Functional localization and competition between the androgen ...

Oncogene (2003) 22, 5602–5613

& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

Functional localization and competition between the androgen receptor and T-cell factor for nuclear b-catenin: a means for inhibition of the Tcf signaling axis David J Mulholland*,1, Jason T Read1, Paul S Rennie1, Michael E Cox1 and Colleen C Nelson1 1

The Prostate Centre, Jack Bell Research, Vancouver, BC, Canada V6 H 3Z6

Recent reports suggest that the b-catenin-T-cell factor (Tcf ) (BCT) signaling pathway is important in the progression of prostate cancer. Evidence suggests that the androgen receptor (AR) can repress BCT-mediated transcription both in prostate cancer and colon cancer cells (Chesire and Isaacs, 2002). In this study, we validate such findings and show that repression of BCT signaling is facilitated by competition between the AR and Tcf. Measurements of the Tcf transcriptional reporter (TOPFLASH) indicated that AR þ DHT-mediated repression can inhibit BCT transcription in the presence of WT and exogenous activating b-catenin (D1–130 bp). Transient transfections in SW480 cells (APCmut/mut) showed that this mode of repression is functionally independent of APCmediated b-catenin ubiquitination. Using a recently developed red flourescent protein (HcRed), we demonstrate novel observations about the nuclear distribution of Tcf. Furthermore, with the use of red (HcRed-AR and HcRed-Tcf) and green fusion proteins (b-catenin-EGFP), we provide morphological evidence of a reciprocal balance of nuclear b-catenin-EGFP (BC-EGFP). By cotransfecting in LNCaP prostate tumor cells and using quantitative imaging software, we demonstrated a 62.0% colocalization of HcRed-AR and BC-EGFP in the presence of DHT and 63.3% colocalization of HcRed-Tcf/BC-EGFP in the absence of DHT. Costaining for activated RNA Pol II (phosphoserine 2) and HcRed-Tcf suggested that Tcf foci contain transcriptional ‘hotspots’ validating that these sites have the capacity for transcriptional activity. Given this apparent androgen-dependent competition for nuclear BC-EGFP, we chose to assess our hypothesis by in vivo and in vitro binding assays. SW480 cells transiently transfected with an AR expression construct, treated with DHT and immunoprecipitated for Tcf showed less associated b-catenin when compared to Tcf precipitates from untreated cells. Furthermore, by treating cells with DHT þ Casodex, we were able to abrogate the androgensensitive AR/b-catenin interaction, in addition to relieving transcriptional repression of the TOPFLASH reporter. In vitro binding assays, with increasing amounts of ARS35, resulted in decreased Tcf S35 association with immunoprecipitated recombinant b-catenin-HIS. These data suggest that in steady-state conditions, AR has the ability to *Correspondence: DJ Mulholland; E-mail: [email protected] Received 7 February 2003; revised 18 May 2003; accepted 19 May 2003

compete out Tcf binding for b-catenin. Finally, using SW480 cells, we show that AR-mediated repression of the BCT pathway has implications for cell cycle progression and in vitro growth. Using FACs analysis, we observed a 26.1% increase in accumulation of cells in the G1 phase of the cell cycle, while in vitro growth assays showed a 35% reduction in viable cells transfected with AR þ DHT treatment. Together, our data strongly suggest that a reciprocal balance of nuclear b-catenin facilitates ARmediated repression of BCT-driven transcription and cell growth. Oncogene (2003) 22, 5602–5613. doi:10.1038/sj.onc.1206802 Keywords: androgen receptor; prostate cancer; coactivator; Tcf signaling; b-catenin

Introduction The androgen receptor (AR) is a member of the liganddependent steroid, nuclear receptor superfamily and is a fundamental regulator of growth, proliferation, differentiation and carcinogenesis in prostate epithelial cells (Quigley et al., 1995). In prostate cancer, the effects of androgens are important in the transition of the disease from an androgen-dependent to androgen-independent state (Quigley et al., 1995). The AR mediates its physiological activities by binding androgens, dissociating from heat shock proteins, dimerization and translocation to the nucleus for binding to androgen-responsive genes such as probasin or prostate-specific antigen (PSA). The transcriptional activity of AR depends not only on its ligand DHT, but also on coregulators including activators and repressors (Ikonen et al., 1997; Yeh et al., 1999; Janne et al., 2000). b-Catenin is a multifunctional protein. In addition to bridging E-cadherin with a-catenin and the actin cytoskeleton, b-catenin and Wnt signaling are potent promoters of oncogenesis (Polakis, 2001). In noncancerous cells, the Wnt signaling members GSK3 and APC prevent cytosolic accumulation of b-catenin and, therefore, negatively regulate T-cell factor (Tcf ) signaling. However, cancer cells may contain defects in negative regulators of the Wnt pathway, ultimately promoting Tcf activation and oncogenesis (Chen and McCormick, 2001). Colon cancer cells often have mutations in either

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5603

adenomatous polyposis coli (APC) or b-catenin, leading to stabilization of b-catenin and activation of downstream Tcf target genes such as c-myc and cyclin D1 (Chen and McCormick, 2001). Mechanisms that either facilitate degradation of b-catenin or the sequestering of b-catenin from Tcf4-binding sites could serve to reduce Tcf/Lef signaling. Previously, it has been shown that nuclear b-catenin implements transcriptional activation by binding to the Tcf transcriptional machinery including a family of the high-mobility group (HMG) factors. The balance between activation and repression of Tcf is mediated by competition between corepressors and TATA box-binding proteins (Carlsson et al., 1993; Barker et al., 2000). Nuclear hormone receptors have been shown to have bipolar transcriptional capabilities, allowing them to either activate or repress expression of target promoters (Damm et al., 1989; Sap et al., 1989; Schulman et al., 1996). Recently, downstream targets have included those of Tcf. For example, it has been shown that bcatenin increases the activity of the retinoic acid receptor (RAR) on RAR-responsive promoters with a reciprocating downregulation of the b-catenin/Tcf signaling pathway by the RAR (Easwaran et al., 1999). Other reports have shown Wnt signaling to be functional in human thyroid cells (Helmbrecht et al., 2001) with triiodothyronine (T3) also having the capacity to silence the b-catenin/Tcf pathway (Miller et al., 2001). With the structural conservation between members of the steroid nuclear receptor family, it is reasonable to hypothesize that similar interactions with the Wnt pathway may exist with other nuclear steroid receptor members. Reports have shown a role for the b-catenin/Tcf pathway in prostate cancer including nuclear localization and b-catenin mutations in primary prostate tumor samples (Chesire and Isaacs, 2002; Chesire et al., 2002). b-Catenin has also been described as a liganddependent coactivator of the AR (Truica et al., 2000; Mulholland et al., 2002; Pawlowski et al., 2002; Yang et al., 2002). Previously, we have shown that translocating AR can provide a means of nuclear entry and accumulation of b-catenin (Mulholland et al., 2002). This mode of b-catenin trafficking has also been shown to hold true in neuronal cells (Pawlowski et al., 2002). Cotrafficking of AR and b-catenin to the nucleus likely has important implications both for AR and Wnt signaling. While nuclear accumulation of b-catenin has been shown to correlate with increased AR transcriptional activity, the effects on Tcf signaling have only begun to be explored. Recently, it has been shown that the AR has the ability to inhibit the b-catenin/Tcf signaling pathway, ligand-dependently (Chesire and Isaacs, 2002). By way of transcriptional reporter assay, this report showed reduced luciferase activity for the Tcf reporter in a ligand-dependent manner in several prostate and colon cancer cell lines. This study also observed reduced transcriptional activity with the use of an AR deletion mutant (for b-catenin binding), suggesting the possibility of a reciprocal balance of nuclear bcatenin between the AR and Tcf. In the present study, we corroborate these results but provide mechanistic

data to support the hypothesis that repression of the bcatenin/Tcf signaling is mediated by ligand-occupied AR that is in competition with Tcf for nuclear b-catenin. Specifically, using transcriptional reporter assays, we show that overexpression of WT Tcf reduced the activity of an AR (ARR3-Luc)-responsive reporter, while overexpression of a DNt Tcf mutant did not have this effect. We employ the use of fluorescent fusion proteins and image analysis software to quantify a reciprocal balance of b-catenin-EGFP between HcRed-Tcf and HcRedAR. We further show that Casodex has the potential to diminish AR-mediated depletion of Tcf transcription and the ability to diminish androgen-sensitive coimmunoprecipitation of endogenous AR and b-catenin. Our data are also novel in showing that AR-mediated reduction of the Tcf signaling pathway can translate into altered cell cycle and reduced cell growth.

Materials and methods Cell culture and transient transfections At 24 h prior to transfection, PC3 (DMEM media), LNCaP (RPMI), HCT116 (McCoy’s) and SW480 (DMEM) cells were cultured in a 5% CO2 incubator in 5% FBS supplemented media with 1% penicillin/streptomycin. Transfections were for dextran-coated 6–16 h in serum-free media and were then treated with charcoal-stripped serum (CSS) with or without ligand for 24 h for PC3, HCT116, and SW480 cells, but for 48 h when using LNCaP cells. R1881 and DHT were obtained from Sigma while Casodex (Bicalutamide) was obtained from Astro Zenica. Plasmids Full-length hAR was PCR amplified using both forward and reverse primers containing BamHI restriction sites and pSVAR0 as a template (Brinkman, University of Rotterdam). Forward and reverse primer sequences were 50 GGATCCG30 and 50 GGATCTT30 , respectively. The PCR product was digested with BamHI and ligated into BamHI sites of pBlueScript plasmid (Stratagene). AR cDNA was excised and subcloned into the BamHI sites of pcDNA3.1. The ligandbinding domain deletion (ARNt/DBD) construct, amino-terminal AR deletion (ARDBD/LBD) construct and the ARR3-Luc reporter were cloned and characterized as previously described (Snoek et al., 1996). The AR-EGFP plasmid was obtained from Dr A Roy (University of Texas Health Science Centre at San Antonio); WT and mutant (D1–130 bp) b-catenin cDNAs from Dr R Moon (Howard Hughes Institute, University of Washington); human DNt-Tcf4 and Tcf construct (pHRhTCF4) from Dr B Vogelstein (Johns Hopkins, Baltimore, USA); and cyclin D1-luc construct from Dr R Pestell (Albert Einstein College of Medicine, NY, USA). The Tcf4 luciferase reporter (TOPFLASH) and a mutated control reporter (FOPFLASH) were purchased from Upstate Biotechnology. HcRed (Clontech) is a recently developed far-red fluorescent protein derived from the coral Heteractis crispa and has been shown to be efficient for colabeling with green fluorescent proteins (Knop et al., 2002). The HcRed-hTcf4 fusion vector was generated by cloning the BamHI fragment of pHR-hTcf4 into the BamHI site of pBlueScript SK- (Stratagene), followed by excision of the reading frame fragment by an EcoRI digest and subsequent cloning into the EcoRI site of the pHcRed1-C1 Oncogene

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5604 vector. The HcRed-AR contains a PCR product digested with BamH1 and ligated into a BamH1 ligated into the pHcRed-C1 vector (Clontech). The b-catenin-HcRed fusion vector was generated by cloning the BamHI-liberated b-catenin coding sequence from pGEX4T1-b-catenin (A Hecht, Max-PlanckInstitute of Immunobiology, Freiburg) into the BamHI site of the pHcRed-C1 vector. The 6  His-tagged hTcf4 bacterial expression vector was generated by subcloning the hTcf4 coding sequence out of the pHR-hTcf4 mammalian expression vector. The pHR-hTcf4 vector was digested with BamHI, and a DNA fragment of B2.4 kb containing the hTcf4 coding sequence was gel extracted. This fragment was cloned into the BamHI site of the pBlueScript SK () plasmid and the resulting clones were sequenced to identify clones in which the EcoRI site of pBlueScript was downstream of the hTcf4 coding sequence. An appropriate clone was isolated and digested with BamHI in the presence of ethidium bromide to prevent cutting of both BamHI sites. The linear plasmid was precipitated to remove ethidium bromide and then digested with EcoRI. The 2.4 kb band corresponding to the hTcf4 coding sequence was gel extracted and cloned into the BamHI/EcoRI sites of the pENTR2B Gateway vector (Invitrogen). The resulting colonies were sequenced to verify the reading frame and then used in an LR recombination reaction as per the manufacturers’ instructions to transfer the hTcf4 coding sequence to the pDEST17 bacterial 6  HIS-tag Gateway Destination vector (Invitrogen). Clones were screened for insert direction and reading frame integrity by sequence analysis. Immunocytochemistry Cells were grown on glass coverslips, transfected with 3 mg of plasmid per well and treated for 48 h with or without ligand. Subsequently, cells were fixed with cold methanol for 3 min, airdried for 10 min and reconstituted in blocking buffer (0.1% Tween PBS/BSA þ 4% NGS) for 20 min. Following blocking, cells were treated with primary and fluorescent secondary antibodies, each for 1–2 h at room temperature with 3  5 min washes (0.1% Tween PBS/BSA þ 1% NGS) after each incubation. Antibodies used for staining included phosphoserine RNA Pol II (Babco). Cells examined for localization of red and green fusion proteins were washed (1  5 min), fixed and mounted with fluorescent mounting media (Vector). Images were captured in Z-series using a Zeiss Axioplan II fluorescence microscope followed by analysis with imaging software (Northern Eclipse, Empix Imaging, Inc.). Z-series images were either deconvolved as a single section (2D interative) or as a Z-series stack (full Iterative) followed by a maximum xy plane projection. Analysis of focal colocalization was also performed with Northern Eclipse and Adobe Photoshop 5.5 software with an assignment of yellow (Y) for a colocalized foci and green (G) or red (R) as non-colocalization (Figure 3). The degree of colocalization was calculated both on a foci level and a total nuclear level. The former method scored an individual foci as either R, G or Y based on the predominating color. Counting on a nuclear level we considered the total R, G or Y within the nucleus, which considered partially overlapping foci containing more than one color. Cells were evaluated using a  63 objective with at least 10 cells/treatment contributing to statistical analysis.

harvested and luciferase activity was evaluated using the Dual Promega Luciferase (Promega). Luciferase assays for PC3 cells were performed in six-well plates while all other cell lines were analysed in 12-well plates, both with a maximum of 10 mg of transfected DNA/plate. For each transfection, the total mass of DNA (mg) was kept equal by the addition of empty vector pcDNA 3.1 (Invitrogen). To normalize for transfection efficiency, several means were used including cotransfection with pRL-TK Renilla vector (Promega) and cotransfection with CMV-EGFP-tagged empty vectors followed by Western blotting for the EGFP. Western blotting/immunoprecipitation Subconfluent cell cultures to be used for Western blotting were washed with PBS and lysed with RIPA lysis buffer (0.1% SDS, 150 mm NaCl, 50 mm Tris, pH 8.0) on ice for 30 min with the addition of a protease inhibitor cocktail (Roche Scientific). Samples were standardized for total protein and separated by SDS–PAGE. Alternatively, cells used for immunoprecipitation were lysed with NP-40 lysis buffer (1% NP-40, 150 mm NaCl, 50 mm Tris, pH 8.0), standardized for total protein content and immunoprecipitated with antibodies (16 h at 41C) and protein A/G beads (Santa Cruz). Following SDS–PAGE and blotting to PVDF (Amersham Life Sciences), membranes were blocked with 5% nonfat milk in TPBS (0.05% Tween-20 in PBS) for 1 h and then incubated for 1–2 h at room temperature with primary antibodies. Antibodies included b-catenin (Transduction Lab., Santa Cruz Biotechnology), AR (ABR, Pharmingen, Santa Cruz Biotechnology), Tcf4 (Santa Cruz, Upstate Biotechnology), HIS-tag (New England BioLabs), myc-tag (Santa Cruz), GFP (Sigma), actin (Sigma) or cyclin D1 (Santa Cruz Biotechnology). Detection was carried out with either anti-mouse-HRP (Santa Cruz) or anti-rabbit-HRP (Santa Cruz) and ECL Western blotting detection reagents (Amersham Life Science). In vitro translation and recombinant proteins Radio-labeled (S35-Met) Tcf4HIS and hAR protein was prepared using the Quick Coupled T7 TNT in vitro transcription/translation kit (Promega). Recombinant b-cateninHIS was expressed in the BL21 Escherichia coli strain and purified as described previously (Mulholland et al., 2002). In vitro binding reactions occurred in binding buffer (20 mm HEPES pH 7.6, 150 mm KCl, 5 mm MgCl2, 1 mm EDTA, 0.05% NP-40) and were precipitated using Ni-NTA beads (Qiagen). Precipitates were washed four times with binding buffer, separated by SDS–PAGE, fixed, amplified with NAMP 100 (Amersham) and exposed to autoradiographic film. FACS analysis SW480 cells were transfected in six-well dishes with and without 10 mg of WT AR treated with increasing concentations of DHT (0, 0.01, 0.025, 0.05, 0.075, 0.1, 1 or 10 nm). Cells were collected using PBS/EDTA (5 mm) and fixed, dropwise, to a final concentration of 70% cold EtOH. Cells were pelleted and resuspended in staining buffer (20 mg/ml RNAse A þ propidium iodide) for 30 min at 371C and then 41C for 60 min. Samples were run in triplicate with a minimum of 10 000 cells/replicate.

Transcriptional assays After 24 h of transfection, cells were cultured in 5% charcoal dextran-treated fetal bovine serum treated with DHT (0, 0.01, 0.025, 0.05, 0.075, 0.1, 1 or 10 nm) dissolved in dH2O, 10 mm pure Casodex or vehicle. At 24 h post-transfection, cells were Oncogene

In vitro growth assays The in vitro growth of SW480 cells was assessed by the in vitro 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (MTT assay) as previously described (Gleave et al.,

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5605 1991). Briefly, 3  103 cells were seeded into each well of a 96well microtiter plate. Cells were transfected with 0.1 mg of DNA per well and treated with ligand for 24 h. Subsequently, cells were treated with 20 ml of MTT (Sigma) in PBS followed by incubation for 4 h at 371C. The absorbance was determined with a microculture plate reader (Becton Dickinson) at 540 nm. Absorbance values were normalized to the values obtained for the vehicle-treated cells to determine the percent of cell viability. Each assay was performed in triplicate.

either cell line, control FOPFLASH values were about 10% of basal TOPFLASH levels. To confirm the ligand dependency of Tcf inhibition, we used AR deletion mutants (Figure 1b), including those coding the aminoand DNA-binding domain regions (Nt/DBD), as well as the DNA-binding region plus ligand-binding region (DBD/LBD). While the Nt/DBD mutant showed little ability to repress TOPFLASH, the DBD/LBD mutant was capable of a 3.5–4-fold repression in the presence of DHT. This suggests that in an AR overexpressed state, the ARNt is dispensable for Tcf repression. Further verification that the LBD is vital for repression is shown by increased relief of AR-mediated repression in cells treated with the pure AR antagonist Casodex (1 and 10 mm), which was able to efficiently relieve AR (DBD/LBD)-mediated repression of TOPFLASH. Having shown that repression is both AR and ligand dependent, we next evaluated the effect of increasing the concentration of DHT. First, we confirmed that our titration of DHT (0, 0.01, 0.025, 0.05, 0.075, 0.1, 1 and 10 nm) was an appropriate titration for the ARR3-Luc reporter. Both PC3s (~) and SW480 cells (’) showed similar trends of ARR3-Luc activation when transfected with AR and treated with increasing

Results Repression of Tcf is both AR and DHT dose dependent but can be relieved by Casodex To evaluate the repressive effect that AR and its physiological ligand had on Tcf signaling, several titrations were performed in PC3 cells. Specifically, we observed decreased TOPFLASH activity with increased amounts of transfected AR ( mg/three wells) with no detectable changes observed in either b-catenin or Tcf4 protein levels (Figure 1a). Basal levels of BCT (TOPFLASH) in untreated PC3 cells were low, while in SW480 cells luciferase counts were 6–8-fold higher. In

a 25

25

20

20

15

15

10

10

5

5

0

0 +

+

WT AR (ug)

0

DHT

+

TOPFLASH

PC3

30

RLU

RLU

b

PC3

30

+

+

+

+

+

1

2

+

+

4

8

0

0

TOPFLASH

+

+

+

+

+

+

+

+

-

+

AR (WT)

+

-

-

-

-

-

AR (Nt/DBD)

-

+

-

-

-

-

AR (DBD/LBD)

-

-

+

+

+

+

DHT

+

+

-

+

+

+

Casodex

-

-

-

-

1µM

10µ M

WB: AR WB: β-Cat WB:Tcf

d

c

90

SW 480=

50

PC3 =

40 30 20 10 0 0

0.01 0.025 0.05 0.075

[nm DHT]

0.1

1

10

TOP FLASH (RLU)

ARR3-Luc (RLU)

60

80

SW 480= PC3 =

70 60 50 40 30 20 10 0 0

0.01 0.025 0.05 0.075

0.1

1

10

[nm DHT]

Figure 1 (a) PC3 cells transiently transfected with AR and treated with 10 nm DHT for 24 h show inhibition of TOPFLASH activity in a dose-dependent manner, but does not affect b-catenin or Tcf4 protein levels. (b) The ARDBD/LBD deletion mutant was sufficient for Tcf repression, while ARLBD is required for repressive activity. Casodex can relieve repression in a dose-responsive manner (1 and 10 mm). (c) Increasing amounts of DHT (0, 0.01, 0.025, 0.05, 0.075, 0.1, 1 and 10 nm) promoted increased ARR3-Luc activity, while (d) the same titration of DHT decreased TOPFLASH activation both in SW480 and PC3 cells. (b–d), 1 mg of AR expression plasmid was transfected per well Oncogene

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5606

DHT concentration (Figure 1c). When these conditions were applied using the TOPFLASH reporter, we observed moderate diminishing of activity in PC3 cells (’, 2.53-fold) and SW480 cells (~, B4–4.5-fold) (Figure 1d). AR þ DHT-mediated repression can overcome b-catenin activating mutants and Wnt targets Using SW480 cells transfected (2 mg/six wells) with either WT b-catenin or an activated b-catenin mutant (D1– 130 bp) to remove the possible GSK3 phosphorylation site, we were able to show that AR plus 10 nm DHT has the ability to repress both forms of b-catenin-driven Tcf signaling with a slightly lesser efficiency for cells containing exogenous, activated b-catenin (Figure 2a). In addition, by using a cell line that is mutant for APC, SW480 cells (APCmut/mut), we verified that DHTmediated repression is distinct from the main route of b-catenin ubiquitination. Experiments using HCT116 cells which harbor activating b-catenin mutation (D codon 45 deletion) were also prone to Wnt repression in the presence of DHT (data not shown). In SW480 cells, the Wnt target, cyclin D1, was inhibited to a similar degree as TOPFLASH with and without the presence of exogenous b-catenin (Figure 2b). A control form of the cyclin D1 reporter (pA3-Luc) showed little fluctuation with treatments (data not shown). By increasing DHT concentration, we were also able to repress cyclin D1 protein levels but with little fluctuation in b-catenin protein levels (Figure 2c). AR showed increased stabilization and a slight increase in protein levels with increased DHT concentrations, while actin showed equal loads. Androgen-dependent localization of HcRed-Tcf with AR-EGFP in LNCaP cells Subsequent to the addition of DHT and the nuclear translocation of the AR fusion, we observed partial colocalization between HcRed-Tcf and AR-EGFP (Figure 3a). Using deconvolution microscopy (DECON) of a single xy plane, to reduce fluorescence contribution from different focal planes, and Northern Eclipse Image analysis software (Empix Imaging, Inc.)/Adobe Photshop 5.5, we calculated colocalization (as demarked with white arrowheads) of the AR and Tcf fusions to be 35%, with both proteins forming approximately 300 foci per nucleus. This value appeared to be consistent between cell lines (LNCaP, SW480 and PC3) subsequent to AR nuclear translocation. Colocalization (Figure 3bi+ii, arrowheads) was deemed as foci that contained yellow (Y, colocalization) rather than red (R) or green (G) (Figure 3biii), which were counted as no colocalization (Figure 3bi, arrows). In carrying out these experiments, LNCaP cells were cotransfected with either b-catenin-EGFP þ HcRed-AR (Figure 3ci) or b-catenin-EGFP þ HcRed-Tcf (Figure 3cii) and treated with or without 10 nm DHT for 48 h. Cell nuclei were evaluated for percentage fusion colors as a function of ligand treatment. Transfecting with b-catenin-EGFP þ Oncogene

Figure 2 (a) Transiently transfected SW480 cells with AR and either WT b-catenin or an activating b-catenin mutant (D1–130 bp) treated with 10 nm DHT showing TOPFLASH repression. (b) SW480 cells transfected with the cyclin D1 reporter show activation by WT b-catenin, but inhibition when AR is introduced. (c) Cyclin D1 protein levels decrease with increasing DHT concentration in AR-transfected SW480 cells, but no change in b-catenin levels was detected. AR levels showed slight stabilization with higher levels of DHT treatment. In all cases 2 mg of AR expression plasmid was transfected per six wells

HcRed-AR and treating with DHT, 62.0% of fusion color appeared yellow, while in the absence of androgen this value dropped to 14.5%, with green and red values changing from 15.5 to 69.0% and 22.5 to 16.5%, respectively. These data validate the ability of b-catenin-EGFP to become associated with HcRedAR in an androgen-dependent manner. We observed

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5607

Figure 3 (a) Distribution of HcRed-Tcf and AR-EGFP foci in transfected LNCaP cells treated with 10 nm DHT viewed in a single xy plane without and with deconvolution (DECON). (b) HcRed-Tcf and AR-EGFP foci colocalized (yellow) in approximately 35% of the fusion signal. Sites that did not colocalize (i, arrows) or did codistribute (i, ii, arrowheads) were used to quantitatively evaluate the degree of focal colocalization using Northern Eclipse Imaging software. Baseline values were set using non-colocalized (R, G) and colocalized (Y) foci (iii) to evaluate b-catenin-EGFP flux to either HcRed-AR or HcRed-Tcf with ( þ DHT) or without (DHT) 10 nm androgen. (c) LNCaP cells were transfected with either (Ci) (B-Cat-EGFP þ HcRed-AR) or (Cii) (B-Cat-EGFP þ HcRed-Tcf ). Cells were treated with or without 10 nm DHT and evaluated for the degree of colocalization (Y) as compared to non-colocalized (G or R). Bar ¼ 5 mm

reduced colocalization of b-catenin-EGFP with HcRedTcf (15.3% Y, 40.2% G, 44.5% R) upon treatment with 10 nm DHT, but an augmentation to 63% Y in cells treated only with vehicle. This is suggestive that bcatenin-EGFP has the capacity to reciprocate between AR and Tcf androgen-dependently. To account for partial overlapping, focal colors and varying number of foci per nucleus, total values of yellow, red or green per nucleus were quantitated as a relative percentage of total focal area per nucleus. Furthermore, a minimum of 10 representative nuclei per cell treatment were evaluated. Standard deviations were included to show statistical variation within each treatment.

Phosphorylated RNA Pol localization to HcRed-Tcf foci To provide support that HcRed-Tcf foci could contain sites of transcriptional activity, SW480 cells were transfected with HcRed-Tcf and stained for a phosphorylated form of RNA Pol II (phospho ser 2) in the presence of 10 nm DHT (Figure 4). Subsequent to deconvolution, we observed an overlap of signals (30%) suggesting that HcRed-Tcf sites or complexes have the capacity to be phosphorylated. These data suggest that these sites are not simply a result of an overexpressed plasmid, but that they may be physiologically and transcriptionally active. Interestingly, we observed little Oncogene

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5608

Figure 4 (a) Colocalization of Tcf-HcRed with phosphorylated (ser 2) RNA Pol II in the presence of DHT (SW480 cells) suggesting that approximately 30% foci could be sites of transcriptional activity. Bar ¼ 5 mm

difference in phospho RNA Pol II staining frequency with and without DHT (data not shown). Confirmation of expression and transcriptional activity of AR-EGFP and HcRed-Tcf Since we chose to evaluate the distribution of both AR and Tcf4 by way of fluorescent fusion vectors, confirmation of correct molecular weight (Figure 5a) and transcriptional activity (Figure 5b) was necessary. Transfections of the AR-EGFP into LNCaP cells showed an expected doublet signifying the endogenous (arrow) and the fusion form of AR (arrowhead) with increased levels present following androgen exposure (10 nm R1881). SW480 cells also showed a band at approximately 135 kDa (arrowhead), indicating a correctly migrating AR-EGFP for an anti-AR Western Blot (a, top panel). Furthermore, low but detectable levels of endogenous AR in both HCT116 (data not shown) and SW480 cells were apparent (arrow). Further, minor fluctuation of AR-EGFP expression occurred as a function of androgen exposure. Expression of endogenous Tcf in SW480 cells was recognized by the presence of a band at about 60 kDa and a tagged form approximately 25 kDa larger upon probing for Tcf protein. The fusion also expressed well in LNCaPs, although considerably lower levels of endogenous Tcf were detected. We further verified the functional activity of HcRed-Tcf in SW480 cells (Figure 5b, right) and observed augmentation of the TOPFLASH reporter, albeit to a slightly lesser extent than with the nontagged form, with only Oncogene

Figure 5 (a) Confirmation of expression and transcriptional activity of AR-EGFP and HcRed-Tcf fusion expression vectors in LNCaP cells and SW480 cells treated with and without 1 nm R1881. Bonds indicate Western Blots for either AR (top panels) or Tcf (lower panels). In transfected cells, the AR and Tcf fusions appear as doublets. (b) Transcriptional reporter assays show androgen-dependent activity of the AR fusion in LNCaP cells and little nonspecific (FOPFLASH) activation with the HcRed-Tcf plasmid in SW480 cells. (c) SW480 cells transfected with AR-EGFP and, as a control, the null HcRed vector (CMV-HcRed) illustrating that Tcf foci do not occur in the presence or absence (not shown) of androgen

slight activation of the mutant control. Similarly, we were able to verify that the AR-EGFP vector could activate the ARR3-Luc reporter in the presence of androgen again with only slightly less activity than the untagged form (Figure 5b, left). To rule out the possibility of false-positive or aggregating fluorescent proteins, we transfected LNCaPs with CMV-HcRed (no insert) either alone or with AR-EGFP. The CMVHcRed showed a distribution that was nonlocalized and without the formation of nuclear foci (Figure 5c). In vivo and In vitro binding assays are suggestive that AR and Tcf compete for nuclear b-catenin Transcriptional and morphological data strongly suggested the possibility of competition between Tcf and AR for a nuclear pool of b-catenin. To evaluate this possibility further, we used transcriptional assays with a WT and a mutant b-catenin-binding Tcf expression

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5609

Figure 6 Evidence from binding assays that AR and Tcf compete for b-catenin. (a) Increased amounts of transfected WT Tcf result in decreased ARR3-Luc activity in SW480 while the non-b-cateninbinding Tcf mutant (DNt Tcf ) did not show this result. Expression, both for the Wt and mutant Tcf, is shown by Western blot. (b) SW480 cells treated with DHT and immunoprecipitated for AR showed more associated b-catenin (arrow), while cells transfected and immunoprecipitated for HIS (Tcf-HIS) showed less associated b-catenin (star). Casodex (5 mm) was able to abrogate the liganddependent formation of an AR/b-catenin complex (solid arrowhead), while also reducing AR-mediated depletion of the Tcf/bcatenin complex (hollow arrowhead). Both DNt-Tcfmyc/b-cateninn and AR/TcfHIS complexes were weakly detected. (c) In vitro binding assays were immunoprecipitated for b-catenin-HIS with constant levels of TcfS35, b-catenin-HIS and increasing levels of ARS35. Higher levels of AR resulted in decreased TcfS35 associated with b-catenin-HIS/Ni-NTA bead complexes, suggesting that, in a pool containing these three species, at a certain concentration AR has the ability to compete for b-catenin. Non-programmed (NP) reticulolysate was used to maintain a constant total pool of lysate

plasmid (DNt Tcf ); coimmunoprecipitations in SW480 cells treated with and without androgen; and an in vitro binding and competition assay (Figure 6). Using the ARR3-Luc reporter, we found that increased levels of full-length Tcf resulted in decreased promoter activity,

while a DNt Tcf deletion mutant (non-b-catenin binding) did not display this trend (Figure 6a). Verification of expression by the WT and mutant Tcf is seen from a Western blot for Tcf and myc tag, respectively (Figure 6a, arrows). While increased Tcf protein levels were observed with the titration, we did not observe any fluctuation in AR levels in the presence of androgen. As part of our evaluation of whether AR and Tcf compete for nuclear bcatenin, we assayed whether we could detect an AR/Tcf complex. As a prerequisite, we verified that we could, in fact, detect an AR/b-catenin complex upon transfection with AR (Figure 6b, arrow). Consistent with our previous studies (Mulholland et al., 2002), we detected more AR/ b-catenin complex from cells treated with DHT, suggesting a ligand-sensitive interaction. We also detected an association with b-catenin and Tcf-HIS, but did not detect this association in the DNt Tcf-myc deletion mutant. Importantly, using this assay, interactions between AR and Tcf were very weak as compared to our detected AR/b-catenin and b-catenin/Tcf complexes both in prostate and colon cancer cells. Nonimmune, control precipitations were only slightly less than those of the AR/Tcf complex, suggesting that only a small fraction of AR associates with Tcf. We wanted to know whether Casodex, which abrogated both HcRed-Tcf focal accumulation (data not shown) and TOPFLASH activity, could alter the binding of AR and b-catenin. By treating cells with 5 mm Casodex (dissolved in EtOH) with and without 5 nm DHT, we observed reduced physical interaction between AR and b-catenin (arrowhead), while also relieving AR-mediated repression of the b-catenin/ Tcf-HIS complex (Figure 6b, star and open arrow heads). To evaluate whether a decreased b-catenin/TcfHIS complex could be directly affected by the AR TcfHIS and ARS35, in vitro translated products were used in series of binding assays with HIS-tagged recombinant b-catenin (Figure 6c). Using Ni-NTA beads to precipitate in vitro binding reactions, we observed that upon increased levels of ARS35 (a–d) there was a corresponding decrease in TcfS35 (a–e) precipitated with the HIS-tagged recombinant b-catenin. Although ARS35 levels up to 10 ml were added per binding reaction, we observed no further increase in detectable AR/b-catenin-HIS binding greater than 6 ml (point d) or 60% of input by volume. Total binding reaction volumes were kept constant using nonprogrammed (NP) reticulolysate. A volume of 10 ml of each probe is shown in Figure 6c (right), indicating that each probe is efficiently labeled and with molecular weights easily separated by SDS–PAGE. Ligand-occupied AR increases G1 arrest and reduces growth of SW480 cells With a reduction in cyclin D1 transcription and protein levels, we assessed whether this translated into altered cell cycle in SW480 cells. To evaluate this, we used FACs analysis of AR transfected and nontransfected SW480 cells treated with increasing concentrations of DHT (Figure 7a). In nontransfected cells, we observed a basal G1 level of 43.2%, which increased to 57% upon the addition of 10 nm DHT for 24 h. In cells transfected with Oncogene

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5610

activation both in prostate cancer cells and in colon cancer cells. We further validated the necessity of the AR ligand-binding domain by way of deletion mutant and followed that Casodex can relieve this inhibition. By using red and green fluorescent fusions in conjunction with in vitro and in vivo binding studies, we provide strong mechanistic data supporting a reciprocal balance of nuclear b-catenin between AR and Tcf sites. Specifically, we showed a novel distribution of Tcf and illustrated its partial colocalization with ligand-occupied AR by way of high-resolution microscopy and image analysis software. By costaining with phosphorylated RNA Pol II, we showed that Tcf foci are likely to be dynamic with the capacity for transcriptional activation. Increased b-catenin-EGFP localization to HcRed-AR sites, but decreased association with HcRed-Tcf foci in the presence of DHT, is strongly suggestive of a limited pool of nuclear b-catenin that has the capacity of shuttling to the more predominating transcription factor. Our data are also novel by linking transcriptional data to changes in the cell cycle and growth. By observing declined transcriptional levels of the Tcf, and the immediate Tcf target cyclin D1, we were able to link AR and DHT to cell cycle mitotic quiescence. We further showed that cell growth can be substantially reduced in cells with hyperactive b-catenin/Tcf signaling. Functional localization of AR, Tcf4 and b-catenin Figure 7 Effect of AR-mediated repression of Tcf signaling on cell cycle and growth. (a) SW480 cells were transfected with AR and treated with increasing DHT concentrations (0, 0.01, 0.025, 0.05, 0.075, 0.1, 1 and 10 nm). Cells transfected with AR and treated with DHT showed an increase of 26.2% in G1 population over non-ARtransfected cells. (b) In vitro cell viability as measured by MTT incorporation showed a decrease of 37% in cells containing exogenous AR and treated with 10 nm DHT over mock-transfected cells treated with vehicle. Casodex (5 mm) abrogated the repressive effects of AR/DHT on SW480 cell growth

AR, resting G1 was 53.1% which increased to 69.8% at maximum DHT concentrations. Therefore, cells containing AR þ DHT showed a G1 increase of approximately 26.4% over mock-transfected and untreated cells. Over the titration of DHT, sub-G1 levels remained relatively constant, suggesting that cell death or apoptosis was not occurring, while S phase population decreased as G1 levels increased. To evaluate whether detected arrest translated into decreased growth, the growth of SW480 cells was evaluated by in vitro mitogenic assay (Figure 7b). Assessing treatments of vehicle alone, 10 nm DHT, AR þ 10 nm DHT, and AR þ 10 nm DHT þ 1 mm Casodex, we observed counts that were consistent with transcriptional and cell cycle data. Treatment with androgen alone promoted a 12% decrease in cell growth, while cells transfected with AR and treated with DHT showed a reduction of 37% over cells treated mock transfected and treated with vehicle alone. Discussion In this study, we validated previous reports (Chesire and Isaacs, 2002) that the AR can repress Tcf transcriptional Oncogene

Previous reports have shown b-catenin to form nuclear morphologies that have included both punctate foci (Simcha et al., 1998) and rods (Simcha et al., 1998; Giannini et al., 2000). It has also been shown that AR forms a mixture of both transcriptional and nontranscriptional foci in the presence of agonist, as seen with high-resolution three-dimensional microscopy (Tomura et al., 2001). Other examples supporting a physiological role for foci include reports that nuclear b-catenin has the capacity to interact with and affect transcriptional regulation of the promyelocytic leukemia (PML) gene (Shtutman et al., 2002). In fact, recently, PML has also been shown to functionally associate with the AR (Rivera et al., 2003), a finding not surprising when considering what is known about the interaction of AR and b-catenin. It has been further suggested that cofactors such as CBP and SRC are necessary components of AR transcriptional foci and that most other steroid receptors form foci as a prerequisite to transcription (Saitoh et al., 2002). Proteins forming nuclear foci are thought to be localized to protein complexes in a highly dynamic manner, whereby exposure to ligands can alter their distributions (Saitoh et al., 2002; Shtutman et al., 2002). Therefore, given what we now know of the dynamic and transient nature of nuclear receptors and cofactors, it is also not surprising that the AR and Tcf could engage in a reciprocal balance of b-catenin. While colocalization data implies interaction, it is in fact, only informative to the micrometer level and does not necessarily mean protein–protein interaction. While we observed a degree of colocalization between AR-EGFP and HcRed-Tcf, we did not detect strong

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5611

physical interaction. Physical AR/b-catenin and Tcf/bcatenin complexes were, however, easily detected. Considering this, one possible interpretation for the close association of AR and Tcf is to facilitate sharing of common transcriptional coactivators and/or repressors including b-catenin. The combination of morphological data with in vitro and in vivo binding studies provides compelling evidence for nuclear competition for b-catenin. Wnt in prostate vs colon cancer cells In this study, we used colon cancer cells to evaluate the potential repressive ability of AR on Tcf/Lef signaling. Colon cancer cells, perhaps, provide the ideal means of testing the potential role of AR to interact with the Wnt pathway, as SW480 s have hyperelevated b-catenin/Tcf signaling, but low levels of endogenous AR. Constitutive Tcf signaling in colon cancer cells is due to defects in the Wnt pathway including defects in the APC protein and b-catenin deletions, resulting in a mainly nuclear b-catenin distribution (Easwaran et al., 1999). When comparing the effects of AR þ DHT on Tcf signaling, we observed similar degrees of repression in SW480 cells (B4.5-fold) and in PC3 cells (3.5–4-fold). The physiological significance of AR repression of the Wnt pathway was assessed using a cyclin D1 transcriptional reporter, cell cycle analysis and cell growth assays. In colon cancer cells, the cyclin D1 reporter showed an AR and DHT dose-dependent repression similar to TOPFLASH. Our data are corroborated by a recent report showing that the AR/b-catenin interaction is ligand androgen-dose-dependent (Song et al., 2003). Paradoxically, we were unable to observe a reduction in the cyclin D1 reporter in prostate cancer cells. We interpret this apparent discrepancy between colon cancer cells and prostate cells in several ways. In colon cancer cells, a predominating Wnt signaling pathway drives cyclin D1. In prostate cancer cells, Wnt signaling is much less prominent and, therefore, cyclin D1 activity is not constitutive. Furthermore, studies have shown cyclin D1 not to be a reliable marker of the AR (Tomura et al., 2001). This is unlike the situation in many breast cancer cell lines where cyclin D1 has been shown to be a reliable marker for cancer prognosis and has been shown to be correlated with estrogen receptor expression (Barnes and Gillett, 1998). In fact, some studies have shown that elevated cyclin D1 levels can in turn inhibit transcriptional activation of AR in a cell cycle-independent manner (Knudsen et al., 1999). An initial stimulation of cyclin D1 by AR could result in a feedback mechanism whereby cyclin D1 inhibits AR. Considering this, it is possible that an initial repression of cyclin D1 occurs (via the Wnt pathway), but is followed by cyclindependent inhibition of AR. While Tcf activity is not high in prostate cancer cells, the function of b-catenin as an AR transcriptional coactivator suggests an important role of the Wnt signaling in prostate cancer progression. Prostate cancer cells show indirect dysregulation of Wnt via PTENmut/mut resulting in elevated Akt levels, reduced GSK3 resulting in increased b-catenin (Cairns et al.,

1997; Davies et al., 1999), which can reside both in the cytosol and in the nucleus (Chesire et al., 2000). Therefore, lesser Tcf activity in prostate cells could be accounted for by distribution of b-catenin, rather than the total amount of cellular b-catenin (Persad et al., 2001). Considering this, molecules that are capable of changing b-catenin distribution, such as the AR (Mulholland et al., 2002), would have the capacity to alter Wnt signaling. With this large pool of cytosolic b-catenin, there is increased opportunity to associate with other binding partners such as the AR which, in the presence of DHT, will cotraffic to the nucleus (Mulholland et al., 2002). AR-driven nuclear b-catenin could then also be directed to AREs and potentially deplete Tcf-binding sites. AR-mediated repression of Tcf is a nuclear event Several lines of evidence suggest that AR-mediated repression is a nuclear event. (1) Cells transfected with an activating b-catenin mutant showed a fivefold increase in TOPFLASH activity over basal levels both in prostate and colon cancer cells (data not shown). However, even under these circumstances, introduction of AR þ DHT could efficiently reduce Tcf signaling. The fact that Tcf repression occurs when AR accumulates in the nucleus, that is, when AR is exposed to DHT, is suggestive of a nuclear event for Tcf silencing rather than a degradative one. (2) By using a cell line that is APCmut/mut, we rule out a main mode of b-catenin degradation (ubiquitination). If the actions of androgen involve targeting cytoplasmic b-catenin for proteosome degradation, we would have observed changes in levels of cytoplasmic, nonphosphorylated b-catenin in cells containing functional APC. (3) Despite the fact that Casodex interferes with AR transcription, it has been well documented to promote nuclear localization of the AR. With the addition of Casodex, we observed a dosedependent relief of AR/DHT-mediated Tcf repression. (4) The reduction of AR transcription (in the presence of DHT) upon overexpression of Tcf provides evidence of competition. These data are consistent with the notion that modulation of the b-catenin-Tcf/Lef assembly may be the mechanism by which AR exerts its repression on Tcf/Lef signaling. To test this further, we evaluated the amount of b-catenin associated with the Tcf/Lef complex with and without AR þ androgen, both in vivo and in vitro. Data achieved with these data corroborate morphological data and our overall hypothesis. Other nuclear receptors and Tcf/Lef signaling It is interesting how two stimulators of cell proliferation (AR/DHT and Wnt) can interact in a repressive manner. One possible interpretation is that androgens may not increase proliferation by, rather, promoting growth and differentiation in prostate epithelia. Prostate cancer cells have cell cycle deregulation compounded by a cell survival response to androgens. The contributions of Wnt signaling in prostate cancer are likely complex, although it is clear that nuclear b-catenin can serve as a Oncogene

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5612

potent AR coactivator. We suggest a scenario whereby b-catenin could be shuttling between Tcf/Lef-binding sites and AR elements. In the absence of androgen, AR resides mainly in the cytosol, while nuclear b-catenin associates with Tcf (Figure 8a). In the presence of androgen, b-catenin could be shuttled by translocating nuclear receptors to both Tcf- and AR-associated response elements to promote coactivation (Figure 8b). Consequently, less b-catenin would be associated with Tcf and more with AR. This would simultaneously lower Tcf activity and augment AR transactivation. Such a model is also considerate of other reports of nuclear receptor repression of Tcf, including the RAR (Easwaran et al., 1999) and the thyroid receptor (Miller et al., 2001), two hormone receptors that are expressed constitutively in the nucleus. In conclusion, our data provide strong support that the AR can inhibit the b-catenin/Tcf signaling axis by competing away nuclear b-catenin. Our findings both corroborate initial reports (1) showing that AR can inhibit Tcf transcriptional activity and augment these observations with mechanistic data showing that Tcf and AR entertain a reciprocal balance of nuclear b-catenin. While Wnt signaling is clearly different, in many aspects, between prostate and colon cancer cells, it is possible that ligand-occupied AR could be important therapeutically in cancers with hyperactive Tcf/Lef transactivation or where this pathway contributes to cancer progression. Interestingly, some reports have even shown differences in AR expression levels between cancerous (less AR) and

Figure 8 Potential mechanism accounting for nuclear hormonemediated repression of b-catenin/Tcf signaling. Subsequent to ARmediated nuclear entry of b-catenin, competition between AR and Tcf could occur for nuclear b-catenin. (a) In the absence of androgen, cells with activated Tcf signaling could show activation of Wnt targets and increased growth. In the presence of androgens, the Tcf/b-catenin signaling axis would be lessened due to depletion of b-catenin, while AR-mediated targets would be activated

noncancerous (more AR) colon tissue (Catalano et al., 2000). It is tempting to speculate that pathologies with low levels of RAR (Chesire et al., 2002), thyroid receptor (Miller et al., 2001) or AR could be a prosurvival adaptation for cancer cells to allow for increased Tcf signaling and, therefore, growth and differentiation.

References Barker N, Morin PJ and Clevers H. (2000). Adv. Cancer Res., 77, 1–24. Barnes DM and Gillett CE. (1998). Breast Cancer Res. Treat., 52, 1–15. Cairns P, Okami K, Halachmi S, Halachmi N, Esteller M, Herman JG, Jen J, Isaacs WB, Bova GS and Sidransky D. (1997). Cancer Res., 57, 4997–5000. Carlsson P, Waterman ML and Jones KA. (1993). Genes Dev., 7, 2418–2430. Catalano MG, Pfeffer U, Raineri M, Ferro P, Curto A, Capuzzi P, Corno F, Berta L and Fortunati N. (2000). Int. J. Cancer, 86, 325–330. Chen RH and McCormick F. (2001). Cancer Res., 61, 4445–4449. Chesire DR, Ewing CM, Gage WR and Isaacs WB. (2002). Oncogene, 21, 2679–2694. Chesire DR, Ewing CM, Sauvageot J, Bova GS and Isaacs WB. (2000). Prostate, 45, 323–334. Chesire DR and Isaacs WB. (2002). Oncogene, 21, 8453–8469. Damm K, Thompson CC and Evans RM. (1989). Nature, 339, 593–597. Davies MA, Koul D, Dhesi H, Berman R, McDonnell TJ, McConkey D, Yung WK and Steck PA. (1999). Cancer Res., 59, 2551–2556. Easwaran V, Pishvaian M, Salimuddin and Byers S. (1999). Curr. Biol., 9, 1415–1418. Giannini AL, Vivanco MM and Kypta RM. (2000). Exp. Cell. Res., 255, 207–220. Gleave M, Hsieh JT, Gao CA, von Eschenbach AC and Chung LW. (1991). Cancer Res., 51, 3753–3761. Oncogene

Helmbrecht K, Kispert A, von Wasielewski R and Brabant G. (2001). Endocrinology, 142, 5261–5266. Ikonen T, Palvimo JJ and Janne OA. (1997). J. Biol. Chem., 272, 29821–29828. Janne OA, Moilanen AM, Poukka H, Rouleau N, Karvonen U, Kotaja N, Hakli M and Palvimo JJ. (2000). Biochem. Soc. Trans., 28, 401–405. Knop M, Barr F, Riedel CG, Heckel T and Reichel C. (2002). Biotechniques, 33, 592–594, 596–598 passim. Knudsen KE, Cavenee WK and Arden KC. (1999). Cancer Res., 59, 2297–2301. Miller LD, Park KS, Guo QM, Alkharouf NW, Malek RL, Lee NH, Liu ET and Cheng SY. (2001). Mol. Cell. Biol., 21, 6626–6639. Mulholland DJ, Cheng H, Reid K, Rennie PS and Nelson CC. (2002). J. Biol. Chem., 277, 17933–17943. Pawlowski JE, Ertel JR, Allen MP, Xu M, Butler C, Wilson EM and Wierman ME. (2002). J. Biol. Chem., 277, 20702–20710. Persad S, Troussard AA, McPhee TR, Mulholland DJ and Dedhar S. (2001). J. Cell. Biol., 153, 1161–1174. Polakis P. (2001). Cell, 105, 563–566. Quigley CA, De Bellis A, Marschke KB, el-Awady MK, Wilson EM and French FS. (1995). Endocr. Rev., 16, 271–321. Rivera OJ, Song CS, Centonze VE, Lechleiter JD, Chatterjee B and Roy AK. (2003). Mol. Endocrinol., 17, 128–140. Saitoh M, Takayanagi R, Goto K, Fukamizu A, Tomura A, Yanase T and Nawata H. (2002). Mol. Endocrinol., 16, 694–706.

AR and Tcf competition for nuclear b-catenin DJ Mulholland et al

5613 Sap J, Munoz A, Schmitt J, Stunnenberg H and Vennstrom B. (1989). Nature, 340, 242–244. Schulman IG, Juguilon H and Evans RM. (1996). Mol. Cell. Biol., 16, 3807–3813. Shtutman M, Zhurinsky J, Oren M, Levina E and Ben-Ze’ev A. (2002). Cancer Res., 62, 5947–5954. Simcha I, Shtutman M, Salomon D, Zhurinsky J, Sadot E, Geiger B and Ben-Ze’ev A. (1998). J. Cell. Biol., 141, 1433–1448. Snoek R, Rennie PS, Kasper S, Matusik RJ and Bruchovsky N. (1996). J. Steroid. Biochem. Mol. Biol., 59(3–4), 243–250.

Song LN, Herrell R, Byers S, Shah S, Wilson EM and Gelmann EP. (2003). Mol. Cell. Biol., 23, 1674–1687. Tomura A, Goto K, Morinaga H, Nomura M, Okabe T, Yanase T, Takayanagi R and Nawata H. (2001). J. Biol. Chem., 276, 28395–28401. Truica CI, Byers S and Gelmann EP. (2000). Cancer Res., 60, 4709–4713. Yang F, Li X, Sharma M, Sasaki CY, Longo DL, Lim B and Sun Z. (2002). J. Biol. Chem., 277, 11336–11344. Yeh S, Chang HC, Miyamoto H, Takatera H, Rahman M, Kang HY, Thin TH, Lin HK and Chang C. (1999). Keio. J. Med., 48, 87–92.

Oncogene