Identification and biochemical characterization of novel putative ...

Vol. 267, No. 8,Issue of'March 15, pp. 5155-5161,1992 Prmted in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Identification and Biochemical Characterization of Novel Putative Substrates for the Epidermal Growth Factor Receptor Kinase* (Received for publication, September 16, 1991)

Francesca FazioliS, Donald P. Bottaro, Liliana Minichiello, Albert0 Auricchio, William T. Wong, Oreste Segatto, and Pier Paolo Di FioreQ From the Laboratory of Cellular and Molecular Biology, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892 ~~

To gain insight into the mechanisms which control the mitogenic response to epidermal growth factor (EGF), we have partially purified and characterized several intracellular proteins which are phosphorylated on tyrosine residues following activation of the epidermal growth factor receptor (EGFR).Partial purification wasachieved by immunoaffinity chromatography using immobilized anti-phosphotyrosine antibodies. Antisera generated against the partially purified proteins were used to identify at least five novel EGFR putative substrates, designated, on the basis of their apparent molecular weight, p97, p68, p61, p56, and p23. All of these proteins became specifically phosphorylated on tyrosine after EGF treatment of intact cells, as assessed by phosphoamino acid analysis, and none of them represented an EGFR degradation product. The phosphorylation of these proteins appeared to be relatively specific for the EGFR. In particular, an EGFR-related kinase, erbB-2 was much less efficient than EGFR at phosphorylating p97, p56,and p23 and incapable of phosphorylating p68. The identification of these novel EGFR putative substrates should lead to a better understanding of the mechanisms controlling the specificity of EGFR-mediated mitogenic signaling.

Several peptide growth factors regulate cellular growth, metabolism, and differentiation by binding to cell surface receptor-tyrosine kinases (1-4). This interaction starts a cascade of events, most likely involving receptor dimerization/ oligomerization (4 and references therein), which culminate in dramatically enhanced receptor catalytic activity (1-4). Active receptor tyrosine kinases are capable of autophosphorylation and the phosphorylation of a number of intracellular substrates (1-4). While receptor autophosphorylation appears to be important in determining the affinity and/or the specificity of the kinase for its cellular substrates (5-8), tyrosine phosphorylation of the latteris thought to represent the initial step of intracellular mitogenic signal transduction. Several receptor-tyrosine kinase substrates have been identified, including the isozyme of phospholipase C (phospholipase C - r ) (9-13), the p2lra.s GTPase activatingprotein(14-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Recipient of an Associazione Italiana per la Ricerca sul Cancro fellowship. § T o whom correspondence and reprint requests should be addressed: Laboratory of Cellular and Molecular Biology, National Cancer Institute, National Institutes of Health, Bldg. 37, Rm. 1D23, Bethesda, MD 20892. Tel.: 301-496-5277.Fax: 301-496-8479.

16), the raf serine-threonine kinase (17, 18), and the p85 subunit of the phosphatidylinositol 3-kinase (19-25). Early attempts to identify receptor-specific signal transduction pathways indicated that a diverse repertoire of effector molecules are used by different growth factor receptors. For instance, the platelet-derived growth factor receptor phosphorylates phospholipase C - r (10, 11) andGTPase activating protein (6, 14, 16) on tyrosine residues; however, other tyrosine kinase receptors do not (14, 26). Therefore, it appears that in addition to receptor expression and/or ligand availability, the specificity of substrate recognition influences the cellular responses to a peptide growth factor. We are interested in elucidating the mechanisms of mitogenic signal transduction by receptors belonging to the er6 subfamily (27). In our previous work, we have demonstrated that two closelyrelated members of this family, the epidermal growth factor receptor (EGFR)' and the product of the er6B2 gene, g~185"'*~-', exerttheir effects on the cell through different pathway(s) (28-31). In fact, they show qualitative and quantitative differences in their ability to activate mitogenic pathways in different target cells under comparable conditions of expression and enzymatic activation (27, 30). Biochemical analysis of known mitogenic transduction pathways, however, failed to reveal any major difference in the ability of EGFR or gp18TbB-'to induce tyrosine phosphorylation of phospholipase C-y, GTPase activating protein, or raf (13). In addition, we detected low stoichiometry of tyrosine phosphorylation (11%of the total pools) of these substrates by both EGFR and gp18YbB.'(13). Furthermore, agenetically engineered EGFR mutant which exhibited dramatically reduced mitogenic activity but no alterations of its intrinsic phosphotransferase activity, was indistinguishable from the wild type EGFR as to its ability to tyrosine phosphorylate phospholipase C-y or GTPase activating protein (32). These results prompted us to hypothesize that other additional signal transduction pathways might exist which are responsible for the different biological effects of members of the erb receptor subfamily. The activation of such alternative signaling pathways should involve early tyrosine phosphorylation of yet unidentified intracellular proteins by active erb receptors. The present studies were undertaken in an attempt to isolate and characterize novel species which are tyrosinephosphorylated following EGFR kinase activation. EXPERIMENTALPROCEDURES

Cell Culture-Genetically engineered NIH-3T3 cells overexpressing EGFR or the chimeric EGFRlerbB-2 receptor (NIH-EGFR and 'The abbreviations used are: EGFR, epidermal growth factor receptor; TK, tyrosine kinase; pTyr, phosphotyrosine; HEPES, 4-(2hydroxyethy1)-1-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethy1enenitrilo)Jtetraacetic acid; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis.

5155

5156

EGFR Substrates

NIH-EGFRlerbB-2, respectively) were described previously (13, 28, 33). These cell lines expressed around 1.5 X lo6 EGFRs or EGFR/ erbB-2s per cell, respectively (13). Cells were maintained in Dulbecco's modified Eagle's medium (GIBCO) supplemented with 10% calf serum (GIBCO). For EGF triggering experiments, subconfluent cell monolayers were incubated for 18 h in Dulbecco's modified Eagle's medium supplemented with transferrin (5 pg/ml; Collaborative Research) and selenium (lo-' M; Sigma) in the absence of serum and then treated with EGF (Upstate Biotechnology) as described in the text. Immunoajjinity Chromatography-Cellswerelysed on ice with buffer containing 1%Triton X-100 (Pierce), 10 mM Tris-HC1 (pH 7.61, 5 mM EDTA, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 100 p~ sodium orthovanadate, 2 mM phenylmethylsulfonyl fluoride, and 50 pg/ml aprotinin (lysis buffer). EGFR was removed from lysates (50 mg of protein) by anti-EGFR affinity chromatography at 4 "C. Anti-EGFR columns were prepared by covalently crosslinkinga monoclonal antibody (Abl, Oncogene Science) directed against the extracellular domain of the EGFR to agarose beads using the Antigen Immobilization kit from Pierce (1 mg of antibody/lO ml of gel). To assess EGFR removal, the columns were eluted with 0.1 M glycine (pH 2.8), and theeluate was neutralized with 50 mM TrisHCl (pH 9.5) and analyzed by SDS-PAGE and immunoblotting as described below. Proteins that did not bindto the anti-EGFR columns were applied to ananti-phosphotyrosine (anti-pTyr) column (3 ml of agarose-bound monoclonal anti-pTyr; Oncogene Science) a t 4 "C. Samples were recycledover the column several times over a period of 2 h. The column was washed with 50 column volumes of lysis buffer, then eluted using 6 mlof lysis buffer supplemented with 10 mM phenyl phosphate. Fractions were collected and analyzed for pTyr content by dot-immunoblot analysis. Dot-Zmmunoblot Analysis-Samples (2 11) of each fraction eluted from theanti-pTyr column were diluted to 200 pl with a buffer containing 30 mM Tris-HC1 (pH 7.6) and 250 mM glycine and boiled for 5 min. Two-fold dilutions of each sample were then spotted onto nitrocellulose filters (pre-equilibrated with buffer) using a microwell filtration apparatus. Wells were washed several times with buffer, and pTyr-containing proteins were detected with an anti-pTyr antibody coupled to '251-proteinA, as described (34). In this and all subsequent immunodetections, three different anti-pTyr antibodies were used a polyclonal serum generated according to Panget al. (35), and two commercially available monoclonal antibodies (Upstate Biotechnology and PY20, ICN), with similar results. The specificity of the immunodetections was confirmed by performing parallel immunostainings in which the anti-pTyrantibodies were absorbed with excess phosphoserine or phosphothreonine (which did not inhibit immunorecognition) or phosphotyrosine (which inhibited immunorecognition; data not shown). Rabbit Immunization-The immunoaffinity-purified pTyr-containing proteins were used to immunize two New Zealand white rabbits as follows. pTyr-containing proteins (25 pg) were mixed with an equal volume of complete Freund's adjuvant and injected subcutaneously, near the inguinal lymph nodes. Subsequent injections (10 pg of antigen in incomplete Freund's adjuvant) were administered subcutaneously 4 weeks after the first injection and were repeated at 3-week intervals. Bleeds were collected 1 2 days after each boost and screened for the production of antibodies. Protein Analysis-Antisera were tested for their ability to recognize putative EGFR substrates by immunoprecipitation from cells metabolically labeled with [32P]orthophosphateand treated with EGF in uiuo. For labeling in vivo with [32P]orthophosphate, serum-starved cells were washed three times with phosphate-free Dulbecco's modified Eagle's medium and incubated for 4 h at 37 "C with 6 ml of the same medium containing 2.5 mCi of [32P]orthophosphate(Du Pont/ New England Nuclear Research, 9000 Ci/mmol). Cells were then treated with EGF (100 ng/ml) for 30 min at 4 "C. Following radiolabeling and EGF stimulation, cell were lysedwith a buffer containing 1%Triton X-100 (Pierce), 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1.5 mM MgC12,5 mM EGTA, 10 mM sodium orthovanadate, 4mM phenylmethylsulfonyl fluoride, 20mM sodium pyrophosphate,andaprotinin (100 pg/ml). Immunoprecipitation was performed for 2 h at 4 "C, and immune complexes were recovered by adsorption to Gamma Bind G-agarose (Genex Corp.). After several washes with buffer containing 0.1% Triton X-100, 20 mM HEPES, 10% glycerol, and 150 mM NaCl, SDS-PAGE sample buffer (30% glycerol, 5% SDS, 0.1 M Tris(pH 6.8), and 0.01% bromphenol blue) was added, and samples were boiled for 5 min. Immunoprecipitated

radiolabeled proteins were then analyzed directly by SDS-PAGE and autoradiography. In some experiments, unlabeled proteins were separated by SDSPAGE, transferred to nitrocellulose filters, and then detected either with colloidal gold (Gold Blot, ISS-Enprotech) or with specific antisera for immunodetection with '251-proteinA as described (34). The antisera used were: two different anti-EGFR antibodies, Ab-1 and E7, a rabbit anti-peptide polyclonal sera recognizing the EGFR carboxyl-terminal tail (residues 1172 to 1186, Ref. 30), and the three different anti-pTyr antibodies described above. Phosphoamino Acid Analysis and Peptide Mapping-Subconfluent cell monolayers in 15-cmdishes were serum-starved and metabolically labeled in uiuo as described above with 1.2 mCi/ml [32P]orthophosphate. Cells were then treated with EGF (100 ng/ml) for 30 min at 4 "C. Proteins were immunoprecipitated from cell lysates as described above, separated by SDS-PAGE, and analyzed by autoradiography. Individual bands were excised from the gels and subjected to phosphoamino acid analysis or V8 protease peptide mapping. Phosphoamino acid analysis was performed as described previously (36). Briefly, peptides were eluted from gel slices by incubation with 200pgof trypsin (tosylphenylalanyl chloromethyl ketone-treated trypsin, Worthington Biochemicals) in 50 mM ammonium bicarbonate (pH 8.0) for 18 h. Eluted material was dried under vacuum and subjected to acid hydrolysis for 2 h at 110 "C using 100 p1 of 6 M HC1 (Pierce, constant boiling grade). HC1 was evaporated under vacuum, and hydrolysates were washed three times with water by repeated vacuum centrifugation, then resuspended in a mixture of phosphoserine, phosphothreonine, and phosphotyrosine (1mg/ml each phosphorylated aminoacid) and subjected to thin layer electrophoresis. Two-dimensional analysis was performed at pH 1.9 for 60 min at 1 kV (first dimension) and at pH 3.5 for 50 min a t 1 kV (second dimension); one-dimensional analysis was performed for 60 min at pH 3.5. Phosphorylated amino acid standards were revealed by ninhydrin staining. V8 protease peptide mapping was performed by the method of Cleveland et al. (37). Individual gel slices were placed in a single well of a 15% SDS-PAGE gel in the presence of 1.25 mg of Staphylococcus V8 protease (Boehringer Mannheim), and gels were typically run for 20 h at 15 mA. The gels were dried and subjected to autoradiography at -70 "C.

RESULTS

Optimization of Substrate Phosphorylation by the EGFR i n Vivo-Our initial efforts were aimed at identifying an optimal system to study tyrosine phosphorylation of cellular substrates by EGFR. Since our final goal was the purification of these proteins, our main concern was to develop an approach yielding high stoichiometry of tyrosine phosphorylation that would allowquantitative recovery of EGFR substrates. Inour previous studies, we have shown that overexpression of EGFR in the normally EGF-responsive NIH/3T3 cell line, even at levels of 2-3 X lo6 receptors/cell (NIH-EGFR cells), did not alter the growth properties of the cells, under standard culture conditions (13, 28, 31). In the presence of EGF, however, NIH-EGFR cells displayed markedly increased DNA synthesis, as compared to mock-transfected NIH/3T3 cells (13, 28, 31). Thus, overexpression of the EGFR amplifies the EGFtriggered mitogenic signal inthese cells. These biological findings were paralleled, at the biochemical level, by a dramatic increase in tyrosine phosphorylation of "putative" EGFR substrates, when NIH/3T3 and NIH-EGFRcells were analyzed under conditions of EGF stimulation(13). We therefore elected to use the NIH-EGFR model system for our experiments. The extent of phosphorylation of the putative EGFR substrates was dependent on the experimental conditions of exposure to EGF i n uiuo. As shown in Fig. 1, an EGF doseresponse analysis indicated that a concentration of50-100 ng/ml (8-16 nM) induced optimal tyrosine phosphorylation of putative substrates in uiuo. Longer exposures of the gel shown in Fig. 1 revealed that differences in substrate phosphorylation over a wide range of EGF concentrations were

EGFR Substrates EG F (ng/ml)

5157

encompasses the major EGFR autophosphorylation site. The E7 serum also recognizes with comparable efficiency pTyrcontaining or the unphosphorylated EGFR(Ref. 30, see also Fib. 2B). Thus, the lack of recognition by the E7 serum of -200 proteins co-migrating with the putative substrates shown in EGFR -B degradaFig. 2 A , demonstrates that the latter are not EGFR -97 tion products. -68 The reasonfor the higher efficiency of substrate phosphorylation at 4 "C, as compared to37 "C, is unclear. At the -43 lower temperature, lack of receptorinternalizationand/or -29 decreased phosphatase(s) activitymay result inmore efficient receptor-activated intracellular phosphorylations. In case, any - 18 the specificity of signal transduction (as assessed by the - 14 nature of the phosphorylated proteins) did not appear to be affected under optimal conditions (EGF stimulation at 100 FIG. 1. Tyrosinephosphorylation of cellularproteinsin more physiological NIH-EGFR cells: EGF dose-response. Serum-starved cells were ng/ml, a t 4 "C for 30 min) as compared to either mock-treated ( 0 lane) or treated with different concentrations conditions (see Figs. 1 and 2 A ) . Therefore, cell lysates were of EGF for 10 min a t 37 "C before lysis. Total cellular proteins (200 obtained under optimal conditionsof EGF stimulationfor all fig) were separated by SDS-PAGE on a 3-27% acrylamide gradient subsequent purification steps. gel and analyzed by anti-pTyr immunoblotting. Molecularweight Purification of Putative EGFR Substrates-Partial purifimarkers ( X lo-") are indicated a t right. The position of tyrosinephosphorylated EGFR is indicatedby the arrow. Comparable results cation of pTyr-containing proteinswas achievedby anti-pTyr were obtained in three different experiments. immunoaffinity chromatography. Prior to affinity purification on anti-pTyr columns, the lysates were depletedof EGFR through chromatography on an anti-EGFR affinity column. A B As evident, in fact, from the immunoblots in Figs. 1 and 2, 37oc 4oc 37oc 4oc theEGFRrepresentedthe major tyrosine-phosphorylated "" protein present in the starting material. Since our final goal y;GF0 1 5 30 1 5 10 15 30 45 0 1 5 30 1 5 1 0 1 5 3 0 4 5 was the generation of polyclonal antibodies against the puri-200fied proteins (see below), we elected to reduce the amount of CEGFR- 97 EGFR which would have represented the major immunogen -68in the preparation. To prepare the anti-EGFR affinity col-43. umn, a monoclonalantibody directed against the extracellular -29domain of EGFR was conjugated to agarose beads (see "Ex- 18 . - 14 . perimental Procedures"). Total cellular proteins from NIHEGFR cells treated with EGF under optimal conditions (50 anti-PTyr anti-EGFR mg from each of four independent preparations) were then FIG. 2. Tyrosinephosphorylation of cellular proteins in cycled over four sequential anti-EGFR affinitycolumns. We NIH-EGFR cells: timecourse of EGF stimulation. Serumremoved more than 80% of the EGFR starved cells were either mock-treated ( 0 lane) or treated with EGF, estimated that this step from the cell lysates. A possible concern regarding this pro100 ng/ml, for the indicated times before lysis. EGF triggering was performed at 37 "C or 4 "C as indicated. Cell proteins (200 pg) were cedure was that receptor-bound substrates might have been fractionated by electrophoresis on a 3-27% acrylamide gradient gel removed from the protein preparations aswell. To check for and then subjected to immunoblot analysis using anti-phosphotyrothis possibility, we eluted the material bound to the antisine ( A ) or EGFR antipeptide seraE7 ( B ) .The position of the 170kDa EGFR is indicated by an arrow. Molecular weight markers (X EGFR column with glycine, 0.1 M,pH 2.8, and analyzed it by lo-:') are indicated between panels A and B. The results shown are immunoblot with anti-pTyr antibodies. As shown in Fig. 3A, representative of three separate experiments. onlythe 170-kDa EGFR specieswas detectedunderour condition of analysis. columns more quantitative than qualitative (data not shown). In fact, Proteins which did not bind to the anti-EGFR were then loaded on an anti-pTyr column (see "Experimental at doses as low as 1-5 ng/ml (0.16-0.8 nM), at which mostly high affinity binding sites would be occupied a t equilibrium, Procedures"), and the pTyr-containing proteins were finally eluted with 10 mM phenyl phosphate.A dot-blot immunoassay the pattern of phosphotyrosine (pTyr)-containing proteins was developed (see "Experimental Procedures") to analyze observed was very similar to that detected by stimulation with eluted fractions for phosphotyrosine content. The fractions higher EGFdoses (data not shown). Thus, it appears that the recruitment of a high number of EGFRs does not markedly with the highest phosphotyrosine content were then pooled and analyzedby SDS-PAGEandconventionalanti-pTyr affect thespecificity of its enzymaticactivity. immunoblotting. As shown in Fig. 3B, a t least 10 putative Fig. 2A shows a time course of EGF triggering (at the optimal dose of 100 ng/ml) a t 37 "C and 4 "C. We consistently substrates were eluted from the anti-pTyrcolumn, whichwere observed higher efficiency of tyrosine phosphorylation of pu- conspicuously enriched when compared to the starting matetative substrates at the latter temperature.signal The detected rial (Fig. 3B). We estimated that this preparation was enin immunoblot with anti-pTyr antibodies was maximal after riched in pTyr-containing proteins a t least 500-1000-fold. 30 min of EGF treatment at 4 "C, then decreased. Again, Recovery was estimated to be -0.06% of the total starting differences in the pattern of pTyr-containing proteinsseemed material; this value isin good agreement with previously of cells stimt o be more quantitative than qualitative. In the same cell published estimates of the pTyr protein content lysates, only the 170-kDa EGFR band was detectable by a n ulated by growth factors or T K oncogenes (38, 39). Blots identical with the oneshown in Fig. 3B were stained EGFR-specific antipeptide serum (E7 serum, Ref. 30) directed against the carboxyl-terminal tail of the receptor(Fig. 2B). It with AuroDye gold stain (see "Experimental Procedures") to should be noted that theepitope recognized by the E7 serum detect non-pTyr-containing proteins contaminating our prep0

,

-1-5._.-10-50 100-

" " "

.

" " " " "

EGFR Substrates

5158

A

1

2

”

3

4

B

”

B

A

1 2 3 4

450

w

v

97 -

97 -

68-

68-

43-

I

PI

E7

PI

I

E7

-EGFR

-

68

-

43

-

97

-

68

-

43

-

4329 -

29

-

18-

1814-

18

-

FIG. 3. Purification of tyrosine-phosphorylated proteins using anti-EGFR and anti-phosphotyrosine affinity chromatography. Panel A, total cellular proteins (50 mg for each of four independentpreparations)prepared from EGF-stimulatedNIHEGFR cells were loaded on foursuccessive anti-EGFRreceptor columns as described under “Experimental Procedures,” to remove the EGFR. The anti-pTyr immunoblot of a typical cell lysate (100 pg) before application to the anti-EGFR column is shown in lane 1 of panel A. Proteins bound to the anti-EGFR columns were eluted with 0.1 M glycine (pH 2.8), and fractions were analyzed by immunoblotting with anti-pTyr antibodies (lanes 2,3, and 4 of panel A are representative of the entire fractionated eluate). Panel B, proteins that did not bind to the anti-EGFR columnwere then applied to an anti-pTyr column, and the pTyr-containing proteins were purified as described under “Experimental Procedures.” Three differentpurified preparations (eachfrom 50 mg of starting material)of proteins eluted from the anti-pTyr column are shown in lunes 2-4. An aliquot (500 ng) of each preparation was fractionated on a 3-27% gradient acrylamide gel and then analyzed by anti-pTyr immunoblotting (lanes 2, 3, and 4 of panel B ) . Lane I shows a typical pTyr-containing protein pattern before the purification procedure (100 pg of total cellular areindicated at the left proteins). Molecularweight markers (X of each panel. The position of tyrosine-phosphorylated EGFR and tyrosine-phosphorylated putative EGFR substrates are indicated by arrowheads at theleft and right sides of panel B, respectively.

I

-

97

29 -

14-

PI

200

200EGFR-.

I

451

”

”

“ ”

PI

200EGFR-,

450

451

FIG. 4. Immunoprecipitation using sera raised against putative EGFR substrates. Lysates prepared from EGF-treated, “’Plabeled cells were immunoprecipitated with sera (450 and 451) obtained from rabbits after immunization( I )or with the corresponding preimmune sera (PI),or with the polyclonal anti-EGFR sera (15’7). Immunoprecipitated proteins were resolved by 10% ( A ) or 7.5% ( B ) SDS-PAGE. Molecular weight markers (X lo-’) are indicated a t the left of each panel. The position of tyrosine-phosphorylated EGFR and phosphorylated proteins specifically recognized by the immune sera are indicated by the arrowheads. Comparable results were obtained in three separate experiments.

As shown in Fig. 4, A and B, serum 451 also recognized several major phosphoproteins. Resolution oflow (Fig. 4A) and high (Fig. 4B) molecular weight species was obtained by varying the acrylamide percentage and lengthsof run in our SDS-PAGE analyses. This allowed us toidentify seven major phosphoproteins recognized by serum 451: p109, p97, p72, p68, p56, p50, and p23, respectively. Of note, none of the phosphoproteins recognized by serum 450 or 451 was efficiently detected in parallel immunoprecipitations with the corresponding preimmune serum(Fig. 4, A and B). The antiEGFR peptide antibody E7 (30) did not recognize any of these phosphoproteins, thus demonstrating that they do not reprearations. Furthermore, lysates obtained from cells metaboli- sent EGFR degradation products (Fig. 4, A and B ) . In addically labeled with [“S]cysteine and[35S]methionine were tion, Staphylococcus aureus V8 proteolytic patterns of individsubjected to the same purification protocol outlined above ual proteins did not match that of EGFR, further proving and then analyzed by SDS-PAGE. After comparing the anti- that the identified putative substrates are not EGFR degrapTyr immunoblots with the gold-stained blots and the 35S- dation products (data not shown). labeling experiments, we estimated that co-purified nonT o prove that the identified putative substrates were spepTyr-containing proteins represented only a minor compo- cifically phosphorylated on tyrosineresidues following EGFR nent ( 4 0 % ) of our preparations (data not shown). activation, phosphoamino acid analysis of the proteins recBiochemical Characterization of EGFR PutativeSubognized by the two sera was performed, before and after EGF strates-To furthercharacterizetheputativeEGFRsubtreatment of NIH-EGFR cells in uiuo. As shown in Fig. 5, in strates obtained as describedabove, we used the purified quiescent, unstimulated cells, all of the putative substrates protein preparations to generate polyclonal antibodies. Two were phosphorylatedonserineand/orthreonine residues, New Zealand rabbits were immunized (see“Experimental whereas there was little if any detectable pTyr. EGF treatProcedures”), and the immune sera (henceforth referred to as ment led to specific tyrosine phosphorylationof p97, p68, p61, sera 450 and 451) were tested for their ability to recognize p56, and p23, whereas little, if any, pSer or pThr increase was phosphoproteins. Lysates fromcells that had been metaboli- observed (Fig. 5). No tyrosine phosphorylation of p72 was cally labeled with [32P]orthophosphate in the presence of EGF detectable, even after growth factorstimulation(datanot were immunoprecipitated with either serum 450 or 451 or shown), whereas phosphoamino acid analysis of p109 and p50 with the corresponding nonimmune sera (prebleeds from the was not obtained due to the low level of ”P incorporation. same animals). As shown in Fig. 4, both sera were able to Specificity of Substrate Tyrosine Phosphorylation by specifically recognize a number of phosphoproteins. Both sera EGFR-We compared the ability of the closely related EGFR recognized a M, = 170,000 species, which was demonstrated and erbB-2 kinases to stimulate the phosphorylation of the by immunodepletion experiments to be EGFR(datanot newly identified substrates. Tothisend, we utilized two shown). Serum 450 also recognized a major phosphoprotein previously described mass populations of NIH-3T3 cells ovof an apparent molecular weight of 61,000 (p61), andseveral erexpressing either thewild type EGFR or anEGFR/erbB-2 other minor phosphoproteins among which was ap23 thatco- chimera containing the extracellular ligand-binding domain migrated with a similar proteinrecognized by serum 451 (Fig. of the EGFR and the intracellular domain of gp185e‘bB-2 (13, 4, A and B). 33). These two cell lines (NIH-EGFR and NIH-EGFRlerbB-

EGFR Substrates

+

-

EGF

5159

+

-

N

" "

PS" PT-

EGF

n n -

+ - +

""

u u EGFR

P97

+

-

+

-

EGF

97 -

f

68 -

+ c p68

p97

+-p56 43 I

I

~ 6 1

I

I

P23

FIG. 5. Phosphoamino acid analysis of putative EGFR substrates. NIH-EGFR cells were metabolically labeled with ['*P]orthophosphate and either mock-treated (-) or treated with EGF (+) as described in the text. After immunoprecipitation of radiolabeled cell lysates with immune sera, phosphoproteins were separated by SDSPAGE, excised from the gels, and digested with trypsin. Elutedtryptic peptides were hydrolyzed, and phosphoamino acids were separated by one (upper panels)- or two (lower panels)-dimensional thin layer electrophoresis. Phosphoamino acids were detected by autoradiographyand identified using standardsstained withninhydrine: p S , phosphoserine;pT, phosphothreonine;pY, phosphotyrosine.p97, p68, p56, and p23 were immunoprecipitated by sera 451. p61 was immunoprecipitated using sera450. EGFR was immunoprecipitated with a monoclonal EGFR antibody. Results are typical and representative of two independent experiments.

2, respectively) bothexpressaround 1.5 x lo6 EGFRs or EGFRlerbB-2s per cell, respectively, and exhibit similar numbers of high ( K d 10"' M) or low (Kd lo-' M) affinity EGF binding sites (13, 33). Under these conditions, therefore, the cellular responses to EGF aresolely dependent on the ability of either kinase to couple with signaling pathways. Lysates were prepared from [3*P]orthophosphate-labeled NIH-EGFR or NIH-EGFRlerbB-2 cells and quantitatively immunoprecipitated with anti-pTyr antibodies. pTyr-containing proteins were then eluted with phenylphosphate, immunoprecipitated asecond time with serum 451, and resolved by SDS-PAGE. Stimulation with EGF was performed under either optimal conditions (100 ng/ml EGF at4 "C for 30 min) or at physiological temperature (100 ng/ml EGF at 37 "C for 10 min). The data presented refer to the optimal conditions, but comparable results were obtained at physiological temperature. As shown in Fig. 6, stimulation of NIH-EGFR cells with EGF yielded the typical pattern of pTyr-containing proteins recognized by serum 451 (indicated by arrowheads). Of note, under these experimental conditions,there was some anti-pTyr recovery of p72 (indicated by an unlabeled arrowhead), a protein in which we failed to detect any pTyr in a phosphoamino acid analysis. At the present, we do not know whether p72 contains, after EGF stimulation, low levels of pTyr which escaped detection by a phosphoamino acid analysis or whether its anti-pTyr recovery is due to interaction with other pTyr-containing proteins. Under identical conditions of EGF triggering, NIH-EGFRlerbB-2 cells exhibited quantitativeand qualitative differences in thepattern of

29 4

p23

FIG. 6. Comparison of substrates phosphorylation by EGFR and an EGFRlerbB-2 chimera. Lysates were prepared from snPlabeled NIH-EGFR and NIH-EGFRlerbB-2cells treated with (+) or without (-) EGF (100 ng/ml for 30 min at 4 "C). Equal amounts of trichloroacetic acid-precipitable counts (10' cpm, corresponding to about 1.0 mg of total protein) were immunoprecipitated with excess anti-pTyr antibodies. The immunoprecipitates were then eluted with phenyl phosphate (10 mM), immunoprecipitated a second time with serum 451, and analyzed by discontinuous gradient SDS-PAGE (top, 7.5%; bottom, 12.5%). Molecular weight markers (X lo-') are indicated a t left. The positions of tyrosine-phosphorylated EGFR substrates recognized by sera 451 are indicated by arrowheads at right. Comparableresults were obtained in three separate experiments. Comparable results were also obtained when the EGF triggering was performed at physiological temperature (100 ng/ml EGF for 10 min at 37 "C).

pTyr-containing proteinsrecognized byserum 451, when compared to NIH-EGFR. As shown in Fig. 6, p97, p56, and p23 seemed to be phosphorylated less efficiently by an active erbB2 kinase than by EGFR, more importantly,there was no detectable p68 phosphorylation upon EGF stimulation of NIH-EGFRlerbB-2. DISCUSSION

Second messenger pathways have been implicated as early mitogenic signal transducers for TK growth factor receptors, particularly the pathwaysactivated by phospholipase C-7 tyrosine phosphorylation/L-a-phosphatidylinositol 4,5-diphosphate breakdown and by p2lrmlGTPase activating protein signaling (9-16,40-42). There is evidence, however, that while these pathwaysmay benecessary, they are notsufficient for receptor-induced mitogenic action. For example, overexpression of phospholipase C-y and subsequent platelet-derived growth factor receptor activation does not lead to increased mitogenic signaling, despite enhanced L-a-phosphatidylinositol 4,5-diphosphate turnover (43). In the case of EGFR-activated mitogenic pathway(s), we have previously shown that signal transductionmechanism(s) otherthan phospholipase C-y tyrosine phosphorylation/L-a-phosphatidylinositol 4,5-diphosphate breakdown andthep2lrasl GTPase activating protein signaling must contribute to the mitogenic response. A 7-amino acid deletion in the EGFR

5160

EGFR Substrates

(EGFR A660-667), in fact, caused a marked decrease in the mitogenic potency of EGFR despite unaltered kinase activity in vitro and in vivo (32). Inparticular, the efficiency of tyrosine phosphorylation of phospholipase C - r in vivo by the EGFR A660-667 was indistinguishable from that of wild type EGFR and, under the conditions of overexpression achieved, no GTPase activating protein tyrosine phosphorylation was detectable by either EGFR or EGFR A660-667 (32). The present studies were, therefore, undertaken to identify alternate signaling pathway(s) which contribute to theEGFRmediated mitogenic response. We approached this question by isolating intracellular proteins phosphorylated on tyrosine upon activation of the EGFR kinase. By taking advantage of polyclonal sera generated against the purified proteins, we were able to identify five proteins, p97, p68, p61, p56,and p23 which were specifically phosphorylated on tyrosine residues immediately upon EGF stimulation. None of these proteins was recognized by anti-peptide sera directed against the carboxyl-terminal domain of the EGFR, and V8 protease mapping confirmed that none were degradation products of the EGFR. These proteins do not appearto correspond to several of the known substrates for tyrosine kinase receptors. By immunoprecipitation, immunodepletion, and immunoblotting techniques, we have determined that p97 is notGTPase activating protein, that none of the newly identified species is phospholipase C - 7 , and thatp68 is notthe raf kinase (data not shown). There have also been reports that members of the src family of tyrosine kinases can be phosphorylated on tyrosine by tyrosine kinase growth factor receptors (44-46). However, the recognition of p68, p61, and p56 by the polyclonal sera 450 and 451 wasnot abolished in immunodepletion experiments performed with anti-src, anti-fgr, or anti-fyn sera (data not shown). Based on migration in SDS-PAGE, it is also unlikely that any of the proteins we have described are the 85-kDa subunits of phosphatidylinositol 3-kinase (23-25), although this warrants further investigation. Together, these data indicate that we have identified novel proteins which are phosphorylated on tyrosine upon EGFR activation and are likely to be part of the EGF-activated signaling pathway, At present, wedo not know whether these proteins directly interact with the EGFR, thereby being phosphorylated or whether they represent “second line” substrates whose phosphorylation is the consequence of secondary tyrosine kinase activation. For this reason, we refer to themasputative substrates. It is to be noted, however, that the kinetics of phosphorylation are consistent with the fact that these proteins represent early direct substrates of the EGFR k’lnase. To achieve quantitative tyrosine phosphorylation of substrates for purification purposes, it was necessary to work under conditions of receptor overexpression (approximately 2 x lo6EGFR per cell). This raises the question of whether the identified pTyr-containing species are physiologically part of the EGFR-specific mitogenic pathway or are recruited as a result of kinase promiscuity associated with overexpression. At least two lines of evidence indicate that this is not the case. First, we have observed that NIH/3T3 cells overexpressing EGFR at various levels (from 2 X 104/cellto 2 X 106/cell) display comparable ED50 values for EGF-induced mitogenesis, despite increased maximal response to EGF with increasing receptor numbers (32). This indicates that theabundance of substrates critical for mitogenesis is not limiting, and that at high levels of EGFR expression the mitogenic stimulus is still routed through a physiological pathway. In addition, we have found that increased EGFRs recruitment, obtained by increasing either EGF doses or the time of exposure, resulted in the phosphorylation of a pattern of substrates similar to

that detected under conditions in which only a few thousand receptors are activated. Together this evidence indicates that the purified pTyr-containing proteins are physiologic substrates of the EGFR. We have previously reported biological evidence that the mitogenic signaling pathways activated by EGFR and gp185”bB-2 must be at least in part different (28,31,32). These two highly related kinases, in fact, exhibited markedly different abilities to trigger mitogenic response in different target cells. In particular,in the NIH-3T3 system, the erbB-2 kinase was at least 100-foldmore potent than EGFR as a transforming gene (13, 28). In our initial efforts to identify differential coupling of these two kinases to mitogenic pathways, however, we did not detect any difference in their relative ability to phosphorylate known signal transducers, including phospholipase C - r and GTPase activating protein (13). In contrast, in this study, we found that, under comparable conditions of activation, the erbB-2 kinase is much less active than EGFR at phosphorylating p97, p56, and p23 and cannot efficiently phosphorylate p68. Thus, our results provide a biochemical correlate for the differential mitogenic coupling of EGFR and erbB-2. Several questions remain as to the functions of p97, p68, p61,p56, and p23.Answers are likely to comefrom the elucidation of their primary sequences, from gene transfer experiments using eukaryotic expression vectors for cDNAs encoding the novel substrates, and from studies of whether monoclonal antibodies directed against these proteins will block the EGF-activated mitogenic pathway in microinjection experiments. In an effort to address these issues, we have screened an NIH-3T3 cell bacterial expression library with sera 450 and 451, and we are presently characterizing a number of cDNAclones. Initial sequence information has revealed no homology with sequences present in GenBank or in the EMBL data base.’ While more work will be needed to unequivocally establish that we have cloned the cDNAs for p97, p68, p61, p56-, and p23, these studies may leadto a better understanding of the relevance of these proteinsin mediating EGF mitogenic actions. Acknowledgments-We are indebted to S. A. Aaronson for continuous support and encouragement. We also thank E. Appella and S. Ullrich for critically reviewing the manuscript. REFERENCES 1. Carpenter, G. (1987) Annu. Reu. Biochem. 66,881-914 2. Yarden, Y., and Ullrich, A. (1988) Annu. Reu. Biochem. 57,443478 3. Williams, L. T. (1989) Science 2 4 3 , 1564-1570 4. Ullrich, A., and Schlessinger, J. (1990) Cell 6 1 , 203-212 5. Anderson, D., Koch, C. A., Grey, L., Ellis, C., Moran, M. F., and Pawson, T. (1990) Science 260,979-981 6. Kazlauskas, A., Ellis, C., Pawson, T., and Cooper, J. A. (1990) Science 247,1578-1581 7. Koch, C. A., Anderson, D., Moran, M. F., Ellis, C., and Pawson, T. (1991) Science 262,668-674 8. Margolis, B., Li, N., Koch, A., Mohammadi, M., Hurwitz, D. R., Zilberstein, A., Ullrich, A., Pawson, T., and Schlessinger, J. (1990) EMBO J. 9,4375-4380 9. Margolis, B., Rhee, S. G., Felder, S., Mervic, M., Lyall, R., Levitzki, A., Ullrich, A,, Zilberstein, A., and Schlessinger, J. (1989) Cell 57,1101-1107 10. Meisenhelder, J., Suh, P., Rhee, S. G., and Hunter, T. (1989) Cell 67,1109-1122 11. Wahl, M. I., Olashaw, N. E., Nishibe, S., Rhee, S. G., Pledger, W. J., and Carpenter, G. (1989) Mol. Cell. Biol. 9, 2934-2943 12. Wahl, M. I., Nishibe, S., Kim, J. W., Kim, H., Rhee, S. G., and Carpenter, G. (1990) J. B i d . Chem. 266, 3944-3948 P. P.Di Fiore and F. Fazioli, unpublished results.

EGFR Su:bstrates 13. Fazioli, F., Kim, U.-H., Rhee, S. G., Molloy, C. J., Segatto, O., and Di Fiore, P. P. (1991) Mol. Cell. Bwl. 11,2040-2048 14. Molloy,C. J., Bottaro, D. P., Fleming, T. P., Marshall, M. S., Gibbs, J. B., and Aaronson, S. A. (1989) Nature 3 4 2 , 711-714 15. Ellis, C., Moran, F. M., McCormick, G. F., and Pawson, T. (1990) Nature 343,377-381 16. Kaplan, D. R., Morrison, D. K., Wong, G., McCormick, F., and Williams, L. T.(1990) Cell 61, 125-133 17. Morrison, D.K., Kaplan, D.R., Rapp, U., and Roberts, T. M. (1988) Proc. Natl. Acad. Sci. U. S. A. 8 5 , 8855-8859 18. Morrison, D. K., Kaplan, D.R., Escobedo, J. A., Rapp, U.R., Roberts, T. M., and Williams, L. T. (1989) Cell 68,649-657 19. Coughlin, S. R., Escobedo, J. A., and Williams, L. T. (1989) Science 243,1191-1194 20. Kazlauskas, A., and Cooper, J. A. (1989) Cell 58,1121-1133 21. Varticovski, L., Druker, B., Morrison, D., Cantley, L., and Roberts, T. (1989) Nature 3 4 2 , 699-702 22. Ruderman, N. B., Kapeller, R., White, M. F., and Cantley, L. C. (1990) Proc. Natl. Acad. Sci. U. S. A. 8 7 , 1411-1415 23. Escobedo, J. A., Navankasattusas, S., Kavanaugh, W. M., Milfay, D., Fried, V. A., and Williams, L. T. (1991) Cell 6 5 , 75-82 24. Skolnik, E. Y.,Margolis, B., Mohammadi, M., Lowenstein, E., Fischer, R., Drepps, A., Ullrich, A., and Schlessinger, J. (1991) Cell 6 6 , 83-90 25. Otsu, M., Hiles, I., Gout, M., Fry, M. J., Ruiz-Larrea, F., Panayatou, G., Thompson, A., Dhand, R., Hsuan, J., Toffy, N., Smith, A. D., Morgan, S. J., Courtneidge, S. A., Parker, D. J., and Waterfield, M. D. (1991) Cell 66, 91-104 26. Downing, J. R., Margolis, B. L., Zilberstein, A., Ashmun, R. A., Ullrich, A., Sherr, C. J., and Schlessinger, J. (1989) EMBO J. 8,3345-3350 27. Hanks, S. K., Quinn, A. M., and Hunter, T. (1988) Science 2 4 1 , 42-52 28. Di Fiore, P. P., Pierce, J. H., Fleming, T. P., Hazan, R., Ullrich, A., King, C. R., Schlessinger, J., and Aaronson, S. A. (1987) Cell 5 1 , 1063-1070 29. Di Fiore, P. P., Pierce, J. H., Kraus, M. H., Segatto, O., Kink, C.

5161

R., and Aarsonson, S. A. (1987) Science 2 3 7 , 178-182 30. Di Fiore, P. P., Segatto, O., Lonardo, F., Fazioli, F., Pierce, J. H., and Aaronson, S. A. (1990) Mol. Cell. Biol. 1 0 , 2749-2756 31. Di Fiore, P. P., Segatto, O., Taylor, W. G., Aaronson, S. A., and Pierce, J. H. (1990) Science 2 4 8 , 79-83 32. Segatto, O., Lonardo, F., Wexler, D., Fazioli, F., Pierce, J. H., Bottaro, D. P., White, M. F., and Di Fiore, P. P. (1991) Mol. Cell. Bwl. 11,3191-3202 33. Lonardo, F., Di Marco, E., King, R. C., Pierce, J. H., Segatto, O., Aaronson, S. A,, and Di Fiore, P. P. (1990) New Biol. 2 , 9921003 34. Segatto, O., Lonardo, F., Pierce, J. H., Bottaro, D. P., and Di Fiore, P. P. (1990) New Biol. 2,187-195 35. Pang, D.T., Sharma, B.R., and Shafer, J. A. (1985) Arch. Bwchem. Biophys. 2 4 2 , 176-186 36. Haring, H. V., Kasuga, M., White, M. F., Crettaz, M., and Kahn, C. R. (1984) Biochemistry 23,3298-3306 37. Cleveland, D. W., Fischer, S. G., Kirschner, M. W., and Laemmli, U. K. (1977) J. Biol. Chem. 262,1102-1106 38. Hunter, T., and Sefton, B. M. (1980) Proc. Natl. Acad. Sci. U. S. A. 77,1311-1315 39. Kanner, S. B., Reynolds, A. B., and Parsons, J. T. (1989) J. Immunol. Methods 120, 115-124 40. Mulcahy, L. S., Smith, M. R., and Stacey, D. W. (1985) Nature 313,241-243 41. Smith, M. R., Ryu, S. H., Suh, P. G., Rhee, S. G., and Kung, H. F. (1989) Proc. Natl. Acad. Sci. U. S. A. 86,3659-3663 42. Smith, M. R., Liu, Y. L., Kim, H., Rhee, S. G., and Kung, H. F. (1990) Science 2 4 7 , 1074-1077 43. Margolis, B., Zilberstein, A., Franks, C., Felder, S., Kremer, S., Ullrich, A., Rhee, S. G., Skorecki, K., and Schlessinger, J. (1990) Science 248,607-610 44. Ralston, R., and Bishop, J. M. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,7845-7849 45. Gould, K., and Hunter, T. (1988) Mol. Cell. Biol. 8 , 3345-3356 46. Kypta, R. M., Goldberg, Y., Ulug, E. T., and Courtneidge, S. A. (1990) Cell 62,481-492