Pro-Leu-Ser/Thr-Pro Is a Consensus Primary Sequence for Substrate Protein Phosphorylation CHARACTERIZATION OF THE PHOSPHORYLATION OF c-myc AND c-jun PROTEINS BY AN EPIDERMAL GROWTH FACTOR RECEPTOR THREONINE 669 PROTEIN KINASE*

A growth factor-stimulated (MAP2-related) protein kinase, ERT, that phosphorylates the epidermal growth factor receptor at Threes has been purified from KB human tumor cells by Northwood and co-workers (Northwood, I. C., Gonzalez, F. A., Wartmann, M., Raden, D. L., and Davis, R. J. (1991) J. Biol. Chern. 266, 15266-15276). The ERT protein kinase has a restricted substrate specificity, and the structural determinants employed for substrate recognition by this enzyme have not been defined. As an approach toward understanding the specificity of substrate phosphoryl- ation, we have used an in vitro assay to identify additional substrates for the ERT protein kinase. In this report we describe two novel substrates: (a) the human c-myc protein at Sere' and (b) the rat c-jun protein at SerZ4'. Alignment of the primary sequences surrounding the phosphorylation sites located within the epi- dermal growth factor receptor (Threes), extracts were prepared using the method of Lee et al. (7). Protein was determined using the Bradford dye binding assay (Bio-Rad) using bovine serum albumin as standard. Protein kinase assays were performed using a procedure that we have previously described (4). The synthetic peptide substrate used was Lys-Arg-Glu-Leu-Val-Glu-Pro-Leu-Thr669-Pro-Ser-Gly-Glu-Ala-Pro-Asn-Gln-Ala-Leu-Leu-Arg. The as- says were performed using 25 mM HEPES (pH 7.4), 10 mM MgC12, 50 p~ [y-32P]ATP (10 pCi/nmol), 1 mg/ml synthetic peptide in a final volume of 25 j11. The reactions were terminated after 20 min at 22 "C by the addition of 10 pl of 90% formic acid containing 50 mM ATP. Control experiments demonstrated that the phosphorylation reaction was linear with time for 30 min. The phosphorylated syn- thetic peptide was isolated by applying 25 pl of the reaction mixture onto phosphocellulose paper (PSI, Whatman) and washing the filters twice in 1 M acetic acid, 4 mM sodium pyrophosphate. Radioactivity was quantitated by measuring the Cerenkov radiation with a Beck- man liquid scintillation counter. Nonspecific incorporation of radioactivity was determined in incubations without the synthetic peptide.

266, 15266-15276). The ERT protein kinase has a restricted substrate specificity, and the structural determinants employed for substrate recognition by this enzyme have not been defined. As an approach toward understanding the specificity of substrate phosphorylation, we have used an in vitro assay to identify additional substrates for the ERT protein kinase. In this report we describe two novel substrates: (a) the human c-myc protein at Sere' and ( b ) the rat c-jun protein at SerZ4'. Alignment of the primary sequences surrounding the phosphorylation sites located within the epidermal growth factor receptor (Threes), Myc (Ser6'), and Jun (Ser246) demonstrated a marked similarity. The observed consensus sequence was Pro-Leu-Ser/ Thr-Pro. We propose that this sequence forms part of a substrate structure that is recognized by the ERT protein kinase.
Treatment of cultured cells with epidermal growth factor (EGF),' platelet-derived growth factor, phorbol ester, or serum causes a marked increase in the state of phosphorylation of the EGF receptor at T h P 9 (1-3). The increase in phosphorylation is caused by the stimulation of a protein kinase activity that can be detected in cell extracts (2). This *These studies were supported in part by Grants CA39240 and GM37845 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisenent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I The abbreviations used are: EGF, epidermal growth factor; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; ERT protein kinase, protein kinase that phosphorylates the EGF receptor a t Thr'"; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid, Jun, the protein product of the c-jun gene; MAP2 protein kinase, growth factor-stimulated protein kinase that phosphorylates microtubule-associated protein 2; IvEyc, the protein product of the c-myc gene.
protein kinase activity is accounted for by two distinct enzymes: 1) the MAP2 protein kinase and 2) a novel serine/ threonine protein kinase that we designate ERT (EGF _Receptor Thrfi69) protein kinase (4).
Recently, the ERT protein kinase was purified from KB human tumor cells (4). In vitro phosphorylation assays demonstrated that the purified ERT protein kinase has a very restricted substrate specificity (4). Indeed, the EGF receptor is the only substrate for the ERT protein kinase that has been characterized in detail (2-5).2 Phosphorylation of the EGF receptor at Thrfifig regulates both tyrosine phosphorylation and receptor endocytosis (5). This phosphorylation of the EGF receptor may therefore be physiologically significant, but it is likely that there are additional substrates for the ERT protein kinase. This is because growth factor-stimulated ERT protein kinase activity can be detected in cells that do not express EGF receptors (2, 6). The marked extent and rapid kinetics of activation of the ERT protein kinase by growth factors suggests that this enzyme may have an important function during signal transduction (2, 6). Regulatory proteins within signal transduction pathways therefore represent potential substrates for the ERT protein kinase.
The purpose of the experiments reported here was to examine the recognition of substrate proteins by the ERT protein kinase. The experimental strategy that we employed was to test potential substrate proteins in an in vitro phosphorylation assay using the purified ERT protein kinase. We report here the identification of two protein substrates: the products of the proto-oncogenes c-myc and c-jun. Alignment of the primary sequences surrounding the phosphorylation sites located within the EGF receptor, Jun, and Myc indicates a consensus sequence Pro-Leu-Ser/Thr-Pro. We propose that this sequence forms part of a substrate structure that is recognized by the ERT protein kinase.

EXPERIMENTAL PROCEDURES
Muteriak-[["'S]Methionine and [3sS]dATP were obtained from Amersham Corp. ["P]Phosphate was from Du Pont-New England Nuclear. [y-"PIATP was prepared using a Gamma-Prep A kit (Promega Biotech) according to the manufacturer's directions. The ERT protein kinase was purified from KB cells as described previously (4). Nuclear extracts were prepared from KB cells as described (7). Taq polymerase was from Perkin-Elmer Cetus. Synthetic peptides were obtained from the Peptide Synthesis Core Facility (University of 'The EGF receptor phosphorylation site ThrGfi9 is conserved in c-erbB2 (38). It is therefore likely that the c-erbB2 gene product is also a substrate for the ERT protein kinase.
Massachusetts Medical School). Restriction enzymes were from Boehringer Mannheim.
Plasmid Construction and Purification of Bacterially Expressed Proteins-Full-length Jun (Jun(1-334)) and a truncated Jun polypeptide containing amino acids 224-334 (Jun(224-334)) were expressed in Escherichia coli as hexahistidine fusion proteins and purified by nickel affinity chromatography as described (8,9). Deletion of the coding sequence corresponding to amino acids 245-247 in Jun(224-334) was achieved using a polymerase chain reaction strategy (IO). The sequence of the mutated gene was confirmed by dideoxynucleotide sequencing (11).
Glutathione S-transferase (GST) fusion proteins were constructed in the vector pGEX-3X (Pharmacia LKB Biotechnology Inc). The coding region of the human c-myc gene corresponding to exon 2 (amino acids 2-251) was isolated from a genomic clone (American Type Culture Collection 41010) using the polymerase chain reaction and the oligonucleotide primers, 5' gcg agg atc ccc ctc aac gtt agc ttc acc aac agg aac 3' and 5' ggg gaa ttc gct cga ggt ggt ggg cgg tgt ctc ctc atg gag cac cag 3'. The amplified DNA was cloned as a BamHI-EcoRI fragment into the vector pGEX-3X. A point mutation was introduced at codon 62 using a polymerase chain reaction strategy (IO), and the sequence was confirmed using the dideoxynucleotide method and synthetic oligonucleotide primers (11). The fusion protein was isolated from bacteria induced for 2 h with 0.4 mM isopropylthiogalactoside by chromatography using glutathione-agarose as described (12). These fusion proteins were designated GST/Myc and GST/ [Alafi2]Myc.
GAL4(1-147) fusion proteins were prepared using the vector pSG424 (13). A DNA fragment of the human c-pyc gene corresponding to amino acids 2-103 was isolated using the polymerase chain reaction and the oligonucleotide primers, 5' gcg agg atc ccc ctc aac gtt agc ttc acc aac agg aac 3' and 5' cag cat cta gat cac cat ctc cag ctg 3'. The amplified DNA was cloned as a blunt-end fragment into pUC13 at the SmaI site. Point mutations were introduced at codons 62, 64, and 67 using a polymerase chain reaction strategy (IO). The sequence of the amplified DNA was confirmed using the dideoxynucleotide method and synthetic oligonucleotide primers (11). The c-myc fragments were then excised from pUC13 by restriction endonuclease digestion and cloned as BamHI-XbaI fragments into the polylinker of the vector pSG424. The fusion proteins were transiently expressed in COS-1 cells and were designated GAL4/Myc, GAL4/ [Ala'j2]Myc, GAL4/[AlaG4]Myc, and GAL4/[Ala67]Myc.
Transient Expression of Fusion Proteins in COS-1 Cells-COS-I cells in 100-mm dishes were maintained in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. Plasmid DNA (10 pg) was transfected using the DEAE-dextran method (14). After 30 h of incubation the culture medium was aspirated and replaced with 5 ml of labeling medium supplemented with 1% fetal calf serum: (a) modified Eagle's medium containing 50 p M [35S] methionine (20 pCi/ml) or ( b ) phosphate-free modified Eagle's medium containing 0.4 mCi/ml [32P]phosphate. After 18 h of additional incubation, the cells were lysed in 25 mM HEPES (pH 7.5), 1% Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 5 mM EGTA, 50 mM NaF, 500 mM NaCl, 100 p M Na3V0,, 10 pg/ml leupeptin, and 1 mM phenylmethysulfonyl fluoride. The lysate was clarified by centrifugation at 100,000 X g, and the supernatant was incubated for 60 min at 22 "C with 5 pl of rabbit anti-GAL4 antiserum prebound to 20 p1 of protein A-Sepharose. The immunoprecipitates were washed three times with lysis buffer, washed once with 25 mM HEPES (pH 7.5), 0.1 mM EGTA, and analyzed by polyacrylamide gel electrophoresis.
Phosphorylation of Synthetic Peptides Based on the Primary Sequence of Jun and Myc by the Purified ERT Protein Kinase-Synthetic peptides were phosphorylated at 22 "C in a buffer containing 25 mM HEPES (pH 7.4), 10 mM MgCl,, 1 mg/ml synthetic peptide, purified ERT protein kinase (-5 ng), and 40 p~ [y-32P]ATP (40 pCi/ nmol) in a final volume of 25 p1. The phosphorylation reaction was terminated after 20 min by the addition of 5 pl of 90% formic acid. The phosphorylated synthetic peptide was isolated by electrophoresis (4 "C) for 3 h at 500 V on a 100-pm cellulose thin layer plate using 30% (v/v) formic acid as solvent. The phosphorylated peptide was located by autoradiography and was isolated from the plate by extraction with 30% formic acid and lyophilized.
Phosphorylation of J u n and Myc Proteins by the ERT Protein Kinase-Jun and Myc proteins (200 ng) were phosphorylated at 22 "C in a buffer containing 25 mM HEPES (pH 7.4), 10 mM MgCL, purified ERT protein kinase (-5 ng), and 50 pM [ T -~P I A T P (10 pCi/nmol) in a final volume of 25 pl. The phosphorylation reaction was termi-nated after 20 min by the addition of 100 pl of Laemmli sample buffer. Phosphorylated proteins were analyzed by polyacrylamide gel electrophoresis (15) and autoradiography.
Phosphoamino Acid Analysis-Phosphoamino acid analysis was performed by partial acid hydrolysis (1 h at 110 "C in 6 M HCl) and thin layer electrophoresis as described (16,17).
Tryptic [32PlPhosphopeptide Mapping-Phosphorylated proteins were eluted from gel slices using sodium dodecyl sulfate, precipitated with trichloroacetic acid, and oxidized with performic acid as described (18). The sample was digested with 1 pg of tosylphenylalanyl chloromethyl ketone-treated trypsin in 100 mM N-ethylmorpholine (pH 8.01, After 5 h, a second addition of trypsin was made, and the incubation was allowed to proceed for an additional 19 h. (In some experiments 100 pg of trypsin was used to confirm the identity of limit tryptic phosphopeptides.) Phosphopeptide mapping was performed by two-dimensional separation on 100-pm cellulose thin layer plates as described (19). The first dimension was electrophoresis in 30% formic acid at 500 V for 2 h at 4 "C. The second dimension was chromatography in butan-1-ol/pyridine/acetic acid/water (15:103:12). Electrophoresis was performed in the horizontal dimension (cathode at right side) and the chromatography was performed in the vertical dimension. The origin is located in the lower left corner of each panel (see figures).
Measurement of ERT Protein Kinase Activity-KB cells were obtained from the American Type Culture Collection and were maintained in Dulbecco's modified Eagle's medium supplemented with 5% calf serum. The KB cells were seeded in 100-mm dishes and were grown to confluence. The cells were washed with serum-free medium and were incubated for 30 min at 37 "C. After treatment for 5 min with 10 nM EGF, the cells were harvested. Cytosolic extracts were prepared by the method of Northwood and Davis (6). Nuclear extracts were prepared using the method of Lee et al. (7). Protein was determined using the Bradford dye binding assay (Bio-Rad) using bovine serum albumin as standard. Protein kinase assays were performed using a procedure that we have previously described (4). The synthetic peptide substrate used was Lys-Arg-Glu-Leu-Val-Glu-Pro-Leu-Thr669-Pro-Ser-Gly-Glu-Ala-Pro-Asn-Gln-Ala-Leu-Leu-Arg. The assays were performed using 25 mM HEPES (pH 7.4), 10 mM MgC12, 50 p~ [y-32P]ATP (10 pCi/nmol), 1 mg/ml synthetic peptide in a final volume of 25 j11. The reactions were terminated after 20 min at 22 "C by the addition of 10 pl of 90% formic acid containing 50 mM ATP. Control experiments demonstrated that the phosphorylation reaction was linear with time for 30 min. The phosphorylated synthetic peptide was isolated by applying 25 pl of the reaction mixture onto phosphocellulose paper (PSI, Whatman) and washing the filters twice in 1 M acetic acid, 4 mM sodium pyrophosphate. Radioactivity was quantitated by measuring the Cerenkov radiation with a Beckman liquid scintillation counter. Nonspecific incorporation of radioactivity was determined in incubations without the synthetic peptide.

RESULTS
A mitogen-stimulated protein kinase (ERT) that phosphorylates the EGF receptor at ThrG9 has been purified from KB human tumor cells (4). To identify substrates for this protein kinase we used an in vitro assay to examine the phosphorylation of purified proteins by the isolated protein kinase. During the initial screening of potential substrates two proteins were found to be markedly phosphorylated Myc and Jun. Further experiments were designed to provide a detailed characterization of the phosphorylation of these proteins.

Jun Is Phosphorylated at Ser246 by the ERT Protein Ki-
nase-To examine the phosphorylation of Jun by the ERT protein kinase we performed an in vitro assay using purified proteins. No protein phosphorylation was observed when the ERT protein kinase was incubated with [y-32P]ATP. The absence of phosphorylation is consistent with previous observations that the ERT protein kinase does not autophosphorylate and that the purified enzyme is not significantly contaminated with exogenous substrates (4). Addition of Jun to the incubation resulted in marked phosphorylation of the Jun protein ( Fig. l A , lune C ) . Phosphoamino acid analysis indicated that Jun was phosphorylated on serine residues ( In order to localize the site(s) of phosphorylation of Jun by the ERT protein kinase, we investigated the phosphorylation of Jun proteins containing NH2-terminal deletions. It was observed that the deletion of residues 1-223 did not block Jun phosphorylation (Fig. lA, lane D). Tryptic ['"Plphosphopeptide maps of the full-length and truncated Jun proteins were identical (Fig. 1C). Thus, the Jun phosphorylation site(s) are located in the COOH terminus of the protein between residues 224 and 334.
Visual inspection of the primary sequence of the COOH terminus of Jun indicated a region that was homologous to the sequence of the EGF receptor surrounding the phosphorylation site Thr'"''. The predicted Jun phosphorylation site was Ser"". The following two experiments were performed to test the hypothesis that Jun was phosphorylated at Ser'.'". ( a ) If Ser"'" is a site of Jun phosphorylation, deletion of this residue would be expected to markedly reduce the phosphorylation of Jun by the ERT protein kinase. It was observed that the deletion of residues 245-247 blocked the phosphorylation of Jun (Fig. L4, lune B ) . This observation is consistent with the hypothesis that Ser24'" is the major site of Jun phosphorylation. ( b ) A synthetic peptide based on the primary sequence of Jun was prepared EEPQTVPEMPGETPPLS"4'PPIDMES-QER. This peptide corresponds to the predicted Jun tryptic peptide that contains Ser""". It was observed that the peptide was a substrate for phosphorylation by the purified ERT protein kinase (data not shown). Phosphoamino acid analysis demonstrated the presence of [:"P]phosphoserine (data not shown). Analysis of the phosphorylated synthetic peptide by two-dimensional peptide mapping demonstrated the presence of three ["Plphosphopeptides (Fig. 2). These phosphopeptides co-migrated during electrophoresis and therefore do not correspond to forms of the synthetic peptide phosphorylated at different numbers of sites. However, the phosphopeptides were resolved by chromatography indicating that the peptides differ in hydrophobicity. In several experiments it was found that the relative yield of the three phosphopeptides was variable (data not shown). This observation suggested that the peptide may be degraded during sample preparation.
One possibility is that the extraction and lyophilization of the

Kinase Substrate Specificity
peptide in formic acid causes partial deamidation of the two glutamine residues present in the synthetic peptide. However, no direct evidence for this hypothesis was obtained.
Comparison of the peptide maps of the phosphorylated synthetic peptide (Fig. 2) with maps of Jun (Fig. 1) indicated a marked similarity. In each case three major phosphopeptides were observed that co-migrated during electrophoresis and were resolved by chromatography. This similarity suggested that the phosphopeptides present in these maps may be identical. To test this hypothesis we performed mixing experiments to examine whether phosphopeptides from the maps co-migrated during two-dimensional separation. In initial experiments it was found that the peptides did not co-migrate (Fig. 2C). However, on close inspection it was observed that the phosphorylated synthetic peptide did co-migrate with a set of three minor peptides present in the maps of Jun (Fig.  2). We therefore considered the possibility that the major phosphopeptides present in the maps of the Jun protein may represent the products of a partial tryptic digestion.:' The electrophoretic mobility of the peptides was consistent with this hypothesis. In order to characterize the limit tryptic ['"PI phosphopeptides we repeated the proteolytic digestion in the presence of a higher concentration of trypsin. It was observed that three limit tryptic phosphopeptides were obtained (Fig.  2 0 ) . These limit tryptic ['"Plphosphopeptides co-migrated with the [:"P]phosphorylated synthetic peptide (Fig. 2F). The observed co-migration suggests an identity between the synthetic peptide and the tryptic peptide containing the Jun phosphorylation site.
We conclude from the results of the deletion studies ( Fig.  1) and comparative ['"Plphosphopeptide mapping (Fig. 2) that Jun is phosphorylated a t Ser"l" by the ERT protein kinase.' Myc Is Phosphorylated ut Ser" by the ERT Protein Kinase-The phosphorylation of Myc by the purified ERT protein kinase was examined in an in vitro assay using a bacterially expressed GST/Myc fusion protein as a substrate.s No phosphorylation of glutathione S-transferase by the ERT protein kinase was observed (data not shown). However, the GST/Myc fusion protein was markedly phosphorylated (Fig.  3A). Phosphoamino acid analysis demonstrated the presence of phosphoserine, but no phosphothreonine or phosphotyrosine was detected (Fig. 3B). To further characterize the phosphorylation we performed phosphopeptide mapping after trypsin digestion of the phosphorylated Myc protein. Fig. 4 shows that two tryptic ["'Plphosphopeptides were resolved by two-dimensional separation on cellulose thin layer plates. Together, these data indicate that Myc contains a t least 1 serine residue that is phosphorylated by the ERT protein kinase.
:I The presence of partially digested peptides in tryptic phosphopeptide maps is not uncommon and has been reported previously. Two examples of proteolytic sites that are cleaved very slowly by trypsin are provided by studies of the phosphorylation of the EGF receptor (17, 39) and the transferrin receptor (19) by protein kinase C .
.'The site of phosphorylation of rat Jun reported here is Ser246. This site is equivalent to Ser'"' in human Jun (33,40). The primary sequence surrounding this phosphorylation site is identical between the rat and human sequences.
In initial experiments a glutathione S-transferase fusion protein was prepared using the full-length Myc protein. However, this protein was found to be insoluble and was not used for in vitro protein kinase assays. Glutathione S-transferase fusion proteins were therefore prepared containing the NH, or the COOH domains of Myc. The COOH domain fusion protein was not phosphorylated by the purified ERT protein kinase (E. Alvarez, unpublished observation). In contrast, the NH, domain fusion protein (GST/Myc) was found to be a substrate (Fig. 3A). T o identify the site(s) of phosphorylation of Myc, we compared the primary sequence of Myc with the sequence of the EGF receptor surrounding Thr"". This analysis suggested that Ser"' was a potential site of Myc phosphorylation. The hypothesis that Myc was phosphorylated a t Ser"' was tested using the following two experimental approaches. ( a ) Ser"' was substituted with an Ala residue, and the effect of this mutation on the phosphorylation of the GST/Myc fusion protein was investigated. Fig. 3A shows that the GST/[Ala"'] Myc fusion protein was not a substrate for phosphorylation by the ERT protein kinase. The lack of phosphorylation of the mutated protein suggests that the site of Myc phosphorylation was Ser"'. ( b ) A synthetic peptide corresponding to the primary sequence surrounding Myc Ser"' was prepared: KKFELLPTPPLSfi'PSRR. This peptide was found to be a substrate for phosphorylation by the ERT protein kinase (data not shown). Phosphoamino acid analysis demonstrated the presence of phosphoserine, but not phosphothreonine or phosphotyrosine (data not shown). A single ["'PJphosphopeptide with a high electrophoretic mobility was detected by twodimensional phosphopeptide mapping (data not shown). The synthetic peptide contains 4 lysine and arginine residues. We therefore digested the phosphorylated synthetic peptide with trypsin and examined the results of the proteolytic digestion by phosphopeptide mapping. Two tryptic phosphopeptides were obtained that were resolved by electrophoresis (Fig. 4). The two peptides were probably the result of partial proteolysis because of the poor exopeptidase activity of trypsin. 3 Mixing experiments demonstrated that the tryptic ["'PJphosphopeptides derived from the synthetic peptide co-migrated with the ['"PJphosphopeptides obtained by trypsin digestion of the GST/Myc fusion protein (Fig. 4). The observed comigration suggests an identity between the tryptic phosphopeptides obtained from the synthetic peptide and the Myc protein.
The results of point mutation (Fig. 3) and comparative phosphopeptide mapping (Fig. 4) demonstrate that S e P is a site of Myc phosphorylation by the ERT protein kinase.
Phosphorylation of Myc Ser6* in Intact Cells-The in vitro phosphorylation of Jun (Ser*") and Myc (Ser") by the ERT protein kinase suggests that these proteins may be phosphorylated in intact cells. However, the demonstration of in vitro phosphorylation does not necessarily imply that this phosphorylation will occur in vivo. It was therefore important to establish whether this phosphorylation could be found in intact cells. Recently, Seg4' -has been demonstrated to be an in vivo site of Jun phosphorylation (20); but little information about the in uivo phosphorylation of Myc is available (21).
We therefore investigated whether Myc Serfi2 was phosphorylated in intact cells.
The experimental approach that we employed was to construct a fusion protein between a fragment of the yeast transcription factor GAL4 (residues 1-147) and Myc. It has previously been documented that this fusion protein is correctly localized in the nucleus and is functional as a transcriptional activator (22). After expression in COS cells," the GAL4/Myc fusion protein was markedly phosphorylated (Fig.  5). In contrast, GAL4(1-147) was not detectably phosphorylated (Fig. 6B). Thus, the observed phosphorylation is specific to the Myc fusion protein. To investigate whether the GAL4/ Myc fusion protein was a substrate for an EGF-stimulated protein kinase we investigated the effect of EGF on the phosphorylation state of the GAL4/Myc fusion protein. Fig.  5 shows that EGF treatment caused a rapid increase in GAL4/ Myc phosphorylation. Phosphoamino acid analysis demonstrated the presence of [:"PJphosphoserine and some ["' PI phosphothreonine (Fig. 6C). The Myc protein was therefore phosphorylated a t more than one site.
Phosphopeptide mapping of the Myc fusion protein isolated ' Control experiments were performed to investigate whether COS cells express the E R T protein kinase. COS cells were treated without and with 10 nM EGF for 5 min at 37 "C. The cells were lysed, and the ERT protein kinase activity was measured as described previously from EGF-treated cells indicated the presence of two major tryptic ["PJphosphopeptides and several minor ["'PJphosphopeptides (Fig. 6D). The two major [""PJphosphopeptides comigrated with the two ["*P]phosphopeptides derived by tryptic digestion of the synthetic peptide KKFELLPTPP-LS"PSRR phosphorylated in vitro by the ERT protein kinase (data not shown). The phosphorylation site located in the synthetic peptide corresponds to Ser"' (Figs. 3 and 4). These data indicate that Serfi* is the major site of phosphorylation of the Myc fusion protein in COS-1 cells.
To confirm that Myc Ser" was phosphorylated in intact cells we investigated the effect of point mutations on the phosphorylation state of the Myc fusion protein. It was observed that the substitution of Serfi4 with Ala caused no significant change in the level of phosphorylation (Fig. 6 B ) or the ["'P]phosphopeptide map obtained after tryptic digestion (Fig. 6D). Similar results were obtained after the replacement of SerG5 with Ala (data not shown). However, replacement of Serfi2 with Ala caused a decrease in the level of phosphorylation of the Myc fusion protein. Tryptic peptide mapping demonstrated that the two major ["P]phosphopeptides observed in maps of the wild-type protein were absent in maps of the mutant [AlafizJMyc fusion protein (Fig. 6D). These data strongly support the hypothesis that Ser"' is phosphorylated in intact cells.
Nuclear Localization of ERT Protein Kinase Activity-The identification of Jun and Myc as in vitro substrates for the ERT protein kinase suggests that these proteins may be substrates for the ERT protein kinase in situ. One important test of this hypothesis is that the ERT protein kinase should be co-localized with its substrates. Jun and Myc are translated on cytoplasmic ribosomes and are subsequently transported into the nucleus. It is therefore possible that the cytosolic ERT protein kinase (4) may phosphorylate Jun and Myc prior to the entry of these proteins into the nucleus. Alternatively, it is possible that the ERT protein kinase may be located in both the nuclear and cytoplasmic compartments of the cell. As an initial approach to address this question we investigated the specific activity of the ERT protein kinase in cytosolic and nuclear extracts isolated from EGF-treated KB cells. The specific activity of the ERT protein kinase was determined to be 5 f 2 fmol/min/mg and 4 f 1 fmol/min/mg in cytosolic and nuclear extracts, respectively (mean f S.D., n = 3). The high specific activity of the ERT protein kinase in nuclear extracts compared with cytosolic extracts suggests that this enzyme may be located in the nucleus as well as the cytoplasm. However, two caveats must be applied to this conclusion. 1) An endogenous inhibitor of the ERT protein kinase (4) makes activity measurements an unreliable method for quan-

Substrate Specificity
titating the level of enzyme expression. 2) There may be significant cross-contamination between the isolated cytosolic and nuclear fractions. A rigorous analysis of the subcellular localization of the ERT protein kinase will require the preparation of specific immunological reagents. These studies are currently in progress in this laboratory.

DISCUSSION
The purified ERT protein kinase has a restricted substrate specificity (4). The structural determinants that are employed by this enzyme for substrate recognition have not been defined. For many protein kinases it has been established that a critical factor for the specificity of phosphorylation is the primary sequence of substrate proteins (23). We therefore examined the primary sequence of the EGF receptor surrounding the phosphorylation site, Thr""". The distinctive feature of this phosphorylation site is the proximity of two proline residues: -Val-Glu-e-Leu-Thr6'9-~-Ser-Gly-. However, no rigorous analysis of this sequence can be achieved by visual inspection. Thus, an alternative approach was required to identify the amino acid residues that may be important for substrate recognition by the ERT protein kinase.
One strategy that has previously been successfully used to identify critical amino acid residues is the comparison of the sequences of proteins that have similar functions. The potential application of this approach to the ERT protein kinase is limited, because the EGF receptor is the only substrate that has been characterized (1-6). A major goal for the study described here was therefore the identification of additional protein substrates. We report here that the protein products of the proto-oncogenes c-jun and c-myc are substrates for phosphorylation by the ERT protein kinase.
Phosphorylation of J u n by the ERT Protein Kinase-The site of in vitro phosphorylation of Jun by the ERT protein kinase was identified as Ser"' ( Figs. 1 and 2). This residue has recently been demonstrated to be phosphorylated in vivo (20): The conservation of Ser246 between members of the Jun family (Fig. 7) suggests that this residue may be significant for Jun function (24). This hypothesis is supported by the observation of a correlation between phosphorylation a t this site and an inhibition of Jun DNA binding activity (20). Thus, the phosphorylation of Jun at Sern4'; may negatively regulate Jun function. It is therefore interestingthat the viral oncogene v-jun contains a point mutation at this phosphorylation site (substitution of Ser with Phe; Fig. 7) and therefore lacks this regulatory pathway that may suppress Jun function (24,25).  (48,49). Agap was introduced to allow the optimal alignment of the sequences. The asterisk indicates the serine residue in the rat c-jun protein (Ser''"') that is phosphorylated by the ERT protein kinase. accounted for by the ERT protein kinase (or by another MAP2-related protein kinase). A possible role for glycogen synthase kinase-3 has also been proposed (20). Further work will be required to rigorously document the identity of the physiologically relevant Jun protein kinase.

The phosphorylation of Jun in intact cells a t Ser246 may be
Phosphorylation of Myc by the ERT Protein Kinase-The site of in vitro phosphorylation of Myc by the ERT protein kinase was identified as Ser6' (Figs. 3 and 4). It is not known if Ser6' represents a site of phosphorylation of Myc in vivo.
To address this question we constructed a GAL4/Myc fusion protein that has previously been demonstrated to be functional as a transcriptional activator (22). It was observed that Ser"' was the major site of phosphorylation of the GAL4/Myc fusion protein in COS-1 cells (Fig. 5 ) . This phosphorylation was blocked by substitution of Ser6' with an alanine residue (Fig. 5). These data strongly implicate Ser6' as a site of Myc phosphorylation i n vivo. It is possible that the in vivo phosphorylation of Ser"' may be accounted for by the ERT protein kinase, but a role for other protein kinases (e.g. the MAP2 protein kinase) cannot be excluded by the data reported here.
The physiological significance of the phosphorylation of Myc at Ser"' is not understood. The phosphorylation site is located in an amino-terminal region of the c-myc protein that is highly conserved between members of the Myc family (Fig. S).7 Deletion studies have established that this aminoterminal region of Myc is required for neoplastic transformation (26) and is functional as a transcriptional activator (22). It is therefore possible that phosphorylation of Ser"' represents a mechanism of regulation of Myc function. Additional work will be required to test this hypothesis.

Identification of a Consensus Primary Sequence for Substrate Phosphorylation by the E R T Protein Kime-Align-
' Ser"' in the human c-myc protein is conserved between members of the Myc family (Fig. 8). However, the L-myc and S-myc proteins contain single amino acid differences within the consensus sequence Pro-Leu-Ser-Pro. The sequence of the L-myc and S-myc proteins are Arg-Leu-Ser-Pro and Pro-Thr-Ser-Pro, respectively. Further work will be required to determine whether the L-myc and S-myc proteins are substrates for phosphorylation by the ERT protein kinase. ment of the primary sequences surrounding the phosphorylation sites located within Myc, Jun, and the EGF receptor demonstrated a marked similarity. The observed consensus sequence was Pro-Leu-Ser/Thr-Pro (Fig. 9). This sequence probably forms part of a substrate structure that is recognized by the ERT protein kinase. It is likely that a three-dimensional substrate structure is recognized rather than simply the primary sequence Pro-Leu-Ser/Thr-Pro. Thus, sequences outside the consensus are probably required for efficient substrate phosphorylation. However, the identity of these residues is unclear. This is because no marked sequence similarity among the EGF receptor, Myc, and Jun was observed in the region flanking the consensus primary sequence (Fig. 9): It should be noted that the consensus sequence Pro-Leu-Ser/Thr-Pro may represent only a single class of substrates for the ERT protein kinase. Consequently, the data reported here do not exclude the possibility that the ERT protein kinase may phosphorylate substrates that do not conform to this consensus. Current studies in this laboratory are focused on the examination of the effects of point mutations in the consensus sequence Pro-Leu-Ser/Thr-Pro on substrate phosphorylation. The results of a systematic analysis of mutations within the consensus sequence will allow a more precise definition of substrate recognition by the ERT protein kinase.
As the ERT protein kinase is a MAP2-related protein kinase (4), it is likely that the consensus primary sequence for phosphorylation (Pro-Leu-Ser/Thr-Pro) may also be relevant to the function of the MAP2 protein kinase.
Identification of Additional Substrates for the .ERT Protein Kinase-The identification of a consensus primary sequence for substrate phosphorylation has the important implication that it provides a rational basis for the prediction of potential substrates for the ERT protein kinase. A search of the National Biomedical Research Foundation Protein Identification Resource data base demonstrated that the Pro-Leu-Ser/ Thr-Pro motif was found in a number of protein sequences. Inspection of the results of the data base search revealed two unexpected findings.
1) The consensus sequence is most frequently found in nuclear proteins. If the phosphorylation of nuclear proteins is physiologically significant it is likely that the ERT protein kinase is present in the nucleus. The results of previous studies have suggested that the protein kinase is located in the cytosol because it is a soluble protein that phosphorylates an integral plasma membrane protein, the EGF receptor (2, 4,6). However, it is possible that the ERT protein kinase may be located in both the nuclear and cytosolic compartments of the cell. This hypothesis is consistent with the results of cell fractionation studies that indicate a high level of protein kinase activity in nuclear extracts. However, a rigorous analysis of the subcellular localization of the ERT protein kinase will require the use of antibody probes. Currently, these studies are limited because antibodies to the ERT protein kinase are not available.
' One similarity between the sequences flanking the Pro-Leu-Ser/ Thr-Pro consensus located in the EGF receptor, Myc, and Jun is that a glutamic acid residue is located close to the NH, terminus of the consensus sequence (Fig. 9). It is possible that this residue is important for substrate recognition, but the location of this residue is not highly conserved. Further work will be required to establish the significance of this observation.
2) Some of the potential substrates exhibit ouerlapping specificity with the cdc2 protein kinase. One example of overlapping substrate specificity is the Myc phosphorylation site Ser6* (Fig. 8). The sequence surrounding this site is Pro-Leu-Ser"-Pro-Ser-Arg and corresponds to the consensus sequence for phosphorylation by the ERT protein kinase (Pro-Leu-Ser-Pro, Fig. 9) and the cdc2 protein kinase (Ser-Pro-X-Arg/ Lys, Ref. 34). This observation is interesting because these protein kinases are active at different stages of the cell cycle. The cdc2 protein kinase is active during mitosis (34), but the E R T protein kinase is rapidly and transiently stimulated in nonmitotic cells by the addition of growth factors (2,6). Thus, the phosphorylation state of Myc a t Ser6* may be regulated by the ERT protein kinase during the G1 phase of the cell cycle and by the cdc2 protein kinase during mitosis. Further studies are required to test this hypothesis, but evidence for the interphase phosphorylation of some cdc2 substrates has been obtained from studies of nuclear lamins (35-37).
Nuclear lamins are phosphorylated by the cdc2 protein kinase at several sites including Ser** and Ser3'* (35). Mutagenesis studies have established that both S e P and Ser3" are required for nuclear envelope breakdown during mitosis (36). Ser"9' is specifically phosphorylated in response to maturation promoting factor (a complex that contains the cdc2 protein kinase), but Ser2* is phosphorylated during both mitosis and interphase (35). Examination of the primary sequence surrounding the mitosis-specific site, Ser3" (30, 31), indicates that it does not correspond to the consensus sequence identified for the ERT protein kinase (Fig. 8). In contrast, S e P is located within the primary sequence Pro-Leu-Ser-Pro-Thr-Arg (30, 31). This sequence contains the consensus for the E R T protein kinase (Pro-Leu-Ser-Pro) and the consensus for the cdc2 protein kinase (Ser-Pro-X-Arg/Lys, Ref. 34). As the cdc2 protein kinase is only activated during mitosis (34), it is unlikely that this enzyme can account for the interphase phosphorylation of nuclear lamins at S e P . A different protein kinase must therefore account for the phosphorylation of SerZ2 observed during interphase (35). The ERT protein kinase represents a candidate enzyme that may account for this activity. This proposal represents a speculative hypothesis that suggests that the ERT protein kinase may exhibit some cdc2-like activity a t early stages of the cell cycle. A rigorous test of this hypothesis is warranted.
Conclusions-The ERT protein kinase is a MAP2-related protein kinase that is markedly and rapidly stimulated by growth factors (2, 4, 6). Substrates phosphorylated by this enzyme include the nuclear proto-oncogene products Myc (Ser6') and Jun (Ser246). A consensus primary sequence for substrate phosphorylation was identified as Pro-Leu-Ser/ Thr-Pro. Proteins containing this consensus sequence represent potential in vivo substrates for the ERT protein kinase. The phosphorylation of regulatory proteins such as Myc and Jun by the ERT protein kinase may represent an important pathway of signal transduction. A goal for further studies will be to test this hypothesis and to define the physiological significance of the phosphorylation during signal transduction.
Note Added in Proof-We have observed that the ERT protein kinase substrates Myc Ser6' and Jun Ser246 are also phosphorylated by the purified MAP2 protein kinase. It is therefore likely that Myc and Jun are phosphorylated by several members of the MAP2 family of protein kinases. Recently, it has been reported that the MAP2 kinase phosphorylates myelin basic protein at Thrg7 within the se-  Biol. Chem. 265, 19728-19735). Myelin basic protein is also phosphorylated at Thrg7 by the ERT protein kinase (Ref. 4). Substrates that have been identified for MAP2 and MAP2-related kinases can therefore be represented by the consensus sequence Pro-(Leu/Arg)-(Ser/Thr)-Pro. Further work will be required to establish the precise substrate specificity of these protein kinases.