Agonist-induced Down-regulation of Endogenous Protein Kinase C α through an Endolysosomal Mechanism*

Background: Mechanisms of endogenous PKC signal termination remain to be fully characterized. Results: Activated endogenous PKCα undergoes dynamin-dependent and -independent endocytic uptake and traffics through early and late endosomes for processing by lysosomes. Conclusion: Lysosomal degradation represents a novel mechanism of desensitizing PKC-mediated signaling. Significance: Multiple degradation mechanisms ensure strict control of the duration of PKC signaling in cells. Protein kinase C (PKC) isozymes undergo down-regulation upon sustained stimulation. Previous studies have pointed to the existence of both proteasome-dependent and -independent pathways of PKCα processing. Here we demonstrate that these down-regulation pathways are engaged in different subcellular compartments; proteasomal degradation occurs mainly at the plasma membrane, whereas non-proteasomal processing occurs in the perinuclear region. Using cholesterol depletion, pharmacological inhibitors, RNA interference, and dominant-negative mutants, we define the mechanisms involved in perinuclear accumulation of PKCα and identify the non-proteasomal mechanism mediating its degradation. We show that intracellular accumulation of PKCα involves at least two clathrin-independent, cholesterol/lipid raft-mediated pathways that do not require ubiquitination of the protein; one is dynamin-dependent and likely involves caveolae, whereas the other is dynamin- and small GTPase-independent. Internalized PKCα traffics through endosomes and is delivered to the lysosome for degradation. Supportive evidence includes (a) detection of the enzyme in EEA1-positive early endosomes, Rab7-positive late endosomes/multivesicular bodies, and LAMP1-positive lysosomes and (b) inhibition of its down-regulation by lysosome-disrupting agents and leupeptin. Only limited dephosphorylation of PKCα occurs during trafficking, with fully mature enzyme being the main target for lysosomal degradation. These studies define a novel and widespread mechanism of desensitization of PKCα signaling that involves endocytic trafficking and lysosome-mediated degradation of the mature, fully phosphorylated protein.

Members of the protein kinase C (PKC) family of serine/ threonine kinases are central players in signal transduction pathways involved in regulation of cell proliferation, differentiation, apoptosis/survival, migration, and invasion as well as receptor trafficking and gene expression (1)(2)(3)(4)(5)(6)(7)(8). PKC␣, a member of the conventional, Ca 2ϩ -dependent class of PKC isozymes (that also includes PKC␤I, -␤II, and -␥), has been implicated in both negative and positive regulation of cell proliferation, survival, and motility in normal cells and tissues (1), and alterations in PKC␣ signaling/expression have been associated with the transformed phenotype in a variety of tumor types (9 -14). Aberrant degradation of PKC␣ has also been associated with various pathologies, including Huntington and Alzheimer (15,16). Strict control of PKC␣ function is achieved by three coordinated mechanisms: (a) phosphorylation on three priming sites in the C-terminal domain (activation loop (Thr 497 ), turn motif (Thr 638 ), and hydrophobic motif (Ser 657 )), which is required for catalytic competence (17,18); (b) binding of Ca 2ϩ and diacylglycerol, which promotes membrane targeting and conformational activation (18); and (c) acute and long term desensitization mechanisms (7,19). Acute termination of PKC␣ signaling may occur through a process termed "reverse translocation," involving return of the enzyme to the cytosol after its translocation and activation at the membrane (20,21). Ligand-induced PKC activity may also be terminated through multisite dephosphorylation by cellular phosphatases such as protein phosphatase 2A and PH domain leucine-rich repeat protein phosphatase (PHLPP) (22)(23)(24)(25). Other studies have indicated that PKC␣ can be acutely inactivated by oxidative mechanisms (26).
It has long been recognized that PKC/PKC␣ signaling can also undergo long term desensitization through down-regulation (27,28). Prolonged activation of PKC isozymes by physiological activators (e.g. hormones, growth factors, G protein-coupled receptor agonists; Refs. 23 and 28 -30) or pharmacological agonists (e.g. phorbol esters, bryostatin 1 (Bryo) 2 ) leads to degradation of these proteins (25,27,(31)(32)(33). Although the steps involved in maturation, activation, and acute desensitization of PKC isozymes have been extensively studied, the pathways underlying long term desensitization of these mole-cules remain poorly understood (19,34,35). Mechanisms that have been put forward include (a) conformation-dependent proteolysis by Ca 2ϩ -dependent proteases (calpains; Ref. 36, 37, and 38), (b) vesicle trafficking-dependent multisite dephosphorylation (23,39) and ubiquitin/proteasome-dependent degradation of the dephosphorylated species (29 -32, 40 -42), and (c) caspase-mediated cleavage. It has been further proposed that the kinase domain is required for recognition by the degradation machinery (25,32,39,40) and that kinase activity may be important for modification of the enzyme itself, increased vesicle trafficking for delivery to a degradation compartment, and activation of protease activity. Together, the available data have been combined to generate the following working model for long term PKC␣ desensitization (19): (a) trafficking of the active enzyme to a perinuclear compartment (19,39,42), (b) dephosphorylation of priming sites by protein phosphatase 2A-like activity (24) and/or by PHLPP1/2 (22), and (c) ubiquitination of the dephosphorylated protein and degradation through a proteasomal pathway (19,40).
Although this model may reflect one pathway of PKC degradation, several lines of evidence from our laboratory and others indicate that there is considerable complexity in the regulation of PKC␣ desensitization processes (35) and that the proposed pathway likely combines aspects of more than one distinct down-regulation mechanism. One source for this confusion may be the widespread use of overexpression strategies to study PKC processing. Our previous studies on endogenous PKC␣ in Bryo-treated IEC-18 intestinal epithelial cells have indicated that at least two mechanisms of PKC␣ degradation can be engaged at the same time. In addition to proteasomal degradation, Bryo induces a novel non-proteasomal, calpain-independent mechanism of PKC␣ processing. In the current study we further characterize this non-proteasomal mechanism of PKC␣ desensitization. We demonstrate that it involves trafficking of activated PKC␣ from the plasma membrane through cholesterol/lipid raft-mediated endocytic pathways that can be dynamin-dependent or -independent, with eventual degradation of the protein via a lysosomal mechanism. Intracellular trafficking of PKC does not require ubiquitination, and the mature, phosphorylated species appears to be the predominant target for lysosomal processing. A better understanding of PKC desensitization mechanisms has important implications for the development of novel approaches to target these kinases in disease (35,43).

MATERIALS AND METHODS
Cell Culture-IEC-18 rat intestinal crypt-like cells (ATCC CRL-1589) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum (FBS), 4 mM L-glutamine, and 5 g/ml insulin. U-87 human glioblastoma cells and HeLa human cervical cancer cells were cultured in DMEM supplemented with 10% FBS and 2 mM L-glutamine; HeLa cells were maintained in the presence of penicillin/streptomycin. L929 mouse fibroblasts were grown in DMEM, 5% FBS. Experiments with genistein were carried out in serum-free DMEM with no additives.
Confocal images are 0.5 M thick optical sections acquired on a Zeiss LSM 510 Meta Confocal Laser Scanning Microscope using a 63ϫ Plan-Apo/1.4 NA oil objective and Zeiss AIM and LSM Image Browser software. Alexa Fluor 488 was visualized using an argon laser at 488 nm, with detection through a 505-550-nm band pass filter. Cy3 was visualized using a HeNe laser at 543 nm and a 575-615-nm band pass filter.
Adenoviral Overexpression-IEC-18 cells on glass coverslips were infected with LacZ-adenovirus or PKC␣-adenovirus at a multiplicity of infection of 5 for 16 h in IEC-18 minimal medium: DMEM supplemented with 2% FBS, 4 mM L-glutamine, and 5 g/ml insulin. The medium was then replaced with IEC-18 complete medium (DMEM supplemented with 5% FBS, 4 mM L-glutamine, and 5 g/ml insulin), and cells were grown to near confluence before treatments.
Plasmids and Transfection-Dominant negative mutant constructs T22N-Rab7-GFP, T17N-Cdc42-GFP, and T27N-Arf6-HA were purchased from Addgene. The dominant negative K44A-dynamin-HA plasmid was a kind gift from Dr. Jeffrey Benovic (Thomas Jefferson University). IEC-18 cells were grown at low confluency on glass coverslips, and transfections were performed using a 3:1 ratio of FuGENE 6 to plasmid DNA as recommended by the manufacturer (Roche Applied Science). Drug treatments and analysis were performed 24 -48 h after transfection.
siRNA-IEC-18 cells were transfected with 100 nM clathrin heavy chain siRNA (Dharmacon) or with 100 nM Silencer Select negative control #1 siRNA (Invitrogen) using Lipofectamine 2000 transfection reagent (Invitrogen) according to manufacturer protocols (Invitrogen). Medium was changed, and transfection with siRNA was repeated after 72 h to ensure full depletion of clathrin heavy chain. Analysis was performed 72 h after the second transfection.
Transferrin and Cholera Toxin Uptake Assay-Transferrin and cholera toxin uptake was examined using previously published protocols (45,46). Briefly, cells were washed twice with PBS to remove all residual serum and incubated 1 h in serumfree DMEM containing 20 mM L-glutamine and 2 nM insulin. 25 g/ml Alexa Fluor 488-conjugated transferrin or 10 g/ml Alexa Fluor 488-conjugated cholera toxin (Invitrogen) were added to the treated cells for 15 min at room temperature, and the medium was then replaced with 37°C medium containing endocytic pathway inhibitors and PKC agonists as indicated. Cells were processed for immunofluorescence after 15-30 min. Where indicated, dynasore was added 30 min before Alexa Fluor 488-conjugated transferrin.

Differential Effects of Agonist Treatment on Endogenous and Overexpressed PKC␣ in IEC-18 Cells
The PKC agonists Bryo and PMA potently activate and then down-regulate PKC␣ in a variety of cell types, and both agents promote subcellular redistribution of the enzyme. Our previous studies demonstrated that these agonists have different effects on the distribution of endogenous PKC␣ in IEC-18 intestinal epithelial cells; Bryo induces both membrane translocation and perinuclear accumulation of the enzyme, whereas PMA predominantly induces plasma membrane association (Fig. 1A) (31). As shown in Fig. 1B, both Bryo and PMA significantly down-regulate endogenous PKC␣ in IEC-18 cells (p Ͻ 0.001 for Bryo/PMA versus control). Down-regulation primarily occurs through a proteasome-dependent pathway (31), as indicated by the inhibitory effects of MG132 at early times of treatment (e.g. 3 h; Fig. 1B, p Ͻ 0.001 for effects of MG132 on down-regulation by Bryo or PMA). Immunofluorescence analysis demonstrated that plasma membrane-associated PKC␣ is the main target for proteasomal degradation, as proteasome inhibition protected the enzyme in this compartment while not affecting levels of internalized protein (Fig. 1C, compare panels b and c and panels d and e). Bryo also engages a slower, non-proteasomal mechanism of endogenous PKC␣ down-regulation, evident after prolonged incubation in the presence of proteasome inhibitors (Fig. 1B, compare lanes 4 -6 and 10 -12; p ϭ 0.005 for the difference between MG132 plus Bryo and MG132 plus PMA at 16 h). The failure of proteasome inhibition to affect the downregulation of internalized protein seen at 2 h (Fig. 1C, panels d and e, and Fig. 1D, compare panels b and d) together with the delayed kinetics of down-regulation of internalized protein relative to membrane-associated enzyme (Fig. 1D) point to the perinuclear region as the site for this slower, non-proteasomal mechanism of agonist-induced degradation.
In contrast to our findings, studies using overexpressed exogenous PKC␣ have pointed to a model in which PKC agonists, including PMA, target internalized enzyme for proteasomal degradation (19,42). The inconsistency between our results and this model raised the possibility that overexpression leads to an altered response of PKCs to agonist treatment. To determine if this is the case, we compared the effects of Bryo and PMA on the subcellular localization of PKC␣ in IEC-18 cells infected with control LacZ-or PKC␣-expressing adenovirus. As expected, Bryo induced translocation of endogenous PKC␣ to both the plasma membrane and the perinuclear region in LacZ-adenovirus infected cells (Fig. 1E, panel c, arrowheads and arrows, respectively), whereas PMA targeted the enzyme exclusively to the plasma membrane (Fig. 1E, panel g, arrowheads). Plasma membrane staining was markedly diminished by 2 h (Fig. 1E, panels e and i), consistent with the previously observed rapid proteasomal degradation of endogenous protein at this location ( Fig. 1C) (31). In contrast, the perinuclear staining seen in Bryo-treated cells was still clearly evident at 2 h (Fig. 1E, panel e, arrows), reflecting the slower, likely non-proteasomal, degradation of internalized protein (31). The fact that both the relocalization and kinetics of degradation paralleled those seen in non-infected cells indicates that adenoviral infection does not in itself affect agonist-induced processing of PKC␣. Bryo similarly promoted plasma membrane and perinuclear accumulation of overexpressed PKC␣ (Fig. 1E, panel d, arrowheads and arrows). However, membrane staining was still evident at 2 h (Fig. 1E, panel f), suggesting saturation of the translocation and/or degradation systems by the overexpressed protein.
Notably, the subcellular distribution of overexpressed protein in PMA-treated cells was markedly different from that seen for the endogenous protein. In addition to translocation to the plasma membrane ( Fig. 1E, panels h and j, arrowheads), exogenous PKC␣ also localized to the perinuclear region (Fig. 1E, panels h and j, arrows). This intracellular accumulation was evident by 30 min, a time at which agonist-induced down-regulation of the endogenous protein is minimal and remained prominent at 2 h, by which time the endogenous protein was extensively down-regulated in PMA-treated cells. Because overexpression alters both the kinetics of degradation and the intracellular trafficking of activated PKC␣, it would appear that the normal agonist-induced processing mechanisms for PKC␣ are easily saturated and that overexpressed protein can be channeled into alternate pathways.
Collectively, these data indicate that, although overexpression systems have yielded valuable information regarding the regulation of PKCs, a true understanding of the mechanisms involved in signal termination necessitates analysis of the endogenous protein. As discussed above, this approach has revealed the existence of a previously unrecognized non-proteasomal mechanism of agonist-induced PKC␣ degradation that appears to occur in a perinuclear compartment. However, the following questions regarding this pathway remained unanswered; what is the mechanism of Bryo-induced PKC␣ perinuclear accumulation? and what is the mechanism of proteasomeindependent agonist-induced PKC␣ degradation?

Bryostatin-induced PKC␣ Perinuclear Accumulation Involves
Endocytic Trafficking-Two possible mechanisms could explain the perinuclear accumulation of PKC␣ seen after Bryo treatment: (a) simultaneous targeting of PKC␣ to the plasma membrane and a perinuclear compartment or (b) initial targeting of the enzyme to the plasma membrane followed by engagement of an endocytic pathway and translocation to the interior of the cell (39,47). Because all endocytic pathways are temperaturesensitive, we manipulated the incubation temperature to explore the role of endocytosis in perinuclear compartmentalization of PKC␣. As shown in Fig. 2, treatment of IEC-18 cells with Bryo at 4°C for 2 h resulted in recruitment of the entire pool of cellular PKC␣ to the plasma membrane (see Fig. 2, panel e). Perinuclear accumulation of the enzyme was only observed after the cells were shifted to 37°C (Fig. 2, panel f). Incubation at 4°C did not affect the diffuse cytoplasmic distribution of PKC␣ in control cells (Fig. 2, panel d), and the enzyme did not relocalize to the plasma membrane when the cells were treated with Bryo at 37°C before being shifted to 4°C (Fig. 2, panel c), excluding a direct effect of low temperature on PKC␣ localization. These data support a mechanism in which PKC␣ is first recruited to the plasma membrane and then translocated to a perinuclear compartment via an endocytic pathway.

Bryostatin-induced PKC␣ Internalization Is Clathrin-independent but Dependent on Cholesterol and Sensitive to
Genistein-Internalization occurs via two main pathways, clathrin-dependent and clathrin-independent lipid raft-dependent endocytosis (48,49) (Table 1). Previous studies of overexpressed (7,39) or endogenous (31,42) protein have indicated that PKC intracellular trafficking can involve clathrin-dependent (7,42) or lipid raft/caveolae-dependent pathways (31,39). To determine whether Bryo-induced intracellular trafficking of PKC␣ involves a clathrin-dependent pathway, clathrin heavy chain expression was silenced in IEC-18 cells using siRNA. Knockdown of clathrin heavy chain, which potently disrupts clathrin-dependent endocytosis (50), was confirmed by Western blotting and immunofluorescence analysis (Fig. 3A, i and ii, panels a and b). As shown in Fig. 3Aii, panels c and d), siRNA-mediated knockdown of clathrin heavy chain completely abrogated clathrin-dependent uptake of transferrin (51,52); however, clathrin deficiency failed to inhibit perinuclear accumulation of PKC␣ in Bryo-treated cells (Fig.  3Aiii, arrows). Consistent with this finding, inhibition of clathrin-dependent intracellular trafficking using potassium depletion (53,54) also had no effect on Bryo-induced internalization (data not shown).   (100 nM Having excluded a role for a clathrin-dependent pathway, we next explored the involvement of clathrin-independent, lipid raft-dependent mechanisms (48,49). Because cholesterol is known to be an important component of clathrinindependent endocytic pathways (55-57) ( Table 1), we tested the cholesterol dependence of Bryo-induced PKC␣ intracellular trafficking by cholesterol depletion and rescue experiments. At concentrations of Յ100 M, the cholesterol sequestering agent nystatin has been shown to block lipid raft-dependent endocytosis while not affecting clathrincoated pits or clathrin-mediated endocytosis (58 -61). This selectivity was confirmed in IEC-18 cells by the inability of nystatin to affect the uptake of transferrin (data not shown). Our previous studies had determined that Bryo-induced internalization of PKC␣ is nystatin-sensitive (31); however, the dependence of this effect on cholesterol was not examined. As seen previously, nystatin treatment completely inhibited Bryo-induced perinuclear accumulation of PKC␣ (Fig. 3Bi, panel d). The ability of excess cholesterol (0.5 mM) to rescue the effect (Fig. 3Bi, panel e) established the cholesterol dependence of PKC␣ internalization. Notably, nystatin treatment also resulted in partial inhibition of Bryo-induced degradation of the enzyme (Fig. 3Bii, p ϭ 0.004), indicating that the cholesterol-dependent internalization pathway directly contributes to PKC␣ degradation after Bryo treatment. Together, the data indicate that Bryo-induced PKC␣ trafficking and subsequent degradation involves a clathrinindependent, lipid raft-dependent mechanism(s) (48,49).
Clathrin-independent/lipid raft-dependent endocytic pathways have also been shown to be sensitive to the isoflavonoid genistein (48, 56 -61) (Table 1), and this agent has been widely used to differentiate between clathrin-dependent and -independent uptake mechanisms (cf. Refs. 48, 49, 62, and 63-66). As shown in Fig. 3C, concentrations of genistein as low as 40 -80 M caused a reduction in Bryo-induced translocation of PKC␣ from the plasma membrane. However, consistent with the concentration of 400 M required for optimal inhibition of clathrin-independent uptake of lactosylceramide in a variety of cell types (48), full blockade of perinuclear accumulation required 300 -600 M genistein. Together, these data indicate that Bryo-induced PKC␣ intracellular trafficking is sensitive to cholesterol depletion and genistein treatment and are, thus, consistent with the involvement of a clathrin-independent/lipid raft-dependent pathway(s).

Clathrin-independent Endocytosis of Endogenous PKC␣ Occurs via Dynamin-dependent and -independent Pathways
and Is Independent of Small GTPase Activity-Clathrin-independent, lipid raft-dependent pathways can be classified based on (a) their requirement for dynamin, which is responsible for scission of vesicles from the plasma membrane during endocytosis (67) and (b) the involvement of the small GTPases, RhoA, ADP-ribosylation factor-6 (Arf6) or Cdc42 (68) ( Table 1). Clathrin-independent, dynamin-dependent pathways include caveolae-mediated endocytosis and RhoA-regulated endocytosis; however, Cdc42-and Arf6-regulated pathways are dynamin-independent (48,49). Two approaches were used to determine the dynamin dependence of PKC␣ internalization: (a) pharmacological inhibition of dynamin activity with dynasore (69,70) and (b) expression of a dominant-negative K44A dynamin mutant (67). The effectiveness of these strategies in blocking dynamin-dependent processes was confirmed by their ability to completely block clathrin-mediated uptake of transferrin and/or caveolar uptake of cholera toxin (Fig. 4A, panels e and f, and B, panels e-l), both of which are dynamin-dependent (45,69,71). Treatment of IEC-18 cells with dynasore consistently, albeit only partially, reduced Bryo-induced PKC␣ internalization. As shown in Fig. 4A, panel d, levels of PKC␣ at the plasma membrane were higher in the presence of dynasore (arrowheads), but perinuclear staining was still evident (arrows). Transfection of cells with the K44A dynamin mutant also led to a partial inhibition of PKC␣ internalization, as indicated by increased retention of the protein at the plasma membrane (Fig. 4B, arrowheads). Because the dynamin-dependent uptake was completely blocked under these conditions, these data indicate that Bryo-induced PKC␣ intracellular trafficking involves both dynamin-dependent and -independent mechanisms.
The dependence of PKC␣ trafficking on small GTPases was determined using C. difficile Toxin B, which inhibits RhoA, Rac1, and Cdc42 function through monoglucosylation of a threonine residue in their nucleotide binding domain (72)(73)(74)(75)(76). Functional blockade of these molecules results in actin condensation and consequent cell retraction and rounding (76,77). These effects could be seen in IEC-18 cells, with concentrations as low as 10 ng/ml inducing cell retraction by 30 min and extensive rounding by 2 h (e.g. Fig. 5A, panels c and g). As shown in Fig. 5A, up to 100 ng/ml Toxin B did not inhibit Bryo-induced plasma membrane translocation (panel d, arrowheads) or    (Fig. 5B, small closed arrows). Taken together, the data indicate that Bryo promotes perinuclear accumulation of PKC␣ via at least two clathrin-independent, lipid raft-dependent pathways: a dynamin-dependent pathway, which likely involves caveolae, and a non-caveolar, small GTPase-independent pathway (cf. Table 1).

Bryostatin-induced PKC␣ Intracellular Trafficking and Nonproteasomal Degradation Does Not Require Ubiquitination-
Recent studies have indicated that monoubiquitination of proteins can provide a signal for endocytosis and intracellular trafficking (79). To determine if Bryo-induced PKC␣ internalization is dependent on ubiquitination, we tested the effects of the E1 ubiquitin activating enzyme inhibitor, UBEI-41 (80). UBEI-41 effectively blocked ubiquitination in these cells, as indicated by its ability to prevent the proteasomal degradation of PKC␣ induced by PMA and short-term Bryo treatment (Fig.  6A, 3 h; p Ͻ 0.05 for the effects of UBEI-41 on Bryo-and PMA-induced down-regulation). As discussed above, proteasomal degradation of activated PKC␣ occurs at the plasma membrane, and inhibition of the proteasome leads to accumulation of PKC␣ at the cell periphery in agonist-treated cells (Fig. 1, B  and C). Thus, consistent with its ability to block proteasomal processing of PKC␣, UBEI-41 also increased the levels of the enzyme at the plasma membrane in Bryo-treated cells (Fig. 6B,  panel d, arrowheads). Notably, however, inhibition of ubiquitination failed to prevent Bryo-induced PKC␣ intracellular trafficking after 2 h of treatment (Fig. 6B, panel d, arrows). Although UBEI-41 showed considerable toxicity with prolonged exposure, it also failed to block the eventual degradation of PKC␣ after long-term Bryo treatment (Fig. 6A, 16 h). The fact that similar levels of PKC␣ were localized to the perinuclear region in the presence and absence of UBEI-41 (Fig. 6B,  panels c and d, arrows) confirms that trafficking of PKC␣ is not noticeably affected by the absence of ubiquitination.

Identification of an Endolysosomal Mechanism of Degradation for Endogenous PKC␣
Bryostatin Induces Trafficking of PKC␣ through the Endolysosomal Pathway-Bryo engages at least two pathways of PKC␣ degradation in IEC-18 cells; that is, a proteasomal pathway that predominates at earlier times of treatment (Ͻ4 h) and a nonproteasomal pathway that is clearly evident in the presence of inhibitors of the proteasome or of ubiquitination (see Figs. 1 and 6). A major mechanism of protein degradation in cells is the endolysosomal pathway. Based on the finding that Bryo promotes endocytosis of PKC␣, endosome/lysosome disrupting agents were used to examine the role of this pathway in nonproteasomal processing of the protein.
In initial studies, we tested (a) the lysosomotropic agent chloroquine, a weak base that inhibits endosomal/lysosomal function by raising the pH of acidic vesicular compartments and inducing vesicle swelling (81)(82)(83)(84)(85)(86) and (b) the amphiphilic amine U18666A, which promotes cholesterol accumulation and swelling in late endosomes and lysosomes, thus altering the trafficking of multivesicular body-associated proteins (87)(88)(89). Immunofluorescence analysis of PKC␣ localization in cells treated with Bryo in the presence of chloroquine clearly demonstrated association of the enzyme with the membrane of swollen vesicles, indicating that internalized endogenous PKC␣ is targeted to the endolysosomal pathway (Fig. 7A, panels b and d, arrows). Similarly, Bryo-stimulated PKC␣ was detected in swollen vesicular structures in U18666A-treated cells (Fig. 7A, panels c and e, arrows).
To identify the PKC␣-associated structures in chloroquineand U18666A-treated cells, dual localization experiments were performed using various markers of specific endolysosomal compartments. EEA1 identifies early endosomes, whereas Rab7 is a marker for late endosomes and multivesicular bodies (90,91). Co-localization of PKC␣ and EEA1 was detected in chloroquine-treated cells (Fig. 7Bi, arrows), indicating that Bryo induces trafficking of PKC␣ to early endosomes. Similarly, colocalization with transfected dominant negative Rab7-GFP (T22N-Rab7-GFP) demonstrated that Bryo-stimulated PKC␣ accumulates in late endosomes/multivesicular bodies in the presence of U18666A (Fig. 7Bii). The association of PKC␣ with early and late endosomes was confirmed by confocal microscopy, which detected colocalization of PKC␣ with EEA1 and T22N-Rab7-GFP in cells that had not been treated with either chloroquine or U18666A (Fig. 7C, i and ii). Notably, confocal analysis also detected association of PKC␣ with lysosomes, as indicated by colocalization of the enzyme with the lysosomal marker LAMP1 (Fig. 7Ciii). Taken together, these findings indicate that Bryo induces trafficking of PKC␣ through early and late endosomal compartments and that the enzyme is ultimately delivered to lysosomes in the perinuclear region.
Bryostatin Induces Endolysosomal Processing of PKC␣ in a Variety of Cell Types-Treatment of IEC-18 cells with a panel of agents that disrupt endolysosomal proteolytic function conferred partial protection from Bryo-induced degradation of endogenous PKC␣ (Fig. 8A; p Ͻ 0.05 for the effects of all inhibitors), providing further evidence that internalized PKC␣ is delivered to the lysosome for degradation. In addition to the lysosomotropic agent chloroquine and U18666A (see Fig. 8A, panels i and iii), significant protection was observed with (a) the macrolide antibiotics concanamycin A and bafilomycin A1, which potently and specifically inhibit vacuolar proton-ATPases but do not cause vesicular swelling (92,93), and (b) leupeptin, which inhibits lysosomal peptidases (84,94,95) (Fig.  8, A, panels ii and iv, and B). This disruption cannot be attributed to blockade of the intracellular trafficking of PKC␣ as observed in RBL-2H3 cells (42), as the protein was translocated to the perinuclear region in cells treated with chloroquine, U18666A, and leupeptin (Figs. 7B and 8B). Furthermore, it cannot be attributed to indirect inhibition of proteasomal degradation because lysosomal agents significantly (p Ͻ 0.02) protect PKC␣ from Bryo-induced degradation in cells treated with proteasome inhibitors: e.g. chloroquine with ALLN (Fig. 8A, panel  i), concanamycin A with lactacystin (panel ii), and leupeptin with MG132 (panel iv). Finally, lysosomal degradation appears to be a major mechanism of PKC␣ processing in the perinuclear region, as indicated by the ability of leupeptin to rescue perinu- clear PKC␣ from degradation while having no protective effects on plasma membrane-associated enzyme (Fig. 8B).
In previous studies we demonstrated that Bryo induces limited accumulation of faster migrating species of PKC␣ in the perinuclear region, which result from a combination of delayed maturation and dephosphorylation of the protein (Figs. 1, 3Bii,  and 8A, arrows, and Ref. 31). As shown in Fig. 3Bii, formation of this species is inhibited by nystatin, which prevents intracellular accumulation of the enzyme. Although this faster migrating species is a target for proteasomal degradation (23), as indicated by its protection by proteasome inhibitors (Fig. 8A), lysosomal agents predominantly protect the slower migrating, fully phosphorylated form of the enzyme (Fig. 8A, arrowheads), indicating that this is a major target for lysosomal degradation. Furthermore, the degree of protection provided by endolysosomal disrupting agents in cells treated with proteasome inhibitors indicates that lysosomal mechanisms represent the predominant mechanism of non-proteasomal Bryo-induced degradation of PKC␣.
To exclude the possibility that endolysosomal processing of PKC␣ in IEC-18 cells is a cell type-specific effect, the ability of endolysosomal disrupting agents to protect PKC␣ from Bryoinduced degradation was explored in L929 murine fibroblasts, U-87 human glioblastoma cells, and HeLa human cervical cancer cells. Immunofluorescence analysis revealed that Bryo induces perinuclear accumulation of PKC␣ in all of these cell lines (data not shown). Importantly, lysosome disrupting agents such as chloroquine and bafilomycin were capable of significantly (p Ͻ 0.02) protecting endogenous PKC␣ (predominantly the mature form) in each of these cell types. Indeed, in some cases protection was more effective than in IEC-18 cells (Fig. 8C). Thus, both lysosomal and proteasomal pathways of PKC␣ degradation exist in different cell types, and the extent to which they are engaged by PKC agonists appears to be cell typedependent as well as agonist-dependent.

DISCUSSION
Although it has long been recognized that prolonged agonist stimulation leads to degradation of PKC isozymes, understanding of the molecular mechanisms involved remains limited (19,34). Furthermore, much of our knowledge has come from the use of overexpression systems, which can result in channeling of PKCs into alternate pathways (Fig. 1E). In previous studies of the effects of agonists on endogenous PKC␣ (31), we established that at least two pathways of degradation of the activated protein can co-exist in cells and demonstrated that a single PKC activator can engage both mechanisms at the same time (e.g. Bryo). These enzyme processing pathways involve proteasomedependent and -independent mechanisms. In the current study we advance the understanding of PKC signal termination by demonstrating that (a) proteasome-dependent and -independent PKC␣ degradation largely occur in different cellular compartments, (b) activated PKC␣ is directed to the perinuclear region of cells via dynamin-dependent and -independent lipid raft-mediated pathways, and (c) perinuclear PKC␣ is mainly degraded in lysosomes. Our data also indicate that caution should be exercised in interpreting findings from studies using overexpressed enzyme.
The current study clearly demonstrates that plasma membrane-associated PKC␣ is a substrate for proteasomal processing. PMA and Bryo both target the entire pool of PKC␣ to the plasma membrane, and the activated enzyme undergoes rapid ubiquitination and proteasome-mediated degradation in this compartment (Figs. 1C and 6B). Blockade of ubiquitination using the E1 ubiquitin-activating enzyme inhibitor UBEI-41 or of the proteasome with proteasome inhibitors rescues the membrane-bound enzyme (Figs. 1 and 6). In IEC-18 cells, this is the sole pathway engaged by PMA and is the predominant processing mechanism at early times of Bryo treatment.
The second pathway, engaged by Bryo but not PMA in IEC-18 cells, involves temperature-sensitive and cholesteroldependent trafficking of active PKC␣ to a perinuclear compartment. This slower pathway is clearly evident in the context of proteasome inhibition, which does not affect internalization of the enzyme. Use of a panel of endolysosome disrupting agents and lysosome protease inhibition provided the first demonstration that PKC␣ can undergo endolysosomal processing, a finding that was further supported by immunofluorescence detection of the enzyme in EEA1-, Rab7-, and LAMP1-positive vesicles (Figs. 7, B and C). Several lines of evidence support an intracellular location for non-proteasomal degradation of PKC␣. First, cholesterol depletion, which inhibits perinuclear accumulation of PKC␣, delays its down-regulation (Fig. 3Bii). Second, the slower kinetics of degradation of perinuclear PKC␣ are consistent with involvement of slower lysosomal processing. Unlike plasma membrane-associated enzyme, which is largely down-regulated by 1 to 2 h of PKC agonist treatment, intracellular PKC␣ persists for longer than 3 h (Fig. 1D). Finally, lysosomotropic agents and leupeptin protect perinuclear PKC␣ from degradation while having no effect on plasma membraneassociated enzyme. Endolysosomal processing does not appear to be a secondary pathway that operates only when the proteasome is impaired or inhibited as (a) endolysosomal blockade could partially rescue PKC␣ from Bryo-induced down-regulation in the presence of a functional proteasome, and (b) combined use of endolysosomal and proteasome inhibitors provided enhanced protection. Rather, the proteasomal and endolysosomal mechanisms represent independent pathways of PKC␣ processing that largely occur at different cellular locations, i.e. the plasma membrane and perinuclear region, respectively.
Importantly, lysosomal processing appears to be a common mechanism for degradation of endogenous PKC␣, as indicated by the protective effects of endolysosomal inhibitors in a variety of cell types (Fig. 8C). Furthermore, in several cell types this pathway seems to be a major mechanism of activation-induced down-regulation of endogenous PKC␣. Although PKC signaling has been implicated in regulation of endolysosomal processing of other proteins (e.g. Refs. 96 and 97-100), to our knowledge this is the first report of a direct role for lysosomal degradation in agonist-induced down-regulation of PKC␣ itself. The failure to detect this pathway in other studies may be attributed to (a) the fact that it is masked at early time points by more rapid proteasomal mechanisms (Fig. 1), (b) the common use of overexpressed, exogenous proteins in studies of PKC down-regulation, which may divert the protein into alternate pathways, (c) agonist-specific differences in the engagement of different pathways, and/or (d) cell type differences.
Further insight into the mechanisms controlling lysosomal processing of activated PKC␣ came from analysis of the pathways mediating Bryo-induced perinuclear accumulation of the enzyme. Immunofluorescence analysis demonstrated that PKC␣ is targeted to EEA1-, Rab7-, and LAMP1-decorated vesicles in cells treated with Bryo; thus, perinuclear accumulation of the enzyme involves trafficking through early and late endosomes/multivesicular bodies, with ultimate delivery to lysosomes. Consistent with this finding, Bryo-induced internalization of PKC␣ is distinct from the previously described intracellular targeting of this isozyme to a recycling (Rab11positive) compartment termed the pericentrion (7); whereas the latter is clathrin-dependent, siRNA-mediated knockdown of clathrin heavy chain had no effect on Bryo-induced perinuclear translocation of PKC␣ (Fig. 3A). Furthermore, Bryo-induced PKC␣ trafficking was blocked by cholesterol depletion and genistein treatment, pointing to a clathrin-independent, lipid raft-dependent mechanism(s) (48,49). Use of a combination of pharmacological inhibitors and dominant negative mutants led to the identification of two distinct pathways of PKC␣ trafficking that differ in their dependence on dynamin (Fig. 4). Because involvement of RhoA was excluded (Fig. 5), the dynamin-dependent pathway likely involves caveolae (48,49). Although this caveolar pathway bears similarities to the dynamin-dependent pathway of PKC␣ translocation and degradation previously identified in PMA-treated PKC␣-overexpressing MCF7 cells (39), a Rab7-positive late endosomal compartment was not implicated in that study. This difference could reflect the involvement of distinct caveaolar pathways in the two systems or to delivery of PKC␣ to late endosomes by the dynamin-independent pathway identified in IEC-18 cells. Although dynamin-independent endocytic pathways remain poorly characterized, the recently identified flotillin-associated pathway (101) represents a candidate mechanism for perinuclear accumulation of PKC␣. The potential involvement of the enzyme in flotillin-associated events is supported by reports that (a) PKC␣ and flotillin are colocalized on mature phagosomes (102), (b) flotillin-1 is required for PKC-mediated endocytosis of the dopamine transporter (103), and (c) both PKC␣ and flotillin have been implicated in regulation of ␤1 integrin trafficking (104,105). Although the precise molecular characteristics of dynamin-independent PKC␣ internalization remain to be fully characterized, our data collectively indicate that Bryo-induced trafficking of PKC␣ to the perinuclear region in IEC-18 cells (a) involves transit through early and late endosomes and (b) occurs via at least two pathways, caveolar endocytosis and a lipid raft-mediated, non-caveolar, dynamin-and small GTPase-independent mechanism. To our knowledge this is the first time that dynamin-independent mechanisms have been implicated in PKC isozyme intracellular trafficking.
A number of studies have pointed to a model of PKC/PKC␣ down-regulation involving dephosphorylation and proteasomal degradation of internalized protein (e.g. Refs. 19, 34, 39, and 42). Consistent with this model, Bryo does induce some dephosphorylation of PKC␣ (31) after its accumulation in the perinuclear region, as confirmed by the ability of nystatin to inhibit the generation of a dephosphorylated species in IEC-18 cells (Fig. 3Bii). This dephosphorylated species is slightly protected by proteasome inhibitors (Fig. 1B), indicating that the internalization 3 dephosphorylation 3 proteasome-mediated degradation pathway is operative in IEC-18 cells. However, the failure of chloroquine or U18666A to promote significant accumulation of a non-phosphorylated species indicates that the internalized protein remains phosphorylated at least as far as late endosomes/multivesicular bodies (Figs. 7 and 8). Furthermore, there is minimal accumulation of non-phosphorylated PKC␣ in the absence or presence of proteasome and lysosome inhibitors even after prolonged Bryo treatment (e.g. 16 -24 h; Figs. 1 and 8), with these inhibitors predominantly protecting the mature, phosphorylated species (e.g. Fig. 8). Thus, it is the phosphorylated form of PKC␣ that is targeted for endosomal trafficking to the lysosome. A role for fully phosphorylated PKC␣ as a target for degradation is consistent with findings with other PKC isozymes (e.g. PKC␦; Ref. 106) and with the observation that PMA induces internalization of phosphorylated PKC␣ in MCF7 cells (39).
In summary, our findings (a) provide additional evidence for the existence of multiple pathways of PKC␣ desensitiza-tion that can occur in different cellular compartments, (b) identify a novel pathway of PKC␣ internalization that is clathrin-independent and does not involve caveolae, and (c) show that activated PKC␣ traffics through early and late endosomal compartments to be degraded by the lysosomal machinery.