Ceramide-induced Apoptosis by Translocation, Phosphorylation, and Activation of Protein Kinase Cδ in the Golgi Complex*

Protein kinase C (PKC), a Ca2+/phospholipid-dependent protein kinase, is known as a key enzyme in various cellular responses, including apoptosis. However, the functional role of PKC in apoptosis has not been clarified. In this study, we focused on the involvement of PKCδ in ceramide-induced apoptosis in HeLa cells and examined the importance of spatiotemporal activation of the specific PKC subtype in apoptotic events. Ceramide-induced apoptosis was inhibited by the PKCδ-specific inhibitor rottlerin and also was blocked by knockdown of endogenous PKCδ expression using small interfering RNA. Ceramide induced the translocation of PKCδ to the Golgi complex and the concomitant activation of PKCδ via phosphorylation of Tyr311 and Tyr332 in the hinge region of the enzyme. Unphosphorylatable PKCδ (mutants Y311F and Y332F) could translocate to the Golgi complex in response to ceramide, suggesting that tyrosine phosphorylation is not necessary for translocation. However, ceramide failed to activate PKCδ lacking the C1B domain, which did not translocate to the Golgi complex, but could be activated by tyrosine phosphorylation. These findings suggest that ceramide translocates PKCδ to the Golgi complex and that PKCδ is activated by tyrosine phosphorylation in the compartment. Furthermore, we utilized species-specific knockdown of PKCδ by small interfering RNA to study the significance of phosphorylation of Tyr311 and Tyr332 in PKCδ for ceramide-induced apoptosis and found that phosphorylation of Tyr311 and Tyr332 is indispensable for ceramide-induced apoptosis. We demonstrate here that the targeting mechanism of PKCδ, dual regulation of both its activation and translocation to the Golgi complex, is critical for the ceramide-induced apoptotic event.

The protein kinase C (PKC) 1 family is a group of phospholipiddependent serine/threonine protein kinases consisting of at least 10 subtypes that can be classified into three subgroups, classical, novel, and atypical (1)(2)(3). The classical PKCs (␣, ␤I, ␤II, and ␥), which have a Ca 2ϩ -binding region (C2 region) and two cysteine-rich regions, are activated by Ca 2ϩ , phosphatidylserine and diacylglycerol, or phorbol esters (4). The novel PKCs (␦, ⑀, , and ), which lack the C2 region, are activated by phosphatidylserine and diacylglycerol or phorbol esters without Ca 2ϩ (5). The atypical PKCs ( and /), which lack the C2 region and have only one cysteine-rich region, are dependent on phosphatidylserine, but are not affected by diacylglycerol, phorbol esters, or Ca 2ϩ (6).
There is increasing evidence showing that PKC is involved in apoptosis (7). Among multiple subtypes of PKC, PKC␦ has been implicated as a pro-apoptotic kinase, mostly by acting as a target of caspase-3 (8), whereas PKC␣ and PKC⑀ inhibit apoptosis by phosphorylating Bcl-2 or by increasing the expression of Bcl-2, respectively (9,10). Apoptotic stimuli such as DNAdamaging agents (11), Fas ligand (12), and etoposide (8,13,14) cause the cleavage of PKC␦ and the accumulation of the constitutively active fragment in the nuclei, and then the catalytic domain of PKC␦ is supposed to induce nuclear fragmentation and cell apoptosis (15). Recent studies have also demonstrated that the cleaved catalytic domain of PKC␦ acts as a sphingosinedependent kinase (16). In addition, it has been reported that PKC␦ is essential for the apoptosis of keratinocytes and LaNCaP cells in response to 12-phorbol 13-myristate acetate (17,18).
Ceramide generated by transient hydrolysis of sphingomyelin has recently emerged as an intracellular lipid mediator implicated in various cellular responses, including programmed cell death (19 -23). The regulation of PKC activity by ceramide has been reported, but the results are still controversial: ceramide has been shown to activate PKC␣ or to inhibit PKC␦ in renal mesangial cells in vitro (24). It has also been reported that ceramide induces the translocation of PKC␦ and PKC⑀ from the membrane to the cytosol (25), of PKC␣ from the cytosol to the membrane (26), or of PKC␦ from the cytosol to the mitochondria (27).
We have reported that ceramide-induced PKC␦-specific translocation to the Golgi complex is accompanied by PKC␦ activation via its tyrosine phosphorylation in HeLa cells (28), although the functional significance of the Golgi-targeted PKC␦ by ceramide is unknown. Furthermore, previous reports have demonstrated that various stimuli result in the tyrosine phosphorylation and activation of PKC␦: PKC␦ is tyrosinephosphorylated by 12-phorbol 13-myristate acetate (29), epidermal growth factor (30), platelet-derived growth factor (31), ligands for the IgE receptor (32), and H 2 O 2 (33,34). Tyrosine residues that are specifically phosphorylated by individual stimulation have been also determined. Tyrosines 52, 64, 155, and 187 of PKC␦ are phosphorylated in response to 12-phorbol 13-myristate acetate (29), platelet-derived growth factor (31), and etoposide (8), and H 2 O 2 treatment results in the phosphorylation of tyrosines 311, 332, and 512 (33). It has also been reported that tyrosines 311 and 332 are phosphorylated by Src family kinases (31,34). However, the activation of PKC␦ by tyrosine phosphorylation is not always accompanied by apoptosis (31,35). In this study, we have determined the tyrosine of PKC␦ that is phosphorylated by ceramide treatment and the significance of both the activation and translocation of PKC␦ in ceramide-induced apoptosis.
Construction of Plasmids Encoding the PKCs-FLAG epitope-tagged expression plasmids of rat PKC␦ were constructed as described (33,34). The constructs encoding green fluorescent protein (GFP)-conjugated rat PKC␦ were previously described (35). GFP was fused to the C terminus of PKC␦. The cDNA encoding the C1B domain (amino acids 231-280) deletion mutant of PKC␦ was generated by PCR using BS391 (rat PKC␦-GFP in pTB701) (35) as the template. The primers were synthesized with an AvaI site at both the 3Ј terminus (amino acid 230) and the 5Ј terminus (amino acid 281) to maintain the amino acids of the joint region. The PCR product were digested with AvaI/EcoRI or AvaI/ BamHI and then subcloned into the EcoRI/BglII sites of BS340 (GFP in pTB701) (35).
Expression of PKC␦ Protein in HeLa Cells-HeLa cells were purchased from the Riken Cell Bank (Tsukuba, Japan) and cultured in minimal essential medium containing 100 units/ml penicillin and 100 g/ml streptomycin with 10% fetal bovine serum. For lipofection, cDNAs (ϳ5.5 g) were transfected into 5 ϫ 10 6 HeLa cells using FuGENE 6 (Roche, Indianapolis, IN) according to the manufacturer's standard protocol.
Observation of PKC␦-GFP Translocation-HeLa cells expressing PKC␦-GFP fusion proteins were spread onto glass-bottomed culture dishes (Mattek Corp., Ashland, MA) and cultured for at least 16 h before observation. The culture medium was replaced with normal HEPES buffer composed of 135 mM NaCl, 5.4 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, and 10 mM glucose (pH 7.3). The fluorescence of GFP was monitored under a Zeiss LSM 510 confocal laser scanning fluorescence microscope at 488 nm excitation with a 505/550-nm bandpass barrier filter. All experiments were performed at 37°C.
Data Analysis of PKC␦-GFP Translocation-After treatment with C 2 -ceramide, the time course of translocation was recorded as a time series of 30 images at 1-min intervals for each experiment. Fluorescence in the Golgi complex, cytoplasm, or nucleoplasm was measured using Zeiss LSM 510 software. Because the PKC␦ distribution in the nucleoplasm scarcely changed, the fluorescence intensity in the nucleoplasm was used as spontaneous photobleaching of GFP fluorescence by scanning. The fluorescence intensity in each region is defined as the fluorescence intensity in the cytoplasm or Golgi complex divided by the fluorescence intensity in the nucleoplasm in a 5-10-m 2 region of interest. For each time point, the Golgi or cytoplasmic fluorescence was calculated in at least five different regions of interest. These values were averaged and plotted to generate a time course of translocation.
Co-detection of Tyrosine Phosphorylation and PKC␦-GFP Translocation by Ceramide-HeLa cells were fixed before or after ceramide treatment with a fixative containing 4% paraformaldehyde and 0.2% picric acid in 0.01 M phosphate-buffered saline (pH 7.4) for 30 min. The cells were then incubated with anti-phosphotyrosine monoclonal antibody (diluted 1:200) and with Alexa 546-labeled goat anti-mouse IgG (diluted 1:1000) for 30 min. The fluorescence of GFP and Alexa 546 was observed under the confocal laser scanning fluorescent microscope. Bar ϭ 20 m. B, inhibition of ceramide-induced apoptosis in HeLa cells by siRNA from endogenous PKC␦. After knockdown of endogenous PKC␦ using siRNA, HeLa cells were stimulated with 10 M C 2 -ceramide. Cell apoptosis induced by C 2 -ceramide was also determined after 24 h by bisbenzimide (Hoechst 33258) staining. C, requirement of PKC␦ for release of cytochrome c (Cyto c) from mitochondria. HeLa cells were transfected with siRNA from human PKC␦. After 2 days, the cells were treated with 10 M C 2 -ceramide. 18 h after treatment, the cells were fixed, and cytochrome c was visualized by immunofluorescence as described under "Experimental Procedures." Furthermore, the cells were stained with bisbenzimide to visualize the nuclei. Bars ϭ 10 m.
For immunoprecipitation, the supernatant of the total fraction was incubated with anti-FLAG monoclonal antibody or anti-GFP polyclonal antibody (diluted 1:100) for 2 h at 4°C and then with protein A-Sepharose for an additional 2 h. Samples were centrifuged at 2000 ϫ g for 5 min at 4°C, and pellets were washed three times with phosphatebuffered saline without Ca 2ϩ or Mg 2ϩ . Finally, the pellets were suspended in 30 l of phosphate-buffered saline without Ca 2ϩ or Mg 2ϩ and used for phosphorylation or immunoblot studies as described below.
Kinase Assay of PKC␦-PKC activity was assayed by measuring the incorporation of 32 P i from [␥-32 P]ATP into substrate as described (34,35). In brief, the kinase activities of FLAG-PKC␦ or PKC␦-GFP fusion proteins immunoprecipitated from HeLa cells after stimulation with C 2 -ceramide or H 2 O 2 (10 l of suspended pellet) were measured with calf thymus histone H1 as the substrate without any PKC activators such as phosphatidylserine or diolein. The reaction mixture contained 20 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 20 M ATP, 15-50 kBq of [␥-32 P]ATP, and 200 g/ml histone H1. The incubation was carried out for 10 min at 30°C.
RNA Preparation and Endogenous PKC␦ Knockdown-21-Nucleotide double-stranded RNAs of human PKC␦ were synthesized as described previously (38). Cells were assayed after 48 h of small interfer-

FIG. 2. Effect of caspase-3 on ceramide-induced apoptosis.
A, effect of Ac-DMQD-CHO on apoptosis in HeLa cells induced by C 2ceramide. After pretreatment with the caspase-3-specific inhibitor Ac-DMQD-CHO (100 M) for 1 h, HeLa cells were treated with 10 M C 2 -ceramide (C2-Cer). Cell apoptosis induced by C 2 -ceramide was also determined after 24 h by bisbenzimide staining. B, immunoblot analysis of active caspase-3 in HeLa cells. After treatment with 10 M C 2 -ceramide for 24 h or with 50 M etoposide for 24 h, total cell lysates from HeLa cells expressing wild-type PKC␦-GFP were subjected to SDS-PAGE and Western blot analysis. The membrane was probed with anti-active caspase-3 antibody.

FIG. 3. Tyrosine phosphorylation sites of PKC␦ in ceramidetreated HeLa cells and effects of Src kinase inhibitors on tyrosine phosphorylation.
A, tyrosine phosphorylation of wild-type or mutant PKC␦ in HeLa cells treated with ceramide. HeLa cells were transfected with FLAG-tagged wild-type or single-or double-point mutant (Y187F, Y311F, Y332F, Y311F/Y332F, and Y512F) PKC␦. After treatment with 10 M C 2 -ceramide (C2-Cer) for 20 min, the tyrosine phosphorylation of FLAG-PKC␦ immunoprecipitated by anti-FLAG monoclonal antibody was analyzed by immunoblotting (IB) with antiphosphotyrosine (anti-pTyr) antibody. B, immunoblot analysis of wildtype and single-and double-point mutant PKC␦-GFP fusion proteins in HeLa cells treated with ceramide using each phosphorylation sitespecific antibody. After treatment with 10 M C 2 -ceramide for 20 min, total cell lysates from HeLa cells expressing wild-type or single-or double-point mutant (Y311F, Y332F, and Y311F/Y332F) PKC␦-GFP were immunoprecipitated by anti-GFP antibody. Immunoblot analysis of these immunoprecipitated PKC␦-GFP fusion proteins was carried out with each phosphorylation site-specific antibody (anti-Tyr(P) 311 (anti-pY311) and anti-Tyr(P) 332 (anti-pY332)). C, effects of Src family tyrosine kinase inhibitors on the tyrosine phosphorylation of PKC␦-GFP induced by C 2 -ceramide. After pretreatment with 30 M PP2 for 15 min or with 1 M herbimycin A for 30 min, HeLa cells overexpressing PKC␦-GFP were treated with 10 M C 2 -ceramide for 20 min. GFPtagged PKC␦ was immunoprecipitated by anti-GFP antibody, and the amount of tyrosine-phosphorylated PKC␦-GFP was measured by immunoblotting with anti-phosphotyrosine (anti-PY) antibody.
ing RNA (siRNA) transfection. For expression of exogenous PKCs, cDNA was transfected after 24 h of siRNA transfection.
Measurement of Cell Apoptosis-After various treatments in serumfree minimal essential medium for 24 h, HeLa cells were collected and fixed with 4% paraformaldehyde and 0.2% picric acid for 15 min. The cells were then stained with 1 ml of bisbenzimide (16 g/ml) in phosphatebuffered saline, and the number of cells showing apoptotic chromatin changes in their nuclei was counted under a fluorescence microscope. Cells with two or more condensed chromatin fragments were considered to be apoptotic. For detection of cytochrome c release from mitochondria, the cells were incubated with anti-cytochrome c antibody and with Alexa 488-labeled goat anti-mouse IgG (Amersham Biosciences).

RESULTS
Role of PKC␦ in the Apoptotic Effect of Ceramide-We first investigated the morphological changes in HeLa cells in response to C 2 -ceramide, a membrane-permeable analog of ceramide. As shown in Fig. 1A, Ͼ80% of the ceramide-treated HeLa cells, but not the C 2 -dihydroceramide (an inactive form of C 2 -ceramide)-treated cells, showed an altered nuclear structure with condensed chromatin fragments upon bisbenzimide staining), indicating the specificity of the apoptotic action of C 2 -ceramide. Possible involvement of PKC␦ in ceramide-induced apoptosis was examined using the PKC␦-specific inhibitor rottlerin and knockdown of PKC␦ by siRNA. Rottlerin significantly inhibited the apoptotic effect of C 2 -ceramide, re-ducing the number of apoptotic cells to 23.5% (Fig. 1A) without affecting the basal level of cell apoptosis. Knockdown of endogenous PKC␦ in HeLa cells using siRNA from human PKC␦ (38) significantly inhibited the apoptotic response to C 2 -ceramide in HeLa cells (Fig. 1B). Furthermore, the release of cytochrome c from mitochondria (27) by C 2 -ceramide treatment was also blocked by knockdown of endogenous PKC␦ (Fig. 1C). These findings strongly suggest that activation of PKC␦ is indispensable for ceramide-induced apoptosis.
PKC␦ can be cleaved by a caspase-dependent process during apoptosis (8,39). We therefore examined the ability of the cell-permeable caspase-3-specific inhibitor Ac-DMQD-CHO (100 M) to block the apoptosis induced by C 2 -ceramide in HeLa cells. As shown in Fig. 2A, pretreatment of the cells for 1 h with Ac-DMQD-CHO failed to inhibit the apoptotic effect of C 2ceramide. In addition, we examined whether active caspase-3 is expressed in C 2 -ceramide-treated HeLa cells. Using Western blot analysis, we found that the active form of caspase-3 was expressed in the etoposide-treated cells, but not in the C 2ceramide-treated cells (Fig. 2B).
PKC␦ Tyrosine Phosphorylation Sites in Ceramide-treated Cells-To determine the phosphorylation sites responsible for ceramide-induced PKC␦ activation, we mutated four tyrosine residues (Tyr 187 , Tyr 311 , Tyr 332 , and Tyr 512 ) in PKC␦ to phenylalanine, as these tyrosine residues have been identified as the phosphorylation sites responsible for PKC␦ activation in response to various stimuli, such as etoposide, platelet-derived growth factor, and H 2 O 2 (8,31,33). The Y187F and Y512F mutants showed significant phosphorylation by ceramide (as much as wild-type PKC␦), whereas mutation of Tyr 311 or Tyr 332 evidently reduced the ceramide-induced phosphorylation, and phosphorylation was undetectable in the double mutant Y311F/Y332F (Fig. 3A). Antibodies against phosphopeptides corresponding to each phosphorylation site (Tyr 311 or Tyr 332 ) demonstrated that both Tyr 311 and Tyr 332 were phosphorylated after ceramide treatment (Fig. 3B). Ceramide-induced tyrosine phosphorylation of PKC␦ was blocked by herbimycin A and PP2, inhibitors of Src family protein-tyrosine kinases (Fig. 3C).
To study the correlation between phosphorylation of Tyr 311 and Tyr 332 and the kinase activity of PKC␦, we measured the kinase activities of immunoprecipitated PKC␦ and its mutants in HeLa cells with and without ceramide treatment. As shown in Fig. 4, C 2 -ceramide increased by 1.5-fold the kinase activities of wild-type PKC␦ and the Y187F and Y512F mutants, but the kinase activities of the Y311F, Y332F, and Y311F/Y332F mutants were not altered. As each single mutation of Tyr 311 or Tyr 332 abolished the ceramide-induced activation, simultaneous phosphorylation of Tyr 311 and Tyr 332 may be essential for PKC␦ activation.
Tyrosine Golgi complex (28). To determine whether the tyrosine phosphorylation of PKC␦ is required for translocation of PKC␦ to the Golgi complex, we investigated the ceramide-induced translocation of GFP-tagged PKC␦, Y311F, Y332F, and Y311F/ Y332F in HeLa cells. Similar to wild-type PKC␦, the unphosphorylatable mutants also translocated to the Golgi complex in response to ceramide (Fig. 5). We concluded that the tyrosine phosphorylation and activation of PKC␦ are not necessary for the translocation of PKC␦ by ceramide stimulation.
As tyrosine phosphorylation is not necessary for the translocation of PKC␦ to the Golgi complex, it is possible that PKC␦ is tyrosine-phosphorylated after its accumulation in the Golgi complex. To investigate whether PKC␦ in the Golgi complex or cytoplasm is tyrosine-phosphorylated after stimulation with C 2 -ceramide, we visualized the tyrosine phosphorylation in HeLa cells expressing PKC␦-GFP after C 2 -ceramide treatment for 20 min using anti-phosphotyrosine antibody. Immunoreactivity for phosphotyrosine (pTyr) accumulated in the perinuclear region and was also found on the plasma membrane as dots (Fig. 6A). pTyr in the perinuclear region co-localized with PKC␦-GFP, but the green fluorescence of PKC␦-GFP in the cytoplasm did not overlap with the red fluorescence of pTyr. This finding suggests that PKC␦ is tyrosine-phosphorylated in the perinuclear region. However, pTyr was also observed in the perinuclear region in ceramide-treated HeLa cells expressing Y311F/Y332F PKC␦-GFP. Because a pTyr immunoreaction was also seen in HeLa cells not expressing Y311F/Y332F PKC␦-GFP (Fig. 6A, arrows), it is possible that the tyrosine phosphorylation of endogenous PKC␦ was detected by anti-pTyr antibody. To distinguish between endogenous and exogenous PKC␦, we utilized the siRNA technique, which reduces the expression of human endogenous PKC␦, but does not affect exogenous rat PKC␦ (Fig. 6B). In HeLa cells with siRNA from human PKC␦, ceramide-induced accumulation of pTyr in the perinuclear region was detected only in cells expressing wildtype PKC␦-GFP, although pTyr-positive dots on the plasma membrane were seen in all cells. The absence of pTyr immunoreaction in the perinuclear region in Y311F/Y332F PKC␦-GFP-expressing cells suggests that PKC␦ accumulated in the Golgi complex by ceramide is phosphorylated at Tyr 311 and Tyr 332 .
But it was still unclear whether the tyrosine phosphorylation of PKC␦ occurred before or after translocation to the Golgi complex. It has been reported that H 2 O 2 treatment induces the tyrosine phosphorylation and activation of PKC␦ without affecting its localization (35). To determine where PKC␦ is phosphorylated, we constructed a PKC␦ mutant lacking the C1B domain (⌬C1B), which could be tyrosine-phosphorylated by H 2 O 2 in the cytoplasm, but failed to translocate to the Golgi complex upon ceramide treatment. As shown in Fig. 7A, C 2ceramide treatment did not translocate ⌬C1B. H 2 O 2 induced the tyrosine phosphorylation and kinase activation of ⌬C1B and wild-type PKC␦ to a similar extent; however, ⌬C1B was neither tyrosine-phosphorylated nor activated by C 2 -ceramide treatment (Fig. 7, B and C). These results show that, after ceramide treatment, PKC␦ accumulated in the Golgi complex via its C1B domain and then was tyrosine-phosphorylated and activated in the compartment.
Translocation of PKC␦ to the Golgi complex occurs upon association of the enzyme with the Golgi complex, but not upon its tight binding to the Golgi membrane (28). The counterbalance of association/dissociation of PKC␦ might be regulated by the tyrosine phosphorylation of PKC␦. We examined the effects of the protein-tyrosine phosphatase (PTP) inhibitor vanadate (200 M) on the activation and accumulation of PKC␦ by ceramide. As shown in Fig. 8A, inhibition of PTP increased the intensity of PKC␦ in the Golgi complex and decreased that in the cytoplasm. Furthermore, vanadate treatment increased the activity of PKC␦ by up to ϳ200% (Fig. 8B).
Targeting of PKC␦ Is Indispensable for Apoptotic Effects in HeLa Cells-To explore the functional interaction between apoptosis and PKC␦ targeting to the Golgi complex accompanied by its tyrosine phosphorylation-mediated activation, we investigated ceramide-induced apoptotic changes in HeLa cells overexpressing PKC␦ or its mutants. Bisbenzimide staining revealed that HeLa cells overexpressing wild-type PKC␦-GFP or Y311F/Y332F PKC␦-GFP exhibited a robust apoptotic response to C 2 -ceramide similar to that of untransfected control cells (Fig. 9). However, endogenous PKC␦ was tyrosine-phosphorylated in the Golgi complex, as demonstrated in Fig. 6A, and may cause apoptosis that is independent of overexpression of PKC␦-GFP. We examined the effects of rat wild-type or Y311F/Y332F PKC␦ on the apoptotic response to C 2 -ceramide in HeLa cells FIG. 6. Tyrosine phosphorylation of PKC␦ in the Golgi complex in ceramide-treated HeLa cells. A, simultaneous visualization of PKC␦-GFP and tyrosine phosphorylation by anti-phosphotyrosine antibody in ceramide-treated HeLa cells. After pretreatment with or without siRNA for knockdown of endogenous PKC␦, HeLa cells were transfected with wild-type PKC␦-GFP or Y311F/Y332F PKC␦-GFP. HeLa cells transfected with PKC␦-GFP or Y311F/Y332F PKC␦-GFP were fixed after treatment with 10 M C 2 -ceramide for 20 min. Cells were immunostained with anti-phosphotyrosine monoclonal antibody, and phosphotyrosine was visualized with Alexa 546-conjugated secondary antibody. The localization of PKC␦-GFP is shown in green (upper panels). Phosphotyrosine (pTyr) is shown in red (middle panels). In the merged images, co-localization of GFP and Alexa 546 signals is shown in yellow (lower panels). Arrows indicate cells that did not express exogenous PKC␦-GFP. Bars ϭ 10 m. B, knockdown of endogenous PKC␦ in HeLa cells using siRNA from human PKC␦. HeLa cells were transfected with siRNA from human PKC␦. 24 h after the siRNA treatment, wild-type PKC␦-GFP or Y311F/Y332F PKC␦-GFP was overexpressed in HeLa cells. 48 h after the siRNA treatment, total cell lysates (40 g) extracted from HeLa cells were immunoblotted with antibody against PKC␦. that were treated with siRNA for knockdown of human endogenous PKC␦. The inhibitory effect of siRNA from human PKC␦ on ceramide-induced apoptosis was restored by overexpression of rat PKC␦. In contrast, Y311F/Y332F PKC␦-GFP could not restore the apoptotic change induced by ceramide (Fig. 9). These results indicate that the phosphorylation of Tyr 311 and Tyr 332 in PKC␦ is necessary for the apoptotic effect of ceramide. We further studied whether the phosphorylation of Tyr 311 and Tyr 332 in PKC␦ is enough for apoptosis or whether the targeting (translocation and activation) of PKC␦ to the Golgi complex is additionally necessary for the apoptotic effect of ceramide. PKC␦-GFP was tyrosine-phosphorylated and activated in the cytoplasm without any translocation upon H 2 O 2 treatment (Fig. 10, A and B). Fig. 10C shows that H 2 O 2 -induced tyrosine phosphorylation and activation of PKC␦ in the cytoplasm did not induce cell apoptosis.

DISCUSSION
Using GFP-tagged PKC subtypes, the dynamic movement of PKC has been visualized in living cells, and the individual function of each PKC subtype in various signaling cascade has been studied (40,41). Translocation of PKC varies depending on PKC subtypes (41), and extracellular stimuli and PKC translocation to the specific intracellular compartment are necessary for the recognition and phosphorylation of their substrates in the compartment (PKC targeting) (42). As the spatiotemporally different translocation of PKC results in distinct HeLa cells were transfected with wild-type PKC␦-GFP or ⌬C1B PKC␦-GFP. After treatment with 10 M C 2 -ceramide for 20 min or with 1 mM H 2 O 2 for 20 min, the tyrosine phosphorylation of these PKC␦-GFP fusion proteins immunoprecipitated by anti-GFP polyclonal antibody was analyzed by immunoblotting (IB) using anti-phosphotyrosine (anti-pTyr) monoclonal antibody. C, effects of ceramide or H 2 O 2 on the kinase activity of wild-type PKC␦-GFP or ⌬C1B PKC␦-GFP assessed by in-cell kinase assay. PKC␦-GFP fusion proteins were immunoprecipitated from HeLa cells overexpressing wild-type PKC␦-GFP or ⌬C1B PKC␦-GFP after treatment with 10 M C 2 -ceramide or with 1 mM H 2 O 2 for 20 min. The kinase activities of the immunoprecipitated PKC␦-GFP fusion proteins were assayed by measuring the incorporation of 32 P i into histone H1 without any PKC activators. Data are expressed as a percentage of the control level (the kinase activity before stimulation). cellular responses (35), this strongly suggests that the targeting mechanisms of PKC subtypes determine their individual roles in cell signaling pathways. We have demonstrated that ceramide (generated by transient hydrolysis of sphingomyelin via receptor-mediated stimulation by extracellular ligand) induces translocation of PKC␦ from the cytoplasm to the Golgi complex and its activation mediated by tyrosine phosphorylation (28). In this study, we explored the importance of PKC␦ and its tyrosine phosphorylation in the induction of apoptosis in HeLa cells upon treatment with ceramide.
Activation of PKC␦ via tyrosine phosphorylation has been reported by several investigators. Tyr 187 , Tyr 311 , Tyr 332 , and Tyr 512 are candidates for the tyrosine residues phosphorylated by various stimuli (8,29,31,33). Substitution of these tyrosine residues with phenylalanine revealed that two tyrosine residues (Tyr 311 and Tyr 332 ) in the hinge region of PKC␦ are the tyrosine phosphorylation sites of PKC␦ in C 2 -ceramide-treated HeLa cells, whereas mutation of Tyr 187 or Tyr 512 did not alter the phosphorylation. It is noteworthy that phosphorylation of both Tyr 311 and Tyr 332 is necessary for the activation of PKC␦. The present finding is in good agreement with the report by Konishi et al. (34) that phosphorylation of Tyr 311 is a critical step in the generation of active PKC␦ in response to H 2 O 2 in vitro. However, the relation between the tyrosine phosphorylation of PKC␦ and its activation is still controversial, and it is possible that other tyrosine residues are phosphorylated at undetectable levels.
Ceramide has been shown to activate tyrosine kinases, including Src (43) and Lck (44 -46), and association of PKC␦ with Src family tyrosine kinases has been reported (30,31,34,(47)(48)(49)(50)(51)(52)(53). The tyrosine phosphorylation of PKC␦ by ceramide was inhibited by inhibitors of Src family protein-tyrosine kinases, suggesting that a certain tyrosine kinase in the Src family phosphorylates PKC␦ in response to ceramide. Because the existence of Src family kinases in the Golgi complex and the phosphorylation of specific substrates by the kinases have been reported previously (54 -56), it is possible that an unidentified tyrosine kinase such as Src phosphorylates PKC␦ in the Golgi complex in HeLa cells after ceramide treatment.
The present results using siRNA and the PKC␦ inhibitor show that activation of PKC␦ is required for ceramide-induced apoptosis. It has been thought that the apoptotic effect of PKC␦ is associated with the cleavage of PKC␦ in the hinge region, followed by the production of the constitutively active catalytic domain. Cleavage of the catalytic domain of PKC␦ by caspase-3 has been reported in cells treated with DNA-damaging agents (11), Fas ligand (12), and etoposide (8,13,14). However, no significant degradation of PKC␦ was observed in HeLa cells treated with ceramide by immunoblot analysis (28). In addition, the active form of caspase-3 was undetectable in ceramide-treated cell homogenates, and pretreatment with the caspase-3-specific inhibitor Ac-DMQD-CHO failed to inhibit the ceramide-induced apoptosis in HeLa cells (Fig. 2). These results suggest that the caspase-3 cascade appears not to be involved in this ceramide-induced apoptotic mechanism mediated by PKC␦ in HeLa cells, although other caspases such as caspase-9 may play some roles in this pathway. It is noteworthy that the caspase-3 cleavage site in PKC␦ is localized between the two tyrosine phosphorylation sites, Tyr 311 and Tyr 332 . It is possible that tyrosine phosphorylation around the caspase-3 cleavage may protect the cleavage, but activates PKC␦ without cleavage, although the molecular mechanism of the activation pathway is unknown. It has been demonstrated that the C1B domain is responsible for the translocation of PKC⑀ to the Golgi complex induced by arachidonic acid and ceramide (57). Deletion of the C1B domain of PKC⑀ abolishes its translocation to the Golgi complex. As PKC⑀ with a C1B domain derived from PKC␦ also shows ceramide-induced translocation to the Golgi complex (57), this suggests that PKC␦ lacking the C1B domain cannot translocate to the Golgi complex in response to ceramide. ⌬C1B is supposed to be activated by phosphorylation of Tyr 311 and Tyr 332 , which are conserved in the deletion mutant. Using the ⌬C1B deletion mutant, we have shown that, in addition to PKC␦ activation, translocation of PKC␦ to the Golgi complex is indispensable for the apoptotic effect of ceramide.
Previous photobleaching studies (28) demonstrated that ceramide modulates the counterbalance of association/dissociation of PKC␦ with the Golgi complex. Ceramide translocates PKC␦ by increasing its association with the Golgi complex in the counterbalance. It is important that inhibition of PTP increased both the activity and accumulation of PKC␦. This finding suggests that PKC␦ association with or dissociation from the Golgi complex is regulated by tyrosine phosphorylation. However, as accumulation of PKC␦ in the Golgi complex was independent of tyrosine phosphorylation (Fig. 5), this suggests that dissociation of PKC␦ from the Golgi complex is regulated by dephosphorylation of PKC␦ by PTP. The small increase in PKC␦ activity (50% increase) (Fig. 4) after ceramide treatment may be due to the fact that most of the PKC␦ in the cytoplasm was inactive and part of the PKC was activated by ceramide because both activation and inactivation of PKC␦ occurred only in the Golgi complex.
The species-specific knockdown of PKC␦ utilized in this study is a useful technique for analyzing the subtype-specific function of PKC. Overexpression of unphosphorylatable PKC␦ failed to show dominant-negative effects on ceramide-induced apoptosis (Fig. 9). This is perhaps due to the fact that a considerable amount of endogenous PKC␦ is present in HeLa cells (enough to induce apoptosis). To elucidate the significance of the tyrosine phosphorylation of PKC␦ for ceramideinduced apoptosis, it was necessary to reduce the amount of endogenous PKC␦. However, knockdown of human endogenous PKC␦ only revealed that endogenous PKC␦ is required for the induction of apoptosis. Knockdown of endogenous PKC␦ was compensated by rat exogenous PKC␦, which is resistant to siRNA from human PKC␦. This compensation demonstrated that Tyr 311 and Tyr 332 are tyrosine-phosphorylated in the Golgi complex after ceramide treatment and then induce apoptosis.
In this work, we studied the importance of the PKC␦ targeting mechanism (translocation and activation) in the induction of apoptosis in HeLa cells upon treatment with ceramide. We have demonstrated that ceramide induces the translocation of PKC␦ to the Golgi complex and its activation via phosphorylation of Tyr 311 and Tyr 332 by Src-like kinases and that this spatiotemporally regulated activation results in the induction of apoptosis. Activation of PKC␦ alone is insufficient for apoptotic events, and the targeting of PKC␦ to the Golgi complex is an essential step in ceramide-induced apoptosis. As described previously (28), the activation of receptors such as interferon-␥ and tumor necrosis factor-␣ receptors results in the translocation and activation of PKC␦, suggesting that apoptosis via PKC␦-specific targeting to the Golgi complex occurs under physiological conditions. Although the target molecule of PKC␦ in this mechanism is not clear, the results strongly suggest that the targeting mechanism of PKC␦ (activation via tyrosine phosphorylation and translocation to the Golgi complex) is indispensable for ceramide-induced apoptosis.