The insulin receptor substrate (IRS)-1 recruits phosphatidylinositol 3 ...

12 downloads 44 Views 251KB Size Report
Received 2 May 2000; revised 23 October 2000; accepted 23. October 2000 ... Class I PI3-Ks are composed of p85 regulatory and p110 catalytic ... Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1). RM Melillo et al. 210.
Oncogene (2001) 20, 209 ± 218 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

The insulin receptor substrate (IRS)-1 recruits phosphatidylinositol 3-kinase to Ret: evidence for a competition between Shc and IRS-1 for the binding to Ret Rosa Marina Melillo1, Francesca Carlomagno1, Gabriella De Vita1, Pietro Formisano1, Giancarlo Vecchio1, Alfredo Fusco2, Marc Billaud3 and Massimo Santoro*,1 1

Centro di Endocrinologia ed Oncologia sperimentale del CNR, c/o Dipartimento di Biologia e Patologia Cellulare e Molecolare `L. Califano', Facolta' di Medicina e Chirurgia, Universita' di Napoli `Federico II', via S. Pansini 5, 80131, Naples, Italy; 2 Dipartimento di Medicina Sperimentale e Clinica, Facolta' di Medicina e Chirurgia di Catanzaro, Universita' di Catanzaro, via T. Campanella 5, 88100, Catanzaro, Italy; 3Laboratoire de Genetique, CNRS UMR5641, 8 avenue Rockfeller, Lyon 69373 Cedex 08, France

Tyrosine 1062 of Ret, which represents an intracytoplasmic docking site for multiple signaling molecules, is essential for Ret-mediated activation of phosphatidylinositol 3-Kinase (PI3-K). PI3-K, in turn, has been implicated in inducing cell survival and neoplastic transformation mediated by Ret. We have examined the mechanisms by which Ret stimulates PI3-K. Here we show that the Insulin Receptor Substrate-1 (IRS-1) is tyrosine phosphorylated and associated with the p85 regulatory subunit of PI3-K in response to Ret activation. IRS-1 coimmunoprecipitates with Ret and co-expression of IRS-1 results in the potentiation of Retmediated activation of Akt(PKB), a bona ®de e€ector of PI3-K. The association with the PTB domain of IRS-1 depends on the phosphorylation of tyrosine 1062 of Ret. The deletion of asparagine 1059 (delN1059) and the substitution of leucine 1061 (L1061P), two Ret mutations identi®ed in families a€ected by congenital megacolon (Hirschsprung's disease), impair the binding of IRS-1 to Ret as well as Ret-mediated Akt(PKB) stimulation. Finally, we show that Shc, which was previously identi®ed as another ligand of Y1062 of Ret, competes with IRS-1 for the binding to Ret pY1062. All together, these ®ndings suggest that IRS-1 is a component of the signaling pathway which leads to Ret-mediated PI3-K activation, a pathway which can be targeted by Hirschsprung-associated Ret mutations. The alternative binding of Shc and IRS-1 to Ret pY1062 can be a system to modulate the activation of di€erent intracellular signaling pathways and to elicit di€erent biological responses following Ret activation. Oncogene (2001) 20, 209 ± 218. Keywords: Ret; tyrosine kinase; IRS-1; PI3-K

*Correspondence: M Santoro Received 2 May 2000; revised 23 October 2000; accepted 23 October 2000

Introduction The Ret transmembrane tyrosine kinase is part of multimeric receptor complexes that bind four structurally related growth factors: GDNF, neurturin, artemin and persephin (Airaksinen et al., 1999). The other component of the receptor complex is a member of a family of glycosyl-phosphatidylinositol (GPI)-membrane anchored molecules, the so-called GFRas. Four di€erent GFRas (GFRa 1, 2, 3 and 4) dictate the ligand speci®city of the complex. GDNF and related molecules are potent neurotrophic factors for several neuronal populations and Ret plays a crucial role in the development of the enteric nervous system and the kidney (Rosenthal, 1999). Somatic rearrangements of the Ret gene are the most typical molecular alteration found in papillary thyroid carcinomas. These rearrangements cause the fusion of the Ret tyrosine kinase encoding domain to the 5'-terminal domain of heterologous genes, thus generating the Ret/PTC oncogenes (Pierotti et al., 1996). On the other hand, activating germ line mutations of Ret are responsible for the familial medullary thyroid carcinoma (FMTC) and multiple endocrine neoplasia (MEN) type 2A and 2B syndromes. MEN2A and MEN2B are characterized by the development of medullary thyroid carcinoma and pheochromocytoma. FMTC consists in the inherited predisposition to develop medullary thyroid carcinoma (Ponder, 1999). Finally, inactivating mutations of Ret cause the impaired development of the enteric nervous system which is responsible for the congenital megacolon or Hirschsprung's disease (HSCR) (Pasini et al., 1996). Activation of the Ret kinase can be achieved either by interaction with its cognate ligand or by di€erent oncogenic mutations. Upon activation, Ret undergoes autophosphorylation on tyrosine residues which, in turn, are docking sites for several signaling molecules. Ret tyrosine 1015 (Y1015) is a docking site for phospholipase Cg (Borrello et al., 1996), tyrosine 905 (Y905) is involved in binding of Grb7 and Grb10 (Pandey et al., 1995, 1996), tyrosine 1096 binds Grb2

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

210

Oncogene

(Alberti et al., 1998), and tyrosine 1062 (Y1062) represents the binding site for at least two molecules, Shc and Enigma (Arighi et al., 1997; Asai et al., 1996; Lorenzo et al., 1997; Durick et al., 1998). By recruiting Grb2/Sos complexes, Shc is involved in the coupling of several receptors to the Ras/MAPK signaling pathway (Pelicci et al., 1992; Schlessinger, 1993). Enigma, is a LIM- and PDZ-domain containing protein which plays an essential role in Ras activation mediated by rearranged Ret/PTC oncoproteins (Durick et al., 1998). Recent reports demonstrate that phosphatidylinositol 3-Kinase (PI3-K) is stimulated by oncogenic Ret and by ligand-stimulated wild type Ret (Murakami et al., 1999a,b; van Weering et al., 1998) and that PI3-K stimulation depends on Ret tyrosine 1062 (SegounCariou and Billaud, 2000; De Vita et al., 2000). Both the oncogenic and the cell survival activities of Ret require coupling to PI3-K; accordingly, the mutation of Y1062 severely impairs the transforming activity of the Ret/MEN2A oncogene (Asai et al., 1996; Segoun-Cariou and Billaud, 2000a) and its ability to promote survival of rat pheochromocytoma PC12 cells (De Vita et al., 2000b). PI3-Ks produce phosphoinositides phosphorylated at the D3 hydroxyl group (phosphatidylinositol 3-phosphate, 3,4-bisphosphate and 3,4,5-trisphosphate), which modulate the activity of several e€ectors, including the serinethreonine kinase Akt(PKB) (Rameh and Cantley, 1999). Akt(PKB) promotes proliferation and inhibits apoptosis following activation (Datta et al., 1999). Class I PI3-Ks are composed of p85 regulatory and p110 catalytic subunits. The p85 subunit has two SH2 domains which mediate PI3-K recruitment to receptor tyrosine kinases (Schlessinger, 1994). The binding of p85 to tyrosine kinases can be either direct or mediated by di€erent docking proteins, such as IRS1, Gab-1 or c-cbl (Rameh and Cantley, 1999). Ret does not show phosphorylation sites in the YXXM motif, which is a selective binding site for the SH2 domains of p85; likely, its binding to PI3-K is indirect and mediated by docking proteins. Gab1 and Gab2 show phosphorylation and PI3-K recruitment upon Ret activation (Murakami et al., 1999b; Hayashi et al., 2000; Besset et al., 2000); it remains still unclear whether Gab family members are the only proteins involved in Ret-mediated PI3-K activation. IRS-1, the Insulin Receptor Substrate 1, belongs to a family of structurally related docking proteins (IRS-1, 2, 3 and 4). IRS-1 binds to the insulin and other receptors through its phosphotyrosine binding (PTB) domain (Zhou et al., 1996). Upon receptor stimulation, IRS-1 is rapidly phosphorylated at many tyrosine positions and engages numerous SH2 proteins including p85, Grb2, Fyn, Nck, phospholipase Cg, and the phosphatase SH-PTP2 (White, 1998). Several phosphorylation sites in YXXM motifs are clustered together in the middle of the IRS-1 protein; accordingly, IRS-1 mediates the direct activation of PI3-K (Myers et al., 1996; Wang et al., 1993). Here we show that IRS-1

binds to the NKLpY(1062) motif of Ret, competes with the binding of Shc to the same docking site, and is involved in Ret-mediated activation of PI3-K.

Results Tyrosine phosphorylation of IRS-1 upon Ret stimulation We have previously generated NIH3T3 cells which express a chimeric EGFR/Ret (E/R) receptor (Figure 1a), whereby the extracellular ligand binding domain of Ret was replaced with that of the EGF receptor (Santoro et al., 1994). These cells represent a suitable system to study Ret signaling as NIH3T3 ®broblasts express relatively low levels of endogenous EGFR and no Ret receptors. The use of this chimeric receptor allows the analysis of signal transduction through Ret independently of the GFRa co-receptors which are capable of autonomous signaling in response to ligand binding (Trupp et al., 1999). To understand how Ret recruits PI3-K we screened for proteins that are tyrosine phosphorylated and associated with the regulatory p85 subunit of PI3-K in response to Ret activation. Lysates from EGFtreated or from untreated cells were immunoprecipitated with antibodies against p85 and the immunocomplexes were probed with anti-phosphotyrosine antibodies. Immunocomplexes obtained from stimulated cells contained several proteins (of 180, 160, and 120 kDa) phosphorylated on tyrosine (Figure 1b). The 160 and the 120 kDa bands were identi®ed as the EGFR/Ret (E/R) and the Grb2-associated binder-1 (Gab-1) respectively, by reprobing the ®lter with the corresponding antibodies (data not shown). Gab-1 has been reported by others to associate with P13-K in cells expressing activated Ret (Murakami et al., 1999a,b). We concentrated our interest on the p180 kDa protein which was not described previously in Ret signaling. As suggested by its molecular mass, the p180 kDa protein was identi®ed as the insulin receptor substrate-1 (IRS-1) by subsequent reprobing of the ®lter (data not shown). Thus, to assess more directly whether IRS-1 couples with Ret, IRS-1 was immunoprecipitated from protein lysates of EGFtreated or untreated cells and the immunoblot was probed with anti-phosphotyrosine and then with antiRet antibodies. The results reported in Figure 1c show that IRS-1 was tyrosine phosphorylated and associated with the activated E/R receptor upon ligand-triggering. Furthermore, IRS-1 exhibited the characteristic mobility retardation as a result of phosphorylation in response to growth factor stimulation (Figure 1c). The physical association of phosphorylated IRS-1 with Grb2 has been reported to occur following insulin stimulation (Skolnik et al., 1993). To assess whether IRS-1 also associates with Grb2 following activation of Ret, we immunoprecipitated IRS-1, Shc or Grb2 from protein lysates of EGF-treated or untreated cells and probed the immunoblot with anti-Grb2 antibodies. The results

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

211

Figure 1 IRS-1 phosphorylation and association with p85 and with activated Ret. (a) A schematic representation of Ret, Ret/ MEN2A and the chimeric EGFR/Ret receptor. CAD: cadherin homologous domain, CYS: cysteine-rich sequence. (b) NIH E/R cells were serum starved and then stimulated with 100 ng/ml EGF. Protein extracts (500 mg) were immunoprecipitated with anti-p85 and analysed by Western blotting with anti-phosphotyrosine antibodies and with anti-p85 antibodies for normalization. Phosphorylated proteins are indicated by arrows. (c) IRS-1 was immunoprecipitated from lysates of cells treated with EGF or left untreated and probed with the indicated antibodies. (d) IRS-1, Grb2 or Shc were immunoprecipitated from lysates of cells treated with EGF or left untreated and probed with anti-Grb2 antibodies. (e) NIH3T3-Ret/MEN2A, -Ret, or parental NIH3T3 cells were serum starved (12 h). Protein extracts (500 mg) were immunoprecipitated with anti-p85 and analysed by Western blotting with anti-phosphotyrosine or with anti-IRS-1 antibodies; another aliquot of the extracts was immunoprecipitated with anti-IRS-1 and immunoblotted with anti-phosphotyrosine or anti-IRS-1

reported in Figure 1d showed that IRS-1 does not associate with Grb2 upon ligand-triggering of the E/R receptor. Conversely, as expected, Shc associated with Grb2 following Ret stimulation. It is possible that Ret promotes the tyrosine phosphorylation of the IRS-1 sites crucial for p85 binding but not of those involved in Grb2 binding. Then, we made use of NIH3T3 cells expressing a constitutively activated Ret mutant, Ret(C634R), i.e. Ret/MEN2A (Figure 1a). The substitution of cysteine 634 with di€erent amino acids (such as arginine) induces Ret activation through disul®de bondmediated dimerization and causes the tumoral MEN2A syndrome (Santoro et al., 1995). As controls, we used both NIH3T3 cells expressing wild type Ret (Carlomagno et al., 1998) and untransfected cells. Protein

lysates were immunoprecipitated with antibodies against p85 and the immunocomplexes were probed with anti-phosphotyrosine or anti-IRS-1 antibodies. Figure 1e shows that p85 is constitutively associated with phosphorylated IRS-1 in cells expressing Ret/ MEN2A but not in control cells. To obtain independent evidence of the p85/IRS-1 interaction in cells expressing activated Ret, we performed a far Western experiment. Protein lysates from Ret-, Ret/MEN2A-expressing or from untransfected NIH3T3 cells were immunoprecipitated with anti-Ret, anti-IRS-1 and anti-Gab1, subjected to SDS ± PAGE, blotted and stained with GST fusion proteins containing either the N- or the C-terminal SH2 domain of p85 followed by anti-GST antibodies. As shown in Figure 2, GST p85/N-SH2 interacted with Oncogene

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

212

Ser 473 and thus active (Bellacosa et al., 1998). As shown in Figure 3c, expression of Ret/MEN2A or of IRS-1 alone induced Akt phosphorylation; this phosphorylation was potentiated (about threefold, as assessed by densitometric scanning) when Ret/ MEN2A, and not when Ret(K-), was co-transfected with IRS-1. Figure 2 IRS-1 and Gab1 interaction with p85/N-SH2. Protein extracts (2 mg) from the indicated cell populations were immunoprecipitated with the indicated antibodies and analysed by far Western blotting with GST p85/N-SH2 (see Materials and methods). The identity of the bands was assessed by subsequent reprobing of the ®lter with anti-Gab1 and anti-IRS-1 antibodies (not shown)

Gab1 and IRS-1 in cells expressing activated Ret but not in control cells. No direct interaction of GST p85/ N-SH2 with Ret/MEN2A was detectable. Similar results were obtained with GST p85/C-SH2 (data not shown). According to the coimmunoprecipitation data (Figure 1e), the Gab1 signal was signi®cantly stronger than the IRS-1 one. IRS-1 enhances Ret-mediated activation of PI3-K To test whether IRS-1 is involved in Ret-mediated PI3-K stimulation, human embryonic kidney 293 cells were transiently transfected with IRS-1, Ret/MEN2A or co-transfected with IRS-1 and Ret/MEN2A. As a control, we co-expressed IRS-1 also with Ret(K-), a kinase-dead Ret mutant. The use of such a mutant was necessary because when expressed in 293 cells, wild type Ret shows signi®cant autophosphorylation levels probably due to protein overexpression (not shown). Protein extracts from the transfected cells were harvested and immunoprecipitated with antiIRS-1 antibodies. The activity of PI3-K in the immunocomplexes was analysed by in vitro kinase assay followed by thin layer chromatography and quantitated by densitometric scanning. A readily detectable PI3-K activity co-precipitated with IRS-1 in Ret/MEN2A-transfected cells but not in untransfected cells. Lower levels of PI3-K activity (about 2.5folds lower) were complexed with IRS-1 in cells transfected with IRS-1 or with Ret/MEN2A alone. A representative assay is shown in Figure 3a and the average results of three di€erent experiments are reported in Figure 3b. This demonstrates that IRS-1 couples with PI3-K activity in cells expressing an active Ret. We then asked whether the over-expression of IRS-1 can potentiate PI3-K activation by Ret. To this aim, we tested Ret/MEN2A-mediated activation of the serine-threonine kinase Akt, a downstream e€ector of PI3-K. 293 cells were transfected with Ret/ MEN2A, IRS-1 or with IRS-1 and either Ret/ MEN2A or Ret(K-) as a control. An epitope tagged Akt (Ha-Akt) was co-expressed in all the samples. Protein extracts were prepared and the degree of Akt activation was assessed by immunoblot with antibodies which detect Akt only when phosphorylated at Oncogene

The NKLpY(1062) motif is essential for Ret association with IRS-1 The PTB domain of IRS-1 mediates its interaction with tyrosine kinase receptors (White, 1998), recognizing protein motifs containing an asparagine residue at position 73 with respect to a phosphorylated tyrosine (Pawson and Scott, 1997). One candidate residue for mediating IRS-1 binding was Ret tyrosine 1062 (Y1062), which is embedded in a NKLpY motif and is essential for the binding of the PTB domain of Shc and of the LIM2 domain of Enigma to Ret (Arighi et al., 1997; Asai et al., 1996; Lorenzo et al., 1997). Y1062 is required for PI3-K binding to Ret and for Ret-mediated transforming and survival signaling (Segoun-Cariou and Billaud, 2000; De Vita et al., 2000a). Thus, 293 cells were transiently co-transfected with IRS-1 and Ret/MEN2A or with IRS-1 and a Ret/MEN2A plasmid carrying the Y1062F mutation. In addition, 293 cells were cotransfected with IRS-1 and Ret/MEN2A constructs harboring the delN1059 and L1061P mutations. These mutations target residues of the NKLpY motif of Ret and have been recently identi®ed in HSCR families. We have previously shown that both delN1059 and L1061P impair Shc association to Ret (Geneste et al., 1999). The Ret constructs are schematically represented in Figure 4a. Forty-eight hours after transfection, cells were serum deprived for 12 h, and then harvested. Protein extracts were immunoprecipitated with anti-IRS-1 and the immunocomplexes were probed with anti-phosphotyrosine or anti-Ret antibodies. As shown in Figure 4b, the Y1062F and the delN1059 HSCR mutation abrogated Ret's ability to associate with IRS-1 and to induce phosphorylation of IRS-1 in living cells, while the L1061P mutation inhibited only partially Ret/IRS-1 association. The expression levels of IRS-1 and of the di€erent Ret plasmids in the transfected populations were comparable (Figure 4b, lower panels). We then asked whether the association with Ret was mediated by the PTB domain of IRS-1. To this aim, protein extracts from Ret/MEN2A, Ret/MEN2A(Y1062F), Ret/MEN2A(L1061P) and Ret/MEN2A(delN1059) expressing cells were subjected to an in vitro `pulldown' assay with a GST-IRS-1(PTB) fusion protein. Bound proteins were detected by immunoblotting with anti-Ret antibodies. As shown in Figure 4c, Ret/ MEN2A associated with GST-IRS-1(PTB). The L1061P mutation strongly reduced such interaction; the Y1062F and delN1059 mutations virtually abrogated the binding. As a control of this experiment and to further address the role of

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

213

Figure 3 IRS-1 expression potentiates PI3-K/Akt activation by RET/MEN2A. (a, b) PI3-K activity in di€erent transfectants. 293 cells were transfected with the indicated constructs. Total cell lysates were immunoprecipitated with anti-IRS-1 antibodies and PI3K catalytic activity was assessed by thin layer chromatography. One representative experiment (a) and the average results of two independent experiments (b). (c) 100 mg of protein lysates from the transfected cells were immunoblotted with anti phosphoAkt (Ser 473) or anti-Akt. (d) Immunoblots were probed with anti-Ret or anti-IRS-1 antibodies for normalization

Y1062 in mediating Ret/IRS-1 interaction, we have generated additional molecular constructs: Ret(C634R)4F, carrying the mutation of Y826, Y1015, Y1029 and Y1062 (the four major Ret autophosphorylation sites) (Liu et al., 1996) and Ret(C634R)3F, in which Y1062 was left wild type. A `pull-down' assay was performed by comparing Ret(C634R) with Ret(K-), Ret(C634)4F and Ret(C634R)3F (Figure 4d). Ret kinase activity resulted essential for Ret/IRS-1 interaction; furthermore, while the 4F mutant virtually lacked any IRS-1 binding activity, the 3F mutant showed interaction levels similar to those exerted by wild type Ret(C634R), further supporting the role of Y1062 in IRS-1 binding. Since, as shown above, IRS-1 participates to Retmediated activation of the PI3-K/Akt pathway, the Ret/MEN2A mutants which are defective in binding IRS-1 could be impaired in the Akt stimulating activity. To test this hypothesis, we co-expressed an epitope tagged Akt construct (Ha-Akt) with Ret/ MEN2A or its mutant derivatives in 293 cells and probed the lysates with the antibody to the phosphorylated form of Akt. Although they retained a residual ability, the Y1062F, L1061P, delN1059 and 4F Ret/MEN2A mutants showed clear defects in Akt activation (Figure 4e). To assess in parallel Akt and

MAPK activation we transfected Ret/MEN2A and its Y1062F mutant in 293 cells and probed the ®lter with the antibody to the phosphorylated form of Akt and with phosphorylation-speci®c anti-MAPK antibodies (Figure 4f). Y1062 resulted to be essential for full activation of both pathways. However, residual Akt and MAPK phosphorylation levels were present in cells transfected with the Y1062F mutant indicating that Ret residues other than Y1062 and docking proteins other than those binding Ret Y1062 may contribute to Ret coupling with such pathways. To further characterize the binding between Ret and the PTB domain of IRS-1, we performed a peptide competition assay. To this end, we used two peptides designed on the Ret NKLY(1062) motif and containing the residue Y1062 either in the phosphorylated (#pp1062) or in the unphosphorylated (#p1062) state. 293 cells were transfected with Ret/MEN2A and the protein extracts were subjected to a `pull-down' assay with GST-IRS-1(PTB) (Figure 5a). The binding between GST-IRS-1(PTB) and Ret/MEN2A was not a€ected by increasing concentrations of the unphosphorylated peptide. The phosphopeptide, instead, was able to displace Ret/MEN2A binding to the recombinant PTB protein. Inhibition of the binding was clearly detected at 10 mm of peptide concentration, and was complete at 50 mm. Oncogene

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

214

Figure 4 The Ret NKLpY motif is essential for the binding to the PTB domain of IRS-1. (a) A schematic diagram of the Ret protein. (b) 293 cells were transiently co-transfected with IRS-1 and the indicated plasmids; (7): cells transfected with IRS-1 alone. The cell lysates (1 mg) were immunoprecipitated with anti-IRS-1 antibodies and the immunocomplexes were probed with either antiphosphotyrosine or anti-Ret antibodies. Aliquots of the same lysates (50 mg) were probed with anti-IRS-1 and anti-Ret antibodies to demonstrate expression levels. (c,d) Cell lysates (1 mg) from 293 cells expressing the indicated plasmids were subjected to a `pulldown' assay with the GST-IRS-1(PTB) fusion protein or GST as a control. Bound proteins were revealed with anti-Ret antibodies (upper panel). Equal Ret protein loading is shown in the lower panel. (e) Ha-Akt was co-expressed with Ret/MEN2A or its mutants in 293 cells. Protein extracts (100 mg) were immunoblotted with an antibody to the phosphorylated activated form of Akt or with anti-Akt. (f) The indicated Ret mutants were expressed in 293 cells. Akt activation was detected as in e. MAPK activation was detected with phospho-MAPK speci®c antibodies. Anti-MAPK was used for normalization

IRS-1 and Shc compete for binding to Ret at Tyr1062 The observation that both Shc and IRS-1 bind to the same phosphotyrosine residue on Ret suggests a model in which Shc and IRS-1 may compete for binding to Ret. This may regulate a cellular switch between the activation of the Ras/MAPK and the PI3-K/Akt cascades, which are triggered (although not exclusively) by Shc and IRS-1, respectively. To test this possibility, competition between Shc and IRS-1 binding to Ret/ MEN2A was assayed in vitro. GST-IRS-1(PTB) was used to precipitate Ret/MEN2A in the presence of increasing concentrations (10, 25 and 50 mg) of soluble Shc(PTB) (Figure 5b) or 50 mg of BSA (Figure 5c). As a further control, we evaluated whether soluble Shc(PTB) competed with the association of Ret/ MEN2A with 5 mg of GST-Src(SH2) (Figure 5c), because we have previously shown that Src binding to Ret does not depend on tyrosine 1062 (Melillo et al., 1999). Soluble Shc(PTB) competes with GST-IRS1(PTB): an almost complete competition was observed Oncogene

with a twofold excess of the soluble protein (corresponding to about ®vefold molar excess). In comparison, 50 mg of BSA did not compete. Furthermore, the soluble Shc(PTB) did not exert any competition when GST-Src(SH2) was used for the `pull-down', indicating that this competition is speci®c and involves tyrosine 1062. Discussion IRS-1 was ®rst identi®ed as a substrate of the insulin receptor (IR) (Myers et al., 1994). More recently, it has been shown that docking molecules of the IRS family are involved in the signaling pathways of several receptors, such as the growth hormone receptor (Liang et al., 1999), the IL-2, -4, -7, -9 and -15 receptors (Johnston et al., 1995; Keegan et al., 1994; Wang et al., 1993; Yin et al., 1995), the IFNa receptor (Uddin et al., 1995) and the TRK family of neurotrophic receptors (Yamada et al.,

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

Figure 5 Peptide competition and competition between the PTB domains of Shc and IRS-1 for the binding to Ret. (a) A pulldown experiment was performed in the presence of increasing concentrations (1, 10 or 50 mM) of two peptides designed on the Ret NKLY motif and containing the Y1062 in the phosphorylated or non-phosphorylated state. The pull-down of Ret/ MEN2A(Y1062F) was performed, as a negative control. An immunoblot was performed to demonstrate equal Ret protein loading. (b) `Pull-down' assay of Ret/MEN2A interaction with IRS-1(PTB) in the presence of soluble PTB domain of Shc. Ret/ MEN2A was transiently expressed in 293 cells. Protein extracts (1 mg) were incubated with GST-IRS-1(PTB)-Sepharose in the presence of increasing amounts of soluble Shc(PTB) or bu€er (7). Bound proteins were revealed with anti-Ret antibodies. (c) The pull-down was performed by incubating protein extracts of Ret/MEN2A transfected 293 cells with GST-IRS-1(PTB)-Sepharose in the presence of BSA (50 mg) or bu€er (7) (right) or by incubating protein extracts (1 mg) with GST-Src(SH2)-Sepharose in the presence of soluble Shc(PTB) or bu€er (7)

1997) but not the ErbB4 or the EGF receptors (Wolf et al., 1995). Mutagenesis of its 18 potential tyrosine phosphorylation sites has shown that IRS-1 serves as a link between the activated receptor and PI3-K (Myers et al., 1996). In this study, we provide evidence that IRS-1 is a component of the signaling pathway that mediates

PI3-K stimulation by the Ret receptor. We found that IRS-1 is tyrosine-phosphorylated and associated with Ret and with the p85 subunit of PI3-K in cells expressing an active Ret. These e€ects are probably not restricted to IRS-1; indeed, we have found that another member of the same family, IRS-2, is tyrosine phosphorylated upon Ret activation (RM Melillo and M Santoro, unpublished). In addition to IRS proteins, we detected another tyrosine-phosphorylated protein in p85 immunoprecipitates with the apparent molecular mass of 120 kDa. This protein has been identi®ed previously by others as the Gab1 adaptor and has been implicated in Ret-mediated PI-3K activation (Murakami et al., 1999a,b). Both co-immunoprecipitation and far Westen experiments demonstrate that, upon Ret activation, Gab1 is a more abundant p85-interacting protein than IRS-1 at least in NIH3T3 cells. It is possible that Gab1 is expressed at higher levels than IRS-1 in NIH3T3 cells and/or that its stoichiometry of phosphorylation upon Ret activation is higher than that of IRS-1. It is likely that Ret uses redundant (or possibly cooperative) mechanisms to activate the PI3K/Akt pathway. In any case, the absence of any detectable direct Ret/p85 interaction points to the essential role played by docking proteins in PI3K recruitment to Ret. Binding of Gab family members to Ret is dependent on Y1062 and Y1096, a residue present only in the long isoform of the receptor (Hayashi et al., 2000; Besset et al., 2000). Indeed, Gab recruitment to Ret seems to be mediated by the Grb2 adaptor; Grb2, in turn, associates to RetY1096 in the long isoform of the receptor and to the ShcRetY1062 complex (Hayashi et al., 2000; Besset et al., 2000). We show that the PTB domain of IRS-1 is sucient to bind Ret in a pull-down assay. The Ret sequence which binds IRS-1 is located in the carboxy-terminal tail of the receptor. This region contains the NKLY motif, which is responsible for the binding of at least three proteins, Shc (Arighi et al., 1997; Asai et al., 1996; Lorenzo et al., 1997), Enigma (Durick et al., 1998), and IRS-1 (this study). While Shc and IRS-1 binding to the NKLY motif is phosphorylationdependent and is mediated by their PTB domains, the binding of Enigma to the same motif is mediated by a LIM domain and is not dependent on phosphorylation. It has been demonstrated that residues that are amino-terminal with respect to the phosphotyrosine are important for binding of speci®c PTB domains (Siegal, 1999). The binding of the Shc PTB domain requires the presence of an hydrophobic residue at position 75, while binding of the IRS-1 PTB domain requires some combinations of hydrophobic residues at positions 76, 77 or 78 (Wolf et al., 1995; Zhou et al., 1996; van der Geer et al., 1999). The NKLpY consensus of Ret could ful®l both requirements showing two hydrophobic residues, Ile and Trp, at positions 75 and 76, respectively, with respect to the phosphorylated tyrosine. We show that the disruption of the NKLY sequence, either by the deletion of N1059 or by the substitution of Y1062 with phenylalanine or L1061

215

Oncogene

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

216

with proline, impairs the binding of Ret to IRS-1. DelN1059 and L1061P have been identi®ed in HSCR patients and inactivate the ability of Ret to bind Shc and to transduce downstream signals (Geneste et al., 1999). It is tempting to speculate that the impaired recruitment of signaling molecules might be responsible of the HSCR disease phenotype in patients carrying these mutations. For instance, the impaired binding of IRS-1 can cause a reduced ability of activating the PI3K pathway and thus mediating neuronal cell survival. According to this hypothesis, we show that mutations of the NKLY motif impair Ret ability of activating the survival mediator Akt. Interestingly, while in most HSCR cases Ret mutations are found in the heterozygous state, the L1061P mutation, which causes only a partial defect in Shc (Geneste et al., 1999) and IRS-1 (this study) coupling, has been found in the homozygous state. The Ret NKLY consensus shows a remarkable ¯exibility, being able to bind several di€erent proteins. Similarly, in the case of the TrkA receptor the NPXY(490) sequence is involved in the binding of two adaptors, namely Shc (van der Geer et al., 1999) and FRS2 (Meakin et al., 1999). It has been shown that FRS2 and Shc compete for their binding to TrkA (Meakin et al., 1999). Here we show that IRS-1 and Shc compete for binding to Ret. This suggests a model in which competition between signaling intermediates may regulate Ret-dependent proliferation, di€erentiation and survival. Shc and IRS-1 are prevalently (although not exclusively) coupled to the Ras/MAPK (Bonni et al., 1999) and the PI3-K/Akt pathway (Datta et al., 1999), respectively. The activation of the Ras/ MAPK pathway is required for cellular proliferation and transformation. Neurotrophic receptors, such as Ret, have the ability of inducing a prolonged and intense activation of the Ras/MAPK pathway, which is required for the switch from mitogenesis to di€erentiation (Marshall, 1995). On the other hand, the PI3-K/ Akt pathway controls multiple biological responses such as mitogenesis, programmed cell death and cellular motility (Datta et al., 1999). The selection of a single Ret site for the binding of both Shc and IRS-1 and thus for conveying signals to the corresponding intracellular pathway might be useful to select the biological response triggered by the activation of the receptor.

Materials and methods Cell cultures and plasmids NIH3T3-E/R (expressing the EGFR/Ret chimeric receptor), NIH3T3-Ret (expressing the wild type Ret receptor) (Carlomagno et al., 1998) and NIH3T3-Ret/MEN2A cells (expressing the Ret(C634R) oncogene) have been described previously (Santoro et al., 1994, 1995) and were grown in Dulbecco's modi®ed Eagle's medium (DMEM) supplemented with 10% calf serum (GIBCO ± BRL, Gaithesburg, MD, USA). 293 cells were from ATCC and were grown in DMEM supplemented with 10% fetal calf serum. Expression vectors Oncogene

for Ret(C634R), i.e. Ret/MEN2A, and for the mutants Ret/ MEN2A(Y1062F), Ret/MEN2A(L1061P) and Ret/MEN2A(delN1059) have been previously described (Geneste et al., 1999). The kinase-dead Ret mutant, Ret(K-), in which the catalytic lysine was mutated to methionine (K758M), the Ret(C634R)4F, in which Y826, Y1015, Y1029 and Y1062 autophosphorylation residues were mutated to phenylalanine and the Ret(C634R)3F, in which only Y826, Y105 and Y1029 were mutated, were generated by site-directed mutagenesis using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA, USA). The mutations were con®rmed by DNA sequencing. Ha-Akt encodes an Ha epitope-tagged wild type Akt protein and was a kind gift of A Bellacosa and P Tsichlis (Bellacosa et al., 1998). The IRS-1 encoding construct, pcDNA3.1 IRS-1, contains the human IRS-1 gene cloned into the pCDNA3.1 vector (Invitrogen, Groningen, The Netherlands) and was a kind gift of D Accili (Voliovitch et al., 1995). Transient transfections Transient transfections were carried out by using the Lipofectamine reagent following manufacturer's instructions (GIBCO ± BRL). Brie¯y, 293 cells were seeded at a density of 1.56106 the day before transfection. Cells were then transfected with 5 mg of each Ret plasmid and, when required, 1 mg of IRS-1 and 1 mg of Ha-Akt and harvested 48 h later. Immunoprecipitation and immunoblotting NIH E/R cells were grown in 100 mm dishes until subcon¯uent and serum starved for 12 h. Cells were then stimulated with EGF (100 ng/ml) for 5 min, washed once with ice-cold phosphate-bu€ered saline (PBS), and subjected to lysis as previously described (Santoro et al., 1994). NIH3T3-Ret/MEN2A were serum starved before harvesting. Cell lysates were incubated with the indicated antibodies for 1 h at 48C. The immunocomplexes were collected with Protein A sepharose CL4B beads (Amersham Pharmacia Biotech, Uppsala, Sweden), immunoblotted on nitrocellulose membranes (Schleicher & Schuell Dassel/Relliehausen, Germany), and incubated with primary antibodies. Detection was achieved by enhanced chemiluminescence as suggested by the manufacturer (ECL, Amersham Pharmacia Biotech). AntiRet polyclonal antibodies have been developed as described (Santoro et al., 1994). Anti-p85, anti-p110, anti-IRS-1, and anti-phosphotyrosine antibodies were purchased from Upstate Biotechnology Inc. (Lake Placid, NY, USA). Anti-Akt, anti-phospho-Akt, anti-MAPK (#9101) and anti-phosphoMAPK (#9102) antibodies were from New England Biolabs Inc. (Beverly, MA, USA). Anti-Grb2 and anti-Shc antibodies were from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Binding assays and far Western The GST-IRS-1(PTB) fusion protein was generated by PCR ampli®cation of the PTB domain of IRS-1 (amino acids 131 ± 331). The PCR product was digested with EcoRI and BamHI, subcloned in the pGEX2T vector (Amersham Pharmacia Biotech), and con®rmed by sequencing. The GST-Shc(PTB) and the GST-Src(SH2) fusion proteins have been described previously (Geneste et al., 1999; Melillo et al., 1999). GST p85/N-SH2 and GST p85/C-SH2 proteins were a kind gift of Dr G Pelicci. Bacterially expressed fusion proteins were produced using standard protocols. For binding assays, cell

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

lysates from transiently transfected 293 cells were incubated with 5 mg of immobilized fusion proteins. Bound proteins were detected by immunoblot analysis with speci®c antibodies. For competition experiments GST-Shc(PTB)-bound Sepharose was digested with 10 units of thrombin overnight at room temperature. The digests were washed twice with PBS containing protease inhibitors and 1 mM EDTA and pooled, and the eluate was used directly. Peptide inhibition assays were performed as above, except that the proteins were incubated in the presence of increasing concentrations of phosphorylated or unphosphorylated peptide. Synthetic peptides were generated using the F-moc synthesis method (Neosystem Groupe SNPE, Strasbourg, France). The two peptides used in these assays were RET pY1062 (TWIENKL(Y)GRIS) and RET ppY1062 (TWIENKL(pY)GRIS). The far Western (immunoblot) experiment was performed as described elsewhere (Matoskova et al., 1995). Brie¯y, blots were blocked in 2% bovine serum albumin (BSA) in TTBS (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.05% Tween 20) for at least 2 h at room temperature and then in reduced glutathione (3 mM) in TTBS with 0.5% BSA (wt/vol) for 1 h at room temperature. Blots were then incubated with the GST p85/N-SH2 and GST p85/C-SH2 proteins (10 nM) in TTBS in the presence of reduced glutathione and BSA. After extensive washing in TTBS, blots were detected with the anity puri®ed anti-GST antibody. PI3-Kinase assay In vitro phosphorylation of PI was carried out in immunocomplexes, as described (Segoun-Cariou and Billaud, 2000). Brie¯y, 293 cells were transiently transfected with Ret(K-) or

Ret/MEN2A either in the presence or in the absence of IRS1. Cells were serum-deprived 12 h before harvesting, and then lysed. Protein extracts (400 mg) were immunoprecipitated with anti-IRS-1 antibodies. Immunocomplexes were washed and incubated with 10 mg phosphoinositides (Sigma, St Louis, MO, USA) and 10 mCi [g-32P]-ATP for 15 min at 228C. The reactions were stopped with 80 ml of chloroform-methanol (1 : 1), centrifuged, and the lower organic phase was applied to a Silica gel thin-layer chromatography plate (Merck, Darmstadt, Germany) in chloroform/methanol/25%NH4OH/ water (43 : 38 : 5 : 7). Chromatography plates were dried, visualized by autoradiography and quantitated by a Molecular Dynamics Phosphorimager.

217

Acknowledgments We thank A Bellacosa and P Tsichlis for the Ha-Akt and D Accili for the IRS-1 plasmid. We are grateful to F Sferratore and AM Cira®ci for technical assistance. This study was supported by the Associazione Italiana per la Ricerca sul Cancro (AIRC), the EC grants BMH4-CT960814 and BMH4-CT97-2157, grants from the MURST, the progetto Biotecnologie 5% of the Consiglio Nazionale delle Ricerche (CNR), and the Ligue Nationale Contre le Cancer. F Carlomagno was supported by a fellowship of the Telethon Foundation. This paper was written while G Vecchio was a Scholar-in-Residence at the Fogarty International Center for Advanced Study in the Health Sciences, National Institutes of Health, Bethesda, MD, USA.

References Airaksinen MS, Titievsky A and Saarma M. (1999). Mol. Cell. Neurosci., 13, 313 ± 325. Alberti L, Borrello MG, Ghizzoni S, Torriti F, Rizzetti MG and Pierotti MA. (1998). Oncogene, 17, 1079 ± 1087. Arighi E, Alberti L, Torriti F, Ghizzoni S, Rizzetti MG, Pelicci G, Pasini B, Bongarzone I, Piutti C, Pierotti MA and Borrello MG. (1997). Oncogene, 14, 773 ± 782. Asai N, Murakami H, Iwashita T and Takahashi M. (1996). J. Biol. Chem., 271, 17644 ± 17649. Bellacosa A, Chan TO, Ahmed NN, Datta K, Malstrom S, Stokoe D, McCormick F, Feng J and Tsichlis P. (1998). Oncogene, 17, 313 ± 325. Besset V, Scott RP and Ibanez CF. (2000). J. Biol. Chem. (in press). Bonni A, Brunet A, West AE, Datta SR, Takasu MA and Greenberg ME. (1999). Science, 286, 1358 ± 1362. Borrello MG, Alberti L, Arighi E, Bongarzone I, Battistini C, Bardelli A, Pasini B, Piutti C, Rizzetti MG, Mondellini P, Radice MT and Pierotti MA. (1996). Mol. Cell. Biol., 16, 2151 ± 2163. Carlomagno F, Melillo RM, Visconti R, Salvatore G, De Vita G, Lupoli G, Yu Y, Jing S, Vecchio G, Fusco A and Santoro M. (1998). Endocrinology, 139, 3613 ± 3619. Datta SR, Brunet A and Greenberg ME. (1999). Genes Dev., 13, 2905 ± 2927. De Vita G, Melillo RM, Carlomagno F, Visconti R, Castellone MD, Bellacosa A, Billaud M, Fusco A, Tsichlis PN and Santoro M. (2000a). Cancer Res., 60, 3727 ± 3731.

De Vita G, Berlingieri MT, Visconti R, Castellone MD, Viglietto G, Baldassarre G, Zannini MS, Bellacosa A, Tsichlis PN, Fusco A and Santoro M. (2000b). Cancer Res., 60, 3916 ± 3920. Durick K, Gill GN and Taylor SS. (1998). Mol. Cell. Biol., 18, 2298 ± 2308. Geneste O, Bidaud C, Vita GD, Hofstra RM, TartareDeckert S, Buys CH, Lenoir GM, Santoro M and Billaud M. (1999). Hum. Mol. Genet., 8, 1989 ± 1999. Johnston JA, Wang LM, Hanson EP, Sun XJ, White MF, Oakes SA, Pierce JH and O'Shea JJ. (1995). J. Biol. Chem., 270, 28527 ± 28530. Hayashi H, Ichihara M, Iwashita T, Murakami H, Shimono Y, Kawai K, Kurokawa K, Murakumo Y, Imai T, Funahashi H, Nakao A and Takahashi M. (2000). Oncogene, 19, 4469 ± 4475. Keegan AD, Nelms K, White M, Wang LM, Pierce JH and Paul WE. (1994). Cell, 76, 811 ± 820. Liang L, Zhou T, Jiang J, Pierce JH, Gustafson TA and Frank SJ. (1999). Endocrinology, 140, 1972 ± 1983. Liu X, Vega QC, Decker RA, Pandey A, Worby CA and Dixon JE. (1996). J. Biol. Chem., 271, 5309 ± 5312. Lorenzo MJ, Gish GD, Houghton C, Stonehouse TJ, Pawson T, Ponder BA and Smith DP. (1997). Oncogene, 14, 763 ± 771. Marshall CJ. (1995). Cell, 80, 179 ± 185. Matoskova B, Wong WT, Salcini AE, Pelicci PG and Di Fiore PP. (1995). Mol. Cell. Biol., 15, 3805 ± 3812.

Oncogene

Ret recruits PI3-Kinase through the insulin receptor substrate-1 (IRS-1) RM Melillo et al

218

Oncogene

Meakin SO, MacDonald JI, Gryz EA, Kubu CJ and Verdi JM. (1999). J. Biol. Chem., 274, 9861 ± 9870. Melillo RM, Barone MV, Lupoli G, Cira®ci AM, Carlomagno F, Visconti R, Matoskova B, Di Fiore PP, Vecchio G, Fusco A and Santoro M. (1999). Cancer Res., 59, 1120 ± 1126. Murakami H, Iwashita T, Asai N, Iwata Y, Narumiya S and Takahashi M. (1999a). Oncogene, 18, 1975 ± 1982. Murakami H, Iwashita T, Asai N, Shimono Y, Iwata Y, Kawai K and Takahashi M. (1999b). Biochem. Biophys. Res. Commun., 262, 68 ± 75. Myers Jr MG, Grammer TC, Wang LM, Sun XJ, Pierce JH, Blenis J and White MF. (1994). J. Biol. Chem., 269, 28783 ± 28789. Myers Jr MG, Zhang Y, Aldaz GA, Grammer T, Glasheen EM, Yenush L, Wang LM, Sun XJ, Blenis J, Pierce JH and White MF. (1996). Mol. Cell. Biol., 16, 4147 ± 4155. Pandey A, Duan H, Di Fiore PP and Dixit VM. (1995). J. Biol. Chem., 270, 21461 ± 21463. Pandey A, Liu X, Dixon JE, Di Fiore PP and Dixit V. (1996). J. Biol. Chem., 271, 10607 ± 10610. Pasini B, Ceccherini I and Romeo G. (1996). Trends Genet., 12, 138 ± 144. Pawson T and Scott JD. (1997). Science, 278, 2075 ± 2080. Pelicci G, Lanfrancone L, Grignani F, McGlade J, Cavallo F, Forni G, Nicoletti I, Grignani F, Pawson T and Pelicci PG. (1992). Cell, 70, 93 ± 104. Pierotti MA, Bongarzone I, Borello MG, Greco A, Pilotti S and Sozzi G. (1996). Genes Chrom. Cancer, 16, 1 ± 14. Ponder BA. (1999). Cancer Res., 59, 1736 ± 1741. Rameh LE and Cantley LC. (1999). J. Biol. Chem., 274, 8347 ± 8350. Rosenthal A. (1999). Neuron, 22, 201 ± 203. Santoro M, Carlomagno F, Romano A, Bottaro DP, Dathan NA, Grieco M, Fusco A, Vecchio G, Matoskova B, Kraus MH and Di Fiore PP. (1995). Science, 267, 381 ± 383.

Santoro M, Wong WT, Aroca P, Santos E, Matoskova B, Grieco M, Fusco A and Di Fiore PP. (1994). Mol. Cell. Biol., 14, 663 ± 675. Schlessinger J. (1993). Trends Biochem. Sci., 18, 273 ± 275. Schlessinger J. (1994). Curr. Opin. Genet. Dev., 4, 25 ± 30. Segoun-Cariou C and Billaud M. (2000). J. Biol. Chem., 275, 3568 ± 3576. Siegal G. (1999). Nat. Struct. Biol., 6, 7 ± 10. Skolnik EY, Batzer A, Li N, Lee CH, Lowenstein E, Mohammadi M, Margolis B and Schlessinger J. (1993). Science, 260, 1953 ± 1955. Trupp M, Scott R, Whittemore SR and Ibanez CF. (1999). J. Biol. Chem., 274, 20885 ± 20894. Uddin S, Yenush L, Sun XJ, Sweet ME, White MF and Platanias LC. (1995). J. Biol. Chem., 270, 15938 ± 15941. van der Geer P, Wiley S and Pawson T. (1999). Oncogene, 18, 3071 ± 3075. van Weering DH, de Rooij J, Marte B, Downward J, Bos JL and Burgering BM. (1998). Mol. Cell. Biol., 18, 1802 ± 1811. Voliovitch H, Schindler DG, Hadari YR, Taylor SI, Accili D and Zick Y. (1995). J. Biol. Chem., 270, 18083 ± 18087. Wang LM, Myers Jr MG, Sun XJ, Aaronson SA, White M and Pierce JH. (1993). Science, 261, 1591 ± 1594. White MF. (1998). Recent Prog. Horm. Res., 53, 119 ± 138. Wolf G, Trub T, Ottinger E, Groninga L, Lynch A, White MF, Miyazaki M, Lee J and Shoelson SE. (1995). J. Biol. Chem., 270, 27407 ± 27410. Yamada M, Ohnishi H, Sano S, Nakatani A, Ikeuchi T and Hatanaka H. (1997). J. Biol. Chem., 272, 30334 ± 30339. Yin T, Keller SR, Quelle FW, Witthuhn BA, Tsang ML, Lienhard GE, Ihle JN and Yang YC. (1995). J. Biol. Chem., 270, 20497 ± 20502. Zhou MM, Huang B, Olejniczak ET, Meadows RP, Shuker SB, Miyazaki M, Trub T, Shoelson SE and Fesik SW. (1996). Nat. Struct. Biol., 3, 388 ± 393.