The 66-kDa Shc Isoform Is a Negative Regulator of the Epidermal Growth Factor-stimulated Mitogen-activated Protein Kinase Pathway*

In addition to tyrosine phosphorylation of the 66-, 52-, and 46-kDa Shc isoforms, epidermal growth factor (EGF) treatment of Chinese hamster ovary cells expressing the human EGF receptor also resulted in the serine/threonine phosphorylation of approximately 50% of the 66-kDa Shc proteins. The serine/threonine phosphorylation occurred subsequent to tyrosine phosphorylation and was prevented by pretreatment of the cells with the MEK-specific inhibitor PD98059. Surprisingly, only the gel-shifted 66-kDa Shc isoform (serine/threonine phosphorylated) was tyrosine phosphorylated and associated with Grb2. In contrast, only the non-serine/threonine-phosphorylated fraction of 66-kDa Shc was associated with the EGF receptor. To assess the relationship between the three Shc isoforms in EGF-stimulated signaling, the cDNA encoding the 66-kDa Shc species was cloned from a 16-day-old mouse embryo library. Sequence alignment confirmed that the 66-kDa Shc cDNA resulted from alternative splicing of the primary Shc transcript generating a 110-amino acid extension at the amino terminus. Co-immunoprecipitation of Shc and Grb2 from cells overexpressing the 52/46-kDa Shc isoforms versus the 66-kDa Shc species directly demonstrated a competition of binding for a limited pool of Grb2 proteins. Furthermore, expression of the 66-kDa Shc isoform markedly accelerated the inactivation of ERK following EGF stimulation. Together, these data indicate that the serine/threonine phosphorylation of 66-kDa Shc impairs its ability to associate with the tyrosine-phosphorylated EGF receptor and can function in a dominant-interfering manner by inhibiting EGF receptor downstream signaling pathways.

The epidermal growth factor (EGF) 1 receptor is one member of a large family of receptor tyrosine kinases that undergoes ligand-stimulated autophosphorylation and can tyrosine phosphorylate a distinct set of endogenous protein substrates (1)(2)(3)(4)(5). The phosphorylation of these tyrosine residues generates recognition motifs for Src homology 2 (SH2) and phosphotyrosine binding (PTB) domains present in various intracellular signaling molecules (6 -12). The association of tyrosine-phosphorylated receptors and substrates with these effector proteins generates multisubunit signaling complexes responsible for downstream biological responsiveness. One important proximal target for the EGF receptor was originally identified as a series of proteins (66, 52, and 46 kDa), termed Shc for Src homology 2/␣-collagen related (13). Although the carboxyl-terminal SH2 domain was originally presumed to be responsible for its association with the tyrosine-phosphorylated EGF receptor, recent studies have demonstrated that the amino-terminal PTB domain is primarily responsible for EGF receptor association (14 -16). Thus, it is now thought that the Shc SH2 domain appears to play a secondary role by enhancing the affinity of interaction between the Shc proteins and the tyrosine-phosphorylated EGF receptor (8,13,15,(17)(18)(19)(20).
In addition to the binding of Shc to the activated EGF receptor, the Shc proteins are also substrates of the EGF receptor and are predominantly tyrosine phosphorylated on tyrosine residues 239, 240, and 317 (21)(22)(23)(24). In particular, phosphorylation of the Shc proteins on tyrosine 317 generates a high affinity docking site for the SH2 domain of the small adapter protein Grb2 (22,23,(25)(26)(27)(28)(29). Grb2 was identified as a 23-kDa growth factor receptor-binding protein which contains a single SH2 domain flanked by two Src homology 3 (SH3) domains (26,30). Whereas the Grb2 SH2 domain is responsible for its association with tyrosine-phosphorylated Shc, the Grb2 SH3 domains direct interactions with SOS, the 150-kDa guanylnucleotide exchange factor for Ras (29,(31)(32)(33)(34)(35)(36)(37)(38)(39). Thus, the EGFstimulated Shc tyrosine phosphorylation induces the formation of a Shc⅐Grb2⅐SOS ternary complex which is localized to the autophosphorylated EGF receptor through the binding of the Shc PTB domain. In this manner, the guanylnucleotide exchange activity of SOS is targeted to the plasma membrane location of Ras, thereby providing a molecular link from the EGF receptor tyrosine kinase to Ras activation (40,41).
The initial cloning of the cDNA for Shc demonstrated that the 46-and 52-kDa species are produced from the use of alternative translation initiation sites within the same transcript (13). This results in an amino-terminal 59-amino acid truncation of the 46-kDa Shc isoform compared with the 52-kDa Shc isoform. In contrast, the 66-kDa Shc species is believed to arise from an alternatively spliced message as carboxyl-terminal antibodies directed against the 46-and 52-kDa Shc proteins cross-react with the 66-kDa species (13,22,(42)(43)(44). In the present article, we have isolated the full-length cDNA for the murine 66-kDa Shc isoform and demonstrate that EGF stimulation results in both the tyrosine and serine/threonine phosphorylation of this Shc isoform. Furthermore, expression of the 66-kDa Shc isoform inhibits EGF stimulation of the ERK family of mitogen-activated protein kinases. Together these data demonstrate that serine/threonine phosphorylation of the 66-kDa Shc isoform provides a limiting function for ERK activation by serving as a negative modulator of the 52-and 46-kDa Shc signaling.

EXPERIMENTAL PROCEDURES
Cell Culture-Chinese hamster ovary cells expressing the human insulin and EGF receptors (CHO/IR/ER) were isolated and cultured as described previously (45). Cells were incubated for 3-12 h in serum-free media and then incubated with and without 20 nM EGF at 37°C for various times as indicated. Cell extracts were prepared by solubilization in 50 mM HEPES, pH 7.4, 1% Triton X-100, 100 mM sodium fluoride, 2.5 mM EDTA, 10 mM sodium pyrophosphate, 2 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 2 M pepstatin, 0.5 trypsin inhibitory units of aprotinin, and 10 M leupeptin. In experiments utilizing the specific MEK inhibitor, PD98059 (kindly provided by Dr. Alan Saltiel, Parke-Davis/Warner-Lambert), cells were pretreated with vehicle (1% dimethyl sulfoxide) or 100 M PD98059 in 1% dimethyl sulfoxide for 1 h at 37°C.
Isolation of the cDNA for the 66-kDa Shc Isoform-To isolate fulllength cDNA encoding the 66-kDa Shc isoform, a cDNA fragment was generated by the polymerase chain reaction from the murine Shc PTB domain (8). This cDNA was labeled by using a Random-Primed Labeling Kit (U. S. Biochemical) with [ 32 P]dCTP (3000Ci/mmol; NEN Life Science Products). Following screening of a 16-day-old mouse embryo library (Novagen, Madison, WI), one clone (321-6) was isolated that differed from the 52-kDa Shc cDNA sequence in the 5Ј end. Sequence analysis of clone 321-6 revealed an additional start site 330 base pairs upstream of the start site from the 52-kDa Shc cDNA. The nature of this clone was verified by subcloning in the RK5 eukaryotic expression vector which utilizes the cytomegalovirus promoter and the SV40 polyadenylation signal and results in high levels of expression in mammalian cells. When overexpressed in HEK293 and CHO cells this construct produced a 66-kDa protein which was detected in multiple Shc immunoblots (GenBank accession number U46956).
Quantitative Transient Transfection by Electroporation-We have previously demonstrated that electroporation can be used to express various cDNAs with 85-100% transfection efficiency in CHO cells (46). Briefly, CHO/IR/ER cells were suspended in 500 l of phosphate-buffered saline with a total of 40 g of the empty vector, the CLDN mammalian expression vector containing the 66-kDa Shc cDNA or the CLDN vector containing the 52-kDa Shc cDNA as described previously (45,46). The cells were then electroporated at 320 volts and 960 microfarads and plated in ␣-minimal essential medium containing 10% serum. Cell debris was removed by replacing media with fresh media 12 and 30 h later. Forty-eight h later the transfected cells were serum starved for 6 -8 h and either untreated or stimulated for various times with 20 nM EGF as described in the figure legends.
Alkaline Phosphatase Treatment-Whole cell lysates (0.3 mg of protein) were incubated 1 h at room temperature with 1,500 units of calf alkaline phosphatase (Sigma) in buffer A (20 mM Tris, pH 7.4, 100 mM NaCl, 1 mM dithiothreitol, 1 mM MgCl 2 , 0.1 mM ZnCl 2 , 15 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride) plus 10 ϫ alkaline phosphatase buffer (Boehringer Mannheim) in a final volume of 100 l. The reaction was stopped by heating the samples at 100°C in SDS-polyacrylamide gel electrophoresis sample buffer.

EGF Stimulation Results in a Decreased
Electrophoretic Mobility of the 66-kDa Shc Isoform-To examine whether growth factors can affect the phosphorylation state of the 66-kDa Shc isoform, we determined the effect of EGF in Chinese hamster ovary cells expressing human EGF receptors (Fig. 1). Immunoblotting of detergent whole cell extracts demonstrated the presence of the 46-, 52-, and 66-kDa Shc isoforms (Fig. 1A, lane 1). Although EGF had no significant effect on the amount or mobility of the 46-and 52-kDa Shc isoforms, there was a time-dependent decrease in SDS-polyacrylamide gel electrophoretic mobility of approximately 50% of the total 66-kDa Shc proteins present in these cells (Fig. 1A, lanes 2-6). The decrease in electrophoretic mobility under these conditions is characteristic of post-translational modification such as serine/threonine phosphorylation. As expected, phosphotyrosine immunoblotting following Shc immunoprecipitation demonstrated EGFstimulated tyrosine phosphorylation of the 46-and 52-kDa Shc species which was persistent for up to 30 min (Fig. 1B). In addition to the relatively slow time-dependent decrease in electrophoretic mobility, the 66-kDa Shc isoform was rapidly tyrosine phosphorylated in response to EGF (Fig. 1B, lanes 1-6). The tyrosine phosphorylation of the 66-kDa Shc protein occurred rapidly (maximum within 1 min), whereas the reduction in electrophoretic mobility was somewhat slower and was not maximal until approximately 5 min (compare Fig. 1, A, lanes  1-4 with B, lanes 1-4).
The Decreased Mobility of the 66-kDa Shc Isoform Results from Serine/Threonine Phosphorylation by a MEK-dependent Pathway-To determine whether the EGF-stimulated decrease in the 66-kDa electrophoretic mobility was, in fact, due to phosphorylation, the Triton X-100 soluble whole cell extracts were subjected to alkaline phosphatase treatment ( Fig. 2A). As observed in Fig. 1, EGF treatment for 5 min resulted in a decreased mobility of approximately 50% of the 66-kDa Shc isoform without any effect on the migration of the 46-and 52-kDa Shc species ( Fig. 2A, lanes 1 and 2). Incubation of these cell extracts with calf intestine alkaline phosphatase did not alter the migration of the 46-, 52-, or 66-kDa Shc isoforms from unstimulated cells ( Fig. 2A, compare lanes 1 and 3). However, alkaline phosphatase treatment of extracts from EGF-stimulated cells resulted in a mobility of the 66-kDa Shc isoform identical to that of the unstimulated cells ( Fig. 2A, lanes 3  and 4).
Recent studies have demonstrated that the insulin-stimulated serine phosphorylation of several proteins involved in Ras activation occurs through a Ras-dependent feedback pathway involving the downstream activation of MEK1 and MEK2 (47,48). In this regard, insulin was observed to specifically stimulate the serine phosphorylation of the 66-kDa Shc isoform but not the 52-or 46-kDa Shc species (47). Similarly, the MEKspecific inhibitor PD98059 prevented the EGF-induced phosphorylation and decrease in SDS-polyacrylamide gel mobility of the 66-kDa Shc isoform, as detected in Shc immunoblots following Shc immunoprecipitation (Fig. 2B, lanes 1-4). As expected in both control and PD98059-treated cells, EGF stimulated the tyrosine phosphorylation of the EGF receptor which co-immunoprecipitated with the 66-kDa Shc species as well as the 52-and 46-kDa Shc isoforms (Fig. 2B, lanes [5][6][7][8]. However, in the absence of the MEK inhibitor, EGF induced the tyrosine phosphorylation of the slower migrating form of the 66-kDa Shc protein (Fig. 2B, lanes 5 and 6). In contrast, inhibition of the serine/threonine feedback phosphorylation by pretreatment with PD98059 resulted in the EGF-stimulated tyrosine phosphorylation of the non-shifted 66-kDa Shc protein (Fig. 2B,  lanes 7 and 8). These data are consistent with the time course of tyrosine phosphorylation occurring prior to the reduction in 66-kDa Shc gel electrophoretic mobility ( Fig. 1) and further suggests that the tyrosine phosphorylation of 66-kDa Shc isoform precedes its serine/threonine phosphorylation.

The Tyrosine-phosphorylated EGF Receptor Predominantly
Associates with the Non-phosphorylated 66-kDa Shc Species-It is well accepted that the tyrosine phosphorylation of Shc generates a consensus binding site for the SH2 domain of the small adapter protein Grb2 (22,23,(25)(26)(27)(28)(29). To assess the association of the 66-kDa Shc protein with Grb2, we next examined the effect of EGF treatment on the co-immunoprecipitation of the Shc proteins with Grb2 (Fig. 3). Immunoprecipitation of cell extracts with a Shc antibody, followed by Shc immunoblotting demonstrated the presence of the 66-and 52-kDa Shc protein isoforms (Fig. 3A, lane 1). In these experiments, the 46-kDa Shc isoform could not be detected in the Shc immunoprecipitates due to the presence of the IgG heavy chain. As previously observed, approximately 50% of the immunoprecipitated 66-kDa Shc isoform displayed a reduction in gel electrophoretic mobility following EGF stimulation for 5 or 30 min (Fig. 3A, lanes 2 and 3). As expected, the Shc proteins were poorly co-immunoprecipitated with a Grb2 antibody from unstimulated cells due to the absence of Shc tyrosine phosphorylation (Fig. 3B, lane 1). In contrast, EGF stimulation resulted in co-immunoprecipitation of both the 52-and 66-kDa Shc isoforms with the Grb2 antibody (Fig. 3B, lanes 2 and 3). However, only the 66-kDa Shc isoform having reduced mobility was found to associate with Grb2, whereas there was essentially no detectable co-immunoprecipitation with the faster migrating 66-kDa species. As a control, the amount of Grb2 protein immunoprecipitated with the Grb2 antibody was similar from the extracts of unstimulated and EGF-stimulated cells (Fig. 3B, lanes 4 -6). These data support the conclusion that only the 66-kDa Shc species with reduced mobility is tyrosine phosphorylated by the EGF receptor and therefore can associate with Grb2.
Recent studies have demonstrated that the PTB domains of the Shc proteins are the major determinants for association with the tyrosine-phosphorylated EGF receptor (14 -16). To determine the relative association of the shifted and nonshifted forms of the 66-kDa Shc proteins with the EGF receptor, we next examined the co-immunoprecipitation of Shc with the EGF receptor (Fig. 3C). In unstimulated cells, immunoprecipitation of the EGF receptor resulted in a small amount of co-immunoprecipitated 66-kDa Shc protein (Fig. 3C, lane 1). Following EGF stimulation there was a marked increase in the amount of 66-kDa Shc associated with the EGF receptor   1-3). B, the extracts were immunoprecipitated with the Grb2 antibody and immunoblotted for Shc (lanes 1-3) or Grb2 (lanes 4 -6). C, the extracts were immunoprecipitated with the EGF receptor antibody and immunoblotted for Shc (lanes 1-3) or EGF receptor (lanes 4 -6).
( Fig. 3C, lanes 2 and 3). However, only the faster migrating, non-phosphorylated form of the 66-kDa Shc protein was detected ( Fig. 3C, lanes 2 and 3). Due to the large amount of nonspecific signal produced by the EGF receptor antibody heavy chain, both the 46-and 52-kDa Shc isoforms were obscured in these immunoblots. To ensure equal immunoprecipitation of the EGF receptor, the EGF receptor immunoprecipitates were also subjected to immunoblotting with the EGF receptor antibody (Fig. 3C, lanes 4 -6). Thus, although the PTB domain directs the association of Shc with the EGF receptor, the EGF-stimulated serine/threonine phosphorylation of the 66-kDa Shc isoform appears to prevent this interaction.
Isolation of the 66-kDa Shc cDNA-The cDNA cloning of the 46-and 52-kDa Shc isoforms demonstrated that these species arise from alternative translational start site usage from the same transcript (13). In addition, since antibodies directed to the smaller Shc proteins cross-react with the 66-kDa Shc isoform, it has been generally thought that the 66-kDa Shc species results from an alternative spliced primary Shc transcript. Therefore, to isolate a cDNA for the 66-kDa Shc isoform we screened a mouse embryo library with a DNA probe derived from the murine Shc PTB domain. Consistent with the 46-, 52-, and 66-kDa Shc species arising from alternative splicing, the entire open reading frame of the murine cDNA encoding for the 52-kDa Shc isoform is encompassed within the cDNA for the 66-kDa Shc transcript (Fig. 4).
During the preparation of this manuscript, the cDNA for the human 66-kDa Shc protein was reported (49). At the protein level the human 66-kDa Shc is 91% identical and 94% similar to the murine sequence (Fig. 4). Within the 110 amino-terminal extension, these proteins are somewhat more divergent with the human amino acid sequence having only 74% identity and 76% similarity with the murine sequence. Nevertheless, out of the 14 potential serine/threonine phosphorylation sites present within the murine amino-terminal extension, 12 are conserved in the human 66-kDa Shc protein.
The 66-and 52-kDa Shc Isoforms Compete for a Limited Pool of Grb2 Molecules-Having isolated the cDNA encoding the 66-kDa Shc isoform, we next examined the effect of increased Shc protein expression on Grb2 association (Fig. 5). This was accomplished by transfection of cells with an empty vector and with mammalian expression plasmids encoding the 52-or 66-kDa Shc isoforms. As previously observed, in mock-transfected cells EGF stimulation resulted in a decreased SDS-gel electrophoretic mobility of approximately 50% of the endogenous 66-kDa Shc protein (Fig. 5, lanes 1-3). Transfection with the cDNA encoding the 52-kDa Shc isoform resulted in markedly increased expression of both the 52-and 46-kDa Shc species (Fig.  5, lanes 4 -6), consistent with the usage of two translation initiation sites within a single transcript encoding for the 52-kDa Shc isoform (13). It is important to note that the relative increase in expression of the 52-and 46-kDa Shc isoforms is under-represented in these immunoblots since 30-fold more cell extract protein was loaded from the mock-transfected cells (Fig. 5, lanes 1-3) compared with either the 52-or 66-kDa Shc transfected cells (Fig. 5, lanes 4 -9). Similarly, the apparent absence of the 66-kDa Shc protein in the cells overexpressing the 52-and 46-kDa Shc isoforms is due to the differences in relatively protein loading. However, longer exposures and/or loading of greater amounts of cell extract protein demonstrated that the increased expression of the 52-and 46-kDa Shc isoforms did not affect either the levels of the endogenous 66-kDa Shc protein or the reduction in the electrophoretic mobility that occurs following EGF stimulation (data not shown). Likewise, transfection with the cDNA encoding the 66-kDa Shc protein resulted in markedly increased expression of the 66-kDa Shc isoform, without any significant change in the levels of the 52or 46-kDa Shc species (Fig. 5, lane 7). The trace amount of lower molecular mass proteins reflect degradation products of the 66-kDa Shc isoform as they do not co-migrate with either the 52-or 46-kDa Shc species. Furthermore, the overexpressed 66-kDa Shc protein displayed an EGF-stimulated decrease in electrophoretic mobility (Fig. 5, lanes 8 and 9). These data demonstrate that the expressed 66-kDa Shc isoform undergoes similar serine/threonine phosphorylation to that of the endogenous 66-kDa Shc isoform and that the 66-kDa Shc transcript is not translated to any appreciable extent into the 52-or 46-kDa Shc proteins.
Consistent with the tyrosine phosphorylation of Shc generating a docking site for Grb2, EGF stimulation of mock-transfected cells resulted in the co-immunoprecipitation of the 66and 52-kDa Shc isoforms with Grb2 (Fig. 6A, lanes 1-3). The 46-kDa Shc isoform was not apparent due to the presence of the IgG heavy chain from the Grb2 antibody. Increased expression of the 52-kDa Shc isoform resulted in a greater amount of the 52-kDa Shc protein that was co-immunoprecipitated with Grb2 (Fig. 6A, lanes 4 -6). However, compared with the mock-transfected cells, there was a marked decrease in the amount of 66-kDa Shc isoform that was co-immunoprecipitated with Grb2. Similarly, increased expression of the 66-kDa Shc protein resulted in a dramatic increase in the amount of this isoform that was co-immunoprecipitated with Grb2 (Fig. 6A,  lanes 7-9). Furthermore, there also was a marked reduction in the amount of 52-kDa Shc that was co-immunoprecipitated with Grb2 compared with the mock-transfected cells. As controls for equal immunoprecipitation of Grb2, the Grb2 immunoprecipitates were also immunoblotted with the Grb2 antibody (Fig. 6B, lanes 1-9). These data indicate that the 52-and 66-kDa Shc proteins compete for a limited pool of cellular Grb2 proteins.
Since serine/threonine phosphorylation of the 66-kDa Shc protein appears to reduce its ability to associate with the EGF receptor, we also compared the effect of increased Shc protein expression on Grb2 binding to the EGF receptor (Fig. 7A). In mock-transfected cells, Grb2 immunoprecipitation resulted in the co-immunoprecipitation of the EGF receptor which increased from 5 to 30 min of EGF treatment (Fig. 7A, lanes 1-3). There was no significant difference in the amount or time dependence between the amount of EGF receptor co-immuno-precipitated with Grb2 in cells overexpressing the 52-kDa Shc isoform (Fig. 7A, lanes 4 -6). However, increased expression of the 66-kDa Shc isoform reduced the amount of EGF receptor that was co-immunoprecipitated with Grb2 at both 5 and 30 min (Fig. 7A, lanes 7-9). To assure equal Grb2 immunoprecipitation, the Grb2 immunoprecipitates were also immunoblotted for Grb2 (Fig. 7B, lanes 1-9). These data further demonstrate that the 66-kDa Shc isoform reduces the extent of Grb2 association with the EGF receptor.
The reduction in the amount of Grb2 that was associated with the EGF receptor could have arisen from direct competition for Grb2 between the EGF receptor and the 66-kDa Shc isoform and/or between the 52-and 66-kDa Shc proteins for the EGF receptor. To address this issue, we determined the amount of the EGF receptor that was co-immunoprecipitated with Shc (Fig. 8A). Although the EGF receptor was clearly co-immunoprecipitated with Shc in mock-transfected cells, it was only barely detectable at the level of exposure necessary to compare the extent of EGF receptor association with the 52and 66-kDa Shc isoforms (Fig. 8A, lanes 1-3). Increased expres-  1-3), the vector encoding for the 52-kDa Shc (lanes 4 -6), or the vector encoding for the 66-kDa Shc isoform (lanes 7-9) as described under "Experimental Procedures." The cells were then either left untreated (lanes 1, 4, and 7) or incubated with 20 nM EGF for 5 (lanes 2, 5, and 8) and 30 (lanes 3, 6, and 9) min at 37°C. Detergent-soluble cell extracts were prepared and 30 g of the cell extracts from the mock transfected cells (Vector) and 1 g from the 52-kDa Shc (52 Shc) and 66-kDa Shc (66 Shc) transfected cells were subjected to immunoblotting with the Shc antibody.
FIG. 6. Increased expression of the 66-kDa Shc protein competes with 52-kDa Shc for association with Grb2. Cells were quantitatively electroporated with the empty vector (lanes 1-3), the vector encoding for the 52-kDa Shc isoform (lanes 4 -6), or the vector encoding for the 66-kDa Shc isoform (lanes 7-9) as described under "Experimental Procedures." The cells were then either left untreated (lanes 1, 4, and 7) or incubated with 20 nM EGF for 5 (lanes 2, 5, and 8) and 30 (lanes 3, 6, and 9) min at 37°C. Detergent-soluble cell extracts were prepared and immunoprecipitated with a Grb2 antibody. The resultant immunoprecipitates were then immunoblotted with an (A) a Shc antibody or (B) a Grb2 antibody.  1-3), the vector encoding the 52-kDa Shc isoform (lanes 4 -6), or the vector encoding the 66-kDa Shc isoform (lanes 7-9) as described under "Experimental Procedures." The cells were then either left untreated (lanes 1, 4, and 7) or incubated with 20 nM EGF for 5 (lanes 2, 5, and 8) and 30 (lanes 3, 6, and 9) min at 37°C. Detergent-soluble cell extracts were prepared and immunoprecipitated with a Grb2 antibody. The resultant immunoprecipitates were then immunoblotted with an (A) EGF receptor antibody or (B) a Grb2 antibody. sion of either the 52-or 66-kDa Shc isoforms resulted in a substantial increase in the extent of EGF receptor that was co-immunoprecipitated compared with the mock-transfected cells (Fig. 8A, lanes 4 -9). In unstimulated cells expressing the 52-kDa Shc isoform, a greater amount of EGF receptor was co-immunoprecipitated than observed in cells overexpressing the 66-kDa Shc isoform (Fig. 8A, lanes 4 and 7). However, following 5 min of EGF stimulation this difference was not as marked and by 30 min there was no significant difference in the amount of EGF receptor protein that co-immunoprecipitated with Shc (Fig. 8A, compare lanes 5 and 6 with lanes 8 and  9). It should also be noted that the basal level of EGF receptor that was co-immunoprecipitated in the cells overexpressing Shc (Fig. 8A, lanes 4 and 7) was only apparent at the high exposure levels necessary to detect the association between the endogenous Shc proteins with the EGF receptor (Fig. 8A,  lane 3).
In any case, overexpression of the 52-kDa Shc isoform also resulted in an increased amount of Grb2 that was co-immunoprecipitated with Shc (Fig. 8B, compare lanes 1-3 with lanes  4 -6). In addition, there was no significant difference in the amount of Grb2 that was co-immunoprecipitated with the total pool of Shc proteins in cells overexpressing either the 52 versus the 66-kDa Shc isoforms (Fig. 8B, compare lanes 4 -6 with  lanes 7-9). The amount of Shc proteins immunoprecipitated under each transfection condition was similar, although due to the high level of 52-and 66-kDa Shc protein levels the amount of endogenous Shc proteins were difficult to visualize at this exposure level (Fig. 8C, lanes 1-9). Together, these data directly demonstrate that increased expression of the 52-and 66-kDa Shc isoforms results in a redistribution of the available Grb2 protein pool without affecting the amount of Grb2 bound to the total combination of Shc proteins. Furthermore, increased expression of the 66-kDa Shc species not only reduces the amount of Grb2 bound to the 52-kDa Shc isoform, but thereby sequesters Grb2 from the EGF receptor.
Increased Expression of the 66-kDa Shc Isoforms Inhibits EGF Stimulation of ERK-Several lines of evidence have indicated that the Shc⅐Grb2⅐SOS complex provides a major route for the activation of the Ras/Raf/MEK/ERK pathway (29,(31)(32)(33)(34)(35)(36)(37)(38)(39). Since the 66-kDa Shc isoform appears to compete with the 52-kDa Shc species for Grb2 binding, we next assessed the potential effect of 66-kDa expression on ERK activation (Fig.  9). In empty vector-transfected cells, EGF treatment resulted in a time-dependent decrease in SDS-polyacrylamide gel electrophoretic mobility of ERK2 and ERK1, characteristic of ERK activation (Fig. 9A, lanes 1-6). The maximal extent of gel shift was detected within 1 min of EGF stimulation and only slightly diminished over the following 30 min. In contrast, in cells overexpressing the 66-kDa Shc isoform, the amount of gelshifted ERK was decreased at early time points following EGF stimulation (Fig. 9A, lanes 7-10). More dramatically, at longer time points (15 and 30 min of EGF treatment), there was almost a complete recovery of ERK back to the basal state (Fig.  9A, lanes 11 and 12). Similarly, immunoblotting with an antibody that only recognizes the phosphorylated, active form of ERK, confirmed that increased expression of the 66-kDa Shc isoform markedly reduced the duration of ERK activation (Fig.  9B, lanes 1-12). This inhibitory function of 66-kDa Shc protein was isoform specific as increased expression of the 52-and 46-kDa Shc species had no significant effect on ERK activation compared with empty vector-transfected cells (Fig. 10). Thus, the increased expression of the 66-kDa Shc isoform appears to primarily accelerate the inactivation rate of ERK following EGF stimulation.  (lanes 1-3), the vector encoding the 52-kDa Shc isoform (lanes 4 -6), or the vector encoding the 66-kDa Shc isoform (lanes 7-9) as described under "Experimental Procedures." The cells were then either left untreated (lanes 1, 4, and 7) or incubated with 20 nM EGF for 5 (lanes 2, 5, and 8) and 30 (lanes 3, 6, and 9) min at 37°C. Detergent-soluble cell extracts were prepared and immunoprecipitated with a Shc antibody. The resultant immunoprecipitates were then immunoblotted with an (A) EGF receptor, (B) a Grb2 antibody, or (C) a Shc antibody.

DISCUSSION
The Shc family of adapter proteins appear to play an important role in mediating proximal receptor tyrosine kinase signals to the distal effector Ras. This is generally thought to occur following the tyrosine phosphorylation of the Shc proteins, which then engage the SH2 domain of Grb2. Since the SH3 domains of Grb2 can associate with the Ras guanylnucleotide exchange factor SOS, a ternary Shc⅐Grb2⅐SOS complex is formed (29,(31)(32)(33)(34)(35)(36)(37)(38)(39). In addition to binding the Grb2⅐SOS complex, Shc contains an amino-proximal PTB domain, which plays a major role in its association with the activated EGF receptor (14 -16). The association of the Shc⅐Grb2⅐SOS complex with the tyrosine-phosphorylated EGF receptor can then activate and/or target the SOS protein by bringing it in close proximity to the plasma membrane location of Ras, thereby effecting guanylnucleotide exchange of GTP for the bound GDP.
The Shc cDNA clone was originally isolated based upon its homology with the SH2 domain of the human c-fes gene (13). More recently, two related murine Shc genes (ShcB and ShcC) have been identified (50). Although the function of the ShcB and ShcC protein products has not been fully established, ShcC is predominantly expressed in the brain and can associated with tyrosine-phosphorylated proteins through both its PTB and SH2 domains. In contrast, the original Shc gene, referred to as ShcA, has been more extensively characterized. The ShcA gene codes for three related mRNAs (2.8, 3.4, and 3.8 kilobases) that translate into protein products of 46, 52, and 66 kDa. The 46-and 52-kDa isoforms result from the use of two alternative translation initiation sites within the single 3.4-kilobase transcript. Since antibodies directed against the carboxyl-terminal region of the 46/52-kDa Shc proteins cross-react with a 66-kDa species, this isoform was predicted to be an alternative splice variant from at least one of the other two transcripts. In confirmation of this hypothesis, our isolated cDNA for the murine 66-kDa Shc isoform contains an identical nucleotide sequence within the open reading frame of the 52-kDa cDNA. However, the 66-kDa Shc sequence has an additional 5Ј 330 nucleotides which encode a 110-amino acid amino-terminal extension.
During our recent studies on insulin regulation of the Ras activation/inactivation cycle, we observed that insulin preferentially stimulates the tyrosine phosphorylation of the 52-kDa Shc isoform and to a lesser extent the 46-kDa Shc isoform, but does not appreciably increase the tyrosine phosphorylation of the 66-kDa Shc species (47). In contrast, EGF was found to effectively induce the tyrosine phosphorylation of all three Shc isoforms as well as inducing the serine/threonine phosphorylation of the 66-kDa Shc protein (45). This difference led us to speculate that the 66-kDa Shc isoform may play a unique functional role in insulin versus EGF-stimulated cells. Although we have not yet been able to identify a signaling function for the 66-kDa Shc isoform in insulin-stimulated cells, expression of the 66-kDa Shc protein appears to both inhibit the EGF activation of ERK and markedly accelerate the rate of ERK inactivation.
While this manuscript was in preparation, Migliaccio et al. (49) cloned the human 66-kDa ShcA cDNA and examined its expression in HeLa and Cos-1 cells. However, our data are somewhat in disagreement with these findings. First, Migliaccio et al. (49) reported that increased expression of the 66-kDa Shc protein resulted in an increased level of the 52-and 46-kDa Shc isoforms. This was interpreted as evidence for internal translational initiation from the 66-kDa Shc mRNA. Under our experimental conditions, increased expression of the 66-kDa Shc protein did not result in any appreciable increase in the amount of the 52-or 46-kDa Shc isoforms. Although a trace amount of lower molecular weight bands were detectable, these are probably proteolytic fragments as they do not co-migrate with either the 52-or 46-kDa Shc bands (Fig. 5). Second, it was reported that increased expression of the 66-kDa Shc had no effect on EGF-stimulated ERK activity (49). However, ERK activation was only assessed at 5 min following EGF treatment, a time point in which we detect a relatively small inhibition of ERK activation. In contrast, we have observed that the major effect of 66-kDa Shc is to accelerate the inactivation of ERK which is only apparent when assayed at later times following EGF stimulation (Fig. 8). Whether these discrepant findings result from differences between the human and murine 66-kDa Shc proteins, cell contexts used to analyze Shc function and/or the experimental design of these studies remains to be determined.
Nevertheless, we have extended these data to provide a possible molecular mechanism by which the 66-kDa Shc isoform can limit ERK activation. In this system, EGF induces the tyrosine phosphorylation of all three Shc isoforms but only stimulates the serine/threonine phosphorylation of the 66-kDa species. Presumably this is due to the presence of several serine and threonine phosphorylation sites that are located adjacent to several proline residues within the 110-amino acid extension. The time course of the 66-kDa Shc protein gel shift versus phosphotyrosine immunoblotting indicates that the serine/ threonine phosphorylation occurs subsequent to tyrosine phosphorylation. Consistent with these data, inhibition of MEK activity prevents the reduction in gel electrophoretic mobility but does not affect tyrosine phosphorylation of the 66-kDa Shc isoform. Together, these data also suggest that only the tyrosine-phosphorylated 66-kDa Shc proteins become serine/ threonine phosphorylated.
The serine/threonine phosphorylation of the 66-kDa Shc protein appears to regulate its association with the tyrosine-phosphorylated EGF receptor. That is, following EGF stimulation approximately 50% of the 66-kDa Shc species were both tyrosine and serine/threonine phosphorylated. However, immunoprecipitation of the EGF receptor only demonstrated the coimmunoprecipitation of the non-gel shifted 66-kDa Shc isoform. In contrast, Grb2 only associated with the tyrosinephosphorylated and gel shifted 66-kDa Shc isoform, supporting the hypothesis that only this species became tyrosine phosphorylated. Furthermore, increased expression of the 46/52-kDa Shc isoforms decreased the extent of Grb2 association with  1-6) or the vector encoding the 52-kDa Shc isoform (lanes 7-12) as described under "Experimental Procedures." The cells were then either left untreated (lanes 1 and 7) or incubated with 20 nM EGF for 1 (lanes 2 and 8), 3 (lanes 3 and 9), 5 (lanes 4 and 10), 15 (lanes 5 and 11), and 30 (lanes 6 and 12) min at 37°C. Detergentsoluble cell extracts were prepared and immunoblotted with (A) the ERK specific antibody or (B) a conformation specific ERK antibody that only detects the activate ERK species. endogenous 66-kDa Shc proteins. Similarly, increased expression of the 66-kDa Shc species decreased the amount of Grb2 bound to the 46/52-kDa Shc proteins. Thus, it appears that the Shc proteins compete for a limited cellular pool of Grb2 molecules.
At present, the physiological role of the 66-kDa Shc isoform remains unknown. Based upon results from increased expression of the 66-and 52/46-kDa Shc isoforms, we can hypothesize that one potential function of the 66-kDa species is to provide for a feedback inactivation of the Ras signaling pathway. It is well established that the Shc proteins, when tyrosine phosphorylated, associate with the Grb2⅐SOS complex through the Grb2 SH2 domain generating a Shc⅐Grb2⅐SOS ternary complex. The formation of the Shc⅐Grb2⅐SOS complex provides one means for the subsequent association with the tyrosine-phosphorylated EGF receptor, primarily through the Shc PTB domain. The subsequent targeting of SOS to the plasma membrane location of Ras then provides a mechanism for Ras activation which subsequently couples to the activation of Raf/ MEK/ERK pathway. At some point during this cascade, the 66-kDa Shc isoform is serine/threonine phosphorylated in a MEK-dependent manner. At present, the nature of this kinase is unknown but several lines of evidence suggest that it is unlikely to be either ERK or MEK itself (47). In any case, this phosphorylation event uncouples the 66-kDa Shc isoform from the EGF receptor while maintaining its association with the Grb2⅐SOS complex. Since increased expression of the 66-kDa Shc reduces the extent of ERK activation and accelerates its inactivation, we speculate that the 66-kDa Shc isoform sequesters the Grb2⅐SOS complex away from Ras. This could result from an uncoupling of the Shc⅐Grb2⅐SOS complex from the EGF receptor or alternatively, the 66-kDa Shc may direct the Grb2⅐SOS complex to a compartment that is incapable of interacting with the Ras/Raf/MEK/ERK cascade.
Alternatively, Migliaccio et al. (49) reported that expression of the 110-amino acid 66-kDa Shc amino-terminal extension (collagen homology 2 domain) alone was sufficient to inhibit the EGF stimulation of c-fos gene expression. Thus it remains possible that this domain, independent of the PTB, SH2, and or tyrosine-phosphorylation sites of Shc, can interact in some manner with other components of the MAP kinase or signal transducers and activators of transcription signaling pathways. In any case, the 66-kDa Shc protein appears to function, at least when overexpressed, by providing a negative signal for the EGF stimulation of the Ras/Raf/MEK/ERK pathway and c-fos gene expression. Whether this mechanism accounts for the differences in the inactivation of Ras following insulin versus EGF receptor stimulation can now be directly tested.