Cross-repression, a Functional Consequence of the Physical ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2002 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 277, No. 21, Issue of May 24, pp. 18501–18509, 2002 Printed in U.S.A.

Cross-repression, a Functional Consequence of the Physical Interaction of Non-liganded Nuclear Receptors and POU Domain Transcription Factors* Received for publication, January 8, 2002, and in revised form, February 11, 2002 Published, JBC Papers in Press, March 12, 2002, DOI 10.1074/jbc.M200205200

Manuel Macias Gonzalez and Carsten Carlberg‡ From the Department of Biochemistry, University of Kuopio, FIN-70211 Kuopio, Finland

Nuclear receptors (NRs) and POU domain factors form two important transcription factor families for which several levels of functional interference have been described. In this study, the adopted orphan receptors constitutive androstane receptor (CAR) and pregnane X receptor (PXR) were found to perform direct protein-protein interactions with Pit-1, a representative POU domain factor. The ligand-dependent interaction profile of Pit-1 with CAR, PXR, and the vitamin D receptor in solution was shown to be that of a corepressor. In the absence of receptor agonist Pit-1 inhibited the complex formation of NRs with the retinoid X receptor on DNA. Also in living cells, Pit-1 and Oct-1, another POU domain factor, behaved like corepressors of NR signaling, and Pit-1-mediated repression was found to involve histone deacetylases. Conversely vitamin D receptor, CAR, and PXR were shown to act as repressors of Pit-1 signaling in different cell lines (MCF-7, HaCaT, and GH4C1). This repression was found to be independent of histone deacetylases and seems to be based on a competition of NRs with coactivator and corepressor proteins for overlaying interaction interfaces on the surface of Pit-1. Taken together this study suggests that cross-repression should occur in all tissues in which POU domain factors and non-liganded NRs meet each other.

The adopted orphan receptors constitutive androstane receptor (CAR,1 NR1I3) and pregnane X receptor (PXR, NR1I2) (1) are members of the nuclear receptor (NR) superfamily that regulates target gene transcription in a ligand-dependent manner. CAR and PXR have a rather broad, overlapping set of ligands that range from natural steroids to xenobiotics and also recognize similar DNA binding sites, referred to as response

* This work was supported by Academy of Finland Grants 50319 and 50331 (to C. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence should be addressed: Dept. of Biochemistry, University of Kuopio, P.O. box 1627, FIN-70211 Kuopio, Finland. Tel.: 358-17-163062; Fax: 358-17-2811510; E-mail: carlberg@messi. uku.fi. 1 The abbreviations used are: CAR, constitutive androstane receptor; 1␣,25(OH)2D3, 1␣,25-dihydroxyvitamin D3; AF-2, activation function-2; androstanol, 5␣-androstan-3␣-ol; DR4, direct repeat spaced by 4 nucleotides; HDAC, histone deacetylase; GH, growth hormone; GST, glutathione S-transferase; NR, nuclear receptor; PPAR, peroxisome proliferator-activated receptor; PXR, pregnane X receptor; RAR, retinoic acid receptor; RE, response element; RXR, retinoid X receptor; T3R, thyroid hormone receptor; TCPOBOP, 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene; TSA, trichostatin A; VDR, 1␣,25(OH)2D3 receptor; NCoR, nuclear receptor corepressor; LUC, luciferase; tk, thymidine kinase; FL, full length; WT, wild type. This paper is available on line at http://www.jbc.org

elements (REs), primarily in promoter regions of the cytochrome P450 gene family (2). CAR and PXR are closely related to each other as well as to another NR superfamily member, VDR (NR1I1), since they share ⬃40% amino acid identity of their ligand binding domain. VDR is bound with high affinity (Kd in the order of 0.1 nM) by its natural ligand 1␣,25-dihydroxyvitamin D3 (1␣,25(OH)2D3) (3), whereas CAR and PXR are bound by an overlapping set of natural and synthetic ligands with rather low affinity (Kd in the order of 1 ␮M) (4). Interestingly CAR differs from most other NRs by displaying in the absence of ligand already a relatively high constitutive activity, which can be reduced by the binding of the inverse agonist 5␣-androstan-3␣-ol (androstanol) (5). CAR and PXR each form a heterodimer with RXR on DR4-type REs (6) but also recognize other RE types. Similarly classical VDR binding sites have only three spacing nucleotides between the receptor binding motifs, but VDR-RXR heterodimers show even higher affinity for DR4-type REs (7, 8). Crystal structure analysis of more than 10 presently characterized NR ligand binding domains has demonstrated a conserved spatial structure formed by 11 or 12 ␣-helices (9). The ligand binding domain has diverse functions; it is not only involved in ligand binding but also in interaction with other NRs for the formation of homo- and heterodimeric complexes, and it is in contact with nuclear mediator proteins, such as coactivators and corepressors, for modulation of transcriptional activities. Contact points for coactivators, such as SRC-1, TIF-2, and RAC3 (10), have been mapped in the activation function-2 (AF-2) domain of helix 12 and in helix 3 (11), and also within the less well understood interaction surface for corepressors, such as NCoR, SMRT (silencing mediator for retinoid and thyroid hormone receptors), and Alien, the AF-2 domain was found to be of importance (12). The classical corepressors act as a specific bridge between transcription factors and histone deacetylases (HDACs), which are enzymes that locally close chromatin (13). POU domain factors, such as Pit-1, Oct-1, and Unc-86, form another large family of transcription factors, which are of special importance during development (14). Since some nuclear receptors are also known to have critical roles in development (15), a possible functional interference between members of the POU domain factor family and NR superfamily has been investigated quite extensively. For example, it had been found that pit-1 gene expression is down-regulated by thyroid hormone (16), that prolactin gene expression is synergistically activated by Pit-1 and either the VDR or the estrogen receptor via neighboring binding sites in the prolactin promoter (17, 18), and that Pit-1 interacts physically with the thyroid hormone receptor (T3R) and retinoid acid receptor (RAR) (19). This study was performed with the aim of attaining a better understanding of the multitude of protein-protein interactions

18501

18502

Cross-repression between NRs and Pit-1

of NRs. Therefore, direct protein-protein interactions between VDR, CAR, and PXR, as representative NRs, and Pit-1, as a representative POU domain factor, were investigated. It was shown that Pit-1 can act as a repressor of NR signaling and that conversely NRs can function as repressors of Pit-1 signaling. This crosswise repressor action, referred to as cross-repression, was demonstrated only for VDR, CAR, PXR, Pit-1, and Oct-1, but it possibly applies to all those members of these two transcription factor families that are coexpressed in the same tissue and that are able to participate in a direct proteinprotein interaction. MATERIALS AND METHODS

Compounds Androstanol was from Steraloids (Newport, RI); 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP) was synthesized and purified according to Honkakoski et al. (20); the HDAC inhibitor trichostatin A (TSA) was from Biomol (Plymouth Meeting, PA); sodium butyrate, rifampicin, and valproic acid (2-propylpentanoic acid) were from Sigma; and 1␣,25(OH)2D3 was kindly provided by L. Binderup (Leo Pharmaceutical Products, Ballerup, Denmark). 1␣,25(OH)2D3 was dissolved in 2-propanol, whereas all other compounds were dissolved in dimethyl sulfoxide (Me2SO); further dilutions were made in Me2SO (for in vitro experiments) or in ethanol (for cell culture experiments).

DNA Constructs Protein Expression Vectors—Full-length cDNAs for human PXR (21), human VDR (22), human RXR␣ (23), human RAR␥ (24), chicken T3R␣ (25), mouse PXR (26), human CAR (27), Xenopus PPAR␥ (28), and human PPAR␣ (29) were subcloned into the T7/SV40 promoter-driven pSG5 expression vector (Stratagene, Heidelberg, Germany); full-length cDNAs for mouse CAR (30), mouse NCoR (31), and human SRC-1 (32) were subcloned into the T7/cytomegalovirus promoter-driven pCMX expression vector; full-length cDNAs for rat Pit-1 (33) and human Oct-1 (34) were subcloned under the control of the Rous sarcoma virus promoter into the vector pUC18; and the full-length cDNA for rat Pit-1 was also subcloned into the T7 promoter-driven vector pBluescript SK(⫺) (Stratagene). Truncated versions of VDR (VDR⌬413–27), PXR (PXR⌬421–35), and CAR (CAR⌬349 –58), which were lacking their AF-2 domain, were generated by introducing via point-directed mutagenesis a stop codon at positions 413, 349, and 421, respectively, and confirmed by sequencing. The T7 promoter-driven constructs are suitable for T7 RNA polymerase-driven in vitro transcription/translation of the respective cDNAs, and the SV40-, cytomegalovirus-, and Rous sarcoma virusdriven constructs were used for overexpression of the respective proteins in mammalian cells. Glutatione S-Transferase (GST) Fusion Protein Constructs—The cDNAs for the NR interaction domains of mouse NCoR (spanning amino acids 1679 –2453) and human SRC-1 (spanning amino acids 596 –790) and for full-length rat Pit-1 were subcloned into the GST fusion protein vector pGEX (Amersham Biosciences). Reporter Gene Constructs—Three copies of the Pit-1 RE from the rat growth hormone (GH) promoter (core sequence 5⬘-CATGAATAAATGTA-3⬘ with core binding motifs in bold) (35) and four copies of the DR4-type RE from the human inducible NO synthase promoter (core sequence 5⬘-GTGGTTCATGCCGGTTCA-3⬘) (36) were fused with the thymidine kinase (tk) minimal promoter driving the firefly luciferase (LUC) reporter gene.

In Vitro Translation and Bacterial Overexpression of Proteins In vitro translated VDR, PXR, CAR, and RXR proteins were generated by coupled in vitro transcription/translation using rabbit reticulocyte lysate as recommended by the supplier (Promega, Mannheim, Germany). By test translation in the presence of [35S]methionine and taking the individual numbers of methionine residues per receptor into account, the specific concentration of the receptor proteins was adjusted to ⬃4 ng/␮l. Bacterial overexpression of GST-Pit-1FL, GST-NCoR(1679 –2453), and GST-SRC-1-(596 –790) was facilitated in the Escherichia coli BL21(DE3)pLysS strain (Stratagene). GST-SRC-1-(596 –790) and GST-Pit-1FL fusion protein expression was stimulated with 0.25 mM isopropyl-␤-D-thio-galactopyranoside for 3 h at 37 °C, and GSTNCoR-(1679 –2453) expression was induced with 1.25 mM isopropyl-␤D-thio-galactopyranoside for 5 h at 25 °C. The fusion proteins were purified and immobilized by glutathione-Sepharose 4B beads (Amersham Biosciences) according to the manufacturer’s protocol. For gel

shift experiments the fusion proteins were eluted by glutathione.

GST Pull-down Assays GST pull-down assays were performed with 50 ␮l of a 50% Sepharose bead slurry of GST-Pit-1FL, GST-NCoR-(1679 –2453), or GST-SRC-1(596 –790) (preblocked with 1 ␮g/␮l bovine serum albumin) and 20 ng of in vitro translated, 35S-labeled NRs in the presence or absence of their respective ligands. Proteins were incubated in immunoprecipitation buffer (20 mM Hepes (pH 7.9), 200 mM KCl, 1 mM EDTA, 4 mM MgCl2, 1 mM dithiothreitol, 0.1% Nonidet P-40, and 10% glycerol) for 20 min at 30 °C. For competition experiments, the peptides CoRNR2NCoR (SNLGLEDIIRKAL with core binding motifs in bold) or CoANR2TIF-2 (KHKILHRLLQDSS) were added before the addition of ligand. In vitro translated proteins that were not bound to GST fusion proteins were washed away with immunoprecipitation buffer. GST fusion proteinbound 35S-labeled NRs were resolved by electrophoresis through 10% SDS-polyacrylamide gels and quantified on a Fuji FLA3000 reader (Tokyo, Japan) using Image Gauge software (Fuji).

Gel Shift Assays Gel shift assays were performed with equal amounts (⬃10 ng) of in vitro translated VDR, CAR, PXR, and RXR and bacterially expressed GST-Pit-1FL (⬃3 ␮g) and GST (as a control). Proteins were incubated on ice for 15 min in a total volume of 20 ␮l of binding buffer (10 mM Hepes (pH 7.9), 150 mM KCl, 1 mM dithiothreitol, 0.2 ␮g/␮l poly(dI-dC), and 5% glycerol). For competition experiments, increasing concentrations of the peptides CoRNR2NCoR, CoANR2TIF-2, or AF-2RXR (TFLMEMLEAPHQMT) were added to the reaction mixture. Approximately 1 ng of 32P-labeled double-stranded oligonucleotides (50,000 cpm) corresponding to one copy of the Pit-1 RE from the rat GH promoter (core sequence 5⬘-CATGAATAAATGTA-3⬘) (35) or the DR4-type RE from the rat Pit-1 promoter (core sequence 5⬘-GAAGTTCATGAGAGTTCA3⬘) (37), respectively, were then added, and the incubation was continued for 20 min at room temperature. Protein-DNA complexes were resolved by electrophoresis through 8% non-denaturing polyacrylamide gels in 0.5⫻ TBE (45 mM Tris, 45 mM boric acid, and 1 mM EDTA (pH 8.3)) and quantified on a Fuji FLA3000 reader using Image Gauge software.

Transfection and Luciferase Reporter Gene Assays MCF-7 human breast cancer cells, HaCaT human immortalized keratinocytes, and GH4C1 rat pituitary tumor cells were seeded into six-well plates (105 cells/ml) and grown overnight in phenol red-free Dulbecco’s modified Eagle’s medium supplemented with 5% (or 10% in the case of GH4C1 cells) charcoal-treated fetal bovine serum. Liposomes were formed by incubating 1 ␮g of the reporter plasmid and the indicated combinations of expression vectors for Pit-1, VDR, CAR, PXR, RXR, NCoR, or SRC-1 (each 1 ␮g if not otherwise indicated) with 10 ␮g of N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP, Roth, Karlsruhe, Germany) for 15 min at room temperature in a total volume of 100 ␮l. After dilution with 900 ␮l of phenol red-free Dulbecco’s modified Eagle’s medium, the liposomes were added to the cells. Phenol red-free Dulbecco’s modified Eagle’s medium supplemented with 15% charcoal-treated fetal bovine serum (500 ␮l, 30% in the case of GH4C1 cells) was added 4 h after transfection. At this time, NR ligands or histone HDAC inhibitors were also added. The cells were lysed 16 h after onset of stimulation using the reporter gene lysis buffer (Roche Diagnostics), and the constant light signal luciferase reporter gene assay was performed as recommended by the supplier (CanberraPackard, Dreieich, Germany). The luciferase activities were normalized with respect to protein concentration, and induction factors were calculated as the ratio of luciferase activity of ligand-stimulated cells to that of solvent controls. RESULTS

The physical interaction between Pit-1 and a selection of 10 different NRs was assessed by GST pull-down assays using bacterially produced GST-Pit-1FL fusion protein immobilized on Sepharose beads and in vitro translated, 35S-labeled NR protein (Fig. 1). This qualitative screening suggested a direct protein-protein interaction between Pit-1 and human VDR, chicken T3R␣, human or mouse PXR, human or mouse CAR, Xenopus PPAR␥, or human PPAR␣ but not between Pit-1 and human RXR␣ or human RAR␥. GST protein alone was not able to interact with these receptors (data not shown).

Cross-repression between NRs and Pit-1

FIG. 1. Pit-1 physically interacts with various but not all NRs. GST pull-down assays were performed with bacterially expressed GSTPit-1FL fusion protein and in vitro translated, 35S-labeled human VDR, human RXR␣, human RAR␥, chicken T3R␣, human PXR, mouse PXR, human CAR, mouse CAR, Xenopus PPAR␥, and human PPAR␣. After precipitation and washing, the proteins were electrophoresed through 10% SDS-polyacrylamide gels. Representative gels are shown. The upper panel shows the NR input, and the lower panel shows the specifically bound proteins.

The interaction of Pit-1 with the PXR and CAR has not been reported yet and was investigated in more detail using human PXR and mouse CAR in reference to human VDR. First, the ligand-dependent interaction of NRs with Pit-1 was compared with their interaction with the corepressor NCoR and the coactivator SRC-1. GST pull-down assays were performed with bacterially expressed GST-Pit-1FL, GST-NCoR-(1679 –2453), and GST-SRC-1-(596 –790) and full-length in vitro translated, 35 S-labeled human VDR, human PXR, and mouse CAR in the absence and presence of their respective ligands (Fig. 2). The interaction between Pit-1 and VDR was found to decrease after addition of the natural VDR agonist 1␣,25(OH)2D3, which resembles the interaction profile of VDR with NCoR but is in contrast to that of VDR and SRC-1 (Fig. 2A). A similar tendency was observed for human PXR and its synthetic agonist rifampicin: the interaction of PXR with Pit-1 was clearly reduced in the presence of ligand, and also the interaction of PXR with NCoR was slightly but significantly reduced, whereas for the interaction of PXR with SRC-1 no significant ligand-dependent effect was found (Fig. 2B). The profile of the liganddependent interaction of mouse CAR with Pit-1 was found to be very similar to that with NCoR and inverse to that of SRC-1 (Fig. 2C). The synthetic agonist TCPOBOP decreased the interaction between CAR and Pit-1 or NCoR but increased the contact of CAR with SRC-1. The inverse agonist androstanol showed the opposite tendency: it slightly increased the interaction between CAR and Pit-1 or NCoR and decreased the contact of CAR with SRC-1. The combined treatment with TCPOBOP and androstanol resulted in effects slightly weaker than that of TCPOBOP alone but still significantly different from the solvent control. In many cases the AF-2 domain of NRs has been shown to mediate their ligand-dependent protein-protein interaction. Therefore, the agonist-triggered interaction of human VDR and mouse CAR with Pit-1 was compared with that of truncated versions of the receptors, which lack their C-terminal AF-2 domains. GST pull-down assays were performed with bacterially expressed GST-Pit-1FL and in vitro translated, 35S-labeled human VDRWT, VDR⌬413–27, CARWT, and CAR⌬349 –58 (Fig. 3). As already shown in Fig. 2, in the presence of the agonists 1␣,25(OH)2D3 and TCPOBOP the interaction of full-length VDR and CAR, respectively, with Pit-1 was clearly reduced. In

18503

contrast, in the absence of agonist the interaction of VDR⌬413–27 and CAR⌬349 –58 with Pit-1 was found to be already at approximately the same low level as the respective fulllength receptor in the presence of agonist but did not respond further to ligand addition. The AF-2 domain has been shown to be critical for the interaction of NRs both with coactivator proteins as well as with corepressor proteins. Therefore the question of whether Pit-1 competes with corepressors and coactivators for the binding to the AF-2 domain of VDR and CAR was addressed. GST pulldown assays were performed with bacterially expressed GSTPit-1FL and in vitro translated, 35S-labeled human VDR and mouse CAR in the presence of increasing concentrations of the peptides CoRNR2NCoR and CoANR2TIF-2 (Fig. 4). The peptides represent specific NR interaction domains of the corepressor NCoR and the coactivator TIF-2, respectively (38, 39). In the absence of ligand, increasing concentrations of CoRNR2NCoR but not of CoANR2TIF-2 were shown to be able to reduce the interaction between Pit-1 and VDR and CAR, i.e. corepressors compete with Pit-1 for binding to both receptors. In contrast, in the presence of agonist, CoANR2TIF-2 but not CoRNR2NCoR was found to compete effectively with Pit-1 for binding to VDR and CAR. As a control, competition experiments were performed with a peptide of the same length and concentration representing the AF-2 domain of human RXR␣ (AF-2RXR). This peptide had no effect on the interaction of Pit-1 with VDR and CAR in either the absence or presence of receptor agonists (data not shown). The previous experiments demonstrated the interaction of Pit-1 with NRs in solution. As the next step this interaction was assessed bound to DNA by performing gel shift experiments with bacterially expressed GST-Pit-1FL on the Pit-1 RE of the rat GH gene in the presence of in vitro translated human VDR, mouse CAR, human PXR, or human RXR␣ (Fig. 5). The RE showed to be specific for Pit-1 since only the addition of Pit-1 but not of any of the four NRs resulted in a detectable protein-DNA complex. Interestingly the preincubation of Pit-1 with VDR, CAR, and PXR resulted in a slower migrating protein-DNA complex and indicated that VDR, CAR, and PXR can bind to DNA-bound Pit-1. In contrast, the addition of RXR was without any effect and confirmed the observation from Fig. 1 that RXR cannot interact with Pit-1. Competition experiments with the corepressor peptide CoRNR2NCoR and the coactivator peptide CoANR2TIF-2 resulted at higher concentrations in a loss of the “supershifted” protein-DNA complex, whereas the same amount of the unspecific peptide AF-2RXR showed no effect. This suggests that corepressor proteins as well as coactivator proteins are able to compete with DNA-bound Pit-1 for the binding to NRs. The inverse question to the point addressed in the previous experiment (Fig. 5) was whether Pit-1 also could interact with DNA-bound NRs. Therefore, gel shift experiments were performed with in vitro translated heterodimers between human VDR and human RXR␣ and between mouse CAR and human RXR␣ and the DR4-type RE of the rat Pit-1 gene (Fig. 6). As reported previously (7), VDR-RXR heterodimers demonstrated a clearly agonist-triggered complex formation on this RE, whereas the DNA binding of CAR-RXR heterodimers was shown to be ligand-independent. Interestingly the preincubation of VDR-RXR and CAR-RXR heterodimers with increasing concentrations of bacterially expressed GST-Pit-1FL and GST alone (as a control) resulted in a Pit-1-specific inhibition of DNA complex formation of the heterodimers only in the absence but not in the presence of receptor agonists. In the previous in vitro experiments (Figs. 2 and 6) the interaction between NRs and Pit-1 looked more like a crosswise

18504

Cross-repression between NRs and Pit-1

FIG. 2. The direct interaction of Pit-1 with NRs resembles that of corepressors but not that of coactivators. GST pull-down assays were performed with bacterially expressed GST-Pit-1FL, GST-NCoR-(1679 –2453), and GST-SRC-1-(596 –790) and full-length in vitro translated, 35 S-labeled human VDR (A), human PXR (B), and mouse CAR (C) in the absence and presence of their respective ligands. GST alone was used as control. After precipitation and washing, the samples were electrophoresed through 10% SDS-polyacrylamide gels, and the percentage of precipitated NRs in respect to input was quantified using a Fuji FLA3000 reader. Representative gels are shown. Columns represent mean values of triplicates, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to respective solvent controls (*, p ⬍ 0.05; **, p ⬍ 0.01). DMSO, Me2SO.

FIG. 3. Critical role of the AF-2 domain in direct interaction between Pit-1 and VDR or CAR. GST pull-down assays were performed with bacterially expressed GST-Pit-1FL and in vitro translated, 35S-labeled human VDR (A) and mouse CAR (B) in the absence and presence of their respective agonists. The NRs were either wild type or truncated by their C-terminal AF-2 domains. GST alone was used as control. After precipitation and washing, the samples were electrophoresed through 10% SDS-polyacrylamide gels, and the percentage of precipitated NRs in respect to input was quantified using a Fuji FLA3000 reader. Representative gels are shown. Columns represent mean values of triplicates, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to respective solvent controls (*, p ⬍ 0.05). DMSO, Me2SO.

repression (here referred to as cross-repression) than a synergistic activation. To study the interaction of Pit-1 with VDR, CAR, and PXR in living cells, luciferase reporter gene assays were performed in MCF-7 and HaCaT cells, which both represent model cell lines for NR signaling. The cells were transiently transfected with a reporter gene construct driven by

three copies of the rat GH Pit-1 RE and expression vectors for rat Pit-1, mouse NCoR, human VDR, mouse CAR, and human PXR (the latter three NRs each in combination with human RXR␣, see Fig. 7). The overexpression of Pit-1 resulted for both cell lines in an ⬃4-fold increase of basal reporter gene activity (column 4). Both in MCF-7 and in HaCaT cells Pit-1 signaling

Cross-repression between NRs and Pit-1

18505

FIG. 4. Pit-1 competes with corepressors and coactivators for binding to VDR and CAR. GST pull-down assays were performed with bacterially expressed GST-Pit-1FL and in vitro translated, 35S-labeled human VDR (A) and mouse CAR (B) in the absence and presence of their respective agonists. Increasing concentrations of the peptides CoRNR2NCoR (SNLGLEDIIRKAL) and CoANR2TIF-2 (KHKILHRLLQDSS) were added before supplying the ligand. After precipitation and washing, the samples were electrophoresed through 10% SDS-polyacrylamide gels, and the percentage of precipitated NRs in respect to input was quantified using a Fuji FLA3000 reader. Representative gels are shown. Columns represent mean values of triplicates, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to receptor without competing peptide (*, p ⬍ 0.05; **, p ⬍ 0.01). DMSO, Me2SO.

FIG. 5. VDR, CAR, and PXR but not RXR bind to DNA-complexed Pit-1. Gel shift experiments were performed with bacterially expressed GST-Pit-1FL (3 ␮g), in vitro translated human VDR, mouse CAR, human PXR, or human RXR␣ (each 20 ng), and the 32P-labeled Pit-1 RE of the rat GH gene (1 ng). As indicated, increasing concentrations of the peptides CoRNR2NCoR, CoANR2TIF-2, and AF-2RXR (TFLMEMLEAPHQMT) were supplied before the addition of DNA. Protein-DNA complexes (Pit-1-DNA and Pit-1-NR-DNA) were separated from free probe through 8% non-denaturing polyacrylamide gels. A representative gel is shown. NS indicates unspecific complexes.

was clearly repressed by the overexpression of NCoR (column 7) and even reduced below basal activity by the overexpression of VDR, CAR, and PXR (columns 10, 13, and 16). The repression of Pit-1 signaling was specific to VDR, CAR, and PXR since

the overexpression of their heterodimeric partner RXR alone showed no effect (columns 19 and 21). The NCoR-mediated repression of Pit-1 signaling could be released by the HDAC inhibitors TSA, sodium butyrate, and valproic acid (columns 8

18506

Cross-repression between NRs and Pit-1

FIG. 6. Pit-1 inhibits ligand-independent NR complex formation on DNA. Gel shift experiments were performed with in vitro translated heterodimers of human VDR and human RXR␣ (A) and of mouse CAR and human RXR␣ (B, each 20 ng) and the 32P-labeled DR4-type RE of the rat Pit-1 gene (1 ng). In the presence and absence of their respective agonists the receptors were preincubated with increasing concentrations (1, 3, and 6 ␮g) of bacterially expressed GST and GST-Pit-1FL. Protein-DNA complexes were separated from free probe through 8% non-denaturing polyacrylamide gels. Representative gels are shown. The percentage of protein-complexed DNA was quantified using a Fuji FLA3000 reader. Columns represent the mean of triplicates, and bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to heterodimers in the absence of GST proteins (*, p ⬍ 0.05).

and 9). In contrast, the HDAC inhibitors showed only minor effects on the NR-mediated repression of Pit-1 signaling (columns 11, 12, 14, 15, 17, and 18). GH4C1 rat pituitary tumor cells endogenously express sufficient amounts of Pit-1 protein so that additional overexpression of the transcription factor was not able to booster Pit-1 signaling (data not shown). Reporter gene assays were performed in GH4C1 cells that were transiently transfected with a reporter gene construct driven by three copies of the rat GH Pit-1 RE and expression vectors for human SRC-1, mouse NCoR, human VDR and its AF-2 deletion mutant VDR⌬413–27, mouse CAR and its AF-2 deletion mutant CAR⌬349 –58, and human PXR and its AF-2 deletion mutant PXR⌬421–35 (Fig. 8). The overexpression of SRC-1 increased the basal Pit-1 activity (column 1) by ⬃40% (column 3), whereas the overexpression of NCoR reduced it by approximately the same amount (column 5). All three tested NRs repressed Pit-1 signaling to a level of 10 –30% basal activity (columns 7, 11, and 15). Again this repression was specific to VDR, CAR, and PXR since the overexpression of RXR showed no effect (columns 19 and 21). AF-2 domain deletion mutants of VDR, CAR, and PXR showed less repression of Pit-1 signaling (columns 9, 13, and 17), which resulted in activities that were approximately double those

observed with the respective full-length receptors (columns 7, 11, and 15). The treatment with the HDAC inhibitor TSA resulted in an ⬃50% increase of basal Pit-1 activity (column 2). The same level of Pit-1 signaling was also reached by TSAtreated, SRC-1- or NCoR-overexpressing cells (columns 4 and 6) and appears to represent the maximal Pit-1 activity in GH4C1 cells. TSA treatment also resulted in a release of NRmediated repression of Pit-1 signaling (columns 8, 12, and 16). However, this inhibition of repression was only incomplete since only activities significantly below the maximal Pit-1 activity were obtained. For the inverse functional experiment, reporter gene assays were performed in MCF-7 cells that were transiently transfected with a reporter gene construct driven by four copies of the human inducible NO synthase DR4-type RE and expression vectors for rat Pit-1, human Oct-1, mouse NCoR, and mouse CAR (Fig. 9). The overexpression of CAR increased basal reporter gene activity on average ⬃10-fold (column 7), whereas the overexpression of Pit-1 (column 3), NCoR (A, column 5), and Oct-1 (B, column 5) alone showed only minor effects on DR4type RE-driven gene activity (column 1). The combined overexpression of CAR and Pit-1 (A, column 9, and B, column 13) or NCoR (A, column 11) resulted in ⬃50% reduced gene activity.

Cross-repression between NRs and Pit-1

18507

FIG. 7. NR-mediated repression of Pit-1 signaling in model cell lines. Luciferase reporter gene assays were performed with extracts from MCF-7 human breast cancer cells (A) and HaCaT human immortalized keratinocytes (B) that were transiently transfected with a reporter gene construct driven by three copies of the rat GH Pit-1 RE ((Pit-1 RE)3-tk-LUC) and the indicated expression vectors for rat Pit-1, mouse NCoR, human VDR, mouse CAR, human PXR, and human RXR␣ (for RXR 1 and 3 ␮g). The cells were treated for 16 h with the indicated HDAC inhibitors (TSA, sodium butyrate (NaBt), and valproic acid (VPA)) or ethanol (as a control). Stimulation of normalized luciferase activity was calculated in comparison with ethanol-induced cells that do not overexpress any proteins. Columns represent the mean of at least triplicates from three independent experiments, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to cells overexpressing only Pit-1 (*, p ⬍ 0.05; **, p ⬍ 0.01).

repressing effect of Pit-1 overexpression on CAR signaling was found to be dose-dependent (B, columns 9, 11, and 13). Oct-1, another prominent POU domain family member, showed comparable repressing effects on CAR signaling (B, columns 15, 17, and 19), and the combined overexpression of Pit-1 and Oct-1 even resulted in super-repression down to 25% of maximal CAR activity (B, columns 21, 23, and 25). However, the HDAC inhibitor TSA was found to release the repressing effect of Pit-1 on CAR signaling completely (compare column 8 with columns 10, 12, and 14). Moreover, the data appear to suggest that TSA releases the repression by Oct-1 only partially (compare column 8 with columns 16, 18, 20, 22, 24, and 26), but this effect was found not to be statistically significant. DISCUSSION

FIG. 8. NR-mediated repression of Pit-1 signaling in rat pituitary cells. Luciferase reporter gene assays were performed with extracts from GH4C1 rat pituitary tumor cells that were transiently transfected with a reporter gene construct driven by three copies of the rat GH Pit-1 RE ((Pit-1 RE)3-tk-LUC) and the indicated expression vectors for human SRC-1, mouse NCoR, human VDR and its AF-2 deletion mutant VDR⌬413–27, mouse CAR and its AF-2 deletion mutant CAR⌬349 –58, human PXR and its AF-2 deletion mutant PXR⌬421–35, and human RXR␣. The cells were treated for 16 h with the HDAC inhibitor TSA or ethanol (as a control). Stimulation of normalized luciferase activity was calculated in comparison with basal Pit-1 activity. Columns represent the mean of at least triplicates from three independent experiments, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to basal Pit-1 activity (*, p ⬍ 0.05; **, p ⬍ 0.01).

Interestingly in the presence of agonist TCPOBOP the overexpression of Pit-1 and NCoR (A, columns 10 and 12) was of no significant effect on CAR signaling (A, column 8), i.e. Pit-1 and NCoR act only in the absence of ligand as corepressors. The

The qualitative screening of 10 different NRs for their ability to perform direct protein-protein interaction with Pit-1 resulted in 80% positive results. Classical endocrine receptors, such as VDR and T3R, as well as adopted orphan receptors, such as CAR, PXR, and PPAR, were found to contact Pit-1 directly. The interaction of Pit-1 with VDR (17), T3R (19), and PPAR (40) has already been shown earlier, whereas the interaction between Pit-1 and CAR or PXR is reported here for the first time. Palomino et al. (19) also have reported a direct contact between Pit-1 and RAR or RXR, which could not been confirmed in this study either in solution (Fig. 1) or bound to DNA (Fig. 5). The ligand-dependent interaction profile of Pit-1 with VDR, CAR, or PXR definitely more closely resembles that of the corepressor NCoR than that of the coactivator SRC-1 since the addition of receptor agonist definitely reduces the Pit-1-NR interaction in solution (Fig. 2). Not all regions of the NR ligand binding domains that contribute to the corepressor interaction interface have been mapped yet, but it seems to be clear that helix 12 with its AF-2 domain is involved (12). Therefore, the finding reported herein that the deletion of the AF-2 domain

18508

Cross-repression between NRs and Pit-1

FIG. 9. Pit-1-mediated repression of CAR signaling. Luciferase reporter gene assays were performed with extracts from MCF-7 cells that were transiently transfected with a reporter gene construct driven by four copies of the human inducible NO synthase DR4-type RE ((DR4)4-tkLUC) and the indicated amounts (in ␮g) of expression vectors for rat Pit-1, human Oct-1, mouse NCoR, and mouse CAR. The cells were treated for 16 h with ethanol (as a control) and either the CAR agonist TCPOBOP (A) or the HDAC inhibitor TSA (B). Stimulation of normalized luciferase activity was calculated in comparison with ethanol-induced cells that do not overexpress any proteins. Columns represent the mean of at least triplicates from three independent experiments, and the bars indicate S.D. Statistical analysis was performed by two-tailed, paired Student’s t test, and p values were calculated in reference to cells overexpressing only CAR (*, p ⬍ 0.05; **, p ⬍ 0.01).

clearly reduces the affinity of NRs for Pit-1 and abolishes the ligand dependence of this interaction (Fig. 3) fits well with the idea that Pit-1 may act as a corepressor to some NRs. Further support of this concept is provided by the observation that the NR interaction domain of NCoR can compete with Pit-1 for binding to VDR and CAR (Fig. 4). This indicates that nonliganded NRs have clearly overlaid interaction interfaces for Pit-1 and NCoR on their surfaces. The interaction of NRs with Pit-1 was found not to interfere with the DNA binding of Pit-1 since NRs can supershift Pit-1DNA complexes (Fig. 5). Interestingly peptides that represent the receptor interaction domain of corepressors as well as those that represent interaction domains of coactivators were shown to be able to prevent these supershifts in the absence of NR agonists, but they did not affect the amount Pit-1-DNA complexes. In turn, this suggests that NRs may be able to inhibit the interaction of Pit-1 proteins with corepressor as well as with coactivator proteins, i.e. NRs may block both repression and activation of Pit-1 activity. Conversely in the absence of ligand Pit-1 is able to inhibit (at least partially) the DNA complex formation of VDR-RXR, CAR-RXR, and PXR-RXR heterodimers (Fig. 6). This effect could be explained by a steric hindrance caused by Pit-1 that either blocks heterodimerization of VDR or CAR with RXR (being essential for effective DNA binding) or prevents the contact between the DNA binding domain of the NRs and DNA. However, the observation that Pit-1 is not able to inhibit the DNA binding of NR heterodimers in the presence of their agonist suggests that, at least in complex with DNA, the agonistic receptor conformation does not interact effectively with Pit-1. This is another indication that Pit-1 acts rather as a repressor than as an activator of NR activity. A repressing effect of NR on Pit-1 signaling as predicted by the in vitro experiments (Figs. 2–5) could be demonstrated in the model cell lines MCF-7 and HaCaT (Fig. 7) as well as in

endogenously Pit-1-responding GH4C1 cells (Fig. 8). The repressing function of NCoR, which is known as a repressor of Pit-1 signaling (41), could be released completely by different HDAC inhibitors. In contrast, the same inhibitors caused only a partial release of NR-mediated repression of Pit-1 signaling. In addition, the functional assays confirmed the important role of the AF-2 domains of VDR, CAR, and PXR for the repressing action of the receptors on Pit-1 signaling, but the partial release of repression through deletion of the AF-2 domain also indicated that helix 12 is only a part of the interaction interface with Pit-1. Taken together the functional assays suggest that NRs do not act as classical repressors that recruit HDACs to locally close chromatin, but they most likely inhibit activation of Pit-1 by preventing its interaction with coactivator proteins. For CAR signaling the repressing potential of Pit-1 on NR activity was demonstrated in the model cell line MCF-7 (Fig. 9). Comparable to NCoR, Pit-1 showed no effect in the presence of receptor agonists and confirmed the results of the in vitro experiments (Fig. 6). Interestingly another POU domain factor, Oct-1, was also found to be able to repress CAR-signaling. This supports the concept of a general interaction potential between members of the POU domain factor family and the NR family. Another interesting aspect is that the repressing effect of Pit-1 on CAR signaling could be released completely by treatment with HDAC inhibitors. This suggests that Pit-1 may be able to recruit HDACs directly (or most likely indirectly). However, this property seems not to be general for POU domain factors since it was found that TSA has only minor effects on Oct-1mediated repression of CAR signaling. It is obvious that not all members of the NR superfamily and the POU domain family are expressed in the same tissue or cell line. This makes it well possible that not all interactions that have been detected in vitro are of physiological relevance. However, as demonstrated in this study, Pit-1 is rather promiscuous in its interaction with NRs so that it will certainly find a

Cross-repression between NRs and Pit-1 partner receptor in every Pit-1-expressing tissue. Moreover, since the interaction of Pit-1 with NRs seems to be largely independent from receptor agonists, it may be not very critical which exact member of the NR superfamily Pit-1 finds as a partner. Pit-1 itself has a very restricted expression pattern and is primarily found in the anterior pituitary gland. The gland shows robust expression of VDR (42) but only low levels of CAR and PXR mRNA expression.2 In contrast, Oct-1 is rather ubiquitously expressed and seems to be able to take a similar repressor role as Pit-1 in tissues such as liver and intestine where high expression levels of CAR and PXR are found (26, 27). For the prolactin gene promoter a synergistic activation by Pit-1 and VDR has been observed, which appears to be in contrast to the results reported in this study (17). However, in the case of the prolactin promoter both transcription factors are able bind to DNA, and the synergistic activation effects were primarily obtained by treatment with the agonist 1␣,25(OH)2D3 and not by VDR overexpression. In contrast, the effects described here were mainly observed in the absence of NR agonists. Cross-repression may apply more likely to those genes that carry a binding site for either a NR or a member of the POU domain family but not for both together in their promoter region. In the rare case, such as the prolactin promoter, where two such binding sites are in close vicinity to each other, mechanisms as described by Castillo et al. (17) may overcome the cross-repression described in this study. In conclusion, this study has demonstrated that Pit-1 can act as a repressor of NR signaling, and conversely NRs have been shown to act as repressors of Pit-1 signaling. This cross-repression is mediated by a direct protein-protein interaction between the NR and Pit-1. Although this principle has been demonstrated only for the NRs VDR, CAR, and PXR and the POU domain family members Pit-1 and Oct-1, the cross-repression described here may apply to all members of the two transcription factor families for which a direct protein-protein interaction can be demonstrated. Acknowledgments—We thank M. Hiltunen for skilled technical assistance, L. Binderup for 1␣,25(OH)2D3, P. Honkakoski for TCPOBOP, and A. Aranda, S. Kliewer, and S. Rhodes for protein expression vectors. REFERENCES 1. Nuclear Receptor Committee (1999) Cell 97, 161–163 2. Honkakoski, P., and Negishi, M. (2000) Biochem. J. 347, 321–337 3. Herdick, M., Bury, Y., Quack, M., Uskokovic, M., Polly, P., and Carlberg, C. (2000) Mol. Pharmacol. 57, 1206 –1217 4. Moore, L. B., Parks, D. J., Jones, S. A., Bledsoe, R. K., Consler, T. G., Stimmel, J. B., Goodwin, B., Liddle, C., Blanchard, S. G., Willson, T. M., Collins, J. L., and Kliewer, S. A. (2000) J. Biol. Chem. 275, 15122–15127 5. Tzameli, I., and Moore, D. D. (2001) Trends Endocrinol. Metab. 12, 7–10 6. Sueyoshi, T., Kawamoto, T., Zelko, I., Honkakoski, P., and Negishi, M. (1999) J. Biol. Chem. 274, 6043– 6046 2

M. Macias Gonzalez and C. Carlberg, unpublished results.

18509

7. Quack, M., and Carlberg, C. (2000) J. Mol. Biol. 296, 743–756 8. Toell, A., Polly, P., and Carlberg, C. (2000) Biochem. J. 352, 301–309 9. Bourguet, W., Germain, P., and Gronemeyer, H. (2000) Trends Pharmacol. Sci. 21, 381–388 10. Leo, C., and Chen, J. D. (2000) Gene (Amst.) 245, 1–11 11. Feng, W., Ribeiro, R. C. J., Wagner, R. L., Nguyen, H., Apriletti, J. A., Fletterick, R. J., Baxter, J. D., Kushner, P. J., and West, B. L. (1998) Science 280, 1747–1749 12. Burke, L. J., and Baniahmad, A. (2000) FASEB J. 14, 1876 –1888 13. Ng, H. H., and Bird, A. (2000) Trends Biochem. Sci. 25, 121–126 14. Ryan, A. K., and Rosenfeld, M. G. (1997) Genes Dev. 11, 1207–1225 15. Altucci, L., and Gronemeyer, H. (2001) Trends Endocrinol. Metab. 12, 460 – 468 16. Sanchez-Pacheco, A., Palomino, T., and Aranda, A. (1995) Mol. Cell. Biol. 15, 6322– 6330 17. Castillo, A. I., Jimenez-Lara, A. M., Tolon, R. M., and Aranda, A. (1999) Mol. Endocrinol. 13, 1141–1154 18. Schaufele, F., Chang, C. Y., Liu, W., Baxter, J. D., Nordeen, S. K., Wan, Y., Day, R. N., and McDonnell, D. P. (2000) Mol. Endocrinol. 14, 2024 –2039 19. Palomino, T., Sanchez-Pacheco, A., Pena, P., and Aranda, A. (1998) FASEB J. 12, 1201–1209 20. Honkakoski, P., Moore, R., Gynther, J., and Negishi, M. (1996) J. Biol. Chem. 271, 9746 –9753 21. Bertilsson, G., Heidrich, J., Svensson, K., Jendeberg, L., Sydow-Backman, M., Ohlsson, R., Postlind, H., Blomquist, P., and Berkenstam, A. (1998) Proc. Natl. Acad. Sci. U. S. A. 13, 12208 –12213 22. Baker, A. R., McDonnell, D. P., Hughes, M., Crisp, T. M., Mangelsdorf, D. J., Haussler, M. R., Pike, J. W., Shine, J., and O’Malley, B. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 3294 –3298 23. Mangelsdorf, D. J., Ong, E. S., Dyck, J. A., and Evans, R. M. (1990) Nature 345, 224 –229 24. Krust, A., Kastner, P., Petkovich, M., Zelent, A., and Chambon, P. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 5310 –5314 25. Sap, J., Munoz, A., Damm, K., Goldberg, Y., Ghysdael, J., Leutz, A., Beug, H., and Vennstro¨ m, B. (1986) Nature 324, 635– 640 26. Kliewer, S. A., Moore, J. T., Wade, L., Staudinger, J. L., Watson, M. A., Jones, S. A., McKee, D. D., Oliver, S. A., Willson, T. M., Zetterstro¨ m, R. H., Perlmann, T., and Lehmann, J. M. (1998) Cell 92, 73– 82 27. Baes, M., Gulick, T., Choi, H.-S., Martinoli, M. G., Simha, D., and Moore, D. D. (1994) Mol. Cell. Biol. 14, 1544 –1552 28. Dreyer, C., Krey, G., Keller, H., Givel, F., Helftenbein, G., and Wahli, W. (1992) Cell 68, 879 – 887 29. Sher, T., Yi, H.-F., McBride, O. W., and Gonzalez, F. J. (1993) Biochemistry 32, 5598 –5604 30. Choi, H. S., Chung, M., Tzameli, I., Simha, D., Lee, Y. K., Seol, W., and Moore, D. D. (1997) J. Biol. Chem. 272, 23565–23571 31. Ho¨ rlein, A. J., Na¨ a¨ r, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., So¨ derstro¨ m, M., Glass, C. K., and Rosenfeld, M. G. (1995) Nature 377, 397– 404 32. Onate, S. A., Tsai, S. Y., Tsai, M. J., and O’Malley, B. W. (1995) Science 270, 1354 –1357 33. Ingraham, H. A., Chen, R. P., Mangalam, H. J., Elsholtz, H. P., Flynn, S. E., Lin, C. R., Simmons, D. M., Swanson, L., and Rosenfeld, M. G. (1988) Cell 55, 519 –529 34. Sturm, R. A., Das, G., and Herr, W. (1988) Genes Dev. 2, 1582–1599 35. Schaufele, F., West, B. L., and Baxter, J. D. (1992) Mol. Endocrinol. 6, 656 – 665 36. Toell, A., Kro¨ ncke, K. D., Kleinert, H., and Carlberg, C. (2002) J. Cell. Biochem. 85, 72– 82 37. Rhodes, S. J., Chen, R., DiMattia, G. E., Scully, K. M., Kalla, K. A., Lin, S.-C., Yu, V. C., and Rosenfeld, M. G. (1993) Genes Dev. 7, 913–932 38. Darimont, B. D., Wagner, R. L., Apriletti, J. W., Stallcup, M. R., Kushner, P. J., Baxter, J. D., Fletterick, R. J., and Yamamoto, K. R. (1998) Genes Dev. 12, 3343–3356 39. Hu, X., and Lazar, M. A. (1999) Nature 402, 93–96 40. Tolon, R. M., Castillo, A. I., and Aranda, A. (1998) J. Biol. Chem. 273, 26652–26661 41. Scully, K. M., Jacobson, E. M., Jepsen, K., Lunyak, V., Viadiu, H., Carriere, C., Rose, D. W., Hooshmand, F., Aggarwal, A. K., and Rosenfeld, M. G. (2000) Science 290, 1127–1131 42. Walters, M. R. (1992) Endocr. Rev. 13, 719 –764