Ras-mutant Cancer Cells Display B-Raf Binding to Ras That Activates Extracellular Signal-regulated Kinase and Is Inhibited by Protein Kinase A Phosphorylation*

Background: How the cAMP-dependent protein kinase PKA regulates B-Raf binding to Ras is not known. Results: PKA inhibits the binding of B-Raf to active Ras via phosphorylation of serine 365 in B-Raf. Conclusion: B-Raf can participate in the Ras-dependent activation of ERK in Ras-mutant cancers, and this is inhibited by PKA. Significance: PKA can block Ras binding to both Raf isoforms C-Raf and B-Raf. The small G protein Ras regulates proliferation through activation of the mitogen-activated protein (MAP) kinase (ERK) cascade. The first step of Ras-dependent activation of ERK signaling is Ras binding to members of the Raf family of MAP kinase kinase kinases, C-Raf and B-Raf. Recently, it has been reported that in melanoma cells harboring oncogenic Ras mutations, B-Raf does not bind to Ras and does not contribute to basal ERK activation. For other types of Ras-mutant tumors, the relative contributions of C-Raf and B-Raf are not known. We examined non-melanoma cancer cell lines containing oncogenic Ras mutations and express both C-Raf and B-Raf isoforms, including the lung cancer cell line H1299 cells. Both B-Raf and C-Raf were constitutively bound to oncogenic Ras and contributed to Ras-dependent ERK activation. Ras binding to B-Raf and C-Raf were both subject to inhibition by the cAMP-dependent protein kinase PKA. cAMP inhibited the growth of H1299 cells and Ras-dependent ERK activation via PKA. PKA inhibited the binding of Ras to both C-Raf and B-Raf through phosphorylations of C-Raf at Ser-259 and B-Raf at Ser-365, respectively. These studies demonstrate that in non-melanocytic Ras-mutant cancer cells, Ras signaling to B-Raf is a significant contributor to ERK activation and that the B-Raf pathway, like that of C-Raf, is a target for inhibition by PKA. We suggest that cAMP and hormones coupled to cAMP may prove useful in dampening the effects of oncogenic Ras in non-melanocytic cancer cells through PKA-dependent actions on B-Raf as well as C-Raf.

Ras-dependent activation of ERKs triggers cell proliferation in both normal and malignant cells. Many human tumors display constitutive activation of the MAP kinase cascade that dictates their high proliferation rate. This activation is commonly achieved by mutations in the family of Ras GTPase that maintain Ras in a basally active (GTP-loaded) state. Indeed, oncogenic Ras mutations are present in nearly 30% of human tumors (1,2). Not surprisingly, blocking the MAP kinase (ERK) cascade downstream of Ras is a well-explored therapeutic strategy (3)(4)(5). For example, the MAP kinase kinase kinase B-Raf is a validated target in human melanomas harboring mutant B-Raf oncogenes (6,7). Interestingly, the use of some Raf inhibitors is contraindicated in patients with melanomas harboring oncogenic Ras mutations (6,7).
Melanoma cells with oncogenic Ras mutations have provided additional insights into the roles of Raf isoforms in human tumors. For example, melanoma cells with oncogenic Ras mutations undergo a switch in the usage of Raf isoforms from B-Raf to C-Raf (8). This novel dependence on C-Raf occurs despite the continued expression of B-Raf within the tumor cells (9). Both the proliferation and basal ERK activation in these Ras-mutant melanoma cells were dependent solely on C-Raf, with no contribution from B-Raf. This was due to the high basal level of ERK activity in these cancer cells and the subsequent ERK-dependent phosphorylation of B-Raf itself, which prevented B-Raf association with oncogenic Ras (9,10). Whether this isoform switch from B-Raf to C-Raf occurs in non-melanocytic tumors that harbor oncogenic Ras mutations is unknown. To determine whether B-Raf contributes to the malignant phenotype of non-melanocytic cancer cells will be important because of the potential of B-Raf-specific therapies and the possibility that signaling from Ras to C-Raf and B-Raf may be differentially inhibited by PKA.
Oncogenic Ras mutations are prevalent in lung and colon cancers (11)(12)(13). Ras mutations occur in 15-30% of non-small cell lung carcinomas (NSCLC) 2 and are associated with a poor prognosis (14). Most Ras mutations arise in either the Kirsten Ras (KRas) or neuroblastoma Ras (NRas) isoforms. NRas and KRas are common in both lung and colon cancers and are thought to induce carcinogenesis through similar functional outputs (13). However, differences between NRas and KRas have been noted in mouse models of colonic cancer (15). These Ras-mutant tumors generally express wild type B-Raf (11). Therefore, it is important to revisit the requirement of B-Raf in this group of Ras-mutant cells. Here we examined a range of Ras-mutant cancer cell lines including H1299 cells, a human NSCLC line, harboring an NRas Q61K mutation, Calu-6, a human NSCLC line (KRas Q61K), and HCT116, a human colorectal carcinoma cell line (KRas G13D).
Although these cells generally express B-Raf, the contribution of B-Raf to oncogenic Ras signaling is not known. We show that both B-Raf and C-Raf contribute to the basal ERK activation seen in these cells. Moreover, both C-Raf and B-Raf were found to be constitutively associated with oncogenic Ras and that the binding of each isoform was inhibited through independent phosphorylations of each kinase by PKA.
Human H-RasV12 (RasV12) in pGEX-4T3 was previously described (21). B-Raf constructs for bacterial expression were generated by PCR amplification. PCR primers containing Sac1 and HindIII sites were designed to amplify DNA encoding amino acids 1-414 of human B-Raf WT or R188L mutant. The PCR products were subcloned into the pET24a vector (Millipore, Inc.) to generate amino-terminal fragments containing the Ras binding domains of B-Raf WT or R188L.
Cell Culture Conditions and Transfections-H1299 cells were provided by Mushui Dai (Oregon Health and Science University). HCT116 cells were provided by Charles Lopez (Oregon Health and Science University). Calu-6 cells were purchased from American Type Culture Collection (Manassas, VA). H1299, HCT116, Calu-6, and Hek293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, and L-glutamine at 37°C in 5% CO 2 . Beas-2b cells were a gift of David Jacoby (Oregon Health and Science University) and were cultured serum-free in 50% DMEM and 50% F-12 medium supplemented with penicillin, streptomycin, and L-glutamine at 37°C in 5% CO 2 . Before treatment, cells were serum-starved in medium with 0.2% FBS overnight and then treated with 10 M forskolin and 100 M 3-isobutyl-1-methylxanthine for 10 min or the indicated times. PACAP38 was used at 500 nM for 10 min. EGF and PDGF were used at 50 ng/ml for 5 min unless otherwise indicated. Phorbol-12-myristate-13-acetate was used at 50 ng/ml. Cross-linking anti-CD3 antibodies were used as described (22). Immunoprecipitations and Western blotting were performed as previously described (23,24). For all transfections, cells were transfected with the indicated plasmids using Lipofectamine 2000 (Invitrogen), and serum was starved before treatment. For shRNA experiments, cells were transfected twice on consecutive days and serum-starved overnight. In all cases, representative Western blots of at least three independent experiments are shown. When shown, statistical significant was evaluated using an unpaired t test.
MTT Proliferation Assay-H1299 and HCT116 cells were plated at 2,500 or 20,000 cells per well in 96-well plates, respectively. 6 -8 wells were used for each condition. After 24 h, cells were serum-starved and treated with F/I and UO126 (10 M). After 3 days, cell density was assessed by an MTT assay, as per the manufacturer's instructions. The absorbance was read at 590 nm using SpectraMax M2 microplate reader. For each cell line, a standard curve was generated to establish a linear range of ODs versus cell number. Relative cell numbers are presented as the percent of cell numbers in the untreated condition. The averages of three independent experiments are shown, and statistical significant was evaluated using an unpaired t test.
Protein Purification and Expression-The plasmids encoding proteins for bacterial expression were transformed into bacterial strain BL21(DE3). Expression of His-BRaf (amino acids 1-414) and His-BRaf R188L (amino acids 1-414) were induced by 1 mM isopropyl-␤-D-thiogalactopyranoside at 37°C for 4 h after an A 600 of 0.6 was reached. The proteins were purified as described (24). RasV12 was expressed and purified as a GST fusion protein, and RasV12 was released after cleavage of the GST peptide, as described previously (21).
In Vitro Binding Assay-100 ng of RasV12 per tube was loaded with 0.1 mM GTP␥S or 1 mM GDP in 200 l of lysis buffer supplemented with 10 mM EDTA at 30°C for 30 min as described (24). The reaction was stopped by the addition of MgCl 2 at 10 mM. 1 g of His-BRaf (amino acids 1-414) wild type or R188L was incubated with RasV12-GTP␥S or RasV12-GDP in the presence of 20 l of His beads (50% slurry) for 1 h at 4°C. The beads were washed 3 times with 1 ml of lysis buffer, boiled in 1ϫ SDS loading dye, and subjected to Western blotting.
Preparation of Primary T Cells-C57/BL6 mice were used to prepare splenic lymphocytes. Experiments on animals were performed according to the ethical guidelines of the Institutional Animal Care and Use Committee at Oregon Health and Science University in accordance with federal regulations regarding approved animal use and care. Cross-linking anti-CD3 antibodies were used as described (22).

RESULTS
Both B-Raf and C-Raf Are Required for Basal ERK Activation in H1299 Cancer Cells-Cancer cells harboring Ras mutations maintain a high basal level of ERK activity that drives their proliferation (13). Moreover, Ras-mutant tumors remain dependent on Ras activation for ERK activation and proliferation (25). This was seen in H1299 cells, as basal ERK activation was blocked by shRNAs for NRas (Fig. 1A).
Ras activation of ERKs requires the action of Raf kinases. H1299, HCT116, and Calu-6 cells all express both B-Raf and C-Raf (Fig. 1B). Importantly, both B-Raf and C-Raf contributed to the basal ERK activity in all cells, as isoform-specific shRNAs to B-Raf and C-Raf each reduced endogenous basal ERK activity ( Fig. 1, C, D, and E). Similar reductions were seen in the basal activation of the ERK kinase MEK (Fig. 1, C, D, and E).
Constitutive Ras Activation Does Not Prevent B-Raf Binding to Ras-The role of B-Raf in maintaining the elevated ERK activity present in H1299 cells was surprising in light of the recent report that B-Raf is incapable of binding to oncogenic Ras in Ras-mutant melanomas (9,10). The inability of B-Raf to bind to Ras in those tumors is a consequence of the ERKdependent phosphorylation of B-Raf on serine 151. Phosphorylation of the Ser-151 site disrupts the Ras binding domain of B-Raf (10,26). The high level of basal ERK activation in Ras-mutant melanomas is thought to maintain this Ser-151 site in a constitutively phosphorylated state that directly interferes with the Ras-dependent activation of B-Raf signaling (10). . Cells were transfected with control shRNA, B-Raf shRNA, or C-Raf shRNA as indicated. Cells were harvested, and endogenous protein levels were assayed by Western blot. The efficiencies of knockdown of endogenous B-Raf and C-Raf are shown in the first and second panels, respectively. MEK activation was measured using phosphorylation-specific antibodies against pMEK (third panel). ERK activation was measured using phosphorylation-specific antibodies against pERK (fourth panel). Total endogenous ERK2 levels are shown in the fifth panel.
Here, we assayed phosphorylation of this site using a MAP kinase substrate antibody that was designed to recognize a large subset of MAP kinase target sites. In H1299 cells, this antibody detected potential ERK-dependent phosphorylations in both transfected and endogenous B-Raf (Fig. 2, A and B), consistent with a basal level of ERK-dependent phosphorylation in B-Raf in these cells. B-Raf is known to be phosphorylated at multiple sites by ERK (26), yet this MAP kinase substrate antibody recognized only phosphorylation of Ser-151 (Ser(P)-151), as demonstrated by the absence of immunoreactivity of this MAP kinase substrate antibody with the mutant B-Raf S151A, both in vivo, and when B-Raf was phosphorylated by ERK2 in vitro ( Fig.   2A). To establish the stoichiometry of phosphorylation of endogenous B-Raf Ser-151 in H1299 cells, we compared the basal levels of p Ser-151 in vivo to the levels of Ser(P)-151 achieved after in vitro phosphorylation of B-Raf by recombinant ERK2. The basal levels of Ser-151 phosphorylation of transfected ( Fig. 2A) and endogenous B-Raf ( Fig. 2 B) were greatly increased upon incubation with active ERK2 kinase in vitro ( Fig. 2A and B), suggesting that Ser-151 was predominantly unphosphorylated in these tumor cells.
An estimate of the stoichiometry of Ser-151 phosphorylation within cells could be made by comparing the Ser(P)-151 levels seen in cells with the maximal level achieved after the complete phosphorylation of B-Raf by active ERK2 kinase in vitro. In Fig.  2B, we show that incubation of immunoprecipitated B-Raf with active ERK kinase in vitro increased the basal level of phosphorylation of Ser-151 8-fold over that seen in cells, suggesting that the stoichiometry of basal phosphorylation in vivo was not greater than 10 -15%. Because in vitro phosphorylation can be incomplete, this may be an overestimate. The low level of basal phosphorylation of B-Raf Ser-151 was also suggested by the finding that wild type B-Raf and B-Raf S151A bound to NRasV12 to similar degrees (Fig. 2C). Therefore, in contrast to melanoma cells, the steady state level of phosphorylation at Ser-151 in these cells was not high enough to prevent a basal level of B-Raf binding to Ras.
To determine whether phosphorylation on Ser-151 diminishes the ability of B-Raf to bind to Ras in H1299 cells, we compared the Ser(P)-151 levels with the entire pool of B-Raf to the pool of B-Raf bound to Ras. A significantly lower level of Ser(P)-151 of B-Raf was detected within a Ras co-IP than in a B-Raf IP containing an equivalent amount of B-Raf protein (Fig. 2D), consistent with previous studies (10,26). Interestingly, the phosphomimetic mutation B-Raf S151E maintained binding to Ras (Fig. 2E), suggesting that this mutation did not faithfully mimic phosphorylation. Taken together, the data suggest that the low level of phosphorylation of B-Raf on Ser-151 permits a high degree of basal B-Raf association with Ras in these cells.
The basal association of B-Raf with Ras could be either direct or indirect through heterodimerization with C-Raf. Here, we examined whether the basal association of B-Raf with activated NRas in these cells was direct. For C-Raf, interaction with Ras in vitro can be blocked by a single change of arginine to leucine (R89L) (27). B-Raf R188L corresponds to this R89L mutation and, like C-Raf R89L, does not bind to Ras-GTP in vitro (Fig.  3A). In H1299 cells cotransfected with FLAG-NRasV12 and B-Raf R188L, B-Raf R188L was absent in NRas immunoprecipitates, confirming that the binding of B-Raf WT to Ras requires a direct interaction (Fig. 3B). The possibility that B-Raf was interacting with NRas through heterodimerization via C-Raf was also ruled out using the dimerization-deficient mutant B-Raf R509H (Fig. 3C).
cAMP/PKA Inhibits Basal ERK Activation and Proliferation in H1299 Cells-Elevations in cAMP levels were triggered by the addition of forskolin, an activator of adenylate cyclase, in the presence of the phosphodiesterase inhibitor 3-isobutyl-1methylxanthine (F/I). In H1299 cells, F/I blocked both basal MEK and ERK activity (Fig. 4A),as well as elevations in ERK activity triggered by EGF (Fig. 4B).
F/I blocked basal ERK activation within 5 min (Fig. 4A). This block was also seen after incubation with the peptide hormone PACAP38 (pituitary adenylate cyclase activating peptide-38) (Fig. 4C), presumably acting through the PAC1 receptor that is expressed on these cells (28). PAC1 receptor is a G proteincoupled receptor linked to the production of cAMP and activation of PKA. In these cells PACAP ability to inhibit ERKs required PKA, as it was blocked by the expression of the PKA inhibitor protein kinase inhibitor (Fig. 4C). In H1299 cells, treatment with F/I also blocked proliferation (Fig. 4D). This block was similar to that seen using the MEK inhibitor UO126. Moreover, F/I had a very small additive effect when combined UO126 (Fig. 4D), suggesting that F/I was functioning largely by blocking ERK signaling. Similar inhibitions of MEK and ERK activity (Fig. 4E) and proliferation (Fig. 4F) by F/I were seen in HCT116 cells.
The inhibition of ERKs by cAMP was also seen in non-cancerous and primary cells, including T lymphocytes, Beas-2b lung epithelial cells, and NIH3T3 cells. In these experiments, ERK activity was triggered by the addition of exogenous stimulators of Ras-dependent signaling (Fig. 5, A, B, and C).
PKA Inhibits Ras Binding to Both C-Raf and B-Raf in H1299 Cells-PKA inhibition of ERKs is thought to be mediated by PKA inhibition of C-Raf binding to Ras (29). However, if B-Raf contributes to ERK activation, any mechanism explaining how PKA inhibits ERKs must include B-Raf as well. Initially, we examined whether B-Raf binding to Ras was a target of inhibition by cAMP/PKA in Ras-mutant cells.
In H1299 cells, endogenous Ras proteins bound constitutively to B-Raf (Fig. 6A) as well as to C-Raf (Fig. 6B), and both interactions were blocked by treatment with F/I (Fig. 6, A and  B). Similar results were seen using transfected NRasV12 in both H1299 and HCT116 cells (Fig. 6, C and D). We also examined Hek293 cells stimulated by growth factors. In Hek293 cells, activation of wild type NRas by EGF induced B-Raf binding, and this was blocked by F/I (Fig. 6E). Therefore, the inhibition of B-Raf binding to Ras by cAMP was also seen during activation of wild type Ras.
B-Raf binding to the small G protein Rap1 has been shown to occur during cAMP-dependent activation of ERKs (30,31). Therefore, we also determined whether B-Raf binding to activated Rap1 was inhibited by cAMP. For these experiments we used RapE63, a constitutively active mutant of Rap1. Unlike that seen for B-Raf binding to RasV12, forskolin treatment had no significant effect on the ability of B-Raf to bind RapE63 (Fig. 7, A-C). This suggests that RapE63, but not Ras, can couple to B-Raf during PKA stimulation of ERKs.

. cAMP inhibits ERK activation and cellular proliferation in H1299 and HCT116 cells.
A, cAMP inhibits the basal ERK activation in H1299. H1299 cells were serum-starved overnight and treated with F/I for the indicated times. MEK activation was measured using phosphorylation-specific antibodies against activated MEK (pMEK) (first panel). ERK activation was detected by Western blot using phosphorylation-specific antibodies (pERK) (second panel). Total levels of ERK2 are shown in the third panel. B, cAMP inhibits growth factor stimulated ERK activation in H1299 cells. Cells were serum-starved overnight and treated with EGF (1 ng/ml) for the indicated times in the presence and absence of 10 min F/I pretreatment. ERK activation (pERK) was detected by Western blot using phosphorylation-specific antibodies (pERK). pERK levels are shown in the top panel. Total levels of ERK2 are shown in the bottom panel. C, PACAP38 inhibits the basal ERK activation in H1299 in a PKA-dependent manner. H1299 cells were transfected with Myc-ERK2 in the presence or absence of protein kinase inhibitor (PKI), a selective peptide inhibitor of PKA. After serum starvation, cells were treated with F/I or PACAP38. Myc-ERK2 activation (pMyc-ERK2) was measured within the Myc IP by Western blot using phosphorylation-specific antibodies. D, cAMP inhibits the proliferation of H1299 cells. H1299 cells were plated, serum-starved, and treated with F/I (F) and/or UO126 (UO) for 3 days, and cell numbers were assessed by MTT assay. Relative cell numbers after 3 days are shown as the percent of cell number compared with the untreated group. Normalized averages of the three different experiments are shown Ϯ S.E. There was statistical significance between all conditions (p Ͻ 0.05) except those marked with ns (not significant). E, cAMP inhibits the basal ERK activation in HCT116 cells. Cells were serum-starved overnight and treated with F/I for the indicated times. MEK activation was measured using phosphorylation-specific antibodies against activated MEK (pMEK) (first panel). ERK activation was detected by Western blot using phosphorylation-specific antibodies (pERK) (second panel). Total levels of ERK2 are shown in the third panel. F, cAMP inhibits cellular proliferation in HCT116 cells. HCT116 cells were plated, serum-starved, and treated with F/I (F), UO126 (UO), or both (F/UO) for 3 days, and cell numbers were assessed by MTT assay. Relative cell numbers after 3 days are shown as the percent of cell number compared with the untreated group. Normalized averages of the three different experiments are shown Ϯ S.E. There was statistical significance between all conditions (p Ͻ 0.05). SEPTEMBER 20, 2013 • VOLUME 288 • NUMBER 38

PKA Phosphorylation of Serine 259 within C-Raf Prevents
Binding to Ras-As mentioned above, C-Raf is a well-established target of PKA inhibition. A number of phosphorylation sites in C-Raf have been suggested to contribute to PKA ability to inhibit Ras signaling to C-Raf, including phosphorylations of serine 259 (Ser(P)-259) and serine 43 (Ser(P)-43) in C-Raf (32)(33)(34). Phosphorylation of Ser-259 serves as an amino-terminal binding site for the adaptor protein 14-3-3 (35). Phosphorylation of Ser-43 of C-Raf by PKA may interfere with C-Raf bind-ing to Ras (36), but others studies have challenged this view (32,34,37).
Here, we confirmed the role of Ser(P)-259 in PKA inhibition of C-Raf binding to Ras in H1299 cells. We show the ability of F/I to inhibit NRas binding to C-Raf was greatly reduced in the C-Raf mutant lacking this phosphorylation site (C-Raf S259A) (Fig. 8A). In contrast, F/I treatment was able to inhibit binding of Ras to C-Raf S43A mutant to the same level as that shown for wild type C-Raf (Fig. 8A), confirming that phosphorylation of FIGURE 5. cAMP inhibits ERK activation in multiple non-cancerous cell types. A, cAMP inhibits the stimulated ERK activation in primary T cells. CD4-positive T cells were harvested from mouse spleen. Cells were plated and stimulated with either anti-CD3 cross-linking or phorbol-12-myristate-13-acetate in the presence and absence of 10 min of F/I pretreatment. ERK activation was detected by Western blot using phosphorylation-specific antibodies (pERK). Total levels of ERK2 are shown as controls. B, cAMP inhibits both the basal and stimulated ERK activation in Beas-2b lung epithelial cells. Cells were plated and stimulated with EGF for 20 min in the presence and absence of 10 min F/I pretreatment. ERK activation was detected by Western blot using phosphorylation-specific antibodies (pERK). Total levels of ERK2 are shown as controls. C, cAMP inhibits the stimulated ERK activation in NIH3T3 cells. Cells were plated and stimulated with PDGF for 5 min in the presence and absence of 10 min F/I pretreatment. MEK activation was measured using phosphorylation-specific antibodies against activated MEK (pMEK) (first panel). ERK activation was detected by Western blot using phosphorylation-specific antibodies (pERK) (second panel). Total levels of ERK2 are shown in the third panel. Ser-259 plays a dominant role in PKA inhibition of C-Raf binding to Ras in these cells. These results are in general agreement with previous studies that phosphorylation of Ser-259 creates a binding site for the adaptor protein 14-3-3, and 14-3-3 binding at that site blocks Ras binding (38 -40).

PKA Phosphorylation of Ser-365 within B-Raf Prevents B-Raf
Binding to Ras-The mechanism by which PKA inhibits B-Raf binding to Ras is not known. We show that PKA inhibition of B-Raf binding to Ras requires phosphorylation of B-Raf on serine 365. B-Raf Ser-365 is analogous to the Ser-259 site in C-Raf that is required for PKA inhibition (32,39). B-Raf S365A, a mutant B-Raf lacking this phosphorylation site, bound to NRasV12 in H1299 cells, but this binding was not inhibited by F/I (Fig. 8B). F/I blocked the activation of ERKs in cells transfected with B-Raf WT but not B-Raf S365A (Fig. 8C). These data suggest that PKA phosphorylation of Ser-365 in B-Raf uncouples NRas from B-Raf and prevents B-Raf from activating ERKs.
The phospho-specific antibody designed against Ser(P)-259 in C-Raf can also be used to monitor Ser(P)-365 in B-Raf (41). Using this antibody, Ser-365 phosphorylation was detected basally, and the level of this phosphorylation was increased upon forskolin treatment (Fig. 8D). Phosphorylation at Ser-365 has been shown to induce the binding of B-Raf to 14-3-3 (35), and F/I also increased the level of 14-3-3 bound to B-Raf (Fig.  8E). Moreover, the absence of phosphorylation at this site in the mutant B-Raf S365A greatly reduced 14-3-3 binding and prevented the increase in 14-3-3 binding induced by forskolin (Fig.  8E). We suggest that the basal binding of B-Raf to Ras occurs within the pool of B-Raf that is basally unphosphorylated at Ser-365. We propose that the PKA-dependent increase in the phosphorylation on Ser-365 upon activation of PKA and the resultant increase in 14-3-3 binding to this phosphorylated site prevents B-Raf from binding to oncogenic Ras. This is similar to the model proposed for the regulation of Ras/C-Raf association via binding of 14-3-3 to phosphorylated Ser-259 within C-Raf (39).

DISCUSSION
Ras-mutant cancers and cell lines maintain their dependence on Ras signaling in a process that is referred to as "oncogene addiction" (25,42,43). Signaling downstream of oncogenic Ras along the MAP kinase cascade can be mediated by either C-Raf and/or B-Raf. We show here that both B-Raf and C-Raf participate in ERK activation in multiple Ras-mutant cancer cell lines. This finding is different from that reported in recent studies showing that B-Raf does not participate in ERK activation in Ras-mutant melanomas (9). The finding that B-Raf can still bind Ras and contribute to ERK activation in some non-melanocytic cancer cells is relevant because it demonstrates that cancers cells harboring Ras mutations are not always dependent on C-Raf signaling. This knowledge is important because of the growing potential of B-Raf-specific therapies.
Studies in melanomas have received much recent attention for the role of Raf dimerization in ERK activation by oncogenic Ras and specific Raf inhibitors (6,44,45). Although it was not the focus of this study, we show that the binding of B-Raf to Ras was independent of B-Raf dimerization, as similar binding to Ras was seen in wild type B-Raf and dimerization-deficient mutants of B-Raf.
The uncoupling of B-Raf from Ras in Ras-mutant melanoma cells is due to the phosphorylation of B-Raf on a specific ERK target site (Ser-151) that lies near the Ras-binding site of B-Raf (10). This site is basally phosphorylated within B-Raf in Rasmutant melanoma cells, and this phosphorylation is thought to prevent it from binding to oncogenic Ras (10). Here, we confirm those results and the results of others (26) that phosphorylation of Ser-151 largely excludes B-Raf from associating with Ras. However, in H1299 cells the basal levels of Ser(P)-151 were quite low. This low level of phosphorylation presumably accounts for the high level of basal association of B-Raf with active Ras.
In non-melanoma Ras-mutant cells, we show findings that B-Raf directly associates with activated Ras and that Ser-151 is not highly phosphorylated. Prior studies in melanoma cells have shown that the phosphorylation of Ser-151 is extremely stable due to its slow rate of dephosphorylation (10). This may be because the dephosphorylation of Ser(P)-151 requires both the phosphatase PP2A and the proline isomerase Pin1 (26). Pin1 expression is elevated in many human cancers (46), including NSCLCs (47,48), and can enhance the transforming ability of Ras activation (49,50). The levels of Pin1 activity may help determine the steady state levels of phosphorylation of Ser-151 seen in different Ras-mutant tumors, influencing the contribution of B-Raf to Ras-dependent signaling.
The reliance of Ras-transformed melanoma cells on C-Raf has been proposed to make the melanoma cells particularly vulnerable to cAMP/PKA inhibition of ERKs and cell proliferation (9, 10). This model suggests that B-Raf activation via Ras is immune to the inhibitory actions of PKA that characterize C-Raf. The finding that B-Raf can bind Ras to participate in Ras activation of ERKs and proliferation in some Ras-mutant tumor cells permits us to ask a longstanding question. Is PKA inhibition of Ras signaling specific for C-Raf? Or is B-Raf equally vulnerable? Indeed, despite reports to the contrary (51), it has long been assumed that, among C-Raf and B-Raf isoforms, only Ras binding to C-Raf is inhibited by PKA (29,52).
This view has arisen in parallel with the parallel view that only B-Raf is capable of mediating a signal from cAMP/PKA to ERKs. Indeed, in many cells types B-Raf is required for ERK activation through hormones coupled to cAMP/PKA, including melanocytes (53), renal collecting duct cells (54), prostate cells (55), chondrocytes (56), thyroid cells (57), renal cells (58), pituitary cells (59), and neuronal cells (31,60,61). There are cases where B-Raf does not seem to be involved, however (62,63).
One PKA phosphorylation site that has been proposed to mediate the selective inhibition of C-Raf by PKA is Ser-43 (33,36). When this site is phosphorylated, this site blocks access to Ras (36). Because the analogous site in B-Raf (Ser-151) is not a PKA target, this finding helped perpetuate the belief that Ras binding to C-Raf and C-Raf is differentially regulated by PKA. We show here that Ser-43 phosphorylation is not required for PKA ability to inhibit C-Raf binding to Ras. In contrast, the phosphorylation of Ser-259 is required for this inhibition of C-Raf binding to Ras by PKA. This finding is in general agreement with Dumaz and Marais (40).
We also show that the mechanism for this uncoupling of B-Raf from Ras is the same as that proposed for C-Raf. The site in B-Raf (Ser-365) that is analogous to Ser-259 is also a target for PKA phosphorylation, and this phosphorylation is required for PKA inhibition of B-Raf binding to Ras as well. Although the similar consequences of PKA phosphorylation of these two sites may not be surprising given the sequence homologies between the two isoforms, it provides formal proof that PKA inhibits the interactions of both C-Raf and B-Raf with Ras.
Why does cAMP inhibit ERK signaling in some cells but activates ERKs in other cells? Given that PKA can inhibit both C-Raf and B-Raf effector pathways coupling Ras to ERKs, it is clear that cAMP/PKA activation of ERKs must utilize a different pathway. Indeed, it has been proposed that the ability of cAMP/PKA to activate or inhibit ERKs in B-Raf-expressing cells is dependent on additional factors. For example, the level of 14-3-3 proteins associating with B-Raf has been proposed to dictate whether PKA can activate or inhibit ERKs (64). Other models predict that the levels of the scaffold molecules including KSR (65) and the PKA anchoring protein AKAP-Lbc (66) will influence how PKA regulates ERKs.
Ras-independent mechanisms of cAMP/PKA activation of ERKs have also been proposed. In many B-Raf-expressing cells, cAMP/PKA activation of ERKs is mediated by the small G protein Rap1 (23, 30, 31, 54 -56, 59, 67-69). We show that PKA does not disrupt the association between Rap1 and B-Raf. Therefore, the mechanism of Ras-dependent inhibition of ERKs by PKA described here may not occur during Rap1 signaling to ERKs.
The presence of oncogenic Ras mutations renders cancer cells dependent on Ras signaling (25). Therefore, PKA-dependent inhibition of Ras signaling to both B-Raf and C-Raf will likely dominate PKA actions on the Ras-dependent basal ERK activity in these cells. This is similar to the inhibitory effect of PKA on growth factor activation of ERKs that has been seen in a number of cell lines (70). In these examples, PKA ability to stimulate ERKs in the absence of added growth factors is masked by the much larger effect of PKA to inhibit ERK activation in the presence of added growth factors.
In summary, we show that oncogenic mutant forms of Ras in human cancers may still bind to B-Raf. This explains why these tumors maintain their dependence on both C-Raf and B-Raf. It also identifies a potential target of PKA inhibition of ERK, the interaction between Ras and B-Raf. Identifying the mechanism by which PKA can disrupt B-Raf binding to Ras has implications for treatment of cancers harboring oncogenic Ras mutations. These studies suggest that cAMP and hormones coupled to cAMP might be effective in dampening the effects of oncogenic Ras in non-melanocytic cancer cells, even in cases where Ras/B-Raf signaling participates in the basal proliferation of these cells. Importantly, mechanisms that dismantle endogenous cAMP signaling would be expected to provide a growth advantage in these cancers, as has been shown in some melanomas (10).