Dominant-negative Mutants of Grb2 Induced Reversal of the Transformed Phenotypes Caused by the Point Mutation-activated Rat HER-2/Neu *

To clarify the role of the Shc-Grb2-Sos trimer in the oncogenic signaling of the point mutation-activated HER-2/neu receptor tyrosine kinase (named p185), we interfered with the protein-protein interactions in the Shc (cid:122) Grb2 (cid:122) Sos complex by introducing Grb2 mutants with deletions in either amino- ( (cid:68) N-Grb2) or carboxyl( (cid:68) C-Grb2) terminal SH3 domains into B104-1-1 cells derived from NIH3T3 cells expressing the point mutation- activated HER-2/neu . We found that the transformed phenotypes of the B104-1-1 cells were largely reversed by the (cid:68) N-Grb2. The effect of the (cid:68) C-Grb2 was much weaker. Biochemical analysis showed that the (cid:68) N-Grb2 was able to associate Shc but not p185 or Sos, while the (cid:68) C-Grb2 bound to Shc, p185, and Sos. The p185-mediated Ras activation was severely inhibited by the (cid:68) N-Grb2 but not the (cid:68) C-Grb2. Taken together, these data demon-strate that interruption of the interaction between Shc and the endogenous Grb2 by the (cid:68) N-Grb2 impairs the oncogenic signaling of the activated p185, indicating that (i) the (cid:68)

To clarify the role of the Shc-Grb2-Sos trimer in the oncogenic signaling of the point mutation-activated HER-2/neu receptor tyrosine kinase (named p185), we interfered with the protein-protein interactions in the Shc⅐Grb2⅐Sos complex by introducing Grb2 mutants with deletions in either amino-(⌬N-Grb2) or carboxyl-(⌬C-Grb2) terminal SH3 domains into B104-1-1 cells derived from NIH3T3 cells expressing the point mutationactivated HER-2/neu. We found that the transformed phenotypes of the B104-1-1 cells were largely reversed by the ⌬N-Grb2. The effect of the ⌬C-Grb2 was much weaker. Biochemical analysis showed that the ⌬N-Grb2 was able to associate Shc but not p185 or Sos, while the ⌬C-Grb2 bound to Shc, p185, and Sos. The p185-mediated Ras activation was severely inhibited by the ⌬N-Grb2 but not the ⌬C-Grb2. Taken together, these data demonstrate that interruption of the interaction between Shc and the endogenous Grb2 by the ⌬N-Grb2 impairs the oncogenic signaling of the activated p185, indicating that (i) the ⌬N-Grb2 functions as a strong dominantnegative mutant, and (ii) Shc/Grb2/Sos pathway plays a major role in mediating the oncogenic signal of the activated p185. Unlike the ⌬N-Grb2, ⌬C-Grb2 appears to be a relatively weak dominant-negative mutant, probably due to its ability to largely fulfill the biological functions of the wild-type Grb2.
Activation of Ras is an important convergence point in the mitogenic signaling pathway of receptor tyrosine kinases (27).
A key upstream pathway leading to Ras activation by receptor tyrosine kinases has recently been established, primarily as a result of studies with the receptors for EGF, platelet-derived growth factor, and insulin (28 -33). The most important components of this pathway include Shc, Grb2, and Sos. Shc stands for SH2 domain-containing ␣2 collagen-related proteins. The Shc family consists of three isoforms (34). The p46 Shc and p52 Shc isoforms come from the same transcript with different translation initiation sites. The p66 Shc species most likely arises from a distinct transcript. Tyrosine phosphorylation of Shc provides a docking site for Grb2 which was originally identified as a growth factor receptor-bound protein (35), a mammalian homolog of Caenorhabditis elegans Sem-5 and Drosophila Drk (36,37). Grb2 is a 24-kDa adaptor protein containing an SH2 domain flanked by two SH3 domains. Through the SH3 domains, Grb2 constitutively associates with Sos (named for the Son of Sevenless gene), a 150-kDa guanylnucleotide exchange factor for Ras (38 -41), by targeting the proline-rich motif at its carboxyl terminus. Upon ligand stimulation, most receptor tyrosine kinases examined to date have been able to induce tyrosine phosphorylation of Shc, which subsequently binds to the SH2 domain of Grb2. The formation of the Shc⅐Grb2⅐Sos ternary complex has been proposed to play an important role in activating Ras (28 -33). Alternatively, the Grb2⅐Sos complex can be directly recruited to the activated EGF receptor (43). Activation of Ras leads to stimulation of downstream kinase cascades, which at least include Raf-1/ MEK/MAPK and MEKK-1/JNKK/JNK pathways (44).
Unlike EGF receptor and other receptor tyrosine kinases, the mutation-activated p185 tyrosine kinase is constitutively active in the absence of exogenously added ligand. Although activation of Ras has been proposed to play an important role in the oncogenic signaling of the mutation-activated p185 (45), coupling of p185 to Ras via Shc⅐Grb2⅐Sos or Grb2⅐Sos or both has not been yet determined. Our recent data and that of others indicated that tyrosine phosphorylation of Shc and formation of the Shc⅐Grb2 complex occurred in transformed NIH3T3 cells that express the mutation-activated p185 and human breast cancer cells that overexpress p185, which suggests that the Shc/Grb2/Sos/Ras pathway may be responsible for transmit-ting the oncogenic signal from the activated p185 (46,55). One way to provide more direct evidence to support this idea is to interfere with the protein-protein interactions involved in this pathway by using dominant-negative mutants and then examine whether these mutants can reverse the transformed phenotypes caused by the activated p185. Grb2 is a central component in this pathway. On the one hand, it binds tightly to Sos through its SH3 domains; on the other hand, it can bind to Shc and probably the activated p185 as well through its SH2 domain. Interestingly, recent studies suggested that the SH2 and SH3 domains of Grb2 functioned independently (47,48). Binding of phosphopeptides to the SH2 domain of Grb2 does not appreciably affect the association of its SH3 domains with proline-rich peptides. Conversely, binding of excessive peptides derived from Sos to Grb2 does not influence the interaction between its SH2 domain with phosphopeptides. We, therefore, reasoned that deleting one of the SH3 domains of Grb2 might create dominant-negative mutants which compete with the endogenous Grb2⅐Sos complex for Shc or activated p185 and block the oncogenic signaling pathway of the activated p185. To test this hypothesis we transfected either an amino-terminal or an carboxyl-terminal SH3 domain deletion mutant of Grb2 into B104-1-1 cells which are transformed NIH3T3 cells expressing the mutation-activated p185. We found that the transformed phenotypes of B104-1-1 were largely reversed by the aminoterminal SH3 domain deletion mutant of Grb2 (⌬N-Grb2). The effect of the carboxyl-terminal SH3 domain deletion mutant (⌬C-Grb2) on phenotypic reversion was much weaker. Biochemical analysis data indicated that the ⌬N-Grb2 functioned as a strong dominant-negative mutant, whereas the ⌬C-Grb2 seemed to be a weak one. These results support the notion that the Shc/Grb2/Sos pathway plays an important role in the oncogenic signaling of the mutation-activated p185.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture-B104-1-1 cells are transformed NIH3T3 cells generated by transfection with the neu oncogene, originally derived from a neuro/glioblastoma cell line (49). Cells were grown in Dulbecco's modified Eagle's medium/Ham's F-12 medium supplemented with 10% calf serum under an atmosphere of 5% CO 2 at 37°C, unless otherwise indicated. Cells transfected with pSV2neo in focus forming assays as indicated below were grown under the same conditions except that G418 (500 g/ml) was added.
Plasmids-The cosmid clone cNeu-104 carries the mutation-activated rat genomic neu (3). ⌬N-Grb2 is an amino-terminal SH3 domain deletion mutant of Grb2. ⌬C-Grb2 is a carboxyl-terminal SH3 domain deletion mutant of Grb2. The cDNAs encoding the ⌬N-Grb2 and ⌬C-Grb2 are driven by the cytomegalovirus promoter and preceded by the hemagglutinin epitope (HA1) tag in expression vector pCGN-Bam, which contains the hygromycin-resistant gene as a selection marker (42,57).
Focus Forming Assays-Focus forming assays were performed as described previously with some modifications (51). cNeu-104 (1 g) was cotransfected into NIH3T3 cells with pSV2neo (0.1 g) and 10 g of plasmids encoding vector alone, ⌬C-Grb2, or ⌬N-Grb2. The filler plasmid, pGEM, was used to ensure that equal amounts of DNA were transfected into cells. Two days after transfection, cells were split 1:4, and duplicate plates were cultured in regular medium for 3-4 weeks while the other set of duplicates were grown in medium containing G418. Foci and G418-resistant colonies were stained in crystal violet solution (1% crystal violet, 20% ethanol in H 2 O). The resulting number of foci from each transfection was corrected for transfection efficiency by dividing by the number of G418-resistant colonies created by the same transfection. Results are expressed as percentage of foci in control transfection with cNeu-104 (100%). Shown here is the average of three individual experiments. Standard deviation is shown by an error bar.
Microfocus Forming Assays-Microfocus forming assays were performed as described previously (52) with some modifications. Exponentially growing NIH3T3, B104-1-1, Vector, Grb2⌬C-11, and Grb2⌬N-11 cells were trypsinized and counted. One hundred fifty cells from each cell line were combined separately with 3.0 ϫ 10 5 NIH3T3 cells and gently mixed in a 6-cm tissue culture plate containing regular medium. Medium was replaced every 3 days over the 2-3-week period of focus formation. Foci were counted as described above. The diameters of individual foci in a random sampling from each plate were also measured.
Colony Formation in Soft Agar-Experiments were carried as previously reported with some modifications (53). Cells (1,000/well) were plated in a 24-well plate in regular medium containing 0.35% Seaplaque-agarose on an underlay of 0.7% agarose. After being cultured for 4 -6 weeks, the colonies were stained with p-iodonitrotetrazolium violet (1 mg/ml) and counted.
Antibodies-The monoclonal antibody c-neu-Ab3 targeting at the carboxyl-terminal domain of p185 and the monoclonal anti-Ras antibody (Y13-259) were from Oncogene Science. The monoclonal antiphosphotyrosine antibody (PY20) was from UBI. The rabbit anti-Shc and anti-Sos polyclonal antibodies and the monoclonal anti-Grb2 antibody were purchased from Transduction Laboratories. The monoclonal anti-HA1 epitope antibody (12CA5) was from Boehringer Mannheim.
Immunoprecipitation and Western Blotting-Unless otherwise indicated, all cells were harvested at subconfluence and lysed in 1% Triton X-100, 150 mM NaCl, 10 mM Tris, pH 7.4, 1 mM EDTA, 0.2 mM sodium vanadate, 0.2 mM phenylmethylsulfonyl fluoride, 0.5% Nonidet P-40, 10 g/ml leupeptin, and 5 g/ml aprotinin. Protein concentrations were determined against standardized controls using the Bio-Rad protein assay kit. Lysates were immunoprecipitated with appropriate antibodies according to the suppliers' specifications, and separated by SDSpolyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose filters (Schleicher & Schuell). Filters were blocked with 5% dry milk powder in TPBS (0.05% Tween 20 in phosphate-buffered saline), incubated with appropriate antibodies, incubated with goat anti-mouse or goat anti-rabbit IgG Fc-horseradish peroxidase (Boehringer Mannheim), and then visualized by chemiluminescence system (ECL).
Stripping of Western Blots-Used immunoblots were stripped by incubation with 62.5 mM Tris⅐Cl, pH 6.8, 2% SDS, and 100 mM ␤-mercaptoethanol at 75°C for 30 min. Filters were then washed twice with TPBS and reprobed with other antibodies.
Analysis of Ras-bound Guanine Nucleotides-The analysis was performed as described in the previous reports (54), with some modifications. 2.5 ϫ 10 4 cells percentimeter squared were starved overnight in phosphate-free medium supplemented with 5% dialyzed serum. Cells were then labeled for 12 h with [ 32 P]orthophosphate (400 Ci/ml) in the same medium. Cells were put on ice, rapidly washed with ice-cold phosphate-buffered saline, and lysed in 50 mM HEPES, pH 7.5, 150 mM NaCl, 20 mM MgCl 2 , 0.5% Nonidet P-40, 0.2 mM Na 3 VO 4 , 0.2 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 5 g/ml aprotinin. Nuclei and cell debris were removed by centrifugation. Ras proteins were immunoprecipitated with Y13-259, a monoclonal antibody to Ras. A goat anti-rat secondary antibody, and protein A-agarose were also applied in the immunoprecipitation. The beads were extensively washed with lysis buffer and proteins were solubilized in 1% SDS at 68°C. The bound guanine nucleotides were chromatographed on polyethyleneimine-cellulose plates in 1.3 M LiCl. The GTP/GDP ratio was determined by a Betascope 603 Blot Analyzer (Betagen, Boston, MA). The GTP-Ras percentage in the B104-1-1 cells was arbitrarily set at 100%. The percent GTP-Ras in other cell lines were standardized against that of the B104-1-1.

Suppression of the Transforming Ability of the Mutationactivated HER-2/neu by SH3 Domain Deletion Mutants of
Grb2-To observe whether the SH3 domain deletion mutants of Grb2 (Fig. 1A) can interfere with the oncogenic signaling pathway of the mutation-activated HER-2/neu receptor tyrosine kinase, we first examined the effect of the Grb2 mutants on the transforming activity of the HER-2/neu oncogene by focus forming assays in which we cotransfected cNeu-104 (a cosmid clone containing mutation-activated rat HER2/neu, see Ref. 3) together with expression vectors encoding either the amino-terminal SH3 domain deletion Grb2 (⌬N-Grb2) or the carboxyl-terminal SH3 domain deletion Grb2 (⌬C-Grb2) into NIH3T3 cells. When the expression vector encoding the ⌬N-Grb2 was cotransfected with cNeu-104, the number of foci caused by cNeu-104 decreased by more than 60%. Cotransfection of the expression vector for the ⌬C-Grb2 could also decrease the number of foci but only by approximately 25%. In control experiments cotransfection of cNeu-104 plus vector backbone alone did not affect the number of foci caused by cNeu-104 (Fig. 1B). Transfection of either expression vectors encoding the ⌬N-Grb2, ⌬C-Grb2, or the vector backbone alone into NIH3T3 cells did not result in any foci (data not shown). The results of focus forming assays indicated that the SH3 domain deletion mutants of Grb2 were able to suppress the transforming ability of the mutation-activated HER-2/neu and that ⌬N-Grb2 was more effective than ⌬C-Grb2.
Phenotypic Reversion of the Transformed B104-1-1 Cells by Stable Transfection of Grb2 Mutants with SH3 Domain Deletion-We next wanted to know whether expression of ⌬N-Grb2 or ⌬C-Grb2 could induce phenotypic reversion of established transformed cells caused by the mutation-activated HER-2/ neu. To do so, we stably transfected expression vectors encoding ⌬N-Grb2, ⌬C-Grb2, or vector backbone into B104-1-1 cells, which are transformed NIH3T3 cells expressing the mutationactivated rat HER-2/neu. After 10 -14 days of hygromycin selection, individual colonies were expanded and characterized for expressing appropriate truncated Grb2 products by West-ern analysis with the monoclonal antibody against the HA1 tag which was fused to the Grb2 mutants. The truncated Grb2 products are expected to be about 22 kDa. We focused our attention on two clones that showed comparable expression levels of ⌬N-Grb2 and ⌬C-Grb2, respectively. As shown in Fig.  2A, Grb2⌬N-11 is the stable transfectant expressing ⌬N-Grb2, whereas Grb2⌬C-11 expresses the ⌬C-Grb2. The expression levels of ⌬N-Grb2 and ⌬C-Grb2 are comparable between these two lines based on the Western blotting using anti-HA1 antibody. The mobility of the ⌬N-Grb2 product is a little slower than the ⌬C-Grb2 protein. Morphology studies indicated that the transfectant with the vector backbone alone looked similar to the transformed parental B104-1-1 cell line (Fig. 2B). In contrast, Grb2⌬N-11 showed morphologic reversal. This was evident by loss of the spindle-shaped morphology and the bright, refractile cell borders that characterize B104-1-1 cells. Moreover, Grb2⌬N-11 cells displayed an ordered growth pattern and contact inhibition, which are properties of nontransformed cells like NIH3T3 cells. The Grb2⌬C-11 cells also exhibited morphology change when compared to the B104-1-1 cells (Fig. 2B). But the morphology change did not reflect a complete reversion like that of the Grb2⌬N-11 cells.
To ascertain whether HER-2/neu was still expressed in the stable transfectants, we performed immunoblot analysis with c-neu-Ab3, a monoclonal antibody specific to p185. As shown in Fig. 2A, the expression level of p185 was comparable between the parental B104-1-1 and its transfectants, indicating that the morphologic reversion seen in Grb2⌬N-11 cells was not due to spontaneous loss of the HER-2/neu gene or down-regulation of HER-2/neu expression by the ⌬N-Grb2 mutant. Similarly, the levels of endogenous Grb2 were comparable between the parental B104-1-1 line and its transfectants along with NIH3T3 cells as shown in Western analysis with anti-Grb2 monoclonal antibody ( Fig. 2A). Both ⌬C-Grb2 and ⌬N-Grb2 were also recognized by the same anti-Grb2 monoclonal antibody ( Fig. 2A).
To evaluate more precisely the phenotypic reversion caused by the SH3 domain deletion Grb2 mutants, we performed microfocus forming assays (52) and soft agar colony formation assays to compare the transformed properties of the parental B104-1-1 cell line and its stable transfectants. As shown in Table I, the focus formation efficiency of Grb2⌬N-11 cells dramatically decreased. For example, the number of foci formed by the Grb2⌬N-11 cells was less than 25% that of the parental B104-1-1 cells. In addition, the size of foci formed by the Grb2⌬N-11 cells were much smaller than the parental B104-1-1 cells. As expected, the Vector control cell line displayed a formation efficiency similar to the B104-1-1 cell line. The Grb2⌬C-11 cell line showed only a 30% reduction in focus formation when compared to the parental B104-1-1 line and the foci were slightly smaller. Consistent with the data from the microfocus forming assays, soft agar colony formation assays also indicated that the transformation potency of the Grb2⌬N-11 cells was significantly weakened, while the Grb2⌬C-11 cells were only moderately affected (Table I). The phenotypic reversion observed in the microfocus forming assays and soft agar colony assays was not due to the decreased growth rates of Grb2⌬N-11 and Grb2⌬C-11 cells demonstrated in in vitro growth rate assays (data not shown), since extending the culture time for these two cell lines in the above assays did not result in extra numbers of foci or colonies (data not shown). Taken together, our data indicated that the transformed phenotypes of B104-1-1 cells could be largely reversed by stable transfection of the ⌬N-Grb2 mutant. The ⌬C-Grb2 mutant had a relatively weak effect.
Both ⌬N-Grb2 and ⌬C-Grb2 Compete with the Endogenous Grb2 for Shc-To investigate the molecular basis of the phenotypic reversion caused by the SH3 domain deletion Grb2 mutants, we first compared the tyrosine phosphorylation profiles between B104-1-1 and its stable transfectants. Immunoblotting analysis indicate that no obvious difference for the profiles of tyrosine phosphorylation was detected between B104-1-1 and Vector, Grb2⌬N-11, or Grb2⌬C-11 (data not shown). In particular, expression of ⌬N-Grb2 or ⌬C-Grb2 did not affect the expression and tyrosine phosphorylation of p185 and Shc (data not shown). We next tested the possibility that the SH3 domain deletion Grb2 mutants may compete with the endogenous Grb2 for Shc and p185. We first examined whether ⌬N-Grb2 and ⌬C-Grb2 could be associated with Shc and p185 by co-immunoprecipitation Western analysis. Anti-HA1 antibody was used to precipitate the ⌬N-Grb2 and ⌬C-Grb2 proteins along with their associated proteins. As shown in Fig. 3, comparable amounts of p52 Shc were detected in the anti-HA1 immunoprecipitates in both Grb2⌬N-11 and Grb2⌬C-11 cells. Immunoblot analysis with anti-HA1 antibody indicated that comparable amounts of ⌬N-Grb2 and ⌬C-Grb2 were precipitated, which suggests that the ⌬N-Grb2 and the ⌬C-Grb2 have a similar affinity for Shc. In contrast, co-precipitation of p185 by the anti-HA1 antibody was only detected in the Grb2⌬C-11 cells. The reciprocal co-immunoprecipitation experiment with anti-p185 antibody also failed to detect a physical association between p185 and the ⌬N-Grb2 in the Grb2⌬N-11 cells (data not shown). These results suggest that the ⌬N-Grb2 may not be able to associate with p185. However, we cannot rule out the possibility that the association is transient and undetectable under our experimental conditions.
Detecting the association between Shc and the truncated a One hundred fifty cells from each line to be tested were mixed with 3.0 ϫ 10 5 NIH3T3 cells in a 6-cm tissue culture plate and grown as described under "Experimental Procedures." The number of foci was counted after 2-3 weeks. The sizes of foci were also measured. The resulting number of foci from various cell lines in each experiment was standardized against that of the parental B104 -1-1 cells (set at 100%). Data shown here are the average of three separate experiments. Standard deviation is also indicated. b Cells were seeded at 1 ϫ 10 3 in 0.35% agarose containing Dulbecco's minimal essential medium/Ham's F-12 with 10% calf serum. Colonies were counted 4 -6 weeks later. The resulting number of colonies from various cell lines was standardized against that of the parental B104 -1-1 cells (set at 100%). Data shown here represent the average of quadruplicate experiments. Standard errors are also shown.
FIG. 3. Association of Shc or p185 with the truncated Grb2 products. Two milligrams of cell extracts from different cell lines were used in immunoprecipitation with monoclonal anti-HA1 antibody. Immunocomplexes or 50 g of lysates from B104-1-1 cells were separated on a 6 -12% gradient SDS-PAGE. The filter was cut into three pieces after transfer. The top portion was probed with c-Neu-Ab3 for associated p185, the middle was incubated with polyclonal anti-Shc antibody for associated Shc, and the bottom portion was incubated with monoclonal anti-HA1 antibody, in order to evaluate equal loading. The band above the truncated Grb2 products was most likely from IgG light chain.
Grb2 proteins prompted us to ask whether the interaction of Shc and the endogenous Grb2 is inhibited in Grb2⌬N-11 and Grb2⌬C-11 cells. To address this issue, we used anti-Shc antibody to precipitate Shc and associated proteins, followed by immunoblotting with anti-Grb2 or anti-Shc antibodies. As shown in Fig. 4, the endogenous Grb2 co-precipitated by the anti-Shc antibody dramatically decreased in both Grb2⌬N-11 and Grb2⌬C-11 cell lines, as compared to that in the parental B104-1-1 and the vector control cell lines. Consistently, coimmunoprecipitation of the ⌬C-Grb2 and ⌬N-Grb2 by anti-Shc antibody was detected by the anti-Grb2 antibody. Equal loading was confirmed by Western analysis with the anti-Shc antibody. These results are consistent with that seen in Fig. 3, which showed that both ⌬N-Grb2 and ⌬C-Grb2 products bound to p52 Shc . We, therefore, concluded that both ⌬N-Grb2 and ⌬C-Grb2 were able to compete with the endogenous Grb2 for Shc.
The Association between Sos and Endogenous Grb2 Was Impaired by the ⌬C-Grb2 but Not the ⌬N-Grb2-Grb2 has been shown to bind constitutively to Sos even in the absence of extracellular stimuli (38). We wondered whether the Grb2/Sos association could be interfered with by ⌬N-Grb2 or ⌬C-Grb2.
To answer this question we used anti-Sos antibody to co-precipitate endogenous Grb2 from NIH3T3, B104-1-1, Vector, Grb2⌬N-11, and Grb2⌬C-11 cells. As expected, the association of Grb2 and Sos was comparable in NIH3T3, B104-1-1, and Vector cells (Fig. 5). In contrast, the association of Sos and the endogenous Grb2 was inhibited in the Grb2⌬C-11 cells. This interference was most likely due to the competition for Sos between the endogenous Grb2 and the ⌬C-Grb2 since the anti-Sos antibody was able to co-precipitate ⌬C-Grb2. In contrast from the ⌬C-Grb2, the ⌬N-Grb2 did not significantly affect the association of Sos and the endogenous Grb2 in the Grb2⌬N-11 cells (Fig. 5). Converse immunoprecipitation with the anti-HA1 antibody also failed to co-precipitate appreciable amounts of Sos in the Grb2⌬N-11 cells (data not shown), suggesting that the association of ⌬N-Grb2 with Sos is very weak. Our results are consistent with a previous report which demonstrated that substitution of Gly-203 with Arg in the carboxyl-terminal SH3 domain of Grb2 had little effect on its binding to Sos in vitro, whereas replacement of Pro-49 with Leu in the amino-SH3 domain (a mutation causing loss of function in C. elegans Sem-5) abrogated this binding (39). However, some previous reports claimed that binding of Grb2 to Sos depended on cooperative actions of the two Grb2 SH3 domains (38,41). This discrepancy may be an issue of binding affinity.
Effect of Grb2 Mutants on Ras Activation-It has been proposed that formation of the Shc⅐Grb2⅐Sos ternary complex plays an important role in Ras activation triggered by activation of the EGF receptor, platelet-derived growth factor receptor, or insulin receptor (28 -33). Given that both ⌬C-Grb2 and ⌬N-Grb2 mutants can compete with the endogenous Grb2 for Shc and that ⌬C-Grb2 can also interact with p185 and Sos (Figs. 3-5), we examined whether signaling from the activated p185 to Ras was affected in Grb2⌬N-11 and Grb2⌬C-11 cells. Cells were labeled with 32 P i , and the guanine nucleotides bound to Ras were analyzed. As expected, the Vector control cells had a similar percentage of GTP-Ras, as compared to the B104-1-1 cells (Fig. 6). In contrast, the percentage of GTP-Ras in the Grb2⌬N-11 cells was less than 35% of that in the parental B104-1-1 cells while the percentage of GTP-Ras decreased to about 75% in the Grb2⌬C-11 cells. These results indicate that interruption of the interaction between Shc and the endogenous Grb2 can inhibit Ras activation by the activated p185 and that the ⌬N-Grb2, which does not bind to p185 or Sos, acts as a strong dominant-negative analog of the endogenous Grb2. The ⌬C-Grb2 mutant appears to be a relatively weak dominant-negative mutant, probably because it is still able to form complexes with Sos, Shc, and p185. The differential inhibition effect of ⌬N-Grb2 and ⌬C-Grb2 on Ras activation is consistent to the different extent of phenotypic reversion of B104-1-1 cells caused by these two Grb2 mutants. These observations support the idea that Ras activation is a key event in the process of transformation induced by the mutation-activated p185.

DISCUSSION
Grb2 consists of a single SH2 domain and two SH3 domains. Previous studies have indicated that Grb2 is a key component of the pathway leading to Ras activation by receptor tyrosine kinases (56). In the present study we tested whether deletion of the amino-or carboxyl-SH3 domain of Grb2 could create dominant-negative mutants which are capable of binding to tyrosine-phosphorylated Shc or mutation-activated p185 but are unable to associate with Sos. We speculated that these mutants could interfere with the recruitment of the Sos⅐Grb2 complex to Shc or p185, leading to inhibition of Ras activation. Our data demonstrated here that the ⌬N-Grb2 functioned as a dominant-negative mutant that suppressed by more than 65% the activation of Ras by the mutation-activated p185 and largely reversed the transformed phenotypes of B104-1-1 cells. The ⌬C-Grb2 appears to be a weak dominant-negative mutant. It down-regulated Ras activation, by only 25%, and slightly induced phenotypic reversion of the B104-1-1 cells. Similar results have been recently obtained for the oncogenic Bcr-Abl tyrosine kinase (57). A ⌬N-Grb2 mutant suppresses Bcr-Ablmediated Ras activation and reverses the transformed phenotype. As shown here, the ⌬C-Grb2 mutant is less effective in reversing Bcr-Abl-induced transformation as compared to the ⌬N-Grb2.
It is of interest to note that the dominant-negative effect of the ⌬C-Grb2 is much weaker than that of the ⌬N-Grb2 even though the ⌬C-Grb2 is able to compete with the endogenous Grb2 for Shc, p185, and Sos while the ⌬N-Grb2 can only bind to Shc. One possible model to explain this phenomenon is shown in Fig. 7. Wild-type Grb2 (wt-Grb2) constitutively binds to Sos mainly through its amino-terminal SH3 domain. The wt-Grb2⅐Sos complex is recruited to the tyrosine-phosphorylated Shc, subsequently resulting in Ras activation. The wt-Grb2⅐Sos may also be directly recruited to the activated p185, which is not shown here in order to simplify the model (discussion seen in the text). When introduced into the B104-1-1 cells, the ⌬N-Grb2 sequesters tyrosine-phosphorylated Shc. Therefore, recruitment of the wt-Grb2⅐Sos complex to Shc is severely impaired. Furthermore, ⌬N-Grb2 cannot appreciably bind to Sos. Thus the complex of Shc and ⌬N-Grb2 cannot lead to Ras activation. Taken together, the ⌬N-Grb2 inhibits Ras activation mediated by the Shc/wt-Grb2/Sos pathway. Unlike the ⌬N-Grb2, the ⌬C-Grb2 can bind to both tyrosine-phosphorylated Shc and Sos. Deletion of the carboxyl-terminal SH3 domain of Grb2 may not significantly affect its biological functions. In other words, the exogenously expressed ⌬C-Grb2 by itself may largely fulfill the functions of the wt-Grb2. Therefore, Ras activation is not dramatically shut down and the transformed phenotypes of the B104-1-1 cells are not obviously reversed even though the endogenous Grb2 is largely competed out from the Shc/Grb2/Sos pathway by the ⌬C-Grb2. This model is indirectly supported by previous studies on the C. elegans Sem-5. The mutation in Sem-5 (sem-5 allele n2195) corresponding to the G203R Grb2 mutant (C-terminal SH3) had a much weaker phenotypic effect in C. elegans than the mutation (sem-5 allele n1619) corresponding to the P49L Grb2 mutant (N-terminal SH3) (59). However, other studies suggest that both P49L and G203R Grb2 were loss-of-function mutants (35). For example, co-microinjection of either P49L or G203R Grb2 protein together with the H-ras protein did not stimulate DNA synthesis in quiescent rat embryo fibroblast cells while coinjection of the wild-type Grb2 and H-ras proteins enhanced DNA synthesis, suggesting that the two SH3 domains of Grb2 constitute an essential functional component of the protein.
These conflicting observations may be explained by the use of different assays in different biological systems. The functional difference between ⌬C-Grb2 and G203R Grb2 is not known currently. We speculate that the ⌬C-Grb2 possesses, at least in part, the biological functions of the wild-type Grb2 since it can bind to Shc, p185, and Sos. The slight down-regulation of Ras activation and partial reversal of transformed phenotypes in Grb2⌬C-11 cells may be due to the relatively lower efficiency of recruitment of the Sos⅐⌬C-Grb2 complex to Shc, as compared to the complex of Sos and the endogenous Grb2. This is supported by the observation seen in Fig. 4 in which comparable amounts of the ⌬C-Grb2 in the Grb2⌬C-11 cells and the endogenous Grb2 in the B104-1-1 cells were co-precipitated by the anti-Shc antibody even though the expression level of the ⌬C-Grb2 was much higher in the Grb2⌬C-11 cells than that of the endogenous Grb2 in B104-1-1 cells (Fig. 2A). This finding suggests that the ⌬C-Grb2 may have a lower affinity for Shc than does the endogenous Grb2. Alternatively, the nucleotide exchange activity of Sos may be relatively weaker in the Sos⅐⌬C-Grb2 complex than in the Sos-endogenous Grb2 complex, probably because of an unknown allosteric effect. Therefore, Ras activation in the Grb2⌬C-11 cells is not as efficient as that in the B104-1-1 cells, leading to slight reversal of the transformed phenotypes caused by the mutation-activated p185.
Previous studies on the EGF receptor signaling pathway indicated that the Grb2⅐Sos complex could be recruited to tyrosine-phosphorylated Shc or directly to the activated EGF receptor (38 -41). The data presented here imply that the Shc/ Grb2/Sos pathway is most likely the dominant one coupling the  7. Hypothetic working modes of wild-type Grb2 and its SH3 domain deletion mutants. Wild-type Grb2 (wt-Grb2) constitutively binds to Sos mainly through its amino-terminal SH3 domain. The Grb2⅐Sos complex is recruited to the Shc that is tyrosine phosphorylated, leading to Ras activation. After introduced into the B104-1-1 cells, the ⌬N-Grb2 can bind to Shc but not Sos. The Shc⅐⌬N-Grb2 complex by itself is unable to trigger Ras activation. On the other hand, ⌬N-Grb2 sequesters Shc. Therefore, the endogenous wt-Grb2 cannot be recruited to Shc. Thus, ⌬N-Grb2 is a dominant-negative mutant of Grb2. In contrast, the ⌬C-Grb2 binds to Sos and Shc. The Shc⅐⌬C-Grb2⅐Sos complex can largely fulfill the functions of the Shc⅐wt-Grb2⅐Sos complex, leading to Ras activation. Since the recruitment efficiency of ⌬C-Grb2⅐Sos by Shc is relatively lower as compared to that of the wt-Grb2⅐Sos, Ras activation is slightly reduced in B104-1-1 transfectant expressing the ⌬C-Grb2. In order to simplify the model, direct recruitment of wt-Grb2 or ⌬C-Grb2 to the mutation-activated p185 is not included in this model, which is described in the text instead. activated p185 to Ras since interference of the interaction between Shc and Grb2 by ⌬N-Grb2 leads to a dramatic inhibition of Ras activation. This idea is consistent with our previous observation that deletion of most of the autophosphorylation sites, including the potential Grb2 binding site on the mutation-activated p185 did not affect its transforming ability, suggesting that direct binding of Grb2 to p185 is not essential for Ras activation (46). Similarly, recent studies using peptide competition and immunodepletion approaches also demonstrated that formation of a complex of EGF receptor with Grb2 was only responsible for a minor part of EGF-stimulated Ras activation while the formation of the Shc⅐Grb2⅐Sos complex played the major role (28,29). Indirect evidence has been shown suggesting that Grb2 binding to tyrosine-phosphorylated Shc is more important than Grb2 binding to the insulin receptor substrate-1 in the activation of Ras in response to insulin (30,31).
Our data, however, also suggest that the Shc/Grb2/Sos pathway may not be the sole pathway that leads to the activation of Ras by the mutation-activated p185. Disruption of the association between Shc and Grb2 by ⌬N-Grb2 is unable to completely inhibit Ras activation or to completely reverse the transformed phenotypes mediated by the mutation-activated p185, suggesting the existence of multiple routes to Ras, which may not be influenced by the ⌬N-Grb2. One conceivable pathway is the direct recruitment of Grb2⅐Sos to the activated p185 since ⌬N-Grb2 appears to be unable to compete with the endogenous Grb2 for p185. Alternatively, the formation of complexes containing Grb2 and phosphorylated proteins other than Shc, which can stimulate the Ras pathway, may not be interfered with by the ⌬N-Grb2. It has been shown that a complex of Syp/SH-PTP2 tyrosine phosphatase and Grb2 can couple platelet-derived growth factor receptors to Ras (60). Recently, a Ras-GAP associated protein, named p62, has been found to form a complex with Grb2 in v-src transformed NIH3T3 cells (61). Interestingly, the presence of the Grb2⅐p62 complex correlates with the phosphorylation of p62 and cellular transformation, suggesting that the Grb2⅐p62 complex may be able to lead to Ras activation. It will be of interest to test whether these complexes exist in the B104-1-1 cells and whether ⌬N-Grb2 and ⌬C-Grb2 are able to interfere with these pathways. Furthermore, the existence of other potential pathways to Ras, in which Grb2 or Sos is not involved, may also account for the absence of complete Ras inhibition and incomplete phenotypic reversion induced by the ⌬N-Grb2. It is now known that mammalian cells contain several Ras guanine nucleotide exchange factors (GEF) apart from Sos. One of these, C3G (named for Crk SH3 binding GEF), can activate Ras in yeast (62). Intriguingly, via its proline-rich domain C3G binds the amino-terminal domain of the adaptor protein Crk (62,63). The Crk⅐C3G complex may thus, like the Grb2⅐Sos complex, couple the oncogenic signal of the mutation-activated p185 to Ras. However, no data exist showing that C3G is an exchange factor for Ras in mammalian cells. On the other hand, an alternative explanation for the failure to completely reverse the transformed phenotypes of B104-1-1 cells by the ⌬N-Grb2 could be that additional, perhaps less efficient, signaling pathways which do not involve Ras are not influenced by the ⌬N-Grb2 and may culminate in cell transformation by the activated p185. Indeed, recent studies have shown that Raf can be activated by the Drosophila torso receptor tyrosine kinase in a Ras-independent pathway (58). Our studies provided direct evidence to support the hypothesis that the Shc/Grb2/Sos pathway plays a major role in the oncogenic signaling of the mutation-activated p185 and may shed light on developing therapeutic agents to block the oncogenic signaling pathway of the p185 oncoprotein.