* This project was supported by Austrian Fonds zur Förderung der wissenschaftlichen Forschung Grants P13638-GEN and P14909-GEN (to C. S.) and MOB-P13707 (to E. O.), the Herzfelder Familienstiftung (to E. O.), and GEN-AU program of the Austrian Government (to C. S.). 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. The on-line version of this article (available at http://www.jbc.org) contains Supplemental Figs. 1 and 2. ‡ Current address: Dept. of Vascular Biology and Thrombosis Research, University of Vienna, Brunnerstrasse 59, A-1235 Vienna, Austria. ¶ Current address: Dept. of Biochemistry, University of Medicine and Dentistry of New Jersey, 683, Hoes Lane, Piscataway, NJ 08854. ∥ Recipient of fellowships from the Fondation pour la Recherche Médicale and the Ligue Nationale contre le Cancer. ** Current address: School of Life Sciences, University of Dundee, Dundee DD1 5 EH, UK.
Histone acetyltransferases and deacetylases are recruited by transcription factors and adapter proteins to regulate specific subsets of target genes. We were interested in identifying interaction partners of histone deacetylase 1 (HDAC1) that might be involved in conferring target or substrate specificity. Using the yeast two-hybrid system, we isolated the repressor of estrogen receptor activity (REA) as a novel HDAC1-associated protein. We demonstrated the in vivo interaction of REA with HDAC1 and characterized the respective domains required for their interaction in vitro. In addition, we found that REA also associates with the class II histone deacetylase HDAC5. In luciferase reporter assays, REA decreased transcription, and this repression was sensitive to the deacetylase inhibitor trichostatin A. Finally, we showed that REA specifically interacts with the chicken ovalbumin upstream binding transcription factors and II. The nuclear receptor chicken ovalbumin upstream binding transcription factor I was found to cooperate with REA and histone deacetylases in the repression of target genes. We, therefore, propose a novel function for REA as a mediator of transcriptional repression by nuclear hormone receptors via recruitment of histone deacetylases.
Eukaryotic DNA is packaged into highly dynamic chromatin, the basic unit of which is the nucleosome. This consists of a central “hub” comprising two of each of the core histone proteins H2A, H2B, H3, and H4, around which are wrapped 146 bp of DNA.
The highly conserved histone proteins are subject to several post-translational modifications, including acetylation, phosphorylation, methylation, and sumoylation. Extensive studies have forged a link between the acetylation state of chromatin and its transcriptional status. Dynamic acetylation and deacetylation of lysine residues within the N-terminal tails of core histones seem to be necessary for a series of crucial nuclear events such as silencing, replication, and correct DNA repair.
Initial research into the relevance of histone acetylation aimed at identifying and characterizing the enzymes involved; that is, the histone acetyltransferases and their antagonists, the histone deacetylases (HDACs).
). An important aspect of HDAC1 function is its target specificity, determined predominantly by the precise composition of HDAC1-containing complexes and the nature of HDAC1-associated proteins. Apart from the well characterized SIN3 and NuRD complexes (Refs.
). Class II enzymes, comprising HDACs 4-7 and HDAC9-10, are generally linked to the regulation of differentiation-related genes, and in contrast to the strictly nuclear class I deacetylases HDAC1 and HDAC2, are found in both the nucleus and cytoplasm of cells (
A large number of transcription factors regulate gene expression through the recruitment of histone deacetylase complexes. Nuclear hormone receptors are well characterized examples of HDAC-recruiting proteins. Initially, the retinoic acid receptors and thyroid receptors were shown to interact indirectly with HDACs through their interaction with N-CoR and SMRT in the absence of ligand (
). Apart from this, specific recruiting factors are required to confer high target specificity upon HDAC1. To this aim, we employed the yeast two-hybrid system to isolate hitherto unknown interaction partners of HDAC1. One of the candidate proteins isolated was the repressor of estrogen receptor activity (REA) (
). By co-immunoprecipitation experiments and glutathione S-transferase (GST) pull-down assays, we established the interaction between REA and both HDAC1 and HDAC5. Moreover, we demonstrated that REA specifically interacts not only with estrogen receptor but also with the orphan nuclear receptors chicken ovalbumin upstream binding transcription factor (COUP-TF) I and II. Our data identify REA as novel HDAC-interacting protein that modulates the activity of a defined subset of nuclear hormone receptors.
Chemicals and Materials—17β-Estradiol and 4-hydroxytamoxifen were obtained from Sigma-Aldrich and resuspended in 96% ethanol. Custom oligonucleotides were purchased from VBC-Genomics, Vienna, Austria.
Cell Culture and Cell Transfection—U2OS, HeLa, 293, Swiss 3T3, and MCF7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum. Embryonic stem cells were grown as previously described (
). Deletions were constructed using standard PCR-based methods. Specific information regarding oligonucleotide sequences is available upon request. The bait plasmid used in two-hybrid screening was constructed by inserting a PCR fragment corresponding to aa 299-482 of mouse HDAC1 into pEG202, downstream of the region encoding the LexA DNA binding domain (DBD). Prey libraries of HeLa cDNA or mouse embryonic brain cDNA were kindly provided by R. Brent and W. Stockinger, respectively. The REA cDNA fragment encoding aa 89-299 was PCR-amplified and subcloned via BamHI/XmaI into pQE30 for expression of His6-tagged REA in the Escherichia coli BL21 DE3 pLysS strain. Correct reading frames were verified by sequencing. REA deletion constructs were PCR-amplified and cloned into adequately digested pGEX4T1 vectors for expression as a GST fusion proteins. GST-HDAC1 constructs have been described elsewhere (
). The mHDA5 expression vector was kindly provided by S. Khochbin. Expression plasmids for COUP-TFI and II were obtained from M. G. Parker, and the GFP-tagged COUP-TFI plasmid was obtained from F. Pakdel. C-terminal and N-terminal deletion mutants of COUP-TFII were obtained by digestion with EagI and BamHI, respectively.
Yeast Two-hybrid Screening—Plasmids and yeast strains used in this study were adapted from the system of R. Brent (
). The yeast strain EGY191, which contains a chromosomally integrated auxotrophic marker (LEU2) and an ectopic lacZ reporter plasmid (pSH18-34), was sequentially transformed with the bait-encoding pEG202-HDAC1 (299-482) plasmid and either the pJG4-5 mouse embryonic brain cDNA or the HeLa cDNA prey libraries. Transformants were plated on media lacking leucine. Galactose inducible LEU+ colonies exhibiting lacZ activity, as determined by filter-lift assays, were further characterized. Colonies harboring interacting prey proteins were used to quantify β-galactosidase activity by an o-nitrophenyl-β-d-galactopyranoside-based liquid assay.
Northern Blot Analysis—mRNA expression was analyzed using the Northern blot sandwich method as described previously (
Antibodies—To raise a specific antibody directed against mouse REA, the His6-tagged fragment (aa 89-299) expressed in E. coli BL21 DE3 pLysS was used to immunize rabbits by standard methods. Antibodies were affinity-purified by overnight incubation on nitrocellulose strips loaded with recombinant His-REA. After washing with 10 mm Tris-HCl, pH 8.0, and 10 mm Tris-HCl, pH 8.0, 0.5 m NaCl, bound antibodies were eluted by sequential incubation with 100 mm glycine, pH 2.5, and 100 mm triethylamine, pH 10.0. Eluates were neutralized by the addition of 1 m Tris-HCl pH 8.0 and tested for specificity by Western blot analysis. The REA antibody described above as well the HDAC1 monoclonal antibody 10E2 and the RbAp48 monoclonal antibody 13D10 are commercially available at Upstate Biotechnology. The estrogen receptor α antibody H-184, the Gal4 DNA-BD monoclonal antibody, and the MTA1 antibody C-17 were purchased from Santa Cruz Biotechnology. Other antibodies used in this study were the HA-specific monoclonal antibodies 12CA5 and 16B12 and the GFP monoclonal antibody from Clontech.
Glutathione S-Transferase Pull-down Assays—GST fusion proteins were expressed in E. coli BL21 DE3 pLysS. GST fusion proteins with HDAC1, HDAC2, and HDAC3 have been described previously (
). GST-SRC-1 and expression vectors for RXRα and TRα were obtained from M. G. Parker. Large scale preparation, purification of the recombinant proteins, and binding to the glutathione beads were performed as previously described (
). Beads coated with GST fusion proteins were incubated with in vitro translated [35S]methionine-labeled full-length REA or HDAC1 protein for 2 h in lysis buffer (20 mm Tris-HCl, pH 8.0, 100 mm NaCl, 1 mm EDTA, 0.5% Nonidet P-40, Complete protease inhibitor mixture). Beads were washed 3 times in lysis buffer and 1 time in radioimmune precipitation assay buffer (150 mm NaCl, 1% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, 50 mm Tris-HCl, pH 8.0) and boiled in SDS-PAGE sample buffer to elute bound proteins. Eluates were then resolved by SDS-PAGE and visualized by autoradiography. To control normalized amounts of immobilized GST and GST fusion proteins, duplicate inputs were resolved by SDS-PAGE and subsequently stained with Coomassie Brilliant Blue solution.
Protein Isolation and Immunoprecipitation—Whole-cell protein extraction and immunoprecipitation experiments were performed as previously described (
). Briefly, equal amounts of protein were incubated with specific antibodies for 1 h at 4 °C in lysis buffer (20 mm Tris-HCl, pH 8.0, 100 mm NaCl, 1 mm EDTA, 0.5% Nonidet P-40, Complete protease inhibitor mixture). Then, 20 μl of 50% v/v protein A-coupled Sepharose bead slurry was added and incubated under gentle agitation overnight at 4 °C. After three washes in lysis buffer, bound proteins were resolved by SDS-PAGE and analyzed on Western blots. Alternatively, one-half of the immunoprecipitates was used directly to assay deacetylase activity, whereas the second half was mixed with SDS-PAGE sample buffer and analyzed on Western blots.
Isolation of Cytoplasmic and Nuclear Protein Fractions—To separate nuclear and cytoplasmic proteins cells were lysed in ice-cold hypotonic buffer (10 mm HEPES, pH 7.9, 10 mm KCl, 1.5 mm MgCl2, 0.3 mm phenylmethylsulfonyl fluoride, 0.5 mm dithiothreitol). After homogenization by passing the suspension through a 20-gauge syringe needle and control of efficiency by phase contrast microscopy, nuclei were pelleted by centrifugation at 2000 × g for 8 min at 4 °C. The supernatant cytoplasmic extract was transferred to a fresh tube. To obtain nuclear proteins, nuclei were resuspended in 1 volume of lysis buffer (20 mm Tris-HCl, pH 8.0, 100 mm NaCl, 1 mm EDTA, 0.5% Nonidet P-40, Complete protease inhibitor mixture) and incubated at 4 °C for 30 min under agitation. The nuclear extract was then cleared from debris by centrifugation at 16,000 × g for 30 min at 4 °C.
Histone Deacetylase Activity Assays—Histone deacetylase activity was measured as previously described (
). In short, protein extracts or immunoprecipitated proteins were incubated with 5 μl of [3H]acetate-labeled chicken erythrocyte histones in a total volume of 30 μl of lysis buffer (20 mm Tris-HCl, pH 8.0, 100 mm NaCl, 1 mm EDTA, 0.5% Nonidet P-40, Complete protease inhibitor mixture) for 1 h at 30 °C. The reaction was stopped by the addition of 36 μl of 1 m HCl, 0.4 m acetate, and released acetate was extracted with 800 μl of ethyl acetate. After centrifugation at 8400 × g for 5 min, 600 μl of the organic phase were counted in 3 ml of toluene liquid scintillation mixture.
Western Blot Analysis and Indirect Immunofluorescence—Protein extracts or immunoprecipitated proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membrane (Schleicher & Schuell). After incubation with specific antibodies, the proteins were detected with the ECL kit from PerkinElmer Life Sciences.
Subcellular localization of HDAC1 and REA was determined by indirect immunofluorescence microscopy (Zeiss Axiovert 135TV microscope and Leica TCS NT confocal microscope) as previously described (
). Nuclear DNA was visualized with 4′,6-diamidino-2-phenylindole or TOTO-3 (Molecular Probes Inc.).
Luciferase Reporter Assays—SV40 thymidine kinase minimal promoter/Gal4-luciferase reporter constructs and Gal4-DBD fusion mammalian expression vectors were generous gifts from D. Eberhard and have been described previously (
). For luciferase reporter assays cells were grown and transfected in 6-well tissue culture dishes. 48 h after transfection, cells were lysed by incubation in 200 μl of luciferase assay buffer (25 mm Tricine, pH 7.8, 0.5 mm EDTA, 0.54 mm sodium tripolyphosphate, 16.3 mm MgSO4·7H2O, 0.1% Triton, 5.6 mm dithiothreitol; 1.2 mm ATP, 46 μm luciferin). Luciferase levels were measured using a Mediators Photoluminometer. Transfection efficiency was assessed by measuring β-galactosidase levels in an 80-μl aliquot of the cell lysate using an o-nitrophenyl-β-d-galactopyranoside-based liquid assay. Additionally, an aliquot of each extract was analyzed on Western blots to evaluate the levels of co-expressed proteins.
REA Interacts with HDAC1 in the Yeast Two-hybrid System—The class I enzyme HDAC1 is present within three large multi-protein complexes, referred to as the SIN3, NuRD (for review, see Ref.
J. Taplick, V. Kurtev, and C. Seiser, unpublished data.
Therefore, we performed a yeast two-hybrid screen using a truncated mouse HDAC1 protein (aa 299-482) as bait with cDNA libraries from human HeLa cells and mouse embryos. Yeast colonies harboring HDAC1-interacting proteins were identified by their inducible viability on media lacking leucine and increased transcription from the lacZ gene. In this manner, we repeatedly identified the repressor of REA (
) as an HDAC1-interacting protein in both the human and the murine cDNA libraries. Sequencing of the corresponding cDNAs revealed that they encode truncated versions of REA encompassing amino acids 89-299 and 103-299 of the human and mouse proteins, respectively. Mouse REA was first identified as the murine B-cell receptor-associated protein 37 (mBAP37), a protein originally isolated through its physical association with the B lymphocyte IgM antigen receptor (
). Complete sequencing of the murine open reading frame revealed that the encoded mouse REA displays 100% identity with the human REA protein (data not shown).
The interaction of HDAC1 with REA was confirmed by back-crossing. Freshly transformed S. cerevisiae strain EGY191 expressing REA (aa 89-299) as bait, and HDAC1 (aa 299-482) as prey showed inducible viability on media lacking leucine (Fig. 1A). In contrast, transformants expressing an unrelated bait protein (LexA-bicoid) or the LexA DNA binding domain alone were not viable under these conditions, thus confirming the specificity of the HDAC1/REA interaction. These results were complemented by the analysis of lacZ transcription levels (Fig. 1B), which showed an ∼6-fold induction of β-galactosidase activity in strains co-expressing REA (89-299) and LexAHDAC1 (299-482) when compared with the LexA DNA binding domain or the unrelated control LexA-bicoid.
Expression and Intracellular Localization of REA—We first looked at the expression pattern of REA in a variety of tissues and cell lines. 32P-Labeled REA cDNA was hybridized to a Clontech Northern blot of poly(A)+ RNA isolated from different mouse tissues. REA mRNA was detectable in several tested tissues, with very high expression in heart, liver, kidney, and to a lower extent, testis (Fig. 1C). HDAC1 mRNA was detected at high levels in testis and kidney and at lower levels in all other tissues tested with the exception of skeletal muscle.
To analyze REA protein expression, we raised an REA-specific polyclonal antiserum using bacterially expressed, purified His6-tagged REA protein (aa 89-299). Specificity of the REA antiserum was confirmed by Western blot analysis and further enhanced by affinity purification against the antigen (data not shown). This affinity-purified antibody detected a single band corresponding to a 37-kDa protein in Western blots of whole cell protein extracts from a variety of established human and mouse cell lines (Fig. 1D). Among those tested were the human osteosarcoma cell line U2OS and the human mammary carcinoma cell line MCF-7. Mouse cell lines analyzed for REA expression were the interleukin-2-dependent T cell line B6.1 and primary erythroblasts as well as embryonic stem cells derived from HDAC1 wild type and homozygous mutant mice. In addition, a signal at 37 kDa was also detected in whole cell protein extracts from rat embryonic fibroblasts, indicating reactivity of the antibody for human, mouse, and rat species. HDAC1 protein was present in all cell lines tested, with the exception of the HDAC1 -/- ES cells.
To establish further the connection between REA and the nuclear enzyme HDAC1, we next determined the subcellular localization of the REA protein. A putative nuclear localization sequence (aa 86-89) has been found through analysis of the polypeptide sequence (
); however, to our knowledge, direct evidence for the presence of REA within the nucleus is lacking. Using the affinity-purified REA antibody, we analyzed the subcellular localization of REA in logarithmically growing MCF-7 cells by confocal microscopy. REA was present in a distinct speckled pattern throughout the cell, with denser signals within the area of the nucleus (Fig. 2A, upper panel). A similar REA distribution pattern was observed in Swiss 3T3 fibroblasts (Fig. 2A, lower panel) and HeLa cells (data not shown). Our data indicate that a subpopulation of REA shares its subcellular localization with the strictly nuclear HDAC1 (
). We were, therefore, interested in knowing whether treatment of MCF-7 cells with the hormone would lead to a re-distribution of REA within the cell. Confocal immunofluorescence studies of MCF-7 cells treated for various periods of time with 10-6m 17β-estradiol or 10-6m 4-hydroxy-tamoxifen showed that REA signal intensity and distribution remained unaffected by the presence of ligand (data not shown).
In a separate approach we analyzed the intracellular localization of REA by Western blot analysis of biochemically fractionated cell extracts. Logarithmically growing MCF-7 cells were harvested and used to prepare cytoplasmic versus nuclear extracts (as described under “Experimental Procedures”). Sequential probing with antibodies specific for REA and HDAC1 confirmed that a significant portion of REA was present in the nuclear fraction, coincident with the localization of HDAC1 (Fig. 2B). Probing with a tubulin-specific antibody indicated only minimal contamination of nuclear fractions with cytoplasmic proteins.
In Vivo Association of REA with HDAC1—To demonstrate the in vivo interaction between REA and HDAC1, we decided to precipitate endogenous REA and analyze co-precipitated proteins for the presence of HDAC1. Whole cell protein extracts prepared from logarithmically growing MCF-7 cells were used for immunoprecipitation with REA antibody or, as a negative control, with the corresponding pre-immune serum. Precipitated proteins were analyzed for associated HDAC1 by Western blotting (Fig. 3A). Duplicate immunoprecipitations were used for HDAC activity assays (Fig. 3B).
As illustrated in Fig. 3A, REA was precipitated by the REA-specific antibody but not by the pre-immune serum. Furthermore, HDAC1 was associated with immunoprecipitated REA, as determined by probing of the Western blot with HDAC1 antibody. Consistent with this finding, HDAC assays using immunoprecipitated fractions showed that considerable levels of HDAC activity were associated with REA (Fig. 3B). These were comparable with activities measured in RbAp48 immunoprecipitations (data not shown). Because RbAp48 is widely regarded as an integral part of the HDAC1 enzyme complex (
), these data indicate efficient recruitment of deacetylases by the REA protein.
Conversely, to analyze REA association with immunoprecipitated HDAC1, whole cell protein extracts were subjected to immunoprecipitation with the HDAC1 monoclonal antibody 10E2 and, as control, with an unrelated monoclonal antibody. Precipitated proteins were then examined by Western blot analysis. As shown in Fig. 3C, REA specifically co-immunoprecipitated with the HDAC1 protein, confirming an in vivo interaction between REA and HDAC1.
In Vitro Interaction between HDAC1 and REA—To define the HDAC1-REA interaction domains within both proteins, we performed GST pull-down experiments. To ensure comparable inputs, duplicates of all GST fusion proteins were separated by SDS-PAGE and stained with Coomassie Brilliant Blue (Supplemental Fig. S1). Full-length HDAC1 and different portions of mouse HDAC1 fused to GST (depicted in Fig. 4A) were bound to glutathione beads and incubated with in vitro translated 35S-radiolabeled REA.
As shown in Fig. 4B, full-length HDAC1 associated with REA, suggesting a direct interaction between these proteins. The region of HDAC1 corresponding to aa 1-303 (delA) exhibited only negligible binding, similar to the signal obtained with GST alone. In contrast, a GST fusion protein comprising the C-terminal 179 aa of HDAC1 was sufficient to mediate interaction with REA (data not shown), comparable with the interaction seen for a full-length GST-HDAC1 fusion protein. This was in accordance with the fact that, in the two-hybrid screen, REA was identified using the C-terminal half of the HDAC1 protein as bait. Further subdivision of the C terminus into two parts (delB and delC) resulted in polypeptides that retained their binding capacity for REA, suggesting the existence of two distinct REA binding domains (Fig. 4B).
To define the HDAC binding region within REA, the REA polypeptide was subdivided into a series of complementary fragments and fused to GST (represented schematically in Fig. 4A). Pull-down experiments were performed with these constructs or full-length REA fused to GST using radiolabeled full-length HDAC1. As can be seen in Fig. 4C, GST-REA interacted with HDAC1, whereas the unrelated control protein LR8 showed no interaction. GST-REAdel9 (aa 185-299) was sufficient for interaction with HDAC1. However, in addition to this C-terminal association region, REA bound HDAC1 via a second domain located within the N-terminal part, witnessed by the binding ability of GST-REAdel1 and GST-REAdel2 (aa 1-85 and 86-146, respectively). The N- and C-terminal interaction domains recruited HDAC1 independently. Binding affinity of the individual domains was markedly lower than that of the full-length protein, indicating that under physiological conditions there might be cooperativity between these two regions for recruitment of HDAC1. GST-del3 (aa 147-184) and GST-del8 (aa 117-184) showed no specific binding, indicating that the central portion of the REA protein (aa 117-184) was expendable for interaction with HDAC1. Furthermore, HDAC1-REA binding was not mediated by DNA, since addition of ethidium bromide (100 μg/ml) to the pull-down reactions had no effect upon HDAC1 precipitation (data not shown).
Analysis of REA-associated deacetylase activity in HDAC1-deficient embryonic stem cells (
) by immunoprecipitation experiments revealed significant HDAC activity associated with REA in the absence of HDAC1 (Supplemental Fig. S2). These data suggested the interaction of other members of the deacetylase family with REA. Therefore, we analyzed the interaction of other deacetylases with the REA protein. As shown in Fig. 4D, both HDAC2 and HDAC3 GST fusion proteins interacted with REA.
Interestingly, members of the HDAC class II family such as HDAC5 also bound to REA in GST pull-down assays (Fig. 4E). When compared with HDAC1, HDAC5 showed in part different specificity with respect to the interaction domains within the REA protein. HDAC5 interacted predominantly with the N terminus of REA (del1), whereas HDAC1 associated with two REA domains (see above). To test whether HDAC5 interacted with REA in an in vivo assay, we expressed FLAG-tagged HDAC5 alone or together with HA-REA in 293 cells. Epitope-tagged REA was precipitated with HA antibody, and the immunoprecipitate was analyzed for the presence of REA with REA-specific antibody. HDAC5 was present in the HA precipitate in the presence of HA-REA but not in the absence of the protein, indicating a specific in vivo interaction between REA and HDAC5 (Fig. 4F).
REA Acts as a Transcriptional Repressor—Although REA was initially described as a repressor of estrogen receptor activity (
), there is to our knowledge only indirect evidence for a repressive function of REA. The prevailing model for the REA mode of function is based upon the competition for binding to ER between REA and the SRC-1 co-activator. Our data showing an association between REA and histone deacetylases indicated that REA could additionally act as a repressor on its own. We, therefore, decided to assess the possible repressive function of this protein using mammalian in vivo reporter gene assays. We cloned full-length REA as a C-terminal fusion to the Gal4 DNA binding domain. This construct was transiently transfected into U2OS cells together with a plasmid carrying a luciferase reporter cassette under the control of a SV40 thymidine kinase promoter preceded by four Gal4 binding sites (
Luciferase levels upon co-transfection of the Gal4 DBD alone were arbitrarily set to 100%. Transfection of the Gal4DBD-REA-expressing construct led to a reduction of luciferase activity to 40% that of the control value (Fig. 5A, gray bars). Based upon the interaction of REA with HDACs, we asked the question of whether the observed repression could be alleviated by treatment with an HDAC inhibitor. Indeed, REA-mediated repression of the reporter cassette was nearly completely abolished by treatment with the deacetylase inhibitor trichostatin A (TSA) (Fig. 5A, black bars), a known specific inhibitor of histone deacetylases. Western blot analysis of Gal4DBD-REA expression levels confirmed that the reduced repressor function of REA in response to TSA treatment was not due to decreased expression of the Gal4DBD-REA fusion protein (Fig. 5B). A fusion construct between Gal4 DBD and GRG4 (
), a member of the mammalian Groucho family known to be independent of deacetylases for its transcriptional repression, was used as control (Fig. 5A). In contrast to REA, GRG4-mediated repression was not affected by treatment with TSA.
A Subset of Nuclear Hormone Receptors Interacts with REA—Next, we examined the specificity of the association between REA and ERα by GST pull-down assays. Consistent with the initial reports characterizing REA as a repressor of estrogen receptor activity (
), GST-REA precipitated radiolabeled ERα in a ligand-independent manner but did not interact with retinoic acid receptor RXRα or thyroid hormone receptor TRα (Fig. 6A). In contrast, the co-activator SRC-1 associated with all three nuclear receptors in the presence of the respective ligand (Fig. 6A).
In addition, we tested the orphan nuclear hormone receptors COUP-TF I and II for interaction with REA. These nuclear receptors have been implicated in the shut-off of major histocompatibility complex I transcription (
). Interestingly, GST-REA had a very high affinity for COUP-TFI and COUP-TFII (Fig. 6B). Although ER-α interacted with both the N-terminal and the C-terminal portion of REA, COUP-TFII preferentially associated with the N-terminal portion of REA (Fig. 6C). In GST pull-down assays, full-length COUP-TFII and a mutant lacking the N terminus (COUP-TFII (56-414)) efficiently associate with in vitro translated REA (Fig. 6D). In contrast, the N-terminal part of COUP-TF did not interact with REA, suggesting that the C-terminal part of the receptor is sufficient for association with the REA protein.
Next, we asked whether REA associates in vivo with nuclear hormone receptors. Although a functional relationship between estrogen receptor and REA was convincingly established (
), evidence for a physical in vivo interaction of nuclear receptors with REA was still missing. To analyze the association of HDAC1 and REA with ERα, we prepared whole cell protein extracts from ER-positive MCF-7 cells. Using an ERα-specific polyclonal antibody, we immunoprecipitated ERα and analyzed associated proteins and co-immunoprecipitated HDAC enzymatic activity. Sequential probing of Western blots with the respective antibodies (Fig. 7A) revealed that ERα was precipitated very efficiently. Furthermore, both REA and HDAC1 were specifically co-precipitated with the ERα-antibody but not with the control antibody. In parallel, we tested the ERα immunoprecipitates for HDAC activity and found that ERα was associated with significant deacetylase activity (Fig. 7B).
To examine a potential in vivo association of the COUP-TF orphan receptor with REA, we expressed HA-REA together with GFP-COUP-TFI or GFP in HeLa cells. As shown in Fig. 7C, immunoprecipitated REA protein was associated with GFP-COUP-TFI but not with GFP alone. Taken together, these data show a specific in vivo interaction of REA with ERα and COUP-TF, suggesting that these nuclear receptors could be specifically targeted by the REA·HDAC complexes.
REA Acts as a Transcriptional Co-repressor of the Orphan Receptor COUP-TFI—Next, we tested the effect of REA on the transcriptional regulator activity of COUP-TFI. COUP-TFs can bind to a wide spectrum of response elements exhibiting variations in the AGGTCA core motif. This allows COUP-TFs to interact with a variety of hormone response elements recognized by other members of the nuclear receptor superfamily. In particular, COUP-TFs were shown to bind estrogen-responsive elements (
) in ER-negative HeLa cells (Fig. 7D and E). Co-expression of REA with COUP-TFI led to significant down-regulation of luciferase activity of the reporter. Additional expression of HDAC5 further increased the repressor effect of COUP-TFI and REA, whereas co-expression of HDAC1 had no enhancing effect. The difference in the effects of co-expressed HDAC1 and HDAC5 is most probably due to the high expression levels of endogenous HDAC1 in HeLa cells. A similar repression upon transfection of REA and HDAC5 was obtained in MCF-7 cells on a cathepsin D reporter construct (data not shown). Together with the findings on REA-HDAC interaction, these results indicate that REA acts as a co-repressor of the orphan receptor COUP-TFI by recruiting histone deacetylases.
REA as an HDAC-associated Protein—The specificity of histone deacetylases as transcriptional repressors is defined by the recruitment of these enzymes by transcription factors or co-repressor proteins. In this study we have identified the repressor of REA as an HDAC1-associated factor, and we have characterized the interaction of REA with histone deacetylases.
The HDAC1 protein contains several functional domains (see Fig. 4A). The central part of the HDAC1 protein encompasses the catalytic domain called HDAC consensus motif that is common to all class I and class III deacetylases (
). The HDAC association domain (within the HDAC1 N-terminal part was previously shown to be required for homodimerization and for interaction with HDAC2 as well as with other components of HDAC1 complexes such as Sin3ASin3B and RbAp48 (
). The data presented in this study indicate that HDAC1 interacts with the co-repressor REA through its C-terminal domain, suggesting that the HDAC1 protein contains specific interaction surfaces for different adapter proteins.
REA most probably interacts directly with HDAC1 via two independent domains located in the N terminus (aa 1-117) and the C terminus (aa 185-299), respectively. The fact that binding activity of the full-length protein is higher than that of the individual domains suggests a cooperativity of these two domains. Interestingly, although isolated in this study as interacting with HDAC1, REA can form stable interactions with other members of the histone deacetylase family, intriguingly even with class II deacetylases, e.g. HDAC5. Notably, the interaction with HDAC5 requires only the N-terminal part of REA. This N-terminal domain, which is also involved in the interaction with HDAC1, contains RD1, one of the two domains previously shown to possess transcriptional repression activity (
). Consistent with these findings, REA immunoprecipitates show remarkably robust deacetylase activity in murine and human cell lines. Taken together, these data indicate that the REA protein has to be considered as a co-repressor that can recruit several members of the HDAC family.
The REA protein was originally identified as BAP37, a factor characterized as a B-cell receptor-associated protein (
). As shown in this study, REA mRNA and protein are expressed in a large variety of tissues and cell lines, arguing against a B-cell-restricted function. A significant portion of REA molecules shows nuclear localization and is found in association with the nuclear deacetylase HDAC1.
REA as Transcriptional Co-repressor of Nuclear Hormone Receptors—Several proteins, including N-CoR, SMRT, SHP, and REA, have been identified as transcriptional co-repressors of ER (
) that exerts its function through competition with co-activators for binding to ER. Subsequent studies also reported a repressive effect of REA upon ligand-bound ER, i.e. association with an estrogen receptor, thereby dampening the estrogen-induced transcriptional activation (
). We propose that REA acts as co-regulator of ERα by recruiting HDACs to nuclear receptor target genes.
As shown by immunoprecipitation experiments, endogenous ERα interacts with REA and HDAC1. Targeting of REA to luciferase reporter constructs confirms an inherent repressive function, partially reversible by treatment with TSA. RD2, the second repression domain of REA, is not required for the interaction with deacetylases and may account for the HDAC-independent repressive activity of REA. Similar HDAC-dependent and independent repressor activities were also observed for other transcriptional regulators, such as Sin3 and Rb (
) show cyclic recruitment of co-activators in response to the hormone. The temporal presence of acetyltransferases such as cAMP-response element-binding protein (CREB)-binding protein (CBP) and p300/CBP-associated factor induced a corresponding cyclic hyperacetylation separated by a transient wave of deacetylation. These results strongly suggest that histone deacetylases are present at ER target promoters even under inducing conditions in the presence of ligands. Several co-regulators of ER, including LCoR, MTA1, and RIP140, recruit histone deacetylases in the presence of estrogen (
Our experiments show an interaction of REA not only with ERα but also with other members of the nuclear receptor family, namely COUP-TFI and COUP-TFII. COUP-TFs bind DNA either as homodimers or as heterodimers with RXR and have been shown to negatively regulate the activation function of other nuclear hormone receptors (for review, see Ref.
) have previously been shown to be involved in the repressive activity of COUP-TFs. In agreement with a repressor function, COUP-TFs associate directly or indirectly with deacetylases, including HDAC1 (
). Our data showing physical and functional interaction between REA and COUP-TFs suggest that REA acts as a more general co-repressor protein, which interacts with a subset of nuclear hormone receptors.
The REA/Prohibitin Family of Co-repressors—Homologues of REA have been found in rodents, C. elegans, and S. cerevisiae as well as in the Cyanobacteria Synechocystis (data not shown), suggesting important cellular functions for these proteins. The mammalian REA homologue prohibitin was identified as a growth inhibitory protein that represses E2F-dependent transcription by binding to the Rb protein (
). These findings together with the data reported in the present study, suggest that REA and prohibitin belong to a family of co-repressor proteins that mediate their function by recruiting deacetylases to distinct sets of transcription factors.
We thank M. Lamers, C. Gausterer, S. Khochbin, M. G. Parker, F. Pakdel, and P. Augereau for providing constructs, G. Brosch for providing radiolabeled histones, and R. Brent and W. Stockinger for providing prey libraries. We are grateful to B. Schuettengruber for expert technical advice and D. Meunier for helpful comments on this manuscript.