Protein Kinase C δ Is Essential for Etoposide-induced Apoptosis in Salivary Gland Acinar Cells*

We have previously shown that parotid C5 salivary acinar cells undergo apoptosis in response to etoposide treatment as indicated by alterations in cell morphology, caspase-3 activation, DNA fragmentation, sustained activation of c-Jun N-terminal kinase, and inactivation of extracellular regulated kinases 1 and 2. Here we report that apoptosis results in the caspase-dependent cleavage of protein kinase C-δ (PKCδ) to a 40-kDa fragment, the appearance of which correlates with a 9-fold increase in PKCδ activity. To understand the function of activated PKCδ in apoptosis, we have used the PKCδ-specific inhibitor, rottlerin. Pretreatment of parotid C5 cells with rottlerin prior to the addition of etoposide blocks the appearance of the apoptotic morphology, the sustained activation of c-Jun N-terminal kinase, and inactivation of extracellular regulated kinases 1 and 2. Inhibition of PKCδ also partially inhibits caspase-3 activation and DNA fragmentation. Immunoblot analysis shows that the PKCδ cleavage product does not accumulate in parotid C5 cells treated with rottlerin and etoposide together, suggesting that the catalytic activity of PKCδ may be required for cleavage. PKCα and PKCβ1 activities also increase during etoposide-induced apoptosis. Inhibition of these two isoforms with Gö6976 slightly suppresses the apoptotic morphology, caspase-3 activation, and DNA fragmentation, but has no effect on the sustained activation of c-Jun N-terminal kinase or inactivation of extracellular regulated kinase 1 and 2. These data demonstrate that activation of PKCδ is an integral and essential part of the apoptotic program in parotid C5 cells and that specific activated isoforms of PKC may have distinct functions in cell death.

can result in reduced salivary gland function or xerostomia.
The critical genes in the apoptotic process have been defined genetically in Caenorhabditis elegans and biochemically in other species (9). These include the Bcl-2 family of proteins, a family of related regulatory proteins, which either promote or suppress apoptosis (10), and the caspases, cysteine proteases that are responsible for initiation and execution of the apoptotic signal (11). Other signaling molecules, including members of the mitogen-activated protein kinase family and protein kinase C (PKC) 1 family, have also been shown to be involved in the regulation of apoptosis (12)(13)(14)(15)(16)(17)(18).
In this report we have focused on the PKC family of enzymes as potential regulators of apoptosis in salivary acinar cells. The PKC family consists of 11 isoforms, whose expression varies between cell types (19,20). Individual isoforms exhibit varying substrate specificity, as well as differences in their subcellular localization and response to specific stimuli (20 -22), arguing that they have specialized roles in cell signaling. A variety of studies indicate that specific isoforms of PKC may be either pro-apoptotic or anti-apoptotic, depending on the stimulus and cell type (23)(24)(25). In support of an anti-apoptotic function, PKC inhibitors are potent inducers of apoptosis in many hematopoietic and neoplastic cells (26 -28), and treatment with phorbol 12-myristate 13-acetate to activate PKC antagonizes apoptosis induced by many agents (29 -31). Recently PKC␣ has been shown to phosphorylate Bcl-2 in vitro, and overexpression of PKC␣ results in increased Bcl-2 phosphorylation and suppression of apoptosis in human pre-B REH cells (32). The atypical PKC isoforms, PKC and PKC, have likewise been shown to protect against apoptosis in many cell types (33)(34)(35).
In support of a pro-apoptotic role for PKC, activation of PKC with phorbol 12-myristate 13-acetate, or overexpression of PKC␣, can induce apoptosis in prostatic carcinoma cells (36,37). Likewise, PKC␣ is activated following induction of apoptosis by genotoxic agents in HL-60 myeloid cells (38). Several laboratories have reported the proteolytic activation of PKC␦ to release a catalytically active fragment in cells induced to undergo apoptosis with ionizing radiation and DNA-damaging drugs (39 -42). Furthermore, in several cell types expression of the PKC␦ catalytic domain induces phenotypic changes indicative of apoptosis (38,39,43). A recent report also shows that cleavage and activation of PKC by caspase-3 occurs in U937 cells in response to agents that induce apoptosis (44).
The present studies were undertaken to ask if specific isoforms of PKC regulate apoptosis in salivary acinar cells in response to genotoxic agents. These studies demonstrate that PKC␦ is activated during etoposide-induced apoptosis in a cell line derived from parotid gland acinar cells and that the activity of this isoform is essential for complete apoptosis in these cells. PKC␣ and PKC␤1 are likewise activated following treatment of parotid acinar cells with etoposide; however, the contribution of these activated isoforms to the apoptotic process appears to be more modest, suggesting that specific isoforms of PKC may have distinct functions in cell death.
Immunoblotting-Adherent and floating cells were scraped into the culture media, collected by centrifugation (3,000 ϫ g for 10 min), washed once with phosphate-buffered saline, and resuspended in 1 ml of JNK lysis buffer (25 mM HEPES, pH 7.5, 20 mM ␤-glycerophosphate, 0.1 mM sodium orthovanadate, 0.1% Triton X-100, 0.3 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 10 mM NaF, and 4 g/ml each aprotinin and leupeptin). The lysate was allowed to sit on ice for 30 min and then clarified by spinning at 12,500 rpm for 5 min in a refrigerated Savant SRF13K microcentrifuge. Protein concentration was determined using a Bradford assay kit purchased from Bio-Rad. Cell lysates (25-50 g) were resolved on a 10% gel, transferred to an Immobilon membrane (Millipore), and immunoblotted with the desired antibody as described previously (46). Enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech) followed by autoradiography was used to detect the signal. Antibodies to all PKC isoforms were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). All anti-PKC antibodies recognize epitopes in the C-terminal portion of the protein. The anti-active ERK2 antibody, which cross-reacts with both phosphorylated ERK1 and ERK2, was obtained from Promega Biotechnology (Madison, WI). An anti-mitogen-activated protein kinase antibody, which cross-reacts with both ERK1 and ERK2, was obtained from Upstate Biotechnology (Lake Placid, NY).
Immunoprecipitation Kinase Assays-PKC␣, PKC␤1, and PKC␦ enzymatic activity was assayed using an immunoprecipitation kinase assay as follows. Cytosolic protein (0.25 or 0.5 mg), prepared as described for immunoblotting, was immunoprecipitated for 4 h at 4°C using 2 g of anti-PKC␣ (C-20), anti-PKC␤1 (C-16), or PKC␦ (C-17) antibody. The antigen-antibody complexes were collected by incubation with Sepharose-protein A (Sigma) for 1 h at 4°C, washed 3 times in JNK lysis buffer, 3 times in 2ϫ kinase buffer (40 mM Tris, pH 7.4, 20 mM MgCl 2 , 20 M ATP, and 2.5 mM CaCl 2 ), and resuspended in 20 l of 2ϫ kinase buffer. To prevent contamination with activated PKC␦, immunoprecipitation of PKC␣ and PKC␤1 was done using cell lysates that were pre-cleared by immunoprecipitation with anti-PKC␦ as described above. Twenty l of reaction buffer (0.4 mg of H1 histone (Sigma), 50 g/ml phosphatidylserine, 4.1 M dioleoylglycerol, and 5 Ci of [␥-32 P]ATP (3000 Ci/mM)) was added, and the samples were incubated for 10 min at 30°C. In some experiments phosphatidylserine and dioleoylglycerol was omitted from the reaction buffer. Reactions were terminated by the addition of 2ϫ SDS sample buffer, boiled, and the reaction products resolved on a 12.5% SDS-polyacrylamide gel. The extent of H1 histone phosphorylation was determined by autoradiography and in some experiments quantified using a PhosphorImager (Molecular Dynamics).
Assay for DNA Fragmentation-DNA fragmentation was assayed using a Cell Death Detection Assay kit from Roche Molecular Biochemicals. This assay detects the appearance of histone-associated low molecular weight DNA in the cytoplasm of cells and was performed in accordance with the manufacturer's recommendations.
Kinase Assay for JNK Activity-The GST-c-Jun-(1-79) expression vector was kindly provided by Dr. Lynn Heasley (University of Colorado Health Sciences Center, Denver, CO), and the fusion proteins were prepared as described (14). JNK activation was assayed using the GST-Jun kinase assay (47). To collect both adherent and floating cells, cells were scraped into the culture media, collected by centrifugation (3,000 ϫ g for 10 min), washed once with phosphate-buffered saline, and resuspended in 1 ml of JNK lysis buffer. The lysate was allowed to sit on ice for 30 min and then clarified by spinning at 12,500 rpm for 5 min in a refrigerated Savant SRF13K microcentrifuge. For the assay a 100-l volume of a 10% suspension of GST-c-Jun-(1-79) was added to 300 g of total cellular protein in a final volume of 1 ml and incubated for 2 h at 4°C. The beads were then washed three times with 20 mM HEPES, pH 7.7, 50 mM NaCl, 2.5 mM MgCl 2 , 0.1 mM EDTA, 0.05% Triton X-100. Forty l of 50 mM ␤-glycerophosphate, pH 7.6, 0.1 mM sodium orthovanadate, 10 mM MgCl 2 , and 20 M ATP containing 10 mCi [␥-32 P]ATP (5000 cpm/pmol in the final reaction) was added to the washed beads, and the reaction was incubated at 30°C for 20 min. The reactions were terminated by the addition of 10 l of 5ϫ SDS sample buffer, boiled, and the reaction products resolved on a 10% SDS-polyacrylamide gel. The position of GST-Jun was determined by staining the gel, and the extent of GST-Jun phosphorylation was determined by autoradiography.

Changes in the Expression of Specific PKC Isoforms
Occur during Etoposide-induced Apoptosis-Apoptosis occurs in a series of well defined steps that involve specific biochemical changes to the cell. We have previously examined the ability of etoposide to induce apoptosis in the parotid C5 cell salivary acinar cell line. 2 These cells were derived from the rat parotid gland and retain many characteristics displayed by parotid acinar cells in vivo (45). Apoptosis was demonstrated by the appearance of cytoplasmic blebbing and nuclear condensation, DNA fragmentation, and caspase-3 activation. 2 In addition, etoposide induced activation of JNK and suppressed accumulation of activated ERK1 and ERK2. 2 The current studies were undertaken to explore the contribution of the PKC family of enzymes to specific events in the apoptotic process. As an initial approach we asked if treatment of parotid C5 cells with etoposide results in changes in the abundance of specific PKC isoforms. Parotid C5 cells strongly express PKC␣, PKC␦, and PKC, whereas much weaker expression of PKC␤1 and PKC⑀ is detected. Expression of PKC␤11, PKC␥, PKC, and PKC is not detectable in this cell line. 3 As shown in Caspase-dependent cleavage and activation of specific PKC isoforms, as well as other signaling molecules, as been previ-ously reported (41,42,44,48). To determine if cleavage of PKC␦ is caspase-dependent, we utilized the cell-permeable, irreversible, caspase inhibitors Z-VAD.FMK and DEVD.FMK. Parotid C5 cells were pretreated with the inhibitors for 30 min prior to the addition of etoposide or 1-␤-D-arabinofuranosylcytosine (ara-C). As seen in Fig. 2, ara-C treatment results in accumulation of a PKC␦ cleavage product which appears to be identical in molecular weight to the product generated in response to etoposide. Inclusion of either caspase inhibitor blocks cleavage of PKC␦ in response to ara-C or etoposide, although Z-VAD.FMK is more potent. Z-VAD.FMK is thought to inhibit apoptosis at an early stage, prior to activation of caspases-2, -3, -6, or -7, which may explain its higher potency (49). Neither caspase inhibitor blocked accumulation of the 40-kDa PKC cleavage product, indicating that generation of this fragment is not caspase-dependent (data not shown). Pretreatment of parotid C5 cells with Z-VAD.FMK or DEVD.FMK likewise did not block the increase in PKC␣ protein accumulation in apoptotic parotid C5 cells (data not shown).
The Activity of PKC␣, PKC␤1, and PKC␦ Increase during Etoposide-induced Apoptosis-To determine if the increased abundance of PKC␣ and PKC␤1 protein, or the cleavage of PKC␦, results in an increase in PKC activity, immunoprecipitation kinase assays were performed on the cell lysates used in Fig. 1. As seen in Fig. 3A, the activity of PKC␦, as determined by its ability to phosphorylate histone H1, increases by 4 h of etoposide treatment, coincident with generation of the cleavage fragment. By 18 h of treatment PKC␦ activity is increased 9-fold as determined by quantification of 3 similar experiments. Notably, this increase in activity does not require the addition of lipid activators to the in vitro reaction (Fig. 3A, ϪLipid), suggesting that the C-terminal cleavage product generated during apoptosis represents the free catalytic domain of the molecule. Since accumulation of the PKC␦ cleavage product is caspase-dependent, we asked if pretreatment of parotid C5 cells with the caspase inhibitor Z-VAD.FMK could block the increase in PKC␦ activity observed. As seen in Fig. 3B, pretreatment with Z-VAD.FMK prior to the addition of etoposide completely blocks the etoposide-induced increase in lipid-independent PKC␦ activity. This indicates that cleavage of PKC␦ is required for the increase in lipid-independent kinase activity seen following etoposide-induced apoptosis.
To determine if the increase in PKC␣ and PKC␤1 protein seen during apoptosis correlates with an increase in kinase activity, immunoprecipitation kinase assays for these isoforms were performed on the cell lysates used in Fig. 1. As seen in Fig.  4A, the activity of PKC␣ increases dramatically following the addition of etoposide. Quantification of histone phosphorylation of 3 similar experiments indicates that PKC␣ activity increases 3-8-fold during etoposide-induced apoptosis. In some experiments the increase in PKC␣ activity appears to exceed the increase in PKC␣ protein expression (see Fig. 1A), suggesting that PKC␣ activity may be regulated by additional mechanisms. As seen in Fig. 4B, PKC␤1 activity increases 5-6-fold following the addition of etoposide. In the case of PKC␤1, however, the increase in kinase activity closely parallels the observed increased in protein abundance (compare Fig. 1 to Fig. 4). Unlike PKC␦, the increased activity of both PKC␣ and PKC␤1 during apoptosis is dependent on the addition of exogenous lipid to the kinase reaction (data not shown).
Inhibition of PKC Suppresses the Morphological Changes Associated with Etoposide-induced Apoptosis-The experiments described above demonstrate that specific isoforms of PKC are activated during etoposide-induced apoptosis. Although activation of PKC by apoptotic stimuli may be important, the fundamental question is whether the activation of these molecules is required for apoptosis to occur. To address the role of PKC␦, PKC␣, and PKC␤1 activation in apoptosis, we have examined the effect of isoform-specific inhibitors on biochemical and morphologic markers of apoptosis. To inhibit specifically PKC␦ in vivo, we have used rottlerin. The IC 50 of rottlerin for inhibition of PKC␦ is reported to be 3-6 M, whereas PKC␣, PKC␤, and PKC␥ are inhibited at significantly higher concentrations (40 M) (50). To inhibit PKC␣ and PKC␤1 we have used the calcium-dependent PKC isoform inhibitor Gö6976. The IC 50 of Gö6976 for inhibition of the PKC␣ and PKC␤1 is reported to be 2-6 nM, whereas no inhibition of PKC␦ is seen even at M concentrations (51).
Microscopically, apoptosis can be monitored by the condensation of the nucleus and cytoplasmic blebbing (52). To determine the effect of inhibition of PKC␦ on appearance of the apoptotic morphology, parotid C5 cells were preincubated with or without rottlerin for 30 min prior to the addition of 50 M etoposide for 18 h. Compared with untreated cells (Fig. 5A), Ͼ80% of parotid C5 cells treated with etoposide demonstrate morphologic changes consistent with apoptosis, and in fact many are detached from the culture dish by this time (Fig. 5B). Pretreatment of parotid C5 cells with 1 M rottlerin prior to the addition of etoposide, however, totally blocks appearance of the apoptotic morphology and cell death (Fig. 5D), whereas rottlerin alone has no effect on cell morphology (Fig. 5C). These results demonstrate that PKC␦ activity is essential for initiation of the morphological changes associated with apoptosis in response to DNA damage. To determine if rottlerin blocks etoposide-induced DNA damage, or the entry of damaged cells into apoptosis, parotid C5 cells were treated with etoposide and 1 M rottlerin for 24 h, washed to remove the drugs, and then placed in drug-free media. Under these conditions the cells undergo apoptosis and by 18 h resemble cells treated with etoposide alone (Fig. 5E). Thus, inhibition of PKC␦ with rottlerin appears to block the entry of cells damaged by etoposide into the apoptotic pathway.
To determine the contribution of PKC␣ and PKC␤1 to the morphologic changes seen during apoptosis, parotid C5 cells were pretreated with Gö6976 for 30 min prior to the addition of activity was assayed by an immunoprecipitation kinase assay using cell lysates that had been precleared with anti-PKC␦ as described under "Materials and Methods." PKC␣ activity was assayed using 0.25 mg of cytosolic protein, and 0.5 mg was used for PKC␤1. H1 histone phosphorylation was assayed in the presence of exogenous lipid. The reaction products were displayed on a 12.5% SDS-polyacrylamide gel. An autoradiogram of the dried gel is shown. This experiment was repeated 3 times with similar results. etoposide. As seen in Fig. 5G, preincubation with 100 nM Gö6976 inhibits the apoptotic morphology induced by etoposide, although less dramatically than pretreatment with rottlerin (compare Fig. 5, D-G). Treatment with Gö6976 alone appears to have no effect on cell morphology (Fig. 5F). Since 100 nM Gö6976 is about 20-fold above the reported IC 50 for PKC␣ and PKC␤1 in vitro, it is possible that other isoforms of PKC, or other kinases, may be inhibited at this dose and may account for the observed phenotype. Alternatively, Gö6976 may be a less effective inhibitor of PKC␣ and PKC␤1 in intact cells, and thus a higher dose may be required to see inhibition of PKC␣ and PKC␤1-dependent events. As discussed above, PKC␦ is not inhibited by Gö6976 even at micromolar concentrations (51) and therefore is unlikely to account for suppression of the apoptotic morphology.
Inhibition of PKC Suppresses DNA Fragmentation in Etoposide-treated Parotid C5 Cells-The breakdown of chromosomal DNA into 200-base pair nucleosomal fragments is characteristic of cells undergoing apoptosis. To determine if inhibition of PKC␦ and/or PKC␣/␤1 suppresses DNA fragmentation, parotid C5 cells were treated with 50 M etoposide, with or without the addition of rottlerin or Gö6976. DNA fragmentation was assayed using an enzyme-linked immunosorbent assay that quantitates cytoplasmic low molecular weight histone-associated DNA. As seen in Fig. 6, pretreatment with rottlerin suppresses etoposide-induced DNA fragmentation by about 30% at 1 M and by almost 50% at 10 M. Pretreatment of parotid C5 cells with Gö6976 prior to the addition of etoposide likewise inhibits DNA fragmentation (Fig. 6), although less effectively than pretreatment with rottlerin. At 10 nM Gö6976, etoposideinduced DNA fragmentation is inhibited by about 10%, whereas at 100 nM Gö6976 it is inhibited by about 40%. These data indicate that PKC␦ and PKC␣/␤1 activity are required for maximal fragmentation of DNA in etoposide-treated parotid C5 cells.
Inhibition of PKC Suppresses Caspase-3 Activation in Etoposide-treated Parotid Cells-We have previously shown that pretreatment with caspase inhibitor, Z-VAD.FMK, prior to the addition of etoposide inhibits biochemical and morphological indicators of apoptosis in parotid C5 cells. 2 Cleavage and activation of PKC␦ is likewise inhibited by Z-VAD.FMK (Figs. 2 and 3B), indicating that activation of at least some caspases must lie upstream of PKC␦ in the apoptotic pathway. To determine if activation of caspases also occurs downstream of PKC␦, we have asked if pretreatment with rottlerin can suppress caspase-3 activity in etoposide-treated parotid C5 cells. In the experiment shown in Fig. 7, parotid C5 cells were treated with etoposide for 18 h, and activation of caspase-3 under these conditions was set at 100%. Pretreatment with 1 M rottlerin prior to the addition of etoposide suppressed caspase-3 activity by Ͼ60%, whereas pretreatment with 2.5 M rottlerin suppressed activity by about 80%. No further suppression of caspase-3 activity was seen at higher concentrations of rottlerin. These results indicate that PKC␦ activity is required for maximal activation of caspase-3 following an apoptotic signal. Since cleavage of PKC␦ itself is inhibited by Z-VAD.FMK, caspase activation may occur both upstream and downstream of PKC␦ activation in the apoptotic pathway. Pretreatment of cells with the PKC␣ and PKC␤1 inhibitor, Gö6976, also suppressed etoposide-induced caspase-3 activity, although less effectively than pretreatment with rottlerin (Fig. 7). Ten nM Gö6976 suppressed caspase-3 activity by 10%, whereas caspase-3 activity was inhibited by 35% at 100 nM Gö6976. A small amount of caspase activation reproducibly occurred in cells treated with 100 nM Gö6976 alone, suggesting that this dose may be a weak inducer of apoptosis. Thus, while caspase-3 activation can occur under conditions where PKC␦, or PKC␣ and PKC␤1 activity is inhibited, maximal caspase-3 activation appears to require these activated isoforms.

Inhibition of PKC␦, but Not PKC␣/␤1, Prevents Sustained Activation of JNK and Suppression of ERK1/2 Activation-
Different members of the mitogen-activated protein kinase family have been demonstrated to be activated in response to stimulation with mitogenic or apoptotic agents. We have previously shown that etoposide induces activation of JNK in parotid C5 cells. 2 To ask if activation of JNK occurs upstream or downstream of PKC␦ activation, parotid C5 cells were pretreated with rottlerin, and JNK activation was assayed at various times after the addition of etoposide using the GST-Jun kinase assay (47). As seen in Fig. 8, lanes 1-7, in cells treated with etoposide alone, activation of JNK is evident by 2 h following the addition of etoposide and is sustained for at least 12 h. In cells pretreated with 5 M rottlerin prior to the addition of etoposide, however, some activation of JNK is apparent at 2 and 4 h (see Fig. 8, lanes 8 and 9), although the sustained activation of JNK appears to be nearly totally blocked (Fig. 8,  lanes 10 -13). No activation of JNK is seen in cells treated with rottlerin alone (Fig. 8, lanes 14 -19). These data suggest that whereas early activation of JNK may occur independent of PKC␦, sustained activation of JNK is likely to be PKC␦-dependent. To ask if JNK activation requires PKC␣ and/or PKC␤1 activity, the same experiment was repeated in cells pretreated for 30 min with 100 nM Gö6976. As seen in Fig. 8 cells with an inhibitor of PKC␣ and PKC␤1 does not block the activation of JNK by etoposide. As seen in Fig. 8, lane 32, a small amount of JNK activation is seen in parotid C5 cells treated with Gö6976 for 12 h, again suggesting that at this dose Gö6976 may be a weak inducer of apoptosis. These results suggest that, in contrast to PKC␦, sustained activation of the JNK pathway does not require PKC␣ or PKC␤1 activity.
We have previously shown that in addition to activating JNK, stimulation of parotid C5 cells with etoposide results in a decrease in the amount of activated ERK1 and ERK2, suggesting that these pathways are reciprocally regulated in apoptotic cells. 2 To ask if this decrease in activated ERK1 and ERK2 is blocked in etoposide-treated cells which are pretreated with rottlerin or Gö6976, activated ERK1 and ERK2 were assayed by immunoblotting using an anti-active ERK antibody which specifically recognizes the phosphorylated (active) forms of these kinases. As seen in Fig. 9A, activated ERK1 and ERK2 can be detected in untreated parotid C5 cells. Although the stimulus responsible for the activation of ERK1 and ERK2 in these cells is not clear, the tissue culture media that the cells are maintained in contains epidermal growth factor that is capable of activating ERKs. As seen in Fig. 9A, following the addition of etoposide, ERK1 and ERK2 activity is initially stimulated and then decreases by 8 h to a level at, or below, that seen in untreated cells. This decrease in activated ERK1 and ERK2 is coincident with the increase in JNK activity shown in Fig. 8, suggesting that these pathways are coordinately regulated. In parotid C5 cells pretreated with rottlerin before the addition of etoposide, however, the decrease in ERK1 and ERK2 activity at 6 -8 h appears to be blocked, with no decline in ERK1 or ERK2 activity apparent even after 12 h of etoposide treatment. Thus PKC␦ appears to be required for the decrease in ERK1 and ERK2 activity seen in etoposide-treated cells. In contrast, pretreatment of parotid C5 cells with Gö6976 FIG. 9. Inhibition of PKC␦, but not PKC␣/␤1, blocks the inactivation of ERK1 and ERK2 in etoposide-treated parotid C5 cells. Subconfluent cultures of parotid C5 cells were treated with 50 M etoposide for 0 -12 h, with or without the addition of 5 M rottlerin or 100 nM Gö6976, or with the inhibitors alone. The PKC inhibitors were added 30 min prior to the addition of etoposide. At the indicated times cell lysates were prepared for immunoblot analysis as described under "Materials and Methods." Panels A and C, 25 g of each cell lysate was resolved on a 10% polyacrylamide gel and immunoblotted with antiacitve ERK2 that cross-reacts with both phosphorylated ERK1 and ERK2. Panels B and D, the immunoblots shown in panels A and C were stripped and reprobed with an anti-ERK antibody that recognizes both ERK1 and ERK2. The positions of both ERK1 and ERK2 are noted on the right side of each panel. Time of stimulation in hours is shown at the top of each lane. had no effect on the decrease in ERK1 and ERK2 activity seen in etoposide-treated cells (Fig. 9C). This is consistent with the inability of Gö6976 to suppress JNK activation in etoposidetreated cells (Fig. 8). Interestingly, treatment of parotid cells with Gö6976 alone resulted in the initial activation and subsequent inactivation of ERK2, again suggesting that Gö6976 is a weak inducer of apoptosis in these cells (Fig. 9C). Uniform loading of the gels was demonstrated by reprobing the blots with an anti-ERK antibody that recognizes both ERK1 and ERK2 (Fig. 9, B and D). There was no change in the amount of either ERK1 or ERK2 over the period examined indicating that the changes in the amount of activated ERK2 did not result from a decrease in the amount of the ERK2 protein. These results suggest that, in contrast to PKC␦ which appears to be required for this event, activation of PKC␣ and/or PKC␤1 most likely occurs parallel to, or downstream of, the suppression of ERK1 activity.
Rottlerin Blocks Activation of PKC␦ in Etoposide-treated Parotid Cells-To determine if the ability of rottlerin to block apoptosis is due to a decrease in the generation and/or activity of the PKC␦ 40-kDa cleavage fragment, the expression of PKC␦ was assayed by immunoblot in etoposide-treated cells pretreated with rottlerin. As seen in Fig. 10A, pretreatment of parotid cells with rottlerin prior to the addition of etoposide inhibits accumulation of the 40-kDa PKC␦ cleavage product. In cells pretreated with 1 M rottlerin prior to the addition of etoposide, accumulation of the cleavage product is blocked significantly, whereas in cells pretreated with 10 M rottlerin the cleavage product is not detectable. In contrast, pretreatment of parotid C5 cells with Gö6976 only slightly suppresses accumulation of the PKC␦ cleavage product (Fig. 10C), an effect which may be secondary to its inhibition of caspase activity. The mechanism by which rottlerin blocks accumulation of the PKC␦ cleavage product is not clear. Since PKC␦ activity is required for caspase-3 activation (see Fig. 7), a decrease in the abundance of the cleavage fragment may be secondary to inhibition of caspase activation. Alternatively, since treatment of parotid cells with 1 or 10 M rottlerin alone results in a decrease in the abundance of the full-length PKC␦ protein (see Fig. 10A), inhibition of PKC␦ may decrease the stability of the PKC␦ protein. Finally, if the kinase activity of PKC␦ is required for its cleavage and activation, rottlerin would be expected to inhibit cleavage directly. To determine if rottlerin inhibits the lipidindependent activation of PKC␦ by etoposide, PKC␦ activity was assayed in parotid C5 cells pretreated with rottlerin prior to the addition of etoposide. As seen in Fig. 10, panel B, pretreatment of parotid C5 cells with rottlerin blocks the increase in PKC␦ activity seen in cells treated with etoposide alone.

DISCUSSION
The activation of specific signaling molecules, including some isoforms of PKC, has been demonstrated in apoptotic cells, suggesting that these activated molecules may function to regulate the apoptotic pathway (12,13,39,44). In this report we show that PKC␦ is activated in a caspase-dependent manner in parotid salivary acinar cells induced to undergo apoptosis by chemotherapeutic drugs. Inhibition of PKC␦ activity partially or totally blocks all parameters of apoptosis examined including appearance of the apoptotic morphology, caspase-3 activation, and DNA fragmentation. PKC␣ and PKC␤1 activities also increase during etoposide-induced apoptosis; however, inhibition of these isoforms results in a much more modest suppression of apoptosis. These data argues that activation of PKC␦ is an integral and essential part of the apoptotic program in parotid C5 cells and that specific activated isoforms of PKC may have distinct functions in cell death.
Proteolytic activation of PKC␦, in which the catalytic domain of the protein is cleaved from the regulatory domain, has been demonstrated in cells induced to undergo apoptosis with the topoisomerase inhibitors etoposide and camptothecin (38), ionizing radiation (41), ara-C and mitomycin C (42), and FAS ligand (43). We show that both cleavage of PKC␦ as well as its lipid-independent activation can be blocked by caspase inhibitors, which also block apoptosis. Furthermore, expression of the PKC␦ catalytic domain in several cell types induces phenotypic changes indicative of apoptosis (38,39,43). Likewise, data from our laboratory show that expression of the catalytic domain of PKC␦, but not a kinase-dead catalytic domain, can induce apoptosis in parotid C5 cells. 3 Although these results demonstrate that activation of PKC␦ by cleavage can effectively initiate the apoptotic program, the question of whether cleavage of PKC␦ is required for apoptosis is still unanswered. In fact, since in our studies PKC␦ activity is required for caspase activation (see Fig. 7), at least some functions of PKC␦ in the apoptotic pathway may be independent of its cleavage by caspase.
To determine the functional consequence of PKC␦ activation in parotid C5 cells we have utilized the PKC␦ specific inhibitor, rottlerin. Pretreatment of parotid C5 cells with rottlerin prior to the addition of etoposide effectively blocks most parameters of apoptosis. However, upon removal of both drugs apoptosis occurs, indicating that the cells have sustained DNA damage but are unable to carry out the apoptotic program. Although rottlerin is thought to inhibit PKC␦ at least in part by competing for ATP binding (50), our data suggest that, in addition to inhibiting activated PKC␦, rottlerin also prevents accumulation of the active PKC␦ cleavage product (Fig. 10). Although this may be due to an effect of rottlerin on protein stability, an alternative explanation is that the kinase activity of PKC␦ is required for its cleavage. Cleavage and activation of MEK kinase 1 (MEKK1) has also been demonstrated during apoptosis, and in this case the kinase activity of MEKK1 has been shown to be required for its cleavage (12).
Of the parameters examined, the morphologic changes associated with apoptosis appear to be the most sensitive to inhibition of PKC␦. The apoptotic morphology is essentially blocked at 1 M rottlerin, a concentration slightly below the reported IC 50 for inhibition of PKC␦ activity (Fig. 5) (50). As depicted in Fig. 11, this suggests that activation of PKC␦ occurs early in the apoptotic pathway and upstream of events that result in the morphologic changes associated with apoptosis. In fact, accumulation of the PKC␦ cleavage fragment and increased PKC␦ activity can be detected by 4 h following the addition of etoposide (Figs. 1 and 3), whereas the apoptotic morphology is not apparent until about 6 h. 2 The activation of caspase-3 in etoposide-treated cells also appears to be quite sensitive to rottlerin, with a maximum inhibition of 80% at 2.5 M rottlerin (Fig. 7). Although some of the structural changes seen during apoptosis, including cleavage of PAK2 and FAK, are thought to be caspase-3-dependent (48), our data suggest that caspase-3 activation alone is not sufficient to initiate these morphologic changes in parotid C5 cells.
DNA fragmentation is likewise suppressed under conditions where PKC␦ activity is inhibited (Fig. 6). Recently Bharti et al. (53) have reported that activated forms of PKC␦ are able to inactivate DNA protein kinase (DNA-PK), an enzyme that is essential for the repair of double-stranded DNA breaks. They suggest that activation of PKC␦ during apoptosis inhibits the ability of DNA-PK to repair DNA damage and thus promotes DNA fragmentation. This hypothesis is supported by our observation that inhibition of PKC␦ with rottlerin partially suppresses DNA fragmentation. Alternatively, since caspase-3 may be important for the inactivation of factors that suppress DNA fragmentation (54,55), inhibition of DNA fragmentation may be secondary to inhibition of caspase-3 activity.
Our studies indicate that both the expression and activity of PKC␣ and PKC␤1 increase following treatment of parotid C5 cells with etoposide. To explore the contribution of this activation to the apoptotic response, we have used the Ca 2ϩ -dependent PKC isoform inhibitor Gö6976. Although caspase-3 activity and DNA fragmentation are each suppressed by about 40% at 100 nM Gö6976, cellular rounding and blebbing are only slightly inhibited. Thus, although activation of PKC␣ and/or PKC␤1 may contribute to DNA fragmentation and caspase-3 activation, these isoforms probably are not required for the morphologic changes associated with apoptosis. One possibility is that activation of these isoforms amplifies specific events in the apoptotic pathway thus ensuring efficient cell demise.
Recent studies indicate that sustained activation of the JNK pathway correlates with the induction of apoptosis by a variety of agents including tumor necrosis factor-␣ (56), isothiocyanates (57), and TRAILl/apo2 (58). We have previously shown that treatment of parotid C5 cells with etoposide results in the caspase-dependent, sustained activation of JNK, as well as a decrease in the level of activated ERK. 2 Our current results demonstrate that sustained activation of the JNK pathway, as well as inactivation the ERK pathway, requires PKC␦ activity, as rottlerin is able to block both events in etoposide-treated cells. In contrast, inhibition of PKC␣ and PKC␤1 activity does not suppress the activation of JNK or inactivation of ERK. These findings support our previous observation that the JNK and ERK pathways are reciprocally regulated in parotid C5 cells during apoptosis. Taken together these results suggest that PKC␦ activation lies upstream of events that regulate JNK activation/ERK inactivation in apoptotic parotid C5 cells (see Fig. 11), whereas activation of PKC␣ and/or PKC␤1 occurs downstream, or independently, of the changes in these pathways.
The family of intracellular signaling molecules whose activity is regulated during apoptosis is increasing rapidly and includes a variety of protein kinases (48). Potential roles for these activated kinases include modulating the apoptotic responsiveness of the cell, as well as amplifying the apoptotic program. Our data demonstrate that one direct or indirect target of activated PKC␦ is caspase-3, since maximal activation of caspase-3 requires PKC␦ activity. This predicts the existence of a positive feedback loop whereby caspase activation of PKC␦ results in the activation of more caspase activity, which in turn contributes to the further activation of PKC␦. Preliminary evidence from our laboratory shows that expression of the PKC␦ catalytic fragment in parotid C5 cells is sufficient to induce caspase-3 activity. 3 In addition, since PKC␦ is essential for the apoptotic morphology, and contributes to DNA fragmentation, it is likely to regulate apoptosis through the activation or inactivation of additional effector molecules. DNA-PK has recently been identified as a substrate for activated PKC␦ in vitro (53). A more thorough understanding of the role of activated PKC␦ in apoptosis awaits the identification of additional substrates for this kinase in the apoptotic pathway. Our data indicate the following in etoposide-induced apoptosis: 1) caspase activation occurs both upstream and downstream of PKC␦ activation; 2) PKC␦ activation is required for the sustained activation of JNK and the inactivation of ERK. We propose that the inverse regulation of these pathways may be important for initiating or sustaining apoptosis. 3) PKC␦ activation is required for the changes in cell morphology that accompany apoptosis and contributes to DNA fragmentation. Although some of the effects of PKC␦ on apoptosis may be mediated via activation of caspase, there are probably additional targets such as DNA-PK. Furthermore, there are likely to be PKC␦independent mediators of apoptosis that remain to be defined. See text for further discussion.