The epidermal growth factor receptor (EGFR) tyrosine kinase recently was identified as providing a link to mitogen‐activated protein kinase (MAPK) in response to G protein‐coupled receptor (GPCR) agonists in Rat‐1 fibroblasts. This cross‐talk pathway is also established in other cell types such as HaCaT keratinocytes, primary mouse astrocytes and COS‐7 cells. Transient expression of either Gq‐ or Gi‐coupled receptors in COS‐7 cells allowed GPCR agonist‐induced EGFR transactivation, and lysophosphatidic acid (LPA)‐generated signals involved the docking protein Gab1. The increase in SHC tyrosine phosphorylation and MAPK stimulation through both Gq‐ and Gi‐coupled receptors was reduced strongly upon selective inhibition of EGFR function. Inhibition of phosphoinositide 3‐kinase did not affect GPCR‐induced stimulation of EGFR tyrosine phosphorylation, but inhibited MAPK stimulation, upon treatment with both GPCR agonists and low doses of EGF. Furthermore, the Src tyrosine kinase inhibitor PP1 strongly interfered with LPA‐ and EGF‐induced tyrosine phosphorylation and MAPK activation downstream of EGFR. Our results demonstrate an essential role for EGFR function in signaling through both Gq‐ and Gi‐coupled receptors and provide novel insights into signal transmission downstream of EGFR for efficient activation of the Ras/MAPK pathway.
Definition of cellular responses such as proliferation or differentiation by a variety of external stimuli involves the regulation of transcriptional events in eukaryotic cells through intracellular signaling cascades, including signaling pathways that activate kinases of the mitogen‐activated protein kinase (MAPK) family (Treisman, 1996). The extensively studied MAPKs of the ERK subfamily become activated in response to growth factors either through receptor tyrosine kinase (RTK)‐ or through G protein‐coupled receptor (GPCR)‐triggered signals (Pagès et al., 1993; Faure et al., 1994; Malarkey et al., 1995; van Biesen et al., 1996b). Transmission of these signals requires the formation of a complex between the Grb2 adaptor protein with the guanine nucleotide exchange factor Sos, which upon recruitment to the plasma membrane allows activation of the small G protein Ras, resulting in subsequent activation of the MAPK pathway (Egan et al., 1993; van Biesen et al., 1995; Dikic et al., 1996). The critical function of Ras in MAPK signal transduction triggered by GPCRs which are coupled either to heterotrimeric Gi or Gq proteins (Crespo et al., 1994) was shown to be sensitive to genistein, suggesting that a tyrosine kinase was involved in this process (van Corven et al., 1993; Sadoshima and Izumo, 1996). Moreover, stimulation of various GPCRs, such as those for lysophosphatidic acid (LPA) (van Biesen et al., 1995), thyrotropin‐releasing hormone (Ohmichi et al., 1994), endothelin‐1 (ET‐1) (Cazaubon et al., 1994), bradykinin (Lev et al., 1995) or thrombin (Chen et al., 1996), rapidly induces tyrosine phosphorylation of the adaptor protein SHC and SHC–Grb2 complex formation, steps that couple both GPCRs and RTKs to Ras (Bonfini et al., 1996; Chen et al., 1996; van der Geer et al., 1996). In addition, a recent report implicated an unidentified 100 kDa tyrosine‐phosphorylated, Grb2‐binding protein in LPA‐induced MAPK activation (Kranenburg et al., 1997).
Recently, we identified the epidermal growth factor receptor (EGFR) as an essential link in the GPCR‐mediated MAPK activation pathway in Rat‐1 fibroblasts treated with the GPCR agonists ET‐1, LPA or thrombin (Daub et al., 1996). Moreover, angiotensin II‐induced platelet‐derived growth factor receptor β (βPDGFR) and thrombin‐stimulated insulin‐like growth factor‐1 receptor (IGF‐1R) tyrosine phosphorylation were reported in primary rat smooth muscle cells (Linseman et al., 1995; Rao et al., 1995), suggesting that transactivation of distinct RTKs might contribute in a cell‐type specific manner to GPCR‐mediated mitogenic signaling. In addition to GPCR activation, a variety of stimuli including cellular stresses such as UV irradiation or calcium‐dependent responses in PC12 cells result in ligand‐independent RTK transactivation (Knebel et al., 1996; Rosen and Greenberg, 1996; Weiss et al., 1997). Interestingly, cytoplasmic kinases of the Src family were also implicated in signal generation by similar stimuli (Devary et al., 1992; Rusanescu et al., 1995). In particular, GPCR‐mediated activation of the Ras/MAPK pathway was reported in some recent studies to involve Src function (Dikic et al., 1996; Luttrell et al., 1996; Sadoshima et al., 1996; Schieffer et al., 1996), which raised the possibility that concerted action of RTKs and Src family kinases might be required to initiate intracellular signaling cascades.
In the present study, we show that ligand‐independent EGFR transactivation occurs in diverse cell types. Employing COS‐7 cells, we demonstrate that Gq‐ and Gi‐coupled receptors can mediate stimulation of EGFR tyrosine phosphorylation and, moreover, that for both GPCR subfamilies this event is essential for the stimulation of SHC tyrosine phosphorylation and MAPK activation. In addition, we present evidence for phosphoinositide 3‐kinase (PI3‐K) function downstream of the EGFR and address the question of whether other tyrosine kinases might be involved in EGFR transactivation and signal transduction.
EGFR transactivation in diverse cell types
We have shown previously that treatment of Rat‐1 fibroblasts with the GPCR ligands ET‐1, LPA and thrombin results in rapid induction of EGFR tyrosine phosphorylation. To substantiate these findings further, we investigated this mechanism in different cell lines. As shown in Figure 1A, stimulation of human HaCaT keratinocytes with thrombin or LPA resulted in enhanced tyrosine phosphorylation of endogenous EGFR. In addition, thrombin or extracellularly applied ATP triggered a comparable response in primary mouse astrocytes (Figure 1B). Although it is presently unclear whether ATP acts through GPCRs or ATP‐gated ion channels (Barnard et al., 1994; Salter and Hicks, 1995), EGFR transactivation might represent an important signaling element in ATP‐evoked responses in astrocytes. Thrombin and LPA were also effective in COS‐7 cells, where these GPCR ligands stimulated tyrosine phosphorylation of endogenous EGFR to a similar extent as 1–3 ng/ml EGF (Figure 1C). Thrombin receptor peptide (TRP) was able to mimic the thrombin response, confirming specific activation of the thrombin GPCR in COS‐7 cells. The LPA vehicle bovine serum albumin (BSA), included as a negative control, showed no effect. For further characterization of the tyrosine‐phosphorylated 170 kDa protein co‐migrating with the EGFR, immunoprecipitations with different anti‐EGFR antibodies were performed and the precipitates were treated with glycosidase. In all these control experiments, the 170 kDa band behaved exactly as the EGFR (data not shown). Moreover, time course experiments revealed that LPA rapidly induced EGFR tyrosine phosphorylation, only slightly lagging behind that induced by 3 ng/ml EGF. The EGF‐induced EGFR tyrosine phosphorylation was more prolonged than that induced by LPA, possibly due to accelerated receptor down‐regulation mediated by LPA‐specific cellular responses (Figure 1D). Taken together, these experiments demonstrate that GPCR‐mediated EGFR transactivation is not restricted to Rat‐1 fibroblasts but occurs in various cell types from different species.
Stimulation of EGFR tyrosine phosphorylation by Gq‐ and Gi‐coupled receptors in COS‐7 cells
To examine the elements involved in EGFR transactivation further, we selected COS‐7 cells, since this cell line has been used extensively to study GPCR signaling and, moreover, serves as a model system to investigate GPCR‐mediated activation of the MAPK pathway (Crespo et al., 1994; Hawes et al., 1995; van Biesen et al., 1995). First, we analyzed the effect of pertussis toxin (PTX) on EGFR transactivation. PTX catalyzes ADP ribosylation of α subunits in the Gi and Go subfamilies of heterotrimeric G proteins, a covalent modification which leads to inactivation. Treatment of cells with PTX partially attenuated LPA‐induced EGFR tyrosine phosphorylation, while the thrombin response was not significantly affected (Figure 2). Notably, PTX reduces the LPA‐stimulated EGFR transactivation to a level similar to that observed after thrombin treatment. Since Go proteins are not expressed in COS‐7 cells (van Biesen et al., 1996a), this indicates that LPA signals through Gi and through PTX‐insensitive G proteins, while thrombin only utilizes PTX‐insensitive pathways. To assess whether PTX‐insensitive G proteins of the Gq subclass are able to mediate EGFR transactivation in COS‐7 cells, we transiently expressed GPCRs known to be coupled predominantly to Gq proteins (Crespo et al., 1994; Offermanns et al., 1994). As shown in Figure 3A, exposure of cells transfected with bombesin receptor (BombR) or M1 muscarinic acetylcholine receptor (M1R) to the respective GPCR ligands resulted in increased EGFR tyrosine phosphorylation, which was not observed upon stimulation of control‐transfected cells. In cells transfected with the Gi‐coupled M2 muscarinic acetylcholine receptor (M2R), GPCR agonist‐enhanced EGFR tyrosine phosphorylation was detectable, although this effect was weaker than that obtained upon Gq‐coupled receptor stimulation (Figure 3B). Moreover, PTX did not affect bombesin‐induced EGFR transactivation, while the carbachol‐induced response mediated by M2R was blocked upon PTX pre‐incubation (Figure 3C). These experiments show that EGFR transactivation may occur through the engagement of both PTX‐insensitive and ‐sensitive pathways. The above experiment, which showed that, in the absence of PTX, LPA induces stronger EGFR tyrosine phosphorylation than thrombin while their responses become similar upon PTX pre‐treatment, suggests that Gi‐mediated and PTX‐insensitive pathways might exert additive effects on EGFR transactivation in COS‐7 cells. To address this question, we co‐transfected Gi‐coupled M2R and Gq‐coupled BombR and stimulated either each receptor alone or both simultaneously. Co‐stimulation with carbachol and bombesin resulted in a more pronounced induction of EGFR tyrosine phosphorylation than observed with either ligand alone, providing evidence for the cooperativity of Gi‐ and Gq‐mediated pathways. Moreover, the combined effect on EGFR transactivation was similar to the response observed upon LPA treatment (Figure 3D).
Identification of Gab1 as Grb2‐interacting protein potentially involved in GPCR‐mediated signaling
We previously reported that several tyrosine‐phosphorylated proteins including the EGFR, SHC and an unidentified 110 kDa protein associate with glutathione S‐transferase (GST)–Grb2 fusion protein upon ET‐1, LPA and thrombin stimulation in Rat‐1 fibroblasts (Daub et al., 1996). Specific association of a similar set of tyrosine‐phosphorylated proteins was also detected when we stimulated COS‐7 cells with LPA or 3 ng/ml EGF, which induces EGFR tyrosine phosphorylation comparable with the LPA response, and subsequently incubated lysates with GST–Grb2 fusion protein (Figure 4A). Pre‐incubation with the EGFR‐specific inhibitor AG1478 prevented LPA‐induced association of tyrosine‐phosphorylated proteins (Levizki and Gazit, 1995), indicating that, in addition to SHC, the 110 kDa protein represents an additional substrate for transactivated EGFR in COS‐7 cells. Moreover, a Grb2‐associating protein of similar molecular weight recently was implicated in LPA‐induced signaling to the Ras/MAPK pathway, but was not identified (Kranenburg et al., 1997). Utilizing an immunodepletion strategy, we were able to identify the 110 kDa protein as Gab1 (Holgado‐Madruga et al., 1996), since the tyrosine‐phosphorylated 110 kDa protein upon LPA treatment was specifically and quantitatively removed when lysates were pre‐cleared with polyclonal anti‐Gab1 antiserum prior to in vitro association with GST–Grb2 (Figure 4B). When we performed immunoprecipitations with anti‐Gab1 antiserum, LPA‐induced stimulation of Gab1 tyrosine phosphorylation and Gab1–Grb2 complex formation were clearly detectable in the absence but not in the presence of the GST–Gab1‐N fusion protein used for antibody generation (Figure 4C). Moreover, both signaling events were abrogated when cells were pre‐incubated with AG1478 in order to block EGFR function (Figure 4D), indicating that Gab1 functions downstream of EGFR in LPA signaling.
Gq‐ and Gi‐coupled receptor‐mediated SHC tyrosine phosphorylation requires functional EGFR
In order to establish that EGFR function represents a general prerequisite for GPCR‐mediated tyrosine phosphorylation of proteins such as SHC, we transfected either the Gq‐coupled BombR or the Gi‐coupled M2R into COS‐7 cells and determined SHC tyrosine phosphorylation and association of immunoprecipitated SHC with the adaptor protein Grb2 upon stimulation with the respective GPCR ligands. Bombesin, acting through its ectopically expressed receptor, stimulated SHC tyrosine phosphorylation as well as SHC–Grb2 complex formation. Both signaling events were blocked when the dominant‐negative EGFR mutant HER‐CD533 was co‐transfected (Kashless et al., 1991; Redemann et al., 1992), while an analogous truncation mutant of the PDGFR (βPDGFR‐CD504) had no inhibitory effect (Figure 5A), although both RTK mutants were expressed at comparable levels (data not shown). Furthermore, the EGFR‐specific inhibitor AG1478 suppressed SHC tyrosine phosphorylation and complex formation with Grb2 upon stimulation of the Gi‐coupled M2R (Figure 5B). Both approaches to inhibit EGFR function appeared to be specific, since SHC tyrosine phosphorylation and association with Grb2 induced by c‐Src overexpression in COS‐7 cells or upon nerve growth factor treatment of PC12 cells were unaffected (data not shown). Thus Gi‐ and Gq‐mediated pathways depend on a functional EGFR to trigger SHC tyrosine phosphorylation, a signaling step considered to be critical for subsequent activation of Ras‐dependent pathways.
EGFR inhibition interferes with MAPK activation by Gq‐ and Gi‐coupled receptors
We previously demonstrated a functional role for EGFR in the GPCR‐mediated activation of the MAPK pathway of Rat‐1 cells upon stimulation with agonists such as LPA. To establish the COS‐7 cell system that allows the investigation of MAPK pathway control elements, we employed a transiently expressed hemagglutinin (HA) epitope‐tagged form of the MAPK ERK2 (HA‐ERK2) and measured phosphorylation of myelin basic protein (MBP). The system was calibrated with 2 ng/ml EGF, which induces EGFR tyrosine phosphorylation comparable with that obtained with LPA and resulted in MAPK activation which was, as expected, inhibited upon co‐transfection of HER‐CD533, while the dominant‐negative βPDGFR‐CD504 mutant had no effect (Figure 6A, left panel). Conversely, MAPK activation through βPDGFR was unaffected by HER‐CD533 but abrogated by βPDGFR‐CD504 co‐expression (Figure 6A, middle panel). These controls confirmed the selective inhibition of EGFR and βPDGFR signaling by HER‐CD533 and βPDGFR‐CD504, respectively. Under these conditions, LPA‐induced MAPK activation was reduced by nearly 80% upon HER‐CD533 co‐expression, while βPDGFR‐CD504 showed no significant effect (Figure 6A, right panel). Co‐transfection of either of the dominant‐negatives RTK mutants, which were expressed at comparable levels as shown by [35S]methionine labeling of cells transfected in parallel followed by nearly quantitative immunoprecipitation (Figure 6B), did not affect expression of HA‐ERK2 (Figure 6A, lower panel). These data clearly confirm our previous Rat‐1 cell experiments, which demonstrated that EGFR transactivation represents an essential step in the signaling cascade that mediates LPA‐induced activation of the MAPK pathway. Using this tightly controlled transient transfection system, we investigated the role of EGFR in Gq‐ and Gi‐mediated MAPK activation. As shown in Figure 6C, the dominant‐negative EGFR mutant nearly abolished MAPK activity stimulated by the Gq‐coupled BombR and inhibited MAPK activation through the Gi‐coupled M2R by 50% (Figure 6C). Comparable results were obtained upon pre‐treatment of cells with AG1478 to block specifically EGFR kinase function (Figure 6D), while pre‐incubation with this tyrphostin did not significantly affect MAPK activation mediated either by co‐transfected IGF‐1R or induced by expression of v‐Abl, an oncogenic variant of the tyrosine kinase c‐Abl. The selectivity of AG1478 was demonstrated using EGFR−/− primary fibroblasts, in which even higher concentrations of the inhibitor did not affect GPCR‐mediated MAPK activation (E.Zwick et al., in preparation). Thus EGFR function is important for Gq‐ and Gi‐mediated MAPK activation and, since in the case of the BombR (data not shown) and M2R (Crespo et al., 1994) this event is suppressed by dominant‐negative Ras, EGFR represents a pathway element upstream of Ras.
Interestingly, comparison of the time course of MAPK activation in COS‐7 cells upon treatment of COS‐7 cells with either LPA or 3 ng/ml EGF revealed that LPA induced MAPK activity more rapidly (Figure 6E). This suggests that during the initial phase of LPA‐mediated MAPK activation, additional LPA‐triggered signals synergize with EGFR‐mediated signaling, leading to a more rapid MAPK activation than upon EGFR activation alone.
PI3‐K is required for MAPK activation but not EGFR transactivation
To assess the role of PI3‐K function in the GPCR‐mediated transactivation signal leading to EGFR tyrosine phosphorylation, we pre‐treated COS‐7 cells with the PI3‐K inhibitor wortmannin (100 nM) prior to LPA or EGF stimulation. As shown in Figure 7A, wortmannin did not affect EGFR tyrosine phosphorylation in response to either different LPA concentrations or 3 ng/ml EGF. Similar results were obtained for M2R‐ or BombR‐mediated EGFR transactivation. and even up to 10 μM wortmannin had no effect on the LPA response (data not shown). Thus PI3‐K is not involved in cross‐talk between GPCRs and the EGFR in COS‐7 cells. Next, we analyzed the effect of the PI3‐K inhibitors wortmannin and LY294002 (Hawes et al., 1996) on GPCR‐mediated MAPK activation. Both inhibitors significantly reduced MAPK activity stimulated through M2R or in response to LPA, as reported previously (Hawes et al., 1996; Lopez‐Ilasaca et al., 1997). Moreover, we found strong reduction of BombR‐mediated MAPK stimulation. Surprisingly, PI3‐K inactivation also markedly reduced MAPK activation induced by low doses of EGF (3 ng/ml), while the response to constitutively active Ras was not significantly affected (Figure 7B). Similar results were obtained when we examined tyrosine phosphorylation of endogenous ERK2 (data not shown). Our findings confirm that PI3‐K function plays an essential role upstream of Ras in GPCR‐mediated MAPK stimulation and, furthermore, indicate an important function downstream of the EGFR. In order to identify a potential site of PI3‐K action downstream of EGFR, we tested whether PI3‐K might interact with GST–Grb2 fusion protein. As shown in Figure 7C, the p85α subunit specifically interacted with GST–Grb2 in unstimulated cells, consistent with previous work (Wang et al., 1995). Interestingly, however, stimulation with LPA or 3 ng/ml EGF enhanced this interaction. Moreover, this response was abrogated when cells were pre‐treated with EGFR‐specific inhibitor AG1478, although the basal level of Grb2‐associated p85α was unaffected by EGFR inactivation. Furthermore, we detected an increase in PI3‐K activity associating with GST–Grb2 upon LPA treatment, which was reversed upon AG1478 pre‐treatment (Figure 7D). Control incubations indicated that most of the detected PI‐K activity reflected PI3‐K activity specifically associated with the GST–Grb2 fusion protein. Although we did not discriminate whether PI3‐K interacts either directly with Grb2 or indirectly via a linker protein such as, for example, Gab1, EGFR‐mediated recruitment of functionally active PI3‐K into Grb2 complexes might represent the critical, PI3‐K inhibitor‐sensitive step necessary for subsequent activation of the MAPK cascade.
Effects of the Src tyrosine kinase inhibitor PP1
Since cytoplasmic tyrosine kinases of the Src family have been implicated recently in GPCR‐mediated MAPK activation (Dikic et al., 1996; Luttrell et al., 1996; Sadoshima et al., 1996; Schieffer et al., 1996), we investigated a possible connection between Src and the EGFR. For this purpose, we utilized the tyrosine kinase inhibitor PP1, which was reported to be specific for Src‐like kinases (Hanke et al., 1996). Pre‐treatment of cells with 10 μM PP1 attenuated tyrosine phosphorylation of endogenous Src. The residual Src tyrosine phosphorylation is most likely due to CSK‐mediated phosphorylation at Tyr527, a non‐autophosphorylation site with inhibitory functions (Luttrell et al., 1997). Expression of a constitutively active Src mutant lacking this tyrosine resulted in increased Src tyrosine phosphorylation which was strongly inhibited upon PP1 pre‐incubation. Furthermore, PP1 suppressed tyrosine phosphorylation of various Src substrates in COS‐7 cells (Figure 8A) and led to moderately increased EGF‐induced EGFR tyrosine phosphorylation, possibly due to an effect on EGFR down‐regulation, while LPA‐induced transactivation was somewhat inhibited (Figure 8B). This indicated that PP1 does not directly inhibit EGFR kinase and further suggests that Src‐like kinases or another PP1‐sensitive kinase partially contribute to EGFR transactivation upon LPA stimulation of COS‐7 cells. Surprisingly, however, PP1 treatment strongly interfered with LPA‐ and EGF‐stimulated SHC and Gab1 tyrosine phosphorylation, Gab1–Grb2 complex formation and MAPK activation (Figure 8C–E). Since PP1 did not affect MAPK activation in response to constitutively active Ras, we conclude that PP1 targets essential signaling elements upstream of Ras and downstream of the EGFR. Moreover, quantification of this data revealed that PP1‐induced inhibition of Gab1 tyrosine phosphorylation was most similar to the effects observed at the level of MAPK, suggesting that this multisite docking protein represents the critical link to EGFR‐dependent activation of the Ras/MAPK pathway in COS‐7 cells. In general, our results establish PP1 as an excellent pharmacological reagent capable of dissociating EGFR tyrosine phosphorylation from immediate receptor‐proximal signaling steps such as SHC and Gab1 tyrosine phosphorylation.
We have demonstrated here that ectopic expression of Gq‐ and Gi‐coupled receptors reconstitutes GPCR agonist‐induced stimulation of EGFR tyrosine phosphorylation in COS‐7 cells. LPA, acting through its endogenous receptor, utilizes both PTX‐insensitive and ‐sensitive pathways leading to EGFR transactivation. While, as mentioned above, it is very likely that heterotrimeric Gi proteins account for the PTX‐sensitive contribution to the LPA response (van Biesen et al., 1996b), it remains to be established whether Gq signaling, although sufficient, indeed mediates the PTX‐insensitive effects on EGFR tyrosine phosphorylation or, alternatively, whether other PTX‐insensitive heterotrimeric G proteins may be involved in LPA signaling. Importantly, LPA treatment or stimulation with low doses of EGF, which activated EGFR to an extent similar to LPA, resulted in comparable tyrosine phosphorylation of Grb2‐interacting proteins such as SHC, indicating a similar signaling capacity of either transactivated or EGF‐activated EGFR in intact cells. Furthermore, LPA induced tyrosine phosphorylation of the docking protein Gab1, which quantitatively accounted for the tyrosine‐phosphorylated 110 kDa protein previously described as associating with Grb2 (Holgado‐Madruga et al., 1996). Conversely, this result demonstrates that tyrosine‐phosphorylated proteins of similar molecular weight, such as, for example, focal adhesion kinase or c‐Cbl, are not present at detectable levels, although these tyrosine kinases can bind to Grb2 (Meisner et al., 1995; Schlaepfer et al., 1997). Moreover, since the docking protein Gab1 plays an important role in RTK‐mediated definition of biological responses (Holgado‐Madruga et al., 1996; Weidner et al., 1996), our identification of Gab1 as a substrate molecule for LPA‐induced tyrosine phosphorylation suggests that this signaling molecule might also be critical for GPCR‐initiated signal transmission.
Focusing on the tyrosine kinase substrate SHC for further analysis, we found induction of tyrosine phosphorylation mediated either through Gq‐coupled BombR or through Gi‐coupled M2R. These findings extend previous results from Lefkowitz and colleagues who attributed GPCR‐mediated SHC tyrosine phoshorylation solely to Gi signaling in COS‐7 cells (van Biesen et al., 1995; Luttrell et al., 1996). In fact, our experiments show that transiently transfected COS‐7 cells are a useful model system to study SHC tyrosine phosphorylation, and our results are consistent with previous data that implicated Gq‐mediated pathways triggered by angiotensin II, bradykinin or thrombin (Cazaubon et al., 1994; Ohmichi et al., 1994; Lev et al., 1995; Chen et al., 1996; Sadoshima and Izumo, 1996) as well as activation of the Gi‐coupled N‐formyl chemoattractant receptor (Ptasznik et al., 1995) in SHC tyrosine phosphorylation and SHC–Grb2 complex formation. Importantly, we found that both signaling events were completely abrogated upon BombR or M2R stimulation when EGFR function was blocked specifically. Thus, EGFR links signals emanating from Gq‐ and Gi‐coupled receptors to the adaptor protein SHC. Although there are conflicting reports on the involvement of SHC in GPCR‐mediated activation of the Ras/MAPK pathway (Chen et al., 1996; Kranenburg et al., 1997; Lopez‐Ilasaca et al., 1997), this adaptor and, perhaps even more importantly, Gab1 appear to play a significant role in the signal transmission pathway via Grb2 and Sos, resulting in activation of the Ras/MAPK signal (van Biesen et al., 1995; Dikic et al., 1996). Consistent with this role for the EGFR, we observed that BombR‐mediated MAPK activation was nearly abolished and the M2R‐mediated response was significantly reduced when we specifically inactivated endogenous EGFR either by expression of the dominant‐negative EGFR truncation mutant HER‐CD533 or by treatment of cells with nanomolar concentrations of the selective EGFR inhibitor AG1478. While there is a consensus that Ras is required for Gi‐mediated MAPK activation, one must point out that previous studies have produced contradictory results, reporting either Ras‐dependent or Ras‐independent MAPK activation through Gq‐coupled receptors in COS‐7 cells (Crespo et al., 1994; Hawes et al., 1995; Pace et al., 1995). In the case of BombR, expression of a dominant‐negative Ras allele interferes with MAPK activation to a similar extent as EGFR inhibition, consistent with the notion that EGFR is positioned upstream of Ras. Moreover, Gq signaling is sufficient to activate Ras (Sadoshima and Izumo, 1996), and further evidence exists that signaling factors such as SHC, Grb2 and Sos, which act upstream of Ras, play essential roles in Gq‐mediated MAPK activation (Chen et al., 1996; Dikic et al., 1996). With respect to the contradictory results about Gq signaling in COS‐7 cells, one possible explanation could be that up to a certain receptor expression level Ras is required and, when expression exceeds this threshold, additional pathways are activated which allow MAPK activation via Ras‐independent routes.
In experiments where we pre‐treated cells with the PI3‐K inhibitor wortmannin, no differences in GPCR‐induced EGFR tyrosine phosphorylation were detected, thus PI3‐K activation does not account for an intermediate step in this pathway. Interestingly, PI3‐K inhibitors not only attenuated MAPK activation in response to GPCR stimulation, but also in response to low doses of EGF, that mimicked, for example, the LPA effect on EGFR tyrosine phosphorylation. Thus, under these conditions, PI3‐K might be required downstream of the EGFR. This assumption is supported further by our finding that recruitment of endogenous, catalytically active p85α/p110 PI3‐K into Grb2‐containing complexes occurs downstream of the EGFR. In addition, the PI3‐Kγ isoform, which is activated by βγ‐subunits of heterotrimeric G proteins, was implicated recently in Gi‐mediated MAPK activation in COS‐7 cells (Lopez‐Ilasaca et al., 1997). However, the involvement of this enzyme in BombR‐mediated or EGF‐stimulated MAPK activation is unlikely since it occurs in a βγ‐independent manner (Faure et al., 1994). Although it is presently unclear how PI3‐K activity might contribute to activation of the MAPK pathway, one potentially interesting clue comes from observations that the PI3‐K product phosphatidylinositol‐3,4,5‐trisphosphate (PIP3) can interact directly with SH2 domains and might thereby contribute to the recruitment of SH2 domain‐containing proteins such as p85, Grb2 or SHC to the plasma membrane (Rameh et al., 1995). Consistent with this hypothesis, partial reduction of SHC tyrosine phosphorylation following GPCR stimulation was detected upon wortmannin treatment (Touhara et al., 1995; Lopez‐Ilasaca et al., 1997). In addition, physical interaction of p85α/p110 PI3‐K with multiprotein complexes containing Grb2, SHC and Gab1 could lead to locally increased PIP3 levels which stabilize membrane association and thereby support further signal transmission. Moreover, the PI3‐K product phosphatidylinositol‐3,4‐bisphosphate (PIP2) binds to the pleckstrin homology (PH) domain of the serine/threonine kinase Akt (Franke et al., 1997). While it is presently unclear whether this finding can be extended to other PH domain‐containing proteins such as Sos or Gab1, it is noteworthy that, in addition to the C‐terminal proline‐rich part of Sos which interacts with Grb2, the N‐terminal region comprising the PH domain is essential for Sos function (McCollam et al., 1995; Byrne et al., 1996). Further, homing of Sos to the plasma membrane might also be supported by βγ‐subunits of heterotrimeric G proteins, which can interact directly with PH domain‐containing proteins such as β‐adrenergic receptor kinase (Shaw, 1996).
Finally, employing the tyrosine kinase inhibitor PP1, which slightly increased EGF‐induced and only moderately inhibited LPA‐induced EGFR tyrosine phosphorylation, we provide evidence that additional signaling factors, most probably tyrosine kinases, act downstream of the EGFR and significantly contribute to activation of the Ras/MAPK pathway. Although PP1 has been reported to be selective for Src‐like kinases, our data must be interpreted carefully since inhibitory effects on other tyrosine kinases such as PYK2 may be possible, which could account for the effects observed. Moreover, there are some discrepancies concerning the role of Src in GPCR‐mediated activation of the Ras/MAPK pathway. This is especially true in view of rather diverging reports regarding the participation of Src in GPCR‐mediated activation of the Ras/MAPK pathway (Dikic et al., 1996; Luttrell et al., 1996; Sadoshima et al., 1996; Schieffer et al., 1996; Kranenburg et al., 1997). It can be anticipated, however, that the identification of PP1 target molecules will provide important insights into the Ras/MAPK activation mechanism not only in response to GPCR stimulation, but also upon direct RTK ligand treatment. It is tempting to speculate that an autophosphorylated EGFR recruits substrates such as SHC which are thereby brought into proximity with and phosphorylated by an additional, PP1‐sensitive tyrosine kinase such as Src. Recently, Luttrell et al. (1977) reported that Src family kinases mediate PTX‐sensitive or βγ‐induced EGFR tyrosine phosphorylation in COS‐7 cells, and suggested that this response does not involve activation of intrinsic EGFR kinase activity because of differential use of receptor phosphorylation sites. In our study, however, both the EGFR inhibitor AG1478, which targets the EGFR kinase domain, and the EGFR truncation mutant HER‐CD533, which selectively blocks endogenous EGFR through formation of inactive heterodimers, fully abrogate EGFR tyrosine phosphorylation without affecting Src‐induced tyrosine phosphorylation (Daub et al., 1996; and data not shown). Moreover, our findings support the conclusion that the signal generation by either transactivation of the EGFR or direct EGF binding are indistinguishable.
While the mechanistic details of GPCR‐mediated EGFR transactivation remain elusive, data obtained in other ligand‐independent RTK activation scenarios such as UV irradiation or oxidant‐induced EGFR tyrosine phosphorylation (Knebel et al., 1996) strongly support the involvement of EGFR‐antagonizing phosphatases whose inhibition could account for the GPCR‐mediated responses observed. Alternatively, interaction with and perhaps even covalent modification by intermediate signaling factors might lead to GPCR‐induced EGFR transactivation. Although the signaling events that immediately precede EGFR transactivation await further investigation, the data presented here not only establish the general significance of EGFR function in signals of diverse GPCR subclasses in diverse cell types, but further identify additional elements of GPCR and RTK signals, thereby providing important and novel insights into the signaling network involving two major classes of cell surface receptors.
Materials and methods
Reagents, antibodies and plasmids
COS‐7 cells were from Genentech Inc. Human HaCaT keratinocytes were a gift from N.E.Fusenig (Boukamp et al., 1988). Primary mouse astrocytes were prepared essentially as described (Jehan et al., 1995). Culture media and Lipofectamine were purchased from Gibco‐BRL. Protein A–Sepharose was from Pharmacia Biotech Inc. PP1 and LY294002 were obtained from Calbiochem. PDGF B/B was from Boehringer Mannheim, and AG1478 (Levitzki and Gazit, 1995) was kindly provided by P.Hirth. All other reagents were obtained from Sigma.
Antibodies purchased were monoclonal anti‐HA antibody (Boehringer Mannheim), rabbit polyclonal anti‐EGFR antibody (Santa Cruz), sheep polyclonal anti‐EGFR antibody (UBI), mouse monoclonal anti‐Grb2 antibody (UBI), mouse monoclonal anti‐PI3‐K antibody (UBI), rabbit polyclonal anti‐ERK2 antibody (Santa Cruz) and mouse monoclonal anti‐Src antibody (Calbiochem). Endogenous EGFR from COS‐7 cells was precipitated with mouse monoclonal antibody 108.1 (Seedorf et al., 1994); mouse monoclonal anti‐PDGFRβ antibody 128D4C10 was a gift from H.J.Bühring. Rabbit polyclonal anti‐SHC antibody has been described (Seedorf et al., 1994). For Gab1 antibody generation, Gab1 cDNA was cloned from a human placenta cDNA library, the amino acids 23–189 were expressed as pGEX fusion protein in Escherichia coli and, after purification, used as antigen in rabbits. Identity with the published sequence (Holgado‐Madruga et al., 1996) was verified by DNA sequencing. Respective secondary antibodies were obtained from Bio‐Rad and Dianova. For immunoblot detection, the ECL system from Amersham was used. Stripping and reprobing of blots were performed according to the manufacturers' recommendations.
All cDNAs used except SrcY527F (Luttrell et al., 1997), which was in a retroviral expression vector (Daub et al., 1996), were in cytomegalovirus‐based expression plasmids containing HER‐CD533 (Redemann et al., 1992), βPDGFR (Vogel et al., 1993), βPDGFR‐CD504 (Strawn et al., 1994), IGF1‐R (Schuhmacher et al., 1993), Ha‐Ras (Lengyel et al., 1995), v‐Abl (Gishinzky et al., 1995), BombR (Lustig et al., 1993), M1R (Crespo et al., 1994) and M2R (Faure et al., 1994). Mouse ERK2 (Her et al., 1991) was fused to an N‐terminal HA epitope as described (Meloche et al., 1992).
Cell culture and transfections
HaCaT keratinocytes and COS‐7 cells were cultured in Dulbecco‘s modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. Primary astrocytes were maintained in DMEM F‐12 containing 10% fetal bovine serum. COS‐7 cells were transiently transfected using Lipofectamine essentially as described (Hawes et al., 1995). For transfection in 12‐well dishes, cells were incubated for 4 h in 0.4 ml of serum‐free medium containing 4 μl of Lipofectamine and 0.8 μg of total plasmid DNA per well. For transfections in 6‐well dishes, 1.0 ml of serum‐free medium containing 10 μl of Lipofectamine and 2.0 μg of total plasmid DNA per well were used. The transfection mixture was then supplemented with an equal volume of medium containing 20% fetal bovine serum and, 20 h later, cells were washed and cultured for a further 24 h in serum‐free medium until lysis. For [35S]methionine labeling, cells were incubated in methionine‐free DMEM during starvation, and 50 μCi/ml [35S]methionine were added 16 h prior to lysis.
Cell lysis, immunoprecipitation, association with fusion proteins and immunoblotting
Prior to lysis, cells grown to 80–90% confluency were treated with inhibitors and agonists as indicated, washed once with phosphate‐buffered saline and then lysed for 10 min on ice in buffer containing 50 mM HEPES pH 7.5, 150 mM NaCl, 1% Triton X‐100, 1 mM EDTA, 10% glycerol, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 10 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride and 10 μg/ml aprotinin. Lysates were pre‐cleared by centrifugation at 13 000 r.p.m. for 10 min at 4°C. Supernatants were diluted with an equal volume of HNTG buffer (Seedorf et al., 1994) and subsequently immunoprecipitated using the respective antibodies and 30 μl of protein A–Sepharose for 4 h at 4°C. Alternatively, lysates were subjected to in vitro associations with either 3 μg of GST–Grb2 or 2 μg of GST as control pre‐bound to 30 μl of gluthathione–agarose beads. Precipitates were washed three times with 0.5 ml of HNTG buffer, suspended in SDS sample buffer, boiled for 3 min and subjected to gel electrophoresis on 7.5% gels or 6–12.5% gradient gels. Following SDS–PAGE, proteins were transferred to a nitrocellulose membrane and immunoblotted.
Epitope‐tagged HA‐ERK2 was immunoprecipitated from lysates obtained from 12‐well dishes using 2.5 μg of 12CA5 antibody; endogenous ERK2 was immunoprecipitated using 0.5 μg of polyclonal anti‐ERK2 antibody. Immunoprecipitates were washed three times with 0.25 ml of HNTG buffer and once with 0.4 ml of kinase buffer (20 mM HEPES pH 7.5, 10 mM MgCl2, 1 mM dithiothreitol, 200 μM sodium orthovanadate). Subsequently, kinase reactions were performed in 30 μl of kinase buffer supplemented with 0.5 mg/ml MBP, 50 μM ATP and 1 μCi of [γ‐32P]ATP for 10 min at room temperature. Reactions were stopped by addition of 30 μl of 2× SDS sample buffer and subjected to gel electrophoresis on 15% gels. The upper parts of the gels were transferred to nitrocellulose membrane and immunoblotted with anti‐ERK2 antibody; labeled MBP on the lower part was quantified using a PhosphorImager (Fuji).
Phosphatidylinositol 3‐kinase assay
PI3‐K assays were performed essentially as described (Wallasch et al., 1995), with some modifications. Briefly, GST–Grb2 in vitro associations were washed twice with HNTG and then three times with 1 ml of buffer containing 50 mM HEPES pH 7.5, 150 mM NaCl, 0.5 mM EGTA, 0.2 mM sodium orthovanadate and 10% glycerol. Then 30 μl of a master mix consisting of 10 μl of 5× buffer (100 mM Tris–HCl pH 7.5, 500 mM NaCl, 25 mM Mg acetate, 25 mM MnCl2, 2.5 mM EGTA), 10 μl of sonicated phosphatidylinositol (0.5 mg/ml in 20 mM HEPES pH 7.5), 5 μl of 2 mM adenosine and 5 μl of H2O were added to the beads followed by a pre‐incubation step for 10 min at room temperature. The kinase reaction was started by addition of 10 μl of 0.1 mM ATP containing 2.5 μCi of [γ‐32P]ATP and performed for 30 min at 30°C. The reaction was stopped by addition of 100 μl of 1 M HCl, and lipids were extracted from supernatants with 200 μl of methanol:chloroform (1:1 v/v). The lipid‐containing organic phase was resolved on oxalate‐coated, thin‐layer chromatography plates (Merck) developed in methanol:chloroform:water:glacial acetic acid:acetone (40:13:7:12:15 by vol.). The plates were then dried and exposed for PhosphorImager (Fuji) analysis.
We are grateful to C.J.Marshall, J.Schlessinger, H.R.Bourne, M.Faure, R.Wetzker, E.Lengyel, M.L.Gishizky, N.E.Fusenig, P.Hirth and H.J. Bühring for kindly providing us with different plasmids and reagents. We thank Cordula Ebner‐von Eschenbach for excellent technical assistance and our colleagues E.Zwick, B.Fernandez, F.U.Weiss, J.Schneckenburger and R.Lammers for various DNA constructs, reagents and stimulating discussions. This work was supported by a grant from Sugen, Inc.
- Copyright © 1997 European Molecular Biology Organization