Inhibition of prostate cancer cell growth by second-site androgen receptor antagonists
Abstract
The impact of ligand binding on nuclear receptor (NR) structure and the ability of target cells to distinguish between different receptor-ligand complexes are key determinants of the pharmacological activity of NR ligands. However, until relatively recently, these mechanistic insights have not been used in a prospective manner to develop screens for NR modulators with specific therapeutic activities. Driven by the need for unique androgen receptor (AR) antagonists that retain activity in hormone-refractory prostate cancer, we developed and applied a conformation-based screen to identify AR antagonists that were mechanistically distinct from existing drugs of this class. Two molecules were identified by using this approach, D36 and D80, which interact with AR in a unique manner and allosterically inhibit AR agonist activity. Unlike the clinically important antiandrogens, casodex and hydroxyflutamide, both D36 and D80 block androgen action in cellular models of hormone-refractory prostate cancer. Mechanistically, these compounds further distinguish themselves from classical AR antagonists in that they do not promote AR nuclear translocation and quantitatively inhibit the association of AR with DNA even under conditions of overexpression. Although the therapeutic potential of these antiandrogens is apparent, it is the demonstration that it is possible, to modulate the interaction of cofactors with agonist-activated AR, using second-site modulators, that has the greatest potential with respect to the therapeutic exploitation of AR and other NRs.
Recently published estimates indicate that prostate cancer accounts for 33% of all new cancer cases diagnosed in males within the United States (1). Fortunately, early detection and advancements in therapeutic options have resulted in an improvement in the overall outcome of this disease. Indeed, a large percentage of prostate cancers present as low to medium risk, require little intervention and exhibit 10-year mortality rates close to 24%. However, patients with advanced and metastatic prostate cancer have a mean survival of 2.5–3 years and usually require aggressive treatment including surgery and/or chemical castration therapy (2). Unfortunately, although often initially successful, metastatic tumors inevitably become resistant to hormonal therapy, a stage of the disease for which there is no curative treatment.
Prostate cancers resistant to any of the currently prescribed hormonal therapies are referred to as “hormone-refractory” or “androgen-independent,” implying that they have progressed beyond the point at which drugs targeting the androgen axis would have clinical utility. However, despite this traditional classification, there is compelling evidence that the androgen receptor (AR) remains a viable therapeutic target in cancers that progress while on traditional castration regimens (3). Notably, a large percentage of patients with hormone-refractory tumors who become resistant to either (i) androgen blockade therapies (LH-RH agonist) or (ii) maximal androgen blockade (MAB, LH-RH agonist plus antiandrogen) often have a favorable response to the substitution of an alternative antiandrogen in the MAB (4, 5). Additionally, the well-described antiandrogen withdrawal syndrome, observed in ≤30% of patients following discontinuation of antiandrogen therapy, implicates AR and its downstream activities in the pathology of these tumors (6). Although the molecular mechanisms underlying this phenotypic response are not fully understood, mutations that confer on AR the ability to recognize antagonists as agonists have been identified in a significant number of patients exhibiting the withdrawal response (7). More recently, it has become clear that up-regulation of AR itself, or some of its attendant co-regulatory proteins, are also associated with progression to the hormone-refractory state in both patients and animal models (8–10). Additionally, the observation, using gene-array technology, that genes shown to be AR-responsive in treatment-naïve tumors are expressed in a constitutive manner in tumors classified as hormone-refractory, provides some of the most compelling evidence to date supporting a role for constitutively active AR in androgen-independent tumors (9). Finally, the clinical efficacy of the second generation AR antagonist, MDV3100, provides strong evidence that therapeutics that target AR by means other than those currently available may have utility in the treatment of hormone-refractory prostate cancer (11).
Although there are recent reports describing the use of AR-cofactor interactions to identify small molecule antagonists of AR (12), traditionally these compounds have been identified by (a) classical ligand binding assays or (b) cellular reporter assays that assess functional antagonism of androgen-dependent transcriptional activation. Although these approaches have been successful in identifying clinically useful AR antagonists, these compounds share a remarkable degree of functional similarity. Of note is the finding that all of the currently available antiandrogens manifest partial agonist activity, the degree of which is influenced by AR (and possibly cofactor) expression (8). It has now been demonstrated that there is a strong relationship between the structure of a specific nuclear receptor-ligand complex and its functional activity (13). Indeed, the estrogen receptor (ER) antagonists ICI182,780 and GW5638, both of which inhibit the growth of tamoxifen refractory breast tumors, induce receptor conformations distinct from tamoxifen and other ER antagonists (14). Thus, using a conformation-directed approach, we sought to identify AR ligands (antagonists or inverse agonists) that would be active in models of hormone-refractory or androgen-independent prostate cancer.
Results
Development of a Conformational-Based Screen for AR Antagonists.
Using a T7-phage display screen intended to identify AR-interacting proteins, we isolated the previously characterized AR-binding protein gelsolin (15). Consistent with published reports, all AR ligands tested, regardless of biological activity, facilitate the interaction of AR with gelsolin in a mammalian 2-hybrid assay (Fig. 1A) (16). Based on the ligand-dependency of the AR/gelsolin interaction and the observation that gelsolin overexpression can increase the partial agonist activities of Casodex (Csx) and hydroxyflutamide (OHF), we reasoned that the AR/gelsolin interaction would be useful as a conformation-based screen for AR antagonists. To this end, we screened a diverse small molecule library of ≈10,000 compounds based on >70 chemical scaffolds for compounds that inhibit the AR/gelsolin interaction to levels approaching that of no hormone in the presence of an EC70 concentration of the synthetic agonist R1881. Of the compounds screened, 87 scored positive in the primary mammalian 2-hybrid-based assay with no demonstrable cell toxicity. The 87 compounds were subsequently tested for their ability to inhibit agonist-mediated transcription of an MMTV-luciferase reporter in LNCaP cells. Two of the 49 compounds found to inhibit AR transcriptional activity in this assay (D36 and D80) (Fig. 1B) were brought forward for a more complete characterization.
Identification of AR ligands. (A) Ligand dependency of AR/gelsolin interaction using the mammalian 2-hybrid assay. HepG2 cells were transfected with AR-VP16 and a Gal4 DBD-Gelsolin fusion protein along with a Gal4-responsive reporter gene and pCMV-βGal. Cells were induced for 48 h with indicated ligand. Data are presented as normalized response that was obtained by normalizing luciferase activity to β-galactosidase activity. Transfection was performed in triplicate and error bars represent standard deviation. (B) Chemical structures of D36 and D80. (C) D36 and D80 disrupt the AR/gelsolin interaction. 293 t cells were transfected with AR-VP16 and a Gal4 DBD-Gelsolin fusion protein and data were normalized as in A. (D) Differential protein interaction defines D36 and D80 as a unique AR ligand class. Mammalian 2-hybrid assay was performed as in A except that Renilla-luciferase was used to normalize for transfection efficiency. Interaction profiles of 24 AR ligands, including D36 and D80, and vehicle controls were generated by using 10 Gal4-interactor fusion proteins. The profiles were analyzed with the Ward hierarchical cluster algorithm using standardized data. The resulting dendrogram and structural activity heatmap demonstrates the relationships between the 6 structure-based clusters.
Because D36 and D80 possess the ability to disrupt the interaction of AR with gelsolin (Fig. 1C), we sought to determine whether these compounds induce a unique conformation in AR. Thus, we tested both compounds in a recently developed AR conformation profiling tool (15). This mammalian 2-hybrid based assay classifies AR ligands by virtue of their ability to modulate the interaction of full-length AR with a set of interacting proteins. Importantly, by using this conformational profiling tool we have demonstrated that the ligand-induced receptor conformation correlates with the biological activity. Thus, we profiled D36 and D80, together with a chemically and biologically diverse set of sentinel AR ligands, by using 10 AR conformational probes described in Table S1. A hierarchical cluster analysis of the conformational profile is shown in Fig. 1D. To avoid signal strength bias, the interaction data for each individual protein was standardized. The individual interaction profiles were then clustered by hierarchical analysis by using the Ward hierarchical cluster algorithm (17). Although there are some similarities in the AR conformation observed in the presence of all antagonists or very weak agonists (Csx, OHF, LG120907, R2, and CPA) (cluster 1, red), it was clear that the compounds we identified most closely resemble the profile of the apo or unliganded receptor (cluster 3, blue). The notable exception to this observation is the anticipated ability of D36 and D80 to disrupt the interaction of AR with gelsolin. We conclude from these studies that D36 and D80 represent a class of antagonists that promote a conformational change in AR that is distinguishable from that induced on binding both Csx and OHF.
D36 and D80 Are Allosteric AR Antagonists.
We determined that both D36 and D80 decrease the maximal R1881 stimulated AR-dependent activation of an MMTV-luciferase reporter transcriptional activity (Vmax) in LNCaP cells although slightly increasing the apparent EC50 (at 30 μM the EC50 is increased less than 3-fold for D36 and 12-fold for D80). These results indicate that these compounds are likely to be functioning as allosteric AR antagonists (Fig. 2A). The Ki for D36 and D80 are calculated to be 9.0 ± 0.37 μM and 17.0 ± 0.86 μM, respectively. The ability of D36 and D80 to disrupt agonist binding was confirmed by 3H-R1881 whole-cell-competition binding assay. Unlabeled OHF, D36, and D80 each effectively compete against R1881 binding in VCaP cells at concentrations near their predicted IC50 (Fig. S1).
D36 and D80 are mixed competitive AR antagonists. (A) D36 and D80 inhibit AR antagonist activity. MMTV-luciferase activity was measured in transfected LNCaP cells with an increasing dose of R1881 and 6 concentrations of competitor as indicated (in μM). Sigmoidal dose–response curves were created in Prism4 (GraphPad). (B) Binding of D36 and D80 to AR as determined by surface plasmon resonance. Full-length recombinant AR saturated with 1 μM R1881 was immobilized onto the surface of a CM5 (Sensorchip SA) at a level of approximately 10,000 response units (RUs). Antagonists were injected at a rate of 20 μL/min at concentrations of 150 μM to 20 μM. (C) Curve fit of the dissociation and association binding phases, separately, to the 67 μM data, assuming a 1:1 (Langmuir) binding model. The measured data are denoted by the solid line and the fit is represented by the dashed line.
The unique pharmacological profiles of both D36 and D80 prompted us to confirm the direct interaction between these compounds and AR by using surface plasmon resonance (SPR). Fig. 2B shows the sensorgrams for D36 and D80 binding to immobilized full-length AR (saturated with R1881, 1 μM). D36 also bound to the LBD alone and with similar affinity as the full-length AR (Fig. S2). However, D80 shows diminished binding to the LBD in comparison to the full-length AR. The kinetic association and dissociation rate constants of kon = 135 (M−1s−1), koff = 0.012 (s−1), Kd = 90 μM for D36, and kon = 665 (M−1s−1), koff = 0.027 (s−1), Kd = 40 μM for D80 were calculated from the sensorgrams assuming a 1:1 (Langmuir) binding model by using Biacore software. Fig. 2C shows representative fits for a 64 μM injection of each compound. The binding affinity of both compounds to full-length AR was similar in the presence of 1 μM DHT, suggesting that the interaction is independent of the saturating agonist. The binding of D36 and D80 as measured by SPR provides evidence for a direct interaction between these compounds and AR.
D36 and D80 Antagonize Endogenous AR Activity.
We next sought to determine whether D36 and D80 could antagonize endogenous wild-type AR transcriptional activity in the prostate cancer cell line, LAPC4 (18). As seen in Fig. 3A, D36 and D80, as well as the control OHF, inhibit R1881-mediated PSA mRNA expression in a dose-dependent manner. In addition to PSA, D36 and D80 were found to inhibit androgen-induced expression of several endogenous AR target genes in LAPC4 and LNCaP prostate cancer cells (Fig. S3). We next tested whether D36 and D80 could inhibit androgen-mediated LAPC4 prostate cancer cell proliferation. A dose-response study of OHF, D36, and D80 both in the presence and absence of 500 pM R1881 is presented in Fig. 3B. As can be seen, OHF is capable of inhibiting androgen-stimulated proliferation but has no effect on the basal, presumably AR-dependent proliferative activity. Similar results are observed with Csx. However, D36 and D80 inhibit both basal and androgen-stimulated proliferation. This activity is not due to cell toxicity as the cell number, measured by relative DNA content, is not reduced to levels below that determined at the time of initial androgen stimulation (represented by the dashed line in Fig. 3B). These results are consistent with the results obtained when AR levels are reduced by using siRNA technology and suggest that D36 and D80 function as AR inverse agonists (19).
D36 and D80 inhibit AR-mediated transcription and proliferation in LAPC4 cells. (A) Inhibition of PSA transcription. LAPC4 cells were induced with ligand as indicated and PSA mRNA levels were measured using real-time PCR. GAPDH was used to normalize RNA expression. (B) OHF, D36, and D80 inhibit AR-mediated LAPC4 proliferation. Cells were plated in 96-well plates and serum-starved for 3 days followed by antagonist treatment in the presence or absence of 500 pM R1881. Proliferation is quantified by measuring dsDNA content after 7 days of compound treatment. The DNA content at the time of antagonist addition is indicated by the dashed line.
D36 and D80 Inhibit AR in a Model of Hormone-Refractory Prostate Cancer.
Considering that D36 and D80 functioned as inverse agonists in the LAPC4 proliferation assay, we next sought to test the ability of the compounds to antagonize AR function in a model of hormone-refractory prostate cancer. In humans as well as in mouse xenograft models, the progression of prostate cancer from hormone-sensitive to hormone-refractory is frequently associated with heightened AR activity often because of receptor overexpression. One consequence of elevated AR levels is the conversion of antagonists, such as Csx and OHF, into partial agonists. Thus, we sought to determine whether D36 and D80 retained their antagonist properties in this model. As previously described, in LNCaP cells infected with retrovirus expressing AR (SRαAR), but not in cells infected with the control virus (SRα), Csx displays weak agonist activity in an MMTV-luciferase reporter assay (8). However, unlike Csx, D36 and D80 display no agonist activity in either the SRα or SRαAR LNCaP cells (Fig. S4). When tested for antagonist activity in the SRα and SRαAR cells, Csx, D36, and D80 suppress R1881- mediated MMTV-Luc transcriptional activity in a dose-dependent manner (Fig. S5 and Fig. 4A). However, both D36 and D80 efficiently inhibit R1881 and Csx-stimulated MMTV-Luc activity in the SRαAR cells (Fig. 4B). These data suggest that D36 and D80 represent a mechanistically distinct class of AR antagonists.
D36 and D80 inhibit AR action in models of hormone refractory prostate cancer. (A) Inhibition of R1881 mediated transcription in SRαAR-infected LNCaP cells. Virally infected LNCaP cells were transiently transfected with MMTV-luciferase and treated with 100 pM R1881 plus the indicated compounds. (B) Inhibition of Csx-mediated transcription in SRαAR-infected LNCaP cells. SRαAR-infected LNCaP cells were transfected with MMTV-Luciferase and 1 μM Csx plus the indicated antagonists. (C) Csx, D36, and D80 inhibit SRα LNCaP proliferation. Proliferation assay was performed and presented as described in Fig. 3B. R1881 was used at 100 pM. (D) Effect of Csx, D36, and D80 on SRαAR LNCaP cell proliferation. R1881 was used at 100 pM. (E) D36 and D80 inhibit VCaP cell proliferation. VCaP proliferation was performed as described for LNCaP and LAPC4 prostate cancer cell lines. Cells were treated with ligands for 7 days as indicated and proliferation was determined by quantification of dsDNA.
In addition to effects on transcription, AR overexpression also imparts on Csx the ability to stimulate LNCaP cell proliferation. As seen in Fig. 4C, SRα infected control LNCaP cells (SRα) proliferate in response to R1881, an activity that can be inhibited in a dose-dependent manner by Csx as well as D36 and D80. Importantly, whereas AR overexpression confers agonist properties to Csx (Fig. 4D), D36 and D80 inhibit R1881-stimulated SRαAR LNCaP cell proliferation despite overexpression of the receptor. This effect is not restricted to the virus-infected LNCaP cells or because of the fact that the cells are expressing mutated and wild-type AR as we observe similar effects in the VCaP cell line, which expresses high levels of wild-type AR (Fig. 4E and ref. 20). Thus, the AR antagonists described here function as AR inverse agonists in normal prostate cancer cells (LAPC4 and LNCaP) and in cell models of prostate cancer expressing high AR levels (SRαAR LNCaP and VCaP).
D36 and D80 Are Mechanistically Distinct AR Antagonists.
In an effort to define the mechanism(s) by which D36 and D80 function as AR inverse agonists, we first sought to determine whether these compounds targeted AR for degradation. Several previously identified AR antagonists, particularly those derived from natural products, function in this manner (21). However, as seen in Fig. 5A and Fig. S6, treatment with 30 μM D36 or D80 for 24 h has minimal effects on AR levels in both LAPC4 and LNCaP cells when normalized to GAPDH expression. Given the modest effects of D36 and D80 on receptor levels, we next asked whether they altered AR cellular localization. Similar to R1881, both Csx and OHF have previously been shown to induce AR nuclear translocation (22). Thus, we tested whether D36 or D80 could promote AR nuclear translocation in either the hormone refractory model SRαAR LNCaP or the control SRα LNCaP cells. As shown in Fig. 5B and unlike R1881, OHF, and Csx, neither D36 nor D80 promote AR nuclear translocation in SRαAR LNCaP cells. Similar results were obtained in the wild-type and SRα-infected LNCaP cells. Moreover, similar effects are seen at lower antagonist concentrations as well as at different time points (Fig. S7). Interestingly, treatment of LNCaP cells with D36 and D80 does not appear to inhibit R1881-mediated AR nuclear translocation (Fig. S8). Thus, unlike the therapeutically important AR antagonists OHF and Csx, neither D36 nor D80 alter AR subcellular localization.
D36 and D80 are mechanistically distinct AR antagonists. (A) D36 and D80 lead to a moderate reduction in AR protein levels. Protein extracts were collected and AR levels were analyzed by western blot following 24-h ligand treatment. AR protein levels were quantified and normalized to GAPDH and expressed relative to the vehicle control. (B) D36 and D80 do not induce AR nuclear translocation. LNCaP SRαAR cells were treated with the indicated ligand for 4 h followed by fractionation of the nuclear and cytoplasmic compartments. αTubulin and PARP were used as controls for the cytoplasmic and nuclear fractions, respectively. (C) D36 and D80 disrupt AR and cofactor association with chromatin. Chromatin immunoprecipitation from SRα and SRαAR infected LNCaP cells was performed 4 h after ligand treatment (1.0 μM Csx, 100 nM R1881, and 30 μM D36 and D80). QPCR quantification was performed by using primers directed to distal PSA enhancer following precipitation with AR, RNA PolII, p300, SRC1, and Trap220 antibodies. IgG was used as a negative control, and the data are presented as percent of immunoprecipitation input.
OHF and Csx both promote AR nuclear translocation and under certain conditions, such as receptor overexpression, induce robust DNA binding and cofactor recruitment subsequent to transcriptional activation (8). To further discern the mechanistic differences between Csx and D36 and D80, we performed chromatin immunoprecipitation in both SRα and SRαAR LNCaP cells and compared the ability of the R1881, vehicle and antagonists, to recruit AR and AR-cofactors to the enhancer region of the PSA gene. As can be seen in Fig. 5C, only R1881 induces AR recruitment to the PSA enhancer in comparison to the vehicle control in the SRα control cells. On the other hand, D36 and D80 both lead to a slight reduction in AR recruitment. Interestingly, Csx did not induce AR recruitment to the PSA enhancer as has been previously reported. This result may be because of the fact that we used 1.0 μM Csx compared with the 10 μM that was used in those studies, suggesting that high concentrations of Csx are required for efficient delivery of AR to DNA (8). As expected, AR overexpression not only enhances R1881- mediated AR recruitment but also promotes the appearance of significant Csx-mediated receptor recruitment. However, even in the SRαAR cells, D36 and D80 do not stimulate but actually inhibited the basal levels (DMSO) of AR recruitment to the PSA enhancer. Consistent with the results from the AR ChIP, D36 and D80 repressed basal Pol II and cofactor (p300, SRC1, and TRAP220) recruitment to the PSA enhancer to levels significantly below the vehicle control. These effects are most pronounced in the SRαAR cells where the basal levels of cofactors present at the PSA enhancer are significantly higher than in the control cells. Similar to Chen et al. (8), we observe that AR overexpression allows for a subset of cofactors to be recruited to the AR/DNA complex by Csx. In our hands, in the SRαAR-infected cells, Csx stimulates RNA polymerase II and TRAP220 recruitment, and has little effect on p300 or SRC1. These results confirm that D36 and D80 function as AR transcriptional inverse agonists at the level of DNA binding and cofactor recruitment, a mechanistically unique mode of antagonism compared with the currently available AR therapeutics.
Discussion
Although it is commonly accepted that nuclear receptor pharmacology/function is dictated by ligand-induced conformational changes in receptor structure, conformation-based screens have only recently been used for drug discovery. In this study, we have exploited the unique nature of the ligand-induced AR/gelsolin interaction to identify 2 AR inverse agonists D36 and D80 that, unlike classical antagonists, allosterically block AR activity. Furthermore, although these compounds disrupt the interaction between AR and gelsolin, they do not appear to dramatically alter the conformation of AR when compared with the apo state. Instead, they are likely interacting with a surface distinct from the traditional ligand-binding pocket potentially disrupting a yet unidentified protein-protein interaction surface on AR that is critical for its function. Importantly, this work demonstrates that the biocharacter of ligands bound to the classical ligand-binding domain of AR receptor can be modulated by compounds that interact with the receptor in a unique manner. This finding presents an opportunity to explore these second-site interactions in the development of a class of modulators that alter the pharmacology of endogenous androgens and other synthetic agonist and antagonists.
Although the use of NR-cofactor interactions as the basis for small molecule screens are not uncommon, they are typically designed using peptides or protein fragments derived from cofactors that engage the AF-2 cofactor-binding pocket (12, 23, 24). Furthermore, these cofactor-based screens are generally performed in vitro using the isolated ligand-binding domain as the target. This approach is based on our current understanding that the predominant ligand-induced transcriptional activation property of most nuclear receptors resides within the AF-2 pocket of the LBD. However, small molecule screens configured in this manner ignore not only the AF-1 transcriptional activation domain, often located in the amino terminus of the receptor, but also subtle changes in the full-length receptor conformation that may reflect substantial changes in biological activity. Additionally, screens of this nature have little ability to discriminate between mechanistically distinct classes of antagonists. In the cases of the ER and AR, this is particularly problematic given the high level of cross resistance observed among different anti-estrogens and antiandrogens observed in breast and prostate cancers, respectively (4, 5, 25). Thus conformation-based screens, such as the AR/gelsolin screen described here, capable of identifying classes of NR antagonists could speed the discovery and development of therapeutics useful in the treatment of hormone-refractory breast and prostate cancers.
Although we have not fully elucidated the molecular mechanisms by which D36 and D80 abrogate AR activity, it is clear that they directly bind AR and block its activity in an allosteric manner. Lack of crystallographic evidence precludes us from determining where these molecules bind AR; however, it is intriguing to speculate that they interact with the recently identified BF-3 surface (12). Ligands engaging the BF-3 pocket (flufenamic acid, Triac, and triiodothyronine) have been demonstrated to allosterically alter the AF-2 binding pocket of AR, thereby weakening the interaction of cofactors with this transcriptional domain. In a similar manner, the results of our conformational profiling assay suggest that D36 and D80 do not induce the formation of an active AF-2 pocket in AR as they do not demonstrate a ligand-dependent interaction with the AF-2 coactivator ARA54. However, although it does not eliminate the BF-3 pocket hypothesis, it is interesting to note that none of the known BF-3-binding compounds alter the interaction of AR with gelsolin in our hands. Given the unique properties of D36 and D80, defining the surface(s) of AR that are engaged by these compounds is an active interest of the laboratory.
The ability of D36 and D80 to inhibit agonist induced transcription and proliferation in the LNCaP-AR overexpressing cells suggest that compounds of this nature, when optimized for affinity, may be viable therapeutics for hormone-refractory prostate cancer. Although the molecular mechanisms promoting the progression of prostate cancer to the hormone-refractory state are not fully understood, it has become clear that up-regulation of AR, is often associated with this transition in patients (9, 10). The relatively modest 2- to 3-fold overexpression of AR in LNCaP cells mimics progression to hormone-refractory disease in both cell-based assays and xenograft models. Mechanistically, AR overexpression not only sensitizes LNCaP cells to low levels of androgens but also allows the classical antagonists, OHF and Csx, to function as partial agonists by enabling them to recruit AR and cofactors to DNA (8). Importantly, neither D36 nor D80 induce AR recruitment to DNA regardless of AR expression levels, suggesting that they would be viable AR antagonists in both hormone-sensitive and hormone-refractory prostate cancer. We are currently optimizing D36 and D80 to test their efficacy in animal models of prostate cancer. Thus, we anticipate the development of conformation-based NR ligand screens, such as the one presented here, could speed the discovery of mechanistically unique therapeutically useful NR modulators.
Materials and Methods
Chemicals and Plasmids.
Testosterone, stanozolol, and oxandrolone were purchased from Steraloids. R1881 was purchased from PerkinElmer. Cyproterone acetate and RU-486 were purchased from Sigma. Hydroxyflutamide was purchased from Toronto Research Chemicals. RTI-6413–001 was synthesized by Edgar Cook (RTI International). All other AR ligands were synthesized and provided by Drs. Philip Turnbull and Timothy Willson at GlaxoSmithKline.
Transfection Assays.
The transfection and compound profiling experiments are detailed in SI Methods.
AR/Gelsolin Screen.
Hek293 cells were plated in DMEM and bulk transfected with Lipofectin (Invitrogen) according to the manufacturer's protocol, using the following DNAs, VP16-AR, 5XGalLuc3, pM-Gelsolin, and phRL-CMV. After 24 h, the cells were trypsinized and 100-μL cells at a concentration of 2.5 × 105/mL were seeded into 96-well plates containing 50-μL DMEM plus either 0.3% DMSO, 0.3% DMSO plus 300 nM R1881, 0.3% DMSO plus 0.3 μM R1881, or 30 μM compound plus 150 pM R1881. Following a 48-h incubation, the cells were lysed and assayed for firefly and renilla luciferase activity.
Whole Cell Competition Binding Assay.
Details of the whole cell competition binding experiment are provided in SI Methods.
Cell Proliferation Assay.
LAPC4, LNCaP, and VCaP cells were maintained in Iscove's DMEM (plus 15% FBS and 0.1 nM R1881), RPMI-1640 (plus 10% FBS), or DMEM (plus 10% FBS), respectively. For proliferation assays, cells were plated in 96-well plates in the appropriate medium supplemented with charcoal-stripped FBS (15% for Iscove's, 10% for RPMI and DMEM) at 10,000 cells/well. Following a 72-h incubation, cells were treated with antagonist in the presence or absence of R1881. Cells were treated with ligands an additional 2 times at 72-h intervals. Twenty-four hours after the final treatment, cell proliferation was quantified by measuring dsDNA content by using the FluoReporter dsDNA quantitation kit (Invitrogen).
Androgen Receptor Expression and Purification.
Full-length AR protein with a biotin tag was expressed in SF9 cells and purified by using a Streptavidin mutein column (Roche) as previously described (26). The ligand binding domain (LBD, amino acids 663–919) of AR was cloned into pGEX-6P. This AR (LBD)-GST fusion protein was expressed in BL21 cells and FPLC purified by using GST-Sepharose (GE Healthcare), followed by gel filtration (HiLoad 16/60 Superdex 75, GE Healthcare).
Surface Plasmon Resonance.
The details of the surface plasmon resonance experiments are provided in SI Methods.
RNA Isolation and Real-Time PCR.
LAPC4 cells were treated for 24 h with ligand and RNA was isolated by using the Aurum total RNA isolation kit (Bio-Rad). RNA (1 μg) was reverse transcribed by using the Bio-Rad iScript cDNA synthesis kit. Real-time PCR was performed by using the Applied Biosystems 7300 instrument and iQ SYBR Green supermix (Bio-Rad). GAPDH expression was used to normalize all real-time data. Real-time primer sequences are listed in Table S2.
Generation of SRα and SRαAR LNCaP Cells.
LNCaP cells were infected with the empty SRα or SRα virus expressing wild-type AR (gift of C. Sawyers, Memorial Sloan–Kettering Cancer Center). Infected cells were selected by growth in RPMI1640 supplemented with 400 μg/mL G418 (Invitrogen).
AR Protein Levels and Localization.
For analysis of AR protein levels LAPC4 and LNCaP cells were plated in medium containing charcoal-stripped FBS for 3 days, followed by ligand treatment for 24 h. Whole-cell protein extracts were collected by using the M-PER reagent (Pierce). Fractionation experiments were performed by using LNCaP cells or LNCaP cells infected with SRα or SRαAR virus. Cells were grown for 3 days in media containing charcoal-stripped FBS and treated with ligand for the indicated time. Cells were fractionated by using the NE-PER reagents (Pierce) according to the manufacturer's protocol. Western blots were performed using the following antibodies; AR (AR441, gift from D. Edwards, Baylor College of Medicine), GAPDH (0411, Santa Cruz), α-tubulin (E-19, Santa Cruz) and PARP-1 (C2–10, Santa Cruz).
Chromatin Immunoprecipitation Assay.
Details of the chromatin immunoprecipitation assay are provided in SI Methods.
Acknowledgments
This work was supported by a grant to D.P.M. from the National Institutes of Health (CA139818).
Footnotes
- 3To whom correspondence should be addressed. E-mail: donald.mcdonnell{at}duke.edu
-
Edited by Bert W. O'Malley, Baylor College of Medicine, Houston, TX, and approved May 28, 2009
-
Author contributions: J.D.J., D.P.M., and J.D.N. designed research; J.D.J., B.M.W., M.A.D., H.C., D.A.D., and J.D.N. performed research; J.D.J., B.M.W., M.A.D., and J.D.N. analyzed data; and J.D.J., D.P.M., and J.D.N. wrote the paper.
-
↵1Present address: Integrated Oncology Solutions Inc., 7020 Kit Creek Road, Suite 130, P.O. Box 13399, Research Triangle Park, NC 27709.
-
↵2Present address: GlaxoSmithKline, 5 Moore Drive, Research Triangle Park, NC 27709-3398.
-
The authors declare no conflict of interest.
-
This article is a PNAS Direct Submission.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0900185106/DCSupplemental.
