The central executioner of apoptosis: multiple connections between protease activation and mitochondria in Fas/APO-1/CD95- and ceramide-induced apoptosis

Abstract

According to current understanding, cytoplasmic events including activation of protease cascades and mitochondrial permeability transition (PT) participate in the control of nuclear apoptosis. However, the relationship between protease activation and PT has remained elusive. When apoptosis is induced by cross-linking of the Fas/APO-1/CD95 receptor, activation of interleukin-1beta converting enzyme (ICE; caspase 1) or ICE-like enzymes precedes the disruption of the mitochondrial inner transmembrane potential (DeltaPsim). In contrast, cytosolic CPP32/ Yama/Apopain/caspase 3 activation, plasma membrane phosphatidyl serine exposure, and nuclear apoptosis only occur in cells in which the DeltaPsim is fully disrupted. Transfection with the cowpox protease inhibitor crmA or culture in the presence of the synthetic ICE-specific inhibitor Ac-YVAD.cmk both prevent the DeltaPsim collapse and subsequent apoptosis. Cytosols from anti-Fas-treated human lymphoma cells accumulate an activity that induces PT in isolated mitochondria in vitro and that is neutralized by crmA or Ac-YVAD.cmk. Recombinant purified ICE suffices to cause isolated mitochondria to undergo PT-like large amplitude swelling and to disrupt their DeltaPsim. In addition, ICE-treated mitochondria release an apoptosis-inducing factor (AIF) that induces apoptotic changes (chromatin condensation and oligonucleosomal DNA fragmentation) in isolated nuclei in vitro. AIF is a protease (or protease activator) that can be inhibited by the broad spectrum apoptosis inhibitor Z-VAD.fmk and that causes the proteolytical activation of CPP32. Although Bcl-2 is a highly efficient inhibitor of mitochondrial alterations (large amplitude swelling + DeltaPsim collapse + release of AIF) induced by prooxidants or cytosols from ceramide-treated cells, it has no effect on the ICE-induced mitochondrial PT and AIF release. These data connect a protease activation pathway with the mitochondrial phase of apoptosis regulation. In addition, they provide a plausible explanation of why Bcl-2 fails to interfere with Fas-triggered apoptosis in most cell types, yet prevents ceramide- and prooxidant-induced apoptosis.

Figure 1

Chronology and cause effect relationship between activation of ICE (or ICE-like) protease(s) and ΔΨ_m disruption. (A) Chronology of the activation of ICE, ΔΨ_m disruption, and nuclear DNA fragmentation in human CEM-C7.H2 lymphoma cells subjected to Fas cross-linking. The frequency of ΔΨ_m ^low cells and of cells exhibiting DNA strand breaks were determined by double staining with the potential-sensitive dye CMXRos and TdT-catalyzed FITC-dUTP incorporation (TUNEL method), as described in Materials and Methods. Note that the TUNEL⁺ population is actually a subset of CMXRos^low cells (see B). Activation of ICE (-like) protease(s) was determined by a fluorogenic substrate containing the ICE cleavage site YVAD (filled symbols), the maximum activity being defined as 100%. Similarly, the activation of CPP32 (-like) protease(s) was determined by means of a fluorogenic substrate containing the cleavage site DEVD (open symbols). (B) Temporal relationship between Fas-induced ΔΨ_m disruption and CPP32 cleavage, as well as DEVDase activation. CEM-C7.H2 cells were cultured during 120 min in the presence of anti-Fas antibody, followed by staining with the ΔΨ_m-sensitive dye DiOC₆(3) plus Annexin V (revealed by phycoerythrin). Cells were then separated in the cytofluorometer into cells with a normal ΔΨ_m (DiOC₆(3)^high Annexin V⁻) or cells with a DiOC₆(3)^low Annexin V⁻ or DiOC⁶(3)^low Annexin V⁺ phenotype (sorting according to Windows), followed by determination of CPP32 cleavage using Western blots (lane 1, unstimulated control cells; lane 2, nonseparated Fas-stimulated cells; lane 3), purified DiOC⁶(3)^high cells; lane 4, purified DiOC₆(3)^low Annexin V⁻ cells; lane 5, purified DiOC⁶(3)^low Annexin V⁺ cells, 8 × 10⁵ cells/lane). Alternatively, cytosols from these cell populations were tested for DEVDase activity in vitro as in A (C) Determination of ΔΨ_m disruption and DNA strand breaks in different cells. CEM-C7.H2 lymphoma cell stably transfected with a Neomycin selection vector (Neo) only (fluorescence displays 1–4), with the crmA cowpox protease inhibitor (graphs 5 and 6), or with a Bcl-2–expressing construct negatively regulated by doxycyclin (graphs 7–12). Cells were either pretreated with doxycyclin (10 ng/ml, 48 h before starting of the experiment) to repress Bcl-2 expression (Bcl-2⁻, graphs 7–9) or left untreated (Bcl-2 ⁺, graphs 10–12), and then subjected to apoptosis induction with C₂ ceramide (50 μM; graphs 9 and 12), anti-Fas (graphs 3, 4, 6, 8, and 11) and/or the ICE inhibitor Ac-YVAD.cmk (50 μM, all during 4 h; graph 4), followed by double staining with CMXRos and the TUNEL method. Neo control cells were treated during 15 min with 100 μM of the protonophore mClCCP, providing a negative control for the CMXRos staining (graph 2). Numbers indicate the percentage of cells in each quadrant. Results are representative for three independent experiments.

Figure 2

A cytosolic factor neutralized by ICE-specific protease inhibitors causes mitochondrial PT in vitro. Isolated hepatocyte mitochondria were exposed to cytosols (final concentration: 100 μg/ml protein) prepared from CEM-C7.H2 lymphoma cells stably transfected with a Neomycin selection vector (Neo) only (graphs 1–4) or cells transfected with the crmA cowpox protease inhibitor (graphs 5 and 6) that were either treated with anti-Fas antibody during 30 min (graphs 2–4, 6) or were left untreated (graphs 1 and 5). Cytosols were tested for their capacity to induce mitochondrial swelling, 100% of swelling being defined as the loss of the OD₅₄₀ observed 5 min after addition of 5 mM atractyloside (A). Arrows indicate addition of the cytosolic extract. Alternatively, the ΔΨ_m was assessed cytofluorometrically on a per-mitochondrion basis of mitochondria treated with the indicated cytosol preparation and then stained with the potential-sensitive dye DiOC₆(3) (B). Treatment with the protonophore mClCCP served as a negative control for DiOC₆(3) staining (dotted line, graph 1 B). The effect of the ICE-specific inhibitor Ac-YVAD.cmk was tested in two different ways. Ac-YVAD.cmk was either used with the cells exposed to αFas (Ac-YVAD.cmk+αFas, graph 3) or, alternatively, was added to the cytosol prepared from Fas-treated cells (αFas+Ac−YVAD.cmk, graph 4).

Figure 3

Recombinant purified ICE is sufficient to induce PT, as well as the release of an apoptosis-inducing factor from mitochondria. Purified liver mitochondria were treated with CFS buffer only (graph 1), purified recombinant human ICE (graphs 2–4), the prooxidant t-BHP (graphs 5 and 6), and/or different protease inhibitors (Ac-YVAD.cmk, graphs 3 and 6 or Ac-DEVD.CHO, graph 4). These reagents were added together to the mitochondria and the following parameters were assessed: large amplitude swelling (A), ΔΨ_m (DiOC⁶(3) staining, 30 min after addition of the reagents) (B), and release of AIF (C). Arrows in A indicate addition of the indicated combination of reagents or buffer only (Control). The dotted line in graph B 1 indicates the negative control of DiOC₆(3) staining obtained in the presence of the ΔΨ_m-dissipating reagent mClCCP. For the determination of AIF release (C), mitochondria were centrifuged (1.5 × 10⁻⁵ g, 1 h) after 5 min of treatment, and the supernatant was incubated for 30 min with purified HeLa nuclei, followed by determination of their DNA content using the flurochrome propidium iodide, as described in Materials and Methods. Percentages detail the percentage of nuclei exhibiting an apparent subdiploidy.

Figure 4

Effect of Bcl-2 hyperexpression on the ICE- or oxidant-induced PT and the release of an AIF from mitochondria. Mitochondria were purified from T cell lymphoma cell lines stably transfected with a human bcl-2 gene under the control of a tetracyclin-repressable promoter that were treated with doxycyclin (10 ng/ml, 48 h) to repress Bcl-2 expression (Bcl-2⁻, graphs 1–3), or left untreated (Bcl-2⁺, graphs 4–6). The inset in graph 4 shows cytofluorometric profiles of isolated mitochondria stained with an anti–hBcl-2-FITC conjugate. These organelles were exposed to CFS buffer only (graphs 1 and 4) human recombinant ICE (graphs 2 and 5), or t-BHP (graphs 3 and 6) as described in the legend to Fig. 3, followed by determination of swelling (A) and the release of AIF (B), which was tested for its capacity to induce hypoploidy in isolated HeLa nuclei. Note that Bcl-2 hyperexpression on the outer mitochondrial membranes does prevent the t-BHP–induced PT and AIF release, yet does not affect the ICE-induced PT and AIF release.

Figure 5

Effect of Bcl-2 on the AIF release triggered by cytosols from ceramide- or Fas-stimulated cells. Cytosols (10⁷ cells/100 μl CFS buffer) were prepared from washed (three times) cells which were either left untreated (control) or treated with C₂ ceramide (50 μM) or anti-Fas during 30 min. These cytosols (5 μl) were added to 25 μl CFS buffer only or CFS buffer containing mitochondria (50 μg protein) from control CEM-C7-H2 cells (Co. mito) or from Bcl-2–transfected cells (Bcl-2 mito), followed by an incubation step of 30 min at 37°C. The supernatants of these cultures (14,000 g, 10 min, 4°C) were added to purified HeLa nuclei (3 × 10⁴ nuclei in 10 μl CFS buffer). After 90 min of incubation at 37°C, nuclei were stained with PI and analyzed for the frequency of hypoploid events. One experiment out of two yielding similar results is shown. Independent control experiments indicate that C₂ ceramide itself does not induce PT in isolated mitochondria at a dose up to 50 μM (not shown).

Figure 6

In vitro effects of AIF on isolated nuclei and mitochondria. (A) Effect of AIF on nuclear ultrastructure. Isolated HeLa nuclei were incubated with purified AIF (100 ng/ml) and/or the AIF inhibitor Z-VAD.fmk (100 μM) during the indicated interval, followed by transmission electron microscopy. Squares measure 8 μm. (B and C) ICE triggers the mitochondrial release of AIF. Mitochondria were treated with CFS buffer (Control, graph and lane 1) or recombinant ICE (graphs and lanes 2–4) in conditions that induce PT (e.g., Figs. 3 and 4), followed by recovery of the mitochondrial supernatant. These supernatants were then tested for their capacity to induce nuclear apoptosis in the absence (graph and lane 2) or presence of 100 μM Ac.YVAD. cmk (graph and lane 3) or Z-VAD. fmk (graph and lane 4). The readout of this system was either the cytofluorometric detection of nuclear hypoploidy (B) or agarose electrophores to detect oligonucleosomal DNA fragmentation (C) (D) Effects of AIF on isolated mitochondria. The same preparations as in B and C (graphs and lanes 1–4); were added to purified liver mitochondria, followed by determination of large amplitude swelling. In addition, purified recombinant AIF was tested for its capacity to induce mitochondrial swelling (graph 5). AIF was either added alone (solid line) or together with 100 μM Z-VAD.fmk (dotted line), as indicated by the arrow.

Figure 7

AIF proteolytically activates CPP32. (A) Induction of CPP32 activity as determined by a fluorogenic substrate. Variable concentrations of AIF-containing supernatants (SN) from Atr-treated mitochondria were added in the presence (open circles) or absence of Z-VAD.fmk (100 μM, filled circles) to constant amounts (100 ng) of recombinant purified CPP32, followed by determination of the DEVDase activity of CPP32 using the fluorogenic substrate Ac-DEVD-amino-4-methylcoumarin. To exclude that Z-VAD.fmk itself might inhibit the DEVDase activity of CPP32, this inhibitor was added together with the fluorogenic substrate after CPP32 had been activated with SN in the absence of any inhibitor (open square). It was also ruled out that the AIF-containing supernatant alone might contain a DEVDase activity (filled square). Purified AIF (100 ng/ml) was also used to activate CPP32 (filled triangle). (B) Induction of CPP32 activity as determined by PARP cleavage. Nuclei were incubated during 90 min at 37°C in the presence or absence of CPP32, AIF-containing supernatant, and/or Z-VAD.fmk, as indicated, followed by Western blotting and immunochemical determination of the cleavage of PARP. Note that only the combination of CPP32 and AIF, but not either of these compounds alone, cause PARP cleavage. (C) Proteolysis of CPP32 by AIF. Recombinant CPP32 protein (10 ng) was exposed to the indicated combination of AIF-containing mitochondrial supernatant (10 μg in 50 μl), Z-VAD.fmk (100 μM) and/or Ac-DEVD.CHO (100 μM), followed by Western blot analysis of CPP32 cleavage.

Figure 8

Hypothetical scenario of Fas-induced apoptosis. After trimerization of the Fas receptor and activation of MACH1/FLICE, depending on the cell type, the ceramide and/or the ICE pathways are activated for death induction. Bcl-2 is an efficient inhibitor of ceramide- (and prooxidant- ) induced mitochondrial PT, yet fails to prevent the ICE-induced PT. PT marks the initiation of the common effector phase of apoptosis and entails the release of mitochondrial intermembrane proteins including AIF and cytochrome c. AIF itself induces PT and thus engages in a self-destructive autoamplification loop. AIF alone and cytochrome c in combination with yet unknown cytoplasmic factors are apoptogenic (i.e., cause DNA condensation and fragmentation by acting on nuclear substrates). In addition, they trigger the activation of CPP32 (and possibly, directly or indirectly, of other proteases). For details and references, consult text.