Characterization of the Repressor Function of the Nuclear Orphan Receptor Retinoid Receptor-related Testis-associated Receptor/Germ Cell Nuclear Factor*

Retinoid receptor-related testis-associated receptor (RTR)/germ cell nuclear factor is a nuclear orphan receptor that plays an important role in the control of gene expression during early embryonic development and gametogenesis. It has been shown to repress transcriptional activation. In this study, we further characterize this repressor function. We demonstrate that RTR can suppress the transcriptional activation induced by the estrogen receptor related-receptor α1 through its response element. The latter is at least in part due to competition for binding to the same response element. In addition, RTR inhibits basal transcriptional activation, indicating that it functions as an active repressor. Mammalian two-hybrid analyses showed that RTR interacts with the co-repressor nuclear co-repressor (N-CoR) but is unable to interact with the co-repressor SMRT or RIP140. Pull-down analyses with glutathioneS-transferase-RTR fusion protein demonstrated that RTR physically interacts with N-CoR in vitro, suggesting a potential role for N-CoR in the transcriptional repression by RTR. To identify the regions in RTR essential for the binding of RTR to N-CoR, the effect of various deletion and point mutations on this interaction was examined. This analysis revealed that this interaction requires the hinge domain, helix 3 as well as the helix 12 region of RTR. The residues Ser246-Tyr247 in the hinge domain, Lys318 in helix 3, and Lys489-Thr490 in helix 12 are identified as being critical in this interaction. Our results demonstrate that RTR can function as an active transcriptional repressor and that this repression can be mediated through interactions with the co-repressor N-CoR. We show that this interaction exhibits several characteristics unique to RTR. Through its repressor function, RTR can suppress the induction of transcriptional activation by other nuclear receptors. These repressor activities may provide important mechanisms by which RTR regulates gene expression during development and gametogenesis.

The nuclear receptor superfamily constitutes a group of ligand-dependent transcriptional factors and a large number of orphan receptors whose ligands have not yet been identified (1)(2)(3). Nuclear receptors share a common modular structure composed of several domains that have functions in DNA binding, ligand binding, nuclear localization, dimerization, repres-sion, and transactivation (4,5). Typically, ligand binding induces a conformational change in the receptor causing dissociation of bound co-repressors, such as nuclear co-repressor (N-CoR) 1 or silencing mediator for retinoid and thyroid hormone receptors (SMRT), and the recruitment of co-activators (4,6). The latter leads to histone acetylation, changes in chromatin conformation, and subsequently to transactivation of target genes and changes in the biological functions of cells.
The orphan nuclear receptor, retinoid receptor-related testisassociated receptor (RTR), also named germ cell nuclear factor (NR6A1; Receptor Nomenclature Committee), has been cloned from mouse (7,8), human (9 -12), zebrafish (13), and Xenopus laevis (14). Sequence comparison has shown that RTR is highly conserved between species, suggesting functional conservation during evolution. RTR has an important role in embryonic development as well as in the adult. It is expressed in embryonic stem cells and differentially regulated during retinoidinduced differentiation of embryonal carcinoma and embryonic stem cells (9,15,16). During embryonic development, expression of RTR mRNA has been observed during several stages of neuronal development (14,17,18). Both RTR mRNA and protein have been detected in the placenta where its expression is restricted to trophoblasts (19). 2 The importance of RTR in development was further demonstrated by targeted disruption of the RTR gene (20). Mouse RTRϪ/Ϫ embryos died between days 10.5 and 11.5 of development and displayed open neural tubes while the critical link between chorion and allantois was not formed. In the adult, RTR expression is much more restricted, and RTR is most abundant in ovary and testis. In the ovary, it was found to be expressed in maturing oocytes before the first meiotic division (7). In the testis, RTR mRNA is differentially regulated during spermatogenesis and present in postmeiotic cells particularly in round spermatids (7,8,(21)(22)(23). These observations suggest a specific role for this receptor in the control of gene expression at this distinct stage of spermatogenesis. Protamine 1 and 2, which are induced in round spermatids, have been identified as potential target genes for RTR regulation (22,24).
RTR displays the common modular structure characteristic for nuclear receptors, but its helix 12 region is unusual in that it does not contain the consensus AF2 sequence ⌽⌽X(E/D)⌽⌽ (⌽ being a hydrophobic amino acid and X being a nonconserved amino acid) (25). RTR binds as a monomer or homodimer to RTR response elements (RTR-REs) consisting of the core motif AGGTCA and to direct repeats of this motif (DR0) (7, 16, 22, 26 -28). Transient transfection assays have indicated that RTR is able to repress basal transcriptional activity of a reporter gene under the control of RTR-REs (22,26,28).
In this study, we further characterized the transcriptional repressor function of RTR. We demonstrate that RTR can antagonize transcriptional activation by the estrogen receptor related-receptor ␣1 (ERR␣1) (29) likely through competition for the same response element. These results indicate that RTR may regulate biological processes by interfering with the transcriptional activation by other nuclear receptors. In addition, we show that RTR can function as an active repressor. To repress transcription RTR must communicate with the basal transcription apparatus either directly or indirectly via interaction with protein intermediates. We demonstrate that RTR is able to interact with the co-repressor N-CoR (30,31) but not with the co-repressor SMRT (32) or RIP-140 (33,34), suggesting a potential role for N-CoR in the transcriptional repression by RTR. The nature of the interaction between nuclear receptors and N-CoR has been reported to differ substantially between receptors (30,(35)(36)(37). To identify the regions and residues in RTR critical in the binding of RTR to N-CoR, we examined the effect of various deletion and point mutations on this interaction. Our study revealed that the hinge domain, helix 3, and the helix 12 region of RTR each are essential in the binding of RTR to N-CoR and demonstrates that the interaction of RTR with N-CoR exhibits several unique characteristics. We believe that these repressor activities will provide important mechanisms by which RTR control biological processes during development and gametogenesis.

EXPERIMENTAL PROCEDURES
Plasmids-The vector pSG5-VP16 containing the VP16 activation domain was obtained from Dr. J. Lehmann (Tularik Inc., San Francisco, CA). The expression plasmids pZeoSV-RTR encoding full-length mRTR and pSG5-VP16RTR encoding the VP16 (activation domain) fused to the full-length mRTR were described previously (22). The Gal4N-RIP13⌬N4 expression plasmid encoding ID-I and ID-II of RIP13/N-CoR was kindly provided from Dr. D. Moore (Baylor College of Medicine, Houston, TX) (31). The expression plasmid pcDNA3.1c-N-CoR encoding ID-I and ID-II of RIP13/N-CoR was created by cloning the EcoRI fragment of Gal4N-RIP13 into pcDNA3.1c (Invitrogen). The different pSG5-VP16RTR deletion mutants were created by placing the VP16 activation domain at the N terminus of various RTR fragments. These fragments were generated by PCR. RTR-specific 5Ј-and 3Ј-primers included either a KpnI or XhoI restriction site, respectively, to allow the PCR fragments to be subcloned into the KpnI and XhoI sites of pSG5-VP16. Details on the length of each deletion are described in the text and figures. pGEX-2TK-RTR, encoding the GST-RTR 149 -495 fusion protein, was constructed by inserting the RTR 149 -495 fragment, generated by PCR, into the BamHI and EcoRI sites of pGEX-2TK (Amersham Pharmacia Biotech). The integrity of all constructs was confirmed by restriction digestion and automatic DNA sequencing. The expression plasmid encoding ERR␣1 and the CAT reporter gene construct (ERR␣1-RE)CAT containing the ERR␣1 response element TCAAGGTCA were kindly provided by Dr. C. Teng (NIEHS, National Institutes of Health) (38,39).
Cell Culture-Chinese hamster ovary (CHO) and CV-1 cells were obtained from American Type Culture Collection and routinely main-tained in Ham's F-12 and Dulbecco's modified Eagle's medium, respectively, supplemented with 10% fetal bovine serum.
Mammalian Two-hybrid Analysis-CHO or CV-1 cells (2 ϫ 10 5 /well) were plated in six-well dishes and 20 h later transfected in Opti-MEM (Life Technologies, Inc.) with various expression and reporter plasmid DNAs as indicated in the legends using Fugene 6 transfection reagent (Roche Molecular Biochemicals). The plasmid ␤-actin-LUC or pCMV␤ was used as an internal control to monitor transfection efficiency. Cells were collected 48 h after transfection and assayed for CAT protein, luciferase, or ␤-galactosidase activity. The level of CAT protein was determined by the CAT enzyme-linked immunosorbent assay kit (Roche Molecular Biochemicals) according to the manufacturer's instructions. Luciferase activity was assayed with a Luciferase kit (Promega). ␤-Galactosidase activity was assayed with a Luminescent ␤-gal kit (CLON-TECH). Transfections were performed in triplicate, and each experiment was repeated at least two times.
Binding Assay Using GST Fusion Proteins-Escherichia coli JM109 cells transformed with pGEX-2TK-RTR or pGEX-2TK were grown at 37°C to mid-log phase, and the synthesis of GST proteins was then induced by the addition of isopropyl-␤-D-thiogalactopyranoside (final concentration, 0.4 mM). After 3 h of incubation, cells were collected, resuspended in phosphate-buffered saline, and sonicated as described by the manufacturer's protocol (Amersham Pharmacia Biotech). Cellular extracts were then centrifuged at 15,000 ϫ g, and the supernatants containing the soluble GST proteins were collected. Aliqouts containing equal amounts of GST or GST-RTR protein were incubated with glutathione-Sepharose 4B beads and washed in phosphate-buffered saline.
[ 35 S]Methionine-labeled N-CoR was obtained by in vitro translation using the TNT-coupled reticulocyte lysate system from Promega. The GST and GST-RTR beads were then incubated in 0.25 ml of binding buffer (20 mM Tris-HCl, pH 7.9, 100 mM KCl, 0.1% Nonidet P-40, 10% glycerol, 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 0.5% nonfat dry milk) with the [ 35 S]methionine-labeled N-CoR. After 1 h of incubation at room temperature, the beads were washed five times in binding buffer and then boiled in 30 l of 2ϫ SDS-polyacrylamide gel electrophoresis loading buffer. Solubilized proteins were separated by 8% SDS-polyacrylamide gel electrophoresis, and the radiolabeled proteins were visualized by autoradiography.
Site-directed Mutagenesis-Point mutations in the hinge domain, helix 3, and C terminus of RTR were introduced using a QuikChange site-directed mutagenesis kit (Stratagene) following the manufacturer's protocol. The pSG5-VP16RTR plasmid was used as parental DNA template. Two oligonucleotide primers were synthesized that are complementary to opposite strands of the vector and contain the desired mutation(s). The oligonucleotide primers were extended during temperature cycling using Pfu Turbo DNA polymerase and the following parameters: 18 cycles for 30 s at 95°C, 30 s at 55°C, and 14 min at 68°C. After temperature cycling, the product was treated with DpnI for 1 h at 37°C to digest the parental DNA template. The nicked vector DNA incorporating the desired mutations was then transformed into E. coli XL10-Gold ultracompetent cells. The authenticity of the mutants was confirmed by automatic DNA sequencing.

Suppression of ERR␣1-mediated Transcriptional Activation by RTR-Previous
studies have demonstrated that RTR is able to bind to response elements with the consensus sequence TCAAGGTCA and to direct repeats of AGGTCA, referred to as DR0 (7, 16, 22, 26 -28). Several receptors, including members of the ERR␣1 subfamily, have been reported to bind response elements similar to RTR-RE (29,38,41). Because RTR and ERR␣1 have been shown to be co-expressed in several cell types, including trophoblasts, embryonic stem cells, and embryonal carcinoma cells (9,19), we analyzed whether RTR would interfere with ERR␣1-induced transcriptional activation. To analyze this, we examined the effect of increasing levels of RTR expression on the transcriptional activation of a CAT reporter by ERR␣1 through a natural ERR␣1-RE (39). CHO cells were co-transfected with an ERR␣1 expression plasmid and an (ERR␣1-RE)-CAT reporter gene construct in the presence or absence of the expression plasmid pZeoSV-RTR. As shown in Fig. 1A, ERR␣1 caused a 2.5-fold increase in transcriptional activation through ERR␣1-RE, whereas RTR strongly inhibited this activation in a dose-dependent manner. Increasing amounts of RTR reduced transactivation to a level severalfold lower than that of the basal (minus ERR␣1) transactivation. A similar repression could be observed when the pZeoSV-RTR expression plasmid was co-transfected into CHO cells together with a CAT reporter plasmid under the control of three consecutive RTR-REs (Fig. 1B). These observations indicate that RTR acts as a repressor and can interfere with the transcriptional activation induced by ERR␣1 and likely other nuclear receptors able to bind to these REs.
Cross-talk between nuclear receptors can occur at different levels of transcriptional control, including competition for the same heterodimerization partners (42), for binding to the same response elements (43), or for shared co-repressors and coactivators (4,6). Because of the sequence similarities between ERR␣1-RE and the consensus RTR-RE, we determined whether the repression by RTR could be due to competition between the two receptors for binding to the same RE. Electrophoretic mobility shift assays demonstrated that both ERR␣1 and RTR were able to bind to ERR␣1-RE (Fig. 1C). The RTRand ERR␣1-oligonucleotide complexes migrated at different positions. EMSA using a combination of RTR and ERR␣1 showed only two complexes that migrated at the same position as the RTR-and ERR␣1-nucleotide complexes. Mammalian two-hybrid analyses using pSG5-VP16RTR and Gal4(DBD)-ERR␣1 indicated that ERR␣1 and RTR did not interact with each other (not shown). These observations are consistent with the concept that ERR␣1 and RTR do not form a heterodimer. Our results further suggest the repression of ERR␣1-induced transactivation by RTR is at least in part due to competition for the same binding site.
Repression of Basal Transcriptional Activation by RTR-Repression of basal transactivation by RTR could be demonstrated in several ways. As shown in Fig Gal4(DBD). These results suggest that this region of RTR harbors an active transcriptional repressor function that inhibits basal transcription, likely through the interaction with co-repressors.
If the repression by RTR is mediated through binding of co-factor proteins (such as co-repressors) that mediate the interactions of RTR with the basic transcriptional machinery, increasing concentrations of RTR would lead to squelching of that repression. To test this concept, Gal4(DBD)-RTR 149 -495 was co-transfected with (UAS) 5 CAT into CHO cells along with increasing amounts of pZeoSV-RTR expression plasmid. As shown in Fig. 2B, overexpression of RTR, which by itself did not affect basal promoter activity, can squelch the transcriptional repression by Gal4(DBD)RTR in a dose-dependent manner, presumably by competing for the limiting amounts of co-repressor(s) in the cell. These results suggest that interactions with co-repressors play a pivotal role in RTR-mediated transcriptional repression.
Analysis of the Interaction of RTR with Different Co-repressors-Repression of transcription by nuclear receptors involves interaction of the receptor with specific co-factors. N-CoR and SMRT are two co-repressors that have been reported to interact with several different nuclear receptors. RIP140 has been reported to be able to function as a co-repressor as well as co-activator (33,34). RIP140 can bind several different nuclear receptors and may repress transcription by competing with co-activators for binding to ligand-bound receptors (33). To determine whether any of these co-repressors could be involved in the transcriptional repression by RTR, we examined by mammalian two-hybrid analysis the interaction of RTR with these co-repressors. CHO cells were co-transfected with (UAS) 5 -LUC reporter plasmid and either Gal4(DBD)-N-CoR, Gal4(DBD)-SMRT, or Gal4(DBD)-RIP-140 in the presence or absence of pSG5-VP16RTR as indicated in Fig. 3. These results showed that N-CoR but not SMRT or RIP-140 was able to interact with RTR under the conditions tested, suggesting that the repression by RTR could be mediated through interactions with the co-repressor N-CoR. RTR was unable to bind the co-activators SRC-1 and CBP (not shown).
To analyze the interaction between RTR and N-CoR in more detail, its dependence on the VP16-RTR concentration as well the ability of RTR to squelch this interaction were examined (Fig. 4). CHO cells co-transfected with Gal4(DBD)-N-CoR and (UAS) 5 -CAT exhibited low reporter activity. This activity was enhanced dramatically by co-transfection with increasing concentrations of pSG5-VP16RTR DNA (Fig. 4A). Cells co-transfected with Gal4(DBD)-N-CoR and pSG5-VP16RTR reached a level of transactivation activity that was about 10-fold higher than that in control cells transfected with Gal4(DBD)-N-CoR or pSG5-VP16RTR alone. Co-transfection of Gal4(DBD)-N-CoR and pSG5-VP16 also did not enhance transactivation above control levels. These results further establish that RTR but not VP16 interacts efficiently with N-CoR. As shown in Fig. 4B, co-transfection with increasing concentrations of pZeoSV-RTR plasmid inhibited the transcriptional activation by VP16RTR. This squelching is could be due to competition between RTR and VP16RTR for binding to N-CoR or other nuclear proteins.
RTR Interacts with N-CoR in Vitro-The interaction between RTR and N-CoR was further examined in vitro by GST pull-down analysis. GST-RTR 149 -495 fusion protein and GST were immobilized on glutathione-Sepharose beads and then incubated with [ 35 S]methionine-labeled N-CoR. After extensive washing, labeled bound proteins were separated by SDS-electrophoresis and visualized by autoradiography. Fig. 5 shows that 35 S-labeled N-CoR was able to bind to GST-RTR but not to GST alone, suggesting that this binding is specific for RTR. This observation demonstrates that N-CoR is able to interact with RTR in vitro and supports the results obtained by twohybrid analysis.
The C-terminal Helix 12 of RTR Is Essential for the Interaction with N-CoR-To identify the domains within RTR that play a role in the interaction with N-CoR, the effect of a series of RTR truncations and point mutations on the interaction with N-CoR was examined by mammalian two-hybrid analysis. In addition, we wanted to examine the role of the unique helix 12 of RTR in this interaction. For this purpose CHO cells were co-transfected with Gal4(DBD)-N-CoR, (UAS) 5 -CAT, and different pSG5-VP16RTR deletion mutants. We first examined the effect of several C-terminal RTR deletion mutants on the interaction with N-CoR (Fig. 6A). RTR 1-149 , containing only the N terminus and the DBD, and RTR  , which also includes the hinge domain, did not induce reporter activity, indicating that these regions are not sufficient to promote interaction with N-CoR. All the smaller C-terminal deletions, even the deletion of the last 9 amino acids, abolished the interaction of RTR with N-CoR. These results indicate that the C terminus of RTR is essential in RTR/N-CoR interactions.
The helix 12 region constitutes the very C-terminal end of RTR (from Lys 482 through Glu 495 ; Fig. 6) (7,8,26). The amino acid sequence of helix 12 of RTR (Fig. 6B) is unique in that it does not contain the nuclear receptor AF-2 consensus sequence ⌽⌽X(E/D)⌽⌽ (25). The importance of helix 12 region in nuclear receptor/N-CoR interactions has been reported to be very much dependent on the type of receptor (30,(35)(36)(37). To further analyze the role of helix 12 in the interaction of RTR with N-CoR, several point mutations were introduced within this region, and their effects on the interaction of RTR with N-CoR examined in mammalian two-hybrid assays. The mutations introduced into the C terminus are shown in Fig. 6B. Only the RTR mutant containing the double mutation K489A,T490A showed a greatly diminished ability to induce reporter activity in twohybrid analysis, indicating the critical role of these amino acids in RTR/N-CoR interactions.
The Hinge Domain of RTR Is Essential for the Interaction with N-CoR-We next examined the effect of several N-terminal deletions and the role of the hinge region and LBD on the interaction of RTR with N-CoR. In RTR, the hinge domain stretches from Gly 149 to Leu 268 , and the LBD stretches from Ser 269 to the C terminus. As demonstrated in Fig. 7A, only the pSG5-VP16RTR deletion construct encoding RTR 149 -495 , containing the hinge and LBD domain, induced reporter gene activity to levels similar to that of full-length RTR, suggesting that this region interacts strongly with N-CoR. Further deletion up to Ser 212 caused a 50 -60% decrease in reporter activity, whereas deletion up to Tyr 240 did not cause any additional

FIG. 5. RTR interacts with N-CoR in GST pull-down assays.
GST and GST-RTR fusion protein were bound to glutathione-Sepharose 4B beads. 35 S-labeled N-CoR synthesized by in vitro translation was incubated with GST-Sepharose and (GST-RTR)-Sepharose beads. After 1 h incubation, the beads were washed extensively. Bound proteins were solubilized and analyzed by SDS-polyacrylamide gel electrophoresis. The radiolabeled proteins were visualized by autoradiography. The input represents 20% of the radiolabeled protein used in the binding assay.
decrease. Transcriptional activation was almost totally abolished with RTR 269 -495 , which lacks the whole hinge domain. The results suggest that two regions in the hinge domain of RTR, one from Gly 149 to Leu 211 and the other from Tyr 240 to Leu 268 , influence its interaction with N-CoR. In agreement with the results obtained in Fig. 6, RTR 149 -268 , containing the hinge domain only, did not promote interaction with N-CoR significantly. The results from Figs. 5 and 6 indicate that both the hinge domain and the helix 12 region at the C terminus of RTR are required for the interaction with N-CoR.
To analyze the importance of the RTR hinge domain in N-CoR binding further, we examined the effect of several point mutations within this region on RTR/N-CoR interaction. Fig.  7B demonstrates that the double mutation S246G,Y247G almost totally abolished the interaction of RTR with N-CoR, whereas the mutations L254A,P255A and S265A,Y266A had little effect on this interaction. These results further support the critical role of this part of the hinge domain in the interaction with N-CoR.
Importance of Helix 3 in RTR to N-CoR Binding-The region in the LBD containing helices 3-5 is moderately conserved among nuclear receptors (26,44,45). Recently, this region has been demonstrated to form the binding surface for the L⌽⌽LL motif in co-activators (46) and is in some receptors also involved in co-repressor binding. In RTR, helices 3-5 constitutes the region between Phe 299 and Val 350 . Fig. 8 shows that the mutation K318A in helix 3 greatly diminished transactivation in two-hybrid analysis, whereas I317A reduced transactivation by about 50%. Several mutations in helix 4 had little effect on the interaction of RTR with N-CoR. These results demonstrate that residue Lys 318 , and to a certain extent Ile 314 , in helix 3 of RTR are critical in N-CoR binding.

DISCUSSION
It is clear from previous observations and from the present study that RTR can function as an active repressor of transcription. In this study, we characterized in more detail this repressor function and identified several regions in RTR that are critical in the interaction with the co-repressor N-CoR. Our results demonstrate that this interaction exhibits differences as well as similarities with those reported for other receptors (30,(35)(36)(37).
Regulation of gene expression by nuclear receptors is complicated by the co-existence of multiple nuclear receptor signaling pathways that can interfere with each other. Cross-talk can involve any step in the mechanism by which the nuclear receptors regulate gene transcription, including competition for the same heterodimerization partners, co-repressors or co-activators, or competition for the same REs. A number of nuclear receptors, including ERRs and SF-1, bind REs that are very similar to RTR-REs (7,22,26,27,47,48). Moreover, some of these receptors have been demonstrated to be co-expressed in several cell types. For example, embryonal carcinoma and embryonic stem cells express RTR, SF-1, and ERRs (7,8,21,47,48). RTR and ERR␣ and -␤ have also been shown to be coexpressed in trophoblasts (19,49,50). 2 Differences in the affinity for the respective DNA element, the expression level of the receptors, and the presence of ligand are contributing factors for the degree of cross-talk between receptors. In this study, we show that RTR and ERR␣1 can bind the same response element and that RTR can suppress the transcriptional activation mediated by ERR␣1 by competing for binding to the same site. This antagonism could be relevant to the control of gene expression in several cell systems. Previously, we reported that RTR expression is down-regulated during retinoid-induced differentiation in embryonal carcinoma F9 cells (9); this decrease in RTR expression could relieve the repression of SF-1-and ERR-dependent transactivation of common target genes. Recent studies have identified protamine 1 and 2 as putative target genes for RTR (22,24). RTR and ERRs could positively and/or negatively control the transcription of these genes. Therefore, cross-talk between RTR and other nuclear receptor signaling pathways may play an important role in the control of gene expression during development and gametogenesis.
We demonstrated that RTR could repress the basal transcriptional activation. The repression of basal transcription could be reversed by increased expression of RTR and is likely due to squelching of the limiting amounts of co-repressor activity in the cell. The observations indicate that RTR can function as an active suppressor of gene transcription.
To repress transcription, nuclear receptors communicate with the basic transcription apparatus indirectly via interaction with protein intermediates, including co-repressors and deacetylases (4,6). RTR repressor activity likely involves interactions of RTR with various co-repressors, some of which may be highly specific for RTR. 3 Two-hybrid analysis demonstrated that RTR is able to interact with the co-repressor N-CoR but not with SMRT or RIP-140. The interaction with N-CoR was confirmed by pull-down analysis and indicates that these two proteins physically interact with each other. These results suggest a potential role for N-CoR in the transcriptional repression by RTR.
Previous studies have implicated a number of different regions in nuclear receptors in co-repressor interactions and shown similarities and important differences in the way nuclear receptors interact with N-CoR. Our study shows that the interaction of RTR with N-CoR has several unique characteristics. To determine which regions in RTR are critical in the interaction between RTR and N-CoR, we introduced a number of different deletions and point mutations in RTR and determined their effect on RTR/N-CoR interactions by two-hybrid analysis. This analysis identified several subdomains in RTR that are essential in RTR/N-CoR interactions. Although the modular structure of RTR is similar to that of other nuclear receptors, RTR has several unique features. Structural analysis has indicated that the hinge domain of RTR stretches from Gly 149 to Leu 268 , whereas its putative LBD starts at Ile 269 with helix 1 and ends right at the C terminus with helix 12. Crystallographic analysis of the ligand binding domain of RTR has indicated that the C terminus (from Lys 482 through Val 495 ) constitutes the H12 region (26). The sequence of H12 is unusual in that it does not contain the AF2 consensus sequence ⌽⌽X(E/D)⌽⌽ (25). Based on these observations it has been suggested that RTR may have a mode of action that is different from that of other receptors (51,52). Our results indicate that deletion of H12 or the introduction of specific point mutations abolish the interaction with N-CoR, suggesting that H12 is essential for and regulates N-CoR binding. Although several studies have indicated that the conformation of the H12 regulates the association of co-repressors and co-activators with nuclear receptors, the nature of this control can differ substantially between receptors (53)(54)(55). The H12 in retinoid X receptor has been found to sterically hinder co-repressor binding (37), whereas deletion of H12 converts apo-retinoid X receptor from a weak repressor to a strong repressor. In the case of estrogen receptor, only antagonist-bound receptor represses transcription and is able to interact with N-CoR (36). Antagonist binding likely causes a shift in the position of H12 of estrogen receptor, thereby exposing the co-repressor binding surface. Deletion of H12 in the TR and RAR receptors enables co-repressors, including N-CoR, to bind even in the presence of 3  ligand (30). In contrast, H12 has been shown to be essential in chicken ovalbumin upstream promoter-transcription factor (COUP-TF)/N-CoR interactions (35,56). Our results with RTR show that deletion of the H12 or the introduction of the double mutation K489A,T490A almost totally abolished the interaction of RTR with N-CoR. These observations indicate that the H12 region is essential in and can control the interaction of RTR with N-CoR. In many receptors, ligand binding causes a conformational change in H12 that results in the dissociation of co-repressors and the creation of a suitable co-activator binding surface. Recent studies have demonstrated that deletion of and mutations in H12 results in a conformation change in the LBD of RTR (26). Such a change in conformation may alter the N-CoR interaction surface in RTR and be responsible for the greatly reduced ability of RTR to bind N-CoR. This conformational change has been reported to also affect the dimerization properties of RTR and to reduce the binding of RTR to RTR-RE (26). Whether the H12 of RTR serves as a structural determinant or interacts directly with N-CoR has yet to be established.
Deletion from the N terminus of RTR demonstrated the importance of the hinge domain in RTR/N-CoR interactions. RTR 269 -495 , which lacks the whole hinge domain, was unable to interact with N-CoR. The double mutation S246G,Y247G almost totally abolished the interaction with N-CoR. Whether the importance of this Ser/Tyr is mainly structural or whether these residues can be phosphorylated and as such have a function in regulating RTR activation or RTR repressor activity has yet to be established. Recently, we have identified a novel repressor protein referred to as RAP80 3 that interacts with RTR in mammalian two-hybrid and pull-down analyses. This protein requires solely the region (Tyr 240 to Leu 268 ) in the hinge domain of RTR for its interaction and in contrast to N-CoR does not need H12 for binding. We have found that this protein can block the binding of N-CoR to RTR and competes with N-CoR for binding to this site. 3 We do not know whether this competition is based on steric hindrance or binding to the same sequence. These observations suggest that this hinge region plays an important role in controlling the interaction of RTR with several different co-repressors and may function as an interaction interface for some co-repressors. Neither the Gly 149 -Leu 211 nor the Tyr 240 -Leu 268 region of the hinge domain of RTR have sequence homology with the CoR box, a sequence in the hinge domain of TR and RAR, shown to be involved in N-CoR binding (30). The CoR box appears to function as structural determinant rather than a direct interface (53,55). Whether the region Tyr 240 -Leu 268 in the hinge domain of RTR interacts directly with N-CoR or controls RTR/N-CoR interactions indirectly has yet to be determined.
Recent studies have demonstrated that helices 3-5 of nuclear receptors form an interaction surface important in the binding of co-activators (53,57). Mutational analysis in this region of the retinoid X receptor and TR receptors have demonstrated the importance of this region also in the binding of co-repressors (54,58). The point mutation K318A in helix 3 of RTR greatly diminishes the binding of N-CoR. This lysine is highly conserved among many nuclear receptors and has been shown to play a key role in the recruitment of co-activators and the ligand-dependent activation of receptors (45). Mutation of the homologous lysine in TR and retinoid X receptor has been found to also abolish their interaction with N-CoR and SMRT (54,58). These observations suggest that in several receptors, and likely RTR as well, helices 3-5 play a key role in the interaction with co-repressors by serving as a binding surface.
In summary, we demonstrated that RTR functions as an active repressor of gene expression and can inhibit transcriptional activation mediated by other nuclear receptors. Our results suggest a potential role for N-CoR in the transcriptional repression by RTR. Deletion and point mutation analysis identified three RTR subdomains, a specific region in the hinge domain, helix 3, and the helix 12 region, that either provide an N-CoR binding surface or control indirectly the interaction with N-CoR. Our study shows that this interaction exhibits several characteristics unique to RTR. These repressor activities may provide important mechanisms by which RTR regulates gene expression during development and spermatogenesis.