Differential interactions of the growth factor receptor-bound protein 2 N-SH3 domain with son of sevenless and dynamin. Potential role in the Ras-dependent signaling pathway.

In this paper, we show that the 36-45 surface-exposed sequence WYKAELNGKD of growth factor receptor-bound protein 2 (Grb2) N-SH3 domain inhibits the interaction between Grb2 and a 97-kDa protein identified as dynamin. Moreover, the peptide GPPPQVPSRPNR from dynamin also blocks the binding of dynamin to the proline-rich recognition platform of Grb2. Mutations in the 36-45 motif show that Glu-40 is critical for dynamin recognition. These observations were confirmed by immunoprecipitation experiments, carried out using ER 22 cells. It was also observed that the proline-rich peptide from dynamin was unable to dissociate the Grb2.Sos complex, whereas the proline-rich peptide from Son of sevenless (Sos) inhibited Grb2. dynamin interaction. A time-dependent stimulation of epidermal growth factor receptor overexpressing clone 22 (ER 22) cells by epidermal growth factor resulted in an immediate increase of the Grb2.Sos complex and a concomitant decrease in Grb2.dynamin. This suggests that the recruitment of Grb2.Sos to the membrane, triggered by epidermal growth factor stimulation, activates the Ras-dependent signaling and simultaneously enhances free dynamin levels, leading to both receptor internalization and endocytotic processes.

Growth factor receptor-bound protein 2 (Grb2), 1 constituted by two Src homology 3 (SH3) domains surrounding one SH2 domain, is an example of adaptors transferring information between extracellular messages monitored by membranebound receptors and intracellular transducers (1). By its SH2 domain, Grb2 recognizes phosphotyrosine residues of stimulated growth factor receptors, such as the EGF receptor, either directly or through additional adaptors such as SH2 domain containing adaptor protein (Shc) (2), whereas by its SH3 domains, Grb2 interacts with several cytosolic proteins such as Sos (Son of sevenless) the exchange factor of Ras (3).
Immunoprecipitation experiments performed on stimulated and unstimulated cells have shown that Grb2⅐Sos exists as a preformed complex (4). Recruited to the cell membrane upon stimulation, the Grb2⅐Sos complex promotes Ras-GTP formation, which in turn is able to bind Raf and thus to induce the activation of transcription factors such as fos, jun, or myc, through a cascade of mitogen-activated protein kinase stimulation (5,6). Site-directed mutagenesis experiments have shown the importance of Grb2 SH3 domains, since mutations P49L and G203R localized in the N-and C-terminal domains, respectively, were shown to block the DNA synthesis induced by activation of the Ras pathway (7,8).
However, the mechanisms involved in the down-regulation of the Ras-activated pathway, triggered by stimulation of the receptor, remain unclear. It has been shown that the stimulation of mitogen-activated protein kinases promotes Sos phosphorylation, which in turn could uncouple the mitogenic signal through dissociation of either Grb2⅐Sos or Grb2⅐receptor complexes (9 -12). The regulation of the activation and inactivation of the Ras-dependent signaling pathway could also occur directly by changes in the levels of the complexes formed between Grb2 and Sos or other ligands. Indeed, through its SH3 domains, Grb2 was reported to interact with different proteins such as dynamin, a GTPase protein involved in the endocytosis process (13,14). However, it is not yet clear whether multiple interactions involving Grb2 proceed as concomitant or successive events in the interruption of the mitogenic signal.
The Grb2 N-SH3 domain has been shown (15)(16)(17) to interact with VPPPVPPRRR, a proline-rich peptide of the C-terminal part of Sos, which has the highest measured affinity for Grb2 (8,18). In this complex, the side chains of Pro-2 and Val-5 in VPPPVPPRRR interact with aromatic residues that constitute a hydrophobic recognition platform, highly conserved in all the SH3 domains known to bind proline-rich peptides (19). In contrast, this aromatic hydrophobic platform has been shown by NMR to be lacking in the structure of the SH3 domain of Ras-GAP (20), a GTPase-activating protein that behaves as a negative regulator (21) or as an effector (22) of Ras. Accordingly, no proline-rich peptide has yet been found to interact with GAP SH3. However, in this SH3 domain, the peptide sequence 317-326 (WMWVTNLRTD), exposed at the surface (20), was shown to play a critical role in the binding of the GAP SH3-binding protein (23). Based on these results, the potential occurrence of two different binding sites on SH3 domains had been proposed (20), an hypothesis, supported by mutational analyses carried out on the SH3 domain of Src (24). Accord-ingly, Schumacher et al. (25) have recently shown that the Src SH3 domain interacts with a cyclic peptide through a binding process involving both the hydrophobic platform and other exposed amino acids such as Trp-119 and Leu-120, which belong to the corresponding region on the 317-326 peptide in  GAP-SH3. A similar recognition mode, involving a motif corresponding to the 317-326 sequence of GAP, seems to occur for the binding of the Vav C-SH3 domain with the protein Ku 70 (26). All these binding motifs are located on a ␤-strand exposed at the surface of the SH3 domains. This is also the case for the sequences 36 -45 of Grb2 N-SH3 (WYKAELNGKD) and 194 -203 (WW-GACHNGQT) of Grb2 C-SH3 domains. Interestingly, whereas the 36 -45 peptide is accessible, the 194 -203 fragment of Grb2 was shown by crystallographic analysis to interact with the N-terminal SH3 domain (27). All these results prompted us to investigate the possible existence of a protein target interacting with the sequence 36 -45 (WYKAELNGKD) of the Grb2 N-SH3 domain. We show in this paper that this peptide is able to block the interaction of Grb2 with a p97 protein, identified as dynamin, whereas the analog, where Glu is replaced by Thr, cannot block this interaction, illustrating its selectivity. Moreover, competition experiments using extracts from ER 22 cells overexpressing the EGF receptor showed that a proline-rich sequence derived from Sos displaces the Grb2⅐dynamin complex, whereas the proline-rich sequence of dynamin, shown to interact with Grb2, is unable to displace the Grb2⅐Sos complex. These results and those obtained from the stimulation of ER 22 cells by EGF suggest that the formation of the Grb2⅐Sos complex might occur at the membrane at the expense of the Grb2dynamin pool. The free dynamin thus generated might interact with other SH3 domain containing proteins, such as amphiphysin, which has been shown to be involved in neuronal receptor endocytosis (28).

EXPERIMENTAL PROCEDURES
Antibodies-The anti-Sos (Sos 1 and Sos 2, IgG3 mAb) and antidynamin (IgG1 mAb) antibodies were purchased from Transduction Laboratories (Lexington, KY). The anti-glutathione S-transferase (GST) monoclonal antibody was purchased from Hybridolab Pasteur Institute (Paris, France). The polyclonal anti-Grb2 used for immunoprecipitation studies was purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
GST Fusion Proteins-DNA sequences corresponding to Grb2 (residues 1-210) were amplified by polymerase chain reaction and cloned into the expression vector pGEX2T at BamHI and EcoRI restriction sites. Bacteria transformed with the recombinant plasmids were grown, induced with isopropyl-1-thio-␤-D-galactopyranoside, and disrupted by sonication (29). The GST-SH3 fusion proteins were purified by affinity chromatography onto glutathione-agarose beads (Sigma) followed by elution with 10 mM reduced glutathione.
Peptide Synthesis and Purification-The peptides were prepared by solid phase synthesis. The protected peptide chains were assembled according to the stepwise solid phase method of Merrifield (45) on an Applied Biosystems 431A peptide synthesizer with hydroxybenzotriazole/dicyclohexylcarbodiimide as coupling reagents. Synthetic peptides corresponding to WYKAELNGKD, WYKATLNGKD, WMWVTNLRTD, GPPPQVPSRPNR, and VPPPVPPRRR were synthesized using 10 eq of amino acids, whereas SH3 domains were synthesized using 20 eq, as described previously (20). The peptides were purified by high pressure liquid chromatography using a C4 (for Grb2 N-SH3 domain) or a C8 (for deca-and dodecapeptides) Vydac 5-mm column diameter (220 ϫ 10 mm) and a linear gradient of B (where A is trifluoroacetic acid 0.1% and B is CH 3 CN 70%, trifluoroacetic acid 0.09%) at a flow rate of 1.5 ml/min with detection at 214 nm. The identity of the peptides was checked by electrospray mass spectroscopy.
Cells and Preparation of Cell Lysates-Adherent ER 22 cells (hamster fibroblasts overexpressing the human epidermal growth factor receptor, a kind gift from Dr. J. Pouyssegur, France) were routinely grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum containing the antibiotic G418 (200 g/ml) and 2 mM L-glutamine (Life Technologies, Inc.) and were incubated at 37°C in 5% CO 2 (23). Cells were grown to confluence and were serum-starved for 24 h.
Cell Stimulation by EGF-Cells were grown to confluence and then serum-starved for 24 h. They were then stimulated with EGF (50 ng/ml) for 1, 5, 15, or 30 min in Dulbecco's modified Eagle's medium, at room temperature (30), and immediately washed three times with ice-cold PBS containing 1 mM Na 3 VO 4 and lysed as described above.
Affinity Precipitated Proteins and Competition Assay-Glutathioneagarose beads (150 l containing 60 g of GST fusion protein) were mixed with lysates (5 mg of cellular proteins), with or without inhibitor peptide, and rocked at 4°C overnight. Beads were washed 4 times with HNG (50 mM Hepes, 150 mM NaCl, 10% glycerol) containing 0.2% of Triton X-100. Affinity precipitated proteins were eluted by boiling in SDS sample buffer for 5 min, fractionated by 7.5% SDS-PAGE, and revealed either by far Western blot or by immunoblotting experiments.
Immunoprecipitation of Proteins-Whole cell lysates (3 mg) were pre-cleared with 50 l of protein A-Sepharose CL-4B (Pharmacia Biotech Inc.) for 1 h at 4°C. Grb2 was then immunoprecipitated by incubation of the supernatant with 4 g of a Grb2 polyclonal antibody overnight at 4°C, in the presence or absence of the peptide (1.5 mM). The resultant immune complexes were precipitated by incubation with 50 l of protein A-Sepharose for 2 h at 4°C. The pellets were washed three times with HNTG, resuspended in SDS sample buffer, and heated at 100°C for 10 min. The immune complexes were then resolved by SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore Corp.). The membranes were then subjected to Western blot as described above.
Inhibition of Grb2 N-SH3 and Purified Dynamin Complexation-Dynamin was purified from rat brains by a novel purification procedure (31). 10 mg of the Grb2 N-SH3 domain (15) were coupled to 1 ml of CNBr-activated Sepharose 4B (Pharmacia), according to Hermanson et al. (32) (coupling yield, 92%). 30 l of Grb2 N-SH3 beads were then preincubated, in the presence or absence of inhibitory peptides, overnight at 4°C in 300 l of PBS buffer. 3 g of purified dynamin were then added and incubated for 2 h at 4°C. Beads were washed 4 times with PBS. Affinity precipitated proteins were eluted by boiling SDS sample buffer for 5 min. After SDS-PAGE separation and transfer, protein was detected by an anti-dynamin Western blot.
Far Western Blot Analysis-As already described (23) precipitated proteins were fractionated by SDS-PAGE (7.5% gel) and transferred onto a PVDF membrane. Nonspecific binding to the filters was blocked by adding 2% skim milk in PBS containing 0.05% Tween 20 for 2 h at room temperature. The filters were then incubated with GST-Grb2 proteins in blocking buffer for 12 h at 4°C. After washing in PBS, 0.05% Tween 20 bound proteins were detected by successive incubations with the anti-GST mAb (0.25 g/ml), peroxidase-conjugated anti-mouse antibody, and revealed by the ECL method (Amersham Corp.).
Immunoblotting Experiments-Precipitated proteins were separated and transferred onto PVDF membrane as described for far Western analysis. After blocking, the membrane was incubated with mAb antibody in blocking buffer for 2 h. After successive washes in PBS, 0.05% Tween 20, bound proteins were detected by incubation with peroxidaseconjugated anti-mouse antibody and revealed as described above by the ECL method. Band intensity was quantified using a Bio-Profil V6.0 Vilber-Lourmat (Marne la Vallée, France). Data are the mean values (Ϯ S.E.) of three independent assays.
In the case of immunoprecipitation, the Western blot against Grb2 was carried out using an anti-mouse Fc antibody to avoid possible cross-reaction with the low chain of the antibody used for immunoprecipitation. In all experiments, Grb2 amounts were constant (data not shown).

RESULTS
Peptide- (36 -45) from Grb2 Inhibits the Association of Grb2 with a p97 Protein-To investigate the hypothesis of the presence of a second binding site on N-SH3 Grb2, the 36 -45 pep-Differences in Grb2 N-SH3 Recognition by Sos and Dynamin tide (WYKAELNGKD) corresponding to the 317-326 sequence of GAP was synthesized and tested for its ability to inhibit potential interactions between Grb2 and putative target proteins. The lysates of Chinese hamster lung fibroblasts overexpressing the human epidermal growth factor (EGF) receptor (ER 22 cells) were used for this purpose. To avoid SH2 interactions, lysates of cells in the G 0 phase were incubated with GST-Grb2 loaded onto agarose-glutathione beads, in the presence or absence of the 36 -45 peptide at a final concentration of 500 M. The proteins complexed with Grb2 were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto a polyvinylidene difluoride (PVDF) membrane, and revealed by a far Western blot with GST-Grb2 as probe. Lane 1 of Fig. 1A shows different bands corresponding to protein ligands of Grb2 in ER 22 cells. As shown in Fig. 1A (lane 2), peptide-(36 -45) from the Grb2 N-SH3 domain significantly reduced the interaction of Grb2 with only one protein with a mass of 97 kDa.
Identification of the p97 Protein as Dynamin-Among the target proteins of the SH3 domains of Grb2, Vav (33) and dynamin (34) have molecular masses ranging from 95 to 100 kDa. Since Vav is exclusively present in hematopoietic cells and dynamin is ubiquitous, the proteins complexed to Grb2 in ER 22 cells lysate were analyzed using an anti-dynamin antibody as a probe. Western blots (Fig. 1B) showed that the band at 97 kDa (lane 1) corresponds mainly to dynamin. Moreover, peptide- (36 -45) from Grb2 N-SH3 at 500 M strongly inhibited the interaction between dynamin and Grb2 (lane 4), whereas neither the peptide-(317-326) from the GAP SH3 domain (lane 3) nor the mutated E40T peptide-(36 -45) (lane 5) were able to modify the level of the Grb2⅐dynamin complex. The same quantity of GST-Grb2, evaluated by Western blot using an anti-GST antibody, was used in both of these experiments. Therefore, the decreased intensity of the p97 signal following addition of peptide-(36 -45) from Grb2 (Fig. 1A, lane 2 or Fig. 1B, lane 4) can be attributed to the inhibition of dynamin binding to Grb2 SH3 domains. Until now, the formation of these complexes has been essentially attributed to the binding of proline-rich motifs with the hydrophobic recognition platform of SH3 domains (34,35). The present results show that dynamin interacts with Grb2 through a dual recognition process also involving the 36 -45 region of the N-SH3 domain.
Comparison of Grb2 Interactions with Dynamin and Sos-The C-terminal region of Sos contains four conserved prolinerich motifs among which VPPPVPPRRR, corresponding to amino acids 1149 -1158, was shown to have the highest affinity for Grb2 and in particular for its N-SH3 domain (8,18). Similarly, in the C-terminal domain of dynamin, the peptide GPP-PQVPSRPNR (amino acids 827-838) was reported to bind Grb2 SH3 domains (34,36). The two proline-rich peptides were therefore tested as competitors for the interactions involving Grb2 and either Sos or dynamin on the ER 22 cells lysate. The complexes were revealed by Western blots using anti-Sos and anti-dynamin antibodies. As shown in Fig. 2A, lane 1, GST-Grb2, loaded onto glutathione-agarose beads, binds Sos and dynamin. Peptide GPPPQVPSRPNR from dynamin, at 500 M, prevented the interaction between Grb2 and dynamin and very weakly (less than 10%) the interaction between Grb2 and Sos ( Fig. 2A, lane 2). In contrast, at the same concentration, the peptide VPPPVPPRRR from Sos completely inhibited the formation of the complexes formed between Grb2 and either Sos or dynamin (Fig. 2A, lane 3). Moreover, the peptide- (36 -45) from Grb2 N-SH3 reduced the Grb2⅐dynamin interaction without modifying the Grb2⅐Sos binding (lane 4). The easier displacement of dynamin from its complex with GST-Grb2 by VPPPVPPRRR is in agreement with the stronger affinity of this proline-rich peptide for Grb2 N-SH3 (K d ϭ 4 M), as com-  -45)). Bound proteins were resolved by 7.5% SDS-PAGE, and replicated PVDF membranes were subjected to Western blot analysis with anti-GST, anti-dynamin, or anti-Sos antibodies followed by incubation with peroxidase-conjugated anti-mouse antibody. The GST staining constitutes the control. The Grb2⅐Sos interaction is inhibited by proline-rich peptides from Sos and dynamin (to a much lower extent) but not by the 36 -45 peptide of Grb2, whereas the Grb2⅐dynamin interaction is inhibited by both proline-rich peptides and by the surface 36 -45 peptide from Grb2. B, the same experiment using the peptide- (36 -45) in a dose-dependent manner shows that Grb2⅐dynamin interaction but not Grb2⅐Sos one is inhibited. pared with that of GPPPQVPSRPNR (K d ϳ300 M), measured by fluorescence (data not shown).
These results are also in favor of an additional role for the region of Grb2 N-SH3 encompassing the sequence 36 -45 in the specific recognition with dynamin. This was confirmed by the dose-dependent selective blockade of the formation of the Grb2⅐dynamin complex by the 36 -45 peptide (Fig. 2B). To confirm the dual recognition mode of Grb2 N-SH3 with dynamin, similar experiments were performed using purified dynamin. The synthetic N-SH3 domain was loaded onto CNBr-Sepharose beads, which were further incubated in PBS buffer with purified dynamin, in the presence of increasing concentrations of inhibitory peptides. As shown in Fig. 3A, both the proline-rich peptide from dynamin and the 36 -45 peptide from Grb2 were able to inhibit, in a dose-dependent manner, the Grb2 N-SH3⅐dynamin interaction. The occurrence of two different binding components on Grb2 N-SH3 for dynamin recognition was confirmed by competition experiments with 300 M Grb2 36 -45 peptide (Fig. 3B, lane 2) or 100 M GPPPQVPSR-PNR from dynamin (Fig. 3B, lane 3) (two sub-effective doses, compared with the control lane 1). Moreover, as shown in Fig.  3B, lane 4, the association of the two peptides produced a clear synergistic inhibition of the complex.
Immunoprecipitation of Grb2⅐Sos and Grb2⅐Dynamin Complexes from ER 22 Cells, Selective Inhibition-To confirm the existence of a dual recognition process between Grb2 N-SH3 and dynamin at the cellular level, competition experiments were performed on ER 22 cells. Cellular lysates in G 0 phase were immunoprecipitated using an anti-Grb2 antibody, in the presence or absence of inhibitory peptides at 1.5 mM. Immunoprecipitated proteins were revealed by Western blotting, using anti-Sos or anti-dynamin antibodies (Fig. 4). In the absence of competitor peptides, Grb2 clearly co-immunoprecipitated with both Sos and dynamin (lane 1). In the presence of the 36 -45 peptide (lane 3), only Grb2⅐dynamin immunoprecipitation was partially inhibited (50% as compared with controls). As shown in lane 2, the 36 -45 peptide containing the E40T mutation was unable to modify Grb2⅐dynamin interactions. This is in accordance with results showing that the E40T mutated N-terminal SH3 domain of Grb2 does not recognize dynamin with a high affinity. 2 Moreover, the proline-rich peptide derived from dynamin GPPPQVPSRPNR inhibited the Grb2⅐dynamin complexation but only very slightly the Grb2⅐Sos interaction (lane 4), whereas VPPPVPPRRR from Sos totally inhibited both Grb2-Sos and Grb2⅐dynamin interactions (lane 5).
The demonstration that the proline-rich peptide from Sos blocks the formation of the Grb2⅐dynamin complex, whereas the proline-rich peptide from dynamin is unable to inhibit the Grb2⅐Sos recognition process at the cellular level (Figs. 2 and  4), raises the question of the biological relevance of these results. Indeed, dynamin seems to play a crucial role in endocytosis and is involved in membrane-bound receptor internalization, through formation of clathrin-coated pits (37), thus constituting a possible Ras signaling regulatory pathway.

Kinetics of Grb2⅐Sos and Grb2⅐Dynamin Complexes Formation Induced by EGF Stimulation in ER 22 Cells-In the ER 22 cells used in this study, preformed complexes between Grb2
and dynamin were easily demonstrated by co-immunoprecipitation with Grb2 antibodies (Figs. 4 and 5). Such preformed complexes have already been observed in unstimulated Chinese hamster ovary cells expressing the insulin receptor (38) but not in unstimulated Madin-Darby canine kidney cells expressing the EGF receptor (39). These complexes were reported to constitute potential storage forms of Grb2 (1). Grb2 immunoprecipitation experiments on ER 22 cells at different EGF stimulation times were therefore performed (Fig. 5). Interestingly, after 1 min of EGF stimulation, a large increase in the levels of the Grb2⅐Sos complex was observed, with a concomitant decrease of the Grb2⅐dynamin complex of approximately 40% (Fig. 5, A and B). A slow restoration of the initial complexes appeared to occur after 30 min of EGF stimulation. These results suggest a displacement of the Grb2⅐dynamin complex by Sos in vivo and concomitantly a release of free dynamin.

DISCUSSION
The aim of this study was to investigate the occurrence of protein target(s) able to recognize the 36 -45 peptide from the Grb2 N-SH3 domain, which is located at the surface of the protein (15,27) as are the corresponding motifs in other SH3 domains such as GAP (20) or Src (40).
We clearly show here, in addition to the hydrophobic prolinerich recognition platform, that the 36 -45 sequence of Grb2 N-SH3 (WYKAELNGKD) plays a critical role in the interaction with dynamin. Indeed the 36 -45 peptide alone was able to inhibit dose-dependently the formation of complexes between dynamin and Grb2 or its N-terminal SH3 domain. Furthermore, illustrating the specificity of the interaction, the nature of the peptide sequence also seems to be important, since the replacement of the aspartic acid in position 40 by a threonine resulted in a peptide unable to compete for the Grb2⅐dynamin complex. In accordance, Grb2 N-SH3 domain mutated as E40T, while keeping similar capacity to bind Sos, shows a strongly decreased capacity to recognize dynamin. 2 Moreover, recent results obtained in our laboratory by using plasmon surface resonance have shown that while Grb2 or its mutant P206L in the C-terminal SH3 have similar affinities for dynamin (K d values of 20 nM), the P49L mutant, characterized by a disruption of the ␤-barrel structure of the N-terminal SH3 domain, has lost its affinity for dynamin. 2 These data support the essential role played by the N-terminal SH3 domain of Grb2 in the binding to dynamin. Thus, the capacity of WYKAELNGKD from Grb2 or the proline-rich peptide from dynamin to inhibit the Grb2⅐dynamin interaction and the synergistic effect of these two peptides (Fig. 3B) demonstrate that they constitute two essential structural parameters in the Grb2⅐dynamin recognition.
In addition, it is noteworthy that regions corresponding to the peptide-(36 -45) from Grb2 in GAP, Src, and Vav SH3 domains seem also to be involved in protein recognition (20,25,26). These results suggest that this accessible sequence located at the S4 ␤-strand of SH3 domains might constitute a potential second binding component involved in their recognition with some targets. Small proline-rich peptides have already been shown to have lower affinities than the corresponding entire proteins for SH3 domains. This is the case for the VPPPVP-PRRR peptide from Sos, which has an affinity of around 4 M for Grb2 (18), whereas the Grb2⅐Sos affinity has been reported to be in the nanomolar range (18). This difference was essentially attributed to thermodynamically unfavorable factors encountered during the binding processes of small flexible ligands (41). In our case, the affinity of the proline-rich peptide derived from dynamin (GPPPQVPSRPNR), which had been shown to be selectively involved in the interaction with Grb2, is low (ϳ300 M) as measured by fluorescence. Nevertheless, this peptide is able to inhibit the interaction of dynamin with the whole Grb2 protein or its N-SH3 domain. Thus, the existence of complementary interactions involving the 36 -45 peptide is undoubtedly important to justify the 20 nM affinity that we have measured between dynamin and Grb2. 2 The importance of such additional interactions has also been proposed, from thermodynamic measurements, to account for the binding affinity of the p85 subunit of PI3 kinase with the SH3 domain of Fyn (41).
Another interesting point is the demonstration that the proline-rich peptide from Sos blocks the formation of the Grb2⅐dynamin complex, while the proline-rich peptide from dynamin is unable to inhibit the Grb2⅐Sos recognition (Figs. 2 and 4), suggesting that an increasing concentration of Sos might locally displace the Grb2⅐dynamin complex in cells containing such a preformed complex (34,38). This is in agreement with the kinetics of the evolution of the Grb2⅐Sos and Grb2⅐dynamin complexes, observed by EGF stimulation for ER 22 cells, as shown in Fig. 5. Indeed, it is conceivable that the increase of Grb2⅐Sos, concomitant with the decrease of Grb2⅐dynamin, might result, at least for a part, from the displacement of dynamin by an increase of Sos concentration, although the affinity of both Grb2⅐Sos and Grb2⅐dynamin are of the same order. Freeing dynamin from its complex with Grb2 might decrease its GTPase activity (13) and thus promote, as was already proposed (42,43), its accumulation under GTPbound form around the neck of endocytotic vesicle budding, an early phase of the coated pit formation in receptor internalization. It is clear that, to prove such an assumption, it would be necessary to analyze more precisely the subcellular concentrations of these complexes and of the free proteins. Moreover, comparative immunofluorescence kinetic analyses of dynamin Grb2 and EGF receptor on ER 22 cells under EGF stimulation could be performed to dissect further the biology of their interactions. Nonetheless, these results provide evidence of an interconnection between both activation and deactivation of Ras pathways. Otherwise, several results suggest that dynamin might interact with the SH3 domain of amphiphysin, in connection with its role in endocytosis (28). Thus, it would be interesting to analyze whether in ER 22 cells an amphiphysinlike protein might be able to trap dynamin near the membrane and promote the receptor endocytosis (44) or whether the Grb2⅐dynamin complex is directly involved in endocytosis as proposed by Wang and Moran (39). FIG. 5. A, immunoprecipitation of Grb2⅐Sos and Grb2⅐dynamin complexes following different EGF stimulation times. ER 22 cell extract was submitted to a Grb2 immunoprecipitation procedure following different EGF stimulation times. The precipitated proteins were then subjected to a Western blot using either anti-Sos (top) or anti-dynamin (bottom) antibody. After short stimulation times an increase of Grb2⅐Sos complex was observed, whereas the Grb2⅐dynamin pool decreased. B, kinetics of Grb2⅐Sos and Grb2⅐dynamin complexes after different EGF stimulation times. Quantification of the band intensities clearly showed that the Grb2⅐Sos complex increased after 1 min of EGF stimulation up to 100%, whereas the Grb2⅐dynamin complex decreased 40%. The curves represent the means (Ϯ S.D.) of three experiments performed in triplicate.