Phosphorylation of Cdc42 Effector Protein-4 (CEP4) by Protein Kinase C Promotes Motility of Human Breast Cells*

Background: Cdc42 effector protein-4 (CEP4) is a substrate of protein kinase C (PKC) in human breast cells. Results: Phosphorylation at two defined positions in CEP4 causes it to dissociate from Cdc42 and consequently to stimulate cell movement. Conclusion: CEP4 phosphorylation at Ser18 and Ser80 by PKC promotes cell movement. Significance: This work describes a novel PKC-stimulated signaling pathway by which human breast cells acquire metastatic potential. Cdc42 effector protein-4 (CEP4) was recently identified by our laboratory to be a substrate of multiple PKC isoforms in non-transformed MCF-10A human breast cells. The significance of phosphorylated CEP4 to PKC-stimulated motility of MCF-10A cells was evaluated. Single site mutants at Ser residues embedded in potential PKC consensus sites (Ser18, Ser77, Ser80, and Ser86) were individually replaced with Asp residues to simulate phosphorylation. Following expression in weakly motile MCF-10A cells, the S18D and S80D mutants each promoted increased motility, and the double mutant (S18D/S80D) produced a stronger effect. MS/MS analysis verified that Ser18 and Ser80 were directly phosphorylated by PKCα in vitro. Phosphorylation of CEP4 severely diminished its affinity for Cdc42 while promoting Rac activation and formation of filopodia (microspikes). In contrast, the phosphorylation-resistant double mutant S18A/S80A-CEP4 blocked CEP4 phosphorylation and inhibited motility of MCF-10A cells that had been stimulated with PKC activator diacylglycerol lactone. In view of the dissociation of phospho-CEP4 from Cdc42, intracellular binding partners were explored by expressing each CEP4 double mutant from a tandem affinity purification vector followed by affinity chromatography, SDS-PAGE, and identification of protein bands evident only with S18D/S80D-CEP4. One binding partner was identified as tumor endothelial marker-4 (TEM4; ARHGEF17), a guanine nucleotide exchange factor that is involved in migration. In motile cells expressing S18D/S80D-CEP4, knockdown of TEM4 inhibited both Rac activation and motility. These findings support a model in which PKC-mediated phosphorylation of CEP4 at Ser18 and Ser80 causes its dissociation from Cdc42, thereby increasing its affinity for TEM4 and producing Rac activation, filopodium formation, and cell motility.

had been genetically engineered to bind an unnatural ATP analog (13,14). MCF-10A cells provide a suitable system for this approach because the endogenous PKC isoforms are expressed at low levels, thereby producing a low background against which the products of the traceable mutant can be detected. Our laboratory investigated substrates of three traceable PKC isoforms (␣, ␦, and ) in MCF-10A cells (5). Findings identified similar protein phosphorylation profiles for ␣ and ␦, whereas that for was radically different. Following MS/MS of selected bands, it was found that the ␣ and ␦ isoforms were associated with a number of Rho GTPase effectors and regulatory proteins, including a protein identified as Cdc42 effector protein-4 (CEP4). Upon further evaluation in vitro with highly pure, recombinant CEP4 and PKC isoforms as well as in cells treated with isoform-specific shRNA reagents, CEP4 proved to be a substrate for all of them. These findings prompted the speculation that individual PKC isoforms may respond to specific stimuli but upon activation may share common protein substrates that relate to a specific phenotype.
CEP4, also known as binder of Rho GTPases-4, is one of five related isoforms, CEPs 1-5 (15), and is expressed ubiquitously in all human adult tissues (16). The existence of this family was discovered as the result of a two-hybrid screen with the GTPase TC10 as bait (15). Consequently, all five CEPs were found to bind with high affinity to the GTP-bound form of Cdc42 but did not bind to the other small GTPases (Rac and RhoA) (15,17). Site-specific mutagenesis established that Cdc42 binds to CEP at its 16-amino acid Cdc42/Rac interaction binding (CRIB) domain (15,18). In the limited published research available on CEPs, the phenotypes associated with these proteins depend on formation of the CEP⅐Cdc42 complex and its role in actin-based membrane protrusions (e.g. pseudopodia), actin filament assembly, and cell-cell contacts. There is one example of CEP3 acting independently of Cdc42 in which it associated with a septin GTPase (19). This protein forms multimeric complexes and has been implicated in cytokinesis, cell polarity, and oncogenesis. Binding of CEP3 to septins disrupted their oligomeric organization. However, this effect by CEP3 was blocked when GTP-Cdc42 was overexpressed, thus providing an example of negative regulation of CEP3 by Cdc42 (19).
In the present work, CEP4 was further investigated as a PKC substrate as well as for the functional consequences of its phosphorylation. By use of site-specific mutagenesis at PKC consensus sites as well as tandem mass spectrometry (MS/MS) of CEP4 phosphorylated by PKC␣ in vitro, key sites of phosphorylation in CEP4 were defined. Site-directed mutants of CEP4 were used to explore the significance of phosphorylation to its interaction with Cdc42, its binding interactions with other proteins, its effects on the actin cytoskeleton, and its role in cell motility. Overall, these studies establish phosphorylated CEP4 as a component of the PKC-mediated motility signaling pathway and as a modulator of actin-based membrane protrusions.

EXPERIMENTAL PROCEDURES
Materials-Human breast epithelial MCF-10A cells were obtained from The Barbara Ann Karmanos Cancer Center. BL21 and DH5␣ bacterial cells, DMEM/F-12 medium, horse serum, and DNA primers were purchased from Invitrogen.
Plasmids-Human CEP4 cDNA was restricted at SgfI and MluI sites in the pCMV6 plasmid. Expression from this plasmid conferred DDK and myc epitope tags at the C terminus of the recombinant protein. By use of PCR, the cDNA insert was amplified with suitable primers for subcloning into the TAP vector (Agilent Technologies) at BamHI and EcoRV restriction sites. The TEM4 shRNA or scrambled control shRNA each was co-expressed with GFP from a bicistronic vector for evaluation of transfection efficiency.
Cell Culture and Transfection-MCF-10A cells were grown in 10-cm or 60-mm BD Falcon culture dishes at 37°C and 5% CO 2 in DMEM/F-12 medium supplemented with 5% horse serum, 1% penicillin/streptomycin, 0.2% Fungizone, 10 g/ml insulin, 0.5 g/ml hydrocortisone, 20 ng/ml epidermal growth factor, and 100 ng/ml cholera toxin. Cells were replated 1 day prior to transfection so that the cell density was 80%. The medium was replaced with fresh complete medium 1 h before transfection. Plasmid DNA was mixed with the PolyExpress transfection reagent in a 1:3 ratio (DNA/reagent) and incubated for 15 min at room temperature. The mixture was added to the cells and incubated at 37°C and 5% CO 2 . After 18 -24 h, the medium was replaced with fresh complete medium. Cells were harvested 48 h post-transfection. At this time point, the transfection efficiency was typically 60 -80%.
For cells transfected with a plasmid encoding an shRNA reagent (TEM4 or scrambled control), the plasmid was transiently transfected into MCF-10A cells with PolyExpress as described above. At 48 h, a plasmid encoding one of the CEP4 mutants or the vector control was transfected, and the cells were incubated at 37°C for an additional 48 h in complete medium. As judged by the GFP signal evident throughout the 96-h period, the transfection efficiency was typically 70 -80%.
Cell Lysis and Western Blot-Cells were disrupted in lysis buffer (Cell Signaling Technology, Inc.) and sonicated three times each for 10 s followed by centrifugation at 10,000 ϫ g for 10 min to remove insoluble debris. The lysate was assayed for protein concentration using a protein dye reagent (Bio-Rad). Samples of known protein concentration were denatured with 5ϫ SDS sample buffer (50% (v/v) glycerol, 1% SDS, 0.05% bromphenol blue, 0.4 M Tris, pH 6.8, and 2 mM dithiothreitol) followed by heating at 95°C for 5 min. Proteins were separated by 8% SDS-PAGE and transferred to a PVDF membrane (Millipore Corp.). The blot was developed with the appropriate antibodies and detected with chemiluminescence reagents (Pierce).
Immunoprecipitation-For MCF-10A cells expressing wildtype (WT) or mutant CEP4, cells were lysed in hypotonic, detergent-free lysis buffer (20 mM Tris, pH 7.4, 2 mM MgCl 2 , 2 mM EGTA, 1 mM DTT, 10 M bisindolylmaleimide, protease inhibitors (1:1000), and phosphatase inhibitors (1:100)) followed by immunoprecipitation with anti-FLAG EZview beads. Anti-FLAG beads (60 l pre-equilibrated in cold TBS) were applied to the lysate and rotated for 2 h at 4°C. The beads were centrifuged at 8200 ϫ g for 5 min and washed three times in hypotonic, detergent-free lysis buffer. Pellets were electrophoresed on 8% SDS-polyacrylamide gels and transferred to a PVDF membrane, and immunoreactive bands were detected with the appropriate antibody.
Binding Assay with CEP4 Mutants and Constitutively Active Q61L-Cdc42-Glutathione S-transferase (GST)-tagged Q61L-Cdc42 was expressed from a bacterial expression vector (pGEX-2T) in BL21 cells. This plasmid was provided by the laboratory of Prof. Gary Bokoch (Scripps Research Institute; Addgene catalog number 12971). Bacterial cells were collected by centrifugation at 5000 ϫ g for 10 min and lysed with 3 ml of B-PER reagent (Fisher Scientific) containing 0.2% (6 l) lysozyme, 0.2% (6 l) DNase I, and 1% protease inhibitor. The bacterial cell lysate was prepared by centrifugation at 15,000 ϫ g for 5 min. The supernatant containing Cdc42-Q61L was added to 500 l of glutathione-Sepharose affinity resin in a Bio-Rad minicolumn pre-equilibrated in wash buffer (50 mM Tris and 150 mM NaCl, pH 8.0), and the column was washed twice with 10 column volumes. The beads with immobilized Cdc42-Q61L were suspended in wash buffer, transferred to an Eppendorf tube, centrifuged at 8000 ϫ g for 5 min at 4°C, and the supernatant was discarded.
MCF-10A cells expressing myc-tagged wild-type or mutant CEP4 were lysed in hypotonic, detergent-free buffer (20 mM Tris, pH 7.4, 2 mM MgCl 2 , 2 mM EGTA, 1 mM DTT, 10 M bisindolylmaleimide, protease inhibitors (1:1000), and phosphatase inhibitors (1:100)). The cell lysates were obtained by centrifugation at 10,000 ϫ g for 10 min and precleared with 60 l of glutathione-Sepharose beads (Fisher Scientific) by rotating at 4°C for 1 h. The precleared MCF-10A cell lysate was incubated with immobilized Q61L-Cdc42 (described above) by rotation at 4°C for 40 min. Following two washes with hypotonic, detergent-free lysis buffer, the bound protein was eluted by boiling with 1ϫ SDS-PAGE sample buffer, resolved by 8% SDS-PAGE, and transferred to a PVDF membrane. Immunochemical analysis with anti-myc was used to determine the amount of myc-tagged CEP4 that was bound to Cdc42.
Isolation of Binding Partners of Pseudophosphorylated CEP4 by the TAP Method-Proteins that were physically associated with CEP4 were isolated by use of a CEP4 double mutant expressed from a TAP vector that conferred two affinity tags at the C terminus of CEP4. The affinity tags enabled binding to a streptavidin-binding resin and a calmodulin-binding resin, respectively (Agilent Technologies). MCF-10A cells grown in three 10-cm plates to 80 -90% confluence were transfected with a TAP vector expressing a CEP4 double mutant (S18A/S80A or S18D/S80D). Forty-eight hours post-transfection, cells were resuspended in 1 ml of hypotonic, detergent-free lysis buffer (20 mM Tris, pH 7.4, 2 mM MgCl 2 , 2 mM EGTA, 1 mM dithiothreitol, 10 M bisindolylmaleimide, 0.1% protease inhibitor mixture, and 1% phosphatase inhibitor mixture) and disrupted by three successive freeze-thaw cycles (10 min at Ϫ150°C and 10 min in cold water). After centrifugation of the cells at 10,000 ϫ g for 10 min, the crude cell lysate was incubated for 2 h at 4°C with 100 l of streptavidin-binding resin (pre-equilibrated with streptavidin binding buffer). The resin and any bound proteins were washed twice with 750 l of streptavidin binding buffer and eluted with 200 l of streptavidin elution buffer for 1 h at 4°C. The eluate was incubated with 50 l of calmodulin-binding resin (pre-equilibrated with calmodulin binding buffer) at 4°C for 3 h and washed twice with 750 l of calmodulin binding buffer. The bound proteins were eluted by boiling with 1ϫ loading buffer at 95°C for 5 min, resolved by SDS-PAGE, and stained with GelCode Blue. Bands of interest were excised and analyzed by MS/MS.
Motility Assay-Cell motility was measured by the cell sedimentation method, which required a 10-well glass slide and a 10-hole metallic manifold (Creative Scientific Methods, Inc.). The cells were applied to the slide through the manifold so that the cells form a tight concentric circle in each well. After incubation overnight at 37°C and 5% CO 2 , the manifold was removed, and the cells were immediately viewed under a Nikon Diaphot microscope (t ϭ 0). The extent of movement was determined in triplicate by measuring the difference in total area at t ϭ 0 and t ϭ 6 h using a camera attached to the microscope (Moticam 2000) and Motic Image software. The results of triplicate measurements were averaged.
Immunocytochemistry-Cells were cultured overnight on polylysine-coated coverslips and then transfected with the myc-tagged CEP4-S18D/S80D or -S18A/S80A constructs. Each coverslip was washed twice in 2 ml of PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, and 2 mM MgCl 2 , pH 6.9) and fixed with 3.7% paraformaldehyde in PBS. Nonspecific sites were blocked for 15 min at 37°C with 2% BSA plus 0.1% sodium azide in TBS-Tween 20. After blocking, the cells were incubated for 2 h at 37°C with an anti-myc tag rabbit monoclonal antibody (Cell Signaling Technology, Inc.) diluted to 1:300. Samples were incubated with FITC-conjugated goat anti-rabbit secondary antibody diluted to 1:300 for 1.5 h at 37°C (Santa Cruz Biotechnology) and washed three times with TBS-Tween 20. Following anti-myc staining, F-actin was stained for 1 h at 37°C with rhodamine-phalloidin diluted to 1:300 (Cytoskeleton, Inc.), and nuclei were stained for 5 min at 37°C with Hoechst 33342 stain diluted to 1:1000 (Invitrogen) before mounting. Imaging was performed with a Plan-Neofluor 63ϫ oil immersion objective lens on a Zeiss Axio Imager-M2 upright fluorescence microscope.
Pulldown Assay of Activated Rac-Activation of Rac was measured in lysates of MCF-10A cells prepared in commercially available buffers (Cytoskeleton, Inc.) as described previously (3). Cell lysates containing a total protein content of 600 g of protein were incubated with p21-activated kinase-binding domain-agarose beads (20 g) at 4°C for 1 h. Following centrifugation, the beads were washed once with wash buffer (25 mM Tris, pH 7.5, 30 mM MgCl 2 , and 40 mM NaCl) followed by centrifugation at 13,000 ϫ g at 4°C for 10 min. The entire sample was loaded, and the pelleted material was resolved by 12% SDS-PAGE, transferred to a PDVF membrane, and probed with anti-Rac.
Tandem Mass Spectrometry Analysis-To identify sites of phosphorylation in pure, recombinant GST-tagged CEP4 (synthesized by a wheat germ system), the protein (1.1 g) was phosphorylated at 30°C for 2 h by pure human PKC␣ (1 g) in vitro (5). The phosphorylated product and untreated control CEP4 were resolved by 8% SDS-PAGE (0.75 mm thickness). After staining with GelCode Blue, the gel slices corresponding to phospho-CEP4 or untreated CEP4 were digested with endo-Lys-C and endo-Glu-C. To identify proteins that co-purified with TAP-CEP4, gel-resolved bands of interest were excised and digested in situ with trypsin. For these experiments, the resulting peptides were analyzed by LC-MS/MS on an LTQ Orbitrap mass spectrometer. All MS/MS spectra were searched against the NCBI database with a probability or significance threshold of p Ͻ 0.05 using the automated Mascot algorithm. Identification of a protein required that two or more MS/MS spectra matched peptides of the same protein entry in the database. MS analysis was performed at the MS and Proteomics Resource at the Yale School of Medicine.
Statistical Analysis-Data are expressed as the mean Ϯ S.D. Each experiment was performed a minimum of three times. The difference between groups was assessed with the Student's t test.

RESULTS
Characterization of CEP4 Proteins Mutated at PKC Consensus Sites-In an initial approach to define the site(s) of CEP4 phosphorylation, we evaluated Ser/Thr residues lying within PKC consensus sites. Whereas a canonical consensus site consists of a Ser or Thr flanked by one or two cationic residues (20), a partial site would be that in which only one side of the phos-phorylated site is adjacent to a cationic residue(s). Upon inspection of the CEP4 primary sequence, Ser 18 , Ser 77 , Ser 80 , and Ser 86 were found to be embedded in either a complete PKC consensus site (Ser 18 ) or partial site (Ser 77 , Ser 80 , and Ser 86 ) and were clustered in the N-terminal region (Fig. 1A). Site-directed mutagenesis of each Ser codon to an Asp codon created a negative charge, thereby simulating the presence of a phosphate group at this position in the protein. Each pseudophosphorylated mutant was expressed bearing both FLAG and myc epitope tags. Following transient transfection, whole cell lysates were analyzed for levels of mutant expression by Western blot analysis using anti-myc (Fig. 1B). It was observed that the mutant proteins were stably expressed at equivalent levels.
MCF-10A cells are weakly motile and express PKC isoforms at low levels. Because PKC activation promotes motility of these cells (4), it was of interest to determine whether any of the pseudophosphorylated mutants of CEP4 elicited increased motility in the absence of PKC activation. Therefore, each of the four mutants was tested for its ability to engender motility of MCF-10A cells relative to the WT and vector control (VC) (Fig.  1C). Compared with the WT, two mutants, namely S18D and S80D, promoted motility of these cells by 1.7-and 2.5-fold, respectively.
To validate that these sites are directly phosphorylated by PKC, in vitro phosphorylation of pure, recombinant CEP4 was performed with PKC␣ (5) followed by MS/MS analysis. It was noted that the phosphorylated form of CEP4 resolved by SDS- where the PKC consensus sites (full or partial) are underlined and the serine residues that are potential phosphorylation sites are identified (red). These sites flank the CRIB domain (green). B, myc-tagged CEP4 mutants were prepared in which an individual serine (S) was replaced with an Asp (D) residue followed by its expression in MCF-10A human breast cells. To demonstrate expression of the mutant proteins relative to a VC, Western blot analysis of whole cell lysates (50 g/lane) was performed with anti-myc (1:1000). Anti-␤actin (1:5000) was used to establish equivalent sample loading. C, each mutant CEP4-encoding plasmid was transfected into MCF-10A cells, and after 48 h, cell motility over 6 h was measured as described under "Experimental Procedures." The values are reported as the average of triplicate measurements ϮS.D. and are representative of three independent experiments. Statistical analysis was performed with the Student's t test that compared each selected mutant with the VC condition (**, p Ͻ 0.00001). Error bars represent S.D. SEPTEMBER 12, 2014 • VOLUME 289 • NUMBER 37 PAGE could be detected as two bands that had a slightly higher molecular mass than the untreated protein (75 kDa) ( Fig. 2A). A Western blot of the reaction product (Fig. 2B) demonstrated that CEP4 displayed a single strong phosphorylation signal when detected with an antibody (Cell Signaling Technology, Inc.) that specifically recognized the phosphorylated PKC con-sensus site (PKC substrate antibody). In contrast, untreated CEP4 gave no detectable phosphorylation signal. For each condition, the stained band at ϳ75 kDa was analyzed by MS/MS. The results of this analysis (Table 1) revealed phosphopeptides in PKC␣-treated CEP4 that contained phosphorylated Ser 18 and Ser 80 as well as additional sites of phosphorylation (see "Discussion"). MS/MS spectra for phospho-Ser 18 and phospho-Ser 80 are shown in Fig. 2C.

Phosphorylation of CEP4 by PKC Drives Motility
In light of the mutagenesis and MS/MS results, preparation of the S18D/S80D (D/D) double mutant and the corresponding phosphorylation-resistant S18A/S80A (A/A) double mutant was carried out. Their expression following transient transfection of MCF-10A cells is shown in Fig. 3A. Western blot analysis demonstrated that each double mutant was equivalently expressed. In Fig. 3B, cells transfected with the double mutants were assayed for motility and showed that the D/D mutant produced 2.5-fold higher motility relative to the A/A mutant, which produced basal motility comparable with that of the vector control.
To determine whether there were additional sites of phosphorylation produced by PKC, the phosphorylation-resis-   tant mutants of Ser 18 and Ser 80 were used to assess the extent of CEP4 phosphorylation in MCF-10A cells treated with DAG-lactone, a cell-permeable PKC activator (21). Each recombinant CEP4 was immunoprecipitated with anti-FLAG, and subsequently the level of phosphorylation was analyzed by Western blot using the PKC substrate antibody. A comparison of WT-CEP4 phosphorylation was made with cells treated with or without DAG-lactone where the unstimulated cells represented the basal condition. The results (Fig. 4A) indicated a strong level of phosphorylation of WT-CEP4 from cells treated with DAG-lactone as compared with WT-CEP4 isolated from unstimulated cells. Furthermore, the signal was decreased to basal levels by each of the single site mutants as well as by the double mutant. These results indicated that Ser 18 and Ser 80 were the principle sites in CEP4 that had undergone phosphorylation by DAG-stimulated PKC isoforms. The impact of the single site and double phosphorylationresistant mutants was also reflected in the degree to which each mutant decreased the motility of cells stimulated by DAG-lactone (Fig. 4B). Blockade of each site individually resulted in substantially decreased motility that was further diminished by the double mutant, thus demonstrating the functional significance of each of the two sites.
Phosphorylation of CEP4 Promotes Its Release from Constitutively Active Q61L-Cdc42-Because CEP4 is known to bind activated (GTP-bound) Cdc42 with high affinity, a central question was whether phosphorylation of CEP4 influenced its ability to engage Cdc42 (15,17). We addressed this question by using the constitutively active mutant of Cdc42 (Q61L-Cdc42), a mutant form that is known to bind CEP4 (15). Following bacterial expression, GST-Q61L-Cdc42 was semipurified by affinity chromatography on glutathione-Sepharose beads. A whole cell lysate was prepared from cells expressing the D/D mutant, A/A mutant, or VC. The lysate was added to the beads (collected as a pellet). Similarly, to stimulate CEP4 phosphorylation via PKC activation, cells expressing WT-CEP4 were treated with 10 M DAG-lactone or DMSO (0.05%, v/v) followed by preparation of cell lysates. Any myc-tagged CEP4 protein that had adsorbed to the immobilized Q61L-Cdc42 was subsequently detected by SDS-PAGE/Western blotting with antimyc. As can be seen in Fig. 5, intracellular phosphorylation of CEP4 via DAG-lactone treatment showed decreased binding to Q61L-Cdc42. Similarly, the D/D mutant bound weakly to Q61L-Cdc42 in contrast with the strongly bound A/A mutant. These results implied that phosphorylation or pseudophosphorylation of CEP4 at Ser 18 and Ser 80 releases it from activated (GTP-bound) Cdc42.
Binding Partners of Pseudophosphorylated CEP4 -To determine potential binding partners of phosphorylated CEP4, TAP was used. Each double mutant was subcloned into the TAP vector to generate CEP4 bearing two affinity peptide tags (at the C terminus) that permit two consecutive affinity chromatography steps on streptavidin-binding and calmodulin-binding resins, respectively. Lysis and chromatography steps were performed under detergent-free conditions that preserved protein-protein interactions between CEP4 and its unknown binding proteins (22). Following the second chromatographic step, proteins immobilized on the beads were resolved by SDS-PAGE, and the gel was stained with GelCode Blue. Fig. 6A demonstrates the expression of each TAP mutant. As shown in Fig.  6B, the D/D mutant revealed three major bands (44, 115-125, and 250 kDa) that were unique to proteins binding to the D/D mutant. Each band was excised and submitted for MS/MS analysis (data not shown). One identified protein (125-kDa band) was TEM4 (also known as ARHGEF17), which interacts with the small GTPases. Another identified protein (250-kDa band) was PARD6G, which is involved in cell polarity and serves as an  adapter that (via PAR3) binds GTP-bound Cdc42, Rac1, and atypical PKC/ (23)(24)(25). In view of their co-purification with D/D-CEP4 but not the A/A-CEP4 mutant, it is possible that these proteins co-localized with phospho-CEP4 in the cell.
Both TEM4 and PARD6G were validated as direct binding partners of CEP4 by performing co-immunoprecipitation of FLAG-tagged CEP4 mutants under the same non-detergent conditions as performed with the TAP mutants. Each pellet was subjected to SDS-PAGE and Western blotting and then probed with the corresponding antibody. When immunoprecipitated with the FLAG-tagged D/D-CEP4, strong bands were detected for PARD6G (60 kDa) and TEM4 (180 kDa) (Fig. 7A). In contrast, when immunoprecipitation was carried out with a lysate expressing the phosphorylation-resistant A/A-CEP4 mutant or VC, these bands were weak or undetectable. The results were consistent with a direct binding interaction between phospho-CEP4 and PARD6G or TEM4. It was further noted in these experiments that D/D-CEP4 co-immunoprecipitated a substantially lower amount of Cdc42 than did the A/A mutant, consistent with the lower affinity of phospho-CEP4 for Cdc42 as found in vitro with recombinant Q61L-Cdc42 (Fig. 5).
Because of previous reports that TEM4 activates RhoA GTPase, binds actin, and promotes migration (26 -28), a possible role for this GEF in phospho-CEP4-mediated motility was addressed by use of a TEM4-specific GFP-shRNA-encoding plasmid. Knockdown of TEM4 was achieved with this reagent as shown in Fig. 7B. While TEM4 knockdown was in effect, transfection of MCF-10A cells with each CEP4 double mutant or VC was performed followed by assay of cell motility. As shown in Fig. 7C, silencing of TEM4 resulted in inhibition by almost 40% of the motile behavior induced by the D/D-CEP4 mutant. Furthermore, the motility of these cells was decreased to the level of motility of cells co-expressing the A/A-CEP4 mutant and TEM4 shRNA. Interestingly, TEM4 silencing had a small stimulatory effect on motility of cells expressing A/A-CEP4, suggesting that TEM4 suppresses some element in the unstimulated pathway. Overall, these results implicate the TEM4⅐phospho-CEP4 complex as an active component in the PKC-induced motility signaling pathway.
Rac Activation Is Induced by Expression of Pseudophosphorylated CEP4 -In view of our previous findings that PKC activity promotes motility by activating Rac (4), we considered the possible involvement of Rac in CEP4-mediated phenotypes. To determine whether phospho-CEP4-induced motility relies on Rac activation, MCF-10A cells expressing the D/D-CEP4 mutant were treated with a Rac-specific inhibitor (NSC23766). This drug blocks the binding of GEFs, thereby preventing GDP-GTP exchange and Rac activation (29). As can be seen in Fig.  8A, highly motile cells expressing the D/D-CEP4 mutant were inhibited by 95% in the presence of 50 M inhibitor, thereby implicating activated Rac1 in the mechanism of cell movement. It is noted that for both Cdc42 and RhoA pulldown assays showed that neither of the CEP4 mutants resulted in a changed FIGURE 6. TAP of pseudophosphorylated CEP4 and its binding partners. Mutants of CEP4 were subcloned into a TAP vector that upon expression conferred epitope tags for isolating each mutant and its associated proteins by affinity chromatography as described under "Experimental Procedures." A plasmid encoding a CEP4 mutant was transfected into MCF-10A cells, and following affinity chromatography, bound proteins were resolved by SDS-PAGE and detected by GelCode Blue staining. A, Western blot analysis of whole cell lysates containing the TAP-CEP4 mutants or the VC probed with anti- myc (1:1000).  activation state. 4 That PKC signaling is apparently exclusive to Rac activation was also observed with another PKC substrate (␣-tubulin) that promotes the motility phenotype (3).
The next step was to determine whether phosphorylated CEP4 acts through TEM4 to activate Rac. For this purpose, cells were transfected with a CEP4 mutant (or VC) and either GFP-TEM4 shRNA or the scrambled control (SC) shRNA. Following preparation of cell lysates, pulldown of activated Rac was performed with p21-activated kinase-binding domain-agarose beads that specifically bound the activated form of Rac. As shown in Fig. 8B, when cells expressed the SC shRNA, the D/D-CEP4 mutant produced a strongly activated Rac signal as compared with the A/A-CEP4 mutant or the VC. It is noted that Rac activation was also observed in cells expressing only the D/D mutant (without SC shRNA transfection). 4 However, when TEM4 was silenced, the Rac signal was undetectable in cells expressing D/D-CEP4. This finding implicated TEM4 in the phospho-CEP4-mediated mechanism by which Rac undergoes activation.
Pseudophosphorylated D/D-CEP4 also produced cell surface projections and altered actin structure. In cells stained with rhodamine-phalloidin, prominent effects by the pseudophosphorylated mutant consisted of actin-based filopodia (microspikes) at the leading edge of Ͼ80% of transfectants as well as increased occurrence of actin stress fibers as shown in Fig. 9A. Far fewer (10%) filopodia were detected in cells expressing the A/A-CEP4 mutant or the VC (Fig. 9B). It is noted that expression of WT-CEP4 (without DAG-lactone treatment) elicited substantially higher numbers of filopodia than did expression of the A/A mutant. The difference in cell surface protrusions by WT-CEP4 and the A/A mutant correlated with their effects on motility (Fig. 3C) and their level of intracellular phosphorylation (Fig. 4A). Treatment of D/D-CEP4 transfectants with a Rac inhibitor (NSC23766) eliminated all actin-based features and correlated with the loss of motility (Fig. 8A). These observations showed that phospho-CEP4 produced activated Rac, and this was the mechanism that caused increased filopodia and motility of MCF-10A cells.

DISCUSSION
In this study, CEP4 was characterized as a new substrate of PKC in non-transformed MCF-10A human breast cells. Sitedirected mutants of CEP4 that had been pseudophosphorylated at both Ser 18 and Ser 80 of this abundant Cdc42-binding protein were found to engender cell motility, a previously unrecognized 4 X. Zhao, unpublished observations.   (Table  1). An important outcome of phosphorylation at these sites was that CEP4 lost its affinity for constitutively active Cdc42 in vitro (Fig. 5) and in cells (Fig. 7A), consequently leading to the acquisition of cell motility (Fig. 3) and the development of actinbased membrane protrusions (filopodia) (Fig. 9). The finding that phospho-CEP4 operated independently of Cdc42 to produce these phenotypes revealed a new dimension in the function and regulation of this effector protein. It is noted that the other four CEP isoforms do not possess PKC phosphorylation sites (17), suggesting that they are refractory to regulation by PKC.
Earlier studies characterized the CEPs as proteins that bind GTP-bound Cdc42 with high affinity (15,17). Cdc42 is bound at the CRIB domain in CEP4 that is located between Ile 27 and Leu 53 (Fig. 1A). Therefore, the two sites of phosphorylation identified here (Ser 18 and Ser 80 ) occurred on nearby serine residues that flank this region. Blocking of either site with an Ala residue resulted in equivalent loss of CEP4 phosphorylation and motility (Fig. 4). Furthermore, coincident blockade (A/A) or pseudophosphorylation (D/D) produced stronger effects. In the absence of an available three-dimensional structure of CEP4, we can only speculate that the negative charge introduced by phosphorylation of these Ser residues alters the threedimensional structure of the CRIB domain and thereby results in severely diminished binding affinity for activated (GTPbound) Cdc42. That Ser 18 and Ser 80 serve as the principal intracellular sites recognized by DAG-stimulated PKC isoforms is supported by the loss of phosphorylation to basal levels in the double phosphorylation-resistant (S18A/S80A) mutant of CEP4 expressed in cells that had been stimulated with DAGlactone (Fig. 4A). The additional sites of phosphorylation revealed by MS/MS analysis ( Table 1) of CEP4 that had been phosphorylated in vitro by PKC␣ (Ser 14 , Ser 72 , and Ser 74 ) also flank the CRIB domain. However, in view of the total blockade of CEP4 phosphorylation by the A/A mutant (Fig. 4A), phosphorylation of these additional sites may not have occurred to a significant extent during intracellular phosphorylation. Nevertheless, these findings provide an avenue for further investigation.
It is evident that phosphorylation at Ser 18 and Ser 80 provides a regulatory mechanism that frees CEP4 to engage in other protein-protein interactions that contribute to motile behavior. The nature of these interactions was addressed by using the TAP method. With this approach, we identified TEM4 (ARH-GEF17) as a protein that co-purifies and co-immunoprecipitates with D/D-CEP4 from whole cell lysates (Fig. 7A), suggesting that they form a complex in the intact cell. TEM4 is a GEF that was previously linked to the regulation of actin structure (26,28,30) and to migration by inhibiting actomyosin contractility at the leading edge of a moving cell (27). The important finding that knockdown of TEM4 was sufficient to suppress both motility (Fig. 7C) and Rac activation (Fig. 8B) induced by the D/D-CEP4 mutant of filopodia was a downstream consequence of D/D-induced signaling was indicated by an almost complete elimination of microspikes following treatment with the Rac-specific inhibitor NSC23766. As a GEF, TEM4 is thought to activate GDP-GTP exchange in RhoA but not in Rac or Cdc42 (30). In view of aberrations that can occur with the small GTPases and their effectors (31), it is possible that TEM4 or the phospho-CEP4⅐TEM4 complex interacts directly with Rac or with an additional, unknown component that affects Rac activation. It is noted that the inhibitor and Rac antibody used in these studies did not differentiate between Rac1 and Rac3, and therefore it is possible that both isoforms function in CEP4related phenotypes, similar to previous findings of their involvement in metastatic phenotypes of human breast cells (32).
During the course of these experiments, Cdc42 was also considered as a participant in one or both CEP4-induced phenotypes following its release from phospho-CEP4. Although principally known to promote cell polarity, Cdc42 was recently reported to suppress migration of breast cancer cells (33). This finding contrasts with the promotility effects of activated Rac1 (34). In the present study, there was no evidence that Cdc42 is involved in the PKC-mediated motility phenotype in agreement with an earlier report (4). It was further noted that adding a Cdc42 inhibitor (EMD-Millipore) resulted in a dramatic increase in cell motility regardless of the phosphorylation state of CEP4. 4 These results implied that activated Cdc42 suppresses motility of MCF-10A cells. Interestingly, the Cdc42 inhibitor also eliminated 50% of D/D-CEP4-induced filopodia. 4 Thus, although our investigations continue to support a major role for activated Rac in driving cell motility, it is likely that Rac and Cdc42 each contributes to cell surface protrusions.
A previous report from our laboratory (5) showed that PKC␣, PKC␦, and PKC (representing the subcategories of conventional, novel, and atypical PKC isoforms, respectively) can each carry out phosphorylation of CEP4 in MCF-10A cells. In the FIGURE 10. Model of PKC-mediated phosphorylation of CEP4 and its consequences. DAG-sensitive PKC isoforms phosphorylate CEP4 at Ser 18 and Ser 80 , leading to its dissociation from Cdc42. Phospho-CEP4 binds TEM4 that in turn stimulates GDP-GTP exchange in Rac1 and leads to downstream events supporting motility and formation of filopodia. A Rac1 inhibitor (I) (NSC23766) that specifically blocks GDP-GTP exchange on Rac1 interrupts this process, leading to complete inhibition of motility induced by phospho-CEP4. GAPs, GTPase-activating proteins. present work, DAG-lactone was used to activate DAG-dependent PKC isoforms to produce intracellular phospho-CEP4 (Figs. 4 and 5). The question of whether the same sites in CEP4 are recognized by DAG-independent isoforms such as PKC was not addressed in the foregoing studies. In this regard, we note that the co-localization of D/D-CEP4 with a PAR6 protein (PARD6G), an adapter that comes into contact with Cdc42 and PKC (via PAR3), suggests a means by which an interaction between dephosphorylated CEP4 and activated PKC may occur, particularly in human breast cells (24). Taken together, our findings support a model (Fig. 10) of an emerging PKC signaling pathway in which PKC-mediated phosphorylation of CEP4 causes its release from GTP-Cdc42 followed by formation of a complex of phospho-CEP4 with a GEF (e.g. TEM4) that results in Rac1 activation, membrane protrusion, and cell movement.
In an earlier effort to elucidate the components of the PKC signaling pathway, we identified ␣-tubulin as a PKC substrate (35). Phosphorylation of ␣-tubulin at a single site (Ser 165 ) resulted in its increased incorporation into microtubules that promoted increased interaction by microtubules with the membrane, activation of Rac, and increased motility (3). Apparently, Rac serves as a point of convergence for at least two PKC substrates, underscoring the importance of Rac-associated GEFs (and GTPase-activating proteins) to both motility signaling and membrane protrusions in breast cancer (6,7). In view of the fact that ␣-tubulin and CEP4 are both highly abundant PKC substrates that promote signaling through Rac, we speculate that their coincident phosphorylation by PKC could lead to continuous GDP-GTP exchange on Rac that consequently gives rise to aggressive cell movement.