Phosphorylation of the Ras Nucleotide Exchange Factor Son of Sevenless by Mitogen-activated Protein Kinase*

Son of sevenless-1 and -2 (Sos-1 and -2) are guanosine nucleotide exchange factors implicated in the activation of Ras by both the insulin and epidermal growth factor signal transduction pathways. Ras appears to function by initiating the activation of cellular protein kinases including mitogen-activated protein ( M A P ) kinases. Sos proteins contain numerous sequences in their carboxyl-terminal regions which correspond to consensus sites for MAP kinase phosphorylation. To examine whether these sites are substrates for MAP kinases, the cDNA encoding Drosophila Sos (dSos) was tagged with se- quences encoding the major antigenic epitope of the influenza virus hemagglutinin (HA) to create a dSosHA fusion construct. dSosHA was transiently expressed in COS-1 cells and immunoprecipitated with anti-HA anti- bodies. When immune complexes were incubated with purified MAP kinase and [ys2P]ATP, a phosphorylated band of 180 kDa was observed when analyzed by SDS-polyacrylamide gel electrophoresis. This band was not present in immunoprecipitations from cells transfected with vector alone. No phosphorylation of the 180 kDa band was seen when immunoprecipitates were incubated with [yS2P]ATP in the absence of MAP kinase. Two dimensional analysis of tryptic peptides from dSosHA phosphorylated by MAP

The abbreviations used are: EGF, epidermal growth factor; GRF, guanine nucleotide-releasing factor; IRS-1, insulin receptor substrate-1; M A P , mitogen-activated protein; PAGE, polyacrylamide gel electrophoresis; GDS, guanine nucleotide dissociation stimulation. Satoh et al., 1990;Burgering et al., 1991;Medema et al., 1991;Thomas et al., 1992;Wood et al., 1992). Stimulation of the tyrosine kinase activities of the receptors results in the rapid release of Ras-bound guanosine diphosphate, which is catalyzed by a class of proteins, the guanosine nucleotide exchange factors. Formation of biologically active Ras occurs by the subsequent binding of GTP. The first guanosine nucleotide exchange factor identified was the Saccharomyces cerevisiae CDC25 protein (Broek et al., 1987). In higher eukaryotes, two different types of proteins with homology to CDC25 have been found. One includes the p140 Ras GRF-like proteins, which in human, rat, and mouse are found primarily if not exclusively in brain tissue (Martegani et al., 1992;Shou et al., 1992;Wei et al., 1992). These exchange factors have not yet been implicated in any tyrosine kinase receptor signaling pathway. The second group includes the Son of sevenless proteins (Sos). Sos was first identified in Drosophila melanogaster, in which genetic studies showed that Sos is downstream of both the Sevenless and EGF receptor tyrosine kinases (Rogge et al., 1991). One human and two murine homologues of Sos have been cloned (Bowtell et al., 1992;Chardin et al., 1993), and unlike the GRF exchange factor, Sos is ubiquitously expressed in mouse and human tissues.
A great deal of recent data have implicated Sos as part of the EGF receptor signaling complex (Buday and Downward, 1993;Egan et al., 1993;Rozakis-Adcock et al., 1993;Li et al., 1993;Gale et al., 1993). Stimulation of the EGF receptor tyrosine kinase results in the formation of a complex between EGF receptor, Sos, and the adapter protein Grb2. The Grb2 protein contains an SH2 domain, which is believed to bind to the EGF receptor autophosphorylation site Y1068, flanked by two SH3 domains, which bind to proline-rich sequences at the COOH terminus of Sos (Rozakis-Adcock et al., 1993;Li et al., 1993). The formation of this complex does not measurably increase the guanosine nucleotide exchange activity that Sos has for Ras i n vitro (Buday and Downward, 1993). However, EGF stimulation results in the translocation of Sos from cytoplasmic to particulate fractions. This suggests that EGF stimulation results in a change in the intracellular location of Sos, bringing it in contact with membrane-associated Ras (Buday and Downward, 1993). In contrast to EGF signaling, activation of the insulin receptor tyrosine kinase results in association of Sos-GRB2 complexes with tyrosine-phosphorylated IRS-1 (insulin receptor substrate-1) and Shc (Baltensperger et al., 1993;Skolnik et al., 1993). Thus in the insulin signaling pathway, Sos guanosine nucleotide exchange activity may be regulated through the translocation of IRS-1 and Shc proteins.
We have recently shown that COS-1 cells cotransfected with Drosophila Sos (dSos) cDNAand human H-Ras cDNAcontain 10 times the amount of GTP-bound R a s than cells transfected with H-Ras alone. Furthermore, dSos binds to IRS-1 only when an active insulin receptor tyrosine kinase is present (Baltensperger et al., 1993). The dSos cDNAwas also found to transform Rat-1 cells (Egan et al., 1993). Taken together these results indicate that dSos protein is functional in mammalian cells.
Phosphoamino acid analysis showed that the dSos protein is phosphorylated on serine and threonine but not tyrosine residues in COS-1 cells (Baltensperger et al., 1993). This indicates that Sos is not a substrate of the insulin receptor itself, but it may be a substrate of serine-threonine kinases downstream of Ras. One group of serine-threonine kinases that are activated by tyrosine kinase signaling pathways through Ras are MAP 4717 "-2 Kinase Phosphorylation of Ras Activator kinases (de Vries-Smits et al., 1992;Medema et al., 1991;Thomas et al., 1992;Wood et al., 1992). The consensus sequence for MAP kinase phosphorylation (Clark-Lewis et al., 1991;Gonzalez et al., 1991) exists seven to nine times in the COOHterminal domains of all Sos proteins. This study addresses whether dSos is a substrate of MAP kinase. The results reported here demonstrate that dSos is in fact phosphorylated by MAP kinase in vitro and suggest this reaction also occurs in intact cells.

EXPERIMENTAL. PROCEDURES
Construction ofpCMV5-dSosHA-A cDNA clone of the D. melanogaster Sos gene (dSos) (Bonfini et al., 1992) in Bluescript was a gift of Utpal Banejee. Sequences encoding the 9-amino acid peptide sequence of the major antigenic epitope of influenza virus hemagglutinin (YPYDVP-DYA) were added to the 3' end of the dSos coding sequence by the polymerase chain reaction (Saiki et al. 1988). This was conducted with dSos digested with HindIII and the following two primers: 5'-TC-

TAGAGGATCCTAAGCGTAATCTGGAACATATGGATATTCT
GTACTTG and 5'-GACGCAGTCGCGCTCGTC.-A 195-base pair fragment was generated and digested with BglI and BamHI. The resulting 130-base pair BglI-BamHI fragment was ligated to a 5.5-kilobase pair HindIII-BglI fragment of Sos, which contains the 5' end of the dSos gene, and to pCMV5 (Andersson et al. 1989) cut with HindIII and BamHI. The resulting construct is denoted pCMV5-dSosHA. To verify the presence of the influenza virus tag in the construct, the 3' end of the coding sequence of pCMV5-dSosHA was determined by the dideoxy method (Sanger et al. 1977) using the Sequenase version 2.0 kit (U. S. Biochemical Corp.) with both primers described above.
'Dansfection, Cell Labeling, and Immunoprecipitation-Plates (6 cm) of COS-1 cells were transfected using N-[1-(2, 3 dioleoyloxy)propyl]-N,N,N-trimethylammoniummethylsulfate (Boehringer Mannheim) according to the manufacturer's instructions using 5 pg of plasmid DNA per plate. Two days after transfection, the cells were washed twice with 3 ml of serum-free Dulbecco's modified Eagle's medium and once with phosphate-buffered saline (1.8 m~ KH,PO,, pH 7.2, 171 m~ NaCl, 1.0 m~ Na,HPO,, and 3.4 m~ KCl). The cells were labeled for 4 h with 10 mCi of 32Pi in 1.5 ml serum-free Krebs-Ringer bicarbonate Hepes buffer (10 m~ Hepes, pH 7.5,120 m~ NaCl, 4.7 m~ KCl, 1.2 m~ MgCl,, 1.2 m~ CaC12, and 24 m~ NaHC03). The plates were washed once with 5 ml of cold phosphate-buffered saline, and cells were lysed in 1 ml cold lysis buffer (20 m~ Hepes, pH 7.9, 50 nm (NH4),S04, 50 mM NaF, 1 mM Na3V04, 1 m~ EDTA, 10% glycerol, 1% Triton X-100, 0.1% Tween 20, 1 m~ dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 5 pg/ml each leupeptin, aprotinin, and pepstatin, and 1 m~ benzamidine). The lysates were spun in a microcentrifuge a t 15,000 x g for 10 min a t 4 "C. The supernatants were precleared by addition of 10 p1 of protein A-Sepharose (Pharmacia LKB Biotechnology Inc.) and incubated on a n end-over-end mixer a t 4 "C for 1 h. The samples were then centrifuged at 15,000 x g for 2 min at 4 "C, and the supernatants were incubated on a n end-over-end mixer with 20 pl of 12CA5 antibody (Wilson et al. 1984) and 30 pl of protein A-Sepharose (Pharmacia) for 16 h. The Sepharose was pelleted by centrifugation at 15,000 x g for 2 min at 4 "C. Pellets were washed five times with cold lysis buffer, and the protein was dissolved in SDS-PAGE sample buffer.
Phosphoamino Acid Analysis and Pyptic Mapping-After separation by SDS-PAGE, the protein was transferred to Immobilon-P membranes (Millipore). The location of Sos was visualized by autoradiography. For phosphoamino acid analysis, protein was hydrolyzed directly on the membrane (Kamps and Sefton, 1989). Relevant pieces were cut from the membranes and heated in 100 pl of 6 N HCl to 110 "C for 1 h. The supernatants were lyophilized and dissolved in 5-10 pl of water with non-radioactive phosphoamino acid markers (1 mg/ml each), and they were applied to cellulose thin-layer plates (Machery-Nagel). Electrophoresis was for 1.5 h a t 1000 V (Hunter and Sefton, 1980). The positions of the markers were determined by ninhydrin staining.
For tryptic mapping, excess protein binding sites were blocked by incubating the membranes in 0.05% Tween 20 in water for 15-30 min, followed by washing three times in water. Relevant pieces were cut from the membranes and incubated overnight with 1 mg of trypsin in 100 pl of 50 m~ NH4COOH. The eluted peptides were lyophilized; excess NH,COOH was removed by redissolving the pellet in water and lyophilizing repeatedly. The peptides were dissolved i n 5 pl of water and applied to 20 x 20-cm cellulose thin-layer plates. Amounts of radioactivity applied to the plates were determined by Cerenkov counting before and after loading. For the first dimension, peptides were sepa-rated by electrophoresis a t pH 1.9 in 88% formic acidacetic acidwater (25:78:897) at 750 V for 2 h. The plates were dried and subjected to chromatography in 2-butanol/pyridin/acetic acidwater (15:10:3:12) in the second dimension.
To determine the phosphoamino acid content of tryptic peptides, relevant areas of cellulose were scraped from thin-layer plates and placed in microcentrifuge tubes. The peptides were eluted by addition of 1 ml of 10 m~ NH4COOH and agitating on an end-over-end mixer for 1 h. The cellulose was spun down by centrifugation for 5 min in a microcentrifuge; the supernatants were lyophilized, and phosphoamino acids were analyzed a s described above.
I n Vitro Labeling of dSosHA-Material from eight 6-cm plates transfected with PCMV5-dSos and from four 6-cm plates transfected with PCMV5 was immunoprecipitated a s described above. The pellets were suspended in 1 ml of Hepes buffer (20 mM Hepes, pH 7.4,lOO mM NaCl, and 1 mM diothiothreitol) with 1 m~ MgC1, and 0.1 m~ ZnCl,, divided in six tubes, centrifuged, and excess fluid carefully aspirated from the pellets. To dephosphorylate dSosHA, the pellets were incubated with 25 units of calf intestine alkaline phosphatase (Sigma) for 30 min a t room temperature. The samples were then washed three times with 1 ml of Hepes buffer containing 10 m~ nitrophenyl phosphate and 10 nm MgCl,. The samples were then incubated with purified human p41 isoform of MAP kinase (kindly provided by Roger J. Davis) and 250 pCi of [y-32PlATP (3000 Ci/mmol) for a n additional 30 min at room temperature. To stop the reactions, the samples were washed once with 1 ml of Hepes buffer and SDS-PAGE sample buffer was added.
To test the efficacy of dephosphorylation, two control and two transfected plates were labeled for 4 h with 2 mCi of 32Pi, immunoprecipitated, and treated with phosphatase as described above.

RESULTS
The consensus sequence for phosphorylation by MAP kinase has been determined to be Pro-X,-Ser/Thr-Pro where X is a neutral or basic amino acid and n = 1 or 2 (Clark-Lewis et al., 1991;Gonzalez et al., 1991). There are seven potential MAP kinase phosphorylation sites in the primary sequence of the dSos protein. To test whether dSos protein is phosphorylated by MAP kinase, dSos was partially purified by immunoprecipitation and used as a substrate for purified MAP kinase in vitro. In order to immunoprecipitate dSos protein, construct PCMV5-dSosHA was engineered by fusing dSos cDNA to sequences encoding the major influenza virus hemagglutinin antigenic epitope. This was subcloned into mammalian expression vector PCMVS (Anderson et al., 1989) to create construct PCMV5-dSosHA. COS-1 cells were transfected with PCMVS-dSosHA, and dSosHA protein was immunoprecipitated with an anti-HA monoclonal antibody. The immunoprecipitates were incubated with purified MAP kinase and [Y-~~PIATP, and the reaction mixture was analyzed by SDS-PAGE. Fig. lA shows that a prominent 180-kDa band is labeled in this reaction. This band is not present in immunoprecipitates from cells transfected with PCMVS alone, nor is it present in dSosHA immunoprecipitates incubated only with [y3'P1ATP. These data strongly suggest that the 180-kDa band phosphorylated by MAP kinase is dSosHA.
Since some sites in proteins isolated from cells may already be phosphorylated, the precipitated dSosHA was treated in some experiments with alkaline phosphatase prior to phosphorylation with MAP kinase. As seen in the last two lanes of Fig. lA, treatment of immunoprecipitates with phosphatase prior to the kinase reaction resulted in an increased phosphorylation of dSosHA by MAP kinase. The efficacy of the phosphatase treatment was verified by examining its effect on dSosHA immunoprecipitated from cells that had been labeled with 32Pi (Fig. 1B). Phosphatase treatment completely removed the radioactivity that had been incorporated into dSosHA from intact 32Pi-labeled cells.
Phosphoamino acid analysis was conducted on the 180-kDa band phosphorylated by MAP kinase. As seen in Fig. lC cubation with MAP kinase slightly increased the overall amount of serine phosphorylation incorporated into dSosHA.
To further characterize dSosHA phosphorylation sites, the 180-kDa protein phosphorylated by MAP kinase in immunoprecipitates was isolated and digested with trypsin. Tryptic peptides were first separated by electrophoresis and then by chromatography in the second dimension. Two-dimensional maps of tryptic peptides isolated from dSosHA with and without treatment of phosphatase prior phosphorylation are shown in Fig. 2 (A and B). Comparison of peptides derived from both conditions revealed several species appearing a t identical positions (Fig. 2, A and B). However, some differences in the pattern of peptides are evident; in particular, peptide 1 (Fig.  2 B ) was absent in tryptic maps prepared from dSosHA that were not treated with phosphatase ( Fig. 2A). This species probably represents a site that is completely phosphorylated in intact cells, and so no further phosphorylation can occur when incubated with MAP kinase. The identities of species were confirmed in mixing experiments in which peptides obtained from phosphatase treated and untreated dSosHA were applied to the same plate and analyzed by two-dimensional separation (data not shown).
The phosphorylation pattern of dSosHA in intact cells was also examined by labeling PCMV5-dSosHA-transfected COS-1 cells with 32Pi and isolating dSosHA by immunoprecipitation. Tryptic maps exhibited four major phosphorylated species and several minor species. Two of the major species (peptides 1 and 2 in Fig. 2C) appeared at positions similar to two phosphopeptides that were seen after MAP kinase phosphorylation of phosphatase-treated dSosHA. To further substantiate this, equal amounts of radioactivity from phosphopeptides obtained from both in vitro and intact cell labelings were mixed and spotted on a thin-layer plate. As seen in Fig. 2 0 , this analysis revealed that the migrations of peptides 1 and 2 obtained by both sources are identical, which strongly suggests that they represent the same phosphorylation sites.
The phosphoamino acid content of the more prominent phosphopeptide species from 32P-labeled dSosHA was determined. Phosphopeptides were eluted from thin-layer plates, partially hydrolyzed with HCl, and analyzed by thin-layer electrophoresis. As illustrated in Fig. 3 A , peptides 1 and 2 ?. + FIG. 2. Tryptic phosphopeptide map of dSosHk A and B, COS-1 cells were transfected with PCMV5-dSosHA and the expressed protein was immunoprecipitated. In B, the precipitated dSosHA was pretreated with alkaline phosphatase, and in both A and B, immunoprecipitates were then phosphorylated with MAP kinase. Immunoprecipitates were separated by SDS-PAGE, and dSosHA was treated with trypsin overnight. The released peptides were separated by electrophoresis (horizontal) and chromatography (vertical), and the plates were subsequently subjected to autoradiography. The sites of application are indicated by the crosses. C , transfected COS-I cells were labeled with "Pi for 4 h, and the labeled dSosHA was immunoprecipitated and analyzed by tryptic mapping. D, 982 cpm of the material shown in B and 1041 cpm of the material shown in C were mixed and separated as before. The two species (indicated by 1 and 2 ) exhibit identical migration whether they are derived from dSos labeled in intact cells or from dSos phosphorylated by MAP kinase in immunoprecipitates. MAP kinase in immunoprecipitates, contained phosphoserine but no phosphothreonine. The slower migrating labeled species represent partial hydrolysis products. As expected, this pattern generally differs for each phosphorylation site in a given protein (Cooper et al., 1983;Martensen, 1984). Interestingly, the patterns of partial hydrolysis products determined for species 1 from both methods of labeling are identical. This is also true for species 2. These data confirm that phosphopeptide species 1 and 2 derived from dSosHA labeled in intact cells represent the same sites that are phosphorylated by purified MAP kinase in dSosHA immunoprecipitates.
One prominent labeled phosphopeptide species labeled in immunoprecipitates contained only threonine (Fig. 3, B and D ) .
Since only one consensus site for MAP kinase phosphorylation in dSosHA contains threonine (Thr-1481), it is likely that this species represents that site. This species did not appear in material derived from dSosHA labeled in intact cells, and its labeling did not require dephosphorylation prior to phosphorylation by MAP kinase. This site, therefore, does not seem to be phosphorylated in intact cells. A summary of the phosphoamino acid content of these and other labeled peptides is given in Fig.   3B.

DISCUSSION
The Son of sevenless proteins are believed to be involved in the activation of Ras by tyrosine kinase receptors (Baltensperger et al., 1993;Buday and Downward, 1993;Li et al., 1993;Rozakis-Adcock et al., 1993;Skolnik et al., 1993). In this communication, we have shown that the Drosophila Son of sevenless protein is a substrate for MAP kinase in vitro. Some of the same sites that are phosphorylated by MAP kinase in vitro are apparently also phosphorylated in intact cells. This is unambiguously demonstrated for phosphopeptide 1 (Fig. 2) because phosphorylation of this peptide in vitro only occurs after dephosphorylation of in vivo phosphorylated dSos. MAP kinase has been shown to be part of a large number of signaling pathways and has numerous targets located in different cellular compartments (Blenis, 1993;Davis, 1993). Proteins such as c-Myc (Seth et al., 1992), p9OrSk (Sturgill et al., 1988), and cytoplasm phospholipase Az (Lih-Ling et al., 1993) are targets of MAP kinase phosphorylation that are downstream in its signal transduction pathway. However, as in the case of Sos, MAP kinase also phosphorylates proteins upstream from it such as MAP kinase kinase (Matsuda et al., 1993), RAF (Anderson et al., 1991;Lee et al., 19911, and EGF receptor (Northwood et al., 1991;Takishima et al., 1991). The functions of MAP kinase-mediated phosphorylations of upstream targets have yet to be elucidated. However, in S. cerevisiae there is some evidence to suggest that a MAP kinase homologue, FUSS, down-regulates its own phosphorylation pathway (Gartner et al., 1992). Thus, it is possible that phosphorylations or dephosphorylations could be involved in either negative or positive feedback mechanisms.
It has been suggested that Sos is phosphorylated on serine and threonine residues in response to EGF or insulin stimulation (Li et al., 1993;Rozakis-Adcock et al., 1993;Skolnik et al., 1993). We have similarly observed a decreased electrophoretic mobility of Sos-1 in 3T3-Ll adipocytes stimulated with insulin or phorbol ester.' In the COS-1 cell system, we have not been able to identify an increase of Sos phosphorylation in response to insulin (Baltensperger, et al., 1993). COS-1 cells have been observed to have high basal levels of MAP kinase activity, and transfection of insulin receptor cDNA into COS-1 cells and subsequent stimulation of insulin leads to no detectable increase of MAP kinase activity in cell extract^.^ Thus, it may not be possible to observe Sos phosphorylation changes in response to insulin in transiently transfected COS-1 cells.
The effect of phosphorylation on Sos function is not known. It is possible that phosphorylation directly effects Sos catalytic activity toward Ras. Wolfman and Macara (1990) report that purification of guanosine nucleotide exchange activity from brain requires the presence of phosphatase inhibitors. However, since brain tissue contains both the Sos and GRF exchange factors, it is not certain which Ras nucleotide exchange activity was measured. The Ral guanosine nucleotide exchange factor ral-GDS is phosphorylated on serines, but treatment of ralGDS with phosphatase does not effect its activity (Albright et al., 1993). Nevertheless, it should be noted that unlike Sos, neither GRF nor ralGDS contain long COOH-terminal proline-rich extensions where the potential MAP kinase sites are located.
Phosphorylation may also affect the association between Sos and GRB2 or other unidentified proteins. In this case, phosphorylation may change the conformation of Sos so that it changes its affinity for binding proteins. A change in phosphorylation state could then affect the intracellular location of Sos, which would then alter its ability to activate Ras. A similar mechanism may exist with CDC25 regulation in S. cerevisiae. The CDC25 becomes hyperphosphorylated, in response to glucose starvation. This hyperphosphorylation occurs concomitantly with a partial relocalization of CDC 25 to the cytoplasm, which reduces its accessibility to membrane-bound Ras. In this case it has been suggested that a downstream element, CAMPdependent protein kinase, down-regulates CDC25 by phosphorylation (Gross et al., 1992). Further studies should determine whether MAP kinase effects Sos regulation in a similar manner.