ZNF479 downregulates metallothionein-1 expression by regulating ASH2L and DNMT1 in hepatocellular carcinoma

Decreased expression of metallothionein-1 (MT-1) is associated with a poor prognosis in hepatocellular carcinoma (HCC). Here, we found that MT-1 expression was suppressed by 14-3-3ε, and MT-1 overexpression abolished 14-3-3ε-induced cell proliferation and tumor growth. We identified that 14-3-3ε induced expression of ZNF479, a novel potential transcriptional regulator by gene expression profiling and ZNF479 contributed to 14-3-3ε-suppressed MT-1 expression. ZNF479 induced the expression of DNMT1, UHRF1, and mixed-lineage leukemia (MLL) complex proteins (ASH2L and Menin), and increased tri-methylated histone H3 (H3K4me3) levels, but suppressed H3K4 (H3K4me2) di-methylation. ZNF479-suppressed MT-1 expression was restored by silencing of ASH2L and DNMT1. Furthermore, ZNF479 expression was higher in HCC tissues than that in the non-cancerous tissues. Expression analyses revealed a positive correlation between the expression of ZNF479 and DNMT1, UHRF1, ASH2L, and Menin, and an inverse correlation with that of ZNF479, ASH2L, Menin, and MT-1 isoforms. Moreover, correlations between the expression of ZNF479 and its downstream factors were more pronounced in HCC patients with hepatitis B. Here, we found that ZNF479 regulates MT-1 expression by modulating ASH2L in HCC. Approaches that target ZNF479/MLL complex/MT-1 or related epigenetic regulatory factors are potential therapeutic strategies for HCC.


Introduction
Metallothioneins (MTs) are a group of low molecular weight (~6 kDa) polypeptides that contain cysteine-rich sequences with a high affinity for heavy metals and oxidative stress 1,2 . The expression levels of MTs are induced by heavy metals, cytokines, and ROS. The MT-1 proteins function as scavengers to directly intercept heavy metals and reactive oxygen species that may damage cells or tissues 2 . MT-1 expression is increased in breast cancer, non-small-cell lung cancer, ovarian cancer, pancreatic carcinoma, renal carcinoma, and skin carcinoma, but decreased in colorectal carcinoma, gastric carcinoma, prostate cancer, and small-cell lung cancer 3 . MT-1 proteins are abundantly expressed in the liver, but are downregulated in hepatocellular carcinoma (HCC) 4,5 . Methylation of the MT-1M and MT-1G promoters is significantly increased in HCC patients, and this is positively correlated with tumor size and the incidence of metastases 6 . Moreover, overexpression of MT-1M could reduce cell proliferation and tumor growth in HCC 7 . Therefore, reduced expression of MT-1 is a potential diagnostic marker for HCC.
The stringent reduction of MT1 expression level in HCC raises the possibility that its promoters and other important cis-elements are epigenetically modified to achieve long-term repression, which can be regulated via specific histone modifications and DNA hypermethylation (Table. 1). The mixed-lineage leukemia (MLL) complex displays histone methyltransferase activity and helps to regulate transcription 8 . Absent, small, or homeotic discslike protein (ASH2L), a component of the MLL complex, functions as an oncoprotein 9 and overexpression of ASH2L stimulates the methyltransferase activity of the MLL complex 10 . ASH2L is essential for tri-methylation of H3K4 (H3K4me3) as lacking of ASH2L in the MLL complex reduces H3K4me3 levels but increases H3K4 dimethylation (H3K4me2) 11 . In addition, MLL complex is recruited to the chromatin via interacts with a histonebinding factor called Menin 8,12 . Expression of Menin is significantly associated with liver fibrogenesis and is correlated with that of ASH2L in HCC 13,14 . Moreover, increased Menin expression has been found to promote hepatocellular carcinogenesis 13 . Therefore, it will be interesting to know how MLL complex and associated factors are involved in the carcinogenesis of HCC.
DNA methyltransferases (DNMTs) contribute to gene regulation in mammalian cells by epigenetic modification. DNMT1 and DNMT3a are involved in the early stages of hepatocarcinogenesis 15 , and increased expression of DNMT1 is significantly correlated with malignancy and poor prognosis of human HCC 16 . It has been shown that treatment with 5-azacytidine (5-AzaC) decreased the association of DNMT1 with the MT-1 promoter, thereby inducing MT-1 expression and restoring MT-1 promoter activity in hepatoma cells 17,18 . Additionally, DNMT1 is a target of microRNA-140 (miR-140), and it has been reported that miR-140 -/mice show increased DNMT1 expression accompanied by reduced MT-1 expression 19 . These results suggest that DNMT1 plays an important role on regulating DNA methylation and MT1 promoter activity. In addition, DNMT1 binds to the CpG islands of the cis-elements through interacting with ubiquitin-like with PHD and ring finger domains 1 (UHRF1, also known as NP95 in mice and ICBP90 in humans), which   [20][21][22][23] . The expression of UHRF1 is increased in HCC and is associated with a poor prognosis as it promotes cell proliferation and metastasis 24 . However, whether UHRF1 expression associates with MT-1 expression has never been elucidated. The Krüppel-associated box (KRAB) domain containing zinc finger proteins (KRAB-ZFPs) are the largest family of transcription repressors in mammals [25][26][27] . Although the functions of most KRAB-ZFPs are unknown, some have been shown to repress transposable elements by interacting with co-repressor tripartite motif-containing 28 (TRIM28; also known as KRAB-associated protein-1 [KAP-1]) 28 . According to sequence homology, zinc finger protein 479 (ZNF479, also known as HKr19) is a potent KRAB-ZFP 29 . ZNF479 was first cloned from a human testis cDNA library 29 . However, the functions or regulation of ZNF479 has never been investigated.
14-3-3 proteins are implicated in regulating multiple cellular and physiological functions 30 . 14-3-3 proteins are overexpressed in HCC and regulate tumor progression [31][32][33][34][35][36] . In this work, we reveal that 14-3-3ε suppressed the expression of MT-1 via inducing ZNF479 expression. siRNA silencing of ZNF479 significantly induced MT-1 expression and attenuated 14-3-3ε-induced cell proliferation and tumor growth. Moreover, the suppression of MT-1 by ZNF479 was mediated by the induction of ASH2L and DNMT1. Results from expression analyses in clinical HCC tissues revealed that expression of ZNF479 is positively correlated with that of DNMT1, UHRF1, ASH2L, and Menin but inversely correlated with MT-1. Finally, we found that correlations between the expression of ZNF479 and its downstream factors are more pronounced in hepatitis B virus (HBV) positive patients. Thus, ZNF479 may play an important role in suppressing MT-1 expression and modulating HCC tumor progression by regulating components of the MLL complex and epigenetic modification.

PDTC induces MT-1 expression, suppresses cell proliferation and tumor growth
While investigating the signaling pathways involved in regulating MT-1 expression, we found that PDTC dosedependently and abundantly induced MT-1 expression (Fig. 1g), and significantly suppressed cell proliferation (Fig. 1h). We also found that peri-tumoral injection of Fig. 1 14-3-3ε-suppressed MT-1 expression contributes to cell proliferation and tumor growth. a Relative expression of MT-1 isoforms in Huh-7 cells was determined by western blot and qPCR analysis in 14-3-3ε-overexpressing stable cells (14-3-3ε). b, c 14-3-3ε cells were transiently transfected with MT-1 overexpression vectors. Expression of cyclin D and MT-1 was determined by western blot analysis and cell proliferation by MTT assay. d Tumor growth and weight in an in vivo xenograft mouse model of 14-3-3ε overexpression versus control cells, and e MT-1M overexpression versus control vector in 14-3-3ε cells. Tumor growth was examined by calculating tumor volume every week and tumor weight at day 35, N = 5 in each group. f 14-3-3ε-overexpressing cells were treated with 10 μg/μl U0126, 10 μM rapamycin, 20 μM LY294002, or DMSO for 24 h. Relative expression of MT-1 was determined by western blot and qPCR analysis. g 14-3-3ε-overexpressing cells were treated with 10-30 μM PDTC for 24 h. Expression of MT-1 was determined by western blot and qPCR analysis (30 μM of PDTC). h 14-3-3ε-overexpressing cells were treated with various concentrations of PDTC for 24-72 h. Cell proliferation was examined by MTT assay. i Nude mice were subcutaneously injected with 2 × 10 6 14-3-3ε-overexpressing cells for 1 week, followed by peri-tumoral injection with vehicle control (PBS) or PDTC (50 and 100 mg/kg) every 2 days for 4 weeks. Tumor growth was examined by tumor volume and tumor weight, N = 5 in each group. j Control and 14-3-3ε-overexpressing cells were transiently transfected with scramble or p65 siRNA for 24 h, then treated with 30 μM PDTC for 24 h. Expression of p65 and MT-1 was examined by western blot analysis. Gene expression of qPCR analysis was normalized to both control cells as well as GAPDH. Actin was used as the loading control for western blotting analysis. Scale bars: mean ± SD. *P < 0.05; **P < 0.01 PDTC into an in vivo xenograft mouse model significantly reduced tumor volume and weight (Fig. 1i). Because PDTC is commonly used as an NFκB inhibitor, we investigated whether NFκB is involved in regulating MT-1 expression. We found higher p65 expression in 14-3-3εoverexpressing cells than in control cells (Fig. 1j). However, siRNA knockdown of p65 exhibited no effect on MT-1 expression (Fig. 1j).
To further investigate the expression of ZNF479 and its regulated factors, we examined changes in expression of ZNF479, DNMT1, UHRF1, ASH2L, Menin, MT-1M, MT-1G, and MT-1H in a clinical HCC cohort (including 50 HBV, 50 HCV, and 50 NBNC) samples. We confirmed that the expression of MT-1 isoforms were decreased in HCC tumors compared to normal tissues (Fig. 5a). The expression of ZNF479 was significantly higher in HCC than that in the normal tissues (Fig. 5a). Unexpectedly, the overall expression of DNMT1, UHRF1, ASH2L, and Menin was lower in HCC than that in the normal tissues (Fig. 5a). Intriguingly, we found a positive correlation between the expression of ZNF479 and DNMT1, UHRF1, ASH2L, and Menin, but an inverse correlation with MT- Expression of MT-1 was determined by western blot analysis. e 14-3-3ε-overexpressing cells were transfected with or without ZNF479 siRNA. Expression of ZNF479 and MT-1 was determined by western blot and qPCR analysis. f 14-3-3ε-overexpressing cells were transfected with ZNF479 siRNA. Cell proliferation was examined by MTT assay. g Huh-7 cells were transfected with a ZNF479 overexpression vector. Cell proliferation was examined by MTT and h colony formation assays. i Cell proliferation was examined by MTT assay in stable cells with ZNF479 shRNA treatment. j Nude mice were subcutaneously injected with 2 × 10 6 cells with ZNF479 shRNA (N = 9) and scramble control (N = 8). Tumor growth was examined by tumor volume and tumor weight. Gene expression of qPCR analysis was normalized to both control cells as well as GAPDH. Actin was used as the loading control for western blotting analysis. Scale bars: mean ± SD. *P < 0.05; **P < 0.01 1M, MT-1G, and MT-1H when analyzed in individual HCC samples (Fig. 5b). The expression of MT-1 isoforms was inversely correlated with ASH2L and Menin, but not significantly correlated with DNMT1 or UHRF1 (Fig. 5c). Finally, we found that the correlations with ZNF479 expression were more prominent in HCC patients in the HBV group than in the HCV and NBNC groups (Fig. 5d).
To further elucidate the expression correlation in patients of HBV group, we examined the protein levels of UHRF1, DNMT1, ASH2L, Menin, ZNF479, and MT-1 by western blotting analysis in HCC (29 samples) with HBV (Fig. S11). The relative protein levels (normalized with GAPDH) of ZNF479, ASH2L, UHRF1, and Menin were significantly higher in HCC than that in the normal tissues (Fig. S12). Although it is not statistically significant, there is a trend that protein levels of DNMT1 was higher but MT-1 was lower in HCC than normal tissues (Fig. S12).
The expression of ASH2L is required for regulating H3K4 methylation, and siRNA knockdown of ASH2L reduces the level of H3K4me3 13 . Results from an earlier study indicated that ASH2L is expressed in 84.8% of HCC cases, but high levels of H3K4me2 are rarely found 38 . These results further confirm that ASH2L expression contributes to the modification of histone methylation in HCC. It has been reported that eliminating ASH2L reduces H3K4me3 but induces H3K4me2 levels 11 . In this study, we found that expression of MT-1 was inversely correlated with H3K4me3. We further examined the binding capacity of H3K4me3 on the MT-1 promoter in ZNF479-overexpressing cells by ChIP assay. We found that overexpression of ZNF479 only slightly influenced the binding capacity of H3K4me2 on the promoters of MT-1M, MT-1G, and MT-1H (Fig. S13). These results suggest that H3K4me3 modulates MT1 expression may mediate through an indirect regulation.
The MLL complex has methyltransferase activity and interacts with histone deacetylases and polycomb proteins, thereby suppressing the expression of HOX genes 39 . SUMO-specific isopeptidase is associated with the MLL complex, enhances the recruitment of ASH2L and Menin to the complex, and consequently turns on the transcriptional activity of HOX genes 40 . In this study, we show for the first time that ASH2L and Menin are upregulated by the 14-3-3ε/ZNF479 axis (Fig. 3b-e). As well, siRNA knockdown of ASH2L significantly induces MT-1 expression (Fig. 3f). MT-1 may be epigenetically suppressed by ZNF479 by inducing ASH2L expression.
Further investigation is needed to elucidate how MT-1 is regulated by ASH2L.
Methylation of histone at the lysine residues plays a crucial role in the epigenetic regulation of gene expression. Although converted expression of H3K4me2 and H3K4me3 are modulated by ASH2L, both of H3K4me2 and H3K4me3 are considered as active factors in the promoter regions 41,42 . In addition to the methylation of H3K4, the symmetric di-methylation of histone H3 at arginine 2 (H3R2) was identified as a histone marker that supports euchromatin maintenance 41 . Results from several studies have indicated that the regulation of H3R2 and H3K4 methylation is mutually exclusive 42,43 . Here, we examined whether the expression of 14-3-3ε and ZNF479 affect di-methylation of H3R2 (H3R2me2). H3R2me2 was downregulated in 14-3-3ε-and ZNF479-overexpressing cells (Fig. S14a and S14b). In contrast, ZNF479 siRNA significantly induced H3R2me2 (Fig. S14c). Our findings of the regulation of histone H3 methylation echo the . Scale bars: mean ± SD. **P < 0.01. c Huh-7 cells were transfected with DNMT1 siRNA for 48 h. Expression of DNMT1 and MT-1 was determined by western blot analysis. Gene expression of qPCR analysis was normalized to both control cells as well as GAPDH. Actin was used as the loading control for western blotting analysis previous hypothesis of antagonism between ASH2Lregulated H3K4 and H3R2 methylation 42,43 .
KRAB-ZFPs regulate epigenetic modification by recruiting KAP-1. KAP-1 acts as a scaffold protein in the silencing complex consisting of histone methyltransferase SETDP1, nucleosome remodeling and deacetylation proteins, heterochromatin protein 1, and DNMTs, thereby impairing transcriptional activation 27 . Here we found that ZNF479 induces ASH2L, Menin, DNMT1, and UHRF1 expression (Figs. 3 and 4). Since ZNF479 contains a KRAB domain that is mostly known to induce transcriptional repression, it seems likely that this effect is indirect. How ZNF479 regulates the expression of DNMT1, UHRF1, ASH2L, Menin, or other MLL-complex proteins remains unclear and needs further investigation.
An earlier study demonstrated that PDTC synergizes with diethyl dithiocarbamate to augment MT-1 expression 44 . Here, we found that PDTC abundantly induced MT-1 expression (Fig. 1g) via an NFκB-independent mechanism (Fig. 1j). We identified that ZNF479 is a potential transcriptional regulator that is downregulated by PDTC (Fig. 2b and Table S1). To investigate whether MT-1 has a role in regulating ZNF479 level through negative feedback, we performed experiments of MT-1 overexpression and examined the expression level of ZNF479. Overexpression of MT-1M, MT-1G, and MT-1H had no significant effect on modulating ZNF479 expression (Fig. S15). Thus, the molecular regulation remains unclear and further investigation is needed to elucidate how PDTC suppresses ZNF479 expression. The function and expression of ZNF479 have never been investigated. Here, we found that the expression level of ZNF479 was significantly higher in HCC (Fig. 5a). Our clinical analysis revealed that expression changes of ZNF479 were positively correlated with DNMT1, UHFR1, ASH2L, and Menin, but negatively correlate with MT-1 (Fig. 5b). Moreover, the expression of MT-1 isoforms was significantly negatively correlated with ASH2L and Menin (Fig. 5c). These results suggest that MLL complex has a more predominant role in regulating MT-1 expression. Interestingly, the correlated expression of ZNF479 with other downstream factors is more prominent in the HBV positive group (Fig. 5d). These results suggest that ASH2L and Menin play an important role in ZNF479downregulated MT-1 expression. Further investigation is needed to elucidate the roles of ZNF479 and the MLL complex in HCC tumorigenesis and tumor progression.

Western blot analysis
Protein expression was analyzed by western blotting analysis described previously 36 . In brief, cells were harvested and protein concentrations were determined using a Bio-Rad Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA, USA). Lysates (30 µg) were separated by SDS-PAGE, then the resolved proteins were transferred to PVDF membranes (Millipore, Bedford, MA, USA). Incubation with primary antibodies (listed in Table S4) was performed overnight at 4 ℃. After washing with PBS-T, membranes were incubated with horseradish peroxidaseconjugated secondary antibodies. The protein signals were determined by enhanced chemiluminescence (Per-kinElmer, Shelton, CT). Quantification of protein expression was determined by western blot and quantified by ImageJ 1.43 u software Cell proliferation analysis and anchorage-independent growth assay Cell proliferation was measured using an MTT assay (Sigma-Aldrich, USA). Absorbance at 570 nm was measured with a reference wavelength of 690 nm. Anchorage-independent growth was assessed by a soft- Fig. 6 Proposed model in this study for how ZNF479 suppresses MT-1 expression by regulating ASH2L and DNMT1 in HCC agar assay. Six-well plates were coated with 2 ml 10% complete medium containing 0.8% low-melting agarose (Lonza, Rockland, ME), then seeded with 5 × 10 3 cells per well. After 3 weeks, cells were stained with crystal violet and colony numbers were counted.

Quantitative real-time PCR (qPCR)
Total RNA was extracted using RNAzol RT (RN190, Molecular Research Center, USA) and cDNAs were synthesized using the PrimeScript RT reagent kit (TAKARA, Japan). qPCR analysis involved use of SYBR Green (Kapabiosystem, Woburn, MA, USA) after probing with specific oligonucleotide primers (Table S5). Applied Biosystems Relative Quantification Manager Software v1.2 was used to analyze the relative gene expression by the comparative cycle threshold (Ct) method. Gene expression was normalized to that of glyceraldehyde-3phosphate dehydrogenase (GAPDH).

Tumor xenograft experiments
The protocol of this study was performed in accordance with the guidelines and regulations approved by the Institutional Animal Care and Use Committee of the National Health Research Institutes. Briefly, 8week-old BALB/c nu/nu nude mice were purchased from the National Laboratory Animal Center, National Health Research Institutes of Taiwan and housed in microisolator cages in a specific pathogen-free facility. A total of 2 × 10 6 14-3-3ε overexpression cells or 4 × 10 6 MT-1M transfected cells were subcutaneously injected into the right flank of nude mice 32 . For pyrrolidine dithiocarbamate (PDTC) treatment, mice (that received 2 × 10 6 14-3-3ε cells for 1 week) were peri-tumorally injected with vehicle control (PBS) or PDTC (50 and 100 mg/kg) 46 every 2 days for 28 days. The tumor volume was determined by sequential caliper measurements of length (L) and width (W) and calculated as LW 2 /2. Mice were sacrificed and tumor weight was measured after cultivation for 5 weeks.
Small-hairpin RNA (shRNA) xenograft experiment pLKO-TRC005-derived ZNF479 small-hairpin RNAs (shRNA) were purchased from the National RNAi Core Facility, Taiwan (target sequences listed in Table S6). shRNAs were transfected into Huh-7 cells, then stabilized with 2 μg/ml puromycin. The knockdown efficiency of the shRNAs was examined by western blot analysis of ZNF479 expression. For the tumor xenograft model, 2 × 10 6 stable cells (ZNF479 shRNA: TRCN0000239328; control: ASN0000000003) were injected into 8-week-old nude mice. Tumor volume was determined every week for 5 weeks, and tumor weight was measured after mice were sacrificed.

Microarray analysis
Gene expression profile with PDTC treatment in 14-3-3ε-overexpressing stable cells was examined by microarray analysis. RNA samples were analyzed using the Affymetrix Human Gene 1.0 ST array (Affymetrix Inc., Santa Clara, CA, USA) according to the manufacturer's recommendations.
Clinical specimens mRNA expression levels were assessed in 300 tissue RNA extractions (including normal liver and HCC) from HCC patients. Patient samples were divided into three groups: 50 HBV, Hepatitis B; 50 HCV, Hepatitis C; and 50 NBNC, neither Hepatitis B nor C. Expression levels of ZNF479, DNMT1, UHRF1, ASH2L, Menin, MT-1M, MT-1G, and MT-1H were examined by qPCR analysis.

Chromatin immunoprecipitation (ChIP)
Interaction of H3K4me3 in the MT-1 promoters was examined using Chromatin Immunoprecipitation Kits (17-10086, Millipore, CA, USA). Briefly, 2 × 10 6 cells were used for each ChIP. DNA-protein complex was crosslinked with 1% formaldehyde (Sigma-Aldrich, USA) for 10 min and then washed with PBS. Quenched cells with 1.25 mM Glycine for 5 min. Cells were collected and lysed in lysis buffer containing protease inhibitor cocktail (Sigma-Aldrich, USA). Cross-linked DNA was sheared to 1000-200 bp in length and immunoprecipitated with 1 μg of anti-H3K4me3 or normal rabbit IgG in 4 ℃ for overnight. Proteins were degraded by proteinase K and genomic DNA was purified using spin columns and eluted in 50 μl of water. Primer sequences used for PCR were listed in Table S7.

Statistical analysis
For clinical samples, changes in gene expression in tumors were determined by assessing the differences between normal and tumor tissues, calculated through the log of mRNA expression in tumor tissue minus normal tissue. These results were then analyzed by a one-sample Student t-test with zero as the test value. A one-way ANOVA test was used when comparing gene expression changes in tumors between different patient groups. To compare expression levels of different genes in the tumors, a Pearson's correlation co-efficient (r) was calculated, and a linear regression formula was provided if 2axis dot plots were drawn. Two-side p < 0.05 is considered as statistically significant. All statistical analyses were performed with SPSS software (version 17.0).