Partial Purification of a Mitogen-activated Protein Kinase Kinase Activator from Bovine Brain IDENTIFICATION AS B-RAF OR A B-RAF-ASSOCIATED ACTIVITY*

A classical biochemical approach was taken to identify mitogen-activated protein kinase kinase (MEK) ac- tivators in bovine brain. Fractionation revealed the presence of one major MEK-stimulating activity that was distinct from c-Raf-1 and MEK kinase. Similar results were obtained using bovine adrenal chromaffin cells, and in both cases, immunoblotting and immunoprecipitation experiments demonstrated co-purification of MEK activator with B-Raf. Partially purified MEK activator stimulated phosphorylation of MEKl on resi- dues tentatively identified as serine 218 and serine 222. Little or no MEK activator was associated with c-Raf-1 in bovine brain or chromaffin cells, although this protein was expressed, suggesting that B-Raf might be the major MEK activator in cells of neural origin. The MAP’ kinase pathway is activated by diverse agonists that stimulate cell division, differentiation, and secretion (re-viewed in Ref. 1). MAP kinase activation requires phosphorylation on both tyrosine and threonine residues (2, 31, reactions that are catalyzed by a novel family of dual-specificity kinases named ME& (MAp or Erk kinases) (4-10). MEK activity is in turn regulated by reversible serinekhreonine phosphorylation (11-14), implying the existence of MEK kinases leupeptin plus 100 ng inactive histidine-tagged MEK1. Reactions were incubated at 30 "C for 20 min, at which time 10 pl of 25 mM Hepes-NaOH, pH 7.5, containing 2 pg of K52R p42""P' and [y3'PIATP (final specific activity of 1000-2000 cpdpmol) was added. Incubations were continued for 20 min, and terminated by the addition of 15 pl of 4 x SDS-PAGE sample buffer. Reaction products were resolved by SDS-PAGE and transferred to ni- trocellulose, and the radioactivity incorporated into K52R p42""pb was quantitated by Cerenkov counting. One unit is that amount of MEKl activator that raises the K52R p42""Pk-phosphorylating activity of 100 ng of histidine-tagged MEKl by I pmoYmin in the standard assay. At each purification step, the pooled activator was assayed with or without exogenous MEK1, and this demonstrated that the K52R p42""pk phos- phorylating activity was absolutely dependent upon the presence of exogenous MEKl in the assays. Control experiments demonstrated that histidine-tagged MEKl activated by brain activator was able to stimulate the kinase activity of wild-type recombinant p42-pk, i.e. the ob- served phosphorylation of K52R p42""pk was on the regulatory threonine and tyrosine residues. Protein was by Coomassie Blue binding (Bio-Rad). electrophoresis Coomassie

A classical biochemical approach was taken to identify mitogen-activated protein kinase kinase (MEK) activators in bovine brain. Fractionation revealed the presence of one major MEK-stimulating activity that was distinct from c-Raf-1 and MEK kinase. Similar results were obtained using bovine adrenal chromaffin cells, and in both cases, immunoblotting and immunoprecipitation experiments demonstrated co-purification of MEK activator with B-Raf. Partially purified MEK activator stimulated phosphorylation of MEKl on residues tentatively identified as serine 218 and serine 222. Little or no MEK activator was associated with c-Raf-1 in bovine brain or chromaffin cells, although this protein was expressed, suggesting that B-Raf might be the major MEK activator in cells of neural origin.
The MAP' kinase pathway is activated by diverse agonists that stimulate cell division, differentiation, and secretion (reviewed in Ref. 1). MAP kinase activation requires phosphorylation on both tyrosine and threonine residues (2, 31, reactions that are catalyzed by a novel family of dual-specificity kinases named ME& ( M A p or Erk kinases) (4-10). MEK activity is in turn regulated by reversible serinekhreonine phosphorylation (11)(12)(13)(14), implying the existence of MEK kinases that function as MEK activators. A number of candidate MEK-activating kinases have been reported, most notably c-Raf-1. A variety of biochemical (15)(16)(17)(18)(19)(20), genetic (9,(15)(16)(17)(21)(22)(23)(24)(25)(26), and regulatory (27,28) evidence points to the importance of c-Raf-1 as a MEK activator. For example, partially purified preparations of c-Raf-1 can activate MEK in vitro (15-17, 20, 29); a rafhomologue in C. elegans has been shown to function "upstream" of MAP kinase and "downstream" of tyrosine kinases (24), and overexpression of mutant inactive forms of c-Raf-1 have been shown to block MAP kinase activation, presumably by a dominant-negative mechanism (26). In addition, c-Mos (30) and MEK kinase (31) have been shown to be potential MEK activators. The latter is a mammalian homologue of the yeast Stell and Byr2 proteins (31), which function as activators of the yeast ME&, Ste7 and Byrl, respectively. GM47332 (to M. J. W.) and DK44199 (to S. J. P. ). The costs of publica-* This work was supported by National Institutes of Health Grants tion 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  The regulation of c-Raf-1 activity is clearly complex (for review, see . The c-Raf-1 protein becomes phosphorylated during activation by kinases whose identities are uncertain but may include protein kinase C isoforms (34)(35)(36). Phosphorylation is apparently necessary for activation of c-Raf-1 (37). c-Raf-1 becomes activated upon activation of p2lrQs, and c-Raf-1 has been shown to bind directly to p21"" (38)(39)(40)(41)(42). However, the Ras-Raf interaction appears insufficient to activate c-Raf-1 kinase activity, and the role of this interaction is likely to recruit c-Raf-1 to its membrane site of activation (43,44). Disruption of the Ras-Raf interaction correlates with inhibition of the MAP kinase cascade in fibroblasts (45)(46)(47) and smooth muscle cells (48).
The studies described above have been carried out predominantly in cultured cells (especially fibroblasts) or in Xenopus oocytes (where Mos was identified as a MEK activator (30)). Moreover, the studies on in vitro activation of MEK suffer from the use of partially purified Raf and/or MEK utilized in high concentrations (where nonspecific reactions might occur). The genetic experiments establish an enzyme sequence, but do not establish proximal partners; and the effectiveness of dominantnegative Raf mutants in blocking MEK and MAP kinase activation may result from the sequestration of p2Yas, thereby preventing access of other Ras-dependent effectors. With notable exceptions (49-511, there is little literature reporting classical purification and identification of endogenous MEK activators from cells or tissues and analysis of Raf isoforms or other MEK activators that might be tissue or agonist specific. Here we report partial purification of a MEK activator from bovine brain and the identification of this activity as B-Raf or a B-Rafassociated activity. B-Raf is a serinekhreonine kinase reportedly expressed primarily in brain and in the nervous system (52,53). B-Raf is 54% homologous to c-Raf-1, with greatest homology in the kinase domain (conserved region 3 or CR3) and CR1 (54). Like c-Raf-1, B-Raf can be rendered oncogenic by mutational events that constitutively elevate Raf kinase activity, indicating a role for these proteins in regulation of cell growth and division (54). Previous reports have indicated an increase in B-Raf phosphorylation and autophosphorylating activity in response to NGF in PC12 phaeochromocytoma and SH-SY5Y neuroblastoma cells (55, 56), although activity against a physiologically relevant substrate could not be examined. NGF treatment results in activation of MEK and MAP kinase (11,57,58) in PC12 cells, and both this stimulation, and subsequent neurite outgrowth are blocked by dominant negative mutants of Ras (59-61).

Construction and Expression ofHistidine-tagged
MEKl-Rat kidney MEKl(5) was tagged at the N terminus with a polyhistidine sequence, allowing purification by Ni'+-chelate chromatography.' Expression was achieved under the control of a cytomegalovirus promoter (62) in CCL39 fibroblasts. Stable transfectants (CCL39hMEKl) were obtained by G418 selection, and MEKl expression was verified by immunoblotting and activity measurements.
Cell Culture"CCL39hMEKl cells were maintained in Dulbecco's modified Eagle's medium supplemented with 5% (v/v) each calf and fetal calf serum. Prior to purification of inactive histidine-tagged MEK1, CCL39hMEK1 cells were incubated overnight in Dulbecco's modified Eagle's medium supplemented with 0.1% (v/v) fetal calf serum. Bovine adrenal chromaffin cells were prepared and maintained as described (63).
Substrate Purification-Recombinant wild-type and kinase-defective (K52R) p42"' "P' were purified to apparent homogeneity as described previously (3,64). Inactive histidine-tagged MEKl was purified from serum-deprived CCL39hMEK1 by sequential Ni'+-chelate and strong anion-exchange chromatography.' Histidine-tagged MEKl purified in this manner was homogeneous by Coomassie and silver staining (not shown) and had a specific activity of approximately 15 unitdmg using the standard assay (see below). This protein could be activated >500fold by active fractions using the standard assay procedure.
Assays-Activation of histidine-tagged MEKl was measured in a two-step assay using kinase-defective (K52R) ~42""~' as the final substrate. 10 pl of each fraction (diluted as necessary) or appropriate immunoprecipitates were mixed with 30 1.11 of a reaction mix to give (final concentrations) 25 m~ Hepes-NaOH, pH 7.5, 10 mM Mg(CH,COO),, 1 m~ D m , 0.1 r m ATP, 1 p~ okadaic acid, 5 pgml leupeptin plus 100 ng inactive histidine-tagged MEK1. Reactions were incubated at 30 "C for 20 min, at which time 10 pl of 25 mM Hepes-NaOH, pH 7.5, containing 2 pg of K52R p42""P' and [y3'PIATP (final specific activity of 1000-2000 cpdpmol) was added. Incubations were continued for 20 min, and terminated by the addition of 15 pl of 4 x SDS-PAGE sample buffer. Reaction products were resolved by SDS-PAGE and transferred to nitrocellulose, and the radioactivity incorporated into K52R p42""pb was quantitated by Cerenkov counting. One unit is that amount of MEKl activator that raises the K52R p42""Pk-phosphorylating activity of 100 ng of histidine-tagged MEKl by I pmoYmin in the standard assay. At each purification step, the pooled activator was assayed with or without exogenous MEK1, and this demonstrated that the K52R p42""pk phosphorylating activity was absolutely dependent upon the presence of exogenous MEKl in the assays. Control experiments demonstrated that histidine-tagged MEKl activated by brain activator was able to stimulate the kinase activity of wild-type recombinant p42-pk, i.e. the observed phosphorylation of K52R p42""pk was on the regulatory threonine and tyrosine residues.
Protein concentration in column fractions was estimated by Coomassie Blue binding (Bio-Rad). The concentration of purified proteins was estimated after gel electrophoresis by Coomassie Blue staining in parallel with bovine serum albumin standards.
Partial Purification of Brain MEKl Actiuator-The following operations were performed at 4 "C. Bovine brain (-300 g) was diced and blended (in two batches) with 1 liter of buffer A (10 m~ Tris-HC1,50 mM NaF, 1.5 mM D m , 1 mM Na,VO,, 3 m~ benzamidine, 2 mM EGTA, pH 7.5 at 4 "C) supplemented with 1 mM phenylmethylsulfonyl fluoride and 5 pg/ml leupeptin in a Waring blender at high speed for 30 s. The resulting extract was homogenized (IO strokes, Dounce homogenizer, loosefitting pestle) and clarified by sequential centrifugation for 45 min at 5000 x g and 30 min at A'. The column was washed with 80 ml of buffer A' and developed with a 100-ml linear gradient from 0-0.5 M NaCl in buffer A'. Fractions 27-29 containing the highest MEKl activator activity were pooled, dialyzed against buffer A supplemented with 50% (v/v) glycerol but without phenylmethylsulfonyl fluoride, and stored a t -20 "C.
Chromatography of Chromaffin Cell Extracts-Chromaffin cells were left untreated or stimulated with nicotine (20 p d for 1 min. Cells were washed rapidly in ice-cold calcium and magnesium-free phosphate-buffered saline, and homogenized (Dounce) in ice-cold buffer A lacking DTT but supplemented with 5 pg/ml leupeptin, 0.08 trypsin-inhibitor unitdml aprotinin, and 0.1% (v/v) 2-mercaptoethanol. The resulting homogenate was clarified by centrifugation for 20 min at 20,000 x g , and 8 mg of protein was applied to a Mono Q column (Pharmacia, HR 5/5). After washing with 10 ml of buffer A, the column was developed with a 30-ml linear gradient from 0-0.5 M NaCl in buffer A. I-ml fractions were collected.
Zmmunoblotting and Immunoprecipitation of Raf Proteins-Aliquots of column fractions or pooled fractions were resolved by SDS-PAGE and transferred to nitrocellulose. Amnity-purified antisera specific for c-Raf-1 (12 C-terminal residues) and B-Raf (19 C-terminal residues) raised against the peptides indicated in parentheses were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Affinity-purified anti-MEK kinase antiserum (raised against N-and C-terminal fragments of MEK kinase) was from Transduction Laboratories (Lexington, KY). Primary antibodies were detected using ECL according to manufacturers instructions (Amersham Corp. ).
To immunoprecipitate Raf family members, pooled MEKl activator from APlQ chromatography (see above) was diluted 20-fold with buffer D (25 mM Hepes-NaOH, pH 7.5,50 mM NaF, 1.5 mM D m , 1 mM Na,VO,, 3 m~ benzamidine) and mixed with protein A-agarose precoupled to 2 pg of anti-A-Raf antibody (raised to C-terminal decapeptide), anti-B-Raf antibody, anti-c-Raf-1 antibody or a matched nonimmune antibody. All antisera were from Santa Cruz Biotechnology. After 2-3 h at 4 "C, the immunoprecipitates were washed 3 times with buffer D, twice with 0.5 M LiCI, 0.1 M Tris-HC1, pH 7.5, and twice with 25 m~ Hepes-NaOH, pH 7.5, 10 mM Mg(CH,COO),; they were halved and assayed with or without histidine-tagged MEKl as described above. Diluted starting material and supernatants following immunoprecipitation were assayed in parallel. Fractions containing MEKl activator following Mono Q chromatography of chromaffin cell extracts (see above) were pooled, and 0.5-ml portions were similarly immunoprecipitated and assayed.
Phosphorylation of MEKl-AP1Q fractions were analyzed for their ability to stimulate phosphorylation of histidine-tagged MEKI. 5 4 column fraction or buffer blank was incubated in a final volume of 40 pl containing 25 mM Hepes-NaOH, pH 7.5, 10 mM Mg(CH,COO),, 1 mM D m , 1 0 pg/ml leupeptin, 1 p~ okadaic acid, 50 p~ [-p3'PlATP (-6000 cpdpmol), with or without -100 ng of purified histidine-tagged MEKI, for 20 min at 30 "C. Reactions were terminated by the addition of 15 pl of 4 x SDS-PAGE sample buffer, and resolved by SDS-PAGE on a 10% gel. MEKl substrate was located by Coomassie Blue staining and autoradiography. For phosphopeptide mapping studies, -700 ng of histidine-tagged MEKl was phosphorylated as above with the following modifications: reaction volume was 200 pl, ATP was used a t -10,000 cpmlpmol, 25 pl of pooled activator from APlQ chromatography or a B-Raf immunoprecipitate replaced the column fractions, and incubation was for 45 min at 30 "C. Reaction products were resolved by SDS-PAGE, transferred to nitrocellulose, and visualized by autoradiography.
Phosphorylation Site Analysis--Tryptic phosphopeptide mapping of nitrocellulose-bound material was essentially as described (651, except peptides were not oxidized with performic acid. Tosylphenylalanyl chloromethyl ketone-treated trypsin was from Worthington (Freehold, NJ). Peptides were separated by electrophoresis in pH 1.9 buffer and by ascending chromatography in phosphochromo buffer (65). Peptides were recovered from the cellulose matrix by extraction with pH 1.9 buffer, coupled (66) to Sequelon aryl amine membrane (Millipore Corp., Bedford, MA), and subjected to repetitive Edman degradation in an Applied Biosystems 470A sequenator as described (67)(68)(69). Edman degradation was performed by Dr John Shannon in the Biomolecular Research Facility at the University of Virginia. Phosphoamino acid analysis was as described (65), except electrophoresis was at pH 2.5 (70). Plates were subject to autoradiography after staining with ninhydrin.

RESULTS
To facilitate detection of MEK activator, we chose to examine a complex tissue where MEK activity (27), and presumably MEK activator activity, is evident in a basal state. A hypotonic extract of bovine brain was assayed for its ability to activate Rrcm histidine-tagged MEKl purified to apparent homogeneity from serum-deprived CCL39hMEK1 fibroblasts. MEKl activation was measured by phosphorylation of homogeneous kinase-defective (K52R) p42"'"pk; assays employing homogeneous recombinant wild-type p42""'pk as substrate confirmed that this phosphorylation resulted in stimulation of p42"'"pk MBP k' mase activity (data not shown).
Bovine Brain Extract Supports Activation of Both MEKl and MEK2-A MEK1-activating activity could be enriched by successive ion-exchange and dye affinity chromatographies (Fig.  11, with a striking increase in apparent yield over the first two steps ( Table I), suggesting the presence of inhibitors in the initial extract. Enhanced phosphorylation of K52R by column fractions was absolutely dependent on the presence of exogenous MEKl in the assays ( Fig. Id and data not shown), and this activation could be abolished by prior incubation of the column fraction at 95 "C for 10 min (data not shown). Recombinant MEK2 could be similarly activated by the same column fractions (data not shown), indicating that the MEK activator shows no marked preference for MEKl or MEK2 in vitro. The minor MEKl activating activity in fractions 84-100 after DEAE-Sephacel chromatography was not analyzed further.
Superose 12 gel filtration of MEK stimulating activity from either the S-Sepharose FF peak or the APlQ peak indicates that the activator has an apparent molecular mass of >400 kDa (data not shown).
B-Raf, but Not c-Raf-1 or MEKKinase, Correlates with MEKstimulating ActivitySince a number of kinases capable of activating MEKl have been described in the literature (15-17, 30, 31, 49, 51), we sought to correlate our activity with candidate enzymes by immunoblotting for these known activators. Aliquots of starting material and column fractions were resolved by SDS-PAGE and immunoblotted using antisera specific for c-Raf-1 and MEK kinase. c-Raf-1 could be detected in the hypotonic extract as a partially resolved doublet of approximately 72 and 74 kDa (Fig. 2a, lane 1 ). Although the majority of c-Raf-1 did not bind to DEAE-Sephacel, a portion was resolved in fractions 64-104 (Fig. 2a)   activating activity (fractions 60-80, see Fig. la). Similarly, MEK kinase (flow-through and fractions 36-64) eluted prior to the majority of MEK activator (Fig. 2b). Thus, weak anionexchange chromatography partially resolves the brain activator from two previously described MEKl activators, c-Raf-1 and MEK kinase. Given the tissue-specific distribution of the Raf-family kinases (52,53), we hypothesized that B-Raf might account for the observed MEKl activator activity. Indeed, immunoblotting with anti-B-Raf antiserum revealed a 90-kDa product in the hypotonic extract (Fig. 2c, lane 1 ) and enrichment of this species in fractions 64-72 from the DEAE-Sephacel column, correlating directly with MEK1-stimulating activity. Furthermore, subsequent chromatography on S-Sepharose FF resulted in co-elution of MEKl activating activity with a form of B-Raf (Fig. 2d). A small amount of c-Raf-1 was also observed in these fractions (Fig. 2e), although a comparison of lanes I, 2 , 8, and 9 between panels d and e clearly indicates enrichment of B-Raf but not c-Raf-1 with MEK1-activating activity. Note that little or no MEK1-activating activity co-eluted with B-Raf in fractions 28-31 of the S-Sepharose FF column (Fig. 2d).
MEKl activator was fractionated further by dye affinity (data not shown) and strong anion-exchange chromatography, resulting in partial resolution of the MEKl activator into two peaks (Fig. IC). Both peaks contain material reactive with B- Clarified extract, pooled activator from DEAE-Sephacel chromatography, S-Sepharose FF flow-through and fractions (Fig. l b ) were immunoblotted with antisera specific for c-Raf-1 ( d ) and B-Raf (e). f, aliquots of clarified extract, pooled activator from each chromatographic step, and APlQ fractions were immunoblotted with antiserum specific for B-Raf. Horizontal bars denote MEKl activator activity, and molecular mass standards (in kDa) are shown to the left.
Raf antiserum (Fig. 2 f ) with the later eluting minor B-Raf peak containing the greatest MEK1-activating activity. c-Raf-1 was not detectable by immunoblotting of these fractions (data not shown).
These data indicate that B-Raf can be resolved into a t least three chromatographically distinct forms, and that MEK1-activating activity correlates with two of these forms.
Immunoprecipitation of MEKl Activator and B-Raf from Pooled Column Fractions-Since the activator was not homogeneous after the chromatographic steps, it was not possible to correlate activity with a stainable protein in the APlQ peaks. Hence, we devised experiments to test the possibility that B-Raf, or a B-Raf-associated activity accounts for the MEKl activator in these fractions. Pooled material from APlQ fractions 27-29 (see "Materials and Methods") was diluted and immunoprecipitated with either control nonimmune antiserum, anti-A-Raf, anti-B-Raf, or anti-c-Raf-1 antiserum. MEKl activator was quantitatively removed from the pooled material by B-Raf antiserum (Fig. 3, lane 8 )  activator in these bovine brain extracts is B-Raf, or B-Rafassociated.
MEKl Activator Correlates with B-Raf in Bovine Adrenal Chromaffin Cells-The results presented above indicated that B-Raf might be a physiologically relevant activator of MEKl in cells of neural origin. To test this hypothesis, we fractionated bovine adrenal chromafin cell extracts by strong anion-exchange chromatography and assayed fractions for MEKl acti- vating activity. Similar MEKl activator activity was detectable in both unstimulated and nicotine-stimulated chromafin cell extracts (Fig. 4u), and in both cases this activity eluted from the column in the same fractions as B-raf ( Fig. 4h and data not shown). Although immunoblotting revealed that a portion of the c-Raf-1 co-purified with the MEKl activator peak (Fig. 4c), immunoprecipitation analysis of the active fractions identified B-Raf as the only detectable MEKl activator (Fig. 5a); MEKl activator activity associated with anti-A-Raf or anti-c-Raf-1 immunoprecipitates did not exceed that seen in the nonimmune control immunoprecipitate (Fig. 5a). Analysis of the supernatants indicated a 72% depletion of MEKl activator following B-Raf immunoprecipitation as compared with 29 and 24% depletion following c-Raf-1 and nonimmune immunoprecipitation, respectively (data not shown). Immunoblotting of the same supernatants indicated -75% depletion of B-Raf and -100% depletion of c-Raf-1 following immunoprecipitation with cognate antisera (Fig. 56). Interestingly, the recovery of MEKl activator in the B-Raf immunoprecipitate was -2.5-fold greater than expected, reminiscent of the apparent increase in yield of MEKl activator following chromatography of the brain extract (see Table I).
MEKl activator in chromaffin cells therefore correlates tightly with the presence of B-Raf protein. These results support the data obtained using bovine brain extracts, and indicate that our failure to measure other MEKl activator activities in brain is probably not due to the pathophysiological status of the bovine brains used in these studies. However, in experiments to date, we have not demonstrated a regulation of B-Raf activity in response to stimulation under conditions where endogenous MEKl activity is regulated. Possible reasons for these results are presented under "Discussion." Brain MEKl Activator Stimulates Phosphorylation of MEKl-Since MEKl is known to be regulated by reversible serinelthreonine phosphorylation, we undertook experiments to address the role of phosphorylation in the activation of MEKl by both partially purified brain activator and B-Raf immunopurified from this pooled material. Purified histidinetagged MEKl was phosphorylated by appropriate fractions from the APlQ column and resolved by SDS-PAGE. Fig. 6 dem-

B-Raf Is a
MEKActivator in Brain 30019 onstrates that MEK1-activating activity co-elutes precisely with an activity that stimulates phosphorylation and electrophoretic retardation of MEKl (compare with Fig. IC). To further investigate this modification, purified MEKl phosphorylated using pooled activator from APlQ fractions 27-29 (see "Materials and Methods") was digested with trypsin, and peptides were recovered for phosphoamino acid analysis and twodimensional phosphopeptide mapping. MEKl phosphorylated by the pooled activator contained phosphoserine, phosphothreonine, and trace amounts of phosphotyrosine (Fig. 7a). Little radioactivity was incorporated into MEKl in the absence of pooled activator (Fig. 7a).
By inspection, phosphopeptide a has properties similar to tryptic phosphopeptides that correlate with activation of MEKl in vivo in PC12 cells (71), and in vitro by immunopurified c-Raf-1 (27). Phosphopeptide a was recovered and subjected to repetitive Edman degradation a s described under "Materials and Methods." Eluent was analyzed for radioactivity after each cycle, and the results are shown in Fig. 8. Two peaks of "P release were observed after cycles 4 and 8. Analysis of the rat kidney MEKl sequence indicates no conventional tryptic phosphopeptides with serine residues a t positions 4 and 8. However, trypsinization of c-Raf-1-phosphorylated recombinant rabbit MEKl generates a serine-phosphorylated peptide with identical properties (20). This observation will be discussed below. DISCUSSION To date, there have been few attempts to define the contribution of various MEK activators in different tissues or in response to various agonists. Most published literature examines the activity of candidate MEK activators without addressing the quantitative role of these enzymes in vivo. We have therefore taken a biochemical approach to identify MEK activators in bovine brain, and our data demonstrate that a MEK1activating activity co-purifies tightly with a form of B-Raf from  '12'13'14'15'16'17'18'19'20 Cycle MEKl activation in uitro. Phosphopeptide a (1683 cpm) was sub- this tissue. Immunoprecipitation analysis strongly suggests that this activity is intrinsic to the B-Raf polypeptide, although we cannot exclude the possibility that a co-precipitating activity is responsible in part or fully for the observed activation of MEK1. Indeed, B-Raf is resolved into at least three chromatographically distinct forms during partial purification of the brain activator, and these are differentially active against MEK1. Furthermore, gel filtration indicates an apparent molecular mass of >400 kDa for the brain MEKl activator, rather larger than that expected for monomeric B-Raf (-90 m a ) . Similarly large sizes have been reported for a MEK activator from Xenopus (41,51) and for c-Raf-1 (72). Conclusive identification of the MEKl activator can be determined only by purification and sequence analysis of the enzyme responsible or perhaps by expression and purification of B-Raf from a heterologous source.
Since the MAP kinase pathway is highly inducible in response to a number of ligands, we hypothesized that a physiological activator of MEKl would display transient activity in response to cell stimulation. Hence, we chose to examine bovine adrenal chromaffin cells where the MAP kinase pathway is rapidly induced in response to the secretagogue nicotine (63). Consistent with the identification of B-Raf as the major MEKl activator in bovine brain, chromaffin cells exhibit a MEKl activator that co-purifies with B-Raf and that can be immunoprecipitated with an antiserum specific for B-Raf, suggesting that B-Raf is also the major detectable MEKl activator in this cell type. However, we have been unable to demonstrate a regulation of MEKl activatorB-Raf activity in response to nicotine under conditions where endogenous MEKl activity was stimulated. Potential explanations for this result include regulation by relocation within the cell, as has been demonstrated for the c-Raf-1 product (43, 44, 721, facilitating interaction of effector and substrate or regulation by a noncovalent mechanism, e.g. associatioddissociation of a regulatory molecule or domain. Our data are suggestive of the latter possibility since the initial purification steps of MEKl activator from bovine brain gives a -4-fold increase in activity. Furthermore, immunoprecipitation of B-Raf from fractionated chromaffin cell extract yields -2.5-fold more MEKl activator activity than predicted, indicating the existence of an inhibitor in the active fractions. This enhancement of activity is unlikely to result from binding of the antibody since soluble anti-B-Raf antibody does not alter the activity of partially purified MEKl activator from bovine brain (data not shown). Cohen and colleagues have described a latent MEKl activator in PC12 cells that becomes active following prolonged storage, regardless of prior cell stimulation (73). The authors suggest that decay of a co-purifying inhibitor may be responsible for this unusual behavior (73).
The possibility that regulation of B-RaUMEK1 activator activity results from relocation of the enzyme following cell stimulation awaits clarification, although no evidence in support of this hypothesis was found by Stephens et al. (55). Furthermore, we have been unable to detect translocation of B-Raf polypeptide or MEKl activator activity to the particulate fraction of bovine adrenal chromaffin cells following stimulation with nicotine (data not shown), although inhibitors of B-Raf or MEKl in this fraction might render enzymatic assays inconclusive. Extraction of chromaffin cells with nonionic detergent also fails to reveal a regulation of B-Raf activity following nicotine stimulation, as measured by immune complex phosphorylation of Escherichia coli-produced glutathione S-transferase-MEKl fusion protein or activation of either fibroblast MEKl or glutathione S-transferase-MEK1 (data not shown).
Our data confirm that activation of MEKl correlates with phosphorylation of the enzyme on serine and threonine residues. Activation of MEKl in vitro with partially purified brain MEK activator, and in vivo in NGF-stimulated PC12 cells (20) correlate with the appearance of a serine-phosphorylated peptide. A peptide with similar properties has been described by Kyriakis et al. (27). We have not conclusively identified the sites utilized by the partially purified brain MEKl activator/ B-Raf, although the sites on MEKl phosphorylated by c-Raf-1 in vitro have been identified genetically (74) and biochemically (20). Alessi et al. (20) isolated a nonconventional tryptic phosphopeptide from c-Raf-1-phosphorylated MEKl and demonstrated, by amino acid sequencing, phosphorylation of serine residues corresponding to 218 and 222 of rat MEK. Our data are consistent with these results since cleavage at the corresponding position in rat kidney MEKl would be expected to yield a peptide with phosphorylation sites on serine residues at positions 4 and 8; phosphoamino acid analysis and repetitive Edman degradation indicate that this is the case. Hence, we suggest that B-Raf activates MEKl through phosphorylation of serine residues 218 and 222.
Serine 218 and 222 are phosphorylated in MEKl isolated from NGF-stimulated PC12 cells (20). The enzyme(s) responsible for this in vivo phosphorylation and activation is unknown, although previous studies in cells of neural origin (55, 56) and our data indicate that B-Raf is a good candidate. Significantly, expression of an activated c-Raf-1 allele in PC12 cells has little effect on the MAP kinase pathway (60), suggesting that a distinct enzyme is the physiological regulator of MEK in these cells.
Relatively little is known regarding the regulation of B-Raf activity. Since NGF-stimulation of MAP kinase in PC12 cells is dependent upon Ras, as determined by studies with dominant negative Ras mutants, one might speculate that B-Raf binds to, and becomes activated upon Ras-GTP. This model could in part explain the differential effects of cyclic AMP elevation in fibroblasts and smooth muscle cells (where MEK activity is inhibited, Refs. 4548) and PC12 cells (where MEK activity is elevated, Ref. 75) since B-Raf lacks serine 43 (531, a proposed site of protein kinase A-mediated inhibition of c-Raf-1 (46). Note however that the Ras-Raf interaction might be regulated independently of protein kinase A-mediated c-Raf-1 phosphorylation (76).
Although we were able to detect both B-Raf and c-Raf-1 by immunoblotting, MEK1-stimulating activity was associated quantitatively with B-Raf following chromatography of bovine brain homogenate or chromaffin cell extracts. Although we have been unable to quantitate the absolute levels of B-Raf and c-Raf-1, this suggests that c-Raf-1 is not a significant MEKl activator in brain or chromaffin cells. If this contention is correct, the differential regulation of these Raf isozymes becomes an interesting question. A-Raf and B-Raf are highly homologous to c-Raf-1 in the kinase domain (CR3) and CR1 (531, the conserved region that encompasses a putative zinc finger domain important for association with GTP-Ras (39-41). However, a third (short) region of sequence homology (94%) between A-Raf and c-Raf-1 is only 47% conserved in B-Raf (531, and sequences surrounding CR2 are largely isozyme specific. The function of this region is uncertain, but linker insertion mutagenesis of CR2 in c-Raf-1 yields a transforming protein (77). One might speculate that CR2 and surrounding B-Raf-specific sequences could in part explain differential regulation of the Raf isozymes.