the human bcl-2 gene or the neomycin (Neo) resistance gene only (38, 39) were kindly provided by Jonathan Ashwell (National Institutes of Health, Bethesda, MD). Cells were cultured in RPMI-1640 medium containing 5% FCS. Apoptosis was induced by culturing cells in the presence of the indicated concentration ofdiazenedicarboxylic acid bis 5N, N-dimethylamide (diamide) or carbonyl cyanide m-chlorophenylhydrazone (mC1CCP; both from Sigma Chemical Co.). DNA fragmentation of non- or ",/-irradiated (10 Gy) thymocytes (106 cells/lane) was monitored after culturing cells for 4 h in the presence of DEX (1 I~M), etoposide (10 ~M; Sigma Chemical Co.), and/or BA (50 btM; purified as described in reference 40), kindly provided by Dr. J.A. Duine (Delft University, Delft, The Netherlands). Cell-free System of Apoptosis. Nuclei from HeLa or 2B4.11 cells were purified on a sucrose gradient, as described (41), and were resuspended in CFS buffer (220 nM mannitol, 68 mM sucrose, 2 mM NaC1, 2.5 mM PO4H2K, 0.5 mM EGTA, 2 mM C12Mg, 5 mM pyruvate, 0.1 mM PMSF, 2 mM ATP, 50 Ixg/nfl creatine phosphokinase, 10 mM phosphocreatine, 1 mM dithiothreitol, and 10 mM Hepes-NaOH, pH 7.4; reagents from Sigma Chemical Co.). Nuclei were conserved at -20~ in 50% glycerol for up to 8 d as described (41, 42). Mitochondria were purified from BALB/c mouse livers, splenocytes, or U937 cells on a Percoll gradient (43) and were stored on ice in B buffer (400 mM mannitol, 10 mM PO4H2K, 5 mg/ml BSA, and 50 mM TtisHC1, pH 7.2) for up to 4 h. For quantitation of nuclear apoptosis, both nuclei (5,000g, 5 rain) and mitochondria (2 • 104g, 3 min) were spun down and washed twice in CFS buffer before being mixed. In standard conditions, mitochondria (500 ng/~l protein final concentration) were cultured at 37~ for 90 min with 103 nuclei per ~1 CFS containing a number of different agents: AIF3, (20 p,M), Atr (5 raM; Sigma Chemical Co.), BA (50 ~M), CaC12 (500 I.zM), mC1CCP (10 p~M), diamide (100 p,M), cyclosporin A (CsA, 10 /a,M; Sandoz AG, Basel, Switzerland), N-methylVal-4CsA (SDZ 220-384, 10 btM; kindly provided by Dr. Roland Wenger, Sandoz), monochlorobimane (MCB; 30 p~M), phosphotyrosine (P; 10 raM), ruthenium red (RR, 100 ~M; Sigma Chemical Co.), ter-butylhydroperoxide (ter-BHP, 50 btM; Sigma Chelnical Co.), ZnC12 (1 raM), AcYVAD-CHO (IL-1]3 converting enzyme [ICE] inhibitor I), AcYVAD-chloromethylketone (ICE inhibitor II), and/or AcDEVD-CHO (inhibitor of CPP32/ Mitochondrial Regulation of Apoptosis Ced3/Yama; Bachem, Basel, Switzerland). Nuclei were stained with 4'-6-diamidino-2-phenylindoledihydrochloride (DAPI; 10 ~M) and examined by fluorescence microscopy (5), or were analyzed by agarose gel electrophoresis (106 nuclei/lane) (44). Cytofluorometric Analysis. For AxI/mdeterminations, isolated mitochondria were incubated for 15 rain at 37~ in the presence of DiOC6(3 ) (80 riM) (45), followed by addition of mC1CCP (50 p~M), BA (50 I~M) and/or Atr (5 raM), and recording of the fluorescence in an Elite cytofluorometer (Coulter Corp., Hialeah, FL) 5 rain later. Loss of nuclear DNA (hypoploidy) was determined by propidium iodine staining of ethanol-fixed cells, as described (46). Large Amplitude Swelling of Isolated Mitochondria. Large amplitude swelling is a colloidosmotic process that is observed among isolated mitochondria undergoing PT in solutions containing low protein concentrations (22). For determination of swelling, mitochondria were washed and resuspended in B buffer (100 ~g protein/10 ~1 buffer), followed by addition of 90 p~M CFS buffer and recording of adsorption at 540 nm in a spectrophotometer (model DU 7400; Beckman Instruments, Inc., Fullerton, CA), as described (26). The loss of absorption induced by 5 mM Atr within 5 rain was considered 100% the value of large amplitude swelling. Characterization of Factors Contained in the Supernatant of Mitochondria. Hepatic mitochondria (1 mg/ml in CFS buffer) were left untreated or were incubated with Atr (5 raM) for 10 rain at room temperature, followed by ultracentrifugation (1.5 • 10s g, 30 rain, 4~ Supernatants were either left untreated or centrifuged through a Centricon 10 membrane (Amicon Inc., Beverly, MA) to separate proteins with an approximate molecular mass of > and <10 kD, following the manufacturers' recommendation. The fraction >10 kD was reconstituted with CFS to the original volume. Supematants (50% volume) were mixed with nuclei (103/~1) in the presence or absence of various antioxidants (50 p~M N-t-butyl-ot-phenylnitrone,230 ~M trolox, 600 IxM L-ascorbate, or 1 mg/ml catalase; Sigma Chemical Co.), and nuclei were stained with DAPI after 90 min of culture at 37~ Results and D i s c u s s i o n Isolated Mitochondria Undergoing P T Induce Nuclear Apoptosis in a Cell-free System. O n e o f the P T pore constituents is the A N T . T h e A N T adopts different molecular conformations w h e n exposed to two specific ligands, Atr and BA (31, 47). Atr favors the opening of the P T pore, whereas BA reduces the probability of P T pore gating (31-35, 47). W e have tested the capacity o f purified mitochondria with open and closed P T pores to induce nuclear apoptosis in a modified cell-free system (4, 42). Mitochondria were purified from the murine liver and were mixed with Hela n u clei in an isotonic buffer containing an A T P regeneration system. O n their own, unmanipulated mitochondria derived from healthy cells are incapable of inducing signs of nuclear apoptosis such as chromatin condensation and endonuclease-mediated D N A fragmentation. However, mitochondria exposed to a dose of Atr that causes PT, determ i n e d either as large amplitude swelling (Fig. 1 A) or as disruption of the AXI~m (Fig 1 B), do induce nuclear apoptosis (Fig. 1 C) in a time- and dose-dependent fashion (Fig. 1 D). Atr does not induce apoptosis itsel~ it only favors the induction o f nuclear apoptosis w h e n mitochondria are present (Fig. 1 C). In the presence o f B A , mitochondria fail to undergo P T in response to Atr (Fig. 1, A and B) and lose the capacity to induce chromatin condensation (Fig. 1 C) and associated oligonucleosomal D N A fragmentation (Fig. Figure !. Regulation of mitochondrial PT and mitochondriamediated nuclear apoptosis by two adenine nucleotide translocator ligands. (A) Effect of atractyloside (Atr) and bongkrekic acid (BA) on mitochondriallarge amplitude swelling. (Arrows) Time points at which BA (50 p~M) and/or Atr (5 mM) were added to mitochondria. (B) A~,,, of Atr- and BA-treated mitochondria, as determined by incorporation of DiOC6(3) into mitochondria, 5 rain after addition of Atr and/or BA. Incubation with the protonophor mC1CCP, which completelydisrupts the A~ m, unravels DiOC6(3) background fluorescence. (C) Chromatin distribution of isolated HeLa nuclei cultured with mitochondria, Atr, and/or BA for 90 rain. Single nuclei stained with DAPI representing the dominant phenotype (I>80%) are shown. (D) Time and dose dependence of nuclear chromatin condensation.Nuclei were incubated during the indicated interval with the specified amount of mitochondria (doses in I~g protein) and/or Atr. (E) Fragmentationof nuclear DNA induced by mitochondria. Purified HeLa nuclei were coculturedwith mitochondria (500 ng/~tl), Atr, and/or BA for 90 min (same conditionsas in C), followedby agarose gel electrophoresis of ethidium bromide-stained DNA. Lane I: Atr only; lane 2: mitochondriaonly; lane 3: mitochondriaplus Atr; and lane 4: mitochondriaplus Atr Flus BA. 1535 Zamzami et al. 1 E). These findings suggest that PT controls the apoptosisinducing capacity ofmitochondria. Mitochondria Lacking mtDNA Can Undergo PT and Cause Nuclear Apoptosis in a Cell-free System. Although most proteins contained in mitochondria are encoded by nuclear genes, a number of proteins including some components of the respiratory chain complexes I, III, and IV, are encoded by the mitochondrial genome. Previously, it has been reported that cell lines lacking mtDNA can undergo fullblown nuclear apoptosis (15), and this finding could give rise to the interpretation that mitochondria are not important for the control ofapoptosis. To challenge this (over)interpretation, we purified mitochondria from cells lacking mtDNA (po cells), as well as from control p+ cells, and tested their apoptosis-inducing potential. As shown in Fig. 2 A, pO mitochondria can undergo large amplitude swelling in response to Atr, exactly as do p+ control organelles. Moreover, in the presence of Atr, pO mitochondria are as efficient inducers of nuclear disintegration, as are control p+ mitochondria (Fig. 2 B). This indicates that all mitochondrial functions critical for apoptosis induction are encoded by nuclear genes. Thus, in accord with the published data (15), the capacity of mitochondria to induce nuclear apoptosis does not depend on the presence of mtDNA. Furthermore, the fact that respiration-deficient po mitochondria (which may be expected to produce less ROS than p+ control organelles) conserve their proapoptotic activity, suggests that R O S do not mediate apoptosis in this cell-free system. Strict Correlation Between PT-associated Swelling and Nuclear Apoptosis Induction. The above results suggest that mitochondria are indeed efficient inducers of nuclear apoptosis, provided that they are undergoing PT. This conclusion is corroborated by the strict correlation between PT and the proapoptotic effect of mitochondria, when PT is induced by a variety of different molecules: Atr, the pro-oxidant terBHP, and calcium ions via the thiol-cross-linking agent diamide or the protonophore mC1CCP (Fig.3). All these re- A B .qA..I 100" iI~II0" pO ~l~ 20" II t li l i time (sec) ii# z2o II @d _Illl. LIts-iL .IL# LAtrl #+ po Figure 2. Mitochondrialswelling and nuclear apoptosis induction do not require mitochondrialDNA. (A) Large amplitude swelling of mitochondria lacking mtDNA. Mitochondria obtained from normal U937 cells (p§ or U937 cellslackingmtDNA (pC)were incubatedwhile assessing absorbanceat 540 nM. (Arrow)Addition of Atr (5 mM). (B) Induction of nuclear apoptosisby mitochondrialacking mtDNA. HeLa nuclei were cultured for 90 rain with Atr and/or mitochondriaobtained from p+ or poU937 cells,followedby determinationof the frequencyofapoptoticor damaged nuclei. 1536 agents are thought to act via distinct mechanisms: Atr by virtue of its capacity to interact with the ANT (31-35); calcium via conformational effects on proteins that are yet poorly understood (19, 20, 22); hydroperoxides via oxidation of mitochondral glutathione and pyridine nucleotides (48); diamide via its thiol-cross-linking action on the ANT(49, 50); and protonophores via dissipation of the proton gradient (A~m), then entailing PT as a secondary phenomenon (24, 27, 28). Control experiments indicate that none of these reagents induces nuclear apoptosis by itself, i.e., in the absence ofmitochondria (not shown). Atrinduced swelling and nuclear apoptosis are efficiently inhibited by BA and CsA, as well as by the thiol reagent MCB. CsA can be substituted for by the nonimmunosuppressive CsA analogue N-methylVal-4-CsA (51), indicating that its PT and apoptosis-inhibitory effect is not mediated via calcineurin. A series of substances previously reported to inhibit apoptosis in a cell-free system (4) can also inhibit both PT and mitochondria-mediated nuclear apoptosis: phosphotyrosine, ZnC12, and A1F 3 (Fig. 3). Other substances (4) have no or little (<20%) inhibitory effects on Atr-induced swelling and apoptosis: calpain inhibitors I and 1I, GTPTS, and ionomycin. Similarly, synthetic tetrapeptide inhibitors of ICE and of CPP32/Yama fail to interfere with PT and PT-dependent nuclear apoptosis (Table 1). Calcium-driven but not Atr-induced PT and apoptosis are selectively inhibited by R R , a specific inhibitor of the mitochondrial calcium uniport (52), underscoring the fact that proapoptotic calcium effects are indeed mediated by mitochondria. As expected (50), the syn-9,10-dioxa-bimanine halogen derivative MCB is particularly effective in inhibiting PT and apoptosis induced by diamide (Fig. 3). In accord with published data (for a review see reference 22), none of the inhibitors used in this study is capable of providing long-term (>30 rain) protection against mitochondrial swelling in response to the whole panel of PT inducers. This probably reflects the profound differences in the molecular mechanisms of PT caused by different inducers (19, 20, 22, 24, 27, 28, 48-50). In synthesis, the strict correlation existing between mitochondrial PT and mitochondria-mediated nuclear apoptosis suggests that PT is indeed a crucial event in the regulation ofapoptosis induction by mitochondria. Moreover, the fact that none of the inhibitory substances (BA, CsA, MCB, R R , phosphotyrosine, ZuC12, A1F3) suppresses PT and apoptosis in response to all PT inducers (Fig. 3) suggests that they do not directly affect nuclei but rather act via PT modulation. Finally, the data summarized in Fig. 3 underscore the complex pharmacology ofmitochondrial PT. Mitochondriafiom Cells UndergoingApoptosis Transfer Nuclear Apoptosis to a Cell-freeSystem. According to several studies (8-10, 12), mitochondrial function is perturbed early during the apoptotic process. Accordingly, mitochondria isolated from hepatocytes exposed in vivo to an apoptosisinducing combination of GaIN and LPS (36, 53), but not control cells treated with GaIN or LPS only, display a reduced uptake of the cationic lipophilic dye 3,3' dihexyloxacarbocyanine iodide (DiOC6[3]) (Fig. 4 A), indicating a Mitochondrial Regulation of Apoptosis mitochondrial sO BA CsASDZMCBRR P ZnCI2AIF3t sO BA CsASDZMCBRR P ZnCI2AIF3s 0 Atractyloside sO BA CsASDZMCBRR P ZnCI2AIF3s ter-BHP ~, .O BA CO, SDZMCBRR P ZnCUAIF3 .0 BA CsASDZMCBRR P ZnCL~AIF3 . ~ Ca ~ BA CO, SDZMCBRR P ZnClaAIF~ mCICCP diamide Figure 3. Correlation between mitochondrial swelling and mitochondrial induction of nuclear apoptosis. Mitochondria from hepatocytes were incubated with the PT inducers Atr (5 re_M), ter-BHP (50 FtM), CaC12 (500 p,M), mCICCP (10 ~M), or diamide (100 ~M) and/or the PT inhibitors BA (50 I~M), CsA (10p~M), N-methylVal-4-CsA (SDZ; 10 I~M), MCB (30 ~M), or R R (100 ~M). Mitochondria were also incubated with P (10 mM), ZnC12 (lmM), or AIF3 (20 ~M). The percentage of condensed nuclei cocultured with mitochondria was recorded after 90 min of incubation at 37~ (open columns, X + SEM of triplicates). Large amplitude swelling was recorded after 5-90 min of culture (blackcolumns). Data are shown for 60 rain, when the correlation between nuclear apoptosis and mitochondrial swelling is optimal. These results are representative of five independent experiments. decrease o f the A~klfm. Such m i t o c h o n d r i a f r o m apoptotic liver cells cause n u c l e a r apoptosis i n vitro (Fig. 4 B). S i m i larly a fraction o f m i t o c h o n d r i a from splenocytes treated i n vivo w i t h the glucocorticoid analogue D E X (10, 37) display a r e d u c e d A ~ m (Fig. 5 A) a n d cause apoptosis o f isolated Hela nuclei in vitro (Fig. 5 B). Thus, m i t o c h o n d r i a from different cell types u n d e r g o i n g apoptosis in vivo are T a b l e 1. e n d o w e d w i t h the capacity o f apoptosis i n d u c t i o n in a ceUfree system. Inhibition of P T Inhibits Nuclear Apoptosis both In Vitro and In Vivo. T h e A N T ligand B A is the agent w i t h the broadest P T - i n h i b i t o r y s p e c t r u m a m o n g all substances tested thus far (Figs. 1 and 3) and is the o n l y P T i n h i b i t o r that is truly specific for a m i t o c h o n d r i a l structure. W e Substances that Fail to Modulate P T and Mitochondria-dependent Nuclear Apoptosis Substance Dose range Calpain inhibitor I Ca]pain inhibitor II ICE inhibitor I ICE inhibitor II AcDEVD-CHO GTP3~S Ionomycin 100 100 100 100 100 100 10 heM-1 m M IxM-1 m M heM-500 p.M p~M-500 p.M p~M-500 p~M p~M-i m M ~ M - 1 0 0 I-~M Inhibition of PT Inhibition of nuclear apoptosis None* None None None None None None None None None None None None None *The indicated substances were employed to modulate the induction of PT in isolated mitochondria induced by the following reagents: Atr, terBHP, calcium, m-C1CCP, and diamide (same conditions and concentrations as in Fig. 3). Absence of inhibition indicates <20% suppression of either mitochondrial swelling (measured at 60 min as in Fig. 3) or nuclear condensation (measured at 90 rain in the same conditions as in Fig. 3), in response to all tested PT inducers. 1537 Zamzami et al. Control 9= GaIN Figure 6. BA inhibits apoptosis of thymocytes induced by various stimuli. Murine thymocytes were exposed to DEX (1 I,M), etoposide (10 IzM), or cultured after y-irradiation (t0 Gy) in the presence or absence of BA (50 }*M). DNA fragmentation of thymocytes (106 cells/ lane) was monitored after a 4-h culture period. Results are representative of three independent experiments. LPS LPS/GalN , iogDiOC6(3) ~ ~ ~ ,~ Figure 4. Mitochondria from apoptotic hepatocytescause nuclear apoptosis in a cell-free system. (A) Reduction of the A~ m in mitochondria from cellsundergoing apoptosisin vivo. Liver mitochondria were obtained from animals treated with an apoptosis-inducing combination of GaIN and/or LPS, followed by DiOCt(3) labeling for A~tIf m a s s e s s m e n t . (B) Mitochondria from apoptotic hepatocytes induce nuclear apoptosis. After the indicated in vivo treatment, liver mitochondria were purified and added to HeLa nuclei in the presence or absence of BA (50 ~M), followed by evaluation ofchromatin condensation as in Fig. 1. Data are representative of two independent experiments. therefore tested the effect o f BA on the m i t o c h o n d r i a mediated transfer o f apoptosis from w h o l e cells undergoing P C D to the cell-free system. As shown in Figs. 4 A and 5 A, B A partially reduces the proapoptotic effect o f m i t o chondria from G a l N / L P S - s t i m u l a t e d hepatocytes o f D E X p r i m e d splenocytes in vitro. This inhibition is sigxtificant: 48 + 10% for G a l N / L P S - t r e a t e d hepatocyte mitochondria and 44 + 9% for D E X - p r i m e d splenocyte mitochondria. In addition, BA is highly efficient ( > 9 0 % inhibition) in preventing the death o f intact thymocytes exposed to a series o f different apoptosis inducers: D E X , irradiation, and topoisomerase inhibition (Fig. 6). These results corroborate the notion that mitochondria are indeed involved in the |' I log DIOC6(3) Figure 5. Transferof apoptosis by mitochondria from apoptotic splenocytes. (A) A~ m disruption of mitochondria from DEX-primed splenocytes. Mitochondria were purified from control splenocytes (co.)or from splenocytes exposed to the glucocorticoid DEX in vivo, followed by staining with DiOCt(3). (B) Mitochondria from apoptotic splenic lymphocytes induce nuclear apoptosis. Mitochondria from control or DEXprimed splenocyteswere mixed with HeLa nuclei either in the absence or in the presence of BA to assess their apoptosis-inducing potential as in Fig. 4 B. Splenic mitochondria were tested at a concentration of 250 ng/ I*1.Typical results out of three independent experiments are shown. 1538 apoptotic cascade in vivo and that mitochondrial P T is both sufficient and necessary to induce nuclear apoptosis. Bcl-2 Inhibits Apoptosis by Preventing Mitochondrial PT. T h e p r o t o - o n c o g e n e product Bcl-2 inhibits apoptosis in response to a n u m b e r o f different stimuli (for a review see reference 54) and prevents b o t h the mitochondrial and the nuclear manifestations o f apoptosis (12). Bcl-2 is localized in the mitochondrial outer membrane and endoplasmatic reticulum, as well as in nuclear membranes (55-57). W i t h i n the mitochondrion, it is found at the i n n e r - o u t e r m e m brane contact site, where P T pores are expected to form (20). T o map the antiapoptotic function o f Bcl-2 either to mitochondria or to nuclei, w e purified these organelles from hBcl-2-transfected murine T cell hybridoma cells (39), as well as from mock-transfected controls. R.econstitution experiments indicate that Atr-treated mitochondria 8 o m hBcl-2-transfected cells fail to provoke nuclear apoptosis (Fig. 7 A) in conditions in which mitochondria from vector-transfected cells (Fig. 7 A) or from hepatocytes constitutively lacking Bcl-2 expression (Figs. 1 and 3) do induce nuclear apoptosis. In contrast, nuclei from hBcl-2-transfected cells readily condense and fragment in the presence o f Atr and control mitochondria (Fig. 7, A and B). Thus, in accord with previous genetic (55, 57) and functional (4) studies, the mitochondrial but not the nuclear localization o f Bcl-2 is critical for its antiapoptotic function. In control experiments, mixtures o f Bcl-2-transfected and control m i tochondria induce apoptosis (Fig. 7 A), indicating that the B c l - 2 - m e d i a t e d inhibition o f apoptosis acts in cis and cannot be attributed to cytosolic Bcl-2 contaminating the m i tochondrial preparation. In addition, isolated mitochondria from Bcl-2-transfected cells are protected against A t r induced PT, i.e., they fail to undergo large amplitude swelling and A ~ m disruption in response to Atr (Fig. 7, C and D). Bcl-2 is a potent inhibitor o f some death pathways, including pro-oxidants (58), but is comparatively inefficient in preventing calcium-induced and antigen r e c e p t o r mediated P C D (38, 39, 59). W e therefore tested whether Bcl-2 w o u l d be a universal inhibitor o f P T or rather, whether it w o u l d have a selective effect. As shown in Fig. 8 A, Bcl-2 prevents large amplitude swelling o f isolated m i tochondria in response to M - C 1 C C P and ter-BHP, but not in response to calcium or diamide. These data underscore that different P T inducers obey different mechanisms; this is also suggested by experiments involving P T inhibitors Mitochondrial Regulation of Apoptosis F i g u r e 7. Mechanism of the antiapoptotic effect of Bcl-2. (A) Functional mapping of the site at which Bcl-2 acts to prevent Atr-induced apoptosis. Nuclei and mitochondria from Bcl-2- or Neo-transfected cells were cocultured in the presence or absence of Atr (5 raM), as indicated by black squares. After 90 rain ofcoculture, nuclei were stained with DAPI and analyzed for apoptotic morphology. (B) Representative nuclei from Bcl-2-transfected cells incubated with the indicated type of mitochondria and/or Atr (same experiment as A). (C) Bcl-2 directly inhibits the Atr-induced large amplitude swelling ofmitochondria. Mitochondria from Neo- or Bcl-2-transfected cells were monitored for large amplitude swelling (as in Fig. 1 A). (Arrows)Repeated addition of 2.5 mM Atr (final concentration 5 raM). (D) Bcl-2 inhibits the Atr-induced disruption of the mitochondrial transmembrane potential. Mitochondria were labeled with DiOC6(3), cultured for 5 min in the presence or absence of 5 mM Atr, and were then analyzed by cytofluorometry. (Fig. 3). T h e pattern o f the bcl-2 effect corresponds most closely to that o f BA, i.e., it inhibits P T i n d u c e d b y Atr (Figs. 3 and 7), m - C 1 C C P , and ter-BHP, b u t n o t calcium or diamide (Figs. 3 and 8). Again, as i n the case o f B A , Bcl2 - m e d i a t e d i n h i b i t i o n o f P T results in the abolition o f the apoptotic potential o f isolated m i t o c h o n d r i a . M o r e i m p o r tantly, the B c l - 2 - d r i v e n i n h i b i t i o n o f m i t o c h o n d r i a l swelli n g (Fig. 8 A) correlates w i t h its apoptosis-inhibitory p o tential in cells. Bcl-2 protects against apoptosis o f T cell h y b r i d o m a cells i n d u c e d b y m - C 1 C C P (Fig. 8 B) and oxidants such as H 2 0 2 (58), yet fails to confer p r o t e c t i o n against diamide (Fig. 8 /3) and C D 3 cross-linking (12, 38, Ali mCICCP _ . ~ u ~ [ ~ . _ ~ 39). Thus, Bcl-2 does n o t p r e v e n t apoptosis w h e n death is i n d u c e d via such agents as diamide (Fig. 8 B) against whose P T - i n d u c i n g potential it does n o t protect (Fig. 8 A). Again, these data are in accord w i t h the hypothesis that Bcl-2 prevents apoptosis b y virtue o f its P T - i n h i b i t o r y potential. A Soluble Factor Released from Mitochondria Undergoing P T Mediates Nuclear Disintegration. As s h o w n above, m i t o c h o n dria u n d e r g o i n g P T i n d u c e apoptotic n u c l e a r disintegration in a cell-free system. W h e r e a s some authors have s h o w n that m i t o c h o n d r i a are necessary to i n d u c e apoptosis in cellfree systems (4, 60), others have f o u n d that cytosolic (organelle-free) extracts m a y be suflficient to i n d u c e n u c l e a r t-BHP o 3o ~ 9o~2otso~so o 3o ~ ~ t 2 o t s m s o o 3o ,o 9o~z.~so~so o 3 o . time ( s e c ) B ~eo ~12.~.~. mCICCP Figure 8. IIBd-2 dlamlde Correlation of the antiapoptotic and the PT-inhibitory effect ofBcl-2. (A) Effect of Bcl-2 on large amplitude swelling of isolated rnitochon- dria. (Arrows)The indicated reagents were added (same concentrations as in Fig. 3), while absorbance at 540 nm was monitored. (B) Spectrum of Bcl-2mediated inhibition of apoptosis in whole cells. Bcl-2, or Neo-transfected T cell hybridoma cells were cultured with the indicated dose ofmC1CCP pr diamide for 6 or 24 h, respectively. The percentage of cells with nuclear hypoploidy was determined after ethanol fixation and staining with propidium iodine. 1539 Zamzami et al. 10oi e~l U't ~Ir ~ 20- lr/A I rill + u I SN from Atr-treated mitochondria Figure 9. Partial characterization of a proapoptotic activity released from Atr-treated mitochondria. Liver mitochondria were incubated in the presence or absence of 5 mM Atr, followed by ultracentrifugation (150,000 g, 30 min). Isolated HeLa nuclei were incubated in the presence of this supernatant to determine the frequency of cells exhibiting apoptotic morphology (90 rain, 37~ Supernatants were heat treated (70~ 5 min) or centrifuged through membranes with a molecular mass exclusion of~10 kD before the test. Alternatively, the antioxidants N-t-butyla-phenylnitrone (50 p.M), trolox (230 I~M), L-ascorbate (600 I~M), or catalase (1 mg/ml) were added to the assay. apoptosis in vitro (5, 42). Prompted by this apparent contradiction, we tested whether mitochondria undergoing PT would release a soluble proapoptotic factor. As shown in Fig. 9, mitochondria treated with Atr release (a) soluble factor(s) into the supernatant (150,000 g, 30 rain) that can induce chromatin condensation in isolated HeLa nuclei. This activity is heat sensitive (70~ 5 rain), has a molecular mass > 1 0 kD, and is not neutralized by antioxidants such as N-t-butyl-cx-phenylnitrone or the water-soluble vitamine E analogue trolox (Fig. 9). In conclusion, at least part of the apoptotic activity of mitochondria is mediated by one or several proteins and does not involve R.OS. P T dependent release of proteins from mitochondria has been reported previously (61). Concluding Remarks As shown in this article, mitochondria from hepatic, m y elomonocytic, or lymphoid cells induce nuclear apoptosis, provided that they undergo PT. Modulation of P T determines the apoptosis-inducing effect of mitochondria in a cell-free system. Moreover, inhibition of P T by BA, a specific ligand of one P T pore constituent, reduces naturally occurring apoptosis, and Bcl-2 apparently functions as an endogenous P T inhibitor. Although these findings establish mitochondrial P T as a critical event in early apoptosis, they do not resolve a number of issues concerning the cellular biology of apoptosis. According to studies performed in Caenorhabditis elegans, at least two gene products, ced-3, which encodes a cysteine protease, and ced-4, whose function is unknown, are required for apoptosis to occur (62). At present, the sequence 1540 of events that eventually link ced-3-like proteases and ced-4 to mitochondria remains unknown. At present, it appears clear that both Bcl-2 (which controls PT; Figs. 7 and 8) and protease activation control two checkpoints of the apoptotic cascade (63). Tetrapeptide inhibitors of the ced-3 homologue C P P 3 2 / Y a m a and of ICE fail to interfere with the induction of P T in isolated mitochondria. Moreover, they fail to inhibit the mitochondria-mediated induction of nuclear apoptosis (Table 1). W h e n thymocyte apoptosis is induced by Fas/CD95 cross-linking, inhibition of ICE prevents both the nuclear manifestations of apoptosis and the A~r~ disruption (Marchetti, P., and G. Kroemer, unpublished results). This may indicate that at least some of the members of the family of ced-3-1ike proteases regulate events that are upstream of mitochondria. At present, h o w ever, our data cannot distinguish between two alternative possibilities. First, the PT and the protease-regulated checkpoints of the apoptotic effector phase could be placed in a serial (hierarchical) fashion. Second, both protease activation and P T could form part of parallel pathways culminating in nuclear apoptosis. It remains largely u n k n o w n h o w Bcl-2 regulates PT on the molecular level. Bcl-2 does not prevent P T as such; it prevents the induction of P T by determined stimuli such as Atr, mC1CCP, and ter-BHP, but not calcium or diamide (Figs. 7 and 8). Bcl-2 could act via direct molecular association with constituents of the P T pore, a possibility that is suggested by the localization of both Bcl-2 and P T pore constituents at inner-outer membrane contact sites (5557). Alternatively, Bcl-2 could affect P T indirectly. Thus, it enhances oxidative phosphorylation (64) and causes mitochondrial inner membrane hyperpolarization (65), which in turn would reduce the probability of P T (24). It has previously been reported that mitochondrial membrane localization is necessary to mediate Bcl-2 suppression o f a p o p t o sis, namely when apoptosis is induced by EIB-defective adenovirus (57) and when it is triggered by IL-3 starvation of IL-3-dependent 32D cells (55). In contrast, in some other systems of apoptosis induction, a mutated Bcl-2 molecule lacking the membrane localization domain (4, 58), as well as the naturally occurring apoptosis-inhibitory Bcl-2 analogue B c l - X A T M (a splice variant of Bcl-X that lacks the transmembrane domain; 66), maintain their antiapoptotic potential. However, the fact that soluble, ubiquitous Bcl-2 still maintains at least part of a its antiapoptotic function does not formally exclude that it acts on the external membrane of mitochondria. T h e present data suggest an intimate linkage between Bcl-2 and mitochondrial regulation. In this context it may be intriguing that the C. etegans bcl-2 homologue, ced-9, is an element of a polycistronic locus that also contains cyt-1, a gene that encodes a protein similar to cytochrome b560 of the mitochondrial respiratory chain complex II (67). Thus both functional and genetic evidence link Bcl-2 to mitochondrial regulation. Irrespective of the exact molecular mechanism by which Bcl-2 affects PT, the finding that Bcl-2 does inhibit PT, at least in response to certain stimuli (Figs. 7 and 8), provides an explanation for hitherto apparently contradictory reports. Mitochondrial Regulation of Apoptosis Bcl-2 hyperexpression has been reported to inhibit the production and/or adverse effects o f 1LOS (58, 68), that in turn, however, are not obligatory for apoptosis (16). In accord with these findings, Bcl-2 prevents oxidant-mediated P T (Fig. 8). Moreover, it prevents the mitochondrial R O S formation that is secondary to P T (12). Thus, Bcl-2 impedes P T as well as two dissociable consequences o f P T : (a) nuclear apoptosis, and (b) mitochondrial uncoupling and superoxide anion generation. A further issue that remains to be elucidated is the m o lecular mechanism by which isolated mitochondria undergoing PT cause nuclear chromatin condensation and endonuclease activation. It appears clear that this mechanism is neither cell type nor species specific, given that, for example, mouse liver mitochondria in P T can promote the apoptotic disintegration of nuclei purified from human fibroblast-like nuclei (Fig. 1). O u r data indicate that mitochondria contain or are associated with (a) pre-formed soluble mediator(s) > 1 0 k D that is/are released after P T and that alone is/are sufficient to cause nuclear apoptosis (Fig. 9). In accord with published experiments performed on intact cells (16, 17), antioxidants do not neutralize this apoptosis inducer (Fig. 9). Thus, R O S that are formed by mitochondria after P T do not participate in the induction o f nuclear apoptosis; this is also indicated by experiments involving p~ cells that lack a functional respiratory chain (15, and Fig. 2). Moreover, it appears improbable that Ced-3-like proteases would be responsible for this apoptosis-inducing activity, given that the mammalian Ced-3 analogue CPP32 per se is not sufficient to induce nuclear apoptosis in a cell-free system (6). Thus, the molecular events linking mitochondrial P T to nuclear apoptosis await further characterization. From the available data, it appears that AXI/m disruption, which presumably is mediated by PT, is a constant feature o f early apoptosis (8-14). Indirect biochemical evidence has previously accused P T to participate in the postischemic or toxin-mediated death o f myocardial cells and hepatocytes (69-72), thus again suggesting that P T is a general regulator of cell death. Indeed, the P T pore is an attractive candidate for a death switch that, once activated, marks a point of no return in PCD. At least six reasons support this concept. First, as shown here, P T is both necessary and sufficient to cause nuclear apoptosis. Second, opening o f P T pores entails multiple potentially lethal alterations of mitochondrial function (loss o f A ~ m, uncoupling of the respiratory chain, hypergeneration o f R O S , and loss of mitochondrial glu- tathione and calcium; 12, 19-21) and thus may initiate pleiotropic death pathways. Moreover, as shown here, P T triggers a nuclear apoptosis effector pathway whose biochemical components remain elusive. Third, the P T pore functions as a sensor for multiple physiological effectors (divalent cations, ATP, ADP, N A D , A@m, pH, thiols, and peptides), thereby integrating information on the electrophysiological, redox, and metabolic state of the cell (19, 20, 73, and Fig. 3). Thus, different death inducers can converge at this level. Fourth, given that a P T pore constituent such as the A N T is essential for energy metabolism, mutations in this apoptosis-regulatory device will be mostly lethal for the cell. In teleological terms, this would have the advantage o f precluding apoptosis-inhibitory (oncogenic) mutations at this level of the apoptotic cascade. Fifth, at least one of the P T constituents, the A N T , is encoded by several members of a gene family that are expressed in a strictly tissue-specific manner (74). Thus, P T pores may be regulated in each cell type in a slightly different fashion. Sixth, P T is endowed with self-amplificatory properties in the sense that loss of matrix Ca 2+ and glutathione, depolarization of the inner membrane, and increased oxidation o f thiols, that result from P T pore opening, all increase the P T pore-gating potential (19-21, 23-29). T h e self-amplificatory property of P T is also underscored by the data presented in this paper. Thus, induction of PT induces A ~ m disruption (Figs. 1 and 7) and, conversely, AxIfm depolarization by mC1CCP causes PT, measured as large amplitude swelling (Figs. 3 and 8). Similarly, oxidant treatment causes P T (Figs. 3 and 8), and P T will ultimately entail mitochondrial generation o f tLOS (12). The fact that some consequences o f P T (e.g., AW m dissipation, R O S generation) themselves may cause P T suggests that P T may engage in a positive feedback loop that contributes to apoptotic autodestruction. Thus, P T would have to respond in an allor-nothing fashion and, once activated, would seal the cell's fate in an irreversible fashion. Accordingly, cells exhibiting an immediate consequence of PT, that is AXI'?m r e d u c t i o n , are irreversibly committed to cell death (10). Apart from these theoretical considerations, the current data suggest that the P T pore occupies a central position in apoptosis regulation. It therefore becomes an attractive target for regulation by pharmacological agents, as well as by endogenous apoptosis regulators belonging to the everexpanding Bcl-2 gene family. We are indebted to Dr. J.A. Duine for the gift of BA; Dr. Javier Naval (University of Zaragoza, Zaragoza, Spain) for p~ cells; Dr. Jonathan Ashwell for Bcl-2-transfected cells; and Dr. Roland Wenger for SDZ 220384. This work was supported by Association pour la Kecherche sur le Cancer, Agence Nationale pour la Recherche sur le SIDA, CNRS, Fondation pour la Recherche M~dicale, Institut National de la Sant~ et de la Recherche M~dicale, NATO, and the Leo Foundation (G. Kroemer). N. Zamzami and S.A. Susin received fellowships from Institut Scientifique Roussel and the Spanish Government, respectively. 1541 Zamzami et al. Address correspondence to Dr. Guido Kroemer, CNRS-UPR420, 19, rue Guy M6quet, B.P.8, F-94801 Villejuif, France. Received for publication 21 September 1995 and in revised form 6 December 1995. References 1. Jacobson, M.D., J.F. Burne, and M.C. Raft. 1994. Programmed cell death and Bcl-2 protection in the absence of a nucleus. E M B O (Eur. Mol. Biol. Organ.)J. 13:1899-1910. 2. Schulze-Osthoff, K., H. Walczak, W. Droge, and P.H. Krammer. 1994. Cell nucleus and DNA fragmentation are not required for apoptosis.J. Cell Biol. 127:15-20. 3. Nakajima, H., P. Golstein, and P.A. Henkart. 1995. The target cell nucleus is not required for cell-mediated granzymeor Fas-based cytotoxicity.J. Exp. Med. 181:1905-1909. 4. Newmeyer, D.D., D.M. Farschon, and J.C. Reed. 1994. Cell-free apoptosis in xenopus egg extracts: inhibition by Bcl-2 and requirement for an organelle fraction enriched in mitochondria. Cell. 79:353-364. 5. Lazebnik, Y.A., S.H. Kaufmann, S. Desnoyers, G.G. Poirier, and W.C. Earnshaw. 1994. Cleavage of poly(ADP-ribose) polymerase by a proteinase with properties like ICE. Nature (Lond.). 371:346-347. 6. Nicholson, D.W., A. All, N.A. Thornberry, J.P. Vaillancourt, C.K. Ding, M. Gallant, Y. Gareau, P.R. Griffin, M. Labelle, Y.A. Lazebnik, et al. 1995. Identification and inhibition of the ICE/CED-3 protease necessary for mammalian apoptosis. Nature (Lond.). 376:37-43. 7. Martin, S.J., and D.R. Green. 1995. Protease activation during apoptosis: death by a thousand cuts? Cell. 82:349-352. 8. Deckwerth, T.L., and E.M. Johnson. 1993. Temporal analysis of events associated with programmed cell death (apoptosis) of sympathetic neurons deprived of nerve growth factor. J. Cell Biol. 123:1207-1222. 9. Vayssi~re, J.-L., P.X. Petit, Y. Risler, and B. Mignotte. 1994. Commitment to apoptosis is associated with changes in mitochondrial biogenesis and activity in cell lines conditionally immortalized with simian virus 40. Proc. Natl. Acad. Sci. USA. 91:11752-11756. 10. Zamzami, N., P. Marchetti, M. Castedo, C. Zanin, J.-L. Vayssi~re, P.X. Petit, and G. Kroemer. 1995. Reduction in mitochondrial potential constitutes an early irreversible step of programmed lymphocyte death in vivo. J. Exp. Med. 181: 1661-1672. 11. Petit, P.X., H. LeCoeur, E. Zorn, C. Duguet, B. Mignotte, and M.L. Gougeon. 1995. Alterations ofmitochondrial structure and function are early events of dexamethasone-induced thymocyte apoptosis.J. Cell Biol. 130:157-167. 12. Zamzami, N., P. Marchetti, M. Castedo, D. Decaudin, A. Macho. T. Hirsch, S.A. Susin, P.X. Petit, B. Mignotte, and G. Kroemer. 1995. Sequential reduction of mitochondrial transmembrane potential and generation of reactive oxygen species in early programmed cell death. J. Exp. Med. 182: 367-377. 13. Castedo, M., A. Macho, N. Zamzami, T. Hirsch, P. Marchetti, J. Uriel, and G. Kroemer. 1995. Mitochondrial perturbations define lymphocytes undergoing apoptotic depletion in vivo. Eur. J. Immunol. 25:3277-3284. 14. Macho, A., M. Castedo, P. Marchetti, J.J. Aguilar, D. Decaudin, N. Zamzami, P.M. Girard, J. Uriel, and G. Kroemer. 1995. Mitochondrial dysfunctions in circulating T lympho1542 cytes from human immunodeficiency virus-1 carriers. Blood. 86:2481-2487. 15. Jacobson, M.D., J.F. Burne, M.P. King, T. Miyashita, J.C. Reed, and M.C. Raft. 1993. Bcl-2 blocks apoptosis in cells lacking mitochondrial DNA. Nature (Lond.). 361:365-369. 16. Jacobson, M.D., and M.C. Raft. 1995. Programmed cell death and Bcl-2 protection in very low oxygen. Nature (Lond.). 374:814-816. 17. Shimizu, S., Y. Eguchi, H. Kosaka, W. Kamlike, H. Matsuda, and Y. Tsujimoto. 1995. Prevention of hypoxiainduced cell death by Bcl-2 and Bcl-xL. Nature (Lond.). 374: 811-813. 18. Kroemer, G., P.X. Petit, N. Zamzami, J.-L. Vayssi~re, and B. Mignotte. 1995. The biochemistry ofapoptosis. FASEB (Fed. Am. Soc. Exp. Biol.)J. 9:1277-1287. 19. Bernardi, P., K.M. Broekemeier, and D.R. Pfeiffer. 1994. Recent progress on regulation of the mitochondrial permeability transition pore; a cyclosporin-sensitive pore in the inner mitochondrial membrane. J. Bioenerg. Biomembr. 26:509517. 20. Zoratti, M., and I. Szab6. 1994. Electrophysiology of the inner mitochondrial membrane.J. Bioenerg. Biomembr. 26:543553. 21. Reed D.J., and M.K. Savage. 1995. Influence of metabolic inhibitors on mitochondrial permeability transition and glutathione status. Biochim. Biophys. Acta. 1271:43-50. 22. Zoratti, M., and I. Szab6, 1995. The mitochondrial permeability transition. Biochim. Biophys. Acta - Rev. Biomembranes. 1241:139-176. 23. Bemardi, P., S. Vassanelli, P. Veronese, R. Colonna, I. Szab6, and M. Zoratti. 1992. Modulation of the mitochondrial permeability transition pore. Effect of protons and divalent cations.J. Biol. Chem. 267:2934-2939. 24. Bemardi, P. 1992. Modulation of the mitochondrial cyclosporin A~ensitive permeability transition pore by the proton electrochemical gradient. J. Biol. Chem. 267:8834-8839. 25. Szab6, I., P. Bernardi, and M. Zoratti. 1992. Modulation of the mitochondrial megachannel by divalent cations and protons.J. Biol. Chem. 267:2940-2946. 26. Petronilli, V., C. Cola, S. Massari, R. Colonna, and P. Bernardi. 1993. Physiological effectors modify voltage-sensing by the cyclosporin A~sensitive mitochondrial permeability transition pore.J. Biol. Chem. 268:21939-21945. 27. Bernardi, P., P. Veronese, and V. Petronilli. 1993. Modulation of the mitochondrial cyclosporin A-sensitive permeability transition pore. I. Evidence for two separate Me 2+ binding sites with opposing effects on the pore open probability. J. Biol. Chem. 268:1005-1010. 28. Petronilli, V., C. Cola, and P. Bernardi. 1993. Modulation of the mitochondrial cyclosporin A-sensitive permeability transition pore. II. The minimal requirements for pore induction underscore a key role for transmembrane electric potential, matrix pH, and matrix Ca2+. J. Biol. Chem. 268:1011-1016. 29. Petronilli, V., A. Nicolli, P. Costantini, R. Colonna, and P. Bernardi. 1994. Regulation of the permeability transition Mitochondrial Regulation of Apoptosis pore, a voltage-dependent mitochondrial channel inhibited by cyclosporin A. Biochim. Biophys. Acta. 1187:255-259. 30. McEnery, M.W., A.M. Snowman, R.R. Trifiletti, and S.H. Snyder. 1992. Isolation of the mitochondrial benzodiazepine receptor: association with the voltage-dependent anion channel and the adenine nucleotide carrier. Proc. Natl. Acad. Sci. USA. 89:3170-3174. 31. Klingenberg, M. 1980. The ADP-ATP translocation in mitochondria, a membrane potential controlled transportJ. Membr. Biol. 56:97-105. 32. Toninello, A., D. Siliprandi, and N. Siliprandi. 1984. On the mechanism by which Mg2+ and adenine nucleotides restore membrane potential in rat liver mitochondria deenergized by Ca 2+ and phosphate. Biochem. Biophys. Res. Commun. 111: 792-797. 33. Halestrup, A.P., and A.M. Davidson. 1990. Inhibition of CaZ+-induced large-amplitude swelling of liver and heart rnitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial-matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenine nucleotide translocase. Biochem.J. 268:153-160. 34. Marry, l., G. Brandolin, J. Gagnon, R. Brasseur, and P.V. Vignais. 1992. Topography of the membrane-bound ADP/ ATP carrier assessed by enzymatic proteolysis. Biochemistry. 31:4058-4065. 35. Majima, E., Y. Shinohara, N. Yamaguchi, Y.M. Hong, and H. Terada. 1994. Importance of loops of mitochondrial ADP/ATP carrier for its transport activity deduced from reactivities of its cysteine residues with the sulflaydril reagent eosin-5-maleimide. Biochemistry. 33:9530-9536. 36. Leist, M., F. Gantner, I. Bohlinger, P.G. Germann, C. Tiegs, and A. Wendel. 1994. Murine hepatocyte apoptosis induced in vitro and in vivo by TNF-alpha requires transcriptional arrest.J. Immunol. 153:1778-1788. 37. Gonzalo, J.A., A. Gonzklez-Garcia, C. Martinez-A., and G. Kroemer. 1993. Glucocorticoid-mediated control of the clonal deletion and activation of peripheral T cells in vivo. J. Exp. Med. 177:1239-1246. 38. Green, D.R., A. Mahboubi, W. Nishioka, S. Oja, F. Echeverri, Y. Shi, J. Glynn, Y. Yang, J. Ashwell, and R. Bissonnette. 1994. Promotion and inhibition of activation-induced apoptosis in T-cell hybridomas by oncogenes and related signals. Immunol. Rev. 142:321-342. 39. Aten, J., P. Prigent, P. Poncet, C. Blanpied, N. Claessen, P. Druet, and F. Hirsch. 1995. Mercuric chloride-induced programmed cell death ofa murine T cell hybridoma: I. Effect of the proto-oncogene Bcl-2. Cell. Immunol. 161:98-106. 40. Lumbach, G.W.M., H.C. Cox, and W. Berends. 1970. Elucidation of the chemical structure of bongkrekic acid. I. Isolation, purification and properties of bongkrekic acid. Tetrahedron. 26:5993-5999. 41. Wood, E.R., and W.C. Earnshaw. 1990. Mitotic chromatin condensation in vitro using somatic cell extracts and nuclei with variable levels of endogenous topoisomerase II. J. Cell Biol. 111:2839-2850. 42. Lazebnik, Y.A., S. Cole, C.A. Cooke, W.G. Nelson, and W.C. Earnshaw. 1993. Nuclear events ofapoptosis in vitro in celt-free mitotic extracts: a model system for analysis of the active phase ofapoptosis.J. Cell Biol. 123:7-22. 43. Boutry, M., and M. Briquet. 1982. Mitochondrial modifications associated with the cytoplasmic male sterility in faba beans. Eur.J. Biochem. 127:129-135. 44. Kawabe, Y., and A. Ochi. 1991. Programmed cell death and 1543 Zamzami et al. extrathymic reduction of V[38 + CD4 + T cells in mice tolerant to Staphylococcus aureus enterotoxin B. Nature (Lond.). 349: 245-248. 45. Petit, P.X., J.E. O'Connor, D. Grunwald, and S.C. Brown. 1990. Analysis of the membrane potential of rat- and mouseliver mitochondria by flow cytometry and possible applications. Eur. J. Biochem. 389-397. 46. Nicoletti, I., G. Migliorati, M.C. Pagliacci, and C. Riccardi. 1991. A rapid simple method for measuring thymocyte apoptosis by propidium iodide staining and flow cytometry. J. Immunol. Meth. 139:271-280. 47. Brandolin, G., A. Le-Saux, V. Trezeguet, G.J. Lauquinn, and P.V. Vignais. 1993. Chemical, immunological, enzymatic, and genetic approaches to studying the arrangement of the peptide chain of the ADP/ATP carrier in the mitochondrial membrane.J. Bioenerg. Biomembr. 25:493-501. 48. Bellomo, G., A. Martino, P. Richelmi, G.A. Moore, S.A. Jewell, and S. Orrenius. 1984. Pyridine-nucleotide oxidation, Ca 2+ cycling and membrane damage during tert-butylhydroperoxide metabolism by rat liver mitochondria. Eur. J. Biochem. 140:1~. 49. Petronilli, V., P. Costantini, L. Scorrano, R. Colonna, S. Passamonti, and P. Bernardi. 1994. The voltage sensor of the mitochondrial permeability transition pore is tuned by the oxidation-reduction state ofvicinal thiols. Increase of the gating potential by oxidants and its reversal by reducing agents. .]. Biol. Chem. 269:16638-16642. 50. Costantini, P., B.V. Chernyak, V. Petronilli, and P. Bernardi. 1995. Selective inhibition of the mitochondrial permeability transition pore at the oxidation-reduction sensitive dithiol by monobromobimane. FEBS (Fed. Eur. Biochem. Sot.) Lett. 362:239-242. 51. Zenke, G., G. Baumann, R. Wenger, O. Hiestand, V. Quesniaux, E. Andersen, and M.H. Schreier. 1993. Molecular mechanisms of immunosuppression by cyclosporins. Ann. N.Y. Acad. ScL 23:330-335. 52. Broekemeier, K.M., R.J. Krebsbach, and D.R. Pfeiffer. 1994. Inhibition of the mitochondrial Ca 2+ uniporter by pure and impure ruthenium red. Mol. Cell. Biochem. 139:33-40. 53. Leist, M., F. Gantner, I. Bohlinger, G. Tiegs, P.G. Germann, and A. Wendel. 1995. Tumor necrosis factor-induced hepatocyte apoptosis precedes liver failure in experimental murine shock models. Am]. Pathol. 146:1220-1234. 54. Gory, S. 1995. Regulation of lymphocyte survival by the Bcl-2 gene family. Ann. Rev. Immunol. 13:513-543. 55. Tanaka, S., K. Saito, andJ.C. Reed. 1993. Structure-function analysis of the Bcl-2 oncoprotein. Addition ofa heterologous transmembrane domain to portions of the Bcl-2[3 protein restores function as a regulator of cell survival. J. Biol. Chem. 268:10920-10926. 56. Krajewski, S., S. Tanaka, S. Takayama, M.J. Schibler, W. Fenton, andJ.C. Reed. 1993. Investigation of the subcellular distribution of the bcl-2 oncoprotein: residence in the nuclear envelope, endoplasmic reticulum, and outer mitochondrial membranes. Cancer Res. 53:4701-4714. 57. Nguyen, M., P.E. Branton, P.A. Walton, Z.N. Oltvai, S.J. Korsmeyer, and G.C. Shore. 1994. Role of membrane anchor domain of Bcl-2 in suppression of apoptosis caused by E1B-defective adenovirus.-]. Biol Chem. 269:16521-16524. 58. Hockenbery, D.M., Z.N. Oltavai, X.-M. Yin, C.L. Milliman, and S.J. Korsmeyer. 1993. Bcl-2 functions in an antioxidant pathway to prevent apoptosis. Cell. 75:241-251. 59. Caron-Leslie, L.A.M., R.B. Evans, andJ.A. Cidlowski. 1994. Bcl-2 inhibits glucocorticoid-induced apoptosis but only partially blocks calcium ionophore or cycloheximide-regulated apoptosis in $49 cells. FASEB (Fed. Am. Soc. Exp. Biol.)J. 8: 639-645. 60. Martin, S.J., D.D. Newmeyer, S. Mathisa, D.M. Farschon, H.G. Wang, J.C. Reed, R.N. Kolesnick, and D.R. Green. 1995. Cell-free reconstitution ofFas-, UV radiation- and ceramide-induced apoptosis. E M B O (Eur. Mol. Biol. Organ.) J. 14:5191-5200. 61. Igbavboa, U., C.W. Zwizinski, and D.R. Pfeiffer. 1989. Release of mitochondrial matrix proteins through a Ca2+requiring, cyclosporin-sensitive pathway. Biochem. Biophys. Res. Commun. 161:619-625. 62. Ellis, R.E., Y. Yuan, and H.R. Horvitz. 1991. Mechanisms and functions of cell death. Annu. Rev. Cell. Biol. 7:663--698. 63. Oltvai, Z.N., and S.J. Korsmeyer. 1994. Checkpoints of dueling dimers foil death wishes. Cell. 79:189-192. 64. Smets, L.A., J. Van den Berg, D. Acton, B. Top, H. van Rooij, and M. Verwijs-Janssen. 1994. BGL-2 expression and mitochondrial activity in leukemic cells with different sensitivity to glucocorticoid-induced apoptosis. Blood. 5:16131619. 65. Hennet, T., C. Richter, and E. Peterhans. 1993. Tumour necrosis factor-alpha induces superoxide anion production in mitochondria ofL929 cells. Biochem.J. 289:587-592. 66. Fang, W., J.J. Rivard, D.L. Mueller, and T.W. Behrens. 1994. Cloning and molecular characterization of mouse bcl-x in B and T lymphocytes.J. Immunol. 153:4388-4398. 67. Hengartner, M.O., and H.R. Horvitz. 1994. C. elegans cell survival gene ced-9 encodes a functional homolog of the 1544 mammalian proto-oncogene bcl-2. Cell. 76:665-676. 68. Kane, D.J., T.A. Sarafian, R. Anton, H. Hahn, E.B. Gralla, J.S. Valentine, T. Ord, and D.E. Bredesen. 1993. Bcl-2 inhibition of neural death: decreased generation of reactive oxygen species. Science (Wash. DC). 262:1274-1277. 69. Crompton, M., H. Ellinger, and A. Costi. 1988. Inhibition by cyclosporin A of a CaZ+-dependent pore in heart mitochondria activated by inorganic phosphate and oxidative stress. Biochem. J. 255:357-360. 70. Andreeva, L., A. Tanveer, and M. Crompton. 1995. Evidence for the involvement of a membrane-associated cyclosporin-A-binding protein in the Ca2+-activated inner membrane pore of heart mitochondria. Eur. J. Biochem. 230: 1125-1132. 71. Mehendale, H.M., R.A. Roth, A.J. Gandolfi, J.E. Klaunig, J.J. Lemasters, and L.R. Curtis. 1994. Novel mechanisms in chemically induced hepatotoxicity. FASEB (Fed. Am. Soc. Exp. Biol.)J. 8:1285-1295. 72. Pastorino, J,G., J.W. Snyder, J.B. Hoek, and J.L. Farber. 1995. Ca 2+ depletion prevents anoxic death of hepatocytes by inhibiting mitochondrial permeability transition. Am J. Physiol. 268:C676-685. 73. Pfeiffer, D.R., T.I. Gudz, S.A. Novgorodov, and W.L. Erdahl. 1995. The peptide mastoparan is a potent facilitator of the mitochondrial permeability transition. J. Biol. Chem, 270: 4923-4932. 74. Walker, J.E., and M.J. Runswick. 1993. The mitochondrial transport protein superfamily. J. Bioenerg. Biomembr. 5:435446. Mitochondrial Regulation of Apoptosis