0013-7227/02/$15.00/0
`Printed in U.S.A,
`
`Endocrinology 148(5):1889-1900
`Copyright © 2002 by The Endocrine Society
`
`A Glucocorticoid-Responsive Mutant Androgen Receptor
`Exhibits Unique Ligand Specificity: Therapeutic
`Implications for Androgen-Independent Prostate Cancer
`
`ARUNA V. KRISHNAN, XIAO-YAN ZHAO*, SRILATHA SWAMI, LARS BRIVEt, DONNA M. PEEHL,
`KATHRYN R. ELY, anp DAVID FELDMAN
`
`Departments of Medicine (A.V.K., X.-Y.Z., S.S., D.F.) and Urology (D.M.P.), Stanford University School of Medicine,
`Stanford, California 94305; and The Burnham Institute (L.B., K.R.E.), La Jolla, California 92037
`
`or activate the AR“). Of the potential antagonists tested,
`The cortisol/cortisone-responsive AR (AR°*) has two muta-
`tions (L701H and T877A) that were found in the MDA PCa
`bicalutamide (casodex) and GR antagonist RU38486 showed
`inhibitory activity. We postulate that corticosteroids provide
`humanprostate cancercell lines established from a castrated
`a growth advantage to prostate cancer cells harboring the
`patient whose metastatic tumor exhibited androgen-indepen-
`promiscuous AR“in androgen-ablated patients and contrib-
`dent growth. Cortisol and cortisone bind to the AR°” with
`ute to their transition to androgen-independence. We predict
`high affinity. In the present study, we characterized the struc-
`that triamcinolone, a commonly prescribed glucocorticoid,
`tural determinants for ligand binding to the AR°*. Our data
`would be a successful therapeutic agent for men with this
`revealed that many of the C17, C19, and C21 circulatingste-
`form of cancer, perhaps in conjunction with the antagonist
`roids, at concentrations that are found in vivo, functioned as
`casodex. We hypothesize that triamcinolone administration
`effective activators of the AR°™but hadlittle or no activity via
`would inhibit the hypothalamic-pituitary-adrenal axis, thus
`the wild-type AR or GRa. Amongthesynthetic glucocorticoids
`suppressing endogenouscorticosteroids, which stimulate tu-
`tested, dexamethasone activated both GRa and AR‘,
`whereas triamcinolone wasselective for GRa. In MDA PCa 2b
`mor growth. Triamcinolone, by itself, would not activate the
`AR°*or promote tumor growth but would provide glucocor-
`cells, growth and prostate-specific antigen production were
`ticoid activity essential for survival. (Endocrinology 143:
`stimulated by potent AR°* agonists such as cortisol or 9a-
`1889-1900, 2002)
`fluorocortisol but not by triamcinolone (which did not bind to
`
`
`HE BIOLOGICAL ACTIONSof androgens are mediated
`by the AR, a memberof the nuclear hormonereceptor
`superfamily (1). The AR has been implicated in the devel-
`opment, growth, and progression of prostate cancer (2-7). In
`someprostate cancers, AR levels are elevated because of gene
`amplification and/or overexpression (8, 9), whereas in oth-
`ers, the AR is mutated (10-12). A numberof mutationsin the
`ARhave been identified in metastatic prostate cancers, and
`these mutations are most frequently located in the ligand-
`binding domain (LBD)ofthe receptor(4, 5, 11-14). ARs with
`LBD mutations, such as the T877A found in the LNCaP
`human prostate cancercell line (15) and in manyprostatic
`cancers(11, 12), exhibit broadenedligand specificity (14-16).
`For example, the T877A mutant ARis capable of responding
`to hydroxyflutamide, progesterone, and estrogens (15), al-
`though the circulating levels of both progesterone and es-
`trogens in men are low and maynotbe clinically significant
`(17). The role of AR mutations in the transition of prostate
`cancers to androgen-independent growth andin the subse-
`quent failure of endocrine therapy is the focus of recent
`studies (2-7).
`Werecently identified an AR with a double mutation
`(L701H and T877A) in its LBD in the human prostate cancer
`
`Abbreviations: AR‘", Cortisol /cortisone-responsive AR; BRFF, Bio-
`logical Research Faculty and Facility; DHT, dihydrotestosterone; FluF,
`9a-fluorocortisol; Ky, dissociation constant; LBD, ligand-binding do-
`main; MMTV, mouse mammary tumor virus; PSA, prostate-specific
`antigen; RBA,relative bindingaffinity; RU486, GR antagonist RU38486.
`
`cell lines MDA PCa 2a and MDA PCa2b,established from
`a bone metastasis of a castrated patient whose prostate cancer
`exhibited androgen-independent growth (18, 19). This
`double-mutant AR bindsthe prostatic androgen, dihydrotes-
`tosterone (DHT), with reduced affinity, compared with the
`wild-type AR or AR with the T877A mutation (20). We have
`also shownthat the double-mutant AR respondsto cortico-
`steroids such as cortisol and cortisone (20). We designated
`this mutant AR as the cortisol/cortisone-responsive AR
`(AR). The AR“is a promiscuous receptor exhibiting re-
`laxed ligand specificity, responding to glucocorticoids, an-
`drogens, progesterone, and E2, but not aldosterone(20, 21).
`In the present study, we investigated the structural re-
`quirements of ligands for the AR“',
`in comparison with
`ligandsfor the human GRa. Wetested naturalsteroids in the
`steroidogenic pathway, as well as synthetic corticosteroids,
`for their potential to act as AR“ligands. The steroids were
`evaluated in functional assays, which included binding to
`AR“ and activation of AR“'-mediated transcription. Se-
`lected corticosteroids were also tested for their ability to
`cause transactivation through the single-mutant L701H AR.
`Theabilities of key steroids to regulate the growth of MDA
`PCa 2b cells, which harbor the AR“, were evaluated; and
`their effects on the androgen-responsive target gene pros-
`tate-specific antigen (PSA) were determined. Structure-
`activity relationships were addressed by studyinga series of
`structurally related steroids.
`Our studies reveal that the AR" can be activated by a
`
`1889
`
`The Endocrine Society. Downloaded frompress.endocrine.org by[$ findividualUser.displayNamet J on 17 October 2044. al 11:39 For personal ase only. No other uses without permission,.All rights reserved.
`
`JANSSEN EXHIBIT 2024
`Mylan v. Janssen IPR2016-01332
`
`JANSSEN EXHIBIT 2024
`Mylan v. Janssen IPR2016-01332
`
`

`

`1890 Endocrinology, May 2002, 143(5):1889 –1900
`
`Krishnan et al. (cid:127) AR Mutations in Prostate Cancer
`
`number of circulating corticosteroids and their precursors.
`Cortisol and 9␣-fluorocortisol (FluF), the most potent ago-
`nists for ARccr, stimulate the growth of MDA PCa 2b cells and
`PSA secretion. The presence of ARccr would therefore pro-
`vide a growth advantage to prostate cancer cells harboring
`these mutations by responding to cortisol and other steroids
`in the steroidogenic pathway and thus contribute to andro-
`gen-independent growth and the progression of prostate
`cancer seen in androgen-ablated patients. The antiandrogen
`bicalutamide (casodex), as well as the GR antagonist
`RU38486 (RU486), acted as antagonists through the ARccr
`and inhibited growth and PSA stimulation in MDA PCa 2b
`cells. Interestingly, the synthetic glucocorticoid triamcino-
`lone was selective for GR␣and did not bind to or activate the
`ARccr. Because triamcinolone did not stimulate the growth of
`MDA PCa 2b cells or increase PSA secretion by these cells,
`it might be useful as a novel therapeutic agent to suppress
`endogenous corticosteroids in patients whose cancers ex-
`press the ARccr mutant receptors.
`
`Materials and Methods
`
`Materials
`
`All steroids were purchased from either Sigma (St. Louis, MO) or
`Steraloids, Inc. (Newport, RI). Tritiated DHT, cortisol, and dexameth-
`asone were obtained from Amersham Pharmacia Biotech (Piscataway,
`NJ). The mouse mammary tumor virus (MMTV) reporter plasmid
`pMMTV-luc and expression vectors (pSG5-AR and pSG5-GR␣) were
`gifts from Dr. Ron Evans (Salk Research Institute, San Diego, CA), Dr.
`Zoran Culig (University of Innsbruck, Innsbruck, Austria), and Dr. Peter
`Kushner (University of California, San Francisco, CA), respectively.
`Biological Research Faculty and Facility (BRFF)-HPC1 medium was
`obtained from Biological Research Faculty and Facility (Ijamsville, MD),
`and DMEM:F12 and LipofectAMINE were from Life Technologies, Inc.
`(Rockville, MD). RU486 was a kind gift from Roussel-Uclaf (Romainville,
`France).
`
`Radioligand-binding assay, Scatchard analysis, and
`competition-binding analysis
`COS-7 cells were transfected with pSG5-AR, pSG5-GR␣, or pSG5-
`ARccr expression vectors using LipofectAMINE (Life Technologies, Inc.)
`(20). After 48 h, cell monolayers were harvested, and high-salt nuclear
`extracts were made as previously described (22, 23). Protein concentra-
`tion of the extract was determined by the method of Bradford (24).
`Binding assays were done as described (22, 23). In a typical binding
`assay, 200 ␮l soluble extract (0.5–1 mg protein/ml) were incubated with
`0 –100 nm of [3H]hormone, for 16 –20 h at 0 C. Bound and free hormones
`were separated by hydroxylapatite. Specific binding was calculated by
`subtracting nonspecific binding obtained in the presence of a 250-fold
`excess of radioinert ligand from the total binding measured in the
`absence of radioinert steroid. Data were expressed as femtomoles of
`bound hormone per milligram of protein.
`Competition-binding assays were performed with extracts of COS-7
`cells expressing ARccr, in the presence of 20 nm [3H]cortisol as the ligand
`and various nonradioactive molecules as competitors at 1-, 10-, and
`100-fold excess.
`
`Reporter assay
`
`CV-1 monkey kidney cells (ATCC, Manassas, VA) were transfected
`with the expression vectors pSG5-ARccr, pSG5-GR␣, or pSG5-L701H AR,
`as well as the reporter MMTV-luc, as previously described (20). Five
`nanograms of pRL-SV40 (Promega Corp., Madison, WI) renilla lucif-
`erase were cotransfected in each sample as an internal control for trans-
`fection efficiency. The cells were treated with various steroids alone or
`in the presence of antagonists, for 16 –30 h, and luciferase activity was
`determined using the dual-luciferase assay system (Promega Corp.).
`
`Cell growth and PSA assays
`
`MDA PCa 2b cells were routinely cultured in BRFF-HPC1 medium
`supplemented with 20% FBS as previously described (18, 19). The BRFF-
`HPC1 medium contains a high concentration of cortisol (hydrocortisone,
`280 nm) as well as DHT at 0.1 nm. To test the effects of ARccr agonists,
`such as cortisol and other steroids, on cell growth and PSA secretion, we
`developed a test medium whose composition was comparable to BRFF-
`HPC1 except for the lack of cortisol and DHT. For these assays, cells were
`seeded in 6-well plates (2 ⫻ 105 cells/well) in BRFF-HPC1 medium. After
`48 h, the BRFF-HPC1 medium was replaced with DMEM:F12 medium
`supplemented with epidermal growth factor (10 ng/ml), insulin (1 ␮m),
`bovine pituitary extract (40 ␮g/ml), cholera toxin (25 ng/ml), phospho-
`ethanolamine (5 ␮m), seleneous acid (30 nm), BSA (250 ␮g/ml), and
`trypsin inhibitor (10 ␮g/ml), along with 20% FBS. We refer to this
`medium as test medium. Various steroids were added at the indicated
`concentrations in test medium. Fresh test medium and compounds were
`replenished every 3 d. The conditioned media were collected, and the
`PSA levels were measured as described (22). DNA content and [3H]thy-
`midine incorporation were assayed as measures of cell proliferation (25).
`The effects of casodex and RU486 on cell growth and PSA were assessed
`in BRFF-HPC1 medium, and their abilities to antagonize the stimulatory
`effects of endogenous cortisol and DHT present in the BRFF-HPC1
`medium were evaluated.
`
`Structural models of the LBDs
`
`Molecular models were based on an AR homology model produced
`in an earlier study (16) using the crystal structure of PR LBD as template
`(Protein Data Base accession code 1A28) (26). After this study was
`initiated, the crystal structure of the human AR LBD was solved (27, 28).
`Because there is a strong structural homology between the template
`structure of PR LBD and AR LBD (the root mean square deviation
`between ␣-carbons is 0.84 Å), predictions about the structural effects of
`mutations can be made from the homology model. Based on the x-ray
`crystallographic findings on the T877A mutant AR, Sack and co-workers
`(28) modeled the double mutant AR (ARccr) bound to DHT (described
`in Ref. 6) and obtained results similar to our modeling data reported in
`this paper.
`The sequences of the PR and AR LBD are 52% identical. A molecular
`model of the mutant ARccr LBD was produced from this model by
`substitution of histidine for leucine at residue 701 and alanine for thre-
`onine at residue 877. The histidine side chain was oriented using a
`rotamer library derived from crystallographically determined protein
`structures (29). For comparison, a homology model of GR␣ LBD (54%
`identical with PR) was constructed, in the present study, using essen-
`tially the same protocol described by McDonald et al. (16). Briefly,
`residues of PR LBD were changed to sequences of GR␣ at homologous
`sites with the program MODELLER (30), and the initial homology model
`was generated automatically. A molecule of cortisol and water mole-
`cules were added based on corresponding positions of steroid rings or
`bound waters in the template. The model was adjusted manually to
`optimize side chain rotamer positions (29) guided by the progesterone
`structure. A few local corrections employed molecular mechanics energy
`minimization using CHARMm (31) within QUANTA 97.0 (Molecular
`Simulations, Inc., San Diego, CA).
`
`Ligand-receptor docking analyses
`
`Molecular coordinates for steroids with crystallographically deter-
`mined structures were retrieved from the Cambridge Crystallographic
`Database (32) for docking analyses to AR, ARccr, or GR␣ LBD pockets.
`Ligands were manually docked into the binding pocket, orienting each
`molecule by superimposition of steroid rings onto the position of pro-
`gesterone in the PR crystal structure (26). Molecular mechanics energy
`minimization calculations using CHARMm were implemented, impos-
`ing harmonic restraints on all nonligand atoms. A number of starting
`positions/configurations were manually generated for each ligand in
`the binding pocket, and the structure with the lowest energy was se-
`lected for further analysis.
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 17 October 2014. at 11:39 For personal use only. No other uses without permission. . All rights reserved.
`
`

`

`Krishnan et al. (cid:127) AR Mutations in Prostate Cancer
`
`Endocrinology, May 2002, 143(5):1889 –1900 1891
`
`Statistical analysis
`
`Data were evaluated by ANOVA using the StatView 4.5 software
`(Abacus Concepts, Inc., Berkeley, CA), and P ⬍ 0.05 was considered
`significant.
`
`Results
`DHT and glucocorticoids bind to the ARccr
`Because ARccr is a mutant AR that responds to cortisol, we
`first determined the binding affinity of androgens and glu-
`cocorticoids for ARccr, and compared the results to the wild-
`type AR and GR␣. Dissociation constant (Kd) values of the
`ARccr, wild-type AR, and GR␣in COS-7 cells were measured
`for the binding of DHT, the major prostatic androgen, cor-
`tisol, the major circulating glucocorticoid, and dexametha-
`sone (a potent synthetic glucocorticoid). Scatchard analyses
`(Fig. 1A) revealed that [3H]DHT, [3H]cortisol, and [3H]dexa-
`methasone bound specifically to ARccr, with Kd values of 10,
`5, and 50 nm, respectively. Compared with the wild-type AR
`(Fig. 1B), the ARccr had a 50-fold reduced affinity for DHT
`⫽ 0.2 nm for wild-type AR vs. 10 nm for the
`binding (Kd
`ARccr). When compared with the GR␣(Fig. 1C), the ARccr had
`a 10-fold higher affinity for cortisol and a 25-fold lower
`affinity for dexamethasone. These results indicate that the
`ARccr has a unique ligand specificity distinct from either
`wild-type AR or GR␣.
`
`Natural corticosteroids are ARccr agonists
`Because the ARccr exhibits a high affinity for cortisol, we
`evaluated a series of cortisol-related steroids for their ability
`to bind to and activate the ARccr (Fig. 2). The structures of
`these steroids are depicted in Fig. 2C. They included natural
`corticosteroids like cortisol (11␤-hydroxycortisone), cortico-
`sterone, and their corresponding precursors (11-deoxycorti-
`sol and 11-deoxycorticosterone, respectively) as well as 18-
`hydroxycorticosterone and the mineralocorticoid hormone,
`aldosterone. We also tested 11␣-cortisol (the biologically in-
`active synthetic stereoisomer of cortisol) and cortisone,
`which has a keto group at the 11 position (the natural me-
`tabolite of cortisol that does not bind to or activate GR␣).
`Competition-binding analyses were performed using
`[3H]cortisol as the ligand and unlabeled steroids at 1, 10, and
`100 molar excess as competitors. The relative binding affinity
`(RBA) values of these steroids for ARccr ranked as follows
`(Table 1): cortisol 100% ⫽ cortisone 100% ⬎ DHT 41% ⬎
`11␣-cortisol 16% ⫽ 11-deoxycortisol 16% ⬎ corticosterone
`11% ⬎ 11-deoxycorticosterone 9% ⬎⬎ aldosterone ⬍1% ⬎⬎
`18-hydroxycorticosterone (⬍ 0.01%).
`These steroids were also tested for their transactivation
`potential, using a cotransfection assay in CV-1 cells. These
`cells were selected because they are devoid of the steroid
`receptors under investigation and lack 11␤-hydroxysteroid
`dehydrogenase (33), the enzyme that catalyzes the reversible
`conversion of cortisol to cortisone. The expression plasmids
`for the wild-type AR, ARccr, or GR␣were cotransfected into
`CV-1 cells with the MMTV-luc reporter. The cells were
`treated with 10 nm of each steroid for 30 h. We observed a
`significant difference in the extent of activation of the lucif-
`erase reporter between ARccr and GR␣. This observation can
`be explained by the fact that wild-type AR has approximately
`
`FIG. 1. The ARccr binds both androgen and glucocorticoids. COS-7
`cells were transfected with expression vectors for ARccr, wild-type AR,
`or GR␣. High-salt extracts from transfected cells were incubated with
`various doses of radioligand, at 0 C, in equilibrium-binding assays, as
`described in Materials and Methods. DEX, Dexamethasone. A, Scat-
`chard plots of [3H]DHT, [3H]cortisol, and [3H]DEX binding to ARccr;
`B, Scatchard plot of [3H]DHT binding to AR; C, Scatchard plots of
`[3H]cortisol and [3H]DEX binding to GR␣.
`
`20% of the maximal transcriptional activity of GR␣ on the
`MMTV-promoter (34). The wild-type AR could be activated
`only by androgens such as DHT and R1881. We tested a
`selected panel of corticosteroids for their ability to cause
`transactivation through the wild-type AR, and none of them
`activated the wild-type AR (data not shown).
`ARccr and GR␣ displayed distinct activation profiles in
`response to the various steroids (Fig. 2, A and B). In these
`transactivation assays, DHT and most of the cortisol-related
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 17 October 2014. at 11:39 For personal use only. No other uses without permission. . All rights reserved.
`
`

`

`1892 Endocrinology, May 2002, 143(5):1889 –1900
`
`Krishnan et al. (cid:127) AR Mutations in Prostate Cancer
`
`TABLE 1. RBA of different molecules for ARccr
`
`Competitor
`
`RBA
`RBA
`Competitor
`11
`17␤-E2
`300
`9␣-Fluorocortisol
`10.6
`Corticosterone
`100
`Cortisol
`9
`T
`100
`Cortisone
`9
`DOC
`65
`R1881
`8
`Progesterone
`60
`17-Hydroxyprogesterone
`⬍1%
`Aldosterone
`41
`DHT
`0.05
`Casodex (bicalutamide)
`30
`Spironolactone
`⬍0.01
`Pregnenolone
`26
`Prednisolone
`⬍0.01
`DHEA
`23
`Prednisone
`⬍0.01
`Androstenedione
`18
`Dexamethasone
`⬍0.01
`16.4 4-Hydroxytamoxifen
`RU486
`⬍0.01
`16.4 ICI 182780
`11-Deoxycortisol
`⬍0.01
`16.4 Triamcinolone
`11␣-Cortisol
`18-Hydroxycorticosterone ⬍0.01
`16
`Hydroxyflutamide
`RBA is expressed as the ratio of concentration of cortisol over
`concentration of competitor, each of which produces a 50% decrease
`in specific [3H]cortisol binding ⫻ 100 (mean, n ⬎ 3). In competition
`binding assays, the dose response curves have been done twice for
`each competing molecule, and the RBA values represent means from
`two experiments.
`DOC, 11-Deoxycorticosterone; DHEA, dehydroepiandrosterone.
`
`steroids, except for 18-hydroxycorticosterone (18B), acti-
`vated the ARccr and induced luciferase activity (Fig. 2A).
`Cortisol and cortisone were the most effective activators of
`the ARccr, inducing reporter levels over 30-fold above the
`basal level. Corticosterone increased reporter levels 20-fold.
`The precursor molecules of cortisol and corticosterone (11-
`deoxycortisol and 11-deoxycorticosterone,
`respectively)
`were also potent ARccr activators. Remarkably, the C11 iso-
`mer of cortisol, 11␣-cortisol, which is inactive through GR␣,
`also increased ARccr-mediated gene transactivation by 13-
`fold. Thus, the ARccr exhibited only limited stereoisomer
`specificity for the C11 position of the corticosteroids. In con-
`trast, only cortisol and corticosterone, both harboring the
`11␤-hydroxyl group, functioned as GR␣ agonists (Fig. 2B).
`Importantly, changing the stereochemistry at C11 of cortisol
`from the naturally occurring (␤) to the synthetic (␣) config-
`uration resulted in a complete loss of GR␣-mediated trans-
`activation, in contrast to the ARccr. Cortisone, which has a
`keto group at the C11 position, had no agonist activity for
`GR␣, as expected. In contrast, it was as effective as cortisol
`(which has a hydroxyl group at C11) in activating the ARccr.
`Overall, these data suggest that the ARccr has an activation
`profile distinct from those of wild-type AR and GR␣and that
`both active glucocorticoids (cortisol and corticosterone) and
`inactive corticosteroids (cortisone and 11␣-cortisol) are po-
`tent activators of the ARccr.
`
`Synthetic glucocorticoids exhibit differential agonist activity
`for the ARccr
`We next tested several commonly prescribed synthetic
`glucocorticoids, which are potent GR␣ agonists, for their
`possible agonist activity via the ARccr. These steroids each
`contain the 11␤-hydroxyl group except prednisone, which
`has a keto group in that position (Fig. 3C). They also contain
`modified A rings that are unsaturated at C1–C2 except the
`mineralocorticoid/glucocorticoid FluF. In competitive bind-
`ing assays (Table 1), FluF exhibited a 3-fold increase in bind-
`ing affinity for the ARccr, compared with cortisol. The potent
`
`FIG. 2. The ARccr differs from the GR␣in structural requirements of
`ligands at the C11 position. CV-1 cells were transfected with expres-
`sion vectors for ARccr (panel A) or GR␣ (panel B) and the reporter
`MMTV-luc as well as renilla luciferase plasmids. Cells were treated
`with the indicated steroids at 10 nM in steroid-depleted medium for
`30 h. Cell extracts were subsequently assayed for luciferase activity
`by the dual-luciferase system (Promega Corp.). Values are given as
`fold activation over activity found in control cells treated with ethanol.
`Data represent the mean of assays performed in triplicate ⫾ SEM.
`Panel C, Chemical structures of the naturally occurring corticoste-
`roids tested in panels A and B. DOC, 11-Deoxycorticosterone; B,
`corticosterone; 18B, 18-hydroxycorticosterone; S, 11-deoxycortisol; F,
`cortisol; E, cortisone; 11aF, 11␣-cortisol.
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 17 October 2014. at 11:39 For personal use only. No other uses without permission. . All rights reserved.
`
`

`

`Krishnan et al. (cid:127) AR Mutations in Prostate Cancer
`
`Endocrinology, May 2002, 143(5):1889 –1900 1893
`
`cortisone (Table 1). Thus, the double bond at C1–C2 in the A
`ring decreased the binding affinity of the steroids for ARccr.
`triamcinolone (9␣-fluoro-16␣-hydroxypred-
`Interestingly,
`nisolone), a potent synthetic glucocorticoid, which has a hy-
`droxyl group in the D ring of the sterol structure replacing
`the C16 methyl group of dexamethasone, did not bind
`to ARccr.
`In transactivation assays, the MMTV-reporter-transfected
`CV-1 cells were treated with each compound at a suboptimal
`concentration (5 nm) to detect differences in agonist activity
`between cortisol and other drugs. All of these synthetic com-
`pounds are known agonists for the GR␣- and activated GR␣-
`mediated transactivation (Fig. 3B). Prednisone, with a keto
`group at C11 position, was inactive through GR␣ as ex-
`pected, because CV-1 cells are deficient in 11␤-hydroxys-
`teroid dehydrogenase, the enzyme that catalyzes the in vivo
`conversion of prednisone to the active molecule pred-
`nisolone with a hydroxyl group at the C11 position. As
`shown in Fig. 3A, FluF had a somewhat greater activity than
`cortisol, consistent with its increased affinity for ARccr (Table
`1). The ARccr, which did not distinguish between a keto or
`hydroxyl group at C11 position, was activated by both pred-
`nisone and prednisolone. Both prednisone and prednisolone
`were comparable with cortisol in ARccr-mediated transcrip-
`tion, although they exhibited lower affinities for binding to
`ARccr. Dexamethasone, which contains a 16␣-methyl group
`and a 9␣-fluoro group in addition to the A-ring double bond,
`showed reduced activity via the ARccr. Interestingly, triam-
`cinolone containing a C16 hydroxyl group did not promote
`ARccr-mediated transactivation. Thus, the C16 hydroxyl
`group seems to abolish ARccr binding and gene activation
`through this receptor.
`In summary, our transactivation studies revealed that
`the following hormones were ARccr agonists: androgens
`[DHT, T, androstenedione, and R1881 (data not shown)];
`corticosteroids (cortisol, cortisone 11-deoxycorticosterone,
`corticosterone, 11-deoxycortisol); synthetic glucocorticoids
`(dexamethasone, prednisone, prednisolone); and the miner-
`alocorticoid/glucocorticoid (FluF). The synthetic glucocor-
`ticoid triamcinolone did not bind to or activate the ARccr.
`
`Casodex and RU486 antagonize
`ARccr-mediated transactivation
`In search of ARccr antagonists that may have therapeutic
`utility in the treatment of prostate cancers harboring this type
`of mutated receptor, we evaluated several known receptor
`antagonists. These included the AR antagonists hydroxyflu-
`tamide and casodex, the GR/PR antagonist RU486, and the
`MR/AR antagonist spironolactone. In competition-binding
`assays (Table 1), these antagonists exhibited significant bind-
`ing to the ARccr. Their RBA values ranked as follows: cortisol
`100% ⬎ spironolactone 30% ⬎ RU486 16.4% ⬎ hydroxyflu-
`tamide 11.3% ⬎⬎ casodex 0.05% (Table 1). Transactivation
`assays demonstrated that both hydroxyflutamide (20) and
`spironolactone (data not shown) functioned as ARccr agonists
`in CV-1 cells, whereas casodex and RU486 acted as antago-
`nists through the ARccr. As shown in Fig. 4, both of these
`antagonists caused significant inhibition of R1881, cortisol,
`FluF, or corticosterone-induced activation of the MMTV-luc
`
`FIG. 3. Synthetic glucocorticoids exhibit differential agonist activity
`for the ARccr. CV-1 cells were transfected with expression vectors for
`ARccr (A) or GR␣ (B) and the reporter MMTV-luc as well as renilla
`luciferase plasmids. Cells were treated with the indicated steroids at
`5 nM in steroid-depleted medium for 30 h. Cell extracts were subse-
`quently assayed for luciferase activity by dual-luciferase system (Pro-
`mega Corp.). Values are given as fold activation over activity found
`in control cells treated with ethanol. Data represent the mean of
`assays performed in triplicate ⫾ SEM. C, Structures of the synthetic
`corticosteroids tested in A and B. TRI, Triamcinolone; PSN, pred-
`nisone; PSL, prednisolone.
`
`glucocorticoids, prednisone (⌬1-dehydrocortisone), pred-
`(⌬1-dehydrocortisol), and dexamethasone (9␣-
`nisolone,
`fluoro-16␣-methylprednisolone), bound to ARccr with bind-
`ing affinities approximately 5-fold lower than cortisol and
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 17 October 2014. at 11:39 For personal use only. No other uses without permission. . All rights reserved.
`
`

`

`1894 Endocrinology, May 2002, 143(5):1889 –1900
`
`Krishnan et al. (cid:127) AR Mutations in Prostate Cancer
`
`FIG. 4. Casodex and RU486 inhibit ARccr-mediated transactivation.
`CV-1 cells were transfected with the ARccr expression vector and the
`reporter MMTV-luc and renilla luciferase plasmids as described in
`Materials and Methods. Cells were treated with the indicated ste-
`roids, at 10 nM, in steroid-depleted medium, in the presence and
`absence of the antagonists casodex (10 ␮M) or RU486 (100 nM) for 16 h.
`Cell extracts were subsequently assayed for luciferase activity by the
`dual-luciferase system (Promega Corp.). Values are given as fold
`activation over the activity found in cells treated with ethanol and
`represent mean ⫾ SEM of three to six determinations. Relative lucif-
`erase activities (MMTV-luc/renilla luc) in cells exposed to ethanol
`were 0.45 ⫾ 0.003, 0.078 ⫾ 0.02, and 0.07 ⫾ 0.001 in control, casodex-,
`and RU486-treated cells, respectively. R1881-, F-, FluF-, and B-
`induced activations of the reporter were significantly lower in casodex
`or RU486-treated cells, compared with control (P ⬍ 0.001–P ⬍
`0.0001).
`
`promoter in CV-1 cells. The degree of inhibition by RU486
`was greater than that produced by casodex, consistent with
`the fact that it exhibited a higher affinity for ARccr than
`casodex (see RBA values in Table 1). Note that triamcinolone
`was inactive and that casodex and RU486 did not exhibit any
`agonist activity in this assay.
`
`Transactivation through the L701H AR. Effects of
`glucocorticoids and casodex
`Earlier studies from our laboratory (20) have shown that
`the presence of the single mutation L701H in the AR confers
`glucocorticoid responsiveness to the mutated AR. The L701H
`AR responds to cortisol, although to a much lower degree
`when compared with the ARccr (20). In the present study, we
`attempted to characterize the responses of the L701H AR to
`key glucocorticoids in transactivation assays using the
`MMTV-luc reporter in the presence and absence of the an-
`tagonist, casodex. The results of these experiments are shown
`in Fig. 5. The L701H AR responded best to the androgen
`R1881 (⬃6-fold induction of the reporter). Both cortisol (⬃2-
`fold) and FluF (⬃3-fold) could elicit responses through the
`L701H AR. The magnitudes of these responses through the
`L701H AR were, however, much lower than their responses
`through the ARccr double mutant (Fig. 4). The other differ-
`ence between the two mutant receptor forms was in their
`ability to respond to corticosterone. Whereas corticosterone
`produced a significant activation of the reporter through the
`(⬃6-fold),
`ARccr
`it did not cause activation through
`the L701H AR. Interestingly, triamcinolone activated neither
`the ARccr (Fig. 4) nor the L701H AR (Fig. 5). The AR antag-
`
`FIG. 5. L701H AR-mediated transactivation; effects of corticoste-
`roids and casodex. CV-1 cells were transfected with the L701H AR
`expression vector, the reporter MMTV-luc, and renilla luciferase plas-
`mids, as described in Materials and Methods. Cells were treated with
`the indicated steroids, at 10 nM, in the presence or absence of 10 ␮M
`casodex, for 16 h, in steroid-depleted medium. Cell extracts were
`subsequently assayed for luciferase activity by the dual-luciferase
`system (Promega Corp.). Values are given as fold activation over the
`activity found in cells treated with ethanol and represent mean ⫾ SEM
`of six determinations. Relative luciferase activities (MMTV-luc/
`renilla luc) in cells exposed to ethanol were 0.029 ⫾ 0.01 and 0.024 ⫾
`0.01 in control and casodex treated cells, respectively. R1881-, F-, and
`FluF-induced activation of the reporter were significantly lower in
`casodex-treated cells, compared with control (P ⬍ 0.01–P ⬍ 0.001).
`
`onist casodex again caused significant inhibition of the
`R1881, cortisol, or FluF-mediated activation of the L701H AR.
`
`Effects of agonists on MDA PCa 2b cell growth
`We next examined the ability of some of the key steroids
`to activate the endogenous ARccr expressed in MDA PCa 2b
`cells and thereby modulate their growth and PSA secretion.
`MDA PCa cells grow best in BRFF-HPC1 medium, which
`contains a high concentration of cortisol (280 nm), and we use
`this medium to routinely culture and passage these cells. To
`test the effects of steroids on growth, the BRFF-HPC1 me-
`dium was replaced, 48 h after plating, with the test medium
`(described in Materials and Methods) lacking cortisol and DHT
`but containing 20% FBS, because the use of serum stripped
`of endogenous steroids and other growth factors could not
`support cell growth. The initial DNA concentrations at the
`beginning of these experiments ranged between 1.6 –2.2 ␮g/
`well. Although the cells grew in the test medium, their
`growth was minimal, and the DNA concentrations at the end
`of the 6-d period were 3.5– 4.5 ␮g/well in various experi-
`ments. Supplementation of the test medium with cortisol or
`use of the BRFF-HPC1 medium with high endogenous cor-
`tisol resulted in substantial increases in cell growth, as de-
`termined by [3H]thymidine incorporation and DNA content
`(Fig. 6). When cultured in BRFF-HPC1 medium, the DNA
`content increased to 10 –15 ␮g/well at the end of 6 d. When
`cortisol (10 –200 nm) was added back to the test medium we
`found increases in