Monocyte chemoattractant protein-1 (MCP-1)/CCL2 secreted by hepatic myofibroblasts promotes migration and invasion of human hepatoma cells

Abstract

The aim of our study was to investigate whether myofibroblasts and the chemokine monocyte chemoattractant protein-1 (MCP-1)/CCL2 may play a role in hepatocellular carcinoma progression. We observed that hepatic myofibroblast LI90 cells express MCP-1/CCL2 mRNA and secrete this chemokine. Moreover, myofibroblast LI90 cell-conditioned medium (LI90-CM) induces human hepatoma Huh7 cell migration and invasion. These effects are strongly reduced when a MCP-1/CCL2-depleted LI90-CM was used. We showed that MCP-1/CCL2 induces Huh7 cell migration and invasion through its G-protein–coupled receptor CCR2 and, to a lesser extent, through CCR1 only at high MCP-1/CCL2 concentrations. MCP-1/CCL2's chemotactic activities rely on tyrosine phosphorylation of focal adhesion components and depend on matrix metalloproteinase (MMP)-2 and MMP-9. Furthermore, we observed that Huh7 cell migration and invasion induced by the chemokine are strongly inhibited by heparin, by β-D-xyloside treatment of cells and by anti-syndecan-1 and -4 antibodies. Finally, we developed a 3-dimensional coculture model of myofibroblast LI90 and Huh7 cells and demonstrated that MCP-1/CCL2 and its membrane partners, CCR1 and CCR2, may be involved in the formation of mixed hepatoma-myofibroblast spheroids. In conclusion, our data show that human liver myofibroblasts act on hepatoma cells in a paracrine manner to increase their invasiveness and suggest that myofibroblast-derived MCP-1/CCL2 could be involved in the pathogenesis of hepatocellular carcinoma.

Varieties of injury in the liver, including alcohol abuse and viral hepatitis, trigger an inflammatory reaction that is part of the wound-healing response. In chronic conditions, this inflammation leads to the development of fibrosis and cirrhosis.1, 2 The last one is in great part responsible for the frequent occurrence of hepatocellular carcinoma (HCC). Myofibroblasts represent a key cell for extracellular matrix (ECM) remodeling that takes place during wound healing, fibrosis and cirrhosis.3 The main source of hepatic myofibroblasts is hepatic stellate cells (HSCs). These cells are located in Disse space and store lipid-soluble vitamin A in a quiescent state. After injury, HSCs undergo an activation or transdifferentiation to myofibroblast-like cells.4 This activated phenotype is characterized by increased proliferation, mobility, contractibility and synthesis of ECM, matrix degrading enzymes and chemokines. Myofibroblasts have also been shown to play a role in cancer cell invasiveness.5–8 Chemokines govern multiple aspects of host defense, leukocyte trafficking and can also stimulate the activation of HSCs.9 They also play an important role in tumor biology because they may influence cell growth, angiogenesis, invasion and metastasis.10–13 Hepatic myofibroblasts produce a variety of chemokines including monocyte chemoattractant protein-1 (MCP-1)/CCL2, regulated on activation normal T cell expressed and secreted (RANTES)/CCL5, macrophage inflammatory protein-1 (MIP-1)-alpha/CCL3, interleukin-8 (IL8)/CXCL8 and growth-related oncogene (GRO)-alpha/CXCL1.14–16 MCP-1/CCL2 is an inducible CC-chemokine whose signals depend on its G protein–coupled receptor (GPCR), CCR2.15, 17 However, MCP-1/CCL2 can also induce biological effects through CCR1, but only at high concentration.18 Like other chemokines, MCP-1/CCL2 binds to heparan sulfate (HS) chains.19

Some lines of evidence support the fact that tumor progression depends on the concerted actions of cellular and extracellular environment.5 The activated myofibroblasts have been demonstrated to play a key role in the formation of ECM and could be involved in tumor progression.20 So, to evaluate the influence of the cellular environment in hepatic context, the biological effects of hepatic myofibroblast LI90 cells induced on human hepatoma Huh7 cell line were investigated.

The aim of this study was to determine whether 1- MCP-1/CCL2 secreted by human hepatic myofibroblasts induces migration and invasion of human hepatoma Huh7 cells, and 2-GPCRs, glycosaminoglycans (GAGs) and proteoglycans (PGs) may be involved in these biological events. Furthermore, a 3-dimensional coculture involving hepatic myofibroblasts and hepatoma cells in vitro was used to investigate the potential role of MCP-1/CCL2 and its partners in the tumoral progression.

ECM: extracellular matrix; GAG: glycosaminoglycan; GPCR: G protein-coupled receptor; HCC: hepatocellular carcinoma; HS: heparan sulfate; HSC: hepatic stellate cell; MCP-1: monocyte chemoattractant protein-1; MMP: matrix metalloproteinase; SDC: syndecan; SPR: surface plasmon resonance

Material and Methods

Cell culture and reagents

Synthetic MCP-1/CCL2 was a gift from Loïc Martin (CEA Saclay, Gif-sur-Yvette, France). Human Ito (fat storing) LI90 cell line were established by Murakami K and purchased from Japanese Collection of Research Bioresources, Osaka, Japan).21 LI90 cell line was established from a hepatic mesenchymal tumor and exhibited characteristics of activated human stellate cells.22–29 This cell line was grown in Dulbecco's modified Eagle medium (DMEM) (Invitrogen, Cergy-Pontoise, France) supplemented with penicillin (50 U/ml), streptomycin (50 μg/ml) and 10% heat-inactivated fetal bovine serum (Invitrogen) at 37°C in 5% CO₂. Cells were subcultivated every 10 days. The medium was changed twice a week. To study the effects of LI90 cell-conditioned medium (LI90-CM), cells were cultured in serum-deprived medium for 24 hours. Human hepatoma Huh7 cells were obtained from ATCC (LGC Standards, Molsheim, France) and grown as described.11, 12

Human chemokine antibody array

To determine the relative expression levels of cytokines (38 chemokines represented twice) released by the LI90 cells, a human chemokine antibody array (RayBio, Tebu Bio, Le Perray en Yvelines, France) was used. Briefly, after the blocking of the membranes, 1 ml of LI90-CM was incubated for 1 hour at room temperature according to manufacturer's instructions. After washing, the membranes were incubated with biotin-conjugated antibodies (Abs) for 1 hour at room temperature. Finally, a HRP-conjugated streptavidin was incubated with the arrayed antibody support for 2 hours at room temperature. The signal was detected by ECL method. The fold changes among samples were analyzed by use of Quantity One software (Biorad, Marne-la-coquette, France). For each spot, the net density gray level was determined by substracting the background gray level from the total raw density gray levels. The data after background substraction were normalized according to positive control densities.

MCP-1/CCL2 quantification by enzyme-linked immunosorbent assay

LI90 cells were serum-deprived for 24 hours. Culture supernatants were tested by enzyme-linked immunosorbent assay (ELISA) for MCP-1/CCL2 according to manufacturer's instructions (R&D Systems, Villejust, France).

Depletion of MCP-1/CCL2 from LI90-CM

Briefly, 500 μl of LI90-CM were incubated with monoclonal anti-human MCP-1/CCL2 Abs (R&D Systems) immobilized on protein G sepharose. After centrifugation, the supernatant was collected and MCP-1/CCL2 level was tested by ELISA assay.

Flow cytometry analysis

To identify the presence of CCR1 and CCR2 receptors on Huh7 cell surface, cells were preincubated for 1 hour at 4°C with anti-CCR1 Abs (10 μg/ml, polyclonal rabbit IgG; VWR, Fontenay-sous-Bois, France) or with anti-CCR2 (10 μg/ml, IgG2b; R&D Systems) or with the respective isotypes. After washing, cells were labeled for 30 minutes at 4°C with streptavidin-Alexa Fluor 488 complex (1:1000, Molecular Probes, Invitrogen). Cells were fixed in 1% paraformaldehyde (PFA) and analyzed with a FACScan (Becton Dickinson, Le Pont-de-Claix, France). To verify the binding of MCP-1/CCL2 on Huh7 cell membrane, a biotinylated MCP-1/CCL2 (B-MCP-1/CCL2, fluorokine, R&D Systems) was used. Cells were incubated with B-MCP-1/CCL2 for 1 hour at 4°C and labeled for 30 minutes at 4°C with avidin-Alexa Fluor 488 complex. Cells were fixed with 1% PFA and analyzed on a FACScan (Becton Dickinson).

To investigate the role of GAGs in MCP-1/CCL2 binding, B-MCP-1/CCL2 was preincubated with heparin (100 μg/ml, low molecular–weight heparin; Sigma-Aldrich, Saint Quentin Fallavier, France) for 2 hours at 37°C, or cells were treated with heparitinase I (5 U/ml), heparitinase III (50 U/ml) and chondroitinase ABC (10 U/ml) (all from Sigma-Aldrich) for 2 hours at 37°C.

Cell migration and invasion assays

Cell migration or invasion was performed with Bio-coat cell migration chambers (Becton Dickinson) as described.11 Briefly, inserts were coated with fibronectin (100 μg/ml, Santa Cruz Biotechnology, Santa Cruz, CA) for migration or Matrigel (320 μg/ml; BD Bioscience Pharmingen, Le Pont de Claix, France) for invasion assay, respectively. The human chemokine MCP-1/CCL2 or LI90-CM or medium depleted in MCP-1/CCL2 was added in the lower chamber. Twenty-four hours latter, cells in the upper chamber were removed, and migrated cells in the lower chamber were fixed with methanol. Three counting methods were then applied to evaluate the quality of the results and to detect experimental bias of counting. In a first set of experiments, cells were stained with Mayer's hemalum and counted manually by 2 different observers who performed the blind data acquisition. In a second set of experiments, micrographs of the migrated cell fields were scanned and quantified with the Scion Imager software (Scion Corp., Frederick, MD). In a third set of experiments, migrated cells were stained with 0.08% crystal violet and then lysed with 10% sodium dodecyl sulfate (SDS) solution. Each sample (100 μl) was transferred to a 96-well plate and the absorbance at 560 nm, representing the number of cells, was measured in a microplate reader (Biorad). All 3 different methods gave the same results (data not shown), and data using a manual cell counting are shown.

In parallel, Huh7 cells were preincubated for 2 hours at 37°C with the Abs: anti-CCR1 (5 μg/ml, rabbit polyclonal IgG, Santa Cruz Biotechnology), rabbit IgG (5 μg/ml; Sigma-Aldrich), anti-CCR2 (5 μg/ml, mouse monoclonal IgG2b, R&D Systems), mouse IgG2b (5 μg/ml; BD Bioscience Pharmingen), anti-HS (5 μg/ml, IgM, BD Bioscience Pharmingen), IgM (5 μg/ml, BD Bioscience Pharmingen), anti-syndecan (SDC)-1 (5 μg/ml, IgG1, clone DL-101, Santa Cruz Biotechnology), IgG1 (5 μg/ml, BD Bioscience Pharmingen), anti-SDC-4 (5 μg/ml, IgG2a, clone 5G9, Santa Cruz Biotechnology), mouse IgG2a (5 μg/ml; BD Bioscience Pharmingen), anti-MMP-9 (10 μg/ml, IgG1, Santa Cruz Biotechnology), IgG1 (10 μg/ml, BD Bioscience Pharmingen) or anti-MMP-2 (5 μg/ml, R&D Systems).

Moreover, Huh7 cells were treated with transduction signal inhibitors: PD98059, Rottlerin, LY294002 and SP600125 (all at 1 μmol/L from Sigma-Aldrich). The Huh7 cells (2.5 × 10⁵) in 0.1% BSA/DMEM were incubated for 24-hour migration or invasion assay. Alternatively, to inhibit the GAG biosynthesis, Huh7 cells were treated with 1 mmol/L β-D-xyloside (βDX) for 48 hours, and then each insert (2.5 × 10⁵ treated Huh7 cells) was incubated in 0.1% BSA/DMEM for 24-hour migration or invasion assay. Before the invasion assays, cell viability in the presence of inhibitors was evaluated by MTT assay to evaluate the cytotoxicity of each compound. For that purpose, cells were incubated with increasing concentrations of inhibitors and then incubated with 0.5 mg/ml MTT solution for 1 hour at 37°C. After MTT withdrawal, the resulting blue formazan crystals were solubilized in DMSO, and absorbance was read at 595 nm.

The percentage of inhibition was [(D1-D2)/D1] × 100. D1 was the difference between the number of untreated cells that migrated toward either LI90-CM or MCP-1/CCL2-depleted LI90-CM or human MCP-1/CCL2 and that of untreated cells that migrated toward the control medium. D2 was the difference between the number of treated cells with several inhibitors that migrated toward either LI90-CM– or MCP-1/CCL2–depleted LI90-CM or human MCP-1/CCL2 and that of cells treated with several inhibitors, which migrated toward the control medium. In some experiments, MCP-1/CCL2 was preincubated with heparin (100 μg/ml) or dextran (100 μg/ml) for 2 hours at room temperature and was further incubated for 24-hour migration or invasion assay. The percentage of inhibition was: [(D1-D3)/D1] × 100, where D3 was the difference between the number of cells that migrated toward MCP-1/CCL2 preincubated with heparin and that of cells that migrated toward heparin alone.

The in vitro migratory activity of Huh7 cells was also measured with a wound migration assay.30 An injury line was created with a single scratch at the center of a Huh7 monolayer (60-80% confluence) with a sterile 1.15-mm–diameter pipette tip. Then Huh7 cells were incubated with either DMEM or human MCP-1/CCL2 or LI90-CM or conditioned medium depleted in MCP-1/CCL2. Huh7 cells were photographed with phase contrast microscopy (Olympus CK40, X10 objective; Olympus America, Center Valley, PA) immediately after the wound and then 48 hours after treatment (15 images at each time point). Distance between cells at both sides of the wound was measured for 5 pairs of cells per image. Results were expressed as a percentage of cell migration at 48 hours versus initial wound width.

RNA interference

MCP-1/CCL2 (5′-GUCACCUGCUGUUAUAACUdTdT forward 5′-AGUUAUAACAGCAGGUGACdTdT-3′ reverse and 5′-GCAGAAGUGGGUUCAGGUdTdT-3′ forward, 5′-AUCCUGAACCCACUUCUGCdTdT reverse) and CCR1 (5′-GCUGUUUCAGGCUCUGAAdTdT-3′ forward and 5′-UUUCAGAGCCUGAAACAGdTdT-3′ reverse; 5′-CAGGGAUUAUAAAGAUUCdTdT-3′ forward and 5′-AGAAUCUUUAUAAUCCCUdTdT-3′ reverse) small interfering RNA (siRNA) sequences were designed and generated as already described.11, 31 Cells were cotransfected with 2 specific siRNA duplex in serum-free medium with jetSI-ENDO (Eurogentec, Seraing, Belgium) following the manufacturer's instructions. In each experiment, the siRNA negative control (snc-RNA, Eurogentec) was used. Cells transfected with siRNAs were used 3 days posttransfection for invasion or migration assay.

Gelatin zymography

Gelatin zymography was performed as described.11, 12 Briefly, Huh7 cells are incubated with or without MCP-1/CCL2 (3 or 50 nmol/L) in serum-free medium for 24 hours. Conditioned medium was resolved on 10% SDS-PAGE, 0.1% gelatin (Sigma-Aldrich), with equal amounts of proteins loaded. After SDS extraction, gelatinolytic activity was developed as described.11

Phosphotyrosine residue immunostaining

Huh7 cells were serum deprived for 24 hours and incubated for 15 minutes at 37°C in 10% fetal calf serum (FCS)-DMEM supplemented or not with MCP-1/CCL2 (3 and 50 nmol/L). Cells were fixed with PFA (1%), and permeabilized in 0.05% Triton X100 (Sigma-Aldrich). Cells were immunostained on phosphotyrosine residues with Tyr(P) mouse Ab (4G10, 10 μg/ml, Cell Signaling, Ozyme, Saint-Quentin en Yvelines, France) and Alexa Fluor 488 goat antimouse IgG (1:400) (Molecular Probes, Invitrogen).

Three-dimensional coculture of Huh7 and LI90 cells

Huh7 cells were stained with a cell tracker orange CMTR (Molecular Probes, Invitrogen). Cells were incubated with 10 μmol/L of CMTR for 20 minutes at 37°C. After centrifugation, cells were suspended in culture medium for 30 minutes at 37°C. Cells were ready to be used in 3-dimensional coculture conditions. To realize the 3-dimensional coculture, 20 × 10³ Huh7 cells or 12 × 10³ LI90 cells were cultured in 3-dimensional “on-Top assay” as already described.32 Briefly, wells of labtek were coated with Matrigel (100 μg/ml, BD Bioscience). Cells were suspended in Matrigel, plated on the precoated surface and incubated for 30 minutes at 37°C. Fifty percent of Matrigel mixture and culture medium was added on these cells. This mixture was changed every 2 days.

In some experiments, before the culture with LI90, Huh7 cells were preincubated for 2 hours at 37°C with Abs: anti-CCR1 (10 μg/ml, mouse IgG2b, R&D Systems), anti-CCR2 mAbs (10μg/ml, mouse IgG2b, R&D Systems), or with the isotype IgG2b (10 μg/ml, BD Bioscience). Alternatively, at the beginning of the 3-dimensional coculture assay, LI90 and Huh7 cell mixture was treated with anti-MCP-1 mAbs (10 μg/ml, mouse monoclonal IgG2b, R&D Systems) or with the isotypes IgG2b (10 μg/ml, BD Bioscience). During the scheduled culture, antibodies were added to the culture medium every day. Cells were then fixed with PFA 4% and stained with monoclonal mouse antihuman α-smooth muscle actin (1:200; Dako, Carpinteria, CA) or its isotype (IgG2a, 1:200, BD Bioscience) and Alexa Fluor 488 goat antimouse IgG (1:400). The coculture experiments were analyzed at day 1, day 2, day 3, day 8 and day 21. Phase contrast images were taken with a phase contrast Olympus CK40 microscope. Fluorescent staining was observed with a Nikon Y-FL fluorescence microscope (Nikon Instruments, Inc., Melville, NY).

Surface plasmon resonance

Optical biosensor experiments were done with a BIAcore 3000 optical biosensor (BIAcore, Orsay, France). Biotinylated heparin was coupled to the surface of a SensorChip SA (immobilized streptavidin for capture of biotinylated ligand) (ICx Technologies, Cambridge, MA). Biotinylated heparin [50 μl of 50 μg/ml in HEPES-buffered saline solution 50 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 3 mmol/L ethylenediamine tetraacetic acid, 0.005% surfactant P-20] was then injected, at 30 μl/min flow rate, to 2 channels of the streptavidin-coated censorship to a resonance unit value of 7000 (a third channel was not loaded and used as negative control). In a typical analysis, MCP-1/CCL2 at a range of concentrations (0, 10, 30, 100, 300 and 1,000 ng/ml) was flowed onto the biotinylated heparin-coated surface at a rate of 30 μl/min for a 5-minute association time, after which the channels were rinsed with the running buffer [10 mmol/L HEPES (pH 7.4), 150 nmol/L NaCl, 3.4 mmol/L ethylenediamine tetraacetic acid, 0.005% Tween 20] to analyze the dissociation phase. After each experiment, the biosensor surface was regenerated with 5 mol/L NaCl (10 μl). Flow cell, temperature, flow rate, sample volume and mixing were selected with the BIAcore control software (BIAcore). Affinities were determined by analysis of the kinetic of the association assuming a 1:1 Langmuir model with BIAevaluation software.

Statistical analysis

For the determination of statistical significance, an analysis of variance test was performed with Statview software (4.5; Abacus Concepts, Berkeley, CA). A p value < 0.05 was used as the criterion of statistical significance.

Results

Hepatic myofibroblast conditioned medium induced human hepatoma Huh7 cell migration and invasion

Human hepatic myofibroblast LI90 cells were incubated for 24 hours in DMEM, and the conditioned medium was used as indirect/chemotactic migratory stimuli (represented by addition in the bottom compartment of a modified Boyden chamber). LI90 cell-conditioned medium (LI90-CM) induced a 3-fold or a 2-fold increase in human hepatoma Huh7 cell migration (Fig. 1a) or invasion (Fig. 1b), respectively, compared to control medium (n = 3, p < 0.05). Huh7 cell invasion and migration induced by hepatocyte growth factor (HGF), used as a positive control, were similar to those induced by LI90-CM (Figs. 1a and 1b). To confirm the data obtained by use of modified Boyden chambers, the influence of LI90-CM on spontaneous Huh7 cell migration was evaluated by use of a wound migration assay. In these conditions, Huh7 cell migration was increased by 42 ± 14% when cells were incubated with LI90-CM compared to the control medium (n = 3, p < 0.05) (Fig. 1c).

Figure 1. Myofibroblast LI90-CM induced the migration (a) and the invasion (b) of human hepatoma Huh7 cells. In both experiments, HGF was used at 20 ng/ml. Results are expressed as means ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05 versus cells chemoattracted toward unconditioned medium (DMEM). (c) (Upper panel) Wound migration assays were performed with a monolayer of Huh7 incubated from left to right with DMEM or with human 3 nmol/L MCP-1/CCL2 or with LI90-CM or with LI90-CM depleted in MCP-1/CCL2. Representative phase contrast photographs were taken 48 hours after treatment. Bar = 40 μm. (Lower panel): Histograms showing Huh7 cell wound migration in each treatment group. The results were expressed as a percentage of cell migration. Basal Huh7 cell migration (DMEM) was arbitrary set to 100%. The data are expressed as means ± SEM. *, p < 0.05 for MCP-1/CCL2 or LI90-CM versus DMEM. **, p < 0.05 for LI90-CM depleted in MCP-1/CCL2 versus LI90-CM. (d) Chemokines secreted by LI90 cell line were analyzed using an antibody array (n = 3). Autoradiographs of the arrays were scanned to determine the density of the protein array position. The value from scans was adjusted based on the intensity of positive control spots on each membrane. Intensities of signal were quantified by densitometry and expressed in arbitrary units.

By the use of an antibody array, we further attempted to identify the chemokines that were strongly secreted by LI90 cell line (Fig. 1d). Among 38 tested chemokines, MCP-1/CCL2, IL8/CXCL8 and GRO chemokines were strongly secreted by LI90 cells. By RT-PCR, we also confirmed the presence of mRNAs encoding for these chemokines in LI90 cells (data not shown). MCP-1/CCL2 was among the most represented in LI90-CM, and its concentration determined by an ELISA assay was of 4.4 ± 0.35 ng/ml (0.51 ± 0.04 nmol/L) in LI90-CM. By contrast, MCP-1/CCL2 was not expressed in Huh7 cells at both transcriptional and translational levels (data not shown).

MCP-1/CCL2 is involved in biological effects induced by LI90-CM

To study the involvement of MCP-1/CCL2 in invasion and migration of Huh7 cells induced by LI90-CM, this medium was depleted specifically in MCP-1/CCL2 to a concentration of 1.8 ± 0.2 ng/ml (0.2 ± 0.02 nmol/L) by immunoprecipitation. In these experimental conditions, the chemotactic migration and invasion of Huh7 cells induced by LI90-CM were dramatically reduced by 61 ± 6.7% and 71 ± 5.6%, respectively (n = 3, p < 0.05) compared to nondepleted LI90-CM (Figs. 2a and data not shown). Similarly, as assessed by wound migration assay, hepatoma cell migration induced by LI90-CM was abolished when this medium was depleted in MCP-1/CCL2 (Fig. 1c). To strengthen the fact that MCP-1/CCL2 could be involved in the chemotactic activities of LI90-CM, MCP-1/CCL2 protein secretion by myofibroblasts was decreased by specific siRNA. When LI90 cells were transfected with MCP-1/CCL2 siRNA, MCP-1/CCL2 concentration in LI90-CM was reduced to 1.8 ± 0.7 ng/ml (0.2 ± 0.08 nmol/L) (58.2 ± 16.4% inhibition as compared to mock-transfected cells; n = 3; p < 0.05). The siRNA transfection had no effect on basal Huh7 cell migration and invasion toward control medium (data not shown). Huh7 cell migration or invasion towards LI90-CM was reduced as compared to conditioned medium from mock- or siRNA negative control (snc-RNA)-transfected LI90 cells (65 ± 8.9% or 75 ± 4.5% inhibition, respectively, as compared to LI-90 CM from mock-transfected cells, n = 3, p < 0.05) (Fig. 2a and data not shown). Similar reduction in Huh7 cell migration induced by LI90-CM was observed in a wound migration assay when Huh7 cells were incubated with conditioned medium from MCP-1/CCL2 siRNA-transfected LI90 cells (data not shown).

Figure 2. MCP-1/CCL2 secreted by LI90 cells is involved in hepatoma Huh7 cell invasion. (a) LI90-CM induced Huh7 cell invasion as compared to unconditioned medium (DMEM). Hepatoma cell invasion toward MCP-1/CCL2–depleted LI90-CM was reduced as compared to control LI90-CM. Cell invasion toward LI90-CM from cells transfected with siRNA MCP-1/CCL2 was reduced as compared to LI90-CM from mock- or snc-RNA-transfected LI90 cells. Results are expressed as mean ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05, versus DMEM, **, p < 0.05, versus LI90-CM, ***, p < 0.05, versus LI90-CM from mock-transfected cells. (b) Flow cytometry analysis of CCR1 and CCR2 on human hepatoma cells. Huh7 cells were incubated with anti-CCR1 Abs or anti-CCR2 mAbs or with polyclonal IgG or IgG2b, and then with Alexa Fluor-488 (AF-488) labeled secondary Abs. Immunolabeling was analyzed by flow cytometry. Reactivity was compared to streptavidin AF-488 alone. Data shown are representative of 3 independent experiments. (c) The preincubation of hepatoma Huh7 cells with anti-CCR1- (5 μg/ml) or anti-CCR2- (5 μg/ml) Abs drastically reduced their invasion induced by LI90-CM as compared to their incubation with the respective isotypes. Results are expressed as means ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05, versus cells migrated towards DMEM. **, p < 0.05, versus cells, preincubated with polyclonal IgG, migrated towards LI90-CM. ***, p < 0.05 versus cells, preincubated with IgG2b, migrated toward LI90-CM.

Given that LI90-CM induces Huh7 cell chemoattraction and that MCP-1/CCL2 signals through its G-protein coupled receptor CCR2 and to a lesser extent through CCR1, we further investigated whether chemotactic activities of LI90-CM could be affected when incubating Huh7 cells with anti-CCR1 or anti-CCR2 Abs. CCR1 and CCR2 were expressed in Huh7 cells at both, transcriptional and translational levels (Fig. 2b and data not shown). By contrast, CCR1 and CCR2 were not expressed in LI90 cells (data not shown). Thereafter, the preincubation of hepatoma cells with anti-CCR1 Abs reduced cell invasion (85 ± 4% inhibition, n = 4, p < 0.05) and cell migration (73 ± 15% inhibition, n = 4, p < 0.05) toward LI90-CM (Fig. 2c and data not shown). In addition, the preincubation of Huh7 cells with anti-CCR2 Abs radically decreased these 2 biological events (n = 4, p < 0.05) (Fig. 2c and data not shown).

MCP-1/CCL2–induced human hepatoma cell invasion and migration

Biotinylated MCP-1/CCL2 (B-MCP-1/CCL2) bound to Huh7 cells (Fig. 3a). The fact that at a concentration of 100 nmol/L the binding of B-MCP-1/CCL2 to the cells was barely detectable (data not shown) may be related to MCP-1/CCL2 aggregation, as described for the CC-chemokine RANTES, which aggregates at high concentration.12

Figure 3. MCP-1/CCL2 induced the invasion of human hepatoma Huh7 cells. (a) Huh7 cells were incubated with biotinylated MCP-1/CCL2 (B-MCP-1) at the indicated concentrations. Binding was analyzed by flow cytometry with streptavidin-Alexa Fluor 488. Data shown are representative of 3 independent experiments. (b) MCP-1/CCL2 induced Huh7 cell invasion in a dose-dependent manner. MCP-1/CCL2 at 0 to 100 nmol/L, as indicated, was used for this experiment. Results are expressed as mean ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05 versus control medium (in the absence of MCP-1/CCL2). (c) The invasion of Huh7 cells induced by MCP-1/CCL2 (3 or 50 nmol/L) was reduced when the cells were preincubated with anti-CCR2 or anti-CCR1 Abs as compared to cells incubated with the control isotypes or with anti-CXCR4 Abs. Invasion induced in the cells preincubated with the control isotypes was arbitrarily set to 100% and MCP-1/CCL2–induced cell invasion in the presence of specific antibodies is shown as a percentage of control. Similar reduced Huh7 cell invasion was observed when cells were transfected with siRNA CCR1 as compared to control snc-RNA-transfected cells. Invasion induced by MCP-1/CCL2 in snc-RNA-transfected control cells was set to 100% and MCP-1/CCL2-induced invasion in siRNA CCR1-transfected cells is shown as a percentage of control. *, p < 0.05, versus control cells preincubated with the isotype controls. **, p < 0.05, versus control cells transfected with snc-RNA. (d) MCP-1/CCL2-mediated Huh7 cell invasion was significantly decreased by the cell preincubation with PD98059 (1 μmol/L), Rottlerin (1 μmol/L), LY294002 (1 μmol/L) or SP600125 (1 μmol/L). Results are expressed as means ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05, versus control cells migrated toward DMEM. **, p < 0.05, versus untreated cells migrated toward MCP-1 (in the absence of inhibitors). The preincubation of Huh7 cells with anti-MMP-9- or anti-MMP-2 Abs (10 μg/ml) diminished the cell invasion towards MCP-1/CCL2 as compared to cells incubated with the respective isotype controls. ***, p < 0.05 versus cells preincubated with the isotype controls. (e) MCP-1/CCL2 induced phosphorylation of focal adhesion components. Huh7 cells incubated or not with MCP-1/CCL2 (3 nmol/L) were examinated by direct immunostaining for phosphotyrosine residues with anti-Tyr(P) 4G10 Abs. Bar = 5 μm. (f) A representative gelatin zymography gel showing increased pro-MMP-9 (98 kDa) and pro-MMP-2 (70 kDa) activities after 24-hours incubation with both MCP-1/CCL2 3 nmol/L and 50 nmol/L solutions versus MCP-1/CCL2-nontreated control cells.

MCP-1/CCL2 (0.01, 0.1, 1, 3, 10, 50, 100 nmol/L) induced the invasion and migration of Huh7 cells in a dose-dependent manner (Fig. 3b and data not shown). Significant effect was already observed for 0.1 nmol/L MCP-1/CCL2. The maximal effect was obtained at 3 nmol/L. Huh7 cell migration induced by 3 nmol/L MCP-1/CCL2 was confirmed by wound migration assay (56 ± 15% increase, n = 3, p < 0.05) (Fig. 1c). These chemotactic effects mainly depended on CCR2 because anti-CCR2 Abs radically decreased these events (100 ± 5% and 90 ± 5.5% respectively, n = 4, p < 0.05) (Fig. 3c and data not shown). The preincubation of Huh7 cells with anti-CCR1 Abs inhibited cell invasion or migration by 30 ± 3% or 40 ± 6.4%, respectively (n = 3, p < 0.05) at 3 nmol/L MCP-1/CCL2 and strongly reduced it at 50 nmol/L MCP-1/CCL2 (100 ± 3% or 90 ± 5%, n = 3, p < 0.05) (Fig. 3c and data not shown). Conversely, the preincubation of cells with the specific isotypes or with anti-CXCR4 Abs had no effect (Fig. 3c and data not shown). To confirm the involvement of CCR1 in Huh7 cell invasion and migration induced by MCP-1/CCL2, we investigated whether the down-regulation of CCR1 by siRNA could affect Huh7 cell chemotaxis. In Huh7 cells transfected with CCR1 specific siRNA, CCR1 protein expression was significantly reduced by 47 ± 4.6% as assessed by flow cytometry (n = 3, p < 0.05) (data not shown). In these experimental conditions, MCP-1/CCL2-induced cell invasion or migration was reduced by, respectively, 60 ± 6 % or 43 ± 11% for 3 nmol/L MCP-1/CCL2 (n = 3, p < 0.05) and 94 ± 5% or 96 ± 3.3% for 50 nmol/L MCP-1/CCL2, (n = 3, p < 0.05) (Fig. 3c and data not shown) compared with snc-RNA-transfected control cells.

To explore the signaling pathways involved in MCP-1/CCL2 mediated-Huh7 cell invasion, cells were incubated with specific pharmacological inhibitors. By the use of the MTT cell viability assay, we first selected the inhibitor concentrations that were not cytotoxic in our conditions (data not shown). Pretreatment of Huh7 cells with PD98059 (a MEK inhibitor) or Rottlerin (a PKC inhibitor) or LY294002 (a PI3K inhibitor) or with SP600125 (a JNK/SAPK inhibitor) reduced the MCP-1/CCL2-mediated cell invasion by 80 ± 3.8%, 85 ± 6.5%, 87 ± 10.5% (n = 3, p < 0.05) or 41 ± 6% (n = 3, p < 0.05) respectively (Fig. 3d and data not shown). Similarly, the preincubation of Huh7 cells with PD98059 or LY294002 or Rottlerin or SP600125 strongly reduced their migration toward MCP-1/CCL2 (95 ± 4.7% or 77 ± 9% or 62 ± 8.6% or 88 ± 2.1% inhibition, n = 3, p < 0.05) (data not shown). Specific pharmacological inhibitors did not affect basal Huh7 cell migration or invasion (data not shown). These results suggest that MAPK, PKC and PI3K pathways are involved in the MCP-1/CCL2-mediated invasion or migration of Huh7 cells.

Focal adhesion components have been demonstrated to play regulatory roles in cell migration. So, to investigate whether these components are activated, Huh7 cells were treated with 3 nmol/L MCP-1/CCL2 for 15 minutes and were examined by indirect immunostaining for phosphotyrosine residues with anti-Tyr(P) (4G10) Abs as shown in Fig. 3e. Membranous staining was more intense in MCP-1/CCL2-treated cells compared to untreated control cells, suggesting that the chemokine induced the phosphorylation of focal adhesion components.

To identify the MMPs involved in Huh7 cell invasion induced by the chemokine, cells were treated with 3 and 50 nmol/L MCP-1/CCL2 and MMP-2 and MMP-9 protein levels were analyzed by gelatin zymography. Interestingly, 3 nmol/L and 50 nmol/L MCP-1/ CCL2 increased the pro-MMP-2 and pro-MMP-9 forms (Fig. 3f). Huh7 cell preincubation with anti-MMP-9 Abs strongly reduced MCP-1/CCL2-mediated hepatoma cell invasion (83 ± 6% inhibition, n = 3, p < 0.05) (Fig. 3d). In addition, theincubation of Huh7 cells with anti-MMP-2 Abs decreased MCP-1/CCL2–mediated cell invasion (74 ± 10% inhibition, n = 3, p < 0.05) (Fig. 3d).

Glycosaminoglycans and proteoglycans are involved in MCP-1/CCL2–induced migration and invasion of Huh7

Chemokines bind to GAGs, which are demonstrated to participate in tumor progression.11, 12 We then confirmed that MCP-1/CCL2 interacts with heparin by surface plasmon resonance (SPR). Therefore heparin was biotinylated and immobilized at 300 RU on the sensorship. MCP-1/CCL2 was then injected over the Biacore heparin surface in a range of concentrations (usually 0-100 nmol/L) to produce a set of sensorgrams from which association and dissociation phases could be analyzed. When the chemokine was injected into biosensor flow cells and flowed over heparin surfaces for 5 min, a typical increase of the SPR response (in RU) versus time was obtained, corresponding to the association phase that was then followed (5 minutes) by a decrease of SPR response, the dissociation phase, when the chemokine was replaced by running buffer (Fig. 4a). Fitting of binding curves gave an association rate constant (k_on) of 9.97 × 10³ M⁻¹ s⁻¹ and a dissociation rate constant (k_off) of 1.39 × 10⁻² s⁻¹. In these experimental conditions, the binding of MCP-1/CCL2 to heparin is characterized by an affinity K_d (k_off/k_on) of 1.39 × 10⁻⁶ mol/L. When MCP-1/CCL2 was flowed over control surfaces (containing streptavidin only), no significant signal was observed (not shown).

Figure 4. Glycosaminoglycans and heparan sulfate proteoglycans are involved in MCP-1/CCL2-mediated Huh7 cell invasion and migration. (a) Surface plasmon resonance analysis of MCP-1/CCL2 heparin interaction. MCP-1/CCL2 was injected over flow cells of a BIAcore sensorship containing streptavidin plus 300 RU of biotinylated heparin. Each set of sensorgrams was obtained by injecting MCP-1/CCL2 at (from bottom to top) 11, 22, 33, 44, 66, 88 or 110 nmol/L. The response in RU was recorded as a function of time. (b) Huh7 cells were treated with heparitinases I and III (Hep'ase) and chondroitinase ABC (C'ase) and then incubated with biotinylated MCP-1/CCL2 (B-MCP-1/CCL2) (80 nmol/L). Binding was analyzed by flow cytometry with streptavidin-Alexa fluor 488. Data shown are representative of 3 independent experiments. (c) B-MCP-1/CCL2 (80 nmol/L) was preincubated or not with heparin (100 μg/ml) and the suspension was added to the cells. (d) MCP-1/CCL2-mediated Huh7 cell invasion was decreased when the chemokine was preincubated with heparin (100 μg/ml) but not with dextran (100 μg/ml). Huh7 cell invasion induced by 3 nmol/L MCP-1/CCL2 was reduced when the cells were treated with βDX (1 mmol/L) as compared with the MCP-1-mediated invasion of untreated Huh7 cells. Results are expressed as means ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05 versus untreated cells migrated toward DMEM. **, p < 0.05, versus untreated cells migrated towards MCP-1. (e) The preincubation of the cells with anti-SDC-1 (5 μg/ml) or anti-SDC-4 (5 μg/ml) mAbs reduced Huh7 cell invasion toward MCP-1/CCL2 as compared to cells preincubated with the control isotypes IgG1 or IgG2a. Results are expressed as means ± SEM of Huh7 cells counted by field for 3 independent experiments. *, p < 0.05, versus cells, preincubated with the respective isotypes, migrated toward DMEM. **, p < 0.05, versus cells, preincubated with IgG1, migrated towards MCP-1. ***, p < 0.05, versus cells, preincubated with IgG2a, migrated towards MCP-1.

Furthermore, the pretreatment of hepatoma cells with heparitinases I, III and chondroitinase ABC strongly reduced the binding of MCP-1/CCL2 to the cells, as well as the preincubation of MCP-1/CCL2 with heparin (100 μg/ml) (Figs. 4b and 4c). Dextran had no effect (data not shown).

To investigate the role of GAGs in the biological effects mediated by MCP-1/CCL2, Huh7 cells were treated with βDX (1 mmol/L) and assayed for MCP-1/CCL2-induced chemotaxis. In these experimental conditions, Huh7 cell invasion and migration were abolished (100 ± 3.1% and 100 ± 4% inhibition, respectively) (n = 3, p < 0.05) (Fig. 4d and data not shown). Furthermore, the incubation of MCP-1/CCL2 with 100 μg/ml heparin diminished hepatoma cell invasion by 84 ± 5% (n = 4, p < 0.05) or migration by 79 ± 9% (n = 4, p < 0.05) whereas dextran had no effect (Fig. 4d and data not shown). Moreover, the preincubation of Huh7 cells with anti-HS Abs strongly decreased MCP-1/CCL2-mediated cell invasion by 97 ± 2.7% (n = 3, p < 0.05) and migration by 93 ± 7.4% (n = 3, p < 0.05) (data not shown). Additionally, the preincubation of LI90-CM with heparin (100 μg/ml) or with anti-HS Abs significantly reduced Huh7 cell migration and invasion toward LI90-CM (data not shown). These data suggest that HS chains are involved in MCP-1/CCL2-induced cell invasion and migration.

Heparan sulfate proteoglycans belonging to the syndecan family have been demonstrated to play a major role in SDF-1/CXCL12- or RANTES/CCL5-mediated biological effects in hepatoma cells.11, 12 Therefore we raised the question whether these proteoglycans could interfere with MCP-1/CCL2-induced hepatoma cell invasion and migration. The preincubation of Huh7 cells with antisyndecan-1 (anti-SDC-1) Abs reduced MCP-1/CCL2-induced cell invasion and migration by 66 ± 9% and 82 ± 11% respectively (n = 4, p < 0.05) (Fig. 4e and data not shown). In the same way, the pretreatment of Huh7 cells with anti-syndecan-4 (anti-SDC-4) Abs decreased cell invasion and migration induced by the chemokine by 70 ± 11% (n = 3, p < 0.05) and by 92 ± 4.7% (n = 3, p< 0.05), respectively (Fig. 4e and data not shown). As a negative control, the preincubation of hepatoma cells with anti-CD44 Abs was performed and this treatment did not affect MCP-1/CCL2-mediated cell chemotaxis (data not shown). All these experiments suggest that SDC-1 and SDC-4 are involved in hepatoma cell invasion and migration induced by 3 nmol/L of MCP-1/CCL2.

Coculture of myofibroblasts and hepatoma cells

To explore the role of MCP-1/CCL2 in the formation of mixed hepatoma-myofibroblast spheroids, we used a 3-dimensional “on-Top assay” as described.32 In a first set of experiments, Huh7 and LI90 cells were plated separately on a Matrigel layer and recovered with another layer of this matrix. To locate the various types of cells in a 3-dimensional coculture, we stained Huh7 cells with the cell tracker CMTR and LI90 cells with anti-α-SMA Abs, this cytoskeleton component being a specific marker of myofibroblast cells. When LI90 cells were cultured in the absence of hepatoma cells, they spread well and became flat. After spreading, LI90 cells extended long cellular processes that protrude into the gel and made contact with other stellate cells via these structures, as described (Fig. 5a).33 When cultured in the absence of LI90 cells, Huh7 cells were first scattered on the surface, and then, they clustered and formed hepatoma spheroids which are spherical multicellular aggregates as already described when hepatocytes were cultured in other matrix (Fig. 5b).34, 35

Figure 5. Three-dimensional coculture of human hepatoma Huh7 and myofibroblast LI90 cells. LI90 (a) and Huh7 (b) cells were cultured separately in 3-dimensional “on Top assay”. LI90 cells were stained with antimonoclonal smooth α-actin Abs (anti-α-SMA) revealed by Alexa Fluor 488 goat antimouse IgG Abs (a, left) and Huh7 cells were stained with a cell tracker orange (CMTR) (b, left). In a and b (right panels), cell nuclei were stained with DAPI. Some LI90 cellular processes are indicated by white arrows. Bar = 30 μm. (c–e) Coculture experiments at day 1, day 2 or day 3: Huh7 cells were immunostained with a cell tracker orange (CMTR) and LI90 cells with anti-α-SMA Abs. The coculture was either untreated (c) or treated with anti-MCP-1/CCL2 mAbs (d) or with both anti-CCR1 and anti-CCR2 mAbs (e). Bar = 60 μm. (f). Phase contrast images of coculture experiments at day 3. Black arrows indicate spheroids in untreated cells or in cells treated with anti-MCP-1/CCL2 mAbs or with both anti-CCR1 and anti-CCR2 Abs. Bar = 60 μm.

When cocultured, 24 hours after plating, Huh7 cells began to form small clusters, whereas LI90 cells were scattered on the matrix (Fig. 5c, day 1). The 2 subsequent days, the clusters of Huh7 cells joined together to form hepatoma spheroids, and Huh7 aggregates were progressively surrounded by LI90 cells. These cells were then located at the periphery of the spheroids (Fig. 5c, day 2 and day 3 and Fig. 5f). This organization was maintained for more than 2 weeks (data not shown). At day 3, the diameter of the mixed hepatoma-myofibroblast spheroids was 107.5 ± 31.3 μm (Fig. 6). To further investigate the role of MCP-1/CCL2 in the formation of mixed spheroids, we treated the cocultures with anti-MCP-1 mAbs (Fig. 5d) or with anti-CCR1 and anti-CCR2 mAbs (Fig. 5e) or with the respective isotypes. The spatial organization of cocultures incubated with the isotypes was similar to that observed in untreated cocultures (data not shown) but was affected in cocultures treated with specific antibodies. Indeed, the small clusters of Huh7 cells, preincubated with either anti MCP-1/CCL2 or both anti-CCR1 and anti-CCR2 antibodies, remained separate and formed smaller hepatoma spheroids than untreated cells at the beginning of the first week (day 1, day 2 and day 3). At this time, most LI90 cells were scattered in the culture and randomly organized around clusters, and therefore the formation of mixed hepatoma-myofibroblast spheroids was scarcely observed (Figs. 5d, 5e and 5f). The diameter of the spheroids, mainly consisting of hepatoma cells without surrounding myofibroblasts was 43.5 ± 7.5 μm or 45.5 ± 10 μm for cocultures treated with anti-MCP-1/CCL2 or with anti-CCR1- and anti-CCR2 mAbs respectively (p < 0.01 as compared to untreated cocultures) (Fig. 6). At day 8, the differences in spheroid morphology and diameter were reduced but still present between cocultures treated with both anti-CCR1 and anti-CCR2 mAbs or with anti-MCP-1/CCL2 mAbs as compared to untreated coculture controls (data not shown). At the end of the assay (day 21), the 3-dimensional organization of all cocultures was similar (data not shown), suggesting that MCP-1/CCL2 and its membrane partners may participate in the formation of mixed hepatoma-myofibroblast spheroids mainly in the first week of experiments.

Figure 6. Quantification of the cell spheroids. (Upper panel): Cell spheroids were photographed at day 3 of coculture experiments of untreated cells or cells treated with anti-MCP-1/CCL2 mAbs or with both anti-CCR1 and anti-CCR2 Abs and the diameters of spheroids (n = 20) were measured. Representative fluorescent images were taken at day 3 of coculture experiments. (Lower panel): Histograms represent means ± SEM of the diameters of the spheroids measured in each coculture condition. *, p < 0.05, versus untreated cells.

Discussion

Accumulating evidence suggests that HCC cell invasion and metastasis proceed by an interaction between tumor cells and resident cells. It has been demonstrated that human liver myofibroblasts act on HCC cells to increase their invasiveness, and these effects are blocked by addition of an antibody to HGF, and all can be reproduced by adding recombinant HGF to tumor cells, suggesting that myofibroblast-derived HGF can be involved in the pathogenesis of HCC.36, 37 Furthermore, it has been shown that HSC recruitment and activation into myofibroblast is under the control of tumoral hepatocytes.20

In our study, we used the LI90 cell line that was established from a human hepatic mesenchymal tumor.21 This cell line exhibited morphological and functional characteristics compatible with those of activated HSCs. Numerous reports have shown that LI90 cell line is a useful tool for in vitro studies of the functions of human HSCs.22–29 We also demonstrated that LI90 cells secreted MCP-1/CCL2 at a concentration previously measured in human primary HSCs.38, 39 Recent evidence has suggested that MCP-1/CCL2 may be an important regulator of liver injury. Secretion of MCP-1/CCL2 by HSCs is up-regulated in several forms of fibrotic liver injury.40, 41 Increased expression of MCP-1/CCL2 has been reported in patients with different types of liver disease and in experimental models of acute and chronic liver injury.9 Mice deficient for CCR2, the major receptor for MCP-1/CCL2, show exacerbation of liver damage during acetaminophen intoxication.42 Conversely, in a model of liver metastasis in CCR2-deficient mice, formation of tumor was attenuated with a concomitant reduction in the accumulation of HSCs, macrophages and neovascularization.43 Furthermore, it was recently demonstrated that MCP-1/CCL2 secreted by adipose tissue may induce steatosis not only by recruiting macrophages but also by acting directly on hepatocytes.44

Here, we demonstrate that the conditioned medium from human myofibroblast LI90 cells strongly stimulated the migration and invasion of human hepatoma Huh7 cells and that the CC-chemokine MCP-1/CCL2 is at least partly involved in these chemotactic activities. Evidence from clinical epidemiology, histopathology, preclinical animal modeling, and in vitro tissue culture studies all point to the possible involvement of MCP-1/CCL2 and CCR2 in cancer pathobiology. MCP-1/CCL2 has been identified as a major chemokine inducing the recruitment of macrophages in human tumors, including those of the bladder, cervix, ovary, lung and breast.45 In bladder and breast cancers, MCP-1/CCL2 was even correlated to the grade of the tumor with highly invasive tumors secreting the highest amounts of MCP-1/CCL2. Besides its protumoral effect via leukocyte recruitment or activation, MCP-1/CCL2 and CCR2 expressions have been documented in various tumor cell types, including breast, pancreas or prostate cancers and some cancer cell lines respond chemotactically to MCP-1/CCL2 in vitro.46, 47 Studies have demonstrated that MCP-1/CCL2 is increased in the sera of various cancer patients as compared to healthy subjects.48, 49 In the liver, we demonstrated that MCP-1/CCL2 serum level mean in patients with chronic hepatic alcoholic or hepatitis C virus-related cirrhosis with or without hepatocellular carcinoma was about 500 ± 200 pg/ml (0.06 ± 0.02 nmol/L).50, 51 Furthermore, primary HSCs have been demonstrated to secrete MCP-1/CCL2 in a range of concentrations from about 0.5 to 1 nmol/L.38, 39 Therefore the concentration observed in LI90 cell conditioned medium may be relevant to the in vivo liver situation.

In the study herein, we demonstrate that MCP-1/CCL2 exerts protumoral effects on hepatoma cells through CCR2 in a paracrine manner, because the chemokine is not secreted by hepatoma cells. CCR1 is involved in MCP-1/CCL2's biological activities in hepatoma cells only at high concentration (up to 50 nmol/L), suggesting that its importance in vivo may be restricted to major physiopathological situations of inflammation or tumoral processes. Therefore we can suggest that CCR2 plays a pivotal role in MCP-1/CCL2-mediated biological effects on hepatoma cells.

Hepatoma cell invasion through the ECM involves cellular detachment, mobility through ECM and basement membrane degradation, which are processes depending on a variety of enzymes that include MMPs. Interestingly, MMP-9 and MMP-2 play major roles in the MCP-1/CCL2-induced cell invasion. This data is in accordance with those demonstrating that both MMPs are involved in HCC progression. Indeed, MMP-2 and MMP-9 have been found to be of prognostic significance in HCC.52 The content of MMP-2, MMP-9 in HCC being higher than that in surrounding liver parenchyma could be used as an important index to judge the invasion and metastasis of HCC.53 We suggest that hepatoma cell migration induced by MCP-1/CCL2 may be related to the activation of focal contacts as demonstrated by the tyrosine phosphorylation of their components induced by the chemokine.

Our data also suggest that MCP-1/CCL2 may play a major role in the spatial coorganization of both hepatoma cells and myofibroblastic cells because the formation of mixed hepatoma-myofibroblast spheroids in a reconstituted extracellular matrix mimicking those observed in vivo is influenced by MCP-1/CCL2 or by its specific membrane receptors. We demonstrate here that blocking the migratory-stimulating molecule MCP-1/CCL2 reduced the formation of mixed hepatoma-myofibroblast spheroids. As it was previously demonstrated that the culture of epithelial cells in a reconstituted extracellular matrix could induce the formation of polarized nonmigratory spheroids,54 our data could appear in some way contradictory. On the one hand, it has been shown that 3-dimensional culture of the immortalized nontransformed human prostatic epithelial BPH-1 cell line in Matrigel recapitulated many structural features and aspects of differentiated phenotype of the prostatic glandular epithelium in vivo.54 Nevertheless, the development of a functional and morphologically correct prostate gland in vitro is also dependent on extracellular matrix and factors from stromal cells and serum.55 On the other hand, recent experimental data revealed a correlation between the ability of cancer cells, such as ovarian cancer cells, to form compact spheroids and to invade/migrate within 3-dimensional matrix, suggesting then a positive relationship between invasiveness and capacity for the formation of compact spheroid.56 Therefore the formation of spheroids does not exclude cell capacity for migration or invasion. With regard to our experimental data, we demonstrate that MCP-1/CCL2 induced Huh7 cell migration (migration wound assay and Boyden chamber experiments) and invasion (Boyden chamber experiments) in a dose-dependent manner through CCR2 and at a lesser extent through CCR1. We also demonstrated that the formation of mixed hepatoma cell-myofibroblast spheroids is affected when the cells were pre-incubated with anti-MCP-1/CCL2 or anti-CCR1 and anti-CCR2 Abs. This effect is mainly observed in the first days of culture (day 1, day 2 and day 3). In the third week of culture (day 21), mixed hepatoma cell-myofibroblast spheroids are not disrupted by cell preincubation with anti-MCP-1/CCL2 or anti-CCR1 and anti-CCR2 Abs and their size and morphology are similar in both untreated and treated cells. These data suggests that blocking MCP-1/CCL2 or its membrane partners could delay the formation of the mixed hepatoma-myofibroblast spheroids.

Furthermore, we suggest that the initial step (day 1 to 3) of hepatoma-myofibroblast spheroid formation mainly occurs as a result of cell migration and cell-to-cell or cell-to-extracellular matrix adhesion. We also suppose that the influence of cell growth during the initial step of mixed spheroid formation is marginal because Huh7 or LI90 cells replicated with a doubling time of about 50 hours or 60 hours, respectively. Furthermore, it was previously demonstrated that MCP-1/CCL2 promotes adhesion events and that cell-to-cell and cell-to-matrix adhesion could also mediate chemokine expression.57, 58 Altogether, our data suggest that MCP-1/CCL2 could participate to the initial step of the formation of hepatoma cell-myofibroblast spheroids. Nevertheless, our data does not exclude that other chemokines that bind to CCR1, such as RANTES/CCL5 or MIP-1/CCL3, could also be involved in the interactions between hepatoma and myofibroblastic cells.

In accordance with previous studies,19, 59 we demonstrate that MCP-1/CCL2 binds to heparin by surface plasmon resonance. Whether MCP-1/CCL2 binding to GAGs interferes with the chemokine activities is partially unknown. It has been demonstrated that soluble heparinlike GAGs can antagonize the leukocyte-activating and migration-promoting properties of MCP-1/CCL2 in vitro.60 It was also demonstrated that a GAG-binding deficient MCP-1/CCL2 mutant retains chemotactic activity in vitro but was unable to recruit cells when administrated intraperitoneally.61 Here, hepatoma cell invasion and migration were drastically reduced by the pre-incubation of the chemokine with heparin or when cellular GAG synthesis is reduced by βDX cell treatment, suggesting that HS chains play major roles in MCP-1/CCL2-mediated biological activities, as described for others chemokines.11, 12 Interestingly, the inhibition of MCP-1/CCL2's chemotactic activities by anti-SDC-1 and anti-SDC-4 Abs suggested that those heparan sulfate proteoglycans could be membrane coreceptors of the chemokine. Given the role of GAGs in modulating MCP-1/CCL2's activities, our data suggest that targeting the specific GAG-chemokine binding could lead to the development of new therapeutic strategy of HCC.

In summary, our data show that human liver myofibroblasts act on HCC cells to increase their invasiveness and suggest that myofibroblast-derived MCP-1/CCL2 could be involved in the pathogenesis of HCC.