The Native Structure of the Activated Raf Protein Kinase Is a Membrane-bound Multi-subunit Complex*

Raf is a mitogen-stimulated protein kinase that functions as a component of the signaling cascade that leads to the stimulation of mitogen-activated protein kinase. Here we show that the native structure of Raf is a large multi-subunit protein complex with an apparent mass of 300400 kDa that interacts with Ras and the mitogen- activated protein kinase kinase Mek. Analysis of the structure of the Raf complex demonstrates that it contains a single Raf protein kinase together with the mo- lecular chaperones hsp90 and p50. The Raf-hsp90-pSO complex was observed in starved cells and in cells acti- vated with serum or phorbol ester. Thus, changes in complex formation with hsp90 and p50 are not required for activation of the Raf protein kinase. However, Raf activation caused by Ras was associated with the trans- location of the cytoplasmic Raf-hsp9O-pSO complex to the cell membrane. Significantly, it is only the mem- brane-bound complex that exhibits increased protein kinase activity. Thus, the Ras-activated Raf protein kinase functions as a membrane-bound multi-subunit complex. Studies of the development of Caenorhabditis elegans and Drosophila melanogaster have provided genetic evidence for a role of Raf in the process of signal transduction by the tyrosine kinase receptors torso (11, sevenless (21, and let-23 (3). Raf is therefore an important mediator of signal transduction by cell surface receptors. Recently, the molecular basis of Raf action has been established. Thus, it is

Studies of the development of Caenorhabditis elegans and Drosophila melanogaster have provided genetic evidence for a role of Raf in the process of signal transduction by the tyrosine kinase receptors torso (11, sevenless (21, and let-23 (3). Raf is therefore an important mediator of signal transduction by cell surface receptors. Recently, the molecular basis of Raf action has been established. Thus, it is now known that Raf is a mitogen-stimulated protein serinehhreonine kinase that functions as a component of the MAP' kinase signal transduction pathway (4).
Growth factor binding to tyrosine kinase receptors causes an increase in the GTP-bound form of Ras via a signaling pathway that includes the SH2/SH3 protein GRB2 and the exchange protein SOS (5)(6)(7)(8). The activated Ras protein can bind to Raf (9-141, a n d it is thought that this interaction is involved in the mechanism of activation of the Raf protein kinase. Indeed, this signaling function of Ras has been demonstrated using in vitro assays (15)(16)(17)(18). The protein kinase C pathway can also lead to Raf stimulation (19, 20). Following activation, the Raf protein kinase functions as a MAP kinase kinase kinase ( M A P K K K ) by * These studies were supported by Grants GM37845 and CA58396 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This with 18 U.S.C. Section 1734 solely to indicate this fact. article must therefore be hereby marked "aduertisement" in accordance ll Investigator of the Howard Hughes Medical Institute. To whom correspondence should be addressed.
phosphorylating and activating the MAP kinase kinase Mek (21)(22)(23) which in turn phosphorylates and activates MAP kinase (24). The stimulated MAP kinase then translocates into the nucleus where many of the physiological targets of the MAP kinase signal transduction pathway are located (25). Together, these data demonstrate that Raf provides one mechanism of signal transduction by tyrosine kinase receptors from the cell surface to the nucleus (4).
Although substantial progress toward understanding the role of Raf in the MAP kinase signal transduction pathway has been achieved, very little is known about the biochemical properties of the Raf protein kinase. The purpose of the study described here was to examine the structure of the Raf protein kinase isolated from tissue culture cells. We report that Raf is a mitogen-stimulated protein kinase that functions within a membrane-bound multi-subunit complex.
EXPERIMENTAL PROCEDURES Materi~ls-[~~PIPhosphate and [3SSlmethionine were purchased from Dupont-New England Nuclear. [-y-32PlATP was prepared using a Gamma-Prep A kit (Promega Biotech) as described by the manufacturer. Polyvinylidene difluoride membranes (Immobilon-P) were obtained from Millipore Corp. PMA and Protein A-Sepharose were from Sigma. Protein G-Sepharose and the Superose 6 gel filtration column HR10/30 were obtained from Pharmacia LKB Biotechnology Inc. Recombinant GST-MAPK and GST-Mek were prepared by bacterial expression and glutathione agarose (Sigma) affinity chromatography using previously described methods (26). The MAP kinase substrate peptide KRELVEPLT669PSGEAPNQALLR was obtained from the Peptide Synthesis Core Facility (University of Massachusetts Medical School).
Antibodies-The rabbit a-Raf antiserum 1558 was prepared using the synthetic peptide peptide CTLTTSPRLPVF (Multiple Peptide Systems Inc., San Diego, CA). The peptide was coupled to bovine serum albumin carrier by incubation with glutaraldehyde, and the crosslinked protein complex was used as an immunogen in New Zealand White rabbits. The rabbit a-Mek antiserum 2880 was raised against the synthetic peptide CPKKKP"PIQLNPNPEG-NHH, (Multiple Peptide Systems Inc., San Diego, CA). The peptide was coupled to keyhole limpet hemocyanin by incubation with maleimidobenzoyl-N-hydroxysuccinimide ester, and the cross-linked protein complex was used as an immunogen in New Zealand White rabbits. Donkey anti-rabbit IgG antibody (Amersham International PLC), sheep a-mouse IgG antibody (Amersham International PLC), sheep a-rat IgG antibody (Amersham International PLC), and goat a-mouse IgM(p) antibody (Life Technologies, Inc.) coupled to horseradish peroxidase were purchased from the suppliers indicated. The monoclonal a-Flag antibody M2 was from IBI (Kodak). The monoclonal a-Ras antibody Y13-259 was obtained from the American Type Culture Collection (Bethesda, MD). The monoclonal a-hsp90 antibody SPA-830 was purchased from StressGen Biotechnologies Corp. (Victoria, British Columbia). The monoclonal a-p50 antibody 3WlB5p50 (27) was provided by Dr. G. Perdew (Perdue University).
Plasmids-The plasmids pCMV-Ras61L and pCMV-Rasl7N were obtained from Dr. L. Kozma (University of Massachusetts Medical School). The plasmid pCMV-Raf was prepared using the vector pCMV5 (28) and a 2114-base pair EcoRI-XbaI fragment of a human c-raf-1 cDNA (29) that was isolated by polymerase chain reaction from a human fibroblast cDNA library. The plasmid pCMV-Flag-Raf was constructed using insertional overlapping polymerase chain reaction (30) to 6695 introduce the Flag epitope (sequence DYKDDDDK, Immunex Corp.) between codons 1 and 2 of the c-raf-I cDNA. The plasmid pCMV-Raf-Flag was constructed using insertional overlapping polymerase chain reaction to fuse the Flag epitope at the carboxyl terminus of the Raf-1 protein. The GST-Mek expression vector (26) was provided by Dr. R. L. Erikson (Harvard University). The plasmid pCMV-Mek was constructed by directional subcloning of the Mek cDNA as a BamHI1 HindIII restriction fragment into the polylinker of the expression vector pCMV5 (BglII and HindIII sites). The GST-MAP kinase expression vector was constructed from pGEX-3X (Pharmacia LKB Biotechnology Inc.) and the human ERK2 cDNA (31). The structure of the recombinant plasmids was confirmed by dideoxy sequencing. Tissue Culture-CHO cells were maintained in Ham's F-12 medium supplemented with 5% (v/v) fetal bovine serum (Life Technologies, Inc.). Cells expressing Flag-Raf fusion proteins were obtained by cotransfecting with pRSVneo using the calcium phosphate technique. Colonies resistant to G418 were isolated and screened for fusion-protein expression by Western blot analysis with the rabbit a-Raf antiserum 1558. Ransient transfection of COS cells was performed using 2 pg of plasmid DNA as described (25).
Cell FractionationSubcellular fractionation was performed essentially as described (32) with minor modifications. Cells were washed with ice-cold Krebs-Ringer HEPES-buffered saline, disrupted by sonication in buffer B and centrifuged for 30 min at 200,000 x g yielding the cytosol and the membrane pellet. The latter was resuspended in buffer B containing 1% Triton X-100, sonicated, and subsequently recentrifuged for 30 min at 200,000 x g in order to obtain a clarified membrane extract.
Metabolic Labeling and Immunoprecipitation-The cells were grown in 100-mm dishes, starved for 18 h in serum-free Ham's F-12 medium, and then labeled by incubation for 4 h in 2 ml of phosphate-free modified Eagle's medium (Flow Laboratories Inc.) supplemented with 1 mCi/ml [32P]phosphate. After washing in ice-cold Krebs-Ringer HEPESbuffered saline, the cells were harvested in 1 ml of buffer A at 4 "C. Clarified supernatant was prepared by sedimenting insoluble material by centrifugation at 15,000 x g for 15 min at 4 "C. The supernatant was precleared by incubation with 20 pl of Protein G-Sepharose for 30 min and then incubated for 1 h at 4 "C with 2 pg of a-Flag monoclonal antibody M2 immobilized on 20 9 of Protein G-Sepharose. The immunoprecipitates were washed once with buffer A, twice with buffer D, and once with buffer E. The samples were boiled for 5 min in 50 pl of two times sample buffer containing 100 m dithiothreitol and analyzed by SDS-PAGE (7% gel).
In experiments using metabolic labeling with [36Slmethionine (4 h), the cells were grown in 6-well multiwell dishes and then incubated with 1 ml of methionine-free modified Eagle's medium (Flow Laboratories Inc.) supplemented with 0.1% (v/v) fetal bovine serum and 50 pCi of [35S]methionine. Immunoprecipitations of detergent extracts (buffer A) of the cells were performed as described above.
Gel Filtration Chromatography-The cells were grown in 150-mm dishes, starved for 18 h in serum-free Ham's F-12 medium, and then treated with 100 n~ PMA for 10 min at 37 "C. The cells were washed twice in Krebs-Ringer HEPES-buffered saline and then harvested in buffer B and passed five times through a 26-gauge needle. Insoluble material was removed by centrifugation at 15,000 x g for 15 min at 4 OC. The clarified supernatant (2.5 m l ) was concentrated 10-fold using a Centricon-30 microconcentrator (Amicon, Beverly, MA). A 200-9 sample (1.5 mg of protein) was loaded on a Superose 6 HR10/30 gel filtration column pre-equilibrated with buffer B at a flow rate of 0.25 ml/min, and 30 fractions (1 ml) were collected. An aliquot of each fraction (0.9 ml) was incubated with 1 pg of M2 antibody prebound to 20 pl of Protein G-Sepharose for 1 h at 4 "C. The beads were washed three times in buffer A, twice in buffer C, and then divided into 2 equal aliquots for Western blot analysis and for the measurement of Raf protein kinase activity. The column was calibrated under identical conditions using protein standards (Pharmacia LKB Biotechnology Inc.).
In Vitro Protein Kinase Assays-Raf protein kinase activity was measured in an immune complex kinase assay using immunoprecipitates (M2 antibody) and recombinant Mek as an exogenous substrate. The immunoprecipitates were washed three times with buffer A and twice with buffer C. The washed Flag-Raf immunoprecipitates were incubated with 50 p~ [y-32PlATP (10 Ci/mmol) and 1 pg of recombinant Mek in a final volume of 100 pl for 30 min at 30 "C. Control experiments demonstrated that the phosphorylation reaction under these conditions was linear for >60 min. The reactions were terminated by transferring 75 pl of the supernatant containing the substrate (GST-Mek) to 75 p l of two times sample buffer. The samples were then boiled for 2 min prior to analysis by SDS-PAGE (7% gel). The gel was dried, and the phosphorylation of GST-Mek was quantitated using a PhosphorImager and Image-Quant software (Molecular Dynamics Inc., Sunnyvale, CA).
Mek activity was measured in an immune complex kinase assay using recombinant MAPK as an exogenous substrate. Mek was immunoprecipitated from detergent extracts (buffer A) of CHO cells by incubation with 5 p1 of rabbit a-Mek antiserum 2880 immobilized on 20 pl of Protein A-Sepharose for 1 h at 4 "C. The immunoprecipitate was washed three times with buffer D, twice with buffer C, and then incubated in buffer C together with 1 pg of GST-MAPK and 50 p~ [ Y -~~P I ATP(10 Ci/mmol) in a final volume of 50 pl for 30 min at 30 "C. The reaction was terminated by boiling with 50 pl of two times sample buffer prior to analysis by SDS-PAGE (7% gel). The gel was dried, and the phosphorylation of recombinant MAPK was quantitated using a PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CAI. The Mek protein kinase activity is presented as the Mekdependent phosphorylation of recombinant MAPK. MAP kinase activity was measured in detergent lysates (buffer A) prepared from CHO cells using the synthetic peptide substrate KRELVEPLTsssPSGEAPNQALLR as described previously (33,34).

RESULTS
In order to facilitate the analysis of the properties of the Raf protein kinase, we have employed an epitope tag to enable the efficient isolation of highly purified Raf from cells with a specific monoclonal antibody. An epitope-tagged Raf protein was constructed by adding the sequence Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys (Flag) at the NH2 terminus of c-Raf to create the protein Flag-Raf. This protein was stably expressed in CHO cells by cotransfection of the expression vector pCMV-Flag-Raf together with a plasmid that confers resistance to -18. In initial experiments, we examined the regulation of Raf by phorbol ester. Western blot analysis demonstrated that a marked electrophoretic mobility shift of the Rafprotein occurred 15-60 min after the treatment of the CHO cells with phorbol ester (Fig.  IA). Raf kinase activity was measured using an immune complex protein kinase assay with recombinant Mek as an exogenous substrate (Fig. 1B). This analysis demonstrated that the Raf kinase activity was markedly and transiently increased after phorbol ester treatment with the maximal increase after 10 min followed by a lower level of kinase activity that was sustained for at least 60 min (Fig. 1B). Comparison of the time course of the decreased electrophoretic mobility of Raf (Fig. 1A) and the increased Raf kinase activity (Fig. 1B) indicates that these processes are poorly correlated and demonstrate that the marked electrophoretic mobility shift is not required for Raf kinase activation.
Mek is phosphorylated and activated by the Raf protein ki-  Mek a s described under 'Experimental Procedures." The protein kina.se activity is presented as arbitrary uniLq obtained by Phosphorlmager analysis. The standard error was less than 1W of thr mran of rach data point.
with a protein kinase cascade mechanism of activation (4. 24). it was observed that phorbol ester treatment caused the rapid stimulation of Raf, Mek. and MAP kinase activity (Fig. 1,. However, the sustained activation of Mek (Fig. IC) and MAP kinase (Fig. 1D) is in contrast to the transient increase in Raf kinase activity (Fig. 1B). This apparent lack of correlation between Raf kinase activity and MAP kinase activity afbr long periods (>15 min) of phorbol ester treatment may be accounted for by the regulation of Mek by mechanisms that are independe n t of Raf (35). or by the regulation of phosphatase activity.

Identification of the Raf Protein Kinase as a High Molecular Weight Protein
Complex-To examine the structure and biochemical properties of the native Raf protein kinase. we exam-D . ined the behavior of Raf during gel filtration chromatography. CHO cells were activated by treatment with phorbol ester for 10 min, and a cytosolic extract containing Raf was prepared for injection onto a Superose-6 HR10130 column. Fig. 2A shows t h a t t h e 74-kDa Raf protein eluted in the early fractions from C.
t h e column with an apparent mass of 300-500 kDa. A similar elution profile of Raf immunoreactivity was observed for extracts prepared from serum-stawed cells (data not shown ). To investigate Raf protein kinase activity, we employed an immune complex assay with recombinant Mek as a n exogenous substrate. Fig. 2Z3 shows that the Raf protein kinase activity also eluted from the gel filtration column with a high apparent mass (300-500 kDa). Significantly, no Raf protein or Raf kinase activity was found to be eluted from the gel filtration column a t the position expected for a Raf monomer (74 kDaJ. Together, these observations suggest that the native structure of basal and activated Raf is a large protein complex. Hsp90 a n d p.50 Are Components of the Raf Protein Kinase Complex-The high apparent mass of the Raf kinase (300-500 kDa) that is indicated by gel filtration chromatography (Fig. 2) may be caused by the association of Raf with other proteins. To test this hypothesis, we examined Raf-associated proteins us- ing a coimmunoprecipitation assay. CHO cells were metabolically labeled with ["Plphosphate or [~'"Slmethionine. Detergent extracts of these cells were prepared, and the Raf proteins were isolated by immunoprecipitation using the monoclonal antibody (M2) that binds to the Flag epitope. The immunoprecipitates were then analyzed by SDS-PAGE. Fig. 3 shows that Raf was specifically immunoprecipitated together with two additional proteins (50 and 90 kDa). Treatment of the CHO cells with serum or phorbol ester caused no significant change in the amount of p50 and p90 associated with Raf detected by coimmunoprecipitation (data not shown). This observation indicates that Raf activation is not associated with a change in the level of Raf complex formation with p90 and p50. This finding is consistent with the observation that activation does not alter the elution profile of Raf during gel filtration chromatography (data not shown).
To identify the Raf-associated p90 and p50 proteins, we performed Western blot analysis of the Raf immunoprecipitates using antibodies directed against specific 90-and 50-kDa proteins. This analysis demonstrated that the Raf-associated 90-kDa protein is hsp90 and that the 50-kDa protein is the hsp90associated p50 protein (Fig. 3).
A Single Raf Protein Kinase Is h a t e d within the Natiue Complex-The dimerization of protein kinase domains has been proposed to be the mechanism of activation of receptor tyrosine kinases (36). It is therefore possible that the mitogen activation of the Raf protein kinase may be mediated by oligomerization. To test this hypothesis we investigated whether the native Raf complex contains one or more Raf protein molecules. The approach that we employed was to investigate the coimmunoprecipitation of Raf molecules. In order to distinguish between different Raf molecules, we constructed a mutated Raf protein (Raf-Flag) in which the epitope recognized by the m-Raf antiserum 1558 was changed by the addition of a synthetic epitope (Flag) to the COOH terminus of Raf. Control experiments demonstrated that Raf-Flag was not immunoprecipitated by the 1558 antiserum and that the wild-type Raf was not immunoprecipitated by the M2 fanti-Flag) monoclonal antibody. We then coexpressed Raf and Raf-Flag in COS cells and immunoprecipitated the Raf proteins from serum-starved and phorbol ester-treated cells using isoform-specific antibodies (1558 and M2). Western blot analysis of the immunoprrcipitates failed to detect the presence of heteromeric complexes containing Raf and Raf-Flag (data not shown). Together, these were incubated in serum-free medium for 18 h and thrn treated with 10'7 serum for defined times at 37 "C. Membrane and c.ytosol fractions were prepared from the cells, and the Raf proteins were isolated by immunoprecipitation with the M2 monoclonal antibody. The level of Raf protein in each fraction was rxamined by Western blot analysis using the a-Raf antibody 1558. The Raf protein is indicated with an arrow at the right side of each panel. Panel R , detergent extracts were prepared from CHO cells incubated with 104, srrum for defined times a t 37 "C, and the Raf proteins were isolated by immunoprecipitation with the M2 monoclonal antibody. The Mrk as described undrr "Experimental Procedures." The protein kinase Rnf protein kinase activity was measured with the exogenous substrate activity is prrsented as arbitrary units obtained by PhosphorImager analysis. The standard error was less than 107 of the mean ofeach data point. data indicate that the native Raf complex contains a single Raf molecule.
Mek Binds to the Raf Protein Kinase Complex-The Raf protein kinase substrate Mek has been demonstrated to interact with the carboxyl terminal domain of Raf in the yeast Saccharomyces cewvisiae using the two-hybrid method (9). We therefore investigated whether the native Raf protein kinase complex was bound to Mek. Western blot analysis of Raf immunoprecipitates failed to detect the presence of Mek (Fig.   4). This observation indicates that Mek is not a stable component of the Raf protein kinase complex. However, the possibility that Mek binds to the native Raf complex is not excluded by these data because the coimmunoprecipitation assay employed may be too stringent to detect low afinity interactions. To test this hypothesis, we used an over-expression strategy to investigate whether the interaction of Mek with the Raf complex can be detected. Fig. 4 shows that Mek can be coimmunoprecipitated with the Raf complex when these proteins are over-expressed in COS cells. Together, these data demonstrate that the native form of the Raf protein kinase interacts with Mek.
Ras-dependent Membrane 'I).anslocation of the Raf Protein Kinase Complex-The Ras protein is an important interrr,ediate in the signal transduction pathway initiated by growth factor receptors that results in M A P kinase activation. Re-cently, the NH2-terminal domain of the R3f protrin bas ht,rn demonstrated to bind directly to the effrctor domain of nctivated GTP-bound Ras (9)(10)(11)(12)(13)(14). This observation suggests that the activation of the Raf protein kinasr may rrquirr thr physlcal interaction of R a f with activatrd Ras. A prediction of this hypothesis is that the activated Raf protrin kinasr in mitogrnstimulated cells should be presrnt on crll mrmbranrs wherr the R a s protein is located. \Ve therrforr rxaminrd thr subcellular distribution of Raf in CHO crlls. Fig. 5 shows that the majority of Raf is present as a soluble c-ytnplasmic protein in starved cells and that only a small amount of R3f is b u n d to membranes. However, treatment of t h r cells with srmm causrd translocation of Raf from the cytosol to the membrane. Thr kinetics of serum-stimulated membrane translocation I Fig. 5A I and Raf kinase activation (Fig. 5R i were ohservrd to tx. similar.
This observation suggests that the association of Raf with membranes is significant for the process of R3f protrin kinase activation.
To examine the role of Ras in the mrmbrane translocation and activation of the Raf protein kinasr, wr rxaminrd the rffrct of Ras expression on the biochemical propertics of R3f. Fig. & I shows that the expression of activated Ras6lL ( 3 7 1 caused a marked increase in the membranr-bund form of the Raf protein kinase and a corresponding decrrase in t h r I r v r l of Raf in the cytosol. Furthermore, it was found that only the mrmbrane-bound form of Raf exhibitrd incrensrd protrin kinasc activity (Fig. 6B ). In contrast, expression of dominant-negative Rasl7N (38) caused no significant change in the activity or subcellular distribution of Raf. Togethrr. thew obsrrvations indicate that Raf protein kinase stimulation requires activated (GTP-bound) Ras and membrane association.
The cytoplasmic Raf protein kinasr rxists as a high molrcular mass complex with hsp90 and p50 (Fig.  3). Wr therrforr investigated whether the mrmbrane translocatinn of R3f rrpresents the binding of the K~f-hsp90-p50 complex to thr mrmbrane or whether the Raf prokin is relrased from thr hsp90-p50 complex prior to membrane association. Wrstem blot analysis demonstrated that hsp90 and p50 wrre coimmunoprrcipitated with the membrane-bound R3f protrin kinasr indicating that i t is the Raf-hspSO-pfiO complex that associatrs with the membrane during Raf protein kinasr activation (Fig. 61.

DISCrSSION
Hsp9O is an essential protein that is required for crll viability (39) and is thought to function as a molrcular chaprronr , 4 0 1 . Biochemical analysis has demonstratrd that hsp99 forms hrteromeric complexes with a 50-kDa protein (~5 0 1 and othrr proteins (27). Examples of proteins that have h r n found in complexes with hsp90 include steroid hormonr rrcrptors ( 4 1 4 3 1 . dioxin receptors (44.45). actin ,461. S r c tyrosinr kinase (47,481, eIF-2u kinase (49-511, and casein kinasr I1 (52). Thr rrsults of this study demonstrate that the R3f kinasc rrprrsrnts an additional example of a protein that is found in complrxes with hsp90 and p50. A similar conclusion has bcrn rrached independently by others (53). Both the basal and activated forms of t h r Raf protein kinase were found in complexrs with hsp90 and p50 ( Fig. 6). Thus, changes in the level of complrx formation hctween Raf and hsp9O-p50 do not account for thr mrchnnism of Raf kinase activation.
The size of the Raf complex ( 3 0 0 0 k D a~ ohsrrvrd during gel filtration chromatography could be accountrd for by t h r 74-kDa Raf protein bound to hsp90-p50. In prrvious studies it has been reported that hsp90 functions as a dimrr (54-561. Thus, the Raf complex is likely to contain two mrrlrculrs C J f hsp90, one (or more) moleculrcs) of p50. and onr Rnf protein kinase. However, the possibility that additional proteins arc' also present in the Raf complex cannot hc rxcludrd by thr data obtained in this study. For example, proteins weakly bound to Raf or proteins labeled poorly with [35S]methionine and [32Plphosphate may not have been detected by the coimmunoprecipitation assay (Fig. 2). One example is provided by Mek which was observed to associate with the Raf complex only when expressed at high levels in COS cells (Fig. 4). A similar weak interaction of Raf with Ras has been observed in coimmunoprecipitation assays when a large excess of recombinant Ras is added to cell extracts (14).
The finding that Raf is present in cells as a large protein complex provides an explanation for the results of previous studies of the biochemical properties of a MAPKKK activity detected in cell extracts. Chromatographic analysis by Matsuda et al. (57) demonstrated the presence of a single major peak of a MAPKKK activity with an apparent mass of approximately 400 kDa. Similarly, Itoh et al. (17) have described a MAPKKK activity with high apparent mass that was required for Ras-induced Mek activation. As Raf functions as a MAP-KKK that interacts directly with Ras, the MAPKKK activities described by Matsuda et al. (57) and Itoh et al. (17) are likely to be accounted for by the Raf-hsp9O-p5O complex described in this report (Fig. 4).
In previous studies a broad spectrum of functional roles for the binding of hsp90 to proteins has been described. For example, the hsp90-p50 complex with Src is a transient intermediate that occurs during biosynthesis of the v-Src (but not c-Src) tyrosine kinase (47). Hsp9O is therefore not a significant regulator of the mature Src kinase. However, an important regulatory role for hsp90 has been described for other proteins. For example, the glucocorticoid receptor is complexed with hsp90, and it has been found that the binding of the receptor to hsp9O is required for hormone-stimulated gene expression (58). In addition, it has been demonstrated that hsp90 binding causes stimulation of the protein kinase activity of casein kinase I1 (52) and eIF-2a kinase (49)(50)(51). Together, these data establish that signaling proteins can be regulated by interaction with hsp9O. The observation that Raf interacts with hsp90-p50 suggests that the Raf-hsp9O-p5O complex is likely to be significant for the physiological function of the Raf protein kinase. However, as both basal and activated Raf are bound to hsp90-p50, it is unlikely that the formation of this complex accounts for Raf kinase activation. Indeed, the level of complex formation by Raf with hsp90-p50 was not found to change during Raf kinase activation.
It is known that Raf can be activated by Ras (59) and that this process is likelv to be mediated by the direct interaction of these proteins (9)(10)(11)(12)(13)(14). However, the detailed molecular basis for the stimulation of Raf kinase activity has not been established. The presence of Raf in a large protein complex that interacts with Ras and Mek suggests that protein-protein interactions may represent the molecular mechanism of regulation of the Raf protein kinase. The importance of these protein interactions for the normal functioning of Raf is consistent with the observed dominant-negative phenotype of defective Raf proteins (60). We observed that both the cytosolic and membranebound forms of the Raf protein kinase are present in a complex with the molecular chaperones hsp9O and p50 (Fig. 6). However, it is only the Ras-induced membrane-bound Raf complex that exhibits increased protein kinase activity (Fig. 6). It is therefore possible that the Ras-mediated translocation of the Raf-hsp9O-p50 complex brings the Raf kinase in proximity to an activating molecule located within the membrane. We propose the hypothesis that Raf activation is mediated by a conformational change in the Raf-hsp9O-p50 complex after Rasinduced recruitment to the membrane.