A novel src- and ras-suppressed protein kinase C substrate associated with cytoskeletal architecture.

We previously identified a novel src- and ras-suppressed gene, 322, encoding a mitogenic regulatory function (Lin, X., Nelson, P. J., Frankfort, B., Tombler, E., Johnson, R., and Gelman, I. H. (1995) Mol. Cell. Biol. 15, 2754-2762). Here, we characterize the 322 gene product as an in vivo and in vitro substrate of protein kinase C (PKC). Hence, we named this product SSeCKS (pronounced essex) for Src Suppressed C Kinase Substrate. Rabbit polyclonal sera raised against glutathione S-transferase (GST)-SSeCKS recognized a myristylated 280/290-kDa doublet in Rat-6 fibroblasts. SSeCKS levels in src- and ras-transformed Rat-6 cells were 15- and 8-fold less, respectively, than those in untransformed cells. Short-term addition of phorbol ester resulted in a 5-fold increase in SSeCKS phosphorylation which was inhibited by bis-indolylmaleimide. In vitro phosphorylation of GST-SSeCKS by purified rabbit brain PKC-α was enhanced by phosphatidylserine and blocked by excess PKC pseudosubstrate inhibitor peptide. GST-SSeCKS bound purified PKC-α or PKC from Rat-6 lysates in a phosphatidylserine-dependent manner. Four SSeCKS domains containing Lys/Arg-rich motifs similar to the PKC phosphorylation site in MARCKS were phosphorylated in vitro by PKC. Immunofluorescence analysis showed SSeCKS present throughout the cytoplasm with enrichment in podosomes and at the cell edge. Short-term addition of phorbol esters caused the movement of SSeCKS from plasma membrane sites to the perinucleus coincident with a loss of actin stress fibers. These data suggest a role for SSeCKS in the control of cellular cytoskeletal architecture.

The ability of the oncogene product encoded by Rous sarcoma virus, p60 v-src , to induce morphological transformation and tumorigenesis is dependent on its intrinsic protein tyrosine kinase activity and ability to recognize cellular substrates. The transforming activity of p60 v-src is controlled by at least four modular domains including a Src-homology 1 (SH1) encoding the kinase domain, an SH2 and an SH3 domain which binds phosphotyrosine residues or polyproline tracts, respectively, on specific signal proteins, and an SH4 motif which directs Nterminal myristylation and thus, association with the plasma membrane (2)(3)(4).
The src oncogene is known to alter cellular signal pathways controlling mitogenesis, cell cycle, and morphology. p60 v-src is thought to initiate mitogenic signal cascades by directly phosphorylating signaling proteins at plasma membrane sites, resulting in the association with and activation of messenger molecules such as GAP, 1 phospholipase-C␥, p21 ras , the c-RAF serine/threonine kinase, MAP kinase kinase, and MAP kinase. Indeed, MAP kinase is activated early after v-src expression (5). This signal ultimately results in the expression and/or modification of "immediate-early" transcription factors such as those comprising the AP-1 complex, NGFI-A, KC, and PC4, followed by induction of DNA replication and cell proliferation (6,7). Recent evidence indicates that v-src can directly activate STAT-3, presumably at plasma membrane sites, resulting in the translocation of STAT-3 to the nucleus where it induces the transcription of mitogen-response genes (8).
In contrast, more recent evidence suggests that src controls cytoskeletal architecture via PKC and GTPase proteins such as Rho and Rac. These pathways may require nuclear events as well as direct effects by p60 v-src in the cytoplasm. For example, p60 v-src affects cell architecture and cell-to-cell interactions by either down-regulating the transcription of several cytoskeletal and extracellular matrix components such as tropomyosin, fibronectin, and type I collagen (9,10), or by directly inducing the phosphorylation of extracellular matrix ligand proteins such as N-cadherin (11) or cytoskeletal proteins such as tropomyosin and talin (12). p60 v-src also phosphorylates and activates the focal adhesion kinase (p125 FAK ), which controls the formation of actin stress fibers and adhesion plaques (12). The association of these fibers with integrins is further modulated by the tyrosine phosphorylation of tethering proteins such as tensin and vinculin (12,13). Because integrins bind directly to extracellular matrix proteins, these events effectively regulate controls on contact inhibition and interactions with neighboring cells. Most importantly, several studies demonstrate an in vivo co-localization of p60 v-src with p125 FAK , PKC␣, vinculin, and actin fiber ends in focal adhesions (12)(13)(14)(15).
We previously characterized a novel gene, 322 (1), isolated in a screen for candidate tumor suppressor or regulatory genes (16). 322 transcription is sustained in confluent, non-dividing fibroblasts, but is suppressed in response to serum growth factors or activation of a Ts-v-src allele. Furthermore, the constitutive overexpression of 322 via retroviral vectors or by stable transfection is toxic, resulting in the selection of variants deleted of the transduced 322 cDNA. These data suggested that 322 may encode a regulator of mitogenesis.
In this paper, we have characterized the protein product of the 322 gene, determined its cellular localization in rat fibro-blasts, and defined it as a PKC substrate in vitro and in vivo. Based on these characteristics, we have named the 322 product SSeCKS for Src Suppressed C Kinase Substrate. Our data suggest a role for SSeCKS in the formation of the actin-based cytoskeleton.

EXPERIMENTAL PROCEDURES
Plasmids-A full-length 322 cDNA was constructed by splicing a 1.2-kb XhoI/BstEII fragment from a 5Ј-rapid amplification of cDNA end clone, p53ext2 (kindly provided by Sue Jaken, W. Alton Jones Cell Science Center, Lake Placid, NY), into a BstEII-partial/XhoI fragment of the clone 13.2.2 322 cDNA (1). The resulting "SSeCKS" cDNA was sequenced using Sequenase 2.0 kits (U. S. Biochemical Corp.) and the data submitted to GenBank in updated form (accession number U23146). GST fusion constructs were produced using pGEX-5x-1 (Pharmacia) and His-tag constructs were produced using pET28 (Novagen). Retroviral constructs of the SSeCKS cDNA were produced in pBABEhygro and packaged in ⍀ e cells as described previously (17). The SSeCKS cDNA was also spliced into pCEV27 (18) containing the Moloney leukemia virus-long terminal repeat promoter, and stably transfected into Rat-6 cells (19) followed by selection in 400 g/ml Geneticin (Life Sciences) as described previously (17).
Production of Immune Sera-After roughly 10 ml of preimmune sera was obtained, two New Zealand giant rabbits were immunized with 150 g each of GST-1322 protein emulsified in an equal volume of complete Freund's adjuvant (Life Technologies). The rabbits were boosted 2-3 times more with 50 -100 g/injection of GST-1322 in incomplete Freund's adjuvant. The specificity of the sera was determined by probing slot blots containing GST protein alone, GST-1322, His 6 -1322, and BL21 lysate alone, followed by incubation with alkaline phosphataselabeled sheep anti-rabbit Ig (Boehringer Mannheim), washing in Western blot buffer (below), and developing with 5-bromo-4-chloro-3-indolyl-1-phosphate/nitro blue tetrazolium (Promega). Both rabbits gave high titers (Ͼ5000) of anti-SSeCKS antibodies. Immunoaffinity-purified anti-SSeCKS antibodies were isolated as follows. Glutathione-Sepharose columns were saturated with either GST or GST-1322, washed, and then treated with 25.5 mM dimethyl pimelimidate cross-linker (Pierce). 10 ml of Rb anti-SSeCKS sera was passed repeatedly over the GST column (the bound antibodies were eluted with glutathione after each round) until all the anti-GST reactivity (as determined by slot blot Western analysis) was removed. The resulting sera was passed over the GST-1332 column, and the bound antibodies were eluted with Tris glycine buffer, pH 2.8, as described previously (20). This fraction was shown by slot blot Western analysis to retain GST-1332 and His 6 -1322 binding at dilutions of 1:1000, and no cross-reactivity to GST protein alone at dilutions of 1:100.
In Vitro Transcription/Translation of SSeCKS-Plasmid DNAs (1 g) containing either the full-length SSeCKS cDNA or the 5.4-kb 13.2.2 cDNA (1) cloned into pBluescript SK II were linearized at the 3Ј-ends of the cDNA inserts (SmaI) and incubated at 30°C for 90 min in a 50 l of coupled transcription/translation reaction (TNT, Promega) containing 50 Ci of translation-grade [ 35 S]methionine (DuPont NEN), according to the manufacturer's specification. 5 l of the resulting protein products were electrophoresed on a 6% SDS-polyacrylamide stacking gel (above). The gels were fixed in methanol/acetic acid (15/7%, respectively), incubated in Amplify (Amersham), and fluorographed with Kodak X-AR film at Ϫ70°C.
In Vivo Phosphorylation Analysis-10 6 Rat-6 cells were incubated overnight in Dulbecco's modified Eagle's medium (Bio-Whittaker) supplemented with 0.5% calf serum (Life Technologies) and then twice for 1 h in Dulbecco's modified Eagle's medium without sodium phosphate (Life Technologies). Labeling was for 2 h in modified essential medium without phosphate supplemented with 150 Ci of [ 32 P]orthophosphate (DuPont NEN). In some cases, phorbol 12-myristate 13-acetate (PMA, 200 nM) was added for various durations at the end of this labeling period. The PKC-specific inhibitor, bis-indolylmaleimide (Boehringer Mannheim; 10 M), was added at the beginning of the labeling period and again when PMA was added. After washing the cells three times with ice-cold phosphate-buffered saline, the cells were lysed in 0.5 ml of RIPA, 150 mM NaCl (17) and analyzed by SDS-PAGE as described above.
Western (Immuno-) Blot Analysis-Cells were washed three times in ice-cold phosphate-buffered saline, lysed in 1 ml/10-cm plate with RIPA buffer containing 150 mM NaCl, vortexed, incubated on ice for 10 min, and then centrifuged at 13,000 ϫ g for 30 min at 4°C to remove debris. 50 -400 g of cell lysate was electrophoresed through 6% SDS-polyacrylamide stacking gels, and then electrophoretically transferred to Immobilon-P. A rapid immunodetection method was followed 2 in which dried blots were not re-wetted, and then processed as described previously (17) using phosphate-buffered saline containing 1% non-fat dry milk (Difco) and 0.05% Tween 20 (Sigma) as the buffer. Alkaline phosphatase-labeled secondary antibodies were either sheep anti-rabbit Ig or sheep anti-mouse Ig (Boehringer Mannheim), and the substrate was room temperature-stabilized 5-bromo-4-chloro-3-indolyl-1-phosphate/ nitro blue tetrazolium solution (Promega). Protein concentrations were determined using a Micro BCA Protein Assay kit (Pierce).
Co-precipitation (Pull-down) Assay-1 mg of lysate from Rat-6 or Rat-6/PKC␣ overexpressor cells (gift of I. B. Weinstein, Columbia University) (22), or 20 ng of purified rabbit brain PKC (Upstate Biologicals, Inc.) were co-incubated with 135 l of glutathione-Sepharose pre-bound to 50 g of GST-1322 for 4 h at 4°C (rotating) in RIPA buffer containing 150 mM NaCl, 5 mM MgCl 2 , and 0.2 mM CaCl 2 . PS was added in some cases at 0.37 mg/ml. The pellets were washed three times and then analyzed by SDS-PAGE and immunoblotting as described above using mouse monoclonal anti-PKC type III (Upstate Biologicals, Inc.).

Subcellular Fractionation of Plasma Membrane and Cytosol
Components-10 6 Rat-6 or Rat-6/PKC␣ over-expressor cells were washed three times in ice-cold Tris-Glu buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM KCl, 1 mM sodium phosphate, 0.1% glucose). The cells were scraped into Tris-Glu, and pelleted by centrifugation at 1,500 ϫ g for 5 min. The cells were swollen on ice for 10 min in 20 mM Tris-HCl, pH 7.4, 10 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 1% Trasylol, and 1 mM phenylmethylsulfonyl fluoride. The cells were Dounce homogenized (40 strokes with pestle B), and then NaCl was added to a final concentration of 100 mM. The nuclei and cell debris were pelleted at 1,500 ϫ g for 10 min (4°C) yielding initial pellet (P1) and supernatant (S1) fractions. The S1 fraction was loaded into polycarbonate tubes and centrifuged in a SW41 rotor (Beckman) for 30 min at 100,000 ϫ g, yielding a secondary pellet (P100), containing plasma membranes, and supernatant (S100). Aliquots of these fractions were analyzed by SDS-PAGE and immunoblotting as above.
Immunofluorescence Analysis-Rat-6 cells were seeded onto sterile 22-mm 2 coverslips at a density of roughly 70% and then incubated overnight or until the cells were confluent for at least 2 days. The coverslips were fixed as described previously (23). After washing in phosphate-buffered saline, the cells were incubated for 1 h with immunoaffinity-purified rabbit polyclonal anti-SSeCKS (above; 1:50 dilution) and rhodamine-labeled phalloidin (1:400; Sigma). Secondary antibodies to detect SSeCKS were fluorescein-labeled anti-rabbit Ig (Boehringer Mannheim). The coverslips were mounted in Aqua-Mount (Lerner Laboratories, Pittsburgh, PA) containing 20 mM p-phenylenediamine (Kodak) as an anti-bleaching agent.
Reagents-All reagents were purchased from Sigma unless indicated otherwise.

RESULTS
SSeCKS Is Identical to a Previously Described "Ͼ200-kDa" PKC Substrate-Recent work from the Jaken laboratory identified several novel substrates of PKC using overlay assays (24). Preliminary sequencing of the Ͼ200-kDa gene showed Ͼ99% homology to our 322 gene sequence, which encodes SSeCKS (GenBank accession U23146). 3 Most importantly, rabbit sera raised against GST/SSeCKS (GST/13.2.2., see Fig. 2) specifically recognizes a 280/290-kDa doublet in Rat-6 cells ( Fig. 1A), comparable in size to the protein recognized by sera raised against the Ͼ200-kDa gene product (25).
A 5Ј-rapid amplification of cDNA end clone product of the Ͼ200-kDa cDNA, p53ext2, provided by Dr. Jaken was spliced to our 5.4-kb cDNA (clone 13.2.2; Ref. 1) in order to construct a full-length 322 cDNA (6.0 kb) as described under "Experimental Procedures." Fig. 1B shows that both the upstream and internal ATG sites are independently recognized in T7 transcription/translation system (TNT, Promega), although the internal site is silenced in the context of the upstream site. Most importantly, the product obtained in vitro from the upstream ATG has a similar mobility in SDS-PAGE to native SSeCKS (ϳ280/290 kDa) in Fig. 1A.
We previously showed that the constitutive expression of the 13.2.2 cDNA (encoding a.a. 387-1594) led to the selection of cells containing deletions of the transduced SSeCKS cDNA copies (1). Thus, we determined whether over-expression of the full-length SSeCKS cDNA, via a retroviral vector (pBABEhygro) or a vector driven by the Moloney leukemia virus-long terminal repeat promoter (pCEV27), was toxic in rodent fibroblasts. As before, the presence of the SSeCKS cDNA severely decreased (Ͼ95%) the frequency of hygromycin-resistant colony formation following either infection with packaged virus or direct DNA transfection of both untransformed and src-trans-formed cells (data not shown). The few colonies that did arise showed highly unstable growth characteristics such as a Ͼ80% decrease in cell viability after trypsinization. These data indicate that high level constitutive expression of the full-length form of SSeCKS is growth inhibitory.
Expression of the 322 ORF Products and Production of Specific Antisera-The predicted structure of SSeCKS using the Chou-Fasman (CF) algorithm (26) was that of an elongated, rod-shaped protein with a concentration of both CF turns and predicted antigenic sites (Jameson-Wolf index; Ref. 27) roughly one-third of the way into the coding sequence (Fig. 2). Fig. 4 shows the inducible expression and purification of a GST-322 fusion protein with an apparent mobility on SDS-PAGE of 160 kDa even though the predicted molecular mass is 81 kDa (including GST). This result is consistent with our previous description of a severely retarded mobility for the 322 ORF product (1), and most likely results from a high concentration of 3 S. Jaken, personal communication.  Fig. 11) and a minor 240-kDa SSeCKS species were recognized. Antibody binding was completely blocked by the addition of GST/13.2.2 (20 g) but not by GST alone (20 g). B, in vitro transcription and translation of SSeCKS. One g of plasmid DNA encoding the full-length SSeCKS cDNA or a N terminally truncated SSeCKS cDNA (clone 13.2.2) were incubated in a coupled T7 transcription/translation reaction (TNT, Promega) containing [ 35 S]methionine as described under "Experimental Procedures." One-tenth of the labeled products were analyzed by SDS-PAGE followed by fluorography. Protein size markers are shown at the left. Note that a shortened version of SSeCKS, synthesized from an internal ATG start site in clone 13.2.2, is not produced in the context of the upstream ATG start site in the full-length SSeCKS cDNA. C, in vivo myristylation of SSeCKS. Lysates from 5 ϫ 10 6 Rat-6 cells incubated overnight in media containing 100 mCi/ml [ 3 H]myristate were analyzed by immunoprecipitation using preimmune or anti-SSeCKS sera, SDS-PAGE, and fluorography (5-month exposure). acidic residues as well as an inherent rod-like structure. The smaller polypeptides purified on glutathione columns are considered C-terminal breakdown products inasmuch as Western blotting using anti-GST sera identified the same band pattern as Coomassie Blue staining (data not shown).
Figs. 1, 3, 9, and 11 show that immune rabbit serum recognizes a 280/290-kDa doublet and a minor 240-kDa SSeCKS form in immunoblots (in Figs. 1 and 3, the 280/290-kDa doublet is unresolved). The 280/290-kDa doublet form is consistent with the in vitro product generated using the TNT system (Fig.  1B) and the bacterially expressed GST-SSeCKS products (Figs. 4 and 6), further suggesting that post-translational modifications are only minor contributors to the molecular masses of the mature SSeCKS products in Rat-6 cells. It is unclear whether the 240-kDa species represents an in vivo utilization of the internal ATG start site described above or a specific proteolytic cleavage product. Interestingly, the 240-kDa species is not readily labeled in vivo with [ 32 P]orthophosphate (Fig. 3, top panel), although predicted casein kinase II and PKC sites are found throughout the SSeCKS ORF. Our previous data (1) showed a single mRNA hybridizing to 322 cDNA, and thus, the multiple forms of SSeCKS are most likely due to protein modifications rather than to multiple SSeCKS allelic products.
SSeCKS contains a predicted N-terminal myristylation signal, MGAGSSTEQR, which conserves the Gly-2 and Ser-6 motifs encoded by retroviral GAGs and the HIV nef product (Ref. 4). 4 Fig. 1C shows that the 280/290-kDa SSeCKS form could be labeled in vivo with [ 3 H]myristate. We assume this post-translational modification facilitates the association of SSeCKS with plasma membrane fractions shown in Fig. 11.
We found it difficult to metabolically label SSeCKS in either subconfluent or confluent cultures using either [ 35 S]methionine/cysteine or [ 3 H]leucine, although p60 c-src was easily labeled in the same lysates (data not shown). This could not be due to a dearth of Met, Cys, or Leu residues in SSeCKS (20, 15, and 86, respectively). In contrast, SSeCKS could be immunoblotted easily under the same conditions, suggesting that its relative rate of de novo synthesis is low. We showed previously that SSeCKS is not glycosylated in an in vitro mammalian translation system (1). The addition of tunicamycin to Rat-6 cells did not alter the electrophoretic mobility of SSeCKS as determined by [ 35 S]methionine/cysteine labeling or Western blotting (data not shown), indicating that SSeCKS is not significantly glycosylated in vivo.
SSeCKS as a PKC Substrate-Activation of PKC by the short-term addition of nanamolar concentrations of phorbol esters is known to result in the rapid phosphorylation of PKC substrates such as MARCKS (28). that this induction is abrogated by the addition of the PKCspecific bis-indolylmaleimide inhibitor, GF-109203X. The PMA-induced phosphorylation effect is apparent in as little as 2 min and lasts at least 10 min (Fig. 3), but wanes with treatments of longer than 60 min (data not shown) although this may reflect down-regulation of PKC (29). This rapid PKCinduced phosphorylation is similar to that of MARCKS (28,30), which is often used as a gauge of PKC activation. Although SSeCKS from quiescent Rat-6 fibroblasts contains no detectable phosphotyrosine, as determined by anti-phosphotyrosine immunoblotting (data not shown), we cannot rule out that it is tyrosine phosphorylated following PMA treatment or activation of Ts-v-src.
We then determined that purified rabbit brain PKC containing ␣, ␤, and ␥ isoforms could phosphorylate GST-322 protein in vitro (Fig. 4, panel B). This phosphorylation is inhibited by the addition of excess PKC pseudosubstrate peptide (a.a. 19 -36) indicating that PKC and not a contaminating kinase is responsible for the phosphorylation. The PKC-specific phospho-rylation of GST-322 paralleled that of myelin basic protein in its dependence on PS (Fig. 5) and Ca 2ϩ (data not shown). PI could supplant PS in this assay although the relative level of GST-322 phosphorylation was roughly 2-fold less than that of myelin basic protein using PI (Fig. 5). As previously reported using myelin basic protein, PC did not stimulate PKC activity of GST-322 (29). Although several C-terminal breakdown products smaller than 70 kDa are present in the preparation of GST-322 (Fig. 4), only the products of Ն70 kDa are phosphorylated in vitro, suggesting that the PKC sites do not map to the extreme C-terminal portion of this construct. Therefore, these data indicate that SSeCKS is both an in vivo and in vitro substrate of PKC.
SSeCKS Binding to PKC-The ability of PKC to phosphorylate GST-322 indicates some level of interaction between these proteins, yet the conditions required for binding are unclear. Thus, we characterized the ability of PKC to bind GST-1322 in vitro. Our results (Fig. 6) indicate that SSeCKS binds both purified PKC and PKC in Rat-6 lysates in a PS-dependent manner. Thus, SSeCKS and PKC most likely interact via a PS bridge, although it cannot be ruled out that there is a lower affinity protein-protein interaction with domains in SSeCKS not encoded by GST-1322. Pre-phosphorylation of GST-1322 by  Fig. 2) were expressed and purified from bacteria as described under "Experimental Procedures" (panel A, Coomassie-stained SDS-polyacrylamide gel). Five g of the GST samples were added to PKC assays containing [␥-32 P]ATP ("Experimental Procedures") in the presence or absence of the PKC peptide inhibitor (19 -36). The products were then bound to glutathione-Sepharose beads, precipitated, and washed, and analyzed by SDS-PAGE and autoradiography (panel B). Protein size markers are indicated on the appropriate sides. Radioactive labeling was detected in GST-322 (160 kDa and breakdown products) only.  Fig. 4, in the presence or absence of various lipids including PS, phosphatidylcholine (PC), or phosphatidylinositol (PI). In some cases, excess PKC peptide inhibitor (19 -36) was added as in Fig. 4. The extent of labeling in the peptide substrates was determined by spotting the reaction products on phosphocellulose discs (Whatman), precipitating peptides with washes of 5% trichloroacetic acid, followed by scintillation counting. PKC decreases this binding at least 10-fold (data not shown), suggesting that phosphorylated SSeCKS has decreased binding affinity for PS.

Identification of in Vitro PKC Phosphorylation Sites on SSeCKS-The consensus motifs for PKC phosphorylation have been identified as (S/T)X(K/R) or (K/R)XX(S/T), with a greater
preference for serine over threonine (31). However, our observations of previously characterized in vivo PKC phosphorylation sites indicates that they typically contain a high concentration of basic residues and at least 2 or 3 of the overlapping phosphorylation motifs described above. Analysis of the SSeCKS sequence yielded 4 such putative phosphorylation sites shown in Table I. These sites share some linear sequence homology and predicted secondary structural similarity with the PKC phosphorylation site in MARCKS. A minimal MARCKS 23-peptide containing this site (32) also binds calmodulin and F-actin (Table I).
In order to determine whether the SSeCKS sequences described in Table I could be phosphorylated by PKC in vitro, polymerase chain reaction products containing these individual sites or several sites in tandem were generated, fused in-frame to GST-expressing vectors (Fig. 2), and checked by sequencing. Fig. 7 indicates that sites 1-4 could be phosphorylated efficiently by purified rabbit brain PKC and that this phosphorylation was blocked by excess pseudosubstrate peptide inhibitor. It is yet unclear whether these sites are utilized during in vivo phosphorylation.
SSeCKS Is Resistant to Heat Denaturation-Besides having predicted rod-like structures, many PKC substrates share peculiar characteristics such as resistance to heat denaturation (31). Fig. 8 shows that the 280/290-kDa form of SSeCKS re-mained soluble after 5 min of boiling in RIPA buffer containing 0.05% SDS. Additionally, boiled SSeCKS retained roughly 50% of its immunoreactivity with rabbit immune serum. These data strengthen the notion that SSeCKS assumes a rod-like structure in vivo. In contrast, GST fusions of SSeCKS are heat-labile PKC substrates, a characteristic most likely conferred by the GST moiety (not shown).
SSeCKS in src-and ras-transformed Cells-The SSeCKS encoding gene, 322, was originally isolated based on its being transcriptionally suppressed in src-transformed NIH3T3 cells (16). We showed subsequently that 322 is down-regulated at least 10-fold at the steady-state RNA level in src-and rastransformed Rat-6 fibroblasts but not in cells transformed by activated raf (1). Fig. 9 shows that the relative level of SSeCKS in src-and ras-transformed Rat-6 fibroblasts is 15-and 8-fold lower, respectively, than in untransformed cells. An additional 305-kDa SSeCKS form found in the ras-and src-transformed cells might represent a modified form of the 280/290-kDa SSeCKS doublet. These data suggest that the relative abundance of SSeCKS in transformed cells is controlled at the transcriptional level.
Cell Localization of SSeCKS-We determined where SSeCKS is found in subconfluent and confluent Rat-6 cells. Immunofluorescence analysis using immunoaffinity-purified ␣-SSeCKS antibody indicates that SSeCKS localizes to the cytoplasm but is enriched at the cell edge, in structures resembling podosomes, and in the perinucleus (Fig. 10). An apparent intranuclear staining of SSeCKS (Fig. 10J) most likely represents deposits of SSeCKS in or on the nuclear cage based on confocal microscopy (data not shown). The association of SSeCKS with cortical actin-like structures (Fig. 10, A, C, and E) and cellular components such as podosomes (Fig. 10I) further suggests a role for SSeCKS in the control of actin-based cytoskeletal architecture.
Previous data indicated that short-term treatment (Յ10 min) of quiescent fibroblasts with PMA led to a rapid detachment of MARCKS from plasma membrane sites into a soluble cytoplasmic compartment, followed by its re-association with membrane structures and progressive movement toward the perinucleus (30). This effect was coincident with a ruffling of actin fibers at the plasma membrane. We determined the effect of PMA on SSeCKS localization. Fig. 10 (A-F) shows that after 10 min PMA treatment, membrane ruffling of actin was apparent in rat fibroblasts, but SSeCKS was still uniformly associated with cortical actin-like structures throughout the cytoplasm. With longer PMA treatment (60 min), SSeCKS localized predominantly to the perinucleus. This delayed movement of SSeCKS toward the perinucleus, when compared with MARCKS, suggests that it is a consequence of exocytosis, which is known to be induced by short-term PMA treatment FIG. 6. Co-precipitation of SSeCKS with PKC. Panel A, GST-1322 fusion protein (see Fig. 2) was expressed and purified from bacteria as described under "Experimental Procedures" (Coomassie-stained SDS-polyacrylamide gel). Panel B, RIPA lysates (1 mg of protein/sample) from Rat-6 or Rat-6/PKC-␣ overexpressor cells, or purified rabbit brain PKC (20 ng; "P-PKC-␣") were incubated for 4 h at 4°C with 50 g of GST-1322 (or GST alone) in the presence or absence of 0.37 mg/ml PS. The proteins were then precipitated with glutathione-Sepharose beads, washed, and Western blotted as described under "Experimental Procedures." The filters were probed with monoclonal antibody specific for PKC type III (UBI). The lane to the right is loaded with 20 ng of purified rabbit brain PKC-␣. IPTG, isopropyl-1-thio-␤-D-galactopyranoside.  (33). We then determined whether PMA treatment causes an initial solubilization of SSeCKS. Fig. 11 shows that the relative level of SSeCKS associated with either membrane or soluble subcellular compartments does not change significantly in Rat-6 cells after 30 min of PMA treatment. We cannot rule out that PMA induces the solubilization of a minor, membraneassociated component of SSeCKS. These data suggest that most of SSeCKS, unlike MARCKS, remains tethered to cytoskeletal structures during movement toward the perinucleus.

DISCUSSION
Our previous preliminary data implicated the 322 gene in the control of mitogenesis or proliferation (1). In this study, we characterize the full-length 322 gene product and demonstrate that it is a novel substrate of PKC.
We previously showed that the over-expression of an Nterminal truncated form of SSeCKS (missing roughly 400 a.a.) was growth inhibitory in untransformed and src-transformed rodent fibroblasts (1). Here we corroborate these findings using the full-length SSeCKS DNA. As before, the few SSeCKStransfected colonies which did proliferate contained heterogeneous deletions of the SSeCKS cDNA. We have recently produced cell lines with inducible expression of SSeCKS in which the background level of expression (repressed state) is equal to that of endogenous SSeCKS. 5 When induced, most lines express only 5-10-fold SSeCKS over background whereas control cells routinely express Ͼ100-fold luciferase reporter over back-ground, suggesting that there is selective pressure against expressing high levels of SSeCKS.
All three SSeCKS forms (240 and the 280/290 kDa doublet) contain multiple potential serine or threonine phosphorylation sites, yet only the doublet species is phosphorylated significantly in vivo. Thus, it is unlikely that the 240-kDa SSeCKS species is a breakdown product or proteolytically-matured form of the protein. As our data indicate that PKC can phosphorylate SSeCKS in vitro and in vivo, it is possible that the 240-kDa form may localize to cellular compartments unavailable to PKCs or other serine/threonine kinases. Indeed, this species is relatively enriched in soluble components (S100) of rat fibroblasts. Our ability to show a relationship between the three SSeCKS protein species using trypic maps has been confounded by a difficulty in metabolically labeling these proteins in vivo.
SSeCKS has a predicted rod-like structure, with a set of turns in the molecule's center flanked by N-and C-terminal tubular domains (Fig. 2). This conformation, as well as the 250 g of total protein from Rat-6, Rat-6/src, and Rat-6/ras (1) cell lysates was analyzed by immunoblotting for SSeCKS as described under "Experimental Procedures." In addition to the 240-kDa (larger arrow) and 280/190-kDa (small arrow) forms of SSeCKS found in untransformed cells, a 305-kDa form was detected in Rat-6/ras cells and to a lesser extent in Rat-6/src cells. The relative level of SSeCKS in srcand ras-transformed cells compared with Rat-6 cells (as defined by densitometry) is 15-and 8-fold lower, respectively. concentration of acidic residues in the N-terminal third of the protein, are most likely responsible for the retarded mobility of SSeCKS in SDS-PAGE. Indeed, bacterially-expressed GST-or His-tag fusions of SSeCKS show similarly retarded mobilities indicating that post-translational modifications do not contribute significantly to the mature form of SSeCKS in mammalian cells. We have also demonstrated that SSeCKS produced in in vitro and in vivo mammalian systems is not glycosylated significantly (1) (this study).
SSeCKS can be phosphorylated by PKC in vivo in response to short-term treatment by phorbol esters (Յ2 min). This phosphorylation is abrogated in the presence of the PKC-specific inhibitor, bis-indolylmaleimide. It is unclear which isoform of PKC phosphorylates SSeCKS in vivo. The majority of PKC in Rat-6 cells is ⑀ and ␦, although ␣ is fairly abundant. 6 Additionally, the specificity of the PKC inhibitor we use (bis-indolylmaleimide) has been established for the so-called Group A PKC isoforms (␣, ␤I, ␤II, and ␥) but has not been tested rigorously for other isoform groups. The ability of the PKC fraction used for our in vitro assays (enriched for ␣, ␤, and ␥) to phosphorylate specific SSeCKS polypeptides is likewise not direct proof that these isoforms are responsible for the in vivo phosphorylation activity.
The SSeCKS coding sequence contains 4 domains of overlapping PKC phosphorylation motifs ((S/T)X(K/R) or (K/R)XX(S/ T)) representing potential phosphorylation sites. Each of these sites (SSeCKS 1-4) can be phosphorylated in vitro by purified rabbit brain PKC in a PS-and Ca 2ϩ -dependent manner. The in vitro phosphorylation of SSeCKS could also be supported by PI but not by PC, confirming previous data on the phospholipid cofactor requirements of PKC␣ (29). Moreover, the binding of SSeCKS to PKC␣ in vitro is PS-dependent which agrees with the PS-dependent binding of PKC by the Ͼ200-kDa protein (24). We are currently determining whether SSeCKS 1-4 are phosphorylated in vivo by activated PKC, and whether the phosphorylation of one or more dominates over others.
The first two PKC phosphorylation sites in SSeCKS (SSeCKS-1 and -2, Table I) contain significant similarities with a 23-mer MARCKS peptide encoding a minimal PKC phosphorylation site as well as binding ability to calmodulin and Factin (32). These SSeCKS sites also are enriched for basic residues, as has been previously reported for other PKC sites (31). In contrast to SSeCKS-1 and -2, whose sequences are not that similar to each other, SSeCKS-3 and -4 share significant sequence and predicted structural homology, although they are less similar to the MARCKS 23-mer PKC site than SSeCKS-1 or -2. This suggests a coordinated or redundant control of the phosphorylation of SSeCKS-3 and -4 in vivo.
We searched the SWISSPROT databank for similarities to the putative serine-phosphorylation sites in SSeCKS 1-4 (Table I), with the requirement that potential phosphoserine residues be retained. No significant similarities to SSeCKS-1 were found. However, the SSeCKS-2 putative PKC site showed 50% identity to a sequence in the retinoic acid receptor-␣ (SWISSPROT: Rra1_Mouse) and the SSeCKS-3/4 consensus peptide showed 46.2% identity to the protein kinase A anchor protein, AKAP-79 (SWISSPROT: Ak79_Human). It is unknown whether these other proteins are phosphorylated by PKC at these sites. However, these similarities to SSeCKS strengthen the notion of a function for SSeCKS at the plasma membrane.
Analysis of the in vitro SSeCKS phosphorylation sites using the HELICALWHEEL program (34) predicts amphipathic helical structures for SSeCKS-1, -2, and -4 but less so for SSeCKS-3. It is difficult to predict whether there is any interplay between these phosphorylations sites as they are separated by between 60 and 100 residues on a proposed rod-shaped 6 R. Krauss (Mount Sinai School of Medicine), personal communication.
FIG. 11. SSeCKS does not enter a soluble cellular component after short-term PMA treatment. Confluent Rat-6 or Rat-6/PKC-␣ overexpressor cell cultures were grown overnight in Dulbecco's modified Eagle's medium lacking calf serum, and the treated with PMA (1.6 M) for 30 min or mock-treated for the same duration. The cells were lysed and spun at low speed (1,500 rpm), yielding a pellet (P1) and supernatant (S1). The S1 component was fractionated by differential centrifugation into plasma membrane (P100) and soluble (S100) subcellular components as described under "Experimental Procedures." 50-g aliquots of P1 and P100, and 25-g aliquots of S100 were then immunoblotted using rabbit anti-SSeCKS as described in the legend to Fig. 3. The SSeCKS isomers (240, 280, and 290 kDa) are shown in relation to a 220-kDa marker protein (myosin heavy chain). molecule. In the case of MARCKS, McLaughlin and Aderem (35) postulate that MARCKS probably associates with plasma membranes via its N-terminal myristyl group and its concentration of positively charged amino acid residues in the PKC phosphorylation site. PKC phosphorylates three serines in this site that align along one axis of a short amphipathic helix. They further postulate that the resulting confluence of electrostatic phosphoserine charges causes MARCKS to detach from plasma membranes sites. Indeed, SSeCKS is enriched at the cell edge and in podosomes (Fig. 10), as was previously demonstrated for MARCKS (28). However, following the activation of PKC, SSeCKS did not detach appreciably from membrane sites or from subcellular fractions enriched for plasma membranes (Fig. 10, A-F, and Fig. 11). This suggests that the phosphoserine charges in SSeCKS are insufficient to counteract its affinity for membranes. It cannot be ruled out that only a minor component of SSeCKS is membrane-associated in the cell and that phosphorylation by PKC induces this component to move to a soluble cytoplasmic compartment.
Although SSeCKS and MARCKS share little sequence similarity past their PKC phosphorylation sites, they share several biochemical and structural characteristics common to other PKC substrates implicated in the regulation of cytoskeletal architecture such as igloo, GAP-43, and neurogranin. These include: (i) a predicted elongated or rod structure, (ii) enrichment for alanine, serine, lysine, and glutamic acid residues, (iii) binding to plasma membranes (GAP-43, for example, is palmitoylated), (iv) association with focal contact sites or cellular processes, (v) predicted or proven phospholipid binding activity, and (vi) predicted or proven calmodulin and F-actin binding domains (29,36,37). Additionally, the over-expression of SSeCKS or MARCKS is growth inhibitory (Ref. 1; this study). 7 This correlates with the increase in SSeCKS and MARCKS expression as cells enter G 0 (1,38). These data suggest that SSeCKS and MARCKS share some overlapping functions and regulatory motifs. However, unlike MARCKS, which is expressed throughout mammalian tissues, and SSeCKS, which is primarily expressed in the brain, genitourinary tract, intestines, and kidney (1), GAP-43, igloo, and neurogranin are brain-specific (29,36,37). Additionally, GAP-43, igloo, and neurogranin, but not MARCKS and SSeCKS, encode PKC phosphorylation sites with the so-called "IQ" motif, KIQAS-FRGH (39).
SSeCKS localizes to focal contact sites (Fig. 10, H and I) known to be enriched for PKC␣, p125 FAK , and actin-binding proteins. These structures mediate the interaction of cytoplasmic actin fibers with extracellular matrices via integrins (12,15). Indeed, SSeCKS seems to associate with a cortical actinlike cytoskeletal matrix in confluent Rat-6 cultures (Fig. 10, A, C, and E). Our preliminary data show binding of SSeCKS to F-actin in vitro. 8 Thus, SSeCKS may be involved in the regulation of actin-based cytoskeletal architecture.
Recent data indicate that actin fiber formation is controlled by rac-and rho-mediated pathways distinct from the raf/ MAP kinase-mediated pathways controlling proliferation (40). SSeCKS transcription is suppressed in src-and ras-but not raf-transformed cells (1). Thus, the raf-independent control of SSeCKS expression parallels the rac-and rho-dependent control of actin-based cytoskeletal architecture. Since actin-based structures regulate cell morphology, motility, metastasis, and cell-to-cell interactions, we are interested in elucidating the role of SSeCKS in controlling these processes in both untransformed and transformed cells.