Glycogen Synthase Kinase-3β and Caspase-2 Mediate Ceramide- and Etoposide-Induced Apoptosis by Regulating the Lysosomal-Mitochondrial Axis

Glycogen synthase kinase-3β (GSK-3β) regulates the sequential activation of caspase-2 and caspase-8 before mitochondrial apoptosis. Here, we report the regulation of Mcl-1 destabilization and cathepsin D-regulated caspase-8 activation by GSK-3β and caspase-2. Treatment with either the ceramide analogue C2-ceramide or the topoisomerase II inhibitor etoposide sequentially induced lysosomal membrane permeabilization (LMP), the reduction of mitochondrial transmembrane potential, and apoptosis. Following LMP, cathepsin D translocated from lysosomes to the cytoplasm, whereas inhibiting cathepsin D blocked mitochondrial apoptosis. Furthermore, cathepsin D caused the activation of caspase-8 but not caspase-2. Inhibiting GSK-3β and caspase-2 blocked Mcl-1 destabilization, LMP, cathepsin D re-localization, caspase-8 activation, and mitochondrial apoptosis. Expression of Mcl-1 was localized to the lysosomes, and forced expression of Mcl-1 prevented apoptotic signaling via the lysosomal-mitochondrial pathway. These results demonstrate the importance of GSK-3β and caspase-2 in ceramide- and etoposide-induced apoptosis through mechanisms involving Mcl-1 destabilization and the lysosomal-mitochondrial axis.


Lysosomal membrane permeabilization (LMP) assay
For LMP assay, cells were treated with 5 μg/ml of acridine orange (AO; Sigma-Aldrich) in serum-free RPMI or DMEM for 15 min at 37°C. After being washed with phosphate-buffered saline (PBS) twice, a flow cytometer (FACSCalibur; BD Biosciences, San Jose, CA) was used in FL-3 channel (>650 nm) following excitation at 488 nm. Meanwhile, cells were observed using a laser scanning confocal microscope (Leica TCS SPII, Nussloch, Germany) equipped with an argon ion laser (488 and 514 nm).

Mitochondrial function assay
For the detection of the loss of mitochondrial transmembrane potential (MTP), cells were incubated with 50 μM rhodamine 123 (Sigma-Aldrich) in culture medium for 30 min at 37°C. After being washed with PBS twice, cells were analyzed by using a flow cytometer (FACSCalibur) in FL-1 channel (515-545 nm) following excitation at 488 nm.

Analysis of cell apoptosis
Cell apoptosis characterized by DNA fragmentation was detected using propidium iodide (PI; Sigma-Aldrich) staining. After fixation with 70% ethanol in PBS, cells were stained with PI/ RNase working solution in PBS containing 40 μg/ml PI and 100 μg/ml RNase A (Sigma-Aldrich) for 30 min at room temperature and then analyzed using flow cytometry (FACSCalibur) with excitation at 488 nm and emission detected in the FL-2 channel (564-606 nm). Samples were analyzed using CellQuest Pro version 4.0.2 software (BD Biosciences), and quantification was carried out with WinMDI version 2.8 software (The Scripps Institute, La Jolla, CA). Small cell debris was excluded by gating for forward scatter. To validate the quantification of apoptotic cells from flow cytometric analysis, cell apoptosis was also detected using 4,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) staining at a concentration of 5 μg/ml at room temperature for 10 min. After being washed with PBS, the cells were visualized under a fluorescent microscope (IX71, Olympus, Japan).

Immunostaining
Cells were fixed with 1% formaldehyde in PBS and permeabilized with 0.01% saponin in PBS. Rabbit polyclonal antibodies against human cathepsin D were used, followed by Alexa 488-conjugated goat anti-rabbit IgG (Invitrogen) staining. Cells were observed using a laser scanning confocal microscope (Leica TCS SPII). DAPI (Invitrogen) was used for nuclear staining.

Western blot analysis
Total cell extracts were separated by SDS-PAGE and then transferred to a polyvinylidene difluoride membrane (Millipore Corporation, Billerica, MA). After blocking, blots were developed with a series of antibodies as indicated. Rabbit antibodies specific for mouse and human caspase-2, -3, -8, poly(ADP-ribose) polymerase (PARP) (Cell Signaling Technology, Beverly, MA), Bid, and tBid (Oncogene, San Diego, CA) were used. Monoclonal antibodies against βactin and LAMP-1 (Sigma-Aldrich) and rabbit antibodies against mouse and human cathepsin D, Mcl-1, and COX IV (Cell Signaling Technology) and rabbit anti-GSK-3β (Santa Cruz Biotechnology) were used. Finally, blots were hybridized with horseradish peroxidase-conjugated goat anti-rabbit IgG or anti-mouse IgG (Calbiochem) and developed using enhanced chemiluminescence (Pierce, Rockford, IL).

Detection of caspase-8 activity
Cellular caspase activation was determined using a caspase-2 and caspase-8 assay kit (Calbiochem) according to the manufacturer's instructions. Optical density (OD) measurements were performed using a Spectra MAX 340PC microplate reader (Molecular Devices, Sunnyvale, CA).

Purification of lysosomes
Lysosomes were isolated using a Lysosome Isolation Kit (Sigma-Aldrich) according to the manufacturer's instructions. Cells equaling 1 ml of packed cell volume were homogenized in extraction buffer (0.25 M sucrose, 20 mM HEPES/KOH, 1 mM EDTA, 2 mM MgCl 2 , and 10 mM KCl) with a Dounce homogenizer. The homogenate was centrifuged at 1,000×g for 10 min, and the resulting supernatant was centrifuged at 20,000×g for 20 min to yield the crude lysosomal fraction (pellet). To further enrich the lysosomes, the crude lysosomal fraction was suspended and diluted with 2.3 M sucrose to a 19% gradient fraction and then loaded on an assembled sucrose gradient (27>22.5>19>16>12>8%) followed by centrifugation at 45,000×g for 3 h without braking. Several bands formed after centrifugation, and the top band, which contained the isolated lysosomes, was carefully removed and saved on ice for further examination.

Statistical analysis
Student's two-tailed unpaired t test or one-way ANOVA analysis followed by Dunnet post-hoc test, as appropriate with commercially available statistical software (SigmaPlot 8.0 for Windows; Systat Software, Inc., San Jose, CA) were performed. Values are expressed as means ± S.D. and a P-value of 0.05 was considered statistically significant.

Ceramide or etoposide induces LMP, MTP reduction, and apoptosis
We used the ceramide analogue C 2 -ceramide (25 μM) or the topoisomerase II inhibitor etoposide (50 μM) to induce apoptosis in mouse 10I T hybridoma cells and human A549 lung epithelial carcinoma cells. The kinetics (Fig 1A top) and dose responses ( Fig 1B) of C 2 -ceramide-induced LMP in 10I cells were shown by staining with AO, a monomeric cationic fluorescent dye, followed by flow cytometric analysis. By staining with the lipophilic cationic fluorochrome rhodamine 123, we found that C 2 -ceramide induced a time-dependent MTP reduction (Fig 1A middle) in 10I cells. Furthermore, PI staining followed by flow cytometric analysis demonstrated that C 2 -ceramide caused 10I cell apoptosis in a time- (Fig 1A, bottom) and dose-dependent manner ( Fig 1B). However, C 2 -dihydroceramide, a stereoisomer of ceramide, did not cause LMP or apoptosis in 10I cells ( Fig 1B). Further results confirmed that either C 2 -ceramide or etoposide caused LMP, MTP reduction, and apoptosis in A549 cells ( Fig  1C). These results show that either ceramide or etoposide can induce lysosomal destabilization and mitochondrial apoptosis.

Ceramide or etoposide induces mitochondrial apoptosis in a lysosomal cathepsin D-regulated manner
Lysosomes contain numerous proteases, such as cathepsin B/L and cathepsin D, which are activated and released following LMP [1, 2,8]. To examine the potential roles of lysosomal cathepsins on ceramide (25 μM)or etoposide (50 μM)-induced mitochondrial apoptosis, by using rhodamine 123 and PI staining, respectively, we found that the cathepsin D inhibitor pepstatin A (25 μM) blocked both C 2 -ceramide-and etoposide-induced MTP reduction (Fig 2A, top) as well as apoptosis (Fig 2A, bottom). To further confirm these results, we introduced siRNA specific for cathepsin D into A549 cells to block cathepsin D expression. The inhibition of cathepsin D expression was observed in cathepsin D siRNA (50 nM)-expressing cells but not in scramble control cells (Fig 2B, left). Both C 2 -ceramide-and etoposide-induced A549 cell apoptosis were blocked in cathepsin D siRNA-transfected cells, as determined by PI staining  followed by flow cytometric analysis (Fig 2B, right). We next investigated whether ceramide or etoposide caused the release of cathepsin D from lysosomes to the cytoplasm. Using a cathepsin D-specific antibody, either C 2 -ceramide or etoposide caused cathepsin D re-localization ( Fig  2C) from lysosomes (characterized by punctate staining) to the cytoplasm (characterized by diffuse staining). These results indicate that either ceramide or etoposide can induce LMP, followed by cathepsin D-regulated mitochondrial apoptosis.
Ceramide or etoposide induces cathepsin D-regulated caspase-8 but not caspase-2 activation followed by mitochondrial apoptosis We previously demonstrated that ceramide caused the sequential activation of caspase-2 and caspase-8 upstream of mitochondrial apoptosis [24]. We next examined the association of cathepsin D inhibition with defective initiator caspase activation. We first inactivated cathepsin D in 10I cells by pretreatment with pepstatin A (25 μM). By using caspase activity detection and Western blot analysis, the results showed that pretreatment with pepstatin A suppressed the activation of caspase-8, Bid, and caspase-3, as well as PARP cleavage, but not caspase-2, in response to either C 2 -ceramide (25 μM) (Fig 3A, left) or etoposide (50 μM) (Fig 3A, right) stimulation. Further confirming these results, cathepsin D siRNA (50 nM)-treated cells were defective in the C 2 -ceramide-induced activation of caspase-8, Bid, caspase-3, and PARP compared to the scramble control. However, cathepsin D siRNA did not inhibit caspase-2 activation (Fig 3B). We then investigated whether cathepsin D caused caspase-8 activation in vitro. Caspase-8 activity assays and Western blot analyses demonstrated that cathepsin D caused caspase-8 activation (Fig 3C). Therefore, cathepsin D is required for either ceramide or etoposide to induce the activation of caspase-8, but not caspase-2, and subsequent apoptotic signals of mitochondrial damage. Inhibiting GSK-3β or caspase-2 reduces ceramide-induced LMP, cathepsin D re-localization, and cell apoptosis We previously demonstrated that GSK-3β activation was essential for ceramide-induced caspase-2 and caspase-8 activation [11]. Because cathepsin D acts upstream of caspase-8 but downstream of caspase-2, we next examined whether caspase-2 and GSK-3β caused lysosomal destabilization. Using AO staining followed by flow cytometric analysis, we found that both the caspase-2 inhibitor z-VDVAD-fmk (10 μM) and the GSK-3β inhibitor LiCl (10 mM) blocked C 2 -ceramide (25 μM) or etoposide (50 μM)-induced LMP in 10I cells (Fig 4A). In addition, confocal microscopic observation confirmed that treating cells with z-VDVAD-fmk or LiCl resulted in a blockade of cathepsin D re-localization from lysosomes to the cytoplasm in C 2 -ceramide-or etoposide-treated A549 cells (Fig 4B). To further confirm the effects of caspase-2 and GSK-3β, a lentiviral-based short hairpin RNA (shRNA) approach was performed. Western blot analysis showed the silencing of caspase-2 (Fig 4C, left top) or GSK-3β (Fig 4C, left bottom) expression in 10I cells treated with specific shRNAs. Caspase-2- (Fig 4C, middle) or GSK-3β-silenced cells (Fig 4C, right) were partially defective in C 2 -ceramide-or etoposide- induced LMP as well as cell apoptosis. These results demonstrate that both ceramide and etoposide can cause GSK-3β-and caspase-2-regulated LMP, cathepsin D re-localization, and cell apoptosis.

Lysosomal Mcl-1 is required for lysosomal stabilization
To further confirm the specific expression of Mcl-1, lysosomes were isolated from untreated 10I or A549 cells. Western blot analysis demonstrated the presence of Mcl-1 in mitochondrial extracts and in lysosomal extracts with no detectable mitochondrial contamination (Fig 6A). To examine the crucial role of Mcl-1 for lysosomal stabilization, artificially forced expression of Mcl-1 in A549 cells was shown by Western blotting (Fig 6B, left). Using AO, rhodamine 123, and PI staining, respectively, results demonstrated that Mcl-1 overexpression caused resistance in response to C 2 -ceramide (25 μM)or etoposide (50 μM)-induced LMP, MTP reduction, and apoptosis (Fig 6B, right). These findings demonstrate that Mcl-1 is also critical for the maintenance on lysosomal stability, which protects cells from ceramide-or etoposide-induced apoptosis.

Discussion
Ceramide or etoposide causes sequential activation of caspase-2 and caspase-8 upstream of mitochondrial apoptosis [24]. GSK-3β is a key regulator of these processes [11,13], and it is of particular interest to know whether there is cross-talk between GSK-3β and caspase-2 before caspase-8 activation. In the present study, as summarized in Fig 6C, we first verified that ceramide-or etoposide-induced lysosomal destabilization is regulated by GSK-3β, caspase-2, and Mcl-1. GSK-3β acts upstream of caspase-2 and synergistically with caspase-2 to induce lysosomal-mitochondrial damage following Mcl-1 destabilization. The mechanisms that underlie GSK-3β-regulated caspase-2 activation and caspase-2-induced Mcl-1 destabilization followed by LMP induction remain unclear. We further demonstrated that lysosomal cathepsin D acts upstream of caspase-8 and Bid activation before mitochondrial damage. Our results showed that Mcl-1 is essential for lysosomal membrane stabilization by inactivating LMP and cathepsin D re-localization.
Ceramide has been recognized as a second messenger for various apoptotic stimuli [21,22]. We previously demonstrated that ceramide and etoposide might induce mitochondrial apoptosis [22,24,25]. In the present study, using inhibitor and siRNA strategies, we showed that lysosomal cathepsin D is the common effector in ceramide-or etoposide-induced apoptotic signaling. The pro-apoptotic role of cathepsin D remains controversial, although the dependence on cathepsin D has been shown with a variety of apoptotic stimuli [8]. However, in stress-induced fibroblast cell apoptosis, which includes anticancer agents, irradiation, CD95, TNF-α, and ceramide, a cathepsin D-independent pathway has also been reported [28]. Therefore, cathepsin D-induced apoptosis depends on the cell type and different stimuli.
In the present study, we first verified the involvement of the lysosomal-mitochondrial axis in ceramide-or etoposide-induced apoptosis. Our results show that cathepsin D acts downstream of caspase-2 but upstream of caspase-8 before mitochondrial damage. Studies have suggested a possible mechanism of cathepsin D-mediated caspase-8 activation [10,29]. Indeed, caspase-8 and a variety of pro-apoptotic proteins, such as Bid and Bax, are also substrates of cathepsin D [8,10,29]. Consistent with these findings, we showed that cathepsin D was activated upstream of caspase-8 and Bid. Therefore, we hypothesize that ceramide or etoposide causes caspase-2 activation and then induces LMP followed by cathepsin D-mediated caspase-8 and Bid activation before mitochondrial apoptosis. Indeed, our current studies demonstrate that GSK-3β mediates ER stress-induced lysosomal apoptosis in leukemia involving caspase-2-induced LMP and cathepsin B relocation, which result in caspase-8 and -3 activation [30]. Furthermore, anesthetic propofol treatment induces GSK-3β-mediated LMP followed by cathepsin B-regulated mitochondrial apoptosis [31]. Combining the results of this study, an essential role of GSK-3β is therefore showed in lysosomal/mitochondrial apoptosis.
Ceramide regulates LMP partly by targeting cathepsin D directly [32]. In general, stressinduced LMP results from the activation of acid sphingomyelinase and ceramide generation, which binds and activates cathepsin D by autocatalytic processing [32,33]. Holman et al.