Tissue specimens

Liver cancer cell lines

Liver tumor-derived cell lines (Bel7402, Bel7404, Bel7405, QGY7701, QGY7703, SMMC7721, Hep3B, HepG2, MHCC-L, MHCC-H, Sk-hep1, Huh-7, PLC, YY-8103 and Focus) and the fetal liver-derived cell line L02 were employed in this study, where MHCC-H and MHCC-L cells were kindly provided by the Cancer Institute affiliated to Zhongshan Hospital, Fudan University. All of these cell lines were grown under standard cell culture conditions in the following media: minimum essential medium Eagle (Sigma, Dorset, UK) supplemented with 10% fetal bovine serum (Life Technologies), 1% L-glutamine (L-glut) and 1% nonessential amino acids (NEAA) in a 5% CO₂-humidified chamber.

Extraction of genomic DNA and total RNA

In this work, genomic DNA was extracted from all available specimens using the DNeasy Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer's recommendation. RNA was extracted using TRIzol solution according to the manufacturer's recommendation. And RNAse-free DNase I was used to remove DNA contamination. Total RNA concentration and quantity were assessed by absorbency at 260 nm using a DNA/Protein Analyzer (DU 530, Beckman, USA).

Semi-quantitative RT-PCR and quantitative real-time RT-PCR

Reverse transcription (RT) was performed in a 20 μl reaction system with 2 μg total RNA treated by DNase I according to the manufacturer's recommendation. Each PCR was generally performed in 35 thermal cycles and then the PCR products were observed by electrophoresis on 2% agarose gel and visualized after staining with ethidium bromide, where β-actin was employed as loading control. The primers of SFRP1 were showed as following: forward, 5'-AAAGCAAGGGCCATTTA GATTAG-3'; reversal, 5'-TTCTGGGCTTGACCTTAATTGTA-3'. The PCR product is 328 bp. β-actin primers are: forward, 5'-TCACCC ACACTGTGCCCATCTACGA-3'; reversal, 5'-CAGCGGAACCGCTC ATTGCCAATGG-3'. The PCR product is 295 bp. To further analyze expression of SFRP1 gene in HCC specimens, the relative mRNA level of SFRP1 was measured by a quantitative real-time RT-PCR using TaKaRa PCR Thermal Cycler Dice Detection System and SYBR green dye (TaKaRa, Japan) in additional 40 paired HCC specimens according to the manufacturer's recommendation. A housekeeping gene, β-actin was used as an internal control. Measurements were repeated thrice to ensure the reproducibility of results. The mRNA level of each gene in each HCC sample was normalized by comparing with the level in corresponding non-cancerous liver [27]. The average Ct value of the β-actin gene was subtracted from the average Ct value of SFRP1 for each sample: SFRP1ΔCt = (Avg. SFRP1 Ct – Avg. β-actin Ct), SFRP1ΔΔCt= (SFRP1ΔCt_HCC -SFRP1ΔCt_non-HCC). The fold change (2^-SFRP1ΔΔCt) of SFRP1 expression relative to β-actin of each HCC sample examined was calculated. The significance level was defined as p value <0.01 [28].

Immunohistochemistry

Four-μm thick sections were deparaffinized and dehydrated, and then treated with methanol containing 0.3% H₂O₂ to inhibit endogenous peroxidase. The slides were incubated with rabbit anti-SFRP1 polyclonal antibody (1:100 dilution, 600-401-475, Rockland Inc.) at 37°C for 2 hrs, and then at 4°C overnight, followed by the incubation with a horseradish peroxidase-conjugated anti-rabbit antibody (Dako Japan Ltd., Kyoto, Japan) at 37°C for 1 h. The signals were detected using Diaminobenzidine Substrate Kit (Vector Laboratories, Burlingame, CA). Counterstaining was performed with hematoxylin. Slides incubated with PBS buffer instead of the primary rabbit antibody were used as negative controls, whereas normal liver in which SFRP1 was known to be strongly positive were used as positive controls in each experiment. In addition, the slides with HCC specimens and corresponding adjacent non-HCC livers were simultaneously examined via the immunohistochemistry staining, and then were assessed by visual inspection and the estimation of the percentage of immunopositive cells. The HCC specimens with less than 10% immunopositive cells were considered as negative. Tissues were graded on a scale of negative (-), low expression (+), high expression (++), or strong expression (+++). If ++ or +++ was observed, the specimen was considered to be strongly positive.

Western blot analysis

Total proteins from cultured cell lines were subjected to protein gel electrophoresis using 12% SDS-PAGE and transferred onto Hybrid- PVDA membrane (Amersham life Science) treated by 20% methanol in Tris-glycine buffer. After blocked in PBS containing 5% BSA, the membrane was incubated for immunoblotting analysis with rabbit anti-SFRP1 polyclonal antibody (600-401-475, Rockland Inc.) by 1:300 dilutions at room temperature for 2 hrs, and then with goat-anti-rabbit secondary antibody for 40 min. Finally, the signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences).

Cell transfection and cell proliferation

All HCC cell lines were grown at 37°C in Dulbecco's modified Eagle's medium (Sigma Chemicals, St. Louis, Missouri, MO) supplemented with 10% fetal bovine serum (Life Technologies), in a 5% CO₂-humidified chamber. To observe the cell proliferation, the recombinant plasmids pcDNA3.0 containing full-length ORF of SFRP1, which was amplified from a normal liver with high fidelity PCR Enzyme (PrimerStar, TaKaRa, Dalian, China), were transiently transfected into target cells using Lipofectamine™ 2000 Transfection Reagent (Invitrogen) according to the manufacturer's instruction. The transfected cells were seeded in 96-well plate at 2 × 10³ cells per well and then cultured for 5–8 d. 10 μl of CCK-8 (Cell Counting Kit-8, Dojindo Laboratories, Kumamoto, Japan) solution was added to each well of per plate, and incubated the plate at 37°C for 1 h. The absorbance at 450 nm was measured to represent the cell viability. To establish the stable offspring cell lines with exogenous SFRP1, above plasmids and empty vectors as control were transfected into SMMC7721 cells and then G418 (Life Technologies, Inc., Paisley, UK) was added to the medium at a final concentration of 700 μg/ml. After 3 weeks, the remaining colonies were individually picked and expanded. The expression of exogenous SFRP1 in these offspring subclones was checked by western blotting using anti-SFRP1 polyclonal antibody. Cell viability was measured to assess the cell proliferation of those stable SMMC7721 subclones with exogenous SFRP1 according to the described method above. All experiments were independently repeated at least three times.

Colony formation

Plasmids pcDNA3. 0 containing SFRP1 or empty vector as control were transfected into Hep3B and YY-8103 cells in 35-mm dishes by Lipofectamine 2000 (Invitrogen) for 24 hrs, and then stripped and plated on 100-mm tissue culture dishes, and then G418 (Life Technologies, Inc., Paisley, UK) was added to the medium at a final concentration of 700 μg/ml. After 3 weeks, the remaining colonies were counted on crystal-violet-stained dishes.

Loss of heterozygosity (LOH) analysis

LOH analysis were here performed using DNA sequencing in 46 pairs of HCC and adjacent non-cancerous livers with the polymorphic microsatellite markers D8S532 (Forward primer: GCTCAAAGCCTCC AATGAC; Reverse primer: GACTTCGTGATCCACCTGC, the size of PCR product is about 240 ~ 260 bp) and D8SAC016868 (Forward primer: AAGTCAAAGGCCAGGGTGT; Reverse primer: TCATGTGTTTCCC AGGAATG, the size of PCR product is about 230 ~ 250 bp), located close to the SFRP1 locus. Amplification was done in 5 μl volumes with 20 ng of genomic DNA, 0.06 μmol/L of each primer (5' fluorescent-labeled primers), 0.2 mmol/L each dinucleotide triphosphate, and 0.2 U hotstar Taq polymerase under the following conditions: 94°C (hot start) for 10 min, followed by 30 cycles at 94°C for 30 sec, at 55°C for 30 sec, and at 72°C for 30 sec, with a final extension at 72°C for 10 min. PCR products were analyzed using ABI 3730 sequencer. LOH was analyzed by determining the fluorescent intensity of each allele and calculating the ratio using peak height as the following formula: $\text{LOH}=\begin{array}{l}\frac{\text{Height of normal allele two}}{\text{Height of normal allele one}}\hfill \\ \frac{\text{Height of tumor allele two}}{\text{Height of tumor allele one}}\hfill \end{array}$, where the longer or shorter alleles were considered to be significantly lost in HCC specimens when an LOH value ≤0.5 or ≥1.5 was observed, respectively.

Treatment of 5-aza-2'-deoxycytidine and trichostatin A

To evaluate whether the genomic DNA methylation can contribute to the re-expression of SFRP1, both 5-aza-2'-deoxycytidine (DAC) (1000 nM) (Sigma), a demethylation reagent, and trichostatin A (TSA) (300 nM) (Wako BioProducts, Richmond, VA), an inhibitor of histone deacetylase, were employed to treat Bel7404, QGY7701 and MHCC-H cells with low expression of endogenetic SFRP1.

Methylation-specific PCR and Bisulfite Sequencing

We treated DNA with bisulfite according to the previous description [29]. Briefly, 1 μg of genomic DNA was denatured by incubation with 0.2 M NaOH. Aliquots of 10 mM hydroquinone and 3 M sodium bisulfite (pH 5.0) were added and the solution was incubated at 50°C for 16 hrs. To analyze the DNA methylation status of the CpG islands of SFRP1 in HCCs and cell lines, Methylation-Specific PCR (MSP) was performed with genomic DNA treated by bisulfite, where specific primers for unmethylated and methylated DNA were designed within the CpG island of SFRP1, as following: Methylation (forward: AGTTAGTGTC GCGCGTTC; reversal: CCGATACCCATACCGACTC) and Unmethylation (forward: GGAGTTGGGGTGTATTTAGTTTG; reversal: CCAATACCCATACCAACTCTACA). The lengths of "M" and "U" product are 299 bp and 247 bp, respectively. In addition, DNA sequencing on PCR products was also carried out to further assess the DNA methylation status of SFRP1 promoter, where the CpG islands-enriched region within SFRP1 promoter was amplified with bisulfite-treated genomic DNA by using the primers (forward: TTTATGGGTTTGTAAGTATGATTTAGG; reversal: ACAAATTAAAC AACACCATCTTCTT). The length of product is 897 bp, and then the PCR products were inserted into pMD 18-T vector (TaKaRa Inc. Japan) for DNA sequencing on ABI 3730 sequencer.

RNA interference using small interference RNA (siRNA)

Two siRNAs against SFRP1 were designed and chemically synthesized (Shanghai GenePharma Co., Shanghai, China) for targeting different coding regions of the gene as following: siRNA-SFRP1_888 (5'- GGCCAUCAUUGAACAUCUCtt-3' and 5'-GAGAUGUUCAAUGAU GGCCtt-3') for nt 888–909 of SFRP1; and siRNA-SFRP1_1094 (5'- GCCACCACUUCCUCAUCAUtt-3' and 5'-AUGAUGAGGAAGUGGU GGCtt-3') for nt 1094–1115 of SFRP1. In addition, a negative control, termed as siRNA_NC (5'-UUCUCCGAACGUGUCACGUtt-3' and 5'-ACGUGACACGUUCGGAGAAtt-3') was also synthesized in this study. All above siRNAs were transfected into SMMC7721 cells to observe the rescue of depressed cell growth triggered by exogenous SFRP1.

SFRP1 was frequently down-regulated in primary HCC

To evaluate the transcriptional expression of SFRP1 in primary HCCs, semi-quantitative RT-PCR was employed to detect the mRNA level of SFRP1 in 120 pairs of HCC specimens and their adjacent non-cancerous liver tissues. The results showed that SFRP1 was frequently down-regulated in 48% (58/120) HCC specimens as compared with adjacent non-cancerous livers, of which the representative RT-PCRs from 24 pairs of HCC samples were showed as Figure 1A. To confirm the absence of genomic DNA contamination, 6 paired HCCs and non-HCCs were randomly selected to detect the expression of SFRP1 and β-actin by RT-PCR. The products of SFRP1 and β-actin were confirmed by DNA sequencing. The resulting data showed that no PCR products were amplified in all of samples without RT (RT-), indicating that there is no genomic DNA contamination in these RNA samples (Fig. 1B). Considering the limitation of RT-PCR method, the mRNA level of SFRP1 was also further evaluated to confirm the down-regulation of this gene in 46 informative cases (Table 1) through real time RT-PCR. Of these 46 cases, 35 (76.1%) HCCs showed at least a 2-fold reduction of the SFRP1 mRNA level as normalized by β-actin level in each sample as compared with that of the corresponding nontumorous livers. The resulting data showed that the transcriptional expression of SFRP1 was significantly reduced in HCC (p < 0.001, Fig. 1C) by comparing between tumor and non-cancerous liver groups. However, the down-regulation of SFRP1 was not statistically correlated with the gender, age (≥60 or <60), or tumor size (≥3 or <3 cm) (p > 0.05, Table 2).

To further confirm the down-regulation of SFRP1 at protein levels, immunohistochemical staining was performed on an additional 100 pairs of HCC specimens and corresponding adjacent non-cancerous livers using tissue array (Shanghai OUTDO Biotech Co, LTD, China), where immunohistochemical-staining intensity was scored on a scale of 1+ to 3+. Interestingly, SFRP1 was found also decreased in 30 (30%) of 100 HCC specimens as compared to adjacent non-cancerous livers (p < 0.05), where the stain intensity of SFRP1 in those non-cancerous livers was generally scored to the scale of 3+. The immunohistochemical staining showed that SFRP1 was mostly anchored in cytoplasm and extracellular matrix (ECM) (Fig. 1D), in coincidence with the character of SFRP1 as a secreted protein. Furthermore, we evaluated the expression level of SFRP1 in available HCC cell lines by RT-PCR. The resulting data showed that SFRP1 was significantly expressed in Bel7405, QGY7703, MHCC-L, Sk-Hep1, HuH-7, PLC, and Focus cell lines, whereas no or weak expression of the gene was found in Bel7402, Bel7404, QGY7701, SMMC7721, Hep3B, HepG2, MHCC-H, L02 and YY-8103 cell lines (Fig. 1E). Together these findings indicated that the down-regulation of SFRP1 could be an important event in oncogenesis of HCC.

Exogenous SFRP1 could inhibit cell growth of HCC cells

To assess whether the down-regulation of SFRP1 could contribute to hepatocarcinogenesis, plasmid pcDNA3.0 with full ORF of SFRP1 under the control of the SV40 promoter were first transiently transfected into YY-8103, a HCC cell line, without the significant expression of endogenetic SFRP1 (Fig. 1E), where the empty vector pcDNA3.0 was used as control (Fig. 2A). The interesting results showed that exogenous SFRP1 can significantly inhibit the cell growth of YY-8103 cells as compared to the vector alone (Fig. 2B), suggested that SFRP1 as a Wnt pathway antagonist could play important roles in regulating negatively the cell growth of HCC-derived cells. To further test the negative effect of SFRP1 on the cell growth of HCC cells, the colony formation assay was performed on another HCC cell line, Hep3B, without the significant expression of endogenetic SFRP1, through the transient cell transfection with the same plasmids, pcDNA3.0 with full ORF of SFRP1 and empty vector (Fig. 2C). After cultured in G418 for 21 days, few colonies can be formed when the cell transfection using pcDNA3.0 with SFRP1, whereas the colony formation was still obvious as the cell transfection with empty vector alone (Fig. 2D). The dramatic reduction of colony formation from Hep3B cells further suggested that SFRP1 could be a potent inhibitor for negatively regulating the cell growth, possibly through the opposed effect of Wnt-β-catenin pathway, which was well-known as an important contributor to the oncogenesis of HCC [30–33]. To further observe the long-term effect of SFRP1 on the cell growth of HCC cells, the same plasmids were stably transfected into SMMC7721 cells, also a HCC cell line. After screening the transfected cells by western blotting assay, five stable offspring subclones were picked up and maintained. Among them, two stable subclones (SMMC7721Z and SMMC7721Y) with significant expression of exogenous SFRP1 were further evaluated by this study (Fig. 3A). The resulting data showed that SMMC7721Z and SMMC7721Y with stable expression of SFRP1 exhibited the depressed cell growth as compared to parent SMMC7721 cells without endogenous expression of the gene (p < 0.01, Fig. 3B). These transient and long-term effects of SFRP1 on different HCC cells support the notion that SFRP1 as a candidate tumor suppressor genes could be involved in hepatocarcinogenesis. To eliminate the artificial growth suppression induced by the overexpression of various genes, we compare the expression level of SFRP1 in SMMC7721Z containing exogenous SFRP1 with that of normal liver through RT-PCR (Fig. 3C). The resulting data showed that the mRNA level of exogenous SFRP1 in SMMC7721Z was not higher than that of human normal liver, implying that the growth suppression of SMMC7721Z could be induced by SFRP1 itself, not by the artificial overexpression of ectopic gene.

Rescue of the suppressive cell growth of exogenous SFRP1 through RNA interference

To further evaluate the contribution of the down-regulation of SFRP1 to oncogenesis of HCC, two small interference RNAs (siRNAs), siRNA-SFRP1_888 for nt 888–909 and siRNA-SFRP1_1094 for 1094–1115 of SFRP1, were designed and chemically synthesized for the knockdown of SFRP1. To test the efficacy of these siRNAs, the siRNAs were transiently transfected into SMMC7721Z cells with exogenous SFRP1 because the endogenetic SFRP1 was difficult to be detected by using western blotting assay, where siRNA-NC was used as a reference control. The resulting data indicated that these two siRNAs can significantly knockdown the exogenous SFRP1, as compared with siRNA-NC (Fig. 3D). Afterward, these siRNAs were further transiently transfected into some HCC cell lines, such as MHCC-L and Sk-hep-1 cells, with the endogenetic SFRP1. However, the growth of these cells had no significant alteration, whatever promotion or suppression of cell growth, implying that those HCC cell lines could be inappropriate to evaluate the effect of SFRP1 on cell growth.

Furthermore, we still chose SMMC7721 cells to evaluate the efficiency of these siRNAs, because the above data indicated that SMMC7721 cells exhibited the response to the exogenous SFRP1. Here, these siRNAs were transiently transfected into stable offspring SMMC7721Z cells with constitutional expression of exogenous SFRP1. Interestingly, both siRNA-SFRP1_888 and siRNA-SFRP1_1094 could obviously promote cell growth of SMMC7721Z cells as compared to siRNA_NC (p < 0.05, Fig. 3E), indicated that the suppressive cell growth of exogenous SFRP1 could be rescued by the RNA interference. The data also further supported that the frequent down-regulation of SFRP1 could contribute to hepatocarcinogenesis via promoting the cell growth of given HCC cells, since the cell growth of some HCC cells was indeed negatively regulated through the expression level of SFRP1.

LOH at the SFRP1 locus in HCCs

To address whether the genetic aberrations could contribute to the down-regulation of SFRP1, genomic imbalance of the SFRP1 locus was evaluated in 46 pairs of HCC specimens with or without SFRP1 expression by using the microsatellite markers D8S532 and D8SAC016868, which were found on chromosome 8p11.2 flanking the SFRP1 locus (Fig. 4A). Among these 46 pairs of HCCs examined, LOH of these two microsatellite markers was identified in 13% (6 of 46) and 6.5% (3 of 46) HCC specimens (Fig. 4B), respectively, where a total of 8 (17.4%) HCCs was considered as involving LOH. Interestingly, SFRP1 was down regulated in seven of eight HCC specimens. However, the overall resulting data suggested that low frequency of LOH at the SFRP1 locus could not be a crucial genetic event in HCCs, although epimutation could occur by the DNA methylation on the remaining allele in those HCC samples when an allele was lost.

DNA methylation status of SFRP1 promoter in HCC cell lines

To assess the DNA methylation level within SFRP1 promoter in these cell lines, MSP was performed on genomic DNA treated by bisulfite with the designed primers. Interestingly, the results showed that DNA hypermethylation within SFRP1 promoter indeed occurred in Bel7402, SMMC7721, Bel7404 and YY-8103 cells without the expression of endogenetic SFRP1, whereas partially or complete unmethylation of the region were found in Bel7405 and Sk-Hep1 cells, respectively (Fig. 5A). Similarly, as a control, we here evaluated the expression of SFRP1 and DNA methylation level within SFRP1 promoter in both fetal and adult livers. The resulting data demonstrated that the obvious DNA hypomethylation of the promoter was significant in the livers, in consistence with the high expression of SFRP1. The findings implied that the methylation status of SFRP1 promoter could be associated with the expression of the gene.

To further investigate whether the epigenetic events could contribute to the down-regulation of SFRP1, both the demethylation agent DAC and the histone deacetylase inhibitor TSA were employed to treat some HCC cell lines, such as Bel7404, QGY7701, MHCC-H cells, without the expression of endogenetic SFRP1 (Fig. 1E). Interestingly, the resulting data showed that SFRP1 was significantly upregulated in Bel7404, QGY7701, MHCC-H cells through the treatment by DAC, not by TSA (Fig. 5B), implying that the DNA methylation status of genomic DNA could be correlated with the dysregulation of SFRP1. To further address the relationship between the methylation status on SFRP1 promoter and the expression of the gene, MSP was performed on these three HCC cell lines to detect the methylation levels of CpG islands within SFRP1 promoter. Expectedly, MSP data suggested that the CpG islands within SFRP1 promoter could be in DNA hypermethylation in all parent cell lines, whereas the unmethylated CpG islands occurred in the cell lines as the treatment of DAC, not TSA, along with the re-expression of endogenetic SFRP1 (Fig. 5C). These findings indicated that the DNA hypermethylation of SFRP1 promoter might be responsible for the silence of SFRP1 transcription.

DNA methylation of the SFRP1 promoter in primary HCCs

To quantify the DNA methylation status of SFRP1 promoter in clinical HCC samples, bisulfite-treated genomic DNA sequencing was employed to analyze the methylation level of CpG islands in three pairs of HCC specimens with the down-regulation of SFRP1 (Fig. 6A). The DNA sequencing on PCR products revealed that the methylation levels of CpG islands around the transcription start site (TSS) of SFRP1 was significantly increased in two HCC specimens, 11C and 38C, as compared to that of non-cancerous livers (p < 0.01) (Fig. 6B), whereas the methylation status in one HCC sample was not significantly changed. The resulting data suggested that the DNA hypermethylation of CpG islands around TSS of SFRP1 can partially contribute to the down-regulation of SFRP1 in some HCC specimens, although other mechanisms, such as LOH and/or other epigenetic alterations, may also contribute to the down-regulation of SFRP1 in HCCs.