Protein Kinase Cα (PKCα) Is Resistant to Long Term Desensitization/Down-regulation by Prolonged Diacylglycerol Stimulation*

Sustained activation of PKCα is required for long term physiological responses, such as growth arrest and differentiation. However, studies with pharmacological agonists (e.g. phorbol 12-myristate 13-acetate (PMA)) indicate that prolonged stimulation leads to PKCα desensitization via dephosphorylation and/or degradation. The current study analyzed effects of chronic stimulation with the physiological agonist diacylglycerol. Repeated addition of 1,2-dioctanoyl-sn-glycerol (DiC8) resulted in sustained plasma membrane association of PKCα in a pattern comparable with that induced by PMA. However, although PMA potently down-regulated PKCα, prolonged activation by DiC8 failed to engage known desensitization mechanisms, with the enzyme remaining membrane-associated and able to support sustained downstream signaling. DiC8-activated PKCα did not undergo dephosphorylation, ubiquitination, or internalization, early events in PKCα desensitization. Although DiC8 efficiently down-regulated novel PKCs PKCδ and PKCϵ, differences in Ca2+ sensitivity and diacylglycerol affinity were excluded as mediators of the selective resistance of PKCα. Roles for Hsp/Hsc70 and Hsp90 were also excluded. PMA, but not DiC8, targeted PKCα to detergent-resistant membranes, and disruption of these domains with cholesterol-binding agents demonstrated a role for differential membrane compartmentalization in selective agonist-induced degradation. Chronic DiC8 treatment failed to desensitize PKCα in several cell types and did not affect PKCβI; thus, conventional PKCs appear generally insensitive to desensitization by sustained diacylglycerol stimulation. Consistent with this conclusion, prolonged (several-day) membrane association/activation of PKCα is seen in self-renewing epithelium of the intestine, cervix, and skin. PKCα deficiency affects gene expression, differentiation, and tumorigenesis in these tissues, highlighting the importance of mechanisms that protect PKCα from desensitization in vivo.

Protein kinase C (PKC) ␣ is a signaling protein that regulates important events in self-renewing tissues, including progression through the cell cycle, differentiation, and cell survival and apoptosis (1)(2)(3)(4)(5). The PKC family comprises three subgroups, termed conventional (cPKC), 3 novel (nPKCs), and atypical PKCs, that have specific cofactor sensitivities as a function of differences in structure (6 -8). PKC␣ is a member of the cPKC subgroup, which also includes PKC␤I/II and PKC␥ (6,9). Catalytic competence of PKC␣ requires phosphorylation on three priming sites in the C-terminal domain (activation loop (Thr 497 ), turn motif (Thr 638 ), and hydrophobic motif (Ser 657 )) (10). The primary physiological activator of PKC␣ and other cPKCs is diacylglycerol (DAG), which binds to the conserved C1 domain in the regulatory region of catalytically competent enzyme (9). DAG is generated through cleavage of phosphatidylinositol 4,5-bisphosphate by phospholipase C (PLC) and by the actions of phospholipase D (PLD) and phosphatidic acid phosphohydrolases on phosphatidylcholine (11). Activation of PLC and PLD by cell surface receptors (e.g. G-protein-coupled receptors and receptor tyrosine kinases) leads to accumulation of DAG and recruitment of cPKCs from the cytoplasm to the plasma membrane. Membrane association of PKCs induces conformational changes that result in kinase activation: thus, membrane association has been widely recognized as an indicator of PKC activity (12). Membrane association and activation of cPKCs also require Ca 2ϩ -dependent binding of phosphatidylserine to the C2 domain (13). The C2 domain of cPKCs has a low intrinsic affinity for Ca 2ϩ ; thus, in the presence of low levels of DAG, activation of cPKCs is dependent on elevated intracellular Ca 2ϩ levels. However, binding of DAG, anionic phospholipids (phosphatidylserine and phosphatidylinositol 4,5-bisphosphate), and Ca 2ϩ is cooperative; therefore, with higher membrane DAG levels, low concentrations of intracellular Ca 2ϩ can support cPKC activation (14 -16). A number of pharmacological PKC agonists, such as phorbol esters and bryostatins, bind the C1 domain of cPKCs with high affinity and can induce membrane translocation and activation of these isozymes independently of DAG (17). The nPKCs, PKC␦, PKC⑀, PKC, and PKC, are also activated by DAG and pharmacological agonists; however, the C2 domain of these enzymes lacks a Ca 2ϩ binding motif, rendering their activation insensitive to Ca 2ϩ (6 -8). The atypical PKCs, PKC and PKC/, lack a C2 domain and do not bind DAG or pharmacological agonists but are instead activated by protein-protein interactions (6 -8).
Acute and long term mechanisms of inactivation of PKC signaling have also been characterized. DAG levels are tightly regulated in the cell; thus, a major mechanism of PKC inactivation following physiological signaling is through rapid metabolism of DAG by DAG kinases and/or DAG lipases (18). The resultant loss of lipid activator (together with the return of intracellular Ca 2ϩ concentrations to basal levels) results in reverse translocation of cPKCs and nPKCs to the cytoplasm where they adopt an inactive conformation that is competent for reactivation upon regeneration of cofactors (19). Notably, reverse translocation of PKCs is not simply a passive diffusion from the membrane but is dependent on PKC activity and is prevented by PKC inhibitors (20). Acute termination of PKC signaling may also involve multisite dephosphorylation by cellular phosphatases (e.g. PP2A and PH domain leucine-rich repeat protein phosphatase) or oxidative mechanisms (19). Another mechanism of inactivation, engaged in response to long term stimulation by physiological activators or pharmacological agonists, is agonist-induced degradation of PKCs. For PKC␣, these long term desensitization mechanisms, which appear to be triggered in a cell type-and agonist-specific manner, include (a) dephosphorylation of activation loop (Thr 497 ), turn motif (Thr 638 ), and hydrophobic motif (Ser 657 ) priming sites followed by proteasomal processing of the dephosphorylated species, (b) ubiquitin/proteasome-dependentdegradationofthemature,fullyphosphorylated enzyme at the plasma membrane, and (c) vesicle trafficking-dependent degradation in lysosomes (19,(21)(22)(23). Agonist-induced degradation of PKCs leads to loss of associated signaling and reversal of their downstream effects in the continued presence of agonists.
Consistent with these reversal mechanisms, agonist treatment often leads to transient activation of PKCs; however, it has long been recognized that many biological responses, such as activation of T cells and mitogenic responses in endothelial and vascular smooth muscle cells, require prolonged activation of PKCs (24 -27). Our studies in intestinal epithelial cells (IECs) have determined that PKC␣ has multiple cell cycle-related effects that lead to growth arrest both in vitro and in vivo and that maintenance of these effects requires sustained activation of the enzyme (1, 2, 28 -31). To further examine the basis for the sustained activation of PKC␣ in vivo, the current study determined the effects of prolonged activation of PKC␣ with DAG in IEC-18 intestinal crypt cells. Our studies demonstrate that PKC␣ is resistant to long term desensitization by prolonged diacylglycerol-induced activation and point to the existence of mechanisms that protect cPKCs from dephosphorylation/degradation under conditions of sustained input signaling.
Plasmids and Transfections-The HA-tagged wild type PKC␣ expression vector (in pGL2) was a gift from Drs. Irwin Gelman and Li-Wu Guo (Roswell Park Cancer Institute, Buffalo, NY). The Y123W mutation was introduced using the QuikChange site-directed mutagenesis kit (Agilent Technologies) using primers ATCACTGTGGGTCACTGCTCTGGG-GACTTATCCATCAAGG and CCTTGATGGATAAGTCC-CCAGAGCAGTGACCCACAGTGAT, and mutagenesis was confirmed by sequencing. Transfections were performed using a 3:1 ratio of X-tremeGENE 9 (Roche Applied Science) to plasmid DNA according to the manufacturer's instructions, and agonist treatments were performed 48 h after transfection.
Immunofluorescence Analysis-Cells grown on glass coverslips were treated as indicated before fixation in 2% formaldehyde, PBS for 15 min at room temperature and processing for immunofluorescence microscopy as we have described previously using 0.2% saponin or 0.05% Triton X-100 for permeabilization (23). Primary antibody dilutions were as follows: rabbit anti-C-terminal PKC␣ at 1:500 and TRITC-conjugated antirabbit secondary antibody at 1:100 or Alexa Fluor 488-conjugated donkey anti-rabbit secondary antibody at 1:800. In experiments involving Triton X-100 extraction, nuclei were stained with Hoechst 33258. Images were obtained using a Zeiss Axioskop epifluorescence microscope with a Hamamatsu C7780 digital camera or a Leica DM5500B epifluorescence microscope with a Leica DFC450 digital camera.
Imaging of Ca 2ϩ -dependent Fura-2 AM Fluorescence-IEC-18 cells grown on glass coverslips were treated with A23187 for 30 min prior to addition of 5 M Fura-2 AM (eBioscience). Cells were incubated with A23187 ϩ Fura-2 AM for an additional 30 min, then rinsed in PBS, and mounted onto glassslides.TheFura-2AMfluorescencewasvisualizedbyepifluorescence microscopy in live cells within 10 min of mounting the coverslips onto slides.
Analysis of Detergent-resistant Membranes-Separation of detergent-resistant membranes (DRMs) on sucrose gradients was essentially as described (33). Briefly, cells grown in two 15-cm plates were treated with vehicle (acetonitrile), 20 M DiC 8 , or 100 nM PMA for 15 min; rinsed two times with ice-cold PBS; and extracted for 30 min on ice in 1 ml of 1% Triton extraction buffer (1% Triton X-100, 50 mM Tris-HCl, 150 mM NaCl, 100 M CaCl 2 , Phosphatase Inhibitor Mixtures 1 and 2 (Sigma), Protease Inhibitor Mixture without metal chelating reagents (Sigma)) containing vehicle, DiC 8 , or PMA as appropriate. Extracts were diluted with an equal volume of 80% (w/v) sucrose in 50 mM Tris-HCl, 150 mM NaCl, 100 M CaCl 2 and overlaid with 2 ml of 30% sucrose and 1 ml of 5% sucrose in the same buffer. Following centrifugation at 250,000 ϫ g for 18 h at 4°C in a Sorval AH-650 rotor, equal fractions were collected from the top of the gradient using an Auto Densi-Flow II (Haake Buchler Instruments Inc.). 20 l of each fraction was subjected to Western blotting for PKC␣, and the relative levels of PKC␣ were quantified using NIH ImageJ software as we have described (22).
Analysis of PKC␣ association with DRMs by immunofluorescence was adapted (34,35). Cells grown on coverslips were treated with vehicle, DiC 8 , or PMA as above prior to extraction for 30 min at 4°C in 150 l of 1% Triton extraction buffer containing the corresponding vehicle or PKC agonist. Cells were then immediately fixed with 2% formaldehyde and processed for PKC␣ immunofluorescence (see above). Images of cells from the different treatments were taken at the same magnification and exposure and processed identically using Adobe Photoshop software. Quantification of immunofluorescence was performed on unprocessed images using NIH ImageJ software and normalized to the number of cells in each field.

Results
Activation and Reverse Translocation of PKC␣ by DiC 8 in IEC-18 Cells-Analysis of the effects of DAG on PKC␣ in IEC-18 cells utilized the synthetic DAG DiC 8 . This short chain DAG was chosen because its use as a mimic of endogenous DAG signaling is long established and is supported by evidence that it (a) rapidly incorporates into the plasma membrane and activates PKCs, (b) is a substrate for DAG-metabolizing enzymes, (c) has essentially the same binding affinities for C1 domains of PKCs as long chain DAGs, and (d) elicits the same physiological responses as phorbol esters (e.g. Refs. 36 -40). Importantly, DiC 8 (a) activates PKC␣ in IEC-18 cells, (b) has the same physiological effects as PMA in these cells, and (c) induces downstream responses in these cells that are consistent with the effects of PKC␣ manipulation in vivo (e.g. Refs. 2, 28 -31, 41, and 42).
The ability of naturally occurring and synthetic DAGs to activate PKC␣ is limited by their rapid metabolism in cell membranes (43,44), which leads to depletion of the agonist and reverse translocation of the enzyme to the cytoplasm (20). This process was characterized in IEC-18 cells by administering a single treatment with a 20 g/ml concentration of the short chain DAG DiC 8 and analyzing PKC␣ localization by immuno-fluorescence microscopy. This was the method of choice for evaluating PKC␣ membrane association because the endogenous protein is fixed in situ for analysis. Subcellular fractionation is highly sensitive to extraction conditions and has resulted in both under-and overestimation of PKC␣ membrane association in response to DAG stimulation (e.g. Refs. 45 and 46). Multiple studies have determined that fluorescence imaging provides a more accurate depiction of the dynamics of PKC signaling molecules (47). As shown in Fig. 1A, i, DiC 8 induced rapid translocation of PKC␣ to the plasma membrane in IEC-18 cells accompanied by clearance of the enzyme from the cytoplasm within 15 min. This effect was indistinguishable from that elicited by the potent PKC agonist PMA and was also seen with 1,2-sn-dioleoylglycerol (Fig. 1B), indicating that DAG can efficiently translocate PKC␣ to the plasma membrane in IEC-18 cells. As expected, the association of PKC␣ with the membrane was transient following a single DiC 8 treatment with the enzyme disappearing from the membrane and gradually accumulating in the cytoplasm 1-3 h following addition of the agonist (Fig. 1A, ii). The effect can be linked to depletion of the short chain DAG from the membrane because reaccumulation of PKC␣ in the cytoplasm was accelerated when DiC 8 was removed from the cells by washing (with complete reversal evident by 1 h; Fig. 1C, panel h). Reverse translocation was also seen following washout of PMA, although cytosolic accumulation was slower than seen with DiC 8 (Fig. 1C, panel i), likely reflecting the slower metabolism of PMA. The appearance of PKC␣ in the cytoplasm following removal of agonist was not primarily the result of accumulation of newly synthesized PKC␣ because it was not affected when new protein synthesis was inhibited with cycloheximide ( Fig. 1C, panels k and l).
In previous studies, we have determined that some PKC agonists, such as bryostatin, can induce lipid raft-dependent endocytosis of membrane-associated PKC␣ in IEC-18 cells (22), and DAG has been shown to be internalized to the endoplasmic reticulum for triglyceride synthesis in some systems (48). To differentiate between endocytic internalization and reverse translocation, cells were treated with the cholesterol-sequestering agent nystatin, which inhibits lipid raft-dependent endocytosis (22,49,50). Nystatin effectively blocked the internalization of PKC␣ elicited by bryostatin ( Fig. 2A, compare panels c and d) but had no effect on the cytosolic reaccumulation of the enzyme following a single DiC 8 treatment ( Fig. 2A, panels g and h). Thus, the return of PKC␣ to the cytoplasm following DiC 8 treatment can be attributed to reverse translocation as a result of metabolism of DiC 8 at the plasma membrane rather than endocytic trafficking.
Because reverse translocation of PKCs is dependent on their kinase activity (20), reaccumulation of PKC␣ in the cytoplasm following DiC 8 treatment is indicative of enzyme activation. To confirm that this is the case in IEC-18 cells, the effect of the PKC inhibitor bisindolylmaleimide I was investigated. As shown in Fig. 2B, bisindolylmaleimide I blocked the return of PKC␣ to the cytosol (compare panels g and h), establishing that DiC 8 treatment promotes both membrane translocation and activation of the enzyme in IEC-18 cells. This finding is supported by the ability of a single DiC 8 treatment to elicit PKC␣-dependent downstream effects in these cells, including activation of 4E-BP1, stimulation of ERK, and down-regulation of cyclin D1, as we have reported previously (31,41,42).
To determine whether DAG-induced PKC␣ activity can be maintained for prolonged periods in IEC-18 cells, cells were subjected to repeated stimulation with DiC 8 at intervals shorter than the time required for reverse translocation to occur (i.e. 10 or 60 min). As long as DiC 8 was replenished before reverse translocation had taken place, repeated additions were able to sustain activation of PKC␣ as reflected in maintenance of the enzyme at the plasma membrane over the full 12-h duration of the treatments (Fig. 3, A and B). Sustained activity was further confirmed by the ability of the enzyme to undergo activationdependent reverse translocation upon removal of DiC 8 by washing at any time during the treatment (data not shown).
Collectively, these data demonstrate that a single DiC 8 treatment leads to rapid membrane association and activation of PKC␣ but that the enzyme reverts to the inactive state between 1 and 3 h of treatment as a result of loss of agonist in the membrane and reverse translocation. Repeated additions of DiC 8 , however, are able to maintain PKC␣ in the active state for prolonged periods of time.
Chronic DiC 8 Stimulation Has Sustained Effects on Downstream Targets of PKC␣ Signaling-Based on the finding that repeated administration of DiC 8 over a 12-h period results in sustained membrane association of PKC␣ (Fig. 3B), we asked whether this effect was accompanied by prolonged downstream signaling. We have previously demonstrated that treatment of IEC-18 cells with PKC agonists results in PKC␣-dependent cell cycle-repressive effects (1, 51), including rapid down-regulationofthetranscriptionalregulatorId1anddephosphorylation/activation of the translational repressor 4E-BP1 (30,42). These effects are transient following PMA or bryostatin treatment with reversal coinciding with agonist-induced degradation of the enzyme (30,42). However, as shown in Fig. 4, the sustained membrane association of PKC␣ induced by repeated addition of DiC 8 was associated with sustained downregulation of Id1 and dephosphorylation of 4E-BP1 (as indicated by the appearance of more rapidly migrating bands on Western blots; Fig. 4, arrows).
Chronic DiC 8 Stimulation Does Not Result in Down-regulation of PKC␣-The ability of DiC 8 to support sustained activation of PKC␣ was surprising because it is well established that prolonged stimulation of PKC isozymes can lead to long term desensitization via degradation of the activated kinase (19,21,22). This down-regulation can be readily seen in PMA-and bryostatin-treated IEC-18 cells: as shown in Fig. 5A, i, PMA-or bryostatin-induced down-regulation of PKC␣ is evident as early as 30 min of treatment and is almost complete by 2 h. However, sustained activation of PKC␣ by repeated addition of DiC 8 failed to down-regulate PKC␣ (Fig. 5A, ii). No down-regulation was seen even when DiC 8 was added as frequently as every 10 min or when activation was extended as long as 12 h (Fig. 5, B and C). Importantly, the failure of DiC 8 to affect PKC␣ levels did not reflect a general inability of the agonist to induce down-regulation of PKCs because the other two DAG-responsive PKCs in IEC-18 cells, novel PKC␦ and PKC⑀, were potently down-regulated by DiC 8 treatment (Fig. 5A, ii).
A role for compensatory increases in protein synthesis in the failure of DiC 8 to down-regulate PKC␣ in IEC-18 cells was excluded by addition of the protein synthesis inhibitor cycloheximide (CHX): as shown in Fig. 5D, levels of PKC␣ remained unchanged following a 6-h incubation with DiC 8 in the presence or absence of CHX (compare the first two lanes on the left with the last two lanes on the right). This finding pointed to an inability of DiC 8 to promote degradation of the enzyme. Ago- nist-induced down-regulation of PKC␣ can involve ubiquitination of the enzyme and degradation by the proteasome or lipid raft-dependent intracellular trafficking and lysosomal processing (19,21,22). Involvement of lysosomal degradation can be excluded because repeated addition of DiC 8 does not induce internalization of the protein (Fig. 3). Furthermore, relocation of PKC␣ to the cytoplasm following a single DiC 8 treatment does not involve an endocytic pathway because it was not affected by nystatin ( Fig. 2A). Induction of proteasomal degradation by DiC 8 was excluded by the failure of the proteasome inhibitor MG132 to affect levels of PKC␣ in DiC 8 -treated cells while strongly protecting PKC␦ and PKC⑀ from degradation induced by the short chain DAG (Fig. 5D). Thus, despite the ability of repeated addition of DiC 8 to maintain long term activation of PKC␣, this short chain DAG fails to trigger activationinduced degradation of the enzyme.
Chronic DiC 8 Stimulation Fails to Trigger Early Events That Target PKC␣ for Degradation-Having determined that prolonged activation of PKC␣ by DiC 8 does not result in degradation of the enzyme, studies were conducted to determine the specific stage(s) along the degradation pathways at which this failure occurred. Two mechanisms have been described for agonist-induced ubiquitination/proteasomal degradation of PKC␣ (19,21,23). One pathway involves activation-dependent priming site dephosphorylation, which both inactivates the kinase and targets it for ubiquitin-dependent proteasomal degradation (9,19). A second pathway involves direct ubiquitination of the active, fully phosphorylated kinase at the plasma membrane (21,23). To determine whether DAG induces ubiquitination of PKC␣, ubiquitinated proteins were immunoprecipitated from lysates of DiC 8 -or PMA-treated cells with antiubiquitin antibody P4D1 and probed for PKC␣. Although PMA treatment produced a smear of high molecular weight PKC␣ species characteristic of the polyubiquitinated enzyme (23, 52), DiC 8 -stimulated PKC␣ showed no mono-or polyubiquitination (Fig. 6A). Thus, DiC 8 does not target PKC␣ for ubiquitination. This does not reflect an inherent inability of DAG to promote ubiquitination/proteasomal degradation of PKCs because the degradation of PKC␦ and PKC⑀ was blocked by the proteasome inhibitor MG132 (Fig. 5D).
Priming phosphorylation of PKC␣ at the activation loop (Thr 497 ), turn motif (Thr 638 ), and hydrophobic motif (Ser 657 ) sites leads to characteristic changes in electrophoretic mobility on SDS-polyacrylamide gels as previously determined using phosphosite-specific antibodies (see Ref. 21). The fully dephosphorylated species can be readily detected on Western blots by its more rapid migration at ϳ76 kDa compared with 80 kDa for the mature/phosphorylated enzyme as can be seen in bryosta- tin-treated cells (Fig. 6B, arrow). Notably, faster migrating species were never detected following DiC 8 treatment (Fig. 6B) even after prolonged treatment (e.g. 12 h; see Figs. 4 and 5, A and B). Although addition of the proteasome inhibitor MG132 clearly protected these unstable species in bryostatin-treated cells (Fig. 6B), no dephosphorylation of PKC␣ was observed following either 2-or 6-h treatment with DiC 8 even in the presence of proteasomal inhibition (Figs. 5D and 6B). Thus, DAG does not appear to induce dephosphorylation of PKC␣ in IEC-18 cells.
Previous studies from our laboratory and others have demonstrated that heat shock proteins (Hsps) play important roles in regulating the phosphorylation and stability of PKC␣ following activation by pharmacological agonists, such as phorbol esters or bryostatin (21, 53-55). Hsp90 protects the activated enzyme from dephosphorylation and proteasomal degradation.
Hsp70/Hsc70 also protects activated PKC␣ from dephosphorylation while facilitating PMA-and bryostatin-induced proteasomal degradation of the fully phosphorylated, mature enzyme. Based on these findings, we examined the effect of the Hsp90 inhibitor 17-AAG (56) and the Hsp70 inhibitor PES (57) on PKC␣ stability during chronic DiC 8 treatment. As shown in Fig.  6C, although inhibition of Hsp90 with 17-AAG accelerated the degradation of PKC␣ in bryostatin-treated cells, it had no effect on the stability of the enzyme in DiC 8 -treated cells. When MG132 was included in the treatment to facilitate detection of unstable dephosphorylated forms of PKC␣, the ability of 17-AAG to enhance PKC␣ dephosphorylation in response to bryostatin was apparent (Fig. 6C, open arrow); however, no accumulation of dephosphorylated species of PKC␣ was seen in DiC 8 -treated cells in the absence or presence of 17-AAG even when MG132 was added. Similarly, PES treatment enhanced the dephosphorylation of PKC␣ in response to bryostatin but not in response to DiC 8 (Fig. 6D). (Note that PES stabilizes the mature form of PKC␣ (arrowhead) in bryostatin-treated cells, reflecting the Hsp70 dependence of proteasomal degradation of this species (21).) Collectively, these findings show that the inability of chronic activation by DAG to elicit dephosphorylation/degradation of PKC␣ in IEC-18 cells does not reflect protective effects of the chaperones Hsp90 or Hsp70. Together with the finding that DiC 8 does not induce endocytic trafficking of PKC␣ to the lysosome (see above), these data indicate that DAG-induced activation of PKC␣ fails to trigger early steps in known desensitization/down-regulation pathways for PKC␣.
Insufficient Intracellular Calcium Does Not Account for the Failure of DiC 8 to Induce PKC␣ Degradation-Our findings demonstrated that DiC 8 does not regulate the stability of PKC␣ in IEC-18 cells while promoting degradation of the nPKCs PKC␦ and PKC⑀. A major difference between PKC␣ and the nPKC isozymes is that DAG stimulation of PKC␣ is sensitive to intracellular Ca 2ϩ levels (58). To test whether low intracellular Ca 2ϩ could account for the lack of DiC 8 -induced PKC␣ degradation, IEC-18 cells were pretreated with the calcium ionophore A23187 (20), which promotes influx of Ca 2ϩ from the medium (medium Ca 2ϩ concentration, 1.8 mM). Calcium levels were monitored using the calcium-binding dye Fura-2 AM, which fluoresces only when bound to calcium ions. Although the Fura-2 AM signal was nearly undetectable in dimethyl sulfoxide-treated control cells (Fig. 7A, panel a), it was significantly enhanced with increasing concentrations of A23187, confirming that A23187 treatment led to elevated intracellular Ca 2ϩ levels in IEC-18 cells (Fig. 7A, panels b-e). Cells pretreated with A23187 were stimulated with DiC 8 every hour to induce PKC activation. As shown in Fig. 7B, no down-regulation of PKC␣ was observed in response to DiC 8 at any of the concentrations of calcium ionophore tested. The failure of DiC 8 to promote PKC␣ degradation cannot, therefore, be attributed to insufficient levels of calcium for optimal translocation/activation of the enzyme.
Affinity for DAG Does Not Account for the Resistance of PKC␣ to Down-regulation by DiC 8 -PKC␦ and PKC⑀ have a higher resting affinity for DAG than PKC␣ and other calcium-dependent PKC isozymes due to an invariant tryptophan at position 22 in the C1b domain that is invariant as tyrosine in cPKCs (60). Mutation of this tyrosine to tryptophan in the cPKCs converts the C1b domain from a low affinity to a high affinity DAGbinding module (60). Thus, to determine whether the difference in DAG affinity accounted for the differential ability of DiC 8 to down-regulate PKC␦/⑀ and PKC␣ in IEC-18 cells, the tyrosine in the C1b domain of PKC␣ (residue 123) was mutated to tryptophan. Wild-type and mutant PKC␣ expression constructs were transfected into IEC-18 cells, and the ability of PMA and DiC 8 to down-regulate the respective proteins was determined (cycloheximide was included in the treatments to avoid complications arising from the ability of PKC agonists to activate the CMV promoter of the expression constructs). As expected, both wild-type and tyrosine-to-tryptophan C1b mutant PKC␣ (Y123W) were down-regulated by PMA (Fig. 7C) as was endogenous PKC␦. However, the tyrosine-to-tryptophan C1b mutant PKC␣ showed the same resistance to downregulation by DiC 8 as the wild-type protein. Thus, low inherent affinity for DAG does not account for the inability of chronic DiC 8 stimulation to elicit down-regulation of PKC␣ in IEC-18 cells.

Phosphatidic Acid and Other Metabolites of DiC 8 Do Not Affect Agonist-induced Down-regulation of PKC␣-Because
DAG is known to be rapidly metabolized in cells, we investigated whether metabolites of DiC 8 may affect degradation of PKC␣. The ability of DAG kinase inhibitors to prolong PKC␣ activation in intestinal epithelial cells (61) indicates that generation of phosphatidic acid through phosphorylation is a major metabolic pathway for down-regulation of DAG-mediated signaling in this cell type. Therefore, the effects of the short chain phosphatidic acid DOPA on the ability of DiC 8 and PMA to induce PKC␣ down-regulation were tested. Addition of either 20 or 40 M DOPA had no discernible effect on the levels of PKC␣ in control or DiC 8 -treated cells and did not affect the down-regulation of the enzyme following PMA treatment (Fig.  8A). To assess the impact of other possible metabolites of DiC 8 on agonist-induced PKC␣ down-regulation, the effect of incubating cells with DiC 8 and PMA in combination was determined. As with DOPA, the presence of DiC 8 did not prevent PMA-induced down-regulation of PKC␣ (Fig. 8B), confirming that interference by metabolites of DAG is not responsible for the inability of DiC 8 to induce down-regulation of the enzyme.

PMA-induced Down-regulation of PKC␣ Involves Targeting of the Enzyme to Detergent-resistant, Cholesterol-sensitive
Membrane Domains-It has long been recognized that a portion of activated PKC␣ can be targeted to specialized membrane domains, including DRMs or lipid rafts (e.g. Refs. 33 and 62-64). The fact that bryostatin targets PKC␣ for internalization from the plasma membrane in IEC-18 cells whereas PMA does not (23) indicates that agonists differ in their ability to promote association of the enzyme with functionally distinct membrane domains. Therefore, two approaches were used to determine the ability of DiC 8 and PMA to localize PKC␣ in DRMs. The first involved sucrose gradient centrifugation of cells extracted with cold Triton X-100. Because membrane binding of classical PKCs is Ca 2ϩ -dependent, 100 M CaCl 2 was included in the extraction buffer and gradients. DiC 8 or PMA was added to the extraction buffers as appropriate to control for any potential metabolism of DiC 8 during processing. Following centrifugation, proteins contained within the more buoyant DRMs collect at the 30-5% sucrose interface near the top of the gradient (fractions 2 and 3), whereas membrane proteins that are extracted by the detergent are found in lower fractions (65). In untreated cells, minimal levels of PKC␣ were found in the DRM-containing fractions 2 and 3; however, as seen by others, PMA treatment led to appearance of the enzyme in these fractions with a corresponding loss of the enzyme from the lower non-DRM fractions (Fig. 9A). Notably, DiC 8 promoted only a slight increase in DRM-associated PKC␣ (Fig. 9A). To address the potential concern that the lower levels of DRM-associated PKC␣ in DiC 8 -treated cells reflected metabolism of DiC 8 during the long centrifugation (18 h) involved in the fractionation protocol, association of PKC␣ with DRMs was also analyzed by immunofluorescence. This second approach allowed for immediate fixation of the cells following a short (30-min) Triton X-100 extraction on ice. To preserve membrane associations during extraction, buffers contained Ca 2ϩ and the appropriate PKC agonist. Consistent with results from density gradient fractionation, very little PKC␣ staining was seen in control cells following Triton X-100 extraction (the presence of cells in the field is confirmed by nuclear Hoechst 33258 staining). Staining in DiC 8 -treated cells was indistinguishable from that in control cells following extraction (Fig. 9B), indicating that DiC 8 targets at most a very minor portion of PKC␣ to DRMs even under these conditions. However, detergent-resistant staining for PKC␣ was clearly evident following PMA treatment (Fig. 9B). Quantification of the fluorescence signal indicated that detergent-resistant staining of PKC␣ represented about 10 -25% of that seen in untreated cells, confirming that PMA targets an appreciable portion of cellular PKC␣ to DRMs.   Having established that PMA and DiC 8 differentially target PKC␣ to DRMs/lipid rafts, the role of DRM domains in PKC␣ down-regulation was examined. Cholesterol is a major component of DRM/lipid rafts, and agents that interfere with membrane cholesterol disrupt raft function (66). Thus, the role of DRM domains was studied using the cholesterol-binding agents methyl-␤-cyclodextrin, nystatin, and filipin (67). These agents affect membrane cholesterol in different ways: methyl-␤-cyclodextrin binds cholesterol in its hydrophobic pocket and extracts it from the membrane, whereas nystatin and filipin sequester cholesterol within the membrane (68). Importantly, cholesterol-disruptive agents do not affect the membrane recruitment of PKC␣ by PMA (23). 4 However, as can be seen in Fig. 9C, all three agents were able to inhibit PMA mediated down-regulation of PKC␣, pointing to lipid rafts as the site for degradation. Thus, the observed differences in targeting to DRMs provide a mechanistic explanation for the inability of DiC 8 to trigger agonist-induced down-regulation of PKC␣.
Partitioning of PKC␣ into lipid rafts may position the enzyme to interact with protein(s) that mediates its long term desensitization. A potential candidate is PLD. PLD has been shown to partition into DRMs, is activated by PMA treatment, and can interact directly with PKC␣ (11, 69 -72). Therefore, a role for PLD in mediating PMA-induced PKC␣ down-regulation was explored using 30 mM 1-butanol, which inhibits the phospholipase activity of PLD and blocks its physical association with PKC␣ (69). As shown in Fig. 9D, 1-butanol   Lysates were then subjected to discontinuous 40%/30%/5% (w/v) sucrose density centrifugation and equal volumes (20 l) of gradient fractions were subjected to Western blot analysis for PKC␣. Graph shows the relative levels of PKC␣ in each fraction as determined by densitometric analysis. B, panels a-c, IEC-18 cells grown on coverslips were treated with vehicle, DiC 8 or PMA and extracted with 1% Triton X-100 for 30 min as above. Immediately after extraction, cells were fixed with formaldehyde and processed for PKC␣ immunofluorescence, and nuclei were stained with Hoechst 33258. Images, acquired at the same magnification and exposure and processed identically, show both PKC␣ (Alexa Fluor 488) and nuclear DNA (Hoechst 33258) staining in grayscale. Retention of cells on the coverslips was confirmed by nuclear Hoechst staining. Scale bars, 10 m. C, IEC-18 cells were treated with vehicle (C) or 100 nM PMA (P) in the presence or absence of the indicated concentrations of filipin, nystatin, or methyl-␤-cyclodextrin (M␤CD). After 4 (i and ii) or 3 h (iii), cells were processed for immunoblotting analysis for the indicated proteins. In i, ii, and iii, a nonspecific band (N.S.) detected by the Upstate Biotechnology, Inc. anti-PKC␣ antibody or Fast Green staining of the immunoblot membrane is included as a loading control. D, cells were treated with vehicle (C), 20 g/ml DiC 8 (D), or 100 nM PMA (P) in the presence or absence of 30 mM 1-butanol as indicated prior to processing for immunoblotting detection of PKC␣ and ␤-actin. All Western blot panels show data from a single blot with dotted lines in C indicating where lanes have been rearranged for clarity. Data are representative of two (A) or at least three independent experiments (B, C, and D).
not involved in the differential ability of PMA and DiC 8 to promote degradation of the enzyme. Current studies are directed at characterization of the membrane domains in which PKC␣ degradation occurs and determining the involvement of the lipid environment and/or other lipid raft/DRM-resident proteins.
DiC 8 Fails to Down-regulate cPKCs in Endometrial Cancer Cells-To determine whether the inability of DiC 8 to induce down-regulation of PKC␣ was restricted to intestinal epithelial cells, the effect of PKC agonists was tested in HEC-1-A and SNG-M endometrial cancer cells. As in IEC-18 cells, PKC␣ was readily down-regulated by PMA and bryostatin in endometrial cancer cells ( Fig. 10 and data not shown), but was resistant to down-regulation by repeated (every 30 min) DiC 8 treatment (Fig. 10). Thus, PKC␣ is insensitive to DAG-induced downregulation in multiple cell types. Although PKC␣ is the only cPKC expressed in IEC-18 cells, HEC-1-A and SNG-M cells also express PKC␤I. Interestingly, this isozyme was also resistant to down-regulation by DiC 8 but not by PMA (Fig. 10). Thus, selective resistance to DAG-induced down-regulation extends to multiple members of the cPKC subfamily.
Sustained Activation of PKC␣ Is Observed in Vivo-Our analysis of PKC␣ in the self-renewing intestinal epithelium revealed that, although PKC␣ is cytosolic in proliferating intestinal crypt cells, it is membrane-associated/activated in all postmitotic cells of the intestinal villus (Fig. 11A) cells that reside in this compartment for 2-3 days before being extruded into the intestinal lumen. Notably, we have previously demonstrated that loss of the enzyme in these cells in PKC␣ knock-out animals is associated with increased levels of Id1, consistent with a sustained repressive effect of PKC␣ signaling in these postmitotic cells. A similar pattern of sustained PKC␣ membrane association/activation is seen in other self-renewing epithelial tissues, including skin (3) and the stratified squamous epithelium of the cervix (Fig. 11B). Thus, consistent with our findings using DiC 8 in vitro, PKC␣ appears to be resistant to activation-induced down-regulation in response to natural stimuli in vivo.

Discussion
Many of the actions of PKC␣ require prolonged signaling despite well established mechanisms for desensitization of the enzyme following sustained activation (19,21,22). Here we show that DAG-stimulated PKC␣ is resistant to known desensitization mechanisms that are triggered by prolonged activation of the enzyme by pharmacological agonists. Sustained membrane association for as long as 12 h fails to target PKC␣ for dephosphorylation/proteolytic processing in DiC 8 -treated cells. This resistance cannot be attributed to an inability of DiC 8 to activate the kinase or to sustain kinase activation for prolonged periods because (a) the extent of membrane translocation of PKC␣ induced by DiC 8 in IEC-18 cells was indistinguishable from that seen with the potent agonist PMA (Fig. 1), (b) in the absence of repeated stimulation with DiC 8 PKC␣ undergoes activity-dependent reverse translocation that is blocked by pharmacological inhibitors of the enzyme (Fig. 2), and (c) PKC␣-specific downstream effects (e.g. Id1 down-regulation and 4E-BP1 activation) are sustained for at least 12 h in cells when DiC 8 is added repeatedly to maintain PKC␣ mem-  brane association/activity (Figs. 3 and 4). The efficacy of DiC 8 was further confirmed by its ability to down-regulate PKC␦ and PKC⑀ in the same cells. The inability of DiC 8 to down-regulate PKC␣ is also not due to effects of a metabolite(s) of this short chain DAG because neither addition of DOPA nor co-incubation with DiC 8 affected the ability of PMA to induce loss of the enzyme (Fig. 8). Finally, the effect does not reflect a general insensitivity of PKC␣ to activation-induced dephosphorylation/down-regulation in the cells used in this study because bryostatin promotes PKC␣ dephosphorylation by 1 h of treatment, and both PMA and bryostatin induce degradation of the enzyme by 2 h (Fig. 5 and Refs. 21 and 23).
Analysis of the effects of DiC 8 indicated that DAG fails to trigger early events that target PKC␣ for degradation. Prolonged DiC 8 treatment did not result in endocytic trafficking of PKC␣, which directs the protein to lysosomes for processing following activation by bryostatin (22). Similarly, sustained activation by DiC 8 did not lead to dephosphorylation or ubiquitination of PKC␣, events that target the enzyme for proteasomal degradation in response to pharmacological agonists (19,21,23). The lack of dephosphorylation in response to DAG is in keeping with the low levels of dephosphorylation of endogenous PKC␣ induced by PMA or bryostatin in IEC-18 cells (21,23). Previous studies from our laboratory and others have determined that Hsp90 and Hsp70 act to protect activated PKC␣ from dephosphorylation and/or degradation in response to PMA or bryostatin treatment (9,21). However, no protective role of these Hsps was seen for DiC 8 -activated PKC␣ (Fig. 6), indicating that the inability of DAG to induce degradation of PKC␣ is due to a failure to trigger event(s) upstream of the effects of these chaperones.
Although prolonged DiC 8 treatment failed to affect PKC␣ levels, it readily down-regulated PKC␦ and PKC⑀ in IEC-18 cells. Based on this finding, we explored unique aspects of the regulation of these isozymes that may underlie the differential effects of prolonged DiC 8 treatment. A major difference between the nPKCs and cPKCs, including PKC␣, lies in the regulation of binding to DAG and anionic lipids. The C1 domains of cPKCs and nPKCs have two DAG binding sites, C1a and C1b. The C1b domain of cPKCs has a 2 orders of magnitude lower affinity for DAG than the C1b domain of PKC␦ and PKC⑀ and cannot mediate kinase activation (60). Although the C1a domains of cPKCs and nPKCs have high affinity for DAG, the C1a domain of unstimulated PKC␣ is masked by interaction with the C2 and catalytic domains and is only released for DAG binding following Ca 2ϩ -dependent interaction of the C2 domain with anionic lipids in the plasma membrane (13,73,74). Thus, activation of PKC␣ is a two-step process in which the enzyme is first recruited to the membrane by the C2 domain, leading to release of the C1a domain for DAG binding. The C2 domain has a low intrinsic affinity for Ca 2ϩ , but Ca 2ϩ binding is enhanced by DAG, phosphatidylserine, and phosphatidylinositol 4,5-bisphosphate in the membrane, allowing activation by DAG even at subphysiological intracellular Ca 2ϩ levels (14 -16, 74). Nonetheless, elevated intracellular Ca 2ϩ concentrations increase the rate of PKC␣ activation in the presence of DAG (74). In contrast to PKC␣, interaction of nPKCs with the plasma membrane is Ca 2ϩ -independent, suggesting that Ca 2ϩ levels may be limiting for optimal activation and eventual desensitization of PKC␣ by DiC 8 in IEC-18 cells. However, our analysis excluded insufficient Ca 2ϩ as the basis for the inability of DAG to induce PKC␣ dephosphorylation or degradation because these effects were still not seen when intracellular Ca 2ϩ levels were increased using the calcium ionophore A23187 (Fig. 7).
The low DAG binding affinity of the C1b domain in cPKCs reflects the presence of an invariant tyrosine in this domain (at residue 123 in PKC␣) that is a tryptophan in the high affinity domain of nPKCs (60). However, differences in the DAG binding affinity of the C1b domain were also excluded as an explanation for the differential effects of DAG because DiC 8 failed to down-regulate a PKC␣ mutant in which tyrosine 123 was replaced by tryptophan (Fig. 7) to convert its C1b domain to a high affinity DAG-binding module resembling that of PKC␦ and PKC⑀ (60).
Our data indicate that association with distinct membrane subdomains plays an important role in the ability of agonists to induce down-regulation of PKC␣. PMA targets a proportion of PKC␣ to DRMs, whereas at most minimal DRM association was seen after stimulation with DiC 8 (Fig. 9). A role for DRM domain(s) in PKC␣ down-regulation was confirmed by the ability of cholesterol-disrupting agents to inhibit PMA-induced down-regulation of the enzyme (Fig. 9). Differential targeting of PKC␣ to distinct membrane domains is in keeping with studies using rat basophilic leukemia cells that demonstrated that the C1a domain of the cPKC PKC␥ is nearly immobilized in the membrane following recruitment by PMA but is diffusible in the membrane when bound to DiC 8 or DAG derived from membrane lipids (75). The selective ability of PMA to target PKC␣ to specific membrane domains likely reflects recruitment of the agonist-bound enzyme rather than partitioning of this agonist into these domains because (a) PMA appears to preferentially target PKC␣ over PKC␦ and PKC⑀ to DRMs in T lymphocytes (76) and (b) partitioning experiments did not detect PKC⑀ in DRMs in PMA-treated IEC-18 cells. 5 The demonstration that both PKC␣ and PKC␤I are resistant to desensitization by chronic DiC 8 treatment (Fig. 10) indicates that the resistance to DAG-induced down-regulation is a common property of cPKCs. Although the structural basis for this resistance remains to be determined, evidence points to a role for the agonist-binding C1 domain. As demonstrated from analysis using fragments of PKC␥, DAG and PMA induce different membrane interactions of the C1a domain of cPKCs (75). The C1 domain of cPKCs is highly conserved but differs in sequence and location from that of nPKCs (6). Notably, structural analysis has indicated that DiC 8 and PMA induce different conformational changes and membrane associations of the C1 domain of PKC␣ (77). Thus, this domain is likely to support distinct interactions dependent on both the bound agonist and PKC class involved. Ongoing studies are directed at determining how structural differences affect membrane compartmentalization and protein-protein interactions that may regulate desensitization of PKC signaling (e.g. binding to receptors for activated C-kinase (RACKs), syndecan-4, and Pin1 (78 -81)).
Collectively, our findings indicate that the cPKCs have unique properties that protect them from dephosphorylation/ degradation under conditions of sustained input signaling. The physiological relevance of this finding is highlighted by in vivo studies in self-renewing epithelial tissues. Analysis of intestinal, epidermal, and stratified squamous epithelial systems indicates that activated, membrane-associated PKC␣ is also protected from down-regulation by natural agonists in vivo. In the mouse intestinal epithelium, PKC␣ is cytosolic/inactive in proliferating cells of the crypt and is activated coincident with growth arrest at the crypt/villus junction as indicated by a clearing of the enzyme from the cytoplasm and its appearance at the plasma membrane ( Fig. 11 and Ref. 30). Notably, this pattern is maintained along the entire length of the villus, indicating that the enzyme remains activated in the face of ongoing signals without any sign of down-regulation for the 2-3 days required for migration of postmitotic cells to the villus tip. Activation of PKC␣ triggers a program of cell cycle withdrawal in intestinal cells that includes rapid down-regulation of Id1 (2,30): that PKC␣ is active in villus cells is indicated by the finding that PKC␣ deficiency results in aberrant expression of Id1 in this compartment (30). Studies by Rozengurt and co-workers (82,83) provide additional evidence for the failure of physiological agonists to induce long term desensitization of PKC␣ in intestinal epithelial cells. Treatment of IEC-18 cells with the G qcoupled receptor agonist angiotensin II promoted PKC␣ activity-dependent feedback inhibition of angiotensin II-induced EGF receptor transactivation (82). Notably, prolonged exposure of IEC-18 cells to angiotensin II for as long as 24 h failed to affect PKC␣ levels or phosphorylation in this system (83). A similar scenario is seen in the epidermis with PKC␣ becoming plasma membrane-associated/activated in the lower spinous layers coincident with growth arrest/terminal differentiation and remaining exclusively localized at the membrane during the several days it takes for cells to transit through this layer (3). Evidence for PKC␣ activity in this compartment was provided by knockdown experiments in an in vitro organotypic epidermis model that led to increased basal and suprabasal proliferation marker expression, decreased differentiation, and reduced epidermal stratification (3). Notably, epidermal PKC␣ is down-regulated within 3 h of topical administration of PMA (e.g. Refs. 84 and 85); thus, the persistence of activated PKC␣ in this tissue reflects selective insensitivity to down-regulation by natural agonists and is not due to rapid replacement of down-regulated enzyme. The fact that PKC␣ expression is lost in human colon and basal cell carcinomas (86,87) together with the finding that PKC␣ knock-out enhances tumorigenesis in the mouse intestine and epidermis (59,88) emphasizes the importance of maintaining PKC␣ signaling for tissue homeostasis. Future studies will address the specific mechanisms involved in protecting DAG-activated PKC␣ from desensitization in tissues.