Epidermal Growth Factor-induced DNA Synthesis KEY ROLE FOR Src PHOSPHORYLATION OF THE DOCKING PROTEIN Gab2*

We have previously demonstrated that phosphatidylinositol 3-kinase (PI3-kinase) is necessary and sufficient to account for epidermal growth factor (EGF)-induced mitogenesis in rat primary hepatocytes. A cytosolic Gab2-containing complex accounts for > 80% of the total EGF-induced PI3-kinase activity (Kong, M., Mounier, C., Wu, J., and Posner, B. I. (2000) J. Biol. Chem. 275, 36035– 36042), suggesting a key role for Gab2 in EGF-induced mitogenesis. Here, we demonstrate that PP1, a selective inhibitor of Src family kinases, blocks the EGF-induced Gab2 tyrosine phosphorylation without inhibiting EGF-induced phosphorylation of the EGF receptor, ErbB3, or Shc. We also show that Gab2 phosphorylation is increased in Csk knockout cells in which Src family kinases are constitutively activated. Furthermore, PP1 blocks Gab2-associated downstream events including EGF-induced PI3-kinase activation, Akt phosphorylation, and DNA synthesis. We demonstrate that Gab2 and Src are constitutively associated. Since this association involves the proline-rich sequences of Gab2, it probably involves the Src homology 3 domain of Src kinase. Mu-tation of the proline-rich sequences in Gab2 prevented EGF-induced Gab2 phosphorylation, PI3-kinase/Akt activation,

Upon ligand binding, the activated epidermal growth factor receptor (EGFR) 1 mediates a number of important biological responses, including the stimulation of cell proliferation, mi-gration, and differentiation (1)(2)(3). Src family kinases are nonreceptor tyrosine kinases and have been described as essential mediators of EGF signaling (4). Like most nonreceptor tyrosine kinases, Src family kinases contain an Src homology 2 (SH2) domain, which binds phosphotyrosine residues, and an Src homology 3 (SH3) domain, which binds proline rich sequences (reviewed in Ref. 5). Several studies have established that Src kinases are required for growth factor-induced mitogenesis such as that effected via receptors for EGF (6 -8), the plateletderived growth factor (9,10), and colony stimulation factor-1 (6). To date, the manner in which Src family kinases participate in effecting the mitogenic response is unclear.
In previous studies, we demonstrated that the activation of PI3-kinase and not mitogen-activated protein kinase (MAPK), is necessary and sufficient to account for EGF-induced mitogenesis (12,20). Although activated PI3-kinase was shown to associate with three phosphotyrosine-phosphorylated proteins (ErbB3, Shc, and Gab2), over 80% was found in a multimeric complex consisting of Gab2-p85-SHP2-Grb2 (12). Confirming the key role of Gab2 was our finding that overexpression of wild type Gab2 (WTGab2) augmented EGF-induced PI3-kinase activity and DNA synthesis, whereas the Gab2 mutant (Gab2⌬p85) lacking p85 binding sites (pYXXM motifs as reviewed in Ref. 21, where pY represents phosphotyrosine) effected no such augmentation. Furthermore, we showed that following EGF treatment, the phosphorylated multimeric Gab2 complex was exclusively cytosolic and did not associate with membranes. Nor did overexpression of the pleckstrin homology domain of Gab2 interfere with EGF-induced Gab2 phosphorylation or mitogenesis. 2 In the present study, we considered the possibility that Gab2 is phosphorylated by a tyrosine kinase other than the EGFR and thus sought to elucidate a link between Src family kinases and Gab2 in the regulation of EGF-induced mitogenesis in primary hepatocytes. Our study demonstrates that Src kinase(s) promote EGF-induced PI3-kinase activation and DNA synthesis through effecting the tyrosine phosphorylation of Gab2. In addition, we found that the proline-rich domains of Gab2 are essential for constitutive Src association with and tyrosine phosphorylation of Gab2. Finally, we demonstrate that these Src-dependent responses are down-regulated by the association of SHP2 with Gab2 and that this latter association is critical to EGF-induced MAPK activation.
Preparation of the Gab2 Mutants-The ⌬SHP2Gab2 was generated by replacing Tyr-603 and Tyr-632 with phenylalanine using the Chameleon double-stranded site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. The mutant was verified by DNA sequencing. A plasmid containing ⌬ProGab2 with a deletion of 348 -355 amino acids and mutations of P500A and R504A was kindly provided by Dr. Morag Park (McGill University, Montréal, Canada). In order to produce the adenoviruses, Gab2 mutants as well as the full-length cDNA (to generate the WTGab2 construct) were subcloned into pShuttle between the NheI and NotI sites using 5Ј-oligonucleotides containing an NheI site and the sequences encoding different tags (hemagglutinin for WTGab2, FLAG for ⌬ProGab2, and Myc for ⌬SHP2Gab2) and 3Ј-oligonucleotides containing a NotI site. The recombinants of adenoviral DNA were further generated using the Adeno-X Expression System (Clontech, Palo Alto, CA) according to the kit's user manual.
Adenovirus Preparation and Titration-Large scale production of recombinant viral particles was performed by infecting 293A cells. The titer of viral particles was determined using the Tissue Culture Infectious Dose 50 (TCID 50 ) method as described in the protocol of the Ad-easy vector system (Qbiogene, Carlsbad, CA).
Cell Culture and Viral Infection-Primary hepatocytes were prepared as previously described (12). Before infection, cells were bathed for 24 h in Dulbecco's modified Eagle's medium/F-12 medium containing 10% fetal bovine serum, 10 mM HEPES, 20 mM NaHCO 3 , 500 IU/ml penicillin, and 500 g/ml streptomycin. Wild type and Csk Ϫ/Ϫ cells, kindly provided by Dr. Philippe Soriano (Fred Hutchinson Cancer Research Center, Seattle, WA) (22) were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. All cells were then infected with 10 MOI of different recombinant adenoviruses for 3 h at 37°C. The infected cells were then serum-starved for 48 h before harvesting.
Immunoprecipitation and Immunoblotting-Cell lysates, prepared as previously described (12), from EGF-treated or nontreated cells were precleared with nonimmune rabbit IgG (Sigma) and protein A-Sepharose for 1 h at 4°C. After centrifugation, the resulting supernatants were incubated for 2 h at 4°C with the antibody indicated in the figure legends. Protein A-Sepharose (50 l of a 50% slurry) was added to each sample, and the incubation was continued for an additional 1 h. Immune complexes were isolated by centrifugation, washed three times in PBS, and boiled in Laemmli sample buffer. Immunoprecipitates or intact samples were subjected to SDS-PAGE, transferred to Immobilon-P membranes (Millipore Ltd., Mississauga, Canada), and immu-noblotted with the indicated first antibody for 90 min followed by a 1-h incubation with horseradish peroxidase-, or 125 I-labeled goat anti-rabbit antibody or goat anti-mouse antibody IgG. Immunoreactive proteins were detected by autoradiography or by ECL system (Amersham Biosciences). Densitometric quantification of the signals was performed using a Bio-Rad densitometer, model GS-700.
[ 3 H]Thymidine Incorporation Assay-After viral exposure, infected cells were serum-starved for 20 h in serum-free medium, and then 100 ng/ml EGF and [ 3 H]methylthymidine (5 Ci/ml) were added to the medium. After an 18-h incubation, cells were rinsed three times with 3 ml of cold PBS, incubated for 15 min at 4°C in 10% trichloroacetic acid, solubilized at room temperature in 1 ml of 1 N NaOH, neutralized with 1 ml of 1 N HCl, and then transferred to scintillation vials and counted for 3 H.
PI3-kinase Activity Assay-Lysates (500 g of protein) from EGFtreated (100 ng/ml EGF for 1 min) or nontreated cells were immunoprecipitated in the presence of protein A-Sepharose, using different antibodies as indicated in the figure legends. Immunoprecipitates were extensively washed, and the Protein A-Sepharose pellets were resuspended in 50 l of kinase assay buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.5 mM EGTA) containing 0.5 mg/ml L-␣-phosphatidylinositol (Avanti Polar Lipids, Inc., Alabaster, AL) and assayed for PI3-kinase activity as previously described (20).

A Key Role for Src Family Kinases in EGF-induced Gab2
Phosphorylation-In the liver, tyrosine-phosphorylated Gab2 is largely localized in the cytosol, could not be demonstrated to associate with the EGFR, and is phosphorylated in a pleckstrin homology domain-independent manner. 2 EGF has been shown to activate tyrosine kinases of the Src family (4), and studies have linked their activation to cell proliferation (6 -10). We tested the hypothesis that Gab2 is a Src substrate by examining the effect of PP1, a selective inhibitor of Src family kinases (23), on EGF-induced Gab2 tyrosine phosphorylation in rat primary hepatocytes. As shown in Fig. 1A (top panels), PP1 reduced EGF-induced tyrosine phosphorylation of Gab2 in a dose-dependent manner. Compared with cells incubated with vehicle only, 1 M PP1 decreased Gab2 phosphorylation by 50% and 20 M PP1 by more than 90%. In contrast, the same doses of PP1 had no effect on EGF-induced tyrosine phosphorylation of EGFR (Fig. 1A, middle panels), ErbB3, or Shc (Fig. 1A, bottom panels). Nor did these doses of PP1 affect insulin-induced tyrosine phosphorylation of IRS-1 or IRS-2 ( Fig. 1B). PP2, another selective inhibitor of Src family kinases, also inhibited the stimulatory effect of EGF on Gab2 tyrosine phosphorylation (data not shown). The specificity of inhibition by PP1 and PP2 of EGF-induced tyrosine phosphorylation of Gab2 suggests that, in rat hepatocytes, this molecule, but not other tyrosine-phosphorylated docking proteins, is a substrate for Src family kinase(s).
All Src family tyrosine kinases are negatively regulated by phosphorylation at a carboxyl-terminal tyrosine (Tyr-527 in Src kinase) carried out by another nonreceptor tyrosine kinase, Csk (C-terminal Src kinase) (24). Cells derived from Csk-deficient embryos exhibit an order of magnitude increase in activity of Src and the related Fyn kinase (22,25). To further confirm the role of Src family kinases on Gab2 phosphorylation, we infected Csk-deficient and wild type mouse embryo fibroblasts with recombinant Gab2 adenovirus and starved the cells for 48 h. Western blot analysis, using a specific phospho-Src antibody (p-Src416) (26), demonstrated that Src kinase is constitutively activated in Csk ؊/؊ cells ( Fig. 2A, top panel). Gab2 tyrosine phosphorylation was then investigated in the absence of EGF stimulation. Immunoprecipitation with ␣Gab2 and Western blotting with anti-phosphotyrosine demonstrated that Gab2 phosphorylation ( Fig. 2A, middle panel) is increased 1.8fold in Csk ؊/Ϫ compared with wild type fibroblasts (Fig. 2B). This study supports the view that Gab2 is a substrate for Src family kinases in vivo.

Gab2 Tyrosine Phosphorylation by Src Family Kinases Is Essential for EGF-dependent Downstream Signaling Events-
In previous work, we showed that over 80% of PI3-kinase, activated following EGF treatment, was associated with tyrosine-phosphorylated Gab2 (12) and that overexpression of Gab2 was sufficient to augment EGF-induced DNA synthesis. 2 In this study, we pretreated hepatocytes with either vehicle (Me 2 SO) or 20 M PP1 for 30 min followed by a stimulation with 100 ng/ml EGF. As shown in Fig. 3A, EGF induced a 20-fold increase in Gab2-associated PI3-kinase activity in control cells. Pretreatment with PP1 abolished this activation, that of Akt, as reflected in Akt-Ser-473 phosphorylation (Fig.  3B), and DNA synthesis (Fig. 3C). Similar results were obtained by treatment of the cells with PP2 (data not shown). Our results demonstrate that EGF-induced tyrosine phosphorylation of Gab2, mediated by Src family kinases, is necessary for EGF-induced DNA synthesis in primary rat hepatocytes and that the tyrosine phosphorylation of ErbB3 and Shc, which is unaffected by PP1 (Fig. 1), cannot supplant the critical require-ment for tyrosine-phosphorylated Gab2 for this response.
Association of Src with Gab2: Significance for Gab2 Tyrosine Phosphorylation and Downstream Signaling-In our previous experiments, demonstrating that Gab2 is a substrate for Src family kinases, we observed a 60-kDa band in anti-Gab2 immunoprecipitates (data not shown). We therefore examined the possibility that Src kinase was associated with its substrate, Gab2. As shown in Fig. 4A (top panel), Src family kinases are Other aliquots were incubated with anti-Gab2 antibody, and the immunoprecipitates were subjected to 7.5% SDS-PAGE followed by immunoblot analysis with anti-phosphotyrosine antibody (middle panel) or anti-Gab2 antibody (bottom panel). B, the level of Gab2 tyrosine phosphorylation, quantified as described under "Experimental Procedures," is expressed in Csk Ϫ/Ϫ cells (solid bar) as -fold over wild type cells (hatched bar) (mean Ϯ S.E.; n ϭ 3; *, p Ͻ 0.01). constitutively associated with endogenous Gab2, and EGF treatment did not appear to augment this association. This suggested that the association may result from the binding of the SH3 domain of Src with proline-rich sequences in Gab2. To test this hypothesis, we generated a recombinant adenovirus containing a Gab2 mutant in which key proline-rich sequences were mutated (⌬ProGab2) (see "Experimental Procedures"). Hepatocytes were infected with either WTGab2 or ⌬ProGab2 recombinant adenovirus and starved for 48 h before EGF treatment. In contrast to WTGab2, overexpressed ⌬ProGab2 manifested no association with Src kinase (Fig. 4B, top panel, lanes  1 and 2 versus lanes 3 and 4), indicating that the proline-rich sequences are essential for mediating the binding of Src family kinases to Gab2. Src kinase phosphorylation was analyzed in anti-Gab2 immunoprecipitates. A tyrosine-phosphorylated band migrating at the same position as Src kinase (60 kDa) was detected in EGF-treated cells infected with WTGab2 but not in EGF-treated cells infected with ⌬ProGab2 (Fig. 4C, top panel). These results suggest that Src kinase is activated upon EGF treatment, although it is constitutively associated with Gab2.
We sought to determine the extent to which the loss of constitutive Src binding influenced Gab2 phosphorylation by Src family kinases. Primary hepatocytes were infected with recombinant WTGab2 or ⌬ProGab2 adenoviruses. As shown in Fig. 5 (top panel), EGF treatment resulted in substantial tyrosine phosphorylation of WTGab2, whereas the phosphorylation of ⌬ProGab2 is reduced by more than 70% (Fig. 5, bottom  panel). These results demonstrate that, in hepatocytes, the proline-rich sequences of Gab2 are important for Src kinasemediated Gab2 phosphorylation.
Since overexpressing ⌬ProGab2 dramatically reduced EGFinduced Gab2 tyrosine phosphorylation, we examined whether this resulted in a decrease of Gab2-dependent downstream signaling. Cells were infected with either control recombinant adenovirus (LacZ) or Gab2 constructs (WTGab2 or ⌬ProGab2). As previously observed, overexpression of WTGab2 potentiated EGF-induced PI3-kinase activation by 25-fold (Fig. 6A, top   panel, lane 2 versus lane 4), whereas overexpression of ⌬Pro-Gab2 reduced EGF-induced PI3-kinase activity by more than 70% (Fig. 6A, top panel, lane 4 versus lane 6). This parallels the association of Gab2 with p85 (Fig. 6A, middle panel). Interestingly, ⌬ProGab2 did not exert a dominant negative effect in that it did not reduce PI3 kinase activity to less than control levels despite being expressed at 20-fold the level of endogenous Gab2. Corresponding to the PI3-kinase activity, we found that EGF treatment augmented Akt serine 473 phosphorylation in cells overexpressing WTGab2 but not in those overexpressing ⌬ProGab2 (WTGab2-versus ⌬ProGab2-infected cells, expressed as a percentage of LacZ-infected cells was 146 Ϯ 4.7 versus 110 Ϯ 4.1 (n ϭ 3, mean Ϯ S.E., p Ͻ 0.01)) (Fig. 6B). Parallel results were obtained when we measured EGF-induced DNA synthesis (Fig. 6C). The observations in Fig. 6 are consistent with a greater sensitivity to EGF of PI3-kinase activation and p85 recruitment than to Akt phosphorylation and DNA synthesis. This could explain the closer correlation between Akt activation and the extent of stimulation of DNA synthesis. These findings indicate that the proline-rich sequences in Gab2, which mediate Src binding, are important for EGF-induced PI3-kinase activation and DNA synthesis.
We then examined the effect of WTGab2 and ⌬ProGab2 on ERK phosphorylation. As shown in Fig. 6D, ERK phosphorylation was barely observed with 0.2 ng/ml EGF in LacZ-infected cells, whereas, at this dose, there was clear phosphorylation of ERK in cells infected with WTGab2. We also demonstrated that ERK1/2 phosphorylation at higher doses of EGF (10 or 100 ng/ml) is much greater than that at 0.2 and 1.0 ng/ml. At these concentrations, we did not observe any effect of overexpressed Gab2 (data not shown). This suggests that overexpressed Gab2 increases the sensitivity of ERK to EGF, which correlated with previous studies showing that at a low dose of EGF (0.25 ng/ml), overexpression of Gab1 potentiated EGF-induced ERK activation in HEK293 cells (27). These results may indicate that, at higher doses of EGF, the influences of Gab2 could be obscured by input(s) from Gab2-independent signaling pathways. Alternatively, a high sensitivity of ERK phosphorylation to EGF may result in its saturation at the 10 ng/ml dose of FIG. 4. Constitutive association of Src with Gab2. A, serumdeprived hepatocytes were treated with (ϩ) or without (Ϫ) 100 ng/ml EGF for 1 min. Cell lysates were incubated with anti-Gab2 antibody, and the immunoprecipitated proteins were subjected to SDS-PAGE (7.5% gel) followed by immunoblotting with anti-Src antibody (top panel) or anti-Gab2 antibody (bottom panel). B and C, hepatocytes were infected with 10 MOI of either recombinant WTGab2 or ⌬ProGab2 adenovirus for 3 h followed by incubation in serum-free medium for 48 h. The hepatocytes were then treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 1 min, and cell lysates were prepared and incubated with anti-Gab2 antibody. The immunoprecipitated proteins were subjected to SDS-PAGE (7.  EGF. As shown in Fig. 6D, ⌬ProGab2 comparably potentiated EGF-induced ERK1/2 activation as WTGab2 at the low dose of EGF (i.e. following EGF (0.2 ng/ml) treatment, ERK activation in WTGab2 and ⌬ProGab2 was 218 Ϯ 4.2 and 236 Ϯ 12, respectively, when expressed as a percentage of values in LacZinfected cells (n ϭ 3, mean Ϯ S.E., p Ͻ 0.05 LacZ versus either construct)). Thus, the reduced level of tyrosine phosphorylation of ⌬ProGab2 would appear to be sufficient to activate ERK1/2 but not PI3-kinase.
SHP2 Is a Negative Effector for PI3-kinase/Akt Activation but Positive for MAPK Activation-Upon EGF treatment, Src activation leads to substantial tyrosine phosphorylation of Gab2 and activation of DNA synthesis. We previously demonstrated that phosphorylated Gab2 associates with the SH2-containing protein-tyrosine phosphatase, SHP2 (12). We considered that this binding might effect dephosphorylation of Gab2 and therefore generated a Gab2 mutant lacking the SHP2 binding sites (⌬SHP2Gab2; see "Experimental Procedures"). Overexpression of ⌬SHP2Gab2 in hepatocytes totally abolished SHP2 binding to Gab2 (Fig. 7A, top panel, lane 2 versus lane 4) and was tyrosine-phosphorylated (Fig. 7B, lane 4 versus lane 6) to an extent 25% greater than WTGab2 (Fig. 7C).
The above observations identify the mechanism by which Src regulates EGF-induced mitogenesis in hepatocytes. Upon EGF binding, Gab2-associated Src is activated, leading to activation of the PI3-kinase cascade and eventually DNA synthesis. This signal appears to be attenuated by SHP2 through dephosphorylation of Gab2. DISCUSSION In response to the binding of EGF to its receptor, Gab proteins become tyrosine-phosphorylated and bind SH2 domaincontaining proteins, including p85, the PI3-kinase subunit, and SHP2 (12,(27)(28)(29)(30). The mechanism by which EGF induces the phosphorylation of Gab docking proteins is key to understanding EGF-induced signaling. Several reports indicate that Hepatocytes were infected with 10 MOI of either recombinant LacZ, WTGab2, or ⌬ProGab2 adenovirus for 3 h followed by incubation in serum-free medium for 48 h. A, serum-deprived hepatocytes were treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 1 min, and cell lysates were prepared, equally divided, and incubated with anti-Gab2 antibody. Immunoprecipitated proteins were assayed for PI3-kinase activity (top panel) or subjected to SDS-PAGE (7.5% gel) followed by immunoblotting with anti-p85 antibody (middle panel). The membrane was stripped and reblotted with an anti-Gab2 antibody (bottom panel). B, serum-deprived hepatocytes were treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 5 min. Cell lysates were subjected to SDS-PAGE (10% gel) followed by immunoblotting with an anti-phospho-Akt473 (top panel), anti-Akt (middle panel), or anti-Gab2 (bottom panel) antibodies. C, virus-infected hepatocytes were incubated in serum-free medium for 24 h followed by the addition of 5 Ci of EGFR is the kinase that phosphorylates Gab1 (31)(32)(33). Furthermore, studies have demonstrated that the pleckstrin homology domain of Gab1 is required for its localization to membranes, thus facilitating its phosphorylation by the EGFR kinase (31,34,35). However, the process involved in phosphorylating Gab2 appears to be different. Thus, as we have previously found, Gab2 is phosphorylated in a pleckstrin homology domain-independent manner; nor was it observed to associate with the EGFR. Moreover, the major multimeric Gab2 complex was found exclusively in the cytosol of rat liver. 2 For these reasons, we examined the possibility that Gab2 is tyrosinephosphorylated by a cytosolic kinase and not the EGFR.
Although previous reports have established that Src family kinases are required for EGF-induced mitogenesis (6 -8), the mechanism by which Src family kinases act has remained unclear. In the present study, we demonstrate that EGF treatment of rat hepatocytes activates Src, leading to Gab2 tyrosine phosphorylation, after which Gab2 recruits SHP2, Grb2, and p85, thus forming a multimeric cytosolic complex. 2 The evidence that Src family kinases are involved in Gab2 tyrosine phosphorylation is based on the inhibitory effects of PP1 and PP2, two widely used specific inhibitors of Src family kinases (36,37). Thus, in the present study, we observed that the inhibition of Gab2 tyrosine phosphorylation was dose-dependent and occurred in the absence of inhibition of tyrosine phosphorylation of EGFR, ErbB3, or the IRS proteins (Fig. 1). Furthermore, using CSK Ϫ/Ϫ mouse embryonic cells, in which Src family kinases are constitutively activated, we confirmed a key role for these kinase(s) in the phosphorylation of Gab2. Consistent with our results are previous studies demonstrating that the target involved in Src-dependent mitogenic activity is a cytosolic molecule accessible to unmyristoylated Src. Thus, Src deletion mutants, lacking the amino-terminal one-third of the molecule, including the membrane binding domain, were shown to still induce cell proliferation (38). It has been known for some time that EGF signaling involves c-Src substrates with molecular sizes of 120 -130, 100, and 75 kDa (39). Further studies have identified p75 as cortactin (40) and p120 -130 as comprising several proteins, including p125FAK (41) and p130CAS (42). However, the identity of p100 has hitherto remained unknown. Our results appear to establish the p100 Src kinase substrate as Gab2.
To elucidate the mechanism by which Gab2 is phosphorylated by Src family kinases, we analyzed the association between these two proteins. In most cases, substrate recognition is dictated primarily by interactions with noncatalytic regions of the Src family kinases such as SH2 and SH3 domains (reviewed in Ref. 5). A constitutive association of Src family kinases with Gab2 was observed, and this association disappeared when the two key proline-rich regions on Gab2 were mutated (Fig. 4), indicating that the SH3 domain of Src is probably involved in this interaction. This is consistent with two previous reports demonstrating that the SH3 domain of Src is required for EGF mitogenic signaling (9,10) and that this domain can bind to Gab2 in an in vitro binding assay (43). The SH3 domains of Src family kinases recognize proline-rich sequences found in a large number of substrates of Src family kinases (44 -46). The SH2 domains of Src family kinase selectively recognize the sequence pYEEI, with a hydrophobic residue at position ϩ3 being an important determinant of binding (47). Interestingly, the Gab2 sequence does not contain a pY-EEI motif, suggesting that Src does not bind to Gab2 through its SH2 domain. Although Src family kinases are constitutively associated with Gab2, the activation of Src kinase is only observed after EGF treatment (Fig. 4C). The mechanism by which Src kinase activation is effected by the binding of EGF to the EGFR remains to be elucidated. In hepatocytes, Src, Fyn, and Yes are the three Src family members expressed (48). However, in the present analysis, we could not identify which isoform(s) phosphorylate Gab2. Indeed, the dose of PP1 we FIG. 7. SHP2 binding negatively regulates EGF-induced Gab2 phosphorylation. A, hepatocytes were infected with 10 MOI of either recombinant WTGab2 or ⌬SHP2Gab2 adenovirus for 3 h, followed by incubation in serum-free medium for 48 h. Cells were then treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 1 min, and cell lysates were incubated with either anti-hemagglutinin or anti-Myc antibody as described under "Experimental Procedures." Immunoprecipitated proteins were subjected to 7.5% SDS-PAGE followed by immunoblotting with anti-SHP2 (top panel), or anti-Gab2 (bottom panel) antibody. B, hepatocytes were infected with 10 MOI of either recombinant WTGab2 or ⌬SHP2Gab2 adenovirus for 3 h followed by incubation for 48 h in serum-free medium. The cells were treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 1 min. Cell lysates were incubated with anti-Gab2 antibody, and immunoprecipitated proteins were subjected to SDS-PAGE (7.5% gel) followed by immunoblotting with anti-phosphotyrosine antibody (top panel). The membrane was stripped and reblotted with anti-Gab2 antibody (bottom panel). The bar graph represents determination of Gab2 tyrosine phosphorylation in EGF-treated cells infected with ⌬SHP2Gab2 versus that in cells infected with WTGab2 (mean Ϯ S.E.; n ϭ 3; *, p Ͻ 0.005). used inhibits all three Src isoforms. Furthermore, the Src antibody we used for Western blots recognizes Fyn and Yes as well as Src.
In this paper, we show that in rat hepatocytes, inhibition of Src family kinases prevents EGF-induced phosphorylation of Gab2 and activation of PI3-kinase/Akt as well as DNA synthesis ( Figs. 1 and 3). Several recent studies have shown that activation of Src family kinases is involved in signal transduction by regulating PI3-kinase activity. Thus, erythropoietin appears to effect erythroid differentiation by effecting Src-dependent tyrosine phosphorylation of and the recruitment plus activation of PI3-kinase to the erythropoietin receptor (49). Also, in MCF-7 cells, estradiol was shown to promote cell cycle progression via a Src-dependent activation of PI3-kinase (50). Finally, in T47D cells, EGF activation of Akt was blocked in parallel with inhibition of Src activation (36). In rat hepatocytes, EGF-dependent Gab2 phosphorylation leads to downstream PI3-kinase/Akt activation and DNA synthesis and is effected by the activation of Src constitutively associated with Gab2 ( Figs. 4 and 6).
SHP2 binds with Gab family proteins through a consensus binding motif (YXX(V/I/L)) located in the C-terminal ends of Gab proteins (15). The function of the Gab-SHP2 interaction has been extensively studied using mutants of Gab family proteins lacking SHP2 binding sites. These studies reveal a general positive role of SHP2 in Gab-mediated signal transduction. For example, SHP2 association with Gab1 is required for Met-dependent morphogenesis (51) as well as EGF and lysophosphatidic acid-induced MAPK activation (27,30). It was also found that SHP2 association with Gab2 is essential for macrophage colony-stimulating factor-induced macrophage differentiation (52). In the present study, we identified a positive role for SHP2 in Gab2-potentiated EGFinduced MAPK activation (Fig. 8D). Recent reports have also identified a negative effect of SHP2 binding on Gab protein tyrosine phosphorylation. Thus, both Gab1 and Gab2 have been identified as substrates for SHP2 using an in vitro phosphatase assay (18). More recently, two studies have shown that inactive forms of SHP2 markedly increased EGFstimulated Gab1 tyrosine phosphorylation and PI3-kinase activity (28,53). The negative regulation of EGF-dependent PI3-kinase activation by SHP2 appears to be through the dephosphorylation of Gab1 p85 binding sites (53). In agreement with this study, we have found that, in rat hepatocytes, a relatively minor increase of ⌬SHP2Gab2 phosphorylation compared with WTGab2 (Fig. 7B) was accompanied by a more noticeable augmentation of PI3-kinase activity and DNA synthesis, suggesting that SHP2 is involved in the dephosphorylation of specific p85 binding sites on Gab2 (Fig. 8, A-C). Moreover, the fact that the ⌬SHP2Gab2 mutant blocks EGFinduced MAPK activation and augments PI3-kinase and DNA synthesis further confirms, in primary hepatocytes, the key role of PI3-kinase and not MAPK in EGF-induced DNA synthesis.
In summary, we have found that, in rat hepatocytes, Src family kinases regulate EGF-induced mitogenesis through association to and phosphorylation of the major cytosolic docking protein, Gab2. Furthermore, we demonstrate that the association of SHP2 with tyrosine-phosphorylated Gab2 leads to dephosphorylation of the latter and a corresponding decrease in EGF-induced DNA synthesis.
FIG. 8. SHP2 binding negatively affects EGF-induced PI3-kinase/Akt activation and DNA synthesis but positively affects MAPK activation. Hepatocytes were infected with 10 MOI of either recombinant LacZ, WTGab2, or ⌬SHP2Gab2 adenovirus for 3 h followed by incubation in serum-free medium for 48 h. A, serum-deprived hepatocytes were treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 1 min, and cell lysates were equally divided and incubated with anti-Gab2 antibody. The immunoprecipitated proteins were either assayed for PI3-kinase activity (top panel) or subjected to SDS-PAGE (7.5% gel) followed by immunoblotting with anti-p85 antibody (middle panel). The membrane was then stripped and reblotted with anti-Gab2 antibody (bottom panel). B, serum-deprived hepatocytes were treated with (ϩ) or without (Ϫ) EGF (100 ng/ml) for 5 min, and cell lysates were subjected to SDS-PAGE (10% gel) followed by immunoblotting with anti-phospho-Akt473 (upper panel), anti-Akt (middle panel), or anti-Gab2 (bottom panel) antibodies. C, virus-infected hepatocytes were incubated in serum-free medium for 24 h followed by the addition of 5 Ci of [ 3 H]methylthymidine without (hatched bars) or with (solid bars) EGF (100 ng/ml) for 18 h. Incorporation of [ 3 H]methylthymidine into DNA was determined as described under "Experimental Procedures." Results are expressed as -fold over control (LacZ adenovirus-infected cells without EGF). The difference between EGF-treated hepatocytes infected with WTGab2 versus ⌬SHP2Gab2 was determined in three separate experiments (mean Ϯ S.E.; *, p Ͻ 0.005). D, serum-deprived hepatocytes were treated with the indicated dose of EGF for 5 min, and cell lysates were subjected to SDS-PAGE (10% gel) followed by immunoblotting with anti-phospho-Erk1/2 (top panel), anti-Erk1/2 (middle panel), or anti-Gab2 (bottom panel) antibodies.