Sequential modification of serines 621 and 624 in the Raf-1 carboxyl terminus produces alterations in its electrophoretic mobility.

The Raf-1 serine/threonine protein kinase plays a central role in many of the mitogenic signaling pathways regulating cell growth and differentiation. The regulation of Raf-1 is complex, and involves protein-protein interactions as well as changes in the phosphorylation state of Raf-1 that are accompanied by alterations in its electrophoretic mobility. We have previously shown that a 33-kDa COOH-terminal, kinase-inactive fragment of Raf-1 underwent a mobility shift in response to the stimulation of cells with serum or phorbol esters. Here we demonstrate that treatment of NIH 3T3 cells or Sf9 cells with hydrogen peroxide (H2O2) also induces the mobility shift of the kinase-inactive Raf-1 fragment. A series of deletion mutants of the Raf-1 COOH terminus were analyzed, and the region required for the mobility shift was localized to a 78-amino acid fragment (residues 566-643). Metabolic labeling revealed that the slower migrating forms of the 33-kDa and of the smaller fragment contained phosphorus. Mutation of a previously characterized phosphorylation site, serine 621, to alanine prevented the mobility shift as well as phosphate incorporation or Src and Ras-dependent kinase activation in Sf9 cells when this mutation was engineered into the full-length Raf-1. Mutation of 621 to aspartate yielded a protein that existed in both the shifted and unshifted forms, demonstrating that a negative charge at 621 was necessary, but not sufficient, for the mobility shift to occur; however, its full-length form was still resistant to activation in the Sf9 system. Additional mutation of nearby serine 624 to alanine blocked the shift, implicating this residue as the site of the second of a two-step modification process leading to the slower migrating form. Co-expression of the 33-kDa fragment with an activated form of mitogen-activated protein kinase kinase in NIH 3T3 led to the appearance of the shifted form in a serum-independent manner. These results demonstrate that a mitogen-activated protein kinase kinase-induced event involving modification of serines 621 and 624 leads to the mobility shift of Raf-1.

Raf-1, the product of the c-raf-1 protooncogene, is a cytoplasmic serine/threonine protein kinase that is activated in response to various mitogenic signals (1,2). Activation of Raf-1 results in a cascade of events involving the phosphorylation and subsequent activation of additional proteins, including mitogen-activated protein (MAP) 1 kinase kinase (MEK) and MAP kinase, which ultimately lead to transcriptional activation and mitogenesis (3,4). Thus the Raf-1 protein kinase plays a central role in cellular signal transduction and is a key player in transmitting signals from the plasma membrane to the nucleus (5,6).
The amino terminus of Raf-1 serves as a regulatory domain, while the catalytic domain resides in the carboxyl terminus, which also may contain regulatory elements (7). Earlier studies showed that the activation of Raf-1 by membrane-associated, protein-tyrosine kinases was dependent on cellular Ras activity (8 -10). Subsequently, it was found that Raf-1 binds directly to Ras, and the Ras-binding site was localized to the amino terminus of Raf-1 (11)(12)(13)(14). The interaction between Raf-1 and Ras results in the re-localization of Raf-1 to the plasma membrane (15)(16)(17), where Raf-1 is converted to a catalytically active form through a still unidentified process. Subsequently, activated Raf-1 kinase activates MEK (18 -21), which in turn activates MAP kinase (22). Activated MAP kinase phosphorylates both cytosolic and nuclear proteins (23), and plays a role in subsequent events, which lead to immediate early gene expression (24).
It has been demonstrated previously that Raf-1 exhibits an electrophoretic mobility shift that occurs after mitogenic stimulation of cells and that this event correlates with a change in the state of Raf-1 phosphorylation (25)(26)(27)(28). Additionally, we have shown that the region of Raf-1 responsible for the serum and phorbol ester-induced shift in gel mobility is localized to a catalytically inactive COOH-terminal fragment of Raf-1 (29). Since the amino-terminal domain of Raf-1 is responsible for direct binding to Ras (14), these results suggested that the modification of Raf-1 can take place in the absence of direct interaction with Ras.
To determine the site and nature of the modification(s) of Raf-1 that accompany the shift in electrophoretic mobility, additional deletion mutants of the carboxyl terminus and point mutants within this region were created and analyzed. From these experiments, the site of modification(s) was further localized to a region of 78 amino acids, within which was a known serine phosphorylation site (Ser-621). Therefore, we analyzed the role of this serine, as well as a neighboring serine (Ser-624), in the modification of Raf-1, through the creation of site-directed mutants of these residues. Metabolic labeling and im-munoprecipitation, as well as immunoblotting of the proteins encoded by these mutants, revealed that the altered electrophoretic mobility of the Raf-1 COOH-terminal fragments occurs following a two-step process that involves the addition of phosphate to two neighboring serine residues.

MATERIALS AND METHODS
Plasmids-Deletion mutants of Raf-1 were constructed to include a COOH-terminal epitope-tagging sequence encoding the last 12 amino acids of the protein kinase C epsilon (⑀) gene to provide rapid detection of the expressed protein. Deletion mutants were obtained by polymerase chain reaction amplification with the Pfu DNA polymerase (Stratagene), using a COOH-terminal ⑀-tagged fragment of Raf-1 (RIII-⑀) encoding amino acids 381-643 (30), as the template. NH 2 -terminal sense primers containing a BamHI restriction site were created at regular intervals originating at amino acid 381, 412, 443, 474, 505, 536, or 566 of Raf-1. A COOH-terminal antisense primer was generated containing the ⑀-tag, along with EcoRI and BamHI restriction sites. Using RIII-⑀ as template and the COOH-terminal ⑀-epitope tagging sequence as primer with various NH 2 -terminal primers, six additional NH 2 -terminally truncated epitope-tagged Raf-1 constructs were created. Several additional COOH-terminal ⑀-epitope tagged constructs were generated using COOH-terminal primers, which originated at amino acids 612, 581, or 550 along with the ⑀-tag, and with the NH 2terminal primer originating at amino acid 381. The full-length Raf-1 mutant plasmid pKSRafS621A was described previously (31). The plasmid encoding activated MEK-1 and a linked hygromycin resistance gene was provided by James C. Stone, University of Alberta, Edmonton (Canada) (32).
The RIII-⑀ single and double point mutants were generated by sequential PCR amplifications (33). The point mutations other than S621A were created using the RIII-⑀ construct as template, and the following sense primers (for S621D, 5ЈCCGGAGCGCTGACGAGC-CATCC3Ј; for S621D/S624A, 5ЈGCGCTGACGAGCCAGCCTTG-CATCGGGC3Ј) along with a COOH-terminal antisense primer containing the ⑀-tag and EcoRI and BamHI restriction sites. RIII-⑀ S621A was generated using the pKSRaf1S621A plasmid as template, the sense primer 5ЈCCGGAGCGCTGCCGAGCCATCC3Ј, and the same antisense primer as above. The PCR product from each of these reactions was then used as the antisense primer to generate RIII-⑀ mutant fragments, which were then cloned into baculovirus expression vectors for expression in Sf9 cells. The presence of point mutations was verified by direct DNA sequence analysis.
Transfection and Zinc-induced Overexpression of Recombinant Proteins in NIH 3T3 Cells-NIH 3T3 cells were maintained at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin/streptomycin, and glutamine. The isolation of NIH 3T3 cell lines expressing epitope-tagged Raf-1 mutants is described elsewhere (30). To induce maximal expression of the recombinant proteins, the transfected cells were changed to serum-free medium supplemented with 20 M zinc acetate (to up-regulate expression from the metallothionine promoter-linked genes) for 48 h prior to stimulation with 10% serum, 100 nM PMA, or 500 M H 2 O 2 . Stable transfectants expressing both RIII-⑀ and activated MEK-1 were isolated following cotransfection of cells with the expression plasmids and selection in neomycin and hygromycin. The Sf9 insect cell line was maintained in spinner flasks in Grace's supplemented medium containing 10% fetal bovine serum and 0.1% pluronic F-68 (Life Technologies, Inc.). To express the Raf-1 deletion mutants, Sf9 cells were co-transfected with the baculovirus transfer vector pEVMOD-Raf-⑀-tag and BaculoGold baculovirus DNA (Pharmingen) using the Lipofectin reagent (Life Technologies, Inc.). Recombinant baculoviruses encoding the various Raf-1 proteins were isolated and enriched according to the manufacturer's protocol (Pharmingen). For Raf-1 point mutants, the Bac-To-Bac Baculovirus expression system for Sf9 cells was used (Life Technologies, Inc.). For experimental purposes, Sf9 cells (2.5 ϫ 10 6 ) were infected with virus and harvested after 48 h.
Preparation of Cell Lysates and Cell Fractionation-For the experiments shown in Figs. 1 and 2, cells were washed twice with ice-cold phosphate-buffered saline, harvested by resuspension of the cell pellet in buffer A (20 mM Tris, pH 7.5, 2 mM EDTA, 2 mM EGTA, containing 1 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, 10 g/ml leupeptin, and 5 mM sodium orthovanadate), and disrupted by Dounce homogenization (100 strokes). Lysates were adjusted to 1 ϫ Laemmli sample buffer concentration and stored at Ϫ70°C. For the experiments shown in Table I and Figs. 3-6, cytosolic extracts of insect cells and mammalian cells were prepared as follows. The cells were washed twice with ice-cold phosphate-buffered saline, harvested by resuspension of the cell pellet in buffer B (0.05% SDS, 20 mM Tris, pH 8.0, 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 0.5% sodium deoxycholate, 2 mM EDTA, containing 1 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, 10 g/ml leupeptin, and 5 mM sodium orthovanadate) for insect cells and buffer C (same as B except 0.1% SDS) for mammalian cells. Cytosolic fractions were isolated by centrifugation of the cell homogenates at 15,000 ϫ g for 10 min at 4°C. Lysates for use in immunoblotting were adjusted to 1 ϫ Laemmli sample buffer concentration and stored at Ϫ70°C.
Immunoprecipitation and Immunoblot Analysis-Immunoprecipitation was performed on cytosolic extracts using anti-PKC⑀ (5 g/ml; Life Technologies, Inc.) and protein A-Sepharose beads (Calbiochem). After incubation for 4 -24 h at 4°C, immunoprecipitates were washed as described previously (34) and processed for SDS-PAGE. Immunoprecipitates were prepared for electrophoresis by the addition of 2 ϫ Laemmli sample buffer with ␤-mercaptoethanol and boiled prior to electrophoresis on an SDS-polyacrylamide gel (12%). The proteins were transferred to nitrocellulose, and immunoblot analysis was performed using 10 g/ml anti-PKC-⑀ antibody as described previously (30).
Metabolic Labeling and Immunoprecipitation of the ⑀-tagged Recombinant Proteins-Cell labeling experiments were performed by first rinsing infected cells once with either methionine-or phosphate-deficient medium. The cells then were incubated either in methioninedeficient medium containing 2.5% dialyzed fetal calf serum and 250 Ci/ml [ 35 S]methionine (Expre 35 S 35 S Label; DuPont NEN), or in phosphate-deficient medium with 2.5% dialyzed fetal calf serum and 1 mCi/ml [ 32 P]orthophosphate (Amersham). After 3 h of labeling, the cells were stimulated by the addition of 500 M hydrogen peroxide (H 2 O 2 ) for the final hour of labeling. The labeled cells were harvested and disrupted, and immunoprecipitation of the ⑀-tagged Raf-1 proteins was carried out as described above.
Raf-1 Protein Kinase Activity Assays-Recombinant human ⑀-epitope-tagged Raf-1 proteins were specifically immunoprecipitated as described above from lysates of singly infected Sf9 cells, or from cells coexpressing Raf-1, p21 ras , and activated pp60 src . The immunoprecipitates were washed three times with lysis buffer, and once with kinase buffer (20 mM Tris, pH 7.4, 20 mM NaCl, 1 mM dithiothreitol, 10 mM MgCl 2 ). Raf kinase activity was assayed by the phosphorylation of kinase-negative MEK, and the autophosphorylation of Raf-1 was assayed by omitting the MEK from the reactions (35). Briefly, the washed immunoprecipitates were incubated in 40 l of kinase buffer containing 100 ng of kinase-negative MEK, 10 M ATP, and 5 Ci of [␥-32 P]ATP at 30°C for 30 min. The reactions were terminated by addition of Laemmli sample buffer and boiling. The samples were separated by SDS-PAGE, and phosphorylation of the kinase-inactive MEK was determined by autoradiography. Alternatively, Raf-1 protein kinase activity was assayed using a coupled assay of extracellular-regulated kinase and MAP kinase activation (21), and the activity was determined by the phosphorylation of myelin basic protein by the activated MAP kinase. Briefly, Raf-1 immunoprecipitates were incubated with 2 g of recombinant MAP kinase (Santa Cruz Biotechnology) in reaction mixtures (40 l) of kinase buffer containing 200 M ATP along with 1 g of recombinant active MEK (Santa Cruz Biotechnology). After 30 min at 30°C, 10 l of the reaction mixtures were added to 40 l of ice-cold kinase buffer containing 100 M Na 3 VO 4 . Ten microliters of the diluted reaction mixture was combined with 30 l of kinase buffer containing 20 g of myelin basic protein and 5 Ci of [␥-32 P]ATP. After incubation for 20 min at 30°C, assays were terminated by the addition of SDS-PAGE sample buffer, the samples were resolved by SDS-PAGE, and phosphoproteins were visualized by autoradiography.

Mobility Shift of a COOH-terminal Raf-1 Fragment in Mammalian and Insect
Cells-Previously, we identified a 33-kDa catalytically inactive COOH-terminal fragment of Raf-1 (amino acids 381-643; designated RIII-⑀), which exhibited a Ras-independent serum-and phorbol ester-induced shift in gel mobility that mimicked the shift observed with full-length Raf-1 (29). Recent interest has focused on the possible role(s) of oxygen free radicals, including hydrogen peroxide (H 2 O 2 ), in the regulation of various signal transduction processes. Oxygen free radicals have been suggested to play a role in regulating several enzymes involved in transmembrane signaling pathways, including protein kinase C, tyrosine-specific protein kinases, and MAPK in certain cell systems (36 -38). Moreover, in one recent report the normal responses of vascular smooth muscle cells to platelet-derived growth factor, including MAPK activation and DNA synthesis, were shown to require H 2 O 2 generation (39).
Since Raf-1 functions as an upstream activator in the MAPK kinase cascade, we investigated whether Raf-1 might also be affected by treatment of cells with H 2 O 2 . To examine the effect of various agents on Raf-1 mobility, quiescent NIH 3T3 cells expressing RIII-⑀ were treated with 10% serum, 100 nM PMA, or 500 M H 2 O 2 . Cytosolic extracts were isolated and separated by SDS-PAGE, then subjected to immunoblot analysis ( Fig. 1). In quiescent cells, only a single electrophoretic form of the RIII-⑀ protein was observed (Fig. 1, lane 1). In contrast, two forms of RIII-⑀ were observed following treatment with serum or phorbol ester, as shown previously, or hydrogen peroxide (Fig. 1, lanes 2-4). The similar migration rates of the upper band induced by each treatment suggested that this form of RIII-⑀ resulted from a similar modification(s) in each instance.
Experiments performed using the Sf9 insect cell system to analyze the activation of Raf-1 have revealed that co-expression of Src, Ras, or PKC with Raf-1 led to its activation (9,10,20,31). To develop a simplified system for the analysis of Raf-1 COOH-terminal modifications, we examined whether treatment of Sf9 cells expressing only RIII-⑀ with H 2 O 2 would lead to a shift in electrophoretic mobility similar to that observed in mammalian cells. After 48 h of infection with an RIII-⑀ expressing baculovirus, readily detectable levels of RIII-⑀ protein were expressed (Fig. 2). Unexpectedly, a doublet corresponding to RIII-⑀ was observed in the untreated cells (Fig. 2, lane 1). The doublet is probably a result of the presence of serum in the infected Sf9 culture. However, the more rapidly migrating form, corresponding to the unmodified form of RIII-⑀, was found in greater amount in the untreated cells. As early as 30 min after H 2 O 2 treatment, a greater amount of the more slowly migrating form of Raf-1 was observed (Fig. 2, lane 2), and after 60 min, almost all of the RIII-⑀ was found in the modified form (Fig. 2, lane 3). Thus, a similar shift to a more slowly migrating form of RIII-⑀ was noted both in insect cells and in mammalian cells following H 2 O 2 treatment.
Deletion Analysis of the Raf-1 COOH Terminus in Sf9 Cells-To further localize the region within the Raf-1 COOH terminus required for modification(s) to occur, which result in a shift in gel mobility, a series of deletion mutants of the RIII-⑀ (amino acids 381-643) protein were created and expressed in Sf9 cells. Six NH 2 -terminal and three COOH-terminal deletion mutants were constructed in 31 amino acid intervals encompassing residues 381-643 of Raf-1 (Table I). After 48 h of infection with the various virus stocks, Sf9 cells were treated with H 2 O 2 for 60 min to fully induce the characteristic shift in electrophoretic mobility. All of the NH 2 -terminal deletion mu-tants exhibited a mobility shift after treatment (Table I), including the smallest polypeptide (RIX-⑀), which encompasses only amino acids 566 -643 of Raf-1. In contrast, all deletions from the COOH terminus resulted in proteins that exhibited only a single band in the absence or presence of H 2 O 2 . Further efforts to define the minimal fragment (RIX-⑀), which still exhibited a band shift, were unsuccessful, probably as a result of instability, since no expression of the smaller proteins was detected (data not shown). We conclude that a region of Raf-1 comprising only amino acids 566 -643 is capable of undergoing modification(s) resulting in a shift in gel mobility upon stimulation of cells with H 2 O 2 .
Phosphorylation State of RIII-⑀ and RIX-⑀-A change in the phosphorylation state of Raf-1 after mitogenic stimulation has been suggested as the explanation for the shift in migration rate of Raf-1 on SDS-polyacrylamide gels (26 (Fig. 3,  lane 2). Following 32 P i labeling, however, the major band ob-  served was the upper band of the RIII-⑀ doublet (Fig. 3, lane 5). These results link the modification responsible for the migration shift of RIII-⑀ to the presence of phosphate in the protein.
Only a low level of 35 S label was incorporated into RIX-⑀, likely due to the low number of methionine and cysteine residues present in this fragment (Fig. 3, lane 3). However, a doublet was visible with much longer exposures of the gel (data not shown). Nonetheless, RIX-⑀ did exhibit significant 32 P i incorporation (Fig. 3, lane 6), and once again, only a single band was observed in the 32 P i -labeled lysates. These results suggested that the change in the rate of migration of the Raf-1 COOHterminal fragments may be associated with a phosphorylation event occurring between residues 566 and 643.
Identification of Possible Sites of Modification Affecting the Electrophoretic Mobility of RIII-⑀-One site of modification that lies within the 566 -643 region is serine 621, which was previously shown to be constitutively phosphorylated in fibroblast cells (31). Additionally, mutation of this site to an alanine resulted in the absence of phosphorylation of Raf-1, and these mutants exhibited a dominant-negative phenotype (7,40). To determine what role serine 621 might play in the mobility shift observed in RIII-⑀ upon mitogenic stimulation, point mutations resulting in a change of this residue to either an alanine (RIII-⑀ S621A) or aspartate (RIII-⑀ S621D) were constructed and expressed in Sf9 cells. Although RIX-⑀ is the smallest fragment that exhibits a mobility shift, the resolution between the two forms is more evident using RIII-⑀. The S621A mutation prevents phosphorylation at site 621, while the negatively charged residue in S621D should mimic the presence of a phosphorylated serine. As expected, mutation of serine 621 to alanine prevented further modification of RIII-⑀, resulting in the presence of only the faster migrating form of the protein (Fig. 4,  lane 2). Surprisingly, however, mutation of serine 621 to aspartate resulted in the presence of a doublet (Fig. 4, lane 3), comparable to the doublet observed for wild-type RIII-⑀ (Fig. 4,   lanes 1 and 5). If phosphorylation of this site alone was responsible for the induced mobility shift, we would have expected to observe the presence of a single, slowly migrating band. Therefore, the modification at Ser-621 appears to be necessary, but not sufficient, for the induced mobility shift of RIII-⑀.
To identify a possible second site of modification, the sequences of different Raf proteins in the region from amino acids 566 -643 were examined for the most evolutionarily conserved residues. One site, serine 624, was of particular interest because of its conservation between the three isoforms of Raf and its proximity to serine 621. To examine what effect this residue might have on the mobility shift of the Raf-1 COOH terminus, we constructed a double point mutant in the RIII-⑀ construct, which changed serine 624 to alanine and serine 621 to aspartate (RIII-⑀ S621D/S624A). Immunoblot analysis of cytosolic extracts from Sf9 cells expressing this double mutant after H 2 O 2 treatment revealed a band pattern that differed from the one that was observed with the single mutant (RIII-⑀ S621D). Only the more rapidly migrating band was observed in lysates containing the RIII-⑀ S621D/S624A double mutant (Fig. 4, lane  4). These data suggest that two modifications occur in the COOH terminus of Raf-1 to result in the electrophoretic mobility shift of RIII-⑀. Apparently, an initial modification occurs at serine 621, and this is required to allow a second modification to occur at serine 624.
Phosphorylation State of RIII-⑀ Proteins Bearing Point Mutations-To determine whether the electrophoretic mobility changes reflected alterations in the phosphate content of Raf-1 proteins, in vivo labeling experiments were performed using the RIII-⑀ single point mutants, RIII-⑀ S621A and RIII-⑀ S621D. Infected Sf9 cells were labeled with either [ 35 S]methionine or [ 32 P]orthophosphate, exposed to H 2 O 2 for 60 min, then lysed, and the lysates were analyzed as for Fig. 3. Metabolic labeling experiments with [ 35 S]methionine showed that adequate levels of protein were expressed for wild-type RIII-⑀ and the two point mutants of RIII-⑀ (Fig. 5, lanes 1-3). A doublet was evident for both wild-type RIII-⑀ and RIII-⑀ S621D, while only a single, more rapidly migrating band was observed for RIII-⑀ S621A (consistent with the immunoblot analysis in Fig.  4). In contrast, when cells were labeled with 32 P i , phosphate was incorporated into both the faster and more slowly migrating forms of wild-type RIII-⑀ protein (Fig. 5, lane 4), while no phosphate was incorporated into the RIII-⑀ S621A protein (Fig.  5, lane 5). Although the RIII-⑀ S621D mutant expressed both forms when [ 35 S]methionine-labeled lysates were analyzed (Fig. 5, lane 3), only the upper band was 32 P-labeled (Fig. 5,  lane 6). This indicates that an alteration requiring a negative charge at site 621 occurred in the RIII-⑀ S621D protein. This modification, which produced the migration shift, also involves the addition of phosphate to the RIII-⑀ protein.
Activated MEK-1 Induces the RIII-⑀ Mobility Shift in Mammalian Cells-Since the activation of Raf-1 appears to involve phosphorylation, we were interested in examining potential protein kinases that might induce the observed mobility shift of RIII-⑀. It has been proposed that the shift in Raf-1 electrophoretic mobility may result from a feedback loop following the activation of the MEK/MAPK pathway (41)(42)(43). To examine the effect MEK-1 might have on the migration of RIII-⑀, we introduced an activated form of MEK-1 (32), which has been shown to induce focus formation in mammalian cells (44,45), into a NIH 3T3-derived cell line expressing RIII-⑀. Immunoblot analysis was performed with proteins extracted from stable transfectants after serum deprivation for 48 h. In the absence of exogenous activated MEK-1, only a single, faster migrating form of RIII-⑀ was observed in serum-deprived cells (Fig. 6, lane  2). However, in serum-deprived transfectants expressing activated MEK-1 (lane 3), a doublet was observed that was comparable to the doublet observed in exponentially growing RIII-⑀ cells (lane 1). Thus, the presence of activated MEK-1 is capable of inducing the mobility shift of RIII-⑀ in mammalian cells.
Effects of Mutations at Amino Acids 621 and 624 on Raf-1 Kinase Activity-To determine whether the Raf-1 modifications involving residues 621 and 624 might alter the kinase activity of the native Raf-1 protein, full-length wild-type Raf-1, or mutant proteins bearing amino acid changes at these residues were expressed in Sf9 cells, and the kinase activity of the immunoprecipitated Raf-1 proteins was assayed (Fig. 7). As the in vitro kinase activity of Raf-1 has been shown to be greatly stimulated when the protein is isolated from Sf9 cells co-expressing pp60 src and p21 ras (9, 31), we analyzed the activity of Raf-1 in triply infected Sf9 cells (Fig. 7, lanes 2-6), as well as in cells expressing Raf-1 alone (lanes 7-11). The phosphorylation of kinase-inactive MEK by Raf-1 (Fig. 7B), as well as Raf-1 autophosphorylation (Fig. 7A) was assayed (35). In addition, we employed a coupled assay (21) in which Raf-1 immunoprecipitates were incubated with wild-type MEK and wild-type MAP kinase, and the activation of MAP kinase was determined by the phosphorylation of myelin basic protein (Fig. 7C, MBP).
In each type of assay, the activity of wild-type Raf-1 was readily detected when the protein was isolated from triply infected cells (lane 2), while only a low level of activity was observed in precipitates from cells infected singly with Raf-1 (lane 7). As previously reported (31), mutation of serine 621 to alanine (S621A) abolished the in vitro kinase activity of Raf-1 (lanes 3 and 8). Unexpectedly, Raf-1 protein with serine 621 changed to aspartate (S621D; lanes 4 and 9) was also defective in its kinase activity. Thus, replacement of the normal serine with a neutral amino acid (alanine) or a negatively charged amino acid (aspartate) at this site has a similar effect on Raf-1 activity. As the modification of serine 624 requires a negative charge at site 621 (see above), we also analyzed the double point mutant in which the S621D mutation was combined with a serine-to-alanine change at amino acid 624 (S621D/S624A; lanes 5 and 6 and lanes 10 and 11). The double mutant protein, which in the RIII-⑀ form is unable to undergo the modifications required to cause a shift in electrophoretic mobility, was also found to be kinase-inactive. In all cases, the wild-type and mutant Raf-1 proteins were expressed and were present in the immunoprecipitates, as detected by immunoblotting (Fig. 7D). We conclude that the presence of a negative charge at amino acid 621 does not yield an activated, or even a kinase-competent Raf-1 protein. Furthermore, preventing the modification of serine 624 by replacing it with alanine does not restore kinase activity to the protein in which serine 621 has been changed to aspartate.

DISCUSSION
Several laboratories have presented data suggesting a correlation between the phosphorylation state of Raf-1 and activation of its kinase activity (31,46,47). While hyperphosphorylation also has been suggested to play a role in the shift in electrophoretic mobility of activated Raf-1 on SDS-polyacrylamide gels (48 -50), the identity of specific site(s) responsible for this shift in gel mobility has remained elusive. Here we present evidence that modification of serine 624, along with phosphorylation of serine 621, is required for the electrophoretic mobility shift of a COOH-terminal fragment (RIII-⑀) comprising residues 566 -643 of Raf-1. The band shift noted with the RIII-⑀ fragment resembles the serum-and PMA-induced shift observed with full-length Raf-1. It is likely that the same modifications we have described here in the context of Raf-1 COOH-terminal fragments also occur with full-length Raf-1, since this protein displays an analogous electrophoretic mobility shift in response to treatment of cells with the same agents.
The regulation of serine 624 appears to be complex; it is a two-step process requiring the initial, obligatory phosphorylation of a neighboring site (serine 621) prior to the modification of serine 624 to produce the characteristic gel mobility shift. While mutation of serine 621 to alanine abolished the mobility shift of RIII-⑀, mutation of serine 621 to aspartate did allow the shift to occur, although only a portion of the S621D protein was in the more slowly migrating form. Mutation of serine 624 to alanine, in the context of S621D, prevented the mobility shift. Thus, a negative charge at residue 621 is necessary for the shift to occur, but it is not sufficient to produce the shift. In vivo phosphate labeling experiments revealed that phosphate was incorporated into both the upper and lower bands of the wildtype RIII-⑀ protein, as well as into the slowly migrating form of the S621D protein, but not into the S621A mutant. These results suggest that a second modification, involving serine 624, is required for the mobility shift, and that this change occurs only after the modification of serine 621.
At least three forms of Raf-1 have been observed when qui-escent cells are mitogenically stimulated (31), 2 suggesting successive post-translational processing of the protein. Similarly, PKC undergoes several phosphorylation events to generate a mature, catalytically active form (51)(52)(53). The first modification of PKC is a transphosphorylation event that activates the kinase, while the second event is an autophosphorylation process that appears to play a role in the subcellular localization of PKC (54). Additionally, an acidic residue must be present at the site of the first modification in order for the second event to occur (55). This requirement for a negative charge in order for a second phosphorylation to occur also has been reported for the regulation of glycogen synthase, CDK2, and MAP kinase (56 -58). Similarly, based on our results, the addition of phosphate moieties on Raf-1 in a successive manner appears to regulate the electrophoretic mobility, and possibly the activity, of this protein.
Serine 621 of Raf-1 has previously been shown to be phosphorylated in quiescent as well as in platelet-derived growth factor-stimulated fibroblasts (31). The importance of this residue for Raf-1 function is suggested by the lack of kinase activity exhibited by Raf-1(S621A) in vitro and in vivo (31,40). Here we have made the unexpected observation that substitution of serine 621 with aspartate (S621D) also yields an inactive kinase. Using a constitutively activated COOH-terminal fragment of Raf-1, other investigators have independently reached the same conclusion (59). The results suggest that either Raf-1 proteins with a negative charge (e.g. a phosphorylated serine) at this site are inactive, or that the aspartate substitution does not completely mimic the phosphorylated serine. Our finding that mutation of serine 621 to aspartate does permit the electrophoretic mobility shift to occur, while the alanine 621 mutation does not, suggests that aspartate at this residue does substitute for at least this aspect of serine 621 function.
Our results raise several questions regarding the regulation of Raf-1 function. One is the nature of the modifications themselves. While the metabolic labeling experiments clearly demonstrated the incorporation of phosphate into RIII-⑀ as a correlate of the mobility shift, the phosphate group could comprise part of a more complex modification rather than phosphorylation per se. Indeed, some investigators have found no effect by either phosphoserine-or phosphotyrosine-specific phosphatases on Raf-1 enzymatic activity (17,29,60). In contrast, more recent studies have shown that treatment of Raf-1 with membrane-associated protein phosphatases can inactivate its kinase activity (47). Thus, while the simplest explanation is that both serine 621 and serine 624 undergo phosphorylation, we cannot exclude the possibility that other modifications occur at these sites. A second and related question is the identity of the enzymes responsible for the modifications of the Raf-1 COOH terminus. There could be a single enzyme that modulates both events, or there might be two distinct enzymes involved. The occurrence of phosphorylation at serine 621 in both quiescent and stimulated cells makes it difficult to speculate about the identity of the kinase(s) responsible. However, the ability of MEK-1, which has a very strict substrate specificity for MAPK (45,61), to induce the Raf-1 shift implies that MAPK, or another kinase activated by MAPK, may be responsible for the modification of serine 624.
A final critical issue is the role of these residues and their modifications in the regulation of Raf-1 activity. Initially, it appeared that the mobility shift of Raf-1 correlated with increased kinase activity of the protein, especially since agents that induced the shift (e.g. serum, PMA) also activated Raf-1 activity. Later reports, however, suggested that Raf-1 activa-2 A. F. Ferrier, unpublished data.

FIG. 7. Catalytic activity of wild-type and mutant Raf-1 proteins measured in vitro.
Panels A-D, lane 1, uninfected; lanes 2 and 7, wild-type Raf-1; lanes 3 and 8, Raf-1 in which the serine residue at 621 was mutated to alanine (S621A); lanes 4 and 9, Raf-1 in which the serine residue at 621 was mutated to aspartate (S621D); lanes 5, 6, 10, and 11, Raf-1 in which the serine residue at 621 was mutated to aspartate and the serine residue at 624 was mutated to alanine (S621D/ S624A). Wild-type and mutant ⑀-epitope-tagged Raf-1 proteins were immunoprecipitated from uninfected Sf9 cells (lane 1), cells expressing Raf-1 proteins alone (lanes 7-11), or cells coexpressing Raf-1 proteins, p21 ras , and activated pp60 src proteins (lanes 2-6), as indicated. A, Raf-1 autophosphorylation was assayed using immunoprecipitated Raf-1 proteins as described under "Materials and Methods." The in vitro 32 Plabeled Raf-1 proteins were separated on 10% SDS-polyacrylamide gels, and radiolabeled proteins were visualized by autoradiography. B, Raf-1 protein kinase assays were performed as for A, but using kinase-negative MEK as phosphoacceptor substrate. Samples were resolved by electrophoresis on 10% SDS-polyacrylamide gels, and phosphoproteins were visualized by autoradiography. C, Raf-1 protein kinase also was assayed using a coupled assay of MAP kinase activation as determined by the phosphorylation of myelin basic protein (MBP). The in vitro phosphorylated 32 P-labeled myelin basic proteins were separated on 12.5% SDS-polyacrylamide gels and visualized by autoradiography. D, immunoblot analysis of the expression of Raf-1 proteins. Immunoprecipitates were prepared in parallel with those assayed for kinase activity, separated by SDS-PAGE, transferred to filters, and probed with anti-PKC⑀ antibody. tion might actually occur prior to the mobility shift, raising the possibility that the mobility shift may in fact reflect a feedback mechanism to turn off Raf-1 (42,43). Thus, it also is possible that the more highly modified (slower migrating) form of Raf-1 may actually be inactive. The experiments described here do not definitively distinguish between these possibilities. Indeed, the lack of kinase activity by Raf-1(S621A) would seem to support the former model, while the lack of activity of Raf-1(S621D) would seem to support the latter. However, the fact that MEK-1 can induce the band shift in the RIII-⑀ fragment when co-expressed in intact NIH 3T3 cells is most easily interpreted in light of the feedback model. Further experiments are required to determine the precise manner in which Raf-1 enzymatic activity is affected by these modifications.