Ligand regulates epidermal growth factor receptor kinase specificity: activation increases preference for GAB1 and SHC versus autophosphorylation sites.

The epidermal growth factor receptor (EGFR) kinase catalyzes phosphorylation of tyrosines in its C terminus and in other cellular targets upon epidermal growth factor (EGF) stimulation. Here, by using peptides derived from EGFR autophosphorylation sites and cellular substrates, we tested the hypothesis that ligand may function to regulate EGFR kinase specificity by modulating the binding affinity of peptide sequences to the active site. Measurement of the steady-state kinetic parameters, K(m) and k(cat), revealed that EGF did not affect the binding of EGFR peptides but increased the binding affinity for peptides corresponding to the major EGFR-mediated phosphorylation sites of the adaptor proteins Gab1 (Tyr-627) and Shc (Tyr-317), and for peptides containing the previously identified optimal EGFR kinase substrate sequence EEEEYFELV (3-7-fold). Conversely, EGF stimulation increased k(cat) approximately 5-fold for all peptides. Thus, ligand changed the relative preference of the EGFR kinase for substrates as evidenced by EGF increases of approximately 5-fold in the specificity constants (k(cat)/K(m)) for EGFR peptides, whereas approximately 15-40-fold increases were observed for other peptides, such as Gab1 Tyr-627. Furthermore, we demonstrate that EGF (i) increased the binding affinity of EGFR to Gab1 Tyr-627 and Shc Tyr-317 sites in purified GST fusion proteins approximately 4-6-fold, and (ii) EGF significantly enhanced the phosphorylation of these sites, relative to EGFR autophosphorylation, in cell lysates containing the full-length Gab1 and Shc proteins. Analysis of peptides containing amino acid substitutions indicated that residues C-terminal to the target tyrosine were critical for EGF-stimulated increases in substrate binding and regulation of kinase specificity. To our knowledge, this represents the first demonstration that ligand can alter specificity of a receptor kinase toward physiologically relevant targets.

Protein-tyrosine kinases, which catalyze the transfer of ␥-phosphate from ATP to tyrosine side chains of peptide and protein substrates, play a central role in cellular signaling. Receptor tyrosine kinases (RTKs) 1 constitute one class of protein kinases that consist of an extracellular ligand binding, a transmembrane, an intracellular tyrosine kinase, and a Cterminal domain that contains tyrosine phosphorylation sites (1). The epidermal growth factor receptor (EGFR, ErbB1, and HER1) is the prototypical member of the ErbB family RTKs, which also includes ErbB2 (HER2, neu), ErbB3 (HER3), and ErbB4 (HER4). Binding of growth factors that are structurally related to EGF induces a process of the receptor dimerization, kinase activation and autophosphorylation (2). EGFR activation results in recruitment and phosphorylation of cytosolic downstream targets such as Grb2-associated binding protein 1 (Gab1) (3), Src homology 2 (SH2) domain, collagen-containing protein (Shc) (4), and phospholipase C␥-1 (PLC␥-1) (5,6). EGFR functions in the proliferation, migration, survival, and differentiation of mammalian cells (2,7), and dysregulation of signaling by EGFR has been implicated in cell transformation and cancer (8,9).
A feature common to many protein kinases is the presence of one or more tyrosines in a segment referred to as the activation loop (A-loop). Phosphorylation of these tyrosines has been shown to be essential for stimulation of catalytic activity and biological functions for a number of RTKs, such as insulin receptor kinase (IRK) (10), fibroblast growth factor receptor (11), and the platelet-derived growth factor receptor (12). Ligand binding to the extracellular domain (ECD) leads to dimerization of monomeric receptors or a rearrangement within the oligomeric receptors resulting in A-loop phosphorylation. In these instances, the unphosphorylated A-loop interacts with the kinase active site blocking the binding of ATP and/or peptide (13,14). The phosphorylated A-loop adopts an open and extended conformation allowing substrate binding and catalysis. However, substitution of the conserved tyrosine (Tyr-845) in the A-loop of EGFR with phenylalanine has no demonstrable effect on the biological function of the receptor (15). A recent crystallographic investigation revealed that the overall structure of the EGFR kinase domain, including conformation of the activation loop, catalytic residues, relative orientation of the ␣C helix, and the conserved salt bridge (between Lys-721 and Glu-738), is similar to those of the phosphorylated active IRK * 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 (16). This suggests that the ligand-induced structural basis for EGFR kinase activation is different relative to other RTKs.
Protein kinase specificity is believed to be critical for selective cellular signaling. The use of synthetic peptides in detailed studies investigating protein kinase specificity and catalysis is widely recognized and supported by crystallographic observations that peptide bound to the active site appears to exist in an extended conformation (17,18). By using a degenerate peptide library, it has been shown previously (19) that each tyrosine kinase has its own optimal peptide substrate, demonstrating that specificity is predominantly based on the residues immediately surrounding the phosphoacceptor tyrosine. In this work the identified optimal peptide for EGFR kinase was EEEEY-FELV, with the tyrosine designated as the zero position and adjacent N-or C-terminal residues designated by minus or plus numbers relative to this tyrosine, respectively. Other peptide model studies (20 -23) have also demonstrated the importance of acidic residues at these positions consistent with the identification of phosphorylation sites in tyrosine kinase protein substrates. By using peptides derived from EGFR, previous steady-state kinetic studies of EGFR kinase demonstrated that EGF increased V max values but did not significantly change peptide binding affinity (24,25).
Here we have tested the hypothesis that binding of ligand to a receptor kinase may not only activate the kinase by increasing k cat but may also alter the binding affinity of select substrate primary sequences to the enzyme-active site. We have performed a systematic study using affinity-purified, epitopetagged full-length EGFR and synthetic peptides corresponding to physiologically relevant phosphorylation sites. Our results revealed that EGF altered the relative preference of the EGFR kinase for substrates as evidenced by EGF-induced increases of ϳ5-fold in the specificity constant for peptides derived from EGFR, whereas ϳ15-40-fold increases were observed for other peptides, such as those derived from the cellular substrates Gab1 (containing Tyr-627) and Shc (Tyr-317). This differential fold increase in specificity constant for substrates was attributable to the fact that EGF lowered the K m and thus increased the binding affinity for these select peptide sequences. Consistent with the results obtained using peptides, we demonstrate that the K m values of EGFR kinase for Gab1 Tyr-627 and Shc Tyr-317 sites in glutathione S-transferase (GST) fusion protein substrates were decreased ϳ4 -6-fold by EGF. Finally, we carried out a competition study between EGFR autophosphorylation and phosphorylation of these sites in Gab1 and Shc using cell lysates and confirmed that EGF stimulation increases the preference of EGFR kinase for these exogenous cellular substrate sites, relative to EGFR autophosphorylation sites. We believe this to be the first demonstration that ligand can alter specificity of a receptor kinase toward physiologically relevant target substrates.

EXPERIMENTAL PROCEDURES
Antibodies and cDNAs-To generate the epitope-tagged EGFR, PCR amplification of a template containing EGFR cDNA in pcDNA 3.1/Mychis (Invitrogen) was performed by using 5Ј-tgtaccatcgatgtctacatga-3Ј and 5Ј-cccaagctttgctccaataaattcactgctttgt-3Ј primers. PCR product was digested with ClaI and HindIII and inserted into the template construct. Wild type, Y627F Gab1, and Shc expression plasmids were generously provided by Toshio Hirano (Osaka University) (26,27) and Silvio Gutkind (NCI, National Institutes of Health), respectively. GST fusion protein constructs were generated by PCR amplification by using primers containing 5Ј EcoRI and 3Ј NotI sites and subcloned into pGEX-6p1. Point mutations were produced by using the QuickChange kit (Stratagene), and the identity of cDNAs was confirmed by DNA sequencing. Monoclonal antibodies against MAPK, phosphotyrosine (PY20), and Shc were obtained from Transduction Laboratories, and monoclonal antibodies against EGFR and Myc were obtained from Neomarkers. Polyclonal phospho-Shc (Tyr(P)-317) and phospho-Gab1 (Tyr(P)-627) antibodies were obtained from Cell Signaling. Antibodies for phospho-MAPK and Gab1 were obtained from New England Biolabs and Upstate Biotechnology Inc., respectively.
Preparation of Peptides-Synthetic peptides were prepared by solidphase synthesis using standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry on an Applied Biosystems 432A peptide synthesizer and purified using preparative reversed-phase high performance liquid chromatography. The identity and purity of peptides were confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry and analytical reversed-phase high performance liquid chromatography.
Generation of Stable 32D Cell Lines-32D cells (28) were transfected with 20 g of DNA by electroporation (960 microfarads, 0.25 kV) using a Bio-Rad Gene Pulser II, cultured in 600 g/ml G418 for 2 weeks, and cloned by fluorescence-activated cell sorting. Briefly, cells were incubated with anti-EGFR antibody (Ab-1, Neomarkers), stained with fluorescein-conjugated secondary antibody, and sorted by using a BD Biosciences FACS Vantage SE into 96-well plate, 1 cell per well. Clones were characterized by Western blotting using anti-EGFR and Myc antibodies.
Western Blotting-Proteins were separated on 8% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were probed with 0.3 g/ml of primary antibody, and detection was performed using the Vectastain ABC Elite kit (Vector Laboratories) and enhanced chemiluminescence and exposed to film.
Purification of GST Fusion Proteins-Transformed BL 21 rosetta bacteria were grown to an A 600 nm of 0.7-0.8 at 37°C and induced with isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 30°C. Cells were centrifuged and lysed using a French press in lysis buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10% glycerol, 1 mM dithiothreitol, 1% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride, 0.2 g/ml pepstatin A, 0.2 g/ml aprotinin, 0.2 g/ml leupeptin). Lysates were clarified by centrifugation and incubated with glutathione-agarose for 1 h at 4°C. After washing three times with the lysis buffer, the bound GST fusion proteins were eluted with lysis buffer containing 10 mM reduced glutathione. The eluates were dialyzed against kinase reaction buffer without ATP. The concentrations of the fusion proteins were measured by Coomassie staining of SDS-PAGE gels using BSA as a standard.
Tyrosine Kinase Assay-All kinase reactions were performed in 25 mM Hepes, pH 7.4, 10 mM MgCl 2 , 2.5 mM MnCl 2 , 50 M sodium orthovanadate, 0.5 mM dithiothreitol, 0.2% Triton X-100, 40 g/ml BSA, 25 M ATP with 62.5 Ci/ml [␥-32 P]ATP (Amersham Biosciences) and 9 nM EGFR in a total volume of 40 l at 25°C. EGFR was preincubated with or without 1 M EGF for 10 min at 25°C. Reactions were initiated by addition of peptide and were stopped with 8.5% H 3 PO 4 . Terminated reaction mixtures were transferred to P30 Filtermats (Wallac) and washed three times with 0.85% H 3 PO 4 and three times with water, and bound radioactivity was quantified with a Trilux Macrobeta scintillation counter (PerkinElmer Life Sciences). Initial rates were determined from the linear portion of the reaction (Ͻ5% substrate converted). Four replicates were measured in all experiments, and kinetic parameters were calculated using the Enzyme Kinetic Module of Sigma Plot (Jandel Scientific).
For GST fusion protein phosphorylation by EGFR, the reactions were stopped with 4ϫ SDS sample buffer and separated by 4 -20% SDS-PAGE. The phosphorylated proteins were detected by antibodies against phospho-Gab1 (Tyr(P)-627) or phospho-Shc (Tyr(P)-317), and chemiluminescence was quantified using a Kodak Image Station 440.
Phosphorylation of Gab1 and Shc in Cell Lysates-Subconfluent COS-7 cells (15-cm dishes) were transfected with 10 g of Gab1 or Shc cDNAs using the DEAE-dextran/chloroquine technique (29). Cells were allowed to recover for 24 h and serum-starved overnight, and lysates were generated as described previously (30). Expression of transfected cDNAs was confirmed by Western blotting analysis (30). To phosphorylate Gab1 or Shc, the cell lysates were mixed with purified EGFR, and the reactions were initiated by the addition of kinase reaction buffer containing ATP. The reactions were terminated by addition of 4ϫ SDS sample buffer and separated by 8% SDS-PAGE. Autophosphorylation of EGFRs was detected by Western blotting using phosphotyrosine antibody (PY20), whereas phosphorylation of Gab1 and Shc sites were detected by antibodies against phospho-Gab1 Tyr-627 and phospho-Shc Tyr-317, respectively.

Expression and Purification of Epitope-tagged EGFR-To
facilitate studies of EGFR tyrosine kinase, we generated stable 32D cell lines expressing either wild type (WT) or kinaseinactive (K721M) epitope-tagged versions of the human EGFR. 32D cells are excellent for this purpose because they grow rapidly in suspension and do not express any endogenous ErbBs (30,31). A cDNA encoding the 1186-amino acid residue full-length EGFR with a 25-amino acid sequence containing a Myc and hexahistidine epitope sequence at the C terminus was generated. Transient transfection experiments comparing EGF-induced signaling in intact cells did not reveal any demonstrable difference between the EGFRs expressed from parental and tagged WT cDNAs (data not shown). 32D cells that express either WT or K721M EGFR were produced as described under "Experimental Procedures." Western blotting analysis of crude lysates relative to a Myc-tagged protein standard indicated that WT and K721M clones each expressed ϳ1.5 ϫ 10 6 EGFRs/cell. As expected, exposure of WT cells to EGF led to phosphorylation of mitogen-activated protein kinase (MAPK; extracellular-signal regulated kinase), whereas no MAPK phosphorylation was observed in K721M cells (Fig. 1A).
WT and K721M EGFRs purified by Co 2ϩ -based immobilized metal affinity chromatography yielded ϳ50 g of EGFR/liter of culture as estimated by silver-stained SDS-PAGE gels and Western blotting with anti-Myc antibody (data not shown). The synthetic 20-mer peptides used in the present study were derived from potential in vivo substrates, containing sequences identical to 17 amino acids surrounding the phosphoacceptor tyrosine, and were named by origin plus tyrosine position. The peptides also contained two and one lysine residues at the N and C termini, respectively, allowing them to bind strongly to phosphocellulose filters upon acidification (21). Using EGFR Tyr-1173, which contained a documented major autophosphorylation site from the C terminus of the EGFR (32) as substrate, WT possessed an activity of 2.8 M/min/M EGFR in the absence of EGF, whereas K721M had no significant detectable activity (Fig. 1B). EGF stimulated the receptor kinase activity in a dose-dependent manner, and a maximal stimulation was reached at an [EGF]/[EGFR] ratio of ϳ1 (Fig. 1B). The development of this assay provided a basis for the quantitative assessment of the catalytic efficiency of the EGFR toward peptide substrates and EGF regulation of this catalysis.

EGF Does Not Affect the Relative Binding Affinities toward EGFR Peptides but Consistently Increases the Catalytic Rate
Constants ϳ5-Fold-EGFR possesses three major and two minor tyrosine autophosphorylation sites located at Tyr-1068, Tyr-1148, Tyr-1173 (24,25,32), Tyr-992, and Tyr-1086 (33)(34)(35), respectively, and peptides were prepared representing these EGFR tyrosines. In addition, a peptide corresponding to Tyr-1114 in the C-terminal tail was generated because of the presence of acidic residue at Ϫ1 position (Table I). Steady-state enzyme kinetic studies provide measurements of relative substrate binding affinities (K m ) and the catalytic rate constant (k cat ). Fig. 2A contains representative Lineweaver-Burk plots of EGFR Tyr-1173 phosphorylation by EGFR at a fixed ATP concentration of 25 M in the absence or presence of EGF. The linearity of the plots confirmed that Michaelis-Menten kinetics was obeyed under either condition. The double-reciprocal plots in the absence or presence of EGF intersected at the abscissa indicating that ligand did not affect the K m value for the pep-tide. However, EGF increased the k cat value ϳ5.6-fold as evidenced by distinct intersections of the plots with ordinate. As summarized in Table I, the K m values for the different peptides ranged from 181 to 665 M, whereas k cat values varied from 1.8 to 3.3 min Ϫ1 in the absence of EGF. EGF stimulation yielded little or no changes in K m values but resulted in 4.0 -5.6-fold increases in k cat values. The specificity constant of an enzyme, k cat /K m , is the best measure for comparison of catalytic efficiencies toward different substrates. EGF activation resulted in an ϳ5-fold increase in k cat /K m values for all EGFR peptides, which was mainly contributed by k cat . The EGFR peptides exhibited specificity constants Ն50 min Ϫ1 mM Ϫ1 in the presence of EGF with the exception of the Tyr-1086 autophosphorylation site peptide (ϳ12 min Ϫ1 mM Ϫ1 ). Similar to the three major EGFR autophosphorylation sites and many tyrosine kinase substrates (23), Tyr-1114 of EGFR is preceded by an acidic residue and possessed a relatively high specificity constant of ϳ50 min Ϫ1 mM Ϫ1 . However, Tyr-1114 has not been found to be  6) were serum-starved, stimulated with 10 nM EGF for 5 min at 37°C, and lysates generated as described previously (30). Lysates were separated in SDS-PAGE and transferred to a polyvinylidene difluoride membrane, and Western blotting was performed with antibodies against Myc, MAPK, and phosphorylated MAPK. B, purified 9 nM WT or K721M was preincubated with various concentrations of EGF, and in vitro kinase assays were performed using 1. phosphorylated in either in vitro reactions or in EGFR from EGF-treated cells (33,36) implying that this site may not be accessible to the active site in the full-length receptor. These findings demonstrate that EGF does not change the binding affinity of EGFR peptides to the enzyme-active site but consistently increases the k cat and specificity constants ϳ5-fold for the EGFR peptides.
EGF Can Alter EGFR Kinase Specificity by Increasing the Binding Affinity of Select Target Sequences to the Enzymeactive Site-The enzymatic specificity of the EGFR kinase toward cellular target sequences was studied by using three peptides containing Gab1 Tyr-627, Shc Tyr-317, or PLC␥-1 Tyr-771. These sites have been documented previously to be major EGFR-mediated direct phosphorylation sites either in phosphorylation reactions in vitro and/or in EGFR isolated from cells stimulated with EGF (6,37,38). Lineweaver-Burk plots for the Gab1 and Shc peptides are provided in Fig. 2, B and C, respectively, and the determined values for the catalytic parameters are summarized in Table II. As observed for EGFR peptides, EGF increased k cat values ϳ5-fold for all of these substrates. Conversely, ligand-induced activation of EGFR significantly decreased K m values for the Gab1 Tyr-627 and Shc Tyr-317 ϳ3-fold as evidenced by the distinct intersections at the abscissa of the double-reciprocal plots in the absence and presence of EGF (Fig. 2, B and C). Thus, the EGF-induced increased binding of these peptides to the EGFR-active site resulted in striking increases in specificity constants. The k cat /K m values for the Gab1 and Shc peptides were increased by ϳ17and ϳ15-fold, respectively, and these stimulations were far greater than observed for any EGFR peptides (Fig. 3). Strikingly, the Gab1 peptide exhibited a very high specificity constant of 979 min Ϫ1 mM Ϫ1 , which was ϳ10-fold greater than the most preferred EGFR peptide (Tyr-992). The EGFR binding affinity for the PLC␥-1 Tyr-771 peptide was modestly increased by EGF (ϳ34%) and resulted in an ϳ9-fold stimulation in specificity constant (Table II).
In addition to peptides derived from cellular EGFR targets, we analyzed catalysis of a peptide whose sequence was optimized for EGFR using a synthetic degenerate peptide library approach (19) which we referred to as OPTIMAL. As shown in Fig. 2D and Table II, the K m value for OPTIMAL was much lower than all tested peptides. EGF stimulation yielded an even more remarkable decrease in K m (7.5-fold), and similar to the other peptides, the k cat increased 5.2-fold. As a result of the EGF-induced concomitant increased binding affinity to the EGFR and increased k cat , the specificity constant for the OP-TIMAL peptide was enhanced nearly 40-fold by EGF (Table II). To determine whether residues flanking the core OPTIMAL sequence could influence phosphorylation, we replaced the core of EGFR Tyr-1148 with the OPTIMAL sequence EEEEYFELV to yield OPTIMAL*. The kinetic parameters for OPTIMAL* were very similar to those for OPTIMAL, indicating that the residues beyond the core sequence had little effect on catalytic efficiency or on EGF-mediated increased binding to the EGFRactive site.
The results presented in Fig. 3 highlight the differential effect that EGF had on the magnitude of increases in specificity constant for the tested peptides. All of the peptides derived from the C-terminal region of the EGFR exhibited ϳ5-fold EGF-stimulated increases in specificity constant, whereas the PLC␥-1 Tyr-771, Shc Tyr-317, Gab1 Tyr-627, and OPTIMAL peptides specificity constants were increased ϳ9-, 15-, 17-, and 40-fold, respectively, by EGF.
These finding demonstrate that the relative preference, and thus specificity, of the EGFR kinase toward peptide substrates can be altered by ligand. Furthermore, the data indicate that EGF-induced regulation of EGFR kinase specificity is mediated by the increased binding affinity of select peptide sequences to the enzyme-active site.
Amino Acid Substitutions within Peptides Reveal Substrate Characteristics Critical to EGF-induced Increased Binding Affinity to EGFR-Measurement of the kinetic parameters for a peptide derived from the EGFR A-loop (Tyr-845) demonstrated that the substrate had a relatively low specificity constant of ϳ6 min Ϫ1 mM Ϫ1 due to a weak binding affinity for the EGFR  (K m of ϳ1 mM) and a modest turnover number of ϳ6 min Ϫ1 in the presence of EGF (Table III). This was particularly striking because the acceptor tyrosine was preceded by glutamate residues at positions Ϫ1, Ϫ3, and Ϫ4, which has been determined to be optimal for EGFR substrates (19). In addition, ligand did not change the binding affinity of the Tyr-845 A-loop peptide toward the EGFR. Therefore, to provide insight into the substrate, structural determinants important to binding and catalysis by EGFR and why EGF regulated binding of select peptides, we prepared A-loop peptides containing sequential substitutions at one or more amino acid positions. First, we replaced the lysine residue at position Ϫ2 with glutamic acid (K843E). This change generated a peptide with residues at positions Ϫ1 to Ϫ4 exactly as in OPTIMAL. The K843E substitution decreased K m ϳ4.5-fold and doubled the catalytic rate constant, but no increased binding affinity of peptide by EGF was observed (Table III).
To investigate the contribution of residues on the C-terminal side of the phosphoacceptor tyrosine, we replaced A-loop residues at positions ϩ1 to ϩ4 with those present in the OPTIMAL sequence. Replacement of His with Phe at ϩ1 (H846F) decreased K m from ϳ0.9 to 0.4 mM and increased k cat by ϳ2-fold, and EGF had no dramatic effect on peptide binding affinity (Table III). However, when positions ϩ2 to ϩ4 in EGFR Tyr-845 H846F were replaced with ELV, K m values for the resulting peptide were decreased, and EGF-stimulated increased binding affinity was observed (ϳ2-fold). However, the ELV replacement did not significantly alter k cat . The additional Lys to Glu substitution at Ϫ2 position generated a peptide containing the core nine amino acids of OPTIMAL within the flanking residues of the A-loop peptide (K843E, H846F, A847E, E848L, and G849V), further decreased K m values, enhanced the EGFinduced increase of substrate binding affinity, and resulted in kinetic properties similar to OPTIMAL and OPTIMAL*. Thus, with regard to EGFR kinase peptide substrates, these results indicated the following: (i) residues positioned both N-and C-terminal to the target tyrosine are important for binding; (ii) the occurrence of EGF-stimulated increased binding affinity is dependent upon substrate primary sequence; and (iii) the Cterminal positions are critical to the enhanced binding of substrate to active site induced by EGF.

EGF Increases the Binding Affinity of EGFR to Gab1 Tyr-627 and Shc Tyr-317 Sites in Purified GST Fusion Proteins-To demonstrate that the EGFR-mediated phosphorylation of Gab1
Tyr-627 or Shc Tyr-317 in protein substrates is also regulated by EGF, we generated purified GST fusion proteins containing either Gab1 or Shc. Phosphorylation of Gab1 Tyr-627 and Shc Tyr-317 in full-length or truncated fusion proteins was detected by specific antibodies directed against these phosphorylated sites. The specificities of the antibodies were confirmed using Gab1 Y627F and Shc Y317F control proteins (see below and Fig. 4). The determined apparent K m values and the fold increase in k cat are summarized in Table IV. The K m values for GST-Gab1 and GST-Gab1-(566 -694) are lower than that for the Gab1 Tyr-627 peptide. However, the phosphorylation of Tyr-627 in either protein or peptide substrates is regulated in a similar manner by EGF. EGF decreased the K m values ϳ5fold and increased the k cat values ϳ3-fold for phosphorylation of the Tyr-627 site in both of the Gab1 GST fusion proteins.
Shc possesses an SH2 domain and a phosphotyrosine binding (PTB) domain. These domains bind phosphotyrosine sites in proteins with high affinity, and as expected GST-Shc had an extremely low K m of 4 nM and was not regulated by EGF. This was consistent with tight binding to the C-terminal region of the autophosphorylated EGFR. However, phosphorylation of Tyr-317 in GST-Shc-(240 -383) that lacks the PTB and SH2 domain had a K m value of 128 M. As observed for the Shc Tyr-317 peptide, EGF decreased the K m for GST-Shc-(240 -383) ϳ3-fold. EGF increased the k cat values of both GST-Shc and GST-Shc-(240 -383) ϳ4-fold. These results demonstrated that EGFR tyrosine phosphorylation of Gab1 Tyr-627 and Shc Tyr-317 in either protein or peptide substrate is regulated by EGF in a comparable manner, and EGF functions to increase binding affinity of the receptor kinase to these target sites.

EGF Increases the Phosphorylation of Gab1 Tyr-627 and Shc Tyr-317 Sites, Relative to EGFR Autophosphorylation, in Cell
Lysates Containing the Gab1 and Shc Proteins-It was reported previously that exogenous substrates have to compete with intrinsic tyrosine phosphorylation sites for the catalytic domain of the EGFR kinase (25,35,39). To test the proposal that ligand-induced activation may alter the relative preference of the receptor kinase toward autophosphorylation sites and downstream substrates, we tested the ability of EGF to regulate the competition between EGFR autophosphorylation  Tables I and II.  and Gab1 or Shc phosphorylation. Purified EGFR was added to lysates derived from COS-7 cells transfected with expression plasmids bearing either full-length Gab1 or Shc cDNAs, and kinase reactions were performed in the absence or presence of EGF (Fig. 4). EGFR autophosphorylation was detected with a phosphotyrosine antibody, whereas the phosphorylation of Gab1 Tyr-627 and Shc Tyr-317 was detected by using specific antibodies against these phosphorylation sites. The time courses of the reactions for Gab1 and Shc are presented in Fig.  4, A and C, respectively. No significant phosphorylation of EGFR, Gab1, or Shc was observed without the addition of exogenous EGFR to the lysates, and no phosphorylation of Gab or Shc was detected in lysates derived from mock-transfected cells (no cDNA; data not shown). Neither Gab1 Y627F nor Shc Y317F phosphorylation was detected after a 5-min reaction confirming that the phospho-Gab1 and phospho-Shc antibodies specifically recognize phosphorylated Tyr-627 in Gab1 and Tyr-317 in Shc (Fig. 4, A and C). Quantification of the phosphorylations demonstrated that the ratio of phosphorylated Gab1 Tyr-627 to autophosphorylated EGFR after 2 min was in-

FIG. 4. Competition between EGFR autophosphorylation and Gab1 or Shc protein substrate phosphorylation in the absence and presence of EGF.
A, COS-7 cells were transfected with wild type (WT) or Y627F Gab1 cDNA. Cell lysates were prepared as described under "Experimental Procedures." Purified EGFR (2 nM final concentration) was incubated with the cell lysates in the absence and presence of 1 M EGF. The phosphorylation reaction was initiated by addition of kinase reaction buffer containing 2 M ATP and terminated by 4ϫ SDS buffer at the indicated time. EGFR autophosphorylation was detected by anti-phosphotyrosine antibody (PY20), whereas Gab1 phosphorylation was detected by phospho-Gab1 (Tyr-627) antibody. The amounts of EGFR and Gab1 in reactions were confirmed by Western blotting using the indicated antibodies. B, ratio of Gab1 Tyr-627 phosphorylation and EGFR autophosphorylation after 2 min in the absence and presence of EGF. The band densities were quantified with Kodak Image Station 440, and the data points represent the means Ϯ S.E. of triplicate experiments. C and D were performed exactly as A and B, respectively, except that Shc cDNAs, Shc, and phospho-Shc Tyr-317 antibodies were used instead of those for Gab1. creased ϳ6-fold by EGF (Fig. 4B). In a similar manner, EGF increased the ratio of phosphorylated Shc Tyr-317 to autophosphorylated EGFR by ϳ3-fold (Fig. 4D). We also found that EGFR autophosphorylation was decreased by the presence of Gab1 or Shc (data not shown). These data indicated that Gab1 and Shc phosphorylation compete with receptor autophosphorylation, and EGF stimulation facilitates phosphorylation of these exogenous protein substrates. The increased preference of the receptor kinase to Gab1 Tyr-627 and Shc Tyr-317 by EGF, relative to EGFR autophosphorylations, is consistent with our measurements demonstrating that EGF decreased the K m values for the Gab1 and Shc sites in either purified peptides or protein substrates. DISCUSSION Activation of the EGFR kinase by ligand leads to phosphorylation of tyrosines in its C terminus and in other cellular targets upon EGF stimulation. Enzymatic activation by ligand may represent an increase in k cat as has been shown for the EGFR (24,25) or in concomitant increases in k cat and decreases in K m as has been demonstrated for the IRK (40). In our study, we have made the novel finding that ligand can also regulate the specificity of a receptor kinase toward physiologically relevant substrates. This concept is most clearly demonstrated by Fig. 3 in which the EGF-stimulated fold increases in specificity constants are compared for 10 peptide substrates and were found to be quite different. EGFR peptides exhibited ϳ5-fold increases in specificity constants by EGF, whereas other peptides had fold increases of up to 40 (Fig. 3). Regardless of the peptide, EGF consistently increased the turnover number, k cat , of EGFR-catalyzed phosphorylation of all the peptides tested in this study ϳ5-fold. Thus, the underlying molecular basis for the differential EGF-mediated fold increases in specificity constant observed for different peptides is because of the increased binding affinity of certain substrate sequences to the EGFR-active site. EGF alters the relative preference (i.e. binding, recognition) of the EGFR kinase toward substrates and therefore can regulate specificity. Gab1 Tyr-627 or Shc Tyr-317 phosphorylation by EGFR kinase in a GST fusion protein was found to have a lower apparent K m value than the corresponding peptides. One possible reason is that either GST-Gab1 or Gab1-(566 -694) possesses multiple identified EGFR-mediated phosphorylation sites (38). The apparent K m app value is shown in Equation 1, where K m is the actual Michaelis constant for the measured sites, and K m i is the Michaelis constants for any of the other sites. 2 Another possible reason is that more complicated inter-actions, such as interactions between the kinase and protein substrates via regions outside the core sequence immediately surrounding the target tyrosine residue, may be involved in the formation of the enzyme-substrate complex (38). The crystallographic investigation has revealed that cAMP-dependent kinase interacts with a peptide inhibitor through multiple binding regions (41). A similar phenomenon has been observed for IRK in which the K m for an insulin receptor substrate-1 (IRS-1) fragment (amino acids 586 -1149) is ϳ7, whereas K m values for peptides derived from the major phosphorylation sites are in the range of 24 -300 M (42). However, the EGFR K m values for Gab1 Tyr-627 in GST-Gab1, -Gab1-(566 -694), and Shc Tyr-317 in GST-Shc-(240 -383) decreased ϳ3-5-fold by EGF, indicating that the phosphorylation of these sites in protein substrates is regulated by EGF in a similar manner as observed in the peptide substrates. The results obtained using peptide substrates reflect, at least partially, the behavior of related phosphorylation sites in protein substrates.
Many RTKs require phosphorylation of one or more conserved tyrosine in the activation loop for full catalytic activity (43). For example, IRK activation loop phosphorylation results in 25-and 40-fold decreases in K m values for ATP and peptide, respectively, and a 7-fold increase in k cat (40). Unlike IRK and other RTKs, the crystal structure of the EGFR catalytic domain revealed that the unphosphorylated A-loop exists in an open extended conformation consistent with an enzymatically active kinase (16). We generated and purified an epitope-tagged EGFR containing the Y845F A-loop (homologous to IRK A-loop Tyr-1162) mutation and found that the EGF-stimulated increased binding affinity for the Gab1 Tyr-627 peptide was maintained (data not shown). Thus, A-loop phosphorylation of the EGFR does not appear to be required for ligand-dependent enhanced binding of substrate.
EGF significantly decreases K m for only certain peptides, including Gab1 Tyr-627, Shc Tyr-317, and peptides containing OPTIMAL sequence (Table II). Sequence analysis of these peptides reveals the presence of hydrophobic residues at ϩ1 and ϩ3 positions. Sequential amino acid substitution within the EGFR A-loop peptide also indicated that residues C-terminal to the target tyrosine play an important role in increased peptide binding affinity by EGF (Table III). As demonstrated by the activated IRK-peptide crystal structure, hydrophobic amino acids at positions ϩ1 and ϩ3 are accommodated in two adjacent hydrophobic pockets on the C-lobe surface of the kinase. The analogous peptide binding surface in the EGFR kinase is composed of Ile-854, Ser-900, Ala-896, Val-852, Met-857, and Ile-862. For EGF to increase the peptide binding affinity, ligand engagement must induce the rearrangement of the binding surface. In some instances this rearrangement does not significantly alter the total strength of the interactions with peptide substrate, and K m is not influenced. Conversely, for select peptides, such as Gab1 Tyr-627 or OPTIMAL, higher affinity interactions are formed in the EGF-activated conformation, probably through the residues C-terminal to the target tyrosine in the peptide substrates. Indeed, the substitution of the residues C-terminal to the tyrosine in EGFR Tyr-845 peptide with those of the optimal peptide leads to the EGF-induced increase in peptide binding affinity. We conclude that the Cterminal positions are critical to the EGF-induced increases in receptor-substrate binding affinity. This control of peptide substrate binding by ligand appears to be a unique property of the EGFR relative to other RTKs.
Our kinetic analysis revealed a very consistent EGF-induced increase in k cat irrespective of their sequence (Tables I-III and data not shown). The simplest explanation for this increase is that binding of EGF to EGFR shifts a pre-existing equilibrium 2 Y. X. Fan and G. R. Johnson, unpublished results. from low to high activity forms. Crystal structures have revealed that EGFR ECD can adopt a closed autoinhibited configuration or an open extended form that can engage ligand (44 -46). EGF binding shifts the equilibrium toward the extended form, and consequently, the closed form converts to the extended form (46). It is reasonable to suggest that the closed and extended configurations of EGFR ECD represent lower and higher k cat forms of kinase domain, respectively. EGF may induce a subtle conformational change or increase flexibility of the kinase domain active site. Changes in flexibility or slight conformational adjustments within the active site can result in dramatic changes in enzymatic activity (47,48). The movement of the ␣C-helix is thought to be critical for the activation of a number of kinases (49). It may also play an important role for the regulation of EGFR kinase activity. The opening of the EGFR ECD results in receptor dimerization that may facilitate propagation of the allosteric conformational signal and stabilize the active form (43).
Because the specificity of protein kinases is believed to be critical for selective cellular signaling, the ability of ligand to alter specificity must also be important. Protein-tyrosine phosphatases control the activities of RTKs and the phosphorylation of downstream substrates of these kinases (50). This would seem to be important for repressing the constitutive tyrosine kinase activity of the EGFR which could lead to dysregulated signaling. Thus, it would appear that EGF functions to shift the equilibrium to more active EGFR to exceed the hurdle of dephosphorylation by protein-tyrosine phosphatases and to generate signaling in a controlled manner. The Gab1 Tyr-627 and Shc Tyr-317 phosphorylation sites are of particular interest because they play important roles in EGF-mediated MAPK activation (51,52). Our finding that EGF activation yielded significantly greater fold increases in specificity constants for the Gab1 and Shc peptides, relative to EGFR autophosphorylation site peptides, suggests that ligand could be responsible for preferentially enhancing phosphorylation of Gab1 or Shc in cells, relative to other sites such as in the EGFR C terminus. Shc can bind directly to the EGFR via its SH2 or PTB domains (53), and clearly, this interaction may facilitate Shc phosphorylation by EGFR. However, the EGF-stimulated increased binding to the EGFR-active site of the amino acid residues surrounding Shc Tyr-317 or Gab1 Tyr-627 may also represent an important signaling event. In addition to our studies with peptides, this was clearly demonstrated by our finding that EGF enhances the phosphorylation of Gab1 Tyr-627 and Shc Tyr-317 in full-length proteins relative to EGFR autophosphorylation. The EGF-induced increased binding of physiologically relevant target sequences in downstream substrates may be required for sufficient levels of phosphorylation to be achieved to exceed the protein-tyrosine phosphatase barrier and thus for signal progression to occur. Furthermore, regulation of EGFR target recognition by ligand may be essential to specific physiological responses because it has been proposed that the magnitude and duration of tyrosine phosphorylation can dictate biological outcome (54). Nevertheless, it will be of interest to determine whether other growth factors or cytokines can alter the substrate specificity of their cognate receptor kinases.