Stromal Cell–Derived Factor-1/Chemokine (C-X-C Motif) Ligand 12 Stimulates Human Hepatoma Cell Growth, Migration, and Invasion

Chemokines are chemotactic cytokines that govern multiple aspects of host defense and inflammation such as hematopoiesis and leukocyte trafficking (1). Chemokines also play an important role in tumor biology because they may influence tumor growth, invasion, and metastasis (2-6). Stromal cell–derived factor 1 (SDF-1)/chemokine (C-X-C motif) ligand 12 (CXCL12), a CXC chemokine that exists mainly in two alternative splicing variants, α and β, is a homeostatic chemokine that signals through chemokine (C-X-C motif) receptor 4 (CXCR4), a G protein–coupled receptor, which in turn plays an important role in hematopoiesis, development, and organization of the immune system (2, 7). However, like other chemokines, this chemokine also binds to glycosaminoglycans (8, 9). Recent studies have indicated that SDF-1 is expressed in some cancer cells (i.e., malignant ovarian and breast cancer cell lines) and is involved in tumor cell migration and metastasis (10, 11).

The syndecans are a family of proteoglycans, which, together with the lipid-linked glypicans, are the major source of heparan sulfate chains at cell surfaces (12, 13). By way of their heparan sulfate, syndecans bind a wide variety of soluble and insoluble ligands, such as extracellular matrix components, cell adhesion molecules, growth factors, cytokines, proteinases, or pathogens such as HIV-1 (13-15).

We recently showed that SDF-1 forms complexes on HeLa cells and human primary lymphocytes or macrophages, which comprise CXCR4 and syndecan-4 (SDC-4; ref. 16). We also showed the occurrence of a heteromeric complex between SDC-4 and CXCR4 at the plasma membrane of these cells. Nevertheless, our data showed that SDF-1 binds directly to SDC-4, which may be a signaling molecule for the chemokine (17).

A number of in vitro and in vivo studies highlight the importance of some chemokines in acute or chronic liver diseases (18-20) and indicate that chemokines may modulate certain biological actions in hepatocytes, including proliferation (20). CXCR4 expression has already been shown in hepatoma cells (21-23).

The aim of the present study was to determine whether SDF-1 induces the growth, migration, and invasion of human hepatoma cells and elucidate the molecular mechanisms of these effects, including the involvement of SDF-1 G protein–coupled receptor, CXCR4, and glycosaminoglycans. We extended our in vitro data by using immunohistochemistry to provide the status of CXCR4 in the liver samples of patients with hepatocellular carcinoma (HCC).

In basal culture conditions, SDF-1α (82.5 ± 21.9 pg/mL) was detected in the culture supernatant of human hepatoma Huh7 cells whereas mRNA encoding for SDF-1 was observed in these cells (Fig. 1A). mRNAs encoding for CXCR4, syndecan-1 (SDC-1), syndecan-2 (SDC-2), and SDC-4 were also observed (Fig. 1A) whereas CXCR4 and the heparan sulfate proteoglycans SDC-1 and SDC-4 were detected at their plasma membrane (Fig. 1B).

Biotinylated SDF-1α bound in a dose-dependent manner to Huh7 cells (Fig. 2A). AMD3100, a CXCR4 antagonist (24), strongly decreased this binding by 66 ± 14% (P < 0.01; n = 3; Fig. 2B) and heparin by 73 ± 19% (P < 0.001; n = 3; Fig. 2C). This suggests that both CXCR4 and glycosaminoglycans are involved in the binding.

To characterize SDF-1 ligands or receptors expressed by Huh7 cells, SDF-1α–containing complexes were collected on anti-SDF-1α–coated beads. Immunoblotting the complexes with anti-CXCR4 monoclonal antibody (mAb) 12G5 revealed 48-kDa proteins (Fig. 2D, lane 1), characterized by an apparent molecular mass close to that reported for CXCR4 (45-48 kDa; refs. 25, 26). Neither immunoreactivity with anti-CCR5 2D7 nor with the isotype was detected (Fig. 2D, lane 2, and data not shown). If this complex was treated with glycosaminidases, 32- and 45-kDa proteins immunoreactive with anti–SDC-4 mAb 5G9 were observed but not with anti–SDC-1 mAb DL-101 nor with the isotype (Fig. 2D, lanes 3-5, and data not shown). The 32 kDa apparent molecular mass is close to that predicted for SDC-4 protein core (13, 27), whereas the 45 kDa molecular mass may represent proteoglycan oligomerization. Such eluted proteins were not detected if the cells were incubated in SDF-1–free buffer (data not shown). Therefore, CXCR4 and SDC-4 coimmunoprecipitate with SDF-1. Whether CXCR4 and SDC-4 associate on Huh7 cells was then investigated; the cells were stimulated or not by SDF-1 and lysed. Lysates were incubated with protein G coated with anti–SDC-4 mAb 5G9. In both cases, proteins immunoreactive with anti-CXCR4 12G5 mAb coimmunoprecipitated with SDC-4 (data not shown). Therefore, a heteromeric complex between SDC-4 and CXCR4 occurs even in the absence of SDF-1. Moreover, we observed that biotinylated SDF-1 directly binds to electroblotted SDC-4 (data not shown), which is consistent with our previous studies (16, 17).

Stimulation of the cells with SDF-1α resulted in a significant, marked, and rapid increase in free radical formation after a 1-min stimulation. Heparin, AMD3100, or heparitinase treatment of the cells abolished this SDF-1α–induced reactive oxygen species production (P < 0.05; Fig. 3A). None of these cell treatments significantly affected basal reactive oxygen species levels.

Stimulation of Huh7 cells with SDF-1α (3 and 125 nmol/L) also significantly increased phosphorylated forms of both extracellular signal–regulated kinase 2 (Erk2; p42) and c-jun NH₂-terminal kinase/stress-activated protein kinase (JNK/SAPK; p54/p46) in a time-dependent manner, reaching a maximum after 15 min of stimulation (Fig. 3B and data not shown). As a positive control, phorbol 12-myristate 13-acetate also significantly activated Erk1/2 and JNK/SAPK kinases (data not shown).

SDF-1α (3 and 125 nmol/L) significantly stimulated Huh7 cell proliferation, as assessed by both crystal violet and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays (P < 0.05; n = 3; Fig. 4A and B). This was partly, but significantly, inhibited by AMD3100 (47 ± 4% inhibition; P < 0.05; n = 3; Fig. 4C). Preincubating SDF-1α with heparin (100 μg/mL) also strongly decreased this effect (83 ± 6% inhibition; P < 0.05; n = 3). Furthermore, in Huh7 cells depleted of proteoglycans by a 3-day incubation in 1 mmol/L 4-methylumbelliferyl-β-d-xyloside (βDX; Fig. 4D), the proliferation induced by SDF-1α was strongly attenuated (67 ± 7% inhibition; P < 0.05; n = 3; Fig. 4C). Therefore, both CXCR4 and glycosaminoglycans are involved in this SDF-1–dependent Huh7 cell growth.

SP600125, a JNK/SAPK pharmacologic inhibitor, partly prevented the proliferative effect induced by 3 nmol/L SDF-1α (54 ± 9% inhibition; P < 0.05; n = 3) whereas the mitogen-activated protein kinase (MAPK)/Erk kinase inhibitor PD98059 had no effect (Fig. 4C).

None of these compounds significantly affected basal cell proliferation.

To discriminate G₀ from G₁ phase, Huh7 cells stimulated with SDF-1α were analyzed for expression of Ki67 proliferation–associated nuclear antigen. This antigen is undetectable in G₀ resting cells; it is expressed in cells entering G₁ and its expression increases with changes in staining pattern during the cell cycle (28). SDF-1α (125 nmol/L) significantly increased the proliferative Ki67 labeling index (Ki67/4′,6-diamidino-2-phenylindole (DAPI)–positive stainings) in these cells, which was 75 ± 4% for untreated cells versus 95 ± 1% for SDF-1α–treated cells (P < 0.05; Table 1); moreover, the Ki67 flow cytometry index [Ki67/isotype immunoglobulin G1 (IgG1) mean fluorescence intensities] increased 3-fold in SDF-1α–treated cells (Table 1).

SDF-1α also affects Huh7 cell cycle status: the addition of SDF-1α (125 nmol/L) to the cells significantly prevented spontaneous DNA degradation in Huh7 cells because their proportion in sub-G₁ decreased from 51.5 ± 9.4% to 25 ± 6.8% (P < 0.001; n = 3). SDF-1α also increased the percentage of Huh7 cells in G₀-G₁ from 28.7 ± 8.4% to 39.4 ± 7.2%, and in S + G₂-M phases from 19.8 ± 4.8% to 35.6 ± 6.9% (P < 0.001; n = 3; Table 2). Therefore, SDF-1 triggers quiescent Huh7 cells from G₀ into cycle, whereas it stimulates the transition of cells already engaged in G₁ to S + G₂-M.

We consequently investigated the chemokine-mediated effect on apoptosis using previously established tumor necrosis factor α (TNFα)–mediated apoptosis in Huh7 cells (29). The percentage of Annexin V–positive cells decreased from 6.8 ± 0.5% to 1.1 ± 0.3% (n = 3; P < 0.05) after a 48-h incubation with the chemokine, suggesting that SDF-1 promotes the survival of these cells (Table 3).

Due to the important role of the CXCR4/SDF-1 axis in cancer metastasis, we investigated whether SDF-1 induces Huh7 cell migration. In these experiments, hepatocyte growth factor (20 ng/mL) was used as a positive control (30). As shown in Fig. 5A, SDF-1α induced Huh7 cell migration in a dose-dependent manner (P < 0.05). This induction was significantly reduced by incubating Huh7 cells with AMD3100 (65 ± 9% inhibition; P < 0.05; n = 3; Fig. 5B). In some experiments, SDF-1α preincubated with heparin was added to the lower chamber. Blocking the heparin-binding site of the chemokine or treating the cells with βDX strongly decreased SDF-1α–induced migration (78 ± 8% and 92 ± 12% inhibition, respectively; P < 0.05; n = 3; Fig. 5B).

SP600125 slightly but significantly reduced this SDF-1α–induced migratory effect by 27 ± 3% (P < 0.05; n = 3; Fig. 5B). PD98059 had no effect.

None of these compounds significantly affected basal migration (data not shown).

Therefore, SDF-1α–induced Huh7 cell migration depends on CXCR4, glycosaminoglycans, and, at least partly, JNK/SAPK pathway activation.

To further characterize the molecular events involved in SDF-1α–induced Huh7 cell migration, SDF-1α–treated cells were examined by indirect immunostaining of phosphotyrosine residues with an anti-Tyr(P) mAb (4G10) and of tyrosine phosphorylation of focal adhesion kinase (FAK) at Tyr³⁹⁷ with a polyclonal anti–FAK-(P)-Tyr³⁹⁷ antibody (Fig. 6B).

The respective stainings occur especially on focal adhesion plaques and were much more intense in SDF-1α–stimulated Huh7 cells compared with controls (Fig. 6A and B). Whereas untreated cells displayed slight actin stress fiber networks and smooth, regular cell borders, SDF-1α causes a change in the reorganization of filamentous actin and induces mostly membrane ruffling at the cell periphery (Fig. 6C).

Finally, FAK was immunoblotted from lysates of SDF-1α–stimulated or unstimulated control cells with anti-FAK antibodies and with anti-FAK-(P)-Tyr⁵⁷⁷ phosphospecific antibodies. The level of tyrosine phosphorylation at FAK-Tyr⁵⁷⁷ from SDF-1α–treated cells was higher as compared with that of SDF-1α–untreated control cells (Fig. 6D; P < 0.05; n = 3).

SDF-1α (3 nmol/L) induced Huh7 invasion into Matrigel (P < 0.05; Fig. 7A). This was inhibited by incubating the cells with AMD3100 (51 ± 17% inhibition; P < 0.05; n = 3; Fig. 7B). AMD3100 itself did not affect basal invasion (data not shown). Strikingly, heparin itself increased it by 50% (data not shown). However, preincubation of SDF-1α with heparin strongly reduced SDF-1α–induced cell invasion (70 ± 15% inhibition; P < 0.05; n = 3). Whereas treatment of cells with βDX did not affect basal invasion, it reduced SDF-1α–dependent cell invasion (82 ± 12% inhibition; P < 0.05; n = 3; Fig. 7B).

Pretreatment of Huh7 cells with PD98059 or SP600125 abolished the increased invasion induced by SDF-1α (Fig. 7B). Therefore, CXCR4, glycosaminoglycans, and MAPK pathways are involved in this SDF-1 biological effect.

Interestingly, SDF-1α (3 nmol/L) increases matrix metalloproteinase-9 (MMP-9) mRNA levels in Huh7 cells (Fig. 7C), as well as MMP-9 precursor, as assessed by zymography (Fig. 7D). Huh7 cell preincubation with anti-MMP-9 mAb resulted in a 42 ± 14% inhibition (P < 0.05; n = 3) of SDF-1α–dependent cell invasion (Fig. 7B).

Because SDF-1 binds to SDC-4 on Huh7 cells, we investigated whether the down-regulation of SDC-4 by RNA interference affects SDF-1–induced biological effects. Huh7 cells were transfected with SDC-4 double-strand RNA (SDC-4 dsRNA) or with a small interfering negative control RNA (snc-RNA) for up to 3 days. Specific SDC-4 RNA interference significantly reduced SDC-4 mRNA and protein levels in Huh7 whereas SDC-1 expression remained unchanged (Fig. 8A and data not shown). SDF-1α–induced Huh7 cell growth was not affected by SDC-4 RNA interference (data not shown). In contrast, in cells transfected with SDC-4 dsRNA, SDF-1α–induced cell migration and invasion were strongly reduced (Fig. 8B and C) compared with snc-RNA–treated control cells (P < 0.05; n = 3). SDC-4 RNA interference did not affect basal migration or invasion (data not shown).

CXCR4 mRNA was clearly expressed in HepG2 and faintly so in Hep3B cells, whereas SDF-1 mRNA was only detected in HepG2 cells (Fig. 9A). CXCR4 was present at the plasma membrane of both cell lines (Fig. 9B). SDF-1α bound to these cells (Fig. 9C). However, whereas SDF-1α induced a slight but significant activation of JNK/SAPK in Hep3B cells, this effect was not detected in HepG2 cells (data not shown). Finally, SDF-1α did not induce the growth or the migration of HepG2 or Hep3B cells (data not shown). In contrast, whereas SDF-1α increased the invasion of Hep3B cells (P < 0.05; n = 3), it did not exert such an effect on HepG2 cells (Fig. 9D).

Liver biopsies from two patients (with hepatitis C or B virus–related cirrhosis and HCC) exhibited a strong membranous CXCR4 staining of the carcinoma cell, irregularly distributed throughout the samples (Fig. 10). In two other patients (one with hepatitis C virus–related cirrhosis and one with alcoholic-related cirrhosis, and HCC), a moderate nuclear staining of scattered carcinoma cells with anti-CXCR4 mAb was detected. No membranous staining was concurrently observed. One patient (hepatitis B virus–related cirrhosis) did not show any staining. Taking all samples into account, no cytoplasmic staining was observed. Isotypic controls were negative.

In the liver, chemokines are leukocyte chemoattractants and can also stimulate key biological processes in hepatic stellate cells, such as activation, proliferation, and migration (31, 32). Recently, the CCL3/CCR1 axis has been shown to be involved in HCC progression (33). CXCR4 expression has been shown in hepatoma cells (21-23). Mitra et al. (21) showed that the human HCC cell line HepG2 expresses CXCR4 but is unresponsive to SDF-1 stimulation because of a defect of the receptor. More recently, it has been shown that a strong expression of CXCR4 in HCC specimens is significantly associated with progressed HCC (23).

Here, we show that Huh7 hepatoma cells express CXCR4 and secrete SDF-1α under basal conditions. Exogenous SDF-1α binds to these cells through CXCR4 and glycosaminoglycans, just as the CXCR4 antagonist AMD3100, or heparin, strongly decreased this binding. Moreover, the SDF-1/CXCR4 axis activates signaling pathways such as Erk2 and JNK/SAPK kinases as previously observed in other tumor cells (34, 35). SDF-1 also induces reactive oxygen species production through CXCR4 in Huh7 cells. Such production may be an intracellular message involved in the chemokine signaling pathways (36). However, glycosaminoglycans interfere with SDF-1α–induced reactive oxygen species production because preincubation of the chemokine with heparin or treatment of the cells with heparitinases reduced ROS level.

We next investigated whether SDF-1 modulates key biological functions in Huh7 cells and explored the underlying molecular mechanisms. As recently described (23), we observed that SDF-1α stimulates Huh7 cell proliferation as assessed by both crystal violet and MTT assays. We showed that this stimulation depends on CXCR4, glycosaminoglycans, and the JNK/SAPK kinase transduction pathway. That SDF-1 triggers G₀ quiescent Huh7 cells in G₁ and S + G₂-M phases of cell cycle, as assessed by the expression of the proliferation-associated nuclear antigen Ki67 and propidium iodide-DNA binding assays, may also explain, at least partly, the proliferative effect of SDF-1α in Huh7 cells. The fact that SDF-1 prevents spontaneous DNA degradation in these cells and decreases the percentage of Annexin V–positive cells in a TNFα-mediated apoptosis experimental model also suggests that SDF-1 behaves as a survival factor for these cells.

The effect of SDF-1 in Huh7 was considerably stronger on migration and invasion than on proliferation and was also mediated through CXCR4. Moreover, whereas SDF-1α–induced Huh7 cell migration, to some extent, depends on JNK/SAPK signaling pathway, SDF-1α invasive effect is very dependent on both Erk2 and JNK/SAPK kinase activations. This involvement of MAPK signaling pathways has been reported in SDF-1 stimulatory effect on human pancreatic (37) or ovarian (38) cancer cell migration and invasion. Cancer cell mobility depends on their interactions with their microenvironment (39). HCC occurs mainly on fibrotic liver and is associated with altered extracellular matrix composition (40). The cleavage of basement membrane boundaries by tumoral cells and invasion is mediated by proteases, such as MMPs (39). MMP-9 overexpression, associated with capsular infiltration and growth of HCC, has been reported (41, 42).

Here, we show that SDF-1α increases MMP-9 expression in Huh7 cells and that MMP-9 inhibition decreases SDF-1α–induced Huh7 cell invasion. Cell attachment to fibrillar extracellular matrix and migration are mediated by structures called “focal adhesions,” which connect the extracellular matrix with the plasma membrane and the underlying actin cytoskeleton (39). Moreover, SDF-1α enhances the tyrosine phosphorylation of FAK, and probably that of the focal adhesion complex components, and reorganizes Huh7 cytoskeleton.

These data strongly suggest the involvement of SDF-1 in the highly regulated processes of HCC invasion and metastasis, which are major determinants in HCC progression and prognosis (43).

Our study also shows that CXCR4 and SDC-4 coimmunoprecipitate with SDF-1 at the plasma membrane of Huh7 cells. Whether two separate complexes or a trimolecular complex is formed on these cells can be hypothesized. However, the fact that CXCR4 coimmunoprecipitates with SDC-4 in the absence or presence of SDF-1 strongly argues for the formation of one complex that comprises SDF-1, CXCR4, and SDC-4. Because biotinylated SDF-1 binds to electroblotted SDC-4, this also suggests a direct binding of SDF-1 to SDC-4.

All biological effects, cell growth, migration, and invasion, induced by SDF-1 on Huh7 cells were significantly affected by preincubating the chemokine with heparin or by treating the cells with β-DX, suggesting the involvement of glycosaminoglycans. Moreover, SDF-1–induced migration and invasion were reduced when SDC-4 cell expression was specifically down-regulated by RNA interference. These results, which agree with those previously observed in HeLa cells (17, 18), suggest that SDC-4 expressed on Huh7 cells may be an auxiliary receptor for SDF-1. Interestingly, previous studies have shown that protein kinase Cδ–dependent phosphorylation of SDC-4 regulates cell migration and that SDC-4 is required for focal adhesion formation (44).

Strikingly, SDC-4 RNA interference did not affect SDF-1–induced Huh7 cell growth, whereas βDX treatment of the cells reduced it. It can therefore be hypothesized that heparan sulfate chains expressed by other proteoglycans may be involved in this effect. In fact, our immunoprecipitation data do not exclude other heparan sulfate proteoglycans, such as SDC-2, from binding to SDF-1. Indeed, SDC-2 is expressed in Huh7 cells and has been shown to play an important role in the tumorigenic activity of numerous tumor cells (45, 46).

Otherwise, we previously showed that SDF-1 induces the shedding of SDC-1 and SDC-4 ectodomains from HeLa cells and that MMP-9 is involved in these events (47). Whether SDF-1 also induces syndecan ectodomain shedding from Huh7 cells is currently under investigation. In this context, the MMP-9 activation induced by SDF-1 in Huh7 cells could be part of an autoregulatory/down-regulation cycle mediated by SDF-1/MMP-9 syndecan ectodomain shedding (48).

In the present study, we observed that the HepG2 and Hep3B hepatoma cell lines, characterized by different p53 status compared with Huh7 cells (49), expressed CXCR4. Indeed, it was recently shown that CXCR4 expression does not depend on p53 status (23). However, in agreement with earlier published data (21, 23), HepG2 cells are unresponsive to SDF-1, and SDF-1 increases Hep3B cell invasion into Matrigel, but not proliferation or migration. Therefore, SDF-1 biological effects on hepatoma cells strongly depend on the cell type.

Finally, to evaluate to what extent our in vitro results may be extrapolated to the in vivo situation, we explored CXCR4 expression in liver samples of patients with HCC. Immunohistochemical staining of HCC liver biopsies displayed either membrane expression or nuclear localization of CXCR4 in hepatocarcinoma cells. In our preliminary study, we did not find any reduced CXCR4 expression compared with the surrounding nontumor tissue, in contrast to others (50). Schimanski et al. (23) recently showed that a strong cytoplasmic expression of CXCR4 in HCC specimens is significantly associated with progressed HCC. In contrast, an earlier published study showed that the degree of CXCR4 expression did not correlate with the clinicopathologic features of HCC (51). Considering the fact that SDF-1 signals through CXCR4 and that SDC-4 may be an auxiliary receptor for the chemokine on hepatoma cells, the relevance of the SDF-1/CXCR4 axis and also of SDC-4 in the liver biopsies of HCC patients could therefore be of interest.

In summary, our data indicate that the SDF-1/CXCR4 ligand receptor axis may play an important role in the pathogenesis of HCC and that a CXCR4 receptor antagonist, such as AMD3100, could inhibit cell growth, migration, and invasion of hepatoma cells. Moreover, glycosaminoglycans modulate the effects of SDF-1 in hepatoma cells. A better understanding of this chemokine effect on HCC development and progression may enable novel chemokine glycosaminoglycan mimetic–based immunomodulating drugs.

Huh7, HepG2, and Hep3B human hepatoma cell lines were grown as described (52). For proteoglycan biosynthesis inhibition, cells were incubated with βDX (Sigma-Aldrich, Saint-Quentin Fallavier, France) for 72 h as described (53).

Cells (10⁵) were incubated with biotinylated SDF-1α (0, 12.5, 40, or 125 nmol/L; gift of F. Baleux, Laboratoire de Chimie, Institut Pasteur, Paris, France) as described (17). In parallel, cells were preincubated for 1 h at 37°C with AMD3100 (1.2-12 μmol/L; Sigma-Aldrich) or biotinylated SDF-1α was preincubated for 2 h at 20°C with heparin (100 μg/mL; low molecular weight heparin, H3149, Sigma-Aldrich). After washing, cells were labeled for 30 min at +4°C with streptavidin-Alexa Fluor 488 complex (1:100; Molecular Probes, Invitrogen, Cergy-Pontoise, France). Flow cytometry data of SDF-1 binding to the cells (B) were expressed as the mean fluorescence intensity of the cells, incubated in the presence of biotinylated SDF-1 (test) minus that of the cells, incubated in the absence of the chemokine (negative control). The percentage of inhibition of SDF-1 binding to the cells, induced by AMD3100 or heparin, was calculated by dividing the difference between B of the cells in the absence of the inhibitor and B of the cells in the presence of the inhibitor (BI) by B of the cells in the absence of the inhibitor, and then multiplying by 100. Results were expressed as the mean percentage of inhibition of at least three independent experiments ± SD. Statistical analysis of the coupled differences between B and BI was done with Student's t test.

CXCR4 or SDC-4 immunostaining for flow cytometric analysis was done using anti-CXCR4 mAb (clone 12G5, BD Bioscience PharMingen, Pont de Claix, France; 10 μg/mL) or anti–SDC-4 mAb (mouse IgG2a; clone 5G9, Santa Cruz Biotechnology, Inc., Santa Cruz, CA; 10 μg/mL) or murine IgG2a (BD Bioscience PharMingen) as described (17).

To assess βDX treatment efficiency, 10⁵ Huh7 cells, pretreated or not with βDX (1 mmol/L) for 72 h, were incubated for 30 min on ice with 10 μg/mL anti–heparan sulfate mAb (clone 10E4, Seikagaku Corporation, Tokyo, Japan) or murine IgM (BD Bioscience PharMingen). Cells were then labeled with FITC-labeled goat anti-mouse immunoglobulin (BD Bioscience PharMingen) and fixed in 1% paraformaldehyde and analyzed on a FACScan (Becton Dickinson, Le Pont-de-Claix, France).

Adherent Huh7 cells were incubated for 1 h at 4°C, with anti-CXCR4 mAb 12G5 (15 μg/mL), anti–SDC-4 mAb 5G9 (10 μg/mL) or murine IgG2a, then labeled as described (17). Alternatively, cells were fixed with paraformaldehyde (1%) and incubated for 1 h at 20°C with anti–SDC-1 mAb B-B4 (10 μg/mL; Serotec, Oxford, United Kingdom) or murine IgG1 (BD Bioscience PharMingen). Cells were observed under a fluorescence microscope (Olympus, Rungis, France).

CXCR4, SDF-1, SDC-1, SDC-2, SDC-4, MMP-9 mRNAs and glyceraldehyde 3-phosphodehydrogenase mRNA were amplified by reverse transcription-PCR (RT-PCR; ref. 18). Specific primers were designed as follows: SDF-1/CXCL12, 5′-CCATGAACGCCAAGGTCGTGGTC-3′ (forward) and 5′-GGGCATGGATGAATATAAGCTGC-3′ (reverse); CXCR4, 5′-AGTATATACACTTCAGATAAC-3′ (forward) and 5′-CCACCTTTTCAGCCAACAG-3′ (reverse); SDC-1, 5′-TCTGACAACTTCTCCGGCTC-3′ (forward) and 5′-CCACTTCTGGCAGGACTACA-3′ (reverse); SDC-2, 5′-GGGAGCTGATGAGGATGTAG-3′ (forward) and 5′-CACTGGATGGTTTGCGTTCT-3′ (reverse); SDC-4, 5′-CGAGAGACTGAGGTCATCGAC-3′ (forward) and 5′-CGCGTAGAACTCATTGGTGG-3′ (reverse); and MMP-9, 5′-AAGATGCTGCTGTTCAGCGGG-3′ (forward) and 5′-GTCCTCAGGGCACTGCAGGAT-3′ (reverse). In some experiments, optimum semiquantitative RT-PCR conditions were established to remain in the linear phase of amplification curve.

Huh7 cells were serum deprived for 48 h. Culture supernatants were tested by ELISA for SDF-1α (R&D Systems, Villejust, France).

Huh7 cells (10⁶) were incubated in the presence or absence of SDF-1α (2 μg) and lysed. Lysates were subjected to immunoprecipitation on protein G-Sepharose beads (Pharmacia, Paris, France), precoated with anti–SDF-1α mAb (goat IgG) or its isotype (both from R&D Systems; each at 2.5 μg), or with anti–SDC-4 mAb 5G9 or its isotype (16). The complexes were electroblotted (17) and revealed with anti-CXCR4 12G5 mAb or, as a negative control, with anti-CCR5 2D7 (BD Bioscience PharMingen). Alternatively, the complexes were treated with heparitinase I (1 units/mL), heparitinase III (15 units/mL), and chondroitinase ABC (5 units/mL) mixture (Sigma-Aldrich) and were revealed with anti–SDC-1 DL-101 mAb (Santa Cruz Biotechnology), anti–SDC-4 5G9 mAb, or their isotypes (all at 1:1,000-1:5,000). After washing, strips were incubated with horseradish peroxidase–conjugated antimouse IgG (1:5,000-1:20,000) and revealed by enhanced chemiluminescence reagent (Amersham Biosciences, Buckinghamshire, United Kingdom). In some experiments, strips were revealed with biotinylated SDF-1α, as described (17).

Cells (10⁵) were stimulated with SDF-1α (3 nmol/L). In parallel, cells were incubated with both AMD3100 (1.2-12 μmol/L) and SDF-1α (3 nmol/L). Alternatively, SDF-1α was preincubated for 2 h at 20°C with heparin (100 μg/mL) and the suspension was added to the cells. In some experiments, cells were pretreated for 2 h with heparitinase I (EC 4.2.2.8; 100 mIU/mL) and heparitinase III (EC 4.2.2.7.; 200 mIU/mL; Sigma-Aldrich; refs. 16, 17).

Cells were then incubated for 30 min at 37°C in the dark with a 10 μmol/L PBS-dichlorofluorescein diacetate solution (Molecular Probes). Unstimulated control cells were incubated in parallel.

Huh7 cells (2.5 × 10⁵) were cultured for 48 h in 0.1% FCS-DMEM and incubated at 37°C for 15 min with SDF-1α (3 and 125 nmol/L). MAPKs were revealed (17) using antibodies specific for phospho-Erk1/2 (p44/p42) [Thr²⁰²/Tyr²⁰⁴] or phospho-JNK/SAPK (p54/p46) [Thr¹⁸³/Tyr¹⁸⁵] or for their total counterparts (all from Cell Signaling, Danvers, MD). Phosphorylated FAK was revealed using polyclonal anti-FAK-(P)-Tyr⁵⁷⁷ antibodies (Cell Signaling). Parallel immunoblotting with anti total FAK polyclonal antibodies (Cell Signaling) was done to confirm equal loading of samples. Quantification of Erk1/2, JNK/SAPK, and FAK phosphorylation was done by using the Scion program after autoradiography scanning.

Cells (5 × 10³) were treated for 48 h at 37°C with SDF-1α (0, 3, and 125 nmol/L), then fixed and incubated for 2 min with 0.08% crystal violet (Sigma). Cell proliferation was assessed by colorimetric assay. Absorbance was read at 595 nm with a microplate reader (model 680, Bio-Rad, Ivry-sur-Seine, France).

Cell viability was measured using the reduction of MTT (Sigma-Aldrich). Cells (5 × 10³) were treated for 48 h with SDF-1α (0, 3, and 125 nmol/L). In parallel, cells were pretreated for 1 h at 37°C with AMD3100 (1.2-12 μmol/L) or for 30 min at 37°C with PD98059 (1 μmol/L) and SP600125 (1 μmol/L; Calbiochem, Fontenay-sous-Bois, France) before the addition of SDF-1α (3 nmol/L). Alternatively, cells were treated with 1 mmol/L βDX for 24 h and 5 × 10³ βDX-pretreated cells were further incubated with 1 mmol/L βDX and 125 nmol/L SDF-1α for 48 h. Cells were then incubated with 0.5 mg/mL MTT for 1 h at 37°C. After MTT withdrawal, the resulting blue formazan cristae were solubilized in DMSO (Merck, Fontenay-sous-Bois, France). Absorbance was read at 595 nm.

Cells were serum deprived for 24 h, incubated for 4 h at 37°C in 10% FCS-DMEM supplemented or not with SDF-1α (125 nmol/L), washed with PBS-0.1% bovine serum albumin, fixed with paraformaldehyde (1%), and permeabilized in 0.05% Triton X-100 (Sigma-Aldrich). Expression of Ki67 proliferation–associated nuclear antigen was assessed with an anti-Ki67 mAb (IgG1; 1:500; Santa Cruz Biotechnology). Cells were then incubated with Alexa Fluor 488 goat anti-mouse IgG (1:400). Ki67 immunostaining was either analyzed directly by flow cytometry or counterstained with 0.1 μg/mL 4′,6-diamidino-2-phenylindole (Sigma-Aldrich) to evaluate nuclei number in each cell field counted.

Cell cycle studies were done by means of DNA-propidium iodide binding. Cells were serum deprived for 24 h and stimulated with SDF-1α (125 nmol/L) for 48 h. After fixation with ethanol, cells were incubated with RNase A (200 μg/mL; Sigma-Aldrich) for 30 min and resuspended in propidium iodide-PBS (10 μg/mL) for flow cytometric analysis.

The percentage of cells undergoing apoptosis was determined using Annexin V detection assay (R&D Systems). Briefly, Huh7 cells were incubated for 48 h under apoptosis-inducing conditions (50 ng/mL TNFα) with or without SDF-1α (125 nmol/L). Cells (10⁵) were then incubated at room temperature for 15 min with fluorescein-conjugated human Annexin V and analyzed.

Cell migration or invasion was done using Bio-coat cell migration chambers (Becton Dickinson). Inserts containing 8-μm pore size filters were coated with fibronectin (100 μg/mL; Santa Cruz Biotechnology) for migration or Matrigel (320 μg/mL BD PharMingen) for invasion assay. After filter blockage with 1 mg/mL bovine serum albumin for 1 h, 2.5 × 10⁵ cells in 0.1% bovine serum albumin-DMEM were added. The chemokine SDF-1α was added to 500 μL of DMEM supplemented with 10% FCS in the lower chamber. After 24 h, cells that had migrated through the filter pores were fixed with methanol, stained with hematoxylin, and counted. In parallel, cells were preincubated for 2 h at 37°C with inhibitors AMD3100 (12 μmol/L), anti-MMP-9 mAb (10 μg/mL, IgG1; Santa Cruz Biotechnology), murine IgG1 (10 μg/mL; BD Bioscience PharMingen), PD98059 (1 μmol/L), or SP600125 (1 μmol/L). Alternatively, cells were treated with 1 mmol/L βDX for 48 h, and for each insert, 2.5 × 10⁵ cells in 0.1% bovine serum albumin-DMEM were further incubated with 1 mmol/L βDX for 24-h migration or invasion assay. The percentage of inhibition was [(D1 − D2) / D1] × 100; D1 was the difference between the number of untreated cells that migrated toward SDF-1α and that of untreated cells that migrated toward the culture medium without SDF-1; D2 was the difference between the number of treated cells that migrated toward SDF-1α and that of treated cells that migrated toward culture medium (D2). Alternatively, SDF-1α was preincubated for 2 h at 20°C with heparin (100 μg/mL). Heparin alone or SDF-1α preincubated with heparin was added to the lower chamber of culture. The percentage of inhibition was [(D1 − D3) / D1] × 100, where D3 was the difference between the number of cells that migrated toward SDF-1α preincubated with heparin and the number of cells that migrated toward heparin alone.

Gelatin zymography was done as described (47). Briefly, Huh7 cells were incubated for 24 h in serum-free medium supplemented or not with phorbol 12-myristate 13-acetate (0.5 μmol/L), used as a positive control, or SDF-1α (3 and 125 nmol/L). Conditioned media were resolved on 10% SDS-PAGE, 0.1% gelatin (Sigma-Aldrich), with equal amounts of proteins loaded. After SDS extraction, gelatinolytic activity was developed in buffer [50 mmol/L Tris-HCl (pH 7.4), 5 mmol/L CaCl₂, 200 mmol/L NaCl, and 0.05% Brij 35; Sigma-Aldrich] at 37°C for 24 h. The gel was stained with Coomassie blue R-250, destained, and scanned.

Huh7 cells were serum deprived for 24 h, incubated for 20 min at 37°C in 10% FCS-DMEM supplemented or not with SDF-1α (125 nmol/L), fixed with paraformaldehyde (1%), and permeabilized in 0.05% Triton X-100 (Sigma-Aldrich). Cells were immunostained on phosphotyrosine residues using Tyr(P) mAb (4G10; 10 μg/mL, Cell Signaling) and Alexa Fluor 488 goat anti-mouse IgG (1:400). Cells were also examined by indirect immunostaining for tyrosine phosphorylation of FAK at Tyr³⁹⁷ using polyclonal anti–FAK-(P)-Tyr³⁹⁷ antibody (Cell Signaling) and Cy3-conjugated goat anti-rabbit polyclonal antibodies (1:400). For visualization of filamentous actin, cells were then exposed to Alexa Fluor 568-phalloidin (1:1,000; Molecular Probes) for 30 min at 37°C.

SDC-4 gene–specific sense and antisense 21-nucleotide single-stranded RNAs with symmetrical two-nucleotide 3′(2′-deoxy)thymidine overhangs were designed as described (17). For RNA interference experiments, dsRNAs were generated by mixing equimolar amounts (50 μmol/L) of sense and antisense single-stranded RNAs in annealing buffer as described (17). Huh7 cells were transfected with 150 nmol/L dsRNA in serum-free medium using Jetsi transfectant reagent (Eurogentec, Seraing, Belgium) following the manufacturer's instructions. Mock cells were cultured in parallel and transfected with the transfection mixture lacking dsRNA. In each experiment, a snc-RNA (Eurogentec) was used. Cells transfected with SDC-4 dsRNA or small interfering negative control dsRNA were used 3 days posttransfection for further analysis.

HCC samples were obtained from five patients with cirrhosis who underwent ultrasound guided biopsy of hepatic nodules. Two patients had hepatitis C virus–related cirrhosis, two had hepatitis B virus–related cirrhosis, and one had alcoholic-related cirrhosis. Four-micrometer-thick paraffin-embedded, alcohol/formalin/acetic acid–fixed liver biopsy sections were deparaffinized and hydrated. Heat antigen retrieval was done by incubating the slides in 10 mmol/L sodium citrate buffer (pH 6) for 20 min. After preincubation with hydrogen peroxide, the slides were incubated overnight at 4°C with primary antibody to CXCR4 (clone 12G5, Zymed, San Francisco, CA) at 1:100 dilution. Labeling was visualized using a streptavidin-peroxidase complex and diaminobenzidine as the chromogen. Slides were counterstained with Mayer's hemalum. Isotypic (IgG2a) negative controls were done on each sample. For positive controls, paraffin-embedded tissue samples of metastatic breast carcinoma were used.

For the determination of statistical significance, ANOVA test was done with the Statview software. P < 0.05 was used as the criterion of statistical significance.