IMP Modulates KSR1-dependent Multivalent Complex Formation to Specify ERK1/2 Pathway Activation and Response Thresholds*

The Ras effector and ubiquitin-protein isopeptide ligase family member IMP acts as a steady-state resistor within the Raf-MEK-ERK kinase module. IMP concentrations are regulated by Ras through induction of autodegradation and can modulate signal/response thresholds by directly limiting the assembly of functional KSR1-dependent Raf·MEK complexes. Here, we show that the capacity of IMP to inhibit signal propagation through Raf to MEK is a consequence of disrupting KSR1 homooligomerization and B-Raf/c-Raf hetero-oligomerization. This impairs both the recruitment of MEK to activated Raf family members and the contribution of Raf oligomers to c-Raf kinase activation. Our observations indicate that human KSR1 proteins promote assembly of multivalent Raf·MEK complexes that are required for c-Raf kinase activation and functional coupling of active kinases to downstream substrates. This property is engaged by IMP for modulation of signal amplitude.

The Raf-MEK 2 -ERK kinase cascade is one of the most extensively studied signal transduction modules on account of its core participation in both normal and pathological cell regulatory networks. A great deal of information has been gathered about the acute participation of the Raf-MEK-ERK signal transduction circuitry in the modulation and/or propagation of a broad range of dynamic cell regulatory events ranging from cellular proliferation and tumorigenesis to differentiation and cell specialization (1)(2)(3). Attention is now necessarily becoming focused on understanding the molecular architecture that allows the core enzymatic components of this cascade to be shared among multiple regulatory pathways yet produce distinct biological outcomes in response to discrete inductive signals. There is accumulating evidence that control of the ampli-tude, duration, and subcellular compartmentalization of ERK activation are all critical determinants of the biological response (4 -7). The discovery and characterization of non-catalytic accessory components, or scaffolds, that generate higher order molecular organization to modulate the assembly, activation, and compartmentalization of MAPK cascades are beginning to reveal mechanisms that generate specificity in the coupling of MAPK activation to appropriate biological responses (8 -11).
We have recently described IMP (impedes mitogenic signal propagation) as a specifier of ERK activation amplitude through inhibition of functional coupling between Raf and MEK kinases (12). The mode of action of IMP is dependent upon the presence of the Raf-MEK-ERK scaffolding protein KSR1 (kinase suppressor of Ras 1) and is a consequence of impairing the capacity of this scaffold to promote signal propagation through Raf to MEK (13). Here, we have examined how IMP association with KSR1 can limit Raf⅐MEK⅐ERK complex formation and function. We find that KSR1 homo-oligomerization is required to couple distinct KSR⅐MEK and KSR⅐B-Raf complexes to allow MEK activation. Furthermore, KSR1 promotes formation of B-Raf/c-Raf hetero-oligomers that contribute to c-Raf kinase activation. IMP association blocks both KSR1 homo-oligomerization and B-Raf/c-Raf hetero-oligomerization to impair both MEK activation and the contribution of Raf oligomers to c-Raf kinase activation. These observations reveal the participation of KSR1 both upstream and downstream of Raf kinase activation in human cells and highlight Raf family oligomerization as a basic control point for modulation of signal amplitude by IMP.
Immunoprecipitation-For co-immunoprecipitation, HEK293 cells were lysed 48 h after transfection in modified radioim-mune precipitation assay buffer (20 mM Tris, pH 8.0, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 20 mM NaF, and protease inhibitors). Lysates were rotated for 20 min at 4°C, cleared by centrifugation at 17,000 ϫ g for 30 min, and immunoprecipitated overnight with antibody-conjugated beads. Immunoprecipitates were washed four times in lysis buffer. Quantitation of immunoblot signals was performed using the National Institutes of Health ImageJ, following signal capture within the linear range using an Alpha Innotech imaging system.
Kinase Assay-c-Raf-BXB was immunoprecipitated and washed three times in modified radioimmune precipitation assay buffer, two times in the same buffer plus 500 mM NaCl, and two times in 25 mM HEPES, pH 7.5, and 10 mM MgCl 2 . The kinase reaction was performed as described previously (12).
Reverse Transcription-PCR-Total RNA was prepared with the High Pure RNA isolation kit (Roche Diagnostics). Reverse transcriptase reaction was done with the SuperScript First-Strand Synthesis System (Invitrogen). PCR amplification was carried out using the LightCycler system (Roche Diagnostics). The following primers were used for PCR amplification: IMP-FW (TGCA-CGGTGTGTCTGGAG) and IMP-RV (GCAAACAGGACACGT-GGT). To quantify the transcripts in this study, parallel experiments were done using the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase as the internal control. Values were normalized using glyceraldehyde-3phosphate dehydrogenase and analyzed using the relative quantification mathematical model (Pfaffl) as described previously (19).

RESULTS
To investigate how IMP limits the capacity of KSR1 to promote productive Raf-MEK interactions and MEK activation, we examined the impact of IMP on KSR1⅐Raf⅐MEK complex formation and function. Immunoprecipitation of epitope-tagged KSR1 from serum-starved HEK293 cells resulted in the specific coprecipitation of endogenous MEK and B-Raf (Fig. 1A). A portion of the MEK present in KSR1 complexes was active independently of exogenous stimuli, as indicated by immunoreactivity with antibodies that selectively recognize the activating phosphorylations on MEK1/2. Expression of IMP reduced the population of active MEK in KSR1 complexes without affecting KSR1 association with MEK or B-Raf. IMP was detected in KSR1 complexes; however, direct immunoprecipitation of IMP, under identical conditions, recovered KSR1 and endogenous MEK but not B-Raf (Fig. 1B). In light of our previous finding that IMP can inhibit Raf⅐MEK complex formation (12), these observations suggested the possibility that IMP impairs the oligomerization of distinct KSR1⅐MEK and KSR1⅐B-Raf complexes to limit the accessibility of MEK for activation by B-Raf. Consistent with this, we found that KSR1 proteins can homo-oligomerize through interactions in the carboxyl-terminal half (Fig. 1, C and D) and that IMP expression blocks this interaction (Fig. 1D).
The capacity of KSR homo-oligomerization to participate in Raf-MEK signal propagation was reminiscent of recent observations describing the role of Raf family oligomerization in MEK activation (17). The participation of KSR1 in Raf family oligomerization is unknown; however, B-Raf/c-Raf hetero-oligomer formation correlates with ERK1/2 pathway activation, and recently discovered low activity B-Raf mutations in melanomas require association with c-Raf to induce stimulus-independent ERK1/2 activation (16,17). Expression of IMP blocked B-Raf⅐c-Raf ( Fig. 2A) complex formation as detected by immunoprecipitation of epitope-tagged B-Raf or by immunoprecipitation of endogenous c-Raf from cells expressing either wildtype B-Raf or the low activity oncogenic variant B-Raf(G596R)  (B) (16). Although 14-3-3 association has been implicated as a prerequisite for B-Raf⅐c-Raf complex formation (16,17,20), IMP impaired Raf hetero-oligomerization without affecting the relative amounts of 14-3-3 coprecipitating with B-Raf ( Fig. 2A). To examine the contribution of IMP to modulation of native B-Raf⅐c-Raf complex formation, endogenous IMP was depleted by RNA interference followed by EGF stimulation (Fig. 2D). B-Raf⅐c-Raf complex formation was induced upon EGF stimulation, and depletion of IMP resulted in an increase in both detectable B-Raf⅐c-Raf complexes (Fig. 2C). These observations indicate that IMP is both necessary and sufficient to limit generation of active B-Raf⅐c-Raf protein complexes.
B-Raf/c-Raf heteroduplexes have been reported to have greater specific activity than corresponding B-Raf or c-Raf homoduplexes or monomers (21). The mechanistic basis of this increased activity is unclear, and the mechanism regulating heteroduplex formation is unknown. We employed c-Raf-BXB, an oncogenic c-Raf variant that lacks the amino-terminal regulatory domain and is constitutively active independently of mitogen stimulation, to examine the consequence of IMP-dependent modulation of Raf complex formation on Raf function in serum-starved cells. As shown in Fig. 3A, c-Raf-BXB interacts with both the wild-type and a kinase-dead B-Raf variant with a K483M substitution (B-Raf-KD). Consistent with published observations (21), B-Raf-KD enhanced the activity of c-Raf-BXB in cells as indicated by the increased accumulation of active MEK (Fig.  3A) and by a significant increase in specific activity as measured in vitro (Fig. 3E). Importantly, unlike c-Raf (12), B-Raf associates with MEK independently of stimulation recruiting both c-Raf-BXB and MEK (Fig. 3B). Therefore, Raf family oligomerization may enhance both Raf kinase activity and substrate availability. IMP expression inhibited B-Raf⅐c-Raf-BXB complex formation together with blocking endogenous MEK activation (Fig. 3, C and D). The capacity of B-Raf-KD to enhance the specific activity of c-Raf-BXB was also impaired by IMP (Fig. 3E).
Using KSR1 null mouse embryonic fibroblasts, we have demonstrated previously that the capacity of IMP to limit signaling through the Raf-MEK-ERK cascade is KSR1-dependent (12). This finding together with those above suggest that KSR1 may  promote B-Raf/c-Raf heteroduplex formation. Consistent with this, we found that EGF-dependent induction of B-Raf⅐c-Raf complexes was undetectable in KSR1 Ϫ/Ϫ mouse embryonic fibroblasts but was restored upon reintroduction of KSR1 at near physiological concentrations (Fig. 4A). In addition, KSR1 expression was sufficient to enhance the recovery of endogenous B-Raf in c-Raf immunocomplexes from HEK293 cells, and this was blocked by IMP (Fig. 4B). This phenotype was evident with both full-length c-Raf and the amino-terminal regulatory domain (c-Raf-NЈ). This observation prompted us to evaluate the impact of IMP on the presence of native c-Raf in KSR1 immunocomplexes. As shown in Fig. 4, C and D, IMP significantly inhibited the association of c-Raf with KSR1. This is in stark contrast to the insensitivity of KSR1⅐B-Raf complexes to IMP expression (Fig. 1A), raising the possibility that c-Raf function may be acutely dependent on scaffolding complexes. To examine this relationship in the context of mitogenic signaling, we tested the relative contributions of IMP and c-Raf to EGFdependent MEK activation. IMP expression was sufficient to impair EGF-induced MEK activation and to reverse the enhanced EGF-induced MEK activation observed upon c-Raf overexpression (Fig. 5A). Importantly, the reduced levels of EGF-stimulated MEK activation, resulting from RNA interference-mediated c-Raf depletion, were insensitive to IMP expression (Fig. 5B), and the enhanced MEK activation normally observed upon IMP depletion was reversed by codepletion of c-Raf (Fig. 5C).

DISCUSSION
IMP is a Ras effector that controls sensitivity of the ERK1/2 kinase cascade to stimulus-dependent activation by limiting assembly of the core enzymatic components (12). IMP interacts with the MAPK scaffold protein KSR1 and requires KSR1 expression to influence MEK activation. KSR1 has been shown to act as an obligate mediator of functional Raf⅐MEK complex formation in a variety of cell systems (11,22). The mechanism by which KSR1 promotes Raf-MEK coupling is unclear but most likely involves direct binding to both Raf and MEK. Our observations suggest that distinct KSR1⅐MEK and KSR1⅐B-Raf complexes are present in cells and that pathway activation requires KSR1 homooligomerization to bring Raf together with its substrate. IMP expression inhibits Raf⅐MEK complex formation in cells, and this is apparently a consequence of blocking KSR1 homooligomerization without affecting the interaction of KSR1 with MEK1 or B-Raf.
Remarkably, IMP has distinct consequences on the mobilization of c-Raf versus B-Raf for mitogenic signaling. In addition to promoting Raf⅐MEK complex assembly, we found that KSR1 promotes B-Raf/c-Raf hetero-oligomerization, and this is likewise impaired by IMP expression. Recently B-Raf/c-Raf heterooligomerization has been revealed as an essential step for maximal c-Raf kinase activation and is required for biological responses to low activity B-Raf variants in melanomas (16,17). However, it is still unclear how B-Raf⅐c-Raf complex formation triggers c-Raf activation and whether this requires B-Raf kinase activity (17,21). We find that kinase-inactive B-Raf has the capacity to drive activation of the c-Raf catalytic domain, sug-gesting that B-Raf association either can facilitate autoactivating c-Raf transphosphorylation events or can recruit additional upstream regulators that facilitate c-Raf kinase activation. Importantly, KSR1 promotes B-Raf/c-Raf hetero-oligomerization, and this is blocked by IMP. As a consequence, we find that IMP reverses the capacity of kinase-dead B-Raf to increase the specific activity of the c-Raf catalytic domain, and IMP uncouples c-Raf from KSR1 complexes.
Current paradigms suggest mammalian KSR1 is primarily responsible for coupling active Raf kinases to their substrates, MEK1 and MEK2 (23,24). In contrast, although Drosophila KSR plays a similar role, it is also clearly required for Raf kinase activation (25,26). Our observations suggest that, like the Drosophila ortholog, human KSR1 acts both upstream and downstream of Raf kinase activation. However, c-Raf is selectively sensitive to this contribution of KSR1 to signal propagation. A simple role for KSR1 proteins in assembly of multivalent Raf and MEK complexes could easily account for this epistatic relationship. The capacity of IMP to limit the formation of such complexes highlights this regulatory step as a key axis of control for signal modulation.