Induction of Apoptosis Is Driven by Nuclear Retention of Protein Kinase Cδ*

Protein kinase Cδ (PKCδ) mediates apoptosis downstream of many apoptotic stimuli. Because of its ubiquitous expression, tight regulation of the proapoptotic function of PKCδ is critical for cell survival. Full-length PKCδ is found in all cells, whereas the catalytic fragment of PKCδ, generated by caspase cleavage, is only present in cells undergoing apoptosis. Here we show that full-length PKCδ transiently accumulates in the nucleus in response to etoposide and that nuclear translocation precedes caspase cleavage of PKCδ. Nuclear PKCδ is either cleaved by caspase 3, resulting in accumulation of the catalytic fragment in the nucleus, or rapidly exported by a Crm1-sensitive pathway, thereby assuring that sustained nuclear accumulation of PKCδ is coupled to caspase activation. Nuclear accumulation of PKCδ is necessary for caspase cleavage, as mutants of PKCδ that do not translocate to the nucleus are not cleaved. However, caspase cleavage of PKCδ per se is not required for apoptosis, as an uncleavable form of PKCδ induces apoptosis when retained in the nucleus by the addition of an SV-40 nuclear localization signal. Finally, we show that kinase negative full-length PKCδ does not translocate to the nucleus in apoptotic cells but instead inhibits apoptosis by blocking nuclear import of endogenous PKCδ. These studies demonstrate that generation of the PKCδ catalytic fragment is a critical step for commitment to apoptosis and that nuclear import and export of PKCδ plays a key role in regulating the survival/death pathway.

PKC␦ 2 is activated by numerous apoptotic stimuli and is emerging as an important modulator of the apoptotic response (1). PKC␦ activity is required for apoptosis induced by UV irradiation, genotoxins, taxol and brefeldin A (2, 3), phorbol ester (4), oxidative stress (5), and death receptors (6). Primary cells isolated from PKC␦-deficient mice are resistant to various apoptotic agents, and apoptosis is suppressed in the irradiated sal-ivary glands of PKC␦-deficient mice, indicating that PKC␦ is required for apoptosis both in vitro and in vivo (7,8).
Recent studies from our laboratory and others have begun to elucidate how PKC␦ is activated in apoptotic cells. PKC␦ translocates from the cytoplasm to the nucleus in response to various apoptotic agents including etoposide, ␥-irradiation, Fas ligand, and interleukin-2 deprivation in T cells (9 -12). We have identified a COOH-terminal bipartite nuclear localization signal (NLS) in PKC␦ and have shown that this sequence is required for both nuclear accumulation and apoptosis (11). PKC␦ is also a caspase substrate, and caspase 3-mediated cleavage generates a constitutively active catalytic fragment (␦CF) that is a potent inducer of apoptosis (13). When transiently expressed, the ␦CF kinase protein localizes to the nucleus in the absence of an apoptotic stimulus and triggers apoptosis; thus, caspase cleavage of PKC␦ may function to amplify the apoptotic response (11,14).
Although ␦CF is sufficient to induce apoptosis, studies in our laboratory have shown that expression of full-length PKC␦ (␦FL) can also induce apoptosis (11). However, it has been difficult to determine whether ␦FL and ␦CF play distinct roles in apoptotic cells as forced overexpression of ␦FL leads to generation of the ␦CF (11). In our current studies we have utilized subcellular localization mutants of PKC␦ to investigate the specific role of ␦FL during apoptosis and to determine the location of caspase cleavage of ␦FL in apoptotic cells. We show that in response to apoptotic agents, ␦FL rapidly and transiently accumulates in the nucleus and that apoptosis requires nuclear retention of PKC␦, which is facilitated by caspase cleavage of ␦FL in the nucleus to generate ␦CF. These results demonstrate that ␦FL functions as an apoptotic sensor by translocating into the nucleus in response to cell damage, whereas caspase cleavage of PKC␦ commits cells to the apoptotic program by facilitating nuclear retention of ␦CF.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-Culture of parC5 cells has been described previously (3). For immunofluorescence studies, cells were grown on 12-mm coverslips in 12 well polystyrene dishes. Subconfluent parC5 cells were transiently transfected 18 -24 h before etoposide treatment using a 3:1 or 6:1 lipid/DNA ratio of FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's protocol. Primary mouse parotid cells were prepared under sterile conditions as previously described (7). Primary cells grew to ϳ80% confluence in 5 days and were used at that time for experiments without further passage. Tissue culture reagents were obtained from Invitrogen or BIOSOURCE (Rockville, MD). Leptomycin B and etoposide were purchased from Sigma-Aldrich, and Z-Val-Ala-Asp(Omethyl)-CH 2 F (zVAD-FMK) is from Enzyme Systems Products, Livermore CA.
Construction of Enhanced Green Fluorescent Protein Plasmids and Site-directed Mutagenesis-pCAG-casp3 and pcasp3-D175A-GFP were a generous gift of Dr. S. Kamada (Kobe University, Japan) (15). The PKC␦ constructs of pGFP-PKC␦, in which the GFP tag was placed on the COOH terminus of the kinase, pGFP-PKC␦ K376R , pGFP-PKC␦ D327A , pGFP-NLM-FL8␦, and pEGFP (Clontech, Palo Alto CA) were previously described (11). pGFP-NTPKC␦, in which the GFP tag was place on the amino terminus of the kinase, was a generous gift from Dr. Stuart Yuspa (National Institutes of Health). To generate pGFP-NLSPKC␦, the SV40-NLS was fused to the NH 2 terminus of pGFP-PKC␦ and pGFP-PKC␦ D327A by PCR using the following primers: 5Ј-GCCTCGAGATGACACCCCCCA-AGAAGAAGCGAAAGGTAGAAGATCCCGAAGCACCC-TTCCTTCGC-3Ј, which contains an XhoI site, and 5Ј-GCTC-TAGAGTCGCGGCCGCTTTACTTGT-3Ј, which contains an XbaI site. XhoI and XbaI double restriction digests were performed on the PCR products along with DNA from the backbone pEGFP-N1 vector. pFLAG PKC␦ K376R was generated from pFLAG CE-PKC␦WT (a gift of Feng Chu, MD Anderson Cancer Center) by PCR site-directed mutagenesis.
Immunoprecipitation and Immunoblot Analysis-Immunoprecipitations and immunoblots were preformed as previously described (3). The mouse monoclonal antibody to GFP was obtained from Zymed Laboratories Inc. (C163, San Francisco, CA). Rabbit polyclonal antibodies were purchased from Santa Cruz Biotechnology (PKC␦ and lamin B) or Abcam (GFP).
Isolation of Cytosolic and Nuclear Fractions-For the experiments in Fig. 1, fractionation was preformed using a Biovision (Mountain View, CA) nuclear/cytosol fractionation kit according to the manufacturer's directions. For Fig. 3B, cell pellets were resuspended in 300 l of nuclei isolation buffer (20 mM HEPES-KOH, 100 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 250 mM sucrose, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, and 10 g/ml each of aprotinin, leupeptin, and pepstatin). The cells were incubated on ice 20 min and then lysed using a Dounce homogenizer (25ϫ). Efficient cell lysis was verified by trypan blue staining. The cell lysate was centrifuged (1000 ϫ g for 5 min), and the pellet (nuclei) and supernatant (cytosol) were collected. The nuclei were washed once in 200 l of nuclei isolation buffer and pelleted by centrifugation, and the resulting supernatant was added to the cytosolic extract. For all experiments, Triton X-100 was added to a final concentration of 1.0% to solubilize the proteins. Protein concentration was quantified using the DC Protein Assay kit (Bio-Rad). In some experiments leptomycin B (Sigma-Aldrich) (10 ng/ml) was added during the last hour of etoposide treatment.
Immunofluorescence Microscopy-To detect endogenous PKC␦, cells were washed with PBS and fixed in 2% paraformaldehyde/PBS for 15 min followed by permeabilization with 0.5% Triton-X, PBS for 5 min at room temperature. Cells were incubated for 1 h in 20 mg/ml bovine serum albumin, PBS before incubation with a rabbit polyclonal PKC␦ primary antibody (#C-17 Santa Cruz Biotechnology) for 1 h. Cells were washed five times with PBS and then incubated with a donkey antirabbit fluorescein isothiocyanate-conjugated secondary antibody (Jackson Immunoresearch Laboratories Inc., West Grove PA) for 1 h. Cells were again washed five times with PBS, and coverslips were mounted with o-phenylenediamine dihydrochloride mounting medium (Sigma). Mounted slides of cells transfected with GFP were prepared identically without the antibody staining. Cells were visualized, and images were collected using SlideBook software (Intelligent Imaging Innovations Inc., Denver, CO) on a Nikon Diaphot TMD microscope equipped for fluorescence with a xenon lamp and filter wheels (Sutter Instruments, Novato, CA), fluorescent filters (Chroma, Battleboro, VT), cooled CCD camera (Cooke, Tonawanda, NY), and a stepper motor (Intelligent Imaging Innovations Inc., Denver CO).
Terminal Deoxynucleotidyltransferase (TdT)-mediated dUTP Nick End Labeling (TUNEL) Analysis and Cell Counts-TUNEL analysis was performed using the in situ Cell Death Detection kit TMR Red (Roche Applied Science) according to the manufacturer's protocol. GFP-positive cells were visualized by immunofluorescent microscopy and counted using a 40ϫ objective. TUNEL-positive cells containing GFP were identified by colocalization with 4Ј,6-diamidino-2-phenylindole dihydrochloride hydrate and by morphology and were quantified as the percent of the total GFP-positive cells per field. Experiments quantifying the percent of cells exhibiting nuclear accumulation of GFP were performed by immunofluorescent microscopy using a 40ϫ objective.

PKC␦ Translocates to the Nucleus before Caspase Cleavage-
Our laboratory and others have previously shown that both ␦FL and ␦CF localize in the nucleus in response to etoposide (10,11). Nuclear translocation is dependent upon a functional NLS; however, nuclear localization is also suppressed by mutation of the caspase cleavage site of PKC␦, suggesting that caspase cleavage facilitates nuclear accumulation (11). To determine whether nuclear translocation of endogenous PKC␦ requires caspase activation, we pretreated parC5 cells with the pan-caspase inhibitor zVAD-FMK before the addition of etoposide. Fig. 1A shows that endogenous PKC␦ is found predominately in the perinuclear compartment of untreated parC5 cells and in the nucleus of etoposide-treated cells, consistent with our previous results (11). Pretreatment with zVAD-FMK largely inhibits nuclear accumulation of PKC␦, suggesting that ␦CF may be the primary form of PKC␦ found in the nucleus. To test this, parC5 cells were treated with etoposide, and the accumulation of ␦FL and ␦CF in nuclear fractions was analyzed by immunoblot. Surprisingly, Fig. 1B (left side) shows that ␦FL also accumulates in the nuclear fraction transiently between 30 min and 2 h of etoposide treatment. Pretreatment with zVAD-FMK does not alter the kinetics of nuclear accumulation of ␦FL. However, nuclear accumulation of ␦CF, which is apparent by 4 -8 h of etoposide treatment, was blocked by zVAD-FMK as expected. These data indicate that nuclear accumulation of ␦FL occurs before generation of ␦CF and independent of caspase activity. Fig. 1C shows the kinetics of the reciprocal distribution of ␦FL between the nucleus and cytoplasm. In contrast to ␦FL, ␦CF distributes between both the nucleus and cytoplasm. Quantification of nuclear ␦FL (Fig. 1D) shows that nuclear accumulation is maximal at 30 min to 2 h, after which it decreases to the basal level. Because the nuclear accumulation of ␦FL is transient and caspase-independent, these results suggest that ␦FL may be actively exported out of the nucleus. Although we have previously shown that nuclear accumulation of PKC␦ depends on a NLS to mediate nuclear import, the mechanism of how PKC␦ is transported out of the nucleus is not known (11). Earlier observations indicated that leptomycin B, a potent inhibitor of the CRM1 exportin pathway, does not alter the nuclear/cytoplasmic distribution of PKC␦ under normal growth conditions (11). To determine whether the exportin pathway facilitates translocation of ␦FL between the nucleus and the cytoplasm in etoposide-treated cells, parC5 cells were treated with leptomycin B (16). As shown in Fig. 1E, leptomycin B resulted in enhanced nuclear accumulation of ␦FL at 30 min, and this was sustained until at least 8 h as compared with the control. Therefore, the decrease in nuclear PKC␦ observed after its initial accumulation is due at least in part to CRM1-mediated nuclear export.
Caspase Cleavage of PKC␦ Occurs in the Nucleus of Apoptotic Cells-Although procaspase 3 is thought to be localized predominantly in the cytoplasm, recent studies by Kamada et al. (15) demonstrate nuclear translocation of active caspase 3 in HepG2 cells treated with ␣-Fas or etoposide. Our data shows that nuclear translocation of ␦FL precedes caspase 3 cleavage, suggesting the presence of caspase 3 in the nucleus. To determine whether caspase 3 translocates to the nucleus in etoposide-treated parC5 cells and if the kinetics of translocation correlate with nuclear accumulation of ␦FL or ␦CF, parC5 cells were transfected with pCAG-casp3 in which GFP is fused to the COOH terminus of caspase 3. Caspase 3 was found predominantly in the nucleus in about 35% of untreated parC5 cells ( Fig. 2A). In agreement with Kamada et al. (15), treatment with etoposide resulted in the progressive FIGURE 1. Nuclear accumulation of ␦FL during apoptosis is transient and caspase-independent. A, ParC5 cells were left untreated or pretreated with 50 M zVAD-FMK for 30 min followed by etoposide for 8 h. Cells were then fixed, permeabilized, and stained with an antibody directed to the COOH terminus of PKC␦, a fluorescein isothiocyanate-conjugated secondary antibody, and visualized by immunofluorescent microscopy. B, ParC5 cells were treated with zVAD-FMK as described above and then followed by treatment with etoposide for the indicated time. Nuclear fractions were prepared and immunoblotted with an antibody directed to the COOH terminus of PKC␦. The top blot shows nuclear translocation of full-length PKC␦ (␦FL), whereas the middle blot shows nuclear accumulation of the catalytic fragment of PKC␦ (␦CF). Note that the middle blot was exposed for a longer time than the upper blot to observe the appearance of ␦CF. The blot was stripped and reprobed with an antibody to lamin B. Note that the initial accumulation of ␦FL occurred 0.5 h later in cells treated with etoposide plus zVAD-FMK, but this finding was not consistent between experiments. C and D, ParC5 cells were treated with etoposide for the indicated times. Nuclear (N) and cytosolic (C) fractions were prepared and immunoblotted with an antibody directed to PKC␦. ␦FL in the nuclear fraction was quantified by densitometry and normalized to lamin B. Data in the graph are the average of two similar experiments. E, ParC5 cells were left untreated and treated with 50 M etoposide for the indicated times (left side). Samples on the right side were treated with 10 ng/ml leptomycin B (Lept B) during the last hour of etoposide treatment. Nuclear fractions were prepared and immunoblotted for PKC␦. The amount of nuclear ␦FL was quantitated by densitometry; data are expressed as -fold increase in protein relative to control. Each experiment shown is representative of two or more experiments.
accumulation of GFP-caspase 3 in the nucleus starting between 1 and 2 h, and by 8 h nearly 80% of the transfected cells had nuclear caspase 3 ( Fig. 2A). To determine whether active caspase 3 or procaspase 3 accumulates in the nucleus, we repeated the experiment shown in Fig. 2A using a mutated form of caspase 3 that cannot be cleaved and activated (pcasp3-D175A-GFP) (15). Fig. 2B shows that the active, but not the inactive form of caspase 3, translocates to the nucleus in etoposide-treated cells. A comparison of the kinetics of nuclear accumulation indicates that ␦FL accumulation in the nucleus precedes that of caspase 3 and that active caspase 3 accumulates before generation of the ␦CF.
The data shown above suggest that caspase cleavage of ␦FL likely occurs in the nucleus of apoptotic cells. Previous studies by Blass et al. (10) used antibodies that recognize distinct NH 2 and COOH epitopes in PKC␦ and showed by immunohistochemistry that both domains of PKC␦ were present in the nucleus. However, this approach does not distinguish ␦FL from ␦CF. To more definitively address if ␦FL is cleaved by caspase 3 in the nucleus, we cotransfected parC5 cells with plasmids that express a PKC␦ protein in which the GFP tag is located on either the COOH or NH 2 terminus. If caspase cleavage occurs in the nucleus, we predicted that both the regulatory domain fragment (␦NF) and ␦CF should be present in the nucleus. Fig.  3A is a schematic showing the sizes of the fragments generated upon cotransfection and subsequent caspase-mediated cleav-age by etoposide. The 70-kDa GFP-tagged COOH-terminal PKC␦ fragment is observed predominately in the nuclear fraction, with some in the cytosolic fraction at 4 and 8 h of etoposide treatment (Fig. 3B). In contrast, the 63-kDa GFP-tagged NH 2 -terminal PKC␦ fragment is only detected in the nuclear fraction. These results indicate that ␦NF and the ␦CF as well as ␦FL can be found in the nucleus and support our hypothesis that the nucleus is the site of caspase cleavage of PKC␦.
If caspase cleavage of PKC␦ occurs in the nucleus of apoptotic cells, we predicted that preventing nuclear localization of PKC␦ should prevent its caspase cleavage. For these experiments we took advantage of our previous finding that GFPtagged PKC␦ can be targeted to the cytoplasm by a mutation of the PKC␦ NLS (11). In the experiment shown in Fig. 3C, left, parC5 cells were transfected with pGFP, pGFP-PKC␦ (GFP-␦WT), or pGFP-NLMFL8␦ (GFP-␦NLMFL8), in which the PKC␦ NLS has been mutated (11). Expression of GFP-  ␦NLMFL8 is exclusively cytoplasmic accumulation in Ͼ99% of transfected parC5 cells (11). Treatment of the transfected cells with etoposide resulted in caspase cleavage of GFP-␦WT as indicated by the presence of ␦CF; however, GFP-␦NLMFL8 was not cleaved by caspase, indicating that nuclear translocation of ␦FL is a prerequisite for its caspase cleavage. Lack of cleavage of GFP-␦NLMFL8 is not due to inhibition of caspase activation, as endogenous PKC␦ was cleaved in pGFP-NLMFL8␦-transfected cells (Fig. 3D). To determine whether nuclear retention of PKC␦ is sufficient to cause caspase cleavage, we fused an SV40 NLS to the NH 2 terminus of PKC␦ (NLS ␦WT) to drive nuclear import. ParC5 cells were transfected with pGFP-PKC␦, pGFP-NLS PKC␦, or pGFP-NLS PKC␦ D327A , a control in which the SV40 NLS is fused to the NH 2 terminus of PKC␦ mutated at the caspase cleavage site. Fusion of the SV40 NLS to PKC␦ resulted in targeting of the protein to the nucleus in Ͼ99% of transfected cells (data not shown). As seen in Fig. 3E, transient expression of ␦WT resulted in the generation of a small amount of GFP-␦CF, consistent with its ability to induce apoptosis (3,17). However, retention of PKC␦ in the nucleus by the addition of an SV40 NLS resulted in a much higher basal level of GFP-␦CF. As expected, constitutive nuclear localization of the caspase cleavage mutant did not result in caspase cleavage. Thus, the nuclear targeting of PKC␦ is sufficient to drive caspase cleavage of PKC␦ in the absence of another apoptotic inducer.
Nuclear Retention of PKC␦ Is Required for Apoptosis-The data presented above demonstrate that nuclear retention of ␦FL is both necessary and sufficient for caspase cleavage, indicating a role for both ␦FL and ␦CF in genotoxin-induced apoptosis. However, it has been difficult to decipher specific functions for these two forms of PKC␦ because under conditions where caspase is activated, both ␦FL and ␦CF are present in the cell. Reconstitution of Pkc␦ Ϫ/Ϫ epithelial cells with wild type PKC␦ or PKC␦ mutants is a powerful model for exploring the function of different domains of this kinase. To address the contribution of caspase cleavage to the proapoptotic function of PKC␦, Pkc␦ Ϫ/Ϫ cells were reconstituted with wild type PKC␦ or an uncleavable form of PKC␦, ␦CM (GFP-PKC␦ D327A ). As we have previously reported, expression of ␦WT in Pkc␦ Ϫ/Ϫ cells completely reconstitutes etoposide-induced apoptosis (Fig. 4A) (7). In contrast, expression of ␦CM is not sufficient to reconstitute apoptosis, indicating that generation of ␦CF is necessary for the apoptotic response. In the experiment shown in Fig. 4B, we asked if expression of ␦CM could induce apoptosis in the context of endogenous PKC␦, since under these conditions ␦CF could be generated from endogenous PKC␦. As seen in Fig. 4B, although expression of ␦WT induces apoptosis, ␦CM is a very weak inducer of apoptosis (11% TUNEL-positive cells compared with 27% of cells transfected with ␦WT) and is not significantly different from expression of GFP alone (9% TUNEL positive). These data indicate that expression of a caspase cleavage-resistant mutant of PKC␦ is insufficient to induce apoptosis even when expressed in the context of endogenous PKC␦. We have previously reported that the ␦CM has reduced accumulation in the nucleus when compared with wild type PKC␦ and attributed this to a lack of ␦CF production (11). To determine whether nuclear retention of full-length ␦CM is sufficient for induction of apoptosis in the absence of etoposide, we gener-ated ␦CM fused to an SV40 NLS (GFP-NLS-␦CM). In contrast to ␦CM, NLS-␦CM is retained exclusively in the nucleus (data not shown). Shown in Fig. 4C, NLS-␦CM and GFP-␦WT, but not ␦CM, are equally efficient at inducing apoptosis. These results indicate that sustained retention of full-length PKC␦ in the nucleus can initiate the apoptotic program and that caspase cleavage of PKC␦ functions primarily to facilitate nuclear retention of the active kinase.
Kinase-negative Mutants of PKC␦ Define Multiple Points of Regulation-Our laboratory and others have shown that expression of kinase-negative full-length PKC␦ (␦KN) protects cells from apoptosis induced from a variety of agents including etoposide (2,10). Likewise, we have shown that a kinase-negative mutant of ␦CF (␦KN-CF) is a potent inhibitor of apoptosis and that inhibition of apoptosis by ␦KN-CF depends upon nuclear localization (11). Although the subcellular localization of ␦KN in etoposide-treated cells has not been explored, it might be expected to localize to the nucleus based on our previous findings (11). To address this question, parC5 cells were transfected with pGFP-PKC␦ (␦WT) or pGFP-PKC␦ K376R (␦KN) and treated with etoposide, and localization of GFP-PKC␦ was analyzed by confocal microscopy. Shown in Fig. 5A, ␦⌲⌵ localizes to the perinuclear region in untreated cells similar to that observed for both ␦WT and endogenous PKC␦ (11). Surprisingly, after etoposide treatment, ␦WT accumulates in the nucleus, whereas ␦KN remains localized to the perinuclear region (Fig. 5A). As expected, cells transfected with pGFP showed a diffuse distribution of the GFP protein throughout the cells both before and after etoposide treatment. Consistent with a requirement for nuclear localization, the data in Fig. 5B indicate that ␦KN is not cleaved by caspase in etoposide-treated cells as no ␦CF is detectable. However, targeting ␦KN to the nucleus with the addition of a SV-40 NLS allows ␦KN to be cleaved by caspase after etoposide treatment, indicating that lack of caspase cleavage is due to its retention in the cytoplasm (Fig. 5C).
Our data suggest that ␦KN and ␦KN-CF inhibit etoposide-induced apoptosis by different mechanisms. Although nuclear localization of ␦KN-CF is necessary to inhibit apoptosis, ␦KN inhibits apoptosis even though it is localized exclusively to the cytoplasm (2). To determine whether suppression of apoptosis by ␦KN and ␦KN-CF are additive, parC5 cells were transiently transfected with pGFP, pGFP-PKC␦-CF K376R (␦KN-CF), pGFP-PKC␦ K376R (␦KN), or pGFP-PKC␦-CF K376R plus pGFP-PKC␦ K376R , and after 18 h the cells were treated with etoposide. Expression of ␦KN suppressed DNA fragmentation in response to etoposide by ϳ60%, in agreement with previous reports from our laboratory (2). Expression of ␦KN-CF resulted in a slightly more robust protection from etoposide-induced apoptosis as seen by an almost 75% reduction in TUNEL-positive cells (Fig. 6A). However, co-transfection with both constructs together did not lead to enhanced inhibition when compared with ␦KN-CF alone, consistent with the possibility that these forms of PKC␦ inhibit distinct steps in the apoptotic pathway.
Based on our finding that ␦FL translocates into the nucleus before caspase activation, we predicted that ␦⌲⌵ may inhibit apoptosis by preventing translocation of PKC␦ into the nucleus. To test this, parC5 cells were transfected with pGFP-PKC␦ or cotransfected with pGFP-PKC␦ (␦WT) and pFLAG-PKC␦ K376R (␦KN). The FLAG-␦KN localized exclusively to the cytoplasm as previously demonstrated for GFP-␦KN (data not shown and Fig. 5A). As shown in Fig. 6B, ␦WT translocated to the nucleus in 46% of cells treated with etoposide. However, coexpression of ␦KN and ␦WT significantly suppressed nuclear translocation of ␦WT (28 and 18% when cotransfected with ␦KN1 and ␦KN2, respectively). These studies indicate that in contrast to ␦KN-CF, which inhibits apoptosis only when it is localized to the nucleus, expression of ␦KN suppresses apoptosis by inhibiting nuclear translocation of PKC␦. These data demonstrate that nuclear translocation of PKC␦ is essential for transduction of the apoptotic signal during genotoxin-induced stress.

DISCUSSION
Molecular mechanisms that regulate the proapoptotic function of PKC␦ are not well understood but are clearly important for cell survival. Furthermore, the respective contribution of PKC␦ and its constitutively activated caspase cleavage product, ␦CF, to apoptotic signaling have not been elucidated. Here we show that the proapoptotic function of PKC␦ is regulated by its nuclear retention. ␦FL accumulates rapidly but transiently in the nucleus in response to apoptotic signals. Transient nuclear accumulation of ␦FL is not sufficient to induce apoptosis; rather, caspase cleavage of PKC␦ facilitates the sustained nuclear accumulation of PKC␦ that is necessary for apoptosis. Targeting of a caspase cleavage mutant of ␦FL to the nucleus is sufficient to induce apoptosis, indicating that it is the sustained accumulation of PKC␦ in the nucleus that is essential for apoptotic signaling and not caspase cleavage per se. Hence, our studies demonstrate that nuclear retention of PKC␦ is a critical step for commitment to apoptosis and suggest that nuclear transport of PKC␦ regulates the survival/death pathway.
We show that apoptotic signaling in etoposide-treated cells is coordinately regulated by ␦FL and ␦CF, with the overall outcome being a sustained increase in nuclear PKC␦. We propose that ␦FL may act as an apoptotic sensor, transmitting cell damage signals to the nucleus. Although nuclear import of PKC␦ requires the COOH-terminal NLS, our current studies indicate that in the absence of an apoptotic signal PKC␦ is retained in the cytoplasm by a mechanism that is dependent upon the regulatory domain of the protein. Thus, in addition to the NLS, posttranslational modifications in the PKC␦ regulatory domain, such as tyrosine phosphorylation, may control nuclear accumulation of ␦FL. An early role for ␦FL in apoptotic signaling is supported by its rapid, caspase-independent accumulation in the nucleus. Although necessary, this step is not sufficient for inducing apoptosis because PKC␦ is rapidly exported. Moreover, ␦FL nuclear accumulation precedes activation of any of the known components of the apoptotic pathway in etoposide-treated parC5 cells, including cytochrome c release and activation of the initiator caspase, caspase 9 (2).
We show that the regulatory domain (␦NF), ␦CF, and active caspase 3 all accumulate in the nucleus during apoptosis, in agreement with previous studies (10,15). Nuclear accumulation of caspase 3 follows accumulation of ␦FL and is coincident with generation of the cleavage product of PKC␦. This sequential regulation may assure that sustained nuclear accumulation of nuclear PKC␦ occurs only when caspase 3 is activated. Studies from our laboratory and others also show that overexpressing ␦FL is sufficient to induce its caspase cleavage and apoptosis. This may be due to the high basal expression of caspase 3 in the nucleus that can cleave PKC␦ when is it overexpressed (Fig.  2), or alternatively, high levels of nuclear ␦FL may activate caspase 3 directly or indirectly. However, because nuclear retention of a caspase cleavage mutant of ␦FL is required for apoptosis, these results indicate that the initiating signal by which ␦FL activates the apoptotic pathway arises in the nucleus. In this regard our data suggest that rapid export of uncleaved ␦FL out of the nucleus is likely to be critical for cell survival. Although we were unable to find a leucine-rich nuclear export sequence in PKC␦, ␦FL could be exported during apoptosis through binding molecules that contain nuclear export sequences. For example, two nuclear export sequences containing proteins c-Abl and Stat1 have each been shown to bind PKC␦ during apoptosis (18 -20).
D'Costa et al. (21) have previously shown that a caspase cleavage-resistant mutant of ␦FL can act as a dominant negative to block UV-induced apoptosis. Likewise, various reports have shown that PKC␦-mediated apoptosis can be inhibited by overexpression of a kinase-negative form of PKC␦ (2,10,22,23). Previously, we showed that a mutation of the NLS within ␦KN-CF results in its retention in the cytoplasm and renders it unable to inhibit apoptosis (11). In this study we show that ␦KN, which is localized exclusively in the cytoplasm, inhibits apoptosis by blocking nuclear translocation of ␦FL. These combined data suggest that ␦KN and ␦KN-CF regulate distinct steps in transduction of the PKC␦ proapoptotic signal that are dependent upon where they are localized in the cell. Because of the differential localization of these dominant-inhibitory PKC␦ mutants, our studies suggest that care should be taken when using these or similar tools to study PKC␦-regulated signal transduction.
Our data suggest that under normal growth conditions, tight regulation of nuclear import and export of ␦FL is required for cell survival. Moreover, a recent study has shown that the B cell growth factor, BAFF, contributes to cell survival at least in part through sequestration of PKC␦ in the cytoplasm (24). We show that sustained nuclear retention occurs only under conditions where caspase is activated, resulting in removal of the regulatory domain and nuclear retention. Because both ␦FL and ␦CF can induce apoptosis under conditions of nuclear retention, ␦FL and ␦CF likely share the same nuclear targets. PKC␦ has been shown to have a multitude of proapoptotic nuclear targets including STAT1, Rad 9, topoisomerase II␣, DNA-dependent protein kinase, and lamin B (9,(25)(26)(27)(28)(29)(30). Moreover, DNA-dependent protein kinase and topoisomerase II␣ are both necessary for apoptosis by DNA-damaging agents associate with both ␦FL and ␦CF (29,31).
Although ␦CF is generated in the nucleus, our studies show that it can also be found in the cytoplasm of apoptotic cells, indicating that ␦CF also exits the nucleus. This observation may explain other published studies reporting that PKC␦ can localize to various cellular compartments in response to many stimuli (4,5,32,33). Because nuclear ␦CF is a potent inducer of caspase activation and apoptosis, cytoplasmic ␦CF may function to amplify apoptosis through direct or indirect modification of the cytoplasmic apoptotic machinery. Recently Sitailo et al. (34) have shown that ␦CF may directly regulate the apoptotic pathway by targeting the antiapoptotic protein, Mcl-1, for degradation. Taken together, our studies suggest that tight regulation of nuclear import and export of ␦FL is critical for cell survival and that caspase cleavage of ␦FL in the nucleus signals an irreversible commitment of cells to apoptosis.