Activation of atypical protein kinase C zeta by caspase processing and degradation by the ubiquitin-proteasome system.

Atypical protein kinase C zeta (PKCzeta) is known to transduce signals that influence cell proliferation and survival. Here we show that recombinant human caspases can process PKCzeta at three sites in the hinge region between the regulatory and catalytic domains. Caspase-3, -6, -7, and -8 chiefly cleaved human PKCzeta at EETD downward arrowG, and caspase-3 and -7 also cleaved PKCzeta at DGMD downward arrowG and DSED downward arrowL, respectively. Processing of PKCzeta expressed in transfected cells occurred chiefly at EETD downward arrowG and DGMD downward arrowG and produced carboxyl-terminal polypeptides that contained the catalytic domain. Epitope-tagged PKCzeta that lacked the regulatory domain was catalytically active following expression in HeLa cells. Induction of apoptosis in HeLa cells by tumor necrosis factor alpha plus cycloheximide evoked the conversion of full-length epitope-tagged PKCzeta to two catalytic domain polypeptides and increased PKCzeta activity. A caspase inhibitor, zVAD-fmk, prevented epitope-tagged PKCzeta processing and activation following the induction of apoptosis. Induction of apoptosis in rat parotid C5 cells produced catalytic domain polypeptides of endogenous PKCzeta and increased PKCzeta activity. Caspase inhibitors prevented the increase in PKCzeta activity and production of the catalytic domain polypeptides. Treatment with lactacystin, a selective inhibitor of the proteasome, caused polyubiquitin-PKCzeta conjugates to accumulate in cells transfected with the catalytic domain or full-length PKCzeta, or with a PKCzeta mutant that was resistant to caspase processing. We conclude that caspases process PKCzeta to carboxyl-terminal fragments that are catalytically active and that are degraded by the ubiquitin-proteasome pathway.

The protein kinase C (PKC) 1 family consists of at least a dozen structurally related phospholipid-dependent serine/threonine protein kinases (1)(2)(3). Two members of the family, zeta () and iota/lambda (/), are atypical (aPKCs), because they lack a functional C1 domain and, therefore, are not activated by C1 ligands such as diacylglycerol and phorbol ester tumor promoters (1)(2)(3). aPKCs are widely expressed in mammalian tissues and cell types (2). aPKCs transduce a variety of extracellular stimuli and thereby modulate the proliferation, malignant transformation, and survival of mammalian cells (2). For example, studies with a kinase-dead, dominant-negative PKC mutant indicate that PKC is required for the mitogenic activation of mouse fibroblasts (4). PKC binds and phosphorylates transcription factor Sp1, which controls the expression of a variety of mammalian genes, including vascular permeability factor/vascular endothelial growth factor (5). By turning on another transcription factor, namely nuclear factor B (NF-B), aPKCs regulate the expression of genes that mediate inflammatory responses and cell survival (6 -10). Activation of NF-B opposes the induction of apoptosis, which suggests that aPKCs transduce one or more survival signals (8 -10). Consistent with this idea, Pongracz et al. (11) found that spontaneously apoptotic U937 cells had about 20% as much PKC as did exponentially growing cells. In contrast to the depletion of PKC protein, induction of apoptosis in NIH-3T3 cells by UV radiation inhibited the activity of aPKCs without changing the aPKC protein level (12). Murray and Fields (13) have shown that PKC/, but not PKC, protects human K562 leukemia cells from apoptosis. Although there is little understanding of how PKC influences apoptosis, PKC activity apparently promotes cell survival (8 -12).
A family of cysteinyl proteases called caspases (Ͼ13 genes), which are related to the ced-3 death gene of Caenorhabditis elegans, are the key players in a pervasive pathway of apoptosis in mammalian cells (14,15). Caspases are among the most specific of proteases with an unusual requirement for cleavage after aspartic acid. Caspases that mediate induction of apoptosis fall into two classes: initiator caspases (e.g. caspase-8, -9), which activate downstream effector caspases (e.g. caspase-3, -6, -7) that process a wide variety of cellular proteins (14,15). Generally processing of target proteins by caspases either activates proapoptotic functions or turns off survival pathways. Widmann et al. (16) have shown that caspase-dependent cleavage of Raf-1 and Akt-1 inhibited their kinase activity and may explain the inhibition of the extracellular signal-regulated and Akt pathways during the progression of apoptosis. Kufe and co-workers (17)(18)(19) found that ionizing radiation or etoposide, which damages DNA and induces apoptosis in U937 cells, produces CF␦, a 40-kDa carboxyl-terminal fragment of PKC␦ that is intrinsically active. Moreover, transfection of HeLa cells with CF␦, but not an inactive CF␦ (K378R) mutant, induced nuclear fragmentation and cell death (19). Analogous results were obtained with a carboxyl-terminal fragment of PKC (20). PKC was processed by caspase-3 (or a closely related caspase) in vitro and in vivo to CF, which induced apoptosis following expression in U937 cells (20). Recently Reyland et al. (21) reported evidence that PKC␦ activity is essential for etoposide-induced apoptosis in rat parotid C5 cells.
Here we identify three caspase cleavage sites of human PKC, which lie in the hinge segment between the regulatory and catalytic domains. Induction of apoptosis in nontransfected parotid C5 cells, or HeLa transfected with epitope-tagged PKC, produced 40-and 50-kDa carboxyl-terminal fragments of PKC and stimulated PKC immune complex kinase activity. Our demonstration of PKC activation following the induction of apoptosis appears to be unprecedented. Caspase processing appears to be an alternative mechanism to phosphorylation for activating the kinase function of PKC. A carboxyl-terminal segment of PKC that contained the catalytic domain was active and formed polyubiquitin conjugates following expression in mammalian cells. The conjugates accumulated in cells treated with Lacta, a selective inhibitor of the 26 S proteasome, which governs the abundance of many influential proteins (22,23). Our results indicate that the free catalytic domain of PKC that is produced by caspases is subject to degradation by the Ub-proteasome pathway, which is a novel mechanism for down-regulation of the kinase lacking the regulatory domain.

EXPERIMENTAL PROCEDURES
Cells and Transfection-Baby hamster kidney (BHK) and HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. The rat parotid salivary gland acinar cell line, parotid C5, was grown on primaria culture dishes (Falcon) in a 1:1 mixture of Dulbecco's modified Eagle's medium:F-12 medium supplemented with 2.5% fetal bovine serum, 5 g/ml insulin, 5 g/ml transferrin, 1.1 M hydrocortisone, 25 ng/ml epidermal growth factor (24). Transfections of BHK were done with 0.5 g of pcDNA3 vector containing lacZ as a negative control, or wild type or mutant PKC using 8 g of LipofectAMINE (Life Technologies, Inc.). Transfected BHK cells were selected with 0.5 mg/ml active G418. Cultures were maintained at 37°C in a humidified atmosphere containing 5% CO 2 and 95% air. HeLa cells were transfected with the pcDNA3.1/GS wild type and mutant PKC constructs using X-tremeGENE Q2 reagent as recommended by the manufacturer (Roche Molecular Biochemicals).
PKC cDNA Constructs-Human PKC cDNA (ATCC; accession number z15108), which was blunt-end-subcloned into the EcoRV site of pcDNA3 (Invitrogen), was kindly provided by Dr. G. Yancey Gillespie (University of Alabama at Birmingham). Site-directed mutagenesis was done with the QuikChange kit according to the instructions of the manufacturer (Stratagene). The following PAGE purified primer pairs (Life Technologies, Inc.) were used (substituted base underlined): D210A, forward, 5Ј-CT TCC GAG GAG ACA GCT GGA ATT GCT TAC;  reverse, 5Ј-GTA AGC AAT TCC AGC TGT CTC CTC GGA AG; D230A,  forward, 5Ј-GAC GAC TCG GAG GCC CTT AAG CCA GTT ATC G;  reverse, 5Ј-C GAT AAC TGG CTT AAG GGC CTC CGA GTC GTC;  D239, forward, 5Ј-CA GTT ATC GAT GGG ATG GCT GGA ATC AAA  ATC TC; reverse, 5Ј-GA GAT TTT GAT TCC AGC CAT CCC ATC GAT  AAC TG. A catalytic fragment of PKC (CAT) consisting of residues 238 -592 and having the D239A mutation was produced by PCR with the following primer pair: forward, 5Ј-ATG GCT GGA ATC AAA ATC TCT CAG; reverse, 5Ј-TCA CAC CGA CTC CTC GGT GGA CAG C. The PCR product was TA cloned into pCR2.1 (Invitrogen), excised with HindIII and SacII, and subcloned into the corresponding sites of pcDNA3.1Bϩ (Invitrogen). A kinase-deficient K281R mutant of PKC and CAT was produced with the QuikChange kit using the following primers: forward, 5Ј-C CAA ATT TAC GCC ATG AGA GTG GTG AAG AAA AGA GC; reverse, 5ЈGC TCT TTT CTT CAC CAC TCT CAT GGC GTA AAT TTG G.
Epitope-tagged wild type and mutant PKC-V5 constructs were prepared from a pcDNA3.1/GS plasmid containing the 1776-bp human PKC open reading frame (H-L14283) with carboxyl-terminal V5 epitope and hexahistidine tags (Invitrogen). A 1.4-kbp NotI/BamHI segment of the open reading frame was excised and replaced with the 1.4-kb fragment of the kinase-deficient or caspase-resistant mutants or wild type PKC constructs described above. A catalytic domain mutant with V5 and hexahistidine tags (CAT-V5) lacking amino acids 16 -239 was produced by PCR from wild type PKC in pcDNA3 with the following primers: 5Ј-TTT AAA GCG GCC GCG GAA TCA AAA TCT CTC; and 5Ј-GAA CCG GGG GAT CCG GAT G. The PCR product was agarose gel-purified and TA-cloned with pCR2.1 (Invitrogen). The 0.7-kb NotI/ BamHI fragment was excised from pCR2.1 and ligated into the corresponding site of the PKC pcDNA3.1/GS plasmid. Similarly, a kinasedeficient K281R CAT-V5 was prepared using the K281R PKC as the PCR template. All mutant PKC cDNA clones were subjected to automated DNA sequencing of both strands, which confirmed that the constructs were correct.

Caspase Treatment of [ 35 S]Met-labeled PKC and Purified
Recombinant PKC in Vitro-Purified recombinant human caspase-2, -3, -4, -6, -7, and -8 were prepared as described (25). [ 35 S]Met (10 Ci/20-l reaction)-labeled PKC was produced by in vitro transcription and translation with the T7 Quick TNT kit (Promega). Caspase treatments were for 30 min at 30°C in buffer (CB) containing 0.1 M Hepes adjusted to pH 7.5 with Tris, 5 mM DTT, 0.5 mM EDTA, 10% (v/v) glycerol, and 5 l of the TNT reaction. The caspase reaction was stopped by addition of SDS sample solution and a 5-min incubation in boiling water.
Caspase Activity-BHK cells were rinsed with ice-cold PBS and disrupted by Dounce homogenization in a buffer containing 20 mM Tris-HCl, pH 8.0, 1 mM NaEGTA, 1 mM DTT, and 10 g/ml each of leupeptin, aprotinin, and pepstatin. The lysate was clarified by centrifugation for 30 min at 4°C at 16,000 ϫ g. Protein concentration was measured by the Bradford method with ␥-globulin as a standard (Bio-Rad). The reaction was started by adding 0.1 mg of lysate to 2 ml of reaction buffer containing 20 mM Tris-HCl, pH 8.0, 2 mM MgCl 2 , and 50 M Ac-DEVD-amc. Fluorescence due to the production of free AMC was continuously recorded at 440 nm (380-nm excitation).
PKC Western Blot and Immunoprecipitation-Transfected cells were rinsed with ice-cold PBS and lysed with a buffer containing (in mM): 10 Tris-HCl, pH 7.4, 1% Triton X-100, 50 NaCl, 50 NaF, 30 sodium pyrophosphate, 5 NaEDTA, 5 NaEGTA, 1 phenylmethylsulfonyl fluoride, 0.1 sodium orthovanadate, and 10 g/ml each of leupeptin and aprotinin. The lysates were homogenized by passage through a 26gauge needle and centrifuged for 30 min at 16,000 ϫ g at 4°C. Protein concentration was measured by the BCA method (Pierce Chemical) with bovine serum albumin as a standard.
For the preparation of cytosol and membrane fractions from parotid C5 cells, they were rinsed with ice-cold PBS and disrupted by Dounce homogenization with C5 lysis buffer which contained (in mM): 20 Tris-HCl, pH 7.5, 0.5 EDTA, 0.5 EGTA, 25 g/ml each of aprotinin and leupeptin. Unbroken cells were removed by centrifugation for 5 min at 1000 ϫ g, and the supernatant was centrifuged for 30 min at 100,000 ϫ g at 4°C to obtain cytosol and particulate fractions. Membrane proteins were extracted from the particulate fraction with C5 lysis buffer containing 1% Triton X-100. Proteins (50 g) were fractionated by SDS-PAGE and subjected to Western blot analysis with the PKC antibody as described (21).
For Western blot analysis a protein sample (30 g) was mixed with SDS sample solution and fractionated by SDS-PAGE (10% gel). Proteins were electrophoretically transferred to a PVDF membrane (Millipore Corp.). The membrane was blocked for 1 h at room temperature with 5% (w/v) nonfat dry milk in Tris-buffered saline (TBS). TBS contained (per liter) 8 g of NaCl, 0.2 g of KCl, and 3 g of Tris base and was adjusted to pH 7.4 at room temperature, and 0.05% (w/v) Tween 20 in the case of TBS containing Tween 20. Membranes were incubated overnight at 4°C with 1 g of PKC antibody (C-20, Santa Cruz Biotechnologies) in 5 ml of TBS containing 1% nonfat dry milk. Membranes were rinsed and processed with HRP-conjugated goat anti-mouse IgG (Transduction Laboratories) and a chemiluminescent substrate as described (26). An HRP-conjugated anti-V5 antibody (Invitrogen) was used for Western analysis of immunoprecipitates of cells transfected with epitope-tagged PKC.
Immunoprecipitations were done with a polyclonal antibody (usually 3 g) to a carboxyl-terminal peptide of (amino acids 576 -592) of rat PKC (C-20, Santa Cruz Biotechnologies) or with 1 g of anti-V5 epitope monoclonal (Invitrogen). The antibody was incubated with the lysate for 3 h at 4°C. Protein A-agarose (20 l of a 50% slurry) was added during the last hour. Immune complexes were washed six times with ice-cold lysis buffer as described (26) and twice with kinase assay buffer as described below.
Immune Complex Kinase Activity-The kinase activity of immunoprecipitated V5-tagged PKC from HeLa cells was assayed in 25 l of buffer, which contained 25 mM HEPES-Tris, pH 7.4, 10 mM MgCl 2 , 20 mM ␤-glycerophosphate, 2 mM DTT, 0.1 mM Na 3 VO 4 , 5 g of MBP, 20 M ATP, and 5 Ci of [␥-32 P]ATP. After 30 min at 30°C the reaction was stopped by adding SDS sample solution and boiling for 5 min. Samples (10 l) were size-fractionated by SDS-PAGE (15% gel) and 32 P-labeled MBP was quantified autoradiographically after fixing and drying the gel. To estimate the relative amounts of the V5 epitope-tagged proteins in the immunoprecipitates, 10-l samples were fractionated by SDS-PAGE (10% gel). Proteins were transferred to a PVDF membrane and immunostained with an HRP-conjugated anti-V5 monoclonal antibody (Invitrogen). Statistical analysis was done by two-tailed Student's t test.
Anti-Ub Western Blot-The Ub monoclonal antibody 6C1 was prepared by immunization of BALB/c mice with bovine erythrocyte Ub, which was coupled to ovalbumin with glutaraldehyde (27). Ascites (10 l/5 ml of TBS containing 1% nonfat dry milk) containing the Ub antibody (IgG 2a class) was used for Western blotting as described previously (27). Pretreatment of the antibody with purified bovine erythrocytes Ub abolished the immunostaining of polyubiquitinylated proteins (27). Lacta was prepared as described previously (28) and used to block protein degradation by the Ub-proteasome pathway (29). The cells were lysed with 95°C SDS lysis buffer, which contained (in mM): and 10 Tris-HCl, pH 7.4, 2 EDTA, 2 EGTA, 5 N-ethylmaleimide, 1% (w/v) SDS, and 50 M ALLN. N-Ethylmaleimide inhibits deubiquitinylating enzymes, and ALLN blocks the proteasome. Protein concentration was measured by the BCA method (Pierce Chemical) with bovine serum albumin as a standard. For immunoprecipitation of PKC and Ub-PKC conjugates, a sample of the SDS lysate (usually 0.5 mg of protein) was diluted 10-fold with immunoprecipitation buffer (IB), which contained (in mM): 1 EDTA, 1 EGTA, 0.2 sodium orthovanadate, 30 sodium pyrophosphate, 50 NaF, 50 M ALLN, 1% Triton X-100, 0.5% Nonidet P-40, and 10 Tris-HCl, pH 7.4. Immunoprecipitation was done with 3 g of the C-20 antibody for 2 h, and 20 l of a 50% slurry of protein A-agarose was added after the first hour. Immunoprecipitates were rinsed twice with IB, twice with IB containing 0.5 M NaCl, twice with lower salt IB (no NaF, sodium pyrophosphate, or sodium orthovanadate), and twice with 10 mM Tris-HCl, pH 7.4. Proteins were extracted with SDS sample solution for 5 min in a boiling water bath, fractionated by SDS-PAGE (10% gel), and transferred to a nylon membrane for immunostaining with the Ub antibody, which was done essentially as described for Western blot analysis of PKC. with apparent masses of 50, 40, and 25 kDa, respectively. Caspase-6 and -8 produced F1 and F3 (Fig. 1).

Processing of [ 35 S]Met-labeled PKC by Purified Caspases-
To identify the sites of caspase cleavage, candidate Asp residues were mutated to Ala and the sensitivity of wild type and PKC mutants to processing by caspase was determined (Fig.  1). These studies show that caspase-3 processed [ 35 S]Metlabeled PKC chiefly at EETD 210 and DGMD 239 , whereas caspase-6 and -8 processed PKC chiefly at EETD 210 ( Fig. 1, B-D). Caspase-7 processed [ 35 S]Met-labeled PKC at EETD 210 and DSED 230 , but not at DGMD 239 (Fig. 1E). These results show that PKC is a substrate of both initiator and effector caspases in vitro and that the processing sites are caspasespecific. Intracellular processing of human PKC occurred chiefly at EETD 210 and DGMD 239 as shown by transfection experiments with wild type PKC, and the mutants with alanine substituted for aspartate at the caspase cleavage sites as discussed below (see Fig. 4).
Caspase Processing and Increased Epitope-tagged PKC Activity following the Induction of Apoptosis by TNF␣ and CHX-To determine the effect of caspase processing on PKC kinase function, HeLa cells were transfected with PKC-V5, which has a carboxyl-terminal V5 epitope tag, and apoptosis was induced by treatment with TNF␣ plus CHX. Induction of apoptosis in HeLa and other cell types requires the combination of TNF␣ plus CHX (30,31). The treatment markedly increased the accumulation of carboxyl-terminal fragments (F1 and F2) of PKC-V5 and decreased the steady-state level of full-length PKC-V5, as determined by Western analysis of anti-V5 immunoprecipitates with an HRP-conjugated V5 antibody ( Fig. 2A). Treatment with the general caspase inhibitor, zVAD-fmk, prevented the depletion of full-length PKC-V5 and accumulation of F1 and F2 (Fig. 2A).
Additionally, we assayed the kinase activity of the anti-V5 immune complexes. Treatment with TNF␣ plus CHX markedly increased immune complex kinase activity (Fig. 2B). Interestingly, treatment with zVAD-fmk, which prevented the conversion of full-length PKC-V5 to the C-terminal fragments (F1 and F2), abolished the increase in kinase activity produced by TNF␣ plus CHX (Fig. 2B). This result suggests that the conversion of full-length PKC-V5 to one or both of the carboxylterminal fragments was responsible for the increase in kinase activity. Also note that the zVAD-fmk treatment decreased F2 and kinase activity of the immune complex from the cells that were not treated with TNF␣ plus CHX (Fig. 2). This result suggests that there was a basal rate of F2 production and disappearance in the untreated cells. Transfection of the cells with a kinase-deficient K281R PKC-V5 mutant produced no detectable kinase activity in untreated cells as expected (Fig.  2B). The kinase-deficient mutant was expressed to a similar extent as wild type PKC,-V5, and the TNF␣ plus CHX treatment depleted full-length K281R PKC-V5 similarly to wild type PKC-V5, although there was much less accumulation of the F1 and F2 fragments of the kinase-deficient mutant ( Fig.  2A). One possible explanation of this result is that the K281R mutant of F1 and F2 is less stable following the TNF␣ plus CHX treatment than wild type F1 and F2. In agreement with this explanation, the TNF␣ plus CHX treatment decreased K281R CAT-V5 more than wild type CAT ( Fig. 2A).
To confirm that the catalytic domain of PKC-V5 was active, HeLa cells were transfected with CAT-V5 or K281R CAT-V5, which are mutants that lack the regulatory domain. As expected, CAT-V5 was active following immunoprecipitation from transfected cells, but K281R CAT-V5 had no detectable activity (Fig. 2B). Note that the amounts of the immunoprecipitated CAT-V5, K281R CAT-V5, and full-length PKC-V5 polypeptides were similar ( Fig. 2A). CAT-V5 had 68 Ϯ 20% (n ϭ 3, p ϭ 0.04) more kinase activity than full-length PKC-V5 based on the ratios of the 32 P-MBP to the immunostained protein bands. Treatment with TNF␣ plus CHX decreased the steady-state level of CAT-V5 and its kinase activity (Fig. 2). Although the mechanism by which treatment with TNF␣ plus CHX induced the disappearance of the catalytic domain of PKC is unknown, it may involve the Ub-proteasome, because the results presented below show that the catalytic domain forms polyubiquitin conjugates and accumulates following blockade of the proteasome.
Production of Catalytic Domain Fragments of Endogenous PKC Activation of PKC Activity in Parotid C5 Cells following the Induction of Apoptosis-To determine if caspase activation evoked the processing of endogenous PKC, rat parotid C5 cells were treated with etoposide for 8 -24 h and PKC was analyzed by Western blot. Treatment with etoposide increased the levels of 40-and 50-kDa carboxyl-terminal fragments of PKC, with the amount of the smaller F2 fragment being greater than F1 (Fig. 3, A and C). The carboxyl-terminal PKC fragments were predominantly in the cytosol fraction, which suggests that they are not strongly membrane-associated (Fig. 3A). Although the antibody used for Western blot analysis cross-reacts with the / isoform of PKC(13), parotid C5 cells lacked detectable PKC/ as determined by Western blot with a PKC/-specific antibody. 2 The carboxyl-terminal fragments in etoposidetreated parotid C5 cells are probably produced by caspase processing of rat PKC at EETD 210 and DGVDG 239 as recently demonstrated in HeLa cells transfected with Myc-tagged PKC (32).
Interestingly, treatment of rat parotid cells with etoposide, which induces apoptosis in these cells (21), increased PKC immune complex kinase activity (Fig. 3B)  (18 h) increased PKC immune complex kinase activity 3.8 Ϯ 1.1 (n ϭ 4, p ϭ 0.04). A general caspase inhibitor, zVAD-fmk (14), abolished accumulation of the carboxyl-terminal PKC fragments in etoposide-treated parotid C5 cells and prevented the increase in the PKC immune complex kinase activity (Fig.  3, B and C). Following the treatment with 25 or 50 M zVADfmk and etoposide, PKC immune complex kinase activity (relative to untreated cells) was not significantly different from that of untreated cells. At 25 M concentration, zVAD-fmk had no effect on PKC immune complex kinase activity, but decreased etoposide-stimulated kinase activity by 86% (n ϭ 3). A caspase-3 selective inhibitor, zDEVD-fmk, reduced PKC immune complex kinase activity of etoposide-treated cells (Fig.  3B). zDEVD-fmk also inhibited the production of the carboxylterminal PKC fragments in etoposide-treated parotid cells, but zVAD-fmk was more potent than zDEVD-fmk in prevent-ing the increase in PKC immune complex kinase activity (Fig. 3B). 2 Accumulation of PolyUb Conjugates of the Catalytic Domain or Full-length PKC following Blockade of the Proteasome-Previously we observed the formation of Ub conjugates of PKC␣ in vitro and intracellularly (26,33). The ␦ and ⑀ isoforms of PKC also appear to be degraded by the Ub-proteasome system (26,34). To determine whether the catalytic domain or full-length PKC formed Ub conjugates, BHK cells were transfected with PKC, CAT, or with lacZ as a negative control. Some cells were treated for 18 h with Lacta, a specific inhibitor of the 26 S proteasome (29), to cause polyubiquitinylated proteins to accumulate. Proteins were immunoprecipitated with an antibody that recognizes the extreme carboxyl-terminal segment of PKC, fractionated by SDS-PAGE, and subjected to Western blot analysis with a monoclonal antibody to Ub (Fig. 4,  top left). Lacta treatment caused the accumulation of a smear of proteins (Ͼ150 kDa) that were immunoprecipitated by the anti-PKC antibody and immunostained by the anti-Ub antibody (Fig. 4, top left). Importantly, the Ͼ150-kDA smear was absent from cells transfected with lacZ instead of PKC or CAT (Fig. 4). These findings show that transfection of BHK cells with PKC or CAT caused Ub conjugates of PKC or carboxyl-terminal fragments of PKC to accumulate following the treatment of the cells with Lacta. No Ub-PKC conjugates were detected in untreated cells (Fig. 4) as expected if the conjugates were rapidly degraded by the proteasome in the absence of Lacta. There was no significant increase in fulllength PKC in BHK cells transfected with PKC compared with those transfected with lacZ (Fig. 4, bottom left, and see Fig. 5C). Lacta treatment caused the accumulation of fulllength PKC and the F1 and F2 fragments (Fig. 4, bottom left,  and Fig. 5C). Additionally, CAT was only detected in Lactatreated cells (Fig. 4, bottom left). Note that PKC comigrated with an endogenous protein band, which is probably the / isoform of PKC. PKC/ has the same electrophoretic mobility as PKC and is recognized by the carboxyl-terminal PKC antibody (13). Moreover, Western blot analysis with antibodies specific for either PKC/ or PKC (Santa Cruz Biotechnologies) showed that BHK cells had endogenous PKC/ and lacked detectable PKC. 3 Lacta treatment increased the steady-state level of ectopic full-length PKC and caused F1 and F2 to accumulate, but had no effect on the level of the endogenous PKC/ (Fig. 4, bottom right). Lacta treatment produced no detectable F1 or F2 in the lacZ-transfected BHK cells (Fig. 4, bottom left), which agrees with the recent reports that PKC/ is not processed by caspases in vitro or intracellularly following induction of apoptosis (32,35).
To determine whether caspase processing was required for the accumulation of Ub conjugates of PKC, BHK cells were transfected with PKC mutants with Ala substitutions at the caspase-processing sites (Fig. 4, right panels). Following Lacta treatment, Ub-PKC conjugates accumulated in cells transfected with wild type PKC or the Ala substitution mutants, including the double mutant, which was not cleaved by caspases (Fig. 4, top right). Ectopic expression of the double mutant produced no detectable PKC fragments in the Lactatreated cells (Fig. 4, bottom right). These results indicate that caspase processing was not required for the production of polyUb-PKC conjugates. The single cleavage site mutants were processed to F1 and F2 as expected in the Lacta-treated BHK cells (Fig. 4, bottom right). Thus, the D210A and D239A mutations prevented the accumulation of F1 and F2, respectively (Fig. 4, bottom right). F1 and F2 PKC fragments were  (Etop). B, the cells were incubated for 18 h in the presence or absence of 50 M etoposide and the indicated concentration of zDEVD-fmk (zDEVD) or zVAD-fmk (zVAD). The cells were lysed, and PKC was immunoprecipitated from a 0.5-mg sample. Histone H1 kinase activity of the immunoprecipitates was assayed as described previously (21). The assay buffer contained 25 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM DTT, 50 M ATP, 1 Ci of [␥-32 P]ATP, 20 g of histone H1 (Sigma, histone type III-SS), and 40 g/ml phosphatidylserine (Avanti Polar Lipids) as indicated. After 10 min at 30°C, the reaction was stopped and [ 32 P]histone was measured as described previously (21). C, the cells were incubated for 18 or 24 h in the absence or presence of 100 M etoposide and 100 M zVAD-fmk as indicated. The cells were lysed, and proteins were subjected to Western blot analysis with the antibody to PKC. F1 and F2 indicate the positions of the 50-and 40-kDa carboxyl-terminal fragments of PKC, respectively. Two additional experiments gave similar results. detected in BHK cells transfected with wild type PKC (Fig. 4,  bottom right, first lane), apparently because BHK cells spontaneously undergo apoptosis shortly after becoming confluent. 3 Expression of the full-length PKC and the single and double caspase cleavage site mutants in HEK 293 cells showed that caspase processing occurred chiefly at EETD 210 and DGMD 239 , as in the case of BHK cells. 3 Treatment of BHK Cells with Lacta or Etoposide Activated Caspase-3-like Activity, but Only Lacta Caused PKC Fragments to Accumulate-Lacta treatment produced a severalfold increase in caspase-3-like activity in the BHK cells, which were transfected with lacZ or PKC (Fig. 5, A and B). The extent of the activation of caspase-3-like activity by Lacta was similar to that produced by etoposide, a topoisomerase II inhibitor, which has been widely used to induce apoptosis (14). The combination of Lacta with etoposide synergistically activated caspase-3-like activity (Fig. 5, A and B). Lacta and etoposide or the combination of the two activated caspase-3-like activity in nontransfected BHK cells similarly to transfected cells. 3 Lacta probably evoked the accumulation of carboxyl-terminal PKC fragments by two mechanisms: 1) inhibition of the proteasome activated caspases (Fig. 5, A and B), as shown previously (36) and 2) the inhibition of the proteasome prevented the degradation of polyubiquitinylated proteins (29), including the carboxyl-terminal PKC fragments (Fig. 4).
Interestingly the accumulation of full-length PKC and the F1 and F2 fragments required treatment with Lacta (Fig. 5C). Thus, full-length PKC and/or the F1 and F2 fragments were probably degraded as rapidly as they were produced in the absence of the proteasome inhibitor. Extraction of BHK cells with Triton X-100 (Fig. 5C) solubilized relatively little of the cross-reacting protein, which had a slightly slower mobility than PKC compared with extraction with SDS (Fig. 4). Treatment of lacZ-transfected cells with etoposide produced a small amount of a polypeptide with a mobility similar to that of the F1. The polypeptide is probably not F1, however, because BHK cells lacked detectable PKC and because Lacta failed to increase the accumulation of the polypeptide (Fig. 5C). 3

DISCUSSION
Atypical PKC, like other PKC isoforms, consists of an amino-terminal regulatory domain (first ϳ240 amino acids), which is connected to the carboxyl-terminal catalytic domain (amino acids ϳ240 -584) via a short hinge segment (1)(2)(3). The results presented here show that both initiator (e.g. caspase-8) and effector (e.g. caspase-3, -6, and -7) caspases efficiently process human PKC in vitro at the hinge segment and generate carboxyl-terminal fragments of ϳ40 and 50 kDa, which contain the catalytic domain. Each of the caspases readily or predominantly cleaved human PKC at EETD2G in vitro, which was not expected, based on the known substrate specificities of the caspases. The substrate specificity of caspase-2, -3, and -7 is DEXD and that of caspase-6, -8, and -9 is (I/L/V)EXD (25,37). Furthermore, studies with peptide substrates have shown that caspase-7 is more efficient than caspase-3 in processing peptides with glutamate at the P4 or P3 positions by six and two times, respectively (25). However, caspase-3 readily processed PKC at EETD2G (Fig. 1). These results support the idea that the structural context of a protein contributes to recognition as a caspase substrate (25). Some of the PKC processing data are consistent with in vitro studies of the substrate specificities of recombinant human caspases. For example, caspase-3 is three times more efficient than caspase-7 at cleaving peptides with a P2 methionine instead of valine (25), and caspase-3, but not caspase-7, cleaved PKC at DGMD2G (Fig. 1). A recent report indicated that Myc-tagged rat PKC expressed in HeLa cells was processed at EETD2G and DGVD2G following the induction of apoptosis by UV radiation (32). The valine for methionine substitution at the caspase site of rat versus human PKC may explain the processing of rat PKC predominantly at DGVD2G instead of EETD2G. 3 Our results appear to be the first demonstration of PKC activation following the induction of apoptosis. Moreover, the activation of PKC depended on caspase processing to carboxylterminal fragments (Figs. 2 and 3). Caspase processing of epitope-tagged or endogenous PKC produced catalytic domain fragments (F1 and F2), which would be expected to dissociate from the autoinhibitory amino-terminal regulatory domain (1-3). CAT-V5, which encodes the catalytic domain with a V5 epitope, was active following expression and immunoprecipitation from HeLa cells (Fig. 2). Furthermore, increased PKC immune complex kinase activity accompanied caspase processing of epitope-tagged or endogenous PKC in parotid C5 or HeLa cells, respectively, following the induction of apoptosis (Figs. 2 and 3). The general caspase inhibitor, zVAD-fmk, which abolished V5-tagged or endogenous PKC processing, prevented the increase in immune complex kinase activity ( Figs. 2 and 3). These results suggest that caspase processing of latent, inactive PKC constitutes a mechanism for activation of the kinase. Additional work is needed to determine whether caspase processing activates PKC independently of phosphorylation by an upstream kinase (38,39), which is plausible because proteolysis between the regulatory and catalytic domains would be expected to dissociate them and thereby relieve the autoinhibition of the kinase function (1)(2)(3).
In contrast to the activation of PKC following the induction of apoptosis in HeLa cells described here, Frutos and coworkers (32) observed that the induction of apoptosis by the exposure of HeLa cells to UV radiation markedly diminished the immune complex kinase activity of Myc-tagged PKC. The decrease in immune complex kinase activity was unrelated to caspase processing of PKC, because the kinase activity of a caspase-resistant PKC mutant was decreased similarly to that of wild type PKC (32). Although the mechanism of kinase activation that accompanies caspase processing is most likely the dissociation of the inhibitory amino-terminal regulatory domain, the biochemical basis for the inhibition of the kinase function following UV radiation is unknown (32). More work is needed to determine whether caspase processing and activation of PKC, which follows the induction of apoptosis, generates executive and/or survival signals.
The present findings appear to be the first to show that an atypical PKC isoform, which lacks a functional diacylglycerolbinding C1 domain, forms Ub conjugates and is degraded by the Ub-proteasome. Polyubiquitinylation reversibly marks proteins for rapid destruction by the 26 S proteasome (22,23). Polyubiquitinylated-PKC conjugates accumulated in BHK cells that were transfected with the catalytic domain or fulllength PKC or with PKC mutants with Ala substitutions for Asp at the caspase cleavage sites (Fig. 4). These results indicate that the carboxyl-terminal catalytic segment of PKC (amino acids 240 -592) is sufficient for ubiquitinylation in transfected BHK cells. Importantly, accumulation of the polyubiquitinylated-PKC conjugates depended on transfection with wild type or a mutant of PKC and blockade of the 26 S proteasome with Lacta (Fig. 4). Without Lacta the cells appear to degrade the Ub-PKC conjugates essentially as rapidly as they are produced. In fact it was difficult to detect the carboxylterminal PKC fragments in the absence of Lacta, even when caspases were activated by treatment with etoposide (Fig. 5C). Although the half-life of PKC has not been reported, the present findings suggest that it is short-lived, at least in BHK cells, because so little of the catalytic domain or full-length PKC accumulated in the transfected unless the proteasome was blocked (Figs. 4 and 5). Polyubiquitinylated-PKC conjugates of the caspase-resistant double mutant accumulated providing the proteasome was blocked by Lacta (Fig. 4). Therefore, ubiquitinylation of a full-length PKC mutant can occur, although additional studies are needed to exclude the possibility that the substitution of the Asp residues with Ala affected ubiquitinylation.
Discreet structural elements, called destruction signals or degrons, which are recognized by the ubiquitinylation E2⅐E3 complex, have been identified in a variety of proteins in fungi and mammalian cells (22,23). Degrons include the cyclin destruction boxes, the ␦ domain of c-Jun, a destabilizing aminoterminal residue (the N-degron), Deg1 and Deg2 of MAT␣2, the DSGXXS motif of IB and ␤-catenin, the Ub conjugation motif of growth hormone receptor, and the PEST elements of ornithine decarboxylase and other proteins (22, 23, 40 -45). Because transfection with the catalytic domain of PKC readily produced Ub conjugates, there appears to be a degron in the catalytic domain. PKC has several prominent PEST elements (amino acids 51-73, 86 -104, 175-190, 190 -211) in the regulatory domain and one in the catalytic domain (amino acids 521-560) (26). Interestingly, the prominent caspase cleavage site of PKC (EETD2G) lies in the middle of the PEST element, which has the highest PEST score of 6.4 (26). If this PEST element constitutes a degron, then processing at EETD2G might alter the stability of the protein. Alternatively, phosphorylation may influence processing of PKC, because PEST sites contain serine and/or threonine residues and because phosphorylation is known to regulate the processing of IB and other proteins by caspases (15,44,46,47). Phosphorylation, in turn, is known to regulate ubiquitinylation. For example, an F box protein recognizes proximal phosphorylated serines of the DSGXXS motif of IB and ␤-catenin (41)(42)(43), and dephosphorylation preserves c-Fos, c-Jun, and c-Mos from destruction by the Ub-proteasome system (40 -43, 48 -50). PKC provides an attractive model for furthering our understanding of the interrelationships between the three major post-translational modifications of phosphorylation, ubiquitinylation, and caspase processing, which are likely to govern the functional impact of this pivotal kinase on cell proliferation, differentiation, malignant transformation, and apoptosis.