HGF/SF and its receptor c-MET play a minor role in the dissemination of human B-lymphoma cells in SCID mice

The MET protooncogene, c-MET, encodes a cell surface tyrosine kinase receptor. The ligand for c-MET is hepatocyte growth factor (HGF), also known as scatter factor (SF), which is known to affect proliferation and motility of primarily epithelial cells. Recently, HGF/SF was also shown to affect haemopoiesis. Studies with epithelial and transfected NIH3T3 cells indicated that the HGF/SF–c-MET interaction promotes invasion in vitro and in vivo. We previously demonstrated that HGF/SF induces adhesion of c-MET-positive B-lymphoma cells to extracellular matrix molecules, and promoted migration and invasion in in vitro assays. Here, the effect of HGF/SF on tumorigenicity of c-MET-positive and c-MET-negative human B-lymphoma cell lines was studied in C.B-17 scid/scid (severe combined immune deficient) mice. Intravenously (i.v.) injected c-MET-positive (BJAB) as well as c-MET-negative (Daudi and Ramos cells) B-lymphoma cells formed tumours in SCID mice. The B-lymphoma cells invaded different organs, such as liver, kidney, lymph nodes, lung, gonads and the central nervous system. We assessed the effect of human HGF/SF on the dissemination of the B-lymphoma cells and found that administration of 5 μg HGF/SF to mice, injected (i.v.) with c-MET-positive lymphoma cells, significantly (P = 0.018) increased the number of metastases in lung, liver and lymph nodes. In addition, HGF/SF did not significantly influence dissemination of c-MET-negative lymphoma cells (P = 0.350 with Daudi cells and P = 0.353 with Ramos cells). Thus the effect of administration of HGF/SF on invasion of lymphoma cells is not an indirect one, e.g. via an effect on endothelial cells. Finally, we investigated the effect of HGF/SF on dissemination of c-MET-transduced Ramos cells. In response to HGF/SF, c-MET-transduced Ramos cells showed an increased migration through Matrigel in Boyden chambers compared to wild-type and control-transduced Ramos cells. The dissemination pattern of c-MET-transduced cells did not differ from control cells in in vivo experiments using SCID mice. Also no effect of HGF/SF administration could be documented, in contrast to the in vitro experiments. From our experiments can be concluded that the HGF/SF–c-MET interaction only plays a minor role in the dissemination of human B-lymphoma cells. © 1999 Cancer Research Campaign

five cell lines tested were c-MET-positive). Furthermore, we found that HGF/SF induced adhesion of c-MET-positive (but not of c-MET-negative) human B-lymphoma cells to the extracellular matrix (ECM) molecules fibronectin and collagen. Adhesion appeared to be mediated by the integrins α 4 β 1 and α 5 β 1 . In addition, HGF/SF promoted migration of c-MET-posit (and not of c-MET-negative) human B-lymphoma cells through Matrigel in Boyden chambers and invasion in rat fibroblast monolayers (Weimar et al, 1997). These results suggest that HGF/SF-c-MET interaction may promote dissemination of lymphoma cells in vivo. Here, we investigated the tumorigenicity of both c-MET-positive and c-MET-negative human B-lymphoma cells in severe combined immune deficient (SCID) mice, with and without administration of human HGF/SF. Human HGF/SF significantly promoted dissemination of c-MET-positive lymphoma cells (BJAB cells) and did not affect dissemination of c-MET-negative tumour cells (Ramos cells). To test the hypothesis that HGF/SF promotes the dissemination of c-MET-positive B-lymphoma cells, we transduced c-MET-negative B-cells (Ramos) with the c-MET gene and studied their capacity to migrate in vitro and in vivo. C-MET transduction resulted in an increased migration of c-METtransduced Ramos cells through Matrigel in response to HGF/SF, whereas HGF/SF did not affect migration of wild-type and control-transduced cells. The dissemination pattern of c-METtransduced, control-transduced, or wild-type Ramos cells was investigated in vivo by injecting the cells in SCID mice.

Tumorigenicity establishment of B-lymphoma cells in SCID mice
C.B-17 scid/scid (SCID) mice were bred and maintained at the Animal Department of The Netherlands Cancer Institute. The mice were kept in isolators under specific pathogen-free conditions and used when 6Ð10 weeks old. All experiments were approved by the Animal Experimental Advisory Board of The Netherlands Cancer Institute.
Mice were injected intravenously (i.v.; in the tail vein) with 5 × 10 6 B-cell tumour cells. In the first pilot experiment HGF/SF (5 µg per mouse) was obtained from R&D Systems, Abingdon, UK (produced in Sf21 insect cells) and was mixed with the tumour cells before i.v. injection, while in all other in vivo experiments, HGF/SF was produced in our laboratory (see below), and was injected subcutaneously (s.c.; into the right flank; 5 µg (or in one experiment 25 µg) per mouse per day; 7 ×; starting on the day of the i.v. tumour injection).
After 3Ð5 weeks, mice appeared ill (signs of paralysis, ovarian tumour, loss of weight, etc.) and all mice in the same experiment were sacrificed and examined. Organs were fixed (in 4% formaldehyde), embedded in paraffin after which sections were stained for human B-leucocytes (monoclonal mouse anti-human leucocyte common antigen [LCA], No M701, Dakopats, Glostrup, Denmark).

HGF/SF production
HGF/SF was produced by Spodoptera frugiperda 9 (Sf9) insect cells using the baculo-gold-virus-producing system as described previously (Yee et al, 1993; HGF/SF cDNA was a gift from Prof T Nakamura). In short, Sf9 insect cells were infected with baculovirus containing the HGF/SF gene, and secreted (±2 µg ml Ð1 as measured by enzyme-linked immunosorbent assay (ELISA) biologically active HGF/SF (tested in proliferation assays and colony assays) with human bone marrow cells on which HGF/SF acts as a synergistic proliferative factor with other growth factors (Weimar, 1998), tested in a proliferation assay with c-MET-transfected NIH3T3 cells; tested in a scattering assay with MDCK cells (Hordijk et al, 1997). HGF/SF in supernatant of the Sf9 cells was purified using a heparin column and eluted with 1.5 M sodium chloride. Usually, 5 µg HGF/SF was injected s.c. per mouse per day, which contained 70 ng endotoxin. In some experiments HGF/SF was denaturated (30 min autoclaved) to administer exactly the same amount of endotoxin in both groups (the autoclaved HGF/SF was no longer biologically active as tested in colony assay with human bone marrow cells). Subcutaneous injections of 5 µg HGF/SF daily for 7 days did not affect the mice. No illness or other dysfunctions were observed.

HGF/SF ELISA
HGF/SF was determined by a HGF-ELISA kit (Institute of Immunology Co. Ltd, Tokyo, Japan). HGF/SF detection is based upon a sandwich method using a solid phase coated anti-human HGF/SF mouse monoclonal antibody and a peroxidase-labelled anti-human HGF/SF mouse monoclonal antibody. Amounts from 0.1 ng ml Ð1 were reliably detected according to the linearity of the standard curve.

Construction of the c-MET vector and transduction
The c-MET coding sequence was cloned from the pMB1 plasmid (generous gift from Prof. T Nakamura) by polymerase chain reaction (PCR) with the following primers: 5′ CGCGCTCGAGC-CATGGAGGCCCCCGCTGTGCTTGCA; 3′ GATCCCGGGCA-CTATGATGTCTCCCAGAAGG.
The PCR product was cut with NcoI (introduced with PCR primer; start codon) and SwaI (downstream of the stop-codon), ligated in pSP73 (with Xho/EV), and after ± 140 bp exchanged with wild-type cDNA (without PCR; with SstII and ClaI). c-MET cDNA (4175 bp) was cut from pSP73 (with Xho/ScaI) and ligated between the XhoI and the SnaBI site of the polylinker from the plasmid LZRS-linker-IRES-EGFP (a modified version of GFP with an enhanced expression (obtained from Clontech Laboratories, Palo Alto, CA, USA) (Heemskerk et al, 1997), to obtain the retroviral vector LZRS-c-MET-IRES-GFP, containing LTR and Ψ packaging sequences from the Moloney murine leukaemia virus (Mo-MuLV) as well as puromycin resistance gene. The retroviral packaging Phoenix cells (generous gift from Dr GP Nolan, Stanford University School of Medicine, CA, USA) were transfected using the calcium phosphate transfection system (Gibco-BRL, Breda, The Netherlands) with the c-MET containing vector. Transfected cells were selected by addition of puromycin (2.5 µg ml Ð1 ; Clontech Laboratories) at day 2. Ten to 14 days after transfection, 6 × 10 6 cells were plated per 10-cm petri dish in 10 ml complete medium without pyromycin. The next day, the medium was refreshed and on the following day retroviral supernatant was harvested, centrifuged and frozen (Ð70°C) in aliquots. Control viral supernatant was collected from cells that had been transfected with the control vector (LZRS-IRES-GFP). Viral supernatants (c-MET-GFP and GFP) were preincubated with dotap (10 µg ml Ð1 ; Boehringer Mannheim, Mannheim, Germany) on ice for 10 min, and incubated overnight with Ramos cells. After 5 days the transduced Ramos cells were sorted by FACS-STAR on GFP expression (Mann et al, 1983;Kinsella and Nolan, 1996).

c-MET-staining of transduced cells
Cytospins were prepared of transduced and wild-type Ramos cells, which were fixed with methanol and stained with a polyclonal antibody directed to the last 28 C-terminal amino acids of c-MET (C-28; Santa Cruz; 250 ng ml Ð1 ) or as a control with a polyclonal control IgG (used at the same concentration as C-28: 250 ng ml Ð1 ; Southern Biotechnology Associates Inc, Birmingham, AL, USA). Cytospins were washed and incubated with fluorescein isothiocyanate (FITC)-labelled goat anti-rabbit. After washing, cells were examined for c-MET staining with a fluorescent microscope.

In vitro invasion assay
Boyden chambers (6-well plates; Becton Dickinson) were used to assess tumour cell invasiveness as described previously by Albini et al (1987). Upper and lower chambers of the Boyden chambers were separated by Matrigel-coated filters with 8-µm pores. After rehydration of the Matrigel in IMDM + 0.1% bovine serum albumin (BSA), cells (5 × 10 5 ) were seeded in the upper chambers in 1.5 ml culture medium (IMDM containing L-glutamin, supplemented with 5% FCS and P/S). HGF/SF (25 ng ml Ð1 ; produced by COH cells; gift from Prof. T Nakamura, Division of Biochemistry, Biomedical Research Center, Osaka University Medical School, Japan) was added to the lower chambers in 2.6 ml of the same culture medium as present in the upper wells. The chambers were incubated overnight at 37°C. All cells that had migrated from the upper to the bottom of the lower chamber were counted using an inverted light microscope with a 20 × ocular.

Statistical analysis
Differences in the number of organs of SCID mice invaded by tumour cells between the different groups was determined using the χ 2 test. Differences were considered to be statistically significant at the 5% level. Means ± standard deviation (s.d.) were calculated from proliferation tests.

Production of HGF/SF and effect on SCID mice
HGF/SF was produced by Sf9 insect cells using the baculo-goldvirus-producing system. This HGF/SF was used for the in vivo experiments described below. First the biological activity of purified HGF/SF was tested in the presence of FCS to enhance the conversion of pro-HGF/SF into active HGF/SF. HGF/SF promoted the proliferation of NIH3T3-cells, transfected with c-MET, in a dose-dependent manner ( Figure 1A). Furthermore, purified HGF/SF promoted colony formation of murine and human haemopoietic progenitor cells. HGF/SF acted synergistic with granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin (IL)-3 ( Figure 1B). Purified HGF/SF induced scattering of MDCK (Hordijk et al, 1997). The biological activity of the Sf9-derived HGF/SF was similar to recombinant HGF/SF produced by Dr T Nakamura ( Figure 1B). The purified HGF/SF from the Sf9 cells was applied for further in vivo studies.
HGF/SF injected s.c. in normal Balb/c mice did not induce marked toxicity in these mice as measured by changes in activity of the mice, haematological cell counts or survival (n = 20; until day 30; results not shown). Also, HGF/SF (5 µg once a day s.c. for 7 days) did not induce toxicity in SCID mice (n = 4; observations for up to 2 months; results not shown).

Expression of c-Met, HGF/SF production cell lines
c-MET expression was determined in the lymphoma cell lines Daudi, Ramos and BJAB using reverse transcription PCR (RT-PCR). c-MET mRNA was detected in the cell line BJAB and not in the cell lines Daudi and Ramos (Weimar et al, 1997). HGF/SF could not be detected (by ELISA) in the supernatant of these three cell lines after 1Ð3 days of culture (results not shown). The c-MET-negative Ramos cell line was transduced with a retroviral vector, containing either green fluorescence protein (Ramos-GFP) or the c-METÐIRESÐGFP (Ramos-c-METÐGFP) sequence. After selection, c-MET-transduced Ramos cells could be identified by both green fluorescence expression as well as by c-MET protein expression (Figures 2 and 3). The biological behaviour of the transfected cells differed from the wild-type Ramos (wtRamos) cells as determined in an in vitro migration assay. In comparison to wtRamos cells or control-transduced (with GFP) Ramos cells, the c-METÐGFP-transduced Ramos cells migrated better in response to HGF/SF (Table 1). Ramos cells (5 × 10 5 ) wild-type, control-transduced (with GFP alone), and c-MET-IRES-GFP-transduced (c-MET/GFP) were mounted in 1.5 ml culture medium in the upper wells of Boyden chambers. Upper and lower wells were separated by Matrigel-coated membranes with 8-µm pores. In the lower chambers 2.6 ml culture medium was present with or without HGF/SF (25 ng ml -1 ). After 16 h, all cells that had migrated to the bottom of the lower well were counted with a 20 × ocular of an inverted microscope.
These in vitro results suggest that the c-METÐHGF/SF interaction promotes dissemination of lymphoma cells in vivo, like has been observed for transfected NIH3T3 cells and several epithelial cells (Cooper et al, 1986;Stoker et al, 1987;Iyer et al, 1989;Rong et al, 1992Rong et al, , 1993Rong et al, , 1994Liu et al, 1992;Giordano et al, 1993;Bellusci et al, 1994).

Effect of HGF/SF on tumorigenicity of c-MET-positive Blymphoma cells in SCID mice
SCID mice were injected i.v. with 5 × 10 6 BJAB cells. HGF/SF was administered s.c. once daily for 7 days. BJAB cells gave tumours in lymph nodes, kidneys, lungs and gonads (Table 2). More mice developed tumours in the different organs when treated with HGF/SF (5 µg per day). Pooling of the six different organs that were investigated of the four mice per group, showed that metastases were formed in 5/24 organs of the mice that were not treated with HGF/SF versus 11/24 organs of the mice that were treated with HGF/SF (mice got ill and were sacrificed 32 days after tumour inoculation) in experiment 1, and that metastases were formed in 6/24 organs of the mice that were not treated with HGF/SF versus 11/24 organs of mice that were treated with HGF/SF (mice got ill and were sacrificed at day 28 and 29) in experiment 2. Adding the results from the two experiments together revealed a significant difference (P = 0.018) in the number of organs with metastases of the mice treated with HGF/SF (5 mg per day) and control mice. In the two experiments, HGF/SF treatment not only promoted dissemination in more mice, but HGF/SF-treated mice also had more lymph node metastases (19 versus eight). Increasing the dose of HGF/SF (from 5 to 25 µg per day × 7 days) did not result in an increased tumorigenicity. Mice treated with 25 µg per day HGF/SF did not have a different dissemination pattern from the control mice (Table 2).

Effect of HGF/SF on tumorigenicity of c-MET-negative B-lymphoma cells in SCID mice
As a control to prove that HGF/SF increases tumorigenicity of c-MET-positive cells by an effect of HGF/SF on the tumour cells and not on e.g. endothelial or other cells leading to an increased invasion, we studied the effect of administration of HGF/SF on the dissemination pattern of c-MET-negative tumours in SCID mice. The Daudi and Ramos cell lines did not express c-MET as detected by RT-PCR. Mice were injected i.v. with 5 × 10 6 tumour cells and 5 µg HGF/SF per day was injected s.c. for 7 days. All mice injected with Daudi cells got sick and were sacrificed at day 32, and mice injected with Ramos cells at day 30. One of the symptoms was paralysis of the hind legs. This symptom was observed in 2/8 mice injected with Daudi cells (one mouse in HGF/SF-treated group and one in the control group) and in 3/8 mice injected with Ramos cells (one mouse in HGF/SF-treated group and two in the control group). Comparing control mice with HGF/SF-treated  Mice were injected i.v. with 5 × 10 6 BJAB cells. Either salt or HGF/SF (5 or 25 mg day -1 ) was administered each day s.c. for 7 days to the mice, starting at the day of tumour injection. a x/y indicates the number of mice with the indicated phenomenon (x) out of the number of mice in each group (y). b The lymph nodes that were involved included: axillary, brachial, poplitial, lumbal, inguinal, caudal and renal; not indicated is the number of lymph nodes in which metastases were formed. c Indicates that the number of organs with metastases is significantly decreased (P = 0.029) compared to the number observed with 5 mg HGF/SF per day and is not significantly different (P = 0.0477) from the number observed without HGF/SF. d indicates that the number of organs with metastases is significantly different (P = 0.018) from the number obtained without HGF/SF. mice revealed that HGF/SF treatment did not promote dissemination of these two cell lines. There is a tendency that with these cell lines, HGF/SF treatment results in fewer tumours; however, these differences are not significant (P = 0.350 for Daudi cells and P = 0.353 for Ramos cells; Table 3). These data suggest that HGF/SF did not affect dissemination of lymphoma cells by an indirect effect on e.g. endothelial cells.

Tumorigenicity after c-MET transduction of c-METnegative B-lymphoma cells
To strengthen our hypothesis that HGF/SF increased tumorigenicity of c-MET-positive cells, we studied the effect of HGF/SF on tumorigenicity of the c-MET-transduced Ramos cells (Ramosc-MET), compared to control-transduced cells (Ramos-GFP), which gave promising results in the in vitro assay (Table 1). Normal HGF/SF (5 µg per mouse per day) and denaturated (30 min autoclaved) HGF/SF were injected s.c. once a day for 7 days. Denaturated HGF/SF was injected instead of salt to exclude the possibility that a low level of endotoxin present in the HGF/SF might influence tumorigenicity of the tumour cells. Denaturated HGF/SF did not have any biological effect in colony assays (data not shown).
Two experiments were performed with four SCID mice per group (Ramos-GFP ± HGF/SF and Ramos-c-METÐGFP ± HGF/ SF). Survival did not differ between the four groups; Ramos-GFP (without HGF/SF) 31 ± 1 days, Ramos-GFP (with HGF/SF) 33 ± 2 days, Ramos-c-METÐGFP (without HGF/SF) 29 ± 1 days, and Ramos-c-METÐGFP (with HGF/SF) 32± 2 days. Twelve out of 16 mice were sacrificed when the hind legs became paralysed. In the second experiments almost all mice got sick on day 30 (paralysis 16/16 mice) and all mice were killed on day 31. The results of the two experiments with regard to dissemination are shown in Table  4. c-MET transduction did not result in a different dissemination pattern. HGF/SF treatment seemed to promote dissemination of Ramos-GFP cells, but the increase was not significant (P = 0.210). HGF/SF treatment also did not significantly affect dissemination of Ramos-c-MET cells in vivo (P = 0.142; Table 4), in contrast to the results of these cells in the in vitro experiments (Table 1).

DISCUSSION
c-MET is expressed on immature B-cells, such as CD38 + CD77 + tonsillar B-cells and CD19 + CD20 Ð B-cells, and not on mature Bcells. c-MET expression can be up-regulated by activation of mature B-cells using stimuli such as CD40-ligand, phorbol 12myristate 13-acetate (PMA) or EpsteinÐBarr virus (EBV) infection (van der Voort et al, 1997;Weimar et al, 1997). We have previously shown that c-MET is expressed in lymph node samples from non-HodgkinÕs lymphoma (NHL) and HodgkinÕs disease patients (Weimar et al, 1997). c-MET expression was found in centroblasts of follicular centre cell NHL (8/11 cases were positive), in centroblasts of large B-cell NHL (7/17 cases were positive), and   Weimar et al, 1997). We found c-MET expression in 2/5 human B-lymphoma cell lines tested (BJAB and Raji were positive; Ramos, Daudi and Jiyoye were negative) (Weimar et al, 1997).
The c-MET ligand HGF/SF is produced by mesenchymal cells, including tonsillar stromal cells and follicular dendritic cells (Van der Voort et al, 1997). For this study we used the c-MET-positive cell line BJAB and the c-MET-negative cell lines Ramos and Daudi for our in vivo experiments, since other investigators have also identified these cell lines as c-MET-positive and -negative respectively (van der Voort et al, 1997;JŸcker et al, 1994). We have previously observed that HGF/SF stimulated the migration and invasion of c-MET-positive cell lines, but not of c-METnegative cell lines (Weimar et al, 1997). In the present study, these results were confirmed. Transduction of Ramos cells with c-MET markedly promoted migration of the transduced cells in response to HGF/SF in an in vitro assay. A response to HGF/SF was only observed when Ramos cells were transduced with c-MET gene (not in control-transduced cells and wild-type cells), suggesting that overexpression of the c-MET gene plays a role in the active migration of lymphoma cells. In a control experiment we determined the effect of HGF/SF on the proliferation of the human Blymphoma cell lines. None of the cell lines that we have used for our studies (BJAB, Daudi, Ramos and c-MET-transduced Ramos cells) showed an altered proliferation by addition of HGF/SF (excluding the possibility that instead of migration, proliferation had been stimulated).
We could not detect the protein HGF/SF by ELISA in the supernatant of the cell lines. Nakamura et al. (1994) have studied 46 lymphoma B-cell lines and could detect production of HGF/SF in only two B-cell lines. Similar to our studies, the cell lines BJAB, RAMOS and Daudi did not produce HGF/SF. Studies in the normal environment of normal B-cells demonstrated that HGF/SF is produced by stromal cells and suggest that HGF/SF is produced by follicular dendritic cells (van der Voort et al, 1997). These data suggest that any effect of HGF/SF on B-cells is mostly due to paracrine production of HGF/SF in the microenvironment of the tumour. Based on these data and the effect of HGF/SF on migration of B-cells in vitro described above, we started these series of experiments to study the effect of (over)expression of c-MET on the dissemination pattern of human lymphoma cells in SCID mice. Murine and human HGF/SF are highly homologous, and human HGF/SF efficiently activates the murine MET receptor, but murine HGF/SF does not appear to activate efficiently the human MET receptor (Bhargava et al, 1992;Rong et al, 1992Rong et al, , 1994. Therefore, human HGF/SF was exogeneously administered to the mice. After i.v. administration, liver and kidney are the most active organs sequestering HGF/SF, but detectable levels of radioiodinated HGF/SF could also be recovered from other highly perfused organs such as spleen and adrenals (Appasamy et al, 1993;Zioncheck et al, 1994). The initial half-life of HGF/SF is 3 min and the elimination half-life is approximately 83Ð114 min. No major toxicities have been reported after i.v. administration of HGF/SF in rats or dogs. The doses applied in this study were equivalent to doses applied in other murine models. In these models, HGF/SF at a dose of 100 µg kg Ð1 protected against renal failure or liver cirrhosis (Matsudo et al, 1995;Ueno et al, 1996;Matsudo et al, 1997).  SCID mice have been used succesfully to study dissemination of human lymphoma cells (Cesano et al, 1991;Mule et al, 1992;Walter et al, 1992;de Kroon et al, 1994;Schwartz et al, 1995;Stroeken et al, 1998). Human B-lymphoma cells (e.g. Ramos and Nalm-6) can bind via their α 4 β 1 integins to murine VCAM-1 molecules, because this interaction is not species-specific (Renz et al, 1994;Tsuzuki et al, 1998). Since we have previously observed that HGF/SF induced adhesion of human c-MET-positive B-cells to fibronectin probably via activation of α 4 β 1 integrins (Weimar et al, 1997), we speculated that i.v. injected human c-MET-positive B-cells could adhere to the murine endothelial cells of the SCID-mice via VCAM-1 molecules expressed on these endothelial cells. We hypothesized that addition of human HGF/SF would increase dissemination (via activation of α 4 β 1 integins on the c-MET-positive human B-cells).
In our experiments, exogenous administered HGF/SF significantly promoted the dissemination of (i.v. injected) BJAB cells; more mice developed tumours and more lymph node metastases were scored (Table 2). These results suggested an effect of HGF/SF on dissemination of c-MET-positive BJAB cells. HGF/SF is known to have multiple activities on vascular endothelial cells. Following stimulation of endothelial cell line (ECV304) with HGF/SF, adherence of colon carcinoma cells to the activated endothelial cells increased. Adhesion was mediated by CD44 receptors, since adhesion could be blocked by anti-CD44 antibodies and HGF/SF up-regulated protein expression of CD44 (Hiscox and Jiang, 1997). Furthermore, HGF/SF promotes the growth of endothelial cells (Bussolino, 1992; and motility of endothelial cells (Bussolino, 1992). To exclude the possibility that the observed effects of HGF/SF on the c-MET-positive BJAB cell line was attributable to an effect of HGF/SF on vascular endothelial cells, we also used the c-METnegative cell lines (Daudi and Ramos). Since HGF/SF did not affect dissemination of these c-MET-negative cell lines, the observed effect of HGF/SF in promoting dissemination of BJAB cells in SCID mice is probably not via an effect on vascular endothelial cells (Table 3).
Overexpression of c-MET has been well documented in various tumours, ranging from melanoma, colectoral carcinoma, thyroid carcinoma, osteosarcoma to leukaemia and lymphoma (Di Renzo et al, , 1995Liu et al, 1992;Natali et al, 1993;JŸcker et al, 1994;Ferracini et al, 1995). The role of overexpression of c-MET on tumorigenicity has been reported for several tumour cell lines. Transfection of murine c-MET in NIH3T3 cells promoted s.c. tumour growth in athymic nude mice, while transfection of human c-MET did not induce tumorigenicity (Rong et al, 1992;Jeffers et al, 1996). The authors explain the differences by the fact that murine HGF/SF does not appropriately stimulate the human c-MET. Co-transfection of human HGF/SF in human c-METpositive NIH3T3 cells did induce tumorigenicity in 89% of the mice. The same group of investigators have studied the effect of c-MET overexpression in the mouse mammary tumour C127 (Jeffers et al, 1996a(Jeffers et al, , 1996b. Transfection of neither c-MET nor HGF/SF promoted tumour formation in mice, while co-transfection of both genes markedly induced s.c. tumors. Based on these data we administered human HGF/SF to our mice injected with human c-MET-transduced Ramos cells. In our experiments with c-MET-transduced Ramos cells, c-MET transduction itself did not affect invasion of organs by lymphoma cells. This could be due to the lack of human HGF/SF. However, also the administration of human HGF/SF did not have any effect on tumorigenicity of the c-MET-transduced Ramos cells. HGF/SF administration also did not affect survival times of the mice. Although we can not exclude the possibility that co-transfection of human HGF/SF in these c-MET-transduced cells, creating an autocrine production of HGF/SF, might have different outcomes, our data indicate that the c-METÐHGF/SF interaction only plays a minor role in tumorigenicity of human B-cell lymphoma cell lines in SCID mice.