MenterDiGiovanni, Scott M. Lippman and David G. Suraokar, Christopher J. Logothetis, JohnShao-Yong Chen, Lirim Shemshedini, Milind Troncoso, Guido Jenster, Akira Kikuchi,Vakar-Lopez, Anita L. Sabichi, Patricia Thomas R. Salas, Jeri Kim, Funda  Androgen Receptor Activitythe Phosphorylation and Suppression of

Is Involved inβGlycogen Synthase Kinase-3Mechanisms of Signal Transduction:

doi: 10.1074/jbc.M309560200 originally published online February 24, 20042004, 279:19191-19200.J. Biol. Chem. 

  10.1074/jbc.M309560200Access the most updated version of this article at doi:

  .JBC Affinity SitesFind articles, minireviews, Reflections and Classics on similar topics on the

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/279/18/19191.full.html#ref-list-1

This article cites 32 references, 11 of which can be accessed free at

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

Glycogen Synthase Kinase-3� Is Involved in the Phosphorylationand Suppression of Androgen Receptor Activity*

Received for publication, August 28, 2003, and in revised form, February 24, 2004Published, JBC Papers in Press, February 24, 2004, DOI 10.1074/jbc.M309560200

Thomas R. Salas‡§, Jeri Kim§¶, Funda Vakar-Lopez�, Anita L. Sabichi‡, Patricia Troncoso�,Guido Jenster**, Akira Kikuchi‡‡, Shao-Yong Chen§§, Lirim Shemshedini§§, Milind Suraokar‡,Christopher J. Logothetis¶, John DiGiovanni¶¶, Scott M. Lippman‡, and David G. Menter‡��

From the Departments of ‡Clinical Cancer Prevention, ¶Genitourinary Medical Oncology, and �Pathology, The Universityof Texas M. D. Anderson Cancer Center, Houston, Texas 77030, the **Department of Urology, Erasmus Medical Center,Rotterdam, P. O. Box 1738, 3000 DR Rotterdam, The Netherlands, the ‡‡Department of Biochemistry, Graduate School ofBiomedical Sciences, Hiroshima University, 1-2-3, Kasumi, Minami-ku, Hiroshima 734-8551, Japan, the §§Department ofBiological Sciences, University of Toledo, Toledo, Ohio 43606, and the ¶¶Department of Carcinogenesis, The Universityof Texas M. D. Anderson Cancer Center Science Park, Smithville, Texas 78957

Kinases can phosphorylate and regulate androgen re-ceptor activity during prostate cancer progression. Inparticular, we showed that glycogen synthase kinase-3�phosphorylates the androgen receptor, thereby inhibit-ing androgen receptor-driven transcription. Con-versely, the glycogen synthase kinase-3� inhibitor lith-ium chloride suppressed the glycogen synthase kinase-3�-mediated phosphorylation of the androgen receptor,thereby enabling androgen receptor-driven transcrip-tion to occur. The androgen receptor hinge and ligand-binding domains were important for both the phospho-rylation and the inhibition of transcriptional activity ofthe receptor by glycogen synthase kinase-3�. Further-more, androgen receptor phosphorylation was aug-mented by LY294002, an indirect inhibitor of proteinkinase B/Akt that inhibits glycogen synthase kinase-3�.We also showed that the mutation of various phospho-rylation sites on glycogen synthase kinase-3� affectedthe ability of these mutants to co-distribute with theandrogen receptor in the cell nucleus, also that bothglycogen synthase kinase-3� and androgen receptorproteins can be found in cell nuclei of prostate cancertissue samples. Because glycogen synthase kinase-3� ac-tivity is suppressed after the enzyme is phosphorylatedby protein kinase B/Akt and Akt activity frequently in-creases during the progression of prostate cancer, nul-lification of the glycogen synthase kinase-3�-mediatedsuppression of androgen receptor activity by Akt likelycontributes to prostate cancer progression.

Androgens and androgen receptors (ARs),1 which like othersteroid nuclear receptors, can be regulated by kinases (1–5). In

the prostate, AR helps maintain the balance between thegrowth and death of prostate cells (4). Structurally, the AR hasa transactivation domain (amino acids 1–556), a DNA-bindingdomain (amino acids 559–627), a hinge region (amino acids622–670), and a ligand-binding domain (amino acids 712–919)(3). During the progression of prostate cancer, however, ARscan become mutated or their regulation compromised, whichoften leads to the development of androgen-independentprostate cancer (4, 6). Androgen-independent prostate canceris a lethal recurring tumor that often develops in patientswho have undergone androgen ablation therapy. Multiplefactors can cause androgen-independent prostate cancer (4),but the most common factor, and the one that helps driveboth normal and dysfunctional receptor activity, is the phos-phorylation of AR.

Data suggest that kinases regulate AR function, but whichkinases are involved remains unresolved. Rigorous phos-phoamino acid analyses of the AR have shown that serinephosphorylation events are more prevalent than either threo-nine or tyrosine suggesting that serine kinases may exert mostof the influence on the AR (1, 7). Many consensus sites arepresent in the AR for serine kinases including mitogen-acti-vated protein kinases, calmodulin-dependent protein kinases,and glycogen synthase kinase (GSK)-3 (8). In one instance,Ser308 in the AR was identified as a phosphorylation site butthe kinase that phosphorylates it remains to be identified (9).Others have shown that serines 81 and 94 become phosphoryl-ated soon after translation and when these sites are mutatedthis limits ligand-induced phosphorylation but not transactiva-tion (7, 10). Also, the mutation of serine 650 reduces transcrip-tional regulation of an androgen response element (ARE) con-taining promoter construct by 30% (10). The importance of Aktkinase activity in prostate cancer has been well documented(11–15), particularly the direct involvement of Akt in phospho-rylating AR. For example, one study showed that the phos-phorylation of AR on Ser210 and Ser790 by Akt stimulatedAR-driven transcriptional activity (16), whereas anotherstudy showed that Akt suppressed activity (17). However,both of these studies using in vitro kinase assays in combi-nation with site-directed mutagenesis, putatively showed

* This work was supported by American Cancer Society Grant TPRN-99-240-01-CNE-1 (to D. G. M.), a University of Texas M. D. AndersonCancer Center Predoctoral Fellow Program in Cancer Prevention, Na-tional Institutes of Health, NCI Grant R25-CA57730 (to T. R. S.), aCapCure grant (to J. D.), and a Kadoorie Charitable Foundation Pros-tate Cancer Research Fellowship (to M. S.). The costs of publication ofthis article were defrayed in part by the payment of page charges. Thisarticle must therefore be hereby marked “advertisement” in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.

§ Both authors contributed equally to this manuscript.�� To whom correspondence should be addressed: The University of

Texas M. D. Anderson Cancer Center, Box 236, 1515 Holcombe Blvd.,Houston, TX 77030. Tel.: 713-792-0626; Fax: 713-792-0628; E-mail:dmenter@mdanderson.org.

1 The abbreviations used are: AR, androgen receptor; ARE, androgenresponse element; GSK-3�, glycogen synthase kinase-3�; Akt, protein

kinase B; R1881, 17�-methyl-17�-hydroxyestra-4,9,11-trien-3-one ormetribolone; DH2O, distilled H2O; C3, pcDNA 3.1 empty vector express-ing PC-3 cells; A103, hAR/pcDNA 3.1 plasmid expressing PC-3 cells;V28, VP16-AR expressing PC-3 cells; PC-3, prostate carcinoma cells;COS-1, African green monkey kidney fibroblast cells; LUC, luciferase;PB, probasin; wt, wild type; HA, hemagglutinin.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 18, Issue of April 30, pp. 19191–19200, 2004© 2004 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

This paper is available on line at http://www.jbc.org 19191

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

FIG. 1. Changes in the phosphorylation state of GSK-3� alter its ability to phosphorylate the AR. A, Western blot analysis wasperformed to compare the phosphorylation status of GSK-3� in AR-deficient (AR�) (PC-3, DU145, and C3) and AR-expressing cells (AR�) (LNCaP,A103, and V28) prostatic carcinoma cells. The total levels of GSK-3� were similar in all cell types (bottom row), and there was little difference inthe (P)Tyr216-GSK-3� levels (middle row). The (P)Ser9-GSK-3� level was higher in the AR� cells, LNCaP, compared with AR� cells, PC-3, DU145,

GSK-3� Inhibits AR Function19192

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

that Ser213 and Ser791 were the amino acids on AR that werephosphorylated by Akt. Limited in vivo analysis further in-dicated that Akt mediated the phosphorylation of AR inCOS-1 cells, but no attempt was made to determine whetherAkt and AR were co-localized within cells (17). In contrastwith these findings, other studies by Gioeli et al. (18) whoused intact cells to perform [32P]orthophosphate labeling invivo followed by peptide mapping and mass spectroscopy,failed to confirm the phosphorylation of AR by Akt on Ser213

and Ser791. These conflicting in vitro findings with regard tothe Akt phosphorylation of Ser213 and Ser791 on AR and thefailure to confirm in vivo that these sites are phosphorylatedled us to postulate that GSK-3� regulates AR as an indirect

consequence of Akt activity. GSK-3� is a likely candidate forregulating AR as an indirect mediator of Akt activity formultiple reasons. One reason is that GSK-3� is known tophosphorylate and mediate the activity of numerous nucleartranscription factors (19); another is that GSK-3� is inacti-vated after being phosphorylated on Ser9 by Akt (19, 20).

We have recently shown that GSK-3� can phosphorylatecyclic AMP response element-binding protein and activate itstranscriptional activity in the nuclei of PC-3 prostate carci-noma cells (21). In the present study, we used prostate car-cinoma cells and COS-1 cells to: 1) examine the phosphoryl-ation of AR by GSK-3� and its effects on AR-mediatedtranscription; 2) determine the AR protein domain that in-

FIG. 2. Activated GSK-3� and AR distribution in prostate cells. We examined the localization of AR (blue) and GSK-3� (green) inR1881-treated prostate cancer cells after counterstaining samples for actin (red). In C3 cells stably expressing an empty transfection vector, no ARwas observed (asterisk). The intracellular distribution of GSK-3� (green) in C3 cells varied with the variant that was transfected into cells. TheGSK-3�wt was distributed in both the nucleus and cytoplasm of C3 cells, whereas the GSK-3��9 variant was predominantly in the nucleus (narrowarrows) and the GSK-3�Y216F variant was found in the cytoplasm (wide arrows). In A103, V28, and LNCaP cells that express AR, the AR (blue)was almost exclusively in the nucleus of R1881-treated cells. The distribution of GSK-3� (green) protein in cells that express AR (A103, V28, andLNCaP) was the same as cells that do not express AR (C3).

C3, or AR� cells derived from an AR� lineage, i.e. A103 and V28 cells (top row). The relative density represents the mean pixel density of the(P)Ser9-GSK-3� normalized to that of the total GSK-3�. B, this diagram shows the structure of HA-tagged GSK-3� and mutants used in this study.1, GSK-3�wt contains an enzymatic site and two critical regulatory phosphorylation sites Ser9 and Tyr216. 2, GSK-3��9 is a deletion mutant thathas the first 9 amino acids removed and is constitutively active because it is missing Ser9, the inhibitory phosphorylation site for Akt. 3,GSK-3�Y216F is an enzymatically inactive point mutant that has an essential tyrosine phosphorylation site mutated to a phenylalanine. 4,GSK-3�K85M is an enzymatically suppressed point mutant that has an essential lysine mutated to a methionine. C, GSK-3� variants are expressedin AR� cells. Only endogenous GSK-3� is expressed in the PCMV4 vector control samples (lanes 1, 5, and 9). In general, the HA tag on theseGSK-3� variants caused them to migrate above the endogenous GSK-3�. GSK-3�wt (lanes 2, 6, and 10) has migrated above the endogenousGSK-3�. Constitutively active GSK-3��9 (lanes 3, 7, and 11) has migrated above and below the endogenous GSK-3�. The inactive point mutantGSK-3�Y216F (lanes 4, 8, and 12) has migrated above the endogenous GSK-3�. Heavy phosphorylation of Ser9 occurs primarily on the endogenousGSK-3� in the presence of all variant proteins. Relative density represents the mean pixel density of the (P)Ser9-GSK-3� normalized to that of thetotal GSK-3�. D, enzymatically active GSK-3� variants phosphorylate AR. Overexpression of constitutively active GSK-3��9 (lanes 3, 8, and 13)in AR� cell lines increased the phosphorylation of AR to its greatest extent, followed by the phosphorylation produced by GSK-3�wt (lanes 2, 7,and 12). Vector controls (PCMV4, lanes 1, 6, and 11) and inactive GSK-3�Y216F (lanes 4, 9, and 14) point mutant-expressing cells showed lessphosphorylation of AR, whereas treatment with GSK-3� inhibitor LiCl (2.0 mM, lanes 5, 10, and 15) actively suppressed the phosphorylation of AR.Relative density represents the mean pixel density of the 32P-labeled AR normalized to that of the total AR.

GSK-3� Inhibits AR Function 19193

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

FIG. 3. The AR is phosphorylated by active forms of GSK-3� in COS-1 cell nuclei. A, the phosphorylation status of GSK-3� was examinedin COS-1 cells. The total levels of GSK-3� (bottom row), (P)Tyr216-GSK-3� (top row), and (P)Ser9-GSK-3� (middle row) were similar to those inLNCaP, A103, and V28 cells. B, COS-1 cells co-transfected with AR and GSK-3� variants showed similar results to those in prostate cancer cells.Only endogenous GSK-3� was expressed in vector control samples (lane 1). GSK-3�wt (lane 2) migrated above the endogenous GSK-3�,constitutively active GSK-3��9 (lane 3) migrated above and below the endogenous GSK-3�, and the inactive point mutant GSK-3�Y216F (lane 4)

GSK-3� Inhibits AR Function19194

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

teracts with and is phosphorylated by GSK-3�; and 3) exam-ine the intracellular co-distribution of AR with mutatedforms of GSK-3�.

EXPERIMENTAL PROCEDURES

Cells and Culture Conditions—The human prostate adenocarcinomacells PC-3, DU145, LNCaP, as well as COS-1 cells, were obtained fromthe American Tissue Type Culture Collection (Manassas, VA). PC-3cells were selected that stably expressed empty pcDNA 3.1 vector (C3cells), hAR/pcDNA 3.1 plasmid (A103 cells), or VP16-AR (V28 cells). Alltumor cells were maintained in Dulbecco’s modified Eagle’s mediumand F-12 medium (mixed 1:1) supplemented with 10% fetal bovineserum.

Western Blot Analysis—Western blot analysis was performed to de-tect various proteins in human prostate cancer cells. Whole cell lysateswere prepared as previously described (22). Proteins were then sepa-rated by SDS-PAGE, electrotransferred to nitrocellulose, and analyzedby chemiluminescence procedures, as previously described (22).

GSK-3� Phosphorylation of AR in Cells—Various GSK-3� constructswere transfected into cells using FuGENE 6 (Roche Applied Science)according to the manufacturer’s instructions. Cells were washed withphosphate free Dulbecco’s modified Eagle’s medium (Invitrogen) andthen incubated with 0.5 mCi/ml [32P]orthophosphate (Amersham Bio-sciences) in the presence of 1 nM 17�-methyl-17�-hydroxyestra-4,9,11-trien-3-one (R1881, otherwise known as metribolone, a synthetic non-metabolizable androgen) for 6 h at 37 °C (23–26). Radiolabeled mediumwas removed and cells were washed 3 times with cold phosphate-buffered saline and then scrape-harvested as previously described (22).Immunoprecipitation studies were performed on 200 �g of total proteinusing rabbit anti-AR antibody (Santa Cruz Biotechnology, Santa Cruz,CA), after which the proteins were examined by SDS-PAGE, electro-transferred to nitrocellulose, and subjected to autoradiography per-formed to detect 32P-labeled AR. Electrotransferred proteins wereprobed with rabbit anti-AR polyclonal or mouse anti-AR monoclonalprimary antibody (Santa Cruz Biotechnology) and visualized using theSuper-Signal chemiluminescence substrate (Pierce).

Co-immunoprecipitation Assays—After starvation prostate cellswere incubated with 1 nM R1881 for 20 min. Following treatment, cellswere harvested and 200 �g of total protein lysate was assayed byimmunoprecipitation/Western blot analysis with either an N- or C-terminal specific AR binding antibody or GSK-3� specific antibody(Santa Cruz Biotechnology). Briefly, protein lysates were precleared byincubating with protein A/G-agarose (Pierce) for 30 min at 25 °C. Spe-cific antibodies (10 �g/ml) were added to lysates and incubated for 16 hat 4 °C, followed by washing 3 times with HEPES-buffered saline.Proteins that were bound to protein A/G were removed using SDSsample buffer and resolved by SDS-PAGE. Samples were then electro-transferred to nitrocellulose and probed with the reciprocal antibodyovernight at 4 °C. Immune complexes were visualized with West Femtochemiluminescence kit (Pierce).

Kinase Assays—pRSET-GST-ARLBD protein was produced in Esch-erichia coli and purified using the B-PER/GST Spin Purification Kit(Pierce) in the presence of 1 nM R1881. Purified GST-ARLBD wasincubated with various concentrations of GSK-3� (New England Bio-labs) in the presence of [�-32P]ATP (Amersham Biosciences) and ana-lyzed by NuPage gradient gel electrophoresis (Invitrogen) followedby autoradiography.

Immunofluorescence Analysis—Immunofluorescence analysis wasperformed as described previously (22). A primary rabbit polyclonalantibody recognizing GSK-3� (Santa Cruz) or mouse monoclonal anti-body recognizing AR (BD Biosciences) was diluted (1:200; v/v) in bufferas previously described (22). The samples were then incubated with ananti-rabbit secondary antibody Alexa-488 (green), an anti-mouse sec-ondary antibody Alexa-350 (blue), and actin probe Alexa-594 phalloidin(red) (Molecular Probes), after which they were examined as previouslydescribed (22).

Luciferase Assays—Transfection experiments were performed intriplicate using FuGENE 6. In one set of experiments, various ARconstructs (ARwt, AR5, AR65, AR104, AR106, and AR126), previously

FIG. 4. Activated GSK-3� suppresses AR-mediated transcrip-tion. A, increasing levels of GSK-3�wt plasmid (0.03, 0.1, and 0.3 �g)inhibited ARwt-driven (0.3 �g) transcriptional activity in the presenceof 1 nM R1881 as measured by ARE2-luciferase reporter (normalized toRenilla luciferase). The vector control (VC) was pCMV4. Lithium chlo-ride (LiCl, 2.0 mM), a GSK-3� inhibitor, was used to eliminate GSK-3�-induced suppression of AR transcriptional activity. Shown is the per-centage of luciferase activity � S.E. data relative to the mean ofsamples treated with R1881 in the presence of ARwt (sixth column fromthe left) taken from two independent experiments involving 6 samplesper condition. B, deletion mutants and point mutants of GSK-3� wereused to determine their effect on ARwt-mediated PB-LUC reporteractivity (normalized to Renilla luciferase). GSK-3�wt and GSK-3��9

suppressed ARwt-mediated PB-LUC transcriptional activity in thepresence of R1881. The vector control and inactive point mutant GSK-3�Y216F had no effect on ARwt-mediated PB-LUC reporter activity inthe absence or presence of ligand. Shown is the percentage of luciferaseactivity � S.E. data relative to the mean of samples treated with R1881in the presence of ARwt (fifth column from the left) taken from twoindependent experiments involving 6 samples per condition.

migrated above the endogenous GSK-3�. The endogenous GSK-3� was phosphorylated on Ser9, but the variant GSK-3� proteins were not. Relativedensity represents the mean pixel density of the (P)Ser9-GSK-3� normalized to that of the total GSK-3�. C, AR (blue) was observed in the nuclei(narrow arrows) of R1881-treated COS-1 cells that were co-transfected with AR and GSK-3� variants. The pattern of GSK-3� (green) expressionin COS-1 cells was the same as that in the cancer cells. AR was present in the cell nuclei with GSK-3�wt and GSK-3��9, but only the AR wasobserved in the nuclei (narrow arrows) of cells transfected with GSK-3�Y216F, which appeared in the cytoplasm (wide arrows). D, the co-transfectionof enzymatically active GSK-3� variants with AR into R1881-treated COS-1 cells increased the phosphorylation of AR. In contrast, vector controls(PCMV4, lane 1) and inactive point mutants GSK-3�Y216F (lane 6) and GSK-3�K85M (lane 7) poorly phosphorylated AR. Enzymatically activeGSK-3�wt (lane 3) and GSK-3��9 (lane 4) increased also the phosphorylation of AR. LY294002 (20 �M) increased the phosphorylation of AR whenit was co-expressed with the empty expression vector (lane 2) or GSK-3��9 (lane 5). E, treatment of COS-1 cells expressing GSK-3�wt withincreasing concentrations of a GSK-3� inhibitor LiCl (lanes 2 and 3) dose dependently suppressed the phosphorylation of AR. Relative densityrepresents the mean pixel density of the 32P-labeled AR normalized to that of the total AR.

GSK-3� Inhibits AR Function 19195

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

FIG. 5. GSK-3� phosphorylates the AR ligand-binding domain. A, this diagram shows the AR structure, deletion mutants, and a pointmutant used in this study. 1, ARwt domains include multiple activation domains (AD) in the N-terminal portion, a DNA-binding domain (DBD),a hinge region, and a ligand-binding domain (LBD) in the C-terminal portion of the molecule. 2, AR5 has the complete N-terminal portion and DBD,whereas the hinge and LBD are removed. 3, AR106 has only a single essential AD and the DBD present, causing it to be constitutively activated(see also Jenster et al. (28)). 4, AR104 has the DBD, hinge, and LBD present but cannot activate transcription. 5, AR126 has the AD removed fromthe N-terminal portion and the DBD but no hinge or LBD, making it transcriptionally inactive. B, deletion mutants and point mutants of the ARwere used to determine the AR protein domains that interacted with GSK-3� in the presence of R1881. AR5 is an N-terminal domain mutant, andAR104 is a C-terminal domain mutant both of which co-immunoprecipitated with GSK-3�. N-terminal-specific and C-terminal-specific antibodieswere immunoprecipitated with the appropriate domains of AR5 and AR104, respectively. C, in reciprocal studies using anti-GSK-3� antibody,

GSK-3� Inhibits AR Function19196

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

described (27, 28) were transfected in combination with wild-typeGSK-3� (GSK-3�wt) (Fig. 5A). In another set of experiments, variousGSK-3� constructs previously described (29) (GSK-3�wt, GSK-3��9,GSK-3�K85M, and GSK-3�Y216F) were transfected in combination withARwt into COS-1 cells. Probasin-luciferase (PB-LUC) and pARE-LUCwere the reporter constructs used to analyze AR-driven transcriptionalactivity. Cells were harvested and analyzed using the LUCLITE system(Packard Instrument Co.) and then quantitated in a 96-well black plateusing a TOPCOUNT NT multiwell plate scintillation detector (PackardInstrument Co.). All results from the luciferase assays were normalizedto Renilla luciferase activity, which was generated from pRLTK (Pro-mega Corp.).

Immunostaining of Paraffin Sections—Immunohistochemistry wascarried out in paraffin-embedded sections (3–4 �m) following deparaf-finization and rehydration in xylene, graded alcohol, and distilled H2O(DH2O), respectively. Antigen retrieval was performed in citrate buffer,pH 6.0, in steam. The endogenous peroxidase activity was quenched byincubating the sections in 3% hydrogen peroxide for 10 min at roomtemperature. After blocking the sections with avidin block followed bybiotin block for 15 min each (Biogenex, San Ramon, CA) and thenrinsing three times between treatments with phosphate-buffered sa-line, protein nonspecific blocking solution (Dako Corp., Carpinteria,CA) was applied and incubated for 20 min and slides were rinsed threetimes with phosphate-buffered saline. Rabbit anti-human GSK-3� (CellSignaling Technology, Inc.) or AR antibodies (Santa Cruz Biotechnol-ogy, Inc.) diluted 1:50 with background reducing diluent (Dako Corp.)were applied to the sections and incubated overnight at 25 °C. Then, thesections were incubated with the appropriate biotinylated secondaryantibody and horseradish peroxidase-streptavidin (Dako Corp.) andrinsed three times with phosphate-buffered saline between steps. Per-oxidase activity was detected by applying 3,3�-diaminobenzidine tetra-hydrochloride containing 0.02% hydrogen peroxide for 10 min. Thesections were counterstained with hematoxylin, rinsed with DH2O,intensified with a saturated lithium carbonate as a bluing agent, andrinsed again with DH2O. The sections were then dehydrated in gradedalcohols, cleared in xylene, and mounted in Permount (Fisher) andcovered with coverslips.

Image Analysis—Digital capture methods were optimized to providethe best signal-to-noise contrast for image analysis. We used a QuantixCCD camera that is driven by IP Labs (Scanalytics Inc., Fairfax, VA)image analysis software on a Power PC G4 computer with 128 megabytesof RAM. This image-acquisition equipment is attached to an IX70 in-verted research light microscope (Olympus America Inc., Melville, NY).

RESULTS

Low GSK-3� Activity in Proliferating Prostate Cancer Cells—We used Western blots to analyze GSK-3� protein expressionand phosphorylation in proliferating prostate cells. The totallevels of GSK-3� protein expression were similar in all theprostate cancer cells with little variation in phosphorylation atTyr216 (a constitutive phosphorylation event that activates thisenzyme). In contrast, phosphorylation at Ser9 suppressedGSK-3� activity and was very high in the androgen-dependentLNCaP prostate cancer cells (Fig. 1A, lanes 2 and 4) but lowerin the androgen-independent PC-3 and DU145 prostate cancercells (Fig. 1A, lanes 1 and 3). Previously we verified thatGSK-3� enzymatic activity inversely correlates with (P)-Ser9-GSK-3� levels in these prostate cancer cells.2

To determine the effect of an androgen-independent cellbackground on AR activity, we selected PC-3 cells that stablyexpress AR from a PC-3 cell background. When we selected

PC-3 prostate cells that stably expressed a control vector (C3cells), AR plasmid (A103 cells) or VP16-AR (V28 cells), Ser9 didnot increase to the same levels in these cells as observed in theandrogen-expressing LNCaP cells (Fig. 1A, lanes 4–7).

Phosphorylation of GSK-3� Variants in AR-expressingCells—We next overexpressed HA-tagged GSK-3� variants(29) (Fig. 1B) containing various point mutations or deletionsin the AR-expressing LNCaP, A103, and V28 cells to determinehow these molecular changes affected GSK-3� expression (Fig.1C). Deletion of the first 9 amino acids of GSK-3� resulted in theformation of the GSK-3��9 protein, which is constitutively activeand cannot be inactivated by Akt. Ser9 phosphorylation waspresent on endogenous GSK-3� but not to the same degree on thetransfected HA-tagged variants (Fig. 1C). The position of theHA-tagged variants was verified using an anti-HA antibody.3

Prevention of GSK-3�-induced Phosphorylation of AR byLiCl—The phosphorylation of AR increased above that in con-trol cells in the presence of GSK-3�wt (Fig. 1D, lanes 2, 7, and12) and to a greater extent in the presence of constitutivelyactive GSK-3��9 (Fig. 1D, lanes 3, 8, and 13). In contrast, thephosphorylation of AR was limited in the presence of PCMV4

(Fig. 1D, lanes 1, 6, and 11) and inactive GSK-3�Y216F (Fig. 1D,lanes 4, 9, and 14). Phosphorylation was further decreased incells treated with 2.0 mM lithium chloride (LiCl), a GSK-3�inhibitor (Fig. 1D, lanes 5, 10, and 15). These data suggest thatGSK-3� is involved in phosphorylating AR.

Distribution of GSK-3� Variants and AR in Cell Nuclei—Weexamined the intracellular distribution of the different GSK-3�(green) proteins, the AR (blue) and actin (red) in immunofluo-rescence studies to determine whether these molecules couldinteract within cells (Fig. 2). Untransfected PC-3 and C3 cellscontained low levels of endogenous GSK-3� that were locatedin both the cytoplasm and nucleus of the PC-3 cells. Overex-pressed GSK-3�wt was also present in both the cytoplasm andnucleus of transfected cells. However, the GSK-3�Y216F proteinwas predominantly found in the cytoplasm, and the GSK-3��9

protein was predominantly found in the nucleus of transfectedcells. In addition, AR was not present in the PC3-C3 controlcells that expressed the empty vector but was present in thenuclei of all R1881-treated AR-expressing cells. In the absenceof R1881 by contrast, much less AR was found in the nuclei ofAR-expressing cells (data not shown). These data show that theintracellular distribution of both AR and GSK-3� did not influ-ence each other. AR was predominantly in the nuclei of R1881-treated prostate cells and was not affected by the distributionof GSK-3�. In contrast, the intracellular distribution ofGSK-3� appears to rely in part on its phosphorylation status,in particular, the absence of the Tyr216 phosphorylation siteprevented the distribution of GSK-3�Y216F into cell nuclei.These data suggest that although the intracellular distributionof AR and GSK-3� do not depend on each other, these twomolecules are capable of co-localizing in cancer cell nuclei,which would enable GSK-3� to phosphorylate AR.

2 T. R. Salas, S. M. Lippman, and D. G. Menter, unpublished data.

3 T. R. Salas, S. M. Lippman, and D. G. Menter, unpublishedobservations.

ARwt co-immunoprecipitated with endogenous GSK-3� in vector-transfected cells and with GSK-3�wt when expressed in transfected cells.Immunoprecipitation experiments were performed twice and gave similar results. D, phosphorylation of AR deletion mutants occurs in intact cells.The phosphorylation of AR5 and AR104 was low in vector-transfected cells. The phosphorylation of AR5 increased slightly and the phosphorylationof AR104 increased greatly when co-transfected with GSK-3��9. E, in vitro kinase assays revealed that increasing levels of purified GSK-3� (1 and5 units) can dose-dependently phosphorylate purified GST-ARLBD (1 �g/lane) in the presence of [�-32P]ATP. F, GSK-3�wt plasmid (0.3 �g)inhibited ARwt-induced PB-LUC reporter activity in the presence of 1 nM R1881 by �50%. AR5, a deletion mutant that contained the entire Nterminus and the DBD but no hinge LBD and AR106, a deletion mutant that contained only the essential N-terminal activation domain, could notbe effectively suppressed by GSK-3�wt either in the absence or presence of R1881. Little PB-LUC signal was observed when AR104 or AR126 weretransfected into cells. Shown is the percentage of luciferase activity � S.E. data relative to the mean of samples treated with R1881 in the presenceof ARwt (third column from the left) taken from two independent experiments involving 6 samples per condition.

GSK-3� Inhibits AR Function 19197

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

Low GSK-3� Activity in COS-1 Cells—We used COS-1 cellsto characterize the interactions between GSK-3� and AR dele-tion variants because they are easy to transfect and do notexpress endogenous AR. First, however, we determined thatthe expression pattern and phosphorylation status of GSK-3�

were similar to those in the LNCaP cells, in that Ser9 wasphosphorylated (Fig. 3A) and the kinase activity was low (datanot shown). Similar to prostate cancer cells, when we expressedGSK-3� variants in COS-1 cells, the endogenous GSK-3� wasphosphorylated on Ser9, but was decreased in the GSK-3�

variants (Fig. 3B).Distribution of GSK-3� Variants and AR in COS-1 Cells—

We examined the intracellular distribution of the differentGSK-3� (green) proteins and the AR (blue) and actin (red)using immunofluorescence analysis (Fig. 3C). The intracellulardistribution of each mutant form was similar to prostate cancercells. Both GSK-3�wt and constitutively active GSK-3��9, butnot the GSK-3�Y216F variant, were present with AR in cellnuclei.

Enhancement of Phosphorylation of AR by EnzymaticallyActive GSK-3� Variants and LY294002—The co-transfection ofAR with GSK-3�wt or GSK-3��9 elevated the phosphorylationof AR in COS-1 cells (Fig. 3D, lanes 3 and 4), whereas theco-expression of AR with PCMV4 or GSK-3�Y216F or GSK-3�K85M did not (Fig. 3D, lanes 1, 6, and 7). The treatment ofcells with the phosphatidylinositol 3-kinase-specific inhibitorLY294002 (20 �M) enhanced the phosphorylation of AR byendogenous GSK-3� in the presence of the empty PCMV4 vec-tor or GSK-3��9 (Fig. 3D, lanes 2 and 5). This is because ofLY294002 effects on endogenous GSK-3�, which when com-bined with GSK-3��9 further elevates the phosphorylation ofAR. In contrast, the addition of LiCl in the presence of GSK-3�wt dose-dependently inhibited its ability to phosphorylateAR (Fig. 3E, lanes 2 and 3). These data illustrate that bymanipulating GSK-3� activation through genetic or pharma-cologic methods, once activated GSK-3� can phosphorylate AR.

Dose-dependent Inhibition of ARE Driven Transcription byGSK-3�—The ARwt was co-transfected with GSK-3�wt at theindicated concentrations or in combination with 0.3 �g of theempty vector control in COS-1 cells (Fig. 4A). In the presence of1 nM R1881, a high affinity synthetic nonmetabolizable andro-gen (23–25, 30), ARE2LUC reporter activity decreased as theGSK-3�wt concentration increased. Conversely, treatmentwith 10 mM LiCl relieved the inhibition of AR-mediated tran-scription produced by GSK-3�.

Inhibition of AR-mediated Transcription by GSK-3�—GSK-3� variants were co-transfected with ARwt in COS-1 cellsto determine their effects on AR driven PB-LUC transcription.In contrast to the ARE2LUC promoter mentioned above (Fig.4A), the PB promoter is a complex promoter from a knownendogenous AR target gene (Fig. 4B). GSK-3�wt and GSK-3��9

substantially inhibited ARwt-mediated PB-LUC transcriptionin the presence of 1 nM R1881 (Fig. 4B). In contrast, the cata-lytically inactive GSK-3�Y216F construct (Fig. 4B) and theGSK-3�K85M (data not shown) did not suppress AR-mediatedtranscription either in the absence or presence of R1881.

Protein-Protein Interactions between GSK-3� and AR—ARmutants containing the various protein domains of the AR wereexpressed in COS-1 cells in the presence of 1 nM R1881 (Fig. 5,A–C). When we used antibodies that recognized either the N- orC-terminal domains of AR to immunoprecipitate AR complexes,the appropriate deletion mutant, either AR5 or AR104, respec-tively, was immunoprecipitated along with the GSK-3� (Fig.5B). These antibodies could also co-immunoprecipitate ARwtalong with GSK-3� (Fig. 5B). In reciprocal immunoprecipita-tion experiments we used the anti-GSK-3� antibody to co-

immunoprecipitate either endogenous GSK-3� or GSK-3�wtalong with the exogenous AR (Fig. 5C).

Phosphorylation of N- and C-terminal AR Domain Mutantsby Constitutively Active GSK-3��9—The phosphorylation ofboth N- (AR5) and C-terminal (AR104) domain AR mutantswas minimal in the presence of empty PCMV4 vector but wasincreased in the presence of GSK-3��9 (Fig. 5D). However, thephosphorylation of the C-terminal domain (AR104) was greaterthan that of the N-terminal domain (AR5). The C-terminaldomain contains at least 3 highly likely consensus sites forphosphorylation by GSK-3�.3

In Vitro Phosphorylation of GST-ARLBD Protein by GSK-3�—Recombinant GST-ARLBD protein was phosphorylated invitro by purified GSK-3� in the presence of [�-32P]ATP (Fig.5E). The phosphorylation of 1 �g of GST-ARLBD increased asthe GSK-3� enzyme concentration increased suggesting thatthe ARLBD can serve as a substrate for GSK-3� enzymeactivity.

Regulation of the Transcriptional Activity of AR DeletionMutants by GSK-3�—We transfected various AR deletion con-structs either alone or with GSK-3�wt into COS-1 cells todetermine which domains of AR interacted with GSK-3� toelicit inhibition. When the ARwt were co-transfected withGSK-3�wt, PB-LUC transcription decreased by �50% in thepresence of R1881 (Fig. 5F). In contrast, AR5, which lacks thehinge region and ligand-binding domain, was not inhibited byGSK-3� co-expression. Similarly, the constitutively activeAR106 also resisted GSK-3� regulation. These data are con-sistent with previous findings that showed AR5 and AR106 actas constitutively activated transcription factors that do notrequire ligand binding to activate transcription (Fig. 5F) (28).In contrast, the AR104 and AR126 constructs that contain aDNA-binding domain but lack an activation domain did notsupport any significant transcriptional activity (Fig. 5F) (28).These data suggest that the complete AR molecule, which

FIG. 6. GSK-3� and AR were present in the cytoplasm andnuclei of prostate cancer tissue. GSK-3� and AR were stained withrabbit anti-human GSK-3� (Cell Signaling Technology, Inc.) or AR(Santa Cruz Biotechnology, Inc.) primary antibodies. Both proteinswere diffusely distributed in the cell cytoplasm or concentrated in thenuclei (arrows) of malignant prostate cells.

GSK-3� Inhibits AR Function19198

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

includes both the activation domain DNA-binding domain andthe ligand-binding domain, is important for the regulation ofAR-driven transcriptional activity by GSK-3�.

GSK-3� and AR Co-distribute in Prostate Cancer Tissue—Immunohistochemical staining of prostate cancer tissue sam-ples showed that GSK-3� and AR were diffusely distributed inthe cytoplasm or concentrated in the nuclei (arrows) of prostatecancer cells (Fig. 6). The co-distribution of GSK-3� and AR inprostate cancer tissue suggests that these molecules are capa-ble of interacting in both the cytoplasmic or nuclear compart-ments of prostate cancer cells.

DISCUSSION

To determine whether Akt indirectly regulates AR throughGSK-3�, we examined the ability of GSK-3� to phosphorylateAR in intact cells. First, we bypassed Akt by using a constitu-tively active GSK-3��9 that cannot be inactivated by Akt. GSK-3��9 phosphorylated AR in LNCaP, A103, and V28 prostatecarcinoma cells that stably expressed the receptor (Fig. 1D) andin COS-1 cells that transiently expressed AR (Fig. 3D). Thephosphorylation of AR increased by using an indirect inhibitorof Akt (LY294002, a phosphatidylinositol 3-kinase inhibitor,Fig. 3D, lane 2), which we previously showed to increase en-dogenous GSK-3� activity in prostate carcinoma cells (21).Furthermore, AR phosphorylation reached its highest level inthe cells that expressed constitutively active GSK-3��9 thatwere also treated with LY294002 to inhibit endogenousGSK-3� activity (Fig. 3D, lane 5). In addition, LiCl, a GSK-3�inhibitor, significantly inhibited the phosphorylation of AR inprostate cancer cells (Fig. 1D, lanes 5, 10, and 15) and in theCOS-1 cells (Fig. 3E). These findings show that phosphoryla-tion of AR involves GSK-3� activity in cells when Akt is by-passed using mutant GSK-3��9 or Akt is inhibited by a phos-phatidylinositol 3-kinase inhibitor.

GSK-3� is able to regulate many signaling pathways becauseit possesses unique structural characteristics that cause it torecognize prephosphorylated substrates (19, 20, 31–33). Impor-tantly, the catalytic efficiency of GSK-3� for a phospho-sub-strate is increased to 400–1000-fold in prephosphorylated asopposed to unphosphorylated substrates (20). Computer anal-ysis has revealed the AR has at least 15 consensus sites, (S/T)XXX(S/T)PO4 (S/T, serine/threonine; X, any residue; PO4,downstream prephosphorylation) that GSK-3� can phosphoryl-ate.3 Three of these consensus sites were found in the hinge-ligand binding domains of AR corresponding to the segment ofthe molecule that we showed was phosphorylated whenGSK-3� was activated both in vivo and in vitro. At least 3putative prephosphorylation consensus sites were observed tobe phosphorylated by Gioeli et al. (18), but they did not observephosphorylation by GSK-3�, likely because it is greatly sup-pressed by Akt in the LNCaP cells they used in this study.

Although GSK-3� has numerous substrates in both the nu-cleus and cytoplasm (19), there is little information on themechanisms that mediate its intracellular distribution. Wefound that eliminating the first 9 amino acids in the HA-tagged�9 mutant of GSK-3� caused this variant to accumulate in thenucleus, which co-distributed with AR (Figs. 2 and 3C). Thisaccumulation of GSK-3��9 was associated with the suppressionof AR-mediated transcription. When Tyr216 on GSK-3� wasconverted to phenylalanine in the point mutant GSK-3�Y216F,this limited the accumulation of this variant in the cell nucleus(Figs. 2 and 3C), which in part prevents this variants ability tosuppress AR activity (Figs. 2 and 3C). As part of this ongoingstudy, we have also examined paraffin-embedded tissue sec-tions from prostate cancer patients using immunohistochemi-cal analysis and found that AR and GSK-3� are capable ofco-distributing in the nuclei of cancer cells (Fig. 6) suggesting

that these two molecules may interact during the progressionof prostate cancer.

We were also able to co-immunoprecipitate AR and GSK-3�,suggesting that there was a physical interaction between them.However, we did not determine what regions of GSK-3� bind tothe AR. Nevertheless, a number of putative GSK-3� consensussites are present in the hinge region and surrounding theligand-binding pocket of AR,3 which may help regulate AR. Insome experiments GSK-3�wt or constitutively active GSK-3��9

showed inhibition of ARwt-driven reporter activity in the ab-sence of R1881 ligand (Fig. 4, A and B), but in others thisinhibition was not as great (Fig. 5F). Whether ligand bindingmay affect GSK-3�wt-mediated regulation of AR function willrequire further study.

Our finding that GSK-3� inhibits AR activity may help ex-plain AR activity in prostate cancer cells. The inactivation ofGSK-3� by Akt and subsequent relief of this inhibition mayalso affect lose androgen sensitivity/dependence of prostatetumor cells after androgen ablation therapy. The resultantkinase-mediated phosphorylation of AR can influence manyfunctions, including nuclear translocation, DNA binding, inter-action with regulatory proteins (1), and, as suggested in thisstudy, inhibition of the AR by phosphorylating the hinge and/orligand-binding domain. Further understanding of how GSK-3�influences AR function and the molecular target sites involvedwill undoubtedly elucidate how androgen-dependent prostatecancer is transformed into androgen-independent prostate can-cer after androgen ablation therapy.

Acknowledgments—We thank Dr. Robert M. Chamberlain for sup-port and advice. We also thank Corinn Rich and Packard InstrumentCo. for their generous donation of Firelight dual luciferase reagents. Wethank Norma Llansa and Gabriella Mendoza for technical support. Wealso thank Hazel Dalton for outstanding expertise and support in per-forming immunohistochemistry. We also thank Dr. VemparalaSubbarayan for helpful suggestions and Betty L. Notzon for editing thismanuscript.

REFERENCES

1. Brinkmann, A. O., Blok, L. J., de Ruiter, P. E., Doesburg, P., Steketee, K.,Berrevoets, C. A., and Trapman, J. (1999) J. Steroid Biochem. Mol. Biol. 69,307–313

2. Jenster, G. (2000) J. Pathol. 191, 227–2283. Jenster, G. (1999) Semin. Oncol. 26, 407–4214. Feldman, B. J., and Feldman, D. (2001) Nature Rev. Cancer 1, 34–455. Nelson, W. G., De Marzo, A. M., and Isaacs, W. B. (2003) N. Engl. J. Med. 349,

366–3816. Logothetis, C. J., Hoosein, N. M., and Hsieh, J. T. (1994) Semin. Oncol. 21,

620–6297. Jenster, G., de Ruiter, P. E., van der Korput, H. A., Kuiper, G. G., Trapman, J.,

and Brinkmann, A. O. (1994) Biochemistry 33, 14064–140728. Kennelly, P. J., and Krebs, E. G. (1991) J. Biol. Chem. 266, 15555–155589. Zhu, Z., Becklin, R. R., Desiderio, D. M., and Dalton, J. T. (2001) Biochem.

Biophys. Res. Commun. 284, 836–84410. Zhou, Z. X., Kemppainen, J. A., and Wilson, E. M. (1995) Mol. Endocrinol. 9,

605–61511. Davies, M. A., Koul, D., Dhesi, H., Berman, R., McDonnell, T. J., McConkey,

D., Yung, W. K., and Steck, P. A. (1999) Cancer Res. 59, 2551–255612. Manin, M., Baron, S., Goossens, K., Beaudoin, C., Jean, C., Veyssiere, G.,

Verhoeven, G., and Morel, L. (2002) Biochem. J. 366, 729–73613. Davies, M. A., Kim, S. J., Parikh, N. U., Dong, Z., Bucana, C. D., and Gallick,

G. E. (2002) Clin. Cancer Res. 8, 1904–191414. Malik, S. N., Brattain, M., Ghosh, P. M., Troyer, D. A., Prihoda, T., Bedolla, R.,

and Kreisberg, J. I. (2002) Clin. Cancer Res. 8, 1168–117115. Paweletz, C. P., Charboneau, L., Bichsel, V. E., Simone, N. L., Chen, T.,

Gillespie, J. W., Emmert-Buck, M. R., Roth, M. J., Petricoin, I. E., andLiotta, L. A. (2001) Oncogene 20, 1981–1989

16. Wen, Y., Hu, M. C., Makino, K., Spohn, B., Bartholomeusz, G., Yan, D. H., andHung, M. C. (2000) Cancer Res. 60, 6841–6845

17. Lin, H. K., Yeh, S., Kang, H. Y., and Chang, C. (2001) Proc. Natl. Acad. Sci.U. S. A. 98, 7200–7205

18. Gioeli, D., Ficarro, S. B., Kwiek, J. J., Aaronson, D., Hancock, M., Catling,A. D., White, F. M., Christian, R. E., Settlage, R. E., Shabanowitz, J., Hunt,D. F., and Weber, M. J. (2002) J. Biol. Chem. 277, 29304–29314

19. Frame, S., and Cohen, P. (2001) Biochem. J. 359, 1–1620. Frame, S., Cohen, P., and Biondi, R. M. (2001) Mol. Cell 7, 1321–132721. Salas, T. R., Reddy, S. A., Clifford, J. L., Davis, R. J., Kikuchi, A., Lippman,

S. M., and Menter, D. G. (2003) J. Biol. Chem. 278, 41338–4134622. Subbarayan, V., Sabichi, A. L., Llansa, N., Lippman, S. M., and Menter, D. G.

(2001) Cancer Res. 61, 2720–272623. Dube, J. Y., Chapdelaine, P., Dionne, F. T., Cloutier, D., and Tremblay, R. R.

GSK-3� Inhibits AR Function 19199

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from

(1978) J. Clin. Endocrinol. Metab. 47, 41–4524. Kaufman, M., Pinsky, L., Hollander, R., and Bailey, J. D. (1983) J. Steroid

Biochem. 18, 383–39025. Traish, A. M., Muller, R. E., and Wotiz, H. H. (1981) J. Biol. Chem. 256,

12028–1203326. Schilling, K., and Liao, S. (1984) Prostate 5, 581–58827. Jenster, G., Trapman, J., and Brinkmann, A. O. (1993) Biochem. J. 293,

761–76828. Jenster, G., van der Korput, H. A., Trapman, J., and Brinkmann, A. O. (1995)

J. Biol. Chem. 270, 7341–734629. Murai, H., Okazaki, M., and Kikuchi, A. (1996) FEBS Lett. 392, 153–16030. Murthy, L. R., Chang, C. H., Rowley, D. R., Scardino, P. T., and Tindall, D. J.

(1984) Prostate 5, 567–57931. Dajani, R., Fraser, E., Roe, S. M., Young, N., Good, V., Dale, T. C., and Pearl,

L. H. (2001) Cell 105, 721–73232. ter Haar, E., Coll, J. T., Austen, D. A., Hsiao, H. M., Swenson, L., and Jain, J.

(2001) Nat. Struct. Biol. 8, 593–59633. Cohen, P., and Frame, S. (2001) Nat. Rev. Mol. Cell. Biol. 2, 769–776

GSK-3� Inhibits AR Function19200

at MD

AN

DE

RSO

N C

AN

CE

R C

EN

TE

R on July 1, 2014

http://ww

w.jbc.org/

Dow

nloaded from