Proc. Natl. Acad. Sci. USAVol. 91, pp. 1270-1274, February 1994Biochemistry

Enzymatic characteristics of the c-Raf-1 protein kinaseTHOMAS FORCE*t, JOSEPH V. BONVENTREf, GISELA HEIDECKER§, ULF RAPP§, JOSEPH AVRUCH¶,AND JOHN M. KYRIAKIS¶*Cardiac, *Renal, and 1Diabetes Units, Medical Services of the Massachusetts General Hospital and the Department of Medicine, Harvard Medical School,Boston, MA 02114; and §Laboratory of Viral Carcinogenesis, National Cancer Institute, Frederick Cancer Research and Development Center,Frederick, MD 21702

Communicated by Alexander Leaf, October 18, 1993

ABSTRACT The c-Raf-1 protein kinase plays a centralrole in the mitogenic response of cells to growth factors,cytokines, and many oncogenes. Despite the critical importanceof this enzyme, very little is known of its biochemical propertiesor mechanisms of regulation. In these experiments, we used theonly candidate physiologic substrate identified as yet forc-Raf-1, mitogen-activated protein kinase kinase (MAPKK), toexamine enzymatic characteristics and candidate modulatorsof c-Raf-1. c-Raf-1 was purified from Sf9 cells infected withrecombinant baculovirus encoding a histidine-tagged c-Raf-1.The Km values of c-Raf-1 for ATP and MAPKK were 11.6 pMand 0.8 pM, respectively, and the stoichiometry of phosphor-ylation of MAPKK by c-Raf-1 was 1.67 mol of phosphate permol of MAPKK. In contrast to prior reports, Mg2+ was thepreferred cation at Mg2+ and Mn2+ concentrations >5 mM.c-Raf-1 substrate specificity was extremely restricted, consis-tent with the identification of only one candidate physiologicsubstrate to date and highlighting the necessity of usingMAPKK rather than artificial substrates in c-Raf-1 activityassays. Of multiple potential substrates tested, the only onephosphorylated to >20% of the level of MAPKK phosphory-lation was myelin basic protein (22%). Heat-denaturedMAPKK was phosphorylated at only 2% the level of nativeMAPKK, indicating that the restricted substrate specificitymay be due to tertiary-structural requirements. We also ex-amined whether c-Raf-1 activity is modulated by lipid bindingto the cysteine finger region in its regulatory domain. Ofmultiple mitogen-stimulated or cell-membrane lipids tested,only phosphatidylserine and diacylglycerol in the presence ofCa2+ (2.5 mM) increased c-Raf-i kinase activity s fantly(1.5-fold). The increase is probably not of physiologic signifi-cance because it was about two orders of magnitude less thanthe stimulation of protein kinase C by these lipids. On gel-filtration chromatography, the peak of c-Raf-i kinase activityand immunoreactivt eluted at a predicted molecular mass of>150 kDa, suggesting that active c-Raf-i (but not inactivec-Raf-1) exists as a multimeric complex. This complex may notinclude p21m, however, because immunoreactive p2l1 wasnot identified in the active fractions.

c-Raf-1 is a ubiquitously expressed serine/threonine kinasethat integrates mitogenic signals from a large number ofgrowth factor receptors, cytokine receptors, several mem-brane-bound oncogenes (for review, see refs. 1 and 2), andsome mitogenic peptides with G protein-linked receptors (3).Despite the critical role c-Raf-1 plays in cellular response tomitogens, very little is known about its enzymatic charac-teristics or its regulation. Enzyme kinetic data have beenlimited largely because, until recently, physiological sub-strates of c-Raf-1 had not been identified. Assays havetypically used immunoprecipitated protein, which is unsuit-able for enzymatic analysis, along with nonphysiologic sub-

strates that were phosphorylated at very low rates (4, 5).Mechanisms of regulation have also been unclear. Becauseno kinase that activates c-Raf-1 has been identified andbecause although p21 binds to the N-terminal regulatorydomain of c-Raf-1 (6, 7), it has not been shown to activatec-Raf-1, another direct protein-protein interaction and, pos-sibly, a lipid-protein interaction with the cysteine fingerregion of the regulatory domain may modulate c-Raf-1 (1, 2).Recently, we and others (3, 8-10) have demonstrated thatoncogenic variants of c-Raf-1 and mitogen-stimulated wild-type c-Raf-1 can phosphorylate and activate mitogen-activated protein kinase kinase (MAPKK) which, in turn,activates mitogen-activated protein kinase (MAPK) (11-13).These results not only identified a physiologic substrate ofc-Raf-1, MAPKK, allowing study of the enzyme character-istics of c-Raf-1 but also identified c-Raf-1 as the mostproximal kinase in the critically important cascade of mito-gen-stimulated kinases, which also includes p85rsk and thetranscription factors p62TCF and c-Jun (14-16). Conse-quently, it is necessary to understand the enzymology andcellular mechanisms involved in regulating c-Raf-1 to under-stand transmission of the mitogenic signal in cells. We usedMAPKK as substrate to characterize the biochemical prop-erties of c-Raf-1.

MATERIALS AND METHODSMaterials. [_32P]ATP was from DuPont-New England

Nuclear. ATP (special quality) was from Boehringer Mann-heim. Prostaglandins F2a (PGF2<,) and E2 (PGE2), 12-hydroxy-[S-(E,Z,Z,Z)]-5,8,10,14-eicosatetraenoic acid, and14(15)-epoxyeicosatrienoic acid were from Cayman Chemi-cals (Ann Arbor, MI). Histones H1, H2A, H2B, H3, and H4were from Boehringer Mannheim. Protein kinase C (PKC)and 1-acetyl-S-farnesylcysteine were from BioMol (Ply-mouth Meeting, PA). Genistein was from ICN. Microtubule-associated protein 2 and 40S ribosomal subunits were pre-pared as described (17-19). The Raf-1 autophosphorylationpeptide (IVQQFGFQRRASDDGKLTD) and Kemptide (LR-RASLG) were from Peninsula Laboratories. Syntide 2(PLARTLSVAGLPGKK) was provided by Edwin Krebs(University of Washington). Epidermal growth factor recep-tor T669 peptide (GVEPLTPSGEAPNQ) was synthesized byusing described methods (20). P81 phosphocellulose paperwas from Whatman. Poly(vinylidene difluoride) membrane(Immobilon) was from Millipore. All other chemicals werefrom Sigma.

Expression and Purification of Histidine-Tagged c-Raf-i.Sf9 cells were infected with a baculovirus encoding histidine-

Abbreviations: MAPK, mitogen-activated protein kinase; MAPKK,mitogen-activated protein kinase kinase; NEM, 1-ethylmaleimide;PtdSer, phosphatidylserine; PKC, protein kinase C; PGF2. andPGE2, prostaglandin F2. and E2, respectively; OAG, 1-oleoyl-2-acetyl-sn-glycerol; BSA, bovine serum albumin.tTo whom reprint requests should be addressed at: Cardiac Unit,Bulfinch 457, Massachusetts General Hospital, Boston, MA 02114.

1270

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

May

10,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 1271

tagged c-Raf-1 (six histidine residues added to the C terminusof the enzyme) alone or simultaneously infected with threeseparate baculoviruses containing histidine-tagged c-Raf-1,v-Src, and Val-12 v-Ras (21) at a multiplicity of infection of10 using described methods (22). Seventy-two hours afterinfection, Sf9 cells were lysed, as in ref. 23, except that 0.4%Triton X-100 was included in the lysis buffer. Purification ofthe histidine-tagged protein was accomplished by loading theclarified extract onto a Ni2+-nitrilotriacetic acid-Sepharosecolumn (Invitrogen) and eluting with imidazole-containingbuffer (23). c-Raf-1 was purified to a specific activity of29,000 units/mg (where 1 unit = 1 pmol of phosphatetransferred to MAPKK per min).

Purification of MAPKK. Bovine brain MAPKK was pre-pared as described (3). Briefly, bovine brain was powdered,homogenized, and centrifuged (100,000 x g for 1.5 hr). Thesupernatant was equilibrated with DEAE-cellulose, and thebreakthrough fraction was collected. The breakthrough frac-tion was subjected to Fast-S Sepharose chromatography.Fractions with peak activity were subjected to DEAE-Cibachron blue-3GA agarose chromatography and finallyconcentrated with Mono S chromatography. This procedureroutinely gave a preparation with a specific activity of 3500units/mg (where 1 unit = 1 pmol of phosphate transferred toMAP kinase per min) with an overall recovery of 6%. On thebasis of Coomassie blue staining (Fig. 1), the preparationpurity was =30% (3). MAPKK concentration in the prepa-ration was estimated by directly comparing intensity of theCoomassie blue-stained MAPKK band with dilutions ofprotein standards of known concentrations.

c-Raf-l Kinase Assays. c-Raf-1 kinase activity was assayedin MgCl2 (10 mM)/[y-32P]ATP (100 ,uM, concentration de-termined gravimetrically; 3000-8000 cpm/pmol)/MAPKK(23 pg/ml) unless otherwise noted in text. Using amounts ofc-Raf-1 between 7.5 ng per assay and 3.75 ng per assay,MAPKK phosphorylation was linear with respect to c-Raf-1concentration and time (up to 20 min). c-Raf-1 was, therefore,assayed using 5.6 ng per assay for 20 min unless otherwisenoted in text. To measure stoichiometry of c-Raf-1 phos-phorylation ofMAPKK, 6 ,ug ofMAPKK/2 mM ATP/750 ngof c-Raf-1 was incubated for 2 hr at 30°C. Reactions werestopped with SDS sample buffer, and the proteins wereresolved by SDS/PAGE. Bands corresponding to the 48-kDaMAPKK or the other substrates tested (see Results) wereexcised, and radioactivity was measured by liquid scintilla-tion counting. All assays using MAPKK as substrate werecorrected for MAPKK autophosphorylation, which was

-97.4 kDa

-66.2

*' 4--- 45

-31

FIG. 1. Coomassie blue-stainedSDS gel of MAPKK preparation.

- 21.5 MAPKK (arrow) was purified as de-scribed. An aliquot of the preparationwas run on a SDS/1Oo acrylamide

- 14.4 gel, and the gel was stained. Molec-ular mass standards are at right.

<10% of c-Raf-1 kinase activity. In assays examining poten-tial lipid modulators of c-Raf-1, lipids were sonicated on icein kinase assay buffer (50 mM (3-glycerol phosphate, pH =7.3/1.5 mM EGTA/1 mM dithiothreitol/0.03% Brij 35) for 1min before being added to the mixture. Sometimes (seeResults), synthetic peptides were used as substrates. Afi-quots of the reaction mixture were spotted onto P81 phos-phocellulose paper and then immersed in a 5% trichloroaceticacid solution/10 mM phosphoric acid. Phosphocellulose pa-pers were washed four times for 30 min, rinsed in acetone,dried, and subjected to liquid scintillation counting.

Gel Filtration Chromatography. Purified histidine-taggedc-Raf-1 (200 1.l) was applied to a 1 x 30 cm Superose 12column (Pharmacia) equilibrated with 20 mM Tris, pH 7.4/2mM EGTA/200mM NaCl/1 mM dithiothreitol/0.05% TritonX-100, and 0.5-ml fractions were collected. The column wascalibrated before each analytic run with ferritin (440 kDa),catalase (232 kDa), aldolase (158 kDa) or IgG (150 kDa),bovine serum albumin (BSA, 67 kDa), and ovalbumin (43kDa) or ribonuclease A (14 kDa). Xylene cyanol was used toindicate included volume.

Immunoblotting. Superose 12 column fractions were run onSDS/10% PAGE and then transferred to Immobilon. Afterblocking, the membranes were exposed to affinity-purifiedanti-Raf antiserum (anti-SP63), anti-Ras antiserum (Onco-gene Science), or a monoclonal anti-GTPase-activating pro-tein antibody (Upstate Biotechnology, Lake Placid, NY).Antibody binding was detected with the ECL immunoblot-ting system (Amersham).

RESULTSMg2+/Mn2+ Dependence. Dependence of c-Raf-1 kinase

activity on Mg2+ and Mn2+ concentrations was determined(Fig. 2). At 5 mM or less, the effect of the two cations onc-Raf-1 activity was equal, but at higher concentrations,c-Raf-1 kinase activity was =50% more with Mg2+ than withMn2+. c-Raf-1 kinase activity was maximal at a Mg2+ con-centration of 10 mM, and all subsequent kinase assays weredone at 10 mM.

Modifiers of c-Raf-1 Kinase Activity. A number of potentialmodifiers of c-Raf-1 kinase activity were examined (Table 1).c-Raf-1 activity was completely inhibited by Zn2+ and Co2+(10 mM). Subsequent incubation with EGTA (10 mM) par-tially reversed this inhibition for Zn2+ only. 1-Ethymaleimide(NEM) inactivated the enzyme, which was prevented byexcess dithiothreitol, suggesting a requirement ofthe enzymefor free sulfhydryl groups. Poly(L-lysine), but not polygluta-mate, markedly inhibited c-Raf-1 kinase activity. ATP wasthe preferred nucleotide substrate; addition of excess unla-beled ATP (1 mM), but not ofGTP or CTP, markedly reduced

01)a)

co-C,

O L

2_>C)6a-

0-

140-

12010080-60-40-20-0

0 1 2 5 10 20 50Concentration, mM

FIG. 2. Dependence of c-Raf-1 kinase activity on Mn2+ vs. Mg2+concentration. Kinase assays with histidine-tagged c-Raf-1 weredone in triplicate in assay buffer with Mn2+ (m) or Mg2+ (o) as noted.ATP was present at 100 ,uM. Data are means ± SEMs.

Biochemistry: Force et al.

Dow

nloa

ded

by g

uest

on

May

10,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

Table 1. Candidate modifiers of c-Raf-1 activityModifier(s) % Control

Heparin (50 units/ml) 113Polylysine (50 pg/ml) 30Polyglutamate (50 pLg/ml) 121Ca2+/calmodulin (250 ,uM/10 pg/ml) 85Staurosporine (1 1uM) 69H-7 (100. ,LM) 74Calphostin (1 ILM) 105Genistein (100 iAM) 89NaF (50 mM) 39ATP (1 mM) 19GTP (1 mM) 112CTP (1 mM) 129CoC12 (10 mM) 0CoC12/EGTA (10 mM/10 mM) 0ZnSO4 (10 mM) 0ZnSO4/EGTA (10 mM/10 mM) 39NEM (2 mM) 14NEM/dithiothreitol (2 mM/5 mM) 92

For CoCl2/EGTA and ZnSO4/EGTA, c-Raf-1 was incubated for20 min at 30°C with the metal followed by EGTA for 15 min beforekinase assay. For NEM and NEM/dithiothreitol, c-Raf-1 was incu-bated with NEM alone or NEM/dithiothreitol for 20 min at 30°C.Dithiothreitol was then added to the NEM-alone tubes for a finalconcentration of 5 mM before kinase assay. c-Raf-1 was incubatedwith all other modifiers for 10 min at 30°C before kinase assay.

incorporation of32PO4 into MAPKK. The nonspecific proteinkinase and phosphatase inhibitor NaF significantly reducedc-Raf-1 kinase activity. Two other relatively nonspecificprotein kinase inhibitors, H7 and staurosporine, inhibitedc-Raf-1 modestly. Calphostin, a PKC inhibitor that interactswith the phorbol ester-binding regulatory domain, andgenistein, a tyrosine kinase inhibitor, were ineffective.Enzyme Kinetics. The Km of c-Raf-1 for ATP was 11.6 ,uM

(Fig. 3, Left). The Km for MAPKK was 0.8 ,M (Fig. 3,Right).

Substrate Specificity. The stoichiometry of phosphoryla-tion of MAPKK by c-Raf-1 was 1.67 mol of phosphate permol ofMAPKK. This observation, combined with the Km forMAPKK, strongly supports the contention that MAPKK is aphysiologic substrate of c-Raf-1 (3, 8-10). A number ofnonphysiologic substrates have been used to examine c-Raf-1kinase activity in immunoprecipitates from mitogen-stimulated cells (4, 5, 24-30). We compared the ability ofc-Raf-1 to phosphorylate MAPKK with its ability to phos-phorylate a variety of nonphysiologic substrates and sub-strates of other protein kinases (Table 2); MAPKK was farsuperior to any other substrate tested. Despite using sub-strate concentrations 6-fold greater than that of MAPKK,phosphorylation was generally only 10% that of MAPKK.

500- 4-

400

~300

-200 2

100

0 1

0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 1 2 3

1/[ATP] (AM-') 1/[MAPKK] (tM-1)

FIG. 3. c-Raf-1 enzyme kinetics. Lineweaver-Burk plots forc-Raf-1 substrates ATP (Left) and MAPKK (Right). All assays weredone with MgC2 (10 mM). (Left) MAPKK was present at 40 pg/mi.Km of ATP was 11.6 ,uM. (Right) ATP was present at 100 pM. Kmof MAPKK was 0.8 AM. Assays were done in duplicate for eachsubstrate concentration, and all data points are shown. Slopes andintercepts were determined by least-squares analysis.

Specifically, the basic substrates-histones H1, H2A, H2B,H3, and H4-were phosphorylated to levels 3-12% that ofMAPKK. Other than MAPKK, myelin basic protein (250,ug/ml), a substrate of the MAP kinases, was the best of thesubstrates tested, but it was phosphorylated at only 22% theMAPKK rate (40 pg/ml). Microtubule-associated protein 2was a poor substrate (9%o of MAPKK phosphorylation).Acidic substrates phosvitin and dephosphocasein were phos-phorylated at 16 and 20o the MAPKK rate. Syntheticpeptides, including Syntide 2 (4, 29) and the Raf-1 autophos-phorylation peptide, a peptide with the sequence ofa putativec-Raf-1 pseudosubstrate or autophosphorylation site (5, 26),were very poor substrates. Phosphorylation of heat-denatured MAPKK (exposed to 550C for 5 min followed bycentrifugation to remove any aggregates) was less than thatof most other substrates tested. These data indicate thatc-Raf-1 has an extremely restricted substrate specificity andthat much of that specificity may be from tertiary-structuralrequirements.

Candidate Lipid Modulators. Analogous to PKC, the reg-ulatory domain of c-Raf-1 contains a cysteine finger regionthat is predicted to bind to lipids (31), suggesting that c-Raf-1activity may, in part, be modulated by lipids (1, 40). Becausedirect activation of c-Raf-1 by p2Vras has not yet beendemonstrated, we examined whether lipids, some producedin response to mitogen stimulation of cells [arachidonic acid,diacylglycerol and its analog, 1-oleoyl-2-acetyl-sn-glycerol(OAG) phosphatidic acid, PGF2a, PGE2, 12-S-HETE, 14(15)-EpETrE; see Table 3], and others that are normal cell-membrane constituents [phosphatidylinositol 4,5-bisphos-phate and phosphatidylserine (PtdSer)] modified c-Raf-1 ki-nase activity. We also examined whether an analog offarnesyl, 1-acetyl-S-farnesylcysteine, modified c-Raf-1 ki-nase activity because prenylation ofp21ras is critical for manyof its biological functions (for review, see ref. 32), suggestingthat the farnesyl group could play a role in c-Raf-1 activation.We examined the effects of these lipids, using both "active"c-Raf-1, derived from Sf9 cells simultaneously infected withrecombinant baculoviruses encoding v-Ras, v-Src, and thehistidine-tagged c-Raf-1, and "inactive" c-Raf-1, derivedfrom Sf9 cells infected with the histidine-tagged c-Raf-1-encoding baculovirus only (Table 3). None of the lipidssignificantly increased kinase activity of the "active"c-Raf-1. The combinations of Ca2+ (2.5 mM), PtdSer (100

Table 2. Raf-1 substrate specificityP04 % of

incorpor- MAPKKation, phosphor-

Substrate fmol/min ylationMAPKK (40 pg/ml) 477 ± 40 100MAPKK (heat denatured) (40 pg/ml) 11 ± 1 2Raf-1 autophosphorylation (10 pg/ml) 6 ± 1 1Raf-1 autophosphorylated peptide (1 mM) 7 ± 1 2Histone H1 (250 pg/ml) 32 ± 2 7H2A (250 pg/ml) 15 ± 1 3H2B (250 pg/ml) 59 ± 2 12H3 (250 pg/ml) 37 ± 1 8H4 (250 pg/ml) 20 ± 1 4MAP-2 (250 pg/ml) 45 ± 7 9Myelin basic protein (250 pg/ml) 106 ± 8 22Phosvitin (250 pg/ml) 78 ± 4 16Dephosphocasein (250 pg/ml) % ± 12 2040S ribosomal subunits (0.4 OD unit) 37 ± 2 8Syntide (1 mM) 5 + 1 1Kemptide (0.4 mM) 7 ± 1 2EGF receptor T669 peptide (1 mM) 4 ± 1 1

All kinase assays were done for 20 min at 30TC as described. EGF,epidermal growth factor; MAP-2, microtubule-associated protein 2.

1272 Biochemistry: Force et al.

3

Dow

nloa

ded

by g

uest

on

May

10,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994) 1273

Table 3. Candidate lipid modifiers"Active" "Inactive"

Lipid Raf (% C) Raf (% C)Arachidonic acid (150 A&M) 98 110DAG (100 pg/ml) 91 102Phosphatidic acid (100 pg/ml) 85 113PIP2 (100 Aig/ml) 99 99PGF2, (150 ,uM) 112 129PGE2 (150 AM) 111 11812(S)-HETE (150 ,uM) 101 10514,(15)-EpETrE (150 pM) 110 105Ca2+ (5 mM) 80 111PtdSer (100 pg/ml) 84 113Ca2+/PtdSer/DAG (2.5mM/100 e/ml/25 pg/ml) 72 151

Ca2+/PtdSer/OAG (2.5mM/100 Pg/ml/10 eg/ml) 74 158

1-Acetyl-S-famesylcysteine(100 Ag/ml) 74 157c-Raf-1 was preincubated with the sonicated lipids for 10 min at

300C before performing the kinase assays. PIP2, phosphatidylinositolbisphosphate; 12(S)-HETE, 12-hydroxy[S,(E,Z,Z,Z)]-5,8,10,14-eicosatetroenoic acid; 14,(15)-EpETrE, 14,(15)-epoxyeicosatrienoicacid;DAG, diacylglycerol. DAGandphosphatidic acidwere ,-arachi-donoyl y-stearoyl.

AM), and either diacylglycerol (25 pg/ml) orOAG (10 ,ug/ml)increased kinase activity ofthe "inactive" c-Raf-1 by 1.5-fold(P < 0.05; Table 3). For comparison, in parallel assays,PtdSer with diacylglycerol or OAG increased the activity ofPKC =110-fold over control.

Gel-Filtration Chromatography. Analytic gel filtration withSuperose 12 showed that the active enzyme eluted as a broadhigh molecular mass peak between ferritin (440 kDa) and IgG(150 kDa) (Fig. 4A). Activity at the predicted molecularweight (coeluting with the 67-kDa BSA standard) was abouthalf that of the peak fraction. The column fractions were alsoimmunoblotted with affinity-purified anti-Raf-1 antibody (an-ti-SP63). Immunoreactive c-Raf-1 coeluted with MAPKK-phosphorylating activity (Fig. 4A). In contrast, the inactiveenzyme eluted from the Superose 12 column at the predictedmolecular mass, coeluting with BSA (Fig. 4B). These datasuggest that active c-Raf-1, but not inactive c-Raf-1, exists asa multimeric complex. Although recent evidence confirmsthat p21 binds to the regulatory domain of c-Raf-1 and,therefore, may modulate c-Raf-1 activity via a direct protein-protein interaction (6, 7), p21ras (and p21ras-GTPase-activating protein) was not identified in the active fractionsby immunoblotting (data not shown).

DISCUSSIONIn this report, we used the only known physiologic substrateof c-Raf-1, MAPKK, to characterize the enzyme kinetics andsubstrate specificity of c-Raf-1 and to explore modulation ofkinase activity by several candidate modifiers. In contrast toprior reports (27), we find that Mg2+ was preferred to Mn2+at concentrations >5 mM. The Km of c-Raf-1 for ATP is 11.6,uM; this Km for ATP was determined at a MAPKK concen-tration just above the Km for MAPKK. The Km for ATPdetermined at higher concentrations of MAPKK could varysomewhat from this value.The Km for MAPKK was 0.8 jLM; this Km, together with a

stoichiometry of phosphorylation of MAPKK by c-Raf-1 of1.67 mol of phosphate per mol of MAPKK, further supportsthe contention, based on experiments showing reactivation ofMAPKK by c-Raf-1 in vitro and complex formation betweenthe catalytic domain of c-Raf-1 and MAPKK with the two-

A1200 -

1000EE 800

> 600

: 400

200 -

0 +6

c-Raf-1 >

2 3 4

8 9 10 11 12 13 14 15 16ml

7 8 9 10 11 12 ml

B

9 10 11 12 13 14 15 16

440 232 158 67

ml

14 kDa

FIG. 4. Analytical gel filtration of histidine-tagged c-Raf-1 fromtriply infected Sf9 cells (A) or from cells infected only with thebaculovirus carrying c-raf-1 (B). (A) Histidine-tagged c-Raf-1 waspurified from Sf9 cells triply infected with baculoviruses encodinghistidine-tagged c-Raf-1, v-Src, and v-Ras and gel-filtered on aSuperose 12 column. The column was developed at 0.5 ml/min, and60 0.5-ml fractions were collected. Kinase assays were done on everyother fraction. Kinase activity of the fractions is presented as cpm of32PO4 incorporated into MAPKK. Elution position of standardproteins is indicated: 1, ferritin (440 kDa); 2, IgG (150 kDa); 3, BSA(67 kDa); 4, ovalbumin (43 kDa). Below graph is an immunoblot ofthe fractions with affinity-purified SP63 used to identify c-Raf-1. (B)Immunoblot of Superose 12 column fractions for inactive c-Raf-1.c-Raf-1 was purified from Sf9 cells infected with the baculovirusencoding only histidine-tagged c-Raf-1 and applied to a Superose 12column. Fractions were immunoblotted with anti-SP63. The molec-ular mass (kDa) and elution position (ml) of standard proteins(ferritin, 440; catalase, 232; aldolase, 158; BSA, 67; ribonuclease A,14) are indicated below.

hybrid system (3, 7-10), that MAPKK is a physiologic sub-strate of c-Raf-1.

c-Raf-1 substrate specificity is extremely restricted. Fif-teen other potential substrates were examined. Some ofthese, notably histones, Syntide 2, and the Raf-1 autophos-phorylation peptide, had been used to examine mitogen-induced activation ofc-Raf-1 in prior studies (4, 5, 24-30). Allmodel substrates were phosphorylated at markedly slowerrates than MAPKK (<22%), despite being at concentrations6- to 7-fold higher than that of MAPKK. This restrictedsubstrate specificity demands that future studies of c-Raf-1activation not use artificial substrates unless they have beenshown equivalent in affinity to MAPKK.Heat denaturation of MAPKK dramatically reduced its

phosphorylation by c-Raf-1. This observation indicates thatthe restricted substrate specificity of c-Raf-1 may be due, inlarge part, to a requirement for specific tertiary-structuralfeatures of the substrate. Heat denaturation ofMAPKK mayeliminate tertiary-structural determinants for c-Raf-1 phos-phorylation, leaving only a primary sequence determinantand thereby reduce MAPKK phosphorylation to levels ob-served for model substrates. The MAP kinases, in addition toa proline-directed substrate specificity, have also been shownto act optimally when substrates possess important tertiary-structure requirements. For example, c-Jun, but not v-Jun(which lacks an N-terminal delta domain), is a MAP kinasesubstrate, even though both c-Jun and v-Jun contain theN-terminal phosphorylation sites (33).

Biochemistry: Force et al.

Dow

nloa

ded

by g

uest

on

May

10,

202

0

Proc. Natl. Acad. Sci. USA 91 (1994)

The restricted substrate specificity of c-Raf-1 is in distinctcontrast to the more "downstream" kinases in the cascade,the MAP kinases, which phosphorylate and activate a num-ber of substrates (e.g., S6 kinase, c-Jun, p62TCF, and phos-pholipase A2) (14-16, 34, 35). Although other physiologicsubstrates of c-Raf-1 may be found, the restricted substratespecificity of c-Raf-1, with only one physiologic substrateidentified to date, may create a "checkpoint" at c-Raf-1/MAPKK through which mitogenic signals, originating from anumber of widely divergent sources, must pass.Bruder et al. (31) have compared the amino acid sequence

of the cysteine finger region of the regulatory domain ofc-Raf-1 to that of several transcription factors and isoformsof PKC and have suggested that the c-Raf-1 cysteine finger ismore likely a lipid-binding than a DNA-binding region.Because no c-Raf-1 kinase kinase that activates c-Raf-1 hasbeen identified, and p2lras binds to the regulatory domain ofc-Raf-1 (6, 7) but has not been demonstrated to activatec-Raf-1, another direct protein-protein or lipid-protein inter-action with the regulatory domain may modulate c-Raf-1activity (1, 2, 40). We examined c-Raf-1 activity in thepresence of normal cell-membrane lipid components, severallipids and representatives of three classes of eicosanoids thatare produced in response to mitogens, and an analog offarnesyl. Only the combination of Ca2+, PtdSer, and diacyl-glycerol or OAG produced a significant increase (1.5-fold) inc-Raf-1 kinase activity. The increase is unlikely to be ofphysiologic significance, however, because the activationdoes not approach the magnitude of activation of c-Raf-1after mitogen stimulation in situ (4- to 20-fold) (3). Bycomparison, the same combinations of lipids increased PKCactivity 110-fold. Phosphatidylinositol trisphosphate, pro-duced by the phosphatidylinositol 3-kinase in response tomany of the growth factors that activate c-Raf-1, also fails toincrease c-Raf-1 kinase activity in vitro under conditions thatstrQngly activate PKC C(J. Exton, personal communication).Many potential lipid modulators were not tested for theirability to activate c-Raf-1, and a c-Raf-1-lipid interactioncould be very specific to one particular lipid. To date,however, lipid activation of protein kinases or phosphataseshas been an effect common to a class or group of lipids(36-38). Although these lipids when tested alone were insuf-ficient to activate c-Raf-1, this does not rule out a role forthese or other lipids, in conjunction with cofactors, such asp2lras (6, 7, 39-41), in c-Raf-1 modulation or a role inlocalizing c-Raf-1 to membranes.

Results from gel-filtration chromatography suggest thatactive c-Raf-1 from Sf9 cells triply infected with baculovi-ruses carrying c-raf-1, v-src, and v-ras, but not inactivec-Raf-1 from cells infected only with the c-raf-l baculovirus,exists, in large part, as a multimeric protein or aggregate andthat this multimeric structure survives the relatively harshelution from the Ni2+ column. The absence of lipid modula-tion and the evidence for a multimeric complex containingactive c-Raf-1 support the contention that protein-proteininteractions rather than lipid-protein interactions may beresponsible for modulation of c-Raf-1 kinase activity. Theidentity of the other proteins in this multimeric complex andtheir relevance to c-Raf-1 activation is unknown. Recent datasuggest that a protein-protein interaction between p2lras andthe cysteine finger-containing regulatory domain of c-Raf-1may directly modulate c-Raf-1 kinase (6, 7). The absence ofp2lras in the active column fractions suggests that if thep21ras-c-Raf-1 interaction is necessary for initial activation, itmay not be necessary to maintain the kinase in an activestate. Alternatively, because the stoichiometry of p21ras-Rafbinding may be relatively low (6), the amount of p2lras in theactive fractions may have been below detection.

We thank Xian-feng Zhang for helpful discussions. This work wassupported by National Institutes of Health Grants DK01986,DK39773, DK38452, and GM46577, a Grant-in-Aid from the Amer-ican Heart Association, and a Searle Research Challenge Grant.

1. Li, P., Wood, K., Mamon, H., Haser, W. & Roberts, T. (1991) Cell 64,479-482.

2. Rapp, U. R. (1991) Oncogene 6, 495-500.3. Kyriakis, J. M., Force, T. L., Rapp, U. R., Bonventre, J. V. & Avruch,

J. (1993) J. Biol. Chem. 268, 16009-16019.4. Kovacina, K. S., Yonezawa, K., Brautigan, D. L., Tonks, N. K., Rapp,

U. R. & Roth, R. A. (1990) J. Biol. Chem. 265, 12115-12118.5. Blackshear, P. J., Haupt, D. M., App, H. & Rapp, U. R. (1990) J. Biol.

Chem. 265, 12131-12134.6. Zhang, X.-f., Settleman, J., Kyriakis, J. M., Takeuchi-Suzuki, E.,

Elledge, S. J., Marshall, M., Bruder, J. T., Rapp, U. R. & Avruch, J.(1993) Nature (London) 364, 308-313.

7. Van Aelst, L., Barr, M., Marcus, S., Polverino, A. & Wigler, M. (1993)Proc. Natl. Acad. Sci. USA 90, 6213-6217.

8. Kyriakis, J. M., App, H., Zhang, X., Banerjee, P., Brautigan, D. L.,Rapp, U. R. & Avruch, J. (1992) Nature (London) 358, 417-421.

9. Howe, L. R., Leevers, S. J., Gomez, N., Nakielny, S., Cohen, P. &Marshall, C. J. (1992) Cell 71, 335-342.

10. Dent, P., Haser, W., Haystead, T. A. J., Vincent, L. A., Roberts, T. M.& Sturgill, T. W. (1992) Science 257, 1404-1407.

11. Ahn, N. G., Seger, R., Bratlien, R. L., Diltz, C. D., Tonks, N. K. &Krebs, E. G. (1991) J. Biol. Chem. 266, 4220-4227.

12. Gomez, N. & Cohen, P. (1991) Nature (London) 353, 170-173.13. Crews, C. M., Alessandrini, A. & Erikson, R. L. (1992) Science 258,

478-480.14. Sturgill, T. W., Ray, L. B., Erikson, E. & Maller, J. L. (1988) Nature

(London) 334, 715-718.15. Pulverer, B., Kyriakis, J. M., Avruch, J., Nikolakaki, E. & Woodgett,

J. R. (1991) Nature (London) 353, 670-674.16. Gille, H., Sharrocks, A. D. & Shaw, P. E. (1992) Nature (London) 358,

414-416.17. Berkowitz, S. A., Katagiri, J., Binder, H. K. & Williams, R. C., Jr.

(1977) Biochemistry 16, 5610-5617.18. Kim, H., Binder, L. I. & Rosenbaum, J. L. (1979) J. Cell Biol. 80,

266-276.19. Price, D. J., Nemenoff, R. A. & Avruch, J. (1989) J. Biol. Chem. 264,

13825-13833.20. Price, D. J., Mukhopadhyay, N. K. & Avruch, J. (1991) J. Biol. Chem.

266, 16281-16284.21. Williams, N. G., Roberts, T. M. & Li, P. (1992) Proc. Natl. Acad. Sci.

USA 89, 2922-2926.22. Piwnica-Worms, H. (1992) in Current Protocols in Molecular Biology,

eds. Ausubel, F., Brent, R., Kingston, R., Moore, D., Seidman, J.,Smith, J. & Struhl, K. (Wiley, New York), pp. 16.8.1-16.11.6.

23. Janknecht, R., De Martynoff, G., Lou, J., Hipskind, R. A., Nordhem, A.& Stunnenberg, H. G. (1991) Proc. Natl. Acad. Sci. USA 88, 8972-8976.

24. Carroll, M. P., Clark-Lewis, I., Rapp, U. R. & May, W. S. (1990)J. Biol.Chem. 265, 19812-19817.

25. Carroll, M. P., Spivak, J. L., McMahon, M., Weich, N., Rapp, U. R. &May, W. S. (1991) J. Biol. Chem. 266, 14964-14969.

26. Baccarini, M., Sabatini, D. M., App, H., Rapp, U. R. & Stanley, E. R.(1990) EMBO J. 9, 3649-3657.

27. Moelling, K., Heimann, B., Beimling, P., Rapp, U. R. & Sander, T.(1984) Nature (London) 312, 558-561.

28. Morrison, D. K., Kaplan, D. R., Rapp, U. & Roberts, T. M. (1988) Proc.Natl. Acad. Sci. USA 85, 8855-8859.

29. Morrison, D. K., Kaplan, D. R., Escobedo, J. A., Rapp, U. R., Roberts,T. M. & Williams, L. T. (1989) Cell 58, 649-657.

30. Siegel, J. N., Klausner, R. D., Rapp, U. R. & Samelson, L. E. (1990) J.Biol. Chem. 265, 18472-18480.

31. Bruder, J. T., Heidecker, G. & Rapp, U. R. (1992) Genes Dev. 6,545-556.

32. Bourne, H. R., Sanders, D. A. & McCormick, F. (1990) Nature (London)348, 125-132.

33. Adler, V., Franklin, C. C. & Kraft, A. S. (1992) Proc. Natl. Acad. Sci.USA P9, 5341-5345.

34. Nemenoff, R. A., Winitz, S., Qian, N.-X., Van Putten, V., Johnson,G. L. & Heasley, L. E. (1993) J. Biol. Chem. 268, 1960-1964.

35. Lin, L.-L., Wartmann, M., Lin, A. Y., Knopf, J. L., Seth, A. & Davis,R. J. (1993) Cell 72, 269-278.

36. Nakanishi, H. & Exton, J. H. (1992) J. Biol. Chem. 267, 16347-16354.37. Nishizuka, Y. (1989) Cancer 63, 1892-1903.38. Zhao, Z., Shen, S.-H. & Fischer, E. H. (1993) Proc. Natl. Acad. Sci.

USA 90, 4251-4255.39. Troppmair, J., Bruder, J. T., App, H., Cai, H., Liptak, L., Szeberenyi,

J., Cooper, G. M. & Rapp, U. R. (1992) Oncogene 7, 1867-1873.40. Cai, H., Erhardt, P., Troppmair, J., Diaz-Meco, M. T., Sithanandam, G.,

Rapp, U. R., Moscat, J. & Cooper, G. M. (1993) Mol. Cell. Biol. 13,7645-7651.

41. Wood, K. W., Sarnecki, C., Roberts, T. M. & Blenis, J. (1992) Cell 68,1041-1050.

1274 Biochemistry: Force et al.

Dow

nloa

ded

by g

uest

on

May

10,

202

0