Phosphorylation of serine 43 is not required for inhibition of c-Raf kinase by the cAMP-dependent protein kinase.

The activity of the serine/threonine kinase c-Raf (Raf) is inhibited by increased intracellular cAMP. This is believed to require phosphorylation with the cAMP-dependent protein kinase (PKA), although the mechanism by which PKA inhibits Raf is controversial. We investigated the requirement for PKA phosphorylation using Raf mutants expressed in HEK293 or NIH 3T3 cells. Phosphopeptide mapping of (32)P-labeled Raf (WT) or a mutant lacking a putative PKA phosphorylation site (serine to alanine, S43A) confirmed that serine 43 (Ser(43)) was the major cAMP (forskolin)-stimulated phosphorylation site in vivo. Interestingly, the EGF-stimulated Raf kinase activity of the S43A mutant was inhibited by forskolin equivalently to that of the WT Raf. Forskolin also inhibited the activation of an N-terminal deletion mutant Delta5-50 Raf completely lacking this phosphorylation site. Although WT Raf was phosphorylated by PKA, phosphorylation did not inhibit Raf catalytic activity in vitro, nor did forskolin treatment inhibit the activity of an N-terminally truncated Raf protein (Raf 22W) or a full-length Raf protein (Raf-CAAX) expressed in NIH 3T3 cells. In contrast, forskolin inhibited the EGF-dependent activation of a Raf isoform (B-Raf), lacking an analogous phosphorylation site to Ser(43). Thus, these results demonstrate that PKA exerts its inhibitory effects independently of direct Raf phosphorylation and suggests instead that PKA prevents an event required for the EGF-dependent activation of Raf.

The mitogen-activated protein kinase (MAPK) 1 cascade is a conserved phosphorylation pathway involved in the transduction of growth and differentiation signals in both simple and complex organisms (reviewed in Ref. 1). In the now established paradigm, peptide growth factors bind to their receptor tyrosine kinases, increasing intracellular protein tyrosine phosphorylation and stimulating the exchange of GDP for GTP in small GTP-binding proteins such as Ras (reviewed in Ref. 2). Raf, a 74-kDa serine/threonine kinase, binds to Ras-GTP at the plasma membrane, thus providing an integral link between the receptor-mediated events and activation of the MAPK cascade (3,4). The activation of Raf is a complex process involving membrane recruitment, interactions with the proteins 14-3-3 (5), hsp90 (6), and regulation by Src family tyrosine kinases (7), protein kinase C␣ (8), the kinase suppressor of Ras (Ksr) (9), and most recently p21-activated kinase (PAK) (10) (for reviews, see Refs. 11 and 12). In addition, Raf activation may also be mediated by cytoskeletal factors that lead to the enhancement Raf but not Ras activity in cells (13). Once activated, Raf phosphorylates and activates MAPK kinase (14,15) (MAPKK or MEK (16)), which in turn phosphorylates MAPK (17,18) as part of the sequence of events leading to the activation of MAPK in mammalian cells (1).
Compounds that increase intracellular cAMP (e.g. glucagon, epinephrine, prostaglandins) are known to oppose the activation of the MAPK cascade in many cell types (for a review, see Ref. 19). This inhibitory effect requires activation of the cAMPdependent protein kinase (PKA) (20), although the actual mechanism by which this is achieved is controversial. Neither MAPK kinase nor MAPK are phosphorylated by PKA, and elevating cAMP does not inhibit receptor tyrosine phosphorylation, phosphatidylinositol turnover, or activation of Ras-GTP in response to growth factor treatment (19 -23). Instead, considerable effort has focused on the serine/threonine kinase c-Raf-1 (c-Raf) as the target of inhibition, and c-Raf is consistently inhibited in cells after elevation of cAMP. PKA phosphorylates substrates containing tandem basic residues upstream of a phosphorylatable serine or threonine residue (24,25), and c-Raf contains a consensus PKA phosphorylation site at serine 43 (RRAS). Serine 43 has been shown to be phosphorylated in vivo (22,26), and although the phosphorylation of this amino acid does not inhibit the catalytic activity of c-Raf (26), it has been proposed to disrupt normal Ras/Raf association and c-Raf activation in vivo (22).
Although this model has been widely accepted, the requirement for Ser 43 phosphorylation has not been directly tested in vivo, and alternative models have been proposed. The studies of Kolch and co-workers (27,28) suggest that the catalytic activity of c-Raf is directly inhibited by PKA and propose that phosphorylation of Ser 621 within the catalytic domain of c-Raf is responsible for this effect. However, this hypothesis has been difficult to test because the autophosphorylation of Ser 621 is required for Raf kinase activity (5), and increased phosphorylation of Ser 621 in response to cAMP has not been demonstrated.
The effect of cAMP on other isoforms of c-Raf has also been examined. However, in comparison with the consistent inhibition found with c-Raf, the influence of cAMP on the related enzyme B-Raf is more complex. There are examples of inhibition (29,30), activation (31), or no effect (32), of cAMP on B-RAF which may depend on the stimulus, the cell type, or the culture conditions (33,34). B-Raf lacks an analogous Ser 43 phosphorylation as found in c-Raf but does contain a similar Ser 621 phosphorylation site (35). In contrast to some of the studies on c-Raf (27,28), B-Raf does not appear to be directly inhibited by PKA phosphorylation (32). The third Raf isoform, A-Raf, is not widely expressed in cells and may be resistant to the inhibitory effects of cAMP (36).
The fact that no single consensus model appears sufficient to explain the regulation of Raf (c-or B-Raf) has led to the search for other targets. The Rap proteins (Rap 1a/b) are phosphorylated by PKA (37), can bind Raf, and block Ras signaling when overexpressed in cells (38,39). Thus, these proteins were originally proposed to be candidates for mediating the cAMP-dependent inhibition of Raf. However, recent studies suggest that Rap is not required for inhibition of MAPK signaling (40) and demonstrate that Rap 1b is activated by cAMP (41). This appears to occur independently of the action of PKA (42) and may play a mitogenic role (43) through the cAMP-dependent activation of B-Raf (31).
Hence, the studies described herein were designed to resolve some of the remaining questions concerning Raf regulation by cAMP. Importantly, we find that deletion of the major PKA phosphorylation site (Ser 43 ) does not impede the cAMP-dependent inhibition of Raf in vivo. Furthermore, we find no evidence for a cAMP-stimulated increase in the phosphorylation of Ser 621 in vivo or a direct inhibitory effect of PKA on Raf either in vitro or on constitutively activated Raf kinases (Raf 22W, Raf-CAAX) expressed in vivo. Finally, examination of an alternative isoform (B-Raf) lacking an analogous Ser 43 phosphorylation site demonstrated that forskolin prevented the EGF-dependent activation of this enzyme, consistent with PKA interrupting an event required for Raf activation rather than inhibiting Raf itself. Cell Cultures-Human embryonic kidney cells (HEK 293) were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and penicillin/streptomycin on 60-or 150-mm plates in a humidified, 5% CO 2 atmosphere. 24 h prior to the experiments, the medium was removed and replaced with DMEM with 0.1% FBS and penicillin/streptomycin. The rat liver epithelial cells (GN4) were grown and harvested as described earlier (44). Sf9 cells were grown and infected with baculovirus as described in Ref. 7.

Materials
Transient Transfection of Cells-Transient transfection of the cells was performed using LipofectAMINE according to the manufacturer's suggested protocol (Life Technologies, Inc.). Briefly, 2 g of plasmid and 2 g of carrier DNA, usually empty pCDNA3 were transfected with 8 l of LipofectAMINE in serum-free DMEM into cells grown on 60-mm cell culture plates for 48 h. For transient transfection of 100-mm plates, 4 g of plasmid and carrier DNA was used. For transient transfection of 150-mm plates, 8 g of plasmid and 8 g of carrier DNA in 100 l of LipofectAMINE were used. Cells grown to 50% confluence were washed once with 1ϫ PBS and once with serum-free DMEM. DNA/Lipo-fectAMINE/DMEM was placed on cells and incubated for 5 h at 37°C. DMEM with 20% FBS and penicillin/streptomycin was then added to the plates and incubated for an additional 24 h. Cells were serumstarved for 24 h with DMEM with 0.1% FBS and penicillin/streptomycin prior to the experiment.
Cell Lysate Preparation-Following drug treatment, cells were washed twice with cold PBS and harvested in 1 ml of ice-cold lysis buffer (20 mM Tris, pH 7.5, 1% Triton, 10% (v/v) glycerol, 137 mM NaCl, 2 mM EDTA, 5 g/ml leupeptin, 10 nM calyculin A, 250 M phenylmethylsulfonyl fluoride, and 150 M sodium orthovanadate). The cell lysates were centrifuged at 15,000 rpm for 10 min, and the resultant supernatants were transferred to a new microcentrifuge tube. Protein concentration of the cell lysates was determined by Coomassie protein assay reagent (Pierce). The same procedure was used to isolate Raf from baculovirus.
Measurement of c-Raf-1 Kinase Activity-2 g of anti-FLAG antibody was added to 1 mg of clarified cell lysate. Samples were rotated for 1.5 h at 4°C. 20 l of protein A beads in PBS were then added to the supernatants and rotated end over end for an additional 40 min at 4°C. The immune complexes were placed on ice and washed twice in cold immunoprecipitation buffer and twice in cold PBS. The remaining supernatant was removed from the protein A beads with a Hamilton syringe. Raf activity was measured by using a coupled enzyme assay containing MEK and extracellular signal-regulated kinase-2 (Erk-2) and quantitating the incorporation of radioactive phosphate into myelin basic protein as described below. The immunocomplexes were incubated in 10 l of a cold ATP mixture (20 mM Hepes, pH 7.3, 10 mM ␤-glycerophosphate, 1.33 mM EGTA, 0.15 mM Na 3 VO 4 , 0.5 mM dithiothreitol, 10 mM MgCl 2 , 0.1 mM ATP, and 0.1 mg/ml bovine serum albumin) and 0.5 g of recombinant MEK for 10 min at 30°C. 1.25 g of recombinant extracellular signal-regulated kinase-2 was added to the samples and incubated for an additional 10 min at 30°C. Finally, 20 l of a radioactive ATP mixture (2 Ci of [␥-32 P]ATP, 10 g of myelin basic protein, 20 mM Hepes, pH 7.3, 10 mM ␤-glycerophosphate, 1.33 mM EGTA, 0.15 mM Na 3 VO 4 , 0.5 mM dithiothreitol, 10 mM MgCl 2 , 0.1 mg/ml bovine serum albumin) was added and incubated for 10 min at 30°C. The reaction was terminated with the addition of 20 l of 100 mM EDTA. Samples were centrifuged at 12,000 ϫ g for 1 min to pellet the beads, and 40 l of the reaction mixture was spotted onto squares (2 ϫ 2 cm) of P-81 paper. P-81 papers were washed in 75 mM phosphoric acid five times (5-10 min), washed once briefly in 100% ethanol, and airdried. The P-81 papers were counted in a Beckman LS 6500 scintillation counter in 0.5 ml of Scintisafe (Fisher) scintillation fluid. Using these conditions, the assay of Raf activity was determined to be linear with respect to time and Raf concentration.
Phosphopeptide Mapping of c-Raf-1 Kinase-293 cells were transiently transfected with WT or S43A Raf-FLAG and grown to confluence in 150-mm plates. Prior to the addition of [ 32 P]orthophosphate, cells were incubated in 12 ml of phosphate-free, 0.1% FBS DMEM for 30 min. 3.0 mCi of [ 32 P]orthophosphate was then added to each plate for 2.5 h. Following labeling, plates were treated with 25 M forskolin in Me 2 SO for 15 min or left untreated. Cells were harvested and Raf-FLAG was immunoprecipitated with 10 g of FLAG antibody and 50 l of protein A beads in PBS. Beads were washed as above. Raf-FLAG was removed from the beads by heating twice in 100 l of sample buffer. Phosphorylated Raf-FLAG was applied to 10% SDS-PAGE, and the gel was fixed in a 25% isopropyl alcohol, 10% acetic acid solution. The gel was then washed three times in water and once in 0.1 M NH 4 HCO 3 , pH 8.5. The gel was dried, and autoradiography was performed to identify radioactive bands. The band corresponding to Raf (74 kDa) was excised from the gel and swelled in a solution containing 100 l of 0.1 M NH 4 HCO 3 , pH 8.5, and 20 g of trypsin and incubated at 37°C overnight. The supernatant was removed and used for two-dimensional phosphopeptide mapping as described previously (45). Peptides were resolved electrophoretically (100 V/14 min) using pH 1.9 buffer in one dimension and chromatographically with isobutyric acid buffer in the second dimension. Autoradiography was performed to visualize the radioactive peptides. In other experiments, synthetic peptides containing the Ser 43 (GYQRRASDDGKLTD) or Ser 621 (LPKINRSASEPSLHRAA) phosphorylation sites were synthesized by Dr. David G. Klapper (University of North Carolina, Chapel Hill). These peptides were phosphorylated with the purified PKA catalytic subunit (New England Biolabs) in a buffer containing 20 mM Tris, pH 7.5, 10 mM MgCl 2 , and 25 M ATP (2 Ci of [␥-32 P]ATP). Following phosphorylation, the peptides were digested with trypsin (0.2 g) in 0.1 M NH 4 HCO 3 , pH 8.5, and acidified with trifluoroacetic acid, and the radioactive peptides were purified by reverse high pressure liquid chromatography on a Vydac C18 (4.6 ϫ 100-mm) column with a linear gradient of 0 -60% acetonitrile in 0.1% trifluoroacetic acid. The amount of Raf peptide phosphorylation was determined by volume analysis of densitometric scans (ImageQuant, Molecular Dynamics, Inc., Sunnyvale, CA).
Immunoprecipitations and Immunoblotting-For immunoprecipitation, 1 mg of cell lysate was incubated with antibody (␣-FLAG or ␣-Raf) for 2 h at 4°C. 20 l of protein A/G-agarose beads were then added and incubated for an additional 1 h at 4°C. Immune complexes were pelleted by centrifugation (12,000 ϫ g for 1 min), washed three times in lysis buffer, and boiled in SDS-PAGE sample buffer. The resulting supernatants were separated by SDS-PAGE, transferred to Immobilon membrane (Millipore Corp.), and incubated with appropriate antibodies. Immunoblots were visualized by ECL according to the manufacturer's instructions (NEN Life Science Products).
Phosphorylation of Raf with PKA in Vitro-Raf was immunoprecipitated and incubated with the purified, active catalytic subunit of PKA (New England Biolabs) in the presence of [␥-32 P]ATP and Mg 2ϩ as described above. Immunoprecipitates were washed with 1ϫ PBS, and Raf kinase was then assayed for activity as described above or analyzed for incorporation of 32 P by SDS-PAGE, autoradiography, and scintillation counting of the radioactive Raf protein. When immunoprecipitated from baculovirus, the amount of Raf protein was estimated by Coomassie staining and comparison with known amounts of bovine serum albumin (0 -25 g) applied to the same SDS-PAGE.

Expression and Phosphorylation of Wild Type and Mutant
Raf Kinases in HEK 293 Cells-c-Raf (Raf) contains an optimal consensus PKA phosphorylation site (RRAS) at Ser 43 and is phosphorylated on this amino acid in vivo (22,26). To investigate the importance of Ser 43 phosphorylation in mediating the inhibition of Raf by cAMP, the phosphorylation and regulation of a Raf mutant protein containing an alanine substitution at this position (S43A) was examined. HEK 293 (HEK) cells were transiently transfected with the cDNA for WT or S43A Raf containing a FLAG epitope tag. The WT and S43A Raf proteins were expressed at similar levels as determined by Western blotting of cell lysates for the FLAG epitope tag (data not shown). HEK cells were metabolically labeled with [ 32 P]orthophosphate, and the phosphorylation of the WT and S43A Raf proteins was examined by two-dimensional tryptic phosphopeptide mapping (45). From untreated WT or S43A Raf cells, two major phosphopeptides (I, II) were obtained after tryptic digestion of Raf (Fig. 1, A-D). One of these peptides (II), co-migrated with a synthetic peptide corresponding to the tryptic peptide containing the Ser 621 autophosphorylation site as described earlier (26) (5) (see upper right-hand corner of Fig.  1E). Treatment of HEK cells with forskolin to increase intracellular cAMP resulted in the appearance of an additional phosphorylated peptide (III) in the WT Raf protein (Fig. 1B). This phosphopeptide (III) co-migrated with a synthetic tryptic peptide corresponding to the serine 43 phosphorylation site (RASDDGK) (Fig. 1E) and was not observed in tryptic digests of untreated WT Raf (Fig. 1A). This phosphopeptide was also absent from tryptic digests of the S43A Raf after forskolin treatment, thus demonstrating the absence of this phosphorylation site in vivo (Fig. 1, C and D). Relative to the peptide (I) that was constitutively phosphorylated (see legend to Fig. 1), forskolin did not increase the phosphorylation of the Ser 621 peptide (II) or significantly change the phosphorylation of other peptides in Raf (Fig. 1, B and D). Thus, these results confirm that Ser 43 is the major forskolin-stimulated phosphorylation site in Raf.
Regulation of Wild Type and S43A Raf by EGF and cAMP-The regulation of the WT and S43A Raf kinases expressed in HEK cells was compared. HEK cells expressing either the WT or S43A Raf were serum-deprived, and the WT or S43 Raf proteins were immunoprecipitated with antibodies to the epitope tag (FLAG) and immunoblotted ( Fig. 2A) or assayed for Raf kinase activity (Fig. 2B) as described under "Experimental Procedures." As shown in Fig. 2A D), and the Raf proteins were immunoprecipitated using anti-FLAG antibody. After SDS-PAGE isolation of the phosphorylated Raf, "in gel" trypsin digestion was performed, and the Raf peptides were analyzed by two-dimensional phosphopeptide analysis and autoradiography. E, synthetic peptides containing the serine 43 and serine 621 residues were phosphorylated with PKA and [␥-32 P]ATP as described under "Experimental Procedures," digested with trypsin, and similarly analyzed. The x axis represents separation by electrophoresis; the y axis represents separation by chromatography as described earlier (45 2B). In parallel with an increase in Raf activity, the migration of both the WT and S43A Raf proteins decreased slightly on SDS-PAGE ( Fig. 2A). The time course and extent of activation of the WT and S43A Raf kinases by EGF was indistinguishable; the peak of Raf activity typically occurred after 5 min of EGF addition and rapidly declined to basal levels thereafter (data not shown).
The effect of increasing intracellular cAMP on the activity of the WT and S43A Raf kinases was determined. Acute forskolin treatment (5-10 min) resulted in greater than 90% inhibition of the EGF-stimulated Raf kinase activity in HEK cells expressing WT Raf (Fig. 2B). Forskolin treatment also inhibited the activation of the S43A Raf kinase to an equivalent extent. Concomitant with the inhibition of Raf activity, the EGF-stimulated shift on SDS-PAGE was abolished (Fig. 2B). The influence of forskolin was independent of the length of exposure to EGF; hence, forskolin treatment did not simply delay the activation of Raf (data not shown). These results indicated that phosphorylation of Ser 43 was unnecessary for inhibition by cAMP. However, to further confirm this, we compared the regulation of a Raf kinase mutant lacking amino acids 5-50 (⌬5-50 Raf) and the entire consensus PKA phosphorylation site. The ⌬5-50 Raf was expressed in HEK cells and treated with EGF as described above. Immunoprecipitation and immunoblotting of lysates for this protein demonstrated that the ⌬5-50 Raf protein migrated faster on SDS-PAGE, consistent with its lower molecular weight ( Fig. 2A). EGF treatment of HEK cells resulted in approximately a 10 -15-fold activation of the ⌬5-50 Raf kinase, similar to that observed with the WT or the S43A Raf kinases (Fig. 2B). Like the results obtained with the WT and S43A Raf, forskolin treatment inhibited the EGF-dependent activation of the ⌬5-50 Raf kinase and the EGFstimulated shift in migration on SDS-PAGE ( Fig. 2A). Thus, these results unequivocally demonstrate that phosphorylation of Ser 43 is not required for the inhibition of Raf by cAMP in vivo. (27,28). Since our studies suggested that phosphorylation of Ser 43 was unnecessary for inhibition of Raf in vivo, we tested whether phosphorylation by PKA resulted in the inhibition of Raf kinase activity in vitro. Active Raf kinase was obtained from Sf9 cells coinfected with Raf and Src baculovirus as shown earlier (7). Incubation of active Raf with the catalytic subunit of PKA and Mg-ATP resulted in a time-dependent increase in the phosphorylation of this protein (Fig. 3A). However, despite stoichiometric phosphorylation (ϳ0.5-1.0 mol/mol) of Raf by PKA in vitro, no detectable loss of Raf kinase activity was observed in these samples (Fig. 3B). Because the Src-activated Raf kinase may be mechanistically different from Raf obtained from EGF-treated cells, we compared the ability of PKA to inhibit the activity of either the WT or the S43A Raf kinases immunoprecipitated from HEK cells. Incubation of Raf from serum-starved or EGFstimulated cells with PKA resulted in phosphorylation of the WT enzyme as expected (data not shown). Again, neither the WT nor S43A Raf kinase activity was inhibited by incubation with PKA under these conditions (Fig. 3, C and D). Hence, these results indicate that PKA does not directly phosphorylate and inhibit the catalytic activity of Raf in vitro.

PKA Does Not Inhibit the Activity of the Catalytic Domain of Raf Expressed in NIH 3T3 Cells (Raf 22W) or Raf-CAAX-To
further determine the effect of PKA on Raf, we examined the regulation of the catalytic domain of Raf kinase expressed in vivo. Raf kinase was immunoprecipitated from NIH 3T3 cells stably transfected with vector (pZip) or a cDNA encoding a constitutively activated Raf protein (Raf 22W, amino acids 305-648). This constitutive activity is due to relief of N-terminal autoinhibition (12), and the Raf 22W cells contained a high level of Raf activity in the absence of serum as reported earlier (14,46). Incubation of the serum-starved Raf 22W cells with forskolin only slightly inhibited the Raf kinase activity in these cells (Fig. 4). In contrast, forskolin completely inhibited the EGF-dependent activation of the endogenous Raf in the vector control NIH 3T3 (pZip) cells as expected (data not shown). Therefore, the catalytic subunit of Raf itself appears to be resistant to inhibition by cAMP in vivo.
Because earlier studies suggested that PKA phosphorylation of Raf on Ser 43 decreased Raf/Ras association (22), we determined the influence of PKA on a Raf mutant that did not require Ras binding for activity. The addition of a farnesyl group to Raf (Raf-CAAX) artificially targets Raf to the plasma membrane; this has been proposed to mimic Ras binding and results in a constitutively active Raf kinase, ultimately resulting in increased cell transformation (47,48). HEK cells were transiently transfected with a hemagglutinin (HA)-tagged Raf-CAAX or wild type HA-tagged Raf, and the Raf kinases were immunoprecipitated (␣-HA tag) and assayed for Raf kinase activity. As expected, Raf-CAAX was constitutively active in the absence of EGF treatment (Fig. 5). EGF treatment did not increase the activity of this kinase, suggesting that Raf-CAAX was already fully activated in these cells. However, a HAtagged WT Raf was activated by EGF to a similar extent as that found with the FLAG-tagged Raf kinases as described above (compare with Fig. 2). The effect of forskolin treatment on the Raf kinases in these cells was compared. As shown in Fig. 5, forskolin resulted in almost a complete inhibition of the WT Raf but was without effect on the activity of the Raf-CAAX kinase.
Regulation of B-and c-Raf by cAMP in Liver Epithelial Cells-Since our results suggested that PKA was not inhibiting c-Raf directly, we compared the effects of cAMP on a related Raf isoform, B-Raf. Although highly homologous to c-Raf throughout the catalytic domain, B-Raf has a longer N-terminal region and lacks an analogous phosphorylation site to serine 43 in c-Raf (35). Rat liver epithelial cells (GN4) express both B-and c-Raf as demonstrated by immunoblotting for these enzymes (data not shown). Using polyclonal antibodies specific for each isoform, we immunoprecipitated B-and c-Raf from serum-starved or EGF-treated GN4 cells and measured Raf kinase activity as described above. In serum-starved cells, B-Raf had considerably higher basal kinase activity than that found with c-Raf (Fig. 6). As expected, EGF stimulated an increase in the kinase activity of both enzymes; however, c-Raf activity was increased to a greater extent than B-Raf. Treatment of GN4 cells with forskolin almost completely inhibited the EGF-stimulated c-Raf kinase and B-Raf kinase activity. In contrast, forskolin did not effect the basal activity of either c-or B-Raf, suggesting that only the EGF-dependent activation was subject to regulation by this treatment (Fig. 6). Forskolin treatment alone did not increase the activity of B-Raf in these cells (data not shown). Thus, these results demonstrate that cAMP prevents the activation of two distinct Raf isoforms in GN4 cells. rylated Raf protein was identified by autoradiography and excised, and the amount of radioactivity was quantitated in a liquid scintillation counter. B, for Raf kinase activity, PKA phosphorylation was first terminated with the addition of 100 M PKI inhibitor peptide, and Raf kinase activity was then measured using the coupled enzyme assay described under "Experimental Procedures." Results are reported as pmol/min/ml. C and D, HEK 293 cells expressing WT (C) or S43A (D) Raf were serum-starved and left untreated as controls or incubated with 40 ng/ml of EGF for 5 min. Cells were harvested, and Raf was immunoprecipitated with anti-FLAG antibody, and the immunoprecipitates were incubated with 4 or 0.4 units of PKA activity and analyzed for activity as described in B. Raf activity is reported as pmol/min/ml and represents the mean Ϯ S.E. of duplicate samples. Shown are representative results of three experiments.

DISCUSSION
Inhibition of the MAPK cascade by cAMP and the consequential activation of PKA is a widely observed phenomenon in many different cell types (19). However, despite this observation, the mechanism by which this occurs is poorly understood and remains controversial. Of particular interest is whether phosphorylation of Raf is required to explain the inhibition of this kinase in vivo. Previous studies have suggested that phosphorylation of Ser 43 is required for Raf inhibition (22,49); however, a direct test of this hypothesis has not been performed in vivo. By examining two different mutants of Raf (S43A or ⌬5-50 Raf), our studies now clearly demonstrate that deletion of this region of Raf is dispensable for normal activation by EGF or for inhibition by cAMP. This is consistent with studies showing that the primary Ras binding domain of Raf includes amino acids 51-131 (see Ref. 12 and references therein) and would imply that Ser 43 is outside the region required for Raf/ Ras binding in vivo. These conclusions are further supported by our findings with B-Raf showing that cAMP inhibits activation of a related isoform lacking an analogous Ser 43 phosphorylation site. Although our phosphopeptide mapping results confirm that Ser 43 is the major site of cAMP-stimulated phosphorylation in c-Raf in vivo (26), the finding that removing Ser 43 does not affect Raf regulation indicates that phosphorylation of this site remains an enigma.
Previously it was suggested that phosphorylation of Raf on Ser 621 was responsible for inhibition of c-Raf kinase in vivo (28). Although the authors demonstrated that PKA phosphorylated Raf on Ser 621 in vitro, evidence for the forskolin or cAMP-stimulated phosphorylation of Ser 621 in vivo was not presented. In our phosphopeptide mapping studies, we observed two constitutively phosphorylated peptides in Raf, one of which co-migrated with the synthetic Ser 621 tryptic peptide, whereas the other most likely corresponds to the Ser 259 autophosphorylation peptide as shown earlier (26). We found no evidence for increased phosphorylation of Ser 621 in response to forskolin treatment of cells. Instead, this peptide was phosphorylated equally in both untreated and forskolin-treated cells, consistent with this amino acid being a major autophosphorylation site in vivo (26) and arguing against Ser 621 as a valid PKA phosphorylation site in vivo. In addition, compared with the Ser 43 peptide, the Ser 621 peptide lacks the optimal pair of basic residues (RSAS) and is a poor substrate for PKA in vitro. Phosphorylation of the Ser 621 peptide occurred at a rate approximately 100 times lower than that observed with the Ser 43 peptide in vitro. 2 Instead, Ser 621 is efficiently phosphorylated by the AMP-activated protein kinase (50). In other studies, phosphorylation of Ser 621 has been shown to be required for both 14-3-3 binding and Raf activity (5). Hence, it seems unlikely that a phosphorylation site that is essential for Raf activity would also be required for inhibition as suggested by Kolch and co-workers (28). These observations combined with our in vivo phosphopeptide mapping studies strongly suggest that phosphorylation of Ser 621 does not mediate the cAMP-dependent inhibition of Raf kinase in vivo.
Although our data demonstrated that phosphorylation of Ser 43 was not required for the inhibition of Raf kinase in vivo, PKA might still prevent Raf/Ras association at the plasma membrane as suggested earlier (22). Our experiments with Raf-CAAX were designed to test this hypothesis, and, consistent with this model, we find that this protein is completely resistant to inhibition by cAMP. Although these results may argue that the primary effect of PKA is to disrupt Ras/Raf association, there are other potential interpretations of these results. For instance, once activated, Raf may be resistant to the effects of PKA. Our results with the PKA phosphorylation of active Raf in vitro or forskolin treatment of the Raf 22W cells would support this model. Using either the baculovirus-expressed Raf, the EGF-activated WT, or S43A Raf from HEK cells, the truncated and active form of Raf expressed in NIH 3T3 cells or Raf-CAAX, we found no evidence for the direct inhibition of Raf kinase by PKA either in vitro or in vivo. In this regard, our results are in agreement with the studies of Whitehurst et al. (51), suggesting that the catalytic subunit of c-Raf is not directly inhibited by PKA.
Instead, our findings would be consistent with cAMP inhibiting an event required for the activation of both B-and c-Raf. This hypothesis would be in agreement with earlier studies showing that cAMP inhibited the Ras-dependent activation of B-Raf (29, 30) and would be supported by our results demon-2 M. Sidovar, unpublished observations. strating that forskolin prevents the EGF-dependent activation of both B-and c-Raf in GN4 cells. The lack of effect of forskolin on the basal activity of either B-or c-Raf in GN4 cells reiterates our findings showing that PKA does not directly inhibit the catalytic activity of Raf. Because the catalytic regions of both B-Raf and c-Raf are highly homologous (35) and because other recent studies have found that B-Raf is resistant to inhibition by PKA (32), we believe argues against a direct effect of PKA on the catalytic domain of Raf. Although these authors (32) did see a partial inhibitory effect of PKA on c-Raf (less than 50%), we do not believe that this incomplete effect accounts for the greater than 90% inhibition of c-Raf activity that we observed in vivo.
Additional targets that were not investigated in this study include the Rap proteins. Although these proteins are phosphorylated by PKA in vivo (37) and can physically antagonize the binding of c-Raf to Ras (52), the most recent studies suggest that PKA phosphorylation may diminish the inhibitory potential of these proteins (53). Furthermore, considerable evidence suggests that cAMP-dependent activation of Rap results in the activation of B-Raf (43) (31). Although we did not observe a stimulatory effect of cAMP on B-Raf in the GN4 cells, 3 this may be explained by cell type differences in the expression of Rap as described by others (34). Thus, these findings, combined with the recent studies showing that Rap is not required for the inhibition of MAPK signaling (40), suggest that Rap proteins do not mediate the inhibition of Raf observed in our studies.
So what might explain the inhibitory effects of cAMP on Raf? Recent studies have shown that PAK3 can function as a Raf kinase, thus contributing to the activation of Raf by growth factors and other stimuli (10,54). Interestingly, studies from Howe et al. 4 show that PAK contains a consensus PKA phosphorylation site and find that PAK is phosphorylated and inhibited in response to forskolin treatment of cells. This observation would be consistent with our findings suggesting that PKA phosphorylation of Raf itself is insufficient to explain the inhibition observed in vivo; however, whether or not the phosphorylation of PAK by PKA is required to account for this regulation remains to be established.