Stimulation of phosphatidylinositol 3-kinase by fibroblast growth factor receptors is mediated by coordinated recruitment of multiple docking

S. H. Ong*, Y. R. Hadari†, N. Gotoh†, G. R. Guy*, J. Schlessinger†‡, and I. Lax†

†Department of Pharmacology and The Skirball Institute, New York University, Medical School, New York, NY 10016; and *Institute of Molecular and Cell Biology, 30 Medical Drive, Singapore 117609

Contributed by J. Schlessinger, March 7, 2001 The docking FRS2 is a major downstream effector that links cific docking proteins such as IRS1 and Gab1, respectively fibroblast growth factor (FGF) and receptors (19–22). In EGF signaling, PI3-kinase is recruited by hetero- with the Ras͞mitogen-activated protein kinase signaling cascade. oligomerization with ErbB3 and by tyrosine phosphorylation of In this report, we demonstrate that FRS2 also plays a pivotal role Gab1; a docking protein that upon tyrosine phosphorylation in FGF-induced recruitment and activation of phosphatidylinositol recruits and activates PI3-kinase (18, 21, 23). Moreover, PI3- 3-kinase (PI3-kinase). We demonstrate that tyrosine phosphoryla- kinase activity recruited through Gab1 was shown to be essential tion of FRS2␣ leads to Grb2-mediated complex formation with the for NGF-induced protection from apoptosis induced by serum docking protein Gab1 and its tyrosine phosphorylation, resulting in deprivation, as well as for neurite outgrowth in PC12 cells the recruitment and activation of PI3-kinase. Furthermore, Grb2 (24–26). Gab1 was originally cloned as a Grb2-associated protein bound to tyrosine-phosphorylated FRS2 through its SH2 domain that is tyrosine phosphorylated in cells upon stimulation with interacts primarily via its carboxyl-terminal SH3 domain with a EGF or insulin (20). Gab1 possesses an N-terminal pleckstrin proline-rich region in Gab1 and via its amino-terminal SH3 domain homology (PH) domain, a Met-binding domain (MBD) that with the nucleotide exchange factor Sos1. Assembly of mediates interactions with the Met receptor and other receptor FRS2␣:Grb2:Gab1 complex induced by FGF stimulation results in tyrosine kinases, and a large C-terminal portion containing activation of PI3-kinase and downstream effector proteins such as multiple tyrosine residues (20, 27). Tyrosine-phosphorylated the S͞T kinase Akt, whose cellular localization and activity are Gab1 is capable of recruiting several SH2 domain-containing regulated by products of PI3-kinase. These experiments reveal a signaling molecules including Grb2, Shp2, PLC␥, Crk1, PI3- unique mechanism for generation of signal diversity by growth kinase, and Nck (20, 24, 28, 29). Recently, a highly homologous factor-induced coordinated assembly of a multidocking protein isoform of Gab1, termed Gab2, was isolated (30, 31). Unlike the complex that can activate the Ras͞mitogen-activated protein ki- broad expression pattern of Gab1, Gab2 is predominantly ex- nase cascade to induce cell proliferation and differentiation, and pressed in hematopoietic cells (30). Several reports have shown PI3-kinase to activate a mediator of a cell survival pathway. that the Gab family of docking proteins are involved in the activation of PI3-kinase in response to growth factors or cyto- ibroblast growth factors (FGFs) play key roles in diverse kines stimulation and by activation of the B cell antigen receptor Fcellular processes including mitogenesis, differentiation, mi- (20, 21, 24, 25, 27, 28, 30, 32–35). gration, and survival (reviewed in refs. 1 and 2). We have In this report, we show that in response to FGF stimulation, FRS2␣ and Gab1 associate indirectly via Grb2 resulting in recently identified a major downstream mediator of signaling tyrosine phosphorylation of Gab1 and activation of the PI3- through the activated FGF receptors (FGFRs), termed FRS2, kinase͞Akt survival pathway. These experiments reveal a mech- that was redesignated FRS2␣ when a highly homologous iso- anism for activation of multiple signaling pathways by coordi- form, FRS2␤, was identified (3, 4). The FRS2 proteins are nated assembly of docking proteins. targeted to the plasma membrane by myristylation at the N terminus and contain a phosphotyrosine-binding domain that Materials and Methods mediates direct interaction with FGF or nerve growth factor Antibodies. Antibodies for FGFR1, FRS2, Grb2, Shc, Sos1, and (NGF) receptors (4–8). The C-terminal region of FRS2 contains pTyr were previously described (3, 4, 36). Anti-Shp2 antibodies multiple tyrosine residues that are phosphorylated by activated were from Transduction Laboratories (Lexington, KY). Anti- FGF or NGF receptors, serving as recognition motifs for the Gab1 antibodies were from Upstate Biotechnology (Lake Placid, SH2 domains of Grb2 and Shp2 (3, 9). Recruitment of both Grb2 NY). Anti-Akt and anti-pS473-Akt antibodies were from New and Shp2 are required for FGF-mediated mitogen-activated England Biolabs. Anti-FLAG (M2) antibodies were from Sigma. protein kinase (MAPK) activation, proliferation of fibroblasts, and neuronal differentiation of PC12 cells (3, 9). Cell Lines. Swiss 3T3, NIH 3T3, 293, COS-1, SH-SY5Y, and Although FGFs and neurotrophins are strongly implicated in L6-FGFR1 cells were cultured in DMEM supplemented with cell survival, the intracellular signaling machinery and mecha- nisms involved are not clear (1, 10, 11). Numerous studies have provided evidence that the PI3-kinase͞Akt (or PKB) signaling Abbreviations: PI3-kinase, phosphatidylinositol 3-kinase; FGFR, fibroblast growth factor cascade activated by multiple growth factors and cytokines receptor; NGF, nerve growth factor; wt, wild type; MBD, Met-binding domain; PH, Pleck- results primarily in cell survival (12–14), as opposed to the strin homology; KD, kinase inactive mutant; MAPK, mitogen-activated protein kinase; ͞ PDGF, platelet-derived growth factor; EGF, epidermal growth factor; GST, glutathione Ras MAPK pathway, which mainly signals to control cell pro- S-transferase; SH2 and 3, Src homology 2 and 3; GDNF, glial-derived neurotrophic factor. liferation and differentiation (15, 16). Although some receptors ‡To whom reprint requests should be addressed. E-mail: [email protected]. such as the platelet-derived growth factor (PDGF) and ErbB3 The publication costs of this article were defrayed in part by page charge payment. This recruit PI3-kinase directly (17, 18), other receptors such as the article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. insulin and epidermal growth factor (EGF) receptors use spe- §1734 solely to indicate this fact.

6074–6079 ͉ PNAS ͉ May 22, 2001 ͉ vol. 98 ͉ no. 11 www.pnas.org͞cgi͞doi͞10.1073͞pnas.111114298 Downloaded by guest on September 24, 2021 10% FBS, 10 mM L-glutamine, and 100 ␮g each of penicillin and streptomycin͞ml, all from GIBCO͞BRL. PC12 cells were grown in DMEM supplemented with 10% FCS and 10% heat- inactivated horse serum (GIBCO͞BRL).

Expression Constructs. The FGFR1 expression vectors for wild- type (wt) or kinase-inactive (KD) mutant were previously de- scribed (36). FRS2␣ vectors for wt or the 6F-FRS2␣ mutant were described in refs. 3 and 4. Mouse Gab1 cDNA was kindly provided by W. Birchmeier (Max Delbru¨ck Center for Molecular Medicine, Berlin, Germany). Gab1 mutants consisting of the PH domain or the MBD or point mutations were generated by standard PCR and subcloning procedures. The Akt plasmid was from P. Cohen (University of Dundee, Dundee, U.K.). Transient transfection of cells was performed by using Lipofectamine (GIBCO͞BRL). Glutathione S-transferase (GST) fusion pro- teins of Grb2 (amino acids 1–217), Grb2-SH2 (amino acid 54–164), Grb2-N-SH3 (amino acid 1–68), Grb2-C-SH3 (amino acid 156–199), Gab1 (amino acids 1–695), and Gab1-MBD (amino acids 450–532) were expressed in Escherichia coli and purified with glutathione-conjugated agarose beads (Sigma) according to established procedures.

Immunoprecipitations, Protein Binding Studies, and PI3-Kinase As- says. The protocols for immunoprecipitations, binding studies Fig. 1. Complex formation between Grb2 and Gab1 in FGF-stimulated cells. with GST fusion proteins and in vitro PI3-kinase assays were as (A) Quiescent Swiss 3T3 cells were unstimulated or stimulated with FGF1 and described previously (3, 4, 17). heparin (100 ng͞ml and 5 ␮g͞ml, respectively) for 10 min. Lysates were immunoprecipitated with anti-Grb2 antibodies and followed by SDS–PAGE Results and immunoblotting with anti-pTyr antibodies. (B) Swiss 3T3 cells were un- stimulated or stimulated with FGF1 and heparin (100 ng͞ml and 5 ␮g͞ml, In a survey of tyrosine phosphoproteins involved in FGF sig- respectively) for 10 min. Cell lysates were incubated with immobilized GST naling, we have identified proteins that coimmunoprecipitate fusion proteins of the N-terminal SH3, SH2, or C-terminal SH3 domains of Grb2. with Grb2 upon FGF stimulation. Lysates from FGF-stimulated Bound proteins were eluted and resolved by SDS–PAGE and then followed by Swiss 3T3 fibroblasts were subjected to immunoprecipitation immunoblotting with anti-pTyr (Upper) or anti-Gab1 antibodies (Lower). (C) with anti-Grb2 antibodies followed by SDS–PAGE and immu- Gab1 binds to the C-terminal SH3 domain of Grb2 through the MBD. 293 cells noblotting with anti-pTyr antibodies. FGF induced the associa- were transfected with the expression vector for Flag-tagged MBD of Gab1. tion of multiple tyrosine-phosphorylated proteins with Grb2 The lysates were incubated with immobilized GST fusion proteins of the SH2, BIOCHEMISTRY (Fig. 1A). Four prominent tyrosine-phosphorylated proteins that N-SH3, or C-SH3 domains of Grb2. Bound proteins were eluted and resolved by SDS–PAGE and followed by immunoblotting with anti-Flag antibodies. were detected include FRS2␣, Shp2, Shc, and a previously undescribed broad band with apparent molecular mass of 120 kDa, termed here p120. To delineate the nature of interaction between p120 and Grb2, Gab1. To further characterize the proteins associating with Gab1 GST fusion proteins of the SH2 or SH3 domains of Grb2 were in FGFR signaling, 293 cells were transfected with expression used to precipitate p120 from lysates of quiescent or FGF- vector for wt FGFR1 resulting in autoactivation, or with expres- stimulated Swiss 3T3 fibroblasts. Bound proteins were eluted, sion vector for kinase-inactive mutant of FGFR1 (KD) as a resolved by SDS–PAGE, and probed by immunoblotting with control, together with expression vectors for FRS2␣, Shc, or specific antibodies. Although the SH2 domain of Grb2 predom- Shp2 as indicated (Fig. 2 A–C). Lysates from the transfected cells inantly precipitated tyrosine-phosphorylated FRS2, the N- and were subjected to pull-down experiments with GST fusion C-terminal SH3 domains of Grb2 apparently precipitated the protein of full-length Gab1 or the MBD. FRS2␣, Shc, and Shp2 p120 species (Fig. 1B Upper). Immunoblotting with specific were precipitated by both full-length Gab1 or the MBD when antibodies revealed that distinct components of the p120 species coexpressed with wt FGFR1 but not when coexpressed with interacted specifically with the N- and C-terminal SH3 domains FGFR1 (KD). When lysates from serum-starved or FGF- of Grb2. The p120 component that precipitated with the C- stimulated NIH 3T3 cells were used for a similar binding assay, terminal SH3 domain of Grb2 was identified by immunoblotting tyrosine-phosphorylated FRS2, Shc and Shp2 were similarly with anti-Gab1 antibodies (Fig. 1B Lower), whereas the p120 found to be precipitated by Gab1 or the MBD alone (data not component bound to the N-terminal SH3 domain of Grb2 shown). We have shown in previous reports that these three remained unidentified. proteins bind directly to the SH2 domain of Grb2 in FGF- Having established that the C-terminal SH3 domain of Grb2 stimulated cells (3, 9). By contrast, FGFR1 was not readily associated with Gab1, we next determined the region on Gab1 that mediates this interaction. A likely candidate was the proline- detected in a same complex with Gab1 or the MBD (data not rich region that resides in the MBD of Gab1 as proline-rich shown). Taken together, these data indicate that Grb2 serves as sequences had been shown to bind to SH3 domains (27). 293 cells an intermediary between Gab1 and FRS2, Shc and Shp2, by were transfected with Gab1-MBD tagged at the N terminus with binding to Gab1 directly through its C-SH3 domain and to the FLAG-epitope, termed Gab1(MBD)-FLAG. The lysate was phosphorylated FRS2, Shc, and Shp2 through its SH2 domain. incubated with immobilized GST fusion proteins of the SH2 or Indeed, in an immunoprecipitate of Gab1 from PC12 cells SH3 domains of Grb2, or with GST alone. Bound proteins were stimulated with FGF, Gab1, Grb2, FRS2, and Shp2 were all eluted, resolved by SDS–PAGE, and probed with anti-FLAG detectable in a same complex (Fig. 2D). Therefore, the inter- antibodies (Fig. 1C). This experiment showed that the C- action of docking proteins with the SH2 domain of Grb2 may be terminal SH3 domain of Grb2 binds specifically to the MBD of a general mechanism to target Grb2:Gab1 complexes to the

Ong et al. PNAS ͉ May 22, 2001 ͉ vol. 98 ͉ no. 11 ͉ 6075 Downloaded by guest on September 24, 2021 Fig. 3. Gab1 and the MBD of Gab1 form a complex with tyrosine- phosphorylated FRS2␣.(A) 293 cells were transfected with expression vectors for wt FGFR1 or kinase-inactive FGFR1 (KD) mutant and wt FRS2␣ or an FRS2␣ mutant in which six tyrosine phosphorylation sites were replaced by pheny- lalanines (6F) as indicated. Equivalent amounts of the total cell lysates were resolved and immunoblotted with anti-FGFR1 (i) or anti-FRS2 antibodies (ii). Alternatively, cell lysates were incubated with immobilized GST fusion pro- teins of the MBD of Gab1 (iii). Bound proteins were eluted and resolved by SDS–PAGE and then followed by immunoblotting with anti-FRS2 antibodies. (B) FRS2 and Gab1 form a complex with tyrosine-phosphorylated NGF or GDNF receptors. Quiescent PC12 cells were unstimulated or stimulated with NGF (100 ng͞ml, 10 min). Lysates were immunoprecipitated with anti-TrkA anti- bodies (Left). Similarly, lysates from unstimulated or GDNF-stimulated (50 ng͞ml, 10 min) SH-SY5Y neuroblastoma cells were immunoprecipitated with anti-Ret antibodies (Right). In both cases the immunoprecipitates were resolved by SDS–PAGE and followed by immunoblotting with anti-pTyr antibodies.

␣ Fig. 2. Gab1 and the MBD of Gab1 form a complex with FRS2 , Shc, or Shp2 that is not detectably tyrosine phosphorylated in cells upon FGF in cells expressing activated FGF receptor. 293 cells were transfected with ␣ expression vectors for kinase-inactive FGFR1 (KD) mutant, wt FGFR1, and stimulation (9). Because 6F-FRS2 is deficient in recruiting FRS2␣, Shc, or Shp2 (A–C) as indicated. Cell lysates were incubated with Grb2 and Shp2 in response to FGF stimulation, this mutant is immobilized GST fusion proteins of full-length Gab1 (G) or the MBD of Gab1 potentially incapable of recruiting Gab1. To test this hypothesis, (M). GST alone was used as a control (C). Bound proteins were eluted and FGFR1 (wt or kinase-inactive mutant) and FRS2␣ (wt or analyzed by SDS–PAGE and immunoblotting with FRS2, Grb2, Shc, Shp2 6F-FRS2␣) were transiently expressed in 293 cells. Cell lysates antibodies (A–C). (D) Quiescent PC12 cells were unstimulated or stimulated were resolved by SDS–PAGE and immunoblotted with anti- ͞ ␮ ͞ with FGF1 and heparin (100 ng ml and 5 g ml, respectively) or EGF (50 FGFR1 or anti-FRS2 antibodies to verify the expression levels ng͞ml) for 10 min. The lysates were immunoprecipitated with anti-Gab1 ␣ antibodies, resolved by SDS–PAGE, and followed by immunoblotting with of FGFR1 and FRS2 , or incubated with immobilized GST anti-pTyr, Gab1, Grb2, FRS2, or Shp2 antibodies. fusion protein of the MBD of Gab1. Bound proteins were eluted, resolved by SDS–PAGE, and probed for the presence of FRS2␣ or the 6F-FRS2␣ mutant by immunoblotting with anti-FRS2␣ plasma membrane for phosphorylation in response to activated antibodies (Fig. 3A). Although wt FRS2␣, when cotransfected growth factor or cytokine receptors. with wt FGFR1, coprecipitated with GST-Gab1-MBD, the 6F-FRS2␣ mutant did not. Similar results were obtained with Grb2 Mediates Complex Formation Between Gab1 and FRS2. In full-length GST-Gab1 whereas GST alone showed no binding response to FGF stimulation, FRS2 forms a complex with Grb2 (data not shown). This experiment demonstrated that the re- directly and indirectly via Shp2 (9, 40). Because Gab1 interacts cruitment of Grb2 to tyrosine-phosphorylated FRS2 mediates with FGFR by an indirect mechanism, we examined the possi- complex formation with Gab1, allowing the formation of a bility of whether the binding of Grb2 and Shp2 via their SH2 ternary FRS2:Grb2:Gab1 complex thus enabling tyrosine phos- domains to tyrosine-phosphorylated FRS2 enables the recruit- phorylation of Gab1 by FGFR. ment of Grb2:Gab1 complex in the vicinity of FGFR, resulting Activation of NGF and glial-derived neurotrophic factor in the tyrosine phosphorylation of Gab1. We have previously (GDNF) receptors also induce tyrosine phosphorylation of reported that FGF stimulation results in the phosphorylation of FRS2␣, resulting in the recruitment of Gab1 by means of Y196, Y306, Y349, and Y392 of FRS2␣, residues that function association to FRS2:Grb2 complex. As shown in Fig. 3B, acti- as binding sites for the SH2 domain of Grb2, Y436 and Y471, vated TrkA immunoprecipitated from NGF-stimulated PC12 residues that function as binding sites for the SH2 domains of cells (Left), and activated Ret immunoprecipitated from GDNF- Shp2 (3, 9). The substitution of these six tyrosine residues by stimulated SH-SY5Y neuroblastoma cells (Right), form a com- phenylalanine resulted in a mutant FRS2␣, termed 6F-FRS2␣, plex with tyrosine-phosphorylated Gab1, FRS2, Shc, and Shp2.

6076 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.111114298 Ong et al. Downloaded by guest on September 24, 2021 NGF- or EGF-induced recruitment of PI3-kinase by Gab1 (21, 24). To investigate whether 3F-Gab1 was also defective in recruiting PI3-kinase in FGF signaling, COS-1 cells transfected with either wt Gab1 or 3F-Gab1, were stimulated with FGF or left unstimulated. Gab1 was immunoprecipitated from the ly- sates with anti-FLAG antibodies and assayed for association of PI3-kinase activity. As shown in Fig. 4B, FGF induced the association of PI3-kinase activity with wt Gab1, but not with the 3F-Gab1 mutant, indicating that the YXXM motifs in Gab1 are indeed responsible for recruitment of PI3-kinase in response to FGF-stimulation. We further confirm that FRS2 is involved in the activation of PI3-kinase by showing that immunoprecipitates of FRS2 from PC12 cells stimulated with FGF contained sig- nificant levels of PI3-kinase activity above the control (Fig. 4C).

Gab1 and Sos1 Associate in a Same Complex with Grb2. We have demonstrated that Gab1 is bound to Grb2 via its C-SH3 domain, as was also shown by others recently (28, 41). This is in contrast to the nucleotide exchange factor Sos1, which binds predominantly to the N-terminal SH3 of Grb2 (42–44). It is therefore possible that a single Grb2 molecule may bind both Gab1 and Sos1 through its C- and N-SH3 domains, respectively. The formation of such a tertiary complex in cells upon growth factor stimulation may enable both Sos1 and Gab1 to be recruited simultaneously to the membrane through association with Grb2͞FRS2 complexes, to activate both the Ras͞MAPK and the PI3-kinase pathways. Alternatively, Sos1 and Gab1 may bind to different pools of Grb2 molecules, resulting in competition between the two molecules for the binding to Grb2. Fig. 4. FGF induces PI3-kinase activity associated with Gab1. (A) Quiescent We proceeded to examine the nature of the interaction among L6-myoblasts stably expressing FGFR1 were unstimulated or stimulated with Grb2, Sos1, and Gab1. Lysates from unstimulated or FGF- FGF1 and heparin (100 ng͞ml and 5 ␮g͞ml, respectively), insulin (100 ng͞ml), stimulated NIH 3T3 cells were immunoprecipitated with anti-Grb2 or PDGF (100 ng͞ml) for 10 min. The lysates were immunoprecipitated with or anti-Sos1 antibodies and followed by SDS–PAGE and immu- anti-Gab1 antibodies. One-half of the immunocomplex was subjected to an in noblotting with anti-Gab1, anti-Sos1, or anti-Grb2 antibodies (Fig. vitro PI3-kinase assay (Upper), and the second half was resolved by SDS–PAGE and immunoblotted with anti-pTyr antibodies (Lower). (B) COS-1 cells were 5A). Although similar amounts of Sos1 were coprecipitated with transfected with expression vectors for Gab1-Flag (wt) or for 3F mutant of Grb2 from lysates of unstimulated or FGF-stimulated cells, the level BIOCHEMISTRY Gab1-Flag, (3F). Lysates from unstimulated or FGF1 and heparin- (100 ng͞ml of Gab1 that coprecipitated with Grb2 or Sos1 was enhanced upon and 5 ␮g͞ml, respectively, 10 min) stimulated cells were immunoprecipitated FGF stimulation. Furthermore, the overexpression of Gab1 did not with anti-Flag antibodies, and the immunocomplexes were assayed for asso- affect the association of Sos1 with Grb2 (Fig. 5B). Similarly, the ciated PI3-kinase activity. (C) Quiescent PC12 cells were unstimulated or overexpression of Sos1 did not affect the association of Gab1 with stimulated with FGF1 and heparin (100 ng͞ml and 5 ␮g͞ml, respectively). The Grb2 (data not shown). We interpret these results as an indication lysates were immunoprecipitated with anti-pTyr, anti-Gab1, or anti-FRS2 that Sos1 and Gab1 associate with Grb2 simultaneously via the N- antibodies and the immunocomplexes assayed for associated PI3-kinase and C-SH3 of Grb2, respectively, allowing the coexistence of Sos1 activity. and Gab1 with Grb2 in a same complex. The increased association of Gab1 with Grb2 upon FGF stimulation, could be caused by Therefore it appears that NGF, GDNF, and FGF use a common additional molecules of Gab1 that are recruited through other mechanism for activation of these effector proteins through their proteins such as Shp2, which binds avidly to phosphorylated Gab1 interaction with the Grb2:FRS2 complex. through its SH2 domains (20, 28, 29), and is itself phosphorylated and capable of binding another population of Grb2͞Gab1 com- plexes. Alternatively, it could be that Gab1 is more highly concen- Gab1 Mediates FGF Stimulation of PI3-Kinase Activity. Tyrosine trated in the vicinity of the cell membrane because of additional phosphorylation of Gab1 induced by EGF, insulin, or NGF interactions with membrane components such as inositol phospho- stimulation has been reported to result in the recruitment of lipids via its PH domain. PI3-kinase, via the binding of the SH2 domains of PI3-kinase to specific pTyr sites on Gab1 (20, 21, 24, 25). To explore the Positive and Negative Regulation of Akt Activity by Overexpression of possibility that FGF-stimulation of PI3-kinase is mediated by wt or Mutant Gab1 and FRS2␣. The products of PI3-kinase have Gab1, lysates from FGF, insulin, or PDGF-stimulated L6 cells been shown to play an important role in the activation of were immunoprecipitated with anti-Gab1 antibodies followed by Akt͞PKB (45–47). We therefore tested whether FGF stimula- analysis of PI3-kinase activity in the immunocomplexes. As tion of cells would result in the activation of Akt as determined shown in Fig. 4A (Upper), FGF and PDGF stimulations result in by immunoblotting with anti-pS-Akt, antibodies that recognize enhancement of PI3-kinase activity in Gab1 immunocomplexes. specifically activated Akt. Lysates from quiescent PC12 or NIH By contrast, insulin stimulation did not result in stimulation of 3T3 cells that were unstimulated or stimulated with FGF were tyrosine phosphorylation of Gab1 and in enhancement of PI3- resolved by SDS–PAGE and then followed by immunoblotting kinase activity in Gab1 immunocomplexes (Fig. 4A Lower). with anti-Akt or anti-pS-Akt antibodies (Fig. 6A). We observed Gab1 contains three potential tyrosine residues (Y448, Y473, that FGF treatment induced significant activation of Akt, con- and Y590) with the consensus YXXM sequences that function sistent with the activation of PI3-kinase described earlier (Fig. as binding sites for the SH2 domain of the p85 subunit of 4). To investigate whether FRS2 and Gab1 play a role in Akt PI3-kinase. The mutation of these tyrosine residues to phenyl- activation, 293 cells were cotransfected with expression vectors alanine, generating the 3F-Gab1 mutant was shown to abolish for FGFR1, FRS2␣, Gab1, and Akt. The expression of FGFR1,

Ong et al. PNAS ͉ May 22, 2001 ͉ vol. 98 ͉ no. 11 ͉ 6077 Downloaded by guest on September 24, 2021 Fig. 5. Gab1 and Sos1 associate with Grb2 in a same complex. (A) Quiescent NIH 3T3 cells were unstimulated or stimulated with FGF1 and heparin (100 ng͞ml and 5 ␮g͞ml, respectively) for 5 min. The lysates were immunoprecipi- tated with Grb2 (i) or Sos1 (ii) antibodies and followed by SDS–PAGE and immunoblotting with Gab1, Sos1, or Grb2 antibodies. (B) The 293 cells were transfected with expression vectors for FGFR1, Sos1, and Gab1 as indicated. Equivalent amounts of total cell lysates were resolved by SDS–PAGE and immunoblotted with anti-FGFR1, Gab1, or Sos1 antibodies. Alternatively, the lysates were immunoprecipitated with anti-Grb2 and followed by SDS–PAGE and immunoblotting with anti-Sos1 or Grb2 antibodies. Fig. 6. FRS2 and Gab1 potentiate FGF-induced activation of Akt. (A) Quiescent ␣ PC12 or NIH 3T3 cells were unstimulated or stimulated with FGF1 and heparin (100 FRS2 , and Gab1 was confirmed by immunoprecipitation and ng͞ml and 5 ␮g͞ml, respectively) for 5 min. Equivalent amounts of total cell immunoblotting with antibodies against these proteins (Fig. 6B). lysates were resolved by SDS–PAGE and then followed by immunoblotting with Equivalent amounts of total cell lysates were resolved by SDS– anti-pS-Akt or anti-Akt antibodies. (B) The 293 cells were transfected with ex- PAGE and then followed by immunoblotting with anti-Akt or pression vectors for FGFR1, FRS2␣, Gab1 and Akt as indicated. The expression of anti-pS-Akt antibodies. This experiment demonstrated that the FGFR1, FRS2␣, and Gab1 was verified by immunoprecipitation with antibodies overexpression of FRS2␣ or Gab1 potentiates FGF-induced Akt specific to these proteins and followed by SDS–PAGE and immunoblotting with stimulation. We next examined whether overexpression of Gab1- the same antibodies. Equivalent amounts of total cell lysates were resolved by MBD, the region mediating the interaction of Gab1 with Grb2, SDS–PAGE and immunoblotted with anti-pS-Akt or anti-Akt antibodies. (C) The would exert a dominant inhibitory effect on FGF-induced Akt MBD of Gab1 inhibits Gab1-potentiated FGFR-induced activation of Akt. 293 cells activation. 293 cells were transfected with Akt alone or together were transfected with the expression vectors for FGFR1, Akt and Gab1-Flag, with FGFR1 and wt Gab1, 3F-Gab1, Gab1-MBD, or the PH Gab1(3F)-Flag, Gab1-MBD-Flag, or Gab1-PH-Flag as indicated. Equivalent domain of Gab1. FGFR1 was transfected at levels that induce amounts of total cell lysate were resolved by SDS–PAGE and followed by immu- noblotting with anti-FGFR1, anti-Flag, anti-Akt, or anti-pS-Akt antibodies. autoactivation, resulting in tyrosine phosphorylation of FRS2␣ and Gab1 and stimulation of PI3-kinase and Akt activities. Equivalent amounts of lysates from the transfected cells were resolved by SDS–PAGE and then followed by immunoblotting activated by growth factors, neurotrophic factors, and cytokines with anti-FGFR1, anti-FLAG, or anti-pS-Akt antibodies (Fig. (reviewed in refs. 14 and 26). Certain growth factor receptors 6C). Although overexpression of wt Gab1 results in enhance- (i.e., PDGF receptor and ErbB3) possess binding sites for direct ment of Akt activity, overexpression of 3F-Gab1, Gab1-MBD, or recruitment of PI3-kinase. Others, such as the insulin receptor, the PH domain decreased FGFR-induced stimulation of Akt. lack these sites and recruit PI3-kinase indirectly through asso- We propose that Gab1-MBD interferes with FGF-induced stim- ciation with IRS proteins; a family of docking proteins contain- ulation of Akt by preventing complex formation between Grb2 ing multiple tyrosine residues that are phosphorylated by acti- and endogenous Gab1; 3F-Gab1 interferes with recruitment vated insulin or insulin-like growth factor I receptors to recruit of PI3-kinases by wt Gab1. The overexpression of the PH do- SH2 domain-containing molecules, such as PI3-kinase and Grb2 main is likely to interfere with membrane translocation by com- peting with Gab1 for binding to phosphatidylinositol 3,4,5- (reviewed in ref. 19). Gab1 and Gab2 represent another family triphosphate. of docking proteins capable of recruiting PI3-kinase in response to growth factor or cytokine stimulation. It has been shown that Discussion activation of NGF, hepatocyte growth factor, or EGF receptors Numerous studies have implicated the PI3-kinase͞Akt-signaling results in tyrosine phosphorylation of Gab1 and its association pathway as a component of a cell survival pathway that is with PI3-kinase (20, 21, 24, 25, 27, 28). Similarly, IL-2 or IL-3

6078 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.111114298 Ong et al. Downloaded by guest on September 24, 2021 stimulation of hematopoietic cells leads to tyrosine phosphory- The ability of tyrosine-phosphorylated FRS2 to mediate FGF- lation of Gab2 and activation of PI3-kinase (30, 32). induced PI3-kinase activation via indirect recruitment of Gab1 Although FGF stimulation has been shown to play a role in cell enables FRS2 to activate many additional effector proteins whose survival (2, 10, 11), it has not been resolved whether the cellular localization and activity are regulated by the reaction ͞ PI3-kinase Akt pathway is involved in this process. Further- product of PI3-kinase. These will increase the repertoire of signal- more, the activated FGF receptor does not appear to recruit ing pathways that are recruited by FGF, NGF, and other growth PI3-kinase directly. In this report, we show that activation of factors that induce the tyrosine phosphorylation of FRS2␣. Be- FGF receptor induces PI3-kinase and Akt activities through the cause Shc and Shp2 are also tyrosine phosphorylated in cells upon recruitment and tyrosine phosphorylation of the docking protein FGF stimulation, leading to their association with the SH2 domain Gab1. Most interestingly, unlike the IRS-docking proteins that of Grb2, we propose that Shc and Shp2 may also contribute toward bind to the insulin receptor directly, the recruitment of Gab1 to the recruitment of the Grb2͞Gab1 complex thus indirectly activat- the FGF receptor occurs indirectly, mediated by the FRS2:Grb2 ing PI3-kinase and its downstream effector proteins. Indeed, it has complex. FGF stimulation results in tyrosine phosphorylation of Gab1 and activation of PI3-kinase. Activation of PI3-kinase is been recently reported that, in response to IL-2 or IL-3 stimulation mediated by binding of the SH2 domain of Grb2 to tyrosine- of hematopoeitic cells, tyrosine-phosphorylated Shc forms a ter- phosphorylated FRS2 and by binding of the C-SH3 domain of nary complex with Grb2 and Gab2, resulting in tyrosine phosphor- Grb2 to a proline-rich region in the MBD of Gab1. The assembly ylation of Gab2 and activation of PI3-kinase (32). These results of Gab1 in a complex with Grb2:FRS2 enables tyrosine phos- together with the experiments described in this report demonstrate phorylation of Gab1 by FGF receptor to generate binding sites that growth factor or cytokine receptors that do not recruit PI3- for the SH2 domain of p85, the regulatory subunit of PI3-kinase, kinase directly may use docking proteins such as FRS2 as a platform thus resulting in recruitment and activation of PI3-kinase. for recruitment of additional adaptors and docking proteins, which Of interest, both the guanine nucleotide releasing factor Sos1 do not interact with the receptors directly. The assembly of such and Gab1, which bind predominantly to the N- and C- terminal large multidocking͞adaptor protein complexes and the effectors SH3 domains of Grb2, respectively, coexist in a same complex that are bound to them may provide a mechanism for generation with Grb2. The formation of such a complex allows the concur- of signal diversity. rent recruitment of both Sos1 and Gab1 to the membrane by the binding of Grb2 to the membrane linked FRS2, eliciting bifur- This work was partially supported by the National Science and Tech- cating signals to activate the Ras͞MAPK and PI3-kinase͞Akt nology Board of Singapore and fellowships from the Human Frontier pathways simultaneously. Science Program Organization.

1. Barde, Y. A. (1989) Neuron 2, 1525–1534. 26. Kaplan, D. R. & Miller, F. D. (2000) Curr. Opin. Neurobiol. 10, 381–391. 2. Goldfarb, M. (1996) Cytokine Growth Factor Rev. 7, 311–325. 27. Weidner, K. M., Di Cesare, S., Sachs, M., Brinkmann, V., Behrens, J. & 3. Kouhara, H., Hadari, Y. R., Spivak-Kroizman, T., Schilling, J., Bar-Sagi, D., Birchmeier, W. (1996) Nature (London) 384, 173–176. Lax, I. & Schlessinger, J. (1997) Cell 89, 693–702. 28. Schaeper, U., Gehring, N. H., Fuchs, K. P., Sachs, M., Kempkes, B. & 4. Ong, S. H., Guy, G. R., Hadari, Y. R., Laks, S., Gotoh, N., Schlessinger, J. & Birchmeier, W. (2000) J. Cell Biol. 149, 1419–1432. Lax, I. (2000) Mol. Cell. Biol. 20, 979–989.

29. Maroun, C. R., Naujokas, M. A., Holgado-Madruga, M., Wong, A. J. & Park, BIOCHEMISTRY 5. Dhalluin, C., Yan, S. K., Plotnikova, O., Lee, W. K., Zeng, L., Kuti, M., M. (2000) Mol. Cell. Biol. 20, 8513–8525. Mujtaba, S., Goldfarb, P. M. & Zhou, M. M. (2000) Mol. Cell 6, 921–929. 30. Gu, H., Pratt, J. C., Burakoff, S. J. & Neel, B. G. (1998) Mol. Cell 2, 729–740. 273, 6. Xu, H., Lee, K. W. & Goldfarb, M. (1998) J. Biol. Chem. 17987–17990. 31. Zhao, C., Yu, D. H., Shen, R. & Feng, G. S. (1999) J. Biol. Chem. 274, 7. Meakin, S. O., MacDonald, J. I., Gryz, E. A., Kubu, C. J. & Verdi, J. M. (1999) 19649–19654. J. Biol. Chem. 274, 9861–9870. 32. Gu, H., Maeda, H., Moon, J. J., Lord, J. D., Yoakim, M., Nelson, B. H. & Neel, 8. Easton, J. B., Moody, N. M., Zhu, X. & Middlemas, D. S. (1999) J. Biol. Chem. B. G. (2000) Mol. Cell. Biol. 20, 7109–7120. 274, 11321–11327. 9. Hadari, Y. R., Kouhara, H., Lax, I. & Schlessinger, J. (1998) Mol. Cell. Biol. 18, 33. Ingham, R. J., Holgado-Madruga, M., Siu, C., Wong, A. J. & Gold, M. R. (1998) 3966–3973. J. Biol. Chem. 273, 30630–30637. 10. Chow, R. L., Roux, G. D., Roghani, M., Palmer, M. A., Rifkin, D. B., 34. Sachs, M., Brohmann, H., Zechner, D., Muller, T., Hulsken, J., Walther, I., Moscatelli, D. A. & Lang, R. A. (1995) Development (Cambridge, U.K.) 121, Schaeper, U., Birchmeier, C. & Birchmeier, W. (2000) J. Cell Biol. 150, 4383–4393. 1375–1384. 11. Desire, L., Courtois, Y. & Jeanny, J. C. (2000) J. Neurochem. 75, 151–163. 35. Hibi, M. & Hirano, T. (2000) Leuk. Lymphoma 37, 299–307. 12. Brunet, A., Bonni, A., Zigmond, M. J., Lin, M. Z., Juo, P., Hu, L. S., Anderson, 36. Mohammadi, M., Dionne, C. A., Li, W., Li, N., Spivak, T., Honegger, A. M., M. J., Arden, K. C., Blenis, J. & Greenberg, M. E. (1999) Cell 96, 857–868. Jaye, M. & Schlessinger, J. (1992) Nature (London) 358, 681–684. 13. Datta, S. R., Dudek, H., Tao, X., Masters, S., Fu, H., Gotoh, Y. & Greenberg, 37. Li, W., Nishimura, R., Kashishian, A., Batzer, A. G., Kim, W. J., Cooper, J. A. M. E. (1997) Cell 91, 231–241. & Schlessinger, J. (1994) Mol. Cell. Biol. 14, 509–517. 14. Datta, S. R., Brunet, A. & Greenberg, M. E. (1999) Dev. 13, 2905–2927. 38. Bazenet, C. E., Gelderloos, J. A. & Kazlauskas, A. (1996) Mol. Cell. Biol. 16, 15. Cowley, S., Paterson, H., Kemp, P. & Marshall, C. J. (1994) Cell 77, 841–852. 6926–6936. 16. Marshall, C. J. (1995) Cell 80, 179–185. 39. Benjamin, C. W. & Jones, D. A. (1994) J. Biol. Chem. 269, 30911–30916. 17. Hu, P., Margolis, B., Skolnik, E. Y., Lammers, R., Ullrich, A. & Schlessinger, 40. Ong, S. H., Lim, Y. P., Low, B. C. & Guy, G. R. (1997) Biochem. Biophys. Res. J. (1992) Mol. Cell. Biol. 12, 981–990. Commun. 238, 261–266. 18. Kim, H. H., Sierke, S. L. & Koland, J. G. (1994) J. Biol. Chem. 269, 41. Lock, L. S., Royal, I., Naujokas, M. A. & Park, M. (2000) J. Biol. Chem. 275, 24747–24755. 31536–31545. 19. Yenush, L. & White, M. F. (1997) BioEssays 19, 491–500. 42. Vidal, M., Montiel, J. L., Cussac, D., Cornille, F., Duchesne, M., Parker, F., 20. Holgado-Madruga, M., Emlet, D. R., Moscatello, D. K., Godwin, A. K. & Tocque, B., Roques, B. P. & Garbay, C. (1998) J. Biol. Chem. 273, 5343–5348. Wong, A. J. (1996) Nature (London) 379, 560–564. 43. Wittekind, M., Mapelli, C., Lee, V., Goldfarb, V., Friedrichs, M. S., Meyers, 21. Rodrigues, G. A., Falasca, M., Zhang, Z., Ong, S. H. & Schlessinger, J. (2000) C. A. & Mueller, L. (1997) J. Mol. Biol. 267, 933–952. Mol. Cell. Biol. 20, 1448–1459. 22. Cunnick, J. M., Dorsey, J. F., Munoz-Antonia, T., Mei, L. & Wu, J. (2000) 44. Goudreau, N., Cornille, F., Duchesne, M., Parker, F., Tocque, B., Garbay, C. J. Biol. Chem. 275, 13842–13848. & Roques, B. P. (1994) Nat. Struct. Biol. 1, 898–907. 23. Kim, H. H., Vijapurkar, U., Hellyer, N. J., Bravo, D. & Koland, J. G. (1998) 45. Alessi, D. R., Andjelkovic, M., Caudwell, B., Cron, P., Morrice, N., Cohen, P. Biochem. J. 334, 189–195. & Hemmings, B. A. (1996) EMBO J. 15, 6541–6551. 24. Holgado-Madruga, M., Moscatello, D. K., Emlet, D. R., Dieterich, R. & Wong, 46. Franke, T. F., Kaplan, D. R., Cantley, L. C. & Toker, A. (1997) Science 275, A. J. (1997) Proc. Natl. Acad. Sci. USA 94, 12419–12424. 665–668. 25. Korhonen, J. M., Said, F. A., Wong, A. J. & Kaplan, D. R. (1999) J. Biol. Chem. 47. Frech, M., Andjelkovic, M., Ingley, E., Reddy, K. K., Falck, J. R. & Hemmings, 274, 37307–37314. B. A. (1997) J. Biol. Chem. 272, 8474–8481.

Ong et al. PNAS ͉ May 22, 2001 ͉ vol. 98 ͉ no. 11 ͉ 6079 Downloaded by guest on September 24, 2021