KSR-1 Binds to G-protein βγ Subunits and Inhibits βγ-induced Mitogen-activated Protein Kinase Activation*

The protein kinase KSR-1 is a recently identified participant in the Ras signaling pathway. The subcellular localization of KSR-1 is variable. In serum-deprived cultured cells, KSR-1 is primarily found in the cytoplasm; in serum-stimulated cells, a significant portion of KSR-1 is found at the plasma membrane. To identify the mechanism that mediates KSR-1 translocation, we performed a yeast two-hybrid screen. Three clones that interacted with KSR-1 were found to encode the full-length γ10 subunit of heterotrimeric G-proteins. KSR-1 also interacted with γ2and γ3 in a two-hybrid assay. Deletion analysis demonstrated that the isolated CA3 domain of KSR-1, which contains a cysteine-rich zinc finger-like domain, interacted with γ subunits. Coimmunoprecipitation experiments demonstrated that KSR-1 bound to β1γ3 subunits when all three were transfected into cultured cells. Lysophosphatidic acid treatment of cells induced KSR-1 translocation to the plasma membrane from the cytoplasm that was blocked by administration of pertussis toxin but not by dominant-negative Ras. Finally, transfection of wild-type KSR-1 inhibited β1γ3-induced mitogen-activated protein kinase activation in cultured cells. These results demonstrate that KSR-1 translocation to the plasma membrane is mediated, at least in part, by an interaction with βγ and that this interaction may modulate mitogen-activated protein kinase signaling.

The Ras signaling pathway affects many aspects of cell physiology, including cell growth, proliferation, movement, and differentiation (1). Recently, KSR-1 1 was identified as a component of the Ras signaling cascade by genetic screens in Drosophila melanogaster and Caenorhabditis elegans (2)(3)(4). Inactivating mutations in the ksr-1 gene blocked the phenotypic effects of activated Ras in these animals, suggesting that KSR-1 is a positive regulator of Ras-mediated signaling. Genetic epistasis experiments in Drosophila demonstrated that KSR-1 acts downstream of Ras but upstream of or parallel to Raf (4).
Mammalian forms of KSR-1 have been identified on the basis of sequence homology (4), but the role of mammalian KSR-1 in Ras-mediated signaling is controversial. Overexpression of KSR-1 in Xenopus oocytes was found by two groups (including ours) to weakly promote MAP kinase activation (5)(6)(7). In one study (6), overexpression of KSR-1 in cultured mammalian cells was found to promote MAP kinase activation; it was shown to inhibit Ras-mediated signaling at the level of MEK activation in several studies (8 -10), and it was found to inhibit Ras-mediated signaling at the level of transcription factor (Elk-1) activation (11). In the absence of loss-of-function studies in mammalian cells, the definitive role of KSR-1 in Rasmediated signaling remains unclear.
Both invertebrate and mammalian forms of KSR-1 consist of a putative amino-terminal regulatory portion and a carboxylterminal serine/threonine kinase domain. Five functional domains of KSR-1 have been identified, including a unique amino-terminal CA1 domain, a proline-rich CA2 domain, a cysteine-rich zinc finger-like CA3 domain, a serine/threonine rich CA4 domain, and the amino-terminal protein kinase CA5 domain (4). KSR-1 is most homologous to Raf-1 kinase, but there is no evidence that KSR-1 can bind to Ras or phosphorylate MEK. Indeed, the in vivo substrate(s) of the kinase domain of KSR-1 is unknown (7,11).
One model of KSR-1 action purports that it is a molecular scaffold that behaves like the budding yeast protein ste5, functionally linking the protein kinases ste11, ste7, and fus3/kss1 (6). Indeed, KSR-1 has been shown to interact with 14 -3-3 protein, Raf-1, MEK, and MAP kinase in coimmunoprecipitation experiments and yeast two-hybrid assays (5-10). It is not clear, however, whether KSR-1 links these associated proteins to promote signal transduction. Mutational analysis of KSR-1 has revealed that separable domains bind to distinct signaling proteins. For example, in vitro binding assays have demonstrated that the CA4 domain of KSR-1 interacts with MAP kinase (12). Furthermore, yeast two-hybrid assays have shown that the CA5 domain of KSR-1 binds to MEK (8,9).
The subcellular localization of KSR-1 is dependent on the activation state of cells. We previously demonstrated that KSR-1 is a cytoplasmic protein in serum-starved cells, but that KSR-1 translocates to the plasma membrane after stimulation with serum (5). The time course of this translocation is similar to that observed for Raf-1 kinase, which binds to activated Ras at the plasma membrane. Work by Michaud et al. (7) has established that the cysteine-rich CA3 domain of KSR-1 is essential for translocation to the plasma membrane. One explanation for this observation is that KSR-1 accompanies Raf-1 to the plasma membrane, but inactive Raf-1 does not bind to KSR-1 (5). Another possibility is that KSR-1 directly binds to a constitutively membrane-bound target after serum stimulation. To further explore this possibility, we performed a yeast two-hybrid screen using KSR-1 as bait.
* 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 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screen-A cDNA encoding the CA2 through CA5 domains (amino acids 195 to 873) of murine KSR-1 (mKSR-1) was inserted into the vector pAS1-CYH (gift of Stephen Elledge, Baylor University, Houston, Texas) as an in-frame fusion with the transactivation domain of GAL4 (pAS1/CA2-5) as described previously (12). A human HeLa cell cDNA two-hybrid library (gift of David Beach, Cold Spring Harbor Laboratory, New York) was screened and pAS1/CA2-5 was used as the bait. A yeast strain (Y190) was cotransfected with pAS1/CA2-5 and the HeLa cell library and yeast were plated onto medium lacking histidine, tryptophan, and leucine (13). Colonies that grew in the absence of histidine were assayed for ␤-galactosidase activity by use of X-gal (5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside) as a substrate. Yeast that contained only the cDNA library plasmid were mated with yeast strain Y189 containing either pAS1/CA2-5 or pAS1/lamin, grown on selective medium, and re-assayed for ␤-galactosidase activity. Clones that specifically interacted with CA2-5 were sequenced and BLAST searches were performed (National Center for Biotechnology Information).
mKSR-1 Deletion Analysis-The CA2 domain of mKSR1 (residues 234 -300), and the CA3 domain (residues 303-397) were inserted into pAS1-CYH as in-frame fusions with the DNA binding domain of GAL4 (residues 1-147) (13). The two-hybrid construct containing the CA5 domain (residues 541-873) of mKSR1 was a gift of Wendy Fantl (Chiron Corporation, Emeryville, CA). Human ␥ 10 was used as an in-frame fusion with the transactivation domain of GAL4 (residues 768 -881) in the vector pGAD GH (13). Human ␥ 2 and ␥ 3 were used as in-frame fusions with the transactivation domains of GAL4 in the vector pACTII (14). Yeast strain Y190 was cotransformed with ␥ subunits and the mKSR-1 mutants or lamin. Colonies were assessed for ␤-galactosidase activity by use of X-gal as a substrate (13).
Mammmalian Expression Constructs-The full-length mKSR-1 cDNA (gift of Marc Therrien and Gerald Rubin, University of California, Berkeley) was subcloned into the pTarget (Promega) mammalian expression vector (4). The human wild-type ␤ 1 and ␥ 3 cDNAs were subcloned into the mammalian expression vector pCB6ϩ (14). A 5Ј-FLAG-tagged version of mKSR-1 was obtained by use of polymerase chain reaction and was subcloned directly into pTarget (Promega) and sequenced. The N17 Ras cDNA was a gift from Dwight Towler (Washington University, St. Louis, MO).
Antibodies-The rabbit polyclonal anti-␤ 1 subunit antibody (BN-1) has been previously described (15). The rabbit polyclonal anti-pan-␤ subunit, the goat polyclonal anti-KSR-1, and the rabbit polyclonal anti-ERK1 antibodies were obtained from Santa Cruz Biotechnology. The alkaline phosphatase-conjugated rabbit anti-goat IgG secondary antibody was obtained from Zymed Laboratories Inc. The alkaline phosphatase-conjugated goat anti-rabbit IgG secondary antibody was obtained from Santa Cruz Biotechnology.
Cell Culture and Transfections-COS7 cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Cells were split 24 h before transfection with the indicated cDNAs. All transfections were carried out with the use of LipofectAMINE Plus reagent (Life Technologies, Inc.) in serum-free medium (OptiMEM, Life Technologies, Inc.).
Coimmunoprecipitation Assays-For triple transfection experiments, 1-3 ϫ 10 6 COS7 cells in 100-mm culture plates were transfected with the cDNAs encoding ␤ 1 (2.5 g), ␥ 3 (2.5 g), and mKSR-1 (2.5 g). Two days later, transfected COS7 cells were lysed in Nonidet P-40 lysis buffer and cleared by low speed centrifugation (5). Goat anti-KSR-1 antibody (Santa Cruz Biotechnology, 1:100 dilution) or goat anti-rabbit FIG. 1. The CA3 domain of mKSR-1 interacts with G-protein ␥ subunits by two-hybrid assay. A, mKSR-1 constructs used for two-hybrid analysis. The full-length mKSR-1 cDNA was used as a template to generate several deletion constructs for two-hybrid analysis that were subcloned as in-frame fusion proteins with the transactivation domain of the GAL4 transcription factor in the vector pAS1. B, detection of interactions between mKSR-1 and ␥ subunits. The cDNAs encoding human ␥ subunits were subcloned as in-frame fusion proteins with the DNA-binding domain of the GAL4 transcription factor in the vector pGAD. The DNA-binding domain and transactivation domain fusion constructs were cotransfected into a yeast strain (Y190) and grown on medium lacking leucine and tryptophan. Colony lifts were analyzed for ␤-galactosidase activity by use of X-gal, and colonies that exhibited detectable activity were scored positive (ϩ).
IgG (Santa Cruz Biotechnology) was added to lysates and incubated for 90 min at 4°C. Protein A/G Agarose (Santa Cruz Biotechnology) was used to immobilize antibody-bound proteins. Immunoprecipitates were washed three times with Nonidet P-40 lysis buffer and analyzed by SDS-PAGE and immunoblotting as above.
Subcellular Fractionation-Two days before fractionation, COS7 cells in 100-mm culture plates were transfected with the cDNAs encoding mKSR-1 (2.5 g) or mKSR-1 (2.5 g) and N17 Ha-Ras (2.5 g) (as described above). Transfected cells were allowed to grow for 24 h and were then serum starved (Dulbecco's modified Eagle's medium with 1% bovine serum albumin) overnight with or without added pertussis toxin (PTX) (Life Technologies, Inc., 200 ng/ml). After serum starvation, some COS7 cells were stimulated with 10 M lysophosphatidic acid (LPA) or 10% FCS for 10 min at 37°C. Cells were washed two times with phosphate-buffered saline, scraped into detergent-free lysis buffer (137 mM NaCl, 50 mM NaF, 6 mM MgCl 2 , 10 mM Tris, pH 7.5, 2 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 4 g/ml aprotinin, and 1 mM Na 3 VO 4 ), and sonicated to disrupt cell membranes. Lysates were cleared by low speed centrifugation (12,000 ϫ g for 5 min), cleared lysates were separated by high speed centrifugation (100,000 ϫ g for 1 h) (16,17). Supernatants were reserved, and pellets were washed twice in detergent-free lysis buffer and resuspended in an equal volume of lysis buffer with 1% Triton X-100. Fractions were resolved by use of SDS-PAGE and immunoblotting as above.
MAP Kinase Assays-Two days before their use, COS7 cells in 6-well dishes (Falcon) were transfected with cDNAs encoding ␤ 1 and ␥ 3 subunits (0.5 g each) with or without the cDNA encoding mKSR-1 (0.5 g) as described above. Cells were serum starved overnight (Dulbecco's modified Eagle's medium with 1% bovine serum albumin) before they were lysed in Nonidet P-40 lysis buffer. Lysates were cleared by low speed centrifugation and then incubated at 4°C for 90 min with murine Bacterially expressed GST fusion proteins were immobilized on glutathione beads, and samples were incubated with NIH/3T3 cell protein lysates. The beads were washed and adherent proteins were analyzed by immunoblotting by use of an an anti-pan-␤ subunit antibody (Santa Cruz Biotechnology). The fusion proteins contained protein fragments corresponding to GST alone (GST), the CA1 and CA2 domains (GST-CA1CA2), the CA3 domain (GST-CA3), the CA4 domain (GST-CA4), and the CA5 domain (GST-CA5) of mKSR-1. B, in vivo association of mKSR-1with ␤ 1 ␥ 3 subunits in transfected cells. COS7 cells were triple transfected with mammalian expression vectors encoding wild-type mKSR-1, human ␤ 1 , and human ␥ 3 . Anti-KSR or control (rabbit IgG) immunoprecipitates were analyzed by immunoblotting by use of an anti-␤ 1 subunit antibody (BN-1). Protein lysates derived from 5 ϫ 10 5 cells were used for each immunoprecipitation. This immunoblot is representative of the results of three separate experiments. C, in vivo association of mKSR-1 with ␤␥ subunits in untransfected cells. Untransfected COS7 cells were cultured in the absence of serum for 24 h and then some were stimulated with 10% fetal bovine serum for 10 min. Anti-KSR or control (rabbit IgG) immunoprecipitates were analyzed by immunoblotting by use of an anti-pan-␤ subunit antibody. Protein lysates derived from 10 7 cells were used for each immunoprecipitation. This immunoblot is representative of the results of four separate experiments.

FIG. 3. Translocation of mKSR-1 from the cytosolic fraction to the membrane fraction of cell lysates.
A, mKSR-1 translocates to the membrane fraction in response to LPA stimulation. COS7 cells that were transfected with wild-type mKSR-1 were cultured in the absence of serum (Unstimulated) and then were stimulated with LPA or 10% FCS. Some cells were preincubated with PTX before LPA stimulation (LPA/PTX). Detergent-free cell lysates were precleared, and then separated by high speed centrifugation (100,000 ϫ g for 1 h). Supernatants (S100) were reserved, and pellets (P100) were resuspended in an equal volume of buffer with added 1% Triton X-100. Fractions were analyzed by immunoblotting by use of an anti-KSR-1 or an anti-pan-␤ subunit polyclonal antibody (Santa Cruz Biotechnology). B, mKSR-1 translocation in response to LPA is blocked by PTX, but not by dominantnegative (N17) Ras. COS7 cells were transfected with wild-type mKSR-1 or with mKSR-1 and N17 Ras. Cells were cultured in the absence of serum, and some were stimulated with LPA or 10% FCS, whereas others were preincubated with PTX before LPA stimulation. Cell lysates were separated and analyzed as described in A. Three subcellular fractionation experiment were downloaded to a computer and analyzed by densitometry with NIH Image software. The data are presented as the mean percent of mKSR-1 in the membrane fraction (membrane fraction ϩ cytosolic fraction ϭ 100%) Ϯ S.E. monoclonal anti-ERK antibody (Santa Cruz Biotechnology, 1:100) in 500-l reactions containing 100 g of total protein. After incubation with anti-ERK antibody, 20 l of protein A/G-agarose (Santa Cruz Biotechnology) was added to each reaction to immobilize antibodybound proteins. Immunoprecipitates were washed twice in Nonidet P-40 lysis buffer to which NaCl had been added (1 M final concentration) and once in 25 mM HEPES (pH 7.45) plus 2 mM phenylmethylsulfonyl fluoride, and 1 mM Na 3 VO 4 . Immunoprecipitates were then resuspended in kinase buffer (25 mM HEPES (pH 7.45), 10 mM MgCl 2 , 1 mM dithiothreitol, and 50 M ATP) and incubated with 2 g of myelin basic protein (Sigma) and 20 Ci [␥-32 P]ATP for 20 min at room temperature. Samples were resolved by SDS-PAGE and visualized by autoradiography.

RESULTS
We performed a yeast two-hybrid screen using a cDNA that encoded the CA2 through CA5 domains of mKSR-1 (CA2-5) as bait with a HeLa cell two-hybrid library. Three positive clones were found to encode the ␥ 10 subunit of heterotrimeric Gproteins (18); the subunit did not interact with the protein kinase Mos or with nuclear lamin on two-hybrid assay. In subsequent experiments with the two-hybrid assay to determine whether the interaction between CA2-5 and ␥ subunits was specific for ␥ 10 , we found that ␥ 2 and ␥ 3 also bound to CA2-5 (Fig. 1). Several additional deletion mutant forms of mKSR-1 were then generated to identify the portion of mKSR-1 that was interacting with these subunits (Fig. 1A). The cys-teine-rich zinc finger-like CA3 domain of mKSR-1 interacted with ␥ 2 , ␥ 3 , and ␥ 10 in the two-hybrid assay (Fig. 1B), whereas neither the proline-rich CA2 domain nor the protein kinase CA5 domain of mKSR-1 interacted with any of the subunits. A construct that contained the CA4 domain was found to be transcriptionally active on its own in yeast.
In living cells, ␤ and ␥ subunits are obligatorily bound to each other (19). We therefore investigated the ability of the isolated CA3 domain of mKSR-1 to bind to ␤␥ subunits derived from cultured cell protein lysates. Recombinant GST fusion proteins that contained the CA1 and CA2 domains, the CA3 domain, the CA4 domain, or the CA5 domain of mKSR-1 were immobilized on glutathione beads and were incubated with NIH/3T3 cell protein lysates. Adherent proteins were analyzed by anti-␤ subunit immunoblotting, because the larger ␤ subunit is more readily detectable on immunoblots than the ␥ subunit, and revealed that ␤␥ subunits bound to immobilized GST-CA3, but not to GST-CA1CA2, GST-CA4, or GST-CA5 ( Fig. 2A). We have previously demonstrated that GST-CA4 specifically interacts with MAP kinase (12), and that GST-CA5 binds to MEK. 2 The ability of KSR-1 to interact with ␤␥ subunits in vivo was examined in coimmunoprecipitation experiments. COS7 cells were triple transfected with ␤ 1 , ␥ 3 , and mKSR-1. Anti-KSR immunoprecipitates obtained from transfected cell protein lysates were analyzed by anti-␤ subunit immunoblotting, and this showed that ␤ 1 ␥ 3 and KSR-1 form a complex in vivo (Fig.  2B). The efficiency of this interaction was determined in three separate experiments by densitometric analysis of anti-KSR and anti-␤ subunit immunoblots: 57% Ϯ 15% (S.E.) of ␤ 1 ␥ 3 bound to KSR-1.
The ability of KSR-1 to interact with ␤␥ subunits in untransfected cells was also investigated. Untransfected COS7 cells were cultured in the absence of serum for 24 h and then some cells were stimulated with 10% fetal calf serum for 10 min. Anti-KSR immunoprecipitates obtained from untransfected cell protein lysates were analyzed by anti-pan-␤ subunit immunoblotting and this showed that ␤␥ and KSR-1 form a complex in serum-stimulated but not in serum-starved cells (Fig. 2C).
The ability of ␤␥ subunits to form a complex with KSR-1 in vivo suggested that the liberation of free ␤␥ subunits on Gprotein activation could cause KSR-1 to translocate to the plasma membrane. We have previously demonstrated that serum stimulation of cultured NIH/3T3 cells results in a redistribution of a significant proportion of KSR-1 from the cytoplasmic fraction of cell lysates to the plasma membrane fraction (5). Because LPA is an important component of serum that binds to G i -coupled receptors, we examined whether LPA could also induce this translocation of KSR-1 (20). Cultured COS7 cells were treated with LPA, and the subcellular localization of KSR-1 was examined by differential centrifugation followed by immunoblotting. LPA treatment of cells resulted in robust redistribution of KSR-1 to the membrane fraction of cell lysates that was blocked by pretreatment with PTX, which specifically inhibits G i (Fig. 3). Because LPA stimulates G i -coupled receptors that can activate Ras (20), we wished to evaluate whether KSR-1 translocation was dependent on Ras activation. Transfection of cells with dominant-negative (N17) Ras did not inhibit LPA-induced redistribution (Fig. 3B) (21).
To evaluate the biological significance of the interaction between ␤␥ subunits and KSR-1, we evaluated MAP kinase activity in transfected cells. Cotransfection of cultured cells with ␤ and ␥ subunits has previously been shown to result in MAP kinase activation in the absence of serum stimulation (22). It has been 2 Heming Xing and Anthony J. Muslin, unpublished observations.

FIG. 4.
A, KSR-1 inhibits ␤␥-induced MAP kinase activation. Anti-ERK1 immunoprecipitates obtained from COS7 cell lysates were analyzed by in vitro kinase assay with myelin basic protein (MBP) used as a substrate. COS7 cell lysates were obtained from control untransfected cells, untransfected cells that were stimulated with 10% FCS, cells that were double-transfected with mammalian expression vectors encoding ␤ 1 and ␥ 3 , or cells that were triple transfected with mKSR-1, ␤ 1 , and ␥ 3 . Three separate triple transfections with mKSR-1, ␤ 1 , and ␥ 3 are depicted. Parallel samples were analyzed by immunoblotting with an anti-␤ 1 subunit polyclonal antibody (BN-1). B, graphical depiction of the results of the in vitro kinase experiment described in A. The autoradiograph was downloaded to a computer and analyzed by densitometry with NIH Image software. The graph is representative of the results of three separate in vitro kinase experiments. demonstrated that free ␤␥ subunits interact with and activate phosphatidylinositol 3-kinase ␥ , and this interaction is thought to eventually lead to activation of Ras and MAP kinase (23). In this study the additional transfection of cultured cells with full-length mKSR-1 markedly inhibited ␤ 1 ␥ 3 -induced MAP kinase activation without affecting ␤ 1 subunit protein levels (Fig. 4). DISCUSSION The subcellular localization of many signaling proteins is highly regulated and is often an important determinant of their activity. Localization is thought to influence activity by increasing the proximity of an enzyme to activating molecules or to substrates. For example, the protein kinase Raf-1 must translocate to the plasma membrane to be fully activated (17,18,24). The protein kinase KSR-1 also translocates from the cytosol to the plasma membrane, but the importance of this event in regulating the activity of KSR-1 has not been determined. In the experiments described here, we evaluated the mechanism of the translocation of KSR-1 from the cytosol to the plasma membrane. By use of the yeast two-hybrid assay we confirmed that KSR-1 can interact with the ␥ 2 , ␥ 3 , and ␥ 10 subunits of heterotrimeric G-proteins. These G-protein subunits are lipid modified and have been shown to be constitutively plasma membrane-bound (19). We also demonstrated that KSR-1 binds to ␤␥ subunits in cultured mammalian cells, and that a ligand that liberates ␤␥ subunits, LPA, can stimulate the translocation of KSR-1 to the plasma membrane. These findings confirm that ␤␥ subunits can mediate the translocation of KSR-1 to the plasma membrane.
One interesting aspect of the interaction between ␥ subunits and KSR-1 is that ␤␥ effectors usually bind directly to the larger ␤ subunit (25)(26)(27). The surface of the ␤␥ dimer that interacts with effectors has been examined by x-ray crystallography, demonstrating that there are several distinct areas that interact with effectors (28 -30), particularly the amino-terminal coiled-coil domain of ␤ subunit, which is immediately adjacent to the amino-terminal domain of the ␥ subunit that also forms a coiled-coil (31)(32)(33). The amino termini of both subunits form a continuous surface that could theoretically interact with effectors although it remains to be determined whether this is the site of interaction with KSR-1.
We demonstrated by two-hybrid assay that the CA3 domain of KSR-1 can bind to ␥ 2 , ␥ 3 , and ␥ 10 . This domain is highly homologous to the cysteine-rich domains of Raf-1, A-Raf, protein kinase C, and citron kinase, and is less homologous to the cysteine-rich domains of diacylglycerol kinase, the racGAP Nchimaerin, and the PTPL1-associated rhoGAP (BLAST search, National Center for Biotechnology Information). Previous investigations have demonstrated that Raf-1 can interact with ␤␥ subunits in vitro and in vivo (34). The budding yeast scaffolding protein ste5 interacts with ␤␥ subunits via its cysteine-rich ring-H2 domain (35,36). It will be important to determine whether other proteins that contain cysteine-rich domains are able to interact with ␤␥ subunits because of the possibility that cysteine-rich domains, as a class, are ␤␥ effectors.
The ability of ␤␥ subunits to bind specifically to the CA3 domain, MEK to bind to CA5 (8,9), and MAP kinase to bind to CA4 (12) supports the hypothesis that mKSR-1 is a scaffolding protein. In budding yeast, ste5 links ␤␥ subunits (ste4, ste18) to the MAP kinase cascade proteins ste11, ste7, and fus3/kss1 and promotes their activation (35,36). In marked contrast to findings with ste5, we found that overexpression of mKSR-1 inhibits ␤␥-induced MAP kinase activation. This discrepancy suggests that mKSR-1 has a unique physiologic role in the regulation of MAP kinase signaling. Our findings complement recent work by other investigators demonstrating that overexpression of mKSR-1 in cultured mammalian cells inhibits serum-and ligand-induced MAP kinase activation (8 -10).