Human liver cancer cell lines (SNU-449, SNU-182, Huh-7, SNU-387, SK-HEP-1, SMMC-7721, PLC/PRF/5 and Hep3B) were obtained from the American Type Culture Collection (Manassas, VA, USA). All cells were maintained in RPMI 1640 or Dulbecco's modified Eagle's medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% fetal bovine serum (HyClone; GE Healthcare Life Sciences, Shanghai, China).

A total of 111 patients with HCC who underwent surgery at the Henan Oncology Hospital (Zhengzhou, China) between 2009 and 2013 were enrolled in the present study. Disease-free liver tissues (n=12) were obtained from liver donors in the same hospital. The present study was approved by the Ethics Committee of Peking University Health Science Center (Beijing, China). Written informed consent was obtained from all patients.

RT-qPCR and quantificational methylation assay were performed as described previously (27). The primers used for this methylation status assay are listed in Table I.

The full-length PCDH9 cDNA with a His-tag at its C-terminal was cloned into the pIRES2-EGFP expression vector (Clontech Laboratories, Inc., Mountainview, CA, USA). The pAAV-U6 vector (Cell Biolabs, Inc. San Diego, CA, USA) encodes two effective shRNA (shRNA1 and shRNA2) against PCDH9. Oligonucleotides of the two shRNAs are presented in Table I.

Cell viability was measured using the 3-(4,5-di-methylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide (MTT) assay as described previously (28).

The anchorage-independent growth capability of cells was measured as described previously (28). The in vivo tumorigenic potential of cells was investigated in 6-week-old male nude mice (weight, ~20 g; n=6). Nude mice were purchased from the Department of Laboratory Animal Science in Peking University Health Science Center (Haidian, Beijing) and housed in the same place. Mice were maintained in specific pathogen-free rooms, using a microisolator (filter bonneted) or pressurized, individually ventilated cages. The animals were kept in an animal room under SPF conditions at a room temperature of 22±1°C, with 55±10% relative humidity. Food and water made freely available to the mice. Sterilized or disinfected bedding and cages were used, as was anything that came into contact with the mice. 6-week-old male nude mice received subcutaneous injections of 5×10⁶ HCC cells (SNU449, Hep3B or PLC/PRF/5) in either side of the posterior flank in a volume of 100 µl. Tumor formation was monitored every week over a 6 week period (SNU449) or 5 weeks period (Hep3B or PLC/PRF/5), depended on the size of the tumor. The tumor volume was calculated by the formula: volume=0.5 × length × width² (in mm), the maximum tumor size was 616.1 mm³ (29).

Western blot analysis was performed as described previously (28). Cells were lysed in RIPA buffer (R0278; Sigma-Aldrich, Merck Millipore, Darmstadt, Germany) and the total cellular protein was resolved on denaturing polyacrylamide gels, transferred to nitrocellulose membranes (Bio-Rad Laboratories, Inc., Hercules, CA, USA), and blotted with primary antibodies. The dilution and catalog no. of primary antibodies used in the present study are listed in Table II. Protein-antibody complexes were visualized using secondary antibodies conjugated with Cy5.5 (catalog. no. RPN998; 1:10,000 dilution; GE Healthcare Life Sciences) and visualized by LI-COR Odyssey IR Imaging System (LI-COR Biosciences, Lincoln, NE, USA).

All statistical analyses were performed with SAS 9.1 software (SAS Institute Inc., Cary, NC, USA). The comparison of the patient genome methylation status was analyzed using the Chi-square test, when n<5 Fisher's exact test was used. The significant differences in the cell growth of clones stably overexpressing PCDH9 and clones transfected with control plasmid were determined using paired t-tests. Differences in tumor growth rate were determined as determined by an analysis of variance followed by Tukey's post hoc test. All tests were 2-sided statistical analyses and P<0.05 was considered to indicate a statistically significant difference.

Downregulation of PCDH9 mRNA was detected in 5 (Huh-7, SMMC-7721, SK-HEP-1, SNU-182 and SNU-449) of the 8 HCC-derived cell lines, compared with the level observed in controls obtained from normal liver tissues (Fig. 1A).

In the aCGH assay performed on a subset of the HCC samples, deletions of the chromosomal region (13q21.32) where PCDH9 located was detected in only certain 25% (6 of 25) tumor tissues analyzed (data not shown). Therefore, to investigate if hypermethylation also contributes to PCDH9 downregulation in HCC, the methylation status of the PCDH9 promoter was examined in HCC cell lines and tumor-derived tissue samples. Subsequently, the CpG island in the promoter of PCDH9 was analyzed and the primer was designed to detect methylation status of PCDH9 as presented in Fig. 1B. Amongst the 8 HCC cell lines examined, SMMC-7721 alone exhibited significant PCDH9 promoter hypermethylation and in this case, treatment of cells with the DNA demethylation reagent 5-Aza (5-aza-2′-deoxycytidine) restored PCDH9 gene expression, however similar treatment of HCC-derived cell lines that did not show hypermethylation of the PCDH9 promoter had no effect on PCDH9 expression (Fig. 1C). In addition, it was identified that treatment of cells with the specific histone deacetylase inhibitor trichostatin (TSA), restored PCDH9 gene expression in Huh-7 and SNU-182 cells. While combined treatment of SMMC-7721 cells with TSA and 5-Aza indicated that the expression of PCDH9 was higher than 5-Aza treatment alone (Fig. 1D), these results indicated that methylation and histone deacetylation of the PCDH9 had effects on PCDH9 expression. In addition, 22% of the HCC tumor-derived tissues tested (24/111 paired samples successfully analyzed) indicated higher levels of methylation in the tumor tissue sample compared with that observed in the corresponding adjacent non-tumor tissues and normal liver tissues, however such differences did not reach statistical significance (P>0.05; Fig. 1E). Nevertheless, the level of PCDH9 mRNA in tumor tissues that exhibited hypermethylation was also identified to be significantly lower than that observed in the un-methylation group (P=0.032; Fig. 1F). These data suggest that DNA hypermethylation can also mediate the downregulation of PCDH9 expression observed in HCC. A further notable result was the significant association of PCDH9 hypermethylation with larger tumor size observed (≥5 cm; P=0.0139) and the more pronounced intrahepatic dissemination (P=0.0312) were identified (Table I).

To further explore the role of PCDH9 in HCC, the HCC-derived cell line SNU-449 that exhibited low levels of endogenous PCDH9 expression was transfected with either a PCDH9 expression constructor or control pIRES2-EGFP plasmid. Separately, two HCC cell lines (Hep3B and PLC/PRF/5) that exhibited high levels of endogenous PCDH9 expression were transfected with either PCDH9 knockdown (shRNA1 and shRNA2) or control pAAV-U6 plasmids. Following G418 selection, stable clones of SNU-449 cells in which the expression of PCDH9 had been increased were obtained, in addition to stable clones of Hep3B and PLC/PRF/5 in which endogenous PCDH9 expression had been suppressed. The overexpression of PCDH9 and suppression of endogenous PCDH9 expression was confirmed by western blot analysis (Fig. 2A).

The ability of PCDH9 expression to affect cell proliferation was analyzed in these over- and underexpressing cell clones using an MTT assay. This indicated that the growth of SNU-449 cells overexpressing ectopic PCDH9 was significantly suppressed compared with that observed in clones transfected with control plasmid (P=0.0125; Fig. 2B). By contrast, the use of shRNA1 and shRNA2 to efficiently knockdown endogenous PCDH9 expression significantly enhanced cellular proliferation in Hep3B-(P=0.0107 and P=0.0110; Fig. 2B) and PLC/PRF/5 (P=0.0118 and P=0.0120; Fig. 2B)-derived clones.

A second in vitro assay measuring the ability of these clones to form colonies in soft agar was also undertaken as an indicator of their tumorigenic potential. This indicated that the SNU-449-derived clones had decreased colony-forming efficiency compared with mock-transfected controls (P=0.0101; Fig. 2C). By contrast the PCDH9 knockdown in the Hep3B derived clones induced the cells to form much larger colonies in soft agar than that observed in controls (P=0.0025 and P=0.0086, respectively; Fig. 2C). Similar results were also obtained with the PLC/PRF/5-derived knockdown clones (P=0.0004 and P=0.0011, respectively; Fig. 2C).

To investigate tumorigenic potential in vivo, a tumor formation assay was conducted in nude mice. A total of 5 weeks after injection, significant differences in average tumor size were observed; 545.8±131.9 mm³ in mice injected with control SNU-449 cells and only 171.5±92.1 mm³ in mice injected with SNU-449 expressing ectopic PCDH9 (P=0.0151; Fig. 2D and E). This is contrasted with the five week post-injection image observed for Hep3B clones in which PCDH9 expression had been knocked-down following transfection with plasmids expressing either shRNA1 or shRNA2, an average 5 week post injection tumor size of 220.3±60.8 mm³ in mice injected with control clones and 372.7±49.8 mm³ for clones expressing shRNA1 (P=0.0364), and 309.2±38.6 mm³ for clones expressing shRNA2 (P=0.0401; Fig. 2D and E). PCDH9-knockdown with either shRNA1 or shRNA2 also resulted in PLC/PRF/5 clones which had significantly increased tumor formation when injected into nude mice compared with that produced following injection of the control counterpart (P=0.0401 and P=0.0456; Fig. 2D and E).

PCDH9 was also identified to have effects on HCC cells in which there was ectopic overexpression or knockdown of the protein PCDH9. Flow cytometry indicated that in SNU-449 cells, PCDH9 overexpression blocked G₁/S transmission with the percentage of cells in G₁ phase increasing from >55% to ~71%, whilst the percentage in S phase decreased from 29 to 16% (Fig. 3A). By contrast, in Hep3B cells, PCDH9 shRNA1 or shRNA2 decreased G₀/G₁ phase from ~55 to 48% or 50%, respectively, whilst the percentage in S phase increased from ~22 to 29% or 28%, respectively (Fig. 3A). Similarly, in PLC/PRF/5 cells, PCDH9 shRNA1 or shRNA2 decreased G₀/G₁ phase from ~60 to 51% or 54%, respectively, whilst the percentage in S phase increased from ~20 to 27% or 26%, respectively (Fig. 3A).

The expression of the cell cycle progression regulators cyclin D1, cyclin E, p21^Waf1/Cip1 and p27^Kip1 were also monitored using a western blot assay. Ectopic expression of PCDH9 in SNU-449 cells significantly increased the levels of p21^Waf1/Cip1 and p27^Kip1, however suppressed cyclin E expression and the protein level of cyclin D1 remained unchanged (Fig. 3B). In addition, in Hep3B and PLC/PRF/5 cells, PCDH9 shRNA1 or shRNA2 decreased the protein levels of p21^Waf1/Cip1 and p27^Kip1, while increasing the cyclin E expression, and the protein level of cyclin D1 was unchanged (Fig. 3C). These results suggested that it may induce G₁ phase arrest by regulating the expression of cell cycle regulators.