Epigenetic DNA Methylation of Antioxidative Stress Regulator NRF2 in Human Prostate Cancer

In the United States, prostate cancer (CaP) is the most commonly diagnosed non–skin cancer and the second leading cause of cancer-related deaths (1). In 2014, it is estimated that 233,000 men will be diagnosed with the disease and 29,480 will die (2). Prostate cancer is one of the most complicated human tumors, and similar to many other malignancies, it arises from progressive genetic and epigenetic alterations, which rapidly change these tumors from a clinically benign state to a malignant one (3, 4). Epigenetic alterations can largely contribute to the malignant transformation and progression of prostate cancer through the deregulation of many genes involved in critical cellular processes, including DNA-damage repair (e.g., GSTP1, GSTM1, and MGMT), hormonal responses (e.g., AR, ERα, ERβ, and RARβ), tumor suppression (e.g., KAI1, factor-β, and DAB2IP), tumor-cell invasion/metastasis (e.g., CDH1 and CD44), apoptosis (e.g., SLC18A2 and TNFRSF10C), and cell-cycle control (e.g., CDKN2A/p16 and RASSF1A; refs. 5, 6). Therefore, the deregulation of the antioxidant defense system has been gaining increased attention because it promotes toxicity and the neoplastic progression of prostate cancer (4, 7, 8).

Erythroid 2p45 (NF-E2)–related factor 2 (NRF2) is a basic-region leucine zipper (bZIP) transcription factor that regulates the expression of many phase II detoxifying/antioxidant enzymes, such as glutathione S-transferase (GST), UDP-glucuronosyltransferase (UGT), heme oxygenase-1 (HO-1), NADP(H):quinone oxidoreductase (NQO), glutamate cysteine ligase (GCL), and γ-glutamylcysteine synthetase (γGCS), by binding in combination with small Maf proteins to antioxidant response elements (ARE) in promoter regions (9). Thus, NRF2 is cytoprotective by reducing the toxicity of reactive intermediates, which protects against oxidative or electrophilic challenges and maintains cellular chemical homeostasis (10, 11). In vivo, NRF2 is dispensable for mouse growth and development (12). Accordingly, Nrf2-deficient (Nrf2^−/−) mice have been demonstrated to markedly exhibit lower expression levels of cellular defense genes in various tissues (4, 13). These mice are inherently more susceptible to oxidative stress-induced diseases and chemically induced DNA damage, which increases their risk of developing certain types of cancer (e.g., stomach, colorectal, and skin) compared with wild-type mice (14–19).

Diverse studies have shown that the levels of NRF2 and members of the GST mu family are decreased in human prostate cancer (20). Recently, the expression of NRF2 in prostate tumors from TRAMP mice was shown to be suppressed epigenetically by promoter CpG methylation and histone modifications in association with MBD2 (8). Interestingly, the treatment of TRAMP C1 cells with the DNA methyltransferase (DNMT) inhibitor 5-aza-2′-deoxycytidine (5-aza) and the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) restored NRF2 expression, representing a promising approach to treating cancer by reversing epigenetic modifications and increasing the expression of NRF2 and its downstream antioxidant and detoxification enzymes (8, 21). Similarly, absent or reduced GSTP1 expression has also been reported in high-grade prostate intraepithelial neoplasias, and methylation of its promoter has been the most frequently detected epigenetic alteration, which occurs in more than 90% of cancer samples and approximately 70% of prostate intraepithelial neoplasia samples (22). In contrast, the loss of GSTP1 is rarely detected in normal prostate or benign prostatic hyperplasia (BPH) tissues (5). Furthermore, the expression and activity of SOD, catalase, and GPx are decreased in prostate cancer tissues, plasma, and erythrocytes (23, 24). Taken together, these studies confirm the pivotal role of NRF2 and its target genes in controlling oxidative stress and chronic inflammatory processes in prostate cancer and the epigenetic mechanisms involved in its regulation.

All the enzymes used in this study were obtained from New England Biolabs Inc. The Dual-Luciferase Assay system and the luciferase reporter vectors pGL 4.75 with cytomegalovirus (CMV) promoter and pGL 4.15 were obtained from Promega. All other chemicals, including dimethyl sulfoxide (DMSO), 5-aza, and TSA, were purchased from Sigma. LNCaP (androgen-dependent prostate cancer cell line from the American Type Culture Collection) cells were maintained in RPMI-1640 with 10% fetal bovine serum (FBS; Gibco) and grown at 37°C in a humidified 5% CO₂ atmosphere. Cells were cytogenetically tested and authenticated before being frozen. Cell line was used no longer than 6 months after resuscitation from freezing. The LNCaP cells were plated on 10-cm plates for 24 hours and then treated with 0.1% DMSO or 2.5 μmol/L 5-aza with 1% FBS-containing medium. The medium was changed every 2 days. For the 5-aza and TSA combination treatment, 500 nmol/L TSA was added to the 5-aza–containing medium on the day 6 and cultured for another 24 hours. The cells were then harvested for DNA, RNA, and protein analyses.

BPH samples (n = 9) were obtained from patients treated for lower urinary tract symptoms by transurethral prostatectomy. Androgen-stimulated prostate cancer (AS-CaP) samples (n = 7) were obtained from patients treated for prostate cancer by prostatectomy. Androgen-deprivation therapy recurrent prostate cancer (ADT-RCaP) tumor samples (n = 11) were obtained by transurethral resection from patients who presented urinary retention from recurrent prostate cancer during ADT-RCaP (25). All samples were obtained from the Roswell Park Cancer Institute (Buffalo, NY). AS-CaP and BPH were enriched to >70% epithelial cells using the standard microdissection of 20-μm frozen step sections adjacent to 8-μm sections identified by H&E staining, which contained >50% epithelial cells as previously described by Berthon and colleagues (26). ADT-RCaP did not require microdissection as it was, on average, composed of 92% malignant cells (25).

MassARRAY quantitative methylation analysis (MAQMA) was performed to interrogate the methylation levels of the NRF2 promoter (Supplementary Fig. S1) in human prostate tissue samples using the MassARRAY Compact system and EpiTYPER software (Sequenom) as previously described (27). Genomic DNA was isolated using the DNeasy Tissue Kit (Qiagen) and subjected to bisulfite conversion. The bisulfite conversion was performed using 750 ng of genomic DNA with the EZ DNA Methylation Gold Kit (Zymo Research Corp.) following the manufacturer's instructions. The CpG sites identified in the 5′-flanking region of human NRF2 gene spanning from position −1552 to +1091 were analyzed using eight locus-specific amplicons (Fig. 1A). The primer sequences used to amplify the converted DNA are described in Supplementary Table S1.

Genomic DNA and sodium bisulfite converted genomic DNA from prostate tissue samples and DMSO- or 5-aza/TSA–treated LNCaP cells were obtained using the same procedures as described above. The converted DNA was amplified by PCR using Platinum PCR SuperMix (Invitrogen) with three specific sets of bisulfite genomic sequencing (BGS) primers spanning three differentially methylated CpG sites from −1530 to −1143, with the translational start site (TSS) referenced as +1 (Supplementary Table S1). The primers were designed using MethPrimer (28). The following PCR amplification conditions were used: 3 minutes at 94°C; 30 seconds at 94°C, 45 seconds at 70/55°C and 1 minute at 72°C for 15 cycles; 30 seconds at 94°C, 45 seconds at 60°C and 1 minute at 72°C for 25 cycles; and 5 minutes at 72°C for 1 cycle. The PCR products were cloned into the pCR4 TOPO vector using a TOPO TA Cloning Kit (Invitrogen). Plasmid DNA from at least 10 colonies from each treatment was prepared using the QIAprep Spin Miniprep Kit (Qiagen) and sequenced (Genwiz; ref. 29).

Immunohistochemical (IHC) staining was performed on a high-density prostate cancer tissue microarray (TMA) containing 27 tissue cores of normal prostate tissue and 160 tissue cores of prostate cancer tissue including adjacent tissue (hyperplasia) and adenocarcinoma tissue with Gleason score from 3–9 from BioMax Company. IHC staining was performed as described previously (30), using primary antibodies directed against NRF2 (1:100 dilution). The results of the IHC analysis (accumulated immunostaining intensity and percentage of positive cells) were determined by using an IHC Image Analysis algorithm to detect both cytoplasmic and nuclei staining using an Aperio ScanScopeR GL system (Aperio Technologies). The NRF2 immunostaining results were scored according to the accumulated intensity as 0, negative; 1+, weak; 2+, moderate; or 3+, strong. All images were quality controlled for accuracy of the tissue recognition software, in addition to accuracy of the image analysis algorithm.

The genomic sequence of the human NRF2 promoter region was retrieved from the human NCBI genome database. Two human NRF2 promoter segments, −1526-1 and −1169-1 (with the TSS referred to as position +1), were amplified from human genomic DNA isolated from normal human prostate cells using the following primers: 1526 forward, 5′-GGTACCACCTAGAGAAAGTAAGCTCTGC-3′; 1169 forward, 5′-GGTACCCATACTTGGAAGTAACAAGGAG-3′; and a common reverse primer, 5′-CTCGAGATGAGCTGTGGACCGTGTGTT-3′. The PCR products were cloned into the pCR4 TOPO vector using a TOPO TA Cloning kit (Invitrogen), digested with KpnI and XhoI and then inserted into the pGL4.15 luc2P/Hygro vector. All the recombinant plasmid sequences were verified by sequencing (Genwiz). The CpG-methylated reporters were generated by treating the reporter plasmids with methyltransferase M.SssI according to the manufacturer's instructions. Briefly, 5 mg of the reporter constructs were incubated with 5 U of M.SssI for 1 hour in NEBuffer 2 (50 mmol/L NaCl, 10 mmol/L Tris–HCl pH 7.9, 10 mmol/L MgCl₂, and 1 mmol/L dithiothreitol) supplemented with 160 mmol/L S-adenosylmethionine at 37°C. Methylated plasmids were purified using the QIAquick PCR Purification Kit (Qiagen), and the concentrations of all plasmids were determined by agarose gel electrophoresis. The efficiency of the methylation reactions was confirmed by digestion using the methylation-dependent HhaI and HpaII restriction endonucleases.

LNCaP cells were plated in 12-well plates for 24 hours and then transfected with 500 ng of the indicated reporter plasmids with GeneJuice (Novagen) according to the manufacturer's instructions. A total of 750 ng of pGL 4.75, which contains a β-galactosidase gene driven by a CMV promoter, was cotransfected as an internal control. After 24 hours of transfection, the cells were lysed in a Dual-Luciferase lysis buffer, and 10-mL aliquots of the cell lysate were assayed using the Dual-Luciferase Assay Kit with a Sirius luminometer (Berthold Technologies). The transcriptional activities of each construct were calculated by normalizing the firefly luciferase activities with the corresponding β-galactosidase enzyme activities and were reported as the fold change in induction relative to the empty pGL 4.15 vector. The values are expressed as the mean ± SD of four separate samples.

Total RNA was extracted from the treated cells using the RNeasy Mini Kit (Qiagen). First-strand cDNA was synthesized from 1 μg of total RNA using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen) according to the manufacturer's instructions. The cDNA was used as the template for real-time PCR (Applied Biosystems ViiA 7 Real-Time PCR System). The sequences of the primers used for cDNA amplification are described in Supplementary Table S1.

The treated cells were harvested using radioimmunoprecipitation assay (RIPA) buffer supplemented with a protein inhibitor cocktail (Sigma). The protein concentrations of the cleared lysates were determined using the bicinchoninic acid (BCA) method (Pierce), and 20 mg of the total protein was resolved by 4% to 15% SDS-PAGE (Bio-Rad). After electrophoresis, the proteins were electro-transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore). The PVDF membrane was then blocked with 5% BSA in PBS-0.1% Tween 20 (PBST) and then sequentially incubated with specific primary antibodies and horseradish peroxidase (HRP)–conjugated secondary antibodies. The blots were visualized with the Super Signal enhanced chemiluminescence (ECL) detection system and documented using the Gel Documentation 2000 system (Bio-Rad). The antibodies were purchased from the following sources: anti-NRF2 from Epitomics Inc.; anti-NQO1, anti-HO1, and anti-β-actin from Santa Cruz Biotechnology; anti-DNTM1, anti-DNMT3a, and anti-DNMT3b from IMGENEX; and anti-HDAC 1–6 from Cell Signaling Technology.

The chromatin immunoprecipitation (ChIP) assays were performed using the MAGnify Chromatin Immunoprecipitation System (Invitrogen) following the manufacturer's instructions. Briefly, LNCaP cells treated with DMSO or 5-aza/TSA for 7 days were washed with PBS and trypsinized. The trypsinized cells were then washed with PBS, and the chromatin from the cells (100,000 cells were used per IP) was cross-linked with 1% formaldehyde for 10 minutes at room temperature and sonicated to generate approximately 200- to 500-bp DNA fragments in lysis buffer. The proteins were then immunoprecipitated with 6 to 10 μL of anti-Pol II, anti-MBD2, anti-MeCP2, anti-trimethyl-histone H3-Lys9 (H3K9me3), and anti-H3Ac antibodies (Millipore) or mouse IgG to capture DNA-bound protein complexes. The enrichment of the eluted DNA was quantified by comparison with the input lysate by qPCR (29) using designed primers covering a 93-bp fragment that is 29 bp upstream of the first CpG site in the NRF2 promoter (Supplementary Table S1).

All experiments were performed at least three times with similar results. Differences in the DNA methylation levels between paired samples were analyzed using the Student t test with the Welch correction factor; the data for the luciferase reporter assay transcriptional activity, mRNA expression, ChIP assay, and percentage of IHC-positive cells were evaluated using the Student t test. A linear regression analysis was used to correlate the accumulated intensity of IHC of NRF2 and tissue groups (normal, hyperplasia, and adenocarcinoma tissue with Gleason score from 3–9). All P values were two-sided, and a P value of <0.05 was considered to be statistically significant.

The genomic sequence of the NRF2 gene (NC_ 000002.11: 178095031-178129859, Homo sapiens chromosome 2, reference GRCh37.p10 primary assembly, including 2.1 kb of its 5′-upstream sequence) was analyzed for CpG islands using the CpG Island Searcher (31). For simplicity, the distal boundary of the TSS (ATG) is defined as position +1. The CpG island of human NRF2 was identified between −764 and +1290 with a GC content of 66.6%, a CpG observed:expected ratio of 0.85 and a total of 205 CpG sites. Thus, the CpG island of the human NRF2 gene comprised the first exon and part of the first intron (Fig. 1A).

To assess the methylation levels in the NRF2 promoter across multiple patients, we performed MAQMA analysis of eight locus-specific PCR products from sodium bisulfate–converted DNA spanning position −1552 to +1091 (Fig. 1B; Supplementary Fig. S1). The analysis revealed specific hypermethylation in prostate tumor tissues (AS-CaP and ADT-RCaP, ranging from 5% to 78% methylation) of the BM-1 locus, which comprised three CpG sites within position −1493 to −1185 (Fig. 1B; P < 0.05). Although the analysis also showed significant differences in other specific promoter loci (e.g., the BM-5 locus), the methylation levels in prostate tumor tissue were low (Fig. 1B). To validate the methylation status of these three specific CpG sites in the NRF2 promoter, we performed BGS in the same set of human tissue samples. Figure 2A shows that CpG sites were differentially methylated in prostate tumors (AS-CaP and ADT-RCaP; 25.3% and 43.6% methylation, respectively) compared with human BPH tissues (19.6% methylation; P < 0.05). In addition, eight downstream CpG sites (−1135 to −917) were examined; however, these sites were poorly methylated (∼2%) and did not display differential methylation levels among the prostate tissues (data not shown).

To investigate the NRF2 expression in clinical samples, we examined the levels of NRF2 protein expression on a high-density prostate cancer TMA, using IHC staining against NRF2 protein (Fig. 2B). The analysis revealed that all sections containing normal, hyperplasia, and adenocarcinoma tissues showed positive NRF2 staining (Fig. 2B). Although both nucleus and cytoplasm show positive staining, majority of the NRF2 staining was observed in cytoplasm. The accumulated intensity of IHC staining of NRF2 revealed a wide range from low to high intensity in all the tissue groups analyzed. As described in the Materials and Methods, the NRF2 immunostaining results, scored according to the accumulated intensity, revealed that the strong staining (with accumulated intensity value scored as 3+) between normal and adenocarcinoma tissues did not show a significant difference (P = 0.9543). However, in the moderate staining (with accumulated intensity value scored as 2+), which included 25 tissue cores of normal prostate tissue and 34 tissue cores of prostate cancer tissue (hyperplasia and adenocarcinoma tissues with Gleason score from 3–9), the accumulated intensity of NRF2 staining showed a decreasing trend from normal tissues to advanced-stage prostate cancer (adenocarcinoma tissue with Gleason score from3–9; Fig. 2B). Likewise, the percentage of positive cells showed similar decreasing trend of NRF2 protein expression with significant differences between normal and advanced prostate cancer tissues (P < 0.05; Fig. 2C). Furthermore, when comparing adenocarcinoma with its adjacent tissues, we observed that the adjacent benign lumina tissue shows a relatively stronger staining (Fig. 2B).

Because promoter CpG demethylation often appears to regulate the transcriptional activation of genes, we examined the functional role of the three specific CpG sites in LNCaP cells with the pGL4.15-1526 and pGL4.15-1169 luciferase reporters, which contained the NRF2 promoter with or without the three CpG sites, respectively (Fig. 3A). To confirm the methylation status of both constructs, the reporter plasmids were methylated in vitro by M.sssI CpG methyltransferase and then digested with the HpaI or HhaII CpG-methylation–sensitive restriction endonucleases (Fig. 3B). When LNCaP cells were transfected with the unmethylated NRF2 promoter/luciferase reporter plasmids containing the three specific CpG sites, there was a 50-fold increase in luciferase reporter expression. Conversely, when the specific CpG sites were absent, the luciferase reporter expression was significantly decreased by 33-fold (Fig. 3C). The transcriptional activity of the methylated NRF2 promoters decreased dramatically (to ∼1-fold) when the specific CpG sites were present in the promoter. Nevertheless, when the three CpG sites were absent, the luciferase activity was decreased to approximately 2-fold (P < 0.05; Fig. 3C). These data suggest that these three CpG sites play a critical role in regulating NRF2 promoter activity. Similar results were also obtained with PC-3 and HeLa cells (Supplementary Fig. S2).

We previously reported that combined treatment with DNMT and HDAC inhibitors 5-aza and TSA, respectively, decreases the CpG methylation levels of the NRF2 promoter in murine prostate cancer TRAMP C1 cells (8). Bisulfite sequencing of genomic DNA from LNCaP cells was performed to examine the methylation levels and determine whether 5-aza/TSA treatment decreases NRF2 promoter methylation. As shown in Fig. 4A, the combined 5-aza/TSA treatment over 7 days in LNCaP cells significantly reduced (P < 0.05) the methylation level of these specific CpG sites (−1530 to −1143) from 62.96% to 32.78%. Similar to human prostate tumor samples, the next eight downstream CpG sites (−1166 to −896) were poorly methylated (∼1%) and did not show any differential methylation change between DMSO and 5-aza/TSA treatments.

To investigate the effect of the 5-aza/TSA treatment on the expression of NRF2 and its related genes in LNCaP cells, we first examined the expression of NRF2, NQO1, and HO-1 by real-time RT-PCR. Accordingly, the gene expression levels of NRF2 and ARE-mediated genes were significantly (P < 0.05) increased in the 5-aza/TSA–treated LNCaP cells compared with the control (Fig. 4B). We next assessed the protein levels of NRF2, NQO1, and HO-1 by Western blot analysis. Figure 4C shows that the protein levels from NRF2 and ARE-mediated genes were in agreement with the previous findings as there were increased protein levels compared with those in the 0.1% DMSO-treated control cells.

We next examined the protein levels of DNMTs and HDACs in LNCaP cells after the control and 5-aza/TSA treatments because DNA methylation is regulated by DNMTs, including DNMT1, DNMT3a, and DNMT3b, and histone methylation is controlled by the HDAC enzymes (4). Figure 5A shows that the 5-aza/TSA treatment dramatically decreased the protein expression of DNMT1, DNMT3a, and DNMT3b by fold changes of approximately 0.5, 0.6, and 0.3, respectively, compared with the control. As expected, HDAC (1–6) protein expression was also decreased after the 5-aza/TSA treatment (Fig. 5B).

DNA methylation has been widely reported to interact with various methyl-CpG–binding domain proteins (MBD) and methyl CpG binding protein 2 (MeCP2; refs. 32, 33). These binding proteins can then also interact with a corepressor complex that includes HDACs, which results in the transcriptional repression of certain genes (34). ChIP assay analysis was performed using specific H3Ac, Pol II, H3K9Me3, MBD2, and MeCP2 antibodies to investigate the proteins that could be potentially bound to the NRF2 promoter in control and 5-aza/TSA–treated LNCaP cells. Interestingly, the 5-aza/TSA treatment of LNCaP cells increased the total amount of H3Ac enrichment and further decreased the total amount of H3K9Me3, MBD2, and MeCP2 at the NRF2 promoter (P < 0.05). In agreement with the above results, Pol II was enriched at the NRF2 promoter (P < 0.05), suggesting a suppression of NRF2 transcription in LNCaP cells basally and that the activation of NRF2 could be reversed by the 5-aza/TSA treatment (Fig. 6).

NRF2 has been widely described as a key transcription factor regulating the activity of many ARE-mediated type II detoxifying/antioxidant enzymes in response to antioxidative/electrophilic stimuli and therapeutic signals (4, 35, 36). In this context, NRF2 has been reported to be downregulated in human prostate cancer and prostate cancer in TRAMP mice, revealing that the loss of NRF2 correlates with the reduced expression of several classes of GSTs, which ultimately leads to elevated ROS levels and the DNA damage that is associated with tumorigenesis (8, 20). Conversely, the aberrant overexpression of NRF2 from gain-of-function mutations or the deregulation of factors regulating NRF2 expression has been detected in various advanced cancer tissues, including lung, esophagus, larynx, gallbladder, skin, pancreas, ovary, and breast (37, 38). Thus, the nuclear accumulation of NRF2 in these cancers leads to higher levels of cytoprotective proteins, including detoxifying/biotransformation enzymes, drug transporters, antioxidants, and antiapoptotic proteins, which results in decreased apoptosis, increased cell survival, and drug resistance to chemotherapy (4, 39).

We have also demonstrated that NRF2 expression is regulated by an epigenetic mechanism during prostate cancer progression in TRAMP mice in vivo and in TRAMP C1 cells in vitro, suggesting that there is a potential role for epigenetic modifications of NRF2 in human prostate cancer (8, 36). However, there are no data showing the molecular mechanisms behind the epigenetic regulation of NRF2. Previous studies have reported the downregulation of GSTP1, RARβ, TNFRSF10C, RASSF1A, and Neurog1 by epigenetic modifications and the presence of many other methylation markers in human prostate cancer (5, 6, 40). In this study, we present data demonstrating that NRF2 is epigenetically regulated through promoter DNA methylation at three specific CpG sites, which were differentially hypermethylated in clinical prostate cancer tissues and human LNCaP cells and significantly decreased in BPH tissues (Figs. 1B, 2A, and 4A). Our IHC analysis of NRF2 expression in prostate tissues (moderate staining, with accumulated intensity value scored as 2+) showed that the accumulated intensity and the percentage of positive cells decreased progressively from normal prostate to adenocarcinoma tissues (adenocarcinoma tissue with Gleason score from 3–9; Fig. 2B and C). We also found that tumor tissues presented adjacent benign or normal lumina tissues that exhibited higher staining of NRF2 compared with the adenocarcinoma cells (Fig. 2B). Importantly, the expression of NRF2 in human prostate cancer has been shown to be significantly reduced in advanced-stage tumors compared with normal prostate in several published human cancer gene expression datasets (ONCOMINE; ref. 41), providing evidence that DNA methylation of the NRF2 promoter could contribute to regulation of transcription activity of NRF2 in advanced stage of human prostate cancer (Supplementary Fig. S3; refs. 42–48).

The functional role of the DNA methylation of the three specific CpG sites in the CpG island of NRF2 was assessed through luciferase reporter constructs and in vitro methylation assays (Fig. 3). The designed pGL-4.15–1169 and pGL-4.15–1526 reporters were active in LNCaP, PC-3, and HeLa cell lines, as shown in Fig. 3C and Supplementary Fig. S2. Interestingly, the promoter sequence containing the three CpG sites from −1526 to −1169 in LNCaP cells repressed the transcriptional activity of the NRF2 promoter when methylated in vitro by CpG methyltransferase (Fig. 3C). Furthermore, this DNA sequence was able to interact with specific repressive factors revealed by ChIP assay analysis (Fig. 6). Interestingly, we have computationally identified seven putative transcription factor–binding sites (TFBS) in the genomic sequence of human NRF2 promoter region comprising two out of the three relevant CpG site (Supplementary Fig. S4). Accordingly, the TFBS identified for the first CpG site were: NF-1 (neurofibromin 1), FOXP3 (forkhead box P3), Pax-5 (paired box 5), and p53 (tumor protein p53); and for the second CpG site: GR-α (glucocorticoid receptor), c-Ets-1 (v-ets avian erythroblastosis virus E26 oncogene homolog 1), and c-Ets-2 (v-ets avian erythroblastosis virus E26 oncogene homolog 2). These observations suggest that these three specific CpG sites could be potentially relevant in the methylation-dependent suppression of NRF2 expression in vitro. Similarly, we described the methylation activity of five specific CpGs sites in TRAMP C1 cells, which controls the transcriptional activity of Nrf2 through the interaction with MBDs and histone modifications (8). Interestingly, we also demonstrated that these epigenetic mechanisms regulating NRF2 expression in prostate cancer TRAMP mice can be reversed by dietary phytochemicals, thereby revealing that specific CpG hypermethylation of the NRF2 promoter could represent a potential molecular biomarker for human prostate cancer prevention (4, 36, 49).

Several studies have revealed that the methylation of CpG sites in gene promoters is a frequent cause of epigenetic gene silencing in prostate cancer cell lines and can be reversed by using DNMT and HDAC inhibitors (5, 50–52). To interrogate the role of epigenetic modifications in the suppression of NRF2 expression and its activation, we treated LNCaP cells with the DNMT and HDAC inhibitors 5-aza and TSA, respectively. The results from this study revealed that the expression of NRF2 and the downstream NQO1 and HO-1 proteins was restored after the 5-aza/TSA treatment (Fig. 4B and C). Our previous observations in prostate cancer TRAMP mice and TRAMP C1 cells confirmed that methylation is important for the regulation of NRF2 expression and that the treatment of TRAMP C1 cells with the DNMT/HDAC inhibitors 5-aza/TSA and dietary phytochemicals, such as curcumin and sulforaphane, restore the expression of NRF2 and induce downstream genes, such as NQO1 (8, 21, 53). Thus, the reactivation of NRF2 in LNCaP cells was consistent with the decreased methylation status and protein expression levels of DNMT1, DNMT3a, and DNMT3b (Fig. 5A).

In this study, we performed ChIP assays to explore the molecular mechanisms underlying the suppression of NRF2 via methylation of the three specific promoter CpG sites. The results showed that 5-aza/TSA–treated LNCaP cells enriched the total amount of RNA Pol II and H3Ac and further decreased the total amount of H3-Lys9 (H3K9me3), MBD2, and MeCP2 at the CpG sites of the NRF2 promoter (Fig. 6), which correlates well with the reexpression of NRF2 and decreased methylation levels (Fig. 4A–C). These observations are in agreement with previous studies in which 5-aza and TSA treatments are able to modulate the methylation-dependent association of various MBDs, such as MBD1–MBD4 and MeCP2, and the interaction of these proteins with a corepressor complex that includes HDACs (34). Ultimately, this process results in gene transcriptional repression (32–34). In previous studies using the same molecular approach in prostate tumors from TRAMP mice, we demonstrated that CpG methylation of the Nrf2 promoter was associated with MBD2, H3Ac, and H3-Lys9 (H3K9me3; ref. 8). Similarly, in LNCaP cells, we found that the MBD2 and MeCP2 proteins and the H3K27me3 epigenetic modification are involved in the transcriptional repression of Neurog1 (40). The above results confirmed our findings in LNCaP cells showing that the enrichment of H3-Lys9 (H3K9me3), MBD2, and MeCP2 is related to transcriptionally repressed chromatin (54). These results are also supported by the increased protein expression of HDACs (HDACs 1–6) in DMSO-treated control LNCaP cells compared with 5-aza/TSA–treated cells (Fig. 5B).

In summary, this study shows for the first time that the in vitro hypermethylation of three specific CpG sites in the NRF2 promoter correlates with reduced NRF2 expression in human prostate cancer cells and the hypermethylated status of the NRF2 promoter in clinical prostate cancer tissues. These findings provide new insights into earlier observations that showed differential NRF2 expression in human prostate tumors (20). Further studies of the methylation profiles from a large number of primary tumor samples will provide important information on the role of methylation of the CpG islands in the NRF2 promoter. Importantly, the NRF2 methylation status may contribute to regulation of transcription activity of NRF2, which could be used as preventive and therapeutic target for the treatment of prostate cancer in future clinical trials.

No potential conflicts of interest were disclosed.

Conception and design: T.O. Khor, F. Fuentes, S. Yegnasubramanian, A.-N.T. Kong

Development of methodology: T.O. Khor, F. Fuentes, A.Y. Yang, Y. Liu, D.J. Smiraglia, A.-N.T. Kong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Fuentes, L. Shu, X. Paredes-Gonzalez, A.Y. Yang, Y. Liu, D.J. Smiraglia, A.-N.T. Kong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Fuentes, L. Shu, X. Paredes-Gonzalez, D.J. Smiraglia, W.G. Nelson, A.-N.T. Kong

Writing, review, and/or revision of the manuscript: T.O. Khor, F. Fuentes, D.J. Smiraglia, S. Yegnasubramanian, W.G. Nelson, A.-N.T. Kong

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Fuentes, A.-N.T. Kong

Study supervision: T.O. Khor, F. Fuentes, A.-N.T. Kong

The authors thank all the members in A.-N.T. Kong's laboratory for their helpful discussion and preparation of this article.

This work was supported, in part, by institutional funds and by R01-CA118947, R01-CA152826, from the National Cancer Institute (NCI), R01AT007065 from the National Center for Complementary and Alternative Medicines (NCCAM) and the Office of Dietary Supplements (ODS; to A.-N. T. Kong).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.