Characterization of a Novel Giant Scaffolding Protein, CG-NAP, That Anchors Multiple Signaling Enzymes to Centrosome and the Golgi Apparatus*

A novel 450-kDa coiled-coil protein, CG-NAP (centrosome and Golgi localized PKN-associated protein), was identified as a protein that interacted with the regulatory region of the protein kinase PKN, having a catalytic domain homologous to that of protein kinase C. CG-NAP contains two sets of putative RII (regulatory subunit of protein kinase A)-binding motif. Indeed, CG-NAP tightly bound to RIIα in HeLa cells. Furthermore, CG-NAP was coimmunoprecipitated with the catalytic subunit of protein phosphatase 2A (PP2A), when one of the B subunit of PP2A (PR130) was exogenously expressed in COS7 cells. CG-NAP also interacted with the catalytic subunit of protein phosphatase 1 in HeLa cells. Immunofluorescence analysis of HeLa cells revealed that CG-NAP was localized to centrosome throughout the cell cycle, the midbody at telophase, and the Golgi apparatus at interphase, where a certain population of PKN and RIIα were found to be accumulated. These data indicate that CG-NAP serves as a novel scaffolding protein that assembles several protein kinases and phosphatases on centrosome and the Golgi apparatus, where physiological events, such as cell cycle progression and intracellular membrane traffic, may be regulated by phosphorylation state of specific protein substrates.

Stimulation of various signaling cascades results in activation of protein kinases and phosphatases, which alter phosphorylation states of their respective substrates, leading to diverse physiological responses. Many serine/threonine protein kinases and phosphatases have relatively broad and overlapping substrate specificity. One of the mechanisms to organize such enzymes into individual signaling pathways is targeting them to discrete subcellular locations by anchoring proteins. For instance, type II cyclic AMP (cAMP)-dependent protein kinase (PKA) 1 is targeted to intracellular compartments through as-sociation of its regulatory subunit RII with protein kinase A anchoring proteins (AKAPs) (1,2). Three types of targeting proteins for protein kinase C (PKC) have been described (3)(4)(5). Three classes of phosphatase-targeting subunits have been identified that are specific for protein phosphatase 1 (PP1), PP2A, and PP2B (6).
Recently, a new class of multivalent adapter proteins that coordinate the location of multienzyme signaling complexes has been identified. For example, the pheromone mating response in yeast proceeds efficiently by clustering the successive members in the mitogen-activated protein kinase cascade on the scaffold protein STE5 (7). AKAP79 anchors not only PKA but also PKC and PP2B at the postsynaptic densities of mammalian synapses (8). AKAP250 (gravin) targets both PKA and PKC to the membrane cytoskeleton and filopodia of cells (9).
PKN is a serine/threonine protein kinase, having a catalytic domain homologous to the PKC family in the C-terminal region and a unique regulatory region in the N-terminal region (10). PKN is activated by a small GTPase Rho (11)(12)(13), unsaturated fatty acids such as arachidonic acid (14,15), and by truncation of the N-terminal regulatory region (14,16). Since PKN represents broad substrate specificity in vitro (10), PKN function may be regulated by intracellular targeting as well as by specific interaction with its substrates. We previously demonstrated that PKN associates with and phosphorylates intermediate filament proteins in vitro (17,18), which may be physiological substrates for PKN. PKN interacts with the actin cross-linking protein ␣-actinin, but does not efficiently phosphorylate it in vitro (19), suggesting that ␣-actinin serves as a scaffolding protein that targets PKN to specific cytoskeletal substrates.
In the present study, a cDNA encoding a novel coiled-coil protein with predicted molecular mass of 450 kDa was identified as a PKN-interacting protein by a yeast two-hybrid screen using the N-terminal regulatory region of PKN as bait. This protein was localized to centrosome throughout the cell cycle and the Golgi apparatus at interphase. Therefore, it was designated CG-NAP (centrosome and Golgi localized PKN-associated protein). CG-NAP interacted with various signaling enzymes including protein kinases (PKN and PKA) and phosphatases (PP1 and PP2A), and thus, may function as a novel multivalent adapter protein at these organelles. * This work was supported in part by grants from the Ministry of Education, Science, Sports and Culture, Japan, the "Research for the Future" program, the Japan Society for the Promotion of Science, the Japan Foundation for Applied Enzymology, and Kirin Brewery Co. Ltd. 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.
The nucleotide sequence reported in this paper has been submitted to the DDBJ/GenBank TM

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening-The N-terminal region (amino acids (aa) 1-540) of PKN fused to the Gal4 DNA binding domain (Gal4bd) was used as bait to screen a million clones of a human brain cDNA library fused to the Gal4 transcription activation domain (Gal4ad) as described (17). Screening to isolate PR130 was performed using a human fetal kidney cDNA library fused to the Gal4ad (CLONTECH) with a fragment of PKN (aa 136 -306) fused to the LexA DNA binding domain (LexAbd) as bait. The yeast expression plasmids for proteins fused to the LexAbd and those to the VP16 transcription activation domain (VP16ad) were constructed by subcloning the corresponding cDNA fragments into pBTM116 and pVP16, respectively.
Isolation of cDNA Clones-The cDNA clones encoding CG-NAP were isolated by screening human neuroblastoma and HeLa cDNA libraries with a 32 P-labeled probe prepared from the insert of clone 2-43. To obtain the entire cDNA sequence, 5Ј-RACE and 3Ј-RACE methods were employed using a Marathon-Ready cDNA library (human hippocampus) according to the manufacturer's instruction (CLONTECH). Mammalian expression plasmid for full-length CG-NAP was constructed by assembling these clones into pTB701-HA (20).
Full-length cDNAs of human RII␣ and human PR130 were obtained by PCR cloning using cDNA libraries of a human lung cancer cell line and human fetal kidney, respectively.
Preparation of Recombinant Proteins in Escherichia coli-Expression plasmids for proteins fused to glutathione S-transferase (GST) were constructed by subcloning the corresponding fragments into pGEX4T (Amersham Pharmacia Biotech). An expression plasmid for the deletion mutant HH tagged with (His) 6 -epitope was constructed by subcloning the corresponding fragment into pRSET A (Invitrogen). GST-fused and (His) 6 -tagged recombinant proteins were expressed in E. coli and purified by using glutathione-Sepharose 4B (Amersham Pharmacia Biotech) and nickel-NTA-agarose (Qiagen), respectively, according to the manufacturer's instruction.
In Vitro Binding Assay-[ 35 S]Methionine-labeled PKNN2 (aa 1-474 of PKN)(17) was incubated with GST-P#2-43 in a buffer containing 20 mM Tris-HCl at pH 7.5, 0.5 mM dithiothreitol, 150 mM NaCl, 0.05% Triton X-100, 1 mM EDTA, and 1 g/ml leupeptin at 4°C for 1 h. After addition of glutathione-Sepharose 4B, the reaction was continued for an additional 30 min. The resin was extensively washed with the same buffer, then bound proteins were eluted, resolved by SDS-PAGE, and the radioactive bands were visualized using a Fuji BAS1000 imaging analyzer.
For in vitro homodimer formation, the purified deletion mutants (His) 6 -tagged HH and GST-fused HH were incubated in a buffer containing 20 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.1% Triton X-100, 3 mM MgCl 2 , 1 mM EDTA, and 1 mM dithiothreitol at 4°C for 1 h. Then proteins bound to glutathione-Sepharose 4B were analyzed by immunoblotting with anti-His antibody.
Cell Culture, Transfection, and Drug Treatments-COS7 and HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% heat-inactivated fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin at 37°C in a humidified 5% CO 2 atmosphere. Mammalian expression plasmids for HA-and FLAG-tagged proteins were constructed by inserting the corresponding cDNA fragments into pTB701-HA and pTB701-FLAG (20), respectively. Plasmids pMhPKN3 (10) and pRc/CMV/PKN-FL (16) were used to express PKN and FLAGtagged PKN, respectively. For transient expression studies, COS7 cells were transfected with expression plasmid(s) by electroporation using GenePulser II (Bio-Rad). To disrupt the Golgi structure, HeLa cells were treated with 20 g/ml nocodazole (Sigma) for 90 min or 10 g/ml brefeldin A (BFA) (Wako, Japan) for 15 min.
Immunoprecipitation and Immunoblotting-Cells were lysed with a buffer containing 20 mM Tris-HCl at pH 7.5, 1% Nonidet P-40, 0.15 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1.5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, and 10 g/ml leupeptin. Cleared lysates were incubated with the appropriate antibody at 4°C for 2 h, then Protein G-Sepharose (Amersham Pharmacia Biotech) was added and the reaction was continued for another 1 h. After extensively washing the resin with the same buffer, the bound proteins were resolved by SDS-PAGE and then subjected to immunoblotting as described (10). Blots were visualized by enhanced chemiluminescence method.
Northern Blotting-Polyadenylated RNA was prepared from HeLa cells using QuickPrep mRNA Purification Kit (Amersham Pharmacia Biotech). The blot of HeLa mRNA (3 g) was incubated with the 32 Plabeled probe prepared from the cDNA insert of clone 2-43, followed by extensive washing. The radioactive band was then visualized using a Fuji BAS1000.
Immunofluorescence Microscopy-Cells grown on cover glasses were extracted with 0.1% Triton X-100 in 80 mM Pipes, pH 6.9, 5 mM EDTA, and 1 mM MgCl 2 at room temperature for 2 min, then fixed with cold MeOH for 3 min, or cells were directly fixed with 3.7% formaldehyde in 0.2 M Na-PO 4 , pH 7.2, and permeabilized using 0.1-0.3% Triton X-100. Cells were blocked with 5% normal donkey serum in phosphate-buffered saline with Tween 20 (20 mM Na-PO 4 , pH 7.5, 150 mM NaCl, and 0.03% Triton X-100), then incubated with the relevant antibody for 1 h at room temperature. Cells were washed with phosphate-buffered saline with Tween 20, and primary antibody is visualized by subsequent incubation with the appropriate secondary antibody conjugated with either rhodamine or DTAF. DNA was visualized by adding 4Ј,6-diamidino-2-phenylindole dihydrochloride at concentration of 0.1 g/ml. The fluorescence of rhodamine and DTAF was observed under a confocal laser scanning fluorescent microscope (Zeiss), the former at 543 nm argon excitation using a 590-nm long pass barrier filter and the latter at 488-nm argon excitation using a 510 -525-nm band pass barrier filter.

FIG. 1. Interaction between PKN and P#2-43.
A, analysis by yeast two-hybrid system. L40 cells were cotransfected with expression plasmids encoding various proteins fused to LexAbd (left) and those fused to VP16ad (right) as indicated in left panel. Right panel shows developments of blue color 1 h after initiating filter assays. Murine tumor suppressor p53 and SV40 large T antigen were used as controls. B, analysis by in vitro binding assay. 35 S-Labeled in vitro translated PKNN2 (aa 1-474 of PKN) was incubated with bacterially synthesized GST or GST-fused P#2-43. After removing aliquots (Input), proteins bound to glutathione-Sepharose 4B were collected (Output). Proteins in "Input" and "Output" preparations were separated on SDS-PAGE followed by visualization of radioactive bands using a Fuji BAS1000. C, analysis by immunoprecipitation. HA-tagged P#2-43 and full-length PKN were coexpressed in COS7 cells and immunoprecipitated (IP) with anti-PKN (␣N2), normal rabbit serum (NRS), anti-HA 12CA5 (␣HA), or normal mouse immunoglobulin (NMIg). P#2-43 and PKN in immunoprecipitates and in extracts (Ϫ) were visualized by immunoblotting with anti-HA 3F10 (␣HA) and ␣N2, respectively.

Yeast Two-hybrid Screen for PKN Interacting Proteins-The
N-terminal region (aa 1-540) of PKN was used as bait to screen a human brain cDNA library by yeast two-hybrid system as described (17). Clone 2-43 contained a 1.3-kilobase pair cDNA insert encoding a novel and partial amino acid sequence, which was named P#2-43. Other combinations of two-hybrid constructs further confirmed the specific interaction between the N-terminal region of PKN and P#2-43 (Fig. 1A).
Primary Structure of CG-NAP-We obtained the presumptive full-length coding sequence from human cDNA libraries of neuroblastoma, hippocampus, and HeLa cells by conventional hybridization screening in combination with 5Ј-and 3Ј-RACE methods. The combined cDNA sequence contained an open reading frame of 11,700 bp encoding a polypeptide of 3,899 amino acids with a predicted molecular mass of 451,803 daltons ( Fig. 2A). We designated this giant protein as CG-NAP (centrosome-and Golgi-localized PKN-associated protein). CG-NAP was a novel protein; however, BLAST search yielded two proteins that are highly homologous to partial regions of CG-NAP: human yotiao (22) and rabbit AKAP120 (23), corresponding to aa 1-1626 and 2049 -3060, respectively (Fig. 3). CG-NAP also represents limited and relatively weak homology with pericentrin (Fig. 3), a centrosomal protein (24). CG-NAP contains four leucine zipper-like motifs ( Fig. 2A) and many stretches of coiled-coil structure (Fig. 2B). These structural features are thought to be involved in association with other proteins and/or homodimerization/homo-oligomerization (25). Homodimerization of P#2-43 was suggested by the yeast twohybrid assay using a combination of P#2-43-LexAbd and P#2-43-VP16ad (data not shown). Therefore, homodimerization of the N-terminal region of CG-NAP was examined in vitro using a deletion mutant HH (aa 17-859). (His) 6 -tagged HH was copurified with GST-fused HH by glutathione-Sepharose (Fig.   FIG. 2. Primary structure and expression of CG-NAP. A, primary structure of CG-NAP. Predicted amino acid sequence of full-length CG-NAP is shown with dark boxes indicating leucine residues in leucine zipper-like motifs. The coding region of the original clone 2-43 is shaded. B, coiled-coil analysis of CG-NAP by COILS program (43). C, homodimerization of the N-terminal region of CG-NAP. GST or GST-fused HH (aa 17-859 of CG-NAP) was incubated with (His) 6 -tagged HH. After removing aliquots (Input), proteins bound to glutathione-Sepharose 4B were collected (Output). Proteins in "Input" and "Output" preparations were analyzed by immunoblotting with anti-His antibody. D, Northern blots of CG-NAP. Polyadenylated RNA from HeLa cells was probed with cDNA insert from clone 2-43. Position of mRNA is indicated by arrowhead. E, immunoblots of recombinant and endogenous CG-NAP. HA-tagged CG-NAP expressed in COS7 cells was immunoprecipitated with 12CA5 (␣HA) or with normal mouse immunoglobulin (NMIg). Immunoprecipitates and extracts of HeLa cells (H) and U937 cells (U) were separated on 4.5% SDS-PAGE, followed by immunoblotting with ␣EE, ␣BH, or control rabbit serum (NRS) as indicated.
2C), suggesting the homodimerization (or homo-oligomerization) of this protein. Furthermore, we independently confirmed the homodimerization by coimmunoprecipitation experiment using different epitope-tagged constructs of CG-NAP deletion mutant (aa 1-1280) coexpressed in COS7 cells (data not shown).
Expression of CG-NAP mRNA and Protein-Northern blots using polyadenylated RNA of HeLa cells revealed a single band longer than 12 kilobase pairs (Fig. 2D). CG-NAP mRNA of the identical size was ubiquitously expressed at low abundance in human tissues (data not shown).
To examine whether the cDNA sequence obtained here encoded the complete CG-NAP, we performed immunoblotting using antiserum ␣EE raised against a CG-NAP fragment EE (Fig. 3). Recombinant CG-NAP expressed in COS7 cells (Fig.  2E, lane 2) co-migrated with the endogenous CG-NAP in HeLa (lane 3) and U937 (lane 4) cells. The size of the band appeared to agree with the calculated molecular mass of 450 kDa. Another antiserum, ␣BH, raised against a different part of CG-NAP (Fig. 3) also detected a band of the same size (Fig. 2E, lane  6). These results indicate that the cDNA sequence obtained here encodes full-length CG-NAP.
Localization of CG-NAP to Centrosome, the Midbody, and the Golgi Apparatus-Subcellular localization of CG-NAP was examined by immunofluorescence analysis using ␣EE in HeLa cells at various phases of the cell cycle (Fig. 4A). Cells were extracted with nonionic detergent before fixation to visualize proteins of low abundance associated with intracellular structures. In interphase cells, CG-NAP was localized to one spot at the perinuclear region and to a dispersed network near the spot (Fig. 4A, a), presumably corresponding to centrosome and the Golgi apparatus, respectively. In mitotic cells, CG-NAP was localized to the spindle poles (Fig. 4A, d, g, and j), and to extremities of the midbody in the cells at telophase/cytokinesis (Fig. 4A, j). Antiserum ␣BH also gave the identical staining (Fig. 4B, c), whereas normal rabbit serum did not (data not shown). Subcellular distribution of CG-NAP was similar in other cell lines, such as SaOS2, TIG1, HEK293, and NIH3T3 cells (data not shown). Localization of CG-NAP to centrosome and the midbody was confirmed by double-staining, using ␣EE or ␣BH with an antibody against a centrosomal protein ␥-tubulin (26) (Fig. 4B). CG-NAP was also co-stained with Golgi 58K protein at perinuclear area (Fig. 4C, a and b). We further examined the relationship between CG-NAP and the Golgi apparatus using the microtubule-destabilizing agent nocodazole (27) and the fungal metabolite BFA (28), which are known to disrupt the Golgi apparatus by distinct mechanisms. Nocodazole treatment of HeLa cells dispersed the perinuclear CG-NAP staining into scattered pattern throughout the cytoplasm (Fig. 4C, d), which is characteristic for the Golgi staining of nocodazole-treated cells. BFA treatment disrupted the perinuclear CG-NAP staining (Fig. 4C, e). BFA-induced Golgi disruption is reversible (28), and the CG-NAP staining was also recovered by incubation in the absence of the drug (Fig. 4C, f). Either treatment did not change the centrosomal staining of CG-NAP. These results indicate that CG-NAP is localized to centrosome throughout the cell cycle, the midbody at telophase, and the Golgi apparatus at interphase.
Immunostaining of recombinant CG-NAP expressed in COS7 cells detected an intense spot at the perinuclear area (Fig. 4D,  a), which was obvious when cells were fixed after detergent extraction (Fig. 4D, b). This spot was colocalized with ␥-tubulin (data not shown), suggesting that recombinant CG-NAP is predominantly localized to centrosome.
Interaction of PKN with Full-length CG-NAP-We next examined the interaction between PKN and full-length CG-NAP. PKN was coimmunoprecipitated with full-length CG-NAP when both proteins were exogenously expressed in COS7 cells (Fig. 5A). Immunofluorescence analysis revealed that a certain population of PKN was localized to centrosome in the cells fixed after detergent extraction (Fig. 5B), although this protein is predominantly located in the soluble fraction of the cells (21). These results suggest that PKN is associated with CG-NAP under physiological condition.
CG-NAP as AKAP-AKAP120 (23) shows high sequence homology with a part of CG-NAP (Fig. 6A), and a putative RIIbinding motif forming an amphipathic helix (29) was conserved in CG-NAP at aa 2540 -2558 (Motif 2 in Fig. 6A). Another RII-binding motif was also found at aa 1438 -1455 (Motif 1 in Fig. 6A). We therefore examined whether these regions could bind to RII␣. Deletion mutants ES (aa 1229 -1917, Fig. 3) and MB (aa 2380 -2876, Fig. 3) were coimmunoprecipitated with RII␣ (Fig. 6B). Furthermore, interaction between RII␣ and full-length CG-NAP was observed using exogenously expressed proteins in COS7 cells (Fig. 6C, lane 3), and using endogenous proteins in HeLa cells (Fig. 6C, lane 5). These data indicate that CG-NAP binds to RII␣ at high affinity and constitutively, and thus, is a novel AKAP. Immunofluorescence analysis de- tected RII␣ in cytosol and in other organelles, when cells were fixed without detergent extraction (Fig. 6D, b). On the other hand, RII␣ was found to be colocalized with CG-NAP in the cells fixed after detergent extraction (Fig. 6D, c and d).
Association of CG-NAP with PP2A-Several AKAPs associate not only with PKA but also with other signaling enzymes including PP2B (8). We have isolated another cDNA clone encoding aa 203-1150 of PR130 as PKN-interacting protein by yeast two-hybrid screen (Fig. 7A). PR130 is one of the B subunits of PP2A, and its mRNA is expressed ubiquitously at low levels (30); however, a heterotrimeric PP2A holoenzyme containing PR130 has not been described. We found that a complex consisting of PR130, the A subunit (PP2A-A), and the catalytic subunit (PP2A-C) was formed in insect cells in an equal molar ratio and exhibited phosphatase activity. 2 These data indicate that PR130 is a functional B subunit of the PP2A holoenzyme.
Next, we examined the interaction between PKN and PR130 by coimmunoprecipitation using full-length proteins. PR130 and PKN associated weakly but significantly (Fig. 7B). Since the efficiency of coimmunoprecipitation was relatively low and direct interaction between PKN and PR130 in vitro was hardly detected by GST-pull down assay (data not shown), we speculated that this interaction is indirect and mediated by some adapter protein such as CG-NAP. Thus, we examined whether PR130 interacted with CG-NAP using COS7 cells expressing PR130. ␣BH coimmunoprecipitated PR130 with endogenous CG-NAP (Fig. 7C, lane 3), suggesting that PR130 associates with PKN through binding with CG-NAP. Deletion analysis of CG-NAP located the binding site for PR130 in the deletion BS (Fig. 7D, lane 5, and Fig. 3).
These results raised the possibility that PP2A holoenzyme could also associate with CG-NAP. We performed immunoprecipitation using COS7 cells coexpressing FLAG-tagged PR130 and HA-tagged BS. Endogenous PP2A-C was coimmunoprecipitated with BS by anti-HA (Fig. 7E, lane 6). Since binding of endogenous PP2A-C with PR130 was confirmed in the same lysate (Fig. 7E, lane 3), it is implicated that PP2A-C and probably PP2A-A were coprecipitated with CG-NAP (in this case, its deletion BS) through binding with PR130.
Association of CG-NAP with the Catalytic Subunit of PP1-By searching for binding motifs for other signaling enzymes, we found a possible PP1 binding motif R/KVXF (31) at aa 1053-1056 of CG-NAP. Immunoprecipitation revealed that the endogenous catalytic subunit of PP1 (PP1-C) interacted with P#2-43 (containing the putative PP1 binding motif) expressed in COS7 cells (Fig. 7F, lane 3), and furthermore, with endogenous CG-NAP in HeLa cells (Fig. 7F, lane 2). These results indicated that PP1-C associates with CG-NAP under physiological condition. DISCUSSION The present study discovered a novel 450-kDa protein CG-NAP ubiquitously expressed in human tissues. CG-NAP is localized to centrosome throughout the cell cycle, the midbody at telophase, and the Golgi apparatus at interphase in cultured cell lines. We have demonstrated that CG-NAP interacts with PKN, RII␣ subunit of PKA, PP2A through its regulatory B subunit PR130, and the catalytic subunit of PP1. Therefore, CG-NAP may function as a scaffolding protein for the subcellular targeting of these enzymes, and thus may be a novel multivalent adapter protein for signaling enzymes as well as a new AKAP.
AKAPs localized to centrosome (AKAP350) (32) and the Golgi apparatus (AKAP85) (33) have been identified by RII overlay of human lymphoblast lysates. The relationship be- tween these AKAPs and CG-NAP remains unclear, since amino acid sequences of these AKAPs are currently unknown. In the course of this study, two novel proteins, rabbit AKAP120 (23) and human yotiao (22), were discovered, both of which represent high sequence homology with partial regions of CG-NAP (Fig. 3). In addition, contiguous BAC clones of human genome, RG293F11 (from chromosome 7q21-22) and GS541B18 (from chromosome 7q21), were found to cover aa 20 -1639 and 1793-3768, respectively, of CG-NAP, suggesting that CG-NAP is encoded by a single gene located at chromosome 7q21-22. AKAP120 may be coded by a rabbit homolog of CG-NAP. Yotiao might be a partial clone or the product of the alternative splicing or post-translational proteolytic processing of CG-NAP.
Phosphorylation of centrosomal proteins is suggested to be involved in the regulation of centrosomal function (34). Various protein kinases and phosphatases are localized to centrosome (6), and some of the kinases are implicated in the regulation of centrosome separation (35,36) and microtubule nucleation (37). Protein phosphorylation is also implicated in mitotic Golgi fragmentation (38) and membrane traffic (39,40). However, it remains largely unknown how protein kinases and phosphatases are localized to these organelles and how these enzymes are coordinated to fulfill the physiological processes. We have demonstrated that a certain population of PKN and RII␣ are localized to centrosome (Fig. 5B) and centrosome/the Golgi apparatus (Fig. 6D, c and d), respectively. In addition, immunocytochemical study of human brain tissues showed that PKN is also enriched in the Golgi bodies (41). Therefore, CG-NAP may represent the candidate that coordinates the location and activity of these enzymes at centrosome and the Golgi apparatus.
Coimmunoprecipitation studies using various deletion mutants of CG-NAP indicated that the four enzymes bind to CG-NAP at distinct sites: PKN on P#2-43, PKA on ES and MB, PP1 on P#2-43, and PP2A on BS (see Fig. 3). Furthermore, PP2A and PKN appears to bind simultaneously to CG-NAP (Fig. 7B). Since we found that dephosphorylation of PKN by PP2A decreased its kinase activity in vitro, 3 it is possible that dephosphorylation of PKN by PP2A closely located on CG-NAP affects the activity and/or localization of PKN under physiological condition. In other words, CG-NAP may provide a scaffold to facilitate interactions among the bound enzymes. On the contrary, another possible function of CG-NAP is that it retains the bound enzymes as inactive pools until they are activated by appropriate signals. Such inhibitory effect is demonstrated in other multiadaptor proteins, AKAP79 on PKC (8) and PP2B (42), and gravin on PKC (9). CG-NAP was phosphorylated by PKN and PKA at distinct regions in vitro (data not shown). Changes in the phosphorylation state of CG-NAP may lead to its dynamic structural alterations, which may result in the changes in its binding affinity with the target compartment(s) or with other enzyme(s). This might explain why the association of PKN, PP2A, or PP1 with full-length CG-NAP is relatively weak compared with that with the deletion fragment of CG-NAP. Studies to elucidate the role of CG-NAP and the bound signaling enzymes in the centrosome and Golgi functions will provide further understanding of the regulatory mechanisms of signal transduction occurring at these organelles.