Tyrosine Phosphorylation of Gsα and Inhibition of Bradykinin-induced Activation of the Cyclic AMP Pathway in A431 Cells by Epidermal Growth Factor Receptor

An increasing amount of experimental data suggest that cross-talk exists between pathways involving tyrosine kinases and heterotrimeric G proteins. In a previous study, we demonstrated that bradykinin (BK) increases the intracellular accumulation of cAMP in the human epidermoid carcinoma cell line A431 by stimulating adenylate cyclase activity via a stimulatory G protein (Gsα) (Liebmann, C., Graneß, A., Ludwig, B., Adomeit, A., Boehmer, A., Boehmer, F.-D., Nürnberg, B., and Wetzker, R. (1996) Biochem. J. 313, 109-118). Here, we present several lines of evidence indicating the ability of epidermal growth factor (EGF) to suppress BK-induced activation of the cAMP pathway in A431 cells via tyrosine phosphorylation of Gsα. Gsα was specifically immunoprecipitated from A431 cells using the anti-αs antiserum AS 348. Tyrosine phosphorylation of Gsα was detectable in EGF-pretreated cells with monoclonal anti-phosphotyrosine antibodies. Additionally, A431 cells were labeled with [32P]orthophosphate in vivo and treated with EGF, and the resolved immunoprecipitates were subjected to amino acid analysis. The results clearly indicate that EGF induces tyrosine phosphorylation of Gsα in A431 cells. Treatment of A431 cells with EGF decreased BK-induced cAMP accumulation in intact cells as well as the stimulation of adenylate cyclase by BK, NaF, and guanyl nucleotides, but not by forskolin. Also, EGF treatment abolished both the BK- and isoprenaline-induced stimulation of guanosine 5′-O-(3-[35S]thiotriphosphate) binding to Gsα. In contrast, the BK-evoked, Gq-mediated stimulation of inositol phosphate formation in A431 cells was not affected by EGF pretreatment. Thus, EGF-induced tyrosine phosphorylation of Gsα is accompanied by a loss of its susceptibility to G protein-coupled receptors and its ability to stimulate adenylate cyclase via guanyl nucleotide exchange. We propose that Gsα may represent a key regulatory protein in the cross-talk between the signal transduction pathways of BK and EGF in A431 cells.

There is mounting evidence indicative of complex, probably cell-specific interactions between signaling pathways involving heterotrimeric G proteins and tyrosine kinases. For example, the stimulation of G protein-coupled receptors modulates key proteins of the mitogen-activated protein kinase pathway via protein kinase C-or protein kinase A-mediated phosphorylation on serine or threonine residues (1)(2)(3). Furthermore, several isoforms of ␣ subunits of G proteins were shown to be phosphorylated in vitro on tyrosine residues by tyrosine kinase receptors such as the epidermal growth factor (EGF) 1 receptor and the insulin receptor or by non-receptor tyrosine kinases of the Src kinase family. EGF was shown to activate cardiac adenylate cyclase via a mechanism requiring both G s␣ and the EGF receptor tyrosine kinase (4,5). EGF was also found to stimulate phospholipase C in rat hepatocytes (6) or phospholipase A 2 in rat kidney (7) in a pertussis toxin-sensitive manner, suggesting an involvement of G i proteins. The molecular mechanism of these receptor tyrosine kinase-mediated effects remained unclear. In reconstituted phospholipid vesicles, the insulin receptor was found to catalyze tyrosine phosphorylation of G i␣ and G o␣ , suggesting the possibility of a direct interaction of receptor tyrosine kinases and G proteins (8). Further progress in this field came from in vitro studies of Hausdorff et al. (9) on direct interactions between the non-receptor tyrosine kinase pp60 c-src and purified G protein ␣ subunits. pp60 c-src was shown to phosphorylate recombinant G s␣ as well as other G ␣ isoforms on tyrosine residues almost stoichiometrically (9). In 1995, the in vitro sites of phosphorylation of G s␣ by pp60 c-src were identified (10). The phosphorylated tyrosine residues are located at N-terminal position 37 and at C-terminal position 377 of G s␣ and thus in regions known to participate in nucleotide exchange and receptor interaction (10,11). Very recently, using purified EGF receptor and recombinant G s␣ , Poppleton et al. (12) demonstrated an activation of G s␣ via phosphorylation on tyrosine residues by EGF receptor kinase. However, little is known about tyrosine phosphorylation of G s␣ in intact cells and how such phosphorylation might affect the function of the G protein. We have recently shown that bradykinin (BK) activates dual pathways in A431 human epidermoid carcinoma cells, i.e. the phospholipase C-␤/protein kinase C pathway and, independently via G s␣ , the cyclic AMP/protein kinase A pathway, which represents a negative feedback loop to the BKinduced protein kinase C activation (13). At the same time, as observed earlier by Hosoi et al. (14), the BK-induced activation of protein kinase C leads to an increased serine/threonine phosphorylation of EGF receptors and, subsequently, to a reduced binding of EGF. These findings prompted us to investigate whether EGF is able to affect (vice versa) the BK signaling pathways in A431 cells.
In this paper, we present experimental evidence that EGF treatment of A431 cells results in both inhibition of BK-induced [ 35 S]GTP␥S binding to G s␣ and BK-induced stimulation of adenylate cyclase activity in A431 membranes as well as inhibition of BK-induced cAMP accumulation in intact A431 cells. Furthermore, we show that EGF stimulation leads to specific tyrosine phosphorylation, but also to an increase in serine/ threonine phosphorylation of G s␣ . This is the first report demonstrating possible functional consequences of G s␣ tyrosine phosphorylation in intact cells.

EXPERIMENTAL PROCEDURES
Cell Culture-A431 human epidermoid carcinoma cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Life Technologies, Inc.) supplemented with 4.5 g/liter glucose, 2 mM glutamine, 7.5% fetal calf serum, and antibiotics. Treatment of intact cells with BK and various agents as outlined in the figure legends was performed with nearly confluent cultures.
Membrane Preparations-An A431 particulate fraction (referred to as "membranes") was prepared as described before (13). Protein concentration was determined according to Lowry et al. (15) or Bradford (16) with bovine serum albumin as a standard.
Adenylate Cyclase Assay-The activity of adenylate cyclase in A431 membranes was determined according to Schultz and Jakobs (17) with slight modifications. Briefly, membranes (80 g of protein/assay) were incubated for 20 min at 25°C in a standard mixture containing [␣-32 P]ATP (ϳ2 ϫ 10 7 cpm/tube), 5 mM MgCl 2 , 1 mM 3-isobutyl-1methylxanthine, 0.1 mM cAMP, 0.1 mM ATP, 5 mM creatine phosphate, 150 units of creatine phosphokinase, 1 mg/ml bovine serum albumin, and 50 mM HEPES, pH 7.2. To inactivate angiotensin-converting enzyme, which degrades BK, the incubation mixture was supplemented with 1 M captopril. Variable additions such as NaF (as [AlF 4 ] Ϫ ), forskolin, GTP␥S, and BK were added as indicated, giving a total assay volume of 100 l. The samples were preincubated for 5 min at 25°C, and the reaction was started by adding the membranes. The incubation was terminated by the addition of 400 l of ZnCl 2 solution (125 mM) followed by 500 l of Na 2 CO 3 solution (125 mM). The samples were centrifuged for 5 min at 10,000 ϫ g, and 800 l of the supernatant fluid were transferred to alumina oxid 90 (E. Merck, Darmstadt, Germany)containing columns. After draining of the sample, labeled cAMP was eluted by the subsequent addition of 2 ϫ 2 ml of Tris-HCl buffer (100 mM), pH 7.5, and [ 32 P]P i was determined by measuring Cerenkov radiation. For each column, cAMP recovery was estimated individually using [ 3 H]cAMP.
Determination of Intracellular Cyclic AMP-A431 cells in 24-well plates were treated with serum-free DMEM overnight, and then the medium was changed to HEPES-buffered serum-free DMEM, which had been adjusted to pH 7.2 just prior usage. Then, cells were exposed to the agents tested in 500 l of serum-free medium supplemented with 100 M 3-isobutyl-1-methylxanthine and 10 M captopril. The reaction was stopped by the addition of 1 ml of ice-cold ethanol (96%, v/v), giving a final concentration of 65% (v/v). The cells were scraped off the plates, and the ethanolic extract was centrifuged at 14,000 ϫ g for 6 min at room temperature. The supernatants containing the extracted cAMP were removed, and the pellets were washed with 500 l of ethanol (65% v/v) and centrifuged as described above. Supernatants were pooled, evaporated to dryness at 60°C, and resolved in 600 l of 0.05 M acetate buffer. Two samples were taken for estimation of cAMP concentration using the cAMP 125 I-labeled scintillation proximity (non-acetylation) assay from Amersham. For each well, the protein content was determined (16).
[ 35 S]GTP␥S Binding-Binding of [ 35 S]GTP␥S to A431 membranes was determined as described previously (18). The reaction mixture contained [ 35 S]GTP␥S (10 5 cpm/assay), 1 M GDP, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 10 M captopril, 50 mM HEPES, pH 7.2, and 15 g of membrane protein in a total volume of 200 l. Further details are provided in the figure legends. The incubation was started by the addition of the membranes and was carried out in quadruplicates for 45 min at 4°C (equilibrium conditions). The reaction was terminated by rapid filtration through Whatman GF/C glass-fiber filters under vac-uum. The filters were washed with 3 ϫ 2 ml of 50 mM HEPES, pH 7.2, containing 5 mM MgCl 2 . The filters were dried and counted for radioactivity. Nonspecific binding was determined in the presence of 10 M unlabeled GTP␥S and represents ϳ40% of total binding.
Determination of Total Inositol Phosphates-BK-induced formation of inositol phosphates in A431 cells was determined as described by Tilly et al. (19). Briefly, cells were prelabeled with 4 Ci/ml [ 3 H]inositol for 24 h in inositol-free DMEM. 2 h prior to stimulation, the medium was changed to serum-free DMEM containing 20 mM HEPES, pH 7.2. Then, the cells were stimulated for 10 min with bradykinin at the concentrations indicated in the presence of 10 mM LiCl 2 . The reaction was terminated by replacing the medium with 1 ml of 10% trichloroacetic acid. Inositol phosphate-containing extracts were washed four times with 2 volumes of water-saturated diethyl ether, neutralized by adding Tris base, diluted to 4 ml with distilled water, and placed on AG 1-X8 columns (200 -400 mesh, formate form, Bio-Rad). The columns were subsequently eluted with 2 ml of water and 5 ml of 60 mM ammonium formate and 5 mM sodium tetraborate, followed by 2 ml of 1.0 M ammonium formate and 0.1 M formic acid for five times, yielding the inositol phosphate fraction. Radioactivity was measured using a Flow-Scint IV scintillator (Packard Instrument Co.).
Immunoprecipitation and Immunoblotting of ␣ s Subunits-Antiserum AS 348 was raised against the peptide sequence RMML-RQYELL, corresponding to C-terminal region 385-394 of ␣ s and characterized as described elsewhere (20,21). RC20 monoclonal antiphosphotyrosine antibodies (Transduction Laboratories) were purchased from Dianova (Hamburg, Germany). Subconfluent A431 cells were preincubated in serum-free DMEM overnight. Then, the medium was changed to fresh serum-free DMEM, and the cells were stimulated with EGF (100 ng/ml) at 37°C in the absence or presence of other additions and for various lengths of time as indicated, followed by the preparation of membranes. Immunoprecipitation of G s proteins was performed according to Laugwitz et al. (22). For several experiments, AS 348 antibodies were covalently coupled to protein A-Sepharose by means of dimethyl pimelimidate. Briefly, 200 l of AS 348 antiserum (or nonimmune serum as a control) and 200 l of protein A-Sepharose (12.5 mg of beads) were incubated for 2 h and subsequently washed three times with phosphate-buffered saline, pH 7.4, and twice with 0.2 M sodium borate, pH 9.0. The beads were resuspended in 0.1 M borate buffer. The antibodies were cross-linked to the beads by adding 5.2 mg of dimethyl pimelimidate and mixing for 30 min at room temperature. Thereafter, the beads were washed with 0.2 M ethanolamine, pH 8.0, and the incubation was continued with 0.2 M ethanolamine, pH 8.0, for 2 h at room temperature. Finally, the beads were washed two times with 0.1 M glycine, pH 3.0, and three times with phosphate-buffered saline. The covalently coupled antibodies were stored in phosphatebuffered saline, pH 7.4, with 0.02% sodium azide at 4°C. Membranes were solubilized in 40 l of 2% (w/v) SDS for 10 min at room temperature. Thereafter, 120 l of precipitating buffer containing 1% (w/v) Triton X-100, 1% (w/v) deoxycholate, 150 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 100 M sodium orthovanadate, and 10 mM Tris-HCl, pH 7.4, were added. To remove insoluble material, the solubilized membranes were centrifuged at 4°C and 12,000 ϫ g for 10 min. Covalently coupled antiserum AS 348 (20 l) or nonimmune serum as a control was added to the supernatants, and the samples were incubated at 4°C for 4 h under constant rotation. Thereafter, the beads were pelleted (14,000 ϫ g, 10 s) and washed twice with 1 ml of washing buffer A containing 1% (w/v) Nonidet P-40, 0.5% (w/v) SDS, 600 mM NaCl, and 50 mM Tris-HCl, pH 7.4, and twice with 1 ml of washing buffer B containing 300 mM NaCl, 10 mM EDTA, and 100 mM Tris-HCl, pH 7.4. The Sepharose beads were resuspended in 40 l of SDS sample buffer, heated for 10 min at 100°C, and centrifuged, and the supernatant was subjected to SDS-polyacrylamide gel electrophoresis (PAGE) on 10% (w/v) acrylamide gels and transferred onto nitrocellulose filters. The blots were further processed as described previously (22) for AS 348 antibodies and according to the instructions of the manufacturer in the case of RC20 anti-phosphotyrosine antibodies.
Metabolic Labeling of A431 Cells and Immunoprecipitation of ␣ s Subunits-Subconfluent A431 cells in 6-well plates (Nunc) were depleted of serum for 24 h. Then, the medium was changed to phosphatefree DMEM medium. After the addition of 0.3 mCi of [ 32 P]orthophosphate/ml of medium, cells were incubated for 3 h at 37°C and 7.5% CO 2 in a humidified atmosphere. After labeling of cells, they were exposed to EGF (100 ng/ml) for 5 min (or not, as indicated), washed twice with 0.5 ml of phosphate-buffered saline containing 0.1% (w/v) bovine serum albumin, and solubilized at 4°C in 160 l/well lysis buffer containing 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% (w/v) Triton X-100, 1% (w/v) deoxycholate, 0.5% (w/v) SDS, 1 mM dithiothreitol, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 0.1 mM sodium vanadate, and 0.1% (w/v) bovine serum albumin. The lysate was cleared by centrifugation at 12,000 ϫ g and 4°C for 10 min. Metabolically labeled G s␣ phosphoprotein was immunoprecipitated by protein A-Sepharosebound anti-␣ s antiserum AS 348 as described above. Immunocomplexes were washed twice with both buffers A and B, heated with SDS sample buffer, and subjected to 10% SDS-PAGE. The ␣ s protein bands were localized in the fixed dried gels by exposure to Biomax film (Eastman Kodak Co.) for 4 h at Ϫ80°C using an intensifying screen.
Phosphoamino Acid Analysis-The section of the dried SDS-polyacrylamide gels corresponding to the position of G s␣ (ϳ100 -200 cpm) was extracted; the protein was precipitated and hydrolyzed; and phosphoamino acid analysis was performed by two-dimensional separation on thin-layer cellulose plates as described by Boyle et al. (23). The thin-layer plates were analyzed using a Bio-Rad Model GS250 Molecular Imager and prolonged exposure (3-4 days) on BI imaging screens.

EGF Counteracts BK-induced Cyclic AMP Accumulation, but Not BK-induced Formation of Inositol Phosphates, in A431
Cells-In intact A431 cells, BK elicited a concentration-dependent increase in intracellular cAMP up to ϳ140% of the basal level (100%) after 20 min of stimulation (Fig. 1). Halfmaximal effects were seen at ϳ3 nM BK. When the cells were pretreated with EGF (100 ng/ml, 5 min), the ability of BK to enhance cAMP accumulation was significantly inhibited (Fig.  1, inset, bar D versus bar B), whereas EGF pretreatment itself had no significant effect on the basal cAMP level (inset, bar C versus bar A). In contrast, in EGF-treated A431 cells, BK further stimulated inositol phosphate formation, but at a higher level. This was probably due to additional stimulation of the phospholipase C-␥ isoform by EGF (Fig. 2). Thus, EGF pretreatment of A431 cells seems to inhibit the stimulatory BK effect on the cAMP system, but not phosphoinositide breakdown, which are activated via the separate G proteins G s and G q , respectively (13,19).
Effect of EGF on Stimulation of Adenylate Cyclase Activity in A431 Membranes-Adenylate cyclase activity was measured in A431 membranes prepared from cells after treatment with EGF (100 ng/ml, 5 min). In these membranes, indeed, the stimulation of adenylate cyclase activity by NaF (0.1 mM), GTP␥S (10 M), Gpp(NH)p (10 M), or BK (1 M) was reduced compared with the effects in membranes prepared from A431 cells not treated with EGF (Fig. 3). The stimulatory effect of forskolin on adenylate cyclase did not significantly differ under either condition (Fig. 3). These results suggest that EGF interferes with activators of adenylate cyclase at the level of the G s␣ protein, which is involved in the activation of adenylate cyclase activity by BK in A431 cells (13).
Influence of EGF Pretreatment of A431 Cells on Functional Activation of G s␣ by BK-As shown in Fig. 4 first increase, half-maximal and maximal stimulations (up to 155% of control) were observed with ϳ0.3 and 1 nM BK, respectively. At higher BK concentrations, [ 35 S]GTP␥S binding was continuously reduced, but was followed by a second increase at BK concentrations of ϳ30 nM for half-maximal binding and 100 nM for maximal binding (up to 175% of control). The first as well as the second increase in [ 35 S]GTP␥S binding induced by BK were completely abolished in the presence of the bradykinin B 2 receptor antagonist Hoe 140, indicating that both effects are mediated via the same BK receptor type (Table I). The reasons for the biphasic curve shape in the BK-stimulated [ 35 S]GTP␥S binding to A431 membranes are not yet known. It should be noted that the basal binding of [ 35 S]GTP␥S to membranes prepared from EGF-treated cells was significantly enhanced (24.0 Ϯ 1.0 fmol/mg of protein; n ϭ 6) compared with that to membranes from untreated cells (18.5 Ϯ 2.2 fmol/mg of protein). To identify that part of the complex pattern of BKinduced [ 35 S]GTP␥S binding to A431 membranes that may correspond to the activation of ␣ s subunits, we studied the effect of anti-␣ s antiserum AS 348 in this assay. AS 348 was raised against the C-terminal region of G s␣ , corresponding to amino acids 385-394 (20), and has been successfully used in A431 cells both to detect G s␣ on Western blots and to prevent the BK-induced stimulation of adenylate cyclase activity (13). Pretreatment of A431 membranes with the anti-␣ s antiserum completely abolished the second phase of [ 35 S]GTP␥S binding evoked by BK concentrations higher than 1 nM, whereas the first stimulatory phase employing subnanomolar concentrations of BK was not affected. In the presence of nonimmune serum, both the first and the second BK-induced increases in [ 35 S]GTP␥S binding were not significantly altered (Fig. 4A). Compared with the effect of BK in the presence of anti-␣ s antibodies, a similar curve shape was obtained when the effect of BK on [ 35 S]GTP␥S binding was studied in membranes pre-pared from EGF-pretreated A431 cells (Fig. 4B). In these membranes, an increase in [ 35 S]GTP␥S binding at subnanomolar BK concentrations was observed, which was attenuated at concentrations higher than 1 nM BK and completely failed at concentrations higher than ϳ3 nM BK. In addition, we studied the effect of the ␤-adrenergic agonist isoprenaline, which is known to activate G s␣ selectively, on [ 35 S]GTP␥S binding to A431 membranes. Isoprenaline was found to stimulate [ 35 S]GTP␥S binding in a concentration-dependent manner (EC 50 ϳ 20 nM), but completely failed to induce this effect in the presence of the anti-␣ s antiserum as well as in membranes prepared from EGF-treated A431 cells (Fig. 5).
EGF-induced Tyrosine Phosphorylation of G s␣ in A431 Cells-Since EGF interferes with activation of adenylate cy- clase by BK at the level of G s␣ , we investigated the possibility that the EGF effect is mediated by tyrosine phosphorylation of G s␣ . At first, we checked whether anti-␣ s antiserum AS 348, which has been successfully used for immunoprecipitation of photolabeled ␣ s subunits in human thyroid membranes (24), may also be useful for immunoprecipitation of G s␣ in A431 cells. Western blots showed that AS 348 recognizes two forms of G s␣ in A431 cells with molecular masses of 45 and 52 kDa (13). Immunoprecipitation with protein A-Sepharose-coupled AS 348 antibodies followed by immunoblotting with AS 348 clearly detected the 45-kDa ␣ s subunit. The long splice variant of ␣ s (52 kDa) was only weakly present and could be superimposed by the heavy chain of the antibodies (Fig. 6, lane 2). AS 348 antibodies blocked by the antigenic peptide (Fig. 6, lane 1) as well as nonimmune serum (lane 3) did not precipitate the 45-kDa protein, demonstrating that immunoprecipitation with the employed antiserum AS 348 is indeed specific. As shown in Fig. 7A, in membranes from EGF-pretreated cells, but not in those from untreated cells, the 45-kDa protein immunoprecipitated with AS 348 was also recognized on Western blots by monoclonal anti-phosphotyrosine antibodies. Anti-phosphotyrosine antibodies failed to detect the 45-kDa band in both membrane fractions after immunoblotting in the presence of an excess of unlabeled phosphotyrosine (Fig. 7B) or after immunoprecipitation with nonimmune serum (Fig. 7C), indicating the specificity of anti-phosphotyrosine detection. Another set of experiments demonstrated that EGF-induced tyrosine phosphorylation of G s␣ was not detectable when EGF receptor tyrosine kinase was inhibited by the EGF receptor-specific tyrphostin AG 1478 (Fig. 8). In an alternative approach, A431 cells were metabolically labeled with [ 32 P]P i and then stimulated with EGF (100 ng/ml), followed by immunoprecipitation with anti-␣ s antiserum AS 348. In AS 348 immunoprecipitates, but not in control precipitates with nonimmune serum, a 45-kDa and a 52-kDa phosphoprotein were detectable, which most likely represent the two phosphorylated isoforms of G s␣ . After treatment with EGF, immunoprecipitated G s␣ was more strongly phosphorylated (1.5-fold increase in radioactivity)   7. Detection of tyrosine-phosphorylated G s␣ with antiphosphotyrosine antibodies. A431 cells were treated with EGF or not as indicated. Membrane preparation, immunoprecipitation with anti-␣ s antiserum AS 348 or nonimmune serum, and SDS-PAGE were performed as described under "Experimental Procedures." After blotting, the nitrocellulose strips were developed with monoclonal antiphosphotyrosine (PY) antibodies (RC20). Shown are the autoluminograms of control membranes and membranes prepared from EGFpretreated A431 cells. A, immunoprecipitation with AS 348; B, immunoprecipitation with AS 348, but immunoblotting with RC20 in the presence of 1 mM phosphotyrosine (PY); C, immunoprecipitation with nonimmune serum (NIS). compared with the control membranes from untreated A431 cells (Fig. 9B). Since, in the experiments with anti-phosphotyrosine antibodies, the 52-kDa isoform was not visible, only the 45-kDa isoform was further analyzed. Phosphoamino acid analysis of the 45-kDa G s␣ revealed that, in the control cells (without EGF treatment), G s␣ was phosphorylated on serine and to some extent on threonine residues, but no phosphotyrosine was detectable. The treatment of A431 cells with EGF resulted in a de novo phosphorylation on tyrosine residues of G s␣ and, additionally, in an increase in basal phosphorylation on serine and threonine residues (Fig. 10, A and B). DISCUSSION In this study, we present evidence that, in A431 cells, EGF induces tyrosine phosphorylation of G s␣ . To our knowledge, this is the first demonstration of G s␣ tyrosine phosphorylation in vivo. The use of two different approaches to detect tyrosine phosphorylation of G s␣ produced complementary sets of data. First, monoclonal anti-phosphotyrosine antibodies are able to recognize G s␣ immunoprecipitates prepared from EGF-pretreated A431 cells. The specificity of this result is demonstrated by results from parallel experiments designed as controls: antiphosphotyrosine antibodies failed to identify G s␣ (i) in nonstimulated cells, (ii) in the presence of the EGF receptor tyrosine kinase inhibitor AG 1478, (iii) in the presence of an excess of unlabeled phosphotyrosine, and (iv) after precipitation with nonimmune serum. Second, phosphoamino acid analysis of immunoprecipitated G s␣ after in vivo labeling of A431 cells with [ 32 P]P i indicated that EGF pretreatment results in the specific appearance of phosphotyrosine, which was absent in control cells. In both control and EGF-treated cells, a relatively large amount of phosphorylated serine and threonine was detected in G s␣ . The serine and threonine phosphorylation was somewhat enhanced in G s␣ from EGF-pretreated A431 cells compared with the basal levels in nonstimulated cells.
Concomitantly with EGF-induced tyrosine phosphorylation of G s␣ , we detected effects of EGF on G s␣ function with three different approaches including the influence of EGF on BKinduced G protein activation, on cAMP accumulation, and on adenylate cyclase activation. The quantitation of [ 35 S]GTP␥S binding to membranes accounts for receptor-induced activation of G proteins. In control cells, [ 35 S]GTP␥S binding to A431 membranes was stimulated in a concentration-dependent manner by both BK and isoprenaline. The effect of isoprenaline was completely and the effect of BK was partly abolished in membranes prepared from A431 cells pretreated with EGF. In the case of BK, two lines of evidence indicate that the ␣ s -mediated FIG. 9. EGF stimulates phosphorylation of G s␣ in intact A431 cells. Cells were labeled with [ 32 P]P i as described under "Experimental Procedures" and treated or not (as indicated) with 100 ng/ml EGF for 5 min at room temperature, and cell extracts were prepared. A, shown are Western blots of A431 whole cell lysates before and after EGF treatment for 5 min. Tyrosine phosphorylation of the 170-kDa EGF receptor (EGFR) was detected with RC20 anti-phosphotyrosine antibodies. RC20 C, RC20 positive control. B, immunoprecipitation with anti-␣ s antiserum AS 348 (first and second lanes) or nonimmune serum (NIS) (third lane) was performed, and the precipitates were resolved by SDS-PAGE. Shown is a representative autoradiogram of four separate experiments.

FIG. 10. Phosphoamino acid analysis of G s␣ phosphoproteins.
Sections corresponding to the labeled G s␣ (shown in Fig. 9) were excised from the dried gels and extracted, and phosphoamino acid analysis was carried out. Separation of the phosphoamino acids by two-dimensional thin-layer electrophoresis as detected with ninhydrin is depicted in A. The relative positions of the phosphoamino acids phosphoserine (PS), phosphothreonine (PT), and phosphotyrosine (PY) are indicated above the panel. In B, the corresponding phosphoimage is shown, which was obtained upon exposure on a PhosphorImager screen for 4 days. The image was subjected to background smoothing using the program NIH Image 1.57. The result is representative of four independent experiments. stimulation of [ 35 S]GTP␥S binding is prevented specifically by EGF. In A431 cells pretreated with EGF, only the G s␣ -mediated stimulation of cAMP accumulation by BK was abolished, whereas the BK-induced stimulation of inositol phosphate formation via G q was not. These findings provide an indirect explanation that only one part of the biphasic stimulation of [ 35 S]GTP␥S binding by BK was prevented by EGF. Furthermore, using anti-␣ s antiserum AS 348, we can show that the part of BK-stimulated [ 35 S]GTP␥S binding that is missing in EGF-pretreated A431 cells exactly corresponds to that part of [ 35 S]GTP␥S binding that is likewise missing in the presence of AS 348.
In general, higher BK concentrations are necessary for both stimulation of adenylate cyclase (EC 50 ϳ 180 nM) and stimulation of [ 35 S]GTP␥S binding (EC 50 ϳ 30 nM) compared with BK-induced cAMP accumulation in intact cells (EC 50 ϳ 2 nM). On the one hand, this might be the result of very different in vitro assay conditions required for optimal BK receptor binding (e.g. pH 6.8) and optimal conditions for adenylate cyclase measurement or [ 35 S]GTP␥S binding (e.g. pH 8.0) as described previously (13,25). On the other hand, there are well known examples of dual signaling elicited by a single receptor, but different agonist concentrations. Thus, the luteinizing hormone receptor requires a 25-fold higher ligand concentration for stimulation of inositol phosphate formation compared with that for stimulation of adenylate cyclase in membranes (26). The inverse relation has been described for tachykinin receptors, where the stimulation of adenylate cyclase was less efficient (about an order of magnitude difference in the effective peptide concentrations) than that of phosphoinositide hydrolysis (27). It may therefore be supposed that either the assay conditions used favor the BK receptor-mediated activation of G q , or alternatively, G q possesses a higher affinity for the BK receptor compared with G s . Nevertheless, the observed correlation of G s␣ phosphorylation and the modulation of its function as a result of EGF stimulation suggest that both events might be causally related.
In this in vivo study, the sites of EGF-induced G s␣ tyrosine phosphorylation have not been determined. However, G s␣ has been shown before to be a substrate of pp60 c-src tyrosine kinase in vitro (9,10). In the study of Moyers et al. (10), Tyr-37 and Tyr-377, which are unique for G s␣ , were identified as the phosphorylation targets of pp60 c-src . Tyr-37 is located in the N terminus known to be involved in nucleotide exchange, and Tyr-377 is located in the C terminus, within a region of G s␣ important for receptor interaction (11). Our results suggest that, in EGF-pretreated A431 cells, the activation of G s␣ by the BK receptor as well as guanyl nucleotides is affected and thus favor both tyrosine residues as phosphorylation targets of the EGF receptor tyrosine kinase.
Very recently, Poppleton et al. (12) demonstrated for the first time the tyrosine phosphorylation of recombinant G s␣ by isolated EGF receptor tyrosine kinase. Under the in vitro conditions used, the phosphorylation of G s␣ occurred exclusively on tyrosine residues and with a stoichiometry of 2 mol of phosphate/mol of G s␣ (12). Tyrosine-phosphorylated G s␣ showed an enhanced GTPase activity, a higher [ 35 S]GTP␥S binding, and an increased degree of adenylate cyclase stimulation after reconstitution with S49 cyc Ϫ membranes (12). These data demonstrate that G s␣ is an excellent substrate for the EGF receptor tyrosine kinase and suggest that the G s␣ tyrosine phosphorylation in A431 cells observed by us is possibly induced directly by this kinase.
At first glance, some of our results obtained with A431 cells stimulated with EGF in vivo seem, however, contradictory to the findings discussed above. Thus, we did not detect an in vitro stimulation of adenylate cyclase in A431 membranes by EGF (data not shown). However, in accordance with Poppleton et al. (12), we also observed a functional activation of G s␣ by EGF reflected in the larger amount of [ 35 S]GTP␥S bound to membranes from EGF-treated A431 cells in the absence of BK. The activation of adenylate cyclase by BK and guanyl nucleotides as well as the receptor-mediated stimulation of [ 35 S]GTP␥S binding were, however, clearly decreased in membranes prepared from EGF-pretreated A431 cells. When both in vitro (12) and in vivo (this study) results are summarized, it might be assumed that the recruitment of G s␣ by EGF receptor tyrosine kinase phosphorylation leads to the stimulation of adenylate cyclase, but, simultaneously, to the inability of G s␣ to become activated by G protein-coupled receptors.
Taken together, we present evidence suggesting that, in A431 cells, a bidirectional cross-talk exists between the signal transduction pathways of BK and EGF. On the one hand, in a fast reaction, BK is able to activate protein kinase C via G q / phospholipase C-␤, subsequently leading to serine/threonine phosphorylation of the EGF receptor and thereby its desensitization (14,19). In a slow reaction, BK can counteract this protein kinase C activation via a G s␣ /adenylate cyclase-mediated pathway (13), which might allow resensitization of the EGF receptor. On the other hand, EGF is capable of inducing tyrosine phosphorylation of G s␣ and of preventing the G s␣mediated activation of the cAMP pathway by BK. Further examples of cross-talk between G protein-coupled receptor pathways and receptor tyrosine kinases have been described recently. EGF-induced tyrosine phosphorylation of protein kinase C-␦ in keratinocytes resulting in its inactivation was reported (28), suggesting a cross-talk between the EGF receptor and the protein kinase C pathway. Also very recently, Daub et al. (29) reported that the EGF receptor in Rat-1 fibroblasts is rapidly tyrosine-phosphorylated in response to the G proteincoupled receptor agonists endothelin and thrombin. These authors postulated that tyrosine kinases contribute in a general or cell-specific way to G protein receptor-mediated mitogenic signaling. Our results emphasize a differential recruitment of G s␣ by G protein-coupled receptors and the EGF receptor as a novel cross-talk mechanism.