Regulation of receptor-mediated protein kinase C membrane trafficking by autophosphorylation.

Signal transduction via protein kinase C (PKC) is closely regulated by its subcellular localization. In response to activation of cell-surface receptors, PKC is directed to the plasma membrane by two membrane-targeting domains, namely the C1 and C2 regions. This is followed by the return of the enzyme to the cytoplasm, a process shown recently to require PKC autophosphorylation (Feng, X., and Hannun, Y. A. (1998) J. Biol. Chem. 273, 26870-26874). In the present study, we examined mechanisms of translocation and reverse translocation and the role of autophosphorylation in these processes. By visualizing the trafficking of wild-type as well as mutant PKCbetaII in live cells, we demonstrated that in response to cell-surface receptor activation, the function of the C1 region is required but not sufficient for recruitment of the enzyme to the plasma membrane. The C2 region is also critical in anchoring the enzyme to the plasma membrane. Furthermore, the inability of a kinase-deficient PKC to undergo reverse translocation was restored by the addition of intracellular calcium chelators, suggesting a role for the C2 region in the persistent phase of translocation. On the other hand, the inability of a C2 deletion mutant (C1 region intact) to translocate in response to agonist was reversed in mutants lacking kinase activity or by mutation of the Ser(660) autophosphorylation site to alanine, suggesting that autophosphorylation of this site is required for opposing the action of the C2 region. Therefore, the membrane-targeting function of the C1 region is facilitated by the C2 region and appears to be opposed by autophosphorylation. Taken together, these findings provide novel evidence of the functional regulation of reversible PKC membrane localization by autophosphorylation, and they show that the dynamic translocation of PKC in response to agonists is tightly regulated in a collaborative fashion by the C1 and C2 regions in balance with the effects of autophosphorylation.

Signal transduction via protein kinase C (PKC) is closely regulated by its subcellular localization. In response to activation of cell-surface receptors, PKC is directed to the plasma membrane by two membranetargeting domains, namely the C1 and C2 regions. This is followed by the return of the enzyme to the cytoplasm, a process shown recently to require PKC autophosphorylation (Feng, X., and Hannun, Y. A. (1998) J. Biol. Chem. 273, 26870 -26874). In the present study, we examined mechanisms of translocation and reverse translocation and the role of autophosphorylation in these processes. By visualizing the trafficking of wild-type as well as mutant PKC␤II in live cells, we demonstrated that in response to cell-surface receptor activation, the function of the C1 region is required but not sufficient for recruitment of the enzyme to the plasma membrane. The C2 region is also critical in anchoring the enzyme to the plasma membrane. Furthermore, the inability of a kinase-deficient PKC to undergo reverse translocation was restored by the addition of intracellular calcium chelators, suggesting a role for the C2 region in the persistent phase of translocation. On the other hand, the inability of a C2 deletion mutant (C1 region intact) to translocate in response to agonist was reversed in mutants lacking kinase activity or by mutation of the Ser 660 autophosphorylation site to alanine, suggesting that autophosphorylation of this site is required for opposing the action of the C2 region. Therefore, the membranetargeting function of the C1 region is facilitated by the C2 region and appears to be opposed by autophosphorylation. Taken together, these findings provide novel evidence of the functional regulation of reversible PKC membrane localization by autophosphorylation, and they show that the dynamic translocation of PKC in response to agonists is tightly regulated in a collaborative fashion by the C1 and C2 regions in balance with the effects of autophosphorylation.
Protein kinase C (PKC) 1 represents a family of serine/thre-onine kinases whose activity is dependent on lipid cofactors and regulators (1). Members of the PKC family are involved in transducing a large variety of signals, including those from cell-surface receptors for hormones, neurotransmitters, and growth factors. These receptors trigger the activation of phospholipase C, which, upon activation, leads to the generation of two important second messengers, diacylglycerol (DAG) and inositol trisphosphate. Inositol trisphosphate elevates intracellular Ca 2ϩ by binding to its intracellular receptors (1)(2)(3). The increased intracellular Ca 2ϩ and membrane DAG promote the trafficking of PKC to the plasma membrane and its subsequent activation (1,4). By catalyzing phosphorylation of various substrates, PKC has been shown to play important roles in a diverse variety of biological processes, including cytoskeletal function, secretion, cell proliferation, differentiation, and gene expression (1,5,6).
Extensive biochemical and immunocytochemical studies have indicated that the biological activity of PKC is intimately regulated by its subcellular localization (7,8). By localizing to specific subcellular locations, PKC effectively responds to its physiological activators such as membrane lipids and gains access to its specific substrates. Activation of the plasma membrane receptor triggers the translocation of PKC from the cytosol to the plasma membrane, resulting from changes in the cellular levels and intracellular localization of Ca 2ϩ and DAG. This is mediated by the amino-terminal regulatory domain in classic PKCs (i.e. PKC␣, -␤I, -␤II, and -␥), which contains an autoinhibitory sequence (i.e. the pseudosubstrate in the V1 region) and two membrane-targeting protein modules, namely the C1 and C2 regions (see Fig. 1) (9). The C1 region contains the binding site for DAG in a phosphatidylserine-dependent mechanism, whereas the C2 region interacts with anionic phospholipids whose binding is allosterically facilitated by the association of Ca 2ϩ with the same region. The interaction of PKC with Ca 2ϩ results in the displacement of the pseudosubstrate in the V1 region from the enzyme catalytic domain, relieving autoinhibition (10). It is believed that the C1 and C2 regions, in the presence of Ca 2ϩ , DAG, and other membrane components, provide membrane-binding sites with very high affinity that recruit the enzyme to the plasma membrane and allow for the maximal activation of PKC (11). Nevertheless, the detailed molecular events that lead to PKC membrane mobilization in response to cell-surface receptor activation have not been fully understood. In particular, it is not clear what specific role the C1 and C2 regions each play in this process.
Increasing lines of evidence support that the phosphorylation of PKC itself is critical for the regulation of its cellular distribution and enzymatic activity (7,12). It has been known for many years that PKC is phosphorylated at multiple serine/ threonine sites during its biogenesis (13,14). In the case of PKC␤II, three major in vivo phosphorylation sites, Thr 500 , Thr 641 , and Ser 660 , have been identified (15)(16)(17). The inability of PKC␤II itself to phosphorylate at Thr 500 has suggested that the phosphorylation of this residue involves a protein kinase distinct from PKC (18). In contrast, Thr 641 and Ser 660 are identified as the residues that are sequentially autophosphorylated following phosphorylation of Thr 500 during PKC maturation (16). Autophosphorylation of PKC has been shown to regulate the enzyme regulatory and catalytic domains by enhancing their binding affinity for Ca 2ϩ and ATP, as well as peptide substrates (19). In addition, it has also been shown to be important for PKC dynamic cellular trafficking, probably by enhancing reverse translocation (20). Therefore, like DAG, Ca 2ϩ , and many other lipid mediators, autophosphorylation provides another effective way to regulate the activity and localization of PKC, but at the level of the enzyme itself.
In previous studies, by visualizing real time cellular trafficking of PKC␤II conjugated to green fluorescent protein (GFP) in live cells, we showed that PKC undergoes a dynamic and reversible redistribution between the plasma membrane and cytoplasm in response to physiological stimuli such as activation of cell-surface G protein-coupled receptors (GPCRs) (21). Shortly after its membrane translocation, PKC dissociates from the plasma membrane and rapidly returns to the cytoplasm. This is different from the effects of phorbol esters such as PMA or Ca 2ϩ ionophores such as A23187, which promote a stable association of PKC with the membrane (21)(22)(23). In addition, we have demonstrated that the autophosphorylation activity of PKC is essential for its membrane dissociation and therefore plays a key role in the reversible translocation of PKC in response to GPCR activation (20). In the present study, we address how PKC autophosphorylation interacts with membrane targeting through the C1 and C2 regions to achieve a balanced and dynamic movement of PKC in live cells upon activation of the G q ␣-coupled angiotensin II type 1A receptor (AT 1A R). Our results support that the GPCR-promoted PKC membrane trafficking consists of at least three distinct steps, i.e. membrane recruitment, membrane anchorage, and membrane dissociation. The C1 and probably the C2 regions are required for recruiting PKC to the plasma membrane upon receptor stimulation. The C2 region then persistently anchors the enzyme to the membrane. When extracellular signals are turned off, the enzyme returns to the cytoplasm, a process dependent on the autophosphorylation status of PKC. Therefore, although each plays a specific role, the functions of the C1 and C2 regions and autophosphorylation are cooperative and interactive, thus ensuring appropriate timing, duration, and magnitude of dynamic PKC trafficking.

EXPERIMENTAL PROCEDURES
Materials-Eagle's minimal essential medium and HEPES were from Life Technologies, Inc. Human embryonic kidney (HEK) 293 cells were provided by American Type Culture Collection. Rabbit polyclonal antibody against PKC␤II was prepared and extensively characterized as described previously (24). [␥-32 P]ATP was purchased from NEN Life Science Products. Restriction enzymes were from Promega or New England Biolabs Inc. Vent DNA polymerase and rabbit anti-phospho-PKC␤(Ser 660 ) polyclonal antibody were purchased from New England Biolabs Inc. Protein A-Sepharose CL-4B was from Amersham Pharmacia Biotech. Phosphatidylserine and sn-dioctanoylglycerol were purchased from Avanti Polar Lipids, Inc. A23187 was purchased from Calbiochem. BAPTA-AM was from Molecular Probes, Inc. All other chemicals were from Sigma.
Cell Culture and Transfection-HEK 293 cells were maintained in Eagle's minimal essential medium supplemented with 10% (v/v) fetal bovine serum in a 5% CO 2 incubator at 37°C. Cells were seeded at a density of 2.0 ϫ 10 6 cells/100-mm dish and transfected using a modified calcium phosphate method with 1-5 g of plasmids. The expression of the hemagglutinin-tagged AT 1A R in the presence of wild-type or various mutant GFP-PKC␤II constructs was assessed by flow cytometry following antibody staining. The levels of the AT 1A R in cells transfected with various mutant GFP-PKC␤II constructs were either equivalent to or higher than that in the presence of wild-type GFP-PKC␤II.
Protein Kinase C Assay-For assaying PKC activity in cells transfected with wild-type or mutant GFP-PKC␤II, the cell lysates (200 g) were first immunoprecipitated with 5 g of rabbit anti-PKC␤II antibody. The immunoprecipitation was performed 48 h after transfection as described previously (21). The PKC autophosphorylation activity was assessed using the vesicle assay for PKC as described previously (26). The phosphorylated proteins were analyzed by SDS-polyacrylamide gel electrophoresis, followed by autoradiography.
Confocal Microscopy-HEK 293 cells were transfected with GFP-PKC␤II or one of its mutants together with the AT 1A R. Twenty-four hours after transfection, the cells were plated onto 35-mm glass-bottom culture dishes (Mattek Corp.) at a density of 4 ϫ 10 5 and incubated for another 24 h for the cells to attach to glass. The cells expressing GFP-PKC␤II or its mutants were observed by confocal microscopy, which was performed on a Zeiss LSM-410 laser scanning microscope using a Zeiss 40ϫ 1.2 NA water immersion lens. The cells were at room temperature in culture medium containing 20 mM HEPES. GFP fluorescent signals were collected sequentially using the Zeiss LSM software time series function with single line excitation (488 nm) with a time interval of 20 or 60 s between two scannings. Angiotensin II, A23187, PMA, and EGTA/BAPTA-AM were applied to the cells during the scannings. Quantitative analysis of relative fluorescence intensity was performed on a Macintosh computer using the public domain NIH Image program. The relative change in plasma membrane fluorescence intensity was calculated according to a previously reported method (27). Briefly, from a series of images recorded before and after angiotensin II stimulation, line intensity profiles across each cell were determined. The relative fluorescence intensity on the plasma membrane was calculated by the formula (I mb Ϫ I cyt )/I cyt , where I mb is the amplitude of the fluorescence signal on the plasma membrane and I cyt is the average cytosolic fluorescent intensity.

Recruitment of Cytoplasmic PKC to the Plasma Membrane by the C1 Region in Response to Receptor Activation-Previously,
we showed that GFP-PKC␤II is fully functional in terms of its kinase activity and its ability to undergo membrane translocation (21). In addition, the inherent fluorescence of GFP allowed us to visualize the trafficking of GFP-PKC␤II in live HEK 293 cells and to demonstrate that autophosphorylation is essential for the reversibility of PKC membrane translocation in response to GPCR activation (20). To further understand the role of autophosphorylation in association with the membrane-targeting function of the C1 and C2 regions in the dynamic translocation of PKC, a series of GFP-PKC␤II mutants were con-structed ( Fig. 1), and their trafficking properties were studied in live HEK 293 cells.
Initial experiments examined the effects of the C1 region mutation on PKC membrane translocation by altering two essential Cys residues in the two cysteine-rich motifs of the C1 region to Ser (Fig. 1). This C1 region mutant (i.e. GFP-mC1), similar to wild-type GFP-PKC␤II, was evenly distributed in the cytoplasm under normal unstimulated conditions ( Fig. 2A). Consistent with the role of the cysteine-rich motifs of the C1 region in DAG binding, the GFP-mC1 mutant lost the ability to translocate to the plasma membrane upon stimulation by 1 M PMA, a phorbol ester commonly used to mimic the action of DAG in targeting PKC to the membrane, but with higher affinity and longer effects ( Fig. 2A). In contrast, GFP-mC1 underwent substantial membrane translocation in response to 10 M A23187, a Ca 2ϩ ionophore that causes a maximal increase in intracellular concentrations of Ca 2ϩ , indicating that the C2 region remained functional in its ability to associate with the membrane (Fig. 2A). As a control, under parallel conditions, wild-type GFP-PKC␤II responded to both PMA and A23187 ( Fig. 2A). To further explore the function of the C1 region in the reversible membrane trafficking of PKC in response to physiological stimuli, GFP-mC1 localization was examined in HEK 293 cells in response to activation of cotransfected AT 1A R by its physiological agonist angiotensin II. GFP-mC1 was found to lack the ability to undergo membrane redistribution in response to angiotensin II (Fig. 2B), whereas wild-type GFP-PKC␤II underwent a rapid and reversible mem-FIG. 1. Scheme of wild-type and mutant GFP-PKC␤II fusion proteins. GFP-PKC␤II was constructed by inserting a GFP cDNA fragment without a stop codon into pBK-CMV-PKC␤II at the 5Ј-end of human PKC␤II cDNA as described previously (21). GFP-K371R is a kinase-deficient mutant constructed and characterized in a previous study (20). GFP-mC1 and GFP-⌬C2 are GFP-PKC␤II mutants deficient in the C1 and C2 region functions, respectively, as described previously (23,25). The double mutants GFP-KR/mC1 and GFP-KR/⌬C2 are deficient in kinase activity as well as the C1 and C2 region functions, respectively, whereas GFP-⌬C2/SA and GFP-⌬C2/SE are C2 region-deficient mutants with altered autophosphorylation properties. The characteristics of membrane translocation of these PKC␤II mutants is summarized in Table I. PS, phosphatidylserine. brane translocation. When quantitated, Ͼ90% of GFP fluorescence from wild-type GFP-PKC␤II was translocated to the plasma membrane and subsequently returned to the cytoplasm within 100 s of agonist stimulation. In contrast, there was no significant agonist-dependent increase in membrane fluorescence intensity in cells expressing GFP-mC1 (Fig. 2C). As the C2 region of GFP-mC1 was functional ( Fig. 2A), the inability of GFP-mC1 to traffic to the plasma membrane suggests that the C1 region is essential for PKC targeting to the plasma membrane following receptor activation and that the C2 region is not sufficient for membrane translocation in response to physiological agonist.
To exclude the possibility that the inability to observe GFP-mC1 mobilization is due to the masking of membrane translocation by very rapid membrane dissociation, we examined the distribution of a kinase-deficient GFP-mC1 mutant (namely GFP-KR/mC1) because loss of kinase activity prevents membrane dissociation in the wild-type enzyme and therefore would serve to trap any translocated enzyme at the plasma membrane (20). GFP-KR/mC1 was generated from the previously described GFP-K371R (20). Converting the conserved Lys at the enzyme ATP-binding site to Arg completely abolishes the ability of PKC to autophosphorylate. Similar to GFP-mC1, when transfected in HEK 293 cells, GFP-KR/mC1 was found mainly in the cytoplasm, and its redistribution to the membrane was induced only by the Ca 2ϩ ionophore A23187, but not by PMA ( Fig. 2A). Furthermore, when examined in HEK 293 cells cotransfected with the AT 1A R and stimulated with angiotensin II, like GFP-mC1, no significant membrane redistribution of GFP-KR/mC1 was observed (Fig. 2, B and C). These results further confirm that the failure of GFP-mC1 and GFP-KR/mC1 to redistribute to the membrane is due to the inability of their C1 regions to respond to DAG. The inability of these mutants to trigger PKC membrane localization indicates that the recruitment of PKC to the plasma membrane in response to receptor activation requires the function of the C1 region.
Regulation of Persistent PKC Anchorage to the Plasma Membrane by the C2 Region and the Opposing Role of Kinase Activity-To investigate the possible function of the C2 region in the transient PKC membrane redistribution stimulated by receptor activation, two C2 deletion mutants (GFP-⌬C2 and GFP-KR/⌬C2) were generated by removing 80 amino acids from the C2 regions of wild-type GFP-PKC␤II and the kinase-deficient mutant GFP-K371R, respectively. In the unstimulated HEK 293 cells, similar to wild-type GFP-PKC␤II, GFP-⌬C2 and GFP-KR/⌬C2 distributed mainly in the cytoplasm (Fig. 3). Stimulation with 1 M PMA induced a membrane redistribution of both GFP-⌬C2 and GFP-KR/⌬C2, whereas 10 M A23187 failed to trigger redistribution of the two mutants (Fig.  3). This indicates that the C1 regions, but not the C2 regions, in GFP-⌬C2 and GFP-KR/⌬C2 are functional.
Consistent with a role of the C2 region in membrane anchoring, upon deletion of the C2 region from wild-type GFP-PKC␤II, the resulting mutant (GFP-⌬C2) demonstrated a dramatic decrease in its ability to undergo membrane transels), and pBK-CMV-GFP-KR/mC1 (lower panels) before (Control) and after stimulation with 0.5 M angiotensin II (AgII) for the indicated time periods. The experiments were performed independently on three different occasions; and each time, 10 -20 cells from independent stimulations were recorded. The micrographs are representative of Ͼ75% of the cells observed. C, shown are the time courses of the relative fluorescence average changes in GFP-PKC␤II, GFP-mC1, and GFP-KR/ mC1 on the plasma membrane following angiotensin II stimulation. Twelve cells overexpressing GFP-PKC␤II, 12 cells overexpressing GFP-mC1, or 12 cells overexpressing GFP-KR/mC1 from three individual experiments were used to determine the relative membrane fluorescence intensity as described under "Experimental Procedures." location in response to angiotensin II compared with the wild-type enzyme (Fig. 4A). Only very weak membrane highlightings were observed in some cells (Fig. 4A, arrows). These results are quantitated in Fig. 4B. Therefore, similar to the C1 region, the C2 region is necessary but not sufficient for translocation of PKC in response to agonists.
A major difference between the C1 and C2 regions became apparent when the interaction of these regions with the kinase activity of the enzyme was investigated. Upon agonist stimulation of HEK 293 cells cotransfected with the AT 1A R, GFP-KR/⌬C2 was observed to translocate rapidly to the plasma membrane within 40 s (Fig. 4A). However, unlike GFP-K371R, which remained on the membrane, GFP-KR/⌬C2 returned to the cytoplasm within ϳ1 min following its membrane translocation (Fig. 4A). The difference between the two mutants was further quantitated by changes in the membrane fluorescence intensity as a function of agonist exposure time (Fig. 4B). Significantly, 120 s after stimulation by receptor agonist, Ͼ90% of GFP-KR/⌬C2 fluorescence was found to return to the cytoplasm, whereas Ͼ90% of GFP-K371R remained on the plasma membrane (Fig. 4B). Therefore, the GFP-KR/⌬C2 mutant behaved similar to the wild-type enzyme and unlike the KR/mC1 mutant.
These results suggested an important role for the kinase activity of the ⌬C2 mutant. Therefore, it became important to demonstrate that PKC␤II mutants lacking the C2 region were capable of undergoing autophosphorylation. GFP-PKC␤II and GFP-⌬C2 were immunoprecipitated with anti-PKC␤II polyclonal antibody from HEK 293 cells overexpressing these two proteins, and their autophosphorylation activities were assessed. Using an in vitro protein kinase assay, it was found that similar to the wild-type enzyme, GFP-⌬C2 was capable of autophosphorylation (Fig. 5A). More important, when the in vivo autophosphorylation status of the kinases was assessed using anti-phospho-PKC␤(Ser 660 ) antibody, which is specific for phosphorylated Ser 660 , a conserved PKC autophosphorylation site (15)(16)(17), both wild-type GFP-PKC␤II and GFP-⌬C2 were found to be phosphorylated to a similar extent at Ser 660 (Fig. 5B), further indicating that GFP-⌬C2 retains the ability to autophosphorylate in cells. On the other hand, only little Ser 660 phosphorylation was detected in cells overexpressing GFP-K371R proteins (Fig. 5B), a PKC␤II mutant deficient in autophosphorylation (17,20). Moreover, as expected from the nature of the antibody, anti-phospho-PKC␤(Ser 660 ) antibody also failed to detect the mutant GFP-S660A, in which Ser 660 was converted to Ala (data not shown). Therefore, the deletion of the C2 region does not interfere with the ability of the enzyme to autophosphorylate. Given that GFP-KR/⌬C2 be-haves, more or less, similar to the wild-type enzyme, the deletion of the C2 region therefore appears to negate the effects of lack of kinase activity on reversal of translocation, strongly suggesting opposing roles for the C2 region and kinase activity in the regulation of translocation.
It is important to note that the kinase-deficient C1 mutant did not show this behavior (Fig. 2). Therefore, this additional role for the C2 region in the regulation of membrane association was only revealed by comparing the different abilities of GFP-K371R, GFP-KR/mC1, and GFP-KR/⌬C2 to be retained on the plasma membrane following AT 1A R-mediated membrane translocation. Given the role of the kinase activity in reversal of membrane translocation and the above results on the opposing effects of the C2 region and the kinase activity, these results suggest a role for the C2 region in anchoring translocated PKC to the plasma membrane, resulting in the persistent membrane localization of PKC in the absence of autophosphorylation.
To test this hypothesis, the function of the C2 region was further evaluated by adding the Ca 2ϩ chelators EGTA and BAPTA-AM, as the binding of the C2 region to plasma membrane lipids is dependent on Ca 2ϩ . EGTA and BAPTA-AM have been demonstrated as effective Ca 2ϩ chelators that block A23187-and fatty acid-induced GFP-conjugated PKC␥ membrane translocation in live COS-7 cells and CHO-K1, cells respectively (23,28). The specificity of their effect was further shown by their inability to inhibit fatty acid-induced membrane translocation of GFP-conjugated PKC⑀, one of the PKC isoenzymes not regulated by Ca 2ϩ (28). Therefore, the effects of EGTA/BAPTA-AM on the membrane association of GFP-K371R were examined in HEK 293 cells cotransfected with the AT 1A R. As shown previously (20), stimulation of the cells by angiotensin II induced a rapid mobilization of GFP-K371R from the cytoplasm to the plasma membrane, and GFP-K371R remained persistently on the membrane (Fig. 6A). However, when the same cells were exposed to 2.5 mM EGTA and 15 M BAPTA, a rapid return of GFP-K371R to the cytoplasm was observed within 1 min (Fig. 6A). The relative membrane fluorescence intensity was quantitated as shown in Fig. 6B. The addition of EGTA/BAPTA-AM triggered the return of Ͼ90% GFP-K371R fluorescence back to the cytoplasm, whereas Ͼ90% of GFP-K371R fluorescence continued to be present on the plasma membrane in control cells not exposed to the Ca 2ϩ chelators (Fig. 6B). This requirement for Ca 2ϩ in the persistent membrane association of GFP-K371R further demonstrates that the Ca 2ϩ -regulated C2 region plays a critical role in the continuous membrane anchoring of PKC␤II once the enzyme is recruited to the plasma membrane.

Functional Cross-talk of Autophosphorylation with the C1 and C2 Regions in Receptor-induced PKC Membrane Trafficking-
The above results showed an important and specific interaction between the C2 region and the kinase activity of PKC. To address the role of autophosphorylation per se in this interaction, we used GFP-S660A, which is mutated in one of two major in vivo autophosphorylation sites at the carboxyl terminus of PKC␤II. Fig. 7 compares the time course of receptormediated membrane trafficking for wild-type GFP-PKC␤II and GFP-S660A. Similar to GFP-PKC␤II, the mutant GFP-S660A was able to localize to the plasma membrane rapidly in response to angiotensin II stimulation. However, it was apparent that the return of the mutant GFP-S660A from the membrane to the cytoplasm was impaired, suggesting a specific role for this autophosphorylation site in the regulation of reverse translocation.
An additional role for autophosphorylation in regulating PKC membrane targeting was investigated by studying the translocation of the C2 deletion mutant and the combined C2 deletion/S660A mutant. If the lack of GFP-⌬C2 membrane translocation was indeed related to its ability to autophosphorylate, mutation of Ser 660 to alanine in GFP-⌬C2 (i.e. GFP-⌬C2/ SA) should result in an enhanced membrane translocation. If not, the results would point to a role for phosphorylation of different substrates rather than autophosphorylation. In HEK 293 cells cotransfected to express the AT 1A R, GFP-⌬C2/SA was observed to undergo a reversible membrane translocation, unlike GFP-⌬C2, whose distribution was unchanged by AT 1A R activation (Fig. 8A). The time frame for the return of GFP-⌬C2/SA appeared to be longer than that of the wild-type enzyme as shown in Fig. 8B, consistent with the finding that GFP-S660A is defective in membrane dissociation (Fig. 7). As a control, mutation of Ser 660 to Glu, which mimics phosphorylated Ser, resulted in a mutant (GFP-⌬C2/SE) that underwent only a very weak membrane redistribution (Fig. 8A), and the phenotype exhibited by GFP-⌬C2/SE was closer to that of GFP-⌬C2 than to that of GFP-⌬C2/SA (Fig. 8B). At the peak of membrane translocation (t ϭ 60 s) in response to AT 1A R activation, the average relative membrane fluorescence intensity for GFP-⌬C2/SA was much higher than those for GFP-⌬C2 and GFP-⌬C2/SE Taken together, these results demonstrate that the absence of GFP-⌬C2 membrane trafficking results from the counteraction between the function of the C1 region and autophosphorylation; and, as a consequence, the removal of an autophosphorylation site to reduce membrane dissociation rescues the ability of GFP-⌬C2/SA to translocate to the membrane. The inhibitory effect of autophosphorylation on PKC membrane association is only obvious in the absence of the C2 region. DISCUSSION The real time visualization of GFP-PKC␤II localization in response to AT 1A R activation in live HEK 293 cells has demonstrated that receptor-mediated PKC membrane trafficking is a rapid and reversible process (21). The present study provides new insights into the molecular mechanisms underlying the dynamic nature of PKC redistribution. Our results indicate that the functions of the C1 and C2 regions and autophosphorylation are all essential and collaborate/interact to achieve the dynamic trafficking of PKC. Specifically, the C1 region is required for recruiting the enzyme to the plasma membrane from the cytoplasm in response to activation of cell-surface receptors, but it is not sufficient. The Ca 2ϩ -regulated C2 region then serves to further anchor PKC persistently on the plasma membrane. The dissociation of PKC from the membrane is dependent on autophosphorylation. In addition, autophosphorylation exerts an inhibitory effect on translocation that is most predominant when the C2 region is not functional, suggesting a role for the C2 region in overcoming the inhibitory effects of autophosphorylation on membrane targeting.
PKC membrane translocation has been extensively studied in cells using nonphysiological reagents (11). Phorbol esters such as PMA are commonly used to mimic the action of DAG in inducing PKC translocation from the cytosol to the plasma membrane. Similarly, Ca 2ϩ ionophores such as A23187 have been used to raise intracellular Ca 2ϩ concentrations. Studies using these reagents have suggested that both the C1 and C2 regions independently target the enzyme to the membrane (11,29). However, the high affinity PKC membrane association mediated by PMA and the high Ca 2ϩ concentrations induced by ionophore do not precisely reflect the PKC membrane trafficking properties stimulated by physiological stimuli such as activation of cell-surface receptors. For instance, the reversibility of receptor-induced PKC membrane trafficking cannot be mimicked by PMA or A23187 (21,23).
By visualizing the trafficking of wild-type and mutant GFP-PKC␤II in live HEK 293 cells, we found that although PKC mutants deficient in the C1 region function still responded to Ca 2ϩ ionophore, they lack the ability to undergo membrane redistribution in response to GPCR activation (Table I and Fig.  2). This not only further indicates that PKC membrane translocation induced by receptor activation is different from that induced by PMA or A23187, but also suggests an important physiological role for the C1 region in receptor-mediated targeting of the enzyme to the membrane. It is likely that the targeting of PKC to the plasma membrane by the C1 region is mediated by DAG (in a phosphatidylserine-dependent manner) (30,31); and therefore, these data provide very strong support for a direct signaling effect for DAG on PKC.
On the other hand, the C2 deletion mutant, which has an intact phorbol ester-responsive C1 region, also fails to translocate, thus showing that the C1 region is not sufficient for PKC translocation. Therefore, the C1 region is necessary but not sufficient for membrane translocation in response to agonists.
The C2 region is commonly believed to be another important membrane-targeting module that interacts with multiple intracellular targets, including phospholipids and PKC-binding proteins (32,33). The functional involvement of the C2 region in PKC membrane translocation has been demonstrated in cells by C2 region-derived peptides that effectively inhibit both hormone-induced translocation of PKC␤II and its signaling functions (34). Structural analysis has revealed that two Ca 2ϩ molecules are coordinated by five aspartate residues in the C2 region of PKC␤II (35,36). This may account for the Ca 2ϩ -dependent regulation of the C2 region. In addition, mutational analysis has identified in the C2 region of PKC␣ two basic residues and two hydrophobic residues that may provide electrostatic and hydrophobic interactions with membrane components, specifically anionic phospholipids (37). Interestingly, the residues involved in binding to Ca 2ϩ and phospholipids are localized to the loops that connect the ␤-sheet sandwich structure of the C2 region. However, despite the increasing structural information on the interaction of the C2 region with its targets, the physiological roles of the C2 region in receptorinduced PKC activation have not been fully understood.
In the present study, we also observed that deletion of the C2 region abrogates agonist-induced translocation. Therefore, the C2 region is necessary for translocation. On the other hand, the C1 deletion mutant, which retains an intact ionophore-responsive C2 region, also fails to translocate, showing that the C2 region is not sufficient for agonist-induced membrane translocation.
The more significant and unexpected set of results came from studies with the deletion of the C2 region from GFP-K371R. The GFP-K371R mutant of PKC␤II loses the ability to dissociate from the membrane. The deletion of the C2 region from this mutant restored the ability of the enzyme to undergo membrane dissociation (Table I and Fig. 4). This indicates that the persistent membrane association of GFP-K371R is a consequence of the C2 region function in the absence of kinase activity (probably through autophosphorylation; see below). This was further supported with studies using intracellular calcium chelators that were able to undo the defect of the K371R mutant in reverse translocation, again showing that the persistence of the enzyme on the membrane is due to calcium effects that are mediated through the C2 region. Therefore, the C2 region serves an additional function to anchor PKC to the membrane until autophosphorylation-dependent membrane dissociation occurs.
The role for autophosphorylation per se in receptor-mediated PKC trafficking was explored in the present study in association with the functions of the C1 and C2 region. Interestingly, we found that the membrane targeting and association properties mediated by the C1 and C2 regions are differentially modulated by the ability of the enzyme to autophosphorylate. It was initially surprising to observe that the C1 region of GFP-KR/⌬C2 was able to target the enzyme to the plasma membrane, whereas GFP-⌬C2, also with a functional C1 region, lost the ability to traffic. As the replacement of Lys 371 with Arg at the ATP-binding site abolishes the ability of GFP-KR/⌬C2 to undergo phosphorylation and autophosphorylation, it is possible that the lack of GFP-⌬C2 translocation results from an inhibitory effect of autophosphorylation on the C1 region function. The linkage between autophosphorylation and the C1 region function was further confirmed by the fact that the replacement of an autophosphorylation residue (Ser 660 ) with Ala (but not Glu) rescued the ability of GFP-⌬C2 to undergo membrane redistribution. Furthermore, the functions of autophosphorylation and the C2 region also are counteractive. The persistent membrane association of PKC regulated by the C2 region was most dominant in the absence of autophosphorylation (i.e. GFP-K371R), and the enzyme membrane dissociation mediated by autophosphorylation was most predominant when the C2 region was not functional (i.e. GFP-⌬C2). Taken together, these results provide direct visual evidence supporting the hypothesis that the receptor-mediated dynamic trafficking of PKC requires the functional collaboration of PKC autophosphorylation with the C1 and C2 regions.
A role for autophosphorylation in regulating PKC membrane association is also in agreement with a recent functional study on autophosphorylation of the calcium-activated PKC AplI from Aplysia using an anti-phospho-PKC antibody specific for  Thr 613 , one of the conserved autophosphorylation residues among members of the PKC family (38,39). Autophosphorylation of AplI was found to remove the enzyme from the plasma membrane by decreasing its affinity for calcium. There are two autophosphorylation sites identified in PKC␤II that are well conserved among the different isoenzymes in the PKC family (15,16). These residues are surrounded by several hydrophobic residues to form an autophosphorylation motif (FXXF(S/T)(F/ Y)), one of which has recently been confirmed using anti-phospho-PKC␤(Ser 660 ) antibody (17). Also, it has been suggested that the autophosphorylated serine or threonine residues together with the surrounding hydrophobic residues modulate the membrane affinity of PKC (19,40). In addition, consistent with our observation that the functions of the C1 and C2 regions are regulated by autophosphorylation, there is evidence indicating that autophosphorylated residues at the carboxyl terminus of PKC may be folded very close to the C1 and C2 regions, as the binding affinity of the enzyme for Ca 2ϩ was altered upon replacement of autophosphorylation residues by alanine (19). The potential mechanisms underlying the functions of the autophosphorylated carboxyl terminus are anticipated as either indirect modulation of the conformation of membrane-targeting protein modules or direct involvement as part of the interface between the membrane and the enzyme (19,34). The molecular details associated with the functions of autophosphorylation await further exploration. Combining our findings with the current knowledge of PKC trafficking, we propose the following model (10,11,30). Briefly, the dynamic membrane trafficking of PKC in response to cellsurface receptor activation occurs in three major steps, namely membrane recruitment, persistent membrane anchorage, and membrane dissociation (Fig. 9). In unstimulated cells, PKC is mainly distributed in the cytoplasm with the pseudosubstrate blocking the catalytic site to prevent the enzyme from substrate binding and catalysis. Upon stimulation by extracellular signals such as the binding of seven-transmembrane G q ␣-coupled receptors with their physiological agonists, the cellular levels of DAG and Ca 2ϩ are increased as a result of activation of phospholipase C-mediated signaling pathway by GPCRs. Consequently, PKC is recruited to the plasma membrane by the C1 region via its binding to DAG and the C2 region through binding to calcium. Following the recruitment of PKC to the membrane, the high affinity membrane anchorage of the recruited enzyme involves the calcium-dependent function of the C2 region or is achieved by the collaboration of the C1 and C2 regions. One important intrinsic property of the G proteincoupled receptor is that for most GPCRs, agonist activation is followed immediately by desensitization of the receptors through phosphorylation by either G protein-coupled receptor kinases or second messenger-activated kinases (41). This inactivation of the receptors results in the subsequent turning off of phospholipase C and a decrease in the cellular DAG and Ca 2ϩ levels. The decrease in these intracellular PKC activators undermines the function of the C1 and C2 regions; and subsequently, the enzyme is removed from the membrane, which is enhanced by autophosphorylation.
An important issue that arises is whether autophosphorylation is regulated or not. Although it cannot be inferred from our study when autophosphorylation of the enzyme occurs, it is believed that PKCs are autophosphorylated during their maturation, independent of any activators (16). If autophosphorylation is stoichiometric in the resting state or under normal growing conditions, then one hypothesis would be that autophosphorylation of PKC keeps the enzyme from associating with the plasma membrane and thus retains the enzyme in the cytoplasm. This is supported by the studies on the Aplysia PKC (39). On the other hand, we found in preliminary studies that autophosphorylation at Ser 660 was enhanced in response to angiotensin II and phorbol esters. 2 This raises the possibility that initial translocation and activation of PKC result in further autophosphorylation, which then provides a further force for reverse translocation and signal termination.
Many important cellular processes, including cell proliferation and differentiation, involve the transduction of extracellular signals to specific intracellular sites. During these processes, signaling molecules, including protein kinases and phosphatases, need to traffic to specific cellular locations for effective action on their downstream targets (42). Due to the dynamic nature of cellular signal transduction, the trafficking of these signaling molecules has been difficult to follow. In the present and previous studies (20), we have utilized GFP con-2 K. P. Becker and Y. A. Hannun, unpublished observations. FIG. 9. Schematic representation of the regulation of receptor-induced PKC membrane trafficking. Briefly, the rapid reversible PKC membrane translocation in response to receptor activation is proposed to consist of at least three steps: membrane recruitment (step I), persistent membrane anchorage (step II), and membrane dissociation (step III), involving the C1 and C2 region functions and PKC autophosphorylation, respectively. See "Discussion" for details. P, phosphate on autophosphorylation residues; kinase, the PKC catalytic domain.
jugates to study the regulatory mechanisms for PKC trafficking and signal transduction. In particular, the real time visualization of GFP-PKC fluorescence provides a direct and sensitive means for assessing the dynamic PKC trafficking between the plasma membrane and cytoplasm in response to the activation and desensitization of G protein-coupled receptors. This represents a novel and effective way to examine the distribution of various signaling molecules in the process of signal transduction. In the case of PKC, although the detailed molecular events in each step of its trafficking still remain obscure, our work provides a framework from which to begin to dissect the different functions of each protein module in PKC membrane trafficking. The recently developed GFP-based molecular tools will certainly be helpful in further dissecting detailed molecular events in the dynamic membrane trafficking of PKC. In addition, with these tools in hand, we are also in a unique position to dissect intracellular events regulating PKC trafficking to intracellular locations other than the plasma membrane and cytoplasm.