137 paraffin-embedded HCC samples and 49 noncancerous paraffin-embedded normal liver samples were obtained from People's hospital of Hunan Province, China. In 137 HCC cases, there were 87 male and 50 female ranging in age from 14 to 79 years (median, 48 years). For the use of these clinical materials for research purposes, prior consents from the patients and approval from the Ethics Committees of People's hospital of Hunan Province were obtained and all the procedures have been performed in compliance with the Helsinki Declaration. All specimens had confirmed pathological diagnosis and were staged according to the 2002 hepatocellular carcinomas staging system of the International Union Against Cancer(UICC).

Immunohistochemistry was performed according to according to standard protocol 20 21 22 Paraffin sections (3 μm) from 137 HCC samples and 49 normal liver samples were deparaffinized in 100% xylene and re-hydrated in descending ethanol series according to standard protocols. Heat-induced antigen retrieval was performed in 10 mM citrate buffer at 100°C for 2 min. Endogenous peroxidase activity and non-specific antigen were blocked with peroxidase blocking reagent containing 3% hydrogen peroxide and serum followed by incubation with rabbit anti-human HDGF antibody (1:100,Proteintech Inc,USA) overnight at 4°C. After washing, the sections incubated with biotin-goat anti-mouse/rabbit antibody at room temperature for 10 minutes were then conjugated with horseradish peroxidase (Maixin Inc, China). The peroxidase reaction was developed with 3, 3-diaminobenzidine chromogen solution in DAB buffer substrate. Sections were visualized with DAB and counterstained with hematoxylin, mounted in neutral gum and analyzed using a bright field microscope.

The immunohistochemically-stained tissue sections were reviewed and scored separately by two pathologists blinded to the clinical parameters. HDGF expression in the nucleus was independently evaluated. Cases in which <90% and >90% of cancer cells at levels greater than or equal to what is observed in the endothelial cells were regarded as HDGF expression index (EI) levels I and II, respectively 18 19 . HDGF EI was determined separately for the nucleus.

The hepatoma cell line HepG2 was acquired from the Cancer Institute, Central South University and cultured in RPMI1640 medium (HyClone Inc, USA) supplemented with 10% fetal bovine serum (FBS) (PAA Laboratories, Inc, Austria) in a 37°C, 5% CO₂ incubator.

Total RNA was extracted from approximately 1 × 10⁶ cells using Trizol reagent (Invitrogen, USA) following the manufacturer's instructions. cDNA was then synthesized from 2 μmug total RNA using oligo(dT)15 as the primer along with the MMLV reverse transcriptase (Takara Inc, Japan). To determine the steady-state mRNA level of HDGF in HepG2 cells, PCR was performed with the following primers: HDGF forward primer 5'-GAGGGTGACGGTGATAAGAA-3', reverse primer 5'-GAAACATTGGTGGCTACAGG-3', and the amplicon size is 377 bps. GAPDH (internal control) forward primer 5'-TTCGCTCTCTGCTCCTC-3', reverse primer 5'-GATGATCTTGAGGCTGTTGT-3', and the amplicon size is 520 bps. The PCR condition was one cycle of denaturation at 94°C for 2 min, 28 cycles of denaturation at 94°C for 20 seconds (s), annealing at 55°C for 30 s and extension at 72°C for 30 s, followed by one cycle of final extension at 72°C for 10 min. After the PCR reaction, the PCR products was loaded on 1% agarose gel and visualized by ethidium bromide staining.

The BLOCK-iT RNAi lentiviral expression plasmid was purchased from Invitrogen Inc. The shRNA sequence targeting HDGF (shHDGF) and a control shRNA sequence (shCtrl) targeting no known human genes was designed using the BLOCK iT RNAi Designer http://www.invitrogen.com as follows: shHDGF, sense 5'-CACCGCCGTGAAATCAACAGCCAAAACGTTGGCTGTTGATTTCACGG-3'; antisense 5'-AAAACCGTGAAATCAACAGCCAACTTTTGGCTGTTGATTTCACGGC-3'. shCtrl, sense 5'-CACCGCCCTGAATTGAACAGCCAAAACGTTGGCTGTTGATTTCACGG-3'; 5'-AAAACCGTGAAATCAACAGCCAACTTTTGGCTGTTCAATTCAGGGC-3'. The shRNA primers were cloned into the lentiviral expression plasmid following the manufacturer's instruction and confirmed by sequencing.

shRNA-expressed lentiviral plasmid (either HDGF-specific or control) was transfected into HepG2 cells using Lipofectamin2000 (Invitrogen) according to the Xie's instructions 23 . 48 h later, the cells were subjected to Blasticidin selection at a final concentration of 1.2 μmug/mL. Following two weeks of selection, stable pooled cells were further expanded for future experiments.

Real-time PCR was performed to measure the knock down efficiency of HDGF mRNA expression using SYBR Premix Ex Taq (Takara, Japan) as described previously 22 . The sense primer: 5' CAGCCAACAAATACCAAGTCT 3', Antisense primer: 5' GTTCTCGATCTCCCACAGC 3'. GAPDH gene was used as an inner control. The sense primer: 5' GAAGGTCGGAGTCAACGG 3', Antisense primer: 5' TGGAAGATGGTGATGGGATT 3'.

Approximately 5 × 10⁶ cells were lysed in RIPA Buffer (50 mM Tris-HCl pH 8.0, 1 mM EDTA pH 8.0, 5 mM DTT, 2% SDS) and total protein concentration determined with BCA assay (Beyotime Inc, China). 60 μmug of total protein were loaded onto 10% SDS-PAGE gel. Antibodies used for Western blot analysis included: rabbit polyclonal anti-HDGF antibody (Proteintech Inc, 1:200), anti-tubulin antibody (Santa Cruz, USA, 1:400), and HRP-conjugated anti-rabbit secondary antibody (Amersham Pharmacia Biotech, 1:50).

Cell growth was determined by MTT assay (Sigma, USA). Briefly, 1 × 10³ cells were seeded into 96-well plate with quadruplicate for each condition. 72 h later, MTT reagent was added to each well at 5 mg/mL in 20 μmuL and incubated for another 4 h. The formazan crystals formed by viable cells were then solubilized in DMSO and measured at 490 nm for the absorbance (A) values.

The anchorage-independent growth of hepatocytes was monitored by the soft agar colony formation assay. In brief, cells (100/well) resuspended in 1.5-ml mixture of 1.2% low-melt agarose and 2×RPMI 1640 (v:v = 1:1) were loaded in triplicate on the top of the solidified bottom agar comprising equal-volume mixture of 0.7% low-melt agarose and RPMI 1640 in 12-well plates. The cells were incubated at 37°C, 5% CO2 for two weeks. The colonies composed of more than 50 Cells were counted.

Cells growing in the log phase were treated with trypsin and resuspended as single-cell solution. Then the cells were counted and 1 × 10⁵ cells in 1 mL of serum-free RPMI 1640 medium were split into the upper chamber of transwell and boyden chamber (8-μmum pore size, BD Biosciences, USA) where the transwell membrane was coated either with (for invasion) or without (for migration) matrigel. Serum-free medium was also added to the lower chamber. The migration assay proceeded in 37°C, 5%CO₂ tissue culture incubator for 12 h and the invasion assay, 18 h. The non-migrated/invaded cells in the upper chamber were removed using cotton swap, and the migrated/invaded cells fixed in methanol and stained with eosin for migrated cells or Gimsa for invaded cells. The cells trapped in or attached to the reverse side of the porous membrane were photographed through × 200 microscope objective, the numbers of migrated cells in at least 5 random fields counted under phase contrast microscope, and the average calculated.

All animal protocols were approved by the Animal Care and Use Committee of Central South University. Nude mice between 4 to 6 weeks were purchased from the Animal Center, Central South University (N = 5). 1 × 10⁶ cells growing in log phase were resuspended in serum-free RPMI 1640 medium and injected subcutaneously into the 4-6 week-old male BALB/c nu/nu mice. To minimize individual difference, shCtrl cells and shHDGF cells were injected symmetrically to the left-right flank of the same mouse. The mice were maintained in a barrier facility on HEPA-filtered racks. The animals were fed with an autoclaved laboratory rodent diet. All animal studies were conducted in accordance with the principles and procedures outlined in the National Institutes of Health (NIH) Guide for the Care and Use of Animals under assurance number A3873-1. 15 days later, the mice were sacrificed with tumors isolated and their sizes and weights measured. Finally, the HDGF expression was examined again by quantitative real-time PCR in implanted nude mice of shRNA-HDGF and control cell groups. GAPDH gene was used as a normalizing control. The designed paired primers were as follow, HDGF forward, 5' CAGCCAACAAATACCAAGTCT3' reverse: 5' GTTCTCGATCTCCCACAGC 3' GAPDHforward, 5' GAAGGTCGGAGTCAACGG 3' reverse, 5' TGGAAGATGGTGATGGGATT 3'. The PCR reaction was carried out in a volume of 25 μl using SYBR Green Mix(Tiangen Inc,China) on MXP3000 Instrument (Stratagene Laboratory). Experiments were repeated three times.

All data were analyzed by SPSS 13.0 software and presented as mean ± SD. The χ² test was applied to analyze the relationship between HDGF expression and clinicopathologic characteristics. Survival curves were plotted by the Kaplan-Meier method and compared by the log-rank test. The significance of various variables for survival was analyzed by the Cox proportional hazards model in the multivariate analysis. One-way ANOVA was applied to test the differences between groups for all in vitro analyses. ANOVA test was used for the in vivo xenograft experiment. A P value of less than 0.05 was considered statistically significant.

We measured the expression level and subcellular localization of HDGF protein in 137 archived paraffin-embedded HCC samples and 49 noncancerous samples using immunohistochemical staining expression. Specific HDGF staining was mainly founded in the nuclei and cytoplasm of noncancerous and malignant epithelial cells. We observed that 53.4% (73/137) archival HCC biopsies showed HDGF EI levels 2 for the nuclei. In comparison, the rate of EI levels 2 for HDGF protein expression was 32.7%(16/49) in the nuclei of normal epithelial cells(Figure 1A-D).

We did not find a significant association of HDGF expression in nuclei of tumor cells with age, gender, HBV infection, smoking, drinking, T classification and distant metastasis in patients with HCC (p > 0.05). Interestingly, we observed that the nuclear HDGF expression was closely correlated with T classification (p < 0.001), N classification (p < 0.001) and clinical stage (p < 0.001) in patients with HCC(Table 1).

To investigate the prognostic value of HDGF for HCC, we assessed the association between HDGF expression and survival duration using Kaplan-Meier analysis with the log-rank test. The log-rank test showed that the survival time of patients with HCC was significantly different between the two groups with HDGF EI level 1 and 2 (p < 0.001). In patients with HCC, the high HDGF expression group had shorter survival, whereas the low HDGF expression group had better survival (Figure 1E).

In addition, N classification and clinical stage were also significantly correlated with survival in Kaplan-Meier analysis and log-rank test (for N classification, p = 0.028; for clinical stage, p = 0.041). To determine whether expression of HDGF is an independent prognostic factor for HCC, we performed multivariate survival analysis of HDGF protein expression and factors including with age, gender, smoking, drinking, HBV infection, T classification, N classification, distant metastasis, or clinical stage in patients with HCC. The results showed that expression of HDGF protein was an independent prognostic factor for HCC (Table 2).

To examine the biological functions of HDGF, we first measured the expression level of endogenous HDGF in HepG2 cells by RT-PCR. As shown in Figure 2A, when the PCR cycle number was controlled (28 cycles) so that both the HDGF and GAPDH products were in the linear range, HDGF showed an expression level comparable to that of the housekeeping gene GAPDH, suggesting HDGF is highly expressed in HepG2 cells, which also indicates that HepG2 is a good model system for studying the functions of endogenous HDGF by loss-of-function approach.

To stably knock down the endogenous expression of HDGF, we applied a lentiviral vector expressing specific shRNA sequence targeting HDGF (shHDGF). As a control, we stably transfected the HepG2 cells with the same lentiviral vector expressing a control shRNA sequence (shCtrl) not targeting any known human genes. By mRNA and protein expression analysis, we found that the shCtrl cells have similar HDGF level as the parental HepG2 cells, which were significantly higher than that in the shHDGF cells (Figure 2B,C).

After successfully knocking down the endogenous expression of HDGF, we first examined its effect on cell growth. As shown in Figure 3, the parental HepG2 cells had a similar growth rate as the shCtrl cells over a seven-day period, while starting from day 3 the growth of shHDGF cells were significantly slower than the former two cells (P < 0.05), suggesting HDGF promotes the growth of HepG2 cells.

We next explored the effect of HDGF on cellular transformation. Since anchorage-independent growth is a hallmark for transformed cells, we measured the growth of different cells on soft agar. Both the parental HepG2 cells and the shCtrl cells formed similar number of colonies on soft agar over a two-week period [(35.3 ± 3.5) vs. (36.7 ± 3.5)]. In contrast, knocking down endogenous HDGF dramatically reduced the number of colonies (9.3 ± 2.0)(p < 0.05), implying an essential role of HDGF in regulating cellular transformation (Figure 4).

Cell migration and invasion are integral steps for the process of tumor development and metastasis. When testing the abilities of HepG2 cells to migrate/invade through 8-μmum pores on the polycarbonate membrane either without or with pre-coated matrigel, we found the knocking down endogenous HDGF significantly reduced the potentials of HepG2 cells to both migrate and invade (p < 0.05), as compared to the parental or shCtrl cells (Figure 5).

In addition to examining the biological functions of HDGF in vitro, we also assessed the in vivo function of HDGF using a xenograft transplantation model. By subcutaneously transplanting the shCtrl or shHDGF cells into nude mice, we monitored the tumor growth over a 15-day period. As shown in Figure 6, by measuring the tumor weights, we found that shHDGF cells gave rise to significantly smaller tumors than shCtrl cells (p < 0.05). Real-time qPCR showed that HDGF expression was obviously reduced in implanted nude mice of shHDGF cell groups compared with control cell groups(Figure 6C).