This Article Abstract Full Text (PDF) Other Versions of this Article: MCB.01297-06v1 27/5/1823    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Klokk, T. I. Articles by Saatcioglu, F. Search for Related Content PubMed PubMed Citation Articles by Klokk, T. I. Articles by Saatcioglu, F.

 Previous Article  |  Next Article 

Molecular and Cellular Biology, March 2007, p. 1823-1843, Vol. 27, No. 5
0270-7306/07/$08.00+0     doi:10.1128/MCB.01297-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Ligand-Specific Dynamics of the Androgen Receptor at Its Response Element in Living Cells

Tove I. Klokk,^1, Piotr Kurys,^1, Cem Elbi,^2, Akhilesh K. Nagaich,^2, Anindya Hendarwanto,² Thomas Slagsvold,¹ Ching-Yi Chang,³ Gordon L. Hager,^2* and Fahri Saatcioglu^1*

Department of Molecular Biosciences, University of Oslo, Postboks 1041 Blindern, 0316 Oslo, Norway,¹ Laboratory of Receptor Biology and Gene Expression, National Cancer Institute, Bethesda, Maryland 20892,² Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, North Carolina 27710³

Received 14 July 2006/ Returned for modification 24 August 2006/ Accepted 8 December 2006

	ABSTRACT

Top Abstract Introduction Materials and Methods Results Discussion References

 
Androgens have key roles in normal physiology and in male sexual differentiation as well as in pathological conditions such as prostate cancer. Androgens act through the androgen receptor (AR), which is a ligand-modulated transcription factor. Antiandrogens block AR function and are widely used in disease states, but little is known about their mechanism of action in vivo. Here, we describe a rapid differential interaction of AR with target genomic sites in living cells in the presence of agonists which coincides with the recruitment of BRM ATPase complex and chromatin remodeling, resulting in transcriptional activation. In contrast, the interaction of antagonist-bound or mutant AR with its target was found to be kinetically different: it was dramatically faster, occurred without chromatin remodeling, and resulted in the lack of transcriptional inhibition. Fluorescent resonance energy transfer analysis of wild-type AR and a transcriptionally compromised mutant at the hormone response element showed that intramolecular interactions between the N and C termini of AR play a key functional role in vivo compared to intermolecular interactions between two neighboring ARs. These data provide a kinetic and mechanistic basis for regulation of gene expression by androgens and antiandrogens in living cells.

	INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

 
Androgens have a critical role in the development and maintenance of the male reproductive system and have roles in physiological and pathological conditions, including the normal prostate and prostate cancer (for reviews, see references 34 and 37). The effects of androgens are mediated by the androgen receptor (AR), which is a ligand-dependent transcription factor that belongs to the nuclear receptor superfamily. Other steroid receptors, such as the progesterone receptor (PR), glucocorticoid receptor (GR), and estrogen receptor (ER), also belong to this family. Upon hormone binding, steroid receptors change conformation, bind hormone response elements (HREs) in target promoters, and initiate gene transcription by the recruitment of chromatin modifying and remodeling complexes, coregulators, and other factors of the basal transcription apparatus (6, 13, 26, 29, 45, 51, 79).

As with other nuclear receptors, AR has three distinct domains: an N-terminal transactivation domain (NTD), a central DNA binding domain, and a C-terminal ligand binding domain (LBD). Upon binding to DNA, sequences found in the NTD (called activation function 1 [AF-1]) and LBD (AF-2) facilitate activation of transcription. Genetic and biochemical experiments have indicated that the LBD of AR interacts with the NTD upon ligand binding (7, 14, 27, 32, 82), which is similar to the results observed for the ER (39). This intramolecular interaction has been shown to be important for optimal receptor activity (7, 12, 14, 32, 82). However, these studies have been performed with truncated versions of the receptors in mammalian or yeast two-hybrid systems or in biochemical experiments in vitro. Therefore, the importance of these intramolecular and possible additional intermolecular (between two AR proteins) interactions for the function of AR with respect to its target gene in vivo has not been directly assessed.

Androgens are required for the growth of prostate cancer in the initial stages; this requirement is the basis for hormonal therapy that is a critical therapeutic option in advanced prostate cancer (30). An integral part of this therapy is the use of antiandrogens to block AR function; for example, the nonsteroid antagonists bicalutamide (Casodex) and flutamide (Eulexin) are two compounds commonly used in prostate cancer therapy today (53). These compounds antagonize AR function by binding to the LBD of AR in competition with the natural agonists testosterone and 5-dihydrotestosterone (12, 47, 61). Even though it is known that the AR-antagonist complex does not activate transcription, it is not completely clear which steps are influenced by antiandrogens in the AR signaling pathway. For example, it was long held (based largely on biochemical and in vitro experiments) that the antagonists may block nuclear import or DNA binding. However, data exist supporting the opposing view (see, for example, references 36 and 47). In fact, it has recently become clear that AR antagonists actually facilitate AR-DNA association but inhibit transcriptional activation via the recruitment of corepressors (68). In support of this view, a recent study demonstrated that antagonist function can be blocked by the disruption of corepressor recruitment (85). It has also been suggested, as for ER (for a review, see reference 25), that antagonists give rise to a different conformation of the LBD compared with the agonists, thereby affecting the interactions of AR with coactivators and corepressors when it is bound to DNA (8). However, modulation of the dynamic properties of AR with respect to its target gene in the presence of different ligands in vivo and its functional consequence have not been studied to date in a living cell.

Until recently, there was little information about the mode of action of nuclear receptors in living cells. The classical view of nuclear receptor function has been that ligand-activated receptors are immobilized on the template as long as the ligand is present in the cellular milieu (5), serving as a platform for the assembly of large transcriptional complexes (13, 48). Recent advances in green fluorescent protein (GFP) technology and quantitative live cell microscopy have led to the discovery of novel principles for nuclear receptor action, leading to the proposal of an alternative model, the "hit-and-run" hypothesis (18, 49, 56, 63, 65). According to this model, the receptor transiently interacts with the promoter, recruits other factors, and is itself dynamically displaced from the promoter (49, 62).

The dynamic interaction of nuclear receptors with their target genes in living cells in response to the presence of various ligands, both agonists and antagonists, has not been quantitatively characterized. Furthermore, it is not clear whether there are inter- and intramolecular interactions when a nuclear receptor is bound to DNA in its transcriptionally active form. In the present study, we systematically investigated the dynamic interactions of AR with its target promoter in living cells compared to nontarget site interactions in the nucleus in response to a complete range of agonists, partial antagonists, and pure antagonists. We determined the ability of AR to selectively recruit Swi/Snf ATP-dependent chromatin-remodeling complex to the target promoter in response to the presence of different AR ligands. We then correlated the changes in AR kinetics to the changes in chromatin remodeling and transcriptional activation. Finally, we used fluorescence resonance energy transfer (FRET) (84) to directly assess possible interactions within and between AR molecules at the AR target gene during transcriptional activation. Thus, our observations provide an integrated kinetic framework for the real-time gene regulatory events that are critical for the in vivo function of AR with respect to its target gene.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

 
Plasmids. PCR was used to clone a 3' fragment of AR cDNA, representing codons ATG (Meth-1) to CTG (Leu-564), into pEGFP-C1 (5' primer, T ATG AAT TCG ATG GAA GTG CAG TTA GGG CTG G; 3' primer, T CAG GCA GGT CTT CTG GGG YGG). The PCR product was digested with EcoRI and KpnI, and the resulting DNA fragment, together with the KpnI-BamHI fragment of pSG5-AR (21), was ligated with the EcoRI-BamHI fragment of pEGFP-C1 (BD Biosciences Clontech). The NheI-XbaI fragment of pEGFP-C1-AR containing GFP-AR cDNA was inserted into pTRE-Tight (BD Biosciences Clontech) at the NheI-XbaI sites to create pTRE-Tight-EGFP-AR, which was used for stable transfection. For expression of GFP-AR-E897A, the KpnI-BamHI fragment of pEGFP-C1-AR was replaced with the KpnI-BamHI fragment of pSG5-AR-M6 (71) to create pEGFP-C1-AR-E897A. The BstEII-XbaI fragment of pTRE-Tight-EGFP-AR was then replaced with the BstEII-XbaI fragment from pEGFP-C1-AR-E897A to create pTRE-Tight-EGFP-AR-E897A, which was used for stable transfection.

The AR portion of pTRE-Tight-EGFP-AR was cut with XbaI and EcoRI (partial) and inserted into the same sites of pECFP-C1 (BD Biosciences Clontech) to create pECFP-C1-AR. The AR sequence from pTRE-Tight-EGFP-AR was cut with XhoI and BamHI and inserted into the same sites of pEYFP-N1 (BD Biosciences Clontech) to create pEYFP-N1-AR. The stop codon was removed, and the frame was corrected by PCR (primers available upon request).

The PvuI-XbaI fragment of pEGFP-C1-AR-E897A was inserted into the same sites of pECFP-C1-AR to create pECFP-C1-AR-E897A, and the PvuI-BamHI fragment of pEGFP-C1-AR-E897A was inserted into PvuI-BamHI sites of pEYFP-N1-AR to create pEYFP-N1-AR-E897A. The stop codon was removed, and the frame was corrected by PCR (primers available upon request).

The NheI-KpnI AR fragment of pEYFP-N1-AR was replaced with the NheI-KpnI ECFP-AR fragment from pECFP-C1-AR to create pEYFP-N1-ECFP-AR, expressing fusion protein ECFP-AR-EYFP. The PvuI-BamHI fragment containing an E897A mutation was transferred from pEGFP-C1-AR-E897A into the same sites of pEYFP-N1-ECFP-AR. The stop codon was removed, and the frame was corrected by PCR (primers available upon request).

CFP-YFP fusion plasmid has previously been described (35). pPUR plasmid (BD Biosciences Clontech) was used without modifications.

The correct sequences of all final constructs were confirmed by sequencing.

Reporter plasmids MMTV-LUC (42) and -285PB-LUC(31) have been described previously.

Agonists and antagonists. R1881 (methyltrienolone) was purchased from Dupont-NEN. Both 5-dihydrotestosterone (DHT) and testosterone (TST) were kind gifts from Jens Berg (hormone laboratory, Aker University Hospital, Oslo, Norway). Cyproterone acetate (CPA) and mifepristone (RU486) were purchased from Sigma, while bicalutamide was obtained from Astra Zeneca. Hydroxyflutamide (OHF) was purchased from Schering-Plough Research Institute, Kenilworth, NJ. All ligands were dissolved in 100% ethanol and used at a working concentration of 10^–8 M (R1881, DHT, and TST) or 10^–6 M (CPA, RU486, bicalutamide, and OHF).

Cell culture and generation of stable cell lines. Stable cell lines expressing GFP-AR and GFP-AR-E897A under the control of the Tet-Off inducible system (24, 69) were obtained as stably transfected derivatives of murine mammary adenocarcinoma cell line 3134. The 3134 cell line contains multiple copies of a bovine papillomavirus-mouse mammary tumor virus (MMTV)-long terminal repeat (LTR)-ras fusion gene (78). The wild-type GFP-AR and mutant GFP-AR-E897A constructs were transfected along with a puromycin resistance plasmid, pPUR, into a Tet-Off cell line (5858 cells). The 5858 cell line was generated by transfecting pTet-Off (Bdbiosciences Clontech) into the 3134 cell line. Colonies were selected in media supplemented with 0.55 µg/ml puromycin (BD Biosciences Clontech) for GFP-AR and 1.1 µg/ml puromycin for GFP-AR-E897A. The cells were maintained in Dulbecco's modified Eagle medium (Gibco) supplemented with 10% fetal bovine serum (Gemini, Woodland, CA), 2 mM L-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 5 mg/ml penicillin-streptomycin, 1 mg/ml G418 (Gibco), 0.55 µg/ml (GFP-AR) or 1.1 µg/ml (GFP-AR-E897A) puromycin, and 10 µg/ml tetracycline (FisherBiotech, Fair Lawn, NJ) at 37°C in 5% CO₂ in a humidified incubator. The cell lines with the inducible expression of GFP-AR or GFP-AR-E897A were named 3108 or 3109, respectively.

Protein extraction and Western analysis. Cells were harvested by scraping in phosphate-buffered saline and centrifugation. Whole-cell extracts were prepared by resuspending the cells in 200 µl of lysis buffer (20 mM HEPES [pH 7.4], 300 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.1% Triton X-100, 0.5 mM dithiothreitol [DTT]) with a protease inhibitor cocktail mix (Calbiochem). After the extraction, the proteins were resolved on a 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (12% SDS-PAGE) gel (Bio-Rad) and transferred to a polyvinylidene difluoride membrane (Bio-Rad). The membrane was blocked, followed by incubation with the primary antibodies for AR (N-20; Santa Cruz) (1:250), GFP (AbCam) (1:500), ß-tubulin (Sigma) (1:1,000), or ß-actin (Calbiochem) (1:30,000). Horseradish peroxidase-linked secondary antibodies (Pierce Biochemicals) and an enhanced luminescence kit (Amersham Pharmacia) were used for the detection of proteins.

Luciferase reporter assay. Cells were grown in six-well culture plates in medium without tetracycline (3108 and 3109 cell lines). One day after plating, the cells were transfected with 100 to 150 ng of luciferase reporter (MMTV-LUC or -285PB-LUC), 10 ng of the appropriate AR construct (3134 cells only), and carrier DNA to a total of 1 µg DNA per well using FuGene6 (Roche) according to the manufacturer's recommendations. At 6 h after transfection, the medium was changed to phenol red-free Dulbecco's modified Eagle medium supplemented with 0.5% charcoal-stripped serum. One day after transfection, the cells were treated with various ligands for 24 h. The cells were harvested and lysed in luciferase cell culture lysis reagent (25 mM Tris-HCl [pH 7.8], 2 mM DTT, 10% glycerol, and 1% Triton X-100), and luciferase activity was determined using a luciferase assay system (Promega) and a Wallac Victor² 1420 Multilabel counter (Perkin Elmer). Protein concentrations were measured using a Bio-Rad protein assay, and luciferase activity was normalized to total protein.

Time-lapse microscopy. Cells were transferred to Lab-Tek one-well chamber slides (Nalge Nunc International, Naperville, IL) for live-cell imaging. The cells were grown in medium without tetracycline for two days prior to the experiment, including one day in phenol red-free medium supplemented with 2% charcoal-stripped serum and one day in phenol red-free medium supplemented with 0.5% charcoal-stripped serum. The cells were observed at 37°C using a Zeiss LSM 510 laser-scanning confocal microscope equipped with a 100x/1.3 numerical-aperture oil immersion objective and a 40 mW argon laser.

RNA FISH and immunofluorescence analysis. Cells were grown on 22-mm-square coverslips placed in six-well culture plates. Cell culture conditions were same as described for the time-lapse microscopy. At the day of the experiment, the cells were treated with the ligands for 45 min (R1881, DHT, TST, and RU486) or 90 min (CPA, bicalutamide, OHF), fixed with 4% paraformaldehyde, and processed for indirect immunofluorescence microscopy combined with RNA fluorescence in situ hybridization (FISH) to detect MMTV transcript as described previously (62). GFP-AR was detected by using a polyclonal anti-GFP antibody (Molecular Probes), and a polyclonal BRM antibody (AbCam) was used for BRM detection. Polymerase II (PolII) was detected using an RNA PolII 8SWG16 monoclonal antibody (Covance). Images were acquired on a Zeiss LSM 510 META or an Olympus FluoView 1000 confocal laser-scanning microscope. The RNA FISH signals were quantified by using MetaMorph software (Universal Imaging, Downingtown, PA) after subtraction of the background nuclear fluorescence as previously described (62). Then, the integrated total RNA FISH intensity was calculated for each condition and normalized to the level of integrated total RNA FISH intensity in untreated cells to obtain relative RNA FISH intensity values. Line scans were created using Olympus FV10-ASW 1.3b software.

FRAP. Cells were transferred to Lab-Tek one- or two-well chamber slides for live cell imaging (Nalge Nunc International, Naperville, IL). Cell culture conditions were same as described above. Fluorescence recovery after photobleaching (FRAP) analysis was carried out on a Zeiss LSM 510 laser-scanning confocal microscope. The stage temperature was maintained at 37°C, and images were captured with a 100x/1.3-numerical aperture oil immersion objective and 40 mW argon laser.

Five single prebleach images were acquired followed by a brief bleach pulse of 160 ms using 458-, 488-, and 514-nm laser lines at 100% laser power (laser output, 50%) without attenuation. Single optical sections were acquired at 490-ms or 96-ms intervals by using a 488-nm laser line with laser power attenuated to 0.2%. Fluorescence intensities in the regions of interest were analyzed, and FRAP recovery curves were generated using LSM software and Microsoft Excel as previously described (15). Briefly, the fluorescence intensity (I_n) in a region of interest was determined as I_n = (I_t – I_bg)/ (T_t – I_bg) x (T_o – I_bg)/(I_o – I_bg), where T_o is the total cellular intensity during prebleach, T_t is the total cellular intensity at time point t, I_o is the average intensity in the region of interest during prebleach, I_t is the average intensity in the region of interest at time point t, and I_bg is the average intensity in an area outside the monitored cell. All of the quantitative data for FRAP recovery kinetics represent means ± standard errors from at least 25 cells imaged in three independent experiments.

To determine the size of total bound fractions, the FRAP method involving bleaching of half of the nucleus in a cell expressing GFP-AR or GFP-AR-E897A was used as described previously (60). Two single prebleach images were acquired and followed by a brief bleach pulse of 400 ms. The recovery of the fluorescence signal in the bleached region and the loss of signal in the unbleached region were monitored simultaneously by time-lapse microscopy. The fluorescence intensity in a region of interest was normalized to the prebleach fluorescence intensity in the region of interest as R = (I_t – I_bg)/(I_o – I_bg) where I_t is the average intensity in the region of interest at time point t, I_o is the average intensity in the region of interest during prebleach, and I_bg is the average intensity in an area outside the monitored cell. We then experimentally determined the size of total bound fraction of AR and AR-E897A in response to the presence of ligands based on the fact that the diffusion time of AR or AR-E897A in the nucleus is shorter than the bleach time used in the experimental conditions. The fluorescence intensity in the unbleached region before bleaching was compared to the intensity seen immediately after bleaching as previously described (60), and the total chromatin-bound fractions were calculated (see Table 1). In all FRAP experiments, signal loss during the recovery period was less than 5% of the initial fluorescence intensity. The bleach extent and depth were confirmed by analyses of three-dimensional image stacks along the z plane of the image axis of fixed cells. All FRAP recovery curves were generated from background subtracted images. Student's t test was used to determine the statistical significance of results (see Fig. 7H).

View larger version (44K): [in this window] [in a new window]   FIG. 7. Increased mobility of transcriptionally deficient AR mutant at the HRE. (A to E) FRAP analysis of GFP-AR-E897A at the MMTV array. The 3109 cells were treated with R1881 (10–8 M) for 45 min. The cells were imaged before and during recovery after bleaching of GFP-AR-E897A. Images collected at 10 s (C), 20 s (D), and 30 s (E) demonstrate the rapid recovery kinetics of GFP-AR-E897A at the MMTV array. Bar in panel E, 4 µm. (F) Quantitative FRAP analysis of GFP-AR-E897A at the MMTV array. Cells were either left untreated or were treated with 10–8 M R1881, 10–8 M DHT, 10–8 M TST, 10–6 M CPA, or 10–6 M RU486 for 45 min or with 10–6 M bicalutamide (Casodex [CAS]) or 10–6 M OHF for 90 min. For each set of conditions, at least 25 cells from three independent experiments were analyzed. (G) Time for 50% recovery of fluorescence (t1/2) for GFP-AR-E897A on the MMTV array in the presence of the various ligands. The t1/2 values were calculated from the recovery curves (n > 25 for each ligand); bars represent standard errors. (H) Quantitative comparison of FRAP recovery kinetics between wild-type AR and the AR-E897A mutant. (I and J) Quantification of BRM loading and RNA FISH signal in 3109 cells treated with the various ligands. The histogram shows relative BRM and RNA FISH intensities from at least 10 cells for BRM loading and 20 cells for RNA FISH for each ligand condition; bars represent standard errors. Quantitative analysis of data was carried out as described in Materials and Methods. The experiment was performed twice with similar results. *, P < 0.05 compared to the results seen in the absence of ligand, as assessed by Student's t test.

In vitro reconstitution of MMTV chromatin. A 1.1-kb PleI/NcoI fragment of MMTV LTR (positions –437 to +674) was immobilized on Dynal magnetic beads as previously described (56). The immobilized fragment was reconstituted into chromatin by use of preblastoderm embryo extract supplemented with mouse histone octamers. The reconstituted chromatin was then incubated in embryo extract buffer {10 mM HEPES [pH 7.6], 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM EGTA, 10% glycerol, 10 mM ß-glycerophosphate, 1 mM DTT, and 1 mM AEBSF [4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride]}with Sarkosyl (0.05% final concentration) at room temperature for 5 min and washed twice with cold EX-N buffer (10 mM HEPES [pH 7.6], 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM EGTA, 10% glycerol, 10 mM ß-glycerophosphate, 1 mm DTT, 0.05% aprotinin, pepstatin, and leupetin, 2 mg/ml bovine serum albumin [BSA]). Reconstituted chromatin was directly used in pulldown experiments with purified AR.

Chromatin and DNA pulldown assays. Prior to the DNA binding assay, MMTV DNA or chromatin was washed twice with binding buffer (20 mM HEPES [pH 7.3], 50 mM NaCl, 10 mM glycerol, 0.5 mm EDTA, 5 mm MgCl₂, 0.1% NP-40, 1 mM DTT, 1 mM aminoethylbenzenesulphonyl fluoride, 1 µg/ml concentrations each of aprotinin, pepstatin, and leupetin, 2 mg/ml BSA) and poly(dI-dC) (10 µg/µl). Purification of hSwi/Snf complexes was performed as described previously (18) with cells expressing the Flag-tagged INI1 subunit (70). The binding reactions were carried out in 50 µl of binding buffer containing 100 ng MMTV DNA or chromatin with or without 1 nM purified androgen receptor bound to DHT, 23.5 µg of HeLa nuclear extract (N.E.) or purified hSWI/SNF (700 ng), and/or 1 mM ATP. After incubation at 30°C for 15 min, the template was washed twice with binding buffer without BSA or poly(dI-dC) and analyzed by 7.5% SDS-PAGE. Western analysis was performed with antisera for AR (441 [sc-7305]; Santa Cruz Biotechnology).

FRET. Acceptor photobleaching FRET analysis on fixed cells was performed as described previously (35). In the presence of FRET, bleaching of the acceptor yellow fluorescent protein (YFP) results in a significant increase in the fluorescence intensity of donor cyan fluorescent protein (CFP). The 3134 cells were transfected with pECFP-C1, pEYFP-N1, CFP-YFP, pECFP-C1-AR, pEYFP-N1-AR, pEYFP-N1-ECFP-AR, or pEYFP-N1-ECFP-AR-E897A. The cells were fixed in 4% paraformaldehyde, washed with phosphate-buffered saline, and imaged on an Olympus FluoView 1000 confocal laser-scanning microscope equipped with a 100x/1.3 numerical-aperture oil objective. Two prebleach and two postbleach images were captured on CFP and YFP channels. Bleaching was done in the YFP channel using a 515-nm laser line at 2% intensity zoomed at x46. Bleaching due to imaging was minimal, since images were collected at low laser intensity (8% of a 458 nm laser and 2% of a 515 nm laser) and bleaching was monitored by comparison of prebleach and postbleach image pairs. Each image was collected first in the CFP channel and then in YFP channel. No cross-talk was detected between YFP and CFP channels during imaging. Fluorescence intensities in all regions of interest were corrected for background fluorescence, and FRET efficiency was calculated according to the following formula as described previously (35): E_F = (I_post – I_pre)/I_post, where I_post is CFP intensity after bleaching and I_pre is CFP intensity before bleaching.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

 
Ligand-dependent transcriptional activity and intracellular distribution of GFP-AR in vivo. To investigate the dynamics of AR on a natural promoter in live cells, we established a stable cell line expressing a GFP fusion of AR under the control of a tetracycline-repressible promoter in the 3134 mammary adenocarcinoma cell line (see Materials and Methods). This cell line contains approximately 200 copies of an MMTV Ras tandem array and about 800 to 1,000 binding sites for GR or PR that enable the direct visualization of GFP-GR and GFP-PR binding to the MMTV promoter in living cells (49, 62). The array in these cell lines exists as a unique repetitive element integrated at a single site in chromosome 4. Development of cell systems with stably integrated arrays of DNA segments containing multiple binding sites allows direct visualization of gene expression events in real time in living cells (33, 49, 50, 75, 83). We took advantage of the fact that AR also binds and activates transcription from the HREs at the MMTV LTR in a ligand-dependent manner. The hormone response of promoters within this array is indistinguishable from that of a single-copy gene, and the chromatin organization of the promoter has been characterized in detail previously (19). Thus, this is an ideal model system for examining the dynamic interactions between a steroid hormone receptor and its target promoter in vivo.

We first developed an inducible cell line in which GFP-AR expression is regulated by tetracycline and confirmed this by Western analysis of total cell lysates by using antibodies against both GFP and AR. As shown in Fig. 1A, there was robust induction of GFP-AR expression resulting in a band of expected size (130 kDa) upon removal of tetracycline from the medium. This cell line was named 3108. The AR antiserum also detected an additional band at around 90 kDa (lane 2) which might represent a proteolytic AR fragment. As this band is much less abundant than the full-length GFP-AR (ratio, 5:1), the nature of this band was not investigated further.

View larger version (39K): [in this window] [in a new window]   FIG. 1. Functional characterization and intracellular localization of GFP-AR. (A) Tetracycline-regulated expression of GFP-AR. Total cell extracts were prepared from GFP-AR-expressing 3108 cells in the presence (lanes 1 and 3) or absence (lanes 2 and 4) of tetracycline. The extracts were probed for GFP-AR by using anti-AR (lanes 1 and 2) or anti-GFP (lanes 3 and 4) antibodies by Western blotting. ß-Actin was used as a loading control (lower panels). (B and C) GFP-AR maintained wild-type response to various AR ligands. 3108 cells were transfected with MMTV-LUC (B) or -285PB-LUC (C) constructs and treated with R1881 (10–8 M), DHT (10–8 M), TST (10–8 M), CPA (10–6 M), RU486 (10–6 M), bicalutamide (10–6 M), or OHF (10–6 M) for 24 h. The cells were harvested, and reporter gene activity was assayed and normalized to total protein amount. The results represent the averages of the results of three replicate experiments, with bars representing standard deviations. The experiments were repeated twice with similar results. (D to K) GFP-AR translocates to the nucleus in response to AR ligands. 3108 cells were treated with ligands as described for panels B and C, and the translocation of GFP-AR was followed by in vivo time-lapse confocal microscopy at 37°C. GFP-AR was completely translocated to the nucleus after 30 min in response to R1881 (E), DHT (F), TST (G), and RU486 (I) and after 120 min in response to CPA (H), bicalutamide (Casodex [CAS]) (J), and OHF (K). No translocation was observed in the absence of ligand (D). Bar in panel K, 4 µm.

It is known that upon ligand binding AR translocates to the nucleus (3, 23, 76, 77). However, most studies have been performed with transiently expressed AR and with only a small set of ligands. Furthermore, we wanted to determine the functionality and the kinetics of GFP-AR nuclear translocation in our inducible cell system. We therefore carried out a detailed time-lapse analysis of GFP-AR translocation in 3108 cells in response to agonists, partial antagonists, and pure antagonists (Fig. 1D to K). All seven ligands caused nuclear translocation of the receptor but with significantly different temporal kinetics (Fig. 1; also see Fig. S1 posted at http://www.imbv.uio.no/gen/groups/fs/AR Dynamics). In the presence of agonists R1881, DHT, and TST (Fig. 1E to G) as well as that of the partial antagonist RU486 (Fig. 1I), the translocation was rapid, with predominantly nuclear localization of GFP-AR within 30 min of ligand addition. In the presence of CPA, bicalutamide, and OHF (Fig. 1H, J, and K) the translocation was significantly slower, and predominantly nuclear localization of GFP-AR was not observed until at least 120 min after ligand addition. The translocation dynamics for all ligands was similar to what was observed for endogenous AR in LNCaP cells (androgen-responsive prostate cancer cell line) as detected by indirect immunofluorescence (reference 77; also see Fig. S2 posted at http://www.imbv.uio.no/gen/groups/fs/AR Dynamics). There were also differences in the distribution of GFP-AR in the nucleus induced by the different ligands (see Fig. S3 posted at http://www.imbv.uio.no/gen/groups/fs/AR Dynamics): while the agonist-bound GFP-AR distributed in multiple bright foci, the antagonist-bound AR showed a diffuse homogenous distribution in the nucleus. Cells treated with the partial antagonists CPA and RU486 had an intermediate pattern, with visibly less formation of foci than the agonists. These data are consistent with previous studies (3, 76, 77). The functional relevance of this heterogeneous distribution induced by the different ligands is currently unclear. Collectively, these results demonstrate that the GFP-AR was functional and maintained the same general mobility properties as endogenous AR.

Ligand-dependent recruitment of GFP-AR to HREs in vivo. In order to examine the ligand-dependent recruitment of GFP-AR to the MMTV promoter in vivo, RNA FISH analysis was performed combined with indirect immunofluorescence microscopy. In the absence of ligand, GFP-AR was mainly distributed in the cytoplasm and no significant binding to the MMTV array was observed (Fig. 2A At the array, there was a low level of basal MMTV transcription, and GFP-AR fluorescence intensity peaks and nascent MMTV transcripts did not coincide (Fig. 2A₃ and 2A₄). In cells treated with agonists R1881, DHT, and TST, a single bright GFP fluorescence signal was detected within the nucleus in addition to the diffuse nucleoplasmic GFP-AR signal (Fig. 2B to D). Overlay of the MMTV RNA FISH and the GFP-AR immunofluorescence data confirmed ligand-dependent recruitment of GFP-AR to the MMTV promoter.

View larger version (36K): [in this window] [in a new window]   FIG. 2. Recruitment of AR to the MMTV array and effect of AR ligands on MMTV transcription in vivo. (A to H) GFP-AR colocalizes with nascent MMTV transcripts, as detected by RNA FISH and indirect immunofluorescence microscopy. The 3108 cells expressing GFP-AR were treated with R1881 (10–8 M), DHT (10–8 M), TST (10–8 M), or RU486 (10–6 M) for 45 min and with CPA (10–6 M), bicalutamide (Casodex [CAS]) (10–6 M), or OHF (10–6 M) for 90 min and were subjected to RNA FISH and immunofluorescence microscopy as described in Materials and Methods. Anti-GFP antibody staining is shown in column 1 (panels A1 to H1), and RNA FISH staining is shown in column 2 (panels A2 to H2). Column 3 (panels A3 to H3) shows an overlay of both signals for each ligand. Colocalization of GFP-AR and RNA FISH signals on the MMTV array is indicated by yellow and was observed for all ligands. The arrows in the overlay images point to the position of line scans. In the line scans in panels B4 to H4, but not in panel A4, the fluorescence intensity peaks for GFP-AR and MMTV transcript coincided, indicating the recruitment of AR to MMTV promoter. (I) Quantification of MMTV RNA transcripts in 3108 cells. The cells were treated with the various AR ligands as described above. The histogram shows the relative RNA FISH intensities for at least 30 cells for each ligand condition, with error bars representing standard errors. Quantitative analysis of RNA FISH data was carried out as described in Materials and Methods. The experiment was performed three times with similar results. Bar in panel H, 4 µm.

Treatment of the cells with the pure antagonists bicalutamide and OHF also resulted in the recruitment of GFP-AR to the MMTV promoter (Fig. 2G to H; also see Fig. S4 posted at http://www.imbv.uio.no/gen/groups/fs/AR Dynamics), which was also confirmed by line scan analysis (Fig. 2G₄ and H₄). Interestingly, the size of GFP-AR fluorescence signal at the array with pure antagonists was smaller than that of the signal seen with pure agonists and partial antagonists, suggesting a decrease in the loading of AR to its binding site in vivo (see Fig. S4 posted at http://www.imbv.uio.no/gen/groups/fs/AR Dynamics). Quantitative RNA FISH analysis further revealed that under these conditions there were only basal levels of MMTV transcription (Fig. 2I). These data demonstrate that all ligands tested recruit GFP-AR to the MMTV promoter but that only the pure agonists and the partial antagonist CPA and, to some extent, RU486 induce transcription in vivo.

Effect of ligands on the nuclear mobility of AR at HREs in living cells. Recruitment of GFP-AR to the MMTV promoter by the various AR ligands enabled us to dissect the kinetic properties of GFP-AR bound to its target site in live cells. To that end, we used the photobleaching technique FRAP. As FRAP kinetics reflects the overall mobility of a protein (60), it can be used to quantitatively measure the kinetics of binding of proteins to chromatin in living cells (49).

To study the binding kinetics of AR at the HREs in vivo, GFP-AR expression was induced in 3108 cells by removal of tetracycline and the cells were treated with the different ligands. GFP-AR bound to the MMTV array was bleached using a brief laser pulse. The recovery of GFP-AR fluorescence signal in the bleached region was monitored using in vivo time-lapse microscopy. The recovery of R1881-bound GFP-AR was very fast, reaching 80% of prebleached levels within 50 s (Fig. 3A to F), and complete recovery was observed around 150 s (data not shown). In the absence of hormone, GFP-AR showed the fastest recovery kinetics, representing the freely diffusing pool of the receptor in the nucleus (Fig. 3F). The AR recovery kinetics was significantly slower when the GFP-AR was bound to agonists compared to the results seen with the pure antagonists. In the presence of R1881, DHT, and TST, the times required to reach half-maximal recovery, the t_1/2, were 3.6, 5.3, and 5.0 s, respectively (Fig. 3G). Interestingly, the t_1/2 for unliganded AR (0.2 s) was highly similar to the t_1/2 for bicalutamide (0.5 s) and OHF (0.5 s), suggesting that the interaction of antagonist-bound AR with a genomic target is very transient (Fig. 3G). The partial antagonists CPA and RU486 showed a somewhat different pattern. The CPA recovery curve lay between those for agonists and antagonists, with a t_1/2 of 1.1 s. RU486, however, showed a recovery that is similar to the agonists, with a t_1/2 of 4.3 s. Under all conditions, the recovery time was still very rapid, indicating a transient interaction between AR and its target HREs. Compared to PR (62) and GR (49), AR had a slower recovery kinetics, suggesting that there are differences in the mechanism of receptor-promoter interactions for AR compared to GR and PR.

View larger version (27K): [in this window] [in a new window]   FIG. 3. Effect of ligands on the exchange of AR at its cognate binding site. (A to E) FRAP analysis of GFP-AR at the MMTV array. The 3108 cells were treated with R1881 (10–8 M) for 45 min. The cells were imaged before and during recovery after bleaching of GFP-AR. Images collected at 10 s (C), 30 s (D), and 60 s (E) demonstrate the rapid recovery kinetics of GFP-AR at the MMTV array. Bar in panel E, 4 µm. (F) Quantitative FRAP analysis of GFP-AR at the MMTV array. Cells were either left untreated or were treated with 10–8 M R1881, 10–8 M DHT, 10–8 M TST, or 10–6 M RU486 for 45 min and with 10–6 M CPA, 10–6 M bicalutamide (Casodex [CAS]), or 10–6 M OHF for 90 min. For each set of conditions, at least 25 cells from three independent experiments were analyzed. (G) Time for 50% recovery of fluorescence (t1/2) for GFP-AR on the MMTV array in the presence of the various ligands. The t1/2 values were calculated from the recovery curves (n > 25 for each ligand condition); error bars represent standard errors.

View larger version (42K): [in this window] [in a new window]   FIG. 4. Ligand-dependent recruitment of BRM chromatin-remodeling complex by AR to the MMTV array. (A to T) Endogenous BRM colocalizes with nascent MMTV transcripts, as detected by RNA FISH and indirect immunofluorescence microscopy. 3134 cells were left untreated (A to D) or treated with 10–8 M R1881 (E to H) or 10–6 M RU486 (M to P) for 45 min or with 10–6 M CPA (I to L) or 10–6 M OHF (Q to T) for 90 min. The cells were subjected to RNA FISH and immunofluorescence microscopy as described in Materials and Methods. Anti-BRM antibody staining is shown in column 1 (panels A, E, I, M, and Q), and RNA FISH staining is shown in column 2 (panels B, F, J, N, and R). Column 3 (panels C, G, K, O, and S) shows an overlay of both signals for each ligand. Significant colocalization of endogenous BRM and RNA FISH signals on the MMTV array is indicated by yellow and was observed only for R1881 and CPA. The arrows in the overlay images point to the positions of line scans. In the line scans in panels H and L, but not those in panels D, P, and T, the fluorescence intensity peaks for BRM and MMTV transcript coincided, indicating the recruitment of BRM to the MMTV promoter. (U and V) Quantification of BRM loading and the RNA FISH signal in 3134 cells treated with the various AR ligands. The histogram shows relative BRM (U) and RNA FISH (V) intensities from at least 20 cells for each ligand condition; bars represent standard errors. Quantitative analysis of data was carried out as described in Materials and Methods. The experiment was performed twice with similar results. Bar in panel S, 4 µm. *, P < 0.05 compared to the results seen in the absence of ligand, as assessed by Student's t test.

View larger version (11K): [in this window] [in a new window]   FIG. 5. Energy-dependent displacement of AR from its binding site during chromatin remodeling in vitro. (A) Diagram for chromatin pulldown assay using chromatin reconstituted onto MMTV DNA attached to streptavidin-coated magnetic beads. (B) A MMTV LTR DNA fragment was reconstituted into chromatin using purified histones. A DNA pulldown assay was carried out by incubation of purified AR-DHT complex with MMTV chromatin (lanes 3 to 8) or naked MMTV DNA (lanes 9 to 14) in the presence of ATP, purified Swi/Snf, or HeLa nuclear extract as indicated. Associated proteins were analyzed by SDS-PAGE and Western blotting using primary antibody against AR.

View larger version (46K): [in this window] [in a new window]   FIG. 6. Functional characterization and intracellular localization of mutant AR with compromised transcriptional activity. (A) Tetracycline (Tet)-regulated expression of GFP-AR-E897A. Total cell extracts were prepared from GFP-AR-expressing 3108 cells (lanes 1 and 2) and GFP-AR-E897A-expressing 3109 cells (lanes 3 and 4) in the presence (lanes 1 and 3) or absence (lanes 2 and 4) of tetracycline. The extracts were probed with an anti-AR antibody by Western blotting. ß-Tubulin was used as a loading control (lower panel). (B and C) GFP-AR-E897A is deficient in transcriptional activity in response to AR ligands. 3109 cells were transfected with MMTV-LUC (B) or -285PB-LUC (C) constructs and treated with R1881 (10–8 M), DHT (10–8 M), TST (10–8 M), CPA (10–6 M), RU486 (10–6 M), bicalutamide (Casodex [CAS]) (10–6 M), or OHF (10–6 M) for 24 h. The cells were harvested, and reporter gene activity was assayed and normalized to total protein amount. The results represent the averages of three replicate experiments, with bars representing standard deviations. These experiments were performed twice with similar results. (D to K) GFP-AR-E897A translocates to the nucleus in response to AR ligands. 3109 cells were treated with ligands at concentrations as described for panels B and C, and the translocation of GFP-AR-E897A was followed by in vivo time-lapse confocal microscopy. GFP-AR-E897A was predominantly localized to the nucleus after 30 min in response to the presence of R1881 (E), DHT (F), TST (G), and RU486 (I), 45 min in response to CPA (H), and 120 min in response to bicalutamide (Casodex [CAS]) (J) and OHF (K). No translocation was observed in the absence of ligand (D). Bar in panel K, 4 µm.

To assess whether AR-E897A has defects in its kinetic and binding properties for targeting genes in vivo, we performed FRAP analysis using the 3109 cells. GFP-AR-E897A was efficiently recruited to the MMTV promoter by all ligands, enabling us to perform FRAP analysis of array-bound receptor in the presence of each ligand (Fig. 7A to F). Quantitative FRAP analysis showed fastest recovery of GFP-AR-E897A in the absence of hormone and in the presence of pure antagonists bicalutamide and OHF (Fig. 7F). The t_1/2 of unliganded GFP-AR-E897A was 0.3 s, and in the presence of bicalutamide and OHF it was 0.2 s (Fig. 7G). In similarity to the wild-type GFP-AR results, FRAP recovery kinetics of GFP-AR-E897A were significantly slower in the presence of pure agonists: the t_1/2 in the presence of R1881, TST, and DHT were 4.6 s, 2.8 s, and 3.7 s, respectively (Fig. 7G). In the presence of partial agonist CPA, GFP-AR-E897A showed faster recovery kinetics compared with that seen in the presence of RU486, with t_1/2 of 0.5 s and 3.1 s, respectively (Fig. 7G). Importantly, for all ligands tested, the recovery kinetics of GFP-AR-E897A at the HRE was faster than that seen with the wild-type GFP-AR (P < 0.001) (comparison shown for R1881, CPA, and OHF in Fig. 7H). Differences in recovery kinetics between wild-type AR and mutant AR-E897A cannot be explained by differences in molecular weight, since the mutant AR-E897A contains a single amino acid substitution rather than a deletion. Thus, although there is recruitment and dynamic exchange of GFP-AR-E897A with the HREs at the MMTV promoter, the kinetics of this interaction is significantly faster than that seen with wild-type GFP-AR. These data demonstrate that AR mobility is not only affected by the nature of the ligand it is bound to but is also linked to its function, i.e., transactivation potential.

Given that antagonist-bound wild-type AR does not recruit BRM to the array and cannot activate transcription (Fig. 4 and 2I), we assessed whether the transcriptionally deficient AR-E897A could do so. The 3109 cells stably expressing GFP-AR-E897A were treated with the various ligands as before and subjected to RNA FISH analysis and indirect immunofluorescence with BRM antibody. RNA FISH analysis revealed significantly impaired agonist-induced transcriptional activity of the mutant AR on the MMTV array compared to wild-type AR results (Fig. 7I; compare with Fig. 2I); this is consistent with the reporter assay results presented in Fig. 6B and C, where AR-E897A displays significantly impaired activity. In keeping with the involvement of chromatin-remodeling complexes for transcriptionally active promoters, GFP-AR-E897A had reduced recruitment of BRM to the MMTV array in the presence of agonist R1881 compared with the wild-type receptor results (Fig. 7J; compare to Fig. 4U). These data further support the idea of the involvement of chromatin-remodeling complexes in transactivation by AR at its response elements during transcription.

Kinetic modeling of AR interaction with HREs in vivo. In order to extract specific quantitative information from the FRAP experiments described above, we used computational kinetic modeling methods (59, 60). Experimental FRAP recovery data for AR and AR-E897A from Fig. 3 and 7 were fitted using a generalized least-squares and classical compartmental approach as described in Materials and Methods. The recovery kinetics of AR and AR-E897A was most accurately fitted by a two-site binding model with a statistically significant coefficient of variation indicating that both receptors were present at the HREs in at least two distinct kinetic populations with distinct binding kinetics (Fig. 8A to F). For simplicity, we named these two populations the "fast" and "slow" fractions. Kinetic properties of AR and AR-E897A on the MMTV array, namely, the off rates, mean residence times, total chromatin-bound fractions, and the numbers and sizes of kinetically distinct receptor populations, were calculated and are included in Table 1. The statistical significance of kinetic parameters is shown in Table2.

View larger version (33K): [in this window] [in a new window]   FIG. 8. Kinetic modeling of AR and AR-E897A interaction with the HRE. The experimental FRAP recovery data for AR and AR-E897A presented in Fig. 3F and 7F were fitted using least-square best fit and classical compartmental approach as described in Materials and Methods. Two-site binding model gave the best fit for both AR and AR-E897A, with statistically significant coefficient-of-variation values. Within the time scale of FRAP experiments, protein synthesis and protein degradation were presumed negligible. For Table 1, kinetic properties of AR and AR-E897A at the MMTV array in response to various ligands were calculated from the fitted data at Fig. 3F and 7F. Total bound fractions of AR and AR-E897A in the nucleoplasm were determined as described in Materials and Methods. For Table 2, the statistical significance of kinetic modeling parameters was expressed as coefficient of variation-percent error relative to average value. Note that the coefficient-of-variation values of koff are identical to the coefficient-of-variation values of residence times, since the residence time = 1/koff. The statistical significance of total bound fractions is expressed as a standard deviation.

We have also calculated kinetic properties of GFP-AR-E897A, a transcriptionally deficient AF-2 mutant of AR. As shown in Table 1, the mean residence time of agonist-bound AR-E897A (R1881 or DHT) at the HREs was significantly longer than that seen with pure antagonist-bound AR (OHF) in both the slow and fast fractions. Even though the overall trends for AR and AR-E897A were similar, there were significant differences. For example, the mean residence time of AR-E897A at the HREs in the slow fraction was significantly shorter than the mean residence time of wild-type AR under all ligand treatment conditions. Similarity of overall mobility trend in response to various ligands between the two receptors could be due to residual transactivation function of AR-E897A (Fig. 6B and C). In summary, transcriptionally deficient AR-E897A has a larger fast fraction size and a smaller slow fraction size at the HREs and it also has a shorter residence time in the slow fraction than that seen with transcriptionally competent wild-type AR at the HREs (Table 1). Finally, we experimentally determined the size of the total bound fraction of AR and AR-E897A in the nucleoplasm in response to the presence of ligands as described in Materials and Methods. More than 92% of both AR and AR-E897A proteins are bound to chromatin at steady state in the nucleoplasm in the presence of various ligands, suggesting that transient DNA binding is a common property of AR and AR-E897A. Error margins and coefficient-of-variance values for all kinetic modeling parameters were less than 10% of the measured values (Table 2).

Wild-type AR, but not an AF-2 mutant, recruits RNA PolII to the target HREs. The data obtained so far showed a tight connection between transactivation potential, recruitment of chromatin-remodeling complexes, and actual transcription in situ. In order to further examine this connection, we assessed the recruitment of the transcriptional apparatus to the HRE array by wild-type and mutant AR. To this end, we used confocal immunofluorescence microscopy and RNA PolII antiserum with ligand-treated 3108 and 3109 cells. As shown in Fig. 9, whereas wild-type AR significantly overlapped with PolII on the array (Fig. 9A), AR-E897A did not (Fig. 9B). This suggests that mutant AR cannot recruit the transcriptional apparatus to the array and therefore cannot induce transcription, resulting in a less-engaged receptor with faster recovery kinetics.

View larger version (66K): [in this window] [in a new window]   FIG. 9. Transcriptionally impaired mutant AR does not recruit RNA PolII to the MMTV array. Endogenous RNA PolII colocalizes with agonist-bound AR at the MMTV array but not with agonist-bound mutant AR. 3108 (A) and 3109 (B) cells were left untreated (data not shown) or were treated with 10–8 M R1881 for 45 min or 10–6 M CPA for 90 min, as indicated, and subjected to immunofluorescence analysis with PolII antibody. Column 1 shows GFP-AR or GFP-AR-E897A expression, column 2 shows staining for PolII, and column 3 shows the overlay of the two images. Colocalization of endogenous PolII and GFP signals on the MMTV array is indicated by yellow in the enlarged square and was observed only for ligand-bound wild-type AR (A) and not for mutant AR (B). For GFP-AR (A), but not GFP-AR-E897A (B), the line scans (column 4) show that fluorescence intensity peaks for GFP and PolII coincided, indicating the recruitment of PolII to the MMTV promoter by wild-type AR and not by mutant AR. Bars, 4 µm.

We generated AR fusion proteins with YFP or CFP at the N terminus or the C terminus or at both termini of wild-type AR or mutant AR-E897A. All of these fusion constructs were first tested for their ability to bind ligand and activate transcription by using the transient transfection assay and the -285PB-LUC reporter. As shown in Fig. 10A, all fusion proteins activated transcription in response to R1881, albeit to a lesser extent for some constructs. Importantly, the doubly labeled ARs with YFP or CFP fused at either end were robust activators of -285PB-LUC, and the AR-E897A mutant fusions were significantly compromised in their transactivation potential compared with corresponding fusions with wild-type AR.

View larger version (25K): [in this window] [in a new window]   FIG. 10. FRET shows predominantly intramolecular interactions of AR at the HRE. (A) AR fusion proteins induce transcription in a ligand-dependent manner. 3134 cells were transiently transfected with -285PB-LUC and the indicated AR constructs. The cells were left untreated or were treated with R1881 (10–8 M) for 24 h. The cell extracts were made, assayed for luciferase activity, and normalized to total protein amount. The results are presented as severalfold change relative to untreated cell results and represent the averages of three replicate experiments, with bars representing standard deviations. The experiments were repeated twice with similar results. (B to E) FRET analysis of 3134 cells expressing CFP-AR-YFP. The cells were treated with R1881 (10–8 M) for 45 min, and acceptor photobleaching FRET was performed at CFP-AR-YFP bound to the MMTV array as described in Materials and Methods. The cells were imaged on the CFP channel and YFP channel before bleaching (B and C) and after bleaching (D and E). The areas marked by a rectangle are enlarged and shown as pseudocolored insets in panels B and D. Bleaching of acceptor YFP (E) resulted in an increase in donor CFP fluorescence (D). Note the increase in AR fluorescence intensity at the array in the pseudocolored inset in panel D compared to that shown in the inset in panel B. Bar in panel E, 4 µm. (G) Quantification of FRET analysis. The 3134 cells expressing indicated fusion proteins were treated with R1881 (10–8 M) for 45 min. Acceptor photobleaching FRET was then performed on the indicated AR fusion proteins, or their combination, bound to the MMTV array. FRET efficiencies were calculated for each condition. The results represent the averages obtained for at least 15 cells for each set of condition, with error bars representing standard errors. *indicates statistical significance of the difference (P < 0.00019) as calculated by Student's t test. (G) Cartoon showing a model for intramolecular interactions in ligand-bound AR molecules creating FRET.

The 3134 cells expressing CFP-AR-YFP were treated with R1881 for 45 min, and YFP acceptor fluorescence was bleached at the MMTV array. The fluorescent signal of CFP and YFP was monitored before (Fig. 10B and C) and after (Fig. 10D and E) bleaching. Upon bleaching, an 18% increase in the donor CFP fluorescence was observed, demonstrating an intramolecular interaction between the N and C termini of AR (Fig. 10F). Acceptor photobleaching FRET of CFP-AR-E897A-YFP at the HREs resulted in approximately 12% FRET efficiency, indicating that the intramolecular interactions in the mutant AR-E897A were significantly impaired compared with wild-type AR results (Fig. 10F).

Coexpression of CFP-AR and AR-YFP produced 3% FRET efficiency at the HREs, indicating that there are intermolecular interactions between AR molecules that are bound to the same promoter, albeit these are significantly less strong than the intramolecular interactions between the N and C termini. Mutation of the AF-2 domain did not alter the magnitude of the intermolecular FRET, i.e., that observed with CFP-AR-E897A and AR-E897A-YFP, suggesting that these interactions may not be directly involved in transcriptional activation. Single-color fusion proteins by themselves, or nonfusion CFP and YFP expressed together, did not display any significant FRET. Taken together, these data suggest that intramolecular interactions play an important role in transactivation by AR.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

 
The ability to detect molecular interactions between AR and its response elements in living cells offers the opportunity to study the mechanisms of gene expression by AR in real time. This study is the first to show targeted recruitment of AR to its binding site within the context of native chromosomes in response to a range of ligands. We have found a rapid and transient interaction of AR with hormone response elements both in vivo and in vitro, supporting the "hit-and-run" model of nuclear receptor action (18, 49, 57, 65). These observations suggest that rapid exchange is a common feature of nuclear receptor action at regulatory elements.

The dynamic interaction of AR with its specific target sites in chromatin is strongly ligand dependent, as is AR's ability to recruit chromatin-remodeling complexes to its target sites. Based on these observations, it is clear that the nature of the ligand not only determines the type of coregulators that are recruited by AR through induction of different conformational changes in the LBD but also significantly affects the dynamics of AR interactions with chromatin and the chromatin-remodeling complexes, resulting in differential effects on AR function on the target gene in vivo. This is in line with modulation of dynamic interactions of a variety of factors with chromatin (1, 4, 43, 46, 58, 67, 75).

Ligand-specific dynamics of AR on HREs. Previous studies of GFP-AR using FRAP found differences in the mobility of unliganded and antagonist-bound AR compared with that of agonist-bound AR in the general nuclear space (16, 17). Here, we have studied the dynamics of AR-chromatin interactions quantitatively when AR is bound and transcriptionally active at its target HRE. There is a significant difference between the dynamics of AR-chromatin associations in the presence of an agonist and that of an antagonist on the HRE array. For example, the t_1/2 of recovery in FRAP analysis on the array is approximately sixfold lower with the AR-bicalutamide complex than with AR-DHT (Fig. 3), which correlates with the transcriptional activity elicited by these ligands (Fig. 2). Kinetic modeling of the FRAP data indicates that there is a significant increase in the rate at which the receptor dissociates from the template (k_off) and a marked decrease in the HRE residence time of AR in the presence of antagonists compared with the results seen when it is bound to agonists (Table 1). These data provide important new insight into the mechanism of action of potent nonsteroidal antiandrogenic drugs: they competitively bind AR and prevent its residence on chromatin long enough to support transcriptional activation.

While assessing AR transcriptional activity by different ligands, we found that whereas all other ligands acted as expected, CPA, which was one of the first antiandrogens used in prostate cancer therapy (10), acted as an agonist in reporter assays as well as in RNA FISH analysis. As CPA is a synthetic derivative of hydroxyprogesterone, it also has progestational and antigonadotrophic properties, and it has also been shown to have agonistic properties in other cell lines (2, 12). The other partial antagonist, RU486, mainly showed antagonistic properties, which is in accordance with previous reports on RU486 function (28, 72). However, although RU486 induced transcription significantly less than the pure agonists, the FRAP recovery curve for RU486 was similar to that of the agonists (Fig. 3). Therefore, the mechanism of antiandrogen action of RU486 seems to be different from that of bicalutamide and OHF, which show a significantly faster recovery at the array. One possible explanation is that the RU486-AR complex associates more strongly with chromatin than the antagonist-AR complex but is not able to recruit the cofactors necessary for transcriptional activation. Consistent with this, RU486-bound AR cannot recruit BRM to the array in the same manner as the agonist-bound AR (Fig. 4). The reason for the delayed recovery of RU486-bound AR compared to antagonist-bound AR needs to be elucidated further.

Another surprising finding concerning the activities of ligands was determined for CPA and RU486 when they were bound to the AF-2 mutant AR-E897A. For one of the reporter constructs (-285-PB-LUC), CPA and RU486 had significant agonist activity comparable to or better than that seen with the agonists (Fig. 6C). These data indicate that partial antagonists can act as agonists, depending on the different contexts, and that this action is at least in part mediated by the nature of the hormone response element and the particular mutation in AR.

What could be the basis for significantly increased mobility of AR when it is bound to antagonists compared with the mobility seen when it is bound to agonists? In some systems, DNA binding is equally avid for AR-DHT and AR-bicalutamide complexes (see, for example, references 47 and 81). This suggests that other proteins that differentially interact with AR in chromatin in the presence of agonists compared with antagonists may be the key determinants in this regard. Another possibility is that AR N-terminal and C-terminal interactions may be involved in stabilizing the interactions with DNA (see below). This may in fact be linked to cofactor interactions, as some coactivators promote whereas some corepressors appear to repress AR N-terminal and C-terminal interactions (44). Further work is necessary to assess these possibilities.

Both our in vivo and in vitro observations showed the transient dynamic exchange of AR on its binding site and existence of a large population of bound AR molecules at steady state throughout the nucleoplasm; this supports the three-dimensional genome-scanning model for chromatin-associated proteins (60). In this model, a large population of bound molecules in the nucleoplasm at a steady state continuously samples the genome by temporary diffusional association and dissociation in order to find their binding sites. This mode of nuclear protein action has been suggested as one of the means of ensuring the availability and targeting of chromatin-associated proteins to their binding sites (59, 60).

Ligand-dependent recruitment of chromatin-remodeling complex by AR. Chromatin-remodeling complexes are involved in gene activation by several members of the nuclear receptor superfamily (see, for example, references 9, 11, 22, 54, and 64). The involvement of the chromatin-remodeling complex Swi/Snf in AR function was demonstrated by ligand-specific recruitment of the ATPase BRM by AR to the MMTV promoter (Fig. 4). Agonists and partial antagonist CPA induced the recruitment of BRM to the MMTV array, while no specific recruitment was observed for RU486 and the pure antagonists OHF and bicalutamide (Fig. 4U). This suggests that when it is agonist bound, AR recruits BRM to the array to induce chromatin remodeling and that this results in longer residence time of the receptor on the template, leading to transcriptional activation, as shown by quantitative RNA FISH, FRAP analysis, and computational kinetic modeling (Fig. 2I, 3F, and 8). In addition, active displacement of AR from MMTV chromatin, but not from naked MMTV DNA, was observed during chromatin remodeling in vitro (Fig. 5B), demonstrating the existence of dynamic interactions between the receptor and its template in vitro as well.

In addition to the ligand-dependent recruitment of BRM to the array, the wild-type AR, but not an AF-2 mutant, recruited the transcription initiation complex to the MMTV promoter indicated by the PolII recruitment. The reporter assays (Fig. 6B and C) and the RNA FISH analysis (Fig. 7I) showed that the mutant AR had retained some activity in presence of agonists, albeit extremely lo