The requirement of specific membrane domains for Raf-1 phosphorylation and activation

Activation of Raf-1 by Ras requires recruitment to the membrane as well as additional phosphorylations, including phosphorylation at serine 338 (S338) and tyrosine 341 (Y341). In this study we show that Y341 participates in the recruitment of Raf-1 to specialized membrane domains called “rafts” which are required for Raf-1 to be phosphorylated on S338. Raf-1 is also thought to be recruited to the small G protein Rap1 upon GTP loading of Rap1. However, this does not result in Raf-1 activation. We propose that this is because Raf-1 is not phosphorylated on Y341 upon recruitment to Rap1. Redirecting Rap1 to Ras-containing membranes or mimicking Y341 phosphorylation of Raf-1 by mutation converts Rap1 into an activator of Raf-1. In contrast to Raf-1, B-Raf is activated by Rap1. We suggest that this is because B-Raf activation is independent of tyrosine phosphorylation. Moreover, mutants that render B-Raf dependent of tyrosine phosphorylation are no longer activated by Rap1. Raf-1 sequential first the phosphorylation of Y341 by membrane-bound Y341 kinases, whose activities induced EGF Ras relocalize the complex specialized microdomains a second on S338 the Raf-1 molecule phosphorylating downstream effectors MEK/ERK. Rap1 lower portion Rap1 activation does not lead to phosphorylation of Raf-1 or activation of MEK/ERK, although Raf-1 is recruited to Rap1-containing membranes. In contrast, B-Raf is constitutively phosphorylated on S445 (S338 equivalent site), and the adjacent tyrosines in Raf-1 are replaced with aspartate residues (447D,448D), and therefore does not need to be phosphorylated upon recruitment to Rap1. In this case, Rap1 is capable of coupling B-Raf to MEK/ERK We suggest that the lack of Y341 activity within Rap1 domains is the limiting step in Rap1’s inability to activate Raf-1. Gray and white circles represent Y341 kinases and S338 kinases, respectively.


INTRODUCTION
The mitogen-activated protein (MAP) kinase family regulates diverse physiological processes including cell growth, differentiation, and death. Activation of one of these MAP kinases (the extracellular signal-regulated kinase, or ERK) is initiated by the recruitment of the MAP kinase kinase kinase Raf-1 to the small G protein Ras, a resident plasma membrane protein. The Ras family consists of three members (Ha-Ras, Ki-Ras, and N-Ras) (1) that display overlapping but distinct patterns of expression and function (2). Ras is tethered to the membrane via a carboxyl CAAX motif containing a cysteine (C) followed by two aliphatic amino acids (A) and a carboxyl-terminal amino acid (X), that directs the attachment of a farnesyl moiety (3,4).
In addition to farnesylation, a second signal assists in correct membrane targeting. For Ha-Ras and N-Ras, this second signal is a palmitoyl moiety that is introduced on a neighboring cysteine. In Ki-Ras, this second site is a polybasic domain. The requirement of Ras' membrane localization for Raf-1 activation has been confirmed by mutating the terminal cysteine in a constitutively active Ras mutant, RasV12, resulting in a mutant that can not activate Raf-1 (5).
Membrane regions rich in cholesterol and sphingolipids, termed "rafts" or detergentinsoluble glycolipid-enriched complexes have been proposed to participate in signaling events by organizing additional molecules, such as c-Src, G protein subunits, and phospholipases, into discrete membrane domains (6). Recent attention has focused on the role of these specialized microdomains in Ras signaling (7)(8)(9)(10). A number of groups have shown that both Ras isoforms Ki-Ras (11) and Ha-Ras are targeted to rafts (7)(8)(9)(12)(13)(14). Others have suggested that Ha-Ras, but not Ki-Ras, can be targeted to raft microdomains (10,15). Targeting of Ras isoforms to specific membrane domains may be determined by the characteristics of the lipid modifications on Ras, as well as other sequences found within the hypervariable region (10). Localization of Ras isoforms to distinct membrane microdomains may influence selectivity of signaling among the Ras isoforms (2). For example, Ki-Ras is thought to couple well to Raf-1 but, unlike Ha- Wild type (CAV/WT) and mutant caveolin-3 (CAV/DGV) plasmids were constructed as per Watson et al (32).
Western blotting and immunoblotting--Cos-7 cells were stimulated and lysates prepared as described. Protein concentrations were determined by using the Bio-Rad protein assay dye reagent according to manufacturer's recommendations. Equal amounts of lysate were immunoprecipitated with either Flag M2 antibody coupled to agarose (Sigma) or anti-Myc antibody (9E10) coupled to agarose (Santa Cruz Biotechnology) where indicated and examined by western blot as previously described (33). Samples were separated by SDS-PAGE and transferred to PVDF membrane. shown. The expression of endogenous N-Ras, Ki-Ras, and Ha-Ras was examined using subtype-specific Ras antisera from Santa Cruz Biotechnology.
Supernatants were then centrifuged at 100,000x g in a Beckman TLA 45 rotor at 4 o C for 30 min. The supernatant was collected and designated the cytosolic fraction (S100) and the pellet was resuspended in 250 µl hypotonic lysis buffer and designated the membrane (P100) fraction.
Sucrose gradients--Membrane microdomains were isolated based on their buoyant density using isopycnic equilibrium sucrose density gradient centrifugation. A non-detergent method for lipid raft isolation was used based on the method of Smart and colleagues (34), but modified for equilibrium centrifugation and sucrose was used instead of Optiprep. Briefly, transfected Cos-7 cells (2x10 6 cells) were rinsed twice in PBS and scrapped into 0.5 ml MES buffer (25 mM MES pH 6.5, 10 mM NaCl, 5 mM Mg2Cl, 10 µg/ml aprotinin, 10 mg/ml leupeptin, 25 SW 55 rotor for 16 hrs at 48,000 rpm. A visible band 3-4mm from the   top was observed after centrifugation and corresponded to the lipid raft/caveolae fraction. Twelve 0.43 ml fractions were collected from the top of the gradient. From each fraction, 40 µl was removed for refractometry and protein determination using the Bradford method (BioRad). The remainder of each fraction was diluted with 1 ml of MES buffer to dilute out the sucrose and membranes and proteins were pelleted at 100,000 x g for 45 min. in a TLA 45 rotor. Pellets were resuspended in Laemmli buffer, separated by SDS-PAGE, and transferred to a PVDF membrane for analysis by immunoblotting. Only the first 10 fractions are shown for each gradient. Similar results were obtained using the sodium carbonate method of raft preparation (10).  PhosphoMycERK2 Assay--For MycERK2 assays, treated and untreated cells were lysed in ERK assay buffer and activation of MycErk2 was detected as described previously (33).

Raf-1 is relocalized to raft microdomains upon EGF stimulation. Activation of
Raf-1 by Ras-dependent signals induced by growth factors is associated with a redistribution of Raf-1 from the cytoplasm to the cell membrane, where it associates with Ras (35). Recent studies suggest that Ras may be localized to specific cholesterolrich membrane microdomains called rafts (11), and upon Ras activation, Raf-1 may also be recruited to rafts (12).
Proteins that localize to cholesterol-rich microdomains (rafts) can be detected within low-density fractions of sucrose density gradients (10,14,36). The method is illustrated in Fig. 1A. The density gradient achieved following equilibrium centrifugation is shown in Fig. 1B with the corresponding protein concentrations. One of these raft proteins, caveolin-1, was used to identify the density of these cholesterolrich raft domains (Fig. 1C, top panel) (6). Using this technique, we show that endogenous Raf-1 was excluded from raft microdomains in untreated cells (Fig. 1C, second panel) and EGF treatment induced the redistribution of endogenous Raf-1 into raft domains (Fig. 1C, third panel). A significant fraction of Raf-1 protein was also detected at higher densities within the gradient.
Similar results were seen using transfected cDNAs encoding wild type Raf-1, tagged with the Flag epitope. We show that wild type Raf-1 was excluded from raft microdomains in untreated cells (Fig. 1D, top panel) and EGF treatment induced the redistribution of wild type Raf-1 into raft domains (Fig. 1D, middle panel). Raft domains can be disrupted by the cholesterol-depleting agent methyl-β-cyclodextrin (CD) (11). In the presence of CD, EGF recruitment of Raf-1 to low density fractions was inhibited (Fig. 1D, bottom panel), confirming that these low density fractions represented cholesterol-rich membrane microdomains. This provides strong evidence that the buoyant fractions containing Raf-1 represented cholesterol-rich membrane 12 microdomains consistent with rafts. Raf-1 could also be redistributed to rafts in cells transfected with constitutively active mutants of Ha-Ras, Ki-Ras and N-Ras (Fig. 1E).
The localization of Ras proteins to rafts is not completely understood and some controversies remain (37), with some groups showing that Ha-Ras, but not Ki-Ras, requires raft localization for full activity (9,15). Cos-7 cells express detectable levels of Ki-Ras and N-Ras ( Fig Recent studies have shown that Ras proteins are localized to specialized raft domains called caveolae (11). These microdomains are enriched for caveolin, and a requirement for caveolin can be assessed using interfering mutants such as CAV/DGV, a truncated form of caveolin-3 (9,10,32). In Cos-7 cells, expression of CAV/DGV was detected within cytoplasmic vesicles, as previously reported (9,10)  Y341 is required for targeting to raft microdomains and phosphorylation of S338. We next examined the requirement of S338 and Y341 in Raf-1 localization.
EGF was able to direct RafS338A into rafts ( Replacing tyrosine with aspartate in the mutant Raf Y341D did not effect EGF's actions The requirement of Y341 phosphorylation for Raf-1 activation can be overcome by targeting to raft domains--Raf-1 can be constitutively targeted to the membrane following the attachment of a Kirsten Ras carboxy-terminal domain (Raf-KiCAAX) (20,21). Here, we examined the localization of a related chimera created by fusing the 14 domain and the hypervariable domain (hvr) ; Raf-HaCAAX(+hvr) (16). As expected, wild type Raf-1 was located within the cytoplasm of resting cells (Fig. 5C, left panel), and the chimera Raf-HaCAAX(+hvr) was present on the plasma membrane (Fig. 5C, middle panel). The chimera Raf-HaCAAX(-hvr) that lacked hvr sequences was also present on the plasma membrane (Fig. 5C, right panel).
Therefore, both S338 phosphorylation and biochemical activity of these chimeras paralleled raft localization.
Furthermore, the introduction of Y341D into Raf-HaCAAX(+hvr) did not significantly increase Raf-1 activity (Fig. 7B, C). These data suggest that one of the functions of Y341 is to localize Raf-1 to specific membrane microdomains permitting efficient phosphorylation of S338 and coupling to downstream effectors.

Rap1 is unable to activate Raf-1 because it can not induce Y341
phosphorylation-The ability of small G proteins to recruit Raf-1 to the membrane is not sufficient for full activation of Raf-1 (22). For example, Rap1 is a small G protein within the Ras family that can associate with Raf-1 but cannot activate it (38). Unlike Ras, which is located at the plasma membrane, Rap1 is located in vesicular membranes 16 contain a distinct CAAX motif that regulate the attachment of geranyl modifications that direct Rap1 to vesicular membranes (42,43). This could be shown using green fluorescent protein (GFP) fusions to the RapE63 protein (GFP-Rap) which, unlike GFP alone (Fig. 8A, a), was localized to perinuclear vesicles within the cytoplasm (Fig. 8A,   b). In contrast, GFP-HaRasV12 was detected at the plasma membrane, consistent with recent reports (19) (Fig. 8A, c). The chimera GFP-RapE63-HaRasCAAX was also present on the plasma membrane, confirming that the carboxyl terminal sequences of HaRas could redirect ectopic proteins (Fig. 8A, d).
Activated Rap1 does not activate Raf1 (44). Rap1's inability to permit S338 phosphorylation and activation of Raf-1 could be partially overcome by swapping Rap1's CAAX domain with that of Ras (RapE63/HaRasCAAX) (Fig. 8B). Although RapE63 could not activate wild type Raf-1, it could activate RafY341D, as measured by S338 phosphorylation and kinase activation (Fig. 8C). These data suggest that the inability of RapE63 to activate Raf-1 was due to the inability of Raf-1 to be correctly phosphorylated when recruited by Rap1, since Rap1 was capable of supporting S338 phosphorylation in the Y341D mutation. These data also suggest that the inability of Rap1 to activate Raf-1 is not just a consequence of the interaction between Rap1 and Raf-1, as has been proposed (44), but may also be dictated by the localization of Rap1.
However, HaRasV12 was better than RapE63-HaCAAX(+hvr) in activating both Raf-1 and RafY341D (Fig. 8B,C), suggesting that sequences within Ras distinct from the carboxy-terminal membrane-targeting domain are critical for maximal activation of Raf-1.
To examine the effect of relocalizing Raf-1 to Rap1-containing membranes, we generated chimeras of Raf-1 fused to the Rap1 carboxy-terminal CAAX motif (Raf-Rap1CAAX) (Fig. 8D). Raf-Rap1CAAX was not constitutively active, and could not be activated (Fig. 8D, middle panel) or phosphorylated on S338 by EGF (Fig. 8D, upper panel). These data suggest that Raf-1 needs to be targeted to specific membranes in order to be activated and that Raf-1 targeting to Rap1-specific membrane domains does not support S338 phosphorylation or activation. The inability of Rap1 to direct the proper phosphorylation of Raf-1/Rap chimeras could be overcome by introducing negative charges into Raf-1 at Y341. Mutation of Y341 to aspartate to generate RafY341D-RapCAAX increased the basal levels of both phosphorylation of S338 and Raf-1 activation compared to Raf-Rap1CAAX (Fig. 8D), which were not further increased by EGF. In Fig. 8E, we show the subcellular localization of GFP-fusion proteins, GFP-Raf-Rap1CAAX (Fig. 8E, a), GFP-RafY341-Rap1CAAX (Fig. 8E, b).
Both chimeras are largely localized to perinuclear regions, with little or no staining detected at the cell surface. Unlike Raf-1, B-Raf can be activated by the small G protein Rap1 (44). Because of this, Rap1 can activate MEK and ERK in B-Raf-expressing cells (45)(46)(47). The ability of Rap1 to activate B-Raf is shown in Fig. 9A (lanes 1-3). Both constitutively active mutants of Ras (RasV12) and Rap1 (RapE63) could activate wild type B-Raf (B-RafWT) (Fig. 9A, lanes 2, 3). In contrast to B-RafWT, expression of a mutant B-Raf in which the aspartic acid residues were mutated to the corresponding tyrosines residues in Raf-1 (B-RafYY) showed no basal phosphorylation on S445 (Fig. 9A, lane 4) sequences were coupled to B-Raf, the resulting chimera, B-Raf-Rap1CAAX, was phosphorylated on p445 and activated ERKs to a similar degree as B-Raf-HaCAAX(+hvr) (Fig. 9B). Although wild type B-Raf was constitutively phosphorylated on S445 and displayed detectable constitutive kinase activity against MEK in vitro (Fig. 9A, lane 1), it could not activate ERKs unless it was targeted to Ras or Rap1 (Fig. 9B), reflecting the requirement of specific membrane targeting for B-Raf's activation of MEK/ERK in vivo. Therefore, we propose that B-Raf's ability to mimic phosphorylation at residues 447 and 448 is critical for its ability to be activated by Rap1.

Multiple Ras isoforms localize Raf-1 to raft microdomains--Both endogenous Ki-Ras
and N-Ras were readily detected in Cos-7 cells and localized to raft domains. In contrast, Ha-Ras was not detected in Cos-7 cells, using isoform-specific antisera. This is consistent with the results of Kranenburg et al. (11), who also detected little or no Ha-Ras in Cos-7 cells. In that study, Ki-Ras, largely, and N-Ras, partially, colocalized with caveolin. Colocalization with caveolin did not appear essential for activation as EGF activated N-Ras but not Ki-Ras in Cos-7 cells (11). We show that the cholesteroldepleting agent CD blocked EGF activation of Raf-1 kinase activity and S338 phosphorylation, consistent with the CD's ability to block EGF's activation of MEK/ERK in these cells (11). Although, it has been shown that CD potentiated activation of Ki-Ras (but not N-Ras), CD completely blocked the coupling of Ras to Raf-1/MEK/ERK (11).

S338 phosphorylation of Raf-1 requires targeting to raft microdomains--Recent
studies suggest that the ability of small G proteins to regulate signaling cascades is dictated not only by the specificity of effector utilization, but also by their subcellular localization (2). Differences in the localization of specific Ras isoforms within rafts has been reported by some (2), but not others (11). Paradoxically, disruption of rafts by CD could completely inhibit coupling to downstream effectors, while actually increasing the GTP loading of selected Ras isoforms (11). This may reflect the need for selected Ras isoforms to shuttle in and out of the raft (10,48). In this study, we focused our attention on the requirement of raft localization not on Ras activation but activation of the proximal downstream effector Raf-1. We show that one of the functions of raft localization is that it permits phosphorylation of Raf-1 on S338.
Localization of Raf-1 to rafts appears to be required for full activation of ERKs (12).
Using the cholesterol-depleting agent CD, we and others (11,13) have shown that 20 disruption of raft microdomains interferes with signaling of Raf-1 to ERKs. Raf-1 activation also requires phosphorylation at serine 338. This activating phosphorylation occurs within the plasma membrane for Raf-1, subsequent to Raf-1's recruitment to Ras (27). Using CD to disrupt rafts, we show that intact rafts are required for proper phosphorylation at 338. Therefore, the requirement of raft localization for full Raf-1 activity is coupled to S338 phosphorylation, extending previous studies showing that the membrane-localized S338 kinase was required for Raf-1 activation by oncogenic Ras (29). A candidate kinase, PAK, has been proposed (49)(50)(51), however, its role has been challenged (52).

CAV/DRG does not disrupt Ras activation of Raf-1 in Cos-7 cells--A number of
studies have demonstrated that at least two types of rafts exist; those that contain caveolin and those that do not (6,11,32). Caveolins are integral components of caveolae, 50-100 nm vesicular invaginations of the plasma membrane involved in vesicular trafficking and cell signaling (53). The role of caveolin in ERK signaling has received much recent attention, and both positive and negative affects on ERKs have been reported (9,11,(54)(55)(56)(57). Ras isoforms appear heterogeneous in their ability to couple to caveolins and to localize to caveolin-containing membranes. For example, in one study, Ki-Ras largely colocalized with caveolin in Cos-7 cells, whereas N-Ras only partially colocalized with caveolin (11). In BHK cells, Ki-Ras was largely excluded from caveolin-containing membrane fractions (10).
The effect of this shift was reflected in the redistribution of caveolin and to a lesser extent of Ha-Ras, with no redistribution of Ki-Ras detected (9). This suggests that Ki-Ras, and to a lesser extent H-Ras, remains localized to low density raft domains even in by guest on March 24, 2020 http://www.jbc.org/ Downloaded from the presence of CAV/DGV. This is likely true for N-Ras as well, since N-Ras only partially colocalizes with caveolin-containing membranes.
We show data that disrupting caveolin function by overexpressing CAV3/DGV did not block the activation of Raf-1 by EGF, Ha-RasV12, Ki-RasV12, or N-RasV12 in these cells. Our finding that CAV3/DGV did not block EGF activation of Raf-1 supports the results of Kranenburg et al. (11), and may reflect the prominent role of endogenous Ki-Ras and N-Ras in EGF's actions in Cos-7 cells. However, the inability of CAV/DGV to block the action of Ha-RasV12 on Raf-1 appears to conflict with the results of Roy, et al. (9). This may reflect differences in the membrane compositions of the BHK cells (9,10) and the Cos-7 cells used in this study. For example, Ki-Ras, which we and others identified in raft microdomains in Cos-7 cells (11), has been localized to non-raft domains in BHK cells (10).
In this study we compared the ability of distinct CAAX motifs to potentiate the phosphorylation and activation of a variety of chimericRaf-1/CAAX proteins whose carboxyl-terminal domains were derived from Ha-Ras or Rap1. We show that a Raf-1 chimera that included the complete carboxyl-terminal membrane targeting domains from Ha-Ras was localized to rafts, showed both constitutive activity and phosphorylation of S338, and activated ERKs. Chimeras containing only the minimal membrane-targeting motif [Raf-HaCAAX(-hvr)], however, had no basal activity. We suggest that the lack of activity of this chimera was a direct consequence of its inability to be phosphorylated on serine 338. The ability of Y341D to restore the raft localization and S338 phosphorylation of Raf-1-HaCAAX(-hvr) and kinase activity argues that localization to specific raft microdomains may be necessary and sufficient for S338 phosphorylation and activation of Raf-1. Recent studies have proposed that hvr sequences help shuttle HaRas out of the rafts in a GTP-dependent fashion (10) and cooperate in effector utilization (16). Differences in the localization of mutant Ha-Ras proteins and Raf/Ras chimeras may be due to the influence of sequences in Ras mutants by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 22 that are absent from the Raf/Ras chimeras. It is also possible that activated Ras shuttles Raf-1 into the raft where it is phosphorylated on S338 and subsequently exits the raft, as suggested by recent studies (10,48).

Phosphorylation of Y341 is required for proper raft localization and subsequent
phosphorylation of S338-Upon EGF stimulation, RafS338A was localized to a raft microdomain. Moreover, RafS338A-HaCAAX(+hvr) was constitutively localized to a raft domain. These data demonstrate that S338 phosphorylation was not required for raft localization, but likely occurs subsequently. This is consistent with a model that Raf-1 activation by Ha-Ras requires post-translational modifications, including S338 phosphorylation that occur within specialized microdomains.
We show here that Y341 phosphorylation of Raf-1was a prerequisite for S338 phosphorylation, consistent with previous results (27). Marais and colleagues also showed that tyrosine phosphorylation by Src enhanced S338 phosphorylation of Raf-1 (27). The data presented here suggest that one of the consequences of Y341 phosphorylation may be the repositioning of Raf-1 near potential S338 kinases. The requirement of Y341 in Raf-1 activation and S338 phosphorylation, however, could be overcome by membrane targeting, suggesting that one of the functions of Y341 phosphorylation is to facilitate proper membrane localization. Indeed, RafY341A mutants were unable to enter rafts upon EGF stimulation, unless linked to ectopic rafttargeting domains.
One explanation for the increased phosphorylation on S338 seen in Y341D mutants is that this reflects the strong cooperativity between the phosphorylations of these sites (50). However, in studies examining the ability of PAK to phosphorylate Raf-1 in vitro, this was not the case (58). Another explanation is that Y341D mutants relocalize Raf-1 to sites of S338 phosphorylation. Phosphorylation of Y341 has been proposed to function in concert with pS338 to provide a negatively charged surface on the Raf-1 protein (27). We suggest that one additional function of phosphorylation of Y341 that is distinct from that of S338, is to target Raf-1 to specific membrane sites that participate in subsequent phosphorylations.

Rap1 association with Raf-1 is not sufficient for the phosphorylation of Y341--The
inability of some small G proteins to activate Raf-1 despite recruiting Raf-1 to the membrane also suggests that recruitment to the membrane is not sufficient for Raf-1 activation. One small G protein that binds Raf-1 without activating it is Rap1 (44).
Chimeric Ras/Rap1 proteins that replace membrane targeting domains of Ras with those of Rap1 are growth inhibitory (42), but this inhibition can be relieved by constitutively active Raf-1, suggesting that the inhibitory effects of this chimera were due to impaired Raf-1 activation.
One proposed function for Ras is the displacement of the 14-3-3 protein from its binding site on residue 259 within Raf-1 (35). The inability to displace 14-3-3 from Raf-1 may explain the inability of selected G proteins to activate Raf-1 (59). However, for Rap1, such a model has been ruled out (59). This suggests that other mechanisms account for the inability of Rap1 to activate Raf-1. Studies have demonstrated that activation of endogenous Rap1 limits Ras activation of Raf-1 (33,38,44). It has been proposed that Rap1 interferes with Ras by trapping the Ras/Raf-1 complex in an inactive conformation (60,61). However, recent studies have demonstrated that Ras and Rap1 occupy distinct subcellular regions (39,41,43), even following Rap1 activation (40).
In part because of its distinct location, Rap1 has been proposed to inhibit Ras activation of Raf-1 by sequestering Raf-1 from Ras. This is consistent with studies showing a loss of Ras/Raf-1 association (and a parallel increase in Rap1/Raf-1 association) upon Rap1 activation (38) This may be due to the lack of specific Y341 kinases within Rap1 domains. Therefore, we propose that Rap1 prevents Raf-1 activation by positioning it away from tyrosine kinases that are required for Y341 phosphorylation. One of the functions of Y341 phosphorylation might be to provide a regulatable interaction with proteins or lipids to participate in proper targeting of Ras/Raf-1 (62,63). Although B-Raf was constitutively active in in vitro kinase assays, we show that membrane recruitment was required to permit B-Raf to activate MEK and ERKs in vivo. Moreover, targeting of B-Raf chimeras via either Rap1-CAAX or Ras-CAAX was sufficient. The ability of B-Raf-Rap1 chimeras to activate ERKs confirms that the requirement for membrane localization for B-Raf activation by small G proteins is less stringent than that of Raf-1. The mutant of B-Raf in which D447D448 was replaced by tyrosines (B-RafYY) behaved like Raf-1; it was no longer activated by Rap1, but retained the ability to be activated by Ras. The unique specificity of Rap1 for B-Raf activation, but not Raf-1 activation, can be largely explained by the distinct requirements of each kinase for specific membrane targeting for phosphorylation and activation. Future studies examining the ability of Rap1 to support additional critical phosphorylations, including T491 and S494 in Raf-1 (T598 and S601 in B-Raf) (28,64) may be informative as well.

B-Raf's lack of dependence on tyrosine phosphorylation accounts for its activation by Rap1--The
It has been proposed that sequences within the cysteine-rich domain (CRD) of Raf-1 and B-Raf dictated the contrasting actions of Rap1 on each Raf isoform (44).
However, the ability of Rap1 to activate RafY341D, as well as the ability of Rap/Ras chimeras to activate wild type Raf-1, both argue strongly that the interactions between Rap1 and Raf-1 are not the only determinants of Raf-1 inhibition. It should be noted that Rap1/Ras chimeras were not as effective as Ras in activating/phosphorylating Raf-1, suggesting that Ras also provides an activation function that is distinct from localization (22,59,65). Furthermore, the lack of activation of B-RafYY by Rap1 suggests that interactions between the B-Raf CRD and Rap1 are also not sufficient to promote activation, although they may be important (44). We propose that the carboxyterminal domain of Rap1 provides specificity to Rap1 signaling in addition to that provided through the interaction between Rap1's effector loop and the Raf CRDs.
In conclusion, we show that Raf-1 phosphorylation at S338 requires membrane targeting of Raf-1 to specific raft microdomains. We propose that tyrosine phosphorylation of Y341 potentiates S338 phosphorylation by facilitating proper membrane localization. This two-step mechanism is outlined in Fig. 10  individual Raf-1 chimeras. Cells transfected with wild type Raf-1 were also treated with EGF as indicated. Flag-containing proteins were recovered by I.P. and examined for S338 phosphorylation (p338, first panel), and phosphorylation of MEK in vitro, as in Fig. 1B (pMEK, second panel). The position of Flag Raf proteins is shown in the third panel (FlagRaf). In the lower two panels, the lysates were subjected to Myc I.P. and the recovered MycERK2 examined for phosphorylation (pMycERK2) or total MycERK levels (MycERK2). induces the recruitment of Raf-1 to plasma membrane-bound Ras. Following this association of Ras with Raf-1 two sequential modifications occur. The first modification is the phosphorylation of Y341 by membrane-bound Y341 kinases, whose activities are induced by EGF and/or Ras activation. This phosphorylation may relocalize the Ras/Raf-1 complex within specialized plasma membrane microdomains where a second phosphorylation on S338 can occur that renders the Raf-1 molecule competent in phosphorylating downstream effectors like MEK/ERK. Rap1 signaling is depicted in the lower portion of the figure. Rap1 activation does not lead to phosphorylation of Raf-1 or activation of MEK/ERK, although Raf-1 is recruited to Rap1-containing membranes. In contrast, B-Raf is constitutively phosphorylated on S445 (S338 equivalent site), and the adjacent tyrosines in Raf-1 are replaced with aspartate residues (447D,448D), and therefore does not need to be phosphorylated upon recruitment to Rap1. In this case, Rap1 is capable of coupling B-Raf to MEK/ERK signaling. We suggest that the lack of Y341 activity within Rap1 domains is the limiting step in Rap1's inability to activate Raf-1. Gray and white circles represent Y341 kinases and S338 kinases, respectively. by guest on March 24, 2020