Specific Androgen Receptor Activation by an Artificial Coactivator*

Transcription activation of steroid receptors, such as the androgen receptor (AR), is mediated by coactivators, which bridge the receptor to the preinitiation complex. To develop a tool for studying the role of the AR in normal development and disease, we constructed artificial coactivators consisting of the transcription activation domains of VP16 or p65/RelA and the AR hinge and ligand-binding domain (ARLBD), which has been shown to interact with the AR N-terminal domain. The artificial VP16-ARLBD and ARLBD-p65 coactivators interacted with the AR N terminus and wild-type AR in an androgen-dependent and androgen-specific manner. VP16-ARLBD and ARLBD-p65 enhanced the AR transactivity up to 4- and 13-fold, respectively, without affecting the expression of the AR protein. The coactivators did not enhance the transcription activity of the progesterone receptor (PR) or the glucocorticoid receptor (GR), showing their specificity for the AR. In addition, to construct PR- and GR-specific coactivators, the VP16 activation domain was fused to the PR and GR hinge/ligand-binding domain. Although VP16-PRLBD and VP16-GRLBDinteracted with the C-terminal portion of steroid receptor coactivator-1, they did not enhance the transcription activity of their receptor. The presented strategy of directing activation domains or other protein activities into the DNA-bound AR complex provides a novel means of manipulating AR function in vitro and in vivo.

Steroid receptors are hormone-dependent transcription factors that regulate the expression of a large variety of genes affecting cell growth, differentiation, development, and homeostasis. How these nuclear receptors activate or repress transcription of target genes has been the focus of much research in the past years. A major breakthrough in our understanding of these activities was the identification of so-called coactivators that bridge the receptors to the preinitiation complex (PIC) (1)(2)(3)(4). 1 These coactivators are recruited into the promoterbound complex by the receptor and facilitate assembly of basal transcription factors into a stable PIC, likely via their activa-tion domains. In addition to this bridging function, some coactivators, including steroid receptor coactivator-1 (SRC-1), cAMP response element-binding protein-binding protein (CBP), p300, and ACTR, can also remodel chromatin by acetylating histones (5)(6)(7)(8). Moreover, the nuclear receptors, ACTR, SRC-1, CBP, and p300 can recruit the p300/CBP-associated factor (9), which also harbors an intrinsic histone acetyltransferase activity (6,7,10,11). The model that emerges from these data takes the two activities of coactivators into account; the liganded steroid receptor binds as a homodimer to the hormone response element and recruits coactivators and p300/CBP-associated factor. These cofactors loosen the nucleosomal structure by targeted histone acetylation. The coactivators can then initiate the stable assembly of the PIC by their bridging function, which results in enhanced rates of transcription initiation by RNA polymerase II (9).
To perform their bridging function, coactivators need to harbor at least two domains: a receptor-interacting domain and a domain that contacts and stabilizes assembly of the PIC. This latter domain will almost certainly be a transcription activation domain that directly or indirectly binds proteins of the PIC. The receptor-interacting domain determines the binding specificity. Except for AR-associated protein 70 (12), none of the known coactivators is specific for a certain nuclear receptor (1,3). In addition, SRC-1 and CBP can mediate transactivity of transcription factors other than nuclear receptors (13)(14)(15).
The objective of this study was to generate an AR-specific and very potent coactivator to manipulate AR function in vitro and in vivo. Such a coactivator could be used as a tool to study the role of the AR in normal development and disease. Moreover, this coactivator could form the blueprint for the construction of artificial corepressors that inhibit AR function in the presence of ligand, of cofactors that direct any other enzyme activity into the AR DNA-bound complex, and of coactivators that are specific for other nuclear receptors.
To form a coactivator that is AR-specific and more potent than the known coactivators, a domain had to be found that interacts only with the AR and not with other nuclear receptors. The first evidence for the existence of such a domain came from the observation that the AR ligand-binding domain (AR LBD ) binds to the AR N terminus in an androgen-dependent manner (16 -18). However, it was unknown whether the AR LBD could also interact with the wild-type AR and whether this binding is AR-specific.
The activation domains are important determinants of coactivator potency and are responsible for the direct interaction with proteins of the PIC and/or recruitment of additional coactivators. There are a few obvious choices for strong activation domains, including the activation domain from viral protein 16 (VP16) (19) and p65/RelA (20). Both have been shown to directly contact proteins of the PIC and associate with coactivators (21-30).
Cell Culture and Transient Transfections-HeLa cells (human epithelial cervix carcinoma (American Type Culture Collection)) were maintained in minimal essential medium supplemented with 5% fetal bovine serum and antibiotics. LNCaP cells (American Type Culture Collection) were maintained in RPMI medium supplemented with 10% fetal bovine serum and antibiotics. 24 h before transfection, 10 5 cells were plated in each well of 12-well dishes in medium containing dextran-coated charcoal-stripped serum. Cells were transfected with 0.3 g of reporter plasmid, 0.1 g of pcDNA-LacZ, 30 ng of receptor construct, and 0.1 g of coactivator plasmid per well using Lipofectin (Life Technologies, Inc.) according to the manufacturer's guidelines. 24 h later, cells were washed and fed with medium containing stripped serum and the indicated hormones. Cells were harvested 14 h later, and cell extracts were assayed for luciferase activity using the luciferase assay system (Promega). Luciferase values were corrected for the ␤-galactosidase internal control. Experiments were performed in triplicate; data are presented as the means Ϯ S.E. of at least three independent experiments.
Immunoblots-Transfected HeLa cells were lysed in 40 mM Tris-HCl pH 7.0, 1 nM EDTA, 4% glycerol, 10 mM dithiothreitol, 2% SDS, and protease inhibitors. Protein concentration of the samples were determined by Bradford assay using 10 g of protein loaded in sample buffer on a 7.5% SDS-polyacrylamide gel as described previously (36). Gels were blotted onto nitrocellulose, and the protein of interest was visualized with the AR-specific F39.4 antibody (37) or Xpress antibody (Invitrogen) and a goat-anti-mouse horseradish peroxidase secondary antibody using the ECL kit (Amersham Pharmacia Biotech). The pcDNA3.1His vectors that were used for the construction of almost all plasmids described above harbor the Xpress epitope tag to which the Xpress antibody is directed.

Construction of Artificial AR Coactivators-To investigate
whether the AR LBD interacts with the wild-type AR, various fusion proteins were constructed containing the AR LBD and the activation domains from VP16 and p65. The various constructs that were used in this study are depicted in Fig. 1. The wildtype AR (AR0) cotransfected with the (ARE) 2 E1b-luciferase reporter in HeLa cells was activated up to 50-fold by 1 nM synthetic androgen R1881 (Fig. 2A). Cotransfection with empty pcDNA3.1His vector, AR LBD , VP16 activation domain, or p65 activation domain did not significantly change the capacity of AR to activate transcription. However, fusion proteins of the AR LBD with VP16 or p65 greatly enhanced AR transactivity. The location of the VP16 activation domain with respect to the AR LBD did not influence the coactivation potential of the VP16-AR LBD and AR LBD -VP16 fusion proteins. We cotransfected all the different coactivators and their components with the reporter plasmid in the absence and presence of R1881 to test their potential effect on the transcription of the reporter in the absence of the AR. As expected, none of the coactivators or their components changed the basal transcription from the (ARE) 2 E1b-luciferase reporter (data not shown). To show that endogenous AR can also be superactivated by coactivators, VP16-AR LBD and AR LBD -p65 were transiently transfected with the reporter plasmid into the prostate cancer cell line LNCaP. LNCaP cells express the AR and are sensitive to androgens with respect to their growth. Transfection of the reporter alone showed that R1881 can activate the endogenous AR (Fig. 2B). Cotransfection of the AR LBD or p65 activation domain did not significantly affect AR activity. However, as was observed in the HeLa cell transfections, the AR LBD -p65 coactivator strongly enhanced the AR up to 9-fold.

Lack of Effect of Artificial Coactivators and Their Components on AR Protein
Expression-To ensure that cotransfection of the AR with the artificial coactivators did not increase AR expression and consequently transcription activity from the reporter, HeLa cells were cotransfected, and the AR protein expression was determined by Western blot analysis. The AR expression was not affected by the coactivators or their components (Fig. 3).
The VP16-AR LBD and AR LBD -p65 Coactivators Are Specifically Androgen-dependent-Because the artificial coactivators harbor the AR LBD , we tested whether the VP16-AR LBD and AR LBD -p65 fusion proteins enhanced the AR in a hormone-dependent manner. The AR5 mutant lacks the AR ligand-binding domain and is constitutively active (32). As expected, the addition of R1881 to the HeLa cells cotransfected with AR5 and the (ARE) 2 E1b-luciferase reporter did not affect AR5 activity (Fig. 4). The artificial AR LBD -p65 coactivator did not enhance AR5 transactivity in the absence of ligand but did superactivate transcription in the presence of 1 nM R1881. This result was repeated with the VP16-AR LBD fusion protein, showing that the coactivators containing the AR LBD depend on ligand for function. To investigate to which part of the AR N-terminal domain the AR LBD binds, we tested two different AR mutants harboring different fragments of the N terminus for enhanced transcription activation by AR LBD -p65. Both AR126 (amino acids 1-370) and AR106 (amino acids 360 -528) were only partially enhanced by the artificial coactivator as compared with AR5, indicating the necessity for both fragments for full AR LBD -p65 interaction. In addition, the binding of the artificial coactivator to the AR DNA-binding domain and ligand-binding domain was analyzed using AR104, which completely lacks the N-terminal domain. The transcription activity of this mutant is very low (32), and in this study AR LBD -p65 did not enhance its transcription (Fig. 4).
We used this same experimental approach to determine whether the fusion of the AR LBD to VP16 or p65 changed ligand specificity. Various agonists (R1881, testosterone, and dihy- FIG. 5. Ligand specificity of the AR LBD -p65 coactivator. HeLa cells were transfected with AR5, (ARE) 2 E1b-luciferase reporter and empty vector or AR LBD -p65 coactivator. Luciferase activity was determined from cell lysates of transfected cells, which were cultured for 16 h in the absence (Ϫ) or presence (ϩ) of 1 nM R1881, 1 nM testosterone (T), 1 nM dihydrotestosterone (DHT), 100 nM estradiol (E2), 100 nM synthetic progestin R5020, 100 nM antiandrogen Casodex, 100 nM antiandrogen hydroxyflutamide (OH-flu), or 100 nM antiandrogen cyproterone acetate (CA). Activities were corrected for a pcDNA3.1His-LacZ internal control and are presented as the percentage of luciferase activity Ϯ S.E. relative to the AR5 activity in the absence of R1881 (first column from the left).

FIG. 6. Superactivation of the AR, PR, and GR by AR LBD -p65.
HeLa cells were transfected with AR, PR, or GR, (ARE) 2 E1b-luciferase reporter and empty vector, or AR LBD -p65 coactivator. Luciferase activity was determined from cell lysates of transfected cells, which were cultured for 16 h in the absence (Ϫ) or presence (ϩ) of 1 nM ligand or a mix of ligands for AR (R1881), PR (R5020), and GR (dexamethasone). Activities were corrected for a pcDNA3.1His-LacZ internal control and are presented as the percentage of luciferase activity Ϯ S.E. relative to the activity of AR (second column from the left), PR (sixth column from the left), or GR (twelfth column from the left) in the presence of ligand. FIG. 7. Functional analysis of VP16 and AR, PR, or GR ligand-binding domain fusion proteins for superactivation of their wild-type receptor. HeLa cells were transfected with AR, PR, or GR, (ARE) 2 E1b-luciferase reporter and empty vector, or VP16-AR LBD , VP16-PR LBD , or VP16-GR LBD fusion proteins. Luciferase activity was determined from cell lysates of transfected cells, which were cultured for 16 h in the absence (Ϫ) or presence (ϩ) of 1 nM ligand or a mix of ligands for AR (R1881), PR (R5020), and GR (dexamethasone). Activities were corrected for a pcDNA3.1His-LacZ internal control and are presented as the percentage of luciferase activity Ϯ S.E. relative to the activity of AR (second column from the left), PR (fifteenth column from the left), or GR (twenty-eighth column from the left) in the presence of ligand. drotestosterone), antagonists (Casodex, hydroxyflutamide, and cyproterone acetate), and other hormones (estradiol and the synthetic progestin R5020) were tested for their ability to induce AR LBD -p65 binding to AR5 (Fig. 5). Only the androgens testosterone, dihydrotestosterone, and R1881 significantly induced AR5 superactivation, showing that the artificial coactivator did not change ligand specificity and remained androgen-specific.
The VP16-AR LBD and AR LBD -p65 Coactivators Are AR-specific-To determine whether the AR LBD would interact only with the AR and not with other steroid receptors, we analyzed the effect of the artificial coactivators on two of the most homologous nuclear receptors, the progesterone receptor (PR) and the glucocorticoid receptor (GR). Cotransfection of the PR or GR with AR LBD -p65 in the presence of R1881 and R5020 or dexamethasone for activation of the PR and GR, respectively, showed no superactivation (Fig. 6), indicating that the AR LBD -p65 coactivator is AR-specific. Note that R1881 is a potent progestin and is able to activate the PR (Fig. 6).
Testing of Artificial PR and GR Coactivators-To determine whether the hinge/ligand-binding domains of the PR and GR can also interact with their receptors, the VP16 activation domain was fused to the PR LBD and GR LBD . Although their construction was almost identical to that of VP16-AR LBD , neither VP16-PR LBD nor VP16-GR LBD enhanced transactivation of its full-length receptor (Fig. 7). For control purposes, we cotransfected the activation domain of VP16 by itself with the different receptors to analyze the effect of this potent activation domain on receptor transactivity. The activity of all receptors was reduced, probably because of sequestering of essential transcriptional cofactors. Interestingly, the GR activity was the most severely reduced (down to 25%). Cotransfection of the GR with VP16-AR LBD or VP16-PR LBD in the presence of the various ligands again showed the transcription-inhibiting effect of VP16 (Fig. 7). However, in the absence of the ligand that interacts with the VP16-AR LBD (R1881) or VP16-PR LBD (R5020), the GR activity was again higher. One explanation for this phenomenon is that in the unliganded VP16-AR LBD and VP16-PR LBD , the VP16 is shielded by the heat shock proteins that associate with the ligand-binding domain. In the presence of ligand, the heat shock proteins dissociate, and VP16 is able to squelch the GR transactivity.
To check correct folding and the ability to bind hormone, we cotransfected VP16-PR LBD and VP16-GR LBD with the receptor interacting-part of the SRC-1 coactivator, which is fused to the GAL DNA-binding domain (DBD) (GAL-SRC-1[1139 -1441]) (38,39). None of the VP16 fusion proteins interacted with the GAL DBD (Fig. 8). However, VP16-AR LBD , AR LBD -VP16, AR LBD -p65, VP16-PR LBD , and VP16-GR LBD bound the GAL-SRC-1(1139 -1441) in a hormone-dependent manner, showing correct expression, nuclear import, folding, and hormone binding. DISCUSSION Our results show for the first time that the AR LBD can interact with the full-length AR but not with the PR or GR. The binding of the artificial coactivators occurred in the AR Nterminal domain and was diminished when the first 360 amino acids or the last 158 residues of the N terminus were deleted, confirming the previous observation that two independent domains (the first 36 amino acids and residues 370 -494) are necessary for full AR LBD binding (40).
Binding of the hinge/LBD to the N terminus or full-length receptor has also been shown for the estrogen receptor ␣ (41) and the PR (42). However, VP16 fusion proteins with the PR and GR hinge/LBD did not enhance the transcription activity of their receptor, indicating that they did not bind the full-length receptor in our mammalian protein-protein interaction system. The folding, ligand binding, and nuclear import of VP16-PR LBD and VP16-GR LBD were correct because they interacted with the receptor-interacting domain of SRC-1 in a ligand-dependent manner (38,39). The interaction of VP16-AR LBD and AR LBD -p65 with SRC-1(1134 -1441) was very weak compared with the binding of VP16-PR LBD and VP16-GR LBD , confirming previously published data showing that the AR binding to SRC-1 is the weakest of all steroid receptors (39). The reason for the inability of VP16-PR LBD and VP16-GR LBD to bind to their cognate receptor is not clear. We speculate that the threedimensional structure of the AR DNA-bound homodimer is different as compared with the PR and GR homodimers, allowing an additional ligand-binding domain to be recruited into the AR complex.
ing FK506 to a short peptide with transactivating properties. FK506 interacts with the FK-binding protein FKBP12, which was fused to the GAL DNA-binding domain, thereby directing the transactivating peptide onto the reporter gene. As a complementary strategy, we show that the AR LBD , which binds the full-length AR also gives us the opportunity to direct proteins into the DNA-bound AR complex. In addition to the transcription activation domains described in this manuscript, the effects of transcription repression domains, DNA-modifying enzymes, and chromatin-remodeling proteins on AR function are currently being tested. These artificial cofactors provide a novel means of manipulating AR activity and AR target gene expression to investigate the role of AR function in normal and disease states.