Calcium and Calpain as Key Mediators of Apoptosis-like Death Induced by Vitamin D Compounds in Breast Cancer Cells*

The active form of vitamin D 3 (1,25(OH) 2 D 3 ) induces an increase in the intracellular free calcium ([Ca 2 (cid:1) ] i ) and caspase-independent cell death in human breast cancer cells. Here we show that the treatment of MCF-7 breast cancer cells with 1,25(OH) 2 D 3 or its chemothera- peutic analog, EB 1089, releases Ca 2 (cid:1) from the endoplasmic reticulum. The increase in [Ca 2 (cid:1) ] i was associated with the activation of a calcium-dependent cysteine protease, (cid:2) -calpain. Interestingly, ectopic expression of a calcium-binding protein, calbindin-D 28k , in MCF-7 cells not only attenuated the elevation in [Ca 2 (cid:1) ] i and calpain activation, but also reduced death triggered by vitamin D compounds. Similarly, the inhibition of calpain activity by structurally unrelated chemical inhibitors increased the survival of the cells and reduces the amount of annexin V-positive cells. Despite the complete absence of effector caspase activation,

. A desired profile of such analogs consists of a high potency for induction of antiproliferative effects, differentiation, and apoptosis combined with reduced effects on systemic calcium homeostasis compared with the parental compound. The vitamin D analog EB 1089 displays such a profile and is currently in phase III clinical trials for the treatment of cancer (5)(6)(7).
Very little is known about the signaling pathways mediating vitamin D-induced cell death. A single family of proteases, the caspases, has until recently been considered the pivotal executioner of all programmed cell death (8). Therefore it is interesting to note that breast cancer cells treated with vitamin D compounds die in the complete absence of effector caspase activation (4). Despite the lack of the caspase activation, dying cells present several characteristics of apoptosis, i.e. rounding, shrinkage, and detachment of cells as well as DNA strand breaks and DNA fragmentation (4,9,10). Furthermore, antiapoptotic proteins Bcl-2 and Bcl-X L can rescue breast cancer cells from death induced by the active form of vitamin D or its analogs (4). This apoptosis-like death program appears also independent of cysteine cathepsins and p53 tumor suppressor protein (4,11). Instead, the elevation in the intracellular free calcium ([Ca 2ϩ ] i ) brought about by vitamin D compounds correlates with the induction of apoptosis in breast cancer cells (9,12,13).
Data from studies employing various pharmaceutical modulators of calcium homeostasis have suggested that the elevation in [Ca 2ϩ ] i is a sufficient signal to induce apoptosis in several model systems, even though it may also have the opposite effect in other systems (14). Further supporting the idea that the elevation in [Ca 2ϩ ] i may mediate apoptosis, studies based on modulated expression of calcium-binding proteins, calbindin-D 28k or glucose-regulated proteins GRP78 and GRP94, have shown that Ca 2ϩ buffering can confer protection against various apoptotic stimuli (15)(16)(17)(18)(19). The calcium-dependent neutral cysteine proteases, calpains, are frequently activated in apoptosis models involving elevated [Ca 2ϩ ] i (20 -22).
Two forms of calpains, -calpain and m-calpain or type I and type II calpain, respectively, are ubiquitously expressed in human cells (23)(24)(25). The active forms of the enzymes consist of a variable large subunit (80 kDa) and a common small subunit (30 kDa). To become active, calpains require an elevation in [Ca 2ϩ ] i , and the autoproteolytic cleavage of the enzymes further enhances their activity. Whereas m-calpain requires Ca 2ϩ at a millimolar range, micromolar concentrations are enough for the activation of -calpain (in vitro; lower in cells). So far no difference in the substrate specificity of the two isozymes has been found. Growing evidence suggests that calpains may play a central role in the execution of apoptosis either upstream or downstream of caspases in, e.g. glucocorticoid-treated and irradiated thymocytes, neuronal cells exposed to various stresses, or MCF-7 breast cancer cells treated with ␤-lapachone (20, 22, 26 -29). Interestingly, caspases and calpains share several substrates (e.g. endogenous calpain inhibitor calpastatin, fodrins, focal adhesion kinase, calmodulin-dependent kinases, actin, vimentin, and keratins), suggesting that these enzymes may execute the cell in a partially indistinguishable manner (23). Calpains can also cleave other potentially interesting apoptosis-related proteins including caspase-12, Bax, Bcl-X L , GRP94, c-Fos, and p53 (19, 20, 30 -32).
This study was undertaken to enlighten the poorly understood signaling pathway mediating vitamin D-induced caspaseindependent death of MCF-7 breast cancer cells. The focus was centered on finding a responsible protease for this form of cell death and characterization with respect to the ultrastructural morphology of the dying cells.

EXPERIMENTAL PROCEDURES
Compounds and Plasmids-1,25(OH) 2 D 3 and EB 1089 were generous gifts from Christina Mørk Hansen and Lise Binderup (LEO Pharmaceutical Products, Ballerup, Denmark). Both compounds were stored at Ϫ20°C and diluted in isopropyl alcohol. Control cells were always treated with the isopropyl alcohol vehicle (1% v/v). TNF was a generous gift from Anthony Cerami (Kenneth Warren Laboratories, Tarrytown, NY). Recombinant human TNF-related apoptosis-inducing ligand (Alexis Corp., San Diego) was used with 2 g/ml anti-FLAG ® M2 monoclonal antibody (F3165, Sigma). N-Acetyl-Leu-Leu-Nle-CHO (calpain inhibitor I) dissolved in ethanol to 20 mM was used at 10 M final concentration. N-Acetyl-Leu-Leu-Met-CHO (calpain inhibitor II) dissolved in dimethyl sulfoxide to 20 mM was used at 20 M final concentration. (2S,3S)-trans-Epoxysuccinyl-L-leucylamido-3-methylbutane ethyl ester (EST or E64d) dissolved in ethanol to 100 mM was used at 100 M final concentration. PD 150606 dissolved in dimethyl sulfoxide to 50 mM was used at 50 M. All were purchased from Calbiochem-Novabiochem Co. Z-Val-Ala-D,L-Asp-fluoromethylketone (zVAD-fmk) was dissolved in ethanol to 10 mM and used at 1 M. The compound was purchased from Bachem (Bubbendorf, Switzerland). Thapsigargin, Pluronic F-127 and fura-2/AM were from Molecular Probes (Eugene, OR).
The pEBS7-calbindin-D 28k plasmid was created by subcloning a 1-kb EcoRI fragment of pGEM-calbindin (kindly provided by Dr. Diane Dowd, St. Louis University, St. Louis, MO) containing the sequence encoding for the complete chicken calbindin-D 28k (33) into the pEBS7 plasmid.
Cell Lines-The MCF-7S8 subclone (34) of MCF-7 breast carcinoma cells with a sensitivity to vitamin D-induced apoptosis comparable with that of parental cells was used throughout the study (4,10). The MCF-cont and MCF-calb28 cells are pools of MCF-7S8 cells transfected with empty pEBS7 vector or pEBS7-calbindin-D 28k , respectively. Cells were transfected by electroporation (960 microfarads, 330 V) and selected for hygromycin (0.15 mg/ml) resistance.
The cells were grown in RPMI 1640 with Glutamax (Invitrogen) supplemented with 5% fetal calf serum (Bio Whittaker Europe, Belgium), 100 units/ml penicillin, and 100 g/ml streptomycin. The transfection did not change the proliferation rate of the MCF-cont and the MCF-calb28 cells compared with the parental cell line under normal growth conditions. The cells were tested regularly and found negative for mycoplasma.
Measurement of Apoptosis and Cell Density-The MTT tetrazolium assay was used to measure the density of viable cells as described previously (4). The Cell Death Detection ELISA TM kit (Roche Molecular Biochemicals) was used to quantitate DNA fragmentation as described previously (11).

Measurement of [Ca 2ϩ ] i and the Release of Ca 2ϩ
Stores from the ER-The [Ca 2ϩ ] i was measured using fluorescence ratiometric high resolution digital imaging employing the Ca 2ϩ indicator fura-2/AM as described previously (13,35). Briefly, cells grown on coverslips were loaded with 2 M fura-2/AM (cell-permeable acetoxymethyl ester) in Dulbecco's phosphate-buffered saline supplemented with 0.1% dimethyl sulfoxide and 0.01% Pluronic F-127 for 30 min at 37°C. After incubation, cells were maintained in phosphate-buffered saline for a 15-min recovery period (to allow the intracellular esterases to cleave the dye completely to the cell-impermeable form) and analyzed directly thereafter, retention never exceeding 30 min. The dynamics of intracellular Ca 2ϩ were assessed with cells placed in the microincubation chamber (37.0°C Ϯ 0.2°C) on a Nikon Eclipse TE-300 inverted microscope equipped for fluorescence digital ratiometric imaging (Fryer Co., Huntley, IL). The images were captured using a Nikon Super Fluor 40ϫ oil immersion objective and CoolSnapFX digital CCD camera (Photometrics, Trenton, NJ), ratioed (340/380 nm excitation, 510 nm emission) on a pixel-by-pixel basis, and stored for analysis.
Mobilization of Ca 2ϩ from the ER stores was accomplished by using thapsigargin, which specifically inhibits ER Ca 2ϩ -ATPase and results in the rapid Ca 2ϩ release from the ER. To measure Ca 2ϩ release from the ER stores, 1 M thapsigargin was added to the cell preparations after recording the basal [Ca 2ϩ ] i for 1-3 min, and the peak values of the [Ca 2ϩ ] i increase were measured. The fluorescence of the Ca 2ϩ signal was calibrated at the end of experiments (35). Ca 2ϩ saturation was achieved by adding 5 M ionomycin in the presence of 5 mM Ca 2ϩ and virtually zero Ca 2ϩ by the further addition of 10 mM EGTA; 5 mM MnCl 2 was subsequently added to obtain background fluorescence. The concentration of vehicle, dimethyl sulfoxide, never exceeded 0.1%; the vehicle did not affect [Ca 2ϩ ] i at this concentration. Image and data analyses were performed as described previously (35,36).
Annexin V Staining and Analysis by FACS-Cells (5 ϫ 10 6 cells/ sample) were washed in phosphate-buffered saline and detached cells from medium, as well as cells detached by trypsinization, were collected, centrifuged, and washed twice in binding buffer (10 mM Hepes pH 7.4, 150 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 ) before incubation in binding buffer supplemented with 1:100 annexin V (Roche Molecular Biochemicals) and 1 g/ml propidium iodide for 20 min at room temperature protected from light. After staining, the cells were handled ice-cold, washed twice in phosphate-buffered saline with calcium and magnesium before analyzing using a Becton Dickinson flow cytometer. The amount of cells presented in the plots was 20,000.
Transmission EM-The cells were grown in Petri dishes and fixed for 10 min in 50% Karnovsky solution at 25°C. Cells were collected by scraping with a rubber policeman, transferred to a centrifugation tube, and centrifuged at 1,500 ϫ g for 5 min. After removal of the supernatant the pellet was washed once and subsequently stored in 70% Karnovsky solution at 4°C until embedding. After two 10-min washings in 0.1 M cacodylate buffer and centrifugation as above, the pellet was embedded in agar, cut into smaller cubes, and fixed in osmium and embedded in Epon according to standard procedures. Ultrathin sections were collected on Formwar-coated copper grids, stained according to standard procedures, and examined in a JEOL 1210 electron microscope.

MCF-7 Cells Expressing Calbindin-D 28k
Are Protected against Apoptosis Induced by 1,25(OH) 2 D 3 -After treatment of MCF-7 cells with 1,25(OH) 2 D 3 , the appearance of apoptotic cells becomes evident first after 3-5 days (4). During this relatively long latent period the cells are growth-arrested, and unknown cellular changes may occur. Previous studies have shown that the treatment of breast cancer cells with vitamin D compounds results in the elevation of [Ca 2ϩ ] i (9,12,13). To test whether this calcium signal plays a role in vitamin D-induced apoptosis, we introduced a cDNA encoding a calcium-binding protein, calbindin-D 28k , into MCF-7 cells. Immunoblot analysis employing anti-calbindin-D 28k antibodies did not reveal a detectable level of the protein in MCF-7 cells left untreated or treated with 100 nM 1,25(OH) 2 D 3 (not shown). Analysis of the MCF-7 cell pool transfected with pEBS7-calbindin-D 28k and selected for hygromycin resistance (MCF-calb28) revealed a single band detectable with the anti-calbindin-D 28k antibody (Fig. 1A). Based on the migration in the SDS-PAGE, the size of the detected protein was double the expected size (56 kDa). Because a single anti-calbindin-D 28k -reacting protein of the same size was also observed in MCF-7 cells transiently transfected with pEBS7-calbindin as well as in MG-63 osteosarcoma cells known to express calbindin-D 28k (16), but not in vectortransfected MCF-7 cells (MCF-cont), the detected protein was likely to be calbindin-D 28k . As analyzed by the MTT cell density assay, 1,25(OH) 2 D 3 treatment for up to 4 days resulted in similar growth inhibition in MCF-calb28 and MCF-cont cells (Fig. 1B). Interestingly, after day 4 of the treatment, the MCF-calb28 cells were, however, clearly more resistant to 1,25(OH) 2 D 3 , suggesting that the antiproliferative effect of the compound was unaffected by the presence of calbindin-D 28k , but the onset of apoptosis occurring after day 4 was inhibited. These data correlated with the microscopic examination showing clearly reduced appearance of cells with apoptotic morphology in MCF-calb28 cells compared with MCF-cont cells (not shown). Although MCF-calb28 cells were almost completely rescued from apoptosis induced by 10 nM 1,25(OH) 2 D 3 , only partial inhibition was observed when a 10ϫ higher concentration of 1,25(OH) 2 D 3 was used. Contrary to this form of cell death, MCF-7 cell apoptosis induced by 1-10 ng/ml TNF or 10 -100 ng/ml TNF-related apoptosis-inducing ligand was not affected by the ectopic expression of calbindin-D 28k (Fig. 1C and data not shown). In both cell lines EB 1089 was clearly more potent in increasing [Ca 2ϩ ] i than 1,25(OH) 2 D 3 . Both vitamin D compounds failed to evoke rapid (from seconds to 1 h) Ca 2ϩ responses (data not shown), and even after a 1-day treatment the [Ca 2ϩ ] i remained unchanged (Fig. 2, B and C).

Calbindin-D 28k Buffers the Elevation of [Ca 2ϩ ] i Induced by Vitamin D Compounds or Thapsigargin-To
Next, we tested whether the calcium signal evoked by vitamin D compounds originated from the ER, which serves as an important intracellular reservoir of Ca 2ϩ . Thapsigargin is a specific mobilizer of the ER Ca 2ϩ stores and causes a rapid and transient increase in [Ca 2ϩ ] i in most cells, including the MCF-7 cells (9,35). Thapsigargin evoked typical Ca 2ϩ responses in both MCF-cont and MCF-calb28 cells, but the magnitude of the [Ca 2ϩ ] i increase was significantly lower in the calbindin-D 28kexpressing cells, indicating that the exogenous calbindin-D 28k is capable of buffering the Ca 2ϩ released from the ER (Fig. 2D). Interestingly, treatment of the MCF-cont cells with 1,25(OH) 2 D 3 or EB 1089 resulted in a decrease in the thapsigargin-induced Ca 2ϩ response in a time-dependent manner (Fig. 2E). These results imply that vitamin D compounds cause a sustained depletion of the ER Ca 2ϩ stores. MCF-calb28 cells treated with 1,25(OH) 2 D 3 or EB 1089 for 1 day responded to thapsigargin by a lower increase in [Ca 2ϩ ] i than MCF-cont cells (Fig. 2D). This effect was, however, not evident at later time points (data not shown). This suggests that the buffering capacity of calbindin-D 28k in the MCF-calb28 cells is exceeded during the sustained treatment of the cells with vitamin D compounds. Overall, these data indicate that 1,25(OH) 2 D 3 or EB 1089 results in increased [Ca 2ϩ ] i by inducing a sustained release of Ca 2ϩ from the ER. Furthermore, the forced expression of calbindin-D 28k increases the Ca 2ϩ buffering capacity of the cells, thus partially counteracting the Ca 2ϩ -mediated signaling induced by 1,25(OH) 2 D 3 and EB 1089.
Vitamin D Compounds Activate Calpain-To investigate whether calcium-dependent proteases, calpains, are activated during treatment of MCF-7 cells with vitamin D compounds, we analyzed protein samples from cells treated with 1,25(OH) 2 D 3 or EB 1089 for characteristic signs of calpain activity by immunoblot analysis. Although caspases and calpains share a number of substrates, including fodrin and calpastatin, their cleavage during treatment with 1,25(OH) 2 D 3 or EB 1089 in MCF-7 cells would indicate calpain activity because effector caspases are not activated in this model (4). As shown in Fig. 3, fodrin processing becomes evident in cells treated with vitamin D compounds for 4 days and increases at days 5 and 6. These data strongly suggest the presence of active calpain, and therefore we subsequently analyzed whether the -calpain or the intracellular inhibitor of calpain, calpastatin, was cleaved (23). MCF-7 cells express a high level of the large subunit of -calpain. In contrast to vehicle-treated control cells, cells treated with 1,25(OH) 2 D 3 or EB 1089 also contained a cleaved form of -calpain, suggesting that the enzyme is, indeed, activated by vitamin D compounds (Fig. 3). Moreover, the level of the calpain substrate, calpastatin, was clearly reduced after 6 days treatment with 1,25(OH) 2 D 3 or EB 1089 (Fig. 3). Overall we conclude that the treatment of MCF-7 cells with 1,25(OH) 2 D 3 or EB 1089 leads to the activation of -calpain occurring in parallel with the onset of apoptosis.
Calpain Inhibitors Partially Inhibit Apoptosis Induced by Vitamin D Compounds-The apoptosis-like cell death induced by vitamin D compounds is associated with DNA fragmentation (10). To test whether the activation of calpain is necessary for this nuclear sign of apoptosis, the effect of protease inhibitors on the degree of vitamin D-induced DNA fragmentation was analyzed by an enzyme-linked immunosorbent assay measuring histone-associated cytoplasmic DNA. As expected, the pancaspase inhibitor zVAD-fmk at a concentration specific for caspases (1 M) could not inhibit the DNA fragmentation induced by a 6-day treatment with 100 nM 1,25(OH) 2 D 3 (Fig.  4A). A 10ϫ higher concentration of zVAD-fmk was also without effect on the survival of 1,25(OH) 2 D 3 -treated MCF-7 cells (data not shown). On the contrary, three structurally different calpain inhibitors, calpain inhibitor II, EST, and PD 150606, all proved partially inhibitory (Fig. 4A). Also, the reduction in cell density in cultures treated with 100 nM 1,25(OH) 2 D 3 for 6 days was partially attenuated in the presence of calpain inhibitors (Fig. 4B). It should be noted that the treatment of the MCF-7 cells with vitamin D compounds reduces cell density in a 6-day assay by two mechanisms, i.e. first by the inhibition of proliferation and then by the induction of cell death (4). Therefore, only a partial protection against cell reduction can be expected even if the latter was completely inhibited; e.g. enforced expression of Bcl-2 that confers complete protection against apoptosis or EB 1089 (right dot blot) displayed 9.6% or 13.6%, respectively. At later stages in this death process after the cells have been annexin V-positive only the membrane integrity is lost, and the cells become propidium iodide-positive too. Presenting the annexin V staining of the cells as a histogram with control treated cells overlaid with 1,25(OH) 2 D 3 -treated cells shows that the whole cell population has changed the degree of annexin V staining (Fig. 4C, lower left panel). Notably, the peak of the cell population has changed and not only the percentage of cells over a threshold value. Dosing the cells with PD 150606 during the last 2 days of the 6-day treatment with vehicle or 1,25(OH) 2 D 3 clearly reduces the annexin V staining for the 1,25(OH) 2 D 3 -treated cells (Fig. 4C, lower middle panel). Under these conditions, PD 150606 conferred protection against 1,25(OH) 2 D 3 -induced membrane flip causing annexin V staining of the cells. This resulted in the 1,25(OH) 2 D 3 -treated cells not being easily discriminated from the control cells as was the case without PD 150606 (Fig. 4C, lower left and middle panels). Using zVAD-fmk under the same conditions could not lower the peak value of annexin V-stained cells treated with 1,25(OH) 2 D 3 , although this concentration clearly reduced the TNF annexin V staining (Fig. 4C, lower right panel; not  shown).
To ensure that the calpain inhibitors really inhibited apoptosis because of the inhibition of -calpain and not because of an undefined side effect, we studied the effects of calpain inhibitor II and PD 150606 in more detail. The extent of fodrin cleavage was studied in cells cotreated with vitamin D compound (6 days) and calpain inhibitor (last 2 days). Under these conditions, the calpain inhibitors did not completely abolish the fodrin cleavage, but both calpain inhibitor II (Fig. 5A) and PD 150606 (Fig. 5B) resulted in a clear reduction in the degradation. This correlates well with the degree of protection against cell reduction and DNA fragmentation (Fig. 4, A and B). We also compared the fodrin cleavage in control and calbindin-D 28k -expressing cells after treatment with 10 or 100 nM 1,25(OH) 2 D 3 . The MCF-calb28 cells show clearly less fodrin cleavage in response to treatment with 1,25(OH) 2 D 3 than the MCF-cont cells (Fig. 5C). Thus, the partial protection conferred by calbindin-D 28k against cell death induced by 1,25(OH) 2 D 3 is also accompanied by the partial inhibition of calpain activity as measured by fodrin cleavage.
Vitamin D Compounds Induce Apoptosis-like Morphological Changes-Because caspase-independent apoptosis with the calpains as central mediators has not been described before, we examined by EM the morphology of the dying MCF-7 cells after treatment with vitamin D compounds. Contrary to control cells (left), cells treated with either EB 1089 (right) or 1,25(OH) 2 D 3 (middle) showed apoptosis-like morphology characterized by condensed fragmented chromatin marginalized along the nuclear envelope, condensed cytoplasma and condensed nuclei (Fig. 6). Furthermore, the dying cells had several light vesicles caused by dilated ER, an effect that could be expected during ER stress occurring after depletion of the calcium store. Such light vesicles were never present in control cells (Fig. 6). DISCUSSION In the present study we have identified a novel death pathway in breast carcinoma cells treated with 1,25(OH) 2 D 3 or its chemotherapeutic analog EB 1089. This pathway involves Ca 2ϩ release from the ER and -calpain activation. Despite the lack of the activation of known caspases, transmission EM images of the MCF-7 cells treated with vitamin D compounds demonstrated an apoptosis-like morphology characterized by condensed cytoplasma, nuclei, and chromatin. Compared with classical caspase-dependent apoptosis models displaying geometrical condensation of chromatin (38), chromatin of the cells treated with vitamin D compounds condensed to lumpier and less compact structures. Similar "incomplete" chromatin condensation has been observed in several other caspase-independent apoptosis models, implying that the activation of caspases may be required for the complete chromatin condensation (39 -44). It should be noted, however, that also TNFtreated tumor cells (including MCF-7 cells) dying in a caspasedependent manner show incomplete chromatin condensation (42,45).
In addition to typical apoptotic changes, the EM of the dying cells revealed dilated ER, indicative of ER stress. The ER may participate in the initiation of apoptosis by at least two different mechanisms, i.e. Ca 2ϩ signaling and unfolded protein re- sponse (14,46). Thus, it is interesting to note that the treatment of the MCF-7 cells with 1,25(OH) 2 D 3 or EB 1089 for 3-5 days caused an elevation in [Ca 2ϩ ] i . This was associated with a sustained reduction in the ER Ca 2ϩ stores, suggesting that vitamin D compounds, indeed, induced a release of Ca 2ϩ from the ER. The late occurrence of the calcium signal correlates well with previous literature, suggesting that vitamin D-induced changes in gene expression are needed for the Ca 2ϩ release to occur in breast cancer cells (12,13,47). Furthermore, the kinetics of Ca 2ϩ release correlated well with the onset of apoptosis suggesting that Ca 2ϩ may play a direct role in the apoptosis induction. In support of this idea, the partial stabilization of [Ca 2ϩ ] i by the enforced expression of calbindin-D 28k partially protected cells against apoptosis induced by vitamin D compounds. Interestingly, modulation of ER Ca 2ϩ fluxes has also been suggested to be one of the modes by which Bcl-2 inhibits apoptosis (49,50). In line with this, Bcl-2 protects the MCF-7 cells and other cancer cells against apoptosis induced by vitamin D compounds (4,51).
Although calbindin-D 28k in some models has been shown to confer protection against apoptosis by directly inhibiting caspases, such a mode of action is, however, unlikely in our model system because the death is independent of known caspases (4,16,17,52). Furthermore, MCF-7 cells expressing calbindin-D 28k were as sensitive as the control cells to death receptor-induced caspase-dependent apoptosis. Instead of inhibiting caspases, the calbindin-D 28k -mediated stabilization of [Ca 2ϩ ] i inhibited the activation of a calcium-dependent neutral cysteine protease, calpain. The activation of calpain upon a 4 -6-day treatment with vitamin D compounds was demonstrated by the cleavage of fodrin and calpastatin, which are specific markers for calpain activation in the absence of parallel effector caspase activation (23). Furthermore, cleavage of the large subunit of -calpain paralleled the cleavage of the substrates. Because no difference in the substrate specificity of the calpain isozymes has been found (23-25), we conclude that vitamin D compounds at least activated -calpain and possibly also other calpains.
The role of calpain in the apoptosis pathway was assessed by employing calpain inhibitors acting by binding either to the active site (calpain inhibitor II and EST) or to the Ca 2ϩ binding site (PD 150606) of the enzyme (43). Remarkably, both types of inhibitors protected MCF-7 cells against DNA fragmentation and cell density reduction induced by vitamin D compounds. Furthermore, PD 150606 also inhibited increased phosphatidylserine exposure at the cell membrane after treatment with 1,25(OH) 2 D 3 as detected by FACS analysis of annexin V-stained cells. It should be noted that calpain inhibitors conferred only a partial protection against apoptosis induced by vitamin D compounds. The lack of complete protection could be caused by subtoxic effects or too low a concentration of the inhibitors (all inhibitors displayed some toxicity during the long term application or at higher doses; not shown). The latter possibility is supported by the fact that the inhibitors reduced the activation of calpains only partially. Alternatively, other effector molecules in addition to calpains may be involved in this apoptosis pathway.
Except for the lack of geometrical condensation of chromatin, our data suggest that calpains can mimic caspases as the major execution proteases in apoptosis-like cell death induced by vitamin D compounds. Previously, calpains have been demonstrated to mediate differentiation and necrosis and to participate in caspase-dependent apoptosis pathways (19 -23, 28, 29, 43, 53-55). In most apoptosis models calpains act upstream of caspases (43). Why calpain does not activate caspases in vitamin D-treated MCF-7 cells is as yet unclear. It is, however, unlikely to be the result of any dominant caspase inhibitor induced by vitamin D compounds because pretreatment of MCF-7 cells with 1,25(OH) 2 D 3 does not inhibit but, in fact, enhances TNF-induced caspase activation (11,56). Furthermore, it is not because of the lack of caspase-3 expression in MCF-7 cells. Vitamin D compounds also fail to induce any measurable effector caspase activity in MCF-7 cells expressing ectopic caspase-3 or in T47D cells expressing high levels of endogenous caspase-3 (4). Treatment of cancer with vitamin D compounds seems to be an attractive suggestion for the future, in particular for the treatment of cancers originating from cells in which calpains are readily activated, but indeed also for combination therapy with agents inducing different apoptotic pathways. The fact that vitamin D compounds induce a form of apoptosis different from the caspase-dependent pathway induced by the majority of anticancer drugs may prove very useful. Defects in caspase-dependent apoptosis signaling pathways are common in cancer cells (57). The problem of resistance may, of course, also hamper vitamin D-based therapies. Long term treatment with vitamin D compounds may in some cancer types induce resistance against themselves via the induced expression of calbindin-D 28K (58). No induction of calbindin-D 28K was, however, detected in MCF-7 cells treated with vitamin D compounds, indicating that this induction is not a universal phenomenon (not shown). Furthermore, putative resistance of renal cancer, osteoblastic tumors, as well as other malignancies with high constitutive expression levels of calbindin-D 28K should be kept in mind. The use of combinations of therapeutical agents that trigger cell death by independent pathways may circumvent resistance problems. Promising in vitro data showing synergistic apoptosis-inducing effects of combination treatments employing vitamin D analogs with agents triggering other death pathways strongly support this idea (11,48,56).