Gab1 Coupling to the HGF/Met Receptor Multifunctional Docking Site Requires Binding of Grb2 and Correlates with the Transforming Potential

Gab1 coupling to the HGF/Met receptor multifunctional docking site requires binding of Grb2 and correlates with the transforming potential

Alberto Bardelli, Paola Longati, Daniela Gramaglia, Maria C Stella and Paolo M Comoglio

Institute for Cancer Research and Treatment (IRCC), University of Torino School of Medicine 10060, Candiolo, Italy

Activation of the HGF receptor, encoded by the c-MET located on chromosome 7. In Tpr-Met, the extra- protooncogene (Met receptor), triggers motility, matrix- cellular domain of Met is replaced by Tpr sequences, invasion and branching morphogenesis in epithelial cells. which provide two strong dimerization motifs (Ro- It has recently been shown that the Met receptor interacts drigues and Park, 1993). Dimerization causes constitu- with Gab-1, an IRS-like adaptor protein, via the docking tive activation of the Met kinase, which acquires site (Y1349VHVNATY1356VNV) known to bind Grb2 and transforming, invasive and metastatic potential (Park multiple SH2-containing signal transducers. Here we et al., 1986; Giordano et al., 1997). show that Gab1 is the major phosphorylation-substrate of Activation of the Met receptor triggers a number of the Met receptor and of its oncogenic variant Tpr-Met.A signaling pathways in target cells. Met signaling is series of point mutations in the docking site established a mediated by a multifunctional docking site comprising direct correlation between the ability to recruit and two tandem arranged phosphotyrosines within the phosphorylate Gab1 and the transforming potential. sequence Y1349VHVNATY1356VNV (Ponzetto et al., Interestingly, the mutations of either Y1356 or N1358 1994). Upon autophosphorylation, these tyrosines abolished the binding of both Grb2 and Gab1 in intact bind and concomitantly activate multiple SH2-contain- cells. Furthermore, peptides designed to block either the ing transducers, including the Grb2/SOS complex, SH2 or the SH3 domains of Grb2 interfered with the phosphatidylinositol 3-kinase (PI 3-kinase), and phos- receptor-Gab1 interaction. These data indicate that Gab1 pholipase C-g (PLC-g) (Graziani et al., 1991, 1993). coupling to the Met receptor requires binding of Grb2 and The multifunctional docking site is an absolute correlates with the transforming potential of Tpr-Met. requirement for the receptor's transforming, scattering, invasive and metastatic ability (Weidner et al., Keywords: HGF/MET-receptor; Gab1; transformation 1995; Ponzetto et al., 1996; Giordano et al., 1997). Using a two-hybrid approach, the HGF receptor has recently been shown to interact with Gab-1 (Weidner et al., 1996). Gab1 is an IRS/DOS-like multi-adaptor Introduction protein originally cloned according to its ability to bind Grb2 (Holgado-Madruga et al., 1996). Gab-1 is Hepatocyte Growth Factor (HGF) is a mesenchymal phosphorylated by both the EGF and the Insulin cytokine capable of inducing proliferation, motility receptors in a ligand dependent fashion. The primary (scattering) and branching morphogenesis in epithelial sequence of Gab1 indicates that it contains a PH cells (Zarnegar et al., 1995). These biological responses domain, a number of consensus binding motifs for depend on the interaction of HGF with its receptor, SH2-containing molecules and a proline-rich region the Met tyrosine kinase. The MET oncogene is which includes two optimal binding sites for the SH3 overexpressed and ampli®ed in a signi®cant propor- domains of Grb2. In the yeast system, the Gab1/Met tion of human tumors (Bardelli et al., 1997). interaction is mediated by the proline-rich region of Interestingly, protein overexpression was found to be Gab1 (Met Binding Domain) and requires phosphor- associated with ampli®cation of the MET gene in only ylation of the Met multifunctional docking site. few primary carcinomas, but in a signi®cant proportion Overexpression of Gab1 in epithelial cells results in of the metastases examined (Di Renzo et al., 1995). constitutive cell-scattering and branching morphogen- Recently, the mapping of the papillary renal carcino- esis (Weidner et al., 1996). ma-associated gene to a region encompassing the met In this work we have investigated the role of Gab1 locus and the subsequent identi®cation of mutations in in Met-mediated transformation and the molecular the MET tyrosine kinase domain of patients aected mechanisms of the Met/Gab1 interaction. We found by this disease, have established a direct genetic that Gab1 plays a central role in Met-signaling leading connection between activation of this receptor and to cellular transformation and that its recruitment the development of renal cancers (Schmidt et al., 1997). requires Grb2. MET was originally identi®ed as a transforming sequence activated by chromosomal rearrangement induced by a chemical carcinogen (Cooper et al., Results 1984; Park et al., 1986). The resulting oncogene (TPR- MET) is a hybrid between 5' sequences of TPR, Gab1 is the major phosphorylation-substrate of the derived from chromosome 1, and 3' sequences of MET, activated Met-receptor

Correspondence: A Bardelli Signaling by the Met receptor and its oncogenic Received 6 August 1997; revised 10 September 1997; accepted 11 counterpart Tpr-Met is mediated by the `multifunc- September 1997 tional docking site' Y1349VHVNATY1356VNV located in Met binding to Gab1 requires Grb2 ABardelliet al 3104 the receptor C-terminal tail. Substitution of the two (Figure 1c). Recently, it has been shown that the Met tyrosines by phenylalanine (Table 1, Tpr-MetY1349 ± 1356F receptor interacts with and phosphorylates Gab-1, a mutant) completely abolishes its transforming potential 110 kDa IRS-like multiadaptor protein (Weidner et al., without aecting the kinase activity (Ponzetto et al., 1996). Therefore, we evaluated the possibility that the 1994). To identify molecules implicated in signaling 110kDa phosphoprotein could be Gab1. We found that through the Met-multifunctional docking site, we have the 110 kDa phosphoprotein is recognized by anti-Gab1 investigated the substrates that associate with and are speci®c antibodies (Figure 1b). We next evaluated the phosphorylated by Tpr-Metwt but not by its signaling- ability of the Met-receptor to induce Gab1 phosphor- inactive counterpart Tpr-MetY1349 ± 1356F. The cDNAs ylation upon ligand stimulation. We found that HGF encoding TPR-METwt and TPR-METY1349 ± 1356F were stimulation of epithelial cells expressing the Met transfected into ®broblasts or epithelial cells and, receptor induces rapid and marked phosphorylation of following cell lysis, the receptor proteins were Gab1 on tyrosine residues which results in decreased immunoprecipitated with Met-speci®c antibodies. The gel-mobility (Figure 1d). These experiments show that proteins associated to the immunoprecipitated Tpr- Gab1 represents the major substrate of both the full- Metwt and Tpr-MetY1349 ± 1356F were visualized by kinase length and the oncogenic forms of the receptor. assays. These analyses clearly demonstrate that a 110 kDa phosphoprotein is the major tyrosine-phos- The Gab1 interaction is abrogated by point mutations phorylation substrate of the activated receptor (Figure which selectively uncouple Grb2 from the Met-receptor 1a). This molecule binds and is phosphorylated only by the Tpr-Metwt and not by the signaling-inactive mutant In the yeast system, the Met/Gab1 interaction is (Tpr-MetY1349 ± 1356F) suggesting its possible involvement mediated by the Met Binding Domain (MBD) of in receptor signal transduction. The 110 kDa protein is Gab1 and requires phosphorylation of the Met speci®cally phosphorylated by Tpr-Met on tyrosine multifunctional docking site (Weidner et al., 1996). residues as shown by phosphoaminoacid analysis We have studied the molecular mechanism of the Met/ Gab1 interaction in mammalian cells using a number of mutants of the receptor multifunctional docking site Table 1 Point mutations in the Met-multifunctional docking site (Table 1). We have previously shown that PI3-kinase 1349 1356 WT YVHVNATYVNV and PLC-g bind to both tyrosine Y and Y and Y1349F FVHVNATYVNV their interaction is abrogated by replacement of the Y1356F YVHVNATFVNV two tyrosines with phenylalanines (Ponzetto et al., Y1349 ± 1356F FVHVNATFVNV 1993, 1994). Grb2 has a strong requirement for an N1358H YVHVNATYVHV H1351N YVNVNATYVNV asparagine in the +2 position and speci®cally interacts with Y1356. Two other mutants, designed to interfere Signaling by the Met receptor is mediated by the sequence Y1349VHVNATY1356VNV which interacts with SH2-containing with the Met/Grb2 interaction, were included in the transducers. Point mutations in the multifunctional docking site experiments: to abrogate Grb-2 binding without were generated by site-directed mutagenesis. Mutated residues are interfering with the other signaling pathways, aspar- underlined agine 1358 (which confers Grb2 speci®city to

a b d Ip anti-Gab1 wt Y1349-1356F wt Y1349-1356F Ctr antibody IP anti-Gab1 193 kD — Met — p110Gab1 p145 —112 kD Gab1 p110Gab1 p110 112 kD — p110Gab1 — — 86 kD Gab-1 — Tpr-Met Blot anti-PTyr 70 kD p65Tpr-Met

— p65 P110 — c p65Tpr-Met 57 kD P-Ser 112 kD — — p110Gab1 BOSC P-Thr P-Tyr

FR Blot anti-Gab1 Figure 1 A 110 kDa protein, indistinguishable from the multiadaptor Gab1, is the major substrate of the activated Met receptor (a) cDNAs encoding wild type (wt) and Y1349 ± 1356F TPR-MET mutants were transfected in Fisher Rat ®broblasts and BOSC epithelial cells. Cell lysates were immunoprecipitated with anti-Met antibodies. After extensive washing, kinase assay was performed in the presence of [32P-g] ATP. The position of the p65Tpr ± Met and p110Gab1 phosphoproteins is indicated. (b) Tpr-MetWt- associated proteins were subjected to a kinase assay as above, eluted and re-immunoprecipitated with control (Ctr) or anti-Gab1 antibodies. (c) Phosphoaminoacid analysis of the p110Gab1 and p65Tpr ± Met phosphoproteins. The positions of standard phosphoaminoacids (P-Ser, P-Thr, P-Tyr) are indicated. (d) Lysates from A549 cells treated (+) or not (7) with 120 ng/ml HGF were immunoprecipitated with anti-Gab1 antibodies, subjected to SDS ± PAGE and blotted with anti-phosphotyrosine antibodies. Two major phosphorylated proteins corresponding to the Met receptor (145Met) and Gab1 (p110Gab1) are present upon HGF stimulation. As a control the same ®lter-blot was probed with anti-Gab1 antibody Met binding to Gab1 requires Grb2 ABardelliet al 3105 tyrosine 1356) was substituted with a histidine (Tpr- proline-rich regions. Peptides blocking either the SH2 MetN1358H); and a mutant with an additional Grb-2 or the SH3 domain of Grb2 were used to directly binding was obtained by replacing histidine 1351 with assess the role of Grb2 in the Met/Gab1 interaction an asparagine (Tpr-MetH1351N). In addition, a kinase- (Table 2). The eects of the peptides on binding of the inactive mutant (Tpr-MetK1110A) was also included as a Met receptor to the isolated MBD domain of Gab1 negative control. Wild-type and receptor mutants were were initially evaluated using GTL-16 cells. In these transiently expressed in epithelial cells and immuno- cells the Met-receptor is phosphorylated on tyrosine precipitated with anti-Met antibodies. The amount of and constitutively binds Grb2 (Ponzetto et al., 1994). endogenous Gab1 associated with the various mutants Cell lysates were preincubated with peptides blocking was assessed using an anti-Gab1 antibody (Figure 2a). The amount of associated Grb2 was also evaluated in a similar experiment using anti-Grb2 antibodies (Figure 2b). We found that Tpr-Metwt but not the kinase- a inactive mutant (Tpr-MetK1110A) interacts with Gab1 indicating that the association is phosphorylation- dependent. Mutation of Y1349 has no eects on

Gab1 binding. Binding is abolished when Y1356 alone wt K1110A Y1349F Y1356F Y1349-1356F N1358H H1351N or in combination with Y1349 is replaced by phenylalanine. Interestingly, the N1358H mutation 193 kD — (Tpr-MetN1358H), which selectively impairs the Met/ Grb2 interaction, also dramatically aects the Met/ Gab1 interaction (Figure 2a and 2b). Binding is 112 kD — p110GAb1 unaected by introduction of a second Grb2 binding site (Tpr-MetH1351N). These data show that point 86 kD — mutations which selectively uncouple Grb2 from the receptor also abrogate binding of Gab1 and suggest 70 kD — that Grb2 acts as an adaptor for Gab1 in the Met signaling system. In the yeast system a proline-rich region of Gab1, Blot anti-Gab1 referred to as the Met Binding Domain (MBD), is both necessary and sucient to mediate coupling with the b Met receptor (Weidner et al., 1996). We have 29 kD — investigated the speci®city of this interaction in Grb2 mammalian cells using the MBD domain expressed as p24 a GST fusion protein and the various Tpr-Met mutants 20 kD — transiently transfected in BOSC cells. GST-MBD immobilized on glutathione-Sepharose was used to Blot anti-Grb2 bind the mutants from cell lysates. The amount of bound receptor was evaluated by Western blotting with c anti-Met antibodies. In agreement with our observa- tions using full-size Gab1, we found that the 70 kD — association is phosphorylation-dependent and is p65Tpr-Met 1349 1356 abrogated when both tyrosines Y and Y are 57 kD — replaced by phenylalanines (Figure 3). Furthermore, the N1358H mutation, which selectively impairs the Met/Grb2 association, also reduces the Met/MBD interaction. Binding is slightly increased by introduc- Blot anti-Met H1351N tion of a second Grb2 binding site (Tpr-Met ). d Interestingly mutation of Y1349, which does not aect interaction of Met with full-size Gab1, reduces binding 70 kD — to the isolated MBD region. These data indicate that p65Tpr-Met Met/Gab1 binding is mediated by the MBD region of Gab1 which can potentially interact with both Y1349 57 kD — and Y1356 and that high eciency Met-MBD interaction requires receptor coupling to Grb2. The Grb2-MBD interaction is likely to occur between the Grb2 SH3 Blot anti-PTyr domains and the proline-rich motifs (PXXPXR) Figure 2 Met-coupling to Gab1 requires a functional Grb2 located within the MBD domain of Gab1. binding site. The cDNAs encoding wild-type and mutated Tpr- Met proteins (Table 1) were transfected in BOSC cells. After immunoprecipitation with anti-Met antibodies, samples were Peptides blocking either the SH2 or the SH3 domains of analysed by SDS ± PAGE followed by Western immunoblotting Grb2 interfere with Met/Gab1 interaction with anti-Gab1 (a) or anti Grb2 (b) antibodies. (c and d) show control Western blots performed with anti-Met and anti- Grb2 is a modular adaptor protein consisting of an phosphotyrosine antibodies respectively. Tpr ± Met proteins are present in similar amounts and phosphorylated at comparable SH2 domain ¯anked by two SH3 domains. While the levels with the exception of the kinase-inactive mutant SH2 domain is responsible for binding phosphorylated (Tpr-MetK1110A). The positions of the p110Gab1, p24Grb2 and tyrosines the SH3 domains mediate association with p65Tpr ± Met proteins are indicated Met binding to Gab1 requires Grb2 ABardelliet al 3106 either the SH2 or the SH3 domains of Grb2, and an receptor was immunoprecipitated from cell lysates in unrelated peptide was used as a control. The MBD the presence of the indicated SH2 or SH3 competing domain of Gab1 was immobilized on glutathione- peptides and the amount of associated Gab1 was Sepharose and then allowed to associate with cell evaluated by Western immunoblotting (Figure 4b). In lysates containing the Met receptor in the presence of agreement with the data obtained using the isolated the SH2 speci®c peptides. The amount of MBD-bound MBD region, we found that in vivo the interaction receptor was evaluated by Western immunoblotting between Met and full size Gab1 can be abolished by (Figure 4a). We found that the peptides which peptides blocking either the SH2 or the SH3 domains uncouple Grb2 from the receptor (Y1356 and Y1349 ± 1356, of Grb2. Table 2) also reduce binding of the receptor to the MBD region of Gab1. A similar experiment was The interaction between the isolated Gab1/MBD domain performed to evaluate the eect of blocking the SH3 and the Met-multifunctional docking site is mediated by domains of Grb2 on the interaction of Met with the Grb2 MBD domain Gab1. Lysates of cells expressing constitutively phosphorylated HGF receptor were In the yeast model the MBD domain of Gab1 preincubated with a SOS proline-rich peptide which appears to bind directly to the Met receptor on binds the Grb2 SH3 domains. The immobilized phosphorylation of tyrosines 1349 and 1356 in the recombinant MBD domain of Gab1 was then allowed multifunctional docking site (Weidner et al., 1996). We to associate with the lysates and the ability of the used real-time BIAcore-analysis to investigate the receptor to interact with the MBD region of Gab1 in mechanism(s) of interaction between the isolated the presence of the SH3 speci®c peptides was MBD domain and the phosphorylated peptide evaluated by Western immunoblotting (Figure 4a). (IGEHpY1349VHVNATpY1356VNVK) corresponding to As for our observation with the SH2 domain, we the multifunctional docking site of the Met-receptor. found that the SH3 domains of Grb2 play a role in Puri®ed fusion proteins (GST alone, GST-MBD or the interaction between the Met receptor and the GST-Grb2) were injected over the immobilized isolated MBD region of Gab1. To prove that the phosphopeptide and association was measured over a same occurs in intact cells we took advantage of the range of protein concentrations. While the Grb2- fact that in GTL16 cells a stable complex exists phosphopeptide interaction was readily detected no between the Met-receptor and Gab1 (Figure 4b). The direct binding was observed with the MBD domain (Figure 5). The MBD region of Gab1 contains two proline rich motifs which match the consensus sequence for binding the SH3 domains of Grb2. Therefore, in the a Met signaling system, Grb2 can potentially act as an adaptor for Gab1 by binding the receptor via the SH2 domain and the MBD region of Gab1 through the SH3 domains (Figure 8). To test this hypothesis we wt K1110A Y1349F Y1356F Y1349-1356F N1358H H1351N performed ordered sequential binding experiments. Puri®ed Grb2 was allowed to bind the immobilized Met-phosphopeptide as described above; when the p65Tpr-Met interaction reached the equilibrium, the MBD domain was allowed to interact with the Grb2/Met-phosphopeptide complex. Fig.5 shows that Grb2 directly binds to the Met multifunctional docking site and can sequentially interact with Gab1. Taken together these experiments demonstrate that Gab1 can be indirectly Blot anti-Met recruited to the receptor via Grb2 (Figure 8).

b Binding and phosphorylation of Gab1 correlates with the transforming potential of Tpr-Met Signaling by the Met-receptor is mediated by the multifunctional docking site (Ponzetto et al., 1994). Tpr-Met p65 Mutations within this region dierentially aect the transforming potential of Tpr-Met, the oncogenic

Blot anti-Met Table 2 Peptides binding the SH2 and SH3 domains of Grb2 Figure 3 Speci®city of the interaction between the MBD domain Met Y1349 peptide EYH*VHVN of Gab1 and Tpr-Met mutants. (a) GST-MBD fusion protein Met Y1356 peptide ATY*VNVK immobilized on glutathione-Sepharose was incubated with lysates Met Y1349 ± 1356 peptide Y*VHVNATY*VNVK of BOSC cells transiently transfected with the wt and mutated SOS peptide SKGTDEVPVPPPVPPRR TPR-MET cDNAs. Associated Tpr-Met proteins were detected Control peptide YVHVNATYVNV by western blotting with anti-Met antibodies. (b) The amount of Tpr-Met proteins present in the lysates was evaluated by SH2-speci®c phosphopeptides are derived from the Met-multifunc- immunoprecipitation with anti-Met antibodies followed by tional docking site. SH3-speci®c peptide corresponds to the SOS Western blotting. The position of the p65Tpr-Met protein is prolin-rich sequence (the SH3-binding motif PXXPXR is underlined). indicated Y*, phosphorylated tyrosine Met binding to Gab1 requires Grb2 ABardelliet al 3107

a peptide peptide peptide 1349 1356 1349-1356 No peptide Y Y Y SOS peptide Control peptide

p145Met

Blot anti-Met

b peptide Figure 5 Grb2 directly binds to the Met-multifunctional docking 1356 site and can sequentially interact with the MBD domain of Gab1. Y Binding of Grb2 and of the MBD domain to the Met

Met SOS peptide multifunctional docking site was performed using a BIAcore Biosensor. The indicated GST-fusion proteins (GST, GST-MBD, M M GST-Grb2) were injected, either singularly or according to the M µ M µ µ µ indicated order, over the immobilized Met phosphopeptide (IGEHpY1349VHVNATpY1356VNVK). Binding was measured No peptide 50 100

50 100 Control peptide over a range of protein concentrations. Relative binding is expressed as Resonance Units (RU)

p110Gab1

that the ability to recruit and phosphorylate Gab1 Blot anti-Gab1 correlates with the transforming potential of the mutants. In particular, wild-type Tpr-Met protein as well as the other transforming mutants associates with, and eciently phosphorylates, Gab1. Conver- p145Met sely, receptor mutants with reduced or absent transforming potential do not interact with Gab1. Moreover, in Tpr-Met transformed cells, Gab1 is constitutively phosphorylated on tyrosine (Figure 7a). Blot anti-Met These data indicate that Gab1 operates in Met- Figure 4 Peptides blocking the SH2 or the SH3 domains of downstream pathways leading to cellular transforma- Grb2 interfere with the Met/Gab1 interaction. (a) GTL16 cells tion. expressing phosphorylated Met receptors were lysed in the presence of phosphatase inhibitors. The MBD domain of Gab1, expressed as a GST fusion protein was immobilized on glutathione-Sepharose and incubated with lysates of GTL-16 Gab1 is constitutively tyrosine-phosphorylated in Met- cells in the presence of peptides binding the SH2 and SH3 transformed cells and binds SH2-signal transducers domains of Grb2 (Table 2). Upon association, the amount of receptor bound to the immobilized MBD domain was detected Gab1 contains a number of potential binding sites for by Western blotting with anti-Met antibodies. (b) Lysates of the SH2 domains of PI3-kinase, PLC-g and SHPTP2/ GTL-16 cells were incubated with anti-Met antibodies in the Syp. In Met-transformed cells Gab1 is constitutively presence of the indicated peptides. Associated Gab1 protein was phosphorylated on tyrosine (Figure 7a). The interac- detected by Western blotting with anti-Gab1 antibodies. A tion of Gab1 with SH2-containing molecules was control Western blotting with anti-Met antibodies was also performed. Position of the Met receptor (p145Met) and Gab1 evaluated in Tpr-Met transformed ®broblasts. Cells (p110Gab1) is indicated were lysed and Gab1 was imunoprecipitated with speci®c antibodies, as a control Gab1 was immunoprecipitated from non transformed ®broblasts. Gab1- associated signal transducers were visualized by counterpart of the receptor (Fixman et al., 1995; Western blotting with speci®c antibodies (Figure 7b). Ponzetto et al., 1996). To assess the role of Gab1 in We found that the amount of p85/PI3-kinase, PLC-g Met-mediated transformation we have compared the and SHPTP2/Syp bound to Gab1 was signi®cantly transforming activity of the various docking-site increased in Tpr-Met transformed cells with respect to mutants (Figure 6a) with their ability to recruit and their normal counterpart. Altogether these experiments phosphorylate Gab1 (Figure 6b). Fibroblasts trans- indicate that in Tpr-Met transformed ®broblasts, Gab1 formed by the cDNA encoding the mutants were lysed acts as a docking protein for SH2 containing and the Tpr-Met proteins were immunoprecipitated transducers and suggest that the pathways in which with Met-speci®c antibodies. Tpr-Met associated Gab1 each of these molecules operate may be relevant to was visualized by in vitro kinase assays. We found Met-mediated transformation. Met binding to Gab1 requires Grb2 ABardelliet al 3108 Discussion Met-signaling is mediated by the multifunctional docking site Y1349VHVNATY1356VNV, which binds a The oncogenic version of the Met receptor, Tpr-Met, number of dierent eectors (Ponzetto et al., 1994). has transforming, invasive and metastatic potential. Replacement of Y1349 and Y1356 with phenylalanine (Tpr-Met1349 ± 1356F mutant) completely abrogates the oncogenic potential of Tpr-Met by impairing recep- a

a IP anti-Gab1 FR wt H1351N

112 kD — p110Gab1

86 kD —

Blot anti-PTyr

112 kD — b wt Y1349F Y1356F N1358H N1351N p110Gab1

Blot anti-Gab1 112 kD — p110Gab1 b IP anti-Gab1 86 kD —

70 kD — Ctr IP FR wt H1351N

p65Tpr-Met p85PI3K 57 kD —

c Blot anti-p85

p80SHPTP2 Ctr antibody IP anti-Gab1

Blot anti-SHPTP2 p110Gab1

p140PLCγ

Figure 6 Correlation between the transforming potential of Tpr- Met mutants with the ability to bind and phosphorylate Gab1. (a) The transforming activity of wt and mutated TPR-MET cDNAs Blot anti-PLCγ was measured by a focus forming assay using Fisher Rat ®broblasts. The values reported are expressed as percentage of Figure 7 Gab1 binds PI3-Kinase, PLC-g and and SHPTP2 in Tpr-Metwt transforming ability and represent the average of three Tpr-Met transformed cells. (a) Lysates of normal (FR) and independent experiments. (b) Lysates of Tpr-Met transformed Tpr-Met transformed ®broblasts (Tpr-Metwt and Tpr-MetH1351N) cells were immunoprecipitated with anti-Met antibodies and were immunoprecipitated with anti-Gab1 antibodies. Phosphory- subjected to a kinase assay in the presence of [32P-g]ATP. Tpr- lated Gab1 was detected by Western blotting with anti- Met (p65Tpr-Met) and the associated Gab1 (p110Gab1) phospho- phophotyrosine antibodies. As a control the same ®lter-blot was proteins are indicated. (c) Proteins immunoprecipitated with anti- probed with anti-Gab1 antibodies. (b) The Gab1-associated SH2 Met antibodies and subjected to kinase assay as above were eluted containing transducers: PI3-Kinase (p85PI3K), SHPTP2 and re-immunoprecipitated with control or anti-Gab1 antibodies. (p80SHPTP2) and PLC-g, (p140 PLC-g) were visualized by Western A p110 phosphoprotein (p110Gab1) was separated as a major band blotting with the indicated antibodies Met binding to Gab1 requires Grb2 ABardelliet al 3109 tor-signaling (Ponzetto et al., 1994; Fixman et al., for Gab1 in the Met-signaling cascade, by binding the 1995, Giordano et al., 1997). receptor via the SH2 domain from one side, and the The initial aim of the present study was to identify MBD region of Gab1 via the SH3 domains on the molecules implicated in signaling through the Met- other. multifunctional docking site. We found that a 110 kDa In the yeast model the MBD domain appears to protein represents a major phosphorylation substrate bind directly to the receptor docking site of the activated receptor. This molecule binds and is Y1349VHVNA-TY1356VNV (Weidner et al., 1996). We phosphorylated by Tpr-Metwt but not by the signaling- used BIAcore-analysis to measure direct binding of inactive mutant (Tpr-MetY1349 ± 1356F). It has been shown the isolated MBD domain to the above phosporylated recently that the Met receptor binds to Gab-1, an IRS/ sequence. While the direct interaction with Grb2 could DOS-like multiadaptor protein (Weidner et al., 1996). be readily detected, no binding was observed with the We found that the Tpr-Met-associated 110 kDa MBD domain. However, ordered sequential binding phosphoprotein comigrates and is recognized by anti- experiments showed that the MBD domain eciently Gab1 speci®c antibodies. associates with the receptor docking site sequence The role of Gab1 in Tpr-Met-mediated transforma- when Grb2 is already bound to Y1356. The negative tion was evaluated using a series of point mutants in results obtained with the Biacore approach were the multifunctional docking that transform ®broblasts surprising, given that the MBD domain binds to with varying eciency. The transforming potential of Met-multifunctional docking in the yeast two-hybrid Tpr-Met mutants correlates with recruitment of Gab1 system (Weidner et al., 1996). One explanation of this in vivo indicating that this multiadaptor protein apparent discrepancy could be that additional operates in Met downstream pathways leading to aminoacids, absent in the peptide used in the cellular transformation. Furthermore, Gab1 is consti- BIAcore experiments, are required. Alternatively a tutively phosphorylated on tyrosine in Tpr-Met longer Gab1 region might be necessary to detect the transformed cells and acts as a docking protein for interaction in vitro. If the MBD domain also binds SH2 containing transducers including PI3-kinase, directly to the two-tyrosine multifunctional docking PLC-g and SHPTP2. The pathways in which each of site, a cooperative `double-hook' model can be these molecules operate could be relevant to Met- proposed for the Met/Gab1 interaction. On one side mediated transformation. Originally identi®ed as a indirect association occurs via Grb2 which selectively Grb2 binding molecule, Gab1 was subsequently found binds Y1356, whereas, on the other, the MBD domain to interact with the Met receptor in the yeast two- of Gab1 might interact directly with Y1349 (Figure 8). hybrid system (Holgado-Madruga et al., 1996; Weidner The `double-hook' mechanism could confer high et al., 1996). The primary sequence shows that Gab1 speci®city to the interaction and would explain why contains a PH domain, a number of potential docking Gab1 binds preferentially to Met but not to other sites for SH2-molecules and a proline-rich region which receptor tyrosine kinases such as Ret and Kit includes two optimal binding sites for the SH3 domains (Weidner et al., 1996). of Grb2. In the yeast system, the Gab1/Met interaction It is known that activation of the MET oncogene is mediated by the proline-rich region of Gab1, referred promotes tumorigenesis in mice and is involved in to as the Met Binding Domain (MBD) which human cancer. The data presented in this paper further binds the multifunctional docking site of the point to a central role for Grb2, including the Met receptor in a phosphorylation-dependent fashion. We have investigated the mechanisms of the Met/Gab1 interaction both in vitro and in mammalian cells using mutants of the receptor docking site (Y1349VHVNATY1356VNV). Binding was phosphorylation dependent and was abolished when both Y1349 and Y1356 were replaced with phenylalanines. Interestingly, we found that two mutations (Y1356F and N1358H) which selectively uncouple Grb2 from the receptor (Ponzetto et al., 1996), also abrogate Gab1 binding in intact cells. Similar results were obtained in in vitro association experiments using the isolated MBD domain of Gab1. Therefore, we conclude that in this cellular system Met-coupling to Gab1 is Grb2 dependent. Grb2 is a modular adaptor protein consisting of an SH2 domain ¯anked by two SH3 domains. While the SH2 domain is responsible for binding phosphorylated tyrosines the SH3 domains Figure 8 Cooperative `double-hook' model proposed for the mediate association with proline-rich regions present in Met/Gab1 interaction. Gab1 can interact both directly and other signalling molecules including SOS, Vav and Abl indirectly with the Met multifunctional docking site. On one side indirect association occurs via Grb2 which acts as a Gab1 (Pawson, 1995). adaptor by binding tirosine 1356 of Met with the SH2 domain To de®ne more clearly the role of Grb2 in the Met/ and the proline rich sequences of Gab1 with the SH3 domains. Gab1 interaction we made use of peptides which bind On the other, the MBD domain of Gab1 might interact to either the SH2 or the SH3 domains of Grb2. The directly with tyrosine 1349 (Weidner et al., 1996). The `double- Met/Gab1 interaction was impaired by peptides hook' mechanism could confer high speci®city to the interaction and would therefore explain why Gab1 binds interfering with both domains. These experiments selectively to Met but not to other receptor tyrosine kinases indicate that Grb2 may eciently act as an adaptor such as Ret and Kit Met binding to Gab1 requires Grb2 ABardelliet al 3110 recruitment of Gab1, in Met-signaling. Approaches Association experiments aimed at interfering with the Met-Grb2 interaction are Wild-type or mutant TPR-MET cDNAs were cloned in therefore likely to be valuable in the control of the the pMT2 vector and transiently transfected in BOSC cells oncogenic properties of the Met receptor. using Lipofectin (BRL). After lysis with EB buer, immunoprecipitation with anti-HGF receptor C-terminal antibodies was performed and samples analysed by SDS ± PAGE and Western blotting with anti-Gab1 antibodies. Materials and methods GST-MBD was immobilized on glutathione-Sepharose (Pharmacia) and incubated with lysates of BOSC cells Reagents, antibodies and cell culture transiently expressing wt or mutant Tpr-Met proteins. After 0.5 M NaCl washing, samples were analysed by All reagents used were from Fluka (FlukaChemie AG) and SDS ± PAGE and Western blotting with anti-Met anti- Sigma (Sigma Chemicals Co). Reagents for SDS ± PAGE bodies. were from Bio-Rad (Biorad Laboratories). Recombinant HGF was obtained from Baculovirus-infected SF-9 cells (Naldini et al., 1991). Anti-phosphotyrosine, anti-GAB1 Peptide competition experiments and anti-p85 antibodies were purchased from UBI, anti Association experiments were performed in the presence of SHPTP2 antibody was kindly provided by Dr Emma Villa- the indicated concentration of the Met-derived phospho- Moruzzi, anti-HGF monoclonal antibodies were obtained peptides (pYVHV, pYVNV, pYVHVNATpYVNV) or the in this laboratory (Prat et al., 1991). BOSC, Fisher Rat and SOS-derived peptide (SKGTDEVPVPPPVPPRR) as pre- A549 cells were from ATCC. GTL-16 cells were obtained viously described (Ponzetto et al., 1993). The SOS peptide as described (Giordano et al., 1989). Cultures of was a kindly provided by Dr Ivan Gout. mammalian cells were maintained in DMEM or RPMI supplemented with 10% FCS in a humidi®ed atmosphere Biacore-Biosensor analysis of 5% CO2-air. Real-time interaction analysis was performed with a BIAlite-Biosensor (Pharmacia) as previously described Cloning and expression of the Met Binding Domain (Panayotou et al., 1993; Ponzetto et al., 1993). Puri®ed Human heart tissue was pulverized in liquid nitrogen and GST-fusion proteins (MBD or Grb2) were desalted in mRNA was prepared with the Guanidine Isothyocianate BIAlite running buer (20 mM Hepes, pH 7.4, 150 mM method (Sambrook et al., 1989). Poly (A)+ mRNA was NaCl, 3.4 mM EDTA, 0.005% Tween20 and 4 mM DTT) isolated with oligo dT-conjugated beads (Dynazyme). One and injected over the avidin-immobilized, biotinylated mgofpoly(A)+ mRNA was used for the synthesis of the phosphopeptide (GGGGIGEHY*VHVNATY*VNVK,

®rst strand of cDNA (reaction mix: 10 mgpd(T)12±18, kindly provided by Dr C Battistini, Pharmacia & Upjon 140 mM potassium chloride, 10 mM magnesium chloride, Preclinical Research) corresponding to the multifunctional 1mM each dNTPs, 4 mM DTT, 25 mM RNAse inhibitor docking site. Association was measured over a range of (Promega), 40U Avian reverse transcriptase (Promega) in protein concentrations. 50 ml). After 1 h at 378C, the reaction was stopped by boiling for 20 min. Residual RNAs were removed by Autokinase assay and phosphoaminoacid analysis adding RNAse A (10 mg/ml), 378C, 20 min. PCRs were performed using 25 ml of the cDNA reaction mix as Immunoprecipitation with anti-Met antibodies was per- template and the following oligonucleotides as primers: formed from lysates of Fisher Rat ®broblasts expressing wt sense primer=5'-AAAATGAATTCCCAATGAATCCCA- or mutant Tpr-Met proteins. Immunoprecipitates were ATTCACCACC-3'; antisense primer=5'-TTCTACTCGAG- incubated with Kinase Buer (KB: 50 mM HEPES

CTGGCTTGACTTTTCTGTCTGGCTTGAGG TTCC-3'. pH 7.5, 150 mM NaCl, 12.5 mM MgCl2) in the presence The underlined residues represent the recognition se- of 20 mM ATP including [32Pg]ATP for 15 min at 378C. quences for the restriction nucleases EcoRI (sense primer) Samples were then analysed by SDS ± PAGE and auto- and XhoI (antisense primer). The PCR fragment was radiography. Bands corresponding to Tpr-Met and Gab1 digested with EcoRI and XhoI, gel-puri®ed and cloned in were cut from the gel and subjected to phosphoaminoacid the GST-fusion-vector pGEX-4T-2 (Pharmacia). Sequen- analysis as described by Cooper et al. (1983). In the re- cing of the PCR product (Sequenase, UBI) showed that no immunoprecipitation experiments, after the kinase reaction mutations were introduced upon ampli®cation. Expression anti-Met immunoprecipitates were eluted by boiling in the of GST-MBD fusion protein was performed in XL1-blue presence of Laemmli Buer without reducing agent (1% cells (Stratagene). SDS, 20 mM Tris-HCl pH 7.4). Samples were then diluted to 0.1% SDS, incubated with anti-Gab1 antibodies and analysed as above. Western immunoblotting

After SDS ± PAGE, proteins were transferred to Hybond- Focus forming assay ECL membranes (Amersham) by high intensity wet blotting. Filters were probed with the appropriate Transfection of the TPR-MET constructs in Fisher rat antibodies and speci®c binding was detected by the ®broblasts was carried out using the DNA-calcium Enhanced Chemiluminescence System (ECL, Amersham). phosphate co-precipitation procedure (CellPhect Transfec- tion Kit, Pharmacia). After transfection, cells were split at very low density and kept in DMEM 5% FCS medium. HGF stimulation of epithelial cells Formation of transformed foci was detected in 2 ± 3 weeks A549 cells were starved in serum-free medium for 24 h, as previously described (Ponzetto et al., 1994) treatedwith120ng/mlHGFandlysedinEBbuer (100 mM Tris pH 7.4, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 1% Triton X-100) containing protease and Acknowledgements phosphatase inhibitors. Immunoprecipitation with anti- We wish to thank S DaroÁ for her enthusiastic help in Gab1 or anti-Met antibodies was performed and samples performing the peptide competition experiments. The analysed by SDS ± PAGE and Western immunoblotting. technical assistance of R. Callipo and G. Petrucelli is Met binding to Gab1 requires Grb2 ABardelliet al 3111 gratefully acknowledged. We thank Dr T Williams for This work was supported by grants from the Italian critical reading of the manuscript, A Cignetto for National Research Council (CNR) and from the Associa- secretarial help and E Wright for editing the manuscript. zione Italiana per la Ricerca sul Cancro (AIRC).

References

Bardelli A, Pugliese L and Comoglio PM (1997). Biochemica Park M, Dean M, Cooper CS, Schmidt M, SJ OB, Blair DG et Biophysica Acta, Review on Cancer, in press. and Vande Woude GF. (1986). Cell, 45, 895 ± 904. Cooper JA, Sefton BM and Hunter T. (1983). Methods Pawson T. (1995). Nature, 373, 573 ± 580. Enzymol., 99, 387 ± 402. Ponzetto C, Bardelli A, Maina F, Longati P, Panayotou G, Cooper CS, Park M, Blair DG, Tainsky MA, Huebne K, Dhand R, Water®eld MD and Comoglio PM. (1993). Mol. Croce CM and Vande Woude GF. (1984). Nature, 311, Cell. Biol., 13, 4600 ± 4608. 29 ± 33. Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P, Di Renzo MF, Olivero M, Giacomini A, Porte H, Chastre E, Giordano S, Graziani A, Panayotou G and Comoglio PM. Mirossay L, Nordlinger B, Bretti S, Bottardi S, Giordano (1994). Cell, 77, 261 ± 271. S, Plebani M, Gespach C and Comoglio PM. (1995). Clin. Ponzetto P, Zhen Z, Audero E, Maina F, Bardelli A, Basile Cancer Res., 1, 147 ± 154. M, Giordano S, Narsimhan R and Comoglio PM. (1996). Fixman ED, Naujokas MA, Rodrigues GA, Moran MF and J. Biol. Chem., 271, 14119 ± 14123. Park M. (1995). Oncogene, 10, 237 ± 249. PratM,NarsimhanRP,CrepaldiT,NicotraMR,NataliPG Giordano S, Ponzetto C, Di Renzo MF, Cooper CS and and Comoglio PM. (1991). Int. J. Cancer, 49, 323 ± 328. Comoglio PM. (1989). Nature, 339, 155 ± 156. Rodrigues GA and Park M. (1993). Mol. Cell Biol., 13, Giordano S, Bardelli A, Zhen Z, Menard S, Ponzetto C and 6711 ± 6722. Comoglio PM. (1997). Proc. Natl. Acad. Sci., in press. Schmidt L, Duh F, Chen F, Kishida T, Glenn G, Choyke P, Graziani A, Gramaglia D, Cantley LC and Comoglio PM. Scherer SW, Zhuang Z, Lubensky I, Dean M, et al. (1997). (1991). J. Biol. Chem., 266, 22087 ± 22090. Nature genetic, 16, 68 ± 73. Graziani A, Gramaglia D, dalla Zonca P and Comoglio PM. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular (1993). J. Biol. Chem., 268, 9165 ± 9168. Cloning: a Laboratory Manual. Cold Spring Harbor Press: Holgado-Madruga M, Emlet DR, Moscatello DK, Godwin Cold Spring Harbor. AK and Wong AJ. (1996). Nature, 379, 560 ± 564. Weidner KM, Sachs M, Riethmacher D and Birchmeier W. Naldini L, Weidner M, Vigna E, Gaudino G, Bardelli A, (1995). Proc. Natl. Acad. Sci. USA, 92, 2597 ± 2601. Ponzetto C, Narsimhan R, Hartmann G, Zarnegar R, Weidner KM, DiCesare S, Sachs M, Brinkmann V, Behrens J Michalopoulos G, Birchmeier W and Comoglio PM. and Birchmeier W. (1996). Nature, 384, 173 ± 176. (1991). EMBO J., 10, 2867 ± 2878. Zarnegar R and Michalopoulos GK. (1995). J. Cell. Biol. Panayotou G, Gish G, End P, Troung O, Gout I, Dhand R, 129, 1177 ± 1180 Fry MJ, Hiles I, Pawson T and Water®eld MD. (1993). Mol. Cell Biol., 13, 3567 ± 3576.