Epigenetic Regulation of Vitamin D 24-Hydroxylase/CYP24A1 in Human Prostate Cancer

Published OnlineFirst June 29, 2010; DOI: 10.1158/0008-5472.CAN-10-0617 Published OnlineFirst on June 29, 2010 as 10.1158/0008-5472.CAN-10-0617

Therapeutics, Targets, and Chemical Biology Cancer Research Epigenetic Regulation of Vitamin D 24-Hydroxylase/CYP24A1 in Human Prostate Cancer

Wei Luo1, Adam R. Karpf1, Kristin K. Deeb1, Josephia R. Muindi2, Carl D. Morrison3, Candace S. Johnson1, and Donald L. Trump2

Abstract Calcitriol, a regulator of calcium homeostasis with antitumor properties, is degraded by the product of the CYP24A1 gene, which is downregulated in human prostate cancer by unknown mechanisms. We found that CYP24A1 expression is inversely correlated with promoter DNA methylation in prostate cancer cell lines. Treatment with the DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine (DAC) activates CYP24A1 expression in prostate cancer cells. In vitro methylation of the CYP24A1 promoter represses its promoter activity. Furthermore, inhibition of histone deacetylases by trichostatin A (TSA) enhances the expression of CYP24A1 in prostate cancer cells. Quantitative chromatin immunoprecipitation-PCR (ChIP-qPCR) reveals that specific histone modifications are associated with the CYP24A1 promoter region. Treatment with TSA increases H3K9ac and H3K4me2 and simultaneously decreases H3K9me2 at the CYP24A1 promoter. ChIP-qPCR assay reveals that treatment with DAC and TSA increases the recruitment of vitamin D receptor to the CYP24A1 promoter. Reverse transcriptase-PCR analysis of paired human prostate samples revealed that CYP24A1 expression is downregulated in prostate malignant lesions compared with adjacent histologically benign lesions. Bisulfite pyrosequencing shows that CYP24A1 gene is hypermethylated in malignant lesions compared with matched benign lesions. Our findings indicate that repression of CYP24A1 gene expression in human prostate cancer cells is mediated in part by promoter DNA methylation and repressive histone modifications. Cancer Res; 70(14); OF1–10. ©2010 AACR.

Introduction expression transgenic rats show low vitamin D3 levels (8, 9); (b) the rate of serum calcitriol clearance after calcitriol ad- 1α,25-Dihydroxycholecalciferol (calcitriol), the active form ministration is delayed in CYP24A1-null mouse (10, 11); and c CYP24A1 of vitamin D3, plays a major role in regulating calcium ( ) pharmacologic inhibition of in combination homeostasis and bone mineralization (1–3). Calcitriol also with calcitriol enhances antitumor effects in human prostate modulates cellular proliferation and differentiation in a cancer and other tumor types (12–15). variety of cell types (4). Calcitriol promotes cell cycle arrest Similar to CYP27B1, a vitamin D–activating enzyme that is and induces cell apoptosis (5). The cytochrome P450 enzyme aberrantly expressed in tumors (16), CYP24A1 is dysregulated 25-hydroxyvitamin D 24-hydroxylase (24-hydroxylase), in a wide range of tumors. Overexpression of CYP24A1 has encoded by the CYP24A1 gene, mediates 24-hydroxylation been reported in several tumors including colon carcinoma α of 1 ,25(OH)2D3 to the much less active vitamin D metabo- (17, 18), ovarian cancer (17, 19), cervical carcinoma (19), lung lites including 1α,24,25(OH)2D3 and its biliary excretory cancer (17, 20), and cutaneous basal cell and squamous cell product calcitroic acid (6, 7). The important role of CYP24A1 carcinoma (21, 22). The overexpression of CYP24A1 is associ- in maintaining vitamin D homeostasis and antitumor effects ated with poor prognosis of some human cancers (17). Upre- is supported by several lines of evidence: (a) CYP24A1 over- gulated CYP24A1 expression may counteract calcitriol antiproliferative activity, presumably by decreasing calcitriol levels. However, microarray expression studies published in Authors' Affiliations: Departments of 1Pharmacology and Therapeutics, ONCOMINE indicate that CYP24A1 is downregulated in pros- 2Medicine, and 3Pathology, Roswell Park Cancer Institute, Buffalo, tate cancer (23). The mechanism(s) underlying the dysregu- New York lation of CYP24A1 in tumors is not clear. Recently, Novakovic Note: Supplementary data for this article are available at Cancer CYP24A1 Research Online (http://cancerres.aacrjournals.org/). and colleagues showed tissue-specific promoter methylation in normal human tissues (24). The CYP24A1 W. Luo and A.R. Karpf contributed equally to this work. gene is methylated in human placenta; no methylation was Corresponding Author: Donald L. Trump, Department of Medicine, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, NY detected in somatic human tissues (24). A lack of expression 14263. Phone: 716-845-5772; Fax: 716-845-8261; E-mail: donald. of CYP24A1 has also been directly associated with DNA [email protected]. methylation of the upstream promoter and the first exon doi: 10.1158/0008-5472.CAN-10-0617 of the CYP24A1 gene in mouse tumor-derived endothelial cells ©2010 American Association for Cancer Research. and in rat osteoblastic ROS17/2.8 cells (12, 25). Differential

www.aacrjournals.org OF1

Luo et al.

methylation of CYP24A1 gene contributed to selective anti- sequencing primers [region 1 (−500 to +6), 5′-AAAATT- proliferative effects of calcitriol treatment on tumor-derived ATTTTAGTTTAGGTTGGGG-3′ and 5′-TCTCCATATTCC- endothelial cells compared with Matrigel-derived endothelial TATACCCAAAAAC-3′;region2(−17 to +609), 5′- cells (12). TTTTTGGGTATAGGAATATGGAGAG-3′ and 5′-CCCAA- The above data suggest that epigenetic mechanisms play a CAATAACCAACTAATAAAAC-3′]wereusedtoamplifythe key role in regulating CYP24A1 expression in normal human CpG islands in the CYP24A1 promoter region. PCR included tissues. However, little is known about the mechanism(s) an initial incubation at 95°C for 3 minutes, followed by 40 cy- underlying the regulation of CYP24A1 expression in human cles of 95°C for 30 seconds, 56°C for 30 seconds, and 72°C for tumors. Given the importance of CYP24A1 in regulating 60 seconds, followed by one cycle of 72°C for 10 minutes. The 1α,25(OH)2D3 levels, we investigated the role of epigenetic PCR products were cloned into the pCR4-TOPO vector using regulation of CYP24A1 in human prostate cancer cell lines the TOPO TA cloning kit (Invitrogen) for sequencing. A total and primary prostate tumors. of 10 clones from each sample were sequenced at the Roswell Park Cancer Institute DNA sequencing core facility, and the Materials and Methods methylation status for each CpG site was determined by as- sessing the presence of T (unmethylated) versus C (methyl- Human tissues and cells ated) at each CpG site. Human prostate cancer cell lines DU145, LNCaP, and PC3 were obtained from the American Type Culture Collection Bisulfite DNA pyrosequencing and maintained in our laboratory. Human DNA and RNA The methylation status of two regions of the CYP24A1 pro- from 30 paired human prostate benign and primary malig- moter in human prostate tissue samples was analyzed by bi- nant lesions were obtained from Department of Pathology, sulfite pyrosequencing. For the first region, the primers used Roswell Park Cancer Institute, and approved by the Institu- were −499F (5′-AAAATTATTTTAGTTTGGTTGGGG-3′)and tional Review Board. −244R (Biotin-5′-AAATAACCCCCAAAAAATCATAC-3′)for amplification and primer 5′-TTAGTAGGGGGAGGG-3′ was Drug treatments used for sequencing. For the second region, the primers used LNCaP, PC3, and DU145 cells were seeded at 1 × 105 cells in a for amplification were +66F (5′-GTGGTTTTAGTTAGAT- six-well plate overnight. For dose-response experiments, cells TTTAGAGG-3′ and +301R (Biotin-5′-CAAAAAAAAC- were treated with 5-aza-2′-deoxycytidine (DAC) at 0.1, 0.25, CAACTAATATAAT-3′) and the sequencing primer was 0.5, 1.0, and 2.0 μmol/L concentrations for 3 days, followed 5′-TTAGTGTAAGGAGGTATTAA-3′. PCR cycling conditions by the addition of calcitriol (100 nmol/L) for an additional were an initial incubation at 95°C for 3 minutes, followed by 24 hours, and cells were harvested. Cells were treated with tri- 40 cycles of 95°C for 30 seconds, 52°C (for region 1) and 51°C chostatin A (TSA) at 50, 100, 200, 300, and 400 nmol/L for (for region 2) for 30 seconds, and 72°C for 30 seconds. Pyro- 8 hours, followed by the addition of calcitriol (100 nmol/L) sequencing was accomplished in duplicate using the PSQ for an additional 24 hours, and cells were harvested. For 96 Pyrosequencing System (Biotage) and was performed DAC and TSA combination treatments, cells were treated two times. with 1 μmol/L DAC for 3 days, followed by the addition of TSA and calcitriol for an additional 24 hours, and cells were CYP24A1 activity assay harvested. Cells were treated with DAC for 72 hours and/or TSA for 8 hours, followed by 100 nmol/L calcitriol for 48 hours, and Quantitative reverse transcriptase-PCR were washed with 1% bovine serum albumin (BSA)-PBS 6 Expression of CYP24A1 mRNA was assessed by quantita- twice. Then, cells (1 × 10 ) were incubated with 25(OH)D3 tive reverse transcriptase-PCR (RT-PCR) using the CYP24A1 (1 μmol/L) in 2 mL of 1% BSA medium for 12 hours. TaqMan Gene Expression Assay (Hs00167999_m1, Applied 25(OH)D3 metabolites from cells and medium were extracted Biosystems). The amplification conditions for both CYP24A1 by liquid-liquid partition with 8 mL of chloroform/methanol and GAPDH were 50°C for 2 minutes, 95°C for 10 minutes, (3:1). Before the extraction, 20 μL of EB1089 (10 μg/mL) were 40 cycles of 95°C for 15 seconds, 60°C for 30 seconds on the added as an internal standard. Dried sample extracts were 7300 real-time PCR system (Applied Biosystems). Gene ex- dissolved in 100 μL of high-performance liquid chromatogra- pression of CYP24A1 was normalized to the GAPDH and phy (HPLC) mobile phase, hexane/isopropanol/methanol in all samples were analyzed in triplicate. Statistical analysis 90:5:5 (v/v/v). Vitamin D3 metabolites and internal standard with Wilcoxon signed-rank test was performed to compare in 50 μL of the extract were separated by HPLC on a 4.6 × 250- the expression levels of CYP24A1 in benign and malignant mm Zorbax SIL column (Agilent Technologies, Inc.) using samples. HPLC mobile phase at a flow rate of 2 mL/min. Metabolites λ were monitored at 265 nm and identified based on retention Bisulfite DNA sequencing time and Diode array collected vitamin D3 chromophore spec- Genomic DNA was isolated using the genomic DNA iso- tral characteristics (UVmax = λ265,UVmin = λ228,UVmax/UVmin = lation kit (Qiagen). DNA (1 μg) was converted with EZ 1.75). CYP24A1 activity was calculated as the area ratio of DNA Methylation Kit (Zymo Research Corporation). Primer 24,25-(OH)2D3: recovered 25-(OH)D3 +24,25-(OH)2D3/12 design was accomplished using Methprimer (26). Bisulfite h/106 cells.

OF2 Cancer Res; 70(14) July 15, 2010 Cancer Research

Epigenetic Regulation of CYP24A1

In vitro CYP24A1 promoter methylation assay Dual-Luciferase Reporter Assay System (Promega) according A 587-bp region of the human CYP24A1 promoter to the manufacturer's instructions. Statistical analysis with gene that was shown (27, 28) to confer vitamin D respon- Student's t test on luciferase values was performed to com- siveness was amplified using primers Xho-CYP24−556F pare the CYP24A1 promoter activities in unmethylated and (5′-GCTAGCCGCAGAAAGCCAAACTTCCTC-3′)andNHeI- methylated CYP24A1 promoter constructs. The experiment CYP24+40R (5′-GCAGCTCGAGAGATGCTGCTGG- was repeated twice to confirm the reproducibility of results. CGCTGCGT-3′) with PfuUltra high-fidelity DNA polymerase (Stratagene). The Xho-andNheI-digested amplicon was Quantitative chromatin immunoprecipitation-PCR cloned into the promoterless luciferase expression vector Quantitative chromatin immunoprecipitation-PCR (ChIP- pGL4.21 (Promega) to produce the plasmid pGL4.21/547/ qPCR) was performed essentially as described by Väisänen +40. Methylation of the CYP24A1 promoter region was done and colleagues (30). Briefly, cells were treated with DAC as described (29). Briefly, the fragments were methylated (1 μmol/L), TSA (300 nmol/L), or both agents. One hour in vitro with SssI, HpaII, and HhaI methylases (New England after adding calcitriol, cells were fixed to cross-link nuclear Biolabs),ornoenzyme(mock),accordingtothemanu- protein to DNA by adding formaldehyde as described previ- facturer's instructions. Methylation efficiency was confirmed ously (30). Chromatin was sheared to an average length of using HhaI, HpaII, and McrBC digestions (New England 500 bp. The chromatin was precipitated with the indicated Biolabs). Methylated or mock-methylated fragments were antibodies. Antibodies against histone H3 dimethylated ly- ligated back into pGL4.21 vector (Promega). PC3 cells were sine 9 (H3K9me2), dimethylated lysine 4 (H3K4me2), and transfected with 100 ng of the differentially methylated acetylated lysine 9 (H3K9ac) were purchased from Upstate constructsalongwith20ngofaRenilla luciferase control Biotechnologies; an antibody against vitamin D receptor construct (Promega). All transfections were carried out in (VDR; sc-1008x) was obtained from Santa Cruz Biotech- triplicate wells of 96-well plates. Twenty-four hours after nologies. ChIP-qPCR primers for the CYP24A1 promoter transfection, cells were treated with calcitriol (100 nmol/L) region 1 and region 2 were −292F (5′-AGCACACCCGGT- for an additional 24 hours and harvested, and firefly and GAACTC-3′) and −152R: (5′-TGGAAGGAGGATGGAGTCAG- Renilla luciferase activities were measured using the 3′); +130F (5′-TTCAAGAGGTCCCCAGACAC-3′) and +333R

Figure 1. CYP24A1 expression and promoter DNA methylation. A, quantitative RT-PCR of CYP24A1 expression in human prostate cancer cells. B, diagram of the CYP24A1 promoter CpG islands. The regions analyzed by bisulfite sequencing and pyrosequencing, along with the number of CpGs contained within each region and the VDREs (VDREp and VDREd), are shown. C, bisulfite sequencing of the CYP24A1 promoter CpG island methylation. The overall methylation percentage from 10 individual clones indicates the total proportion of methylated CpGs in the CYP24A1 promoter region in a given sample, taking into account all sequenced alleles.

www.aacrjournals.org Cancer Res; 70(14) July 15, 2010 OF3

Luo et al.

(5′-AGTCGGGGCTTAACGATTCT-3′). ChIP-qPCR was per- encompasses the promoter region and contains two known formed using Power SYBR Green PCR Master Mix (Applied VDR response elements (VDRE): the proximal element VDRE Biosystems) with the following cycling parameters: 95°C for (VDREp; −172/−143) and the distal element VDRE (VDREd; 10 minutes, 40 cycles of 95°C for 15 seconds, and 60°C for −293/−273; Fig. 1B; ref. 27). To examine whether CYP24A1 1 minute. Copy number was determined using a standard is differentially methylated in these cell lines, we conducted curve containing serial dilutions (107–100 copies) of the bisulfite sequencing on 506-bp region 1 and 627-bp region 2 CYP24A1 DNA amplicon. Samples were run in triplicate of the CYP24A1 5′-CpG island promoter (Fig. 1B). This assay and data were normalized to 5% input DNA amplifications revealed that the overall methylation of CYP24A1 promoter after subtraction of signals obtained from antibody isotype regions in LNCaP and PC3 cells is 24% and 50%, respectively. control. ChIP was repeated twice to confirm the reproduc- The hypermethylation is particularly evident in region 2 of ibility of results. the CYP24A1 gene. In contrast, the CYP24A1 5′-CpG island was completely unmethylated in DU145 cells, which is con- Results sistent with constitutive CYP24A1 expression in this cell line (Fig. 1C). These data suggest that promoter DNA methylation Correlation of CYP24A1 expression in human prostate status may regulate differential CYP24A1 expression in pros- cell lines with promoter DNA methylation status tate cancer cells. To investigate the mechanism(s) regulating CYP24A1 mRNA expression in prostate cancer, we analyzed LNCaP, Activation of CYP24A1 gene expression by epigenetic PC3, and DU145 cells, which express variable levels of modulatory drugs in prostate cancer cells CYP24A1. Quantitative RT-PCR showed that LNCaP and To determine the role CYP24A1 gene methylation in PC3 cells express CYP24A1 weakly in the presence of calci- CYP24A1 expression, we treated LNCaP and PC3 cells with triol, whereas DU145 cells express CYP24A1 at high baseline DAC, a DNA methyltransferase inhibitor. Quantitative RT- and show further induction following calcitriol treatment PCR revealed that DAC elicited a dose-dependent induction (Fig. 1A). We further tested whether CpG island methylation of CYP24A1 expression in LNCaP and PC3 cells that was de- could account for differential CYP24A1 expression in the pendent on calcitriol treatment (Fig. 2A). To confirm the prostate cell lines. Analysis of the human CYP24A1 gene by demethylation effect of DAC, bisulfite sequencing was done Webgene indicated that it contains two CpG islands (580 on treated cell lines. Overall methylation of the CYP24A1 and 1,232 bp, respectively) that span the transcriptional start promoter in LNCaP and PC3 cells was reduced from 24% site (Fig. 1B; ref. 31). The 5′ CpG island of the CYP24A1 gene to 5% and from 50% to 13%, respectively (Figs. 1C and 2B).

Figure 2. Activation of CYP24A1 by DAC treatment. A, quantitative RT-PCR of CYP24A1 expression in cells treated with varying concentrations of DAC followed by calcitriol. Columns, mean values of triplicate data points; bars, SD. B, bisulfite sequencing of the CYP24A1 promoter CpG island methylation in DAC treated cells. The overall methylation percentage indicates the total proportion of methylated CpGs in the CYP24A1 promoter region in a given sample, taking into account 10 sequenced alleles.

OF4 Cancer Res; 70(14) July 15, 2010 Cancer Research

Epigenetic Regulation of CYP24A1

Figure 3. Activation of CYP24A1 gene expression and enzymatic activity by TSA alone and in combination with DAC. A, quantitative RT-PCR of CYP24A1 expression in cells treated with varying concentrations of TSA followed by calcitriol. Columns, mean values of triplicate data points; bars, SD. B, quantitative RT-PCR of CYP24A1 expression in cells treated with TSA in combination with DAC followed by calcitriol. Columns, mean values of triplicate data points; bars, SD. C, CYP24A1 activity in cells treated with DAC and/or TSA.

To determine whether histone deacetylation is also in- lines, combination treatment of DAC and TSA causes a sig- volved in repression of CYP24A1 expression, we treated nificant increase in CYP24A1 activity (Fig. 3C). However, LNCaP and PC3 cells with TSA, a histone deacetylase inhib- it is not clear why the differences in the enzyme activity itor. Quantitative RT-PCR revealed that TSA elicited a dose- between DU145 cells and the other two cells lines are less dependent induction of CYP24A1 expression in LNCaP and than the relative differences at the mRNA expression level. PC3 cells that was dependent on calcitriol treatment We speculate that the enzyme activity may be affected by (Fig. 3A). Combination treatment with DAC and TSA induced other endogenous factors and that the mRNA and protein a significant increase in CYP24A1 mRNA and protein expres- expression of CYP24A1 may closely, but not exactly, parallel. sion in LNCaP and PC3 cells (Fig. 3B, top and middle; Sup- plementary Fig. S1). In contrast, DU145, which has an Repression of human CYP24A1 gene promoter activity unmethylated promoter, showed no increase in CYP24A1 by DNA methylation expression with DAC treatment but a robust effect with To examine whether DNA methylation directly represses TSA treatment (Fig. 3B, bottom). The slightly less CYP24A1 CYP24A1 promoter activity, we cloned a 587-bp fragment of expression with combination treatment of DAC and TSA the 5′ end of CYP24A1, which contains 46 CpG sites, into a compared with TSA alone could be due to the additive tox- luciferase reporter construct. We methylated the cloned in- icity of this approach in DU145. To test whether the in- sert (29) using SssI methylase (methylation of 46 CpGs), creased CYP24A1 expression by epigenetic modulators is HpaII methylase (methylation of 14 CpGs), and HhaI meth- associated with increased CYP24A1 activity, we performed ylase (methylation of 4 CpGs). SssI methylase methylates all enzyme activity assay by HPLC. Increased CYP24A1 enzyme 5′-CpG-3′ sites, HpaII methylase methylates only the CpG activity was observed in LNCaP and PC3 cells treated with within the sequence 5′-CCGG-3′, and HhaI methylase methy- TSA (Fig. 3C). In addition, there was no significant increase lates only the CpG within the sequence 5′-GCGC-3′. Proper of enzyme activity in DU145 cells treated with DAC, which methylation of inserts was confirmed by digesting with the has hypomethylated CYP24A1 promoter (Fig. 3C). In all cell restriction enzymes McrBC, HpaII, and HhaI (Fig. 4A).

www.aacrjournals.org Cancer Res; 70(14) July 15, 2010 OF5

Luo et al.

CYP24A1 promoter activity was tested by transfection of the ChIP-qPCR analysis using a VDR-specific antibody revealed luciferase construct containing methylated CYP24A1 promot- that there was significantly increased binding of VDR to er. Prior methylation with SssI, HpaII, or HhaI methylases in the CYP24A1 promoter in PC3 cells on treatment with DAC PC3 cells repressed CYP24A1 promoter activity (Fig. 4B). Re- and TSA, in addition to calcitriol (Fig. 5C). These data pro- pression was methylation dose dependent in that SssI, which vide functional evidence that decreased VDR binding to the methylates all CpG sites, showed the greatest repression, and CYP24A1 promoter in PC3 cells, due to hypermethylation and HhaI, which methylates four CpG sites, showed the least re- a repressive histone modifications, contributes to the suppression on CYP24A1 promoter activity (Fig. 4B). pression of CYP24A1 expression in prostate cancer cells.

Histone modifications associated with the CYP24A1 Expression and methylation of CYP24A1 in primary promoter region normal and cancerous human prostate tissues The association of CYP24A1 promoter DNA methylation We analyzed CYP24A1 expression in human normal pros- and gene silencing in relation to histone modifications has tate, benign prostatic hyperplasia, prostate carcinoma, and not been investigated previously. To clarify the role of histone metastatic prostate cancer using microarray expression modifications in the regulation of the CYP24A1 gene, we studies published in ONCOMINE (23). CYP24A1 expression examined the active histone marks H3K4me2 and H3K9ac was significantly lower in prostate cancer compared with and the repressive histone mark H3K9me2 in the chromatin normal prostate in three of four (P < 0.01) microarray expres- associated with the CYP24A1 promoter region using ChIP- sion studies (Fig. 6A). However, VDR expression was not qPCR. The results showed similar histone acetylation patterns downregulated in studies published in ONCOMINE (data and H3 methylation levels in LNCaP and PC3 cells with not shown). We obtained 30 matched human prostate benign relative high H3K9me2 compared with DU145 cells (Fig. and malignant prostate samples. Gleason scores were 3 + 3 5A). In contrast, the active histone markers (H3K4me2 and (n =6),3+4(n = 15), 4 + 3 (n =7),4+5(n = 1), and 5 + 4 H3K9ac) were highest in DU145 cells, as expected (Fig. 5A). (n = 1). Quantitative RT-PCR revealed that CYP24A1 expres- We also examined whether treatment with DAC and/or TSA sion was significantly downregulated in prostate malignant changed the profile of histone modifications in the CYP24A1 lesions compared with benign lesions (P = 0.0336; Fig. 6B promoter. The ChIP-qPCR assay revealed that TSA treatment and C). Bisulfite pyrosequencing assays for the CYP24A1 gene induced increases of H3K9ac and H3K4me2 and a decrease of were designed to span VDREd and VDREp (21) and a puta- H3K9me2 in PC3 cells, giving rise to an active pattern of tive regulatory sequence located in exon 1 within the associ- histone modifications following treatment (Fig. 5B). Similar ated CpG island. Pyrosequencing revealed that the CYP24A1 histone modification changes were also observed in LNCaP gene was significantly hypermethylated in malignant lesions cells after treatment (data not shown). as compared with matched benign lesions (Fig. 6D). There was no correlation of the level of CYP24A1 expression and Increased recruitment of VDR to the CYP24A1 promoter methylation with Gleason score. To determine whether in PC3 cells treated with epigenetic modulatory drugs malignant samples with decreased CYP24A1 expression We next examined whether these epigenetic changes affect correlated with CYP24A1 hypermethylation, we built a 2 × 2 the binding of the VDR to the CYP24A1 promoter region. contingency table by dividing the 30 samples based on the

Figure 4. Repression of CYP24A1 promoter activity by DNA methylation. A, methylation of the CYP24A1 gene promoter region in vitro. Following in vitro methylation with SssI, HpaII, or HhaI methylases, CYP24A1 promoter fragments were digested with McrBC (a methylation-specific restriction enzyme) or with HpaII or HhaI (methylation-sensitive restriction enzymes) to confirm the methylation status of the CYP24A1 promoter construct. B, dual luciferase activity assay of differentially methylated CYP24A1 promoter constructs following transfection into PC3 cells. Luminescence values were normalized to Renilla luciferase. Columns, mean from triplicate data points relative to the mean promoterless control vector; bars, SD. Results are representative of two independent experiments.

OF6 Cancer Res; 70(14) July 15, 2010 Cancer Research

Epigenetic Regulation of CYP24A1

Figure 5. CYP24A1 promoter–associated histone modifications and VDR recruitment. A, ChIP-qPCR of H3K9me2, H3K4me2, and H3K9ac at the CYP24A1 promoter. The amplification value from immunoprecipitated DNA was normalized to 5% input. Columns, mean from triplicate data points; bars, SD. B, ChIP-qPCR of H3K9me2, H3K4me2, and H3K9ac at the CYP24A1 promoter in PC3 cells treated with DAC and/or TSA. The amplification value from immunoprecipitated DNA was normalized to 5% DNA input. Columns, mean from triplicate data points; bars, SD. C, ChIP-qPCR analysis of VDR recruitment to the CYP24A1 promoter in PC3 cells treated with DAC and/or TSA. The mean values from triplicate data points are plotted.

CYP24A1 expression change (≥1.5-fold down versus other) ond, the CYP24A1 promoter is hypermethylated in human and methylation change (≥2.0-fold up versus other). Ten of prostate cell lines that weakly express CYP24A1 (LNCaP and the 15 samples with CYP24A1 downregulation in malignant PC3), whereas it is completely unmethylated in a human pros- lesions have increased CYP24A1 DNA methylation, whereas tate cell line with high CYP24A1 expression (DU145). Third, only 4 of the other 15 samples have increased CYP24A1 treatment of low CYP24A1 expression prostate cell lines with DNA methylation. Fisher's exact test shows that decreased DNA methyltransferase inhibitor induces CYP24A1 expression CYP24A1 expression is significantly associated with increased in a dose-dependent manner, which coincides with DNA CYP24A1 DNA methylation (P < 0.05). hypomethylation of the CYP24A1 promoter. However, no increased induction of CYP24A1 expression was observed with Discussion DAC treatment in DU145, which has hypomethylated CYP24A1. Furthermore, in vitro methylation of the CYP24A1 In this study, we showed that both DNA hypermethylation promoter directly represses its promoter activity. Finally, and repressive histone modifications at the CYP24A1 pro- ChIP-qPCR shows that treatment of PC3 cells with DAC in- moter were shown to repress CYP24A1 gene expression. Epi- creases the recruitment of VDR to the CYP24A1 promoter. genetic repression also impairs the recruitment of VDR to Data obtained from prostate cell lines were further supported the CYP24A1 regulatory regions, providing a plausible mech- by the analysis of human matched prostate benign and malig- anism for how the repressive epigenetic markers exert their nant lesion samples. The expression of CYP24A1 was signifi- effects on CYP24A1 expression. cantly downregulated in malignant prostate tissues compared This study indicates a direct molecular relationship be- with benign prostate tissues. This finding was consistent with tween 5′CpG island methylation and downregulation of the data obtained from ONCOMINE (23). We further show CYP24A1 expression in human prostate cancer. This conclu- that the decreased expression of CYP24A1 in malignant sion is supported by the following observations. First, the tissues was significantly associated with hypermethylation presence of a classic 5′CpG island in the CYP24A1 promoter of the CYP24A1 promoter, showing a relationship between suggests a role for DNA methylation–based regulation. Sec- 5′-CpG island methylation and reduced CYP24A1 expression

www.aacrjournals.org Cancer Res; 70(14) July 15, 2010 OF7

Luo et al.

in prostate tumor specimens. These observations suggest that histone modifications function in combination to silence im- DNA methylation may serve as a mechanism for controlling portant tumor suppressor genes in human cancers (34, 35). CYP24A1 expression in human cancers. Studies have shown that hypermethylated promoters often In the human CYP24A1 promoter, VDREs are located with- involve recruitment of methyl binding domain proteins and in the hypermethylated regions. Therefore, hypermethylation histone-modifying enzymes such as histone deacetylase (36–39). of these regions identified in this study could lead to a chro- Binding of these proteins leads to a chromatin closed state matin state in which VDR is prevented from binding to the and prevents binding of transcription factors like VDR to VDREs, which results in transcription silencing of CYP24A1. target promoters. Although evidence in the current study This model is supported by our observations that recruited and previously reports (12, 24, 25, 40) supports promoter VDR on the CYP24A1 promoter is increased in PC3 cells trea- DNA hypermethylation–mediated CYP24A1 suppression, ted with DAC or TSA. H3K9me2, a repressive histone modi- quantitative RT-PCR analysis of human prostate tissues fication, was high in LNCaP and PC3 cells, which have low reveals that the level of CYP24A1 methylation does not entirely CYP24A1 expression. In contrast, low H3K9me2 and high account for strong inverse correlation with the level of H3K9ac and H3K4me2 in DU145 cells, which have the highest CYP24A1 expression in human prostate tissues (Fig. 6C CYP24A1 expression, suggest an association of an open chro- and D). Our results suggest that histone code modifications matin structure with active gene expression (32, 33). Notably, could provide an alternative regulatory mechanism that we found that treatment of CYP24A1 hypermethylated affects CYP24A1 gene expression in vitro and in vivo (41–43). prostate cells with histone deacetylase inhibitor and DNA Analysis of the CYP24A1 gene exon 1 region by NUBIScan methyltransferase inhibitor induces CYP24A1 expression in (44) identified a putative VDRE. We found that treatment a dose-dependent manner, accompanied by an active pattern with DAC or TSA increases the recruitment of VDR to this of histone modifications. These results are consistent with putative VDRE region, which is methylated in LNCaP and the hypothesis that aberrant DNA methylation and repressive PC3 cells (Figs. 1C and 5C). These results suggest that the

Figure 6. CYP24A1 expression and promoter DNA methylation in normal and cancerous prostate tissues. A, analysis of ONCOMINE database for CYP24A1 expression in human prostate tumors (T) compared with normal prostate tissue (N). Dots represent maximum and minimum values; whiskers represent 90th and 10th percentile values; and horizontal lines represent 75th, 50th, and 25th percentile values. B, quantitative RT-PCR analysis of CYP24A1 expression in human matched prostate malignant and benign lesions. Columns, mean values of triplicate data points; bars, SD. C, comparison of CYP24A1 expression in human matched prostate malignant and benign lesions by Wilcoxon signed-rank test. D, pyrosequencing of the methylation status in human matched prostate malignant and benign lesions. Mean ± SD values of duplicate data points are indicated. Results are representative of two independent experiments.

OF8 Cancer Res; 70(14) July 15, 2010 Cancer Research

Epigenetic Regulation of CYP24A1

region of downstream of the transcriptional start site may be potentially rendering the tumor less sensitive to vitamin D. involved in the recruitment of CYP24A1 regulators. Further In addition, decreased activity of extrarenal 1α-hydroxylase studies are in progress to investigate this possibility. in prostate cancer may be also responsible for the increased Taken together, these data define promoter DNA meth- proliferation of the cancer cells (47). In conclusion, the data ylation and altered histone codes as a key mechanism obtained from this study contribute to our understanding of controlling CYP24A1 expression in human prostate cancers. the mechanisms regulating CYP24A1 gene expression in Information obtained from this study has implications for malignant and normal human tissues and facilitate the opti- developing strategies to optimize the vitamin D antitumor mization of antitumor therapy of vitamin D in combination activity. Based on the apparent role of CYP24A1 as a key en- with repression of CYP24A1 in prostate cancer. zyme to degrade the activity of vitamin D, one could envision that direct inhibition of CYP24A1 function could enhance the Disclosure of Potential Conflicts of Interest antitumor activity of vitamin D (14, 45, 46). CYP24A1 expression is heterogeneous in prostate cancer (Fig. 6B). Prostate No potential conflicts of interest were disclosed. cancer patients with high expression of CYP24A1 may not be as responsive to vitamin D analogue therapy, presumably Acknowledgments due to the ability of CYP24A1 in prostate tumors to decrease CYP24A1 We thank Drs. Alan Hutson and Song Liu for statistical assistance, Dr. 1,25(OH)2D3 levels. Decreased expression of in some Moray J. Campbell for helpful discussions, Dr. Yingyu Ma for helping with malignant prostate tumor tissue is caused, in part, by meth- figure preparation, and Rui-Xian Kong, Smitha James, and Petra Link for ylation and histone modification associated with the expert technical assistance. CYP24A1 promoter. Furthermore, Chung and colleagues reported that induction of CYP24a1 by hypomethylating agents Grant Support may attenuate responses to vitamin D in mouse tumor- NIH/National Cancer Institute grants CA067267, CA085142, and CA095045. derived endothelial cells that do not express CYP24 due to The costs of publication of this article were defrayed in part by the payment epigenetic silencing (12). Therefore, the use of DNA methyl- of page charges. This article must therefore be hereby marked advertisement in transferase inhibitors and histone deacetylase inhibitors in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. – vitamin D based therapy for prostate cancer should be Received 02/19/2010; revised 04/26/2010; accepted 05/14/2010; published avoided due to the reactivation of CYP24A1 expression, OnlineFirst 06/29/2010.

References 1. Haussler MR, Haussler CA, Jurutka PW, et al. The vitamin D hormone vitamin D-24-hydroxylase: delineation of pathways involving 1α,24- and its nuclear receptor: molecular actions and disease states. dihydroxyvitamin D2 and 1α,25-dihydroxyvitamin D2. Biochim J Endocrinol 1997;154 Suppl:S57–73. Biophys Acta 2006;1761:221–34. 2. Holick MF. Vitamin D and bone health. J Nutr 1996;126:1159–64S. 12. Chung I, Karpf AR, Muindi JR, et al. Epigenetic silencing of CYP24 in 3. DeLuca HF, Schnoes HK. Vitamin D: recent advances. Annu Rev tumor-derived endothelial cells contributes to selective growth Biochem 1983;52:411–39. inhibition by calcitriol. J Biol Chem 2007;282:8704–14. 4. Pols HA, Birkenhager JC, Foekens JA, van Leeuwen JP. Vitamin D: a 13. Swami S, Krishnan AV, Peehl DM, Feldman D. Genistein potentiates modulator of cell proliferation and differentiation. J Steroid Biochem the growth inhibitory effects of 1,25-dihydroxyvitamin D3 in DU145 Mol Biol 1990;37:873–6. human prostate cancer cells: role of the direct inhibition of CYP24 5. Chung I, Wong MK, Flynn G, Yu WD, Johnson CS, Trump DL. enzyme activity. Mol Cell Endocrinol 2005;241:49–61. Differential antiproliferative effects of calcitriol on tumor-derived 14. Yee SW, Campbell MJ, Simons C. Inhibition of vitamin D3 metabo- and Matrigel-derived endothelial cells. Cancer Res 2006;66: lism enhances VDR signalling in androgen-independent prostate 8565–73. cancer cells. J Steroid Biochem Mol Biol 2006;98:228–35. 6. Makin G, Lohnes D, Byford V, Ray R, Jones G. Target cell meta- 15. Trump DL, Muindi J, Fakih M, Yu WD, Johnson CS. Vitamin D bolism of 1,25-dihydroxyvitamin D3 to calcitroic acid. Evidence for compounds: clinical development as cancer therapy and prevention a pathway in kidney and bone involving 24-oxidation. Biochem J agents. Anticancer Res 2006;26:2551–6. 1989;262:173–80. 16. Deeb KK, Trump DL, Johnson CS. Vitamin D signalling pathways 7. Beckman MJ, Tadikonda P, Werner E, Prahl J, Yamada S, DeLuca in cancer: potential for anticancer therapeutics. Nat Rev Cancer HF. Human 25-hydroxyvitamin D3–24-hydroxylase, a multicatalytic 2007;7:684–700. enzyme. Biochemistry 1996;35:8465–72. 17. Anderson MG, Nakane M, Ruan X, Kroeger PE, Wu-Wong JR. 8. Hosogane N, Shinki T, Kasuga H, Taketomi S, Toyama Y, Suda T. Expression of VDR and CYP24A1 mRNA in human tumors. Cancer Mechanisms for the reduction of 24,25-dihydroxyvitamin D3 levels Chemother Pharmacol 2006;57:234–40. and bone mass in 24-hydroxylase transgenic rats. FASEB J 2003; 18. Bareis P, Bises G, Bischof MG, Cross HS, Peterlik M. 25-Hydroxy- 17:737–9. vitamin D metabolism in human colon cancer cells during tumor 9. Kasuga H, Hosogane N, Matsuoka K, et al. Characterization of progression. Biochem Biophys Res Commun 2001;285:1012–7. transgenic rats constitutively expressing vitamin D-24-hydroxylase 19. Friedrich M, Rafi L, Mitschele T, Tilgen W, Schmidt W, Reichrath J. Anal- gene. Biochem Biophys Res Commun 2002;297:1332–8. ysis of the vitamin D system in cervical carcinomas, breast cancer and 10. Masuda S, Byford V, Arabian A, et al. Altered pharmacokinetics of ovarian cancer. Recent Results Cancer Res 2003;164:239–46. 1α,25-dihydroxyvitamin D3 and 25-hydroxyvitamin D3 in the blood 20. Parise RA, Egorin MJ, Kanterewicz B, et al. CYP24, the enzyme that and tissues of the 25-hydroxyvitamin D-24-hydroxylase (Cyp24a1) catabolizes the antiproliferative agent vitamin D, is increased in lung null mouse. Endocrinology 2005;146:825–34. cancer. Int J Cancer 2006;119:1819–28. 11. Masuda S, Strugnell SA, Knutson JC, St-Arnaud R, Jones G. 21. Mitschele T, Diesel B, Friedrich M, et al. Analysis of the vitamin D sys- Evidence for the activation of 1α-hydroxyvitamin D2 by 25-hydroxy- tem in basal cell carcinomas (BCCs). Lab Invest 2004;84:693–702.

www.aacrjournals.org Cancer Res; 70(14) July 15, 2010 OF9

Luo et al.

22. Reichrath J, Rafi L, Rech M, et al. Analysis of the vitamin D system antagonist gene in human renal cell carcinoma cell lines. Int J Cancer in cutaneous squamous cell carcinomas. J Cutan Pathol 2004;31:224–31. 2008;123:535–42. 23. Rhodes DR, Yu J, Shanker K, et al. ONCOMINE: a cancer microarray 36. Bakker J, Lin X, Nelson WG. Methyl-CpG binding domain protein 2 database and integrated data-mining platform. Neoplasia 2004;6:1–6. represses transcription from hypermethylated π-class glutathione 24. Novakovic B, Sibson M, Ng HK, et al. Placenta-specific methylation S-transferase gene promoters in hepatocellular carcinoma cells. of the vitamin D 24-hydroxylase gene: implications for feedback J Biol Chem 2002;277:22573–80. autoregulation of active vitamin D levels at the fetomaternal interface. 37. Lopez-Serra L, Ballestar E, Fraga MF, Alaminos M, Setien F, J Biol Chem 2009;284:14838–48. Esteller M. A profile of methyl-CpG binding domain protein occu- 25. Ohyama Y, Kusada T, Yamasaki T, Ide H. Extensive methylation of pancy of hypermethylated promoter CpG islands of tumor CpG island of CYP24 gene in osteoblastic ROS17/2.8 cells. Nucleic suppressor genes in human cancer. Cancer Res 2006;66:8342–6. Acids Res Suppl 2002;249–50. 38. Nan X, Ng HH, Johnson CA, et al. Transcriptional repression by the 26. Li LC, Dahiya R. MethPrimer: designing primers for methylation methyl-CpG-binding protein MeCP2 involves a histone deacetylase PCRs. Bioinformatics 2002;18:1427–31. complex. Nature 1998;393:386–9. 27. Chen KS, DeLuca HF. Cloning of the human 1α,25-dihydroxyvitamin 39. Jones PL, Veenstra GJ, Wade PA, et al. Methylated DNA and MeCP2 D-3 24-hydroxylase gene promoter and identification of two vitamin recruit histone deacetylase to repress transcription. Nat Genet 1998; D-responsive elements. Biochim Biophys Acta 1995;1263:1–9. 19:187–91. 28. Tashiro K, Ishii C, Ryoji M. Role of distal upstream sequence in 40. Khorchide M, Lechner D, Cross HS. Epigenetic regulation of vitamin vitamin D-induced expression of human CYP24 gene. Biochem D hydroxylase expression and activity in normal and malignant Biophys Res Commun 2007;358:259–65. human prostate cells. J Steroid Biochem Mol Biol 2005;93:167–72. 29. Woloszynska-Read A, James SR, Link PA, Yu J, Odunsi K, Karpf AR. 41. Bachman KE, Park BH, Rhee I, et al. Histone modifications and DNA methylation-dependent regulation of BORIS/CTCFL expression silencing prior to DNA methylation of a tumor suppressor gene. in ovarian cancer. Cancer Immun 2007;7:21. Cancer Cell 2003;3:89–95. 30. Väisänen S, Dunlop TW, Sinkkonen L, Frank C, Carlberg C. Spatio- 42. Cedar H, Bergman Y. Linking DNA methylation and histone modi- temporal activation of chromatin on the human CYP24 gene fication: patterns and paradigms. Nat Rev Genet 2009;10:295–304. promoter in the presence of 1α,25-dihydroxyvitamin D3. J Mol Biol 43. Jenuwein T, Allis CD. Translating the histone code. Science 2001; 2005;350:65–77. 293:1074–80. 31. Milanesi L, D'Angelo D, Rogozin IB. GeneBuilder: interactive in silico 44. Podvinec M, Kaufmann MR, Handschin C, Meyer UA. NUBIScan, prediction of gene structure. Bioinformatics 1999;15:612–21. an in silico approach for prediction of nuclear receptor response 32. Zhang Y, Reinberg D. Transcription regulation by histone meth- elements. Mol Endocrinol 2002;16:1269–79. ylation: interplay between different covalent modifications of the 45. Schuster I, Egger H, Herzig G, et al. Selective inhibitors of vitamin D core histone tails. Genes Dev 2001;15:2343–60. metabolism-new concepts and perspectives. Anticancer Res 2006; 33. Martin C, Zhang Y. The diverse functions of histone lysine 26:2653–68. methylation. Nat Rev Mol Cell Biol 2005;6:838–49. 46. Schuster I, Egger H, Astecker N, Herzig G, Schussler M, Vorisek G. 34. Kondo Y, Shen L, Issa JP. Critical role of histone methylation in Selective inhibitors of CYP24: mechanistic tools to explore vitamin D tumor suppressor gene silencing in colorectal cancer. Mol Cell Biol metabolism in human keratinocytes. Steroids 2001;66:451–62. 2003;23:206–15. 47. Chen TC. 25-Hydroxyvitamin D-1 α-hydroxylase (CYP27B1) is a 35. Kawamoto K, Hirata H, Kikuno N, Tanaka Y, Nakagawa M, Dahiya R. new class of tumor suppressor in the prostate. Anticancer Res DNA methylation and histone modifications cause silencing of Wnt 2008;28:2015–7.

OF10 Cancer Res; 70(14) July 15, 2010 Cancer Research

Epigenetic Regulation of Vitamin D 24-Hydroxylase/CYP24A1 in Human Prostate Cancer

Wei Luo, Adam R. Karpf, Kristin K. Deeb, et al.

Cancer Res Published OnlineFirst June 29, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-10-0617

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2010/06/28/0008-5472.CAN-10-0617.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2010/06/23/0008-5472.CAN-10-0617. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.