Characterization of structural domains of the human epidermal ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 1984 by The American Society of Biological Chemists, Inc.

Voi. 259, No. 18, Issue of September 25, pp. 11534-11542,1984 Printed in U. S.A.

Characterization of Structural Domains of the Human Epidermal Growth Factor Receptor Obtained by Partial Proteolysis*

(Received for publication, December 12,1983)

Michael Chinkers$ and Joan S. Brugge From the Department of Microbiology, State University of New York ut Stony Brook, Stony Brook, New York 11 794

Partial cleavage with trypsin has been used to study the structure of the epidermal growth factor (EGF) receptor purified from human carcinoma cells. Follow- ing affinity labeling of the receptor with 12’I-EGF or the ATP analogue 5’-p-fluorosulfonyl benzoyl[14C] adenosine, metabolic labeling with [35S]methionine, [3H]glucosamine, or [32P]orthophosphate, or in vitro autophosphorylation with [y3’P]ATP, tryptic cleav- age defines the following three regions of the 180-kDa receptor protein: 1) a 125-kDa trypsin-resistant do- main which contains sites of glycosylation, EGF bind- ing, and an EGF-specific threonine phosphorylation site; 2) an adjacent 40-kDa fragment which contains serine and threonine phosphorylation sites and is fur- ther cleaved to a 30-kDa trypsin-resistant domain; and 3) a terminal l5-kDa portion of the receptor that con- tains the sites of tyrosine phosphorylation and is de- graded to small fragments in the presence of trypsin. Both the 125- and 40-kDa regions of the EGF receptor appear to be required for receptor-associated protein kinase activity since separation of these regions by tryptic cleavage abolishes this activity, and both re- gions are specifically labeled with an ATP affinity analogue, suggesting that both are involved in ATP binding. Additional 63- and 48-kDa phosphorylated fragments are generated upon trypsin treatment of EGF receptor from EGF-treated cells. The potential usefulness of partial tryptic cleavage in studying the EGF receptor and the possible biological function of the 30-kDa trypsin-resistant fragment of the receptor are discussed.

EGF’ is an M , 6045 polypeptide hormone that is mitogenic for a wide variety of cell types in vivo and in vitro (reviewed in Ref. 1). The biological effects of EGF are mediated by a receptor present in the plasma membrane of target cells (2- 4). The EGF receptor is an M, 180,000 transmembrane protein that is modified by glycosylation and phosphorylation (5-9). It has recently been shown that the receptor contains an

* This work was supported by Grants CA28146 andCA27951 from the National Cancer Institute. 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.

3 Supported by Postdoctoral Fellowship CA 06991 from the Na- tional Cancer Institute. Present address, Department of Pharmacol- ogy, Vanderbilt University, School of Medicine, Nashville, TN 37232.

The abbreviations used are: EGF, epidermal growth factor; DME, Dulbecco’s modified Eagle’s medium; 5’-p-FS02B~[’~C]Ado, 5 9 - fluorosulfonylbenzoyladenosine; EGTA, ethylene glycol bis(@-ami- noethyl-ether)-N,N,N’,N’-tetraacetic acid; NaDodS04, sodium do- decyl sulfate; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

intrinsic tyrosine-specific protein kinase activity that is stim- ulated several-fold following EGF binding (10-14). This EGF- stimulated protein kinase activity is thought to be important for EGF-induced cell proliferation, since the proliferative effects of a number of tumor viruses are mediated by tyrosine- specific protein kinases (15). However, the stimulation of protein kinase activity by EGF may not be sufficient for mitogenesis (16). In addition to stimulating receptor-associ- ated protein kinase activity, the binding of EGF to its receptor on target cells results in rapid endocytosis and lysosomal degradation of the hormone-receptor complex (17-19). This has led to speculation that proteolytic degradation products of the EGF receptor might act as intracellular signals for mitogenesis (18, 20) and to examination of the proteolysis products of EGF receptor labeled by cross linking to ‘T-EGF (21, 22). Such studies have defined a proteolytic domain structure of the EGF-binding portion of the receptor molecule, but the possible domain structure of other portions of the receptor has received little attention.

In this report, we use partial proteolytic digestion with trypsin to separate domains of the EGF receptor. These domains were then further characterized in order to map sites of EGF binding, glycosylation, and phosphorylation in the presence and absence of EGF. These studies have led to the localization of all the in vivo tyrosine phosphorylation sites within the terminal 15 kDa of the receptor and demonstrated new sites of phosphorylation in cells treated with EGF. In addition, we describe a new trypsin-resistant domain of the receptor which could function as a second messenger for EGF action.

EXPERIMENTAL PROCEDURES

Muterials”A431 human carcinoma cells were obtained from Dr. Stanley Cohen (Vanderbilt University) and were grown in DME supplemented with 10% calf serum. An antiserum that specifically precipitates the EGF receptor was prepared by the method of Haigler and Carpenter (23), using shed A431 membrane vesicles (12) as antigen. EGF, lZ5I-EGF grade, was purchased from KOR Biochemi- cals. Tosylphenylalanyl chloromethyl ketone-trypsin was purchased from Worthington. Trasylol was from FBA Pharmaceuticals. Chy- motrypsin, lentil lectin-agarose, and a-casein were from Sigma. Staphylococcal V-8 protease was from Miles. [35S]Methionine (1260 Ci/mmol) was from Amersham Corp. Carrier-free (32P)orthophos- phate and crude [ Y - ~ ~ P ] A T P were from ICN. 5’-p-FSOzBz[14CJAdo (43 mCi/mmol), D-[3H]-glucosamine (19 Ci/mmol), and ‘T-EGF (prepared using chloramine-T) were from New England Nuclear.

Metabolic Labeling-Cells were labeled at 37 ”C as follows: for [35S] methionine, 0.1 mCi/ml for 4 h in methionine-free DME; for [3H] glucosamine, 0.1 mCi/ml for 6 h in growth medium; for [3ZP]ortho- phosphate, 0.3-3.0 mCi/ml for 4 h in phosphate-free DME. EGF, when used, was present at 100-150 ng/ml for the entire labeling period. 32P-labeled EGF receptor from sparse or confluent cultures yielded identical tryptic peptide fragmentation patterns. However, since tyrosine phosphorylation may be inhibited at high cell density (9), sparse cultures were used for experiments involving phosphoa- mino acid analysis.

11534

EGF Receptor Domains 11535

Immunoprecipitation and Partial Proteolysis of the EGF Receptor- Cultures were placed on ice, washed three times with phosphate- buffered saline, and lysed for 10 min in buffer A (50 mM sodium phosphate, 5 mM EDTA, 1 mM EGTA, 10 mM NaF, 10 p~ Na3V0,, 1% Triton X-100, 10% glycerol, pH 7.4). After scraping the plates with a rubber "policeman," the lysates were transferred to centrifuge tubes and clarified for 30 min a t 50,000 X g. The EGF receptor was then immunoprecipitated by incubation for 30 min on ice using an excess of antiserum followed by an excess of formali-fixed Staphylo- coccus aureus (24). Following centrifugation, the bacterial pellet was washed several times with buffer A and resuspended in a small volume of buffer A. Aliquots of this suspension were incubated a t 25 "C for 10 min with an equal volume of a 2 X trypsin solution in water, yielding the final trypsin concentrations indicated in the text (final volume, usually 20 pl). The reaction was generally stopped by addition of Laemmli sample buffer (25).

In Vitro Phosphorylation Reactions-To examine tryptic cleavage of in vitro-phosphorylated EGF receptor, the receptor was immuno- precipitated from unlabeled cells as described above, and the bacterial pellet was washed and resuspended in buffer B (20 mM Hepes, 100 mM NaCI, 0.2% Triton X-lOO,lO% glycerol, pH 7.4). To this suspen- sion were added MnC12 (final 1 mM) and [-p3'P]ATP (15 pCi, final 9 p~ in 100 pl). After incubation for 5 min at 25 'C, the bacteria were washed three times with 1.0 ml of buffer B and resuspended in a small volume of buffer B. Aliquots of this suspension were treated with trypsin as described above.

To assay the ability of trypsin-treated receptor to phosphorylate casein, unlabeled receptor was immunoprecipitated and treated with trypsin described above, and the reaction was stopped by the addition of 2 pl of Trasylol to each 20-pl reaction mixture. 10 pg of casein were added to each sample, and phosphorylation reactions were carried out for 10 min a t 25 "C after addition of MnCI2 to 1 mM and [y-"PIATP to 1 p~ (10 pCi/sample). The reactions were stopped by boiling in Laemmli sample buffer, and the samples were analyzed by NaDodS0,-polyacrylamide gel electrophoresis on a 10% gel. Follow- ing staining, drying, and autoradiography, the labeled casein bands were excised from the gel and their "P content determined by liquid scintillation counting.

Lubeling with 5'-p-FSO&["C]Ado-Two 100-mm plates of cells were lysed in 2.0 ml each of 20 mM Hepes, 1 mM EGTA, 1% Triton X-100, 10% glycerol, pH 7.4, and the lysates were clarified as de- scribed above. The clarified lysates were incubated for l h on ice with 100 pl of packed lentil lectin-agarose beads with resuspension every 5 min. The beads were washed 3 times with buffer B, and receptor was eluted for 30 min on ice by addition of 300 pl of 0.4 M a-methyl mannoside in buffer B, with resuspension every 5 min. Following removal of the beads by centrifugation, the eluate was filtered through glass wool and transferred to a tube a t 25 "C. The following were then added to the eluate: 10 pl of 100 mM MnC12, 5 pl (500 ng) of EGF, and 35 pl of 5'-p-FS02Bz["C]Ado. The reaction was allowed to proceed for 1 h at 25 "C, following which 65-pl aliquots of the mixture were incubated for 10 min a t 25 "C with 10 pl of trypsin or water. Reactions were then stopped by boiling in Laemmli sample buffer.

Direct Linkuge of IzI-EGF to the Receptor-Chloramine-T-acti- vated '251-EGF was used to affinity label the receptor's EGF-binding site by a modification of a published procedure (26). 1.0 pCi of "'I- EGF in 30 pI of phosphate-buffered saline/O.l% bovine serum albu- min was added to 30 pl of 20 mM Hepes, pH 7.4, containing 150 pg of A431 membrane vesicle protein. Vesicles were prepared as de- scribed (12). Following a 3.5-h incubation a t 37 "C, the vesicles were solubilized in 1.0 ml of buffer A and the receptor was immunoprecip- itated as described above. Labeling of the receptor was blocked by the presence of unlabeled EGF in the reaction mixture.

NaDodS0,-Polyacrylamide Gel Electrophoresis and Fluorography- Samples were generally analyzed on 9% polyacrylamide gels (0.24% bis), except 10% gels were used to analyze casein phosphorylation and 7.5% gels were used to separate the 180- and 165-kDa forms of the EGF receptor in preparations for phosphoamino acid analysis. Preparative gels were dried directly following electrophoresis, while analytical gels were stained and destained before drying. Autoradiog- raphy of samples containing 32P or '=I was sometimes enhanced using Lightning Plus intensifying screens (General Electric), while samples containing "S, "C, or 3H were treated with 1 M salicylate (27) before exposure to Kodak XAR-5 or XRP-1 x-ray film. Molecular weight standards used were: from Sigma, myosin (200,000), phosphorylase b (94,000), bovine serum albumin (68,000), ovalbumin (43,000); from

Bethesda Research Laboratories, a-chymotrypsinogen (25,700), and 0-lactoglobulin (18,400).

Peptide Mapping-Partial proteolysis of receptor fragments during electrophoresis was carried out on 12.5% gels using the method of Cleveland et al. (28), omitting the incubation without current in the stacking gel.

Phosphoamino Acid Analysis-Following localization by autoradi- ograpy, labeled bands were excised from dried gels, eluted, subjected to partial hydrolysis in 6 N HCI, and analyzed by paper electrophoresis a t pH 3.5 as previously described (29). Our electrophoresis proceeds long enough to resolve uridine monophosphate (a contaminant of in uiuo-labeled material) from phosphotyrosine. However, as a precau- tion, all in uiuo-labeled samples found to contain phosphotyrosine by this procedure were also analyzed by two-dimensional paper electro- phoresis using the buffer systems of Hunter and Sefton (30).

RESULTS

Tryptic Domains of the EGF Receptor-In our initial exper- iment aimed at mapping the EGF receptor, we examined conditions under which trypsin might cleave the receptor into structural domains. The EGF receptor of A431 human carci- noma cells was metabolically labeled with [3sS]methionine and purified from cell lysates by immunoprecipitation with a specific antiserum. The receptor in the immune complex was then treated with increasing doses of trypsin, and the reaction products were analyzed by NaDodS04-polyacrylamide gel electrophoresis and fluorography (Fig. 1). Treatment of the 180-kDa receptor (Fig. 1, lane a) with trypsin generated fragments having apparent M, values of 16,500, 12,500, 4,000 and 3,000; the 125- and 30-kDa fragments were relatively resistant to further trypsin action (Fig. 1, lanes 6-4. The 40- and 30-kDa fragments were released from the immune com-

a b c d

41 2

440

430

FIG. 1. Partial tryptic digestion of [S5S]methionine-labeled EGF receptor. EGF receptor was immunoprecipitated from labeled cells, treated with trypsin in the immune complex, and analyzed by NaDodS04-polyacrylamide gel electrophoresis and fluorography as described under "Experimental Procedures." The receptor was treated with: a, 0; b; 1; c, 5; or d, 20 pg/ml of trypsin.

11536 EGF Receptor Domains

plex, while the 165- and 125-kDa fragments remained anti- body-bound (not shown). No further major fragments were observed upon increasing the trypsin concentration to 200 d m l .

The tryptic fragments observed in Fig. 1 appear to represent proteolytic domains of the EGF receptor since treatment with chymotrypsin or staphylococcal V-8 protease, enzymes with quite different specificities, generates fragments of similar size (not shown). The relative protease resistance of these domains is not dependent on antibody protection in the immune complex since we have obtained identical fragments following trypsin treatment of soluble EGF receptor purified from [“S]methionine-labeled A431 cells by chromatography on EGF-agarose or lentil-agarose (not shown).

Because estimates of the molecular mass of the intact human epidermal growth factor receptor have varied from 155 (9) to 186 kDa (31) and estimates of the molecular mass of a degradation product commonly found in membrane prepara- tions have varied from 145 (21) to 170 kDa (22), it is important to note that our 180-kDa receptor co-migrates on NaDodSO, gels with the intact EGF receptor band phosphorylated in uitro in shed A431 membrane vesicles (12), while our 165-kDa fragment migrates slightly more slowly than the 150-kDa EGF receptor band phosphorylated in uitro in A431 membranes prepared as described by Carpenter et al. (8).

Relationships between Major Tryptic Fragments of the EGF Receptor-An experiment was performed to explore the top- ographical relationships between the major tryptic fragments of the EGF receptor. Two possible arrangements are drawn schematically in Fig. 2a. 1) The 40-kDa fragment might arise by cleavage of the 165-kDa fragment to the 125-kDa fragment, with subsequent cleavage of the 40-kDa fragment to the 30- kDa fragment. 2) Alternatively, since the estimates of M, by NaDodS04-polyacrylamide gel electrophoresis could be inac- curate, the 40-kDa fragment might arise by cleavage of the 180-kDa receptor to the 125-kDa fragment while the 30-kDa fragment could originate by cleavage of the 165-kDa fragment to the 125-kDa fragment. The experiment presented in Fig. 2b was designed to distinguish between these two possibilities. EGF receptor, immunoprecipitated from cells labeled with [:“S]methionine, was subjected to treatment with various concentrations of trypsin in the immune complex (0, 1, 2.5, or 10 pglml). The four bacterial pellets were then washed to remove any 30- and 40-kDa fragments which had been formed. Each pellet was then resuspended and divided into another four aliquots, each of which was treated with a different concentration of trypsin (0, 1, 5, or 20 pglml). The reaction products of this second trypsin treatment were analyzed by

:aDodSO4-polyacry1amide gel electrophoresis and fluorogra- phy (Fig. 26). As expected (see Fig. 1). increasing doses of trypsin during the first treatment cleaved the 180-kDa recep- tor to 165-kDa and then 125-kDa fragments, and any 40-kDa and 30-kDa fragments generated a t this time were completely removed by washing the bacterial pellet (Fig. 2b, lanes I , 5 , 9 , and 13). The material in Fig. 2b, lanes 1,5,9, nd 13, represents the starting material for the second trypsin treatment. During this second treatment, the usual 165-, 125-, 40-, and 30-kDa fragments were generated from the sample which had been treated only with water during the first incubation (Fig. 26, lanes 1-4). As the trypsin dose was increased in the first incubation, the starting material for the second trypsin treat- ment contained lower and lower ratios of 180-kDa receptor to 165-kDa fragment (Fig. 2b, lanes 1 , 5, and 9). Nevertheless, the ratio of 40-kDa to 30-kDa fragment generated in the second incubation remained approximately constant (Fig. 26, lanes 3, 7, and 1 1 ) . This observation would seem to exclude

a I. 4 0 k

125k 3 0 k 15k

r ,- - ’ ’ l8Ok

165k

2. 4 0 k

125k 30 k

+ + I

180k

I 165k

1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 16

FIG. 2. Successive trypsin treatments of [S6S]methionine- labeled EGF receptor. a, two possible arrangements of the receptor and its tryptic fragments are drawn schematically. Arrows represent sites of tryptic cleavage. b, immunoprecipitated [%]methionine- treated EGF receptor in the immune complex was treated with 0 (lunes 1-4), 1 (lunes 5-8),2.5 (lunes 9-12), or 10 (lunes 13-16) pg/ml of trypsin as described under “Experimental Procedures.” Each of the four samples was then washed several times and divided into a further four aliquots, each of which was treated in a second digestion with 0 (lunes 1,5, 9, and 131, 1 (lunes 2 ,6 , 10, and 14), 5 (lunes 3, 7, 11, and 15), or 20 (lunes 4,8,12, and 16) pg/ml of trypsin as described under “Experimental Procedures.” Samples were then subjected to analysis by NaDodS0,-polyacrylamide gel electrophoresis and fluo- rography.

the second model in Fig. 2a. Further, since depletion of the 180-kDa receptor (Fig. 26, lanes 1 and 5 ) does not reduce the amount of 40- and 30-kDa fragments generated during trypsin treatment (Fig. 26, compare lanes 3 and 4 with lanes 7 and B), and since depletion of the 165-kDa fragment (Fig. 2b, lanes 1 , 5, 9, and 13) results in a progressive loss of the 40- and 30- kDa fragments (Fig. 2b, compare lanes 3 and 4 with lanes 7 and 8, lanes 11 and 12, and lanes 15 and 16), we conclude that the 40- and 30-kDa fragments both arise from cleavage of the 165-kDa receptor fragment to the 125-kDa fragment, as in model 1 described above. However, we cannot exclude the

EGF Receptor Domains 11537

possibility that the 40- and 30-kDa fragments are derived directly from the cleavage of a very small subpopulation of highly trypsin-resistant 180-kDa receptor molecules that be- have differently than the bulk of the receptor molecules shown in Fig. 2b.

Mapping of Functional Sites within Receptor Tryptic Do- mains-Using a variety of labeling techniques in conjunction with partial tryptic digestion of immunoprecipitated EGF receptor, we have mapped various functional sites of the receptor to its different tryptic domains (Fig. 3).

The EGF-binding site of the receptor was labeled by “direct linkage” to ‘2sI-EGF (Fig. 3a), and sites of receptor glycosy- lation were labeled by in vivo incorporation of [‘H]glucosa- mine (Fig. 36). In both cases, only the 165- and 125-kDa tryptic fragments of the receptor contained radiolabel. Similar results were obtained when the receptor was metabolically labeled with [‘H]galactose or [‘H]fucose (not shown). These results indicate that the sites of EGF binding and glycosyla- tion are contained within the 125-kDa domain of the receptor. This probably corresponds to the 115- to 130-kDa EGF- binding domain previously described (21, 22).

To identify the sites of phosphorylation i n uiuo, the receptor was labeled metabolically with [“P]orthophosphate (Fig. 3c). Only the 165-, 40-, and 30-kDa tryptic fragments of the “P- labeled receptor contained radiolabel; the 125-kDa domain is not phosphorylated under these conditions. Thus, the sites of receptor phosphorylation are contained within the 40-kDa tryptic domain. These sites of phosphorylation are examined in more detail in a later section.

To identify the sites of autophosphorylation in vitro, unla- beled EGF receptor in the immune complex was incubated in the presence of [y-”PIATP and Mn2+ (Fig. 3d). Upon subse- quent trypsin treatment, the bulk of the 32P incorporated in uitro was rapidly converted to low molecular weight material which migrates with the dye front. In addition, some radio- label was found in the 165-kDa receptor fragment, in a number of trypsin-labile 25- to 40-kDa fragments, and in two previ- ously undetected trypsin-resistant fragments of mass 25 and 27 kDa (Fig. 3d). These fragments may represent nonphysi-

a 1 2 3 4 b

1 2 3 4

ological sites of phosphorylation, since they are not observed upon trypsin treatment of receptor labeled with [ 3 2 P ] ~ r t h ~ - phosphate in living cells (Fig. 3c).

Because the sites of receptor phosphorylation observed i n vivo and i n vitro were quite different and because in uitro sites of phosphorylation might conceivably be quite sensitive to alterations in reaction conditions, we performed the i n vitro phosphorylation reactions under many different conditions. Very similar patterns of tryptic digestion were obtained from receptor phosphorylated using a variety of incubation times, temperatures, and ATP concentrations, using soluble EGF receptor purified by chromatography on EGF-agarose or lentil lectin-agarose, or using EGF receptor phosphorylated i n vitro in intact membrane vesicles prior to solubilization and im- munoprecipitation. Similar patterns were also obtained in the absence or presence of EGF. Thus the differences we have observed between EGF receptor phosphorylation sites i n uivo and in vitro are not due to arbitrarily chosen reaction condi- tions in vitro.

We have also examined the ATP-binding site of the EGF receptor, which Buhrow et al. (13, 14) have shown may be specifically affinity labeled in A431 membrane preparations using the ATP analogue ~ ’ - ~ - F S O ~ B Z [ ’ ~ C ] A ~ O . We have been able to specifically label immunoprecipitated EGF receptor with 5’-p-FS02Bz[’4C]Ado (not shown). However, since we found that the immunoprecipitated IgG heavy chain was nonspecifically labeled with 5’-p-FS02B~[’~C]Ado, we used lentil lectin-purified EGF receptor to map the receptor’s ATP- binding site (Fig. 3e). Following affinity labeling with 5”p- FS02Bz[’4C]Ado, the lectin-purified receptor was subjected to partial tryptic digestion, NaDodS04-polyacrylamide gel electrophoresis, and fluorography (Fig. 3e). Label was incor- porated into the 165-, 125-, 40-, and 30-kDa fragments, as well as some low molecular weight material. All labeling was prevented by prior heat inactivation of the receptor kinase activity (not shown). These observations suggest that the kinase ATP-binding site may involve more than one struc- tural domain of the EGF receptor. Alternatively, the labeling of more than one structural domain could indicate the pres-

C 1 2 3 4

440

430

d 1 2 3 4

e 1 2 3 4

L

440

430

FIG. 3. Partial tryptic digestion of EGF receptor labeled at various functional sites. Receptor was labeled with the following radioactive ligands as described under “Experimental Procedures”: a, ‘=I-EGF; b, [3H] glucosamine; c, [32P]orthophosphate; d, [-y-32P]ATP, e, ~ ’ - ~ - F S O ~ B Z [ ’ ~ C ] A ~ O . I n each part, receptor was then digested with 0, 1, 5, or 20 pg/ml of trypsin, in lanes 1-4, respectively, before being analyzed by NaDodS04- polyacrylamide gel electrophoresis and autoradiography.

11538 EGF Receptor Domains

ence of more than one ATP-binding site. Kinase Assay of the Trypsin-treated EGF Receptor-It re-

cently has been shown that the tyrosine-specific protein ki- nase activity of two different avian sarcoma virus-transform- ing proteins resides in a 30-kDa protease-resistant domain (32,33). In order to determine whether the 30-kDa protease- resistant domain of the EGF receptor might, by analogy, be active as a protein kinase, we examined the ability of immu- noprecipitated EGF receptor to phosphorylate casein follow- ing partial tryptic digestion. Phosphorylation of casein in this reaction occurred exclusively on tyrosine (data not shown). When the receptor was partially degraded to the 165-kDa fragment by treatment with 1 pg/ml of trypsin, 79% of the initial casein kidney activity was still observed. This is con- sistent with previous observations that a similar fragment is active as a protein kinase (12,21). However, when the receptor was treated with 5 or 20 pg/ml of trypsin, kinase activity was reduced to 6 and 0.7%, respectively, of control levels. When ["?3SJmethionine-labeled receptor was used as a source of ki- nase in order to permit monitoring of the extent of digestion, the loss of activity correlated with the cleavage of the 165- kDa fragment to the 125-, 40-, and 30-kDa fragments. Similar results were obtained using [Val']angiotensin I1 as a substrate, and none of the receptor fragments smaller than 165 kDa was capable of autophosphorylation. Similar results were also obtained using chymotrypsin in place of trypsin. The require- ment for the entire 165-kDa fragment for kinase activity is consistent with, but does not prove, the hypothesis that the kinase ATP-binding site could involve portions of both the 125- and 40-kDa domains (above). Thus, the EGF receptor differs from certain viral tyrosine kinases that have protease- resistant catalytic domains.

Effect of EGF on Sites of Receptor Phosphoylation-The analysis of the partial tryptic peptides of "P-labeled EGF receptor shown in Fig. 3, c and d, suggested that the sites of phosphorylation were all clustered within one region of the molecule. However, since EGF stimulates the phosphorylation of its receptor both in vitro (8) and in vivo (9), it was of interest to examine the distribution of phosphate in receptor labeled in the presence of EGF. Others have reported that EGF treatment increases the total phosphate incorporation into receptor but does not increase the number of sites of receptor phosphorylation in vivo (9) or in vitro (31). These previous studies examined total tryptic peptides of the recep- tor by two-dimensional analysis on thin layer plates. As expected from the results of Gates and King (31), we did not observe any differences in the pattern of partial tryptic phos- phopeptides (analyzed by NaDodS04-polyacrylamide gel elec- trophoresis) when lectin-purified EGF receptor was phospho- rylated in vitro in the presence or absence of EGF (not shown).

In contrast, the pattern of partial tryptic phosphopeptides from receptor labeled with ["P]orthophosphate in vivo was markedly altered when cells were exposed to EGF during the labeling period (Fig. 4). In this experiment, the EGF receptor was immunoprecipitated from clarified lysates of control or EGF-treated cells, treated with increasing doses of trypsin, and analyzed by NaDodS0,-polyacrylamide gel electropho- resis and autoradiography as above for Fig. 3c. EGF treatment caused a substantial stimulation of receptor phosphorylation (Fig. 4, lanes a and h). The 165-, 40-, and 30-kDa phospho- rylated tryptic fragments of receptor from untreated cells (Fig.4, lanes b-d) were also detected after trypsin treatment of receptor from EGF-stimulated cells; however, the 30- and 40-kDa phosphopeptides released from the receptor upon treatment with 5-20 pg/ml of trypsin were altered in migra- tion following EGF treatment (Fig. 4, lanes e-g). The 40-kDa

4 6 3

4 4 8

4 4 0 c

4 3 0

FIG. 4. Effect of EGF on the pattern of partial tryptic phos- phopeptides from [3ZP]orthophosphate-labeled receptor. EGF receptor was immunoprecipitated from cells labeled with ['*P]ortho- phosphate in the absence(lanes a-d) or presence (lanes e-h) of EGF. Receptor in the immune complex was then treated with 0 (lanes a and h), 1 (lanes b and g), 5 (lanes c and f), or 20 ( l anes d and e ) , pgl ml of trypsin and analyzed by NaDodS0,-polyacrylamide gel electro- phoresis and autoradiography.

tryptic fragment from EGF-treated receptor migrated more slowly than that from untreated receptor, and the 30-kDa fragment migrated as a doublet, the lower band of which co- migrated with the 30-kDa fragment from control cells. In addition, three new fragments of mass 125, 63, and 48 kDa were observed when receptor from EGF-treated cells was treated with trypsin (Fig. 4, lane g).

In occasional experiments, the 63 and 48-kDa fragments, but not the 125-kDa fragments, have been generated from receptor from untreated cells. We were initially concerned that some of these fragments might come from the minor 200- kDa phosphoprotein that contaminates our immunoprecipi- tates. However, experiments in which the "P-labeled bands were excised from the gel and subjected to liquid scintillation counting indicated that the 200-kDa contaminant did not contain enough counts to account for the new fragments from EGF-treated cells. Further, phosphorylation of this band, unlike generation of these fragments, is not affected by EGF in most experiments.

As a first approach to analyzing the relationships between the various phosphopeptides, all of the 32P-labeled receptor fragments shown in Fig. 4 were mapped by partial proteolysis during electrophoresis in the presence of chymotrypsin or staphylococcal V-8 protease. Unfortunately, in several such experiments, only the 40- and 30-kDa fragments generated well resolved cleavage maps. The intact receptor and 165 and 125-kDa fragments were poorly digested, while the 48- and 63-kDa fragments formed only indistinct smears upon further proteolysis under these conditions. Fig. 5 shows the partial proteolytic peptide maps which resulted from digestion of the 40- and 30-kDa phosphopeptides from control and EGF- treated cells with chymotrypsin or staphylococcal V-8 pro- tease during electrophoresis. All of these 30- and 40-kDa phosphopeptides appear to related, since they all share com- mon cleavage products. This is consistent with our conclusion, drawn from the results shown in Fig. 26, that the 30-kDa

EGF Receptor Domains 11539

a b C d e 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

FIG. 5. Mapping of the 30- and IO-kDa phosphorylated re- ceptor fragments by partial proteolysis during electrophore- sis. The 30- and 40-kDa fragments of 32P-labeled EGF receptor from control or EGF-treated cells were excised from a preparative gel similar to that shown in Fig. 4. The gel pieces containing these fragments were then analyzed by partial proteolysis during electro- phoresis as described under “Experimental Procedures.” Present dur- ing the electrophoresis were: (a ) 100 ng of staphylococcal V-8 pro- tease, ( b ) 10 ng of V-8 protease, (c) no protease, ( d ) 50 ng of chymotrypsin, or (e) 500 ng of chymotrypsin. The proteins analyzed in each panel were: 1 , the 40-kDa fragment from control cells; 2, the 40-kDa fragment from EGF-treated cells; 3, the 30-kDa fragment from control cells; 4, the lower band of the 30-kDa doublet from EGF- treated cells; and 5, the upper band of the 30-kDa doublet from EGF- treated cells.

fragment is derived from the 40-kDa fragment. It is interest- ing that, using either V-8 protease or chymotrypsin in the Cleveland mapping procedure, fragments of very similar size are produced, including a cleavage ofthe 40-kDa fragments to 30 kDa. Because these enzymes have different specificities for cleavage, both from each other and from trypsin, the results imply that considerable tertiary structure remains intact in both the 40- and 30-kDa fragments even following treatment with 2% NaDodSO,. Thus, the structure of this portion of the EGF receptor molecule is very stable.

In addition to the common cleavage products of the various fragments, new phosphopeptides are generated from the 40- and 30-kDa fragments from EGF-treated cells (Fig. 5). This suggests the possible existence of new sites of phosphorylation of these tryptic fragments in EGF-treated cells. To further explore the possibility of new sites of receptor phosphoryla- tion in EGF-treated cells, we examined the phosphoamino acid composition of each of the major tryptic fragments of the receptor labeled with [32P]orthophosphate in EGF-treated or untreated cells. The intact receptor contained phosphoserine, phosphothreonine, and phosphotyrosine whether labeled in the presence or absence of EGF; the relative proportions of the various phosphoamino acids were similar to those previ- ously reported by Hunter and Cooper (9) (70% phosphoserine, 20% phosphothreonine, 10% phosphotyrosine). The 165-, 40-, and 30-kDa fragments contained both phosphoserine and phosphothreonine, but not phosphotyrosine in either control or EGF-treated cells. These observations suggested that the site of tyrosine phosphorylation in vivo is located on the end of the receptor that is removed upon cleavage to the 165-kDa fragment. When we analyzed the new fragments which were specifically phosphorylated in the presence of EGF, phospho-

threonine alone was found in the 125-kDa fragment, while primarily phosphoserine and traces of phosphothreonine were found in the 63- and 48-kDa fragments. Trace amounts of phosphotyrosine were also detected in the 63-kDa fragments, but we are not certain that the low amounts of phosphotyro- sine recovered were not due to contamination by a minor nonspecific band seen in our immunoprecipitates.

The simplest interpretation of these data is that EGF induces phosphorylation of new sites on the receptor, includ- ing a threonine phosphorylation site on the 125-kDa domain that is not phosphorylated a t all in control cells. Alternatively, it is conceivable that EGF-enhanced phosphorylation of re- ceptor sites phosphorylated a t low levels in control cells alters the sensitivity of various areas of the receptor to tryptic cleavage, leading to the generation of the new fragments.

Recovery of Phosphotyrosine-containing Peptides of the EGF Receptor-Because all phosphotyrosine was lost from the EGF receptor upon cleavage ofthe 180-kDa receptor to a 165- kDa fragment, we tentatively concluded that the site(s) of tyrosine phosphorylation were located in the 15-kDa region that is removed during this cleavage and degraded (above). Alternatively, since we had not actually recovered phospho- tyrosine from trypsin-treated receptor material, it seemed possible that the loss of phosphotyrosine from the receptor during trypsin treatment could have been due to the presence of a phosphotyrosine-specific phosphatase in our trypsin prep- aration. Therefore, it was necessary to perform an experiment to test this possibility before concluding that sites of tyrosine phosphorylation are removed from the receptor by proteolytic cleavage to the 165-kDa fragment. Such an experiment is presented in Fig. 6.

Cells were labeled with [32P]orthophosphate in the presence of EGF to maximize tyrosine phosphorylation of the receptor. EGF receptor was then immunoprecipitated from the labeled cells, and the bacterial-bound immune complexes were resus- pended in a small volume of ammonium bicarbonate buffer. Aliquots of this suspension were then incubated in the absence or presence of trypsin for 10 min, and the reaction was stopped by the addition of Trasylol. At this time, the bacteria were sedimented, and aliquots of the material that had been re- leased from the immune complexes into the supernatant were either subjected directly to NaDodS04-polyacrylamide gel electrophoresis or were lyophilized, hydrolyzed with 6 N HCl, and analyzed for phosphoamino acids by two-dimensional paper electrophoresis. Material that had remained bound to the bacteria in the immune complex was eluted with Na- DodSO,, and aliquots of this eluate were either subjected directly to NaDodS04-polyacrylamide gel electrophoresis or were precipitated with trichloroacetic acid, hydrolyzed with 6-N HCl, and analyzed for phosphoamino acids by two- dimensional paper electrophoresis.

Analysis of aliquots of these samples by NaDodS0,-polya- crylamide gel electrophoresis is presented in Fig. 6a. In the absence of trypsin, the phosphorylated receptor remains bound in the immune complex (Fig. 6a, lane 3); no 32P-labeled material was detectable in the supernatant (Fig. 6a, lane I). Following trypsin treatment, all phosphorylated receptor ma- terial is released from the immune complex (Fig. 6a, lane 4 ) and is detectable in the supernatant as the 30-kDa receptor fragment (Fig. 6a, lane 2) and as small peptides at the dye front (not shown; the dye front was run off of this gel).

Two-dimensional phosphoamino acid analyses of aliquots from the same samples are presented in Fig. 6, 6 and c. Analysis in the first dimension at pH 1.9 was performed to separate phosphoserine from phosphothreonine and phospho- tyrosine (Fig. 66). The phosphothreonine/phosphotyrosine-

11540 EGF Receptor Domains

a 1 2 3 4

1 2 = 3 4 m 0 P-thr

136

e e P”

0 0

FIG. 6. Determination of phosphoamino acids present in the total reaction mixture following trypsin treatment of in vivo-phosphorylated EGF receptor. EGF receptor was immunoprecipitated from cells labeled with [3ZP]orthophosphate in the presence of EGF and washed as described under “Experimental Procedures.” Following a further wash in 50 mM ammonium bicarbonate and resuspension in a small volume of the same buffer, the receptor was treated with 0 or 20 pg/ml of trypsin as described under “Experimental Procedures.” The reaction was stopped by the addition of Trasylol, and the bacterial pellet was collected by centrifugation. Aliquots of the supernatant were boiled in Laemmli sample buffer and subjected to NaDodS0,-polyacrylamide gel electrophoresis or were lyophilized and analyzed for phosphoamino acids as described under “Experimental Procedures.” Material in the immune complex was eluted from the bacteria with Laemmli sample buffer and either subjected to NaDodS0,- polyacrylamide gel electrophoresis or to trichloroacetic acid precipitation and phosphoamino acid analysis. a, NaDodS0,-polyacrylamide gel electrophoresis and autoradiography. Lune I , material released from the immune complex in the absence of trypsin; fane 2, material released from the immune complex in the presence of trypsin; fane 3, material remaining antibody bound in the absence of trypsin; lane 4, material remaining antibody bound in the presence of trypsin. 6, pH 1.9 electrophoresis and autoradiography of partial acid hydrolysates. Numbering of lanes as in a; positions of origin and marker phosphoamino acids are indicated. c, pH 3.5 electrophoresis and autoradiography of the combined phosphothreonine/phosphotyrosine spots excised from the paper shown in b; numbering of lanes is as in a.

containing spot was subjected to a second electrophoresis a t pH 3.5 to separate these phosphoamino acid species (Fig. 6c) . As expected, no phosphoamino acids were recovered from the material released from the immune complex in the absence of trypsin (lane I) or in the material remaining antibody-bound following trypsin treatment (lane 4) ; these samples lacked phosphoproteins as analyzed by NaDodS0,-polyacrylamide gel electrophoresis (Fig. 6a). However, similar ratios of free phosphate, phosphoserine, phosphothreonine, and phospho- tyrosine were found in the hydrolysates of intact receptor that remained immune complex bound in the absence of trypsin (lane 3) and of the 30-kDa and low molecular mass receptor fragments released from the immune complex following tryp- sin treatment (lane 2). Approximately 5% of the total phos- phoamino acids (phosphoserine plus phosphothreonine plus phosphotyrosine) present in each of these two samples was recovered as phosphotyrosine. Since the 30-kDa receptor frag- ment does not contain phosphotyrosine (above), we believe that the phosphotyrosine recovered in the trypsin-treated sample must be derived from small peptides present in this sample. We conclude that our trypsin preparation is not contaminated with a phosphotyrosine-specific phosphatase, since trypsin treatment does not result in appreciable hydrol- ysis of phosphotyrosine; identical percentages of phosphoty- rosine are recovered from untreated and trypsin-treated re- ceptor. It appears, therefore, that the loss of phosphotyrosine from the receptor when it is cleaved to the 165-kDa fragment

is due to the proteolytic removal of the phosphotyrosine- containing receptor peptides, which are too small to be re- solved on the 9% NaDodSO, gels presented above.

DISCUSSION

The experiments described herein confirm and extend pre- vious findings suggesting a domain structure for the EGF receptor. A 115- to 130-kDa protease-resistant EGF-binding domain has previously been identified (21,22); we now show that this domain also contains sites of glycosylation and is phosphorylated on threonine in EGF-treated cells. The pos- sible function of this EGF-activated phosphorylation is not known, but the existence of EGF binding, glycosylation, and phosphorylation sites on the same receptor fragment suggests that this fragment traverses the plasma membrane. We have also identified a 40-kDa domain of the EGF receptor that contains serine and threonine phosphorylation sites. This domain appears to retain considerable tertiary structure even following electrophoresis in NaDodSO, gels, based on the peptide-mapping experiment presented in Fig. 5; the structure of this receptor fragment may thus be very stable. Finally, we have presented data suggesting that the terminal 15-kDa fragment of the receptor, known to be protease labile, contains the in uiuo and in vitro sites of tyrosine phosphorylation. A working model of the EGF receptor based on the current results is presented in Fig. 7.

It is of interest that the 165-kDa receptor fragment, which

EGF Receptor Domains 11541

l 25k 40 k 15k

*P P P P

e..... ATP7

FIG. 7. Proteolytic domains of the EGF receptor. This figure represents a summary of the structural data on the sites of EGF binding, glycosylation, and in vivo incorporation of phosphate (P) on serine (Ser) , threonine (Thr), or tyrosine ( T y r ) residues. *Prepresents a site phosphorylated only when cells are incubated with EGF. The order of sites within each domain is not known.

is active as a protein kinase, lacks the major site(s) of tyrosine phosphorylation. This contrasts with the situation found for the viral tyrosine kinase transforming proteins which contain the tyrosine acceptor site within the catalytic domain (32- 34) . Indeed, phosphorylation at tyrosine has been suggested to play an important role in activating the insulin receptor tyrosine-specific protein kinase activity (35), and Purchio et al. have presented evidence that phosphorylation of pp60"'"n a tyrosine residue within the amino half of this molecule causes a 5- to &fold increase in enzymatic activity (36). Such phosphorylation may not be important for the kinase activity associated with the EGF receptor. It is also of interest that the protein kinase activity of the EGF receptor does not appear to be contained within one structural domain; portions of both the 125- and 40-kDa domains are required for activity. This contrasts with the observation that the tyrosine kinase activity of three viral transforming proteins is contained within a protease-resistant domain (32-34).

Linsley and Fox previously observed that when A431 cell membranes were phosphorylated in vitro using [-p3'P]ATP and then treated with trypsin, 32P-labeled peptides of 30-40 kDa were generated (21). Although these peptides might have been derived from any of a number of 32P-labeled membrane proteins, their size encouraged speculation that they were derived from the EGF receptor by the mechanism diagrammed in Fig. 2a of this paper (model 2). We have found similar peptides released after trypsin treatment of crude membrane fractions labeled in uitro. However, it is unlikely that the phosphopeptides formed under these conditions correspond to the 30- and 40-kDa receptor fragments we have described here, since these fragments are not phosphorylated in vitro under a wide variety of conditions. It is likely that the peptides derived from phosphorylated membrane preparations are de- rived from nonreceptor phosphoproteins and that their simi- larity in size to the receptor domains we have described here is coincidental.

Our data concerning the phosphoamino acid content of EGF receptor fragments are in apparent conflict with data reported by Carlin and Knowles (37). These authors found that when EGF receptor metabolically labeled with ["PI orthophosphate was cleaved by cellular proteass following cell lysis, the removal of a 20-kDa fragment from the receptor resulted in the loss of phosphoserine from the receptor, but not of phosphothreonine or phosphotyrosine. We report here that a similar cleavage with trypsin removes all phosphoty- rosine from the receptor, but not phosphoserine or phospho- threonine. The reasons for these different results are not clear, but could involve the different methods for analyzing phosphoamino acids, as well as the different buffers used for cell lysis and the different proteases acting on the EGF receptor.

We believe that partial cleavage with trypsin as described here may be of general use in mapping studies of the EGF

receptor. We have shown the use of this technique in mapping sites of EGF binding, glycosylation, phosphorylation, and ATP binding. In addition, this technique could be used to map regions of the receptor to which various anti-receptor antibodies bind or regions within which new covalent modi- fications or functional activities might be located.

It is intriguing to speculate that the 30-kDa trypsin-resist- ant receptor fragment described here could be biologically active and might be generated in living cells upon transport of the internalized EGF-receptor complex into lysosomes. EGF-stimulated tyrosine phosphorylation may produce the early biological effects of EGF but does not appear to be sufficient to induce DNA synthesis (16). As postulated by Das and Fox, a fragment released from the EGF receptor by the action of lysosomal proteases might act as a second messenger to induce DNA synthesis (18, 20). The relative protease resistance of the 30-kDa receptor domains makes it an attrac- tive candidate for such a second messenger. Experiments are in progress to deterine whether this fragment is present in EGF-treated cells.

Acknowledgments-We are grateful to Sandi Burns and Phyllis Leder for excellent secretarial assistance.

REFERENCES 1. Carpenter, G., and Cohen, S. (1979) Annu. Rev. Biochem. 48,

2. Hollenherg, M. D., and Cuatrecasas, P. (1973) Proc. Natl. Acnd. Sci. U. S. A. 70,2964-2968

3. Carpenter, G., Lembach, K. J., Morrison, M. M., and Cohen, S. (1975) J. Biol. Chern. 250,4297-4304

4. Schreiber, A. B., Lax, I., Yarden, Y., Eschhar, Z., and Schlessin- ger, J. (1981) Proe. Natl. Acad. Sci. U. S. A. 78,7535-7539

5. Das, M., Miyakawa, T., Fox, C. F., Pruss, R. M., Aharanov, A,, and Herschman, H. R. (1977) Proc. Natl. Acad. Sci. U. S. A.

6. Carpenter, G., and Cohen, S. (1977) Biochem. Biophys. Res.

7. Hock, R. A., Nexo, E., and Hollenherg, M. D. (1979) Nature

8. Carpenter, G., King, L., Jr., and Cohen, S. (1979) J. Biol. Chem.

9. Hunter, T., and Cooper, J. A. (1981) Cell 24, 741-752

193-216

74,2790-2794

Commun. 79,545-552

(Lond.) 277, 403-405

254.4884-4891

10. Cohen, S., Carpenter, G., and King, L., Jr. (1980) J. Bid. Chem.

11. Ushiro, H., and Cohen, S. (1980) J. Biol. Chem. 255,8363-8365 12. Cohen, S., Ushiro, H., Stoscheck, C., and Chinkers, M. (1982) J.

13. Buhrow, S. A., Cohen, S., and Staros, J. V. (1982) J . Bid. Chern.

14. Buhrow, S. A., Cohen, S., Garbers, D. L., and Staros, J. V. (1983) J. Bwl. Chem. 258,7824-7827

15. Bishop, J. M., and Varmus, H. (1982) in RNA Turnor Viruses (Weiss, R., Teich, N., Varmus, H., and Coffin, J., eds) pp. 999- 1108, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York

16. Schreiber, A. B., Yarden, Y., and Schlessinger, J. (1981) Biochem. Bwphys. Res. Cornmun. 191,517-523

17. Carpenter, G., and Cohen, S. (1976) J. Biol. Chem. 71,159-171 18. Das, M., and Fox, C . F. (1978) Proc. Natl. Acad. Sei. U. S. A . 75,

255,4834-4842

Biol. Chern. 257, 1523-1531

257,4019-4022

2644-2648 19. Haigler, H. T., McKanna, J. A., and Cohen, S. (1979) J. Bioi.

Chem. 81,382-395 20. Fox, C. F., and Das, M. (1979) J. Supramol. Struct. 10, 199-214 21. Linsley, P. S., and Fox, C. F. (1980) J. Supramol. Struct. 14,

22. O'Keefe, E. J., Battin, T. K., and Bennett, V. (1981) J. Supramol.

23. Haigler, H. T., and Carpenter, G . (1980) Biochirn. Biophys. Acta

24. Kessler, S. W . (1975) J. Imrnunol. 115, 1617-1624 25. Laemmli, U. K. (1970) Nature (Lord.) 227, 680-685

461-471

Struct. Cell. Bwchem. 15, 15-27

598,314-325

11542 EGF Receptor Domains 26. Comens, P. G., Simmer, R. L., and Baker, J. B. (1982) J. Bwl.

27. Chamberlain, J. P.11979) Anal. Biochem. 98,132-135 28. Cleveland, D. W., Fischer, S. G., Kirschner, M. W., and Laemmli,

U. K. (1977) J. Biol. Chem. 252,1102-1106 29. Brugge, J. S., and Darrow, D. (1982) Nature (Lond.) 2 9 5 , 250-

253 30. Hunter, T., and Sefton, B. M. (1980) Proc. Nutl. Acad. Sci. U. S.

31. Gates, R. E., and King, L. E., Jr. (1982) Biochem. Biophys. Res.

Chem. 267,42-45

A. 77,1311-1315

Commun. 105,57-66

32. Brugge, J. S., and Darrow, D. (1984) J. Bwl. Chem. 259, 4550-

33. Levinson, A. D., Courtneidge, S. A., and Bishop, J. M. (1981)

34. Weinmaster, G., Hinze, E., and Pawson, T. (1983) J. Virol. 46,

35. Rosen, M., Herrera, R., Olowe, Y., Petrozzelli, L. M., and Cobb,

36. Purchio, A. F., Wells, S. K., and Collett, M. S. (1983) Mol. Cell.

37. Carlin, C. R., and Knowles, B. B. (1982) Proc. Nutl. Acad. Sci. U.

4557

Proc. Natl. Acad. Sci. U. S. A. 78, 1624-1628

29-41

M. H. (1983) Proc. Nutl. Acad. Sci. U. S. A. 80,3237-3240

BWl. 3,1589-1597

S. A. 79,5026-5030