Protein kinase A blocks Raf-1 activity by stimulating 14-3-3 binding and blocking Raf-1 interaction with Ras.

Cyclic AMP (cAMP) blocks Raf-1 activation by stimulating its phosphorylation on serine 43 (Ser43), serine 233 (Ser233), and serine 259 (Ser259). We show here that phosphorylation of all three sites blocks Raf-1 binding to Ras.GTP in vivo and that cAMP stimulates binding of 14-3-3 proteins to Ser233 and Ser259. We also show that Raf-1 and protein kinase A (PKA) form a complex in vivo that is disrupted by cAMP and that ablation of PKA by use of small interfering RNA blocks phosphorylation by cAMP. The ability of PKA to block Raf-1 activation is ablated by the PKA inhibitor H89. These studies suggest that Raf-1 and cAMP form a signaling complex in cells. Upon activation of PKA, Raf-1 is phosphorylated and 14-3-3 binds, blocking Raf-1 recruitment to the plasma membrane and preventing its activation.

In mammalian cells, growth factors, mitogens and hormones stimulate the activity of the extracellular signalregulated protein kinases ERK1 and ERK2. 1 The ERKs control cell growth, differentiation, and survival through the phosphorylation of multiple substrates (for review, see Ref. 1). ERKs are activated by the mitogen-activated protein kinase kinases, MEK1 and MEK2, which are activated by the MEK kinases, Raf-1, A-Raf, and B-Raf. Raf-1 activation is a complex process (for review, see Ref. 2), the first step of which is direct binding to active Ras (Ras.GTP), resulting in Raf-1 recruitment to the plasma membrane. Phosphorylation of serine 338 and tyrosine 341 in the N-region and threonine 491 and serine 495 in the activation segment is essential for Raf-1 activation (3)(4)(5)(6). All of these phosphorylation events occur at the plasma membrane, so Ras-mediated recruitment of Raf-1 to the plasma membrane is essential for its activation.
14-3-3 proteins also play a key role in Raf-1 regulation. These small (ϳ30 kDa) acidic dimers bind to and regulate the activity of many proteins by binding to short phosphorylated peptide motifs (7). Two 14-3-3 motifs are present in Raf-1, centered on Ser 259 and serine 621 (Ser 621 ) (2). Binding to Ser 621 appears to be essential for Raf-1 activation, whereas binding to Ser 259 appears to suppress activity. It was recently shown that binding of 14-3-3 to Ser 259 antagonizes Raf-1 recruitment to the plasma membrane and prevents its activation by the Ras related proteins TC21 and R-Ras (8,9).
Cyclic AMP (cAMP) is another negative regulator of Raf-1. When cellular cAMP levels increase, Raf-1 becomes phosphorylated on three sites (Ser 43 , Ser 233 , and Ser 259 ). These sites work independently to suppress Raf-1 activity so all three must be mutated to prevent Raf-1 inactivation by cAMP in cells (8,10,11). How these sites work is unclear. PKA phosphorylation of Ser 43 prevents the isolated N terminus of Raf-1 from binding to Ras.GTP in vitro (12), but mutation of Ser 43 to alanine does not overcome the effects of cAMP in vivo (8,10,11). Although 14-3-3 is known to bind to phosphorylated Ser 259 , its role in cAMP mediated suppression of Raf-1 activity is not known and the mechanism of action of Ser 233 is unknown. Finally, it has not been established that PKA mediates these cAMP effects in cells, because cAMP activates several effectors (13), and in some cells, Ser 259 can be phosphorylated by protein kinase B (14,15).
Here we re-examine Raf-1 regulation by cAMP. Our data suggest that Raf-1 and PKA form a signaling complex in cells. When PKA is activated, it phosphorylates Raf-1, and stimulates recruitment of 14-3-3, preventing Raf-1 recruitment to the plasma membrane and subsequently blocking its activation.
RNA Interference-Synthetic small interfering RNA (siRNA) probes were from Dharmacon Research Inc. (Lafayette, CO). The following sequences were used: PKA C␣, AAGTGGTTTGCCACGACTGAC; PKA C␤, AAGAGTTTCTAGCCAAAGCCA; scrambled, AACCGTCGATTTCAC-CCGGG. Exponentially growing cells were transfected with 20 pmol of PKA C␣ plus 20 pmol of PKA C␤ siRNA or 40 pmol of the scrambled control.
Generation of p233-specific Antibodies-Synthetic peptide Raf-1 pS233 (SQHRYpSTPHAF; single amino acid code, where pS is a phosphorylated serine) was coupled to keyhole limpet hemocyanin (374817; Cal-* This work was supported by Medical Research Council Grant G9900391. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: ERK, extracellular signal-regulated kinase; MEK, mitogen-activated protein kinase/extracellular signalregulated kinase kinase; PKA, protein kinase A; IBMX, isobutylmethylxanthine; PDGF, platelet-derived growth factor; HA, hemagglutinin; siRNA, small interfering RNA; EGF, epidermal growth factor; GST, glutathione S-transferase; AKAP, protein kinase A anchoring protein; RBD, Ras binding domain. biochem) using gluteraldehyde (G5882; Sigma), and antibodies were raised in rats using standard protocols (18). For competition experiments, antibodies were preincubated with synthetic peptides (10 M) for 2 h at room temperature prior to use. Control peptides were: Raf-1 S233 (SQHRYSTPHAF) and Raf-1 pS43 (QRRApSDDGK).

Phosphorylation on Ser 43 , Ser 233 , and Ser 259 Disrupts Raf-1
Binding to Ras.GTP-To study Raf-1 phosphorylation on Ser 43 , Ser 233 , and Ser 259 , endogenous Raf-1 was immunoprecipitated from NIH3T3 cells using antibody M40091.G and subjected to Western blotting using phospho-specific antibodies. M40091.G was raised to amino acids 1-240 of Raf-1, and we previously demonstrated that its binding to Raf-1 is not affected by Ser 43 , Ser 233 , or Ser 259 phosphorylation (11). Ser 259 phosphorylation was examined using a commercial phospho-specific antibody (p259), and we have recently described a phospho-specific antibody for Ser 43 (p43; Ref. 11). We have also generated a rat phospho-specific monoclonal antibody to Ser 233 (p233). This antibody bound strongly to myc-epitope tagged Raf-1 (mRaf-1) that had been transiently expressed in COS cells treated with forskolin (25 M) and IBMX (500 M; Fig 1A). Forskolin activates adenylate cyclase, and IBMX inhibits phosphodiesterases, so these agents work in concert to increase intracellular cAMP levels. The antibody bound only weakly to Raf-1 from untreated cells (Fig. 1A), and binding was ablated when Ser 233 was mutated to alanine (Fig. 1A). It was also blocked when the antibody was preincubated with the phosphorylated peptide immunogen (Raf-1 pS233 ) but not by the corresponding unphosphorylated peptide (Raf-1 S233 ) or an unrelated phosphorylated peptide (Raf-1 pS43 ) (Fig. 1A). Thus the antibody is specific for phosphorylated Ser 233 .
Using these antibodies, we examined the kinetics of phosphorylation of Ser 43 , Ser 233 , and Ser 259 on endogenous Raf-1 in NIH3T3 cells. In serum-starved cells, Ser 43 and Ser 233 , were very weakly phosphorylated (Fig. 1B). Both became strongly phosphorylated within 60 s of forskolin/IBMX treatment and remained phosphorylated for up to 60 min (Fig. 1B). By contrast, Ser 259 was phosphorylated in untreated cells, but its phosphorylation still increased 2-3-fold following forskolin/ IBMX treatment (Fig. 1B). We next examined how Ser 43 , Ser 233 , and Ser 259 phosphorylation affected Raf-1 binding to Ras.GTP in vivo. For these studies, mRaf-1 was transiently expressed in COS cells together with HA-epitope tagged K-Ras (HAK-Ras). HAK-Ras was immunoprecipitated using monoclonal antibody 12CA5, and the complexes were probed for the associated mRaf-1 by Western blotting with monoclonal antibody 9E10. In serum-starved COS cells, mRaf-1 did not bind to HAK-Ras, but a complex formed rapidly following EGF treatment (Fig. 2, lanes 2 and 4). Thus, the association between these transfected proteins is dependent on Ras activation by growth factors. In agreement with our data for the endogenous proteins (11), this binding was completely blocked when the cells were pretreated with forskolin and IBMX for 10 min (Fig.  2, lanes 4 and 5), a time that stimulates efficient phosphorylation of Raf-1 on all three sites (Fig. 1B). When Ser 43 was mutated to alanine ( S34A mRaf-1), EGF still induced Raf-1 binding to Ras, but forskolin/IBMX could only partially inhibit the binding (Fig. 2, lane 9). Similarly, S259A mRaf-1 binding was still induced by EGF, and forskolin/IBMX suppression of this binding was blunted (Fig. 2, lanes 14 -17). Finally, EGF also stimulated binding of S233A mRaf-1 to HAK-Ras, but forskolin/ IBMX pretreatment was relatively effective at blocking this interaction (Fig. 2, lanes 10 -13). When all three sites were mutated, the ability of forskolin/IBMX to block binding to EGF activated Ras was completely abolished (Fig. 2, lanes 18 -21). , and protein extracts were prepared. myc-epitope-tagged proteins were captured with monoclonal antibody 9E10 for immunoblotting with phopho-specific antibody p233 (upper panels). The p233 monoclonal antibody was preincubated with competing peptides where indicated. Raf-1 levels were verified by blotting with the monoclonal antibody M40091.G (lower panel). B, NIH3T3 cells were serum-starved (24 h) and treated with forskolin/IBMX (F/I) for the indicated times when cell extracts were prepared. Endogenous Raf-1 was immunoprecipitated with monoclonal antibody M40091.G for blotting with phospho-specific antibodies to Ser 43 (p43), Ser 233 (p233), and Ser 259 (p259). Blots were reprobed with M40091.G to reveal total Raf-1 (lower panel). Thus, in vivo phosphorylation of each site appears to block the interaction of Raf-1 with Ras.GTP independently of the other two sites.
The above data show that Ser 259 is the major cAMP-regulated 14-3-3 binding site on Raf-1 in vivo, but that another site(s) exists, so we tested whether 14-3-3 binds to Ser 43 or Ser 233 . Mutation of Ser 43 to alanine did not affect binding of 14-3-3 to Raf-1 (data not shown). However, 14-3-3 binding to S233A mRaf-1 was reduced compared with wild-type mRaf-1, both in untreated and in forskolin/IBMX treated cells (Fig. 3,  lanes 8 and 9). Finally, HA14-3-3 binding to mRaf-1 was completely ablated when both Ser 233 and Ser 259 were mutated to alanine (Fig. 3, lanes 14 and 15). These data show that there are two cAMP regulated 14-3-3 binding sites in the N terminus of Raf-1, Ser 233 and Ser 259 . Ser 259 appears to be the major 14-3-3 binding site and Ser 233 a minor site.
PKA Mediates the Suppression of Raf-1 by cAMP-Finally, we tested whether the effects of cAMP were mediated by PKA. First, we examined whether Raf-1 and PKA form a complex under physiological conditions. Endogenous Raf-1 was immunoprecipitated from untreated NIH3T3 using antibody M40091.G, and the immunocomplexes were examined for PKA by Western blotting for the PKA catalytic subunit, C␣. We found that Raf-1 immunoprecipitated from untreated cells contained PKA C␣, but the interaction between these proteins was disrupted when cells were treated with forskolin/IBMX (Fig.  4A). Furthermore, the PKA regulatory subunit, RII␣, was also bound to Raf-1 in these cells, but unlike C␣, RII␣ binding was not disrupted when the cells were treated with forskolin/IBMX (Fig. 4B).
Next we examined the effects of PKA on Raf-1 kinase activity. Raf-1 activity was measured using an immunoprecipitation kinase cascade assay in which GST-MEK, GST-ERK, and myelin basic protein were used as sequential substrates (11). PDGF (50 ng/ml, 5 min) strongly activated Raf-1 in NIH3T3 cells, but this activation was blocked when cells were pretreated with forskolin/IBMX (Fig. 4C and see Ref. 11). However, when the cells were pretreated with forskolin/IBMX in the presence of the PKA inhibitor H89 (10 M, 30-min pretreatment), Raf-1 activation by PDGF was no longer blocked by forskolin/IBMX pretreatment (Fig. 4C).
Since H89 is not completely specific for PKA (19), we also used RNA interference to ablate the catalytic subunits of PKA in NIH3T3 cells. There are two catalytic subunits in mice, C␣ and C␤ (20), so we designed siRNA probes selective for each. Treatment of NIH3T3 with these probes resulted in ablation of expression of both subunits within 72 h, whereas the scrambled control probe did not affect expression of either (Fig. 4D). The C␣ and C␤ probes did not affect expression of Raf-1 or the PKA regulatory subunit RII (Fig. 4D). Ablation of C␣ and C␤ completely blocked cAMP stimulated phosphorylation of Raf-1 on Ser 43 , Ser 233 , and Ser 259 (Fig. 4D). Thus, we show for the first time that PKA is responsible for stimulating Raf-1 phosphorylation in response to cAMP in vivo. DISCUSSION A number of studies have demonstrated that agents that stimulate cAMP production in cells block Raf-1 signaling to ERK (8,11,12,21,22). However, since cAMP can activate at least three effectors (PKA, exchange factors for small G-proteins and ion exchange channels (13)), it is possible that the cAMP effects are not mediated by PKA. Here we provide unequivocal proof of the involvement of PKA. We demonstrate that when the PKA catalytic subunits are ablated using siRNA, cAMP does not stimulate phosphorylation on Ser 43 , Ser 233 , and Ser 259 . Although this does not prove that PKA directly phosphorylates Raf-1 in cells, it demonstrates that it is necessary. Furthermore, PKA can phosphorylate synthetic peptides representing each site in vitro, demonstrating that these are bona fide PKA sites (Ref. 11 and data not shown). We also show here that endogenous Raf-1 and PKA form a complex that is disrupted when cAMP levels in cells are elevated, and, finally, the PKA inhibitor H89 rescues Raf-1 activation in the presence of forskolin/IBMX. Although this compound inhibits several kinases (19), taken together these data demonstrate that PKA is directly involved in mediating Raf-1 phosphorylation and thereby suppressing its activation.
PKA is regulated by a diverse family of proteins called the protein kinase A anchoring proteins (AKAPs). These scaffolds bind to the PKA regulatory subunits and regulate the subcellular localization of the PKA holoenzymes. They also bind to PKA substrates, co-localizing kinase and substrate and thereby integrating diverse signaling pathways by coordinating phosphorylation of specific substrates (23). cAMP releases the catalytic subunits from this complex and it then phosphorylates local substrates. The discovery of a complex between Raf-1 and PKA is therefore particularly interesting, because it provides a molecular explanation for the integration of these two pathways. The data suggest that a specific AKAP keeps these two kinases co-localized so that when cAMP is elevated PKA can phosphorylate Ser 43 , Ser 233 , and Ser 259 on Raf-1 and block signaling to MEK. Consistent with this model is the observation that whereas PKA C␣ binding to Raf-1 in cells is disrupted by cAMP, RII␣ subunit binding was not, indicating that the putative AKAP remains bound to Raf-1 even when PKA is FIG. 3. 14-3-3 binds to both Ser 233 and Ser 259 . mRaf-1 (WT), S233A mRaf-1 (233A), S259A mRaf-1 (259A), or S233A,S259A mRaf-1 (233A259A) were transiently co-expressed with 14-3-3 (14-3-3) in COS cells as indicated. The cells were serum-starved (24 h) and left untreated or treated with forskolin/IBMX (F/I) for 10 min as indicated, and cell extracts were prepared for 14-3-3:Raf-1 binding assays. The level of Raf-1 and 14-3-3 in 10% of the cell extracts is shown in the lower two panels and the immunoprecipitated 14-3-3 in the second panel.
Raf-1 co-precipitated with 14-3-3 is shown in the first panel. Raf-1 was revealed with antibody 9E10 and 14-3-3 with the 12CA5 antibody. activated. We are currently attempting to identify this AKAP. Our data also show that PKA does not mediate basal phosphorylation of Ser 43 , Ser 233 , or Ser 259 , because PKA catalytic subunit ablation did not affect these events (Fig. 4D).
We have previously demonstrated that Raf-1 only completely escapes the effects of cAMP when Ser 43 , Ser 233 , and Ser 259 are all mutated to alanine (11). We show here that mutation of each also weakly restores binding of Raf-1 to Ras.GTP in vivo, although Ser 233 was the least efficient in this respect. These data suggest that each site works independently to block binding to Ras, explaining why they work independently to block Raf-1 activation. The mechanism of action of Ser 43 is unclear, but this site is just N-terminal to the Ras binding domain (RBD) of Raf-1 (amino acids 53-112 (24)) and so may work by steric hindrance or by inducing a conformational change in the RBD. Ser 233 and Ser 259 appear to act primarily by recruiting 14-3-3 to the N terminus of Raf-1.
Previous workers have proposed a model of Raf-1 regulation by 14-3-3. Since 14-3-3 binding to Ser 621 is required for activation, and binding to Ser 259 inhibits Raf-1 activity, and because 14-3-3 is a dimer, this model proposes that 14-3-3 binds to both sites simultaneously and folds Raf-1 into an inactive conformation (2). Ras.GTP would displace 14-3-3 from CR2, simultaneously unfolding Raf-1 and recruiting it to the plasma membrane for activation. However, we show here that cAMP stimulates Ser 259 phosphorylation by 2-3-fold and 14-3-3 binding by about 3-fold in vivo. Thus, in untreated cells, only about 30% of Ser 259 appears to be phosphorylated and bound to 14-3-3, so the majority of Raf-1 cannot be in this 14-3-3-maintained inactive conformation. It has also been suggested that cAMP stimulates Ser 621 phosphorylation and thereby blocks Raf-1 activity through an unknown mechanism (25). However, we did not find such a role. Ser 621 phosphorylation was not elevated in forskolin/IBMX-treated cells, and Ser 621 seems to be a rather weak 14-3-3 binding site, because binding was only visible in the S233A,S259A mRaf-1 double mutant when the Western blot was strongly overexposed (data not shown).
Based on our studies, we propose a different model for Raf-1 regulation by 14-3-3 in our cells. We and others recently demonstrated that 14-3-3 binding to Ser 259 antagonises Raf-1 recruitment to the plasma membrane (8,9). We propose that the majority of Raf-1 in resting cells is not phosphorylated on Ser 259 , and 14-3-3 is not bound to the N terminus of Raf-1, so this population can be recruited to the plasma membrane for activation. The remaining Ser 259 phosphorylated population has 14-3-3 bound to its N terminus, so it cannot be recruited to the membrane or activated unless Ser 259 is dephosphorylated by protein phosphatase 2A (26 -28). We further propose that Raf-1 is in a complex with PKA that may be mediated by AKAPs and that cAMP releases the catalytic subunit, which then phosphorylates Raf-1 on Ser 43 , Ser 233 , and Ser 259 . Phosphorylated Ser 43 blocks membrane recruitment by directly interfering with Raf-1 binding to Ras.GTP. Ser 233 and Ser 259 also block membrane recruitment, but by recruiting 14-3-3 to the Raf-1 N terminus, and 14-3-3 binds with high affinity due to the proximity of the Ser 233 and Ser 259 sites. Thus, the three sites work in unison to suppress Raf-1 activity when cAMP is elevated.
These data fit the model recently proposed by Michael Yaffe about 14-3-3 binding (7). He observed that 14-3-3 often binds to FIG. 4. PKA binds to Raf-1 and mediates its phosphorylation. A, NIH3T3 cells were treated as described in the legend to Fig. 1B, and endogenous Raf-1 was immunoprecipitated with monoclonal antibody M40091.G for blotting with PKA C␣ and Raf-1 (M40091.G antibody). The level of PKA C␣ in 10% of the cell extracts is shown in the bottom panel. B, NIH3T3 cells were treated as described in the legend to Fig.  1A, and the Raf-1 was immunoprecipitated, and the immunocomplexes were blotted for Raf-1 and PKA RII␣ as indicated. The levels of Raf-1 and PKA RII␣ in 10% of the cell extracts is shown. C, Raf-1 was immunoprecipitated for kinase activity assay from untreated NIH3T3 cells (Ϫ), or cells treated with forskolin/IBMX (F/I), PDGF, H89, or the combinations indicated. D, NIH3T3 cells were transiently transfected with scrambled siRNA (Scrambled) or siRNA directed against PKA C␣ and PKA C␤ (PKA C␣ϩC␤) as indicated. 48 h later, the cells were serum-starved (24 h) and left untreated or treated with forskolin/IBMX (F/I) for 10 min as indicated, and extracts were prepared to analyze the level of PKA C␣, PKA C␤, and PKA RII (first three panels). Endogenous Raf-1 phosphorylation (last four panels) was analyzed as described in the legend to Fig. 1. two sites on the same protein, one of which is a dominant site or "gatekeeper." The gatekeeper site is phosphorylated in unstimulated cells, the secondary site is not. Thus, each 14-3-3 dimer is only bound to its client through one of its two binding pockets. However, when the secondary site on the client protein becomes phosphorylated, the second 14-3-3 binding pocket engages onto that client rapidly due to the high local concentration induced by proximity. Since 14-3-3 is extraordinary rigid, it is thought to behave as a molecular anvil deforming its bound client protein while it undergoes only minimal structural alterations (7,29). Our data suggest that Ser 259 is the gatekeeper site, while Ser 233 is the secondary site, which is consistent with the fact that while Ser 259 conforms to a strong 14-3-3 consensus (RSTpSTP; important residues are underlined) Ser 233 conforms to a weak 14-3-3 consensus ( 231 RYpSTP 235 ) (30). When both are phosphorylated, according to the model, 14-3-3 would bind to the N terminus of Raf-1 with high affinity and induce a conformational change that would serve to completely block membrane recruitment, possibly because it also interfers with binding to Ras.GTP. Thus, although either site can block Raf-1 activation, together they provide a much stronger block, and the main effect of cAMP is to inhibit Raf-1 membrane recruitment rather than to specifically block its kinase activity.
In our model, it is still unclear what role Ser 621 binding plays in Raf-1 regulation and what the relationship between Ser 621 and Ser 259 binding is. However, our data provide a mechanistic explanation for how phosphorylation of Ser 43 , Ser 233 , and Ser 259 block Raf-1 activity and support a model in which all three sites can work independently to block Raf-1 activation by cAMP in vivo. Our results directly implicate PKA in this event for the first time.