Constitutive Association of c-N-Ras with c-Raf-1 and Protein Kinase C e in Latent Signaling Modules*

Phorbol ester stimulation of the MAPK cascade is be-lievedtobemediatedthroughtheproteinkinaseC(PKC)-dependentactivationofRaf-1.AlthoughseveralstudiessuggestthatphorbolesterstimulationofMAPKisinsen-sitivetodominant-negativeRas,arequirementforRasinRaf-1activationbyPKChasbeensuggestedrecently.Wenowdemonstratethatinnormal,quiescentmousefibroblasts,endogenousc-N-Rasisconstitutivelyassoci-atedwithbothc-Raf-1andPKC e in a biochemically silent, but latent, signaling module. Chemical inhibition of novel PKCs blocks phorbol 12-myristate 13-acetate (PMA)-mediated activation of MAPKs. Down-regulation of PKC e protein levels by antisense oligodeoxyribo-nucleotides blocks MAPK activation in response to PMA stimulation, demonstrating that PKC e activity is required for MAPK activation by PMA. c-Raf-1 activity in immunoprecipitated c-N-Ras z c-Raf-1 z PKC e complexes is stimulated by PMA and is inhibited by GF109203X, thereby linking c-Raf-1 activation in this complex to PKC activation. These observations suggest that in quiescent cells Ras is organized into ordered, inactive signaling modules. Furthermore, the regulation of the MAPK cascade by both Ras and PKC is intimately linked, converging prewashed with P21 buffer containing 1 m M ATP, and washed extensively with the same buffer. [ 32 P]Phosphate incorporation was determined by scintillation counting. N-Ras-Raf-1 Destabilization Conditions— CHAPS-solubilized rat brain extracts (30) were used as a source of N-Ras. These lysates were exchanged with 100 m M GTP in the presence of 25 m M EDTA for 1 h at 4 °C. Excess Mg 2 1 was added to stabilize the N-Ras-GTP. Glutathione- agarose precoupled (and washed) with GST fused to the Ras-binding domain of Raf-1 (GST-RBD) was added to the lysates. Following a 1-h incubation at 4 °C the beads were washed and then incubated in the indicated concentrations of buffered NaCl (1 h at 4 °C). The beads were again washed, and the amount of N-Ras associated with the GST-RBD was assessed by Western analysis. A similar experiment was performed in which the preformed N-Ras-RBD complexes were incubated (1 h at 4 °C) with buffer at the indicated pH values (ranging from 4 to 9) rather than increasing NaCl concentrations. The GST-RBD-agarose was washed, and the amount of N-Ras associated with the RBD was deter- mined by Western analysis. GTP Binding Assays— GTP binding assays were performed with [ a - 32 P]GTP in p21 buffer supplemented with 25 m M EDTA and 100 m M GTP. These assays were done in two different ways. First we tested whether the exchange was inhibited at buffer pH values ranging from 4–9. This was done simply by altering the pH of the assay buffer and performing the filter binding assay as described previously (30). Second, we tested whether the Ras-GTP complex itself was stable at pH values ranging from 4 to 9. In this assay the GTP binding was performed at pH 7.4 in the standard binding buffer described above. Excess Mg 2 1 was added to stabilize the complex; the pH was altered, and excess unlabeled GTP was added to quench any labeled GTP released by the change in pH. Afte

The regulation of the Raf-1 serine/threonine kinase is a highly complex and poorly understood process (1,2). Raf-1 activation is a critical component of the proliferative response. Raf-1 activation regulates the MAPK 1 cascade, a linear kinase cascade comprising the MAPKs, ERK1, and -2, which are the physiological targets for MEK1, the substrate for Raf-1 (3)(4)(5). Raf-1 can be activated in response to both oncogenes and receptor tyrosine kinase activation. Raf-1 activity is regulated by phosphorylation on serine and tyrosine residues (6 -8), with both PKC and Src implicated as Raf-1 regulatory kinases. Raf-1 activity is also subject to negative regulation by PKAmediated phosphorylation of serine 621 in the kinase domain (9,10) and serine 43 in the Ras-binding domain (11), in response to elevations in intracellular cAMP levels. The direct interaction of Raf-1 with Ras (12,13), plasma membrane phospholipids (14,15), and molecular chaperone proteins of the hsp90 (16) and 14-3-3 families (17) also regulates Raf-1 activity possibly through contributions to Raf-1 dimerization and the masking of critical regulatory phosphorylation and phosphatidylserine-binding sites (reviewed in Ref. 2). Other proteindirected mechanisms of Raf-1 regulation have been identified recently. These appear to function through a poorly understood mechanism of membrane targeting and the organization of Raf-1 with its effectors. These include KSR (18 -20), which may function as a scaffolding protein in organizing the interaction of Raf-1 with its substrate MEK1, CNK (21), which targets Raf-1 to cell-cell contacts, and RKIP (22) which inhibits Raf-1 signaling by disrupting the interaction of Raf-1 with its substrate, MEK1.
The Raf kinases contribute to the complex process of cell proliferation through their regulation of the MAPK cascade. Persistent stimulation of this cascade by activated Raf protein can result in cellular transformation (reviewed in Ref. 23). Recent work by Mason et al. (8) demonstrated that the expression of oncogenic Ras in cells induced phosphorylation of Raf-1 at serine 338, a consensus site for PKC-mediated phosphorylation, whereas activated Src induced Raf-1 phosphorylation predominantly at tyrosine 341; Raf-1 phosphorylation was also dependent upon its association with the plasma membrane. Mitogen-induced Raf-1 activity required phosphorylation on both Ser-338 and Tyr-341, which cooperate to induce full Raf-1 activation. Under specific conditions, Tyr-341 phosphorylation appeared to direct Ser-338 phosphorylation, since Raf-1 harboring a Y341A substitution was not phosphorylated on Ser-338 in cells expressing activated Ras and Src. Although coexpression of activated Ras and Src induced Ser-338 phosphorylation that was dependent upon phosphorylation of Tyr-341, mitogen-induced (PMA or EGF) Ser-338 phosphorylation could occur in the absence of Tyr-341 phosphorylation. This observation suggested that two distinct Raf-1 kinases might phosphorylate Raf-1 on Ser-338 and contribute to its activation in vivo. Recently, in addition to PKC, PAK3, a kinase downstream of the Ras-related Rho GTPases, Cdc42 and Rac, was also identified as a putative Raf-1 kinase by its ability to phosphorylate Raf-1 on Ser-338 in vitro (24).
PKC activation by phorbol esters or diacylglycerol (DAG) acti-vates the MAPK cascade, presumably through stimulating Raf-1 activity (25,26). Several PKC family members phosphorylate and activate Raf-1 in vitro indicating that the actions of PKC may directly impinge upon the MAPK cascade. Ser-338 has been identified as a putative PKC phosphorylation site on Raf-1 (7,8). Both classical (cPKC) and novel (nPKC) PKCs have the potential to phosphorylate and activate Raf-1. Both cPKC and nPKC respond to DAG or phorbol ester and have been suggested as Raf-1 regulatory kinases (25)(26)(27)(28). There is little evidence, however, as to whether Raf-1 activity is directly regulated by distinct PKC isozymes. Schonwasser et al. (26) demonstrated that although cPKC␣, nPKC, and the atypical (DAG-and Ca 2ϩ -independent) PKC stimulate MEK1 and ERK2 activity, only PKC␣ andwere able to directly stimulate Raf-1 activity. However, a study by Ueda et al. (29) suggested that the expression of activated nPKC␦, but not activated cPKC␣ or nPKC⑀, activated the MAPK pathway in a c-Raf-1-dependent manner. Although these studies demonstrated some degree of PKC isotype specificity in Raf-1 regulation, the use of activated PKC constructs did not clearly define an unequivocal role for distinct PKC isotypes in Raf-1 regulation.
The interaction and regulation of Raf-1 by Ras (30,31) is the best characterized Ras-dependent signaling cascade. In vivo, activated Ras binds Raf-1, localizing the kinase to the plasma membrane (14) where it is subsequently phosphorylated and activated (6,8). The physical interaction of Ras with Raf-1 also provides a structural requirement for efficient Raf-1 activation. Tamada et al. (13) described a Ras activator region mutant, Ras(V45E), which, although it bound efficiently to Raf-1 and recruited the kinase to the plasma membrane, failed to stimulate efficiently Raf-1 activity. These data indicated that not only did Ras interact with the Ras-binding domain (RBD) of Raf-1 but that the interaction of Ras with the Raf-1 cysteinerich domain was also critical in Raf-1 activation.
Previous studies in this laboratory demonstrated a Ras isoform dependence in the regulation of c-Raf-1 activation in C3H10T1/2 mouse fibroblasts transformed by the minimal expression of oncogenic Ha-Ras. These studies demonstrated that N-Ras has a higher affinity for c-Raf-1 than did Ha-Ras in vivo (32). Cells transformed with oncogenic N-Ras had a persistently elevated MAPK activity that was resistant to the actions of growth factor and receptor tyrosine kinase antagonists, unlike those transformed by oncogenic Ha-Ras. This suggested that distinct aspects of an Ha-Ras-transformed cell phenotype, including persistent stimulation of MAPK activity, were the consequence of autocrine growth factor signaling (32,33) and not the physical association of c-Raf-1 with Ha-Ras. Indeed, contrary to expectations, oncogenic Ha-Ras was not associated with c-Raf-1. Oncogenic N-Ras, however, was sufficient in itself to stimulate the MAPK cascade through its stable association with c-Raf-1. Further analysis demonstrated that in Ha-Rastransformed cells, c-Raf-1 was associated with wild-type c-N-Ras and that antisense mediated down-regulation of c-N-Ras protein levels in these cells ablated their elevated MAPK activity and inhibited their serum-independent growth. These observations supported the concept that in this cell background, c-N-Ras was the Ras isoform that specifically regulated c-Raf-1 and the MAPK cascade.
We have now identified the existence of signaling complexes containing c-N-Ras, c-Raf-1, and PKC⑀ that are biochemically silent, insofar as possessing no significant Raf-1 activity, in quiescent, normal, non-transfected C3H10T1/2 mouse fibroblasts. In light of the recent observations of Marais et al. (34) who demonstrated that the interaction between Raf-1 and Ras was necessary for PKC to activate Raf-1, we have now extended those observation to show that these biochemically silent com-plexes of c-N-Ras, c-Raf-1, and PKC⑀ are latent and are directly activated by phorbol esters. We postulate that these latent complexes represent physiological signaling modules that are a target for the phorbol ester-mediated activation of c-Raf-1 through the activation of PKC⑀.

EXPERIMENTAL PROCEDURES
Materials-PVDF membrane, enhanced chemiluminescence (ECL) reagent, [␥-32 P]ATP (3000 Ci/mmol), and [␣-32 P]GTP were purchased from Amersham Pharmacia Biotech. PKC inhibitors GF109203X and Gö6976 and the cell-permeable Ca 2ϩ chelator, BAPTA-AM, were purchased from Calbiochem. Kinase-dead GST-MEK1(K97A) was purchased from StressGen Biotechnologies Corp. The phorbol ester, phorbol 12-myristate 13-acetate (PMA), Percoll, protein G-Sepharose, and the Sephacryl S200HR gel filtration media were purchased from Sigma. Protein A-Sepharose was purchased from Repligen. Mouse EGF was purchased from Austral Biologicals. Recombinant PKC⑀ and PKC␣ were purchased from Panvera Corp. and Upstate Biotechnology, Inc., respectively. The PKC⑀ and PKC␣ substrate peptides were purchased from BIOSOURCE International and Santa Cruz Biotechnology, respectively. PKC⑀ phosphorothioate-modified sense and antisense oligodeoxynucleotides (ODN) were synthesized and purified by high pressure liquid chromatography (Sigma). Tissue culture reagents were purchased from Mediatech. Fetal calf serum and newborn calf serum were purchased from Atlanta Biologicals. All other reagents were molecular biology grade.
Antibodies-Raf-1 and PKC isozyme-specific monoclonal antibodies were purchased from Transduction Laboratories. N-Ras and phospho-ERK monoclonal antibodies were purchased from Santa Cruz Biotechnology. The anti-MAPK rabbit polyclonal antibody was purchased from Upstate Biotechnology, Inc. The N-Ras-specific polyclonal antisera used for immunoprecipitation studies have been described previously (32). Horseradish peroxidase-conjugated anti-rabbit and anti-mouse secondary antibodies were purchased from Transduction Laboratories and Kirkegaard & Perry Laboratories, respectively. The phospho-Ser-338 Raf-1 rat monoclonal antibody was a kind gift from Richard Marais, Institute of Cancer Research, London, UK.
Western Blot Analysis-Immunoblot analysis was performed according to standard protocols. After resolution on SDS-PAGE gels, proteins were electroblotted to PVDF membranes and blocked using 10% (w/v) nonfat dry milk in TBS containing 0.1% (v/v) Tween 20 (TBST). Membranes were washed with TBST and incubated with the appropriate dilution of primary antibody in TBST containing 10% (v/v) newborn calf serum. Membranes were then extensively washed with TBST before being incubated with the appropriate horseradish peroxidase-conjugated secondary antibody. The membranes were again extensively washed, and the immunoreactive bands were detected by ECL.
Immunoprecipitations-Equal amounts of whole cell lysate or solubilized plasma membrane proteins were incubated with 50 l of anti-N-Ras sera or non-immune sera or with 2.5-5 g of purified antibodies at 4°C. After 2 h, the lysates were microcentrifuged at 14,000 rpm at 4°C, and the clarified lysates were transferred to clean tubes containing either protein A-Sepharose or protein G-Sepharose, in excess, and the antibody complexes were precipitated for an additional hour at 4°C. Immunoprecipitates were washed extensively with TBST, denatured by the addition of Laemmli buffer, and resolved on SDS-PAGE gels.
Preparation of Plasma Membranes-Plasma membranes were prepared using a modified protocol as described by Smart et al. (54). Briefly, cells were washed, scraped, and pelleted by low speed centrifugation in Tricine buffer (20 mM Tricine, 250 mM sucrose, 1 mM EDTA, pH 7.8). The cell pellet was resuspended in 0.75 ml of Tricine buffer and disrupted with 20 strokes in a 7-ml Dounce. The cell debris was pelleted by centrifugation at 5,000 rpm in a refrigerated microcentrifuge at 4°C for 10 min. The supernatant was removed and stored on ice. The cell debris was resuspended in 0.75 ml of Tricine buffer, homogenized, and recentrifuged as described. The supernatant was removed and combined with the first supernatant. The supernatant was then layered on a 30% (v/v) Percoll gradient prepared in Tricine buffer and centrifuged at 90,000 ϫ g for 30 min at 4°C. Following centrifugation through the 30% Percoll gradient, the distinct, plasma membrane band was collected and resuspended in Tris-buffered saline (TBS), pH 7.6, and c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester pelleted by centrifugation at 100,000 ϫ g for 1 h at 4°C. The plasma membrane was collected and washed twice with TBS followed by centrifugation at 14,000 rpm in a microcentrifuge at 4°C. The pelleted plasma membranes were solubilized in P21 buffer (20 mM MOPS, 200 mM sucrose, 5 mM MgCl 2 , 1 mM EDTA, pH 7.6) containing 1% (w/v) CHAPS (CHAPS buffer) plus protease and phosphatase inhibitors for 20 min on ice. The insoluble material was pelleted by centrifugation at 14,000 rpm in a microcentrifuge at 4°C. Enrichment of plasma membranes was monitored by Western analysis for the plasma membrane marker protein, the ouabain-sensitive Na ϩ /K ϩ ATPase versus the mitochondrial specific protein Porin.
Sephacryl S200HR Gel Filtration Chromatography-Plasma membranes were prepared from 72-h starved C3H10T1/2 cells as described. Equal amounts of plasma membrane protein were incubated with either 2 g of control (Con) antibody or with a PKC⑀-specific antibody (P14820; Transduction Laboratories) for 2 h at 4°C followed by high speed centrifugation to remove protein aggregates. The soluble protein was then subjected to gel filtration chromatography at 4°C through a 1-m Sephacryl S200HR column equilibrated in TBS, pH 7.6, containing 0.2% (w/v) CHAPS. Eluted protein was collected as 0.95-ml fractions for analysis by SDS-PAGE.
PKC Inhibitor Studies-All inhibitors were reconstituted in Me 2 SO. C3H10T1/2 cells were serum-starved for 72 h. Prior to mitogen stimulation, cells were either incubated overnight with 5 M PMA (chronic PMA) or for 1 h prior to mitogen stimulation with the cPKC and nPKC inhibitor, GF109203X (10 M), the cPKC inhibitor, Gö6976 (10 M), or with the Ca 2ϩ chelator, BAPTA-AM (25 M). Cells were pretreated for 30 min with the PKC␦-specific inhibitor Rottlerin (15 M). Cells were stimulated with either 10 ng/ml EGF or 5 M PMA for 3 min. Cell lysates were prepared, and equal amounts of protein were resolved on 10% SDS-PAGE gels and immunoblotted for activated MAPKs using an antibody that recognizes phosphorylated, activated ERK1 and ERK2. The specificity of the PKC inhibitors was confirmed, in vitro. 200 ng of recombinant PKC␣ and PKC⑀ were incubated in P21 buffer containing either 10 M GF109203X or Gö6976 or Me 2 SO solvent alone, 400 ng of PKC⑀ or PKC␣ peptide substrate, 10 M ATP, 10 Ci of [␥-32 P]ATP (3000 Ci/mmol), 10 M phosphatase inhibitors, 10 M CaCl 2 (PKC␣ buffer only), 50 ng/ml PMA and were incubated at 30°C for 30 min. Reactions were terminated by the addition of 120 l of 10% (v/v) phosphoric acid; the reactions were spotted onto P81 paper and washed extensively with 0.5% (v/v) phosphoric acid. The P81 paper was dried with acetone, and the samples were processed for counting in a scintillation counter.
PKC⑀ Antisense ODN Studies-PKC⑀ antisense ODN sequences were derived from the mouse PKC⑀ cDNA sequence and selected to recognize areas of minimal secondary structure and base pairing using the RNA folding program Mulfold. A specific 18-mer PKC⑀ sense ODN and complementary antisense sequences were synthesized as phosphorothioatemodified, high pressure liquid chromatography-purified DNA ODNs. The 18-mer sense ODN sequence is cagacgttccttttggac, and its complementary antisense sequence is gtccaaaaggaacgtctg. These sequences correspond to bases 115-132 of the published mouse PKC⑀ cDNA sequence (NM005400). Confluent C3H10T1/2 cells were transiently transfected for 5 h with 1 M sense or antisense PKC⑀-specific ODN in the presence of the cationic lipid, LipofectACE (Life Technologies, Inc.). Cells were allowed to recover overnight and then serum-starved for 48 h before being stimulated with 5 M PMA for 3 min. Cells were lysed in 400 l of CHAPS buffer, and the lysates were analyzed for PKC expression and MAPK activity.
MAPK Activity Assay-MAPK activity was measured in a direct kinase assay. Briefly, MAPK was immunoprecipitated from equal amounts of cell protein for 2 h using a MAPK-specific rabbit polyclonal antibody precoupled to protein A-Sepharose. Immunoprecipitates were washed with P21 buffer and then incubated in 50 l of P21 buffer containing 10 Ci of [␥- 32  Phorbol Ester-mediated Activation of Raf-1-c-N-Ras immunoprecipitates were washed as described above. Immunoprecipitates were then washed once in Buffer A (50 mM Tris-Cl, pH 7.5, 15 mM MgCl 2 , 10 M phosphatase inhibitors, 0.01 mg/ml leupeptin, 2 mM MnCl 2 ) and incubated in 50 l of Kinase Buffer B (Buffer A containing 25 M ATP, 30 Ci of [␥-32 P]ATP, 2.5 g of GST-MEK1(K97A), 25 g/ml phosphatidylserine, and 50 ng/ml PMA) for 30 min at 30°C. Additional calcium was not added to the buffers. 10 M GF109203X was included in some reactions to inhibit PKC activity. Kinase reactions were terminated by the addition of 50 l of 2ϫ Laemmli buffer, and phosphorylated proteins were resolved on 8% SDS-PAGE gels and visualized on a Molecular Dynamics StormImager.
PKC⑀ Activity toward Kinase-dead GST-MEK1(K97A)-1-5 ng of purified, recombinant human PKC⑀ was incubated with 2.0 g of GST-MEK1(K97A) in Kinase Buffer B with or without phosphatidylserine and PMA for 30 min at 30°C. In parallel reactions, 2 ng of recombinant human PKC⑀ was incubated with 10 -100 ng of PKC⑀ peptide substrate (NH 2 -ERMRPRKRQGSVRRRV-OH) in kinase buffer B containing PMA for 30 min at 30°C. After 30-min reactions were transferred to 0.22 M GS filters (Millipore), prewashed with P21 buffer containing 1 mM ATP, and washed extensively with the same buffer. [ 32 P]Phosphate incorporation was determined by scintillation counting.
N-Ras-Raf-1 Destabilization Conditions-CHAPS-solubilized rat brain extracts (30) were used as a source of N-Ras. These lysates were exchanged with 100 M GTP in the presence of 25 mM EDTA for 1 h at 4°C. Excess Mg 2ϩ was added to stabilize the N-Ras-GTP. Glutathioneagarose precoupled (and washed) with GST fused to the Ras-binding domain of Raf-1 (GST-RBD) was added to the lysates. Following a 1-h incubation at 4°C the beads were washed and then incubated in the indicated concentrations of buffered NaCl (1 h at 4°C). The beads were again washed, and the amount of N-Ras associated with the GST-RBD was assessed by Western analysis. A similar experiment was performed in which the preformed N-Ras-RBD complexes were incubated (1 h at 4°C) with buffer at the indicated pH values (ranging from 4 to 9) rather than increasing NaCl concentrations. The GST-RBD-agarose was washed, and the amount of N-Ras associated with the RBD was determined by Western analysis.
GTP Binding Assays-GTP binding assays were performed with [␣-32 P]GTP in p21 buffer supplemented with 25 mM EDTA and 100 M GTP. These assays were done in two different ways. First we tested whether the exchange was inhibited at buffer pH values ranging from 4 -9. This was done simply by altering the pH of the assay buffer and performing the filter binding assay as described previously (30). Second, we tested whether the Ras-GTP complex itself was stable at pH values ranging from 4 to 9. In this assay the GTP binding was performed at pH 7.4 in the standard binding buffer described above. Excess Mg 2ϩ was added to stabilize the complex; the pH was altered, and excess unlabeled GTP was added to quench any labeled GTP released by the change in pH. After 1 h at 4°C, the reactions were then filter through 0.22-micron filters as described above. Bound radioactivity was determined by scintillation counting.

Determination of the Guanine Nucleotide State of c-N-Ras in Purified Plasma
Membranes-Purified plasma membranes were prepared as described earlier (54). The plasma membrane pellet was solubilized in p21 containing 1% CHAPS and clarified by centrifugation. The supernatant was divided into 5 aliquots. One aliquot was directly incubated with prebound GST-RBD. The remaining aliquots were incubated in pH 4 buffer for 1 h. At this point 2 aliquots were neutralized (using 1.0 M MOPS, pH 7.4) and incubated with glutathione-agarose (without the GST-RBD) or glutathione-agarose-GST-RBD. The remaining two aliquots were exchanged with either GMPPNP or GDP in the presence of excess EDTA. Excess Mg 2ϩ was added along with the agarose-GST-RBD and MOPS buffer to neutralize the pH. The samples were incubated for 1 h at 4°C and the beads washed, and the amount of N-Ras associated with the GST-RBD was analyzed by Western analysis.

RESULTS
c-N-Ras and c-Raf-1 Are Stably Associated in Quiescent, Normal Mouse C3H10T1/2 Fibroblasts-We had previously identified a Ras isoform-specific association of c-Raf-1 with N-Ras in cells transformed by either oncogenic Ha-Ras or N-Ras (32). This association of c-Raf-1 with c-N-Ras was also observed when c-N-Ras was immunoprecipitated from quiescent, normal mouse C3H10T1/2 fibroblasts under conditions of serum starvation (Fig. 1A). Complexes of c-N-Ras and Raf-1 were also c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester detected in serum-starved NIH3T3, intestinal epithelial cells, and Rat2 fibroblasts (data not shown). The c-Raf-1 in these c-N-Ras immunoprecipitates, however, did not possess significant activity, as measured by the ability of the co-immunoprecipitated c-Raf-1 to phosphorylate kinase-dead GST-MEK1 (Fig. 1B), a physiological substrate for Raf-1 (3). Prior EGF stimulation, however, resulted in c-N-Ras-associated c-Raf-1 being active. The apparent reduction in the electrophoretic mobility of c-Raf-1 associated with c-N-Ras immunoprecipitates upon EGF stimulation observed in Fig. 1A most probably reflects EGF-induced changes in Raf-1 phosphorylation. This is similar to the electrophoretic mobility of the c-Raf-1 that coimmunoprecipitates with c-N-Ras in Ha-Ras-transformed C3H10T1/2 (11A) cells, whose constitutive MAPK activity is dependent upon autocrine EGF receptor activation (33). Since inactive c-Raf-1 appeared to be constitutively associated with c-N-Ras in quiescent cells, we sought to determine whether the kinase activity of this inactive c-Raf-1 could be stimulated.
The current perception of Ras-regulated Raf-1 signaling is that upon Ras activation, whether through point mutation or the actions of receptor-driven guanine nucleotide exchange factor (e.g. SOS) activity, active Ras binds cytosolic c-Raf-1, thereby recruiting the kinase to the plasma membrane (14). The physical interaction of Ras with c-Raf-1 and its juxtaposition at the plasma membrane facilitates c-Raf-1 activation. However, Raf-1 was detected in purified plasma membranes isolated from serum-starved, quiescent C3H10T1/2 cells (Fig.  1C), and this plasma membrane association is in the absence of any significant MAPK activity. Upon subsequent PMA or EGF stimulation there is an increase in plasma membrane c-Raf-1 levels, indicating that mitogen stimulation induces some recruitment of Raf-1 to the plasma membrane. The observations that Raf-1 is already membrane-associated in quiescent cell plasma membranes and that it is present in a biochemically inactive complex with Ras might poise the Raf-1 in this signaling module in a unique position to be rapidly activated prior to, or in tandem with, mitogen-induced Ras activation and subsequent Raf-1 recruitment and activation. Since the association of Raf-1 with Ras alone is not sufficient to stimulate Raf-1 activity, the inactive c-N-Ras⅐Raf-1 complexes were analyzed further for putative Raf-1 activators.
PKC Co-immunoprecipitates with c-N-Ras and c-Raf-1-Further immunoblot analysis of plasma membranes from quiescent C3H10T1/2 cells also identified the presence of both PI3-OH kinase and PKC (data not shown). Both PKC and PI3-OH kinase have been described as having significant effects on c-Raf-1 activity. Several PKC isozymes are able to phosphorylate Raf-1 in vitro, stimulating its activity (25,26,35). PI3-OH kinase appears to be able to exert both positive and negative regulatory effects on Raf-1 activity through the phosphatidylinositol 3,4,5-trisphosphate-stimulated activation of phosphoinositide-dependent kinase (reviewed in Ref. 36) and protein kinase B/Akt (37), respectively. Since the PKC-mediated activation of Raf-1 appears to require Ras, we were prompted to further investigate the relationship between PKC and the inactive c-N-Ras⅐Raf-1 complex.
Since both EGF and PMA stimulate the MAPK cascade, we analyzed the status of the c-N-Ras⅐Raf-1 complexes from unstimulated and EGF-and PMA-stimulated cells. Again, c-Raf-1 co-immunoprecipitated with c-N-Ras from serum-starved control and EGF-stimulated cells. c-Raf-1 was also present in the c-N-Ras immunoprecipitates from PMA-stimulated cells ( Fig.  2A, upper panel). PMA stimulation, like EGF stimulation, can induce a reduction in c-Raf-1 electrophoretic mobility. Overnight treatment with PMA down-regulates the levels of DAGdependent cPKC and nPKCs. This chronic PMA treatment blocks MAPK activation in serum-starved C3H10T1/2 cells stimulated with PMA but not with EGF (see Fig. 4A). Therefore, to determine whether PKC was a component of the inactive c-N-Ras⅐Raf-1 signaling complex, c-N-Ras immunoprecipitates were probed for PKC using a mixture of PKC isoformspecific monoclonal antibodies. Western blot analysis detected a band of ϳ90 kDa in c-N-Ras immunoprecipitates from both quiescent and EGF-stimulated cells ( Fig. 2A, lower panel). This 90-kDa band was also present in c-N-Ras immunoprecipitates from PMA-stimulated cells, although at a much reduced level. c-N-Ras immunoprecipitates from chronic PMA-treated cells did not contain the 90-kDa PKC band (Fig. 2B, upper panel), although c-Raf-1 still co-immunoprecipitated with c-N-Ras (Fig. 2B, lower panel). These data suggest that a phorbol esterdown-regulatable PKC isozyme(s) associates with c-N-Ras⅐Raf-1 complexes. The reduction in the level of PKC that co-immunoprecipitates with c-N-Ras after PMA stimulation may reflect a rapid down-regulation of PKC in the complexes. The absence of the 90-kDa PKC in the c-N-Ras immunoprecipi- 1. A, c-Raf-1 is associated with c-N-Ras in serum-starved cells. C3H10T1/2 cells were serum-starved for 48 h and stimulated with 10 ng/ml EGF for 3 min. Ras was immunoprecipitated from cell lysates using a Ras isoform-specific antisera (32), and the immunoprecipitates were resolved on 6% SDS-PAGE gels, transferred to a PVDF membrane, and blocked. Membranes were immunoblotted for c-Raf-1, and immunoreactive bands were visualized by ECL. Arrows indicate the relative positions of Raf-1 in quiescent and mitogen-stimulated cells. IP, immunoprecipitating antisera; NI, non-immune sera; N-Ras, N-Rasspecific antisera; Ha-Ras, Ha-Ras-specific antisera. The data are representative of at least five independent experiments. B, c-Raf-1 that co-immunoprecipitates with c-N-Ras prior to EGF stimulation is inactive. c-N-Ras was immunoprecipitated from serum-starved C3H10T1/2 cells or from EGF-stimulated cells. Immunoprecipitates were assayed for Raf-1 activity in a direct kinase assay using kinase-dead GST-MEK1(K97A) as a Raf-1-specific substrate. Kinase reactions were resolved on SDS-PAGE gels and transferred to PVDF membrane. 32 P-Phosphorylated GST-MEK1 was detected using a Molecular Dynamics StormImager. The data are representative of at least three independent experiments. C, the plasma membrane of quiescent cells contains c-Raf-1. Plasma membrane preparations from control (Con), serumstarved C3H10T1/2 cells, and from cells stimulated for 3 min with 10 ng/ml EGF or 5 M PMA were solubilized in CHAPS buffer, and 40 g of protein were resolved on 8% SDS-PAGE gels and immunoblotted for Raf-1. The characteristic mitogen-induced change in Raf-1 electrophoretic mobility seen in 6% SDS-PAGE gels (A) is not as evident in 8% SDS-PAGE gels. Immunoreactive bands were visualized by ECL. The data are representative of two independent experiments. c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester tates from chronic PMA-treated cells indicates that the association of Raf-1 with c-N-Ras is independent of PKC association. We next determined whether the Raf-1 associated with c-N-Ras was indeed phosphorylated through a PKC-dependent mechanism. Serine 338 has been identified by others as a site of PKC phosphorylation on Raf-1. Serum-starved C3H10T1/2 cells FIG. 2. A, PKC and c-Raf-1 co-immunoprecipitate with c-N-Ras. c-N-Ras was immunoprecipitated from whole cell lysates prepared from control, serum-starved C3H10T1/2 cells (Con) and from cells stimulated with either 10 ng/ml EGF or 5 M PMA for 3 min. Immunoprecipitates were resolved on a 6% SDS-PAGE gel and immunoblotted for Raf-1 (upper panel). Immunoprecipitates were also probed for PKC using a mixture of PKC isoform-specific monoclonal antibodies (lower panel). The positions of Raf-1 and PKC are indicated. The data are representative of two independent experiments. B, chronic PMA treatment down-regulates c-N-Ras-associated PKC. As described above, parallel cell cultures were treated with 5 M PMA overnight (chronic PMA treatment) to down-regulate cPKC and nPKC isoforms. Cells were then stimulated with either 10 ng/ml EGF or 5 M PMA for 3 min. c-N-Ras was immunoprecipitated from cell lysates and the immunoprecipitates immunoblotted for c-Raf-1 and PKC. C, PMA treatment results in serine 338 phosphorylation of Raf-1 within the c-N-Ras ⅐Raf-1 complex. Cells were untreated or treated with 5 M PMA Ϯ the MEK inhibitor U0126 (10 M) for 3 min. Cell lysates were prepared as described above. Lysates were immunoprecipitated for N-Ras and probed with an antibody directed against the phosphorylated serine at position 338 of Raf-1. D, classical and novel PKC expression in C3H10T1/2 cells. Cell lysates were prepared from 72-h starved cells with or without overnight treatment with 5 M PMA. Equal amounts of protein were resolved on 8% SDS-PAGE gels, transferred to PVDF membranes, and immunoblotted for distinct PKC isoforms as indicated. A Raf-1 Western blot was included as a protein loading control. E, c-Raf-1 co-immunoprecipitates with membrane-associated PKC␦ and PKC⑀ . (i) c-Raf-1 and the indicated PKCs were immunoprecipitated (2.5 g of antibody per immunoprecipitation) from 50 g of membrane-enriched or cytosolic fractions prepared from serum-starved cells. The immunoprecipitates were resolved on 8% SDS-PAGE gels, transferred to PVDF membrane, and immunoblotted for Raf-1. The position of Raf-1 is indicated. IP, immunoprecipitating antibody; NI, non-immune IgG. The immunoprecipitation of PKC␦ and PKC⑀ from the membrane-enriched fraction in E(i) was confirmed by Western blot in E(ii). (iii) Raf-1 is readily immunoprecipitated from cytosolic but not membrane fractions. Cytosolic and membrane-enriched fractions (50 g) were immunoprecipitated with the c-Raf-1-specific monoclonal antibody (2.5 g). Washed immunoprecipitates were resolved on 8% SDS-PAGE gels and immunoblotted for c-Raf-1. F, c-N-Ras and PKC⑀ co-immunoprecipitate. Upper panel, c-Raf-1, PKC⑀, and PKC␦ were immunoprecipitated from membrane-enriched fractions as above. Immunoprecipitates were resolved on a 16.5% SDS-PAGE gel and immunoblotted for c-N-Ras. The position of Ras (21 kDa) is indicated. Lower panel, c-N-Ras was immunoprecipitated from 30 g of plasma membrane prepared from serum-starved C3H10T1/2 cells. The immunoprecipitate was resolved on a 8% SDS-PAGE gels and immunoblotted for PKC⑀. Immunoreactive bands were visualized by ECL. The position of PKC⑀ is indicated. IP, immunoprecipitating antibody; NI, non-immune antibody.
were treated with PMA, lysed, and immunoprecipitated with antisera against c-N-Ras. The immunoprecipitates were then analyzed for the presence of phosphoserine 338 on Raf-1. Fig.  2C demonstrates a clear increase in the amount of Ser(P)-338-Raf-1 associated with c-N-Ras following treatment with phorbol esters. Inclusion of the MEK inhibitor U0126 failed to reduce the level of Raf-1 phosphorylation, suggesting that the phosphorylation does not result from a MAPK feedback loop. Similar to the results reported by Marias et al. (34), PKCmediated activation of Raf-1 does occur in the context of a pre-existing Ras⅐Raf-1 complex.
Since PKC appears to be a component of inactive c-N-Ras⅐Raf-1 complexes, cPKC and nPKC expression in C3H10T1/2 cells was determined by Western blot analysis. C3H10T1/2 cells express cPKC␣; neither cPKC␤ nor cPKC␥ was detected in the lysates. nPKC␦ and nPKC⑀ were also present in C3H10T1/2 cell lysates (Fig. 2D), whereas nPKC was poorly expressed, if at all. PKC␣, -␦, and -⑀ have each been implicated as Raf-1 activators. Overnight treatment with PMA, as expected, down-regulated the expression of PKC␣, -␦, and -⑀ isoforms but had no effect upon c-Raf-1 expression.
It has been reported previously that PKC⑀ (90 kDa) is associated with Raf-1 in Rat6 fibroblasts transformed by the overexpression of PKC⑀ (28,38). Since a 90-kDa PKC appeared to co-immunoprecipitate with c-Raf-1 and c-N-Ras, we sought to determine whether endogenous PKC⑀ was constitutively associated with endogenous c-Raf-1 in C3H10T1/2 cells; c-Raf-1 was found associated with both PKC⑀ and PKC␦ immunoprecipitates from membranes isolated from serum-starved C3H10T1/2 cells (Fig. 2E(i)). The amounts of PKC␦ and PKC⑀ immunoprecipitated from the membrane preparation is indicated in Fig.  2E(ii). PKC␣ immunoprecipitates did not appear to co-immunoprecipitate c-Raf-1. In contrast, PKC␦ and PKC⑀ immunoprecipitates from a cytosolic fraction did not appear to be associated with a significant amount of c-Raf-1 (Fig. 2E(i)), although weak Raf-1 signals could be detected after longer film exposures (data not shown). The data, however, strongly suggest that in quiescent cells, PKC␦ and -⑀ are only associated with membrane-bound Raf-1. PKCs were not detected in c-Raf-1 immunoprecipitates from membranes. This may, however, reflect significant differences in the ability of the Raf-1 antibodies to immunoprecipitate endogenous c-Raf-1 from cytosolic and membrane fractions. Endogenous c-Raf-1 could be readily immunoprecipitated from cytosolic extracts but not membrane extracts (Fig. 2E(iii)) and may reflect differences in the macromolecular nature of cytosolic and membrane-associated c-Raf-1 or the masking of the c-Raf-1 epitope upon plasma membrane association.
When PKC␦, and PKC⑀ immunoprecipitates from membranes were immunoblotted for c-N-Ras, only the PKC⑀ immunoprecipitate contained a 21-kDa immunoreactive band consistent with c-N-Ras (Fig. 2F, upper panel). Parallel analysis of the purified antibodies confirmed that the immunoreactive 21-kDa band in the PKC⑀ lane does not correspond to variable mobility of the IgG light chain (data not shown). Conversely, PKC⑀ was detected in c-N-Ras immunoprecipitates (Fig. 2F,  lower panel). Similar experiments, however, failed to detect PKC␦ in c-N-Ras immunoprecipitates (data not shown). These observations imply that in quiescent cell membranes, both endogenous PKC␦ and PKC⑀ are closely associated with c-Raf-1. However, only PKC⑀ appears to be associated with c-Raf-1 in a complex that contains c-N-Ras.
Antibodies to PKC⑀ Increase the Molecular Mass of Plasma Membrane-associated c-Raf-1 and c-N-Ras-The previous coimmunoprecipitation data suggest that there might be ternary complex between c-N-Ras, Raf-1, and PKC⑀ in serum-starved C3H10T1/2 cells. These data, however, do not exclude the possibility that there are individual complexes composed of c-N-Ras⅐Raf-1, PKC⑀⅐Raf-1, and c-N-Ras⅐PKC⑀. To address this issue, the interaction between PKC⑀ and the c-N-Ras⅐Raf-1 complex was further analyzed by Sephacryl S200HR gel filtration chromatography. Purified plasma membrane proteins were prepared from serum-starved C3H10T1/2 cells and solubilized with 1% (w/v) CHAPS. The CHAPS-solubilized plasma membranes were incubated with a control or PKC⑀-specific antibody for 2 h prior to separation on a Sephacryl S200 HR column. Fractions were collected and resolved on 6 -15% gradient SDS-PAGE gels, and the elution profiles of Raf-1 and c-N-Ras in each fraction were analyzed by Western blotting. In control antibody treated CHAPS solubilized plasma membranes, c-N-Ras and c-Raf-1, co-eluted with an apparent molecular mass of greater than 150 kDa (Fig. 3A). This is identical to their observed molecular mass in the absence of added an- FIG. 3. A, PKC⑀-specific antibodies increase the molecular mass of plasma membrane-associated c-Raf-1 and c-N-Ras. Plasma membranes were prepared from 72-h starved C3H10T1/2 cells, and equal amounts of CHAPS-solubilized plasma membrane protein were incubated with control (Con) antibody or with a PKC⑀-specific antibody (P14820; Transduction Laboratories) for 2 h at 4°C, followed by high speed centrifugation to remove protein aggregates. The soluble protein was then subjected to gel filtration chromatography at 4°C through a 1-m Sephacryl S200HR column equilibrated in TBS, pH 7.6, containing 0.2% (w/v) CHAPS. 0.95-ml fractions were collected, and 100-l aliquots taken from every second fraction for separation on 6 -15% gradient SDS-PAGE gels. Gels were transferred to PVDF membrane and immunoblotted for c-N-Ras and c-Raf

c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester
tibody and in the presence of a PKC␣-specific antibody (data not shown). The mobility of these proteins roughly correlates to their additive molecular weights 180,000 (90 kDa(PKC) ϩ 70 kDa(Raf-1) ϩ 20 kDa(Ras)). Incubation with the anti-PKC⑀ antibody, however, induced a large change in the molecular mass of most of the c-N-Ras and c-Raf-1. Densitometric analysis of these c-Raf-1 and c-N-Ras fractions indicated a significant overlap in the control and PKC⑀ antibody-treated samples (Fig. 3B), suggesting that a portion of the c-N-Ras⅐Raf-1 complexes is complexed with PKC⑀. These finding support the hypothesis that endogenous c-N-Ras, c-Raf-1, and PKC⑀ are constitutively and stably associated with each other in the plasma membranes of serum-starved C3H10T1/2 cells.
Phorbol Ester Regulates MAPK through the Novel PKC⑀ -Since both classical and novel PKCs have been implicated as Raf-1 regulatory kinases, we investigated the role of PKC⑀ in c-Raf-1 regulation in C3H10T1/2 cells. cPKC and nPKC activities were inhibited either through chronic phorbol ester-mediated PKC down-regulation or by PKC class-specific inhibitors. Chronic phorbol ester treatment blocked MAPK activation in response to PMA but not EGF stimulation (Fig. 4A(i)). Pretreatment of cells with 25 M BAPTA-AM, a cell-permeable Ca 2ϩ chelator (39), did not block ERK activation in response to PMA stimulation (Fig. 4A(ii)), indicating that changes in intracellular calcium, including the activation of Ca 2ϩ -dependent PKC␣, were not necessary for PMA-mediated ERK activation. A 1-h pretreatment with 10 M GF109203X, an inhibitor of both classical and novel PKCs (40,41), blocked MAPK activation in response to PMA stimulation but not EGF stimulation (Fig. 4B(i)). Pretreatment of serum-starved cells with the specific PKC␣ and PKC␤I inhibitor, Gö6976 (41), failed to block MAPK activation in response to either PMA or EGF stimulation ( Fig. 4B(i)). EGF also stimulated ERK activation in chronic PMA-treated cells that were preincubated with either GF10920X or Gö6976 (Fig. 4B(ii)), demonstrating that EGFmediated activation of the MAPK cascade occurred independently of cPKC or nPKC activation. The specificity of the PKC class-specific inhibitors was confirmed, in vitro, using recombinant PKC␣ and PKC⑀. GF109203X completely blocked PKC␣ and PKC⑀ activity, whereas Gö6976 inhibited only PKC␣ (Fig.  4C). Similar experiments were performed using the PKC␦specific inhibitor, Rottlerin (42) (Fig. 4D). Pretreatment with Rottlerin failed to block PMA-mediated MAPK activation in serum-starved C3H10T1/2 cells. From these studies we can conclude that PMA-mediated stimulation of the MAPK cascade occurs through a Ca 2ϩ -independent, phorbol ester-dependent, and GF10920X-sensitive novel PKC(s) that is not PKC␦. These observations agree with the previous data in implicating PKC⑀ as the specific PKC isoform that activates Raf-1 upon phorbol ester stimulation.
The role of PKC⑀ in regulating the MAPK cascade was directly addressed through the specific down-regulation of PKC⑀ protein by antisense ODNs. Cationic liposomes (43,44) were used to transiently transfect C3H10T1/2 cells with 1 M PKC⑀ sense or antisense ODNs for 5 h. The cells were then serumstarved for 48 h. Cells were lysed following 3 min of stimulation with PMA, and the activity of immunoprecipitated MAPK was measured using MBP as a substrate. PMA stimulated a robust activation of MAPK in cells incubated with PKC⑀ sense ODN (Fig. 4E(i)), identical to that seen in untreated cells (data not shown). In contrast, MAPK activity was severely abrogated in cells treated with the PKC⑀ antisense ODN where the levels of PKC⑀ protein were reduced by the antisense ODN (Fig. 4E(ii)). Western blots of cell lysates confirmed the specificity of the PKC⑀ ODN. PKC␣ protein levels were not significantly affected by PKC⑀ ODN treatment nor were the levels of c-Raf-1 protein, a PKC-related kinase (data not shown). There was a strong correlation between the degree of reduction in PKC⑀ protein levels and PMA-stimulated MAPK activity; a 50 -76% ODNmediated reduction in PKC⑀ protein levels correlated with a 59 -87% reduction in PMA-stimulated MAPK activity (Fig.  4E(iii)). These data would appear to confirm PKC⑀ as a critical component of the mechanism by which PMA activates MAPK.
Phorbol Ester Stimulates Raf-1 Activity and Its Phosphorylation on Ser-338 in c-N-Ras Immunoprecipitates-In order to address the functional significance of the endogenous, signaling complexes of c-N-Ras, c-Raf-1, and PKC⑀, PMA was directly added to immunoprecipitated complexes from serum-starved cells. c-Raf-1 activity in the complex was then determined using a kinase-dead GST-MEK1(K97A) substrate in a direct kinase assay. As observed previously, c-N-Ras immunoprecipitates from serum-starved cells contained no detectable c-Raf-1 activity (Fig. 5A). However, the inclusion of PMA in the reaction mix stimulated Raf-1 activity. The PMA-mediated activation of c-Raf-1 could be inhibited by pre-incubating the immunocomplexes with GF109203X. In contrast, GF109203X failed to block c-Raf-1 activity in c-N-Ras immunoprecipitates from cells stimulated with PMA prior to cell lysis (data not shown).
In control experiments, recombinant human PKC⑀ did not phosphorylate the GST-MEK1 substrate, even in the presence of PMA (Fig. 5B(i)), demonstrating that the PKC⑀ present in the c-N-Ras⅐Raf-1⅐PKC⑀ immunocomplexes cannot use GST-MEK1 as a substrate. The activity of the recombinant PKC⑀ used in the in vitro assay was confirmed using a PKC⑀-specific peptide substrate in parallel assays (Fig. 5B(ii)).
In subsequent experiments, the c-Raf-1 associated with c-N-Ras immunoprecipitates was analyzed, by Western blot, for phosphorylation at serine 338, which is phosphorylated upon PMA stimulation in vivo (7,8) and by PAK3 in vitro (24). C3H10T1/2 cells were made quiescent by serum deprivation, and c-N-Ras was immunoprecipitated as described previously. The washed immunoprecipitates were either left untreated, incubated with PMA, or preincubated with PMA plus 10 M GF109203X on ice before incubation at 30°C for 30 min. The c-N-Ras-associated c-Raf-1 was then analyzed for Ser-338 phosphorylation using a phospho-(Ser-338)Raf-1-specific antibody (8). Western blot analysis demonstrates that the c-Raf-1 that co-immunoprecipitates with c-N-Ras from quiescent cells is not phosphorylated on Ser-338. Incubation of these c-N-Ras immunoprecipitates with PMA resulted in a strong phospho-(Ser-338)Raf-1 signal, and this phospho-(Ser-338)Raf-1 signal was almost completely inhibited by the inclusion of GF109203X in the reaction mixture (Fig. 5C). Therefore, PMA stimulates c-Raf-1 activity in inactive c-N-Ras⅐Raf-1 complexes by inducing the phosphorylation of Raf-1 on the critical, PKC consensus, Ser-338 phosphorylation site. Taken together, the data support the hypothesis that these biochemically inactive, constitutive, complexes of c-N-Ras, c-Raf-1, and PKC⑀ are latent and represent physiologically relevant signaling entities awaiting selective activation signals.
Determination of the Guanine Nucleotide(s) Bound to c-N-Ras within the Latent c-N-Ras⅐Raf-1⅐PKC⑀ Complex-There are two currently used protocols to determine the amount of Ras-GTP within a cellular environment. The first involves metabolically labeling cells with [ 32 P]orthophosphate, immunoprecipitating the Ras proteins and analyzing the bound nucleotides by TLC analysis. Whereas this method has been useful for determining the relative ratio of GTP to GDP bound for total Ras proteins, we have found that this method does not give a sufficient signal to noise ratio when examining the GTP-bound state of specific Ras isoforms (data not shown). The second method involves using the Ras-binding domain (RBD) of Raf-1 c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester FIG. 4. A, PMA-mediated MAPK activation is inhibited by chronic phorbol ester treatment. (i) 72-h serum-starved control and chronic PMA-treated C3H10T1/2 cells were stimulated with 5 M PMA or 10 ng/ml EGF for 3 min. Cell lysates were prepared, and equal amounts of protein were resolved on 10% SDS-PAGE gels and immunoblotted for activated MAPK using an antibody that recognizes phosphorylated, activated MAPK (Santa Cruz Biotechnology). Immunoreactive bands were visualized by ECL. Activated MAPK proteins (pERK1 and -2) are indicated (lower panel). Equal protein loading was confirmed by immunoblotting for ERK1 protein levels (upper panel). The data are representative of three independent experiments. (ii) a calcium-independent PKC mediates MAPK activation in response to PMA stimulation. Serum-starved C3H10T1/2 cells were incubated for 1 h with 25 M BAPTA-AM, a cell-permeable calcium chelator, prior to stimulation with 5 M PMA for 3 min. Equal amounts of cell lysates were resolved on SDS-PAGE gels and immunoblotted for activated MAPKs as described. The data are representative of three independent experiments. B, a novel PKC mediates MAPK activation in response to PMA stimulation. (i) 72-h serum-starved cells were incubated with 10 M GF109203X (GFX) or 10 M Gö6976 (Gö) 1 h prior to stimulation with either 10 ng/ml EGF or 5 M PMA for 3 min. Cell lysates were prepared and analyzed for activated MAPKs as described above. The positions of activated pERK1 and pERK2 are indicated (lower panel). Equal protein loading was confirmed by immunoblotting for ERK1 protein levels (upper panel). (ii) EGF stimulates MAPK activation independently of nPKC or cPKC. Chronic PMA-treated C3H10T1/2 cells were preincubated with 10 M of either GF109203X or Gö6976. Cells were then stimulated with 10 ng/ml EGF for 3 min, and cell lysates were prepared and resolved on 8% SDS-PAGE gels. Lysates were analyzed for phosphorylated ERK1 and ERK2 as described above (lower panel). Equal protein loading was confirmed by immunoblotting for ERK1 protein levels (upper panel). C, Gö6976 is a specific inhibitor of Ca 2ϩ -dependent PKCs. Recombinant cPKC␣ and nPKC⑀ (200 ng) were incubated in kinase buffer containing 10 Ci of [␥-32 P]ATP and 400 ng of PKC peptide substrates in the presence or absence of either 10 M GF109203X or Gö6976 for 30 min at 30°C. The reaction was stopped, and the incorporation of radiolabeled phosphate into the substrates was quantitated by scintillation counting. The data are representative of two independent experiments performed in triplicate. D, a PKC␦-specific inhibitor does not block PMA-mediated MAPK activation. 72-h serum-starved C3H10T1/2 cells were pretreated with 15 M Rottlerin prior to treatment with PMA for 3 min (5 M). Cells were harvested and lysed, and the level of phosphorylated MAPK was determined as described previously. Raf-1 was blotted to confirm equal loading. E(i), antisense-mediated down-regulation of PKC⑀ protein inhibits PMA-stimulated MAPK activation. C3H10T1/2 cells transiently transfected with 1 M sense (S) or antisense (AS) PKC⑀-specific ODN were allowed to recover overnight and then serum-starved for 48 h before being stimulated with 5 M PMA for 3 min. Cells were lysed in 400 l of lysis buffer, and MAPK was immunoprecipitated from 300 l of lysate. The activity of the immunoprecipitated MAPK was measured in a direct kinase assay using MBP as a substrate, and MBP phosphorylation was visualized and quantitated on a Molecular Dynamics StormImager. AS(1) and AS(2) represent two independent determinations with PKC⑀ antisense ODN. The antisense experiments were performed twice, each in triplicate. The results shown are representative of these replicates. In each experiment the percent reduction in the MAPK activity was always higher than the percent reduction in PKC⑀ expression. E(ii), PKC⑀ c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester to trap Ras-GTP. This is then analyzed by Western analysis for the amount of Ras bound to the GST-RBD fusion protein. The problem with this particular approach is that this method will not detect Ras-GTP bound to effector proteins, i.e. Ras-GTP has only one effector-binding site. To determine whether the c-N-Ras within the latent c-N-Ras⅐Raf⅐PKC⑀ complex was bound with either GTP or GDP, we set up a model system using c-N-Ras from rat brain lysates and the GST-RBD to destabilize the Ras-Raf interaction without altering the guanine nucleotide associated with the N-Ras protein. We tested two different conditions, increasing salt concentrations (Fig. 6A) and pH extremes (Fig. 6B). The interaction between N-Ras and Raf-1 is not destabilized by increasing salt concentrations (Fig. 6A). We did find, however, that the c-N-Ras⅐Raf complex could be destabilized by exposure to either pH 9 or pH 4 buffer (Fig. 6B). We then tested whether Ras-GTP itself was stable under either of these pH extremes. This was tested using two slightly different protocols. In Fig. 6C, we analyzed the ability of recombinant c-N-Ras to exchange nucleotides in buffer at the indicated pH values. Although there is some variability in the In parallel reactions, c-N-Ras immunoprecipitates were incubated in kinase buffer containing 10 M GF109203X. All reactions were incubated on ice for 10 min prior to incubating at 30°C for 30 min. The kinase reaction was stopped by the addition of 50 l of 2ϫ Laemmli buffer. Reactions were resolved on 8% SDS-PAGE gels, and phosphorylated GST-MEK1 was visualized on a Molecular Dynamics StormImager. The data are representative of at least three independent experiments. B, GST-MEK1 is not a PKC⑀ substrate. (i) 2 g of kinase-dead GST-MEK1 was incubated with 1-5 ng of recombinant PKC⑀ in a kinase reaction, with or without PMA and phosphatidylserine for 30°C for 30 min. (ii) in parallel, 10 -100 ng of PKC⑀ substrate was incubated with 2 ng of recombinant PKC⑀ in a kinase reaction to confirm the activity of the recombinant PKC⑀ kinase. After the reactions were terminated, the incorporation of [ 32 P]phosphate into substrate peptide was determined by scintillation counting. C, PMA stimulates the phosphorylation of Raf-1 on serine 338. Cell lysates prepared from serum-starved C3H10T1/2 cells were split over several c-N-Ras immunoprecipitates, and c-N-Ras was immunoprecipitated as described previously. In identical reactions as described above, the c-N-Ras immunoprecipitates were then incubated in kinase buffer with ATP but without [␥-32 P]ATP or kinase-dead GST-MEK1(K97A) for 30 min at 30°C. Reactions were terminated and resolved on 6% SDS-PAGE gels, transferred to PVDF membrane, and immunoblotted for phosphorylated Raf-1 using a phospho-(Ser-338)Raf-1-specific rat monoclonal antibody. Immunoreactive bands were visualized by ECL.
antisense ODNs specifically down-regulate PKC⑀ protein levels. Two 40-l aliquots of cell lysate from the PKC⑀ ODN-treated cells as described above were resolved on 8% SDS-PAGE gels and immunoblotted for PKC⑀ (upper panel) and PKC␣ (lower panel). Immunoreactive bands were visualized by ECL. E(iii), quantitation of PKC⑀ antisense ODN effects on PKC⑀ expression and PMA-stimulated MAPK activity. The expression of PKC⑀ and PKC␣ protein in sense and antisense ODN-treated cells was quantitated by densitometry using the image analysis program, NIH Image. PKC⑀ expression in sense ODN-treated cells was given a value of 100%. PMA-stimulated MAPK activity in sense and antisense ODN-treated cells was quantitated using a Molecular Dynamics StormImager, with the MAPK in sense ODN-treated cells represented as 100% activity. The data are representative of duplicate experiments, each performed in triplicate.

c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester
numbers, it is clear that extended incubation of c-N-Ras at pH values 4 or 9 did not inactivate the Ras protein with respect to its ability to exchange guanine nucleotides. We also tested whether Ras-GTP-Mg 2ϩ complexes were stable to these pH conditions. In this experiment, c-N-Ras-[␣-32 P]GTP was preformed in pH 7.4 buffer. The pH was then changed, and excess unlabeled GTP was added. After 1 h under these conditions, the reactions were analyzed for the amount of bound [␣-32 P]GTP. We observed no differences in the off-rates of the labeled GTP at the extreme pH values compared with the control assays performed at pH 7.4 (data not shown).
Plasma membranes from 72-h serum-starved C3H10T1/2 cells were prepared and solubilized with 1% CHAPS as described under "Experimental Procedures." As expected, direct incubation of the solubilized plasma membranes with the GST-RBD fusion protein did not result in detectable amounts of c-N-Ras association with the RBD moiety. This is in agreement with our observations that the c-N-Ras in the plasma membranes are in latent, pre-existing complexes. Upon treatment of the solubilized plasma membranes with pH 4 buffer, we then detected significant amounts of c-N-Ras capable of binding the GST-RBD protein. We observed no increase in the signal upon exchange with GMPPNP, suggesting that all the c-N-Ras in the plasma membranes of serum-starved C3H10T1/2 cells is associated with GTP. As expected, exchanging the solubilized plasma membrane with GDP resulted in no detectable association of c-N-Ras in the GST-RBD pull-down assay. These data support the conclusion that the c-N-Ras in the latent c-N-Ras⅐Raf-1⅐PKC⑀ complex is bound with GTP, even after 72 h of serum starvation. DISCUSSION There is considerable evidence supporting the concept that the components of signaling cascades are organized into ordered signaling modules through their association with scaffolding, adapter, or chaperone proteins. This juxtaposition of effector proteins and their substrates facilitates their tight regulation, specificity of action, and their compartmentalization within the cell into discrete environments. Examples of ordered signaling modules includes the AKAP complex that co-localizes PKA with its effectors and substrates (45,46), the JIP scaffold protein that organizes the components of the c-Jun NH 2 -terminal kinase pathway (47,48), and MP1 that organizes and regulates the interaction of MEK with ERK1 (49). Similarly, proteins that regulate the subcellular localization and organization of Raf-1 have been identified; the association of connector enhancer of KSR (21) with Raf-1 targets the kinase to cell-cell contacts, whereas the association of KSR with Raf-1 is postulated to regulate both Raf-1 activity and its interaction with MEK (18,20,50).
We have generated several pieces of data that support the hypothesis that c-N-Ras, Raf-1, and PKC⑀ exist in a latent ternary complex in serum-starved mouse fibroblasts. First, all the co-immunoprecipitation data are consistent with the exist-FIG. 6. Determination of the guanine nucleotide bound to c-N-Ras within the latent c-N-Ras⅐Raf-1⅐PKC⑀ complex. A, stability of c-N-Ras⅐GST-RBD complexes to increasing NaCl concentrations. Membranes from a rat brain extract were loaded with GMPPNP in the presence of excess EDTA. Mg 2ϩ was added to stabilize the Ras-GMP-PNP complexes. This was then incubated with glutathione-agarose beads precoupled with the GST-RBD fusion protein. Following a 1-h incubation, the glutathione-agarose GST-RBD beads were washed and then incubated with the indicated NaCl concentrations for 1 h at 4°C. The c-N-Ras⅐GST-RBD complexes immobilized to glutathione-agarose beads were washed and analyzed for the presence of N-Ras by Western analysis. B, stability of c-N-Ras⅐GST-RBD complexes to changes in pH. Similar to the experiments described in A except the c-N-Ras⅐GST-RBD immobilized to glutathione-agarose beads was treated with buffers at the indicated pH values for 1 h at 4°C. The beads were then washed and analyzed for the presence of N-Ras by Western analysis. C, stability of recombinant c-N-Ras to changes in pH. c-N-Ras, purified from Escherichia coli, was incubated in GTP-binding buffers containing [␣-32 P]GTP at the indicated pH values. After 30 min at room temperature the reactions were added to quench buffer and subsequently filtered through 0.22-micron nitrocellulose filters. The amount of radioactivity associated with the c-N-Ras was determined by scintillation counting. D, determination of the guanine nucleotides associated with c-N-Ras in the latent c-N-Ras⅐Raf-1⅐PKC⑀ complex. Purified plasma membranes were prepared and solubilized in 1% CHAPS as described previously. The solubilized proteins was separated into 5 equal aliquots (5 g of protein per sample). One aliquot was incubated with the GST-RBD coupled to glutathione-agarose in the absence of any further treatment. The remaining protein samples had their pH lowered to 4 with acetate buffer. After 1 h at 4°C, one sample had glutathioneagarose beads added (without the GST-RBD, indicated by the minus sign) and the pH readjusted to 7.4 with MOPS buffer. A second sample was incubated with the glutathione-agarose beads precoupled with GST-RBD at pH 7.4. The remaining two samples were exchanged with either GMPPNP or GDP in the presence of excess EDTA for 1 h, Mg 2ϩ was added followed by the addition of glutathione-agarose beads precoupled with GST-RBD at pH 7.4. All the samples were washed and analyzed for the presence of c-N-Ras by Western analysis. The data are representative of two separate experiments.
c-Raf-1 in c-N-Ras⅐PKC⑀ Complexes Activated by Phorbol Ester ence of such a ternary complex. Second, the gel filtration analysis confirms that both c-N-Ras and Raf-1 are associated with at least one additional protein of roughly 70 -100 kDa. The PKC⑀ antibody-dependent shift in molecular weight of both c-N-Ras and Raf-1 implicates PKC⑀ as the additional protein within the c-N-Ras⅐Raf-1 complex. These data are further supported by the spatial orientation implicated by the in vitro phosphorylation of Raf-1 by PKC⑀ in c-N-Ras immunoprecipitates. The cumulative data do not completely rule out the possibility that rather than a ternary complex between c-N-Ras, Raf-1, and PKC⑀, there are a series of other complexes that coincidentally move at the same molecular weight upon gel filtration analysis. We feel, however, the number of different complexes that must exist to explain our observations in these terms is very unlikely. Given these considerations, we feel our data support the existence of a ternary, latent complex between c-N-Ras, Raf-1, and PKC⑀ in quiescent, serum-starved C3H10T1/2 cells.
Our data suggest that components of Ras-regulated signaling pathways may be ordered into signal modules prior to activation. Here we demonstrate that, even in quiescent cells, endogenous c-N-Ras, c-Raf-1, and PKC⑀ are found in a latent signaling complex and that c-Raf-1 activity in the complex can be directly stimulated by PMA. This complex, in situ, may therefore represent a physiological target for DAG-mediated MAPK activation. Although PKC␦ was also found to be constitutively associated with membrane-bound c-Raf-1, c-N-Ras did not appear to be a component of this complex. However, this does not completely exclude a role for Ras in PKC␦-regulated c-Raf-1 signaling since the precise role of the four endogenous Ras isoforms (Ha-, N-, Ki-Ras A, and B) in Raf-1 regulation, in vivo, has not been fully determined. These data, however, might suggest that the regulation of Raf-1 activity by PKC␦ and PKC⑀ may be in response to distinct agonists. Similarly, the regulation of Raf-1 activity by distinct Ras isoforms may also occur in response to distinct agonists or biological events, in order to elicit specific responses such as proliferation, migration, or differentiation.
The PKC⑀ antisense data presented here would indicate that endogenous PKC⑀ (but not PKC␦) is required for PMA (and consequently DAG)-mediated stimulation of the MAPK cascade in C3H10T1/2 cells. The data also provide further evidence regarding the convergence of the Ras and PKC pathways in c-Raf-1 regulation. More importantly, signaling modules containing components from supposedly distinct signal transduction pathways and common effector proteins, such as c-Raf-1, have the potential to be rapidly activated and regulated by a diverse range of stimuli.
The role of Ras in the activation of Raf-1 by PKC has proven to be controversial, partly through the contradictory results obtained using dominant-negative (S17N)Ras to study the role of Ras in this process. Several studies suggested that PKCmediated activation of the MAPK cascade was independent of Ras activation, since PKC-stimulated MAPK activity appeared to be insensitive to dominant-negative (S17N)Ras (29,51). Fucini et al. (52) recently demonstrated that the activation of MAPK by PMA, unlike insulin, appeared to occur in the absence of any significant Ras activation. By using the Rasbinding domain of Raf-1 to trap newly formed Ras-GTP, they found that PMA did not result in a significant level of Ras activation, unlike the efficient trapping of Ras-GTP by the RBD domain in insulin-stimulated cells. This contrasted the findings of Barnard et al. (7) and Marais et al. (34) who demonstrated that the association between Ras and Raf-1 was required for PKC to activate Raf-1. These reports, however, can now be reconciled by the possibility that PMA stimulates MAPK in the context of activating the PKC⑀ in the inactive, latent c-N-Ras⅐Raf-1⅐PKC⑀ complexes. The presence of preformed c-N-Ras-GTP⅐Raf-1 complexes would preclude the requirement for the generation of de novo c-Ras-GTP (which is blocked by (S17N)Ras) in response to PMA stimulation. Since Ras effector binding is a mutually exclusive process, where Ras can only associate with one effector when active, the c-N-Ras-GTP present in the inactive c-N-Ras⅐Raf-1⅐PKC⑀ signaling complex complex would be resistant to trapping by the RBD domain. This was clearly demonstrated by our ability to destabilize the c-N-Ras⅐Raf-1 complex at pH 4.0 and then detect Ras-GTP using the GST-RBD pull-down assay. Interestingly, a specific role for the c-N-Ras isoform in the regulation of the MAPK cascade by phorbol ester was also suggested by Malumbres and Pellicer (53) who reported that cells deficient in N-Ras had a decreased response to PMA. Moreover, the critical role for PKC⑀ in the activation of MAPK by PMA supports the hypothesis that these inactive c-N-Ras⅐Raf-1⅐PKC⑀ complexes are unique, physiological targets for phorbol ester and therefore, putatively, diacylglycerol-mediated MAPK activation.
This is the first report to document a functional complex of Ras-GTP-effector complex in quiescent cells. In fact, our observation that there exists a pool of complexed Ras-GTP in quiescent cells highlights the deficiencies in the current tools to make an accurate assessment of Ras-GTP levels. This complex, because of its stability, does not bind to the GST-RBD in the now commonplace pull-down assays. In addition, as demonstrated by Marias et al. (34), this preformed complex would not be blocked by expression of the typical dominant-negative Ras Asn-17 protein, which ties up exchange factors and prevents the generation of newly formed Ras-GTP. The identification of a latent c-N-Ras⅐GTP-dependent signaling complex raises a number of questions relative to the current Ras paradigm. Is it the generation of new Ras-GTP that is important or possibly the cycling of Ras with successive GTP molecules? It is not clear from published work whether the generation of a single molecule of Ras-GTP, in this case c-N-Ras GTP, results in the activation of a single Raf-1 molecule or many Raf-1 molecules. It will be interesting to examine the relationship between this latent c-N-Ras-GTP⅐Raf-1⅐PKC⑀ and the generation of newly formed Ras-GTP in response to extracellular factors.