Modulation of Androgen Receptor Transactivation by FoxH1 A NEWLY IDENTIFIED ANDROGEN RECEPTOR COREPRESSOR *

Androgen signaling plays key roles in the development and progression of prostate cancer, and numerous ongoing studies focus on the regulation of androgen receptor (AR) transactivity to develop novel therapies for the treatment of androgen-independent pros-tatecancer.FoxH1,amemberoftheForkhead-box(FOX)genefam- ily of transcription factors, takes part in mediating transforming growth factor- (cid:1) /activin signaling through its interaction with the Smad2 (cid:1) Smad4 complex. Using a series of experiments, we found that FoxH1 repressed both ligand-dependent and -independent transactivation of the AR on androgen-induced promoters. This action of FoxH1 was independent of its transactivation capacity and activin A but relieved by Smad2 (cid:1) Smad4. In addition, the repression of the AR by FoxH1 did not require deacetylase activity. A protein-protein interaction was identified between the AR and FoxH1 inde-pendently of dihydrotestosterone. Furthermore, a confocal micro-scopicanalysisofLNCaPcellsrevealedthattheinteractionbetween immunoprecipitated with an anti-AR or normal rabbit immunoprecipitated with anti-Myc

Prostate cancer is a significant cause of morbidity and mortality worldwide. Androgens play major roles in promoting the development and progression of prostate cancer (1)(2)(3), and therefore, androgen ablation and blockade of androgen actions through the androgen receptor (AR) 3 have been the cornerstones of treatments for advanced prostate cancer. Despite these regimens, prostate cancer invariably progresses to a fatal, androgen-refractory state (4,5). However, although such relapsed tumors are androgen-independent, they are still dependent on the AR for their growth and survival (4, 6 -8). Therefore, identification of the precise mechanisms underlying the regulation of AR function is of critical importance for the design and development of novel therapies and pharmaceutical targets for treating prostate cancers.
The AR shares a characteristic structure with other members of the steroid hormone receptor family (comprised of receptors for estrogens (ERs), progesterone (PR), glucocorticoids (GR), and mineralocorticoids), namely a variable NH 2 -terminal transactivation domain (NTD) possessing an activation function 1 (AF-1) domain, a highly conserved zinc finger-type DNA binding domain (DBD), and a ligand binding domain (LBD) that usually contains a second activation domain (AF-2) (9, 10). AF-1 functions in a ligand-independent manner, whereas the activity of AF-2 requires cognate ligand binding (9,11,12). Upon activation by ligands, the AR translocates to the nucleus, where it binds to androgen response elements and regulates the transcription of target genes. Moreover, it has become clear that the transactivity of nuclear receptors, including the AR, is regulated by coregulator proteins that enhance (coactivators) or reduce (corepressors) the target gene transcription by various mechanisms (10,13,14). Although most of the AR coregulators identified to date have been coactivators, it is conceivable that AR corepressors are also required for precise and efficient regulation of the AR activity in cells (13,15). Therefore, further characterization of AR corepressors may provide insights into the signaling events that occur within prostate cancer and pave the way to the development of individualized therapies.
Activins, which are members of the TGF-␤ superfamily, are composed of two ␤ subunits, ␤A and ␤B, which form activin A (␤A␤A), activin B (␤B␤B), and activin AB (␤A␤B) (16). In addition to their stimulatory effects on pituitary follicle-stimulating hormone synthesis, activins have also been implicated in the control of many other cellular processes, including growth and tumorigenesis (17,18). The presence of activin A and its receptors in the prostate (19 -22) and the ability of activin A to inhibit prostate cancer cells grown in culture (23)(24)(25)(26) suggest an important role for activin A in the regulation of prostatic growth. Moreover, activin A has been shown to induce the expression of prostate-specific antigen (PSA), prostatic acid phosphatase, and the AR (26), genes that are also induced by androgen. Although the molecular mechanism through which activin A regulates gene expression and growth in the prostate has not yet been fully elucidated, cross-talk between activin A and androgen-mediated signaling pathways may play critical roles in these processes (27).
Smads are a family of proteins that function as effectors of the TGF-␤/activin-signaling pathway. Ligand addition induces the phosphorylation of specific receptor-regulated Smads (R-Smads), which then oligomerize with the common mediator Smad4 and move into the nucleus. Once there, the R-Smad⅐Smad4 complex interacts with a vari-ety of transcription factors and coregulators and becomes targeted to a diverse array of gene promoters (28,29). Interestingly, some coregulators of Smads, such as AP-1 (30), CBP/P300 (31,32), and TGIF (33), can also regulate AR-mediated transactivation (34 -37). FoxH1 (also known as FAST-1), a member of the Forkhead-box (FOX) gene family of transcription factors, plays an important role in mediating TGF-␤/activin signals through its interaction with the Smad2⅐Smad4 complex (38 -40). We hypothesized that FoxH1 may also function as a coregulator to regulate the AR transactivation potential and, therefore, be involved in the cross-talk between activin A and androgen-mediated signaling. In the present study we found that FoxH1 could repress ligand-dependent and -independent transactivation of the AR on androgen-induced promoters. However, the repression of the AR by FoxH1 was not alleviated by activin A, indicating that FoxH1 was dispensable for the stimulatory effect of activin A on PSA expression (26,27,41). Nonetheless, our results demonstrate that FoxH1 is a new corepressor of the AR.
AR and FoxH1 expression vectors for mammalian two-hybrid assays were subcloned into the pBIND and pACT expression vectors, respectively. pFoxH1-Myc and pFoxH1 H83R -Myc were constructed by subcloning the FoxH1 and FoxH1 H83R cDNAs into pcDNA3.1/Myc-His, respectively. pFoxH1-GFP was constructed by inserting the FoxH1 cDNA into pEGFP-N1 (BD Biosciences Clontech, Palo Alto, CA). The validity of the structure of each construct was confirmed by DNA sequencing and Western blot analysis of transfected COS-7 cells.
RNA Preparation and RT-PCR-Total RNA was extracted using ISOGENE (Wako, Osaka, Japan) according to the manufacturer's instructions. The concentration and purity of the RNA were determined spectrophotometrically. Next, 5 g of total RNA was reversetranscribed into first-strand cDNA using a SuperScript III kit (Invitrogen) in a final volume of 20 l. To analyze the expression of FoxH1 in prostate cancer cell lines, a sensitive RT-PCR was performed using previously described primers (39). The PCR was carried out in a 50-l reaction mixture containing 2.5 mM MgCl 2 , 0.3 mM dNTP, and 2.5 units of Taq DNA polymerase (Promega) under the following conditions: 35 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 1 min. Aliquots of the PCR products were electrophoresed in 2% agarose gels containing 0.5 mg/ml ethidium bromide and then photographed under UV light using a positive/negative instant film (Polaroid 665; Nippon-Polaroid, Tokyo, Japan). The authenticity of each PCR product was confirmed by sequencing.
Transactivation Assays-The human prostate cancer cell lines LNCaP and PC-3 were maintained in RPMI 1640 medium (Sigma) supplemented with 10% fetal bovine serum (Sera Laboratories Ltd., Sussex, UK). COS-7 monkey kidney cells and ALVA41 human prostate cancer cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The cells were cotransfected with the indicated expression vectors in 24-well plates using an Effectene transfection kit (Qiagen K. K., Tokyo, Japan) according to the manufacturer's protocol. The cells were then incubated in RPMI 1640 or Dulbecco's modified Eagle's medium containing 0.5% dextrancoated charcoal-stripped fetal bovine serum and vehicle (0.1% ethanol) or ligands as indicated. After 24 h, the firefly and Renilla luciferase activities were assayed using the Dual-Luciferase Reporter assay system (Promega) according to the manufacturer's protocol in a Minilumat LB9507 (Berthold Technologies, Bad Wildbad, Germany). The results were normalized for the internal Renilla control and presented as the relative luciferase activity. All transfection experiments were carried out in triplicate wells and repeated at least three times using two sets of plasmids prepared separately. Data were calculated as the mean Ϯ S.D.
Stable Transfection of LNCaP Cells with pFoxH1-Myc, Semiquantitative RT-PCR, and Western Blotting-LNCaP cells were cultured in six-well plates and then transfected with pFoxH1-Myc using an Effectene transfection kit. After 4 weeks of culturing and selection with 400 g/ml of Geneticin (Invitrogen), 6 colonies were harvested. After limited dilution, 2 independent FoxH1-Myc-expressing clones (designated LNCaP/FoxH1-1 and LNCaP/FoxH1-2) were identified by Western blotting with an anti-Myc antibody (1:1000; sc-40; Santa Cruz Biotechnology, Inc.) and further maintained as stable cell lines in RPMI 1640 supplemented with 200 g/ml Geneticin. Cells stably transfected with the empty vector (LNCaP/Vector) served as a control.
The effect of stably expressing FoxH1 on the endogenous PSA level in the absence or presence of 100 nM dihydrotestosterone (DHT) was investigated by semiquantitative RT-PCR and Western blotting, respectively. For semiquantitative RT-PCR, total RNA was extracted and reverse-transcribed as described above. Preliminary experiments were conducted to ensure linearity for the semiquantitative procedures. Hotstart PCR was performed by heat-activating AmpliTaq Gold DNA polymerase (PerkinElmer Life Sciences) at 95°C for 10 min. Optimized cycling condition was 30 cycles (for PSA) or 26 cycles (for glyceraldehyde-3-phosphate dehydrogenase) of 1 min at 94°C, 1 min at 56°C, and 1 min at 72°C. The primer sequences specific for PSA and glyceraldehyde-3-phosphate dehydrogenase were described previously (48,49). For Western blotting, the cells were harvested, and the protein concentration of each sample was measured using a BCA protein assay kit (Pierce). Aliquots containing 20 g of protein were separated in 10% SDS-PAGE gels and transferred onto nitrocellulose membranes. Next, the membranes were probed with the anti-human PSA antibody A67-B/E3 (1:200; sc-7316; Santa Cruz Biotechnology Inc.) or an anti-␤-actin antibody (1:500; AC-74; Sigma). Bands were visualized using an alkaline phosphatase system.
Microscopy and Imaging Analyses-Microscopy and imaging analyses were performed essentially as described previously (44 -46). The cells were imaged for green fluorescence by excitation with the 488-nm line from an argon laser, and the emission was viewed through a 496 -505-nm band pass filter.
Immunoprecipitation and Western Blotting-COS-7 cells were cotransfected with 2 g of pFoxH1-Myc or pFoxH1 H83R -Myc together with 2 g of the pCMVhAR or pCMV parent vector in 6-well plates using Lipofectamine (Invitrogen) according to the manufacturer's instructions. Transfected cells were cultured in the presence or absence of 10 nM DHT for 24 h and then harvested in celLytic TM -M (Sigma) containing 1ϫ Complete Mini-EDTA-free protease inhibitor mixture (Roche Applied Science). The protein concentration of each sample was measured and adjusted to 1 mg/ml. Next, an aliquot (160 g) of each lysate was incubated with 3 g of the anti-AR antibody C-19 (sc-815; Santa Cruz Biotechnology Inc.) or normal rabbit IgG as a control in TNE buffer (10 mM Tris-HCl, pH 7.8, 1% Nonidet P-40, 1 mM EDTA, 150 mM NaCl) containing 1ϫ Complete Mini-EDTA-free protease inhibitor mixture and then further incubated with 25 l of protein G magnetic beads (New England Biolabs Inc., Beverly, MA) at 4°C for 2 h on a rotating platform. The beads were collected using a magnet, and the bound proteins were eluted in 1ϫ SDS-PAGE sample buffer and subjected to 10% SDS-PAGE. A Western blot analysis was performed using the anti-Myc antibody (1:500) as described above, and the positive bands were detected using an enhanced chemiluminescence detection system (Amersham Biosciences) and a VersaDoc TM imaging system (Bio-Rad).
Mammalian Two-hybrid Assays-Mammalian two-hybrid assays (Promega) were mainly performed according the manufacturer's protocol, with some modifications. NIH3T3 cells were transiently cotransfected with the indicated vectors in 24-well plates using the Superfect transfection reagent (Qiagen K. K.) in the presence or absence of 10 nM DHT. Luciferase activities were assayed as described above.
Statistical Analysis-Statistical significance was determined by onefactor analysis of variance followed by a post hoc test (Fisher's protected least significant difference test). p Ͻ 0.05 was considered statistically significant.

Expression of FoxH1 in Prostate Cancer
Cell Lines-A previous study (39) reported that human FoxH1 gene expression was ubiquitous and could be detected in all normal human tissues tested as well as in several cancer cell lines. However, whether FoxH1 is expressed in LNCaP cells as well as in other prostate cancer lines has yet not been elucidated. Therefore, to investigate the role of FoxH1 in the cross-talk between activin A-and androgen-mediated signaling in prostate cancer, we initially used the primers described in the above-mentioned study (39) to investigate FoxH1 expression in the prostate cancer lines ALVA41, DU145, LNCaP, and PC-3 by RT-PCR. These primers spanned a 100-bp intron and discriminated between the spliced (423 bp) and unspliced (523 bp) products. As shown in Fig. 1A, a 423-bp band was detected in all 4 prostate cancer cell lines. Moreover, three of the cell lines also contained the unspliced product, which may arise from either genomic DNA or unprocessed transcripts.
FoxH1 Represses Both Ligand-dependent and -independent Transactivation of the AR in LNCaP Cells-Next, we examined the possible roles of FoxH1 expression in ligand-dependent and -independent transcription of the PSA promoter induced by endogenous AR by cotransfecting LNCaP cells, the most commonly used androgen-sensitive prostate cancer cell line, with a FoxH1 expression plasmid and a PSA-luciferase reporter gene. As shown in Fig. 1B, DHT activated the AR in a concentration-dependent manner, and an ϳ10-fold higher induction was observed in the presence of 1 M DHT compared with vehicle treatment. Cotransfection of the FoxH1 expression plasmid brought about a marked repression of the DHT-induced AR activation at all concentrations of DHT tested (Fig. 1B), and the repression was dose-dependent (Fig. 1C). Furthermore, FoxH1 completely blocked the stimulatory effects of progesterone and the anti-androgens cyproterone acetate and hydroxyflutamide on the AR in LNCaP cells, which contains a mutation (T877A) that results in alterations of the specificity and sensitivity of the receptor to these molecules (50,51) as well as the ligand-independent transactivation of the AR by interleukin-6 (52-54) and forskolin (55, 56) (Fig. 1D). In these and subsequent experiments, the inhibitory effect of FoxH1 on the AR transactivity was not due to a reduced AR expression level, since immunoblot analyses of extracts from the transfected cells revealed comparable amounts of immunoreactive protein (Fig. 1E). Together, these results demonstrate that FoxH1 represses both liganddependent and -independent transactivation of endogenous AR in LNCaP cells.
To further explore the relevance of FoxH1 in AR-mediated transactivation, we examined the relationship between FoxH1 expression and endogenous PSA expression in LNCaP cells. pFoxH1-Myc was stably transfected into LNCaP cells, and 2 isolated clones expressing similar protein levels of FoxH1-Myc, LNCaP/FoxH1-1, and LNCaP/FoxH1-2 were established (Fig. 1F, top panel) to investigate the effect of FoxH1 on PSA expression. As expected, stable expression of FoxH1 resulted in a parallel reduction of PSA expression in either mRNA or protein levels in both the absence and presence of DHT but had little effect on the levels of the housekeeping genes glyceraldehyde-3-phosphate dehydrogenase and ␤-actin (Fig. 1F, middle and bottom panels), indicating that FoxH1 could regulate the expression of endogenous androgen-responsive genes.
Repression of FoxH1 Is Not Rescued by Activin A but Is Relieved by Smad2/4 Proteins-Because FoxH1 mediates transcriptional responses to TGF-␤/activin in a ligand-, receptor-, and Smad-dependent fashion (39)and activin A has been shown to induce PSA expression in LNCaP cells (26,27,41), we next investigated whether activin A and its effectors Smad2⅐Smad4 could alleviate the repression of the AR by FoxH1 in LNCaP cells. Consistent with previous reports, activin A induced transcription from the PSA promoter, and the PSA promoter induction after treatment with both DHT and activin A was additive compared with the values observed with either reagent alone ( Fig. 2A). Unexpectedly, FoxH1 could still repress the AR-mediated transactivation of the PSA promoter in the presence of activin A, indicating that FoxH1-mediated inhibition was independent of activin A in LNCaP cells and that FoxH1 was dispensable for the stimulatory effect of activin A on PSA expression (26,27,41). On the other hand, in agreement with a previous report (57) neither Smad2 nor Smad4 alone had a signifi-  cant effect on the AR-dependent transcription, whereas coexpression of Smad2 or Smad4 resulted in complete or partial relief of the repression by FoxH1, respectively (Fig. 2B). These results suggest that Smad2⅐Smad4 may be negative regulatory factors for the repression of the AR by FoxH1 through competition with the AR for binding to the limiting cellular FoxH1.

FoxH1-mediated Repression of AR
Repression of FoxH1 Is Neither Cell-type-Nor Promoter Contextdependent-To further explore the biological significance of the FoxH1-mediated repression, we examined the repression effect of FoxH1 on the wild-type AR in COS-7 and PC-3 cells using both pPSA-LUC and pMMTV-LUC as reporter genes. As shown in Fig. 3, FoxH1 as well as its mutant FoxH1 H83R , which has a mutation in the DNA binding Forkhead domain and completely lacks transcriptional activity (39), still significantly repressed AR-regulated transcription from either the PSA or MMTV promoter in both COS-7 and PC-3 cells. Notably, the basal transcription from the reporter genes was also reduced in the presence of FoxH1 but only in the presence of the AR (Fig. 3). In addition, the internal control Renilla luciferase activity was not influenced by FoxH1. Therefore, the repression by FoxH1 was specific for the AR and not a general transcriptional inhibition. These observations suggest that the inhibitory effect of FoxH1 on the AR is neither cell type-nor promoter context-dependent and is independent of its transactivity.
Selective Repression by FoxH1-Several of the corepressors identified to date, such as N-CoR and SMRT, can repress the transcriptional activity of other steroid hormone receptors as well as that of the AR (58 -61). Therefore, it is important to investigate whether the repression by FoxH1 is specific for the AR or a more general phenomenon. To address this question, we tested GR, ER-␣, ER-␤, and PR isoforms in PC-3 cells under similar experimental conditions to those used for the AR. As shown in Fig. 4, these receptors showed ligand-dependent transactivation in the presence of their appropriate ligands, and cotransfection of FoxH1 resulted in obvious repressions of ER-␣ and ER-␤ as well as of the AR. In contrast, FoxH1 showed no significant repression of the transactivation of the GR and PR isoforms, demonstrating that the repression of the AR by FoxH1 did not result from competition with the AR for binding to the hormone response elements. In these experiments, the internal control Renilla luciferase activity was stable. Collectively, these results suggest that FoxH1 may be involved in multiple regulatory processes mediated by steroid hormone receptors.

Repression of the AR by FoxH1 Does Not Require Deacetylase Activity-
Recent studies have shown that the transcriptional repression of the AR by some corepressors, such as TGIF (37) and ARR19 (62), is mediated through histone deacetylase pathways. In this regard, we examined whether trichostatin A (TSA), a specific inhibitor of histone deacetylase activity (63), had any influence on the repression of the AR by FoxH1 in LNCaP cells. Consistent with a previous report (64), treatment with 10 Ϫ7 M TSA slightly increased the DHT-induced AR transactivity by about 30% (Fig. 5). However, TSA had no significant effect on the FoxH1-mediated inhibition at any of the doses tested, indicating that the repression of the AR by FoxH1 does not require deacetylase activity and that FoxH1 may repress AR-mediated transactivation through other mechanisms.
FoxH1 Abrogates Nuclear Foci Formation by the AR-We previously reported that DHT-bound AR formed foci in the nucleus, which were correlated with AR-mediated transactivations (44,45). More recently, it was reported that the Tob-mediated suppression of AR activity may result from inhibition of the AR foci formation (65). Therefore, the prevention of AR foci formation may be responsible for the repression by FoxH1. To verify this hypothesis, we first examined the intracellular localization of the FoxH1 protein. As expected, the majority of FoxH1-GFP was homogeneously distributed in the nucleus in all the cell lines examined (Fig. 6, A-C). No significant changes were observed in the subcellular distribution of FoxH1-GFP before and after the addition of DHT or activin A. Consistent with our previous report (44), after cotransfection of LNCaP cells with pAR-GFP and the empty pCMV vector, the majority of AR-GFP was located homogeneously in the cytoplasm in untreated control cells (Fig. 6D), whereas treatment with 10 nM DHT resulted in nuclear translocation and fine foci formation by the AR in the nucleus (Fig. 6E). However, the addition of activin A (25 ng/ml) had little effect on the distribution of either unliganded or liganded AR (Fig. 6, F and G). In contrast, cotransfection with FoxH1 resulted in disruption of the DHT-induced foci formation, and AR-GFP was distributed homogeneously in the nucleus (Fig. 6H). In agreement with the transactivation assays, treatment with activin A did not recover the disruption of the DHT-induced foci formation by FoxH1 (Fig. 6I). Furthermore, replacement of pCMVFoxH1 by pCMVFoxH1 H83R produced similar results (Fig. 6, J and K). Notably, and consistent with our previous report (45), CBP, a general integrator for nuclear receptors, was distributed in a mixed pattern with fine foci formation in a diffuse background in the nucleus (Fig. 6L). Cotransfection of FoxH1 had no obvious effect on the nuclear distribution of CBP (Fig. 6M), demonstrating that the disruption of foci formation by FoxH1 was specific for the AR rather than a general phenomenon.
FoxH1 Interacts Physically with AR-As a result of the above observations, we examined whether there was a physical interaction between the AR and FoxH1 using a coimmunoprecipitation analysis. pFoxH1-Myc was cotransfected into COS-7 cells with the empty pCMV vector or pCMVhAR. Cell extracts were immunoprecipitated with normal rabbit IgG (negative control) or an anti-AR antibody (C-19) followed by Western blotting with an anti-Myc monoclonal antibody. As shown in Fig. 7A, upper panel, FoxH1 protein was detected in anti-AR immunoprecipitates from cells cotransfected with the AR and FoxH1 in the absence or presence of 10 nM DHT but not in those from cells transfected with the AR or FoxH1 alone or when normal rabbit IgG was used. A similar result was obtained when pFoxH1-Myc was replaced by pFoxH1 H83R -Myc (Fig. 7A, lower panel). These data indicate that either FoxH1 or FoxH1 H83R can form a specific complex with the AR.
Involvement of AF-1 in the Interaction of the AR with FoxH1-To further define the individual domains of the AR involved in binding to FoxH1, mammalian two-hybrid assays were carried out. The AR constructs consisted of amino acids 1-919 (pBIND-AR), amino acids 1-660 (pBIND-AR-NTD/DBD), and amino acids 615-919 (pBIND-AR-LBD) fused to the DBD of GAL4, whereas full-length FoxH1 was fused to the VP16 activation domain (pACT-FoxH1). As shown in Fig.  7B, the pG5-LUC reporter was induced after cotransfection of pBIND-AR-NTD/DBD and the empty pACT vector but not after cotransfection of the pBIND-AR-LBD construct, consistent with previous evidence that AF-1 in the NTD is responsible for most of the AR transactivation (66). Further induction was observed when pACT-FoxH1 was cotransfected with either pBIND-AR or pBIND-AR-NTD/DBD but not pBIND-AR-LBD. The addition of DHT (10 nM) further enhanced the transcription by about 2-fold in the case of pBIND-AR, whereas no significant induction was observed in the case of pBIND-AR-LBD. Moreover, pBIND-AR-NTD/DBD was found to have even stronger induction ability with FoxH1 than the full-length AR. Interestingly, when ANT-1, which specially enhances AR transactivation through a direct interaction with AR-AF-1 (46), was coexpressed, the induction of pG5-LUC by pACT-FoxH1 and pBIND-AR-NTD/DBD was completely blocked, suggesting that ANT-1 competed with FoxH1 for bind-  ing to AR-AF-1. These data suggest that AF-1 may be involved in the interaction of the AR with FoxH1 in both the presence and absence of androgens.
FoxH1 Represses AR-AF-1 Function-Because our data implicated AF-1 as the site of FoxH1 corepressor activity, we investigated whether AF-1-dependent transactivation was sensitive to FoxH1 repression. As shown in Fig. 7C, when pBIND-AR-NTD (amino acids 1-532) was cotransfected together with pG5-LUC, there was a marked increase in the luciferase activity which was dramatically reduced when FoxH1 was coexpressed. These results provide additional evidence that AF-1 may be involved in the interaction between the AR and FoxH1 in cells.

DISCUSSION
To date many coregulators of the AR have been identified and characterized (13,15). Compared with the coactivators, the AR corepressors identified are relatively fewer and less well characterized. The data obtained in the present study demonstrate a new function for FoxH1, which was expressed in LNCaP cells and the other prostate cancer cell lines tested, as a corepressor that attenuates the transactivation poten-tial of the AR. This effect occurred in both the presence or absence of activin A, indicating that activin A was not involved in the repression of the AR by FoxH1 and, conversely, that FoxH1 was dispensable for the stimulatory effect of activin A on PSA expression (26,27,41). However, the possible roles of the FoxH1-mediated repression of the AR in activin signaling in prostate cancer require further investigation since Smad2/4 can rescue the repression of the AR by FoxH1, and a mutually antagonistic interaction between activin and androgen signaling has been shown to be involved in modulating the expressions of cell cycle regulatory proteins such as Rb, E2F-1, and p27 (27), thereby playing an important role in the regulation of prostate cancer cell growth. Hence, further clarification of the effect of the FoxH1-mediated repression of the AR on the expressions of these cell cycle regulatory proteins would help to elucidate the molecular mechanism through which activin A regulates gene expression and growth in the prostate.
The AR shares hormone response element sequences in the DNA with GR and PR (67). In this regard, the NH 2 -terminal region, which varies among these receptors, is considered to be responsible for the cell-and ligand-specific regulation of their target genes (66, 68, 69). The AR is thought to be quite unique among the nuclear receptor superfamily members, since most if not all of its activities are mediated via the ligand-independent constitutive activity of the AF-1 (66). The fundamental role of the AR-AF-1 was further supported by our recent finding (43) that the absence of an AR-AF-1-specific coactivator resulted in androgen insensitivity syndrome. Our current observations show that AF-1 is involved in the interaction of the AR with FoxH1, which may explain why the inhibitory effect of FoxH1 on steroid receptors is selective. Because FoxH1 is expressed in mammary gland tissue (39), it is of interest to further elucidate whether FoxH1 is also able to interact with ER in breast cancer cell lines, such as MCF-7, and whether it is involved in the regulation of TGF-␤/activin signaling during the growth of breast cancer.
In addition to SRC-1 (70) and CBP/P300 (71), which interact with both AF-1 and AF-2 of the AR, many coregulators, such as Hey1 (72), SMRT (60), SMAD3 (57), AES (73), cyclin D1 (74), ANT-1 (46), ARA24 (75), and ARA160 (76), have also been shown to interact preferentially with the AF-1. Interestingly, STAT3, a critical signaling molecule required for ligand-independent transactivation of the AR by interleukin-6, also associates with the AR via AF-1 (53,77,78). Because androgen levels do not fluctuate dramatically in adult males, the relative levels of coactivators versus corepressors binding to the AR may play a critical role in modulating AR function (15,79). Indeed, Hey1 (72) and SMRT (60) have been shown to attenuate AR transcriptional activity through interrupting the interaction between the AR and SRC-1. Similarly, we demonstrated that ANT-1 competed with FoxH1 for binding to the AR. Therefore, interrupting the interactions between the AR and its coacti-vators may be responsible for the inhibitory effects of FoxH1 on the ligand-dependent and -independent transactivation of the AR. In addition, FoxH1, as well as Hey1, could attenuate the AR transactivation potential in both the presence and absence of androgens, suggesting that therapeutic interventions based on AR-coregulator interactions could be designed to block both androgen-dependent and -independent growth of tumor cells.
We previously reported a sensitive confocal laser microscopy approach that can clearly distinguish the transcriptionally active and inactive forms of the AR in vivo (44). More recently, we reported that transfer to common compartments (foci) of the nucleus and complex formation with coactivators, such as SRC-1, TIF2, and CBP, may be essential processes for eliciting the transactivation function of the AR (45). In the present study we observed that FoxH1 had no effect on the nuclear translocation of the AR but specifically blocked the DHT-induced foci formation by the AR in LNCaP cells. Therefore, the simplest explanation for our current finding that FoxH1 represses the AR-mediated transactivation is that FoxH1 maybe compete with AR coactivators, such as SRC-1 and ANT-1, for binding to the AR, thereby inhibiting the formation of the transcriptionally active complex. However, further studies are required to elucidate the precise mechanism of the FoxH1-induced repression of AR-mediated transactivation. Moreover, it will be interesting to investigate whether disruption of the foci formation is also involved in the repression of AR-mediated transactivation by other corepressors, such as Hey1 and SMRT.
In conclusion, this is the first report to demonstrate that association of FoxH1 with the AR can inhibit the transactivation potential of the AR. . Physical interaction between FoxH1 and the AR. A, coimmunoprecipitation of the AR with FoxH1 or FoxH1H83R was performed as described under "Experimental Procedures." Whole cell extracts were immunoprecipitated with an anti-AR antibody (C-19) or normal rabbit IgG (NRI), and the immunoprecipitated fractions were analyzed by immunoblotting with an anti-Myc antibody. B, involvement of AF-1 in the interaction of the AR with FoxH1. Mammalian two-hybrid assays were carried out to test for an in vivo interaction between the AR and FoxH1. NIH3T3 cells were cotransfected with a DNA mixture containing 100 ng of pG5-LUC, 1.5 ng of pRL-SV40, 75 ng of pACT-FoxH1, or an equimolar amount of the pACT parent vector, 25 ng of pBIND-AR or equimolar amounts of pBIND-AR-NTD/DBD, pBIND-AR-LBD or the pBIND parent vector, and 225 ng of pcDNA3-ANT-1 or an equimolar amount of the pcDNA3 parent vector. The total DNA was adjusted to 600 ng/well with pBSKϩ DNA. After transfection the cells were treated with 0.1% ethanol or 10 nM DHT. C, FoxH1 represses AR-AF-1 function. NIH3T3 cells were cotransfected with 100 ng of pG5-LUC, 1.5 ng of pRL-SV40, 50 ng of pCMV-FoxH1, or an equimolar amount of pCMV parent vector and 10 ng of pBIND-AR-NTD or an equimolar amount of the pBIND parent vector. The total DNA was adjusted to 600 ng/well with pBSKϩ DNA. Approximately 24 h after transfection the cells were harvested, and the luciferase activity was measured.
Further studies of the expression of FoxH1 in prostate cancer at different stages and its role in the mutually antagonistic effects of androgen and activin A will provide fresh insights into the biology of prostate cancer and may lead to the development of new treatments.