![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Previous Article | Next Article 
Molecular and Cellular Biology, May 2007, p. 3390-3404, Vol. 27, No. 9
0270-7306/07/$08.00+0 doi:10.1128/MCB.02411-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Center for Cell Signaling,1 Department of Biochemistry and Molecular Genetics,2 Department of Microbiology, School of Medicine, University of Virginia, Charlottesville, Virginia3
Received 23 December 2006/ Accepted 14 February 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In addition to ligand binding, nuclear receptors can be regulated by signal transduction pathways. Kinases including those controlled by growth factor-dependent pathways act directly or indirectly on a variety of nuclear receptors (37). Kinases reported to regulate androgen receptor (AR)-dependent transcription include the mitogen-activated protein kinases (p42/44, p38, and Jun N-terminal protein kinase), protein kinase A, and protein kinase C (8). The mitogen-activated protein kinases p38 and Jun N-terminal protein kinase also regulate the nucleocytoplasmic distribution of AR (15). Determining exactly how kinases regulate nuclear receptor transcription activity has been challenging because of cross talk between signaling pathways and because gene expression pathways contain numerous potential kinase targets (25). Perhaps the best-characterized example in this regard is phospho-regulation of the estrogen receptor (ER). Direct phosphorylation of ER provides the basis for an estrogen-independent pathway that promotes coactivator recruitment (18, 41).
Information addressing how dephosphorylation contributes to nuclear receptor regulation is limited. Experiments with the black sponge-derived toxin okadaic acid (OA) have implicated phosphatases in the control of glucocorticoid receptor (GR) localization and activity (11, 39). The phosphatases targeted by OA in these experiments include protein phosphatase 2A (PP2A) and PP5. Additionally, it has been shown that knock-down of PP5 protein levels promotes the cytoplasmic distribution of GR (9), suggesting that nuclear transport of GR is regulated by PP5. Despite these and other intriguing observations linking phosphatases to nuclear receptor function, evidence that a phosphatase directly dephosphorylates a nuclear receptor and that dephosphorylation is causal to a change in transcription activity has not been unequivocally demonstrated.
PP2A is one of the most abundant phosphatases in mammalian cells. The PP2A core enzyme is a heterodimer that contains a catalytic C subunit and a structural A subunit. The PP2A A/C heterodimer is targeted to protein substrates by a third subunit known as a B subunit (21). We recently reported that phosphatase PP2A is a component of AR complexes isolated from simian virus 40 (SV40)-transformed cells (48). Unexpectedly, PP2A binding to AR occurs through an atypical substrate targeting reaction in that it does not involve a PP2A B subunit. Rather, PP2A targeting in this system relies on A-subunit contact with AR, an interaction that is induced by SV40 small t antigen (ST). ST is known to bind directly to HEAT (for Huntington, elongation factor, A subunit, TOR) repeats (1) in the A subunit, and ST binding causes displacement of B subunits and reduces PP2A catalytic activity (32, 35). PP2A, once transferred from ST to AR, is catalytically active. Phospho-site-specific antibodies were used to show that PP2A dephosphorylates five phosphoserines in the AF-1 region of AR (48). ST expression reduces AR transactivation of luciferase-based reporter genes, but it increases ER-dependent transcription (48). These data suggested that either AR or a component of the AR transcription complex is a PP2A substrate that, when dephosphorylated, results in a reduction of transcription.
In the present study we describe the mechanism of ST-dependent PP2A transfer onto AR. We present evidence that ST binding induces a conformational change that is transduced 
76 Å through tandem HEAT repeats in the PP2A A subunit. The conformational changes generate an AR binding site in C-terminal HEAT repeats in the PP2A A subunit. A remarkable feature of the ST-induced conformational change in the PP2A A subunit is that it discriminates between different ligand-bound forms of AR. Thus, the A subunit can function as a targeting subunit for the PP2A A/C heterodimer, and HEAT repeats in the PP2A A subunit can adopt a structure that is capable of recognizing relatively subtle differences in substrate structure. Recent crystallographic analysis of PP2A has revealed a flexible region in the C terminus that includes HEAT repeats 13 to 15 (5, 17, 46, 47). We found that deletion of HEAT repeat 13 is sufficient to abolish binding to AR. In the course of analyzing the AR domains that mediate PP2A interactions, we made several observations that address the more general question of how ligand binding increases the level of nuclear receptor phosphorylation. It has generally been assumed that ligand binding to nuclear receptors increases phospho-site availability to kinases. We show, however, that kinases can access phospho-sites in AR prior to androgen addition. Phosphatase access and/or targeting to AR was found to be dependent on the LBD. Our data suggest that ligand binding increases the phosphorylation state of multiple AF-1 phospho-sites by repressing the action of phosphatases on AR. We propose that phosphatase action is a key determinant of AR phosphorylation state.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Other antibodies used and their sources are as follows. PAb108 (American Type Culture Collection) is a mouse MAb against SV40 ST that also cross-reacts with Jamestown Canyon virus (JCV) ST and was used in immunoblotting. PAb430 is a mouse MAb that recognizes native SV40 ST and was used to immunoprecipitate SV40 ST-PP2A complexes. PP2A C-subunit antibodies used include a mouse monoclonal 1D6 (Upstate) and a rabbit polyclonal antibody (29). PP2A A-subunit antibodies used include a goat polyclonal antibody (SC-6112; Santa Cruz) and a rat monoclonal antibody (MRT-204R; Covance). The rabbit polyclonal antibody against the C subunit and the goat polyclonal antibody against the A subunit were used in immunoblotting analyses; all antibodies against the PP2A A or C subunit were tested in electrophoretic mobility shift assays (EMSAs). MU014-UC is a rabbit polyclonal antibody against prostate-specific antigen (PSA; BioGenex). DM1A is a mouse MAb against tubulin (Sigma). 16B12 is a mouse MAb against the hemagglutinin (HA) epitope (Covance). M2 is a mouse MAb against the Flag epitope (Sigma).
Transfection reagents used were Fugene6 (Roche) for PC-3, Cos7, 293, and 293T cells and Transfectin (Bio-Rad) for PC-3 and LNCaP cells. The AR ligands used were 5-dihydrotesterone (DHT), androstenedione (ASD), dehydroepiandrosterone (DHEA), estradiol, flutamide (Flut), and hydroxylflutamide (HO-Flut), all from Sigma; synthetic androgen methyltrienolone (R1881) and the radiolabeled form [3H]R1881 from Perkin-Elmer; and bicalutamide (Casodex) from Fisher Scientific. Immunoblotting was performed with peroxidase-labeled secondary antibodies by enhanced chemiluminescence (Pierce). Optimal film exposures were scanned, normalized to a blotting control, and quantified using ImageJ.
IP assays. Cells were suspended in 5 volumes of Triton lysis buffer (0.5% Triton X-100, 20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and a 5 µg/ml of aprotinin, leupeptin, and pepstatin A) and incubated on ice for 20 min with occasional mixing. The lysates were clarified by centrifugation (18,000 x g for 15 min), and the supernatants were rotated with antibody-protein G beads (4 µg of antibody/10 µl of packed beads) at 4°C for 4 h. After five washes with Triton lysis buffer, AR complexes were either eluted by incubating the beads with synthetic peptide (20 µg/ml) at room temperature for 60 min or by resuspending the beads in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer at 95°C. AR complexes were separated by SDS-PAGE followed by Western blot analysis and/or silver staining by standard methods. Immunoblot analysis of PP2A subunits that coimmunoprecipitate with AR was performed with antibodies to the A subunit, the C subunit, or both. Because of the stability of the A/C heterodimer, C-subunit detection is a reliable indicator of the presence of the A subunit.
Androgen binding assays in vivo. PC-3 cells were grown in RPMI phenol red-free (PRF) medium containing 5% fetal bovine serum (pretreated with charcoal and dextran). The cells were infected with adenovirus (Ad)-HisAR, an Ad encoding human AR with a deletion of residues 1 to 37 and an N-terminal His6 tag (kindly provided by William Walker) and with Ad-ST, an Ad encoding SV40 ST (kindly provided by K. Rundell) (33). The infected cells were washed three times with phosphate-buffered saline (PBS) and then incubated with labeled synthetic androgen (0.02 to 2 nM of [3H]R1881) in serum-free medium for 60 min at 37°C. The cells were washed three times with PBS to remove unbound androgen. Bound androgen was extracted from the cells using ice-cold ethanol and measured in a scintillation counter. Nonspecific binding was measured in the presence of a 400-fold excess of unlabeled R1881. Scatchard analysis was used to determine the binding affinity.
Androgen and PP2A dissociation assays in vivo. Ad-infected cells were washed three times with PBS and incubated with 2 nM [3H]R1881 (or 100 nM ASD) in serum-free medium for 60 min at 37°C. The cells were then washed three times with PBS and incubated with 1 µM cold R1881 or 100 µM bicalutamide for the times indicated in the figure legends (0 to 3 h). The cells were again washed twice in PBS. The bound androgen was extracted with ice-cold ethanol, and the counts were measured in a scintillation counter. Alternatively, proteins were extracted from the cells using Triton lysis buffer followed by IP using AR441. The amount of [3H]R1881 bound to AR was measured in a scintillation counter. Proteins isolated by IP were subjected to SDS-PAGE, and the levels of AR and PP2A were determined by immunoblotting.
Androgen/PP2A dissociation assays in vitro. AR complexes were isolated by IP from AR and ST-expressing cells treated with R1881. AR complexes immobilized on antibody-protein G beads were incubated in Triton lysis buffer at room temperature for 3 h to allow androgen and PP2A dissociation. Following centrifugation, the initial bead fraction, the postdissociation bound fraction, and the dissociated fraction were all treated with SDS-PAGE loading buffer and subjected to immunoblot analyses.
Transcription assays. Transcription was measured by use of a dual-luciferase reporter assay system (Promega). Cells were grown in RPMI PRF medium with 5% fetal bovine serum (charcoal and dextran stripped) in 12-well dishes. The cells were transfected with the indicated plasmids according to protocols appropriate for each cell type. Each well typically received a total of 245 or 345 ng of DNA, which included 145 ng of reporter plasmid and 100 or 200 ng of AR plasmid; the balance was empty vector pcDNA3. The luciferase reporters used in this study contained the 6-kb PSA promoter and androgen response element (ARE) I or III from PSA (7) fused to a minimal thymidine kinase promoter. After 24 h of transfection, the medium was changed, and the cells were treated with AR ligands for an additional 24 h. Cells were then washed once with PBS and lysed in passive lysis buffer (Promega). Luciferase activities in lysates were measured on a Berthold LB 953 luminometer. Firefly luciferase activities were normalized to Renilla luciferase activities (plotted as firefly activity/Renilla activity); assays were performed in triplicate, and the results are representative of at least three experiments. Error bars stand for standard deviation. Immunoblotting was used to verify that differences in transcription were not due to AR expression levels. Phosphorylation site mutants of AR were constructed with alanine substitutions (serines 81, 94, 256, 308, and 424) or with aspartate substitutions (serines 81, 94, 256, 308, 424, and 650) by a combination of PCR, restriction digestion, and ligation methods, and all were confirmed by DNA sequencing.
EMSA.
Oligonucleotides representing C3(1)-ARE DNA (6) were end labeled with [32P]dATP (Perkin-Elmer). EMSAs (20-µl reaction mixtures) contained 10 mM Tris-HCl (pH 7.6), 50 mM NaCl, 5 mM MgCl2, 10 nM R1881, 0.05% Triton X-100, 0.5 mM EDTA, 2 mM dithiothreitol, 100 pmol of single-stranded DNA, 1 µg of poly(dI-dC), 0.05 mg/ml bovine serum albumin, 15% glycerol, and either purified AR complexes or recombinant glutathione S-transferase-AR(DBD-Hinge). Reactions were supershifted with an anti-AR hinge region antibody (1 µg). Reaction mixtures were preincubated at room temperature for 5 to 10 min, and AR-ARE binding was initiated by the addition of the labeled probes (40,000 cpm/reaction). The reaction mixture was incubated at room temperature for 15 to 20 min and analyzed by nondenaturing PAGE (4% gel). The gel was run in 0.25x Tris-borate-EDTA buffer and run at 100 V at 4°C. The gels were vacuum dried on 3MM paper and analyzed by using a phosphorimager. There is no supershift of the AR complexes in the assays using the antibodies against the PP2A A subunit or C subunit, probably because none of these antibodies recognizes native protein by IP (our unpublished observations).
Xenograph analysis. LNCaP cell lines stably expressing SV40 ST, ST-myc, or vector alone were prepared by hygromycin selection. LNCaP cell lines (5 x 106 cells) were mixed with 100 µl of matrigel and injected subcutaneously into the flanks of 6- to 8-week-old intact male nude mice (NCI athymic nude mice). Tumors were grown for 11 to 16 weeks, and tumor size was monitored by caliper measurements. Some mice were surgically castrated 4 days prior to tumor harvest. Tumors were cut into small pieces, ground in liquid nitrogen using a mortal and pestle, and resuspended in 5 volumes of homogenization buffer (250 mM sucrose, 50 mM Tris-HCl [pH 8.0], 50 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, and 5 µg/ml each of aprotinin, leupeptin, and pepstatin A). The mixture was homogenized on ice using a hand-held polytron (five 2-s pulses with 10-s pauses between bursts). Homogenates were separated into supernatant and pellet fractions by centrifugation (18,000 x g for 15 min at 4°C). The supernatant was supplemented with 0.5% Triton X-100 and then subjected to IP using anti-AR antibody G122-134 and AR441 (1:1). The AR complexes were separated by SDS-PAGE and analyzed by immunoblotting.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
We constructed a mutant form of PP2A A subunit that lacks only HEAT repeat 13. The HEAT repeat 13 deletion mutant of the PP2A A subunit interacted with ST, but it did not undergo transfer onto androgen-bound AR (Fig. 1D). Thus, HEAT repeat 13 is critical for AR to bind the PP2A A subunit. Deletion of HEAT repeat 13 also disrupted C-subunit binding to the A subunit (Fig. 1D). This result is not surprising, given the biochemical and crystallographic evidence that the C subunit directly contacts HEAT repeats 11 to 15 in the A subunit (5, 35).
PP2A transfer reaction is highly specific for SV40 ST. We investigated whether a homologous ST from human polyomavirus can substitute for SV40 ST in mediating PP2A transfer to AR. JCV causes multifocal leukoencephalopathy in brain, but it is also found in other tissues including the prostate (49). Under conditions where SV40 ST induced PP2A transfer to AR, transient JCV ST expression did not result in PP2A binding to AR (Fig. 2A). Although JCV ST is 65% identical to SV40 ST and we determined that JCV ST binds the PP2A A subunit (Fig. 2B), it fails to induce the A-subunit conformation that is necessary for PP2A transfer to AR. We obtained the same result whether antibodies were used to immunoprecipitate AR or HA-tagged A subunit (Fig. 2C).
|
, ß, 
1, 
, or 
), however, did not promote PP2A transfer to AR (Fig. 2D). The B subunit from the B' family (B') also did not promote PP2A loading onto AR (Fig. 2E).
PP2A interaction with the AR LBD.
AR lacking the AF-1 domain is functional for PP2A binding, but further deletion of the DBD results in a loss of PP2A binding (48). We interpreted this as evidence that PP2A binding to AR may be dependent on the DBD, either because the DBD provides part of the PP2A binding site or because the DBD stabilizes LBD structure. To distinguish between these possibilities, we deleted the DBD from AR (
DBD) and tested it for PP2A binding in the co-IP assay. In the presence of ST and androgen, PP2A was transferred to full-length AR and to the DBD form of AR but not to the LBD form of AR (Fig. 3A). This result together with previous mapping data indicates that the LBD is necessary for PP2A binding, and it rules out an essential function for the DBD in this interaction. As the LBD alone is not sufficient for PP2A binding, we infer that the agonist conformation of the LBD that is recognized by PP2A requires a conformation that is stabilized by interdomain interactions that can be provided by either the N-terminal AF-1 region or the DBD.
|
Effect of PP2A on AR binding to the ARE. ST expression inhibits AR-dependent transcription from the PSA promoter (48). To determine whether the inhibitory effect of ST is due to a reduction in AR binding to DNA, we analyzed AR complexes isolated by immunoaffinity (Fig. 4A and B) in an EMSA. AR binding to DNA was analyzed with the well-characterized C3(1) ARE (6) and quantified on a phosphorimager. Native AR complexes isolated from cells without or with ST coexpression, and consequently with or without bound PP2A, shifted the C3(1) ARE, and both types of complexes were supershifted by an antibody raised against the hinge region of AR (Fig. 4B and C). However, the amount of C3(1) ARE shifted by the AR-PP2A complex was about 68% of the amount shifted by the AR complex that lacked the PP2A heterodimer. The result shown is representative, as the average amount of binding in the presence of PP2A in four experiments was 67.5% ± 10.4%. The fact that the presence of PP2A does not induce a gel shift resolved in this system suggests that either the size of the heterodimer (A plus C subunit, 100 kDa) is inconsequential relative to the size of the native AR complex or that PP2A has dissociated from the AR-DNA complex. The small reduction in DNA binding could be due to AR dephosphorylation or a structural effect of PP2A binding.
|
Androgen dissociation correlates with PP2A release from AR. Since androgen-free AR does not bind PP2A, we hypothesized that structural changes that occurred in the LBD after androgen dissociation might provide the basis for PP2A release from AR. To test this hypothesis, we designed a cell-based assay that measures the rates of both androgen and PP2A dissociation from AR. Cells expressing AR and ST were initially incubated with [3H]R1881 to allow androgen and PP2A binding. The cells were then treated with excess anti-androgen bicalutamide for up to 180 min. A large excess of bicalutamide was added to the cells during the chase period since the bicalutamide-bound form of AR does not bind to PP2A (48). This rendered the reaction unidirectional and allowed us to measure the dissociation rate of both androgen and PP2A from AR in vivo. AR complexes isolated at each time point by IP were analyzed by immunoblotting for AR and PP2A content (Fig. 5A) and by scintillation counting for bound radiolabeled androgen (Fig. 5B). The data for androgen dissociation and PP2A dissociation were each fit with a single exponential decay curve, indicating that both dissociation reactions obey first-order dissociation kinetics (Fig. 5B and C). There is a statistically significant correlation (R2 = 0.986) between the amounts of PP2A and androgen that are bound to AR during the time course of dissociation (Fig. 5D). Our data suggest that androgen dissociation is correlated with PP2A dissociation from AR, but we considered the formal possibility that PP2A remained bound to the androgen-free AR and that bicalutamide binding induced a conformation in the LBD that stimulated PP2A release. We addressed this possibility by showing that PP2A dissociates from AR in the absence of bicalutamide in vitro (Fig. 5E) to an extent that is comparable to that measured in the presence of bicalutamide in vivo (Fig. 5A).
|
|
|
ST-dependent PP2A transfer to AR in a xenograft model of prostate cancer. We prepared stable lines of LNCaP cells expressing ST for xenograft studies that would enable us to test whether androgen status influences PP2A loading onto AR in vivo. ST-myc-expressing cells were generated as a control line, since a minimal C-terminal myc tag inhibits the PP2A transfer function of ST (48). We used IP analysis to verify that PP2A transfer onto AR in the stable lines grown in culture requires androgen (R1881) and functional ST (Fig. 8A). The cell lines were next grown as subcutaneous tumors in nude mice, and 4 days prior to tumor harvest, a subset of mice bearing LNCaP-ST tumors was castrated to deplete testicular androgens. This would allow us to test whether a PP2A transfer reaction could occur in the presence of ST in the context of an animal model and whether AR can adopt an androgen-bound conformation in an animal following castration. AR complexes were isolated from the tumor extracts by IP and then probed for the presence of AR-PP2A complexes. As a functional readout of PP2A binding to AR, we also examined the phosphorylation state of phospho-sites in the AR AF-1 region that are dephosphorylated by PP2A in cells grown in tissue culture. In the LNCaP-ST xenograft harvested from the intact mouse, the C subunit of PP2A was present in the AR complex, indicating that ST mediates PP2A transfer to AR in this model system (Fig. 8B, lane 2). ST-myc failed to mediate PP2A transfer onto AR (Fig. 8B, lane 1), which is consistent with results obtained in cultured cells (48). The presence of PP2A in the AR complex was correlated with a reduction in AF-1 phosphorylation. Unexpectedly, the level of PP2A bound to AR from the LNCaP-ST tumor from the castrated mouse was comparable to the level observed in the tumor from the intact mouse (Fig. 8B, lane 3). Thus, under castration conditions, while castrate levels of androgen in the nude mouse are insufficient to maintain growth of a number of androgen-dependent cell lines and xenografts, AR in tumor xenografts is still capable of forming a conformation that can bind stably to PP2A. This shows that in LNCaP cells propagated as a xenograft, AR acquires a conformation that, by the biochemical criterion of PP2A binding, is like the conformation induced by DHT, R1881, and ASD. This suggests that castrate levels of androgen (including testicular androgen DHT and testosterone and adrenal androgen ASD) are sufficient for driving the androgen-bound conformation of AR. It is also possible that the milieu provided by the xenograft allows activation of signaling pathways that promote the androgen-bound conformation of AR.
|
We tested whether known phospho-sites are accessible to kinases by expressing AR in 293 cells and treating the cells with androgen and/or the phosphatase inhibitor OA. The AR was then isolated by IP and analyzed using phospho-site-specific antibodies. As shown previously, androgen treatment increased the phosphorylation level of Ser81, Ser256, Ser308, and Ser424 (Fig. 9A, +R1881) (16, 48). The phosphorylation levels of Ser16 and Ser94 displayed little change with androgen treatment, consistent with apparent constitutive phosphorylation of these sites (48). Remarkably, the level of phosphorylation of all AF-1 sites in AR showed a large increase in the presence of a concentration of OA (400 nM) that inhibits phosphatases PP2A and PP5 (Fig. 9A, lane 3). Cells treated with OA and androgen (Fig. 9A, lane 4) showed little additional increase in the level of AR phosphorylation. From these data, we conclude that AF-1 phosphorylation sites in AR are kinase accessible prior to androgen binding. PP2A binding to AR does not preclude kinase access to the sites since AR-PP2A complexes are highly phosphorylated in the presence of OA (Fig. 9C).
|
LBD form of AR that was comparable to the level of phosphorylation in full-length AR induced by androgen (Fig. 9C). Also, Ser424 phosphorylation in the LBD form of AR was phosphorylated to the same extent as in full-length AR. In contrast to these sites, the level of phosphorylation of Ser256 was clearly not induced when the LBD was deleted. We interpret these data as evidence that the LBD of AR is critical for efficient phosphatase interaction with AR because its removal results in constitutive phosphorylation of at least two sites that are subject to dephosphorylation in androgen-free AR. Thus, androgen binding either reduces phosphatase targeting to AR or induces an AR conformation that protects Ser81 and Ser308 from dephosphorylation. Phosphorylation of Ser256 is strongly dependent on the LBD for reasons related either to phospho-site conformation or kinase targeting to AR or simply because Ser256 in LBD AR is dephosphorylated more efficiently than it is phosphorylated. |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Mechanism of PP2A binding to AR. The PP2A core enzyme is a heterodimer that contains a catalytic C subunit and a structural A subunit. In previous work we showed that the A subunit of PP2A contacts AR, and we presented data suggesting that ST-induced conformational changes in the A subunit are critical for PP2A A/C heterodimer to bind the androgen-bound form of AR (48). ST binds directly to the A subunit (34), and we found that ST binding to the A subunit in vitro alters the protease sensitivity of sites both within and outside of the ST interaction site (48). Work from other laboratories had mapped the ST binding site to HEAT repeats 3 to 6 in the N-terminal region of the A subunit (34). We found that deletion of HEAT repeats 3 to 5 abrogated ST-induced PP2A binding to AR (Fig. 1), suggesting that HEAT repeats 3 to 5 (or the interface between HEAT repeats 3 and 5) are critical elements of the ST binding site (Fig. 10A). Our deletion analysis also revealed that removal of HEAT repeats 13 to 15 in the C-terminal region of the A subunit eliminates PP2A binding to AR. Moreover, we determined that deletion of HEAT repeat 13 is sufficient to disrupt PP2A A-subunit binding to AR. Together, these results indicate that ST binding to N-terminal HEAT repeats elicits a conformational change in the C-terminal HEAT repeats that creates binding site for AR.
|
Crystallographic studies of PP2A (5, 17, 46) provide a context for considering how ST contact with the A subunit might facilitate PP2A binding to the AR. Each of the 15 HEAT repeats in the PP2A A subunit consists of anti-parallel A and B helices that are connected by a short turn. The HEAT repeats are stacked such that the A helix in HEAT repeat 2 contacts the A helix in HEAT repeats 1 and 3, and so forth. Likewise, the B helix in the HEAT repeat 2 contacts the B helix in HEAT repeats 1 and 3. The overall structure is an elongated, left-handed superhelix with A helices on the convex surface and B helices on the concave surface with dimensions of 45 Å by 65 Å (17). The approximately parallel arrangement of A helices with A helices (and B helices with B helices) is not, however, maintained throughout the entire superhelix. Discontinuities in helix stacking are found between HEAT repeats 3 and 4 and between HEAT repeats 12 and 13. Because of these discontinuities, the PP2A A subunit can be divided into three regions consisting of HEAT repeats 1 to 3, 4 to 12, and 13 to 15, each of which is believed to form a stable structural unit (17). As shown in Fig. 10A, superimposition of two A-subunit molecules from the crystallographically asymmetrical unit reveals a striking degree of alignment of HEAT repeats 5 to 12, but there is apparent conformational flexibility of the N-terminal and C-terminal HEAT repeats. The apparent flexibility of HEAT repeats 13 to 15 is further emphasized by a comparison of the A-subunit structure with the structures of the PP2A core and holoenzymes (5, 46). It seems noteworthy that the N-terminal and C-terminal regions of the A subunit that display conformational flexibility are the same regions that bind reversibly to ST and AR binding regions of PP2A A subunit.
We propose that ST binding to N-terminal HEAT repeats causes a conformational change that is transduced through HEAT repeats 6 to 12, causing a subtle conformational change in the C-terminal HEAT repeats that reveals an AR binding site (Fig. 10A). Because ST is not detected in the AR-PP2A complex (48), AR binding to the C-terminal HEAT repeats in the A subunit may elicit a structural change that is transduced back through HEAT repeats 6 to 12, resulting in a conformational change in the N-terminal HEAT repeats that releases ST. The available data are consistent with the concept that the architecture of the A subunit is uniquely suited for protein binding and for propagating conformational changes from N to C and from C to N (Fig. 10A).
A second important aspect of the PP2A-AR interaction is that a specific androgen-induced conformation of AR is required for the interaction. Androgen-free AR does not bind to PP2A in the ST-dependent transfer reaction, and of all androgen agonists and antagonists tested, only the synthetic androgen R1881, DHT, and ASD were capable of generating an AR conformation compatible with PP2A binding. The requirement for an androgen-bound conformation points to the LBD as the likely binding site for the PP2A A subunit. This view is corroborated by deletion analysis showing that AR lacking the AF-1 or DBD is still functional for PP2A binding, but removal of the LBD abolishes PP2A binding. We also observed a correlation (R2 = 0.986) between dissociation of androgen from AR and PP2A release, evidence that PP2A release occurs when AR returns to the androgen-free conformation.
AR conformation and transcription. The stability of the AR-PP2A interaction enabled us to examine whether AR adopts a conformation that binds PP2A, which the data argue is an androgen-bound conformation, under conditions that are relevant to prostate cancer. The T877A mutation broadens the ligand recognition properties of the AR LBD, and, as a consequence, antagonists including Flut and HO-Flut function as agonists that promote AR-dependent transcription (13, 38, 42). These observations led us to hypothesize that Flut and HO-Flut might, in the context of the T877A mutation, generate the agonist-type conformation of AR. This was not the case, however, as these and other ligands activated AR-dependent transcription of reporter genes and endogenous PSA without generating an apparent androgen-bound conformation of AR. Thus, five different ligands (Flut, HO-Flut, bicalutamide, DHEA, and estradiol) supported T877A AR-dependent transcription without generating the androgen-bound conformation of AR. Our conclusion appears to differ from the conclusion drawn by the Chang and McDonnell (2). Using a mammalian two-hybrid assay to score the interaction between a coactivator peptide (D30) and AR, it was concluded that HO-Flut could, in the context of the T877A mutation, generate the agonist conformation of AR (2). A potential explanation for the difference in interpretation is that PP2A and the D30 coactivator peptide may register different structural features of the antagonist-bound LBD.
The AR-PP2A co-IP assay also allowed us to formally test whether AR adopts an androgen-bound conformation in the presence of castrate levels of androgen. The impetus for this experiment was that castrated mice provide the setting for developing and propagating androgen-independent human prostate tumors. In some prostate cancer models, castrate levels of androgen are insufficient to fully activate AR (50). Our analysis, which was performed in LNCaP cells grown as xenografts in mice, revealed that the conformation of AR that binds PP2A is maintained even after the mice were castrated. This resulted in ST-dependent PP2A transfer to AR and dephosphorylation of phospho-sites in the AF-1. Our data suggest that castrate levels of androgen are sufficient to promote the androgen-bound conformation of AR, though it is possible that signal transduction in the xenograft promotes an AR conformation that binds PP2A through another pathway. We speculate that androgen-sensitive prostate tumor growth might reflect the involvement of an androgen-sensitive component that has a higher dissociation constant for androgen than AR (19).
An alternative view of how ligand can regulate the AR phosphorylation state. Ligand binding increases the phosphorylation state of nuclear receptors, including those that bind steroid hormones. Phospho-site-specific antibodies have been used to show that phosphorylation levels at sites in the AF-1 region of AR (Ser81, Ser213, Ser256, Ser308, Ser424, and Ser650) (40, 48), GR (Ser203 and Ser211) (43), progesterone receptor (Ser102, Ser 294, and Ser345) (51), and ER (Ser118) (22) increase upon the addition of cognate ligand. The tacit assumption in these studies has been that ligand binding induces a conformational change that exposes AF-1 phosphorylation sites (Fig. 10B, model I). That kinase accessibility to AF-1 sites would be enhanced in response to ligand binding seems reasonable, given the evidence that AF-1 regions of nuclear receptors display ligand-regulated interactions with the LBD (28). According to this view, ligand binding to the C-terminal LBD could promote changes in the N-terminal AF-1 structure that either (i) increase AF-1 solvent exposure, (ii) release proteins that otherwise repress kinase action on the AF-1, or (iii) promote AF-1 kinase recruitment to AR.
Our data suggest an additional mechanism for how the phosphorylation state of the AF-1 region could increase upon ligand treatment (Fig. 10B, model II). We found that treating cells with a concentration of OA that inhibits a subset of phosphatases was sufficient to promote a high level of phosphorylation of five sites in the AF-1 region of AR. Most of these sites in AR undergo robust androgen-dependent phosphorylation. We also observed that phospho-site Ser16, which displays a basal level phosphorylation and is affected only slightly by ligand addition, shows a significantly higher level of phosphorylation in the presence of OA. Our data provide compelling evidence that in cells, these phosphorylation sites are kinase accessible prior to ligand binding. This leads us to propose that the low levels of phosphorylation observed in the AF-1 sites of AR in the absence of androgen are maintained by the constitutive action of one or more phosphatases (Fig. 10B, model II). This could be through the action of phosphatases that are stably bound to AR, phosphatases that interact transiently with AR, or both. According to this model, androgen binding would induce a conformation in AR that is resistant to phosphatase action, either because the phospho-sites become inaccessible to phosphatase or because changes in AR structure are incompatible with efficient phosphatase targeting. We favor the latter scenario since removal of the LBD from AR results in a high level of Ser81 and Ser308 phosphorylation even without OA treatment.
The LBD appears to play a dual role in regulating phosphatase action on AR. In the absence of androgen, the LBD facilitates AF-1 dephosphorylation because the ligand-free form of AR is highly compatible with phosphatase interactions. In the presence of ligand, the LBD appears to repress dephosphorylation either because the ligand-bound form of AR is incompatible with phosphatase interactions or because the ligand-bound form of AR actively represses dephosphorylation. Even though the phospho-sites are kinase accessible prior to androgen binding, the LBD may also make a positive contribution to Ser256 phosphorylation. Ser256 phosphorylation in the LBD mutant of AR was extremely low compared to androgen-bound AR. This could be because the LBD is important, directly or indirectly, for kinase recognition of the Ser256 phospho-site.
The candidates for mediating constitutive dephosphorylation of unliganded AR are the OA-sensitive phosphatases PP5 and PP2A. PP5 is known to bind GR-associated Hsp90, and through this interaction PP5 can influence GR subcellular localization and transactivation function (9, 11, 39). It is not known, however, whether GR is a direct substrate for PP5-dependent dephosphorylation. By extension, PP5 is predicted to bind AR-associated Hsp90, but to our knowledge this interaction has not yet been demonstrated. Proposing that PP2A is a constitutive phosphatase for ligand-free AR seems at odds with our finding that PP2A binds specifically to the androgen-bound conformation of AR. But it should be noted that this is a special circumstance that occurs in the presence of SV40 ST, and it is correlated with alterations in PP2A A-subunit structure. It is possible that SV40 has coopted a constitutive pathway that normally mediates dephosphorylation of ligand-free AR. ST binding to PP2A shifts the pathway to a dependence on androgen-dependent changes in AR structure and results in androgen-dependent transfer and dephosphorylation of AR. The biological significance of this reaction is not yet clear. It has been shown that combining SV40 ST and large-T-antigen expression in a mouse model of prostate cancer results in a highly metastatic phenotype, whereas SV40 large-T-antigen expression is only tumorigenic (14) (23). Whether ST expression and modulation of AR phosphorylation are causal to some aspect of the metastatic phenotype remains an open question.
In summary, we have shown that protein conformation plays a critical role in determining phosphatase action on AR. Understanding the balance between kinase and phosphatase actions on AR should help explain how extracellular signals are transduced from the plasma membrane to the nucleus and provide the context for defining how alterations in these pathways may contribute to the development of prostate cancer.
|
ACKNOWLEDGMENTS |
|---|
This work was funded by the NIH.
|
FOOTNOTES |
|---|
Published ahead of print on 26 February 2007.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2. Chang, C. Y., and D. P. McDonnell. 2002. Evaluation of ligand-dependent changes in AR structure using peptide probes. Mol. Endocrinol. 16:647-660.
3. Chen, F., K. Knecht, C. Leu, S. J. Rutledge, A. Scafonas, C. Gambone, R. Vogel, H. Zhang, V. Kasparcova, C. Bai, S. Harada, A. Schmidt, A. Reszka, and L. Freedman. 2004. Partial agonist/antagonist properties of androstenedione and 4-androsten-3ß,17ß-diol. J. Steroid Biochem. Mol. Biol. 91:247-257.[CrossRef][Medline]
4. Chen, J., J. Kim, and J. T. Dalton. 2005. Discovery and therapeutic promise of selective androgen receptor modulators. Mol. Interv. 5:173-188.
5. Cho, U. S., and W. Xu. 2006. Crystal structure of a protein phosphatase 2A heterotrimeric holoenzyme. Nature 445:53-57.
6. Claessens, F., L. Celis, B. Peeters, W. Heyns, G. Verhoeven, and W. Rombauts. 1989. Functional characterization of an androgen response element in the first intron of the C3(1) gene of prostatic binding protein. Biochem. Biophys. Res. Commun. 164:833-840.[CrossRef][Medline]
7. Cleutjens, K. B., H. A. van der Korput, C. C. van Eekelen, H. C. van Rooij, P. W. Faber, and J. Trapman. 1997. An androgen response element in a far upstream enhancer region is essential for high, androgen-regulated activity of the prostate-specific antigen promoter. Mol. Endocrinol. 11:148-161.
8. Daaka, Y. 2004. G proteins in cancer: the prostate cancer paradigm. Sci. STKE 2004:re2.
9. Dean, D. A., G. Urban, I. V. Aragon, M. Swingle, B. Miller, S. Rusconi, M. Bueno, N. M. Dean, and R. E. Honkanen. 2001. Serine/threonine protein phosphatase 5 (PP5) participates in the regulation of glucocorticoid receptor nucleocytoplasmic shuttling. BMC Cell Biol. 2:6.[Medline]
10. Defranco, D. B., A. P. Madan, Y. Tang, U. R. Chandran, N. Xiao, and J. Yang. 1995. Nucleocytoplasmic shuttling of steroid receptors. Vitam. Horm. 51:315-338.
