MECHANISM OF GENOMIC CROSSTALK BETWEEN ANDROGEN AND GLUCOCORTICOID RECEPTORS
IN PROSTATE CANCER CELLS
Hanna Laakso Master’s Thesis
University of Eastern Finland Institute of Biomedicine, School of Medicine, Faculty of Health Sciences 12.6.2020
UNIVERSITY OF EASTERN FINLAND, Faculty of Health Sciences School of Medicine, Master´s Programme in Biomedicine
Hanna Laakso: Mechanism of Genomic Crosstalk Between Androgen and Glucocorticoid Receptors in Prostate Cancer Cells
Master’s thesis: 32 pages, 2 pages Supplementary Data
Supervisors: Ville Paakinaho, PhD & Jorma Palvimo, Professor 12.6.2020
Keywords: glucocorticoid receptor, androgen receptor, prostate cancer, chromatin accessibility, assisted loading, FOXA1, BRD4
ABSTRACT
Chromatin represents a barrier for transcription factors in the regulation of gene expression.
For optimal gene regulation transcription factors either bind to pre-accessible chromatin sites or they are required to penetrate closed chromatin sites. Previous models of transcription factor action have indicated that steroid receptor binding to chromatin in breast and prostate cancer (PC) requires the activity of pioneer factors. However, according to the latest knowledge, steroid receptors can also function as pioneer factors, making chromatin accessible to other DNA binding proteins. Preliminary results in the research group indicate, that androgen receptor (AR) activation increases chromatin accessibility and glucocorticoid receptor (GR) binding in PC cells. The mechanism behind this effect is unclear. In this thesis we investigated the role of two cofactors, BRD4 and FOXA1, in the crosstalk between AR and GR with genome-wide methods ChIP- and ATAC-seq in a PC cell model. By inhibiting BRD4 or FOXA1 activity, we affirmed BRD4 to be a cofactor in AR-mediated binding of GR to chromatin and found silencing of FOXA1 to drastically reduce GR chromatin occupancy.
Deeper understanding of these mechanisms is important as steroid receptors influence PC development, progression, and drug resistance. The role of AR has been well studied, however currently there is little understanding on GR activity in PC. The results presented in this thesis improve our knowledge of the intricate relationship between steroid receptors and cofactors in PC cells.
3
INTRODUCTION
Prostate cancer (PC) is the most common diagnosed cancer in men, and sixth leading cause of cancer death, and despite improved prognosis in many countries, it still remains a topical issue in the aging population (Siegel, et al. 2019). Normally androgens are needed for the development of primary and secondary sex characteristics of males, such as the development and growth of the prostate. In PC, the development, progression, and drug resistance are driven by the dysregulation of androgen receptor (AR) signaling (Basu & Tindall, 2010). First forms of treatment are prostatectomy and androgen deprivation therapy (ADT), which includes chemical or surgical castration in order to reduce serum androgen levels, as well as administration of antiandrogens to block AR signaling. This combination of treatments is initially effective in decreasing tumor volume; however, more than 30 % of patients relapse (Shipley, et al. 2017) and progress into castration resistant prostate cancer (CRPC), called so due to non-responsiveness to ADT (Handle, et al. 2019). Despite the overall high survival rate in PC (98%), patients with CRPC have an average survival of only three years (Roviello, et al.
2016; Siegel, et al. 2019).
There are many suspected reasons that contribute to development of CRPC, one being selective pressure caused by inhibition of the AR signaling axis. Despite the non-responsiveness to ADT, CRPC progression is still driven by the activation of AR target gene expression (Basu &
Tindall, 2010). This continued expression is maintained despite castration levels of serum androgens via adaptations in the AR gene (amplification, overactivity, splice variants), AR bypass by other transcription factors (TFs), or other crosstalk mechanisms. Different mechanisms of ADT resistance in PC have been reviewed recently in more detail by Tucci, et al. (2018). One mechanism of AR bypass in CRPC is by the glucocorticoid receptor (GR): GR can take over many AR’s target genes due to selective pressure caused by antiandrogen treatment, such as enzalutamide (Semenas, et al. 2013), and blunt the effects of ADT (Arora, et al. 2013; Isikbay, et al. 2014). Arora, et al. (2013) were the first to suggest bypass of AR blockade through upregulation of GR in CRPC as a mechanism of drug resistance. They induced enzalutamide resistance in PC cells and showed that GR antagonism was enough to regain therapeutic effect. This has been proven to be the case in other studies as well (Isikbay, et al. 2014; Puhr, et al. 2018).
The steroid receptor (SR) family includes AR, GR, the estrogen receptor (ER), the progesterone receptor (PR) and the mineralocorticoid receptor (Shiota, et al. 2019). SRs are a highly
4 conserved subfamily of nuclear receptors, that control a variety of biological functions in the body, such as metabolism, inflammation and development by regulating gene expression (Truong & Lange, 2018). They function as hormone-dependent TFs which upon ligand activation bind to response elements and regulate gene transcription. Arora, et al. (2013) demonstrated that GR is able to drive expression of certain AR target genes independent of AR, for instance, the PSA (KLK3) gene that is used to evaluate response to ADT. The GR- mediated regulation of AR target genes could be explained by their structural similarity, but also by the composition of their regulatory proteins, cofactors. It is known that AR and GR share overlapping target sites (Sahu, et al. 2013), but in addition they also share several cofactors (Arora, et al. 2013; Lempiäinen, et al. 2017). Cofactors include proteins like coregulators, histone modifiers, chromatin remodelers and other TFs, that help regulate the context-specific transcription of genes. The goal of this project is to investigate the crosstalk between AR and GR, and the different TFs and coregulators in these processes. The results from this project could increase our knowledge of the deeply intertwined function of AR and GR, and more importantly, contribute to the improvement of PC treatment.
For optimal gene regulation TFs either bind to pre-accessible chromatin sites or they are required to penetrate closed chromatin sites. For SRs, the binding of the cognate ligand to its receptor in the cytosol causes a conformation change and release from chaperone proteins, relocalization into the nucleus and formation of oligomers (Presman, et al. 2016). These oligomers bind to specific DNA elements called response elements and either induce or repress transcription of target genes via the recruitment of coregulators with histone- and chromatin modifying activities (Lempiäinen, et al. 2017). Previous models of TF action have indicated that SR binding to closed chromatin requires the activity of pioneer factors, such as the Forkhead Box A 1(FOXA1) (Jozwik & Carroll 2012; Pihlajamaa, et al. 2015). As a pioneer factor, it opens chromatin for other TFs independent of ATP-dependent chromatin remodeling complexes (Zaret & Carroll, 2011). According to the latest knowledge, SRs also possess pioneer-like function, making chromatin accessible for other DNA binding proteins (Paakinaho, et al. 2019; Swinstead, et al. 2016a) via a mechanism termed dynamic assisted loading (Swinstead et al. 2018).
In dynamic assisted loading, an initiating factor first binds to a closed chromatin region and recruits ATP-dependent chromatin remodelers. Followed by the decompaction of chromatin, a secondary factor binds to the now accessible region (Fig. 1). This often works in both ways for the factors involved; the secondary factor can initiate the binding of the other factor on other
5 regions on the chromatin (Swinstead, et al. 2018). Furthermore, even though the initiating and secondary factors bind to the same sequence on chromatin, they do not compete for binding (Voss, et al. 2011). The assisted loading differs from the classic pioneer model of action, as the factors involved do not have chromatin remodeling capability of their own, unlike pioneer factors, but instead they recruit proteins that do (Swinstead, et al. 2016a). For instance, in breast cancer (BC) under progesterone and estrogen dual treatment, Mohammed, et al. (2015) reported reprogramming of ER binding sites by PR-mediated chromatin remodeling via EP300 and FOXA1. Similarly, Miranda, et al. 2013 found that GR binding causes a certain subset of ER response elements to become accessible via recruitment of chromatin remodeling proteins, like the SWI/SNF complex. Our main hypothesis is that possibly like the SR crosstalk in BC, in PC AR and GR are dependent on each other for a subset of binding sites, where AR facilitates the binding of GR. The mechanism by which AR assists the binding of GR is not known.
However, it is known that activated AR increases the binding of GR by altering chromatin accessibility. To perform this, AR is required to bind to closed chromatin sites and increase accessibility potentially through the recruitment of cofactors and other TFs.
Figure 1. Assisted loading model. (1) Initiator Protein A (yellow circle) binds to response element (RE) located on nucleosomal chromatin, where protein B (purple circle) is unable to bind. (2) Followed by the binding of Protein A, a chromatin remodeling complex is recruited and (3) chromatin is locally decompacted. (4) Protein B is able to bind to accessible binding site.
FOXA1 is a known pioneer factor for multiple SRs (Jozwik & Carroll 2012; Pihlajamaa, et al.
2015) most well-known for amplification of ER signaling in BC (Hurtado, et al. 2011). In hormonal cancers, FOXA1 is connected to the deregulation of ER and AR (Robinson &
Carroll, 2012; Yang & Yu, 2015) and facilitation of chromatin accessibility as a part of chromatin relaxation (Robinson & Carroll, 2012). The loss of chromatin structure, known as chromatin relaxation (Corces, et al. 2018; Flavahan, et al. 2017; Urbanucci, et al. 2017), is seen
6 as a hallmark of cancer. The genomic distribution of FOXA1 has been shown to be regulated by GR and ER in BC (Swinstead, et al. 2016a), and to colocalize with active AR on chromatin (Belikov, et al. 2013), suggesting that not only does it pioneer the binding of other factors, but that it also plays a role in the crosstalk of SRs.
The FOXA1 DNA-binding domain (DBD), called the winged helix due to its two wing-like structures, resembles the linker histones H1 and H5. One method of FOXA1 mediated decompaction of chromatin is by displacing the linker histones from chromatin, enabling the binding of TFs to nucleosomal sites. The functionality of FOXA1 DBD also shares resemblance to those of other TFs, allowing FOXA1 to bind to and regulate a variety of chromatin sites (Zaret & Carroll, 2011). In ER expressing BC cells, ER and FOXA1 co-occupy on over half of sites (Lupien, et al. 2008), and silencing of FOXA1 results in loss of ER binding and ER-mediated proliferation (Hurtado, et al. 2011). The relationship of FOXA1 and AR is more complex: depending on the cellular context, FOXA1 can either promote or repress AR target gene expression. AR and FOXA1 have been shown to interact through their DBDs upon liganded AR activation, thus helping in the regulation of AR-target gene expression (Gao, et al. 2003) at cell-specific sites (Lupien, et al. 2008).
FOXA1 mutations have been connected to PC progression and a less favorable prognosis (Adams, et al. 2019; Parolia, et al. 2019). As FOXA1 has been shown to be a factor in CRPC (Robinson, et al. 2014), studies on AR and FOXA1 occupancy as well as FOXA1 depletion in PC cells, such as VCaP cells, have been done before. For instance, Sahu, et al. (2011) identified three differently FOXA1-responsive AR binding sites. These sites can be divided into those independent of FOXA1, those pioneered by FOXA1, and those that are masked by FOXA1 and functional upon FOXA1 depletion. In another study they reported that FOXA1 depletion also lead to a shift in AR- as well as GR-binding sites (Sahu, et al. 2013). Moreover, out of the overlapping AR-GR binding sites, most have FOXA1 binding motifs (Arora, et al. 2013).
Taken together, this could suggest that FOXA1 has an important role in regulating chromatin accessibility and consequently the binding of TFs at these sites.
Another way to either allow or restrict binding of proteins to regulatory sites is regulation of chromatin accessibility by post-translational modifications (PTMs) of the nucleosomal histones. These PTMs include acetylation, phosphorylation, methylation or ubiquitination of certain residues histones (Bannister & Kouzarides, 2011). By changing the positive charge of histones, PTMs like acetylation weaken the DNA-histone and histone-histone interactions.
7 This allows the movement of histones or complete eviction from a nucleosome. The nucleosome instability followed by histone eviction causes chromatin decompaction (Devaiah, et al. 2016).
PTMs are catalyzed by so-called writer or eraser proteins, such as histone acetyltransferases and deacetylases which remodel chromatin by adding or removing acetyl groups from histone lysine residues, respectively. PTMs can either directly affect the interactions of DNA and chromatin proteins or they can function as markers for reader proteins that recognize specific PTMs and serve as adaptors to recruit factors like chromatin remodelers or TFs (Taniguchi, 2016; Yun, et al. 2011). For example, the Bromodomain-containing protein 4 (BRD4) recognizes and binds to acetylated lysine residues on histones H3 and H4, and recruits chromatin remodelers and transcription initiation factors (Taniguchi, 2016). Furthermore, histone acetyl transferase activity as well as histone eviction by BRD4 has also been noted (Devaiah, et al. 2016). This indicates that BRD4 does not only recognize PTMs but actively participates in the remodeling of chromatin. Interestingly, the action of PTMs and ATP- dependent chromatin remodelers, such as SWI/SNF complexes (Clapier et al. 2017; Swinstead et al. 2016b), in the regulation of chromatin accessibility seem to be intertwined (Chatterjee et al. 2011; Local et al. 2018).
Based on structural similarities, bromodomain-containing proteins have been divided into subfamilies. The BRD4 belongs to the Extra-Terminal Domain (BET) family together with BRD2, BRD3, and BRDT. For a more in-depth review on the BET family protein structures, see the recent article by Zaware and Zhou (2019). The BRD4 interacts with multiple DNA binding proteins, such as the positive transcription elongation factor b, which in turn facilitates RNA Polymerase II dependent transcription. Due to the ubiquitous expression of BRD4 and its role in proliferation, BET inhibitors have been researched as potential treatment for multiple diseases and cancers (Padmanabhan, et al. 2016, Perez-Salvia & Estelle, 2017). This has led to development of BET-targeting drugs, such as JQ1 and I-BETs (Filippakopoulos, et al. 2010).
Both JQ1 and I-BET762 (Mirguet, et al. 2013) competitively inhibit the histone binding of BET proteins. Inhibition of BET proteins has been found to block cell cycle progression and transcription (Filippakopoulos, et al. 2010, Tan, et al. 2018).
The BRD4 and AR chromatin co-occupancy in prostate cancer (VCaP) cells has been studied before by Asangani, et al. (2014). In their study, it was shown that AR and BRD4 interact directly on chromatin via the BRD4s BD1/2 domains, and that BET protein inhibition with JQ1
8 lead to disruption of the AR-BRD4 interaction. In a recent study Lambert, et al. (2019) showed that JQ1 treatment lead to rapid changes in interactomes of each BET protein, while not all interactions were affected, possibly because they are mediated in mechanisms not related to BD1/BD2 activity. By interacting with AR, BRD4 facilitates transcription and increased chromatin accessibility (Urbanucci, et al. 2017). Consequently, the overexpression of BRD4 in advanced PC enables the chromatin binding of AR in low androgen environment such as created with ADT. Urbanucci, et al. (2017) reported that AR overexpressing CRPC cells, such as VCaPs, were most sensitive to BRD4 inhibition, which suggests dependency in AR driven transcription and chromatin accessibility. Furthermore, Asangani, et al. (2014) reported that BRD4 inhibition affected AR activity more than direct AR antagonism in CRPC, suggesting it to be a powerful drug target in CRPC.
The role of the AR has so far been the main focus as the driving force behind PC development and progression. However, an interest in other proteins related to the regulation of AR activity and target gene expression in PC has risen due to the challenges of acquired drug resistance in PC therapies. When taken into consideration that GR agonists, such as dexamethasone, are administered to cancer patients as part of treatment due to their anti-inflammatory properties, the comprehensive understanding of GR activation in CRPC is still lacking (Handle, et al.
2019). The current understanding of AR and GR action in PC is mainly based on research in enzalutamide treated cells, and little information on GR interactions in antiandrogen naïvePC cells is currently available. The individual action of key receptors like ER and AR in hormonal cancers has been well studied, but the need for better understanding of SR interactions and crosstalk is evident due to high rates of acquired drug resistance.
This project is based on research done by Ville Paakinaho on the pioneer activity of AR and GR in PC and will expand the research done in the research group. As known cofactors of AR, FOXA1 and BRD4 mediate the transcriptional activity of AR by chromatin remodeling and recruiting other transcriptional regulators. Both interact with AR and so have been studied as druggable targets in PC, as their expression has been associated with poor prognosis (Asangani, et al. 2014, Urbanucci, et al. 2017, Sahu, et al. 2011). Based on this notion, inhibition studies on FOXA1 and BRD4 were completed to study the recruitment of these cofactors in the binding of GR to its response elements in a CRPC cell line model. To achieve this, we used chromatin immunoprecipitation and assay for transposon-accessible chromatin followed by deep sequencing (ChIP-seq, ATAC-seq). First, we determined the genome-wide binding of BRD4
9 and FOXA1 following hormone induction, and subsequently thechanges in the occupancy of AR and GR caused by inhibition or depletion of either protein.
MATERIALS AND METHODS
Cell culture
Endogenous AR-expressing prostate cancer cell line VCaP was obtained from American Type Culture Collection (ATCC, #CRL-2876) and grown in Dulbecco’s Modified Eagle Medium (DMEM, Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS) and 1%
penicillin/streptomycin (P/S). Cells were seeded 1:3 every four days on 10-cm plates in upkeep medium (37 °C, 5% CO2). Cells were detached with 0.25% trypsin containing 0.5 M EDTA.
In experiments, stripped medium (DMEM, 2.5% FBS, 1% P/S) or transfection medium (DMEM, 2.5% FBS) were used.
Antibodies
Antibodies used for ChIP-seq were rabbit polyclonal anti-FOXA1 (Abcam, cat# ab170933), GR (Santa Cruz Biotechnology, cat# sc-1003), rabbit polyclonal AR (K183, Karvonen, et al.
1997) and BRD4 (Bethyl Laboratories, cat# A301-985A). Same anti-FOXA1 antibody as well as GAPDH (Santa Cruz Biotechnology, cat# sc-25778) were used for immunoblotting, together with appropriate secondary goat anti-rabbit antibody (Invitrogen, #31460).
Chemicals
Dexamethasone (Dex) and dihydrotestosterone (DHT) were purchased from Sigma Aldrich (cat# D1756, cat# A8380), as was the BRD4 inhibitor I-BET762 (cat# SML1272).
Silencing of FOXA1 by RNA interference
For reverse silencing of FOXA1, SMARTpool: ON-TARGETplus FOXA1 siRNA and ON- TARGETplus Non-targeting Pool (Horizon Discovery, cat# D-001810-10-50) were used at 60 nM final concentration with Lipofectamine RNAiMAX transfection reagent (Life Technologies, cat# 13778030) according to manufacturer’s instructions. For reverse- transfection, siRNA and OPTIMEM were added first together with RNAiMAX, gently mixed by swirling and incubated for 10-15 minutes RT. Cells suspension in transfection medium was added to the plates, 5.5 x 106 cells per 10-cm plate. For both ChIP-seq and ATAC-seq, cells were incubated for 72 h before medium change to stripped medium for another 72 h, after which the cells were exposed to 100 nM steroid (Dex, DHT, Dex+DHT) or vehicle (ethanol) for 1 h.
10 Immunoblotting
Immunoblotting was used to verify silencing of FOXA1 after transfection. Immunoblotting was done according to standard protocol from whole cell protein extracts (Rytinki, et. al. 2011).
Cells were seeded on a 6-well plate, 0.5 x 106 cells per well, and reverse silenced as described previously. Cells were incubated for 72 h before harvesting. Cells were washed with TBS supplemented with protease inhibitor cocktail (PIC) (Sigma-Aldrich, cat# 11836145001) and replicates scraped into one 1.5 ml Eppendorf. After centrifuging, the pellet was suspended into SDS buffer supplemented with PIC. Samples were heated for 5 min at 95 °C and sonicated for 2 x 10 s on ice. After adding 2 % beta-mercaptoethanol, samples were heated again as before.
A gel with 10% running gel and 4% top gel was prepared and sample or 10-250 kDa PageRuler Plus Protein Ladder (Thermo Fischer, cat# 26620) pipeted into wells. Proteins were blotted onto a 0.45 µm nitrocellulose membrane (thermo Fischer, cat# 88018) at 250 mA for 1 h, and immunoblotted over night with primary antibodies for GAPDH or FOXA1.
Corresponding secondary antibody was used. Antibodies were diluted 1:5000 for GAPDH, 1:1000 for FOXA1 in 5 % fat-free milk in wash buffer (1x TBS, 0.1% Tween) and 1:10000 dilution in wash buffer for the secondary antibody. Pierce ECL Western Blotting Substrate (Thermo Scientific) and ChemiDoc imaging system (Bio-Rad) were used for imaging the membranes.
Chromatin Immunoprecipitation (ChIP)
ChIP-seq was used to study changes in DNA binding of AR, GR and FOXA1 in steroid treated (Dex, DHT, Dex+DHT) cells before and after treatment with specific inhibitor of BRD4 or silencing of FOXA1. ChIP was done as described previously (Paakinaho, et al. 2014).
Briefly, VCaP cells were seeded 5 x 106 cells per 10-cm plate in upkeep medium and grown for 72 h. Medium was changed to stripped medium 48 h before ChIP. Cells were exposed to 100 nM Dex, 100 nM DHT or both, or an ethanol vehicle for 1 h, and immunoprecipitated with ChIP-grade antibodies for AR, GR, FOXA1 or BRD4. For silencing of BRD4 or FOXA1, cells were treated either with 1 µM bromodomain inhibitor I-BET762 1 h prior to hormone treatment, or cells were reverse transfected to silence FOXA1 as describe earlier. A DMSO control was used for inhibitor treatment, and a non-target siRNA for transfection cell samples.
Cells were fixed in 1% formaldehyde for 10 min at room temperature and washed twice with ice-cold PBS. The cell suspension was centrifuged, and the pellet resuspended RIPA (1% NP- 40, 0.5% Sodium deoxyxholate, 0.1% SDS in 1x PBS, containing 1% PIC). Chromatin was
11 sonicated to approximately fragment sizes 200-500 bp with Bioruptor 300 (Diagenode), after which the samples were pelleted at 4 °C and supernatant saved.
VCaP cells treated with steroids (Dex, Dex+DHT) were immunoprecipitated with anti- FOXA1, anti-BRD4, anti-AR, and anti-GR to establish binding patterns in the different treatments. The BRD4 or FOXA1 silenced samples were immunoprecipitated with antibodies for AR and GR. For all antibodies, 2 µg of antibody was used per sample, except 3 µl for AR (unknown concentration of antibody) coupled overnight to Dynabeads (Invitrogen, cat# 10004D). Two technical replicates per treatment group with approximately 10 ng of DNA per sample.
ChIP-qPCR analysis
DNA fragments from immunoprecipitation were amplified with primers for known GR- and AR binding loci at FKBP5 (-3), PSA enhancer as well as control RHO with LightCycler® 480 (Roche) and LightCycler® 480 SYBR Green I Master (Roche, cat# 04707516001). The qPCR data was analyzed for relative gene expression using the 2^-deltaCT fold enrichment method and normalized to respective control. Primer sequences are shown in Supplementary Table 1.
Assay for Transposon-accessible Chromatin with sequencing (ATAC-seq)
ATAC-seq was performed to study the general changes in chromatin accessibility after inhibition of BRD4 or depletion of FOXA1. ATAC-seq protocol has been modified and integrated from multiple publications (Buenrosto, et al. 2015; John, et al. 2011; Swinstead, et al. 2016; Paakinaho, et al. 2019).
Briefly, cells were seeded 5.5 x 106 cells per 10-cm plate either in upkeep medium (BRD4 inhibition) or transfection medium (RNAi) for 72 h after which medium was changed into stripped medium for 48 h. Cells were treated with steroids (and inhibitor) as in ChIP. Cells were washed with room temperature PBS and detached with 2 ml per plate of warm Accutase (Sigma Aldrich, cat# A6964) supplemented with steroids (Dex, DHT) and inhibitor for BRD4 silenced cells. After neutralizing the Accutase with medium, cells were pelleted at 4 °C and washed in 5 ml of ice-cold PBS (+PIC) and centrifuged again. Per sample, 5 x 106 cells/ml were resuspended in Buffer A (15 mM NaCl, 60 mM KCl, 1 mM EDTA, 0.5 M EGTA, 0.5 mM Spermidine, 1x PIC) containing IGEPAL. Cells were incubated on ice for 10 mins, and centrifuged. Pellets were washed once with Buffer A without IGEPAL and ATAC-RSB Buffer (10 mM NaCl, 10 mM Tris-HCl, 3 mM MgCl2). Nuclei were counted using Trypan Blue (Thermo Fisher, cat# 15250061) staining and approximately 100,000 nuclei were taken into
12 the transposition reaction done with Nextera DNA Library Preparation Kit (Illumina, cat#
FC-121-1030). Samples were purified according to manufacturer’s instructions using the Monarch PCR Sample Purification Kit (NEB, cat# T1030). PCR amplification was done using the NEBNext High-Fidelity 2x PCR Master Mix (NEB, cat# M0541) in a PCR thermal cycler (Biometra TRIO). For calculating the optimal cycle number for library preparation (Buenrosto, et al. 2015), qPCR (Roche, LightCycler® 480) was used. Size selection to remove fragments
<150 bp or >800 bp was done with SPRIselect beads (Beckman Coulter, cat# B23317) according to manufacturer’s instructions. Size selection verification, library concentration and size were analyzed using Agilent Bioanalyzer High Sensitivity DNA Analysis kit (Agilent, cat# 5067-4626).
Library preparation and data analysis
ChIP-seq libraries were generated according to manufacturer’s protocol using NEBNext Ultra II DNA Library Prep Kit (E7645L, NEB).In general, two biological replicate samples were sequenced in the EMBL Genomics Core Facility (Heidelberg, Germany) using Illumina NextSeq 500; 75SE for ChIP-seq, 40PE for ATAC-seq.
ChIP-seq read filtering and aligning to hg38 genome was performed as previously described (Paakinaho, et al. 2014, Toropainen, et al. 2015). For ATAC-seq data, after filtering low quality reads as with ChIP-seq data, paired-end samples were aligned to hg38 genome using Bowtie2 (Langmead & Salzberg 2012). Alignment was performed with end-to-end sensitive mode allowing no mismatches. Downstream data analysis was performed using HOMER (Heinz, et al. 2010). Peaks in each dataset were called using findPeaks with style factor, FDR < 0.01, >
25 tags, > 4 fold over control sample and local background. EtOH sample was used as control sample for AR and GR ChIP-seq, and input sample from VCaP cells was used as control for BRD4 and FOXA1 ChIP-seq. DESeq2 (Love, et al. 2014) through getDiffrentialPeaksReplicates.pl was used to isolate differential binding peaks (FDR < 0.05, fold change > 2) between the single and dual hormone treatments (C1-C2). Aggregate plots and heatmaps were generated with 10 bp or 20 bp bins surrounding ±1 kb area around the center of the peak. All plots were normalized to 10 million mapped reads and further to local tag density, tags per bp per site. Box plots represented log2 tag counts. De novo motif searches were performed using findMotifsGenome.pl with the following parameters; 200 bp peak size window, strings with 2 mismatches, binomial distribution to score motif p-values, and 50 000 background regions. Genome browser tracks were generated using makeUCSCfile, converting the obtained bedGraph to TDF using igvtool and visualized with IGV 2.3
13 (http://software.broadinstitute.org/software/igv/home) (Robinson, et al. 2011). The following publicly available sequencing datasets were used: GSE55062 for BRD2, BRD3, and BRD4 ChIP-seq treated with DMSO or I-BET762 from VCaP cells (Asangani et al. 2014).
RESULTS
Androgen exposure induces the binding of FOXA1 and BRD4 to AR-dependent GR binding sites in VCaP cells
As indicated above, previous models of TF action have indicated that SR binding to closed chromatin requires the activity of pioneer factors, such as FOXA1. However, according to the latest knowledge, SRs also possess pioneer-like function. This capability enables SRs to induce chromatin accessibility for other DNA binding proteins via a mechanism termed assisted loading. As they share similar response elements, AR and GR can bind almost interchangeably to each other’s response elements, yet they do not compete for them. Rather, results from the research group show that AR binds prior to a subset of sites that GR alone could not bind to due to inaccessible chromatin (detailed below), possibly similar to other SR dynamics studied in BC, like the GR-induced binding of ER.
The differential GR binding peaks (FDR 0.5%, see Materials and Methods section for details) between single (5α-dihydrotestosterone [DHT] or dexamethasone [Dex]) and dual hormone (DHT and Dex) treatments in VCaP cells were divided into two clusters, cluster (C)1 and (C)2, based on the genome-wide binding of GR. The C1 (4938 sites) includes GR binding sites where GR can bind only when AR is activated and bound prior GR. The C2 (4922 sites) includes GR binding sites where GR can bind without AR activity, although binding of GR is enhanced with AR at these sites. In the presence of an agonist, AR can bind to both independently of GR (Fig. 2, Supplementary Fig. 1).
In the previous research done by Paakinaho, FOXA1 along with NR3C (GR, NR3C1; AR, NR3C4) binding motifs in VCaP cells were enriched at C1 and C2. This suggests a possible role in FOXA1 mediating GR binding at these sites together with AR (Fig. 2). To investigate the role of FOXA1, as well as BRD4, in GR binding to chromatin, we used genome-wide methods such as ChIP-seq and ATAC-seq to investigate TF/cofactor chromatin binding and chromatin accessibility, respectively.
14 Figure 2. Motif enrichment analysis. (A) Heatmaps represent ChIP-seq signals ±1 kb around the peak center for C1 (AR-dependent GR binding) and C2 (AR-independent GR binding) clusters. VCaP cells were immunoprecipitated for AR and GR in cells treated 1 h with either 100 nM Dex or 100 nM DHT, or both. (B) De novo motif analysis of AR-induced GR sites (C1) compared to GR independent sites (C2) in VCaP cells. Top 4 enriched motifs are shown: NR3C (GR/AR), FOXA1, HOXB13 and ERG. P-value and % of sites with motif are shown.
First, to assess the possible changes in occupancy induced by AR and GR activity, we determined the genomic binding of FOXA1 and BRD4 in cells treated with activating hormones of AR and GR (DHT and Dex) by ChIP. ChIP-qPCR was used to validate the immunoprecipitation samples (Fig. 3). For this purpose, primers for FKBP5(-3) and PSA enhancer were used representing clusters C1 and C2, respectively. Both regions are known to harbor binding sites for AR and GR. Primer for androgen and glucocorticoid independent gene RHO was included as a negative control region.
For BRD4, its binding to both loci is induced the most by AR agonist treatment (Fig. 3A), which was expected due to their known interactions in VCaP cells. Moreover, BRD4 occupancy is highest in FKBP5(-3), which represents C1, whereas FOXA1 occupies the C2 representing loci (PSA enhancer) more in all hormone treatments (Fig. 3B). This seems to complement the motif enrichment analysis, which shows higher occupancy of FOXA1 motifs at C2 sites, as well as the probability of BRD4 being involved in AR-mediated GR binding.
GR agonist alone has practically no effect on the BRD4 binding, and it can only induce FOXA1 binding at PSA enhancer site.
As the preliminary ChIP-qPCR results confirmed the occupancy of BRD4 and FOXA1 at the C1 and C2 loci, we next sought to confirm if they have a role in AR-mediated binding of GR on these sites. To answer this question, we depleted cells of FOXA1 using specific siRNA or
A B
15 inhibited BRD4 with specific inhibitor and studied the binding of AR and GR. As expected, inhibition of BRD4 reduces AR occupancy (Fig. 3C), which is complementary to published results (Supplementary Fig. 2). As for GR, interestingly, the inhibition of BRD4 and FOXA1 depletion both seemed to increase GR occupancy in Dex single treatment (Fig. 3D).
Figure 3. Validation of immunoprecipitation with ChIP-qPCR. Primers for FKBP5(-3) and PSA enhancer were used for (A) BRD4 ChIP-qPCR and (B) FOXA1 ChIP-qPCR. Both FKBP5(-3) and PSA enhancer regions are known to contain binding sites for AR and GR. (C) AR ChIP-qPCR and (D) GR ChIP-qPCR for same treatments and primers for BRD4/FOXA1 depleted VCaP cells. Cells were depleted of FOXA1 using 60 nM specific siRNA (siFOXA1) or exposed to control siRNA (siNON) for 72 h, or BRD4 inhibited with 1 h treatment with 1 µM I-BET762 (iBET) or treated with control solvent (DMSO). Treatments were with vehicle (EtOH), 100 nM Dex, 100 nM DHT, or both for 1 h. Data is presented as fold change to FKBP5(-3) EtOH treated sample.
In order to gain genome-wide information we proceeded with deep sequencing of the samples.
Validation of the sequencing samples are shown in Supplementary Figure 3. In Figure 4A, the differential binding peaks for GR, AR, FOXA1 and BRD4 between ethanol (EtOH) vehicle, GR agonist, AR agonist and dual treatments in C1 and C2 are shown as heatmaps. The combined signal of C1 and C2 sites with all treatments and factors are shown as aggregate plots
D C
A
B
16 in Figure 4B. The ATAC-seq supports the notion that GR binds independent of AR to sites located on pre-accessible chromatin. This seems to be in line with published results (John, et al. 2011). In comparison, AR activation leads to increased accessibility especially on AR- dependent GR binding sites, suggesting that AR-mediated chromatin remodeling is required for GR binding at C1 sites.
Figure 4. Chromatin binding of the glucocorticoid receptor, androgen receptor, FOXA1 and BRD4 in VCaP cells. ChIP-seq and ATAC-seq for hormone treated VCaP cells. Cells were treated 1 h with vehicle (EtOH), 100 nM Dex, 100 nM DHT or both previous. (A) Heatmaps represent ChIP-seq signals ±1 kb around the peak center for C1 (AR-dependent GR binding) and C2 (AR-independent GR binding) clusters. (B) Aggregate plots for tag density for each cluster with different colors indicating treatments. Results are also available as box plots in Supplementary Fig. 1.
Both BRD4 and FOXA1 occupy chromatin on C2 sites prior to hormone treatment, whereas on C1 sites their occupancy is greatly increased with AR ligand treatment (Fig. 4-5). Dex alone has only a slight effect to the FOXA1 and BRD4 chromatin binding. The increase in FOXA1 and BRD4 binding by DHT treatment could indicate that active AR induces accessibility at receptor binding sites for GR by recruiting chromatin modifier proteins BRD4 and FOXA1 on closed chromatin. In comparison, as indicated by ChIP-seq, FOXA1 and BRD4 already occupy pre-accessible sites (Fig. 5). Interestingly, GR activation with ligand decreased BRD4 occupancy in both regions, while dual agonist treatment showed synergistic increase, especially in C1.
Taken together, these data suggest that FOXA1 and BRD4 likely have a role in AR induced chromatin binding of GR. To further study their influence in these processes, we either depleted VCaP cells of FOXA1 using specific siRNA or inhibited BRD4 with specific inhibitor.
A
B
17 Subsequently, ChIP-seq for AR and GR was done to study the changes in receptor chromatin occupancy before and after silencing/inhibition.
Figure 5. FOXA1 and BRD4 are recruited to FKBP5(-3) by the androgen receptor, while they occupy PSA enhancer prior hormone induction. Normalized ChIP-seq sequencing tracks for (A) FKBP5 and (B) KLK3 (PSA) gene loci. VCaP cells were treated with vehicle (EtOH), 100 nM Dex, 100 nM DHT or both previous (Dex+DHT) for 1 h. Samples were immunoprecipitated with anti-BRD4 (purple) or anti-FOXA1 (orange). The FKBP5(-3) and PSA enhancer are highlighted with a blue outline.
Loss of BRD4 activity inhibits AR-mediated GR genome binding in VCaP cells
First, we studied the effect of BRD4 inhibition on GR and AR occupancy. Inhibition of BRD4 was achieved with bromodomain specific inhibitor I-BET762 (IC50 32.5-42.5 nM) (Asangani et al. 2014, Mirguet, et al. 2013). In addition to inhibition of activity, Asangani, et al. (2014) have shown with ChIP-seq that genome-wide chromatin binding of BRD2, -3 and -4 decreased with I-BET762 treatment (Fig. 6A).
Expectedly, inhibition of BRD4 did not have as drastic an effect on AR binding, as it did for GR. For AR, only a small decrease in occupancy can be seen (Fig. 6B-C, middle panel), like what has been observed by Asangani et al. 2014 (Supplementary Fig. 2). In keeping mind that AR recruits cofactor BRD4 potentially assisting in the binding of GR, inhibition of BRD4
A
B
18 results in decreased binding of GR, especially in the C1 sites (Fig. 6B-C, left panel). More specifically, in BRD4 inhibited cells there is a decrease in occupancy in both clusters on Dex+DHT induced binding sites, where the binding of GR is enhanced in the presence of active AR, but less so in those induced by Dex alone. This finding further supports the hypothesis of BRD4 being important in AR-mediated binding of GR. Example genome browser tracks for C1 and C2 sites are shown in Figure 7.
Subsequently, we wondered if the reduction in GR occupancy could be caused by compaction of chromatin due to the loss of chromatin remodeling activity by BRD4. To assess this, we performed ATAC-seq to study the changes in chromatin accessibility followed by inhibition of BRD4 (Fig. 6B-C, right panel). The ATAC-seq showed that overall only a small decrease in accessibility is followed by BRD4 inhibition, mainly in Dex+DHT treated C2 sites.
Figure 6. BRD4 inhibition affects glucocorticoid receptor occupancy separate from chromatin accessibility.
BRD4 was inhibited with 1 h treatment of 1 µM I-BET762 (IBET), followed by 1 h treatment with vehicle (EtOH), 100 nM Dex, 100 nM DHT, or both. Heatmaps represent ChIP-seq signals ±1 kb around the peak center of C1 and C2 sites. (A) Aggregate plot generated from BRD2/3/4 ChIP-seq data from Asangani, et al. (2014) for I-BET762 treated cells and DMSO control in VCaP cells treated with 500 nM I-BET762 for 6 h and 10 nM DHT for another 12 h. (B) Heatmap of AR and GR ChIP-seq peaks of BRD4 inhibited cells (iBET) and DMSO control, as well as ATAC-seq. (C) Aggregate plots for tag density for each cluster of iBET/DMSO samples with different colors indicating treatments.
We examined the same C1 and C2 representing regions from before [FKBP5(-3) and PSA enhancer] after sequencing as ChIP-seq and ATAC-seq peak tracks (Fig. 7 and 8). Without inhibition, Dex+DHT treatment induces binding of GR to FKBP5(-3), but interestingly, BRD4
A
B
C
19 inhibition lead to increased binding in Dex single treatment (Fig. 7A). On the PSA enhancer, however, BRD4 inhibition decreased GR occupancy in both treatments, but more so in dual treatment. AR binding on FKBP5 or PSA enhancer was not as affected by BRD4 inhibition, showing only slight decrease in receptor binding on both loci. According to ATAC-seq, there was little to no change in chromatin accessibility on either loci followed by BRD4 inhibition (Fig. 8). This indicates that the changes in GR occupancy are not caused by changed chromatin accessibility, but in fact due the lack of BRD4 activity. Based on these observations, it is likely that BRD4 indeed has a role in mediating AR-dependent GR occupancy on C1 sites.
Figure 7. BRD4 inhibition decreases androgen receptor mediated glucocorticoid receptor occupancy on FKBP5(-3) and PSA enhancer. Normalized ChIP-seq sequencing peak tracks visualized with IGV 2.3 for (A) FKBP5 and (B) KLK3 (PSA) gene loci. VCaP cells were treated with 1 µM I-BET762 (I-BET) for 1 h, followed by treatments with a vehicle (EtOH), Dex, DHT or both previous (Dex+DHT) for another 1 h. Samples were immunoprecipitated with anti-AR (gray background) or anti-GR (white background). The FKBP5(-3) and PSA enhancer are highlighted with a blue outline.
A
B
20 Figure 8. Chromatin accessibility is not drastically changed by BRD4 inhibition on FKBP5(-3) or PSA enhancer. Normalized ATAC-seq sequencing peak tracks visualized with IGV 2.3 for (A) FKBP5 and (B) KLK3 (PSA) gene loci. VCaP cells were treated with 1 µM I-BET762 (I-BET) for 1 h, and another 1 h with vehicle (EtOH), Dex, DHT or both previous (Dex+DHT). I-BET762 treatment is light purple, and DMSO control dark purple. The FKBP5(-3) and PSA enhancer are highlighted with a blue outline.
FOXA1 depletion diminishes GR genome binding in VCaP cells
The depletion of FOXA1 was achieved with FOXA1 specific siRNA (Sahu, et al. 2011;
Malinen, et al. 2017). Immunoblotting with anti-FOXA1 was done to validate silencing of FOXA1 (Fig. 9A). Due to a technical issue, Dex+DHT dual treated siNON AR ChIP-seq samples were not used for further analysis, as previous data shows that DHT and DHT+Dex treatments cause AR-binding of similar magnitude in VCaP cells (cf. AR in Fig. 4B to Fig. 9C).
However, the data was displayed in heatmap and aggregate plots together with data from the other samples.
Depletion of FOXA1 had a drastic effect on GR occupancy, especially in AR-independent GR binding sites (C2), where FOXA1 is normally bound prior to any hormone treatment. On these sites, the binding of GR is almost non-existent after FOXA1 depletion, as shown in
A
B
21 Figures 9B-C. It could be that the absence of FOXA1 leads to condensed chromatin where GR cannot bind by itself. The same effect, although not quite so drastic, can be seen in AR- dependent binding sites (C1). Occupancy of AR is also reduced upon FOXA1 depletion in both clusters, but considerably less than that of GR. Differences in DHT treatment between FOXA1 depleted cells and control show AR occupancy to decrease in both C1 and C2, but especially in C2, where occupancy is almost halved. The more drastic effect on GR binding could be indicative of FOXA1 having a greater role in GR than AR binding.
Interestingly, the GR occupancy in FOXA1 depleted cells reduced in both Dex and Dex+DHT treatments compared to control, whereas in BRD4 inhibited cells, mostly dual treatment showed reduction. This could be due to FOXA1 being generally important in facilitating receptor binding as it has a role in opening and maintaining open chromatin across the genome, and a level of decrease in accessibility is expected following its depletion.
Figure 9. Silencing of FOXA1 drastically decreases glucocorticoid receptor occupancy. VCaP cells were depleted of FOXA1 with 60 nM FOXA1 specific siRNA and non-specific control for 72 h, followed by treatment with a vehicle (EtOH), 100 nM Dex, 100 nM DHT, or both for 1 h. (A) Immunoblotting with anti-FOXA1 and control protein GAPDH was used to verify the depletion of FOXA1 after RNA interference. (B) Heatmap of AR and GR ChIP-seq peaks of FOXA1 depleted cells (siFOXA1) or control (siNON). Heatmaps represent ChIP-seq signals ±1 kb around the peak center of C1 and C2 sites. (C) Aggregate plots for tag density for each cluster of siFOXA1/siNON samples with different colors indicating different treatments.
To investigate the effect of FOXA1 depletion to chromatin accessibility, ATAC-seq was performed. Closer examination of the chromatin accessibility with ATAC-seq followed by FOXA1 depletion shows that accessibility on the FKBP5(-3) and PSA enhancer loci is curiously increased (Fig. 11). Despite that, occupancy of both GR and AR was decreased (Fig. 10). FOXA1 depletion noticeably decreased GR occupancy on both loci, while AR
A
B
C
22 occupancy was less effected. However, as the other FOXA1 siRNA samples had technical issues, we cannot be sure if the DHT ATAC-seq samples are truly representative.
Figure 10. Silencing of FOXA1 diminishes glucocorticoid receptor occupancy on FKBP5(-3) and PSA enhancer. Normalized ChIP-seq sequencing peak tracks visualized with IGV 2.3 for (A) FKBP5 and (B) KLK3 (PSA) gene loci. VCaP cells were treated with 60 nM FOXA1 specific siRNA (siFOXA1) or non-target control (siNON) for 72 h, followed by treatments with a vehicle (EtOH), Dex, DHT or both previous (Dex+DHT) for another 1 h. Samples were immunoprecipitated for AR (gray background) or GR (white background). The FKBP5(-3) and PSA enhancer are highlighted with a blue outline. AR ChIP-seq Dex+DHT samples were not included.
Figure 11. Chromatin accessibility followed by silencing of FOXA1 on FKBP5(-3) and PSA enhancer.
Normalized ATAC-seq sequencing peak tracks visualized with IGV 2.3 for (A) FKBP5 and (B) KLK3 (PSA) gene loci. VCaP cells were treated with 60 nM FOXA1 specific siRNA (siFOXA1) or non-target siRNA (siNON) for 72 h, followed by 1 h treatment with DHT. The FKBP5(-3) and PSA enhancer are highlighted with a blue outline.
ATAC-seq siNON EtOH, siNON Dex, siNON Dex+DHT, siFOXA1 EtOH, siFOXA1 Dex, and siFOXA1 Dex+DHT samples were not included.
B A
B
A
23
DISCUSSION
SRs like AR and ER are key drivers of hormonal cancers. Given their pivotal role in the development, progression and drug resistance of BC and PC, the function of the receptors in these cancers have been heavily investigated. However, the molecular mechanisms of GR-AR crosstalk in PC are still largely unknown, even though studies have shown that AR and GR are highly similar in their regulation of transcriptional targets (Arora, et al. 2013) as well as protein- protein interactions (Lempiäinen, et al. 2017). In this study we mapped the mechanism of GR- AR crosstalk in PC by investigating the possible role of two cofactors, BRD4 and FOXA1, in AR-mediated binding of GR on chromatin. BRD4 and FOXA1 both have chromatin modifying capability and are known to interact with SRs, contributing to the deregulation of AR and ER in PC and BC (Robinson & Carroll, 2012; Urbanucci, et al. 2017; Yang & Yu, 2015).
Our results indicate that both seem to have a role in GR occupancy in VCaP cells: GR binding was most sensitive to BRD4 inhibition in AR-GR coactivated cells, while FOXA1 depletion affected GR occupancy strongly in all treatments. Our chromatin accessibility analyses indicated that the decrease in GR genomic binding could not be completely explained by the alterations in openness of chromatin. Upon inhibition of BRD4 activity, the chromatin accessibility at GR binding sites decreased less than the binding of the receptor. However, we could not verify this in the case of FOXA1 depletion since samples did not yield reliable results.
Our results thus suggest that BRD4 and FOXA1 assist in the genomic binding of GR in ways that are not directly linked to chromatin remodeling. Improved understanding of these mechanisms would be required to better understand disease progression and to find alternative drug targets for SR driven hormonal cancers. Interestingly, both BRD4 and FOXA1 are highly sought targets for drug therapies. Indeed, very recently several compounds that inhibit BET protein, including BRD4, activity have been developed for the treatment of inflammatory diseases and several cancers, including PC (Faivre et al. 2020, Gilan et al. 2020). In the case of FOXA1, currently no known drugs exist to inhibit its activity. However, Azeria Therapeutics (https://www.azeriatherapeutics.com/) is currently developing a small molecule drug to inhibit FOXA1 activity.
The crosstalk of GR with another SR, the ER, in BC has been studied in more depth. Studies on the coactivation of GR and ER have shown that they can alter genomic binding of each other by inducing chromatin remodeling via the recruitment of a secondary factor (Miranda, et al.
2013). This is known as assisted loading (Fig. 1), a mechanism that allows TFs to modulate chromatin binding patterns of other TFs and has been demonstrated for multiple SRs (Tonsing-
24 Carter, et al. 2019, Voss, et al. 2011). In support of the assisted loading model, Vahrenkamp, et al. (2018) showed that chromatin accessibility increased followed by GR-ER coactivation in an endometrial cancer cell line. In another study, GR-ER coactivation in BC similarly caused significant chromatin remodeling on ER binding sites (Tonsing-Carter, et al. 2019).
Preliminary results in the Palvimo research group support the hypothesis that there is a similar mechanism for AR-GR crosstalk in VCaP cells, where AR facilitates GR occupancy in a subset of sites in AR-GR coactivated cells by modulating chromatin accessibility indirectly via secondary factors. It is noteworthy to point out that assisted loading mechanism is not restricted solely to cancers, since it has been observed also in hepatic fasting response (Goldstein et al.
2017).
Our results for AR genomic binding followed by BRD4 inhibition follow a similar trend found in a study by Asangani, et al. (2014). They showed that the inhibition of BRD4 lead to decreased AR binding in BRD4 inhibitor treated VCaP cells. In their study, AR occupancy decreased more (~15% vs. ~40% decrease) due to longer treatment times used compared to our study: 6 h inhibitor treatment and hormone induction for 12 h, as opposed to our 1 h followed by 1 h treatment times. However, the AR binding in our control sample was over 2-fold better than the AR binding in Asangani, et al. (2014). This could also explain the differences between the studies. In the end, experiments with the exact same conditions would have to be performed to verify these discrepancies. Shah, et al. (2017) reported that BRD4 inhibition primarily affected GR-biased genes, showing only a moderate effect on AR target gene expression. We did not measure transcript or protein expression, but instead found GR occupancy to decrease more than AR, especially on sites bound by GR in the presence of active AR. As BRD4 is known to mediate AR activity in CRPC (Urbanucci, et al. 2017), our results could indicate that BRD4 is needed for AR-mediated GR binding on a subset of sites, potentially affecting GR target gene expression as well.
The decrease in AR occupancy caused by FOXA1 depletion can possibly be explained partly due to redistribution of AR binding and loss of pioneering by FOXA1, as FOXA1 has been shown to inhibit AR binding to chromatin on some loci, while increasing binding on others (Jin, et al. 2014; Sahu, et al. 2011). Our results show that besides the known pioneer activity of FOXA1 in recruiting SRs, FOXA1 can also be recruited by AR to a subset of binding sites.
Once it is depleted, GR occupancy on these sites is diminished. FOXA1 maintains chromatin accessibility and GR preferably binds to open chromatin (John, et al. 2011), which helps to explain why GR occupancy decreases upon loss of FOXA1 activity in literature (Sahu, et al.
25 2013). However, based on our results we cannot draw definitive conclusions as to how FOXA1 mediates the binding of GR before further studies are performed, as we were not able to fully study the changes in chromatin accessibility followed by depletion of FOXA1. It is also possible that FOXA1 influences GR binding through different mechanisms at AR-independent versus AR-dependent sites. Based on other published data on SR crosstalk, we speculate that assisted loading of GR by tethering with BRD4 and/or FOXA1 could be one option to explain our findings.
Miranda, et al. (2013) reported that the GR assisted ER binding sites were enriched in binding motifs for a TF called Activator Protein-1 (AP-1) in a mouse BC cell line. They proposed an assisted loading model, where GR recruits chromatin remodelers, after which AP-1 binds and facilitates the binding of ER by tethering. Similar role for AP-1 in chromatin association of coactivated GR-ER has been reported in other studies as well (Biddie, et al. 2011; Vahrenkamp, et al. 2018; West, et al. 2016). As the BC cell lines used in the paper by Miranda, et al. (2013) do not express FOXA1 and the VCaP cells in turn do not significantly express AP-1, we propose that these proteins could contribute in a similar manner to the cell-type specific regulation of genomic GR binding in the respective cell lines. Indeed, crosstalk between GR and FOXA1 is observed in FOXA1-expressing BC cells (Swinstead, et al. 2016a, Karmakar, et al. 2013). In addition, the role of FOXA1 in GR crosstalk with other SRs has been determined in hormonal cancers before.
Historically, GRs role in inflammatory diseases and hematologic cancers has been established, and synthetic GR agonists are widely used due to their anti-inflammatory and immunosuppressive effects. For many cancers, synthetic glucocorticoids are used to treat adverse effects e.g. from chemotherapy (Pufall, 2015). However, GR can also contribute to progression of many diseases such as cancers. It was shown recently that glucocorticoids can induce metastasis and immunosuppression of BC favoring cancer progression (Obradović, et al. 2019). Thus, it could be detrimental to the BC patient to treat adverse effects with glucocorticoids. Hence, it is important to understand the mechanism of how GR operates in these diseases. In PC the hypothesis of GR being a driver of drug resistance is strongly supported by studies consistently showing that GR antagonism is enough to regain drug sensitivity in AR-inhibition resistant PC cells (Arora, et al. 2013; Isikbay, et al. 2014; Puhr, et al. 2018). While research has so far concentrated on studying individual SRs, recently it has become clear that SRs interact via many mechanisms, contributing to progression and drug resistance in hormonal cancers (Arora, et al. 2013; D'Amato, et al. 2016; Miranda, et al. 2013).
26 In the context of PC, AR has been well studied (Basu & Tindall, 2010), while considerably less is known about the impact of GR activity, especially in ADT naïve cells. In this thesis, we studied an ADT naïve CRPC cell line and gained new information on the mechanism of GR- AR crosstalk and the proteins that are part of it.
As TF interactions on chromatin are often transient and complex, future studies on whether the cofactors we studied bind to chromatin prior to GR or if they co-occupy need to be completed to better understand the nature of these interactions. This can be achieved with methods like single molecule tracking with fluorescence microscopy as has been done previously to study the GR-ER crosstalk (Paakinaho, et al. 2017; Swinstead, et al. 2016a). Also, the role of other possible cofactors that we did not study in this thesis need to be researched in the future. Other potential cofactors could be discovered with proteomic approaches (Lempiäinen, et al. 2017).
Mapping of these mechanisms will open new drug target possibilities to compliment directly AR targeting therapies in treatment of PC and help us understand the contribution of SR crosstalk in the progression of hormonal cancers.
ACKNOWLEDGEMENTS
This study was done at the University of Eastern Finland and funded by Prof. Jorma Palvimo’s laboratory. I’d like to give special thanks to my main supervisor Ville Paakinaho (PhD) for the research topic, as well as overseeing the practical execution and writing during this project. I am also thankful to research group members for help and support in the laboratory. Final thanks to Prof. Jorma Palvimo for his insight on the subject.
27
REFERENCES
Adams EJ, Karthaus WR, Hoover E, Liu D, Gruet A, Zhang Z, Cho H, DiLoreto R, Chhangawala S, Liu Y, Watson PA, Davicioni E, Sboner A, Barbieri CE, Bose R, Leslie CS, Sawyers CL (2019) FOXA1 mutations alter pioneering activity, differentiation and prostate cancer phenotypes. Nature. 571(7765):408-412. doi: 10.1038/s41586-019-1318-9.
Arora VK, Schenkein E, Murali R, Subudhi SK, Wongvipat J, Balbas MD, Shah N, Cai L, Efstathiou E, Logothetis C, Zheng D & Sawyers CL. (2013) Glucocorticoid receptor confers resistance to antiandrogens by bypassing androgen receptor blockade. Cell. 155(6):1309-22.
Asangani IA, Dommeti VL, Wang X, Malik R, Cieslik M, Yang R, Escara-Wilke J, Wilder- Romans K, Dhanireddy S, Engelke C, Iyer MK, Jing X, Wu YM, Cao X, Qin ZS, Wang S, Feng FY & Chinnaiyan AM (2014) Therapeutic targeting of BET bromodomain proteins in castration-resistant prostate cancer. Nature. 510(7504): 278-82. DOI: 10.1038/nature13229.
Bannister AJ & Kouzarides T (2011) Regulation of chromatin by histone modifications. Cell Res. (3):381-95. doi: 10.1038/cr.2011.22.
Basu S & Tindall DJ (2010) Androgen Action in Prostate Cancer. HORM CANC. 1(5):223–
228. DOI: 10.1007/s12672-010-0044-4.
Belikov S, Öberg C, Jääskeläinen T, Rahkama V, Palvimo JJ & Wrange Ö (2013) FoxA1 corrupts the antiandrogenic effect of bicalutamide but only weakly attenuates the effect of MDV3100 (Enzalutamide™). Mol Cell Endocrinol. 2013 Jan 5;365(1):95-107. doi:
10.1016/j.mce.2012.10.002.
Biddie SC, John S, Sabo PJ, Thurman RE, Johnson TA, Schiltz RL, Miranda TB, Sung MH, Trump S, Lightman SL, Vinson C, Stamatoyannopoulos JA, Hager GL (2011) Transcription factor AP1 potentiates chromatin accessibility and glucocorticoid receptor binding. Mol Cell.
43(1):145-55. doi: 10.1016/j.molcel.2011.06.016.
Buenrostro JD, Wu B, Chang HY & Greenleaf WJ (2015) ATAC‐seq: A Method for Assaying Chromatin Accessibility Genome‐Wide. Curr Protoc Mol Biol. 109:21–29. DOI:
10.1002/0471142727.mb2129s109
Chatterjee N, Sinha D, Lemma-Dechassa M, Tan S, Shogren-Knaak MA & Bartholomew B (2011) Histone H3 tail acetylation modulates ATP-dependent remodeling through multiple mechanisms. Nucleic Acids Res. 39(19):8378–8391. doi:10.1093/nar/gkr535
Clapier CR, Iwasa J, Cairns BR & Peterson CL (2017) Mechanisms of action and regulation of ATP-dependent chromatin-remodelling complexes. Nat Rev Mol Cell Biol. 18(7):407–422.
doi:10.1038/nrm.2017.26
Corces MR, Granja JM, Shams S, Louie BH, Seoane JA, Zhou W, Silva TC, Groeneveld C, Wong CK, Cho SW, Satpathy AT, Mumbach MR, Hoadley KA, Robertson AG, Sheffield NC, Felau I, Castro MAA, Berman BP, Staudt LM, Zenklusen JC, Laird PW & Curtis C (2018) Cancer Genome Atlas Analysis Network, Greenleaf WJ, Chang HY. The chromatin accessibility landscape of primary human cancers. Science. 362(6413):eaav1898. doi:
10.1126/science.aav1898.
D'Amato NC, Gordon MA, Babbs B, Spoelstra NS, Carson Butterfield KT, Torkko KC, Phan VT, Barton VN, Rogers TJ, Sartorius CA, Elias A, Gertz J, Jacobsen BM & Richer JK (2016)
28 Cooperative Dynamics of AR and ER Activity in Breast Cancer. Mol Cancer Res. 14(11):1054- 1067. doi: 10.1158/1541-7786.MCR-16-0167.
Devaiah BN, Case-Borden C, Gegonne A, Hsu CH, Chen Q, Meerzaman D, Dey A, Ozato K
& Singer DS (2016) BRD4 is a histone acetyltransferase that evicts nucleosomes from chromatin. Nat Struct Mol Biol. 23(6):540-8. doi: 10.1038/nsmb.3228.
Faivre EJ, McDaniel KF, Albert DH, et al. (2020) Selective inhibition of the BD2 bromodomain of BET proteins in prostate cancer. Nature. 578(7794):306‐310.
doi:10.1038/s41586-020-1930-8
Filippakopoulos P, Qi J, Picaud S, Shen Y, Smith WB, Fedorov O, Morse EM, Keates T, Hickman TT, Felletar I, Philpott M, Munro S, McKeown MR, Wang Y, Christie AL, West N, Cameron MJ, Schwartz B, Heightman TD, La Thangue N, French CA, Wiest O, Kung AL, Knapp S & Bradner JE (2010) Selective inhibition of BET bromodomains. Nature.
468(7327):1067-73. doi: 10.1038/nature09504.
Flavahan WA, Gaskell E & Bernstein BE (2017) Epigenetic plasticity and the hallmarks of cancer. Science. 357(6348):eaal2380. doi: 10.1126/science.aal2380.
Gao N, Zhang J, Rao MA, Case TC, Mirosevich J, Wang Y, Jin R, Gupta A, Rennie PS, Matusik RJ (2003) The Role of Hepatocyte Nuclear Factor-3α (Forkhead Box A1) and Androgen Receptor in Transcriptional Regulation of Prostatic Genes. Mol. Endocrinol. 17(8):
1484–1507, https://doi.org/10.1210/me.2003-0020
Gilan O, Rioja I, Knezevic K, et al. (2020) Selective targeting of BD1 and BD2 of the BET proteins in cancer and immunoinflammation. Science. 368(6489):387‐394.
doi:10.1126/science.aaz8455
Goldstein I, Baek S, Presman DM, Paakinaho V, Swinstead EE, Hager GL (2017) Transcription factor assisted loading and enhancer dynamics dictate the hepatic fasting response. Genome Res. 27(3):427-439. doi: 10.1101/gr.212175.116.
Handle F, Prekovic S, Helsen C, Van den Broeck T, Smeets E, Moris L, Eerlings R, Kharraz SE, Urbanucci A, Mills IG, Joniau S, Attard G & Claessen F (2019) Drivers of AR indifferent antiandrogen resistance in prostate cancer cells. Sci Reports. 9:13786
Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P, Cheng JX, Murre C, Singh H &
Glass CK (2010) Simple combinations of lineage-determining transcription factors prime cis- regulatory elements required for macrophage and B cell identities. Mol Cell. 38(4):576-89.
DOI: 10.1016/j.molcel.2010.05.004.
Hurtado A, Holmes KA, Ross-Innes CS, Schmidt D & Carroll JS (2011) FOXA1 is a key determinant of estrogen receptor function and endocrine response. Nat Genet. 43(1):27-33. doi:
10.1038/ng.730.
Isikbay M, Otto K, Kregel S, Kach J, Cai Y, Vander Griend DJ, Conzen SD, and Szmulewitz RZ (2014) Glucocorticoid Receptor Activity Contributes to Resistance to Androgen-Targeted Therapy in Prostate Cancer, Horm Cancer. 5(2):72–89.
Jin HJ, Zhao JC, Wu L, Kim J, Yu J (2014) Cooperativity and equilibrium with FOXA1 define the androgen receptor transcriptional program. Nat Commun. 5:3972. doi:
10.1038/ncomms4972.
29 John S, Sabo PJ, Thurman RE, Sung MH, Biddie SC, Johnson TA, Hager GL &
Stamatoyannopoulos JA (2011) Chromatin accessibility pre-determines glucocorticoid receptor binding patterns. Nat Genet. 43(3):264-8. DOI: 10.1038/ng.759.
Jozwik KM & Carroll JS (2012) Pioneer factors in hormone-dependent cancers. Nat. Rev.
Cancer. 12(6):381-5.
Karmakar S, Jin Y & Nagaich AK (2013) Interaction of Glucocorticoid Receptor (GR) with Estrogen Receptor (ER) α and Activator Protein 1 (AP1) in Dexamethasone-mediated Interference of ERα Activity. J. Biol. 288: 24020-24034. doi:10.1074/jbc.M113.473819 Karvonen U, Kallio PJ, Jänne OA & Palvimo JJ (1997) Interaction of androgen receptors with androgen response element in intact cells. Roles of amino- and carboxylterminal regions and the ligand. J.Biol.Chem. 272: 15973-15979. doi: 10.1074/jbc.272.25.15973
Lambert JP, Picaud S, Fujisawa T, et al. (2019) Interactome Rewiring Following Pharmacological Targeting of BET Bromodomains. Mol Cell. 73(3):621–638.e17.
doi:10.1016/j.molcel.2018.11.006
Langmead B & Salzberg SL (2012) Fast gapped-read alignment with Bowtie 2. Nat Methods.
9(4):357-9. DOI: 10.1038/nmeth.1923.
Lempiäinen JK, Niskanen EA, Vuoti KM, Lampinen RE, Göös H, Varjosalo M & Palvimo JJ (2017) Agonist-specific Protein Interactomes of Glucocorticoid and Androgen Receptor as Revealed by Proximity Mapping. Mol Cell Proteomics. 16(8):1462-1474. doi:
10.1074/mcp.M117.067488.
Local A, Huang H, Albuquerque CP, Singh N, Lee AY, Wang W, Wang C, Hsia JE, Shiau AK, Ge K, Corbett KD, Wang D, Zhou H & Ren B (2018) Identification of H3K4me1-associated proteins at mammalian enhancers. Nat Genet. 50(1):73-82. doi: 10.1038/s41588-017-0015-6.
Love MI, Huber W & Anders S (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15(12):550. DOI: 10.1186/s13059-014-0550-8.
Lupien M, Eeckhoute J, Meyer CA, Wang Q, Zhang Y, Li W, Carroll JS, Liu XS & Brown M (2008) FoxA1 translates epigenetic signatures into enhancer-driven lineage-specific transcription. Cell. 132(6):958-70. doi: 10.1016/j.cell.2008.01.018.
Malinen M, Niskanen EA, Kaikkonen MU & Palvimo JJ (2017) Crosstalk between androgen and pro-inflammatory signaling remodels androgen receptor and NF-κB cistrome to reprogram the prostate cancer cell transcriptome. Nucleic Acids Res. 45(2):619–630.
Miranda TB, Voss TC, Sung MH, Baek S, John S, Hawkins M, Grøntved L, Schiltz RL &
Hager GL (2013) Reprogramming the chromatin landscape: interplay of the estrogen and glucocorticoid receptors at the genomic level. Cancer Res. 73(16):5130-9. DOI: 10.1158/0008- 5472.CAN-13-0742.
Mirguet O, Gosmini R, Toum J, Clément CA, Barnathan M, Brusq J, Mordaunt JE, Grimes RM, Crowe M, Pineau O, Ajakane M, Daugan A, Jeffrey P, Cutler L, Haynes AC, Smithers NN, Chung C, Bamborough P, Uings IJ, Lewis A, Witherington J, Parr N, Prinjha RK &
Nicodème E (2013) Discovery of Epigenetic Regulator I-BET762: Lead Optimization to Afford a Clinical Candidate Inhibitor of the BET Bromodomains. J. Med. Chem. 56(19):7501- 7515. DOI: 10.1021/jm401088k
Mohammed H, Russell IA, Stark R, et al. (2015) Progesterone receptor modulates ERα action in breast cancer. 523(7560):313–317. doi:10.1038/nature14583
