A

prchfRntnt©

K

1

usgmpaiaab

0d

Available online at www.sciencedirect.com

Journal of Steroid Biochemistry & Molecular Biology 108 (2008) 121–131

A new highly androgen specific yeast biosensor, enablingoptimisation of (Q)SAR model approaches

Toine F.H. Bovee a,∗, Jos P.M. Lommerse b, Ad A.C.M. Peijnenburg a,Elsa Antunes Fernandes a, Michel W.F. Nielen a

a Department of Safety & Health, RIKILT-Institute of Food Safety, P.O. Box 230, 6700 AE Wageningen, The Netherlandsb Department of Molecular Design & Informatics, N.V. Organon, P.O. Box 20, 5340 BH Oss, The Netherlands

Received 26 March 2007; accepted 15 May 2007

bstract

Recently we constructed recombinant yeast cells that express the human androgen receptor (hAR) and yeast enhanced green fluorescentrotein (yEGFP), the latter in response to androgens. When exposed to 17�-testosterone, the concentration where half-maximal activation iseached (EC50) was 50 nM. Relative androgenic potencies (RAP), defined as the ratio between the EC50 of 17�-testosterone and the EC50 of theompound, were 1.7, 1.2 and 0.008 for 19-nortestosterone, tetrahydrogestrinone and 17�-estradiol respectively. Steroids representative for otherormone receptors, like estrone, 17�-ethynylestradiol, and diethylstilbestrol for the estrogen receptor and corticosterone and dexamethasoneor the glucocorticoid receptor, showed no agonistic response. Only compounds known to exert androgenic effects give a response. DeterminedAPs were in line with results obtained from optimised QSAR model calculations and demonstrated that Saccharomyces cerevisiae showedo metabolism of test compounds and displayed no crosstalk from endogenous hormone receptors. The suitability of this bioassay to verifyhe outcomes of (Q)SAR models to predict the activities of different steroids was further examined by studies with steroid isomers and a

umber of designer steroids, confirming that the 17�-hydroxyl group, 3-keto group and 5�-steroidal framework are extremely important forhe activity of the androgenic steroid.

2007 Elsevier Ltd. All rights reserved.

eywords: Androgen receptor; Biosensor; Crosstalk; Green fluorescent protein; Isomers; Metabolism; Steroids; QSAR modelling

Acgk

bahyts

. Introduction

Chemical and immunological methods are commonlysed to detect steroid hormones in environmental and foodamples, clinical practice, or for doping control. Due to thereat variety of chemicals with hormone-like activity, theseethods have the drawback that they only quantify the com-

ound of interest and are not able to determine biologicalctivity of unknown compounds and their metabolites, thisn contrast to biological assays. Animal in vivo assays suchs the Herschberger and Allen–Doisy tests are highly valu-

ble to assess the overall biological effect of a compound,ut are generally not suited for large-scale screening [1,2].

∗ Corresponding author. Tel.: +31 317 475598; fax: +31 317 417717.E-mail address: toine.bovee@wur.nl (T.F.H. Bovee).

teuara

960-0760/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.oi:10.1016/j.jsbmb.2007.05.035

lternatively, receptor based transcription activation assaysan be used to detect all compounds having affinity for aiven receptor. This feature is very helpful in detecting bothnown and unknown compounds.

Several assays have been developed for this purpose, usingoth mammalian and yeast cells. In general transcriptionctivation assays based on mammalian, or more particularuman cell lines have been shown to be more sensitive thaneast based assays, and may be able to identify compoundshat require human metabolism for activation into their activetate. However, yeast based assays have several other advan-ages. These include low costs, easy handling, lack of knownndogenous receptors that may compete with the activitynder investigation (no crosstalk), and the use of media that

re devoid of steroids. Furthermore, yeast cells are extremelyobust and survive extracts from dirty sample matrices suchs sediments, urine and feed [3,4].

1 emistry

ew(cisptitAmnocTaPdb[calaLwsasAcaeoeA

yrenctutwow(wccsd

2

2

twtdpaaAdGaAT(at(as5aala(55aa(dSfs((ro

2e

e[kU

22 T.F.H. Bovee et al. / Journal of Steroid Bioch

Especially in the case of androgens, the lack of knownndogenous receptors in yeast is a great advantage comparedith mammalian cell lines, as androgen responsive elements

AREs) can also be activated by the progesterone and gluco-orticoid receptor (PR and GR). To avoid potential crosstalkn mammalian cell lines, lots of efforts were made to con-truct an ARE that is specific and no longer inducible by therogesterone or glucocorticoid receptor [5–7]. However, upill now such an ARE does not exist and it is doubtful whethert will be found, as the consensus progesterone and glucocor-icoid responsive elements (PRE/GRE) equal the consensusRE. Moreover, the GR is normally expressed in all mam-alian cell types. So far this resulted in cell lines that are

ot specific for androgens and also respond to progesteroner glucocorticoids, like the Chinese hamster ovary (CHO)ell line and the T47-D human breast carcinoma cell line.he high interference of glucocorticosteroids and of dex-methasone in particular with GR and of progestagens withR can be suppressed by the addition of onapristone, a non-imerising PR and GR antagonist making the androgen testased on the CHO-AR cell line more specific for androgens8]. The AR-LUX in the T47-D human breast carcinomaell line responds to androgens, but also to progesteronend the glucocorticoid dexamethasone. Moreover, this celline has a relative high EC50 value of 115 nM for 5�-DHTnd showed no response to 17�-testosterone [9]. The TM-uc T47-D human breast cell line constructed by Willemsenas transfected with the MMTV-Luc reporter plasmid and

howed maximum responses with 100 nM 17�-trenbolonend 100 nM progesterone. Dexamethasone did not elicit aignificant response, but was only tested at 1 nM [10]. A newR-CALUX test, based on a human U2-OS osteosarcoma

ell line, was developed [11]. This cell line is very sensitivend the EC50 for 17�-testosterone is about 1 nM. Dexam-thasone still gave a response, but the maximum responsebtained with this compound was only 8%. However, 17�-stradiol, a compound that is known to be a strong ER andR agonist, did not give a response in this assay.This paper reports the performance of a new developed

east androgen bioassay that expresses yEGFP as measurableeporter protein in response to androgens. The lack of knownndogenous receptors in yeast enabled us to use the strongon-specific consensus ARE sequence, which is actually aommon HRE that is recognized by the androgen, proges-erone and glucocorticoid receptor and can therefore not besed in mammalian cell lines expressing more than one ofhese receptors. Exposures to all kind of different compoundsere performed to demonstrate the suitability and specificityf this new yeast androgen biosensor. In addition, exposuresith the isomers of androstanediol (8), dihydrotestosterone

4) and androsterone (4) and a number of designer steroidsere carried out in order to explore the effects of structural

haracteristics and to verify the outcomes of QSAR modelalculations. Besides the crosstalk, also the metabolism ofteroid hormones in both yeast and mammalian cells will beiscussed.

mDw(

& Molecular Biology 108 (2008) 121–131

. Materials and methods

.1. Chemicals

Chemicals to prepare growth media, to perform PCR, toransform bacteria and yeast and for the isolation of DNAere as described earlier [12]. Actinomycin D, corticos-

erone, coumestrol, cycloheximide, dehydroepiandrosterone,examethasone, 17�-estradiol, 17�-estradiol, estrone, ethylaraben, 4-hydroxytamoxifen, medroxyprogesterone 17-cetate, 19-nortestosterone, progesterone, 17�-stanozololnd zearalenone were from Sigma (St. Louis, MO, USA).trazine, fenarimol, linuron, p,p′-methoxychlor, procymi-on and vinclozolin were obtained from Dr. EhrenstorfermbH (Augsburg, Germany), p,p′-DDE, o,p′-DDT, kepone

nd 2,4,5-trichlorophenoxyacetic acid were ordered fromlltech, but also came from Dr. Ehrenstorfer GmbH.he following compounds were obtained from Steraloids

Newport, RI, USA): 17�-boldenone, methylboldenone,ndrosterone (5�-androstane-3�-ol-17-one), transandros-erone (5�-androstane-3�-ol-17-one), etiocholanolone5�-androstane-3�-ol-17-one), epietiocholanolone (5�-ndrostane-3�-ol-17-one), 4-androstenedione, diethylstilbe-trol, 5�-dihydrotestosterone (5�-androstane-17�-ol-3-one),�-dihydrotestosterone (5�-androstane-17�-ol-3-one), 5�-ndrostane-17�-ol-3-one, epiallodihydrotestosterone (5�-ndrostane-17�-ol-3-one), 17�-ethynylestradiol, methy-testosterone, 17�-testosterone, 17�-testosteronend 17�-trenbolone. The 17�-dihydroandrosterone5�-androstane-3�-17�-diol) and its seven isomers�-androstane-3�-17�-diol, 5�-androstane-3�-17�-diol,�-androstane-3�-17�-diol, etiocholan-3�,17�-diol (5�-ndrostane-3�-17�-diol), etiocholan-3�,17�-diol (5�-ndrostane-3�-17�-diol), etiocholan-3�,17�-diol5�-androstane-3�-17�-diol) and etiocholan-3�,17�-iol (5�-androstane-3�-17�-diol) were also obtained fromteraloids. The tetrahydrogestrinone (THG) was a giftrom Prof. M. Thevis (DSHS, Cologne, Germany). Copper-ulfate and dimethyl sulfoxide were obtained from MerckDarmstadt, Germany), methyltrienolone from Perkin-ElmerUSA) and 16�-stanozolol from NMI (Australia). Allestriction endonucleases and corresponding buffers werebtained from New England Biolabs (NEB, England, UK).

.2. Construction of the p403-GPD-hAR receptorxpression vector

The yeast strain and the used plasmids were as describedarlier for the construction of our yeast estrogen bioassay12]. Yeast cells expressing the human androgen receptor,indly provided by Prof. D.P. McDonnell (Duke University,SA), were grown overnight in minimal medium supple-

ented with l-leucine (MM/L) at 30 ◦C. ChromosomalNA was isolated [12] and full length human AR cDNAas obtained using the Expand High Fidelity PCR System

Boehringer Mannheim). The sequence of the 5′-primer was:

emistry

5cA5cs2rwwcdc

2r

es

sdcbsfiCr

2

wepcgtImTper

fwCeg

2b

Mt3tmFw1Des1atewtwwy

2

c[ifTehbrtLamaplce

T.F.H. Bovee et al. / Journal of Steroid Bioch

′-GCTCTAGAATGGAAGTGCAGTTAGGGCTGGG-3′ontaining a restriction site for Xba I just before theTG start codon. The sequence of the 3′-primer was:′-GCGGATCCTCACTGGGTGTGGAAATAGATGGG-3′ontaining a restriction site for BamH I just after the TGAtop codon. This PCR generated a full-length ds cDNA of763 bp of the human AR gene with 5′-Xba I and 3′-BamH Iestriction sites. The 2763 bp full length hAR PCR productas isolated from a 1% low-melt agarose gel and cleavedith Xba I and BamH I. This hAR cDNA was ligated into the

orresponding site of the p403-GPD yeast vector. Plasmidigestion controls and PCR controls revealed several goodlones.

.3. Construction of the p406-ARE2-CYC1 yEGFPeporter vector

A set of complementary oligonucleotides (O1 and O2),ach with two consensus ARE sequences (underlined), wereynthesised.

(O1)5′-AAAGTCAGAACAGCATGTTCTGATCAAAT-CTAGAAGATCCAAAGTCAGAACAGCATGTTCTGA-TCAAACTCGAGCAGATCCGCCAGGCGTGTATATAT-AGCGTGGATGG-3′.(O2)5′-CCATCCACGCTATATATACACGCCTGGCGG-ATCTGCTCGAGTTTGATCAGAACATGCTGTTCTG-ACTTTGGATCTTCTAGATTTGATCAGAACATGC-TGTTCTGACTTTAGCT-3′.

When a solution with both complementary DNAequences, 2.5 �M of each, was heated at 95 ◦C and cooledown to room temperature in 2 h, it gave a ds DNA with twoonsensus AREs with a 5′-Sac I sticky end and a 3′-Msc Ilunt end. This ds DNA was cloned into the correspondingite of the p406-CYC1 vector. Subsequently, yEGFP obtainedrom a Hind III/Sal I double digestion of pyEGFP was clonedn the corresponding Hind III/Sal I sites of the p406-ARE2-YC1 reporter construct. Plasmid digestion and PCR controls

evealed several good clones.

.4. Stable transformation of yeast cells

Transformation of yeast K20 (Ura−, His− and Leu−)as performed by the Lithium-acetate protocol as described

arlier [12]. First, the yeast was transformed with the406-ARE2-CYC1-yEGFP reporter vector, integrated at thehromosomal location of the Uracil gene via homolo-ous recombination. Therefore, prior to transformation,he reporter vector was linearised by cutting with Stu, which has a unique restriction site in the URA3arker gene. Transformants were grown on MM/LH plates.

his yeast reporter strain was then transformed with the403-GPD-hAR receptor expression vector, which was lin-arised by cleavage with Nsi I, which has a uniqueestriction site in the HIS3 marker gene (Histidine). Trans-

hrat

& Molecular Biology 108 (2008) 121–131 123

ormants were grown on MM/L plates and PCR controlsere used to select clones that contain the p406-ARE2-YC1-yEGFP reporter and the p403-GPD-hAR receptorxpression construct, both stably integrated in the yeastenome.

.5. Streamlined yEGFP assay with the yeast androgeniosensor

The day before running the assay, a single colony from aM/L agar plate was used to inoculate 10 ml of the selec-

ive MM/L medium. This culture was grown overnight at0 ◦C with vigorous orbital shaking. At the late log phase,he yeast AR cytosensor was diluted in the selective MM/L

edium to an OD value at 604 nm between 0.08 and 0.12.or exposure, aliquots of 200 �l of this diluted yeast cultureere pipetted into each well of a 96-well plate and 2 �l of a7�-testosterone or other stock solution in DMSO was added.MSO only controls were included in each experiment and

ach sample concentration was assayed in triplicate. Expo-ure was performed for 24 h at 30 ◦C and orbital shaking with25 rpm. Fluorescence was measured at 0 and 24 h directly inCytoFluor Multi-Well Plate Reader (Series 4000, PerSep-

ive Biosystems) using excitation at 485 nm and measuringmission at 530 nm. The fluorescence signal was correctedith the signals obtained with the supplemented MM con-

aining DMSO solvent only. Densities of the yeast cultureere determined by measuring the OD at 630 nm, but thisas only done to check whether a sample was toxic for theeast cells.

.6. QSAR modelling

Free energies of the ligand–receptor complexes were cal-ulated using the Schrodinger Maestro software package13]. To end this, the atomic coordinates for the ligand bind-ng domain structure of the androgen receptor were takenrom the Protein Data Bank [14] (PDB entry code 1i37).he structure was corrected by flipping the carbamoyl moi-ties of Asn 705 and Glu 711 by 180◦, ensuring importantydrogen bonds with testosterone and one with the loopetween helix 11 and 12. Subsequently the complex wasefined using the Maestro’s Protein Preparation module. Inotal 39 ligands (from Tables 1 and 2) were processed inigPrep. Using the induced fit docking workflow, all lig-nds were docked in the receptor in double precision (XP)ode, followed by Prime energy minimisation of the amino

cid side chains. In certain cases several distinct dockingoses were automatically kept. Finally, MM-GBSA calcu-ations [15] were applied on all minimised protein–ligandomplexes to estimate the free energy of binding. Entropicffects are neglected in this procedure. However, this will

ardly influence the overall results since this study mainlyegards rather similar and rigid steroids. For each lig-nd, the lowest calculated free energy of the complex wasaken.

124 T.F.H. Bovee et al. / Journal of Steroid Biochemistry & Molecular Biology 108 (2008) 121–131

Table 1EC50 concentrations and relative androgenic potencies (RAP) of compounds in the RIKILT yeast androgen bioassay expressing yEGFP in response to androgens

Compound Qualitative responsefor AR agonisma

Commentsb EC50 (nM) inthe RAAc

RAPd

17�-Testosterone (17�-T)e Pos. (11/11) Strong AR agonist 76 1.017�-Testosterone (17�-T) n.r. n.r.5�-Dihydrotestosterone (5�-DHT)e Pos. (21/21) Strong AR agonist, weak ER agonist 33 2.3Androsterone n.r. n.r.4-Androstenedione (AD)* Pos. (3/3) Strong AR agonist 7200 0.011Dehydroepiandrosterone (DHEA) n.r. n.r.

Methyltestosterone Pos. (2/2) AR and ER agonist 75 1.019-Nortestosterone (nandrolone) 44 1.717�-Boldenonee 510 0.15Methylboldenone 580 0.13Stanozolol 450 0.1716�-Hydroxystanozolol n.r. n.r.17�-Trenbolonee Pos Binds strongly to AR 52 1.5Methyltrienolone (R1881)e Pos. (8/8) AR agonist 54 1.4Tetrahydrogestrinone (THG)e AR agonist 65 1.2

17�-Estradiol (17�-E2)e Pos. (10/11) AR agonist and antagonist, strong ERagonist

9000 0.0084

Estrone (E1)e Pos. (2/2) AR agonist, strong ER agonist n.r n.r.17�-Estradiol (17�-E2)e Neg. (1/1) ER agonist n.r. n.r.17�-Ethynylestradiol (EE2)e Neg. (1/1) Strong ER agonist n.r. n.r.Diethylstilbestrol (DES)e Neg. (2/2) Strong ER agonist n.r. n.r.4-Hydroxytamoxifen**,e Neg. (1/1) ER antagonist n.r. n.r.Progesterone*,e Pos. (7/9) 1700 0.045Medroxyprogesterone acetate (MPA)e Pos. (4/4) Weak AR agonist 1500 0.051Corticosteronee Neg. (1/1) Binds weakly to AR n.r. n.r.Dexamethasonee Pos. (3/4) AR agonist n.r. n.r.Actinomycin D Neg. RNA synthesis inhibitor n.r. n.r.Atrazine Neg. (1/1) n.r. n.r.Coumestrol Neg. (1/1) ER agonist n.r. n.r.Cycloheximide** Neg. Protein synthesis inhibitor n.r. n.r.Ethyl-4-hydroxybenzoate (ethyl paraben) Neg. Binds weakly to ER n.r. n.r.Fenarimol Neg. Aromatase inhibitor n.r. n.r.Linuron Pos. (1/1) Weak AR agonist and antagonist n.r. n.r.Zearalenone Neg. (1/1) ER agonist n.r. n.r.p,p′-DDE Pos. (2/3) Weak AR agonist and antagonist n.r. n.r.o,p′-DDT Neg. (1/1) Weak AR and ER antagonist, weak ER

agonistn.r. n.r.

Kepone** Neg. (2/2) Binds to AR and ER n.r. n.r.p,p′-Methoxychlor Neg. (1/1) AR antagonist, weak ER agonist n.r. n.r.Procymidone Neg. (1/1) AR antagonist n.r. n.r.2,4,5-Trichlorophenoxyacetic acid Neg. Weak ER agonist n.r. n.r.Vinclozolin Neg. (1/1) AR antagonist n.r. n.r.

n.r. = no response.a Qualitative response for AR agonism across all MCRG studies. These data were obtained from NIH Publication No: 03-4503.b Comments obtained from NIH Publication No: 03-4503.c The EC50 is the concentration giving half-maximum response.d The relative androgenic potency (RAP) is defined as the ratio between the EC50 of 17�-testosterone and the EC50 of the compound.e Obtained from Bovee et al. [16].

one are00 nM,

3

3

trar

iii

* The maximum response obtained with 4-androstenedione and progester** These compounds were toxic to yeast. Cycloheximide was toxic above 10 �M.

. Results

A recombinant yeast cell was constructed that expresses

he human androgen receptor and yeast enhanced green fluo-escent protein (yEGFP) as a reporter protein in response tondrogens. Compared to other yeast based bioassays, both theeceptor construct as well as the reporter construct was stably

cspi

about 50 and 35% of that of 17�-testosterone respectively.kepone was toxic above 1000 nM and 4-hydroxytamoxifen was toxic above

ntegrated into the yeast genome by the use of yeast integrat-ng plasmids. PCR controls were carried out to check thentegration of the vectors into the yeast genome. These PCR

ontrols demonstrated that our new yeast androgen cytosen-or contains the p406-ARE2-CYC1-yEGFP reporter and the403-GPD-hAR receptor expression construct, both stablyntegrated in the yeast genome [16].

T.F.H. Bovee et al. / Journal of Steroid Biochemistry & Molecular Biology 108 (2008) 121–131 125

Table 2EC50 concentrations and relative androgenic potencies (RAP) of the eightisomers of androstanediol, the four isomers of 5�-DHT and the four isomersof androsterone in the RIKILT yeast androgen bioassay

Compound EC50 (nM) RAP

17�-Testosterone 50 1.017�-Testosterone n.r. n.r.

5�-Androstane-3�-17�-diol 230 0.22a

5�-Androstane-3�-17�-diol 330 0.15a

5�-Androstane-3�-17�-diol n.r. n.r.5�-Androstane-3�-17�-diol n.r. n.r.5�-Androstane-3�-17�-diol 15000 0.0033a

5�-Androstane-3�-17�-diol n.r. n.r.5�-Androstane-3�-17�-diol n.r. n.r.5�-Androstane-3�-17�-diol n.r. n.r.

5�-Androstane-17�-ol-3-one(5�-dihydrotestosterone, 5�-DHT)

22 2.3

5�-Androstane-17�-ol-3-one 1000 0.050a

5�-Androstane-17�-ol-3-one 2000 0.025a

5�-Androstane-17�-ol-3-one n.r. n.r.

5�-androstane-3�-ol-17-one n.r. n.r.5�-Androstane-3�-ol-17-one

(androsterone)n.r. n.r.

5�-androstane-3�-ol-17-one n.r. n.r.5�-Androstane-3�-ol-17-one

(ethiocholanolone)n.r. n.r.

a These compounds reach a maximum response that is lower than themaximum response obtained with 17�-testosterone. The maxima obtainedwith 5�-androstan-3�-17�-diol, 5�-androstan-3�-17�-diol, 5�-androstan-3a

3c

sdtapnv1htceacpg3hsvta

Fig. 1. Response of the yeast androgen biosensor to different substances.Exposure to 17�-testosterone (17�-T), 5�-dihydrotestosterone (DHT), 17�-testosterone (17�-T), progesterone (P), dexamethasone (Dex), 17�-estradiol(17�-E2), tetrahydrogestinon (THG) and 17�-boldenone (Bold) was startedbtd

sttta

gp1est1tsrgHtogti

tgio15

�-17�-diol, 5�-androstan-17�-ol-3-one and 5�-androstan-17�-ol-3-onere about 85, 60, 35, 50 and 80%, respectively.

.1. Dose–response curves obtained with differentompounds in the yeast androgen biosensor

Typical dose–response curves for several natural andynthetic androgens are shown in Fig. 1. It shows that 5�-ihydrotestosterone (5�-DHT), 17�-testosterone (17�-T),etrahydrogestrinone (THG) and 17�-boldenone are potentndrogens. All of them caused a dose-related increase in theroduction of green fluorescent protein. 17�-Testosterone didot give a response. Obviously, the 17�-OH configuration isery important for androgenic activity. Fig. 1 also shows that7�-estradiol and progesterone give a response. The femaleormone 17�-estradiol gives a full dose–response curve, buthe maximum of the curve is reached at a 500 times higheroncentration than that of 17�-testosterone. Note that 17�-stradiol is a strong ER agonist, but is also known as an ARgonist and an AR antagonist. In 10 out of 11 mammalianell reporter gene systems 17�-estradiol was shown to dis-lay androgenic activity as well [17] (Table 1). Progesteroneives a response, but the maximum response is only about5% of that of 17�-testosterone and is reached at a 25 timesigher concentration. Also progesterone is known to possessome androgenic properties. It shows androgenic effects in

ivo [18] and displays low binding to the AR [19]. Accordingo the NIH publication [17], progesterone showed androgenicctivity in seven out of nine mammalian cell reporter gene

Foe

y adding to 200 �l of a yeast culture, an aliquot of 2 �l of a stock solu-ion of the compound in DMSO. Fluorescence was determined after 24 h asescribed. Fluorescence signals are the mean of a triplicate with S.D.

ystems (MCRG systems). In the in vivo Hershberger assayhe relative androgenic potency of progesterone, comparedo 17�-testosterone, was 0.064 [11]. The corticosteroids cor-icosterone and dexamethasone showed no response in ourssay.

Table 1 shows the calculated EC50, i.e. the concentrationiving a half-maximum response, and the relative androgenicotency (RAP), defined as the ratio between the EC50 of7�-testosterone and the EC50 of the compound, for sev-ral compounds. The yeast androgen bioassay showed goodensitivity towards all androgens tested. Steroids representa-ive for other hormone receptors, like estrone, 17�-estradiol,7�-ethynylestradiol (EE2) and diethylstilbestrol (DES) forhe estrogen receptor and corticosterone and dexametha-one for the glucocorticoid receptor, showed no agonisticesponse. Only 17�-estradiol, progesterone and medroxypro-esterone acetate (MPA) gave a clear agonistic response.owever, as mentioned above for 17�-estradiol and proges-

erone, also MPA is known to exert androgenic effects. Allther compounds tested, mainly pesticides, showed no andro-enic activity in our new yeast androgen bioassay. Together,hese data demonstrate that this new yeast androgen bioassays both sensitive and very specific.

The relatively high RAP of 5�-DHT was expected. It ishe most potent endogenous androgen and in many andro-en target tissues, e.g. in prostate and skin, 17�-testosterones reduced to 5�-DHT by 5�-reductase. A small proportionf all endogenous 17�-testosterone (0.2%) is converted to7�-estradiol by aromatase, mainly in adipose tissues. Both�-reduction and aromatisation are irreversible processes.

ig. 2 shows some molecular structures of steroids and partf the testosterone metabolism. The relatively great differ-nce between the potencies of 17�-testosterone and 5�-DHT

126 T.F.H. Bovee et al. / Journal of Steroid Biochemistry & Molecular Biology 108 (2008) 121–131

nic stero

ir1aild

date1

te(bovtss

Fig. 2. Molecular structure of isomeric androge

ndicates that our Saccharomyces cerevisiae host has no 5�-eductase activity. The great difference between the RAPs of7�-testosterone and 17�-estradiol indicates that the yeastlso lacks aromatase activity. In addition, the great differencen the RAP between 4-androstenedione (AD) and estrone, theatter is not active, also indicates that our S. cerevisiae hostoes not exhibit any aromatase activity.

Compared to the NIH publication [17], there are a fewiscrepancies. According to the NIH publication, estrone is

n AR agonist that showed androgenic activity in two out ofwo mammalian cell reporter gene (MCRG) systems. How-ver, some mammalian cells are able to convert estrone into7�-estradiol and vice versa. This conversion is ascribed

ocnH

ids and selected synthetic androgenic steroids.

o 17�-hydroxysteroid dehydrogenase 3 (17�-HSD 3). Thisnzyme is responsible for the high relative estrogenic potencyREP) of estrone in the ER-CALUX test with the T47-Dreast cancer cells, resulting in a reported REP for estronef 1.0 [20]. This 17�-HSD 3 enzyme also catalyses the con-ersion of 4-androstenedione (AD) to 17�-testosterone andhat of 5�-androstanedione to 5�-DHT. The facts that estronehowed no androgenic response in the yeast androgen bioas-ay (Table 1) and showed a relative estrogenic potency (REP)

f 0.2 in our yeast estrogen assay [12], indicate that theonversion of estrone to 17�-estradiol by 17�-HSD 3 doesot happen in our S. cerevisiae host. The absence of 17�-SD 3 activity is supported by the great difference in the

emistry & Molecular Biology 108 (2008) 121–131 127

dbtratisv[roma1tp

toyasbactEp[n

Dssetgoyrdacm

3i

osvmcp

Fig. 3. Response of the yeast androgen biosensor to the eight different iso-mers of 5-androstane-(3,17)-diol. Exposure to 17�-testosterone (17�-T) andthe eight isomers was started by adding to 200 �l of a yeast culture, an aliquotodt

ohgtacaaaMi3htf1ahbciArcrabae

t

T.F.H. Bovee et al. / Journal of Steroid Bioch

etermined RAPs of 5�-androstanedione and 5�-DHT andy the observation that AD showed a dose–response curvehat reached only 50% of the maximum response that iseached with 17�-testosterone and showed a low relativendrogenic potency of 0.011 (Table 1). The characterisa-ion of AD as a strong AR agonist in the NIH publications thus probably due to its conversion to 17�-testosterone,ince microsomal and cytoplasmic 17�-HSDs that can con-ert AD are reported from testicles, prostate, liver and lungs21]. At least 11 types of mammalian 17�-HSDs have beeneported and were reviewed [21]. Although the ability to carryut 17�-reduction was also reported for a great variety oficroorganisms, very few microbial 17�-HSDs were isolated

nd characterised. Documented and most investigated are the7�-HSDs from Pseudomonas testosteroni and Mycobac-erium sp. Et1, but there are no proves that this enzyme isresent or active in Saccharomyces cerevisiae [22].

Dexamethasone was also described as an AR agonist inhe NIH publication, showing androgenic activity in threeut of four mammalian assays, but was found negative in oureast androgen bioassay. However, these MCRG systems useMMTV-Luc reporter construct or an ARE-Luc reporter con-

truct. Both the MMTV and the ARE sequence are recognisedy the progesterone and glucocorticoid receptor (PR and GR)nd this means that the response found in these mammalianell reporter gene systems is probably due to crosstalk, ashe GR is normally expressed in all mammalian cell types.ven competitive AR binding assays using prostate cytosolreparations suffer from crosstalk of other nuclear receptors23]. Only binding assays using recombinant AR protein doot suffer from this possible artefact [24].

The other discrepancies are the pesticides linuron and p,p′-DE. Both are weak AR agonists and antagonists in MCRG

ystems, but showed no activity in our yeast androgen bioas-ay. It is possible that these compounds fail to show theirffect in yeast cells or cannot enter the yeast cells. However,he latter seems unlikely, as all active compounds alreadyave a response after 4 h of exposure (data not shown). More-ver, p,p′-DDE was shown to exert its estrogenic activity ineast only after a prolonged exposure of 6 days [25]. Theeason for this delayed response was not clear, but was notue to permeability or solubility. With respect to the mech-nism behind the estrogenicity and androgenicity of theseompounds, it is of interest to examine the possible role ofetabolic conversions or crosstalk in the MCRG systems.

.2. Dose–response curves obtained with the differentsomers of androstanediol, 5α-DHT and androsterone

Fig. 3 shows the dose–response curves of the eight isomersf androstanediol. The calculated EC50 and RAP values arehown in Table 2. Table 2 also shows the EC50 and RAP

alues of the four isomers of 5�-DHT and of the four iso-ers of androsterone. The molecular structure of all studied

ompounds and isomers can be derived from Fig. 2. Com-ared to the four isomers of 5�-DHT and the eight isomers

iific

f 2 �l of a stock solution of the compound in DMSO. Fluorescence wasetermined after 24 h as described. Fluorescence signals are the mean of ariplicate with S.D.

f 5-androstane-3,17-diol, the four isomers of androsteroneave a keto group ( O) at position 17 instead of a hydroxylroup ( OH). As all the four isomers of androsterone failedo show a response, it demonstrates that the hydroxyl groupt position 17 is essential for androgenic activity. Since etio-holanolone is not active and AD displays a RAP of 0.011, itlso demonstrates that our S. cerevisiae host does not displayny significant 3�-hydroxysteroid dehydrogenase (3�-HSD)ctivity, as this enzyme converts etiocholanolone into AD.oreover, when looking in detail to the activity of the four

somers of 5�-DHT and the eight isomers of 5-androstane-,17-diol, it becomes clear that the active compounds have theydroxyl group in the 17�-position, while all isomers withhe hydroxyl group in the 17�-position are not active, exceptor 5�-androstane-17�-ol-3-one. It is thus obvious that the7�-hydroxyl group is extremely important for androgenicctivity. These data are in line with findings that the 17�-ydroxyl group is crucial for hydrogen bonding in the ligandinding pocket of the AR. It forms H-bonds with the sidehains of Asn 705 and Thr 877. These interactions, shownn Fig. 4 for 17�-testosterone, between ligand and residuessn 705 (N-terminal region of H3) and Thr 877 (C-terminal

egion of H11) appear to be crucial to the conformationalhanges that pull the N-terminal region of H3 and C-terminalegion of H11 toward the ligand binding pocket, which servess part of the mechanism to close the hydrophobic ligandinding pocket upon ligand binding [26,27]. The structurend function of the AR is reviewed [28] and a lot of researchffort is spend lately on this topic [29–32].

The results in Table 2 also show that the reduction ofhe 3-keto group to a 3-hydroxyl group, like in the eight

somers of 5-androstane-3,17-diol, is not favourable for activ-ty. These findings with yeast are again in line with generalndings, as the 3-keto group forms H-bonds with the sidehains of residues Gln 711 and Arg 752 directly or indirectly

128 T.F.H. Bovee et al. / Journal of Steroid Biochemistry

tloiidttioaAt88twogrt

hha5iFpitS

3

we

ltRhfchFadttwdD3Tsige

4

sCt1imlTa

kiagoaoFtboianot

Fig. 4. The 17�-testosterone-AR hydrogen-bonding network.

hrough a H2O molecule, which is illustrated in Fig. 4. Whenooking in detail again to the activity of the four isomersf 5�-DHT and the eight isomers of 5-androstane-3,17-diol,t is obvious that the stereochemistry at position 5 is verymportant for androgenic activity. 5�-Androstane-3�-17�-iol is active, but compared to 5�-androstane-3�-17�-diolhe 5�-isomer is not as potent and has a RAP that is about 65imes lower than that of the 5�-isomer. Moreover, the max-mum response obtained with 5�-androstane-3�-17�-diol isnly about 40% of the maximum that is reached with 5�-ndrostane-3�-17�-diol and 17�-testosterone (see Fig. 3).lso 5�-androstane-17�-ol-3-one is active, but compared

o 5�-androstane-17�-ol-3-one (5�-DHT) the 5�-isomer is0 times less potent and the maximum response is about0% of the maximum that is reached with 5�-DHT or 17�-estosterone (see Table 2). These data show that compoundsith the hydrogen group in the 5�-position are not activer much less active than the stereo-isomers with the hydro-en group in the 5�-position. The 5�-steroidal frameworkesults in a more straight steroidal structure, which better fitshe androgen ligand binding pocket.

Although the reduction of the 3-keto group to an alco-ol is not favourable for activity, the conformation of theydroxyl group at position 3 seems also important. The 5�-ndrostane-3�-17�-diol isomer is much more active than�-androstan-3�-17�-diol and 5�-androstane-3�-17�-diols a little more potent than 5�-androstane-3�-17�-diol (seeig. 3 and Table 2). It seems that for activity there is a slightreference for the hydroxyl group in the 3�-position. Thiss an equatorial position where the oxygen atom is nearbyhe position where the oxygen of the 3-keto group would be.imilar results were found earlier [24].

.3. QSAR modelling

Fig. 5 shows the correlation between the RAP determinedith the yeast androgen bioassay and the calculated free

nergy after ligand-docking and energy minimisation of the

ta11

& Molecular Biology 108 (2008) 121–131

igand–receptor complex. For 36 out of the 39 compoundsested there is a good correlation between the determinedAP and the free energy. The more potent compounds allave a free energy that is lower than −50 kcal/mol. Theree energy of 5�-androstane-17�-ol-3-one (−48.4) nicelyonfirms the determined RAP (0.05) of the only 17�-ydroxy compound being active on the androgen receptor.or 17�-ethynylestradiol, 16�-hydroxystanozolol and dex-methasone a relatively low free energy has been calculated,espite the fact that no response has been measured inhe bioassay, thus fine-tuning of the free energy calcula-ions seems necessary here. We performed calculations inhich 17�-ethynylestradiol and dexamethasone were pre-icted to be inactive, but then progesterone, stanozolol,HEA, 5�-androstane-3�-ol-17-one and 5�-androstane-�,17�-diol showed deviant correlations (data not shown).he calculated free energy for stanozolol (−51.3) demon-trates that the pyrazole moiety is also able to form strongnteractions with both Arg 752 and Glu 711, as the 3-ketoroup does in 17�-testosterone, which is confirmed by thexperimental RAP (0.17).

. Discussion

The relative androgenic potencies (RAPs) of the designerteroids can be explained by their steroidal structure.ompared to 17�-testosterone, the 17�-position in methyl-

estosterone is alkylated to block the metabolism of the7�-hydroxyl group and to improve the oral bioavailabil-ty. Small steric substitutions at the 17�-position, such as aethyl group, have no effect on the activity, but increasing the

ength of the alkyl chain resulted in decreased activity [33].hus the RAP of 17�-methyltestosterone of 1.0 in our yeastndrogen bioassay is in line with what could be expected.

The actions of androgens in the reproductive tissues arenown as the androgenic effects, while the nitrogen retain-ng effects of androgens in muscle and bone are known as thenabolic effects. It is almost impossible to separate the andro-enic and anabolic effects of AR ligands due to their reliancen a single receptor. However, another class of syntheticndrogens was developed as anabolic steroids as a resultf efforts to minimize undesirable androgenic side effects.or 19-norandrogens, removal of the 19-methyl group was

hought to result in the reduction of the androgenic activityut retention of the anabolic activity. But, although the ratiof anabolic versus androgenic activity of 19-norandrogens ismproved, most 19-norandrogens have both greater anabolicnd androgenic activities [34]. The RAP of 1.7 for 19-ortestosterone (nandrolone) was thus expected. The RAPf R1881 was 1.4. Just like methyltestosterone, R1881 con-ains a methyl group at the 17�-position that has no effect on

he activity, but the 19-methyl group is removed and R1881lso contains three double bonds, one of which alters the1�-position (see Fig. 2). The first alteration, removal of the9-methylgroup, is associated with an increased androgenic

T.F.H. Bovee et al. / Journal of Steroid Biochemistry & Molecular Biology 108 (2008) 121–131 129

F -dockinR

apaHbTpitRoaofHgrhim11

at

tcmpoag1

bga52NH7ticP

ig. 5. Correlation between the calculated free energy (kcal/mol) after ligandAP of compounds in the yeast androgen bioassay.

ctivity and the second alteration, the alteration of the 11�osition, is associated with a reduction of the androgenicctivity. It is therefore difficult to predict the overall effect.owever, just like our findings, R1881 is mostly found toe more potent than 17�-testosterone. The RAP of 1.2 forHG was also expected. The structure of this designer com-ound is similar to that of R1881, but THG contains an ethylnstead of a methyl group at the 17�-position and it washus expected that THG would be a little less potent than1881 (see Fig. 2). Compared to 17�-testosterone and a lotf other steroids, THG also contains an ethyl group instead ofmethyl group at position 18. To our best knowledge, effectsf substitutions at this position were only described once andocus on binding energy calculations through modelling [35].owever, it was speculated that this additional 18-methylroup allows THG to form more numerous contacts with theeceptor via hydrophobic interactions, resulting in a slightlyigher affinity for the hAR. The structure of 17�-trenbolones also similar to that of R1881, but 17�-trenbolone is notethylated at the 17�-position and thus it was expected that

7�-trenbolone would be as potent as R1881. The RAP of

.5 is thus in line with the expectation.

Compared to 17�-testosterone, 17�-boldenone is onlyltered at the A-ring. It contains an extra double bond inhe A-ring between the carbon 1 and 2 position. This dis-

cbti

g mineralising the energy of the ligand–receptor complex and the determined

orts the 5�-steroidal framework and thus the RAP of 0.15an be expected. In line with this is the RAP of 0.13 forethylboldenone, as compared to 17�-boldenone this com-

ound is only methylated at the 17�-position. The low RAPf 0.011 for 4-androstenedione (AD) was again as expected,s compared to 17�-testosterone, containing a 17�-hydroxylroup, this compound contains a keto group at position7.

According to our QSAR calculations, stanozolol onlyinds weakly to the AR, but is known as a strong andro-enic and anabolic agent. To our best knowledge, the structurectivity relation of stanozolol is not discussed. Compared to�-DHT, stanozolol contains a pyrazole moiety at position–3 instead of the keto group at position 3. However, the two-atoms in the pyrazole ring can probably also form a good-bond network with the carbamoyl moiety of residue Gln11 and the guanidinio group of Arg 752 (see Fig. 4) and thushe RAP of 0.17 seems possible. The 16�-hydroxystanozolols inactive. Compared to stanozolol, 16�-hydroxystanozololontains an additional hydroxyl group at the 16�-position.robably both hydroxyl groups can form H-bonds with side

hains Asn 705 and Thr 877 (see Fig. 4). This is confirmedy the free energy calculations, which indicate good bindingo the receptor. At this moment it is difficult to explain thenactivity of this compound.

1 emistry

natwrtTeaaEpOsawOTtaowntheyhiwtpwid

A

i6t

R

[

[

[

[

[

[

[

[

[

[

30 T.F.H. Bovee et al. / Journal of Steroid Bioch

Overall it can be concluded, that S. cerevisiae has little oro aromatase, 5�-reductase, 3�-HSD or 17�-HSD activitynd displays no crosstalk from endogenous hormone recep-ors. Due to the relative simplicity of our yeast bioassayith regard to metabolic conversions and lack of endogenous

eceptors, the new yeast based androgen bioassay was foundo provide a very specific prediction of androgen activity.he often found androgenic activities of estrone and dexam-thasone in mammalian cell line based androgen bioassaysre probably due to a metabolic conversion to 17�-estradiolnd crosstalk from the glucocorticoid receptor, respectively.ven competitive AR binding assays using prostate cytosolreparations suffer from crosstalk of other nuclear receptors.nly binding assays using recombinant AR protein do not

uffer from this possible artefact. The new developed yeastndrogen bioassay shows RAPs of compounds that are in lineith (Q)SAR findings, confirming the importance of the 17�-H group, the 5�-steroidal framework and the 3-keto group.his assay is a simple, reliable and probably the best tool

o check and optimise QSAR model approaches to predictctivities of different isomers and designer steroids. More-ver, the yEGFP assay can be performed completely in 96ell plates within 4 h and does not need cell wall disruptionor does it need the addition of a substrate. As a result, theest is not only sensitive, but also rapid and convenient withigh reproducibility and small variation. The robustness andase of the yeast cells in combination with the qualities ofEGFP, ensure that the assay will be suited to be used as aigh through put system. As structure and function of the ARs such an important research field, not the least because of aorld wide increase in the occurrence of sex organs related

umours, researchers must be aware of the potential misinter-retations from (Q)SAR model results based on data obtainedith mammalian cells, both in vivo and in vitro. Therefore,

t might be time for a renaissance of structure based rationalrug designs based on yeast biosensors.

cknowledgements

This project was financially supported by the Dutch Min-stry of Agriculture, Nature and Food Quality (project number.72027.01). We would like to thank Patrick Mulder for hisechnical help drawing the structures.

eferences

[1] A. Allen, E.A. Doisy, An ovarian hormone. Preliminary report on itslocation, extraction and partial purification and action in test animals,J. Am. Med. Assoc. 81 (1923) 819–821.

[2] L.G. Herschberger, E.G. Shiply, R.K. Meyer, Myotrophic activity of19-nortestosterone and other steroids determined by modified leva-

tor and muscle method, Proc. Soc. Exp. Biol. Med. 83 (1953) 175–180.

[3] H.E. Witters, C. Vangenechten, P. Berckmans, Detection of estrogenicactivity in Flemish surface waters using an in vitro recombinant assaywith yeast cells, Water Sci. Technol. 43 (2001) 117–123.

[

& Molecular Biology 108 (2008) 121–131

[4] T.F.H. Bovee, G. Bor, H.H. Heskamp, L.A.P. Hoogenboom, M.W.F.Nielen, Validation and application of a robust yeast estrogen bioas-say for the screening of estrogenic activity in animal feed, Food Add.Contam. 23 (2006) 556–568.

[5] A. Haelens, G. Verrijdt, L. Callewaert, V. Christiaens, K. Schauwaers,B. Peeters, W. Rombauts, F. Claessens, DNA recognition by the andro-gen receptor: evidence for an alternative DNA-dependent dimerization,and an active role of sequences flanking the response element on trans-activation, Biochem. J. 369 (2003) 141–151.

[6] P.L. Shaffer, A. Jivan, D.E. Dollins, F. Cleassens, D.T. Gewirth, Struc-tural basis of androgen receptor binding to selective androgen responseelements, PNAS 101 (2004) 4758–4763.

[7] J. Brodie, I.J. McEwan, Intra-domain communication between theN-terminal and DNA-binding domains of the androgen receptor: modu-lation of androgen response element DNA binding, J. Mol. Endocrinol.34 (2005) 603–615.

[8] W.G. Schoonen, G.J. Vermeulen, G.H. Deckers, P.M. Verbost, H.J.Kloosterboer, Antiprogestins: their mechanisms of action and the con-sequences for compound selection by in vitro and in vivo studies, Curr.Top. Steroid Res. 2 (1999) 15–54.

[9] B.M.J. Blankvoort, E.M. De Groene, A.P. Van Meeteren-Kreikamp,R.F. Witkamp, R.J.T. Rodenburg, J.M.M.J.G. Aarts, Development ofan androgen reporter gene assay (AR-LUX) utilising a human cell linewith an endogenously regulated androgen receptor, Anal. Biochem. 298(2001) 93–102.

10] P. Willemsen, M. Scippo, G. Maghuin-Rogister, J.A. Martial, M.Muller, Enhancement of steroid receptor-mediated transcription forthe development of highly responsive bioassays, ABC 382 (2005)894–905.

11] E. Sonneveld, J.A.C. Riteco, H.J. Jansen, B. Pieterse, A. Brouwer, W.G.Schoonen, B. Van der Burg, Comparison of in vitro and in vivo screen-ing models for androgenic and estrogenic activities, Toxicol. Sci. 89(2006) 173–187.

12] T.F.H. Bovee, J.R. Helsdingen, I.M.C.M. Rietjens, J. Keijer, L.A.P.Hoogenboom, Rapid yeast estrogen bioassays stably expressing humanestrogen receptors � and �, and green fluorescent protein: a comparisonof different compounds with both receptor types, JSBMB 91 (2004)99–109.

13] Schrodinger: Glide version 4.0 and Prime version 1.5, LLC, New York,2005.

14] J.S. Sack, K.F. Kish, C. Wang, R.M. Attar, S.E. Kiefer, Y. An, G.Y.Wu, J.E. Scheffler, M.E. Salvati, S.R. Krystek Jr., R. Weinmann, H.M.Einspahr, Crystallographic structures of the ligand-binding domains ofthe androgen receptor and its T877A mutant complexed with the naturalagonist dihydrotestosterone, Proc. Natl. Acad. Sci. U.S.A. 98 (2001)4904–4909.

15] P.D. Lyne, M.L. Lamb, J.C. Saeh, Accurate prediction of the rel-ative potencies of members of a series of kinase inhibitors usingmolecular docking and MM-GBSA scoring, J. Med. Chem. 49 (2006)4805–4808.

16] T.F.H. Bovee, J.R. Helsdingen, A.R.M. Hamers, M.B.M. Duursen,M.W.F. Nielen, L.A.P. Hoogenboom, A new highly specific and robustyeast androgen bioassay for the detection of agonists and antagonists,Anal. Bioanal. Chem. 389 (2007) 1549–1558.

17] NIH Publication No: 03-4503, ICCVAM Evaluation of In Vitro Meth-ods for Detecting Potential Endocrine Disruptors, 2003.

18] D.C. Collins, The role of hormonal contraceptives: sex hormone recep-tor binding, progestin selectivity, and the new oral contraceptives, Am.J. Obstet. Gynecol. 170 (1994) 1508–1513.

19] U. Fuhrman, R. Krattenmacher, E.P. Slater, K.H. Fritzmeier, The novelprogestin drospirenone and its natural counterpart progesterone: bio-chemical profile and antiandrogenic potential, Contraception 54 (1996)

243–251.

20] L.A.P. Hoogenboom, L. De Haan, D. Hooijerink, G. Bor, A.J. Murk,A. Brouwer, Estrogenic activity of estradiol and its metabolites in theER-CALUX assay with human T47D breast cells, APMIS 109 (2001)101–107.

emistry

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[35] K. Pereira de Jesus-Tran, P. Cote, L. Cantin, J. Blanchet, F. Labrie, R.

T.F.H. Bovee et al. / Journal of Steroid Bioch

21] M.V. Donova, O.V. Egorova, V.M. Nikolayeva, Steroid 17�-reduction by microorganisms—a review, Process Biochem. 40 (2005)2253–2262.

22] T.L. Rizner, M. Zakelj-Mavric, Characterization of fungal 17beta-hydroxysteroid dehydrogenases, Comp. Biochem. Physiol. Biochem.Mol. Biol. 127 (2000) 53–63.

23] C.L. Waller, B.W. Juma, L.E. Gray, W.R. Kelce, Three-dimensionalquantitative structure-activity relationships for androgen receptor lig-ands, Toxicol. Appl. Pharmacol. 137 (1996) 219–227.

24] H. Fang, W. Tong, W.S. Branham, C.L. Moland, S.L. Dial, H. Hong,Q. Xie, R. Perkins, W. Owens, D.M. Sheehan, Study of 202 natural,synthetic, and environmental chemicals for binding to the androgenreceptor, Chem. Res. Toxicol. 16 (2003) 1338–1358.

25] W. Dhooge, K. Arijs, I. D’Haese, S. Stuyvaert, B. Versonnen, C.Janssen, W. Verstaete, F. Comhaire, Experimental parameters affectingsensitivity and specificity of a yeast assay for estrogenic compounds:results of an interlaboratory validation exercise, Anal. Bioanal. Chem.386 (2006) 1419–1428.

26] W. Bourguet, P. Germain, H. Gronemeyer, Nuclear receptorligand-binding domains: 3D structures, molecular interactions andpharmacological implications, Trends Pharmacol. Sci. 21 (2000)381–388.

27] X. Zhang, X. Li, G.F. Allan, A. Musto, S.G. Lundeen, Z. Sui, Discoveryof indole-containing tetracycles as a new scaffold for androgen receptorligands, Bioorg. Med. Chem. Lett. 16 (2006) 3233–3237.

28] W. Gao, C.E. Bohl, J.T. Dalton, Chemistry and structural biology ofandrogen receptor, Chem. Rev. 105 (2005) 3352–3370.

29] L. Callewaert, N. Van Tilborgh, F. Claessens, Interplay between twohormone-independent activation domains in the androgen receptor,Cancer Res. 66 (2006) 543–553.

& Molecular Biology 108 (2008) 121–131 131

30] J. Jaaskelainen, A. Deeb, J.W. Schwabe, N.P. Mongan, H. Martin, I.A.Hughes, Human androgen receptor gene ligand-binding-domain muta-tions leading to disrupted interaction between the N- and C-terminaldomains, J. Mol. Endocrinol. 36 (2006) 361–368.

31] D. Kazmin, T. Prytkova, C.E. Cook, R. Wolfinger, T.M. Chu, D.Beratan, J.D. Norris, C.Y. Chang, D.P. McDonnell, Linking ligand-induced alterations in androgen receptor structure to differentialgene expression: a first step in the rational design of selectiveandrogen receptor modulators, Mol. Endocrinol. 20 (2006) 1201–1217.

32] V. Georget, W. Bourgeut, S. Lumbroso, S. Makni, C. Sultan, J.C.Nicolas, Glutamic acid 709 substitutions highlight the importanceof the interaction between androgen receptor helices H3 and H12for androgen and antiandrogen actions, Mol. Endocrinol. 20 (2006)724–734.

33] P.M. Matias, P. Donner, R. Coelho, M. Thomaz, C. Peixoto, S. Macedo,N. Otto, S. Joschko, P. Scholz, A. Wegg, S. Basler, M. Schafer, U.Egner, M.A. Carrondo, Structural evidence for ligand specificity inthe binding domain of the human androgen receptor. Implicationsfor pathogenic gene mutations, J. Biol. Chem. 275 (2000) 26164–26171.

34] B.J. Attardi, S.A. Hild, J.R. Reel, Diemethandrolone undecanoate:a new potent orally active androgen with progestational activity,Endocrinology 147 (2006) 3016–3026.

Breton, Comparison of crystal structures of human androgen receptorligand-binding domain complexed with various agonists reveals molec-ular determinants responsible for binding affinity, Protein Sci. 15 (2006)987–999.