Flt-1, a Receptor for Vascular Endothelial Growth Factor, Has Transforming and Morphogenic Potentials

Oncogene (1998) 16, 2585 ± 2595  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Flt-1, a receptor for vascular endothelial growth factor, has transforming and morphogenic potentials

Yoshiro Maru, Sachiko Yamaguchi and Masabumi Shibuya

Department of Genetics, Institute of Medical Science, University of Tokyo, Japan

A paradox of Flt-1, a tyrosine kinase receptor for been reported to stimulate DNA synthesis in response vascular endothelial growth factor (VEGF), is that the to VEGF (Millauer et al., 1993). However, Flt-1 does ligand cannot activate the receptor to stimulate growth not mediate cell growth when introduced into of cells that exogenously overexpress the receptor. In NIH3T3 cells or into porcine aortic endothelial cells order to ®nd Flt-1 kinase-dependent biological systems, that express no VEGF receptors (Seetharam et al., we obtained for the ®rst time activated forms of the Flt-1 1994; Waltenberger et al., 1994). The Flt-1-mediated kinase in a ligand-independent manner. Replacement of biological activities of VEGF elucidated so far the ABL sequences in the human leukemia oncoprotein include tissue factor production in endothelial cells BCR ± ABL with the cytoplasmic domain of Flt-1 (Clauss et al., 1996), nitric oxide release in (BCR ± FLT) followed by a retroviral random mutagen- trophoblast cells (Ahmed et al., 1997) and migration esis scheme gave constitutively active arti®cial chimera of monocytic cells (Barleon et al., 1996). More BCR ± FLTm with mutations within the Flt-1 sequence. recently Ziche et al. have claimed that the Flt-1- Like BCR ± ABL it could, but not the original BCR ± speci®c ligand placenta growth factor (PlGF)-1, when FLT, transform Rat1 ®broblasts, abrogate cytokine puri®ed from overexpressing eukaryotic cells, was as dependence in Ba/F3 cells, and induce neurite-like potent as VEGF to induce angiogenesis in the rabbit structures in neuronal PC12 cells. Interestingly, Rat1 cornea and the chick chorioallantoic membrane cells transformed by BCR ± FLTm formed tube-like assays (Ziche et al., 1997). These informations raise structures in basement membrane matrix. BCR ± FLTm a fundamental question of whether the Flt-1 tyrosine retroviruses may be a very useful tool to investigate an kinase activity can promote cell growth or it has as yet uncovered functions of the Flt-1 kinase. other biological functions. In order to address to this issue, we attempted to Keywords: Flt-1; VEGF; BCR ± ABL; tyrosine kinase; obtain activated forms of the Flt-1 kinase in a ligand- transformation independent fashion. The initial step for activation of receptor tyrosine kinases requires dimerization or oligomerization of receptors by ligand binding (Ullrich and Schlessinger, 1990; Heldin, 1995). The Introduction recent discovery of the chimeric transforming onco- protein Tel ± PDGFR suggests that the tyrosine kinase Angiogenesis is involved in many physiological as domain of the PDGFR family members could be well as pathological circumstances including wound constitutively activated by the fusion of helix ± loop ± healing, organ regeneration, cyclic proliferation of helix domain of an ets-related transcription factor Tel uterine endometrium, metastasis or neovascularization with a dimerizing potential (Golub et al., 1994). The of cancer, retinopathy in diabetes mellitus and so tetramer domain of the amino-terminal BCR fragment forth (reviewed in Folkman and Shing, 1992). One of with an alpha helical coiled-coil structure is essential the directly-acting mitogens involved in this process is for all biological activities for another transforming vascular endothelial growth factor (VEGF) (Ferrara tyrosine kinase BCR ± ABL (McWhirter et al., 1993a; and Davis-Smyth, 1997). Two separate VEGF Maru et al., 1996a). We replaced the ABL tyrosine receptors of platelet derived growth factor receptor kinase domain of P185 BCR ± ABL with that of Flt-1 (PDGFR) family have been identi®ed; Flt-1 (Shibuya and expected the activation of the Flt-1 kinase. et al., 1990; De Vries et al., 1992; Shibuya, 1995) and Surprisingly, the resulting arti®cial chimera BCR ± KDR/Flk-1 (Terman et al., 1992; Millauer et al., FLT had no biological activity as judged by Rat1 cell 1993). Recent studies on mice de®cient in either of transformation or growth factor independence in both these receptors suggested that Flt-1 may regulate the of which BCR ± ABL is active. Then we made our assembly of endothelial cells (Fong et al., 1995) and e€orts to obtain oncogenically mutated forms of the KDR/Flk-1 plays an important role in vasculogenesis BCR ± FLT by taking advantage of the highly and blood island formations (Shalaby et al., 1995). mutagenic nature of retrovirus life cycle (Goga et KDR/Flk-1 expressed in non-endothelial cells has al., 1993). Because retrovirally transduced oncogenic forms of c-kit or EGF receptor, v-kit or v-erbB, respectively, have multiple mutations in addition to the amino-terminal truncation (Nilsen et al., 1985; Qiu et al., 1988). Here we report the establishment of retroviruses expressing activated forms of BCR ± FLT Correspondence: Y Maru Received 28 October 1997; revised 11 December 1997; accepted 11 and show their transforming as well as morphogenic December 1997 potentials. Retroviruses expressing activated forms of VEGF receptor YMaruet al 2586 changes in its activity (Afar et al., 1994) and therefore Results can no longer serve as a control. The resulting arti®cial chimeras were named as BCR ± FLT and BCR ± TRK, BCR ± TRK is active in vivo while BCR ± FLT is not respectively. When they were expressed in the transient The ABL kinase domain and downstream sequences in 293T transfection system, BCR ± TRK was shown to BCR ± ABL were replaced with the intracellular be tyrosine-phosphorylated (Figure 1b, lane 8) at a portion of Flt-1 (Figure 1a). For an experimental high level comparable to that in BCR ± ABL. This control we also used the receptor for nerve growth suggests that other kinases of receptor type can factor Trk because its biological activities based on potentially be activated by fusion to BCR in the molecular analysis have been well studied (Martin- presence of the ABL SH3/SH2 domains. Surprisingly, Zanca et al., 1989; Coulier et al., 1990; Obermeier et however, anti-phosphotyrosine (PY) Western blotting al., 1994; Stephens et al., 1994). We avoided simple showed complete absence of PY-containing proteins in fusions of the BCR sequences with oligomerizing cells transfected with BCR ± FLT (Figure 1b, lane 7). ability to the kinase domain of the receptors. Because In order to examine whether or not BCR ± FLT can BCR ± ABL with a deletion in the SH2 domain has function as an enzyme in vitro, immunopuri®ed chimeric proteins were subjected to in vitro kinase assays followed by anti-PY Western blotting. As shown in Figure 1b (lanes 9 ± 12), autophosphorylation of a BCR ± FLT was detected, indicating that the Flt-1 kinase retains enzymatic activity even in the context of BCR ± FLT but that the activity may be down- regulated in vivo.

Establishment of retroviruses expressing activated forms of BCR ± FLT The facts that the Flt-1 kinase can hardly be activated in the context of BCR ± FLT in vivo but that it still retains the kinase activity in vitro may partly re¯ect extremely weak autophosphorylations or activation of Flt-1 extrinsically expressed in NIH3T3 cells (Seethar- am et al., 1994). They also suggest that the BCR tetramer domain-mediated oligomerization alone may not be sucient for activation of the Flt-1 tyrosine kinase. Considering VEGF-induced tyrosine phosphor- ylations of PLCg and GAP (GTPase-activating protein) for Ras in Flt-1-expressing NIH3T3cells, the Flt-1 kinase activity, although weak, can potentially b transmit signals but the activity may not be high enough to exert biological e€ects in the usual tissue culture system (Seetharam et al., 1994; Shibuya, 1995). MOCK MOCK MOCK BCR-FLT BCR-FLT BCR-FLT BCR-ABL BCR-TRK BCR-ABL BCR-TRK BCR-ABL BCR-TRK Therefore we challenged to obtain activated forms of 1234 5678 910 11 12 BCR ± FLT by utilizing the retroviral random muta- KD KD KD 205 — 205 — 205 — genesis scheme described before (Goga et al., 1993). Retroviruses have a highly mutagenic property in natural conditions and many of the well-known 116 — 116 — 116 — oncogenes virally transduced from host cells contain 80 — 80 — 80 — gain-of-function mutations (Nilsen et al., 1985; Qiu et al., 1988). Rat1 cells infected with helper-free retro- viruses expressing BCR ± FLT fused either in-frame or 49 — 49 — purposely out-of-frame were plated into soft agar, and we expected that Rat1 cells infected with oncogenically α BCR α PY α PY in vivo in vivo mutated BCR ± FLT retroviruses could give rise to an anchorage-independent growth. In three independent Figure 1 (a) Schematic representation of BCR ± TRK and soft agar assays performed in this fashion, four BCR ± FLT. Intracellular domains of nerve growth factor receptor (NGF-R or TRK) and vascular endothelial growth colonies (clone 1 ± 4) were observed in BCR ± FLT factor receptor (VEGF-R or FLT-1) were substituted for the ABL with an in-frame fusion while BCR ± FLT with an out- sequences in BCR ± ABL. Y490 and Y785 in TRK indicate of-frame fusion always gave no colonies and BCR ± tyrosine residues responsible for binding to Shc and PLCg, ABL uncountable number of colonies. Cells derived respectively. Arrows show the junctions between the ABL SH2 domain and the fused receptor sequences. TM=transmembrane from each of the four colonies plated back into liquid domain. (b) BCR ± TRK is activated while BCR ± FLT is not. culture revealed transformed morphology (data not Total protein lysates (lanes 1 ± 8) or anti-BCR immunoprecipitates shown). They also formed colonies when replated into after in vitro kinase reactions (lanes 9 ± 12) from 293T cells soft agar. Integrated retroviruses with BCR ± FLT transfected with vector alone (MOCK) (lanes 1, 5, 9), P185 cDNAs that were expected to contain activated BCR ± ABL (lanes 2, 6, 10), BCR ± FLT (lanes 3, 7, 11), and BCR ± TRK (lanes 4, 8, 12) were subjected to anti-BCR (lanes 1 ± mutations were rescued from those transformed clones 4) or anti-phosphotyrosine (lanes 5 ± 12) Western blotting by infection with replication-competent Molony Retroviruses expressing activated forms of VEGF receptor YMaruet al 2587 Murine Leukemia Virus. Reinfection of Rat1 cells with BCR ± FLT, a protein of around 150 KD (pp150) was the rescued viruses gave G418-resistant transformed strongly tyrosine-phosphorylated in every clone. This foci that formed colonies in soft agar (data not shown). protein is not BCR ± FLT itself that is posttranslation- To con®rm the expression of the BCR ± FLT ally modi®ed, because it could not be detected by anti- proteins in Rat1 clone 1 ± 4, anti-BCR Western BCR Western blotting (Figure 2a) and because an blotting was performed (Figure 2a). The expression exactly same-sized phosphoprotein was also observed levels of BCR ± FLT in non-transformed cells that in clone 2 that expresses a short-sized BCR ± FLT express the original BCR ± FLT and in transformed (Figure 2b). clones were roughly equal, and were also similar to Southern blot analysis of the Rat1 clones has those in BCR ± ABL-transformed cells (Figure 2a). The revealed that BCR ± FLT was integrated into the rat protein size of BCR ± FLT in clone 2 is approximately genome with a single copy (Figure 2d). Considering the 20 KD smaller than that of other clones (Figure 2a, sizes of bands in EcoRI-digested DNAs, clones 1, 3 lane 4). Anti-PY Western blotting revealed that and 4 are possibly identical. Polymerase chain reaction autophosphorylation intensities of activated BCR ± (PCR) analysis of DNAs and reverse transcriptase- FLTs and BCR ± ABL were approximately the same mediated PCR (RT ± PCR) of RNAs of those Rat1 (Figure 2b). Interestingly, tyrosine phosphorylation of clones showed that the Flt-1 sequence has a point GAP-associated p62Dok (Carpino et al., 1997; mutation at nucleotide 2743 (T to A) (amino acid 832 Yamanashi and Baltimore, 1997), which is a good Ser to Thr) in clones 1, 3 and 4 and that clone 2 has indicator for the BCR ± ABL-mediated transformation, mutations at nucleotide 3103 (G to T) (amino acid 952 was totally absent in the activated forms of BCR ± Gly to Cys) and at nucleotide 3835 (G to T) (amino FLT. In addition to autophosphorylation bands of acid 1196 Glu to stop). These data strongly support the

a c P185 BCR-ABL (clone 1) P185 BCR-ABL (MIX) BCR-FLT (clone 1) BCR-FLT (clone 2) BCR-FLT (clone 3) BCR-FLT (clone 4) BCR-FLT MOCK BCR-FLT (MIX) BCR-FLT 1 2345678 KD 205 — P185 BCR-ABL P160 cBCR d 116 — BCR-FLT αBCR Western C1 C2 C3 C4 C1 C2 C3 C4 PLASMID PLASMID MOCK 80 — MOCK Kb 12 345678910 11 12 13 14 23 — b 10 — 12345678 KD 6.5 — 205 — 4.5 — pp150 2.5 — 116 — 2.2 —

80 — 0.6 — pp62 EcoR l Xba l 49 — αPY Western

Figure 2 (a) and (b) Expression of activated forms of BCR ± FLT. Total protein lysates from Rat1 cells infected with MOCK (lane 1), the original BCR ± FLT retrovirus (lane 2, BCR ± FLT (MIX)), or transformed Rat1 clones from each of the four colonies (lane 3 ± 6), mixed population of P185 BCR ± ABL-transformed cells (lane 7, BCR ± ABL (MIX)), or a representative P185 BCR ± ABL clone from an agar colony (lane 8) were subjected to anti-BCR (a) or anti-PY Western blotting (b). Arrows indicate BCR ± ABL, BCR ± FLT (note a smaller protein size in clone 2 (lane 4)), endogenous P160 cBCR, phosphotyrosine-containing proteins around 150 KD (pp150) and GAP-associated tyrosine-phosphorylated proteins p62Dok (pp62). (c) and (d) Retrovirally transferred BCR ± FLT DNAs are integrated at a single site. (c) Restriction map of the BCR ± FLT cDNA cloned into retroviral expression vector (see Materials and methods and Figure 1). The 5' Xb or E to E fragment (designated as BCR) contains the BCR ®rst exon, ABL SH2, and SH3 domain sequences. The 3' E to Xb fragment (Flt-1) contains the kinase domain of the receptor Flt-1. Xb=XbaI, E=EcoRI. A thick bar indicates the DNA probe used for Southern blot analysis in (d). (d) Southern blot analysis of Rat1 cells grown from each of the four soft agar colonies in Figure 2a (C1 indicates clone 1). Ten mg of genomic DNAs from Rat1 cells (lanes 5, 12), or from transformed colonies (lanes 1 to 4, 8 to 11 for C1 to C4, respectively), 10 pg (lanes 6, 13) or 100 pg (lanes 7, 14) of BCR ± FLT DNAs in the retroviral expression vector were digested by either EcoRI (lanes 1 to 7) or XbaI (lanes 8 to 14), and were subjected to Southern blot analysis using 3' Flt-1 (SacI/XbaI) probe Retroviruses expressing activated forms of VEGF receptor YMaruet al 2588 interpretation of the data in Southern blot analysis and as v-ABL (Figure 3a). Although morphologies of explain the smaller protein size in clone 2. However, transformed Rat1 cells by each retrovirus were BCR and ABL sequences containing the SH2 and SH3 indistinguishable, a unique morphology were observed domains had no mutations. in cells transformed by BCR ± FLTm when cells were Thus we have established two retroviruses that starved of serum for 24 ± 48 h. They showed an express BCR ± FLT with oncogenic mutations. Since anastomosing reticular pattern that is morphologically the carboxy-terminal truncation in the activated form similar to capillary-like networks formed by endothelial of BCR ± FLT derived from clone 2 may result in loss cells (Folkman and Haudenschild, 1980; Montesano et of functions that belong to this region, we did most of al., 1983) (Figure 3b, 4), while cells transformed by the following experiments with the activated form other retroviruses appeared apoptotic without any derived from clone 1 that contains the full-length changes in cell alignment. BCR ± FLT and named it BCR ± FLTm. Tyrosine phosphorylation and recruitment of proteins in BCR ± TRK and BCR ± FLTm have transforming cells that express BCR ± TRK and BCR ± FLTm abilities Anti-BCR Western blotting of the transformed Rat1 It is known that Trk was initially identi®ed as a chimeric ®broblasts showed that BCR ± ABL, BCR ± FLTm and oncoprotein with tropomyosin that can transform BCR ± TRK were autophosphorylated (Figure 4, lanes NIH3T3 cells (Martin-Zanca et al., 1986). The 1 ± 10). The absence of tyrosine phosphorylations of recombination constitutively activates the Trk tyrosine GAP-associated p62Dok was not only reproduced in kinase. Given the fact that BCR ± TRK expressed in the cells reinfected with the BCR ± FLTm retrovirus but transient 293T transfection system was biochemically also was observed in BCR ± TRK-transformed cells activated (Figure 1), both BCR ± TRK and BCR ± (Figure 4, lanes 8 and 9). Tyrosine phosphorylations of FLTm were expected to have transforming potentials. p62Dok were reported in many tyrosine kinases When retrovirally expressed in Rat1 ®broblasts, they including v-ABL, v-FPS, v-SRC, epidermal growth induced not only morphological changes (Figure 3b) factor receptor (EGF-R), PDGF-R and Flt-1 but not but also anchorage-independent growth in soft agar. in Trk (Carpino et al., 1997; Ellis et al., 1990; Maher, Judging from the number of macroscopic colonies in 1988; Seetharam et al., 1994). However, anti-PY soft agar assays performed with roughly equivalent Western blotting of anti-GAP immunoprecipitates titers of retroviruses, both of them were as transforming from BCR ± FLTm or BCR ± TRK-expressing cells also revealed that tyrosine phosphorylation of p62Dok was hardly detectable (Figure 4, lanes 13 and 14). Presently we have no explanation for the 3. BCR-TRK a di€erence in p62Dok phosphorylation between Flt-1- expressing NIH3T3 cells and BCR ± FLTm-expressing Rat1 cells. Interestingly, another GAP-associated protein p190 (Settleman et al., 1992) was tyrosine phosphorylated in all of the transformed cells. Tyrosine phosphorylation of GAP appeared to correlate with that of p190 (Ellis et al., 1990) but it was under detection in BCR ± FLTm-transformed cells. b 1. MOCK 4. BCR-FLTm We then examined tyrosine phosphorylation of potential substrates or recruitment of proteins in BCR ± FLTm-transformed cells. We focused on three signal transduction molecules of biological importance: GRB-2, Shc, and PLCg. The tyrosine residue 177 (Y177) of BCR ± ABL is phosphorylated in BCR ± ABL and it transduces a Ras-mediated signal through recruitment with an SH2/SH3-containing adapter molecule GRB-2 that is complexed to SOS, a GDP/ 2. BCR-ABL 5. v-ABL GTP exchanger for Ras (Pendergast et al., 1993). Shc is highly tyrosine-phosphorylated in BCR ± ABL-trans- formed cells and makes an alternative Ras pathway (Puil et al., 1994; Goga et al., 1995; Maru et al., 1996a). Previously we have shown that PLCg could be tyrosine-phosphorylated with VEGF stimulation in NIH3T3 cells that overexpress Flt-1 (Seetharam et al., 1994). In Trk it is known that Shc and PLCg associate with the autophosphorylated receptor at Figure 3 Transforming potentials of BCR ± TRK and BCR ± FLTm. (a) Number of macroscopic colonies in soft agar assay. Y490 and Y785, respectively (Obermeier et al., 1994; Means of three experiments are shown. (b) A unique transformed Stephens et al., 1994). Both tyrosine residues are morphology found in BCR ± FLTm-expressing Rat1 cells. Rat1 demonstrated to be essential for neuronal differentia- cells transformed by retroviruses expressing the indicated kinases tion of PC12 cells. Since they are retained in BCR ± were starved of serum for 36 h in usual tissue culture dishes. Note TRK (Figure 1a), we expected that it may serve as a the transformed morphologies in all of them (2 ± 5) but anastomosing reticular patterns accompanied by realignment of good control to monitor the protein interactions cells were observed only in BCR ± FLTm (4) discussed above. Retroviruses expressing activated forms of VEGF receptor YMaruet al 2589 v-ABL v-ABL BCR-FLTm BCR-FLTm BCR-FLTm BCR-TRK MOCK MOCK MOCK v-ABL BCR-TRK BCR-ABL BCR-ABL BCR-ABL BCR-TRK KD 12345 678910 11 12 13 14 15 205 — KD KD 205 — 205 — P190

116 — 116 — GAP 116 —

80 — 80 — 80 —

P62

49 — 49 — IP: α GAP 49 — Blot: α PY Total Total Blot: α BCR Blot: α PY

IP: α GAP 116 — Blot: α GAP

Figure 4 Absence of p62Dok tyrosine phosphorylations in BCR ± FLTm- or BCR ± TRK-transformed Rat1 cells. Total cell lysates or anti-GAP immunoprecipitates from Rat1 cells transformed by MOCK (lanes 1, 6, 11), BCR ± ABL (lanes 2, 7, 12), BCR ± FLTm (lanes 3, 8, 13), BCR ± TRK (lanes 4, 9, 14), and v-ABL (lanes 5, 10, 15) were subjected to anti-BCR (lanes 1 ± 5), anti-PY (lanes 6 ± 10, upper panel of lanes 11 ± 15), anti-GAP (lanes 11 ± 15, lower panel) Western blotting. PY-containing bands of p190, p62 and GAP are indicated

Shc was tyrosine-phosphorylated in all of the comigrates with BCR ± FLTm was detected also in transformed cells (Figure 5a). Recruitment of autopho- cells transformed by activated BCR ± FLT derived sphorylated BCR ± TRK with Shc was evident as from clone 2 (Figure 6b, lanes 1 ± 3). Another reason expected (Figure 5a and b, lane 4). Although BCR ± is that anti-Flt-1 immunoprecipitates from cells FLTm bound to Shc was not detectable in anti-BCR expressing BCR ± FLTm (from clone 1) contained Western blotting, anti-PY Western blotting revealed a both BCR ± FLTm and pp150 while those from clone faint band of BCR ± FLTm (Figure 5a and b, lane 3). 2 showed the truncated form of BCR ± FLT but not Although both BCR ± ABL and v-ABL have been pp150, although in both cases roughly equivalent shown to bind to Shc (Puil et al., 1994; Ra€el et al., amounts of PLCg were recruited with the activated 1996), it was undetectable in this stringent condition. forms of BCR ± FLT (Figure 6b, lanes 4 ± 6). These Anti-GRB-2 Western blotting of anti-Flt-1 and anti- data suggest that pp150 is complexed to BCR ± FLTm Trk immunoprecipitates revealed the binding of GRB-2 through the carboxy-terminal Flt-1 sequences 1197 ± to BCR ± FLTm as well as to BCR ± TRK (Figure 5c). 1338, and that PLCg is recruited with BCR ± FLTm The binding of GRB-2 in BCR ± ABL is predominantly through other sequences. mediated by Y177 in the BCR sequence (Pendergast et al., 1993; Afar et al., 1994). In addition to Y177, Biological activities of BCR ± FLTm and BCR ± TRK BCR ± TRK also eciently recruits Shc which can bind distinct from those of BCR ± ABL to GRB-2. There is a potential GRB-2 binding site in the carboxy-terminal Flt-1 sequence (1213 YVNA) In addition to the transformation assay of Rat1 which is conserved in human, mouse, and rat (Shibuya, ®broblasts (Lugo and Witte, 1989), the biological 1995). This site may be involved in but certainly is not activities of BCR ± ABL can be tested in several required for GRB-2 binding in BCR ± FLTm because convenient assays, which include abrogation of growth the activated BCR ± FLT derived from clone 2, which factor dependence in interleukin-3 (IL-3)-dependent has a stop codon at 1196, was also recruited with cell line Ba/F3 (Daley and Baltimore, 1988), transfor- GRB-2 (data not shown). mation of mouse bone marrow cells (McLaughlin et PLCg was also tyrosine-phosphorylated in every al., 1987), and partial induction of neurite extension in transformant but its recruitment was observed only in neuronal PC12 cells (Maru et al., 1996b). Although BCR ± FLTm and BCR ± TRK (Figure 6a). The both BCR ± FLTm and BCR ± TRK abolished IL-3 reported molecular size of PLCg is 145 KD but we dependence in Ba/F3 cells, neither of them could do not suppose that pp150 that appeared in Figure 2b transform mouse bone marrow cells in primary culture is PLCg by two reasons. One is the di€erence in the (Table 1). In the absence of NGF, both BCR ± FLTm apparent molecular size. PY-containing PLCg which and BCR ± TRK induced neurite-like structures that Retroviruses expressing activated forms of VEGF receptor YMaruet al 2590 ac BCR-FLTm MOCK v-ABL BCR-ABL BCR-TRK 12345 KD 205 — v-ABL BCR-FLTm MOCK/TOTAL MOCK MOCK MOCK α BCR-ABL BCR-TRK IP: Shc 12345678 Blot: α PY 49 —

116 — BCR-TRK 32 — GRB-2 80 — 27 —

Blot: Shc α GRB-2 18 — 49 —

b1 2 3 4 5 IP: α ABL α TRK α FLT

205 — IP: α Shc Blot: α BCR

116 — BCR-TRK

80 — Figure 5 Shc phosphorylation and GRB-2 binding. (a) and (b) Anti-Shc immunoprecipitates from Rat1 cells transformed by MOCK (lane 1), BCR ± ABL (lane 2), BCR ± FLTm (lane 3), BCR ± TRK (lane 4), and v-ABL (lane 5) were subjected to anti-PY (a) or anti-BCR (b) Western blotting. Tyrosine-phosphorylated Shc and BCR ± TRK (a), and BCR ± TRK recruited with Shc (b) were indicated by arrows. (c) Anti-ABL (lanes 1 ± 3), anti-TRK (lanes 4, 5), and anti-Flt-1 (lanes 6, 7) immunoprecipitates from Rat1 cells transformed by MOCK (lanes 1, 4, 6), BCR ± ABL (lane 2), v-ABL (lane 3), BCR ± TRK (lane 5), and by BCR ± FLTm (lane 7), and total cell lysates from MOCK (lane 8) were subjected to anti-GRB-2 Western blotting

ab clone 1 clone 2 clone 1 clone 2 MOCK MOCK 123 456 v-ABL BCR-TRK MOCK BCR-FLTm BCR-ABL KD KD 12345 KD 205 — 205 — 205 —

PLC γ 116 — IP: α Flt-1 116 — BCR-TRK Blot: α PY IP: α PLCγ 116 — 80 — 80 — Blot: α PY

12345 80 — 456 α γ 205 — IP: PLC 205 — Blot: α BCR IP: α Flt-1 BCR-FLT Blot: α PLCγ 49 — 116 — BCR-TRK 116 — IP: α PLCγ Blot: α PY

Figure 6 Recruitment of PLCg and pp150. (a) Anti-PLCg immunoprecipitates from Rat1 cells transformed by MOCK (lane 1), BCR ± ABL (lane 2), BCR ± FLTm (lane 3), BCR ± TRK (lane 4), and v-ABL (lane 5) were subjected to anti-PY (upper panel) and anti-BCR (lower) Western blotting. (b) Anti-PLCg (lanes 1 ± 3) and anti-Flt-1 (lanes 4 ± 6) immunoprecipitates from Rat1 cells (MOCK, lane 1) and original Rat1 colonies (clone 1, lane 2; clone 2, lane 3) described in Figure 2 were subjected to anti-PY (lanes 1 ± 3), anti-PY (lanes 4 ± 6, upper panel) and anti-PLCg (lanes 4 ± 6, lower panel) Western blotting Retroviruses expressing activated forms of VEGF receptor YMaruet al 2591 were morphologically similar to those previously 1995), PC12 cells transformed by v-ABL were round or reported in BCR ± ABL (Figure 7b,c,d). BCR ± TRK- spindle-shaped and showed extensions of short transformed PC12 cells showed long neurites with less processes (Figure 7e). prominent branching in the absence of NGF but the Since Rat1 cells transformed by BCR ± FLTm induced morphology was certainly di€erent from that showed capillary-like networks on usual tissue in terminally di€erentiated PC12 cells by NGF culture dishes when starved of serum, we plated the addition (Figure 7c,f). Although activation of MAP cells onto basement membrane matrix (Matrigel) kinase kinase has been shown to be necessary and which is often utilized to study lumen formations sucient for PC12 di€erentiation (Cowley et al., 1994) by endothelial cells (Folkman and Haudenschild, and constitutive activation of MAP kinases has been 1980; Montesano et al., 1983). Interestingly they reported in v-ABL-expressing cells (Raitano et al., formed cell aggregates of 100 ± 200 mm in diameter with radiating cords that were observed neither in the control Rat1 cells (Figure 8a, b) nor in other Rat1 Table 1 Transformation of hematopoietic cells cell transformants shown in Figure 3 (data not No. IL-3-independent Bone marrow cultures shown). Histological sections revealed that Rat1 Retroviral stock cultures Transformed/total cells, whether transformed or not, invaded into MOCK 0/96 0/3 Matrigel. Typical apoptoic features like nuclear BCR±ABL 96/96 3/3 fragmentation and vacuolar degeneration were ob- BCR±FLTm 53/96 0/4 served in the center of aggregations of BCR ± FLTm- BCR±TRK 45/96 0/4 transformed cells (Figure 8c). In contrast both v-ABL 96/96 4/4 surrounding cells that attached to Matrigel and The ability of Ba/F3 cells to be rendered IL-3-independent for cord-forming cells were alive, and showed structural growth following infection by BCR±FLTm and BCR±TRK retro- viruses. Wells were scored positive when they showed outgrowth of continuity. Sections of cords revealed that a few cells IL-3 independent cells after 4 weeks. Bone marrow cells in primary participated in the formations of lumen-like structures culture from BALB/c mice were infected with the retroviruses. with neither Matrigel nor cell debri in the inner space Cultures were scored positive when lymphoid outgrowth was (Figure 8d). Although Rat1 ®broblasts were originally observed after 4 weeks. The number of cultures positive over total derived from embryo (Freeman et al., 1973), we is shown. The retroviral titers were normalized to give roughly equivalent expression following infection to Rat1 ®broblasts (see assume that they have a potential to form lumen-like Figures 3 and 4) structures like those in endothelial cells.

MOCK BCR-FLTm (a) MOCK (d) BCR-FLTm a

b

(b) BCR-ABL (e) v-ABL

c

(c) BCR-TRK (f) NGF ADDITION d

Figure 8 Rat1 cells transformed by BCR ± FLTm show lumen- like structures in Matrigel. Rat1 cells transformed by MOCK (left panel) and BCR ± FLTm (right) were plated onto Matrigel and Figure 7 Morphological changes in transformed PC12 cells. were incubated for 2 weeks. Phase contrast (a) and Nomarski PC12 cells transformed by MOCK (a), BCR ± ABL (b), BCR ± (higher magni®cation than a)(b) microscopic analyses are shown. TRK (c), BCR ± FLTm (d) and by v-ABL (e) and cells treated Hematoxylin and eosin staining of paran sections (c, d). A scale with NGF (f) bar in D indicates 5 mm Retroviruses expressing activated forms of VEGF receptor YMaruet al 2592 when we treated those cells with a blocking antibody Functions of full length Flt-1 with or without mutation against murine VEGF, it could not in¯uence their overexpressed in NIH3T3 cells growth (data not shown). Although we do not know We have previously reported that full length Flt-1 the precise molecular mechanism, we assume that overexpressed in NIH3T3 cells (Flt-1/NIH3T3) could overexpression by itself could also confer transforming transmit biochemical signals upon VEGF stimulation potentials upon Flt-1. Because expression levels of Flt- but not exert growth stimulatory e€ects (Seetharam et 1 correlated with transformed morphologies (data not al., 1994). In order to examine whether or not the gain- shown) and because NIH3T3 cells that expressed of-function mutation that BCR ± FLTm acquired could vector alone did not show transformed phenotypes in¯uence functions of the original Flt-1, Flt-1 with the (Figure 9b). Although colony formation by Flt-1- mutation was reconstructed (Flt-1m) and was ex- expressing cells was VEGF-independent, we observed pressed in NIH3T3 cells (Flt-1m/NIH3T3) at a level VEGF-dependent biological phenotypes when they comparable to that of Flt-1/NIH3T3 reported pre- were plated onto Matrigel. In the absence of VEGF, viously (Seetharam et al., 1994) (Figure 9a, upper NIH3T3 cells that overexpressed wild type Flt-1 panel). VEGF-induced autophosphorylation activities formed a cyst-like structure (Figure 9c, panel 2). It were 10 ± 20% higher in Flt-1m than in Flt-1 (Figure appeared to be morphologically similar to what we 9a, lower panel). Then we expected VEGF-dependent observed in Rat1 ®broblasts transformed by BCR ± colony formation in soft agar in Flt-1m/NIH3T3 cells. FLTm (see Figure 8 for comparison). However, the Surprisingly, however, NIH3T3 cells that overex- central space contained Matrigel and a small number pressed the original Flt-1 showed anchorage-indepen- of cells that were not apoptotic (Figure 9c, panel 4). dent growth in soft agar even in the absence of VEGF Addition of VEGF appeared to abolish the continuity and addition of VEGF in soft agar had almost no of surrounding cell layers and to promote cellular e€ect on the number of colonies (Figure 9b). Flt-1m/ contact with adjacent cyst-like structures (Figure 9c, NIH3T3 cells gave approximately 20% higher colony panel 2). Histological sections revealed that cyst-like numbers than Flt-1/NIH3T3 cells but again VEGF structures were no longer observed in the presence of could not further increase the number of colonies. VEGF and cells appeared to migrate in every direction. Since transformed tumor cells have been reported to No signi®cant di€erences were observed in this frequently produce VEGF, we ®rst thought about morphogenic activity between NIH3T3 cells over- activation of Flt-1 by autocrine mechanism. However, expressing Flt-1 and Flt-1m.

c a VEGF (-) VEGF (+) 1 Flt–1 Flt–1m MOCK VEGF –––++ + 123456 KD 205 —

116 — 2

205 —

116 —

3 b

4

Figure 9 VEGF-induced activities of Flt-1 and Flt-1m. (a) Total cell lysates (upper panel) and anti-Flt-1 immunoprecipitates (lower panel) from NIH3T3 cells expressing MOCK (lanes 1, 2), Flt-1 (lanes 3, 4), or Flt-1m (lanes 5, 6) before (lanes 1, 3, 5) and after (lanes 2, 4, 6) VEGF stimulation were subjected to anti-Flt-1 (upper panel) and anti-phosphotyrosine (lower panel) Western blotting, respectively. (b) Number of soft agar colonies formed by NIH3T3 cells expressing MOCK, Flt-1 or Flt-1m in the presence or absence of VEGF. Means of two experiments are shown. The bottom numbers correspond to those shown in (a). (c) Phase contrast microscopic analysis (panels 1, 2) and histological section (panels 3, 4) of NIH3T3 cells expressing MOCK (panels 1, 3) or Flt-1 (panels 2, 4) 7 days after plating onto Matrigel in the presence (right panel) or absence (left panel) of VEGF Retroviruses expressing activated forms of VEGF receptor YMaruet al 2593 Discussion in cells expressing any one of those three kinases. Biological signi®cance of the dissociation in tyrosine We have shown here that the cytoplasmic portion of phosphorylation between p62Dok and p190 needs to Trk containing the tyrosine kinase domain can be be investigated. Both of them are phosphorylated in v- activated by fusion to the cassette of the BCR ®rst ABL-transformed cells (Ellis et al., 1990). v-ABL exon/the ABL SH3 and SH2 domains (BCR ± TRK), mutants with carboxy-terminal truncations are com- while that of Flt-1 cannot (BCR ± FLT). The Trk promised in the ability to transform lymphoid cells but oncogene was initially identi®ed in a human colon not NIH3T3 ®broblasts. This correlates well with low cancer as a fused oncogene with truncated tropomysin levels of tyrosine phosphorylation in p62Dok (Parmar (Martin-Zanca et al., 1986). Although the Trk can be and Rosenberg, 1996). Although BCR ± TRK and activated in experimental systems by point mutation, BCR ± FLTm could transform Rat1 ®broblasts as in-frame deletion, and duplication of the kinase eciently as v-ABL, they could not transform domain, the tropomysin sequences are presumed to lymphoid cells of mouse bone marrow. Consistent induce dimerization (Coulier et al., 1990). Recently it with the previous report on the inability of v-ABL/Src was shown that the oligomerizing property of the chimera to transform lymphoid cells (Mathey-Prevot transcription factor Tel activates both PDGFR and and Baltimore, 1988), activated receptor kinases Trk ABL in human leukemias with speci®c chromosome and Flt-1 could not substitute for ABL in the context translocations (Golub et al., 1994, 1996). The of BCR ± ABL to transform lymphoid cells. oligomerizing property of BCR ®rst exon-encoded Autophosphorylated BCR ± FLTm and pp150 are sequence can function in a similar fashion. The the two major phosphotyrosine-containing proteins in subsequent random mutagenesis scheme utilizing the BCR ± FLTm-transformed cells (Figure 2b). We highly mutagenic property of retroviruses has enabled assume that one candidate for pp150 is an inositol us to obtain activated forms of BCR ± FLT. Although tetraphosphate and phosphatidylinositol 3,4,5-trispho- this is the ®rst time that the Flt-1 kinase was activated sphate 5-phosphatase (SHIP) which is induced to to exert transforming abilities, activation mechanism in associate with Shc by multiple cytokines (Damen et BCR ± FLTm still remains to be elucidated. The Ser to al., 1996). Anti-PY Western blotting of anti-Shc Thr mutation found in BCR ± FLTm is adjacent to the immunoprecipitates from the transformed Rat1 cells ATP binding GXGXXG motif (Shibuya et al., 1990). showed that a tyrosine-phosphorylated protein around PDGFR, Fms and c-Kit have threonine at this 150 KD associated with Shc (Figure 5). IL-3 depen- position, whereas none of the Flt family members, dence in Ba/F3 cells is abrogated by BCR ± FLTm Flt-1, KDR/Flk-1, and Flt-4 (Pajusola et al., 1992) has (Table 1) and pp150 can also be detected in IL-3- threonine at this position. Retrovirally transduced independent cells that express BCR ± FLTm (data not oncogenic forms of c-Kit or EGFR, v-Kit or v-ErbB, shown). Although recruitment of SHIP to Shc is respectively, have multiple mutations in addition to the induced by cytokines, the SHIP transcripts have been amino-terminal truncation (Nilsen et al., 1985; Qiu et shown to be detected not only in bone marrow but also al., 1988). The Gly to Cys mutation in the kinase insert in many other non-hematopoietic tissues (Damen et al., region derived from clone 2 may induce dimerization 1996). of the kinase domain by a disul®de bond. The kinase While BCR ± ABL has a hematopoietic cell specifi- domains of EGFR and p185c-neu can associate as dimers city for transformation, both BCR ± FLTm and BCR ± to create an ecient signaling assembly and to result in TRK appear to have unique biological features that oncogenic cellular transformation (Murali et al., 1996). may be related to the biology of the original receptor Homodimerization and constitutive activation are also forms. Compared with partial neurite extensions by observed in the erythropoietin receptor with exoplas- BCR ± ABL and BCR ± FLTm in the PC12 system, mic domain mutation converting Arg to Cys BCR ± TRK caused rather long extensions of neurites (Watowich et al., 1992). in the absence of NGF. Rat1 cells transformed by We have found no signi®cant di€erences among BCR ± FLTm not only formed capillary-like network BCR ± ABL, BCR ± FLTm and BCR ± TRK in tyrosine structures in the usual liquid culture system when phosphorylation of Shc, PLCg and p190, and in deprived of serum but also displayed 3-dimensional recruitment of the Ras-signaling adaptor molecule lumen-like structures in basement membrane matrix. GRB-2. However, Rat1 cells transformed by BCR ± This implies the presence of kinase-speci®c substrates TRK or BCR ± FLTm did not contain tyrosine- that are responsible for those biological phenotypes. phosphorylated p62Dok that were coimmunoprecipi- Expression of the receptor Flt-1 is restricted to table with Ras-GAP. We have previously reported endothelial cells and macrophages (Shibuya, 1995; VEGF-induced tyrosine phosphorylation of p62Dok in Barleon et al., 1996). VEGF has been shown to Flt-1-expressing NIH3T3 cells (Seetharam et al., 1994). promote the formation of capillary-like networks in Since p62Dok is highly phosphorylated in BCR ± ABL, endothelial cells (Ferrara and Davis-Smyth, 1997). Our the fusion with BCR sequences by itself may not be data that VEGF-activated Flt-1 could transmit neither responsible for the loss of its phosphorylation. We morphogenic nor growth stimulatory signal in the assume that subcellular localization might explain the usual tissue culture system or in soft agar but could di€erence. The carboxy-terminal tail of ABL is induce cell realignment in Matrigel suggest that Flt-1 involved in actin cytoskeleton-binding (McWhirter might cooperate with signals from extracellular matrix. and Wang, 1993b; van Etten et al., 1994). But However, there has been no reports on the Flt-1- immunostaining of BCR ± FLT or BCR ± TRK in the mediated VEGF activity for the morphogenic behavior 293T overexpression system revealed that they are not of endothelial cells. This may be due to experimental associated with actin micro®laments (data not shown). diculties in endothelial cells. For example, endothelial However, p190 appeared to be tyrosine-phosphorylated cells, whether in primary culture or established as cell Retroviruses expressing activated forms of VEGF receptor YMaruet al 2594 lines, are usually destined to form lumens in was cloned into EcoRI/HindIII sites in pSRaMSVtkneo extracellular matrix. It may be useful if there are (Pendergast et al., 1991). endothelial cells that lack the ability to form lumens. Overexpression of receptor form of Flt-1 or BCR ± Retroviral stock, infection of cells and agar colony assay FLTm in such cells might enable us to test if the Retroviruses were obtained by the transient 293T cell activated forms of Flt-1 kinase is in fact responsible for cotransfection procedure as described (Pendergast et al., the endothelial morphogenesis. 1991). Soft agar colonies derived from cells infected with the BCR ± FLT retrovirus were plated back into liquid culture. By infection with the replication-competent Molony Murine Leukemia Virus BCR ± FLT retroviruses Materials and methods were rescued as described by Goga et al. (1993) and stocked for subsequent experiments. Cells

Rat1 cells and 293T cells were cultured in Dulbecco's Southern blot analysis and PCR analysis modi®ed Eagle medium (DMEM) with a 10% fetal calf serum (FCS). PC12 cells were maintained with DMEM/ Southern blot analysis was performed as described 10% FCS/5% horse serum. Ba/F3 cells were cultured in previously with a 3' Flt-1 fragment (3612 SacI to 4262 RPMI/10% FCS supplemented with 10% conditioned XbaI) as a probe (Shibuya et al., 1990). For reverse medium from actively growing WEHI-3B cells as a source transcriptase-mediated polymerase chain reactions (RT ± of interleukin-3 (IL-3). Transformation assays of Rat1 cells PCR) and sequencing analyses of activated forms of (Lugo and Witte, 1989), PC12 cells (Maru et al., 1996b), BCR ± FLT, the following primers were designed: 2845R mouse bone marrow cells (McLaughlin et al., 1987), and (nucleotide position 2845/reverse primer) (5'-CCT- Ba/F3 cells (Daley and Baltimore, 1988) and transfection CTTTCAGCATTTTCACAGCCA-3'), 3020R (5'-TTGCT- of 293T cells were performed as described previously CTTGAGGTAGTTGGA-3'), 3181R (5'-CAGACTTT- (Maru et al., 1996a). NIH3T3 cells overexpressing Flt-1 TATCTTCCTGAAAGCC-3'), 3450R (5'-AGCCATCC- and their culture conditions were reported previously ATTTCAGAGGAAGTC-3'), 3724R (5'-TTGAAGCA- (Seetharam et al., 1994). Matrigel was purchased from AATCACCTAGTTTTTC-3'), 3900R (5'-GAAAGCATT- Beckton Dickinson and 300 ml was used in 24-wells culture TACATATCTGACATC-3'), 4078R (5'-TGGTTCTAGAC- plates. Approximately 103 to 104 cells were plated onto TAGTCAATCTTGAGCGAGG-3'), 3025N (5'-GACTT- Matrigel. VEGF at 100 ng/ml was added to cell layers in ATTTTTTCTCAACAAGGAT-3'), 3331N (5'-ATTCT soft agar colony assay or Matrigel cultures. TTTATCTGAGAACAA-3'), 3865N (5'-GAAGCTCT- GATGATGTCAGATATG-3'). For sequencing the SH2 and SH3 domains from ABL, primers BCR4N (5'-TCG- Molecular construction TGGGCGTCCGCAAGACCG-3'), ablN (5'-GCTCCG- Construction of BCR ± FLT was performed as follows. GGTCTTAGGCTATAATC-3') and ablR (5'-GACGTA- P¯MI site (2705) of Flt-1 (Shibuya et al., 1990) was GAGCTTGCCATCAGAAG-3') were designed. blunted by Klenow and either 8 bp (GGAATTCC) or 10 bp (CCGAATTCGG) EcoRI linkers were added. The 5' Immunoprecipitation and Western blot analysis BCR ± ABL XbaI/KpnI fragment from pGEM4/p185 BCR ± ABL and the KpnI/EcoRI fragment from Anti-BCR and anti-Flt-1 antibodies were described (Maru pGEX3X-(1021 ± 1120)P210[SH2] (Pendergast et al., 1991) and Witte, 1991; Seetharam et al., 1994). Anti-Trk were ligated. This XbaI/EcoRI fragment was fused with the antibody was kindly provided by Dr S Nakamura at EcoRI/XbaI fragment containing the Flt-1 kinase domain. National Institute of Neuroscience. Anti-phosphotyrosine The fusion is in-frame when 8 bp EcoRI linkers were used antibody (PY20) was purchased from ICN, anti-GRB-2 but out-of-frame with 10 bp linkers. Construction of antibody from Transduction Laboratory, anti-Shc, anti- BCR ± TRK was started with pLM6 kindly provided by GAP and anti-PLCgl antibodies were from Upstate Dr M Barbacid. EcoRI linkers (12 bp CCGGAATT- Biotechnology. Procedures for immunoprecipitation and CCGG) were added to BalI site (1487) (Martin-Zanca et Western blot analysis were described previously (Maru et al., 1989) and EcoRI sites in pUC118 were converted to al., 1996a). Anti-VEGF neutralizing antibody was pur- HindII sites. The EcoRI/HindIII fragment containing the chased from R&D Systems Inc. TrkkinasedomainwasfusedtotheXbaI/EcoRI fragment described above. The resulting BCR ± FLT (XbaI/XbaI fragment) and BCR ± TRK (XbaI/HindIII) sequences were Acknowledgements cloned into pSRaMSVtkneo vector whose EcoRI site was We thank Dr M Barbacid for the Trk cDNA, Dr S converted to XbaI(Maruet al., 1996a). The KpnI/BamHI Nakamura for anti-Trk antibody, This work was supported fragment containing the entire P160v-ABL-coding se- by Grant-in-Aid for Special Project Research on Cancer- quence in pAcC12/v-ABL (Maru and Witte, 1991) was Bioscience 04253204 from the Ministry of Education, cloned into KpnI/BamHI sites in pUC18. Then the v-ABL Science and Culture of Japan and by a research grant sequence was cut out as an EcoRI/HindIII fragment and from The Yakult Foundation.

References

Afar DEH, Goga A, McLaughlin J, Witte ON and Sawyers Clauss M, Weich H, Breier G, Knies U, Rockl W, CL. (1994). Science, 264, 424 ± 426. Walternberger J and Risau W. (1996). J. Biol. Chem., Ahmed A, Dunk C, Kniss D and Wikes M. (1997). Lab. 271, 17629 ± 17634. Invest., 76, 779 ± 791. Coulier F, Kumar R, Ernst M, Klein R, Martin-Zanca D and Barleon B, Sozzani S, Zhou D, Weich HA, Mantovani, A Barbacid M. (1990). Mol. Cell. Biol., 10, 4202 ± 4210. and Marme D. (1996). Blood, 87, 3336 ± 3343. Cowley S, Paterson H, Kemp P and Marshall CJ. (1994). Carpino N, Wisniewski D, Strife A, Marshak D, Kobayashi Cell, 77, 841 ± 852. R, Stillman B and Clarkson B. (1997). Cell, 88, 197 ± 204. Retroviruses expressing activated forms of VEGF receptor YMaruet al 2595 Daley GQ and Baltimore D. (1988). Proc. Natl. Acad. Sci. Murali R, Brennan PJ, Kieber-Emmons T and Greene MI. USA, 85, 9312 ± 9316. (1996). Proc. Natl. Acad. Sci. USA, 93, 6252 ± 6257. Damen JE, Liu L, Rosten P, Humphries RK, Je€erson AB, Nilsen T, Maroney PA, Goodwin RG, Rottman FM, Majerus PW and Krystal G. (1996). Proc. Natl. Acad. Sci. Crittenden LB, Raines MA and Kung H-J. (1995). Cell, USA, 93, 1689 ± 1693. 41, 719 ± 726. De Vries C, Escobedo JA, Ueno H, Houck K, Ferrara N and Obermeier A, Bradshaw RA, Seedorf K, Choidas A, Williams LT. (1992). Science, 255, 989 ± 991. Schlessinger J and Ullrich A. (1994). EMBO J., 13, Ellis C, Moran M, McCormick F and Pawson T. (1990). 1585 ± 1590. Nature, 343, 377 ± 381. Pajusola K, Aprelikova O, Korhonen J, Kaipainen A, Ferrara N and Davis-Smyth T. (1997). Endocrine Rev., 18, Pertovaara L, Alitalo R and Alitalo K. (1992). Cancer 4±25. Res., 52, 5738 ± 5743. Folkman J and Haudenschild C. (1980). Nature, 288, 551 ± Parmar K and Rosenberg N. (1996). J. Virol., 70, 1009 ± 556. 1015. Folkman J and Shing Y. (1992). J. Biol. Chem., 267, 10931 ± Pendergast AM, Muller AJ, Havlik MH, Maru Y and Witte 10934. ON. (1991). Cell, 66, 161 ± 171. Fong G-H, Rossant J, Gertsenstein M and Breitman ML. Pendergast AM, Quilliam LA, Cripe LD, Bassing CH, Dai Z, (1995). Nature, 376, 66 ± 70. Li N, Batzer A, Rabun KM, Der CJ, Schlessinger J and Freeman AE, Gilden RV, Vernon ML, Wolford RG, Gishizky ML. (1993). Cell, 75, 175 ± 185. Hugunin PE and Huebner RJ. (1973). Proc. Natl. Acad. Puil L, Liu J, Gish G, Mbamalu G, Bowtell D, Pelicci PG, Sci. USA, 70, 2415 ± 2419. Arlinghaus R and Pawson T. (1994). EMBO J., 13, 764 ± Goga A, McLaughlin J, Pendergast AM, Parmar K, Muller 773. A, Rosenberg N and Witte ON. (1993). Mol. Cell. Biol., Qiu F, Ray P, Brown K, Barker PE, Jhanwar S, Ruddle FH 13, 4967 ± 4975. and Besmer P. (1988). EMBO J., 7, 1003 ± 1011. Goga A, McLaughlin J, Afar DEH, Sa€ran DC and Witte Ra€el GD, Parmar K, and Rosenberg N. (1996). J. Biol. ON. (1995). Cell, 82, 981 ± 988. Chem., 271, 4640 ± 4645. Golub TR, Barker GF, Lovett M and Gilliland DG. (1994). Raitano AB, Halpern JR, Hambuch TM and Sawyers C. Cell, 77, 307 ± 316. (1995). Proc. Natl. Acad. Sci. USA, 92, 11746 ± 11750. Golub TR, Goga A, Barker GF, Afar DEH, McLaughlin J, Seetharam L, Gotoh N, Maru Y, Neufeld G, Yamaguchi S Bohlander SK, Rowley JD, Witte ON and Gilliland DG. and Shibuya M. (1994). Oncogene, 10, 135 ± 147. (1996). Mol. Cell. Biol., 16, 4107 ± 4116. Settleman J, Narasimhan V, Foster LC and Weinberg RA. Heldin C-H. (1995). Cell, 80, 213 ± 223. (1992). Cell, 69, 539 ± 549. Lugo TG and Witte ON. (1989). Mol. Cell. Biol., 9, 1263 ± Shalaby F, Rossant J, Yamaguchi TP, Gertsenstein M, Wu 1270. X-F, Breitman ML and Schuh AC. (1995). Nature, 376, Maher PA. (1988). Proc.Natl.Acad.Sci.USA,85, 6788 ± 62 ± 66. 6791. Shibuya M, Yamaguchi S, Yamane A, Ikeda T, Tojo A, Martin-Zanca D, Hughes SH and Barbacid M. (1986). Matsushime H and Sato M. (1990). Oncogene, 5, 519 ± 524. Nature, 319, 743 ± 748. Shibuya M. (1995). Adv. in Cancer Res., 67, 281 ± 316. Martin-Zanca D, Oskam R, Mitra G, Copeland T and Stephens RM, Loeb DM, Copeland TD, Pawson T, Greene Barbacid M. (1989). Mol. Cell. Biol., 9, 24 ± 33. LA and Kaplan DR. (1994). Neuron, 12, 691 ± 705. Maru Y and Witte ON,. (1991). Cell, 67, 459 ± 468. Terman BI, Dougher-Vermazen M, Carrion ME, Dimitrov Maru Y, Afar DE, Witte ON and Shibuya M. (1996a). J. D, Armellino DC, Gospodarowicz D and Bohlen P. Biol. Chem., 271, 15353 ± 15357. (1992). Biochem. Biophy. Res. Comm., 187, 1579 ± 1586. Maru Y, Witte ON and Shibuya M. (1996b). Exp. Cell Res., Ullrich A and Schlessinger J. (1990). Cell, 61, 203 ± 212. 229, 438 ± 445. Van Etten RA, Jackson PK, Baltimore D, Sanders MC, Mathey-Prevot B and Baltimore D. (1988). Mol. Cell. Biol., Matsudaira PT and Janmey PA. (1994). J. Cell Biol., 124, 8, 234 ± 240. 325 ± 340. McLaughlin J, Chianese E and Witte ON. (1987). Proc. Natl. Waltenberger J, Claesson-Welsh L, Siegbahn A, Shibuya M Acad. Sci. USA, 84, 6558 ± 6562. and Heldin C-H. (1994). J. Biol. Chem., 269, 26988 ± McWhirter JR, Galasso DL and Wang JYJ. (1993a). Mol. 26995. Cell. Biol., 13, 7587 ± 7595. Watowich SS, Yoshimura A, Longmore GD, Hilton DJ, McWhirter JR and Wang JYJ. (1993b). EMBO J., 12, 1533 ± Yoshimura Y and Lodish H. (1992). Proc. Natl. Acad. Sci. 1546. USA, 89, 2140 ± 2144. Millauer B, Wizigmann-Voos S, Schnurch H, Martinez R, Yamanashi Y and Baltimore D. (1997). Cell, 88, 205 ± 211. Moller NPH, Risau W and Ullrich A. (1993). Cell, 72, Ziche M, Maglione D, Ribatti D, Morbidelli L, Lago CT, 835 ± 846. Battisti M, Paoletti I, Barra A, Tucci M, Parise G, Montesano R, Orci L and Vassalli P. (1983). J. Cell Biol., 97, Vincenti V, Granger HJ, Viglietto G and Persico MG. 1648 ± 1652. (1997). Lab. Invest., 76, 517 ± 531.