Importance of C1B domain for lipid messenger-induced targeting of protein kinase C.

The molecular mechanisms by which arachidonic acid (AA) and ceramide elicit translocation of protein kinase C (PKC) were investigated. Ceramide translocated epsilonPKC from the cytoplasm to the Golgi complex, but with a mechanism distinct from that utilized by AA. Using fluorescence recovery after photobleaching, we showed that, upon treatment with AA, epsilonPKC was tightly associated with the Golgi complex; ceramide elicited an accumulation of epsilonPKC which was exchangeable with the cytoplasm. Stimulation with ceramide after AA converted the AA-induced Golgi complex staining to one elicited by ceramide alone; AA had no effect on the ceramide-stimulated localization. Using point mutants and deletions of epsilonPKC, we determined that the epsilonC1B domain was responsible for the ceramide- and AA-induced translocation. Switch chimeras, containing the C1B from epsilonPKC in the context of deltaPKC (delta(epsilonC1B)) and vice versa (epsilon(deltaC1B)), were generated and tested for their translocation in response to ceramide and AA. delta(epsilonC1B) translocated upon treatment with both ceramide and AA; epsilon(deltaC1B) responded only to ceramide. Thus, through the C1B domain, AA and ceramide induce different patterns of epsilonPKC translocation and the C1B domain defines the subtype specific sensitivity of PKCs to lipid second messengers.

All PKCs possess an amino-terminal regulatory domain and a catalytic domain in the carboxyl terminus. The regulatory domain of the PKCs contains a variable region 1 (V1), a pseudosubstrate motif (PS), and a conserved region 1 (C1). The V1 of ⑀PKC has been reported to be a selective inhibitor of ⑀PKC translocation (11,12). In the resting state, the PS is bound in the active site of the catalytic domain, keeping the enzyme inactive by blocking the catalytic site. The binding of activators to the regulatory domain causes a conformational change which releases the PS from the active site and activates the enzyme (13). DG and phorbol ester binding have been localized to the C1 domain (2,8,14,15). Additionally, the C1 domain mediates protein-protein interactions: that of ⑀PKC binds actin (16 -18).
The C1 domain of cPKCs and nPKCs have two cysteine-rich loops (C1A and C1B), each consisting of ϳ50-amino acids including six cysteine and two histidine residues arranged in a zinc finger motif. The C1B of cPKCs and nPKCs showed strong phorbol esters binding, but all C1A except for ␥PKC showed very weak affinity for phorbol esters (19). GFP-tagged C1A-C1B or C1A translocated to the plasma membrane in response to receptor or phorbol esters stimuli, whether significant plasma membrane translocation of C1B was only observed in phorbol esters stimulation (20). In addition, distinct roles for the C1A and C1B domains in the activation of the enzyme have been shown (21). These results suggest that the C1A and C1B domains of PKCs are functionally distinct.
The activity of PKC can be regulated not only by DG and phorbor ester but also by other lipids such as arachidonic acid (AA) (22,23) and ceramide (24,25). Like DG/phorbol esters, these lipid second messengers also induce translocation of PKCs. Immunoblot analysis and immunocytochemistry in fixed cells have shown that AA induces translocation of ⑀PKC (26) and ceramide translocates ⑀PKC and ␦PKC from the plasma membrane to the cytoplasm (27). Using green fluorescent protein (GFP)-tagged PKCs and live cell imaging, we have shown that AA translocates ⑀PKC, but not ␦PKC, from the cytoplasm to the Golgi complex (28,29) and that ceramide translocates ␦PKC from the cytoplasm to the Golgi complex (30). However, little is known about the mechanism underlying these translocations. Here we identified the intramolecular domains of ⑀and ␦PKC that respond to ceramide and AA to clarify the molecular mechanisms responsible for the lipids-dependent translocation of nPKCs.

EXPERIMENTAL PROCEDURES
Materials-Arachidonic acid and C 6 -ceramide were purchased from Doosan Serdary Research Laboratories (Englewood Cliffs, NJ) and Molecular Probes, Inc. (Eugene, OR), respectively. 12-O-Tetradecanoylphorbol-13-acetate (TPA) was obtained from Sigma. All the other chemicals used were of analytical grade.
Cell Culture-COS-7 and CHO-K1 cells were purchased from the Riken cell bank (Tsukuba, Japan) and Health Science Research Re-* This work was supported by a grant from the Ministry of Education, Culture, Sports, Science and Technology in Japan, a grant-in-aid for Scientific Research on Priority Areas(C), Advanced Brain Science Project, from Ministry of Education, Culture, Sports, Science and Technology in Japan, the Uehara Memorial Foundation and Sankyo Foundation of Life Science, and the Hyogo Science and Technology Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Transfection of the GFP-tagged PKCs-CHO-K1 cells were transfected using 3 l of FuGENE TM 6 Transfection Regent (Roche Molecular Biochemicals) and 1 g of DNA according to the manufacturer's protocol. Transfected cells were cultured at 37°C for 16 -48 h prior for imaging.
Construction of Plasmids Encoding the GFP-tagged PKCs-The constructs encoding GFP-conjugated rat ⑀PKC (⑀PKC) and ␦PKC (␦PKC) were previously described (28,31). The cDNA for the GFP-tagged proteins used in these studies are diagrammed in Fig. 5; the primers used are shown in Table I. The cDNA encoding the amino-terminal deletion mutants of ⑀PKC were generated by PCR using BS 495 (rat ⑀PKC in pCR TM 2.1) (28) as the template. The primers were synthesized with BglII sites on the both 5Ј and 3Ј terminus to facilitate subcloning. cDNAs encoding domain-deleted and point-mutated ⑀PKCs were produced using the ExSite TM PCR-based Site-directed Mutagenesis kit (Stratagene) with BS495 as a template. Chimeras of ⑀PKC containing ␦C1B (⑀(␦C1B)) was produced by two-step PCR using two plasmids as templates at one reaction. For the first step, BS495 and BS751 (rat ␦PKC in pCR TM 2.1) (31) were used as the templates with ␦R845/⑀F882 as the primers using the ExSite TM PCR-based Site-directed Mutagenesis kit. The product of the first reaction was a chimera having the ␦PKC regulatory domain and the ⑀PKC kinase domain (BS758). For the second step, BS758 and BS495, and ⑀R718/␦F682 were used as the templates and the primers, respectively. Similarly, chimeras of ␦PKC containing ⑀C1B (␦(⑀C1B)) was generated by two-step PCR. BS495 and BS751 as the templates and ⑀R881/␦F846 as the primers were used for the first step to produce a chimera having the regulatory domain of ⑀PKC and the kinase domain of ␦PKC (BS759). For the second step, BS759/BS751 and ␦R683/⑀F720 were used as the templates and primers, respectively.
The PCR products for deletion mutants of ⑀PKC, point-mutated ⑀PKCs and ⑀(␦C1B) were digested with BglII, and subcloned into the BglII site of the EGFP expression vector (BS340). ␦(⑀C1B) was digested with EcoRI/BamHI and subcloned into the EcoRI/BglII sites of BS340. All PCR products were sequenced prior to use.
Confocal Microscopy-CHO-K1 cells transfected with the GFPtagged PKCs were cultured for 16 -48 h for maximal GFP expression. The media was then replaced with Ringer's solution composed of 135 mM NaCl, 5.4 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, and 10 mM glucose, pH 7.3. Translocation of the GFP-tagged PKCs was triggered by the addition of the various stimuli to the Ringer's solution to obtain the appropriate final concentration. All experiments were done at 37°C. The GFP fluorescence was monitored by confocal laser scanning fluorescent microscopy (Carl Zeiss, Jena, Germany) at 488-nm argon excitation with a 515-nm long pass barrier filter. Time series images were recorded before and after stimulation.
Fluorescent Recovery after Photobleaching Study (FRAP)-After recording 1 or 2 images, a 10 ϫ 10 pixel subregion of the cells was scanned with the maximal power of the 488-nm laser for 30 s to bleach the fluorescence. To monitor the recovery of fluorescence in the photobleached spot, a time series of 30 -50 images was taken with 2.25-s time intervals. The fluorescence intensity of the region was quantified for each image in the time series using LSM510 softwear (Carl Zeiss).
Co-detection of the Golgi Network and GFP-tagged ⑀PKC-Texas Red-conjugated wheat germ agglutinin was used to monitor the Golgi network. GFP-tagged ⑀PKC-transfected cells were stimulated with 100 M AA or 10 M C 6 -ceramide, the cells were fixed and treated with 0.3% Triton X-100 and 10% normal goat serum. 0.5 g/ml Texas Red-conjugated WGA (Molecular probes, Leiden, Netherlands) in PBS-T was then added to label the Golgi complex. The fluorescence of Texas Red and GFP were observed by a confocal laser scanning fluorescent microscopy. Texas Red was visualized using 588-nm argon excitation and a 590-nm long pass filter, GFP was detected at 488-nm excitation using a 510 -525 nm band pass filter.

RESULTS
Effects of Ceramide, AA, and TPA on ⑀PKC Translocation-Ceramide has been shown to translocate ␦PKC from the cytoplasm to the Golgi complex in HeLa cells (30), but the effect of ceramide on ⑀PKC translocation has not been examined. The ability of C 6 -ceramide (ceramide), a membrane permeable ceramide analog, to alter subcellular localization of ⑀PKC in CHO-K1 cells was investigated and compared with that of AA and TPA.
In resting cells, wild type ⑀PKC-GFP (⑀PKC) was detected throughout the cytoplasm, but was excluded from the nuclei. Within the cytoplasm, ⑀PKC was diffusely distributed with slight enrichment in the perinuclear region (Fig. 1A, before). After treatment with ceramide (10 M), ⑀PKC accumulated in the perinuclear region (Fig. 1A, left), peaking within 20 min and remaining for more than 60 min; the cytoplasmic fluorescence did not change significantly after ceramide stimulus. In contrast, the homogeneous fluorescence of ⑀PKC in the cytoplasm became heterogeneous within 1 min after AA addition (100 M) with a diffuse accumulation of fluorescence apparent in the perinuclear region (Fig. 1A, center). The timing of the AA-induced perinuclear accumulation was similar to that seen upon ceramide treatment, reaching a maximum within 20 min and remaining at least for 60 min. In contrast, TPA (1 M) translocated ⑀PKC from the cytoplasm to the plasma membrane within 10 min, where it remained for at least 60 min ( Fig. 1A, right).
The pattern of ⑀PKC concentration in response to ceramide resembles the Golgi complex staining induced by AA (Fig. 1B, bottom) (28). To test the hypothesis that ceramide induces translocation to the Golgi complex, CHO-K1 cells transfected with ⑀PKC were stimulated with ceramide (10 M), fixed, and the Golgi complex was visualized with Texas Red-conjugated wheat germ agglutinin (WGA) (Fig. 1B, upper). Intense fluorescence of Texas Red was seen in the perinuclear region (Fig.  1B, Cer, center). This staining resembled the perinuclear concentration of GFP (Fig. 1B, Cer, left). A merged image verified that the fluorescence of Texas Red and GFP co-localized in the perinuclear region (Fig. 1B, Cer, right). These results indicate that ceramide and AA both induced the translocation of ⑀PKC to the Golgi complex, although the pattern of ⑀PKC accumulation was slightly different (compare AA and Cer in Fig. 1B). The differences were not due to an effect of ceramide (10 M) or AA (100 M) on the structure of Golgi complex, since the shape of the organelle visualized by GFP-tagged galactosyltransferase (32) was not altered by 30 min treatment with each lipid (data not shown).
TPA Treatment and FRAP Identify Differences in the Association of ⑀PKC with the Golgi Complex after Ceramide and AA Treatments-To compare the relative strength of ⑀PKC-Golgi association in response to ceramide or AA, we tested the effect of subsequent TPA treatment on ⑀PKC localization. Ceramide was added to CHO-K1 cells for 10 min to induce ⑀PKC translocation to the Golgi complex (Fig. 2, upper). Subsequent treatment with 1 M TPA redistributed ⑀PKC from the Golgi complex to the plasma membrane within 25 min (Fig. 2, upper). Cells treated with AA followed by TPA retained significant perinuclear staining for more than 40 min after TPA stimulus although some ⑀PKC was translocated to the plasma membrane (Fig. 2, bottom). These results suggest that the AAmediated ⑀PKC-Golgi complex association is tighter than that induced by ceramide.
To further probe the interaction of ⑀PKC with the Golgi complex, we used FRAP analysis. After photobleaching in the Golgi complex, the fluorescence in bleached or unbleached area was measured at 2-8-s intervals for 1-3 min (Fig. 3, A and B). In ceramide-treated cells, the fluorescence of ⑀PKC in the photobleached area (Fig. 3A, blue) recovered to 80% within 30 s. The fluorescence in an unbleached Golgi complex area (Fig. 3A, yellow) did not change significantly over the course of the experiment. In contrast, the fluorescence in an unbleached region of the cytosol faded to 60% (Fig. 3A, red). These results suggest that after ceramide treatment, the fluorescence of the bleached area in the Golgi complex was recovered from the cytoplasm.
In cells treated with AA for 20 min, the fluorescence in the photobleached Golgi complex area (Fig. 3B, blue) recovered maximally by 60 s. However, this level was only 50% of the original signal. The recovery was accompanied by a corresponding decrease in the fluorescence level in an unbleaced area of the Golgi complex (Fig. 3B, yellow). In contrast to the result in the ceramide-treated cell, the fluorescence in the unbleached cytosol was not significantly altered (Fig. 3B, red), suggesting that recovery in AA-treated cells was the result of the redistribution of the ⑀PKC present in the Golgi complex.
AA and C 6 -ceramide Differently Regulate Translocation of ⑀PKC-To determine whether one lipid mediator could alter  the distribution of ⑀PKC induced by the other, cells were stimulated sequentially with AA and C 6 -ceramide. AA translocated ⑀PKC to the perinuclear region and heterogeneous fluorescence was detected in the cytoplasm (Fig. 4A). A subsequent application of C 6 -ceramide eliminated the accumulation of ⑀PKC around nucleus and produced homogeneous cytoplasmic staining within 20s. By 10 min, the perinuclear staining returned (Fig. 4A). Magnified images (bottom row, Fig. 4A) revealed that the Golgi complex staining elicited by ceramide and AA are different, suggesting that AA and ceramide selectively target ⑀PKC to different compartments of the Golgi complex. Interestingly, AA failed to alter the localization of ⑀PKC induced by ceramide (Fig. 4B). However, TPA was able to translocate ⑀PKC to the plasma membrane after sequential treatment of ceramide and AA (Fig. 4B), indicating that the ⑀PKC had not lost the ability to translocate.
C1B Is the Only Responsible Domain for the Translocation of ⑀PKC Induced by Ceramide and AA-We constructed cDNAs encoding a series of GFP-tagged deletion mutants of ⑀PKC to identify the domains of ⑀PKC required for the translocation by ceramide, AA, and TPA. Fig. 5A and Table I summarize the structures of the mutants and primers used to generate them. Immunoblotting of fusion proteins with anti-GFP antibody verified that molecular weight of each GFP-tagged mutant was appropriate and no significant degradation products were detected (Fig. 5B).
First, we compared the intracellular distribution of the deletion mutants with that of wild ⑀PKC in resting CHO-K1 cells. Although all deletion mutants were localized in the cytoplasm as was full-length ⑀PKC, differences in the intracellular distributions of some mutants were apparent (Fig. 6, before). For example, ⌬V1-PS and ⌬V1-PS-C1A were localized heterogeneously in the cytoplasm. When the regulatory domain was deleted (⌬V1-PS-C1A-C1B), the GFP fluorescence was homo-geneous in the cytoplasm with no accumulation in the perinuclear region. ⌬PS was localized heterogeneously in the cytoplasm with prominent accumulation in the perinuclear region. In contrast, deletion of ⌬C1A, ⌬C1B, or the entire C1 did not significantly change the distribution compared with ⑀PKC.
We further studied the ceramide-and AA-induced translocation of ⑀PKC mutated in the C1A and C1B domain (33). Mutation of 11th proline residue to glycine in the C1A or C1B domain of ␦PKC decreases the affinity of PDBu binding; mutation of 17th cysteine to glycine abrogates PDBu binding. Thus, we created two C1A mutants of ⑀PKC (P180G, C186G) and two C1B mutants (P253G, C259G) which are predicted to weaken or lack PDBu binding to C1A and C1B domain (Fig.  5A). These mutants had the predicted molecular weights (Fig.  5B). Before the stimulation, the two C1A mutants (P180G, C186G) were expressed heterogeneously in the cytoplasm with some accumulation of fluorescence present in the perinuclear region (Fig. 7, before). The C1B mutants (P253G, C259G) were homogeneously distributed throughout the cytoplasm (Fig. 7, before). The C1A mutants (P180G, C186G) translocated similarly to wild type in response to ceramide or AA (Fig. 7). In contrast, distribution of the C1B mutants (P253G, C259G) were not altered by ceramide (10 M) or AA (100 M) treatment. TPA (1 M) translocated all mutants to the plasma membrane with a pattern similar to that of wild type (Fig. 7). These results confirm that the translocation induced by TPA can be mediated through either C1A or C1B, but that the C1B domain is essential for AA-and ceramide-induced translocation.
C1B Domain of ⑀PKC or ␦PKC Determines the Sensitivity to AA-induced Translocation of the PKC Subtypes-Ceramide translocates both ⑀PKC and ␦PKC (30); AA translocates ⑀PKC but not ␦PKC (29). Our data demonstrate that the C1B domain of ⑀PKC is important for the ceramide-and AA-induced translocation (Figs. 6 and 7). Taken together, these results suggest that the C1B domains of ⑀PKC and ␦PKC determine their sensitivity to ceramide and AA. To test this hypothesis, we determined the effect of ceramide and AA on the translocation of GFP-conjugated chimeras of ⑀PKC and ␦PKC. Chimeras of ⑀PKC containing the C1B domain of ␦PKC (⑀(␦C1B)) and ␦PKC having the C1B domain of ⑀PKC (␦(⑀C1B)) were made as described under "Experimental Procedures." Western blots of the expressed chimeras with anti-GFP antibody showed the appro-priate molecular weights and no degradation products were detected (Fig. 5B).
Both ⑀(␦C1B) and ␦(⑀C1B) were expressed in the cytoplasm and enriched in perinuclear structures (Fig. 8). Ceramide induced translocation of both chimeras, as well as ⑀PKC and ␦PKC, to the perinuclear region (Fig. 8). In contrast, ␦(⑀C1B) and ⑀PKC, but not ⑀(␦C1B) and ␦PKC, accumulated in the perinuclear region upon addition of AA (100 M). Additionally, like ⑀PKC, ␦(⑀C1B) could be concentrated in the Golgi complex by sequential treatment with ceramide followed by AA (data not shown). Taken together, these results demonstrate that the C1B domain of ⑀PKC is responsive to both ceramide and AA while the C1B domain of ␦PKC mediates ceramide, but not AA-stimulated translocation. Thus, although the C1B domains of ⑀PKC and ␦PKC are very homologous, subtle differences in their sequences and/or structures determine their differential sensitivity to AA.

FIG. 5. The schematic structures of the ⑀PKC constructs (A) and their molecular weight (B).
A, the left column shows schematic composition of ⑀PKC and mutants: V1 (variable region 1), PS (pseudosubstrate), C1 (conserved region 1). The primers used to produce respective constructs are listed in the right column. Their sequences were provided in Table I. B, COS cells were transfected with the indicated plasmids and the molecular weight of the proteins were verified by Western blotting technique using an anti-GFP antibody. The predicted molecular weights are listed in the right column.

DISCUSSION
The importance of ceramide and AA as lipid messengers has only recently begun to be appreciated. Ceramide is involved in such processes as cell differentiation (34), outgrowth of neurons (35), apoptosis (27,36,37), and long-term depression of synaptic transmission (38). AA has been shown to act as a retrograde transmitter (39) in the generation of long-term potentiation, as a regulator of ion channels (40), a mediator of cell death (25,41) and is known to be necessary for superoxide generation (42). Both ceramide and AA can regulate the activity and/or translocation of PKC. AA activates ⑀PKC but not ␦PKC in vitro (22), and inhibits the ceramide-induced activation of PKC (25). O'Flaherty (43) demonstrated that ␣PKC, ␤PKC, and ␦PKC can be translocated by low concentrations of AA. In contrast, Oancea et al. (20) reported that AA inhibits translocation of the C1A domain of ␥PKC. Ceramide translocates ␦PKC and ⑀PKC from the plasma membrane to the cytoplasm (27), and ␣PKC from the cytoplasm to the membrane (24). Taken together, these reports suggest that both ceramide and AA play important roles in signal transduction and implicate their involvement in the regulation of subtype-specific activation or translocation of PKCs.
In this study, we showed that ceramide translocates ⑀PKC from the cytoplasm to the perinuclear region and identified this region as the Golgi complex by WGA staining (Fig. 1). We have previously shown that AA also induces the translocation of ⑀PKC to the Golgi complex (Fig. 1B) (28). In those studies, 10 M ceramide and 100 M AA were used to detect translocation of PKCs clearly and constantly, although the translocation to the Golgi complex could be detected even at 25 M AA and 1 M ceramide. The concentrations of these lipids might be relatively higher than that of physiological condition. It, however, is noteworthy that ⑀PKC is translocated to the Golgi complex when ceramide is generated by receptor stimuli with tumor necrosis factor-␣ as seen in the case of exogenous ceramide stimulation (data not shown), and that ⑀PKC accumulates in the Golgi complex in the brain (data not shown). These findings suggest that translocation of ⑀PKC to the Golgi complex occurs under physiological conditions.
Although both ceramide and AA translocated ⑀PKC from the cytoplasm to the Golgi complex, the pattern of localization was subtly, but distinctly, different. First, in ceramide-treated cells ⑀PKC was concentrated in the well defined Golgi complex with uniform distribution in the cytosol. In contrast, upon AA treatment ⑀PKC accumulated in a diffuse pattern around the nucleus with heterogeneous fluorescence in the cytosol. Second, TPA application after ceramide or AA revealed differences in the dissociation of ⑀PKC from the Golgi complex. The ceramidestimulated interaction of ⑀PKC with the Golgi complex was transient, as shown by the TPA-induced relocalization from the Golgi complex to the plasma membrane. On the other hand, interaction of AA-stimulated ⑀PKC was strong enough to resist being translocated by TPA stimuli. Third, FRAP analysis also revealed distinct interaction of ⑀PKC with the Golgi complex. The fact that, in ceramide-treated cells, fluorescence recovery in the Golgi complex was coincident with decreased cytosolic fluorescence suggests the ⑀PKC exchanges with the cytosolic pool. As staining in the unbleached regions of the Golgi complex did not change, it is unlikely that there is significant movement of ⑀PKC in the Golgi complex in response to ceramide. On the other hand, after AA treatment, the recovery came from unbleached regions of the Golgi complex rather than the cytosol. This suggests that AA mediates a tight association of ⑀PKC with the Golgi complex resulting in low exchange with cytosolic ⑀PKC pools. Finally, AA-translocated ⑀PKC was sensitive to redistribution by ceramide but AA did not alter the ceramide-translocated ⑀PKC. This indicates that AA-mobilized ⑀PKC is responsive to ceramide, but the ceramide-treated ⑀PKC cannot be further translocated by AA. Taken together, these differences imply that distinct mechanisms are involved in the translocation of ⑀PKC mediated by ceramide and AA. Similarly, different effects of ceramide and AA on PKC translocation have been reported for ␥PKC-C1A (20). In those studies, pretreatment with AA reduced the DG-induced translocation of ␥PKC-C1A to the plasma membrane but pretreatment with ceramide had no effect on the translocation.
To identify the domains of ⑀PKC necessary for AA-and ceramide-induced ⑀PKC translocation, we constructed a series of deletion mutants and studied their translocation characteristics. Loss of the V1, PS, and/or C1A domains did not alter translocation in response to ceramide or AA as compared with ⑀PKC. However, deletion of the C1B domain rendered the mutants insensitive to both ceramide and AA. These results indicate that the C1B domain is necessary for the translocation induced by both ceramide and AA. Although the mechanisms causing the distinct translocation are unknown, they may include differences in phosphorylation, interaction partners, and/or specific conformation changes.
The fact that the C1B, but not the C1A domain, is involved in the AA-and ceramide-induced translocation suggests that C1A and C1B have different roles in translocation. Even in the case of TPA, difference between C1A and C1B was observed. Unlike ceramide and AA, TPA induced translocation of both ⌬C1A and ⌬C1B but the mutants lacking both C1A and C1B (⌬V1-PS-C1A-C1B and ⌬C1A-C1B) were insensitive to TPA, indicating that either C1A or C1B can mediate TPA-induced translocation. However, the translocation of ⌬C1B was weaker than that of ⌬C1A (Fig. 6). This is consistent with the report that the C1B domain of ⑀PKC has higher affinity for phorbol esters than the C1A domain (19). In addition, several reports suggest distinct contributions of C1A and C1B domain in the regulation of PKC. For example, Shieh et al. (44,45) used mutants of ␣PKC lacking either C1A or C1B and showed no differences in TPA stimulated activity, suggesting that TPA regulates ␣PKC activity via either C1A or C1B. In contrast, mezerein regulation occurs predominantly via the C1A. Second, Bogi et al. (46) reported that translocation of ␦PKC by PMA requires the C1B domain but not C1A, although C1A and C1B domains of ␣PKC have equivalent roles for the PMA-induced translocation. Finally, the ␥PKC-C1A fragment was preferentially translocated to the plasma membrane compared with the ␥PKC-C1B or ␥PKC-C1AC1B fragment upon treatment of rat basophilic leukemia cells with IgE or ligands of PAF receptor. (20). Thus, there is a considerable body of literature consistent with our findings that C1A and C1B domains differentially regulate PKC translocation.
We used point mutations in the C1A (P180G and C186G) and C1B (P253G and C259G) domains to confirm that the C1B domain is responsible for the AA-or ceramide-induced translocation, and that either the C1A or the C1B domain is sufficient for the TPA-induced translocation. The proline mutants have a decreased affinity for PDBu, and the cysteine to glycine mutation eliminates PDBu binding (33). Ceramide and AA translocate the C1A mutants, but not the C1B mutants; TPA translocates all. Collectively, these results provide strong evidence that the ⑀C1B domain is required for ceramide-and AA-stimulated translocation, while TPA has a less stringent  requirement, needing only one cysteine-rich loop of C1 domain, in either C1A or C1B, for membrane localization.
Like ⑀PKC, ceramide induces the translocation of ␦PKC from the cytosol to the Golgi complex (30). Unlike ⑀PKC, ␦PKC is not sensitive to AA (29). What then, are the differences between ⑀PKC and ␦PKC? Our results show that the ⑀C1B is necessary for AA-induced translocation of ⑀PKC. Can the differences between the C1B domains of ⑀PKC and ␦PKC account for their differential sensitivity to AA? If so, the chimera of ␦PKC having the C1B domain of ⑀PKC, (␦(⑀C1B)), should translocate in response to AA, but the chimera of ⑀PKC having the C1B domain of ␦PKC, (⑀(␦C1B)), should not. ␦(⑀C1B), but not ⑀(␦C1B), translocated in response to AA. This difference is not due to a general nonresponsiveness of the ⑀(␦C1B) as it does translocate in response to ceramide (Fig. 8). Instead the data suggests inherent differences between the ␦C1B and ⑀C1B. Specifically, our data demonstrates that ⑀C1B domain responds to both ceramide and AA, but ␦C1B is sensitive only to ceramide. These differences in the C1B domain may determine the subtypespecific responses of ⑀and ␦PKC to lipid second messengers.
How the differences between ⑀PKC and ␦PKC to the lipid second messengers contribute their physiological roles? For example, AA is thought to be one of retrograde transmitters (39), and AA indeed facilitates long-term potentiation by the enhancement of synaptic transmission in the hippocampus (47)(48)(49). Enzymologically, ⑀PKC is activated with AA even in the absence of DG, although ␦PKC is not activated with AA at all (22). ⑀PKC is enriched in hippocampus and cerebral cortex and is localized mainly in the presynaptic terminals (50). Taken together, AA-induced activation of ⑀PKC, but not ␦PKC might be in involved in the expression of hippocampal longterm potentiation. On one hand, receptor stmulations with ␥-interferon, tumor necrosis factor, and vitamin D 3 result in the production of ceramide, leading to apoptosis (27,36,37) or cell differentiation (34). It is possible that these physiological phenomena via ceramide are mediated by ␦or ⑀PKC. In fact, Sawai et al. (27) reported that ceramide translocated ␦and ⑀PKC, not ␣, ␤2, ␥, nor PKC, to the cytoplasm, resulting in apoptosis in human leukemia cells (27). These results indicate that not only DG but also ceramide and AA regulate the activity and distribution of each PKC subtype, contributing to the subtype-specific physiological roles in long-term potentiation or apoptosis. In other words, even though several PKC isoforms are expressed in the same cell, each subtype of PKC can be regulated by specific activators and play a subtype-specific role in various signal transduction.
In conclusion, ceramide and AA translocate ⑀PKC to the Golgi complex by distinct mechanisms involving the C1B domain. In contrast, TPA requires only C1A or C1B domain for translocation. The subtle differences in the C1B domains of ⑀PKC and ␦PKC apparently account for their differential sensitivity to AA. These results indicate that different domains of PKC mediates translocation in response to different second messengers and the distinct characteristics of the domain determine the subtype-specific translocation, thereby contributing to the subtype-specific function. Although ceramide translocated both chimeras as same as ⑀PKC and ␦PKC, AA translocated ⑀PKC and ␦(⑀C1B), but not ⑀(␦C1B) nor ␦PKC. Constructs that translocated are designated "ϩ," while those that did not move were scored as "Ϫ." Scales, 5 m.