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

J. Biol. Chem., Vol. 276, Issue 51, 48196-48205, December 21, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/51/48196    most recent M104320200v1 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 Lin, H.-M. Articles by Cheng, S.-y. Search for Related Content PubMed PubMed Citation Articles by Lin, H.-M. Articles by Cheng, S.-y. Social Bookmarking                      What's this?

An Isoform of Branched-chain Aminotransferase Is a Novel Co-repressor for Thyroid Hormone Nuclear Receptors^*

Huei-Min Lin, Masahiro Kaneshige, Li Zhao, Xiaoyong Zhang, John A. Hanover, and Sheue-yann Cheng^§

From the Gene Regulation Section, Laboratory of Molecular Biology, NCI and  Laboratory of Cell Biochemistry and Biology, NIDDK, National Institutes of Health, Bethesda, Maryland 20892-0851

Received for publication, May 14, 2001, and in revised form, September 13, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The functions of thyroid hormone receptors (TRs) are regulated by a host of co-regulatory proteins. Tissue-specific expression of these co-regulators leads to distinct expression patterns and regulation of thyroid hormone (T3) target genes in tissues. Previously we have found that human colon carcinoma RKO cells exhibit strong T3-independent transcriptional activity. We therefore searched for co-regulatory proteins in RKO cells using a yeast two-hybrid system with the intact TR1 as bait. One of the three positive clones, designated as P3, was identified to be an isoform of human mitochondria branched-chain aminotransferase (BCATm). P3 was a spliced variant of BCATm with an internal 12-amino acid deletion near the carboxyl-terminal region and was abundantly expressed in RKO cells. The expressed protein localized both to the mitochondria and the nucleus of transfected CV1 cells. P3 physically interacted with TR1 in a T3-independent manner that led to the inhibition in binding of TR1 to thyroid hormone-responsive element. P3 not only enhanced the repressor activity of the unliganded TR but also repressed the ligand-dependent activation of TR. This repression was reversed by treatment of cells with trichostatin A, suggesting that in addition to the inhibition of DNA binding, the repression activity of P3 on TR may also be mediated by histone deacetylase activity. Thus, unlike the currently known co-repressors, P3 is a novel ligand-independent co-repressor for TR.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Thyroid hormone receptor (TR)¹ is a member of the nuclear hormone receptor superfamily that acts as ligand-dependent transcription factor to control cell proliferation, differentiation, and homeostasis (1). Two distinct genes, TR and TR, give rise to four hormone-binding TR isoforms (1, 1, 2, and 3) by alternative splicing (1, 2). The tissue specificity and differential expression of TR isoforms indicate that they play distinct functional roles in vivo (3, 4). This is demonstrated by the finding that TR-knockout mice show very different phenotypes from TR1-deficient mice. TR regulates transcription by binding to thyroid hormone response element (TRE) as a homodimer or heterodimer in the promoter region of target genes. The transcriptional activity of TR relies on the types of TRE as well as on T3. In the presence of T3, TR functions as a transcriptional activator that binds to positive TREs and as a repressor that binds to negative TREs.

Modulation of gene expression by TR involves the coordination of a network of co-regulatory proteins, including co-activators and co-repressors. Binding of ligand to TR induces a conformational change that promotes dissociation of the co-repressors and recruitment of the co-activators, leading to gene activation. Unlike other nuclear hormone receptors, TR and retinoic acid receptors are able to bind to their target genes in the absence of ligands and actively repress transcription (5-8). The repression is mediated by a silencing domain in the carboxyl terminus (5, 9-11) that also harbors several other functions, including hormone binding, dimerization, and hormone-dependent activation. Although the mechanisms by which TR repress basal transcription are not fully understood, studies suggest that TR interacts with various regulatory proteins to exert its repression functions. Interaction with components of basal transcription machinery, including TATA binding protein and TFIIB, and disruption of the pre-initiation complex formation were shown to be involved in gene silencing by TR (8, 12-14). However, mutations in the hinge region of TR that completely abolished ligand-independent repression did not interfere with the interaction of TFIIB (15), suggesting that interaction with transcription factors is not solely responsible for gene repression by TR. Investigations of active repression by unliganded TR and retinoic acid receptor have led to the identification of several co-repressors, N-CoR (15), SMRT (16-18), and Alien (19), by a yeast two-hybrid screen. The repressive activity of N-CoR and SMRT is manifested through their interaction with Sin3 that associates with histone deacetylase (HDAC) to form a co-repressor complex. Recently, isolation and characterization of HDAC3, HDAC5, and HDAC7 showed that SMRT and N-CoR-mediated repression is also promoted by HDAC in a Sin3A-independent fashion (20-23). Deacetylation of histones is thought to condense chromatin that then becomes less accessible to transcription factor, leading to gene repression.

To explore the mechanism of repression by TR, we searched for co-regulatory protein in human colon carcinoma RKO cells by yeast two-hybrid screen using intact TR1 as bait. In this study, we report the identification and characterization of a novel co-repressor for TR that interacts with TR in a hormone-insensitive manner. This gene, designated as P3, encodes an isoform of human mitochondria branched-chain aminotransferase (BCATm) that catalyzes the first step in degradation of branched chain amino acids (24, 25). We provide evidence that P3 cDNA is derived from BCATm by alternative splicing in human cell lines. Importantly, we show that P3 was localized not only to the mitochondria but also that a significant fraction of P3 was localized in the nucleus.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Plasmids-- The following plasmids have been described previously: pCLC51, pCLC61, and pCDMTR2, which express human TR1, TR1, and rat TR2, respectively (20, 27); pTK-Pal-Luc, which contains two TRE-Pal-binding sites and a luciferase gene under the control of thymidine kinase promoter; pCEP4F contains the CMV early promoter and an ATG followed by the FLAG epitope (20); pCJ3, pJL8, pJL5, and pCJ4, which contain full-length and truncated TR1 cDNA downstream of the phage T7 promoter (27).

The pCMV-P3 plasmid, carrying both the T7 and cytomegalovirus promoters, was constructed as follows. First, a double-stranded oligonucleotide (5'-CCATGGCTCGAGGATCCGAATTCCAGATCTATGCTAGCCCTAGG-3'), containing multiple cloning sites and an ATG signal (underlined), was ligated to the pExpress-shutter vector at NcoI and BamHI sites. Subsequently, pExp-P3 was constructed by releasing the P3 cDNA from pGAD10-P3 and subcloning it into the pExpress-shutter vector digested with EcoRI. The P3 fragment (HindIII-NheI) was cut out from the pExp-P3 vector and ligated to the HindIII-NheI site of the pCMV-TR1 to produce the pCMV-P3.

To construct pGST-P3, P3 cDNA was isolated and cloned into the EcoRI site of the pGEX-6P-1 vector (Amersham Pharmacia Biotech), in-frame with the glutathione S-transferase (GST) polypeptide. pCEP4F-P3 was generated by isolating the P3 cDNA with SalI digestion and subcloning it into the XhoI site of the pCEP4F vector to produce an in-frame fusion protein with FLAG tagged at the amino terminus. All constructions were verified by restriction enzyme digestions and DNA sequencing.

Yeast Two-hybrid Assay-- The human colon carcinoma RKO cDNA library in pGAD 10 vector (CLONTECH Laboratories, Inc.) was constructed according to the manufacturer's protocol (28). The full-length TR1 was inserted into the NdeI and EcoRI site of pAS2-1 vector to generate Gal4DNA-BD-TR1 as the bait. Both plasmids were transformed into yeast Y190 strain and selected on histidine(), leucine(), and tryptophan() medium. After 4 days, colonies were picked and confirmed by -galactosidase assay. The cDNA from positive clones were isolated and subjected to sequencing.

Cell Culture and Transient Transfection-- Monkey kidney CV1 cells and human colon carcinoma RKO cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS), 100 µg/ml penicillin, 0.25 µg/ml streptomycin, and 2 mM L-glutamine at 37 °C in 5% CO₂. To prepare thyroid hormone-depleted FBS (Td-FBS), 500 ml of FBS was incubated with 20 mg/ml activated charcoal and 50 mg/ml anion exchange resin (AGX-8, Bio-Rad) for 24 h followed by centrifugation and ultrafiltration.

3 × 10⁵ cells were seeded into a 6-well plate 1 day before transfection in DMEM with 10% Td-FBS. Cells were transfected with 1 µg of pTK-Pal-Luc, 0.25 µg of TR, and P3-expressing vector using FuGENE 6 transfection reagent (Roche Molecular Biochemicals). Empty vectors were used to supplement equal amounts of DNA in each transfection. After overnight incubation, cells were replaced with fresh DMEM with 10% Td-FBS and induced by 100 nM T3. After 24 h, cells were lysed with reporter lysis buffer (Promega) and assayed for luciferase activity and total protein. All experiments were performed in duplicates and repeated 3-5 times. The results shown are the means ± S.E.

Northern Blot Analyses-- Total RNA was isolated from cell lines by RNA miniprep kit (Qiagen) according to the manufacturer's protocol. Ten µg of RNA was analyzed in 1% agarose gel containing 2.2 M formaldehyde. A Northern blot containing RNA from normal human tissues was purchased from CLONTECH Laboratories. The -³²P-labeled probes were prepared by random prime kit (Stratagene).

RT-PCR Analysis-- RNA was isolated from tissue culture cells by TRIzol reagent (Life Technologies, Inc.). The first-stranded cDNA synthesis was catalyzed by SuperScript II Reverse Transcriptase (Life Technologies, Inc.) using 5 µg of total RNA and poly(dT) as a primer. The target cDNA was amplified with a pair of primers: BCAT-5' primer, 5'-ACGCCCCCGCTGAATGGTGTTATC, and BCAT-3' primer, 5'-GTGCTGGCGTGACGAGATGCTACG. PCR was performed as follows for 35 cycles: denaturation at 94 °C for 30 s, annealing at 60 °C 66 °C for 30 s, and amplification at 74 °C for 30 s. The amplified products were analyzed in 3% agarose gel in TAE buffer (Tris acetate, 1 mM EDTA, pH 8.3)

GST Pull-down and Co-immunoprecipitation Assay-- Expression of GST, GST-TR1 was induced by 1 mM isopropyl--D-thiogalactopyranoside in BL21 cells. The expressed proteins were purified and immobilized by glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) according to the manufacturer's protocol. In vitro translated proteins were synthesized by TNT-coupled reticulocyte lysate system (Promega), and the translation efficiency was verified by SDS-polyacrylamide gel electrophoresis and autoradiography. For binding assay, beads were incubated with in vitro translated [³⁵S]methionine-labeled proteins (intact TR1 and its truncated proteins lacking domain A/B, A/B +C, or A/B + C + D) (27) using TNT Kit (Promega, Madison, WI) in 500 µl of buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.1% Lubrol, 2 mM dithiothreitol, 0.05% BSA, 5% glycerol, and protease inhibitors) at 4 °C overnight with constant rotation. Subsequently, beads were washed 4 times in 25 mM Tris-HCl, pH 7.4, 2.5% sucrose, 2.5 mM EDTA, 250 mM NaCl, and 1% Lubrol. The bound proteins were eluted out by boiling the beads in SDS sample buffer and analyzed on SDS-polyacrylamide gels. The proteins were detected by autoradiography.

The physical interaction of P3 with domain E of TR1 was further evaluated by co-immunoprecipitation assay. In vitro translated [³⁵S]methionine-labeled domain E (T7 expression plasmid, pCJ4) prepared similarly as described above (10 µl lysates) and P3 (10 µl lysates) were incubated in 100 µl of buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.1% Lubrol, 2 mM dithiothreitol, 0.05% BSA, 5% glycerol, and protease inhibitors) at 20 °C for 2 h. Monoclonal anti-TR antibody (mAbC4; 2 µg; 29) or an irrelevant antibody (MOPC; 2 µg; Sigma) was added and incubated for an additional 2 h at 4 °C. The mixture was further incubated with 20 µl of protein G-agarose beads (Roche Molecular Biochemicals) for 1 h at 4 °C. The beads were spun down and washed 4 times with buffer containing 25 mM Tris-HCl, pH 7.4, 2.5% sucrose, 2.5 mM EDTA, 250 mM NaCl, and 1% Lubrol. The bound proteins were eluted out by boiling the beads in SDS sample buffer and analyzed on SDS-polyacrylamide gels. The proteins were detected by autoradiography.

Immunoprecipitation and Western Blot Analysis-- Cells were washed in phosphate-buffered saline and lysed in lysis buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) containing a complete protease inhibitor mixture (Roche Molecular Biochemicals). Cell lysate was clarified by centrifugation at 14,000 rpm for 5 min at 4 °C. For immunoprecipitation, 300 µg of cell extract was incubated with mAbC4 (29) or anti-FLAG M2 antibodies (Sigma) for 2 h at 4 °C and followed by incubation with protein G-agarose beads (Roche Molecular Biochemicals) overnight. Beads were washed 5 times in lysis buffer and boiled in 2× Laemmli sample buffer. The immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis and transferred to Immobilon membranes (Millipore). The bolt was probed with mouse anti-FLAG M2 antibody, mAbC4, or a rabbit polyclonal antibody against sheep BCATm provided by Susan Hutson (from Bowman Gray School of Medicine, Winston-Salem, NC) (30). After washing, the reactive proteins were detected by ECL reagent (Pierce).

Electrophoresis Mobility Shift Assay (EMSA)-- A double-stranded oligonucleotide containing the F2-TRE-binding sequence was prepared by annealing two synthetic complementary oligomers (5'-AAGGGATCCTTATTGACCCCAGCTGAGGTCAAGTTACG-3' and 5'-AGGAGGATCGTAACTTGACCTCAGCTGGGGTCAATAA-3'). The oligonucleotide was labeled by Klenow fragment of DNA polymerase I in the presence of [-³²P]dCTP. About 0.2 ng of probe (3-5 × 10⁴ cpm) was incubated with in vitro translated proteins, 0.15 µg of single-stranded DNA, and 1 µg of nonspecific competitor poly(dI-dC) in 10 µl of binding buffer (10 mM HEPES, pH 7.9, 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 4% Ficoll). Full-length P3 and TR1 were prepared by in vitro transcription/translation kit (wheat germ TNT, Promega). Reactions were carried out at room temperature for 30 min, and complexes were resolved on 6% non-denaturing polyacrylamide in 0.5× TBE (45 mM Tris-HCl, 45 mM boric acid, 0.5 mM EDTA). After electrophoresis, the gel was vacuum-dried and subjected to autoradiography.

To establish the specificity of the effect of P3 on the binding of TR1 to TRE, we also carried out EMSA similarly as described above except in the presence of bovine serum albumin (0.1-1 µg) or the in vitro translated steroid hormone co-activator-1 (1 and 4 µl) (31) as controls. EMSA were also carried out similarly using purified Gal repressor (GalR) and a ³²P-labeled 140-bp DNA containing its target binding sequence, O_I (32), in the presence or absence of P3.

Immunofluorescence and Confocal Localization-- CV-1 cells were seeded in Nunc chamber slides (Fisher, catalog number 2565401) with a density of 2.5 × 10⁴ cells/well. After culturing the cells for 24 h, cells were transfected with FLAG-P3 expression plasmid (pCEPF-P3; 1 µg/well) using Fugene6 (1 µg/50 µl of Opti-MEM) according to the manufacturer's instructions. After 5 h, an equal volume of T3 (100 nM) containing medium was added to the cell cultures. After culturing the cells for 16 h, cells were washed with phosphate-buffered saline and fixed with 3.7% formaldehyde for 30 min at room temperature followed by permeabilization with 0.1% Triton X-100. After washing with phosphate-buffered saline (three times), cells were treated with the anti-FLAG M2 monoclonal antibody (Sigma) at a dilution of 1:20,000 in 5% BSA. Detection was carried out with fluorescein isothiocyanate-labeled goat anti-mouse IgG (20 µg/ml in 5% BSA; Jackson Laboratories). All immunofluorescence localization experiments were performed using Nunc chamber slides. Mitochondria were visualized using the Mitochondrion-selective Probe Mitotracker Red CMXRos (Molecular Probes) and visualized using the rhodamine channel. Cells were examined using an Ultraview Confocal microscope on a Zeiss TV200 microscope (krypton-argon laser).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Characterization of Human P3-- To search for TR-interacting proteins, a yeast two-hybrid system was used with the intact TR1 as bait. The screening led to the isolation of several positive clones from the human RKO cDNA library. DNA sequence analysis of one clone showed an open reading frame of 1143 bp encoding for a protein of 381 amino acids. We referred this cDNA clone as P3 (Fig. 1). While searching the NCBI data base for sequence homology, we discovered that P3 is highly homologous to members of the BCATm family. Among these members, the human BCATm cDNA encodes a mature protein of 366 amino acids that is preceded by a mitochondrial targeting sequence of 27 amino acids (Fig. 1B). DNA sequence alignment of the P3 and human BCATm (GenBank^TM accession number U68418) shows that they share 97% identity except for a 36-bp gap (corresponding to nucleotides 1030-1065 of BCATm) located near the carboxyl terminus of the P3 (Fig. 1A). This suggests that P3 and BCATm may originate from the same gene.

View larger version (40K): [in this window] [in a new window]   Fig. 1.   Comparison of P3 with human BCATm. A, comparison of DNA sequences of P3 and human BCATm. P3 DNA sequence is aligned with human BCATm (GenBankTM accession number U68418) using GCG program. Only the carboxyl-terminal sequences are shown. Identical sequences are aligned and dots represents gaps in P3. The asterisk denotes the deduced stop codon for translation. Primers used in RT-PCR analysis are underlined. The potential 5' and 3' splice sites in human BCATm are indicated in bold letters. B, schematic representation of the human BCATm, P3, and FLAG-P3 proteins. The numbers show the position of the indicated amino acids. The arrow indicates the predicted cleavage site of the mitochondrial targeting sequence (shown as the black box). The predicated splicing junctions are shown as the gray box. A FLAG tag shown as the striped box was added to the amino terminus of the P3 to produce the FLAG-P3.

Analysis of the flanking sequence of the gap region revealed the consensus sequence of the 5' and 3' splice sites (5'-GT and AG-3') for RNA processing. To explore further the possibility that the P3 mRNA resulted from alternative splicing of the human BCATm mRNA, we isolated total mRNA from RKO cells and synthesized the first stranded cDNA by reverse transcription. To enrich the target genes, PCR was performed using primers complementary to the flanking sequences of the gap region. Reactions were carried out at different annealing temperatures to ensure specific gene amplification. Two specific bands corresponding to the predicted sizes for the P3 (345 bp) and the BCATm (381 bp) were detected in all PCRs (Fig. 2A). Evidently, the BCATm mRNA is the dominant form in RKO cells, ~10 times more abundant than the spliced P3 mRNA. To determine whether the splicing event occurs in other human cell lines, total mRNA isolated from human HeLa, Saco-2, MCF7, and 293 cell lines was analyzed accordingly. Similarly, BCATm mRNA was the major form, and detectable P3 mRNA was observed in these cell lines under the conditions stated (data not shown), suggesting that the splicing event is not unique in RKO cells.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Differential expression of P3 and BCATm in human tissues and cell lines. A, detection of the P3 and BCATm mRNA by RT-PCR. Total RKO RNA was synthesized into cDNA by reverse transcription and amplified as described under "Experimental Procedures." PCRs were performed at different annealing temperatures as indicated at the top. Sizes of DNA 1-kilobase pair (kb) markers are shown on the left. B, Northern blot analysis of BCATm/P3 in human tissues and cell lines. Ten µg of total RNA samples from selected human tissues and cell lines were separated on formaldehyde gel and hybridized with the 32P-labeled P3 probe. The size markers are indicated on the right. C, detection of endogenous BCATm/P3 and transfected P3 proteins. Lysates from untransfected human cell lines (lanes 1-5) and CV1 cells transfected with FLAG-P3 expression vector or empty vector (lanes 6 and 7, respectively) were analyzed by Western blotting using rabbit anti-BCATm. The loading amount for each cell line is 30 µg except 15 µg for RKO cells.

To explore the expression of BCATm/P3 mRNA in tissues and cultured cells, Northern blot analysis was performed using intact P3 cDNA as a probe. Whereas the P3 and BCATm mRNAs are produced by alternative splicing, they only differ 36 bp in size. As expected, only one major mRNA transcript of about 1.7 kilobase pairs was detected in all normal human tissues and cell lines. The BCATm/P3 was expressed at higher levels in human heart, placenta, skeletal muscle, and pancreas, whereas lower levels were observed in brain, lung, liver, and kidney (Fig. 2B). A significantly higher level of BCATm/P3 was detected in human RKO cells than in rat growth hormone-producing cell (GC) and monkey CV1 cell lines. The low signal in GC and CV cells did not result from lack of sequence homology because the human BCATm shares more than 80% identity in DNA sequence with other cloned mammalian BCATm (33).

Because the human BCATm is highly homologous to sheep BCATm, a rabbit antiserum against mature sheep BCATm protein was used to detect the expression of BCATm and P3 (33). The anti-BCATm antibody detected a 42-kDa band corresponding to the predicted size for endogenous BCATm in several human cell lines, including HeLa, 293, MCF7, RKO, Saos-2 cells, and monkey CV1 cells (Fig. 2C, band a, lanes 1-3 and 5-7). Consistent with the results of RT-PCR and Northern blotting, BCATm and P3 were expressed more abundantly in RKO cells than in other cell lines.

To demonstrate that the cloned P3 cDNA encoding a protein is antigenically related to BCATm, the P3 cDNA was cloned into the pCEP-F expression vector, which encodes a amino-terminal FLAG-tagged P3 fusion protein and transfected into CV1 cells. Surprisingly, two bands were detected by anti-BCATm antibody. The lower band co-migrated with the endogenous BCATm/P3 (band a, lanes 6 and 7) and the upper band migrates approximately at the size of 45 kDa (band b, lane 6). When the bolt was re-probed with anti-FLAG, only the upper band was detected (data not shown), suggesting that the FLAG-P3 expressed an isoform of BCATm that was not proteolytically processed in CV1 cells (Fig. 1B). The failure of cleavage most likely results from interference of the FLAG peptide fused upstream of the mitochondrial targeting sequence (Fig. 1B). Taken together, these data confirm that P3 encodes an isoform of human BCATm and is derived from the BCATm gene by alternative splicing.

Binding of P3 to TR1 in Vitro-- P3 was identified as a TR1-interacting protein by the yeast two-hybrid screening. To confirm the physical interaction between P3 and TR in vitro, we used bacterially expressed GST-TR1 and in vitro ³⁵S-labeled P3 in the binding assay. In agreement with the yeast two-hybrid result, P3 bound strongly to GST-TR1 (Fig. 3A, lanes 3 and 4) but not to GST alone (Fig. 3A, lanes 1 and 2), indicating that the association was specific. Interestingly, addition of 1 µM T3 ligand did not disrupt the interaction between GST-TR1 and P3 (lane 3 versus 4). TR1 consists of several functional domains: the amino-terminal domain (A/B), a highly conserved DNA-binding domain (C), and a hormone-binding domain (D and E) (see also Fig. 3D of Ref. 1). To identify the domain that is required for interaction with P3, we used in vitro translated TR1-truncated proteins (27) in a reciprocal GST pull-down assay. All TR1-truncated proteins were efficiently translated in vitro, and equal amounts these proteins were used in the binding assays. Full-length TR1 and all truncated TR1 bound to GST-P3 (Fig. 3B, lanes 2, 4, 6, and 8) but not to GST alone (Fig. 3B, lanes 1, 3, 5, and 7). These results indicate that the hormone-binding domain E is the critical binding region (see also Fig. 3D). To further support that domain E was responsible for the binding of TR1 to P3, co-immunoprecipitation assay was also carried out by using anti-TR antibody, mAbC4 (29). Fig. 3C shows that in the presence of both P3 and domain E (see also Fig. 3D), P3 was associated with the immunoprecipitated domain E. Lane 4 is a control to indicate that an irrelevant monoclonal antibody, MOPC, did not precipitate any proteins. These results provide additional support to show that domain E of TR1 is the binding site for P3.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Interaction of P3 with TR1 in vitro and in vivo. A, pull-down assay of P3 by GST-TR1. GST or GST-TR1 was incubated with in vitro translated 35S-labeled P3 in the presence or absence of 1 µM T3. B, mapping of the P3 interaction domain of TR1. GST or GST-P3 was incubated with 35S-labeled TR1 mutants. The GST-pull-down assay was carried out as described in A. The lanes are marked. C, the binding of P3 to domain E as shown by co-immunoprecipitation assay. Ten µl each of 35S-labeled domain E and P3 were prepared by in vitro translation and co-immunoprecipitated by using anti-TR mAbC4 as described under "Experimental Procedures." Lane 4 is a control to indicate that an irrelevant monoclonal antibody, MOPC, did not precipitate any proteins. D, schematic representation of TR1-truncated proteins. The amino-terminal domain (A/B), DNA binding domain (C), hinge region (D), and hormone binding domain (E) of TR1 are indicated. The designations for the T7 expression vectors for the intact TR1 and its truncated proteins are also indicated. E, in vivo association of P3 with unliganded TR1. CV1 cells were transfected with the indicated expression vectors of TR1 and FLAG-P3 in the absence of T3. Lysates were immunoprecipitated with anti-FLAG antibody before being immunoblotted for TR. Ten percent of the input was used to detect the expression level of TR1 and P3 as shown in direct Western. F, association of endogenous BCATm/P3 with transfected TR1. Lysates from TR1-transfected CV1 cells were immunoprecipitated with either control IgG or anti-TR antiserum. Samples were immunoblotted with anti-BCATm antibody.

We next determined whether P3 could associate with TR1 in intact cells. Thus, CV1 cells were transfected with plasmids encoding TR1 and FLAG-P3 in the absence of T3. TR1 co-precipitated with FLAG-P3 using an antibody against the FLAG epitope (Fig. 3E, lane 3). To monitor the expression of TR1 and FLAG-P3, lysates from the transfected cells were analyzed by direct Western blotting (Fig. 3E, lanes 4-6). We estimated ~2% of the expressed TR-forming complex with FLAG-P3 in CV1 cells. Similar association of FLAG-P3 with TR1 in CV1 cells was observed in the presence of T3 (data now shown). The T3-independent association of P3 with TR1 detected in cells is consistent with the results found by GST pull-down assay in vitro.

Finally, we examined interaction of endogenous BCAT/P3 with transfected TR1. Endogenous BCAT/P3 was precipitated with anti-TR antibody in cells transfected with TR1 but not with empty vector (Fig. 3F, lane 1 versus 2). No BCATm/P3 was precipitated with normal mouse IgG in cells transfected with TR1 (Fig. 3F, lane 3). Together, these results demonstrate that TR1 interacts with P3, through the hormone-binding domain, in a ligand-insensitive manner.

P3 Localizes to the Nucleus and Mitochondria of Transfected Cells-- The intracellular localization of P3 was examined by immunofluorescence after transfecting CV1 cells with FLAG-P3 expression plasmid. As indicated by the arrows (Mi) in Fig. 4A, P3 was localized to the mitochondria. Importantly, P3 was also found to localize in the nucleus (indicated by arrows, Nu). To confirm further the localization of P3 in the mitochondria, we carried out similar immunofluorescence experiments except that living cells were first treated with Mitotracker (a mitochondrion-selective red marker) before immunostaining with anti-FLAG for FLAG-P3. Fig. 4B shows the merged image (indicated by yellow), indicating substantial coincidence of confocal labeling by the Mitotracker (D, viewed using red rhodamine channel) and anti-FLAG P3 (C, viewed using green fluorescein channel).

View larger version (70K): [in this window] [in a new window]   Fig. 4.   Expression and immunofluorescent localization of FLAG-P3 in the nucleus and mitochondria. CV1 cells were transfected with expression plasmid for FLAG-P3 in the absence of T3. Cells were fixed with 3.7% formaldehyde and permeabilized with 0.1% Triton X-100. Immunofluorescence images were obtained as described under "Experimental Procedures." A, using confocal microscopy, P3 was localized to both nuclei (arrows, Nu) and to extended cytoplasmic structures morphologically resembling mitochondria (arrows, Mi). B, co-localization of P3 with mitochondria using MitoTracker Red CM-H3Xros. Living cells were incubated with MitoTracker for 10 min according to the manufacturer's instructions and fixed. Localization of FLAG-P3 was carried out as described under "Experimental Procedures." This image represents the merge of the rhodamine (red) and fluorescein (green) channels showing substantial co-localization in the confocal microscope as represented by yellow. C, localization of P3 using anti-FLAG (fluorescein; green). D, localization of mitochondria using the MitoTracker (Rhodamine channel; red). The bars at the lower left in A-D are 10µm.

P3 Enhances the Silencing Mediated by TR1 and Functions as a Transcriptional Repressor-- To study the ability of P3 to influence the silencing activity of TR1, CV1 cells were transiently transfected with a luciferase reporter containing 2 copies of inverted repeat TRE (TRE-pal) together with the TR1 and the FLAG-P3 expression vectors in the absence of T3. As expected, unliganded TR1 behaved as a repressor of the TRE-mediated transcription (Fig. 5A, bars 1 versus 2). FLAG-P3 itself repressed the TRE-luciferase activity by 3-fold (Fig. 5A, bar 3). When FLAG-P3 was co-expressed with TR, the repression was synergistically enhanced (15-fold, bar 4 versus 2 and 3). These data show P3 as a transcriptional co-repressor that augments the silencing mediated by TR1.

View larger version (14K): [in this window] [in a new window]   Fig. 5.   P3 harbors intrinsic repression activity. A, P3 enhancing the silencing effect of unliganded TR1. Luciferase reporter together with TR1 expression plasmid, FLAG-P3, or empty vector was transfected into CV1 cells. Fold repression was calculated as relative to cells co-transfected with empty vectors. B, repression of P3 on TRE-mediated transcription. CV1 cells were transfected with pTK-Pal-Luc reporter (1 µg) and the FLAG-tagged P3 expression vector (0.1, 0.2, and 0.4 µg). Fold repression was calculated relative to cells transfected with reporter construct. Data are means ± S.D. (n = 3).

The repression effect shown in Fig. 5A raises the question whether P3 itself harbors intrinsic silencing activity. CV1 cells were transiently transfected with FLAG-P3 expression plasmid along with the TRE-luciferase construct. Expression of FLAG-P3 repressed basal transcription from the TRE-luciferase reporter in a dose-dependent fashion (Fig. 5B), indicating that P3 harbors intrinsic silencing activity.

P3 Represses TR Transactivational Activity-- To determine whether P3 represses the ligand-dependent transactivation activity of TR, FLAG-P3 was co-transfected with TR1 and TRE-luciferase reporter into CV1 cells. The luciferase activity was activated by TR1 in the presence of T3. However, the activation of luciferase activity by TR1 was greatly reduced by increasing the expression plasmids of FLAG-P3 (Fig. 6A). To see whether TR1 expression was affected by the presence of FLAG-P3, cell lysates from each transfection were subjected to Western blot analysis. The expression level of TR1 was not affected by the co-expression of P3 (data not shown), ruling out the possibility that the reduced activation of luciferase reporter is a consequence of lower expression of TR.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Repression of TR1 transcriptional activity by P3. A, P3 represses T3-dependent activation by TR1. CV1 cells were transfected with pTK-Pal-Luc, TR1 expression plasmid, and increasing amounts of FLAG-P3. After transfection, cells were induced by 100 nM T3 for 24 h before harvesting. The relative reporter activity in the absence of T3 was set as 1. The reporter plasmid pTK-Pal-Luc (1 µg) together with pCMV-P3 (0.2 or 0.4 µg) and TR1 (0.2 µg) expression vectors were transiently transfected into CV1 (B) and RKO (C) cells. The fold activation is calculated relative to that of reporter activity in the absence of T3, which was set to 1. Data represent means ± S.D. (n = 3).

To ask whether the P3-mediated repression depends on the backbone of expression vector, we generated a second P3 expression vector by cloning the P3 cDNA into the pcDNA3.1 vector. As shown in Fig. 6B, TR1 alone repressed the basal transcription of TRE-luciferase in the absence of T3 and activated luciferase activity in the presence of T3. Co-transfection of pCMV-P3 further enhanced the gene silencing by the unliganded TR. In the presence of T3, P3 significantly repressed the transactivation activity of TR1 in CV1 cells. Comparable but less significant repression was seen in RKO cells (Fig. 6C). The lower repression effect of P3 on TR activity could be attributed to the high expression of endogenous BCATm/P3 in RKO cells (Fig. 2C), suggesting the extent of repression is cell type-dependent. Similarly, P3 also repressed TR transcriptional activity on reporters containing Lys-TRE (an inverted TRE) in CV1 cells (data not shown).

Previously, we mapped the hormone binding domain and DNA binding domain of TR1 as the critical interaction regions with P3. Sequence comparison shows that TR isoforms share high homology in their DNA binding and ligand binding domains. However, very little or no sequence similarity exists in the amino-terminal activation domain among the TR isoforms. We therefore determined whether P3 represses the transcriptional activity of other TR isoforms. As shown in Fig. 7, the repression effect of P3 was in the order TR2 > TR1 > TR1, indicating that the extent of the repression by P3 is TR isoform-dependent.

View larger version (24K): [in this window] [in a new window]   Fig. 7.   Repression of transcriptional activity of TR isoforms by P3. CV1 cells were transfected with pTK-Pal-Luc reporter (1 µg), TR expression vector (0.2 µg), and pCMV-P3 (0.4 µg) or empty vector. As indicated, cells were treated with 100 nM T3 for 24 h. The fold activation is calculated relative to that of reporter alone without adding T3, which was set to 1. Data represent means ± S.D. (n = 3).

P3 Inhibits the Binding of TR1 to TRE-- To understand the mechanisms by which P3 repressed the T3-dependent transactivation, we first analyzed the TR-TRE complexes in response to P3 using EMSA. TR1 and P3 proteins were generated in vitro by a rabbit reticulate lysate transcription/translation kit, and the amounts of the translated proteins were determined by SDS-polyacrylamide gel electrophoresis. As shown in Fig. 8A, binding of TR1 homodimer to ³²P-labeled TRE probe was enhanced as an increasing amount of TR1 was added to the reactions (lanes 2 and 3). However, TR1 binding was greatly reduced when increasing amounts of P3 were present (lanes 6-9). Unlike TR1, P3 did not show any direct binding to TRE (lanes 4 and 5), indicating that inhibition of TR1 binding to TRE by P3 is not resulting from a direct competition for the same DNA-binding site.

View larger version (37K): [in this window] [in a new window]   Fig. 8.   Binding of TR1 to TRE is specifically inhibited by P3. A, inhibition of TR1 homodimer binding to TRE by P3. The 32P-labeled TRE was incubated with in vitro translated TR1 (lanes 2 and 3), P3 (lanes 4 and 5), or TR1 in the presence of increasing concentrations of P3 (lanes 6-9; 1-4 µl, respectively), and subjected to EMSA as described under "Experimental Procedures." An equal amount of unprogrammed lysates was used as control shown in lanes 1 and 10. The same amount of TR1 (1 µl) of translated lysates was used in lane 2 and lanes 6-9. TR1 homodimer complex is indicated by arrows. B, the binding TR1 to TRE is inhibited by P3 but not by BSA or SRC-1. Binding of the in vitro translated TR1 (1 µl) to 32P-labeled TRE without (lane 1) or with in vitro translated P3 (lysates 1 and 4 µl in lanes 3 and 4, respectively). Binding of the in vitro translated TR1 (1 µl) to 32P-labeled TRE was in the presence of BSA (0.1, 0.5, and 1 µg in lanes 5-7, respectively) or in the presence of in vitro translated SRC-1 (lysates 1 and 4 µl in lanes 8 and 9, respectively). Lane 10, no binding of SRC-1 (4 µl) to 32P-labeled TRE. C, P3 does not affect the binding of GalR to its target DNA element. Binding of purified GalR (10 nM) to 32P-labeled 140-bp DNA fragment carrying its binding sequence OI (1 nM; lane 2); no binding of P3 (4 µl) to OI (lane 3); P3 (4 µl) had no effect on the binding of GalR to OI (lane 4). Lane 1 shows the control in which only lysates and DNA probe were present. Only nonspecific (NS) band was detected as marked. D, reduction of TR1/RXR heterodimer binding to TRE by P3. Assay was performed as described in A. Increasing amounts of RXR were incubated with in vitro translated TR1 (lanes 2 and 3) or P3 (lanes 4 and 5); the inhibition of TR1-RXR complex formation by increasing concentrations of P3 (1, 2, and 4 µl for lanes 6-8, respectively). An equal amount of TR1 was used in the binding reactions where indicated. Arrows mark the TR1 homodimer, monomer, and heterodimer complexes.

To establish the specificity of the inhibitory effect of P3 on the binding of TR1 binding to TRE, we carried out similar EMSA in the presence of increasing concentrations of BSA or the steroid hormone co-activator-1 (SRC-1; Fig. 8B). As shown lanes 5-7 in Fig. 8B, the presence of increasing amounts of BSA had no effect on the binding of TR1 binding to TRE, whereas the increasing amounts of P3 decreased the binding of TRE to TR1 (lanes 2-4 of Fig. 8B). As an additional control, we also carried out EMSA in the presence of increasing amounts of SRC-1 in the absence of T3 (lanes 8 and 9 of Fig. 8B), and no effect on the binding of TR1 to TRE (lanes 8 and 9 versus lanes 2-4). Lane 10 of Fig. 8B shows that SRC1 itself did not bind to TRE.

To demonstrate further the specificity of the inhibitory effect of P3 on the binding of TR1 to TRE, we used a well characterized DNA-binding protein, the Gal repressor (GalR; see Ref. 32). Lane 2 of Fig. 8C shows that the purified GalR bound to its target DNA element, O_I (32), whereas P3 did not bind to the GalR-binding element O_I (lane 3 of Fig. 8C). Lane 4 shows that same amount of P3 that inhibited the binding of TR1 to TRE (see lane 9 of Fig. 8A and lane 4 of Fig. 8B) did not affect the binding of GalR to O_I. Lane 1 was a control to indicate that only a nonspecific band was detected when the DNA probe was incubated with the unprogrammed lysates. These data provide additional control to indicate that P3 did not affect the binding of other DNA-binding proteins to its target DNA element.

Heterodimerization of TR and RXR has been shown to modulate TR transactivation activity. To assess whether P3 affected TR1/RXR binding to the Lys-TRE sequence, bacterially expressed RXR was incubated with TR1 in the binding reactions (Fig. 8D). Heterodimer of TR1 and RXR increased as more RXR was added (lanes 2 and 3 of Fig. 8D). Incubation of P3 and RXR did not result in any binding to the TRE probe (lanes 4 and 5 of Fig. 8D). As shown in lanes 6 and 7 (Fig. 8D), binding of TR1/RXR heterodimers to Lys-TRE was decreased by P3 in a concentration-dependent manner. Taken together, these data suggest that the suppression action of P3 on TR transcriptional activity is, at least in part, due to inhibition of TR1 binding to TRE by P3.

Role of HDAC in P3 Repression-- To investigate whether the P3-mediated repression of TR1 involves HDAC, we transfected CV1 cells with TRE-luciferase reporter with TR1 alone or TR1 and FLAG-P3 and treated cells with T3 and 100 nM trichostatin A (TSA), a specific inhibitor of HDAC. In the absence of TSA, co-transfection of P3 resulted in a 7-fold reduction of T3-dependent activation by TR1. P3-mediated repression of TR1 activation was significantly relieved by TSA treatment (Fig. 9A). We asked whether the silencing activity of P3 also involves HDAC activity. As shown in Fig. 9B, P3-mediated repression of TRE-luciferase activity was dramatically reversed by TSA. These results suggest that the repressor activity of P3 may also be mediated by HDAC activity.

View larger version (13K): [in this window] [in a new window]   Fig. 9.   Reversal of P3 repression by histone deacetylase inhibitor TSA. A, pTK-Pal-Luc and TR alone or together with FLAG-P3 expression vector was transfected into CV1 cells. After transfection, cells with 100 nM T3 and in the presence or absence of 100 nM TSA for 24 h. Relative activation was determined relative to cells transfected with TR in the absence of T3. B, CV1 cells were transfected pTK-Pal-Luc and FLAG-P3 or empty vector. Luciferase activity was measured in the absence or presence of 100 nM TSA.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The redundancy of regulatory function is commonly observed in the biological systems. We therefore used a yeast two-hybrid screen to search for additional co-regulators for TR1 in human RKO cells. Here we identified a novel co-repressor for TR, designated as P3. Unlike SMRT, N-CoR, and Alien which only associate with TR in the absence of ligand (15), P3 physically interacts with TR in a T3-independent manner. P3 enhances TR-mediated silencing and represses T3-dependent transcriptional activity of TR. Taken together, our data show that P3 represents a member of a new class of co-repressor.

We identified P3 as a shorter isoform of the human BCATm. The human BCATm genes have been isolated and mapped to chromosome 19 (33), but the genomic organization is not yet explored. In this study, DNA sequence analysis shows the P3 is identical to the BCATm except for a 36-bp deletion at the carboxyl-terminal region. We hypothesized that P3 may have originated from the human BCATm gene. This is confirmed by the identification of RNA spliced sites at the junction of the gap region (Fig. 1A). This was further confirmed by RT-PCR analysis that yielded two fragments corresponding to the predicated sizes for BCATm and P3 cDNA (Fig. 2A). In support of our findings, it was reported that three different isoforms of BCATm were detected in rat tissues by Western blot analysis (25).

In mammals, there are two BCAT isoenzymes encoded by two different genes, a BCATm and a BCAT cytosolic (BCATc) form, to catalyze the first step in the catabolism of branched chain amino acids, leucine, isoleucine, and valine, to -keto acids (24, 25, 34). BCATm is the dominant isoenzyme and is expressed in all tissues. In contrast, BCATc is highly expressed during embryogenesis but is limited to only a few adult tissues.

Multiple functions have been ascribed to BCATm. It has been shown to catalyze the transamination reaction as well as to transport branched-chain -keto amino acids (35). The finding that BCATc is highly expressed in proliferating cells but not in arrested cells, whereas BCATm is expressed constitutively through the cell cycle (36, 37) implicates that BCATm may act as a housekeeping gene. Recently, yeast BCATm that shares 49% homology to its human homolog was shown to regulate cell growth and transition of G₁ to S phase in the cell cycle (38, 39). Yeast cells deleted of BCATm show a faster growth rate and a shorter G₁ phase. In contrast, cells depleted of BCATc do not show any growth phenotype. This suggests that BCATm may be involved in maintaining normal cell growth. It is not clear how BCATm regulates the cell cycle progression.

In this study, we discovered that P3, a shorter isoform of BCATm, can act as a co-repressor for TR-mediated transcription. This repressor function is supported by the findings that P3 not only localized to the mitochondria, but also localized to the nucleus to affect the function of TRs. However, the mechanism by which P3 mediates repression is still unclear. We observed that P3 reduced the binding of TR1 homodimer and heterodimer to TRE in gel shift assay. The decreased binding activities are not likely mediated by a direct competition of P3 for TRE binding because P3 fails to bind to TRE. It is possible that the association with P3 induced a conformation change of TR1 in that it lost the binding ability to TRE. This mode of action is not without precedent. Previously, we also found that the tumor suppressor, p53, physically interacts with TR1 and inhibits the binding of TR1 to TREs (26). Interestingly, we further discovered that P3 itself harbors an intrinsic repression activity. Our results show that the silencing activity of P3 is reversed by TSA, an HDAC inhibitor. This suggests that P3 may recruit complexes containing HDAC to repress basal transcription. However, we cannot rule out the possibility that P3 could interact with basal transcription factors and interfere with the assembly of pre-initiation complex. Meanwhile, we have shown that TSA treatment reversed the repressive effect of P3 on TR transcription activation in the presence of ligand. This may be explained by the finding that the interaction of TR with P3, which has recruited HDAC complexes, was not disrupted by the binding of ligand. It is possible that the P3-HDAC complexes interfere with the binding of co-activators to TR and thus prevent further activation. These data provide a link between P3-mediated repression and HDAC activity. Taken together, our results suggest that P3 acts as a repressor for TR-mediated transcription, most likely via multiple mechanisms. The elucidation of these mechanisms awaits future studies.

	ACKNOWLEDGEMENTS

We thank Dr. Xuguang Zhu for the construction of RKO cDNA library and assistance in the identification of the P3 clone. We thank Dr. Edward Seto for the pCEP-F vector and Dr. William Chin for the pCDMTR2 expression vector. We are grateful to Dr. Susan Hutson for the antiserum against sheep BCATm. We thank Drs. S. Sanskar and S. Semsey for providing us with the purified GalR and DNA binding fragment and Dr. Dona C. Love for assistance with the Mitotracker experiments and confocal microscope.

	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: Laboratory of Molecular Biology, Bldg. 37, Rm. 2D-24, 37 Convent Dr. MSC 4255, NCI, National Institutes of Health, Bethesda, MD 20892-4255. Tel.: 301-496-4280; Fax: 301-480-9676; E-mail: sycheng@helix.nih.gov.

Published, JBC Papers in Press, September 26, 2001, DOI 10.1074/jbc.M104320200

	ABBREVIATIONS

The abbreviations used are: TR, thyroid hormone receptor; BCATc, branched-chain aminotransferase cytosolic; BCATm, BCAT mitochondrial; FBS, fetal bovine serum; GST, glutathione S-transferase; HDAC, histone deacetylase; Td-FBS, thyroid hormone-depleted FBS; TRE, thyroid hormone response element; TSA, trichostatin A; RT-PCR, reverse transcriptase-polymerase chain reaction; BSA, bovine serum albumin; EMSA, electrophoresis mobility shift assay; bp, base pair; CMV, cytomegalovirus; DMEM, Dulbecco's modified Eagle's medium; mAb, monoclonal antibody; RXR, retinoid X-receptor.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES