Transmodulation of Epidermal Growth Factor Receptor Function by Cyclic AMP-dependent Protein Kinase*

Binding of epidermal growth factor (EGF) to its receptor (EGFR) augments the tyrosine kinase activity of the receptor and autophosphorylation. Exposure of some tissues and cells to EGF also stimulates adenylyl cyclase activity and results in an increase in cyclic AMP (cAMP) levels. Because cAMP activates the cAMP-dependent protein kinase A (PKA), we investigated the effect of PKA on the EGFR. The purified catalytic subunit of PKA (PKAc) stoichiometrically phosphorylated the purified full-length wild type (WT) and kinase negative (K721M) forms of the EGFR. PKAc phosphorylated both WT-EGFR as well as a mutant truncated form of EGFR (Δ1022–1186) exclusively on serine residues. Moreover, PKAc also phosphorylated the cytosolic domain of the EGFR (EGFRKD). Phosphorylation of the purified WT as well as EGFRΔ1022–1186 and EGFRKD was accompanied by decreased autophosphorylation and diminished tyrosine kinase activity. Pretreatment of REF-52 cells with the nonhydrolyzable cAMP analog, 8-(4-chlorophenylthio)-cAMP, decreased EGF-induced tyrosine phosphorylation of cellular proteins as well as activation of the WT-EGFR. Similar effects were also observed in B82L cells transfected to express the Δ1022–1186 form of EGFR. Furthermore, activation of PKAc in intact cells resulted in serine phosphorylation of the EGFR. The decreased phosphorylation of cellular proteins and diminished activation of the EGFR in cells treated with the cAMP analog was not the result of altered binding of EGF to its receptors or changes in receptor internalization. Therefore, we conclude that PKA phosphorylates the EGFR on Ser residues and decreases its tyrosine kinase activity and signal transduction both in vitro andin vivo.

Binding of epidermal growth factor (EGF) to its receptor (EGFR) augments the tyrosine kinase activity of the receptor and autophosphorylation. Exposure of some tissues and cells to EGF also stimulates adenylyl cyclase activity and results in an increase in cyclic AMP (cAMP) levels. Because cAMP activates the cAMP-dependent protein kinase A (PKA), we investigated the effect of PKA on the EGFR. The purified catalytic subunit of PKA (PKAc) stoichiometrically phosphorylated the purified full-length wild type (WT) and kinase negative (K721M) forms of the EGFR. PKAc phosphorylated both WT-EGFR as well as a mutant truncated form of EGFR (⌬1022-1186) exclusively on serine residues. Moreover, PKAc also phosphorylated the cytosolic domain of the EGFR (EGFRKD). Phosphorylation of the purified WT as well as EGFR⌬1022-1186 and EGFRKD was accompanied by decreased autophosphorylation and diminished tyrosine kinase activity. Pretreatment of REF-52 cells with the nonhydrolyzable cAMP analog, 8-(4-chlorophenylthio)-cAMP, decreased EGF-induced tyrosine phosphorylation of cellular proteins as well as activation of the WT-EGFR. Similar effects were also observed in B82L cells transfected to express the ⌬1022-1186 form of EGFR. Furthermore, activation of PKAc in intact cells resulted in serine phosphorylation of the EGFR. The decreased phosphorylation of cellular proteins and diminished activation of the EGFR in cells treated with the cAMP analog was not the result of altered binding of EGF to its receptors or changes in receptor internalization. Therefore, we conclude that PKA phosphorylates the EGFR on Ser residues and decreases its tyrosine kinase activity and signal transduction both in vitro and in vivo.
Epidermal Growth Factor (EGF) 1 is responsible for a variety of biological effects ranging from mitogenesis (1) to influences on glucose metabolism (2). Upon binding of EGF to its receptor, the epidermal growth factor receptor (EGFR) undergoes dimerization and displays tyrosine kinase activity. This leads to autophosphorylation of the EGFR as well as phosphorylation of intracellular substrates (3). In addition, we have shown that, in the heart, EGF increases contractility and heart rate by augmenting cAMP accumulation (4), a result of stimulation of adenylyl cyclase activity via activation of G s␣ (4 -6). Additionally, we have also demonstrated that EGF-elicited stimulation of adenylyl cyclase activity requires the tyrosine kinase activity of the EGFR and may involve phosphorylation of G s␣ (7,8). The cyclic AMP formed in response to EGF can activate the heterotetrameric cAMP-dependent protein kinase A (PKA) by binding to its two regulatory subunits, thereby dissociating them from the catalytic subunits (9). The catalytic subunit of PKA (PKAc) can then phosphorylate a variety of intracellular proteins (9).
The EGFR is a 170-kDa glycoprotein with a single transmembrane domain. Its intracellular domain is susceptible to phosphorylation on various residues. Thus, the five autophosphorylation sites in the cytosolic domain of EGFR are located at tyrosine residues 992, 1068, 1086, 1148, and 1173 (10 -12). Furthermore, the EGFR is subject to regulation by other kinases. Hence, protein kinase C (PKC) phosphorylates the EGFR at Thr-654, thereby decreasing the intrinsic tyrosine kinase activity of the EGFR (13,14). Phosphorylation of Ser-1002 by p34 cdc2 (15) and of Ser-1046/1047 on the EGFR by calmodulin-dependent kinase II (16) is likewise associated with inhibition of tyrosine kinase activity. The EGFR can also be phosphorylated on Thr-669 by mitogen-associated protein kinase (MAPK; Refs. 17 and 18). In addition, in vivo pp60 c-src phosphorylates the EGFR on three tyrosines (Tyr-845, -891, and -820) that are not the autophosphorylation sites (19,20). These novel phosphotyrosines may provide docking sites for SH2 domain-containing proteins, which would explain the enhancement of the mitogenic response to EGF observed in pp60 c-src -overexpressing cells (20).
Previous in vitro studies have shown that the EGFR is a substrate for phosphorylation by PKA (21,22). However, to date, the functional significance of the phosphorylation has not been described. Since we and others have previously shown that EGF can activate adenylyl cyclase and increase cAMP accumulation in several tissues (4,(23)(24)(25), in the present study we have investigated the possibility that PKA may modulate EGFR tyrosine kinase activity. Our data show that phosphorylation of the EGFR by PKA on serine residues leads to decreased tyrosine kinase activity and diminished autophosphorylation of the EGFR both in vitro and in vivo.

MATERIALS AND METHODS
Purification of Wild Type (WT), K721M, and ⌬1022-1186 Forms of EGFR-Using the procedures described previously (26) wild type EGFR was purified from A431 cells. Likewise, using the same method (26), the EGFR K721M and EGFR⌬1022-1186 were purified from B82L cells transfected to express these forms of the receptor. The final step in this method involved elution of the EGFR with EGF from an EGFR antibody affinity column.
Phosphoamino Acid Analysis of EGFR-Bands corresponding to the 32 P-labeled proteins were excised from polyvinylidene difluoride membranes and subjected to phosphoamino acid analyses as described previously (8).
Stoichiometry of EGFR Phosphorylation by PKAc-This was achieved by pursuing two approaches. Both approaches involved decreasing the autophosphorylation of the EGFR. First, we employed the purified, kinase negative form of the EGFR (EGFR K721M; Ref. 27). This receptor (10 ng) was phosphorylated as described above. In the second approach, the tyrosine kinase activity of the wild type EGFR (100 ng) was inactivated by incubation with 1 mM N-ethylmaleimide (28) for 15 min in the phosphorylation buffer described above except that vanadate and dithiothreitol were absent. After addition of dithiothreitol (2 mM) to block further modifications of sulfhydryl groups, PKAc (480 units/ml) and [␥-32 P]ATP (10 M) were added to phosphorylate the EGFR as described above. Following SDS-PAGE (7.5% acrylamide), the bands corresponding to the EGFR were excised, and radioactivity associated with the receptor was determined. From the known specific radioactivity of ATP in the reaction mixture, the moles of phosphate incorporated into known amount of receptor were determined.
Detection of Tyrosine Phosphorylated Cellular Proteins and Active EGFR in Cell Lysates-B82L cells expressing EGFR⌬1022-1186 (12) and REF-52 cells were plated at a density of 2 ϫ 10 5 cells per 35-mm dish and allowed to grow for 24 h. The cells were then serum starved overnight and treated with 100 M 8-(4-chlorophenylthio)-cAMP (8-CPT-cAMP; Sigma) for the indicated times before addition of 100 nM EGF. Cells were harvested in 2ϫ Laemmli sample medium, and the samples were boiled at 100°C for 5 min. An aliquot (20 l) of the proteins in the Laemmli sample medium was diluted with water (80 l) and mixed with 400 l of 0.1 M sodium phosphate, pH 7.2, to precipitate the SDS. Standards of bovine serum albumin were similarly treated. The supernatant from this mixture was utilized to determine protein concentrations using the Bio-Rad Protein Assay reagent. Equal amounts of protein (15-45 g) were then subjected to SDS-PAGE. Western blots were performed as described previously (8). Either a polyclonal (Zymed Laboratories Inc.) or a monoclonal (PY-20; ICN Biomedicals Inc.) anti-phosphotyrosine antibody (both at 1:1000 dilution) was used. To detect activated EGFR, immunoblots of cellular proteins were performed with anti-active EGFR antibody (Transduction Laboratories). To ensure that equal amounts of proteins from REF-52 cells were loaded on gels, the blots were reprobed with anti-PKAc-␣ antibody (Santa Cruz).
Serine Phosphorylation of the EGFR by PKA in Intact Cells-B82L cells (4 ϫ 10 5 cells per 35-mm dish) expressing the K721M form of the EGFR were treated with and without 8-CPT-cAMP for 30 min. The EGFR was then immunoprecipitated from equal amounts of cell lysates (400 g of protein) with anti-EGFR antibody (Transduction Laborato-ries) as described before (7). Following separation of proteins in the immunoprecipitate by SDS-PAGE, Western analysis was performed with anti-phosphoserine antibody (clone A49, Alexis Corp., 1:250 dilution) according to the manufacturer instructions. The blot was reprobed with anti-EGFR antibody (Santa Cruz) to ensure that equal amounts of EGFR were immunoprecipitated.
Tyrosine Kinase Assays-Purified WT, EGFR⌬1022-1186, or EGFRKD was preincubated with the indicated amounts of PKAc in phosphorylation buffer described above for 30 min at 30°C. Thereafter, 1 mM Val-5-angiotensin II (Novabiochem) and 1.5 Ci of [␥-32 P]ATP were added and phosphorylation was allowed to proceed for 10 min at room temperature in a total volume of 25 l. The reactions were stopped with 25 l of 20% trichloroacetic acid. After centrifugation, the supernatants were spotted onto P81 Whatman filter disks, rinsed five times for 2 min each in 75 mM phosphoric acid, dried, and counted for 32 P incorporation into Val-5-angiotensin II.
Ligand Binding and Internalization of EGFR-B82L cells expressing either the WT or ⌬1022-1186 forms of EGFR were plated in 24-well plates (30,000 cells/well) and serum starved overnight. The cells were washed with 0.5 ml of ice-cold Krebs-Henseleit buffer modified to contain 20 mM Hepes, pH 7.4, and incubated on ice for 25 min with the same buffer containing 0.5 mg/ml bovine serum albumin. Thereafter, 125 I-EGF (50 pM) was added, and incubation was continued on ice for a further 2 h. Cells were then washed three times with 0.5 ml of ice-cold Krebs-Henseleit buffer and solubilized in 0.5 ml of 1 N NaOH. Nonspecific binding was determined in the presence of excess (1 M), unlabeled EGF. To monitor internalization of the EGFR, the method of Honnegar et al. (29) was pursued. Essentially, following binding of 125 I-EGF to cells as described above, the cells were incubated at 37°C for 30 min and then washed with Krebs-Henseleit buffer as described above. Cell surface-associated 125 I-EGF was removed and counted with 0.5 ml ice-cold 0.5 M acetic acid and 150 mM NaCl. Thereafter, cells were washed once with Krebs Henseleit buffer and lysed at 37°C in 0.5 ml of 1N NaOH to determine the amount of internalized 125 I-EGF.

RESULTS AND DISCUSSION
Because EGF can increase cAMP accumulation in the heart and other tissues (4 -6, 23-25), we postulated that akin to the negative regulation of the EGFR by protein kinase C (14), calmodulin-dependent protein kinase II (16) and p34 cdc2 (15), PKA may alter the functional activity of EGFR. To address this hypothesis, initially, we determined the effect of pure PKAc on autophosphorylation of the purified WT and ⌬1022-1186 forms of EGFR; in the ⌬1022-1186 form of EGFR, four of the five autophosphorylation sites are deleted. Fig. 1A shows that PKAc decreases the incorporation of 32 P label from [␥-32 P]ATP into the WT-EGFR, whereas 32 P incorporation into EGFR⌬ 1022-1186 remained unchanged in the presence of PKAc. Similar results, i.e. decreases in 32 P incorporation, were obtained when the WT-EGFR was incubated with the PKA holoenzyme activated by 8-CPT-cAMP (not shown). This is in contrast with the results of Rackoff et al. (21) and Ghosh-Dastidar et al. (22) who showed an increase in 32 P incorporation from [␥-32 P]ATP into the purified EGFR after addition of PKA. Possible explanations for this difference are as follows. First, Rackoff et al. (21) incubated the EGFR with PKA for just 30 -60 s, which may be too short to reach steady state phosphorylation of the EGFR on tyrosine and serine/threonine residues. Second, in the experiments of Rackoff et al. (21) and Ghosh-Dastidar and Fox (22), vanadate, an inhibitor of phosphotyrosine phosphatases was absent. Under these conditions, tyrosine phosphatases, which may co-purify with the EGFR, would decrease tyrosine phosphorylation of the EGFR, thereby allowing the net increase in PKA-mediated phosphorylation to be monitored. In contrast, in our experiments that employed a longer incubation time and vanadate, the phosphorylation observed with 32 P incorporation would represent the sum of tyrosine and serine/ threonine phosphorylation of the EGFR. Therefore, to determine whether PKAc altered tyrosine phosphorylation of the EGFR, experiments similar to those in Fig. 1A were performed with unlabeled ATP. The samples were then subjected to SDS-PAGE and Western analyses with anti-phosphotyrosine anti-body. As demonstrated by data in Fig. 1B, incubation of the WT-EGFR with PKAc decreased tyrosine phosphorylation of the receptor. Similarly, autophosphorylation of the ⌬1022-1186 form of the EGFR was also decreased in the presence of PKAc. Thus, the decrease in 32 P incorporation in WT-EGFR observed in the presence of PKAc (Fig. 1A) represents a decrease in autophosphorylation of the receptor. Likewise, because tyrosine phosphorylation of the one site (Tyr-992) on the EGFR⌬1022-1186 is decreased (Fig. 1B), in the presence of a serine or threonine phosphorylation of this receptor by PKAc, no net change in 32 P incorporation would be evident (Fig. 1A), i.e. phosphorylation of one serine or threonine residue by PKAc on EGFR⌬1022-1186 would compensate for the loss of phosphorylation at Tyr-992.
To determine whether PKAc phosphorylates serine or threonine residues on the EGFR, phosphoamino acid analyses were performed. Fig. 2A shows that phosphorylation of the WT-EGFR occurs predominantly on serine residues, as also found by Rackoff et al. (21) and Ghosh-Dastidar and Fox (22). Furthermore, Fig. 2A shows that the appearance of phosphoserine in the presence of PKAc is accompanied by a decrease in phosphotyrosine. PKAc also phosphorylates EGFR⌬1022-1186 on serine residues (Fig. 2B), and this increase in serine phosphorylation is accompanied by the loss of tyrosine phosphorylation. These data support the contention that, in experiments performed with EGFR⌬1022-1186 and [␥-32 P]ATP, no net difference in 32 P incorporation into the receptor is observed in the presence of PKAc (Fig. 1A, right panel) because of a gain of serine phosphorylation and concomitant loss of tyrosine phosphorylation. Because both the WT and EGFR⌬1022-1186 are phosphorylated on serine residues by PKAc, the data in Fig. 2 demonstrate that the PKAc phosphorylation site(s) is(are) not located in the region between amino acids 1022 and 1186. In additional experiments, we determined whether the purified, cytosolic region of the EGFR (amino acids 645-1186; EGFRKD) is also phosphorylated by PKAc. As demonstrated by data in Fig. 3A, phosphorylation of the purified EGFRKD by PKAc resulted in a mobility shift of the EGFRKD on polyacrylamide gels. These data coupled with the findings with the wild type and ⌬1022-1186 forms of the EGFR (Figs. 1 and 2) demonstrate that the serine residue(s) on the EGFR, which is(are) Phosphoproteins were separated on a 7.5% SDS-PAGE gel and visualized by autoradiography. Note that the autoradiograph with WT-EGFR was exposed for 2 h only, whereas the autoradiograph with EGFR⌬1022-1186 was exposed overnight. PKAc autophosphorylation is evident only after long exposures. Panel B, phosphotyrosine content of WT and ⌬1022-1186 forms of EGFR in the presence and absence of PKAc. The WT (left) and ⌬1022-1186 (right) EGFR were phosphorylated by PKAc (480 units/ml) in the presence of unlabeled ATP (10 M) as described under "Materials and Methods." Phosphoproteins were separated by SDS-PAGE (7.5% acrylamide) and transferred to nitrocellulose. Western blotting was performed with anti-phosphotyrosine antibody as described under "Materials and Methods." Representatives of at least three similar experiments are shown. phosphorylated by PKAc, resides between amino acids 644 and 1022.
Because the wild type EGFR is phosphorylated both on Tyr as well as Ser residues in the presence of PKAc, and because tyrosine phosphorylation of the receptor decreases in the presence of PKAc (Fig. 2, A and B), the stoichiometry of phosphorylation was difficult to monitor. To circumvent this problem, we employed two approaches. First, phosphorylation of the purified, kinase negative form of the EGFR (EGFR K721M) was monitored. Second, the WT-EGFR was incubated with N-ethylmaleimide to inhibit its tyrosine kinase activity (28). Thereafter, the PKAc-mediated phosphorylation was monitored. As demonstrated in Figs. 3, B and C, autophosphoryla-tion of either the K721M form of the EGFR or the wild type receptor in the presence of N-ethylmaleimide was minimal. Under these conditions, PKAc phosphorylated both the kinase negative EGFR (EGFR K721M, Fig. 3B) and WT-EGFR with a stoichiometry of 0.74 and 0.87 mol of P i per mol of EGFR, respectively.
Because incubation of the WT and ⌬1022-1186 forms of EGFR in the presence of PKAc decreased tyrosine phosphorylation of both receptors (Figs. 1B and 2), it would appear that PKAc-elicited phosphorylation of EGFR on serine residues modulates the tyrosine kinase activity of the EGFR. Therefore, to directly evaluate the functional significance of phosphorylation of EGFR by PKAc, we studied the influence of PKAc on the tyrosine kinase activity of the WT and ⌬1022-1186 forms of EGFR. Fig. 4 demonstrates that PKAc, in a concentration-dependent manner, inhibited tyrosine kinase activity of both types of EGFR. Moreover, as demonstrated in Fig. 4C, PKAc also inhibited the tyrosine kinase activity of the EGFRKD. Together, the data in Figs. 1-4 demonstrate that the stoichiometric serine phosphorylation of the EGFR by PKAc is accompanied by a decrease in tyrosine kinase activity of the receptor. Moreover, the site of serine phosphorylation on the EGFR is located between residues 644 -1022.
Next we investigated whether or not activation of PKA in vivo alters EGF-induced tyrosine phosphorylation of cellular proteins and activation of the EGFR. For this purpose, cells were incubated with 8-CPT-cAMP, a nonhydrolyzable cAMP analog, that activates PKA (30). Fig. 5A(i) illustrates that in REF-52 cells, in the absence of 8-CPT-cAMP, the addition of EGF resulted in a marked increase in tyrosine phosphorylation of cellular proteins. The most prominent tyrosine phosphorylation in response to EGF was observed in proteins of molecular masses of ϳ180 and ϳ70 kDa (Fig. 5A(i)). Treatment of REF-52 cells with 8-CPT-cAMP resulted in a marked decrease in EGFelicited tyrosine phosphorylation of these proteins (Fig. 5A(i)). The decrease in tyrosine phosphorylation of cellular proteins in the presence of 8-CPT-cAMP cannot be attributed to differences in protein loading because reprobing the same blot with anti-PKAc antibody showed that the amount of PKAc in each lane was the same (lower panel of Fig. 5A(i)). Moreover, as assessed with the anti-active EGFR antibody, EGF-elicited activation of the EGFR in REF-52 cells was also markedly attenuated in the presence of 8-CPT-cAMP ( Fig. 5A(ii)); immunoprecipitation of the EGFR from REF-52 cells treated with 8-CPT-cAMP also showed a decrease in EGF-elicited autophosphorylation of the immunoprecipitated EGFR (not shown). Similarly, in B82L cells transfected to express EGFR⌬1022-1186, activation of PKA by 8-CPT-cAMP for different times also led to a decrease in tyrosine phosphorylation of proteins (Fig.  5B). Thus, 5 min after treatment of cells with 8-CPT-cAMP, a decrease in EGF-mediated tyrosine phosphorylation of cellular proteins was observed (Fig. 5B). Treatment of these cells with forskolin, which directly activates adenylyl cyclase and increases cAMP levels (31), yielded similar results (not shown). To determine whether or not activation of PKA in intact cells phosphorylates the EGFR on serine residues, the experiment depicted in Fig. 5C was performed. Essentially, B82L cells transfected to express the kinase-negative form of the EGFR (EGFR K721M) were treated with 8-CPT-cAMP for 30 min in the absence of EGF. Thereafter, the cells were lysed, and EGFR was immunoprecipitated. Following separation of proteins in the immunoprecipitate, Western analysis with anti-phosphoserine antibody was performed. As shown in Fig. 5C, treatment of cells with 8-CPT-cAMP resulted in serine phosphorylation of the immunoprecipitated EGFR; the amount of receptor immunoprecipitated from cells treated with and without 8-CPT-

FIG. 3. PKAc phosphorylates the purified cytosolic domain of the EGFR (EGFRKD) and stoichiometrically phosphorylates the kinase negative (K721M) form of the EGFR and kinase inactivated WT-EGFR.
Panel A, purified, cytosolic domain of the EGFR (10 ng) was incubated in the presence of PKAc (480/ml) as described under "Materials and Methods." Proteins were subjected to SDS-PAGE and detected by autoradiography. The phosphorylation of EGFRKD by PKAc alters its migration (indicated by **) on the gel as seen for many phosphoproteins. A representative of three experiments is shown. Panel B, purified EGFR K721M (10 ng) was incubated in the presence of PKAc (480 units/ml, Promega) and [␥-32 P]ATP (10 M) for 10 min at room temperature in the reaction mixture described under "Materials and Methods." The reactions were terminated by the addition of Laemmli sample medium. Proteins were separated by SDS-PAGE (7.5% gels) and subjected to autoradiography. The bands corresponding to EGFR K721M in the presence and absence of PKAc were excised and counted for 32 P content. A representative of three similar experiments is shown. Stoichiometry of phosphorylation was 0.74 mol of P i per mol of EGFR. Panel C, purified EGFR (100 ng) was incubated with N-ethylmaleimide (1 mM) for 15 min as described under "Materials and Methods." PKAc (480 units/ml) was then added to the incubation containing all ingredients of the phosphorylation mixture. Reactions were terminated after 30 min and proteins separated by SDS-PAGE. Autoradiography was performed to detect the phospho-proteins. The EGFR band was excised from the gel, and the 32 P label incorporated was determined by counting. Stoichiometry of phosphorylation was 0.87 mol of P i per mol of EGFR. Note that autophosphorylated PKAc is visible on the autoradiographs. cAMP was the same (Fig. 5C). The data in Fig. 5 demonstrate that activation of PKA in intact cells results in serine phosphorylation of the EGFR, decreased activation of the receptor (Fig.  5A(ii)) and decreased phosphorylation of cellular proteins in response to EGF. Moreover, the data in Fig. 5C demonstrate that activation of PKA in intact cells can phosphorylate the EGFR on serine residues in the absence of EGF. Thus, the EGFR does not have to be activated by its ligand for phosphorylation by PKA.
One possible reason for a decrease in autophosphorylation of EGFR in intact cells in the presence of the PKA activator 8-CPT-cAMP is that PKA-elicited phosphorylation of the EGFR may alter binding of EGF to its receptor or enhance receptor internalization, thereby decreasing the number of cell surface EGF receptors. To address this possibility, we monitored the binding of 125 I-EGF and internalization of the ligand in B82L cells expressing EGFR⌬1022-1186 that had been pretreated with or without 8-CPT-cAMP. These experiments were performed with 125 I-EGF concentrations of 50 pM, the K D of EGFR for EGF, because any change in binding affinity would be most pronounced at this concentration of the ligand. Essentially, our data demonstrated that, in B82L cells expressing EGFR⌬1022-1186, the binding of EGF was not altered (Table I). Likewise, pretreatment of B82L cells expressing EGFR⌬1022-1186 with 8-CPT-cAMP did not alter receptor internalization (Table I). These data indicate that the decrease in EGFelicited receptor autophosphorylation observed in cells preincubated with 8-CPT-cAMP is not the result of a modification in either EGF binding to its receptors or a change in cell surface EGFR numbers because of receptor internalization. Thus, the effects of PKA activation in intact cells reflect the in vitro findings which demonstrate that PKA, by phosphorylating EGFR on serine residues, decreases its protein-tyrosine kinase activity.
Regulation of EGFR tyrosine kinase activity by phosphorylation has been described for PKC (13,14), calmodulin-dependent protein kinase II (16), and p34 cdc2 (15). In all of these cases, phosphorylation decreases tyrosine kinase activity. Our data presented here add to the list of protein kinases that regulate EGFR and, for the first time, demonstrate an inhibitory effect of PKA on the early steps of the EGF signaling cascade both in vivo and in vitro. To date, studies of the inhibitory influence of the cAMP-PKA pathway on EGFR signaling have focused mainly on events downstream of Ras, more specifically on MAPK. Thus, it is known that activation of the cAMP-PKA pathway can have inhibitory (32-34), stimulatory (35), or no (36) effects on the activation of MAPK by EGF, depending on the cell type used. This variability is probably because of the various Raf isoforms expressed in the different cell types (37). However, these studies in which intracellular cAMP levels were raised to modulate MAPK activation in response to EGF failed to show an inhibitory effect on WT-EGFR autophosphorylation (32,33,38,39). This is in contrast with our data (Fig.  5) which demonstrate that PKA can interfere with the EGFR signaling cascade at the level of the EGFR itself. One explanation for this discordance may be that the cell lines overexpressing the WT-EGFR, which were used in a number of the studies concerning MAPK activation, may not be suitable for detecting this inhibitory effect of PKA because the amount of EGFR far exceeds the amount of PKA. Indeed, in B82L cells that overexpress the WT-EGFR, we have also not observed any significant decrease in EGF-elicited tyrosine phosphorylation of cellular proteins in the presence of 8-CPT-cAMP (not shown). In any event, our findings, together with those of others (32)(33)(34) suggest that, in some cells, PKA can inhibit EGF-mediated activation of MAPK by attenuating both the EGFR kinase as well as interactions of Raf and Ras, and these two mechanisms may act in a mutually reinforcing manner.
Interestingly, the EGFR is not the only receptor tyrosine kinase target for regulation by PKA. Hence, pretreatment of cells with cAMP-elevating agents increases serine/threonine phosphorylation of the insulin receptor and decreases its insulindependent tyrosine kinase activity by 50% (40). This decrease in insulin receptor tyrosine kinase activity by cAMP-elevating agents has been confirmed in in vitro experiments which demonstrated that PKA phosphorylates the insulin receptor and decreases its tyrosine kinase activity (41). These findings coupled with our observations would suggest that PKA may play a more generalized role in regulation of receptor protein-tyrosine kinase signaling.
Although, at present, the identity of the serine residue(s) on EGFR which is(are) phosphorylated remains unknown, there are some sites that can be discarded. Thus, because PKA phosphorylates and modulates the activity of the WT-EGFR, EGFR ⌬1022-1186, and cytosolic domain (EGFRKD) of the receptor similarly, it would appear that the functionally important serine residue(s) which is(are) phosphorylated and modulates EGFR kinase activity must reside in the cytosolic region, i.e. between amino acids 644 and 1022. Within this region of the EGFR, serine residues 671, 967, 971, and 1002 have been shown to be phosphorylated (15,42). Among these, however, only the phosphorylation of Ser-1002 by p34 cdc2 has been shown to decrease EGFR tyrosine kinase activity (15). The functional significance of phosphorylation of Ser-671, -967, and -971 remains to be determined (42). Notably, however, none of these sites including Ser-1002 conforms to the PKA consensus sequence. Nevertheless, the identity of the precise serine residues on EGFR that are phosphorylated by PKA and that alter EGFR kinase activity forms the subject of future investigations.
In conclusion, we have demonstrated that PKA can phosphorylate EGFR on serine residue(s) and that this phosphorylation is accompanied by a decrease in EGFR protein-tyrosine kinase activity both in vitro and in vivo. In view of our previous findings that EGFR kinase activity is required to stimulate adenylyl cyclase activity (7) and increase intracellular cAMP levels (5), the findings presented here would suggest that activation of PKA can, in a feedback regulatory manner, attenuate EGFR tyrosine kinase activity. The cells were then treated with EGF (100 nM) for the indicated time period. Cells were then lysed in Laemmli sample medium. After determination of proteins in the Laemmli sample medium as described under "Materials and Methods," equal amounts of cellular proteins (45 g) were separated by SDS-PAGE (7.5% acrylamide) and transferred to nitrocellulose. Western blots were developed with antiphosphotyrosine antibody PY-20 (top panel). The proteins whose tyrosine phosphorylation was most markedly altered by EGF are indicated by arrows. To ensure that the amount of protein loaded was the same, the same blot was reprobed with anti-PKAc-␣ antibody (Santa Cruz; bottom panel). Panel A(ii), same as panel A(i) except that anti-active EGFR antibody (Transduction laboratories; was employed. Representatives of three similar experiments are shown. Panel B, time course of 8-CPT-cAMP-mediated attenuation of EGF-elicited cellular tyrosine phosphorylation. B82L cells expressing EGFR⌬1022-1186 were incubated with 8-CPT-cAMP (100 M) for the indicated times and then exposed to EGF (100 nM) for 10 min. Cell lysates (40 g of protein) were subjected to Western analyses with anti-phosphotyrosine antibody. The migration of the major phosphoprotein, pp170, is shown. Panel C, B82L cells expressing the kinase negative mutant (K721M) of the EGFR were treated with and without 8-CPT-cAMP (100 M) for 30 min. Cells were lysed, and EGFR was immunoprecipitated as described under "Materials and Methods." Proteins in the immunoprecipitates were subjected to SDS-PAGE and immunoblotted (IB) with anti-phosphoserine antibody (clone A49) as described under "Materials and Methods." The same blot was reprobed with anti-EGFR antibody to ensure that equal amounts of EGFR were immunoprecipitated.