INTRODUCTION

The insulin and epidermal growth factor (EGF)¹ receptors are members of a family of receptor tyrosine kinases that phosphorylate a variety of intracellular effector proteins (1, 2). One such family of effector proteins, termed Shc (for Src homology 2 (SH2) domain-containing 2 collagen-related), is tyrosine-phosphorylated on Tyr³¹⁷, which generates a recognition site for the SH2 domain of Grb2 (3, 4, 5, 6). Grb2 is a 23-kDa adapter protein containing a single SH2 domain flanked by two Src homology 3 (SH3) domains (7, 8). The amino-terminal Grb2 SH3 domain is the predominant determinant mediating the basal state association with the proline-rich domain of SOS, a 170-kDa guanylnucleotide exchange factor for the p21 GTP binding protein Ras (8, 9, 10, 11, 12, 13). The receptor-mediated tyrosine phosphorylation of Shc and assembly of a Shc-Grb2-SOS ternary complex is generally thought to be the major pathway leading to Ras activation (14, 15).

In addition to this acute activation of Ras (maximal stimulation within 1 min of insulin or EGF treatment), there is a slower recovery of Ras back to the inactive GDP-bound state (16, 17, 18). The rate of Ras inactivation is dependent on several factors including cell type, receptor number, and the particular stimulating growth factor examined. In the case of insulin and EGF treatment, Ras inactivation is generally complete within 10-30 min even in the continuous presence of ligand (16, 17, 18, 19, 20, 21, 22, 23). Recent studies have demonstrated that insulin treatment results in a feedback serine/threonine phosphorylation of SOS, which directly correlates with the dissociation of the Grb2-SOS complex (24, 25). Furthermore, inhibition of this feedback phosphorylation of SOS preserves the interaction between Grb2 and SOS and prolongs the time that Ras remains in the active GTP-bound state (16, 17). Although EGF stimulation also induces a serine/threonine phosphorylation of SOS and a transient activation of Ras, there is no dissociation of the Grb2-SOS complex (19, 26, 27, 28). In contrast, EGF apparently induces a dissociation of the Grb2-SOS complex from tyrosine-phosphorylated Shc (27, 28). Thus, although the disruption of the Shc-Grb2-SOS complex appears to be one mechanism that may be partially responsible for the inactivation phase of Ras following growth factor stimulation, insulin and EGF utilize different molecular pathways to accomplish the uncoupling of the Grb2-SOS activation signal.

However, recent studies have also demonstrated that in addition to Grb2 other small adapter proteins including Nck and CrkII, through their respective SH3 domains, also associate with the proline-rich carboxyl-terminal domain of SOS (29, 30, 31). Furthermore, increased expression of Nck has been observed to enhance Ras-dependent signal transduction (29). Similarly, expression of v-Crk in PC12 cells augments growth factor-stimulated Ras activation and neuronal differentiation (32, 33). In addition, expression of SH2 and SH3 loss of function point mutants inhibits growth factor stimulation of Ras and neurite outgrowth (31, 32, 33, 34). Together these data provide direct evidence suggesting that alternative pathways to Grb2 may also contribute to SOS-mediated Ras activation.

To examine the interactions between Grb2-SOS with that of the Nck-SOS and CrkII-SOS complexes during the Ras inactivation phase following growth factor stimulation, we have determined the effects of insulin and EGF to regulate the association state of SOS with these adapter proteins. The data presented in this study demonstrate that although EGF does not induce the dissociation of either Grb2 or Nck from SOS, EGF is fully capable of dissociating the CrkII-SOS complex. Thus, these results suggest that in addition to the Grb2-SOS complex, the CrkII-SOS complex is involved in the EGF regulation of the Ras activation/inactivation cycle.

EXPERIMENTAL PROCEDURES

CHO cells expressing both the human insulin and EGF receptors were obtained as described previously (35). These cells were maintained in minimal Eagle's medium containing nucleotides plus 10% fetal bovine serum. The cells were grown to confluence and incubated in serum-free medium for 8 h prior to the addition of 100 nM insulin or 20 nM EGF at 37 °C for the times indicated in the figure legends.

Whole cell extracts were prepared by detergent solubilization in a lysis buffer (20 mM Hepes, pH 7.4, 1% Triton X-100, 3 mM MgCl₂, 2 mM EDTA, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 10 µM leupeptin, 10 µg/ml aprotinin, and 1.5 µM pepstatin) for 1 h at 4 °C. Immunoprecipitations were performed by dilution of the detergent-solubilized cell extracts 10-fold in lysis buffer without Triton X-100 and incubation with 4 µg of an SOS polyclonal antibody (Transduction Laboratories), Grb2 polyclonal antibody (Santa Cruz), CrkII polyclonal antibody (Santa Cruz), or Nck monoclonal antibody (Upstate Biotechnology, Inc.) for 2 h at 4 °C. The samples were then incubated with protein A-Sepharose or protein G(+)-Sepharose for 1 h at 4 °C. The resulting immunoprecipitates were subjected to SDS-polyacrylamide gel electrophoresis and Western blotted (enhanced chemiluminescence detection kit, Amersham Corp.) using the SOS, Grb2, CrkII, Nck, or PY20-HRP phosphotyrosine antibody (Transduction Laboratories) as indicated in the individual figure legends.

GST-Nck and GST-CrkII cDNAs were kindly provided by Dr. Lewis T. Williams (University of California, San Francisco, CA) and Dr. Bruce Mayer (The Children's Hospital, Boston, MA), respectively. The GST-Grb2 cDNA was obtained as described previously (24). All three fusion proteins were isolated and bound to glutathione-Sepharose beads (25 µl bed volume). The glutathione-Sepharose-bound GST-fusion proteins were then incubated for 1 h at 4 °C with whole cell lysates (50 mM Hepes, pH 7.8, 1% Triton X-100, 2.5 mM EDTA, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium vanadate, 1 mM phenylmethylsulfonylfluoride, 2 µM pepstatin, 0.5 trypsin inhibitory units of aprotinin, and 10 µM leupeptin) isolated from the CHO/IR/ER cells. The Sepharose beads were pelleted and washed three times with phosphate-buffered saline and solubilized in Laemmli sample buffer. The amount of SOS associated with the immobilized GST-fusions protein was determined by immunoblotting with an SOS antibody. Quantification of signal intensity was performed in the linear range of the film using NIH Image version 1.5 software.

RESULTS

Previous studies have demonstrated that SOS can form complexes with the small adapter proteins Grb2, Nck, and CrkII (8, 9, 10, 11, 12, 13, 29, 30, 31). To estimate the relative binding affinities of SOS for these different adapters, we compared the binding of SOS to GST-Grb2, GST-Nck, and GST-CrkII fusion proteins in vitro (Fig. 1). Incubation of CHO/IR/ER cell extracts with various concentrations (1-100 µg) of GST-Grb2 demonstrated a dose-dependent and saturatable binding of SOS (Fig. 1A, lanes 1-6). Similarly, incubations of cell extracts with different amounts of GST-Nck also resulted in the saturatable binding of SOS (Fig. 1A, lanes 7-12). However, the apparent affinity of SOS for GST-Grb2 was somewhat greater than for GST-Nck, which was particularly evident at the lower concentrations of the fusion proteins (e.g. compare lane 1 with lane 7 and lane 2 with lane 8). In an analogous manner, GST-CrkII was also observed to specifically bind SOS from the CHO/IR/ER cell extracts with a reduced affinity compared with GST-Grb2 (Fig. 1B). Transformation of these data into a saturation binding curve indicated that GST-Grb2 had an approximate 2.5-fold greater binding affinity compared with both GST-Nck and GST-CrkII (data not shown).

It has been well documented that the SH3 domains of Grb2, Nck, and CrkII are responsible for the basal state association with the proline-rich carboxyl-terminal domain of SOS (8, 9, 10, 11, 12, 13, 29, 30, 31, 36). However, in addition to the SH3 domains, these adapter proteins also contain SH2 domains for interaction with various phosphotyrosine docking sites (37). To identify the phosphotyrosine docking proteins that are associated with Grb2 and CrkII in vivo, the CHO/IR/ER cells were incubated in the absence or presence of either 100 nM insulin or 20 nM EGF for 30 min at 37 °C. Whole cell detergent extracts were then prepared and immunoprecipitated with either a Grb2 or a CrkII antibody (Fig. 2). As expected, phosphotyrosine immunoblots of whole cell extracts demonstrated the specific insulin-stimulated tyrosine phosphorylation of IRS1 and the insulin receptor  subunit (Fig. 2, lanes 1 and 2). In contrast, EGF stimulation resulted in the tyrosine phosphorylation of the EGF receptor and an approximately 120-130-kDa protein (Fig. 2, lane 3). Immunoprecipitation of Grb2 followed by phosphotyrosine immunoblotting of extracts from insulin-stimulated cells resulted in the co-immunoprecipitation of IRS1 and the 52-kDa Shc isoform (Fig. 2, lanes 4 and 5). EGF stimulation resulted in the co-immunoprecipitation of Grb2 with the tyrosine-phosphorylated EGF receptor and both the 52- and 46-kDa Shc isoforms (Fig. 2, lane 6). Consistent with these data, the 52-kDa Shc isoform has been reported to be a significantly better substrate for the insulin receptor kinase than the 46-kDa species, whereas the EGF receptor can effectively tyrosine-phosphorylate these two Shc isoforms equally well (35).

In contrast to Grb2, insulin stimulation resulted in a smaller amount of tyrosine-phosphorylated IRS1 associated with CrkII (Fig. 2, lanes 7 and 8). In addition, there was no detectable association of CrkII with tyrosine-phosphorylated Shc proteins. Similarly, EGF stimulation resulted in a lesser extent of CrkII association with the tyrosine-phosphorylated EGF receptor (Fig. 2, lane 9). However, CrkII was predominantly bound to a 120-130-kDa EGF-stimulated tyrosine-phosphorylated protein that is probably either p120 c-Cbl or p130^cas (38, 39). In addition, CrkII itself appeared to undergo a small degree of tyrosine phosphorylation following both insulin and EGF stimulation, consistent with a recent study demonstrating an insulin-like growth factor-1-stimulated CrkII tyrosine phosphorylation (40). We were unable to perform a similar analysis with our Nck antibody, since it was found to interfere with the ability of Nck to interact with other effector proteins. Nevertheless, these data demonstrate that although both Grb2 and CrkII bind to SOS, they apparently adapt SOS to different subsets of tyrosine-phosphorylated proteins in either insulin- or EGF-stimulated cells.

Since Grb2 and CrkII associate with an overlapping subset of tyrosine-phosphorylated proteins, we next examined the potential formation of multisubunit complexes containing both Grb2 and CrkII (Fig. 3). In the basal state, a small amount of Grb2 was detected in the CrkII immunoprecipitate, which was not significantly affected by insulin stimulation (Fig. 3A, lanes 1 and 2). In contrast, EGF stimulation for 30 min resulted in an increased amount of Grb2 that was co-immunoprecipitated with the CrkII antibody (Fig. 3A, lane 3). This was not due to differences in CrkII immunoprecipitation, since equivalent amounts of CrkII protein were present under these conditions (Fig. 3A, lanes 4-6). It should be noted that we typically detect a strong immunoreactivity for the established 40-kDa CrkII protein and a weak signal from a 42-kDa band that cross-reacts with the CrkII antibody. In any case, to confirm these findings we also examined the ability of the Grb2 antibody to co-immunoprecipitate CrkII (Fig. 3B). Although CrkII was also detected in the Grb2 immunoprecipitates from unstimulated cells, again insulin had no effect (Fig. 3B, lanes 1 and 2). However, following EGF stimulation there was a significant increase in the amount of Grb2 immunoprecipitated CrkII proteins (Fig. 3B, lane 3). As controls, equivalent amounts of Grb2 protein were immunoprecipitated by the Grb2 antibody as detected by Grb2 immunoblotting (Fig. 3B, lanes 4-6). These data suggest that multisubunit complexes are formed following EGF stimulation, which simultaneously contain both Grb2 and CrkII.

We and others have recently reported that insulin stimulation results in a time-dependent serine/threonine phosphorylation of SOS concomitant with a dissociation of the Grb2-SOS complex (24, 25). However, although EGF stimulation also results in a serine/threonine phosphorylation of SOS, the Grb2-SOS complex remains stably associated (18, 19, 26, 27, 28). To document this observation under the same experimental conditions, we performed a series of co-immunoprecipitation using cell extracts from control, insulin-, and EGF-stimulated cells (Fig. 4). Immunoblotting of cell extracts with a Grb2 antibody demonstrated identical levels of Grb2 protein in untreated or cells stimulated with 100 nM insulin or 20 nM EGF for 30 min (Fig. 4A, lanes 1-3). As expected, immunoprecipitation of SOS from control cell extracts resulted in the co-immunoprecipitation of Grb2 (Fig. 4A, lane 4), whereas following insulin stimulation there was a significant reduction in the amount of Grb2 protein that was co-immunoprecipitated with the SOS antibody (Fig. 4A, lane 5). In contrast to insulin, EGF stimulation did not significantly decrease the ability of the SOS antibody to co-immunoprecipitate Grb2 (Fig. 4A, lane 6). To ensure that equal amounts of the SOS protein were immunoprecipitated, the SOS immunoprecipitates were subjected to SOS immunoblotting (Fig. 4A, lanes 7-9). It should be noted that the small decrease in mobility of the SOS protein from the insulin and EGF-stimulated cells was characteristic of serine/threonine phosphorylation (24, 25, 26, 27, 28).

To further document the ability of insulin but not EGF to induce the dissociation of the Grb2-SOS complex, a similar analysis was performed using a Grb2 antibody to co-immunoprecipitate the SOS protein (Fig. 4B). SOS immunoblots demonstrated approximately equivalent amounts of SOS protein in the extracts from unstimulated, insulin-stimulated, and EGF-stimulated cells (Fig. 4B, lanes 1-3). Immunoprecipitation with a Grb2 antibody resulted in the co-immunoprecipitation of SOS protein but which was markedly reduced in the extracts from the insulin-stimulated cells (Fig. 4B, lanes 4 and 5). As observed in the SOS immunoprecipitates, EGF stimulation had no significant effect on the ability of the Grb2 antibody to co-immunoprecipitate the SOS protein (Fig. 4B, lane 6). Grb2 immunoblotting of the Grb2 immunoprecipitates was performed to document equivalent efficiencies of Grb2 immunoprecipitation under these conditions (Fig. 4B, lanes 7-9).

Since the adapter protein Nck can also associate with SOS, we next examined the effect of insulin and EGF treatment on Nck-SOS interactions (Fig. 5). Nck immunoblotting of extracts from control, insulin-stimulated, or EGF-stimulated cells demonstrated the presence of a single band in corresponding to the 47-kDa Nck protein (Fig. 5, lanes 1-3). Immunoprecipitation of SOS from unstimulated cell extracts resulted in the co-immunoprecipitation of Nck, which was reduced in extracts from insulin-stimulated cells (Fig. 5, lanes 4 and 5). Similar to Grb2-SOS interactions, EGF stimulation had no significant effect on the co-immunoprecipitation of Nck with the SOS antibody (Fig. 5, lane 6). The band above the Nck protein in lanes 4-6 corresponds to the heavy chain of the SOS antibody, which cross-reacts with the Nck antibody in these immunoprecipitates. SOS immunoblotting of the SOS immunoprecipitates demonstrated that the insulin-stimulated decrease in Nck association was not due to differential immunoprecipitation or expression of the SOS protein (Fig. 5, lanes 7-9). It should also be noted that we were unable to co-immunoprecipitate SOS with the available Nck antibody, since this antibody was directed to the Nck SH3 domain responsible for SOS binding (data not shown). Nevertheless, these data demonstrated that insulin induces the dissociation of both the Grb2-SOS and Nck-SOS complexes, whereas EGF stimulation does not have any significant effect on their association state.

Our in vitro binding results indicated that CrkII has a similar affinity for SOS as Nck (Fig. 1). Therefore, we next examined the effect of insulin and EGF on the association state of CrkII-SOS by co-immunoprecipitation (Fig. 6). CrkII immunoblotting of CHO/IR/ER cell extracts demonstrated the presence of the major 40-kDa CrkII protein (Fig. 6A, lane 1). As expected, 30 min of insulin or EGF treatment had no effect on the levels of expressed CrkII protein (Fig. 6A, lanes 2 and 3). As observed for Grb2 and Nck, immunoprecipitation of SOS resulted in the co-immunoprecipitation of CrkII, which was decreased in extracts isolated from insulin-stimulated cells (Fig. 6A, lanes 4 and 5). However, in contrast to Grb2 and Nck, EGF stimulation resulted in a decreased co-immunoprecipitation of CrkII with the SOS antibody (Fig. 6A, lane 6). This was not due to differential immunoprecipitation of SOS based upon SOS immunoblotting of the same SOS immunoprecipitates (Fig. 6A, lanes 7-9).

To confirm this result, we determined the effect of insulin and EGF on the co-immunoprecipitation of SOS with the CrkII antibody. As controls, SOS immunoblotting demonstrated equal amounts of SOS protein in the extracts isolated from control, insulin-stimulated, and EGF-stimulated cells (Fig. 6B, lanes 1-3). However, CrkII immunoprecipitation of extracts from both insulin- and EGF-stimulated cells resulted in a marked decrease in the amount of co-immunoprecipitated SOS protein compared with the unstimulated cell extracts (Fig. 6B, lanes 4-6). In addition, CrkII immunoblots of the CrkII immunoprecipitates confirmed equivalent immunoprecipitation under these experimental conditions (Fig. 6B, lanes 7-9). Thus, these data demonstrate fundamental differences among the Grb2-SOS, Nck-SOS, and CrkII-SOS complexes (i.e., insulin can induce the dissociation of all three adapter proteins from SOS, whereas EGF has no effect on the Grb2-SOS or Nck-SOS complexes). In contrast, EGF appears to specifically induce only the dissociation of the CrkII-SOS complex.

We and others have recently observed that the time dependence of insulin-stimulated dissociation of the Grb2-SOS correlated with the inactivation of Ras from the GTP-bound to the GDP-bound state (16, 17, 18, 19, 24, 28). However, compared with insulin, EGF stimulation resulted in a more rapid Ras inactivation despite the inability of EGF to uncouple the Grb2-SOS complex (18). We therefore examined the time course of EGF stimulation on the interactions between CrkII-SOS and Grb2-SOS (Fig. 7). As previously observed, immunoprecipitation of SOS resulted in the co-immunoprecipitation of CrkII (Fig. 7A, lane 1). Following 1 min of EGF stimulation, there was a marked decrease in the amount of co-immunoprecipitated CrkII that was completely dissociated from 3-30 min (Fig. 7A, lanes 2-6). As expected, in the same immunoprecipitate there was no significant difference in the amount of Grb2 that was co-immunoprecipitated with SOS (Fig. 7B, lanes 1-6). Similarly, immunoblotting of the SOS immunoprecipitates also demonstrated that the SOS protein was equivalent under all of these conditions (Fig. 7C, lanes 1-6). These data suggest that the interaction between CrkII and SOS may play a more significant role for Ras inactivation following EGF stimulation compared with the Grb2-SOS complex.

DISCUSSION

It is generally accepted that tyrosine kinase receptor activation of Ras results from the substitution of bound GDP with GTP by the guanylnucleotide exchange factor SOS. Although the mechanism by which SOS function is regulated by upstream signaling events has not been completely established, several models have been proposed. In the basal state, the carboxyl-terminal region of SOS is associated with the SH3 domains of Grb2 (10, 41, 42). The initial tyrosine phosphorylation of a receptor (i.e. the EGF receptor) and/or downstream signaling proteins (i.e. IRS1 or Shc) generates a specific docking site for the SH2 domain of Grb2 (5, 8, 42, 43). In this manner a functional ternary complex is generated between a phosphotyrosine effector protein, Grb2 and SOS. Recently, the expression of plasma membrane-targeted forms of SOS has been reported to induce the constitutive activation of Ras (44, 45). Based upon these data it has been postulated that in the basal state the Grb2-SOS complex was sequestered from Ras and that tyrosine phosphorylation of effector proteins such as Shc and/or the EGF receptor itself might function in the appropriate targeting of the Grb2-SOS complex to the plasma membrane location of Ras. The importance of the adapter function of Grb2 in this process has been further suggested by loss of function mutants in Caenorhabditis elegans. In this system, the worm equivalent of Grb2, termed Sem5, when mutated in either its SH2 or SH3 domain fails to generate Ras-dependent signals required for vulval development (46). In contrast, several studies have demonstrated that the carboxyl-terminal domain of SOS functions as an autoinhibitory domain, which may be derepressed by the binding of Grb2. For example, expression of a carboxyl-terminal SOS deletion mutant resulted in the constitutive activation of Ras with a concomitant transformed phenotype (47, 48). In Drosophila, a combination of the catalytic domain with the pleckstrin homology and/or Dbl domain(s) was sufficient to partially restore R7 cell fate development in SOS deletion mutations (49). Although we cannot reconcile these apparent contradictory findings at present, clearly substantial data has been obtained to indicate an important role for Grb2-SOS interactions in regulating the activation state of Ras.

In any case, following growth factor stimulation, Ras is also quickly inactivated and typically returns to the inactive state within 10-60 min depending upon the particular stimulatory growth factor as well as the cellular context (16, 17, 18, 19, 20, 21, 22, 23). Recently, we and others have observed that in addition to the rapid formation of a Shc-Grb2-SOS ternary complex, insulin stimulation also resulted in a time-dependent dissociation of the Grb2-SOS complex (16, 17, 18, 19, 24, 28). The dissociation of the Grb2-SOS complex was found to correlate with the inactivation of Ras from the GTP-bound to the GDP-bound state (16, 17). Furthermore, inhibition of Grb2-SOS dissociation, by preventing SOS phosphorylation, prolonged the time that Ras remained bound to GTP.

Although the uncoupling of the Grb2-SOS complex may account, at least in part, for the inactivation of Ras following insulin stimulation, this does not appear to be the case for EGF. In contrast to insulin, EGF stimulation does not affect the Grb2-SOS, which remains persistently associated despite the serine/threonine phosphorylation of SOS (19, 27, 28). Nevertheless, in the same cell context EGF stimulation also induced a rapid activation/inactivation of Ras, which in fact occurred more quickly than that following insulin stimulation (18). The inactivation of Ras following EGF stimulation has been suggested to result from dissociation of the Grb2-SOS complex from tyrosine-phosphorylated Shc rather than from an uncoupling of Grb2 from SOS (27, 28). Although this remains a potential mechanism for Ras inactivation, this has not been observed in other studies (17, 19). Furthermore, the cellular amount of the Shc-Grb2-SOS ternary complex is relatively low compared with the total Grb2-SOS pool (18, 50).

Recently, several studies have reported that in addition to Grb2, the proline-rich carboxyl-terminal domain of SOS can also direct its association with the SH3 domains of the adapter proteins Nck and CrkII (29, 30, 31, 32, 33, 34). The nck gene was originally isolated from a human melanoma cDNA library and encodes a 47-kDa protein composed of three SH3 domains followed by one SH2 domain (51). On the other hand, CrkII is found as a predominant 40-kDa protein that contains a single SH2 followed by two SH3 domain (52). Despite similarities in the modular structure of these molecules, overexpression of either Nck or CrkII can transform mammalian cells (52, 53, 54). In contrast, Grb2 does not appear to have transforming potential, suggesting different biological functions between these different adapter proteins (7).

To determine the relative binding potential of these adapter proteins to SOS, we compared the in vitro binding of SOS with GST-Grb2, GST-Nck, and GST-CrkII fusion proteins. These data indicated that all three adapters could bind to SOS with a similar K_d, although Grb2 appeared to have a slightly higher affinity for SOS (~2.5-fold) than either Nck or CrkII. Nevertheless, co-immunoprecipitation of SOS with both Nck and CrkII clearly demonstrated that these adapter proteins are also associated with SOS in vivo. Our data also demonstrated that, in addition to the Grb2-SOS complex, insulin stimulation resulted in a dissociation of the Nck-SOS and CrkII-SOS complexes. Surprisingly, however, EGF stimulation resulted in a rapid dissociation of the CrkII-SOS complex without any detectable effect on the association states of Grb2-SOS or Nck-SOS. In fact, the time dependence of CrkII-SOS dissociation was significantly faster than observed for the insulin-stimulated dissociation of Grb2 from SOS and more closely paralleled the EGF-dependent inactivation of Ras (18).

At present the molecular basis for the ability of insulin but not EGF stimulation to induce the dissociation of the Grb2-SOS complex is not known. Recently, we have observed that the persistent plasma membrane targeting of the Grb2-SOS complex to the EGF receptor itself accounts for the inability of EGF to uncouple Grb2 from SOS. This finding was based upon expression of a targeting-defective Grb2 mutant in which the SH2 domain was rendered inactive and by mutations of the tyrosine autophosphorylation acceptor sites of the EGF receptor responsible for directing its interaction with the Grb2-SOS complex (28). In this regard, the data presented in this study demonstrate that the major tyrosine-phosphorylated protein that associates with Grb2 following EGF stimulation was the EGF receptor itself. This is consistent with our previous observation that EGF primarily stimulates the association of the Grb2-SOS complex with the tyrosine-phosphorylated EGF receptor. However, in EGF-stimulated cells the predominant tyrosine-phosphorylated CrkII binding protein is a 120-130-kDa species that is most likely either p120 c-Cbl or p130^cas (38, 39). In either case, the EGF receptor itself does not appear to be the major CrkII-SOS binding protein. Thus, in contrast to the association of the Grb2-SOS complex with the EGF receptor, interaction of the CrkII-SOS complex with either p120 c-Cbl or p130^cas allows for a the dissociation of the complex following growth factor stimulation.

In summary, the data presented in this study demonstrate that insulin stimulation induces a dissociation of the Grb2-SOS, Nck-SOS, and CrkII-SOS complexes consistent with a role in the return of Ras to the basal GDP-bound state following the initial phase of Ras activation. Although EGF stimulation does not affect the association state of either the Grb2-SOS or the Nck-SOS complex, there is a rapid time-dependent dissociation of the CrkII-SOS complex. Thus, the interaction between CrkII and SOS probably plays a more significant role in the inactivation of Ras following EGF stimulation than does the Grb2-SOS complex.