Capicua suppresses hepatocellular carcinoma progression by controlling the ETV4–MMP1 axis

TISSUE MICROARRAY AND IMMUNOHISTOCHEMISTRY

Two liver cancer tissue microarrays with liver tumors and adjacent normal liver tissues (LV1221 and LV6161) were purchased from Biomax (MD, USA). Formalin-fixed, paraffin-embedded specimens were deparaffinized and stained with rabbit polyclonal anti-CIC antibody. Each sample stained with anti-CIC antibody was scored as negative (–), weak (+), moderate (++), or strong (+++) according to the staining intensity. These scores were determined independently by two pathologists. The scoring by the pathologists was done in a blinded manner.

INDUCTION OF HCC IN MICE

To induce HCC, diethylnitrosamine (DEN; Sigma-Aldrich, MO) was injected intraperitoneally into 2-week-old male mice (5 μg/g). For tumor formation analysis, mice were sacrificed to prepare liver tissues at 8 months after DEN treatment. Externally visible tumors (>1 mm) on the liver were counted and measured. Livers were microdissected into tumor and nontumor sections and stored at −80°C until analyzed by quantitative RT-PCR.

Lung metastasis and survival were analyzed at 15 months after DEN treatment. The survival of mice was recorded weekly. After 15 months, mice were sacrificed, and their lungs were dissected, paraffin-embedded, and used for hematoxylin and eosin staining. Serial sections of entire lung tissues were conducted. Four sections per lung tissue were chosen for hematoxylin and eosin staining. The total number of metastasized tumor lesions was determined from the hematoxylin and eosin–stained sections and used for calculation of the average number of tumor foci in a lung tissue for each genotype.

Other assays used in this study are described in the Supporting Information.

INVERSE CORRELATION BETWEEN CIC PROTEIN LEVELS AND HCC SEVERITY

Because the role of CIC in HCC development and progression has not yet been determined, we analyzed The Cancer Genome Atlas (TCGA) data sets for patients with HCC (Supporting Table S1) in order to gain insight on the relevance between CIC and HCC pathogenesis. CIC mRNA levels were not down-regulated but rather increased in HCC tissue samples compared with normal liver tissues (Fig. 1A). However, the survival rate was significantly decreased in HCC patients with low levels of CIC (lower 20%, n = 74) compared to those with high levels of CIC (upper 20%, n = 74) (Fig. 1B; Supporting Table S2). Studies have demonstrated that activation of epidermal growth factor receptor and its downstream signaling molecules, which promote tumorigenesis and cancer metastasis, inactivates CIC through either degradation or cytoplasmic translocation.20, 21 Moreover, we have shown that CIC protein levels were dramatically decreased in prostatic adenocarcinoma.35 Thus, we examined CIC protein levels in normal liver and HCC tissues on tissue microarrays by immunohistochemistry using anti-CIC antibody. Reduced expression of CIC was more frequently observed in HCC samples than in normal liver tissues (Fig. 1C). To directly address whether CIC expression decreases in HCC tissues at the protein level, but not at the mRNA level, we examined expression profiles of both CIC protein and CIC mRNA in the same tissue samples of normal liver and HCC of different pathological stages (Supporting Table S3). Most HCC tissues showed reduction in CIC protein levels compared with normal liver tissues, whereas the CIC mRNA level was not correlated with the CIC protein level in each tissue sample (Fig. 1D), suggesting the posttranscriptional regulation of CIC expression in HCC cells. Taken together, these data indicate an association of CIC levels with HCC progression.

CIC SUPPRESSES HCC PROGRESSION

To test whether CIC has a suppressive function in HCC progression, we examined the effect of CIC overexpression on HCC cell growth and invasion. We chose SK-HEP-1 and MHCC-97H cells, which are highly metastatic and aggressive HCC cell lines,37, 38 for the experiments because they express relatively low levels of CIC compared with other HCC cell lines (Supporting Fig. S1). Forced expression of either the short or the long form of CIC suppressed cell proliferation, invasion, and migration in both cell lines (Fig. 2A-D). Then, we tested whether suppression of CIC expression has the opposite effects. HCC cells that stably express short hairpin RNA against CIC (shCIC) had increased proliferation rate and invasive and migratory activity compared with control HCC cells (Fig. 2E-H). CRISPR-Cas9-mediated knockout of CIC also promoted HCC cell proliferation and invasion (Supporting Fig. S2). We confirmed these results in vivo using xenograft mouse models. Control and shCIC-expressing SK-HEP-1 or MHCC-97H cells were subcutaneously injected into either posterior flank of the same nude mice, respectively, and the tumor volume was measured every week. The CIC-deficient HCC cells grew more rapidly and formed larger tumor mass than control cells (Fig. 3A). To compare metastatic activity between control and CIC-deficient HCC cells, the cells were intravenously injected into nude mice, followed by quantification of green fluorescent protein signal, which is expressed from short hairpin RNA–expressing lentiviral vectors (pGIPZ), in lung tissues. The CIC-deficient HCC cells had a higher degree and frequency of metastasis to lung than control cells (Fig. 3B,C). Taken together, these findings indicate that CIC functions as a negative regulator in HCC progression.

INCREASED LUNG METASTASIS AND LETHALITY IN Cic-DEFICIENT MICE TREATED WITH DEN

To better understand the in vivo effect of CIC deficiency on HCC progression, we generated mice with a specific deletion of Cic alleles in hepatocytes (Cic^f/f;Alb-Cre, Cic liver-specific knockout [LKO]) (Fig. 4A) and induced liver cancer in these mice by treatment with DEN. Wild-type (WT; Cic^f/f) and Cic LKO mice were intraperitoneally injected with DEN at 2 weeks of age and subjected to analyses of tumorigenesis, lung metastasis, and viability (Fig. 4B). Tumor formation on liver tissues was comparable between WT and Cic LKO mice at 8 months of age (Fig. 4C). However, lung metastasis was substantially increased in Cic LKO mice at 15 months of age (Fig. 4D). Moreover, about 30% of Cic LKO mice died after 1 year of age, whereas none of the WT mice did (Fig. 4E). These results demonstrate that reduction in CIC expression can critically contribute to the promotion of HCC progression, which is consistent with the finding that the survival rate was decreased in HCC patients with low levels of CIC (Fig. 1B).

ETV4 IS A CRITICAL CIC TARGET THAT PROMOTES HCC PROGRESSION

Many studies have shown that PEA3 group genes, which include ETV1, ETV4, and ETV5, are direct target genes of CIC14, 19, 25, 26 and that overexpression of these genes promotes proliferation and invasion of various types of cancer cell.33 However, the role of PEA3 group transcription factors in HCC progression has not yet been comprehensively understood. Therefore, we analyzed expression profiles of PEA3 group genes and association of these gene expression levels with lethality in patients with HCC using the TCGA database (Supporting Tables S1 and S4). Among three genes, only ETV4 levels were increased in all stages of HCC cells with statistical significance compared with normal liver cells (Fig. 5A). On the other hand, overall survival rates of HCC patients were inversely correlated with the levels of all PEA3 group genes (Fig. 5B), suggesting that PEA3 group transcription factors might also function as a tumor promoter in the context of HCC progression.

Next, we examined whether CIC regulates expression of PEA3 group genes in HCC cells. The levels of PEA3 group genes in control and shCIC-expressing HCC cells were determined by quantitative RT-PCR analysis (Fig. 5C). Among three genes, ETV4 levels were most significantly up-regulated in three different HCC cell lines with CIC RNA interference (Fig. 5C). Knockdown of CIC did not increase levels of PEA3 group genes in HepG2 cells (Fig. 5C), suggesting that CIC differentially regulates its target gene expression in a cell type–dependent manner, which is consistent with previous findings.27, 35, 39 Furthermore, the up-regulation of ETV4 expression in CIC knockdown HCC cells was confirmed at the protein level (Fig. 5D). Increases in Etv4 levels were also most apparent in both normal liver and DEN-induced tumor tissues from Cic LKO mice compared with those in Etv1 and Etv5 (Fig. 5E-G). Overall, these data suggest that, among PEA3 group genes, ETV4 is a major target gene of CIC in hepatic cells.

Given that ETV4 had the highest relevance to HCC progression (Fig. 5A,B) and that ETV4 expression was most significantly regulated by CIC in HCC cells (Fig. 5C), we focused on the role of ETV4 in HCC progression. We first tested whether ETV4 has HCC-promoting activity. Overexpression of ETV4 indeed increased cell proliferation, invasion, and migration in HCC cells (Fig. 6A-D; Supporting Fig. S3). We next examined whether the increased cell proliferation, invasion, and migration in CIC-deficient HCC cells were due to derepression of ETV4. Knockdown of ETV4 completely blocked the CIC deficiency–mediated promotion of HCC progression (Fig. 6E-H), demonstrating that ETV4 is a key target gene of CIC in the regulation of HCC progression.

REGULATION OF MMP1 EXPRESSION BY THE CIC–ETV4 AXIS IN HCC CELLS

MMPs promote cancer progression in various ways, including destruction of extracellular matrix, activation of growth factors, suppression of apoptosis, and induction of angiogenesis.40 Therefore, MMPs are the principal mediators of cancer progression and are frequently used as biomarkers for various types of cancer. There are 23 members of the MMP family in humans. Studies have revealed that PEA3 group transcription factors activate expression of various MMP genes and that most MMP genes harbor ETS binding elements in their promoters.26, 41 To gain insight on which MMPs are critically involved in HCC progression, we analyzed the relevance of each MMP to HCC progression using the TCGA database (Supporting Tables S1 and S5), as we did for CIC and PEA3 group genes. Among 23 MMP genes, levels of MMP1, MMP9, MMP10, MMP11, MMP12, and MMP14 were significantly higher in HCC tissues than in normal liver tissues (Fig. 7A; Supporting Fig. S4). Analysis of survival rates revealed that levels of MMP1, MMP7, MMP10, MMP12, MMP16, and MMP26 were inversely correlated with survival rates of HCC patients with statistical significance (Fig. 7B; Supporting Fig. S5). Thus, these analyses identified MMP1, MMP10, and MMP12 as MMP genes strongly associated with the promotion of HCC progression.

Next, we investigated which of the selected MMP genes are regulated by the CIC–ETV4 axis in HCC cells. Among MMP1, MMP10, and MMP12, only MMP1 expression was significantly induced by ETV4 overexpression in both SK-HEP-1 and MHCC-97H HCC cell lines (Fig. 7C). We confirmed that ETV4 enhances MMP1 promoter activity by luciferase assay using an MMP1 promoter–containing reporter construct (Fig. 7D). We further examined the regulation of MMP1 expression by CIC in HCC cells. Knockdown of CIC significantly up-regulated levels of MMP1, but neither MMP10 nor MMP12, in SK-HEP-1 and MHCC-97H HCC cell lines (Fig. 7E). Consistent with this result, overexpression of CIC down-regulated MMP1 expression (Supporting Fig. S6). On the other hand, expression of MMP13, which belongs to the interstitial collagenase family as MMP1 does and shares high amino acid identity (86%) with MMP1,42 was not significantly affected by either ETV4 overexpression or CIC RNA interference in HCC cells (Supporting Fig. S7), suggesting that the CIC–ETV4 axis might selectively regulate expression of MMP1 rather than all members of the MMP subfamily with similar biochemical and pathological properties. Levels of Mmp1a, a mouse homolog of human MMP1, were also up-regulated in liver tumors from 15-month-old Cic LKO mice treated with DEN compared to those from WT controls, while levels of other Mmp genes including Mmp2, Mmp9, and Mmp12, which have been implicated in the promotion of HCC progression,43, 44 were comparable (Supporting Fig. S8). Overall, these results suggest that MMP1 might be a key effector MMP protein that functions downstream of the CIC–ETV4 axis in the context of HCC progression.

To directly address whether the increased expression of MMP1 in CIC-deficient HCC cells was due to derepression of ETV4, we examined levels of MMP1 in control and shCIC-expressing HCC cells treated with either control small interfering RNA (siRNA) or siRNA against ETV4. Knockdown of ETV4 certainly restored MMP1 expression to the normal level in CIC-deficient HCC cells (Fig. 7F). Taken together, these data suggest that MMP1 expression can be regulated by the CIC–ETV4 axis in the process of HCC progression.

CIC DEFICIENCY PROMOTES HCC PROGRESSION THROUGH MMP1 OVEREXPRESSION

Because we identified MMP1 as a critical downstream target gene of the CIC–ETV4 axis in HCC cells, we finally determined whether the induction of MMP1 expression contributed to the enhanced cancer progression in CIC-deficient HCC cells. We transfected control and shCIC-expressing HCC cells with either control siRNA or two different siRNAs against MMP1 and examined cell proliferation, invasion, and migration. Knockdown of MMP1 completely suppressed the increased cell proliferation and invasive and migratory activity in shCIC-expressing HCC cells (Fig. 8A-D), demonstrating that increased expression of MMP1 indeed critically contributed to the CIC deficiency–mediated promotion of HCC progression.