Apaf-1 Is a Mediator of E2F-1-induced Apoptosis*

E2F-1 is capable of promoting both cell cycle progression and apoptosis. The latter is important for suppressing untoward expansion of proliferating cells. In this study, we investigated its underlying mechanisms. E2F-1-induced apoptosis was accompanied by caspase-9 activation and inhibited by a specific inhibitor of caspase-9 in K562 sublines overexpressing E2F-1. E2F-1 enhanced the expression of Apaf-1 without the cytosolic accumulation of cytochrome c. Apaf-1-deficient melanoma cell lines were resistant to E2F-1, indicating that Apaf-1 is an essential element of E2F-1-mediated apoptosis. Finally, we isolated the promoter region of the Apaf-1 gene and found a putative binding site for E2F. A chromatin immunoprecipitation assay revealed that E2F-1 bound to Apaf-1 promoter upon E2F-1 overexpression, suggesting that Apaf-1 is under transcriptional regulation of E2F-1. These data demonstrate a novel mechanism of apoptosis in which an increase in Apaf-1 levels results in direct activation of caspase-9 without mitochondrial damage, leading to the initiation of a caspase cascade.

E2F is a family of transcription factors that control G 1 /S transition of eukaryotic cells by regulating the expression of various growth-related genes (1). Target genes of the E2F family include those encoding enzymes for DNA synthesis (DNA polymerase ␣, thymidine kinase, thymidylate synthase, dihydrofolate reductase, and ribonucleotide reductase), regulators of DNA replication (HsOrc1, Cdc6, MCM5, MCM6, and proliferating cell nuclear antigen), and components of the cell cycle machinery (Cdc2, Cdk2, cyclins A, D1, D2, and E, E2F-1, E2F-2, pRB, p107, and the Myc and Myb families) (2). Among E2F family members, E2F-1 is unique in its ability to induce apoptosis, which may play a role in the cellular homeostasis of multicellular organisms (3)(4)(5).
In response to mitogenic stimuli, E2F-1 is induced in quiescent cells and promotes both cell cycle progression and apoptosis (6). The apoptosis-inducing ability of E2F-1 is important for suppressing untoward expansion of proliferating cells and, thus, provides an internal defense mechanism against tumor development. The importance of E2F-1-induced apoptosis under physiological conditions is clearly demonstrated by spontaneous development of multiple tumors in mice lacking E2F-1 (7,8). In addition, it has been reported that deregulation of E2F-1 activity contributes to enhanced proliferation and resistance to cytotoxic drugs in human melanoma cells (9). E2F-1 is also involved in the accelerated apoptosis of hematopoietic progenitor cells, which is considered the major mechanism of bone marrow failure in myelodysplastic syndrome (10). To clarify the molecular basis of these diseases, it is essential to understand the precise mechanisms of E2F-1-induced apoptosis.
There are a few reports dealing with the mechanisms of E2F-1-mediated apoptosis. First, Bates et al. (11) reported that E2F-1 induces transcriptional activation of ARF, which stabilizes p53 by sequestering MDM2, a ubiquitin ligase for p53. The stabilization of p53 results in facilitation of p53-dependent apoptosis. Obviously, this is not the sole mechanism of E2F-1induced apoptosis, because E2F-1 can trigger cell death in p53-deficient tumors, including most leukemia cell lines (12). p53 is also shown to be dispensable for E2F-1-mediated cell death by genetic approaches (13). Second, the lack of NF-B activation has been described as a mechanism of increased sensitivity of E2F-1-overexpressing cells to apoptotic stimuli through the death receptor pathways (14). However, this mechanism is not applicable to all cell types, because E2F-1 can induce apoptosis in a death receptor-independent manner in certain settings. Most recently, three groups demonstrated the involvement of the p53 homologue p73 in E2F-1-mediated apoptosis (15)(16)(17). Although p73 is known to cause apoptotic cell death, its underlying mechanism is still unclear (18). In view of the fact that p73 is also a transcription factor, it is possible that other direct mediators act downstream of p73 in E2F-induced apoptosis.
To explore the mechanisms of E2F-1-induced apoptosis, we established K562 sublines that can overexpress E2F-1 conditionally. Using this system, we found that E2F-1 was capable of activating caspase-9 to initiate the caspase cascade without mitochondrial damage. We further demonstrated that the activation of caspase-9 was caused by up-regulation of Apaf-1, which is a direct transcriptional target of E2F-1.

MATERIALS AND METHODS
Cell Lines and Cell Culture-The human immature myeloid cell line K562 was provided by Dr. Carmen B Lozzio (University of Tennessee Medical Center) and maintained in RPMI 1640 medium supplemented with 10% fetal calf serum. The human melanoma cell lines Malme-3M and SK-MEL-5 were purchased from ATCC (Manassas, VA). G-361 and Colo 679 were obtained from the RIKEN Cell Bank (Tsukuba, Japan), and MMAc was provided by Dr. Hirosi Katayama (Katayama Dermatology Clinic, Gunma, Japan).
Reagents-All of the chemicals were purchased from Sigma unless otherwise stated. Cell-permeable fluoromethylketone-derivatized peptides corresponding to specific binding sites for caspases were used as caspase inhibitors (R & D Systems, Minneapolis, MN). Stock solutions were made with Me 2 SO at 10 mM and used at a final concentration of 100 M in growth medium (the final Me 2 SO concentration was 1%). The colorimetric assay kits for caspase-8 and -9 were also purchased from R & D Systems. Establishment of K562 Sublines That Conditionally Overexpress E2F-1-K562 sublines in which full-length E2F-1 is inducible by the addition of isopropyl-␤-D-thiogalactoside (IPTG) 1 were established using the LacSwitch mammalian expression system (Stratagene, La Jolla, CA). Briefly, we first transfected the eukaryotic Lac repressor-expressing vector p3ЈSS into the immature hematopoietic cell line K562 using electroporation and established stable transformants by selection with hygromycin. The Lac operator-containing vector pOPI3 carrying fulllength E2F-1 cDNA was then transfected into the p3ЈSS transformants and E2F-1-overexpressing clones (referred to as KpEf1 in this study) were isolated in the presence of both hygromycin and G418. Using the same procedure, we established K562 stable transformants that conditionally overexpress full-length E2F-4 and E2F-6, and K562 sublines carrying an empty pOPI3 vector.
Plasmids-An expression plasmid for E2F-1 was constructed by inserting full-length E2F-1 cDNA (nucleotides 82-1408) into the EcoRI/ Ecl136II sites of the pIRES2-EGFP bicistronic mammalian expression vector (CLONTECH, Palo Alto, CA). The structure of the construct was verified by restriction mapping and direct DNA sequencing.
Isolation and Analysis of the Apaf-1 Promoter-We searched the human genome project data base at the Sanger Centre (Hinxton, Cambridge, UK) and found that the 5Ј-untranslated region of the Apaf-1 gene is included in the human chromosome 12 clone RP11-453P16 (DDBJ/EMBL/GenBank TM accession number AC026112). Based on the information obtained, the putative promoter region of the Apaf-1 gene (nucleotides Ϫ1136 to ϩ34; see Fig. 7A for the sequence) was amplified by PCR and inserted into pCAT-basic vector (Promega, Madison, WI) to generate a reporter plasmid. Promoter activity was determined by transient transfection into KpEf1 cells and chloramphenicol acetyltransferase assays as previously reported (19). A search for the putative binding sites of known transcription factors was performed with the TFSEARCH program (cbrc.jp/research/db/TFSEARCHJ.html) based on the TRANSFAC data base (20). To define CpG islands, the ratio of observed/expected CpG was calculated according to the following formula: (number of CpG/number of G ϫ number of C) -N, where N is the total number of nucleotides in the analyzed sequence (21).
Transient Transfection and Fluorescence Microscopy-The cells were seeded at 5 ϫ 10 5 cells/ml on coverslips in 60-mm culture dishes, and the plasmids were transfected with an Effectene transfection reagent (Qiagen, Valencia, CA) according to the manufacturer's instructions. After 72 h of transfection, the cells on coverslips were stained with Annexin-V-Cy3 (Medical & Biological Laboratories, Nagoya, Japan) and examined under a fluorescence microscope using specific filter sets to distinguish green fluorescent protein (GFP) and Annexin-V-Cy3 (Olympus U-MWIBA for GFP and U-MWG for Cy3).
Flow Cytometry-The cell cycle profile was determined by staining DNA with propidium iodide in preparation for flow cytometry with the FACScan/CellQuest system (Becton-Dickinson, San Jose, CA). The size of the sub-G 1 fraction was calculated as a percentage with the Mod-FitLT 2.0 program (Verity Software, Topsham, ME). Annexin-V-positive cells were detected according to the manufacturer's protocol.
DNA Fragmentation Assay-DNA fragmentation was quantitatively measured by detection of cytosolic oligonucleosome-bound DNA with the cell death detection enzyme-linked immunosorbent assay kit (ELISA) (Roche Molecular Biochemicals) (22). Briefly, the cytosolic fraction (13,000 ϫ g supernatant) of the cells was used as an antigen source in a sandwich ELISA with a primary anti-histone antibody coated to microtiter plates and a secondary anti-DNA antibody conjugated with peroxidase. Double absorbance at 405 and 495 nm (A 405 nm / A 495 nm ) was measured against substrate solution as a blank. The data are expressed as the fold increases over the values of untreated exponentially growing cells.
Preparation of Cytosolic Proteins-The cytosolic proteins were isolated according to the method of Yang et al. (23). Briefly, the cell pellets were washed once with ice-cold phosphate-buffered saline and resus-pended in five volumes of buffer A (20 mM Hepes-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride) containing 250 mM sucrose. The cells were homogenized with a Teflon homogenizer and centrifuged twice at 750 ϫ g for 10 min at 4°C. The supernatants were centrifuged at 10,000 ϫ g for 15 min at 4°C, and the resulting mitochondria pellets were resuspended in buffer A containing 250 mM sucrose. The supernatants of the 10,000 ϫ g spin were further centrifuged at 100,000 ϫ g for 1 h at 4°C, and the resulting supernatants were used as cytosol fractions.
Confocal Laser Microscopy-The cells were collected on glass slides using a Cytospin centrifugator (Shandon Scientific, Cheshire, UK) and fixed in 4% paraformaldehyde in phosphate-buffered saline. In situ staining for cytochrome c was carried out as previously reported (24) using anti-cytochrome c monoclonal antibody (6H2.B4; Pharmingen, San Diego, CA) and Alexa 488-conjugated antibody to mouse immunoglobulin (Molecular Probes, Eugene, OR) as primary and secondary antibodies, respectively. For co-labeling of mitochondria, we used two different rabbit polyclonal antibodies: anti-OPA1 and an antibody against 62-kDa mitochondrial protein (AB3598; Chemicon International, Temecula, CA). Anti-OPA1 was generated by immunizing rabbit with the GTPase domain of OPA1 (25), a dynamin-related protein localized at the intermembrane space of mitochondria. 2 Goat antibody to rabbit immunoglobulin conjugated with Cy3 (Amersham Biosciences) was used as a secondary antibody. In situ staining for Bax was performed similarly using anti-Bax (clone 3; Transduction Laboratories) and anti-OPA1 antibodies. Confocal microscopic analysis was done as described previously (26).
Measurement of Mitochondrial Transmembrane Potential-Mitochondrial transmembrane potential was measured with a MitoCapture mitochondrial apoptosis detection kit (Medical & Biological Laboratories). MitoCapture is a rhodamine derivative that aggregates in mitochondria and generates a bright red fluorescence in intact cells (27). When the mitochondrial membrane potential is altered, MitoCapture dye remains in the cytoplasm without aggregation, giving off a green fluorescence. The fluorescent signals were distinguished by flow cytometry using propidium iodide (FL-2) and fluorescein isothiocyanate (FL-1) channels.
Chromatin Immunoprecipitation Assay-A chromatin immunoprecipitation assay was performed according to the method of Moreno et al. (31) with minor modifications. Approximately 5 ϫ 10 7 cells were fixed in growth medium with 1% formaldehyde at room temperature for 10 min. After washing with phosphate-buffered saline, the cells were disrupted in cell lysis buffer (5 mM PIPES, pH 8, 85 mM KCl, and 0.5% Nonidet P-40) containing protease inhibitor complex (Roche Diagnostics, Mannheim, Germany) on ice for 5 min. Then, the nuclei were collected by microcentrifugation, resuspended in nuclei lysis buffer (50 mM Tris-HCl, pH 8, 10 mM EDTA, and 1% SDS) containing protease inhibitor complex, and sonicated on ice at 10 s pulses six times to an average length of 500 -1000 bp. The samples were centrifuged at 14,500 rpm for 10 min, and the supernatants were stored at Ϫ70°C. The chromatin suspensions (100 l each; an equivalent of 5 ϫ 10 6 cells) were added to immunoprecipitation dilution buffer (167 mM NaCl, 16.7 mM Tris-HCl, pH 8, 1.2 mM EDTA, 0.01% SDS, 1.1% Triton X-100, 20 g/ml salmon sperm DNA, and 50 g/ml yeast tRNA) containing 5 g of one of the 1  . After incubation at 4°C for 2 h, the mixtures were further rocked with 20 l of goat anti-rabbit IgG-conjugated magnetic beads (Miltenyi Biotec, Auburn, CA) for 1 h, and the immune complexes were recovered in a magnetic separator. The immunoprecipitants were washed three times each with four different buffers, followed by elution with 50 mM NaHCO 3 and 1% SDS. The eluents were heat-treated, ethanol-precipitated, digested with proteinase K, extracted with phenol/chloroform, and finally resuspended in 30 l of Tris/EDTA buffer (pH 8). An aliquot of 3 l of the final suspension was subjected to semiquantitative PCR as described previously (32).

RESULTS
Apoptosis Is a Relatively Late Onset Event in E2F-1-overexpressing Cells-To investigate the mechanisms of E2F-1-induced apoptosis, we established a K562 subline (KpEf1) in which full-length E2F-1 is inducible by the addition of IPTG using the LacSwitch mammalian expression system. Upon the addition of IPTG, E2F-1 protein increased more than 5-fold after 6 h of culture, confirming the inducible overexpression of E2F-1 (Fig. 1, lower panel). Using this cell line, we first examined the effects of E2F-1 on the cell cycle profile and apoptosis by flow cytometry. As a control, we simultaneously analyzed a K562 stable transformant carrying an empty vector (Mock), which did not overexpress E2F-1 in the presence of IPTG (Fig.  1, upper panel). At 12 h after the addition of IPTG, a significant increase in the number of cells in the S phase was observed in an E2F-1-overexpressing clone but not in mock-transfected cells, followed by a relative increase in the number of cells in the G 2 /M phase at 24 h. After 72 h, apoptosis, as judged by the appearance of a sub-G 1 fraction in DNA histograms, was readily induced in KpEf1 cells, whereas apoptosis was not observed in the control (Fig. 1) and K562 cells overexpressing either E2F-4 or E2F-6 (data not shown). The induction of apoptosis by E2F-1 was confirmed by other methods including DNA fragmentation ELISA and Annexin-V staining (data not shown). These results suggest that among E2F family mem-bers, E2F-1 specifically induces cell cycle progression and apoptosis simultaneously but with different kinetics, and the induction of apoptosis is a relatively late onset event.
Involvement of Caspases-9, -6, and -3 in E2F-1-induced Apoptosis-To determine the pathways of apoptotic signal transduction activated by E2F-1, we examined the effects of specific inhibitors of caspases on E2F-1-induced apoptosis. KpEf1 cells were pretreated with cell-permeable caspase-binding peptides and then cultured in the presence of IPTG. The percentage of cells undergoing apoptosis was measured by calculating sub-G 1 fractions on DNA histograms (Fig. 2, A and B) and DNA fragmentation ELISA (Fig. 2C) after 72 h. E2F-1-induced apoptosis was significantly suppressed by a general caspase inhibitor VAD as well as specific inhibitors for caspases-9, -6, and -3.
To confirm the activation of caspases during E2F-1-mediated apoptosis, we examined the expression and cleavage of caspases in E2F-1-overexpressing cells by immunoblotting. As shown in Fig. 3, the amount of procaspase-9 (48 kDa) decreased after 24 h of culture with IPTG in KpEf1 cells with active fragments of caspase-9 (37 kDa/17 kDa) appearing after 48 h. We simultaneously measured the activity of caspase-9 using colorimetric assay kits and found that it indeed increased in E2F-1-overexpressing cells in accord with the processing of procaspase-9 (data not shown). With similar kinetics, procaspases-6 (34 kDa) and -3 (32 kDa) were processed into a large active subunit (17 kDa) in these cells. No apparent effect on other initiator (caspase-8 and -2) and executioner (caspase-7) caspases was observed. These results suggest that E2F-1 activates caspase-9, which in turn processes caspases-6 and -3 to initiate a caspase cascade, leading to apoptotic cell death.
E2F-1-induced Caspase-9 Activation Coincides with Up-regulation of Apaf-1-The late onset of caspase-9 activation indicates that E2F-1 is not directly involved in the process but can initiate a caspase cascade through transcriptional regulation of the molecules that activate caspase-9. The candidates for such molecules include CARD-containing adapters, Apaf-1 (28) and Nod-1/CARD4 (29,33), and an inhibitor of apoptosis-binding factor Smac/DIABLO (30,34). We screened the expression of these genes in E2F-1-overexpressing clones by Northern blotting and found that E2F-1 increased the abundance of Apaf-1 mRNA after 24 h of culture (Fig. 4A). In addition, Nod-1/ CARD4 mRNA was transiently up-regulated in an E2F-1-dependent manner, and the level of Smac/DIABLO transcript increased over time in both mock-transfected and E2F-1-overexpressing cells. To verify the induction of Apaf-1 by E2F-1, we examined the expression of Apaf-1 protein in E2F-1-overexpressing clones by Western blotting. Apaf-1 protein was barely detectable in untreated KpEf1 cells but was readily induced 48 h after the addition of IPTG (Fig. 4B).
It has been shown that Apaf-1 activates caspase-9 by interacting with procaspase-9 and promoting its oligomerization and cleavage in the presence of dATP and cytochrome c (35,36). When mitochondria are damaged by certain apoptotic stimuli such as anticancer drug treatment, cytochrome c is released from the mitochondria into the cytosol, where it triggers oligomerization of Apaf-1 and activation of caspase-9 (23, 37). We therefore examined whether E2F-1 induces cytoplasmic translocation of cytochrome c in association with the increase in Apaf-1. Cytosolic proteins were serially isolated from E2F-1overexpressing cells and subjected to immunoblotting for the evaluation of cytochrome c content. As shown in Fig. 4B, E2F-1 did not increase the amount of cytochrome c in the cytosol up to 96 h of culture, whereas adriamycin immediately induced a cytosolic accumulation of cytochrome c in these cells. To corroborate the result of immunoblotting, we examined the intracellular distribution of cytochrome c using confocal laser mi- croscopy. In untreated KpEf1 cells, cytochrome c was stained in a perinuclear punctate pattern indicative of intramitochondrial localization (Fig. 4C, Control). Intramitochondrial localization of cytochrome c was confirmed by co-labeling with mitochondrial markers. During apoptosis triggered by adriamycin, the distribution of cytochrome c became completely diffuse, reflecting the translocation of cytochrome c into the cytosol (Fig. 4C,  ADR). In contrast, the pattern of cytochrome c distribution remained the same as that of the control in E2F-1-overexpressing KpEf1 cells up to 48 h of E2F-1 induction (Fig. 4C, E2F-1).
Most if not all cells retained identical patterns, even after 72 h of culture when there was the initial signs of cell death, indicating that cytochrome c translocation is not a prerequisite for apoptosis following E2F-1 activation. Next, we measured mitochondrial transmembrane potential using a rhodamine-derivative dye to assess mitochondrial damage. As shown in Fig. 4D, mitochondrial transmembrane potential was not severely affected during E2F-1-induced apoptosis, although it was lost in dead cells at day 3. These data suggest that the E2F-1-induced up-regulation of Apaf-1 is not a direct result of mitochondrial damage or massive cytosolic accumulation of cytochrome c. It is highly likely that Apaf-1 mediates E2F-1-induced apoptosis as a transcriptional target of E2F-1.
Finally, we investigated why E2F-1 did not induce cytoplasmic translocation of cytochrome c in KpEf1 cells. Upon DNA damage, p53 is stabilized by phosphorylation and induces transcriptional activation of Bax, which in turn acts on mitochondrial membrane to trigger the release of cytochrome c (38). We therefore examined the expression of p53 and Bax in KpEf1 cells. As shown in Fig. 5A, p53 was not detectable in this cell line even after the induction of E2F-1 overexpression. We found that the lack of p53 protein was attributable to the defect of mRNA expression caused by a point mutation of the p53 gene (data not shown). Consistent with the absence of p53, the amount of Bax was not increased by E2F-1 in KpEf1 cells (Fig.  5A). Furthermore, we also demonstrated the absence of mitochondrial translocation of Bax using confocal microscopy. Bax was detected in both the cytoplasm and the nucleus in untreated KpEf1 cells, whereas it accumulated in the cytoplasm and became co-localized with mitochondria after adriamycin treatment (Fig. 5B). The change in Bax distribution was not observed after E2F-1 overexpression up to 72 h, indicating that Bax is not directly involved in cell death caused by E2F-1. These data may explain why cytochrome c release does not occur following E2F-1 overexpression in this cell line.

Apaf-1-defective Cells Are Relatively Resistant to E2F-1-induced
Apoptosis-To confirm the involvement of Apaf-1 in E2F-1-induced apoptosis, we examined whether E2F-1 could elicit apoptosis in Apaf-1-defective cells. For this purpose, we first screened the expression of Apaf-1 in human melanoma cell lines, in some of which Apaf-1 is inactivated because of hypermethylation of an enhancer element of the Apaf-1 gene (39), by immunoblotting. Among the five cell lines examined, the Apaf-1 level was very low in G-361 and below the detection limit in SK-MEL-5 (Fig. 6A). E2F-1 was transiently overexpressed in these cell lines by transfecting a bicistronic vector that induces simultaneous expression of E2F-1 and GFP. At 72 h after transfection, we determined the percentage of GFPpositive cells undergoing apoptosis by fluorescence microscopy. The data from three independent experiments are summarized in Fig. 6B. The inducibility of apoptosis by E2F-1 was significantly lower in the two cell lines with little or no expression of Apaf-1 than in the other three melanoma cell lines with intact Apaf-1 expression. There was no difference in the proportion of apoptotic fraction among the five cell lines when an empty pIRES2-EGFP vector was transfected (Fig. 6B). These results indicate that E2F-1-induced apoptosis is at least in part mediated through Apaf-1-dependent pathways.
E2F-1 Binds to a Putative E2F-binding Site of Apaf-1 Promoter-Finally, we examined whether Apaf-1 is directly under the transcriptional regulation of E2F-1. To this end, we isolated the putative promoter region of the Apaf-1 gene by a data base-oriented approach (Fig. 7A). The promoter activity and E2F-1 responsiveness of the isolated region were confirmed by transient transfection of the pCAT basic vector containing PCR-amplified sequences between nucleotides Ϫ1136 and ϩ34 into KpEf1 cells (data not shown). This region was found to be GC-rich; The percentage of G and C for the entire sequence is 66%, and the ratio of observed/expected CpG is 0.79. In partic-FIG. 4. E2F-1 increases Apaf-1 levels without affecting mitochondrial transmembrane potential. A, total cellular RNA was isolated from KpEf1 and mock-transfected K562 cells at the indicated time points after the addition of IPTG and subjected to Northern blotting for Apaf-1, Nod-1, and Smac mRNA expression. Ethidium bromide-stained 28 and 18 S rRNAs are shown as a loading control. B, KpEf1 cells were cultured in the presence of IPTG for up to 96 h. Whole cell lysates and cytosol fractions were prepared at the indicated time points and subjected to immunoblotting for Apaf-1 and cytochrome c, respectively. ␤-Actin was used as a loading control. ADR, cytosol fraction isolated from KpEf1 cells treated with adriamycin at 1 g/ml for 24 h. C, KpEf1 cells were cultured in the absence (Control) or presence of IPTG (E2F-1) for up to 72 h or treated with adriamycin as above (ADR). The cells were double-stained with anti-cytochrome c and anti-mitochondrial protein antibodies as described under "Materials and Methods." Adriamycin incorporated in cellular DNA gave off a bright red fluorescence in the nuclei. D, KpEf1 cells were cultured in the absence (Control) or presence of either IPTG (E2F-1) or adriamycin (ADR) and stained with a MitoCapture dye to measure mitochondrial transmembrane potential at the indicated time points. FL1 and FL2 signals indicate mitochondrial transmembrane potential and the levels of apoptotic cells, respectively. The data shown are representative of multiple independent experiments. ular, the sequence between nucleotides Ϫ316 and Ϫ103 shows an extremely high G and C content (79%) with an observed/ expected CpG ratio of 1.21, indicating that this region constitutes a CpG island or is part of a large CpG island. Because a typical TATA motif is not present in the Apaf-1 promoter, it is likely that methylation of the CpG island serves as the major mechanism of transcriptional regulation of Apaf-1. A search for putative binding sites of known transcription factors revealed that there is a consensus sequence for E2F between nucleotides Ϫ532 and Ϫ526 (Fig. 7A). In addition, there are several GC boxes, some of which are located around the CpG island, and two p53-binding regions at Ϫ765 to Ϫ739 and Ϫ604 to Ϫ572.
We then examined whether E2F-1 binds to the E2F consensus site upon the transcriptional activation of Apaf-1 caused by E2F-1 overexpression in vivo using the chromatin immunoprecipitation assay. As shown in Fig. 7B, E2F-1 bound to the Apaf-1 promoter in KpEf1 cells 24 h after the induction, whereas no binding of other E2F family proteins was observed, except for a small amount of PCR product detected in E2F-4 immunoprecipitants from untreated KpEf1 cells. These results suggest that E2F-1 directly regulates transcription of the Apaf-1 gene in a positive manner, at least in E2F-1-overexpressing cells. DISCUSSION In this study, we investigated the mechanisms of E2F-1induced apoptosis using K562 sublines that can overexpress E2F-1 conditionally. When E2F-1 was overexpressed in these cells, apoptosis was readily induced after 72 h, following a transient increase in the S phase at 12 h. This confirms the notion that E2F-1 promotes cell cycle progression and apoptosis simultaneously to suppress untoward expansion of proliferating cells (3)(4)(5)(6)12).
We found that E2F-1 was capable of activating caspase-9 to initiate a caspase cascade in the absence of mitochondrial damage. The activation of caspase-9 is usually triggered by the release of cytochrome c from damaged mitochondria in response to certain apoptotic stimuli such as anticancer drugs, ultraviolet radiation, and serum deprivation (40). Subsequently, cytochrome c binds to Apaf-1 and, in association with adenine nucleotides, facilitates a conformational change of Apaf-1 to expose its CARD domain (35,36). Apaf-1 oligomerizes through the exposed CARD domain and recruits procaspase-9 by a homophilic interaction involving CARDs, which results in catalytic autoactivation of caspase-9. However, E2F-1 activates caspase-9 through a different pathway; it bypasses the translocation of cytochrome c from mitochondria to cytosol and directly induces autoactivation of caspase-9 via the increase in intracellular Apaf-1 levels. The involvement of other direct activators of caspase-9, including a CARD-containing adapter Nod-1/CARD4 (29,33) and an inhibitor of apoptosis-binding factor Smac/DIABLO (30,34), is to be determined because E2F-1 also modulates the expression of these molecules. Upregulation of Apaf-1 is also observed during apoptosis caused by the adenoviral oncogene E1A, but it occurs as a secondary event of cytoplasmic accumulation of cytochrome c and therefore is distinct from that of E2F-1 (41). Although the mecha- The arrows correspond to the sequences used for PCR primers for chromatin immunoprecipitation assay. The nucleotide sequence shown here has been deposited in the DDBJ/EMBL/GenBank TM data base under accession number AB070829. B, chromatin suspensions were prepared from KpEf1 cells before (T-0) and after 24 h (T-24) of culture with IPTG and subjected to immunoprecipitation with the indicated antibodies. The resulting precipitants were subjected to PCR using a specific primer pair corresponding to the nucleotide positions of Ϫ618 to Ϫ597 and Ϫ428 to Ϫ407 of the Apaf-1 promoter. PCR was carried out in the presence of [ 32 P]dCTP, and the amplified products were visualized by autoradiography after 8% polyacrylamide gel electrophoresis. The representative data of 30 PCR cycles are shown. No DNA, PCR was done without DNA templates. Input, prior to the first wash, 200 l of the supernatant was saved for each time point and used for PCR after proteinase treatment and ethanol precipitation. nism by which caspase-9 is activated by the increase in Apaf-1 is at present unclear, a high concentration of Apaf-1 may increase the probability of physical interaction between procaspase-9 molecules even in the absence of mitochondrial damage, resulting in oligomerization and subsequent autoactivation of procaspase-9. This hypothesis is substantiated by studies in which forced expression of Apaf-1 solely promoted apoptosis in human myeloid leukemia cell lines (42,43).
Finally, we provide evidence that Apaf-1 is a direct transcriptional target of E2F-1 by analyzing the 5Ј-untranslated regions of the Apaf-1 gene. This finding is consistent with the results of a recent effort to identify E2F-responsive genes by global analysis of gene expression using high density oligonucleotide arrays; Apaf-1 is one of the newly identified E2F-1-inducible genes (44). Interestingly, E2F-2 and E2F-3 did not up-regulate Apaf-1 in this analysis, which is consistent with the fact that E2F-1 has the strongest ability in promoting apoptosis among E2F family members (5). However, there are some reports suggesting that other E2F proteins, especially E2F-2 and E2F-4, are also capable of inducing apoptosis (45,46). The mechanisms underlying apoptosis induced by E2F-2 and E2F-4 may be different from that of E2F-1, because E2F-2 and E2F-4 cannot transactivate Apaf-1 (44,47). Because E2F-4 was not able to induce apoptosis in our system (data not shown), it is likely that E2F-4-mediated apoptosis involves p53-dependent pathways.
The structure of the Apaf-1 promoter is characteristic; it lacks either a typical TATA motif or initiator elements and is devoid of canonical binding sites for known transcription factors except E2F, Sp-1, and p53. Instead, the Apaf-1 promoter shows an extremely high G and C content (Ͼ60%) with an observed/expected CpG ratio of greater than 0.8, which fulfills the requirement of a CpG island (21). These features suggest that methylation of the CpG island serves as the major mechanism of transcriptional regulation of Apaf-1. Indeed, Apaf-1 expression is defective in approximately half of melanoma cell lines, which is reversed by 5-aza-2Ј-deoxycytidine, an inhibitor of methylation (39). This implies that the Apaf-1 gene is silenced by methylation of the CpG island, and the modulation of the methylation status alleviates the repression to induce Apaf-1 transcription. In this study, we demonstrated the binding of E2F-1 to the Apaf-1 promoter in E2F-1-overexpressing cells using the chromatin immunoprecipitation assay. This result indicates that E2F-1 participates in the derepression of the Apaf-1 gene probably by affecting promoter methylation. The mechanisms by which E2F-1 modulates the methylation status of the Apaf-1 promoter are currently under investigation in our laboratory.
While this manuscript was being prepared, an almost identical observation was published by Moroni et al. (48). Moreover, Fortin et al. (49) reported that p53 up-regulates Apaf-1 as a direct transcriptional target during neuronal cell death. Notably, the former study clearly demonstrated that cytoplasmic translocation of cytochrome c is accompanied by E2F-1-induced apoptosis, which contradicts our present finding. This contradiction may stem from the difference in the p53 status of host cells used in each study; p53 expression is defective in our cell line, whereas Moroni et al. (48) used primary human fibroblasts and U2OS cells, both of which possess normal p53. Although our finding may be unique to p53-deficient cells, it instead makes two important points clear. First, E2F-1 can transactivate the Apaf-1 gene in the absence of p53. This is clinically important because most cancer cells lost the function of p53. Second, it is suggested that the E2F-1/Apaf-1 pathway (growth-associated intrinsic route) works in parallel with the p53/Bax/cytochrome c pathway (DNA damage-triggered extrinsic route) in promoting apoptotic cell death.
In conclusion, these data demonstrate a novel mechanism of apoptosis in which an increase in Apaf-1 levels induced by E2F-1 results in direct activation of caspase-9 without mitochondrial damage, leading to the initiation of a caspase cascade. Because both E2F-1 and Apaf-1 are frequently deregulated in various pathologic conditions, this finding may contribute to a better understanding of the pathophysiology of many diseases.