Control of Protein Kinase C Activity, Phorbol Ester-induced Cytoskeletal Remodeling, and Cell Survival Signals by the Scaffolding Protein SSeCKS/GRAVIN/AKAP12*

Background: SSeCKS/GRAVIN/AKAP12 is a major protein kinase C substrate and binding protein. Results: SSeCKS binding of PKC via two homologous domains is required for attenuation of PKC activity and transduction of phorbol ester-induced cytoskeletal remodeling and survival signals. Conclusion: SSeCKS controls PKC signaling via direct scaffolding interactions. Significance: PKC signaling is regulated by the spatiotemporal scaffolding activity of SSeCKS. The product of the SSeCKS/GRAVIN/AKAP12 gene (“SSeCKS”) is a major protein kinase (PK) C substrate that exhibits tumor- and metastasis-suppressing activity likely through its ability to scaffold multiple signaling mediators such as PKC, PKA, cyclins, calmodulin, and Src. Although SSeCKS and PKCα bind phosphatidylserine, we demonstrate that phosphatidylserine-independent binding of PKC by SSeCKS is facilitated by two homologous SSeCKS motifs, EG(I/V)(T/S)XWXSFK(K/R)(M/L)VTP(K/R)K(K/R)X(K/R)XXXEXXXE(E/D) (amino acids 592–620 and 741–769). SSeCKS binding to PKCα decreased kinase activity and was dependent on the two PKC-binding motifs. SSeCKS scaffolding of PKC was increased in confluent cell cultures, correlating with significantly increased SSeCKS protein levels and decreased PKCα activity, suggesting a role for SSeCKS in suppressing PKC activation during contact inhibition. SSeCKS-null mouse embryo fibroblasts displayed increased relative basal and phorbol ester (phorbol 12-myristate 13-acetate)-induced PKC activity but were defective in phorbol 12-myristate 13-acetate-induced actin cytoskeletal reorganization and cell shape change; these responses could be rescued by the forced expression of full-length SSeCKS but not by an SSeCKS variant deleted of its PKC-binding domains. Finally, the PKC binding sites in SSeCKS were required to restore cell rounding and/or decreased apoptosis in phorbol ester-treated LNCaP, LNCaP-C4-2, and MAT-LyLu prostate cancer cells. Thus, PKC-mediated remodeling of the actin cytoskeleton is likely regulated by the ability of SSeCKS to control PKC signaling and activity through a direct scaffolding function.

trast, "receptors for activated C kinases" (RACKs) bind only activated forms of PKCs and thus are thought to promote phosphorylation of substrates. A third regulatory mechanism involves the direct interaction of certain PKC isoforms with signaling structures. For example, PKC␣ is enriched in caveolae, whereas PKC␤ and -⑀ bind to F-actin. A small group of scaffolding proteins including SSeCKS and AKAP79 are capable of binding PKC, PKA, calmodulin, PS, and phosphoinositol bisphosphates and thus may play a central role in regulating several cross-talking pathways (8).
The list of known PKC substrates includes signal transduction proteins thought to work at plasma membrane sites (e.g. MARCKS and SSeCKS), cytoskeletal proteins (GAP43, adducin, fascin, ERM proteins (ezrin, radixin, moesin), AFAP-110, myosin light chain, and vinculin), and nuclear structural proteins (e.g. lamin B). Many of these substrates tether microtubules to plasma membrane sites or facilitate actin fiber assembly. For example, the ability of PKC to phosphorylate myosin light chain directly affects the ability of the myosin light chain kinase to phosphorylate myosin light chain (9), resulting in altered cell motility and cell shape change (9 -13). However, little is known regarding the specific mechanism by which PKC substrates mediate cell shape change or the resulting biologies associated with cancer progression such as invasiveness, proliferation, and survival. Prostate cancer (CaP) progression has been linked to the activation and/or overexpression of specific PKC isoforms such as ␣, ␤, ⑀, , and (14,15). However, the transgenic expression of PKC⑀, but not PKC␣ or -␦, was sufficient to induce CaP (16). Further confounding the issue, PKC␣ promotes proliferation of androgen receptor-negative PC-3 CaP cells, whereas it along with PKC␦ promotes apoptosis (17) in the androgen-responsive cell line LNCaP (18 -22). Our laboratory identified SSeCKS (for Src-suppressed C kinase substrate) as a substrate of PKC whose expression is severely downregulated in src-and ras-transformed cells (23,24). Subsequently, Jaken and co-workers (25,26) identified STICK72, which encodes a "Ͼ200-kDa" PKC substrate with PS binding activity, as identical to SSeCKS. SSeCKS has many similarities to the well studied PKC substrate MARCKS, although they have quite different sequences and molecular weights. These include transcriptional suppression by v-src, PS-induced binding to PKC, N-terminal myristoylation, a predicted rodlike shape, heat stability, severely retarded mobility on SDS-PAGE, Ca 2ϩ -dependent calmodulin binding, and PMA-induced translocation to perinuclear sites (24,27). We recently showed that SSeCKS can associate with PKC␣, -␦, and -⑀ isozymes when overexpressed (28).
Here, we map PS-independent PKC-binding domains on SSeCKS and show that this binding inhibits PKC kinase activity. Furthermore, in the absence of SSeCKS, PMA-induced PKC activation failed to induce cell shape change and cytoskeletal reorganization. The finding that full-length SSeCKS, but not an SSeCKS mutant deleted of its PKC-binding domains, can rescue PMA-induced cell shape change in MEF and prostate cancer cells and can prevent PMA-induced apoptosis in LNCaP cells strongly suggests that SSeCKS mediates PKC-induced cytoskeletal signaling through a direct physical interaction with PKC.
GST Fusion and TAT-SSeCKS Protein Preparation-GST-SSeCKS fusion proteins were expressed in BL21-pLysS bacteria by induction with 1 mM isopropyl ␤-D-thiogalactopyranoside at 30°C for 1 h and purified as described previously (32) using 250 mM NaCl, 50 mM HEPES, pH 7.9, 5 mM EDTA, 0.1% Nonidet P-40, 0.5 mM DTT, 1 mM PMSF, and 20 mM reduced glutathione for protein elution from glutathione-Sepharose columns (GE Healthcare). TAT-SSeCKS was expressed in BL21-pLysS bacteria by induction with isopropyl ␤-D-thiogalactopyranoside at 30°C for 3 h followed by resuspension of bacterial pellets in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, and 10 mM imidazole), addition of lysozyme (0.2 mg/ml final concentration), and incubation for 30 min on ice followed by sonication. The lysates were boiled for 3 min and centrifuged at 14,000 ϫ g for 20 min at 4°C; His-TAT-SSeCKS protein was retained in the supernatant based on the predicted lack of SSeCKS secondary structure (32).
Immunoprecipitation (IP)-Cells were harvested, washed twice with ice-cold PBS, incubated for 30 min on ice in 0.2-0.5 ml of Nonidet P-40 lysis buffer (140 mM NaCl, 20 mM NaPO 4 , pH 7.8, 5 mM EDTA, 1% Nonidet P-40, 10 g/ml aprotinin, 10 g/ml leupeptin, 10 g/ml pepstatin, 10 mM sodium fluoride, 1 mM DTT, and 1 mM PMSF) to allow cells to swell, and then lysed by forcing the cell suspension five times through a hypodermic needle (25 gauge). The lysates were cleared by centrifugation (14,000 rpm, 4°C, 10 min) whereupon the supernatant (500 g of protein/sample) was precleared with a 10-min incubation at 4°C with 50 l of Protein A/G-agarose (Thermo/Pierce, Rockford, IL) and then incubated 2 h to overnight with 25 l of Protein A/G-agarose-Ab beads (Protein A/G preincubated for 1 h with 5-10 g of IP Ab and then washed twice with 1 ml of lysis buffer) plus lysis buffer to bring the final volume to 500 l. The pellets were washed thrice with lysis buffer and boiled for 3 min in 20 l of 2ϫ Laemmli buffer, and associated proteins were separated by SDS-PAGE. Transfer of the protein to PVDF and IB analysis were as above.
Pulldown Assay-GST fusion proteins (1 g each) were diluted in 200 l of ice-cold Binding Buffer (20 mM HEPES, pH 7.9, 1 mM MgCl 2 , 40 mM KCl, 0.1 mM EDTA, 0.1% Nonidet P-40, 1 mM DTT, and 1 mM PMSF) and then incubated with 25 l of washed glutathione-Sepharose TM 4B beads (GE Healthcare) with gentle agitation at 4°C for 1 h. The beads were washed thrice with radioimmune precipitation assay buffer and once with Binding Buffer (1 ml/wash). After resuspending in 245 l of Binding Buffer, either 5 l of in vitro transcribed/ translated (TNT, Promega, Madison, WI) protein or 500 g of cell lysate were added, and the mixture was agitated gently at 4°C for 2 h, washed four times (1 ml each) with Binding Buffer, boiled in Laemmli buffer, and then analyzed by SDS-PAGE.
Lipid/Protein Overlay Assays-Nitrocellulose membranes containing PS (Sigma-Aldrich) were produced as follows. PS was dissolved in chloroform and then diluted with an equal volume of 6% Triton X-100 in water followed by sonication thrice (5 s each). 10 l of various PS concentrations were spotted and dried at room temperature (RT) for 1 h followed by storage at 4°C in the dark. PS/phosphatidylethanolamine/ phosphatidylglycerol/phosphatidic acid membrane strips (Echelon Biosciences Inc., Salt Lake City, UT) were blocked with 3% BSA in PBS for 1 h at RT and then incubated for 1 h with the probe protein (1 g/ml) in blocking buffer. After three washes (5 min/wash) with PBS and 0.1% Triton X-100 (PBS/T), the membranes were processed with primary and secondary antibodies in standard IB (above).
PKC/SSeCKS Overlay Assay-Purified GST or GST-SSeCKS fusion proteins (1 g/slot in 250-l volumes) were loaded onto PVDF membranes using a Slot-Blot apparatus (Bio-Rad) whereupon the membranes were dried at room temperature and stored at 4°C. Membranes were blocked with 3% BSA in PBS for 0.5 h at room temperature and then washed three times with PBS/T. Membranes were incubated with purified PKC (1 g/ml) in 3% BSA in PBS/T for 1 h at RT, washed, and processed for PKC-specific IB.
In Vitro Transcription and Translation-One microgram of plasmid DNA was added to T7-TNT reaction assays (Promega) containing 25 l of rabbit reticulocyte lysate, 2 l of TNT reaction buffer, 1 l of amino acid mixture minus methionine (1 mM), 2 l of [ 35 S]methionine (10 Ci/l), 1 l of RNase inhibitor (40 units/l), 1 l of T7 RNA polymerase, and double distilled H 2 O to make 50 l total and incubated at 30°C for 90 min.
PKC Kinase Activity Assay-Total cellular PKC kinase activity was measured using a solid phase enzyme-linked immunoabsorbent assay kit (Assay Designs/Stressgen, Ann Arbor, MI) according to the manufacturer's specifications. Absorbance of the chromogenic substrate product was measured at 450 nm, and the relative PKC kinase activity was calculated as follows: (average sample absorbance Ϫ average blank absorbance)/volume (l) ϫ dilution factor. PKC␣ kinase activity either from IPs or using baculovirus-expressed enzyme (Biomol) was assayed in 60-l reactions containing PKC␣ in 20 l of kinase buffer (50 mM HEPES, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 2.5 mM EGTA, 1 mM sodium fluoride, 0.1 mM Na 3 VO 4 , and 10 mM ␤-glycerophosphate) plus 40 l of kinase buffer containing 5 g of myelin basic protein (MBP) or histone H1 substrate and 5 Ci of [␥-32 P]ATP. The reactions were incubated at 30°C for 30 min and terminated by adding SDS sample buffer followed by boiling for 3 min. The reaction products were analyzed by SDS-PAGE and phosphorimaging (Typhoon ImageQuant, GE Healthcare).
PMA-induced Cell Shape Change-Phorbol 12-myristate 13-acetate (Avanti Polar Lipids Inc., Alabaster, AL) was dissolved in dimethyl sulfoxide as a 5 mM stock solution and stored in aliquots at Ϫ20°C. Cultures of MEF were plated 18 h before PMA treatment, which was 100 nM for various lengths of time.
Immunofluorescence Analysis-Cells plated on glass coverslips (22 mm 2 ) were fixed at Ϫ20°C for 20 min with 60% acetone and 3.7% paraformaldehyde in PBS (for F-actin staining) or in 4% paraformaldehyde and 0.5% Triton X-100 at room temperature for 10 min followed by two PBS washes. The slides were blocked with 3% nonfat dry milk in PBS for 30 min at room temperature. F-actin was stained with FITC-phalloidin (Sigma), and nuclei were stained with DAPI (Invitrogen; 1:500). Fluorescent images were captured using a Nikon TE2000-E inverted microscope equipped with a Roper CoolSnap HQ charge-coupled device camera.
Apoptosis/Cell Death Assay-Apoptosis and cell death were measured, respectively, in LNCaP cells labeled with phycoerythrin-Annexin V and 7-aminoactinomycin D using a V-PE Apoptosis Detection kit (BD Biosciences/Pharmingen) according to the manufacturer's specifications. The stained cells were detected in the Roswell Park Cancer Institute Flow and Image Cytometry Core Lab (Paul Wallace, Director) using a BD Biosciences FACSCalibur and FCS Express 3.0 software (De Novo Software, Los Angeles, CA).

PS-dependent and -independent SSeCKS/PKC Binding-The
original finding that SSeCKS binds PKC comes from overlay assays using high, non-physiologic levels (10 -100 g/ml) of PS (25). We also showed (32) that the PS-independent co-IP of SSeCKS and PKC could be enhanced by the addition of high levels of PS (123 M). In contrast, Piontek and Brandt (36) showed that the ability of GRAVIN (the human SSeCKS/ AKAP12 orthologue) to co-immunoprecipitate PKC␣ or -␤II was not enhanced by PS. Thus, we investigated whether SSeCKS encodes distinct PS-dependent and -independent binding activities and if so whether these were encoded by different SSeCKS protein domains. Nitrocellulose strips containing increasing concentrations of PS were incubated with either equal amounts of GST, GST-PKC␣, or His-SSeCKS, and after stringent washing, the blots were immunoblotted for either GST or SSeCKS. SSeCKS and PKC␣ bound levels of PS (Fig. 1A) that are physiologically relevant given that 16 -20% of the phospholipids of a cell are PS and that cell membranes typically consist of 6 -10% PS (37). SSeCKS also bound phosphatidic acid and phosphatidylglycerol with similar affinities but failed to bind phosphatidylethanolamine, phosphatidylcholine, phosphatidylinositol, sphingomyelin, or diacylglycerol (Fig. 1B). As  30)) grown in the absence of tetracycline were immunoprecipitated with NRS, anti-SSeCKS, or anti-PKC␣ Ab, and then the IPs were probed by IB using the reciprocal Ab. E, left panel, autoradiography of 35 S-labeled full-length PKC␣ produced in TNT reactions; right panel, pulldown of labeled PKC␣ proteins using beads containing NRS or SSeCKS (after IP from S24 cell lysates). F, beads with GST or GST-PKC␣ (Coomassie-stained in the bottom panel) were used to precipitate His-SSeCKS protein in the presence or absence of PS (20 g/ml) followed by SSeCKS IB (top panel). a control, the His-DHRS12-P1 protein encoding a mitochondrial dehydrogenase (a gift from Lionel Coignet, Roswell Park Cancer Institute) failed to bind to any of the lipids, indicating that the His-SSeCKS binding was not due to the shared His-TAT N-terminal tag. Taken together, these data suggest that SSeCKS binding to PS may have physiologic significance.
In an attempt to determine which PKC isoform classes could associate with SSeCKS in a PS-independent manner, IPs of SSeCKS from lysates of SSeCKS-overexpressing NIH3T3 cells (30) were immunoblotted with PKC isoform-specific Ab or with a pan-PKC serum that recognizes PKC␣, -␤, and -␥ isoforms. These cell lysates are not PS-free given that roughly twothirds of cellular PS is not membrane-associated (37). SSeCKS readily associated with several cPKC (␣, ␤I, ␤II, and ␥) and nPKC (␦ and ) isoforms without the need for added PS (Fig.  1C), although no association with the atypical PKC isoform was detected (not shown). As a negative control, IP with normal rabbit serum (NRS) resulted in no reactivity to a pan-PKC serum. Thus, SSeCKS binding of PKC is not isoform-specific. The SSeCKS/PKC association could be demonstrated by recip-rocal co-IPs from SSeCKS-overexpressing NIH3T3 lysates (Fig.  1D) by pulldown of 35 S-labeled PKC␣ using SSeCKS beads (versus NRS beads) (Fig. 1E) or by pulldown of overexpressed SSeCKS using GST-PKC␣ versus GST beads (Fig. 1F). Note that in the latter example, the addition of PS failed to enhance the association of SSeCKS and PKC␣.
The PS-independent binding sites on SSeCKS were identified by performing pulldown assays with bait beads containing fusions of GST to non-overlapping SSeCKS fragments of roughly 200 aa ( Fig. 2A, bottom panel) incubated with cell lysates from HEK293T cells forced to overexpress PKC␣. After probing these pulldowns with anti-PKC␣, two SSeCKS regions exhibited association: aa 553-720 and 721-900 ( Fig. 2A, top  panel). The inability of other GST-SSeCKS fusions and of GST alone to pull down PKC␣ indicates that the binding by the 553-720 and 721-900 regions was specified by their SSeCKS sequences. To further determine whether these regions are required for SSeCKS-PKC binding, an in-frame aa 553-900 SSeCKS deletion mutant was produced ("⌬553-900"). We then assessed the binding to GST or GST-PKC␣ beads by 35 S-labeled full-length and ⌬553-900 SSeCKS products synthesized in coupled in vitro transcription/translation systems. Whereas the full-length SSeCKS product could be pulled down by the GST-PKC␣ beads, loss of the aa 553-900 domain ablated this binding (Fig. 2B).
Because the pulldown assays in Fig. 2, A and B, contain endogenous levels of PS from cell lysates, we sought to determine whether the two SSeCKS domains, aa 553-720 and 721-900, could bind PKC as purified proteins. Thus, produced slot blots containing equal amounts (1 g/slot) of GST or GST-SSeCKS fusion protein were overlaid with various PKC preparations (or GST as a negative control) followed by IB with various PKC Abs. A commercial enzymatically active PKC␣ enzyme preparation (Biomol) bound to both domains (Fig. 2C), whereas bacterially expressed GST-PKC␣ bound only to 721-900 (Fig. 2B). Importantly, the GST-PKC␣ is not enzymatically active (supplemental Fig. S1) likely because of the absence of phosphorylation in the activation loop (1). PKC␣ labeled with 35 S in TNT reactions only bound the 721-900 SSeCKS domain (Fig. 2C). In our hands, this form of PKC had no discernable PKC activity (data not shown), and thus, there is a correlation between the lack of enzymatic activity and failure to bind the aa 553-720 SSeCKS domain. Interestingly, preincubation of the bacterial GST-PKC␣ with 100 g/ml PS, which marginally increased its enzymatic activity level (supplemental Fig. S1), induced binding to both 553-720 and 721-900 (Fig. 2C). It is noteworthy that enzymatically active PKC⑀ only bound to the aa 721-900 domain (first panel on the right), and we speculate that because this isoform has a much lower affinity for PS than PKC␣ (38) it does not undergo a PS-mediated conformation change that would allow it to recognize the aa 553-720 domain. Nonetheless, in contrast to previous results by Piontek and Brandt (36) using pulldowns, this result plus the data in Fig. 1C and our recently published data (28) show that under these saturated binding conditions PKC⑀ can bind to at least one SSeCKS domain. Taken together, our results strongly suggest that the 721-900 domain on SSeCKS binds PKC irrespective of enzyme activity, whereas the 553-720 domain binds a second PKC conformer associated with enzyme activity and possibly PS binding in the case of PKC␣.
Although neither of the PKC binding SSeCKS regions shares homology with the PS-independent PKC-binding domain on RACK (39), they share a 28-aa sequence, EG(I/V)(T/S)XWXS-FK(K/R)(M/L)VTP(K/R)K(K/R)X(K/R)XXXEXXXE(E/D), having 43% identity and 71% homology (Fig. 2D, upper panel). To show that the two homologous regions facilitate the PKC␣ binding activity of the 553-720 and 721-900 domains, GST fusions were produced lacking these regions (⌬596 -605 and ⌬745-753). Beads loaded with GST or these SSeCKS domains were used to pull down overexpressed PKC␣ from 293T cell lysates as in Fig. 2A. Loss of either homologous region ablated the PKC␣ binding activity of the respective 553-720 and 721-900 domains (Fig. 2E). Moreover, PKC␣ binding by each domain could be competed by incubating with molar excess levels of peptides covering either homologous sequence ("599" or "748"; Fig. 2D, lower panel) but not by a non-homologous SSeCKS peptide ("835") (Fig. 2E). These data are consistent with SSeCKS encoding two independent yet homologous PKC␣ binding sites.
Effect of Culture Confluence on SSeCKS/PKC Association-Data from our laboratory and others suggest that the scaffolding activity of SSeCKS for PKC is antagonized by the phosphorylation of SSeCKS by PKC and other mitogen-activated kinases (32). Although the abundance of SSeCKS protein rises precipitously in confluence (40), the relative phosphorylation of SSeCKS decreases (41). Moreover, PMA or serum can induce serine phosphorylation of SSeCKS in cycling cells but not in confluent cultures (41). The increasing abundance of SSeCKS protein in confluent versus subconfluent WT-MEF correlated with decreased levels of total cellular PKC activity (Fig. 3A). Although PKC␣ proteins levels were unaffected by confluence in WT-MEF or in KO-MEF isolated from an SSeCKS-null mouse model (42), KO-MEF had roughly half the PKC␣ levels of WT-MEF (Fig. 3, A (upper panel) and B). Importantly, however, subconfluent KO-MEF exhibited roughly 3-fold more total PKC activity (based on an ELISA using a universal PKC substrate; Fig. 3A) or PKC␣ activity (based on the in vitro kinase activity of immunoprecipitated PKC␣ against MBP substrate; Fig. 3B) than WT-MEF, and this activity increased in confluent KO-MEF (Fig. 3A). More PKC␣ co-immunoprecipitated with SSeCKS from lysates of confluent cells (Fig. 3C, inset panel), and that population had roughly 40% less relative kinase activity than the co-immunoprecipitated PKC␣ from subconfluent lysates (Fig. 3C). PMA was capable of activating PKC to similar extents in subconfluent cultures of KO-and WT-MEF (Fig.  3D), ruling out the possibility of selective sensitivity in the KO-MEF. (Note that cultures of KO-MEF become contacted-inhibited (not shown), and thus, the lack of SSeCKS or the lack of PKC attenuation is not sufficient to override contact inhibition.) Taken together, these data suggest that the relatively underphosphorylated SSeCKS found in confluent cells has increased scaffolding activity for PKC and that SSeCKS binding in confluent cells inhibits PKC activity.
SSeCKS Inhibits PKC Activity-To investigate whether SSeCKS could directly inhibit PKC activity, 293T cells were co-transfected with HA-PKC␣ plus expression vectors for pEGFP or SSeCKS constructs, and then PKC␣ activity was monitored in HA IPs using the in vitro kinase/MBP assay described in Fig. 3B. Whereas full-length (FL) SSeCKS inhibited PKC␣ activity 2.5-fold, deletion of the two major PKC-binding regions (aa 553-900) abrogated this effect (Fig. 3E). In contrast, deletion of an upstream region (aa 2-553) shown not to encode PKC binding activity (Fig. 2, A and C) did not decrease PKC inhibition.
To determine whether SSeCKS scaffolding activity for PKC was responsible for the confluence-mediated PKC attenuation, KO-MEF were co-transfected with expression vectors encoding GFP fusions of either full-length SSeCKS or SSeCKS deleted of its PKC-binding domains (⌬553-900) (Fig. 3F). After isolation of the transfectants by FACS for GFP, aliquots of the cells were cultured under subconfluent conditions for an additional 24 h, and then cell lysates were assayed for total PKC activity. Full-length but not ⌬553-900 SSeCKS could rescue PKC attenuation (Fig. 3G), indicating that SSeCKS likely attenuates PKC during contact inhibition through a direct scaffolding activity.
We sought to determine whether the PKC kinase-inhibitory activity of SSeCKS mapped to the two binding motifs, aa 596 -605 and 745-753. Thus, the in vitro kinase activity of purified, enzymatically active PKC␣ against MBP substrate was assessed in the presence of a molar excess (25ϫ) of either GST or GST-SSeCKS proteins. Whereas the individual PKC-binding domains in SSeCKS, aa 553-720 or 721-900, could inhibit PKC␣ activity, deletion of the aa 596 -605 and 745-753 motifs in these proteins, respectively, abrogated this inhibition (Fig.  3H). As a control, the addition of a C-terminal SSeCKS domain encoding the PKA-Regulatory Subunit II-binding (AKAP) motif (43) failed to inhibit the kinase activity. Given that SSeCKS is also a major PKC substrate (26,32), the GST and GST-SSeCKS fusion proteins from the PKC␣ kinase assay described in Fig. 3H were phosphorimaged. Only the aa 533-720 domain of SSeCKS was robustly phosphorylated (Fig. 3I). Importantly, although it was in molar excess, the relative phosphorylation of this SSeCKS domain was roughly 12% of that of the MBP substrate (data not shown). The fact that deletion of aa 596 -605 severely decreased PKC␣-mediated phosphorylation of the aa 553-720 domain along with the inability of the aa 721-900 domain to be phosphorylated (Fig. 3I) strongly suggests that PKC␣ phosphorylates Ser-613 (Ser-627 in rat and Ser-614 in human) and not a motif shared with the aa 741-769 PKC-binding domain (Fig. 2D). Indeed, phosphorylation of SSeCKS Ser-613 has been documented in at least 25 proteomics studies on PhosphoSitePlus. Taken together, these data suggest that SSeCKS inhibits PKC activity via direct scaffolding to its two PKC-binding domains, aa 596 -605 and 745-753, and not via substrate competition.
SSeCKS/PKC Association Required to Transduce PMA-induced Cytoskeletal Reorganization-It is well established that PKC activation by short term PMA treatment induces major remodeling of the actin cytoskeleton in many cell types, starting with the collapse of actin fiber networks to the perinucleus and a loss of peripheral focal adhesions, resulting in increased cell rounding (1). Given that SSeCKS is an in vitro and in vivo PKC substrate (26,32,44) and that short term PMA treatment induces a co-translocation of SSeCKS and actin fibers to the perinucleus (32,36,45), we addressed whether KO-MEF were defective in PMA-induced PKC translocation, cytoskeletal remodeling, and cell rounding. Although KO-MEF exhibit increased PMA-induced PKC activation (Fig. 3D), immortalized KO (imKO)-MEF (28) showed little plasma membrane translocation of GFP-PKC isozymes or cell rounding after short term PMA treatment, whereas PKC␣, -␤I, -␥, -␦, or -⑀ moved to the cell periphery in PMA-treated immortalized WT-MEF concomitant with increased cell rounding (Fig. 4, arrows). As a control, pEGFP-transfected immortalized WT (imWT)-MEF showed a PMA-induced cell shape change but no membrane enrichment of GFP. These data strongly suggest that SSeCKS is required to scaffold activated PKC isozymes to sites of activity such as those involving substrate phosphorylation regulating actin cytoskeleton remodeling.
PMA induced cell rounding in a dose-dependent manner in confluent cultures of WT-MEF treated for 1 h with 100 nM, showing cell rounding in Ͼ80% of cells (counted in three independent fields of at least 200 cells in two independent experiments; Fig. 5A, top panel). Cell rounding was defined as the loss of flat cytoplasmic surfaces extending from the perinuclear area and the concomitant increase in phase-contrast refractility. In contrast, KO-MEF treated with 10 nM-1 M PMA showed minimal cell shape change with only slight cell rounding at 10 M PMA (Fig. 5A, bottom panel). Treatment of subconfluent cultures for even shorter periods (15-30 min) with 100 nM PMA yielded similar results, namely cell rounding in WT-MEF but little effect in KO-MEF (data not shown), consistent with our previous finding that culture confluence suppresses the ability of PMA to activate PKC (41). Longer treatment with 200 nM PMA induced cell rounding and the loss of F-actin stress fibers after 45 min in imWT-but not in imKO-MEF (Fig. 5B). Moreover, PMA failed to cause disassembly of F-actin stress fibers in the KO-MEF (Fig. 5C).
To address whether the PKC scaffolding activity of SSeCKS is required for PMA-induced cell shape change, KO-MEF transiently co-transfected with expression vectors for SSeCKS plus EGFP were treated with 100 nM PMA for 30 min. Rounded versus flat GFP-expressing cells were scored by simultaneous phase-contrast and fluorescence microscopy (Fig. 5D, supplemental Fig. S2, and supplemental Fig. S4 for real time video microscopy) with the untransfected cells acting as internal controls. The expression of FL SSeCKS significantly restored the ability of PMA to induce cell rounding (Fig. 5, D and E). In contrast, expression of the ⌬553-900 SSeCKS mutant failed to restore PMA-induced cell rounding; mutants lacking either PKC-binding domain, aa 596 -605 or 745-753, partially restored cell rounding, whereas a mutant lacking both was incapable of restoration ( Fig. 5E and supplemental Fig. S3). This suggests that the two SSeCKS PKC-binding domains work independently in cells. Taken together, these findings are consistent with a role for SSeCKS in mediating PKC-induced cytoskeletal reorganization through a direct association with PKC.
We investigated whether the deficiency in SSeCKS levels in rat and human CaP cell lines that we reported previously (31,46) might play a direct role in facilitating PMA-induced cytoskeletal reorganization. The relative SSeCKS protein level in MAT-LyLu (MLL) cells, an androgen-responsive rat CaP cell line, was roughly 8-fold lower than in the untransformed, SV40 T antigen immortalized rat prostate epithelial cell line EP12 (47) (Fig. 6A, left panel). The forced re-expression of FL SSeCKS induced cell flattening prior to PMA treatment (Fig. 6B) consistent with our previously reported role for SSeCKS in re-establishing an actin-based cytoskeletal architecture and cell-cell adherence in MLL cells (31). Addition of PMA to MLL/vector cells (Fig. 6B, "Ϫ SS") induced only marginal cell shape change between 20 and 60 min of treatment, whereas even short term (20 min) treatment of the SSeCKS-re-expressing cells (Fig. 6B, "ϩ SS") caused a significant increase in cell rounding. These findings suggest that SSeCKS may facilitate the ability of PMA to induce cytoskeletal reorganization in MLL cells.
To strengthen this notion, we reproduced this analysis in human LNCaP cells, which exhibit severe down-regulation of SSeCKS (25-fold) compared with the untransformed, SV40 T antigen immortalized human epithelial cell line P69SV40T as we reported previously (48). Even with this down-regulation, SSeCKS is likely to exert some regulatory functions in this cell type because SSeCKS protein levels were still induced by culture confluence (Fig. 5A, right panel). In contrast, the metastatic, androgen-independent derivative, LNCaP-C4-2 (49), showed no confluence-induced SSeCKS. PMA treatment of LNCaP or LNCaP-C4-2 cells induced severe cell flattening (Fig.   FIGURE 4. Defective plasma membrane translocation of PKC isozymes in SSeCKS-null cells. imWT-MEF or imKO cells transfected with GFP fusion constructs of PKC isozymes (versus GFP alone) were treated with 100 nM PMA for 30 min and then imaged. Membrane translocation of PKC is shown by arrows. 6, C and E) and apoptosis (Fig. 6D) in agreement with previous publications (21, 22, 50 -52). However, the re-expression of FL SSeCKS ␣ or ␤ isoform rescued these cells from PMA-induced apoptosis (Fig. 6D), and FL ␣SSeCKS, but not the SSeCKS[⌬553-900] mutant, could facilitate PMA-induced cell rounding (Fig. 6E). Taken with the data for MLL cells and KO-MEF, these findings indicate that SSeCKS is sufficient to facilitate PMA-induced cell shape change likely through a direct interaction with PKC.

DISCUSSION
The phenomenon of PMA-induced actin cytoskeletal remodeling has been well described in the literature for some time, and there is clear evidence that the reorganization of the cytoskeletal architecture, mediated in major part through the activation of PKC isoforms, controls oncogenic processes such as proliferation, cell survival, and invasiveness (1). Here we show that a major PKC substrate, SSeCKS/GRAVIN/AKAP12, is required in MEF and CaP cells to mediate PMA-induced cell shape change and in the case of CaP cells to suppress PMAinduced apoptosis. Moreover, we show that a region in SSeCKS containing at least two PKC-binding domains is required to rescue the PMA effect on the actin cytoskeleton. These data strongly suggest that PKC signals to the cytoskeleton are mediated through a direct scaffolding interaction between SSeCKS and PKC. Consistent with this notion, SSeCKS-null MEF were defective in PMA-induced cytoskeletal reorganization and cell shape change. This deficiency was unaffected by cell confluence, and moreover, these cells were not defective in PKC kinase activation by PMA, although they were defective in the ability to translocate PKC isozymes to the plasma membrane.
Previous attempts to demonstrate PKC-SSeCKS binding involved overlay protocols requiring high, non-physiologic levels of PS, specifically 100 -400 g/ml (25,44). Indeed, we showed previously that the addition of high levels of PS enhanced the formation of PKC-SSeCKS complexes in pulldown experiments from endogenous protein lysates (32). Here, we confirm that SSeCKS binds PS independently and at levels likely to be physiologically relevant. The affinity of SSeCKS for PS is likely facilitated by the potential PS-binding motif, 506 FSSSGLKKLSGKKQR 520 , which is homologous (bold residues) to the known PS-binding motif in PKC, FXFX-LKXXXXKXXR (53). Interestingly, SSeCKS also binds phosphatidic acid and phosphatidylglycerol albeit at affinities 2-4fold lower than for PS. It should be noted that the direct binding affinity of SSeCKS for PS is at least 2 orders of magnitude lower than what is required to facilitate the overlay experiments described above. This suggests that the PKC/SSeCKS association in the overlay assays may be indirect, i.e. facilitated by the independent association of PKC and SSeCKS to PS in structures promoted by high concentrations of PS (e.g. micelles). Indeed, our current results indicate that the addition of PS did not enhance the overall binding level between bacterially pro- duced GST-PKC␣ and His-SSeCKS, although it did induce binding site preference (below). This is the first description of a PS-independent protein-protein interaction between PKC and SSeCKS. A role for PS in facilitating PKC/SSeCKS association in cells cannot be ruled out, however, because we show reciprocal co-IP between SSeCKS and various cPKC or nPKC family members from cell lysates, which likely contain physiologic levels of free PS (non-membranous). This agrees with previous findings from Piontek and Brandt (36) in which endogenous human SSeCKS (GRAVIN) and PKC␣, -␤I, and -␤II could be co-immunoprecipitated from lysates of NT2-N1 neurons without the addition of PS.
Our data identify two independent PKC-binding domains on SSeCKS, aa 596 -605 and 745-753, which share the common motif EG(I/V)(T/S)XWXSFK(K/R)(M/L)VTP(K/ R)K(K/R)X(K/R)XXXEXXXE(E/D). The aa 745-753-containing domain bound all three preparations of PKC␣ we tested irrespective of kinase activity. This domain also bound a commercial enzymatically active preparation of PKC⑀, a finding that contrasts with a previous report (36), although it should be noted that our conditions are at saturating levels. In contrast, enzymatically inactive preparations of PKC␣ failed to bind to the aa 596 -605-containing domain, whereas the enzymatically active commercial PKC␣ preparation bound this domain. Moreover, the preincubation of bacterially expressed GST-PKC␣ with high PS levels, a treatment that marginally activates PKC, induced binding to the aa 596 -605-containing domain. Thus, the 596 -605 domain might have higher affinity for enzymatically active or PS-stimulated conformations of PKC␣.
SSeCKS association with PKC␣ increases in confluent cultures, a condition in which SSeCKS protein accumulation is induced significantly (40) and in which SSeCKS phosphorylation by PKC (following PMA addition) is minimized (54). Our data suggest that SSeCKS binding under these cellular conditions inhibits PKC␣ activity. This is based on our finding that the ability of the aa 553-720 and 721-900 domains to inhibit PKC␣ activity in vitro was dependent on retention of their PKC-binding domains (i.e. deletion of either aa 596 -605 or 745-753 abrogated the inhibitory activities). The finding that total cellular PKC activity in SSeCKS-null MEF was more than double that found in WT-MEF strengthens the notion that a function of SSeCKS in cells is to attenuate PKC activation through its scaffolding activity.
Our model suggests that SSeCKS regulates PKC activity in a dynamic, spatiotemporal process: during contact inhibition when SSeCKS-PKC binding is saturated, SSeCKS inhibits PKC activity, whereas in mitogenically activatable cells, SSeCKS binding to PKC facilitates its translocation to sites where it can phosphorylate substrates involved in cytoskeletal remodeling and cell survival. Continued mitogenic stimulation results in the phosphorylation of SSeCKS by PKC and other kinases, antagonizing its binding to PKC and acting as a negative feedback to disengage PKC from mediating signals to the actin cytoskeleton. This might explain why in CaP lacking SSeCKS PMA fails to induce the cytoskeletal remodeling that regulates normal cell cycle progression, resulting in a reactive program of apoptosis. In this context, SSeCKS down-regulation promotes CaP malignancy by facilitating PKC-induced invasiveness and proliferation at the cost of increased apoptosis.