Apoptosome-independent pathway for apoptosis. Biochemical analysis of APAF-1 defects and biological outcomes.

Induction and execution of apoptosis programs are generally believed to be mediated through a hierarchy of caspase activation. By using two cellular variants obtained from the L1210 cell line (L1210/S and L1210/0), we have shown previously that staurosporine induces apoptotic cell death through both caspase-dependent and caspase-independent pathways. Both pathways normally coexisted in L1210/S cells, whereas L1210/0 cells lacked the ability to activate caspases despite the confirmed presence of both procaspase-3 and -9. Here we show that this defect in caspase activation is not due to mechanisms such as an absence of cytochrome c release, the expression of non-functional caspases, or the presence of an endogenous inhibitor but results from the loss of apoptosis protease activator protein-1 (APAF-1) expression. This absence of APAF-1 protein results from multiple alterations at both genomic and transcriptional levels. However, although this lack of APAF-1 delays the apoptotic program, it does not hamper its execution. Importantly, in these cells, apoptosis develops not only in an APAF-1-independent way but also in the absence of caspase-3 and -9 activation. Altogether these findings provide evidence that apoptosis may occur through alternative signaling pathways independent of APAF-1 expression and totally dissociated from any caspase processing. Therefore, the L1210/0 variant sub-line provides a valuable tool for the elucidation of these pathways.

Induction and execution of apoptosis programs are generally believed to be mediated through a hierarchy of caspase activation. By using two cellular variants obtained from the L1210 cell line (L1210/S and L1210/0), we have shown previously that staurosporine induces apoptotic cell death through both caspase-dependent and caspase-independent pathways. Both pathways normally coexisted in L1210/S cells, whereas L1210/0 cells lacked the ability to activate caspases despite the confirmed presence of both procaspase-3 and -9. Here we show that this defect in caspase activation is not due to mechanisms such as an absence of cytochrome c release, the expression of non-functional caspases, or the presence of an endogenous inhibitor but results from the loss of apoptosis protease activator protein-1 (APAF-1) expression. This absence of APAF-1 protein results from multiple alterations at both genomic and transcriptional levels. However, although this lack of APAF-1 delays the apoptotic program, it does not hamper its execution. Importantly, in these cells, apoptosis develops not only in an APAF-1-independent way but also in the absence of caspase-3 and -9 activation. Altogether these findings provide evidence that apoptosis may occur through alternative signaling pathways independent of APAF-1 expression and totally dissociated from any caspase processing. Therefore, the L1210/0 variant subline provides a valuable tool for the elucidation of these pathways.
Apoptosis is a major form of cell death essential to maintain homeostasis in multicellular organisms. It can be activated by a variety of stimuli including death receptor ligation, chemotherapeutic drugs, and UV-or ␥-irradiation. Resistance to apoptosis has been established as one important mechanism of antitumor treatment failure (1).
Apoptotic signals are generally believed to be mediated through a hierarchy of caspase activation. Caspases, members of the cysteine protease family, are synthesized as relatively catalytically inactive proenzymes and need to be activated by proteolytic cleavage at internal aspartate residues (2)(3)(4)(5)(6)(7)(8)(9). Two pathways of caspase activation during apoptosis have been described. The first one is mediated by death receptors, such as Fas or tumor necrosis factor receptors, controlled by caspases-8/10, which in turn activate downstream effector caspases such as caspase-3 and caspase-7. In the second pathway, diverse apoptotic signals converge at the mitochondrial level, inducing the release of cytochrome c from the mitochondria into the cytosol (10). Once in the cytosol, cytochrome c binds to its cytosolic partner APAF-1 1 (apoptotic protease activating factor-1), the human homologue of Caenorhabditis elegans apoptotic protein CED-4, and induces the oligomerization of APAF-1-cytochrome c complex in a dATP/ATP-dependent manner (11,12). This multimeric complex, named "apoptosome," is sufficient to recruit the initiator caspase, procaspase-9, to the complex and induces procaspase-9 autoactivation (13). The activated caspase-9 is released from the apoptosome and subsequently initiates a caspase cascade involving the effector caspases such as caspases-3, -6, and -7 (14,15). Once active, these caspases cleave various cellular targets, ultimately leading to cell death. Many inducers of apoptosis, such as ␥-irradiation, anticancer drugs, staurosporine, and growth factor withdrawal, were shown to activate cell death independent of death receptor pathways, thus presumably directly activating the mitochondrial intrinsic apoptotic pathway (11, 16 -18). Both the death receptor and mitochondrial pathway can be interconnected via the activation of the proapoptotic Bcl-2 family member, Bid, through caspase-8, possibly serving as an amplification loop (19).
Deficiency of the essential components of the mitochondrial apoptotic pathway renders cells resistant to apoptotic stimulation, as shown in gene-disrupted mice (20 -23). Recently, it was reported that low levels or deficiency in APAF-1 protein can determine sensitivity to apoptosis downstream of mitochondrial events suggesting that regulation of APAF-1 may be important for apoptosis regulation (24,25). However, it seems that caspase activation and subsequent apoptosis could still occur in an APAF-1-independent pathway (26 -29).
Although much progress has been made in dissecting the routes to activation of caspases (30), evidence for caspaseindependent cell deaths was only recently demonstrated by a number of reported cases, including ours (see Refs. 31-33 and reviewed in Ref. 34). However, there has been some controversy over the nature of these caspase-independent deaths. It is generally believed that these deaths are of a necrotic type. Furthermore, it has been difficult to elucidate the components and significance of these pathways because they are often intimately intermingled with and obscured by the ubiquitous caspase pathways.
By using two cellular variants obtained from an L1210 cell line (L1210/S and L1210/0), we showed that staurosporineinduced apoptosis operates through both a rapid caspase-dependent and a delayed caspase-independent pathway depending on the enzymatic equipment of the cell (33). The two pathways normally coexist in L1210/S cells, whereas only the caspase-independent pathway is active in L1210/0 cells due to a general defect in the caspase activation. We identify here that this defect is due to the loss of a functional APAF-1 protein, which results from alterations at both the genomic as well as the transcriptional level. However, despite this loss and the subsequent absence of caspase activation, apoptosis develops, although delayed.

EXPERIMENTAL PROCEDURES
Reagents-Staurosporine was purchased from Roche Diagnostics. Stock solution (2.0 mM staurosporine in Me 2 SO) was stored at Ϫ20°C. Proteinase K and DNase-free RNase were obtained from Roche Diagnostics, and the DNA molecular weight marker (123-bp ladder) was from Invitrogen. Electrophoresis grade agarose, polyvinylidene difluoride membranes (Immobilon-P), and the ECL detection system were from Amersham Biosciences. Bovine cytochrome c, dATP, 3,3Јdihexyloxacarbocyanine (DiOC 6 (3)) were obtained from Sigma.
Cell Viability, Morphologic and Biochemical Assessment of Apoptosis-Apoptosis was monitored by cell morphology changes after May-Grü nwald-Giemsa staining; typical condensed and fragmented nuclear morphology was visible after staining with 4Ј,6-diamidino-2-phenylindole (DAPI, Roche Diagnostics) and internucleosomal DNA fragmentation. For measurement of DNA fragmentation, DNA was isolated from 2 ϫ 10 6 cells and separated by agarose gel electrophoresis as described previously (37).
Western Blot Analysis-For the detection of cytosolic cytochrome c and apoptosis-inducing factor (AIF), L1210/S and L1210/0 cells were treated with 5 M staurosporine for 3 and 12 h, respectively, and analyzed by Western blotting. Briefly, 1 ϫ 10 7 cells were collected, washed twice in ice-cold phosphate-buffered saline (PBS), and resuspended in buffer A (250 mM sucrose, 20 mM Hepes-KOH, 10 mM KCl, 1.5 mM EGTA, 1.5 mM EDTA, 1 mM MgCl 2 , 1 mM DTT, and a mixture of protease inhibitors). Cells were then transferred to a 2-ml Dounce-type homogenizer and allowed to swell for 20 min on ice. Cells were then disrupted with 30 strokes of B-type pestle. The cell homogenates were centrifuged at 800 ϫ g for 10 min at 4°C. Supernatants were further centrifuged at 22,000 ϫ g for 15 min at 4°C. The resulting supernatants (cytosolic fraction) were removed and stored at Ϫ70°C until required. For the detection of caspase processing, 2 ϫ 10 6 cells were washed twice with PBS and resuspended in 100 l of RIPA buffer (1% Triton X-100, 10% deoxycholate, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% SDS, 0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin) on ice. After 10,000 ϫ g centrifugation at 4°C for 15 min, the supernatants were harvested. Protein concentration in cellular extracts was determined using the Bradford (Bio-Rad) protein assay reagent. For detection, samples (15 g/lane) were fractionated on a 10, 12, or 15% SDS-PAGE in a Tris-glycine running buffer and blotted on polyvi-nylidene difluoride membranes. The loading homogeneity and transfer efficiency were checked by staining the membrane with red Ponceau and the gel with Coomassie Blue. The membranes were preblocked 2 h at room temperature in PBS containing 5% non-fat dry milk powder and then immunoblotted overnight with the primary antibody. Mouse polyclonal anti-caspase-3 (Transduction Laboratories), anti-cytochrome c (Pharmingen), rabbit polyclonal anti-caspase-8 (Pharmingen), anticaspase 9 (Cell Signaling, Ozyme), and anti-APAF-1 (either from Pharmingen or from Alexis, Coger, Paris, France), and anti-AIF (Santa Cruz Biotechnology) were used as primary antibodies at 1:1000 dilution for detection of caspases and APAF-1, at 1:250 dilution for detection of cytochrome c, and at 1:2000 for detection of AIF. After washing, the membranes were then incubated with the respective peroxidase-conjugated secondary antibody for 1 h. Detection was performed using the chemiluminescence procedure (ECL, Amersham Biosciences), according to the manufacturer's recommendations.
Measurement of Caspase Activity-Cells (2 ϫ 10 6 ) were pelleted, washed with PBS, pH 7.2, and lysed in 50 mM Tris-HCl, pH 7.5, 0.03% Nonidet P-40, 1.0 mM DTT. Lysates were centrifuged at 14,000 rpm for 15 min at 4°C. Protein determination was done using the Bradford assay. Assays were set up in flat bottom 96-well plates containing 0.2 mM of the caspase-3-like peptide substrate (DEVD-pNa, N-acetyl-Asp-Glu-Val-Asp-p-nitroanilide, Alexis, Coger, Paris, France) in a caspase reaction buffer (100 mM Hepes, pH 7.5, 10% sucrose, 0.1% CHAPS, 10 mM DTT) and 0.01 ml of protein extract (20 -50 g) in a total volume of 0.1 ml. Assays were performed at 37°C. Release of pNa was detected by periodic readings of absorbance at 405 nm taken against a blank containing buffer and peptide alone (i.e. no extract) from 0 to 5 h to mark the linearity of the enzymatic reaction in time. Enzyme activities were measured as initial velocities and expressed as relative intensity/ min/mg total protein within the linear range of the response. In some experiments a specific peptide substrate for granzyme B was used (IETD-pNa, Alexis, Coger, Paris, France).
Cell-free Reactions-Cells were collected, washed twice in PBS, and resuspended in CEB hypotonic buffer (20 mM Hepes-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1 mM DTT, 100 M phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 2 g/ml aprotinin). Cells were then transferred to a 2-ml Dounce-type homogenizer and allowed to swell for 20 min on ice. Cells were then disrupted with 30 strokes of B-type pestle. Lysates were then transferred to Eppendorf tubes and were centrifuged at 15,000 ϫ g for 15 min at 4°C. The supernatant was removed and frozen at Ϫ70°C until required. Activation of cell extracts was induced by addition of 5 M bovine cytochrome c and 1 mM dATP to extracts. Aliquots were then withdrawn for subsequent SDS-PAGE/ Western blot and caspase assay. For activation of caspases by granzyme B, cell extracts were prepared in the caspase assay buffer and incubated with 2.8 ng/ml of granzyme B (Calbiochem and VWR International).
In some experiments, before addition of cytochrome c and dATP, extracts from L1210/0 or L1210/S cells were mixed together or with extracts from melanoma cells (SKMel) known to be "APAF-1-negative" cells (generously provided by Dr. Soengas at Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). We then assessed caspase activation by monitoring DEVD-pNa hydrolysis.
Preparation and Activation of Naive and Immunodepleted Cell Lysates-For the immunodepletion experiments, lysates from L1210/S cells prepared in CEB buffer were incubated for 2 h with 50 l of protein A/G plus agarose beads (Santa Cruz Biotechnology), precoated with 3% bovine serum albumin coated with antibodies against caspase-9. Immunodepleted supernatants were collected by centrifugation. In vitro activation of caspases in both normal and immunodepleted lysates was initiated by incubating lysates (6 mg/ml) with cytochrome c (5 M) and dATP (1 mM) at 37°C. In the reconstitution assay equivalent amounts of L1210/0 lysates and L1210/S-immunodepleted lysates were mixed before the addition of cytochrome c and dATP.
Reverse Transcription (RT)-PCR and Nucleotide Sequencing-RNA was isolated from both L1210/0 and L1210/S cells using the guanidinium isothiocyanate-based Trizol reagent according to the manufacturer's instructions (Invitrogen). We then synthesized APAF-1 cDNAs by RT-PCR using oligo(dT) 15 and random hexamer priming. The amplification of APAF-1 cDNA was performed with primer pairs presented in Table I. PCR products were purified by agarose gel electrophoresis, cloned in PGEM-T easy vector (Promega, Charbonnières-les Bains, France), and sequenced.

RESULTS
Defective Caspase Activation in L1210/0 Cells-As shown previously (33), staurosporine treatment resulted in an early cell death (within 3 h) in L1210/S cells, whereas in L1210/0 cells, death occurred only after 12 h. In both instances, death was of the apoptotic type, as indicated by the presence of morphological features such as cell shrinkage and condensation of chromatin and biochemical markers such as DNA ladder (Fig. 1A) and annexin V staining (data not shown). Despite this similar ability to induce apoptosis in both cell lines, staurosporine triggered caspase activation in L1210/S but not in L1210/0 cells, as indicated by the measurement of caspase-3-like activity and the absence of efficient processing of both initiator (caspase-8, -2, and -9) and the effector caspase-3 (Fig. 1B). Note that in L1210/S, the level of the procaspase-8 remained unchanged, indicating that in this cell line, staurosporine treatment did not activate this caspase. Furthermore, as shown previously (33), procaspase-7 was not expressed in L1210/0 cells, and procaspase-6 was barely detectable and not efficiently processed. This apparently caspase-independent pathway of apoptosis observed in L1210/0 cells was not specific to staurosporine as it is also induced by other drugs (cisplatin and etoposide) that activated caspases in L1210/S cells (data not shown). Starting from this, we performed a detailed dissection of molecular events in both staurosporine-treated L1210/S and L1210/0 cells in order to identify potential defects in staurosporine-induced caspase activation in L1210/0 cells. ⌬ m Reduction Develops in Staurosporine-induced Apoptosis of Both L1210/S and L1210/0 Cells-An early event of many types of apoptosis is the opening of mitochondrial pores or megachannels, a process known as the mitochondrial permeability transition (PT) (40). One of the consequences of PT is the disruption of the mitochondrial inner transmembrane potential. Therefore, the cytofluorometric determination of ⌬ m during staurosporine-induced apoptosis in both L1210/0 and L1210/S was performed using the potential-sensitive probe DiOC 6 (3). Treatment of L1210/S with 5 M staurosporine caused an initial increase of ⌬ m as early as 30 min after the onset of treatment ( Fig. 2A). This increase could reflect the transient mitochondrial hyperpolarization occurring before or concomitantly to cytochrome c release, and preceding caspase activation and subsequent membrane depolarization, as reported by several groups (41)(42)(43)(44). Then a pronounced reduction of ⌬ m began by 1 h, and after 3 h of treatment, most of the population showed loss of ⌬ m signal. In L1210/0 cells, however, the reduction of ⌬ m was first detected at 8 h, and only 30% of the cell population presented the decrease in ⌬ m after 12 h of treatment, even though the same morphological and biochemical characteristics of apoptosis were observed in at least 80% of the cells. The fact that ⌬ m loss was delayed and less intense in L1210/0 compared with L1210/S cells may suggest that caspase activity could be necessary for a stronger disruption of ⌬ m during apoptosis.
Release of Cytochrome c and AIF in Both L1210/S and L1210/0 Cells during Staurosporine-induced Apoptosis-Loss of the mitochondrial transmembrane potential, which results in mitochondrial PT, has been shown to cause mitochondrial release of apoptogenic proteins, including cytochrome c and AIF (45,46). Cytochrome c can then activate downstream caspases amplifying the apoptotic signal (3,10,11,47), whereas AIF acts in a caspase-independent fashion by causing chromatin condensation and large scale DNA fragmentation. Thus, we tested the potential of staurosporine to induce cytochrome c and AIF translocation to the cytosol in both L1210/0 Procedures." B, whole cell extracts (15 g) prepared from both cell lines were also analyzed for procaspase (procasp)-8, -2, -9, -6, -7, and -3 processing by Western blotting as described under "Experimental Procedures" (upper panel). Levels of procaspase-2, -3, -6, -7, -8, and -9 along with the processed caspase-2, -3, and -9 subunits are shown. Activation of caspase-2, -3, and -9 was characterized by a reduction in the intensity of the proforms (43, 32, and 46/47 kDa, respectively) and the appearance of a band at 36 (for caspase-2), 17 (for caspase-3), and 37/35 kDa (for caspase-9). Caspase-6 and -7 activation led to the disappearance of the 32-kDa band. Procaspase-8 was not processed in both treated cell lines. Caspase-3-like enzyme activity was assayed by the measurements of DEVD-pNa hydrolysis by spectrophotometry, as described under "Experimental Procedures" (lower panel). and L1210/S cells. Cytosolic extracts prepared from staurosporine-treated cells were subjected to SDS-PAGE and Western blot analysis for both cytochrome c and AIF (Fig. 2B). Although no detectable cytochrome c was found in the cytosolic extract of untreated cells, staurosporine triggered the translocation of both cytochrome c and AIF to the cytosol of both L1210/S and L1210/0 cells. Even though AIF release to the cytosol occurred almost similarly in L1210/S and L1210/0 staurosporine-treated cells, a possible involvement of this protein in the caspaseindependent pathway for apoptosis that developed in staurosporine-treated L1210/0 cells cannot be excluded. However, as AIF-induced apoptosis has been shown to be accompanied by a large scale DNA fragmentation but not oligonucleosomal DNA fragmentation, this result would indicate that additional events are required to explain DNA laddering observed in L1210/0-treated cells. This hypothesis is supported by the fact that serine protease inhibitors, including 3,4-dichloroisocoumarin and N-tosyl-L-phenylalanine chloromethyl ketone, prevent DNA laddering in staurosporine-treated cells but have no effect on cell death. 2 Cytochrome c Is Not Able to Activate Caspases in L1210/0 Cells-Despite its similar ability to induce release of mitochondrial cytochrome c in both L1210/S and L1210/0 cells, in L1210/0 cells staurosporine was unable to trigger efficient processing of both the initiator caspase-9 and the effector caspase-3 and induction of a caspase-3-like activity (see Fig. 1B). These observations were extended by showing that cell-free extracts from L1210/0 cells were fully resistant to the activation of caspases by 5 M cytochrome c and 1 mM dATP, although in a parallel experiment, caspase-3-like activity was induced in activated extracts from L1210/S cells (Fig. 2C). Altogether, these results show that, as reported previously in other situations (18,48,49), in L1210/0 cells the release of cytochrome c may occur in a caspase-independent manner. Caspases-3 and -9 Expressed in L1210/0 Cells Can Be Processed in Vitro by Granzyme B-Although expressed in L1210/0, at the same level as in L1210/S cells, procaspases-3 and -9 could not be activated either by cytochrome c/dATP in L1210/0 cell-free extracts nor by staurosporine treatment of whole cells (see Fig. 1B and 2C). To ensure further that this absence of caspase processing was not the result of mutations leading to the generation of uncleavable and/or non-functional caspases, cellular extracts from both L1210/S and L1210/0 cells were incubated in vitro in the presence of granzyme B. Granzyme B has been shown to activate directly or indirectly most Simultaneous electrophoresis of bovine cytochrome c (cytc) is shown. C, measurement of caspase-3-like enzyme activity in cell-free extracts prepared from L1210 cells. The lysates were dATP-and cytochrome c-activated for 2 h at 37°C. DEVDase activity was analyzed as described under "Experimental Procedures." procaspases in an in vitro system (50). In the presence of granzyme B, procaspases-3 and -9 were processed as shown by Western blot, and a caspase-3 like activity is generated as shown by DEVD-pNA cleavage (Fig. 3). This result indicates that caspase-3 activity is not defective in L1210/0 cells and can be generated provided a direct processing of procaspases occurs.
In order to determine whether caspase-9 is active in these particular cells, can be recruited into a functional apoptosome complex, and can then initiate other caspase activation, we immunodepleted an L1210/S lysate for this caspase and mixed this caspase-9-depleted lysate with L1210/0 lysates before activation by cytochrome c and dATP. Depletion of procaspase-9 from L1210/S extracts was confirmed by Western blot analysis and DEVDase activity (Fig. 4, lanes 3 and 4). Depletion of procaspase-9 abolished cytochrome c/dATP-inducible activation of L1210/S extracts, consistent with caspase-9 being required for caspase-3 activation in this cell system. Addition of caspase-9-immunodepleted L1210/S extracts to L1210/0 extracts reconstituted L1210/0 procaspase-9 processing and the subsequent procaspase-3 activation in the presence of cytochrome c and dATP (lane 6), indicating that procaspase-9 from L1210/0 cells can be recruited in a functional apoptosome, and excluding that a diminished amount of caspase-9 in the apoptosome would be responsible for the dysfunctional apoptosome activation as described recently (51). Altogether, these results demonstrate that caspase-9 and -3 activities are intact in the L1210/0 cell line, and suggest that the L1210/0 defect is likely to be located downstream from the release of cytochrome c but upstream to the activation of procaspase-9 and -3.
Defective APAF-1 Protein Expression in L1210/0 Cells-It has been shown that cytochrome c released from mitochondria promotes the activation of caspase-9 through APAF-1, which forms a complex with caspase-9 in the presence of dATP and cytochrome c, two co-factors for APAF-1 function (11). It was recently reported that APAF-1 inactivation by genetic loss or inhibition may constitute a significant mode of resistance to apoptosis in cancer cells (24,25,52). Given that 1) in staurosporine-treated L1210/0 cells, cytochrome c was released without subsequent activation of procaspase-9, and 2) this procaspase is expressed and can be processed and activated in this cell line, we hypothesize that the defect in L1210/0 cells might stay at the level of the APAF-1 protein. Therefore, the expression of APAF-1 was examined in both L1210/S and L1210/0 cells by immunoblotting using a polyclonal anti-APAF-1 antibody directed against the N-terminal domain of the human protein (Pharmingen) and compared with that of melanoma cells reported to be APAF-1-negative cells (52) and with a monoclonal anti-APAF-1 antibody directed against the N-terminal domain of the mouse protein (Alexis, Coger, Paris, France). Fig. 5 (top) showed that this 130-kDa protein is highly expressed in L1210/S cells. In contrast, in L1210/0 cells, the level of APAF-1 was less than that of the APAF-1-negative cells. Furthermore, the longer exposure time of the immunoblot revealed in L1210/0 extracts, in addition to the normal-size protein, an abnormal protein of smaller molecular weight, which was not present in L1210/S extracts (Fig. 5, bottom). We verified that APAF-1 expression did not recover after 5 M 5-azacytidine treatment of L1210/0 cells for 72 h, thus excluding a possible transcription silencing of APAF-1 through methylation as already reported for some melanoma and leukemia cells (52, 53) (data not shown).
Identification of Multiple Variants of APAF-1 Transcripts in L1210/0 Cells by RT-PCR-We next investigated whether in L1210/0 cells an APAF-1 defect at the mRNA level might account for the loss of expression of the protein. We performed RT-PCR with a number of primer pairs covering the entire coding region (Fig. 6A and Table I) to amplify overlapping regions of APAF-1 cDNA. For all primer pairs, a band was recovered from PCR in both L1210/S and L1210/0 cells that corresponded to the expected PCR product (Fig. 6B).
However, by using the 7AS oligonucleotide as the forward primer, all the primer pairs produced an additional shorter fragment, which was amplified from L1210/0 cDNA only. L1210/0 cDNA from exon 2 to 22 was then amplified by the 1S/7AS set of primers (data not shown). The expected fragment of 3102 bp and the shorter PCR product obtained were excised from the gel, cloned, and sequenced in their entirety. As a control, the 3102-bp fragment amplified from the parental L1210/S cDNA was also cloned and sequenced. The sequence of the 3102-bp fragment, obtained from L1210/0 and L1210/S cDNA, was identical to that reported previously and deposited in the GenBank TM data base for the murine APAF-1 (accession number AF064071) (54). It contained both the 11 amino acids inserted between amino acids 98 and 99 after the caspase recruitment domain (CARD) and the additional WD-40 repeat (sequence of extra 43 amino acids between amino acids 811 and 812 corresponding to exon 17Ј) in the C terminus. This additional WD-40 repeat has been reported to be necessary for cytochrome c-and dATP-dependent procapase-9 activation (13,55,56). Sequencing of the shorter fragment obtained only from L1210/0 revealed that it lacked exons 20 and 21 (Fig. 6C).
Surprisingly, the cloning experiment led us to the identification of another shorter fragment obtained only from L1210/0 cDNA. This fragment lacked not only exons 20 and 21 but also exon 10 and 11. It likely represents a minor form of APAF-1 transcript because it was not possible to detect it by PCR amplification with adequate flanking primers (data not shown). The sequence of the fragments obtained from the amplification of the 3Ј-end of L1210/0 and L1210/S cDNAs using the primer pair (7S/11AS) was identical to that reported previously (54). Therefore, this study has led to the identification in L1210/0 cells of two altered APAF-1 transcripts (⌬20 -21 and ⌬10 -11/⌬20 -21) coexisting with the wild-type form.
Apaf-1 Gene Deletion in L1210/0 Cells-Because the two new variant APAF-1 transcripts were truncated at the same position and correspond exactly to total exon deletions (⌬20 -21) preserving in phase coding sequences, it was important to determine whether the L1210/0 APAF-1 RNA variants were generated from the same APAF-1 allele and resulted from defects at the genomic level or from exon skipping during mRNA splicing. Southern blot analyses were performed comparatively in L1210/0 and L1210/S (Fig. 7). The genomic DNAs were restricted with several different enzymes, separated on an agarose gel, and transferred to a nylon membrane. The membrane was hybridized with two 32 P-labeled probes generated first by amplification of a fragment using the 7S/11AS primer pair, cloned in PGEM-T vector, and then cut either by StuI/ BglII (restriction site in exon 19 and 24, respectively) or BglII/ SpeI (restriction in exon 19 and in PGEM-T vector, respectively). Autoradiography of the blots obtained using the StuI/BglII probe revealed the presence of rearranged fragments in DNA from L1210/0 cells in addition to the expected normal sized fragments visible in DNA from L1210/S cells. In contrast, no rearrangement was found on autoradiography obtained using the BglII/SpeI probe. These results indicate that the altered forms of APAF-1 transcript, identified in L1210/0 cells, reflect a genomic deletion within the upstream flanking intron of exon 20 and the downstream flanking intron of exon 21 resulting in the absence of exons 20 and 21. Therefore, the two new variant transcripts were generated from the same APAF-1-mutated allele.
Supplementation Studies Restore Cytochrome c/dATPdependent Caspase Activation in L1210/0 Cells-Despite several attempts, all the experiments aimed to restore cytochrome c/dATP-dependent caspase activation by normal full-length APAF-1 transfections reproducibly failed. This may be due to the death-inducing property of functional APAF-1, which may have forced the elimination of the cells that express higher levels of the protein; only low expressing cells remained viable, and this viability could be also reinforced by a dominant-negative effect of mutated APAF-1. Therefore, the absence of in vivo supplementation reinforces our conclusion that mutated APAF-1 abrogated the caspase-dependent pathway for apoptosis. Noteworthy, even in transient expression transfections, the expression of functional APAF-1 was not sufficient to restore normal phenotype. An alternative possibility is that a not yet identified factor is needed to restore caspase activation in L1210/0 cells.
To reinforce our conclusion that the absence of caspase activation in L1210/0 cells results from this loss of APAF-1 expression, a whole set of in vitro mixing experiments was carried out. Various amounts of L1210/0 extracts were mixed with dilutions of L1210/S extracts, which expressed high level of APAF-1 protein (Fig. 8A). It was verified that at each dilution, cytochrome c/dATP-dependent caspase activity is low. However, mixing them to various amounts of L1210/0 extracts induced a restoration of cytochrome c/dATP-dependent caspase activity that seems to be dependent on the amount of L1210/0 extracts. This restoration of caspase activity was confirmed by the immunodetection of caspase-3 cleavage (Fig. 8B, compare the 3rd lane with the 1st and 2nd lanes and the 6th lane with the 4th and 5th lanes). As a control, mixing extracts from melanoma cells, known to be APAF-1-negative cells, with equivalent amounts of L1210/0 extracts did not restore cytochrome c/dATPdependent caspase activation (Fig. 8C), indicating that both L1210/0 and melanoma "APAF-1-defective" cells shared a similar functional defect. Even though it is not excluded that some unknown factor(s) present in the L1210/S extracts could contribute to restoring caspase activation in L1210/0 extracts, altogether these data confirm that the defect in caspase activation is linked to APAF-1 protein expression in L1210/0 cells. DISCUSSION Currently, there is some controversy in the literature as to the actual role played by caspases in apoptosis. Numerous results suggest that caspases may form a hierarchical, although redundant, network that may function as an amplifier for a given apoptotic stimulus (5,57,58). APAF-1 was shown to participate as an adaptor molecule in the sequential activation of caspase-9 and caspase-3 (11,14,59). The importance of APAF-1 has been highlighted by reports (23,54,60) that demonstrated that APAF-1 or caspase-9 deficiency results in embryonic lethality due to defective neuronal apoptosis. Importantly, recent reports (24,25,51) described that loss of APAF-1 function resulted in resistance to cytochrome c/dATP-dependent caspase activation and subsequently to apoptosis. However, in one of these reports (24), cytochrome c-dependent caspase activation was only delayed. In the other report (25), the resistance to this caspase activation was attributed to the loss of APAF-1 activity;
in this case, as the cell lines retained APAF-1 expression, the involvement of a specific inhibitor was suggested.
Here we demonstrate by a stepwise functional dissection of the apoptotic signaling in L1210/0 mouse leukemia cells that apoptosis still operates in an APAF-1-defective context. Of major interest for ongoing studies is indeed the fact that this lack of a functional APAF-1 protein was associated with resistance to caspase activation. The molecular basis of the APAF-1 functional defect was assigned to both a genomic alteration affecting exon 20 and exon 21, and generation of alternative splice variants by exon 10 and exon 11 skipping. These alterations affect important functional domains of APAF-1 and could explain the biological features of these cells.
1) The deletion of exons 20 -21, present in both forms was due to a disruption of the Apaf-1 gene, and these cells contained both a wild-type and a mutated allele of APAF-1. This deletion resulted in the elimination of two of the 13 WD-40 repeats, whose importance in the regulation of the apoptosome had been underlined experimentally in several reports (61,62). Indeed, the genetically engineered mutant cells carrying deletion in the WD-40 region failed to activate procaspase-3 (12), like the spontaneous APAF-1 mutated cell line we have investigated. Moreover, our work defines more precisely a minimal deletion in the WD-40 region associated to APAF-1 dysfunction.
2) In addition to the ⌬20 -21 gene deletion, the APAF-1-(⌬20 -21/10 -11) transcript lacks exon 10 and the exon 11. These missing exons (leading to the deletion of residues 456 -537 of the mouse APAF-1 protein) normally encoded most of the linker between the Ced-4 homology domain and the WD repeat region. The importance of this linker region in the proteolytic processing of the procaspase-9 molecule has already been demonstrated using a mutated human APAF-1 protein (deletion of residues 456 -559) generated in vitro (61). The absence of exons 10 and 11 in this variant transcript is likely to result from an alternative splicing because no rearrangement or deletion of genomic DNA in this region was detected. Such a transcript, accumulating defects in two important functional domains of APAF-1, has never been reported so far.
Altogether, the genomic analysis and the expression profile of APAF-1 transcripts in the L1210/0 cells support the conclusion that the two APAF-1 alleles are expressed, one is normal and the other bears a mutation that deletes exon 20 and exon 21. Moreover, by alternative splicing, the mutated allele gen-    , lanes 3 and 6). At both dilutions of L1210/S extracts, cleavage of procaspase-3 was either undetectable (lane 1) or very weak (lanes 4 and 5). An additional control mixing L1210/S extract with the same amount of bovine serum albumin (BSA) as L1210/0 extract was done in order to eliminate a possible effect of the total protein concentration in the mixture (lane 5). C, equal amounts of the L1210/0 (L/0) were mixed with APAF-1-defective melanoma cell extracts (SKMel 94), activated with cytochrome c plus dATP for 2 h at 37°C, and then assayed for DEVDase activity. Open bars and gray bars indicate each control non-mixed cell extracts treated with buffer and cytochrome c plus dATP for 2 h at 37°C, respectively. the protein could alter APAF-1 function. Defects in APAF-1 protein expression in these cells are not yet understood. Although the wild-type allele of Apaf-1 gene was expressed, it remains unclear why the intensity of the corresponding band was so low. Similarly, the additional APAF-1 band of lower molecular weight detected on Western blot, in L1210/0 but not in L1210/S cells, and only after a longer exposure time of the APAF-1 immunoblot, might correspond to the mutated form of APAF-1. We speculate that the expression of these altered APAF-1 proteins might generate a general alteration of the stability of other associated proteins, including the wild-type APAF-1 protein and partners in the formation of the apoptosome as already described in other cases (63,64). This level of regulation has never been envisioned so far but should open a wide field of promising investigation on apoptosome formation.
The Biological Outcomes of APAF-1 Activity Defects-Altogether these molecular and biochemical observations led us to predict that the altered forms of APAF-1 transcripts identified in L1210/0 cells are likely to be non-functional and could explain the defect in APAF-1 activity in these cells. Furthermore, because in these new forms of transcripts many sequences of normal APAF-1 are preserved, including CARD, Ced-4 homology, and nucleotide-binding domains, it could be postulated that they should act as dominant-negative for full-length apoptosome. This effect might explain the reproducible and specific failure of the APAF-1 transfection assays performed in order to restore cytochrome c/dATP-dependent activation of caspase in L1210/0 cells. However, because in vitro supplementation of L1210/0 extracts with L1210/S extracts cells was able to restore cytochrome c/dATP-dependent activation of caspase, it indicates that in vivo other factor(s), in addition to APAF-1, could be involved.
Despite evidence for a central role of APAF-1 in the initiation of caspase cascade in a great number of cell models, there are an increasing number of reports demonstrating that alternative mechanisms for effector caspase activation exist independently of APAF-1 and the involvement of the apoptosome (26,27). However, these alternative mechanisms could not account for the results observed in L1210/0 cells, because in these cells a complete defect in the activation of effector caspases was demonstrated. Also important and in contrast with several previous works (24,25,51), despite the defect in APAF-1 protein identified in L1210/0 cells, the apoptotic cell death is delayed but not impaired, indicating that apoptosis can develop through an alternative signaling pathway independent of effector caspase activation and apoptosome functions. The initiator caspase-2 has been shown to act upstream of the mitochondria (65)(66)(67) and to mediate the release of cytochrome c and other apoptotic factors from the mitochondria. Even though procaspase-2 was processed in L1210/S-treated cells, it was not in L1210/0-treated cells. This observation can be explained by the defect of caspase-3 and -9 activation because caspase-2 processing has been shown to be dependent both on caspase-3 and -9 activity. The involvement of caspase-2 in L1210/0-induced apoptosis cannot totally be excluded because it has been shown recently (29,68) that the activation of initiator caspases, including caspase-2, may occur, at least in vitro, without processing of the precursor molecule. However, as no cleavage of cytosolic Bid protein, a known substrate of this caspase (65), was observed in staurosporine-treated L1210/0 cells (data not shown), this indicates that additional mechanisms are required for apoptosome-independent apoptosis in L1210/0.
Altogether, these results indicate that other modes of apoptosis induction exist when both APAF-1 protein is lost and caspase activation is defective. This strengthens the notion that multiple pathways can mediate apoptotic cell death depending on the enzymatic equipment of the cells.
Despite this fine dissection of the apoptotic pathway, the complete knowledge of all events that temporally antecedes and accompanies the morphological changes of apoptosis is still evolving, with many newly discovered regulatory steps. In this respect, besides revealing another model of apoptosis in which caspases do not seem to play a crucial role, the L1210/0 cell line provides a valuable tool to investigate APAF-1 and alternative apoptosome-independent signaling pathways for apoptosis and uncover new molecules regulating them.