A Cellular Conformation-Based Screen for AndrogenReceptor InhibitorsJeremy O. Jones and Marc I. Diamond*Departments of Neurology and Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, California94143-2280

ABSTRACT The androgen receptor (AR), amember of the steroid nuclear receptor family oftranscription factors, regulates a wide range ofphysiological processes. Androgen signaling isalso associated with numerous human diseases,including prostate cancer. All current anti-androgen therapies reduce ligand access to AR,whether by competitive antagonism or inhibitionof androgen production, but are limited by ac-quired resistance and serious side-effects. Thus,novel antiandrogens that target events subse-quent to ligand binding could have importanttherapeutic value. We developed a high through-put assay that exploits fluorescence resonanceenergy transfer (FRET) to measure ligand-inducedconformation change in AR. We directly com-pared this assay to a transcription-based assayin a screen of FDA-approved compounds andnatural products. The FRET-based screen identi-fied compounds with previously unrecognizedantiandrogen activities, with equivalent sensitiv-ity and superior specificity compared to areporter-based screen. This approach can thusimprove the identification of small molecule ARinhibitors.

T he androgen receptor (AR) is a mem-ber of the nuclear receptor (NR) super-family, which consists of a large

group of ligand-regulated transcription fac-tors (1). AR is expressed in many tissues andinfluences an enormous range of physi-ologic processes such as cognition, musclehypertrophy, bone density, and prostategrowth (2). AR signaling is directly linked tonumerous human disorders including be-nign prostatic hyperplasia, alopecia, andhirsutism. AR also drives prostate carcinomaproliferation, even in the setting of andro-gen ablation therapies, and is thus the ma-jor therapeutic target for this malignancy (3).Existing therapies seek to prevent ligandbinding to AR, whether by direct competi-tion or by reduction of serum hormone lev-els with GNRH agonists or 5-� reductase in-hibitors. New classes of AR inhibitors areneeded and could have broad therapeuticapplications.

AR signaling is complex and highly regu-lated (Figure 1, panel a). Prior to binding itsnatural ligand, dihydrotestosterone (DHT),AR associates with a complex of cytoplas-mic factors and molecular chaperones thatmaintain it in a high-affinity ligand bindingconformation (4, 5). Ligand binding inducesan intramolecular conformation change inAR that brings the N- and C-termini intoclose proximity. Using fluorescence reso-nance energy transfer (FRET) (Figure 1,panel b), we previously determined thatthis occurs with a t1/2 � 3.5 min in live cells(6) but does not occur in cell lysates(Figure 1, panel c), suggesting that confor-mation change is not protein autonomous

but depends on additional cellular factors.Following ligand binding, AR concentrates inthe nucleus, where it binds DNA as a ho-modimer at specific androgen response ele-ments (AREs) to regulate gene expression.Transcriptional control by AR results fromcomplex interactions with positive (coactiva-tor) and negative (corepressor) factors (7)(Figure 1, panel a). The receptor is recycledback to the cytoplasm in a highly regulatedprocess that is independent of receptor deg-radation (8).

AR is regulated by cross-talk pathwaysthat may include post-translational modifi-cations such as phosphorylation, sumola-tion, and acetylation (7). For example, HER-2/neu kinase, keratinocyte growth factor,insulin-like growth factor-1, epidermalgrowth factor, and cytokines such as IL-6can activate AR and minimize or possibly ne-gate the requirement for ligand (9–12). In-hibiting these and other regulatory path-ways may provide alternative methods toblock AR activity.

Most screening assays to identify AR in-hibitors indirectly measure AR activity usingreporter genes. These are potentially vulner-able to nonspecific inhibition at multiplesteps interposed between the initial activat-ing event (ligand binding) and the final read-out (gene activation) (Figure 1, panel a). Asan alternative, we focused on AR ligand-induced conformation change as a highlyspecific, proximal molecular event in AR sig-naling. We developed a FRET-based assayto monitor AR conformation change in livecells that is amenable to high-throughputscreening (HTS) (Figure 1, panel b).

*Corresponding author,marc.diamond@ucsf.edu.

Received for review March 10, 2008and accepted May 21, 2008.

Published online June 27, 2008

10.1021/cb800054w CCC: $40.75

© 2008 American Chemical Society

LETTER

ACS CHEMICAL BIOLOGY • VOL.3 NO.7 www.acschemicalbiology.org412

In previous work, we have shown thatthe fusion of full length AR to cyan fluores-cent protein (CFP) and yellow fluorescentprotein (YFP), termed C-AR-Y, maintains itsbasic transcriptional activity (6). C-AR-Y canbe used to measure ligand-induced intramo-lecular conformation change in real time: insingle cells by microscopy or in cell mono-layers using a fluorescence plate reader(FPR) (6). Here we have compared this FRET-based assay to a transcription-based sys-tem in an attempt to identify new classes ofAR inhibitors. We have carried out a screen

of FDA-approved drugs and natural prod-ucts and have identified compounds withpreviously unidentified antiandrogenactivities.

RESULTS AND DISCUSSIONCreation and Characterization of FRET

and Transcription-Based Reporters. C-AR-Ywas stably expressed in HEK293 cells(HEK293/C-AR-Y), a human kidney cell linethat does not express AR, and LAPC4 cells(LAPC4/C-AR-Y), an androgen-dependentprostate cancer cell line with endogenous

AR expression (13). The stable cell lines ex-hibited DHT-induced FRET signal with a char-acteristic dose�response (Figure 1, pan-els d and e). The calculated Z values forthese stable cell lines in the FRET assay were0.6 (LAPC4) and 0.5 (HEK293) (14). For thetranscriptional reporter system, HEK293cells were transfected with vectors express-ing full-length human AR, an androgen-responsive firefly luciferase (MMTV-luc),and an androgen-insensitive SV40promoter-driven renilla luciferase (pRL-SV40). These cells also exhibited a charac-

CFP

FRET

AR YFP

No FRET

DHT

CFP

YFP

b c

FRE

T/do

nor

0.1

0.4

0.2

0.3

50 200100 150Min

1 hr prior

4 hr prior

1 min post

no DHT

d

FRE

T/do

nor

0.1

0.4

0.2

0.3

0 301 3[DHT] (nM)

LAPC4 FRET

10

e

FRE

T/do

nor

0.1

0.4

0.2

0.3

0 301 3[DHT] (nM)

HEK293 FRET

10

f

Luci

fera

se u

nits

1

4

2

3

0 30.1 0.3[DHT] (nM)

HEK293 Txn

1

Ligand bindingconformational change

apo-AR Active AR

Nuclearlocalization

Co-factorrecruitment

Gene activation

Transcription

RNA

Translation

Protein

a

Figure 1. Conformation vs transcription readouts of AR activity. a) Reporter gene assays require multiple steps of cellularactivity to produce a signal, beginning with conversion of apo-AR from an inactive to an active state. b) Fusion of cyanfluorescent protein (CFP) or yellow fluorescent protein (YFP) to the N- and C-termini of AR creates a conformational re-porter that produces a FRET signal upon hormone activation. c) AR ligand-induced conformational change will not occurin a cell-free extract but is stable if this change is induced prior to cell lysis. HEK293 cells expressing CFP-AR-YFP weretreated with 10 nM DHT 1 h or 4 h prior to lysis, immediately after lysis, or were left untreated. Prior treatment of cellswith DHT produced a stable FRET signal after lysis, indicating that AR assumed an active conformation. Treatment afterlysis does not trigger the same conformational change, indicating the requirement for an intact cell. d�f) LAPC4/C-AR-Y cells, HEK293/C-AR-Y cells, or HEK293 cells transiently transfected with AR and MMTV-luciferase were evaluatedwith a dose�response to DHT. Each cell line exhibited a characteristic DHT response, whether by FRET or luciferaseactivity.

LETTER

www.acschemicalbiology.org VOL.3 NO.7 • 412–418 • 2008 413

teristic dose�response (Figure 1, panel f)with a Z value of 0.6, comparable to the FRETassays.

Screening for Novel AR Antagonists:Conformation versus Transcription. Sincethe FRET assays specifically monitor AR con-formational change, we predicted that they

might detect novel antiandrogens with agreater degree of specificity than a tradi-tional transcription reporter assay. Conse-quently, the assays were compared in ascreen of the NINDS compound collection(www.msdiscovery.com) (15, 16), whichconsists of 1040 FDA-approved drugs and

natural products. This librarywas chosen for its small size,the potential to rapidly intro-duce hits into the clinic and be-cause many of the compoundshave previously annotated func-tions, which can facilitate theidentification of cellular targets.

Cells were treated with 10nM DHT for 24 h in the pres-ence of library compounds.Each compound was tested induplicate on separate plates at10 �M. Control wells on eachplate included no DHT (baselinesignal) and DHT without librarycompound (maximal signal).Hydroxy-flutamide (OH-F, 1 �M),a competitive antagonist of DHTknown to inhibit the AR N�C in-teraction (6, 17), was used as apositive control. After 24 h, nor-malized luciferase or fluores-cence signals were measured.

We established a basic algo-rithm to filter the data and vali-date compounds from each ofthe primary screens (Table 1). Tobe selected for further analysis,a compound had to function inboth replicates of the primaryscreen, each reducing the sig-nal by at least 1 standard devia-tion from the mean of the maxi-mal DHT-induced signal. This

eliminated compounds with a strong effectin only one replicate. Compounds with toxiceffects were eliminated based on loss ofthe raw fluorescence signals in the FRET as-says and loss of renilla luciferase activity inthe transcription assay. Approximately5�10% of the library compounds wereeliminated due to toxicity in one of the as-says (Table 1). Since a goal of this screenwas to identify new classes of AR inhibitors,all known competitive antagonists, includ-ing all steroidal compounds, were dis-missed from further analysis. Compoundsthat passed these filtering requirementswere ranked on the basis of antiandrogenactivity.

Each assay effectively sorted compoundsaccording to efficacy (Figure 2, panel a). Allthree systems were equally sensitive tocompetitive antagonists, each identifyingthree of the four known AR antagonistswithin the library. However, the FRET assayappeared to be more stringent than thetranscription-based assay. In HEK293/C-AR-Y cells, 95% of the compounds wereeliminated at 4 standard deviations (SD)from the mean maximal signal. In LAPC4/C-AR-Y cells, the same stringency wasachieved at 3 SD. Similar stringency wasnot obtained until 6 SD for the reporter tran-scription assay (Figure 2, panel a). This im-plies that the conformational assay is lesssensitive to nonspecific cellular perturbationthan the reporter transcription assay. TheFRET-based assay should thus improvecompound detection in larger screens, witha higher specificity and similar sensitivity.

The top 50 hits (�5%) in each assaywere re-examined using a dose-titration tovalidate their activities. Approximately30�40% of the top 50 hits demonstrateda dose response in each primary assay

TABLE 1. Screening strategy and distribution of hits through the screening processa

Compounds scoring positive

Screening step LAPC4 FRET HEK293 FRET HEK293 transcription

I Toxic compounds eliminated 996 898 985II Dose�response in primary assay (top 50) 14 18 17III AR transcription in LAPC4 cells 13 14 10

aC-AR-Y stable cells or transfected transcription reporter cells were cultured in duplicate in the presence of 10 nM DHT and 10 �M of library com-pounds. Cytotoxic compounds and compounds without consistent activity in duplicate trials were eliminated (I). The top 50 hits from each assaywere evaluated in detail with a dose�response study (II). Compounds that displayed a classic dose�response were considered “validated” hits inthe primary assay. This constituted 28�36% of the hits from the primary screen. The validated hits were then tested for efficacy in a secondary as-say of endogenous AR transcriptional activity (III). The number of compounds validated by effects on endogenous AR transcription in LAPC4 cells isindicated.

HEK293 transcription

LAPC4 FRET

HEK293 FRET

700

0

100

200

300

400

500

600

Com

poun

ds

1 72 3 4 5 6Screening stringency (SD from mean)

a

HEK293transcription

HEK293 FRET LAPC4 FRET5

38 4

2 2

10

b

576

364

142

494

257

79

374

7844

228

10 24

101

6 18 41 15 19

Figure 2. Comparison of stringency and overlap of differentassays. a) Hits from the primary screens were comparedin terms of reduction of signal by standard deviation (SD)from positive control cells treated with DHT alone. Thenumber of compounds remaining at various stringencies(SD below the mean) was tabulated. The HEK293 transcrip-tion assay was the most sensitive to nonspecific inhibi-tion by test compounds, as cutoff stringencies of 5�6 SDbelow the mean were required to sort the top 5% of hits.HEK293/C-AR-Y cells sorted the top 5% of hits at 3�4 SDbelow the mean. LAPC4/C-AR-Y cells were the least sensi-tive to nonspecific inhibition by test compounds, with thetop 5% of hits identified at 2�3 SD below the mean.b) The validated hits from each assay were compared inthe other assays. Three compounds were active in all threeassays.

414 VOL.3 NO.7 • 412–418 • 2008 www.acschemicalbiology.orgJONES AND DIAMOND

(Table 1). To determine the efficacy of eachin a secondary assay of endogenous AR ac-tivity, the compounds were evaluated inLAPC4 cells transfected with MMTV-luc/

pRL-SV40. Most, but not all, of the com-pounds that were validated in the primaryassays were effective in this model of endo-genous AR activity (Table 1). The FRET as-

says were more predictive of efficacy in thissecondary model than the transcription as-say. The most efficacious compounds,pyrvinium pamoate (PP) (EC50 � 12 nM)

TABLE 2. List of hits validated vs endogenous ARa

Identified in primary screen

Compound (related compounds grouped) Annotated function LAPC4 FRET HEK293 FRET HEK293 Transcription

Pyrvinium pamoate Antihelminthic X X XThiabendazole Antihelminthic XHarmol HCl (�-carboline) BDZ receptor inverse agonist XHarmaline (�-carboline) BDZ receptor inverse agonist XClozapine GABA receptor antagonist XClonazepam GABA receptor antagonist XEsculin (coumarin) Anticoagulant, vitamin K1 epoxide inhibitor X XWarfarin (coumarin) Anticoagulant, vitamin K1 epoxide inhibitor XPeucedanin (coumarin) Anticoagulant, vitamin K1 epoxide inhibitor XScopoletin (coumarin) Anticoagulant, vitamin K1 epoxide inhibitor X XSulfaquinoxaline Vitamin K1 epoxide inhibitor XEmetine Protein synthesis inhibitor XMelatonin Central nervous system depressant XXylazine Adrenergic receptor antagonists XPhenoxybenzamine HCl Adrenergic receptor antagonists XMitomycin C Antibiotic, Antineoplastic XBleomycin Antibiotic, Antineoplastic XTeniposide Antibiotic, Antineoplastic XOxyquinoline Antiseptic XPomiferin Flavonoid XGedunin HSP90 inhibitor XParthenolide Nonsteroidal anti-inflammatory XFenofibrate Antilipemic XProbucol Antilipemic XTriacetin Triglyceral antifungal XExalamide Antifungal XMemantine HCl Dopamine agonist XApomorphine HCl Dopamine agonist XAminopyridine Potassium channel blocker XAcetyl tryptophan Protease inhibitor XDiffratic acid Unknown XZoxazolamine Muscle relaxant XDioxybenzone Unknown XPimethixene maleate Unknown X

aCompounds identified in secondary analyses as having activity against endogenous AR in an LAPC4 transcription assay were ranked according toefficacy. Annotated functions were gathered from the literature. The rightmost columns indicate which primary screening methodology initially identi-fied the compounds.

LETTER

www.acschemicalbiology.org VOL.3 NO.7 • 412–418 • 2008 415

and harmol hydrochloride (HH) (EC50 �

106 nM), had higher potency than classicalcompetitive antagonists OH-F (EC50 � 130nM) and bicalutamide (BiC) (EC50 � 1600nM). Four of the five most efficacious com-pounds were discovered using the FRET as-say; the fifth was a top hit in all threescreens (Table 2). Inspection of the primarydata revealed that the four compoundsidentified in the FRET assay also scoredpositive in the HEK293 transcription assaybut were not among the top 50 hits, demon-strating that the noise from such an assaycan hinder the detection of truly effective an-tiandrogens and that use of the FRET assayof conformation change may improve thedetection of antiandrogens.

When validated compounds from eachassay were subsequently examined for ac-tivity in the other two assays, 55�82% ofcompounds scored positive in at least oneother assay (Figure 2, panel b). Three vali-dated compounds scored positive in allthree assays, though two of them had notappeared in the top 50 hits of all three pri-mary screens. Interestingly, not all com-pounds were effective in both FRET assays,implying cell-specific targets that influenceAR conformation change.

Noncompetitive AR Inhibitors. No vali-dated hits have structures similar to knownAR ligands or inhibitors, suggesting thatnone would compete for ligand binding.We tested this idea by evaluating a repre-sentative sample of validated hits using awhole-cell radioligand competition assay.HEK293/C-AR-Y cells were incubated with[3H] DHT and cold competitors (Figure 3).

Unlabeled DHT and the competitive antago-nist hydroxyflutamide (OH-F) effectivelycompeted with [3H] DHT for binding. Notest compounds examined effectively com-peted for ligand binding, even at 3�100�

the concentration necessary to fully inhibitAR activity. While we cannot rule out a com-petitive mechanism for all validated hits,our data suggest that most hits from thisscreen function by a noncompetitive mech-anism to inhibit AR activity.

Potential Mechanisms. Coumarin deriva-tives, including esculin, peucedanin, scopo-letin, and warfarin, all inhibited AR activity.Coumarins are widely prescribed anticoagu-lants that inhibit vitamin K1 epoxide to inter-fere with the clotting pathway (18). Interest-ingly, a number of coumarins have beenpreviously determined to have activityagainst prostate cancer. Decursin inducedcell-cycle arrest in prostate cancer derivedcells (19), and coumarin itself produced tu-mor regression in clinical trials of metastaticprostate cancer (20). Sulfaquinoxaline, an-other compound identified in this study, isan antimicrobial that also has activityagainst vitamin K1 epoxide (21). The con-nection between vitamin K1 epoxide and ARactivity is not immediately obvious but war-rants further investigation, as thestructure�activity relationship of thesecompounds suggest a novel mechanism toinhibit AR activity.

A number of inhibitors of �-aminobutyricacid (GABA)/benzodiazepine (BDZ) recep-tors also inhibited AR conformation changeand transcriptional activity. Two of the mosteffective AR inhibitors identified in this

study, the �-carbolines harmol and harma-line, are well-documented inverse agonistsof BDZ receptors (22, 23). Clonazepam, aGABAA selective antagonist (24), alsoscored positive, further implicating theGABA/BDZ receptors in the regulation ofAR. It has also been shown that peripheralBDZ receptors (PBR), which are involved insteroid biosynthesis, are not expressed innormal human prostate but are expressedspecifically in hyperplastic prostate cells,implicating these receptors in the uncon-trolled growth of prostate tissue (25). Under-standing the relationship between AR andBDZ receptors could lead to novel inhibitoryapproaches for AR.

Several compounds known to influencethe activity of adrenergic receptors (AdRs)were also identified in the screens. Phenoxy-benzamine, a nonselective but irreversibleinhibitor of �-AdR (26), and xylazine, an�-AdR agonist (27), both had mild antian-drogen activities. AdRs are highly expressedin prostate tissue. Notably, �-AdR antago-nists are used to treat benign prostatic hy-perplasia (28, 29) and have been proposedas treatments for prostate cancer (30, 31). Aconnection between AdRs and AR may in-volve signaling through the G-�s subunit ofa G-protein coupled receptor (GPCR) andprotein kinase A, using cAMP as an interme-diate (32). Cross-talk between AR and AdRsignaling pathways thus may have impor-tant therapeutic implications.

Drug Discovery. The robust nature of theFRET-based assay suggests that it could beused to screen larger libraries. Theconformation-based approach can likely beadapted also to identify modulators of othernuclear receptors, including the estrogenreceptor (33). Further, the ability ofmicroscope-based systems to measure ARconformational change and dimerizationwith subcellular resolution could make pos-sible the identification of highly specificregulators of receptor function. Addition-ally, the application of fluorescence read-outs of AR conformation in live cells could

DHT

log[drug] (nM)

Rel

ativ

e co

mpe

titio

n

0.2

1.4

0.4

0.6

0.8

1.0

1.2

−2 −1 0 1 2 3 4 5

Thiabendazole

OH-FHarmaline

Emetine

Scopoletin

Xylazine

Clozapine

Gedunin

Mitomycin C

Figure 3. Validated compounds are noncompetitive AR inhibitors. HEK293/C-AR-Y cells were in-cubated with 3nM [3H] DHT and the indicated compounds. The ability of each compound to com-pete for binding is expressed relative to the no competition value, which was set to 1. Hy-droxyflutamide (OH-F), a competitive antagonist, and unlabeled DHT effectively competed with[3H] DHT for binding. None of the eight AR inhibitors competed for ligand binding.

416 VOL.3 NO.7 • 412–418 • 2008 www.acschemicalbiology.orgJONES AND DIAMOND

be exploited to identify genetic modifiers ofreceptor function at a single-cell level viaflow cytometry. This work validates the con-cept that conformation-based screening willyield novel AR inhibitors that mightotherwise be missed by conventionaltranscription-based approaches.

Conclusion. Cell-based assays havemanifest utility to study nuclear receptors.This work illustrates how a conformationalchange too complicated to reproduce invitro may be exploited to discover novel in-hibitors, many of which may function bymechanisms distinct from competitive an-tagonists. The ability to identify such “or-thogonal” inhibitors is greatly improved bythe use of cell-based assays, which permitinhibitors of nonreceptor factors and pro-vide information that cannot be obtainedwith simple ligand-binding studies (34). In-deed, this general approach was recentlyvalidated in another study based on anandrogen-deprived gene expression profile,which identified inhibitors of hsp90, in addi-tion to three compounds also identified inour screen (35). Multiple such orthogonalregulators of AR activity may exist andshould facilitate discovery of cellular path-ways that control this process. Last, the inhi-bition of cross-talk pathways that modulateAR signaling in a synergistic fashion may al-low significant dose reductions to reducetoxicity for treatments of AR-associateddisease.

MATERIALS AND METHODSCell Culture. HEK293 cells were maintained in

Dulbecco’s modified Eagle’s medium supple-mented with antibiotics and 5% fetal bovine se-rum (FBS). LAPC4 cells were maintained in phenol-red free RPMI 1640 media supplemented withantibiotics and 10% FBS. After transfection usingLipofectamine (Invitrogen) reagents, 293/C-AR-Yand LAPC4/C-AR-Y cell lines were isolated fromsingle colonies formed under hygromycin selec-tion. Cells were transferred to media containing5% charcoal-stripped FBS 48 h prior to FRET ortranscription assays. The NINDS collection waspurchased from Microsource in 96 well plates inDMSO. Pyrvinium pamoate was purchased fromMP biomedicines, and all other compounds werepurchased from Sigma.

Transcription Assays. Pools of cells were trans-fected using Lipofectamine Plus reagents (Invitro-gen) with plasmids containing full-length AR,MMTV-luciferase and pRL-SV40 (Promega). The fol-lowing day, the cells were trypsinized and trans-ferred to 96 well plates along with 10 nM DHT andlibrary compounds at 10 �M with a BioMek FX liq-uid handling robot (Beckman-Coulter). Twentyfour hours later, luciferase activity was measuredusing the Dual luciferase assay kit (Promega).Mean-effect plots (log[compound] vs log[fractionaleffect]) were generated to determine the EC50 val-ues for each compound. Microsoft Excel was usedto calculate the statistics for a line using the “leastsquares” method. The F statistic was used to de-termine whether the observed relationship be-tween the dependent and independent variablesoccurred by chance. Only data with an r2 valuegreater than 0.95 and an F value that was greaterthan that indicated by the F table for � � 0.05were used for analysis. The EC50 values are de-rived from four independent experiments.

FRET Assays. FRET assays were performed as de-scribed previously (36). Briefly, cells stably ex-pressing C-AR-Y were transferred to black, clear-bottomed 96 well plates along with DHT andlibrary compounds. The cells were fixed in 4%paraformaldehyde and read in PBS on amonochromator-based fluorescence plate reader(Safire, Tecan, Inc.). Each plate contained untrans-fected, positive, and negative controls. FRET:do-nor ratios were calculated following backgroundsubtraction and correction for acceptor (YFP) con-tribution to the FRET signal.

Radioligand Competition Binding Assay.HEK293/C-AR-Y cells were seeded in 24-wellplates in phenol-red free media containing 5%charcoal-stripped FBS. After 3 days, media were re-placed with serum-free media containing 3 nM[3H] DHT in the absence or presence of 0.1�1000-fold molar excess of unlabeled competitor ligandsfor 90 min at 37 °C. Cells were washed with phos-phate buffer, and bound ligand was extracted inethanol for 30 min at RT and detected using a scin-tillation counter.

Acknowledgment: This work was supportedby NIH-5F32CA123750 (J.O.J.), the NIH/NCI1R01CA131226-01 (M.I.D.), the Prostate CancerFoundation (M.I.D.), and the Sandler Family Sup-porting Foundation (M.I.D.).

REFERENCES1. Katzenellenbogen, J. A., and Katzenellenbogen, B. S.

(1996) Nuclear hormone receptors: ligand-activatedregulators of transcription and diverse cell re-sponses, Chem. Biol. 3, 529–536.

2. Gelmann, E. P. (2002) Molecular biology of the an-drogen receptor, J. Clin. Oncol. 20, 3001–3015.

3. Scher, H. I., and Sawyers, C. L. (2005) Biology of pro-gressive, castration-resistant prostate cancer: di-rected therapies targeting the androgen-receptor sig-naling axis, J. Clin. Oncol. 23, 8253–8261.

4. Cardozo, C. P., Michaud, C., Ost, M. C., Fliss, A. E.,Yang, E., Patterson, C., Hall, S. J., and Caplan, A. J.(2003) C-terminal Hsp-interacting protein slows an-drogen receptor synthesis and reduces its rate ofdegradation, Arch. Biochem. Biophys. 410, 134–140.

5. Georget, V., Terouanne, B., Nicolas, J. C., and Sul-tan, C. (2002) Mechanism of antiandrogen action:key role of hsp90 in conformational change andtranscriptional activity of the androgen receptor,Biochemistry 41, 11824–11831.

6. Schaufele, F., Carbonell, X., Guerbadot, M., Born-graeber, S., Chapman, M. S., Ma, A. A., Miner, J. N.,and Diamond, M. I. (2005) The structural basis ofandrogen receptor activation: intramolecular and in-termolecular amino-carboxy interactions, Proc. Natl.Acad. Sci. U.S.A. 102, 9802–9807.

7. Poletti, A. (2004) The polyglutamine tract of andro-gen receptor: from functions to dysfunctions in mo-tor neurons, Front. Neuroendocrinol. 25, 1–26.

8. Tyagi, R. K., Lavrovsky, Y., Ahn, S. C., Song, C. S.,Chatterjee, B., and Roy, A. K. (2000) Dynamics of in-tracellular movement and nucleocytoplasmic recy-cling of the ligand-activated androgen receptor in liv-ing cells, Mol. Endocrinol. 14, 1162–1174.

9. Craft, N., Shostak, Y., Carey, M., and Sawyers, C. L.(1999) A mechanism for hormone-independentprostate cancer through modulation of androgen re-ceptor signaling by the HER-2/neu tyrosine kinase,Nat. Med. 5, 280–285.

10. Klocker, H., Culig, Z., Hobisch, A., Cato, A. C., andBartsch, G. (1994) Androgen receptor alterations inprostatic carcinoma, Prostate 25, 266–273.

11. Yeh, S., Lin, H. K., Kang, H. Y., Thin, T. H., Lin, M. F.,and Chang, C. (1999) From HER2/Neu signal cas-cade to androgen receptor and its coactivators: anovel pathway by induction of androgen targetgenes through MAP kinase in prostate cancer cells,Proc. Natl. Acad. Sci. U.S.A. 96, 5458–5463.

12. Ueda, T., Bruchovsky, N., and Sadar, M. D. (2002)Activation of the androgen receptor N-terminal do-main by interleukin-6 via MAPK and STAT3 signaltransduction pathways, J. Biol. Chem. 277, 7076–7085.

13. Klein, K. A., Reiter, R. E., Redula, J., Moradi, H., Zhu,X. L., Brothman, A. R., Lamb, D. J., Marcelli, M., Bell-degrun, A., Witte, O. N., and Sawyers, C. L. (1997)Progression of metastatic human prostate cancer toandrogen independence in immunodeficient SCIDmice, Nat. Med. 3, 402–408.

14. Zhang, J. H., Chung, T. D., and Oldenburg, K. R.(1999) A simple statistical parameter for use in eval-uation and validation of high throughput screen-ing assays, J. Biomol. Screening 4, 67–73.

15. Finkelstein, R., Miller, T., and Baughman, R. (2002)The challenge of translational research�a perspec-tive from the NINDS, Nat. Neurosci. 5, 1029–1030.

16. Abbott, A. (2002) Neurologists strike gold in drugscreen effort, Nature 417, 109.

17. Wong, C. I., Zhou, Z. X., Sar, M., and Wilson, E. M.(1993) Steroid requirement for androgen receptordimerization and DNA binding. Modulation by in-tramolecular interactions between the NH2-terminaland steroid-binding domains, J. Biol. Chem. 268,19004–19012.

LETTER

www.acschemicalbiology.org VOL.3 NO.7 • 412–418 • 2008 417

18. Hirsh, J., Dalen, J. E., Deykin, D., and Poller, L. (1992)Oral anticoagulants. Mechanism of action, clinicaleffectiveness, and optimal therapeutic range, Chest102, 312S–326S.

19. Yim, D., Singh, R. P., Agarwal, C., Lee, S., Chi, H., andAgarwal, R. (2005) A novel anticancer agent, decur-sin, induces G1 arrest and apoptosis in humanprostate carcinoma cells, Cancer Res. 65, 1035–1044.

20. Mohler, J. L., Gomella, L. G., Crawford, E. D., Glode,L. M., Zippe, C. D., Fair, W. R., and Marshall, M. E.(1992) Phase II evaluation of coumarin (1,2-benzopyrone) in metastatic prostatic carcinoma,Prostate 20, 123–131.

21. Preusch, P. C., Hazelett, S. E., and Lemasters, K. K.(1989) Sulfaquinoxaline inhibition of vitamin Kepoxide and quinone reductase, Arch. Biochem. Bio-phys. 269, 18–24.

22. Braestrup, C., and Nielsen, M. (1993) Discovery ofbeta-carboline ligands for benzodiazepine recep-tors, Psychopharmacol. Ser. 11, 1–6.

23. Hadjipavlou-Litina, D., Garg, R., and Hansch, C.(2004) Comparative quantitative structure-activityrelationship studies (QSAR) on non-benzodiazepinecompounds binding to benzodiazepine receptor(BzR), Chem. Rev. 104, 3751–3794.

24. Greenblatt, D. J., Miller, L. G., and Shader, R. I. (1987)Clonazepam pharmacokinetics, brain uptake, andreceptor interactions, J. Clin. Psychiatry 48, 4–11.

25. Bribes, E., Carriere, D., Goubet, C., Galiegue, S., Ca-sellas, P., and Simony-Lafontaine, J. (2004) Immuno-histochemical assessment of the peripheral benzo-diazepine receptor in human tissues, J. Histochem.Cytochem. 52, 19–28.

26. Te, A. E. (2002) A modern rationale for the use ofphenoxybenzamine in urinary tract disorders andother conditions, Clin. Ther. 24, 851–861 and dis-cussion on p 837.

27. Fyffe, J. J. (1994) Effects of xylazine on humans: a re-view, Aust. Vet. J. 71, 294–295.

28. Furuya, S., Kumamoto, Y., Yokoyama, E., Tsuka-moto, T., Izumi, T., and Abiko, Y. (1982) Alpha-adrenergic activity and urethral pressure in prostaticzone in benign prostatic hypertrophy, J. Urol. 128,836–839.

29. Kyprianou, N. (2003) Doxazosin and terazosin sup-press prostate growth by inducing apoptosis: clini-cal significance, J. Urol. 169, 1520–1525.

30. Tahmatzopoulos, A., Rowland, R. G., and Kyprianou,N. (2004) The role of alpha-blockers in the manage-ment of prostate cancer, Expert Opin. Pharmaco-ther. 5, 1279–1285.

31. Tahmatzopoulos, A., and Kyprianou, N. (2004) Ap-optotic impact of alpha1-blockers on prostate can-cer growth: a myth or an inviting reality? Prostate 59,91–100.

32. Kasbohm, E. A., Guo, R., Yowell, C. W., Bagchi, G.,Kelly, P., Arora, P., Casey, P. J., and Daaka, Y. (2005)Androgen receptor activation by G(s) signaling inprostate cancer cells, J. Biol. Chem. 280, 11583–11589.

33. Cvoro, A., Paruthiyil, S., Jones, J. O., Tzagarakis-Foster, C., Clegg, N. J., Tatomer, D., Medina, R. T., Ta-gliaferri, M., Schaufele, F., Scanlan, T. S., Dia-mond, M. I., Cohen, I., and Leitman, D. C. (2007)Selective activation of estrogen receptor-beta tran-scriptional pathways by an herbal extract, Endocri-nology 148, 538–547.

34. Kenakin, T., and Onaran, O. (2002) The ligand para-dox between affinity and efficacy: can you be thereand not make a difference? Trends Pharmacol.Sci. 23, 275–280.

35. Hieronymus, H., Lamb, J., Ross, K. N., Peng, X. P.,Clement, C., Rodina, A., Nieto, M., Du, J., Stegmaier,K., Raj, S. M., Maloney, K. N., Clardy, J., Hahn, W. C.,Chiosis, G., and Golub, T. R. (2006) Gene expres-sion signature-based chemical genomic predic-tion identifies a novel class of HSP90 pathway mod-ulators, Cancer Cell 10, 321–330.

36. Desai, U. A., Pallos, J., Ma, A. A., Stockwell, B. R.,Thompson, L. M., Marsh, J. L., and Diamond, M. I.(2006) Biologically active molecules that reducepolyglutamine aggregation and toxicity, Hum. Mol.Genet. 15, 2114–2124.

418 VOL.3 NO.7 • 412–418 • 2008 www.acschemicalbiology.orgJONES AND DIAMOND