Targeting DNMTs to overcome enzalutamide resistance in prostate cancer

Abstract

Prostate cancer is the second leading cause of cancer death among men in the United States. The androgen receptor (AR) antagonist enzalutamide is a FDA-approved drug for treatment of patients with late-stage prostate cancer and is currently under clinical study for early-stage prostate cancer treatment. After a short positive response period to enzalutamide, tumors will develop drug resistance. In this study, we uncovered that DNA methylation was deregulated in enzalutamide-resistant cells. DNMT activity and DNMT3B expression were upregulated in resistant cell lines. Enzalutamide induced the expression of DNMT3A and DNMT3B in prostate cancer cells with a potential role of p53 and pRB in this process. The overexpression of DNMT3B3, a DNMT3B variant, promoted an enzalutamide-resistant phenotype in C4–2B cell lines. Inhibition of DNA methylation and DNMT3B knockdown induced a re-sensitization to enzalutamide. Decitabine treatment in enzalutamide-resistant cells induced a decrease of the expression of AR-V7 and changes of genes for apoptosis, DNA repair and mRNA splicing. Combination treatment of Decitabine and enzalutamide induced a decrease of tumor weight, Ki-67 and AR-V7 expression and an increase of cleaved-caspase3 levels in 22Rv1 xenografts. The collective results suggest that DNA methylation pathway is deregulated after enzalutamide resistance onset and that targeting DNA methyltransferases restores the sensitivity to enzalutamide in prostate cancer cells.

INTRODUCTION

Prostate cancer (PCa) is the top diagnosed cancer and the second leading death-causing cancer in American men (1). Patients with prostate cancer are well managed in the early stages of the disease. Prostate cancer is driven by androgen receptor (AR) signaling pathway which is activated upon the binding of the ligand, dihydrotestosterone, to the receptor, resulting in regulation of downstream gene expression (2). Approaches targeting the androgen signaling pathway have been used in the clinic to treat prostate cancer patients (3). However, after a short-lived response to androgen deprivation therapies, patients become resistant to these approaches, leading to a more aggressive form of the disease ending in death. Enzalutamide, an AR antagonist, is FDA-approved for the treatment of prostate cancer patients across different stages of the disease (4). Although enzalutamide showed benefits in treating post-chemotherapy CRPC patients, almost all will become nonresponse to the antiandrogen, resulting in limitations of their options (5). It is important to study and identify the underlying mechanisms by which prostate cancer cells develop resistance to enzalutamide. These efforts will allow researchers to discover biomarkers of resistance predisposition and develop advanced therapeutic targets.

DNA methylation is an epigenetic mark, deposited on cytosine molecules by a family of DNA methyltransferases (DNMTs). In promoter regions, methylcytosines promote gene silencing by interplaying with transcription factors and components of the chromatin remodeling machinery (6–8). DNA methyltransferases are heavily spliced, resulting in a complicated network of proteins implicated in this process (9–11). In prostate, DNA methylation changes are higher in prostate cancer samples than those in normal samples (12–14). Many tumor suppressor genes, such as APC, RASSF1A, GSTP1, and INK4A, have been found to be hypermethylated and silenced during prostate cancer initiation and progression (15,16). However, its role in enzalutamide resistance is unexplored and poorly understood. In this study, we aim to investigate the role of DNA methylation in enzalutamide resistance by exploring the transcriptome and methylome of enzalutamide-resistant prostate cancer cells. We also aim to explore the effect of enzalutamide on DNA methyltransferases and the role of these enzymes and their splice variants in the response to enzalutamide.

MATERIALS AND METHODS

Mammalian cell lines

LNCaP and C4–2B were used as enzalutamide-sensitive cell lines, whereas MR49F, C4–2B-MDVR and 22Rv1 were considered enzalutamide-resistant cell lines. For this study, LNCaP and C4–2B were paired with their enzalutamide-resistant counterparts MR49F and C4–2B-MDVR, respectively. 22Rv1 was used as an independent cell line with no enzalutamide-sensitive pair. The LNCaP and 22Rv1 lines were purchased from ATCC. ATCC uses short tandem repeat (STR) profiling for testing and authentication of cell lines. The C4–2B line was acquired from the M.D. Anderson Cancer Center. The identity of C4–2B cell line was verified by the “Characterized Cell Line Core Facility,” MD Anderson Cancer Center, through short tandem repeat DNA profiling. Enzalutamide-resistant lines MR49F was gifted from Dr. Amina Zoubeidi at the Vancouver Prostate Cancer Center and the cell line was tested and authenticated by whole-genome and whole-transcriptome sequencing on Illumina Genome Analyzer IIx platform (17). C4–2B-MDVR was gifted from Dr. Allen Gao at the University of California Davis (18). Short Tandem Repeat (STR) profiling, provided by ATCC was used for testing and authentication of C4–2B-MDVR cell line. All cell lines were maintained in RPMI 1640 medium supplemented with 10% (v/v) Fetal Bovine Serum at 37°C in a humidified incubator with 5% carbon dioxide. MR49F and C4–2B-MDVR were constantly maintained in medium supplemented with 10 and 20 μM enzalutamide, respectively. All cells were within 50 passages and Mycoplasma were detected every 3 months using MycoAlert PLUS Mycoplasma Detection Kit (Lonza, LT07–705).

MTT Assay

To evaluate cell viability, plates were primed with Poly-D-lysine hydrobromide (purchased from Millipore Sigma), followed by seeding of cells at densities ranging between 1000 to 3000 cells/well. On the following day, cells were transfected or treated with respective drugs. At the end of incubation time, cells were then treated with 0.5 mg/ml MTT (3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) for an hour. After supernatant was removed, DMSO was added to dissolve crystals, followed by measurement of optical density at a wavelength of 570 nm.

22Rv1 mouse xenografts

Castrated NSG mice were subjected to a right flank subcutaneous inoculation of 7.5 × 10⁵ 22Rv1 cells suspended in PBS and mixed with Matrigel (1:1). Tumor-bearing mice were randomly assigned into 4 groups: vehicle, decitabine, enzalutamide and combination. Mice received treatment on a 4-day ON, 2-days OFF regimen for 4 cycles. Mice received the vehicle and the enzalutamide 20 mg/kg/day injections through oral gavage and the vehicle and the decitabine 0.5 mg/kg/day injections through intraperitoneal injections. Upon necropsy, tumor size, tumor weight and body weight were measured. All the animal experiments described in this study were approved by the Purdue University Animal Care and Use Committee.

Immunohistochemistry

Slides were incubated with 3% hydrogen peroxide in water for 5 minutes, rinsed with TBST and incubated in 2.5% normal goat serum for 20 minutes. Excess reagent was blown off and either Ki67 (Cell Marque, 275R-16) at a dilution of 1:100 (0.364 μg/mL) or Cleaved Caspase 3 (Cell Signal Tech, 9661) at 1:200 (2.6 μg/mL) were incubated with the slides for 30 minutes. The negative control slide was stained with Rabbit IgG (Vector Labs, I-1000) at a concentration of 1:5000 (1 μg/mL) for 30 minutes. Slides were rinsed twice in TBST and a goat anti-rabbit secondary (Vector Labs, MP-7451) was applied for 30 minutes. Slides were rinsed twice in TBST and Vector ImmPACT DAB (Vector Labs, SK-4105) was applied for 5 minutes. Slides were rinsed in water and transferred to a Leica Autostainer XL for hematoxylin counterstain, dehydration and coverslipping.

DNMT activity assay

LNCaP, MR49F, C4–2B and C4–2B-MDVR cells were seeded after being removed from enzalutamide maintenance medium for one passage. LNCaP and C4–2B cells were then treated with DMSO, 5 μM and 10 μM enzalutamide. After 5 days, cells were collected and washed once with PBS. Cell pellets were subjected to the NE-PER^™ Nuclear and Cytoplasmic Extraction kit where cytoplasmic and nuclear fractions were separately isolated. 10 μg of nuclear extracts were used as input amount for the EpiQuik^™ DNMT Activity/Inhibition ELISA Easy Kit (Colorimetric). Samples were processed according to the manufacturer’s instructions and optical density was measured at a wavelength of 450 nm.

DNA methylation quantification assay

LNCaP, MR49F, C4–2B and C4–2B-MDVR cells were seeded after being removed from enzalutamide maintenance medium for one passage. LNCaP and C4–2B cells were then treated with DMSO, 5 μM and 10 μM enzalutamide. After 5 days, cells were collected and washed once with PBS. Cell pellets were subjected to the DNeasy Blood & Tissue kit where genomic DNA was extracted. 200 ng of genomic DNA were used as input amount for the MethylFlash^™ Methylated DNA Quantification Kit (Colorimetric). Samples were processed according to the manufacturer’s instructions and optical density was measured at a wavelength of 450 nm.

Statistical analysis

Standard 2-tailed Student t tests and One-Way ANOVA were performed to analyze statistical significance of the results. Two-way ANOVA tests were performed to analyze statistical significance of datasets with grouped observations. A p-value of less than 0.05 indicates statistical significance. All graphs and statistical analyses were completed on Prism Graphpad.

Data Availability

The RNA sequencing data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE184168.

More details can be found in Supplementary Materials and Methods

RESULTS

Enzalutamide treatment induces an increase of DNMT activity and DNA methylation in enzalutamide-resistant PCa cell lines

In our previous study, we compared gene expression in enzalutamide-sensitive and -resistant cell lines by RNA sequencing (19). We observed a higher number of downregulated genes in the resistant cell lines than those in sensitive cell lines. These data suggested that a repressive mechanism might play a role in promoting transcriptional silence of genes upon the onset of resistance. To investigate whether DNA methylation plays a role in enzalutamide resistance in PCa cells, we first investigated DNMT activities in a panel of enzalutamide-sensitive and -resistant PCa cell lines. Nuclear extracts of LNCaP, MR49F, C4–2B, and C4–2B-MDVR were subjected to DNMT activity assay. Results showed a significant and consistent increase of DNMT activities in MR49F (4.67 OD/h/mg) and C4–2B-MDVR (9.35 OD/h/mg) compared to their parental lines, respectively (Fig. 1A). To investigate whether enzalutamide is directly involved in the increase of DNMT activity, LNCaP and C4–2B cells were treated with enzalutamide for 5 days. As shown, enzalutamide treatment promoted significant increases of DNMT activities in LNCaP and C4–2B cells (Figs. 1B and 1C). These results suggest that enzalutamide treatment induces an increase of DNMT activity in PCa cells. To investigate the levels of DNA methylation, we quantified the levels of 5-mC in those cell lines. The data showed that MR49F (2.18%) had a higher level of global DNA methylation than that in LNCaP cells (1.23%) (Fig. 1D). We also observed a very limited increase of global 5-mC in C4–2B-MDVR versus C4–2B (2.13% vs. 1.86%) (Fig. 1D). To test whether enzalutamide treatment can affect global DNA methylation, LNCaP and C4–2B were treated with enzalutamide for 5 days, followed by measurement of 5-mC level. The results indicated an increase of the levels of 5-mC in LNCaP cells. However, only treatment with 5 μM enzalutamide induced a significant increase of DNA methylation (1.23% vs. 2.35%) (Fig. 1E). No significant changes were observed in global DNA methylation in C4–2B post-enzalutamide treatment at both doses (Fig. 1F). These results suggest that the association between DNMT activity and global DNA methylation can be observed in androgen-dependent cell line but not in androgen independent cell line, implicating discrepancies in the regulation of DNMTs and DNA methylation in these cell lines.

Figure 1. DNA methyltransferase activity and DNMT3B are overrepresented in enzalutamide resistant cell lines.

A, D, G-J, LNCaP, MR49F, C4–2B, C4–2B-MDVR and 22Rv1 cells were seeded and cultured for 5 days in regular medium. B, C, E and F, LNCaP and C4–2B cells were treated with DMSO, 5 or 10 μM Enzalutamide for 5 days. A-C, Nuclear extracts were collected and DNMT activity was measured. D-F, Genomic DNA was collected and DNA methylation was measured. G-I, RNA was extracted and followed by cDNA synthesis. Transcript levels of DNMT1, DNMT3A and DNMT3B were measure using qRT-PCR. J, Cells were subjected to protein extraction and followed by western blot. Two-tailed student t-test was used to evaluate statistical significance *p<0.05 and **p<0.01.

DNMT3B is the predominant DNA methyltransferase overexpressed in enzalutamide-resistant PCa cell lines

Changes of DNA methyltransferases directly correlate with cellular DNMT activity and DNA methylation (20). DNMT activity directly correlates with the expression of DNMT1 and/or DNMT3B in primary and metastatic transgenic adenocarcinoma of the mouse prostate (TRAMP) mice tumors (21). To understand the underlying mechanisms contributing to the increase of DNMT activity in enzalutamide-resistant cell lines, we used qRT-PCR and western blot to study the transcript and protein levels of these enzymes in LNCaP, MR49F, C4–2B, C4–2B-MDVR, and 22Rv1 cell lines. Transcript levels of DNMT1 were significantly increased in MR49F compared to LNCaP cells (Figs. 1G). The increase of DNMT3A is mild in C4–2B-MDVR in comparison with C4–2B cells (Figs. 1H). DNMT3B was observed to be upregulated in MR49F and C4–2B-MDVR compared with their parental lines (Figure 1I). For the protein level, DNMT1 expression was not significantly changed between the sensitive and resistant cell lines (Fig. 1J). However, DNMT3B was significantly upregulated in resistant cell lines (Fig. 1J). DNMT3B7, a DNMT3B variant, was also significantly upregulated in resistant cell lines compared to their sensitive counterparts. The DNMT3A variant, DNMT3A1, was significantly overexpressed in MR49F compared to LNCaP cells (Fig. 1J). No significant changes were observed for the expressions of DNMT3A1 and DNMT3A2 in C4–2B versus C4–2B-MDVR cells. These data support a correlation between enzalutamide resistance and the overexpression of DNMT3B, indicating the role of DNMT3B activity in enzalutamide resistance. In addition, upregulation of DNMT3A projects another layer of complexity in the comparisons between LNCaP and MR49F.

Enzalutamide treatment and AR knockdown induce expressions of DNMT3A and DNMT3B in PCa cells

The effect of enzalutamide on the expression of DNA methyltransferases is poorly studied. To understand the mechanism contributing to the increase of DNMT activity upon enzalutamide treatment, we assessed changes of the expression of DNMTs after AR knockdown and treatment with enzalutamide. LNCaP, MR49F, C4–2B, and C4–2B-MDVR cells were transfected with siRNAs targeting a scrambled sequence (siCtrl) or AR (siAR). Then, transfected cells were treated with DMSO or 5 μM enzalutamide for 3 days. The results showed that enzalutamide treatment alone induced an increase of the expression of DNMT3A and DNMT3B but not DNMT1 (Figs. 2A and 2B). Additionally, knockdown of AR induces an increase of the levels of DNMT3A and DNMT3B (Figs. 2A and 2B). These results suggest that targeting AR induces the expression of DNMT3A and DNMT3B, accounting for the increase of DNMT activity observed in LNCaP and C4–2B. All the data in figures 1 and 2 suggest that enzalutamide treatment induces overexpression of different isoforms of methyltransferases, resulting in enzalutamide resistance in prostate cancer in turn.

Figure 2. Enzalutamide induces and increase in the expression of DNMT3B and DNMT3A in prostate cancer cells.

A, LNCaP, MR49F, C4–2B and C4–2B-MDVR cells were transfected with siCtrl or siAR, treated with DMSO or 5 μM Enzalutamide for 3 days, and subjected to protein extraction, followed by western blot B, Band intensity of DNMT3A and DNMT3B was quantified using ImageJ. C, Total RNA was extracted from LNCaP and C4–2B cells treated with DMSO and 5 μM Enzalutamide and followed by qRT-PCR. Two-tailed student t-test was used to evaluate statistical significance *p<0.05, **p<0.01 and ***p<0.001.

Tumor suppressors pRB and p53 are directly involved in the regulation of the expression of DNA methyltransferases in mammalian cells. While pRB and p53 binding to and inhibiting the activity of SP1/SP3 and E2F1, the expression of SP1/SP3 and E2F1 directly correlates with the expression of DNMTs (22). To understand the mechanisms by which DNMT3A and DNMT3B expression is induced upon enzalutamide treatment, we investigated the expressions of E2F1, SP3, MDM2, and pRB. Our data showed that MDM2 protein levels were elevated in C4–2B-MDVR compared to C4–2B cells, but its expression in MR49F was lower than that in LNCaP cells (Fig. S1A). E2F1 expression was consistently increased in MR49F and C4–2B-MDVR cells compared to the parental lines (Fig. S1A). To test whether this pathway is directly involved in the enzalutamide-mediated overexpression of DNMT3B, we treated C4–2B-MDVR cells with enzalutamide and screened for any changes of MDM2. Enzalutamide-treatment induced an increase of MDM2 accompanied with an increase of DNMT3B (Fig. S1B). We then tested whether the increase of DNMT3B was mediated by MDM2-p53 interaction. MR49F cells were incubated with 5 μM enzalutamide, followed by treatment with the MDM2-p53 interaction inhibitor Nutlin-3a. The data showed that the increase of DNMT3B was abolished by the inhibition of MDM2 in MR49F cells (Fig. S1C). These results suggest that the enzalutamide-mediated increase of DNMT3B and the overall upregulation of DNMT3A and DNMT3B in enzalutamide-resistant cells are mediated by p53.

Ectopic expression of DNMT3B3 promotes enzalutamide resistance in PCa cells

The roles of DNMT3B variants are poorly understood in the context of prostate cancer. To understand the functions of different DNMT3B variants in response to enzalutamide treatment, we first compared the expression of DNMT3B1, 3, 4, 5, 7 and 8 in enzalutamide-resistant cells versus -sensitive cells. The comparisons revealed similar observations regarding the expression of the different DNMT3B isoforms investigated. DNMT3B2 expression was reduced in both MR49F and C4–2B-MDVR compared to LNCaP and C4–2B, respectively (Figs. S2A and S2B). In addition, both DNMT3B3 and DNMT3B7 showed a consistent and significant increase in the resistant cell lines compared to their sensitive counterparts (Figs. S2A and S2B). To test whether overexpression of DNMT3B3 and DNMT3B7 are directly involved in the response to enzalutamide treatment, C4–2B cells were transfected with different DNMT3B isoforms (Fig. S2C). Transfected C4–2B cells were then exposed to increasing doses of enzalutamide and cell viability was assessed after 6 days. The results show that C4–2B cells overexpressing DNMT3B3 or DNMT3B3 plus DNMT3B7 exhibits a significant decrease in sensitivity to enzalutamide compared to cells transfected with the empty vector (Fig. S2D). The differences in cell viability were observed starting from a concentration of 1 μM enzalutamide. These data suggest that overexpression of DNMT3B3 may play a direct role in the response to enzalutamide and promotes an enzalutamide-resistant phenotype in PCa cells.

Decitabine treatment reestablishes the response to enzalutamide in enzalutamide-resistant PCa cell lines

Decitabine, a DNA methylation inhibitor, has been historically investigated at high doses, where it has been shown to induce DNA damage, leading to its cytotoxic effect in mammalian cells (23,24). To eliminate some of the cytotoxic effects of decitabine in our study, we used decitabine at sub-micromolar concentrations. We first asked whether DNA methylation inhibition can restore the sensitivity to enzalutamide in the resistant cells. To this end, we assessed the apoptotic response of a pre-treatment with decitabine followed by exposure to enzalutamide. LNCaP, MR49F, C4–2B, and C4–2B-MDVR were pre-incubated with 0, 250 or 1000 nM decitabine for 5 days, followed by a treatment with DMSO, 10, 20 or 40 μM of enzalutamide for another 3 days. Pre-treatment with 250 and 1000 nM decitabine enhances the apoptotic response to enzalutamide in C4–2B-MDVR cells, demonstrated by a significant increase of cleaved-PARP and cleaved-caspase 3 levels upon double-treatment compared to single-treatment (Fig. 3A). In MR49F cells, a similar response was observed in the group treated with the highest concentration of decitabine. In contrast, LNCaP and C4–2B cells didn’t exhibit significant increases in the levels of the cleaved apoptotic proteins assessed. Pre-treatment with decitabine induced a significant decrease in the expression of DNMT1, DNMT3A and DNMT3B in all cell lines. These results suggest that inhibition of DNA methylation in enzalutamide-resistant cells induces an increase of the sensitivity to enzalutamide.

Figure 3. Inhibition of DNA methylation in enzalutamide-resistant cells restores response to enzalutamide.

A, The indicated cells were pre-treated with 0, 250 and 1,000 nM decitabine for 5 days, then incubated with DMSO (0) or 10, 20 or 40 μM enzalutamide for 3 days, and harvested for western blot B, The indicated cells were seeded in 96 well plates, treated with DMSO, decitabine, enzalutamide or the combination of decitabine and enzalutamide, and subjected to MTT assay Two-way ANOVA was used to evaluate statistical significance ***p<0.001.

To investigate the effect of decitabine treatment in combination with enzalutamide on cellular proliferation, LNCaP MR49F, C4–2B, and C4–2B-MDVR cells were subjected to MTT assays. Our results show that the exposure of resistant cells to decitabine plus enzalutamide induces a significant decrease in cell proliferation compared to exposure to single agents (Fig. 3B). C4–2B cells exhibited a similar response to the enzalutamide-resistant cells. However, in LNCaP cells we didn’t see any significant differences between the groups treated with the combination compared to the groups treated with single agent (Fig. 3B). Overall, these data suggests that the combination of enzalutamide and decitabine promotes apoptosis of enzalutamide-resistant cells.

Knockdown of DNMT3B reestablishes the response to enzalutamide in enzalutamide-resistant PCa cell lines

To examine the involvement of the individual DNA methyltransferases in response to enzalutamide, cells were transfected with siRNAs, and followed by treatment with enzalutamide. Knockdown of DNMT3B combined with enzalutamide treatment increased the levels of cleaved-PARP exclusively in MR49F and C4–2B-MDVR cells (Fig. 4A). The knockdown of DNMT1 and DNMT3A did not produce an increase of the levels of the cleaved apoptotic protein. In LNCaP and C4–2B cells, knockdown of DNMT3B alone produced a significant increase of cleaved-PARP levels (Fig. 4A). These results suggest that DNMT3B plays a critical role in antagonizing enzalutamide-mediated apoptosis in the resistant cells. Also, we observed the most significant decrease of cell viability and proliferation in MR49F and C4–2B-MDVR cell lines upon knockdown of DNMT3B and treatment with enzalutamide (Fig. 4B).

Figure 4. The combination of DNMT3B knockdown and enzalutamide induces an increase in apoptosis and a decrease in cell growth in MR49F and C4–2B-MDVR cells.

A, The indicated cells were transfected with siCtrl, siDNMT1, siDNMT3A and siDNMT3B, treated with DMSO (0) or 20 or 40 μM enzalutamide for 3 days, and harvested for western B, The indicated cells were seeded in 96 well plates, transfected with siCtrl, siDNMT1, siDNMT3A and siDNMT3B, treated with 10 μM enzalutamide, and subjected to MTT assay Two-way ANOVA was used to evaluate statistical significance ***p<0.001.

To better understand which of the DNA methyltransferases play(s) a major role in the resistance to enzalutamide, MR49F and C4–2B-MDVR cells were transfected with siRNA against scrambled sequence, DNMT1, DNMT3A and DNMT3B, followed by treatment with increasing doses of enzalutamide. The results showed that knockdown of DNMT3B induced significant decrease of cell proliferation in response to increasing doses of enzalutamide (Figs. S3A and S3B). The collective results suggest that DNMT3B plays a pivotal role in the response to enzalutamide in the resistant cells.

Decitabine treatment decreases the levels of AR-V7 in enzalutamide-resistant cell lines

Studies have previously shown that the promoter of the AR gene can be methylated to regulate expression of AR-V7 in different prostate cancer cells (25,26). Therefore, we explored the effect of the combination treatment on the AR signaling pathway in enzalutamide-resistant cell lines. We treated C4–2B-MDVR and 22Rv1 with decitabine and enzalutamide. After 5 days of treatment, protein levels of key players in the AR signaling pathway were examined. Decitabine treatment induced a consistent decrease of AR-V7 expression in both C4–2B-MDVR and 22Rv1 cell lines (Figs. S3C and S3D). In addition, a decrease of AR expression was observed in in both cell lines. Of interest, decitabine also induced the expression of PSA in C4–2B-MDVR and 22Rv1 cell. To further investigate the role of DNMT3B in AR-V7 expression, we depleted DNMT3B, followed by qRT-PCR to assess transcriptions of AR-V7 and its target gene, UBE2C. As shown in Fig. S3E, AR-V7 and UBE2C were significantly reduced upon DNMT3B depletion. These results suggest that treatment with decitabine may increase the response to enzalutamide by decreasing the expression of AR and AR-V7 in these cells.

Decitabine suppresses tumor growth of 22Rv1-derived xenograft

To investigate the effect of decitabine combined with enzalutamide on tumor growth in vivo, castrated NSG mice were inoculated with 22Rv1 cells and treated for 3 weeks. We observed a significant decrease of the tumor volume in mice treated with enzalutamide and decitabine compared to the vehicle and the single-treatment groups (Fig. 5A). The combination ofenzalutamide with decitabine induced a significant decrease of tumor weight (Figs. 5B, 5C). There were no significant toxicities induced by the treatments shown (Fig. 5D). Decitabine treatment induced a significant decrease of the expressions of DNMT1, DNMT3A and DNMT3B, as expected. AR-V7 expression in the group treated with enzalutamide and decitabine was significantly lower than that in the vehicle group (Figs. 5E, 5F). No significant changes were observed for AR (Fig. 5E). However, the expression of AR was slightly lower in the combination-treated group than that in the vehicle group.

Figure 5. Inhibition of DNA methylation in combination with enzalutamide induces a decrease in tumor growth in a 22Rv1 xenograft model.

Pre-castrated NSG mice (n = 6 per group) were inoculated with 22Rv1 cells (0.75 × 10⁶), treated with vehicle control, 0.5 mg/kg decitabine, 20 mg/kg enzalutamide, or the combination of decitabine and enzalutamide at the frequency of treatment 4 days on and 2 days off. A, Tumor volume was measured every 3 days. B, Tumor weight was measured using a standard scale upon necropsy. C, A representative picture of the tumors upon necropsy. D, General toxicity was assessed by body weight measurement at day 22 before necropsy. E, Proteins were extracted from tumors followed by western blot Two-way ANOVA was used to evaluate statistical significance *p<0.05, **p<0.01 and ***p<0.001.

Immunohistochemistry analyses showed an increase of the number of apoptotic cells, reflected by the increase in the intensity of cleaved-caspase3 staining, in the combination treatment group (Figs. S4A and S4B). In addition, Ki-67 staining was significantly decreased in the combination group, indicating a decrease of cellular proliferation in the tumors (Figs. S4A and S4C). These in vivo data recapitulate our observations in vitro. These observations suggest that inhibiting the DNA methylation with decitabine, in combination with enzalutamide can prohibit tumor growth by increasing cell apoptosis and inhibiting cell proliferation.

Decitabine reverses the expression of key genes in the enzalutamide-resistant cell line

To study the impact of DNMT3B in enzalutamide resistance, we designed an RNA sequencing experiment depicted in Fig. 6A. In this experiment, we compared enzalutamide-sensitive cell lines, LNCaP and C4–2B to the enzalutamide-resistant cell lines, C4–2B-MDVR as comparison 1 (C1) and comparison 2 (C2), respectively (Table S1). Then we compared untreated C4–2B-MDVR cells to C4–2B-MDVR cells treated with decitabine or decitabine plus enzalutamide as comparison 3 (C3) and comparison 4 (C4), respectively (Table S1). In these comparisons, we explored the changes of gene expression upon different treatments. This approach allowed us to identify high interest genes and pathways involved in responses to treatments. The expression of these highly interest genes were explored in C3 and C4 to identify any reversal in gene expression compared to C1 and C2. These analyses allowed us to identify a subset of genes that changed in resistant cells compared to sensitive cells after decitabine treatment. The identification of these players is important to understand underlying mechanisms of resistance to enzalutamide in prostate cancer.

Figure 6. Bioinformatics analyses to identify genes and pathways changed during enzalutamide-resistance development.

A, Scheme depicting the comparison between enzalutamide-sensitive and enzalutamide resistant cells and the comparison between enzalutamide-resistant cells pre- and post-decitabine treatment using RNA-sequencing. B, Venn diagram showing an overlap of DE (FDR ≤ 0.05) genes between DESeq2 and edgeR methods. C & D, Decitabine treatment of C4–2B-MDVR cells reverses the enrichment of critical gene sets within the HALLMARK set of genes. While C shows GSEA from the HALLMARK gene sets in the C2 comparison (C4–2B vs. C4–2B-MDVR), D is GSEA from the HALLMARK gene sets in the C3 comparison (C4–2B-MDVR DMSO vs. C4–2B-MDVR 500 nM Dec). E & F, Decitabine treatment of C4–2B-MDVR cells reverses the expression of key genes with potential role in enzalutamide resistance. E, Heatmap depicting the expression of genes that have a log2FC>1 in C1 and C2 and a log2FC<−1 in C3. F, Heatmap depicting the expression of genes that have a log2FC<−1 in C1 and C2 and a log2FC>1 in C3. Gene sets were separated in Excel followed by heatmap generation on Diplayr.com.

We first performed a Gene Set Enrichment Analysis and explored the hallmark gene set collection in C2 and C3 (Fig. 6B). GSEA analyses showed significant enrichment of signature genes accounting for enzalutamide resistance (Fig. 6C) and induced by decitabine treatments (Fig. 6D). Our results show a reversal in the enrichment of gene signatures upon decitabine treatment. Genes that were positively enriched in enzalutamide-sensitive cells, such as HALLMARK_MYC_TARGETS_V1, HALLMARK_MYC_TARGETS_V2, HALLMARK_E2F_TARGETS and HALLMARK_G2M_CHECKPOINT, became negatively enriched in resistant cells after treatment with decitabine (Figs. 6C and 6D). In contrast, HALLMARK_ANDROGEN_RESPONSE gene set was found to be negatively enriched in enzalutamide-resistant cells. Decitabine treatment reversed this enrichment and promoted an overexpression of androgen response genes. These data suggest that decitabine treatment may result in an increased sensitivity to enzalutamide. Moreover, gene sets such as HALLMARK_APOPTOSIS and HALLMARK_DNA_REPAIR were not significantly changed between sensitive and resistant cells. However, decitabine treatment stimulated the overexpression of apoptotic genes in resistant cells and induced a decrease of DNA repair genes (Fig. 6D). These observations suggest that the genes are involved in the decitabine-mediated response to enzalutamide in resistant cells.

To narrow down the list of genes, we isolated two subsets of genes depicted in the heatmaps in Figs. 6E and 6F. In Fig. 6E, the heatmap represents 74 genes in C1 and C2 with Log2FC>1 and in C3 with Log2FC<−1. In Fig. 6F, the heatmap represents 307 genes in C1 and C2 with Log2FC<−1 and in C3 with Log2FC>1. Among genes shown in Fig. 6E, we identified ABCB11, MYB, WNT10B, WT1, MCM7, BRCA1 and PAX1 those are known to play a pro-survival and pro-resistance role in cancer. In contrast, in Fig. 6F, we discovered ALPK2 and AIFM2 that play a pro-apoptotic role in cancer and may be potentially involved in promoting an enzalutamide-resistant phenotype in prostate cancer cells. These observations were validated by qRT-PCR in comparison between enzalutamide-resistant and –sensitive cell lines (Fig. S5). The bioinformatics data support the results in vitro and in vivo.

DISCUSSION

With the increasing use of enzalutamide for the treatment of prostate cancer and breast cancer patients, it is vital to understand the mechanisms by which cells develop resistance to this anti-androgen. In this study, we investigated the role of DNA methyltransferases in the resistance to enzalutamide and the effect of enzalutamide treatment on DNA methyltransferases. It is known that dynamic changes in DNA methylation and DNMT expression occur upon prostate cancer initiation and progression (27). It has been shown that bicalutamide, a first generation anti-androgen, induces changes in DNA methylation and an increase in the expression of DNMTs in prostate cancer cell lines (28). However, the effect of enzalutamide on DNA methylation and DNMT expression are still unknown. More interestingly, the functional impact of changes of DNA methylation are not studied in the context of established enzalutamide resistance.

In comparison of enzalutamide-resistant cell linesto their sensitive counterparts, we found a significant increase of DNMT activity (Fig. 1A). The increase in DNMT activity resulted in an increase of global DNA methylation (Fig. 1D). By evaluating the expression of different DNMT3B variants, we identified an overexpression of DNMT3B3 and DNMT3B7 in enzalutamide-resistant cell lines. DNA methyltransferases are alternatively spliced or driven by alternative promoters in mammalian cells, leading to the identification of more than 30 different isoforms (11). “mRNA splicing” was identified as the top 1 overrepresented REACTOME pathway. Many genes were involved in mRNA splicing, including Vezf1 and RBM. Vezf1 expression was higher in enzalutamide-resistant than that in –sensitive cell lines. Vezf1 increases the expression of the alternatively spliced DNMT3B variants by binding to the intronic regions of DNMT3B and promoting the accumulation of RNA-pol II (29). In addition, RBM involved in alternative splicing are deregulated in the C1 and C2 comparisons. RBM10 regulates the expression of DNMT3B isoforms in mouse embryonic fibroblasts (30). Other splicing factors, such as Splicing Factor (SR) proteins and Serine and Arginine Rich Splicing Factor (SRSF) proteins are deregulated in the resistant cell lines. These observations suggest that the deregulation of the mRNA splicing machinery in enzalutamide-resistant cells accounts for the changes in the expression DNMT3B variants.

Studies have shown the involvement of the MDM2/p53/SP1-SP3 axis and the MDM2/pRB/E2F1 axis in the regulation of the expression of DNMT1, DNMT3A and DNMT3B in mammalian cells (22). MDM2, an E3 ubiquitin ligase, promotes the ubiquitination and proteasomal degradation of p53 and pRB (31,32). p53 and pRB inhibit the transcriptional activation function of SP1/SP3 and E2F1 (31). We investigated the involvement of these pathways in the increased expression of DNMT3A and DNMT3B. Our results showed an increase of the expression of E2F1 in the resistant compared to the sensitive cell lines (Fig. S1A). The increase of E2F1 expression in MR49F and C4–2B-MDVR correlates with the expression of DNMT3B. Additionally, our bioinformatics results revealed an increase of Wilm’s Tumor 1 (WT1) expression in C1 and C2 comparisons. WT1 promotes the transcriptional activation of DNMT3A in mammalian cells (33). It’s not clear whether WT1 has any role in the regulation of DNMT3B expression in prostate cancer cells.

The effect of enzalutamide on DNA methylation and expression and activity of DNMT was investigated in enzalutamide-sensitive cell lines. Our findings suggest that enzalutamide promotes an increase of the activity of DNMTs in LNCaP and C4–2B cells (Figs. 1B and 1C). The increase of DNMT activity in LNCaP correlates with an increase of global 5-mC levels after enzalutamide treatment (Fig. 1E). Enzalutamide treatment induced increases of DNMT3B and DNMT3A (Figs. 2A–2C). Enzalutamide induced increases of the expression of DNMT3B1, DNMT3B3 and DNMT3B7 (Fig. 2C). These results suggest that long-term enzalutamide treatment induces an increase of the expression of DNMT3A and DNMT3B, of which DNMT3B becomes stably overexpressed upon the onset of enzalutamide resistance.

We then investigated the mechanism accounting for the increase of DNA methyltransferase expression following enzalutamide treatment. Treatment of C4–2B-MDVR cells with enzalutamide promotes an increase of MDM2, which directly correlates with the expression of DNM3B (Fig. S1B). Nutlin-3a inhibits the binding of MDM2 to p53, promoting the activity of the tumor suppressor (34). Treatment of MR49F cells with Nutlin-3a abolished the enzalutamide-induced increase of DNMT3B in the cells. Collectively, these results suggest a role for p53 in the enzalutamide-regulated DNMTs in prostate cancer cells. However, additional experiments need to be done to confirm the direct involvement of p53 in this response and eliminate any alternative effects.

The effect of DNMT3B variants on enzalutamide response are not studied. DNMT3B3 and DNMT3B7 are catalytically inactive forms and their expression correlates with an aberrant DNA methylation profile (35, 36). The inactive DNMT3B isoform, DNMT3B3, has been shown to act as a binding partner of DNMT3A and DNMT3B (36). The accessory role of DNMT3B3 is believed to be associated with the recruitment of DNA methyltransferases to genomic sites (36). DNMT3B7 expression correlates with the tumorigenesis and cancer aggressiveness across different tissue types (37,38). The functional significance of the increased expression of DNMT3B3 and DNMT3B7 was investigated in C4–2B cells. Specifically, the impacts of DNMT3B3 and DNMT3B7 on enzalutamide response were assessed. Enzalutamide treatment of C4–2B cells overexpressing DNMT3B3 and DNMT3B3 plus DNMT3B7 had a significantly lesser impact on cell proliferation. These results show that mostly DNMT3B3 plays a pro-tumorogenic role in prostate cancer cells by promoting enzalutamide resistance.

The increase of DNMT activity and level in enzalutamide-resistant cell lines prompted us to investigate their roles in promoting resistance to the anti-androgen. DNA methyltransferases have been shown to promote drug resistance in cancers including prostate cancer (39,40). Researchers have previously shown that treatment with the cytidine analog, azacytidine, restores the sensitivity and reverses drug resistance in a number of cases (39,41). In agreement, our results showed that decitabine plus enzalutamide treatment induced an increase of apoptosis (Fig. 3A). These results correlate with the findings from RNA-sequencing data. In C3, the GSEA results suggested an enrichment of apoptotic gene signatures in C4–2B-MDVR cells upon decitabine treatment (Fig. 6D). Furthermore, pro-apoptotic proteins AIFM2 and ALPK2 are include in the genes exhibiting a significant increase post decitabine treatment. We also found that decitabine treatment decreased the expression of an anti-apoptotic gene, BCL2. All the evidences support a role of decitabine in promoting apoptosis in combination with enzalutamide in enzalutamide resistant cells.

Cytidine analogs get incorporated into DNA molecules and form a covalent bond with all 3 DNA methyltransferases, promoting their proteasomal degradation (42). To test which component of the DNMTs plays the most pivotal role, we knocked down all three methyltransferases and tested the cells’ responses to enzalutamide. Our results indicated that DNMT3B knockdown in combination with enzalutamide treatment promoted an increase of apoptotic markers and significantly inhibited cell proliferation in enzalutamide-resistant cells (Figs. 4 and S3). These results indicate that DNMT3B plays a significant role in the resistance to enzalutamide in prostate cancer cells which further supports our findings in Fig. 1.

Enzalutamide targets the androgen signaling pathway by inhibiting the androgen receptor (4). AR aberrations are among well-known mechanisms by which cells acquire resistance to antiandrogen (43). We investigated the effects of inhibition of DNA methylation and enzalutamide treatment on AR signaling. In Figs. S3C and S3D, we showed that decitabine treatment induced a significant decrease of AR-V7 and AR in enzalutamide-resistant cells. Also, our DEG results showed that components of the mRNA splicing pathway were involved in the splicing of AR and AR-V7 in C4–2B-MDVR cells. These factors were shown to be involved in the alternative splicing of AR and generation of AR-V7 (44).

In our in vivo experiment, we used the primary enzalutamide-resistant cell line, 22Rv1, to generate xenograft tumors in castrated mice to assess the efficacy of decitabine plus enzalutamide. Results showed in Figs. 5 and S4 recapitulated our in vitro findings in MR49F and C4–2B-MDVR cells. The combination of enzalutamide and decitabine decreased tumor volume, tumor weight, AR-V7 expression, and KI-67 staining and increased cleaved-caspase 3 staining compared to vehicle-treated samples.

Our sequencing and bioinformatics data offered insights into the mechanisms of enzalutamide-resistance to and decitabine-mediated restoration of drug sensitivity. Our GSEA data indicated a deregulation of pathways that can be pursued in the future to identify individual genes involved in the resistance to enzalutamide (Fig. 6). We analyzed the upstream regulators using the list of DEGs we generated. In the analysis, we observed that AR and dihydrotestosterone were inhibited in C4–2B-MDVR compared to C4–2B cells. After decitabine treatment, AR inhibition was reversed. These results suggest that AR signaling is activated, providing a rationale for combination of decitabine and enzalutamide in enzalutamide-resistant cells.

Finally, we generated heatmaps to illustrate genes that changed during enzalutamide resistance and upon decitabine treatment (Figs. 6E and 6F). We identified 74 genes contributing to enzalutamide resistance which could be rescued by decitabine treatment (Fig. S6 and Table S2). These results suggest that the upregulation of the subsets in C4–2B-MDVR cells can be reversed through targeting the DNA methylation pathway. ABCB11, MYB, WNT10B, WT1, MCM7, BRCA1 and PAX1 are among the genes downregulated.

In conclusion, this study shows that enzalutamide promotes epigenetic changes in prostate cancer cells, leading to aberrations in gene expression and signaling pathways and finally the onset of enzalutamide resistance. DNMT3B, specifically DNMT3B3, plays a central role in this procedure (Fig. S7). Decitabine treatment restores the sensitivity to enzalutamide in resistant prostate cancer cells in vitro and in vivo. Components of the apoptosis, DNA repair and AR signaling pathways are influenced by changes of DNA methylation in enzalutamide-resistant prostate cancer cells.