Elucidation of molecular events mediating induction of apoptosis by synthetic retinoids using a CD437-resistant ovarian carcinoma cell line.

Retinoids have great promise in the area of cancer therapy and chemoprevention. Although some tumor cells are sensitive to the growth inhibitory effect of all-trans-retinoic acid (ATRA), many ovarian tumor cells are not. 6-((1-Admantyl)-4-hydroxyphenyl)-2-naphthalenecarboxylic acid (CD437) is a conformationally restricted synthetic retinoid that induces growth arrest and apoptosis in both ATRA-sensitive and ATRA-resistant ovarian tumor cell lines. To better understand the mechanism by which CD437 induces apoptosis in ovarian tumor cell lines, we prepared a cell line, CA-CD437R, from the ATRA-sensitive ovarian cell line, CA-OV-3, which was resistant to CD437. We found that the CD437-resistant cell line was also resistant to the induction of apoptosis by tumor necrosis factor-alpha but not resistant to the induction of apoptosis by another synthetic retinoid, fenretinide N-(4-hydroxyphenyl)retinamide. We also show that this cell line remains ATRA-sensitive and exhibits no deficiencies in RAR function. Analysis of this CD437-resistant cell line suggests that the pathway for induction of apoptosis by CD437 is similar to the pathway utilized by tumor necrosis factor-alpha and different from the pathway induced by the synthetic retinoid, fenretinide N-(4-hydroxyphenyl)retinamide. The CA-CD437R cell line is a valuable tool, permitting us to further elucidate the molecular events that mediate apoptosis induced by CD437 and other synthetic retinoids. Results of experiments utilizing this cell line suggest that the alteration responsible for resistance of CA-CD437R cells to CD437 induced event maps after the activation of p38 and TR3 expression, prior to mitochondrial depolarization, subsequent release of cytochrome c and activation of caspase-9 and caspase-3.

Retinoids are a group of natural and synthetic analogs of vitamin A that have been shown to play an important role in cell differentiation and proliferation (1,2). Because retinoid treatment inhibits the growth of a variety of epithelial cancers, retinoids have great promise in the area of cancer chemother-apy and chemoprevention (3)(4)(5). However, many tumors have been found to be resistant to the growth inhibitory activity of natural retinoids. More recently, several synthetic retinoids such as 6-((1-admantyl)-4-hydroxyphenyl)-2-naphthalenecarboxylic acid (CD437) 1 and fenretinide N- (4-hydroxyphenyl)retinamide (4-HPR) have been shown to inhibit the growth and induce apoptosis in both all-trans-retinoic acid (ATRA)-sensitive and -resistant tumor cell lines. Thus determination of the mechanism by which these synthetic retinoids inhibit growth and induce apoptosis of ovarian tumors could lead to the development of more effective cancer treatments.
CD437 is a conformationally restricted synthetic retinoid that has been studied in a number of cancer models including non-small cell lung carcinoma, cervical, breast, and ovarian carcinomas (6 -15). This retinoid is currently being evaluated in a number of clinical trials. Our laboratory has previously reported that CD437 inhibits growth and induces apoptosis in ATRA-resistant (SK-OV-3) as well as ATRA-sensitive (CA-OV-3) ovarian carcinoma cells lines (6,16). CD437 has been reported to be an RAR-␥-/RAR-␤-selective ligand when used at apoptotic inducing concentrations (10 Ϫ6 M) (6,7,15,17,18). Our previous studies (16) demonstrated that the effects of CD437 on ovarian carcinoma cell growth and induction of apoptosis is at least partially RAR-dependent. 4-HPR, another synthetic retinoid, also induces growth arrest and apoptosis in both the ATRA-sensitive, CA-OV-3, and ATRA-resistant, SK-OV-3, ovarian tumor cell lines. Although, the mechanism of action of 4-HPR is not fully understood, it has been reported that this compound could induce apoptosis and growth inhibition by both RAR-dependent and -independent pathways (19 -25). We also have found that in CA-OV-3 and SK-OV-3 ovarian tumor cells, 4-HPR acts independent of RAR function or activation. 2 In this investigation we describe the isolation and characterization of an ovarian carcinoma cell line, CA-CD437R, which has been made resistant to the apoptotic effects of CD437. This cell line, isolated from the CA-OV-3 ovarian tumor cell line, is resistant to the effects of CD437 treatment, but is not resistant to apoptosis induced by 4-HPR. Interestingly this cell line is also resistant to the apoptotic effects of TNF-␣. Our results suggest that these two retinoids use separate apoptotic inducing pathways. This cell line is a valuable tool, which can be used to compare molecular mechanisms leading to apoptosis induced by CD437 and 4-HPR in ovarian carcinoma cells.

MATERIALS AND METHODS
Cell Lines and Culture Conditions-CA-OV-3 cells were obtained from the American Type Culture Collection (Rockville, MD). Stock cultures of CA-OV-3 and CA-CD437R were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 units/ml penicillin and streptomycin, and 100 units/ml nystatin. All cell lines were routinely split at a ratio of 1:10 weekly. The plates were incubated in an atmosphere of 98% humidified 5% CO 2 .
Retinoids and TNF-␣-ATRA was kindly supplied by Hoffmann-La Roche (Nutley, New Jersey). CD437 was a generous gift of Galderma Research and Development (Sophia Antipolis, France). 4-HPR was generously provided by R. W. Johnson Pharmaceutical Research Institute (Raritan, NJ). Retinoid stock solutions (10 Ϫ3 M) prepared in Me 2 SO were stored at Ϫ70°C. Recombinant human TNF-␣ was purchased from Promega (Madison, WI). All procedures involving retinoids were carried out under subdued light. In each experiment, control cultures were treated with an equivalent amount of Me 2 SO.
Isolation of the CA-CD437R Cell Line-The CA-CD437R cell line was isolated from the CA-OV-3 ovarian tumor cell line. CA-OV-3 cells were cultured for 6 months in complete medium with increasing concentrations of CD437. The concentration of CD437 synthetic retinoid was progressively increased from 10 Ϫ10 M to a concentration of 10 Ϫ6 M over the 6-month period of time. The CA-CD437R cell line has been cultured for over 18 months, during this time every 14 days the medium was supplemented with 10 Ϫ6 M CD437 for a period of 5 days to ensure continued resistance. The CA-CD437R cell line did not revert to a CD437-sensitive phenotype similar to the parental CA-OV-3 cell line upon removal of selective pressure. Before CA-CD437R cells were used in experiments, CD437 selective pressure was applied to stock CA-CD437R cultures, for a period of 48 h. The stock cultures were then trypsinized and plated in complete medium without CD437 as needed for each specific experiment.
Cell Proliferation-Ovarian carcinoma cells were seeded onto 100-mm tissue culture plates at a density of 2 ϫ 10 4 cells/plate. The cells were incubated for 48 h prior to treatment. The medium, with or without varying concentrations of retinoids, was changed every 2 days. On the days indicated, cells were washed with phosphate-buffered saline, trypsinized, and counted using a hemocytometer. Cells were assessed for viability using trypan blue. All cell counts were repeated in triplicate.
Apoptosis ELISA-Detection of apoptosis in ovarian cancer cells was performed via an ELISA as described by Salgame et al. (26). The ELISA uses a capture monoclonal antibody (LG11-2), which is specific for residues 1-25 of the amino-terminal domain of histone H2B, and a biotinylated detection monoclonal antibody (PL2-3), which is specific for the nucleosome subparticle composed of histones H2A, H2B, and DNA. The presence of histones in the cytoplasm is indicative of apoptosis. The ELISA was quantitated by measuring absorbance at 405 nm using a V max Kinetic Reader (Molecular Devices, Sunnyvale, CA).
RAR Transcriptional Activity-RAR transcriptional activity was assayed using a pRARE-CAT reporter plasmid (a generous gift from Dr. R. Evans of the Salk Institute, La Jolla, CA) and a commercially available pCMV-␤-galactosidase plasmid (Clontech). Transient transfection was performed using the Effectene transfection kit from Qiagen (Valencia, CA). Briefly, 5 ϫ 10 5 cells were seeded into 60-mm tissue culture plates in complete DMEM and incubated at 37°C in a humidified CO 2 incubator 1 day prior to transfection. 1 g of pRARE-CAT plasmid DNA and 1 g of pCMV-␤-galactosidase plasmid DNA was diluted in the DNA condensation buffer to a final volume of 150 l. The mixture was centrifuged and mixed with 25 l of Effectene transfection reagent. The medium was then removed from the culture plates, the cells were washed once with phosphate-buffered saline, and 4 ml of fresh complete DMEM was added. 1 ml of complete DMEM was added to the DNA mixture and the mixture was added dropwise onto the cells. The cells were incubated at 37°C for 24 h at which time the cells were treated with either 10 Ϫ6 M ATRA or Me 2 SO carrier. After an additional 24 h the cells were harvested and assayed for chloramphenicol acetyltransferase activity (CAT) (27) and ␤-galactosidase activity (28). CAT was normalized with respect to ␤-galactosidase activity to control for efficiency of transfection. All experiments were repeated in triplicate.
Caspase Activity Assays-The activity of caspase proteases was measured using an ApoAlert CPP32 colorimetric assay kit (Clontech). Briefly, whole cell lysate from ϳ4 ϫ 10 6 ovarian cells was incubated for 1 h at 37°C with the caspase-specific substrate in the provided reaction buffer. Cleavage of the caspase-specific substrate by active caspase resulted in the liberation of pNA (p-nitroanilide) into solution. Release of pNA was quantitated spectrophotometrically by measuring absorbance at 405 nm using a V max kinetic reader (Molecular Devices, Sunnyvale, CA). All experiments were repeated in triplicate. The following specific caspase substrates were utilized in these studies: Ac-DEVD-pNA (caspase-3/7) (Clontech), Ac-VEID-pNA (caspase-6), Ac-IETD-pNA (caspase-8) and Ac-LEHD-pNA (caspase-9) (Biomol, Plymouth Meeting, PA). Caspase Inhibitor Studies-Ovarian carcinoma cells were seeded onto 100-mm tissue culture plates at a density of 2 ϫ 10 4 cells/plate. The cells were incubated for 48 h prior to treatment. 1 h prior to retinoid treatment, 20 g/ml of the caspase-3-like inhibitor DEVD-cmk (Bachem, King of Prussia, PA) was added. The effect of inhibiting caspase-3-like activity on retinoid-induced growth arrest and apoptosis was determined by direct cell counting and/or apoptotic ELISA. All experiments were repeated in triplicate.
Oxygen Consumption Measurements of Cultured Cells-The rate of oxygen consumption was measured using a YSI model 5300 biological oxygen monitor that employs a Clark-type oxygen electrode (YSI Inc., Yellow Springs, OH). 6 ϫ 10 6 cells were trypsinized and suspended in 3 ml of complete DMEM culture medium in a 3-ml respiration chamber with continuous stirring. The respiration chamber was placed in a circulating water bath at 37°C and the oxygen electrode was inserted into the respiration chamber eliminating any air space between the oxygen electrode and the surface of the culture medium. The oxygen electrode was calibrated using complete DMEM without cells. The rate of oxygen consumption was measured over 10-min time periods and the respiration rates were normalized (nanomole of O 2 /min/10 6 cells) assuming 220 M O 2 concentration in air-saturated medium at 37°C (29). Retinoid and control treatments were added directly to the culture medium through a channel in the electrode holder, using a Hamilton syringe, without separating the electrode from the surface of the culture medium. Measurements were carried out under low light conditions. CD437 treatments were done at a concentration of 10 Ϫ6 M. Treatments greater than 1 h were performed as reported in Hail et al. (29) with the modification that the number of cells tested was 6 ϫ 10 6 cells. Cell numbers were adjusted using numbers determined by direct cell counts after respiration determinations. 35 S-PCR Analysis of Mitochondrial DNA-A PCR method reported by Wei et al. (32) was used to detect mitochondrial DNA sequences. This method was employed to determine the effects on mitochondrial number associated with CD437 exposure for long periods of time and to determine the relative proportion of mitochondrial DNA in the CA-CD437R cell line as compared with the CA-OV-3 cell line. To determine the linear range of the PCR reactions, the reactions were carried out for 37, 40, and 43 cycles, as recommended in Wei et al. (32), in a thermal cycler using primers specific for mitochondrial DNA (L1-5Ј-AACATAC-CCATGGCCAACCT-3Ј and H1-5Ј-GGCAGGAGTAATCAGAGGTG-3Ј). The L1-H1 primers were used for the amplification of a 533-bp fragment from the total mitochondrial DNA. The first cycle involved a 3-min denaturation at 94°C, 3-min annealing at 55°C, and a 1-min extension at 72°C. The other cycles were as follows: denaturation of 40 s at 94°C, annealing for 40 s at 55°C, and a 50-s extension at 72°C. Two l of mitochondrial DNA was amplified in a 50-l reaction mixture containing 200 M of each dNTP, 0.4 M of each primer, 1 unit of TaqDNA polymerase (Promega, Madison, WI), 50 mM KCl, 1.5 mM MgCl 2 , and 10 mM Tris-HCl, pH 8.3. We modified the original method by adding 1 l of 35 S-labeled dATP (10 mCi/ml) to the reaction mixture. The proper size and relative amounts of PCR product was determined by phosphorimaging (Cyclone Phosphor System and Optiquant Image Analysis software, Packard Instrument Co., Meridian, CT).
CMXRos Cell Staining-A method described by Poot et al. (30) was used to determine the permeability of mitochondria after treatment with 4-HPR and CD437. After a 12-h treatment with retinoids, cells were stained with MitoTracker Red CMXRos (8-(4Ј-chloromethyl)phenyl-2,3,5,6,11,12,14,15-octahydro-1H,4H,10H,13H-diquinolizino-8Hxanthylium chloride) from Molecular Probes (Eugene, OR). Culture medium was added to the cells and the fluorescent mitochondrial probe CMXRos was added at a final concentration of 0.25 g/ml for 15 min at room temperature. After two rinses for 2 min each with culture medium the cells were wet mounted with culture media and sealed with silicone. Images were visualized using a fluorescent microscope (Nikon Eclipse TE300, Japan) and camera (Nikon Digital Camera DXM1200, Japan). 35 S-RT-PCR Analysis of TR3 Expression-RT-PCR was used to de-CD437 Resistance tect TR3 expression using the Advantage RT-for-PCR kit (Clontech). In brief, the RT reaction was carried out with 1 g of total RNA, from cells treated for 8 h with CD437, in H 2 O to a total volume of 12.5 l into which was added 1 l of oligo(dT) 18 primer, 4 l of 5ϫ reaction buffer (50 mM Tris-HCl, pH 8.3, 75 mM KCl, and 3 mM MgCl 2 ), 1 l of dNTP mixture (10 mM each), 0.5 l of RNase inhibitor, and 1 l of Moloney murine leukemia virus-reverse transcriptase. The mixture was incubated for 1 h at 42°C, then heated for 5 min at 94°C to stop the reaction. The cDNA mixture was diluted to 100 l. The PCR reaction was carried out in a thermal cycler using primers specific for human TR3 (5Ј-TCATGGACGGCTACACAG-3Ј and 5Ј-GTAGGCATGGAAT-AGCTC-3Ј) (31). To determine the linear range of the reaction, samples are removed from the PCR reaction at 20, 25, and 30 cycles. These primers were used for the amplification of a 517-bp fragment of the TR3 cDNA. The cycles were as follows: denaturation of 45 s at 94°C, annealing for 45 s at 60°C, and a 2-min extension at 72°C. Five l of the cDNA mixture was amplified in a 50 l of reaction mixture containing 200 M of each dNTP, 0.4 M of each primer, 1 unit of TaqDNA polymerase (Promega), 50 mM KCl, 1.5 mM MgCl 2 , 10 mM Tris-HCl, pH 8.3, and 1 l of 35 S-labeled dATP (10 mCi/ml). Quantitation of the PCR product was determined by phosphorimaging (Cyclone Phosphor System and Optiquant Image Analysis software, Packard Instrument Co.).
Antibodies Used for Western Blotting-Differences in the level of a variety of mitochondrial proteins associated with apoptosis including Bid, Bad, Bax, Bcl-2, and Bcl-x (Transduction Laboratories), caspase-3, caspase-12, and cytochrome c (Santa Cruz Biotechnologies) were determined by Western blotting as we have described previously (33).
Antisense Caspase-12 Treatment-An antisense approach was used to determine the importance of caspase-12 function after treatment with CD437 using a modification of a method described by Nakagawa (34). The sequence of the antisense caspase-12 oligo was: 5Ј-thiol modification-TGTCCTCCTGGCCGCCATGGCTGT-3Ј. A scrambled caspase-12 oligo was used as a control: 5Ј-thiol modification-GTCGCT-CTGTACGCCTGTGCCTG-3Ј. In brief, 5 ϫ 10 5 cells were plated into 100-mm culture dishes in complete DMEM. Twenty-four hours later the cells were treated with antisense caspase-12 oligonucleotides or the control oligonucleotide representing a scrambled caspase-12 sequence. 1 h after the addition of oligonucleotides, retinoids were added to the culture. The cells were assessed for reduction of caspase-12 protein expression by Western blot, viability by direct cell count, and induction of apoptosis by ELISA.
MAP Kinase Inhibitors and in Vivo Kinase Assay-To determine the role of p38 MAP kinase activation in induction of apoptosis the cellpermeable p38 MAP kinase inhibitors SB 203580 and PD 169316 (Calbiochem, San Diego, CA) were added to culture medium 1 h prior to CD437 or 4-HPR treatments. A commercially available kit was used to measure MAP kinase activity (MAPK In Vivo Kinase Assay kit Clontech). Briefly, 4 ϫ 10 5 cells/well were cultured in 6-well plates. These cells were transfected with the p38 target protein ATF, a negative control or a positive control using the Effectene (Qiagen, Valencia, CA) transfection protocol. The cells were co-transfected with luciferase reporter and pCMV-␤-galactosidase (Clontech). Cells were treated with CD437, 4-HPR, or the p38 inhibitor PD 169316 (Calbiochem) 24 h after transfection. After 6 h cell lysates were collected and luciferase activity was measured using a luminometer (Femtomaster FB 12, Zylux Corp., Oak Ridge, TN). The results of the luminometer measurements were normalized by ␤-galactosidase levels.

CA-CD437R Cells do Not Exhibit Growth Arrest or Apoptosis after CD437 Treatment-
We have previously demonstrated that CD437 treatment induced apoptosis in the CA-OV-3 ovarian cancer cell line (6,16). The CA-CD437R cell line was generated by culturing CA-OV-3 cells in the presence of increasing concentrations of CD437 over a period of 6 months. CA-CD437R cells were treated with a concentration of CD437 of 10 Ϫ6 M because this concentration was previously found to be the most effective at inducing apoptosis in ovarian tumor cell lines (6,16). Fig. 1A shows that over a treatment period of 120 h, the CA-CD437R cell line did not undergo apoptosis as determined by apoptosis ELISA. For comparison, the parental cell line CA-OV-3 did undergo apoptosis after treatment with CD437. It should be noted that we have treated CA-CD437R cells for as long as 7 days with CD437 and have failed to observe apoptosis. Thus, these results clearly indicate that the cell line selected over the 6-month treatment period is resistant to CD437 and not merely less sensitive, requiring longer treatments.
Whereas CA-CD437R cells were resistant to CD437-induced apoptosis it is possible that these cells might still be sensitive to growth arrest by CD437. It has been reported by our laboratory and others that CD437 treatment is able to induce growth arrest in a number of tumor cells (6 -15). Parental CA-OV-3 and CA-CD437R cell lines were treated with 10 Ϫ6 M CD437 and viable cell counts were determined daily following treatment. Fig. 1B shows that CA-CD437R cells were not growth arrested by CD437 treatment. As expected, treatment with CD437-induced growth inhibition and reduction in cell number of parental CA-OV-3 cells. Moreover, resistance to growth inhibition did not change with increased time of exposure to CD437. These results are consistent with our previous findings that CA-CD437R cells have been selected to be resistant to the effects of CD437 treatment. Taken together, our results using the apoptotic ELISA and direct cell counts indicate that the CA-CD437R cell line is resistant to both the growth inhibitory and apoptotic effects of CD437.
Stability of the CA-CD437R Cell Line-The results described above indicate that the CA-CD437R cell line is resistant to CD437-mediated growth inhibition and apoptosis. We next wished to determine the stability of this cell line when the selective pressure of CD437 treatment was removed for extended periods of time. If this cell line remained resistant after removal of the selective pressure it is likely that resistance to CD437 results from a permanent change to the genome of the CA-OV-3 cells. CA-CD437R cells are normally cultured with CD437 selective pressure 5 of every 14 days. CD437 was removed from the CA-CD437R cell culture for extended periods of time, and then these cells were exposed to CD437 for 5 days and assayed for growth inhibition and induction of apoptosis. We found that CD437 treatment of the CA-CD437R cell line does not induce growth inhibition or apoptosis even when selective pressure is removed for as long as 84 days (data not shown). Thus the CA-CD437R cell line does not revert back to a CD437-sensitive phenotype when grown for extended periods of time in the absence of CD437.
Sensitivity of the CA-CD437R Cell Line to ATRA, 4-HPR, and TNF-␣-To determine whether the CA-CD437R cell line was also resistant to ATRA we determined the effect of ATRA treatment on cell growth and their ability to undergo RAR transcriptional activation. Fig. 2A shows that CA-CD437R cells remain sensitive to growth inhibition by 10 Ϫ6 M ATRA. To further determine whether RAR function was altered in these cells, the ovarian tumor cell lines were transfected with a RARE-␤-CAT reporter plasmid. Twenty-four h after transfection the cells were treated with 10 Ϫ6 M ATRA. No distinguishable difference in RAR-mediated transcriptional activation could be discerned between the parental CA-OV-3 and the CA-CD437R cell lines (Fig. 2B).
To further characterize the CA-CD437R cells we next determined the sensitivity of this cell line to other apoptotic inducing agents. The agents chosen for study, fenretinide (4-HPR) and TNF-␣, both induce apoptosis in the CA-OV-3 parental cell line. 4-HPR was used at a concentration of 10 Ϫ5 M and TNF-␣ treatments were carried out at a concentration of 50 ng/ml. The CA-CD437R cell line, which is resistant to CD437, was also resistant to the induction of apoptosis by TNF-␣ but was not resistant to the induction of apoptosis by 4-HPR as measured by an apoptotic ELISA (Fig. 3, A and B).
Caspase Activity in CA-CD437R Cells after Retinoid Treatment-To begin to understand the mechanism by which retinoids such as CD437 and 4-HPR induce apoptosis, we next examined the activation of a number of effector caspases including caspase-3/caspase-7 and caspase-9 activity. Fig. 4, A and B, shows that no caspase-3-like activity could be detected in the CA-CD437R cell line, in response to CD437. However, the level of caspase-3 activity assayed in CA-CD437R cells following 4-HPR treatment was similar to that induced by both CD437 and 4-HPR in the CD437-sensitive parental cell line CA-OV-3. Furthermore, treatment with DEVD-cmk (a caspase-3-like inhibitor) 1 h prior to retinoid treatment significantly inhibits 4-HPR-induced caspase-3-like activities in the CA-CD437R cell line. These results demonstrate that the caspase-3-like protease is not altered in CA-CD437R cells. We also investigated the activation of other caspases in these cell lines after treatment with CD437 and 4-HPR. In the 24 h after treatment no caspase-6 or caspase-8 activity could be detected in either CA-OV-3 or the CA-CD437R cells (data not shown). Similar to our results with caspase-3-like activity, caspase-9 activity was present in both cell lines after treatment with 4-HPR and could be detected in the CA-OV-3 cell line after CD437 treatment (Fig. 4, C and D). However, no caspase-9 activity could be detected following CD437 treatment of the CA-CD437R cell line. These results suggest that the alteration in CA-OV-3 cells, which is responsible for CD437 resistance, must map upstream of caspase-3-like and caspase-9 activation.
Caspase-12 has been reported to be important for induction of apoptosis following endoplasmic reticulum stress (34 -37). Because retinoids are lipophilic and might act via an endoplasmic reticulum-mediated process, we wished to determine whether there was a difference in levels or activation of caspase-12 in CA-CD437R cells versus CA-OV-3 cells following CD437 treatment. Interestingly, Western blot analysis showed that the CA-CD437R cell line expressed only 10% of the caspase-12 observed in the parental CA-OV-3 cells (Fig. 5A).
However, the levels of caspase-12 did not change following CD437 treatment (Fig. 5A). To determine whether this reduction in caspase-12 levels was responsible for the resistance of CA-CD437R cells to CD437 we determined the effect of reducing caspase-12 levels on the ability of CD437 to induce apoptosis in parental CA-OV-3 cells. Fig. 5B shows that treatment of CA-OV-3 cells with antisense to caspase-12 (50 g/ml) substantially reduced the level of caspase-12 within 6 h after treatment. Fig. 5C shows that pretreatment with concentrations of antisense caspase-12, which significantly reduced caspase-12 protein levels, did not prevent the induction of apoptosis as measured by an apoptotic ELISA (Fig. 5C), or the reduction of cell number (Data not shown) even when assayed 120 h after CD437 treatment (Fig. 5D). These results indicate that reduction of caspase-12 levels in CA-OV-3 cells does not in and of itself result in resistance to induction of apoptosis by CD437.
CMXRos Cell Staining-Apoptosis characterized by induction of caspase-9 activity followed by activation of caspase-3 has been reported to be associated with permeabilization of the mitochondrial membrane (38 -40). Because CD437 and 4-HPR both induced caspase-9 and caspase-3 in the CA-OV-3 cell line, we next examined the mitochondrial permeability of CA-OV-3 and the CA-CD437R cells using the probe CMXRos. Mitochondria accumulate the CMXRos probe unless the membranes of the mitochondria are depolarized. The micrographs in Fig. 6 clearly show that compared with vehicle-treated control cells (Fig. 6, A and B) the mitochondria of CA-OV-3 did not accumulate the CMXRos stain following treatment with either CD437 (Fig. 6, C and D) or 4-HPR (Fig. 6, E and F). This indicated that the induction of apoptosis by these retinoids involves depolarization of the mitochondrial membranes. In contrast, CD437 treatment of the CA-CD437R cell line did not induce the depolarization of the mitochondrial membranes (Fig. 7, C and D). As expected, 4-HPR was found to induce depolarization in both CA-CD437R cells and parental CA-OV-3 cells as indicated by the fact that the mitochondria were no longer able to accumulate CMXRos (Figs. 6, E and F, and 7, E and F). It should be noted that the use of a second similar stain, DIOC (6) gave the same results (data not shown). Thus, CD437 fails to induce mitochondrial membrane depolarization in the CA-CD437R cells indicating that the alteration in the CA-CD437R cells maps upstream of mitochondrial permeabilization. 35 S-RT-PCR Analysis of TR3 Expression-Increased expression of TR3 has been reported to occur in cells induced to undergoing apoptosis by a number of agents including CD437 (8,31,(41)(42)(43)(44)(45). Because the CMXRos staining indicated a depolarization of the mitochondrial membrane in conjunction with reports that CD437-induced apoptosis can be associated with the increased expression of TR3 and the eventual translocation of TR3 to the mitochondria, we next examined TR3 expression in our cells (8,31,(43)(44)(45). We used a semiquantitative RT-PCR to measure expression of TR3 after 8 h of CD437 treatment. This method was employed to determine whether CD437 treatment induced expression of TR3 in CA-OV-3 cells and to determine whether differential expression of TR3 was responsible for the resistance of the CA-CD437R cell line to the effects of CD437. As indicated in Fig. 8, both the parental CA-OV-3 cell line and the CA-CD437R cell line express TR3 and exhibited an increase in the expression of TR3 following CD437 treatment as measured by 35 S-RT-PCR. This increase was similar for both cell lines. The data were normalized to glyceraldehyde-3-phosphate dehydrogenase levels as determined by 35 D), or 4-HPR (E and F) and then stained with CMXRos to determine mitochondrial membrane potential. The lack of retention of CMXRos (red) occurs in mitochondria, which loses membrane potential and is indicative of apoptosis induction. For comparison purposes cells are shown in bright field in panels A, C, and E. Images were visualized at ϫ1000 using a fluorescent microscope and digital photo camera.
indicate that expression of TR3 is not impaired in the CA-CD437R cell line following CD437 treatment.
Oxygen Consumption Measurements of Ovarian Carcinoma Cells-It has recently been reported that CD437 could decrease cellular oxygen consumption in human cutaneous squamous cells and that reduction of mitochondria numbers by ethidium bromide treatment resulted in resistance to CD437 (29). We compared the oxygen consumption rate in the parental ovarian carcinoma cell line, CA-OV-3, and the CA-CD437R cell line. We found that the resting rate of oxygen consumption in the CA-CD437R cells was 65-70% lower than the parental CA-OV-3 cells (Fig. 9A). Treatment of the parental CA-OV-3 cell line with CD437 resulted in a decrease in the rate of cellular oxygen consumption by ϳ75% (Fig. 9A). This reduction in the rate of oxygen consumption continued to decrease for 6 h. In contrast, CD437 treatment of CA-CD437R cells does not further reduce oxygen consumption. To determine whether the lower resting rate of oxygen consumption of the CA-CD437R cell line compared with the parental CA-OV-3 cells was because of a reduction in the number of mitochondria in CA-CD437R cells, a 35 S-PCR method was used to detect mitochondrial DNA sequences. Fig. 9B shows that the relative amount of mitochondrial DNA and thus the relative number of mitochondria was similar in both cell lines indicating that the reduced oxygen consumption by the CA-CD437R cell line was not because of a reduction in the number of mitochondria. This suggests that the alteration in CA-CD437R cells resulting in resistance to CD437-induced apoptosis maps at the level of the mitochondria and most likely involves events that regulate oxygen consumption.
Western Blotting for Proteins Associated with the Induction of Apoptosis-Because the resistance of the CA-CD437R cell line to CD437 appeared to involve mitochondria function but was not because of a decrease in the number of mitochondria, we examined the expression of a number of mitochondrial proteins associated with apoptotic induction. Western blot analysis indicated that no differences in the levels of expression of Bad, Bid, Bax, and Bcl-x mitochondrial proteins occur between the CA-OV-3 and CA-CD437R cell lines (data not shown). Whereas levels of the mitochondrial protein Bcl-2 were slightly greater in the parental cell line and were observed to increase only after 4-HPR treatment, Bcl-2 levels did not change after CD437 or 4-HPR treatment of the CA-CD437R cell line (Fig. 10A). Likewise cytochrome c was released from the mitochondria to the cytoplasm after treatment of the CA-OV-3 cell line with CD437 ( Fig. 10B) but not after treatment of the CA-CD437R cell line (data not shown). 4-HPR treatment of both the parental CA-OV-3 and the CA-CD437R cell lines induced cytochrome c release at similar levels (data not shown).
MAP Kinase (p38)-It has been reported that activation of p38 MAP kinase, following treatments that induce apoptosis, can lead to the activation of caspase-3 and the depolarization of the mitochondrial membrane (39, 46 -48). Fig. 11A shows that 6 h of CD437 treatment can induce the activity of p38 in both the CA-OV-3 and CA-CD437R cell lines. In fact p38 activity was induced to a greater extent in CA-CD437R cells following retinoid treatment. We used p38 MAP kinase inhibitors in conjunction with retinoid treatment to determine whether the inhibition of p38 activity would prevent retinoid-induced apoptosis (Fig. 11B). The apoptosis ELISA was used to measure the extent of apoptosis after 5 days treatment with the synthetic retinoids and MAP kinase inhibitors. We found that both p38 inhibitors, PD169316 and SB202190, could inhibit the induction of apoptosis by CD437 in the CA-OV-3 cell line. In contrast, the MAP kinase inhibitors did not inhibit 4-HPRinduced apoptosis. These results confirm that CD437 and 4-HPR induce apoptosis by completely different pathways. Moreover because CD437 can induce p38 activity in CA-CD437R cells, the alteration in CA-CD437R cells responsible for resistance to apoptosis must map downstream of the activation of p38 by CD437. DISCUSSION Using a cell line selectively resistant to the growth suppressive and apoptotic effects of CD437 treatment, we have determined the following new information about the molecular events responsible for induction of apoptosis by CD437: 1) the pathway utilized by CD437 to induce apoptosis appears to be similar to that induced by TNF-␣ but not that induced by another apoptotic inducing retinoid, 4-HPR. 2) CD437-induced apoptosis in ovarian carcinoma cells is dependant upon caspase-3 and caspase-9 but not caspases-6, caspase-8, or caspase-12. 3) CD437-induced apoptosis in ovarian carcinoma cells is dependent and involves a change in mitochondrial oxygen consumption, depolarization of the mitochondria, and subsequent release of cytochrome c, resistance of CA-CD437R cells to CD437 does not involve a reduction in ATRA responsiveness or RAR activity, mitochondrial number, or ability to activate p38 kinase.
To characterize the mechanism involved in induction of apoptosis by CD437 we explored the response of a CD437-resistant cell line (CA-CD437R) to other agents that induce apoptosis. The parental CA-OV-3 ovarian tumor cell line is sensitive to apoptosis induced by treatment with 4-HPR and TNF-␣. Surprisingly, the CA-CD437R cell line was not responsive to the effects of TNF-␣ treatment but was just as sensitive as the FIG. 9. Oxygen consumption measurements of cultured cells after treatment with CD437 and 4-HPR. A, the rate of oxygen consumption was measured using a Clark-type oxygen electrode. 6 ϫ 10 6 cells were trypsinized and suspended in 3 ml of DMEM culture medium in a 3-ml chamber with continuous stirring at 37°C. The oxygen electrode was calibrated using complete DMEM without cells. The rate of oxygen consumption was measured over 10-min time periods at 1, 3, 6, and 12 h and the respiration rates were normalized (nanomole of O 2 /min/10 6 cells) assuming 220 M O 2 concentration in air-saturated. CD437 (10 Ϫ6 M) and control treatments were added directly to the culture medium. Cell numbers were adjusted using numbers determined by direct cell counts after respiration determinations. CA-OV-3 was treated with Me 2 SO (q) and 10 Ϫ6 M CD437 (E), and the CA-CD437R cell line was treated with Me 2 SO () and 10 Ϫ6 M CD437 (ƒ). The data represent the mean Ϯ S.D. of three separate experiments performed in triplicate. B, a 35 S-PCR method was employed to determine the relative proportion of mitochondrial DNA in the CA-CD437R cell line as compared with the CA-OV-3 cell line. The PCR reaction was carried out as indicated under "Materials and Methods" and the image of the PCR products at three different cycles and the relative band density was determined by phosphorimaging. CA-OV-3 cell line to 4-HPR treatments. This suggests a common mechanism for the induction of apoptosis by TNF-␣ and CD437 treatments. Furthermore, the ability of 4-HPR to induce apoptosis in CA-CD437R cells indicates a clear divergence in the pathway of apoptosis induction by these two synthetic retinoids.
Because retinoids are known to function by activation of RARs and induction of apoptosis by CD437 is at least partially RAR dependent in CA-OV-3 cells (16), it was important to determine whether RAR function had been altered in the CA-CD437R cell line. Two approaches were used to determine whether the RAR response was altered: sensitivity to ATRA growth inhibition and transcriptional transactivation of a reporter plasmid construct that contained the ATRA responsive RAR-␤-RARE. We found that CA-CD437R cells responded to ATRA-induced growth inhibition and RARE-dependent transcriptional activation in a manner identical to that of the parental CA-OV-3 cells. These experiments show that RAR function and the sensitivity to ATRA-induced growth inhibition has not been altered in the CA-CD437R cell line. This suggests that the event responsible for CA-CD437R resistance to CD437 maps downstream of RAR activation and does not involve elements common to both the ATRA and CD437 pathways. This is consistent with the work of Ponzanelli et al. (49) who showed that a promyelocytic leukemia cell line resistant to CD437 retained sensitivity to ATRA and the RAR-␣ selective synthetic retinoid AM580.
Recent studies in other cell models to determine the molecular events that constitute the apoptotic pathway induced by CD437 have suggested that the mechanism of CD437 action is extremely complex and at times contradictory. For example, CD437 was shown to induce growth arrest and apoptosis in tumor cell lines via a caspase-dependent pathway in some tumor cells and via caspase-independent pathways in others (50,51). CD437-induced apoptosis is associated with the activation and degradation of many caspase isoforms including caspase-3, -6, -7, and -9 (38,49,52). In addition, treatment with CD437 has been shown to modulate the expression of apoptoticrelated proteins such as Bax, Bcl-2, DR5, p53, and p21 (WAF1/ CIP1) in lung, breast, and small cell carcinoma cells (8,13,38,51,53).
We have found that in ovarian carcinoma cells the induction of apoptosis by both CD437 and 4-HPR is caspase-dependent. Addition of the caspase-3-like inhibitor, DEVD-cmk, 1 h prior to the addition of CD437 or 4-HPR blocked apoptosis induced by these retinoids. We could not detect caspase-3-like activity in the CA-CD437R cells after treatment with CD437 or TNF-␣.
To determine whether lack of caspase-3 or -7 was responsible for CD437 resistance in CA-CD437R cells, we used 4-HPR to determine whether the caspase-3-like proteases were present and functional in this cell line. Measurements of the activity of the caspase proteases using the caspase-3-like substrate DEVD-pNA mixed with 4-HPR-treated cell lysates indicates that 4-HPR induces a caspase-3-like protease activity. Moreover, treatment of CA-CD437R cells with DEVD-cmk protected these cells from 4-HPR-induced apoptosis. Together these two findings show that caspase-3-like proteases are functional and unaltered in the CA-CD437R cell line and are thus not responsible for the resistance of these cells to CD437-induced apoptosis.
We also looked at other caspases that are known to function upstream of caspase-3 activation. Assays for caspase-8 and -6 after treatment showed that neither CD437 nor 4-HPR activated these caspases. Caspase-12 has been reported to be activated upstream of caspase-3-like activity during endoplasmic reticulum stress (34 -37). Moreover CA-CD437R cells express less caspase-12 protein than parental CA-OV-3 cells. However, reduction of caspase-12 levels in parental CA-OV-3 cells did not protect them from CD437 or 4-HPR-induced apoptosis.
Reports in other cell models systems have indicated that CD437 and 4-HPR treatments cause an apoptosis-related depolarization of the mitochondrial membrane (38, 49, 54 -56). Consistent with these reports, we show that depolarization also occurs in both CA-OV-3 and CA-CD437R cells after treatment with 4-HPR but that CD437 induced depolarization of the mitochondria only in the parental CA-OV-3 cells and not in the CA-CD437R cells. Depolarization of the mitochondrial mem- FIG. 11. CD437-induced apoptosis requires MAP kinase (p38) activity. A, MAP kinase (p38) activity was detected after CD437 treatment in the CA-OV-3 cell line but not the CA-CD437R cell line. CA-OV-3 and CA-CD437R cells (4 ϫ 10 5 cells/well) were cultured in 6-well plates. The next day cells were transfected with a luciferase reporter to measure p38 activity and a pCMV-␤-galactosidase to determine transfection efficiency. Twenty-four h after transfection cells were treated with CD437 or 4-HPR and/or the p38 inhibitor PD169316. After 12 h of treatment cell lysates were collected and luciferase activity was measured using a lumenometer. UVB-treated cells were used as a positive control. ␤-Galactosidase levels normalized the luminometer results. The data represent the mean Ϯ S.D. of three separate experiments performed in triplicate. B, MAP kinase (p38) inhibitors modulate CD437-but not 4-HPR-induced apoptosis in CA-OV-3 and CA-CD437R cells. CA-OV-3 and CA-CD437R ovarian tumor cells (5 ϫ 10 5 ) were plated in 100-mm culture dishes with 10 ml of DMEM with 10% FBS, and treated with either 10 Ϫ6 M CD437 or 10 Ϫ5 M 4-HPR and MAP kinase inhibitor, 20 g/ml PD169316 or 10 g/ml SB202190. The apoptosis ELISA, described under "Materials and Methods," was performed on 5-day treated cells to determine the extent of apoptosis induction. Me 2 SO and the retinoids alone were used as controls. The data represent the mean Ϯ S.D. of three separate experiments performed in triplicate.
brane results in the release of caspase-9, cytochrome c, and APAF-1, which form a complex capable of activating caspase-3, -6, and -7 (57). It has been suggested that cytochrome c may be central to CD437-induced caspase activation (38,49,52,53,55). Consistent with these reports, we determined that cytochrome c is released after treatment of CA-OV-3 cells with CD437 but not after treatment of CA-CD437R cells.
Caspase-9 has been shown to be released after mitochondrial depolarization in conjunction with cytochrome c and APAF-1, prior to activation of caspase-3. Parental CA-OV-3 ovarian carcinoma cells treated with a caspase-9 inhibitor (LEHD-cho) 1 h prior to the addition of CD437 or 4-HPR blocked apoptosis induced by these retinoids. Measurements of the activity of the caspase-9 proteases using the substrate LEHD-pNA mixed with lysates from either CD437-or 4-HPR-treated CA-OV-3 cells but not CA-CD437R cells indicates that both retinoids induce caspase-9 protease activity. Treatment of CA-CD437R cells with LEHD-cmk protected these cells from 4-HPR-induced apoptosis, caspase-9 does not appear to be altered in the CA-CD437R cell line and therefore is not responsible for resistance to CD437.
Our studies show that the CA-CD437R cell line is resistant to apoptosis induced by TNF-␣. This suggests that CD437induced apoptosis utilizes a similar mechanism as TNF-␣. Cross-resistance of ovarian tumor cell lines to multiple apoptotic agents has been reported previously. Kumar et al. (58) found that a cell line selected for resistance to paclitaxel was also resistant to CD437. The fact that CA-CD437R cells contain functional caspase-3-like and caspase-9 protease activation and activity suggests that the alteration responsible for CD437 resistance lies upstream of caspase-3-like and caspase-9 activation. A recent report suggests that resistance to CD437 of a leukemia cell line mapped upstream of cytochrome c release and activation of caspases (49). The release of cytochrome c is dependent on release from the mitochondria often through a depolarization of the mitochondrial membrane (55,59). CMXRos staining demonstrated that mitochondria were depolarized in CA-OV-3 cells but not CA-CD437R cells after CD437 treatment. Likewise, cytochrome c was found to be released in CD437-treated CA-OV-3 cells but not in CA-CD437R cells. However, because 4-HPR treatment of CA-CD437R cells did result in mitochondrial depolarization and release of cytochrome c, this step in the apoptotic pathway is not defective and thus not responsible for CD437 resistance.
TR3 has been associated with the depolarization of mitochondria through increased expression, translocation to the cytosol, and targeting of the mitochondria (8, 31, 43-45, 60 -62). In light of these findings it was logical to determine whether a lack of TR3 expression could be the cause of resistance to the effects of CD437 in the CA-CD437R cell line. Our data indicate that both cell lines express TR3 and that there was a comparable response in the increase of TR3 expression occurring in both the parental CA-OV-3 and the CA-CD437R cell lines after CD437 treatment. Whereas we did not examine translocation of TR3, it would appear, at least based on our expression analysis, that a TR3 defect is not likely to be responsible for CA-CD437R cell line resistance to CD437.
Hail et al. (29) reported that resistance to CD437 treatment occurs in cutaneous squamous carcinoma cells if mitochondria numbers are reduced. Our studies do not show that the CA-CD437R cell line has a reduced number of mitochondria compared with parental CA-OV-3 cells but do exhibit significantly lower oxygen consumption after CD437 treatment. This suggests that the resistance seen in the CA-CD437R cell line might involve a defect in the electron transport chain.
TNF-␣ has been reported to activate tyrosine kinases of cellular stress-induced pathways consequently leading to activation of c-Jun NH 2 -terminal kinase (JNK) and p38 MAPK (63)(64)(65). Recent reports suggest that activated p38 MAPK is associated with apoptosis and regulates the release of cytochrome c from the mitochondria (39,46,48). Because CD437 has been reported to activate JNK, this suggests that the JNK pathway may be a common link for apoptosis induced by TNF-␣ and CD437. Our in vivo p38 MAP kinase activity assays show that activity occurred in both the CA-OV-3 and CA-CD437R cell lines after CD437 treatment. These data indicate that the resistance to CD437 in the CA-CD437R cell line maps downstream of MAP kinase activation.
A model for the pathway by which CD437 induces apoptosis in CA-OV-3 cells is shown in Fig. 12. We propose that CD437 induces apoptosis by a similar mechanism as has been reported for TNF-␣. As indicated in our model (Fig. 12), CD437 is able to induce the activation of the p38 MAPK after 6 h treatment. As reported in the literature, p38 MAPK can activate the MEF2 transcription factor (66 -69). Because it has been reported that a MEF2 site is located on the TR3 gene promoter (66), MEF2 could be responsible for the increase in TR3 expression that we Our results show that the CA-CD437R cell line has increased levels of TR3 after CD437 treatment but no mitochondrial depolarization. This indicates that the alteration in the pathway exhibited by the CA-CD437R cell line maps downstream of TR3 mRNA induction and upstream of mitochondrial depolarization. Our results also show that, in contrast to CD437, no p38 MAPK activity was induced after 4-HPR treatments. In addition MAP kinase inhibitors failed to protect CA-OV-3 cells from 4-HPR-induced apoptosis. Thus, it would appear that the difference between 4-HPR-induced apoptosis and CD437-induced apoptosis occurs upstream of mitochondrial depolarization and activation of p38 MAP kinase. observed 8 h after treatment with CD437. The translocation of TR3 has been reported to occur after CD437 treatment. TR3 translocation can induce the depolarization of mitochondrial membranes (8, 31, 43-45, 60 -62). We demonstrate that CD437 treatment results in mitochondrial depolarization. This occurs after activation of p38 and is followed by the release of cytochrome c (55,59). Depolarization of the mitochondrial membrane is associated with the release of cytochrome c, APAF-1, and procaspase-9. These proteins when released from the mitochondria will form a complex that results in the activation of caspase-9. Active caspase-9 will cleave procaspase-3 and result in the activation of caspase-3. Caspase-3 activity is indicative of the final stages of apoptosis induction. This is consistent with our findings that both caspase-3 and caspase-9 activity is induced by CD437 and that apoptosis occurs after this activation. Future studies will focus on identification of molecular events, which follow activation of p38 MAPK and precede mitochondrial depolarization.
Interestingly, neither cell line had p38 activity after 4-HPR treatment. This finding indicates that in ovarian carcinoma cells the divergence in the pathways induced by these two retinoids occurs early in the pathway prior to induction of p38 MAP kinase activity. Our findings that suggest divergence of the pathways that CD437 and 4-HPR utilize to induce apoptosis is consistent with the findings by Appierto et al. (70) who developed an ovarian carcinoma cell line that was resistant to 4-HPR but not resistant to ATRA or CD437. Moreover, we believe that demonstration of the fact that CD437 and 4-HPR induce apoptosis via different pathways will have potential therapeutic implications in that ovarian tumors resistant to one retinoid may not be resistant to the other.