Hepatocellular carcinoma (HCC) is one of the most

Transcription

1 GASTROENTEROLOGY 2013;144: SPOCK1 Is Regulated by CHD1L and Blocks Apoptosis and Promotes HCC Cell Invasiveness and Metastasis in Mice YAN LI, 1,2 LEILEI CHEN, 1,2,3,4 TIM HON MAN CHAN, 1,2 MING LIU, 1,2 KAR LOK KONG, 1,4 JI LIANG QIU, 4 YAN LI, 4 YUN FEI YUAN, 4, and XIN YUAN GUAN 1,2,4, 1 Department of Clinical Oncology, 2 State Key Laboratory for Liver Research, The University of Hong Kong, Hong Kong, China; 3 Cancer Science Institute of Singapore, National University of Singapore, Singapore; and 4 State Key Laboratory of Oncology in Southern China, Sun Yat-Sen University Cancer Center, Guangzhou, China BACKGROUND & AIMS: Chromodomain helicase/adenosine triphosphatase DNA binding protein 1-like (CHD1L) is an SNF2-like transcription factor involved in the development of human hepatocellular carcinoma (HCC). Sparc/ osteonectin, cwcv, and kazal-like domains proteoglycan 1 (SPOCK1) is up-regulated by CHD1L; we investigated its role in hepatocellular carcinogenesis. METHODS: We investigated interactions between SPOCK1 and CHD1L using electrophoretic mobility shift and luciferase reporter assays. Levels of SPOCK1 messenger RNA (mrna) and protein were measured in samples of HCC and adjacent nontumor liver tissues (135 pairs) and compared using Pearson correlation coefficients. Effects of SPOCK1 overexpression and silencing were determined in HCC cell lines (QGY-7703, PLC-8024, BEL-7402, and QGY-7701). RESULTS: The CHD1L protein bound directly to the promoter region (nt-1662 to 34) of SPOCK1 and activated transcription. Levels of SPOCK1 mrna and protein were increased in 60% of human HCC samples, compared with nontumor live tissues, and was associated significantly with clinical stage. Levels of SPOCK1 mrna were increased among tumors that became metastatic, compared with those that did not, and among patients with shorter overall and disease-free survival times. Ectopic expression of SPOCK1 in HCC cells increased proliferation, foci formation, and colony formation in soft agar; these cells also formed larger xenograft tumors, more rapidly, in nude mice than control HCC cells. Silencing SPOCK1 expression with short hairpin RNA had the opposite effects. We found that SPOCK1 prevents apoptosis of HCC cells by activating Akt, to block release of cytochrome c and activation of caspase-9 and caspase-3; these effects were reversed with an Akt inhibitor. HCC cells that overexpressed SPOCK1 expressed higher levels of matrix metallopeptidase 9, were more invasive in Matrigel assays, and formed more metastatic nodules in immunodeficient mice than control HCC cells. CONCLUSIONS: CHD1L activates expression of SPOCK1, which activates Akt signaling to block apoptosis and invasion by HCC cells, in culture and in mice. Levels of SPOCK1 increase with progression of human HCC. SPOCK1 might be used as a prognostic factor or therapeutic target. Keywords: Liver Cancer; Tumor Progression; Gene Regulation; Signal Transduction. Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide and ranks as the third leading cause of cancer death. Annually, 1.5 million people are newly diagnosed with or die of HCC, and this number continues to increase. 1,2 Similar to other cancers, the hepatocellular phenotype progressively evolves into dysplastic hepatocytes with an accumulation of genetic and epigenetic changes. 3 In particular, the duplication of chromosome 1q21 is one of the most frequent alterations associated with early HCC development. 4,5 Because DNA amplification represents an important mechanism in the activation of proto-oncogenes, one or more oncogenes may harbor the 1q21 amplicon. One oncogene, chromodomain helicase/adenosine triphosphatase DNA binding protein 1-like (CHD1L), has been identified within this amplicon and plays a critical role in cell-cycle control, apoptosis, DNA repair, and metastasis. 6 9 Moreover, in a CHD1L transgenic mouse model, transgenic CHD1L expression induced the spontaneous formation of tumors. 10 Together, these data imply that CHD1L is involved in more than one regulatory pathway, which partly can be explained by its role as an SNF2-like transcription factor. 9 Further study of the CHD1L transcriptionally regulated network would aid in the elucidation of the molecular pathogenesis of HCC. Because HCC is a multistep process, further study also would help to link the early onset of chromosome 1q21 amplification with subsequent heterogeneous genetic changes. Authors share co-senior authorship. Abbreviations used in this paper: AJCC, American Joint Committee on Cancer; BAD, BCL2-associated agonist of cell death; bp, base pair; cdna, complementary DNA; CHD1L, chromodomain helicase/adenosine triphosphatase DNA binding protein 1-like; ChIP, chromatin immunoprecipitation; DIG, digoxin; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; MMP, matrix metallopeptidase; mrna, messenger RNA; OS, overall survival; PARP, poly(adenosine diphosphate-ribose) polymerase; PI3K, Phosphatidylinositol 3-kinase; qrt- PCR, quantitative real-time polymerase chain reaction; shrna, short hairpin RNA; shspock1, short hairpin Sparc/osteonectin, cwcv, and kazal-like domains proteoglycan 1; SNF2, sucrose nonfermentable 2; SPARC, secreted protein, acidic, cysteine-rich; SPOCK1, Sparc/osteonectin, cwcv, and kazal-like domains proteoglycan 1; STS, staurosporine; TUNEL, terminal deoxynucleotidyl transferase mediated deoxyuridine triphosphate nick-end labeling by the AGA Institute /$

2 180 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 To explore the regulatory network underlying CHD1Linduced hepatocarcinogenesis, CHD1L-regulated transcripts were characterized by a complementary DNA (cdna) microarray. One up-regulated gene, sparc/osteonectin, cwcv, and kazal-like domains proteoglycan 1 (SPOCK1), was selected for further study. SPOCK1 encodes a matricellular glycoprotein belonging to a novel Ca(2 )-binding proteoglycan family. Members of this protein family, which share a similar N-terminus, follistatin-like domain, and C-terminus, are involved in cell proliferation, adhesion, and migration. 11 Other members of this family include SPARC, TESTICAN-2, and TESTICAN-3; of these 3, SPARC has been well studied in various cancers. Increasing evidence has emphasized the importance of SPARC in regulating proliferation, cell-cycle progression, apoptosis, adhesion, and cell-matrix interaction. 12 SPOCK1 recently was shown to be overexpressed in gastrointestinal neuroendocrine carcinomas and prostate cancer. 13,14 More intriguingly, clinicopathologic analysis revealed that SPOCK1 may be involved in glioblastoma invasion. 15 However, the underlying mechanism of SPOCK1 overexpression is far from clear. Even less is known about the function and mechanism by which SPOCK1 contributes to cancer development and progression. In view of the structural similarity between SPOCK1 and SPARC, it is of great interest to investigate the role of SPOCK1 in cancer development and progression. In the present study, we identify the mechanism mediating the overexpression of SPOCK1 in HCC by showing that CHD1L binds the SPOCK1 promoter region. The clinical significance of SPOCK1 overexpression was assessed, and its oncogenic function was shown further in in vitro and in vivo studies. With a focus on its anti-apoptotic and modulatory cell-matrix interaction functions, the molecular mechanism linking an increase in SPOCK1 expression to cancer progression also was investigated. Materials and Methods Patients, Specimens, and Cell Lines Primary HCC samples (135 pairs) and their adjacent nontumor liver tissues were collected from patients who underwent hepatectomy at Sun Yat-Sen University Cancer Center (Guangzhou, China). None of these patients received preoperative chemotherapy or radiotherapy. The samples used in this study were approved by the Committees for Ethical Review of Research Involving Human Subjects at the Sun Yat-Sen University Cancer Center. The mean age of the 135 HCC patients included in this study was 48 years (range, y). The median follow-up period was 34.3 months (range, mo). The majority (109 of 135) of HCC patients had hepatitis B virus infection. Of the 26 HCC patients without hepatitis B virus infection, 1 patient was hepatitis C surface antigen positive, 3 patients had a history of cigarette smoking and/or habitual alcohol consumption, 10 patients had cirrhosis, and the remaining 14 patients displayed no evident etiologic factors such as inherited metabolic diseases. Normal liver tissues were obtained from donor liver that had not been used for transplantation (Queen Mary Hospital, Hong Kong). The HCC cell lines Huh7, QGY-7703, PLC-8024, BEL-7402, and QGY-7701 were obtained from the Institute of Virology at the Chinese Academy of Medical Sciences (Beijing, China). Laboratory Methods See the Supplementary Materials and Methods section for detailed experimental procedures. Detection of Expression Generally, quantitative real-time polymerase chain reaction (qrt-pcr) was used to detect messenger RNA (mrna) expression and Western blotting was used to detect protein expression. Protein expression on paraffin-embedded sections was detected by immunohistochemical staining (IHC). Interactions Between CHD1L and SPOCK1 The regulatory mechanism of the CHD1L SPOCK1 interaction was investigated by chromatin immunoprecipitation (ChIP) assay, electrophoretic mobility shift assay, and luciferase reporter assay. In Vitro Functional Assays In vitro tumorigenic abilities were assessed by XTT cell proliferation assay (Roche Diagnostics, Indianapolis, IN), foci formation assay, and colony formation assay in soft agar. In vitro metastatic abilities were assessed by invasion assay. In Vivo Functional Assays In vivo tumorigenic ability was investigated in a xenograft mouse model. In vivo metastatic ability was estimated by tail vein assay. Apoptosis Apoptotic effects were detected by a terminal deoxynucleotidyl transferase mediated deoxyuridine triphosphate nickend labeling (TUNEL) assay, mitochondrial membrane potential assay, and flow cytometry. Statistics SPSS (version 16.0; SPSS, Chicago, IL) was used for statistical analysis. The Pearson correlation coefficients were used to evaluate the positive correlation between CHD1L and SPOCK1 in the clinical samples. The mrna level of SPOCK1 in paired nontumor and tumor tissues was compared with a paired Student t test. An independent Student t test was used to compare the mean value of any 2 preselected groups. The Pearson 2 test was used to analyze the association of SPOCK1 expression with clinicopathologic parameters. Kaplan Meier plots and log-rank tests were used for survival analysis. Univariable and multivariable Cox proportional hazard regression models were used to analyze independent prognostic factors. The data are presented as the mean standard deviation of 3 independent experiments. A P value less than.05 was considered statistically significant. Results The Expression of SPOCK1 Is Correlated Positively With CHD1L Expression As reported in our previous study, CHD1L is an SNF2-like family member that can bind to a putative DNA-binding motif (C/A)C(T/A)T(T/A/G)T and transcriptionally regulate the expression of the corresponding

3 January 2013 SPOCK1 IN HCC DEVELOPMENT 181 gene. 9 To identify genes potentially regulated by CHD1L, a cdna microarray was used to compare the gene expression profiles between cells transfected with CHD1L or empty vector (unpublished data). One up-regulated gene, SPOCK1, was selected for further study. First, we tested the expression correlation between SPOCK1 and CHD1L in QGY7703 and Huh7 cells. As shown in previous studies, 6,7 the level of CHD1L expression in QGY7703 cells was the lowest among the HCC cell lines and similar to that in the immortalized normal liver cell line LO2. By contrast, Huh7 cells showed a higher level of CHD1L expression that was comparable with pathologic status. Therefore, we tested the effect of CHD1L overexpression in QGY7703 cells and down-regulation in Huh7 cells. SPOCK1 expression was up-regulated by CHD1L in QGY7703 cells after transient transfection with a CHD1L construct (Figure 1A). In Huh7 cells, SPOCK1 was down-regulated after CHD1L was silenced by RNA interference, suggesting that SPOCK1 expression was modulated in a CHD1L-dependent manner (Figure 1A). A significantly positive correlation (Pearson correlation coefficient, 0.619; P.001) between the expressions of CHD1L and SPOCK1 was detected by qrt-pcr in 135 pairs of HCC specimens (Figure 1B). Consistently, a correlation between the protein levels of SPOCK1 and CHD1L also was detected by Western blot analysis (Pearson correlation coefficient, 0.736; P.015; Figure 1C). CHD1L Binds to the 5= Upstream Region of SPOCK1 and Activates SPOCK1 Transcription To determine if CHD1L is able to bind directly to the promoter region of the SPOCK1 gene, the software MatInspector Professional (Genomatix, Munich, Germany) was used to search potential CHD1L binding sites in the SPOCK1 promoter. Five CHD1L potential binding sites ( 237 base pair [bp], 711 bp, 801 bp, 1189 bp, and 1512 bp) were identified within a 2-kb region upstream of the promoter region of SPOCK1 (Figure 1D). ChIP-PCR assays then were used to verify that CHD1L physically interacts with these predicted binding sites on SPOCK1. All 4 DNA fragments containing different CHD1L binding motifs could be detected in CHD1Limmunoprecipitated DNA fragments but not in IgGimmunoprecipitated controls (Supplementary Figure 1). Electrophoretic mobility shift assays were performed to further confirm the binding of the DNA fragments by the CHD1L protein. As shown in Figure 1E, CHD1L specifically bound DIG-labeled fragments A, B, C, and D. To determine if these interactions activated SPOCK1 transcription, a dual luciferase reporter assay was performed. The luciferase activities of pgl3-spock1-fe (FE, -1662/ 34) were increased significantly in cells co-transfected with pcdna3.1-chd1l compared with cells co-transfected with pcdna3.1 (Figure 1F). These results show that CHD1L can activate SPOCK1 transcription by binding to the 5= upstream region of SPOCK1. Clinical Significance of SPOCK1 in HCC To determine the prevalence and clinical significance of SPOCK1 in HCC, expression of SPOCK1 mrna in 8 normal livers and 135 pairs of HCCs (tumor and corresponding nontumor tissue) was compared by qrt-pcr. The expression of SPOCK1 gradually increased during HCC pathogenesis from the normal to adjacent nontumor liver tissues and to HCCs (Figure 2A). SPOCK1 overexpression (defined as a 2-fold increase in tumor tissue) was detected in 92 of 135 (68%) HCCs. The relative expression level of SPOCK1 was significantly higher in tumor tissues compared with their nontumor counterparts (P.0001, paired Student t test; Figure 2A). Western blotting showed that down-regulation of SPOCK1 protein was detected in 39 of 60 (65%) randomly selected HCCs. Statistical analysis revealed that HCC tissues expressed a significantly higher level of SPOCK1 protein than adjacent nontumor tissues (P.006, paired Student t test, Figure 2B). IHC staining was used to study the expression pattern of SPOCK1 in paraffin sections from normal liver and paired HCC tissues. The expression of SPOCK1 was significantly higher in tumor tissues compared with normal livers and their adjacent nontumor tissues (Figure 2C). Interestingly, in some cases, increased expression of SPOCK1 was observed in tumor cells at the edge of the tumor (Figure 2C, panels 4 and 5). A clinicopathologic association study in 135 HCCs found that overexpression of SPOCK1 was associated significantly with advanced clinical stage (Pearson 2 test, P.020; Table 1) and metastasis (Pearson 2 test, P.024; Table 1). HCC patients who developed metastasis after hepatectomy showed a significantly higher expression level of SPOCK1 than those without metastasis (P.0081, Student t test; Figure 2D), which implies that SPOCK1 may play a role in metastasis. More intriguingly, overexpression of SPOCK1 was correlated significantly with shorter overall survival (OS) (log rank, 6.492; P.011; Figure 2E) and shorter disease-free survival of patients (log rank, 4.277; P.039; Figure 2E). Multivariate Cox regression analysis further revealed that SPOCK1 was an independent prognostic marker for the OS time of HCC patients (hazard ratio, 3.713; 95% confidence interval, ; P.027; Supplementary Table 1). SPOCK1 Shows Strong Tumorigenic Ability To explore its role in tumorigenicity, SPOCK1 was cloned into an expression vector and stably transfected into the HCC cell lines QGY-7703 and PLC The expression of SPOCK1 in SPOCK1 transfectants was confirmed by Western blot analysis (Figure 3A). The tumorigenic ability of SPOCK1 was assessed by cell proliferation, foci formation, and soft agar assays. Compared with empty vector transfected cells, SPOCK1-transfected cells showed increased growth rates (Figure 3B), higher foci formation frequencies (Figure 3C), and greater colonyforming abilities in soft agar (Figure 3D).

4 182 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 Figure 1. CHD1L activates SPOCK1 transcription by binding to the 5= upstream region of SPOCK1. (A) QGY-7703 cells were transiently transfected with an ectopic CHD1L-expressing construct. Huh7 cells were treated with sichd1l or scramble small interfering RNA. The relative expression of CHD1L and SPOCK1 was detected by qrt-pcr; 18S ribosomal RNA was used as an internal control. The bars represent the mean standard deviation of 3 independent experiments. (B) The correlation between CHD1L and SPOCK1 expression was detected by qrt-pcr in 135 pairs of HCCs and matched nontumor tissues with linear regression lines and Pearson correlation significance (P.001, Pearson 2 test). (C) The protein levels of CHD1L and SPOCK1 were detected by Western blot analysis in 5 representative primary HCC tissues (T) and their paired nontumor (N) tissues. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a loading control. The Western blot data were quantified using ImageJ software (National Institutes of Health, Bethesda, MD). The expression levels of CHD1L relative to those of SPOCK1 after normalization were plotted in the line chart. (D) Schematic diagram representing the distribution of the CHD1L-binding loci (ellipse) in the regulatory region of SPOCK1. The 5 individual fragments represent different 5= upstream sequences of SPOCK1. (E) Electrophoretic mobility shift assay was used to detect the interaction between CHD1L and SPOCK1 double-stranded DNA probes. NE, nuclear extract. (F) Fragment E was subcloned into the pgl3-basic vector (pgl3-spock1) and co-transfected with pcdna3.1- CHD1L or empty vector for dual luciferase assays. The results were normalized to pgl3-basic activity, and the pgl3-control vector was used as a positive control.

5 January 2013 SPOCK1 IN HCC DEVELOPMENT 183 Figure 2. Clinical significance of SPOCK1 in human HCC. (A) Scatterplots of the relative expression of SPOCK1 in normal liver, nontumor tissues, and their matched HCC counterparts. Black lines, mean standard deviation. (B) SPOCK1 protein expression levels in 8 representative primary HCC tissues (T) and their paired nontumor (N) tissues was evaluated via Western blot analysis. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a loading control. SPOCK1 bands were quantified with ImageJ software (National Institutes of Health) and are shown in the bar chart after normalization. (C) Representative image of IHC staining with an anti-spock1 antibody (from left to right: normal liver; nontumor tissue; paired tumor tissue; HCC tissue, 200 magnification; HCC tissue, 400 magnification). (D) Scatterplots of the relative expression of SPOCK1 in patients with metastasis or without metastasis. Black lines, mean standard deviation. (E) Kaplan Meier overall survival curve (left panel) and disease-free survival curve (right panel) of HCC patients correlated with SPOCK1 expression. SPOCK1( ), patients with SPOCK1 overexpression; SPOCK1( ), patients without SPOCK1 overexpression. To further investigate the in vivo tumorigenic ability of SPOCK1, empty vector and SPOCK1-transfected cells were injected subcutaneously into the left and right dorsal flank of nude mice (n 10), respectively. Tumors induced by SPOCK transfectants (tumor formation in 10 of 10 mice) showed significantly shorter latency and larger mean tumor volume than tumors induced by Vec-7703 cells (tumor formation in 1 of 10 mice) (Figure 3E). A similar result was observed when SPOCK1-transfected PLC-8024 cells were used in the xenograft mouse experiment. Compared with the control Vec-8024 cells (tumor formation in 2 of 5 mice), SPOCK1-transfected cells (tumor formation in 4 of 5 mice) showed a significantly larger mean tumor volume (Figure 3E). Short Hairpin RNA Mediated SPOCK1 Silencing Abolishes the SPOCK1 Tumorigenic Effect We next examined whether SPOCK1 is required for the tumorigenic phenotypes of HCC cells by silencing SPOCK1 expression with short hairpin RNA

6 184 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 Table 1. Association of SPOCK1 Overexpression With Clinicopathologic Features in 135 Primary HCCs Features Total Overexpression SPOCK1 expression Normal expression Sex Male (63.4%) 41 (36.6%) Female (73.9%) 6 (26.1%).335 Age, y (64.9%) 39 (35.1%) (66.7%) 8 (33.3%).867 Hepatitis B surface antigen a Negative (73.1%) 7 (26.9%) Positive (63.4%) 37 (36.6%).353 Serum -fetoprotein level, ng/ml (59.4%) 28 (40.6%) (72.4%) 16 (25.7%).125 Tumor size, cm (64.1%) 14 (35.9%) (66.7%) 30 (33.3%).778 Cirrhosis Absent (60.5%) 15 (39.5%) Present (67.8%) 28 (32.2%).430 Vascular invasion Absent (64%) 41 (36%) Present (80%) 3 (20%).220 Differentiation Well/moderate (65.7%) 24 (34.3%) Poor (62.5%) 18 (37.5%).720 Tumor nodule Single (62%) 38 (38%) Multiple (82.1%) 5 (17.9%).046 Tumor stage (AJCC) I (59.6%) 36 (40.5%) II/III (81.1%) 7 (18.9%).020 Metastasis Absent (56.1%) 29 (43.9%) Present (74.6%) 17 (25.4%).024 NOTE. Statistical significance (P.05) is shown in bold. a Partial data are not available, and the statistic was based on available data. P value (shrna) against SPOCK1. The introduction of shrna into the HCC cell lines BEL-7402 and QSG-7701 dramatically decreased the expression level of SPOCK1 relative to control cells expressing scrambled shrna (Figure 4A). Functional assays revealed that SPOCK1 depletion could reverse the tumorigenic phenotype in shspock1 cells by inhibiting the cell growth rate (Figure 4B), foci formation efficiencies (Figure 4C), and colony formation in soft agar (Figure 4D) compared with control cells. In in vivo xenograft experiments, solid tumors were visible in 4 of 5 mice injected with shctl-7402 cells, whereas only 1 of 5 of mice injected with shspock cells formed tumors within 4 weeks (Figure 4E). Collectively, these data indicate that SPOCK1 possesses strong tumorigenicity both in vitro and in vivo. SPOCK1 Inhibits Apoptosis in HCC Cells To explore the molecular mechanism involved in SPOCK1-enhanced tumor cell survival, the effect of SPOCK1 on apoptosis was investigated. TUNEL assays revealed that SPOCK1 inhibited apoptosis in the presence of staurosporine (STS). The apoptotic index of SPOCK1 transfectants was significantly lower than that of vector transfectants after a 6-hour exposure to STS (Figure 5A). We next evaluated whether knockdown of SPOCK1 expression could reverse this phenotype. Six hours after STS stimulation, the apoptotic indexes of knockdown control cells (shctl-7402) and SPOCK1- silenced cells (shspock c1 and shspock c2) were 6% and 35%, respectively (Figure 5A). These results indicate that silencing SPOCK1 not only restored the cellular response to apoptotic stimulation but also rendered SPOCK1-defective HCC cells more vulnerable to STS-induced apoptosis compared with control cells. SPOCK1 Exerts an Anti-apoptotic Function Through the Akt Cyt C Caspase-9 Caspase-3 Pathway Tumor cells resist cell death through either the disruption of apoptotic processes or the activation of survival signals. In general, survival signals are mediated by the PI3K-Akt pathway. 16 Deregulation of Akt phos-

7 January 2013 SPOCK1 IN HCC DEVELOPMENT 185 Figure 3. SPOCK1 has strong tumorigenic ability. (A) Ectopic expression of SPOCK1 was detected in SPOCK1-transfected cells by Western blot. -actin was used as a loading control. (B) The cell growth rates were determined with an XTT proliferation assay (*P.05; **P.001; independent Student t test). Representative images of (C) foci formation and (D) colony formation in soft agar induced by Vec-7703, SPOCK1-7703, Vec-8024, and SPOCK cells. The numbers of foci and colonies were calculated and are depicted in the bar chart. The values indicate the mean standard deviation of 3 independent experiments (*P.05; **P.001; independent Student t test). (E) Representative examples of tumors (indicated by arrows) formed in nude mice injected with the indicated cells. Tumor growth curves are summarized in the line chart. The average tumor volume was expressed as the mean standard deviation of 5 mice. (*P.05; **P.001; independent Student t test). phorylation represents an important anti-apoptotic mechanism in various cancers. Activated Akt can phosphorylate a wide variety of substrate proteins, including BAD, a pro-apoptotic member of the Bcl-2 protein family whose function is suppressed by phosphorylation. BAD inactivation maintains the integrity of the mitochondrial membrane, which in turn blocks cytochrome c release and the subsequent activation of caspase-9, caspase-3, and poly(adenosine diphosphate-ribose) polymerase (PARP). 17 Therefore, we examined whether SPOCK1 inhibits apoptosis via Akt phosphorylation. Phosphorylated Akt was increased in SPOCK1-transfected cells compared with control cells and persisted for a longer period of time upon STS stimulation (Figure 5B). Activated Akt subsequently regulates Bcl-2 family proteins, which affect mitochondrial membrane permeability. As a result, SPOCK1-transfected cells maintained a high inner mitochondrial transmembrane potential ( m ) (shown in red fluorescent), whereas most of the Vec-7703 control cells underwent a proapoptotic mitochondrial permeability transition (Figure 5C). SPOCK1-induced Akt phos-

8 186 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 Figure 4. shrna-mediated SPOCK1 silencing abolishes the tumorigenic effect of SPOCK1. (A) BEL-7402 and QGY-7701 cells were treated with scrambled shrna or shrna against SPOCK1. Western blot analysis and qrt-pcr were performed to detect SPOCK1 expression; -actin and 18S were used as loading controls, respectively. (B) The cell growth rates were determined with an XTT proliferation assay (*P.05; **P.001; independent Student t test). Representative images of (C) foci formation and (D) colony formation in soft agar induced by shctl-7402, shspock1-7402, shctl-7701, and shspock cells. The numbers of foci and colonies were calculated and are depicted in the bar chart. The values indicate the mean standard deviation of 3 independent experiments (*P.05; **P.001; independent Student t test). (E) Representative examples of tumors (indicated by arrows) formed in nude mice injected with the indicated cells. The tumor incidence rate during the 4-week observation period is shown in the table section of panel E. phorylation protected the mitochondrial membrane from STS-induced collapse, thereby blocking cyt c release. As a result, the subsequent cleavage of pro-caspase-9, procaspase-3, and PARP all were suppressed in SPOCK1-overexpressing clones (Figure 5B). The anti-apoptotic phenotype and Akt phosphorylation were reversed when SPOCK1 was silenced in shspock cells. Reduced phosphorylated Akt in SPOCK1 knockdown cells led to m collapse (Figure 5B, green), whereas most control Con-7402 cells maintained their m (Figure 5C, red fluorescence). Concomitantly, cleaved forms of pro-caspase-9, pro-caspase-3, and PARP increased more rapidly in SPOCK1 knockdown cells than in control cells (Figure 5B). An Akt1 Inhibitor Abolishes the Preferential Survival of SPOCK1-Overexpressing HCC Cells via the Akt and BAD Pathways To further confirm the importance of the Akt pathway in the increased survival of SPOCK1-overexpressing HCC cells, we assessed the ability of an Akt1 inhibitor to abolish SPOCK1-induced apoptotic resistance. The Akt1 inhibitor reduced Akt activity and subsequent BAD phosphorylation in a dose-dependent manner (Figure 5D). Cells were pretreated with 80 mol/l Akt1 inhibitor for 24 hours before the addition of the apoptosis inducer STS. After STS treatment, the amount of apoptosis was assessed quantitatively by flow cytometry after staining with Annexin V fluorescein isothiocyanate and pro-

9 January 2013 SPOCK1 IN HCC DEVELOPMENT 187

10 188 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 pidium iodide. Similar to the TUNEL results, the flow cytometry histogram (Figure 5E) showed that SPOCK1 transfectants were resistant to STS in the absence of the Akt1 inhibitor. Interestingly, pre-incubation with the Akt1 inhibitor completely inhibited the preferential survival effect induced by SPOCK1 overexpression in cells (Figure 5E and F). The reversal of SPOCK1-mediated apoptotic resistance by the Akt inhibitor provides additional evidence supporting the role of this pathway in the increased survival of SPOCK1-overexpressing HCC cells. SPOCK1 Promotes Tumor Invasion and Metastasis To investigate the effects of SPOCK1 overexpression on metastasis, an in vitro Matrigel invasion assay (Becton Dickinson, Bedford, MA) and an in vivo experimental metastasis assay were performed. The Matrigel invasion assay showed that the invasive capability of SPOCK cells was greater than that of Vec-7703 cells (Figure 6A, left). By contrast, silencing SPOCK1 expression by shrna in BEL-7402 cells abolished the invasiveness of the shspock cells (Figure 6A, right). These results indicate that SPOCK1 increases cell invasion, which we further validated in vivo. The experimental metastasis assay was performed by injecting HCC cells intravenously into severe combined immunodeficient Beige mice to mimic cell metastasis through circulation. Nine weeks after injection, the metastatic modules that formed on the surface of the lungs and liver were counted. The number of metastatic nodules formed on the surface of the liver ( and , respectively) was significantly higher in mice injected with SPOCK cells than in mice injected with Vec-7703 cells ( ; P.05, independent Student t test; Figure 6B). Metastatic lesions in the lungs were detected by histologic study (Figure 6C). SPOCK1 IHC staining further confirmed that the lesions were caused by extravasation and subsequent tumor growth of SPOCK1-transfected (positive staining) HCC cells into the liver (Figure 6C). SPOCK1 Regulates Extracellular Matrix Remodeling Because cell invasion involves the degradation of basement membrane extracellular matrix proteins and matrix metallopeptidase (MMP)-2 and MMP-9 are the primary MMPs responsible for this process, 18 we determined if SPOCK cells secreted a higher level of MMP-2 or MMP-9 than Vec-7703 cells. Although no significant difference in MMP-2 expression was observed, MMP-9 mrna expression was higher in SPOCK-7703 cells than in Vec-7703 cells (Figure 6D). Moreover, Gelatin zymography assay showed that MMP-9 activity in SPOCK conditioned medium was dramatically higher than that in Vec-7703 conditioned medium (Figure 6E). To further confirm the importance of the increase of MMP-9 in the increased invasion of SPOCK cells, we assessed the ability of an MMP-9 inhibitor to prevent the invasion of SPOCK cells through the Matrigel Matrix. Treatment with the MMP-9 inhibitor significantly inhibited the invasion ability of SPOCK cells in a dose-dependent manner (Figure 6F). Discussion Amplification of 1q21 is an early event and is detected in more than 60% of human HCCs. 4,5 CHD1L, a putative oncogene isolated from this frequently amplified region, has been shown to exert profound effects on the initiation of HCC pathogenesis As a member of the SNF2-like family of transcription factors, CHD1L affects a broad spectrum of cellular processes. In the present study, a cdna microarray was performed to unravel the intricate CHD1L-regulated network and identified a novel oncogene, SPOCK1. Although SPOCK1 has been reported to be overexpressed in several other carcinomas, 13,17 little is known about the underlying mechanism. This study showed the mechanism involved in SPOCK1 overexpression: CHD1L binds to the 5= upstream region of SPOCK1 and subsequently activates SPOCK1 transcription. Because the amplification of 1q is one of the most recurrent DNA copy number changes in ovarian cancer, prostate cancer, breast cancer, small-cell lung cancer, and non small-cell lung cancer, 19,20 this 1q amplification CHD1L overexpression SPOCK1 up-regulation axis also may be relevant to these malignances. In vitro and in vivo assays both showed that SPOCK1 had strong tumorigenic function. Additional experiments revealed that SPOCK1-enhanced tumor cell sur- 4 Figure 5. SPOCK1 exerts anti-apoptotic function via the Akt cyt c caspase-9 caspase-3 pathway. (A) After treatment with STS, apoptosis was determined by TUNEL assay. Apoptotic cells were stained with fluorescein isothiocyanate labeled DNA strand breaks (green signal), and nuclei were counterstained with 4=,6-diamidino-2-phenylindole (blue signal). The apoptotic index, defined as the percentage of apoptotic cells, was calculated and is summarized in the bar chart. The values represent the mean standard deviation of 3 independent experiments (*P.05; **P.001; independent Student t test). (B) After treatment with STS for the indicated time, the levels of phosphorylated Akt (P-Akt), total Akt, caspase-9, caspase-3, and PARP were detected in Vec-7703, SPOCK1-7703, shctl-7402, and shspock cells by Western blot analysis. -actin was used as a loading control. (C) Representative images of JC-1 dye staining. JC-1 dye aggregates at high m mitochondria (red) or forms monomers at low m mitochondria (green). (D) SPOCK cells were incubated for 24 hours with different concentrations of Akt1 inhibitor varying from 0 to 80 mol/l. The levels of phosphorylated Akt (ser473) and BAD (ser136) were detected by Western blot analysis. -actin was used as a loading control. (E) Vec-7703 and SPOCK cells were treated with STS (1 mol/l) alone or in combination with Akt1 inhibitor (80 mol/l). Apoptosis in Vec-7703 and SPOCK cells was compared at the indicated times by flow cytometry. Cells stained with Annexin-V fluorescein isothiocyanate were counted as apoptotic, and the apoptotic index was defined as the percentage of apoptotic cells. (F) Six hours after STS treatment, the apoptotic index of Vec-7703 and SPOCK cells was calculated and is summarized in the bar chart. The values represent the mean standard deviation of 3 independent experiments (*P.05; independent Student t test).

11 January 2013 SPOCK1 IN HCC DEVELOPMENT 189 Figure 6. SPOCK1 promotes tumor invasion and metastasis. (A) A Matrigel invasion assay was performed to study the invasion ability of Vec-7703, SPOCK1-7703, shctl-7402, and shspock cells. The number of invaded cells was calculated and is depicted in the bar chart. The values indicate the mean standard deviation of 3 independent experiments (*P.05; **P.001; independent Student t test). (B) An in vivo experimental metastasis assay was performed to evaluate the effect of Vec-7703 and SPOCK cells on tumor metastasis. Representative images of livers derived from severe combined immunodeficient mice after tail vein injection of Vec-7703 or SPOCK cells. The right panel summarizes the numbers of metastatic nodules. The values for the individual mice are shown above the bars; the values by group are also denoted (*P.05, **P.001; independent Student t test). (C) Representative images of H&E staining of lung tissues from severe combined immunodeficient mice injected with Vec-7703 or SPOCK cells (200 magnification, upper panel). Representative images of IHC staining of liver tissues from severe combined immunodeficient mice (200 magnification, lower panel). (D) The relative expression of MMP-9 and SPOCK1 was detected in empty vector and SPOCK1-transfected cells by qrt-pcr; 18S ribosomal RNA was used as an internal control. (E) Gelatin zymography was performed to quantify MMP-9 activity in Vec-7703 and SPOCK cells. The gelatin digested by MMP-9 was visualized as a clear band on a dark background. MMP-9 activity was quantified by ImageJ software (National Institutes of Health) and is shown in the bar chart. The value is expressed as the average of 2 independent experiments. (F) Representative images of cells invading through the Matrigel in the presence of MMP-9 inhibitor at concentrations of 0 20 mol/l. The number of invaded cells was calculated and is depicted in the bar chart. The values indicate the mean standard deviation of 3 independent experiments (*P.05, **P.001; independent Student t test).

12 190 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 vival may be attributable to its anti-apoptotic ability. The data presented here show that SPOCK1 contributes to the anti-apoptotic effect through the activation of the Akt pathway, which subsequently inhibits the cyt c caspase-9 caspase-3 pathway. Inhibition of apoptosis is one of the major mechanisms in cancer development and ultimately leads to the expansion of neoplastic cells with deregulated proliferation and accumulation of genetic instability and mutations. 21 Therefore, SPOCK1 overexpression may be an early and critical event that propels uncontrolled tumor cell expansion during HCC initiation. Moreover, because chemotherapeutic drugs and irradiation work primarily by inducing apoptosis, 22 defects in apoptosis are an important clinical problem in chemotherapy. HCC is one of the most chemoresistant cancers, with a reported response rate varying from 0% to 20%. 23 The contribution of SPOCK1 to the activation of Akt in HCC cells may provide further rationale for the inclusion of SPOCK1 as a target modulator of chemosensitivity. In addition to its tumorigenic roles, the present study also showed that SPOCK1 induces tumor invasion and metastasis. Studies have revealed that the invasive edge of the tumor is the most active area in local invasion. 24 Microscopic examination of tissue samples from cancer patients and animals points to increased expression of VEGF and MMPs at the leading edge of the primary tumor. 24,25 In the present study, SPOCK1 expression was increased at the edges of HCC tumors. This observation not only reflects the association of increased SPOCK1 expression with the most motile and polarized tumor cells but also implicates SPOCK1 in the induction of metastasis. As shown in our in vitro studies, SPOCK1 expression increased MMP-9 expression and activity. MMP-mediated extracellular matrix and basement membrane degradation is an important proteolytic event in metastasis, particularly during tumor cell invasion into surrounding tissues, vascular infiltration, and extravasation. 24,26 The increased SPOCK1 expression at the edges of tumors may induce extensive extracellular matrix remodeling, allowing individual tumor cells or cohorts of tumor cells to undergo directional migration and leave the edge of the tumor mass. This finding corroborates a published report showing that CHD1L is overexpressed at the edges of tumors and in cells invading surrounding tissue and blood vessels. 9 As a newly identified downstream target of CHD1L, increased SPOCK1 expression may be induced by CHD1L at the edges of tumors. Interestingly, the versatile protein Akt also has been reported to play a role in cancer cell metastasis via MMP-9 modulation. 27 It remains to be investigated whether the invasive aspects of SPOCK1 are related to Akt. In the present study, we showed that SPOCK1 could inhibit apoptosis and promote cancer invasion. Because SPOCK1 belongs to the Ca (2 ) -binding proteoglycan family, some of these effects may be mediated by the glycan segment of SPOCK1. Increasing evidence has shown that glycan specifically can interact with growth factors, chemokines, and matrix architecture. 28 Cancer cells may usurp these properties to gain a survival advantage and invade throughout the organism. For example, the glycan segment of perlecan effectively can protect fibroblast growth factor 2 from proteolytic degradation and potentiate its angiogenic role. 29 In addition to the steady-state properties of the glycan, changes in the glycan segment, such as glycosylation changes and depolymerization, also affect cancer development. Heparanase-induced depolymerization can release fibroblast growth factor 2 from perlecan to facilitate vascular sprouting during angiogenesis. 30 Some of these characteristics of perlecan could be shared by other pericellular proteoglycans such as agrin, collagen type XVIII, and SPOCK1. However, whether SPOCK1 performs its functions by working alone or in concert with other ligands or elements of the matrix architecture such as fibronectin is unknown. If SPOCK1 works with other partner molecules, it will be important to identify the portion of the proteoglycan that mediates the interaction. Further studies of the glycan segment of SPOCK1 will be necessary to expand our knowledge of SPOCK1 and establish a basis for developing pharmaceutical agents that target this molecule. A clinical association study found that overexpression of SPOCK1 was associated significantly with advanced tumor stage and shorter OS time of HCC patients. Cox proportional hazard regression analysis further identified SPOCK1 as an independent marker for poor prognosis. Because SPOCK1 is a secreted protein that is detected at very low expression levels among normal vital tissue specimens, SPOCK1 overexpression in HCC may serve as a biomarker for early detection and precise prognoses. A better understanding of the oncogenic mechanisms of SPOCK1 during HCC initiation and progression may have implications for future patient treatment. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at and at dx.doi.org/ /j.gastro References 1. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN Int J Cancer 2010;127: Parkin DM, Bray F, Ferlay J, et al. Global cancer statistics, CA Cancer J Clin 2005;55: Thorgeirsson SS, Grisham JW. Molecular pathogenesis of human hepatocellular carcinoma. Nat Genet 2002;31: Kusano N, Shiraishi K, Kubo K, et al. Genetic aberrations detected by comparative genomic hybridization in hepatocellular carcinomas: their relationship to clinicopathological features. Hepatology 1999;29: Guan XY, Fang Y, Sham JS, et al. Recurrent chromosome alterations in hepatocellular carcinoma detected by comparative genomic hybridization. Genes Chromosomes Cancer 2000;29: Ma NF, Hu L, Fung JM, et al. Isolation and characterization of a novel oncogene, amplified in liver cancer 1, within a commonly amplified region at 1q21 in hepatocellular carcinoma. Hepatology 2008;47:

13 January 2013 SPOCK1 IN HCC DEVELOPMENT Chen L, Hu L, Chan TH, et al. Chromodomain helicase/adenosine triphosphatase DNA binding protein 1-like (CHD1l) gene suppresses the nucleus-to-mitochondria translocation of nur77 to sustain hepatocellular carcinoma cell survival. Hepatology 2009; 50: Ahel D, Horejsi Z, Wiechens N, et al. Poly(ADP-ribose)-dependent regulation of DNA repair by the chromatin remodeling enzyme ALC1. Science 2009;325: Chen L, Chan TH, Yuan YF, et al. CHD1L promotes hepatocellular carcinoma progression and metastasis in mice and is associated with these processes in human patients. J Clin Invest 2010;120: Chen M, Huang JD, Hu L, et al. Transgenic CHD1L expression in mouse induces spontaneous tumors. PLoS One 2009;4:e Bradshaw AD, Sage EH. SPARC, a matricellular protein that functions in cellular differentiation and tissue response to injury. J Clin Invest 2001;107: Tai IT, Tang MJ. SPARC in cancer biology: its role in cancer progression and potential for therapy. Drug Resist Updat 2008; 11: Leja J, Essaghir A, Essand M, et al. Novel markers for enterochromaffin cells and gastrointestinal neuroendocrine carcinomas. Mod Pathol 2009;22: Wlazlinski A, Engers R, Hoffmann MJ, et al. Downregulation of several fibulin genes in prostate cancer. Prostate 2007;67: Colin C, Baeza N, Bartoli C, et al. Identification of genes differentially expressed in glioblastoma versus pilocytic astrocytoma using Suppression Subtractive Hybridization. Oncogene 2006;25: Datta SR, Brunet A, Greenberg ME. Cellular survival: a play in three Akts. Genes Dev 1999;13: Igney FH, Krammer PH. Death and anti-death: tumour resistance to apoptosis. Nat Rev Cancer 2002;2: Malemud CJ. Matrix metalloproteinases (MMPs) in health and disease: an overview. Front Biosci 2006;11: Knuutila S, Bjorkqvist AM, Autio K, et al. DNA copy number amplifications in human neoplasms: review of comparative genomic hybridization studies. Am J Pathol 1998;152: Saramaki OR, Porkka KP, Vessella RL, et al. Genetic aberrations in prostate cancer by microarray analysis. Int J Cancer 2006;119: Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000; 100: Herr I, Debatin KM. Cellular stress response and apoptosis in cancer therapy. Blood 2001;98: Nerenstone SR, Ihde DC, Friedman MA. Clinical trials in primary hepatocellular carcinoma: current status and future directions. Cancer Treat Rev 1988;15: Deryugina EI, Quigley JP. Matrix metalloproteinases and tumor metastasis. Cancer Metastasis Rev 2006;25: Onogawa S, Kitadai Y, Tanaka S, et al. Expression of VEGF-C and VEGF-D at the invasive edge correlates with lymph node metastasis and prognosis of patients with colorectal carcinoma. Cancer Sci 2004;95: Fidler IJ. The pathogenesis of cancer metastasis: the seed and soil hypothesis revisited. Nat Rev Cancer 2003;3: Kim D, Kim S, Koh H, et al. Akt/PKB promotes cancer cell invasion via increased motility and metalloproteinase production. FASEB J 2001;15: Blackhall FH, Merry CL, Davies EJ, et al. Heparan sulfate proteoglycans and cancer. Br J Cancer 2001;85: Jiang X, Couchman JR. Perlecan and tumor angiogenesis. J Histochem Cytochem 2003;51: Vlodavsky I, Friedmann Y. Molecular properties and involvement of heparanase in cancer metastasis and angiogenesis. J Clin Invest 2001;108: Received May 4, Accepted September 18, Reprint requests Address requests for reprints to: Xin Yuan Guan, PhD, Department of Clinical Oncology, The University of Hong Kong, 21 Sassoon Road, Hong Kong, China. xyguan@hkucc.hku.hk; fax: Yun-Fei Yuan, MD, State Key Laboratory of Oncology in Southern China, Sun Yat-sen University Cancer Center, 651 Dongfeng East Road, Guangzhou , China. yuanyf@mail.sysu.edu.cn; fax: Conflicts of interest The authors disclose no conflicts. Funding Supported by grants from the National Basic Research Program of China (2012CB967001), the Hong Kong Research Grant Council Collaborative Research Funds (HKU5/CRF/08, HKU7/CRG09, and HKBU5/CRG/10), and the Theme-based Research Scheme fund (T12-403/11).

14 191.e1 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 Supplementary Materials and Methods Reagent AKT1 inhibitor (1L6-hydroxymethyl-chiro-inositol-2-(R)-2-Omethyl-3-O-octadecyl-sn-glycerocarbonate) and MMP-9 inhibitor I were purchased from Calbiochem (San Diego, CA). qrt-pcr Total RNA was purified using TRIzol (Invitrogen, Carlsbad, CA) and cdna was synthesized using the Transcriptor High Fidelity cdna Synthesis Kit (Roche) according to the manufacturer s instructions. qrt-pcr was performed using the SYBR Green PCR Kit (Applied Biosystems, Foster City, CA) and an ABI PRISM 7900 Sequence Detector (Applied Biosystems). The amplification plots were analyzed using SDS software (Applied Biosystems) and the threshold cycle (Ct) was measured during the exponential amplification phase. The relative expression level of the target gene is given by 2 CT ( C T C target T C 18S T ) and normalized to the relative expression detected in the corresponding control cells, which was defined as 1.0. For correlation study, the expression level (defined as fold change) of CHD1L and SPOCK1 is given by 2 CT ( C T C tumor T C nontumor T ). All reactions were performed in duplicate. Primer sequences are listed in Supplementary Table 2. RNA Interference A small interfering RNA (20 nmol/l) against CHD1L (Ambion, Austin, TX) was transfected into cells in 6-well plates using Lipofectamine 2000 (Invitrogen) according to the manufacturer s instructions. Forty-eight hours after transfection, gene silencing effects were measured by qpcr analysis. Chromatin Immunoprecipitation By using an EZ-Magna ChIP G kit (Upstate Biotechnology, Lake Placid, NY), GFP-tagged-CHD1L-binding DNA fragments were pulled down with the anti-gfp antibody B-2. Mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA) was used as a negative control. The predicted CHD1L binding sites were confirmed by ChIP- PCR. The primers used to amplify the precipitated DNA fragments are listed in Supplementary Table 3. Electrophoretic Mobility Shift Assay Fragments containing the SPOCK1 promoter region were amplified by PCR with digoxin (DIG)-labeled deoxyuridine triphosphate (Roche) in addition to deoxynucleoside triphosphates. Nuclear extracts were prepared with the NucBuster Protein Extraction kit (Novagen, Madison, WI). Electrophoretic mobility shift assay was performed with 10 g of nuclear extracts and 50 ng of DIG-labeled or unlabeled probes, as previously described. 9 Dual-Luciferase Reporter Assay A 1696-bp fragment from the 5= upstream promoter region of SPOCK1 ( bp) that contains 5 predicted CHD1L binding sites was cloned into the pgl3 basic vector (Promega, Madison, WI). This SPOCK1 promoter reporter construct (pgl3-spock1) then was transfected into cells with the CHD1L expression plasmid or empty pcdna3.1 vector (Invitrogen). Firefly and Renilla luciferase activities were determined by the Dual- Luciferase reporter assay kit (Promega) 48 hours after transfection. The pgl3-control vector (Promega) was used as a positive control. Plasmid Constructs and Transfection To evaluate the tumorigenic ability, full-length SPOCK1 cdna was cloned into the pcdna3.1 expression vector (Invitrogen) and transfected into QGY-7703 cells and PLC-8024 cells independently. Stable SPOCK1- expressing clones were selected for 2 weeks using Geneticin (Roche), and the expression level of SPOCK1 was determined by q-pcr and Western blot analysis. Empty vector transfected cells (Vec-7703 or Vec-8024) were used as control. Establishment of SPOCK1 Stably Knockdown Cells The scrambled shrna plasmid pretrosuper (prs) and the SPOCK1-specific shrna expression prs vectors (prs-shspock1) were provided by OriGene Technologies (Rockville, MD). Two constructs against SPOCK1 (shspock1) were used: shspock1-1, GGATGTTCAA- CAAGTTGGACATGAACTAT; shspock1-2, TACAAAGC- CACACAGTGCCACGGCAGCAC. The prs-shspock1 or the scrambled shrna construct was transfected into BEL-7402 and QSG-7701 cells using Lipofectamine 2000 (Invitrogen). Stable SPOCK1 knockdown clones were selected for 2 weeks with puromycin (Origene, Rockville, MD). In Vitro Tumorigenic Assays XTT cell proliferation assay (Roche Diagnostics) was performed according to the manufacturer s instructions. Anchorage-dependent growth was assessed by foci formation assay. Briefly, cells were seeded in a 6-well plate. Colonies ( 50 cells/colony) were stained by Giemsa (Invitrogen) and counted at day 7. Anchorageindependent growth was assessed by colony formation ability in soft agar. Briefly, cells were suspended in soft agar mixture (Dulbecco s modified Eagle medium, 10% fetal bovine serum, and 0.35% SeaPlaque agarose) and subsequently were overlaid on the solidified 0.5% agar base. After 3 weeks, colonies ( 10 cells) were counted under the microscope in 10 fields per well. Triplicate independent experiments were performed.

15 January 2013 SPOCK1 IN HCC DEVELOPMENT 191.e2 In Vivo Tumorigenicity Assay The in vivo tumorigenic ability of SPOCK1 was investigated in a xenograft mouse model. Briefly, control cells (Vec-7703, Vec-8024, and shctl-7402) were injected subcutaneously into the left dorsal flank of BALB/ cann-nu (nude) mice, and SPOCK1-expressing cells (SPOCK1-7703, SPOCK1-8024) or knockdown cells (shspock1-7402) were injected into the right dorsal flank of the same animal. Over a 4-week period after injection, tumor formation was monitored by measuring the tumor volume weekly. Tumor volume was calculated as 0.5 l w 2, where l is the length (long axis), and w is the width (short axis) of the tumor. Terminal Deoxynucleotidyl Transferasemediated dutp Nick-end Labeling Cells were treated with the universal apoptosis inducer STS (1 mol/l; Sigma, St. Louis, MO). Apoptotic cells at different time points (3 and 6 hours after exposure to STS) were detected using the In Situ Cell Death Detection Kit according to the manufacturer s protocol (Roche). Mitochondrial Membrane Potential Assay The mitochondrial membrane potential ( m ) was detected with a MitoPT JC-1 detection kit (Immunochemistry Technologies, Bloomington, MN) according to the manufacturer s protocol. Briefly, cells were seeded onto coverslips and cultured to 80% confluence before STS treatment. At different time points after STS stimulation (45 minutes to 2 hours), the cells were washed twice with phosphate-buffered saline and then incubated with the m -sensitive dye JC-1 at 37 C for 15 minutes. Flow Cytometry STS (1 mol/l) was added for different times (0, 3, and 6 h) to trigger apoptosis. Cells then were collected and stained with fluorescein isothiocyanate conjugated Annexin-V and propidium iodide as provided by the Annexin- V Fluos Staining Kit (Roche) according to the manufacturer s instructions. Analysis was determined by flow cytometry on a FACScan (Becton-Dickinson, Bedford, MA). Invasion Assays Invasion assays were performed with BioCoat Matrigel Invasion Chambers (8- m pore size) according to the manufacturer s instructions (Becton-Dickinson). The number of cells that invaded through the Matrigel was counted in 10 fields under a 20 objective lens. Triplicate independent experiments were performed. In Vivo Metastasis Assay Four- to 5-week-old severe combined immunodeficient Beige mice were used, and all animal procedures were approved by Committee on the Use of Live Animals in Teaching and Research (CULATR). Briefly, cells (2 experimental groups including Vec-7703 and SPOCK1-7703) were injected intravenously through the tail vein into each mouse. All of the mice were euthanized 9 weeks after injection. Tumor nodules that formed on the lung and liver surfaces were macroscopically detected and counted. After the experiment, the livers and lungs excised from the mice were embedded in paraffin, and sections (5 m) of the tissues were stained with H&E to visualize the structure. Gelatin Zymography Conditioned medium was collected and concentrated in an Amicon Ultra-4 centrifugal filter device (10kD) (Millipore, Bedford, MA) according to the manufacturer s instructions. MMP-9 protein expression in conditioned medium was measured by Novex Zymogram Gels (Invitrogen) according to the manufacturer s protocol. MMP-9 activity was visualized as negative staining with Coomassie Brilliant Blue (Sigma-Aldrich, St. Louis, MO) at a size of 92 kilodaltons. This assay was performed in 3 independent experiments. Antibodies and Western Blotting Mouse anti-chd1l, -actin, and rabbit anti- SPOCK1 antibodies were purchased from Abcam (Cambridge, MA). The rabbit anti-akt, phospho-akt (Ser473), BAD, caspase-3, caspase-9, and PARP were all purchased from Cell Signaling Technology (Danvers, MA). The mouse anti-gfp (B-2), phosphor-bad (Ser136), and IgG were purchased from Santa Cruz Biotechnology. Briefly, quantified protein lysates were resolved on sodium dodecyl sulfate polyacrylamide gel electrophoresis gel, transferred onto a polyvinylidene difluoride membrane (Millipore, Billerica, MA), and immunoblotted with primary antibody. After incubation with a secondary antibody, blots were visualized by enhanced chemiluminescence (GE Healthcare, Buckinghamshire, UK). IHC Staining Immunohistochemical assays were performed on paraffin-embedded sections. Briefly, sections were deparaffinized and rehydrated. Antigen retrieval was performed using a steamer for 15 minutes in 10 mmol/l citrate buffer (ph 6.0). The endogenous peroxidase activity was blocked with 3% hydrogen peroxide (H 2 O 2 ) for 10 minutes; then normal goat serum was applied for 10 minutes to prevent nonspecific binding. Sections were incubated successively with primary antibody against SPOCK1 (1:500 dilution; Novus Biologicals, Littleton, CO) at 4 C overnight. Then slides were reacted with biotinylated goat anti-rabbit IgG (1:100 dilution; Santa Cruz Biotechnology) for 30 minutes and streptavidinperoxidase conjugated for 30 minutes at room temperature. Finally, the 3, 5-diaminobenzidine Substrate Kit (Dako, Carpinteria, CA) was used for color development followed by Mayer s hematoxylin counterstaining. Isotope-matched human IgG was used in each case as a negative control.

16 191.e3 LI ET AL GASTROENTEROLOGY Vol. 144, No. 1 Supplementary Table 2. Primer Sequences for Real-Time Quantitative PCR Name Primer Sequence SPOCK1 Forward 5=-GGACCCATCCAAGGACCC-3= Reverse 5=-GGCTTGCACTTGACCAAATTC-3= CHD1L Forward 5=-GGTGGAGTTGGCATGAACTT-3= Reverse 5=-CACTCAACTGGAGGTCAGCA-3= MMP-9 Forward 5=-CCATGAGTCCCTGGCAG-3= Reverse 5=-GTAGTATTGTAGGTATGA-3= 18S Forward 5=-CTCTTAGCTGAGTGTCCCGC-3= Reverse 5=-CTGATCGTCTTCGAACCTCC-3= Supplementary Figure 1. A ChIP-PCR assay was performed to detect the existence of indicated DNA fragments pulled down by GFP antibodies (B-2) against GFP tagged CHD1L protein. RNA polymerase antibody was used as a positive control and IgG was used as a negative control. Supplementary Table 1. Cox Proportional Hazard Regression Analysis for Overall Survival Univariable analysis Multivariable analysis Clinicopathologic features HR (95% CI) P HR (95% CI) P Sex ( ) ( ).013 Age ( ) ( ).583 Underlying liver reserve ( ) ( ).348 Type of surgery ( ) ( ).705 Complication from surgery ( ) ( ).055 Comorbidities ( ) ( ).299 Cell differentiation ( ) ( ).114 Tumor staging (AJCC) ( ) < ( ).500 Overexpression of SPOCK1 in tumor ( ) ( ).027 CI, confidence interval; HR, hazard ratio.