PPAR δ agonist GW0742 interacts weakly with multiple nuclear receptors including the vitamin D receptor

Abstract

A high throughput screening campaign was conducted to identify small molecules with the ability to inhibit the interaction between the vitamin D receptor (VDR) and steroid receptor coactivator 2. These inhibitors represent novel molecular probes to modulate gene regulation mediated by VDR. The peroxisome proliferator-activated receptor δ (PPARδ) agonist GW0742 was among the identified VDR-coactivator inhibitors and has been characterized herein as a pan nuclear receptor antagonist at concentrations higher than 12.1 µM. The highest antagonist activity for GW0742 was found for VDR and the androgen receptor (AR). Surprisingly, GW0742 behaved as PPAR agonist/antagonist activating transcription at lower concentration and inhibiting this effect at higher concentrations. A unique spectroscopic property of GW0742 was identified as well. In the presence of rhodamine-derived molecules, GW0742+ increased fluorescence intensity and fluorescence polarization at an excitation wavelength of 595 nm and emission wavelength of 615 nm in a dose dependent manner. The GW0742-inhibited NR-coactivator binding resulted in a reduced expression of five different NR target genes in LNCaP cells in the presence of agonist. Especially VDR target genes CYP24A1, IGFBP-3 and TRPV6 were negatively regulated by GW0742. GW0742 is the first VDR ligand inhibitor lacking the secosteroid structure of VDR ligand antagonists. Nevertheless, the VDR-meditated downstream process of cell differentiation was antagonized by GW0742 in HL-60 cells that were pretreated with the endogenous VDR agonist 1,25-dihydroxyvitamin D₃.

Nuclear receptor (NR) ligands represent one of the largest class of currently FDA approved drugs for a wide variety of human diseases.¹ The 48 identified members of this superfamily of transcription factors share a similar protein structure and bind, in case of the steroid hormone receptors, to endogenous ligands with a similar steroid structure.² During the past decade, many new synthetic NR ligands with diverse scaffolds have been introduced, especially for the peroxisome proliferation activated receptors (PPARα (3363 ligands), PPARγ (3891 ligands), and PPARδ (1729 ligands).³ Although the structural optimization of NR ligand has been guided by affinity to a given NR, it is generally accepted that ligand selectivity among NRs has to be optimized as well in order to reduce side effects. The discovery that NR–coactivator binding is required for NR-mediated transcription led to a more stringent concept of selectivity for NR ligands.^{4, 5} NR ligand drug candidates should not only bind selectively to a certain NR but also selectively modulate recruitment of particular nuclear receptor coactivators.⁶ Many NR ligands with improved selectivity have been recently introduced but the sheer number of NRs and affiliated coactivators that govern their specific gene regulation pattern has complicated the selectivity assessment of these ligands.⁷

A new concept to selectively modulate NR function involves the development of small molecules that directly inhibit the interaction between NRs and coregulators instead of interacting with the receptor ligand binding pocket to allosterically modulate interactions of receptor with its target proteins.⁸ The development of an inhibitor selective towards a particular coregulator-NR interaction could be highly useful for the vitamin D receptor (VDR).⁹ VDR binds to its endogenous ligand, 1,25-dihydroxyvitamin D₃ (1,25-(OH)₂D₃), with high affinity,¹⁰ mediating changes in transcription of genes responsible for cell differentiation, proliferation, and calcium homeostasis.¹¹ Based on its biological function, the VDR has been identified as an important pharmaceutical target for the treatment of metabolic disorders, skin diseases, cancer, autoimmune diseases, and cardiovascular diseases.¹² VDR, similar to other NRs, contains several functional domains, including a ligand-binding domain (VDR-LBD), which mediates ligand-dependent gene regulation.¹³ VDR binds DNA as a heterodimer with the retinoid X receptor (RXR).¹⁴ In the unliganded state, VDR is associated with corepressor proteins, which repress transcription of VDR target genes.¹⁵ In the presence of 1,25-(OH)₂D₃, the VDR-LBD undergoes a conformational change which prevents corepressor binding and permits interactions with coactivators, resulting in the formation of a multi-protein complex responsible for VDR-mediated transcription.¹⁶

VDR agonists have been developed to treat metabolic bone diseases and proliferative skin disorders.¹² In contrast to the large number of reported VDR agonists, only a limited number of VDR ligand antagonists have been described with the ability to allosterically inhibit the interactions between VDR and its coactivators.^17–24 All of these antagonists are based on the secosteroid structure of 1,25-(OH)₂D₃. Mita et al. introduced the first reversible small molecule direct inhibitors of VDR–coactivator interactions with moderate NR selectivity.²⁵ The reported compounds inhibited both VDR-mediated and estrogen receptor β-mediated transcription. Recently, we reported the first irreversible direct inhibitors of the VDR–coactivator interaction.²⁶ This new class of compounds selectively inhibited the interaction between VDR and coactivator SRC2²⁷ among six other NR–coactivator interactions tested and reduced the expression of VDR target gene TRPV6 in DU-145 prostate cancer cells in the presence of 1,25-(OH)₂D₃.

Herein, we describe the evaluation of compound GW0742, which was identified during a high throughput screen (HTS) to identify small molecules that inhibit the interaction between VDR and nuclear receptor coactivators 2. GW0742 was introduced by GlaxoSmithKline in 2003 as a highly selective agonist for the peroxisome proliferator activated receptor δ (PPARδ).²⁸ Since then, GW0742 has been investigated in cell-based assays and in vivo to understand the role of PPARδ in hypertension,^{29, 30} diabetes,^{31, 32} inflammation,^{33, 34} obesity,³⁵ and cancer.^36–38 Interestingly, PPAR agonists have been shown to inhibit the transcription of VDR target gene CYP24A1 in the presence of 1,25-(OH)₂D₃.³⁹ The results presented herein show that GW0742, at micromolar concentrations, behaves as an antagonist of VDR and other nuclear receptors.

Experimental Procedures

Reagents

1,25-(OH)₂D₃ (calcitriol) was purchased from Endotherm, Germany; GW0742 was purchased from Tocris, triiodothyronine (T3), GW7647, Rosiglitazone, dihydrotestosterone (DHT), Bexarotene, and estradiol were purchased from Sigma. LG190178 was synthesized using a published procedure.⁴⁰

Labeled coactivator Peptides

Peptides, such as SRC2–3 (CLQEKHRILHKLLQNGNSPA),⁴¹ SRC2-2 [CEKHKILHRLLQDSS], DRIP2 [CNTKNHPMLMNLLKDNPAQD] were purchased and labeled with cysteine-reactive fluorophores, such as fluorescein maleimides, Texas-Red maleimides and Alexa Fluor 647 maleimides, in DMF/PBS 50:50. After purification by HPLC, the corresponding labeled peptides were dissolved in DMSO and stored at −20°C.

Protein Expression and Purification

The VDR-LBDmt DNA was kindly provided by D. Moras⁴² and cloned into pMAL-c2X vector (New England Biolabs). A detailed expression and purification protocol for VDR was reported.⁴¹ Detailed expression and purification of protocols for PPARγ-LBD, TRβ-LBD, and AR-LBD were reported as well.⁴³

High Throughput FP Assay

The HTS was carried out at the NIH Chemical Genomic Center (NCGS) Bethesda. For details see [AID:504847/pubchem].

VDR–SRC2–3 inhibition confirmation assay

These assays were conducted in 384 well black polystyrene plates (Corning) using a buffer (25 mM PIPES (pH 6.75) 50 mM NaCl, 0.01% NP-40, 2% DMSO, VDR-LBD protein (0.6 µM), LG190178 (5 µM), and Alexa Fluor 647-labeled SRC2–3, Texas Red-labeled SRC2–3, or Fluorescein-labeled SRC2–3 (7.5 nM) and optional mercaptoethanol 1 mM or 100 mM as indicated). Small molecule transfer into 20 µl assay solution was accomplished using a stainless steel pin tool (V&P Scientific), delivering 100 nl of a 10 mM compound solution. Fluorescence polarization was detected after 2h at an excitation/emission wavelength of 650/665 (Alexa Fluor), 595/615 nm (Texas Red), 495/520 (Fluorescein). Three independent experiments were carried out in quadruplicate and data was analyzed using nonlinear regression with variable slope (GraphPrism).

VDR Ligand Competition Assay

Ligand antagonism was determined by using a FP assay (PolarScreen, Invitrogen), which employs a rhodamine-labeled VDR ligand. Two independent experiments were conducted in quadruplicate and data was analyzed using nonlinear regression with variable slope (GraphPrism).

Transcription assays

Briefly, HEK 293T cells (ATTC) were cultured in 75 cm2 flasks using DMEM/High Glucose (Hyclone, #SH3024301), non-essential amino acids, HEPES (10 mM), penicillin and streptomycin, and 10% of dialyzed FBS (Invitrogen,#26400–044). At 70–80 percent confluency, 2 ml of untreated DMEM containing 1.56 µg of NR plasmid, 16 µg of a luciferase reporter gene, Lipofectamine™ LTX (75 µl), and PLUS™ reagent (25 µl) was added. After 16 hours, the cells were harvested with 0.05% Trypsin (3 ml) (Hyclone, #SH3023601), added to 15 ml of DMEM high Glucose (Hyclone, #SH3028401), non-essential amino acids, sodium pyruvate (1 mM), HEPES (10 mM), penicillin and streptomycin, and 2% percent charcoal treated FBS (Invitrogen, #12676–011), and spun down for 2 minutes at 1000 rpm. The cell were resuspended in the same media and plated in sterile cell culture treated black 384 well plates with optical bottom (Nunc, #142761) at a concentration of 15,000 cells per well, which were treated with a 0.25% solution of Matrigel (BD Bioscience, #354234) beforehand. After 2 hours, plated cells were treated with small molecules in vehicle DMSO, followed by a 16 hours incubation time. The following NR agonists were used: 1,25(OH)₂D₃ (10 nM), GW7647 (30 nM), Rosiglitazone (300 nM), GW0742 (50 nM), DHT (10 nM), Bexarotene (200 nM), T3 (10 nM), and estradiol 10 nM). Transcription was determined using Bright-Glo (Promega). Cell viability was determined using Cell-Titer-Glo (Promega) for identical treated cells. Controls for cell viability were 3-dibutylamino-1-(4-hexyl-phenyl)-propan-1-one (100 µM in DMSO) (positive) and DMSO (negative). Three independent experiments were conducted in quadruplicate and data was analyzed using nonlinear regression with variable slope (GraphPrism).

Western Blot of in vitro binding reactions between SRC2 bearing all three NIDs and VDR-LBD in the presence of GW0742

GST fusion to the SRC2 bearing all three NIDs were expressed in Escherichia coli BL21. Cultures were grown to OD600 = 0.5–0.6 at 22 °C and induced with 0.5 mM isopropyl-D-thiogalactoside for 12 h. The cultures were centrifuged (1000 x g), and bacterial pellets were resuspended in 20 mM TRIS, pH 7.4, 200 mM NaCl, 1 mM NaN₃, 0.5M EDTA, 1 mM DTT, protein inhibitors cocktail (Roche) and sonicated. Debris was pelleted by centrifugation (100,000 x g). The supernatant was incubated with glutathione-Sepharose 4B beads (Amersham Biosciences) and washed. Protein on bead was stored with 10% glycerol at –20°C. Each pull-down reaction was carried out in 100 µl buffer (25 mM PIPES (pH 6.75) 50 mM NaCl, 0.01% NP-40, 2% DMSO) using 100 nM calcitriol, 10 µM VDR-LBD-MPB, and GW0742. After 2 hours at rt, 15 µl of SRC2-beads was added to each reaction followed by 30 minutes incubation. The reaction was filtered, washed with buffer (100 µl) and eluted from the bead using a buffer and 10 mM reduced glutathione. Separation was carried out using SDS-PAGE followed Western blotting using standard procedures with anti-MBP (E8032S, New England BioLabs) and anti-mouse IgG-Tr (sc-2781, Santa Cruz).

Semi-quantitative real time PCR

Briefly, LNCaP cells (ATTC) were cultured in 75 cm2 flasks using DMEM/High Glucose (Hyclone, #SH3024301), non-essential amino acids, HEPES (10 mM), penicillin and streptomycin, and 10% dialyzed FBS (Invitrogen, #26400–044). LNCaP cells were incubated at 37 °C with GW0742 (20 µM) in the presence or absence of DHT (10 nM), Rosiglitazone (5 µM), Triiodothyronine (10 nM), or 1,25(OH)₂D₃ (10 nM) for 18 h in a 6 well plate. Cells of each well were harvested using 0.3 ml of 0.05% Trypsin (Hyclone, #SH3023601) and added to DMEM media (1 mL). The cell suspension was spun down for 2 min at 1000 rpm, media was removed and the cell pellet was resuspended in RTL buffer (RNAeasy kit, Qiagen) with the addition of mercaptoethanol. The cells were lysed using a QIAshredder (Qiagen) and total RNA was isolated using RNAeasy kit (Qiagen). A QuantiFast SYBR Green RT-PCR Kit (Qiagen) was used for the real time PCR following manufacturer’s recommendations. Primers used in these studies are as follows: GAPDH FP 5’-accacagtccatgccatcac-3’, GAPDH RP 5’-tccaccaccctgttgctgta-3’; TRPV6 FP 5’-ACTGTCATTGGGGCTATCATC-3’, TRPV RP 5’-CAGCAGAATCGCATCAGGTC-3’; IGFBP3 FP 5’-CGCCAGCTCCAGGAAATG-3’, IGFBP3 RP 5’- GCATGCCCTTTCTTGATGATG-3’; UGT1A1 FP 5’-GCCATGCAGCCTGGAATT-3', UGT1A1 RP 5'-GGCCTGGGCACGTAGGA-3'; PSA FP 5'-TCTGCCTTTGTCCCCTAGAT-3', PSA RP 5'-AACCTTCATTCCCCAGGACT-3'; BTG2 FP 5'-AGGGTAACGCTGTCTTGTGG-3', BTG2 RP 5'-TACAGTTCCCCAGGTTGAGG-3'; ANGPTL4 FP 5'-AGGGTAACGCTGTCTTGTGG-3', ANGPTL4 FP 5'-TACAGTTCCCCAGGTTGAGG-3'. Real-time rt-PCR was carried out on a Mastercycler (Eppendorf). We used the ΔΔCt method to measure the fold change in gene expression of target genes. Standard errors of mean were calculated from 4 biological independent experiments performed in triplicates.

HL60 differentiation assay

HL60 cells (ATTC) were cultured in 75 cm² flasks using RPMI-1640 (ATTC, #30–2001), non-essential amino acids, penicillin and streptomycin, and 10% dialyzed FBS (Invitrogen, #26400–044). The cells were treated with 1,25(OH)₂D₃ (50 nM) or vehicle DMSO and plated at a concentration of 2M cell/ml in 384 well plates with optical bottom (Nunc 142761). Cells were incubated with different concentrations of GW0742 for 4 days at 37°C. 4 µl of a nitro blue tetrazolium chloride (5.4 mg/ml) and phorbol-12-myristate-13-acetate (PMA) (100 µg/ml) was added to each well and incubated for 30 minutes at 37°C. The amount of oxidized nitro blue tetrazolium was quantified at a wavelength of 550 nm applying a 4×4 scan of full well area using the Tecan M1000. Two independent experiments were conducted in quadruplicate and data was analyzed using nonlinear regression with variable slope (GraphPrism).

Results

A small molecule screen of about 390,000 compounds at different concentrations was conducted in 1536-well black polystyrene plates using fluorescence polarization (FP) [AID:504847/pubchem]. The HTS identified inhibitors of the interaction between VDR and fluorescently labeled peptide SRC2–3, representing the NR interaction domain of the steroid receptor coactivator 2 (SRC2). The screen was carried out in the presence of VDR-LBD protein (800 nM), VDR agonist LG190178⁴⁰ (5 µM) and Alexa Fluor 647 labeled coregulator peptide SRC2–3 (7.5 nM). Fluorescence polarization was detected at an emission wavelength at 650 nm and excitation wavelength of 665 nm. 1938 compounds exhibited IC₅₀ values of 40 µM and lower in the primary screen (Figure 1, A). Furthermore, these compounds were validated with two alternative FP assays employing Texas Red-labeled SRC2–3 (excitation/emission 595/615 nm) [AID:602201/pubchem] and Fluorescein-labeled SRC2–3 (excitation/emission 495/520 nm) [AID:602200/pubchem] (Figure 1, B and C).

Figure 1.

IC₅₀ values of 1938 hit compounds using VDR-LBD and SRC2–3 labeled with: A Alexa Fluor 647; B Texas-Red in correlation with Alexa Fluor 647; C Fluorescein in correlation with Alexa Fluor 647.

Further characterization of the 1938 initial hits revealed that 69% of the compounds exhibited IC₅₀ values of 40 µM and lower in the FP assay employing VDR-LBD and Texas Red-labeled SRC2–3 (Figure 1, B). A good correlation between compound activities in this assay and validation assays was observed. 83% of the 1938 hit compounds also exhibited IC₅₀ values of ≤ 40 µM in the FP assay employing Fluorescein-labeled SRC2–3 (Figure 1, C). Also in this case, a good correlation with the assay employing SRC2–3 Alexa Fluor 647 was observed.

Based on these results, 747 compounds were selected with an emphasis on diverse scaffolds and functionality. In order to delineate compounds that might inhibit coactivator binding by irreversibly reacting with cysteine residues of the VDR-LBD protein, the VDR–SRC2–3 FP assay was carried out in the presence of 1 mM or 100 mM 2-mercaptoethanol (ME). The hypothesis here is that electrophilic compounds are more likely to react in the presence of excess nucleophilic ME than with protein nucleophilic protein residues such as cysteine. We introduced this method for the electrophilic compound DHPPA, which showed reduced thyroid receptor binding in the presence of 100 mM ME.⁴⁴ The results of both assays are depicted in Figure 2.

Figure 2.

Inhibition for the VDR-SRC2–3 Alexa Fluor 647 interaction by small molecules (50 µM) in the presence of 1 mM or 100 mM of 2-mercaptoethanol (ME).

Interestingly, we observed significant reduction of the majority of compounds to inhibit the interaction between VDR and SRC2–3 in the presence of 100 mM ME in comparison with 1 mM ME (Figure 2, B). This suggests that a large number of our initial hits may inhibit VDR–coactivator interactions via modification of surface cysteine residues. However, activities of a significant minority of compounds were not influenced by a higher concentration of ME (100 mM) and among these was GW0742 (Figure 3, A).

Figure 3.

VDR–GW0742 interactions. A chemical structure of GW0742; B FP VDR ligand competition assay; C Fluorescence interaction between GW0742 and rhodamine-labeled VDR ligand.

The following activities were measured for GW0742 during the HTS campaign: 1. VDR–SRC2–3 inhibition (FP assays) (IC₅₀ value): 14 µM (Alexa Fluor 647), 25.1 µM (Fluorescein), inconclusive (Texas-Red); 2. VDR transcription assay (VDR–GeneBLAzer, Invitrogen): The IC₅₀ value was 26 µM at a maximum response of 30% as defined by the absence of VDR agonist 1,25(OH)₂D₃ [AID:602202, pubchem]; 3. Toxicity in HEK293T cells (CellTiter-Glo, Promega): 18% cell death at a concentration of 45.8 µM [AID:602204, pubchem].

In order to determine that GW0742 is interacting with VDR and not the coactivator a commercially available VDR ligand binding FP assay was used employing VDR and a rhodamine-labeled VDR ligand (Figure 3, B).

Using increasing concentrations of GW0742, competition of VDR–ligand binding was confirmed with an IC₅₀ value of 8.7 ± 1.7 µM. Interestingly, partial inhibition was observed with respect to 100% free probe (55 mP) measured in the absence of VDR (Figure 3, B (square)). Prompted by this result and by the fact that the FP assay employing Texas-Red SRC2–3 was inconclusive during the HTS campaign, we analyzed the interaction between GW0742 and rhodamine-labeled VDR ligand in the absence of VDR (Figure 3, C). Surprisingly, we found that both fluorescence intensity and fluorescence polarization of the rhodamine-labeled VDR ligand increased at higher concentrations of GW0742. Similar results were found for the Texas-Red labeled SRC2–3 in the presence of GW0742 using the same excitation and emission wavelength (results not shown). GW0742 by itself showed no significant fluorescence under these conditions. Based on these results, we concluded that GW0742 inhibited the interaction between VDR and rhodamine-labeled VDR ligand, but that IC₅₀ values measured with Alexa Fluor 647 labeled SRC2–3 are a more reliable measure of the inhibitory activity of the ligand. No change of fluorescence intensity and fluorescence polarization was detected at an emission wavelength at 650 nm and excitation wavelength of 665 nm for Alexa Fluor 647 labeled SRC2–3 and increasing concentrations of GW0742 (results not shown). In addition, we investigated the interaction between GW0742 and SRC2 using isothermal titration calorimetry (ITC) but no specific interaction was observed (supporting information).

FP assays were carried out to determine the selectivity of GW0742 with respect to its ability to inhibit the interaction between other nuclear receptors and Alexa Fluor 647-labeled coactivator peptides. These included VDR and SRC2–3, thyroid hormone receptor β (TRβ) and SRC2–2, peroxisome proliferation activated receptor γ (PPARγ) and DRIP2, derived from coactivator DRIP205⁴⁵, and AR and SRC2–3. The results are depicted in Figure 4.

Figure 4.

Summary of nuclear receptor-coactivator inhibition in the presence of GW0742 using FP. A Inhibition of the VDR–SRC2–3 interaction; B Inhibition of the TRβ–SRC2–2 interaction; C Inhibition of the PPARγ–DRIP2 interaction; D Inhibition of the AR–SRC2–3 interaction. The conditions for different NRs were as follows: VDR-LBD (0.8 µM), Alexa Fluor 647 labeled SRC2–3 (7 nM), and LG190178 (2 µM); TRβ-LBD (1 µM), Alexa Fluor 647 labeled SRC2– 2 (7 nM), and triiodothyronine (10 nM); PPARγ-LBD (5 µM), Alexa Fluor 647 labeled DRIP2 (7 nM), and rosiglitazone (5 µM); AR-LBD (5 µM), Alexa Fluor 647 labeled SRC2–3 (7 nM), and dihydrotestosterone (10 nM) were incubated with GW0742 for 2h.

The inhibition of VDR-SRC2–3 binding in the presence of GW0742, demonstrated during the HTS campaign, was confirmed with a calculated IC50 value of 27.2 ± 2.7 µM (Figure 4, A). The ability of GW0742 to disrupt the TR–SRC2–2 and the PPARγ–DRIP2 interactions was modest with IC50 values of 59.9 ± 9.5 µM and >86 µM, respectively (Figure 4, B and C). The strongest GW0742 inhibition was observed for the AR–SRC2–3 interaction with an IC50 value of 6.6 ± 1.5 µM (Figure 4, D).

Many other groups have investigated GW0742 in respect to different nuclear receptor binding using a variety of assays. These include PPARα, PPARγ and PPARδ using cell-based transactivation assays (alkaline phosphatase as reporter enzyme), where GW0742 exhibited agonistic EC50 values of 1.1 ± 0.109 µM (PPARα), 2.0 ± 1.3 µM (PPARγ), and 0.001± 0.002 µM (PPARδ), respectively. Antagonistic activity of GW0742 was also observed in HTS campaigns using a GeneBLAzer assays (Invitrogen) employing the estrogen receptor α (ERα, AID:588513, pubchem), the glucocorticoid receptor (GR, AID:588533, pubchem), PPAR δ (AID:588535, pubchem), PPARγ (AID588537, pubchem), the retinoid X receptor (RxR, AID:588546, pubchem), and TRβ (AID:588547, pubchem). Activity of GW0742 found for the majority of these nuclear receptors was in the micromolar range and exhibited inconclusive dose-response curves in most cases.

Therefore, an array of nuclear receptor mediated transcription assays was used to confirm the pan nuclear receptor–coactivator inhibition caused by high concentrations of GW0742 in the FP assay. For VDR, TRα, and TRβ, HEK293T cells were transfected with nuclear receptor expression vector and luciferase reporter under control of the cognate nuclear receptor response elements or target gene promoters. For all other nuclear receptors, one-hybrid assays employing nuclear receptor LBDs fused to the DNA binding domain (DBD) of GAL4 and a luciferase reporter plasmid driven by five copies of the GAL4 binding site fused to a tk promoter were employed. The assays were carried out in 384-well plates and the amount of luciferase formed was quantified after 18 hours using Bright-Glo. The transcription assays were carried out in the presence and absence of endogenous ligands or synthetic agonists using different concentrations of GW0742. The EC₅₀ and IC₅₀ values are summarized in Table 1.

Table 1.

Evaluation of GW0742 in different nuclear receptor reporter assays.

Entry	Nuclear Receptor	Agonist EC50 (µM)11	Antagonist IC50 (µM)11
1	VDR	10	14.7 ± 1.51
2	PPARα	1.3 ± 0.3	37.4 ± 8.22
3	PPARγ	2.8 ± 0.7	20.2 ± 5.43
4	PPARδ	0.0037 ± 0.0014	21.6 ± 4.94
5	AR	10	12.1 ± 5.35
6	RxRα	10	22.9 ± 3.86
7	TRα	10	31.4 ± 4.07
8	TRβ	10	25.8 ± 5.28
9	ERα	10	21.3 ± 7.29

¹1,25(OH)₂D₃(10nM)

²GW7647 (30nM)

³Rosiglitazone (300nM)

⁴GW0742 (50nM)

⁵DHT (10nM)

⁶Bexarotene (200nM)

⁷T3 (10nM)

⁸T3 (10nM)

⁹estradiol (10nM)

¹⁰no activation detected

¹¹three independent experiments were conducted in quadruplicate and data was analyzed using nonlinear regression with variable slope (GraphPrism).

As expected, GW0742 was only able to activate transcription mediated by PPARα, PPARγ, and PPARδ among the nuclear receptors tested. The measured EC₅₀ values, 1.3 ± 0.3 µM (PPARα), 2.8 ± 0.7 µM (PPARγ), and 0.0037± 0.0014 µM (PPARδ) (Table 1, entries 2, 3, and 4), were in agreement with previously reported results ²⁸. In contrast, inhibition of transcription was found for all nuclear receptors within an IC₅₀ value range of 12.1 to 37.4 µM. Interestingly, GW0742 exhibited significantly lower IC₅₀ values for VDR and AR relative to other nuclear receptors with IC₅₀ values of 12.1 µM and 14.7 µM, respectively (Table 1, entries 1 and 5). The cell viability in the presence of GW0742 was quantified under the same conditions using CellTiter-Glo (data not shown). At a GW0742 concentration of 37.5 µM, 93 ± 3 % of the cell population was viable and at a concentration of 18.7 µM of GW0742 no toxicity was observed. This observation is in agreement with the HTS data described before.

For PPARs, GW0742 exhibited an agonistic effect at lower concentration and an antagonistic effect at higher concentrations as shown for PPARγ for the FP assay (Figure 4) and transcription assay depicted in Figure 5. GW0742 EC₅₀ values determined in both assays are very similar with 2.6 µM (FP assay) and 2.8 µM (transcription), respectively. At GW0742 concentrations of >20 µM, we observed significant PPARγ-mediated transcriptional inhibition (Figure 5, B), whereas significant inhibition between PPARγ-LBD and DRIP2 was observed for GW0742 at higher concentrations (>86 µM).

Figure 5.

Agonistic and antagonist effect of GW0742 with respect to PPARγ. A FP assay using PPARγ-LBD and Alexa Fluor 647-labeled DRIP2; B One-hybrid transcription assay employing PPARγ.

GW0742 also inhibited the interaction between VDR-LBD and an SRC2 fragment bearing three nuclear interaction domains in a western pull-down assay (Figure 6). Control experiments indicated that SRC2 binds to VDR in the presence of VDR ligand LG190178 (Figure 6, lane 8) but not in the absence of ligand (Figure 6, lane 9). The faint bands in the absence of ligand and VDR were caused by nonspecific binding of VDR antibody to SRC2 beads. The VDR–SRC2 interaction was blocked in a dose dependent manner by GW0742 (Figure 8, lane 1–7). Although significant inhibition of the VDR–SRC2 interaction was observed at 80–120 µM GW0742, residual bands could still be detected partially due to unspecific binding of the VDR antibody to SRC2 beads as observed for the controls.

Figure 6.

Inhibition of the VDR–SRC2 interaction by GW0742 analyzed by a Western pull-down assay.

Figure 8.

Inhibition of differentiation of HL-60 cells induced by 1,25(OH)₂D₃ (50 nM) in the presence of GW0742.

LNCaP cancer cells express different nuclear receptors including AR⁴⁶, PPARγ⁴⁷, TRβ⁴⁸ and VDR⁴⁹. Furthermore, steroid receptor coactivator 1, 2 and 3 and DRIP205, essential for transcriptional activation, are expressed in these cells.^{50, 51} Based on these reports, we investigated the ability of GW0742 to inhibit nuclear receptor transcriptional activity by quantifying the RNA levels of different target genes in the presence and absence of endogenous or synthetic ligands (Figure 7).

Figure 7.

Gene regulation by GW0742 (20 µM, or 50 µM for H) in LNCaP cells after 18 hours (or as indicated) in the presence and absence of NR ligands. A and B DHT (10 nM); C Rosiglitazone (5 µM); D Triiodothyronine (10 nM); E-H 1,25(OH)₂D₃ (10 nM). Standard errors of mean were calculated from four biological independent experiments performed in triplicates. Stars represent P<0.05 (*), P<0.01 (**), P<0.001 (***) (Student’s t-test).

For AR, expression of UDP glucuronosyltransferase 1 family polypeptide A cluster (UGT1A1)⁵² and the prostate-specific antigene (PSA)⁵³ were upregulated in the presence of DHT (Figure 7, A and B). LNCaP cells treated with DHT (10 nM) and 20 µM GW0742 showed a significant inhibitory effect. PPARγ target gene angiopoietin-related protein 4 (FIAF/ANGPTL4)⁵⁴ was induced by rosiglitazone (5 µM) and this response was unaffected by GW0742 (20 µM) (Figure 7, C). TR target gene B-cell translocation gene 2 (BTG2)⁵⁵ was down-regulated in the presence of T3 as shown in Figure 7, D. This effect was reversed by GW0742 (20 µM) confirming the inhibitory action of GW0742 upon TR demonstrated in other assays. Finally, VDR target genes IGFBP-3^{56, 57} and TRPV6^{58, 59} (Figure 7, E and F) were strongly induced by 1,25(OH)₂D₃ (10 nM) and GW0742 selectively inhibited this effect. For VDR target gene CYP24A1,⁶⁰ we observed an induction at 6 hours and 18 hours in the presence of 1,25(OH)₂D₃. Interestingly, only at the earlier time point of 6 hours and a concentration of 50 µM GW0742 a significant inhibition of this induction was determined (Figure, 7, G and H).

As mentioned earlier, VDR governs expression of genes responsible for differentiation of human promyelocyte HL-60 cell into monocytes by 1,25(OH)₂D_3.^{61, 62} VDR antagonists with a secosteroid structure inhibit this effect.^18,19,22 The established differentiation assay employs nitro blue tetrazolium chloride to colorimetrically quantify the amount of cellular superoxide in the presence of phorbol-12-myristate-13-acetate (PMA) that function as a stimulant of the respiratory burst.⁶³ HL-60 cells were incubated for 4 days with and without 1,25(OH)₂D₃ (10 nM) and GW0742 in the a dose dependent manner. The results are shown in Figure 8.

A strong HL-60 cell differentiation was observed in the presence of 1,25(OH)₂D₃. A significant reduction of the 1,25(OH)₂D₃-induced differentiation was observed for GW0742 at concentration of higher than 37.5 µM. Significant HL-60 cell death was only observed for a GW0742 concentration of 150 µM.

Discussion

This study reports a successful HTS campaign to identify new inhibitors of VDR–coactivator interaction. A novel strategy to delineate potential irreversible inhibitors, which are likely to react with nucleophilic protein residue, from a pool of hit compounds was introduced by executing a secondary screen in the presence of a high concentration of the nucleophile ME. A concentration of 100 mM ME was essential to significantly reduce the number of hit compounds.

GW0742, a PPARδ agonist, was identified among the VDR–coactivator inhibitors that were not influenced by ME. Interestingly, GW0742 exhibited characteristics of a pan nuclear receptor antagonist at concentrations higher than 12.1 µM, as demonstrated for biochemical FP assays and cell-based transcription assays. While GW0742 inhibited activity of a large number of nuclear receptors, albeit at micromolar concentrations, it inhibited VDR and AR activity most effectively. Interestingly, GW0742 also behaved as an agonist/antagonist for PPARα, PPARγ, and PPARδ, activating transcription at low concentrations and inhibiting this effect at higher concentrations.

The mechanism by which GW0742 inhibits activity of VDR and other nuclear receptors is not absolutely clear. Unexpectedly, we found that GW0742 inhibits the interaction between VDR and its small molecule ligand. While ligand competition appears incomplete, we suspect that our ligand binding assay underestimates the effectiveness of GW0742 competition because of an unexpected spectroscopic property of GW0742 discovered in this study. In the presence of rhodamine derived compounds an increase of fluorescence intensity was observed at higher GW0742 concentrations using an excitation wavelength of 595 nm and emission wavelength of 615 nm. The interaction between GW0742 and rhodamine compound also increase fluorescence polarization at the same excitation and emission wavelength. This unusual spectroscopic behavior of GW0742 might partly mask its ability to compete with the fluorescent VDR agonist and could also represent an underlying reason for a number of false positive FP HTS hits. Thus, this data is suggestive of a conventional antagonist mode of action, in which GW0742 interacts with the VDR ligand binding site and allosterically disrupts the interaction between VDR and SRC2.

Results with PPARs, in which low concentrations of GW0742 display agonist behaviors and high concentrations inhibit this effect, suggest that other mechanisms may be at play. Similar behaviors have been reported for 4-hydroxytamoxifen (HT) in regard to estrogen receptors (ERs). A crystal structure of ERβ showed one HT molecule bound to the ligand binding site and another at the coactivator binding site.⁶⁴ As a result, ER was activated at low HT concentrations by ligand-induced conformational changes that allowed coactivator recruitment but ER-coactivator interaction was inhibited at higher HT concentrations by competition for ER coactivator binding.⁶⁵ A similar mechanism might be one explanation for the agonist/antagonist behavior of GW0742 in regard to PPAR binding. There are other possibilities; PPARs possess large ligand binding pockets that can accommodate more than one ligand.⁶⁶ Thus, high GW0742 concentrations could result in unusual PPAR/ligand stoichiometries that could trigger inactive receptor conformations. Clarification of this issue will require further investigation.

Regardless of the mechanism of action, our studies revealed that GW0742 is effective in cell culture models. The therapeutic ratio (efficacy: toxicity) for GW0742 is relative small for VDR-mediated processes (5:1) but very large for PPARδ (20000:1). The downstream effect of nuclear receptor–coactivator inhibition by GW0742 resulted in the suppression of VDR genes IGFBP-3 and TRPV6 in LNCaP cells in the presence of 1,25(OH)₂D₃. Other than in the prostate, the proteins encoded by these genes are highly expressed in the placenta.⁶⁷ Therefore, GW0742 treatments might be problematic during pregnancy. As observed by other groups, the classic VDR target gene CYP24A1 was moderately upregulated by 1,25(OH)₂D₃ in LNCaP cells.⁴⁹ The inhibition of CYP24A1 transcription induced by 50 µM GW0742 at only the earlier time point is likely due to different expression levels of CYP24A1 over time, which has been shown in PC-3 cells.⁶⁸ In addition, we found the inhibition of transcription of AR target genes UGT1A1 and PSA in the presence of GW0742 in LNCaP when stimulated with DHT. The proteins encoded by these genes are found in increased levels in patients with prostate cancer and prostatitis. Thus, GW0742 might have beneficial effects for both conditions. BTG2, a TRβ target gene, encodes the coregulator BTG2, which has been shown to inhibit cell cycle progression at the G1 checkpoint by repressing cyclin D1 transcription.⁶⁹ In the presence of T3, BTG2 is down-regulated in LNCaP cells but the T3 suppression is inhibited in the presence of 20 µM GW0742. The regulation for PPAR target gene ANGPL4, which plays a significant role in fat metabolism, was inconclusive at the GW0742 concentration tested. ANGPTL4 activation in the presence of Rosiglitazone and GW0742 was reported for HT29 colon cancer cells although at a GW0742 concentration of 1 µM.⁷⁰ Thus, high local concentrations of GW0742 have a strong influence on the gene regulation mediated by PPARs and other nuclear receptors. The activation of transcription by 1,25(OH)₂D₃ induces differentiation of HL-60 cells, a process which was shown to be inhibited by GW0742 in a dose dependent manner. Overall, we have demonstrated the interactions between GW0742 and different NRs with diverse transcriptional and biological effects underlining the possibility that GW0742 might share a privileged scaffold recognized by many members of the NR superfamily. In respect to VDR, GW0742 is the first ligand antagonist lacking a secosteroid structure, which inhibits the transcriptional activation of VDR target genes in the presence of 1,25(OH)₂D₃. Because of the fact that the three dimensional folding of VDR and resulting interactions with coregulators are highly depended on the structure of the bound ligand,⁷¹ we expect that GW0742 will exhibit different biological responses than current secosteroid VDR antagonist. We are currently in the process to develop new VDR-selective antagonists based on GW0742 in order to determine the downstream effect of VDR–coactivator inhibition without influencing PPAR signaling.

Abbreviations

VDR

vitamin D receptor

NR

nuclear hormone receptor

FP

fluorescence polarization

SRC2

steroid receptor coactivator 2

SRC2–3

third NR interaction domain of the steroid receptor coactivator 2 peptide

TRPV6

transient receptor potential vanilloid type 6 gene

CYP24A1

1,25-dihydroxyvitamin D3 24-hydroxylase

1,25-(OH)₂D₃

1,25-dihydroxyvitamin D₃, DBD, DNA binding domain

LBD

ligand-binding domain

RxR

retinoid X receptor

HTS

high throughput screening

DMSO

dimethyl sulfoxide

HEK293T

human embryonic kidney cells

AR

androgen receptor

ERα

estrogen receptor alpha

TRβ

thyroid receptor beta

PPARα

peroxisome proliferator-activated receptor α

PPARγ

peroxisome proliferator-activated receptor γ, PPARδ, peroxisome proliferator-activated receptor δ

DRIP205

VDRR-interacting protein 205

GST

glutathione-S-transferase

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

CYP24A1

1,25-dihydroxyvitamin D3 24-hydroxylase gene

NCGC

NIH Chemical Genomic Center

ME

mercaptoethanol

T3

triiodothyronine

DHT

dihydrotestosterone

GAL4

yeast transcription activator protein GAL4

HT

4-hydroxytamoxifen

UGT1A1

UDP glucuronosyltransferase 1 family polypeptide A cluster

PSA

prostate-specific antigene

ANGPTL4

angiopoietin-related protein 4

BTG2

B-cell translocation gene 2

IGFBP-3

insulinlike growth factor-binding protein

HL-60

human promyelocytic leukemia cell

PMA

phorbol-12-myristate-13-acetate

DMF

dimethylforamide

PBS

Phosphate-buffered saline

HPLC

high pressure liquid chromatography

PIPES

piperazine-N,N’-bis(2-ethanesulfonic acid

NP-40

Tergitol-type NP-40

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

FBS

Fetal bovine serum

DMEM

Dulbecco’s Modification of Eagle’s Medium

NID

nuclear receptor interaction domain

EDTA

Ethylenediaminetetraacetic acid

DTT

Dithiothreitol

MBP

maltose binding protein

LNCaP

human prostate cancer cells

PMA

phorbol-12-myristate-13-acetate