Membrane targeting and cytoplasmic sequestration in the spatiotemporal localization of human protein kinase C alpha.

In order to map the molecular determinants that dictate the subcellular localization of human protein kinase C alpha (hPKCalpha), full-length and deletion mutants of hPKCalpha were tagged with the green fluorescent protein (GFP) and transiently expressed in GH3B6 cells. We found that upon thyrotropin-releasing hormone (TRH) or phorbol 12-myristate 13-acetate stimulation, hPKCalpha-GFP was localized exclusively in regions of cell-cell contacts. Surprisingly, PKCalpha failed to translocate in single cells despite the presence of TRH receptors, as attested by the TRH-induced rise in intracellular calcium concentration in these cells. TRH-stimulated translocation of hPKCalpha-GFP from the cytoplasm to cell-cell contacts was a biphasic process: a fast (measured in seconds) and transient phase, followed by a slower (approximately 1 hour) and long lasting phase. The latter and the translocation induced by phorbol 12-myristate 13-acetate absolutely required the N-terminal V1 region. In contrast to the full-length hPKCalpha, the N-terminal regulatory domain alone or associated with the V3 hinge region was spontaneously and uniformly localized at the plasma membrane of single and apposed cells. However, treatment with the calcium chelator BAPTA/AM induced a differential cytoplasmic/nuclear redistribution of the regulatory domain, depending on its association with V3, which suggests the existence of a mechanism controlling the cytoplasmic sequestration of inactive hPKCalpha and involving the V3 region. By using other deletion mutants, we were able to map the sequence required for this sequestration to the C2+V3 regions. This work points to the existence of a complex interplay between membrane targeting and cytoplasmic sequestration in the control of the spatiotemporal localization of hPKCalpha.

Protein kinase C (PKC) 1 is a term coined to designate a family of isoforms that play key roles in the processes of proliferation/apoptosis, differentiation, or hormone release and the function of which is regulated at multiple levels: transcription, phosphorylation, and subcellular targeting.
Subcellular targeting of PKC, and particularly that of the conventional PKC ␣, ␤, and ␥ isoforms, is linked to enzyme activation. Indeed, inactive PKC is mostly cytoplasmic, whereas activated PKC translocates to various membrane compartments, such as the plasma membrane. Physiological activation of the conventional PKC is associated with an increase in diacylglycerol (DAG) and intracellular Ca 2ϩ concentrations (1), which results from the activation of a seven transmembrane receptor coupled to phospholipase C␥ via a heterotrimeric G protein (2). The increase in Ca 2ϩ is thought to be essential for translocation, although it can be bypassed by the phorbol ester phorbol 12-myristate 13-acetate (PMA) (3). The increase in DAG concentration at the plasma membrane allows additional conformational changes to achieve PKC activation at its targeting site. When inactive, PKC is in a "closed" conformation due to the interaction of the pseudosubstrate sequence with the catalytic site (4,5). Upon activation, this interaction is disrupted, unmasking the regions of the enzyme involved in translocation and accumulation at the targeting site. Although numerous studies have been aimed at determining which region of the protein is involved in each step, it is not clear yet whether or not membrane sequestration results from an increase in the affinity of PKC for membrane phospholipids (essentially phosphatidylserine), or if it involves an interaction with a "shuttle" protein. Nevertheless, it appears that the NH 2 -terminal regulatory domain is essential for translocation, whereas the catalytic domain seems to be less important, even though it may influence PMA-induced PKC translocation (6). The amino acid sequences responsible for the interaction with phospholipids are located within the C1 and in the C2 regions (7). This has been demonstrated both by conventional biochemistry and direct microscopic observation in the living cell by use of the green fluorescent protein (GFP) tag (3, 8 -11). Accumulation of the enzyme at its targeting site may also result from an interaction with anchoring proteins (12). Mochly-Rosen and co-workers (13,14) have cloned the first cDNA encoding a receptor for activated protein kinase C (RACK1), recently described as an inhibitor of Src tyrosine kinase activity and NIH 3T3 cell growth (15). The interaction between PKC and RACK1 involves the C2 region (16), also known for its high affinity for Ca 2ϩ (17,18). More recently, using the V1 region of PKC⑀ as a bait in the two hybrid system, ␤Ј-COP has been characterized as a RACK for this PKC isoform (19). It is noteworthy that although it does not bind Ca 2ϩ , the V1 region of PKC⑀ shares homologies with the C2 region of conventional PKCs (20). Additional studies provide evidence for the involvement of V1 in PKC subcellular targeting: the V1 region of atypical PKC contains a protein-protein interacting motif that mediates its interaction with p62 (21), and the involvement of the V1 region of PKC␥ in the translocation kinetics of PKC␥ has recently been evidenced by Oancea and Meyer (22).
In the present study, human PKC␣ (hPKC␣)-GFP and GFPtagged deletion mutants were transiently expressed in the pituitary GH3B6 cell line, stimulated by PMA or by thyrotropin-releasing hormone (TRH), used as a physiological stimulus. We show here the following: 1) the exclusive localization of hPKC␣-GFP at the interface between two apposed stimulated cells; 2) the biphasic nature of translocation upon TRH stimulation (rapid and reversible versus slow and long lasting) and the dependence on V1 for the slow and long lasting phase; 3) the absence of hPKC␣-GFP translocation in single cells despite the presence of TRH receptors as attested by the induced calcium increase in these cells; and 4) the key role played by the C2-V3 region in the control of the cytoplasmic sequestration of the hPKC␣-GFP.
Construction of Plasmids Encoding Fusion Proteins-The constructs used here are presented in Fig. 2. The full-length hPKC␣ cDNA and its various deletion mutants with an EcoRI site at their 5Ј-terminus and a KpnI site at their 3Ј-terminus were produced by polymerase chain reaction using hPKC␣ cDNA subcloned in the pBabe vector as a template.
The various polymerase chain reaction fragments encoding fulllength or deletion mutants of hPKC␣ were gel-purified, digested with EcoRI and KpnI, and then fused in-frame to GFP by ligation into EcoRI and KpnI digested pEGFP-N1 vector. The sequences of ligated polymerase chain reaction fragments were checked by DNA sequencing, and no mutations were detected.
Cell Culture, Transfection, and Observation of Fusion Protein Localization-GH3B6 cells were cultured in Ham's F-10 medium supplemented with 2.5% (v/v) fetal bovine serum and 15% (v/v) horse serum both heat inactivated at 56°C for 1 h. Transient transfection of GH3B6 cells was performed with ExGen 500 according to the manufacturer's standard protocols. Briefly, cells were seeded at 25,000 cells per well in six-well dishes (Falcon) for 18 h before transfection. Immediately before transfection, fresh culture medium (2.5 ml) was added to the cells. Five l of ExGen 500 stock solution were diluted in 50 l NaCl 150 mM. The desired amount of plasmid DNA was also diluted in 50 l of NaCl 150 mM in order to have 10 linear polyethylenimine nitrogens per DNA phosphate. The two solutions were then mixed. After 10 min, the transfection mixture was added to the cells. The six-well dishes were then centrifuged for 5 min at 280 ϫ g and maintained for 4 h at 37°C. The medium was then removed and replaced by fresh culture medium. The fluorescence of fusion proteins was observed 48 h after transfection. The localization of fusion proteins in living cells was examined by conventional (long term treatment) or confocal (short term treatment) fluorescence microscopy. Confocal laser scanning microscope was equipped with an Ar/Kr laser (Odyssey XL with InterVision 1.4.1 software, Noran Instruments Inc., Middleton, WI) as described by Guérineau et al. (23). At the time of observation, the culture medium was replaced by a buffer composed of 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 10 mM HEPES, glucose 6 mM, pH 7.4.
Localization of hPKC␣ fusion proteins was monitored under basal conditions and after application of 100 nM PMA or 100 nM TRH for the times indicated. Although GH3B6 cells are clonal cells, we observed that not all of the cells behaved identically concerning translocation. Absence of translocation of a particular fusion protein was assumed when it could not be observed in any cells; however, successful translocation was never observed in 100% of the cells. For long term treatments, at least 100 cells were observed in each experiment. For short term treatments, at least 10 cells were observed in each experiment. All experiments were repeated between 3 and 10 times.
Immunoprecipitation-hPKC␣-GFP, ⌬(V1)hPKC␣-GFP, and GFP constructs were transiently transfected into GH3B6 cells. Forty-eight hours after transfection, cells were washed in cold PBS and incubated in RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin, 1 g/ml pepstatin) at 4°C by gently rocking. Cells were then scrapped and collected in microcentrifuge tubes. Lysates were precleared with 50 l of protein A beads, incubated for 10 min at 4°C by gentle rocking, and centrifuged at 14,000 ϫ g for 10 min at 4°C. Supernatants were collected. Two mg of cell proteins were mixed with 2 g of GFP antibody, and this reaction mixture was incubated at 4°C overnight. Immunocomplexes were captured by adding 50 l of protein A beads. This mixture was gently rocked at 4°C overnight. After centrifugation and washing of the beads with 800 l of 50 mM Tris, pH 7.4, supplemented with 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 g/ml pepstatin, immunocomplexes were resuspended either in 50 l of 50 mM Tris, pH 7.4, for further kinase activity assay or in 50 l of Laemmli buffer (24) for Western blot analysis.
PKC␣ Catalytic Activity Measurement-Catalytic activity of hPKC␣ purified from Sf9 cells and that of immunoprecipitated hPKC␣-GFP, ⌬(V1)hPKC␣-GFP, or GFP were measured with the S17R peptide as substrate (SLKKRSGSFSKLRASIRR). The S17R peptide corresponds to the PKC phosphorylation site of clone 35F (25). The amount of each protein used for catalytic activity assay was estimated before the assay by Western blot analysis with a PKC␣ antibody.
Activity was measured in the presence of 20 M S17R, 1 M EGTA, 10 M PMA, 5 mM MgAc, 25 M ATP, 1 mM dithiothreitol, 1 nM [␥-32 P]ATP (specific activity, 30 Ci/mmol) (Amersham Pharmacia Biotech), 10 g/ml PS (26). The reaction was prepared in the absence or presence of 1.2 mM calcium. Reaction was started by incubation at 30°C for 5 min and stopped at 0°C for 5 min. One-half volume of each reaction mixture was dropped down on phosphocellulose paper P81 (Whatman) squares, which were subsequently washed twice for 10 min in 0.01 M phosphoric acid, once in acetone for 30 s, and in petroleum ether for 10 s and then air dried. The paper squares were transferred to scintillation vials and counted.
Cell Fractionation and Western Blot Analysis-Transiently transfected GH3B6 cells were separated into soluble and membrane fractions. Untreated or 100 nM PMA-treated cells were washed with cold PBS followed by scrapping in homogenization buffer (10 mM Tris, 2 mM EDTA, 0.01% phenylmethylsulfonyl fluoride, 20 g/ml leupeptin). Cells were then homogenized in a glass Dounce homogenizer and centrifuged for 30 min at 14,000 rpm. Supernatants were collected; they corresponded to soluble fractions. Pellets were resuspended in homogenization buffer supplemented with 1% (v/v) Nonidet P-40 and incubated for 45 min on ice. This corresponds to membrane fractions.
For immunoblotting, soluble and membrane fractions were subjected to 10% SDS-polyacrylamide gel electrophoresis and electrophoretically transferred onto nitrocellulose membranes. Nonspecific binding sites were blocked by incubation with Tris-buffered saline (50 mM Tris, 150 mM NaCl, pH 7.4) containing 10% powdered milk for 1 h at room temperature. Membranes were then incubated with anti-PKC␣ (1:2000) or with anti-GFP (1:1000) antibody overnight at 4°C or for 1 h at room temperature, respectively. After washing with Tris-buffered saline containing 0.1% Tween, membranes were incubated with anti-mouse IgGperoxidase antibody (1:4000) for 1 h at room temperature. Immunoreactive bands were visualized with chemiluminescence detection kit.
Production and Purification of hPKC␣ from Sf9 Cells-T-Flasks containing 15 ϫ 10 6 Sf9 cells were incubated for 3 days at 27°C with 5 ϫ 10 7 pfu/ml of recombinant baculovirus encoding hPKC␣ (27). Infected cells were resuspended in medium, pelleted, and washed three times in PBS, pH 7.4. Cells were sonicated in homogenization buffer (10 mM Tris, pH 7.6, 2 mM EDTA, 0.01% phenylmethylsulfonyl fluoride, 20 g/ml leupeptin), and hPKC␣ was partially purified according to Birman et al. ( (23). Cells were visualized with a 63 ϫ 0.9 numerical aperture achroplan water immersion objective lens (Zeiss). The larger slit (100 m) was used, giving bright images with a 3.1 m axial resolution. Cells were loaded with the Ca 2ϩ -sensitive fluorescent probe fluo-3 by exposure to 50 M fluo-3 acetoxymethyl ester (fluo-3/ AM, Molecular Probes, Eugene, OR) by incubation for 30 min at 37°C in a humidified incubator. Fluo-3 was excited through a 488-nm band pass filter, and the emitted fluorescence was collected through a 515-nm barrier filter. [Ca 2ϩ ] i changes were expressed as the ratio F/F min , where F min was the minimum fluorescent intensity measured during the recording (30 images/s). Acquired data were then processed for analysis using Igor 3.14 software (Wavemetrics Inc., Lake Oswego, OR). Three separate experiments were performed, and in each experiment, a minimum of 10 fields with both single and contacting cells were analyzed for [Ca 2ϩ ] i changes. Fig. 1A shows that the transfected hPKC␣-GFP, visualized in the living cell by fluorescent microscopy, was found in the same subcellular compartments as is the rat endogenous PKC␣, as revealed by an anti-PKC␣ antibody. Both the rat endogenous PKC␣ and the hPKC␣-GFP were cytoplasmic under basal con- There was no translocation of PKC␣ in isolated cells regardless of the presence of the GFP tag. Scale, 1 cm ϭ 10 m. B, the GFP tag does not affect activity of hPKC␣. Protein kinase C catalytic activity was assessed by measuring the incorporation of 32 P from [␥-32 P]ATP into the S17R substrate in the presence of 10 g/ml PS and 10 M PMA. The experiment was performed in the presence or in the absence of 1.2 mM calcium. hPKC␣-GFP and its deletion mutant ⌬(V1)hPKC␣-GFP fusion proteins phosphorylate S17R, whereas GFP has no detectable kinase activity regardless of the conditions. Three separate experiments gave identical results. Western blot (WB) analysis of hPKC␣ and immunoprecipitated hPKC␣-GFP and ⌬(V1)hPKC␣-GFP (revealed with the anti-PKC␣ antibody) or GFP fractions (revealed with the anti-GFP antibody) indicates the amount of each protein used for PKC activity assay. hPKC␣ was extracted from Sf9 cells infected with a baculovirus containing the cDNA encoding hPKC␣. ditions, whereas under stimulation with 100 nM PMA for 60 min, both translocated to a restricted area of the plasma membrane, localized at the interface between apposed cells. Stimulation-induced translocation was never observed in isolated cells (at least 100 cells observed). In contrast, GFP was evenly distributed between the cytoplasmic and nuclear compartments (see Fig. 6B), and its localization was not affected by PMA treatment (data not shown).

The Exclusive Localization of hPKC␣ at the Cell-Cell Contacts upon PMA Stimulation Is Not Affected by the GFP Tag-
The fact that PKC␣ localization was not affected by the GFP tag on localization correlated with the lack of effect of the GFP tag on catalytic activity. This was shown by comparing the properties of hPKC␣ extracted from Sf9 cells to those of hPKC␣-GFP immunoprecipitated from transiently transfected GH3B6 cells. The contents of hPKC␣, hPKC␣-GFP and GFP in cell extracts and immunoprecipitates were first evaluated by Western blot analysis (Fig. 1B). The catalytic activities of hPKC␣-GFP and hPKC␣ were then measured in the presence of PMA and phosphatidylserine, in the presence or absence of Ca 2ϩ . As shown in Fig. 1B, the catalytic activities of both hPKC␣-GFP and hPKC␣ were increased upon Ca 2ϩ addition, whereas GFP alone was, as expected, catalytically inactive.
The GFP fusion proteins used in transient transfection experiments are schematically represented in Fig. 2A, and the Western blot shown in Fig. 2B shows that the translated products are of the expected size. Transfection efficiency approximated 25% in GH3B6 cells, although the efficiency was higher for the smaller sized constructs.
hPKC␣ Translocation to the Plasma Membrane Is Biphasic upon TRH Stimulation and Monophasic upon PMA Exposure- Fig. 3 illustrates the time-dependent translocation of hPKC␣-GFP to the plasma membrane of living contacting cells upon stimulation with 100 nM TRH. Translocation always occurred at cell-cell contacts but never in single cells, regardless of the duration of TRH stimulation. In agreement with recent publications (3, 8 -11, 22, 29), the stimulation of a seven transmembrane domain receptor (the TRH receptor in the present study) induced a rapid (8 s in all of the cells observed; n Ͼ 20) but transient translocation of hPKC␣-GFP (Fig. 3A). This first rapid phase was followed by a relocation of hPKC␣-GFP to the cytoplasm within 60 s. Unexpectedly, however, a second phase occurred within 1 h of stimulation, and hPKC␣-GFP was still located at the plasma membrane after 3 h (data not shown). Deletion of the V1 region (⌬(V1)hPKC␣-GFP) abolished the second phase but did not affect the first phase of translocation (Fig. 3B), suggesting a possible role for V1 in the mechanisms leading to the long lasting association of hPKC␣ with the membrane. It is noteworthy that V1 deletion has no effect on the catalytic activity of hPKC␣ (Fig. 1B). Under 100 nM PMA stimulation, translocation started after 10 min and was long lasting, as shown by the Western blot performed with cytosol (Fig. 3C, s) and membrane (m) after 1 h of stimulation. When V1 was deleted, the PMA-induced translocation was totally abolished, regardless of the duration of PMA treatment, as shown by Western blot analysis (Fig. 3C). These results suggest that the mechanisms underlying the early translocation induced by TRH are different from those involved in the late translocation induced either by TRH or by PMA.

An Increase in Intracellular Calcium Concentration Is Not Sufficient to Induce hPKC␣-GFP Translocation to the Plasma
Membrane-In order to establish whether the absence of translocation of PKC␣ in isolated cells could be attributed to a deficiency in TRH receptors, we measured the variations in intracellular calcium concentration in both isolated and apposed cells upon TRH stimulation. As shown in Fig. 4, TRH elicited a rise in calcium concentration of similar amplitude in apposed cells (Fig. 4B) and in isolated cells (Fig. 4C). All of the cells observed showed this intracellular calcium rise, and in particular all of the isolated cells. This implies that TRH receptors are indeed present in isolated cells; their absence cannot account for the lack of PKC␣ translocation. Intracellular calcium rise, which is known to be necessary for translocation, thus may not be sufficient for hPKC␣-GFP translocation in GH3B6 cells.

Removal of the Catalytic Domain Results in the Constitutive and Uniform Localization of the Regulatory Domain ϩ V3
Region at the Plasma Membrane-When inactive, native PKC␣ localizes exclusively in the cytoplasm. Activation results in the "opening" of the enzyme due to disruption of the interaction between the N-terminal pseudosubstrate sequence and the Cterminal catalytic site. Deletion of the catalytic domain may therefore unmask regions of the RD involved in translocation. We found that the RDϩV3-GFP (Fig. 5A) was spontaneously located at the plasma membrane of single as well as apposed cells, with no preferential localization at cell-cell contacts and no change in localization upon treatment with PMA (Fig. 5A) or TRH (data not shown). The results obtained by direct microscopic observation were confirmed by Western blot analysis (Fig. 5B).
The importance of Ca 2ϩ for the association of the RDϩV3 with the plasma membrane was evidenced in transiently transfected cells incubated for 15 min in the presence of the Ca 2ϩ chelator BAPTA/AM. In single cells, the treatment with BAPTA/AM induced a dissociation of the RDϩV3-GFP construct from the membrane. A similar phenomenon was observed in apposed cells, except that fluorescence persisted on cell-cell contacts (Fig. 5A). We concluded that accumulation of the RDϩV3 at the plasma membrane may result from both Ca 2ϩ -dependent and Ca 2ϩ -independent mechanisms and that it contains all of the sequences involved in the accumulation of PKC␣ at the plasma membrane, including at cell-cell contacts.
Inactive hPKC␣ Is "Sequestered" in the Cytoplasm: Involvement of the C2-V3 Region-Similarly to the RDϩV3-GFP construct, the RD-GFP construct spontaneously located at the plasma membrane of single (data not shown) as well as apposed cells (Fig. 6A). Likewise, in single cells, the treatment with BAPTA/AM induced a dissociation of the RD-GFP construct from the membrane, just as it did for the RDϩV3-GFP construct (data not shown). The same thing happened in apposed cells at the exception of the cell-cell contacts (Fig. 6A). However, Fig. 6A also shows that in the absence of Ca 2ϩ , the subcellular relocalization of the RD-GFP is not identical to that of the RDϩV3-GFP. In the presence of V3, the fluorescence relocates from the membrane to the cytoplasm whereas in the absence of V3, the fluorescence relocates both in the cytoplasm and the nucleus upon treatment of cells with BAPTA/AM. This prompted us to hypothesize that there is a sequestration of hPKC␣ in the cytoplasm and that this may involve the V3 region. When V3 alone was fused to GFP, or when the C-terminal half of C2 was added to V3, the fluorescence was both cytoplasmic and nuclear (Fig. 6B), as for GFP alone. However, when the entire C2 domain was fused to V3, the resulting fluorescence was exclusively cytoplasmic. It is important to note that the exclusion of fusion proteins from the nucleus is not size-dependent because C2ϩV3-GFP, which could not be found in the nucleus, is smaller (45 kDa) than RD-GFP (59 kDa), which is both cytoplasmic and nuclear. This means that the minimum sequence responsible for the cytoplasmic sequestration of PKC␣ is contained in a region containing C2 ϩ V3.
The V1 region Is Involved in hPKC␣ Localization at the Plasma Membrane-We have shown in Fig. 3, B and C, that V1 deletion abrogates long term TRH-and PMA-dependent translocation of hPKC␣-GFP to the plasma membrane but does not affect the first TRH-dependent phase of translocation or the selective accumulation of hPKC␣-GFP at the interface between apposed cells. Fig. 7A shows that removal of the catalytic domain in addition to the V1 deletion has no additional effect on translocation. Indeed, the targeting selectivity of ⌬(V1)RDϩV3-GFP was maintained upon TRH stimulation (a few seconds), and there was no translocation upon long term TRH or PMA stimulation; this was confirmed by Western blot analysis (Fig. 7B). Therefore, the catalytic domain is required neither for the first translocation phase nor for maintenance of targeting selectivity during this translocation phase. Identical results were obtained when the pseudosubstrate sequence was deleted in addition to V1 and the catalytic domain (data not shown).
The fact that the spontaneous presence of the RDϩV3 at the plasma membrane is abolished when V1 is deleted may indicate a major role for V1 in addressing hPKC␣-GFP to the plasma membrane, in the same manner as a nuclear localization sequence does. However, the results shown in Fig. 8A are not consistent with this hypothesis, because V1-GFP was uniformly distributed throughout the cell and was not present at the plasma membrane.
Alternatively, if V1 is a binding site for an anchoring protein, V1 overexpression may inhibit, at least partially, the translocation of the endogenous PKC␣. However, in V1-GFP expressing cells stimulated for 1 h with PMA, the endogenous PKC␣ translocated normally (Fig. 8A), a result confirmed by Western blot analysis (Fig. 8B). Thus, if V1 interacts with a protein required for the spatiotemporal localization of PKC␣, this association is not rate-limiting. DISCUSSION For any given protein, the process of subcellular translocation, which describes the fact that the protein goes from one cellular location to another, may be arbitrarily divided into several defined and controlled steps: the early signals, including that telling which route to take; the route taken; and the final destination. Proteins such as PKC that do not contain a peptide signal are cytoplasmic when inactive and require an active mechanism to be addressed to specific subcellular compartments upon stimulation. The present study aimed at understanding the structure-localization relationship of hPKC␣. To this end, intracellular displacements of various hPKC␣-GFP fusion proteins were assessed in living GH3B6 cells. Our main findings can be summarized as follows. 1) Upon physiological stimulation of the TRH receptor, the translocation of hPKC␣ to the plasma membrane displays two phases, each phase being controlled by different mechanisms. 2) Translocation is observed at cell-cell contacts, but not in isolated cells, although TRH induces a similar rise in intracellular calcium concentra- In control cells, the regulatory domain associated or not associated with the V3 region is mainly at the plasma membrane. Upon calcium chelation, RDϩV3-GFP redistributed in the cytoplasm, whereas RD-GFP redistributed in the cytoplasm and in the nucleus. Scale, 2 cm ϭ 10 m. B, GH3B6 cells transiently transfected with V3-GFP or associated with the C terminus of the C2 region (C 236 -289 -V3-GFP) or with the entire C2 domain (C2ϩV3-GFP). Cytoplasmic sequestration of GFP needed both the V3 and the C2 regions. Note that the presence of a fusion protein in the nucleus is not dependent on the size of the protein because RD-GFP, which can enter the nucleus, is larger than C2ϩV3-GFP, which is only cytoplasmic. Note that GFP itself was found uniformly distributed in the cells. Scale, 2.4 cm ϭ 10 m. The Early Signals-In our rat pituitary-derived experimental model, the GH3B6 cells, we showed that translocation and accumulation of PKC␣ (both endogenous and hPKC␣-GFP) only occurs in contacting cells, never in single cells. Furthermore, we showed that the absence of translocation in single cells is not due to the lack of functional TRH receptors, because intracellular Ca 2ϩ concentration does rise similarly in single and in apposed cells. The increase is rapid and transient, as previously reported (30). Therefore, the fact that there is no translocation in single cells suggests that the rapid translocation induced by TRH may require another signal in addition to Ca 2ϩ . This result challenges the dogma recently proposed by Oancea and Meyer (22), presenting "PKC as a molecular machine for decoding calcium and DAG signals." The authors showed that repetitive calcium spikes triggered by stimulation of the Fc⑀RI receptor induce a parallel repetitive translocation of GFP-tagged PKC␥ to the plasma membrane of rat basophilic leukemia 2H3 cells. The fact that we did not observe translocation of hPKC␣ at the plasma membrane of single GH3B6 cells, as it should be if the relationship between calcium and PKC translocation were sufficient, may be attributable to the different isoforms involved or to the different cell lines used. However, unlike PKC␣, which is a ubiquitous isoform, PKC␥ is a strictly brain-specific isoform. The fact that PKC␥ was transfected in cells that normally do not express this isoform may be relevant: in the absence of isoform-specific and tissue-specific PKC-interacting proteins, translocation may strictly depend on variations in DAG and Ca 2ϩ concentrations. The nature of the additional signal required for translocation of PKC␣ in GH3B6 contacting cells needs to be identified.
Because the RD alone spontaneously accumulates uniformly at the plasma membrane, the RD may contain the required membrane interacting sequences, possibly the C2 region that binds calcium (17,18) and the C1 region that binds DAG. In addition, the spontaneous accumulation of the RD-GFP at the plasma membrane also suggests that the increase in DAG and intracellular Ca 2ϩ concentrations required for disrupting the interaction between the pseudosubstrate and the catalytic site may not be required for "translocation." Which Route to Take?-The uniform accumulation of the RD at the plasma membrane, by contrast to the native protein accumulation at the interface between apposed cells, indicates that selectivity of translocation and sequestration in the cytoplasm have been lost. It is as if the RD alone ignored the signal telling which route to choose, despite the fact that the sequences capable of interacting with this unknown signal are most probably in the RD because (a) ⌬(V1)RDϩV3-GFP translocates at the interface of apposed cells upon stimulation with TRH, as does the endogenous PKC␣; and (b) RD-GFP localized at the interface between cells upon treatment with BAPTA/AM.
We have seen that RD and RDϩV3 are constitutively located at the plasma membrane. Upon Ca 2ϩ removal, RDϩV3 is shifted back to the cytoplasm, whereas RD alone is uniformly distributed between cytoplasm and nucleus (at the exclusion of the cell-cell contacts). This is compatible with V3 being involved in the cytoplasmic sequestration of hPKC␣ under basal conditions. This hypothesis is further supported by the finding that under basal conditions, ⌬(V1)RD is uniformly distributed within the cell, whereas ⌬(V1)RDϩV3 is exclusively cytoplasmic (data not shown). The V3 region is, however, necessary but not sufficient because cytoplasmic sequestration requires the C2 region. What mechanism could account for cytoplasmic sequestration? One possible explanation is that the C2-V3 region of hPKC␣ is a site of specific interaction with a cytoplasmic protein. This would be in agreement with the hypothesis of Mochly-Rosen and Gordon (12) speculating that PKC, in its inactive conformation, may interact with a receptor for inactive kinase C.
The Final Destination-How to explain the observation that hPKC␣ accumulates at the interface between apposed cells? It could be mediated either by a direct interaction with the membrane or via anchoring proteins. However, the lack of BAPTA/AM effect on RD accumulation at cell-cell contacts makes a direct interaction of hPKC␣ with the plasma membrane rather unlikely. Indeed, previous studies have shown that the interaction of PKC with phospholipidic vesicles is rapidly disrupted in the presence of a calcium chelator (31). The interaction between PKC␣ and anchoring proteins, such as RACK1, has been evidenced in the MDA-MB-231 cell line (32). Such an interaction is supposed to maintain an active pool of enzyme in the vicinity of its substrates. RACK1 is known to interact with the integrin ␤ subunit (33), which is part of the adhesion receptors that mediate attachment of cells to the extracellular matrix. The selective localization of PKC␣ at the cell interface and its absence from adhesion foci in GH3B6 cells is not in favor of the existence of a direct interaction between PKC␣ and RACK1. Thus, the presence of hPKC␣ at the inter- face between cells is more likely mediated via an as yet unidentified protein. The membranes located at the contact between cells contain the multiproteic complexes (gap junctions and the catenin/cadherin system) involved in cell-cell communication. Localization of PKC␣ in the apposed membranes may indicate a privileged role for PKC␣ in this type of communication, as already suggested by different studies showing for instance the direct role of PKC in the control of gap junction permeability (34,35).
Two Phases of Translocation-Up to now, PKC translocation has been widely accepted as being both monophasic and transient. Here, we report for the first time that PKC␣ can display a two phase kinetics upon chronic TRH stimulation: a fast (measured in seconds) and transient phase being followed by a slower (approximately 1 hour) and long lasting phase. Speculative mechanisms underlying the late translocation phase may include TRH receptor recovery from desensitization (36,37), or alternatively, it may indicate that a new protein needs to be synthesized or matured. In addition, one major difference between the first and the second phase of translocation is the time that hPKC␣-GFP remains at the targeting site: seconds for the first phase, hours for the second, implying that proteolytic-dependent down-regulation might occur only during the second phase.
In our study, one prominent result is that deletion of V1 from PKC␣ abolishes both the second phase of TRH-induced translocation and the PMA-induced translocation in GH3B6 cells. The potential role of V1 in translocation/accumulation had previously been addressed for PKC␥ transfected into rat basophilic leukemia 2H3 cells (22). The authors of that study showed that V1-deleted PKC␥ translocates faster than the wild type enzyme upon phorbol 12,13-dibutyrate stimulation. These opposite results could be explained by the different isoforms studied, which differ mainly in the variable regions (such as V1 or V3), the different cell lines used, or the absence of specific PKC-interacting proteins.
The results obtained with the ⌬(V1)RDϩV3 construct, that is, abolition of the spontaneous accumulation of the RDϩV3 at the plasma membrane, suggests that V1 may play a direct role during translocation. However, not only is V1 not a membrane targeting sequence by itself (because it does not address GFP to the membrane) but, in contrast to PKC⑀ (38), V1 overexpression does not prevent translocation of the endogeneous PKC␣. Therefore, if the role of V1 in targeting is exerted through an interaction with a protein, this interaction is not rate-limiting.
In conclusion, these observations as a whole provide a new insight into the spatiotemporal localization of hPKC␣, the complexity of this process reflecting the complexity of each step: sequestration, targeting, and accumulation. Our work provides arguments for 1) the existence of two phases of translocation upon stimulation by TRH, differently controlled and involving the V1 variable region; 2) the requirement of a signal other than calcium that together with calcium triggers translocation; and 3) the existence of a controlled cytoplasmic sequestration involving the C2 and V3 regions.