The Tumor Suppressor Gene Hypermethylated in Cancer 1 Is Transcriptionally Regulated by E2F1

HIC1 is a bona fide tumor suppressor as shown in Hic1^+/− mice that develop different spontaneous, age- and gender-specific malignant tumors (1). The HIC1 gene is located on the short arm of human chromosome 17 (17p13.3) telomeric to the p53 tumor suppressor gene. Chromosome 17 is frequently altered in human cancers and its allelic loss at the p53 locus is often combined with mutations of the remaining p53 allele. Interestingly, the 17p13.3 region is frequently deleted or hypermethylated in human cancers, such as breast cancer (2), colorectal cancer (3), and hepatocellular carcinomas (4, 5). Moreover, hypermethylation in this chromosomal region often leads to inhibition of HIC1 expression, and overexpression of HIC1 in different cancer cell lines decreases their clonogenic survival (6).

The HIC1 gene encodes a transcriptional repressor belonging to the broad complex, tramtrack, and bric à brac/poxviruses and zinc finger (BTB/POZ) transcription factor family. Currently, three transcriptionally repressed HIC1 targets are known: (a) the type III NAD⁺-dependent histone/protein deacetylase Silent Information Regulator 2a homologue1 (SIRT1) is involved in regulating cellular senescence and longevity (7); (b) the fibroblast growth factor binding protein (8) enhances FGF-mediated biochemical and biological events specifically during blood vessel growth; and (c) the proneural transcription factor Atonal Homolog 1 (Atoh1) is essential for cerebellar growth and development (9).

Previous studies have shown, that HIC1 plays a role in the p53-dependent apoptosis circuitry in response to DNA damage and p53 was identified as transcriptional inducer of HIC1. It has been reported that HIC1 indirectly modulates p53-dependent DNA-damage responses by transcriptional repression of the p53 deacetylase SIRT1 (7). We showed that HIC1 mRNA is induced by p53 on UV irradiation (10). These observations are supported by the fact that enforced HIC1 expression leads to growth arrest and reduced survival of glioblastoma, breast cancer, and adenocarcinoma cell lines (6).

In this article, we show that HIC1 is a new transcriptional target of the cell cycle and apoptosis regulator E2F1. E2F1 induces HIC1 via two E2F DNA binding sites within the TATA-box containing HIC1 P0 promoter. We further show that E2F1 binds to this promoter region in vivo and induces endogenous HIC1 mRNA expression. Furthermore, HIC1 expression is induced by E2F1 regardless of HIC1 P0 promoter hypermethylation. Moreover, using RNA interference to inhibit E2F1 expression, we show that E2F1 is necessary for HIC1 mRNA induction on etoposide treatment. Lastly, we show that endogenous E2F1 protein directly binds to the HIC1 promoter region in DNA damage responses to etoposide treatment.

To date, p53 is the only known transcriptional regulator of HIC1 (6, 10, 11). To find new transcriptional regulators of HIC1, we analyzed a 3.2-kb HIC1 genomic fragment upstream of exon 2 for putative transcription factor binding sites using MatInspector software (12). We found several putative E2F DNA binding sites spread over the whole genomic fragment analyzed (Fig. 1A, left). Cotransfection of the longest HIC1 promoter construct (pGL3.HIC1prm.1) that comprises all three HIC1 promoters (P2 GC-rich, P1 GC-rich, and P0 TATA) together with E2F1 into H1299 lung cancer cells led to a 20-fold induction of the HIC1 reporter construct compared with the empty pcDNA3.1 control vector (Fig. 1A, right). Analysis of HIC1 promoter deletion constructs (pGL3.HIC1prm2-pGL3.HIC1prm6), pointed to three putative E2F binding sites within the HIC1 P0 promoter located 24, 126, and 235 bp downstream of the TATA-box. Individual mutations identified two active E2F responsive elements. Mutation of either active E2F-site was sufficient to markedly decrease E2F1-induced HIC1 promoter activity. Additionally, mutations of both active E2F binding sites further decreased HIC1 promoter activity on E2F1 expression (Fig. 1B). We thus identified two new E2F responsive elements (ERE1 and ERE2). Interestingly, ERE1 and ERE2 are conserved among human, mouse, and rat genomic sequences, as shown by BLAST sequence comparisons (Fig. 1C).

To verify if the transcription factor E2F1 would bind in vivo to the identified putative E2F binding sites within the HIC1 P0 promoter, a chromatin immunoprecipitation (ChIP) assay in TIG3 cells stably expressing the estrogen receptor (ER) or a ER-E2F1 fusion protein was done. TIG3-ER-E2F1 cells were treated with 4-hydroxytamoxifen (OHT) to induce translocation of the ER-E2F1 fusion protein to the nucleus. The cells were then cross-linked and the DNA-protein complexes immunoprecipitated with an anti-E2F1 antibody or an unrelated anti-Flag antibody as negative control. The precipitated DNA was amplified by PCR using primers in the E2F1-responsive region. Significant enrichment of E2F1 on the HIC1 P0 promoter was seen (Fig. 2A). As a control for the binding of E2F1, its association with the E2F target gene, CDC6 (13), was measured (Fig. 2B).

Using promoter reporter and ChIP assays, we identified two E2F binding sites placed within the HIC1 P0 promoter region, which bind to E2F1 in vivo. We next asked whether activation of E2F1 in TIG3-ER-E2F1 fibroblasts would increase endogenous HIC1 mRNA levels. To this end, TIG3-ER-E2F1 fibroblast cells were incubated with 600 nmol/L OHT for 4 hours. Activation of E2F1 resulted in a 12-fold induction of endogenous HIC1 mRNA (Fig. 3A, left). Successful activation of E2F1 was confirmed by measuring mRNA expression of the E2F1 target gene Cyclin E1 (CCNE1; Fig. 3A, right; ref. 14) that was induced 17-fold. Similarly, activation of endogenous E2F in TIG3-ER-E1A fibroblasts resulted in a 6- and 4.5-fold induction of HIC1 and CCNE1 mRNA, respectively (Fig. 3B). Expression of E1A competes with endogenous E2F for binding to the retinoblastoma protein and thus derepresses E2F target genes.

Moreover, expression of E2F1 in Hep3B hepatocellular carcinoma cells using adenoviral vectors resulted in an 18-fold induction of endogenous HIC1 mRNA levels compared with Hep3B cells infected with a control adenovirus expressing the luciferase gene (Fig. 3C, left). We used p53-negative Hep3B cells to exclude that induction of HIC1 by E2F1 occurs indirectly via the E2F1-ARF-p53 pathway. Activation of CCNE1 mRNA after adenoviral expression of E2F1 in Hep3B cells was measured as a control (Fig. 3C, right).

Different reports indicate that the HIC1 promoter is epigenetically silenced in a wide variety of human cancers, as for example in primary human hepatocellular carcinomas (4, 5). Of note, it was shown that in most cell lines, dense methylation of the HIC1 P0 promoter was sufficient to completely silence HIC1 mRNA expresssion. Treatment of Hep3B hepatocellular carcinoma cells with 0.5 or 3.0 μmol/L 5′aza-2′deoxycytidine methyltransferase inhibitor for 2 days increased HIC1 mRNA 1.6- and 2.3-fold, respectively (Fig. 4A). With the technique of bisulfite genomic sequencing, we confirmed that the HIC1 P0 promoter is indeed hypermethylated in Hep3B cells including CpGs within the two E2F binding sites (Fig. 4B).

Chen et al. (7) showed that HIC1 plays a role in etoposide-induced DNA damage responses. Furthermore, it has been reported that E2F1 is stabilized in response to DNA damage and this stabilization would lead to the activation of a subset of E2F1 target genes (15, 16). Treating Hep3B cells with 40, 80, and 120 μmol/L etoposide for 24 hours leads to a 2.7-, 4.8-, and 8-fold induction of HIC1 mRNA levels, respectively (Fig. 5A). To examine whether HIC1 mRNA is induced via E2F1 in Hep3B cells on exposure to etoposide, we treated two different Hep3B cell lines expressing small hairpin (sh)RNAs targeting E2F1 and a nontargeting shRNA control cell line with 80 μmol/L etoposide. Inhibition of E2F1 in the two Hep3B E2F1 knock-down cell lines diminished induction of HIC1 mRNA on etoposide treatment by 40% and 60%, respectively, compared with a nontargeting shRNA control cell line. Efficient E2F1 knockdown in Hep3B cells was shown by Western blotting (Fig. 5B). To investigate if endogenous E2F1 directly activates HIC1 expression on etoposide treatment, we did a ChIP assay in Hep3B cells treated with 80 μmol/L etoposide for 24 hours. Cross-linked DNA-protein complexes were immunoprecipitated with an anti-E2F1 antibody, or with an unrelated anti-IgG antibody as negative or with an anti–acetyl-histone H3 antibody as positive control. Precipitated DNA was PCR amplified with primers in the E2F-responsive region. Significant enrichment of the endogenous E2F1 protein on the HIC1 P0 promoter was seen (Fig. 5C). As an E2F1 binding control, promoter association with CDC6 (13), a well-known E2F target, was measured.

These observations further support our finding that the tumor suppressor gene HIC1 is a direct transcriptional target of E2F1.

In this article, we identified the cell cycle and apoptosis gene E2F1 as a new transcriptional regulator of HIC1. E2F1 is a member of the E2F transcription factor family involved in different cellular processes such as cell cycle progression, DNA replication, oncogenic transformation, and in apoptosis responses. E2F1 can promote cell growth or induces apoptosis dependent on the amount of active E2F1 in the cell. For example, E2F1 is activated in response to oncogenic hyperproliferation and can induce apoptosis in a p53-dependent or p53-independent manner as part of a failsafe antiproliferative mechanism (17-19).

Our results show that E2F1 directly activates the transcription of HIC1 mRNA as shown with promoter deletion, E2F binding site mutation, and in vivo binding experiments. Individual mutations of the two functional E2F-binding sites, ERE1 and 2, in the HIC1 P0 promoter led to a 53% and 42% reduction of reporter activity, respectively. Mutating both functional E2F binding sites further reduced promoter activity. These results are comparable with results obtained from other E2F1 target genes such as CDC25A or DNMT1 (20, 21), but we cannot completely exclude that other E2F binding sites further upstream might contribute to HIC1 regulation.

Interestingly, Zhang et al. (22) recently published that HIC1 directly binds to the E2F1 promoter and inhibits its expression. The HIC1-mediated transcriptional inhibition of E2F1 and consequently of E2F-responsive genes led to growth suppression. The observation of HIC1-mediated E2F1 repression together with our results showing activation of HIC1 by E2F1 might point to a novel negative feedback mechanism.

Recent studies showed that ectopic expression of E2F1 can induce apoptosis in different tumors (23-25). Additionally, it was clearly shown that adenoviral expression of E2F1 results in growth suppression and increased responsiveness of tumor cells to chemotherapy (23, 26). Moreover, different studies suggest that overexpression of E2F1 could be a valuable tool for cancer therapy in tumor cells (reviewed in ref. 27). It was also shown that E2F1 exerts its apoptotic effects through p53-dependent and p53-independent mechanisms (17, 18). In addition, HIC1 expression leads to reduced survival of different cancer cell lines (6). We now show that expression of E2F1 in p53^−/− Hep3B cells using adenoviral vectors results in induction of endogenous HIC1 mRNA level. We speculate that E2F1 could exert its apoptotic function partially through the activation of HIC1, in a p53-independent manner. Moreover, HIC1 mRNA is inducible in Hep3B cells on E2F1 expression, although methylation analysis revealed that the HIC1 P0 promoter bearing the E2F binding sites is hypermethylated. This fact indicates that E2F1 induces HIC1 mRNA levels in Hep3B cells independent on the methylation status of the HIC1 P0 promoter. Importantly,hypermethylation of the P0 promoter is sufficient to silence HIC1 expression (28). In line with our methylation results, two studies measured HIC1 promoter hypermethylation in different morphologic grades of the liver and found increasing HIC1 promoter methylation from normal liver tissue, to precancerous liver tissue (showing chronic hepatitis or cirrhosis), to primary hepatocellular carcinoma, but no significant correlation between hypermethylation and HIC1 mRNA expression levels was seen (4, 5). These results indicate that mechanisms other than hypermethylation contribute to the low HIC1 expression seen in hepatocellular carcinoma.

We further show that HIC1 mRNA levels are induced in Hep3B cells on etoposide treatment. Previously, Chen et al. (7) pointed out a certain role of HIC1 in the p53-dependent apoptosis response to DNA-damage. The fact that E2F1 is involved in p53-dependent and p53-independent DNA-damage responses (29-32) raises the possibility that HIC1 mRNA induction is a direct E2F1 effect on etoposide induced DNA damage. Indeed, silencing of E2F1 with the help of two different lentivector delivered shRNAs diminished induction of HIC1 mRNA by 40% and 60%, respectively, in p53^−/− Hep3B cells on etoposide treatment. Further investigations revealed direct binding of endogenous E2F1 protein to the HIC1 P0 promoter after etoposide treatment of Hep3B cells. We show for the first time that E2F1 activates HIC1 expression on etoposide induced DNA damage in a p53-independent fashion.

Taken together, we identified HIC1 as a new transcriptional target of E2F1. We found two functional E2F responsive elements within the TATA-box containing HIC1 P0 promoter that mediate HIC1 transcription. Using Hep3B E2F1 knockdown cells, we further show, that E2F1 is required for etoposide-induced up-regulation of HIC1 expression. Additionally, we show that E2F1 directly binds to the HIC1 P0 promoter region on etoposide treatment. These findings indicate that HIC1 is involved in the p53-independent DNA-damage reponse of E2F1. Lastly, because HIC1 promoter hypermethylation cannot prevent induction of HIC1 by E2F1, HIC1 might represent an important effector of E2F1-mediated cancer therapy.

H1299 non–small lung cancer and Hep3B hepatocellular carcinoma cells were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH. Tamoxifen-inducible TIG3-ER-E2F1 and -E1A cells were generated as described elsewhere (33). Cells were maintained in RPMI 1640 or DMEM supplemented with 10% FCS, 50 U/mL penicillin, and 50 μg/mL streptomycin in a 5% air humified atmosphere at 37°C. OHT was dissolved in ethanol at 1mg/mL and stored protected from light at −20°C (Sigma-Aldrich). The human hepatocellular carcinoma cell line Hep3B was treated with 5′aza-2′deoxycytidine (Sigma-Aldrich) in DMSO for 2 d.

The HIC1 promoter deletion constructs pGL3.HIC1prm 1, 2, 5, and 6 were generated as described (10). HIC1prm.6 m1, 2, 3, and 1/2 constructs with mutated E2F-binding sites were generated using the QuikChange XL Site-Directed Mutagenesis kit (Stratagene) and the following oligonucleotides: m1, 5′-CGCGCCGGGCCCGGCCTTAAAAATCTGCCCGCCACGGCAGC-3′; m2, 5′-GGCCAGGGCGGCGCCAGGATATCCACCGCGCTCCCCTCCT'-3; m3, 5′-CCCTACTTGGGTAAAGTTCGATATCGGGCTCGAGATCTGCGATC'-3. The site-specific mutations were confirmed by DNA sequencing.

H1299 non–small cell lung cancer cells were transfected with 400 ng HIC1 promoter reporter, 400 ng E2F1 expression, and 10 ng of phRL-TK plasmid using Lipofectamine 2000 (Invitrogen). Cells were lysed 24 h after transfection and Luciferase activity was measured using the Dual-Luciferase Reporter Plasmid System (Promega). Results, expressed relative to a value of 1.0 for cells transfected with empty vector, are the means of 2 replications, and error bars represent SDs.

For ChIP assays, OHT-treated TIG3-ER and TIG3-ER-E2F1 cells or Hep3B cells treated for 24 h with 80 μmol/L etoposide, and an antibody against E2F1 (Sc-193) were used. An unrelated anti-Flag antibody (F3165; Sigma) or an unrelated anti-IgG antibody (Stratagene) were used as negative controls and an anti–acetyl-histone H3 antibody (Stratagene) as a positive control. ChIP assays were done and analyzed as described previously (34). Conventional and quantitative PCR were done using the following primers: HIC1 promoter, forward 5′-ACCGAGGGTTGACAGCCC-3′ and reverse 5′-AGTAGGGAGAAAGGGAGCCG-3′; CDC6 promoter, forward 5′-AAAGGCTCTGTGACTACAGCCA-3′ and reverse 5′-GATCCTTCTCACGTCTCTCACA-3′.

Total RNA was extracted using the RNeasy Mini kit (Qiagen), and total RNA was reverse transcribed using random primers (Roche Diagnostics) and M-MLV reverse transcriptase (Promega). PCR and fluorescence detection were done using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems).

For quantification of HIC1 mRNA, Taqman Gene Expression Assay Hs00359611_s1 (Applied Biosystems) or SYBR Green and the following primers fw 5′-CGACGACTACAAGAGCAGCAGC-3′ and rev 5′-CAGGTTGTCACCGAAGCTCTC-3′ were used. Cyclin E1 mRNA was quantified using Taqman Gene Expression Assay Hs01026536_m1 (Applied Biosystems) or SYBR Green and the following primers: fw 5′-TGCAGAGCTGTTGGATCTCTGTG-3′ and rev 5′-GGCCGAAGCAGCAAGTATACC-3′. For the housekeeping gene GAPDH, we used the following primers and probe: 5′-GAAGGTGAAGGTCGGAGT-3′ and 5′-GAAGATGGTGATGGGGATTT-3′ at 900 nmol/L and 5′-FAM-CAAGCTTCCCGTTCTCAGCC-TAMRA-3′ at 200 nmol/L. Primers and probe for the housekeeping genes ABL and HMBS have been described (10). All measurements were done in duplicates, and the arithmetic mean of the Ct-values was used for calculations: target gene mean Ct-values were normalized to the respective housekeeping gene, and then to the experimental control. Obtained values were exponentiated 2^−ΔΔCt to be expressed as n-fold changes in regulation compared with the experimental control (35).

Genomic DNA was extracted and bisulfite converted using the MethylampTM DNA modification kit (Epigentek). Primers for PCR amplification, cloning, and sequencing have been described elsewhere (36).

pLKO.1-puro lentiviral vectors expressing shRNAs targeting E2F1 (shE2F1_1: NM_005225.1-502s1c1 and shE2F1_2: NM_005225.1-948s1c1) and the nontargeting control shRNA vector (SHC002) were purchased from Sigma-Aldrich. These vectors contain a puromycin antibiotic resistance gene for selection of transduced mammalian cells. Lentivirus production and transduction were done as described (10, 37). AdCMV-E2F1 and AdCMV-Luc vectors were used as described (25).

Immunoblotting and protein extraction have been described previously (38, 39). Anti-E2F1 (sc-251; Santa Cruz Biotechnology) was used as primary and donkey anti-rabbit horseradish peroxidase–conjugated IgG (Amersham) as secondary antibody.

No potential conflicts of interest were disclosed.

We thank Deborah Shan and Gustav Arvidsson for expert technical support.