- Email: [email protected]
Direct evidence that hepatocyte growth factor-induced invasion of hepatocellular carcinoma cells is mediated by urokinase
Journal of Hepatology 1999; 30:511-518 Printed in Denmark • All rights reserved Munksgaard. Copenhagen
Copyright © European Association for the Study of the Liver 1999
Journal of Hepatology ISSN 0168-8278
Direct evidence that hepatocyte growth factor-induced invasion of hepatocellular carcinoma cells is mediated by urokinase Arnaud Monvoisin, V6ronique Neaud, Victor De L6dinghen, Liliane Dubuisson, Charles Balabaud, Paulette Bioulac-Sage, Alexis Desmouli~re and Jean Rosenbaum Groupe de Recherches pour l'Etude du Foie, Universit~ Victor Segalen Bordeaux 2, Bordeaux, France
Background~Aims: We have shown that hepatocyte growth factor secreted by human hepatic myofibroblasts increased the in vitro invasion of the hepatocarcinoma cell line HepG2 through Matrigel. Our aim in this study was to evaluate the role of urokinase in this process. Methods: Expression of urokinase in HepG2 cells was measured by Northern blot and zymography, and plasminogen activation was shown by a chromogenic substrate assay. Cell invasion was assayed on Matrigelcoated filters. Urokinase and urokinase receptor transcripts in hepatocarcinoma were detected by reverse transcription-polymerase chain reaction. Activated hepatocyte growth factor was detected by Western blot with a hepatocyte growth factor-]/chain-specific antibody. Results: HepG2 cells expressed urokinase mRNA and secreted active urokinase. Urokinase expression was enhanced by hepatocyte growth factor at the protein
and mRNA level. Notably, cell-surface-associated urokinase was increased 22-fold by hepatocyte growth factor. Hepatocyte growth factor also increased urokinase receptor mRNA expression. B428, a urokinase inhibitor, decreased by up to 70% HepG2 invasion induced by myofibroblasts and by 90% that induced by recombinant hepatocyte growth factor. This was not due to a decrease in the generation of activated hepatocyte growth factor by myofibroblasts. Finally, all 17 hepatocarcinoma samples tested expressed urokinase and urokinase receptor transcripts. Conclusion: Hepatocyte growth factor-dependent, myofibroblasts-induced invasion of HepG2 cells is secondary to the induction of urokinase expression on tumor cells.
EPATOCELLULARcarcinoma (HCC) is the main type of liver cancer and is developed from hepatocytes. HCC is associated with a high rate of intrahepatic invasion. The strorna of HCC is infiltrated with numerous myofibroblasts (1). Using an in vitro co-culture model, we have shown that human liver myofibroblasts strongly promoted invasion of two HCC cell lines through Matrigel-coated filters (2). We found that this effect was mediated by hepatocyte growth factor/scatter factor (HGF/SF), secreted by myofibroblasts. HGF/SF is a heparin-binding polypeptide, trans-
lated as a 92-kDa single-chain inactive precursor. HGF/SF is proteolytically activated to form two subunits, alpha (60 kDa) and beta (32-36 kDa), linked by a unique disulfide bond. HGF/SF is a multifunctional protein with mitogenic (3), motogenic (4) and morphogenic (2,5) properties. The specific HGF/SF cell receptor is a 190 kDa transmembrane tyrosine kinase receptor encoded by the proto-oncogene c-met (6). Numerous studies have shown that HGF/SF/c-Met signalling can induce the invasiveness and metastatic potential of several carcinoma cell types (7,8). The molecular mechanisms underlying HGF-induced invasion are not well known. Thus, although HGF has been shown to induce the secretion of urokinase-type plasminogen activator (u-PA) (9,10) and matrix metalloproteinases (11), the role of these proteases in HGF-induced invasion has not been ascertained. Urokinase is able to activate plasminogen to plasmin
H
Received 8 June; revised 5 October; accepted 20 October 1998
Correspondence: Jean Rosenbaum, Groupe de Recherches pour l'Etude du Foie, Universit6 Victor Segalen Bordeaux 2, 146 rue L6o Saignat, 33076 Bordeaux, France. Tel: 33 5 57571771; Fax: 33 5 56514077; E-mail: [email protected]
Key words: Cancer; Hepatocyte growth factor; Invasion; Liver; Urokinase
511
A. Monvoisin et al.
which possesses a broad-spectrum proteolytic activity. u-PA is synthesized and secreted as a single-chain proenzyme (pro-u-PA) (12) that is cleaved in a two-chain active form. Pro-u-PA and u-PA bind to the u-PA receptor (u-PAR) (13), a heavily glycosylated protein attached to the cell membrane by a glycosyl phosphatidyl inositol anchor. Binding of pro-u-PA to u-PAR allows for u-PA activation, although a recent study suggested that uPAR-independent activation can also occur (14). In this study, we have investigated whether urokinase is involved in HepG2 cells invasion in vitro when induced by liver myofibroblasts or HGE
Materials and Methods Materials Culture medium, additives and Moloney Murine Leukemia Virus Reverse Transcriptase (MMLV-RT) were from Gibco-BRL (Life Technologies, Cergy-Pontoise, France). Human AB serum was from Centre R6gional de Transfusion Sanguine (Bordeaux, France). The RNeasy mini kit was from Qiagen (Les Ulis, France), HybondTM-N + membrane and ECL kit from Amersham (Les Ulis, France). [a-32p]dCTP and [y_32p] dATP were from ICN Biomedicals (Orsay, France). The Ready-to-go random priming kit was from Boehringer-Mannhein (Meylan, France). Taq polymerase was from Promega (Charbonni6res, France). Plasminogen and the plasmin substrate $2251 were from Chromogenix (M61ndal, Sweden), human urokinase from Sanofi Winthrop (Gentilly, France). Heparin-Sepharose was from Pharmacia-Biotech (Orsay, France), and polyvinylidene difluoride membranes from Millipore (Saint-Quentin en Yvelynes, France). Polycarbonate filters were from Falcon (Becton Dickinson, Le Pont de Claix, France) and Matrigel basement membrane matrix from Collaborative Biomedical Products (Bedford, MA, USA). Recombinant human HGF (rhHGF) was from R&D Systems (Oxon, UK). Chemicals were from Sigma Aldrich (St Quentin Fallavier, France). The HGF-fl chain rabbit polyclonal antibody was a generous gift from Dr. A. Galvani (Pharmacia-UpJohn, Milan, Italy). The u-PA inhibitor B428 was a kind gift from Dr. B.A. Littlefield of Eisai Research Institute (Andover, MA, USA) (15). B428 was made as a 40 mM stock concentration in dimethyl sulfoxyde, then diluted into water to 4 mM, and finally into Dulbecco's modified Eagle's medium (DMEM). In our experiments, the final concentration of DMSO did not exceed 0.0375%. Cells and cell culture The HepG2 human HCC cell line was cultured in DMEM containing 10% fetal calf serum. Human hepatic myofibroblasts (MF) were obtained from explants of non-tumoral liver resected during partial hepatectomy and characterized as myofibroblasts as previously described (16,17). MF were grown in DMEM containing 5% fetal calf serum, 5% pooled human AB serum and 5 ng/ml recombinant human epidermal growth factor. Northern blot Total RNA was prepared from cultured HepG2 cells using the RNeasy mini kit. 15 and 20/tg, respectively for u-PA and u-PAR mRNA expression, were electrophoresed on a 0.8% agarose/formaldehyde gel and transferred to a HybondTM-N+ membrane, using a Vacuum Blotter (Appligene, Illkirch, France). Prehybridization and hybridization were carried out at 65°C in Na2HPO4/H3PO4 0.5 M, pH 7.2, SDS 7%, EDTA 1 raM, BSA 1 g/100 ml, ssDNA 100/tg/ml. The membrane was hybridized with a cDNA probe for human u-PA or u-PAR, labeled with [c~-32p] dCTP. The blots were washed in stringent conditions and exposed for autoradiography. The u-PA and u-PAR plasmids used were a kind gift from Dr B. S. Nielsen (The Finsen
512
Institute, Copenhagen, Denmark) and included nucleotides 32q527 and 497-1081 for u-PA and u-PAR, respectively (18). To control for variations in RNA loading and transfer, the blots were rehybridized with a 28 S rRNA-specific oligonucleotide probe labeled with [7-32p] dATE The autoradiographic signals were quantified with the NIH Image 1.60 software on a Macintosh computer. Reverse-transcription polymerase ehain reaction Total RNA was prepared from HepG2 cells and from frozen tumoral liver samples collected from 17 patients undergoing liver resection or transplantation for hepatocellular carcinoma. One microgram of RNA was reverse-transcribed in a 50 #1 volume using MMLV-RT. Two microliters of the reaction were used for amplification with the following specific primers: 5'-CAG ATC TGA TGC TCT TCA GCT3' (uPA sense primer), 5'-GTG ACT TCA GAG CCG TAG TAG-3' (uPA antisense primer), 5'-GCC CTG GGA CAG GAC CTC TG-3' (uPAR sense primer), 5'-CAT TGA TTC ATG GGG CCT CGG C-3' (uPAR antisense primer). Thirty-five cycles of PCR were performed, consisting each of denaturation for 1 min at 94°C, annealing for 1 min at 58.5°C for u-PA (65°C for u-PAR), and elongation for 2 min at 72°C. PCR was performed in 50 id of reaction buffer containing 50 mM KC1, 10 mM Tris-HC1, pH 9, 0.1% Triton X-100, 2.4 mM MgC12, 0.4 mM dNTPs, 1.25 U Taq polymerase and 10 pmol of primers. Ten microliters of the reaction were analyzed by agarose gel electrophoresis. The sizes of the predicted products are 747 and 613 bp, respectively, for u-PA and u-PAR. For negative controls, reverse transcriptase was omitted during the RT procedure, or PCR was performed without cDNA. Secreted urokinase assays and zymography HepG2 cells were seeded in 24-well dishes at 2 ×105 cells per well in DMEM 2% FCS. After 6 h, the cells were rinsed once with Gey's balanced salt solution (GBSS) CaZ+/Mg2+, once with an acid buffer (50 mM glycine, 100 mM NaC1, pH 3) to remove cell surface-bound urokinase, and twice with serum-free DMEM. Finally, cells were incubated in 0.5 ml serum-free DMEM for 48 h. Experiments were done in triplicate. Conditioned media (CM) were first centrifuged at 600×g for 4 min at 4°C to remove cellular debris, then at 15000×g for 5 min at 4°C and stored at -20°C. Six microliters of the supernatants were electrophoresed in a 10% SDS-polyacrylamide gel containing 1 mg/ml gelatin and 12.5/~g/ml plasmin-free plasminogen. After washing with 2.5% Triton X-100, the gel was incubated at 37°C for 48 h in 100 mM glycine, 20 mM EDTA, pH 8.3. The gel was finally stained with 0.5% Coomassie blue. Proteolytic activity was detected as a white zone in a dark field. As a control, the same experiment was done in a gel without plasminogen. Purified human urokinase was used as a positive control. To explore the effect of HGF on u-PA synthesis by HepG2 cells, HepG2 cells were seeded as above and incubated in 1 ml serum-free DMEM with or without rhHGF at 100 ng/ml. After 48 h, CM were collected and cells were rinsed twice with 100 /tl of serum-free DMEM. Endogenous cell surface-associated u-PA was eluted at room temperature for 3 min with 100/zl of the acid buffer described above. The eluates were neutralized with 25/d of 0.5 M Tris-HC1, pH 7.8 and centrifuged. The cell layer was finally processed for DNA measurement according to Labarca & Paigen (19). Following normalization for DNA content of the cell layer, u-PA activity in supernatants and eluates was assayed, u-PA activity on the zymography was quantified with the NIH Image 1.60 software. Plasminogen activation HepG2 cells were seeded as described above and were incubated for 24 h in 500/tl of serum-free DMEM with or without rhHGF at 100 ng/ml. At the end of the incubation, media were removed and cells were rinsed with phosphate buffer saline (PBS). One hundred microliters of plasmin-free plasminogen at 250/~g/ml in 0.15 M NaC1, with or without 15/~M B428, were added to the cells for 3 h at 37°C. Supernatants were collected and centrifuged and the cells were rinsed with PBS. Plasmin bound to the cell layer was then eluted with 120 /A of 3 mM tranexamic acid in PBS Ca2+/Mg2+ (20). Plasmin activity
Urokinase and H C C invasion was assayed in supernatant and eluate samples by incubating 120/zl of each sample with 50/A of the plasmin-sensitive chromogenic substrate $2251 at 4 mM, for 2 h 30 min in a 96-well plate. Plasmin generated was evaluated by its amidolytic activity on $2251 (21). The cell layer was processed for D N A measurement. Cell invasion assay Tests were performed as described previously (2) with slight modifications. Eight-micrometer polycarbonate pore size filters were coated with a uniform layer o f Matrigel basement membrane matrix (16/~g/ cm2). Myofibroblasts 50× 10 3 w e r e seeded in the lower compartment of the system and 45× 103 HepG2 ceils were seeded onto the filters in 2°,/0 FCS D M E M . The final concentration of FCS was 0.4%. In some experiments, r h H G F (100 ng/ml) was used instead of myofibroblasts in the lower compartment. After 48 h, the cells on the upper surface o f the filter were wiped with a cotton swab. Filters were fixed for 10 min with methanol and stained with hematoxylin. Cells that invaded the lower surface of the filter were counted under a photonic microscope at a final magnification of 320. Results were expressed as number of HepG2 cells invaded per filter. Experiments were done in triplicate and results are shown as mean_+l SD. Cytotoxicity assay Myofibroblasts were seeded into 24-well dishes and HepG2 cells into 96-well dishes at 50× 103 or 45× 103 cells per well, respectively, in 2% FCS D M E M with or without B428. Cells were incubated for 48 h at 37°C. After this time, medium was removed and cells were incubated with D M E M containing 1 mg/ml of 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT). After 2 h, cells were solubilized with DMSO and the optical density at 540 nm was measured in a microplate reader. Measure of HGF activation Confluent monolayers of M F were washed with serum-free medium and incubated for 2 h with serum-free medium. This medium was discarded and the ceils were then further incubated for 24 h in serumfree medium, in the absence or in the presence of 15/~M B428. Medium was collected, centrifuged, and heparin-binding proteins were adsorbed to heparin-Sepharose, as described (2). Heparin-bound proteins, or r h H G F were analyzed by Western blot as previously described (2), using the H G F fl-chain-specific antibody, diluted 1:2500.
Results Urokinase and urokinase receptor expression by HepG2 cells Northern blot analysis showed that HepG2 cells expressed a single 3.1 kb transcript corresponding to uPA m R N A (Fig. 1A). RT-PCR analysis showed that HepG2 cells also expressed u-PA receptor m R N A (Fig. 1B). Controls where the enzyme was omitted during the RT step or where cDNA was omitted for the PCR step were negative. The ability of HepG2 cells to secrete urokinase was determined by zymography on a SDS-polyacrylamide gel containing plasminogen and gelatin. Gel lysis was detected with CM from HepG2 cells (Fig. 1C). This was indeed plasminogen-dependent lysis since it was not found when the experiment was done in a gel without plasminogen (data not shown). The lysis activity had a relative mobility which corresponded to approximately 55 kDa and co-migrated with purified human urokinase. No bands corresponding to t-PA were observed. Finally, this activity
was inhibited when incubation of the gel was done in the presence of 100 ~tM B428, a urokinase-specific inhibitor derived from amiloride (data not shown). Taken together, these results indicate that HepG2 ceils are able to produce active urokinase. Effect of HGF/SF on u-PA and u-PAR expression by HepG2 cells Northern blot analysis showed that H G F increased the u-PA mRNA level in HepG2 cells. Following normalization with 28 S RNA, a 6-fold increase was observed after 6 h and 2.2-fold at 24 h (Fig. 2). Moreover, incubation of HepG2 cells with HGF/SF led to an in-
A
3,1 kb ----~
B
1
C
I
2
3
2
Fig. 1. Expression of u-PA and u-PAR mRNA in HepG2 cells. A, Total RNA from HepG2 cells was analyzed by Northern blot with a u-PA cDNA probe. B, RT-PCR analysis of HepG2 RNA with u-PAR-specific primers. Lane 1: molecular size markers; Lane 2: negative control; Lane 3:HepG2 cells. C, Zymography of HepG2 cells conditioned medium for u-PA detection. Lane 1:HepG2 cells; lane 2: purified human urokinase (10 mU).
6 hrs -
24 hrs ÷
-
÷
u-PA mFINA
28S
---~
Fig. 2. Effect of HGF on u-PA mRNA expression byHepG2. HepG2 cells were incubated in the absence ( - ) or the presence (+) of human recombinant HGF, and total RNA was extracted at the time indicated. The Northern blot was hybridized sequentially with a u-PA and a 28 S probe. 513
A. Monvoisin et al.
A
_
A
÷
6000 5000 4000
B o ~
3000
ooo
Z.~ Fig. 3. Effect of H G F on u-PA production by HepG2 cells. u-PA activity in the supernatant (A) and on the cell surface (B) of HGF-stimulated ( + ) or unstimulated ( - ) HepG2 cells was assayed by zymography in triplicate. One experiment representative of three is shown.
lOOO
0
MF : B428(gM)
:
-
+
+
+
+
"
"
6
10
15
B 6 hrs
24 hrs
6000
i
-
+
-
+
UPAR1 mRNA
5000
l
4000 UPAR2 mRNA
o I~w
3000 2000
Z.. =
1000
28S --~
0 Fig. 4. Effect of H G F on u-PAR m R N A expression byHepG2. HepG2 cells were incubated in the absence ( - ) or the presence ( + ) of human recombinant HGF, and total R N A was extracted at the time indicated. The Northern blot was hybridized sequentially with a u-PAR and a 28 S probe.
crease in the expression of active u-PA. u-PA expression in the cell supernatant was enhanced by 2.8_+ 1.4fold (mean_+ 1 SD), while cell-surface associated u-PA was increased by 22_+2.4-fold after HGF-treatment (p<0.01, Student's t-test) (Fig. 3). As shown on Fig. 4, H G F also upregulated the level of the 1.4 kb u-PAR m R N A at 6 and 24 h. This transcript, called uPAR 1, codes for the full-length cell surface u-PAR (13). On the other hand, H G F had only a 514
HGF: B428(l.tM)
. :
-
+ "
+ 15
Fig. 5. Effect of urokinase inhibitor B428 on HepG2 cells invasion in vitro. HepG2 were seeded in invasion chambers coated with Matrigel as described in Materials and Methods. Experiments were done using either M F (A) or human recombinant HGF at 100 ng/mI final concentration (B) in the lower chamber. Experiments with M F were done in the presence of different concentrations ofB428. Experiments with HGF were done with B428 at 15/~M. The total number of cells migrating through the filter was counted. Results are expressed as mean +_1 SD of three experiments done in triplicate.
Urokinase and HCC invasion
minor effect on the level of the 1.1 kb u-PAR mRNA. This transcript, called uPAR2, is a variant generated by alternative splicing described by Pyke et al., which probably encodes a soluble form of uPAR (22). Effect of HGF/SF on cell-surface activation of plasminogen After addition of purified human plasminogen to HGF/SF-stimulated HepG2 and unstimulated HepG2 cells, plasmin activity could be recovered in the supernatant and as a cell-bound fraction. Plasmin activity was enhanced by HGF-treatment by 2.9_+0.7 (mean_+ 1 SD; p<0.02, Student's t-test) and 3.5___l.l-fold (p<0.05) in the supernatant and on the cell surface, respectively, as compared with control cells. When B428 was added together with plasminogen, plasminogen activation was inhibited in the supernatant and on the cell surface by 87.6 and 96.2%, respectively. Effect of B428 on HepG2 cells invasion in vitro We finally tested whether u-PA was involved in HepG2 cells invasivity. As shown on Fig. 5A, co-incubation of HepG2 cells with hepatic myofibroblasts led to a 77fold increase in invasivity. Addition of B428 caused a dose-dependent inhibition of the HepG2 cells invasion induced by myofibroblasts. At 15/zM B428, the effect of myofibroblasts was reduced by 69.5%. Similar resuits were obtained when rhHGF was substituted for myofibroblasts, rhHGF stimulated the invasion of HepG2 cells 120-fold. Addition of B428 at 15/zM decreased this effect by 91% (Fig. 5B). No inhibition of HepG2 cells invasion was found with DMSO, the solvent of B428 (data not shown). The decreased invasiveness was not due to a cytotoxic effect of B428 on HepG2 or myofibroblasts, since B428 did not modify MTT uptake by these two cell types at the concentrations used (data not shown). Effect of B428 on HGF activation As u-PA is known to be able to activate pro-HGF into its active heterodimeric form, we checked whether B428 prevented the generation of the H G F heterodimer by ME Fig. 6A shows that addition of B428 to cultured MF did not modify the amount of fl-chain in the cultures. In addition, we also showed that a significant part of the rhHGF used in this study is already in active form (Fig. 6B). Finally, ELISA assays demonstrated that B428 did not modify H G F secretion by MF (not shown). u-PA and u-PAR expression in human HCC Total R N A was extracted from 17 human HCC samples and assayed by RT-PCR with u-PA and u-PAR
A
1
13
2
Fig. 6. Effect of B428 on HGF activation. A. HeparinSepharose-concentrated conditioned medium from MF was analyzed by Western blot with an antibody to the fl chain of HGF. Lane 1: control MF-CM; Lane 2: CM prepared in the presence of 15 IzM B428. B. rhHGF standard (50 ng). Arrows point to the pro-HGF above, and the HGF-fl chain doublet below. The arrowhead indicates a non-specific signal due to the large amount of bovine serum albumin present in the solution of rhHGF.
A
B
1 2 8 4 8
$
78
g 10 11 12 18 14 16 18 17 18 1 |
Fig. 7. Expression of u-PA and u-PAR mRNA in HCC. cDNA reverse transcribedfrom total RNA extracted from HCC samples was amplified using either u-PA (A) and uPAR-specific (B) primers. Lane 1: molecular size markers; Lane 2: negative control; Lane 3 to 19: HCC samples.
primers. As shown on Fig. 7, both transcripts were detected in every sample. Controls were negative.
Discussion In this study, we have shown that the human HCC cell line HepG2 expresses u-PA and its receptor, and that u-PA is responsible for the in vitro invasivity of HepG2 cells across Matrigel, when induced by co-culture with liver MF or by purified human H G E u-PA expression was demonstrated by Northern blot and zymography. 515
A. Monvoisin et al.
Zymography showed a single band of proteolytic activity that co-migrated with purified human u-PA and which was abolished in the presence of B428, a specific u-PA inhibitor derived from amiloride. No other plasminogen activator species was detected in HepG2-conditioned medium. Membrane-bound u-PA was biologically active since it was able to promote conversion of plasminogen to plasmin. This activity was up to 96% inhibited by B428, indicating clearly that u-PA accounted for almost all of the plasminogen activator activity in our model. These results differ from those of Wojta et al. (23), who were unable to detect u-PA expression in HepG2 cells, using Northern blot and a specific ELISA. The reasons for this discrepancy could be a lack of sensitivity of their assays or the use of a different subclone of HepG2. In further experiments, we demonstrated that the expression of u-PA by HepG2 cells at the mRNA and protein level was enhanced by exposition of the cells to HGE HGF specially increased the amount of membrane-bound u-PA by more than 20-fotd. This is potentially important as it is membrane-bound u-PA that is specifically involved in the pericellular proteolysis necessary for cell invasion (24,25). Our data also show that H G F treatment increases the expression of u-PAR transcripts in HepG2 cells. Such an increase in u-PAR expression likely participates in the relatively selective increase in membrane-bound u-PA, by localizing u-PA to the cell surface. Up-regulation of u-PA by HGF has previously been demonstrated in other cell types, such as M D C K cells (10), or human leiomyosarcoma cells transfected with a HGF expression vector (9). Recently, Kamiyama et al. have shown that H G F upregulated uPA antigen secretion in 1 out of 3 human HCC cell lines (26). However, the increase was very modest, about 1.2-fold over baseline. Moreover, in all of the above studies, no data allow the increased u-PA expression to be linked to the pro-invasive effect of HGE Such a link has been established in our study since we have shown that the u-PA-specific inhibitor B428 blocked HepG2 cells invasivity, when induced by coculture with ME or by recombinant human HGE We have previously shown that MF-induced invasivity was dependent on the secretion of H G F by ME since it was completely inhibited by anti-HGF antibodies (2). We have now further established that HGF-dependent, MF-induced, HepG2 invasiveness is secondary to the induction of u-PA expression on tumor cells. B428 inhibited rHGF-induced invasion more efficiently than MF-induced invasion. This suggests that, besides HGF, MF secrete other pro-invasive molecules that cooperate with H G F and act via a u-PA-independent pathway. The downstream mechanism of action of u-PA in 516
our system has not been completely elucidated, u-PA could promote invasion through plasminogen conversion into plasmin, with subsequent direct degradation of extracellular matrix proteins, or activation of the latent forms of matrix-metalloproteinases (27). Another possibility is that B428 acts mainly by preventing the u-PA-mediated conversion of pro-HGF to its active heterodimeric form (28). Our data argue against this hypothesis since the amount of HGF fl-chain, a corollary of HGF activation, was unchanged in the presence of B428, indicating that activation of HGF in these cultures is largely u-PA-independent. Moreover, B428 inhibited the pro-invasive effect of recombinant H G F as efficiently as that of myofibroblasts, and we have found that a large fraction of the commercial recombinant H G F was already in heterodimeric form. The mechanism of HGF-activation in MF-conditioned medium is unknown but could be due to proteolysis by the HGF activator (29). Finally, B428 could act by inhibiting the activation of the pro-form of matrixmetalloproteinase-2 (MMP-2) by urokinase (30). MMP-2 is involved in extracellular matrix breakdown and invasion (31), and has recently been shown to be overexpressed in human HCC (32). Our data also indicate that u-PA is actively transcribed in human HCC. We show too that u-PA receptor mRNA is expressed in human HCC. This is in good agreement with a recent report showing, using in situ hybridization and immunohistochemistry, that uPAR is present in tumoral hepatocytes along the tumorstroma interface (33). While this manuscript was being submitted, De Petro et al. confirmed the expression of both transcripts in HCC and showed, using a quantitative RT-PCR assay that u-PA trancript levels were much higher in HCC than in the surrounding non-tumoral liver. This is compatible with our own data, using in situ hybridization (manuscript in preparation). They were also able to correlate high u-PA levels with a poor prognosis. Altogether, these results suggest that the u-PA pathway is implicated in HCC. Thus, u-PA inhibitors could conceivably be of benefit in limiting HCC invasiveness in the clinical setting. Interestingly, the u-PA inhibitor B428 used in our study has shown anti-invasive effects in animal models of breast or prostatic cancer (34,35). We are currently testing its efficacy in a rat HCC model.
Acknowledgements We wish to express our deep thanks to Bruce A. Littlefield for providing B428, Boye Schnack Nielsen for the plasmids, Arturo Galvani for the HGF fl-chainspecific antibody, and Genevieve N'Guyen for her help in starting this study. This work was supported by
Urokinase and H C C invasion
grants from Comit6s de la Gironde et de la Dordogne from the Ligue Nationale Franqaise contre le Cancer, Association pour la Recherche sur le Cancer, Groupement des Entreprises Franqaises pour la Lutte contre le Cancer, and Conseil r6gional d'Aquitaine. AM was a recipient of a fellowship from the Comit6 de la Dordogne from the Ligue Nationale Fran~aise contre le Cancer.
References 1. Enzan H, Himeno H, Iwamura S, Onishi S, Saibara T, Yamamoto Y, et al. a-smooth muscle actin positive perisinusoidal stromal ceils in human hepatocellular carcinoma. Hepatology 1994; 19: 895-903. 2. Neaud V, Faouzi S, Guirouilh J, Le Bail B, Balabaud C, BioulacSage P, et al. Human hepatic myofibroblasts increase invasiveness of hepatocellular carcinoma cells: evidence for a role of hepatocyte growth factor. Hepatology 1997; 26: 1458-86. 3. Michalopoulos G, Houck KA, Dolan ML, Luetteke NC. Control of hepatocyte replication by two serum factors. Cancer Res 1984; 44: 4414-9. 4. Stoker M, Gherardi E, Perryman M, Gray J. Scatter factor is a fibroblast-derived modulator of epithelial cell mobility. Nature 1987; 327: 239-42. 5. Montesano R, Matsumoto K, Nakamura T, Orci L. Identification of a fibroblast-derived epithelial morphogen as hepatocyte growth factor. Cell 1991; 67: 901-8. 6. Bottaro DP, Rubin JS, Faletto DL, Chan AML, Kmiecik TE, Vande Woude GE et al. Identification of the hepatocyte growth factor receptor as the c-met proto-oncogene product. Science 1991; 251: 802-4. 7. Hasegawa Y, Yamamoto M, Maeda S, Saitoh Y. Hepatocyte growth factor and its receptor/c-MET regulate both cell growth and invasion of human pancreatic cancer. Int J Oncol 1995; 7: 877-81. 8. Matsumoto K, Date K, Shimura H, Nakamura T. Acquisition of invasive phenotype in gallbadder cancer cells via mutual interaction of stromal fibroblasts and cancer cells as mediated by hepatocyte growth factor. Jpn J Cancer Res 1996; 87: 702-10. 9. Jeffers M, Rong S, Vande Woude GE Enhanced tumorigenicity and invasion-metastasis by hepatocyte growth factor/scatter factor-Met signalling in human cells concomitant with induction of the urokinase proteolysis network. Mol Cell Biol 1996; 16:111525. 10. Pepper MS, Matsumoto K, Nakamura T, Orci L, Montesano R. Hepatocyte growth factor increases urokinase-type plasminogen activator (u-PA) and u-PA receptor expression in Madin-Darby canine kidney epithelial cells. J Biol Chem 1992; 267: 20493-6. 11. Uchiyama A, Essner R, Doi F, Nguyen T, Ramming KP, Nakamura T, et al. Interleukin 4 inhibits hepatocyte growth factorinduced invasion and migration of colon carcinomas. J Cell Biochem 1996; 62: 443-53. 12. Kasai S, Arimura H, Nishida M, Suyama T, Primary structure of single-chain pro-urokinase. J Biol Chem 1985; 260: 12382-9. 13. Roldan AL, Cubellis MV, Masucci MT, Behrendt N, Lund LR, Dano K, et al. Cloning and expression of the receptor for human urokinase plasminogen activator, a central molecule in cell surface, plasmin dependent proteolysis. EMBO J 1990; 9: 467-74. 14. Carmeliet P, Moons L, Dewerchin M, Rosenberg S, Herbert JM, Lupu E et al. Receptor-independent role of urokinase-type plasminogen activator in pericellular plasmin and matrix metaUoproteinase proteolysis during vascular wound healing in mice. J Cell Biol 1998; 140: 233-45. 15. Towle M J, Lee A, Maduakor EC, Schwartz E, Bridges A J, Littlefield BA. Inhibition of urokinase by 4-substituted benzo[b]-
16.
17.
18.
19. 20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
thiophene-2-carboxamidines: an important new class of selective synthetic urokinase inhibitors. Cancer Res 1993; 53: 2553-9. Win KM, Charlotte F, Mallat A, Cherqui D, Martin N, Mavier P, et al. Mitogenic effect of TGFI31 on human Ito cell in culture: evidence for mediation by endogenous platelet derived growth factor. Hepatology 1993; 18: 137-45. Blazejewski S, Pr6aux AM, Mallat A, Broch6riou I, Mavier P, Dhumeaux D, et al. Human myofibroblast-like cells obtained by outgrowth are representative of the fibrogenic cells in the liver. Hepatology 1995; 22: 788-97. Pyke IC, Kristensen P, Ralfkiaer E, Grondahl-Hansen J, Eriksen J, Blasi E et al. Urokinase-type plasminogen activator is expressed in stromal cells and its receptor in cancer cells at invasive loci in human colon adenocarcinomas. Am J Pathol 1991; 138: 1059-67. Labarca C, Paigen K. A simple, rapid, and sensitive DNA assay procedure. Anal Biochem 1980; 102: 344-52. Stephens RW, P611~inen J, Tapiovaara H, Leung KC, Sim PS, Salonen EM, et al. Activation of pro-urokinase and plasminogen on human sarcoma cells: a proteolytic system with surface-bound reactants. J Cell Biol 1989; 108: 1987-95. Lottenberg R, Christensen U, Jackson MC, Coleman PL. Assay of coagulation proteases using peptide chromogenic and fluorogenic substrates. Methods Enzymol 1981; 80: 341-81. Pyke C, Eriksen J, Solberg H, Schnack Nielsen B, Kristensen P, Lund LR, et al. An alternatively spliced variant of mRNA for the human receptor for urokinase plasminogen activator. FEBS Lett 1993; 326: 69--73. Wojta J, Nakamura T, Fabry A, Hufnagl P, Beckmann R, McGrath K, et al. Hepatocyte growth factor stimulates expression of plasminogen activator inhibitor type 1 and tissue factor in HepG2 cells. Blood 1994; 8: 151-7. Cohen RL, Xi XP, Crowley CW, Lucas BK, Levinson AD, Shuman MA. Effects of urokinase receptor occupancy on plasmin generation and proteolysis of basement membrane by human tumor cells. Blood 1991; 78: 479-87. Hollas W, Blasi E Boyd D. Role of the urokinase receptor in facilitating extracellular matrix invasion by cultured colon cancer. Cancer Res 1991; 51: 3690-5. Kamiyama T, Une Y, Uchino J, Hamada J. Hepatocyte growth factor enhances the invasion activity of human hepatocellular carcinoma cell lines. Int J Oncol 1998; 12: 655-9. Baramova EN, Bajou K, Remacle A, L'Hoir C, Krell HW, Weidle UH, et al. Involvement of PA/plasmin system in the processing of pro-MMP-9 and in the second step of pro-MMP-2 activation. FEBS Lett 1997; 405: 157-62. Naldini L, Tamagnone L, Vigna E, Sachs M, Hartmann G, Birchmeier W, et al. Extracellular proteolytic cleavage by urokinase is required for activation of hepatocyte growth factor/scatter factor. EMBO J 1992; 11: 4825-33. Miyazawa K, Shimomura T, Kitamura A, Kondo J, Morimoto Y, Kitamura N. Molecular cloning and sequence analysis of the cDNA for a human serine protease responsible for activation of hepatocyte growth factor. J Biol Chem 1993; 268: 10024-8. Keski-Oja J, Lohi J, Tuuttila A, Tryggvason K, Vartio T. Proteolytic processing of the 72 000-Da type IV collagenase by urokinase plasminogen activator. Exp Cell Res 1992; 202: 471-6. Abe T, Mori T, Kohno K, Seiki M, Hayakawa T, Welgus HG, et al. Expression of 72 kDa type IV collagenase and invasion activity of human glioma cells. Clin Exp Metast 1994; 12: 296304. Musso O, Th6ret N, Campion JP, Turlin B, Milani S, Grappone C, et al. In situ detection of matrix metalloproteinase-2 (MMP-2) and the metalloproteinase inhibitor TIMP2 transcripts in human primary hepatocellular carcinoma and in liver metastasis. J Hepatol 1997; 26: 593-605. Morita Y, Hayashi Y, Wang Y, Kanamaru T, Suzuki S, Kawasaki K, et al. Expression of urokinase-type plasminogen activator receptor in hepatocellular carcinoma. Hepatology 1997; 25: 85661.
517
A. Monvoisin et al. 34. Xing RH, Mazar A, Henkin J, Rabbani A. Prevention of breast cancer growth, invasion, and metastasis by antiestrogen tamoxifen alone or in combination with urokinase inhibitor B-428. Cancer Res 1997; 57: 3585-93.
518
35. Alonso DE Farias EE Ladeda V, Davel L, Puricelli L, Bal De Kier Joff6 E. Effects of synthetic urokinase inhibitors on local invasion and metastasis in a routine mammary tumor model. Cancer Res 1996; 40: 209-23.
Copyright © European Association for the Study of the Liver 1999
Journal of Hepatology ISSN 0168-8278
Direct evidence that hepatocyte growth factor-induced invasion of hepatocellular carcinoma cells is mediated by urokinase Arnaud Monvoisin, V6ronique Neaud, Victor De L6dinghen, Liliane Dubuisson, Charles Balabaud, Paulette Bioulac-Sage, Alexis Desmouli~re and Jean Rosenbaum Groupe de Recherches pour l'Etude du Foie, Universit~ Victor Segalen Bordeaux 2, Bordeaux, France
Background~Aims: We have shown that hepatocyte growth factor secreted by human hepatic myofibroblasts increased the in vitro invasion of the hepatocarcinoma cell line HepG2 through Matrigel. Our aim in this study was to evaluate the role of urokinase in this process. Methods: Expression of urokinase in HepG2 cells was measured by Northern blot and zymography, and plasminogen activation was shown by a chromogenic substrate assay. Cell invasion was assayed on Matrigelcoated filters. Urokinase and urokinase receptor transcripts in hepatocarcinoma were detected by reverse transcription-polymerase chain reaction. Activated hepatocyte growth factor was detected by Western blot with a hepatocyte growth factor-]/chain-specific antibody. Results: HepG2 cells expressed urokinase mRNA and secreted active urokinase. Urokinase expression was enhanced by hepatocyte growth factor at the protein
and mRNA level. Notably, cell-surface-associated urokinase was increased 22-fold by hepatocyte growth factor. Hepatocyte growth factor also increased urokinase receptor mRNA expression. B428, a urokinase inhibitor, decreased by up to 70% HepG2 invasion induced by myofibroblasts and by 90% that induced by recombinant hepatocyte growth factor. This was not due to a decrease in the generation of activated hepatocyte growth factor by myofibroblasts. Finally, all 17 hepatocarcinoma samples tested expressed urokinase and urokinase receptor transcripts. Conclusion: Hepatocyte growth factor-dependent, myofibroblasts-induced invasion of HepG2 cells is secondary to the induction of urokinase expression on tumor cells.
EPATOCELLULARcarcinoma (HCC) is the main type of liver cancer and is developed from hepatocytes. HCC is associated with a high rate of intrahepatic invasion. The strorna of HCC is infiltrated with numerous myofibroblasts (1). Using an in vitro co-culture model, we have shown that human liver myofibroblasts strongly promoted invasion of two HCC cell lines through Matrigel-coated filters (2). We found that this effect was mediated by hepatocyte growth factor/scatter factor (HGF/SF), secreted by myofibroblasts. HGF/SF is a heparin-binding polypeptide, trans-
lated as a 92-kDa single-chain inactive precursor. HGF/SF is proteolytically activated to form two subunits, alpha (60 kDa) and beta (32-36 kDa), linked by a unique disulfide bond. HGF/SF is a multifunctional protein with mitogenic (3), motogenic (4) and morphogenic (2,5) properties. The specific HGF/SF cell receptor is a 190 kDa transmembrane tyrosine kinase receptor encoded by the proto-oncogene c-met (6). Numerous studies have shown that HGF/SF/c-Met signalling can induce the invasiveness and metastatic potential of several carcinoma cell types (7,8). The molecular mechanisms underlying HGF-induced invasion are not well known. Thus, although HGF has been shown to induce the secretion of urokinase-type plasminogen activator (u-PA) (9,10) and matrix metalloproteinases (11), the role of these proteases in HGF-induced invasion has not been ascertained. Urokinase is able to activate plasminogen to plasmin
H
Received 8 June; revised 5 October; accepted 20 October 1998
Correspondence: Jean Rosenbaum, Groupe de Recherches pour l'Etude du Foie, Universit6 Victor Segalen Bordeaux 2, 146 rue L6o Saignat, 33076 Bordeaux, France. Tel: 33 5 57571771; Fax: 33 5 56514077; E-mail: [email protected]
Key words: Cancer; Hepatocyte growth factor; Invasion; Liver; Urokinase
511
A. Monvoisin et al.
which possesses a broad-spectrum proteolytic activity. u-PA is synthesized and secreted as a single-chain proenzyme (pro-u-PA) (12) that is cleaved in a two-chain active form. Pro-u-PA and u-PA bind to the u-PA receptor (u-PAR) (13), a heavily glycosylated protein attached to the cell membrane by a glycosyl phosphatidyl inositol anchor. Binding of pro-u-PA to u-PAR allows for u-PA activation, although a recent study suggested that uPAR-independent activation can also occur (14). In this study, we have investigated whether urokinase is involved in HepG2 cells invasion in vitro when induced by liver myofibroblasts or HGE
Materials and Methods Materials Culture medium, additives and Moloney Murine Leukemia Virus Reverse Transcriptase (MMLV-RT) were from Gibco-BRL (Life Technologies, Cergy-Pontoise, France). Human AB serum was from Centre R6gional de Transfusion Sanguine (Bordeaux, France). The RNeasy mini kit was from Qiagen (Les Ulis, France), HybondTM-N + membrane and ECL kit from Amersham (Les Ulis, France). [a-32p]dCTP and [y_32p] dATP were from ICN Biomedicals (Orsay, France). The Ready-to-go random priming kit was from Boehringer-Mannhein (Meylan, France). Taq polymerase was from Promega (Charbonni6res, France). Plasminogen and the plasmin substrate $2251 were from Chromogenix (M61ndal, Sweden), human urokinase from Sanofi Winthrop (Gentilly, France). Heparin-Sepharose was from Pharmacia-Biotech (Orsay, France), and polyvinylidene difluoride membranes from Millipore (Saint-Quentin en Yvelynes, France). Polycarbonate filters were from Falcon (Becton Dickinson, Le Pont de Claix, France) and Matrigel basement membrane matrix from Collaborative Biomedical Products (Bedford, MA, USA). Recombinant human HGF (rhHGF) was from R&D Systems (Oxon, UK). Chemicals were from Sigma Aldrich (St Quentin Fallavier, France). The HGF-fl chain rabbit polyclonal antibody was a generous gift from Dr. A. Galvani (Pharmacia-UpJohn, Milan, Italy). The u-PA inhibitor B428 was a kind gift from Dr. B.A. Littlefield of Eisai Research Institute (Andover, MA, USA) (15). B428 was made as a 40 mM stock concentration in dimethyl sulfoxyde, then diluted into water to 4 mM, and finally into Dulbecco's modified Eagle's medium (DMEM). In our experiments, the final concentration of DMSO did not exceed 0.0375%. Cells and cell culture The HepG2 human HCC cell line was cultured in DMEM containing 10% fetal calf serum. Human hepatic myofibroblasts (MF) were obtained from explants of non-tumoral liver resected during partial hepatectomy and characterized as myofibroblasts as previously described (16,17). MF were grown in DMEM containing 5% fetal calf serum, 5% pooled human AB serum and 5 ng/ml recombinant human epidermal growth factor. Northern blot Total RNA was prepared from cultured HepG2 cells using the RNeasy mini kit. 15 and 20/tg, respectively for u-PA and u-PAR mRNA expression, were electrophoresed on a 0.8% agarose/formaldehyde gel and transferred to a HybondTM-N+ membrane, using a Vacuum Blotter (Appligene, Illkirch, France). Prehybridization and hybridization were carried out at 65°C in Na2HPO4/H3PO4 0.5 M, pH 7.2, SDS 7%, EDTA 1 raM, BSA 1 g/100 ml, ssDNA 100/tg/ml. The membrane was hybridized with a cDNA probe for human u-PA or u-PAR, labeled with [c~-32p] dCTP. The blots were washed in stringent conditions and exposed for autoradiography. The u-PA and u-PAR plasmids used were a kind gift from Dr B. S. Nielsen (The Finsen
512
Institute, Copenhagen, Denmark) and included nucleotides 32q527 and 497-1081 for u-PA and u-PAR, respectively (18). To control for variations in RNA loading and transfer, the blots were rehybridized with a 28 S rRNA-specific oligonucleotide probe labeled with [7-32p] dATE The autoradiographic signals were quantified with the NIH Image 1.60 software on a Macintosh computer. Reverse-transcription polymerase ehain reaction Total RNA was prepared from HepG2 cells and from frozen tumoral liver samples collected from 17 patients undergoing liver resection or transplantation for hepatocellular carcinoma. One microgram of RNA was reverse-transcribed in a 50 #1 volume using MMLV-RT. Two microliters of the reaction were used for amplification with the following specific primers: 5'-CAG ATC TGA TGC TCT TCA GCT3' (uPA sense primer), 5'-GTG ACT TCA GAG CCG TAG TAG-3' (uPA antisense primer), 5'-GCC CTG GGA CAG GAC CTC TG-3' (uPAR sense primer), 5'-CAT TGA TTC ATG GGG CCT CGG C-3' (uPAR antisense primer). Thirty-five cycles of PCR were performed, consisting each of denaturation for 1 min at 94°C, annealing for 1 min at 58.5°C for u-PA (65°C for u-PAR), and elongation for 2 min at 72°C. PCR was performed in 50 id of reaction buffer containing 50 mM KC1, 10 mM Tris-HC1, pH 9, 0.1% Triton X-100, 2.4 mM MgC12, 0.4 mM dNTPs, 1.25 U Taq polymerase and 10 pmol of primers. Ten microliters of the reaction were analyzed by agarose gel electrophoresis. The sizes of the predicted products are 747 and 613 bp, respectively, for u-PA and u-PAR. For negative controls, reverse transcriptase was omitted during the RT procedure, or PCR was performed without cDNA. Secreted urokinase assays and zymography HepG2 cells were seeded in 24-well dishes at 2 ×105 cells per well in DMEM 2% FCS. After 6 h, the cells were rinsed once with Gey's balanced salt solution (GBSS) CaZ+/Mg2+, once with an acid buffer (50 mM glycine, 100 mM NaC1, pH 3) to remove cell surface-bound urokinase, and twice with serum-free DMEM. Finally, cells were incubated in 0.5 ml serum-free DMEM for 48 h. Experiments were done in triplicate. Conditioned media (CM) were first centrifuged at 600×g for 4 min at 4°C to remove cellular debris, then at 15000×g for 5 min at 4°C and stored at -20°C. Six microliters of the supernatants were electrophoresed in a 10% SDS-polyacrylamide gel containing 1 mg/ml gelatin and 12.5/~g/ml plasmin-free plasminogen. After washing with 2.5% Triton X-100, the gel was incubated at 37°C for 48 h in 100 mM glycine, 20 mM EDTA, pH 8.3. The gel was finally stained with 0.5% Coomassie blue. Proteolytic activity was detected as a white zone in a dark field. As a control, the same experiment was done in a gel without plasminogen. Purified human urokinase was used as a positive control. To explore the effect of HGF on u-PA synthesis by HepG2 cells, HepG2 cells were seeded as above and incubated in 1 ml serum-free DMEM with or without rhHGF at 100 ng/ml. After 48 h, CM were collected and cells were rinsed twice with 100 /tl of serum-free DMEM. Endogenous cell surface-associated u-PA was eluted at room temperature for 3 min with 100/zl of the acid buffer described above. The eluates were neutralized with 25/d of 0.5 M Tris-HC1, pH 7.8 and centrifuged. The cell layer was finally processed for DNA measurement according to Labarca & Paigen (19). Following normalization for DNA content of the cell layer, u-PA activity in supernatants and eluates was assayed, u-PA activity on the zymography was quantified with the NIH Image 1.60 software. Plasminogen activation HepG2 cells were seeded as described above and were incubated for 24 h in 500/tl of serum-free DMEM with or without rhHGF at 100 ng/ml. At the end of the incubation, media were removed and cells were rinsed with phosphate buffer saline (PBS). One hundred microliters of plasmin-free plasminogen at 250/~g/ml in 0.15 M NaC1, with or without 15/~M B428, were added to the cells for 3 h at 37°C. Supernatants were collected and centrifuged and the cells were rinsed with PBS. Plasmin bound to the cell layer was then eluted with 120 /A of 3 mM tranexamic acid in PBS Ca2+/Mg2+ (20). Plasmin activity
Urokinase and H C C invasion was assayed in supernatant and eluate samples by incubating 120/zl of each sample with 50/A of the plasmin-sensitive chromogenic substrate $2251 at 4 mM, for 2 h 30 min in a 96-well plate. Plasmin generated was evaluated by its amidolytic activity on $2251 (21). The cell layer was processed for D N A measurement. Cell invasion assay Tests were performed as described previously (2) with slight modifications. Eight-micrometer polycarbonate pore size filters were coated with a uniform layer o f Matrigel basement membrane matrix (16/~g/ cm2). Myofibroblasts 50× 10 3 w e r e seeded in the lower compartment of the system and 45× 103 HepG2 ceils were seeded onto the filters in 2°,/0 FCS D M E M . The final concentration of FCS was 0.4%. In some experiments, r h H G F (100 ng/ml) was used instead of myofibroblasts in the lower compartment. After 48 h, the cells on the upper surface o f the filter were wiped with a cotton swab. Filters were fixed for 10 min with methanol and stained with hematoxylin. Cells that invaded the lower surface of the filter were counted under a photonic microscope at a final magnification of 320. Results were expressed as number of HepG2 cells invaded per filter. Experiments were done in triplicate and results are shown as mean_+l SD. Cytotoxicity assay Myofibroblasts were seeded into 24-well dishes and HepG2 cells into 96-well dishes at 50× 103 or 45× 103 cells per well, respectively, in 2% FCS D M E M with or without B428. Cells were incubated for 48 h at 37°C. After this time, medium was removed and cells were incubated with D M E M containing 1 mg/ml of 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT). After 2 h, cells were solubilized with DMSO and the optical density at 540 nm was measured in a microplate reader. Measure of HGF activation Confluent monolayers of M F were washed with serum-free medium and incubated for 2 h with serum-free medium. This medium was discarded and the ceils were then further incubated for 24 h in serumfree medium, in the absence or in the presence of 15/~M B428. Medium was collected, centrifuged, and heparin-binding proteins were adsorbed to heparin-Sepharose, as described (2). Heparin-bound proteins, or r h H G F were analyzed by Western blot as previously described (2), using the H G F fl-chain-specific antibody, diluted 1:2500.
Results Urokinase and urokinase receptor expression by HepG2 cells Northern blot analysis showed that HepG2 cells expressed a single 3.1 kb transcript corresponding to uPA m R N A (Fig. 1A). RT-PCR analysis showed that HepG2 cells also expressed u-PA receptor m R N A (Fig. 1B). Controls where the enzyme was omitted during the RT step or where cDNA was omitted for the PCR step were negative. The ability of HepG2 cells to secrete urokinase was determined by zymography on a SDS-polyacrylamide gel containing plasminogen and gelatin. Gel lysis was detected with CM from HepG2 cells (Fig. 1C). This was indeed plasminogen-dependent lysis since it was not found when the experiment was done in a gel without plasminogen (data not shown). The lysis activity had a relative mobility which corresponded to approximately 55 kDa and co-migrated with purified human urokinase. No bands corresponding to t-PA were observed. Finally, this activity
was inhibited when incubation of the gel was done in the presence of 100 ~tM B428, a urokinase-specific inhibitor derived from amiloride (data not shown). Taken together, these results indicate that HepG2 ceils are able to produce active urokinase. Effect of HGF/SF on u-PA and u-PAR expression by HepG2 cells Northern blot analysis showed that H G F increased the u-PA mRNA level in HepG2 cells. Following normalization with 28 S RNA, a 6-fold increase was observed after 6 h and 2.2-fold at 24 h (Fig. 2). Moreover, incubation of HepG2 cells with HGF/SF led to an in-
A
3,1 kb ----~
B
1
C
I
2
3
2
Fig. 1. Expression of u-PA and u-PAR mRNA in HepG2 cells. A, Total RNA from HepG2 cells was analyzed by Northern blot with a u-PA cDNA probe. B, RT-PCR analysis of HepG2 RNA with u-PAR-specific primers. Lane 1: molecular size markers; Lane 2: negative control; Lane 3:HepG2 cells. C, Zymography of HepG2 cells conditioned medium for u-PA detection. Lane 1:HepG2 cells; lane 2: purified human urokinase (10 mU).
6 hrs -
24 hrs ÷
-
÷
u-PA mFINA
28S
---~
Fig. 2. Effect of HGF on u-PA mRNA expression byHepG2. HepG2 cells were incubated in the absence ( - ) or the presence (+) of human recombinant HGF, and total RNA was extracted at the time indicated. The Northern blot was hybridized sequentially with a u-PA and a 28 S probe. 513
A. Monvoisin et al.
A
_
A
÷
6000 5000 4000
B o ~
3000
ooo
Z.~ Fig. 3. Effect of H G F on u-PA production by HepG2 cells. u-PA activity in the supernatant (A) and on the cell surface (B) of HGF-stimulated ( + ) or unstimulated ( - ) HepG2 cells was assayed by zymography in triplicate. One experiment representative of three is shown.
lOOO
0
MF : B428(gM)
:
-
+
+
+
+
"
"
6
10
15
B 6 hrs
24 hrs
6000
i
-
+
-
+
UPAR1 mRNA
5000
l
4000 UPAR2 mRNA
o I~w
3000 2000
Z.. =
1000
28S --~
0 Fig. 4. Effect of H G F on u-PAR m R N A expression byHepG2. HepG2 cells were incubated in the absence ( - ) or the presence ( + ) of human recombinant HGF, and total R N A was extracted at the time indicated. The Northern blot was hybridized sequentially with a u-PAR and a 28 S probe.
crease in the expression of active u-PA. u-PA expression in the cell supernatant was enhanced by 2.8_+ 1.4fold (mean_+ 1 SD), while cell-surface associated u-PA was increased by 22_+2.4-fold after HGF-treatment (p<0.01, Student's t-test) (Fig. 3). As shown on Fig. 4, H G F also upregulated the level of the 1.4 kb u-PAR m R N A at 6 and 24 h. This transcript, called uPAR 1, codes for the full-length cell surface u-PAR (13). On the other hand, H G F had only a 514
HGF: B428(l.tM)
. :
-
+ "
+ 15
Fig. 5. Effect of urokinase inhibitor B428 on HepG2 cells invasion in vitro. HepG2 were seeded in invasion chambers coated with Matrigel as described in Materials and Methods. Experiments were done using either M F (A) or human recombinant HGF at 100 ng/mI final concentration (B) in the lower chamber. Experiments with M F were done in the presence of different concentrations ofB428. Experiments with HGF were done with B428 at 15/~M. The total number of cells migrating through the filter was counted. Results are expressed as mean +_1 SD of three experiments done in triplicate.
Urokinase and HCC invasion
minor effect on the level of the 1.1 kb u-PAR mRNA. This transcript, called uPAR2, is a variant generated by alternative splicing described by Pyke et al., which probably encodes a soluble form of uPAR (22). Effect of HGF/SF on cell-surface activation of plasminogen After addition of purified human plasminogen to HGF/SF-stimulated HepG2 and unstimulated HepG2 cells, plasmin activity could be recovered in the supernatant and as a cell-bound fraction. Plasmin activity was enhanced by HGF-treatment by 2.9_+0.7 (mean_+ 1 SD; p<0.02, Student's t-test) and 3.5___l.l-fold (p<0.05) in the supernatant and on the cell surface, respectively, as compared with control cells. When B428 was added together with plasminogen, plasminogen activation was inhibited in the supernatant and on the cell surface by 87.6 and 96.2%, respectively. Effect of B428 on HepG2 cells invasion in vitro We finally tested whether u-PA was involved in HepG2 cells invasivity. As shown on Fig. 5A, co-incubation of HepG2 cells with hepatic myofibroblasts led to a 77fold increase in invasivity. Addition of B428 caused a dose-dependent inhibition of the HepG2 cells invasion induced by myofibroblasts. At 15/zM B428, the effect of myofibroblasts was reduced by 69.5%. Similar resuits were obtained when rhHGF was substituted for myofibroblasts, rhHGF stimulated the invasion of HepG2 cells 120-fold. Addition of B428 at 15/zM decreased this effect by 91% (Fig. 5B). No inhibition of HepG2 cells invasion was found with DMSO, the solvent of B428 (data not shown). The decreased invasiveness was not due to a cytotoxic effect of B428 on HepG2 or myofibroblasts, since B428 did not modify MTT uptake by these two cell types at the concentrations used (data not shown). Effect of B428 on HGF activation As u-PA is known to be able to activate pro-HGF into its active heterodimeric form, we checked whether B428 prevented the generation of the H G F heterodimer by ME Fig. 6A shows that addition of B428 to cultured MF did not modify the amount of fl-chain in the cultures. In addition, we also showed that a significant part of the rhHGF used in this study is already in active form (Fig. 6B). Finally, ELISA assays demonstrated that B428 did not modify H G F secretion by MF (not shown). u-PA and u-PAR expression in human HCC Total R N A was extracted from 17 human HCC samples and assayed by RT-PCR with u-PA and u-PAR
A
1
13
2
Fig. 6. Effect of B428 on HGF activation. A. HeparinSepharose-concentrated conditioned medium from MF was analyzed by Western blot with an antibody to the fl chain of HGF. Lane 1: control MF-CM; Lane 2: CM prepared in the presence of 15 IzM B428. B. rhHGF standard (50 ng). Arrows point to the pro-HGF above, and the HGF-fl chain doublet below. The arrowhead indicates a non-specific signal due to the large amount of bovine serum albumin present in the solution of rhHGF.
A
B
1 2 8 4 8
$
78
g 10 11 12 18 14 16 18 17 18 1 |
Fig. 7. Expression of u-PA and u-PAR mRNA in HCC. cDNA reverse transcribedfrom total RNA extracted from HCC samples was amplified using either u-PA (A) and uPAR-specific (B) primers. Lane 1: molecular size markers; Lane 2: negative control; Lane 3 to 19: HCC samples.
primers. As shown on Fig. 7, both transcripts were detected in every sample. Controls were negative.
Discussion In this study, we have shown that the human HCC cell line HepG2 expresses u-PA and its receptor, and that u-PA is responsible for the in vitro invasivity of HepG2 cells across Matrigel, when induced by co-culture with liver MF or by purified human H G E u-PA expression was demonstrated by Northern blot and zymography. 515
A. Monvoisin et al.
Zymography showed a single band of proteolytic activity that co-migrated with purified human u-PA and which was abolished in the presence of B428, a specific u-PA inhibitor derived from amiloride. No other plasminogen activator species was detected in HepG2-conditioned medium. Membrane-bound u-PA was biologically active since it was able to promote conversion of plasminogen to plasmin. This activity was up to 96% inhibited by B428, indicating clearly that u-PA accounted for almost all of the plasminogen activator activity in our model. These results differ from those of Wojta et al. (23), who were unable to detect u-PA expression in HepG2 cells, using Northern blot and a specific ELISA. The reasons for this discrepancy could be a lack of sensitivity of their assays or the use of a different subclone of HepG2. In further experiments, we demonstrated that the expression of u-PA by HepG2 cells at the mRNA and protein level was enhanced by exposition of the cells to HGE HGF specially increased the amount of membrane-bound u-PA by more than 20-fotd. This is potentially important as it is membrane-bound u-PA that is specifically involved in the pericellular proteolysis necessary for cell invasion (24,25). Our data also show that H G F treatment increases the expression of u-PAR transcripts in HepG2 cells. Such an increase in u-PAR expression likely participates in the relatively selective increase in membrane-bound u-PA, by localizing u-PA to the cell surface. Up-regulation of u-PA by HGF has previously been demonstrated in other cell types, such as M D C K cells (10), or human leiomyosarcoma cells transfected with a HGF expression vector (9). Recently, Kamiyama et al. have shown that H G F upregulated uPA antigen secretion in 1 out of 3 human HCC cell lines (26). However, the increase was very modest, about 1.2-fold over baseline. Moreover, in all of the above studies, no data allow the increased u-PA expression to be linked to the pro-invasive effect of HGE Such a link has been established in our study since we have shown that the u-PA-specific inhibitor B428 blocked HepG2 cells invasivity, when induced by coculture with ME or by recombinant human HGE We have previously shown that MF-induced invasivity was dependent on the secretion of H G F by ME since it was completely inhibited by anti-HGF antibodies (2). We have now further established that HGF-dependent, MF-induced, HepG2 invasiveness is secondary to the induction of u-PA expression on tumor cells. B428 inhibited rHGF-induced invasion more efficiently than MF-induced invasion. This suggests that, besides HGF, MF secrete other pro-invasive molecules that cooperate with H G F and act via a u-PA-independent pathway. The downstream mechanism of action of u-PA in 516
our system has not been completely elucidated, u-PA could promote invasion through plasminogen conversion into plasmin, with subsequent direct degradation of extracellular matrix proteins, or activation of the latent forms of matrix-metalloproteinases (27). Another possibility is that B428 acts mainly by preventing the u-PA-mediated conversion of pro-HGF to its active heterodimeric form (28). Our data argue against this hypothesis since the amount of HGF fl-chain, a corollary of HGF activation, was unchanged in the presence of B428, indicating that activation of HGF in these cultures is largely u-PA-independent. Moreover, B428 inhibited the pro-invasive effect of recombinant H G F as efficiently as that of myofibroblasts, and we have found that a large fraction of the commercial recombinant H G F was already in heterodimeric form. The mechanism of HGF-activation in MF-conditioned medium is unknown but could be due to proteolysis by the HGF activator (29). Finally, B428 could act by inhibiting the activation of the pro-form of matrixmetalloproteinase-2 (MMP-2) by urokinase (30). MMP-2 is involved in extracellular matrix breakdown and invasion (31), and has recently been shown to be overexpressed in human HCC (32). Our data also indicate that u-PA is actively transcribed in human HCC. We show too that u-PA receptor mRNA is expressed in human HCC. This is in good agreement with a recent report showing, using in situ hybridization and immunohistochemistry, that uPAR is present in tumoral hepatocytes along the tumorstroma interface (33). While this manuscript was being submitted, De Petro et al. confirmed the expression of both transcripts in HCC and showed, using a quantitative RT-PCR assay that u-PA trancript levels were much higher in HCC than in the surrounding non-tumoral liver. This is compatible with our own data, using in situ hybridization (manuscript in preparation). They were also able to correlate high u-PA levels with a poor prognosis. Altogether, these results suggest that the u-PA pathway is implicated in HCC. Thus, u-PA inhibitors could conceivably be of benefit in limiting HCC invasiveness in the clinical setting. Interestingly, the u-PA inhibitor B428 used in our study has shown anti-invasive effects in animal models of breast or prostatic cancer (34,35). We are currently testing its efficacy in a rat HCC model.
Acknowledgements We wish to express our deep thanks to Bruce A. Littlefield for providing B428, Boye Schnack Nielsen for the plasmids, Arturo Galvani for the HGF fl-chainspecific antibody, and Genevieve N'Guyen for her help in starting this study. This work was supported by
Urokinase and H C C invasion
grants from Comit6s de la Gironde et de la Dordogne from the Ligue Nationale Franqaise contre le Cancer, Association pour la Recherche sur le Cancer, Groupement des Entreprises Franqaises pour la Lutte contre le Cancer, and Conseil r6gional d'Aquitaine. AM was a recipient of a fellowship from the Comit6 de la Dordogne from the Ligue Nationale Fran~aise contre le Cancer.
References 1. Enzan H, Himeno H, Iwamura S, Onishi S, Saibara T, Yamamoto Y, et al. a-smooth muscle actin positive perisinusoidal stromal ceils in human hepatocellular carcinoma. Hepatology 1994; 19: 895-903. 2. Neaud V, Faouzi S, Guirouilh J, Le Bail B, Balabaud C, BioulacSage P, et al. Human hepatic myofibroblasts increase invasiveness of hepatocellular carcinoma cells: evidence for a role of hepatocyte growth factor. Hepatology 1997; 26: 1458-86. 3. Michalopoulos G, Houck KA, Dolan ML, Luetteke NC. Control of hepatocyte replication by two serum factors. Cancer Res 1984; 44: 4414-9. 4. Stoker M, Gherardi E, Perryman M, Gray J. Scatter factor is a fibroblast-derived modulator of epithelial cell mobility. Nature 1987; 327: 239-42. 5. Montesano R, Matsumoto K, Nakamura T, Orci L. Identification of a fibroblast-derived epithelial morphogen as hepatocyte growth factor. Cell 1991; 67: 901-8. 6. Bottaro DP, Rubin JS, Faletto DL, Chan AML, Kmiecik TE, Vande Woude GE et al. Identification of the hepatocyte growth factor receptor as the c-met proto-oncogene product. Science 1991; 251: 802-4. 7. Hasegawa Y, Yamamoto M, Maeda S, Saitoh Y. Hepatocyte growth factor and its receptor/c-MET regulate both cell growth and invasion of human pancreatic cancer. Int J Oncol 1995; 7: 877-81. 8. Matsumoto K, Date K, Shimura H, Nakamura T. Acquisition of invasive phenotype in gallbadder cancer cells via mutual interaction of stromal fibroblasts and cancer cells as mediated by hepatocyte growth factor. Jpn J Cancer Res 1996; 87: 702-10. 9. Jeffers M, Rong S, Vande Woude GE Enhanced tumorigenicity and invasion-metastasis by hepatocyte growth factor/scatter factor-Met signalling in human cells concomitant with induction of the urokinase proteolysis network. Mol Cell Biol 1996; 16:111525. 10. Pepper MS, Matsumoto K, Nakamura T, Orci L, Montesano R. Hepatocyte growth factor increases urokinase-type plasminogen activator (u-PA) and u-PA receptor expression in Madin-Darby canine kidney epithelial cells. J Biol Chem 1992; 267: 20493-6. 11. Uchiyama A, Essner R, Doi F, Nguyen T, Ramming KP, Nakamura T, et al. Interleukin 4 inhibits hepatocyte growth factorinduced invasion and migration of colon carcinomas. J Cell Biochem 1996; 62: 443-53. 12. Kasai S, Arimura H, Nishida M, Suyama T, Primary structure of single-chain pro-urokinase. J Biol Chem 1985; 260: 12382-9. 13. Roldan AL, Cubellis MV, Masucci MT, Behrendt N, Lund LR, Dano K, et al. Cloning and expression of the receptor for human urokinase plasminogen activator, a central molecule in cell surface, plasmin dependent proteolysis. EMBO J 1990; 9: 467-74. 14. Carmeliet P, Moons L, Dewerchin M, Rosenberg S, Herbert JM, Lupu E et al. Receptor-independent role of urokinase-type plasminogen activator in pericellular plasmin and matrix metaUoproteinase proteolysis during vascular wound healing in mice. J Cell Biol 1998; 140: 233-45. 15. Towle M J, Lee A, Maduakor EC, Schwartz E, Bridges A J, Littlefield BA. Inhibition of urokinase by 4-substituted benzo[b]-
16.
17.
18.
19. 20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
thiophene-2-carboxamidines: an important new class of selective synthetic urokinase inhibitors. Cancer Res 1993; 53: 2553-9. Win KM, Charlotte F, Mallat A, Cherqui D, Martin N, Mavier P, et al. Mitogenic effect of TGFI31 on human Ito cell in culture: evidence for mediation by endogenous platelet derived growth factor. Hepatology 1993; 18: 137-45. Blazejewski S, Pr6aux AM, Mallat A, Broch6riou I, Mavier P, Dhumeaux D, et al. Human myofibroblast-like cells obtained by outgrowth are representative of the fibrogenic cells in the liver. Hepatology 1995; 22: 788-97. Pyke IC, Kristensen P, Ralfkiaer E, Grondahl-Hansen J, Eriksen J, Blasi E et al. Urokinase-type plasminogen activator is expressed in stromal cells and its receptor in cancer cells at invasive loci in human colon adenocarcinomas. Am J Pathol 1991; 138: 1059-67. Labarca C, Paigen K. A simple, rapid, and sensitive DNA assay procedure. Anal Biochem 1980; 102: 344-52. Stephens RW, P611~inen J, Tapiovaara H, Leung KC, Sim PS, Salonen EM, et al. Activation of pro-urokinase and plasminogen on human sarcoma cells: a proteolytic system with surface-bound reactants. J Cell Biol 1989; 108: 1987-95. Lottenberg R, Christensen U, Jackson MC, Coleman PL. Assay of coagulation proteases using peptide chromogenic and fluorogenic substrates. Methods Enzymol 1981; 80: 341-81. Pyke C, Eriksen J, Solberg H, Schnack Nielsen B, Kristensen P, Lund LR, et al. An alternatively spliced variant of mRNA for the human receptor for urokinase plasminogen activator. FEBS Lett 1993; 326: 69--73. Wojta J, Nakamura T, Fabry A, Hufnagl P, Beckmann R, McGrath K, et al. Hepatocyte growth factor stimulates expression of plasminogen activator inhibitor type 1 and tissue factor in HepG2 cells. Blood 1994; 8: 151-7. Cohen RL, Xi XP, Crowley CW, Lucas BK, Levinson AD, Shuman MA. Effects of urokinase receptor occupancy on plasmin generation and proteolysis of basement membrane by human tumor cells. Blood 1991; 78: 479-87. Hollas W, Blasi E Boyd D. Role of the urokinase receptor in facilitating extracellular matrix invasion by cultured colon cancer. Cancer Res 1991; 51: 3690-5. Kamiyama T, Une Y, Uchino J, Hamada J. Hepatocyte growth factor enhances the invasion activity of human hepatocellular carcinoma cell lines. Int J Oncol 1998; 12: 655-9. Baramova EN, Bajou K, Remacle A, L'Hoir C, Krell HW, Weidle UH, et al. Involvement of PA/plasmin system in the processing of pro-MMP-9 and in the second step of pro-MMP-2 activation. FEBS Lett 1997; 405: 157-62. Naldini L, Tamagnone L, Vigna E, Sachs M, Hartmann G, Birchmeier W, et al. Extracellular proteolytic cleavage by urokinase is required for activation of hepatocyte growth factor/scatter factor. EMBO J 1992; 11: 4825-33. Miyazawa K, Shimomura T, Kitamura A, Kondo J, Morimoto Y, Kitamura N. Molecular cloning and sequence analysis of the cDNA for a human serine protease responsible for activation of hepatocyte growth factor. J Biol Chem 1993; 268: 10024-8. Keski-Oja J, Lohi J, Tuuttila A, Tryggvason K, Vartio T. Proteolytic processing of the 72 000-Da type IV collagenase by urokinase plasminogen activator. Exp Cell Res 1992; 202: 471-6. Abe T, Mori T, Kohno K, Seiki M, Hayakawa T, Welgus HG, et al. Expression of 72 kDa type IV collagenase and invasion activity of human glioma cells. Clin Exp Metast 1994; 12: 296304. Musso O, Th6ret N, Campion JP, Turlin B, Milani S, Grappone C, et al. In situ detection of matrix metalloproteinase-2 (MMP-2) and the metalloproteinase inhibitor TIMP2 transcripts in human primary hepatocellular carcinoma and in liver metastasis. J Hepatol 1997; 26: 593-605. Morita Y, Hayashi Y, Wang Y, Kanamaru T, Suzuki S, Kawasaki K, et al. Expression of urokinase-type plasminogen activator receptor in hepatocellular carcinoma. Hepatology 1997; 25: 85661.
517
A. Monvoisin et al. 34. Xing RH, Mazar A, Henkin J, Rabbani A. Prevention of breast cancer growth, invasion, and metastasis by antiestrogen tamoxifen alone or in combination with urokinase inhibitor B-428. Cancer Res 1997; 57: 3585-93.
518
35. Alonso DE Farias EE Ladeda V, Davel L, Puricelli L, Bal De Kier Joff6 E. Effects of synthetic urokinase inhibitors on local invasion and metastasis in a routine mammary tumor model. Cancer Res 1996; 40: 209-23.