MOLECULAR AND CELLULAR BIOLOGY, Nov. 1992, p. 5152-5158 Vol. 12, No. 11 0270-7306/92/115152-07$02.00/0 Copyright X 1992, American Society for Microbiology Tumorigenicity of the met Proto-Oncogene and the for SING RONG,1 MYRIAM BODESCOT,1t DONALD BLAIR,2 JOYCE DUNN,2 TOSHIKAZU NAKAMURA,3 KENSAKU MIZUNO,3 MORAG PARK,4 ANDREW CHAN,5 STUART AARONSON,5 AND GEORGE F. VANDE WOUDE1* ABL-Basic Research Program' and NCI-Laboratory ofMolecular Oncology,2 NCI-Frederick Cancer Research and Development Center, P.O. Box B, Frederick, Maryland 21702; Department ofBiology, Faculty of Science, Kyushu University 33, 6-10-1 Hakozaki, Fukuoka 812 Japan3; Ludwig Institute for Cancer Research, Montreal, Quebec H3A JAI, Canada4; and Laboratory of Cellular and Molecular Biology, Division of Cancer Etiology, National Institutes ofHealth, Bethesda, Maryland 20892 Received 11 March 1992/Returned for modification 20 April 1992/Accepted 17 August 1992 The met proto-oncogene is the tyrosine kinase growth factor receptor for hepatocyte growth factor/scatter factor (HGF/SF). It was previously shown that, like the oncogenic tpr-met, the mouse met proto-oncogene transforms NIH 3T3 cells. We have established NIH 3T3 cells stably expressing both human (Met"') and mouse (Met"') met proto-oncogene products. The protein products are properly processed and appear on the cell surface. NIH 3T3 cells express endogenous mouse HGFISF mRNA, suggesting an autocrine activation mechanism for transformation by Metm'. However, the tumor-forming activity of Met"u in NIH 3T3 cells is very low compared with that of Met"", but efficient tumorigenesis occurs when Met"u and HGF/SF"h are coexpressed. These results are consistent with an autocrine transformation mechanism and suggest further that the endogenous murine factor inefficiently activates the tumorigenic potential of Meteu. The tumorigenicity observed with reciprocal chimeric human and mouse receptors that exchange external -binding domains supports this conclusion. We also show that HGF/SFe' expressed in NIH 3T3 cells produces tumors in nude mice.

The met protooncogene was originally identified as an dissociates and induces the motility of epithelial cells and activated oncogene in a human osteosarcoma cell line (HOS) can promote invasiveness of a number of human carcinoma treated with the chemical carcinogen N-methyl-N'-nitro-N- cell lines in vitro (57). Recently, SF was shown to be nitrosoguanidine (7). Its activation involved a chromosomal identical to HGF (14, 15, 27, 35, 56). HGF/SF is now known rearrangement that linked tpr sequences on 1 to be a paracrine mediator of epithelial morphogenesis (28) with the tyrosine kinase domain of met on chromosome 7 and a pleiotropic factor acting as a mitogen and motogen for (38). The met gene was shown to be amplified and overex- a variety of epithelial and endothelial cells (23, 49). This pressed in spontaneous transformants of NIH 3T3 cells (8, widespread effect of HGF/SF is not surprising, since Met is 21) and in human gastric carcinoma cell lines (16, 26, 41). widely distributed in adult and embryonic tissues as well as The primary met proto-oncogene product is a 150-kDa in established cell lines (11, 18, 20, 24, 38, 53). precursor, which is glycosylated to generate a 170-kDa It was previously shown that spontaneous mouse NIH 3T3 (p170"') protein. The mature form of the receptor is gener- cell tumors have amplified met sequences and that overex- ated by cleavage of p170"' to yield a disulfide-linked pression of the mouse met proto-oncogene (metmu) trans- 140-kDa e subunit (p1401?w) and a 45-kDa a subunit (p45Pet). forms NIH 3T3 cells (8, 24). It was subsequently shown that The a subunit is extracellular, while the e subunit encom- fibroblast cells produce endogenous HGF/SF (49), suggest- passes a major portion of the extracellular domain, the ing that metmu transforms cells via an autocrine mechanism. transmembrane domain, and the intracellular tyrosine kinase To further study the mechanism of met transformation, we domain (6, 17, 18). Recent data suggest that a-P het- have expressed the human (met"") and metmu proto-onco- erodimeric structures are clustered on the cell surface (13). in NIH 3T3 cells and have tested their tumorigenicity The ligand for the met proto-oncogene product has been in nude mice (3). We show that NIH 3T3 cells expressing identified as hepatocyte growth factor (HGF) (5, 34). HGF methu were poorly tumorigenic in this assay compared with was originally identified as a potent mitogen for hepatocytes NIH 3T3 cells expressing met"u. However, efficient tumor (31) and purified as a disulfide-linked heterodimer consisting formation was observed when met"u was coexpressed with of a 69-kDa a subunit and a 34-kDa e subunit (32). Molecular HGF/SFhu. Moreover, we show that NIH 3T3 cells trans- cloning of HGF cDNA revealed that HGF is derived from a fected with HGFISFT' were tumorigenic in nude mice. single-chain precursor and has a structure similar to that of plasminogen in its N-terminal kringle domains and C-termi- nal serine protease-like domain (30, 33). Independently, MATERIALS AND METHODS scatter factor (SF) was identified as a secretory protein of fibroblasts and cDNA plasmid constructs and cell lines. All met cDNA smooth muscle cells (45, 46, 51, 52) that plasmids were constructed in PMB1, a derivative (without the polylinker sequences) of pMEX (a gift from Dionisio Martin-Zanca [37]) that contains the long terminal repeat * Corresponding author. promoter from Moloney murine sarcoma virus and the t Present address: Laboratoire d'Oncologie Moleculaire, Institut polyadenylation signal of simian virus 40. The meth" plasmid Gustave-Roussy, 94805 Villejuif cedex, France. was constructed by replacing an internal 300-bp EcoRI 5152 VOL. 11, 1992 met TRANSFORMATION 5153

fragment with the 250-bp EcoRI fragment of pOK in the with a stream of nitrogen. The lodo-Gen was then dissolved 4.6-kb methu sequence containing the open reading frame in 1 M Tris with 10 mM EDTA (pH 7.5) and added to the cell (39, 44). The metmu plasmid contains the entire 4.6-kb mouse pellet. Nal'I (0.5 mCi) was added to the reaction mixture met open reading frame (24). Chimeric human-mouse met for 10 min. The labeled cells were washed three times with constructs were made by using the conserved PvuII site phosphate-buffered saline, lysed with RIPA buffer, and (amino acid 807). The HGFhU plasmid was constructed by subjected to immunoprecipitation analysis. inserting the 2.3-kb BamHI-KpnI fragments of the human Western immunoblot analysis. Near-confluent cells in 100- HGF sequence into the BamHI-KpnI sites of pMEX (33, 37). mm-diameter dishes were washed twice with cold TBS (10 NIH 3T3 490 cells were grown in Dulbecco modified Eagle mM Tris [pH 8.0], 150 mM NaCl) and lysed in 1 ml of lysis medium (DMEM; GIBCO) with 8% calf serum (GIBCO). buffer (25 mM Tris-HCl [pH 8.0], 100 mM NaCl, 50 mM DNA transfection. The calcium phosphate method of DNA NaF, 1% Triton X-100, 10 p,g of aprotinin per ml, 10 jig of transfection (7) was carried out by mixing plasmid DNA (2 leupeptin per ml, 1.25 mM phenylmethylsulfonyl fluoride, 1 p,g in 75 ,ul of water containing 8 ,ug of calf thymus carrier mM vanadate). Cell lysates were immunoprecipitated with DNA) with 75 p,l of 0.67 M CaCl2. This mixture was added SP260 or C28 peptide antibody and subjected to Western dropwise to 0.15 ml of solution H (0.27 M NaCl, 0.01 M KCI, analysis, using either antiphosphotyrosine (anti-P-Tyr) 4G10 0.0014 M Na2HPO4- 7H20, 0.012 M dextrose) with contin- (a gift from Deborah Morrison and David Kaplan [291), 19S uous agitation. After 30 min at room temperature, the monoclonal antibody, or SP260 peptide antibody. 1 I-pro- mixture was added to the cells, which were 70% confluent on tein A (Amersham) was used to detect positive bands 35-mm-diameter dishes (7). Cells were incubated at 37°C for according to the manufacturer's instructions. 4 h and then treated with 15% (vol/vol) glycerol in solution H Nude mouse tumor assay. The assay was performed as for 2 min. For G418 selection, cells were incubated with previously described (3). Transfected and G418-selected DMEM and 8% calf serum overnight and subsequently NIH 3T3 cells (106) were washed twice and resuspended in transferred to three 60-mm-diameter dishes. After a 24-h 0.1 ml of serum-free medium. The cells were injected sub- incubation, cells were fed twice weekly with medium con- cutaneously into the backs of weanling athymic nude mice taining 400 p,g of G418 (GIBCO) per ml. (BALB/c) (Harlan Sprague Dawley, Inc.). Tumor formation Immunoprecipitation. Near-confluent cells were labeled was monitored each week for up to 10 weeks. Tumors were for 4 to 6 h with 0.25 mCi of Translabel (ICN) (1 ml/35-mm- explanted when they reached 15 mm in size, and the tumor diameter dish) in DMEM lacking methionine and cysteine cells were subjected to immunoprecipitation analysis. (GIBCO). The labeled cells were lysed in 0.5 ml of RIPA Soft-agar assay. The soft-agar growth assay was carried buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% out by a modification of the method of Blair et al. (4). Briefly, sodium dodecyl sulfate [SDS], 0.15 M NaCl, 0.02 M NaPO4 trypsinized cells were suspended at 2 x 105 and 2 x 104 cells [pH 7.2], 1 mM phenylmethylsulfonyl fluoride, 2 mM EDTA, in 8 ml of DMEM (GIBCO) with 10% calf serum and 0.24% 50 mM NaF, 30 mM sodium pyrophosphate). Clarified purified agar (Difco) and quickly transferred to a duplicate lysates were immunoprecipitated with appropriate antibod- 60-mm-diameter dish containing a hardened base layer of ies at 4°C overnight. Immunoprecipitates were complexed to DMEM with 10% calf serum and 0.27% agar. Plates were fed protein G-Sepharose (GIBCO) and then washed twice with with 2 ml of DMEM with 10% calf serum and 0.27% agar at RIPA buffer and high-salt buffer (1 M NaCl, 10 mM Tris-HCl weekly intervals. Colonies were counted microscopically [pH 7.2], 0.5% Triton X-100). The immunocomplexes were after 3 weeks at 37°C. solubilized by boiling in SDS sample buffer in the presence of 5% P-mercaptoethanol. Samples were analyzed by SDS- RESULTS polyacrylamide gel electrophoresis (PAGE) followed by treatment with fluorographic enhancer (Amplify; Amer- Expression of Methu and Met"u in NIH 3T3 cells. Plasmids sham) and fluorography with an intensifying screen at containing met proto-oncogene cDNA were cotransfected -700C. with pSV2neo into NIH 3T3 cells, and G418-resistant trans- Pulse-chase analysis. Near-confluent cells were labeled for fected cells were screened for Met expression by immuno- 45 min with 0.25 mCi of Translabel (1 ml135-mm-diameter precipitation analysis. Of approximately 50 lines examined, dish) in DMEM lacking methionine and cysteine. The cells several expressed high levels of either the human or mouse were washed twice, chased with complete medium for 0.5, 2, Met product. One example of each is shown in Fig. 1A (lanes and 4 h, lysed, and subjected to immunoprecipitation anal- 2 and 3, respectively). We find little or no expression of the ysis. endogenous Metmu in the G418-resistant cells expressing 19S mouse monoclonal antibody was generated against the Methu (data not shown). Also, in pulse-chase experiments, bacterially expressed p50 form ofmethu (13, 18). C28 peptide Methu and Metmu products are appropriately processed in antibody was raised in rabbits by immunization with the NIH 3T3 cells (data not shown). Moreover, human and C-terminal 28 amino acids of Methu (18) (NH2-Ala-Pro-Tyr- mouse Met are localized on the cell surface. We labeled Pro-Ser-Leu-Leu-Ser-Ser-Glu-Asp-Asn-Ala-Asp-Asp-Glu- intact cells with Na125I and showed by immunoprecipitation Val-Asp-Thr-Arg-Pro-Ala-Ser-Phe-Trp-Glu-Thr-Ser-COOH). of the lysates that both forms, pl4OP7t and p170fet, were SP260 peptide antibody is a rabbit antiserum directed against iodinated (Fig. 1B). Thus, human or mouse Met expressed in the C-terminal 21 amino acids of Metmu (24) (NH2-Cys-Val- NIH 3T3 cells is correctly processed and localized on the Ala-Pro-TTyr-Pro-Ser-Leu-Leu-Pro-Ser-Gln-Asp-Asn-Ile- cell surface. Asp-Gly-Glu-Gly-Asn-Thr-COOH). A3.1.2 is a monoclonal These analyses also indicate that p170?e' can arrive at the antibody against human recombinant HGF (immunoglobulin cell surface uncleaved. Iodination of p170fet did not occur in G, subclass G2a). lysed cells, since under similar iodination conditions, the Surface iodination. Near-confluent cells were labeled with cytoplasmic Tpr-Met oncoprotein was not iodinated (18; Na1"I in the presence of Iodo-Gen (Pierce). Twenty micro- data not shown). Furthermore, p170m?' expressed in Oka- liters of the Iodo-Gen reagent (10 mg/ml in chloroform) was jima cells, a human gastric carcinoma cell line that overex- added to the bottom of a 1-dram (3.697-ml) vial and dried presses Methu, was also labeled by surface iodination (data 5154 RONG ET AL. MOL. CELL. BIOL.

A B TABLE 2. Tumorigenicity of NIH 3T3 cells transfected with methU with or without HGFISFPU cDNA 1 2 3 1 2 3 4 No. of mice with kDa P. v t kDa TransfectedTransfectdgene(s)tumors/no,gene(s) tested Ltny(kLatency (wk) - 200 - j 200 ... - neo 0/6 69- neo, methu, HGF/SFhu 17/19 4-6 % 92.5- 9255- neo, HGF/SFIhU 3/7 7

6g ^ 69-

t r _w 46- 46 - 5 of met1"-transfected NIH 3T3 cells is that Methu receptor 30- 30- activation by endogenous HGF/SFmU may not provide a sufficient signal. We therefore tested whether transfection of FIG. 1. Met products in transfected NIH 3T3 cells. (A) Immu- both met"u and HGF/SF"h cDNAs would increase tumori- noprecipitation analysis. Cells were metabolically labeled with genicity through an autocrine mechanism. These analyses [35S]methionine and [35S]cysteine for 5 h, and cell lysates were immunoprecipitated with either 19S monoclonal antibody (lanes 1 show that NIH 3T3 cells cotransfected with methu and and 2) or SP260 peptide antibody (lane 3). Immunoprecipitates were HGF/SF"u are highly tumorigenic (Table 2). Moreover, the solubilized in SDS sample buffer, subjected to SDS-PAGE analysis tumor cells showed higher levels of expression of both and fluorography, and exposed on film for 40 h. Lanes: 1, cells Meth" and HGF/SFhu than did the parental lines (Fig. 2A and transfected with pSV2neo only; 2, cells transfected with methu; 3, C, respectively; lanes 2, 4, and 5). None of the tumor cells transfected with metru. (B) Cell surface iodination of Met. explants showed increased levels of Metmu expression (Fig. Near-confluent methu-transfected NIH 3T3 cells (lanes 1 and 3) and 2B). We found that HGF/SFhu-transfected NIH 3T3 cells metOu-transfected NIH 3T3 cells (lanes 2 and 4) were labeled with also produced several tumors, but the levels of HGF/SFhu Na'25I in the presence of Iodo-Gen (Pierce). The labeled cells were expressed in tumor cell explants was not as high as the levels lysed with RIPA buffer, subjected to immunoprecipitation with either 19S monoclonal antibody (lanes 1 and 2) or SP260 peptide expressed in tumors from the metzu-HGF/SF"u cotransfec- antibody (lanes 3 and 4), and exposed on film for 16 h. Arrows tion experiments (data not shown and Fig. 2C). In one of five indicate the positions of p170"' and p140(ef. HGF/SFhu tumors, we detected elevated levels of endoge- nous Metmu. Activation of Methu and Metmu in NIH 3T3 cells. The results presented above provided evidence that Met trans- not shown), but the ratio of pl70m7' to p140fe, labeled was forming potential could be activated via an autocrine mech- less than in NIH 3T3 cells. anism and that species specificity was responsible for the Tumorigenicity of Met in NIH 3T3 cells. We observed that inefficient transformation by Methu. We therefore examined NIH 3T3 cell cultures expressing Metmu, but not Methu, whether the Methu and Metmu expressed in these cells were transformed (24) (see Table 3; also data not shown). reacted with anti-P-Tyr. Extracts from cell lines expressing This finding was confirmed by testing for the tumorigenicity Methu and Metmu were subjected to immunoprecipitation of the cultures in nude mice following transfection and with C28 peptide antibody (for Methu) or SP260 peptide G418 selection. NIH 3T3 cells expressing Metmu were antibody (for Metmu) and Western analyses with either highly tumorigenic (Table 1), while cells expressing Methu anti-P-Tyr (Fig. 3A, row a) or 19S monoclonal antibody and were low in tumorigenicity. We conclude that the Methu SP260 peptide antibody (Fig. 3A, row b). These analyses is poorly tumorigenic in NIH 3T3 cells. By Northern show that Metmu is much more reactive with anti-P-Tyr than (RNA) hybridization analyses (data not shown), we detect is Methu (Fig. 3A). While we cannot directly compare the HGF/SF mRNA expression in nontransfected NIH 3T3 relative amounts of Metmu and Methu, the data suggest that cells and in cells expressing either methu or metmu. Thus, there is a much higher anti-P-Tyr reactivity with the Metmu transformation by metmu may be mediated via an autocrine product (Fig. 3A, row a). We performed pulse-chase analy- mechanism. ses to examine the stability of the Met receptors. The Tumorigenicity of NIH 3T3 cells cotransfected with met"u turnover rate of Metmu or Methu when coexpressed with and HGF/SFIU. One explanation for the low tumorigenicity HGF/SFhu was much faster than that of Methu alone (Fig. 3B). Collectively, these results suggest that activation of Methu by mouse HGF/SF is inefficient. Tumorigenicity of chimeric human-mouse Met in NIH 3T3 cells. To examine whether the ligand-binding domain influ- TABLE 1. Tumorigenicity of NIH 3T3 cells transfected with enced tumorigenicity, we generated chimeric human-mouse methu or metmU cDNA Met receptor molecules and tested their tumorigenicity in Transfected No. of mice with nude mice (Table 3). We used a conserved PvuII site in the gene(s)a tumors/no. tested Latency (wk) external domain adjacent to the transmembrane coding sequences to make these recombinants. We found that when neo 0/19 neo, methu 2/41b 5 the mouse external ligand-binding domain was linked to the neo, metmu 17/17 3-5 human transmembrane and tyrosine kinase domains, the chimeric receptor displayed tumorigenic activity equivalent a Cells (106) were washed twice with serum-free medium and injected to that of Metmu (Table 3). Explants of these tumors showed subcutaneously on the backs of weanling athymic nude mice. Tumor forma- levels of the chimeric Met that was recog- tion was monitored each week for up to 10 weeks. increased protein b These two tumors displayed higher levels of endogenous Metmu and lower nized with the antibody directed against the human tyrosine levels of Methu than did those of the parental line. kinase domain (Fig. 4, lanes 2 and 3). No evidence for Metmu VOL. 11, 1992 met TRANSFORMATION 5155 A B C M 1 2 3 4 5 6 M 1 2 3 4 5 6 1 2 3456 kDa kDa kDa M 200 - 200 - is235 200 - -- _8- =O. 92.5 - 92.5-- --__ 92.5- am 69- 69- 69- ._ j a

- 45- - 45- %-.. -- 45-

30- 30- 30- --______

FIG. 2. Characterization of Met and HGF/SF in NIH 3T3 tumor explants. Tumor cells were explanted and metabolically labeled with [35S]methionine and [35S]cysteine for 6 h. Cell lysates were immunoprecipitated with either 19S monoclonal antibody (A) or SP260 peptide antibody (B). A 0.25-ml aliquot of 6-h supernatants was concentrated threefold in a Centricon apparatus (molecular weight cutoff of 10,000; Amicon); the volumes were adjusted to 0.3 ml with RIPA buffer, and the samples were immunoprecipitated with HGF monoclonal antibody A3.1.2 (C) and exposed on film for 16 h. Lanes: 1 and 3, samples from two different lines of cotransfected cells before injection; 2, tumor explant derived from the cells analyzed in lane 1; 4 and 5, tumor explants derived from the cells analyzed in lane 3; 6, sample prepared from control NIH 3T3 cells. Arrows indicate the positions of pl70mlw and p140"'- (A and B) and the positions of the 87-kDa (precursor), 69-kDa, and 34-kDa HGF polypeptides (C).

amplification was observed in these tumors (data not three human cell lines but has no effect on EMT6 cells (2). shown), and the chimeric product was recognized by West- Similar species specificity of HGF/SF has been reported for ern analysis with anti-P-Tyr (Fig. 4, lane 5). the activation of Met and its potential substrates in signal In contrast to the high tumorigenicity of the chimera with transduction studies (12). These studies do not exclude the the mouse Met external ligand-binding domain, the recipro- possibility that very high levels of ligand lead to activation of cal human N-terminal-mouse C-terminal chimera was the heterologous receptor. Furthermore, mouse CSF-1 does poorly tumorigenic (Table 3). However, when this chimera not bind to the human CSF-1 receptor with high affinity (47), was cotransfected with HGF/SFU cDNA, efficient tumor and even though NIH 3T3 cells synthesize CSF-1, only formation was observed (Table 3). We also observed that coexpression of both human CSF-1 and CSF-1 receptor can cells expressing the Metmu-Methu chimera, before injection lead to transformation and tumorigenicity of transfected into nude mice, were transformed and formed colonies in NIH 3T3 cells (48). soft agar. In contrast, cells expressing the Methu-Metmu The p170m?t precursor is localized on the cell surface of chimera did not display a transformed phenotype unless NIH 3T3 cells. The unprocessed precursor has previously coexpressed with HGF/SF. Therefore, we conclude that the Metmu external ligand-binding domain is a major factor in determining tumorigenicity. A. 1 2 3 4 B. 1 2 3 4

_ DISCUSSION a. ;S a. _ Cell lines that simultaneously express growth factors with their cognate receptors elicit tumorigenic behavior via auto- b.5II --IV "We-4-. crine growth factor stimulation (50, 54). The transforming activities of epidermal growth factor receptor (10, 40, 43, 55) and colony stimulatory factor 1 (CSF-1) receptor (48) in NIH 3T3 cells have been shown to be factor dependent. Here we C. 4 show that NIH 3T3 cells express endogenous HGF/SFmu, FIG. 3. Activation of Methu and Met"m in NIH 3T3 cells. (A) Met which provides an explanation for the transforming activity protein reactivity with anti-P-Tyr. Near-confluent methu-transfected of the metm' proto-oncogene (24). We also show that Metm" cells (lanes 1 and 2) and metu-transfected cells (lanes 3 and 4) were expressed in NIH 3T3 cells is readily recognized by anti-P- lysed in lysis buffer and immunoprecipitated with C28 peptide Tyr, suggesting that transformation may occur via autocrine antibody for Methu (lanes 1 and 2) or SP260 peptide antibody for activation of the receptor. In contrast, NIH 3T3 cells ex- Metmu (lanes 3 and 4). After being dissolved in SDS buffer, samples pressing Methu are generally not tumorigenic, and Methu were separated by SDS-PAGE on 7.5% gels, transferred to Immo- reacts poorly with anti-P-Tyr. However Methu is activated bilon-P (Millipore), and probed with anti-P-Tyr (row a), 19S mono- clonal antibody (row b, lanes 1 and 2), or SP260 peptide antibody when it is coexpressed with HGF/SFh . One explanation for these results is that Meth" and HGF/SF"" interact with low (row b, lanes 3 and 4). (B) Pulse-chase analysis. Cells were meta- bolically labeled with [3 S]methionine and [3 Slcysteine for 45 min affinity. Species specificity of the HGF/SF ligand has previ- (lane 1) and then chased for 0.5 h (lane 2), 2 h (lane 3), and 4 h (lane a ously been reported (2, 12). Thus, purified SFmu induces 4). Met was immunoprecipitated with SP260 peptide antibody (row scattering response in the mouse mammary carcinoma cell a) or 19S monoclonal antibody (rows b and c), subjected to electro- line EMT6 but does not induce scattering in the human phoresis as described above, and exposed on film for 16 h. Rows: a, carcinoma cell lines A253, FaDu, and YaOVBix 2NMA (2). cells transfected with metmu; b, cells cotransfected with methu and On the other hand, purified SFhu induces scattering in all HGF"u; c, cells transfected with met"u. 5156 RONG ET AL. MOL. CELL. BIOL.

TABLE 3. Transforming properties of NIH 3T3 cells transfected with human-mouse chimeric cDNAs with or without HGF/SFhu cDNA b No. of mice with Latency Transformed Colony formation Transfected gene(s)a met DNA structure tumors/no. tested (wk) phenotype in soft agar metmu 11/12 3-5 + + methu 0/7 metmu-methu (PvuII) I= xx 6/7 6-7 + + methu-metmu (PvuII) XXXXXXXX 1/11 5 - NDc methu-metmu (PvuII) X X X 6/7 3-6 + ND + HGF/SF"u a All genes were cotransfected with neo. b met DNA of mouse (X) or human (-) origin. c ND, not determined. been shown to be expressed on the cell surface of Meth"- critical review of the manuscript. We also thank Joan Hopkins for expressing insect cells together with the cleaved Met recep- typing the manuscript. tor (35). Moreover, p170tm' was reactive with anti-P-Tyr This research was sponsored in part by the National Cancer along with pl40?w in Western analysis (Fig. 3A). We have Institute under contract N01-CO-7410I with ABL. expressed Methu in mouse cells that do not express endog- REFERENCES enous HGF/SF. In these cells, pl70fee (and p140Pee) reacted 1. Aaronson, S. A. 1991. Growth factors and cancer. Science with anti-P-Tyr only when exogenous HGF/SF is added 254:1146-1153. (data not shown). Thus, p170net may, in the absence of 2. Bhargava, M., A. Joseph, J. Knesel, R. Halaban, Y. Li, S. Pang, cleavage, bind HGF/SF and contribute to the receptor signal I. Goldberg, E. Setter, M. A. Donovan, R. Zaraegar, G. A. transduction pathway. The uncleaved insulin receptor pre- Michalopoulos, T. Nakamura, D. Falleto, and E. M. Rosen. 1992. cursor has been shown to bind insulin, indicating that Scatter factor and hepatocyte growth factor: activities, proper- cleavage is not required to generate an insulin-binding site ties and mechanisms. Cell Growth Differ. 3:11-20. (36). Moreover, mutation at the insulin receptor gene in the 3. Blair, D. G., C. S. Cooper, M. K. Oskarsson, L. A. Eader, and protein cleavage site resulted in the expression, on the cell G. F. Vande Woude. 1982. New method for detecting cellular transforming genes. Science 218:1122-1125. surface, of an uncleaved proreceptor that has low affinity for 4. Blair, D. G., M. A. Hull, and E. A. Finch. 1979. The isolation insulin (58). and preliminary characterization of temperature-sensitive trans- A number of growth factors, such as platelet-derived formation mutants of Moloney sarcoma virus. Virology 95:303- growth factor (A and B), epidermal growth factor, trans- 316. forming growth factor a, basic and acidic fibroblast growth 5. Bottaro, D. P., J. S. Rubin, D. Faletto, A. M.-L. Chan, T. E. factor, and fibroblast growth factor 5 have been shown to be Kmiecik, G. F. Vande Woude, and S. A. Aaronson. 1991. The oncogenic (1, 9, 19). We show that HGF/SFhU was tumori- hepatocyte growth factor receptor is the c-met protooncogene genic in NIH 3T3 cells. These results also suggest that while product. Science 152:802-804. HGF/SFm' is inefficient in activating Methu, HGF/SFhU may 6. Chan, A. M.-L., H. W. S. King, E. A. Peakin, P. R. Tempest, J. be more in Metmu. Hilkens, V. Kroozen, D. R. Edwards, A. J. Wills, C. S. Cooper, efficient activating and P. Brooke. 1988. Characterization of the mouse met pro- tooncogene. Oncogene 2:593-599. ACKNOWLEDGMENTS 7. Cooper, C. S., M. Park, D. G. Blair, M. A. Tainsky, K. Huebner, C. M. Croce, and G. F. Vande Woude. 1984. Molecular cloning We are grateful to Jim Dobbs, Oscar Smith, and Louise Cromwell of a new transforming gene from a chemically-transformed for their assistance, to Anne Arthur for editing, and to Renping human cell line. Nature (London) 311:29-33. Zhou, Ilan Tsarfaty, Thomas Kmiecik, and Donna Faletto for 8. Cooper, C. S., P. R. Tempest, P. M. Beckman, C.-H. Heldin, and P. Breakers. 1986. Amplification and overexpression of met gene in spontaneously transformed NIH/3T3 mouse fibroblast. 1 2 3 4 5 EMBO J. 5:2623-2628. kDa 9. Cross, M., and T. M. Dexter. 1991. Growth factors in develop- ment, transformation, and tumorigenesis. Cell 64:271-280. 200- = 10. DiFore, P. P., J. H. Pierce, T. P. Fleming, R. Hazan, A. Ulirich, 92.5 C. R. King, J. Schlessinger, and S. A. Aaronson. 1987. Overex- pression of the human EGF receptor confers an EGF-dependent 69- transformed phenotype to NIH/3T3 cells. Cell 51:1063-1070. 11. DiRenzo, M. F., R. P. Narsimhan, M. Olivero, S. Bretti, S. 46- Giordano, E. Medico, P. Gaglia, P. Zara, and P. M. Comoglio. 1991. Expression of the Met/HGF receptor in normal and neoplastic human tissues. Oncogene 6:1997-2003. 30 12. Faletto, D. L., D. R. Kaplan, D. 0. Halverson, E. R. Rosen, and G. F. Vande Woude. In I. D. Goldberg and E. R. Rosen (ed.), FIG. 4. Characterization of Met chimeric protein in NIH 3T3 Hepatocyte growth factor/scatter factor (HGF/SF) and the tumor explants. Cells were metabolically labeled with [35S]methio- c-met receptor. Signal transduction in c-met mediated motoge- nine and [35S]cysteine for 6 h, and cell lysates were immunoprecip- nesis, in press. Birkhauser Verlag Co., Cambridge, Mass. itated with 19S monoclonal antibody (lanes 1 to 4). Lanes: 1, 13. Faletto, D. L., I. Tsarfaty, T. E. Kmiecik, M. Gonzatti, T. parental G418-resistant cells transfected with mouse N-terminal- Suzuki, and G. F. Vande Woude. 1991. Evidence for noncova- human C-terminal chimeric met; 2 and 3, tumors formed after lent clusters of the c-met protooncogene product. Oncogene injection of cells analyzed in lane 1; 4, NIH 3T3 control cells; 5, 7:1149-1157. anti-P-Tyr Western analysis of cells analyzed in lane 1 after immu- 14. Furlong, R. A., T. Takehara, W. G. Taylor, T. Nakamura, and noprecipitation with C28 peptide antibody. J. S. Rubin. 1991. Comparison of biological and immunochem- VOL. 11, 1992 met TRANSFORMATION 5157

ical properties indicates that scatter factor and hepatocyte of the receptor encoded by the protooncogene c-met. Oncogene growth factor are indistinguishable. J. Cell Sci. 100:173-177. 6:501-504. 15. Gherardi, E., and M. Stoker. 1990. Hepatocytes and scatter 35. Naldini, L., K. M. Weidner, E. Vigna, G. Gaudino, A. Bardell, factor. Nature (London) 346:228. C. Ponzetto, R. P. Narsimhan, G. Hartmann, R. Zarnegar, G. K. 16. Giordano, S., M. F. DiRenzo, R. Ferracini, L. Chiado-Plat, and Michalopoulos, W. Birchmeier, and P. M. Comoglio. 1991. P. M. Comoglio. 1988. p145, a protein with associated tyrosine Scatter factor and hepatocyte growth factor are indistinguish- kinase activity in a human gastric carcinoma cell line. Mol. Cell. able ligands for the met receptor. EMBO J. 10:2867-2878. Biol. 8:3510-3517. 36. Olson, T. S., and M. D. Lane. 1987. Post-translational acquisi- 17. Giordano, S., C. Ponzetto, M. F. DiRenzo, C. S. Cooper, and tion of insulin-binding activity by the insulin proreceptor. J. P. M. Comoglio. 1989. Tyrosine kinase receptor indistinguish- Biol. Chem. 262:6816-6822. able from the c-met protein. Nature (London) 339:155-158. 37. Oskam, R., F. Coulier, M. Ernst, D. Martin-Zanca, and M. 18. Gonzatti-Haces, M., A. Seth, M. Park, T. Copeland, S. Oroszlan, Barbacid. 1988. Frequent generation of oncogenes by in vitro and G. F. Vande Woude. 1988. Characterization of the tpr-met recombination of trk protooncogene sequences. Proc. Natl. oncogene p65 and the met protooncogene p140 protein-tyrosine Acad. Sci. USA 85:2964-2968. kinase. Proc. Natl. Acad. Sci. USA 85:21-25. 38. Park, M., M. Dean, C. S. Cooper, M. Schmidt, S. J. O'Brien, 19. Heldin, C.-H., and B. Westermark. 1989. Growth factors as D. G. Blair, and G. F. Vande Woude. 1986. Mechanism of met transforming proteins. Eur. J. Biochem. 184:487-496. oncogene activation. Cell 45:895-904. 20. Higuchi, O., and T. Nakamura. 1991. Identification and change 39. Park, M., M. Dean, K. Kaul, M. J. Braun, M. A. Gonda, and in the receptor for hepatocyte growth factor in rat liver after G. F. Vande Woude. 1987. Sequence of met protooncogene partial hepatectomy or induced hepatitis. Biochem. Biophys. cDNA has features characteristic of the tyrosine kinase family Res. Commun. 176:599-607. of growth-factor receptor. Proc. Natl. Acad. Sci. USA 84:6379- 21. Hudziak, R. M., G. D. Lewis, W. E. Holmes, A. Ullrich, and 6383. H. M. Shepard. 1990. Selection for transformation and met 40. Pierce, J. H., M. Ruggiero, T. P. Fleming, P. O. DiFiore, J. S. protooncogene amplification in NIH/3T3 fibroblasts using tumor Greenberger, L. Varticovski, J. Schlessinger, G. Rivera, and S. neurosis factor a. Cell Growth Differ. 1:129-134.17. Aaronson. 1988. Signal transduction through the EGF receptor 22. Hudziak, R. M., J. Schiessinger, and A. Ullrich. 1987. Increased transfected in IL-3-dependent hematopoietic cells. Science 239: expression of the putative growth factor receptor pl85''R 628-631. causes transformation and tumorigenesis of NIH/3T3 cells. 41. Ponzetto, C., S. Giordano, F. Peverali, G. Della Valie, M. Abate, Proc. Natl. Acad. Sci. USA 84:7159-7163. G. Vaula, and P. M. Gmoglio. 1991. C-met is amplified but not 23. Igawa, T., S. Kanda, H. Kanetake, Y. Saitoh, A. Ichihara, Y. mutated in a cell line with an activated met tyrosine kinase. Tomita, and T. Nakamura. 1991. Hepatocyte growth factor is a Oncogene 6:553-559. potent mitogen for cultured rabbit tubular epithelial cells. Bio- 42. Prat, M., T. Crepaldi, L. Gandino, S. Giordano, P. Longati, and chem. Biophys. Res. Commun. 174:831-838. P. Comoglio. 1991. C-terminal truncated forms of Met, the 24. Iyer, A., T. E. Kmiecik, M. Park, I. Daar, P. Blair, K. J. Dunn, hepatocyte growth factor receptor. Mol. Cell. Biol. 11:5954- P. Sutrave, J. N. Ihie, M. Bodescot, and G. F. Vande Woude. 5962. 1990. Structure, tissue-specific expression, and transforming 43. Riedel, H., S. Massozlia, J. Schlessinger, and A. Ullrich. 1988. activity of the mouse met protooncogene. Cell Growth Differ. Ligand activation of overexpressed epidermal growth factor 1:87-95. receptors transforms NIH/3T3 mouse fibroblasts. Proc. Natl. 25. Kan, M., G. H. Zhang, R. Zarnegar, G. Michalopoulos, Y. Acad. Sci. USA 85:1477-1481. Myoken, W. L. McKeehan, and J. L. Stevens. 1991. Hepatocyte 44. Rodriguez, G. A., M. A. Naujokas, and M. Park. 1991. Alterna- growth factor-hepatopoietin A stimulates the growth of rat tive splicing generates isoforms of the c-met receptor tyrosine proximal tubule epithelial cells (rpte), rat non-parenchymal liver kinase which undergo differential processing. Mol. Cell. Biol. cells, human cells, mouse keratinocytes, and stimu- 11:2962-2970. lates anchorage-independent growth of SV40-transformed rpte. 45. Rosen, E. M., I. D. Goldberg, B. M. Karinski, T. Buckhdz, and Biochem. Biophys. Res. Commun. 174:331-337. D. W. Vimter. 1989. Smooth muscle releases an epithelial scatter 26. Kmiecik, T. E. Unpublished data. factor which bind to heparin. In Vitro Cell Dev. Biol. 25:163- 27. Konishi, T., T. Takehara, T. Tsuji, K. Ohsato, K. Matsumoto, 173. and T. Nakamura. 1991. Scatter factor from human embryonic 46. Rosen, E. M., L. Meromsky, E. Setter, D. W. Vinter, and I. D. lung fibroblasts is probably identical to hepatocyte growth Goldberg. 1990. Smooth muscle-derived factor stimulates mo- factor. Biochem. Biophys. Res. Commun. 180:765-773. bility of human tumor cells. Invasion Metastasis 10:49-64. 28. Montesano, R., K. Matsumoto, T. Nakamura, and L. Orci. 1991. 47. Roussel, M. F., J. R. Downing, C. W. Rettenmier, and C. J. Identification of a fibroblast-derived epithelial morphogen as Sherr. 1988. A point mutation in the extracellular domain of the hepatocyte growth factor. Cell 67:901-908. human CSF-1 receptor (c-fins protooncogene product) activates 29. Morrison, D. K., D. R. Kaplan, J. A. Escobedo, J. R. Rapp, its transforming potential. Cell 55:979-988.38. T. M. Roberts, and L. T. Williams. 1989. Direct activation of the 48. Roussel, M. F., T. J. Dull, C. W. Rettenmier, P. Ralph, A. serine/threonine kinase activity of Raf-1 through tyrosine phos- Ullrich, and C. J. Sherr. 1987. Transforming potential of the phorylation by the PDGF-j receptor. Cell 58:649-657. c-fins protooncogene (CSF-1 receptor). Nature (London) 325: 30. Nakamura, T. 1992. Structure and function of hepatocyte 549-552. growth factor. Prog. Growth Factor Res. 3:67-86. 49. Rubin, J. S., A. M.-L. Chan, D. P. Bottaro, W. H. Burgess, 31. Nakamura, T., K. Nawa, and A. Ichihara. 1984. Partial purifi- W. G. Taylor, A. C. Cech, D. W. Hirschfield, J. Wong, T. Miki, cation and characterization of hepatocyte growth factor from P. W. Finch, and S. A. Aaronson. 1991. A broad-spectrum serum of hepatectomized rats. Biochem. Biophys. Res. Com- human lung fibroblast derived mitogen is a variant of hepatocyte mun. 122:1450-1459. growth factor. Proc. Natl. Acad. Sci. USA 88:415-419. 32. Nakamura, T., K. Nawa, A. Ichihara, A. Kaise, and T. Nishino. 50. Sporn, M. B., and A. B. Robert. 1985. Autocrine growth factors 1987. Subunit structure of hepatocyte growth factor from rat and cancer. Nature (London) 313:745-747. platelets. FEBS Lett. 224:311-318. 51. Stoker, M., M. Gherardi, and M. Perryman. 1987. Scatter factor 33. Nakamura, T., T. Nishizawa, M. Haglya, T. Seki, M. Shimon- is a fibroblast-derived modulator of epithelial cell mobility. ishi, A. Sugimura, K. Tashiro, and S. Shimizu. 1989. Molecular Nature (London) 327:239-242. cloning and expression of human hepatocyte growth factor. 52. Stoker, M., and M. Perryman. 1985. An epithelial scatter factor Nature (London) 342:440443. released by embryo fibroblasts. J. Cell Sci. 77:209-213. 34. Naldini, L., E. Vigna, R. P. Narsimhan, G. Gandino, R. Zarne- 53. Tajima, H., 0. Higuchi, K. Mizuno, and T. Nakamura. 1992. gar, G. K. Michalopoulos, and P. M. Comoglio. 1991. Hepato- Tissue distribution of hepatocyte growth factor receptor and its cyte growth factor (HGF) stimulates the tyrosine kinase activity exclusive down-regulation in a regenerating organ after the 5158 RONG ET AL. MOL. CELL. BIOL.

injury. J. Biochem. 111:401-406. the identity of human scatter factor and human hepatocyte 54. Ullrich, A., and J. Schlessinger. 1990. Signal transduction by growth factor. Proc. Natl. Acad. Sci. USA 88:7001-7005. receptors with tyrosine kinase activity. Cell 61:203-212. 57. Weidner, K. M., J. Behrens, J. Vandekerckhove, and W. Birch- 55. Velu, T. J., L. Benguinot, W. C. Vass, M. C. Willingham, G. T. meier. 1990. Scatter factor: Molecular characteristics and effect Merline, I. Pastan, and D. R. Lowy. 1987. Epidermal growth on the invasiveness of epithelial cells. J. Cell Biol. 111:2097- factor-dependent transformation by a human EGF receptor 2108. protooncogene. Science 237:1408-1410. 58. Williams, J. F., D. A. McClain, T. J. Dull, A. Ullrich, and M. 56. Weidner, K M., N. Arakaki, G. Hartmann, J. Vanderkerck- Olefsky. 1990. Characterization of an insulin receptor mutant hove, S. Weingart, H. Rieder, C. Fonatsch, H. Tsubouchi, T. lacking the subunit processing site. J. Biol. Chem. 265:8463- Hishida, Y. Daikuhara, and W. Birchmeier. 1991. Evidence for 8469.