Oncogene (2000) 19, 1509 ± 1518 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Sustained recruitment of phospholipase C-g to Gab1 is required for HGF-induced branching tubulogenesis

Philippe Gual*,1, Silvia Giordano1, Tracy A Williams1, Stephane Rocchi2, Emmanuel Van Obberghen2 and Paolo M Comoglio1

1Institute for Cancer Research and Treatment (IRCC), University of Torino Medical School, Str. Prov. 142, Km 3.95, 10060 Candiolo, Italy; 2Institut National de la Sante et de la Recherche (INSERM) U145, Faculte de MeÂdecine, avenue de Valombrose 06107, Nice ceÂdex 2, France

A distinctive property of Hepatocyte Growth Factor terminal domain of the b-subunit contains a two- (HGF) is its ability to induce di€erentiation of tubular tyrosine docking site (tyrosines 1349 and 1356) structures from epithelial and endothelial cells (branching (Ponzetto et al., 1994), which, upon autophosphoryla- tubulogenesis). The HGF receptor directly activates PI3 tion, binds multiple SH2-containing transducers, in- kinase, Ras and STAT signalling pathways and cluding phosphatidylinositol 3-kinase (PI 3-kinase) phosphorylates the adaptator GRB2 Associated Binder- (Graziani et al., 1991), the transcriptional factor STAT 1 (Gab1). Gab1 is also phosphorylated in response to (Boccaccio et al., 1998), the docking protein Gab1 Epidermal Growth Factor (EGF) but is unable to induce (Weidner et al., 1996) and the adaptators SHC (Pelicci tubule formation. Comparison of 32P-peptide maps of et al., 1995) and GRB2 (Ponzetto et al., 1994). A weak Gab1 from EGF- versus HGF-treated cells, demon- direct binding of phospholipase C-g 1 (PLC-g 1) was strates that the same sites are phosphorylated in vivo. also found (Ponzetto et al., 1994). However, while both EGF and HGF induce rapid Gab1 belongs to the insulin receptor substrate-1 tyrosine phosphorylation of Gab1 with a peak at (IRS-1) -like family, that also comprises IRS-2/3, 15 min, the phosphorylation persists for over 1 h, only p62DOK, DOS, Gab2, Dok-R and FRS2 (Herbst et in response to HGF. Nine tyrosines are phosphorylated al., 1996; Raabe et al., 1996; Yamanashi and by both receptors. Three of them (Y307, Y373, Y407) Baltimore, 1997; Carpino et al., 1997; Gu et al., bind phospholipase C-g (PLC-g). Interestingly, the 1998; Jones and Dumont, 1998; Kouhara et al., 1997; overexpression of a Gab1 mutant unable to bind PLC- Yenush and White, 1997). Upon receptor-induced g (Gab1 Y307/373/407F) did not alter HGF-stimulated tyrosine phosphorylation, these substrates act as cell scattering, only partially reduced the growth docking proteins for several SH2-containing transdu- stimulation but completely abolished HGF-mediated cers. It has been shown that Gab1 is phosphorylated tubulogenesis. It is concluded that sustained recruitment in response to insulin (Holgado-Madruga et al., 1996), of PLCg to Gab1 plays an important role in branching a number of growth factors (EGF, HGF, NGF) tubulogenesis. Oncogene (2000) 19, 1509 ± 1518. (Weidner et al., 1996) (Holgado-Madruga et al., 1996, 1997) and cytokines (IL 3 and 6, Interferon a and b, Keywords: PLC-g; Gab1; HGF; tubulogenesis erythropoietin, thrombopoietin, granulocyte-macro- phage colony-stimulating factor, stem cell factor) (Takahashi-Tezuka et al., 1998; Lecoq-Lafon et al., Introduction 1999; Nishida et al., 1999) and T and B cell antigens (Nishida et al., 1999). Gab1 contains a pleckstrin The biological e€ects of hepatocyte growth factor homology (PH) domain, a proline rich region that (HGF) are mediated by Met, its speci®c tyrosine kinase includes two optimal binding sites for the SH3 receptor. As a result of stimulation of cell-dissociation, domains of GRB2 and 16 potential tyrosine-phos- motility and polarization, HGF induces the formation phorylation sites (Holgado-Madruga et al., 1996). In of three-dimensional branched tubular structures vivo, the HGF-receptor preferentially recruits Gab1 (Montesano et al., 1991). Tubulogenesis is not elicited via a bridge with GRB2 (Nguyen et al., 1997; Bardelli by other tyrosine kinase receptors, such as the et al., 1997a). Upon HGF or EGF stimulation, epidermal growth factor (EGF) receptor which is GRB2, PLC-g 1, SHPTP 2 and PI3 kinase associate expressed and functionally active in most cells to the respective receptors via Gab1 (Weidner et al., responsive to HGF. 1996; Holgado-Madruga et al., 1996; Bardelli et al., Met displays a dimeric structure consisting of an 1997a). extracellular a-subunit disulphide linked to a trans- With the aim of understanding why branching membrane b-subunit. Interaction of the ligand with the tubulogenesis is induced by HGF and not by EGF, receptor stimulates b-subunit tyrosine kinase activity, we have investigated the role of Gab1 in the induction leading to autophosphorylation of the receptor and of this speci®c response. It has been previously shown tyrosine phosphorylation of cellular substrates. The C- that overexpression of Gab1 alone is sucient for the induction of this speci®c response (Weidner et al., 1996) and that the disruption of Gab1 binding with Met abrogated the HGF-mediated branching tubulo- *Correspondence: P Gual Received 27 September 1999; revised 20 January 2000; accepted 8 genesis response (Weidner et al., 1996; Maroun et al., February 2000 1999). On the other hand, Gab1 is also phosphorylated PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1510 by the EGF receptor (Weidner et al., 1996), but this The comparison of the tryptic-peptide maps of GST- event is not followed by branching tubulogenesis. Gab1 phosphorylated in vitro either by the HGF or In the present study, we show that in the response to EGF receptors also showed no marked qualitative HGF or EGF, Gab1 is phosphorylated in vivo on the di€erence, indicating that Gab1 phosphorylated by same residues. However, a sustained activation of either receptor could recruit the same set of SH2 signaling pathways downstream to Gab1 (as a result containing transducers (data not shown). To identify of its sustained phosphorylation) is achieved only in the phosphorylated tyrosine residues, we performed response to HGF. We also show that, among these kinase assays with puri®ed receptors (immunoprecipi- pathways, the recruitment of PLC-g 1 to Gab1 while tated from overexpressing cells) and 16 potential dispensable for cell scattering and growth, is an peptide substrates corresponding to di€erent domains absolute requirement for HGF-mediated-tubulogenesis. of Gab1. Each peptide contained one putative tyrosine phosphorylation site. Among 16 peptides only nine were phosphorylated by the EGF and HGF receptors, namely peptides containing the tyrosine residues 285, Results 307, 373, 407, 448, 473, 590, 628 and 660 (Table 1). No striking di€erences in the level of peptide phosphoryla- Gab-1 is phosphorylated on the same residues by HGF and tion by both kinases were observed. Some of these EGF receptors peptides contain serine and/or threonine residues; It has been previously shown that overexpression of however, the peptide phosphorylation detected is not Gab1 alone is sucient for the induction branching due to the presence of a serine/threonine kinase tubulogenesis (Weidner et al., 1996). The fact that both associated with the receptor since we have previously the HGF receptor and the EGF receptor phosphor- shown that under these experimental conditions, serine ylate Gab1 but only HGF mediates branching and threonine phosphorylation of Gab1 are not tubulogenesis suggests that Gab1 could play a di€erent detected (Bardelli et al., 1997a). However, among these role in signal transduction for each receptor. To nine peptides, the peptides containing tyrosines 628 address this question, we compared the pattern of and 660 were phosphorylated to a higher extent by the Gab1 phosphorylation induced in response to HGF HGF receptor. Tyrosines 448, 473 and 590 have been versus that induced by EGF. For this purpose, we shown to bind the p85 protein (Rocchi et al., 1998; performed tryptic-peptide maps of Gab1 phosphory- Maroun et al., 1999), whereas tyrosine residue 628 lated in vivo either in response to HGF or EGF. binds the phosphatase SHPTP 2 (Rocchi et al., 1998; Comparison of the di€erent phosphopeptide maps Lehr et al., 1999). Phosphotyrosines in the Y-D-X-P showed no marked qualitative di€erences (Figure 1). and Y-X-I-P motifs are potential binding sites for Nck This indicates that Gab1 phosphorylation induced by SH2 and PLC-g 1 domains, respectively. Thus, HGF and EGF occurs at the same residues. tyrosines Y307 (Y-D-I-P) and Y407 (Y-D-I-P) are potential binding sites for both PLC-g 1 and Nck proteins, while tyrosine Y373 (Y-C-I-P) constitutes a potential binding site for PLC-g 1 alone. Proteins able to bind phosphotyrosines 285 and 660 have not yet been identi®ed.

Table 1 Nine synthetic peptides corresponding to di€erent domains of Gab1 and containing a single tyrosine residue are phosphorylated by both HGF and EGF receptors Phosphorylation assay with: Peptide amino Phosphorylated HGF EGF acid sequence residue receptor receptor

ADGELYTFNTP Y285 260 189 HVSISYDIPPT Y307 154 170 DTDSSYCIPPP Y373 186 219 SSQDCYDIPRT Y407 260 169 Figure 1 Phosphopeptide map analysis of Gab1. COS-7 cells ELDENYVPMNP Y448 233 258 overexpressing wild type Gab1 were labeled with 32P-orthopho- IQEPNYVPMTP Y473 222 227 sphate for 3 h in DMEM without phosphate and then stimulated DSEENYVPMNP Y590 174 150 with EGF (100 ng/ml) or HGF (100 units/ml) for 10 min. DKQVEYLDLDL Y628 543 610 Phosphorylated Gab1 was immunoprecipitated with speci®c DERVDYVVVDQ Y660 476 235 antibodies. Pellets were washed with lysis bu€er, resuspended in Laemmli sample bu€er and separated by SDS ± PAGE using an Clari®ed lysates from COS-7 cells overexpressing either the HGF or 8% resolving gel. 32P-Gab1 was extracted from SDS ± PAGE gels EGF receptors were incubated with their respective antibodies or and treated with trypsin. 32P-peptides were then separated by two- with non-immune serum preadsorbed on protein G-Sepharose. After dimensional electrophoretic analysis on silica thin-layer plates. 2 h at 48C, the pellets were washed twice and the peptides (250 mM) The plates were dried and subjected to autoradiography. Samples and the phosphorylation bu€er were added. The phosphorylation from HGF and EGF treated cells are shown in panels A and B, reactions were stopped after 45 min by spotting a constant volume of respectively. In panel C, both samples were mixed before being each sample onto phosphocellulose papers, which were dipped into analysed. A representative experiment is shown that was 1% (vol/vol) orthophosphoric acid. The values given represent the performed in duplicate. The directions of electrophoresis and means of two experiments and are expressed as the percentage of the chromatography are indicated and the points of sample origin are basal peptide phosphorylation (in the presence of non-immune indicated by arrows serum)

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1511 Taken together, these results indicate that the EGF kinase-dependent AKT (Franke et al., 1997). There- and HGF receptors phosphorylate the same tyrosine fore, we examined AKT activation levels in response residues on Gab1 and can potentially recruit the same to HGF and EGF stimulation. The total cell lysates set of SH2 or PTB containing proteins. were probed with a speci®c anti-phospho AKT antibody and total levels of AKT were determined by reprobing the same membrane with anti-AKT HGF induces sustained Gab-1 phosphorylation antibodies. As for Gab1 phosphorylation, EGF We then stimulated epithelial cells with HGF and induced a transient AKT activation, in contrast to EGF to assess whether these growth factors could HGF, which produced sustained activation of AKT induce di€erent time-courses of Gab1 tyrosine phos- for over 60 min (Figure 2b). The Gab1 phosphoryla- phorylation. The immunoprecipitated Gab1 was tion state and AKT activation, in response to HGF analysed by anti-phosphotyrosine Western blotting. were returned to basal levels within 180 min. The total level of Gab1 protein was determined by Thus, while both EGF and HGF promote Gab1 stripping the membrane and reprobing with speci®c phosphorylation, only HGF can sustain a long-lasting antibodies. The time-course of phosphorylation dif- activation of signaling pathways downstream of fered according to the stimulus (Figure 2a). In Gab1. response to EGF, we observed transient phosphoryla- tion of Gab1 that returned to basal levels within The phosphorylation of tyrosines Y307, Y373, Y407 in 60 min; in contrast, in response to HGF, the Gab1 generates specific docking sites for PLC-g 1 phosphorylation state of Gab1 was sustained for 60 min before decreasing. Since Gab1 plays a critical As mentioned, it is known that upon autophosphoryla- role in HGF-mediated PI3 kinase activation (Maroun tion, tyrosine residues 448, 473 and 590 are docking et al., 1999), sustained phosphorylation of Gab1 sites for the PI3 kinase adaptator p85 (Rocchi et al., should result in the prolonged activation of the PI3 1998; Maroun et al., 1999), whereas tyrosine residue 628 is the docking site for the phosphatase SHPTP 2 (Rocchi et al., 1998; Lehr et al., 1999). To identify signal transducer(s) binding to Gab1 at the tyrosine residues 307, 373 and 407, we replaced them with phenylalanines in di€erent combinations (Y307/373F, Y307/407F, Y373/407F and Y307/373/407F). Wild type and mutated forms of Gab1 were phosphorylated by a constitutively active form of Met (Tpr-Met). Proteins were pulled-down after interaction with PLC-g 1 (GST- N/C SH2 domains of PLC-g 1). As shown in Figure 3a, Gab1 wild type was very eciently pulled down, showing strong interaction with PLC-g 1. In contrast, the Gab1 triple mutant, Gab1 Y307/373/407F, here referred to as Gab1 DPLC-g, did not interact with GST- PLC-g 1. Finally, the Gab1 double mutants, Gab1 Y307/373F, Y307/407F and Y373/407F, were pulled- down to a lesser extent compared to Gab1 Wt. This indicates that tyrosine Y407 constitutes the major site for PLC-g 1 association of Gab1 and that Gab1 triple mutant is no longer able to recruit PLC-g 1. To con®rm in vivo the absence of PLC-g 1 recruitment by the Gab1 triple mutant, we measured the association of PLC-g with either Gab1 Wt or DPLC-g from HGF-treated cells. The presence of PLC-g associated with Gab1 was detected only with Gab1 Wt (Figure 3b). The phosphorylation state of PLC-g associated to Gab1 was also determined by reprobing the same membrane Figure 2 HGF, but not EGF, induces sustained Gab1 and with anti phosphotyrosine antibodies (data not shown). AKT phosphorylation. (a) one aliquot of each clari®ed lysate Since PLCg activation is correlated to its phosphoryla- (700 mg of total protein) was used for immunoprecipitation of tion (Kim et al., 1991), this indicates that PLCg Gab1 followed by anti-phosphotyrosine Western blotting. The recruited to Gab1 is activated. nitrocellulose membrane was subsequently stripped and re- probed with anti-Gab1 antibodies. Phosphorylated Gab1 and Tyrosines Y307 and Y407 are also potential binding total Gab1 were quanti®ed by densitometry using Molecular sites for Nck proteins. To determine whether the Nck Dynamics ImageQuant software. Data are expressed as the protein constitutes a potential partner of Gab1, we relative increase of phosphorylated Gab1 over basal levels and performed co-immunoprecipitation experiments with are normalized for amounts of total Gab1. (b) a second aliquot (100 mg of total protein) of each total cell lysate was analysed by speci®c antibodies and lysates from 293 cells co- anti-phospho AKT Western blotting. The nitrocellulose mem- overexpressing Gab1 Wt and Tpr-Met. No co- brane was subsequently stripped and re-probed with anti-AKT immunoprecipitation was detected (Figure 3c), indicat- antibodies. Phosphorylated AKT and total AKT were quanti®ed ing that Nck did not associate either with Tpr-Met or as described above and data are expressed as the relative increase of Phospho-AKT over basal levels normalized for total with Gab1. This indicates that phosphorylated tyr- AKT. In both a and b, each point is the mean of three osines Y307, Y373 and Y407 bind PLC-g 1 but not independent experiments+s.e.mean Nck.

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1512

Figure 3 The tyrosines 307, 373 and 407 of Gab1 generate docking sites for PLC-g but not for Nck. (a) Clari®ed lysates from COS-7 cells expressing Tpr-Met and Gab1 wild type or mutated forms (as indicated) were incubated with GST, GST-C/N SH2 domains of PLC-g (5 mg) preadsorbed on glutathione-Sepharose. After 2 h at 48C, the pellets were washed twice. These samples were analysed by SDS ± PAGE using an 8% resolving gel and transferred to nitrocellulose ®lters for probing with anti-Met or anti- Gab1 antibodies. We show a representative experiment that had been performed in triplicate. (b) and (c): Co-immunoprecipitation assays with 293 cells were performed as described in Materials and methods

Gab1 DPLC-g abolishes HGF-mediated branching Clonal MDCK cell lines expressing either MYC-tagged tubulogenesis wild type Gab1 or MYC-tagged triple mutant (Gab1 DPLC-g) were selected and characterized for Gab1 MDCK cells in culture form tight epithelial mono- expression, cell scattering, cell growth, and tubulogen- layers with junctional complexes. The addition of HGF esis. induces the disruption of intercellular junctions and Four independent clones expressing wild type and stimulates cell motility (Stoker, 1989). Moreover, in four clones expressing DPLC-g Gab1 were isolated response to HGF, MDCK cells migrate in tridimen- (Figure 4a). The ability of the clones to induce the cell sional collagen gel and form long and branched tubules scattering response to HGF was measured. As shown (Montesano et al., 1991). Since Gab1 plays an in Figure 4b, the expression of the Gab1 DPLC-g important role in these morphogenic e€ects (Weidner mutant did not alter the cell scattering response in the et al., 1996; Maroun et al., 1999), we investigated the presence of HGF. In agreement with this data, the role of the recruitment of the PLC-g 1. To do this, we incubation of the MCDK cells with a speci®c PLC mutated the speci®c sites for PLC-g 1 binding, Y307, inhibitor (U-73122) did not alter cell scattering in the Y373 and Y407 by replacement with phenylalanines. presence of HGF.

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1513 The cell growth of these clones was also evaluated. tubules, whereas cells expressing Gab1 DPLC-g formed As shown in Figure 4c, while the overexpression of only cysts. Most of the Gab1 DPLC-g cells grown in Gab1 wild type was ine€ective, overexpression of Gab1 collagen died in the absence of HGF. As expected, DPLC-g only partially reduced the slope of the growth tubulogenesis was speci®c to HGF stimulation: indeed, curve in response to HGF. This indicates that PLC-g control cells and cells expressing either Gab1 Wt or recruitment by Gab1 contributes, but is not an Gab1 DPLC-g formed only cysts in response to EGF absolute requirement, for HGF-mediated cell growth. (Figure 5a). To demonstrate that indeed PLC-g Finally, we tested the ability of these cells to form activation is required for tubule formation, MDCK tubules in collagen. In the presence of HGF, control cells were incubated with a PLC inhibitor (U-73122) cells and cells expressing wild type Gab1 formed either in the absence or in the presence of HGF. As

Figure 4 Expression of Gab1 DPLC-g in MDCK cells partially reduces the cell growth without altering cell scattering in response to HGF. (a) Clari®ed lysates from clonal cell lines of MDCK cells stably expressing either MYC-tagged wild type or MYC-tagged triple mutant (Gab1 DPLC-g) were incubated with an antibody to MYC preadsorbed on protein G-Sepharose. After 4 h at 48C, the pellets were washed twice and samples were analysed by Western blotting with anti-Gab1 antibodies. Cell scattering (b) and cell growth (c) assays with MDCK clones were performed as described in Materials and methods. A representative experiment for one clone of each group is shown

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1514

Figure 5 Expression of Gab1 DPLC-g in MDCK cells impairs HGF-mediated branching tubulogenesis. MDCK cell lines were grown in collagen for 12 h. HGF (100 units/ml), EGF (100 ng/ml) and U73122 (2 mM in 0.1% of DMSO) were added as indicated. The medium was changed every two days. After 7 ± 10 days the branching tubulogenesis responses were evaluated. (a) shows a representative experiment for one clone of each group. (b) shows a representative experiment that had been performed in triplicate

shown in Figure 5b, HGF stimulation in the presence divided into three phases: the ®rst (cell scattering) is of the inhibitor resulted in a signi®cant inhibition of most fully dependent on PI3 kinase and Rac activation branching tubulogenesis. Since we did not observe a (Graziani et al., 1991; Ridley et al., 1995; Royal and complete inhibition, this raises the possibility that Park, 1995; Khwaja et al., 1998); the second (cell another protein, apart from PLCg, could bind to Y307, growth) requires stimulation of the ras-MAP kinase 373, 407 and play some role in tubulogenesis. None- cascade (Ponzetto et al., 1996) and the third phase theless, these results show that recruitment of PLC-g1 (induction of cell polarity and tubulogenesis) is via Gab1 plays a major role in HGF-dependent dependent on the STAT pathway (Boccaccio et al., branching tubulogenesis. 1998). Recent studies have shown that overexpression of the docking protein Gab1 induces constitutive, ligand-independent, branching tubulogenesis (Weidner Discussion et al., 1996). Under physiological conditions, Gab1 is recruited and phosphorylated by the HGF receptor In epithelial cells, activation of the HGF receptor (Weidner et al., 1996); disruption of this association induces a speci®c response, called branching tubulo- impairs HGF-mediated branching tubulogenesis (Mar- genesis, which is not elicited by other tyrosine kinases oun et al., 1999). These data indicate that the such as the EGF receptor. The mechanisms controlling activation of one or more Gab1-dependent signaling this complex HGF-mediated response have not yet pathways is required for this HGF response. It has been fully elucidated. However this process can be been shown that disruption of the Gab1 associated PI3

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1515 kinase activity did not completely abolish tubule sponses. PLC-g has been involved in growth factor formation upon HGF receptor activation, suggesting induced mitogenesis, di€erentiation, development, me- that other signaling pathways involving Gab1 are tabolism, secretion, contraction and sensory perception required for ecient tubulogenesis (Maroun et al., (Wang et al., 1998; Noh et al., 1995; Thackeray et al., 1999). Gab1, however, is also phosphorylated by the 1998). Recently, it has been shown that activation of EGF receptor (Holgado-Madruga et al., 1996) but this PLC-g pathways is essential for triggering nerve growth is not followed by branching tubulogenesis. cone guidance (Ming et al., 1999). PLC-g 1 forms One hypothesis to explain this apparent paradox could stable complexes with tyrosine kinase receptors such as be that the HGF and the EGF receptors phosphorylate the EGF and PDGF receptors via its SH2 domains Gab1 on di€erent tyrosine residues, thus recruiting (Anderson et al., 1990; Margolis et al., 1990). After di€erent SH2- or PTB-containing transducers and activation by tyrosine phosphorylation (Kim et al., activating speci®c signaling pathways. This, however, 1991), PLC-g hydrolyzes phosphatidylinositol-4-5-bis- does not seem to be the case, since we have found that the phosphate to form inositol-1,4,5 triphosphate (IP3) and residues of Gab1 phosphorylated in response to EGF diacylglycerol (DAG). IP3 stimulates the release of and HGF are identical. Nine domains of Gab1 contain- calcium from internal stores (Lipp and Niggl, 1996; ing a single tyrosine are phosphorylated in vitro by both Berridge et al., 1998; Nishizuka, 1992) and DAG kinases. Three tyrosines (Y448, Y473, Y590) bind p85, activates protein kinase C (PKC) (Wahl and Carpenter, the regulatory subunit of PI3 kinase, while Y628 binds 1991). Here, we show that HGF-receptor recruits the tyrosine phosphatase SHPTP 2 (Rocchi et al., 1998; indirectly PLC-g. Interestingly we show that PLC-g is Lehr et al., 1999). In addition, sequence analysis revealed not required for cell scattering or motility. Activation potential binding sites for Nck (Y307, Y407) and for of the PI3 kinase, Ras and Rac pathways are thus PLC-g 1 (Y307, Y373, Y407). Despite the report that sucient for cell motility in response to HGF Nck is phosphorylated in response to HGF in A549 cells (Takaishi et al., 1994; Hartmann et al., 1994; Ridley (Kochhar and Iyer, 1996), we could not detect the et al., 1995; Royal and Park, 1995; Royal et al., 1997; formation of Nck/Gab1, Nck/Tpr-Met or Nck/Met Bardelli et al., 1997b). Moreover, the recruitment of complexes. This suggests that Nck does not interact PLC-g via Gab1 contributes to cell growth but is not directly with Met or Gab1. Nck phosphorylation in an absolute requirement. response to HGF could be due to a downstream kinase. Since cells expressing the Gab1 DPLC-g mutant died Here we show that tyrosine 407 is a major PLC-g 1 in collagen (Figure 5) or in soft agar (data not shown), binding site. Tyrosine 373 also binds PLC-g 1 but to a the activation of Gab1-associated PLC-g 1 appears to lesser extent and tyrosine 307 very weakly. We can not be required for serum-mediated anchorage-independent rule out the possibility that these phosphotyrosines also growth. Previous studies have shown that the product bind other SH2 or PTB containing proteins. The ability of PLC-g (IP3) mediates the release of calcium from of a single phosphorylated tyrosine of a docking internal stores and the correct balance of calcium protein to recruit multiple transducers has already between the endoplasmic reticulum and the mitochon- been described (Sun et al., 1993, 1996; Rocchi et al., dria is necessary for cell survival (Berridge et al., 1998; 1995). However, we show that the Gab1 triple mutant Porter et al., 1998; Dolmetsch et al., 1998). The loss of does not recruit in vivo PLC-g 1 in response to HGF. this equilibrium, for example by a decrease in IP3 Although the HGF receptor does not phosphorylate production, could be lethal for the cell grown in speci®c Gab1 docking sites compared to the EGF collagen. Moreover, the addition of HGF can prevent receptor, we observed a strong di€erence in time course apoptosis induced by loss of anchorage through PI3 between the two growth factors: sustained Gab1 kinase-dependent activation of AKT, which has been phosphorylation in response to HGF in contrast to a described to play an important role for cell survival transient phosphorylation in response to EGF. This (Franke et al., 1997; Kau€mann-Zeh et al., 1997; di€erence could be due to a number of mechanisms. Dudek et al., 1997). In agreement with this hypothesis, These could involve di€erences in the kinetics of kinase a previous study has shown that expression of a Gab1 activation (Naldini et al., 1991; Aharonov et al., 1978; mutant unable to bind PI3 kinase enhanced apoptosis McIntyre et al., 1995), in the half-life of the ligand- and decreased the protective e€ect of NGF (Holgado- receptor complex at the surface (Naka et al., 1993; Madruga et al., 1997). Mizuno et al., 1993; Kozu et al., 1997; Helin and In conclusion, the branching tubulogenesis response Beguinot, 1991) or di€erential activation of phospho- requires the activation of di€erent pathways such as tyrosine phosphatases such as SHPTP-2 (Holgado- the PI3 kinase, Ras (Khwaja et al., 1998; Fournier et Madruga et al., 1996; Bardelli et al., 1997a; Lehr et al., al., 1996) STAT- (Boccaccio et al., 1998) and, as 1999). Transient versus sustained activation of speci®c demonstrated in this study, PLC-g. We would like to signaling pathways has already been shown to play an suggest that a delicate balance of activation of these important role in inducing biological responses. For signaling pathways is required for tubulogenesis and example, sustained activation of the MAP kinase the alteration of this equilibrium could abrogate this pathway was found to be critical for neuronal cell speci®c response. HGF and EGF can trigger the same di€erentiation in the response to NGF versus EGF number of signaling pathways, but only HGF ± (Marshall, 1995; Traverse et al., 1992). stimulating the sustained phosphorylation of Gab1 ± In this paper, we show that recruitment of PLC-g to could provide the optimal balance for the required Gab1 is required for branching tubulogenesis re- activation of multiple signaling pathways.

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1516 Materials and methods and separated by SDS ± PAGE using an 8% resolving gel. The gel pieces corresponding to the 32P-labeled Gab1, pre- Materials and antibodies localized by autoradiography, were excised and incubated in 50 m NH HCO (pH 8) for 2 h at 378C. CaCl at a ®nal All reagents used were from Fluka (FlukaChemie, AG, USA) M 4 3 2 concentration of 1 m and trypsin at a ®nal concentration of and Sigma (Sigma Chemicals Co.). Reagents for SDS ± M 50 mg/ml were added for 12 h at 378C. For each sample, the PAGE were from Bio-Rad (Biorad Laboratories). [32P] eluted phosphopeptides were lyophilized, washed with H O, orthophosphate was purchased from Amersham Italia. 2 and dissolved in NH3 (N/1000). Phosphopeptides were Recombinant HGF was obtained from baculovirus-infected separated by two-dimensional analysis on silica thin-layer SF-9 cells cultured in medium supplemented with 2% FCS. plates, as described (Gual et al., 1998). The plates were dried Epidermal growth factor was purchased from Sigma (St and subjected to autoradiography. Louis, MO, USA). The PLCs inhibitor (U-73122) was purchased from Biomol. The monoclonal anti-Met antibodies DQ 13 were puri®ed from ascite ¯uids of hybridoma from Receptor kinase assay mice, and kindly provided by M Prat and R Albano. Anti- phosphotyrosine, anti-Nck, anti-Gab-1, anti-PLC-g 1 and Antibodies anti-HGF or -EGF receptors (4 mg/sample) were anti-EGF receptor antibodies were purchased from UBI incubated with protein G-sepharose for 1 h at 48C. The (Lake Placid, NY, USA). Anti-phospho-AKT and anti-AKT pellets were washed twice with 50 mM HEPES, 150 mM antibodies were from Biolabs (New England) and the Anti- NaCl, pH 7.6. Lysates from transfected cells were incubated MYC antibodies were from Calibiochem (New England). with the corresponding antibodies for 2 h. The pellets were washed twice with 20 mM HEPES, 100 mM NaCl containing 1% (vol/vol) of Triton X-100. The peptides (250 mM) were Plasmid construction mixed with pellets containing the immunopuri®ed HGF or The mouse Gab1 cDNA subcloned into pBluescript-SK was EGF receptors. The phosphorylation reaction was initiated a gift from Dr W Birchmeier. The mouse Gab1 cDNA was by addition of 20 mM HEPES, 100 mM NaCl, 5 mM MnCl2, 32 subcloned into pcDNA3 (Invitrogen, CA, USA). All point 5mM MgCl2,15mM [g- P] (2.5 Ci/mmol) ATP for HGF mutations of Gab1 were generated by site-directed mutagen- receptor and 20 mM HEPES, 100 mM NaCl, 10 mM MnCl2, 32 esis using the Stratagene QuikChange Kit (La Jolla, CA, 5mM MgCl2,15mM [g- P] (2.5 Ci/mmol) ATP for EGF USA) and the presence of point mutations was veri®ed by receptor. The phosphorylation reactions were stopped after DNA sequence analysis. To construct the expression vector, a 45 min by spotting a constant volume of each sample onto Myc epitope sequence was fused to the 5' end of wild type phosphocellulose papers, which were dipped into 1% (vol/ and mutated Gab1 cDNA in pBluescript KS, and the Not-1 vol) orthophosphoric acid. fragments containing epitope-tagged Gab1 were subcloned into the pCEV 29.1 expression vector. The wild type Tpr-Met Phosphorylation of Gab1 and AKT in intact cells cDNA was subcloned into the pMT2 vector. Con¯uent A549 human lung carcinoma cells growing in 145- mm culture dishes were starved in DMEM supplemented Cell culture with 10 mM glutamine for 15 h before addition of EGF 293 cells, COS-7 cells, A549 lung carcinoma cells and Madin- (100 ng/ml) or HGF (100 ng/ml) for di€erent times. Cells Darby Canine Kidney (MDCK) cells were purchased from were subsequently washed with ice-cold bu€er containing ATCC (American Type Culture Collection). Cells were 50 mM HEPES, 150 mM NaCl, 10 mM EDTA, 100 mM NaF cultured in Dulbecco's modi®ed Eagle's medium supplemen- and 2 mM sodium orthovanadate (pH 7.4) before solubiliza- ted with 5% fetal calf serum (Sigma Chemical Co., St Louis, tion for 30 min at 48C in lysis bu€er. One aliquot of each clari®ed lysate (700 mg of total protein) obtained after USA) and maintained at 378C in a 5% CO2-humidi®ed atmosphere. MDCK stable cell lines expressing recombinant centrifugation (15 min at 15 000 g at 48C) was incubated forms of Gab1 were obtained by transfection of each for 3 h at 48C with antibodies preadsorbed on protein-A pCEV29.1 expression plasmid using the calcium phosphate Sepharose (4 mg anti-Gab1 antibody/sample). Pellets washed precipitation technique (CellPhect, Pharmacia, Uppsala, with lysis bu€er and a second aliquot (100 mg of total Sweden). The selection was performed in growth medium protein) of each total cell lysate were resuspended in Laemmli supplemented with 0.8 mg/ml of G-418 sulfate (geneticin; sample bu€er and separated by SDS ± PAGE using an 8% GIBCO BRL Life Technologies, Inc., Gaithersburg, USA). resolving gel, before transfer to a nitrocellulose membrane. After two weeks, the expression of the respective recombinant The membrane was blocked with saline bu€er (10 mM Tris, proteins was tested by Western blot analysis. 140 mM NaCl, pH 7.4) containing 5% (wt/vol) BSA for 2 h at 228C and probed with the appropriate antibodies (Mouse anti-phosphotyrosine antibodies and rabbit anti-phospho Phosphopeptide map analysis of Gab1 AKT antibodies at 1 mg/ml). Speci®c binding was detected After the serum-depletion period, COS-7 cells overexpressing by an Enhanced Chemiluminescence System (ECL, Amer- wild type Gab1 were washed twice in culture medium without sham). In some cases, the membrane was stripped for 30 min phosphate and incubated for 3 h in this medium enriched at 508Cin62mM Tris, 100 mM 2-mercaptoethanol, and 2% with [32P]-orthophosphate (0.8 mCi/ml). EGF (100 ng/ml) or (wt/vol) SDS, and reprobed with the indicated antibodies. HGF (100 units/ml) were added at the end of the labeling period. After 10 min, cells were washed with ice-cold bu€er Production of fusion proteins containing 50 mM HEPES, 150 mM NaCl, 10 mM EDTA, 100 mM NaF and 2 mM sodium orthovanadate (pH 7.4) The pGEX-PLC-g 1 (NH2- or COOH- SH2 domains) was before solubilization for 30 min at 48C in lysis bu€er (50 mM expressed in Escherichia Coli. Expression of recombinant HEPES, 150 mM NaCl, 10 mM EDTA, 100 mM NaF, 2 mM proteins was induced with 0.1 mM isopropyl-b-D-thiogalacto- sodium orthovanadate, 0.5 mM phenylmethylsulfonyl ¯uor- pyranoside for 3 h. Bacteria were lysed with 20 mM Tris, ide, 100 U/ml aprotinin, 20 mM leupeptin and 1% (vol/vol) 1mM NaCl, 0.2 mM EDTA, 0.2 nM EGTA, 1 mg/ml Triton X-100, pH 7.4). The clari®ed lysates obtained after lysozyme, 0.5 mM phenyl-methylsulfonyl ¯uoride, 100 U/ml centrifugation (15 min at 15 000 g at 48C) were incubated for aprotinin, 20 mM leupeptin (pH 7.4) for 30 min in ice, 3 h at 48C with antibodies preadsorbed on protein-A sonicated three times, and frozen in liquid nitrogen. Lysates Sepharose (4 mg anti-Gab1 antibody/sample). Pellets washed were centrifuged (30 000 g for 30 min at 48C), and the with lysis bu€er were resuspended in Laemmli sample bu€er supernatants were incubated with glutathione-Sepharose

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1517 (Pharmacia) for 1 h at 48C. The fusion proteins were eluted glutheraldeyde and stained with crystal violet. The stained with 50 mM glutathione and 100 mM HEPES (pH 8). cells were solubilized in 10% acetic acid and absorbance at 595 nm was counted in a microplate reader (Perkin-Elmer). Pull down of Gab1 wild type and mutated forms by GST fusion proteins Scattering and branching tubulogenesis assays COS-7 cells co-overexpressing wild type or mutated Gab1 MDCK transfectants (26103 cells per well) were plated in with Tpr-Met were lysed with Lysis bu€er containing DMEM containing 10% serum in 96 well plates. The medium protease inhibitors and the phosphatase inhibitor sodium was changed after 5 h and HGF (100 units/ml) and the orthovanadate. These lysates were incubated either with GST inhibitor of the PLC (U-73122 used at 2 mM in 0.1% of (5 mg) or with a mix of NH2- and COOH- SH2 domains of DMSO) were added as indicated. After 12 h, the cells were PLC-g (2.5 mg of each) obtained as GST fusion proteins ®xed with glutheraldeyde and stained with crystal violet. The immobilized on glutathione-Sepharose (Pharmacia). After 2 h morphogenic activity of the MDCK cultures was analysed as incubation at 48C, Sepharose beads were washed with lysis described previously (Montesano et al., 1991). Brie¯y, the bu€er and bound proteins were separated on SDS ± PAGE cells were harvested from cultures using trypsin-EDTA and using an 8% resolving gel and transferred to nitrocellulose suspended at a concentration of 105 cells per ml in collagen ®lters (Hybond, Amersham, UK) for probing with anti-Met solution containing type I collagen gel (3 mg/ml) (Collabora- or anti-Gab1 antibodies. tive Biomedical Products; Becton Dickinson Labware), 10X DMEM (Gibco BRL), and 0.5 M HEPES (pH 7.4). Aliquots (100 ml) of the cell suspension were dispensed in 96-well Co-immunoprecipitation assays microliter plates and allowed to gel for 15 min at 378C before After serum-depletion, 293 cells overexpressing Myc-tagged adding 200 ml of DMEM with 20% FCS. After 12 h, HGF wild type or MYC-tagged mutated Gab1 were stimulated (100 units/ml), EGF (100 ng/ml) and U-73122 (2 mM in 0.1% with either HGF (100 units/ml) or FCS (0.4%) for 10 min. of DMSO) were added as indicated. The medium was Concerning the cells overexpressing wild type Gab1 and Tpr- changed every two days. After 7 ± 10 days the branching Met, the serum starvation step was performed for 48 h. In all tubulogenesis responses were evaluated. cases, the cells were subsequently washed with ice-cold bu€er containing 50 mM HEPES, 150 mM NaCl, 10 mM EDTA, 100 mM NaF, and 2 mM sodium orthovanadate (pH 7.4) before solubilization for 30 min at 48C in lysis bu€er. One Abbreviations aliquot of each clari®ed lysate (700 mg of total protein) was DAG, diacylglycerol; EGF, epidermal growth factor; incubated for 3 h at 48C with the appropriate antibodies Gab1, GRB2 associated binder-1; GRB-2, growth factor preadsorbed on protein-G Sepharose (4 mg antibody/sample). receptor-bound protein 2; GST, glutathione S-transferase; Pellets washed with lysis bu€er and a second aliquot (100 mg HGF, hepatocyte growth factor; IP3, inosotol-1,4,5 tripho- of total protein) of each total cell lysate were resuspended in sphare; NGF, nervous growth factor; p85, the 85 kDa Laemmli sample bu€er and separated by SDS ± PAGE using regulatory subunit of PI 3 kinase; PDGF, platelet-derived an 8% resolving gel, before transfer to a nitrocellulose growth factor; PI 3 kinase, phosphoinositide 3 kinase; membrane. The membrane was blocked with saline bu€er PLC, phospholipase C; SH2, Src homology 2; SHC, Src containing 5% (wt/vol) BSA for 2 h at 228C and probed with homology/collagen; SHPTP, SH2 containing Phosphotyr- the appropriate antibody (Mouse anti-phosphotyrosine anti- osine phosphatase; STAT, signal transducer and activator bodies, rabbit anti-Gab1 antibodies, mouse anti-Myc anti- of the transcription. bodies and mouse anti-PLC g 1 antibodies at 1 mg/ml; and rabbit anti-Nck antibodies at 0.1 mg/ml). Speci®c binding was detected by an Enhanced Chemiluminescence System (ECL, Amersham). In some cases, the membrane was stripped for Acknowledgments 30 min at 508Cin62mM Tris, 100 mM 2-mercaptoethanol, We are grateful to Dr W Birchmeier for mouse Gab1 and 2% (wt/vol) SDS, and reprobed with the indicated cDNA,DrPPDiFioreforGST-PLCg fusion proteins antibodies. and Dr T Miki for pCEV29.1 plasmid. The excellent technical assistance of L Palmas, R Albano and G Petruccelli is gratefully acknowledged. We thank Anto- Cell growth assay nella Cignetto for secretarial help and Elaine Wright for MDCK transfectants (26103 cells per well) were plated in editing the manuscript. This work was supported by 5% serum in 96 well plates. The cell medium was changed BIOMED EC grant no. BMH4-98-3852 to PM Comoglio. after 12 h and HGF (100 units/well) was added as indicated. P Gual and TA Williams are supported by Marie Curie After 12 h, 24 h, 48 h, 72 h the cells were ®xed with TMR fellowships.

References

Aharonov A, Passovoy DS and Herschman HR. (1978). J. Dolmetsch RE, Xu K and Lewis RS. (1998). Nature, 392, Supramol. Struct., 9, 41 ± 45. 933 ± 936. Anderson D, Koch CA, Grey L, Ellis C, Moran MF and Dudek H, Datta SR, Franke TF, Birnbaum MJ, Yao R, Pawson T. (1990). Science, 250, 979 ± 982. Cooper GM, Segal RA, Kaplan DR and Greenberg ME. Bardelli A, Longati P, Gramaglia D, Stella MC and (1997). Science, 275, 661 ± 665. Comoglio PM. (1997a). Oncogene, 18, 3103 ± 3111. Fournier TM, Kamikura D, Teng K and Park M. (1996). J. Berridge MJ, Bootman MD and Lipp P. (1998). Nature, 395, Biol. Chem., 271, 22211 ± 22217. 645 ± 648. Franke TF, Kaplan DR and Cantley LC. (1997). Cell, 88, Boccaccio C, AndoÁ M, Tamagnone L, Bardelli A, Michieli P, 435 ± 437. Battistini C and Comoglio PM. (1998). Nature, 391, 285 ± Graziani A, Gramaglia D, Cantley LC and Comoglio PM. 288. (1991). J. Biol. Chem., 266, 22087 ± 22090. Carpino N, Wisniewski D, Strife A, Marshak D, Kobayashi Gu H, Pratt JC, Burako€ SJ and Neel BG. (1998). Mol. Cell, R, Stillman B and Clarkson B. (1997). Cell, 88, 197 ± 204. 2(6), 729 ± 740.

Oncogene PLC-g requirement for HGF-mediated tubulogenesis P Gual et al 1518 Gual P, Baron V, Lequoy V and Van Obberghen E. (1998). Nishida K, Yoshida Y, Itoh M, Fukada T, Ohtani T, Endocrinology, 139, 884 ± 893. Shirogane T, Atsumi T, Takahashi-Tezuka M, Ishihara K, Hartmann G, Weidner KM, Schwarz H and Birchmeier W. Hibi M and Hirano T. (1999). Blood, 93, 1809 ± 1816. (1994). J. Biol. Chem., 269, 21936 ± 21939. Nishizuka Y. (1992). Science, 258, 607 ± 614. Helin K and Beguinot L. (1991). J. Biol. Chem., 266, 8363 ± Noh DY, Shin SH and Rhee SG. (1995). Biochim. Biophys. 8368. Acta, 1242, 99 ± 113. Herbst R, Carroll PM, Allard JD, Schilling J, Raabe T and Pelicci G, Giordano S, Zhen Z, Salcini AE, Lanfrancone L, Simon MA. (1996). Cell, 85, 899 ± 909. Bardelli A, Panayotou G, Water®eld MD, Ponzetto C, Holgado-Madruga EM, Emlet DR, Moscatello DK, Godwin Pelicci PG and Comoglio PM. (1995). Oncogene, 10, AK and Wong AJ. (1996). Nature, 379, 560 ± 564. 1631 ± 1638. Holgado-Madruga M, Moscatello DK, Emlet DR, Dieterich Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P, R and Wong AJ. (1997). Proc. Natl. Acad. Sci. USA, 94, Giordano S, Graziani A, Panayotou G and Comoglio PM. 12419 ± 12424. (1994). Cell, 77, 261 ± 271. Jones N and Dumont DJ. (1998). Oncogene, 17, 1097 ± 1108. Ponzetto C, Zhen Z, Audero E, Maina F, Bardelli A, Basile Kau€mann-Zeh A, Rodriguez-Viciana P, Ulrich E, Gilbert ML, Giordano S, Narsimhan R and Comoglio PM. C, Co€er P, Downward J and Evan G. (1997). Nature, 385, (1996). J. Biol. Chem., 271, 14119 ± 14123. 544 ± 548. Porter VA, Bonev AD, Knot HJ, Heppner T, Stevenson AS, Khwaja A, Lehmann K, Marte BM and Downward JD. KleppischT,LedererWJandNelsonMT.(1998).Am. J. (1998). J. Biol. Chem., 273, 18793 ± 18801. Physiol., 274, C1346 ± C1355. Kim HK, Kim JW, Zilberstein A, Margolis B, Kim JG, Raabe T, Riesgo-Escovar J, Liu X, Bausenwein BS, Deak P, Schlessinger J and Rhee SG. (1991). Cell, 65, 435 ± 441. Maroy P and Hafen E. (1996). Cell, 85, 911 ± 920. Kochhar KS and Iyer AP. (1996). Cancer Lett., 104, 163 ± Ridley AJ, Comoglio PM and Hall A. (1995). Mol. Cell. 169. Biol., 15, 1110 ± 1122. Kouhara H, Hadari YR, Spivak-Kroizman T, Schilling J, Rocchi S, Tartare-Deckert S, Mothe I and Van Obberghen Bar-Sagi D, Lax I and Schlessinger J. (1997). Cell, 89, E. (1995). Endocrinology, 136, 5291 ± 5297. 693 ± 702. Rocchi S, Tartare-Deckert S, Murdaca J, Holgado-Madruga Kozu A, Kato Y, Shitara Y, Hanano M and Sugiyama Y. M, Wong AJ and Van Obberghen E. (1998). Mol. (1997). Pharm. Res., 14, 1228 ± 1235. Endocrinol., 12, 914 ± 923. Lecoq-Lafon C, Verdier F, Fichelson S, Chretien S, Royal I, Fournier TM and Park M. (1997). J. Cell. Physiol., Gisselbrecht S, Lacombe C and Mayeux P. (1999). Blood, 173, 196 ± 201. 93, 2578 ± 2585. Royal I and Park M. (1995). J. Biol. Chem., 270, 27780 ± Lehr S, Kotzka J, Herkner A, Klein E, Sietho€ C, Knebel B, 27787. Noelle V, Bruning JC, Klein HW, Meyer HE, Krone W Stoker M. (1989). J. Cell. Physiol., 139, 565 ± 569. and Muller-Wieland D. (1999). Biochemistry, 38, 151 ± Sun XJ, Crimmins DL, Myers MGJ, Miralpeix M and White 159. MF. (1993). Mol. Cell. Biol., 13, 7418 ± 7428. Lipp M and Niggl E. (1996). Prog. Biophys. Mol. Biol., 65, Sun XJ, Pons S, Asano T, Myers MGJ, Glasheen E and 265 ± 296. White MF. (1996). J. Biol. Chem., 271, 10583 ± 10587. Margolis B, Li N, Koch A, Mohammadi M, Hurwitz DR, Takahashi-Tezuka M, Yoshida Y, Fukada T, Ohtani T, Zilberstein A, Ullrich A, Pawson T and Schlessinger J. Yamanaka Y, Nishida K, Nakajima K, Hibi M and (1990). EMBO J., 9, 4375 ± 4380. Hirano T. (1998). Mol. Cell. Biol., 18, 4109 ± 4117. Maroun CR, Holgado-Madruga M, Royal I, Naujokas MA, Takaishi K, Sasaki T, Kato M, Yamochi W, Kuroda S, Fournier TM, Wong AJ and Park M. (1999). Mol. Cell. Nakamura T, Takeichi M and Takai Y. (1994). Oncogene, Biol., 19, 1784 ± 1799. 9, 273 ± 279. Marshall CJ. (1995). Cell, 80, 179 ± 185. Thackeray JR, Gaines PC, Ebert P and Carlson JR. (1998). McIntyre BS, Birkenfeld HP and Sylvester PW. (1995). Cell Development, 125, 5033 ± 5042. Prolif., 28, 45 ± 56. Traverse S, Gomez N, Paterson H, Marshall C and Cohen P. Ming GL, Song HJ, Berninger B, Inagaki N, Tessier-Lavigne (1992). Biochem. J., 288, 351 ± 355. M and Poo MM. (1999). Neuron, 23, 139 ± 148. Wahl M and Carpenter G. (1991). Bioessays, 13, 107 ± 113. Mizuno K, Higuchi O, Tajima H, Yonemasu T and Wang Z, Gluck S, Zhang L and Moran MF. (1998). Mol. Nakamura T. (1993). J. Biochem. (Tokyo), 114, 96 ± 102. Cell. Biol., 18, 590 ± 597. Montesano R, Matsumoto K, Nakamura T and Orci L. Weidner KM, DiCesare S, Sachs M, Brinkmann V, Behrens J (1991). Cell, 67, 901 ± 908. and Birchmeier W. (1996). Nature, 384, 173 ± 176. Naka D, Shimomura T, Yoshiyama Y, Sato M, Ishii T and Yamanashi Y and Baltimore D. (1997). Cell, 88, 205 ± 211. Hara H. (1993). FEBS Lett., 329, 147 ± 152. Yenush L and White MF. (1997). Bioessays, 19, 491 ± 500. Naldini L, Vigna E, Narsimhan RP, Gaudino G, Zarnegar R, Michalopoulos GK and Comoglio PM. (1991). Oncogene, 6, 501 ± 504. Nguyen L, Holgado-Madruga M, Maroun C, Fixman ED, Kamikura D, Fournier T, Charest A, Tremblay ML, Wong AJ and Park M. (1997). J. Biol. Chem., 272, 20811 ± 20819.

Oncogene