The juxtamembrane, cytosolic region of the epidermal growth factor receptor is involved in association with alpha-subunit of Gs.

Previously, we have demonstrated that epidermal growth factor (EGF) can stimulate adenylyl cyclase activity via activation of Gs in the heart. Moreover, we have recently shown that Gsα is phosphorylated by the EGF receptor protein tyrosine kinase and that the juxtamembrane region of the EGF receptor can stimulate Gs directly. Therefore, employing isolated cardiac membranes, the two-hybrid assay, and in vitro association studies with purified EGF receptor and Gsα we have investigated Gsα complex formation with the EGF receptor and elucidated the region in the receptor involved in this interaction. In isolated cardiac membranes, immunoprecipitation of EGF receptor was accompanied by co-immunoprecipitation of Gsα. In the yeast two-hybrid assay, the cytosolic domain of the EGF receptor and the N-terminal 64 amino acids of this region (Met644-Trp707) associated with Gsα. However, interactions of these regions of the EGF receptor with constitutively active Gsα were diminished in the two-hybrid assay. Employing purified proteins, our studies demonstrate that the EGF receptor, directly and stoichiometrically, associates with Gsα (1 mol of Gsα/mol of EGF receptor). This association was not altered in the presence or absence of ATP and therefore, was independent of tyrosine phosphorylation of either of the proteins. Peptides corresponding to the juxtamembrane region of the receptor decreased association of the EGF receptor with Gsα. However, neither the C-terminally truncated EGF receptor (Δ1022-1186) nor a peptide corresponding to residues 985-996 of the receptor altered association with Gsα, thus indicating the selectivity of the G protein interaction with the juxtamembrane region. Interestingly, peptides corresponding to N and C termini of Gsα did not alter the association of Gsα with the EGF receptor. Consistent with the findings from the two-hybrid assay where constitutively active Gsα poorly associated with the EGF receptor, in vitro experiments with purified proteins also demonstrated that activation of Gsα by guanosine 5′-3-O-(thio)triphosphate decreased the association of G protein with the EGF receptor. Thus we conclude that the juxtamembrane region of the EGF receptor, directly and stoichiometrically, associates with Gsα and that upon activation of Gsα this association is decreased.

The pleiotropic actions of epidermal growth factor (EGF) 1 are elicited by stimulation of a number of second messenger systems by the ligand-activated EGF receptor (1,2). In addition to its well documented effects on the mitogen-activated protein kinase cascade (3) and phospholipase C␥ (4), EGF has also been demonstrated to regulate the cAMP second messenger system (5). We have previously demonstrated that in cardiac myocytes EGF elevates cAMP accumulation (6) by augmenting the activity of adenylyl cyclase (7,8) and that this increase in cAMP accumulation also augments the beating rate and contractility in intact hearts (9). EGF stimulates cardiac adenylyl cyclase by activation of the ␣-subunit of the stimulatory GTP-binding protein, G s␣ (8), and the protein tyrosine kinase activity of the EGF receptor is important in this modulation (10). More recently, we have demonstrated that a 13-amino acid sequence in the cytosolic, juxtamembrane, region of the EGF receptor is important for activation of G s and stimulation of adenylyl cyclase (11). In addition to the activation of G s by the juxtamembrane region of the receptor (11), we have also shown that the EGF receptor protein tyrosine kinase can phosphorylate G s␣ on tyrosine residues and that this phosphorylation of G s␣ increases its ability to stimulate adenylyl cyclase (12). Thus, the combined actions of the juxtamembrane region of the activated EGF receptor on G s (11) and phosphorylation of G s␣ by the EGF receptor protein tyrosine kinase (12), may, in a mutually reinforcing manner, amplify the signaling events leading to adenylyl cyclase stimulation.
The juxtamembrane region of the EGF receptor has also been shown to be important for determining specificity of mitogenic signaling as well as determining substrate specificity. Thus, the deletion of 8 amino acids in the juxtamembrane region of the EGF receptor (amino acids 660 -667) alters the mitogenic activity of EGF (13). In addition, the mutation of one amino acid residue in this region (Arg 662 ) of the EGF receptor alters the mitogenic signaling and pattern of protein phosphorylations within the cells without altering the protein tyrosine kinase activity of the receptor (14). Likewise, the juxtamem-brane region (Met 644 -Gly 666 ) of the EGF receptor is also important for the association of phosphatidylinositol-4 kinase and phosphatidylinositol 4-phosphate 5-kinase (15). This latter region encompasses the 13-amino acid region that activates G s (11). Because the juxtamembrane region of the EGF receptor has been demonstrated to be important for activation of G s (11), as well as binding of proteins, and determining substrate specificity for the EGF receptor protein tyrosine kinase (13)(14)(15), the studies described herein were performed to characterize the association of EGF receptor with G s␣ , and to elucidate the region(s) in the receptor that are important for such an association. Our studies demonstrate that in cardiac membranes, the activated EGF receptor associates with G s␣ . Moreover, our data from yeast two-hybrid assays, as well as from in vitro association studies with purified proteins (EGF receptor and G s␣ ), demonstrate that the juxtamembrane region of the EGF receptor is important for association with G s␣ and that the GDP-liganded G s␣ , not GTP⅐G s␣ , is the preferred form of the G protein which physically interacts with the EGF receptor. Interestingly, the association between EGF receptor and G s␣ is stoichiometric, direct (not involving adapter protein(s)), and does not involve the major phosphorylation sites on the EGF receptor.

Isolation of Cardiac Membranes and
Incubations-Hearts excised from male rats of the Harlan Sprague Dawley strain (180 -200-g body weight) were homogenized in medium containing 5 mM Tris-HCl, pH 7.4, 250 mM sucrose, and 1 mM EGTA. Cardiac membranes were isolated from the homogenate by the methods previously described (7,8). The cardiac membranes (300 g of protein) were then incubated in adenylyl cyclase assay mixture described previously (7,8) in the presence and absence of EGF (100 nM). Following incubation for 30 min at room temperature, the reactions were terminated by addition of lysis buffer described below and immunoprecipitations performed (see below).
Construction of Plasmids Encoding Chimeric Proteins-Essentially, this assay was performed using the plasmids and yeast strains provided in the Matchmaker TM kit (Clontech Laboratories Inc.). Employing the full-length human EGF receptor cDNA as template (gift from Dr. Gordon Gill, University of California at San Diego), and primers corresponding to nucleotides 1932-1953 (sense strand; primer sequence: 5Ј-GGGGAATTCATGCGAAGGCGCCACATCGTTCGG-3Ј) and 3538 -3561 (complementary strand; primer sequence: 5Ј-AGGTCGACG-GATCCCTCATGCTCCAATAAATTCACTGCT-3Ј), the complete cytosolic region of the EGF receptor (amino acids 644 -1186; nucleotides 2188 -3816) was generated by PCR. The 5Ј primer introduced an EcoRI site (underlined above) and the 3Ј primer was tagged with BamHI and SalI sites downstream (underlined above). The addition of the unique EcoRI site at the 5Ј end facilitated the in-frame cloning of the cDNA corresponding to the cytosolic region of EGF receptor into the plasmids pGAD424 and pGBT9; these constructs are referred to as pGAD424-EGFR C and pGBT9-EGFR C , respectively.
The plasmids pGAD424 and pGBT9 contain the activating domain and binding domain, respectively, of the GAL4 gene and expression is under the control of the yeast alcohol dehydrogenase promoter. The BamHI site at the 3Ј end (introduced by PCR) along with an internal BamHI site at nucleotide 2121 in the EGF receptor cDNA facilitated the truncation of the chimeric constructs in plasmids pGAD424-EGFR C and pGBT9-EGFR C so that only N-terminal 64 amino acids (Met 644 -Trp 707 ) in the juxtamembrane region of the cytosolic domain of the EGF receptor were expressed as fusion proteins with the activating and binding domains of GAL4, respectively. These chimeric constructs in the plasmids expressing the short form of the EGF receptor cytosolic domain are referred to as pGAD424-EGFR CJM and pGBT9-EGFR CJM . In addition, using a 5Ј primer corresponding to amino acids Gly 695 -Phe 699 (sequence: 5Ј-CAAAGTCCCGGGCTCCGGTGCGTTC-3Ј) tagged with a SmaI site (underlined) and 3Ј complementary primer corresponding to nucleotides 3796 -3816 described above, constructs pGAD424-EGFR C⌬JM and pGBT9-EGFR C⌬JM were also generated. These latter constructs encoded all of the cytoplasmic region (amino acids 694 -1186) of the EGF receptor devoid of the juxtamembrane region (amino acids 645-694).
Employing the full-length G s␣ cDNA as template (obtained from Dr. Alfred Gilman, University of Texas Southwestern Medical Center) and primers corresponding to nucleotides 1-27 (sense strand; sequence:  5Ј-ATTCTAGACCGTCGACCCATGGGCTGTCTCGGAAACAGC-3Ј)  and 1122-1143 (complementary strand; sequence: GAGGTTGTCGAC-TTAGAGCAGCTCATACTGACG 3Ј), SalI restriction endonuclease sites (underlined in sequences above) were added on the 5Ј and 3Ј ends, respectively. The SalI site at the 5Ј end facilitated the in-frame cloning of the G s␣ cDNA into the plasmids pGAD424 and pGBT9 to generate plasmids pGAD424-G s␣ and pGBT9-G s␣ , respectively; numbering system for nucleotides and amino acids used here are those for the short form of G s␣ (16). The constitutively active form of G s␣ (Q213L) (17) was obtained by a two-step PCR. First, using G s␣ cDNA as template along with 5Ј sense primer mentioned above and a 3Ј complementary primer corresponding to nucleotides 628 -648 of G s␣ with a T 3 A substitution at nucleotide 638 (to substitute glutamine for leucine; sequence: TTC-ATCGCGCAGGCCGCCCAC), a PCR product was generated. Similarly, using a sense primer corresponding to nucleotides 628 -650 of G s␣ with an A 3 T substitution at nucleotide 638 and the 3Ј complementary primer to G s␣ mentioned above, another PCR product was generated.
Using the PCR products of G s␣ with the A 3 T and T 3 A substitutions in the coding and noncoding strands, respectively, and using the 5Јmost sense and 3Ј-most primers mentioned above, the full length G s␣ cDNA coding for the constitutively active form of the protein was generated by a second round of PCR. This latter cDNA encoding the constitutively active, Q213L, form of G s␣ (G s␣ *) was then cloned into the SalI site of pGAD424 and pGBT9 vectors to generate plasmids pGA-D424-G s␣ * and pGBT9-G s␣ *, respectively. The mutation of G s␣ at nucleotide 638 was confirmed by di-deoxynucleotide sequencing method (18). Likewise, all of the plasmid constructs were sequenced to confirm that the cloning of the appropriate cDNAs was in-frame for transcription and devoid of any mistakes resulting from PCR. Two-hybrid Assay-The two-hybrid assay was performed employing the HF7c strain of yeast (Clontech Laboratories Inc.) transformed with the various constructs in plasmids pGAD424 and pGBT9. Transformation of cells was performed as described by Clontech Laboratories Inc. in the Matchmaker TM kit. The transformed yeast cells were grown on plates containing either medium devoid of L-leucine (Leu Ϫ ) and Ltryptophan (Trp Ϫ ) or medium in which L-histidine as well as L-leucine and L-tryptophan (Leu Ϫ /Trp Ϫ /His Ϫ ) had been omitted. The plates were incubated at 30°C for 3 days. Several of the colonies from transformants were then individually streaked out onto new plates containing the corresponding media.
To monitor ␤-galactosidase activity in transformants growing on Leu Ϫ /Trp Ϫ /His Ϫ medium, colonies were individually grown in liquid medium lacking the three amino acids. After overnight growth in this medium, when the cells had reached an A 600 nm of approximately 0.3, the cells were harvested by centrifugation, resuspended in complete medium and grown for an additional 3 h (A 600 ϭ ϳ0.5). At this point, the cells were harvested by centrifugation, washed with buffer containing 60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 1 mM KCl, 1 mM MgCl 2 , pH 7.0, and then lysed in the same buffer supplemented with Triton X-100 (final concentration, 0.2%). The cell lysates were subjected to two freeze-thaw cycles with liquid nitrogen and assayed for ␤-galactosidase activity employing the chemiluminescence kit obtained from Clontech Laboratories Inc.
Purification of G s␣ and EGF Receptor-The BL21(DE-3) strain of Escherichia coli transformed with the plasmid pQE-60, containing cDNA encoding the 45-kDa form of bovine G s␣ , was obtained from Alfred Gilman (University of Texas Southwestern Medical Center, Dallas, TX). Expression of G s␣ was induced with isopropyl ␤-D-thiogalactopyranoside, and the protein was purified essentially as described by Graziano et al. (19). EGFR was purified from A431 cells as described previously (20).
In Vitro Association of EGF Receptor and G s␣ -Purified EGF receptor (33 ng) and recombinant G s␣ (25 ng) were incubated in a medium (final volume ϭ 10 l) containing 5 mM HEPES-NaOH, pH 7.4, 5 mM MgCl 2 , 2 mM MnCl 2 , 20 g/ml aprotinin, 20 g/ml leupeptin, 1 mM dithiothreitol, and 1 M EGF with or without 10 M ATP at room temperature for 60 min. The mixture was supplemented with 500 l of immunoprecipitation buffer containing the following (final concentration): 25 mM Tris, pH 7.4, 0.5% Nonidet P-40, 1% Triton X-100, 1 mM EDTA, 150 mM NaCl, 20 g/ml aprotinin, and 20 g/ml leupeptin. To immunoprecipitate the EGF receptor, monoclonal anti-EGF receptor antibody (EGFR1, Amersham Corp.) was added to the above mixture to reach a final concentration as 2-3 g/ml. To immunoprecipitate G s␣ , 5 l of CS1 antiserum raised against the C terminus decapeptide (8) were added. As a control for EGFR1, an irrelevant mAb BBC-4 against adenylyl cyclase (provided by Dr. Thomas Pfeuffer, University of Wü rzburg, Germany) was used. Likewise, nonimmune rabbit serum was employed as a control for CS1 antiserum. The mixture was then incubated at 4°C overnight with constant rolling. Pansorbin suspension (20 l of 10% suspension, Calbiochem), which had been prewashed with immunoprecipitation buffer, was then added, and the mixture was further incubated at 4°C for 1 h with constant rolling. The samples were centrifuged at 14,000 ϫ g for 1 min, and the resulting pellets were washed once each in 500 l of buffer containing high salt (25 mM Tris, pH 7.4, 500 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40), medium salt (25 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40), and no salt (25 mM Tris, pH 7.4, 0.1% Nonidet P-40). The final pellet was resuspended in 30 l of 2 ϫ Laemmli sample buffer, heated, and subjected to SDS-polyacrylamide gel electrophoresis. The association between G s␣ and EGF receptor was detected by employing two different approaches. First, since we have shown that the EGF receptor can phosphorylate G s␣ (12), the incubation was performed in the presence of 10 M [␥-32 P]ATP, 2 mM MnCl 2 , and 5 mM MgCl 2 . Following incubation for 1 h at room temperature, the respective proteins were immunoprecipitated with either the EGFR1 mAb or CS1 antiserum, separated by SDS-PAGE, and the phosphorylated G s␣ and EGF receptors were detected by autoradiography on Kodak X-Omat film. In the second approach, following separation of proteins in the immunoprecipitate by SDS-PAGE and after transfer onto nitrocellulose, the proteins were detected by Western analyses with either EGFR1 mAb (CS1 immunoprecipitations) or CS1 antiserum (EGFR1 immunoprecipitates) employing the Amersham ECL system. Whenever peptides were employed, these were added in the initial incubation.

RESULTS AND DISCUSSION
Our initial studies to characterize the association between EGF receptor and G s␣ were performed in isolated membranes derived from rat hearts. This system was selected because we have previously demonstrated that, in membranes isolated from rat hearts, EGF stimulates adenylyl cyclase activity via activation of G s␣ (7,8). Essentially, rat heart membranes were incubated under the conditions of adenylyl cyclase activity assay in the presence or absence of ATP, with and without EGF. Reactions were terminated by addition of lysis buffer and the EGF receptor was immunoprecipitated. Following separation of proteins by SDS-PAGE, Western analysis with anti-G s␣ antiserum, CS1, demonstrated the presence of G s␣ in the EGF receptor immunoprecipitates of incubations performed in the presence of EGF and ATP, but not in EGF receptor immunoprecipitates of incubations conducted in the absence of EGF (Fig. 1). Moreover, ATP was also required to observe association of the EGF receptor with G s␣ (Fig. 1). The presence or absence of EGF and/or ATP in the incubation did not alter immunoprecipitation of the EGF receptor (Fig. 1, bottom panel). We have previously demonstrated that the EGF receptor protein tyrosine kinase activity is important for EGF to stimulate adenylyl cyclase activity in cardiac membranes (10). These earlier findings, along with the requirement for ATP and EGF to observe EGF receptor association with G s␣ (Fig. 1), suggest that the activation of the EGF receptor and perhaps autophosphorylation of the receptor alter its conformation to make sites accessible for association with G s␣ . Indeed, Cadena et al. (21) have shown that, upon activation and autophosphorylation of the EGF receptor, the receptor undergoes a conformational change from a compact to a more extended form.
To further characterize the association between EGF recep-tor and G s␣ and to define the regions of the receptor which interact with the G protein, additional experiments were performed employing the yeast two-hybrid system (22). Yeast cells (HF7c strain) were transformed with plasmids pGAD424-EGFR C or pGBT9-EGFR C and pGBT9-G s␣ or pGAD424-G s␣ , respectively; these EGF receptor constructs expressed only the cytosolic region of the receptor (Met 644 -Ala 1186 ) as chimeric proteins with either the activating (pGAD424 plasmid) or binding (pGBT9 plasmid) domains of GAL4 gene product. Controls were performed with either the two plasmids alone (not shown), or with the chimeric construct in one plasmid and the other plasmid devoid of either the EGF receptor or G s␣ cDNA ( Fig. 2A). Since the plasmids pGAD424 and pGBT9 encode for the LEU2 and TRP1 cDNAs, respectively, growth of cells on Leu Ϫ /Trp Ϫ medium indicated the successful transformation of cells with both plasmids (right-hand panels in Fig. 2, A-C). Likewise, since the GAL4 activating and binding domains have to come in proximity of each other to initiate transcription of the HIS3 and lacZ genes, growth of transformants on Leu Ϫ / Trp Ϫ /His Ϫ medium and expression of ␤-galactosidase activity indicated a productive interaction between the proteins being tested. As shown in Fig. 2 (panels A-C), transformation of yeast HF7c cells with the various constructs in plasmids pGAD424 and pGBT9 allowed comparable growth of cells on Leu Ϫ /Trp Ϫ medium. However, when these same transformants were grown on Leu Ϫ /Trp Ϫ /His Ϫ medium, growth to different extents was observed in only some of the transformants. Thus, in the latter medium, transformation with plasmids expressing the entire cytoplasmic region (EGFR C ) ( Fig. 2A) 1. Association of G s␣ with the EGF receptor in isolated cardiac membranes. Membranes (300 g protein) were incubated in the presence and absence of ATP (100 M) and/or EGF (100 nM) for 30 min as described under "Materials and Methods." Reactions were terminated by addition of lysis buffer and EGFR was immunoprecipitated using EGFR1 mAb. Following separation of proteins in the immunoprecipitate, Western analysis was performed with anti-G s␣ antibody, CS1 (top panel). The migration of G s␣ and IgG are indicated. The blot was stripped and subjected to Western analysis with anti-EGF receptor antibody (bottom panel) to ensure that immunoprecipitation of the EGFR was the same under different experimental conditions. gion of the EGF receptor in plasmid pGBT9 (Fig. 2B), and the G s␣ in plasmid pGAD424, robust cell growth was observed. Although growth of cells on Leu Ϫ /Trp Ϫ /His Ϫ medium was observed when the EGF receptor cytoplasmic domain (EGFR C ) or the N-terminal 64 amino acids of this region (EGFR CJM ) and G s␣ were expressed in the reciprocating plasmids, the yeast did not grow as well as in the former combination (Fig. 2, A and B). The better interaction between cytosolic regions of the EGF receptor as chimeras with the binding domain of GAL4 and G s␣ or constitutively active form of this G protein (G s␣ *) as a chimera with GAL4 activating domain as compared to the reciprocal chimeras of these proteins (Fig. 2C) is not unusual and similar observations have previously been reported for other proteins (reviewed in Fields and Sternglanz (23)).
In the presence of G s␣ *, as indicated by growth of yeast cells on Leu Ϫ /Trp Ϫ /His Ϫ medium, the interactions with EGF receptor cytosolic regions, EGFR C and EGFR CJM , was diminished (Fig. 2, A and B). Notably, when interactions between G s␣ and the cytosolic region of EGF receptor in which the N-terminal amino acids 645-694 had been deleted (constructs pGAD424-EGFR C⌬JM and pGBT9-EGFR C⌬JM ) were tested (Fig. 2C), no growth on Leu Ϫ /Trp Ϫ /His Ϫ medium was observed in any of the combinations of the plasmids, except the positive control with pGBT9-EGFR C and pGAD424-G s␣ . That these cells were successfully transformed with the various constructs is demonstrated by growth of all transformants on Leu Ϫ /Trp Ϫ medium (Fig. 2C, right panel).
The differences in activity of ␤-galactosidase in the various transformants were also consistent with the findings described above with growth in medium lacking histidine (Fig. 2, panel  D). Hence, both the entire cytosolic domain of the EGF receptor (EGFR C ) and the N-terminal 64 amino acids of this region (EGFR CJM ) demonstrated greater expression of ␤-galactosidase activity in the presence of G s␣ as compared to G s␣ * (Fig.  2D). Moreover, as observed for cell growth, transformation of cells with G s␣ (wild type and constitutively active) in plasmid pGAD424 and EGF receptor constructs in pGBT9 yielded higher ␤-galactosidase activities than expression of these proteins in the reciprocal vectors (Fig. 2D). Since controls performed with either EGFR cDNA constructs (EGFR C and EGFR CJM ) or G s␣ constructs paired with the empty plasmids did not grow on Leu Ϫ /Trp Ϫ /His Ϫ medium, ␤-galactosidase ac-  A and B). Individual colonies (six each) of yeast transformed with the various constructs were grown overnight in liquid medium lacking the three amino acids. Growth of cells was then continued in complete medium for 3 h and ␤-galactosidase activity was monitored as described under "Materials and Methods" section. Data are presented as mean Ϯ S.E. of six determinations. Student's unpaired t test analysis was performed to determine the significance of change between G s␣ (wild type) versus G s␣ * (active) interactions with the entire cytosolic region of the EGF receptor and its N terminus 64-amino acid constructs. tivity in these cells could not be monitored. Likewise, since transformants with plasmids constructs encoding the cytosolic region of EGF receptor without the N terminus 50 amino acids (deletion of residues 645-694; EGFR C⌬JM ) also did not grow in the absence of histidine, ␤-galactosidase activity could not be monitored. These findings in the two-hybrid system therefore suggest that the EGF receptor and G s␣ interact with each other and at least a portion of the 64 amino acids (residues 644 -707) in the juxtamembrane region of the receptor are required for this interaction. Additionally, the data from the yeast twohybrid experiments (Fig. 2, B and D) demonstrate that the juxtamembrane region of the EGF receptor interacts better with the wild type G s␣ as compared to the constitutively active form of this protein, G s␣ *.
In order to determine whether the association between the EGF receptor and G s␣ is direct (i.e. not involving other proteins) and to further delineate the region(s) on the EGF receptor which associate with G s␣ , in vitro association experiments were performed employing the purified EGF receptor and G s␣ . Since G s␣ is phosphorylated by the EGF receptor (12), experiments were performed wherein, after phosphorylation of G s␣ , either the EGF receptor or the G s␣ were immunoprecipitated, and 32 P-labeled proteins in the immunoprecipitate were detected by autoradiography following separation by SDS-PAGE. As demonstrated by the data in Fig. 3, immunoprecipitation of the EGF receptor resulted in co-immunoprecipitation of G s␣ . Likewise immunoprecipitation of G s␣ was accompanied by the presence of EGF receptor. In controls performed with either an irrelevant monoclonal antibody (BBC-4) or nonimmune serum, neither EGF receptor nor the G s␣ was immunoprecipitated (Fig. 3A). Similarly, in additional control experiments with EGF receptor or G s␣ alone in the incubation mixture, the anti-EGF receptor antibody (EGFR1) did not immunoprecipitate G s␣ , and the anti-G s␣ antiserum (CS1) did not immunoprecipitate EGF receptor (see e.g. Fig. 3B). The requirement for ATP to observe association between the EGF receptor and G s␣ in experiments with cardiac membranes (Fig. 1) would suggest that autophosphorylation of the EGF receptor is important for interaction with the G protein. Therefore, to further elucidate the role of phosphorylation of EGF receptor and/or G s␣ , if any, in association of the two proteins, incubations were performed in the presence or absence of unlabeled ATP. Following immunoprecipitation of EGF receptor, the G s␣ in the immunoprecipitate was detected by Western analysis with CS1 antiserum. The data in Fig. 3B demonstrate that the amount of G s␣ coimmunoprecipitated with EGF receptor was the same whether or not ATP was present; the phosphorylation states of G s␣ and EGF receptor in these experiments were confirmed by probing the Western blot with anti-phosphotyrosine antibodies (not shown). Thus, the data in Fig. 3B demonstrate that the association of the EGF receptor with G s␣ is independent of ATP and the phosphorylation state of the EGF receptor and G s␣ . Notably, the purified EGF receptor preparation contains EGF (ϳ5-8 M), and therefore, experiments in the absence of the growth factor are not possible. However, since the purified EGF receptor which is not tyrosine phosphorylated does not require ATP to associate with G s␣ , it would appear that the ligand bound, nonphosphorylated, pure EGF receptor is already in a conformation in which the site(s) on the receptor are accessible for binding the G protein. Conversely, in intact cardiac membranes, the presence of ATP is required to observe EGF-dependent association of the G protein with the receptor (Fig. 1). These latter data suggest that in its native membrane environment, as demonstrated by Cadena et al. (21), the EGF receptor changes its conformation upon autophosphorylation and assumes a structure in which the G s␣ binding sites are accessible.
The experimental strategy described in Fig. 3A also allowed the determination of stoichiometry of the EGF receptor and G s␣ association. In these studies, the EGF receptor (33 ng) and G s␣ (500 ng) were incubated in the presence of [␥-32 P]ATP, and phosphorylation of both proteins was allowed to reach stoichiometry as described previously (12). An aliquot of the incubations was directly applied to SDS-PAGE, and following separation of known amounts of proteins, the stoichiometry of phosphorylation of each protein was calculated. This also allowed the determination of specific radioactivity of each protein. Consistent with our previous report (12) the stoichiometry of phosphorylation of G s␣ was 2 mol of P i /mol of G s␣ , and that of the EGF receptor was 4.5 mol of P i /mol of receptor. The remainder of the incubation was immunoprecipitated with the anti-EGF receptor antibody, and following separation of proteins by SDS-PAGE, the radioactivity associated with EGF receptor and G s␣ was determined. By dividing the radioactivity associated with EGF receptor and G s␣ in the immunoprecipitate by the specific radioactivity of each protein (obtained from the stoichiometry of phosphorylation), the amount of each protein in the immunoprecipitate was calculated. By this method, the stoichiometry of association between the two proteins was determined to be 0.9 mol of G s␣ associated with 1 mol of EGF receptor, i.e. the association between the receptor and G s␣ is stoichiometric. Because the phosphorylation states of the EGF receptor and G s␣ were not important for association of these proteins (Fig. 3B), all subsequent studies were performed in the absence of ATP. The juxtamembrane region of the EGF receptor activates G s (11) and this region of the receptor has also been reported to bind proteins and determine substrate specificity with respect to the proteins that are phosphorylated (13)(14)(15). Moreover, our data from the two-hybrid assay suggested that the juxtamembrane region of the EGF receptor is important for association with G s␣ (Fig. 2). Therefore, additional in vitro association studies were aimed at elucidating whether or not the juxtamembrane region of the EGF receptor is important for physical interactions with G s␣ . In our approach we employed peptides EGFR-13 and EGFR-14, which correspond to amino acids 645-657 and 679 -692 in the juxtamembrane domain of the EGF receptor (11). We have previously demonstrated that EGFR-13 is a potent activator of G s (11). As demonstrated by the data in Fig. 4A, in the presence of EGFR-13 and EGFR-14 the amount of G s␣ which co-immunoprecipitated with the EGF receptor was markedly diminished. Densitometric analyses demonstrated a 95% decrease in association in the presence of EGFR-13 and EGFR-14. This decrease in EGF receptor-G s␣ association was not the result of decreased EGF receptor immunoprecipitation, since the latter remained constant as determined by Western analysis of the immunoprecipitate with the anti-EGF receptor antibody (Fig. 4B). Previously we have shown that phosphorylation of EGFR-13 on threonine residue corresponding to Thr 654 in the EGF receptor diminishes its ability to activate G s and stimulate adenylyl cyclase (11). Likewise, phosphorylation of EGFR-13 (P-EGFR-13) also abolished the ability of the peptide to compete for association of G s␣ with the EGF receptor (Fig. 4D). These findings demonstrate that the effects of EGFR-13 are specific. Since both EGFR-13 and EGFR-14 are basic peptides, additional controls were per-formed with polylysine and polyarginine (Fig. 4D). Neither polyarginine nor polylysine competed for the association between EGF receptor and G s␣ (Fig. 4D). These findings, along with the observation that neither a peptide corresponding to amino acids 985-996 of the EGF receptor (Fig. 4C) nor three other peptides corresponding to sequences in G s␣ (discussed later) decreased the association of EGF receptor with G s␣ , demonstrate that the effects of EGFR-13 and EGFR-14 are specific. Thus the data with EGFR-13 and EGFR-14 demonstrate that the juxtamembrane region of the EGF receptor (amino acids 645-692) is important for association with G s␣ . Previously we demonstrated that micromolar concentrations of EGFR-13 are required to activate G s (11). However, in the present study millimolar concentrations of the peptide are required to compete for the association of EGF receptor with G s␣ (Fig. 4, A and C). This difference in concentration probably relates to the fact that in the G s activation studies (11) only the peptide and G s were present. On the other hand, in the present study, the peptide is being utilized to compete for association between the receptor and G s␣ . Thus it would appear that the affinity of the full-length EGF receptor for G s␣ is high, and therefore, high concentrations of a peptide corresponding to a sequence within the receptor are required for competition. Previously we have also shown that the 13-amino acid region (EGFR-13; amino acids 645-657), but not the 14-amino acid region (EGFR-14; amino acids 679 -692), is important for activation of G s by the EGF receptor (11). This coupled with the competition studies with the peptides EGFR-13 and EGFR-14 (Fig. 4, A and C) suggest that within the juxtamembrane domain of the EGF receptor, regions which activate G s (e.g. amino acids 645-657) (11) as well as other regions not involved in FIG. 4. Juxtamembrane region of the EGF receptor is important for association of EGF receptor with G s␣ . Panel A, purified EGF receptor and G s␣ were incubated in the absence of ATP and with or without the peptides EGFR-13 and EGFR-14. Proteins were immunoprecipitated with the anti-EGF receptor mAb EGFR1. Proteins in the immunoprecipitate were detected by anti-G s␣ antiserum CS1. Panel B, the Western blot in panel A was probed with mAb EGFR1 to ensure equal immunoprecipitation of the EGF receptor. Panel C, experiments were performed as described for panel A, except the ability of the peptide corresponding to the residues 985-996 of the EGF receptor to decrease association of EGF receptor and G s␣ was also tested. Additionally, the ability of the C terminus truncated EGF receptor (⌬1022-1186) to co-immunoprecipitate G s␣ was also compared with the full-length EGF receptor. Panel D, the ability of 1 mM each of peptides EGFR-13, phospho-EGFR-13 (P-EGFR-13) in which the threonine residue corresponding to Thr 654 in the EGF receptor was phosphorylated, polylysine, and polyarginine to compete for EGF receptor-G s␣ association was tested as described for experiments in panel A.
Although the studies with peptides corresponding to the juxtamembrane region of the EGF receptor suggest that this region of the receptor is important for association with G s␣ (Fig.  4, A and C) these experiments do not completely rule out the participation, if any, of the C terminus region of the receptor that harbors the autophosphorylation sites (24 -26) to which SH2 domain-containing proteins such as Shc and phospholipase C␥ bind (3,4). Therefore, to determine whether or not the C terminus of the EGF receptor is involved in G s␣ association, experiments were performed with the purified fulllength and truncated EGF receptor (⌬1022-1186). In the truncated receptor all amino acids after threonine 1022 are deleted. In these studies equal amounts of the full-length receptor and truncated receptors were employed. As shown in Fig. 4C, the amount of G s␣ co-immunoprecipitated with the truncated receptor was the same as that associated with the full-length EGF receptor. Moreover, a peptide corresponding to amino acids 985-996 of the EGF receptor did not decrease association between EGF receptor and G s␣ (Fig. 4C). Therefore, the data in Fig. 4 demonstrate that the juxtamembrane region, but not the C terminus region, of the EGF receptor is involved in its association with G s␣ .
The importance of the juxtamembrane region of the EGF receptor in association with G s␣ (Fig. 4) and activation of G s by this region (11) would suggest that upon activation of G s , the ␣-subunit does not associate with the receptor. Although some support for this contention is provided by the data with yeast two-hybrid assay (Fig. 2, A, B, and D), additional experiments with purified proteins were performed to determine whether or not the active, GTP␥S-bound, form of G s␣ associates with the EGF receptor. As demonstrated by the data in Fig. 5, in immunoprecipitates of the EGF receptor, in the presence of GTP␥S (1 M), the co-immunoprecipitation of G s␣ was reduced by an average of 85%. GTP␥S did not alter the immunoprecipitation of the EGF receptor (not shown). Notably, the effects of GTP␥S are specific since neither GDP (1 M, Fig. 5) nor another nucleotide such as ATP (Fig. 3) affected the co-immunoprecipitation of G s␣ with the EGF receptor. The data in Fig. 5 are also consistent with the findings from the two-hybrid assay which compared the association of the wild type and constitutively active G s␣ with EGF receptor (Fig. 2) and demonstrate that upon activation, G s␣ is no longer associated with the EGF receptor.
Employing chimeric proteins, antibodies, and peptides, several studies have shown that the C terminus region of G s␣ is important for activation of the G protein by receptors (see e.g. Refs. 8,24,and 25). Indeed, employing CS1 antiserum, which is directed against the C terminus decapeptide of G s␣ , we have previously shown that this region of G s␣ is important for its activation by the EGF receptor and ␤-adrenergic receptors (8). More recently, employing peptides, studies from Hamm's laboratory have shown that the activation of G s␣ by ␤-adrenergic receptors can be obliterated by peptides corresponding to the C terminus of G s␣ . Therefore, to determine whether or not the C terminus regions of G s␣ is also important for association with EGF receptor, in experiments similar to those described in Fig.  4, the ability of peptides corresponding to amino acid residues 15-29, 354 -372, and 371-380 of G s␣ to compete for the association of the EGF receptor with G s␣ was investigated. Although the C terminus of G s␣ is important for its activation by EGF receptor in terms of stimulation of adenylyl cyclase (8), none of the peptides tested altered the association of G s␣ with the EGF receptor (data not shown). These data (not shown) indicate that the region of the G s␣ molecule which is involved in association is not necessarily the same region that is important for activation of this G protein by receptors in terms of mediating a signal. Moreover, our findings that the three peptides corresponding to sequences in G s␣ at concentrations as high as 1 mM did not alter association of the EGF receptor with G s␣ (not shown) indicate that the effects of EGFR-13 and EGFR-14 as observed in Fig. 4 are specific. Three-dimensional structure studies of the transducin and G i␣ proteins have demonstrated that the N terminus of the ␣-subunits of heterotrimeric G proteins interacts with the ␤␥-subunits (26,27). Additionally, the N terminus region of ␣-subunits of G proteins has been implicated to interact with receptors (26). However, the inability of a peptide corresponding to amino acids 15-29 of G s␣ to compete for association between the EGF receptor and G s␣ (not shown) suggests that the region of G s␣ , which is important in interactions with ␤␥-subunits (26,27) and which may interact with receptors, is not the region important in association with the EGF receptor. Given our observation that GTP␥S decreases the association of G s␣ with the EGF receptor and the findings from the three-dimensional structures of G protein ␣-subunits (28,29) that GTP binding most markedly changes the conformation in the three switch regions (28,29), it is tempting to speculate that perhaps one of the switch regions and/or a domain proximal to this switch is involved in association with the EGF receptor. Whatever the case, presently, the precise region of G s␣ which associates with EGF receptor remains unknown and forms the subject of further investigations.
The association of single transmembrane protein tyrosine kinase receptors with adapter proteins that participate in the signaling to serine/threonine kinases has been well documented (see Refs. 30 and 31 for reviews). However, the interactions of this family of receptors with G protein-mediated processes has been less well defined. Likewise, although a large number of studies have demonstrated that regions in G protein ␣-subunits or domains in heptahelical receptors which activate these G proteins are important for transducing signals (8, 24 -29, 32, 33), to date the association of G protein ␣-subunits with heptahelical receptors has only been demonstrated for the angiotensin AT2 receptor (34). However, even in the case of the angiotensin AT2 receptor, the region of the receptor that asso- FIG. 5. GTP␥S decreases the association of G s␣ with the EGF receptor. Panel A, purified EGF receptor and G s␣ were incubated in the presence of GDP or GTP␥S (1 M each) as described in Fig. 3. The EGF receptor was immunoprecipitated with EGFR1 mAb and G s␣ coimmunoprecipitation was detected with CS1 antiserum. Panel B, densitometric analyses of the amount of G s␣ co-immunoprecipitated with the EGF receptor in the presence and absence of GTP␥S. Data are the mean Ϯ S.E. of three determinations. ciates with G i␣2 and G i␣3 remains unknown (34). Moreover, whether or not the association of AT2 receptor with G i␣2 and G i␣3 is direct or involves another protein also remains to be investigated. In this respect, the data presented in this study are the first to show the direct, and stoichiometric, association of a G protein ␣-subunit with a receptor which can activate this G protein (8,11,12). Although the association of G i␣ with EGF receptor has been suggested (35), in that study, phospholipase C␥ was shown to be associated with EGF receptor as well as G i␣ . Therefore, whether the association of G i␣ with the EGF receptor is direct or indirect via phospholipase C␥ remains unknown. Similarly, although Nishimoto et al. (36) have demonstrated the association of G o with amyloid precursor protein (APP), heterotrimeric G o was demonstrated to be associated with APP, and therefore, whether or not the G o␣ -subunit associates directly with APP is also not known. Since the regions within the EGF receptor and the ␤-adrenergic receptors that activate G s are similar (11,32) and because the region (EGFR-13) in the EGF receptor that activates G s also modulates association of the receptor with G s␣ , it is tempting to speculate that the similar motif in the third cytosolic loop of the ␤-adrenergic receptor would also be involved in association of that receptor with G s␣ .
Interestingly, both EGFR-13 and EGFR-14 compete for the association between EGF receptor and G s␣ . Since the association of the two proteins is stoichiometric (1 mol of G s␣ :1 mol of EGF receptor), it is possible that there are at least two contact sites for G s␣ on the EGF receptor; one site at amino acids 645-657 (EGFR-13) and the other at amino acids 679 -692 (EGFR-14). Moreover, because the addition of either EGFR-13 or EGFR-14 effectively decreased the association between EGF receptor and G s␣ , it would appear that the loss of contact of G s␣ at one of the two sites on the EGF receptor diminishes the affinity for the other site. Alternatively, in the presence of one of the peptides corresponding to the EGF receptor juxtamembrane domain, the conformation of the G s␣ is altered so that it loses the ability to bind at any point on the juxtamembrane region of the EGF receptor. The identification of the critical amino acid residues involved in association with the G s␣ will be facilitated by future experiments involving site-directed mutagenesis of the receptor.
In conclusion, the studies described herein demonstrate that the cytosolic, juxtamembrane region of the EGF receptor encompassing sequences corresponding to EGFR-13 and EGFR-14 (Ϫ48 amino acids; Arg 645 -Lys 692 ) is important for the direct, and stoichiometric, association with G s␣ . Since the first 13 amino acids in this region (Arg 645-657 ) are also important for activation of G s (11) and because the EGF receptor protein tyrosine kinase can phosphorylate G s␣ on tyrosyl residues (12), it is possible that association of G s␣ with the juxtamembrane region of the EGF receptor is important for its activation. Further support for the latter contention is derived from the observation that expression of the constitutively active form of G s␣ in the two-hybrid assay (Fig. 2) or the addition of GTP␥S (Fig. 5) decreases the interaction of EGF receptor with G s␣ . Another interesting finding from the experiments described herein is that the region on G s␣ that is involved in association with the EGF receptor is different from the region (C termi-nus), which is important for activation of G s␣ by receptors in terms of mediating signals to adenylyl cyclase (8). The definition of the precise regions on G s␣ which associate with the EGF receptor forms the subject of future experiments.