Characterization of calcium release-activated apoptosis of LNCaP prostate cancer cells.

Apoptosis inhibition rather than enhanced cellular proliferation occurs in prostate cancer (CaP), the most commonly diagnosed malignancy in American men. Therefore, it is important to characterize residual apoptotic pathways in CaP cells. When intracellular Ca(2+) stores are released and plasma membrane "store-operated" Ca(2+) entry channels subsequently open, cytosolic [Ca(2+)] increases and is thought to induce apoptosis. However, cells incapable of releasing Ca(2+) stores are resistant to apoptotic stimuli, indicating that Ca(2+) store release is also important. We investigated whether release of intracellular Ca(2+) stores is sufficient to induce apoptosis of the CaP cell line LNCaP. We developed a method to release stored Ca(2+) without elevating cytosolic [Ca(2+)]; this stimulus induced LNCaP cell apoptosis. We compared the apoptotic pathways activated by intracellular Ca(2+) store release with the dual insults of store release and cytosolic [Ca(2+)] elevation. Earlier processing of caspases-3 and -7 occurred when intracellular store release was the sole Ca(2+) perturbation. Apoptosis was attenuated in both conditions in stable transfected cells expressing antiapoptotic proteins Bclx(L) and catalytically inactive caspase-9, and in both scenarios inactive caspase-9 became complexed with caspase-7. Thus, intracellular Ca(2+) store release initiates an apoptotic pathway similar to that elicited by the dual stimuli of cytosolic [Ca(2+)] elevation and intracellular store release.

the most commonly diagnosed malignancy in American men and is the second most common cause of death due to cancer, with estimates that 37,000 men died from CaP in 1999 (5).
Successful CaP therapies have been shown to induce tumor regression via apoptosis; however, therapy-resistant cells fail to die in response to treatment (6). It is therefore important to characterize the residual functional apoptotic pathways in therapy-resistant CaP cells. Perturbation of calcium ion (Ca 2ϩ ) homeostasis is a common occurrence in CaP cells induced to undergo apoptosis by a variety of agents, including chemotherapeutics (7) and androgen ablation (8,9).
Although the influx of Ca 2ϩ across the cell membrane results in a dramatic increase in cytosolic [Ca 2ϩ ], this influx is often invoked by a less readily detectable cytosolic [Ca 2ϩ ] elevation as a consequence of organellar Ca 2ϩ store release (10). For example, the endoplasmic reticulum (ER) serves as an intracellular Ca 2ϩ store that is released into the cytosol in response to a variety of stimuli, including binding of inositol 1,4,5trisphosphate (IP 3 ) to its cognate ER-localized receptor (IP 3 R) (11). The release of ER Ca 2ϩ stores in turn induces a Ca 2ϩ influx across the plasma membrane, also known as the Ca 2ϩ release-activated Ca 2ϩ current (I CRAC ) (12). Thus, the release of ER Ca 2ϩ stores serves to trigger a larger perturbation in cytosolic [Ca 2ϩ ].
Studies with an IP 3 R-deficient Jurkat T-lymphocyte cell line underscore the importance of ER Ca 2ϩ stores in cellular signaling. In contrast to the parental cell line, these cells are defective in antigen-specific T-cell signaling and are resistant to multiple inducers of apoptosis, including Fas ligand, CD3 antibody, glucocorticoids, and irradiation (13). These results are of interest because Jurkat cells have been used to study the role of caspases in apoptosis. Caspases are aspartate-specific cysteine proteases that constitute the effector arm of cell death (14). Upon receipt of an apoptotic stimulus, large pro-domain initiator caspases (such as caspase-8 and -9) are activated by autoproteolysis, which converts the single polypeptide zymogen to form the active dimeric protease. Caspase-8 is activated when it is recruited by the adapter molecule FADD (Fas/APO-1-associated death domain protein) to death receptors upon extracellular ligand binding (15,16). Caspase-9 is activated when cells are subjected to intracellular insults that result in the release of cytochrome c from damaged mitochondria. In the presence of cytochrome c the ATPase adapter molecule Apaf-1 binds to and activates caspase-9 in a reaction requiring ATP or dATP hydrolysis (17,18). Initiator caspases amplify the death signal by proteolytically processing and activating downstream small pro-domain effector caspases (such as caspase-3 and -7). Activated effector caspases subsequently cleave a variety of apoptotic substrates, resulting in the biochemical and morphological changes that characterize apoptosis (19).
Because IP 3 R-deficient Jurkat cells are incapable of mobilizing Ca 2ϩ stores and are resistant to apoptosis, the question has arisen as to whether release of ER Ca 2ϩ is by itself sufficient to induce apoptosis. To address this we chose to study thapsigargin (TG)-induced [Ca 2ϩ ] perturbations in the human CaP cell line LNCaP, an androgen-sensitive CaP cell line derived from a lymph node metastasis (20). TG is the most potent and specific irreversible inhibitor of ATP-dependent Ca 2ϩ pumps localized in the ER (sarcoplasmic/endoplasmic reticulum calcium ATPase (SERCA) pumps) (21). As a consequence of SERCA pump inhibition, ER Ca 2ϩ stores are rapidly released into the cytosol via ER Ca 2ϩ leak channels, thereby activating an I CRAC -mediated sustained elevation of cytosolic [Ca 2ϩ ] (22). To study the effects of ER Ca 2ϩ store release independent of I CRAC -mediated cytosolic [Ca 2ϩ ] elevation, we developed a technique to hold cytosolic [Ca 2ϩ ] constant after TG-mediated ER store release. This release of ER [Ca 2ϩ ] was found to be sufficient to induce apoptosis of LNCaP cells. To determine whether the apoptotic pathway elicited by ER Ca 2ϩ store release differed from that elicited by the dual perturbations of cytosolic [Ca 2ϩ ] increase and ER Ca 2ϩ store release, we compared the caspase activation profiles evoked by each stimulus as well as the effects of antiapoptotic proteins.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagent Treatment Conditions-LNCaP human prostatic carcinoma cells were grown in RPMI 1640 cell culture medium supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin, and 1% L-glutamine. For dose-response studies, 1 ϫ 10 5 cells were seeded in duplicate 6-well plates. Indicated concentrations of the drugs were achieved by adding 1000ϫ stock solutions dissolved in Me 2 SO. Untreated cell culture medium and medium with 0.1% Me 2 SO were used as controls. Cells were trypsinized and counted at the appropriate time point using trypan blue dye exclusion to mark viable cells. Jurkat cells were cultured in RPMI 1640 cell culture medium supplemented with 10% heat-inactivated fetal bovine serum, 1% penicillin-streptomycin, and 1% L-glutamine.
DNA Extraction and Analysis-Internucleosomal DNA fragmentation as a consequence of TG treatment was evaluated using a modification of the original method (23). LNCaP cells treated with 10 nM PMA served as the positive control for inducing internucleosomal DNA fragmentation (24). LNCaP cells cultured in 0.1% Me 2 SO and untreated cells served as the negative control. In brief, 2 ϫ 10 6 cells were seeded in duplicate 100-mm plates, treated as indicated, and collected at the appropriate time point. Cells were lysed (10 volumes of 25 mM sodium acetate buffer, pH 6.6) and centrifuged (12,400 ϫ g, 45 min) to remove insoluble DNA. The supernatant containing soluble DNA was precipitated (1 volume of 0.13 M NaCl in 70% ethanol) and recovered by vacuum centrifugation. The DNA was then dissolved (50 L of 10 mM Tris buffer with 1 mM EDTA and 1% SDS, pH 7.4), extracted (4 ϫ 50 L, 24:1 chloroform/isoamyl alcohol) and resolved by electrophoresis on an agarose gel (1.8% for 4 h at 75 V). Fragment size was calibrated using a 1 Kb ladder (Life Technologies, Inc.).
Electron Microscopy-Cells were seeded onto 12-well plates, treated as indicated, and fixed with Karnovsky's fixative followed by postfixation in osmium tetroxide in 0.1 M Sorensen's buffer. After fixation, cells were dehydrated in a graded series of acetone and embedded in epoxy resin. Sections were cut en fasse and stained with both uranyl acetate and lead citrate. Electron micrographs were photographed using a Phillips EM-400 electron microscope.
Flow Cytometric Analysis of Hypodiploid DNA-60-mm dishes were plated with 1 ϫ 10 6 cells, treated as indicated, and harvested at the appropriate time point. Floating cells were combined with trypsinized adherent cells and were fixed and analyzed in the Genentech Cytometry Laboratory following standard techniques (25). For analysis of LNCaP cells transfected with antiapoptotic proteins, multiple time points for each condition were assessed and results reported are for those time points that gave the most significant results. ] analysis of using Indo-1, cells were plated to 60% confluence on 4-well chamber slides (Nunc, Napierville, IL). The cells were prepared and analyzed in the Genentech Cytometry Laboratory following established protocols using a heated slide chamber attachment (26). Measurements with Fura-2 were achieved following a modified protocol of Grynkiewicz et al. (27). In brief, 1 ϫ 10 6 cells were plated onto 9 ϫ 22-mm poly-L-lysinecoated glass coverslips in 35-mm plates and washed with 2 ml of HBSS (Life Technologies, Inc.). HBSS was warmed to 37°C for preparation of all solutions and washes. A Fura-2/acetoxymethyl ester (AM) stock (Molecular Probes, Eugene, OR) was prepared in 20% Pluronic F-127 (Molecular Probes) in anhydrous Me 2 SO and was dissolved in HBSS for a final concentration of 3 M Fura-2/AM. Cells were loaded (30 min, 37°C) in 2.75 ml of this solution and washed (2 ϫ 2 ml of HBSS). Fluorescence ratios (340/380-nm excitation) were measured using a spectrophotometer supplied with a xenon lamp (model F-2000, Hitachi, San Jose, CA) with a 510-nm emission recording in an unheated chamber with a stir bar. Calcium concentrations were determined by using a calcium calibration buffer kit (no. 2, Molecular Probes) following the manufacturer's suggested protocol. When cells were loaded with BAPTA/AM, 1000ϫ to 5000ϫ BAPTA/AM stock solutions were prepared in anhydrous Me 2 SO and dissolved in HBSS, and the cells were incubated (30 min at 37°C) and washed in 2 ml of HBSS prior to Fura-2/AM or Indo-1/AM loading. All [Ca 2ϩ ] measurements were performed in HBSS having [Ca 2ϩ ] matched to the cell culture medium [Ca 2ϩ ], because [Ca 2ϩ ] measurements in the culture medium, even without phenol red, produced lower signals (data not shown).
Generation of Stable Cell Lines-Stable LNCaP cell lines were generated using CaPO 4 precipitation-mediated transfection. Following transfection, cells were cultured in supplemented RPMI 1640 culture medium with 360 g/ml Geneticin (Life Technologies, Inc.).
Immunoprecipitation and Western Blot Analysis-For caspase analysis, monoclonal antibodies against the active subunits of caspase-3, -7, and -9 were generated at the Genentech (GNE) Hybridoma Facility. For positive controls, four flasks, each containing 30 ml of 90% confluent Jurkat cells, were treated for 2 h with 100 ng/ml anti-Fas antibody (Upstate Biotechnology, Lake Placid, NY), and four equivalent flasks of untreated Jurkat cells were used as negative controls. Flasks of treated or untreated Jurkat cells were pooled and lysed in 3.2 ml of lysis buffer (50 mM HEPES, pH 7.2, 0.1% Nonidet P-40, 120 mM NaCl) containing protease inhibitor tablets (Roche Molecular Biochemicals) on ice for 30 min followed by three freeze-thaw cycles. For each experimental time point, five 15-cm plates of 70 -80% confluent wild-type LNCaP cells were treated as indicated, harvested at the appropriate time point, pooled, and lysed as described for Jurkat cells.
For Western blot analysis of caspase-3, 60 l of each lysate was centrifuged (15,300 rpm, 15 min, 4°C) and 5ϫ sample buffer containing 0.1% ␤-mercaptoethanol was added to the supernatant. After boiling, 35 l of each sample was resolved on 4 -20% acrylamide gels, transferred to nitrocellulose, and incubated in a mixture of anti-caspase-3 antibodies for immunoblotting (5 g/ml GNE clone 2358 and 0.3 g/ml clone E-8, Santa Cruz Biotechnology, Inc.).
Caspase-7 and -9 were detected by a sequential immunoprecipitation/immunoblotting protocol. For immunoprecipitation 785 l of cell lysate was centrifuged (as above), the supernatant was divided into two tubes, lysis buffer was added to a final volume of 1.0 ml per tube, and 10 g/ml appropriate antibody (caspase-7 GNE clone 2347 or caspase-9 GNE clone 2352) was added. Samples were rotated for a minimum of 2 h at 4°C, 15 l of protein AϩG beads (Calbiochem) was added, and samples were rotated for an additional 2 h. The beads were recovered by centrifugation, washed three times in lysis buffer, boiled in sample buffer to elute bound proteins, and the divided samples were recombined and resolved on 4 -20% (caspase 7) or 10 -20% (caspase-9) acrylamide gels. Immunoprecipitated caspase-7 was detected by immunoblotting using 1 g/ml anti-caspase-7 antibody (GNE clone 2347) and caspase-9 using a monoclonal antibody mixture (gift from Yuri Lazebnik, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY plus 4 g/ml PharMingen clone B40).
For expression analysis of transfected cell lines, standard Western blot protocols were followed as above using 40 g of cellular lysates and a monoclonal antibody to Bclx L (2 g/ml, Pharmingen), the anticaspase-9 antibody mixture described above, or a polyclonal antibody to neomycin phosphotransferase II for control pcDNA3-transfected cells (15 g/ml, 5 Prime 3 3 Prime, Inc., Boulder, CO). Positive controls included Bclx L ⌬TM or caspase-9 recombinant protein and lysate from a caspase-9 DN stable transfected LNCaP cell for neomycin phosphotransferase II analysis. Wild-type LNCaP cell lysate served as the negative control in all cases.

RESULTS AND DISCUSSION
TG Induces LNCaP Cell Apoptosis-To determine the optimal TG concentration for inducing LNCaP cell death, lethality curves as functions of both dose and time were established (Fig.  1A). Most cell death occurred between 24 and 48 h using 100 nM TG. Based on these results and on phase-contrast microscopic evaluation of cellular morphology, treatment with 100 nM TG was chosen as the standard concentration for inducing LNCaP cell death. This is consistent with the concentration needed for complete SERCA pump blockade (29). We used the structurally unrelated SERCA pump inhibitor cyclopiazonic acid (CPA) to confirm that SERCA pump blockade was indeed the mechanism responsible for TG-induced LNCaP cell death. At 100 M, a concentration also consistent with complete SERCA pump blockade (29), CPA induced LNCaP cell death following a similar time course as TG-induced lethality (data not shown). Because the only similarity shared between TG and CPA is the ability to inhibit SERCA pumps (29), we concluded that the [Ca 2ϩ ] perturbation induced by SERCA pump inhibition was likely responsible for inducing apoptosis in LNCaP cells.
Because cell death can occur via apoptosis or necrosis, it was important to determine which mechanism was deployed in TG-treated LNCaP cells. Apoptosis is characterized by a num- ber of morphological and biochemical events that distinguish it from necrosis. Whether a cell undergoes apoptotic or necrotic demise is cell type-and stimulus-specific (30). TG-induced cell death was therefore assessed by several measures. These included: 1) electron microscopic analysis, which revealed morphological changes characteristic of apoptosis, including nuclear fragmentation and plasma membrane blebbing (Fig. 1B); 2) internucleosomal DNA fragmentation, the biochemical hallmark of apoptosis (Fig. 1C); and 3) flow cytometric analysis of DNA, demonstrating the progressive accumulation of hypodiploid DNA (Fig. 1D) (Fig. 2B). This was consistent with previous reports that the buffering capacity of BAPTA is overwhelmed with time (31). Extracellular BAPTA concentrations greater than 30 M were unsatisfactory because they induced LNCaP cell death (data not shown). Because it was also noted that the transient buffering capacity of BAPTA diminished after 6 -8 h, possibly due to intracellular compartmentalization, export, or hydrolysis (26) (data not shown), we decided to replenish BAPTA every 6 h, which maintained an effective cytosolic BAPTA concentration (Fig. 2). Because BAPTA loading alone was ineffective in preventing TG-induced elevation of cytosolic [Ca 2ϩ ], we decided to reduce I CRAC -mediated Ca 2ϩ influx using econazole and NiCl 2 , reagents that block depletion-activated Ca 2ϩ channels (32). Econazole and NiCl 2 were unsatisfactory, because concentrations inhibiting TG-induced Ca 2ϩ influx were lethal over 12-24 h (50 M and 5 mM, respectively) and lower concentrations were ineffective (data not shown). However, these studies indicated that Ca 2ϩ store release-activated Ca 2ϩ channels are present and functional on LNCaP cell membranes (33,34).
Because econazole and NiCl 2 were unsatisfactory for attenuating TG-induced Ca 2ϩ influx, EGTA was added to the extracellular medium to reduce the [Ca 2ϩ ] gradient across the plasma membrane. To determine if LNCaP cells could be grown in reduced extracellular [Ca 2ϩ ], cells were cultured in medium having [Ca 2ϩ ] ranging from 6 to 165 M, and viability was assessed after 60 h. The TG-induced I CRAC was negligible in extracellular [Ca 2ϩ ] up to 120 M (Fig. 2C), and decreased viability was detected in medium having 80 M [Ca 2ϩ ] or less (data not shown). Thus, the range of extracellular [Ca 2ϩ ] that permitted cell viability without contributing to a TG-induced cytosolic [Ca 2ϩ ] increase was 90 -120 M. However, as expected, the approximately 0.1-M increase in cytosolic [Ca 2ϩ ] resulting from intracellular store release was not attenuated by extracellular EGTA (Fig. 2C). Given this, we decided to use a combination of extracellular EGTA and intracellular BAPTA to maintain a complete cytosolic [Ca 2ϩ ] clamp. Culturing cells in RPMI with 110 M Ca 2ϩ and 7 M extracellular BAPTA/AM were therefore determined to be the optimal parameters for clamping cytosolic [Ca 2ϩ ] (Fig. 2D).
Calcium Clamp Conditions Do Not Deplete Intracellular Calcium Stores-Because high concentrations of Ca 2ϩ chelators have been shown to deplete intracellular Ca 2ϩ stores (34), it was important to determine whether the [Ca 2ϩ ] clamp conditions led to Ca 2ϩ store release. This was accomplished by exploiting the properties of reagents that specifically block depletion-activated Ca 2ϩ channels to result in I CRAC such as econazole and NiCl 2 (32). Following treatment with 5 mM NiCl 2 , the TG-induced elevation of cytosolic [Ca 2ϩ ] due to I CRAC is diminished (Fig. 3A). Therefore, we reasoned that the attenuation of I CRAC -mediated Ca 2ϩ influx with NiCl 2 can be used to detect the presence of intracellular Ca 2ϩ stores, because by definition I CRAC only occurs after intracellular Ca 2ϩ stores have been released. When cells were cultured under the [Ca 2ϩ ] clamp conditions for 24 h and treated with TG, as expected no fluctuation in cytosolic [Ca 2ϩ ] occurred, although the depletionactivated Ca 2ϩ channels likely opened in response to TG-induced Ca 2ϩ store release. However, when extracellular Ca 2ϩ was added, the cytosolic [Ca 2ϩ ] rapidly increased presumably via Ca 2ϩ influx through the already opened depletion-activated Ca 2ϩ channels, which overwhelmed the chelating ability of the cytosolic BAPTA (as in Fig. 2D). That the Ca 2ϩ influx occurred via depletion-activated Ca 2ϩ channels was confirmed by Ca 2ϩ influx attenuation with NiCl 2 . As an additional control, cells were cultured under the clamp conditions for 24 h, and extracellular Ca 2ϩ was added prior to adding TG (Fig. 3C) ] clamp yet did not deplete intracellular Ca 2ϩ stores or induce apoptosis, it was possible to determine whether Ca 2ϩ release from intracellular stores was critical for TG-induced LNCaP cell death. No protection against TG-induced apoptosis was conferred by clamping cytosolic [Ca 2ϩ ], indicating that LNCaP cell death was likely induced by the release of intracellular Ca 2ϩ stores (Fig. 4). Bearing this in mind, it is noteworthy to recall that loading LNCaP cells with high (Ͼ30 M) extracellular concentrations of BAPTA, which has been reported to cause the release of ER Ca 2ϩ stores in the absence of cytosolic [Ca 2ϩ ] elevation (34), also induced LNCaP cell death. These results are in contrast to a previous report that TG-induced apoptosis of prostate cancer cells is due to elevation of cytosolic [Ca 2ϩ ] rather than ER Ca 2ϩ store depletion (35). This discrepancy could be due to the different properties of Dunning rat AT-3 androgen-independent prostate cells and androgen-sensitive LNCaP human prostate cancer cell lines used in the study or to the different criteria used for assessing apoptosis.
Caspase Processing in LNCaP Cells-Because caspases are the enzymes responsible for the execution phase of apoptosis, it was of interest to determine which caspases were activated, the relative timing of their activation, and whether differences occurred in caspase processing when apoptosis was induced by TG under the [Ca 2ϩ ] clamp conditions versus control [Ca 2ϩ ] conditions. Processing of caspase-9, the initiator caspase associated with intracellular apoptotic stimuli, occurred by 24 h under both the control [Ca 2ϩ ] and the [Ca 2ϩ ] clamp conditions (Fig. 5A). The effector caspases-3 and -7 were also processed by 12 h under the [Ca 2ϩ ] clamp conditions and by 36 h with control [Ca 2ϩ ] (Fig. 5, B and C). That processing of the effector caspases is detected earlier than the initiator caspase-9 under the [Ca 2ϩ ] clamp conditions is possibly a reflection of relative antibody sensitivity but could also imply the existence of a mechanism that directly activates the small pro-domain effector caspases-3 and -7. Regardless, the release of intracellular Ca 2ϩ stores in the absence of cytosolic [Ca 2ϩ ] elevation results in earlier caspase activation than the dual perturbations of sustained cytosolic [Ca 2ϩ ] increase and intracellular Ca 2ϩ store release. It is important to note that no caspase activation was detected when cells were cultured for 36 h under the [Ca 2ϩ ] clamp conditions (Fig. 5).
Inhibition of TG-induced Apoptosis by Antiapoptotic Proteins-Because TG-induced caspase activation was earlier when cells were cultured under the [Ca 2ϩ ] clamp conditions versus under control [Ca 2ϩ ] conditions, we further character-ized the death pathways activated by these stimuli to determine if they were distinct. We generated stable transfected LNCaP cell lines expressing the specific inhibitor of apoptosis Bclx L and a dominant negative form of caspase-9 (caspase-9 DN-FLAG), where the catalytic cysteine had been altered to an alanine (28). Bclx L is an antiapoptotic member of the Bcl2 family of proteins and is thought to function either by directly inhibiting Apaf-1-mediated activation of caspase-9 or by attenuating release of mitochondrial cytochrome c (36,37). These stable transfected cell lines, as well as a pcDNA3 vector-transfected LNCaP cell line control, were treated with TG under control [Ca 2ϩ ] conditions or under the [Ca 2ϩ ] clamp conditions, and apoptosis was assessed by quantitation of hypodiploid DNA.
The protection afforded by the antiapoptotic proteins was consistent with the profiles of caspase activation (Fig. 6). Caspase-9 DN reduced apoptosis in cells treated with TG under control [Ca 2ϩ ] and [Ca 2ϩ ] clamp conditions. However, based on our previous studies indicating that caspase-9 DN potently blocks apoptosis (36), we would have predicted a greater degree of protection than that observed in these studies. This is, however, consistent with the possibility of direct activation of effector downstream caspases, a situation where caspase-9 DN would be of limited effectiveness given that it is thought to work by blocking the ability of endogenous caspase-9 from interacting with downstream small pro-domain caspases. Bclx L , a potent inhibitor of apoptosis that is thought to function at many levels, provided roughly equal levels of protection in cells treated under both [Ca 2ϩ ] conditions.
Caspase-9 DN Interacts with Caspase-7-We questioned whether the ability of caspase-9 DN to inhibit apoptosis, albeit partially, was related to its forming an activation-dependent complex with downstream effector caspases. Such complexes formed from the binding of catalytically inert caspase-9 DN would potentially be more stable than those formed with the To permit better visualization of the processed fragments of caspases-9 and -3, the film was exposed for a longer duration than that for the corresponding zymogen. A, caspase-9 proteolytic fragments are detected in LNCaP cells after 24  wild-type protease, which would facilitate "trapping" of an interacting substrate. Indeed, caspase-7 was found complexed to caspase-9 DN in the TG-treated transfected cell line but not in TG-treated wild-type LNCaP cells or in untreated cells under both the control [Ca 2ϩ ] and [Ca 2ϩ ] clamp conditions (Fig. 7). These results are consistent with the notion that, on receipt of a death signal, caspase-9 DN binds to caspase-7 and presumably inhibits its processing by catalytically active caspase-9.
The data obtained in these studies support the contention that intracellular Ca 2ϩ store release in the absence of cytosolic [Ca 2ϩ ] elevation is sufficient to induce apoptosis of LNCaP prostate cancer cells. Furthermore, this intracellular perturbation results in earlier caspase activation than that elicited by the dual insults of cytosolic [Ca 2ϩ ] increase and intracellular [Ca 2ϩ ] store release. Mechanisms by which these perturbations result in apoptosis are uncertain, although recently a pathway has been elucidated in hippocampal neurons involving calcineurin-mediated dephosphorylation of the Bcl2 family member BAD (Bcl2-associated death promoter), which displaces Bclx L from Apaf-1 to permit caspase-9 activation (38). Because Ca 2ϩ is a second messenger involved in multiple signaling pathways, it is likely that induction of apoptosis due to TGinduced [Ca 2ϩ ] perturbation is also multifactorial.
Our studies also show that caspase-9 is one initiator protease activated during TG-induced cell death. Caspase-9 is required for the execution of intracellular apoptotic stimuli, which is clearly demonstrated by caspase-9 knockout mice. Embryonic stem cells and fibroblasts derived from these mice are resistant to ultraviolet and gamma irradiation and their thymocytes are resistant to dexamethasone and gamma irradiation (39). How these intracellular stimuli, including [Ca 2ϩ ] perturbation, activate caspases remains unclear. were cultured for the indicated duration with 100 nM TG or without TG treatment (untreated). Cellular lysate proteins were immunoprecipitated (IP) with 10 g/ml anti-FLAG antibody (FLAG) or were not immunoprecipitated (Ϫ), resolved on acrylamide gels, and transferred to nitrocellulose membranes, which were probed with an anticaspase-7 monoclonal antibody (see "Experimental Procedures"). A, control [Ca 2ϩ ] conditions. B, [Ca 2ϩ ] clamp conditions. Under both conditions, caspase-7 forms a complex with caspase-9 DN after TG treatment.
FIG. 6. Evaluation of hypodiploid DNA in TG-treated stable transfected LNCaP cell lines. Bcl-x L is a more potent inhibitor of TG-induced LNCaP apoptosis than caspase-9 DN under both control cell culture conditions and under [Ca 2ϩ ] clamp conditions. The relative expression level of each transfected gene compared with its expression in wild-type LNCaP cells is indicated. However, it has been shown that caspase-9 and Apaf-1 are essential downstream components of p53-induced apoptosis (40). The p53 tumor suppressor protein has been implicated as a sensor of intracellular perturbations and is involved in regulating cell cycle arrest and apoptosis (41). The cyclin-dependent kinase inhibitor p21 is a downstream effector of p53 and causes an arrest in the G 1 phase of the cell cycle by at least two mechanisms: dephosphorylation of the retinoblastoma protein (Rb) (42) and complexing to proliferating cell nuclear antigen (PCNA) (43). Our previous studies have shown that TG treatment of LNCaP cells results in p21 induction, progressive Rb dephosphorylation, and p21-PCNA complex formation (44,45). Flow cytometric analyses performed in our current studies are consistent with these results, showing an accumulation of cells in G 1 -phase and a reduction of S-phase cells (data not shown). It is therefore possible that p53 is a sensor of TG-induced [Ca 2ϩ ] perturbations that induces cell cycle arrest and caspase-9 activation.
That p53 is involved in TG-induced apoptosis is also suggested by the finding that treating LNCaP cells with TG under the [Ca 2ϩ ] clamp conditions leads to earlier caspase activation than treatment under control [Ca 2ϩ ] conditions. It has been shown that the TG-induced elevation of cytosolic [Ca 2ϩ ] transiently suppresses p53-induced apoptosis, possibly by mimicking mitogenic stimuli (46). Indeed, our data also indicate that lower concentrations of TG result in LNCaP cell proliferation relative to the untreated control at early time points (Fig. 1A). This suppression of p53-induced apoptosis is dependent on extracellular [Ca 2ϩ ], suggesting that release of the intracellular stores does not result in a sufficient elevation of cytosolic [Ca 2ϩ ] to suppress apoptosis. It would therefore be expected that under the [Ca 2ϩ ] clamp conditions that abolish the TGinduced cytosolic [Ca 2ϩ ] increase, the temporary p53-mediated suppression of apoptosis would be inhibited and result in earlier caspase activation.
In summary, we have shown that TG-induced [Ca 2ϩ ] perturbations can induce apoptosis of LNCaP prostate cancer cells. By developing a technique to isolate the TG-induced release of ER Ca 2ϩ stores from the elevation of cytosolic [Ca 2ϩ ], we found intracellular Ca 2ϩ store release is by itself sufficient to induce apoptosis and that this stimulus results in earlier caspase processing than the dual stimuli of ER Ca 2ϩ store release and cytosolic [Ca 2ϩ ] elevation. However, once activated, the apoptotic pathways induced by intracellular Ca 2ϩ store release or by Ca 2ϩ store release coupled with elevated cytosolic [Ca 2ϩ ] are similar, as indicated by a similar degree of protection via antiapoptotic proteins and the formation of a complex of caspase-9 DN and caspase-7 in both scenarios.