HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 275, Issue 25, 19401-19408, June 23, 2000

This Article Abstract Full Text (PDF) All Versions of this Article: 275/25/19401    most recent M002472200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Klein, E. S. Articles by Chandraratna, R. A. S. Search for Related Content PubMed PubMed Citation Articles by Klein, E. S. Articles by Chandraratna, R. A. S. Social Bookmarking                      What's this?

Recruitment of Nuclear Receptor Corepressor and Coactivator to the Retinoic Acid Receptor by Retinoid Ligands

INFLUENCE OF DNA-HETERODIMER INTERACTIONS^*

Elliott S. Klein^§, Jenny W. Wang, Berket Khalifa, Stacey A. Gavigan, and Roshantha A. S. Chandraratna^§¶

From Retinoid Research, Departments of  Biology and ^¶ Chemistry, Allergan Pharmaceuticals, Irvine, California 92715

Received for publication, March 22, 2000

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Ligand activation of retinoic acid receptors (RARs) involves coordinated changes in their interaction with coregulatory molecules. Binding of the agonist all-trans-retinoic acid to the RAR results in increased interaction with coactivator molecules as well as a decreased interaction with corepressor molecules. Thus, an all-trans-retinoic acid antagonist might function either by preventing agonist induction of such events or, additionally, by actively increasing repression via corepressor recruitment. We demonstrate that the repression of the transcriptional activity of a constitutively active RAR-VP-16 chimeric receptor by the inverse agonist AGN193109 requires a functional Co-R box and that binding of this ligand to RAR leads to an increased interaction with the corepressor N-CoR both in glutathione S-transferase pull-down and yeast two-hybrid analyses. Detection of nuclear receptor corepressor (N-CoR) association with RAR was greatly facilitated by inclusion of a RARE oligonucleotide in coimmunoprecipitation analyses, a result of an increase in association of the ternary complex consisting of RAR, RXR, and DNA. Similarly, this DNA-dependent increase in heterodimer formation likewise resulted in an increase in agonist-mediated recruitment efficiency of the coactivator SRC-1. Under conditions which favor ternary complex formation, a RAR neutral antagonist is distinguished from an inverse agonist with respect to corepressor recruitment as is a RAR partial agonist distinguished from an agonist with respect to coactivator recruitment. These results indicate that it is possible to design RAR ligands with distinct recruitment capabilities for coregulators, both coactivators as well as corepressors. In addition, using this recruitment assay, we show that SRC-1 and the related coactivator molecule ACTR associate with the ternary complex via utilization of different helical motifs within their conserved receptor interaction domains.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Binding of ATRA¹ to the RARs, members of the nuclear hormone receptor superfamily, results in coordinated changes in the associations of the RAR with various cofactors. In the absence of hormone, RAR has been shown to be associated with corepressor molecules such as N-CoR and SMRT (1, 2). Upon binding ATRA, corepressor association is disrupted and interaction with coactivator molecules occurs. Various nuclear receptor coactivators have been described including CBP/p300 (3, 4), SRC-1 (5), TIF-2/Grip1 (6, 7), ACTR/RAC3/p/CIP (8-10). Models for ligand-mediated receptor transactivation have thus consisted of transcriptionally repressed receptors in the absence of hormone, due to corepressor association, followed by derepression and activation upon hormone addition. Alternatively, maximum corepressor interaction may not occur in the unliganded state. If so, the binding of certain ligands such as antagonists could lead to corepressor recruitment. Furthermore, RAR antagonists could conceivably vary in their mechanism of action ranging from simple competitive displacement of agonist to a more active process involving recruitment of nuclear corepressor.

Inverse agonists have been described for -adrenergic receptors (11), opioid receptors (12), serotonin receptors (13), dopamine receptors (14) as well as for members of the steroid receptor superfamily (15, 16). These active antagonists have the ability to lower the constitutive activity, in the opposite direction exerted by agonist, of these receptors which occurs either naturally or results from overexpression or mutation. For G-protein-coupled receptors, inverse agonists have been proposed to stabilize a receptor conformation which has decreased affinity for G-protein. The existence of corepressor molecules offers a potential mechanism to account for inverse agonism observed for nuclear receptors. Thus, in a manner analogous to agonist-mediated coactivator recruitment, binding of an antagonist to a nuclear receptor may lead to an increase in the association with a corepressor.

We have recently identified novel synthetic RAR ligands which function as antagonists of ATRA (16-18). One of these antagonists, AGN193109, exhibits inverse agonist activity as evidenced by its repression of the basal activity of a chimeric RAR-VP-16 in the absence of added agonist (16). In this report we demonstrate that interaction between the nuclear corepressor N-CoR and the RAR can be increased by RAR antagonists. In the course of these experiments, we developed a method which facilitates ternary complex formation by RAR, RXR, and DNA. The association of coregulator molecules with the ternary complex is greatly facilitated relative to RARs in isolation. This increased association allowed us detect ligand-mediated recruitment of N-CoR to the three different RAR isoforms and distinguish RAR antagonists in regard to their respective corepressor recruitment capabilities. Similarly, this method allowed for RAR agonist, partial agonist, and antagonist to be distinguished in respect to coactivator recruitment. Furthermore, we demonstrate that coactivators SRC-1 and ACTR utilize different motifs within their receptor interaction domains for association with the ternary complex. Our results support the hypothesis that the target for these nuclear receptor ligands is the RAR·RXR·DNA ternary complex.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

DNA Constructs-- The plasmids pRS-RXR (19), RAR-VP-16, ER-RXR, ERE-tk-luciferase (16), and pBKS-N-CoR-C' (2) have been previously described. The plasmids pGEM3Z-hRAR5', pGEM-hRAR5', and pGEM3Z-hRAR5', containing full-length cDNAs, were kindly provided by Dr. Bill Lamph, Ligand Pharmaceuticals. pcDNA3.1-hRAR-V5, was constructed by replacing the RAR stop codon with a XbaI site in the plasmid pGEM3Z-hRAR5' using PCR, followed by insertion of the EcoRI-XbaI fragment into the plasmid pcDNA3.1-V5/HisA (Invitrogen). The plasmid pcDNA3.1-RAR-V5 was constructed by replacing the RAR stop codon with a XbaI site in the plasmid pGEM-hRAR5' using PCR, followed by insertion of the SacI-XbaI fragment into the EcoRV and XbaI sites of the plasmid pcDNA3.1-V5/HisA. pcDNA3.1-hRAR-V5 was constructed via insertion of the KpnI-ApaI fragment of the plasmid pGEM3Z-hRAR5' into the plasmid pcDNA3.1-V5/HisB (Invitrogen), followed by insertion of the C-terminal portion of the RAR cDNA via the incorporation of an ApaI site at the RAR stop codon using PCR. For RAR-VP-16(AHT), amino acids 223, 224, and 227 (Ala, His, and Thr) of RAR were converted to Gly, Gly, and Ala, respectively, by PCR mutagenesis within the SacI and EcoRV sites. After substitution of this fragment into RAR-VP-16, the VP16 and RAR N-terminal domains were reintroduced using the SacI fragment of RAR-VP-16. For pAS-N-CoR, PCR mediated introduction of EcoRI sites into the N-CoR cDNA, adjacent to the initiator methionine codon and the ApaI site located at nucleotide position 403 (GenBank accession number U35312), allowed insertion of the N-CoR N termini in-frame with the Gal4 DNA-binding domain (DBD) in plasmid pAS2-1 (CLONTECH Laboratories). The C-terminal encoding portion of N-CoR cDNA was inserted using the internal ApaI site and the SalI site located in the polylinker of pAS2-1. pAS-RXR was made via insertion of EcoRI sites into the mouse RXR cDNA by PCR and insertion of this fragment, encoding the ligand-binding domain (amino acids 147-410), in-frame with the Gal4 DBD of the plasmid pAS2-1 (CLONTECH). pACT-RAR was constructed via PCR mediated introduction of BamHI sites next to the start and stop codons of the human RAR cDNA followed by insertion into the BamHI site of the plasmid pACT2 (CLONTECH). pACT-RAR (AHT) was derived from pACT-RAR via the substitution of the relevant SacI and EcoRV cDNA fragment from RAR-VP16-AHT. Vectors for bacterial expression of glutathione S-transferase (GST)-RAR (amino acids 1-454) and -RXR (amino acids 1-462) fusion proteins were generated via PCR mediated introduction of NotI and BamHI (RAR) or NotI and SmaI (RXR) restriction sites into receptor cDNAs and insertion in-frame with the GST moiety of the plasmid pGEX-KN (20).

Proteolytic Protection Analysis of in Vitro Translated RARs-- Limited proteolytic digestion of RARs was carried out as described previously (21). ³⁵S-Labeled RARs were generated using the T-N-T transcription-translation system (Promega) programmed with pGEM3Z-hRAR5', pGEM-hRAR5', or pGEM3Z-hRAR5'. 3 µl of programmed lysate was mixed with 5 µl of buffer A (8 mM Tris-HCl, pH 7.4, 20 mM KCl, 4 mM dithiothreitol, and 8% glycerol) and 1 µl of retinoid (1 µM final) in vehicle (ethanol) or vehicle alone. After incubating on ice for 1 h, 1 µl of trypsin (250, 500, or 1000 µg/ml) was added and the lysates were incubated for 10 min at room temperature. After addition of loading buffer, samples were electrophoresed in an 8% polyacrylamide-SDS gel and visualized by autoradiography.

Electrophoretic Mobility Shift Analysis (EMSA)-- In vitro translated hRAR, -, and - (see above) and mRXR, programmed from pBSK-RXR (19) were generated using the T-N-T system (Promega). 2 µl of RAR and 2 µl of RXR lysates were preincubated in a total of 20 µl of 1 × EMSA buffer (20 mM Hepes, pH 7.8, 80 mM KCl, 1 mM dithiothreitol, 0.1% Nonidet P-40, 6% glycerol) containing 1 µg of poly(dI-dC:dI-dC). One hour after addition of ligand (1 µM final) a ³²P-labeled RARE (5'-agctttcaggtcaccaggaggtcagaa-3') was added. After 30 min, samples were run on a 6% polyacrylamide gel containing 0.5 × TBE and subjected to autoradiography.

Transfections-- Analysis of ligand regulation of RAR-VP-16 and RAR-VP-16(AHT) was performed as described previously (16). 4 × 10⁵ CV-1 cells per well of a 12-well plate (Costar) were transiently transfected via calcium phosphate precipitation (22) with 0.5 µg of pERE-tk-luciferase (containing the estrogen-regulated element of the Xenopus vittelogenin A2 gene (23) inserted into the plasmid tk-luciferase (24)), 0.1 µg of pCH110, 0.1 µg of ER-RXR expression vector, and 0.2 µg of the chimeric expression vector RAR-VP-16 or RAR-VP-16(AHT). ER-RXR contains the hormone-binding domain (amino acids 181 to 458) of RXR (19) fused downstream from the estrogen receptor A/B and DNA-binding domains (25). Eighteen hours after introduction of the DNA precipitants, cells were rinsed with phosphate-buffered saline (PBS) and fed with Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% activated charcoal-extracted fetal bovine serum (Gemini Bio-Products). Cells were treated for 18 h with the indicated retinoids. After rinsing with PBS cells were lysed and luciferase activity was measured as described previously (26). For whole cell extracts, CV-1 cells were cultured with Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% activated charcoal-extracted fetal bovine serum (Gemini Bio-Products) before transfection. At a density of 40~60% (15-cm plate, Falcon), cells were transiently transfected with 15 µl of FuGene 6 Transfection Reagent (Roche Molecular Biochemicals) with 0.5 µg of pRS-RXR, and 5 µg of pcDNA3.1-hRAR-V5, pcDNA3.1-hRAR-V5, or pcDNA3.1-hRAR-V5 per plate. After 2 days, cells were rinsed twice with PBS and lysed in cold NET buffer (20 mM Tris-Cl, pH 8.0, 200 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 10% glycerol) containing protease inhibitors, homogenized by QIAshredder (Qiagen), and clarified by centrifugation.

GST Pull-down Analyses-- Expression and purification of GST fusion proteins was performed in BL-21 bacteria as described previously (27). ³⁵S-Radiolabeled N-CoR-C' (amino acids 1629-2453) and RAR were produced using the T-N-T-coupled transcription-translation system (Promega) programmed with pBKS-N-CoR-C' and pGEM3Z-hRAR5', respectively. After binding to glutathione-agarose beads (Sigma), GST fusion proteins were resuspended in 50 µl of CHAPS buffer (8 mM Tris phosphate, pH 7.4, 120 mM KCl, 8% glycerol, 4 mM dithiothreitol, 0.5% CHAPS (Calbiochem)) to which 2 µl of retinoid in vehicle or vehicle alone (1 µM final), 2 µl of 10% bovine serum albumin, and 145 µl of CHAPS buffer was added. After incubation on ice for 30 min, 1 µl of in vitro translated ³⁵S-N-CoR-C' was added and tubes were incubated for 20 min at 20, 30, or 37 °C. After washing 3 times with 1 ml of cold CHAPS buffer, bound proteins were eluted in SDS-polyacrylamide gel electrophoresis loading buffer, electrophoresed through a 8% polyacrylamide gel, and visualized by autoradiography. Quantitation of bound proteins was performed using a Molecular Dynamics PhosphorImager.

Yeast Two-hybrid Analyses-- pAS2-1 and pACT2 based vectors (CLONTECH) were transformed into Y190 yeast using the instructions provided by the manufacturer. -Galactosidase activity was measured, using o-nitrophenyl--D-galactopyranoside (Sigma) as a substrate (28), in freeze-fractured extracts of yeast grown overnight at 30 °C in the presence or absence of the indicated retinoids (1 µM final).

Immunoprecipitations and Western Analyses-- 1 mg of transfected CV-1 whole cell extract was used for each immunoprecipitation. Cell lysates were incubated with retinoids on ice for 1 h. Where indicated, annealed double-strand oligonucleotides (DR-5 RARE: 5'-agctttcaggtcaccaggaggtcagaa-3'; G-5-G mutant RARE: 5'-agcttagagaacaccgaaagaacacta-3') were added prior to ligand addition and incubated on ice for 30 min. After 1-h incubation on ice with primary antibody (mouse anti-V5, Invitrogen), Protein G-agarose (Sigma) was added and samples were rocked overnight at 4 °C. After washing with ice-cold NET buffer, immunoprecipitants were resolved on SDS-polyacrylamide gels (4-12%) followed by Western blotting. Membranes were probed with the indicated antibodies in PBS-T buffer (PBS with 0.1% Tween 20) containing 5% nonfat dry milk, and washed in PBS-T buffer. Primary antibodies were obtained from Santa Cruz Biotech (number SC553, rabbit anti-RXR; number SC1609, goat anti-N-CoR) Upstate Biotechnology (number 05-490, mouse anti-ACTR) and Affinity BioReagents (number MA1-840, mouse anti-SRC1). Where indicated, ACTR peptides (LXD1, LESKGHKKLLQLLTCSSDDRGH; LXD2, LLQEKHRILHKLLQNGNSPAEV; LXD3, KKKENNALLRYLLDRDDPSDAL) or SRC-1 peptides (LXD1, KYSQTSHKLVQLLTTTAEQQLR; LXD2, SLTERHKILHRLLQEGSPSDIT; LXD3, KESKDHQLLRYLLDKDEKDLRS) were added before the addition of retinoids to a final concentration between 20 and 40 µM.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

AGN193840 and AGN193109 are potent antagonists of ATRA at RAR (16). Differing from AGN193840 only by the substitution of a methyl substituent for a fluorine on the phenyl ring (see Fig. 1A), AGN193109 is distinguished by its ability to repress the elevated basal activity of a chimeric receptor, RAR-VP-16, containing the transactivation domain of the herpes simplex viral protein VP-16 (16), as well as its agonist-like effect on the expression of a subset of retinoid-sensitive genes in primary human keratinocytes (16, 29). Antagonists for various nuclear receptor family members have been shown to induce receptor conformations distinct from those induced by agonist using limited protease digestion (21, 30, 31). As has been previously demonstrated, binding of the RAR selective antagonist Ro-41-5253 (21) to in vitro translated RAR followed by trypsin digestion results in a protected fragment 26 kDa in size compared with a 30-kDa species conferred by the natural hormone ATRA (Fig. 1B). In contrast, binding of the inverse agonist AGN193109 to all three RARs results in a proteolytic protection pattern equivalent to that produced by ATRA (Fig. 1B). Similar analyses using the neutral antagonist AGN193840 gave identical results as AGN193109 (data not shown). However, the RAR conformations induced by AGN193109 and ATRA must be distinct due to the inability of AGN193109 to trans-activate RARs (17) or to mediate coactivator interaction with the RAR in yeast two-hybrid studies (data not shown) or in co-immunoprecipitations from cellular extracts (see below).

View larger version (35K): [in this window] [in a new window]   Fig. 1.   A, structures of AGN193840 and AGN193109. Previously described (16) relative affinities (Kd, nM) and antagonist activity (IC50, nM) for the RARs are as follows. For RAR, AGN193840 Kd = 85 ± 45, IC50 = 42 ± 17; and AGN193109 Kd = 16 ± 5, IC50 = 9 ± 1. For RAR, AGN193840 is a partial agonist Kd = 52 ± 30; and AGN193109 Kd = 7 ± 3, IC50 = 24 ± 3. For RAR, AGN193840 Kd = 82 ± 35, IC50 = 22 ± 12; and AGN193109 Kd = 7 ± 1, IC50 = 5 ± 1. B, partial proteolytic digestion of in vitro translated RARs in the presence of the inverse agonist AGN193109 does not distinguish it from the agonist ATRA. RARs were digested with increasing concentrations (black ramps) of trypsin as described under "Experimental Procedures" in the presence of 1 µM ATRA, Ro-41-5253, AGN193109, or vehicle alone. Far left lanes contain 14C-labeled high molecular weight standards (Life Technologies, Inc.). Undigested RARs (lanes 0), are indicated by the arrows. Proteolytic resistant fragments are indicated by asterisks.

Dimerization of the RAR with its heterodimeric partner RXR is required for efficient DNA binding and transactivation (32). Addition of either AGN193840 or AGN193109 did not inhibit RAR/RXR binding to a RARE using EMSA (Fig. 2A). To further address this question we analyzed RAR-RXR heterodimerization in a yeast two-hybrid system consisting of a Gal4 (DBD)-RAR fusion protein and a Gal4 activation domain (AD)-RXR fusion protein. -Galactosidase activity in yeast containing both of these expression plasmids is increased over that in yeast having only the Gal4(DBD)-RAR plasmid, demonstrating the heterodimerization between the RAR and RXR moieties. Neither agonist (TTNPB), AGN193109, nor AGN193840 addition altered this degree of RAR-RXR interaction (Fig. 2B). Thus, binding of these retinoids to the RAR does not disrupt heterodimer formation or DNA binding.

View larger version (49K): [in this window] [in a new window]   Fig. 2.   RAR antagonists AGN193840 and AGN193109 do not disrupt RAR-RXR heterodimerization or DNA binding. A, EMSA analysis of RAR (lanes 2, 5, and 8), RAR (lanes 3, 6, and 9), and RAR (lanes 4, 7, and 10) heterodimerized to RXR bound to a 32P-labeled RARE after incubation with vehicle (ethanol) or vehicle containing AGN193840 or AGN193109 (1 µM final). Migration of bound RAR/RXR and free probe are indicated by the arrows. B, -galactosidase activity of Y190 yeast containing pAS-RXR alone or together with pACT-RAR were determined after growth in the presence of the indicated retinoid (1 µM final).

The existence of corepressor molecules (1, 2) which interact with the RAR, and other members of the nuclear receptor superfamily, offer a possible explanation of the inverse agonism seen for ligands such as AGN193109. To confirm that the trans-repressive effect of an inverse agonist upon the activity of RAR-VP-16 is due to recruitment of a corepressor to the RAR, we mutated the corepressor (CoR) box in the RAR-VP-16 context. Mutations in this domain, located in the hinge region immediately C-terminal to the DNA-binding domain, result in loss of functional N-CoR binding to RAR (2). While AGN193109 treatment leads to a dose-dependent trans-repression of RAR-VP-16 activity, mutation of the CoR box in RAR-VP16-AHT results in basal activity which is refractory to this inverse agonist (Fig. 3A). This requirement of a functional CoR box for inverse agonist activity is consistent with recruitment of corepressor to the RAR upon binding AGN193109.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Recruitment of N-CoR to RAR by the antagonist/inverse agonist AGN193109. A, repression of RAR-VP-16 transcriptional activation by the RAR inverse agonist AGN193109 requires a functional CoR box. CV-1 cells transfected with either RAR-VP-16 (closed squares) or RAR-VP-16 (AHT) (closed triangles) were treated with AGN193109 as described under "Experimental Procedures." 100% activity equals the mean luciferase activity ± S.E. of triplicate determinations normalized to -galactosidase activity, in extracts prepared from vehicle only treated cells. B, interaction of N-CoR with RAR is increased by AGN193109. 35S-Labeled N-CoR was incubated with GST-RAR in the presence of vehicle alone or vehicle with 1 µM AGN193109 or TTNPB at the indicated temperatures. Ligand mediated change in N-CoR retention (fold effect) relative to vehicle is indicated below each lane. For comparison, interaction of 35S-labeled RAR with GST-RXR was also performed (fourth lane from the left). See "Experimental Procedures" for details. C, -galactosidase activity of Y190 yeast containing the plasmids pAS-N-CoR and pACT-RAR (gray columns) treated with vehicle (CTL) or the indicated ligands (1 µM final). Black columns represent -galactosidase activity in Y190 yeast containing pAS-N-CoR and pACT-RAR-(AHT) treated as indicated.

In order to analyze ligand modulation of RAR-N-CoR interaction, in vitro translated N-CoR was incubated with a GST-RAR fusion protein immobilized to glutathione-agarose (Fig. 3B). When incubations were performed at 20 or 30 °C, TTNPB treatment led to decreased interaction between N-CoR and GST-RAR relative to vehicle alone. In contrast AGN193109 treatment resulted in an increased interaction at both incubation temperatures. When incubations were performed at 37 °C, the detectable interaction in the absence of added ligand was considerably reduced. However, the addition of AGN193109 led to a robust increase in the interaction between N-CoR and GST-RAR. Similar analysis using the neutral antagonist AGN193840 did not indicate it to be different than AGN193109 in its ability to increase RAR-N-CoR interaction (data not shown). Consistent with these findings, treatment of yeast expressing full-length Gal4(DBD)-N-CoR and Gal4(AD)-RAR fusion proteins with AGN193109 resulted in an increase in -galactosidase activity (Fig. 3C). Mutation of the CoR box in Gal4(AD)-RAR(AHT) abrogated this effect. Therefore, binding of this RAR inverse agonist to RAR leads to a receptor conformation which has an increased affinity for the corepressor. Similar to our observations in GST pull-down studies (above), the neutral antagonist AGN193840 was essentially equivalent to AGN193109 in its ability to increase the interaction between RAR and N-CoR in the yeast two-hybrid system (Fig. 3C).

We wished to compare the ability of AGN193840 and AGN193109 to modulate N-CoR interaction with RAR in the context of the RAR/RXR heterodimer. While initial studies demonstrated a transfected V5-epitope-tagged RAR to be well expressed in CV-1 cells, we were unable to detect co-immunoprecipitated endogenous N-CoR in the presence or absence of AGN193109 (data not shown). This was also the case for cells co-transfected with both RAR-V5 and RXR (Fig. 4B, lane 1). In contrast, addition of TTNPB to lysates prepared from CV-1 cells transfected with RAR-V5 alone (data not shown) or together with RXR (Fig. 4A) resulted in co-immunoprecipitation of the coactivator SRC-1. However, this agonist mediated SRC-1-RAR interaction was further increased by the inclusion of a synthetic double stranded oligonucleotide containing a DR-5 RARE in the immunoprecipitation. This RARE mediated increase was concentration dependent (Fig. 4A, lanes 2-5) and was specific for the DR-5 RARE; a mutated DR-5 in which the RARE half-sites has been modified to that of a glucocorticoid receptor half-site, G-5-G, did not result in an increase in SRC-1 co-immunoprecipitation (Fig. 4A, lane 7). Analysis of RXR in these immunoprecipitates indicated that the DR-5 mediated increase in RAR-SRC-1 interaction was associated with a parallel increase in immunoprecipitated RXR. Due to this apparent RARE/heterodimer increase in SRC-1 interaction with RAR, we analyzed AGN193840 and AGN193109 for their ability to recruit N-CoR to the RAR-V5/RXR heterodimer in the presence of the DR-5 RARE (Fig. 4B). Under these conditions, N-CoR association with RAR was negligible both in the absence of ligand (Fig. 4B, lane 2) and after treatment with the RAR agonist TTNPB (lane 5). In contrast, N-CoR association with the ternary complex was detectable in the presence of 193109 (Fig. 4B, lane 4). In comparison, the amount of N-CoR association after treatment with 193840 (Fig. 4B, lane 3) was reduced to only 43% relative to 193109 treatment. As such, AGN193840 exhibits partial inverse agonism at the RAR·RXR·DR-5 ternary complex relative to AGN193109. Detection of AGN193109-mediated N-CoR-RAR interaction required the presence of a DR-5 RARE as use of a related G-5-G DNA element (Fig. 4B, lane 1) failed to do so. Analysis of N-CoR association with RAR and RAR containing ternary complexes further substantiated the difference in corepressor recruitment capability of the above two RAR antagonists. However, for these RAR isoforms, the difference between these antagonists was more qualitative than that demonstrated for RAR. In contrast to that shown for RAR, the amount of N-CoR co-immunoprecipitated with unliganded RAR was readily detectable (Fig. 4C, lane 1). Expression levels of the V5-epitope-tagged RAR and RAR are comparable (data not shown) and do not account for this apparent difference between isoforms for N-CoR association in the unliganded state. This level of RAR-N-CoR interaction was further increased by AGN193109 but not by AGN193840. Rather, 193840 treatment resulted in a small decrease in N-CoR association as compared with the unliganded receptor. As previously demonstrated, the synthetic agonist TTNPB resulted in abrogation of RAR-N-CoR association. For RAR, N-CoR association with the unliganded ternary complex was not obvious (Fig. 4C, lane 5). Only treatment with AGN193109 led to an increase in N-CoR co-immunoprecipitation with the RAR containing ternary complex. Under identical conditions the neutral antagonist AGN193840 did not increase N-CoR recruitment to RAR. Results of multiple N-CoR recruitment analyses at RAR, - and -containing ternary complexes are shown in Fig. 4D. Thus, the antagonists AGN193109 and 193840 confer distinct corepressor recruitment capabilities to the RARs in the ternary complex. We propose that the DNA dependent increase in RAR/RXR heterodimerization and the resulting increase in ligand-mediated coregulator recruitment in vitro reflects an apparent increase in the affinity of coregulators for the ternary complex as compared with isolated RAR. Under these conditions, RAR inverse agonists and neutral antagonists can be differentiated quantitatively.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   A DR-5 RARE mediates an increase in RAR-RXR heterodimerization and ligand recruitment of coregulators. A, RAR was immunoprecipitated from transfected CV-1 whole cell extracts in the absence (, lanes 1 and 6) or presence (+, lanes 2-5 and 7) of 1 µM TTNPB and increasing amounts of a DR-5 RARE (0-0.5 µg). For comparison, a mutated DNA element representing two glucocorticoid receptor half-sites separated by 5 base pairs (G-5-G) was used (lane 7). Omission of DNA resulted in TTNPB-induced SRC-1 detection identical to that of lane 7 (data not shown). Anti-SRC-1 and RXR antibodies were used to detect co-immunoprecipitated SRC-1 and RXR, respectively. Molecular weight standards are indicated on the left. B, RAR was immunoprecipitated (lanes 1-5) from transfected CV-1 whole cell extracts in the presence (+) of the indicated ligands (1 µM final) and in the presence of 0.5 µg of the DR-5 RARE (lanes 2-5) or G-5-G (lane 1). CV-1 whole cell extract (50 µg) was run in lane 6. Anti-N-CoR and anti-RXR antibodies were used to detect recruitment of N-CoR and heterodimerization with RXR, respectively. N-CoR recruitment by 193840 (lane 3) was 46% relative to that recruited by 193109 (lane 4). See D for results of multiple experiments. C, RAR (lanes 1-4) and RAR (lanes 5-8) were immunoprecipitated from transfected CV-1 whole cell extracts in the presence (+) of the indicated ligands (1 µM final) and 0.5 µg of the DR-5 RARE. Anti-N-CoR and anti-RXR antibodies were used to detect recruitment of N-CoR and heterodimerization with RXR, respectively. N-CoR recruitment to the RAR containing ternary complex by 193840 (lane 2) was 41% relative to that recruited by 193109 (lane 3). For RAR, N-CoR coimmunoprecipitation required treatment with AGN193109 (lane 7). See D for results of multiple experiments. D, N-CoR recruitment by the indicated ligands was measured at RAR, -, and - containing ternary complexes (RAR·RXR·DR5). Values represent the mean (n = 3) ± S.E. of independent immunoprecipitations where the amount of N-CoR recruited by 193109 = 100% recruitment. ND, not detectable.

AGN193840 and -193109 can be further distinguished with regard to coactivator recruitment at RAR. While AGN193109 lacks any detectable agonist activity at the three RARs, AGN193840 exhibits partial agonist activity at RAR selectively (16). As shown in Fig. 5A, AGN193840 treatment activates the RAR/RXR heterodimer with a maximal efficacy of 31% compared with that of the agonist ATRA. Treatment with AGN193109 failed to transactivate this heterodimer. We next determined whether measurement of coactivator recruitment could account for these results. To that end, we utilized the DNA modified coimmunoprecipitation method to measure SRC-1 recruitment to RAR·RXR·DR5 complex after treatment with these ligands. As shown in Fig. 5A, the amount of Src-1 recruited by AGN193840 was 5% of that recruited by ATRA, in agreement with the partial agonist identity of this ligand at the RAR -isoform. AGN193109 did not recruit SRC-1. Thus, a partial agonist exhibits diminished SRC-1 recruitment to the ternary complex relative to that seen with a full agonist.

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Analysis of coactivator recruitment to RAR·RXR·DR5 ternary complexes. A, comparison of the transactivation and SRC-1 recruitment properties of ATRA and AGN193840 at RAR. CV-1 cells were co-transfected with pcDNA3-RAR-P-GR, pRS-RXR, the reporter plasmid MTV4X(R5G)-luciferase, and the -galactosidase expression plasmid pCH110 as described previously (16). Cells were treated with 1 µM of the indicated ligands. 100% activity equals the mean luciferase activity ± S.E. of triplicate determinations normalized to -galactosidase activity, in extracts prepared from ATRA-treated cells. For analysis of SRC-1 recruitment to the RAR·RXR·DR-5 complex, whole cell lysates prepared from CV-1 cells transfected with pcDNA-RAR-v5 and pRS-RXR were treated with 1 µM of the indicated ligands. Immunoprecipitations and detection of SRC-1 were performed as outlined under "Experimental Procedures." B, differential utilization of LXDs by SRC-1 and ACTR at the same ternary complex. RAR (lanes 1-5) or RAR (lanes 6-9) were immunoprecipitated from transfected CV-1 cells in the absence (lane 1) or presence (lanes 2-9) of 1 µM ATRA as described under "Experimental Procedures." Where indicated +, LXD peptides specific to either SRC-1 or ACTR were included. The resulting effect on SRC-1 (top) and ACTR (bottom) recruitment is shown.

The DNA modified coimmunoprecipitation procedure also allows measurement of coactivators other than SRC-1. ACTR is an additional p160 coactivator family member which, similar to SRC-1, has been shown to interact with a variety of nuclear receptor family members. As shown in Fig. 5B (lane 2), treatment with the RAR agonist ATRA results in recruitment of endogenously derived ACTR (bottom) and SRC-1 (top) to the RAR·XR·DR5 ternary complex. Furthermore, this method can be used to distinguish these related coactivator molecules in their mode of interaction with these receptors. The domains within these coactivators which are required for interaction with nuclear receptors have been mapped to a central receptor interacting region containing three "LXXLL" domains (LXDs) (10, 33). These LXDs, exhibit an amphipathic helical structure and make direct contact with the coactivator interaction domain of the receptor which is formed upon binding agonist (34). We addressed the importance of each of the three LXDs in mediating coactivator-receptor interaction via competition with synthetic peptides specific for LXD1, LXD2, or LXD3 in the DNA modified immunoprecipitation. As shown in Fig. 5B, ATRA mediated recruitment of SRC-1 to the RAR·RAR·DR5 ternary complex (lanes 2-5) required both LXD2 and LXD3 as it was completely abolished by addition of SRC-1 peptides specific for LXD2 or LXD3, whereas the LXD1 peptide had only a weak effect. In contrast, ACTR recruitment required both LXD1 and LXD2 as it was abolished by ACTR peptides LXD1 or LXD2 but not LXD3. This difference in LXD utilization by SRC-1 and ACTR was also observed for the RAR·RXR·DR5 ternary complex (Fig. 5B, lanes 6-9). Thus, while the receptor interacting domains of SRC-1 and ACTR are well conserved, with the LXD1-LXD3 region exhibiting 52% similarity (including conservative amino acid changes), these coactivator molecules utilize distinct LXD interfaces for interaction with identical receptor complexes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The prediction that ligand activation of nuclear receptors is mediated by coactivator protein recruitment was based on numerous mutation analyses and mapping of the AF-2 domains of nuclear receptors (35-38). While this has been subsequently proven with the cloning of a variety of nuclear receptor coactivator proteins (reviewed in Ref. 39), the identification of the corepressor molecules N-CoR (2) and SMRT (1) has further increased the complexity involved in ligand modulation of these transcription factors. Originally identified via their interaction with retinoic acid and thyroid hormone receptors, these negative cofactors have been proposed to account for the repressive effect of unliganded receptors on reporter constructs (40, 41). Corepressor interaction has now been reported for ER (42, 43) and PPAR (44) as well as for other diverse transcription factors such as Pit-1(45), Msx-1 (46), and Notch (47). Notably, two of these reports (43, 45) demonstrated N-CoR-ER association to be antagonist dependent. Therefore, it is perhaps more accurate to classify antagonists such as these as inverse agonists to distinguish them from antagonists which do not recruit corepressor. A recent example of the latter is the PPAR partial agonist/antagonist (GW0072) which does not recruit corepressor (44). Androstene and androstendione function as naturally occurring inverse agonists of the constitutively active nuclear receptor CAR although interaction of this receptor with corepressor has yet to be demonstrated (15).

Mechanisms which could account for the action of a RAR antagonist include disruption of RXR heterodimerization and/or DNA binding, blockade of coactivator interaction and recruitment of corepressor. While AGN193840 and AGN193109 do competitively displace [³H]ATRA from the RARs (16, 17) they do not alter RAR/RXR heterodimerization or binding to a RARE DNA element in a gel shift assay. Furthermore, heterodimerization of RAR and RXR is not altered by these ligands in a yeast two-hybrid system. Analysis of the induction of a proteolytic-resistant RAR polypeptide fragment by these antagonists failed to detect a conformation change distinct from that mediated by ATRA. A similar finding has been reported for other RAR antagonists (48). This is in contrast to earlier reports using the progesterone antagonist RU-486 (30), the RAR-specific antagonist Ro-41-5253 (21), as well as various ER antagonists (31). While AGN193840 and AGN193109 are not distinguished from ATRA in this respect, they do, however, induce specific conformation changes as evidenced by the increase in resistance to proteolytic digestion compared with vehicle alone. Thus, rather than simply displacing agonist, binding of these antagonists results in a change in RAR conformation. An antagonist-specific conformation change has been demonstrated in crystallographic analyses of the ER in the presence of tamoxifen (49). In this case, antagonist binding results in conformation changes within the ER ligand-binding domain which are similar to those induced by agonist, but which result in blockade of the coactivator interaction domain.

Antagonists for several G-protein-coupled receptors have been categorized either as neutral antagonists or inverse agonists based upon their ability to inhibit basal receptor activity (11). By analogy, we previously designated AGN193109 and AGN193840 as RAR inverse agonist and neutral antagonist, respectively, based on their differing abilities to repress the elevated basal transcriptional activity of RAR-VP-16 (16). Specifically, while the degree of repression mediated by AGN193109 was greater than that of 193840, the latter did repress RAR-VP-16 to a small extent and in retrospect can be characterized as a partial or incomplete inverse agonist at RAR. While these antagonists lead to an increase in the association of N-CoR with RAR in the present GST pull-down and yeast two-hybrid interaction studies, our detection of endogenous N-CoR association with immunoprecipitated RAR in cell extracts required the addition of a DR-5 RARE DNA element. Inclusion of this RARE results in a significant increase both in RXR heterodimerization as well as in ligand (agonist as well as antagonist) mediated recruitment of coregulators to the RAR. These findings suggest the RAR is subject to allosteric modification upon its association into the heterodimeric complex bound to DNA. A similar structural modification has been proposed for RXR in which its association with RAR on DR-5 elements results in allosteric inhibition of ligand binding (50). These findings and those of Mouchon et al. (51) suggest that analyses of ligand induced receptor-cofactor interactions in the absence of the complete functional unit, the RXR-RAR heterodimer bound to DNA, may be incomplete. Using this DNA-modified immunoprecipitation technique, analysis of corepressor association with RAR/RXR bound to DNA consistently demonstrated a greater N-CoR recruitment with the inverse agonist AGN193109 compared with the neutral antagonist AGN193840. This difference provides an explanation for the relatively poor efficacy of AGN193840 in repressing the transcriptional activity of RAR-VP-16 compared with AGN193109 (16). AGN193109 is further distinguished from 193840 in that only the former recruited N-CoR to immunoprecipitated RAR/RXR and RAR/RXR heterodimers bound to a DR-5 RARE element. Thus, AGN193019 is an inverse agonist at all three RARs. In contrast, AGN193840 is a neutral antagonist at RAR, a partial agonist at RAR and a partial inverse agonist at RAR. Whether it may be possible to further differentiate AGN193840 and AGN193109 with the use of different DNA elements and/or heterodimeric partners remains to be determined. The DNA modified coimmunoprecipitation recruitment procedure also allows quantitative analysis of coactivator recruitment by partial agonists, correlating with their distinct transactivation properties. As such we demonstrate that the RAR selective partial agonist AGN193840 has reduced capacity, relative to ATRA, to recruit SRC-1 to the RAR·RXR·DR5 ternary complex.

The DNA modified coimmunoprecipitation procedure allows simultaneous measurement of ligand-mediated recruitment of different coactivator molecules to the ternary complex. The significance of the interaction of different members of the p160 family of coactivators to a given RAR is at this time not well understood. Using LXD specific peptides, we were able to show the requirement of LXD2 and LXD3 for agonist-mediated SRC-1 interaction. This result is in agreement with the peptide microinjection studies of McInerney et al. (52). Interestingly, interaction of ACTR with the same ternary complex also utilizes LXD2 but, in contrast to that of SRC-1, in combination with LXD1. The spacing between LXD domains may play a role in this selective use of these amphipathic helices. In SRC-1, the sequence between LXD 1-2 and LXD 2-3 is 52 and 54 amino acids in length, respectively. In ACTR, these intra-LXD sequences are 59 and 48 amino acids in length, respectively. Thus, the ACTR LXD2-3 and the SRC-1 LXD1-2 regions may be too short for interaction at the ternary complexes tested. Alternatively, acidic and basic residues in or nearby the LXD domains may play a role. Consistent with its common utilization for both of these coactivators, the arrangement of charged residues surrounding LXD2 (from position 16 to +15) is well conserved between ACTR and SRC-1. However, in comparing LXD1 domains, there is a 14-amino acid insert between a conserved group of acidic residues and the LXXLL motif in ACTR which is absent in SRC-1. Juxtaposition of these acidic residues closer to the core LXXLL in the SRC-1 LXD1 may play a active role in disallowing the use of this LXD. This utilization of different interaction domains among these two conserved members of the p160 family of coactivators begins to illuminate subtle differences which could have a consequence for the association of other molecules with the complex. As such, ACTR recruited to the ternary complex would have LXD3 available for interaction with another molecule while the availability of LXD1 in SRC-1 could result in an interaction with a different cofactor. Thus, an ACTR-associated ternary complex may be functionally distinct than one associated with SRC-1.

The observed negligible detection of N-CoR association with immunoprecipitated ternary complexes containing RAR or RAR, in the absence of retinoids, was surprising given that such association was readily apparent in GST-RAR pull-down analyses. While this result is somewhat in contradiction to proposed models of corepressor-RAR interaction, it is in contrast to our results using RAR, the RAR isoform used in previous reports (1, 2). The requirement of DNA and RXR for RAR-N-CoR interaction likely reflects an allosteric modification in the RAR affording the necessary tertiary structure for corepressor interaction. Under these conditions, the antagonist dependence for N-CoR association for RAR, but not for RAR, indicates that different RAR isoforms have different set points for corepressor interaction. Similarly, our detection of N-CoR-RAR association was dependent on the presence of the inverse agonist AGN193109. Analogous results have been described for the ER in which association with N-CoR, in the absence of added DNA, is dependent on the presence of antagonist (43, 45). Whether different heterodimeric partners or different DNA response elements can alter the set point for corepressor interaction remains to be tested.

The existence of endogenous nuclear receptor ligands with corepressor recruiting activities is an intriguing possibility. Alternatively, corepressor molecules may represent a means to fine tune receptor activity as well as to interconnect different nuclear receptor pathways and the ability of certain synthetic ligands of these receptors to increase corepressor recruitment may be a pharmacological phenomenon. Thus, interaction between RAR and corepressor can be modulated by ligands in both a negative (agonist) as well as a positive (inverse agonist) manner. Analogous to a recent report of distinct conformation changes in ER and ER by different ER modulators (53), our findings indicate that RAR antagonists can be designed which vary in their ability to actively recruit corepressor. As such, the opportunity exists for the design of a spectrum of ligands with varying abilities to recruit cofactor molecules and which, as a result, have different biological activities.

	ACKNOWLEDGEMENTS

We thank Dr. Michael G. Rosenfeld and Dr. Andreas Horlein for their generous gift of the N-CoR cDNA and Mary Pino for help in performing the proteolytic protection analyses.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The 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: Mail code RD-3D, 2525 Dupont Dr., Irvine, CA 92715-9534. Tel.: 714-246-4895; Fax: 714-246-6207; E-mail: klein_elliott@allergan.com.

Published, JBC Papers in Press, April 20, 2000, DOI 10.1074/jbc.M002472200

	ABBREVIATIONS

The abbreviations used are: ATRA, all-trans-retinoic acid; RAR, retinoic acid receptors; N-CoR, nuclear receptor corepressor; SMRT, silencing mediator of retinoid and thyroid receptors; PCR, polymerase chain reaction; DBD, DNA-binding domain; GST, glutathione S-transferase; EMSA, electrophoretic mobility shift assay; RXR, retinoic X receptor; RARE, retinoic acid receptor element; ER, estrogen receptor; PBS, phosphate-buffered saline; LXD, LXXLL domain; CHAPS, 3- [(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid; TTNPB, {(E)-4-[2-(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthelenyl)-propen-1yl} benzoic acid.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore