PICK1, an Anchoring Protein That Specifically Targets Protein Kinase Cα to Mitochondria Selectively upon Serum Stimulation in NIH 3T3 Cells*

PICK1 binds to protein kinase Cα (PKCα) through the carboxylate-binding loop in its PDZ (PSD95/Disc-large/ZO-1) domain and the C terminus of PKCα. We have previously shown that PICK1 modulates the catalytic activity of PKC selectively toward the antiproliferative gene TIS21. To investigate whether PICK1 plays a role in targeting activated PKCα to a particular intracellular compartment in addition to regulating PKC activity, we examine the localization of PICK1 and PKCα in response to various stimuli. Double staining with organelle markers and anti-rPICK1 antibodies reveals that PICK1 is associated with mitochondria but not with endoplasmic reticulum or Golgi in NIH 3T3 cells. Deletion of the PDZ domain impairs the mitochondria localization of PICK1, whereas mutations in the carboxylate-binding loop do not have an effect, suggesting that PICK1 can bind PKCα and mitochondria simultaneously. Upon serum stimulation, PICK1 translocates and displays a dense ring-like structure around the nucleus, where it still associates with mitochondria. A substantial portion of PKCα is concomitantly found in the condense perinuclear region. The C terminal-deleted PKCα fails to translocate and remains a diffuse cytoplasmic distribution, indicating that a direct interaction between PICK1 and PKCα is required for PKCα anchoring to mitochondria. 12-O-Tetradecanoylphorbol-13-acetate stimulation, in contrast, causes translocation of PKCα to the plasma membrane, whereas the majority of PICK1 remains in a cytoplasmic punctate pattern. Deletion at the C terminus of PKCα has no effect on 12-O-tetradecanoylphorbol-13-acetate-induced translocation. These findings indicate a previously unidentified role for PICK1 in anchoring PKCα to mitochondria in a ligand-specific manner.

Targeting of signaling molecules to specific intracellular sites through interactions with anchoring proteins allows activation of particular pools of protein kinases and phosphatases and thus plays a crucial role in determining the specificity of signal transduction cascades (1,2). The anchoring proteins not only bring kinases/phosphatases to the proximity of their substrates but can also modulate the activity of these enzymes upon binding (3).
Protein kinase C (PKC) 1 is a family of at least 11 isozymes that have been implicated in a variety of cellular responses (4,5). Upon activation, each isozyme differentially translocates to distinct subcellular structures including plasma membrane, cytoskeleton, and other organelles (1,6). Translocation is a cell type-specific event. In addition, different stimuli cause PKC isozymes to move to different intracellular sites in the same cell (7,8). A number of proteins, including RACKs, myristoylated alanine-rich C kinase proteins (MARCKs), and PICKs, have been found to bind activated PKC (9 -11), but yet only a few are shown to be responsible for selective targeting of activated PKC isozymes to particular subcellular compartments in cells. RACK1 (receptors for activated protein kinase C) selectively binds the active form of ␤IIPKC (12,13). Treatment of Chinese hamster ovary cells with phorbol 12-myristate 13-acetate causes movement of ␤IIPKC and RACK1 to the same intracellular sites that resemble Golgi apparatus (13). Peptides that prevent binding of ␤PKC to RACK1 inhibit insulin-induced ␤PKC translocation and function in Xenopus oocytes and phorbol 12-myristate 13-acetate-induced hypertrophy in cardiac myocytes (12,14), demonstrating a correlation between PKC redistribution and cellular responses and the importance of anchoring proteins in mediating PKC functions. Activated PKC⑀ co-localizes with the coatomer protein ␤Ј-COP in cardiac myocytes and binds to Golgi membranes in a ␤Ј-COP-dependent manner. ␤Ј-COP is thus identified as a PKC⑀-selective RACK (15). PICK1, a PDZ-containing protein, is first cloned as a PKC␣binding protein by the yeast two-hybrid screening (11). The association of PICK1 and PKC␣ was subsequently demonstrated by different approaches including co-immunoprecipitation (16 -18). PDZ domains are protein-interacting motifs implicated in association with plasma membrane, cell-cell junctions, cytoskeletal proteins, and signaling molecules (19,20). Studies have revealed a complexity in PDZ-target interactions (19,20). PICK1 selectively binds to PKC␣ through interaction with the QSAV sequence at the extreme C terminus of PKC␣ (16). The carboxylate-binding loop within the PDZ domain of PICK1 is required for the interaction (16). In addition to PKC␣, PICK1 interacts and co-localizes with several membrane proteins including Eph receptor tyrosine kinases and ephrin-B ligands (21), AMPA receptor GluR2 (22) metabotropic receptor mGluR7a (17), the dopamine transporter (23), and ion channels (24). Functionally, PICK1 can induce clustering of Eph receptor and its ligands (21), AMPA receptors (22), and mGluR7a (17) in heterologous expression systems. PICK1 is also shown to target PKC␣ to AMPA receptor clusters in hippocampal neurons (18) and to regulate phosphorylation of mGluR7 by PKC (17). It is therefore proposed that PICK1 plays a role in synaptic transmission mediated by PKC␣ in the central nervous system and that the function of PICK1 depends on the cell types and its interacting partners.
Functions of PICK1 in cells other than neurons are not yet documented; however, certain features of PICK1 suggest that functions of PICK1 should not be limited to synaptic transmission in neurons only. PICK1 interacts with, in addition to neuronal membrane-bound proteins, class I ADP-ribosylation factors, which are essential for vesicle formation in Golgi apparatus (25), and the immediate-early gene TIS21, whose expression is induced by a variety of extracellular stimuli including those known to activate PKC (26,27). Moreover, phosphorylation of TIS21 by PKC is selectively modulated by PICK1 (27). What further supports this notion is that the message and protein of PICK1 are ubiquitously expressed in all tissues examined (11,22). Thus, a full understanding of PICK1 functions will require further investigation in cells other than neurons.
In this study, we examine the intracellular localization of endogenous PICK1 in NIH 3T3 cells and its role as a PKC␣ targeting component in response to different stimuli. We show for the first time that PICK1 is specifically localized to mitochondria and that an intact PDZ domain is essential for the mitochondria localization. We demonstrate that activated PKC␣ co-localizes with PICK1 selectively upon serum stimulation and that the co-localization requires a direct interaction between these two proteins. This indicates that PICK1 targets PKC␣ to mitochondria and is likely to be responsible for mediating PKC␣ functions associated with mitochondria. 12-O-Tetradecanoylphorbol 13-acetate (TPA) treatment causes a translocation of PKC␣ to the plasma membrane that appears to be independent of PICK1 binding. These findings suggest that the differential translocation of PKC isozymes upon distinct stimulations may be achieved through differential responses of targeting molecules such as PICK1 to the extracellular stimuli.
The pcDNA3PKC␣ plasmid was also used as a template for construction of the pcDNA3PKC␣dQSAV plasmid by PCR.
Intracellular Distribution of Endogenous PICK1 Proteins-NIH 3T3 cells were routinely cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. For immunostaining, the cells were seeded on poly-D-lysine-coated coverslips 1 day before experiments, fixed in 4% paraformaldehyde for 5-10 min, permeabilized with Tris-buffered saline, pH 7.4, and 1% bovine serum albumin containing 0.1% Triton X-100 for 30 min and then postfixed with acid ethanol for 10 min, and the non-specific binding was further blocked in Trisbuffered saline containing 0.1% Triton X-100 for 60 min. Cells were then incubated overnight at room temperature with the polyclonal anti-rPICK1 antibody (27) diluted in Tris-buffered saline containing 0.01% Triton X-100 (1:50 to 1:500), washed with Tris-buffered saline (pH 7.4) for three times, and incubated with either rhodamine-or Cy3-conjugated goat anti-rabbit IgG (1:600 and 1:400, respectively) (Jackson ImmunoResearch Laboratories) for 30 min to visualize PICK1. The samples were counterstained for DNA with Hoechst 33258 (1 g/ml) for 30 min along with the secondary antibodies. Coverslips were mounted on slides with Fluoromount G (Southern Biotechnology Associates, Inc.). The cells were imaged using a fluorescent microscope (Olympus BX50) or a Leica TCS SP2 spectral confocal and multiphoton system. The image processing was performed using the Meta View software (Universal Imaging Corp.).
The mitochondria were identified by staining with a monoclonal anti-cytochrome oxidase subunit I (a protein located in the mitochondria inner membrane) antibody, with the antibody diluted 1:50 (Molecular Probes), and the ER was identified by staining with a monoclonal anti-Bip/Grp78 (an ER-resident protein) antibody, with the antibody diluted 1:25 (BD Transduction Laboratories). The secondary antibody consisted of FITC-conjugated anti-mouse IgG antibodies, diluted 1:150 to 1:175 (Jackson ImmunoResearch Laboratories). For visualization of Golgi, cells were transfected with pEYFP-Golgi (Clontech) in which the EYFP (enhanced yellow fluorescent protein) was fused to an anchoring signal peptide that targets the fusion protein to the trans-medial region of the Golgi apparatus. Fluorescence from EYFP was observed directly on a fluorescent microscope. In some cases, cells were incubated with brefeldin A (10 g/ml), which disrupted Golgi structures, for 30 min to confirm the Golgi localization of EYFP. The endogenous RACK1 protein was detected using monoclonal mouse anti-RACK1 antibodies, diluted 1:50 (BD Transduction Laboratories).
Stimulation of Cells-NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum to 70 -80% confluency, serum-starved overnight, and then stimulated with either 20% fetal bovine serum or TPA (Sigma) (200 ng/ml) for various times, as indicated.
Transfection and Detection of Heterologously Expressed Proteins-NIH 3T3 cells were seeded on coverslips overnight and transfected with various plasmids using LipofectAMINE Plus reagent (Invitrogen) in accordance with the manufacturer's instruction. The wild type and mutant FLAG-tagged rPICK1 proteins were detected with anti-FLAG polyclonal antibodies (1:50) (Abcam Ltd.) and visualized with rhodamine-conjugated anti-rabbit IgG antibodies (1:600) (Jackson ImmunoResearch Laboratories). In some instances, the anti-cytochrome oxidase subunit I monoclonal antibody (1:50) and the FITC-conjugated antimouse IgG antibody (1:200) were used simultaneously to locate mitochondria. The PKC␣ was detected with monoclonal anti-PKC␣ antibodies (BD Transduction Laboratories), which recognize PKC␣ specifically.
Antibodies-The rabbit polyclonal anti-rPICK1 antibody was generated against the recombinant rPICK1 protein as described previously (27). The mouse monoclonal anti-FLAG M5 antibody was from Sigma, and the rabbit polyclonal anti-FLAG epitope antibody was from Abcam Ltd. Monoclonal mouse anti-RACK1, monoclonal anti-Bip/Grp78, and monoclonal anti-PKC␣ antibodies were from BD Transduction Laboratories. The mouse monoclonal anti-cytochrome oxidase subunit I antibody was from Molecular Probes. Rhodamine-, FITC-, and Cy3-conjugated anti-rabbit or anti-mouse IgG antibodies were from Jackson ImmunoResearch Laboratories.

The PICK1 Protein Resides in Cytoplasm in a Punctate
Pattern-To examine the intracellular distribution of PICK1 in NIH3T3 cells, we have performed immunofluorescence staining using anti-rPICK1 antibodies. The PICK1 protein resided mainly in the cytoplasm with a punctate pattern (Fig. 1A). The sera preabsorbed with recombinant rPICK1 did not detect PICK1 any more (Fig. 1B), suggesting a specific recognition to PICK1. At higher magnifications, PICK1 protein typically exhibited as a rod-like network structure in the cytoplasm, with more concentration in a broad area surrounding the nucleus without significant indication of plasma membrane association (Fig. 1C). These results suggested that the PICK1 protein was associated with subcellular compartments in unstimulated cells. Similar distribution patterns were also observed in Rat-1 cells, H460 and H1299 (data not shown), indicating a common distribution pattern of PICK1 in these cells. The well characterized, PKC␤II-specific targeting protein RACK1 displayed a distinct diffuse distribution pattern throughout the cytoplasm (Fig. 1D). Our results suggested that these two PKC-binding proteins have distinct roles in mediating PKC functions.
The PICK1 Protein Is Associated with Mitochondria but Not with ER or Golgi Apparatus-To identify which organelles the PICK1 protein is associated with, we localized ER, Golgi, and mitochondria in NIH 3T3 cells using organelle-specific markers. Mitochondria, detected with the anti-cytochrome oxidase subunit I antibody, displayed a rod-like or network structure in the cytoplasm, which was identical to that of the PICK1 protein ( Fig. 2A). The structure was also recognized by the antibody to cytochrome c, an intermembrane resident of mitochondria, or Mito Tracker dye, whose uptake depends on mitochondria membrane potential in live cells (data not shown). The mitochondria and PICK1 signals overlapped almost completely, confirming the mitochondria localization of the PICK1 protein ( Fig. 2A). Staining of NIH 3T3 cells with the anti-GRP78 (an ER-resident protein) antibody displayed a fine punctate/reticular pattern surrounding the nucleus, a typical ER distribution (Fig. 2B). The pattern was distinct from that of PICK1, indicating that PICK1 did not associate with ER. The EYFP-Golgiexpressing cells, which expressed enhanced yellow fluorescent protein fused to a signal sequence targeting to the trans-medial Golgi, revealed intense fluorescence of a compact juxtanuclear structure, a characteristic feature of Golgi apparatus (Fig. 2C). Cells treated with brefeldin A, which disrupted Golgi structure, displayed a dramatic change of EYFP fluorescence into a diffuse distribution in the cytoplasm (Fig. 2C). The distribution of PICK1 remained unchanged in cells treated with brefeldin A (Fig. 2C). Taken together, these results indicated that PICK1 resided in mitochondria but not in ER or the Golgi apparatus.
The PDZ Domain-deleted rPICK1 Fails to Associate with Mitochondria-In addition to the PDZ domain, primary sequence analysis of PICK1 also predicts the existence of a coiledcoil domain (amino acids 139 -166) and an acidic region (amino acids 380 -390). Since no known mitochondria targeting sequence has been identified in rPICK1, we speculated that mitochondria association might be mediated by an interaction with mitochondria-resident protein(s) residing in the outer membrane. To examine which sequence region is required for the mitochondria localization of PICK1, we transfected NIH 3T3 cells with pCMV2FLAG-rPICK1, pCMV2FLAG-rPICK1 Ala-142-Ser-416 , or pCMV2FLAG-rPICK1 Met-1-Leu-352 for expression of the wild type and truncated rPICK1s with a FLAG epitope tag at the N terminus. Their intracellular distribution was examined by indirect immunofluorescence using anti-FLAG antibodies as primary antibodies and fluorophoreconjugated anti-mouse IgG antibodies as secondary antibodies. The wild type FLAG-rPICK1 displayed a punctate pattern in the cytoplasm resembling the one that was observed for endogenous PICK1 and was found to be coincident with mitochondria as detected with antibodies against cytochrome oxidase subunit I (Fig. 3), whereas the PDZ-deleted rPICK1 Ala-142-Ser-416 , which still contains the coiled-coil domain, displayed a dense, juxtanuclear pattern, mostly concentrated in one or two regions that were not co-localized with the mitochondrial marker cytochrome oxidase (Fig. 3) indicating a loss of mitochondria association. The rPICK1 Met-1-Leu-352 , which contains an intact PDZ domain but lacks the C-terminal 64 residues, including the acidic region, displayed a mitochondria distribution pattern similar to the wild type (Fig. 3). These results indicated that the PDZ domain was essential for mitochondria association of PICK1, whereas the C-terminal region was dispensable and the coiled-coil region was not able to sustain mitochondria association.
Many PDZ domains bind directly to the C-terminal ends of their target molecules through the carboxylate-binding loop. Less commonly, PDZ domains bind internal sequences rather than the C-terminal sequences (19,20,28). To test whether association of PICK1 with mitochondria is mediated by the carboxylate-binding loop, lysine 27 and aspartate 28 residues, which were shown to be essential for PICK1-PKC␣ interactions (16), were substituted with alanines. Interestingly, the K27A,D28A mutant rPICK1 displayed a distribution pattern similar to that of the wild type and was coincident with mitochondria markers (Fig. 3). These results indicated that the lysine and aspartate residues in the carboxylate-binding loop were not essential for mitochondria association of PICK1 and suggested that PICK1-mitochondria association may be mediated through a non-conventional PDZ-ligand interaction.
The PICK1 Protein Displays a Dynamic Redistribution and PKC␣ Concomitantly Translocates and Co-localizes with PICK1 upon Serum Stimulation-Activation of the protein kinase C family is accompanied with intracellular redistribution, and each isozyme displays a distinct redistribution pattern. To investigate whether PICK1 has a role in targeting or localizing active protein kinase C␣ to a particular subcellular compartment in NIH 3T3 cells, we first examined the localization of PICK1 in response to serum. Before stimulation, more than 95% of cells displayed a cytoplasmic punctate pattern of PICK1 (Fig. 4A). The PICK1 protein redistributed to a perinuclear region and displayed a dense ring-like structure when NIH 3T3 cells were stimulated with serum (Fig. 4A). Dynamic studies revealed that the redistribution started as early as 5 min (30% of cells with ring-like PICK1 structure) after stimulation, reached its peak (60 -70% of cells with ring-like PICK1 structure) at around 15-30 min and returned gradually to the unstimulated, cytoplasmic punctate pattern at 240 min (96% of cells) (Fig. 4, B and A). Cells simultaneously stained with the anti-rPICK1 antibody and the anti-cytochrome oxidase subunit I antibody revealed that PICK1 was still associated with mitochondria in the perinuclear region after serum stimulation (Fig. 4B). On the other hand, RACK1 remained in a diffuse, cytoplasmic distribution pattern, a pattern the same as that observed without stimulation (Fig. 1D), in the same cell where PICK1 translocated to the perinuclear region 30 min after serum stimulation (Fig. 4C). This demonstrated that the ␤IIPKC-binding protein RACK1 and the PKC␣-binding protein PICK1 have distinct responses to the same stimulation in NIH 3T3 cells with respect to their intracellular localization. Our results support the notion that isozyme-specific targeting proteins mediate specific functions of each isozyme and that the mitochondria association and the redistribution upon serum stimulation were events specific to PICK1.
To investigate whether PICK1 directs PKC␣ to mitochondria in the perinuclear region in NIH 3T3 cells after activation, we expressed PKC␣ and examined its intracellular localization by indirect immunofluorescence. NIH 3T3 cells were co-stained with antibodies specific to PKC␣ and the anti-rPICK1 antibody. PKC␣ stained diffusely throughout the entire cytoplasm in unstimulated cells, whereas PICK1 displayed a punctate mitochondria pattern (Fig. 4D, control). However, after cells were stimulated with serum, a substantial portion of PKC␣ translocated to the perinuclear region where it co-localized with PICK1 (Fig. 4D, serum). As shown with RACK1 in Fig. 4C, the distribution of endogenous PKC␤ remained unchanged after serum stimulation (data not shown), indicating that the co-movement caused by serum stimulation was specific to PICK1 and PKC␣.
TPA Stimulation Causes a Translocation of PKC␣ to the Plasma Membrane but Does Not Affect the Cytoplasmic Distribution of PICK1-To investigate whether the co-localization of PKC␣ and PICK1 is a common event upon activation of PKC␣, we further examined the intracellular distribution of PICK1 and PKC␣ in response to stimulation of TPA, a potent PKC activator. After TPA treatment of NIH 3T3 cells, the distribution pattern of PICK1 was, in contrast to serum stimulation, largely unchanged (Fig. 5A). These results indicated that the redistribution of PICK1 to the perinuclear region was a serumspecific response and that translocation of PICK1 was dependent on the nature of extracellular stimuli. However, TPA treatment caused a translocation of PKC␣ to the plasma membrane (Fig. 5B, upper panel), in the same cell, as documented. The distribution pattern of PICK1 in cells either expressing PKC␣ (Fig. 5B, upper panel) or not expressing PKC␣ (Fig. 5B, lower  panel) was similar, indicating that overexpression of PKC␣ alone did not cause redistribution of PICK1. These results indicated that the co-movement of PICK1 and PKC␣ was an event specific to serum stimulation.

FIG. 3. Mitochondria association of PICK1 is dependent on the PDZ domain.
FLAG-tagged rPICK1 and its various mutants, as shown in the schematic diagram, were expressed in NIH 3T3 cells, and the cellular localization of rPICK1 proteins was detected with polyclonal anti-FLAG antibodies followed with rhodamine-conjugated anti-rabbit IgG antibodies. Mitochondria were detected with monoclonal anti-cytochrome oxidase subunit I antibodies followed with FITC-conjugated anti-mouse IgG antibodies simultaneously. Deletion of the entire PDZ domain (rPICK1 Ala-142-Ser-416 ) resulted in a loss of the cytoplasmic punctate distribution pattern. Deletion of the C-terminal region and the K27A,D28A mutation did not affect the intracellular distribution of PICK1. White speckled pattern with black dots, PDZ domain; speckled and cross-hatched pattern, coiled-coil domain; black speckled pattern with white dots, acidic region.

PKC␣ That Lacks the C-terminal PDZ-binding Motif Fails to
Translocate to the Perinuclear Region upon Serum Stimulation-To investigate the molecular mechanism of the targeting of PKC␣ to the perinuclear region in response to serum stimulation, we constructed PKC␣ mutant that lacked the 4 amino acid residues, QSAV, at the extreme C terminus. These 4 amino acid residues are not only the binding site for PICK1 but also provide the sequence specificity that discriminates PKC␣ from other PKC isozymes that do not interact with PICK1. The wild type PKC␣ co-localized with PICK1 in the perinuclear region after serum stimulation (Fig. 6, A and B), as shown in the previous figures. However, the QSAV-deleted PKC␣ failed to translocate to the perinuclear region after serum stimulation (Fig. 6D), whereas PICK1 still translocated in the same cell (Fig. 6C). After TPA treatment, the QSAV-deleted PKC␣ mutant was found predominantly in the plasma membrane (Fig.  6H) as seen for the wild type PKC␣ (Fig. 6F), indicating that the QSAV-deleted PKC␣ mutant was responsive to TPA but FIG. 4. Serum stimulation induces a dynamic redistribution of PICK1 and co-translocation of PKC␣ to the perinuclear region in NIH 3T3 cells. Serum stimulation was carried out in NIH 3T3 cells for various times, as indicated (A and B). PICK1 was detected with polyclonal anti-rPICK1 antibodies followed with rhodamine-conjugated anti-rabbit IgG antibodies. To quantify cells that displayed either a cytoplasmic punctate pattern or a perinuclear ring-like structure of PICK1, an average of 500 cells were counted under microscope for each time point. B is a representation of five separate experiments. In some cases, cells were double-stained with anti-rPICK1 antibodies and anti-cytochrome oxidase antibodies to show mitochondria association (B, upper panel). In C, cells were stimulated with serum for 30 min. RACK1 proteins was detected using monoclonal anti-RACK1 antibodies (1:50). In contrast to PICK1, the intracellular distribution of RACK1 did not change by serum stimulation. In D, NIH 3T3 cells were transfected with plasmids expressing wild type PKC␣ and stimulated with 20% serum 36 h after transfection. The localization of PKC␣ was detected with monoclonal anti-PKC␣ antibodies (1:500) and FITC-conjugated anti-mouse IgG antibodies (1:175). PKC␣ co-localized with PICK1 in the perinuclear region after serum stimulation. Fluorescent images were observed using a fluorescence microscope. lost its capacity to translocate upon serum stimulation. TPA treatment did not change the distribution of PICK1 under these conditions (Fig. 6, E and G). Since PICK1 still moved in cells where mutant PKC␣ failed to when stimulated with serum, it was strongly suggested that PICK1 recruited PKC␣ to the perinuclear region when the kinase was activated by serum stimulation through interactions of the PDZ domain of PICK1 and the C-terminal QSAV of PKC␣. Taken together, our results suggested a unique role of PICK1 in targeting active PKC␣ to mitochondria in a ligand-specific manner. The PICK1 protein may serve as a switch point in PKC-mediated pathways that lead to specific cellular processes in response to different stimuli. DISCUSSION PICK1 may mediate functions unique to mitochondria, as indicated by our results that PICK1 is in mitochondria but not in ER, Golgi apparatus, or plasma membrane. An identical distribution pattern of PICK1 observed in NIH 3T3, Rat-1, H460, and H1299 cells suggests a conservation of intracellular localization and, potentially, functions in these cells. Mitochondria association is through the PDZ domain near the N terminus, but mutations in the putative carboxylate-binding loop that abolish binding to PKC␣ (16) have no effect on its mitochondria localization. These findings indicate that PICK1 can associate with mitochondria and simultaneously interact with PKC␣ and thus suggest a role for PICK1 in mediating PKC␣ targeting to mitochondria. This notion is confirmed because of the fact that upon activation, a substantial amount of PKC␣ is targeted to mitochondria in the perinuclear region, where it co-localizes with PICK1.
As a targeting protein for C kinases, PICK1 is unique in its association with a particular intracellular organelle in unstimulated cells. RACK1 is not localized to any specific organelle in Chinese hamster ovary cells, NG108-15 (neuroblastoma and glioma hybrid cells) (13), or NIH3T3 cells (Fig. 1D). Activation of PKC by NPA (a dopamine D2 agonist) induces movement of both ␤IIPKC and RACK1 to Golgi-like structures in Chinese hamster ovary cells expressing the dopamine D2 receptor. The translocation requires generation of second messengers for both ␤IIPKC and RACK1 (13). ␤Ј-COP, a coatomer protein essential for Golgi budding and vesicle trafficking, is a selective anchoring protein for activated PKC⑀ (15). Binding of ␤Ј-COP to Golgi membrane is dependent on activated ADPribosylation factor and accompanied by a corresponding in- crease in PKC⑀ binding (15). These indicate a stimulus-dependent movement of RACK1 and ␤Ј-COP to particular intracellular sites. Unlike these known anchoring proteins for PKC, PICK1 is clearly associated with mitochondria in unstimulated cells as well as cells stimulated with either serum or TPA (Figs. 4 and 5). Thus, association of PICK1 with mitochondria is independent of extracellular stimulation. PICK1 appears to serve as a docking site in mitochondria for PKC␣ rather than co-moving with PKC␣ from cytoplasm to mitochondria. However, anchoring of PKC␣ by PICK1 to mitochondria is still a ligand-specific event since only serum causes a translocation of PKC␣ to mitochondria (Figs. 4 and 5). Unveiling of the mechanism underlying this ligand-specific anchoring of PKC␣ by PICK1 will be of particular importance in dissecting the distinct function of PKC␣ upon different stimulations and the signaling pathway leading to it. Mutant PKC␣ with a deleted C-terminal QSAV sequence that abolishes interaction with PICK1 (16) fails to translocate on serum stimulation (Fig. 6). Serum stimulation may cause changes that lead to a "competent" conformation favoring interactions via the carboxylate-binding loop of PICK1 and the C-terminal QSAV peptide of PKC␣. The PICK1 protein may serve as a switch point in PKC-mediated pathways that lead to specific cellular processes in response to different stimuli. We propose that a direct interaction between PKC and its specific anchoring proteins, such as PKC␣ and PICK1, is one of the mechanisms sustaining the mitochondria-specific localization and the mitochondria-associated function of PKC␣.
The physiological significance of the association of PKC␣ and mitochondria upon serum stimulation remains to be investigated. Data indicate that PKC isozymes are localized on mitochondria and functionally associated with mitochondria under various conditions including apoptosis. Enhanced resistance to apoptosis induced by therapeutic drugs is reported in human pre-B REH cells when pretreated with the PKC agonist bryostatin-1 (29). The enhanced chemoresistance is coincident with the increased mitochondrial localization of PKC␣ and augmented phosphorylation of mitochondrial Bcl2, a key antiapoptotic protein (29). These results support a role for mitochondrial PKC␣ in Bcl2 phosphorylation and suppression of apoptosis in REH cells. Translocation is probably essential to bring PKC to the proximity of its substrates, such as the antiapoptotic Bcl-2 or the proapoptotic Bax, on mitochondria and therefore affects the apoptotic activity of mitochondria. Phosphorylation of mitochondria substrate(s) essential for cellular responses to serum may be achieved via recruitment of PKC␣ to mitochondria by PICK1.
Changes in intracellular mitochondria distribution are observed in various cellular processes such as during oocyte maturation, fertilization and embryo development (30,31), viral infection (32), and apoptosis (33,34). Clustering of mitochondria to a particular region may reflect a different local demand on energy and calcium flux or the need to facilitate translocation of mitochondrial proteins to a specific compartment. Targeting of PKC␣ to the perinuclear region by PICK1 upon serum stimulation may represent a need for translocation of phosphorylated mitochondria substrates or PKC␣ itself to the nucleus upon serum stimulation.
The PDZ domain of PICK1 is required for association with both PKC␣ (16) and mitochondria (Fig. 3), but they appear to be through distinct modes of interaction because only PKC␣ requires the carboxylate-binding loop of PICK1 for interaction. PDZ domains are widespread protein interaction domains that bind primarily the C-terminal carboxylate group in a sequencespecific way (19,20,28). In addition to the C-terminal peptide, PDZ domains can also bind internal non-C-terminal sequences or other PDZ. Functionally, PDZ domains have emerged as scaffolds for the organization and assembly of protein complexes at specific subcellular locations, particularly at the plasma membrane (20,28). PDZ-containing proteins frequently interact with several different partners or transmembrane proteins simultaneously. Assembly of protein complex can be attributed to the diverse modes of PDZ-target interaction. The PDZ in neuronal nitric oxide synthase can bind directly to the PDZ of ␣1-syntrophin (35). Structural analysis reveals that there are two interaction surfaces in the PDZ of neuronal nitric oxide synthase (36). A ␤-finger motif docks in the syntrophin peptide-binding groove leaving the canonical peptide-binding groove of neuronal nitric oxide synthase free for interaction with the conventional C-terminal peptide ligand. PICK1, which contains only one PDZ domain, may interact with PKC␣ and mitochondria through different modes. There is no known mitochondria targeting sequence found in PICK1, so PICK1 probably associates with mitochondria via interaction with a mitochondria outer membrane protein or protein complex through PDZ-mediated interactions. Since PICK1 can self-associate through the coiled-coil domain (37), it is also possible that the PDZ from one PICK1 molecule interacts with PKC␣ and that from the other PICK1 molecule in the dimer interacts with the mitochondria outer membrane protein.
The co-localization of activated PKC␣ with PICK1 is specific to serum stimulation since TPA has no significant effect on the localization of PICK1, whereas it causes a translocation of PKC␣ to the plasma membrane. Taken together, we have demonstrated that PICK1 is a specific targeting protein for PKC␣ that selectively recruits activated PKC␣ to mitochondria, thereby allowing specific phosphorylation events such as phosphorylation of PICK1 (27) to occur in a ligand-dependent way. Our findings presented here indicate a novel, previously unrecognized regulatory role for PICK1 as a specific mitochondria anchoring protein of PKC␣.