A subset of Suz12/PRC2 target genes is activated during HBV replication and liver carcinogenesis associated with hepatitis B virus X protein

Abstract

Chronic hepatitis B virus (HBV) infection is a major risk factor for developing liver cancer, and the HBV X protein (pX) has been implicated as a cofactor in hepatocyte transformation. We have shown that HBV replication as well as in vitro transformation by pX are associated with induction of the mitotic polo-like kinase 1 (Plk1) and down-regulation of the chromatin remodeling components Suz12 and Znf198. Herein, we demonstrate the same inverse relationship between Plk1 and Suz12/Znf198 in liver tumors from X/c-myc bitransgenic mice and woodchuck hepatitis virus (WHV)-infected woodchucks. Employing these animal models and the HBV replicating HepAD38 cells we examined the effect of Suz12/Znf198 down-regulation on gene expression. Genes analyzed include hepatic cancer stem cell markers BAMBI, DKK1,2, DLK1, EpCAM, MYC, and proliferation genes CCNA1, CCND2, IGFII, MCM4–6, PLK1, RPA2 and TYMS. Suz12 occupancy at the promoters of BAMBI, CCND2, DKK2, DLK1, EpCAM and IGFII was demonstrated by chromatin immunoprecipitation in untransformed hepatocytes, but was markedly reduced in pX-transformed and Suz12 knockdown cells. Accordingly, we refer to these genes as “Suz12 repressed” genes in untransformed hepatocytes. The Suz12 repressed genes and proliferation genes were induced in HBV-replicating HepAD38 cells, and interestingly, they exhibited distinct expression profiles during HCC progression in X/c-myc bitransgenics. Specifically, CCND2, EpCAM and IGFII expression was elevated at the proliferative and preneoplastic stages in X/c-myc bitransgenic livers, whereas BAMBI and PLK1 were over-expressed in hepatic tumors from X/c-myc bitransgenics and WHV-infected woodchucks. Importantly, most of these genes were selectively up-regulated in HBV-induced HCCs.

Conclusion

The distinct expression profile of the identified Suz12 repressed genes in combination with the proliferation genes hold promise as biomarkers for progression of chronic HBV infection to HCC.

Chronic hepatitis B virus (HBV) infection is a major factor in the pathogenesis of hepatocellular carcinoma (HCC) (1, 2). Despite availability of a HBV vaccine, the World Health Organization reports approximately 400 million people are chronically infected with HBV. Effective treatment regimens for chronic hepatitis B and HCC remain an important challenge (3–5). To reduce liver cancer, new biomarkers are needed for molecular classification and prognosis of HCC and for developing efficient therapies.

Pathogenesis of HBV-mediated HCC involves chronic liver inflammation (6) and effects of the weakly oncogenic HBV X protein (pX) (7, 8). In a cell culture model of non-transformed hepatocytes (9), pX activates cellular mitogenic pathways (10, 11), promotes DNA damage induced by DNA re-replication (12), and activates Plk1 (13). In turn, activated Plk1 mediates checkpoint adaptation, generating partial polyploidy and transformation (13, 14). Significantly, Plk1 is elevated in liver tumors from chronic HBV patients (15). Likely substrates of Plk1 relevant to HBV-mediated HCC pathogenesis are proteins Znf198 and Suz12 which are involved both in pX-mediated transformation and HBV replication (16).

Znf198 stabilizes the LSD1-CoREST-HDAC1 co-repressor complex that removes histone modifications associated with transcriptional activation (17). In addition, the ZNF198 gene, located on chromosome 13q12.11, is frequently affected by loss of heterozygosity in early onset HCC and correlates with high tumor grade in HBV-induced tumors (18, 19).

Suz12 is an essential component of the Polycomb Repressive chromatin remodeling Complex 2 (PRC2) that mediates the repressive trimethylation of H3 on Lys27, H3K27me3, (20–22). In human embryonic fibroblasts, more than 1000 genes are transcriptionally silenced by PRC2 (23). A subset of Suz12/PRC2 target genes that include EpCAM, IGFII, DKK1,2, DLK1, MYC, MCM5, RPA2, CCNA1 and CCND2 are over-expressed in human HCCs, based on microarray studies of human liver tumors (reviewed in ref. 24). EpCAM and DLK1 are expressed in hepatoblasts (24, 25) and together with DKK1,2 and MYC are up-regulated in hepatic cancer initiating/stem cells (26, 27) and HCCs of poor prognosis derived from chronically HBV-infected patients (28–30). Furthermore, MYC, MCM5, RPA2, CCNA1 and CCND2, the proliferation gene cluster, are also over-expressed in such HCCs (15, 26–31). These observations suggest deregulation of PRC2 activity is involved in HCC pathogenesis.

The mechanism that regulates selection of genes repressed by the PRC2 complex is not yet understood. PRC2 interacts with long noncoding RNAs and short RNAs transcribed from the 5′ end of PRC2 repressed genes (32, 33). These PRC2/RNA interactions suggest mechanisms of PRC2-mediated regulation of gene expression. For example, PRC2 regulates expression of the imprinted DLK1 gene in mouse ESCs via direct interactions with the noncoding RNA Gtl2 acting as cofactor (32). DLK1 is expressed in hepatic progenitors and hepatoblasts (24) and over-expressed in hepatic cancer stem cells from HBV-infected patients with HCC (26). More interestingly, HBV-mediated liver tumors associated with EpCAM positivity and poor prognosis (C3 type) exhibit enhanced expression of a cluster of microRNAs encoded by the imprinted DLK1-DIO3 region (34). Together, these observations suggest loss of PRC2 activity is involved in HBV-mediated hepatocarcinogenesis.

By independent approaches, we have identified down-regulation of the PRC2 component Suz12, during pX-mediated transformation and HBV replication in vitro (16). Accordingly, we hypothesized that genes repressed by the Suz12/PRC2 complex will be activated during HBV-mediated hepatocarcinogenesis and HBV replication. To test this hypothesis, we examined expression of HCC-relevant Suz12 target genes (24) during pX-mediated transformation and HCC pathogenesis, as well as in the presence of HBV replication. These HCC-relevant, Suz12 regulated genes include EpCAM, IGFII, DKK1,2, BAMBI, DLK1, MYC, MCM5, RPA2, CCNA1 and CCND2 (24). Herein we demonstrate induction of these genes: i) in an in vitro model of pX-mediated transformation, ii) in liver tumors from X/c-myc bitransgenic mice (7) and woodchucks chronically infected by woodchuck hepatitis virus (WHV), and iii) in the presence of HBV replication (35). Significantly, the distinct expression profile of these Suz12 target genes during liver cancer development in the X/c-myc bitransgenics suggests that they could serve as biomarkers for HBV-mediated disease progression.

EXPERIMENTAL PROCEDURES

Cell lines

4pX-1 (9), 4pX-1-SUZ12^kd (Suz12 knockdown), 4pX-1-ZNF198^kd (Znf198 knockdown) and 4pX-1^GIPZ (vector control cell line for 4pX-1-SUZ12^kd and 4pX-1-ZNF198^kd) (16)are tetracycline-regulated pX-expressing cell lines, grown as described (9, 16). 4pX-1^GIPZ is equivalent to 4pX-1 cell line (9, 16). pX-transformed cultures were derived from 4pX-1 cells as described (14). The HepAD38 cell line that supports HBV replication by removal of tetracycline was grown as described (35). HBV replication was quantified as described (16).

Real-Time PCR

cDNA was synthesized from 2.0 μg total RNA using iSCRIPT cDNA synthesis kit (Bio-Rad). Real-time PCR reactions were performed in triplicates and normalized to GAPDH. Primer sequences are listed in supplementary section.

Preparation of Whole Cell Extract (WCE) and Immunoblotting

Mouse or woodchuck liver was homogenized in RIPA buffer (50 mM Tris pH7.4, 150 nM NaCl, 1 % NP-40, 1 mM EDTA, 100 mM NaF, 50 mM glycerol phosphate, 1 mM sodium orthovanadate, 1 mM PMSF, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin), sonicated on ice for 10 seconds and clarified by centrifugation (12,000 rpm, 15 minutes at 4°C). WCE (10–20 μg) were analyzed by SDS-PAGE and immunoblotted using standard protocols. Antibodies: Plk1, 1:1000 (Abcam), Suz12, 1:1000 (Abcam 12073), Znf198, 1:500 (Abcam), phospho-Plk1 T210, 1:500 (Abcam), phospho-Plk1 S137, 1:500 (Abcam), actin, 1:2000 (Sigma).

Chromatin Immunoprecipitation (ChIP) assays were performed as described (11) employing 3.0 μg ChIP-validated Suz12 antibody (Abcam 12073). Immunoprecipitated DNA was quantified by real-time PCR. Primer sequences are listed in supplementary section.

Statistical analyses to evaluate gene expression differences in HCC subgroups and normal liver were performed by Student’s t-test (two-tailed). Outlier enrichment analysis in HCC subgroups were calculated by Chi square test.

RESULTS

Elevated Plk1 and reduced protein levels of Znf198 and Suz12 during HCC pathogenesis in X/c-myc bitransgenic mice

We previously identified the genes SUZ12 and ZNF198 as negative regulators of pX-mediated transformation (16). During oncogenic transformation of pX-expressing hepatocytes (14), SUZ12 and ZNF198 exhibit a progressive decrease in protein but not mRNA levels, while Plk1 protein levels increase (16). Elevated Plk1 and reduced protein levels of Suz12 and Znf198 also occurs in cell lines derived from HBV-induced liver tumors and in HBV-replicating HepAD38 cells (16). To establish the importance of these molecules in HBV pX-mediated hepatocarcinogenesis, we examined protein levels of Plk1, Znf198 and Suz12 employing the X/c-myc bitransgenic mouse model which exhibits accelerated HCC development due to pX expression in the liver (7).

X/c-myc bitransgenics exhibit enhanced hepatocyte proliferation at 2 weeks, preneoplastic liver lesions at 4 months, and macroscopic liver tumors starting at 7–9 months (7). Immunoblots of liver lysates from 2-week and 4-month X/c-myc bitransgenic mice exhibited elevated Plk1 protein compared to WT mice (Fig. 1A). Likewise, increased Plk1 was observed in tumor (T) vs. peri-tumoral/normal (N) liver tissue at 12 months. On the other hand, protein levels of Znf198 were significantly reduced (p<0.01) as early as 2 weeks in X/c-myc bitransgenic livers, and remained reduced in T vs. N liver tissue. Suz12 protein levels exhibited a significant reduction in most tumors at 12 months, compared to peri-tumoral tissue (Fig. 1A–C).

Fig. 1.

Increased protein levels of Plk1 and reduced levels of Znf198 and Suz12 during liver cancer pathogenesis in X/c-myc bitransgenics. A. Immunoblots of liver lysates from WT and X/c-myc bitransgenics at the indicated ages employing antibodies for Plk1, Znf198, Suz12 and actin. A representative immunoblot is shown from at least three independent WCE preparations. Protein levels of Plk1, Znf198 and Suz12 were quantified vs. actin by ImageJ software. B. Quantification of protein levels of Plk1, Znf198 and Suz12, in liver lysates from 2-week and 4-month WT and X/c-myc bitransgenic mice, is the average of values from all tested animals in each group, from at least three independent WCE preparations. Error bars indicate +/− standard error of the mean. * indicates p<0.01 (student’s t test) C. Quantification of protein levels of Plk1, Znf198 and Suz12, in liver lysates from individual 12-month X/c-myc bitransgenics, in paired tumor vs. peri-tumoral tissue. The average from at least three independent experiments is shown. Error bars indicate +/− standard error of the mean. * indicates p<0.01 and ** p<0.05 (student’s t test).

To understand the role of pX in accelerating HCC pathogenesis in X/c-myc bitransgenics, we first determined the level of X mRNA by real-time PCR. As shown previously (7), highest expression of X was detected in X/c-myc bitransgenic liver at 2 weeks, decreasing to significantly lower levels by 4 months (Fig. 2A). Next, we determined by immunoblots the level of Plk1 and activated Plk1 (phospho-T210 or phospho-S137 form) in liver of 4-month WT, X and c-myc monotransgenics, and X/c-myc bitransgenics. Compared to WT animals, levels of Plk1 and active Plk1 were increased in liver of X and c-myc monotransgenics as well as X/c-myc bitransgenics. In agreement with the role of Plk1 in mitosis (36), levels of pH3 an early mitotic marker exhibited a concomitant increase, while protein levels of Suz12 and Znf198 exhibited a modest reduction in X and c-myc monotransgenics and a more pronounced reduction in X/c-myc bitransgenics (Fig. 2B and C). These results are consistent with effects of both c-myc (37) and X (10) in stimulating cell cycle progression and proliferation, and with the effect of pX in accelerating liver tumorigenesis of c-myc monotransgenics (7).

Fig. 2.

Enhanced expression and activation of Plk1 in X/c-myc bitransgenics. A. Real-time PCR quantification of X mRNA, employing liver tissue from the indicated X/c-myc bitransgenics. Results are from three independent RNA preparations, analyzed by real-time PCR in triplicates with GAPDH as internal control. B. Immunoblots of liver lysates from 4-month WT, X and c-myc monotransgenic mice, and X/c-myc bitransgenics, employing indicated antibodies. A representative experiment is shown from two independent WCE preparations. C. Quantification of indicated proteins from immunoblots shown in B. vs. actin is by ImageJ software, from two independent WCE preparations. Error bars indicate +/− standard error of the mean.

Enhanced protein level of Plk1 and reduced level of Suz12 in liver tumors of chronically WHV-infected woodchucks

To assess the relevance of the observations derived from the X/c-myc mouse model, we performed similar analyses using liver tumors from woodchucks with established chronic WHV infection. Experimental infection of neonate woodchucks with WHV closely models HBV-induced pathogenesis, including development of HCC (38). We employed liver tumors from WHV-infected woodchucks previously characterized by Jacob and coworkers, and classified as well-differentiated (T₁), less-differentiated (T₂), and least-differentiated (T₃), based on anaplastic features and tumor size (39).

Protein levels of Plk1 and Suz12 were examined by immunoblots. Information about Znf198 was hampered by lack of cross-reactivity of the Znf198 antibody with woodchuck protein. We observed increased levels of Plk1 in tumor vs. peri-tumoral liver, and decreased levels of Suz12 in most tumors (Fig. 3). Thus, opposite changes in Plk1 and Suz12 expression occur during hepatocarcinogenesis induced by chronic hepadnaviral infection, similar to HCC pathogenesis in X/c-myc bitransgenic mice (Fig. 1).

Fig. 3.

Increased protein levels of Plk1 and reduced levels of Suz12 in hepatic tumors from WHV-infected animals. Immunoblots of Plk1, Suz12 and actin using lysates from WHV-induced liver tumors. Tumors were classified as well-differentiated HCC (T₁), less-differentiated (T₂), and least-differentiated (T₃) based on anaplastic features and tumor size (39). Relative intensity of indicated protein bands was quantified vs. actin by ImageJ software. A representative assay is shown from at least three independent WCE preparations.

Expression of Suz12/PRC2 repressed genes during pX-mediated transformation

Since Suz12 is down-regulated in cellular (14) and animal models (Figs 1 and 3) of hepadnaviral hepatocarcinogenesis, we investigated whether genes known to be repressed by the Suz12/PRC2 complex become expressed during transformation. The PRC2 complex silences numerous genes in human embryonic fibroblasts including BAMBI, CCNA1, CCND2, DKK1, 2, DLK1, EpCAM, IGFII, MCM5, MYC and RPA2 (23). Moreover, independent studies have confirmed EpCAM and IGFII expression is regulated by dynamic changes in H3K27me3 (40, 41). In this study we examined expression of this subset of Suz12 repressed genes, because they are over-expressed in hepatic cancer stem cells (26–28) and in HCCs of poor prognosis (29), reviewed in (24). First, we quantified by real-time PCR their mRNA levels in pX-transformed cells (14), Suz12 knockdown 4pX-1 cells (4pX-1-SUZ12^kd) used as positive control, and untransformed 4pX-1^GIPZ cells (16). We observed increased expression in pX-transformed cells and 4pX-1-SUZ12^kd cells, relative to 4pX-1^GIPZ cells (Fig. 4A and supplementary Table 1).

Fig. 4.

Identification of Suz12/PRC2-repressed genes in untransformed hepatocytes. A. quantification by real-time PCR of Suz12/PRC2 target genes using total RNA isolated from pX-transformed cells (14), 4pX-1-Suz12^kd (16) and 4pX-1^GIPZ cells (16) expressing pX by tetracycline removal for 24h. Results are from three independent RNA preparations, analyzed by real-time PCR in triplicates with GAPDH as internal control. Error bars indicate +/− standard error of the mean. B. Chromatin immunoprecipitation (ChIP) assays with Suz12 antibody in 4pX-1^GIPZ cells, pX-transformed, and 4pX-1-SUZ12^kd cells. ChIP-derived DNA was amplified by PCR with primer pairs (see supplementary information) for indicated genes. Real-time PCR quantification of indicated genes is expressed as percent reduction in Suz12 occupancy in pX-transformed and 4pX-1-Suz12^kd cells relative to untransformed 4pX-1^GIPZ cells. Error bars indicate +/− standard error of the mean. C. Quantification by real-time PCR of Suz12-repressed genes using total RNA isolated from pX-transformed cells (14), 4pX-1-Znf198^kd (16) and 4pX-1^GIPZ cells (16) expressing pX by tetracycline removal for 24h. Results are from three independent RNA preparations, analyzed by real-time PCR in triplicates with GAPDH as internal control. Error bars indicate +/− standard error of the mean. D. Chromatin immunoprecipitation (ChIP) assays with Suz12 antibody in 4pX-1^GIPZ cells, pX-transformed and 4pX-1-Znf198^kd cells. ChIP-derived DNA was amplified by PCR with primer pairs (see supplementary information) for indicated genes. Real-time PCR quantification of indicated genes is expressed as percent reduction in Suz12 occupancy in pX-transformed and 4pX-1-Znf198^kd cells relative to untransformed 4pX-1^GIPZ cells. Error bars indicate +/− standard error of the mean.

To determine whether these genes are directly repressed by Suz12/PRC2 complex in untransformed 4pX-1^GIPZ cells, we employed chromatin immunoprecipitation (ChIP) assays with ChIP-validated Suz12 antibody, as described (11). In comparison to untransformed 4pX-1^GIPZ cells, Suz12 occupancy was reduced at the promoters of CCND2, DKK2, EpCAM and IGFII in pX-transformed cells and 4pX-1-SUZ12^kd cells (Fig. 4B). Suz12 occupancy at the BAMBI promoter exhibited a 20% reduction in pX-transformed cells relative to 4pX-1^GIPZ cells, and nearly a 10% reduction with DLK1 promoter. By contrast, absence of Suz12 association with CCNA1, MYC and RPA2 promoters was independent of pX-mediated transformation, because this was also observed in untransformed 4pX-1 cells (Supplementary Fig. 1). We conclude BAMBI, CCND2, DKK2, DLK1, EpCAM and IGFII genes are transcriptionally repressed, to varying degrees, by Suz12/PRC2 complex in untransformed 4pX-1^GIPZ cells.

Suz12/PRC2 complex is tethered via long intergenic RNA (lincRNA) to LSD1-CoREST-HDAC1 complex stabilized by Znf198 (42). Since protein levels of Znf198 are significantly reduced in proliferative stage of X/c-myc bitransgenic liver (Fig. 1A), we examined effect of Znf198 knockdown on expression of the identified Suz12-repressed genes. Employing the Znf198 knockdown 4pX-1 cell line (4pX-1-ZNF198^KD) described earlier (16), we quantified expression of Suz12-repressed genes by real-time PCR (Fig. 4C), and Suz12 occupancy by ChIP assays (Fig. 4D). With the exception of CCNA1 and CCND2, the Suz12-repressed genes are induced in Znf198 knockdown cells (Fig. 4C). Conversely, with the exception of CCNA1, MYC and RPA2, Suz12 occupancy at these promoters was reduced in the context of Znf198 knockdown (Fig. 4D).

Differential expression of Suz12-repressed genes during HCC pathogenesis in X/c-myc bitransgenics

Next, we investigated the expression profile of Suz12 target genes BAMBI, CCND2, DKK2, DLK1, EpCAM and IGFII and proliferation genes CCNA1, MCM4–6, MYC, PCNA, PLK1, RPA2 and TYMS during liver tumorigenesis in X/c-myc bitransgenics (Fig. 5A and B, and supplementary Table 2). CCND2, EpCAM and IGFII were highly up-regulated in liver of 2-week and 4-month X/c-myc bitransgenics relative to WT liver. By contrast, at 12 months, fold induction of CCND2 and EpCAM was smaller when comparing tumor vs. peri-tumoral liver, while expression of BAMBI, DLK1 and IGFII progressively increased from proliferative to HCC stage at 12 months (Fig. 5A). Proliferation genes MCM5, MCM6, RPA2, TYMS and PLK1 exhibited enhanced expression at preneoplastic stage (4 months) (Fig. 5B). Interestingly, expression of PLK1 significantly increased in HCCs (Fig. 5B).

Fig. 5.

Expression of the Suz12 repressed genes during distinct stages of HCC pathogenesis in X/c-myc bitransgenics. Quantification by real-time PCR of the indicated Suz12 repressed genes in A., and the proliferation gene cluster in B., using total RNA isolated from liver of X/c-myc bitransgenics at the indicated stages of HCC pathogenesis. Total RNA isolated from liver tissues shown in Fig. 1A, was pooled and analyzed by real-time PCR. Each PCR reaction was performed in triplicates from three independent RNA preparations. Quantification is relative to GAPDH used as internal control. Error bars indicate +/− standard error of the mean.

Enhanced expression of Suz12-repressed genes during HBV replication

Protein levels of Suz12 and Znf198 are down-regulated in HBV replicating cells, and siRNA knockdown of Suz12 and Znf198 increases HBV replication (16). Accordingly, we hypothesized the Suz12 repressed genes might be up-regulated during HBV replication. Employing the HepAD38 cell line that supports HBV replication following tetracycline removal (35), we quantified viral replication as well as expression of Suz12-regulated and proliferation genes, 10 days after induction of HBV replication (Fig. 6). Also, because our studies indicate that Plk1 down-regulates Suz12 (16), we investigated the effect of Plk1 inhibition on viral replication and expression of the proliferation cluster and Suz12 target genes.

Fig. 6.

Enhanced expression of the Suz12 repressed genes and proliferation genes in the presence of HBV replication. A. Quantification of HBV replication in HepAD38 cells that support HBV replication following removal of tetracycline (35). HepAD38 cells were grown with (+) or without (−) tetracycline for 10 days, i.e. without (−) or with (+) HBV replication, respectively. HBV genome equivalents were quantified by real-time PCR of DNA isolated from purified intracellular virions as described (16). Effect of DMSO addition (5.0 μl) on HBV replication, added on day 9 for 24 h (1 day) or day 8 for 48 h (2 days), as indicated, prior to cell harvesting. 250 nM or 500 nM BI 2536 dissolved in DMSO (5.0 μl) added on day 9 for 24 h (1 day) or day 8 for 48 h (2 days), as indicated, prior to cell harvesting. Results represent the average of three independent experiments. Error bars indicate +/− standard error of the mean. B. Immunoblots of Suz12 and Znf198 using lysates of HepAD38 cells grown with (+) or without (−) tetracycline for 10 days and treated for 24 h prior to harvesting with 500 nM BI2536. Relative intensity of indicated protein levels was quantified by ImageJ software vs. actin. A representative experiment is shown from three independent experiments. Quantification by real-time PCR of the proliferation gene cluster in C., and Suz12 repressed genes in D. using total RNA isolated from HepAD38 cells grown with (+) or without (−) tetracycline for 10 days and treated for 24h prior to harvesting with 500 nM BI2536. Quantification is relative to GAPDH used as internal control. Results are from three independent RNA isolations. PCR assays were performed in triplicates. Error bars indicate +/− standard error of the mean.

Treatment with Plk1 inhibitor BI 2536 (43) for 1 or 2 days, added on day 9 or day 8 of HBV replication, respectively, suppressed HBV replication, quantified by measuring HBV genome equivalents in purified intracellular virions (Fig. 6A), immunoblots of HBV core antigen (Supplementary Fig. 2A), and PCR quantification of viral RNA (Supplementary Fig. 2C). BI 2536 addition inhibited Plk1 (Supplementary Fig. 2B) but did not induce apoptosis, determined by unchanged levels of pro-caspase-3 and caspase-3 (Supplementary Fig. 2A), while protein levels of Znf198 and Suz12 were increased (Fig. 6B and Supplementary Fig. 2C). Under these conditions of Plk1 inhibition, we quantified expression of the proliferation cluster and Suz12-repressed genes on day 10 of HBV replication (Fig. 6C and D). Expression of both groups of genes increased in the presence of HBV replication (Fig. 6C and D). BI 2536-mediated Plk1 inhibition significantly reduced expression of all proliferation genes (Fig. 6C), demonstrating the link between Plk1 activity and cell cycle progression. Likewise, the Suz12 repressed genes DKK2, EpCAM and IGFII were significantly decreased by BI 2536 treatment (Fig. 6D).

Enhanced expression of SUZ12-repressed genes in hepatic tumors from WHV-infected woodchucks

To determine the relevance of the observations derived from the 4pX-1 cell line and the X/c-myc bitransgenics (Figs. 1 and 4), we quantified expression of the Suz12-repressed genes and proliferation genes in WHV-infected liver tumors. RNA from T₁, T₂, and T₃ liver tumors used in Fig. 3, was analyzed by real-time PCR. Both groups of genes were induced in WHV-infected HCCs in comparison to peri-tumoral tissue (Fig. 7 and Supplementary Table 3). The proliferation cluster and in particular PLK1 were greatly induced in less-differentiated T₂ and T₃ tumors (Fig. 7A), and EpCAM expression was significantly elevated in all hepatic tumors (Fig. 7B). Importantly, expression of BAMBI, similar to HCCs from X/c-myc bitransgenics, was also induced in WHV-infected liver tumors.

Fig. 7.

Enhanced expression of the proliferation gene cluster and the Suz12 repressed genes in hepatic tumors from WHV infected animals. Tumors were classified (39) as well-differentiated HCC (T₁), less-differentiated (T₂), and least-differentiated (T₃). Total RNA isolated from individual paired liver tissues, tumor vs. peri-tumoral, shown in Fig. 2, was analyzed by real-time PCR. A., Quantification of the proliferation gene cluster and B., of the Suz12 repressed genes. Quantification is relative to GAPDH used as internal control. Results are the average of three independent experiments, each PCR assay performed in triplicates. The fold induction of each gene with standard error of the mean is shown in supplementary Table 3.

To assess whether the observations reported above are linked to gene expression profiles in human HBV-induced HCC, we examined expression of Suz12 target genes and proliferation genes in a microarray-based transcriptome database from the study by Boyault et al (29). Table I shows expression levels of these genes in HBV-positive and -negative HCC vs. normal liver. Expression of MCM7, PCNA, PLK1 and RPA1 is significantly increased in HBV-induced HCCs. The Suz12-regulated genes DKK1, DLK1, IGFII and EpCAM showed marked up-regulation in HBV-positive HCC, but the values did not reach statistical significance, suggesting elevated expression of these genes in HBV-positive HCC is not homogeneous (ref. 29 and supplementary Fig. 3). To verify that over-expression of these genes was significantly higher in HBV-positive HCC compared to HBV-negative HCC, samples were ranked by gene expression intensity, and a contingency Table was generated comparing the number of samples with expression level ranked among the top 10 or 20% of the whole HCC series (90^th and 80^th percentile, respectively, in Table I). This analysis revealed that liver tumors from HBV-infected patients display significantly elevated expression of DKK1, DLK1, IGFII, EpCAM, MCM6, PLK1 and RPA1, thereby supporting the data obtained from the animal models of viral hepatocarcinogenesis.

Table I.

Expression of proliferation and Suz12 target genes in HBV-positive and HBV-negative HCC

Probe set	Gene Symbol	HBV-positive HCC (n=16)		HBV-negative HCC (n=43)		** Above/below 90th percentile			** Above/below 80th percentile
		T/NT*	p-value (t-test)	T/NT*	p-value (t-test)	HBV+ HCC (n=16)	HBV− HCC (n=43)	p-value (x2)	HBV+ HCC (n=16)	HBV− HCC (n=43)	p-value (x2)
203204_at	BAMBI	1.5	0.224	1.4	0.464	2/14	4/39	0.78	3/13	9/34	0.85
205889_at	CCNA1	0.8	0.017	0.8	0.007	2/14	4/39	0.78	4/12	8/35	0.59
208712_at	CCND2	0.7	0.190	0.6	0.022	3/13	3/40	0.18	6/10	6/37	0.046
204602_at	DKK1	16.4	0.202	2.1	0.521	5/11	1/42	0.001	7/9	5/38	0.006
209560_s_at	DLK1	7.7	0.320	7.0	0.728	4/12	2/41	0.022	4/12	8/35	0.59
210881_s_at	IGFII	6.0	0.176	1.2	0.833	6/10	0/42	<0.001	6/10	6/37	0.046
222037_at	MCM4	1.7	0.087	1.4	0.174	2/14	4/39	0.78	3/13	9/34	0.85
216237_s_at	MCM5	1.5	0.139	1.3	0.384	1/15	5/38	0.54	5/11	7/36	0.2
201930_at	MCM6	1.7	0.056	1.6	0.242	1/15	5/38	0.54	6/10	6/37	0.046
210983_2_at	MCM7	1.8	0.019	1.7	0.174	0/16	6/37	0.12	4/12	8/36	0.59
202431_s_at	MYC	0.8	0.534	0.9	0.704	2/14	4/39	0.78	3/13	9/34	0.85
201202_at	PCNA	1.7	0.020	1.7	0.102	1/15	5/38	0.54	3/13	9/34	0.85
201429_s_at	PLK1	1.2	0.010	1.1	0.061	3/13	3/40	0.18	7/9	5/38	0.005
201529_s_at	RPA1	1.8	0.003	1.4	0.084	2/14	4/39	0.78	6/10	6/37	0.045
201839_s_at	EpCAM	4.8	0.212	0.5	0.436	4/12	2/41	0.022	8/8	4/39	0.001

Fold-change >4

^*T/NT=Tumor/Non-tumor

HBV⁺ = HBV-positive

p<0.05

HBV⁻ = HBV-negative

p<0.01

^**Above/below indicates the number of samples above/below the given percentile

DISCUSSION

Protein levels of Plk1 and chromatin remodeling components Suz12 and Znf198 exhibit an opposite relationship both during pX-mediated transformation and HBV replication (16). Herein we show that this opposite relationship between Plk1 vs. Suz12 and Znf198 is also observed in animal models of hepadnavirus-mediated hepatocarcinogenesis, including X/c-myc bitransgenic mice (7) and woodchucks chronically infected with WHV (38). In X/c-myc bitransgenics, expression of X in liver cooperates with c-myc, reducing liver tumor latency by 2–3 months (7). Since X monotransgenics do not develop liver tumors, this animal model has established the weak oncogenic potential of HBV X protein. On the other hand, chronic WHV infection in woodchucks recapitulates HBV infection and HBV-mediated hepatocarcinogenesis in humans (38, 44–45).

Increased Plk1 is observed in both animal models of liver tumorigenesis (Figs. 1 and 3). Increased Plk1 expression and activation is also detected in liver of X and c-myc monotransgenics (Fig. 2), in agreement with the role of both c-myc and pX in proliferation. Lower Plk1 activation in X vs. c-myc monotransgenics is congruent with the weak oncogenic potential of pX (7). In X/c-myc bitransgenics, expression and activation of Plk1 is further increased, as well as down-regulation of Suz12 and Znf198, demonstrating a cooperative effect between c-myc and pX. This cooperative effect could involve direct Plk1-mediated down-regulation of Suz12 and Znf198 or induction of microRNAs targeting Suz12 and Znf198. For example, miR-200 shown to target Suz12 (47) is significantly up-regulated in human cirrhotic liver (C. Rogler personal communication). Further studies are needed to understand the mechanism of this cooperation between X and c-myc, and the down-regulation of Suz12 and Znf198 proteins.

Znf198 protein levels are significantly reduced at the proliferative liver-stage of X/c-myc bitransgenics (Fig. 1). On the other hand, Suz12 protein levels are slightly reduced in preneoplastic stage at 4 months, and significantly reduced in most liver tumors of both animal models (Figs. 1 and 3). The variation in Suz12 protein levels in tumor vs. peri-tumoral tissue suggests multiple mechanisms contribute to PRC2 inactivation. For example, EZH2, an essential component of PRC2, is inactivated by Akt-mediated phosphorylation (48). Significantly, Akt activation is a hallmark of poor prognosis HBV-mediated HCCs (29). Other mechanisms that inactivate PRC2 include loss of JARID required for association of PRC2 with target genes (49), or induction of H3K27me3 demethylase, as in Epstein Bar virus-infected cells and Hodgkin’s lymphoma (50). Furthermore, combined effects of loss of Suz12/PRC2 and Znf198/LSD1-CoREST-HDAC1 complexes may contribute to oncogenic transformation. This notion is supported by studies showing that these complexes are linked via the lincRNA HOTAIR, thereby coupling repressive H3K27me3 mediated by PRC2 with demethylation of transcriptionally activating H3K4me3 (42). Indeed, Suz12 occupancy is reduced in promoters of select Suz12-repressed genes in the context of Znf198 knockdown, thereby allowing their expression (Fig. 4).

Re-expression of Suz12-repressed genes in animal models of X-mediated liver cancer and in HBV replicating cells

Although Suz12/PRC2 silences more than 1,000 genes in human embryonic fibroblasts (23), in this study we analyzed a subset of Suz12/PRC2 repressed genes relevant to HBV-mediated liver cancer (24). One group of Suz12/PRC2 repressed genes includes proliferation genes over-expressed in human HCCs (15, 29), and the other, genes over-expressed in hepatic cancer stem cells (26, 27). Indeed, ChIP assays with Suz12 antibody demonstrated that BAMBI, CCND2, DKK2, DLK1, EpCAM and IGFII are direct targets of transcriptional repression by the Suz12/PRC2 complex in untransformed hepatocytes (Fig. 4B). Interestingly, these Suz12-repressed genes and the proliferation genes exhibit distinct expression in X/c-myc bitransgenic liver during progression to HCC (Fig. 5). Expression of CCND2, EpCAM and IGFII increased in the liver of 2-week X/c-myc bitransgenics (proliferative stage) whereas CCNA1, MCM6, RPA2, TYMS and PLK1 exhibit highest expression at the preneoplastic stage (4 months). At the HCC stage, Suz12 target genes BAMBI, DLK1 and IGFII and proliferation marker PLK1 were significantly elevated. In HCCs from WHV-infected woodchucks expression of BAMBI, PLK1 and also EpCAM was elevated (Fig. 7). Lack of EpCAM induction in HCCs from X/c-myc bitransgenics (Fig. 5) is likely due to their well-differentiated status and non-metastatic potential, whereas reduced expression of CCNA1, CCND2 and MYC in HCCs of X/c-myc bitransgenics (Fig. 5) maybe due to comparison with peri-tumoral liver. Lastly, analysis of a human HCC transcriptome database (29) shows increased expression of the Suz12-regulated genes DKK1, DLK1, IGFII and EpCAM, and proliferation genes MCM6, PLK1 and RPA in HCCs from HBV-infected patients, supporting our conclusions derived from animal models of viral hepatocarcinogenesis. More human HCC samples must be analyzed to conclusively determine the biomarker potential of these genes.

Enhanced expression of the proliferation cluster and Suz12 target genes also occurs in HBV replicating HepAD38 cells (Fig. 6). Importantly, inhibition of Plk1 by BI2536 significantly reduced viral replication as well as expression of both the proliferation cluster and select Suz12 target genes (Fig. 6), suggesting Plk1 inhibitors could exert antiviral and anticancer effects.

In summary, the distinct expression profile of genes repressed by Suz12/PRC2 in untransformed hepatocytes in combination with PLK1 expression distinguish the proliferative and preneoplastic stages from the HCC stage in both animal models of HCC pathogenesis. The analysis examining expression of these genes in human HCCs (29) also suggests that these genes could serve as prognostic biomarkers of subtypes of HBV-mediated HCC.

Supplementary Material

Supp Fig S1. Supplementary Figure 1.

Chromatin immunoprecipitation (ChIP) assays with Suz12 antibody or IgG as indicated in 4pX-1 cells, pX-transformed cells (14), and 4pX-1-Suz12^kd cells (16). The ChIP-derived DNA was amplified by PCR with primer pairs (see supplementary information) for indicated genes and PCR products were analyzed by agarose gel electrophoresis.

Supp Fig S2. Supplementary Figure 2.

HepAD38 cells that support HBV replication following removal of tetracycline (35), were grown with (+) or without (−) tetracycline for 10 days, i.e. without (−) or with (+) HBV replication, respectively. A. 250 nM or 500 nM BI 2536 added on day 9 for 24 h (day 1) or day 8 for 48 h (day 2), respectively, as indicated, prior to cell harvesting. WCE isolated without (−) or with (+) addition of BI 2536 were immunoblotted for HBV core antigen, caspase 3 and cleaved/active caspase 3. B. Immunoblots of active Phospho-Plk1 (T210) and Plk1 using WCE isolated from HepAD38 cells grown in the presence of 250 nM BI 2536 for 2 days or 500 nM BI 2536 for 1 day, as indicated. C. 250 nM or 500 nM BI 2536 added on day 8 for 48 h (2 days) or day 9 for 24 h (1 day), respectively, as indicated, prior to isolation of total RNA. Expression of viral RNA was quantified by real-time PCR employing primer pairs described by Zhang et al, Hepatology 53, 1476–1485, 2011. Results are the average from two independent WCE preparations. Error bars indicate +/− standard error of the mean. D. Quantification of protein levels of Suz12 and Znf198 from immunoblots shown in Fig. 5B, using lysates of HepAD38 cells grown with (+) or without (−) tetracycline for 10 days and treated for 24 h prior to harvesting with 500 nM BI2536. Quantification of indicated proteins is by ImageJ software vs. actin. Results are the average from two independent WCE preparations. Error bars indicate +/− standard error of the mean.

Supp Fig S3. Supplementary Figure 3.

Analysis of microarray data from the study by Boyault et al (29). Relative induction of indicated genes is individual liver tumors with high HBV titer (H), low HBV titer (L), no HBV titer (N), and non-tumor normal liver (n).

Supp Table S1. Supplementary Table 1.

Expression of indicated genes during oncogenic transformation of X-expressing cells. P2-P5 cultures are derived by consecutive passages of partially polyploid pX-expressing cells, isolated by live cell-sorting from the immortalized 4pX-1 cell line, as described (14). P2 and P3 cultures are not transformed, whereas P4 and P5 cultures are pX-transformed cells (14). 4pX-1-Suz12^kd is a Suz12 knockdown cell line (16), used as positive control. Fold induction of indicated genes was quantified relative to expression in untransformed 4pX-1 cells. Results are from at least three independent RNA preparations used in real-time PCR reactions. Each PCR reaction was performed in triplicates and quantified relative to GAPDH used as internal control, +/− standard error of the mean.

Supp Table S2. Supplementary Table 2.

Fold induction of indicated genes in the liver of 4-month old X monotransgenics, c-myc monotransgenics and X/c-myc bitransgenics in comparison to WT mice of the same age, quantified by real-time PCR. Results are from at least three independent RNA preparations used in real-time PCR reactions. Each PCR reaction was performed in triplicates and quantification was relative to GAPDH used as internal control, +/− standard error of the mean..

Supp Table S3. Supplementary Table 3.

Fold induction of the proliferation gene cluster and Suz12-repressed genes quantified in tumor vs peri-tumoral paired tissues from liver of individual WHV-infected animals. The woodchuck identification number is from the study by Jacob et al (39). Results are from at least three independent RNA preparations used in real-time PCR reactions. Each PCR reaction was performed in triplicates and quantification was relative to GAPDH used as internal control, +/− standard error of the mean.

LIST OF ABBREVIATIONS

HBV

hepatitis B virus

HCC

hepatocellular carcinoma

WCE

whole cell extract

Plk1

Polo-like kinase 1

PCR

polymerase chain reaction

PRC2

polycomb repressive complex 2

Suz12

suppressor of zeste 12 homolog (Drosophila)

Znf198

zinc finger, MYM-type 2

LSD1-Co-REST-HDAC1

lysine-specific demethylase 1-corepressor of RE1-silencing transcription factor-histone deacetylase 1

BAMBI

BMP and activin membrane-bound inhibitor

EpCAM

epithelial cellular adhesion molecule

IGFII

Insulin-like growth factor 2

DKK2

Dickkopf homolog (xenopus) 2

DLK1

Delta-like homolog (Drosophila) 1

CCND2

Cyclin D2

CCNA1

Cyclin A1

PCNA

proliferating cell nuclear antigen

MCM4–6

minichromosome maintenance 4–6

TYMS

thymidylate synthase

RPA2

replication protein A2