By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
Hepatocyte growth factor (HGF) is the most powerful hepatotrophic factor identified so far. However, the ability of HGF to promote tumor cell “scattering” and invasion raises some concern about its therapeutic safety. We compared the therapeutic efficacy of HGF with that of Metron Factor-1 (MF-1), an engineered cytokine derived from HGF and the HGF-like factor macrophage stimulating protein (MSP), in mouse models of acute and chronic liver injury. At the same time, we tested the ability of HGF and MF-1 to promote tumor growth, angiogenesis, and invasion in several mouse models of cancer. We show that (1) MF-1 and HGF stimulate hepatocyte proliferation in vitro; (2) MF-1 and HGF protect primary hepatocytes against Fas-induced and drug-induced apoptosis; (3) HGF but not MF-1 induces scattering and matrigel invasion of carcinoma cell lines in vitro; (4) HGF but not MF-1 promotes migration and extracellular matrix invasion of endothelial cells in vitro; (5) MF-1 and HGF prevent CCl4-induced acute liver injury as measured by alanine aminotransferase (ALT) levels, histology, terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) analysis, and phospho-histone-3 immunostaining; (6) MF-1 and HGF attenuate liver fibrosis caused by chronic CCl4 intoxication and promote regeneration as measured by Sirius red staining, alpha-smooth muscle actin immunostaining, and Ki-67 analysis; (7) HGF but not MF-1 promotes tumor growth, angiogenesis, and metastasis in a variety of xenograft models; (8) HGF but not MF-1 promotes intrahepatic dissemination of hepatocarcinoma cells injected orthotopically. Conclusion: These data suggest that MF-1 is as effective as HGF at preventing liver injury and at promoting hepatocyte regeneration, but therapeutically safer than HGF because it lacks proangiogenic and prometastatic activity. (HEPATOLOGY 2008;47:2010–2025.)
Hepatocyte growth factor (HGF), also known as scatter factor, is a pleiotropic cytokine that masters a characteristic biological program known as “invasive growth.”1 This complex process results from the harmonic coordination of several biological activities, including cell proliferation, motility, survival, and morphogenesis.2–5 While HGF is secreted by cells of mesenchymal origin, its high-affinity receptor, the Met tyrosine kinase, is expressed by a broad range of tissues, including epithelial, endothelial, hemopoietic, muscle, and neuronal cells. HGF/Met signaling is essential during embryo development, during which it drives uterus invasion by the trophoblast, neo-angiogenesis in the developing organs, and migration of differentiating precursor cells of diverse origins.6–8 Homozygous deletion of the hgf or met gene leads to in utero embryo death because of placental and hepatic defects.9–11
In the adult, the HGF pathway is probably silent under normal conditions, but it is resumed during particular biological processes requiring reorganization of organ architecture, including tissue regeneration and wound healing.12–15 Such reorganization also occurs in pathological conditions, for example, during tumor progression, where HGF is known to sustain cancer cell proliferation, survival, and motility, as well as to promote host vessel remodeling.5, 16, 17 HGF/Met signaling is one of the most frequently activated tyrosine kinase pathways in human cancer.18 In a significant fraction of human tumors, Met is found activated by different mechanisms, including receptor overexpression, point mutation, and autocrine HGF stimulation.1, 5 Met overexpression often results in increased sensitivity to environmental HGF. For example, it has been shown that hypoxia induces transcriptional activation of the c-met proto-oncogene, thus sensitizing cells to HGF stimulation and increasing tumor invasion.19 Similarly, point-mutated forms of Met found in hereditary and sporadic papillary renal carcinomas and in malignant lesions of the gastrointestinal tract lie in a semi-activated state, needing ligand stimulation to display their full transforming potential.20, 21
The ability of HGF to prevent cell death and to promote tissue regeneration makes this pleiotropic cytokine an ideal candidate for therapeutic use in regenerative medicine. Indeed, recombinant HGF—administered as a protein or as a gene—has been shown to display trophic and healing properties in a variety of organs, including liver,22 kidney,23 lung,24 skin,25 intestine,26 peripheral nerves,27 central nervous system,28 heart,29 and hemopoietic system.30 HGF has also been shown to increase graft survival in transplantation experiments.31 However, the profound involvement of the Met receptor in tumor biology and the proangiogenic and prometastatic activity of its ligand raise some concerns about the therapeutic safety of HGF. The separation of the trophic, beneficial effects of HGF from its proinvasive, adverse properties is a much debated topic in regenerative medicine32 and a key point in order to harness its healing potential for the clinic.33
We address this issue by comparing the therapeutic efficacy of HGF with that of metron factor-1 (MF-1), an engineered cytokine derived from HGF and the HGF-like factor macrophage-stimulating protein (MSP), in mouse models of acute and chronic liver injury. MF-1 is a partial agonist of Met and was shown to lack proinvasive activity in vitro.33 We show that, although both HGF and MF-1 effectively prevent liver injury and promote hepatocyte regeneration, only HGF, and not MF-1, drives tumor angiogenesis and dramatically increases the metastatic potential of cancer cells.
Mouse primary hepatocytes were prepared from 7-week-old to 8-week-old male Balb-c mice (Sankyo Labo Service, Tokyo, Japan) by the Percoll density gradient method.34 Freshly prepared cells were plated in type I collagen-coated dishes (Sigma, St. Louis, MO), cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 10−7 mol/L dexamethasone, 10 ng/mL epidermal growth factor, and 0.5 μg/mL insulin (all from Sigma) for 24 hours, and then subjected to the various analyses as indicated. MLP29 mouse liver precursor cells have been described.35 MDA-MB-435 human melanoma cells, A549 human lung carcinoma cells, and HepG2 human hepatocarcinoma cells were obtained from ATCC (Rockville, Maryland). Cells were maintained in Dulbecco's modified Eagle's medium (MDA-MB-435 and HepG2) or in Roswell Park Memorial Institute medium (A549) supplemented with 1% glutamine and 10% fetal bovine serum. CBO-140-C12 mouse hepatocarcinoma cells, a gift of Dr. Ikuo Saiki (University of Toyama, Toyama, Japan), were cultured as described.36 Human umbilical vein endothelial cells (HUVECs) were prepared and grown as previously described.37 LOC human kidney epithelial cells and T47D human breast carcinoma cells have been described.33
Recombinant Factors and Lentiviral Vectors.
Engineering of MF-1 has been described.33 The complementary DNA encoding for MF-1 or human HGF (Gene Bank # M73239) was subcloned into the pRLL.CMV lentiviral vector38 by standard genetic engineering techniques. Lentiviral stocks were produced as previously described.39 Viral p24 antigen concentration was determined by the human immunodeficiency virus-1 p24 core profile enzyme-linked immunosorbent assay (ELISA) kit (NEN Life Science Products, Boston, MA) according to the manufacturer's instructions. Cells (MDA-MB-435, A549, CBO-140-C12) were transduced in six-well plates (105 cells/well in 2 mL medium) using 40 ng/mL p24 in the presence of 8 μg/mL polybrene (Sigma) as described.39 Conditioned medium of lentiviral vector-transduced MDA-MB-435 cells containing approximately 100 nM MF-1 was subjected to standard immobilized metal ion affinity chromatography procedures, and MF-1 was purified to homogeneity as described.33 Recombinant human HGF and MSP were purchased from R&D Systems (Minneapolis, MN).
Immunoprecipitation, Western Blotting, and ELISA.
Anti-mouse Met, anti-human Met, anti-human Ron, anti-mouse Cyclin D1, anti-mouse Cdk4, anti-mouse glyceraldehyde 3-phosphate dehydrogenase, and anti-human HGF (cross-reacting with MF-1) polyclonal antibodies were obtained from Santa Cruz Biotech (Santa Cruz, CA). Anti-mouse Ron polyclonal antibodies were purchased from R&D Systems. Anti-phospho-AKT [serine-threonine AKT/protein kinase B] and anti-phospho-mitogen-activated protein kinase [p42-44/extracellular signal-regulated kinase (ERK)] monoclonal antibodies were obtained from Cell Signaling Technologies (Beverly, MA). Anti-phospho-tyrosine monoclonal antibodies were purchased from UBI (Lake Placid, New York). Affinity-purified polyclonal anti–MF-1 antibodies (nonreacting with HGF) were obtained by immunizing New Zealand rabbits with recombinant MF-1 (outsourced to Medical & Biological Laboratories, Nagoya, Japan). Immunoprecipitation and Western blotting was performed using extraction buffer (EB) as described.33 MF-1 concentration in medium or plasma was determined by sandwich ELISA using affinity-purified polyclonal anti–MF-1 antibodies for both capture (nonbiotinylated) and revealing (biotinylated) by standard antibody techniques.
In Vitro Biological Assays.
For Met and Ron phosphorylation, signal transduction, and cell cycle analysis, primary hepatocytes were plated as described in six-well plates and then incubated in mitogen-free medium (without epidermal growth factor and containing only 2% fetal bovine serum) for 24 hours. After stimulation with MF-1, HGF, or MSP, cells were harvested at the indicated times. Immunoprecipitation and Western blotting was performed as previously described33 with the appropriate antibodies. For mitogenic assays, primary hepatocytes were plated as above in 24-well plates, deprived of growth factors for another 24 hours, and then stimulated with the indicated factor concentrations. After 8 hours, 3[H]-thymidine (NEN Life Sciences Products) was added to each well (1.5 μCi/well), and cells were further incubated at 37°C for 16 hours. 3[H]-thymidine incorporation was determined by liquid scintillation counting. Survival assays were performed by pre-incubating primary hepatocytes with the indicated concentrations of HGF, MSP, or MF-1 for 24 hours. Next, cells were challenged with either 40 ng/mL anti-Fas antibodies (Medical and Biological Laboratories) for 14 hours, or with 30 nM staurosporine (Sigma) for 1 hour. Apoptosis was determined by the free nucleosome method (Cell Death Detection ELISAPLUS, Boehringer, Mannheim, Germany). For the scatter assay, cells were seeded in 96-well plates (1000 cells/well) in the presence of increasing factor concentrations (0-500 ng/mL). One day after, cells were analyzed by microscopy and photographed. Matrigel invasion assays were performed in Transwell chambers as described.19 For competition experiments, A549 cells were stimulated with 50 ng/mL HGF or MSP plus 500 ng/mL MF-1. Matrigel invasion was determined as described. Migration assays with HUVECs (4 × 104/well) were also performed in Transwell chambers using fibronectin instead of Matrigel (20 μg/well; Sigma). Bromodeoxyuridine incorporation assays with HUVECs were performed using a chemiluminescence ELISA kit (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instructions. For extracellular matrix invasion assays, HUVECs were seeded onto a layer of Matrigel (1.4 mg per well, Becton Dickinson) in 48-well plates (5 × 104 cells/well). After stimulation with 100 ng/mL MF-1, HGF, or MSP, cells were photographed and the length of pseudopodia was determined using ImageProPlus 4.0 imaging software (Media Cybernetics, Silver Spring, MD).
Acute Liver Injury Model.
All animal procedures concerning analysis of liver injury were performed in the Animal Facility of the University of Toyama Medical School following institutional guidelines and were approved by the Ethical Commission of the Faculty of Medicine. All animals received humane care according to the criteria outlined in the “Guide for the Care and Use of Laboratory Animals” prepared by the National Academy of Sciences and published by the National Institutes of Health (NIH publication 86-23 revised 1985). A micro-osmotic pump (Alzet, Cupertino, CA) set to release 20 μg/day recombinant MF-1 or HGF was implanted into the abdominal cavities of 7-week-old to 8-week-old male Balb/c mice (Sankyo Labo Service; five animals/group). Recombinant factors were dissolved in a special delivery vehicle.23 Empty osmotic pumps were implanted into control animals. One day later, a 10% solution of CCl4 in olive oil (Sigma) was injected subcutaneously in all mice to induce acute liver damage (0.1 mL/mouse). At regular intervals after acute intoxication (6, 12, 24, 48, and 72 hours), a blood sample was taken before mice were euthanized for analysis. Alanine aminotransferase (ALT) levels were determined as described.40 Livers were fixed, embedded in paraffin, and processed for histological analysis. Serial sections were stained with hematoxylin-eosin and subjected to histological evaluation by an independent pathologist not informed of sample identity. Terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) staining was performed using the ApopTag Plus kit (Chemicon, Temecula, CA). Anti-phospho-histone-3 antibodies for immunohistochemical analysis were purchased from Cell Signaling Technology. Control animals receiving factor-containing osmotic pumps but not CCl4 for a period of approximately 1 month (3 animals/group) were also analyzed with the same procedure and did not reveal any obvious sign of pathological conditions.
Chronic Liver Injury Model.
Adult (7-week-old to 8-week-old) male C57BL/6J mice were given a biweekly subcutaneous dose of a 10% solution of CCl4 in olive oil (0.2 mL per mouse) for 10 weeks. On week 6, mice were implanted intraperitoneally with micro-osmotic pumps containing MF-1 or HGF as described above. Pumps were replaced with new ones every week. Control mice were implanted with osmotic pumps containing vehicle alone. On week 10, a blood sample was taken from each mouse right before suppression, and livers were extracted for analysis. Plasma ALT levels were determined as described. Livers were fixed, embedded in paraffin, and processed for histology. Serial liver sections were stained with hematoxylin-eosin and subjected to histological evaluation as described. Sirius red staining was performed as described.40 Anti–alpha-smooth muscle actin and anti-Ki67 antibodies for immunohistochemical analysis were purchased from DAKO (Copenhagen, Denmark) and Ylem (Rome, Italy), respectively.
Subcutaneous Tumor Model.
All animal procedures concerning analysis of HGF-induced or MF-1–induced tumor growth, angiogenesis, and metastasis were performed in the Animal Facility of the Institute for Cancer Research and Treatment, Candiolo, Torino, Italy, after approval by the Ethical Commission of the University of Torino and by the Italian Ministry of Health. All animals received humane care as indicated. Analysis was performed as described.38 Briefly, MDA-MB-435 and A549 were transduced with HGF or MF-1 lentivectors as described, and stable cell lines secreting similar levels (approximately 70 pmol/106 cells/24 hours) of recombinant factors were established. Factor concentration in cell medium was determined by both ELISA and western blotting as described. Lentiviral vector–transduced tumor cells were injected subcutaneously (3 × 106 cells/mouse) into the right posterior flank of 6-week old immunodeficient nu−/− female mice on Swiss CD-1 background (Charles River Laboratories, Calco, Lecco, Italy). Tumor volume was calculated as described.38 At the end of the observation period, tumors were extracted for analysis and divided into two parts. One half was embedded in Tissue-Tek optimal cutting temperature compound (Sakura Finetek, Torrance, CA) and immediately frozen in liquid nitrogen. Frozen tumor sections were used for tissue analysis. The second half of the tumor was used for preparation of protein extracts as described.41 Immunofluorescence analysis of tumor vessels was performed on frozen sections using anti–platelet/endothelial cell adhesion molecule-1 (CD31 endothelial marker) rat monoclonal antibody (Pharmingen, San Diego, CA). Lung metastases were contrasted with black India ink and counted under a stereoscopic microscope. After scoring, lungs were fixed, embedded in paraffin, and processed for histology. Serial sections were stained with hematoxylin-eosin.
Intrahepatic Metastasis Model.
Mouse CBO-140-C12 liver carcinoma cells36 secreting similar levels (50 pmol/106 cells/24 hours) of HGF or MF-1 were obtained by lentiviral vector transduction as described. An equal number (0.75 × 106 cells/mouse) of lentiviral vector–transduced cells were injected orthotopically into the left hepatic lobe of CD-1 nu−/− mice through a single injection site. After 1 month, animals were suppressed and livers extracted for analysis. Organs were analyzed by stereoscopic microscopy immediately after extraction and photographed. After intrahepatic metastasis scoring, livers were fixed, embedded into paraffin, and processed for histology. Serial sections were stained with hematoxylin-eosin and subjected to histological analysis by an independent pathologist not informed of sample identity.
Statistical significance was determined using a two-tail homoscedastic Student t test (array 1, control group; array 2, experimental group; n = 3-6 depending on the experiment). For all data analyzed, a significance threshold of P < 0.05 was assumed. In all figures, values are expressed as mean ± standard deviation, and statistical significance is indicated by a single (P < 0.05) or double (P < 0.01) asterisk. The data generated in vitro are representative of at least two distinct experiments conducted in triplicate. In the case of in vivo data, the number of mice employed for each experiment is six, except for the orthotopic model, in which five mice per group were used.
Signaling of Metron Factor-1 in Primary Hepatocytes.
MF-1, a chimera between the N domain and two first kringles of MSP and the N domain and two first kringles of HGF (Fig. 1A), was produced in mammalian cells as described in Materials and Methods and purified to homogeneity by affinity chromatography (Fig. 1B). Because MF-1 binds to both the HGF receptor (Met) and the MSP receptor (Ron), we determined Met and Ron expression in mouse primary hepatocytes. The mouse liver cell line MLP2935 was used as a positive control. Immunoprecipitation and Western blotting using anti-Met and anti-Ron antibodies showed that mouse hepatocytes express both receptors (Fig. 1C). Human hepatocytes also express Ron as determined by immunohistochemistry.42 Therefore, the liver is a potential target of MF-1. To test this hypothesis, we stimulated mouse primary hepatocytes with increasing concentrations of MF-1, immunoprecipitated cell extracts with anti-Met or anti-Ron antibodies, and determined receptor phosphorylation by Western blotting with anti-phosphotyrosine antibodies. HGF and MSP were used as controls. Total receptor levels were determined with anti-Met and anti-Ron antibodies. As previously shown,33 MF-1 activated Met and Ron less potently compared with HGF or MSP, respectively (Fig. 1D). Furthermore, consistent with the notion that MF-1 contains a high-affinity binding site for Met and a low-affinity binding site for Ron,33 its ability to activate Met was much more pronounced compared with its effect on Ron.
Next, we determined the ability of MF-1 to activate key signaling pathways downstream of Met and Ron. Primary hepatocytes were stimulated as described and then lysed at different times. Cell extracts were analyzed by western blotting, using anti-phospho-ERK and anti-phospho-AKT antibodies. Total protein amount was normalized using anti-glyceraldehyde 3-phosphate dehydrogenase antibodies. As shown in Fig. 1E, all factors induced activation of the ERK and AKT pathways. However, MF-1–induced ERK and AKT phosphorylation was less rapid and pronounced compared with that induced by HGF or MSP.
Metron Factor-1 Promotes Hepatocyte Proliferation and Survival.
To determine whether MF-1 elicited a biological effect in primary hepatocytes, we performed a mitogenic assay, which is a standard assay to test HGF activity. Freshly isolated cells were plated in collagen-coated dishes, incubated in mitogen-free medium, and then stimulated with increasing concentrations of MF1, HGF, or MSP. DNA synthesis was determined by 3[H]-thymidine incorporation. As shown in Fig. 2A, both HGF and MF-1 were mitogenic on primary hepatocytes, although—consistent with our analysis of signal transduction—HGF was more potent than MF-1. In contrast, MSP elicited only a marginal effect on DNA synthesis. To further characterize the mitogenic effect of MF-1 on hepatocytes, we analyzed the levels of cyclin D1 and CDK-4 after factor stimulation. Cells were stimulated as described and lysed at different time points. Cell extracts were analyzed by western blotting using anti-cyclin D1 and anti–CDK-4 antibodies. Total protein amount was normalized using anti–β-actin antibodies. This analysis revealed that MF-1 stimulation induces expression of these two key G1-to-S transition regulators, although consistently to a slightly lower extent compared with HGF (Fig. 2B). In contrast—consistent with its inability to promote DNA synthesis—MSP did not significantly induce cyclin D1 or CDK-4.
Although the antiapoptotic activity of HGF on hepatocytes is well established,43 the role of the MSP/Ron pathway is more controversial.44, 45 To test whether MF-1 promotes hepatocyte survival, we pre-incubated primary cells with different concentrations of MF-1, HGF, or MSP, and then challenged cells with anti-Fas antibodies or staurosporine. Cell death was determined by the free nucleosome method. Remarkably, MF-1 protected hepatocytes against apoptosis very effectively, to an extent comparable if not superior to HGF (Fig. 2C, D). MSP promoted cell survival at low doses only, whereas higher concentrations were ineffective. Western blot analysis of apoptosis-induced caspase-3 cleavage confirmed the results obtained by the free nucleosome method (data not shown).
Metron Factor-1 Does Not Promote Tumor Cell Motility and Invasion.
MF-1 was shown to lack proinvasive activity on mouse liver precursor cells.33 Because a major concern for the clinical use of HGF is its proinvasive activity on cancer cells, we tested whether MF-1 could induce invasion of several tumor cell lines. HepG2 human hepatocarcinoma cells, A549 human lung carcinoma cells, and MDA-MB-435 human melanoma cells, which express both Met and Ron (Fig. 3A), were subjected to (1) a “scatter” assay and (2) a Matrigel invasion assay. For the former assay, cells growing in compact colonies were stimulated with different concentrations of MF-1, HGF, or MSP. Scatter activity was quantified by determining the minimal factor concentration (in other words, 1 scatter unit) required to achieve a visible scattering effect. In these experiments, MF-1 did not elicit any significant change in cell morphology, whereas HGF and MSP promoted cell scattering with different efficiency depending on the cell line (Fig. 3B). For the Matrigel invasion assay, cells were seeded onto a layer of Matrigel in a Transwell plate and then stimulated with different concentrations of MF-1, HGF, or MSP. Invasion was quantified by determining the number of cells that had migrated through the Matrigel layer. In this assay as well, MF-1 did not display any significant proinvasive activity on tumor cells (Fig. 3C).
Because the lack of proinvasive activity is central to the potential clinical application of MF-1, we designed further experiments that allowed a more thorough characterization of this biological aspect. First, we determined whether MF-1 could affect the ability of HGF and MSP to promote invasion. To this end, we stimulated A549 cells with HGF or MSP in the absence or presence of MF-1 and measured their ability to invade Matrigel as described. This analysis revealed that MF-1 potently antagonizes HGF-induced Matrigel invasion (Fig. 3D). This is consistent with the notion that both HGF and MF-1 compete for the same high-affinity binding site on the Met receptor.33 In contrast, MF-1 did not substantially affect MSP-induced cell migration, thus confirming that the biologically determinant binding site in MSP is the one contained in its β-chain.46 Next, we performed an additional Matrigel invasion assay using LOC human epithelial kidney cells and T47D human breast carcinoma cells, which express Met only and Ron only, respectively.33 As shown in Fig. 3E, HGF and MSP were effective when their specific receptor was present, whereas MF-1 did not promote invasion in either system. All together, these data suggest that MF-1, in contrast to HGF and MSP, does not promote tumor cell motility and invasion, and that this lack of proinvasive activity is independent of Met or Ron expression on target cells.
Metron Factor-1 Does Not Induce Endothelial Cell Migration and Morphogenesis.
A key biological feature of HGF that significantly contributes to tumor growth promotion is its potent proangiogenic activity.38 Neo-angiogenesis depends on three fundamental processes, all of which must be properly induced to achieve new vessel formation: endothelial cell migration, endothelial cell proliferation, and extracellular matrix remodeling.47 To determine the ability of MF-1 to promote angiogenesis, we thus performed three different assays that recapitulate these processes in vitro using HUVECs. Endothelial cells express both Met and Ron as determined by western blot analysis (Fig. 3A). In a migration assay using fibronectin-coated Transwell filters, HGF and MSP promoted endothelial cell migration through the fibronectin layer, whereas MF-1 was totally ineffective (Fig. 4A). In contrast, and consistent with the results obtained with hepatocytes, MF-1 induced mitogenesis of HUVECs as measured by bromodeoxyuridine incorporation, although to a reduced extent compared with HGF. MSP did not significantly promote DNA synthesis in this assay (Fig. 4B). Finally, in a pseudopodia elongation assay that measures the ability of endothelial cells to invade the extracellular matrix, HGF and MSP promoted pseudopodia growth while MF-1 was completely inactive (Fig. 4C). Therefore, MF-1 possesses only one of the three key proangiogenic activities displayed by HGF, at least in vitro.
Metron Factor-1 Stability In Vitro and In Vivo.
We next set to study MF-1 suitability as a drug. To be able to quantify MF-1 concentration in vitro and in vivo, we developed specific polyclonal anti–MF-1 antibodies as described in Materials and Methods. Affinity-purified anti–MF-1 immunoglobulins specifically detected 10 ng antigen by western blotting and down to 0.5 ng/mL antigen by ELISA (data not shown). To determine MF-1 stability, we dissolved recombinant MF-1 in a standard vehicle buffer used for HGF delivery23 at a concentration of 50 ng/mL. Protein solution was incubated at 37°C for several days and its concentration determined by ELISA at different times. As shown in Fig. 5A, recombinant MF-1 was extremely stable in these conditions, because more than 90% of the protein was still present after 6 days of incubation. Similar results were obtained by western blot analysis (not shown).
Next, MF-1 stability was tested in vivo. The protein preparation described was injected intravenously into adult male Balb-c mice (10 μg MF-1/mouse), and blood samples were collected from injected mice at different times (1, 2, 4, 12, and 24 hours). MF-1 concentration in plasma was measured by ELISA (Fig. 5B). Remarkably, this analysis revealed that the half-life of MF-1 in plasma is approximately 1 hour, a value significantly higher than that reported for recombinant HGF (2.4 minutes48). In any case, these data suggest that intravenous injection is perhaps not the most suitable method for MF-1 delivery, because it would be necessary to repeat protein injection very frequently to achieve a constant drug concentration in plasma.
To overcome this problem, we employed microosmotic pumps specifically designed for recombinant protein delivery. Mice were implanted intraperitoneally with osmotic pumps set to release 20 μg recombinant factor per day. Starting from the day of the operation, blood samples were taken at regular intervals (12, 24, 48, 72, and 96 hours), and the concentration of MF-1 in plasma was determined by ELISA as discussed. As shown in Fig. 5C, plasma MF-1 concentration in these animals remained substantially constant from as early as 12 hours after implantation, with an overall average of 14 ± 2.3 ng/mL. Considering that endogenous HGF is present in mouse plasma at a concentration of approximately 0.5 ng/mL,49 these levels are potentially significant from a biological viewpoint.
Metron Factor-1 Prevents CCl4-Induced Acute Liver Injury in Mice.
Using the osmotic pump-mediated delivery method described, we next compared the biological activities of MF-1 and HGF in vivo. Because MSP was not mitogenic, only displayed a weak antiapoptotic effect on hepatocytes, and yet robustly promoted tumor and endothelial cell motility and invasion, we excluded it from further investigation. Five groups of male Balb-c mice (five mice/group) were implanted intraperitoneally with osmotic pumps as described. Mice bearing empty osmotic pumps were used as a control. One day after operation, mice were injected subcutaneously with a lipophilic solution of CCl4 to induce acute liver damage. Each group of mice was examined at a different time. Right before suppression, a blood sample was taken from each animal and plasma analyzed for ALT content (Fig. 6A). On autopsy, livers were collected and processed for histological analysis (Fig. 6B, row 1). Liver sections were stained by the TUNEL method to assess the extent of CCl4-induced apoptosis (row 2) and using anti-phospho-histone-3 antibodies to determine hepatocyte proliferation (row 3).
This analysis revealed that MF-1 is at least as effective as HGF at preventing acute experimental liver damage and at promoting hepatic regeneration. Twelve hours after acute intoxication, ALT levels in control mice were found increased by more than 400 times (12,364 ± 3044 IU/L). MF-1 prevented this increase by 72% (P < 0.001) and HGF by 67% (P < 0.001). One day after CCl4 administration, livers from control mice displayed extensive necrosis around the central vein (46.0% ± 5.4% of analyzed section area). In mice treated with MF-1 or HGF, the extension of zonal necrosis around the central vein was drastically reduced (23.9% ± 2.6% and 24.3% ± 3.1%, respectively; P < 0.05 for both). Control liver sections contained a high number of TUNEL-positive apoptotic cells (19. 8 ± 2.1 cells/field). Liver sections from MF-1 and HGF-treated mice contained 49.6% (P < 0.05) and 50.7% (P < 0.05) less apoptotic cells compared with controls, respectively. At day 2, very few hepatocytes (1.00 ± 0.89 cells/field) in the control group were positive for phospho-histone-3 (an M-phase marker), indicating that liver regeneration had not started yet in these animals. In contrast, liver sections from the MF-1 and HGF groups contained 5.50 ± 1.64 (P < 0.01) and 8.17 ± 2.23 (P < 0.01) phospho-histone-3–positive cells per field, unveiling a higher rate of hepatocyte division.
Metron Factor-1 Ameliorates CCl4-Induced Liver Fibrosis.
These data suggest that MF-1 is a potent hepatotrophic factor in a model of acute liver failure. To further explore its therapeutic potential, we tested its ability to prevent or treat liver fibrosis associated with chronic hepatic intoxication. Adult male C57BL/6J mice were given a biweekly subcutaneous dose of CCl4 for 10 weeks. On week 6, mice were implanted intraperitoneally with microosmotic pumps containing MF-1 or HGF as described. Pumps were replaced every week. Mice implanted with osmotic pumps containing vehicle alone were used as negative controls. On week 10, a blood sample was taken from each mouse and plasma analyzed for ALT levels. In contrast to the acute model, chronically intoxicated control mice had a plasmatic ALT concentration only approximately 2 times higher than normal, nontreated mice (in other words, approximately 30 IU/L;49). Remarkably, however, both MF-1 and HGF did almost completely prevent this increase in enzymatic activity (Fig. 7A).
On autopsy, liver were extracted and processed for histology. Fibrotic tissue was highlighted by Sirius red staining (Fig. 7B, row 1) and alpha-smooth muscle actin immunostaining (row 2). In control livers, fibrotic areas were clearly visible around central veins, with some fibers extending to other central (C-C bridges) or portal (C-P bridges) veins. In both MF-1–treated and HGF-treated mice, fibrotic areas appeared significantly reduced in both intensity and length. Finally, we determined hepatocyte proliferation by immunochemical analysis of Ki67 expression (row 3). Proliferation index (Fig. 7C) in livers from control mice was very low (0.057 ± 0.008). In mice treated with HGF, the proliferation index increased approximately 2 times (0.114 ± 0.015), and a similar effect was achieved by MF-1 (0.123 ± 0.010). Thus, in a chronic fibrosis model as well, MF-1 displays protective and healing properties comparable to those displayed by HGF.
Metron Factor-1 Does Not Promote Tumor Growth and Metastasis.
To compare the therapeutic safety of MF-1 and HGF, we analyzed their ability to promote tumor growth and metastasis in experimental mouse models of cancer. To this end, we undertook a lentiviral vector-based ex vivo approach that allows determination of the effects of a given recombinant protein on both the tumor and its microenvironment.38 We engineered transgenic lentiviral vectors encoding for MF-1 or HGF and then used these viruses to transduce two different tumor cell lines expressing both Met and Ron (MDA-MB-435 and A549). Lentiviral vector-transduced cells secreted recombinant factors into the medium at a rate of approximately 200 pmol/106 cells in 72 hours as determined by ELISA and western blotting (data not shown). Equal number of cells expressing MF-1 or HGF were injected subcutaneously into CD-1 nu−/− mice, and animals were monitored for tumor development. Cells transduced with an empty lentiviral vector were used as controls. After approximately 8 weeks, experimental tumors were extracted for analysis, and lungs were contrasted with India ink to highlight metastases. After metastasis scoring, lungs were processed for histological analysis.
These experiments confirmed that HGF potently promotes xenograft growth in both cell systems38, 41 (Fig. 8A). At the time of autopsy, tumors expressing HGF were approximately 5.9 times (MDA-MB-435, P < 0.01) and 2.3 times (A549, P < 0.05) larger than control tumors. In contrast, tumors expressing MF-1 were only slightly larger than controls (MDA-MB-435, 1.7 times; A549, 1.1 times), and these differences were not statistically significant (P > 0.05 in both cases). HGF and MF-1 were expressed at similar levels as determined by both immunofluorescence analysis of tumor sections and western blot analysis of tumor lysates (data not shown). Consistent with our in vitro analysis, immunofluorescence staining of tumor sections with anti-CD31 antibodies revealed that HGF but not MF-1 promoted tumor angiogenesis (Fig. 8B). HGF increased microvascular density of A549 tumors by 4.6 times (P < 0.01), whereas MF-1 had substantially no effect. Similar results were observed in MDA-MB-435 tumors (Fig. 8C).
Lung metastases were scored under a stereoscopic microscope. This analysis revealed that HGF expression in tumors increased metastasis incidence in both cell systems. In control animals, only two (MDA) or one (A549) of six mice bore visible metastases. In the HGF group, virtually all animals were affected by metastatic lesions (MDA, 6/6; A549, 5/6). In contrast, MF-1 expression did not substantially affect metastasis incidence in either model (MDA, 1/6; A549, 1/6). HGF expression also increased the mean number of metastases per mouse (MDA, 63 times, P < 0.01; A549, 4 times, P < 0.05), whereas MF-1 had essentially no effect (Fig. 8D). Histological analysis of serial lung sections stained with hematoxylin-eosin confirmed these results. In control animals and in the MF-1 group, metastatic cells were very rare and were found only in those lungs that were identified positive under the stereoscopic microscope. In contrast, lungs from HGF mice were riddled with small metastatic lesions, most of them of the embolic type (Fig. 8E).
Metron Factor-1 Does Not Induce Intrahepatic Dissemination of Hepatocarcinoma Cells.
We undertook a similar, lentiviral vector-based approach to determine the effect of HGF and MF-1 on intrahepatic dissemination of cancer cells. Mouse CBO-140-C12 liver carcinoma cells36 were stably transduced as described with lentiviral vectors encoding HGF, MF-1, or no factor as a control, and factor secretion into the medium was analyzed by ELISA and western blotting (Fig. 9A). An equal number (0.75 × 106 cells/mouse) of lentiviral vector-transduced cells were injected orthotopically into the left hepatic lobe of CD-1 nu−/− mice through a single injection site. After 1 month, animals were suppressed and livers were analyzed by stereoscopic microscopy. Tumor masses were identified as white knots clearly distinguishable from the dark red background. Most livers contained a primary tumor in the injection site; livers of the HGF group contained the largest tumors (Fig. 9B). In addition to the primary site, smaller, secondary colonies derived from intrahepatic tumor cell dissemination could be identified in mice injected with HGF-expressing cells, but not in controls or in animals injected with cells expressing MF-1 (Fig. 9C). Several large extrahepatic metastases, typically adherent to the inner wall of the peritoneum, could also be found in some animals of the HGF group (Fig. 9D), suggesting that HGF—but not MF-1—dramatically enhances the metastatic potential of CBO-140-C12 hepatocarcinoma cells.
Following microscopy analysis, livers were processed and analyzed by histology. This allowed determination that primary tumors developed in the injection sites of control and MF-1 animals were (1) mainly located underneath the visceral mucosa and thus very superficial and (2) characterized by a nonnecrotic, giant cell inflammatory reaction typical of extraneous bodies (Fig. 9E). In contrast, HGF-expressing lesions were deep, invasive, rich of mitoses, and prone to give rise to embolic metastases. Dissemination of tumor cell emboli often elicited secondary ischemia of healthy hepatocytes (Fig. 9F). Only one metastatic nodule in a single animal was found in the MF-1 group, and this nodule was scarcely necrotic, poor in mitoses, and rich in psammoma bodies, thus unveiling a low proliferation rate and an active regression process (Fig. 9G). All in all, our tumorigenesis experiments suggest that HGF expression promotes tumor growth, drives tumor angiogenesis, and fosters metastatic dissemination of cancer cells. In contrast, in the same systems, similar levels of MF-1 have a substantially neutral effect.
The results presented in this study show that MF-1, an engineered factor derived from HGF and the HGF-like factor MSP, prevents or attenuates liver damage and promotes hepatocyte regeneration in mouse models of acute and chronic liver injury. They also show that MF-1, in contrast to HGF, does not elicit the invasive growth program typically induced by full Met activation that leads to tumor invasion, angiogenesis, and metastasis as measured by several in vitro invasion assays and in three different mouse models of cancer.
HGF is a pleiotropic factor that elicits a very complex genetic program that comprises several biological activities. Some of these activities are definitely beneficial for therapeutic use, including its ability to protect cells against injury and to promote tissue regeneration. Indeed, HGF has been shown to display healing properties in mice for a variety of pathological conditions,22–31 and it is an ideal candidate for therapeutic use in regenerative medicine, particularly in the hepatology field.50 Other activities of HGF, including its ability to promote cell “scattering” and migration, are essential for embryo development but may result adversely in a therapeutic setting. In fact, overwhelming experimental and clinical evidence point at a key role of HGF/Met signaling in tumor invasion, angiogenesis, and metastasis.
The Met receptor is expressed—and often overexpressed—by virtually all solid tumors and by endothelial cells in the tumor microenvironment.1, 5 Systemic (endocrine) or local (paracrine or autocrine) HGF expression strongly promotes the spreading of cancer cells and enhances the formation of spontaneous metastases in mice.38, 41, 51–54 Chronic HGF expression has also been shown to accelerate tumor induction by mutagenic agents.55, 56 Furthermore, HGF is one of the most powerful proangiogenic factors identified,57 which robustly drives tumor angiogenesis.38, 41, 55 Importantly, HGF and Met are established targets for cancer therapy. A plethora of HGF/Met inhibitors have been shown to antagonize tumor growth, angiogenesis, and metastasis in preclinical cancer models, and some of them have already entered the clinical stage.58 There is therefore little doubt if any at all that activation of the Met receptor on both cancer cells and on cells in the tumor microenvironment is causally linked to tumor progression.
This fact poses a pungent safety issue when thinking of HGF as a therapeutic agent. Will the beneficial effects of HGF on organ healing and tissue regeneration be accompanied by unwanted, iatrogenic side effects leading to tumor onset or spreading? This concern is amplified by the fact that most patients that are eligible for HGF therapy are likely to bear latent neoplastic or preneoplastic lesions, as in the case of liver cirrhosis or hepatitis C.
In this study we address this issue by showing, experimentally, that it is feasible to separate the healing, beneficial properties of HGF from its proinvasive, adverse effects by protein engineering. MF-1 is a chimeric factor derived from HGF and MSP that, unlike its parental factors, is completely devoid of any proinvasive activity, yet it does promote hepatocyte proliferation and survival. Because HGF and MSP actually synergize in inducing invasive growth,35 it is unlikely that the lack of proinvasive activity by MF-1 is attributable to concomitant activation of Met and Ron. Consistent with this, our results show that the absence of one of the two receptors does not rescue the ability of MF-1 to promote invasion. A more likely mechanism underlying this biological dieresis is, conceivably, the ability of MF-1 to only partially activate Met and Ron (in other words, less intensively) and to elicit a less sustained (more transient) downstream signaling. This idea is corroborated by the observation that partial agonists of Met, including ligand-mimetic antibodies59 and HGF-derived engineered factors,60 also promote cell survival without sustaining invasion (our unpublished results).
It therefore appears that partial activation of Met and Ron is sufficient for protecting cells against apoptosis and for promoting tissue regeneration but is not enough for inducing invasive growth. This notion is “good news” from a therapeutic viewpoint and in particular for regenerative medicine. Based on this idea, it will be possible to design “smart” therapeutic agents—of which MF-1 is a prototype—that activate Met and Ron only to the extent that is needed to achieve a healing effect, thus reducing the risk of promoting tumor progression to a minimum.
In addition to its inability to induce invasive growth and angiogenesis, MF-1 also presents other advantages over HGF. First, MF-1 is a monomeric protein that does not need proteolytic activation. In contrast, HGF is dimeric but is secreted as a monomeric precursor that must be cleaved by specific proteases to acquire biological function. This complicates the process of its pharmacological preparation as a recombinant protein, and it makes the success of a gene therapy–based approach dependent on the activity of endogenous proteases. Second, MF-1 is extremely stable and has a higher half-life than HGF in vivo. The mechanism underlying this difference is not clear. One possible explanation is that the lack of the serine protease–like domain (in other words, the beta-chain of HGF) prevents MF-1 from being targeted by proteases of the blood clotting cascade. Alternatively, the N-terminal portion of MSP could somehow “mask” the negative charges responsible for glycosaminoglycan binding on the N-domain and first kringle of HGF. This thesis is supported by the observation that HGF, when administered as a recombinant protein, remains entrapped in the injection site, whereas MF-1 becomes readily available in the plasma. In conclusion, based on the data presented here, we suggest that MF-1 is a safer and valid alternative to HGF for clinical use in the treatment or prevention of hepatic disorders or other diseases that are responsive to HGF therapy.