Interplay between Calcium, Diacylglycerol, and Phosphorylation in the Spatial and Temporal Regulation of PKCα-GFP*

The function of protein kinase C (PKC) is closely regulated by its subcellular localization. We expressed PKCα fused to green fluorescent protein (PKCα-GFP) and examined its translocation in living and permeabilized cells of the human parotid cell line, HSY-EB. ATP induced an oscillatory translocation of PKCα-GFP to and from the plasma membrane that paralleled the appearance of repetitive Ca2+ spikes. Staurosporine attenuated the relocation of PKCα-GFP to the cytosol and caused a stepwise accumulation of PKCα-GFP at the plasma membrane during ATP stimulation. Diacylglycerol enhanced the amplitude and duration of the ATP-induced oscillatory translocation of PKCα-GFP. Ionomycin induced a transient translocation of PKCα-GFP to the plasma membrane despite the continuous elevation of cytosolic Ca2+. The ionomycin-induced transient translocation of PKCα-GFP was prolonged by staurosporine, diacylglycerol, and phorbol myristate acetate. Experiments using permeabilized cells showed that staurosporine or the elimination of ATP and Mg2+ decreases the rate of dissociation of PKCα-GFP from the membrane. Diacylglycerol slowed the dissociation of PKCα-GFP from the membrane regardless of the Ca2+ concentration. The effect of diacylglycerol was attenuated by ATP plus Mg2+ at low concentrations of Ca2+ (<500 nm) but not at high concentrations of Ca2+ (>1000 nm). These data suggest a complex interplay between Ca2+, diacylglycerol, and phosphorylation in the regulation of the membrane binding of PKCα.

Protein kinase C (PKC) 1 isoforms make up a large family of serine/threonine protein kinases and play a key role in transducing numerous signals generated by growth factors, hormones, and neurotransmitters (1)(2)(3). All of the isoforms identified to date share in common an amino-terminal regulatory domain linked to a carboxyl-terminal kinase domain. Based on their cofactor dependence, which is dictated by the structure of the regulatory domain, PKCs are generally classified into three groups: conventional PKCs (␣, ␤I, ␤II, and ␥ subtypes), novel PKCs (␦, ⑀, , and subtypes), and atypical PKCs (, , and ) (3,4). Conventional PKCs (cPKCs) are activated by Ca 2ϩ and diacylglycerol (DAG), whereas novel PKCs are activated by DAG alone. The activation of atypical PKCs is not well understood.
Biochemical and immunocytochemical studies have indicated that the biological activities of PKCs are closely regulated by their subcellular localization (2,4,5). cPKCs contain two conserved membrane-targeting modules, the C1 and C2 domains, within the regulatory domain. The C1 domain binds DAG or phorbol esters, whereas the C2 domain binds acidic lipids such as phosphatidylserine in a Ca 2ϩ -dependent manner (6,7). In addition it has been found that phosphorylation plays an essential role in the activation and translocation of these enzymes (8 -13). The present study addresses the question of how these regulatory mechanisms are integrated for the activation of cPKCs.
Recently, GFP and its variants have been used as expressed fluorescent tags for signaling proteins and their mutants. The development of GFP-tagged PKCs makes it possible to monitor the spatio-temporal dynamics of PKC translocation through real-time visualization in living cells (14). This new strategy has revealed that the translocation of PKC to the plasma membrane occurs transiently and reversibly in response to physiological stimuli such as the activation of phospholipase C-coupled receptors (6,12,15,16). Here we express GFP-tagged human PKC␣ (PKC␣-GFP) in HSY-EB cells and examine the roles of Ca 2ϩ , DAG, and kinase activities in the dynamics of this enzyme during agonist-induced Ca 2ϩ oscillation. We have also used permeabilized cells to more directly and quantitatively analyze the interplay of Ca 2ϩ , DAG, and phosphorylation in the membrane binding of PKC␣-GFP.

EXPERIMENTAL PROCEDURES
Reagents-1,2-Dihexanoyl-sn-glycerol (DiC6) was obtained from Sigma. 1,2-Dioctanoyl-sn-glycerol (DiC8) was from Biomol (Plymouth Meeting, PA). Staurosporine and phorbol 12-myristate 13-acetate (PMA) were from Wako Pure Chemicals (Osaka, Japan). Ionomycin was from Calbiochem-Novabiochem. These reagents were dissolved in dimethyl sulfoxide as 200ϫ stock solutions and diluted to the desired final concentrations shortly before the experiment. [␥- 32  Cell Culture and Transfection-The HSY human parotid cell line, a generous gift from Dr. Mitsunobu Sato (Tokushima University, Japan), was subcloned by using a dilution plating technique, and six clones were obtained. One of these subclones, HSY-EB, was used for imaging experiments in which the cells were cultured for 1 day in sample chambers consisting of 7 ϫ 7-mm plastic cylinders glued to round glass coverslips with silicone rubber adhesive. Transient transfections were performed using a CalPhos mammalian transfection kit (CLONTECH, Palo Alto, CA) with 0.4 g/ml plasmid (pPKC␣-EGFP or pEGFP-C3; CLONTECH) according to the manufacturer's instructions.
Stable transfections of HSY-EB cells with the pPKC␣-EGFP vector were established using LipofectAMINE 2000 (Invitrogen). G418 selection (0.5 mg/ml) was initiated 24 h after the transfection. After 1 week, cells were expanded for an additional week without G418. Colonies of PKC␣-GFP-expressing cells were screened by fluorescence microscopy. One of the stable HSY-EB transfectants (PKC␣-GFP/EB2 cells) was used in phosphorylation experiments. All cells were cultured in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F12 medium supplemented with 10% newborn calf serum, 2 mM glutamine, and 100 g/ml each of penicillin and streptomycin (all from Invitrogen) as described previously (17).
Visualization of PKC Translocation and [Ca 2ϩ ] i in Intact Cells-Experiments were carried out on cells cultured for at least 3 days after transfection. Cells were incubated with 3 M Fura Red/AM (Molecular Probes, Eugene, OR) in HBSS-H for 30 min at room temperature. After loading Fura Red, the cells were washed and incubated in HBSS-H without BSA. Drug addition was performed by replacing HBSS-H buffer with experimental buffers containing various reagents.
The distribution of PKC␣-GFP and [Ca 2ϩ ] i were monitored simultaneously with confocal microscopy using a Leica TCS SP system (Leica, Heidelberg, Germany) equipped with a 40ϫ PL Fluotar objective. Confocal images were obtained by excitation at 488 nm and dual emission wavelengths at 495-550 nm for GFP and at 600 -700 nm for Fura Red. Time series of 50 -150 confocal images were recorded for each experiment at time intervals of between 2.5 and 10 s. Unless stated otherwise, experiments were performed independently on at least four different occasions, and in each experiment, recordings were obtained from 2-10 cells.
Analysis of the Dissociation of PKC␣-GFP in Permeabilized HSY-EB Cells-The dissociation of PKC␣-GFP was assessed using saponin-permeabilized HSY-EB cells in ICM. The concentration of free Ca 2ϩ in the medium ([Ca 2ϩ ] m ) was calculated from the pH and the amount of potassium, sodium, calcium, EGTA, ATP, and magnesium by the method of Fabiato and Fabiato (18). In this experiment, PKC␣-GFPtransfected cells first were permeabilized with ICM-1000S. Cells were then exposed to test media consisting of ICM with variable [Ca 2ϩ ] m . The time course of the dissociation of PKC␣-GFP from the permeabilized cells was monitored at room temperature by confocal microscopy with excitation at 488 nm and emission at 495-550 nm. The data were then analyzed by a nonlinear least-squares fit to a monoexponential equation, PKC dissociation ϭ A ϫ (1 Ϫ exp Ϫkt ), where A is the maximum amplitude of PKC dissociation (fractional PKC dissociation) for that particular experimental condition, t is time, and k is the rate constant for PKC dissociation.
Phosphorylation of PKC␣-GFP in Permeabilized HSY-EB Cells-PKC␣-GFP/EB2 cells were seeded at a density of 7.5 ϫ 10 5 cells/60-mm dish and cultured for 3 days. Cells were washed with HBSS-H and then permeabilized with ICM-1000S for 5 min and kept for 5 min in ICM-250S containing 100 M MgCl 2 . Permeabilized cells were then incubated with 800 l of phosphorylation medium (ICM-250S containing 100 M MgCl 2 , 50 M ATP, and 50 Ci/ml of [␥-32 P]ATP) at room temperature. After a 10-min incubation, the medium was corrected and mixed with 400 l of ice-cold 3ϫ stop solution. The cells remaining on the dish were suspended in 400 l of ice-cold 3ϫ stop solution and then added to 800 l of phosphorylation medium. Both samples were centrifuged at 10,000 ϫ g for 5 min, and the resulting supernatants were used for immunoprecipitations.
Immunoprecipitation of PKCa-GFP-A 1000-l sample aliquot was supplemented with 100 l of 1% BSA and 10 l (2.5 g) of anti-human PKC␣ monoclonal antibody (BD Transduction Laboratories) and incubated at 4°C overnight. The amount of antibody used in this experiment is sufficient to immunoprecipitate ϳ20% of the PKC␣-GFP protein expressed by confluent PKC␣-GFP/EB2 cells in a 60-mm dish. Protein A beads (Sigma) were prewashed with 1ϫ stop solution plus 0.1% BSA and added to the samples (10 l of protein A beads/tube). After an additional hour of incubation, beads were collected by centrifugation and washed five times with 1ϫ stop solution. The tube was changed for the last spin. Protein retained by the washed beads was then eluted with 30 l of electrophoresis sample buffer, and 20 l of each sample were subjected to electrophoresis on 3-8% NuPAGE Tris acetate gels (Novel Experimental Technology). Gels were silver-stained, dried, and visualized by autoradiography using x-ray films.

ATP-induced Oscillatory Translocation of PKC␣-GFP-The
stimulation of phospholipase C-coupled receptors results in the generation of inositol 1,4,5-trisphosphate-induced Ca 2ϩ signals and DAG signals. Our previous study demonstrated that the activation of P 2 -purinoceptors induces a typical Ca 2ϩ oscillation in HSY cells (19). Ca 2ϩ and DAG are known as regulators for the translocation of cPKCs. We therefore examined the dynamic translocation of PKC␣-GFP during ATP-induced Ca 2ϩ oscillations. Earlier studies have shown that a C-terminal GFP tag does not affect the catalytic activity or the cofactor dependence of PKC␣ (15,20). To monitor the [Ca 2ϩ ] i and the translocation of PKC␣-GFP simultaneously, a calcium indicator, Fura Red, was loaded into PKC␣-GFP-expressing HSY-EB cells. Fig. 1 shows typical images and time courses of the translocation of PKC␣-GFP versus the calcium spikes. In unstimulated cells, PKC␣-GFP was evenly distributed throughout the cytosol (Fig. 1, a1), but stimulation with 10 -100 M ATP resulted in oscillations in [Ca 2ϩ ] i (Fig. 1, b1-b4 and c) and simultaneous oscillatory translocations of PKC␣-GFP to and from the plasma membrane ( Fig. 1, a1-a4 and c). The extent of the ATP-induced translocation of PKC␣-GFP was 15-30% of the maximal translocation induced by ionomycin (Fig. 1, a5). Oscillatory translocation of PKC␣-GFP was also observed when the cells were stimulated with 10 -100 M carbachol (data not shown). When HSY-EB cells expressing a control plasmid, EGFP, were loaded with Fura Red, the GFP fluorescence patterns were not changed by ATP, carbachol, or ionomycin (data not shown). Similar oscillatory translocation of fluorescent PKC␣ fusion proteins has been reported in human embryonic kidney cells stimulated with histamine (16).
Effect of Staurosporine and DAG on ATP-induced Oscillatory Translocation of PKC␣-GFP-The subcellular localization of cPKCs is known to be affected by DAG (1). In addition, it has been suggested that the kinase activity of PKC itself plays an essential role in the relocation of PKC␤II (12,21). We therefore examined the effect of a membrane-permeable DAG, DiC8, and the PKC inhibitor staurosporine on the ATP-induced oscillatory translocation of PKC␣-GFP. Staurosporine itself had no effect on the distribution of PKC␣-GFP (Fig. 2b), but when cells were pretreated with staurosporine, PKC␣-GFP accumulated at the plasma membrane in a stepwise manner in response to the repetitive ATP-induced Ca 2ϩ spikes (Fig. 2). This result suggests that the oscillatory translocation of PKC␣-GFP is driven by the interplay of Ca 2ϩ signals and kinase activities.
The addition of DiC8 to the cells decreased the frequency of the ATP-induced repetitive Ca 2ϩ spiking. This effect of DiC8 is probably due to suppression of phosphoinositide metabolism through the activation of PKCs (19). DiC8 markedly increased the extent of the translocation of PKC␣-GFP (Fig. 3, a-g). The traces in Fig. 3, i and j, clearly demonstrate that externally added DiC8 enhanced the amplitude of the oscillatory translocation of PKC␣-GFP (Fig. 3i) without any enhancement in the amplitude of the Ca 2ϩ spikes (Fig. 3j). DiC8 alone did not change the localization of PKC␣-GFP (Fig. 5c). To the best of our knowledge, this is the first demonstration that the lipid mediator DAG can act as an amplitude modulator for the oscillatory translocation of PKC␣. The expanded time course revealed that in the absence of DiC8, PKC␣-GFP rapidly dissociated from the plasma membrane when each Ca 2ϩ spike was terminated (Fig. 3k). In contrast, after the addition of DiC8, the dissociation of PKC␣-GFP was delayed after the Ca 2ϩ spike (Fig. 3l). In the presence of DiC8, the half-maximal relocation (recovery of PKC␣-GFP fluorescence in the cytosol) was delayed by 5-10 s after the half-maximal drop in [Ca 2ϩ ] i (recovery of Fura Red fluorescence).
The effects of DiC8 on the ATP-induced Ca 2ϩ oscillation and the oscillatory translocation of PKC␣-GFP were analyzed quantitatively (Fig. 4). DiC8 increased the intervals between Ca 2ϩ spikes (Fig. 4a), but the amplitude (Fig. 4b) and duration (Fig. 4c) of each Ca 2ϩ spike was not altered even with the highest concentration (100 M) of DiC8. These results clearly demonstrate that DiC8 did not alter the pattern of each Ca 2ϩ spike. On the other hand, the amplitude and duration of the PKC␣-GFP translocation were increased by DiC8 in a concentration-dependent manner. The delay of the half-maximal relocation of PKC␣-GFP after the half-maximal recovery of Fura Red fluorescence was increased by DiC8 at concentrations above 3 M (Fig. 4d). Similar effects were also observed with another DAG analog, DiC6 (data not shown).
Effects of Staurosporine, DiC8, and PMA on Ionomycin-induced Translocation of PKC␣-GFP-To clarify the effect of Ca 2ϩ on the translocation of PKC␣-GFP, we manipulated the Ca 2ϩ response using ionomycin-treated cells. The treatment of ionomycin continuously elevated [Ca 2ϩ ] i , and if extracellular Ca 2ϩ was removed, the ionomycin treatment lowered [Ca 2ϩ ] i to below the resting level (Fig. 5a5, closed circles). Unlike the [Ca 2ϩ ] i response, ionomycin induced a transient translocation of PKC␣-GFP; the maximal translocation occurred within 30 s (Fig. 5a2), and 50 -90% of the translocated PKC␣-GFP relocated to the cytosol within 2 min after the stimulation (Fig.  5a3). This implies the presence of a specific mechanism by which PKC␣-GFP relocates to the cytosol during [Ca 2ϩ ] i elevation. Subsequent removal of extracellular Ca 2ϩ resulted in the complete recovery of PKC␣-GFP to the cytosol (Fig. 5a4).
We next examined the effect of staurosporine and DAG on ionomycin-induced translocation of PKC␣-GFP. When cells were pretreated with staurosporine, PKC␣-GFP translocated to the plasma membrane in response to the ionomycin-induced [Ca 2ϩ ] i elevation but did not relocate to the cytosol until the removal of extracellular Ca 2ϩ (Fig. 5b). These results confirm that the relocation of PKC␣ is dependent on PKC kinase activity.
Treatment with DiC8 alone did not change the localization of PKC␣-GFP (Fig. 5c). However, in DiC8-pretreated cells that were then stimulated with ionomycin, PKC␣-GFP translocated to the plasma membrane and remained there during the period of [Ca 2ϩ ] i elevation (Fig. 5d), relocating to the cytosol only when Ca 2ϩ had been removed. These results indicate that Ca 2ϩ is required for DiC8 to exert its influence on the translocation of PKC␣-GFP. Unlike DiC8, PMA alone induced a slow translocation of PKC␣-GFP to the plasma membrane; the PMAinduced translocation became detectable at 5 min or later and was completed at 10 -15 min (Fig. 5e). When ionomycin was added 30 s after application of PMA, PKC␣-GFP was translocated to the plasma membrane (Fig. 5f) and did not relocate even after the removal of extracellular Ca 2ϩ . The fact that the PMA-mediated stable binding of PKC␣-GFP to the plasma membrane was established within 2 min in ionomycin-treated cells (compared with 10 min in the absence of ionomycin) suggests that the binding of PMA to PKC␣-GFP was markedly accelerated by Ca 2ϩ . Recent studies on PKC␥ have proposed that cPKC isoforms are sequentially activated by Ca 2ϩ and DAG (6). Based on this well established "sequential model," our results can be explained as follows. The initial binding of Ca 2ϩ to the C2 domain of PKC␣-GFP recruits the enzyme to the plasma membrane and makes the C1 domain accessible to DAG. The subsequent binding of DAG to the C1 domain enables the continuous binding of PKC␣-GFP to the plasma membrane. Although PMA is a functional analog of DAG, its action is persistent because of its slow dissociation from cell membranes and its metabolic stability (22,23).
Analysis of the Distribution of PKC␣ Using Saponin-permeabilized Cells-The data presented above suggest that kinase activities and/or DAG modulate the Ca 2ϩ -dependent translocation of cPKCs. The interaction of these regulatory mechanisms, however, is largely unresolved. To explore the question of how kinase activities and DAG modulate the Ca 2ϩ -dependent binding of PKC␣-GFP to biological membranes, we initially attempted to directly examine the effect of Ca 2ϩ on the binding of PKC␣-GFP to the plasma membrane using saponin-permeabilized cells. As the Ca 2ϩ concentration within permeabilized cells quickly equilibrates to the Ca 2ϩ concentration of the ICM ([Ca 2ϩ ] m ), we expected that PKC␣-GFP would translocate to the plasma membrane upon the permeabilization. In fact, 50 -70% of the PKC␣-GFP fluorescence translocated to the membrane upon permeabilization in ICM containing 1000 nM Ca 2ϩ and 100 g/ml saponin (Fig. 6, a, b, and f). The translocated PKC␣-GFP stably bound to the membrane for at least 10 min (Fig. 6, c and f) but was rapidly released and lost into medium with the subsequent exposure to "Ca 2ϩ -free" ICM ( Fig. 6, d-f) in which [Ca 2ϩ ] m was calculated to be ϳ1.5 nM. This Ca 2ϩ -dependent fraction of the change in fluorescence (depicted as dF) directly demonstrates the Ca 2ϩ -dependence of PKC␣-GFP in its binding to cell membranes.
We then tested the effect of PMA (Fig. 7a) and staurosporine (Fig. 7b) on the release of PKC␣-GFP from the membrane. When permeabilized cells were pretreated for 30 s with 1 M PMA, the dissociation of PKC␣-GFP in the Ca 2ϩ -free ICM was strongly blocked. Similar results were also obtained when cells were permeabilized with streptolysin O (data not shown). This result is expected, given the experiment using intact ionomycin-treated cells (Fig. 5e) where PMA blocked the relocation of PKC␣-GFP in the absence of extracellular Ca 2ϩ , and strongly suggests that the PMA-dependent binding between the C1 domain of PKC␣-GFP and the plasma membrane is preserved in permeabilized cells. Because the effect of PMA on the binding is mediated by the C1 domain of PKC␣-GFP, we expected that the effect of DiC8 on PKC␣-GFP could also be examined in this permeabilized cell system. PKC␣-GFP was not released from the permeabilized cells in ICM containing 250 nM [Ca 2ϩ ] m lacking ATP and Mg 2ϩ but was readily released by the addition of ATP and Mg 2ϩ (Fig. 7b). The addition of either ATP or Mg 2ϩ did not release PKC␣-GFP at 250 nM [Ca 2ϩ ] m (data not shown). The ATP/Mg-dependent release of PKC␣-GFP was strongly blocked by staurosporine (Fig  7b). These data directly indicate that the kinase activities accelerate the release of PKC␣-GFP.
It has been suggested that Ca 2ϩ -mediated membrane binding of cPKCs is attenuated by autophosphorylation (12,21,24). Thus, we examined whether the ATP/Mg-dependent release of PKC␣-GFP is due to its phosphorylation. To directly determine the phosphorylation of PKC␣-GFP, released and membranebound PKC␣-GFP were immunoprecipitated with anti-PKC␣ antibody (Fig. 8A), and their incorporation of the 32 P from radioactive ATP was analyzed by autoradiography (Fig. 8B). Although the phosphorylation of membrane-bound PKC␣-GFP was detected, the additional phosphorylation of the released PKC␣-GFP was not recognized. This result suggests that the ATP/Mg 2ϩ -dependent release of PKC␣-GFP is not directly mediated by the phosphorylation of PKC␣-GFP. M PMA (f) were exposed to 5 M ionomycin in the presence or absence of extracellular Ca 2ϩ . Confocal images of PKC␣-GFP were obtained before (a1, b1, d1, and f1) and 30 s (a2, b2, d2, and f2), 120 s (a3, b3, d3, and f3), and 220 s (a4, b4, d4, and f4) after stimulation with 5 M ionomycin (Iono). Extracellular Ca 2ϩ was removed 180 s after stimulation. In control experiments, images were acquired before (c1 and e1), and 2 (c2 and e2) and 20 min (c3 and e3) after treatment with DiC8 (c) and PMA (e).  Figs. 6 and 7 clearly indicate that the effect of Ca 2ϩ and PMA on PKC␣-GFP membrane binding and the role of kinase activities in facilitating the release are preserved in saponin-permeabilized cells. To further analyze the interplay of Ca 2ϩ , DiC8, and the kinase activities for the binding of PKC␣-GFP to cell membranes, permeabilized cells were exposed to various test media, and the resulting time courses of the release of PKC␣-GFP were analyzed quantitatively. As shown in Fig. 9, A and B, the time courses conform well to a single exponential decay (see "Experimental Procedures"). Accordingly, we have used least-squares curve-fitting techniques to determine the initial rate of PKC release, thereby quantitating our results. The calculated initial rates are plotted as a function of [Ca 2ϩ ] m in Fig. 9C.

Effect of DiC8 on the Ca 2ϩ -and ATP/Mg-dependent Dissociation of PKC␣-GFP-The experiments depicted in
In the absence of ATP/Mg, the release rate gradually decreased as the [Ca 2ϩ ] m increased (Fig. 9A, a-c), and release was blocked almost completely by 100 or 250 nM [Ca 2ϩ ] m (Fig.   9A, d and e). The addition of DiC8 slowed the release of PKC␣-GFP in the presence of 1.5, 25, and 50 nM [Ca 2ϩ ] m (Fig. 9A, a-c,  and C). The half-maximal Ca 2ϩ concentrations (the concentration required to reduce the release rate by 50%) in the absence and presence of DiC8 were estimated to be 12.1 and 13.7 nM, respectively (Fig. 9C). DiC8 decreased the release rate by ϳ50%. These results clearly demonstrate that DiC8 attenuates the release of PKC␣-GFP without affecting the efficacy of Ca 2ϩ . Thus, DiC8 and Ca 2ϩ appear to act independently in mediating the binding of PKC␣-GFP to cell membranes.
In the presence of ATP/Mg, the release rate also decreased with increased [Ca 2ϩ ] m (Fig. 9B, a-e), whereas the Ca 2ϩ concentrations required to prevent the release were markedly higher than those required in the absence of ATP/Mg. Halfmaximal Ca 2ϩ concentrations in the absence and presence of DiC8 were estimated to be 57.9 and 66.8 nM, respectively. Thus, the efficacy of Ca 2ϩ in inducing the binding of PKC␣-GFP to cell membranes was reduced to ϳ20% by ATP/Mg. Furthermore, unlike in the absence of ATP/Mg, in the presence of ATP/Mg the effect of DiC8 was altered by [Ca 2ϩ ] m . Although DiC8 failed to slow the release of PKC␣-GFP in the presence of ATP/Mg and [Ca 2ϩ ] m lower than 500 nM (Fig. 9, B and D), it decreased the release rate to ϳ50% in the presence of high [Ca 2ϩ ] m (above 1000 nM) and ATP/Mg. This result clearly demonstrates that in the presence of ATP/Mg, DAG selectively potentiates the effect of high concentrations of Ca 2ϩ on the membrane binding of PKC␣-GFP. It appears, therefore, that in intact cells, this interplay between Ca 2ϩ , DAG, and ATP/Mg may underlie the DAG-dependent enhancement of the translocation of PKC␣-GFP during agonist-induced Ca 2ϩ oscillations. DISCUSSION We have demonstrated that agonist-induced Ca 2ϩ oscillations cause a parallel oscillatory translocation of PKC␣-GFP to and from the plasma membrane. This oscillatory translocation of PKC␣-GFP is accomplished by an interplay between Ca 2ϩ and kinase activities, where the translocation to the membrane is regulated by the increase in [Ca 2ϩ ] i and the relocation to the cytosol is regulated by the kinase activity in addition to the decrease in [Ca 2ϩ ] i . More strikingly, we found that DAG increases the amplitude of the oscillatory translocation of PKC␣ by delaying the dissociation of PKC␣-GFP, so that the translocation overcomes the relocation. To the best of our knowledge, this is the first evidence for DAG as an amplitude modulator for the oscillatory translocation of cPKCs.
It has been proposed that DAG delays the dissociation of PKC␥ and leads to a persistent membrane translocation of PKC␥ during high-but not low-frequency Ca 2ϩ spikes (6). This proposal points to the possibility that PKC␥ serves as a molecular machine for decoding Ca 2ϩ oscillations in combination with the DAG signal. However, while we have shown that DAG delayed the dissociation of PKC␣ after the termination of a Ca 2ϩ spike, we did not observe persistent membrane translocation of PKC␣ during agonist-induced Ca 2ϩ oscillations after DAG had been added. The difference between the present and previous studies may be attributable to the different cell lines used or to the different PKC isoforms involved. It is possible that the PKC␣ and PKC␥ isoforms have differential functions; for example, PKC␣ may act as an amplifier, whereas PKC␥ may serve as a decoder of Ca 2ϩ oscillation.
Regarding the physiological relevance of this amplitude modulation, we postulate that DAG signals play a crucial role in recruiting PKC␣ to specific sites within the plasma membrane where phosphoinositide metabolism is locally activated. Unlike the rapid diffusion of the resulting Ca 2ϩ signals, DAG signals would diffuse more slowly, remaining localized for a longer time in the specific site (25). Indeed, in macrophage cell lines, the generation of local DAG signals (26) and the concomitant accumulation of PKC␣ and PKC␦ to the phagosomal membranes have been suggested (27)(28)(29). Recent reports using GFP-tagged PKC␣ have shown that PKC␣ accumulates in regions of cell-cell contact (20) and at focal spots of the plasma membrane (15). These authors have also speculated that there is a requirement for some other unidentified signal(s) in addition to Ca 2ϩ . We anticipate that the spatial and temporal regulation of cPKCs is controlled by the interplay of Ca 2ϩ , DAG, and phosphorylation. This type of restricted signaling process to small subregions of the cell might be relevant in polar cells such as epithelial cells, chemotactic cells, and neuronal cells (25).
In experiments using ionomycin, we demonstrated that PKC␣-GFP translocates to the plasma membrane in a transient manner despite the continuous elevation of [Ca 2ϩ ] i . Preincubation with staurosporine blocks the relocation of PKC␣-GFP to the cytosol during [Ca 2ϩ ] i elevation, indicating that the relocation of PKC␣-GFP is dependent on kinase activities. More importantly, it is interesting to note that staurosporine requires [Ca 2ϩ ] i elevation to enhance binding of PKC␣-GFP to the plasma membrane and cannot block the relocation of PKC␣-GFP to the cytosol after the removal of Ca 2ϩ . This suggests that the phosphorylation of PKC or other proteins exerts its effect by modulating the affinity of PKC␣-GFP for Ca 2ϩ . In agreement with this observation, we found that membranebound PKC␣-GFP in permeabilized cells was released in ATP/ Mg-dependent manner. It has been shown that PKC␤II requires autophosphorylation at Thr-641 and Ser-660 to be released from membranes (12,21). Homologous autophosphorylation sites are also present in PKC␣ (10,13). Although membrane-bound PKC␣-GFP was phosphorylated, we did not find that phosphorylation of PKC␣-GFP increased after ATP/ Mg-dependent release. This experiment suggests that, in addition to the autophosphorylation of PKC, the phosphorylation of proteins other than PKC␣ may be involved in the ATP/Mg-dependent release of membrane-bound PKC␣. Additional work will be required to explore this possibility.
It is generally known that the effect of DAG and Ca 2ϩ on the translocation of cPKC isoforms are mediated by the C1 and C2 domains, respectively. In addition, a number of aspects of PKC regulation have been linked to their phosphorylation sites (10,12,13,21,24). Molecular mechanisms involved in the translocation of cPKCs have been extensively examined with mutants that lack phosphorylation site(s) and functional regulatory domain(s). Although these studies provide important findings concerning the function of the regulatory domains and phosphorylation sites, the details of how these domains interact and the role of phosphorylation plays in the membrane translocation of cPKCs remain largely unclear.
To attempt to resolve these issues, we directly examined the effects of Ca 2ϩ , DAG, and ATP/Mg on the binding between PKC␣-GFP and cell membranes using saponin-permeabilized cells. Because our initial data showed that saponin-permeabilized cells preserved the responsiveness of the translocation of PKC␣-GFP to Ca 2ϩ , PMA, DiC8, and ATP/Mg (Figs. 6 and 7), the membrane interaction of PKC␣-GFP is apparently intact in this permeabilized cell system. We then extended this technique to analyze the process by which PKC␣-GFP dissociates from the cell membranes in permeabilized cells (Fig. 9). This simple technique allows us to directly assess the roles PKC regulators and their interactions play in the binding of PKC to biological membranes in close to physiological conditions. We learned the following: 1) DAG supports the binding between PKC␣-GFP and membranes regardless of Ca 2ϩ concentrations; 2) ATP/Mg decreases the effect of Ca 2ϩ on the membrane binding of PKC␣-GFP and thereby accelerates the release of PKC␣-GFP; 3) ATP/Mg attenuates the DAG-mediated membrane binding of PKC␣-GFP at lower Ca 2ϩ concentrations (Ͻ500 nM); (4) high concentrations of Ca 2ϩ (Ͼ1000 nM) restore the DAG-mediated membrane binding of PKC␣-GFP in the presence of ATP/Mg. These results directly support some of the current hypotheses discussed below and provide new insights into the collaborative regulation of PKC␣ by Ca 2ϩ , DAG and phosphorylation.
We found that DAG prolongs the Ca 2ϩ -mediated binding of PKC␣-GFP to the plasma membrane, whereas DAG alone exerts no effect on the translocation of cytosolic PKC␣-GFP in intact cells. These observations support the proposition that cPKCs require Ca 2ϩ -mediated translocation to become accessible to DAG (6,30). However, unlike the translocation of cytosolic PKC␣-GFP, the release of the membrane-bound PKC␣-GFP in permeabilized cells was slowed by DAG regardless of the Ca 2ϩ concentration, suggesting that it is the binding of cPKCs to phospholipids, rather than to Ca 2ϩ , that makes these enzymes accessible to DAG. In addition, this implies that the Ca 2ϩ -binding site and the DAG-binding site of membranebound PKC␣ do not interact directly, as suggested by studies on PKC␤II (31,32). Given that DAG-mediated membrane binding of PKC␣-GFP is attenuated by ATP/Mg when the [Ca 2ϩ ] m is lower than 500 nM but the DAG-mediated binding is restored in the presence of ATP/Mg with high concentrations of Ca 2ϩ , it appears that DAG and phosphorylation collaborate to make this enzyme selectively responsive to high Ca 2ϩ concentrations. Although the underlying mechanisms have yet to be clarified, the present study provides the first detailed insight into the way Ca 2ϩ , DAG, and phosphorylation concordantly regulate the oscillatory translocation of cPKCs.