DZNeP

Aristeromycin and DZNeP cause growth inhibition of prostate cancer via induction of mir-26a

Abstract

Most prostate cancers initially respond to androgen deprivation therapy, but then progress from androgen- dependent to androgen-independent prostate cancers. In the present study, a differential cytotoxicity screen of hormone-resistant prostate cancer LNCaP-hr cells and the parental LNCaP-FGC cells against normal MRC5 fibroblast cells, identified a small molecule compound, Aristeromycin (a derivative of 3-deazaneplanocin A (DZNeP)). The molecular target was shown to be S-adenosylhomocysteine hydrolase (AHCY), which catalyzes reversible hydrolysis of S-adenosylhomocysteine (SAH) to adenosine and L-homocysteine. DZNeP and Aristeromycin showed high inhibitory activity against AHCY. Treatment of the prostate cancer cells with DZNeP led to SAH accumulation and decreased levels of homocysteine and histone H3K27 methylation. SAH accumulation and cell growth inhibition were confirmed after siRNA-mediated AHCY knockdown. To further understand why AHCY inhibitors decreased prostate cancer cell growth, we performed microRNA expression profiling with LNCaP-hr cells. Mir-26a, which is involved in regulation of EZH2 expression, was upregulated in Aristeromycin-treated LNCaP-hr cells. A reporter assay established with the EZH2 3′-UTR confirmed that transfection of microRNA precursor molecules for miR-26a decreased the EZH2 3′-UTR luciferase activity. Meanwhile, an antisense microRNA inhibitor for miR-26a recovered the luciferase activity. The present findings suggest, at least in part, that miR-26a induced by an AHCY inhibitor can regulate oncogenic EZH2 expression, and could thus be an important mechanism of action for AHCY inhibitors in the treatment of prostate cancer.

1. Introduction

One of the major features of prostate cancer is its progression from androgen dependence to androgen independence. Previous investiga- tions have shown that androgen-independent prostate cancer cell lines are sensitive to androgens, but maintain their growth, proliferation, and prostate-specific antigen production under androgen deprivation conditions. So far, no effective therapy is currently available for androgen-independent prostate cancer. Therefore, more effective ther- apeutic approaches are needed.

S-adenosylhomocysteine hydrolase AHCY is an enzyme that cata- lyzes reversible hydrolysis of S-adenosylhomocysteine (SAH) to ade- nosine and L-homocysteine. SAH hydrolysis by AHCY is the only source of L-homocysteine in mammals (Chiang et al., 1996). Inhibition of AHCY results in accumulation of SAH, an inhibitor of S-adenosyl- methionine (SAM)-dependent methyltransferases. Upregulation of AHCY gene expression in colorectal carcinoma compared with unin- volved colon mucosa was reported in independent studies involving immunohistochemistry (Fan et al., 2011) or transcript profiling with cDNA microarrays (Birkenkamp-Demtroder et al., 2002; Cardoso et al., 2007; Notterman et al., 2001). Epigenetics are known to contribute to prostate cancer malignancy. These epigenetic alterations are reversible, rendering them attractive targets in prostate cancer therapy.

3-deazaneplanocin A (DZNeP) was reported to increase SAH levels and decrease SAM levels in human T-cell lymphoma H9 cells (Gordon et al., 2003). DZNeP also improved the survival of NOD/SCID mice with HL-60 leukemia compared with vehicle (Fiskus et al., 2009). Furthermore, DZNeP-treated DU145 cells formed tumors with a significantly slower growth rate than untreated cells (Crea et al., 2011). The downstream targets of AHCY among methyltransferases, including histone methyl- transferases and DNA methyltransferases, were clarified. Enhancer of zeste homolog 2 (EZH2), was shown to be a major downstream target of AHCY inhibition by DZNeP in cancer cells (Miranda et al., 2009). EZH2 plays an essential role in epigenetic regulation. Since EZH2 is over- expressed and acts as an oncogene in prostate cancer, and is associated with poor clinical outcomes of prostate cancer patients, it has been proposed as a target for prostate cancer therapy (Bracken et al., 2003; Rhodes et al., 2003; Varambally et al., 2002; Yu et al., 2007). EZH2 degradation (H3K27me inhibition) induced by DZNeP activates EZH2 target genes to induce apoptosis and tumor growth inhibition (Fiskus et al., 2009; Tan et al., 2007). These reports raise the possibility that targeting AHCY may result in the development of effective therapies for various cancers and diseases. However, the degradation mechanism of EZH2 protein in cancer cells remains unclear.

The objective of this study was to evaluate the mechanism on how AHCY inhibition result in EZH2 down regulation and develop a ther- apeutic target for androgen independent prostate cancer. Here we report the identification of a potential hit compound from a cytotoxic compound library, and further identify AHCY as a hit target for cancer therapy. In addition, to evaluate the mechanism for EZH2 protein depletion, and the relationship between microRNA expression and anticancer activity of this AHCY inhibitor, microRNA expression was profiled after cell treatment with Aristeromycin. MicroRNAs regulate EZH2 expression by modulating protein translation. Recently, the contribution of microRNAs to cancer development is increasingly being reported. Furthermore, microRNA expression changes could be clinical biomarkers for inhibitors against cancer therapeutic targets.

2. Material and methods

2.1. Cloning and protein purification of human full-length AHCY

The full-length coding sequence of human AHCY is identical to NCBI accession number NM_000687. This sequence was subcloned into the pENTR221 vector (Life Technologies, Carlsbad, CA). An N- terminally His-tagged AHCY-encoding recombinant baculovirus was generated using a BaculoDirect baculovirus expression system (Life Technologies) and baculovirus-infected insect cells. Briefly, Sf9 insect cells (Novagen) were cultured in SF900-II medium (Life Technologies) supplemented with 5% fetal bovine serum (FBS) and 50 µg/ml gentamicin (Life Technologies). When the Sf9 cell density reached 2 × 106 cells/ml, the cells were infected with baculovirus and cultured for 3 days. The cells were harvested by centrifugation (4000×g, 10 min, 4 °C) and the cell pellets were frozen at − 80 °C. The AHCY protein was isolated from the cell pellets by affinity chromatography on Ni-NTA superflow columns (Qiagen, Hilden, Germany) followed by gel- filtration chromatography on HiLoad 26/600 Superdex 200 pg columns (GE Healthcare, Buckinghamshire, England) equilibrated with 1× Tris-buffered saline (pH 7.4), 10% glycerol, and 150 mM NaCl.

2.2. Human AHCY enzyme assays using a thiol-reactive fluorescent probe

AHCY activity was measured by an in vitro assay using black 384- well plates (Cat. no. 784076; Greiner BioOne, Frickenhausen, Germany). The AHCY enzyme assay was based on the principle of fluorescence emission following incubation of a reactive dye with the products of an AHCY-catalyzed reaction, adenosine and homocysteine, produced by hydrolysis of SAH through the hydrolase activity of AHCY. The reactive dye, a thiol-reactive fluorescent probe, bound to homo- cysteine through its thiol-containing moiety.

The assay was performed in the 384-well plates (6 µl/well) with recombinant human AHCY enzyme. Briefly, the enzyme reaction was performed in reaction buffer consisting of 25 mM Tris-HCl (pH 7.5), 0.05 mM NAD, 1 mM EDTA, 0.01% Tween-20%, and 0.01% BSA. The final concentration of the SAH substrate was 50 μM or 1000 μM, and that of AHCY was 3 nM. Preincubation of compounds and enzyme was carried out for 60 min at room temperature. After 30 min of incubation at room temperature, the reaction was terminated by addition of 0.4% SDS solution. As the reagent for detection, a thiol-reactive fluorescent probe was added to the wells at a final concentration of 2 μM. The fluorescence intensity (excitation at 400 nm, emission at 465 nm) was measured using a SpectraMax fluorescence plate reader (Molecular Devices, Sunnyvale, CA). IC50 values were derived by fitting a sigmoidal dose-response curve to a plot of assay readout over inhibitor concentration. All fits were performed with GraphPad Prism 5.03 software (GraphPad Software, San Diego, CA).

2.3. Cell culture

The human prostate cancer cell line LNCaP-FGC and human fetal lung fibroblast cell line MRC-5 (American Type Culture Collection, Rockville, MD) were grown in RPMI 1640 medium supplemented with 10% FBS (Trace Scientific Ltd., Melbourne, Australia) at 37 °C in a humidified atmosphere containing 5% CO2.

LNCaP-hr cells, generated in vitro by culturing LNCaP-FGC cells in androgen-depleted medium, were a kind gift from Dr. Takahito Hara (Takeda Pharmaceutical Company Ltd., Shonan, Japan) (Hara et al., 2003). There is a possibility that chronic androgen deprivation induces the development of a hypersensitive androgen receptor (AR) that responds to extremely low concentrations of androgens. LNCaP-hr cells were maintained in phenol red-free RPMI 1640 medium with 10% dextran-coated charcoal (DCC)-FBS (Hyclone/Thermo Scientific). DCC-FBS was prepared by incubating 500 ml of FBS with 25 ml of 5% DCC solution (Sigma, St Louis, MO) and 0.05% T-70 dextran (Pharmacia, Uppsala, Sweden) in Dulbecco’s PBS (Wako, Osaka, Japan) at 45 °C for 30 min, followed by collection of the supernatant by centrifugation at 1800×g for 30 min. The cell line was maintained at 37 °C in a humidified atmosphere containing 5% CO2. The LNCaP-hr cells clearly grew more slowly in the absence of androgen than in the presence of androgen.

2.4. Cell growth and viability of LNCaP-FGC, LNCaP-hr, and MRC5 cells

Cell viability was measured by ATP production in viable cells using the CellTiter-Glo® Luminescent Cell Viability Assay (Cat. no. G7573; Promega, Madison, WI). Briefly, LNCaP-FGC cells (5 × 103 cells/well) and LNCaP-hr and MRC5 cells (1.5 × 104 cells/well) were seeded in white 96-well plates (Cat. no. 3917; Corning, MA), and then drug treatments were initiated at 1 or 10 μM. The drugs were dissolved in DMSO and added to the medium to give a final DMSO concentration of 0.3%. The cells were exposed to the compounds for 72 h, and relative cell viability was assessed. After 15 min of reagent/sample incubation, the lumines- cence was measured by an EnVision Plate Reader (PerkinElmer, Boston, MA). Data analysis was carried out with GraphPad Prism 5.03 software using nonlinear regression for a sigmoidal dose-response curve with a variable slope. The IC50 value was defined as the concentration that inhibited the tumor cell growth by 50%.

2.5. Homocysteine assay and SAH ELISA in LNCaP-FGC cells

Cells were seeded into 6-well plates (Cat. no. 354413; Becton Dickinson, Franklin Lakes, NJ) at 1.5 × 106 cells/well and treated with various concentrations of compounds for 48 h in a humidified 5% CO2 incubator. Homocysteine was measured using high-performance liquid chro- matography (HPLC) by SRL Inc. (Tokyo, Japan). For the SAH ELISA, all reagents were supplied in an Axis Homocysteine EIA Kit (Cat. No. AX51301; Axis Shield, Dundee, Scotland). Cell supernatants (25 µl/ well) were placed in SAH-coated 96-well plates. Next, 200 µl of monoclonal mouse anti-SAH-antibody was added to each well and incubated at room temperature for 30 min. After three washes with washing buffer (phosphate buffer containing 0.01% merthiolate, 0.01% Tween-20%, and 0.01% BSA), 100 µl of horseradish peroxidase-con- jugated anti-mouse-antibody was added to each well and incubated at room temperature for 20 min. After three further washes with washing buffer, 100 µl of tetramethylbenzidine and 0.8 M sulfuric acid were added to each well and incubated for 10 min. The absorbances of the wells were measured at 450 nm by the EnVision Plate Reader. Serial dilutions of an SAH standard ranging from 0 to 2 μM were evaluated using the same competitive assay format.

The inhibition rate was calculated on the basis of 0% control wells containing DMSO-treated cell supernatants and 100% control wells containing culture medium. Data analyses was carried out with GraphPad Prism 5.03 software using nonlinear regression for a sigmoidal dose-response curve with a variable slope.

2.6. Immunoblot analysis

LNCaP-FGC and LNCaP-hr cells were plated in 96-well plates (Cat. no. 3585; Costar) at 6 × 103 cells/well (100 µl). After 24 h, the cells were treated with 5 μM DZNeP and incubated for 72 h at a CO2 incubator. LNCaP-FGC cells were treated with 1 μM of DZNeP and with or without 100 nM Anti-miR26 a Mix (hsa-miR-26a-5p, hsa-miR26a-2-3) for 96 h. The cells were lysed in 50 µl of lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na3VO4, 1 µg/ml leu- peptin (Cat. no. 9803; Cell Signaling Technology, Danvers, MA)) contain- ing protease inhibitors (Cat. No. P5726; Sigma-Aldrich, St Louis, MO). The proteins were subjected to 5–20% SDS-PAGE, and electrophoretically transferred to PVDF membranes using an iBlot Gel Transfer System (Life Technologies). After incubation for 1 h with a blocking buffer (Cat. no. UK-B80; DS Pharma Biomedical Co. Ltd., Osaka, Japan) at room temperature, the membranes were incubated with primary antibodies against human EZH2, embryonic ectoderm development protein (EED), suppressor of zeste 12 homolog (SUZ12) and anH3K27me3, followed by horseradish peroxidase-conjugated anti-rat, anti-rabbit, or anti-mouse IgG secondary antibodies (Cell Signaling Technology). Anti-beta Actin HRP conjugate (Wako, Tokyo, Japan) was used for Beta Actin proteins. Antibody-bound proteins were detected using an enhanced chemilumi- nescent detection system (GE Healthcare), and scanned using an ImageQuant LAS-4000 Image Analyzer (GE Healthcare).

2.7. AHCY siRNA assay in LNCaP-FGC cells

LNCaP-FGC cells were plated in 96-well plates (Cat. no. 3585; Costar) or BD BioCoat™ Poly-D-Lysine Cellware 96-well plates (Cat. no. 356620; BD Biosciences) at 6 × 103 cells/well (100 µl). After 24 h, the cells were transfected with siRNAs using RNAi MAX Reagent (Life Technologies) according to the manufacturer’s instructions. A non- targeted siRNA (AllStars Negative Control siRNA; Cat. no. 102728; Qiagen) and a siRNA targeting AHCY were purchased from Applied Biosystems (Foster City, CA). At 72 h after transfection, cell viability was assessed by ATP production in viable cells using the CellTiter-Glo® Luminescent Cell Viability Assay.

For the SAH detection assay in LNCaP-FGC cells, all reagents were supplied in an EPIgeneous™ Methyltransferase Assay Kit (Cat. no. 62SAHPEB; CisBio, Codolet, France). Cells were inoculated in 96-well plates at 3000 cells/well and incubated at 37 °C for 72 h in a humidified 5% CO2 incubator. Cell supernatants (5 µl/well) were placed in 384-well plates (Cat. no. 784075; Greiner BioOne). Next,
1.5 µl of detection buffer was added to the plates and incubated for 10 min at room temperature with gentle shaking. SAH-red or anti-SAH Eu cryptate was diluted 8-fold or 50-fold in detection buffer, respec- tively, and 3 µl of each solution was added to the cell supernatants.
The plates were incubated for 60 min at room temperature, and measured for their absorbance by the EnVision Plate Reader (excitation at 320 nm and emission donor at 615 nm or emission acceptor at 665 nm). The inhibition rate was calculated on the basis of 0% control wells containing DMSO-treated cell supernatants and 100% control wells containing culture medium.

2.8. Quantitative RT-PCR measurement of AHCY and GAPDH mRNA

The expression levels of AHCY and reference gene GAPDH were evaluated by real-time PCR (RT-PCR). The RT-PCR primers for AHCY and GAPDH were purchased from Applied Biosystems (Waltham, MA).

Fig. 1. (A) Flowchart showing an overview of strategy. Each part of the screen is depicted with the number of positive hit compounds taken to the next stage of the screen. (B) Chemical structures of Aristeromycin and DZNeP.

Cells were inoculated in 96-well plates at 3000 cells/well and incubated at 37 °C for 72 h in a humidified 5% CO2 incubator. Preparation of cell lysates was conducted using a Cell Amp™ Direct RNA Prep Kit (Cat. No. 3732; Takara-Bio, Shiga, Japan) according to the manufacturer’s instructions. Briefly, after cells were washed with washing buffer, 50 µl of processing solution was added and the plates were mixed for 5 min at room temperature. The reaction mixtures were incubated in 384- well V-bottom plates for 5 min at 75 °C and then held at − 80 °C.
Reverse transcription reactions were conducted using a TaqMan RNA-to-CT 1-Step Kit (Cat. no. P/N4392938; Applied Biosystems) according to the manufacturer’s instructions. A reaction mixture consisting of 5 µl of TaqMan Mix (2×), 0.25 µl of RT Enzyme Mix (40×), 0.5 µl of primers mix (20×), 2.25 µl of double distilled water, and 2 µl of lysate was prepared in MicroAmp Optical 384-well reaction plates (Applied Biosystems). The final reaction volume was adjusted to 10 µl with nuclease-free water. RT-PCR was performed using a 7900HT Fast Real-Time PCR system and recorded by 7900HT SDS 2.3 software (Applied Biosystems). For quantitative RT-PCR, samples were subjected to reverse transcription at 42 °C for 5 min. The cycle parameters for the PCR were 95 °C for 10 s (enzyme activation), followed by 40 cycles of denaturation at 95 °C for 5 s and annealing/ extension at 60 °C for 30 s, and a final hold at 4 °C. All reactions were run in four replicates. The expression of the AHCY gene relative to the GAPDH gene was determined using the ΔΔCt method. The threshold cycle (Ct) was defined as the fractional cycle number at which the fluorescence passed the fixed threshold.

Fig. 2. Growth inhibitory activity by Aristeromycin in LNCaP-FGC, LNCaP-hr, and MRC5 cells. (A–C) In vitro growth inhibitory assays were performed in LNCaP-FGC (A), LNCaP-hr (B), and MRC5 (C) cells. The cells were exposed to Aristeromycin for 72 h, and the relative cell viability was assessed by ATP production in viable cells using the CellTiter-Glo® Luminescent Cell Viability Assay (n = 6). LNCaP-FCG cell proliferation was shown in the presence of Aristeromycin up to 168 h.

Fig. 3. Inhibition of AHCY enzyme activity by Aristeromycin and DZNeP. (A) AHCY enzyme assays with a thiol-reactive fluorescent probe were established using human recombinant full-length AHCY. The time courses of AHCY activity in the presence of 0–10 nM AHCY and 10 µM SAH are shown. (n = 2). (B) Determination of kinetics constants for SAH. The experiments were performed under assay conditions of 3 nM AHCY and increasing concentrations of SAH for 60 min of incubation. The Km value for SAH was 48 μM (n = 2). (C, D) Dose-dependent inhibition of 3 nM AHCY by Aristeromycin (C) or DZNeP (Din the presence of SAH (50 or 1000 μM)) (n = 2).

2.9. MicroRNA profiling

LNCaP-hr cells (3 × 106 cells/flask) were treated with 5 μM Aristeromycin for 48 h. Total RNA isolation and small RNA enrichment were performed with a mirVana miRNA Isolation Kit (Cat. no. 1560; Ambion, Austin, TX) according to the manufacturer’s instructions. One million cells were used for purification of total RNA. The purified total RNA was eluted in 100 µl of elution buffer, and immediately chilled on ice before reverse transcription reactions.

Fig. 4. SAH and homocysteine levels in the culture medium of LNCaP-FGC cells after AHCY inhibitor treatment. (A) LNCaP-FGC cells were treated with 0.01–1 μM Aristeromycin (left) and DZNeP (right) for 48 h. The culture medium was collected and the SAH levels were measured by EIA using an anti-SAH antibody. Data are presented as mean ± S.D. (n = 4). (B) Homocysteine levels in the culture medium of LNCaP-FGC cells measured by HPLC after 48 h of treatment with DZNeP. DZNeP dose- dependently inhibited homocysteine from cells with an IC50 of 64.4 nM (n = 2).

A TaqMan MicroRNA Reverse Transcription Kit (Cat. no. 4366596; Applied Biosystems) and Megaplex™ RT Primers Human Pool A and B (Cat. no. 4399966, 4399968; Applied Biosystems) were used to synthesize complementary DNA for the TaqMan MicroRNA Array (TaqMan® Human MicroRNA A + B Array v2.0; Cat. no. 4398965, 4398966; Applied Biosystems). The final reverse transcription reaction consisted of 3 µl (450 ng) of total RNA and 4.5 µl of RT reaction mix. cDNA was synthesized as described in the manual for the Megaplex™ Primers without pre-amplification. The samples were subjected to reverse transcription with the following protocol: 40 cycles of 2 min at 16 °C, 1 min at 42 °C, 1 s at 50 °C, and 5 min at 85 °C.

The TaqMan Human MicroRNA A and B Array v2.0 has a 384-well format, and contains 377 + 290 human microRNAs. The array was loaded with a mixture comprising 450 µl of TaqMan Universal PCR Master Mix, No AmpErase UNG (Cat. No. 4324018; Applied Biosystems), 6 µl of Megaplex RT product, and 444 µl of H2O. The TaqMan Human MicroRNA Array was performed using the 7900HT Fast Real-Time PCR system and recorded by the 7900HT SDS 2.3 software. The cycle parameters for the PCR were 95 °C for 10 min (enzyme activation), followed by 40 cycles of denaturation at 95 °C for 15 s and annealing/extension at 60 °C for 60 s, and a final hold at 4 °C. All reactions were run in four replicates.
The TaqMan Human MicroRNA Array experiments of LNCaP-hr cells were normalized by the median Ct of all microRNAs assayed. The mean fold differences were calculated by normalizing the relative expression (ΔΔCt values) in LNCaP-hr cells treated with Aristeromycin to that in untreated LNCaP-hr cells. Mean fold differences below 0.5 or above 2.0 were considered to represent a significant difference between the treated and untreated cells.

Fig. 5. Effects of DZNeP on EZH2 and H3K27 trimethylation in LNCaP cells. (A) LNCaP-FGC cells and LNCaP-hr cells were treated with 5 µM DZNeP for 72 h. (B) LNCaP-FGC cells were treated with 1 μM of DZNeP and with or without 100 nM Anti- miR26 a Mix (hsa-miR-26a-5p, hsa-miR26a-2-3) for 96 h. The cells were then harvested, and measured for their EZH2, EED, SUZ12, H3K27me3 and Beta Actin levels by immunoblot analysis.

2.10. Quantitative RT-PCR measurement of microRNAs

LNCaP-FGC cells were seeded in 96-well plates at 1.5 × 104 cells/well and incubated for 24 h. Some cells were treated with 0.1–10 μM DZNeP. After 48 h, the expression levels of miR-101a, miR-26a-2*, and reference small RNA molecule RNU6B were evaluated by real-time PCR. The real- time PCR primers for these miRNAs were purchased from Applied Biosystems. Reverse transcription with the miRNA-specific primers was performed with a TaqMan MicroRNA Reverse Transcription Kit (Cat. no. 4366596; Applied Biosystems), followed by a real-time PCR protocol using miRNA-specific TaqMan primers according to the manufacturer’s protocol. Briefly, the reverse transcription reactions contained RNA samples including purified total RNA, 50 nM stem–loop RT primer, 1× RT buffer, 0.25 mM each dNTP, 3.33 U/µl of MultiScribe reverse transcriptase, and 0.25 U/µl RNase inhibitor. The 15-µl reactions (750 ng total RNA) were incubated in 384-well plates for 30 min at 16 °C, 30 min at 42 °C, 5 min at 85 °C, and then held at 4 °C.

Real-time PCR was performed using the standard TaqMan PCR Kit protocol. Each reaction consisted of 17.67 µl of TaqMan Universal PCR Master Mix, No AmpErase UNG, 1 µl of 20× individual TaqMan MicroRNA Assay Mix, and 1.33 µl of cDNA. The cycle parameters for the PCR were the same as those used for the TaqMan Human MicroRNA Array. The microRNA TaqMan assay experiments demon- strated that RNU6B was highly expressed in various cell lines. The expression of each microRNA relative to RNU6B was determined using the ΔΔCt method for the microRNA of interest. The threshold cycle (Ct) was defined as the fractional cycle number at which the fluorescence passed the fixed threshold.

2.11. Luciferase reporter assay

LNCaP-FGC cells were seeded in 96-well plates at 1.5 × 104 cells/well and incubated for 24 h. The cells were then transfected with an EZH2 3′- UTR luciferase plasmid (Switch Gear Genomics, Menlo Park, CA) using the FuGENE HD transfection reagent. After 24 h, pre-miR-101a, pre- miR-26a-2* or anti-miR-26a mix (26a, 26a-1*, 26a-2*) (Ambion, Austin, TX, USA) were transfected using the siPORT NeoFX transfection reagent (Thermo Fisher Scientific, Waltham, MA). Some cells were treated with DZNeP. At 48 h after transfection, luciferase activity was assayed using a Steady-Glo Luciferase Assay System (Promega). Relative cell viability was assessed by ATP production in viable cells using the CellTiter-Glo® Luminescent Cell Viability Assay. After 15 min of reagent/sample incuba- tion, the luminescence was measured by an EnVision Plate Reader (PerkinElmer, Boston, MA). Data analysis was carried out with GraphPad Prism 5.03 software.

3. Results

3.1. High-throughput screening of a cytotoxic compound library

After confirming the reproducibility of the cell viability assay, we screened a cytotoxic compound library for small molecule inhibitors using the optimized assay conditions. Each test compound was assayed at a concentration of 1 or 10 μM. A summary of our screening strategy is shown in Fig. 1A. A diverse library of approximately 8159 cytotoxic small compounds at Takeda Pharmaceutical Company was screened in the work flow. Of the 8159 compounds, 58 hit compounds caused more than 50% inhibition at 1 μM in LNCaP-hr cells, and had more potent inhibitory activity in MRC5 cells. The initial primary screen identified growth inhibiting compounds with a hit rate of 0.7% (58 compounds/ 8159 compounds). The 58 hit candidate compounds were subjected to IC50 determination assays.

Fig. 6. Effects of AHCY-specific knockdown by siRNA transfection. The cells were exposed to the AHCY siRNA for 72 h. (A)The relative expression levels of AHCY were evaluated by real-time PCR in LNCaP-FGC cells after AHCY silencing (n = 3). (B) SAH levels in LNCaP-FGC culture medium were measured by EPIgeneous™ Methyltransferase Assay Kit using an anti-SAH antibody (n = 6). (C) In vitro growth inhibitory assays were performed in LNCaP-FGC cells. and the relative cell viability was assessed by ATP production in viable cells using the CellTiter-Glo® Luminescent Cell Viability Assay (n = 6).

Fig. 7. MicroRNA expression patterns in LNCaP-hr cells after treatment with Aristeromycin. The fold changes of microRNAs in LNCaP-hr cells after 48 h of treatment with 5 μM Aristeromycin are presented. The ratios for Aristeromycin-treated LNCaP-hr cells/untreated LNCaP-hr cells were obtained in four independent experiments (n = 3).

Fig. 8. Expression of miR-101a and miR-26a-2* in LNCaP cells. LNCaP-FGC cells were seeded in 96-well plates at 1.5 × 104 cells/well and incubated for 24 h. Some cells were treated with 0.1–10 μM DZNeP. After 48 h, the relative expression levels of miR-101a and miR-26-2* were evaluated by real-time PCR (n = 3).

Of the 58 compounds, 19 compounds were known anticancer targets showing sigmoidal dose-response inhibition in LNCaP-hr and LNCaP-FGC cell growth. Prioritization of the hit compounds was performed after the screening and identified Aristeromycin, the structure of which is shown in Fig. 1B. After a search for structu- rally-related compounds, Aristeromycin was known as an AHCY inhibitor (Jackson et al., 2012), and found to show similarity structure with DZNeP. Aristeromycin had IC50 values (95% confidence interval) of 3.2 μM (2.9–3.6 μM) for LNCaP-FGC cell growth and 0.88 μM (0.74–1.0 μM) for LNCaP-hr cell growth (Fig. 2A, B). DZNeP showed similar IC50 values against these cells, at 0.50 μM (0.35–0.73 μM) for LNCaP-FGC cell growth and 26.4 nM (4.44–15.6 nM) for LNCaP-hr cell growth. The cytotoxicity of Aristromycin was in a concentration and tim-dependent fashion. These two compounds did not show any inhibitory activity against MRC5 cell growth. LNCaP-FCG cell prolif- eration was shown in the presence of Aristeromycin up to 168 h (Fig. 2D).

3.2. Assay development and enzymological characterization of an AHCY inhibitor

Inhibitory activity of AHCY inhibitors was assessed by fluorescence assays using a thiol-reactive fluorescent probe. To determine the incubation time needed to work within the linear range of the enzymatic reaction, the time course was measured with a reaction mixture containing 0–10 nM AHCY and 10 μM SAH (Fig. 3A). The enzyme reaction was monitored for up to 60 min. As a result, the working concentration for enzymes was set at 3 nM and the reaction time was fixed at 30 min for future assays to retain a sufficient signal- to-noise ratio of more than 34.

For enzymological characterization of AHCY, the kinetics constant was assessed by varying the concentration of SAH. The Km for SAH was found to be 48 µM (Fig. 3B). Thus, the working concentration of substrates was set at 50 µM SAH to enhance the sensitivity. The inhibitory effect of Aristeromycin, DZNeP, was assessed to demonstrate the assay sensitivity of the system. As shown in Fig. 3C, without preincubation, the IC50 value (95% confidence interval) of Aristeromycin against AHCY was determined to be 38.5 nM (37.8–39.3 nM) at 50 μM SAH (approximately equal to the Km), but 271 nM (212–248 nM) at 1000 µM SAH (20× Km) (Fig. 3D). DZNeP showed similar IC50 values against AHCY enzyme activity, being 103 nM (90– 118 nM) at 50 µM SAH and 756 nM (647–883 nM) at 1000 µM SAH (20× Km). Aristeromycin and DZNeP showed an approximately 7-fold or 7.3-fold shift in the IC50 upon raising the SAH concentration from 50 µM to 1000 µM. A simple competitive inhibition pattern of Aristeromycin or DZNeP was confirmed at the two different concentrations of SAH. With 60 min of preincubation, the mean IC50 values of Aristeromycin and DZNeP at 50 μM SAH were 12.7 nM (9.3–17.3 nM) and 7.9 nM (4.9–12.8 nM), respectively. These results indicate that Aristeromycin and DZNeP are two SAH-competitive inhibitors with slow-binding kinetics to AHCY.

3.3. Characterization of Aristeromycin and DZNeP

To determine whether AHCY enzyme inhibition by compounds also occurs in cells, the potency of Aristeromycin or DZNeP was investi- gated in cellular assays. To elucidate the mechanism of inhibition, Aristeromycin and DZNeP were examined SAH concentrations. The SAH level was increased by approximately 70.3-fold or 61.4-fold for 1 µM Aristeromycin or DZNeP at 48 h after treatment compared with the level in untreated cells (Fig. 4A). DZNeP caused a dose-dependent decrease in homocysteine by HPLC and consequently a dose-depen- dent increase in SAH (Fig. 4B). Aristeromycin also showed dose- dependent homocystein decrease by MBL enzyme cycling kit (Medical & Biological Laboratories Co.,Ltd, Tokyo, Japan).

Reduced cell viability caused by DZNeP suggested that these compounds could have a tumor-suppressive role through an increase in SAH or decrease in homocysteine in cancer cells. The DZNeP- mediated decreases in expression of Polycomb repressive complex 2 (PRC2) (EZH2, SUZ12, and EED) were also evaluated in the prostate cancer cells. As expected, DZNeP depleted PRC2 (EZH2, SUZ12, and EED) in LNCaP-FGC or LNCaP-hr cells, accompanied by decreased levels of H3K27me3 (Fig. 5A). These down regulation of EZH2 by DZNeP was recovered with transfection of an anti-miR-26a mixture (26a, 26a-1*, 26a-2*) (Fig. 5B).

3.4. AHCY siRNA knockdown

To evaluate AHCY-mediated SAH levels in LNCaP-FGC cells, the cells were depleted of AHCY by siRNA transfection for 3 days and then measured for cell viability, AHCY mRNA, and SAH by the HTRF method. The results confirmed that AHCY mRNA expression decreased in a dose-dependent manner (Fig. 6A). The siRNA-mediated knock- down of AHCY effectively increased SAH expression in LNCaP-FGC cells (Fig. 6B), and caused dose-dependent growth inhibition of the cells (Fig. 6C).

3.5. Differential expression of microRNAs in LNCaP-hr cells after Aristeromycin treatment

The LNCaP-hr cell line was used to compare the difference associated with the anticancer mechanism after Aristeromycin treatment. To investigate the role of microRNAs in Aristeromycin-treated cells, we first searched for microRNAs that showed differential expres- sion using the TaqMan MicroRNA Array.As shown in Fig. 7, the expressions of some microRNAs were changed in LNCaP-hr cells after 48 h of treatment with 5 μM Aristeromycin. About 40% of the microRNA probes gave detectable signals in at least one of the samples analyzed in triplicate. Using a 3.5-fold difference as the cut-off, 69 differentially expressed microRNAs were identified, comprising 32 that were upregulated and 37 that were downregulated in LNCaP-hr cells compared with untreated cells. Among the microRNAs with lower transcript levels in Aristeromycin-treated LNCaP-hr cells, MiR-16-1* was the most highly downregulated. Meanwhile, miR-512-3p was the most significantly upregulated. The list of microRNAs showing increased expression also contained members of the miR-26a cluster.

Fig. 9. Analysis of miR-101a and miR-26a effects on the EZH2 3′-UTR in reporter assays. LNCaP-FGC cells were seeded in 96-well plates at 1.5 × 104 cells/well and incubated for 24 h. The cells were then transfected with the EZH2 3′-UTR luciferase plasmid using the FuGENE HD transfection reagent. After 24 h, pre-miR-101a and pre-miR-26a-2* (0, 30, 60, 120 nM) or anti-miR-26mix (26a, 26a-1*, 26a-2*) at 60 nM were transfected using the siPORT NeoFX transfection reagent. Some cells were treated with 0.1–10 μM DZNeP. At 48 h after transfection, luciferase activity was assayed using the Steady-Glo Luciferase Assay System. **P < 0.01, **P < 0.05, ***P < 0.001, versus negative control cells by Dunnett's test (n = 3). To verify the accuracy of the TaqMan MicroRNA Array results, we measured the expression of individual microRNAs using a real-time quantitative TaqMan RT-PCR method designed to detect mature microRNA sequences. As shown in Fig. 8, the results revealed an increase in miR-26a expression, while miR-101a had slight activity at 10 µM. 3.6. Effects of pre-miR-101a and miR-26a-2* on EZH2 expression To evaluate whether EZH2 is a direct target of miR-26a, we performed an EZH2 3′-UTR reporter assay involving luciferase activity. There were two predicted binding sites for the seed sequences of miR- 101a and one predicted miR-26a-2* target site in the 3′-UTR of the EZH2 mRNA (Cao et al., 2010). When increasing amounts (30, 60, 120 nM) of pre-miR-101a and miR-26a-2* were transfected into LNCaP-FGC cells, the reporter construct with the EZH2 3′-UTR showed luciferase activity (Fig. 9A, left). DZNeP inhibited the EZH2 3′-UTR luciferase activity in LNCaP-FGC cells (Fig. 9A, left). Meanwhile, the reporter construct without the EZH2 3′-UTR comple- tely lost the response to transfected pre-miR-101a and miR-26a-2* or DZNeP treatment (Fig. 9A, right). Cell viability was not influenced by treatment of pre-miRs, anti-miRs and reporter construct. Cotransfection of an anti-miR-26a mixture (26a, 26a-1*, 26a-2*) and EZH2 3′-UTR abolished the inhibition of luciferase activity (Fig. 9B). These findings suggest that miR-101a and miR-26a-2* individually or miR-26a induced by DZNeP can partially repress EZH2 expression. 4. Discussion A differential cytotoxicity screen for hormone-resistant prostate cancer LNCaP-hr cells and the parental LNCaP-FGC cells against normal MRC5 fibroblast cells identified Aristeromycin, a derivative of DZNeP. This compound did not show growth inhibition of MRC5 cells. Treatment of cancer cells with DZNeP caused accumulation of SAH and decreases in homocysteine and histone H3K27 methylation. As AHCY gene overexpression has been observed in several cancers, AHCY is considered a therapeutic target in cancer treatment. EZH2 was reported to be a major downstream target of AHCY inhibition by DZNeP in cancer cells (Miranda et al., 2009). EZH2 is degraded by the proteasome in acute myeloid leukemia resistant cells, and proteasome inhibitors restore EZH2 protein levels (Göllner et al., 2017). EZH2 degradation (H3K27me inhibition) causes EZH2 target genes to induce apoptosis and tumor growth inhibition (Fiskus et al., 2009; Tan et al., 2007). In our growth inhibition assay, DZNeP showed no sign of toxicity against normal fibroblast cells, and thus some EZH2 target genes may be differentially expressed between cancer cells and normal cells. PRC2 includes at least three core components, EED, SUZ12, and EZH2. Increased expression of PRC2 core subunits EED and EZH2 was reported in breast cancer lymph node metastatic tumor cells in a microarray network analysis (Yu et al., 2012). Inhibition of PRC2 activity was shown to modulate tumor suppressor gene expression. DZNeP treatment decreased global H3K27 methylation levels, degraded PRC2 components, and upregulated tumor suppressors including FOXO32, TGFB1, IGFBP3 (Tan et al., 2007), FHIT, HIC1, p21, and RASSF1A (Kemp et al., 2012). While overexpression of EZH2 coincided with decreased levels of miR-101 and miR-26a (Kemp et al., 2012; Vishnubalaji et al., 2015), the detailed mechanisms for EZH2 downregulation (Fiskus et al., 2009) and growth inhibition by DZNeP treatment remained unknown. Therefore, to eluci- date one of the mechanisms responsible for the development of microRNA expression changes during growth of prostate cancer cells, we evaluated the microRNA expression profiles of androgen-independent LNCaP-hr cells using TaqMan MicroRNA Microarray analyses.
LNCaP-hr cells treated with Aristeromycin exhibited alterations in microRNA profiles that differed from those in untreated cells.

Specifically, we found that the expression levels of miR-135b, miR-449b, miR-433, and miR-512-3p were upregulated by 3.2, 3.0, 8.8, and 50-fold, respec- tively, in Aristeromycin-treated cells compared with untreated cells. A recent report suggested that these microRNAs play a role in the control by regulating AR gene expression levels. Among these miRNAs, miR-135b and miR-449b were characterized as andro-miRs based on their ability to suppress AR expression by directly binding to target sequences in the AR 3′-UTR (Östling et al., 2011). AR signaling plays a central role in prostate cancer cell growth and survival (Heinlein and Chang, 2004). AR expres- sion is maintained throughout prostate cancer progression, and the majority of androgen-independent or hormone-refractory prostate cancers express AR. Downregulation of AR expression through these mechanisms in addition to androgen ablation, such as modulation of signal transduc- tion pathways, may delay prostate cancer progression.

The present study also showed that miR-26a, which is involved in regulation of EZH2 expression, was upregulated in Aristeromycin- treated LNCaP-hr cells. A reporter assay established with the EZH2 3′-UTR confirmed that transfection of microRNA precursor molecules for miR-26a decreased luciferase activity of the EZH2 3′-UTR. Meanwhile, an antisense microRNA inhibitor for miR-26a recovered the luciferase activity. Our findings at least partly suggest that miR-26a induced by an AHCY inhibitor can regulate oncogenic EZH2 expression, and could thus be one of an important mechanism of action of AHCY inhibitors in the treatment of cancer. In addition, since miR-26a was reported to be downregulated during management of colorectal cancer (Vishnubalaji et al., 2015), this miR-26a expression change could be a candidate clinical biomarker for AHCY inhibitors.

In conclusion, our study extensively screened a cancer-specific target, AHCY, and evaluated the antiproliferative effects of AHCY inhibitors in prostate cancer cells based on the microRNA expression profiles. Our findings provide significant contributions toward understanding of the anticancer properties of AHCY and toward the future development of therapeutic agents against cancer.