Apoptosis Induction by Activator Protein 2α Involves Transcriptional Repression of Bcl-2*

Activator protein 2α (AP-2α) induces cytotoxicity by inducing cell cycle arrest and apoptosis. In this study we investigated the mechanism of apoptosis induction by AP-2α. We found that AP-2α induced apoptosis efficiently in cells treated with benzyloxycar-bonyl-IETD-fluoromethyl ketone or FADD-silenced cells but failed to do so in benzyloxycarbonyl-LEHD-fluoromethyl ketone-treated or apoptosis protease activation factor-1 (Apaf1)-silenced cells, suggesting the central role of mitochondria in AP-2α-induced apoptosis. In good correlation, cells overexpressing AP-2α showed a reduction in mitochondrial membrane potential (Δψm), cytochrome c and Smac/DIABLO release into cytosol, and Bax translocation into mitochondria. We found that the pro-apoptotic protein Bax is important for AP-2α-induced apoptosis as adenovirus AP2 failed to induce apoptosis in HCT116 Bax–/– cells. However, we found the IAP (inhibitor of apoptosis) inhibitor Smac/DIABLO may have a limited role in AP-2α-induced apoptosis as we found the IAP member Survivin down-regulated by AP-2α. Although the total Bax level remains unaltered, we found a time-dependent increase in the activated form of Bax in adenovirus AP2-infected cells. In addition, we show that AP-2α transcriptionally represses Bcl-2 by binding to its promoter both in vitro and in vivo and that this is essential for AP-2α-induced apoptosis as ectopic expression of Bcl-2 efficiently inhibited apoptosis induced by AP-2α. Furthermore, we show that chemotherapy-induced endogenous AP-2α down-regulates Bcl-2 and induces apoptosis in an AP-2α-dependent manner. Moreover, we demonstrate that inhibition of okadaic acid or staurosporine-sensitive pathways in AP-2α overexpressing breast cancer cells resulted in AP-2α-dependent apoptosis induction. These results suggest that AP-2α induces apoptosis by down-regulating Bcl-2 and utilizing a bax/cytochrome c/Apaf1/caspase 9-dependent mitochondrial pathway.

served proline and aromatic amino acid-rich helix transactivation domain near the N terminus and binds to a consensus DNA sequence, 5Ј-GCCNNNGGC-3Ј (4 -8). AP-2␣ has been shown to regulate many genes involved in variety of biological functions (9).
Several lines of evidences indicate that AP-2␣ may act as a tumor suppressor gene. AP-2␣ gene is located in chromosome position 6p22, a region of frequent loss of heterozygosity in breast and other cancers (10). Diminished AP-2␣ function has been correlated with N-Ras oncogene-mediated transformation (11). The functions of AP-2␣ have been shown to be regulated by SV40 T antigen and adenovirus E1A oncoproteins (1,12). In addition, reduced or loss of AP-2␣ expression has been reported in human cancers of breast, ovary, colon, skin, brain, and prostate (13)(14)(15)(16)(17)(18)(19)(20). In good correlation, expression of dominant negative mutant AP-2␣ resulted in increased invasiveness and tumorigenicity (21).
Overexpression of AP-2␣ by transient transfection into cultured cells has been shown to induce p21 WAF1/CIP1 and inhibit cellular DNA synthesis and colony formation (22). Significant correlation between AP-2␣ expression and p21 WAF1/CIP1 has been observed in breast cancer, colorectal cancer, and malignant melanoma (15,18,23). The growth inhibitory activity of AP-2␣ has also been shown to be mediated through direct interaction with p53 (24). Results from our laboratory suggest that adenovirus-mediated overexpression of AP-2␣ inhibits growth of cancer cells by inhibiting cellular DNA synthesis and inducing apoptosis (25). Furthermore, our recent work establishes that AP-2␣ is induced in cancer cells upon treatment with chemotherapeutic drugs, which contributes to chemosensitivity because the simultaneous inhibition of AP-2␣ by siRNA increases the chemoresistance. In addition, the re-expression of epigenetically silenced AP-2␣ in breast cancer resulted in enhanced chemosensitivity and loss of tumorigenicity upon chemotherapy in an AP-2␣-dependent manner (26). These results point out the importance of apoptosis induction by AP-2␣ for its functions, particularly the role in chemosensitivity.
The molecular mechanism by which AP-2␣ induces apoptosis is not known. In the present study we have analyzed the pathways and various molecules involved in AP-2␣-induced apoptosis. We found that AP-2␣-induced apoptosis requires primarily the mitochondrial pathway involving a bax/cytochrome c/Apaf1/caspase 9-dependent mechanism. We also found that AP-2␣ binds to the Bcl-2 promoter leading to its transcriptional down-regulation and is essential for the apoptosis induction by AP-2␣. In addition, we provide evidence that the overexpressed AP-2␣ (perhaps functionally inactive) in certain breast cancer cells can be made functional by inhibiting survival signaling pathways sensitive to okadaic acid and staurosporine.

EXPERIMENTAL PROCEDURES
Plasmids, Adenoviruses, siRNA, Cell Lines, and Culture Conditions-pCEP4/AP-2␣ was generated by cloning AP-2␣ cDNA as a BamHI/ HindIII fragment originally derived from pSG5/AP-2␣, which was kindly provided by Dr. P. Kannan (Metro Health Medical Center, Case Western Reserve University). pD5neo-bcl-2 was obtained from Dr. S. Pervaz Hussain, NIH, through Dr. Apurva Sarin, National Center for Biological Sciences, Bangalore. Human cancer cell lines SW480, HCT116 WT, and H460 were described previously (25,27). HCT116 Bax Ϫ/Ϫ cells were kindly provided by Dr. Bert Vogelstein (Johns Hopkins University) (28). Ad-LacZ and Ad-AP2 lack both E1A and E1B but carry ␤-galactosidase and AP-2␣, respectively (25). FADD and Smac/DIABLO-specific siRNA were obtained from Dharmacon as siGENOME TM SMARTpool reagent, which contains a pool of four different double-stranded RNA oligonucleotides (siRNA) directed against the particular gene. Apaf1 siRNA is described previously (29) and was also bought from Dharmacon as a duplex. Lamin siRNA was purchased as siGLO Lamin A/C siRNA (human/mouse/rat) from Dharmacon. The cells were transfected using Oligofectamine (Invitrogen) at a 100 nM concentration of siRNA. Bcl-2 promoter construct, Bcl-2prom (1.9 kilobases)-Luc, (Ϫ1291 to ϩ610) was a kind gift of Dr. Fritz Aberger, University of Salzburg, Salzburg, Austria (30). GST-AP-2␣ DNA binding domain was constructed by cloning the region of AP-2␣ between amino acids 165 and 437 into TA cloning vector from which it was released as BamHI and then fused in-frame with GST in pGEX-5X-1.
MTT Assay and Stable Colony Suppression Assay-MTT assay was carried out as described previously (25). For colony suppression assay, the cells were transfected with pCEP4 and pCEP4/AP-2␣ plasmids and selected on hygromycin (250 g/ml).
Generation of Bcl-2 Stable Cell Lines-SW480 cells were transfected with Bcl-2 plasmid (pD5neo-bcl-2) and selected for G418 resistance (800 g/ml). Medium was changed periodically every third day. Resistant colonies were cloned and screened for Bcl-2 expression by Western blot analysis. Positive clones were checked for clonal purity and used for further experiments.
Preparation of Mitochondrial and Cytosolic Fractions-Cells were lysed in buffer (250 mM sucrose, 20 mM HEPES, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors, which contained antipain, leupeptin, pepstatin A, and chymostatin). Lysis buffer was added and incubated for 30 min on ice. In addition the cells were homogenized by 30 strokes in a 22-guage needle. Homogenates were centrifuged at 750 ϫ g for 10 min at 4°C. Supernatants were centrifuged at 10,000 ϫ g for 15 min at 4°C to collect the mitochondrial pellets. The supernatant after mitochondrial pelleting was used as a cytosolic fraction.
Luciferase Assay-Transfections were carried out either by using Escort III (Sigma) as per the manufacturer's recommendations. In all transfections 1 g of pCMV-LacZ DNA was added to normalize the transfection efficiency variation between samples. Luciferase and ␤-galactosidase assays were performed as described previously (22).
Chromatin Immunoprecipitation Assay-Chromatin immunoprecipitations were performed as described previously (34). Briefly, subconfluent SW480 cells that were infected with Ad-AP2 for 24 h were treated with formaldehyde at a final concentration of 1% for 7 min at AP-2␣ Represses Bcl-2 to Induce Apoptosis room temperature. Chemical cross-linking was terminated by the addition of glycine to a final concentration of 0.125 M followed by additional incubation for 5 min. After a wash with cold phosphate-buffered saline, cells were suspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 1% SDS, and 10 g/ml antipain, leupeptin, pepstatin A, and chymostatin and kept at 4°C for 10 min. The chromatin was then sonicated as described (34) so that the total DNA is sheared to 200 -1000 bases in length. After centrifugation, the supernatant was diluted 10-fold with dilution buffer (16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl, 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, and 10 g/ml antipain, leupeptin, pepstatin A, and chymostatin). Diluted chromatin was precleared with protein A-agarose beads saturated with bovine serum albumin (1 mg/ml) and salmon sperm DNA (500 g/ml) and then incubated overnight at 4°C using anti-AP-2 (C-18, Santa Cruz) or anti-mouse lamin B (Oncogene) and immunoprecipitated with protein A-agarose beads. The beads were extensively washed, and then chromatin was eluted from beads by incubation in elution buffer (0.1 M NaHCO 3 , 1% SDS) for 2 h. Cross-links were then reversed by overnight incubation at 65°C in elution buffer containing 300 mM NaCl and 30 g of RNase A/ml. A 20% equivalent amount of diluted chromatin was similarly processed without immunoprecipitation and noted as "input" afterward. DNA samples were then purified by phenol-chloroform extraction and ethanol precipitation and further analyzed by PCR for the region of interest. The primers used to amplify a 173-bp fragment spanning Ϫ129 to ϩ44 of Bcl-2 promoter are as follows: upstream primer, 5-CGGTTGGGATTCCTGCGGATT-3Ј, and downstream primer, 5-AATTGCATAAGGCAACGATCCC-3Ј.

AP-2␣ Induces Apoptosis through an Intrinsic Pathway-AP-2␣
overexpression is highly cytotoxic to cancer cells as it inhibits cellular DNA synthesis leading to G 1 cell cycle arrest and induces apoptosis. To study the relative role of intrinsic versus extrinsic pathway in AP-2␣-induced apoptosis, we examined the cells infected with AP-2␣ adenovirus (Ad-AP2) for activated forms of major effector caspases and their importance. In all our experiments where recombinant adenoviruses were used, a control adenovirus expressing ␤-galactosidase (Ad-LacZ) was also used. Infection of H460 human lung carcinoma cells with Ad-AP2 resulted in efficient induction of apoptosis as seen by the appearance of cells with less than a 2N amount of DNA, representing apoptotic cells, by 48 h of virus infection (Fig. 1A) and efficient PARP cleavage in a time-dependent fashion (Fig. 1B). AP-2␣ overexpression activated caspase 3, 8, and 9 as seen by the appearance of cleaved forms of these caspases in Ad-AP2-but not in Ad-LacZ-infected cells (Fig. 1C). In good correlation, AP-2␣-induced caspase activation also coincided with PARP cleavage (Fig. 1C). We also found that caspase activation is essential for AP-2␣-induced apoptosis as the pan caspase inhibitor, z-VADfmk, inhibited apoptotic cell death in Ad-AP2-infected cells ( Fig. 2A). In addition, z-VAD-fmk treatment completely reversed the suppression of colony formation by AP-2␣ ( Fig. 2B; compare pCEP4/AP-2␣ ϩ z-VADfmk with pCEP4/AP-2␣).
Because both caspases 8 and 9 were activated in Ad-AP2-infected cells, we next determined the relative roles of the extrinsic (caspase 8-mediated) and intrinsic (caspase 9-mediated) pathways in AP-2␣induced apoptosis. To address this point, we used caspase 8-and 9-specific inhibitors as well as an siRNA approach. The caspase 9-specific inhibitor (z-LEHD-fmk), but not caspase 8-specific inhibitor (z-IETDfmk), efficiently inhibited AP-2␣-induced apoptosis as seen by the reduced percentage of cells with less than a 2N content of DNA ( Fig.  2A). In good correlation, colony suppression by AP-2␣ is completely reversed by z-LEHD-fmk but not by z-IETD-fmk (Fig. 2B). These results suggest the primary utilization of caspase 9, but not caspase 8, by AP-2␣ to induce apoptosis. To confirm this fact, we analyzed the role of FADD (the adapter protein, which acts upstream of caspase 8) and Apaf1 (which normally complexes with cytochrome c and caspase 9 to form an active apoptosome complex) in AP-2␣-induced apoptosis. We found that AP-2␣ is able to induce apoptosis efficiently in cells where the expression of FADD is silenced by siRNA (see supplemental Fig. S1). On the other hand, AP-2␣ failed to induce apoptosis in Apaf1-silenced cells (see below). Transfection of Apaf1-specific siRNA reduced the levels of Apaf1 transcript but not lamin transcript (Fig. 2C, lane 3) and Apaf1 protein (Fig. 2D, compare lane 4 with lane 3). Similarly, lamin siRNA reduced the lamin but not Apaf1 transcript (Fig. 2C, lane 2) and Apaf1 protein (Fig. 2D, lane 3). Neither siRNA inhibited the level of GAPDH transcript (Fig. 2C) and actin protein (Fig. 2D). Although Ad-AP2 induced apoptosis in the cells transfected with lamin-specific siRNA efficiently, it failed to do so in the cells transfected with Apaf1-specific siRNA (Fig. 2E). The percentage of apoptotic cells in Ad-AP2 virusinfected cells was reduced from 49.48% in the lamin siRNA-transfected group to 6.23% apoptotic cells in the AP-2␣ siRNA-transfected group at 72 h after virus infection. These results suggest that caspase 8-mediated extrinsic pathway is dispensable, whereas caspase 9-dependent intrinsic pathway is required in AP-2␣-induced apoptosis.
Essential Role of Mitochondria in AP-2␣-induced Apoptosis-To understand better the importance of the intrinsic pathway of apoptosis, we investigated the role of mitochondria in AP-2␣-induced apoptosis by measuring mitochondrial membrane potential (⌬ m ) as a measure of mitochondrial outer membrane permeabilization (MOMP) by using the potentiometric dye JC-1 (33) in Ad-AP2-infected cells. JC1 is a fluorescent cationic dye that aggregates in mitochondria in healthy cells and fluoresces red, whereas in apoptotic cells, wherein the mitochondrial membrane potential is reduced, it is diffused throughout the cells and assumes a monomeric form that fluoresces green. Because Ad-AP2 expresses green fluorescent protein also, which would interfere with JC1-based measurement of mitochondrial membrane potential, we decided to overexpress AP-2␣ in the tetracycline-based (Tet-off) conditional expression system. AP-2␣ expressed from the tetracycline-responsive promoter in stable SW480 cell clones was induced by infecting them with an adenovirus expressing tetracycline responsive transcriptional activator (Ad-tTA) in the absence of tetracycline (26). Overexpression of AP-2␣ in SW480 cells resulted in 2.5-fold reduction in ⌬ m in comparison to control cells (Fig. 3A, compare bar 4 with bar 3). As expected, treatment of SW80 cells with valinomycin, which was used as a positive control because it induces apoptosis by disrupting ⌬ m , resulted in 2.5-fold reduction in ⌬ m (Fig. 3A compare bar 2 with bar 1).
MOMP results in release of soluble mitochondrial intramembrane proteins such as cytochrome c and the IAP antagonist Smac/DIABLO (35,36). To monitor the release of mitochondrial proteins into cytosol in AP-2␣ overexpressing cells, we separated cytosolic and mitochondrial extracts from Ad-LacZ-and Ad-AP2-infected cells by biochemical fractionation and monitored the release of cytochrome c and Smac/ DIABLO into cytosol. AP-2␣ overexpression resulted in release of cytochrome c and Smac/DIABLO into cytoplasm (Fig. 3B) as against control cells infected with Ad-LacZ. The quality of fractionation was verified by monitoring the specific subcellular markers (␤-actin as a cytoplasm-specific marker and oxidative complex 1 protein as a mitochondria-specific marker).
Next we sought to find out the importance of cytochrome c and Smac/ DIABLO release into cytosol for AP-2␣-induced apoptosis. Because our results (see above) suggest that Apaf1 and caspase 9, which requires complex formation with cytosolic cytochrome c for their function, are essen-  tial for AP-2␣-induced apoptosis, we conclude that cytochrome c release is essential for AP-2␣-induced apoptosis. However, we found that the IAP inhibitor Smac/DIABLO may have a very limited role in AP-2␣-induced apoptosis (see below). The apoptosis induction by AP-2␣ in cells where Smac/DIABLO is silenced by siRNA was reduced by 33-45% (see supplemental Fig. S2). In addition, we found that AP-2␣ down-regulated the expression of IAP protein Survivin, thus making Smac/DIABLO dispensable for AP-2␣-induced apoptosis (see supplemental Fig. S2). Thus, these results suggest that mitochondria play a major role in AP-2␣-induced apoptosis.
Role of Bax in Apoptosis Induction by AP-2␣-Because MOMP is mainly mediated and controlled by Bcl-2 family members (36), we then analyzed the role of pro-apoptotic Bcl-2 family member Bax. First we determined whether AP-2␣-induced MOMP is accompanied by Bax relo-  (Fig. 3B). In addition, to find out the importance of Bax in AP-2␣-induced apoptosis, we analyzed the ability of AP-2␣ to induce apoptosis in HCT116 bax Ϫ/Ϫ , which is a somatic . Total RNA was isolated after 24 h of infection and subjected to quantitative real time RT-PCR analysis for AP-2␣, Bcl-2, and GAPDH transcripts. The -fold change for each gene was calculated as the ratio between specific transcript levels in the Ad-AP2-and Ad-LacZ-infected samples and was normalized to the levels of GAPDH transcript. The normalized -fold change for AP-2␣ in H460 and SW480 cells was found to be 38.80 and 12.13, respectively (data not shown). C, total RNA was made from the Tet-Off AP-2␣ clone at the indicated time points in the presence (uninduced) or absence (induced) of tetracycline (1 g/ml) and subjected to quantitative real-time RT-PCR analysis for AP-2␣, Bcl-2, and GAPDH transcripts. The -fold change for each gene was calculated as the ratio between specific transcript levels in the Ϫtet and ϩtet conditions for each time points and was normalized to the levels of GAPDH transcript. D, SW480 colon cancer cells were transfected with 5 g of Bcl-2prom (1.9 kilobases) (indicated by ϩ) and pSG5 vector or pSG5/AP-2␣ expression plasmid at indicated amounts. Transfections were carried out by using Escort III (sigma). Lysates were prepared and analyzed for luciferase reporter activity 24 h post-transfection as described in under "Experimental Procedures." pUC18 plasmid DNA was added to keep the total amount of DNA constant per transfection. 1 g of the plasmid expressing ␤-galactosidase was transfected in each well and was further used as an internal control for normalization of transfection efficiency variations. E, electrophoretic mobility shift assay was performed using purified GST-AP-2 DNA binding domain (amino acids 165-437) as described in under "Experimental Procedures." Competitions were performed by adding increasing amounts (0.2, 1, and 2 pmol) of specific (Bcl-2 wt and SV40 AP2 wt) or nonspecific (SV 40 AP2 m) competitor as indicated (left panel). On the right competition was carried out by adding increasing amounts (0.2, 1, and 2 pmol) of specific (Bcl-2 wt) or nonspecific (Bcl-2 m) competitor. F, chromatin from cross-linked subconfluent Ad-AP2-infected cells was immunoprecipitated with antibodies specific for AP-2␣ (lane 5) or nonspecific antibody lamin B (lane 4). bcl-2 sequence was detected by PCR analysis of eluted DNA. A known amount of chromatin from each sample was removed before immunoprecipitation and used for PCR amplified (Input; lane 3). Plasmid DNA carrying Bcl-2 promoter (Bcl-2prom (1.9 kilobases)) was used as a positive control for PCR (lane 2). A schematic of the amplified region Bcl-2 promoter is shown on the right. GLI-bs, glioma-associated oncogene homolog 1 binding sequence.
knockout for both alleles of Bax and is derived from HCT116 cells (28). We confirmed the absence of bax protein in HCT116 Bax Ϫ/Ϫ cells by Western blot analysis (Fig. 3C; compare lane 2 with lane 1). Ad-AP2 infection resulted in efficient apoptosis induction in HCT116 cells but not in the Bax Ϫ/Ϫ derivative as evidenced by reduced PARP cleavage (Fig. 3D) and reduced sub-G 1 cells (Fig. 3E) upon Ad-AP2 virus infection. The percentage of apoptotic cells is 58.80% in Ad-AP2-infected HCT116 cells as against 5.98% in Ad-AP2-infected HCT116 Bax Ϫ/Ϫ cells by 72 h (Fig. 3E). These results suggest that Bax is important for AP2␣-induced apoptosis.

AP-2␣ Induces Apoptosis by Transcriptionally Down-regulating Bcl-2-
The relative amounts or equilibrium between the pro-and anti-apoptotic proteins of Bcl-2 family is believed to influence the susceptibility of cells to a death signal (37). With the hypothesis that AP-2␣ may

. Ectopic expression of Bcl-2 inhibits AP-2␣-induced apoptosis.
A, SW480 cells were either transfected with the indicated combinations of pCEP4 (1 g), pCEP4/AP2␣ (1 g), pCDNA3 (10 g), and pCDNA3/Bcl-2 (pD5neo-bcl-2) (10 g) plasmid DNA. Note that Bcl-2 cDNA (pCDNA3/Bcl-2) or the empty expression vector (pCDNA3) was added in 10 times excess. All the transfected samples were selected for hygromycin resistance, which is encoded by pCEP4 vector with periodic medium change for 2 weeks. The plates were then stained with Coomassie Blue, the colonies containing greater than 50 cells were counted from five representative fields, and the average values are shown. B, total cell lysates from clones of SW480 cells stably transfected with either pCDNA3 (lane 1) or Bcl-2 vector (lanes 2 and 3) were subjected to Western blot analysis for Bcl-2 and actin proteins. C, SW480/Neo or SW480/Bcl-2 #3 cells were infected with Ad-LacZ or Ad-AP2 at a 20 m.o.i. Cells were harvested at the indicated time points and subjected for Western blot analysis for AP-2␣, PARP, and actin proteins. D, SW480/Neo or SW480/Bcl-2 #3 cells were infected with Ad-LacZ or Ad-AP2 a 20 m.o.i. Cells were harvested at the indicated time points and subjected to FACS analysis. Apoptotic population is quantitated and shown at the right.
transcriptionally regulate Bcl-2 family members to induce apoptosis, we determined the levels of various Bcl-2 family members in AP-2␣ overexpressing cells. Because our results suggest that Bax is essential for AP-2␣-induced apoptosis, we first examined the levels of Bax in Ad-AP2-infected cells. We found no change in Bax transcript levels (data not shown) and Bax protein levels (Fig. 3F) between Ad-LacZ-and Ad-AP2-infected cells. It has been shown that translocation of Bax into the outer mitochondrial membrane involves a conformational change that exposes the N terminus and the hydrophobic C terminus that targets mitochondria (38). To examine the possibility that Bax might be activated by conformational change in AP-2␣ overexpressing cells, we performed immunoprecipitation experiments with anti-Bax 6A7 antibody that recognizes only the conformationally changed Bax protein. As seen in Fig. 3F, Bax underwent a conformational change at 24 h after Ad-AP2 virus infection, which further increased in a time-dependent fashion. Bax conformational change in AP-2␣ overexpressing cells also correlated well with the release of cytochrome c into cytosol, which occurs by 24 h, and it precedes apoptosis induction, which occurs at 48 h after virus infection as seen by the appearance of apoptotic cells documented by flow cytometry, PARP cleavage, and caspase 3, 8, and 9 activation ( Fig. 1, A-C). Thus, these results indicate that, although Bax protein level does not change, Bax undergoes a conformational change to form an active form in AP-2␣-overexpressing cells, thus playing a major role in AP-2␣-induced apoptosis.
We then focused our attention on other members of Bcl-2 family belonging to both anti-apoptotic and BH3-only groups. Although most of them remain unaltered in their transcript levels in AP-2␣ overexpressing cells, we found that the level of Bcl-2 protein, the prototype anti-apoptotic member of Bcl-2 family, is reduced drastically in a timedependent fashion in SW480 cells overexpressing AP-2␣ (Fig. 4A). We also found that Bcl-2 transcript level is decreased in H460 and SW480 cells upon Ad-AP2 virus infection by 1.95-and 2.96-fold, respectively (Fig. 4B). In addition, upon overexpression of AP-2␣ in a tet-off-inducible system, Bcl-2 transcripts decreased 1.59-fold by 12 h, which further decreased to 1.95-fold by 24 h after tetracycline removal (Fig. 4C). The time-dependent decrease in Bcl-2 transcript levels also correlated well with the increasing levels of AP-2␣ transcripts after tetracycline removal (Fig. 4C). The transcriptional repression of Bcl-2 by AP-2␣ appears to be due decreased promoter activity as exogenous expression of AP-2␣ decreased the Bcl-2 promoter activity in a dose-dependent manner (Fig. 4D).
Next, to determine whether transcriptional repression of Bcl-2 by AP-2␣ is due to a direct effect, we analyzed the Bcl-2 promoter for the presence of AP-2␣ binding sequence. We found an AP-2 binding motif 5Ј-CTCTCCCCGC-3Ј at Ϫ48 to Ϫ39 of Bcl-2 promoter. We carried out electrophoretic mobility shift assay with 32 P-end-labeled Ϫ62 to Ϫ39 Bcl-2 probe and bacterially expressed GST-AP-2␣ fusion. The AP-2 binding sequence located between Ϫ62 to Ϫ39 of Bcl-2 promoter bound to GST-AP-2␣ (Fig. 4E, left panel, lane 2) and is specific as it is competed by unlabeled Bcl-2 oligonucleotide as well as by oligonucleotides containing the AP-2 binding site of the SV40 enhancer but not by oligonucleotides containing mutated SV40 AP-2 binding site (Fig. 4E,  left panel, compare lanes 3-5, 6 -8, and 9 -11 with lane 2, respectively). The specificity of AP-2␣ binding to Ϫ48 to Ϫ39 Bcl-2 promoter sequence is also confirmed by the fact that the unlabeled Bcl-2 oligo carrying mutated AP-2 binding motif (5Ј-CTCTCCCCGC-3Ј is mutated to 5Ј-TATTCCCAAA-3Ј) failed to compete efficiently (Fig.  4E, right panel, compare lanes 6 -8 with 3-5). In addition, to find out whether AP-2␣ binds Bcl-2 promoter in vivo, we analyzed a 173-bp fragment that spans Ϫ129 to ϩ44 carrying the AP-2 binding sequence ( Fig. 4F) by chromatin immunoprecipitation assay. In SW480 cells infected with Ad-AP2, AP-2␣ bound to this fragment efficiently and is specific as a nonspecific antibody (lamin) failed to bind (Fig. 4G, compare lanes 5 and 4). These results conclusively prove that AP-2␣ transcriptionally represses Bcl-2 by binding directly to its promoter.
To find out whether Bcl-2 down-regulation is necessary for AP-2␣induced apoptosis, we determined whether ectopic expression of Bcl-2 could block cytotoxicity and apoptosis induced by AP-2␣. First, we checked the effect of Bcl-2 on the ability of AP-2␣ to suppress the colony formation. Although AP-2␣ expression vector alone efficiently inhibited colony formation, it failed to inhibit colony formation when cotransfected along with Bcl-2 expression vector (Fig. 5A, compare bar  3 with bar 4). As expected, Bcl-2 expression vector alone did not inhibit colony formation (Fig. 5A, bar 2). We then tested the ability of Ad-AP2 to induce apoptosis in Bcl-2 stable cell lines derived from SW480 cells. Two independent Bcl-2 stable clones of SW480 (#3 and #15) expressed increased amounts of Bcl-2 in comparison to a vector stable clone (SW480/Neo) (Fig. 5B, compare lanes 2 and 3 with lane 1). AP-2␣ failed to induce apoptosis efficiently in the Bcl-2 stable cell line as seen by reduced PARP cleavage and reduced apoptotic cells (Fig. 5, C and D). Although Ad-AP2-infected SW480/Neo cells showed efficient PARP cleavage by 48 h, SW480/Bcl-2 #3 clone, upon Ad-AP2 infection, showed no PARP cleavage by 48 h and very minimal PARP cleavage by   (Fig. 5C). Similarly, Ad-AP2 infection resulted in a severalfold increase in apoptotic cells in SW40/Neo cells in a time-dependent manner ( Fig. 5D; 0.78% at 24 h is increased to 24.89% by 48 h and 42.91% by 72 h). However, Ad-AP2 infection failed to induce any significant apoptosis in SW480/Bcl-2 #3 clone ( Fig. 5D; 2.10% at 24 h is increased to 5.73% by 48 h and 6.54% by 72 h). As expected in both cell lines, Ad-LacZ infection did not result in apoptosis induction (Fig. 5D). Taken together, we conclude that down-regulation of Bcl-2 is essential for AP-2␣-induced apoptosis.
Chemotherapy-induced Endogenous AP-2␣ Down-regulates Bcl-2 and Induces Apoptosis-We have previously shown that AP-2␣ is an important determinant of cancer cell chemosensitivity (26). Both chemotherapyinduced endogenous AP-2␣ and methylation inhibitor (5aza2dC)-induced re-expression of endogenous AP-2␣ modulated the chemosensitivity. To demonstrate the physiological relevance of apoptosis induction by AP-2␣, we analyzed apoptosis induction by AP-2␣ upon chemotherapy treatment or 5aza2dC treatment. Upon treatment of SW480 cells with adriamycin or cisplatin resulted in increase in endogenous AP-2␣ (Fig. 6A Okadaic Acid or Staurosporine Treatment Induces AP-2␣-dependent Apoptosis in Breast Cancer Cells-An inefficient proteosomal degradation pathway has been found to be responsible for high levels of AP-2␣ in certain breast cancer-derived cells (39). We hypothesized that the growth inhibitory functions of AP-2␣ is inhibited in these cells by survival signaling, inhibition of which might restore its functions. We used the breast cancer cell line MDA-MB-453 cell line, which carries elevated levels of AP-2␣ (Fig. 7, C and D, lane 1) as previously reported (39). We found that treatment of MDA-MB-453 cells with staurosporine or okadaic acid resulted in apoptosis induction, as evidenced by an increase in the percentage of cells with less than 2N content of DNA (Fig. 6, A

DISCUSSION
In this study we demonstrate that AP-2␣-induced apoptosis utilizes mainly the mitochondrial pathway of apoptosis, which is demonstrated by the following experiments. We found that the addition of caspase 9 inhibitor or siRNA silencing of Apaf1 completely abrogated the apoptosis induction by AP-2␣, whereas the addition of caspase 8-specific inhibitor z-IETD-fmk or siRNA silencing of the adapter protein FADD did not have any significant effect on AP-2␣-induced apoptosis. In addition, AP-2␣-overexpressing cells showed MOMP, which is accompanied by reduction in mitochondrial membrane potential (⌬ m ), release of cytochrome c and Smac/DIABLO into cytosol, and translocation of Bax into mitochondria. Although cytochrome c is found to be essential, Smac/DIABLO may play a limited role in AP-2␣-mediated apoptosis. Moreover, we found AP-2␣ failed to induce apoptosis in cells lacking Bax. Although the total Bax levels remain unchanged in AP-2␣-overexpressing cells, we found the presence of conformationally changed active form of Bax. We also found that AP-2␣ binds to the Bcl-2 promoter, specifically leading to transcriptional repression of Bcl-2, and is essential for AP-2␣-induced apoptosis as ectopic expression of Bcl-2 blocked the same. Finally, we show that in cancer cells overexpressing non-functional AP-2␣, apoptosis could be induced by blocking okadaic acid-or staurosporine-sensitive survival signaling pathways in an AP-2␣dependent manner.
The apoptotic signaling pathway that leads to caspase activation can be subdivided into two major categories: death receptor-mediated and mitochondria-mediated pathways (40,41). We provide multiple evidence for the intrinsic pathway along with mitochondria that play an important role in AP-2␣-induced apoptosis. Neither caspase 8 inhibitor nor siRNA silencing of death receptor adaptor protein, FADD, significantly affected AP-2␣-induced apoptosis. On the contrary, caspase 9 inhibitor completely inhibited the apoptosis induced by AP-2␣. In addition, AP-2␣ adenovirus-infected cells showed a reduction in mitochondrial membrane potential, which is suggestive of permeabilization of the mitochondrial outer membrane. This was also correlated with the release of mitochondrial proteins cytochrome c and Smac/DIABLO into cytosol in AP-2␣-overexpressing cells.
The permeabilization of the mitochondrial outer membrane leads to cell death because of the release of soluble mitochondrial intramembrane proteins such as cytochrome c and the disruption of essential mitochondrial functions (36). Upon release from mitochondria, cytochrome c binds to Apaf1, leading to the formation of apoptosome complex, which then recruits procaspase 9 leading to its activation through oligomerization (42) Apart from cytochrome c, another group of proteins released from mitochondria upon MOMP are IAP antagonists including Smac/DIABLO, HtrA2/Omi, and GSPT1/eRF3. Although IAPs act as a brake for apoptosis by inhibiting activated caspases, IAP antagonists bind IAPs to inhibit them in a competitive manner. Smac/ DIABLO is also released along with the cytochrome c during apoptosis to neutralize the inhibitory activity of IAPs and promote cytochrome c-dependent caspase activation (35). Our results indicate that both cytochrome c and Smac/DIABLO are released from mitochondria in AP-2␣-overexpressing cells. However, cytochrome c release seems to be essential for AP-2␣-induced apoptosis as the siRNA-based silencing of its downstream target Apaf1 resulted in the complete abrogation of apoptosis in Ad-AP2-infected cells. Unlike Apaf1, silencing of Smac/ DIABLO resulted in only partial reduction of apoptosis in AP-2␣-over- expressing cells. This suggests the possibility that IAPs may have only a limited role in controlling the apoptosis induced by AP-2␣. Indeed, we found that the transcript levels of Survivin, a member of IAP family, is reduced severalfold in cells overexpressing AP-2␣. Antisense oligonucleotide-based inhibition of Survivin in H460 lung cancer cells has been shown to sensitize the cells to radiation (43). p53 has also been shown to transcriptionally down-regulate Survivin by binding to its promoter, which is believed to contribute to p53-dependent apoptosis (44). Therefore, we conclude that cytochrome c release is essential and IAP inhibitor is less essential as the IAP Survivin is down-regulated by AP-2␣ for apoptosis induction by AP-2␣.
The Bcl-2 family of proteins plays a critical role in the mitochondrial pathway of apoptosis (36). The antiapoptotic Bcl-2 family proteins like Bcl-2 and Bcl-XL inhibit the release of certain pro-apoptotic factors from mitochondria, whereas the pro-apoptotic Bcl-2 family molecules (BH3 only and multidomain BH1-3 proteins) induce the release of mitochondrial apoptogenic molecules into the cytosol. Evidence suggests that the multidomain pro-apoptotic proteins Bax and Bak are absolutely required for apoptosis induction by diverse intrinsic death stimuli (45). Our results clearly demonstrate an important role for Bcl-2 family members in AP-2␣-induced apoptosis. Although the total Bax levels (both transcript and protein) did not change, Bax is found to be activated by conformational change leading to its translocation into mitochondria in Ad-AP2-infected cells. Moreover, we found that Bax is essential for apoptosis induction by AP-2␣. We found the levels of Bcl-2 transcript and protein reduced severalfold in AP-2␣-overexpressing cells. Furthermore, we show that Bcl-2 down-regulation by AP-2␣ involves direct binding AP-2␣ onto an AP-2 binding motif located between Ϫ48 to Ϫ39 of Bcl-2 promoter. We also provide evidence for the fact that Bcl-2 down-regulation by AP-2␣ is necessary for it to induce apoptosis. Forced expression of Bcl-2 into cells, either from a transiently transfected cDNA or stably integrated copy of Bcl-2, resulted in complete abrogation of apoptosis induction by AP-2␣. Therefore, we propose that the mechanism of apoptosis induction by AP-2␣ may involve transcriptional down-regulation of Bcl-2, which indirectly may promote Bax activation. In addition to this, repression of the IAP family member Survivin by AP-2␣ may also play a role in AP-2␣-induced apoptosis. AP-2␣ has been shown to negatively regulate the expression of several genes, some of which involves direct binding of AP-2␣ to their promoter sequences like K3 keratin and vascular endothelial growth factor (19,46). Thus, we conclude that AP-2␣ induces apoptosis by down-regulating Bcl-2 and utilizes mitochondrial pathway of apoptosis induction.
We also provide evidence for physiological relevance of apoptosis induction by AP-2␣. Chemotherapy-induced endogenous AP-2␣ in SW480 cells resulted in Bcl-2 down-regulation and apoptosis induction in an AP-2␣-dependent manner. Similarly, chemotherapy-mediated increase of re-expressed AP-2␣ in 5aza2dC-treated MDA-MB-231 cells resulted in down-regulation of Bcl-2 and apoptosis induction in a AP-2␣dependent manner.
Although the endogenous level of AP-2␣ in most cell lines has been found to be low, increased levels have been reported in certain breast cancer cells (39). Because AP-2␣ is a potent growth inhibitory protein, the high levels seen in these cells might reflect a functionally inactive state. Adenovirus oncogene E1A has been shown to stabilize the tumor suppressor protein p53 while simultaneously inhibiting its transcriptional activation function (47). Similarly, regulation of AP-2␣ at posttranscriptional level has been reported previously. AP-2␣ activity has been shown to be stimulated by 12-O-tetradecanoylphorbol-13-acetate or okadaic acid treatment via a posttranscriptional mechanism (48,49). Similarly, protein kinase A-induced phosphorylation of AP-2␣ has been shown to increase the ability of AP-2␣ to transactivate apolipoprotein E (APOE) promoter (50). Our results show that treatment of MDA-MB-453 cells with okadaic acid or staurosporine resulted in AP-2␣-dependent apoptosis (Fig. 6). Protein phosphatase 2A, which is inhibited by okadaic acid, is known to regulate many kinases particularly involved in survival signaling. For example, protein phosphatase 2A regulates Akt/ protein kinases B and C and mTOR, all of which promote cell survival (51)(52)(53). Similarly, protein kinase C, which is inhibited by staurosporine, has been implicated in neoplastic transformation (54), regulation of apoptosis (55), and multidrug-resistant phenotype (56). Because our results show that okadaic acid and staurosporine induce AP-2␣-dependent apoptosis, it is possible that protein phosphatase 2A-and protein kinase C-regulated growth promoting, anti-apoptotic, and tumorigenic signaling pathways could lead to AP-2␣ functional inactivation in AP-2␣-overexpressing cancer cells. Further studies are in progress to dissect the pathways that leads to inactivation of AP-2␣.