Leukemia/Lymphoma-related Factor, a POZ Domain-containing Transcriptional Repressor, Interacts with Histone Deacetylase-1 and Inhibits Cartilage Oligomeric Matrix Protein Gene Expression and Chondrogenesis*

Mutations in the human cartilage oligomeric matrix protein (COMP) gene have been linked to the development of pseudoachondroplasia and multiple epiphyseal dysplasia. We previously cloned the promoter region of the COMP gene and delineated a minimal negative regulatory element (NRE) that is both necessary and sufficient to repress its promoter (Issack, P. S., Fang, C. H., Leslie, M. P., and Di Cesare, P. E. (2000) J. Orthop. Res. 18, 345–350; Issack, P. S., Liu, C. J., Prazak, L., and Di Cesare, P. E. (2004) J. Orthop. Res. 22, 751–758). In this study, a yeast one-hybrid screen for proteins that associate with the NRE led to the identification of the leu-kemia/lymphoma-related factor (LRF), a transcriptional repressor that contains a POZ (poxvirus zinc finger) domain, as an NRE-binding protein. LRF bound directly to the NRE both in vitro and in living cells. Nine nucleotides (GAGGGTCCC) in the 30-bp NRE are DNA, one-tenth of the DNA was PCR-amplified using forward primer 5 (cid:2) -GTAGTCAATAGGCCTGG-GAAGA and reverse primer 5 (cid:2) -GGAAAGGGAGTCAGAACTGAAG. Thirty-five cycles of PCR at 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s were performed. PCR products were analyzed on 1% agarose gel. Reporter Gene Assay— cells grown to (cid:1) 50% confluence in 35-mm culture dishes were transfected with 1 (cid:4) g of reporter construct ( (cid:3) 1925COMPluc, (cid:3) 1775COMPluc, (cid:3)(cid:4) 1925COMPluc, (cid:3) 592COMPluc, or 4xNREpGL2-Promoter) along with 1 (cid:4) g of pSVGal plasmid (internal control) and various amounts of pFLAG-LRF expression plasmid. To examine whether HDACs are involved in the LRF-mediated repression, various amounts of trichostatin A (TSA) (see C were added to the medium. At 48 h after transfection, the cultures were harvested and lysed. Luciferase assays were performed using 20 (cid:4) l of cell extract and 100 (cid:4) l of luciferin substrate (Promega). (cid:1) -Galactosidase assays were performed using a (cid:1) -galactosidase assay kit (Tropix Inc., Foster City, CA) following the manufacturer’s protocol. (cid:1) -Galactosidase and lucifer- ase activities were measured using a Mini-Lum luminometer (Bioscan, Inc., Washington, D. C.). Magnetic Bead Assay— The magnetic bead assay was performed as (50). Briefly, a 592-nucleotide proximal region ( (cid:3) 592 to (cid:3) 1) and a 400-nucleotide distal region ( (cid:3) 1925 to (cid:3) 1526, -ACAGTCTTCTGAGTG-GCAGTGA-3 rat GAPDH. PCR products were visualized on 1% agarose gels containing 0.1 mg/ml ethidium bromide using ultraviolet The identity of each targeted PCR amplification product was confirmed by DNA sequence analysis of gel-purified bands Inc.). Immunoblot Analysis— To characterize anti-LRF serum, 20 (cid:4) g of total cell lysates prepared from HEK293 cells transfected with pFLAG- LRF were mixed with

tenocytes, and ligament cells, is a fundamental molecular event of both embryonic development and repair of cartilage, ligament, tendon, and bone (1,2). After commitment to the chondrocyte lineage, mesenchymal cells undergo condensation, cease expression of type I collagen, and differentiate into a chondrocytic phenotype characterized by expression of collagen types II, IX, and XI and the proteoglycan aggrecan (1,2). During this process, there appears to be transition cells between type I collagen (expressing mesenchymal cells) and type II collagen (expressing chondrocytes) that are characterized by lack of expression of type II collagen and abundant expression of cartilage oligomeric matrix protein (COMP) 1 (3)(4)(5)(6)(7)(8)(9)(10)(11). These cells may represent musculoskeletal precursor cells that have the potential subsequently to differentiate into a variety of musculoskeletal cell types; however, little is known about the generation of these potential precursor cells.
COMP is synthesized by chondrocytes, osteoblasts, tenocytes, and ligament cells, but not by undifferentiated mesenchymal cells (3)(4)(5)(6)(7)(8)(9)(10)(11)(33)(34)(35)(36)(37)(38). To delineate cis-elements in the COMP promoter necessary for expression in any of these tissues, we cloned the murine COMP promoter and identified cis-elements necessary for expression in the chondrocytic cell line Swarm rat chondrosarcoma (RCS) (35). We have shown that COMP mRNA and protein are expressed in RCS cells, but not in NIH3T3 fibroblasts. A COMP promoter fragment containing ϳ1.9 kb of 5Ј-flanking sequence is specifically active in RCS cells. In cell culture experiments, deletion analysis of the distal region of the COMP promoter identified a silencer region situated between Ϫ1775 and Ϫ1716 that specifically binds protein complexes expressed in non-chondrocytic cells, but not in RCS cells. Competition gel shift experiments localized the binding site to within 30 bp (Ϫ1775 to Ϫ1746; negative regulatory element (NRE)). This site is necessary and sufficient to repress COMP expression in fibroblast cell lines (35,39).
Yeast one-hybrid screening has proven to be an effective tool to identify DNA-binding proteins (40,41), including the rat fibronectin gene (42) and Fgf3 promoter (43). To identify proteins that interact with the NRE in the promoter region of the COMP gene, we screened a yeast expression cDNA library using a tandem repeat of four NREs as bait. These experiments identified the leukemia/lymphoma-related factor (LRF) transcriptional repressor as a binding protein at the NRE. LRF is a nuclear protein with an N-terminal POZ (poxvirus zinc finger) domain and a C-terminal Krü ppel-like zinc finger DNA-binding domain (44). LRF was found to be the mouse counterpart of human FBI-1 (factor that binds to the HIV-1 inducer of short transcripts) (45)(46)(47) and the rat osteoclast-derived zinc finger (OCZF) protein (48), with identical functionally important molecular domains.
Many transcription factors repress transcription by recruiting histone deacetylases (HDACs) to chromatin. HDACs are classified into two groups based on structural and functional similarities. Class I HDACs, including HDAC1, are expressed in the nuclei of cells in most tissues (49). In this study, we discovered the transcriptional repressor LRF as a novel regulator controlling COMP gene expression and chondrogenesis and demonstrate that HDAC1 is involved in LRF-mediated gene transcription. LRF, which associates with HDAC1, is the first transcription factor found to bind directly to the COMP gene promoter and to regulate COMP gene expression and chondrogenic differentiation.
The COMP-specific luciferase expression reporter constructs (Ϫ1925COMPluc, Ϫ1775COMPluc, and Ϫ592COMPluc) have been described previously (35). The Ϫ⌬1925COMPluc plasmid, in which the NRE was deleted, was generated by fusion PCR. Two primers with deletion flanking sequences complementary to one another (5Ј-GACA-GCAGGGAGAGGTTGTCCAATTGTAAGAGCCCCAGC-3Ј and 5Ј-CTT-ACAATTGGACAACCTCTCCCTGCTGTCCCCTCCCTA-3Ј) were used in combination with end primers 5Ј-CGGGGTACCACTGGAGGCCTG-GAGGAG-3Ј and 5Ј-GTTGGGCCCTAAAGGGAGCTGTGGGAAAG-3Ј to generate fragments on either side of the region to be deleted. The two fragments were then annealed to one another in a second PCR using the primers at each end of the sequence. The resultant fragment was digested with KpnI and ApaI and subcloned into the KpnI/ApaI sites in Ϫ1925COMPluc to generate the Ϫ⌬1925COMPluc plasmid. The 4xN-REpGL2-Promoter plasmid contains four tandem repeats of the NRE inserted upstream of the SV40 promoter. This construct was generated by subcloning PCR-amplified tandem repeats of four 30-bp sequences (Ϫ1775 to Ϫ1746) into the KpnI/XhoI sites in the polylinker of the pGL2-Promoter vector (Promega, Madison, WI). The mammalian expression constructs pFLAG-LRF and pFLAG-LRF⌬POZ were kindly provided by Drs. R. L. Widom and A. Zelent (44). All constructs were verified by DNA sequencing.
Yeast One-hybrid Screening-A one-hybrid library screen (Clontech) was performed according to the manufacturer's protocol. The pHISi-4NRE, pHISi-1-4NRE, and pLacZi-4NRE bait plasmids were linearized and integrated into the genome of yeast strain YM4271 by homologous recombination. The yeast strain was then transformed with a pPC86 vector-based mouse embryo day 10.5 cDNA library (Invitrogen). Approximately 2.5 ϫ 10 6 yeast transformants were plated on synthetic complete medium lacking histidine and tryptophan and supplemented with 45 mM 3-amino-1,2,4-triazole. Selected clones were subjected to the ␤-galactosidase assay using the colony lift filter method. Plasmids from putative positive clones were recovered from yeast and individually transformed into E. coli DH10B cells for amplification. To eliminate false-positive results, these plasmids were separately introduced into yeast cells containing either bait or the p53-binding site (Clontech), and transformants were tested by ␤-galactosidase assay. Inserts were sequenced, and the DNA sequences were used to query the GenBank TM / EBI Data Bank.
Preparation and Characterization of Affinity-purified Anti-LRF Antibodies-The GST-LRF-(207-277) fusion protein, encoding a segment (aa 207-277) between the POZ domain and zinc finger motifs of LRF, was expressed in E. coli DH5␣, purified on a glutathione-Sepharose column, and subjected to preparative scale SDS-PAGE. The major band was excised and used to immunize rabbits for polyclonal antiserum production (Zymed Custom Antibody, Zymed Laboratories Inc., South San Francisco, CA). To affinity-purify anti-LRF antibodies, the anti-GST activity in the rabbit antiserum was depleted using GST protein immobilized on glutathione-agarose beads. The depleted serum was incubated with Affi-Gel-10 beads (Bio-Rad) to which purified GST-LRF-(207-277) was covalently linked. The bound antibodies were eluted from the beads with 0.15 M glycine buffer (pH 2.5) and immediately neutralized with 1.5 M Tris-HCl buffer (pH 8.0) (55). The affinitypurified antibodies were characterized using in vitro translated LRF in a rabbit reticulocyte lysate (Promega).
Preparation of Nuclear Extracts-Nuclear extracts were prepared from C3H10T1/2 cells and RCS cell lines transfected with the pFLAG-LRF expression plasmid. Cells were harvested by trypsinization, washed with phosphate-buffered saline, pelleted, and resuspended in lysis buffer (10 mM Tris-HCl (pH 8.0), 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, proteinase inhibitors, and 0.3% Nonidet P-40). After 5 min on ice, the lysates were centrifuged at 1000 ϫ g for 5 min at 4°C, and the pelleted nuclei were washed with lysis buffer without Nonidet P-40. The nuclear pellet was resuspended in an equal volume of nuclear extraction buffer (20 mM Tris-HCl (pH 8.0), 420 mM NaCl, 1.5 mM MgC1 2 , 0.2 mM EDTA, and 25% glycerol), and NaCl was added to obtain a final concentration of 400 mM. After incubation at 4°C for 10 min, the nuclei were centrifuged at 25,000 ϫ g for 5 min. The supernatant fraction was used as the nuclear extract.
Electrophoretic Mobility Shift Assay (EMSA)-Complementary oligonucleotides for the NRE and its serial mutants (see Fig. 5A for sequences) were synthesized and annealed by heating to 70°C for 5 min and then slowly cooled to room temperature. Probes were prepared by end labeling oligonucleotides with [␥-32 P]ATP and T4 polynucleotide kinase. The binding reaction was achieved by preincubating nuclear extracts with 1 g of poly(dI-dC)/poly(dI-dC) (Amersham Biosciences) in buffer containing 20 mmol/liter HEPES (pH 7.9), 70 mmol/liter NaCl, 5 mmol/liter MgCl 2 , 0.05% Nonidet P-40, 10% glycerol, 0.5 mmol/liter dithiothreitol, and 5 mol/liter p-amidinophenylmethylsulfonyl fluoride at room temperature for 20 min as described previously (51). Three nanograms of end-labeled probes were added to the reaction mixture containing the nuclear extract and incubated for 15 min at room temperature. For competition experiments, excess unlabeled DNA was incubated with the reaction mixture for 15 min before the addition of the radiolabeled probe. In supershift assays, 5 g of anti-FLAG monoclonal antibody M2 (Sigma) were incubated with the reaction mixture for 15 min before the addition of the radiolabeled probe. The samples were subjected to 4% PAGE in 0.5% buffer (45 mM Tris borate and 1 mM EDTA) at 15 V/cm for 3 h at room temperature. The gel was dried, and autoradiography was performed at Ϫ70°C.
Chromatin Immunoprecipitation (ChIP)-In vivo binding of LRF to the NRE of the COMP promoter was investigated using the ChIP assay kit (Upstate Biotechnology, Inc., Lake Placid, NY). Confluent human embryonic kidney (HEK) 293 cells were transfected with the COMPspecific reporter construct Ϫ1925COMPluc and expression plasmid pFLAG-LRF. These transformed HEK cells and murine C3H10T1/2 progenitor cells were cultured on 10-mm dishes and then treated with formaldehyde (1% final concentration) to cross-link LRF to the DNA. Cells were washed with cold phosphate-buffered saline and lysed in SDS lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris-HCl (pH 8.1)). The lysate was sonicated to shear DNA to a length between 200 and 1000 bp. The sonicated supernatant was diluted 10-fold with ChIP dilution buffer (0.01% SDS, 1% Triton X-100, 2 mM Tris-HCl (pH 8.1), and 150 mM NaCl) and incubated with anti-FLAG antibody (Stratagene, La Jolla, CA), anti-LRF or anti-HDAC1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), or preimmune serum overnight at 4°C with rotation. To collect DNA/proteins complexes, salmon sperm DNA/ protein A-agarose slurry was added to the mixture and incubated for 1 h at 4°C with rotation, and the DNA/protein A-agarose was pelleted by centrifugation. After extensive washing of the pellet with a series of wash buffers, the pellet was dissolved with 250 l of elution buffer and centrifuged to remove the agarose. The supernatant was treated with 20 l of 5 M NaCl and heated to 65°C for 4 h to reverse the LRF-DNA cross-link. After treatment with EDTA and proteinase K, the supernatant was extracted with phenol/chloroform and precipitated with ethanol to recover the DNA. For PCR of the COMP promoter region using the chromatin-immunoprecipitated DNA, one-tenth of the DNA was PCR-amplified using forward primer 5Ј-GTAGTCAATAGGCCTGG-GAAGA and reverse primer 5Ј-GGAAAGGGAGTCAGAACTGAAG. Thirty-five cycles of PCR at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s were performed. PCR products were analyzed on 1% agarose gel.
Reporter Gene Assay-RCS cells grown to ϳ50% confluence in 35-mm culture dishes were transfected with 1 g of reporter construct (Ϫ1925COMPluc, Ϫ1775COMPluc, Ϫ⌬1925COMPluc, Ϫ592COMPluc, or 4xNREpGL2-Promoter) along with 1 g of pSVGal plasmid (internal control) and various amounts of pFLAG-LRF expression plasmid. To examine whether HDACs are involved in the LRF-mediated repression, various amounts of trichostatin A (TSA) (see Fig. 10C) were added to the medium. At 48 h after transfection, the cultures were harvested and lysed. Luciferase assays were performed using 20 l of cell extract and 100 l of luciferin substrate (Promega). ␤-Galactosidase assays were performed using a ␤-galactosidase assay kit (Tropix Inc., Foster City, CA) following the manufacturer's protocol. ␤-Galactosidase and luciferase activities were measured using a Mini-Lum luminometer (Bioscan, Inc., Washington, D. C.).
Magnetic Bead Assay-The magnetic bead assay was performed as described previously (50). Briefly, a 592-nucleotide proximal region (Ϫ592 to Ϫ1) and a 400-nucleotide distal region (Ϫ1925 to Ϫ1526, containing the NRE and used as a positive control) of the COMP promoter were labeled with biotin using the Promega random-primed DNA labeling kit (DNA probes). Five-picomole aliquots of the DNA probe were conjugated to 20 l of streptavidin-coupled magnetic beads (Promega) in 20 mM HEPES (pH 7.9), 100 mM NaCl, 0.5 mM EDTA, 10% glycerol, and 0.01% Nonidet P-40 and incubated with 500 g of nuclear extracts prepared from RCS cell lines transfected with expression plasmid pFLAG-LRF in 300 l of the same buffer at room temperature for 20 min. Bound FLAG-UBF1 was detected by SDS-PAGE and immunoblotting with anti-FLAG antiserum (Santa Cruz Biotechnology).
Generation of LRF Stable Lines-Murine C3H10T1/2 progenitor cells were cultured in tissue culture dishes in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum, 0.2 mM L-glutamine, and antibiotics. C3H10T1/2 cells were plated 1 day before transfection at a density of 1.5 ϫ 10 5 cells/30-mm dish. Transfection was carried out using Superfect transfection reagent (QIAGEN Inc., Valencia, CA) following the manufacturer's instructions. The LRF expression plasmid pFLAG-LRF or the empty pFLAG-F4 vector (51) was added together with the selective plasmid pMiniHgh (provided by Dr. X. Y. Fu), which conferred hygromycin-resistance upon the cells, at a ratio of 10:1. Two days after transfection, cells were split into 100-mm dishes at a density of 10 5 cells/dish in 10 ml of Dulbecco's modified Eagle's medium containing hygromycin at 200 g/ml. After 14 days in selective medium (medium changed every 3 days), cloning cylinders dipped in sterile vacuum grease were placed around individual colonies, and 0.3 ml of 0.1% trypsin and EDTA were added to dissociate the colonies from the plates. Cells were plated in 12-and 6-well plates and 35-and 100-mm dishes for expansion in Dulbecco's modified Eagle's medium containing hygromycin at 200 g/ml. The FLAG-tagged LRF levels were examined by immunoblotting using anti-FLAG antibody.
To examine COMP protein expression in the BMP-2-induced chondrogenic differentiation of C3H10T1/2 cells, 25 g of cell lysates prepared from micromass cultures of cloned LRF stable lines, control lines, and parental C3H10T1/2 cells maintained in Ham's F-12 medium in the presence of 100 ng/ml recombinant BMP-2 for 7 days were used to perform an immunoblot assay with rabbit polyclonal antiserum to COMP (1:500 dilution), followed by horseradish peroxidase-conjugated anti-rabbit IgG (1:1000 dilution).
Alcian Blue Staining-To assess the extent of chondrogenesis, micromass cultures of cloned LRF stable lines, control lines, and parental C3H10T1/2 cells maintained in Ham's F-12 medium containing 10% fetal calf serum in the presence of 100 ng/ml recombinant BMP-2 for 14 days were fixed in Kahle's fixative, washed with water, and stained overnight with 1% Alcian blue 8GX (Sigma) (56).
Co-immunoprecipitation-Approximately 500 g of cell lysates prepared from LRF stable lines were incubated with anti-LRF (25 g/ml) or anti-HDAC1 (20 g/ml) antibody or with control rabbit IgG (25 g/ml) for 1 h, followed by incubation overnight with 30 l of protein A-agarose (Invitrogen) at 4°C. After washing five times with immunoprecipitation buffer, bound proteins were released by boiling in 20 l of 2ϫ SDS loading buffer for 3 min (51). Released proteins were examined by Western blotting with anti-HDAC1 antibody, and the signal was detected using the ECL chemiluminescence system (Amersham Biosciences).

RESULTS
The Yeast One-hybrid System Identifies the Transcriptional Repressor LRF as an NRE-binding Protein-To identify the repressors that bind to the NRE located at Ϫ1775 to Ϫ1746 in the mouse COMP gene promoter, yeast one-hybrid screening was carried out according to the Matchmaker one-hybrid protocol. For this purpose, four tandem repeats of the NRE sequence (AGCCTGGGAGAGGGTCCCTGCCCTATGGAA) from the COMP gene promoter were ligated into the yeast integration and reporter vectors pHISi and pLacZi to generate pHISi-4NRE and pLacZi-4NRE, respectively. Each pHISi-4NRE and pLacZi-4NRE reporter construct was linearized and integrated sequentially into the genome of the competent yeast strain YM4271. The resulting yeast cells with integrated pHISi-4NRE and pLacZi-4NRE were used for one-hybrid screening with a mouse embryonic cDNA/VP16 activation domain fusion library. Screening of 2.5 ϫ 10 6 clones from a mouse embryonic cDNA library generated seven positive clones. Positive colonies were selected on synthetic complete medium lacking histidine and tryptophan and supplemented with 45 mM 3-amino-1,2,4triazole and transformed into E. coli DH5␣. Each cDNA insert was sequenced and analyzed by a BLAST search of the Gen-Bank TM /EBI Data Bank. Two of the positive clones were identical and contained a 1793-bp insert. The nucleotide sequences of these clones contain an open reading frame of 1698 bp, with the first ATG surrounded by an appropriate Kozak consensus sequence (57). The predicted open reading frame encodes a protein of 566 amino acid residues that is identical to mouse LRF (GenBank TM /EBI accession number AF086830) (44). LRF is the mouse counterpart of human FBI-1 (47) and the rat OCZF protein (48), with identical matches in three functionally important domains (alignment not shown): the N-terminal POZ domain (aa 1-120), four Cys 2 -His 2 zinc fingers (aa 379 -478), and a nuclear localization sequence (aa 483-496, CXX-VXXRXXRKXXX) (44,47,48,58). LRF is a highly conserved transcription factor; mouse LRF and rat OCZF are 90 and 96% identical at the nucleotide and amino acid levels, respectively, whereas mouse LRF and human FBI-1 sequences are 89% identical at the amino acid level (data not shown).
The yeast one-hybrid assay was repeated to verify the binding of LRF to the NRE. The plasmid encoding p53 (control) and LRF fused to the VP16 activation domain were retransformed into the yeast cells with integrated pHISi-4NRE, pHISi-53BE (p53-binding elements), pLacZi-4NRE, and pLacZi-53BE. The resulting HIS ϩ clones were subjected to ␤-galactosidase and growth phenotype assays (Fig. 1). The results show that p53 bound to the p53-binding elements, but not to the NRE, and that LRF bound to the NRE, but not to the p53-binding elements based on the activation of the lacZ reporter gene (Fig. 1A) and growth on the culture plates lacking histidine and tryptophan but containing 45 M 3-amino-1,2,4-triazole (Fig. 1B).
Preparation and Characterization of Anti-LRF Antibodies-The GST-LRF-(207-277) fusion protein, encoding a segment (aa 207-277) between the POZ domain and zinc finger motifs of LRF, was expressed in bacteria, purified on a glutathione-Sepharose column; and subjected to preparative scale SDS-PAGE. The major band was excised and used to immunize rabbits (Zymed Custom Antibody). To generate affinity-purified anti-LRF antibodies, the GST-absorbed anti-LRF serum was incubated with Affi-Gel-10 beads to which purified GST-LRF-(207-277) was covalently linked. The bound antibodies were eluted at low pH and neutralized.
Western blotting was performed using in vitro translated LRF and nuclear extracts prepared from C3H10T1/2 progenitor cells to characterize the anti-LRF antiserum (Fig. 2B). The results show that affinity-purified anti-LRF antibodies specifically recognized endogenous LRF in C3H10T1/2 cells, with a similar size as in vitro translated LRF (compare lanes 2 and 3); no bands were resolved in the control reaction using empty vector as template (lane 1).
LRF Binds to the NRE in the COMP Promoter in Vitro-Interactions identified in yeast must be verified using independent procedures because 1) the yeast one-hybrid system may produce false interactions, and 2) true interactions in yeast in which post-translational modifications are absent might be lost in the mammalian cells if post-translational modifications are involved in the associations. We first confirmed the binding of LRF to the NRE in EMSA (Fig. 3). Incubation of the 32 P-labeled NRE probe with the nuclear extracts prepared from RCS cells transfected with the mammalian expression plasmid pFLAG-LRF resulted in a specific LRF Binds to the NRE in the COMP Gene Promoter in Living Cells-To determine whether LRF also binds to the NRE in vivo, we performed ChIP assays, which are important for defining interactions of factors with specific DNA elements in living cells. ChIP was first carried out in HEK293 cells transfected with the COMP-specific reporter construct Ϫ1925COM-Pluc and plasmid pFLAG-LRF. After cross-linking with formaldehyde, cell lysates were immunoprecipitated with control IgG (negative control) or with anti-FLAG or anti-LRF antibody, and the DNA purified from this coprecipitation was analyzed by PCR with PCR primers spanning the NRE in the COMP promoter. As shown in Fig. 4A, we observed a clear PCR product using DNA isolated from immunoprecipitated complexes with both anti-FLAG (lane 2) and anti-LRF (lane 3) antibodies, but not with control IgG (lane 1), indicating that FLAG-tagged LRF binds to the NRE of the COMP promoter in transfected living cells.
ChIP also provides information on native chromatin structure and the endogenous factors bound to regulatory elements in different functional states (37). ChIP assays were performed in C3H10T1/2 progenitor cells, which express LRF (Fig. 2). Endogenous protein⅐DNA complexes were then covalently cross-linked with formaldehyde, and after cell lysis, the chromatin was sheared by sonication. Endogenous LRF⅐DNA complexes were immunoprecipitated with either preimmune serum (negative control) (Fig. 4B, lane 2) or anti-LRF antibody (lane 3), and the DNA purified from these coprecipitations and input DNA (positive control) (lane 1) were analyzed by PCR. We observed amplified products of the COMP promoter from input DNA and DNA isolated from anti-LRF antibody-precipitated (but not from preimmune serum-precipitated) complexes. These results demonstrate that endogenous LRF binds to the native NRE in the COMP gene.
Identification of the LRF-binding Element (LBE) in the NRE-Once the interaction between LRF and the NRE became clear, we sought to identify the LBE in the 30-bp NRE by performing EMSA again using wild-type and serial mutant NRE probes. All probes were incubated with nuclear extracts prepared from pFLAG-LRF-transfected RCS cells; the binding of LRF to various probes is summarized in Fig. 5A and shown in Fig. 5B. Mutants 1 and 5, in which the first six nucleotides (AGCTGG) and the last six nucleotides (ATGGAA) were mutated, respectively, bound to LRF as strongly as the wild-type NRE probe (Fig. 5B, compare lanes 1, 2, and 6); however, mutants 2 and 3, in which the second six nucleotides (GGA-GAG) and the third six nucleotides (GGTCCC) were altered, respectively, totally abolished the binding of LRF (lanes 3 and 4), clearly demonstrating that these 12 nucleotides (GGAGAG GGTCCC) are essential for binding to LRF. Interestingly, mutant 4, in which the fourth six nucleotides (TGCCCT) were replaced, still bound to LRF, but the binding intensity was weaker (lane 5). Twelve nucleotides (GGAGAGGGTCCC) identified above in the NRE contain a typical consensus binding site (G(A/G)GGG(T/C)(C/T)(T/C)(C/T)) (59) of FBI-1, a human counterpart of mouse LRF. To confirm whether the flanking sequences of the consensus binding site are involved in the binding of LRF to the NRE, an additional three mutants of NRE (lanes 7-9) were constructed and tested. Mutant 6, in which three 5Ј-nucleotides (GGA) flanking the "consensus sequence" were mutated, bound to the LRF as well as the wild-type probe, indicating that these three nucleotides are not required for LRF-NRE association. Mutants 7 and 8, in which the first three nucleotides (TGC) and the last three nucleotides (CCT) of six 3Ј-nucleotides flanking the consensus sequences were changed, respectively, still bound to LRF with reduced affinity, as did mutant 4, in which these six nucleotides were replaced. These data indicate that the identified core DNA-binding site for LRF is in accordance with the published consensus sequence for FBI-1. It is also possible that the six 3Ј-nucleotides of this consensus site might enhance the association of LRF with the consensus sequences.
LRF Inhibits COMP-specific Reporter Construct Activities and Endogenous COMP Gene Expression-To determine whether LRF represses transcription of the COMP promoter using reporter gene assays, four COMP-specific reporter gene plasmids (Ϫ1925COMPluc, Ϫ1775COMPluc, Ϫ⌬1925COM-Pluc, and Ϫ592COMPluc) were generated in which segments with or without an NRE from the 5Ј-flanking region of the COMP gene were linked to the upstream end of a region encoding luciferase in the pGL2-Basic vector, and one NRE-specific reporter construct (4xNREluc) was generated in which a tandem repeat of four NREs was inserted upstream of the SV40 promoter in the pGL2-Promoter vector (Fig. 6A). We transfected RCS cells with these reporter constructs together with the mammalian expression plasmid pFLAG-LRF. As shown in Fig. 6B, LRF produced Ͼ90% inhibition of the reporter constructs with an NRE (Ϫ1925COMPluc and Ϫ1775COMPluc), and this repression was dose-dependent (Fig. 6D); as expected, LRF also significantly repressed the activity of 4xNREpGL2-Promoter (which contains four tandem repeats of NRE) (Fig.  6C), indicating that the NRE is sufficient for LRF binding in the transfected cells. Given that LRF also repressed the activity of two other reporter constructs without an NRE (Ϫ⌬1925COMPluc and Ϫ592COMPluc), but to a lesser degree (ϳ40% inhibition) (Fig. 6B), we next examined whether LRF also binds to the proximal region of the COMP promoter using a magnetic bead assay. Both a 592-bp proximal region of the COMP promoter (Ϫ592 to Ϫ1) and a 400-bp distal region of the COMP promoter (Ϫ1925 to Ϫ1526) were labeled with biotin. Streptavidin-coupled magnetic beads conjugated to these biotin-labeled probes were incubated with nuclear extracts expressing pFLAG-LRF. Bound FLAG-LRF was detected by immunoblotting with anti-FLAG antibody. As shown in Fig. 6E, the NRE-containing distal region bound to LRF, but the proximal region did not, indicating that LRF cannot directly bind to the proximal region of the COMP promoter.
To confirm whether exogenous expression of LRF represses expression of the endogenous COMP gene, we transfected RCS cells, which express COMP, with the pFLAG-LRF expression plasmid and performed RT-PCR to determine the COMP mRNA levels. As shown in Fig. 6F, COMP gene expression was inhibited in the pFLAG-LRF-transfected RCS cells, in contrast with untransfected cells and cells transfected with the control pCMV vector.
Generation of an LRF Stable Line in Which LRF Is Constitutively Overexpressed-To determine whether stable overexpressed LRF affects the expression of endogenous COMP, we generated LRF stable lines in murine C3H10T1/2 progenitor cells, which undergo chondrogenic differentiation in the presence of inducers for chondrocyte differentiation. C3H10T1/2 cells were transfected with either an empty vector (for generating a control line) or the expression plasmid pFLAG-LRF together with the selective marker pMiniHgh, which confers hygromycin resistance upon transfected cells; the resultant clones were selected and amplified, and the level of ectopically expressed FLAG-LRF was examined by Western blotting with the anti-FLAG probe. As shown in Fig. 7, FLAG-LRF was clearly expressed in three selected LRF stable clones, i.e. clones 1, 9, and 11 (lanes 4 -6, respectively), and absent in both parental C3H10T1/2 cells (lane 1) and the control stable lines pc1 and pc2 (lanes 2 and 3, respectively).
LRF Overexpression Inhibits Endogenous COMP Gene Expression-To determine the effects of stable overexpressed LRF on the expression of COMP, we first performed RT-PCR using micromass cultures of C3H10T1/2, control, and LRF stable lines exposed to recombinant BMP-2 for 1 week (Fig. 8A). COMP mRNA was expressed in the BMP-2-treated micromass cultures of parental C3H10T1/2 cells and the control stable lines pc1 and pc2 (lanes 1-3, respectively), whereas BMP-2induced COMP expression in the course of chondrogenic differentiation was totally inhibited in LRF stable lines (lanes 4 -6). We next performed Western blotting with anti-COMP antibodies using the same micromass cultures (Fig. 8B). In agreement with PCR data, the COMP protein was expressed in the control stable lines (lanes 1-3), but not in the LRF stable lines (lanes 4 -6). Taken together, these results demonstrate that overexpression of LRF represses endogenous BMP-2-induced COMP expression.
LRF Overexpression Inhibits Chondrocyte Differentiation in C3H10T1/2 Micromass Cultures-In addition to COMP, several other extracellular matrix molecules of cartilage, such as aggrecan and collagen types II, IX, X, and XI, are hallmarks for monitoring the process of chondrogenic differentiation (60,61). Type X collagen is indicative of hypertrophic chondrocytes (62,63). We next performed RT-PCRs using the mRNAs isolated from micromass cultures of C3H10T1/2, control, and LRF stable lines exposed to recombinant BMP-2 for 1 week with the specific primers for collagen types II, IX, X, and XI; aggrecan; and GAPDH (used as an internal control). As shown in Fig. 9A, aggrecan and collagen types II, IX, X, and XI were expressed in the BMP-2-treated cultures (lanes 1-3), but expression of collagen type II, IX, and X mRNAs during BMP-2-induced chondrogenic differentiation was completely repressed in the presence of ectopically expressed FLAG-LRF (lanes 4 -6). Although aggrecan and type XI collagen mRNAs were still positive in the LRF stable lines (lanes 4 -6), these levels were lower compared with the controls (lanes 1-3).
We also performed Alcian blue staining to confirm the re-pression of BMP-2-induced chondrogenesis by LRF overexpression using the micromass cultures of C3H10T1/2, control, and LRF stable lines exposed to recombinant BMP-2 for 2 weeks (Fig. 9B). Positive staining in the control lines (upper) was FIG. 6. Inhibition by LRF of COMPspecific reporter constructs and endogenous COMP gene expression. A, schematic structures of four COMP-specific reporter constructs and one NREspecific reporter construct. The indicated segments from the 5Ј-flanking region of the COMP gene were linked to a DNA segment encoding luciferase (Luc; open  boxes). Numbers indicate distances in nucleotides from the first nucleotide of intron 1. The NRE is represented by ovals. B, LRF inhibits COMP-specific reporter construct activities. The indicated reporter gene was transfected into RCS cells with 4 g of pFLAG-LRF expression plasmid as well as a pSVGal internal control plasmid. At 48 h after transfection, the cultures were lysed, and the ␤-galactosidase and luciferase activities were determined. The activity of Ϫ1925COMPluc was arbitrarily set to 1 in the absence of LRF. rel., relative. C, LRF represses NRE-containing pGL2-Promoter activity. 4xNREpGL2-Promoter (containing four tandem repeats of NRE) was transfected into RCS cells with or without 4 g of pFLAG-LRF expression plasmid as well as a pSVGal internal control plasmid. D, LRF-mediated repression is dose-dependent. The reporter gene Ϫ1925COMPluc was transfected into RCS cells together with the indicated amount of pFLAG-LRF expression plasmid as well as a pSVGal internal control plasmid. E, LRF does not bind to the proximal region of the COMP promoter (magnetic bead assay). Five picomoles of the biotin-tagged proximal region (Ϫ592 to Ϫ1) and distal region (Ϫ1925 to Ϫ1526) of the COMP promoter were conjugated to streptavidin-coupled magnetic beads and incubated with 500 g of nuclear extracts expressing FLAG-LRF. The FLAG-LRF bound to the beads was subjected to SDS-PAGE and immunoblotting with anti-FLAG antiserum. The FLAG-LRF band is indicated by the arrow. F, exogenous expression of LRF reduces COMP gene expression in RCS cells (RT-PCR). RNAs isolated from untransfected RCS cells or from RCS cells transfected with 30 g of pFLAG-LRF expression plasmid or control pCMV vector were reverse-transcribed, and the mRNA expression of COMP and GAPDH (serving as an internal control) was examined by RT-PCR. strongly inhibited in the LRF stable lines (lower). The results from these assays show that stable overexpression of LRF inhibited BMP-2-induced chondrogenesis in C3H10T1/2 micromass cultures.
LRF Associates with HDAC1, and Its Repression Requires Deacetylase Activity-LRF contains a POZ domain, which is a conserved protein-protein interaction motif found in many transcription factors, and the most striking common property of the POZ domain-containing transcription factors is their ability to repress transcription via interaction with other key regulatory proteins such as HDACs (64,65). A co-immunoprecipitation assay was first performed to determine whether LRF binds to HDAC1 in vivo (Fig. 10A) The interaction between LRF and HDAC1 was confirmed using an in vitro protein-protein interaction assay (an in vitro GST pull-down assay) (Fig. 10B). Briefly, affinity-purified GST as well as N-terminally (aa 51-482) and C-terminally (aa 1-432) truncated HDAC1 fused to GST that was immobilized on glutathione-Sepharose beads were incubated with in vitro translated LRF and, after washing, resolved by Western blotting. Similar to the GST control (lane 2), C-terminally truncated GST-HDAC1-(1-432) did not pull down the LRF protein (lane 4), whereas N-terminally truncated GST-HDAC1-(51-482) efficiently pulled down LRF (lane 3), indicating that the C-terminal (but not N-terminal) 50 amino acids of HDAC1 contain LRF-binding domains. To narrow down the LRF-binding sequences in the C terminus of HDAC1, two additional mutants were generated, and the same in vitro binding assay was performed (Fig. 10C). Removal of 15 amino acids from the C terminus (compare GST-HDAC1-(51-467) with GST-HDAC1-(51-482)) did not affect interaction, nor did further removal of the other 15 amino acids from the C terminus (GST-HDAC1-(51-452)). Given that mutant HDAC1-(1-432) failed to bind LRF (Fig. 10B, lane 4), it appears that 20 amino acids (aa 433-452) of HDAC1 are essential for interacting with LRF.
To determine whether the LRF⅐HDAC1 protein complex is detectable in the NRE of the COMP promoter, ChIP assays were performed in C3H10T1/2 progenitor cells, which express both LRF and HDAC1. Endogenous HDAC1⅐LRF⅐DNA complexes were immunoprecipitated with control IgG (negative control) (Fig. 10D, lane 3), anti-LRF antibody (positive control) (lane 2), or anti-HDAC1 antibody (lane 4), and the DNA purified from these coprecipitations and input DNA (lane 1) were analyzed by PCR. We observed amplifications of COMP promoter DNA from input DNA and DNAs isolated from both FIG. 7. Expression of FLAG-LRF in LRF stable lines. C3H10T1/2 cells transfected with the expression plasmid pFLAG-LRF or empty vector were incubated with hygromycin. Individual colonies were isolated, and FLAG-LRF expression was analyzed by immunoblotting. Twenty micrograms of lysate prepared from parental C3H10T1/2 (10T1/2) cells, control lines (pc1 and pc2), and LRF stable lines (clones 1, 9, and 11) were subjected to SDS-PAGE and immunoblotting with anti-FLAG antibody. The band corresponding to FLAG-LRF is indicated by the arrow.

FIG. 8. Effects of LRF overexpression on BMP-2-induced COMP expression in C3H10T1/2 micromass cultures. A, RT-PCR.
Total RNA was prepared from micromass cultures of control lines (C3H10T1/2 (10T1/2), pc1, and pc2) or LRF stable lines (clones 1, 9, and 11) in the presence of 100 ng/ml recombinant BMP-2 for 7 days, and the mRNA expression of COMP and GAPDH (serving as an internal control) was examined by RT-PCR. Similar results were obtained in repeated studies. B, Western blot assay. Twenty-five micrograms of cell lysates prepared from the same cultures as A were used to perform immunoblot assay with anti-COMP polyclonal antiserum, followed by horseradish peroxidase-conjugated anti-rabbit IgG. The COMP band is indicated by the arrow.

FIG. 9. Effects of LRF overexpression on BMP-2-induced chondrogenesis in C3H10T1/2 micromass cultures.
A, total RNA was prepared from micromass cultures of control lines (C3H10T1/2 (10T1/ 2), pc1, and pc2) or LRF stable lines (clones 1, 9, and 11) in the presence of 100 ng/ml recombinant BMP-2 for 7 days. The mRNA expression of collagen types II, IX, X, and XI; aggrecan; and GAPDH (serving as an internal control) was examined by RT-PCR. Similar results were obtained in repeated studies. B, shown are the results from whole mount Alcian blue histochemistry. Staining was performed on the micromass cultures of control lines (C3H10T1/2, pc1, and pc2) or LRF stable lines (clones 1, 9, and 11) treated with 100 ng/ml recombinant BMP-2 for 14 days.
anti-LRF and anti-HDAC1 antibody-precipitated (but not control IgG-precipitated) complexes. These results demonstrate that the LRF⅐HDAC1 complex is detectable in the NRE of the COMP gene promoter.
To examine whether HDACs are required for the LRF-mediated repression of the COMP reporter construct, RCS cells transfected with the COMP-specific reporter construct Ϫ1925COMPluc and the mammalian expression plasmid pFLAG-LRF were cultured in the presence of different concentrations of TSA for 48 h, and luciferase and ␤-galactosidase activities were measured. As shown in Fig. 10E, LRF inhibition of the COMP promoter was attenuated when TSA was added to the medium, demonstrating that HDACs are involved in the LRF-mediated repression of COMP promoter activity.
We next tested whether COMP gene expression in C3H10T1/2 cells can be stimulated by TSA. As revealed by RT-PCR (Fig. 10F), COMP mRNA was detectable in the TSAtreated (but not untreated) C3H10T1/2 cells, indicating that TSA induces COMP gene expression and that this induction is dose-dependent. DISCUSSION Our previous studies identified an NRE located between Ϫ1775 and Ϫ1746 of the 5Ј-flanking region of the murine COMP gene (35,39). In this study, a yeast one-hybrid screen was used to identify transcription factors that bind to this NRE. We present comprehensive evidence that LRF, which contains a POZ domain, a conserved 120-amino acid motif involved in transcriptional repression and dimerization, and four Krü ppel-like zinc finger DNA-binding motifs (44), binds directly to the NRE in the promoter of the COMP gene. We also present evidence that repression of the COMP promoter by LRF blocks both endogenous COMP gene expression as well as chondrogenesis of C3H10T1/2 micromass cultures. The molecular mechanism underlying LRF-mediated gene transcription appears to involve the association of LRF with HDAC1.
Although a 9-bp guanidine-rich LBE (GAGGGTCCC) in the 30-bp NRE (identified via an EMSA with a series of mutant NRE probes) of the mouse COMP promoter is essential for binding to LRF, the following six nucleotides (GGTCCC) also appear to be involved in the association between LRF and the . DNA recovered from the immunoprecipitation was amplified by PCR. Input DNA (lane 1) was used as a positive control. E, repression by LRF is sensitive to the HDAC inhibitor TSA. RCS cells were transfected with 2 g of Ϫ1925COMPluc with or without 4 g of pFLAG-LRF expression plasmid as well as a pSVGal internal control plasmid in the presence or absence of the indicated concentrations of TSA. Luciferase values were corrected for transfection efficiency with ␤-galactosidase values. The change in repression (n-fold) was calculated relative to that in cells transfected with Ϫ1925COMPluc alone. F, TSA treatment stimulates COMP gene expression in C3H10T1/2 cells. RNAs isolated from C3H10T1/2 cells treated with various amounts of TSA, as indicated, were reverse-transcribed, and the mRNA expression of COMP and GAPDH was examined by RT-PCR. LBE because its mutation reduces binding intensity (Fig. 5). These results suggest that these six nucleotide (GGTCCC) may act as the surrendering matrix sequence of the core-binding sequence and facilitate the binding of LRF/FBI-1 to DNA. Sequence analysis revealed that the human COMP gene promoter also contains a typical consensus site (GGGGGCCTC, Ϫ1294 to Ϫ1302) (59) for binding LRF/FBI-1, indicating that FBI-1 is very likely to repress COMP gene expression in human as LRF does in mouse.
It has also been reported that FBI-1 binds to DNA in a flexible manner (58,59) and that OCZF, the rat homolog of LRF, binds to both the cKrox tandem site and the Egr-1 single site, albeit with weaker affinity than their cognate proteins (48). The cKrox gene family, including human cKrox-␣, cKrox-␤ (another name for FBI-1), and cKrox-␥ as well as the murine cKrox-␤ homolog, LRF, was found to repress transcription driven by promoters for the type I and II collagen, fibronectin, and elastin genes (66,67). LRF also inhibits the BMP-2-induced endogenous expression of COMP; collagen types II, IX, X, and XI; and aggrecan ( Figs. 8 and 9), suggesting that LRF acts as a general transcriptional repressor that regulates multiple extracellular matrix genes, probably via binding to the regulatory regions of those genes. It remains to determine whether each of these genes has LBE or LBE-like sequences in the regulatory regions and, if not, to identify the alternative LBEs in the promoters of these genes.
In addition to producing dose-dependent inhibition of the NRE-containing COMP-specific reporter genes, LRF also repressed the activities of two other COMP-specific reporter constructs without an NRE (Ϫ⌬1925COMPluc and Ϫ592COM-Pluc) to a lesser degree (ϳ40% inhibition) (Fig. 6B), which raises the possibility that LRF may also bind to DNA elements other than the NRE in the COMP promoter. A magnetic bead assay (50) was performed to address this possibility. In this assay, the NRE-containing distal region retained LRF, but the proximal region did not, demonstrating that the LRF does not directly bind to the proximal region of the COMP promoter (Fig. 6E). Thus, inhibition of NRE-lacking COMP-specific reporter constructs by LRF might be due to the indirect regulation by LRF of the COMP promoter; for example, LRF may exert its effects via cooperating with some as yet unidentified transcription factors that directly associate with the proximal region of the COMP promoter.
LRF contains a POZ domain, which is a well defined, conserved protein-protein interaction motif found in many transcription factors, oncogenic proteins, and ion channel proteins and in some actin-associated proteins (68,69). The most striking and common property of the POZ domain-containing transcription factors is their ability to repress transcription (64,65,68,70,71). The ability of the domain to interact with other key regulatory proteins such as corepressor proteins and other transcription factors appears to be important for repression (64,65). In particular, the POZ domains of human PLZF (promyelocytic leukemia zinc finger transcription factor) and Bcl-6 (B-cell lymphoma transcription factor-6) have been shown to interact with other factors as well as with HDAC (65,71). It has been suggested that "chromatin remodeling" by the HDAC complex recruited by the POZ domain represses transcription (65,72,73). LRF binds to HDAC1 through an Asp/Glu-rich 20-amino acid unique segment (aa 433-452) in the C terminus of HDAC1 (Fig. 10, B and C) other than the LXCXE-like motif (IACEE, aa 415-419) that is required for interactions between HDAC1 and the pocket protein family, including Rb, p107, and p130 (74 -76). LRF inhibition of the COMP promoter was attenuated when TSA was added to the medium, demonstrating the potential involvement of HDACs in the LRF-mediated re-pression of COMP promoter activity. Given that TSA could not completely overcome LRF inhibition, the independent pathways of HDAC might be also involved in LRF-mediated gene repression.
It appears that LRF functions as a universal transcriptional repressor and regulates multiple biological processes. FBI-1, the human homolog of LRF (first purified as a cellular factor that binds to the human immunodeficiency virus inducer), has been shown to modulate human immunodeficiency virus type 1 Tat transactivation and repression of some cellular genes (45,47,58). OCZF, the rat homolog of LRF, is highly expressed in osteoclasts, and antisense OCZF cDNA suppresses the formation of osteoclast-like multinucleated cells in bone marrow culture (48). These results suggest that OCZF plays an important role in the late stage of osteoclastogenesis. LRF was reported to repress the promoter activity of several extracellular genes (66). Our unbiased genetic screen and EMSA as well as ChIP showed that LRF directly binds to the NRE in the promoter region of the COMP gene and associates with HDAC1, and our functional assays demonstrated that LRF regulates COMP gene expression and chondrogenesis of high density micromass cultures of C3H10T1/2 cells.