In silico discovery and validation of potent small-molecule inhibitors targeting the activation function 2 site of human oestrogen receptor α
- 4.7k Downloads
Current approaches to inhibit oestrogen receptor-alpha (ERα) are focused on targeting its hormone-binding pocket and have limitations. Thus, we propose that inhibitors that bind to a coactivator-binding pocket on ERα, called activation function 2 (AF2), might overcome some of these limitations.
In silico virtual screening was used to identify small-molecule ERα AF2 inhibitors. These compounds were screened for inhibition of ERα transcriptional activity using stably transfected T47D-KBluc cell line. A direct physical interaction between the AF2 binders and the ERα protein was measured using biolayer interferometry (BLI) and an ERα coactivator displacement assay. Cell viability was assessed by MTS assay in ERα-positive MCF7 cells, tamoxifen-resistant (TamR) cell lines TamR3 and TamR6, and ERα-negative MDA-MB-453 and HeLa cell lines. In addition, ERα inhibition in TamR cells and the effect of compounds on mRNA and protein expression of oestrogen-dependent genes, pS2, cathepsin D and cell division cycle 2 (CDC2) were determined.
Fifteen inhibitors from two chemical classes, derivatives of pyrazolidine-3,5-dione and carbohydrazide, were identified. In a series of in vitro assays, VPC-16230 of the carbohydrazide chemical class emerged as a lead ERα AF2 inhibitor that significantly downregulated ERα transcriptional activity (half-maximal inhibitory concentration = 5.81 μM). By directly binding to the ERα protein, as confirmed by BLI, VPC-16230 effectively displaced coactivator peptides from the AF2 pocket, confirming its site-specific action. VPC-16230 selectively suppressed the growth of ERα-positive breast cancer cells. Furthermore, it significantly inhibited ERα mediated transcription in TamR cells. More importantly, it reduced mRNA and protein levels of pS2, cathepsin D and CDC2, validating its ER-directed activity.
We identified VPC-16230 as an ERα AF2-specific inhibitor that demonstrated promising antiproliferative effects in breast cancer cell lines, including TamR cells. VPC-16230 reduced the expression of ERα-inducible genes, including CDC2, which is involved in cell division. We anticipate that the application of ERα AF2 inhibitors will provide a novel approach that can act as a complementary therapeutic to treat ERα-positive, tamoxifen-resistant and metastatic breast cancers.
KeywordsMCF7 Cell Root Mean Square Deviation Virtual Screening Fulvestrant Carbohydrazide
Biotinylated oestrogen receptor-alpha
Binding function 3
Cell division cycle 2
Effective concentration needed to achieve 80% of the maximum response
Epidermal growth factor receptor
Electronic high-throughput screening
Oestrogen response element
Foetal bovine serum
Glyceraldehyde 3-phosphate dehydrogenase
Human epidermal growth factor receptor type 2
Half-maximal inhibitory concentration
Insulin-like growth factor 1 receptor
Ligand Explorer module
Molecular Operating Environment
Peroxisome proliferator-activated receptor gamma, coactivator 1-alpha
Quantitative real-time reverse transcriptase polymerase chain reaction
Time-resolved fluorescence resonance energy transfer
Breast cancer (BCa) is the most common type of noncutaneous malignancy and the leading cause of cancer-related death in women worldwide [1,2]. A total of 232,340 new cancer cases and 39,620 cancer deaths were projected to occur in the United States alone in 2013 . Approximately 75% of BCa are classified as oestrogen receptor-alpha (ERα)-positive. Abnormal ERα-mediated activity is the characteristic feature of most of these BCa . The hormone oestradiol (E2) binds to ERα to regulate a wide range of genes associated with proliferation, survival and invasion of breast tumour cells [5,6]. For this reason, the aim of current therapies is to either reduce E2 levels or block signalling through ERα. For the last 30 years, tamoxifen has been the standard treatment for ERα-positive BCa in premenopausal women and for postmenopausal women who have relapsed on aromatase inhibitors. However, most of the patients with advanced disease develop tamoxifen resistance, and one-third of the patients given adjuvant treatment will develop recurrent disease within 15 years of surgery (acquired resistance) due to the progression of the surviving tumour cells to a resistant state [7-9].
Although the factors responsible for development of resistance are not fully understood, several potential mechanisms have been proposed . Altered expression and/or modification of growth factor receptors known to cross-talk with the ERα signalling pathway, such as epidermal growth factor receptor (EGFR), human epidermal growth factor receptor type 2 (HER2) and insulin-like growth factor 1 receptor (IGF-1R) [11-13], and their downstream kinases, such as extracellular signal-regulated kinase 1/2, p38, Akt and p21-activated kinase [14,15], have been shown to correlate with tamoxifen resistance. It is noteworthy that in biopsies from patients with BCa who relapsed on tamoxifen, ERα expression was maintained in more than 50% of cases , and up to 80% of metastases from ERα-positive primary tumours retain ERα expression [17,18]. In addition, 20% of patients with resistant disease responded to a second-line treatment of either aromatase inhibitors or fulvestrant . Collectively, these studies suggest the continuing involvement of ERα coregulatory proteins and cross-talk between the ERα pathway and other growth factor and kinase pathways in resistance. Thus, ERα signalling remains an important therapeutic target in resistant disease. Based on these findings, several combination therapies have been proposed [20-22]. However, the problem of resistance remains unresolved, despite the use of the additional endocrine agents. There is a clear unmet need to develop an entirely new class of ERα inhibitor drugs with alternative mechanisms of action to replace or supplement existing treatments.
Previously, it has been shown that it is possible to inhibit ERα transcriptional activity by blocking the association of coactivators at the AF2 site. Several groups have identified small-molecule inhibitors of the ERα AF2 site in the past by different approaches [28-37]. However, such compounds have not yet been shown to inhibit the growth of tamoxifen-resistant (TamR) cells. Moreover, the small-molecule inhibitors identified to date demonstrate only moderate micromolar potencies against the ERα and are not yet suitable for clinical application.
In this study, we employed our established computational drug discovery pipeline [38-40] to identify more potent and selective inhibitors of ERα that specifically bind to the AF2 site and inhibit the interaction between ERα and its coactivator proteins. Our lead compound significantly inhibited ERα transcriptional activity by blocking the association of coactivators at the AF2 site. It demonstrated good antiproliferative activity and reduced the expression of ERα-dependent genes. Importantly, it also inhibited the growth of TamR cell lines in an ERα-specific manner.
Preparation of the protein structure for docking
Virtual screening was carried out on the ERα crystal structure [PDB:3UUD] (1.60 Å resolution) . To prepare the protein structure for docking, all solvent molecules were deleted and bond orders for the ligand and the protein were adjusted. The missing hydrogen atoms were added, and side chains were then energy-minimised using the OPLS-2005 force field (as implemented by using Maestro software ). The ligand-binding region was defined by a 12 Å box centred on the crystallographic ligand of 3UUD. No van der Waals scaling factors were applied; the default settings were used for all other adjustable parameters.
The ZINC database version 8.0  was used for virtual screening against the ERα AF2 site. The compounds were imported into a molecular database using the Molecular Operating Environment (MOE) version 2012 . Hydrogen atoms were added after these structures were ‘washed’ (a procedure including salt disconnection, removal of minor components, deprotonation of strong acids and protonation of strong bases). The following energy minimisation was performed with the Merck molecular force field 94x, as implemented by the MOE, and optimised structures were exported into the Maestro suite in .sdf format.
Virtual screening, consensus scoring and voting
Initially, approximately 4 million compounds were docked into the AF2 site using the Glide SP module. Next, we redocked about 2 million molecules, which had a Glide docking score <5.0, into the same binding cavity using the electronic high-throughput screening (eHiTS) docking module. A total of 5 × 105 structures, which received eHiTS docking scores below the 3.0 threshold, were identified for further in in silico refinement.
The determined docking poses of the 5 × 105 selected compounds were evaluated by (1) Glide docking score; (2) eHiTS docking score; (3) predicting inhibitory constant (pKi) of protein-ligand binding using MOE SVL scoring.svl script to improve accuracy of the prediction of energies of hydrogen bonds and hydrophobic interactions; (4) calculating rigorous docking scores, defined by the Ligand Explorer (LigX) module of the MOE package, which accounts for receptor/ligand flexibility; and (5) computing the root mean square deviation (RMSD) between docking poses generated by the Glide and eHiTS programs to quantify their docking consistency. On the basis of these sorted output values from the above four procedures, each molecule would then receive a binary 1.0 vote for every ‘top 10% appearance’. The final cumulative vote (with the maximum possible value of 5) was then used to rank the training set entries. On the basis of the cumulative count, we selected the most highly voted (5 × 103) molecules and subjected their docking poses to visual inspection.
After this final selection step, we formed a list of 100 compounds that were purchased and tested experimentally. Among these 100 compounds, 14 were found to be active and demonstrated the ability to displace the coactivator peptide from the target AF2 site. Small molecules were purchased from established suppliers, including ASINEX (Moscow, Russia; compound VPC-13002), ChemBridge (San Diego, CA, USA; compounds VPC-16007, VPC-16003, VPC-16004, VPC-16222 and VPC-16223), ChemDiv (San Diego, CA, USA; compound VPC-16019), TimTec (Newark, DE, USA; compounds VPC-16041, VPC-16046, VPC-16038, VPC-16040 and VPC-16021) and Vitas-M Laboratory (Apeldoorn, the Netherlands; compounds VPC-16230, VPC-16225, and VPC-16236).
T47D-KBluc and MDA-MB-453 cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA). The MCF7 and HeLa cell lines were a gift from Dr Sandra Dunn (Division of Hematology and Oncology, Department of Pediatrics, University of British Columbia, Vancouver, BC, Canada). TamR cell lines TamR3 and TamR6 were kindly provided by Dr Euphemia Leung (University of Auckland, New Zealand) . Cells were cultured at 37°C in a humidified incubator with 5% CO2. The cell lines were maintained in the following culture media: for MCF7, phenol red–free RPMI 1640 (Gibco/Life Technologies, Grand Island, NY, USA) supplemented with 10% foetal bovine serum (FBS) (Gibco/Life Technologies); for T47D-KBluc, phenol red–free RPMI 1640 containing 4.5 g/L glucose (Sigma-Aldrich, St Louis, MO, USA), 10 mM 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (Sigma-Aldrich), pH 7.5, 1 mM sodium pyruvate (Life Technologies), 0.2 U/ml insulin (Sigma-Aldrich) and 10% FBS; for MDA-MB-453 and HeLa, Dulbecco’s modified Eagle’s medium (HyClone Laboratories/GE Healthcare Life Sciences, Logan, UT, USA) supplemented with 10% FBS (Gibco/Life Technologies); and for TamR3 and TamR6, phenol red–free RPMI 1640 containing 10% charcoal-stripped serum (CSS) (Gibco/Life Technologies) and 1 μM tamoxifen (Sigma-Aldrich).
Chemicals and antibodies
17β-Oestradiol, 4-hydroxytamoxifen (4-OHT) and tamoxifen were obtained from Sigma-Aldrich. E2 was dissolved in 100% ethanol. 4-OHT, tamoxifen and test compounds were dissolved in dimethyl sulphoxide (DMSO). Cold peroxisome proliferator-activated receptor gamma, coactivator 1-alpha (PGC-1α) peptide (EAEEPSLLKKLLLAPANTQ) was synthesised by Elim Biopharmaceuticals (Hayward, CA, USA).
The rabbit monoclonal anti-pS2 antibody (EPR3972) was obtained from Abcam (Cambridge, UK). The mouse monoclonal antibody against cell division cycle 2 (CDC2) was purchased from (JI-04-00640; RayBiotech, Norcross, GA, USA). The rabbit polyclonal antibody for α-actin (A2066) and mouse monoclonal anti-cathepsin D antibody (C0715) were obtained from Sigma-Aldrich.
Luciferase oestrogen receptor-alpha transcriptional assay
ERα-positive T47D-KBluc human BCa cells were grown in phenol red–free RPMI 1640 supplemented with 10% CSS for 5 days. The cells were seeded on a 96-well plate (2 × 104 cells/well). After 24 hours, the cells were treated with either the test compounds or 4-OHT in the presence of 1 nM E2. The test compounds were screened at two concentrations: 12 μM and 30 μM. 4-OHT was added at a final concentration of 5 μM. For generation of dose–response curves, the compounds were added at a range of 0.1 to 50 μM and 4-OHT was added at a range of 0.000006 to 3 μM. The medium contained 0.1% (v/v) ethanol and 0.1% (v/v) DMSO. Twenty-four hours after treatment, the medium was aspirated and the cells were lysed by adding 50 μl of 1× passive lysis buffer (Promega, Madison, WI, USA). The plates were placed on a shaker at room temperature for 15 minutes and then subjected to two freeze–thaw cycles to help lyse the cells. Next, 20 μl of the lysate from each treatment was transferred onto a white, 96-well, flat-bottomed plate (Corning Life Sciences, Corning, NY, USA), and the luminescent signal was measured after adding 50 μl of the luciferase assay reagent (Promega) on a TECAN M200 PRO microplate reader (Tecan, Männedorf, Switzerland). Differences in growth were normalised against total protein concentration, which was measured by bicinchoninic acid (BCA) assay.
To rule out binding at HBS, dose–response curves of test compounds (0.1 to 50 μM) and of 4-OHT (0.000095 to 50 μM) were generated in the presence of a set of higher concentrations (1, 10, 50 and 100 nM) of E2 following the same procedure described above.
For transient transfection, we used pGL2.TATA.Inr.luc plasmid, which contains three copies of vitellogenin oestrogen response element (ERE) upstream of the TATA promoter (3X ERE TATA luc; Addgene plasmid 11354). This is the same plasmid used to construct pGL2.TATA.Inr.luc.neo, used to create the stable cell line T47D-KBluc.
TamR3 and TamR6 cells were grown in phenol red–free RPMI 1640 supplemented with 10% CSS and 1 μM tamoxifen. The cells were seeded on a 96-well plate (2 × 104 cells/well). After 24 hours, the cells were cotransfected with 50 ng each of the ERα-responsive luciferase plasmid and a constitutive Renilla reporter (to normalise for variations in transfection efficiency) using TransIT-2020 reagent (Mirus Bio, Madison, WI, USA). Cells were treated the next day with the test compounds in the presence of 1 nM E2. The compounds were added in a twofold dilution ranging from 0.1 to 50 μM. Tamoxifen was added at concentrations ranging from 0.000095 to 6 μM, and fulvestrant was added in the range of 0.000095 to 50 μM. The medium contained 0.1% (v/v) ethanol and 0.1% (v/v) DMSO. Twenty-four hours after treatment, the medium was aspirated and the cells were lysed by adding 50 μl of 1× passive lysis buffer (Promega). Luciferase activities were assayed using the Dual-Luciferase Reporter Assay System (Promega).
Time-resolved fluorescence resonance energy transfer oestrogen receptor-alpha coactivator assay
Peptide displacement was assessed with the LanthaScreen TR-FRET ER Alpha Coactivator Assay kit (PV4544; Life Technologies) as per the instructions of the manufacturer. The compounds were tested in the range of 0.1 to 50 μM, and cold PGC-1α was added at threefold dilutions ranging from 1.8 to 50 μM.
For the peptide competition assay, the compounds were tested in the range of 0.05 to 400 μM in the presence of three different concentrations (250, 500 and 1000 nM) of fluorescein-PGC-1α peptide and the recommended concentrations of glutathione S-transferase (GST)-tagged ER-LBD (7.25 nM) and terbium (Tb)-labelled anti-GST antibody (5 nM). Briefly, a twofold serial dilution of the test compounds was prepared at 100× final concentration in DMSO. The compounds were diluted 50-fold in complete assay buffer (assay buffer containing 5 mM dithiothreitol) to get a 2× final concentration and 2% DMSO. The GST-tagged ER-LBD was prepared at 4× final concentration in complete assay buffer, and a 4× fluorescein-PGC-1α/4× Tb anti-GST antibody/4× effective concentration needed to achieve 80% of the maximum response (EC80) E2 mix was prepared separately in complete assay buffer. The EC80 of E2 was determined to be 6.1 μM in this assay. Ten microlitres of the diluted test compounds were added to a black, flat-bottomed, 384-well plate, followed by addition of 5 μl of the 4× ER-LBD mix. A 4× concentration of fluorescein-PGC-1α/4× Tb anti-GST antibody/4× EC80 E2 mix was added last. The plate was incubated at room temperature for 2 hours, and time-resolved fluorescence resonance energy transfer (TR-FRET) was analysed on a Synergy 4 hybrid microplate reader (BioTek, Winooski, VT, USA) with the settings at 340-nm excitation and 495- and 520-nm emission. The emission ratio (520:495) was analysed and plotted.
Oestradiol displacement assay
E2 displacement was assessed with the PolarScreen ER Alpha Competitor Assay Green kit (P2698; Life Technologies) as per the instructions of the manufacturer. For screening purposes, the compounds were tested at 20 μM in the presence of 25 nM full-length ERα and 4.5 nM fluorescein-labelled E2 (Fl-E2). For E2 ligand competition assay, a twofold serial dilution of the test compounds was prepared at a 100× final concentration in DMSO. The compounds were diluted 50-fold in assay buffer to get a 2× final concentration and 2% DMSO. Fifty microlitres of the diluted test compounds were added to a 50-μl mixture containing 2× full-length ERα and Fl-E2 in each well to obtain final concentrations of 3 to 150 μM of test compound in the presence of 25 nM full-length ERα and 4.5 nM Fl-E2. Unlabelled E2 was tested at concentrations ranging from 0.01 to 1,000 nM. After a 2-hour incubation, polarisation was measured as per the instructions of the manufacturer on a Tecan F500 microplate reader.
Bio-layer interferometry assay
The direct reversible interaction between small molecules and the ER-α was quantified by BLI using an Octet RED apparatus (Pall ForteBio, Menlo Park, CA, USA). The LDB of the biotinylated oestrogen receptor-alpha (bERα) was produced in situ with AviTag technology (Avidity, Aurora, CO, USA). The AviTag sequence (GLNDIFEAQKIEWHE) was incorporated at the N-terminus of the ERα LBD (peptides 302 to 552). A six-residue histidine tag was incorporated at the C-terminus of the ERα LBD for purification of the protein. Escherichia coli strain BL21 containing both biotin ligase and ERα LBD vectors was induced with 0.5 mM isopropyl-β-d-1-thiogalactopyranoside in the presence of 0.02 mM E2 and 0.15 mM biotin at 16°C overnight. The bacteria were then lysed by sonication, and the resulting lysate was purified by immobilised metal ion affinity chromatography with nickel agarose beads and cation exchange chromatography (HiTrap SP; GE Healthcare Life Sciences, Pittsburgh, PA, USA). Purified and biotinylated protein (bERα LBD at 0.05 mg/ml) was bound to the Super Streptavidin Sensors (Pall ForteBio) overnight at 4°C in assay buffer (20 mM Tris, pH 7.5, 500 mM NaCl, 0.2 mM tris(2-carboxyethyl)phosphine, 0.02 mM E2, 5% glycerol and 5% DMSO). The compounds were dissolved in the assay buffer in a twofold dilution series ranging from 3.1 to 100 μM. In all experiments, a known AF2-interacting peptide—PGC-1α (Elim Biopharmaceuticals)—was used as a control to confirm the functionality of the bERα LBD.
Cell proliferation was determined using the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulphophenyl)-2H-tetrazolium (MTS) assay. Cells were seeded in 96-well plates at a density of 5 × 103 cells/well. MCF7, TamR3, TamR6, MDA-MB-453 and HeLa cells were seeded in their respective media. On the following day, the cells were treated with test compounds (0.2 to 50 μM) in the presence of 1 nM E2 and incubated at 37°C in 5% CO2. After 96 hours, 30 μl of MTS reagent (CellTiter 96 AQueous One Solution Cell Proliferation Assay reagent; Promega) were added and incubated for 90 minutes at 37°C in 5% CO2. The production of formazan was measured at 490 nm.
mRNA levels were analysed by quantitative RT-PCR (qRT-PCR). For this purpose, serum-starved MCF7 cells were seeded onto six-well plates at a density of 6 × 105 cells/well. After 24 hours, the cells were treated with the either VPC-16230 or 4-OHT in the presence or absence of 1 nM E2. RNA was isolated after 24 hours with TRIzol reagent (Life Technologies) and purified with the RNeasy Mini Kit (QIAGEN, Valencia, CA, USA). The purified mRNA was quantified using a NanoDrop spectrophotometer (NanoDrop, Wilmington, DE, USA). RNA (0.5 μg) was reverse-transcribed using the iScript synthesis kit (Bio-Rad Laboratories, Hercules, CA, USA). cDNA product (100 ng) was added to the primer mix. The final concentration of the primers was 5 pM. The sequences of the primers used in the qRT-PCR experiments were as follows: pS2, forward 5′-TTGTGGTTTTCCTGGTGTCA-3′ and reverse 5′-GCAGATCCCTGCAGAAGTGT-3′; cathepsin D, forward 5′-CAGAAGCTGGTGGACCAGAAC-3′ and reverse 5′-TGCGGGTGACATTCAGGTAG-3′; CDC2, forward 5′-ACTGGCTGATTTTGGCCTTG-3′ and reverse 5′-TTGAGTAACGAGCTGACCCCA-3′; glyceraldehyde 3-phosphate dehydrogenase (GAPDH), forward 5′-TGCACCACCAACTGCTTAGC-3′ and reverse 5′-GGCATGGACTGTGGTCATGAG-3′. The fold change in expression of the gene was calculated using the 2−ΔΔCt method with GAPDH as the internal control.
MCF7 cells were serum-starved in phenol red–free RPMI 1640 containing 10% CSS for 5 days. The cells were then seeded onto a six-well plate at a density of 6 × 105 cells/well and treated the following day with the test compounds in the presence of 1 nM E2. After 24 hours, the medium was aspirated, and the cells were washed with ice-cold phosphate-buffered saline. Cells were lysed in 1× radioimmunoprecipitation assay buffer containing one tablet of cOmplete protease inhibitor cocktail (Roche Life Science, Indianapolis, IN, USA). Cell debris was pelleted by centrifugation at 15,000 g for 10 minutes at 4°C. The supernatants were collected and quantified using the BCA assay. In each case, 25 μg of protein was loaded onto 15% (v/v) SDS-PAGE gels, separated and transferred to polyvinylidene fluoride membrane. Membranes were incubated with pS2, cathepsin D and CDC2 antibodies or control α-actin antibody. Bound antibodies were detected using horseradish peroxidase–conjugated secondary antibodies. Chemiluminescence was detected with an Amersham ECL detection kit (GE Healthcare Life Sciences), and bands were visualised using the G:BOX imager (Syngene, Frederick, MD, USA).
Data were analysed, and dose–response curves generated, using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA, USA). A P-value <0.05 was considered significant.
In a series of previous studies, we successfully targeted coactivator binding pockets AF2 and binding function 3 (BF3) on the human androgen receptor (AR) and developed several potent AR inhibitors [38-40]. These compounds demonstrated significant antiproliferative effects on a variety of prostate cancer cell lines, including those that are resistant to conventional antiandrogens. Furthermore, several structures of the inhibitors within the intended AR AF2 and BF3 target sites were resolved by X-ray crystallography and deposited in the Protein Data Bank [PDB:4HLW, PDB:2YLO, PDB:2YHD, PDB:3ZQT, PDB:2YLP, PDB:2YLQ]. Because ERα belongs to the same nuclear receptor superfamily, and shares the same domain structural organisation, as the AR, we adopted a similar approach to exploit the ERα AF2 target to develop specific ERα inhibitors. X-ray crystallography studies at the AF2 site have revealed that, despite high sequence homology between the AF2 sites of nuclear receptors, these sites possess different surface shapes and electrostatic characteristics, which may be exploited to achieve selective target binding . These differences allowed us to design inhibitors specific to the ERα AF2 site.
In silico virtual screening for potential activation function 2 binders
The AF2 site represents a hydrophobic groove on the ERα surface flanked by charged amino acids that are essential for the binding of coactivators (Figure 1A). As it is a protein–protein interaction site, AF2 is a challenging target; nevertheless, it offers an attractive option for direct inhibition of ERα activation. Using our in-house computational drug discovery pipeline, we conducted a virtual screen of more than 4 million purchasable compounds from the ZINC database  to identify a list of potential AF2 binders. The in silico pipeline included large-scale docking, in-site rescoring and consensus voting procedures (for details, see Figure 1B).
We began by collecting and adjusting all chemical structures to their proper protonation state, checking for errors and then docking them into the ERα AF2 pocket using the Glide SP program with default settings . For this purpose, we used the 3UUD crystal structure of the ERα with 1.6 Å resolution . The resulting set of 2 million compounds that received a dock score ≤5.0 were redocked into the 3UUD structure using another docking protocol—eHiTS —with the corresponding docking score threshold set to 3.0. This step helped to reduce a large set of candidate AF2 binders to 5 × 105.
Next, to identify the most consistently predicted binding orientations of the compounds, RMSD was calculated between the docking poses generated by the Glide SP and eHiTS protocols. Only molecules with docking poses with RMSD values <2.0 Å were selected for further analysis. Furthermore, the selected ligands were subjected to additional on-site scoring using the LigX program and the pKi prediction module of the MOE . With this information, a cumulative scoring vote of five different parameters (Glide SP score, eHiTS score, LigX score and pKi predicted by the MOE) was generated with each molecule, receiving a binary 1.0 score for every ‘top 10% appearance’. The final cumulative vote was then used to select 5,000 compounds that consistently demonstrated high predicted binding affinity toward the targeted AF2 site. These compounds were then visually inspected, and a list of the 100 most promising candidates was determined for purchasing and biological testing.
Effects of lead compounds on inhibition of transcriptional activity of oestrogen receptor α in breast cancer cell lines
Effect of active compounds on coactivator recruitment to the activation function 2 site
Binding of the identified small molecules to the AF2 site should inhibit E2-dependent coactivator peptide recruitment to this site. To test this hypothesis, the AF2 binders were analysed by using the LanthaScreen TR-FRET ER Alpha Coactivator Assay kit. Tb-labelled anti-GST antibody indirectly labels the ERα LBD by binding to a GST tag on the protein. Binding of the agonist E2 to the ERα causes conformational changes that result in an increase in the affinity of ERα for a fluorescently labelled coactivator peptide, fluorescein-PGC-1α. PGC-1α has been shown to interact with the AF2 site of ERα in an agonist-dependent manner . The close proximity of fluorescein-PGC-1α to the terbium-labelled antibody causes an increase in the TR-FRET signal. When a compound binds to the AF2 site, the recruitment of the coactivator peptide is blocked, causing a decrease in the TR-FRET signal, which is measured as a ratio of emission at 520 nm to 495 nm.
Of 15 chemicals tested, 9 demonstrated effective blocking of AF2–PGC-1α interaction in a concentration-dependent manner, and their IC50 values were determined to range between 2 and 20 μM (Additional file 1: Table S1 and Additional file 2: Table S2). This suggests that the compounds bind to the AF2 site, thereby inhibiting coactivator recruitment. The small molecules VPC-13002, VPC-16225 and VPC-16230, which were potent in cellular assays, demonstrated IC50 values of 2.46, 3.76 and 2.98 μM, respectively (Figure 2C). The cold PGC-1α peptide used as a control in this assay showed a dose-dependent decrease in the FRET signal with increases in concentration (Figure 2D).
Active compounds do not bind to the oestrogen receptor-alpha hormone-binding site
Active compounds show direct binding to the oestrogen receptor-alpha ligand-binding domain
To confirm that the identified inhibitors directly bind to the ERα LBD, we cloned and purified the ERα LBD by fusion with an AviTag at the N-terminus and a six-residue histidine tag at the C-terminus. The ERα LBD was biotinylated on the AviTag by a biotin ligase expressed by the bacterial cells cotransformed with the biotin ligase plasmid pBirACm. The bERα LBD was purified by nickel-affinity chromatography and immobilised on streptavidin-coated biosensor tips. The interaction between small molecule and protein was measured in real time as a shift in the interference pattern of the incident light. A response profile was generated on the Octet RED system.
Compounds VPC-16225 and VPC-16230 reduce the growth of MCF7 cells, including tamoxifen-resistant cells
Compounds VPC-16225 and VPC-16230 inhibit oestrogen receptor-alpha in tamoxifen-resistant cells
VPC-16230 inhibits expression of oestrogen receptor-alpha-driven genes in MCF7 cells
Because the AF2 pocket plays a pivotal role in mediating ERα function, targeting this site offers a rich opportunity for the discovery of new BCa drugs. Inhibiting ERα in this manner offers several advantages over conventional anti-oestrogens for the treatment of TamR and metastatic BCa. First, by targeting a site other than the HBS, AF2 inhibitors should be effective in hormone-resistant BCa, because mutations at the HBS should have no effect on the efficacy of AF2 inhibitors. Furthermore, because all conventional anti-oestrogens bind to the same site (that is, the HBS), they cannot be used therapeutically in combinations. By targeting a different site on the ERα, a given AF2 inhibitor could be taken concurrently with the current HBS-directed anti-oestrogen(s). This could potentially be very important for decreasing the time to cancer remission in patients with BCa. Similar to highly active antiretroviral therapy, by taking several complementary therapeutics simultaneously, the rate of drug resistance should be greatly reduced and will lead to an improvement in the overall survival rate of patients with advanced BCa.
Molecular docking-based analysis of the reported compounds
With the use of in silico modelling, we have identified 15 structures that offer two novel classes of molecular scaffolds capable of inhibiting ERα–coactivator interaction at the AF2 site. These in silico-identified AF2 binders belong to two distinctive types: derivatives of pyrazolidine-3,5-dione (Additional file 1: Table S1) and carbohydrazide (Additional file 2: Table S2).
It should be noted that VPC-16003 and VPC-16004 exhibited a tenfold decrement in peptide displacement and cell inhibition compared with VPC-13002. We can speculate that the replacement of indole- with methyl- or chlorobenzene resulted in losing critical hydrogen bond and hydrophobic interactions. This may account for its lowered peptide displacement activities (IC50 > 25 μM versus 2.46 μM for VPC-13002). Because of having the same indole moiety at R1 positions and similar hydrophobic groups at R2 positions, compounds VPC-16019, VPC-16041, VPC-16046 and VPC-16038 demonstrated peptide displacement activities in a comparable range (3.16 to 7.98 μM) with VPC-13002. Compound VPC-16040, with a fourfold benzene substitution, lost its peptide displacement activity fivefold because of the decrease of hydrophobicity at the substitution position. Among all of these compounds, VPC-16021 was the weakest peptide displacer (>100 μM), with poor cell inhibition due to the unavailability of an NH of indole moiety to make a critical hydrogen bond interaction with Gln375. Moreover, lacking hydrophobic groups on the benzene ring hampers its activity. Even though some derivatives of this chemical type demonstrated promising ER-inhibitory activity and effectively block ER coactivator recruitment, they failed to reduce the growth of MCF7 cells. Moreover, VPC-13002 demonstrated toxicity on ER-negative cell lines (Figure 7A). On the basis of the above-described considerations, this class of compounds was not considered in further analysis.
The analysis of the binding orientations of carbohydrazide derivatives, VPC-16230 and VPC-16225 (peptide displacement IC50 = 2.98 and 3.76 μM, respectively), inside the target illustrates possible formation of hydrogen bonds with the residues of AF2 ‘charge-clamp’ regions. They perfectly occupy the binding regions of the LXXLL motif and disrupt key interactions between the coactivator and AF2 residues (see Figure 10C for VPC-16230). According to our docking models, VPC-16230 is likely to be anchored by three hydrogen bonds with Lys362 and Gln375. In addition, 5-bromo-3-ethoxy-2-hydroxyphenyl and -benzodioxene moieties of VPC-16230 form strong hydrophobic contacts with Val368, Ile358, Leu372 and V376 (Figure 10D). Notably, bulky substitution on the phenyl ring of VPC-16236 did not allow the compound to form tight binding. Hence, it retained lower peptide displacement activity. It should be noted that the employed peptide displacement assay directly quantifies interactions between the chemicals and the ERα AF2 site and thus provides firsthand evidence of the ER AF2 binding.
Experimental analysis of the reported compounds
VPC-13002, VPC-16225 and VPC-16230 appeared to be very potent inhibitors of ERα transcriptional activation in cells and AF2 coactivator binding in vitro, demonstrating explicit dose–response behaviour (Figure 2A, C). These compounds show a rightward shift in their dose–response curves upon increasing the concentration of the coactivator peptide (Figure 3A–C). This confirms that the compounds are in fact binding at the AF2 site. Further investigation of these molecules using BLI binding assessment, E2 displacement assay and qRT-PCR experiments allowed us to characterise VPC-16230 as the most promising ERα AF2 binder.
In luciferase assays, the IC50 of VPC-16230 was established as 5.8 μM (in the presence of 1 nM E2). If we assume that this activity is due to the binding to the HBS part of the receptor, then the binding affinity of the compound is 1/5,800th that of E2. This means that in direct ligand competition assays with Fl-E2, the compound should be able to compete with the fluorescently labelled E2 and its binding affinity should be close to 1/5,800th that of E2. However, when tested in the presence of 4.5 nM Fl-E2, the compound failed to displace the fluorescently labelled ligand, even at concentrations as high as 150 μM (binding affinity approximately 1/30,000th that of E2). The EC50 of unlabelled E2 for displacement of Fl-E2 was established at 4.2 nM in this assay. Moreover, if VPC-16230 bound to the HBS, we would observe a shift in its IC50 values in the presence of higher concentrations of E2. This was not the case in T47D-KBluc cells, where the inhibition by the compound was tested in the luciferase reporter assay in the presence of 1, 10, 50 and 100 nM E2. The dose–response curve of the compound did not shift significantly with the IC50, varying only from 6.1 μM to 6.5 μM with a 100-fold increase in E2 concentration. However, 4-OHT presented a rightward shift of IC50 values from 0.004 to 0.5 μM, in proportion to a 100-fold increase in E2 concentration. Taken together, these observations confirm that the mechanism of action of VPC-16230 is not related to HBS binding and is likely to be through AF2 interaction.
Tamoxifen resistance remains a fundamental cause of therapeutic failure in BCa therapy. This creates a challenge for researchers in the field of BCa drug discovery. Studies have shown that the majority of resistant tumours retain dependence on ERα and E2 for survival. In this scenario, AF2-directed compounds could prove critical in suppressing tumour growth. To test this hypothesis, we began by evaluating our compounds in two MCF7-derived TamR cell lines: TamR3 and TamR6. The results of the E2 dose–response experiments with a reporter assay (see Additional file 4: Figure S2), and the effect of fulvestrant on these cells (Figure 8) confirmed that these cell lines retain a functionally active ER. The E2 responsiveness, even in the presence of tamoxifen, suggests that ligand-dependent coactivator recruitment at the AF2 site remains functional in TamR3 and TamR6 cells. Because our lead compounds VPC-16225 and VPC-16230 showed promising inhibition of ER transcriptional activity in an AF2-specific manner, we anticipated that these compounds would also be able to reduce the growth of these resistant cells. When tested in MTS assays, VPC-16225 and VPC-16230 demonstrated promising antiproliferative effects in MCF7, TamR3 and TamR6 cells and had no effect on ER-independent MDA-MB-453 and HeLa cells. This is the first evidence of AF2-directed molecules with an inhibitory effect on TamR cells. Our compounds significantly inhibited the expression of a transiently transfected E2-responsive luciferase reporter in both TamR3 and TamR6 cells, further confirming that the reduction in the growth of TamR3 and TamR6 is via ER inhibition through AF2-directed action.
A comparison of the binding modes of AF2 ligands and tamoxifen (Additional file 5: Figure S3) revealed distinct differences in the structural conformation of ER. Tamoxifen allosterically affects coactivator recruitment by competitively inhibiting association between E2 and ER HBS. This causes conformational changes in the receptor that eventually lead to movement of helix 12 in such a way that it prevents the formation of the AF2 site. Because E2 continues to activate ERα, the AF2 site is still accessible and the compounds can bind the pocket and inhibit the receptor in TamR3 and TamR6 cells. This provides a possible explanation why AF2 inhibitors are effective in these cells, as observed in our assays.
The downstream effect of ER inhibition should be reflected as downregulation of ER target genes. VPC-16230 could significantly decrease E2 stimulated mRNA expression levels of ERα target genes such as pS2, cathepsin D and CDC2 in MCF7 cells. However, when treated in the absence of E2, the compound did not show any significant effect on the expression levels of these genes. This was expected because E2 is required for the AF2 pocket to be formed and because, in its absence, the compounds will not bind to the AF2 site and therefore will not show any effect. It is well established that E2 stimulates the proliferation of MCF7 cells, and it was observed that CDC2, a cell cycle–associated gene whose expression was increased threefold upon E2 stimulation, was significantly reduced by VPC-16230. This observation correlates with the inhibition of proliferation by VPC-16230 of MCF7 cells in MTS assays. Because these genes are ERα-dependent, their inhibition corroborates the idea that blocking coactivator binding is a promising approach to inhibit ER activity in BCa cells.
Overall, our results indicate that targeting the AF2 site with small molecules represents a feasible therapeutic approach that allows successful inhibition of growth of TamR BCa cells. Although the potency of our lead AF2 inhibitor, VPC-16230, is less than that of tamoxifen, one must note that this is an early-stage compound and that its activity can be enhanced by further lead optimisation efforts.
In BCa, progression to a hormone-resistant phenotype in the presence of an ERα antagonist is possibly maintained through selection of cells with epigenetic or mutational changes that bypass the inhibitory action of the drug [48,51,52]. Using a combination of virtual screening and biochemical assays, we have identified several compounds that bind to the AF2 site of human ERα and directly inhibit its interaction with coactivators. The present study is the first successful attempt to identify direct disruptors of ERα transactivation using an in silico virtual screening approach, and this resulted in the identification of 15 AF2-selective low micromolar inhibitors. On the basis of these results, more potent ERα inhibitors should be investigated, with the goal of developing novel therapeutic strategies that can act as complementary approaches to treat ERα-positive, TamR and metastatic BCa.
This work was supported by funding from Canadian Cancer Society Research Institute grant F12-03271. The authors thank Mohamed Sheriffdeen for helping with in vitro experiments. The authors thank Dr D McDonnell (Duke University, Durham, NC, USA) for the 3X ERE TATA luc plasmid. The tamoxifen-resistant MCF7 sublines TamR3 and TamR6 were obtained from the cell line collection maintained by the Auckland Cancer Society Research Centre, Faculty of Medical and Health Sciences, University of Auckland, New Zealand.
- 8.Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet. 2005;365:1687–717.Google Scholar
- 21.Leary AF, Drury S, Detre S, Pancholi S, Lykkesfeldt AE, Martin LA, et al. Lapatinib restores hormone sensitivity with differential effects on estrogen receptor signaling in cell models of human epidermal growth factor receptor 2-negative breast cancer with acquired endocrine resistance. Clin Cancer Res. 2010;16:1486–97.CrossRefPubMedGoogle Scholar
- 40.Munuganti RSN, Leblanc E, Axerio-Cilies P, Labriere C, Frewin K, Singh K, et al. Targeting the binding function 3 (BF3) site of the androgen receptor through virtual screening. 2. Development of 2-((2-phenoxyethyl)thio)-1H-benzimidazole derivatives. J Med Chem. 2013;56:1136–48.CrossRefPubMedGoogle Scholar
- 41.Delfosse V, Grimaldi M, Pons JL, Boulahtouf A, le Maire A, Cavailles V, et al. Structural and mechanistic insights into bisphenols action provide guidelines for risk assessment and discovery of bisphenol A substitutes. Proc Natl Acad Sci U S A. 2012;109:14930–5.CrossRefPubMedPubMedCentralGoogle Scholar
- 42.Schrödinger. Maestro software. Schrödinger, New York. 2008. http://www.schrodinger.com/Maestro/. Accessed 6 Mar 2015.
- 44.Chemical Computing Group. MOE: Molecular Operating Environment. http://www.chemcomp.com/. Accessed 4 Mar 2015.
- 48.Tcherepanova I, Puigserver P, Norris JD, Spiegelman BM, McDonnell DP. Modulation of estrogen receptor-α transcriptional activity by the coactivator PGC-1. J Biol Chem. 2000;275:16302–8.Google Scholar
- 49.Frasor J, Danes JM, Komm B, Chang KCN, Lyttle CR, Katzenellenbogen BS. Profiling of estrogen up- and down-regulated gene expression in human breast cancer cells: insights into gene networks and pathways underlying estrogenic control of proliferation and cell phenotype. Endocrinology. 2003;144:4562–74.CrossRefPubMedGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.