Autophosphorylation and protein kinase C phosphorylation of the epidermal growth factor receptor. Effect on tyrosine kinase activity and ligand binding affinity.

The effect of autophosphorylation and protein kinase C-catalyzed phosphorylation on the tyrosine-protein kinase activity and ligand binding affinity of the epidermal growth factor (EGF) receptor has been studied. Kinetic parameters for the phosphorylation by the receptor kinase of synthetic peptide substrates having sequences related to the 3 in vitro receptor autophosphorylation sites (tyrosine residues 1173 (P1), 1148 (P2), and 1068 (P3)) were measured. The Km of peptide P1 (residues 1164-1176) was significantly lower than that for peptides P2 (residues 1141-1151) or P3 (residues 1059-1072). The tyrosine residue 1173 was also the most rapidly autophosphorylated in purified receptor preparations, consistent with previous observations for the receptor in intact cells (Downward, J., Parker, P., and Waterfield, M. D. (1984) Nature 311, 483-485). Variation in the extent of receptor autophosphorylation from 0.1 to 2.8 mol of phosphate/mol of receptor did not influence kinase activity or EGF binding affinity either for purified receptor or receptor in membrane preparations. Phosphorylation of the EGF receptor by protein kinase C was shown to cause a 3-fold decrease in the affinity of purified EGF receptor for EGF and to reduce the receptor kinase activity. In membrane preparations, phosphorylation of the EGF receptor by protein kinase C resulted in conversion of high affinity EGF binding sites to a low affinity state. This suggests that activation of protein kinase C by certain growth promoting agents and tumor promoters is directly responsible for modulation of the affinity of the EGF receptor for its ligand EGF. The regulation of the EGF receptor function by protein kinase C is discussed.

(P2), and 1068 (P3)) were measured. The K, of peptide P1 (residues 1164-1176) was significantly lower than that for peptides P2 (residues 1141-1151) or P3 (residues 1059-1072). The tyrosine residue 1173 was also the most rapidly autophosphorylated in purified receptor preparations, consistent with previous observations for the receptor in intact cells (Downward, J., Parker, P., and Waterfield, M. D. (1984) Nature 311, [483][484][485]. Variation in the extent of receptor autophosphorylation from 0.1 to 2.8 mol of phosphate/mol of receptor did not influence kinase activity or EGF binding affinity either for purified receptor or receptor in membrane preparations.
Phosphorylation of the EGF receptor by protein kinase C was shown to cause a %fold decrease in the affinity of purified EGF receptor for EGF and to reduce the receptor kinase activity. In membrane preparations, phosphorylation of the EGF receptor by protein kinase C resulted in conversion of high affinity EGF binding sites to a low affinity state. This suggests that activation of protein kinase C by certain growth promoting agents and tumor promoters is directly responsible for modulation of the affinity of the EGF receptor for its ligand EGF. The regulation of the EGF receptor function by protein kinase C is discussed.
The binding of EGF' to specific cell surface receptors initiates, in certain cells, a series of biochemical events which can induce cell proliferation by an as yet unclear mechanism (1, 2). A number of different experimental approaches, including the determination of the primary amino acid sequence, have been used to develop a working model for the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: EGF, epidermal growth factor; EGFR-P, epidermal growth factor receptor-phosphorylated preparation; EGFR-C, epidermal growth factor receptor-control preparation; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate; PMA, 4B-phorbol 12p-myristate 13a-acetate; HPLC, high performance liquid chromatography; PTH, phenylthiohydantoin. structure of the receptor. In the model, the receptor polypeptide is divided into four distinct domains (3)(4)(5)(6). Extracellular is an amino-terminal domain which binds EGF and this is linked through a putative single transmembrane domain to the cytoplasmic region of the receptor. This region has a domain with tyrosine-protein kinase activity which is able to autophosphorylate a carboxyl-terminal domain of the receptor. The amino acid sequence of the cytoplasmic receptor kinase domain is homologous to the conserved sequences of the src family of transforming proteins; the majority of which possess a protein-tyrosine kinase activity (4). It has been proposed that expression of a truncated EGF receptor by the v-erb B oncogene of avian erythroblastosis virus may induce transformation through generation of a ligand-independent receptor signal (3). The protein-tyrosine kinase activity (7), which is also associated with the v-erb-B protein (8-lo), is the sole known intrinsic EGF receptor enzymatic activity and has consequently been the focus of much investigation with the goal of identifying physiological substrates that are important in signal transduction.
Autophosphorylation of the EGF receptor may represent a primary or regulatory event with respect to signal transduction. Investigations in our laboratory have shown that purified EGF receptor phosphorylates the receptor polypeptide at three distinct tyrosine residues which have been located close to the receptor carboxyl terminus at residues 1173, 1148, and 1068 (5). Tyrosine residue 1173 is the most extensively modified tryosine residue in intact A431 cells treated with EGF (5). Additionally, in intact cells the EGF receptor is phosphorylated on serine and threonine residues as well as on tyrosine residues (5), indicating that other protein kinases can alter the phosphorylation state and presumably the function of this receptor. The only protein kinases that have been implicated in these phosphorylations are protein kinase C (12-15) and the cyclic AMP-dependent protein kinase (16, 17). While there is reason to doubt the physiological significance of the phosphorylation by the cyclic AMP-dependent protein kinase, there is good evidence for a functional role for the protein kinase C phosphorylation. The receptor site phosphorylated by protein kinase C has previously been identified by comparative peptide mapping and indirect sequencing (18,19). Work employing phorbol esters and synthetic diacylglycerols indicates that not only may protein kinase C be involved in the phosphorylation of the receptor in vivo, but also provides circumstantial evidence that this may underlie the regulation of receptor affinity (transmodulation) (20).
In this study the site of the EGF receptor phosphorylated by protein kinase C has been directly sequenced. The effect of this phosphorylation and of the tyrosine autophosphorylations on the protein-tyrosine kinase activity and EGF bind-ing affinity of the receptor in both purified and membrane systems has been studied to elucidate the role of these phosphorylation events in the control of cell proliferation.

Materials
Peptides PI, P2, P3, and RS were synthesized by Cambridge Research Biochemicals, and peptide RS was also obtained from Peninsula Laboratories. Human epidermoid carcinoma A431 cells were grown as described in Ref. 21 and plasma membranes from these cells were isolated by the method of Thom et al. (22). The EGF receptor was purified from these cells using monoclonal antibody 9A as described previously (23). [y-''P]ATP (>5000 Ci/mmol) was from Amersham. EGF was prepared from adult male mouse submaxillary glands (24) and iodinated using the glucose oxidase method (25). Protein kinase C was purified from bovine brains (26).

Methods
Tyrosine Kinase Assay-For measurement of tyrosine kinase activity in solubilized A431 cell plasma membranes, 10 pg of membrane protein was incubated with 50 mM Hepes, pH 7.4, 5% glycerol, 0.2% Triton X-100, 150 mM NaCI, 2 mM MnC12, 12 mM MgC4, 100 p M Na3V04, and 20 p~ [y-32P]ATP (5 Ci/mmol) in a total volume of 50 11 with varying concentrations of substrate peptide. Reactions were carried out for 2 min at 30 or 0 "C in the presence or absence of 200 nM EGF as stated. The reaction was stopped by the addition of 45 pl of 5% trichloroacetic acid followed by 5 pl of 30 mg/ml bovine serum albumin. The mixture was kept at 0 "C for 15 min then centrifuged at 10,000 X g for 15 min at 4 "C. 50 pl of each supernatant was spotted onto phosphocellulose paper squares (1.5 cm, Whatman P l l ) which were then washed in 30% acetic acid (3 X 10 min), and adsorbed radioactivity was measured. When purified receptor was used, incubations were carried out as above, except membranes were replaced by 2.5 ng of receptor protein. To determine the amount of ["PI phosphate incorporated into purified receptor in the course of a kinase assay, the trichloroacetic acid-precipitated protein pellet was redissolved in 50 pl of 1 M NaOH, and the suspension was heated at 60 "C for 3 min, vortexed vigorously, and spotted onto Whatman 3 " paper squares (1.5 cm). These were washed in 10% trichloroacetic acid (3 X 30 min), and adsorbed radioactivity was measured.
Time Course of the Incorporation of Phosphate into Each Autophosphorylation Site of the EGF Receptor-100 pg of A431 Thom membranes (0.7 pmol of EGF receptor) were incubated in the phosphorylation buffer described above, without ATP but including 1 pg/ ml EGF, for 30 min at 18 "C (total volume 100 pl). [y-3'P]ATP (100 Ci/mmol) was added (final concentration 5 p~) and the mixture was incubated at 0 "C between 0.5 and 5 mi n. The reaction was stopped by the addition of 30 pl of 4 X SDS gel sample buffer. Each mixture was run on a 7% SDS-polyacrylamide gel, and the EGF receptorcontaining region of the gel was cut out and washed with 85% acetone, 5% acetic acid, 5% triethylamine, 5% water (3 X 30 min). The gel slice was then incubated with 500 pl of 100 mM NH4HC03 containing 1 mg/ml DPCC-treated trypsin (Sigma) for 48 h a t 37 "C. The released peptides were applied to a Vydac C-18 reverse phase HPLC column in 0.1% trifluoroacetic acid and eluted with an acetonitrile gradient (as described in Ref. 5).
Preparation of Prephosphorylated EGF Receptors-Receptor was purified in active form as described previously (23) except that the monoclonal antibody 9Afcohnn material carrying immobilized receptor after washing was divided into two equal parts by mass. One part (EGFR-P) was incubated with 100 p~ [y-32P]ATP (0.05 Ci/ mmol), 50 mM Hepes, pH 7.4, 5% glycerol, 0.2% Triton X-100, 150 mM NaCl, 2 mM MnC12, 12 mM MgC12, 100 p~ Na,VO, for 2 h at 4 "C. The other part (EGFR-C) was incubated in a similar solution which lacked ATP. At the end of this incubation the two columns were washed and eluted as previously described (23). The extent of receptor phosphorylation was estimated from measurement of the amount of receptor present using radioimmunoassay and of the amount of 32P label incorporated into trichloroacetic acid-precipitable protein; this gave a figure of 1.6 mol of phosphate/mol of receptor for EGFR-P. The amount of unlabeled phosphate present in the receptor was estimated by chemical analysis (27); the total stoichiometry was found to be 0.5 mol of phosphate/mol of receptor for the control preparation EGFR-C and was assumed to be mostly on serine and threonine (11).

Preparative Phosphorylation of EGF Receptor by Protein Kinase
C-EGF receptor was purified from 5 X 10' A431 cells using the monoclonal antibody R1 as described (3). After the immobilized receptor was washed, the column was incubated with 1 0 0 p~ unlabeled ATP in the same buffer as described above for prephosphorylation of receptor on monoclonal antibody 9A. After 2 h at 4 "C the column was again washed and then incubated with 20 mM Hepes, pH 7.2, 10 mM MgCI,, 0.5 mM CaC12, 100 pg/ml phosphatidyl serine, 10 pg/ml diolein, lOOpM [Y-~'P]ATP (10 Ci/mmol), 1000 units of protein kinase C in a volume of 10 ml for 1 h at 18 "C. The column was then washed again and receptor eluted as described previously (3).
Purification and Sequence Analysis of the Peptide from the EGF Receptor Phosphoryhted by Protein Kinase C-The protein kinase Cphosphorylated receptor was fully reduced and alkylated, further purified by gel permeation HPLC, and cleaved with cyanogen bromide and the digest was fractionated by gel permeation HPLC as described previously (3). The major phosphopeptide peak eluted was dialyzed against 50 mM NKHCQ,, lyophilized, and redissolved in 10 p1 of 100 mM NH4HC03, pH 7.6, which contained 10 pg/ml Staphylococcus aureus V8 protease. The digestion was continued for 48 h at 37 "C and then the digest was loaded onto a Vydac C18 reverse phase HPLC column equilibrated in 0.1% trifluoroacetic acid. A linear gradient from 0 to 40% acetonitrile was run in 40 min at a flow rate of 1 ml/ min and 0.5-ml fractions were collected. The major phosphopeptide peak from this separation was lyophilized and the amino-terminal operated as described by Hewick et al. (28). Phenylthiohydantoin sequence was determined using a gas phase sequencer assembled and (PTH) derivatives of amino acids were analyzed by HPLC as described in Ref.

3.
Phosphoamino Acid Analysis-The identity of the phosphoamino acid in phosphorylated peptides and proteins was determined by onedimensional electrophoresis at pH 3.5 (7).
Protein and Peptide Concentration Determinations-The concentration of EGF receptor was determined both by radioimmunoassay (25) and by amino acid analysis using a Beckman System 630 High Performance Analyzer. Peptide and EGF concentrations were also determined by amino acid analysis.

'=1-EGF Binding to Solubilized Receptor Phosphorylated by Protein
Kinase C-This was determined by radioimmunoassay (25): 0.45 pmol of purified EGF receptor was added to an incubation solution containing 10 mM MgC12, 20 mM NaF, 100 p~ ATP (where applicable), 50 mM Hepes, pH 7.4, 0.01% Triton X-100, 1 mM CaC12, 100 pg/ml phosphatidylserine, 5 units of purified protein kinase C (where applicable), and 2 pg of monoclonal antibody R1 in a total volume of 0.2 ml, along with varying amounts of lZ5I-EGF (0.1-200 nM), added 15 min after the other reagents. This mixture was incubated at 18 "C for 1 h, then 10 p1 of a 1:l slurry of Protein A-Sepharose (Pharmacia) was added and the tube was tumbled for 20 min. The Protein A-Sepharose was washed by centrifugation and resuspension and counted for y radiation as described previously (25). Backgrounds were measured by replacing the R1 antibody with 10 p1 of a 1:lOO dilution of normal mouse serum. Each point was determined in duplicate with background subtracted.

"'I-EGF Binding to EGF Receptor in Membranes Phosphorylated by Protein
Kinase C-Due to the presence of phospholipid in the plasma membranes, proteolytically activated protein kinase C was used in these experiments in order that control incubations could be carried out in the presence of inactive protein kinase C. Protein kinase C (150 units/ml) was incubated in 10 mM Tris, pH 8.0, 0.5 mg/ml bovine serum albumin, 20 mM P-mercaptoethanol, and 4 pg/ ml trypsin for 3 min at 30 "C before the reaction was stopped with a 50-fold excess of trypsin inhibitor. For control incubations trypsin inhibitor was added before the trypsin. The binding reaction was performed in a total volume of 200 pl of 10 mM MgCI2, 20 mM NaF, 100. p M ATP, 50 mM Hepes, pH 7.4, 3 units of activated protein kinase C with 50 pg of A431 plasma membrane protein (0.35 pmol of EGF receptor) and varying amounts of lz5I-EGF (added 15 min after the other reagents). The mixture was incubated for 1 h at 18 "C, applied to glass fiber filters (Whatman GFC), and washed with 3 x 20 ml of 10 mM Hepes, pH 7.4, containing 1 mg/ml bovine serum albumin. The filters were then counted for y radiation. Backgrounds were estimated by the inclusion of a 100-fold excess of unlabeled EGF in the incubation.  ing sequences identical to those surrounding the three major in vitro autophosphorylation sites of the EGF receptor (5) were synthesized (Table I, Pl-P3). Two arginine residues not present in the analogous EGF receptor sequence were added to the amino terminus of each peptide, thus allowing them to adhere to phosphocellulose paper at low pH (29). Peptides Pl-P3 and peptide RS, a tyrosine kinase substrate peptide containing the autophosphorylation site 0fpp60"-'~ (30), were used as substrates for the EGF receptor kinase. The kinetic parameters for the phosphorylation of these peptides in the presence or absence of EGF by the receptor are shown in Table I. In each case phosphoamino acid analysis of the phosphorylated peptide showed that only tyrosine was phosphorylated (not shown). The assays were performed using 10 pg of solubilized A431 membrane protein at 30 "C as described under "Experimental Procedures." Although the Vmax values determined were similar, there were significant differences in the K, values for each peptide. Peptide P1 had a lower K,,, than the other peptides, possibly indicating that the phosphorylation site from which it is derived would be a "better" substrate than the other phosphorylation sites for the autophosphorylation reaction. P1 is known to be the major tyrosine phosphorylation site utilized on the EGF receptor in whole cells (5). Ordered Autophosphorylation of the EGF Receptor-It is known that the incorporation of 1 phosphate/receptor by autophosphorylation is kinetically favored over the second and third site phosphorylations (31). Due to the low K,,, of peptide P1 and the use of site P1 in vivo it seemed possible that in vitro this site was also favored. To investigate this, the incorporation of [32P]phosphate into receptor was studied in the presence of EGF using solubilized A431 cell plasma membranes at 0 "C with 5 PM ATP. The reaction was stopped and tryptic peptides of the EGF receptor were analyzed by HPLC (see "Experimental Procedures"). In Fig. 1 the amount of label incorporated into each site is shown plotted against the time of incubation. It can be seen that site P1 is phosphorylated more rapidly than sites P2 and P3, indicating that phosphorylation at P1 is kinetically preferred in vitro, consistent with the situation observed in intact cells.

Synthetic Autophosphrylation Site Peptides as Substrates for the Receptor Protein-Tyrosine Kinase-Peptides contain-
The Effect of Autophosphorylatwn on EGF Receptor Function-Although the autophosphorylation of the EGF receptor has been studied in detail, the consequences of this modification on the kinase activity of the receptor and its EGF binding affinity have not been characterized. To investigate these parameters, EGF receptor was purified from A431 cells using the anti-carbohydrate monoclonal antibody 9A (23). Two preparations of receptor were made: a control preparation ("EGFR-C") which was not autophosphorylated and a phosphorylated preparation ("EGFR-P") which was extensively autophosphorylated while immobilized on the antibody matrix (see "Experimental Procedures"). Autophosphoryla- tion of EGFR-P was carried out in the absence of EGF for 2 h; this has been shown to give a pattern of phosphorylation similar to that resulting from briefer incubations in the presence of EGF (5). The extent of autophosphorylation of EGFR-P was 1.6 mol of tyrosine phosphate/mol of receptor. The two preparations contained the same concentration of receptor as measured by radioimmunoassay (25) or by amino acid analysis. The kinetic parameters of these two preparations with respect to the phosphorylation of the synthetic peptide P1 were determined and are shown in Table 11. Under all of the conditions used, the two preparations showed very similar enzymatic activities as judged by peptide K, and V,,, values.
Since significant levels of autophosphorylation of the receptor may be taking place during the time period used for the kinase assay, it was possible that the control "unphosphorylated" preparation (EGFR-C) had effectively become phosphorylated. In order to investigate this possibility, it was therefore necessary to measure the level of phosphorylation at the end of the assay. The rate of peptide P1 phosphorylation at varying concentrations of peptide for the preparations EGFR-P and EGFR-C in the presence or absence of EGF and at 30 or 0 "C is shown in Fig. 2. Also shown in Fig. 2 are the stoichiometries of phosphate newly incorporated into the receptor during the course of the assay. It is evident that significant levels of autophosphorylation can occur under these conditions and thus autophosphorylation is inhibited

Epidermal Growth Factor Receptor Phosphorylation 14541
by high concentrations of peptide substrate.
To investigate the effect of the role of autophosphorylation on the tyrosine protein kinase activity of the EGF receptor, the data shown in Fig. 2 was used to calculate V, , values which are shown in Fig. 3 expressed as a function of receptor phosphorylation. Since the prephosphorylated receptor preparation EGFR-P had been shown to contain 1.6 mol of phosphate/mol of receptor, this value was added to that incorporated during the assay. Results shown in Fig. 3 demonstrate that over widely varying levels of phosphorylation, from 0.1 mol of phosphate/mol of receptor to greater than 2.8 mol of phosphate/mol of receptor there is no significant difference in the V, , for the phosphorylation of exogenous substrate either in the presence or absence of EGF. It therefore appears that the level of autophosphorylation does not influence this activity, at least in purified preparations of solubilized receptor.
Since the affinity of receptor for EGF changes greatly upon solubilization (25), presumably as a result of conformational changes in the receptor, the kinase activity towards exogenous substrate peptide P1 was characterized for EGF receptor in  Table' I11 show that the level of receptor phosphorylation did not effect the activity of the kinase whether the receptor was in membranes or purified in solution.

Receptor Phosphorylation by Protein Kinase C-The cal-
cium-and phospholipid-dependent protein kinase, protein kinase C, has been shown to phosphorylate the EGF receptor (12,18). This phosphorylation site has been located by comparative peptide mapping to threonine 654 (18,19). In order The effect of receptor autophosphorylation on the exogenous kinase activity, 11. The data presented in Fig. 2 has been used to predict the V,, values for the kinase reaction with respect to peptide under the conditions described in the legend to Fig. 2. The rate of the kinase reaction, V, at a given substrate concentration, [SI, and a known K,,, for peptide (see Table 11) allow the value of V, . to be calculated from the Michaelis-Menten equation to positively establish this assignment, we have directly determined the amino acid sequence of the phosphorylated peptide derived from purified EGF receptor which had been p h o s p h o~l a t~ by protein kinase C. The human EGF receptor from A431 cells was purified using a monoclonal antibody R1 aelnity column (3), and the immobilized receptor was washed and then incubated with unlabeled ATP for 2 h at 4 "C to allow the tyrosine autophospho~lation to occur. ATP was removed by washing and protein kinase C, calcium,  4. Deter~nation of the sequence of the site at which protein kinase C phosphorylates the EGF receptor. Receptor was purified and phosphorylated by protein kinase C and digested with cyanogen bromide as described under "Experimental Procedures." A, separation of cyanogen bromide peptides on a TSK 3,000 sizing column. Sotid line, optical density profile at 280 nm of cyanogen bromide peptides eluted from the TSK 3,000 column. 0.3-ml fractions were collected and counted for 32P label; the broken line shows the profile of 32P label/fraction. The positions at which standard molecular weight marker proteins run on this system are marked with arroms. B, reverse phase HPLC purification of a V8 digest of the major phosphopeptide peak from A. The solid line shows the optical density profile of the eluted peptides at 206 nm. The broken line shows the profile of 32P label per fraction. C, fractions corresponding to the major phosphopeptide peak from B were pooled, lyophilized, and sequenced using a gas phase sequencer (see "Experimental Procedures"). The yield of PTH amino acid (PTH-Qu) at each cycle is plotted against residue number. The identity of the residue is written above each peak in single letter code.
phosphatidylserine, and [Y-~TIATP were added as described under "Experimental Procedures." After 1 h at 18 "C, the column was again washed, receptor eluted and purified as previously described (3). Amino acid analysis showed that 210 pmol of pure EGF receptor were obtained. This material was digested with cyanogen bromide and the resulting peptides were fractionated by gel permeation HPLC in guanidine solution (Fig. 4A). The major 32P-labeled component was found to contain phosphothreonine while the two minor peaks corresponding to higher molecular weight peptides contained phosphotyrosine and the minor smaller molecular weight peak

Epidermal Growth Factor
Receptor Phosphorylation 14543 was free phosphate (data not shown). Fractions corresponding to the phosphothreonine-containing; peak were pooled, digested with V8 protease, and analyzed on a reverse phase HPLC column using 0.1% trifluoroacetic acid buffers and an acetonitrile gradient (Fig. 4B). Fractions containing the 32Plabeled peptide were pooled, and the amino-terminal sequence was determined on a gas phase sequencer (see "Experimental Procedures"). The yields and identities of phenylthiohydantoin derivatives of amino acids at each cycle are shown in Fig.  4C. The identification of 7 residues at the amino terminus of this peptide made it possible to unequivocally assign threonine 654 as the target for protein kinase C. When EGF receptor from A431 cells metabolically labeled with [32P]orthophosphate was digested by this protocol, the only phosphopeptide found in receptor from tumor promoter (PMA) treated cells which was absent in receptor from control cells comigrated with the peptide sequenced here on reverse phase HPLC at pH 2.0 and 6.5 (data not shown).
Functional Effect of EGF Receptor Threonine 654 Phosphorylation-The effects of protein kinase C phosphorylation on ligand binding and protein-tyrosine kinase activity of the EGF receptor were studied. First, the exogenous kinase activity of protein kinase C phosphorylated, and control receptor preparations at varying concentrations of EGF (see Fig. 5 ) were determined. Purified EGF receptor from A431 cells (23) was incubated with calcium, phosphatidylserine, and MgATP with or without purified protein kinase C using concentrations of EGF from 0 to 160 nM (as described under "Experimental Procedures"). The kinase activity towards the pp6PC-derived tyrosine kinase substrate peptide (RS in Table I) was measured in each case; this peptide was used because peptides P1, P2, and P3 acted as substrates for protein kinase C. It was found that the response of the EGF receptor protein-tyrosine kinase activity to EGF was attenuated following phosphorylation by protein kinase C. Thus, it required a %fold higher concentration of EGF to obtain a 50% activation of the EGF receptor protein-tyrosine kinase after phosphorylation of receptor by protein kinase C. The maximal kinase activity at saturating EGF concentrations was not, however, greatly altered protein kinase C phosphorylation caused a reduction of only 19 k 7% (mean and S.E., n = 3).
These results imply that phosphorylation of the EGF receptor at threonine 654 may directly decrease the affinity of the receptor for EGF. This was confirmed by Scatchard analysis (32) of the binding of lZ5I-labeled EGF to solubilized purified receptor using a radioimmunoassay (25). Scatchard plots for ligand binding to unphosphorylated EGF receptor, autophosphorylated receptor, and receptor phosphorylated by purified protein kinase C are shown in Fig. 6. While the affinity of the receptor for EGF was not changed by autophosphorylation (Kd = 16.4 nM autophosphorylated, 14.9 nM unphosphorylated), phosphorylation by protein kinase C caused a significant decrease in the apparent affinity (Kd = 46.6 nM). Thus, the affinity of the EGF receptor for its ligand in solution is decreased 3-fold when receptor is phosphorylated by protein kinase C, in agreement with the datapresented in Fig. 5 .
Since the affinity of the EGF receptor for its ligand is known to be greatly reduced upon solubilization (25), the effect of protein kinase C phosphorylation on the receptor in membrane preparations was studied. The binding of lZ5I-EGF to a plasma membrane preparation from A431 cells was investigated using a filter assay (see "Experimental Procedures''). Fig. 7 shows Scatchard plots generated for binding of EGF to nonphosphorylated receptor, autophosphorylated receptor, and for protein kinase C phosphorylated receptor (0.8 mol of phosphate/mol of receptor at Thr 654). The total number of receptor binding sites was unaffected by phosphorylation. The Scatchard plots for receptor in the absence of protein kinase C are curvilinear: this may indicate the presence of both low (Kd = 7 nM) and high affinity binding sites (Kd = 0.8 nM). For receptor which has been phosphorylated by prbtein kinase C, the Scatchard plot is virtually linear suggesting that the high affinity binding sites have been converted to low affinity sites by phosphorylation of the receptor; the total number of binding sites remained constant.

DISCUSSION
A detailed study of the effect of phosphorylation on the EGF receptor protein-tyrosine kinase and its ligand binding properties has been presented. Peptides containing sequences identical to the major in uitro autophosphorylation sites were synthesized and used as substrates for the EGF receptor protein-tyrosine kinase. The peptide containing the sequence surrounding tyrosine 1173 appeared to be the preferred substrate as deduced from the observed lower @-fold) apparent K,. This result was consistent with the observation that tyrosine 1173 is also the most rapidly autophosphorylated site on the intact purified receptor. In addition, tyrosine 1173 is the phosphorylation site that is most extensively phosphorylated in intact A431 cells in response to EGF (5). The apparent preference for the phosphorylation of receptor tyrosine 1173 in uitro thus appears to correlate with the situation observed for receptor in uiuo.
Our study of the effects of autophosphorylation on proteintyrosine kinase activity have shown that over a wide range of stoichiometry of phosphorylation (0.1-2.8 mol of phosphate/ mol of receptor) there is no apparent change in either EGF binding to receptor or of receptor protein-tyrosine kinase activity. These results were obtained with both purified detergent-solubilized EGF receptor and also with receptor in A431 cell membrane preparations. This lack of response to autophosphorylation is consistent with previous published studies. For example, thiophosphorylation of the receptor does not greatly alter receptor kinase activity (33) and it has been shown that there is no lag phase in the phosphorylation of A431 cell membrane proteins in response to EGF (34), again indicating that autophosphorylation does not appear to activate the EGF stimulated protein-tyrosine kinase activity.
The lack of effect of autophosphorylation on the activity of the EGF receptor protein-tyrosine kinase is interesting when the effects of autophosphorylation on certain protein-tyrosine kinases of the src family are considered. Thus, although the Receptor Phosphorylation kinase activity of pp6OV-"" itself is thought to be activated by phosphorylation on tyrosine residues (35), removal of the major site of autophosphorylation (tyrosine 416) by mutagenesis does not alter the kinase activity of pp60""" (36, 37), perhaps indicating that the regulatory autophosphorylation is occurring at a second site. Different results were reported in studies of P130Pag~fp". In this case, deletion of the major autophosphorylation site, tyrosine 1073 (homologous to src tyrosine 416), caused a 5-fold decrease in kinase activity (38). In the case of the insulin receptor, autophosphorylation on tyrosine generates an active receptor protein-tyrosine kinase that is no longer dependent upon insulin (39,40). The insulin receptor is phosphorylated at 2 or 3 tyrosines (40,41), none of which have been assigned in the primary structure. In the insulin receptor, there is no domain analogous to the Cterminal domain which includes the autophosphorylation sites of the EGF receptor (42). Thus, it seems that there is no general or consistent response to autophosphorylation among these different protein-tyrosine kinases. In the case of the EGF receptor, one may suppose that autophosphorylation plays some as yet undefined role in the function of the receptor. One possible effect of autophosphorylation might be to alter the protein substrate specificity of the kinase. In this study, we have presented data based on the activity directed towards low molecular weight substrates and we cannot of course rule out an effect where autophosphorylation of the receptor may alter access of larger protein substrates to the catalytic site.
With respect to the role of the receptor as a protein kinase, the specific activity of the purified EGF receptor proteintyrosine kinase is 1035 units/mg at saturating substrate concentrations. This value is similar to those observed for purified pp6OV-" and Abelson protein-tyrosine kinases and is similar to the specific activity of a number of serine/threoninespecific protein kinases. The turnover number for this protein relative to the catalytic subunit of the CAMP-dependent protein kinase (2500 units/mg) is 1.7. The well-characterized role of the CAMP-dependent protein kinase as a regulator of intracellular protein phosphorylation and the similar phosphorylating potential of the purified EGF receptor suggests that the EGF receptor may indeed function as a protein tyrosine kinase in uiuo. In this respect, we have been unable to demonstrate the phosphorylation of phosphatidylinositol employing purified EGF receptor and would conclude that this is not an activity intrinsic to the receptor?
We have determined the amino acid sequence of the site on the EGF receptor phosphorylated by protein kinase C (threonine 654). This direct observation confirms the conclusion of Hunter et al. (18) and Davis and Czech (19) where in uitro phosphorylated synthetic peptides based on the sequence around Thr 654 were compared with phosphopeptides from the receptor by peptide mapping. Our data show that phosphorylation of the EGF receptor at Thr 654 directly reduces receptor affinity for its ligand EGF, both in purified preparations and in membrane systems. For purified solubilized receptor, phosphorylation by protein kinase C induces an absolute change in ligand binding from high to low affinity, while for receptor in membranes the curvilinear Scatchard plot becomes straight, indicating that conversion of high affinity to low affinity sites takes place within a mixed population or, conceivably, indicating that a homogenous population of high affinity sites interacting with negative cooperativity has been converted to a population of noninteracting low affinity sites. This implies that the loss of high affinity J. Downward, M. D. Waterfield, and P. Parker, unpublished observations. EGF binding sites seen on treatment of A431 cells and fibroblasts with tumor promoters (14,43) or diacylglycerol (15) is caused by the phosphorylation of the EGF receptor on Thr 654 by activated protein kinase C. Our experiments have not included tumor promoters or diacylglycerols as activators of protein kinase C and therefore indicate that these amphiphilic compounds do not directly contribute to the drop in affinity.
The loss of high affinity EGF binding sites observed when Swiss 3T3 cells are treated with PDGF (20,44), bombesin (45) or vasopressin (46) ("transm~dulation~~) is therefore probably mediated by the direct action of protein kinase C on the EGF receptor polypeptide itself; all these agents will stimulate phosphatidylinositol turnover and may therefore activate protein kinase C.
On treatment of A431 cells with tumor promoters, there is a large reduction in the level of phosphotyrosine in the EGF receptor (50% at 0.1 ng/ml PMA) (43). However, protein kinase C phosphorylation of the EGF receptor in vitro reduces its kinase activity at saturating EGF levels by only 20%. This apparent contradiction could be explained by assuming either that additional constraints exist for the receptor in the cell membranes which are lost on solubilization, or that a finely balanced cycle of autophosphorylation and dephosphorylation by a protein-tyrosine phosphatase exists which would amplify such changes in the kinase activity. It should also be emphasized that at subsaturating concentrations of EGF, the receptor protein-tyrosine kinase activity is suppressed by greater than 20%. Thus, at 14 nM EGF ( K for EGF) the effect of phosphorylation of threonine 654 is to decrease the receptor kinase activity by approximately 60% (see Fig. 5).
The protein kinase C phosphorylation site on the EGF receptor (threonine 654) lies in the cytoplasmic domain, 10 residues C-terminal to the putative transmembrane sequence, within a highly basic stretch of amino acids which contains 8 out of 13 basic residues. The proximity of this sequence to the transmembrane domain suggests that these positively charged residues may interact with negatively charged head groups in the phospholipid bilayer. This interaction could be significantly perturbed by the introduction of a phosphate group at threonine 654. Such an effect may underlie the functional changes associated with the phosphorylation of the receptor by protein kinase C. While our results indicate that phosphorylation of the receptor at Thr 654 causes a reduction in EGF binding affinity, it does not completely explain the presence of apparently high and low affinity receptors in intact cells. If A431 cells are labeled to equilibrium with [32P]orthophosphate and the EGF receptor is purified in the presence of phosphatase inhibitors, there is no detectable phosphorylation at Thr 654.' Nevertheless, the receptor on these cells displays both high (5-10%) and low (90-95%) affinity states. It is therefore necessary to propose that factors in addition to the phosphorylation state of threonine 654 are involved in the distribution between high and low affinity states of the EGF receptor. There are many possible explanations for this situation; however, it should be emphasized that in vitro with purified receptor one can observe an affinity change on phosphorylation by protein kinase C (see Fig. 6) indicating that in this situation the change in affinity is due to an alteration in the intrinsic physical state of the receptor and not to interactions of the receptor with other proteins. Perhaps the simplest explanation for these observations would be that the EGF receptor exists in a dynamic equilibrium between different aggregated forms, for example monomer and dimer. These forms could exhibit differing affinities for EGF, such that monomeric receptor would bind with high affinity and dimeric receptor with low affinity. In this event one can calculate an equilibrium constant (Keg) given the total number of receptors/cell and the relative amounts of high and low affinity sites. Thus, from Ref. 20 we would calculate Keq = 4.4 X [concentration units]-' for Swiss 3T3 cells. This equilibrium constant has dimensions of l/concentration; it is evident, therefore, that changes in the total concentration of receptor will affect. the monomer/dimer ratio. To take the extreme case of A431 cells for example, where there are 3.3 X lo6 receptors/cell (15), and using the equilibrium constant derived for 3T3 cells, one can calculate that the proportion of receptors in a high affinity state (monomeric) would be 6.5% (accounting for the larger surface area of Swiss 3T3 cells relative to A431 cells). This value is surprisingly close to that determined empirically and would be consistent with the hypothesis. Receptor-receptor interaction provides not only one of many possible solutions to the paradox outlined above but also a basis for the interactions between the intracellular and extracellular domains of the receptor. Elaborating this simple model, one might predict that EGF also induces changes in the receptor aggregation state. Thus, EGF would bind initially (although not exclusively) to its high affinity site, creating a conformational change in the extracellular domain that would promote interation between such domains. This draws into close proximity the intracellular domains and produces in the presence of bound EGF an active tyrosine kinase. This model is meant to provide a conceptual basis with which to probe the communication between extracellular and intracellular domains of the EGF receptor. If such interactions occur, it may transpire that autophosphorylation also plays a functional role at this quaternary level. Biophysical studies are at present being employed to elucidate these points.