We obtained paired liver tissues (tumor vs. non-tumor) from 45 patients with HCC who received hepatic resection between July 2008 and January 2009 at Beijing You’an Hospital. One part of resected tissues was fixed in formalin and embedded in paraffin for histological examination. The remained liver tissues were snap-frozen and stored at −80°C for real-time PCR and western blot analyses. Clinical characteristics of these patients are shown in Table I. HCC was confirmed by histological examinations according to the AASLD guidelines proposed in 2005 (18). This study was approved by Research Ethics Committee, You’an Hospital affiliated with Capital Medical University. Written informed consent was obtained from each patient.

Four-μm paraffin-embedded sections were deparaffinized and rehydrated. Antigen was retrieved using heat-induced epitopes in 10 mM citrate buffer, pH 6.0 and endogenous peroxidases were blocked for 5 min using 0.3% H₂O₂. Then, the sections were incubated with mouse monoclonal anti-GPC3 antibody (diluted 1:100, Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight in a cool room. Next day, the slides were applied for horseradish peroxidase-labeled secondary antibody (Zhongshan Golden Bridge, Beijing, China) after three PBS washes. To evaluate GPC3 immunostaining, the percentage of positive cells and intensity of staining were analyzed. The intensity of immunostaining was semi-quantitatively estimated according to the following scores: 0, no positive staining; 1, positive staining was weak yellow; 2, positive staining was yellow; and 3, strong positive staining was brown. The mean percentage of positive tumor cells was determined after counting positive GPC3 cells in five fields under ×400 magnification. The scores were defined as follows: 0, ≤10%; 1, 10%–25%; 2, 25%–50%; 3, 50–75%; and 4, >75%. The final GPC3 immune score was calculated as staining intensity × mean percentage of positive tumor cells. We defined the scores ≤4 as low expression of GCP3 and 5–12 as high expression of GPC3.

Total RNA was purified from frozen HCC tissues using an RNeasy Mini kit (Qiagen, Hilden, Germany). Total RNA (1 μg) was reverse transcribed using a Reverse-Transcription system (Promega, Madison, WI, USA). Transcript levels of GPC3 cDNA were quantified using real-time PCR (Rotor-Gene Q, Qiagen, Germany). The primers used to amplify GPC3 were: forward, 5′-TTCTCAACAACGCCAATA-3′, and reverse, 5′-GATGTAGCCAGGCAAAGC-3′. Amplicon expression in each sample was normalized to β-actin. The primers used to amplify β-actin were: forward, 5′-ACCCACACTGTGCCCATCTA-3′, and reverse, 5′-GCCACAGGATTCCATACCCA-3′. After normalization, expression of GPC3 was quantified using a 2^−ΔΔCt calculation. PCR was performed in duplicate and three independent experiments were performed.

The human HCC cell lines HepG2, Hep3B and Huh7 were obtained from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). HepG2, Hep3B and Huh7 were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco Life Technology, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS; Hyclone, Rockford, IL, USA), 100 μg/ml streptomycin, and 100 U/ml penicillin (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) in a humidified atmosphere of 5% CO₂ at 37°C. For experiments, cells at 70% confluence were serum-starved 24 h before treatment. To block the ERK pathway, cells were preincubated with PD98059 (Sigma-Aldrich, St. Louis, MO, USA) for 3 h before GPC3 (Sino Biological Inc., Beijing, China) treatment.

Western blotting was performed as described previously (19). Briefly, total cellular proteins were extracted using cell lysis buffer (20 mM Tris/HCl, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS and protease inhibitor cocktail, pH 7.5). The protein concentrations of the lysates were determined according to the bicinchoninic acid (BCA) method using a protein assay kit (Pierce Biotechnology, Rockford, IL, USA). Cell or liver tissue lysates containing 100 μg protein were generated and total proteins were separated by standard SDS-PAGE, followed by transfer to polyvinylidene difluoride membranes (Millipore, Billerica, MA, USA). Membranes were incubated with primary antibodies against GPC3 (diluted 1:500, Santa Cruz Biotechnology), MMP9 (diluted 1:1,000, Cell Signaling Technology, CA, USA), E-cadherin (diluted 1:1000, Santa Cruz Biotechnology), α-SMA (diluted 1:1,000, Cell Signaling Technology), β-catenin (diluted 1:1,000, Cell Signaling Technology), Snail (diluted 1:1,000, Cell Signaling Technology), p-ERK (diluted 1:1,000, Cell Signaling Technology), ERK (diluted 1:1,000, Cell Signaling Technology), or Sonic hedgehog (Shh; diluted 1:500, Santa Cruz Biotechnology). Next, membranes were incubated with an appropriate horseradish peroxidase-conjugated secondary antibody. Reactive bands were detected using enhanced chemiluminescence reagents (Applygen Technologies Inc., Beijing, China). To ensure equal loading of samples in each lane, membranes were stripped and reprobed with an anti-glyceraldehyde 3-phosphate dehydrogenase antibody (GAPDH; diluted 1:10,000; Kang Chen, Shanghai, China). The relative densities of the protein bands were quantitatively determined using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

The migration of cells was examined with scratch assays. When cells grew to 90% confluency, a scratch wound in the monolayer was made using a pipette tip. After washing away all detached cells with PBS, the remaining cells were treated with 0.1 μg/ml GPC3. The wound distances were measured by microscope at 0, 24, and 48 h after treatment. Each test was performed in triplicate.

Serum-starved cells were resuspended in DMEM free of FBS and seeded in a 24-well plate at a concentration of 10⁵ cells/well on inserts with 8-μm filter pores, which were either uncoated (migration assay; BD Falcon™ Cell Culture Insert, San Jose, CA, USA) or Matrigel-coated (invasion assay; BD Biocoat™ Matrigel™ Invasion Chamber). As a chemoattractant, 1% FBS was added to the lower chamber while FBS-free DMEM with 0, 0.05, 0.1 or 0.2 μg/ml GPC3 or 0, 25, 50 or 100 nM PD98059 was put into the upper chamber, respectively. After 24 h, cells were fixed and stained with methanol and hematoxylin. Migrated or invaded cells were counted at ten random fields per triplicate filter under an inverted microscope whereas non-migrating and non-invading cells were removed from the upper surface by wiping with a cotton swab. Each test was performed in triplicate.

Clinicopathological parameters in patients with GPC3 high/low were compared using Pearson’s χ² test or Fisher’s exact test. Continuous variables were analyzed with one-way analysis of variance. The results of multiple observations are presented as the means ± SDs of at least three independent experiments. The Kaplan-Meier method was used to determine survival probability and differences between groups were assessed using the log-rank test. Overall survival (OS) was defined as the interval between surgery and death or the last follow-up time-point. Statistical analyses were conducted using SPSS version 17.0 software (SPSS Inc., Chicago, IL, USA).

To examine whether GPC3 involved in HCC progression, GPC3 mRNA and protein levels were assessed in liver tissues from patients with different HCC stages. The Barcelona Clinic Liver Cancer (BCLC) system was applied to classify the disease stage (20). Among the 45 pathologically diagnosed HCC patients, 13 had BCLC-stage A, 14 BCLC-stage B, and 18 BCLC-stage C. The expression of GPC3 was examined in all HCC tissues using an anti-GPC3 antibody. Immunostaining of GPC3 was detected in the majority of hepatocyte membranes and cytoplasms of most HCC cancer cells (Fig. 1A). Notably, both mRNA and protein expression of GPC3 was remarkably elevated in HCC patients with BCLC-stage C compared to those with BCLC-stage A or B (Fig. 1B and C). Real-time PCR and western blot analyses showed that GPC3 was upregulated on average ≤2.4-fold in patients with BCLC-stage C compared to those with BCLC-stage A or B (P=0.003, Fig. 1B and C). There were no significant difference of GPC3 expression between patients with BCLC-stages A and B (P>0.05; Fig. 1B and C), or the paracancerous tissues of different stages (P>0.05; Fig. 1B and C).

As mentioned in Materials and methods, we divided 45 patients into GPC3^high and GPC3^low groups based on GPC3 immune positivity. Table II shows that 22 patients (48.9%) belonged to GPC3^high group and 23 (51.1%) GPC3^low group, respectively. Correlations between GPC3 expression and clinicopathological features were further analyzed. Consistent with the findings based on real-time PCR and western blotting, there was a significant difference in GPC3 expression levels between BCLC-stages C and A (P=0.002) or B (P=0.001), while there was no significant difference between BCLC-stages A and B. In addition, high levels of GPC3 expression were strongly correlated with the rate of vascular invasion (P<0.001). No other clinical characteristics, including age, gender, hepatitis B virus infection, serum AFP level, tumor size, tumor number, and Child-Pugh class, were related to the expression of GPC3. We then investigated whether the GPC3 expression is associated with the prognosis of HCC patients. The 5-year overall survival rate of HCC patients with low GPC3 expression levels reached 80.8% whereas only 31.8% of the HCC patients with high GPC3 expression were still surviving at the end of the follow-up (P=0.042, Mantel-Cox test; Fig. 1D).

HCC progression and metastasis were highly associated with phenotypic alterations of cancer cells, particularly EMT (21,22). In this study, we measured the expression of the epithelial cell markers, e.g., E-cadherin and β-catenin, and the mesenchymal markers such as MMP9 and α-SMA, as well as Snail, a transcription factors associated with EMT. Western blot analyses showed that the expression of Snail, MMP9, β-catenin and α-SMA in cancerous tissues with BCLC-stage C were significantly increased compared to those with BCLC-stages A or B (P<0.05, Fig. 2A). By contrast, expression of E-cadherin in cancerous tissues with BCLC-stage C was lower than in those with BCLC-stages A or B (P<0.05; Fig. 2A). There was a positive correlation between the expression levels of GPC3 and MMP9, β-catenin, and α-SMA, but a negative correlation between the expression of GPC3 and E-cadherin (Fig. 2B). These results strongly suggest that GPC3 might contribute to EMT of HCC.

To further clarify the potential association between GPC3 expression and HCC metastasis, we investigated the expression of GPC3 and the metastatic capacity in HCC cell lines e.g., HepG2, Hep3B and Huh7. Migration and invasion assays indicated that HepG2 cells had the highest and Huh7 cells had the lowest metastatic capacity among three HCC cell lines (P<0.05; Fig. 3A). Western blotting showed that HepG2 cells expressed the highest levels of GPC3, whereas Huh7 cells expressed the lowest (Fig. 3C). Alternatively, HepG2 cells expressed higher levels of α-SMA and lower levels of E-cadherin compared to Hep3B and Huh7 cells (Fig. 3C and D).

In addition to the cell surface expression, GPC3 can also be detected in the extracellular environment after being released from the cell membrane by Notum, a lipase that cleaves GPI anchors (4). Next, the impact of soluble exogenous GPC3 on HCC cellular motility was studied. Scratch-assays displayed an obviously greater migration capacity of both HepG2 and Hep3B cells when they were treated by exogenous GPC3 compared to untreated cells (Fig. 4A). Additionally, in Boyden chamber and Matrigel invasion assays, exogenous GPC3 induced dynamic migration and invasion of both HepG2 and Hep3B cells in a dose-dependent manner (Fig. 4B and C). Furthermore, GPC3 treatment significantly reduced expression of the epithelial cell markers such as E-cadherin and increased the mesenchymal markers MMP9 and α-SMA in a dose-dependent manner in both HepG2 and Hep3B cells.

Given the critical roles of the mitogen-activated protein kinase (MAPK) (23) and Sonic hedgehog (Shh) signaling pathways in EMT (24), we investigated the impact of GPC3 in the two pathways. We found that ERK activity was strongly elevated in HCC patients with BCLC-stage C (P<0.05; Fig. 5A). A positive correlation between GPC3 expression and phospho-ERK was observed in HCC tissues by regression analysis, with a coefficient (R) of 0.803 (P<0.001; Fig. 5B). In vitro, the exposure of cells to soluble GPC3 promoted the phosphorylation of ERK in a dose-dependent manner in both HepG2 and Hep3B cells (Fig. 5C). To further confirm whether GPC3-induced cell migration and invasion is via the ERK pathway, we applied PD98059, an inhibitor of MEK (a kinase that activates ERK), to HepG2 cells prior to GPC3 treatment. We observed that PD98059 could decrease HepG2 cell migration and invasion. Furthermore, this inhibitor significantly attenuated the GPC3-induced migration of HepG2 cells in a dose-dependent manner (25 nM, 82.80±16%; 50 nM, 35.19±6%; and 100 nM, 24.32±6%). In an invasion assay, PD98059 reduced the GPC3-induced invasion capacity of HepG2 cells in a dose-dependent manner (25 nM, 73.00±6.74%; 50 nM, 41.70±3.07%; and 100 nM, 35.54±2.13%; Fig. 6A). Notably, PD98059 significantly attenuated GPC3-induced EMT. As shown in Fig. 6B, E-cadherin was increased whereas MMP-9, α-SMA and Snail were decreased under PD98059 treatment. Besides the ERK pathway, we also investigated whether the Shh pathway was involved in GPC3-induced EMT. Unexpectedly, no significance was obtained on Shh expression among different stages (P>0.05; Fig. 5A). The correlation of Shh expression with GPC3 was weaker compared to ERK (Fig. 5B). In vitro, GPC3 treatment could not markedly increase the Shh expression (Fig. 5C).