Molecular Endocrinology, doi:10.1210/me.2003-0020
Molecular Endocrinology 17 (8): 1484-1507
Copyright © 2003 by The Endocrine Society
The Role of Hepatocyte Nuclear Factor-3
(Forkhead Box A1) and Androgen Receptor in Transcriptional Regulation of Prostatic Genes
Nan Gao,
Jianfeng Zhang,
Mira A. Rao,
Thomas C. Case,
Janni Mirosevich,
Yongqing Wang,
Renjie Jin,
Aparna Gupta,
Paul S. Rennie and
Robert J. Matusik
Department of Cell and Developmental Biology (N.G., J.Z., R.J.M.), Vanderbilt University Medical Center; Department of Urologic Surgery (N.G., J.Z., T.C.C., J.M., Y.W., R.J., A.G., R.J.M.), and the Vanderbilt Prostate Cancer Center, Vanderbilt University; Department of Cancer Biology (T.C.C., J.M., Y.W., R.J., A.G., R.J.M.), and the Vanderbilt Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee 37232; and The Prostate Centre at Vancouver General Hospital (M.A.R., P.S.R.), Vancouver, British Columbia V6H 3Z6, Canada
Address all correspondence and requests for reprints to: Robert J. Matusik, Ph.D., Department of Urologic Surgery, Vanderbilt University Medical Center, Nashville, Tennessee 37232. E-mail: Robert.matusik{at}vanderbilt.edu.
 |
ABSTRACT
|
|---|
Androgens and mesenchymal factors are essential extracellular signals for the development as well as the functional activity of the prostate epithelium. Little is known of the intraepithelial determinants that are involved in prostatic differentiation. Here we found that hepatocyte nuclear factor-3
(HNF-3), an endoderm developmental factor, is essential for androgen receptor (AR)-mediated prostatic gene activation. Two HNF-3 cis-regulatory elements were identified in the rat probasin (PB) gene promoter, each immediately adjacent to an androgen response element. Remarkably, similar organization of HNF-3 and AR binding sites was observed in the prostate-specific antigen (PSA) gene core enhancer, suggesting a common functional mechanism. Mutations that disrupt these HNF-3 motifs significantly abolished the maximal androgen induction of PB and PSA activities. Overexpressing a mutant HNF-3 deleted in the C-terminal region inhibited the androgen-induced promoter activity in LNCaP cells where endogenous HNF-3 is expressed. Chromatin immunoprecipitation revealed in vivo that the occupancy of HNF-3 on PSA enhancer can occur in an androgen-depleted condition, and before the recruitment of ligand-bound AR. A physical interaction of HNF-3 and AR was detected through immunoprecipitation and confirmed by glutathione-S-transferase pull-down. This interaction is directly mediated through the DNA-binding domain/hinge region of AR and the forkhead domain of HNF-3. In addition, strong HNF-3 expression, but not HNF-3ß or HNF-3
, is detected in both human and mouse prostatic epithelial cells where markers (PSA and PB) of differentiation are expressed. Taken together, these data support a model in which regulatory cues from the cell lineage and the extracellular environment coordinately establish the prostatic differentiated response.
|
INTRODUCTION
|
|---|
THE PROSTATE IS a male accessory sex organ that is found exclusively in mammals. This organ produces various components found in the semen. Androgens are required to initiate prostatic development, to establish prostatic morphogenesis, and to begin secretory activities. In the mouse, the developing fetal testis begins to produce androgens at embryonic d 12.5–13 (E12.5–13) with peak production at E17–18, during which the earliest signs of prostatic formation are observed (1, 2). Ablation or surgical removal of fetal testes during the ambisexual period inhibits the development of prostate and other male internal sex organs (3). Androgens act directly through androgen receptor (AR) to elicit their effects (4). Testicular feminization mice, which express a nonfunctional AR, never develop a prostate, even though their testes produce adequate amounts of testosterone (3, 5). In addition, prostatic epithelial cells that do not express functional AR, also do not express characteristic secretory markers (6). These observations demonstrated the essential requirement of androgens and AR for prostate development and function. Interestingly, tissue recombination experiments have shown that the urogenital epithelium from testicular feminization mice develops normal prostate gland histology when combined with wild-type urogenital mesenchyme (UGM), although these urogenital epithelium contain a defective AR (7). Also, it has been observed that the induction of prostatic epithelial buds occurs before AR expression in wild-type prostatic epithelial cells (8, 9, 10). These results argue that the epithelial AR is not required for prostatic determination, although it is essential for the production of androgen-dependent secretory proteins at later stages (6).
Extensive tissue recombination studies have proven that mesenchymal-epithelial interactions play a key role in directing prostate development (3, 11). Epithelial cells from other urogenital sinus derivatives (bladder, urethra, and vagina) can be instructively induced by UGM to give rise to prostatic tissue (3, 12, 13, 14, 15). Similar results have been obtained using human bladder epithelial cells (16). The ductal-acinar structures that form in such experiments histologically resemble prostatic epithelium and produce prostate-specific antigen (PSA) (16). Nevertheless, the response of epithelia to inductive mesenchyme is limited by the developmental repertoire of the germ layer origin of the epithelium (17). For example, mesodermally derived seminal vesicle epithelium responds to either urogenital sinus or seminal vesicle mesenchyme by generating seminal vesicle. In contrast, the endodermally derived epithelia of the prostate, bladder, or urethra respond to the same inductive mesenchyme by generating prostatic tissue. Therefore, it is apparent that the cell developmental program is determined by both epithelial and mesenchymal factors. Although androgens and mesenchymal factors are critical for prostatic development, epithelial factors that directly participate in this process are still unknown.
Hepatocyte nuclear factor-3 proteins (newly named as forkhead box A proteins) are a group of endoderm-related developmental factors (HNF-3, HNF-3ß, and HNF-3) that belong to the forkhead box transcription factor family (18, 19, 20). In mice, HNF-3ß is first expressed in the endoderm progenitor during gastrulation (E6.5), immediately followed by the expression of HNF-3 (before organogenesis, E7–8) and HNF-3 (19, 21). The targeted null mutation of HNF-3ß gene results in embryonic lethality due to the absence of endodermal progenitor cells (22), whereas homozygous mutation of HNF-3 gene leads to perinatal death due to pancreatic defects (23). All three HNF-3 genes are restrictively expressed in the endoderm-derived organs in the adult and are involved in endodermal differentiation (19). In vivo footprinting using mouse embryo liver cells showed that HNF-3 factors, among a few of the earliest transcription factors, bind to the enhancer of liver-specific albumin gene before this gene is activated (19, 24). This binding of HNF-3 is believed to provide developmental competence to the target gene, which becomes activated during hepatic induction when additional liver- enriched transcription factors are recruited (19, 25, 26). Although HNF-3 proteins were first discovered in liver (27), the highest HNF-3 mRNA level was detected in the prostate gland by comparison of 16 human tissues (28). A novel expression pattern of HNF-3 in the epithelium of the bladder, urethra, and prostate has also been reported (28, 29). However, in the urogenital system, target genes that are regulated by HNF-3 have not been identified. The role of HNF-3 in the urogenital system remains to be defined.
In the adult prostate, fully differentiated luminal epithelial cells secrete androgen-dependent prostate-specific proteins, such as probasin (PB) in the rat and PSA in the human. Activation of these characteristic genes in the epithelium signifies that prostatic differentiation has occurred. The transcription factors that regulate these genes may also be involved in the process of prostate development. Therefore, a better understanding of the transcriptional regulation of these prostate-specific genes is likely to provide insight into organ development. In the present study, characterization of the PB promoter has led to the identification of two HNF-3 cis-regulatory motifs, which are immediately adjacent to functional AR binding sites (ARBSs). We have observed a similar organization of the androgen response element (ARE) adjacent to HNF-3 sites in the PSA gene enhancer and the prostatic acid phosphatase (PAP) gene promoter/enhancer, suggesting that a common functional mechanism is involved. DNA mutation studies showed that these HNF-3 sites are essential for maximal androgen induction of both PB and PSA gene. Chromatin immunoprecipitation (ChIP) assay confirmed that HNF-3 occupies the PSA enhancer in vivo. In addition, we detected a direct protein-protein interaction between HNF-3 and AR. These data suggest that an endodermal transcription factor HNF-3 and a steriod receptor AR may coordinately participate in the assembly of a nucleoprotein complex, which directs prostatic differentiated function. The regulatory mechanism between HNF-3 and AR may also apply to the transcriptional control of other prostatic epithelial cell-specific genes.
|
RESULTS
|
|---|
Two HNF-3 Binding Sites Are Identified on the PB Promoter
The rat PB is a prostate-specific and androgen-regulated gene, which has been extensively characterized (30, 31, 32, 33, 34, 35, 36, 37, 38). It has been shown that a 12-kb fragment of the PB 5'-flanking sequence, a -426/+28-bp fragment as well as an artificial PB promoter, androgen-response region (ARR)2PB, which contains an extra copy of -244/-96 bp fused in front of the -286/+28-bp region, is able to specifically target genes to transgenic mouse prostate (31, 33, 35, 36, 39, 40, 41). Therefore, the minimal fragment that confers prostate specificity was mapped to the 315-bp region (-286/+28) (36), which contains two functional ARBSs (ARBS-1 at -236/-223 and ARBS-2 at -140/-117) (30). Based on the hypothesis that additional key cis-acting elements remain to be uncovered, an extensive mutation analysis was performed. In this study, we report two critical HNF-3 motifs that are essential for promoter activity.
The first cis-acting element, designated as R1, is located at -253/-237 bp, immediately upstream to ARBS-1. The second element, designated as R2, is located at -126/-107 bp, downstream and slightly overlapping ARBS-2 (Fig. 1A
). To identify putative transcription factor (TF)-binding sites, we employed web-based search engines TESS (www.cbil.upenn.edu/tess) and TFSEARCH (www.cbrc.jp/research/db/TFSEARCH.html). The search was restricted to a maximum of 20% mismatch within an element length of six nucleotides (nt) or greater. Either a 9-nt sequence (-252/-244 bp) located in R1 or a 9-nt sequence (-121/-113 bp) located in R2 highly matched with the HNF-3 consensus binding sequence (5'-TRTTTRYTY-3') (20). The first 7-nt 5'-T(A/G)TT(T/G)(G/A)(T/C)-3', extracted from various known HNF-3-regulated gene promoters (20), perfectly matched with the TATTTGT motif in R2 (Fig. 1A). Figure 1B shows a typical EMSA using nuclear extract prepared from LNCaP human prostate cancer cells, which express HNF-3 (Fig. 1C). The TTRs (Fig. 1B) is a strong HNF-3 binding unit on the liver transthyretin (TTR) gene promoter and was originally used to affinity purify HNF-3 proteins (27, 42, 43). Strong complexes were formed when the TTRs consensus HNF-3 binding site was used (Fig. 1B, lane 2), and these complexes were reduced with the addition of antibodies against HNF-3 or HNF-3ß (lanes 3 and 4). RT-PCR using primers specific for HNF-3ß failed to detect the expression of HNF-3ß in LNCaP cells (data not shown), indicating the effect caused by HNF-3ß (M20) antibody may be due to the cross-reactivity of this antibody because HNF-3 and HNF-3ß have highly homologous C termini. Four complexes, designated as a, b, c, and d, were formed with radiolabeled PB-127/-102, which contains R2 (Fig. 1B, lane7). Three complexes, designated as A, B, and C, were formed with PB-257/-232, which contains R1 (Fig. 1B, lane 20). HNF-3 antibody distinctly abolished the binding of two R2 complexes (lane 8, b and c) and R1 complex B (lane 21), whereas mock antibodies had no visible effect (lanes 10–16 and 23). A 200-fold molar excess of cold TTRs oligonucleotide completely competed off the complexes b and c (lane 17), indicating binding specificity. An increasing competitor (400- to 600-fold) further affected complex a (lanes 18 and 19), suggesting that a higher-order HNF-3 complex might be contained in complex a. Because complex a was not affected by HNF-3 antibody (lane 8), it indicates that this complex may have higher stability and the epitope in this complex may not be accessible to HNF-3 antibody. Western blot in Fig. 1C shows that HNF-3 protein was strongly detected in LNCaP and PC3, but weakly detected in DU145 cells in comparison with the positive control HepG2 cells, and the negative control Hela or Cos-1 cells. The expression of HNF-3 mRNA in these three prostatic cell lines (LNCaP, PC3, and DU145) was confirmed by RT-PCR (data not shown). Similar EMSA results in Fig. 1B were obtained when PC3 nuclear extract was used (data not shown).
View larger version (100K):
[in this window]
[in a new window]
|
Fig. 1. Identification of Two HNF-3 Binding Sites in the PB Promoter
A, Location of HNF-3 motifs. Mutation study revealed two key cis-acting elements R1 and R2 (dashed lines). Each contains a potential HNF-3 binding motif (bold arrows). The direction of the arrow indicates the sense or the antisense strand that matched the consensus sequence. Two adjacent ARBSs (ARBS-1 and ARBS-2) are underlined. Sequences in italic indicate the nucleotide replacements in mutation assays. B, EMSA. LNCaP nuclear extracts were incubated with radiolabeled TTRs (consensus HNF-3 binding sequence), PB-127/-102, or PB-257/-232 probes. Strong complexes formed with TTRs (lane 2). HNF-3 or HNF-3ß antibody reduced the corresponding band (lanes 3 and 4). Four complexes, a, b, c and d, were formed with R2 (lane 7). Complexes b and c were specifically removed by HNF-3 or HNF-3ß antibody (lanes 8 and 9), but not other mock antibodies (lanes 10–16). molar excess (200-fold) of cold TTRs oligonucleotides completely competed off complexes b and c (lane 17), whereas 400- and 600-fold molar excess of TTRs also affected complex a (lanes 18 and 19). Three complexes, A, B, and C, were formed with R1 (lane 18). Complex B was specifically eliminated by HNF-3 antibody (lane 19). C, Western blot. Whole-cell lysates (20 µg) from HepG2, Hela, Cos-1, LNCaP, PC3, and DU145 cells were resolved on SDS-PAGE and probed with HNF-3 antibody. D, A radiolabled oligonucleotide (ARBS2/R2, Table 1 ) containing both ARBS-2 and R2 was used in EMSA with 10 µg LNCaP nuclear extract. With the presence of an ARBS, two HNF-3 complexes (arrows) still formed with R2, because both complexes were reduced or removed by HNF-3 antibody (lane 1 vs. 2), or eliminated by mutating the HNF-3 motif in ARBS2/R2-M (lane 7, Table 1). Both complexes were not affected by 100-, 200-, and 500-fold molar excess of cold ARBS-2 site (lanes 4–6), whereas several
|
|
View larger version (58K):
[in this window]
[in a new window]
|
Fig. 1A. Continued
other complexes (asterisk) were sensitive to the competition. E, Southwestern blot. Nuclear extracts from LNCaP and PC-3 cells were separated on SDS-PAGE, transblotted to nitrocellulose membranes, and probed with radiolabeled 2xR2 (lanes 1 and 3), 2xmR2 (lanes 2 and 4) or 2xAR-con probes (lane 5). A Western blot using HNF-3 antibody was performed as a parallel control. Arrows indicate migration of proteins bound to the respective probes or antibody.
|
|
Because ARBS-2 slightly overlaps with the HNF-3 motif in R2, it is necessary to determine 1) whether HNF-3 still binds R2 when the ARBS-2 site is occupied by AR, 2) whether the HNF-3/R2 interaction is dependent on adjacent AR binding. To address these questions in vitro, we used a radiolabeled probe (Table 1, ARBS2/R2) containing both ARBS-2 and R2. Figure 1D shows that, among the multiple complexes formed with LNCaP nuclear extract, two complexes were specifically reduced or removed by HNF-3 antibody (lane 1 vs. 2), whereas both complexes were disrupted by mutating the HNF-3 motif in ARBS2/R2 string (lane 3 vs. 7 and Table 1, ARBS2/R2-M1). Competition assay, using 100-, 200-, and 500-fold molar excess of cold ARBS-2 site (lane 3 vs. 4–6), showed that several complexes (indicated as asterisks) were sensitive to the competitor, whereas none of the HNF-3 complexes were competed off. These results demonstrated in vitro that HNF-3/R2 binding can occur concomitantly with AR/ARBS-2 interaction, but HNF-3 binding is independent of AR/ARBS-2 interaction.
In Fig. 1E, Southwestern blot was used to determine the relative molecular masses of R2-interacting proteins. LNCaP and PC-3 cell nuclear extracts were resolved on sodium dodecyl sulfate (SDS)-polyacrylamide gels and transferred to nitrocellulose membranes. The blots were probed with 32P-labeled oligonucleotides containing two tandem copies of wild-type R2 (2xR2). For both LNCaP and PC-3 cells, two proteins with molecular masses of approximately 52 kDa and 46 kDa were detected (lanes 1 and 3). Similar blots probed with 2xmR2 (containing two copies of mutant R2, see Table 1 for sequence) showed that the binding of both proteins was distinctly reduced in LNCaP cells or completely blocked in PC-3 cells (lanes 2 and 4), indicating that the binding is sequence specific. A radiolabeled oligonucleotide containing two tandem copies of AR consensus binding site (Table 1), serving as a positive control, was hybridized specifically to a 110-kDa protein (the size of human AR), but not the low molecular mass proteins (lane 5). A parallel Western blot using HNF-3 antibody showed a specific band around 52 kDa (lanes 6 and 7), suggesting that the 52-kDa protein detected by Southwestern blot was HNF-3. The 46-kDa protein detected in both LNCaP and PC3 might be an HNF-3 minor degradation product that escaped antibody detection but still retained DNA binding activity. However, the presence of a higher-order protein complex on R2 (complex a in Fig. 1B, lane 7) suggests that additional DNA-binding protein on R2 may exist.
In Vitro Synthesized HNF-3 Binds R1 and R2
Because crude nuclear extracts were used in EMSA and Southwestern experiments, it was necessary to confirm the results with HNF-3 synthesized by in vitro transcription/translation (TNT). Figure 2 shows that synthesized HNF-3 specifically bound to the radiolabeled R2 in EMSA (Fig. 2A, lane 3), in contrast to the TNT blank control (lane 2), which only formed a nonspecific band. Three mutant HNF-3 proteins deleted in the N-terminal (
N), C-terminal (C), or both terminal regions [FH (forkhead)] still showed binding activities (lanes 4–6), because all mutants still contain the FH DNA-binding domain (DBD). The binding of HNF-3 was completely removed by an antibody against the HNF-3 C terminus (Fig. 2B, lane 2 vs. 3). The same antibody could not eliminate the complex formed by the mutant HNF-3 deleted in the C terminus (lane 5 vs. 6). In addition, a mutant probe (mPB-127/-102 bp) containing a mutated HNF-3 motif also affected the specific binding (lane 2 vs. 4). Similar results were observed when using R1 (PB-257/-232 bp) sequence. As shown in Fig. 2C, HNF-3/R1 interaction is relatively weaker in contrast to HNF-3/R2 (Fig. 2C, lane 2 vs. 4), because R1 has a 1-bp mismatch with HNF-3 cognate binding site, whereas R2 is a perfect match. The presence of multiple HNF-3 sites with different affinities has been proposed to be relevant for transcriptional modulation (44). These in vitro experiments confirmed the identification of two HNF-3 cis-regulatory elements adjacent to ARBSs in the PB proximal promoter.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 2. In Vitro Synthesized HNF-3 Binds R1 and R2
A, Wild-type HNF-3 and three mutant proteins (N, FH, and C) were synthesized by in vitro TNT, and incubated with radiolabeled PB-127/-102 probe. In contrast to a nonspecific band (arrowhead) formed with the TNT blank control (lane 2), wild- type HNF-3 and HNF-3C formed a strong complex (lanes 3 and 6), whereas N and FH (arrowhead) showed weak binding ability (lanes 4 and 5, overexposure). B, The binding of HNF-3 with PB-127/-102 was eliminated by an antibody raised against the HNF-3 C terminus (lane 3 vs. 2). The same antibody did not affect the complex formed by HNF-3C, which has a deletion in the C terminus (lane 6 vs. 5). The HNF-3 binding was also affected by mutating the TRTTTGY motif (lane 4 vs. 2). C, Similar results but weaker binding affinity was observed when PB-257/-232 was used as a probe (lane 4 vs. 2).
|
|
HNF-3 Is Expressed in Prostate Epithelial Cells
Two previous studies have reported the mRNA expression profile of HNF-3 in the mouse and rat prostates by using in situ hybridization and Northern blot (28, 29). We examined the HNF-3 protein expression pattern in the human and mouse prostate by immunohistochemical and Western blot analyses. In Fig. 3, immunohistochemistry showed a clear nuclear staining of HNF-3 in both mouse (panels A and B) and human (panels C and D) prostate luminal epithelial cells. HNF-3ß and HNF-3 were not detected (data not shown). Consistently, Western blot showed that HNF-3 is expressed in all mouse prostate lobes, with highest levels of expression in the ventral prostate (VP). This pattern is consistent with a previous study that showed that HNF-3 mRNA levels in the VP are 14 times more abundant than in the liver (28). Results from RT-PCR using primers specific for HNF-3, HNF-3ß, or HNF-3 are consistent with immunohistochemistry and Western blot (data not shown). The absence of HNF-3ß expression in prostate has been reported before (29), and this is different from the expression patterns in other organs where HNF-3 and HNF-3ß are often coexpressed (42).
HNF-3 Binding Sites Are Essential for Maximum Androgenic Induction
The mechanism of HNF-3 transcriptional regulation, in some part, involves nucleosome disruption (24, 45). Although transiently transfected DNA cannot assemble the same higher-order chromatin structure as genomic DNA, certain levels of nucleosome-mediated regulation can be observed on transiently introduced DNA (46, 47). Studies using transient transfection have demonstrated that HNF-3 relieves nucleosome-mediated transcriptional repression of a liver specific -Fetoprotein gene (47). To examine the biological role of the two HNF-3 binding motifs, we compared the activities of mutant PB promoters with wild-type promoter in transfection assays. PCR-based site-directed mutagenesis was used to generate point mutations in HNF-3 binding sites (R1 and R2). As seen in Fig. 4A, three mutant promoters were generated. Two of them, M1 and M2, contain mutant HNF-3 motif in R2. The third one, M3, contains mutant HNF-3 motif in R1. EMSA in Fig. 4B demonstrated that M1 completely abolished HNF-3 complexes a, b, and c. M2 slightly affected complexes b and c, but distinctly removed complex a. Addition of HNF-3 antibody further removed complex b and c, indicating that M2 affects the formation of a higher-order complex, which was competed by the HNF-3 consensus sequence (Fig. 1B, lanes 18 and 19). All mutant, as well as a wild-type PB reporter, constructs were transfected into LNCaP cells and luciferase activities were compared. Experiments were repeated at least six times. Figure 4C is a representative experiment showing that the maximal androgen induction of PB promoter activity was significantly affected in either mutant R1 or mutant R2. In contrast to R1 motif, R2 may have a greater biological relevance because both M1 and M2 mutations almost completely abolished PB activity (P < 0.001). The significant decrease in M2 activity suggests that complex a (Fig. 4B), which is in a higher order, is biologically important. The reduction in M3 activity is consistent with a previous study (48) conducted by another group, who demonstrated a significant loss in PB activity by deleting a region overlapping with R1. Similar results were obtained when these transfection experiments were performed in PC3 cells (data not shown). In addition, EMSA in Fig. 4D demonstrated in vitro that a purified GST-AR-DBD fusion protein bound to these mutant ARBS-1/R1 or ARBS-2/R2 strings (Table 1 for sequences) with similar intensity as to the wild-type sequence, indicating that these mutations generated in HNF-3 motifs do not directly abrogate the AR/ARBS interaction. Thus, the activity loss in PB promoter (Fig. 4C) is not directly due to the loss of AR/ARBS binding. However, such mutations may destabilize the formation of a complete nucleoprotein complex in these regions, and this will be further discussed in this paper. These data strongly suggested that two HNF-3 motifs are required for the maximal androgenic induction in PB gene.
View larger version (59K):
[in this window]
[in a new window]
|
Fig. 4. Two HNF-3 Motifs Are Essential for Maximum Androgenic Induction in PB
A, Wild-type PB promoter sequences containing boxed HNF-3 motifs were mutated (underlined lowercase characters) as indicated in M1–M3. B, EMSA was conducted using 10 µg LNCaP cell extracts incubated with either wild-type PB-127/-102 or mutant R2 probes (lanes 1–3). M1 disrupted the binding of three complexes (lane 2, a, b, and c), whereas only the higher-order complex a was eliminated by M2 (lane 3). Addition of HNF-3 antibody further removed complexes b and c (lane 4). C, LNCaP cells were transiently transfected with luciferase reporter constructs (0.2 µg/well) containing either wild-type or mutant (M1–M3) PB promoters. Each well also received 0.0125 µg of Renilla vector. Cells were treated with or without 10-8 M DHT for 24 h before harvest. Background activities of cell lysates with no DNA transfection were subtracted from the data obtained in the experimental group, before the normalization with Renilla activities. Results are presented as relative luciferase activity. Data shown here are a representative from at least six independent experiments in triplicate. Error bars indicate SD values. Asterisk indicates where P < 0.01 as compared with the androgen-induced activity in wild-type (WT) reporter. D, Probes (ARBS2/R2-M1, -M2, and ARBS1/R1-M3, Table 1) containing mutant HNF-3 motifs were used in EMSA to determine their interactions with a purified GST-AR-DBD protein described in Materials and Methods.
|
|
Overexpressing a Mutant HNF-3 Inhibits PB Activity
It has been suggested that both the N- and C-terminal regions of HNF-3 proteins have transactivation activities (49, 50). Recent studies using in vitro nucleosome assembly showed that the transcriptional modulation by HNF-3 is mediated through its C-terminal region (24). We generated two mutant recombinant HNF-3 proteins (N141-466, C1-294) with deletions in either the N- or C-terminal regions. These mutant HNF-3 proteins, as well as a wild-type HNF-3 expression vector (0.2 µg/well), were cotransfected into LNCaP cells with the ARR2PB-Luc reporter construct. Luciferase activities were measured and compared. As shown in Fig. 5A, overexpression of wild-type HNF-3 did not significantly increase the androgen induction of PB activity, indicating that the endogenous HNF-3 level is sufficient for the maximal promoter response. In contrast to wild-type and N141-466 proteins, overexpressing C1-294 significantly inhibited the total PB activity (Fig. 5A, P < 0.01), and this inhibition was dose responsive (Fig. 5B). In contrast, transfection of same amount of C1-294 did not significantly inhibit the activity of the mutant PB promoter (Fig. 5C), which contains a nonfunctional HNF-3 site (M1 in Fig. 4C). Finally, Western blot showed that transient expression of these HNF-3 proteins in LNCaP cells did not result in a detectable change in the endogenous AR protein levels (Fig. 5D), suggesting that the reduction in PB promoter activity was not due to reduced AR levels. These results strongly indicated a dominant negative mechanism for C1-294, and further supported the regulatory role of HNF-3 C-terminal region in transcriptional modulation.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 5. Overexpressing a Mutant HNF-3 Inhibits PB Activity
A, An empty vector as well as expression vectors corresponding to wild-type HNF-3, N141-466, or C1-294 (0.2 µg per well) were cotransfected with a PB luciferase reporter into LNCaP cells. Cells were treated with or without 10-8 M R1881 for 24 h followed by luciferase assays. B, Different amounts of C1-294 (0.05 µg, 0.15 µg, or 0.25 µg per well) were cotransfected into LNCaP cells with PB reporter. Luciferase activities were determined after 24 h incubation in the presence or absence of androgen. C, In contrast to panel B, C1-294 (0.05 µg, 0.15 µg, or 0.25 µg per well) was cotransfected into LNCaP cells with a mutant PB reporter (M1, in Fig. 4). Error bars indicate SD values. Asterisk indicates P < 0.01 in comparison with androgen-induced activities in control wells, which were transfected with empty vector. D, No significant change in AR protein level was detected by Western blot in cells transiently transfected with wild-type HNF-3, N141-466, or C1-294.
|
|
Identification of Two Functional HNF-3 Binding Sites in PSA Core Enhancer
The results obtained for rat PB gene prompted us to examine whether HNF-3 is also required for the regulation of human PSA gene. It has been reported that the core enhancer region (-4.2/-3.8 kb) of the PSA gene, but not the proximal promoter, is essential and sufficient for androgen regulation and prostate specificity (51, 52, 53, 54). Six androgen response elements (AREs) are located in this region (55). Among these AREs, ARE III (at -4154/-4132 bp) shows the highest AR affinity and biological activity, because mutations in ARE III significantly abolished PSA enhancer activity (55). Interestingly, we identified two strong HNF-3 binding motifs in this enhancer region (Fig. 6A). The first site, designated as PSA1, is located at -4122/-4109 bp, immediately downstream of the ARE III. The second site, designated as PSA2, is located at -4028/-4005 bp, between ARE IIIA and ARE IIIB (Fig. 6A). In vitro synthesized HNF-3 bound to both elements (Fig. 6B, lane 2 vs. 3 and 6E, lane 3 vs. 4). Similar to the results for PB, mutant HNF-3 proteins (N, C, and FH) containing the FH domain also showed binding activity (Fig. 6B, lanes 4–6). In addition, the binding of HNF-3 was supershifted by HNF-3 antibody (lane 3 vs. 7). Therefore, both elements were confirmed to be authentic HNF-3 sites by these in vitro experiments.
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 6. Identification of HNF-3 Binding Motifs in the PSA Core Enhancer
A, Two HNF-3 binding sequences (bold arrow) were identified in the PSA core enhancer (-4.2/-3.9 kb) region. The first element (-4122/-4109), designated as PSA1, is adjacent to ARE III (underlined). The second one (-4028/-4005), designated as PSA2, is located in the middle of ARE IIIA and ARE IIIB (underlined). B, EMSA. In vitro synthesized wild-type HNF-3 and mutant proteins (N, FH, and C) were able to bind PSA1 (lanes 3–6) as compared with the TNT blank control (lane 2). The HNF-3 binding was supershifted by the HNF-3 antibody (lane 3 vs. 7). C, Concomitant DNA binding of AR and HNF-3. Radiolabeled oligonucleotide containing both ARE III and PSA1 was incubated with a constant amount (0.1 µg) of a purified GST AR-DBD protein. A slow migrating ternary complex (AR/HNF-3/DNA) (lanes 5–8) was formed as addition of increasing amounts (1, 2, 3, and 4 µl) of in vitro synthesized HNF-3, in contrast to HNF-3 alone (lane 3) or AR alone (lane 4), indicating that HNF-3 and AR can occupy DNA concomitantly. The ternary complex was disrupted by addition of HNF-3 antibody (lane 9). D, EMSA using LNCaP nuclear extract shows the formation of HNF-3/PSA1 complexes (lane 2), which were disrupted by HNF-3 antibody (lane 3). The probe PSA1 (see Table 1 for sequence) contains binding sites for both AR and HNF-3. E, In vitro synthesized wild-type HNF-3 also interacts with PSA2 (lane 4).
|
|
Because PSA1 is closely adjacent to ARE III, it was intriguing to ask whether HNF-3 and AR can bind to DNA concomitantly. Thus, an oligonucleotide (Table 1) containing both ARE III and PSA1 was radiolabeled and incubated with constant amounts of a purified GST-AR-DBD fusion protein. A specific AR-DBD/DNA complex was formed (Fig. 6C, lane 4). As increasing amounts of HNF-3 proteins were added to the reaction, a specific HNF-3/DNA complex as well as a slow-migrating band (AR/HNF-3/DNA ternary complex) became stronger (lanes 5–8). Addition of HNF-3 antibody disrupted the complex (lane 9). Similarly, in EMSA using LNCaP nuclear extract, HNF-3 antibody disrupted a strong complex that formed with this oligonucleotide (Table 1, PSA1) containing both HNF-3 and AR binding sites (Fig. 6D, lane 2 vs. 3). These data indicate that AR and HNF-3 can bind to respective DNA binding sites concomitantly. Thus, the binding of one protein does not negatively affect the adjacent binding of another protein. However, the dislocation of HNF-3 from the target binding site (Fig. 6C, lane 9) may destabilize a higher-order protein complex associated with this entire regulatory region, which may include AR.
A similar mutation assay was performed to determine the biological relevance of these HNF-3 motifs in PSA enhancer. Figure 7A shows two mutant PSA reporter constructs, mPSA1-EP and mPSA2-EP, which were generated by PCR and transfected into LNCaP cells. Primers used for PCR are shown in Table 1. Luciferase activities were measured and compared with wild-type construct. Figure 7B shows that point mutations that abolish HNF-3 binding in either PSA1 (Fig. 6C, lane 2) or PSA2 (lane 4) significantly affected the maximal androgen-induced PSA activities (Fig. 7B, P < 0.01). A 95% reduction of activity in the PSA enhancer has been reported previously by deleting the region corresponding to PSA1 (53). Similarly, the direct interaction between AR and ARE III remained intact when mutant oligonucleotide (Table 1, mPSA1), containing both ARE III and mutant HNF-3 motif, was used in EMSA (Fig. 7D), indicating that the loss of PSA activity is not directly due to the loss of AR/ARE III binding. These results suggest that both HNF-3 motifs are essential for maximal PSA induction by androgen.
HNF-3 Binds PSA Enhancer in Vivo
Because LNCaP cells express endogenous PSA gene, ChIP was performed to investigate the in vivo association of HNF-3 with the PSA enhancer. Figure 8A is a schematic diagram of the PSA regulatory region. Five previously described DNA fragments (56) corresponding to the distal region, ARE III, middle region, ARE II, and ARE (2) were tested in the experiment. The ARE III region (-4170/-3978 bp) is a 192-bp fragment that covers both HNF-3 motifs we identified in this study (Fig. 8A). To determine the occupancy of HNF-3 on active vs. inactive PSA chromatin status, LNCaP cells were initially grown in RPMI medium 1640 supplemented with 5% charcoal/dextran-treated fetal bovine serum. After 3 d of cultivation, cells were either treated with 10-8 M dihydrotestosterone (DHT) or maintained in androgen-depleted medium. Soluble chromatin was prepared after formaldehyde treatment of cells. Specific antibodies against HNF-3 or AR were used to immunoprecipitate antigen-bound genomic DNA fragments. The DNAs were amplified by PCR using specific primers (56) spanning the tested regions. After 48 h of DHT treatment, PSA expression was induced as compared with the undetectable level in untreated cells (Fig. 8B). In contrast, the expression levels of HNF-3 protein were almost identical in androgen-treated and untreated cells. In agreement with a previous study (56), DHT induced the recruitment of AR onto multiple AREs, but not control regions (Fig. 8D). In contrast to AR, HNF-3 constantly occupies the ARE III region independently of DHT treatment (Fig. 8D). These ChIP assays provided in vivo evidence for the association of HNF-3 with PSA enhancer and also indicated that the occupancy of HNF-3 alone does not result in the transactivation of PSA gene.
View larger version (73K):
[in this window]
[in a new window]
|
Fig. 8. HNF-3 Occupies PSA Enhancer in Vivo
A, Schematic diagram of PSA gene 5'-upstream region. Six DNA fragments corresponding to the distal, ARE III, middle, ARE II, and ARE (2 ) regions were amplified in ChIP assays using primers described previously (56 ). B, LNCaP cells were initially grown in RPMI 1640 with 5% charcoal/dextran-treated fetal bovine serum for 3 d. Cells were then incubated for another 48 h with or without 10-8 M DHT treatment. Western blot was performed to determine the expression levels of PSA and HNF-3. C, Formaldehyde-cross-linked chromatins were sheared by sonication into an average size of 0.5–1.0 kb. D, After an overnight IP with AR or HNF-3 antibodies, formaldehyde cross-linking was reversed, and DNA fragments were extracted, followed by PCR amplification (Materials and Methods).
|
|
HNF-3 Interacts with AR
Steroid receptors play a pivotal role in transcriptional control of numerous genes. In addition to functioning through binding specific steroid response DNA elements, steroid receptors interact with many coregulators as well as other DNA-binding transcription factors. It has been reported that glucocorticoid receptor (GR)-mediated activation of multiple liver-specific genes requires HNF-3ß (57, 58, 59, 60, 61, 62). Recent studies have shown that estrogen receptor (ER) interacts with three FH transcription factors, FKHR, FKHRL1, and AFX (63, 64). AR is a member of the steroid receptor family and shares many similarities with GR and ER in terms of structure and mechanism of function. The observation of close proximity of AR and HNF-3 binding sites prompted us to examine whether these two proteins can interact with each other. For this purpose, IP was performed in an AR-Hela cell line (55), in which a flag-tagged full-length AR was stably integrated. Because Hela cells do not express HNF-3 (Fig. 1C), we transfected an HNF-3 mammalian expression vector (pcDNA3.1D/V5-HNF-3) and a control LacZ expression vector (pcDNA3.1D/V5-LacZ) into these cells. Western blot using anti-V5 confirmed the expression of V5-tagged LacZ and HNF-3 proteins in transfected cells (Fig. 9A, lanes 3 and 4 vs. 1 and 2). Cells were grown in medium with 10-8 M DHT. All cell lysates were immunoprecipitated with an anti-flag M2 affinity gel in the presence of ethidium bromide (EB, 0–100 µg/ml), which was used to disrupt DNA-protein interactions (65). The gel was washed before Western blot analysis. As seen in Fig. 9B, flag-AR was detected in all AR-Hela precipitates (lanes 6–10) in contrast to normal Hela precipitate (lane 5). HNF-3 (lanes 14–16) but not LacZ (lane 13) was detected in the AR-Hela precipitates, indicating that flag-AR specifically immunoprecipitates with HNF-3. It is significant that this interaction was resistant to the presence of EB (lanes 15 and 16), ruling out the possibility that it was mediated through DNA. Next, IP was performed in LNCaP cells to examine this interaction in a physical context in which AR and HNF-3 proteins express at endogenous levels. LNCaP cells were either grown in androgen-depleted medium for at least 3 d or grown in the presence of 10-8 M DHT. Cell lysates (1 mg per IP reaction) were immunoprecipitated with anti-AR- conjugated protein G-Sepharose beads in the presence of EB. Similar IPs were performed using HNF-3 antibody instead of anti-AR. Each reaction was performed in the presence of 1% NP-40 and 1 mg BSA to quench nonspecific binding. Figure 9B unambiguously demonstrated a physical interaction of AR and HNF-3 in DHT-treated cells in contrast to the androgen- depleted condition. This interaction was resistant to the presence of EB at a concentration of 100 µg/ml, demonstrating a protein-protein interaction (Fig. 9B). These IP experiments strongly suggested that AR and HNF-3 might be involved in the assembly of a multinucleoprotein complex under a physical context.
View larger version (41K):
[in this window]
[in a new window]
|
Fig. 9A. Continued
C-terminal domains designated as region 2 and 3 are in striped boxes. D, Five purified GST-AR fusion proteins (20 µg for each) were bound to glutathione agarose beads, followed by incubation with V5-labled HNF-3 protein synthesized in vitro. Western blot was performed to determine the HNF-3-interacting domain in AR. E, The GST AR fusion proteins DBD/Hinge (50 ng), NT/DBD (250 ng), and DBD/LBD (150 ng) were used in EMSA to determine their DNA binding activities with ARBS-1 on a molar basis. F, The expression of eight V5-labeled HNF-3 subdomains as well as a LacZ protein was confirmed in Western blot (lanes 1–9). These proteins were in vitro synthesized and incubated with the GST-bound AR NT/DBD, DBD/Hinge, and DBD/LBD to determine the AR-interacting region in HNF-3. G, The interaction of AR-DBD/Hinge and HNF-3 FH domain was resistant to the presence of 25–150 µg/ml of EB. LBD, Ligand-binding domain; NT, N-terminal domain.
|
|
Responsible Domains for AR/HNF-3 Interaction
An in vitro GST pull-down assay was performed to confirm AR/HNF-3 interaction as well as to determine the interacting regions. A full-length HNF-3 protein labeled with a C-terminal V5 epitope was synthesized in vitro. Five GST-AR fusion proteins containing different AR subdomains (Fig. 9C) were purified as described previously (66). Figure 9D is a GST pull-down experiment showing that the AR DBD/hinge region (524–649 amino acids) alone is sufficient to mediate the interaction with HNF-3. Experiments were repeated at least five times. In our experiments, the ARNT/DBD showed a weaker interaction with HNF-3 as compared with ARDBD/Hinge or ARDBD/LBD (Fig. 9D), suggesting that the AR N terminus might have a negative effect on the interaction. Although the biological relevance of this effect is currently unknown, this effect was obviously not due to the diminished activity of ARNT/DBD protein, because it bound to the ARBS-1 sequence (on PB promoter) to a similar extent as ARDBD/Hinge or ARDBD/LBD on a molar basis (Fig. 9E). In Fig. 9, C and F, in vitro synthesized HNF-3 fragments were used to map the AR-interacting region(s) in the HNF-3 protein. Most interestingly, the FH domain (amino acids 141–294) alone was sufficient to mediate a strong interaction with AR (Fig. 9F, lanes 13, 22, and 31), whereas the N- or C-terminal domain alone did not bind with AR. Figure 9G demonstrated that the interaction between ARDBD/Hinge and HNF-3 FH domain was insensitive to the presence of EB (0–150 µg/ml), supporting a direct protein-protein interaction. The ligand effect on this interaction was also examined (data not shown) and the results suggested that, under these in vitro conditions, the interaction between ARDBD/LBD and HNF-3 could occur without ligand. This was not surprising because in vitro purified GST-ARDBD/LBD fragments may not act identically as in vivo wild-type AR, in terms of protein folding. Although these in vitro binding experiments cannot exactly reflect the physical conditions, they identified regions responsible for AR/HNF-3 interaction.
|
DISCUSSION
|
|---|
In this study, we report that an extracellular signal mediator, AR, and a tissue-restricted FH protein, HNF-3, coordinately participate in prostatic gene regulation. HNF-3/FH proteins are a group of transcription factors that share remarkable sequence similarity over their DBD, termed a FH or winged-helix domain (19, 20, 67, 68). The HNF-3/FH proteins are involved in the differentiation of endoderm-derived tissues. This feature has been strikingly conserved among the metazoans (19). The Drosophila homeotic protein, FH, which is equivalent to HNF-3, was found to be essential for the development of the terminal region including the foregut and hindgut (67); the PHA-4 protein, which is a Caenorhabditis elegans HNF-3 equivalent protein, is also required for the formation of terminal gut structures in the worm (44). In mammals, HNF-3 proteins (HNF-3, HNF-3ß, and HNF-3) were originally discovered as liver-enriched factors, because of their ability to bind to the TTR gene promoter (27, 42). In early embryo development, HNF-3 proteins are believed to provide developmental competence to numerous liver-specific genes through their high-affinity DNA elements. This mechanism is essential for the activation of these target genes in later stages (19, 26). The role that HNF-3 and some other developmental factors play in organ differentiation has been proposed as genetic potentiation (19, 26, 69).
We observed strong nuclear staining of HNF-3 in both human and mouse prostate epithelial cells (Fig. 3, C–F). The prostate epithelium, like the epithelium of other urogenital sinus derivatives (bladder and urethra), is derived from the endodermal layer of the embryo. In contrast, Wolffian duct derivatives (epididymis, ductus, deferens, and seminal vesicle) are derived from the mesoderm (17, 70). Understanding the role of HNF-3 in the prostate may provide an insight into the differences in epithelial patterning and developmental programs among prostate and Wolffian-derived organs, which are all androgen dependent (3, 70). The cell type-dependent expression of HNF-3 and other endodermal factors could easily provide an explanation why prostatic differentiation markers such as PSA can be instructively induced in bladder epithelial cells under proper extracellular signals (16), whereas such differentiation, under the same signals, cannot occur in a mesoderm-derived context (3).
Using in vitro gel shift and mutation assays, functional HNF-3 sites were identified in both PSA and PB gene-regulatory regions (Figs. 1, 2, and 6). ChIP assay unambiguously demonstrated that HNF-3 binds to PSA enhancer in vivo (Fig. 8). In our experiments, HNF-3 binding sites were further identified in other prostate-specific gene-regulatory regions. As shown in Fig. 10A, the proximal promoter of the rat prostatic acid phosphatase gene (rPAP) contains an HNF-3 binding sequence at -113/-103 bp. This sequence, as well as an upstream ARE, is well conserved in its human homolog hPAP gene promoter (71). In the hPAP upstream enhancer (-1258/-779 bp), a G/T-rich region (-1183/-1151 bp) contains two HNF-3 binding sites with a putative ARE in the middle. This region also exists in the rPAP enhancer (Fig. 10A), except that only one HNF-3 binding site was found. Importantly, the hPAP enhancer (-1258/-779 bp) is related to the cell-specific expression of hPAP gene (48). In addition, two HNF-3 motifs (at -184/-173 bp and -141/126 bp, Fig. 10A) were found in the rat prostatic steroid-binding protein C1 (PBP C1) gene promoter (72). We confirmed all of these HNF-3 sites experimentally by EMSA (data not shown). The sequences used in EMSA are shown in Table 1. Figure 10B is a schematic diagram showing a striking similarity in the organization of binding motifs for HNF-3 and AR in these highly prostate-specific genes, which suggests that a common functional mechanism may apply.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 10. Similar Organization of HNF-3 and AR Binding Motifs in Prostate Gene-Regulatory Regions
A, A summary of HNF-3 binding motifs identified in this study. B, Schematic diagram showing remarkable similarity in the organization of AR and HNF-3 binding motifs in different prostate gene regulatory regions.
|
|
To develop a complex eukaryote requires differential expression of more than 30,000 genes/proteins in precise spatial and temporal patterns (73). An organism uses transcriptional synergy to achieve such diversity, to maintain cell specificity, and to respond to the environment dynamically (73). A combinatory use of a subgroup of general transcription factors, signal-dependent factors, and cell-specific factors could lead to large number of regulatory possibilities, ensuring that every gene is precisely controlled (74). In the prostate, AR is a signal-dependent nuclear receptor mediating extracellular signals, whereas HNF-3 is a cell type-limited genetic potentiator (19). A synergy between these classes of factors has been suggested in other studies (74), and HNF-3 plus AR may be involved in prostatic differentiation. In an in vitro DNA footprinting study, a prostate-specific protein was reported to bind a DNA sequence in PB promoter (48). Interestingly, this region overlaps the R1 site we reported in Fig. 1A. Similarly, previous gel shift experiments have discovered several cell-specific factor-binding elements in the PSA core enhancer (51). Among these elements, two of them completely cover the two HNF-3 motifs shown in Fig. 6A (PSA1 and PSA2), whereas others overlap with binding sites for GATA proteins (75), which belong to another endodermal transcription factor family (19). Consistently, the PSA1 site in Fig. 6A, which is adjacent to ARE III, was protected by a cell-specific protein in a previous DNA footprinting study (53). Interestingly, the PB promoter (-286/+28 bp), the PSA enhancer (-4.2/-3.8 kb), and the hPAP enhancer (-1278/-779 bp), where HNF-3 binding sites are located, have all been implicated in prostate-specific regulation (36, 51, 54, 76). These observations strongly suggest that HNF-3 and AR may coordinately participate in the transcriptional regulation of other prostate epithelial cell-specific genes.
X-ray crystallography revealed that the structure of the HNF-3 FH domain resembles the globular domain of linker histone H1 (77). In vitro nucleosome assembly experiments further showed that HNF-3 is able to open compacted chromatin by displacing linker histone in an ATP-independent manner (24, 45). This mechanism of HNF-3 action is mediated through its high-affinity DNA-binding sites as well as the binding of its C terminus to histones H3 and H4 (24). We used transient transfection experiments to examine the role of HNF-3 in prostatic gene regulation. Although transiently transfected DNA does not appear to assemble the same higher-order chromatin structure observed with genomic DNA, certain aspects of nucleosome-mediated regulation can be observed on transiently expressed DNA (46, 47). This has been shown by the studies of Crowe et al. (47), who demonstrated that HNF-3 relieves nucleosome-mediated transcription repression of a liver-specific gene in transient transfection experiments. In our study, mutations of these HNF-3 binding motifs significantly abolished PB and PSA gene activities (Figs. 4 and 7). Overexpression of a mutant HNF-3 protein (C1-294) suppressed PB promoter activity in a dose-responsive and binding site-dependent manner (Fig. 5). These results are consistent with the assumption that some level of nucleosome assembly occurs on the introduced plasmid. Also, these data are in agreement with previous studies showing that overproduction of mutant HNF-3 proteins deleted in a similar region suppressed multiple liver gene expressions (78, 79). However, in contrast to AR, which binds to ARE in a ligand-dependent manner, HNF-3 occupies the PSA enhancer in the absence of androgen (Fig. 8). The DNA binding of HNF-3 does not seem to be directly associated with PSA activation, which eventually requires the recruitment of ligand-bound AR onto the multiple AREs (Fig. 8). Thus, HNF-3, by itself, is not a strong transactivator; however, the occupancy of HNF-3 in the regulatory region may confer certain transcriptional potential to target prostatic genes that is required for the later recruitment of additional factors as well as activation by a strong and rate-limiting activator (ligand-bound AR) during prostatic differentiation.
Transcriptional regulation depends not only on the interactions between DNA-binding proteins and their respective cis-regulatory elements but also on the interactions among these proteins and with other components of the transactivation machinery. In addition to interacting with various type I or type II coregulators, steroid receptors also make contact with other DNA binding proteins, resulting in modulation of transcriptional activity (80). In the present work, AR was found to directly interact with HNF-3 (Fig. 9), because such interaction was resistant to the presence of a DNA intercalator, EB (Fig. 9, A, B, and G), which serves as a good indicator of DNA-dependent and DNA-independent protein associations (65). In fact, AR has been found to interact with a number of DNA binding transcription factors including activator protein 1
TOP
ABSTRACT
INTRODUCTION
