Oncogene (2006) 25, 2433–2443 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE GRB2-mediated recruitment of GAB2, but not GAB1, to SF-STK supports the expansion of Friend virus-infected erythroid progenitor cells

HE Teal1,5,SNi1,5,JXu1, LD Finkelstein2, AM Cheng3, RF Paulson1, G-S Feng4 and PH Correll1,5

1Department of Veterinary and Biomedical Science, Center for Molecular Immunology and Infectious Disease, The Pennsylvania State University, University Park, PA, USA; 2NHGRI, NIH, Bethesda, MD, USA; 3Washington University, St Louis, MO, USA and 4The Burnham Institute, San Diego, CA, USA

Friend virusinducesthe development of erythroleukemia model with which to study the signaling events that in mice through the interaction of a viral glycoprotein, direct development through progressive stages gp55, with a truncated form of the Stk receptor tyrosine of transformation in vivo. FV is a complex of two kinase, short form-Stk (Sf-Stk), and the EpoR. We have viruses, the replication-defective spleen focus-forming shown previously that the ability of Sf-Stk to participate virus (SFFV) which expresses gp55, the pathogenic in the transformation of Friend virus-infected cells re- component, and the replication-competent helper virus, quires the kinase activity and Grb2-binding site of Sf-Stk. Friend murine leukemia virus (F-MuLV) (Ben-David Here we show that Grb2 heterozygous mice exhibit de- and Bernstein, 1991). In the early stages of FV infection, creased susceptibility to Friend erythroleukemia and that erythroid progenitor cells are infected resulting in the expansion of erythroid progenitors in response to infection polyclonal expansion of the infected cells (Mirand et al., requiresthe C-terminal SH3 domain of Grb2. A fusion 1968; Tambourin and Wendling, 1971; Horoszewicz in which the Grb2-binding site in Sf-Stk is et al., 1975; Liaoand Axelrad, 1975). The late stage of replaced by Gab2, supports the growth of progenitors erythroleukemia in FV-infected mice is characterized by from mice lacking Sf-Stk, whereasa Sf-Stk/Gab1 fusion subsequent mutations in p53 combined with the inser- protein does not. Gab2 is expressed in spleens from Friend tional activation of the Spi-1 oncogene, resulting in virus-infected mice, co-immunoprecipitates with Sf-Stk acute erythroleukemia (Mowat et al., 1985; Moreau- and is tyrosine phosphorylated in the presence of Sf-Stk. Gachelin et al., 1988). Mice with a targeted deletion in Gab2 are less susceptible Several loci have been identified that control suscept- to Friend erythroleukemia and the expansion of erythroid ibility toFV infection(Axelrad, 1969). The Fv1 progenitor cells in response to infection can be rescued by and Fv4 affect the ability of FV to infect the early expression of Gab2, but not Gab1. Taken together, these erythroid progenitor cells. Another set of genes, W, Sl, f data indicate that a Sf-Stk/Grb2/Gab2 complex mediates and Fv2 are required for the expansion of infected the growth of primary erythroid progenitor cellsin progenitor cells and the progression of the early stages response to Friend virus. of erythroleukemia. The host Fv2 acts in a cell Oncogene (2006) 25, 2433–2443. doi:10.1038/sj.onc.1209288; autonomous manner and determines the proliferative published online 12 December 2005 response of infected erythroblasts to gp55 (Lilly, 1970; Behringer and Dewey, 1985). Previously, we demon- Keywords: Friend virus; erythroleukemia; Fv2; Sf-Stk strated that Fv2 encodes the Met-related macrophage- stimulating 1-receptor (Mst1r) tyrosine kinase, also known as stem cell-derived tyrosine kinase receptor (Stk) (Persons et al., 1999). Mice homozygous for the resistant allele of Fv2 (Fv2rr), including C57BL/6 mice, Introduction fail toexpress a naturally occurring, N-terminally truncated form of Stk, called short form-Stk (Sf-Stk). The development of leukemia proceeds in response to a An internal promoter within the Stk locus drives Sf-Stk series of events resulting in transformation, however, the expression which lacks the N-terminal ligand-binding signals that govern the progression of disease through domain of the full-length Stk, but retains the transmem- these discrete phases are largely unknown. Friend brane and tyrosine kinase domains. Enforced expression erythroleukemia virus (FV) provides an experimental of Sf-Stk in C57BL/6 mice is sufficient to confer FV susceptibility in Fv2rr mice (Persons et al., 1999). Previous work has shown that gp55 interacts with the Correspondence: Dr PH Correll, 115 Henning Building, Department EpoR, and that coexpression of EpoR and gp55 in IL-3- of Veterinary Science, University Park, PA 16802-3500, USA. dependent BaF/3 cells results in factor-independent cell E-mail: [email protected] growth (Li et al., 1990; D’Andrea, 1992). Sf-Stk has also 5These authors contriuted equally to this work. Received 9 June 2005; revised 27 October 2005; accepted 31 October been shown to co-immunoprecipitate with gp55 in 2005; published online 12 December 2005 hematopoietic cells resulting in constitutive activation Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2434 of Sf-Stk (Nishigaki et al., 2001), and studies from our 4 laboratory clearly demonstrate that Sf-Stk kinase 3.5 activity is essential for Epo-independent growth of 3 infected cells (Finkelstein et al., 2002). Interestingly, *** the chicken homologue of Stk, c-sea, is the cellular 2.5 homologue of the avian retroviral oncoprotein v-sea, 2 which causes erythroblastosis and anemia in chickens 1.5 (Hayman et al., 1985). v-sea is a 155 kDa glycoprotein, 1 which is cleaved into85 and 70 kDa subunits and linked Spleen weight (g) by a disulfide bridge (Hayman et al., 1985). The 85 kDa 0.5 env subunit of v-sea mediates interaction between SEA 0 molecules resulting in autophosphorylation and consti- Grb2 +/+ Grb2 +/- Grb2 +/+ Grb2 +/- tutive activation of the tyrosine kinase domain (Smith uninfected infected et al., 1989). Thus, the ability of viral envelope Figure 1 Grb2 þ /À mice are less susceptible toFriend virus in to induce erythroleukemia through the constitutive vivo. Grb2 þ /À mice (8 weeks old) on a sensitive BALB/c activation of this family of RTKs is conserved across background and wild-type BALB/c control mice were infected species. with FV. Spleens were weighed 2 weeks postinfection. **Po0.01. In vitro infection of primary bone marrow cells with FV results in Epo-independent erythroid colony forma- tion. In previous analysis of Sf-Stk signaling during FV 2003). In order to determine the susceptibility of infection in vitro, we demonstrated that tyrosine 1337, Grb2 þ /À mice on a sensitive BALB/c background to which binds Grb2, is required for the formation of Epo- FV in vivo, 8-week-old Grb2 þ /À and wild-type control independent erythroid colonies by FV (Finkelstein et al., mice were infected with FV and spleens were harvested 2 2002). In addition, mutation of the asparagine following weeks later. The average spleen size of infected Grb2 þ /À this tyrosine (YVNV to YVHV), which eliminates Grb2 animals was significantly reduced when compared to binding without affecting the binding of other Src wild-type controls, suggesting that Grb2 may be 2 (SH2) domain-containing proteins, abro- required for the expansion of FV-infected cells in the gates FV-induced cell transformation in vitro, which spleen (Figure 1). In vitro FV infection of primary bone further underscores the key role of Grb2-dependent marrow cells from sensitive mice has previously been signals (Fournier et al., 1996; Finkelstein et al., 2002). shown to result in Epo-independent erythroid colony Here we show that Grb2 haploid insufficiency decrea- formation (Finkelstein et al., 2002). In order to ses the susceptibility of mice to Friend erythroleukemia determine whether Grb2 is required for this response, in vivo. Furthermore, we demonstrate that primary we infected bone marrow cells from Grb2 þ /À and wild- erythroblasts from Grb2 þ /À mice exhibit reduced type control mice with FV and assessed the ability of -independent colony formation in response these cells to form Epo-independent erythroid colonies toFriend virus infection in vitro, and that this defect can in response to FV, compared with the number of Epo- be rescued by retroviral introduction of wild-type Grb2, dependent colonies formed in the absence of FV but not a Grb2 mutant containing a mutation in the infection. Bone marrow from wild-type controls gave C-terminal SH3 domain of Grb2. This domain binds the rise to approximately equal numbers of Epo-dependent Gab family of adaptor proteins and we demonstrate and FV-induced Epo-independent colonies; however, here that Sf-Stk co-immunoprecipitates with Gab2 and bone marrow cells from Grb2 þ /À mice yielded an Gab2À/À mice are largely resistant toFriend virus average of about threefold fewer Epo-independent infection in vivo. Furthermore, introduction of a Sf-Stk/ colonies compared to the number of Epo-dependent Gab2 fusion protein into bone marrow lacking Sf-Stk colonies (Figure 2d). supports Epo-independent colony formation. These data demonstrate that a Sf-Stk/Grb2/Gab2 complex The C-terminal SH3 subdomain of Grb2 is critical for mediates the efficient expansion of Friend virus-infected FV-induced erythroid colony formation cells both in vitro and in vivo. Previously we demonstrated that exogenous expression of Sf-Stk in C57BL/6 bone marrow cells, which fail to express endogenous Sf-Stk, could rescue the ability of these cells to form Epo-independent colonies following Results FV infection in vitro (Finkelstein et al., 2002). However, a mutant form of the receptor which lacked the ability to Grb2 þ /À mice exhibit reduced susceptibility to FV bind Grb2 could not rescue this response. Similarly, to Mice with a targeted deletion in the gene-encoding Grb2 determine whether the defect in the Grb2 þ /À mice can are embryonic lethal (Cheng et al., 1998), however, be rescued by wild-type Grb2, we transduced bone haplo-insufficiency results in several defects in the adult marrow cells from Grb2 þ /À mice with an MSCV- animal including impaired negative selection of T cells, a based retroviral vector expressing wild-type and mutant block in cardiac hypertrophy and fibrosis in response to forms of Grb2 (Figure 2a) fused to GFP. Equivalent pressure overload and resistance to neointima formation protein expression in the packaging cells was confirmed following carotid injury (Gong et al., 2001; Zhang et al., by Western blot analysis with anti-GFP (Figure 2b), and

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2435 Grb2-GFP a bc32 W36K-GFP W193K-GFP W36K W193K GFP Grb2 SH3 SH2 SH3 Grb2 Vector Event W36K SH2 SH3 W36K

SH3 SH2 W193K W193K 100 101 102 103 EGFP de25

Epo Epo 80 20 FV FV

60 15

40 10 ** ** 20 * Number of BFU-E 5 Number of BFU-E 0 0 Grb2 +/+ Grb2 +/- GFP Grb2-GFP W36K-GFP W193K-GFP Figure 2 The C-terminal SH3 subdomain of Grb2 is required for FV-induced colony formation. (a) Schematic of wild-type and mutant Grb2 proteins used in this study. (b) Expression of wild-type and mutant Grb2 in 293T cells. (c) Total bone marrow from Grb2 þ /À and wild-type control mice were infected with the indicated retroviral vectors, and infection efficiency was determined by flow cytometry for GFP. (d) Wild-type and Grb2 þ /À bone marrow was plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence of either FV or Epo (1 U/ml). (e) Total bone marrow from Grb2 þ /À mice was transduced with vector or MSCV-neo-EGFP- Grb2 with or without SH3 domain mutations and plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence of FV or Epo (1 U/ml). BFU-E were stained with acid-benzidine and scored on day 5. Data are representative of three independent experiments and s.e. bars denote the mean7s.d. of three replicates.**Po0.01. equivalent infection efficiency was confirmed by flow capable of rescuing the ability of these cells to form analysis for GFP of infected bone marrow cells BFU-E in response to FV infection. Conversely, the (Figure 2c). While control infections resulted in a expression of the Grb2 C-terminal SH3 subdomain reduced percentage of FV-induced Epo-independent mutant, W193K, in Grb2 þ /À bone marrow failed to colonies as expected, bone marrow from Grb2 þ /À support Epo-independent erythroid colony formation mice expressing wild-type Grb2 supported Epo-inde- in response to FV infection (Figure 2d). These pendent BFU-E colony formation in response to FV results suggest that the C-terminal SH3 domain of infection at levels comparable to the number of Epo- Grb2 is essential for the ability of FV to drive the dependent colonies (Figure 2e). These data support our Epo-independent growth of these cells. previous conclusion that Grb2 binding to Sf-Stk is required for FV-mediated erythroid colony formation. Grb2 is an adapter protein composed of a central SH2 Gab2 is a downstream target of Sf-Stk signaling domain and two flanking SH3 domains. The SH2 Previous work has demonstrated that the adapter domain is responsible for recruitment of Grb2 to proteins, Gab and SLP76, interact with the C-terminal tyrosine-phosphorylated receptors. The N-terminal SH3 subdomain of Grb2 (Lewitzky et al., 2001). The SH3 domain recognizes a P-x-x-P-x-R-binding motif, Gab family of adapter proteins consists of three which is found in a number of signaling proteins, includ- members in mammalian cells, Gab1, Gab2 and Gab3. ing SOS and c-Cbl, while the consensus-binding motif All three Gab proteins are composed of an N-terminal for the C-terminal SH3 domain is P-x-x-x-R-x-x-K-P PH domain, two proline-rich regions and numerous (Kohda et al., 1994; Lock et al., 2000; Lewitzky et al., tyrosines that, upon phosphorylation, serve as docking 2001). In order to map the subdomains of Grb2 that are sites for signaling proteins including p85 and SHP-2 required for FV-induced Epo-independent growth of (Hibi and Hirano, 2000; Liu and Rohrschneider, 2002). erythroid progenitor cells, we mutated a conserved These adapter proteins are thought to play a role in the tryptophan in either the N- or C-terminal SH3 domain amplification of signals initiated by RTKs. To determine to lysine, which abrogates the ability of the SH3 whether the Gab family of proteins could be a target domains to bind ligand. Retroviral-mediated expression of Grb2 in FV-infected erythroblasts, we first examined of the Grb2 N-terminal SH3 subdomain mutant, W36K, the expression of Gab1, Gab2 and Gab3 in FV- in bone marrow from Grb2 þ /À animals was fully infected spleens of BALB/c mice by reverse transcriptase

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2436 (RT)–PCR. All three Gab genes were found to be The interaction of Gab2 and Myc/Sf-Stk in primary expressed in FV-infected spleens (data not shown). infected cells was confirmed using spleens from FV- These data suggest that Gab family members are infected animals (Figure 3d). expressed in the cells that repopulate the spleen in the course of FV infection and could therefore be targets for the C-terminal SH3 domain of Grb2 in FV-infected cells. Recruitment of Gab2, but not Gab1, by Sf-Stk is sufficient Both Gab1 and Gab2 have been demonstrated to to support the growth of primary erythroblasts in response signal downstream of the MET family of receptor to Friend virus infection tyrosine kinases. While Gab1 binds directly to the MET In order to determine whether the recruitment of Gab and RON receptors through the Met-binding domain proteins by Sf-Stk is sufficient to support Friend virus- (MBD), Gab2, which lacks a MBD, is indirectly induced growth, we generated Sf-Stk/Gab fusions in recruited toRTKs throughits interactionwith Grb2. which the C-terminal docking site tyrosines Y429 and Overexpression of the MBD from Gab1, which blocks Y436 of Sf-Stk were replaced with the coding sequence recruitment of Gab1 to MET and RON, in primary of Gab1 or Gab2 (Figure 4a). We have shown that Y436 erythroid progenitor cells failed to inhibit colony is essential for the growth of primary erythroblasts in formation in response to Friend virus infection (data response to FV infection in vitro. Therefore, the absence not shown). Conversely, Gab2 has been shown to play a of this docking site tyrosine in the context of the fusion critical role in the transformation of cells by the chicken proteins would prevent recruitment of other signaling homologue of STK, c-Sea. Therefore, we analysed the molecules to this site. We removed the PH domain of ability of Gab2 to co-immunoprecipitate with Sf-Stk. Gab1 and Gab2 in these constructs because this domain 293T cells were transfected with plasmids expressing is generally involved in recruiting these scaffolding myc-Sf-Stk and mGab2 and cell lysates were immuno- proteins to the signaling complex, and this should be precipitated with anti-myc or anti-Gab2 and probed accomplished by fusion with Sf-Stk. Both of these fusion with antiphosphotyrosine, anti-myc or anti-Gab2 proteins were expressed at similar levels and efficiently (Figure 3a). Data from these studies clearly indicate induced the activation of Akt, as determined by its that Gab2 co-immunoprecipitates with Sf-Stk in these ability to phosphorylate GSK, and phosphorylation of cells and that Gab2 becomes tyrosine phosphorylated in Erk in transiently transfected 293T cells (Figure 4b). the presence of Sf-Stk. We failed to detect a similar Retroviral-mediated expression of the Sf-Stk/Gab2 interaction when Sf-Stk was co-transfected with Gab1 in fusion in cells from C57Bl/6 mice lacking endogenous this system (Figure 3b), and did not observe phospho- Sf-Stk, rescued the ability of these cells to form Epo- rylation of Gab1 in these cells (Figure 3b and c). independent colonies in response to Friend virus

ab123 4 123 4 IP Blot IP Blot Myc/Sf-Stk Gab2 α-Myc α-p-Y α-Myc α-p-Y Myc/Sf-Stk Gab1 α-Gab1 α-p-Y Gab2 α-Gab2 α-p-Y Myc/Sf-Stk α α Myc/Sf-Stk α-Gab2 α-Myc -Gab1 -Myc α α Myc/Sf-Stk -Myc -Myc Myc/Sf-Stk α-Myc α-Myc α-Gab2 α-Gab2 Gab2 Gab1 α-Gab1 α-Gab1

cd

123 IP Blot Gab2 α-Gab2 α-p-Y IP Blot Sf-stk+Gab1 Gab1+MSCV Myc/sf-stk+MSCV MSCV stk Myc/Sf-Stk α-p-Y α-Myc α-Gab2 α-stk

α α Gab1 -p-Y -Gab1 Gab2 α-Gab2 α-Gab2

Figure 3 Gab2 is present in spleens from FV-infected mice and co-immunoprecipitates with Sf-Stk. (a) 293T cells were transiently transfected with plasmids expressing (1) Myc-Sf-Stk and Gab2, (2) Myc-Sf-Stk, (3) Gab2 and (4) vector control. Cell lysates were immunoprecipitated with anti-Myc or anti-Gab2 antibodies and subsequently analysed by SDS–PAGE and Western blotting with antiphosphotyrosine, anti-Gab2 or anti-Myc antibody. (b)293T cells were transiently transfected with plasmids expressing (1) Myc-Sf-Stk and Gab2, (2) Myc-Sf-Stk, (3) Gab2 and (4) vector control. Cell lysates were immunoprecipitated with anti-Myc or anti-Gab1, and subsequently analysed by SDS–PAGE and Western blotting with antiphosphotyrosine, anti-Gab1 and anti-Myc. (c) 293 cells were transiently transfected with (1) Sf-Stk and Gab1, (2) Gab1, (3) Sf-Stk and (4) vector control. Cell lysates were immunoprecipitated with antiphosphotyrosine, resolved by SDS–PAGE and probed with anti-Myc or anti-Gab1. (d) Splenocytes from day14 FVP-infected mice were immunoprecipiated with anti-Gab2 and blotted with antiphosphotyrosine, anti-Stk or anti-Gab2 as indicated. Lane 1, uninfected wild-type mice; Lane 2, FVP-infected wild-type mice; Lane 3, FVP-infected Gab2À/À mice.

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2437 a b MSCV Sf-Stk Sf-/Stk-Gab1 Sf-/Stk-Gab2 YY P-GSK Myc Shp2 Myc/SfStk p85 p85 p85 Shp2 Total Akt

PH Gab 1 or 2 P-Erk Y Y YY Y

Myc Myc/SfStk- Total Erk Gab1 or 2 Myc-tag

cd 30 Epo 60 25 FvP 50 Epo 20 * 40 FvP

15 30

10 20 *** *** Number of BFU-e 5 Number of BFU-e 10

0 0 MSCV Sf-Stk/Gab1 Sf-Stk/Gab2 MSCV Sf-Stk/Gab2 Sf-Stk/Gab2(KD) Figure 4 Retroviral expression of a Sf-Stk-Gab2 fusion protein supports FV-induced colony formation in C57BL/6 mice. (a) Schematic of WT and fusion Sf-Stk and Gab proteins (b) Total bone marrow from C57Bl/6 mice was transduced with MSCV- vector, MSCV-myc-Sf-Stk/Gab1 and MSCV-myc-Sf-Stk-Gab2. Transduced cells were plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence/absence of FV or Epo (1 U/ml). BFU-E were stained with acid-benzidine and scored on day 5. The data is representative of three independent experiments and the s.e. bars denote the mean7s.d. of three replicates. (c) 293T cells were transiently transfected with empty vector, myc-Sf-Stk, myc-Sf-Stk/Gab1 or myc-Sf-Stk/Gab2. Cell lysates were probed with antiphospho-Akt, anti-Akt, antiphospho-Erk, anti-Erk and anti-Myc. (d) Total bone marrow from C57Bl/6 mice was transduced with Sf-Stk/Gab2 or KDSf-Stk/Gab2. Cells were plated and BFU-E were analysed as described in (b). *Po0.05, ***Po0.001. infection. However, expression of the analogous Sf-Stk/ transformation of erythroid progenitor cells. Protein Gab1 fusion protein in these cells repeatedly failed to expression of the mutant proteins was confirmed in the support the growth of FV-infected erythroblasts packaging cells with anti-Myc (Figure 5c and d), and (Figure 4c). The kinase activity of Sf-Stk was required equal infection efficiency in primary bone marrow cells for the ability of Sf-Stk/Gab2 to support Epo-indepen- was confirmed by flow cytometry with anti-Myc dent colony formation as demonstrated using a kinase- following cell permeabilization (Figure 5e). Interest- dead fusion protein (Figure 4d), suggesting that Sf-Stk ingly, mutation of either the p85 or Shp2-binding sites in kinase activity is required for more than the recruitment Gab2 resulted in both reduced Erk phosphorylation and of Gab2 to the docking site tyrosines. Akt phosphorylation in 293T cells compared with the We and others have shown previously that activation wild-type Sf-Stk/Gab2 fusion protein (Figure 5f). of the Erk- and PI3K-signaling pathways by this family of receptors is required for cellular transformation. Both Gab1 and Gab2 contain three p85-binding sites, Gab2 mediates the efficient expansion of Friend resulting in PI3K activation, and two Shp2-binding virus-infected cells in vitro and in vivo sites that lead to the activation of Erk through an as yet To further study the potential role of Gab2 in the unidentified mechanism. Therefore, we mutated either growth of Friend virus-infected cells, we examined the the p85- or Shp2-docking sites in Gab2 in the context of effect of a targeted mutation in Gab2 on the progression the Sf-Stk/Gab2 fusion protein and tested the ability of of Friend disease. Mice homozygous for a deletion in these mutants to support Friend virus-induced colony Gab2 on a mixed genetic background were crossed to formation. Mutations in the three p85-binding sites the Fv2-sensitive BALB/c strain of mice and resultant (Figure 5a) or two Shp2-binding sites (Figure 5b) heterozygous mice were bred to obtain progeny with resulted in reduced colony number, but not colony size varying alleles of Gab2 and Fv2 (Figure 6a and b). The (data not shown), in response to FV infection, indicating mice were injected with FV i.v. and spleen size was that these twosignaling pathways play a key rolein the determined 2 weeks later. Data from these studies

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2438 a c e Myc-Sf-Stk-Gab2

p85 Myc-Sf-Stk-Gab2∆p85 ∆ 100 50 Myc-Sf-Stk-Gab2∆Shp85 Epo 40 FvP 30

Myc-Sf-Stk-Gab2 Myc-Sf-Stk-Gab2 20 Myc-SfStk Event 10

Number of BFU-e ** 0

-10 MSCV MycSfStk Myc/SfStk- 100 101 102 103 -Gab2 Gab2∆p85 FITC

b d f 20 Epo 1 2 3 4 FvP

Shp2

15 ∆ α-p-Erk

10 α-Erk 5 **

vector Myc-Sf-Stk-Gab2 Myc-Sf-Stk-Gab2 Number of BFU-e 0 α-p-Akt

-5 α-Akt MSCV MycSfStk Myc/SfStk- -Gab2 Gab2∆Shp2 Figure 5 Efficient colony formation of Friend virus-infected cells in the presence of Sf-Stk-Gab2 requires the p85- and Shp2-binding sites of Gab2. (a) Total bone marrow from C57BL/6 mice was transduced with empty vector, Myc-Sf-Stk/Gab2 or Myc-Sf-Stk/Gab2- harboring Y to F mutations at the three p85-binding sites (Myc-Sf-Stk/Gab2Dp85). Transduced cells were plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence/absence of FV or Epo (1U/ml). (b) Total bone marrow from C57BL/6 mice was transduced with empty vector, Myc-Sf-Stk/Gab2 or Myc-Sf-Stk/Gab2 harboring Y to F mutations at the two Shp2-binding sites (Myc-Sf-Stk/ Gab2DShp2). Transduced cells were plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence/absence of FV or Epo (1 U/ ml). BFU-E were stained with acid-benzidine and scored on day 5. The data shown is representative of three independent experiments and s.e. bars denote the mean7s.d. of three replicates. **Po0.01. (c) Expression of Myc-Sf-Stk, Myc-Sf-Stk/Gab2 and Myc-Sf-Stk/ Gab2Dp85 in 293T cells. (d) Expression of Myc-Sf-Stk/Gab2 and Myc-Sf-Stk/Gab2DShp2 in 293T cells. (e) Total bone marrow from C57BL/6 mice was transduced with vector (shaded area), Myc-sf-stk/Gab2, Myc-sf-stk/Gab2Dp85 or Myc-sf-stk/Gab2DShp2. Transduced cells were permeabilized and stained with anti-Myc antibody, followed by an FITC-conjugated secondary antibody. (f) 293T cells were transiently transfected with (1) vector control, (2) Myc-Sf-Stk/Gab2Dp85, (3)Myc-Sf-Stk/Gab2DShp2 or (4) Myc- Sf-Stk/Gab2. Erk phosphorylation was examined using antiphospho-Erk. Blots were stripped and reprobed for total Erk. Akt was immunoprecipiated with anti-Akt, and resulting lysates were blotted with antiphospho-Akt. Blots were stripped and reprobed for Akt.

indicate that, while the Gab2 mutation had no effect on defect (Figure 7b). Both Gab1 and Gab2 were efficiently the response of Fv2-resistant strains of mice to Friend expressed in 293T-packaging cells (Figure 7c). Mutation virus, Gab2À/À animals on a Fv2-sensitive background of either the p85- or Shp2-binding sites in wild-type exhibited significantly reduced spleen size following Gab2 resulted in reduced colony formation (Figure 7d), infection with Friend virus when compared to wild-type reflecting the previous results using the chimeric and Gab2 þ /À mice (Figure 6c). These data support a proteins. Taken together, these data demonstrate that role for Gab2 in the progression of Friend virus-induced recruitment of Gab2, but not Gab1, to Sf-Stk is erythroleukemia in vivo. sufficient to support the growth of primary erythroblasts Todetermine whether Gab2 regulates the expansion in response to FV. of infected erythroid progenitor cells in vitro, we infected bone marrow cells from wild-type and Gab2À/À mice with Friend virus and assessed colony formation in the presence or absence of Epo. While bone marrow cells Discussion from wild-type controls formed Epo-independent colo- nies in response to FV infection as expected, the virus Grb2 has emerged as a central adapter molecule in RTK failed to induce colony formation following infection of signaling in both invertebrates and mammals. Grb2 is bone marrow cells from Gab2À/À mice (Figure 7a). The recruited to phosphotyrosine motifs through a central defect in Gab2À/À erythroblasts could be rescued by SH2 domain, leading to the activation of the MAPK exogenous expression of wild-type Gab2, however, and PI3K pathways via binding of flanking SH3 expression of Gab1 in these cells did not rescue this domains to SOS and a number of adapter proteins,

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2439 a b including the Gab family. While gene-targeting experi- ments demonstrated a requirement for Grb2 in endo- derm differentiation at day 4, this defect could be Gab2 +/+ Gab2 -/- Gab2 +/- Fv2 s/r Fv2 s/s Fv2 r/r rescued by a Grb2/SOS fusion protein indicating that the primary role of Grb2 in early development may be to activate the ras/MAPK pathway (Cheng et al., 1998). Subsequent studies have revealed gene dosage-depen- c 4.0 dent functions for Grb2. A hypomorphic mutation, 3.5 Gab2 +/+, Fv2 resistant analogous to one characterized in the Caenorhabditis 3.0 elegans Grb2 ortholog Sem-5, which reduces binding of Gab2 -/-, Fv2 resistant 2.5 the SH2 domain to phosphotyrosine, results in perinatal *** 2.0 Gab2 +/+, +/-, Fv2 sensitive lethality, and fibroblasts from these mice showed a 1.5 Gab2 -/-, Fv2 sensitive defect in Erk activation and Gab1 tyrosine phosphor- 1.0 ylation (Stern et al., 1993; Saxton et al., 2001). In this

Spleen weight (grams) weight Spleen Average 0.5 report, we show that Grb2 þ /À animals exhibit 0.0 012345 decreased susceptibility toFriend erythroleukemia, adding to the growing list of defects observed in the Figure 6 Gab2À/À mice exhibit reduced susceptibility toFriend adult animal due toGrb2 haplo-insufficiency. Futher- virus in vivo. Gab2À/À mice on a mixed genetic background were crossed with sensitive BALB/c mice and the Fv2 locus was more, our data indicate that the primary role of Grb2 in genotyped. Mice (8–10 weeks old) of Gab2À/À, þ /À, and WT this processis torecruit Gab2 tothe Sf-Stk receptor. genotype (a)onFv2-sensitive and resistant backgrounds (b) The Gab (Grb2-associated binder) family of adapter were injected with 200 ml FVP virus supernatant collected from proteins are the mammalian homologues of Drosophila FP63 cells. Spleens were weighed 2 weeks postinjection (c). Lanes: (1) Gab2 þ / þ , Fv2 resistant, (2) Gab2À/À, Fv2 resistant, (3) DOS (daughter of sevenless) and C. elegans Soc1, which Gab2 þ / þ , þ /À, Fv2 sensitive, (4) Gab2À/À, Fv2 sensitive. are critical for signaling downstream of RTKs and thus ***Po0.001. development of the fly and nematode, respectively (Herbst et al., 1996; Raabe et al., 1996; Schutzman et al., 2001). There are three mammalian Gabs and the

ab60

50 40

40 Epo Epo FvP 30 FvP 30 20 20 Number of BFU-e

Number of BFU-e * 10 10

0 0 Gab2-/- Gab2+/+

MSCV Gab1 Gab2

cd35 Epo 30 FvP 25 MSCV Gab1 Gab2 ** 20 * 15 10 Blot Gab1/2 * Number of BFU-e 5 0 MSCV Gab2 Gab2(∆p85) Gab2(∆Shp2) Figure 7 Gab2À/À erythroblasts fail to form colonies in response to FV in vitro.(a) Total bone marrow from wild-type and Gab2À/À mice was incubated with or without FVP and plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence or absence of Epo. (b) Total bone marrow from Gab2À/À mice was transduced with empty vector, Gab1 or Gab2 followed by infection with FVP and plated in methylcellulose containing IL-3 (2.5 ng/ml) in the presence or absence of Epo. BFU-E were stained with acid-benzidine and scored on day 8. The data shown are representative of three independent experiments and s.e. bars denote the mean7s.d. of three replicates. (c) Expression of Gab1 and Gab2 in transiently transfected 293 cells. (d) Total bone marrow from Gab2À/À mice was transduced with empty vector, Gab2, Gab2Dp85 or Gab2DShp2. Cells were plated and BFU-E were assessed as described in (b). *Po0.05, **Po0.01.

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2440 critical nature of these proteins in RTK signaling in Friend virus-infected erythroid cell lines (Nishigaki mammals was highlighted by the phenotype of the Gab1 et al., 2000). However, the observation that Epo knock-out mice. These mice are embryonic lethal, stimulation of EpoR results in tyrosine phosphorylation exhibiting developmental defects similar to those ob- of Gab1 in HCD57 and UT7 cells (Wickrema et al., served in mice homozygous mutant for the MET RTK, 1999) could account for its phosphorylation in Friend with which Gab1 interacts directly through its MBD virus-infected cells. Therefore, our data do not rule out a (Bladt et al., 1995; Uehara et al., 1995; Sachs et al., role for Gab1 in this process, but only suggest that Gab1 2000). While Gab2 and Gab3 lack a MBD, all of the does not mediate the expansion of primary erythroblasts Gab proteins share a proline-rich domain that serves as in vitro following Friend virus infection. Further in vivo a consensus Grb2-binding site (Gu and Neel, 2003). studies will be required to assess the potential role of Gab1 recruitment tothe EGF receptoris dependent Gab1 in the progression of Friend disease. upon the Grb2-binding site, and cytokine receptors that In addition to the Grb2-binding site, all Gab proteins utilize the bc chain recruit Gab2 through a SHC–Grb2– contain an N-terminal PH domain with sequence Gab2 interaction (Gu et al., 2000; Lock et al., 2000; similarity to the Btk PH domain which recognizes Rodrigues et al., 2000). Our data indicate that Sf-Stk primarily PIP3, the lipid products of PI3K (Isakoff recruits Gab2 through an indirect interaction with Grb2, et al., 1998). Therefore, the requirement for PI3K and that the ability of Sf-Stk/Grb2 to recruit Gab2 is a signaling in the transformation of cells by FV could critical mediator in the progression of Friend disease. also reflect a role for this signaling pathway in the However, the Gab2À/À mice are not completely recruitment of Gab2. A PH domain deletion mutant of resistant toFV-induced disease as demonstrated by the Gab1 which fails to support Met-induced morpho- enhanced spleen size when compared with FV2-resistant genesis, fails totranslocatein a PIP3-dependent manner, controls. The residual growth of infected cells in the and a Met/Gab1 fusion protein lacking the PH domain Gab2À/À animals could be due to compensation by fully supports this process (Maroun et al., 1999a, b). other adaptor proteins, or the retained ability of Grb2 to Conversely, while mutations in the PH domain of DOS recruit SOS. Interestingly, transformation by BCR/Abl inhibit its ability tofunctionin rescue experiments, this has alsorecently been showntodepend onthe mutation does not affect its ability to localize to the recruitment of a Grb2/Gab2 complex, suggesting the membrane suggesting that the PH domain of DOS potential of a broader role for Grb2/Gab2 in the may also play a role in propagating downstream development of myeloid (Sattler et al., 2002). signals (Dickson et al., 1992; Bausenwein et al., 2000; All Gab family members contain multiple p85- Schutzman et al., 2001). While overexpression of a PH binding sites that become phosphorylated upon receptor domain from Gab1 blocked the ability of FV-infected activation (Liu and Rohrschneider, 2002), and we bone marrow to form Epo-independent colonies (un- demonstrate here that the p85-binding sites in Gab2 published observation), deletion of the PH domain of are required for efficient cytokine-independent growth Gab2 in our fusion proteins did not appear to affect the of Friend virus-infected erythroblasts. The ability of ability of Sf-Stk/Gab2 to support the Epo-independent Gab proteins to amplify Sf-Stk signaling through the growth of infected cells. It is therefore likely that the PH activation of PI3K is consistent with previous findings domain is required for the recruitment of Gab2 to the from our laboratory that demonstrate a requirement for signaling complex, but not for generating further PI3K activation in the transformation of cells by FV signaling events. (Park and Hayman, 1999; Finkelstein et al., 2002). Our The activation of MAPK is also required for the data also extend previous studies demonstrating con- transformation of cells by FV, and previous studies have stitutive tyrosine phosphorylation of Gab2 and its demonstrated that while Gab2-mediated transformation association with p85 in FV-infected cells grown in the of cells in response to v-sea requires PI3K activation, absence of Epo (Lecoq-Lafon et al., 1999; Wickrema efficient transformation also requires binding of Gab2 to et al., 1999; Nishigaki et al., 2000). In these cells, PI3K is bind the SH2-containing protein phosphatase, SHP-2 constitutively active, and PI3K activity, but not EpoR (Ischenko et al., 2003). In addition, constitutively active phosphorylation, is required for the proliferation of forms of SHP-2 have been implicated in the progression these cells in the absence of Epo (Nishigaki et al., 2000). of some human leukemias (Neel et al., 2003). All Gab Furthermore, v-sea, which induces erythroleukemia in proteins contain two tandem tyrosines in the C-terminal chickens, preferentially interacts with Gab2 and is portion of the protein which, upon phosphorylation, essential for cellular transformation induced by v-sea result in the recruitment of SHP-2. In our studies, (Ischenko et al., 2003), and mutagenesis studies revealed mutation of these two tyrosines in the context of the a critical role in v-sea-induced transformation for PI3K Sf-Stk/Gab2 fusion protein, reduced the ability of this activation (Park and Hayman, 1999). fusion to support the growth of primary erythroblasts in We were surprised by the observation that Gab1 does response to Friend virus. Mutation of these tyrosines, not associate with Sf-Stk, is not phosphorylated in but not the p85-binding sites, in DOS and Soc1 also response to Sf-Stk expression and does not support the failed to rescue loss-of-function alleles, suggesting a growth of primary erythroblasts downstream of Sf-Stk. critical role for the recruitment of SHP-2 by Gab Gab1 has previously been shown to bind human (Herbst et al., 1999; Schutzman et al., 2001). Further- Ron through its MBD(van den Akker et al., 2005), more, SHP-2-binding sites in Gab are critical for the and Gab1 is alsoconstitutivelyphosphorylatedin induction of branching morphogenesis in epithelial cells

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2441 by Met (Maroun et al., 2000; Schaeper et al., 2000). loci, the primers are 50-GGTGGGTTTAACGGTTAGGG-30 SHP-2 contains two tandem SH2 domains and the (forward) and 50-TCTGGGCTCTGCCTCCTTAT-30 (re- N-terminal SH2 domain serves an autoinhibitory function verse). The PCR parameters are: 951C 1 min followed by 30 in the absence of binding to phosphorylated tyrosines on cycles of 951C30s,551C 30 s and 721C 30 s. The PCR products adjacent proteins (Feng, 1999). Recruitment of SHP-2 then loaded on 12.5% acrylamide gel electrophoresis to visualize the DNA fragment by eherdium bromide staining. by Gab most likely has the dual function to activate the The sensitive allele amplifies a 53 bp fragment and resistant PTP activity of SHP-2 and to bring SHP-2 in proximity allele 50 bp. There are three bands for the heterozygous mice. toits substrates at the membrane. Interestingly, like the BALB/c and C57BL/6 mice were obtained from Jackson Grb2-SOS-Ras-MAPK pathway, recent studies have Laboratories. All research involving the use of mice were suggested that the Grb2-Gab1-SHP-2 pathway results in performed in strict accordance with protocols approved by the the activation of Erk, and may play a role in the Institutional Animal Care and Use Committee of Pennsylvania enhanced or sustained activation of this signal (Cunnick State University. For in vivo infections with Friend virus, wild- et al., 2001). type and mutant mice were tail-vein injected with 200 ml of FV. The three mammalian Gab proteins, despite their high At 2 weeks postinjection, mice were killed and spleens degree of homology, appear to play largely non- removed. The spleen was weighed with an analytical balance and measured in grams. overlapping roles in vivo (Wolf et al., 2002). Mice with targeted deletions in Gab2 and Gab3, unlike Gab1, are viable. Gab2-deficient mice display defects in allergic Plasmid construction responses due to defective IgE receptor signaling in bone All the restriction and modification enzymes used here were marrow-derived mast cells, whereas Gab3-deficient mice from New England BioLabs, unless otherwise stated. The Grb2 cDNA was subcloned in-frame with EGFP into the XhoI appear normal to date (Gu et al., 2001; Seiffert et al., site of the pEGFP vector (Clontech) and subsequently the 2003). Furthermore, while both Gab1 and Gab2 have 1.4 kb EGFP-tagged Grb2 was subcloned in the HpaI site of similar effects on the activation of Erk by growth factor the MSCV-neo retroviral vector (provided by A Henderson, stimulation, they have reciprocal effects in mediating Pennsylvania State University) to produce MSCV-neo-EGFP- Elk induction (Zhao et al., 1999). In this study, the Grb2. Grb2 mutants MSCV-neo-EGFP-Grb2W36 K, MSCV- resistance of Gab2À/À mice toFriend virus was not neo-EGFP-Grb2W193 K and MSCV-neo-EGFP-Grb2W36, absolute, supporting the existence of other compensat- 193 K were produced by site-directed mutagenesis using the ing signals, perhaps provided by other Gab family Stratagene QuickChanget kit (Stratagene), following the members. However, the exogenous expression of Gab1 manufacturer’s protocol. Using MSCV-neo-EGFP-Grb2 as failed to rescue the Epo-independent growth of Friend template, the mutations were introduced with the following oligos: MSCV-neo-EGFP-Grb2W36 K, N-terminal SH3 mu- virus-infected erythroblasts from Gab2À/À mice in tation: sense-50-CGAAGAATGTGATCAGAACAAGTACA vitro. It will therefore be of future interest to take AGGCAGAGCTTAATGG-30; MSCV-neo-EGFP-Grb2W193 K, advantage of this system in order to examine the C-terminal SH3 mutation: sense-50-GGATAACTCAGACCC functional similarities and/or differences in signaling CAACAAGTGGAAAGGAGCTTGC-30. MSCV-neo-myc/Sf- through the Gab family of adaptor proteins and to map Stk was described previously (Finkelstein et al., 2002). the domains of Gab2 responsible for the transformation The murine Gab2 coding sequence was RT–PCR amplified of primary hematopoietic cells. from mouse liver total RNA (RNAeasy Kit from Qiagen) using the forward primer 50-CGGCGGGCTCCAGTTTAG CCG-30 and reverse primer 50-CCCCTTCATTACAGCTT GGCACCC-30. The PCR product was cloned into pcDNA3.1 Materialsand methods (Invitrogen) at the EcoV site toproducepcDNA3.1-Gab2. To generate the Myc-Sf-Stk and Gab2 fusion, the Myc/Sf-Stk Mice fragment was amplified from MSCV-neo-Myc/SF-Stk using Fv2 Grb2 þ /À mice, on a -sensitive BALB/c background, were primers 50-gcaggatcccatcgatttaaagc-30 (forward) and 50-atgc et al genotyped as previously described (Cheng ., 1998). tagcagtgaggccactacctgc-30 (reverse) and the Gab2 fragment Gab2À/À mice on a mixed genetic background were crossed was amplified from pcDNA3.1-Gab2 with primers 50-atgc for one generation onto BALB/c. Genotypes of Gab2 knock- tagcCGGCTTCAATCAGGCTGAAGAG-30 (forward) and out mice (generated by GS Feng, The Burnham Institute, to be vector primer 50-ctagaaggcacagtcgaggctg-30 (reverse). The described elsewhere) were determined by PCR with the fragments were digested with proper enzymes, purified and appropriate primers. For preparing genomic DNA, about three-way ligated intopcDNA3.1 vector.The fused gene was 5 mm of mouse tails were clipped and lysed in lysis buffer then cut out from pcDNA3.1 vector with PmeI and subcloned (100 mM Tris-Cl, pH 8.0, 5 mM EDTA, 0.1% SDS, 100 ml intoMSCV-neoat the HpaI site to produce MSCV-neo-Myc/ 1 Proteinase K) at 55 C overnight. After centrifugation, add 2 SF-Stk-Gab2. The mutations in Myc/Sf-Stk-Gab2Y3F were volume of ethanol to cleared lysate to precipitate the genomic introduced by site-directed mutagenesis (Stratagene Quick- DNA. The pellet was washed with 75% ethanol; air dried and Changet Kit) following the manufacturer’s protocol. The resuspended in 100 mlof10mM Tris-Cl pH 8.0. For the 0 template was MSCV-neo-Myc/SF-Stk-Gab2 and the primers genotype of Gab2, primer pair 5 -GGCTTACAGAGCCTCT 0 0 0 (sense) were as follows: 5 -ccagctctgatgacaactTcgtgcccatgaacc GACCCAG-3 (forward) and 5 -CTGCTATACCTCTGC 0 0 0 0 cagg-3 (Y441F); 5 - gacaactcccagagtgtctTcatccccatgagcccagg-3 CATGGTGGG-3 (reverse) were used. The PCR parameters (Y465F); 50-ggagacagtgaggagaactTtgtccctatgcaaaaccc-30 (Y574). are: 951C 1 min followed by 30 cycles of 951C30s,551C30s and 721C 1 min and 30 s. The PCR products were loaded onto 1% agarose gel for eletrophoresis and visualized with Retrovirus production etherdium bromide staining. The wild-type allele amplifies a 293T cells were transiently transfected by either CaPO4 1.5 kb fragment and null allele a 0.7 kb fragment. For the Fv2 coprecipitation (Wigler et al., 1979) or TransIT-293

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2442 transfection reagent (Mirus corporation) using 1–2 mg pEco triplicate with or without 1 U/ml Epo (R&D Systems). and the appropriate MSCV-neo constructs (1–10 mg). For Cultures were incubated for 2–8 days in 5% CO2 at 371C. CaPO4 coprecipitation, cells were transfected for 16 h at 371C, Erythroid colonies (BFU-E and CFU-E) were visualized by where the media was then aspirated, changed, and cells were acid-benzidine staining as previously described (Finkelstein grown for an additional 30 h at 371C prior to harvest of the et al., 2002). For in vitro infection of bone marrow, cells were viral supernatant. For TransIT-293 transfection, cells were harvested from various strains of mice and washed in PBS. transfected for 48–72 h at 371C prior to harvest of the viral Bone marrow was infected with viral supernatant from the supernatant. Protein expression of MSCV-EGFP, MSCV- transient transfections for 20 h, as previously described (Finkel- myc/Sf-Stk-EGFP, MSCV-neo-EGFP-Grb2, MSCV-neo- stein et al., 2002). The cells were then infected with FVP and EGFP-Grb2W36K, and MSCV-neo-EGFP-Grb2W193K was suspended in methocult 3234 containing IL-3 (2.5 ng/ml) verified by Western blot, by probing with antibodies against (Peprotech), either in the presence or absence of Epo (1 U/ml). the EGFP- and myc-tagged fusions (Clontech and Cell Signaling, respectively). RT–PCR Expression of Gab1, 2 and 3 mRNA in total bone marrow, Antibodies, immunoprecipitation, and Western blotting liver, spleen, and FV infected spleen was determined by Mouse anti-Myc (1:1000 dilution), mouse anti-phospho- RT–PCR. Total RNA was obtained from the above samples tyrosine (1:1000 dilution) and mouse anti-GFP (1:2000 using the RNeasy kit (Qiagen). cDNA was generated by dilution) antibodies were purchased from Cell Signaling. GeneAmp RNA PCR kit components where reverse tran- Rabbit anti-Gab2 antibody (1:200 dilution) was purchased scription of RNA was followed by PCR amplification of from Upstate Biotechnology. Anti Gab2 was used to detect the cDNA (Applied Biosystems). For Gab1 gene the primers are Gab2 protein, anti-GFP antibody was used to detect Grb-GFP 50-GGAGGTGTCTCGGGTGAAGAGC-30 (sense) and 50-CG fusion and mutants, and anti-Myc antibody was used to detect GCAGAGGCGACGGCATG-30 (antisense) which amplify a the expression of Myc/sfStk, Myc/sfStk-Gab2 fusion protein 0.8 kb fragment, for Gab2 gene the primers are 50-ATGTCC and mutants. CAACCACTCCTCTCTCAGC-30 (sense) and 50-CCATAGC 293T cells were transiently transfected with plasmids CAGCAGGGTAGAAGAAC-30 (antisense) which amplify expressing desired protein. At 24 h post-transfection, cells a 0.43 kb fragment, and for Gab3 gene the primers are were lysed in lysis buffer (1% Digitonin, 150 mM NaCl, 0.4 mM 50-CCCAGTGCTGAAGACAGCTATGTGC-30 (sense) and EDTA, 2 mM Na3VO4,10mM NaF, and 2 mM PMSF). After 50-atcagtccatgcctgcttggtgctc-30 (antisense) which amplify a 20 min centrifugation, the cleared cell lysates were incubated 0.6 kb fragment. 1 with antibodies and protein A-sepharose at 4 C overnight. The Of the reaction mixture, 10% was electrophoresed on a 1% immunoprecipitates were analysed by reduced SDS–PAGE agarose gel (Fisher Scientific) and stained with ethidium and transferred toPVDF membrane. The blotswere incubated bromide (International Biotechnologies). RT cycling para- with antibodies and detected with ECL (Amersham). meters were: 5 min at 701C, 40 min at 401C, 5 min at 991C, then hold at 41C. PCR cycling parameters were: 1 min at 951C, 20 s Colony assay and in vitro infection at 941C, 30 s at 621C, 1 min at 721C, for 40 cycles. For Epo-independent colony analysis, total bone marrow cells from Grb2 þ /À and wild-type control mice were harvested and incubated with supernatant from cells expressing poly- Acknowledgements cythemia-inducing FV (derived from FP63 cells, Alan Bern- stein, Mount Sinai Hospital, Toronto, ON, Canada) and This work was supported by Grants NIH (R01-L66471) to DMEM (10% FBS, P/S, L-glu) on ice for 1 h. Cells were then PHC, NIH(R01-HL66208) toGSF, LLS scholaraward to plated in methocult M3534 (Stem Cell Technologies) in PHC and ACS (PF-01-121-LIB) toLDF.

References

Axelrad A. (1969). Proc Can Cancer Conf 8: 313–343. Gu H, Maeda H, Moon JJ, Lord JD, Yoakim M, Nelson BH Bausenwein BS, Schmidt M, Mielke B, Raabe T. (2000). Mech et al. (2000). Mol Cell Biol 20: 7109–7120. Dev 90: 205–215. Gu H, Neel BG. (2003). Trends Cell Biol 13: 122–130. Behringer RR, Dewey MJ. (1985). Cell 40: 441–447. Gu H, SaitoK, Klaman LD, Shen J, Fleming T, Wang Y et al. Ben-David Y, Bernstein A. (1991). Cell 66: 831–834. (2001). Nature 412: 186–190. Bladt F, Riethmacher D, Isenmann S, Aguzzi A, Birchmeier C. Hayman MJ, Kitchener G, Vogt PK, Beug H. (1985). Proc (1995). Nature 376: 768–771. Natl Acad Sci USA 82: 8237–8241. Cheng AM, Saxton TM, Sakai R, Kulkarni S, Mbamalu G, Herbst R, Carroll PM, Allard JD, Schilling J, Raabe T, Simon Vogel W et al. (1998). Cell 95: 793–803. MA. (1996). Cell 85: 899–909. Cunnick JM, Mei L, Doupnik CA, Wu J. (2001). J Biol Chem Herbst R, Zhang X, Qin J, Simon MA. (1999). EMBO J 18: 276: 24380–24387, Epub 2001 Apr 25. 6950–6961. D’Andrea AD. (1992). Cancer Surv 15: 19–36. Hibi M, HiranoT. (2000). Leuk Lymphoma 37: 299–307. Dickson B, Sprenger F, Morrison D, Hafen E. (1992). Nature Horoszewicz JS, Leong SS, Carter WA. (1975). J Natl Cancer 360: 600–603. Inst 54: 265–267. Feng GS. (1999). Exp Cell Res 253: 47–54. Isakoff SJ, Cardozo T, Andreev J, Li Z, Ferguson KM, Finkelstein LD, Ney PA, Liu QP, Paulson RF, Correll PH. Abagyan R et al. (1998). EMBO J 17: 5374–5387. (2002). Oncogene 21: 3562–3570. IschenkoI, PetrenkoO, Gu H, Hayman MJ. (2003). Oncogene Fournier T, Kamikura D, Teng K, Park M. (1996). J Biol 22: 6311–6318. Chem 271: 22211–22217. Kohda D, Terasawa H, Ichikawa S, Ogura K, Gong Q, Cheng AM, Akk AM, Alberola-Ila J, Gong G, Hatanaka H, Mandiyan V et al. (1994). Structure 2: Pawson T et al. (2001). Nat Immunol 2: 29–36. 1029–1040.

Oncogene Requirement for Grb2/Gab2 in Friend erythroleukemia HE Teal et al 2443 Lecoq-Lafon C, Verdier F, Fichelson S, Chretien S, Rodrigues GA, Falasca M, Zhang Z, Ong SH, Schlessinger J. Gisselbrecht S, Lacombe C et al. (1999). Blood 93: (2000). Mol Cell Biol 20: 1448–1459. 2578–2585. Sachs M, Brohmann H, Zechner D, Muller T, Hulsken J, Lewitzky M, Kardinal C, Gehring NH, Schmidt EK, Walther I et al. (2000). J Cell Biol 150: 1375–1384. Konkol B, Eulitz M et al. (2001). Oncogene 20: 1052–1062. Sattler M, Mohi MG, Pride YB, Quinnan LR, Malouf NA, Li JP, D’Andrea AD, Lodish HF, Baltimore D. (1990). Nature Podar K et al. (2002). Cancer Cell 1: 479–492. 343: 762–764. Saxton TM, Cheng AM, Ong SH, Lu Y, Sakai R, Cross JC LiaoSK, Axelrad AA. (1975). Int J Cancer 15: 467–482. et al. (2001). Curr Biol 11: 662–670. Lilly F. (1970). J Natl Cancer Inst 45: 163–169. Schaeper U, Gehring NH, Fuchs KP, Sachs M, Kempkes B, Liu Y, Rohrschneider LR. (2002). FEBS Lett 515: 1–7. Birchmeier W. (2000). J Cell Biol 149: 1419–1432. Lock LS, Royal I, Naujokas MA, Park M. (2000). J Biol Chem Schutzman JL, Borland CZ, Newman JC, Robinson MK, 275: 31536–31545. Kokel M, Stern MJ. (2001). Mol Cell Biol 21: Maroun CR, Holgado-Madruga M, Royal I, Naujokas MA, 8104–8116. Fournier TM, Wong AJ et al. (1999a). Mol Cell Biol 19: Seiffert M, Custodio JM, Wolf I, Harkey M, Liu Y, Blattman 1784–1799. JN et al. (2003). Mol Cell Biol 23: 2415–2424. Maroun CR, Moscatello DK, Naujokas MA, Holgado- Smith DR, Vogt PK, Hayman MJ. (1989). Proc Natl Acad Sci Madruga M, Wong AJ, Park M. (1999b). J Biol Chem USA 86: 5291–5295. 274: 31719–31726. Stern MJ, Marengere LE, Daly RJ, Lowenstein EJ, Kokel M, Maroun CR, Naujokas MA, Holgado-Madruga M, Wong AJ, Batzer A et al. (1993). Mol Biol Cell 4: 1175–1188. Park M. (2000). Mol Cell Biol 20: 8513–8525. Tambourin P, Wendling F. (1971). Nat New Biol 234: Mirand EA, Steeves RA, Lange RD, Grace Jr JT. (1968). Proc 230–233. Soc Exp Biol Med 128: 844–849. Uehara Y, Minowa O, Mori C, Shiota K, Kuno J, Noda T Moreau-Gachelin F, Tavitian A, Tambourin P. (1988). Nature et al. (1995). Nature 373: 702–705. 331: 277–280. van den Akker E, van Dijk T, Parren-van Amelsvoort M, Mowat M, Cheng A, Kimura N, Bernstein A, Benchimol S. Grossman K, Schaeper U, Toney-Early K et al. (2005). (1985). Nature 314: 633–636. Blood 103: 4457–4465. Neel BG, Gu H, PaoL. (2003). Trends Biochem Sci 28: Wickrema A, Uddin S, Sharma A, Chen F, Alsayed Y, 284–293. Ahmad S et al. (1999). J Biol Chem 274: 24469–24474. Nishigaki K, Hanson C, Ohashi T, Thompson D, Muszynski Wigler M, Pellicer A, Silverstein S, Axel R, Urlaub G, Chasin K, Ruscetti S. (2000). J Virol 74: 3037–3045. L. (1979). Proc Natl Acad Sci USA 76: 1373–1376. Nishigaki K, Thompson D, Hanson C, Yugawa T, Ruscetti S. Wolf I, Jenkins BJ, Liu Y, Seiffert M, Custodio JM, Young P (2001). J Virol 75: 7893–7903. et al. (2002). Mol Cell Biol 22: 231–244. Park C, Hayman M. (1999). J Biol Chem 274: 7583–7590. Zhang S, Ren J, Khan MF, Cheng AM, Abendschein D, Persons D, Paulson R, Loyd M, Herley M, Bodner S, Muslin AJ. (2003). Arterioscler Thromb Vasc Biol 23: Bernstein A et al. (1999). Nat Genet 23: 159–165. 1788–1793, Epub 2003 Jul 3. Raabe T, Riesgo-Escovar J, Liu X, Bausenwein BS, Deak P, ZhaoC, Yu DH, Shen R, Feng GS. (1999). J Biol Chem 274: Maroy P et al. (1996). Cell 85: 911–920. 19649–19654.

Oncogene