Dysfunctional Apoptosome Activation in Ovarian Cancer

INTRODUCTION

Epithelial ovarian cancer is the leading cause of death of women with gynecological cancers and is the fourth leading cause of cancer death among women in the United States. Primary cytoreductive surgery followed by platinum based chemotherapy (e.g., cisplatin, carboplatin) is the standard of care for treatment of ovarian cancer. This treatment is associated with high response rates, but the disease is characterized by recurrence and the subsequent development of resistance to chemotherapy. Only a minority of patients survive 5 years. Resistance to chemotherapy has been associated with decreased susceptibility to apoptosis, introducing the possibility that cell death determinants may influence the outcome of treatment (1) .

The family of mammalian caspases (aspartate-specific cysteine proteinases) represents the effector arm of the apoptotic program. Caspases are expressed in cells as inactive precursors, which are processed following stimulation with apoptotic signals into mature forms composed of a tetramer of two large and two small subunits (2) . Apoptosis occurs via a proteolytic cascade in which upstream initiator caspases (caspase-8, -9, and -10) activate downstream effector caspases (caspase-3, -6, and -7; Refs. 2, 3, 4 ). In vivo, this results in the cleavage of target proteins and the demise of the cell. The initiator caspases are activated through intrinsic autocatalytic activity in conjunction with other proteins with which they form complexes known as apoptosomes (5) .

Two distinct pathways leading to apoptosis have been delineated. Both converge on the activation of downstream caspases. The extrinsic pathway involves ligation of death receptors such as Fas or tumor necrosis factor receptor-1 by their ligands, resulting in the recruitment of procaspase-8 or procaspase-10 to the receptor complex through the adaptor Fas-associated death domain (2) . Following oligomerization, procaspase-8 or procaspase-10 are self-activated presumably through induced proximity of procaspase molecules (6) . These active species then cleave their downstream targets, which include effector caspases. A second pathway independent of death receptors leads first to mitochondrial release of cytochrome c. In the presence of dATP/ATP, cytochrome c binds to Apaf-1, 3 which then oligomerizes and binds procaspase-9, resulting in the formation of an apoptosome containing Apaf-1 and caspase-9 (5 , 7) . Biochemical studies have shown that Apaf-1 oligomerizes to a M_r 700,000-1,400,00 complex that contains both procaspase-9 and processed caspase-9 (7, 8, 9) . There is some controversy over whether the apoptosome also contains fully processed caspase-3 and caspase-7 (7 , 9) . Apaf-1 oligomerization allows proximity and enzymatic self-activation of bound caspase-9. There is controversy over whether the enzymatic activity of caspase-9 is attributed to cleaved caspase-9 bound to the apoptosome (8) or if processed caspase-9 that has been released from the apoptosome is also enzymatically active (10) .

Evaluation of the function of the apoptosome complex in tumor specimens has been limited. Recently, investigators have shown that Apaf-1 is inactivated in malignant melanoma and that its inactivity may contribute toward the aggressive, chemoresistant phenotype associated with this disease (11) . In the present study, we assess the role of the Apaf-1/caspase-9 apoptosome in epithelial ovarian cancer cell lines as well as in primary ovarian tumors.

MATERIALS AND METHODS

Cell Lines and Immunoblotting.

Ovarian cancer cell lines A2780, CaOV3, MDAH 2774, and SKOV3, and Jurkat cells were obtained from the American Type Culture Collection. Heat shock of Jurkat cells was performed as described (12) . Cytosolic extracts were prepared as described (12) . Immunoblotting analysis was performed with relevant antibodies. Apaf-1 was detected with a monoclonal antibody obtained from Yuri Lazebnik (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). Monoclonal antibodies for caspase-9, Hsp70, and Hsp90 were obtained from Stressgen (Victoria, Canada). The monoclonal antibody for XIAP was obtained from BD Transduction (Franklin Lakes, NJ). The monoclonal antibody for β-tubulin was obtained from Sigma Chemical Co.-Aldrich (St. Louis, MO). Western blots were developed by enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ). Densitometry analysis was performed using Scion Image Beta 4.02 software from Scion Corp (Frederick, MD).

In Vitro Chemotherapy-induced Cell Death and Caspase Activation Assays.

Ovarian cancer cells were maintained in DMEM supplemented with 10% fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 mg/ml; Life Technologies, Inc., Carlsbad, CA). Cells were plated, and 24 h after plating, 10 μg/ml cisplatin (Bristol-Myers Squibb, Princeton, NJ) was added to the culture medium. The percentage of apoptotic cells was determined at the indicated time points by nuclear propidium iodide staining as described previously (14) . In this assay, apoptotic cells exhibit a sub-Go DNA content resulting from DNA fragmentation (15) . For caspase activation assays, cells were harvested with trypsin and then double stained with propidium iodide and a carboxyfluorescein-labeled caspase substrate for caspase-9 (FAM-LEHD-FMK; Intergen, Purchase, NY). Processing of this substrate was measured by flow cytometry. Analysis was performed on a FACScan flow cytometer (Becton Dickinson, Mountain View, CA).

Patient Sample Preparation.

Patient samples were obtained from patients with primary or recurrent epithelial ovarian cancer admitted to the University of Michigan Medical Center. In conjunction with institutional review board approval, samples were obtained at the time of primary surgery or at the time of paracentesis for diagnosis of recurrent disease. For solid tumor samples, the tumor tissue was minced and incubated with collagenase type 3 (Worthington Biochemical Corp, Lakewood, NJ) at 37°C. Cellular material was pelleted, washed with M199 medium, RBCs were lysed by hypotonic lysis, and cell pellet aliquots were frozen in liquid nitrogen. For ascites samples, cells were pelleted and processed as described above. Primary ovarian surface epithelial cells were collected from normal ovaries using the scrape technique as described (16) . Cytosolic lysates were prepared as described (17) .

In Vitro Caspase-9 and Caspase-3 Assay.

Cytosolic extracts were prepared from cell lines and primary tumor cultures as described (17) . The in vitro caspase-9 and caspase-3 assays were performed as described previously (17 , 18) . Reactions were stopped with 5 × SDS loading buffer [250 mm Tris-Cl (pH 6.8), 500 mm DTT, 10% SDS, 0.5% bromophenol blue, and 50% glycerol], boiled, and loaded onto a 12% polyacrylamide/SDS gel. Gels were dried and exposed for autoradiography.

Reverse Transcription-PCR Analysis of Cell Lines and Human Samples.

RNA was extracted from human embryonic kidney 293T, A2780, MDAH 2774, CaOV3, and SKOV3 cell lines, and from human ovarian cancer samples using a commercially available kit (Qiagen, Valencia, CA). cDNA was prepared using these samples as templates according to protocols provided with a commercially available kit (Promega, Madison, WI). Apaf-1 cDNAs were detected using the specific primers: C1, 5′-CAGCTGATGGAACCTTAAAGC-3′and C2, 5′-GTCTGGTCATCAGAAGATGTC-3′, and PCR conditions as described (18) .

Fractionation of Cytosolic Extracts by Gel Filtration.

S-100 cytosolic extracts were prepared as described (17) and incubated at 30°C for 60 min in the presence or absence of 10 μg/ml cytochrome c and 1 mm dATP. Lysate (2 mg) was loaded on a Superdex-200 HR gel filtration column (Amersham Pharmacia Biotech) pre-equilibrated with Buffer A [20 mm HEPES-KOH (pH 7.5), 10 mm KCl, 1.5 mm MgCl₂, 1 mm EDTA, 1 mm EGTA, 1 mm DTT, and 0.1 mm phenylmethylsulfonyl fluoride] at a flow rate of 0.5 ml/min using a Bio-Rad BioLogic HR Workstation. The column was calibrated with an Amersham Pharmacia Biotech HMW gel filtration protein standards kit (thyroglobulin M_r = 669,000, ferritin M_r = 440,000, catalase M_r = 232,000, and BSA M_r = 66,000). After discarding the majority of the void volume, fractions of 400 μl were collected. Aliquots of 50 μl from each fraction were resolved by SDS-PAGE followed by immunoblotting with monoclonal antibodies for Apaf-1, caspase-9, XIAP, Hsp70, and Hsp90.

RESULTS

Differential Response to Cisplatin-induced Apoptosis and Caspase Activation in Ovarian Cell Lines.

We tested the sensitivity of A2780, MDAH 2774, CaOV3, and SKOV3 ovarian cell lines to cisplatin-induced apoptosis. After incubation with the drug (10 μg/ml), the apoptotic fraction was determined by flow cytometry at 24 and 48 h after treatment. By 48 h, cisplatin induced apoptosis in 50–70% of cells in the A2780, MDAH 2774, and CaOV3 cell lines (Fig. 1A) ⇓ . In contrast, almost 90% of the SKOV3 cells were resistant to cisplatin-induced apoptosis after 48 h of treatment (Fig. 1A) ⇓ . To determine whether chemotherapy-induced apoptosis correlates with caspase activation, cells were analyzed for caspase-9 activity. By 24 h, 12–56% of A2780, 2774, and CaOV3 cells demonstrated caspase-9 activity, which increased to 35–56% after 48 h of treatment (Fig. 1B) ⇓ . However, in the chemoresistant SKOV3 cell line, <3% of the cells showed detectable caspase-9 activity even after 48 h of treatment (Fig. 1B) ⇓ . To directly assess caspase-9 activation, cytosolic extracts from cell lines were immunoblotted for caspase-9 after treatment with cisplatin. A2780, MDAH 2774, and CaOV3 cells all demonstrated cleavage of endogenous caspase-9 indicated by the presence of the intermediate p35 proteolytic fragment (Fig. 1C) ⇓ . In contrast, treatment of SKOV3 cells with cisplatin resulted in minimal processing of endogenous caspase-9 (Fig. 1C) ⇓ .

The Inability to Undergo Cisplatin-induced Caspase-9 Activation Correlates with Deficient Apaf-1 Activity in Ovarian Cell Lines.

Apaf-1 is a cytosolic factor that activates caspase-9 (4) . We next tested whether the differential response of ovarian cell lines to cisplatin-induced caspase-9 activation correlated with the ability to process procaspase-9 in vitro as a surrogate for Apaf-1 activity. Cytosolic extracts were incubated with in vitro translated ³⁵S procaspase-9, and caspase cleavage was determined after addition of cytochrome c and dATP. As shown in Fig. 2A ⇓ , cytosolic extracts from 293T, A2780, 2774, and CaOV3 cells induced the activation of procaspase-9 in a cytochrome c and dATP dependent fashion as indicated by the appearance of the intermediate p35 proteolytic fragment. In contrast, extracts from SKOV3 did not activate procaspase-9 (Fig. 2A) ⇓ .

To assess the function of endogenous caspase-9 in the cell lines, we next tested the ability of the cytosolic lysates to process in vitro translated ³⁵S procaspase-3. As shown in Fig. 2B ⇓ , cytosolic extracts from 293T, A2780, 2774, and CaOV3 cells induced the activation of procaspase-3 in a cytochrome c- and dATP-dependent fashion. Extracts from SKOV3 did not activate procaspase-3.

We performed immunoblotting to determine the levels of Apaf-1 in the ovarian cell lines. Apaf-1 was detected in all of the cell lines tested (Fig. 2C) ⇓ . Expression levels of Apaf-1 varied slightly but did not correlate with the ability of the cytosolic extracts to activate procaspase-9, procaspase-3, or with the pattern of chemosensitivity.

Cell Extracts from Primary Ovarian Tumors Are Deficient in Apaf-1 Activity.

We next examined the ability of cell extracts derived from primary ovarian tumors to activate procaspase-9 using the in vitro assay. Twenty-eight ovarian carcinoma samples were harvested and cytosolic extracts were prepared. Thirteen samples were derived from patients with primary ovarian cancer before treatment with chemotherapy, and 15 samples were harvested from patients with recurrent ovarian cancer after treatment with chemotherapy. Sixteen samples were obtained as ascites, and 12 samples were obtained as solid tumor biopsies. Twenty-one of 28 samples analyzed (77%) were deficient in Apaf-1 activity in that they demonstrated no detectable ability to activate procaspase-9 in the presence of cytochrome c and dATP (Fig. 3A) ⇓ . Cytosolic extracts from control 293T cells and normal ovarian epithelial cells activated procaspase-9 (Fig. 3A) ⇓ . Five of 13 of the primary ovarian tumor samples and 2 of 15 of the recurrent ovarian tumor samples induced the activation of procaspase-9 in a cytochrome c- and dATP-dependent fashion.

Similarly, a subset of patient samples was analyzed for the ability to cleave in vitro-translated ³⁵S procaspase-3 in the presence of cytochrome c and dATP (Fig. 3B) ⇓ . Three patient samples with the ability to cleave radiolabeled procaspase-9 (Fig. 3A ⇓ , #1, 4, and 6) and 4 patient samples that demonstrated no ability to cleave radiolabeled procaspase-9 (Fig. 3A ⇓ , #24, 33, 39, and 40) were selected. Two of the 3 samples that demonstrated procaspase-9 activation were not able to cleave procaspase-3. All of the samples that did not process procaspase-9 also did not cleave procaspase-3. Although a larger number of patient samples will need to be analyzed to make definitive conclusions about the clinical significance of deficient apoptosome activity in this disease, these results suggest that in the majority of primary and recurrent ovarian cancers, the apoptosome is nonfunctional.

Apaf-1 Levels Do Not Correlate with Ability to Activate Caspase-9 in Primary Ovarian Tumors.

We next determined the levels of Apaf-1 protein in extracts of the primary ovarian tumors to determine whether differences in the amount of Apaf-1 present explains the results presented above. Immunoblotting analysis showed that Apaf-1 was expressed in all of the primary tumor samples (Fig. 4A) ⇓ . Although the amounts of Apaf-1 varied, protein levels did not correlate with cytosolic activation of procaspase-9 (Fig. 3) ⇓ .

Several isoforms of human Apaf-1 have been described based on differential splicing of the Apaf-1 gene (19) . Apaf-1XL contains an insertion of 43 amino acids creating an additional WD-40 repeat in the COOH terminus (Fig. 4B) ⇓ required for functional activity (18) . Therefore, we determined whether ovarian cell lines and primary tumors express sequences encoding this isoform. As seen in Fig. 4C ⇓ , all of the tumor samples analyzed expressed both Apaf-1 isoforms (with and without the additional WD-40 repeat), although the transcript corresponding to the Apaf-XL isoform was most abundant. This pattern of Apaf-1 expression is similar to that observed in normal human tissues (18) . Although differences in relative amounts of Apaf-1 isoforms were seen between tumor samples, these differences were not predictive of functional or defective Apaf-1 activity.

Expression of Heat Shock Proteins and XIAP Does Not Correlate with Functional Activity of the Apoptosome in Ovarian Cancer.

Recently, several investigators have described negative regulation of the Apaf-1 apoptosome complex by Hsp70 and Hsp90 (12 , 13) . Hsp70 appeared to bind to Apaf-1 but not to procaspase-9, preventing recruitment of caspases to the apoptosome complex (12) . Hsp90 appeared to inhibit cytochrome c-mediated oligomerization of Apaf-1 and thereby activation of procaspase-9 (13) . XIAP has been shown to inhibit caspase-9 processing of caspase-3 by binding to and inhibiting processed mature caspase-9 (20) . Immunoblot analysis of 293T, A2780, MDAH 2774, and CaOV3 cell lines revealed expression of Hsp70, Hsp90, and XIAP; however, expression levels of these proteins did not correlate with functional activity of Apaf-1 (Fig. 5, A, B, and E) ⇓ . Similarly, immunoblot analysis of primary ovarian carcinoma samples revealed expression of Hsp70, Hsp90, and XIAP in all of the samples, but again, expression levels did not correlate with functional activity of Apaf-1 (Fig. 5, C, D, and F) ⇓ .

SKOV3 Cells Retain the Ability to Form Apaf-1 Oligomers.

To determine whether the nonfunctional apoptosome activity of SKOV3 cells (inability to cleave in vitro translated caspase-9) is explained by defective formation of the approximately M_r 700,000–1,400,00 Apaf-1 oligomer described previously (7, 8, 9) , we used size exclusion chromatography to fractionate cytosolic extracts from two representative ovarian cancer cell lines. After activation with cytochrome c and dATP to induce Apaf-1 oligomerization or a similar treatment in the absence of cytochrome c and dATP as a control, fractions separated by gel filtration chromatography were analyzed for Apaf-1 expression by immunoblotting with anti-Apaf-1 antibody. In the absence of cytochrome c and dATP, both A2780 and SKOV3 cell lines demonstrated Apaf-1 elution in fractions corresponding to ∼M_r 200,000 protein, which is consistent with the monomeric form of this protein (Fig. 6, B and D) ⇓ . Incubation of extracts with cytochrome c and dATP resulted in a shift in the Apaf-1 elution profile so that the majority of Apaf-1 eluted in fractions 3–7, corresponding to approximately M_r 700,000–1,400,00 (Fig. 6, A and C) ⇓ . These results indicate that the nonfunctional apoptosome activation of SKOV3 is not attributable to the inability of Apaf-1 to oligomerize in response to cytochrome c and dATP. Preliminary experiments using extracts from representative patient samples show similarly that in samples with functional (#4) and nonfunctional (#17) apoptosome activation, Apaf-1 is able to oligomerize in response to cytochrome c and dATP (data not shown).

SKOV3 Cells Express a Diminished Amount of Caspase-9 in the Apoptosome.

The apoptosome has been characterized as an Apaf-1-containing complex that recruits and activates procaspase-9, which then activates caspase-3 (19) . To determine whether the Apaf-1 apoptosome contains activated caspase-9, immunoblot analysis was performed on the fractions containing the Apaf-1 oligomer for the cell lines A2780 and SKOV3. As seen in Fig. 7A ⇓ , in the A2780 cell line after dATP activation, ∼50% of the procaspase-9 was processed to its catalytically active large subunits (p35/p37), which eluted with the approximately M_r 700,000–1,400,000 apoptosome complex. Fractions 15–19 contain procaspase-9, and cleaved caspase-9 that has been released from the Apaf-1 apoptosome. In contrast, in the SKOV3 cell line, a negligible amount of procaspase-9 was processed to its catalytically active large subunits (p35/p37) and subsequently coeluted with the approximately M_r 700,000–1,400,000 apoptosome complex (Fig. 7C) ⇓ . Most of the caspase-9 remained in its proform and eluted in fractions 15–19. These results suggest that in the chemoresistant SKOV3 cell line, dysfunctional apoptosome activation is associated with a decrease in the amount of cleaved caspase-9 bound to the Apaf-1 apoptosome.

To determine whether cleaved caspase-9 released from the Apaf-1 apoposome remains enzymatically active and can subsequently cleave caspase-3, fractionated S100 cytosolic lysates from the cell lines A2780 and SKOV3 were incubated with in vitro translated ³⁵S-labeled caspase-3 in the presence of cytochrome c/dATP. In the A2780 cell line, only fractions 3–7 containing the Apaf-1 oligomer as well as cleaved caspase-9 were able to cleave caspase-3. Fractions 15–19, containing monomeric caspase-9 along with cleaved caspase-9 that had been released from the Apaf-1 apoptosome, did not cleave caspase-3. In the SKOV3 cell line, none of the fractions were able to cleave caspase-3. These findings are consistent with previous work by investigators that showed that cleaved caspase-9 is active and can activate downstream caspase-3 only when it is bound to the Apaf-1 apoptosome (8) .

As mentioned above, several investigators have reported negative regulation of the Apaf-1 apoptosome complex by Hsp70, Hsp90 (12 , 13) , and XIAP (20) . To determine whether these proteins play a role in the function of the Apaf-1 apoptosome complex in ovarian cancer, cytosolic fractions from A2780 and SKOV3 activated with cytochrome c and dATP were separated by gel filtration chromatography and were analyzed for expression of Hsp70, Hsp90, and XIAP by immunoblotting. Expression of Hsp70 was equivalent between the A2780 and SKOV3 extracts across all of the fractions. Hsp90 and XIAP did not coelute with the Apaf-1 apoptosome (fractions 4–6) in either cell line, with the majority eluting in fractions 6–11. These results suggest that Hsp70, Hsp90, and XIAP do not interfere with the formation of the Apaf-1 oligomer in ovarian cancer (data not shown).

DISCUSSION

Alterations in the regulation of apoptosis may contribute both to the pathogenesis and the development of chemoresistance in ovarian cancer. Chemotherapy-induced apoptosis is thought to occur via assembly of a cytochrome c-dependent apoptosome complex containing the proteins Apaf-1 and caspase-9 (3) . Several studies have indicated that cells lacking procaspase-9 or Apaf-1 exhibit resistance to chemotherapy (11 , 21, 22, 23) . However, evaluation of the expression of Apaf-1 and procaspases -2, -3, -7, -8, and -9 in leukemia patients did not correlate with clinical response to therapy (24) . In this study, we show that the majority of primary ovarian tumors and a subset of ovarian cell lines are deficient in Apaf-1 activity. In tumor specimens, processing of procaspase-9 was only detected in 7 of 28 samples. Recognizing that processed caspase-9 may be enzymatically active only when retained in the apoptosome (8) , we additionally analyzed a subset of those tumor specimens that cleaved procaspase-9 to determine whether they were capable of processing procaspase-3. Interestingly, 2 of 3 samples were unable to process procaspase-3. The absence of procaspase-3 activation suggested that although procaspase-9 was cleaved, it was not retained in the apoptosome complex.

In the ovarian cell lines, deficient Apaf-1 activity in vitro was linked to reduced caspase-9 activation in vivo and reduced susceptibility to cisplatin-induced apoptosis. The regulation of caspase-9 activity is complex, but it is thought to be primarily controlled by Apaf-1. Cytochrome c (released from damaged mitochondria) and dATP/ATP are critical factors that regulate the activity of Apaf-1 (4) . Because our assay for Apaf-1 activity was performed in the presence of cytochrome c and dATP, our results suggest that deficient Apaf-1 activity in ovarian tumors is attributable to a defect in the apoptotic pathway downstream of cytochrome c release. However, it is possible that in ovarian tumor cells additional defects contribute to deregulation of the Apaf-/caspase-9 pathway. For example, certain ovarian tumors overexpress Bcl-x_L (25) , which inhibits caspase-9 activation, by preventing the release of cytochrome c from mitochondria (26 , 27) .

One possible explanation for deficient Apaf-1 activity in ovarian tumors is the presence of altered forms of Apaf-1 or caspase-9. Other investigators have described the presence of an alternatively spliced short form of caspase-9 (termed caspase-9b) that acts as a dominant-negative fashion (28) . However, immunoblotting analysis of the ovarian cancer cell lines tested did not reveal the presence of any short isoform of caspase-9 (Fig. 1C) ⇓ . Deletion or silencing of Apaf-1 has been reported in metastatic melanomas (11) . However, in our work all of the tumors and cell lines evaluated expressed Apaf-1 protein. It is possible that ovarian tumors contain subtle genetic alterations such as point mutations that impair the activity of Apaf-1. Despite this possibility, other investigators have failed to find consistent Apaf-1 mutations in a variety of cancers including lymphoma, and pancreatic, germ cell, cervical, or colorectal carcinomas (29 , 30) . Alternative splicing can influence the ability of Apaf-1 to activate caspase-9 (18) . Isoforms expressing 12 WD-40 COOH-terminal repeats demonstrate a decreased cytochrome c-/dATP-dependent caspase activation as compared with isoforms containing 13 WD-40 repeats (18) . All of the ovarian cancer cell lines analyzed expressed RNA encoding Apaf-1 isoforms with both the 12 and 13 WD-40 COOH-terminal repeats, suggesting that the deficient Apaf-1 activity is not attributable to the absence of the functional isoform of Apaf-1.

Inhibitors of the apoptosome including Hsp70, Hsp90, and XIAP have been described recently (12 , 13 , 20) . Deregulated expression of any of these inhibitors could explain the deficiency of Apaf-1 activity in ovarian tumors. Hsp70 has been shown to bind to Apaf-1 but not to procaspase-9, and it appears to prevent the recruitment of caspases to the apoptosome complex (12) . Hsp90 has been shown to inhibit cytochrome c-mediated oligomerization of Apaf-1 and subsequent activation of procaspase-9 (13) . XIAP has been shown to bind to and inhibit processed, mature caspase-9 (20) . Expression levels of these proteins in ovarian cell lines and primary tumors did not correlate with functional activity of Apaf-1 (Fig. 5) ⇓ . Moreover, Hsp70 did not coelute with the Apaf-1 oligomer in either the A2780 or the SKOV3 cell line, and despite the expression of Hsp90 in both cell lines, the Apaf-1 oligomer still formed normally (data not shown; Fig. 6 ⇓ ). Similarly, XIAP did not coelute with the Apaf-1 oligomer in either the A2780 or the SKOV3 cell line (data not shown).

In the chemoresistant SKOV3 cell line, which demonstrates deficient Apaf-1 activity, we show that the ability to form the Apaf-1 apoptosome is retained; however, there is a decrease in the amount of cleaved caspase-9 bound to the Apaf-1 apoptosome. There is some controversy concerning the enzymatic activity of processed caspase-9 that has been released from the apoptosome (8 , 10) . Our data indicates that only cleaved caspase-9 bound to the Apaf-1 apoptosome can activate downstream caspase-3 (Fig. 7) ⇓ . The regulation and activation of caspase-9 in the apoptosome complex is poorly understood. There are a number of possibilities to explain the reduction in the amount of cleaved caspase-9 (and subsequently the diminished enzymatic activity of caspase-9) bound to the Apaf-1 apoptosome in the SKOV3 cell line. Failure of recruitment of caspase-9 to the apoptosome could explain the phenotype seen in this cell line. Such a factor has been described (e.g., Hsp70); however, our results indicate that Hsp70 is not involved in the deficient apoptosome activity seen in ovarian cancer. It is possible that another inhibiting factor that interferes with the recruitment of procaspase-9 to the apoptosome is overexpressed in ovarian cancer. Another possibility is that procaspase-9 is normally recruited to the apoptosome but is not retained. Our data demonstrates that only procaspase-9 that is recruited, activated, and retained in the apoptosome is enzymatically active. We see evidence of processed caspase-9 that has been released from the apoptosome and coelutes with the monomer form of procaspase-9 in the SKOV3 cell line (Fig. 7) ⇓ . Thus, another possibility is that that there may be an unknown cytosolic factor in ovarian cancer that either fails to retain activated caspase-9 in the apoptosome or that accelerates release of activated caspase-9 from the apoptosome.

In summary, our studies show that the majority of primary ovarian tumors and a subset of ovarian cell lines are deficient in Apaf-1 activity. Deficient Apaf-1 activity is linked with resistance to cisplatin-induced apoptosis in an ovarian cancer cell line. The deficient Apaf-1 activity in ovarian cancer may be attributed to a reduction in the recruitment of caspase-9 to the apoptosome or a diminished ability to retain caspase-9 in the apoptosome. This mechanism, which leads to failed apoptosis, will need to be confirmed in primary tumor samples; however, it may play an important role in resistance to chemotherapy in the clinical setting. Additional experiments to elucidate the molecular explanation for deficient Apaf-1 activation in ovarian cancer are under way.