Kinase Suppressor of Ras Signals through Thr269 of c-Raf-1*

We recently established a two-stage in vitro assay for KSR kinase activity in which KSR never comes in contact with any recombinant kinase other than c-Raf-1 and defined the epidermal growth factor (EGF) as a potent activator of KSR kinase activity (Xing, H. R., Lozano, J., and Kolesnick, R. (2000)J. Biol. Chem. 275, 17276–17280). That study, however, did not address the mechanism of c-Raf-1 stimulation by activated KSR. Here we show that phosphorylation of c-Raf-1 on Thr269 by KSR is necessary for optimal activation in response to EGF stimulation. In vitro, KSR specifically phosphorylated c-Raf-1 on threonine residues during the first stage of the two-stage kinase assay. Using purified wild-type and mutant c-Raf-1 proteins, we demonstrate that Thr269 is the major c-Raf-1 site phosphorylated by KSR in vitro and that phosphorylation of this site is essential for c-Raf-1 activation by KSR. KSR acts via transphosphorylation, not by increasing c-Raf-1 autophosphorylation, as kinase-inactive c-Raf-1(K375M) served as an equally effective KSR substrate. In vivo, low physiologic doses of EGF (0.001–0.1 ng/ml) stimulated KSR activation and induced Thr269 phosphorylation and activation of c-Raf-1. Low dose EGF did not induce serine or tyrosine phosphorylation of c-Raf-1. High dose EGF (10–100 ng/ml) induced no additional Thr269phosphorylation, but rather increased c-Raf-1 phosphorylation on serine residues and Tyr340/Tyr341. A Raf-1 mutant with valine substituted for Thr269 was unresponsive to low dose EGF, but was serine- and Tyr340/Tyr341-phosphorylated and partially activated at high dose EGF. This study shows that Thr269 is the major c-Raf-1 site phosphorylated by KSR. Furthermore, phosphorylation of this site is essential for c-Raf-1 activation by KSRin vitro and for optimal c-Raf-1 activation in response to physiologic EGF stimulation in vivo.

The mammalian raf-1 gene was first identified as the normal cellular counterpart of v-raf, the transforming gene of murine sarcoma virus (1). Two other related members of this family, A-raf and B-raf, were discovered subsequently (2,3). Raf proteins display three conserved domains: a cysteine-rich aminoterminal domain (CR1), a serine/threonine-rich domain (CR2), and a C-terminal kinase domain (CR3) (1, 4 -7). The Raf-1 serine/threonine kinase is a central component in many signal-ing pathways, functioning as a bridge to connect upstream activated tyrosine kinases and Ras to downstream serine/threonine kinases (8 -10). Upon activation, Raf-1 phosphorylates and activates MEK1, 1 resulting in propagation of the signal to MAPK. MAPK in turn phosphorylates several regulatory proteins in the cytoplasm and nucleus to alter the program of transcription and translation (11,12).
Activation of Raf-1 in vivo is initiated when cytoplasmic Raf-1, through the Ras-binding domain within CR1, couples to active GTP-bound Ras and is recruited to the plasma membrane. After binding Ras, c-Raf-1 becomes activated by a complex and still incompletely understood mechanism involving phosphorylation and interaction with membrane lipids (13)(14)(15)(16)(17)(18). Confirmation of an essential role for phosphorylation in regulating Raf-1 kinase function derives from the identification of critical in vitro and in vivo phosphorylation sites and mutational analysis of these sites. In resting cells, Raf-1 is phosphorylated exclusively on serine (Ser 43 , Ser 621 , and Ser 624 ) and threonine (Thr 268 ) residues. Serine and threonine phosphorylation can be further increased at these sites and other selected sites (Ser 259 , Ser 388 /Ser 389 , Ser 497 /Ser 499 , and Thr 269 ) upon stimulation with mitogenic stimuli and cytokines and is often accompanied by tyrosine phosphorylation at positions 340 and 341 (6, 19 -32). The relative contribution of these sites to Raf-1 activation is cell type-and stimulus-specific.
KSR (kinase suppressor of Ras), a novel member of the Ras/MAPK pathway, was originally identified and characterized in Drosophila melanogaster and Caenorhabditis elegans to function as a positive modulator of Ras function either upstream of or parallel to Raf-1 (33)(34)(35)(36). The isolation of murine and human KSR homologs with a high level of sequence identity (33) suggested that KSR signaling is evolutionarily conserved. Since its discovery as a component of the Ras pathway, much effort has been invested to elucidate the precise role of this molecule in the regulation of the MAPK cascade. However, a confusing and often contradictory picture of the role of this protein in signal transduction has emerged. The presence of an Arg in kinase subdomain II in mouse and human KSR instead of the conserved Lys normally involved in ATP binding in mammalian kinases (33) has led to the suggestion that mammalian KSR might not even function as a kinase. In fact, based on the inability of a number of groups to detect a KSR kinase activity in vitro (33,37,38) and the capacity of KSR to associate with numerous proteins (38,39), it has been proposed that the primary mode of KSR signaling is via protein-protein interaction (the scaffolding model).
In our experience, however, KSR consistently displays kinase activity in an in vitro assay in which the entire MAPK cascade is reconstituted with individual recombinant components (30,40,41). To validate our original observations, we recently established a two-stage in vitro assay for KSR kinase activity (41). During the first stage, highly purified KSR is incubated with c-Raf-1 in a reaction mixture containing ATP. In the second stage of the assay, activated c-Raf-1 is separated from KSR and incubated with a reaction mixture containing unactivated MEK, unactivated MAPK, and a human GST-Elk-1 fusion protein. Elk-1 phosphorylation serves as a specific readout for MAPK activation. Thus, KSR never comes in contact with any recombinant kinase other than c-Raf-1 in this assay. Using this approach, we demonstrated that KSR, purified to homogeneity, retained its activity to reconstitute c-Raf-1-dependent MAPK activation in vitro. Furthermore, functional integrity of the KSR kinase domain appeared necessary since substitution of conserved aspartates involved in phosphotransfer with alanines to generate a kinase-inactive KSR (D683A/D700A or Ki-KSR) abrogated MAPK signaling. Critically, endogenous human KSR from A431 cells was capable of signaling c-Raf-1-dependent activation of the MAPK pathway in vitro, indicating that KSR kinase activity is not an artifact of overexpression, but a property intrinsic to this protein. Moreover, we defined EGF as a potent activator of KSR kinase activity. This study, however, did not address the mechanisms of c-Raf-1 stimulation by activated KSR.
Our prior study provided experimental evidence for KSR phosphorylation of Thr 269 , located within the CR2 domain of c-Raf-1, as a mechanism for c-Raf-1 activation in response to tumor necrosis factor-␣ and ceramide (30). KSR failed to phosphorylate or activate a Raf-1 mutant in which Thr 268 , a putative autophosphorylation site, and Thr 269 were both substituted with valine resides (Raf-1(T268V/T269V)). To examine the mechanism of c-Raf-1 activation by KSR in response to EGF stimulation and the significance of Thr 269 in c-Raf-1 activation, we generated recombinant human FLAG-c-Raf-1 and FLAG-Raf-1(T269V) (FLAG-TV-Raf) and purified them to homogeneity. Using our newly defined two-stage in vitro KSR activity assay, we show that Thr 269 is the major c-Raf-1 site phosphorylated by KSR and that transphosphorylation of this site is essential for c-Raf-1 activation by KSR in vitro. Furthermore, Thr 269 appears to be required for optimal activation of c-Raf-1 by KSR in response to physiologic EGF stimulation in vivo.

MATERIALS AND METHODS
Cell Culture and EGF Treatment-COS-7 cells (American Type Culture Collection) were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.), penicillin, and streptomycin at 37°C in 5% CO 2 (41). For EGF stimulation, cells were placed in serum-free medium for 12 h prior to the treatment.
For EGF studies, COS-7 cells were plated at a density of 1.5 ϫ 10 6 cells in 150-mm plates (Corning Inc.) and grown overnight to ϳ70% confluence. Culture medium was replenished with fresh medium 1 h before transfection, and cells were transfected with 5 g of DNA/plate using FuGene TM 6 transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. At 48 h post-transfection, cells were placed in serum-free medium for 12 h prior to treatment with EGF (0.001-100 ng/ml; Upstate Biotechnology, Inc.) for 3 min. Cells were then harvested in Nonidet P-40 lysis buffer (25 mM Tris (pH 7.5), 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin/soybean trypsin in-hibitor, and 5 mM NaVO 4 ). The homogenate was centrifuged at 10,000 ϫ g for 5 min at 4°C, the supernatant collected and precleared with protein A/G-agarose (Amersham Pharmacia Biotech), and protein content was measured using BCA Reagent A (Pierce). Lysates were divided into aliquots and stored at Ϫ80°C for subsequent use. KSR and c-Raf-1 expression were determined by Western blot analysis as described (41).
Immunoprecipitation of KSR and c-Raf-1-FLAG-tagged proteins were immunoprecipitated from 500-g (or as otherwise indicated) COS-7 lysates using 60 l of agarose-conjugated mouse anti-FLAG Ab M2 (Sigma, catalog no. 1205) as previously described (41). KSR or Raf-1 immune complexes were collected by centrifugation and washed five times with Nonidet P-40 lysis buffer containing 1.0 M NaCl to remove contaminating proteins (see below). The beads were subsequently washed once with reaction buffer (100 mM Tris (pH 7.5), 25 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM dithiothreitol, and 1 mM NaVO 4 ) before measuring kinase activity.
Immunoaffinity Purification of Human FLAG-c-Raf-1, FLAG-TV-Raf, and FLAG-c-Raf-1(K375M)-Lysates were generated from four 150-mm plates of COS-7 cells transfected with FLAG-c-Raf-1, FLAG-TV-Raf or FLAG-c-Raf-1(K375M) and diluted to 1 g/l with Nonidet P-40 lysis buffer. Raf-1 proteins were immunoprecipitated with 1 ml of agarose-conjugated anti-FLAG Ab M2 at 4°C for 3 h. c-Raf-1 immune complexes were collected by centrifugation and washed five times with Nonidet P-40 lysis buffer containing 1.0 M NaCl and once with ice-cold phosphate-buffered saline. The beads were then resuspended in 5 ml of phosphate-buffered saline and transferred to an equilibrated Poly-Prep chromatography column (Bio-Rad), and the gel bed was washed twice with 5 ml of phosphate-buffered saline. FLAG-Raf proteins were competitively displaced with 5 volumes of 0.5 ml of phosphate-buffered saline containing 100 g/ml FLAG octapeptide (Sigma). 5 l of each fraction was resolved by 10% SDS-PAGE, and proteins were silverstained using the Bio-Rad silver staining kit. The identity of eluted proteins was determined by Western blot analysis using a rabbit antihuman c-Raf-1 Ab (Upstate Biotechnology, Inc., catalog no. 06-901) and mouse anti-FLAG Ab M2 (Sigma, catalog no. F-3165). The Raf-1-containing fractions were pooled, concentrated with Microcon 3 columns (Amicon, Inc.), dialyzed overnight with storage buffer (50 mM Tris-HCl (pH 7.5), 0.1 mM EGTA, 0.1 mM EDTA, 0.1% 2-␤-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride, and 50% glycerol), and protein content was measured using BCA Reagent A.
Two-stage KSR Activity Assay and Raf Activity Assay-KSR kinase activity was measured by the two-stage assay as previously described (41). Briefly, in the first stage of the assay, immunopurified KSR was incubated at 30°C in 15 l of reaction buffer containing 60 M ATP, 7.5 mM MgCl 2 , and the indicated amounts (10 and 100 ng) of either human His-c-Raf-1 coexpressed with Ras and Lck in Sf9 insect cells (Upstate Biotechnology, Inc.) or FLAG-c-Raf-1 or FLAG-TV-Raf prepared as described above. After 10 min, KSR-containing beads were pelleted (14,000 ϫ g) for 3 min at 4°C, and 10 l of supernatant containing activated c-Raf-1 was added in the second stage of the assay to 20 l of reaction buffer containing 60 M ATP, 7.5 mM MgCl 2 , 0.1 g of unactivated murine GST-MEK1 (Upstate Biotechnology, Inc.), 1.0 g of unactivated murine GST-ERK2/MAPK (Upstate Biotechnology, Inc.), and 2 g of human GST-Elk-1 fusion protein (New England Biolabs). After 20 min, the reaction was stopped by addition of 10 l of Laemmli buffer. For some studies, kinase-inactive MEK1(K97M-MKK1) was used as substrate to assess Raf-1 activity in the second stage of the assay. Phosphorylated Elk-1 or MEK1 was resolved by 7.5 or 10% SDS-PAGE as indicated and visualized by Western blot analysis using rabbit antiphospho-Elk-1(Ser 383 ) Ab (catalog no. 9181, BioLabs) or anti-phospho-MEK1(Ser 217 /Ser 221 ) Abs (catalog no. 9121, BioLabs), respectively. For studies examining the activity of Raf-1 directly immunoprecipitated from COS-7 cells, the first stage of the assay was omitted, and 30 l of reaction mixture was used. It should be noted that a prior study showed that although triply transfected His-c-Raf-1 is active, it is only partially activated and can be further activated 6 -8-fold by KSR (28).
Phosphorylation of c-Raf-1 by KSR in Vitro-c-Raf-1 phosphorylation was determined by 32 P labeling and by Western blot analysis of threonine, serine, and tyrosine phosphorylation. For radiolabeling, immunoprecipitated KSR was washed and incubated for 30 min at 30°C in 30 l of reaction buffer containing 60 M ATP, 7.5 mM MgCl 2 , 30 Ci of [␥-32 P]ATP (3000 Ci/mmol), and 300 ng of commercially available Hisc-Raf-1, or FLAG-c-Raf-1 or FLAG-TV-Raf prepared in our laboratory. After 30 min, KSR-containing beads were pelleted as described above, and Raf-1 was quantitatively immunoprecipitated from 25 l of supernatant with 2 g of rabbit anti-human c-Raf-1 Ab. Raf-1 immune complexes were captured by 40 l of protein A/G beads (Amersham Pharmacia Biotech), resolved by 7.5% SDS-PAGE, and autoradiographed. For threonine, serine, and tyrosine phosphorylation studies, preparation of KSR immune complexes and Raf-1 phosphorylation reactions were carried out as described above, however, [␥-32 P]ATP was omitted from the reaction mixture. Raf-1 immunoprecipitated from the reaction mixture was resolved by 7.5% SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. Raf-1, phosphorylated on threonine, serine, or tyrosine residues, was visualized by Western blot analysis using a mouse anti-phosphothreonine Ab (Zymed Laboratories Inc., catalog no. 13-9200), a rabbit anti-phosphoserine Ab (Zymed Laboratories Inc., catalog no. 61-8100), or a rabbit anti-c-Raf-1(Tyr(P) 340 / Tyr(P) 341 ) Ab (BIOSOURCE, catalog no. 44-506), respectively, according to manufacturers' instructions.
Phosphatase assay-For these studies, 300 ng of purified human FLAG-c-Raf-1 was treated with 4 g of Type III potato acid phosphatase (Sigma) in the presence of 20 M each aprotinin and leupeptin for 30 min prior to or after KSR phosphorylation. Phosphatase treatment was terminated with 500 l of cold Nonidet P-40 lysis buffer containing 10 mM sodium vanadate. Thereafter, FLAG-c-Raf-1 was quantitatively immunoprecipitated with 40 l of anti-FLAG beads at 4°C for 2 h. Dephosphorylated c-Raf-1 was then used in either the first or second stage of the KSR activity assay as described above. For studies employing pre-dephosphorylated c-Raf-1, KSR remained throughout both stages of the assay, which had no discernible effect on reconstitution of MAPK signaling.

KSR Activates c-Raf-1 in Vitro by
Phosphorylation-Our previous investigation established a two-stage in vitro assay for KSR kinase activity (41). During the first stage, highly purified KSR is incubated with c-Raf-1 in a reaction mixture containing ATP. In the second stage of the assay, activated c-Raf-1 is separated from KSR and incubated with a reaction mixture containing unactivated GST-MEK, unactivated GST-MAPK, and a human GST-Elk-1 fusion protein. Thus, KSR never comes in contact with any recombinant kinase other than c-Raf-1 in this assay. Using this procedure, we demonstrated that KSR, washed to homogeneity with high salt, retained its activity toward c-Raf-1, indicating that the kinase activity was intrinsic to KSR. Furthermore, we showed that KSR kinase activity was enhanced by EGF stimulation.
Our previous study, however, did not address the mechanism of c-Raf-1 stimulation by activated KSR (41). To investigate whether activated KSR stimulates c-Raf-1 by direct phosphorylation, COS-7 cells transiently transfected with FLAG-KSR were treated for 3 min with 50 ng/ml EGF. We showed previously that this EGF dose and duration of treatment induce maximal KSR activation (41). Thereafter, KSR was immunoprecipitated, washed with high salt, and incubated with recombinant human His-c-Raf-1 substrate that had been coexpressed with Ras and Lck in Sf9 insect cells in a reaction mixture containing [␥-32 P]ATP as described (28,30). After the phosphorylation reaction, KSR-containing beads were pelleted, and His-c-Raf-1 was quantitatively immunoprecipitated from the supernatant to separate it from potential contaminating proteins in the commercial His-c-Raf-1 preparation (see Fig. 2A). 32 P-Labeled c-Raf-1 was resolved by 7.5% SDS-PAGE and autoradiographed (Fig. 1B). The amount of c-Raf-1 used in the assay was titrated so that in the absence of KSR, Raf-1 autophosphorylation was barely detectable (lanes 2 and 3). Under these conditions, a slight increase in c-Raf-1 phosphorylation measured by incorporation of radioactive phosphate was visualized after a 30-min incubation with KSR immunoprecipitated from untreated cells (lane 4). However, KSR immunoprecipitated from EGF-treated COS-7 cells markedly increased the incorporation of radioactive phosphate into c-Raf-1 (lane 5). Immunoprecipitation from the reaction mixture with normal rabbit IgG rather than the anti-c-Raf-1 Ab confirmed the specificity of the anti-c-Raf-1 Ab (lanes 1 and 6).
Activation of c-Raf-1 kinase is tightly regulated by multiple phosphorylation events. In resting cells, c-Raf-1 is phosphorylated exclusively on several serine and threonine residues. Serine and threonine phosphorylation can be further increased at selected sites upon stimulation with mitogenic stimuli and cytokines and is often accompanied by tyrosine phosphorylation at positions 340 and 341 (6, 20 -32). To investigate the contribution of threonine, serine, and tyrosine phosphorylation to the overall increased c-Raf-1 phosphorylation by KSR in vitro, Western blot analysis was performed using anti-phosphothreonine, anti-phosphoserine, and anti-phosphotyrosine anti- In vitro phosphorylation of c-Raf-1 by KSR immunoprecipitated (IP) from EGF-treated COS-7 cells was determined by 32 P labeling (B) and by Western analysis of threonine (C) and serine and tyrosine (not shown) phosphorylation. For these studies, COS-7 cells, transfected with mouse FLAG-KSR (B and C) and FLAG-Ki-KSR (C), were treated with 50 ng/ml EGF for 3 min. KSR was then immunoprecipitated from 500 g of lysate, purified to homogeneity, and used to phosphorylate human His-c-Raf-1 as described under "Materials and Methods" and summarized in A. Briefly, immunopurified KSR was incubated in a reaction mixture containing commercially available human His-c-Raf-1 with (B) or without (C) [␥-32 P]ATP. After 30 min, KSR-containing beads were pelleted by centrifugation; 25 l of supernatant (sup) containing c-Raf-1 was collected; and His-c-Raf-1 was precipitated with 2 g of rabbit anti-human c-Raf-1 Ab. In B, 32 P-labeled c-Raf-1 immune complexes were resolved by 7.5% SDS-PAGE and autoradiographed (lanes 2-5). Immunoprecipitation with normal rabbit IgG served as a control for the specificity of anti-c-Raf-1 Ab (lanes 1 and 6). In C, threonine phosphorylation was detected by Western blot (WB) analysis using a mouse anti-phosphothreonine Ab. The lower panel shows that c-Raf-1 levels in each lane were similar by Western analysis. These data represent one of three similar experiments each.
bodies. Treatment of COS-7 cells with 50 ng/ml EGF for 3 min markedly increased the capacity of immunoprecipitated KSR to increase c-Raf-1 threonine phosphorylation when reconstituted in vitro (Fig. 1C, lane 6). However, serine and tyrosine phosphorylation of c-Raf-1 by EGF-activated KSR was not detected (data not shown). Furthermore, Ki-KSR, whether isolated from untreated or EGF-treated COS-7 cells, failed to support threonine phosphorylation of recombinant His-c-Raf-1 (lanes 3 and 4), indicating that the kinase activity of KSR was necessary for c-Raf-1 phosphorylation. It should be noted that the level of c-Raf-1 was similar in all incubations (Fig. 1C, lower panel). These results suggest that threonine phosphorylation may be involved in c-Raf-1 activation by activated KSR and that the kinase activity of KSR is obligatory.
Immunoaffinity-purified Human FLAG-c-Raf-1 Can Support KSR-mediated MAPK Activation in Vitro, and Thr 269 of c-Raf-1 Is Required-Thr 269 is located within the CR2 domain of c-Raf-1, a region documented to critically regulate c-Raf-1 phosphorylation and activation (6). We previously proposed that this threonine might be important for phosphorylation and activation of c-Raf-1 by KSR in response to tumor necrosis factor-␣ and ceramide, as KSR failed to phosphorylate or activate a Raf-1 mutant in which Thr 268 , a putative autophosphorylation site, and Thr 269 were both substituted with valine residues (Raf-1(T268V/T269V)) (30). To further explore the possible regulation of c-Raf-1 by threonine phosphorylation and the significance of Thr 269 to c-Raf-1 activation, we generated a FLAG-tagged valine-substituted Thr 269 mutant (FLAG-Raf-1(T269V) or FLAG-TV-Raf). We expressed and isolated this mutant and wild-type recombinant human FLAG-c-Raf-1 from COS-7 cells by immunoaffinity chromatography. To purify these proteins to homogeneity, FLAG-c-Raf-1 and FLAG-TV-Raf immune complexes were subjected to five washes with Nonidet P-40 lysis buffer containing 1.0 M NaCl, loaded onto a polypropylene column, and then eluted using the FLAG octapeptide. The purity of our c-Raf-1 preparations was compared with that of commercial recombinant human His-c-Raf-1 from Sf9 cells after resolution by 10% SDS-PAGE and silver stain- ing. As shown in Fig. 2A, ϳ60% of the commercial His-c-Raf-1 preparation consisted of impurities as assessed by densitometric analysis. In contrast, high salt-washed and immunoaffinitypurified human FLAG-c-Raf-1 and FLAG-TV-Raf were homogeneous and free of contaminants (second and third lanes). Prolonged staining failed to detect any visible contaminants (data not shown). The availability of highly purified KSR immune complexes and recombinant Raf-1 proteins allowed us to directly assess the requirement of Thr 269 for KSR-mediated activation of MAPK in vitro.
To investigate the contribution of Thr 269 to threonine phosphorylation of c-Raf-1 by KSR in vitro, the capacity of our highly purified recombinant FLAG-c-Raf-1 and FLAG-TV-Raf to support c-Raf-1 threonine phosphorylation by KSR was determined by Western analysis using the anti-phosphothreonine antibody, as in Fig. 1C. Fig. 2B (lane 1) shows that autophosphorylation of recombinant human FLAG-c-Raf-1 on threonine residues was very low. Immunoprecipitates from COS-7 cells transfected with vector alone or from KSR-overexpressing cells left untreated failed to confer c-Raf-1 threonine phosphorylation (lanes 2-5). However, EGF treatment of COS-7 cells significantly increased the capacity of immunoprecipitated KSR to increase c-Raf-1 threonine phosphorylation in vitro (lane 6). These results are consistent with prior studies showing that recombinant human FLAG-c-Raf-1, purified to homogeneity, could be activated in vitro (26,(42)(43)(44)(45). Homogeneous FLAG-TV-Raf, however, was not phosphorylated by KSR preparations from EGF-treated or untreated cells (lanes 7 and 8). Furthermore, an equivalent amount of Ki-KSR (data not shown), whether isolated from untreated or EGF-treated COS-7 cells, failed to support threonine phosphorylation of recombinant FLAG-c-Raf-1 (lanes 9 and 10), indicating the necessity of kinase activity of KSR for c-Raf-1 phosphorylation. These studies indicate that the kinase activity of KSR and Thr 269 of c-Raf-1 are required for Raf-1 phosphorylation by KSR in vitro.
To determine whether Thr 269 phosphorylation regulates c-Raf-1 activity, we examined the capacity of purified FLAG-c-Raf-1 or FLAG-TV-Raf to support KSR-mediated activation of MAPK using the two-stage kinase assay as described above (41). A high (100 ng) and a low (10 ng) dose of purified recombinant human FLAG-c-Raf-1 or FLAG-TV-Raf was tested using KSR isolated from lysates of EGF-treated COS-7 cells (Fig. 3A). Immunoprecipitates from COS-7 cells transfected with vector alone failed to confer activity onto recombinant FLAG-c-Raf-1,

FIG. 3. Purified human FLAG-c-Raf-1 supports KSR-mediated MAPK activation in vitro, and Thr 269 of c-Raf-1 is required.
A, the capability of 10 or 100 ng of purified FLAG-c-Raf-1 or FLAG-TV-Raf to support KSR signaling of MAPK activation in vitro was assessed by the two-stage KSR activity assay as previously described (41). Briefly, in the first stage of the assay, FLAG-tagged KSR immunopurified from EGF-treated COS-7 cells was incubated in 15 l of reaction buffer containing 60 M ATP, 7.5 mM MgCl 2 , and the indicated amounts (10 and 100 ng) of either His-tagged or triply transfected human c-Raf-1 or FLAG-tagged Raf-1 prepared as described under "Materials and Methods. as measured by reconstitution of the MAPK cascade using Elk-1 phosphorylation as readout (lanes 1-3). KSR immunoprecipitated from EGF-treated COS-7 cells, when incubated in the first stage of the assay without Raf-1, also yielded no effect (lane 4). However, KSR immunoprecipitated from EGF-treated COS-7 cells activated 10 ng of FLAG-c-Raf-1 about as effectively as it activated an equal amount of triply transfected commercial His-c-Raf-1 ( lane 5 versus lane 7). Increasing the amount of recombinant human FLAG-c-Raf-1 in this assay to 100 ng yielded no further increase in KSR-initiated reconstitution of the MAPK cascade in vitro (lane 6), indicating that 10 ng of FLAG-c-Raf-1 was optimal under our assay conditions. These results show that immunoaffinity-purified recombinant human FLAG-c-Raf-1, although free of contamination, is replete in mediating KSR signaling. In contrast, substitution of threonine with valine at amino acid 269 of c-Raf-1 (lane 8) abrogated KSR-mediated Raf-1-dependent MAPK activation. The lack of signaling by 10 ng of TV-Raf was not due to an insufficient quantity of TV-Raf in the assay, as 100 ng of FLAG-TV-Raf also failed to elicit MAPK activation (lane 9).
To provide evidence that it is the threonine phosphorylation of c-Raf-1 by KSR that dictates activation, we dephosphorylated c-Raf-1 before or after KSR treatment using potato acid phosphatase, prior to reconstitution of the MAPK cascade. As shown in Fig. 3B, although potato acid phosphatase treatment abrogated KSR activation of c-Raf-1 (lane 6 versus lane 5), c-Raf-1 dephosphorylated prior to KSR treatment could still be partially activated (lane 7). These data show that KSR phosphorylates c-Raf-1 on Thr 269 and that this phosphorylation site is essential for c-Raf-1 activation by KSR.
In a prior study, we showed that endogenous KSR from A431 cells, which signal constitutively through the EGF receptor, was constitutively active (41). Endogenous KSR purified from unstimulated A431 cells, like overexpressed KSR from EGFtreated COS-7 cells, phosphorylated c-Raf-1 on threonine residues and increased c-Raf-1 activity (data not shown). (In fact, endogenous KSR isolated from A431 cells is ϳ1.6-fold more active than FLAG-tagged KSR isolated from EGF-treated COS-7 cells.) Endogenous KSR from A431 cells, however, neither phosphorylated nor activated TV-Raf (data not shown). Thus, the phosphorylation of Thr 269 and the activation of c-Raf-1 in vitro by KSR are not an artifact of KSR transfection, but a property intrinsic to this protein.
Physiologic Doses of EGF Activate KSR in Vivo-Recently, Wennstrom and Downward (46) proposed that low dose EGF activated the Ras/MAPK signaling pathway by a different mechanism than high dose EGF in COS-7 cells. To better understand the involvement of KSR in intact cells, we reevaluated the dose dependence of EGF to enhance KSR kinase activity using the two-stage assay. For these studies, we treated COS-7 cells, transiently transfected with FLAG-KSR or vector, with varying doses of EGF (0.001-100 ng/ml for 3 min), isolated KSR from lysates, and reconstituted the MAPK cascade using our homogeneous preparation of human c-FLAG-Raf-1 (41). As anticipated, immunoprecipitates that lack KSR, from cells transfected with vector only, did not activate the MAPK cascade (Fig. 4A, lanes 1 and 2). However, EGF conferred dose-dependent activation of KSR kinase activity (lanes  4 -8). The maximal KSR activity was achieved with 0.1 ng/ml EGF (lane 6). This dose is within the physiologic range for EGF-stimulated mitogenesis (47). (Note that this and subsequent studies resolved phosphorylated Elk-1 by 10% SDS-PAGE to separate multiply phosphorylated Elk-1 forms (see below).) To determine whether KSR activation corresponded to c-Raf-1 threonine phosphorylation in vivo, the phosphorylation pattern after low and high dose EGF was determined in cells transiently transfected with FLAG-c-Raf-1 in addition to Myc-KSR. Myc-tagged KSR constructs were employed to avoid coimmunoprecipitation of KSR with c-Raf-1. Fig. 4B shows that cells cotransfected with Myc-tagged KSR and FLAG-c-Raf-1 displayed minimal threonine phosphorylation (upper panel, lane 3), which was markedly increased upon treatment with 0.1 ng/ml EGF, a dose that maximally activated KSR (upper panel, lane 4). (Note that cells transfected with c-Raf-1 alone displayed no threonine phosphorylation (Fig. 5B, lane 3); the modest threonine phosphorylation observed in Fig. 4B due to KSR coexpression is examined in detail in Fig. 5B.) High dose EGF (100 ng/ml), which induced no further increase in KSR kinase activity, also induced no further increase in threonine phosphorylation (Fig. 4B, upper panel, lane 8 versus lane 4).
In contrast to enhanced threonine phosphorylation, at low dose EGF, there was no effect on serine or tyrosine phosphorylation. Neither cells expressing c-Raf-1 alone (data not shown) nor cells coexpressing Myc-KSR displayed c-Raf-1 tyrosine phosphorylation (Fig. 4B, middle panel, lane 3). Furthermore, low dose EGF had no obvious effect on tyrosine phosphorylation (middle panel, lane 4), whereas high dose EGF, as often reported (21,23,48,49), effected substantive Tyr 340 / Tyr 341 phosphorylation (lane 8). As described by others (6,22,25,29,50), c-Raf-1 was phosphorylated on serine residues in resting cells whether expressed alone (data not shown) or in cells coexpressing Myc-KSR (lower panel, lane 3). As with tyrosine phosphorylation, low dose EGF had no effect on serine phosphorylation, but increased serine phosphorylation at high doses (lower panel, lane 4 versus lane 8). These studies indicate Substitution of Thr 269 of c-Raf-1 Blocks Raf-1 Activation by KSR in Vivo-We next assessed the requirement of Thr 269 for c-Raf-1 activation by KSR in vivo using the 0.1 ng/ml EGF dose. Fig. 5A (lanes 1 and 2) shows that anti-FLAG-Raf-1 immunoprecipitates from COS-7 cells transfected with vector alone displayed no significant effect to reconstitute MAPK activation. As reported by others (23,26,42,44), immunoprecipitated FLAG-c-Raf-1 alone, in the absence of EGF treatment, also exhibited no apparent kinase activity (lane 3). Upon stimulation with 0.1 ng/ml EGF, FLAG-c-Raf-1 activity was significantly increased (lane 4). Multiple supershifted forms of phosphorylated Elk-1 were observed, reflecting different degrees of Elk-1 phosphorylation. Yang et al. (51) similarly observed multiply phosphorylated forms of Elk-1 in an in vitro kinase assay employing MAPK isolated from EGF-treated COS-1 cells. Increased c-Raf-1 activity was also detected after EGF stimulation by direct phosphorylation of Ki-MEK1(K97M-MKK1) (data not shown). In contrast, substitution of Thr 269 with Val abrogated the enhanced kinase function of FLAG-c-Raf-1, rendering TV-Raf incapable of transducing an EGF signal (lane 6). These data support a potentially important role of Thr 269 in c-Raf-1 function under physiologic conditions in vivo.
The capacity of KSR to activate c-Raf-1 in vivo was further evaluated by coexpression of Myc-KSR and FLAG-c-Raf-1. Myc-tagged KSR coexpressed with FLAG-c-Raf-1, in the absence of EGF, substantially enhanced the ability of immunoprecipitated FLAG-c-Raf-1 to activate the MAPK cascade in vitro (Fig. 5A, lane 7), comparable to that of FLAG-c-Raf-1 when expressed alone and immunoprecipitated from EGFstimulated COS-7 cells (lane 4). Moreover, EGF treatment markedly augmented the capacity of KSR to activate c-Raf-1, producing multiple supershifted Elk-1 bands of high intensity (lane 8). In contrast to wild-type KSR, Ki-KSR failed to support c-Raf-1 activation with or without EGF treatment (lanes 9 and 10). In fact, Ki-KSR appeared to serve a dominant-negative function, blocking EGF-induced c-Raf-1 activation (compare lanes 4 and 10). These results are consistent with our published study that showed that KSR mediated and Ki-KSR prevented ceramide-induced apoptosis through Ras and c-Raf-1 in BADoverexpressing COS-7 cells (40). These results clearly indicate that the kinase function of KSR is required for it to act as a c-Raf-1 activator in vivo as well as in vitro (Figs. 1 and 2). The significance of Thr 269 in c-Raf-1 activation by KSR in response to EGF stimulation was further investigated in COS-7 cells cotransfected with wild-type KSR and TV-Raf. As observed in the in vitro studies, this single amino acid substitution abrogated c-Raf-1 activation by KSR in vivo (Fig. 5A, lanes 11 and  12). It should be noted that cotransfection of Myc-tagged KSR and FLAG-tagged Raf-1 constructs did not affect the level of their respective expression in these studies (Fig. 5A, lower  panel).
If phosphorylation of Thr 269 is necessary for c-Raf-1 activation by KSR, it is anticipated that TV-Raf would be defective in this event. Although c-Raf-1 overexpressed in COS-7 cells was not phosphorylated on threonine residues (Fig. 5B, lane 3), KSR coexpression, which resulted in c-Raf-1 activation, con- ferred threonine phosphorylation (lane 7). Low dose EGF treatment, which enhanced c-Raf-1 activation in the presence of KSR, further increased threonine phosphorylation (lane 8). However, Ki-KSR failed to support c-Raf-1 threonine phosphorylation with or without EGF treatment (lanes 9 and 10). Moreover, as in Fig. 5A, Ki-KSR appeared to serve a dominantnegative function, abrogating c-Raf-1 phosphorylation (Fig. 5B,  lane 10 versus lane 4). In contrast, TV-Raf, which could not be activated by KSR with or without EGF, failed to display threonine phosphorylation (lanes 6 and 12). TV-Raf is not globally inactive however, as high dose EGF (100 ng/ml) induced serine and Tyr 340 /Tyr 341 phosphorylation of TV-Raf and stimulated partial activation (data not shown). Collectively, these results strongly argue that KSR signals through Thr 269 to activate c-Raf-1 in response to physiologic EGF stimulation in vivo.

DISCUSSION
This investigation extends our studies on the role of KSR in c-Raf-1 activation using highly purified reagents. In our prior investigation, we showed, using a two-stage assay in which KSR comes in contact only with commercially available His-c-Raf-1, that purified KSR from EGF-treated cells activates c-Raf-1 (41). Here, we show that EGF-activated KSR phosphorylated c-Raf-1 during the first stage of this assay, specifically on threonine residues. The mechanism of KSR-mediated c-Raf-1 phosphorylation appears to be via transphosphorylation rather than increased c-Raf-1 autophosphorylation, as Ki-c-Raf-1(K375M) effectively served as a KSR substrate. Phosphorylation appears to be required for c-Raf-1 activation, as subsequent dephosphorylation with potato acid phosphatase reversed the KSR effect. Consistent with a requirement for threonine phosphorylation in c-Raf-1 activation by KSR, c-Raf-1 in which Thr 269 , the putative KSR phosphorylation site, has been substituted with valine could be neither phosphorylated nor activated by KSR in vitro. In these latter studies, a great deal of attention was paid to utilizing c-Raf-1 reagents purified to homogeneity, as the commercially available material was contaminated with numerous unknown proteins. However, Thr 269 phosphorylation alone may be insufficient for activation, as bacterially expressed c-Raf-1 cannot be activated by KSR, 2 suggesting that an unknown post-translational modification of c-Raf-1 primes it for KSR activation. Consistent with this notion, c-Raf-1, dephosphorylated prior to KSR treatment, could still be partially activated by KSR-dependent phosphorylation. Whether priming represents an irreversible postphosphorylation conformational change in c-Raf-1 by kinases other than KSR or an unknown post-translational modification will require further investigation. That Thr 269 was also required for c-Raf-1 activation in vivo was determined by overexpression of wild-type or TV-Raf-1 protein with or without kinase-active or -inactive KSR and then stimulation of cells with EGF. In contrast to wild-type c-Raf-1, TV-Raf-1 could not be activated by physiologic concentrations of EGF or by concomitant overexpression of kinase-active KSR. Furthermore, the combination that delivered the largest activation, coexpression of wild-type KSR with c-Raf-1 and then treatment with physiologic doses of EGF, was ineffective in the absence of Thr 269 . However, TV-Raf is not globally inactive, as it can be fully activated in vivo by 12-O-tetradecanoylphorbol-13-acetate (30) or partially activated by super-physiologic doses of EGF. These studies provide evidence that Thr 269 is required for KSR-dependent c-Raf-1 activation in vivo.
Evidence that the kinase activity of KSR is required for c-Raf-1 activation in vivo as well as in vitro was established. Ki-KSR, in which two conserved aspartates involved in phos-pho-transfer were substituted with alanine residues, did not increase c-Raf-1 phosphorylation or activation in vitro, indicating that simple contact of c-Raf-1 with KSR was insufficient for either event. However, this investigation does not rule out a direct interaction between these proteins, as both proteins used in the first stage of the reconstitution assay were purified to homogeneity. Similarly, Ki-KSR neither activated c-Raf-1 in vivo nor supported EGF-dependent c-Raf-1 activation. In fact, Ki-KSR blocked EGF-induced c-Raf-1 activation, consistent with prior studies in COS-7 cells that showed that Ki-KSR served a dominant-negative function, antagonizing ceramideinduced apoptosis signaled through c-Raf-1 (40).
Activated KSR, immunoprecipitated from COS-7 cells treated with EGF, phosphorylated c-Raf-1 in vitro exclusively on threonine residues (Figs. 1 and 2). Under our assay conditions, increases in serine and tyrosine phosphorylation of Raf-1 by activated KSR were not observed. This is consistent with the recognition that mammalian KSRs contain the motif YI(L)APE in subdomain VIII, which is conserved among serine/threonine kinases (33). Direct tyrosine phosphorylation of Raf-1 by KSR was not expected since mammalian KSR does not contain the conserved sequence HKDLR required for tyrosine kinase activity. This latter sequence is present in both C. elegans and D. melanogaster KSR (33), suggesting that KSR activation of C. elegans and Drosophila Raf might occur through a fundamentally different mechanism than in the mammalian systems.
It is noteworthy that maximal activation of KSR as measured by reconstitution of its activity in vitro occurred with physiologic doses of EGF (0.1 ng/ml). Recently, it has become clear that these low doses activate a specific high affinity subset of EGF receptors to signal Ras/MAPK-mediated mitogenesis in vivo (46). Moreover, the high EGF doses commonly used to stimulate cells in culture often yield antiproliferative signals (52)(53)(54)(55)(56)(57). Strength of stimulation as a determinant in the assembly of different downstream signaling complexes has been reported during platelet-derived growth factor-and EGF-induced MAPK activation in COS-7 cells (58). For EGF, Wennstrom and Downward (46) recently suggested that low dose EGF may activate an Shc-Grb2-Sos complex that is not EGF receptor-associated to induce Ras activation, whereas high dose EGF stably recruits Shc and Grb2 to the EGF receptor. Furthermore, a protein kinase C-mediated pathway through c-Raf-1 to ERK (46, 59, 60) may be initiated only at high EGF concentrations sufficient for phosphorylation of phospholipase C␥. Whether receptor-associated adaptor complexes and protein kinase C activation are involved in the Tyr and Ser phosphorylation, respectively, observed at high EGF doses requires further investigation.
In this study, we further define the mechanism of KSR activation of c-Raf-1. In vitro, KSR from EGF-stimulated cells phosphorylates and transactivates human c-Raf-1. The kinase activity of KSR and Thr 269 of c-Raf-1 are the minimal requirements for this event. Similarly, these properties are required for optimal c-Raf-1 activation in vivo. That physiologic EGF is sufficient for maximal KSR signaling implies that this mechanism is involved in proliferative responses upon signaling through the EGF receptor. A better understanding of the mechanism of KSR as a Ras effector and Raf-1 activator in vitro should help elucidate the biological function of this protein in vivo.