The Androgen Receptor Can Promote (cid:1) -Catenin Nuclear Translocation Independently of Adenomatous Polyposis Coli*

We provide evidence that the androgen receptor (AR) can promote nuclear translocation of (cid:1) -catenin in LNCaP and PC3 prostate cancer cells. Using AR-expressing cells (LNCaP) and non-AR-expressing cells (PC3) we showed by time course cell fractionation that the AR can shuttle (cid:1) -catenin into the nucleus when exposed to exogenous androgen. Cells exposed to the synthetic androgen, R1881, show distinct, punctate, nuclear co-localization of the AR and (cid:1) -catenin. We further showed that the AR does not interact with adenomatous polyposis coli or glycogen synthase kinase-3 (cid:1) and, therefore, conclude that androgen-mediated transport of (cid:1) -catenin occurs through a distinct pathway. The minimal necessary components of the AR and (cid:1) -catenin required for binding nuclear accumulation of (cid:1) -catenin nuclear import appears to be the DNA/ligand binding regions and the Armadillo repeats of (cid:1) -catenin. We also employed a novel DNA binding assay to illustrate that (cid:1) -catenin has the capacity to bind to the probasin promoter in an AR-dependent manner. The physiological relevance of AR-mediated transport of (cid:1) -catenin and binding to an AR promoter appeared to be a substantial increase in AR transcriptional reporter activity. AR-mediated import represents a novel mode of nuclear accumulation of (cid:1) -catenin. The Arm repeats of (cid:1) -catenin binding sites

N-terminal domain (AR Nt ), which is least conserved (5)(6)(7). The AR Nt contains a ligand-independent transcriptional activating function whereas the AR Ct contains one that is ligand-dependent (8). The ligand binding domain of nuclear receptors interact with a variety of other proteins following ligand binding (9), which has the potential to augment or modulate transcriptional response. The transcriptional activity of the AR is largely determined by the presence or absence of other co-factors, including co-activators, which enhance AR activity, and corepressors, which repress AR activity. Examples of previously identified co-activating molecules of the AR include CBP, SRC1, and TIF-2 (10 -12).
There is strong documentation to suggest steroid receptor shuttling upon exposure to the cognate ligand. Such studies have pertained to the AR (13)(14)(15)(16), glucocorticoid receptor (GR) (17), estrogen receptor (ER) (19), mineralocorticoid receptor (20), and thyroid receptor (TR) (21). These receptors show a certain degree of trafficking either to or from the nucleus but also in a subnuclear fashion. Those that show a strong migration to the nucleus upon exposure to ligand are termed "translocating receptors" and can be contrasted with receptors that are constitutively nuclear (4). The ER shows expression that is mainly nuclear in the absence of ligand (22), whereas the AR and GR show a distribution that is both cytoplasmic and nuclear (23,22). Curiously, there are varying reports as to relative abundance of AR in the cytoplasm (24) and in the nucleus (6) in many cell types. Upon receptor stimulation by DHT, or potent analogues of DHT, including R1881, the AR will dissociate from heat-shock proteins, translocate to the nucleus, and form transcriptionally active DNA-protein complexes (4). In general, this two-step model for steroid hormone receptor action can be applied to the AR and GR whereby the unliganded receptor is localized in the cytoplasm and upon ligand binding undergoes conformational change that permits translocation to the nucleus. This homodimerization leads to initiation of target gene regulation (15). Within the nucleus most members of the receptor superfamily form focal accumulations within the nucleus in the presence of ligand (4).
Proteins that are carried or shuttled with steroid receptors into the nucleus have not been thoroughly explored. Therefore, with the ability of the AR to translocate to the nucleus, we hypothesize that the AR could co-traffic other molecules to the nucleus. Examples of this phenomenon are few if any. Cytoplasmic to nuclear and subnuclear trafficking could allow for formation of multiprotein-DNA complexes and AR transcriptional activation (15). Given the documented ligand-dependent relationship between the AR and ␤-catenin (25), we hypothesize that ␤-catenin could be part of a complex that translocates to the nucleus as a pre-requisite to forming transcriptionally active, nuclear complexes.
␤-Catenin is a multifunctional protein. Specifically, it plays a central role in cell adhesion by its association with E-cadherin * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
An intriguing feature of Wnt signaling is the manner in which ␤-catenin is actively translocated to the nucleus. Although ␤-catenin does not have a nuclear localization signal (32), APC (adenomatous polyposis coli) can act as a nuclearcytoplasmic shuttling protein (32). Such studies have shown that alteration of the amino nuclear export sequence on APC could accumulate nuclear ␤-catenin and concluded that APC can shuttle between the nucleus and cytoplasm while directing ␤-catenin to functionally important locations. APC also contains two nuclear localization signals that are necessary for optimal nuclear APC activity (33) and likely for its tumor suppressor function. Recently, studies have shown that nuclear export of ␤-catenin can occur independent of the CRM1 export protein and suggested that there could be alternative pathways associated with ␤-catenin transport (34). Previous studies have also shown that ␤-catenin can localize to the nucleus independent of the shuttling protein Ran (35).
Little is known about the functional contribution of the recently identified ligand-dependent interaction between ␤-catenin and the AR. In this study, 1) we used confocal microscopy and a novel DNA binding assay to provide evidence that ␤-catenin binds in an AR ligand-dependent manner, via the AR, to an androgen-regulated promoter; 2) we defined the domains of the AR and ␤-catenin as they related to protein interactions and related this to transcriptional activity; 3) we demonstrated that AR can translocate ␤-catenin to the nucleus in an AR liganddependent fashion as a distinct pathway that is independent of APC; 4) finally, we identified the structural components of the AR and ␤-catenin that are necessary and sufficient for co-translocation.
Plasmids-AR luciferase reporter, AR, and GR expression constructs were constructed as previously described (Snoek et al. (8)). ␤-catenin cDNAs were obtained both from Dr. Berry Gumbiner (Memorial Sloan-Kettering Cancer Center, New York, NY) and Dr. Randall Moon (Howard Hughes Medical Institute, University of Washington). All remaining receptor expression constructs were obtained from Dr. Ronald Evans (Salk Institute, La Jolla, CA).
Immunofluorescence-Cells were grown in 5% FBS in RPMI on glass coverslips, fixed in cold methanol for 10 min, and air-dried. Cells were reconstituted in 4% normal serum in 0.1% Tween 20/phosphate-buffered saline for 20 min. Primary polyclonal ␤-catenin (sc-8199) antibodies and monoclonal AR (DNA binding domain, 15071A) antibodies were used at a dilution of 1:100 and incubated for 1-2 h at 37°C followed by 3 ϫ 10-min washes. Secondary antibodies conjugated to fluorophores were used at a 1:100 dilution and were incubated for 1 h at 37°C followed by 3 ϫ 10-min washes. Coverslips were mounted with mounting media containing DAPI (Vector Laboratory) on glass slides. Confocal microscopy images were obtained used a Bio-Rad 1024 system and were digitally compiled using IMAGE software (National Institutes of Health).
Cell Fractionation and Time Course Study-Subsequent to transfection and a least 12-h growth in charcoal-stripped serum, LNCaP, PC3, and HeLa FLAG-AR cells were treated with 10 nM ligand for up to 60 min. Prior to harvest, cells were washed once in cold PBS and separated into cytoplasmic and nuclear fractions using the Nuclear and Cytoplasmic Extraction Reagent (Pierce) at 10-min intervals. Fractions were assayed for total protein using the BCA protein assay (Pierce, 23223).
In Vitro Translation-Plasmid cDNA was transcribed and translated in vitro using the rabbit reticulocyte lysate TnT (Promega) system. A standard reaction was used consisting of 40 l of TnT Quick Master (SP6 or T7 promoter), 2 l of cold 2 mM methionine (1000 Ci/mol at 10 mCi/ml), 1-2 g of plasmid DNA made up to a final volume of 50 l with nuclease-free water. Reactions were incubated at 30°C for 90 min and either used immediately for binding reactions or stored at Ϫ20°C. The efficiency of in vitro translation reactions were assessed by monitoring [ 35 S]methionine incorporation by SDS-PAGE and radiography following equilibration in ENHANCE (Cat. NEF981, PerkinElmer Life Sciences). Gels were washed 2 ϫ 10 min with dH 2 O, dried, and exposed to radiographic film.
GST-Pull Downs-Recombinant proteins were labeled with [ 35 S]methionine (Promega TnT) in a volume of 50 l. GST-bound beads were equilibrated with binding buffer (20 mM HEPES, pH 7.6, 150 mM KCl, 5 mM MgCl 2 , 1 mM EDTA, 0.05% Nonidet P-40, and protease inhibitors) and incubated with lysate for 2 h at 4°C. Complexes were washed 4ϫ with 1 ml of cold binding buffer, boiled in Laemmli sample buffer, and separated by SDS-PAGE. Gels were enhanced, dried, and exposed to film.
Acrydite Capture of DNA-binding Complexes-The ACDC assay (36) was carried out by incubating recombinant His-tag androgen receptor DNA binding domain (His-tag AR DBD ) without DNA or with NF-1 acrydite binding sites or ARE acrydite binding sites (ARE-ac). The nuclear factor-1 (NF-1) acrydite served as a negative binding control. Binding reactions were polymerized in the well of a 5% polyacrylamide acrydite gel, run on a 15% SDS-PAGE and probed with anti His-tag antibody. LNCaP nuclear extracts (NE) treated with 10 nM R1881 were incubated with ARE-ac and polymerized inside the well of a 5% polyacrylamide gel. After electrophoresis, the specifically bound protein was extracted from the acrydite gel, separated by SDS-PAGE (8.5%), and probed with an anti-AR antibody (DBD epitope). NE were incubated without DNA or with acrydite probasin (P␤-ac) promoter (Ϫ286 to ϩ28) or the pBluescript acrydite multiple cloning site (MCS-ac) created by PCR amplification using specific primers with a 5Ј-acrydite moiety attached.
Luciferase Reporter Assays-PC3 and LNCaP cells were plated in 6-well dishes and incubated overnight at 37°C. Cells were transfected with a total of no more than 4 g of DNA per well for 8 -16 h using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. After transfection, cells were incubated in charcoal-stripped serum containing 10 nM R1881, dexamethasone, estradiol, all-trans-retinoic acid, or tri-iodothyronine for at least 16 h. All assays were carried in at least triplicate and evaluated using the Dual Luciferase Reporter system (Promega).

Ligand-dependent Co-localization of ␤-Catenin and AR-
The localization of AR ( Fig. 1, ai) and ␤-catenin (aii) was viewed with confocal microscopy using antibodies to the DNA binding domain of the AR and Arm repeats/Ct portion of ␤-catenin in fixed LNCaP cells that were treated with and without R1881 for 60 min (Fig. 1). Images captured by confocal microscopy and compiled in images analysis software suggest that in the absence of androgen the AR (bi) stained diffusely throughout the cell, whereas ␤-catenin staining (bii) was predominately found at cell borders, diffusely in the cytoplasm, and within the nucleus. In the presence of ligand we observed a specific, punctate staining pattern of AR (ai), which was strongly nuclear. ␤-Catenin also showed nuclear localization (aii) in cells upon ligand addition and co-localized in many instances with AR (arrowheads) whereas in some instances it did not (arrows). Levels of ␤-catenin at cell borders were not observed to change in response to AR ligand, whereas cytoplasmic levels appeared to decrease moderately. Nuclei were stained using DAPI (Fig.  1, aiv and biv). AR DBD/LBD Interacts with the ␤-Catenin Arm Repeats-As a prerequisite to assessing the trafficking properties of the AR, we determined the relative affinity of binding between the AR and ␤-catenin. To do this we used recombinant AR proteins, including AR Nt , AR DBD/LBD , AR DBD , and AR Nt/DBD (Fig. 2a). ␤-Catenin 35 S-labeled deletion mutants included Nt-(1-140), Arm repeats-(140 -664), and Ct-(664 -781) domains (Fig. 2b) translated in vitro from Xenopus expression constructs. Although relatively weak interactions were detected between the ␤-Cat Nt , or ␤-Cat Ct , and any portion of the AR (only slightly greater than the GST-negative control), strong interactions were detected between the ␤-catenin Arm repeats and AR (Fig.  2c). The Arm repeats showed moderate interactions with the AR DBD but increased affinity with the AR DBD/LBD suggesting a role both for the DBD and LBD in binding AR and ␤-catenin. Despite the differences detected in our physical mapping using our recombinant truncations we were unable to detect a significant difference (Ͻ10%) between AR-GST/␤-catenin binding affinity in the presence or absence of ligand. We have attributed this to potential masking of interaction sites by chaperone proteins in the cell in absence of ligand, which is not present in the recombinant system. In general, these data support transcriptional assays, immunoprecipitation studies, and shuttling time courses.
Ligand-bound AR Promotes Nuclear Translocation of Cytosolic ␤-Catenin-To determine if ␤-catenin was able to translocate to the nucleus in an androgen-dependent manner, we performed a series of cell fractionations on LNCaP cells and HeLa-FLAG AR cells over a time course of androgen exposure (Fig. 3). Cytoplasmic and nuclear compartments were efficiently separated as determined by probing nuclear fractions for pan-Cadherin and probing cytoplasmic fractions for total histone (Fig. 3A). The low levels of pan-Cadherin in nuclear fractions and histone in the cytosol suggested an efficient separation of cellular compartments. We further demonstrated equal total protein loads through a time course by assessing total cytosolic actin and nuclear histone (Fig. 3A). Non-transfected LNCaP cells that were treated with R1881 and were fractionated into cytoplasmic and nuclear compartments at 10-min intervals over 60 min (10,20,30,40,50, and 60 min) showed a moderate accumulation of endogenous, nuclear ␤-catenin (Fig. 3B). We observed a similar increase in nuclear AR. Densitometry evaluation of nuclear accumulation of AR and ␤-catenin suggest a similar rate of increase. Loading controls (actin) remained constant (Fig. 3Bii). A less noticeable decrease in cytosolic ␤-catenin was observed when compared with changes in nuclear ␤-catenin levels. This is likely accounted for by the large amount of ␤-catenin that remained stationary; that is, that does not migrate upon addition of AR ligand. Interestingly, the shift in ␤-catenin found in cytosolic and nuclear compartments was considerably more noticeable in LNCaP cells, which were transiently transfected with tagged expression constructs for ␤-catenin (Fig. 3C). Specifically, by transfecting LNCaP cells with either HA-tagged and myctagged ␤-catenin constructs, we observed a greater amount of nuclear accumulation of the de novo synthesized ␤-catenin as compared with non-transfected cells. Densitometry analysis (Fig. 3Cii) showed a similar trend in nuclear accumulation of ␤-catenin and illustrated a plateau at 50-min post-ligand addition suggesting a decreased rate of nuclear import. We further chose to consider whether AR-dependent movement of ␤-catenin could occur using other AR ligands, including the physiological androgen, DHT. DHT, like its analogue, R1881, promoted nuclear accumulation of both the AR and ␤-catenin although to a slightly lesser extent than with R1881 (data not shown). We next investigated whether other non-prostate can- In the absence of R1881, AR staining (bi) appears diffuse and throughout the cytoplasm whereas that for ␤-catenin (bii) is localized at the cell membrane, in the cytosol, and in the nucleus. When serial images collected by confocal microscopy were digitally compiled, the distribution of the AR appeared punctate and nuclear (ai). Similarly, ␤-catenin (aii) showed a punctate pattern that co-localized in some instances (arrowheads) whereas in others (arrows) did not (overlay ϭ aiii, biii; DAPI ϭ aiv, biv) (bar ϭ 5 m).
cer cell lines, including stably AR-FLAG-transfected HeLa cells could mediate AR-mediated ␤-catenin nuclear translocation. To do this we isolated cell fractions at 0 and 60 min post-ligand addition and immunoprecipitated AR, FLAG, and ␤-catenin (Fig. 3D). We observed a trend of nuclear accumulation of ␤-catenin similar to that in LNCaP cells, although ␤-catenin movement from the cytosol was not as apparent as neither was the overall movement of the FLAG-AR. We further assessed the relative amounts of nuclear and cytosolic AR-associated ␤-catenin in these cells by immunoprecipitation of FLAG tagged AR (Fig. 3D). In the absence of androgen, there were greater amounts AR in the cytosol as judged by anti-FLAG and anti-AR antibodies whereas greater amounts in the nuclear fractions of cells were treated with ligand. We found detectable ␤-catenin with AR and FLAG-AR immunoprecipitates in the absence of androgen confirming the presence of a constitutive interaction as observed with LNCaP immunoprecipitations (Fig. 3D). We further demonstrated that there were detectable amounts of the AR/␤-catenin complex in nuclei without androgen but substantially higher in its presence (Fig. 3D).
The AR DBD/LBD Is Necessary and Sufficient for Nuclear Translocation of ␤-Catenin in PC3 Cells-Having found evidence that the AR can translocate ␤-catenin to the nucleus in LNCaP and HeLa cells, we chose to ascertain whether other nuclear receptors have this capability in PC3 prostate cancer cells. To do this we used a prostate cancer cell line that dose not express the androgen receptor. When PC3 cells were transiently transfected with RAR (Fig. 4A), ER (Fig. 4B), GR (Fig.  4C), TR (Fig. 4D), or ␤-catenin (HA-tag) we observed an inability to move ␤-catenin to the nucleus with ligand exposure. Although the GR shows an ability to translocate to the nucleus upon exposure to ligand, co-trafficking with ␤-catenin was not detected in this cell line despite minor fluctuation in GR cellular levels.
With ␤-catenin translocation being detected most pronounced with the AR we chose to assess ␤-catenin movement using various AR deletion mutants (Fig. 4, E-Hi). In PC3 cells the empty, control expression vector (PRC/CMV) and that of the amino terminus (Nt) of the AR showed little ability to move ␤-catenin to the nucleus (Fig. 4F). However, constructs expressing the AR Nt/DBD showed a more prominent cellular expression but did not show ligand-dependent changes in nuclear distribution of itself or ␤-catenin as a function of ligand (Fig.  4G). When cells were transfected with constructs expressing both the AR LBD and the AR DBD ligand-dependent translocation of ␤-catenin was readily apparent (Fig. 4Hi). This indicated to us the necessity of both the DNA binding domain and the ligand binding domain of the AR for efficient ␤-catenin nuclear translocation. Densitometry indicated a coincident movement between the AR and ␤-catenin. Additionally, accumulation in PC3 cells was similar to that in LNCaPs in that both appeared to plateau at 50 -60 min post-ligand addition. We further demonstrated that the Arm repeats are both necessary and sufficient for AR-dependent nuclear translocation of Catenin. Constructs expressing the Nt and Ct components of Catenin showed little, if any, fluctuation between compartments whereas the Arm repeats showed accumulation to the nucleus over 60 min (Fig. 4I).
AR-mediated Translocation of ␤-Catenin Is Distinct from APC/GSK3-To determine if AR-mediated import of ␤-catenin was a distinct pathway, we assessed whether the AR had the ability to interact with APC and GSK3 in LNCaPs (whole cell lysates) treated with and without R1881. As controls we probed AR immunoprecipitations for AR (Fig. 5a) and ␤-catenin (Fig.  5b). We detected little fluctuation in AR levels as a function of hormone treatment but observed a ligand sensitive interaction between AR and ␤-catenin, which is more ␤-catenin-associated with AR in the presence of androgen. Immunoprecipitated GSK3 showed a distinct band at about 45 kDa but was not detected in AR immunoprecipitates. Although APC immunoprecipitates showed some degradation at ϳ180 and 200 kDa, major species were detected at 300 kDa. These immunoprecipitates also did not contain detectable AR either with or without ligand.
␤-Catenin and the AR Complex Directly on the Probasin Promoter-Acrydite experiments operate on the premise that protein specifically bound to a DNA sequence during a binding reaction will remain intact after being exposed to a electrophoretic current. Such species were then denatured and separated by SDS-PAGE. Our results showed that the androgen receptor from nuclear receptor extracts treated with R1881 bound specifically to ARE-acrydite (ARE-ac) PCR products (Fig. 6). This was verified by the presence of the HIS-tag component of the AR DBD in the elution from the ARE-ac binding reaction (Fig. 6a). AR antibodies also reacted strongly with androgen response elements and with only a small amount of nonspecific binding with negative controls, including a binding reaction for NF-1 and binding reactions without DNA (Fig.  6bi). Fig. 6bii indicates the extent of recovery of the androgen receptor in nuclear extract binding reactions and was measured by comparing the reactivity with antibodies toward the DNA binding region of the AR with that which is specifically bound to the ARE-ac DNA sequence. Fig. 6c suggests that, when binding reactions were probed for ␤-catenin using a promoter sequence known to contain four cooperative ARE (Ϫ286/ ϩ28), high amounts of ␤-catenin were detected. Only small amounts of ␤-catenin were detected in negative control binding reactions, including the multiple cloning site-acrydite PCR reaction (MCS-ac) or binding reactions without DNA. These data strongly support the presence of a ligand-dependent transcriptionally active AR/␤-catenin complex directly associated with the probasin promoter. Treating LNCaP nuclear extracts with 2 M NaCl and separating the non-extracted nuclear components by SDS-PAGE show a greater amount of both AR and ␤-catenin retained in the presence of ligand (Fig. 6d). This suggests that both are incorporated into the nuclear matrix to a greater extent in the presence of ligand.
The Arm Repeats and AR LBD Are Sufficient for Ligand-dependent Interaction of AR and ␤-Catenin-Using PC3 cells we sought to evaluate the various components of the AR and ␤-catenin that were necessary for ligand-dependent transcriptional activation. To do this, we transfected AR and ␤-catenin truncations into PC3 cells and monitored ARR3-Luc activity as a function of increasing ␤-catenin. We observed that AR truncations containing only Nt provided little transcriptional response whereas AR constructs expressing the Nt plus the DBD showed constitutive activation and some augmentation by ␤-catenin. Cells transfected with both the LBD and the DBD showed a dose-and ligand-dependent transcriptional response to increasing ␤-catenin (Fig. 7A). This suggests the requirement of the AR LBD for ligand-dependent co-activation of the AR by ␤-catenin as well as for AR-mediated translocation of ␤-catenin. Cells transfected with only the AR Nt or control vector (PRC/CMV) showed relatively little transcriptional activity. Transfections using constructs coding for only the Arm repeats (Fig. 7B) were still able to enhance the AR transcriptional response using the ARR3-Luc reporter in a ligand-dependent manner but not to the same extent as the full-length ␤-catenin. Deletion of the Arms diminishes the ability of ␤-catenin to augment AR transcription considerably. DISCUSSION Androgen receptor trafficking of ␤-catenin is a particularly attractive hypothesis for several reasons. First, ␤-catenin does not have an identified nuclear localization signal making it dependent upon other chaperone molecules for nuclear import. Second, the nuclear interactions between the AR and ␤-catenin are ligand sensitive. Third, AR-mediated transport of ␤-catenin appears to be distinct from APC transport of ␤-catenin. Fourth, transcriptional activity of AR promoters is augmented with increased levels of transfected ␤-catenin when in the presence of ligand.
In this study, we provided evidence for a novel and distinct mechanism by which ␤-catenin can enter the nucleus of AR expressing cancer cell lines. We showed that upon exposure to androgen the AR is capable of shuttling ␤-catenin to the nucleus, thus providing a means by which ␤-catenin can augment the transcriptional activity of AR reporters. This mode of import appears to be functionally independent of the ␤-catenin transporter protein, APC. Immunofluorescence studies strongly suggest co-localization of endogenous AR and ␤-catenin at punctate complexes in the nucleus of LNCaP prostate cancer cells in the presence of R1881. Co-immunoprecipitation studies suggest a cytosolic AR/␤-catenin interaction but not with APC or GSK3. Further evidence for an AR/␤-catenin complex that can directly bind to an AR-regulated promoter was provided using a novel DNA binding assay. Finally, we provided evidence that the ␤-catenin Armadillo repeats, and the AR DBD/LBD sequence is sufficient for ligand-dependent nuclear translocation and transcriptional activation.
AR and ␤-Catenin Can Co-localize at Transcriptionally Active, Multiprotein Nuclear Complexes-Using confocal laser microscopy we showed that ␤-catenin can co-localize with AR  -catenin (a), GSK3 (b), and APC (c) in LNCaP cells treated with (؉) and without (؊) R1881. Although AR interacted with ␤-catenin in a ligand-sensitive manner, an interaction with GSK3 and APC was not detected. Control immunoprecipitations for AR (a) and APC (c) showed species at 120 and 300 kDa, respectively.
FIG. 6. Specific retention of proteins to androgen-regulated promoter regions using the acrydite capture of DNA complexes (ACDC) method. a, retention of the His-tag AR DBD was observed only for the specific ARE-ac binding site not for the non-cognate N1-ac binding or in the absence of DNA. bi, The ACDC method was used to capture the full-length AR and a proteolytic cleavage product of the AR that binds DNA. c, the ACDC assay was used to capture ␤-catenin from LNCaP nuclear extracts treated with 10 nM R1881. Nuclear extracts were incubated without DNA or with acrydite probasin (P␤-ac) promoter (Ϫ286 to ϩ28) or the pBluescript acrydite multiple cloning site (MCS-ac). Retention of ␤-catenin for the AR-containing LNCaP NE was observed only for P␤-ac, which contains four known AREs. No retention was observed with the nonspecific MCS-ac or in the absence of DNA. upon addition of ligand, it is reasonable to assume that there is dynamic subnuclear trafficking of these receptors. Focal accumulations of nuclear proteins in androgen-treated cells likely consist of transcriptionally active protein complexes with the AR, ␤-catenin, and other associated co-factors, including SRC, CBP, and TCF4 being present. These studies are currently being addressed. The co-localization of the AR and ␤-catenin was not exclusive as seen by confocal microscopy suggesting that there are many AR complexes unoccupied by ␤-catenin and, similarly, ␤-catenin complexes that are unoccupied by the AR. The co-localization of ␤-catenin with the AR also implies that ␤-catenin could be involved with the transcriptional machinery of LNCaP cells. This prospect was explored with the use of a novel DNA binding assay. Using an acrydite polymer, we were able to show specific AR-dependent binding of ␤-catenin to the probasin promoter. This suggests that ␤-catenin could have the capacity to modulate the cell cycle in prostate cancer cells in an androgen-dependent fashion, likely by altering levels of downstream known AR-modulated transcription factors such as c-myc (39) and the cyclins (40). It is probable that ␤-catenin regulates downstream transcription factors by acting as an AR co-activator in a pro-survival manner. Similarly, the AR could have an influence on the Wnt pathway as other steroid receptors such as the RAR and TR have been shown to repress the Wnt pathway. Such a ligand-dependent interplay might suggest sharing of nuclear ␤-catenin between various promoter sites.
Steroid Receptor Nuclear Translocation of ␤-Catenin by the Androgen Receptor-AR-mediated translocation of ␤-catenin was considerable both with light level experiments and biochemical data with the majority of migrating AR reaching the nucleus by 60 min after the addition of ligand. Time-course experiments extended beyond this time point (data not shown) did not yield significant amounts of AR or ␤-catenin movement between compartments. In both endogenously expressed AR cells (LNCaP) and transfected AR-expressing cells (PC3), time series cell fractionations showed a modest, ligand-dependent, nuclear translocation of ␤-catenin. The movement of ␤-catenin was, in general, much less obvious in the cytosolic fraction. We attribute this to the fact that the percentage of ␤-catenin associated with the AR and able to move to the nucleus is relatively small compared with the total cytosolic pool of ␤-catenin. Additionally, it is likely that the amount of "free," non-bound, ␤-catenin varies considerably between different cell lines. The issue of different cytoplasmic pools of ␤-catenin has been raised in previous studies and could be functionally important in AR transcription. Previous reports (41) have identified distinct pools of catenins, which are suggestive of a balancing between APC and AR interactive ␤-catenin. It would be interesting to evaluate how other cytoplasmic pathways are altered as a function of the removal of androgen-associated ␤-catenin upon its nuclear translocation. Certainly, prior to an achieved equilibrium of cellular ␤-catenin, there would be perturbation to the Wnt pathway and Wnt-associated transcription factors. An equally interesting notion is the concept of different nuclear compartments of ␤-catenin of which are likely in a constant state of flux as nuclear receptors and TCF4-related complexes could trade off available ␤-catenin.
Time series experiments showed relatively little, if any, movement of ER between the cytosol and the nucleus. This finding supports previous reports showing that the ER is found predominantly in the nucleus with or without the presence of ligand (22). Given that we did not detect nuclear translocation of TR and RAR, the mechanism as to how ␤-catenin augments their transcriptional activity (42,43) requires future investigation. Although we were not able to detect trafficking of ␤-catenin with these receptors, we cannot rule out the possibility that during the initial translation and modification of these receptors some ␤-catenin is brought to the nucleus. Such a form of Catenin trafficking would not be detectable with the techniques used in this study. The finding that GR does not facilitate ␤-catenin translocation in the same manner as AR does could be a function of the cell lines used in the present studies; that is, cells that express high levels of GR likely show a greater ability to translocate ␤-catenin as compared with prostate cancer cell lines. Future studies using such cell lines will help elucidate if a functional relationship exists between GR and ␤-catenin.
We sought to determine the minimally required components for translocation of ␤-catenin to the nucleus. By transiently transfecting a series of deletion mutants into PC3 cells we able to determine that ␤-catenin requires the AR DBD region for binding but requires the AR DBD/LBD for significant ligand-dependent nuclear translocation. These observations are consistent with previous reports describing the necessity of the ligand binding domain for nuclear localization. ␤-Catenin showed a small degree of nuclear movement in cells that were AR Nt/DBDtransfected but not in prominent, ligand-dependent manner as observed when cells were transfected with AR DBD/LBD . By GST fusion protein interactions we determined that the AR DBD region is capable of a strong receptor/␤-catenin interaction without the AR LBD indicating that minor amounts of ␤-catenin, although not detected by our methods, may be carried into the nucleus by the AR DBD independent of the AR DBD . When we examined the relative affinities between the AR and ␤-catenin as a function of the presence of ligand presence, we did not observe any difference, likely due to the absence of chaperone proteins that would otherwise mask binding. We further investigated what the minimal requirement of AR and ␤-catenin domains for nuclear translocation and found that ligand-dependent movement could be achieved using transfected AR DBD/ LBD and the Arm repeats of ␤-catenin. The Arm repeats of ␤-catenin are known binding sites for other molecules, includ- Upon ligand occupation heat-shock proteins (Hsp) dissociate from the AR allowing for augmentation of ␤-catenin binding. Hsp dissociation also exposes the AR nuclear localization site, which could promote translocation of an AR-␤-catenin complex into the nucleus and binding to one or more AR promoters.
ing E-Cadherin, TCF, CBP, and fascin. The elucidation of the exact region of Arm repeat binding to a region such as the DNA binding domain of steroid receptors is currently being investigated.
LNCaP and PC3 prostate cancer cells are in many respects good systems to demonstrate the hypothesis of AR-mediated nuclear translocation of ␤-catenin. This is mainly due to constitutively low activity of GSK-3 and highly active Akt (18), a regulator of GSK-3. Irregular GSK-3 and Akt levels can be attributed to the mutated tumor suppressor phosphatase with tensin homology, which both cell types carry. Furthermore, the relative amounts of unphosphorylated ␤-catenin are likely higher than other cell types where GSK-3 activity are at higher endogenous levels implying that any additional cytosolic ␤-catenin could have the potential for other binding partners such as cytosolic steroid receptors. By using tagged ␤-catenin constructs, we were more easily able to monitor the movement of this exogenous ␤-catenin. Western blotting for the transiently transfected tagged ␤-catenin showed a much more pronounced translocation of ␤-catenin, likely because the large amount of endogenous ␤-catenin, by our detection methods, previously masked its movement. Although AR-dependent movement of ␤-catenin appears to be functional independent of functional APC, the continual movement of ␤-catenin from the cytoplasm to the nucleus and back to the cytoplasm in cancer cells suggests that these pathways can be functionally independent but can also draw from similar pools of ␤-catenin.
Our proposed model of AR mediated transport of ␤-catenin (Fig. 8) shows that upon ligand occupation heat-shock proteins (Hsp) also dissociate from the AR allowing for augmentation of ␤-catenin binding. Hsp dissociation also exposes the AR nuclear localization site, which could promote translocation of an AR-␤-catenin complex into the nucleus and binding to one or more AR promoters. Our finding of an AR-mediated trafficking pathway leads to the logical implication for a role in oncogenesis whereby stimulation of the Wnt pathway could promote greater cytosolic ␤-catenin and therefore greater AR transcriptional activity. Increased nuclear ␤-catenin could result in altered cell cycle or increased differentiation.
Future studies will likely include investigations of how the AR and other nuclear receptors interact with the Wnt pathway. With the high degree of conservation between the DNA binding regions within the nuclear receptor family, it is likely that there are many commonalities between the AR, GR, and ER with respect to how they could influence Wnt-related genes as a function of ligand exposure. Similarly, the ability of ␤-catenin to augment transcriptional activity of other nuclear receptors is a significant finding and will likely lead to further studies in specific tissues such as the thyroid-targeted tissues, breast and liver. With the apparent similarities in how ␤-catenin may interact with the currently investigated nuclear receptors, it is likely there are many caveats to how ␤-catenin could influence individual receptor pathways. Ultimately, such interactions may lead to more invasive studies whereby repression of Wnt members could attenuate steroid receptor associated transcriptional activation. A key to elucidating how the Wnt pathway interacts with nuclear receptors lies in understanding the likely dynamic interplay between TCF components, nuclear receptor promoters, and ␤-catenin, but also shared co-factors such as CBP and many, yet, undefined ones with the potential to facilitate both pathways.