Clin. Exp. Metastasis, 1998, 16, 253–265

11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

C-met activation is necessary but not sufficient for liver colonization by B16 murine melanoma cells Shuo Lin, Dario Rusciano, Patrizia Lorenzoni, Guido Hartmann*, Walter Birchmeier*, Silvia Giordano†, Paolo Comoglio† and Max M. Burger Friedrich Miescher Institut, Basel, Switzerland, *Max Delbrück Centrum, Berlin Buch, Germany, † Istituto per la ricerca e la curo del cancro, Divisione di Oncologia Molecolare, Torino Italy. (Received 28 July 1997; accepted in revised form 22 September 1997)

Metastasis to the liver is a frequent event in clinical oncology, the molecular mechanisms of which are not fully understood. We have recently reported a consistent overexpression of c-met in B16 melanoma cells selected in vivo for enhanced liver metastatic ability. In this study we address the question as to whether constitutive activation of c-met is a necessary and sufficient condition for enhanced liver colonization by B16 melanoma cells. Different levels of c-met expression and/or activation in B16 cells were achieved by subcloning, or by c-DNA transfection with either HGF/SF or the oncogenic form of c-met (tpr-met). Metastatic ability of the different populations was then evaluated in vivo by the lung colonization (experimental metastasis) assay. Results indicate that c-met (but not tpr-met) activation in B16 melanoma cells may increase their liver colonizing potential, probably by enhancing motility and invasion in response to paracrine interactions with its ligand. C-met expression per se, however, is not able to change the organ specificity of the cells. C-met activation appears instead to be required at later stages of liver colonization by B16 melanoma cells, in order to enhance their site–specific metastatic ability. © 1998 Lippincott-Raven Publishers Keywords: B16 melanoma, c-met, HGF/SF, liver metastasis

Introduction The liver is the most common site for blood-borne metastatic disease. Liver metastases are seen in more than 40% of gastrointestinal tumors, 24% of melanomas, 15% of lung cancers and 4% of breast cancers [1]. Cancer metastases are usually derived from a specialized population of cells within the primary tumor [2]. However, experimental evidence suggests that malignant cells rarely possess all the necessary properties to give rise to metastatic colonies autonomously; rather, they may derive from specific interactions with normal host cells those Address correspondence to: D. Rusciano, Friedrich Miescher Institut, P.O. Box 2543, CH-4002 Basel, Switzerland. Tel: (+41) 61 6978571; Fax: (+41) 61 6973976; E-mail: [email protected].

© 1998 Lippincott-Raven Publishers

activities which assist accomplishment of the metastatic cascade [3,4]. Therefore, the organ colonization pattern of a given tumor will depend on the one hand on the specific properties of metastatic cells, and on the other hand on how this repertoire can be complemented by host cells at different organ sites [5]. Hepatocyte Growth Factor/Scatter Factor (HGF/SF) was originally characterized as a potent mitogen for hepatocytes [6], and a humoral factor involved in hepatic regeneration [7]. Further studies have revealed that HGF/SF is widely distributed in the body, where it can affect virtually every tissue [8]. The cellular receptor for HGF/SF has been identified as the product of the c-met protooncogene [9]. Further studies have shown that ligand-mediated Clinical & Experimental Metastasis Vol 16 No 3

253

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

stimulation of c-met may trigger multiple activities in target cells through activation of its tyrosine kinase domain [10,11]. Growth, motility and invasion, morphogenesis, and angiogenesis are all influenced by c-met activation [8], thus making it a suitable candidate for the regulation of the metastatic phenotype. In fact, activation of an autocrine loop in NIH 3T3 cells (already expressing HGF/SF) after transfection of c-met, is able to induce an invasive, tumorigenic and metastatic phenotype in the recipient clones [12–14]. In the B16 melanoma model system we have reported that selection for enhanced liver colonization resulted in a cell line (B16–LS9) with adhesion and organ distribution properties very similar to either the parental or the lung-specific counterpart (B16–F10). However, B16–LS9 had different growth properties, which appeared to allow efficient growth primarily in the liver [15]. Next, we described that selection for liver (but not for lung) colonization ability consistently resulted in cell populations overexpressing a constitutively active c-met [16]. Constitutive activation of c-met in liver-specific B16–LS9 cells was not consequent upon the activation of an autocrine loop; it rather resulted from a higher density of receptor molecules on the plasma membrane [17]. However, despite the presence of constitutively active c-met, B16–LS9 cells were still responsive to HGF/SF stimulation, and their response appeared to be enhanced compared with the parental line B16–F1, expressing lower levels of c-met [16]. This indicates that paracrine interactions may occur in vivo, where they might influence the organotropism of B16 cells. To address the question as to what extent the enhanced liver colonizing ability of B16–LS9 cells was dependent on their overexpression of a constitutively active c-met, we chose three different approaches. To analyze the effects of different c-met expression levels in B16 cells, we cloned both B16–F1 and B16–LS9 cell populations, and characterized the clones for c-met expression and activation, and their organ colonizing ability. Next, we analyzed the organ colonization pattern of parental B16–F1 cells transfected with either HGF/SF (to activate an autocrine loop), or with the oncogenic form of c-met (tpr-met), which lacks the extracellular and transmembrane domains, so that its constitutive activation is completely independent of the presence of the ligand [18]. Results indicated that c-met activation by itself could enhance metastatic potency of B16 cells either to the lung or to the liver, depending on the genetic background in which it was acting (B16–F1 or –LS9), while a clear influence on the organotropism of B16 cells was not evident. 254 Clinical & Experimental Metastasis Vol 16 No 3

Materials and methods Cell culture and cloning B16 melanoma cells were routinely cultured in DMEM with 10% FCS and antibiotics (penicillin and streptomycin), in a humidified incubator at 37°C and 10% CO2. Subculturing was done two to three times per week, detaching the cells by a short (5 min) incubation at 37°C in Ca2+ and Mg2+-free PBS (CMF–PBS) containing 0.02% EDTA. Cell cultures were free of mycoplasma contamination, as detected by Hoechst staining. Cloning was performed by limiting dilution, plating an average of 0.5 cells/well in 96 multiwell plates. Clones developing from a single cell were expanded and conserved as frozen stocks for further analysis. Seventy-nine B16–F1 and 81 B16–LS9 clones were characterized for c-met expression and tyrosine phosphorylation, as described below. Cell extracts and Western blotting Cell monolayers were prepared by plating cells at a density of 8 ´ 104/cm2 in normal culture medium. Fifteen to 20 h later, cells were extracted by a 20 min incubation with cold CHAPS buffer (5mM MgCl2, 1 mM EGTA, 100 mM NaCl, 10% glycerol, 1% CHAPS w/v, 25 mM HEPES pH 7.4) containing orthovanadate 1 mM and protease inhibitors (25 mg/ml of both aprotinin and leupeptin, Sigma; 1 mM AEBSF, Calbiochem). Extracts were clarified by centrifugation, diluted with reducing Laemmli’s sample buffer and boiled. Proteins were determined in separate aliquots by the DC–Lowry assay (Bio Rad). Similar amounts of proteins were separated by SDS–PAGE and immunoblotted onto PVDF membranes (Immobilon, Millipore) with specific antibodies (anti-phosphotyrosine monoclonal antibody 4G10 was a kind gift of Kurt Ballmer, FMI; rabbit polyclonal antibody against c-met was obtained from Santa Cruz Biotechnology). Peroxidase-labeled secondary antibodies (DAKO) were detected by the luminol reaction (Amersham) on Hyperfilm (Amersham). Transfection with HGF/SF Two million B16–F1 cells were transfected with 10 mg of the expression plasmid pBat–SF/HGFtag containing the cDNA coding for the human HGF/SF [19], and co-transfected with 1 mg of pSV2-neo, containing the selectable marker neomycin resistance, by the DNA–calcium phosphate method. Controls received pSV2-neo alone. Twenty-four h after transfection, cells were placed in selective

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

medium containing 2 mg/ml G418 (Sigma). After 14 days 20 G418 resistant cell colonies were cloned using cloning cylinders. The expression of HGF/SF in these clones was tested by measuring the scattering activity of supernatants on MDCK cells [19]. One clone with a scattering activity of 10 scatter units/ml (F1–51) and two pools of clones with scattering activities between 1 and 3 scatter units/ml (F1–P1) or below 1 scatter unit/ml (F1–P2) were chosen for further analysis. Expression of HGF/SF by the above cell lines was also evaluated by Northern (not shown) and Western blot analysis. Conditioned medium produced under serum-free conditions by each of the lines was partially purified by heparin-sepharose chromatography. The heparinbound fraction was loaded on a SDS polyacrylamide gel under non-reducing conditions, and immunoblotted to PVDF with specific anti-HGF/SF antibodies (kindly donated by Ermanno Gherardi, Cambridge). The biological activity of secreted HGF/SF on B16 melanoma cells was confirmed by its ability to stimulate directed cell motility of B16–LS9 cells in a Boyden chamber assay, as described [16]. Transfection with tpr-met The tpr-met construct, cloned in the pLXSN vector, which also contains the selectable neomycin marker, was transfected into B16–F1 cells using the DNA–calcium–phosphate co-precipitation procedure (CellPhect Transfection Kit, Pharmacia). Cells were selected for 15 days with 2.0 mg/ml of G418, resistant cells were pooled, and the presence of the tpr-met protein at 65 kD was confirmed by the autokinase assay. Proteins from transfected and parental B16–F1 cells were extracted with either DIM buffer (50 mM PIPES, 300 mM Saccarose, 100 mM NaCl, 5 mM EGTA, 5 mM MgCl2, 0.1 mM ZnCl2, 1 mM orthovanadate, pH 7.4) containing 1% Triton-X100, or CHAPS buffer, as indicated in the figure legend, and precipitated with monoclonal antibodies against the cytoplasmic tail of the human c-met receptor [20], coupled to Protein A–Sepharose. Immunocomplexes were washed twice with the appropriate buffer (DIM or CHAPS), and once with phosphorylation buffer (25 mM HEPES, pH 7.2, 100 mM NaCl, 5 mM MgCl2). The kinase assay was performed in 20 ml of phosphorylation buffer, containing 10 mCi [g–32P]ATP (specific activity 7000 Ci/mmole; Amersham). The reaction was carried out for 10 min at 37°C, and stopped by adding concentrated boiling Laemmli’s buffer. Eluted proteins were subjected to 10% SDS–PAGE followed by autoradiography.

Tumorigenicity and organ colonization assays B16 cell lines and clones were detached when close to confluency by EDTA treatment (see above) taking care to obtain a single cell suspension with more than 95% viable cells (as judged by trypan blue exclusion), resuspended in CMF–PBS at 2 ´ 105/ml, and 0.25 ml (containing 5 ´ 104 cells) of the suspension was slowly injected in the tail vein of syngeneic C57Bl/6 male mice through a 25-guage needle. Three weeks later, animals were sacrificed, and internal organs carefully examined for the presence of tumor colonies with the aid of a dissecting microscope (Zeiss). For the tumorigenic assay, cells were prepared as above, and 105 cells in 100 ml were injected subcutaneously. Tumor growth was evaluated by weighing the excised tumor 20 days after cell inoculation. When required, s.c. tumors or organ colonies were also explanted into tissue culture by finely mincing the tumor in culture medium. After no more than two passages in vitro, extracts were prepared in CHAPS buffer and analyzed for c-met expression and tyrosine phosphorylation (see above). Animal care conformed to NIH guidelines. Statistical analysis A two-tailed Mann–Whitney rank sum test was used to compare animal groups injected with different cell lines [21]. Two-tailed Student’s t-test was used to evaluate cell response in vitro. Significance of data is indicated as follows: *P , 0.05; **P , 0.005; ***P , 0.001; ****P , 0.0005.

Results Colonizing potential of B16 clones Clones derived from parental B16–F1 and liver-specific B16–LS9 cells were characterized by Western immunoblotting for their relative c-met expression and its constitutive activation, as inferred from tyrosine phosphorylation levels [16]. Several clones were chosen for further studies, and Figure 1 shows their characterization. It appears that at relatively low or high levels of c-met expression (lower panels) there is a corresponding amount of c-met activation and tyrosine phosphorylation (upper panels), consistent with our previous finding that c-met represents the main tyrosine kinase activity in these cells [16]. However, at intermediate levels of c-met expression (such as in F1 clones 6D4, 7H3, 7C2 and 5F10), its tyrosine phosphorylation appears to be variable, suggesting that there is a threshold for phosphorylation, which might depend on the one hand on the membrane density of the receptor, which is critical for constitutive activation [17], and on the other hand on Clinical & Experimental Metastasis Vol 16 No 3

255

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 1. C-met expression and activation in B16–F1 and B16–LS9 clones. Thirty mg of total cell extracts from different clones were separated on a 6% SDS–PAGE under reducing conditions, blotted to PVDF and immunostained with antiPY monoclonal antibody 4G10 (upper panels), or with anti–met peptide polyclonal antibodies (bottom panels). The identity of the phosphorylated band at 145 kDa as the b chain of c-met (here indicated by the arrowhead on the left) was demonstrated previously [15]. The phosphorylated band cluster visible below c-met spans between 110 and 130 kDa. Both the uncleaved c-met at 170 kDa, and the b chain at 145 kDa are detected by a polyclonal antibody against the carboxy-terminal peptide of the murine form of c-met (bottom panel). The relative amount of c-met expression and activation of different F1 clones is indicated by the number of ‘+’. LS9 cells and derived clones are all considered ++++.

the activity of a cytosolic tyrosine phosphatase which appears to be involved in regulating c-met activation [17]. C-met expression appeared to be more heterogeneous in parental B16–F1 than in liver–specific B16–LS9 cells, since we could identify clones with different c-met expression levels among B16–F1 cells (left panels), whilst only clones with high c-met expression were found amongst B16–LS9 cells (right panels). Therefore, the effects of c-met expression on the organ colonizing ability of B16 cells could be better evaluated with B16–F1 than with B16–LS9 clones. Ten F1 clones characterized by different amounts of c-met as shown in Figure 1, and five randomly chosen B16–LS9 clones (underlined in Figure 1) all with high c-met expression, were tested for tumorigenicity and organ colonizing ability. All clones were able to produce primary subcutaneous (s.c.) tumors, with no highly significant differences amongst B16–F1 or B16–LS9 derived clones. Tumor growth appeared to be independent of c-met expression levels and reflected the behavior of the parental population (Figure 2A). In fact, B16–LS9 cells and clones grew very poorly after s.c. implantation, with most of the tumors after 3 weeks weighing less than 256 Clinical & Experimental Metastasis Vol 16 No 3

1 g, while the majority of tumors derived from B16–F1 cells and clones were greater than 1 g, consistent with previous reports [15]. Clones were also injected intravenously (i.v.) into syngeneic C57 Bl/6 mice to evaluate their organ colonizing ability. As expected, parental B16–F1 cells gave few colonies in both lung and liver, while liver-selected B16–LS9 cells colonized the liver with high efficiency and the lung very poorly [15]. B16–F1 clones with low c-met expression (+) resembled the parental population, colonizing both lung and liver with low efficiency. Clone 6D4 showed an anomalous behavior both in its c-met expression and phosphorylation (despite a relatively high level of expression, phosphorylation was very low) and tumorigenic properties, since it was tumorigenic after s.c. implantation, but not after tail vein injection. B16–F1 clones with higher c-met expression (++/+++) showed on average an increased efficiency in lung colonization, but were incapable of producing a detectable amount of liver colonies. B16–LS9 clones, all expressing high levels of c-met (++++), conserved the phenotype of parental B16–LS9 cells, giving a high number of liver colonies and a low number of lung colonies. Clone 4D11 was

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 2. Tumorigenicity and organ colonizing potential of B16 clones. (A). Tumorigenicity of each clone was evaluated by injecting 105 cells s.c. and weighing resulting tumors 3 weeks later. The total number of injected mice is indicated in parentheses. (B,C) Organ colonizing ability of each clone in the lung (B) and in the liver (C) was estimated after tail vein injection of 5 ´ 104 cells and colony counting 3 weeks later. Statistical significance of the data was assessed as described in Materials and methods. Each void symbol (0) represents a mouse with a number of colonies as indicated on the ordinate axis. Horizontal bars indicate the average number of colonies per organ per clone (written in close proximity to the bar). The relative amount of c-met expression and activation of the different clones is reported at the bottom line of each figure.

a notable exception, which produced the two s.c. tumor growths over 4 g in weight, and colonized the lung much better than the liver. Therefore, c-met expression correlated better with the general efficiency of organ colonization than with its specificity. Figure 3 shows further analysis of c-met expression and activation in B16–F1 clones explanted in tissue culture from s.c. tumors and organ colonies. B16–F1 clones 2D6, 6D6 and 6G3 stably expressed low levels of c-met in vitro; however, tumor-derived cells consistently overexpressed c-met (bottom panels). In cells derived from liver colonies (and to a lesser extent in cells derived from lung colonies or s.c. primary tumors) c-met also showed an elevated degree of constitutive activation, as inferred from tyrosine phosphorylation (upper panels). Results with B16–F1 clone 6H6 illustrate the tendency observed with other B16–LS9 and B16–F1 clones (not shown) with high c-met expression. The elevated levels of c-met expression and activation were conserved both in lung colonies and s.c. primary tumors (no liver colonies were produced by B16–F1 clones with high c-met expression). Organ colonization potential of B16–F1 cells expressing HGF/SF Three different B16–F1 cell sublines were derived following transfection with HGF/SF cDNA: F1–51 is a clonal population expressing the highest amount of factor; F1–P1 is a pool of clones with moderate expression of scattering activity; F1–P2 is a pool of neomycin-resistant clones with almost undetectable factor expression. Figure 4A illustrates the amount of HGF/SF secreted in conditioned medium by each of the sublines, thus confirming at the protein level that clone F1–51 was the best producer, F1–P1 secreted lower but detectable amounts, while undetectable amounts of factor were present in F1–P2 conditioned medium. Similar results were also

obtained by Northern blot analysis (not shown). The secreted factor was biologically active, as demonstrated by its chemotactic activity on B16–LS9 cells (not shown), which are known to scatter and move in response to HGF/SF [16]. Moreover, isolated colonies developing in vitro from F1–51 and F1–P1 cells had a more scattered morphology compared with parental B16–F1 or F1–P2 cell-derived colonies Clinical & Experimental Metastasis Vol 16 No 3

257

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 3. C-met expression and activation in cells explanted in vitro from tumors growing at different sites. Thirty mg of extracts from cells established in tissue culture from colonies in liver (Li), lung (Lu), lymph nodes (Ly), kidney (Ki), or from primary subcutaneous tumors (Sc) were separated on a 6% SDS–PAGE under reducing conditions, blotted to PVDF and immunostained with anti-PY monoclonal antibody 4G10 (upper panels), or with anti-met peptide polyclonal antibodies (bottom panels). The same amount of extract from original clones and from B16–LS9 and –F1 populations were also loaded as internal references for each blot.

(not shown), thus indicating an autocrine effect of the factor secreted in the medium. Figure 4B shows the analysis of c-met expression and activation in the different sublines. It is interesting to observe that, as a consequence of cloning, different amounts of c-met were expressed by these sublines (lower panel). F1–51 and F1–P1 cells, despite expressing respectively less and similar amounts of c-met compared with parental F1 cells (lower panel), showed a detectable activation of the b subunit of c-met at 145 kDa (upper panel), due to autocrine stimulation. F1–P2 expressed the highest level of cmet and the lowest level of HGF/SF, thus suggesting that most c-met activation is not due to activation of an autocrine loop, but most likely due to the high receptor density on the membrane [17]. All three sublines were tumorigenic after s.c. inoculation (Figure 4C). Primary s.c. tumor growth was fast and efficient for F1–51 and F1–P1, with most of the 258 Clinical & Experimental Metastasis Vol 16 No 3

tumors after 3 weeks weighing at least 1 g. In contrast, tumors produced by F1–P2 grew somewhat slower, with a growth rate closer to that observed for B16–LS9 than for B16–F1, with 57% of the tumors still less than 1 g at the end of the experiment. The organ colonization after tail vein injection is shown in Figure 5A and B. In contrast to B16–F1 (and comparable with B16–LS9) all three sublines colonized the liver somewhat better than the lung. Both F1–51 and F1–P1 were poor lung colonizers (Figure 5A), and did not show any significant increase in their liver colonization ability with respect to parental B16–F1 cells (Figure 5B). F1–P2, on the other hand, colonized the liver significantly better than parental B16–F1 (P = 0.0006), but still to a lesser extent than B16–LS9 cells (P = 0.001). Therefore, no straightforward correlation could be found between autocrine production of HGF/SF and

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 4. Phenotype of B16–F1 cells transfected with the human HGF/SF gene. (A) Secretion of HGF/SF in conditioned medium. Serum-free conditioned media from the indicated cell sublines were partially purified on heparinsepharose, eluates loaded on a non-reducing 10% SDS–PAGE gel, and immunoblotted to PVDF with anti-HGF/SF antibodies. (B). C-met expression and activation in different sublines. Thirty mg of total cell extracts from different sublines were evaluated by immunoblotting as described in the legend to Figure 1. (C). Tumorigenicity of different sublines were evaluated by tumor weight 3 weeks after subcutaneous injection of 105 cells. The numbers of injected animals are indicated in parentheses. Statistical significance of the data was assessed as described previously. The relative expression of both c-met and HGF/SF is reported on the bottom line.

liver colonization ability. Further analysis of HGF/SF production in cells explanted in tissue culture from tumors or organ colonies (Figure 5C) showed that expression of the exogenous cDNA was better conserved in liver colonies than in tumors growing at other sites. This was most notable in s.c tumors, since the size of lung colonies only allowed the evaluation of one colony. Even in the case of F1–P2 cells, originally producing undetectable amounts of factor (Figure 4A), a clear positivity could be seen only in cells derived from a liver colony. Therefore, these data suggest that although activation of the autocrine

loop in B16 cells is not enough to determine a clear shift towards liver colonization, it can alter their colonization ability, and might be helpful at later stages of colony growth, particularly in the liver. Organ colonization of B16–F1 cells expressing tpr-met The expression level of functional tpr-met in the selected B16–F1 cell population was determined by the autokinase assay. Figure 6A (upper panel) shows the activity measured in transfected cells (F1–TPR) in comparison with parental B16–F1 cells (showing, Clinical & Experimental Metastasis Vol 16 No 3

259

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 5. Organ colonizing potential of B16–F1 sublines expressing different amounts of HGF/SF. Organ colonization ability in the lung (A) and in the liver (B) was estimated after tail vein injection as described in the legend to Figure 2. Statistical significance of the data was assessed as described previously. Symbols used are described in the legend to Figure 1. (C). Production of HGF/SF in conditioned media by cells explanted in vitro from colonies in liver (li) and lung (lu), or from primary subcutaneous tumors (sc), was evaluated after partial purification on heparin–sepharose and immunoblotting as described in the legend to Figure 4A.

as expected, no detectable activity), and a positive control represented by Fischer rat fibroblasts transfected with the same tpr-met cDNA. Expression of c-met in transfected B16–F1 cells is shown in the bottom panel of Figure 6A, and appears not to be changed with respect to parental B16–F1 cells. The presence of an active tpr–met resulted in an increased motility of the transfected B16–F1 population (Figure 6B), similar to the effect of a constitutively active c-met in B16–LS9 cells. Tumor growth after s.c. implantation was more comparable with B16–F1 than B16–LS9, with all the tumors reaching at least 1 g within 3 weeks (Figure 6C). Organ colonization analysis after tail vein injection is shown in Figure 7A and B. F1–TPR 260 Clinical & Experimental Metastasis Vol 16 No 3

colonized the liver somewhat better than the lung. Both organs were, however, colonized with an efficiency similar to the parental B16–F1 cell population. Further evaluation of tpr-met activity in cells explanted in vitro from tumors and colonies growing at different sites (Figure 7C) indicated that expression of the exogenous cDNA was conserved at all sites, although colonies developing in the lung usually expressed increased amounts with respect to the parental B16–F1 population. This apparent selection in the lung for cells with higher tpr-met activity is in agreement with the observation that an increase of c-met constitutive activation in B16–F1 clones appeared to enhance lung, but not liver colonization (Figure 2).

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Figure 6. Phenotype of B16–F1 cells transfected with human tpr–met cDNA. (A) (Upper panel) Tpr-met activity in parental and transfected B16–F1 cells was evaluated in cell extracts prepared in DIM buffer, 1% Triton-X100, by an autokinase assay. Fischer rat fibroblasts transfected with the same tpr-met cDNA were used as a positive control. This cell line was specifically chosen because of its very low level of spontaneously transformed foci. (Lower panel) Cmet expression was detected in 30 mg of a cell extract by immunoblotting as described in the legend to Figure 1. (B) Migratory activity of transfected B16–F1 cells was measured in 8 mm pore Transwell chambers after 48 h in the absence of any chemotactic agent. The number of cells crossing the filter was estimated by crystal violet staining. Statistical comparison of the data was done by the Student’s t-test, and significance is indicated by the number of asterisks (Materials and methods). (C) Tumorigenicity of transfected cells was evaluated as described previously. Statistical significance was assessed by the Mann–Whitney rank sum test, and is indicated by asterisks as before. The relative level of c-met expression and activation in the different populations is reported on the bottom line.

Figure 7. Organ colonizing potential of B16–F1 cells expressing tpr-met. Organ colonizing ability in the lung (A) and in the liver (B) was estimated after tail vein injection as described in the legend to Figure 2. Symbols are as in Figure 1. (C) Tpr-met expression in cells explanted in vitro from tumor colonies in lung or liver, or from primary subcutaneous tumors was evaluated in a CHAPS extract by the autokinase assay.

Clinical & Experimental Metastasis Vol 16 No 3

261

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

Discussion HGF/SF is ubiquitously distributed in mammalian tissues [22], with high expression levels found in lung, liver and kidney [23,24]. Increased amounts of HGF/SF have been observed during regeneration of these organs [25–27], or in cases of liver disease [28]. HGF/SF accumulation has been detected in the extracellular matrix of cancerous and inflammatory lesions of lung, liver and pancreas [29], while enhanced levels of soluble HGF/SF have been associated with metastasis to the pleura [30] and the liver [31]. On the other hand, HGF/SF receptor (c-met) has been found to be overexpressed in several malignancies [32], thus suggesting that a paracrine interaction between c-met and HGF/SF could have a relevant role in promoting metastatic dissemination, mostly to the lung or the liver. Our recent finding that c-met was consistently overexpressed only in liver–selected B16 melanoma cells [16] prompted us to investigate to what extent this overexpression was facilitating the liver-specific phenotype. The analysis of c-met in B16-derived clones showed that B16–F1 cells (which colonize both lung and liver with low efficiency) are widely heterogeneous for c-met expression and activation. Liverselected B16-LS9 cells appear to be much more homogeneous, yielding no clones with low c-met expression (Figure 1), hence giving a preliminary indication that elevated levels of c-met could be specifically required for efficient liver colonization. However, in contrast to colon cancer cells, in which in vitro selection for elevated expression of EGFr resulted in a cell population with enhanced liver colonizing ability [33], further analysis of the organ colonization pattern of B16–F1 and –LS9 clones (Figure 2) indicated that c-met overexpression did not specifically enhance B16 liver colonizing ability, but both lung and liver depending on the genetic background in which it occurs. B16–F1 cells with high c-met expression colonized the lung with increased efficiency, consistent with the higher concentration of HGF/SF present at that site [23, 24, Rusciano, unpublished]. B16–LS9 cells appeared to be directed to the liver by a mechanism different from c-met overexpression. However, once cells reach the liver, elevated levels of c-met expression might facilitate colonization. Accordingly, c-met over-expression has been correlated with tumor progression rather than specific sites of metastatic disease in several human malignancies [34–40]. In the case, though, of a tumor with a high propensity for liver metastasis such as human colorectal cancer, c-met expression has been shown to be dramatically 262 Clinical & Experimental Metastasis Vol 16 No 3

increased more specifically in liver metastatic isolates, thus suggesting a late involvement of c-met activation in hepatic metastasis [41]. A similar late involvement is also suggested by our data, since liver metastases, even when derived from B16–F1 clones with very low c-met expression, tended to display increased levels of expression and constitutive activation of the HGF/SF receptor (Figure 3). This increase was not due to an intrinsic instability of the clonal population, because the amount of c-met expression and activation in B16–F1 clones with low c-met is stable in tissue culture for at least 30 passages (not shown); rather it suggests inductive events (now under investigation) related to tumor cell growth in different environments. Data obtained with B16 clones are further supported by the analysis of tpr–met or HGF/SF transfected B16–F1 cells. Tpr–met was identified by its ability to transform NIH3T3 cells [42], and its presence has been described in gastric cancer [43] and other human tumor cell lines [44]. Transfection of tpr–met cDNA in mouse or rat fibroblast cell lines induced metastatic ability, apparently confined to the lungs [13; Silvia Giordano, personal communication]. B16–F1 transfected with tpr–met showed the same organ colonizing ability as the parental B16–F1 cells (Figure 7A). However, further enrichment for tpr–met expression in metastatic colonies appeared to take place more consistently in the lung than at other sites (Figure 7C), suggesting again that increased c-met activity in B16–F1 cells may favor lung colonization. Tpr-met lacks both the extracellular and transmembrane domains of c-met, thus remaining confined inside the cell. It becomes constitutively activated because of the presence of the tpr’ dimerization motif fused with the cytoplasmic portion of c-met [45]. The fact that an increased activity of the oncogenic form of c-met has little influence on either the organ colonization pattern of B16 cells, or its efficiency, may suggest that specific interactions in the target organ between the c-met receptor and its ligand have a higher impact on the metastatic phenotype, than a constitutive activity which is totally independent of the presence of any external ligand. In fact, activation of an autocrine circuit by transfection with HGF/SF had a greater effect on the colonizing capacity of recipient B16–F1 cells, leading to a decrease in lung colonies, making the liver their preferential target. Of note is the fact that F1–P2 cells, despite their high levels of c-met expression and activation — expected to increase their lung colonization potential (like F1 clones with high c-met) — colonized the liver better than the parental F1 population, thus suggesting an influence

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

of the activation of the autocrine circuit on their organ preference. Moreover, the slower growth of this cell line as primary subcutaneous tumors (Figure 4C) might suggest, similar to B16–LS9 [15], a stronger dependence on paracrine growth stimulation for efficient growth, which results in a more selective pattern of metastatic distribution [5]. Accordingly, clone LS9–4D11, which – alone among LS9 clones – produced the two s.c. growths over 4 g in weight, colonized the lung much better than the liver (Figure 2). Activation of an autocrine loop by transfection of c-met into NIH3T3 fibroblast cells has been shown to be able to induce an invasive [14], tumorigenic [12] and metastatic [13] phenotype. Similarly, potentiation of the autocrine loop in a human sarcoma cell line (SK–MLS–1) by transfection of HGF/SF resulted in enhancement of tumorigenicity and acquisition of the metastatic phenotype [46]. These results were, however, obtained in cell lines which were not (NIH3T3) or were only poorly (SK–MLS–1) tumorigenic in the nude mouse model system. Accordingly, overexpression of other growth factors or their receptors in experimental systems has been reported to induce a malignant phenotype in recipient cells which originally were not, or were only marginally malignant. NIH3T3 cells were rendered tumorigenic and metastatic after transfection of either secreted forms of bFGF [47,48], or cerbB2/neu [49]. The latter was also able to enhance the metastatic potential of human SCLC cells [50], and to induce malignant mammary tumors in transgenic mice [51]. B16 melanoma cells, on the contrary, are already tumorigenic and metastatic in syngeneic mice, and are able to colonize both the lung and the liver [15]. It might thus be conceivable that further activation in these cells of a growth factor receptor like c-met, which is already expressed at detectable levels by parental B16–F1 cells (Figure 1), may not have the same dramatic effects on their metastatic phenotype as is observed in nonmetastatic or poorly metastatic cell lines in which receptor activation is newly induced. Taken together, results presented here indicate that constitutive activation of c-met (but not tpr–met) enhances the metastatic efficiency of B16 melanoma cells (probably through a further paracrine interaction with its cognate ligand in the lung or in the liver). It is not, however, the leading cause of liver targeting, although it seems to be required at later stages in the development of liver colonies. Other factors, as yet unknown, are likely to cooperate with c-met to induce liver specificity in the B16 melanoma model. Therefore, when c-met activation occurs in

the right genetic background, such as in B16–LS9 cells, it may contribute to their organotropism, giving them a specific advantage for liver colonization, while reducing their growth ability at other sites [15; Rusciano, unpublished]. Induction of cell motility and invasion [16], and modulation of the growth rate (Rusciano, unpublished) are obvious candidates (now under investigation) for the specific effects exerted by hypersensitivity to HGF/SF stimulation in B16–LS9 cells.

Acknowledgements This work was supported by the E.C. Grant no. BMH1–CT94–1651, and the BBW grant N. 94.0130 awarded by the Swiss Federal Office for Education and Science.

References 1. Kemeny NE, Kemeny M and Lawrence TS. 1995, Liver metastasis. In: Abeloff MD, Armitage JO, Lichter AS and Niederhuber JE, eds. Clinical Oncology. New York: Churchill Livingstone, pp. 679–707. 2. Fidler IJ and Hart IR. 1982, Biological diversity in metastatic neoplasms: origins and implications. Science, 217, 998–1003. 3. Sargent NSE, Oestreicher M, Haidvogl H, Madnick HM and Burger MM. 1988, Growth regulation of cancer metastases by their host organ. Proc Natl Acad Sci USA, 85, 7251–55. 4. Nicolson GL. 1991, Tumor and host molecules important in the organ preference of metastasis. Sem Cancer Biol, 2, 143–54. 5. Rusciano D and Burger MM. 1992, Why do cancer cells metastasize into particular organs? BioEssays, 14, 185–94. 6. Nakamura T, Nishizawa T, Hagiya M et al. 1989, Molecular cloning and expression of human hepatocyte growth factor. Nature, 342, 440–43. 7. Michalopoulos GK and Zarnegar R. 1992, Hepatocyte growth factor. Hepatol, 15, 150–55. 8. Zarnegar R and Michalopoulos GK. 1995, The many faces of Hepatocyte Growth Factor, from hepatopoiesis to hematopoiesis. J Cell Biol, 129, 1177–80. 9. Bottaro DP, Rubin JS, Faletto DL et al. 1991, Identification of the hepatocyte growth factor receptor as the c-met protooncogene product. Science, 251, 802–4. 10. Weidner KM, Sachs M and Birchmeier W. 1993, The met receptor tyrosine kinase transduces motility, proliferation and morphogenetic signals of scatter factor/hepatocyte growth factor in epithelial cells. J Cell Biol, 21, 145–54. 11. Komada M and Kitamura N. 1993, The cell dissociation and motility triggered by scatter factor/hepatocyte growth factor are mediated through the cytoplasmic domain of the c-met receptor. Oncogene, 8, 2381–90. Clinical & Experimental Metastasis Vol 16 No 3

263

S. Lin et al. 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

12. Rong S, Bodescot M, Blair D et al. 1992, Tumorigenicity of the met proto-oncogene and the gene for hepatocyte growth factor. Mol Cell Biol, 12, 5152–58. 13. Rong S, Segal S, Anver M, Resau JH and Vande Woude GF. 1994, Invasiveness and metastasis of NIH 3T3 cells induced by met-hepatocyte growth factor/scatter factor autocrine stimulation. Proc Natl Acad Sci USA, 91, 4731–35. 14. Giordano S, Zhen Z, Medico E, Gaudino G and Comoglio PM. 1993, Transfer of motogenic and invasive response to scatter factor/hepatocyte growth factor by transfection of human met proto-oncogene. Proc Natl Acad Sci USA, 90, 649–53. 15. Rusciano D, Lorenzoni P and Burger MM. 1993, Paracrine growth response as a major determinant in liver-specific colonization by in vivo selected B16 murine melanoma cells. Invasion Metastasis, 13, 212–24. 16. Rusciano D, Lorenzoni P and Burger MM. 1995, Expression of constitutively activated hepatocyte growth factor/scatter factor receptor (c-met) in B16 melanoma cells selected for enhanced liver colonization. Oncogene, 11, 1979–87. 17 Rusciano D, Lorenzoni P and Burger MM. 1996, Constitutive activation of c-met in liver metastatic B16 melanoma cells depends on both substrate adhesion and cell density and is regulated by a cytosolic tyrosine phosphatase activity. J Biol Chem, 271, 20763–69. 18. Dean M, Park and Wande Woude GF. 1987, Characterization of the rearranged tpr-met oncogene breakpoint. Mol Cell Biol, 7, 921–24. 19. Hartmann G, Naldini L, Weidner KM et al. 1992, A functional domain in the heavy chain of scatter factor/hepatocyte growth factor binds the c-met receptor and induces cell dissociation but not mitogenesis. Proc Natl Acad Sci USA, 89, 11574–78. 20. Prat M, Narsimhan RP, Crepaldi T et al. 1991, The receptor encoded by the human c-met oncogene is expressed in hepatocytes, epithelial cells and solid tumors. Int J Cancer, 49, 323–28. 21. Leoncini L, Pacenti L, Rusciano D et al. 1988, Correlation between differentiation and lung colonization by retinoic acid-treated F9 cells as revealed by the expression pattern of extracellular matrix and cell surface antigens. Am J Pathol, 130, 505–14. 22. Wolf HK, Zarnegar R and Michalopoulos GK. 1991, Localization of hepatocyte growth factor in human and rat tissue. Hepatol, 14, 488–94. 23. Tashiro K, Hagiya M, Nishizawa T et al. 1990, Deduced primary structure of rat hepatocyte growth factor and expression of the mRNA in rat tissue. Proc Natl Acad Sci USA, 87, 3200–4. 24. Kagoshima M, Kinoshita T, Matsumoto K and Nakamura T. 1992, Developmental changes in hepatocyte growth factor mRNA and its receptor in rat liver, kidney and lung. Eur J Biochem, 210, 375– 380. 25. Zarnegar R, DeFrances MC, Kost DP, Lindroos P and Michalopoulos GK. 1991, Expression of hepatocyte growth factor mRNA in regenerating rat liver after partial hepatectomy. Biochem Biophys Res Commum, 177, 559–65. 26. Yanagita K, Matsumoto K, Sekiguchi K et al. 1993, Hepatocyte growth factor may act as a pulmotrophic 264 Clinical & Experimental Metastasis Vol 16 No 3

27.

28. 29.

30.

31. 32. 33.

34.

35.

36.

37. 38.

39. 40.

41.

42. 43.

factor on lung regeneration after acute lung injury. J Biol Chem, 268, 21212–17. Kawaida K, Matsumoto K, Shimazu H and Nakamura T. 1994, Hepatocyte growth factor prevents renal failure and accelerates renal regeneration in mice. Proc Natl Acad Sci USA, 91, 4357–61. Tomiya T, Nagoshi S and Fujiwara K. 1992, Significance of serum hepatocyte growth factor levels in patients with hepatic failure. Hepatol, 15, 1–4. Yoshinaga Y, Matsuno Y, et al. 1993, Immunohistochemical detection of hepatocyte growth factor/scatter factor in human cancerous and inflammatory lesions of various organs. Jpn J Cancer Res, 84, 1150–58. Eagles G, Warn A, Ball RY et al. 1996, Hepatocyte growth factor/scatter factor is present in most pleural effusion fluids from cancer patients. Br J Cancer, 73, 277–381. Taniguchi T, Toi M, Inada K et al. 1995, Serum concentrations of hepatocyte growth factor in breast cancer patients. Clin Cancer Res, 1, 1031–34. Di Renzo MF, Narshiman RP, Olivero M et al. 1991, Expression of the Met/HGF receptor in normal and neoplastic human tissues. Oncogene, 6, 1997–2003. Radinsky R, Risin S, Fan D et al. 1995, Level and function of epidermal growth factor receptor predict the metastatic potential of human colon carcinoma cells. Clin Cancer Res, 1, 19–31. Grigioni WF, Fiorentino M, D’Errico A et al. 1995, Overexpression of c-met protooncogene product and raised Ki67 index in hepatocellular carcinomas with respect to benign liver conditions. Hepatol, 21, 1543–46. Natali PG, Nicotra MR, Di Renzo MF et al. 1993, Expression of the C-Met/HGF receptor in human melanocytic neoplasms, demonstration of the relationship to malignant melanoma tumor progression. Br J Cancer, 68, 746–50. Tuck AB, Park M, Sterns EE, Boag A and Elliott BE. 1996, Coexpression of Hepatocyte Growth Factor and receptor (met) in human breast carcinoma. Am J Pathol, 148, 225–32. Di Renzo MF, Olivero M, Katsaros D et al. 1994, Overexpression of the Met/HGF receptor in ovarian cancer. Int J Cancer, 58, 658–62. Hasegawa Y, Yamamoto M, Maeda S and Saitoh Y. 1995, Hepatocyte Growth Factor and its receptor cmet regulate both cell growth and invasion of human pancreatic cancer. Int J Oncol, 7, 877–81. Humphrey PA, Zhu X, Zarnegar R et al. 1995, Hepatocyte Growth Factor and its receptor (c-met) in prostatic carcinoma. Am J Pathol, 147, 386–96. Ferracini R, Di Renzo MF, Scotlandi K et al. 1995, The MET/HGF receptor is overexpressed in human osteosarcomas and is activated by either a paracrine or an autocrine circuit. Oncogene, 10, 739–49. Di Renzo MF, Olivero M, Giacomini A et al. 1995, Overexpression and amplification of the Met/HGF receptor gene during the progression of colorectal cancer. Clin Cancer Res, 1, 147–54. Cooper CS, Park M, Blair DG et al. 1984, Molecular cloning of a new transforming gene from a chemically transformed human cell line. Nature, 311, 29–33. Soman NR, Correa P, Ruiz BA and Wogan GN. 1991, The tpr–met oncogenic rearrangement is present and

C-met and liver metastasis 11111 2 3 4 5 6 7 8 9 10111 1 2 3 4 5 6 7 8 9 20111 1 2 3 4 5 6 7 8 9 30111 1 2 3 4 5 6 7 8 9 40111 1 2 3 4 5 6 7 8 9 50111 1 2 3 4111

44.

45.

46.

47.

expressed in human gastric carcinoma and precursor lesions. Proc Natl Acad Sci USA, 88, 4892–96. Soman NR, Wogan GN and Rhim JS. 1990, Tpr-met oncogenic rearrangement, detection by pcr amplification of the transcript and expression in human tumor cell lines. Proc Natl Acad Sci USA, 87, 738–42. Gonzatti-Haces M, Seth A, Park M et al. 1988, Characterization of the TPR-MET oncogene p65 and the MET protooncogene p140 protein–tyrosine kinases. Proc Natl Acad Sci USA, 85, 21–25. Jeffers M, Rong S and Vande Woude GF. 1996, Enhanced tumorigenicity and invasion–metastasis by HGF/SF-met signalling in human cells concomitant with induction of the urokinase proteolysis network. Mol Cell Biol, 16, 1115–25. Yayon A and Klagsbrun M. 1990, Autocrine regulation of cell growth and transformation by basic fibroblast growth factor. Cancer Metastasis Rev, 9, 191–202.

48. Damen JE, Greenberg AH and Wright JA. 1991, Transformation and amplification of the K–fgf protooncogene in NIH3T3 cells, and induction of metastatic potential. Biochim Biophys Acta, 1097, 103–10. 49. Yu D, Hamada J, Zhang H, Nicolson GL and Hung MC. 1992, Mechanisms of c-erbB2/neu oncogeneinduced metastasis and repression of metastatic properties by adenovirus 5 E1A gene products. Oncogene, 7, 2263–70. 50. Yu D, Wang SS, Dulski KM et al. 1994, C-erbB2/neu overexpression enhances metastatic potential of human lung cancer cells by induction of metastasis–associated properties. Cancer Res, 54, 3260–66. 51. Guy CT, Webster MA, Schaller M et al. 1992, Expression of the neu protooncogene in the mammary epithelium of transgenic mice induces metastatic disease. Proc Natl Acad Sci USA, 89, 10578–82.

Clinical & Experimental Metastasis Vol 16 No 3

265