Oncogene (2000) 19, 2386 ± 2397 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Dominant negative Met reduces tumorigenicity-metastasis and increases tubule formation in mammary cells

Michal Firon1, Miriam Shaharabany1, Rom T Altstock1, Judith Horev1, Armand Abramovici2, James H Resau3, George F Vande Woude3 and Ilan Tsarfaty*,1

1Department of Human Microbiology, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel, 69978; 2Department of Pathology, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel, 69978; 3ABL-Basic Research Program NCI- Frederick Cancer Research and Development Center, PO Box B, Frederick, Maryland, MD 21702-1201, USA

Activation of the Met tyrosine kinase growth factor Yang et al., 1995). HGF/SF is a paracrine factor, receptor by its ligand HGF/SF has been shown to produced primarily by mesenchymal cells, which increase in vitro invasiveness in epithelial cell lines. To induces mitogenic, motogenic and morphogenic study the e€ect of Met-HGF/SF signaling in breast changes. The diverse biological e€ects of HGF/SF are cancer cells, we transfected met, hgf/sf and dominant all mediated by Met, which is preferentially expressed negative (DN) forms of met into the poorly di€erentiated on epithelial cells (Je€ers et al., 1996a). In vivo, this metastatic murine mammary adenocarcinoma cell line receptor-ligand pair is essential for normal embryolo- DA3. These cells express moderate levels of endogenous gical development (Bladt et al., 1995; Schmidt et al., Met, which is rapidly phosphorylated in response to 1995). Null mutations in both HGF/SF (Schmidt et al., HGF/SF treatment. Met+hgf/sf transfection results in 1995) and Met (Uehara et al., 1995) are embryonic signi®cantly increased tumorigenic and metastatic activ- lethal. ity in vivo accompanied by reduced tubule formation. We have shown that Met-HGF/SF signaling plays DA3 cells transfected with DN forms of Met (DN-DA3) an important role in epithelial tissue morphogenesis exhibit reduced Met phosphorylation following exposure and lumen formation (Tsarfaty et al., 1992, 1994). to HGF/SF. Furthermore, as compared to the parental HGF/SF has been shown to induce tubular branching cells, the DN-DA3 cells exhibit diminished in vitro of mammary cell lines in collagen gels (Berdichevsky et scattering and invasiveness, while in vivo they display al., 1994; Soriano et al., 1995) and to be involved in the greatly reduced tumorigenicity and spontaneous metas- development of mammary tubular structures in vivo tasis. Tumors emanating from DN-DA3 cells injected to (Pepper et al., 1995). While Met-HGF/SF signaling BALB/C mice are highly di€erentiated and display clearly plays a role in normal cellular processes, this extensive tubule formation. These results suggest that signaling pathway has also been implicated in tumor Met-HGF/SF signaling is a determining factor in the development and progression. Met-HGF/SF signaling delicate balance between di€erentiation/tubule formation can increase tumorigenicity (Rong et al., 1992), induce and tumorigenicity-metastasis. Oncogene (2000) 19, cell motility (Stoker et al., 1987), enhance in vitro 2386 ± 2397. invasiveness (Giordano et al., 1993; Je€ers et al., 1996a; Matsumoto et al., 1994; Rong et al., 1994) Keywords: Met; HGF/SF; dominant negative; tubulo- and in vivo metastasis (Je€ers et al., 1996b; Rong et al., genesis; breast cancer cells 1994; Rosen et al., 1994). In addition, Met-HGF/SF signaling can increase production of proteases (Rong et al., 1994) and urokinase (Je€ers et al., 1996b) which Introduction are associated with extracellular matrix/basal mem- brane degradation important for metastasis. HGF/SF Breast cancer is a multi-step process, which proceeds is a potent angiogenic factor (Bussolino et al., 1992; from pre-neoplastic disease through invasive cancer, Rosen et al., 1993) that induces blood vessel formation lymphatic spread and metastasis (Benz et al., 1995). A in tumors originating from injection of human breast number of growth factors and their receptors have cancer cells into the mammary fat pads of nude mice been implicated in breast tumor formation and (Lamszus et al., 1997). Furthermore, HGF/SF trans- progression. Their involvement in mammary carcino- genic mice develop a broad array of histologically genesis has been associated with alteration in their distinct tumors of both mesenchymal and epithelial number, structure, or sub-cellular location (Cunha, origin (Takayama et al., 1997). Activating mutations in 1994). the tyrosine kinase domain of Met have been identi®ed Recently, attention has focused on the role of Met in both hereditary and sporadic forms of human tyrosine kinase growth factor receptor and its ligand papillary renal carcinoma (Schmidt et al., 1997). hepatocyte growth factor/scatter factor (HGF/SF) in Constitutive activation of growth factor receptors mammary development and carcinogenesis (Pepper et through autocrine ligand stimulation frequently occurs al., 1995; Tsarfaty et al., 1992; Yamashita et al., 1994; in human cancers (Ethier, 1995). We previously demon- strated that when Met and its ligand are co-expressed in a variety of mouse and human cells, the cells become *Correspondence: I Tsarfaty highly tumorigenic in nude mice through an autocrine Received 5 June 1998; revised 24 February 2000; accepted 1 March mechanism (Cortner et al., 1995; Rong et al., 1992, 2000 1995). Additional reports describe Met-HGF/SF auto- Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2387 crine loops in murine mammary carcinoma cells inducing signaling can alternate the fate of the cells between in vitro invasiveness (Je€ers et al., 1996a; Rahimi et al., cell di€erentiation (tubulogenesis) and invasive-meta- 1996). NIH3T3 cells co-transfected with met and hgf/sf static phenotype. form tumors in nude mice. These mesenchyme-derived tumor cells and explants acquire lumen-like morphology in vitro and in vivo. Furthermore, histopathological Results examination of paran-embedded sections of tumor derived from these cells shows that in addition to Expression of HGF/SF, Met and DN-Met in DA3 cells lumenal structures, they contain carcinoma-like focal areas (Tsarfaty et al., 1994). Met expression in DA3 cells was determined by Met-HGF/SF signaling has been implicated in a wide Western blot (WB) analysis with SP260 rabbit anti- range of cancers including breast cancer (Pepper et al., peptide antibody. Moderate levels of p140met were 1995; Tsarfaty et al., 1992; Yamashita et al., 1994; Yang detected (Figure 1A, lane 1). This band was not et al., 1995), but its exact role is not clearly understood. evident in the presence of SP260 immunizing peptide Loss of heterozygosity (LOH) for the c-met locus in (Figure 1A, lane 2), con®rming the speci®city of the breast cancer patients was described (Bieche et al., 1992; antibody. The in¯uence of HGF/SF on Met phosphor- Deng et al., 1994) and has been shown to correlate with a ylation was determined by immunoprecipitation (IP) higher risk of relapse, metastatic disease and reduced using SP260 followed by WB analysis using anti- overall survival (Bieche et al., 1992). We have shown that phosphotyrosine antibody (anti-pTyr). Low levels of Met is expressed and activated in normal breast ducts phosphorylated Met were detected in the untreated and reduced in some breast tumors (Tsarfaty et al., DA3 cells (Figure 1A, lane 3). A 5-minute exposure to 1992). On the other hand, Met expression levels were HGF/SF increased (410 fold) Met phosphorylation shown to be signi®cantly greater for patients with breast (Figure 1A, lane 4). Thus, Met is present and is rapidly cancer than in benign breast disease (Nagy et al., 1995). phosphorylated in response to HGF/SF treatment. Transgenic expression of tpr-met led to development of In order to investigate the e€ects of Met-HGF/SF mammary hyperplasia and tumors (Liang et al., 1996). signal transduction modulation on mammary cells, Other studies show elevated HGF/SF level is a strong DA3 cells were transfected with human hgf/sf (DA3H), and independent predictor of recurrence and reduced murine met (DA3M), or both human hgf/sf and murine survival in human breast cancer (Yamashita et al., 1993, met (DA3MH). DA3 cells transfected with expression 1994). Invasive breast cancers have signi®cantly higher vector containing only neo (DA3N) served as control. HGF/SF levels as compared with ductal carcinoma in In addition to the clones isolated, we pooled at least 50 situ, and invasive cancers that had spread to axillary resistant colonies (clones) of DA3 cells transfected with lymph nodes exhibited signi®cantly higher HGF/SF hgf/sf (PL-DA3H) or with met+hgf/sf (PL-DA3MH) content than did invasive cancers without regional in order to avoid misleading conclusions resulting from spread (Yao et al., 1996). We have recently shown that clonal variation. We evaluated HGF/SF expression either increased or decreased Met expression in breast levels in the supernatants of the transfected cells by tumors indicates poor prognosis (Tsarfaty et al., 1999). heparin-sepharose precipitation and WB analysis. Receptor dimerization has been proposed to be Expression of HGF/SF was not detected in DA3N necessary for receptor tyrosine kinase activation (Yard- supernatant (Figure 1Bb, lane 1), while PL-DA3H and en and Schlessinger, 1987). Receptor dimerization is PL-DA3MH supernatants exhibited high levels of followed by intramolecular transphosphorylation of the HGF/SF (Figure 1Bb, lanes 2 and 3, respectively) with receptor sub-units, which initiates intracellular signal HMH cell supernatant used as positive control (Figure transduction. In vivo, expression of certain mutated 1Bb, lane 4). Immuno¯uorescence (IF) analysis also receptors, referred to as dominant-negative receptors, showed no expression of HGF/SF in DA3N cells can disrupt receptor signaling by hetero-dimerization (Figure 1Ba, 1), while DA3H cells exhibited high levels with the wild type receptors (Amaya et al., 1991) or by of HGF/SF (Figure 1Ba, 2). DA3MH cells displayed sequestering ligand (Robinson et al., 1995). In either the highest HGF/SF expression levels of all the case, the consequence is a signi®cant reduction in transfected clones (Figure 1Ba, 3). ligand-receptor association-induced activity. We transfected DA3 cells with forms of Met that To further elucidate the contribution of Met-HGF/ were partially deleted at the tyrosine kinase (TK) SF signaling to breast cancer, we examined the e€ect of domain (pMBTDN-TDN cells) or contained substitu- both up- and down-regulation of Met expression and tion of three tyrosine residues (Y1347, Y1354 and activity on the tumorigenic and invasive potential of a Y1363) with phenylalanine in the docking site mammary cell line, using the DA3 cell line as a model (pMFDN-F1DN cells) as potential dominant negative system. Up-regulation of Met activity was accom- forms of Met. Mutation of these docking site tyrosines plished by co-transfection with expression vectors for was previously demonstrated to abrogate Met activity Met and HGF/SF. The down-regulation of Met (Ponzetto et al., 1994; Weidner et al., 1995). In activity was achieved by constructing two dominant- addition to the selected individual clones transfected negative forms of Met (DN-Met): One was truncated with pMFDN, we pooled at least 50 neo-resistant at the kinase domain; the other produced by the clones (PL-FDN) in order to avoid misleading substitution of three tyrosines in the docking site conclusion resulting from clonal variation. We also (F1DN). We found that a reduction in Met signaling in constructed an additional TDN-Met expression plas- DA3 cells (via DN-Met) induced in vitro and in vivo mid with a in¯uenza hemagglutinin (HA) tag (pTDN- tubulogenesis, whereas an elevation in Met signaling HA). DA3 cells were transfected with pTDN-HA and induced enhanced motility, tumorigenicity and metas- at least 50 neo-resistant clones were pooled (PL-TDN- tasis. Our results demonstrate that Met-HGF/SF HA). We also tested several TDN-HA clones.

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2388

Figure 1 (A) Met expression and activation in DA3 cells. DA3 cell lysates were IP using SP260 antibody (1) or SP260 antibody competed with the SP260 immunizing peptide (2) and WB with SP260 antibody. Untreated DA3 cells (3) or DA3 cells treated with HGF/SF (80 u/ml) for 5 min (4) were lysed in the presence of sodium orthovanadate, cell extracts were IP with SP260 antibody and immunoblotted with anti-phosphotyrosine antibody (4G10). Met is expressed in DA3N cells and is rapidly phosphorylated in response to HGF/SF treatment. (B) (a) IF detection of HGF/SF expression in DA3 transfectants. DA3N (1), DA3H (2) and DA3MH (3) cells were stained with rabbit anti recombinant human HGF/SF antibody, labeled with donkey anti rabbit antibody conjugated to FITC and analysed using CLSM. HGF/SF is not detected in DA3N cells while DA3H cells express moderate levels of HGF/SF, and DA3MH cells express greater levels (magni®cation 6260). (b) WB of HGF/SF expression in DA3 transfectants. Supernatants from DA3N (1), PL-DA3H (2), PL-DA3MH (3) and HMH (4) cells were incubated with heparin-sepharose (HS). Following incubation pellets were suspended in sample bu€er, resolved on 12% SDS ± PAGE and analysed by Western blot analysis using rabbit anti HGF/SF antibody diluted 1 : 350. HGF/SF is secreted from PL-DA3H and PL-DA3MH cells. (C) Analysis of truncated dominant negative met mRNA transcription. Ethidium bromide stained agarose gel of 3' met RT ± PCR on RNA extracts from DA3N cells (1) and TDN cells (2); PCR product from 3' met primers on full-length met cDNA plasmid pMB11 (3) and pMBTDN plasmid (4) served as controls. A band of 1004 bp, corresponding to WT met, appears in RNA extracts of DA3N cells, while RNA extracts of TDN cells also produced a band of 352 bp, indicating the expression of TDN mRNA. (D) Expression of truncated and docking-site tyrosine-mutated dominant negative Met. IP and WB using anti Met (SP260) antibody was preformed on cell lysates of 293 (1), 293-Met (2), 293-TDN (3), 293-FDN (4), DA3N (5) and TDN (6). 293-TDN and TDN cells exhibit a 120- kD band, corresponding to the truncated form of Met, 293-TDN cells also display a 140 kD precursor band and TDN cells also express the 170- and 140 kD bands of endogenous Met. 293-FDN cells express high levels of the docking-site tyrosine-mutated Met

We proceeded to verify the transfections. TDN-Met transiently transfected with pMBTDN (293-TDN) or (pMBTDN) transcription in TDN cells was determined pMB11 (293-Met). These cells were subjected to IP and by RT ± PCR. The transcribed pMBTDN DNA is WB analysis using SP260 anti Met Ab. A band of 651 bp shorter than the wild-type cDNA. RT ± PCR of approximately 120 kDa appears only in the 293-TDN DA3N yielded the expected fragment of 1004 bp cells (Figure 1D, lane 3) but not in 293 cells (Figure (Figure 1C, lane 1) comparable to the PCR product 1D, lane 1) or in 293-Met cells (Figure 1D, lane 2). IP obtained with wild-type met cDNA plasmid (pMB11) and WB analysis using SP260 anti Met Ab shows that (Figure 1C, lane 3). RT ± PCR of TDN cells produced a corresponding 120 kD band appears in TDN cells three fragments (Figure 1C, lane 2): one of 1004 bp, (Figure 1D, lane 6) but not in DA3 cells (Figure 1D, corresponding to the endogenous Met, another of lane 5). Hence, the truncated Met is expressed at both 352 bp, indicating the deletion, comparable to the PCR the mRNA and protein levels. Truncated Met expres- product obtained with DN met plasmid (pMBTDN) sion in the TDN-HA clones was examined by IP of the (Figure 1C, lane 4), and a third of *700 bp, which is cell lysate with SP260 anti Met Ab and subsequent WB currently under investigation. with monoclonal anti HA Ab. A band of approxi- To verify truncated Met expression, 293 human mately 120 kD appears only in the TDN-HA clones embryonic kidney cells, which do not express endo- (Figure 2C, lanes 3 ± 10) but not in DA3N cells (Figure genous Met detectable with the SP260 antibody, were 2C, lanes 1 and 2) verifying the expression of truncated

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2389 HA labeled Met. Expression of wild-type Met was Met which remained high at 10 and 20 min exposures similar in all TDN-HA clones as demonstrated by IP (Figure 2Ba). After 60 min exposure the levels decreased of the cell lysate with SP260 anti Met Ab and by about 50% (results not shown). Although DA3MH subsequent WB with SP260 (Figure 2E). cells exhibit a slightly higher basal level of phosphorylated Expression of tyrosine-mutated Met (FDN) was Met, 5 min treatment with HGF/SF still invoked a more dicult to verify since the mutated Met is marked increase (5 ± 10-fold) in Met phosphorylation indistinguishable from wild-type Met by WB. However, (Figure 2Bb). TDN (Figure 2Bc) and F1DN (Figure 2Bd) the construct was transiently transfected into 293 cells cells exhibit dramatically reduced HGF/SF-induced Met and its expression was examined by IP and WB with phosphorylation compared to DA3N cells. These results SP260. The transfected 293-FDN cells exhibited high demonstrate the dominant negative e€ect of the truncated mutated Met levels (Figure 1D, lane 4) as opposed to and tyrosine docking site mutated Met. Similar results the parental 293 cells (Figure 1D, lane 1). Hence, the were obtained with the TDN-HA clones, which exhibited FDN protein is expressed in cells. greatly reduced Met phosphorylation following HGF/ SF-treatment (Figure 2D, lanes 3 ± 10) compared to DA3N cells (Figure 2D, lanes 1 and 2). HGF/SF-induced Met phosphorylation in transfectants We proceeded to determine the e€ect of the various Proliferation of DA3 and transfectants following Met transfections on Met phosphorylation in DA3 cells activation exposed to HGF/SF. DA3N, DA3MH and the DA3 cells transfected with the DN forms of met, were We examined cell proliferation of DA3 and the various treated with HGF/SF for various times. Subsequently, transfectants. Cells (2500/well) were incubated with or the cell extracts were examined for Met expression without HGF/SF (80 u/ml) for 96 h. Proliferation was (Figure 2A) and phosphorylation (Figure 2B). determined using MTT assay (Figure 3A). Phosphorylated Met was not detected in untreated Exposure to HGF/SF had no e€ect on the DA3N cells (Figure 2Ba). DA3N cells exposed for 5 min proliferation rate of the various transfectants, with to HGF/SF showed a marked increase in phosphorylated the exception of DA3M, which displayed a 31%

Figure 2 Met expression and phosphorylation in DA3 and transfectants. DA3N (a), DA3MH (b), TDN (c) and F1DN cells (d) were incubated with 80 u/ml of HGF/SF for 0, 5, 10 and 20 min, lysed in the presence of sodium orthovanadate, Cell lysates were: (A) IP and WB using anti Met (SP260) antibody. (B) IP with SP260 antibody and WB using anti-phosphotyrosine antibody (4G10). Met is expressed in DA3N and transfected cells. DA3MH cells exhibited increased basal Met phosphorylation, while expression of either type of DN-Met dramatically reduced HGF/SF-induced Met phosphorylation. Expression and phosphorylation of truncated Met in TDN-HA clones. DA3N (1, 2), DHA1 (3, 4), DHA2 (5, 6), DHA4 (7, 8) and DHA5 (9, 10) cells, untreated (1, 3, 5, 7, 9) or incubated with 80 u/ml of HGF/SF for 10 min (2, 4, 6, 8, 10) were lysed in the presence of sodium orthovanadate. Cell lysates were: (C) IP with SP260 antibody and WB using anti-HA antibody. TDN-HA clones express HA-tagged truncated Met while DA3N cells do not. (D) IP with SP260 antibody and WB using anti-phosphotyrosine antibody (4G10). Met phosphorylation is reduced in TDN- HA-transfected cells compared with DA3N cells. (E) IP and WB with SP260

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2390 reduction in proliferation rate following HGF/SF model for HGF/SF-induced cell motility) exhibiting treatment (P50.00008, t-test). When compared to disruption and scattering of epithelial colonies. HGF/ untreated DA3N, an increased growth rate was SF also induced scattering of DA3M, DA3H and observed for untreated TDN and F1DN cells of 49% DA3MH cells (data not shown) similar to that of the (P50.001) and 86% (P50.0001), respectively. In order parental DA3 cells. However, negligible scattering was to further assess the e€ect of the reduction of Met observed in HGF/SF-treated PL-TDN-HA and PL- activity compared to normal/high Met activity, we FDN cells (Figure 3B, 4, 6 respectively) compared to divided the di€erent transfections into two groups: untreated cells (Figure 3B, 3, 5 respectively). Similarly, Met+ (DA3N, DA3M, DA3H and DA3MH cells) and no signi®cant scattering reaction was observed in DA3-DN (TDN and F1DN cells). DA3-DN cells, HGF/SF-treated TDN and F1DN cells (data not without HGF/SF treatment, displayed a signi®cantly shown). increased proliferation rate as compared with untreated Invasiveness of DA3 and transfectants was evaluated Met+ cells (P50.00001, t-test). by Boyden chamber basement membrane matrigel assay (Albini et al., 1987) (Figure 3C). HGF/SF- treated DA3 cells migrated through the matrigel-coated Scattering and in vitro invasiveness of DA3 and ®lter (Figure 3C, 1, columns 1 ± 3), while no sponta- transfectants neous migration was detected in the absence of HGF/ Motility and invasiveness are characteristics of malig- SF. HGF/SF-treated DA3MH cells e€ectively migrated nant cells. DA3 treated with HGF/SF (80 u/ml) through the matrigel-coated ®lter (data not shown). display ®broblast-like `scattered' morphology (Figure HGF/SF-induced invasiveness of TDN and F1DN cells 3B, 2) as compared to appearance of untreated cells was signi®cantly reduced in comparison with parental (Figure 3B, 1). This change in morphology is similar DA3 (Figure 3C, 1, columns 4 ± 6 and 7 ± 9, respec- to that of HGF/SF-treated MDCK cells (the classical tively).

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2391 Quanti®cation of migration revealed that TDN cells compared to the parental cell line, as determined by exhibit a 56% (P50.0001) and F1DN cells a 88% positive area (see Materials and methods) (Figure 3C, (P50.000004) reduction of in vitro invasiveness 2). These results further substantiate that the DN

Figure 3 (A) Proliferation rate of DA3 and transfectants. DA3N and transfectants were seeded into 96-well microtiter plates and incubated with (vertical line bars) or without HGF/SF (80 u/ml) (gray bars) for 96 h. Graph shows OD results of MTT analysis. Proliferation rate of TDN and F1DN cells is signi®cantly increased compared to DA3N cells. (B) Scattering of DA3 and transfectants. DA3 (1, 2), PL-TDN-HA (3, 4) and PL-FDN (5, 6) cells were seeded in 96-well microtiter plates and incubated for 24 h without (1, 3, 5) or with 80 u/ml of HGF/SF (2, 4, 6). Cells were ®xed, stained with Giemsa and examined by inverted microscope (magni®cation6230). (C) In vitro invasiveness of DA3 and transfectants. 1. DA3N (1 ± 3); TDN (4 ± 6) and F1DN (7 ± 9) cells were submitted in triplicate to Boyden chamber assay using basement membrane matrigel coated ®lter. HGF/SF (80 u/ml) was added (+) or not added (7) to the lower wells. Following 24 h incubation, the membrane was removed and its upper surface wiped. Filter was ®xed and stained with Giemsa. HGF/SF-induced invasiveness of TDN and F1DN cells is signi®cantly reduced in comparison with DA3N cells. 2. The stained ®lter was scanned and positive area quantitized using image analysis software as described in Materials and methods. The averaged positive area and standard deviation are shown, substantiating the reduced invasiveness of TDN and F1DN cells. (D) In vitro tubule formation by DA3 and transfectants. 26105 DA3N (1, 2), TDN (3, 4), F1DN (5, 6) and MDCK cells (7, 8) were seeded in collagen gels. HGF/SF (80 u/ml) was added (2, 4, 6, 8) or not added (1, 3, 5, 7) to wells. Following 5 days incubation, collagen gels were ®xed and imaged by Nomarski, di€erential interference contrast microscopy (N/DIC). HGF/SF induces tubule formation in TDN and F1DN cells, but not in DA3N cells (magni®cation6170)

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2392

Figure 4 (A) Tumorigenic potential of DA3 and transfectants. Ten mice were intramuscularly injected with 56104 cells of each clone and tumor formation was monitored in the di€erent groups (^ DA3N, & DA3M, ~ DA3H, * DA3MH, X TDN, * F1DN). Tumorigenicity of TDN and F1DN cells is signi®cantly reduced compared to DA3N cells, and signi®cantly increased in DA3M, DA3H and DA3MH cells. (B) Tumor sections from mice injected with DA3 transfectants. Photomicrograph of H&E stained sections prepared from paran blocks of primary tumors excised from mice injected with: DA3N (1); DA3MH (2); TDN (3); F1DN (4) cells. (Magni®cation6190). (C) Metastasis formation in mice injected with DA3 and transfectants. Photomicrograph of H&E-stained sections prepared from paran blocks of lungs from mice injected with DA3 transfectants. Mice were intramuscularly injected with 56104 DA3N (1), DA3M (2), DA3H (3), DA3MH (4), TDN (5) and F1DN (6) cells 56 days prior to being sacri®ced. (Magni®cation6130). (D) Survival analysis of mice injected with DA3 and transfectants. Ten mice were intramuscularly injected with 56104 cells for each clone. Survival analysis for the di€erent groups were calculated (^ DA3N, & DA3M, ~ DA3H, * DA3MH, X TDN, * F1DN). Life spans were signi®cantly shorter in DA3M- and DA3MH-injected mice and signi®cantly longer in TDN- and F1DN-injected mice

forms of Met abrogate Met signal transduction, presence of HGF/SF, branching morphogenesis leading to reduced motility and invasiveness. occurs in TDN and F1DN cells, which leads to outgrowth of long extensions (Figure 3D, 4 and 6, respectively) similar to treated MDCK cells (Figure In vitro tubule formation by DA3 and transfectants 3D, 8). Using CLSM we observed that the branches Tubule formation is a major characteristic of contained lumens (data not shown). However, HGF/ di€erentiated mammary cells. Tubule formation in SF treatment did not induce tubule formation in collagen was assessed for the di€erent transfectants DA3N cells (Figure 3D, 2) and DA3MH (data not following HGF/SF treatment. Untreated DA3N, shown). These results indicate that reduction of Met TDN and F1DN cells grow as cysts in the collagen activity enables DA3 cells to form branching tubular cultures (Figure 3D, 1, 3 and 5, respectively). In the structures.

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2393 Tumor formation by DA3 and transfectants occupying an average of 8 ± 9% of the lung area. DA3MH lung sections also contained very large foci. In vivo tumorigenicity of DA3 cells and transfectants The metastatic cells surrounded the alveolar space in was determined by intra-muscular injection of the solid clumps, with no tubular structure (Figure 4C, 4) various transfected cells into BALB/C mice. DA3 and with metastases occupying an average of 25 ± 27% of transfectants were each injected (56104 cells) to 10 the lung area. In contrast, TDN-injected mice lung mice, which were subsequently monitored twice weekly sections contained few small foci. The metastatic tissue for tumor formation. One month follow-up revealed: was mainly comprised of stroma, with indication of palpable tumors developed in 80% (eight of 10) of capsule formation (Figure 4C, 5) with metastases mice injected with DA3N and of DA3M-injected mice, occupying an average of 0.1% of the lung area. In 100% (10 out of 10) of DA3H-injected mice and of F1DN lung sections, a single carcinoma-like focus was DA3MH-injected mice, but in only 40% (four of 10) of observed. The focus, which displayed little tubular TDN-injected mice and 20% (two of 10) of F1DN- di€erentiation, was located in the lung periphery injected mice. We analysed the rate of tumor formation (Figure 4C, 6) with metastases occupying an average and overall survival for statistical signi®cance using of 0.1% of the lung area. Kaplan ± Meier plots (Figure 4A). This analysis shows Additional six groups of 10 mice were injected as that DA3H and DA3M cells exhibit a statistically previously described. The mice were monitored over signi®cant increase in tumor formation as compared a period of 85 days. DA3M and DA3MH-injected with DA3N cells; the highest rate of tumor formation mice had the shortest life spans, while the survival was observed in mice injected with DA3MH cells. A rates for TDN and F1DN-injected mice were signi®cant reduction in tumor formation was found in remarkably longer. Results were analysed using mice injected with TDN cells, and the lowest Kaplan ± Meier statistical analysis and were deter- tumorigenicity was observed in F1DN-injected mice. mined signi®cantly di€erent (P=0.0001 Log-rank) Log rank analysis shows a statistically signi®cant (Figure 4D). di€erence between the di€erent groups (P=0.0001). Fifty-six days after injection, eight mice of each group were sacri®ced, tumors were excised and Discussion histopathological examination was carried out on H&E-stained tumor sections. Tumors in mice injected Loss of normal tubular and glandular structures is one with DA3N cells were pleomorphic and irregular in of the primary characteristics of breast tumors. Rather size and shape. These tumors exhibited numerous than forming organized tubular structures, tumor cells mitotic ®gures, few micro-lumena and evidence of proliferate and ®ll ducts and ductules. A poor degree increased vascularity (Figure 4B, 1). DA3MH tumor of di€erentiation or tubular organization correlates cells were the least di€erentiated, arranged in sheets with a poor prognosis in invasive ductal carcinoma and fragments with less columnar di€erentiation (Clark, 1996). Understanding the molecular mechan- (Figure 4B, 2). However, the tumors in mice injected isms underlying tubule formation and its abrogation with TDN cells were highly di€erentiated and formed during carcinogenesis/tumor progression may shed numerous lumenal as well as glandular structures light on the cellular processes leading to breast cancer. (Figure 4B, 3). F1DN-injected mice developed tumors Others and we have shown that activation of Met that were even more di€erentiated, containing mature signal transduction induces lumen/tubule formation. columnar structures with increased stroma between the However, up-regulation of the same signal transduction glandular elements and larger columnar cells (Figure also induces invasiveness and metastasis. In order to 4B, 4). examine the role of Met in the balance between The lungs of the sacri®ced mice were evaluated for tumorigenesis/metastasis and di€erentiation/tubulogen- spontaneous metastasis formation by weight and esis, we created a model to investigate the e€ects of morphological/histopathological inspection. The weight Met signaling modulation on mammary cells, both in of lungs from DA3MH-injected mice was 48% vivo and in vitro. Our model system is based on the (0.32+0.13 g) and DA3M-injected mice was 19% carcinogen-induced mouse mammary cell line DA3. (0.25+0.04 g) greater than those of DA3N-injected These cells are highly tumorigenic and metastatic, mice (0.21+0.03 g). Lung weight of mice injected with poorly di€erentiated and express moderate levels of TDN and F1DN cells was 13% (0.19+0.04 g) and 8% endogenous Met. Up-regulation of Met signal trans- (0.20+0.04 g) less than DA3N-injected mice, respec- duction was achieved by co-transfection of met and tively. Lung weight of DA3H-injected mice was similar hgsf/sf into DA3 cells; down-regulation of the signaling to that of DA3N-injected mice (0.21+0.05 g). These was achieved by transfection with DN forms of the results were substantiated by the examination of H&E Met receptor. stained histological sections of the lungs from ®ve mice The DN receptor is a powerful tool in the study of of each group. Examination of DA3N-injected mice receptor action and signal transduction (Ullrich and lung sections revealed few foci of moderately di€er- Schlessinger, 1990). It has been demonstrated that TK entiated tubular carcinoma (Figure 4C, 1) with truncated receptors such as the insulin receptor, EGF metastases occupying an average of 1 ± 2% of the lung receptor (Qian et al., 1994) and FGF receptor (Amaya area. In DA3M-injected mice sections the normal lung et al., 1991) can act as DN and suppress the function structure was largely destroyed by the metastatic tissue of WT receptors. We constructed two DN forms of (Figure 4C, 2) with metastases occupying an average of Met: One consists of a deletion in the TK domain; the 5 ± 7% of the lung area. Many foci, some very large, other has three mutated tyrosine residues (Y1347, were observed in DA3H lung sections; the tubular Y1354 and Y1363) in the docking site. Both forms of structure was minimal (Figure 4C, 3) with metastases DN-Met were found to reduce the in vitro motility and

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2394 invasiveness, as well as in vivo tumorigenic and cancer cells. Our results, which demonstrate the metastatic potential of the transfected DA3 cells. signi®cance of Met in the invasive process, are The molecular mechanism of the DN e€ect caused corroborated by other publications showing that most by the truncated TK-Met receptor is not entirely clear. lymph node metastases have higher Met expression Since the level of TDN Met receptor expression is not than primary tumors (Beviglia et al., 1997). In other high, it is less conceivable that the DN e€ect is due to cancers it was shown that distant metastases express competition for the ligand. It is more conceivable that high levels of Met compared to the primary tumor, dimerization of the truncated receptor with the WT suggesting that the presence of Met receptor may receptor is required for DN action. These results contribute to the acquisition of an invasive phenotype indicate that receptor dimerization is essential for WT (Di Renzo et al., 1994, 1995; Natali et al., 1993). Met activity. This is substantiated by previous work, Tumors resulting from the DA3-DN cells were highly which demonstrates that a leucine zipper-mediated di€erentiated and contained glandular structures, while dimerization of the tpr-met product is essential for injection of DA3 with increased Met signaling resulted the transforming activity of the met oncogene (Ro- in undi€erentiated tumors with fewer and smaller drigues and Park, 1993). glandular structures. These results imply that loss of To our knowledge, these results are the ®rst to tubule formation may be reversible via alteration of demonstrate that a receptor that harbors mutations in Met signal transduction. the docking site can serve as a DN molecule. Phosphor- Although we show down-regulation of Met phos- ylation of tyrosines in the docking site is a crucial step in phorylation by the DN-Met, the molecular mechanisms tyrosine receptor signal transduction. Mutations in these underlying the DN mediated biological e€ects are as tyrosines abrogate substrate association with the recep- yet unclear. These biological e€ects may be caused tor, blocking downstream events and rendering the solely by the DN-induced reduction in Met phosphor- receptor inactive. It has been shown that mutations in ylation; another possibility is that the DN form Met Y1349 and Y1356 can reduce or abrogate the associated with the WT receptor activates alternate binding of Met substrates PLCg, GRB-2 and pp60c-src to signal transduction pathways responsible for tubule Met and reduce Tpr-Met transforming ability (Ponzetto formation. Possibly, the di€erentiative signaling path- et al., 1994). Moreover, mutation of Met tyrosine ways induce the expression of di€erent proteins, for residues Y1311, Y1347, Y1354 and Y1363 reduced or example; cell ± cell and cell ± matrix proteins in DN ± abolished ligand-induced cell motility and branching Met cells, while met+hgf/sf-transfected cells express morphogenesis (Weidner et al., 1995). However, it is not proteases. Recently it was shown that tubulogenesis is entirely clear how the docking site tyrosine-mutated dependent on the STAT pathway: Met binds and receptor serves as a DN, since the TK region is intact and phosphorylates STAT which subsequently translocates thus theoretically it is still capable of cross-phosphor- to the nucleus, binds to the speci®c promoter element ylating the endogenous WT receptor. As suggested for SIE and modulates the expression of genes which lead the TK-mutated DN, the docking site-mutated DN to di€erentiation, including c-fos and waf-1 (Boccaccio activity can be explained by competition with the WT et al., 1998). Thus, a possible explanation for our receptor for the ligand or by dimerization with the WT results could be that an up-regulation in Met signaling receptor, which might sterically interfere with substrate results in increased signaling in the Ras-Src pathway, binding to the WT receptor counterpart, thereby while reduction in Met signaling via the DN receptor reducing or altering signal transduction. induces signaling through the STAT pathway. Increased Met activation via met and/or met+hgf/sf We have recently shown that breast cancer patients transfection into DA3 cells resulted in increased tumor with either up-regulation or down-regulation of Met and metastasis formation, as well as increased expression in the tumor have poor prognosis. We mortality rate of mice injected with these transfected hypothesize that while cells with high Met expression cells, compared to mice injected with control cells. have a greater invasive and metastatic potential, Met These observations are in concordance with previous can act adversely at low levels by not allowing cells to work showing that cells co-expressing HGF/SF and fully di€erentiate into tubular structures (Tsarfaty et Met display increased invasiveness in vitro (Je€ers et al., AQCH in press). The model system presented in al., 1996a). Co-expression of Met and HGF/SF is this paper indicates that Met activity can change the abundant in invasive breast carcinoma as compared fate of the cells and explain the dual function of Met in with benign breast tissue (Jin et al., 1997). Moreover, it breast cancer. This hypothesis is substantiated by was shown that Met is up-regulated in breast cancer recent work on branching in the Drosophila tracheal biopsies and that Met as well as HGF/SF levels can system. Regulation of branching in this system is serve as prognostic markers for breast cancer (Ghous- achieved by modulation of signal transduction levels of soub et al., 1998). the FGF-receptor homolog breathless (Btl), expressed DA3-DN cells display dramatically reduced HGF/ on tracheal cells (Hacohen et al., 1998). SF-dependent motility (scattering) and invasiveness in In our system, up-regulated Met signaling leads to vitro as compared to the parental cells and met+hgf/sf increased tumorigenesis/metastasis, while down-regu- transfected cells. Furthermore, in vivo we observed lating the signal via the DN receptor results in dramatic reduction in tumor and metastasis formation increased tubulogenesis. Thus, alterations in Met accompanied by improved survival in mice injected signaling can shift the balance between di€erentiation with DN-Met cells. DA3-DN cells proliferate slightly and invasiveness, changing the cells' fate from non- faster than parental DA3 or hgf-sf transfected cells. di€erentiated invasive cells to gland-forming cells. These results indicate that invasiveness, rather than Using a therapeutic modality that induces cell di€er- proliferation rate, has greater signi®cance in determin- entiation has been suggested for many malignancies. ing the tumorigenic-metastatic potential of breast Down-regulating Met activity in patients with high

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2395 Met activity can lead to reduced tumorigenicity ± HA clones) or pMFDN (PL-FDN) and F1DN) and pSTneo metastasis and increased di€erentiation, providing a plasmids (1 : 1) or with pSTneo (DA3N) (Rong et al., 1992). basis for new therapeutic strategies. Such strategies The plasmids were suspended in bu€er solutions and added may include gene therapy using a dominant negative to the DA3 cells as recommended by manufacturer. After form of the receptor as presented in this paper. 18 h of incubation at 378C, the transfection mixture was replaced with DMEM medium containing 10% FCS. After However, a more pragmatic approach would be to 72 h incubation at 378C, the transfected cells were divided ®nd substances, such as HGF/SF competitors or 1 : 5, incubated in DMEM medium containing 10% FCS and speci®c Met phosphorylation inhibitors, capable of 100 mg/ml of G418 (Gibco ± BRL) and resistant clones were impeding Met activity. isolated or pooled. Transient transfection: 293 cells were transfected with pMB11 (293-Met), pMBTDN met (293-TDN) or pMFDN (293-FDN) using the calcium co-precipitation technique Materials and methods (Gorman et al., 1982). Cell lines, animals, cDNA plasmids and antibodies D1-DMBA-3 is a cell line derived from a poorly di€er- Immunofluorescence staining and confocal analysis entiated mammary adenocarcinoma induced in BALB/C mice Eight well Lab-Tek chamber slides (Nunc, Denmark) were by dimethylbenzanthracene. Limiting dilution cloning pro- seeded with 104 cells per well in medium containing 10% duced the cell line designated DA3 (Fu et al., 1990). DA3 and FCS. Cells were ®xed in absolute methanol for 10 min and transfectants were grown in DMEM (Gibco ± BRL Gaithers- consequently permeabilized with cold acetone for 10 min. burg, MD, USA) supplemented with 10% heat-inactivated Following blocking (1% BSA, 10% normal donkey serum in fetal calf serum (FCS) (Gibco ± BRL). MDCK epithelial cells phosphate bu€ered saline ± PBS) for 10 min at room (type 2) provided by Dr KE Mostov (University of California temperature, cells were incubated with primary antibody San Francisco, San Francisco, CA, USA) were grown in diluted 1 : 50 for 1 h at room temperature. Following three DMEM (Gibco ± BRL) supplemented with 5% heat-inacti- washes in PBS, cells were incubated with donkey anti rabbit vated FCS (Gibco ± BRL). HMH cells ± NIH3T3 cells or anti mouse antibody conjugated either to ¯uorescein transfected with human met and human hgf/sf ± were isothiocyanate (FITC) or Texas Red (Jackson Immuno maintained as described previously (Rong et al., 1992). 293 Research Laboratories, INC West Grove, PA, USA) diluted and transfectants were grown in DMEM (Gibco ± BRL) 1 : 50 for 1 h at room temperature. Cells were washed three supplemented with 10% heat-inactivated FCS (Gibco ± BRL). times in PBS and coverslipped with gel mount (Biomeda, Human hgf/sf plasmid was constructed by inserting the Foster City, CA, USA). 2.3 kb BamHI ± KpnI fragment of the human HGF/SF Fluorochrome-labeled cells were analysed using a 410 Zeiss sequence (Nakamura et al., 1989) into the BamHI-KpnI site (Oberkochen, Germany) confocal laser scanning microscope of pMEX (Rong et al., 1992). The following antibodies were (CLSM) with the following con®guration: 25 mW HeNe used: Rabbit polyclonal SP260 anti murine Met peptide lasers (633 nm) and Krypton Argon (488, 568 and 647 nm) antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) laser lines. When comparing ¯uorescence intensity, identical (Gonzatti-Haces et al., 1988), Rabbit polyclonal anti parameters were used for each image (e.g., scanning line, recombinant human HGF/SF antibody (Rong et al., 1993), laser light, contrast brightness). Images were printed using and anti-phosphotyrosine 4G10 mAb (UBI, NY, USA). Codonics dye sublimation color printer.

Construction of dominant negative met expression vectors Mitogenic assay Mouse met cDNA was used to construct DN form of Met by Cells (2500/well) were seeded into 96-well microtiter plates deleting part of the met tyrosine kinase domain which spans with or without HGF/SF (80 u/ml) and incubated for 24 ± between base pairs 3247 and 3984 or by point mutating of three 96 h. Following incubations, 25 ml of MTT (3-(4,5-di- tyrosines (Tyr) in the Met docking site. The pMB11 plasmid, methylthiazol-2-yl)-2,5 diphenyl-tetrazolium bromide, Sigma) containing the full-length mouse met cDNA (Rong et al., 1992) were added to 100 ml of medium in each well and plates were was partially cleaved by BglII restriction enzyme and self-ligated incubated for 2 h. Subsequently, 100 ml of isopropa- omitting base pairs 2880 ± 3531, thus removing a crucial part of nol+0.13% HCl were added to each well and mixed the met tyrosine kinase domain (pMBTDN). The plasmid vigorously. The product of the reaction was determined by containing the full-length mouse met cDNA was point-mutated using an ELISA reader (Bio-tek instruments, Winooski, VT, at three di€erent positions (pMFDN) as previously described USA) at wavelengths of 570/630 nm. (Je€ers et al., 1997). In short, we mutated the Tyr in position 1347 to Phenylalanine (Phe) (TAC to TTC, nucleotide 4039 ± 4041 in mouse met sequence), Tyr in positions 1354 to Phe (TAT Scatter assay to TTT, nucleotide 4060 ± 4062 in mouse met sequence) and the Scatter assay was carried out as previously described (Rosen Tyr in positions 1363 to Phe (TAT to TTT, nucleotide 4087 ± et al., 1989; Stoker et al., 1987). Assay medium (DMEM with 4089 in mouse met sequence). Mutations were con®rmed by 5% calf serum, GIBCO) with or without HGF/SF (80 u/ml) sequencing. pTDN-HA plasmid was constructed by partial was added in 150 ml to 96 well plates (Costar). Cells (5000 restriction using BglII removing base pairs 2880 ± 3531 and cells/well) were added in 150 ml assay medium and incubated ligation with an oligonucleotide pair (positive and negative overnight. Cells were ®xed, air dried, stained with Giemsa strand) containing two copies of the in¯uenza hemagglutinin and examined for scattering (spreading and dispersion of epitope. epithelial colonies). The highest dilution at which scatter e€ect was observed was de®ned as 0.5 HGF/SF scatter unit DNA transfection (u) per ml. Stable transfection: The Lipofectin reagent kit (Gibco ± BRL) was used for DNA transfection. DA3 cells at 70% con¯uence In vitro invasion assay were either co-transfected with 4 mg human hgf/sf (DA3H), Invasion assays were performed as previously described by pMB11 (DA3M), human hgf/sf and pMB11 (DA3MH), Albini et al. (1987) in 96-well chemotaxis chambers (Neuro- pMBTDN (TDN), pMBTDN-HA (PL-TDN-HA and TDN- Probe, MD, USA). Lower and upper wells were separated by

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2396 Nucleopore ®lters (8 mm pore size) coated with Matrigel centrifugation and 1 mg cell lysate protein was IP with SP260 (1 mg/mm2, Collaborative Research). Eighty u/ml of HGF/SF anti Met Ab. The immunoprecipitates were subjected to WB diluted in 30 ml DMEM containing 1 mg/ml BSA was placed analysis using either SP260 antibody (1 : 500) or 4G10 anti-p- in lower wells. Cells (3.66104) in DMEM-BSA were placed Tyr antibody (1 : 1500). Visualization was achieved using in upper wells and chambers were incubated overnight. Non- HRP-conjugated anti-mouse IgG antibody or HRP-conju- invading cells remaining on the upper surface were removed gated protein A (1 : 5000) (Amersham, Arlington Heights, IL, with a cotton swab and invading cells that migrated to the USA), ECL reaction and exposure to X-ray ®lm (Fuji, lower surface of ®lter were ®xed with cold methanol and Japan). stained with Giemsa (Sigma). The stained ®lter was scanned and the area occupied by stained cells (upper threshold 214; In vivo tumor formation and survival analysis lower threshold 1) was analysed using Optimas image analysis software (Seattle, WA, USA). The average and Cells were trypsinized, washed twice and resuspended in PBS. standard deviation was calculated for each transfectant and 56104 cells in 0.2 ml PBS were injected intra-muscularly into the statistical di€erence was calculated using Anova. the leg of BALB/C mice. Tumor formation was monitored twice weekly. After 56 days, mice were sacri®ced and tumor and lung were collected and ®xed in formalin, embedded in Tubule formation in collagen paran, sectioned, stained with hematoxylin and eosin 26105 DA3N, TDN and F1DN cells were seeded in 24-well (H&E) and analysed by light microscopy using histopatho- plates in 0.5 ml of 0.6 mg/ml collagen. HGF/SF (80 u/ml) logical morphological criteria. was added (+) or not added (7) to appropriate wells. Following 5-day incubation, collagen gels were ®xed using Statistical analysis 4% paraformaldehyde+0.1% triton and imaged by Nomars- ki, di€erential interference contrast microscopy (N/DIC). Analyses of tumor-free and overall survival were performed with Kaplan ± Meier (Kaplan and Meier, 1958) and tests for di€erences between survival curves were made with the log- Western blot analysis of HGF/SF rank statistic using SAS (SAS Institute Inc., SAS/STAT Supernatants (8 ml) were collected from con¯uent cells, Software: Changes and Enhancements through Release 6.11, ®ltered and incubated with 0.2 mg/ml heparin sepharose Cary, NC, 1996). (HS) (Pharmacia-Bio-tec, Uppsala, Sweden) for 2 h at room Tests for statistical di€erences between groups were temperature. Following incubation the HS was washed63 performed using Student's t-test and Anova (Microsoft with PBS. Pellets were suspended in sample bu€er, resolved Excel). in 12% SDS ± PAGE and analysed by Western blot analysis using rabbit anti HGF/SF antibody diluted 1 : 350. Visualiza- tion was achieved using HRP-conjugated anti-mouse IgG antibody (1 : 5000) (Amersham, Arlington Heights, IL, USA), ECL reaction and exposure to X-ray ®lm (Fuji, Japan). Acknowledgments This work was performed in partial ful®llment of the requirements for a PhD degree of Michal Firon, Sackler Immunoprecipitation and Western blot analysis of Met and Met Faculty of Medicine, Tel Aviv University, Israel. We wish tyrosine phosphorylation to thank Dr Tom Kmiecik and Dr Sonia Mendlovic for Near-con¯uent cells were treated with HGF/SF (80 u/ml) for their professional assistance. We thank Dr Ira Daar and 0, 5, 10, 20, 30 and 60 min at room temperature. Cells were Dr Eli Golomb for critical review of this work. Partially washed twice with cold PBS and lysed in 1 ml lysis bu€er supported by a grant from the United States ± Israel (20 mM Tris-HCl, pH 7.8, 100 mM NaCl, 50 mM NaF, 1% Binational Science Foundation, Grant # 93-00072, the NP40, 0.1% SDS, 2 mM EDTA, 10% glycerol) with protease Israel Science Foundation founded by the Israel Academy inhibitor cocktail (Boehringer Mannheim, Germany) and of Sciences & Humanities and by the National Cancer 1mM sodium orthovanadate. Cell lysates were clari®ed by Institute, DHHS, under contract with ABL.

References

Albini A, Iwamoto Y, Kleinman HK, Martin GR, Aaronson Clark GM. (1996). Diseases of the breast. Harris JR, SA, Kozlowski JM and McEwan RN. (1987). Cancer Res., Lippman ME, Morrow M and Hellman S (eds). 47, 3239 ± 3245. Lippincott-Raven: Philadelphia, pp. 461 ± 485. Amaya E, Musci TJ and Kirschner MW. (1991). Cell, 66, Cortner J, Vande Woude GF and Rong S. (1995). Epithelial- 257 ± 270. Mesenchymal Interaction in Cancer. (eds). Goldberg ID Benz CC, Brandt BH and Zanker KS. (1995). Gene, 159, 3± and Rosen EM. pp. 89 ± 121. Basel: Birkhuse Verlag. 7. Cunha GR. (1994). Cancer, 74, 1030 ± 1044. Berdichevsky F, Alford D, D'Souza B and Taylor Papadimi- Deng G, Chen LC, Schott DR, Thor A, Bhargava V, Ljung triou J. (1994). J. Cell Sci., 107, 3557 ± 3568. BM, Chew K and Smith HS. (1994). Cancer Res., 54, 499 ± Beviglia L, Matsumoto K, Lin CS, Ziober BL and Kramer 505. RH. (1997). Int. J. Cancer, 74, 301 ± 309. Di Renzo MF, Olivero M, Giacomini A, Porte H, Chastre E, Bieche I, Champeme MH, Matifas F, Hacene K, Callahan R Mirossay L, Nordlinger B, Bretti S, Bottardi S, Giordano and Lidereau R. (1992). Lancet, 339, 139 ± 143. S, Plebani M, Gespach C and Comoglio PM. (1995). Clin. BladtF,RiethmacherD,IsenmannS,AguzziAand Cancer Res., 1, 147 ± 154. Birchmeier C. (1995). Nature, 376, 768 ± 771. Di Renzo MF, Olivero M, Katsaros D, Crepaldi T, Gaglia P, Boccaccio C, Ando M, Tamagnone L, Bardelli A, Michieli P, Zola P, Sismondi P and Comoglio PM. (1994). Int. J. Battistini C and Comoglio PM. (1998). Nature, 391, 285 ± Cancer, 58, 658 ± 662. 288. Ethier SP. (1995). J. Natl. Cancer Inst., 87, 964 ± 973. Bussolino F, Di RM, Ziche M, Bocchietto E, Olivero M, Fu Y, Watson G, Jimenez JJ, Wang Y and Lopez DM. NaldiniL,GaudinoG,TamagnoneL,Co€erAand (1990). Cancer Res., 50, 227 ± 234. Comoglio PM. (1992). J. Cell Biol., 119, 629 ± 641.

Oncogene Dominant negative Met inhibits tumorigenesis-metastasis M Firon et al 2397 Ghoussoub RAD, Dillon DA, D'Aquila T, Rimm EB, Rong S, Donehower LA, Hansen MF, Strong L, Tainsky M, Fearon ER and Rimm DL. (1998). Cancer, 82, 1513 ± Je€ers M, Resau JH, Hudson E, Tsarfaty I and Vande 1520. Woude GF. (1995). Cancer Res., 55, 1963 ± 1970. Giordano S, Zhen Z, Medico E, Gaudino G, Galimi F and Rong S, Je€ers M, Resau JH, Tsarfaty I, Oskarsson M and Comoglio PM. (1993). Proc. Natl. Acad. Sci. USA, 90, Vande Woude GF. (1993). Cancer Res., 53, 5355 ± 5360. 649 ± 653. Rong S, Segal S, Anver M, Resau JH and Vande Woude GF. Gonzatti-Haces M, Seth A, Park M, Copeland T, Oroszlan S (1994). Proc. Natl. Acad. Sci. USA, 91, 4731 ± 4735. and Vande Woude GF. (1988). Proc. Natl. Acad. Sci. Rosen EM, Goldberg ID, Kacinski BM, Buckholz T and USA, 85, 21 ± 25. Vintner DW. (1989). In Vitro Cell Dev. Biol., 25, 163 ± 173. Gorman CM, Merlino GT, Willingham MC, Pastan I and Rosen EM, Grant DS, Kleinman HK, Goldberg ID, Howard BH. (1982). Proc. Natl. Acad. Sci. USA, 79, Bhargava MM, Nickolo€ BJ, Kinsella JL and Polverini 6777 ± 6781. P. (1993). Symp.Soc.Exp.Biol.,47, 227 ± 234. Hacohen N, Kramer S, Sutherland D, Hiromi Y and Rosen EM, Knesel J, Goldberg ID, Jin L, Bhargava M, Krasnow MA. (1998). Cell, 92, 253 ± 263. Joseph A, Zitnik R, Wines J, Kelley M and Rockwell S. Je€ers M, Rong S, Anver M and Vande Woude GF. (1996a). (1994). Int. J. Cancer, 57, 706 ± 714. Oncogene, 13, 853 ± 856. Schmidt C, Bladt F, Goedecke S, Brinkmann V, Zschiesche Je€ers M, Rong S and Vande Woude GF. (1996b). Mol. Cell W, Sharpe M, Gherardi E and Birchmeier C. (1995). Biol., 16, 1115 ± 1125. Nature, 373, 699 ± 702. Je€ers M, Schmidt L, Nakaigawa N, Webb CP, Weirich G, Schmidt L, Duh FM, Chen F, Kishida T, Glenn G, Choyke Kishida T, Zbar B and Vande Woude GF. (1997). Proc. P, Scherer SW, Zhuang Z, Lubensky I, Dean M, Allikmets Natl. Acad. Sci. USA, 94, 11445 ± 11450. R, Chidambaram A, Bergerheim UR, Feltis JT, Casade- Jin L, Fuchs A, Schnitt SJ, Yao Y, Joseph A, Lamszus K, vallC,ZamarronA,BernuesM,RichardS,LipsCJ, Park M, Goldberg ID and Rosen EM. (1997). Cancer, 79, Walther TM, Tsui LC, Geil L, Orcutt ML, Stackhouse T, 749 ± 760. LipanJ,SlifeL,BrauchH,DeckerJ,NiehansG,Hughson Kaplan EL and Meier P. (1958). J. Am. Statistical Soc., 53, MD, Moch H, Storkel S, Lerman MI, Linehan WM and 457 ± 481. Zbar B. (1997). Nat. Genet., 16, 68 ± 73. Lamszus K, Jin L, Fuchs A, Shi E, Chowdhury S, Yao Y, Soriano JV, Pepper MS, Nakamura T, Orci L and Polverini PJ, Laterra J, Goldberg ID and Rosen EM. Montesano R. (1995). J. Cell Sci., 108, 413 ± 430. (1997). Lab. Invest., 76, 339 ± 353. Stoker M, Gherardi E, Perryman M and Gray J. (1987). Liang TJ, Reid AE, Xavier R, Cardi€ RD and Wang TC. Nature, 327, 239 ± 242. (1996). J. Clin. Invest., 97, 2872 ± 2877. Takayama H, LaRochelle WJ, Sharp R, Otsuka T, Kriebel P, Matsumoto K, Matsumoto K, Nakamura T and Kramer Anver M, Aaronson SA and Merlino G. (1997). Proc. RH. (1994). J. Biol. Chem., 269, 31807 ± 31813. Natl. Acad. Sci. USA, 94, 701 ± 706. Nagy J, Clark JS, Cooke A, Campbell AM, Connor JM, Tsarfaty I, Resau JH, Rulong S, Keydar I, Faletto DL and Purushotham AD and George WD. (1995). J. Surg. Vande Woude GF. (1992). Science, 257, 1258 ± 1261. Oncol., 60, 95 ± 99. Tsarfaty I, Rong S, Resau JH, Rulong S, da Silva PP and Nakamura T, Nishizawa T, Hagiya M, Seki T, Shimonishi Vande Woude GF. (1994). Science, 263, 98 ± 101. M, Sugimura A, Tashiro K and Shimizu S. (1989). Nature, Tsarfaty I, Alvord WG, Resau JH, Altstock RT, Lidereau R, 342, 440 ± 443. Bieche I, Bertrand F, Horer J, Klabansky RL, Keydar I NataliPG,NicotraMR,DiRM,PratM,BigottiA, and Vande Woude GF. (1999). AQCH, 397 ± 408. Cavaliere R and Comoglio PM. (1993). Br.J.Cancer, Uehara Y, Minowa O, Mori C, Shiota K, Kuno J, Noda T 68, 746 ± 750. and Kitamura N. (1995). Nature, 373, 702 ± 705. Pepper MS, Soriano JV, Menoud PA, Sappino AP, Orci L Ullrich A and Schlessinger J. (1990). Cell, 61, 203 ± 212. and Montesano R. (1995). Exp. Cell Res., 219, 204 ± 210. Weidner KM, Sachs M, Riethmacher D and Birchmeier W. Ponzetto C, Bardelli A, Zhen Z, Maina F, Dalla ZP, (1995). Proc. Natl. Acad. Sci. USA, 92, 2597 ± 2601. Giordano S, Graziani A, Panayotou G and Comoglio Yamashita J, Ogawa M and Beppu T. (1993). Res. Commun. PM. (1994). Cell, 77, 261 ± 271. Chem. Pathol. Pharmacol., 82, 249 ± 252. Qian X, Dougall WC, Hellman ME and Greene MI. (1994). Yamashita J, Ogawa M, Yamashita S, Nomura K, Kur- Oncogene, 9, 1507 ± 1514. amoto M, Saishoji T and Shin S. (1994). Cancer Res., 54, Rahimi N, Tremblay E, McAdam L, Park M, Schwall R and 1630 ± 1633. Elliott B. (1996). Cell Growth Di€er., 7, 263 ± 270. Yang Y, Spitzer E, Meyer D, Sachs M, Niemann C, Robinson ML, MacMillan-Crow LA, Thompson JA and Hartmann G, Weidner KM, Birchmeier C and Birchmeier Overbeek PA. (1995). Development, 121, 3959 ± 3967. W. (1995). J. Cell Biol., 131, 215 ± 226. Rodrigues GA and Park M. (1993). Mol. Cell Biol., 13, YaoY,JinL,FuchsA,JosephA,HastingsHM,Goldberg 6711 ± 6722. ID and Rosen EM. (1996). Am.J.Pathol.,149, 1707 ± Rong S, Bondescot M, Blair D, Dunn J, Nakamura T, 1717. Mizuno K, Park M, Chan A, Aaronson S and Vande Yarden Y and Schlessinger J. (1987). Biochemistry, 26, Woude GF. (1992). Mol. Cell Biol., 12, 5152 ± 5158. 1443 ± 1451.

Oncogene