The signaling pathway coupling epidermal growth factor receptors to activation of p21ras.

Epidermal growth factor (EGF) treatment causes autophosphorylation of the epidermal growth factor receptor (EGFR) leading to increased guanine nucleotide exchange factor (GEF; Sos) activity and enhanced formation of p21ras-GTP. The connection of the EGFR to p21ras activation can occur through binding of Grb2.Sos complexes to the EGFR or through the adaptor protein Shc via EGFR.Shc.Grb2.Sos multimeric complexes. Therefore, we investigated the importance of Shc in coupling the EGFR to activation of ras GEF (Sos). EGF treatment led to rapid tyrosine phosphorylation of Shc. Although phosphorylated EGFR can bind to both Shc and Grb2, the predominant linkage was observed between EGFR and Shc. Similarly, more Grb2 was associated with Shc than with EGFR after EGF stimulation. Immunoprecipitation of Shc from EGF-stimulated cells removed almost all EGFR-associated Grb2. Furthermore, immunodepletion of Shc proteins from membrane fractions of EGF-stimulated cells removed 93% of the ras GEF activity, whereas, precipitation of EGFR had only a small effect on ras GEF activity. These data indicate that coupling to Shc provides the major pathway linking activated EGFRs to Grb2.Sos and stimulation of the p21ras pathway.

In this report, we evaluated the relative contributions of the EGFR.Grb2.Sos or EGFR.Shc.Grb2.Sos pathways in mediating EGF-induced GEF activation of p21"". Our results indicate the importance of Shc as an adapter protein transducing biologic signals from activated EGFR to the p21"" pathway.

Cell Lines and Materials-Rat1 fibroblasts were maintained in
Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 50 ng/ml gentamycin (17). The p21"" probe (c-Ha-Ras) was a gift from Dr. Alan Wolfman (Cleveland Clinic Foundation). EGF was purchased from Life Technologies, Inc. r3H1GDP (32 Ciinmol) was from DuPont-NEN. Electrophoresis reagents were from Bio-Rad. Enhanced chemiluminescence reagents were from Amersham Corp. A monoclonal anti-phosphotyrosine antibody (pY201, a polyclonal and a monoclonal anti-Shc antibody, and a monoclonal anti-Grb2 antibody were from Transduction Laboratories (Lexington, KY). Polyclonal Grb2 antibodies were from Upstate Biotechnology (Lake Placid, N Y ) and Santa Cruz (Santa Cruz, CA). A polyclonal anti-EGF receptor antibody was kindly provided by Dr. Stuart J. Decker (Parke-Davis Pharmaceuticals, MI). All other routine reagents were purchased from Sigma.
The cell lysates were centrifuged to remove insoluble materials. The supernatants (50 pg of protein) were used for immunoprecipitation with the indicated antibodies for 5 h at 4 "C. The entire precipitate and all of the remaining supernatant proteins were then separated by SDS-PAGE and transferred to Immobilon-P by electroblotting. For immunoblotting, membranes were blocked and probed with specified antibodies. Blots were then incubated with horseradish peroxidase-linked second antibody followed by enhanced chemiluminescence detection, according to the manufacturer's instructions (Amersham Corp.) (18). Based on immunoprecipitation and Western blot results, we estimate the efficiency of precipitation as 80-90% for anti Shc, 7 0 4 0 % for anti Grb2, and 80-90% for anti EGFR.
Measurement of GTP-and GDP-bound pBl""-As described previously (191, cells were  serum starved for 16 h, labeled with 132Plorthophosphate, and stimulated with 130 nM EGF for varying times. After cell lysis, Ras was immunoprecipitated, and the nucleotides were eluted from the immunoprecipitate and separated by thin layer chromatography. Measurement of GEF Activity in Membranes-Cells were starved for 16 h in serum-free Dulbecco's modified Eagle's medium. The cells were then treated with 130 nM EGF at 37 "C for 2 min. The cells were then collected in a buffer containing 50 mM Hepes, 150 mM NaCI, 10 mM MgCI,, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na,HPO,, 1 mM Na,VO,, 10 pg/ml leupeptin, 10 pg/ml aprotinin, 1 mM dithiothreitol, pH 7.5. The cells were disrupted by 20 strokes of a tight fitting Dounce homogenizer. The homogenate was centrifuged a t 3,000 rpm in an Eppendorf 5402 centrifuge a t 4 "C for 3 min to remove the nuclear fraction. The supernatants were re-centrifuged a t 220,000 x g a t 4 "C for 60 min. The particulate fraction was suspended in a buffer containing 0.05% SDS, 0.1% Triton X-100, 50 mM Hepes, 150 mM NaCI, 10 mM MgCl,, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na,HPO,, 1 mM Na,VO,, 10 pg/ml leupeptin, 10 pg/ml aprotinin, 1 mM dithiothreitol, 100 PM GTP, 100 PM GDP, pH 7.5, and sonicated at 4 "C for 30 s. For immunodepletion studies, the extract was immunoprecipitated with specified antibodies a t 4 "C for 3 h, and the supernatants were used for GEF activity assay. The GEF activity in the membranes was determined by measuring the dissociation of protein-bound I3H1GDP radioactivity using a nitrocellulose filter binding assay as described previously (19). Fig. 1 shows the time course of EGF-stimulated tyrosine phosphorylation of Shc. To assess Shc phosphorylation, cell lysates were immunoprecipitated with anti-Shc antibody, and the precipitates were immunoblotted with anti-phosphotyrosine antibody. As shown in Fig. lA, the 46-, 52-, and 66-kDa Shc isoforms were tyrosine-phosphorylated upon EGF stimulation, and the 52-kDa isoform was the major phosphorylated species. Peak phosphorylation of Shc was observed by 30 s and declined after 2.5 min. These results are summarized in Fig. 1B. It is clear that tyrosine phosphorylation of Shc proceeds rapidly, consistent with the notion that activated EGF receptors bind to Shc directly (12)(13)(14)(15). The time course of EGFstimulated p21m"-GTP formation is shown in Fig. lC, and the functional form of this time course is consistent with the kinetics of Shc phosphorylation.

RESULTS
It has been shown that both Shc and Grb2 can associate with the EGFR following EGF stimulation (12)(13)(14)(15), and therefore the time course and the amount of Shc and Grb2 association with the EGFR were assessed. After EGF treatment, cell lysates were immunoprecipitated with anti-Shc antibody, and the precipitates and supernatants were immunoblotted with antiphosphotyrosine antibody. As seen in Fig. 2 4 , a 175-kDa phosphoprotein band was co-precipitated with the anti-Shc antibody, and the identity of this band as the EGFR was confirmed by using specific EGFR antibodies (data not shown). The EGFR.Shc association was rapid, with peak complex formation detected at 30 s, declining thereafter ( Fig. 2 A ) . Thus, the time course of this EGFR-Shc complex formation was comparable to the kinetics of Shc phosphorylation as seen in Fig. 1. Similar experiments were performed using anti-Grb2 antibody. Following EGF stimulation, cell lysates were immunoprecipitated with anti-Grb2 antibody, and the precipitates and supernatants were immunoblotted with anti-phosphotyrosine antibody. Tyrosine-phosphorylated EGFRs were co-precipitated by the anti-Grb2 antibody, but the magnitude of EGFR.complex formation was much less than for EGFR.Shc complexes (Fig. 2 B ) . For example, a t 1-2.5 min, -55% of EGFR precipitates with anti-Shc, compared to -20% with anti-Grb2. EGF treatment can also lead to tyrosine phosphorylation of erbB2 (81, which is present in rat fibroblasts (20). ErbB2 has a reported mobility of -185 kDa on SDS-PAGE (201, which should allow distinction from the 175-kDa EGFR. Interestingly, as seen in Fig. 2, there is a variable EGF stimulated phosphoprotein band, mostly in isoform is summarized as the percentage of maximal tyrosine phosphorylation. Results are representative of two separate experiments. C , cells were stimulated with EGF for the indicated times, p21"" was precipitated from the cell lysates, and nucleotides were eluted from the precipitate and then separated by thin layer chromatography. Results are expressed as the increment of GTP/(GTP + GDP) x 100% over basal. the supernatants, which runs above the EGFR, consistent with erbB2. By Western blotting with anti-EGFR antibody we have confirmed that the 175-kDa phosphoprotein band is the EGFR and no higher molecular mass bands were identified. Nevertheless, we cannot exclude the possibility that a small amount of erbB2 is co-mingled in the EGFR band.
We next assessed the association of Grb2 with Shc andor the EGFR. After EGF treatment, cell lysates were immunoprecipitated with anti-Shc or anti-EGFR antibody, and the precipitates and supernatants were immunoblotted with anti-Grb2 antibody. Formation of complexes containing Shc.Grb2 (Fig.  3 A ) andor EGFR.Grb2 (Fig. 3B) was rapid. Although the kinetics of Grb2 association with Shc or EGFR were similar, a substantially larger amount of Grb2 associated with Shc than with EGFR.
Another way to quantitate the association of Grb2 with Shc and EGFR is to conduct sequential immunoprecipitation studies using anti-Shc and anti-EGFR antibodies. After EGF stimulation, cell lysates were first immunoprecipitated with the anti-Shc antibody, and the remaining supernatants were reimmunoprecipitated with anti-EGFR antibody. The anti-Shc and anti-EGFR immunoprecipitates were then immunoblotted with anti-Grb2 antibody. As can be seen in Fig. 4, there was negligible association of Grb2 with Shc or EGFR in the basal state. After EGF stimulation, substantial association of Grb2 with Shc was observed in the anti-Shc immunoprecipitates (Fig. 4, lane 3), whereas there was minimum association of Grb2 with EGFR after the cell lysates were first immunodepleted by anti-Shc antibody (Fig. 4, lane 4 ) . By reversing the order of immunoprecipitation, the presence of Grb2 within complexes containing EGFR could be demonstrated (Fig. 4, lane 7).  (lanes 1 , 2 , 5 , 6 ) , or with EGF (lanes 3 , 4 , 7, 8) for 1 min. Cell lysates were immunoprecipitated with anti- Shc (lanes 1 4 ) or anti-EGFR (lanes 5-81 antibody. The anti-Shc and anti-EGFR supernatants were then re-immunoprecipitated with anti- EGFR (lanes 2 and 4 ) or anti-Shc (lanes 6 and 8) antibody, respectively. The anti-Shc and anti-EGFR precipitates were separated by SDS-PAGE and immunoblotted with anti-Grb2 antibody. Molecular mass of Grb2 (25 kDa) is shown by an arrow.

GEF activity in membrane fraction after immunodepletion
Serum-starved cells were stimulated with 130 nM EGF for 2 min a t 37 "C. GEF activity in the membrane fraction before and after immunoprecipitation with specified antibodies was determined by measuring the loss of protein-bound ['HIGDP radioactivity. Results are expressed as the percentage of reduction from total GEF activity (100%) by immunodepletion and are shown as the mean e S.E. of four separate experiments.

Antibody
Reduction of GEF activity Most likely these represent EGFR-Shc.Grb2 containing complexes. However, a much greater amount of Grb2 could be precipitated by anti-Shc antibody even when the lysates were first immunodepleted by anti-EGFR antibody (Fig. 4, lane 8).
It has been shown that Sos contains GEF activity toward p21"" in vitro and that stimulation with EGF does not alter total cellular GEF activity (14). Rather, a major proportion of Sos is translocated from the cytosol to the plasma membrane fraction after EGF stimulation (14). Consequently, we measured GEF activity in membrane fractions following EGF stimulation and found that EGF increased by 2.3-fold the ability of the membrane fraction to enhance ras guanine nucleotide dissociation (from 19.4 2 4% to 45.6 2 5% [3H]GDP released).
It has been suggested that preformed Grb2.Sos complexes exist in cells (6,(12)(13)(14)(15), and consistent with this, we found that immunodepletion by anti-Grb2 antibody removed 88.4 2 4% of the EGF-stimulated membrane GEF activity (Table I). EGF stimulation leads to direct association of these complexes with either Shc or EGFR (12-E), and to evaluate the relative magnitudes of these effects, we measured GEF activity in the membranes before and after immunoprecipitation of the fractions with anti-Shc andor anti-EGFR antibody. As can be seen in Table I, precipitation of Shc from EGF-stimulated preparations removed 92.5 2 2% of the total membrane GEF activity, while immunoprecipitation by anti-EGFR antibody only removed 39.3 2 17% of total membrane GEF activity. DISCUSSION EGF and other growth factors stimulate the formation of p21mr-GTP, and this plays an important role in mediating the overall mitogenic response (2, 3). For EGF, this occurs largely by stimulation of GDP dissociation from p21"", with replacement by GTP (8-101, and this guanine nucleotide exchange is mediated by Sos proteins (4-6). Since Grb2 exists in preformed complexes with Sos in unstimulated cells (6,(12)(13)(14)(15), Grb2 is also a critical signaling molecule connecting EGFRs to p21". Recent studies have clearly established the importance of Grb2 in growth factor action. For example, microinjection of Grb2, together with H-rus protein, into quiescent rat embryo fibroblasts resulted in enhanced DNA synthesis (11). In addition, microinjection of an anti-Grb2 antibody into normal rat kidneyderived fibroblasts inhibited EGF and platelet-derived growth factor stimulation of cell cycle progression (21). Therefore, a key to understanding how EGF causes increased p2lrn"-GTP lies in the identification of the upstream linkages which connect the EGFR to Grb2.Sos.
After EGF stimulation, there are four possible mechanisms of molecular linkage from the EGFR to Sos and stimulation of p21"", and these are summarized schematically in Fig. 5. First, the tyrosine phosphorylated EGFR could bind directly to the Grb2 SH2 domain forming an EGFR.Grb2.Sos complex (Fig.  5A). However, it is unlikely that this mechanism is a major one, a t least in Rat1 fibroblasts, since anti-EGFR antibody precipitated a relatively small amount of Grb2, compared to anti-Shc antibody, and, equally importantly, the anti-EGFR antibody precipitated a negligible amount of Grb2 from lysates which had been already precipitated with anti-Shc antibody. Second, the EGFR could complex with Grb2.Sos complexes via Shc, forming a multimeric unit consisting of EGFR.Shc-Grb2.Sos (Fig. 5B). Our data provide strong evidence that this is an important component of EGF signaling, since anti-Shc antibody co-precipitates a large amount of phosphorylated EGFR, as well as much of the cellular Grb2 and GEF activity. Likewise, after anti-Shc antibody precipitation, when the remaining EGFRs were precipitated with anti-EGFR antibody, only a negligible amount of Grb2 and GEF activity was recovered. Our results also exclude the possibility that appreciable amounts of She alone bind to EGFRs without forming Shc.Grb2.Sos complexes, since almost all tyrosine phosphorylated Shc was coprecipitated by anti-Grb2 antibody (data not shown). Third, it is possible that a single EGFR can concomitantly associate with both Grb2-Sos and Shc.Grb2.Sos complexes (Fig. 5 0 . Obvi-ously, it is experimentally difficult to separate this situation from the EGFR.Shc.Grb2.Sos complexes as in Fig. 5B. However, in both of these cases, Shc would remain a quantitatively more important linkage between the EGFR and Grb2.Sos complexes, since anti-Shc antibody precipitates the greater amount of EGFR and GEF activity. Finally, following EGF stimulation, Shc.Grb2.Sos complexes could exist, independent of the EGFR (Fig. 5 0 ) . Our data provide strong evidence for this possibility, since following immunodepletion of cell lysates from EGFstimulated cells with anti-EGFR antibody, anti-Shc antibody was able to co-precipitate a relatively large amount of Grb2. In addition, anti-Shc antibody precipitation removed far more GEF activity than did simple anti-EGFR antibody precipitation, consistent with the presence of Shc.Grb2.Sos complexes independent of the EGFR.
Taken together, our data are consistent with the formulation that following autophosphorylation of EGFR, Shc binds to the EGFR through its SH2 domain (12)(13)(14)(15). This facilitates tyrosine phosphorylation of Shc and association of Shc with Grb2-Sos complexes which then, in turn, mediate the formation of p21""-GTP within a multimeric EGFR.Shc.Grb2.Sos complex (12)(13)(14)(15). However, we also find that a substantial amount of Shc.Grb2.Sos complexes exist in the absence of associated EGFR. This could mean that during the process of EGF action, dissociation of Shc.Grb2-Sos from the EGFR occurs, creating the scenario depicted in Fig. 50. Alternatively, it is possible that EGF stimulation could lead to Shc phosphorylation, and subsequent formation of Shc.Grb2.Sos complexes, without any direct interactions between the EGFR and Shc. This would involve an intermediate tyrosine kinase, which is stimulated by the activated EGFR, and phosphorylates Shc directly. This formulation is consistent with recent observations showing that truncated EGFRs, which lack all of the major autophosphorylation sites, still induce Shc tyrosine phosphorylation and complex formation of Shc with Grb2 (22). Even if such a pathway exists, involving an intermediary tyrosine kinase situated between the EGFR and Shc, our results clearly demonstrate the presence of direct linkage between Shc and the EGFR, indicating that both pathways leading from the EGFR to Shc phosphorylation could be operative.
Increasing evidence has accrued in several systems indicating that Shc is the important adaptor molecule linking Grb2.Sos to surface receptors. Thus, it has been reported that Shc is a linking molecule coupling activated T cell receptors (23), interleukin-2 receptors (24), and Trk receptors (25) to Grb2. Shc has also been reported as the predominant coupling molecule linking insulin receptors to Grb2SOS and activation of the ras pathway (26,27). Furthermore, Pronk et al. (26) have found Grb2 and Sos in Shc immunoprecipitates from EGFtreated cells, consistent with the current results implicating Shc as a key adaptor molecule in EGF action. In addition, Grb2 forms complexes with the platelet-derived growth factor receptor by binding to Syp, also called SHFTP2, PTPBC, or PTPlD, which in turn binds to the platelet-derived growth factor receptor through its SH2 domain (28). Taken together with the current results, the association of Grb2.Sos complexes with membrane receptors via Shc, or possibly other adaptor molecules such as Syp, appears to be the physiologically relevant mechanism for activation of ~21"". This line of reasoning is also consistent with our recent studies which demonstrate the functional role of Shc in mediating EGFs biologic action. Thus, we have conducted single-cell microinjection studies showing that microinjection of anti-Shc antibody or Shc SH2 GST fusion proteins into living Rat1 fibroblasts inhibited EGF-induced cell cycle progression by 80% (18).
It is important to consider how activation of this pathway leads to increased formation of p21""-GTP. Clearly, two possibilities exist. First, formation of the multiprotein complexes could serve to translocate Grb2.Sos complexes to the cytoplasmic side of the plasma membrane, where Sos would then gain access to membrane anchored p21"". Alternatively, formation of Grb2.Sos complexes with Shc could lead to activation of the catalytic activity of Sos to mediate GDP dissociation. The current results (Fig. 41, as well as those of others, demonstrating translocation of Sos or GEF to the plasma membrane, plus our recent finding that EGF stimulation does not change total cellular GEF activity, are more consistent with the former translocation hypothesis2 (14).
It is of interest to note that in our studies, the EGF-induced Shc phosphorylation peaked at 1-2 min and then declined by -80% by 20 min. The time course of Shc.Grb2 complex formation (Fig. 3 A ) showed a similar pattern (Fig. l), but only declined by -35% by 20 min. From these differences, it is possible to speculate that within Shc.Grb2 complexes, since the Shc phosphotyrosine is buried in the Grb2 SH2 domain, the phospho-Shc within these complexes may be less accessible to phosphatases and relatively protected from dephosphorylation. Clearly, future experiments will be necessary to evalute this notion.
In summary, our studies indicate the importance of Shc in forming the molecular linkage between EGFR and Grb2.Sos complexes. Following EGF stimulation, Shc associates with and becomes tyrosine-phosphorylated by the EGFR. This leads to the formation of EGFR.Shc.Grb2.Sos complexes. Free Shc.Grb2.Sos complexes also exist, consistent with either dissociation of Shc.Grb2+3os from the EGFR, or a parallel pathway containing a n intermediary tyrosine kinase which phosphorylates Shc following EGFR activation. Either scenario would highlight the importance of Shc as an adaptor protein linking the EGFR to the p21"" pathway. In previous studies, it has been shown that Shc is the major adaptor molecule linking the activated insulin receptor to Grb2.Sos complexes and the p21" pathway (26,27). In combination with the current results, this suggests the general principle that adaptor proteins like Shc are the dominant mechanisms linking activated growth factor receptors to the downstream components of the ~21"" pathway.