Stage-specific Expression of Myelin Basic Protein in Oligodendrocytes Involves Nkx2.2-mediated Repression That Is Relieved by the Sp1 Transcription Factor*

The homeodomain-containing protein Nkx2.2 is critical for the development of oligodendrocyte lineage cells, but the target genes of Nkx2.2 regulation have not been identified. In the present study, we found that the myelin basic protein gene is one of the genes that is regulated by Nkx2.2. Expression of Nkx2.2 represses the expression of myelin basic protein in oligodendrocyte progenitors. Two regulatory elements in the myelin basic protein promoter were identified and found to interact with Nkx2.2 in vitro. Despite their sequence divergence, both sites were involved in the Nkx2.2-mediated repression of the myelin basic protein promoter. Binding of Nkx2.2 also blocked and disrupted the binding of the transcriptional activator Purα to the myelin basic protein promoter. Additionally Nkx2.2 recruited a histone deacetylase 1-mSin3A complex to the myelin basic protein promoter. We also found that the transcription factor Sp1 was able to compete off the binding of Nkx2.2 to its consensus binding site in vitro and reversed the repressive effect of Nkx2.2 in vivo. Our data revealed a novel role for Nkx2.2 in preventing the precocious expression of myelin basic protein in immature oligodendrocytes. Based on this study and our previous reports, a model for myelin basic protein gene control is proposed.

Oligodendrocytes are the myelin-producing cells in the central nervous system. Differentiation of the progenitor cells into mature myelinating cells involves the activation of a genetic program that leads to expression of a set of genes encoding proteins important for the elaboration of the myelin membrane. The expression of these genes is at least partially regulated by transcription factor activity. These regulators of transcription include activators, repressors, and their cofactors. Transcriptional regulation may also involve chromatin modifications.
One of the genes that is activated during differentiation of oligodendrocytes is the gene encoding myelin basic protein (MBP). 1 This gene is exclusively expressed in mature oligoden-drocytes, and its expression is mainly regulated at the transcriptional level. It provides an ideal model to investigate the mechanism of transcriptional regulation during cellular differentiation.
Myelination begins in the mouse at ϳ9 days after birth, peaks at about postnatal day 17, and declines to a steady state by 2 months of age. MBP is encoded by a single gene in the mouse and human. The sequence of the MBP promoter from mouse, rat, and human has been determined, and it is highly conserved, especially in the core promoter region (1). In mouse brain, MBP mRNA is detected in oligodendrocytes at the end of the first postnatal week, peaks at 18 days, and remains at low levels in the adult (2). Thus expression of MBP is temporally regulated and is tightly correlated with myelination in the developing central nervous system.
Previous work from our laboratory and the laboratories of others has shown that several different transcription factors are involved in activation of MBP expression in differentiating oligodendrocytes. Among these is the Sp1 family of transcription factors. Sp1 plays an essential role in the regulation of MBP expression through binding to the GC-rich region of the MBP promoter (3)(4)(5). Through interaction with Sox10, Sp1 activates and contributes to the tissue-specific expression of MBP in the central nervous system (6). Other transcription factors also found to play a role in the regulation of MBP expression include nuclear factor B (7), Pur␣ (4), myelinating glia-enriched DNA binding activity protein (8), NF1 (3,9,10), and the AP-1 family of transcription factors (11).
Nkx2.2, a member of the homeodomain transcription factor gene family, has been shown to be involved in oligodendrocyte specification (12,13) and differentiation (14). However, the target genes of Nkx2.2 are still unclear. In the present study, we found that expression of Nkx2.2 repressed the expression of MBP in CG4 cells. Repression of MBP expression by Nkx2.2 involved disruption of the binding of a transcriptional activator and recruitment of the histone deacetylase 1 (HDAC1)-mSin3A corepressor complex to the MBP promoter. We also showed that Sp1 was able to compete off the binding of Nkx2.2 to the MBP promoter in vitro and reverse the effect of Nkx2.2 on the MBP promoter in vivo. These results suggested that Nkx2.2 plays a role in the regulation of MBP expression and that coordination of various transcription factors including Nkx2.2 and Sp1 defines the temporal and specific expression of MBP in oligodendrocyte lineage cells.

MATERIALS AND METHODS
Cell Culture-CG4 cells were cultured and induced to differentiate as reported previously (5). Briefly cells were plated in dishes coated with poly-L-ornithine and cultured at 37°C in an atmosphere of 5% CO 2 and 100% humidity. Cells were maintained in growth medium consisting of Dulbecco's modified Eagle's medium containing 30 nM sodium selenite, 50 ng/ml insulin, 50 g/ml transferrin, 1 mg/ml bovine serum albumin, 5 ng/ml platelet-derived growth factor, 10 ng/ml basic fibroblast growth factor, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C. Fresh platelet-derived growth factor and basic fibroblast growth factor were added to the cultures every other day. To induce differentiation, cells were switched to differentiation medium (DM) consisting of Dulbecco's modified Eagle's medium containing 30 nM sodium selenite, 5 g/ml insulin, 50 g/ml transferrin, 10 nM biotin, 10 nM hydrocortisone, 30 nM 3,3Ј,5Ј-triiodo-L-thyronine, 20 nM progesterone, 2 mM glutamine, 1 mg/ml bovine serum albumin, 100 units/ml penicillin, and 100 g/ml streptomycin.
DNA Transfection and Luciferase Assays-CG4 cells were plated in 35-mm dishes and cultured as described above. Transfections were done in growth medium using GenePorter (Gene Therapy Systems, San Diego, CA) according to the manufacturer's protocol. A combination of 1.0 g of MBPLuc construct and 2.0 g of expression plasmid or control vector (unless otherwise specified) was transfected into each dish of cells. Four hours after transfection, the medium was changed to DM. For treatment with trichostatin A (TSA) (Calbiochem), the medium was changed to DM containing TSA or Me 2 SO. After 48 h, the cells were rinsed once with phosphate-buffered saline and lysed by the addition of reporter lysis buffer (Promega, Madison, WI). Luciferase activity was measured in a Packard TopCount luminometer. The samples were also assayed for protein content using Bio-Rad protein assay reagent and for luciferase activity using Steady-Glo luciferase assay reagent (Promega). Luciferase activity was normalized to the protein concentration.
Site-specific Mutagenesis-The eukaryotic expression plasmid pBAT-NkxNQ was made using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's protocol. pBAT-NkxNQ encodes a mutant form of Nkx2.2 containing a single amino acid mutation (Asn to Gln) in the homeodomain. The prokaryotic expression plasmid pQE30-NkxNQ was made using the same strategy. Reporter gene constructs containing a mutated 5Ј-flanking region of the MBP gene in which the Nkx2.2 binding sites were altered were also made using the QuikChange site-directed mutagenesis kit. All constructs were subjected to sequencing to confirm the mutations.
Protein Purification-Plasmids encoding His-Nkx2.2, His-NkxNQ, or His-Pur␣ were transformed into Escherichia coli strain BL21. For protein preparation, 500 ml of LB medium containing 50 g/ml ampicillin was inoculated with a 25-ml overnight culture and incubated with shaking at 37°C for 3-5 h until the A 595 reached 0.6. Recombinant protein expression was induced by addition of isopropyl 1-thio-␤-Dgalactopyranoside to a final concentration of 1 mM. The culture was further incubated at room temperature for 2 h. Bacteria were collected by centrifugation at 5000 ϫ g for 5 min at 4°C. The supernatant was removed, and the pellet was completely resuspended in 10 ml of IMAC-5 (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 10% glycerol, 5 mM imidazole) containing a protease inhibitor mixture (1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 1 g/ml aprotinin). Cells were lysed by sequential addition of lysozyme to a final concentration of 1 mg/ml and Triton X-100 to a final concentration of 0.1% (v/v) and then incubated on ice for 30 min with occasional stirring. Lysates were then subjected to pulse sonication on ice. The lysate was centrifuged at 10,000 ϫ g for 30 min at 4°C, and the supernatants were collected and filtered through a 0.45-m syringe filter. His-Bind resin (Novagen, Madison, WI) equilibrated with IMAC-5 was charged with 5 mg/ml NiCl 2 . The column was then equilibrated with 10 volumes of IMAC-5 following which the cell lysates were loaded. The His-tagged protein was eluted with a step gradient of 20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride containing 5, 20, 50, 100, and 200 mM imidazole. Fractions were collected and analyzed on a 10% SDS-polyacrylamide gel that was stained with Coomassie Blue R-250. Fractions containing the target protein were pooled and dialyzed against 25 mM HEPES, pH 8.0, 10 mM Tris-Cl, pH 8.0, 10% glycerol, 150 mM NaCl at 4°C overnight. Protein concentrations were determined using Bio-Rad protein assay reagent. Protein samples were stored at Ϫ70°C.
DNase I Footprinting-Ϫ1323 MBPLuc plasmid was digested with StuI and HindIII to generate the MBP promoter fragment containing sequences from Ϫ265 to ϩ30. The fragment was purified and labeled with [␣-32 P]dATP by end-filling using Klenow fragment. The probe (20,000 cpm) was incubated with ϳ0.5 g of purified Nkx2.2 in binding buffer (50 mM Tris-HCl, pH 8.0, 100 mM KCl, 12.5 mM MgCl 2 , 1 mM EDTA, 50 g/ml poly(dI-dC), 20% glycerol, 1 mM dithiothreitol) in a total volume of 25 l and incubated at 4°C for 30 min. MgCl 2 (10 mM) and CaCl 2 (5 mM) were added, and the reaction was incubated for 1 min at room temperature. DNase I was then added to the reaction at a final concentration of 0.625, 1.25, 2.5, or 5 g/ml. A control tube without DNase I was also treated similarly. After further incubation for 1 min at room temperature the reaction was stopped by adding 75 l of stop solution (20 mM EDTA, pH 8.0, 1% SDS, 0.2 M NaCl, 125 g/ml yeast tRNA) and extracted with an equal volume of phenol:chloroform. The aqueous phase was transferred to a fresh microcentrifuge tube, and the nucleic acids were precipitated with 2.5 volumes of chilled ethanol. The precipitated nucleic acids were dissolved in 10 l of formamide dye, mixed by vigorous shaking, and boiled for 5 min. A Maxam-Gilbert G reaction was also performed using the labeled MBP promoter fragment according to standard procedures (20). Samples were run on a denaturing 8% polyacrylamide sequencing gel and visualized by autoradiography.
Electrophoretic Mobility Shift Assays (EMSAs)-Synthetic oligonucleotides were annealed and end labeled with [␥-32 P]ATP using T4 polynucleotide kinase (MBI Fermentis, Hanover, MD). The Sp1 protein used in these experiments was purified from HeLa cells (Promega). Approximately 0.5-5 ng (10,000 -20,000 cpm) of labeled probe was incubated for 30 min on ice with ϳ0.5 g of recombinant protein (unless otherwise specified) in binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.5 mM EDTA, 1 mM MgCl 2 , 4% glycerol, 0.5 mM dithiothreitol) containing 2 g poly(dI-dC) in a total volume of 40 l. When included as competitors, unlabeled double-stranded oligonucleotides were added in excess to the reaction together with the labeled probe. Complexes were resolved on 3.5% nondenaturing polyacrylamide gels in 0.25ϫ Tris borate-EDTA (0.022 M Tris, 0.022 M boric acid, 0.5 mM EDTA). Gels were dried, and complexes were detected by phosphorimaging on an Amersham Biosciences Typhoon.

Nkx2.2 Is Down-regulated as CG4 Cells
Are Induced to Terminally Differentiate-CG4 cells are derived from primary cultures of rat oligodendrocyte precursor cells (21). They proliferate in the presence of platelet-derived growth factor and basic fibroblast growth factor. Upon withdrawal of growth factors, these cells are capable of differentiating into mature oligodendrocytes. After being transplanted into the central nervous system, they function as oligodendrocytes in vivo to myelinate axons (22). Nkx2.2 is expressed in oligodendrocyte precursor cells and is important for oligodendrocyte specification in the developing spinal cord. We therefore first investigated the expression of Nkx2.2 in CG4 cells by Western blotting. Nkx2.2 was expressed at elevated levels in proliferating cells and was down-regulated when these cells undergo terminal differentiation (Fig. 1). After 3 days of culture in differentiation medium, Nkx2.2 was not detectable, whereas the expression of MBP was significantly elevated (Fig.  1). Thus the level of Nkx2.2 expression was inversely correlated with the levels of differentiation markers such as 2Ј,3Ј-cyclic nucleotide 3Ј-phosphohydrolase and MBP. This expression pattern is similar to those reported in primary oligodendrocytes in vivo (23). The levels of the transcription factors Sp1 and Sox10 also increase in differentiating CG4 cells and primary oligodendrocytes (5,6). These data suggested that CG4 cells represent an accurate model in which to study Nkx2.2 expression and function during differentiation.
Nkx2.2 Inhibits MBP Promoter Activity-To explore the function of Nkx2.2 in the regulation of MBP expression, reporter gene constructs containing either 1323 or 105 bp of the MBP promoter upstream of the luciferase gene were made ( Fig.  2A). CG4 cells were cotransfected with each of the MBP-luciferase constructs and either a control plasmid or a plasmid encoding Nkx2.2. After transfection the cells were switched to differentiation medium for 2 days before harvesting for luciferase assays. The results of these transfection experiments are shown in Fig. 2B. Compared with control, expression of Nkx2.2 caused a dramatic decrease of luciferase activity driven by either the Ϫ1323 or Ϫ105 MBP promoter (Fig. 2B). Transfection experiments using a different expression vector encoding Nkx2.2 gave rise to the same results (data not shown). Similar repression effects were also obtained by co-transfection of reporter gene constructs and Nkx2.2 plasmids into CG4 cells under growing conditions (data not shown). Therefore, it appeared that expression of Nkx2.2 strongly repressed MBP promoter activity in both proliferating and differentiating cells. These data also suggested that the Ϫ105 MBP promoter contains regulatory elements that respond to the expression of Nkx2.2.
Nkx2.2 Binds to the MBP Promoter through Two Distinct Binding Sites-The Nkx family of proteins has been shown to modulate gene expression through interacting with multiple cis-elements within various target genes. It is likely that the repression of the MBP promoter by Nkx2.2 also shares a similar mechanism. To test this hypothesis, we first did DNase I footprinting to identify the cis-elements involved in the binding of Nkx2.2 to the MBP promoter. His-tagged Nkx2.2 was expressed in E. coli, and recombinant protein was purified. Purified protein was incubated with a 32 P-labeled fragment of the Ϫ1323 to ϩ30 (Ϫ1323Luc) or Ϫ105 to ϩ30 (Ϫ105Luc) were subcloned upstream of the luciferase gene (LUC) in pGL3Basic. B, CG4 cells were transfected with Ϫ1323Luc or Ϫ105Luc plus either empty vector (Control) or an Nkx2.2 expression plasmid. Transfections were done in growth medium using GenePorter. After transfection, the medium was changed to DM. After 48 h in DM, cells were harvested, and luciferase activity was measured and normalized to protein concentration. Data are the mean Ϯ S.E. from at least three replicates. *, p Ͻ 0.05 with respect to control of same reporter construct.
MBP promoter containing sequences between Ϫ300 and ϩ30 and subjected to increasing amounts of DNase I digestion. In this region of the MBP promoter, two sites were found to be protected from DNase I digestion by Nkx2.2. These are shown in Fig. 3A in the boxes. Interestingly both binding regions are located downstream of Ϫ105 in the MBP promoter, which is consistent with luciferase data showing that this length of promoter could still be repressed by Nkx2.2. Inspection of the sequences protected by Nkx2.2 revealed a conserved Nkx2.2 binding site, ACTTGA (Ϫ83 to Ϫ78), within site A (Fig. 3B, underlined), while the second site showed no sequence homology to the consensus binding site (Fig. 3B, site B).
Binding of Nkx2.2 to site A in the MBP promoter was also confirmed by EMSAs using recombinant Nkx2.2. Two probes were used in these experiments. One was an MBP promoter fragment spanning Ϫ95 to Ϫ76 containing the consensus Nkx2.2 binding site. The other was a probe with mutations in the consensus binding site (Fig. 4A). Nkx2.2 was capable of binding to the probe containing the consensus Nkx2.2 site in the wild type MBP promoter context (Fig. 4B, lane 2). This binding was specific since it could be competed by unlabeled oligonucleotides containing the consensus sequence but not by oligonucleotides containing nonspecific sequences (data not shown). However, there was no complex formed when the probe containing a mutation in the consensus sequence was used as a probe (Fig. 4B, lane 4). This indicated that the mutation in the consensus binding site completely disrupted the binding of Nkx2.2 to this region of the MBP promoter.
To pinpoint the Nkx2.2 binding sequences in site B of the MBP promoter, we made a series of probes corresponding to the sequences between Ϫ27 and Ϫ10 of the MBP promoter. Each oligonucleotide contained a different mutation in site B (Fig. 5A). EMSAs were performed using these probes and recombinant Nkx2.2. The results are shown in Fig. 5B.
Nkx2.2 did bind to this region of the MBP promoter as shown in lane 1, which is consistent with our DNase I footprinting assay. The binding was also specific as it could be competed by unlabeled specific competitor oligonucleotides but not by other nonspecific competitors (data not shown). Mutation of the sequence GGCCCTC led to partial or complete loss of Nkx2.2 binding (Fig. 5B, lanes 2-4), while mutation of bases other than this sequence had no significant effect on binding (Fig. 5B, lane 5). These data indicated that Nkx2.2 bound to site B through the GGCCCTC sequence.
Nkx2.2 Has a Higher Binding Affinity for Site A than Site B, but Both Sites Are Involved in Nkx2.2-mediated Repression of the MBP Promoter-To assess the relative binding affinity of Nkx2.2 for sites A and B, we did a competitive binding assay using either unlabeled site A-containing oligonucleotides or unlabeled site B-containing oligonucleotides to compete the binding of Nkx2.2 to a labeled probe containing site B. Both competitors were capable of effectively competing with the probe for Nkx2.2 binding (Fig. 6). However, the cold competitor containing site A was a better competitor of the binding of Nkx2.2 to the probe than the site B-containing competitor (Fig.  6B). Similar experiments were also performed using unlabeled competitors to compete the binding of Nkx2.2 to site A-containing probes with similar results (data not shown). Taken together, these results indicated that Nkx2.2 bound with a higher affinity to site A-containing sequences than to site Bcontaining sequences.
To test whether either one or both sites are involved in the Nkx2.2-mediated repression of the MBP promoter, transient transfection experiments were performed. Reporter gene constructs driven by either the wild type MBP promoter or the MBP promoter containing mutations in one or the other Nkx2.2 binding sites were used. The mutations in site A (CTT mutated to ACA) or site B (CCC mutated to AAA) were those shown to completely disrupt the binding of Nkx2.2 to the MBP promoter (Figs. 4 and 5). Expression of Nkx2.2 was still able to repress promoter activity when either site A or site B was mutated in the MBP promoter. However, the amount of repression was not as great as with the wild type promoter (Fig. 7). However, when both binding sites were mutated Nkx2.2 was no longer able to repress MBP promoter activity. Instead there was a slight increase in luciferase activity, which may result from the titration of corepressors away from the promoter by Nkx2.2. These data indicated that both sites A and B were involved in Nkx2.2mediated repression of the MBP promoter.
The Homeodomain of Nkx2.2 Is Required for Binding and Repression of the MBP Promoter-The Nkx2 family of transcription factors contains three highly conserved regions including the homeodomain, the tin domain, or the N-terminal decapeptide and the Nkx2-specific domain (24,25). The homeodomain of this family is a highly conserved DNA binding domain and is believed to recognize the DNA sequence 5Ј-CAAG-3Ј (26). A single amino acid replacement in the homeodomain can abrogate binding activity and inactivate the protein (26). This mutation does not interfere with the nuclear localization signals of Nkx2.2 (27). To test whether the homeodomain of Nkx2.2 mediates the binding to the MBP promoter, His-tagged Nkx2.2 and the mutant form, His-NkxNQ, containing the mutation Asn to Gln, were expressed in E. coli. The recombinant proteins were purified and used in EMSAs. The purified proteins were incubated with 32 P-labeled probes containing either site A or site B as described in Fig. 6. Wild type Nkx2.2 bound to both probe A and probe B (Fig. 8A). However, an equal amount of NkxNQ did not bind to either probe (Fig.  8A). This indicated that the homeodomain mediated the binding of Nkx2.2 to both the consensus and non-consensus binding sites in the MBP promoter. Consistent with the binding results, in transient transfection experiments wild type Nkx2.2 repressed the promoter, while expression of NkxNQ had no significant effect on the activity of the MBP promoter (Fig. 8B). These experiments demonstrated that the homeodomain of Nkx2.2 was required for binding and repression of the MBP promoter in CG4 cells.
Binding of Nkx2.2 to Site B Displaces the Binding of Pur␣ to the MBP Promoter-Pur␣ was first identified as a singlestranded DNA-binding protein involved in DNA replication (28). Since then, it has been shown that Pur␣ also binds to RNA and double-stranded DNA and is essential for various cellular activities including gene transcription (29 -31). Pur␣ is found to be highly abundant in brain and stimulates transcription of MBP (32). In CG4 cells, Pur␣ together with the Sp1 family of transcription factors synergistically activates MBP gene transcription (4). Interestingly the binding element for Pur␣, as for Nkx2.2, is located between Ϫ27 and Ϫ10 of the MBP promoter. Thus it is possible that Pur␣ and Nkx2.2 compete for binding to the MBP promoter. To test this possibility, recombinant Pur␣ was first purified from bacteria and tested for binding to the MBP promoter in EMSA experiments. As shown in Fig. 9, Pur␣ did bind to the Ϫ27 to Ϫ10 region of the MBP promoter (Fig.  9B), which is consistent with previous reports (4). A close examination of the binding sequence revealed that Pur␣ binds to the GGCCC sequence. The first two bases of this sequence were important for binding because mutation of these bases resulted in a complete loss of binding (Fig. 9B). It therefore appears that the Pur␣ binding site overlaps the Nkx2.2 binding site in this region of the MBP promoter (Fig. 9C).
To determine whether Pur␣ and Nkx2.2 compete for binding to their respective sites in the MBP promoter, we used EMSA experiments to test whether Pur␣ had any effect on the binding of Nkx2.2 and vice versa. As shown in Fig. 10, both Nkx2.2 and Pur␣ bound to the Ϫ27 to Ϫ10 region of the MBP promoter independently (Fig. 10). When Nkx2.2 was prebound to the MBP promoter by first incubating Nkx2.2 with the probe, addition of Pur␣ to the reaction did not disrupt the formation of the Nkx2.2-DNA complexes (Fig. 10). However, if Pur␣ was incubated with the probe first, addition of Nkx2.2 to the reaction appeared to disrupt the Pur␣-DNA complexes. This effect was dose-dependent (Fig. 10). These results indicated that Nkx2.2 was able to compete off Pur␣ that was prebound to site B in the MBP promoter. Since Pur␣ functions as a transcriptional activator in the regulation of MBP transcription, Nkx2.2 may repress MBP gene expression in proliferating oligodendrocytes by preventing the binding of Pur␣ to the promoter.
Interaction of Nkx2.2 with HDAC1 Also Contributes to the Repression of the MBP Promoter-Recruitment of corepressor complexes to target genes is one of the important mechanisms of transcriptional repressor-mediated gene repression. Other transcription factors in the homeodomain family have been found to recruit corepressor complexes to target genes through interaction of the homeodomain with HDAC1 (19,33,34). To test the possibility that Nkx2.2 represses expression from the MBP promoter through interaction with a corepressor, FLAGtagged HDAC1 and Nkx2.2 were ectopically expressed in CG4 cells. The interaction between Nkx2.2 and HDAC1 was evaluated using a co-immunoprecipitation assay. When HDAC1 was precipitated, Nkx2.2 was also found in the precipitates (Fig.  11A). When the reverse experiment was performed and Nkx2.2 was precipitated, HDAC1 could also be found (Fig. 11B), suggesting that Nkx2.2 and HDAC1 form a stable complex in vivo in CG4 cells.
HDAC1 is known to be one of the essential components of a number of corepressor complexes (35,36). It is most commonly known to form a stable complex with mSin3A and mediate transcriptional repression (18,37). Our experiments also showed that mSin3A formed a stable complex with HDAC1 and Nkx2.2 in CG4 cells because it could be co-immunoprecipitated with HDAC1 (Fig. 11A) or Nkx2.2 (Fig. 11B). This indicated that Nkx2.2 formed a complex with an HDAC1-mSin3A complex to repress the MBP promoter.
To further investigate whether HDAC activity is required for Nkx2.2 to repress the MBP promoter, the effect of TSA, a potent HDAC inhibitor, on the transcriptional repressor activity of Nkx2.2 was examined. The ability of Nkx2.2 to repress transcription of Ϫ105 MBPLuc was compromised in the presence of a low concentration of TSA and completely abolished at a higher concentration (Fig. 11C). Therefore, it appeared that HDAC1 activity was absolutely necessary for Nkx2.2-mediated repression of MBP promoter activity in CG4 cells. These data suggested that recruitment of an HDAC1-mSin3A corepressor complex by Nkx2.2 also contributed to its repression of the MBP promoter in CG4 cells.
Binding of Sp1 Displaces Binding of Nkx2.2 to the Consensus Site and Reverses the Nkx2.2-mediated Repression of the MBP Promoter-Our previous work has shown that the proximal GC box of the MBP promoter is essential for binding of the Sp1  2-5). In lanes 6 -10, the probe was first incubated with Pur␣ and then incubated without (lane 6) or with increasing amounts of Nkx2.2 (lanes 7-10). The arrows on the right indicate the position of the protein-DNA complexes formed by either Nkx2.2 or Pur␣. family of transcription factors and activation of the MBP promoter (5). A less conserved GC box, several bases downstream of the conserved one, is also involved in Sp1 binding. Both sites are involved in the synergistic activation of the MBP promoter by Sp1 and the glial specific transcription factor, Sox10 (6). The conserved Nkx2.2 binding site is located between the two Sp1 binding elements (Fig. 12A). To investigate how Sp1 and Nkx2.2 might coordinately regulate the expression of MBP, we first evaluated the binding of Sp1 and Nkx2.2 to the MBP promoter. Purified Sp1 from HeLa cells and recombinant Nkx2.2 were used in EMSAs. The probe used in these experiments encompassed the sequences between Ϫ105 and Ϫ52 of the MBP promoter. This region contains two Sp1 binding sites and the consensus Nkx2.2 binding site. As expected, both Sp1 and Nkx2.2 bound to the probe independently (Fig.  12B). When the probe was preincubated with Sp1, addition of Nkx2.2 to the binding reaction had no significant effect on the formation of the Sp1-DNA complexes (Fig. 12B). However, if the probe was preincubated with Nkx2.2, addition of Sp1 to the binding reaction led to a significant loss of the Nkx2.2-DNA complexes. Similar experiments were also performed using probes containing Nkx2.2 binding site B. Sp1 did not bind to this region of the MBP promoter and had no significant effect on the binding of Nkx2.2 to site B (Fig. 12C). This effect was not likely to have resulted from direct protein-protein interaction between Sp1 and Nkx2.2 because there is currently no evidence that these two proteins can interact.
When oligodendrocytes are induced to differentiate, the expression of Sp1 increases (5). Since increased Sp1 can compete off the binding of Nkx2.2 to the MBP promoter in vitro, activation of the MBP promoter in differentiating cells might, at least in part, result from the displacement of Nkx2.2 from the MBP promoter by Sp1. To test this possibility, CG4 cells were cotransfected with a constant amount of plasmid encoding Nkx2.2 and increasing amounts of plasmid encoding Sp1 together with the Ϫ105 MBPLuc reporter construct. As seen before, expression of Nkx2.2 led to repression of MBP promoter activity, while expression of Sp1 alone led to activation of the MBP promoter (Fig. 13A). When Sp1 and Nkx2.2 were coexpressed in these cells repression of MBP promoter activity by Nkx2.2 was reversed by Sp1. This effect was Sp1 concentrationdependent (Fig. 13A). Expression of either Sp1 or Nkx2.2 had no effect on the expression of the other as shown by Western  4). The asterisk marks the position of the IgG from the immunoprecipitation. B, the experiment was performed as in A except that immunoprecipitation was done using antibody to Sox10 or Nkx2.2. Western blotting was done for mSin3A, FLAG-tagged HDAC1, and Nkx2.2. The double asterisks mark the position of the IgG from the immunoprecipitation. C, inhibition of histone deacetylase activity blocks the ability of Nkx2.2 to repress the MBP promoter. CG4 cells were transfected with Ϫ105Luc plus either empty vector (Control) or an Nkx2.2 expression plasmid. Transfections and luciferase assays were performed as in Fig. 2 except that, after transfection, TSA was added to the medium at a final concentration of 5.0 or 10 g/ml as indicated. The same volume of Me 2 SO was added to control cultures. *, p Ͻ 0.01 with respect to control. IP, immunoprecipitation. blotting (Fig. 13B). This effect was specific to Sp1 because expression of other transcriptional activators, such as Sox10, was unable to overcome the repressive activity of Nkx2.2 (data not shown).

DISCUSSION
Role of Nkx2.2 in Oligodendrocytes-Nkx2.2 belongs to a family of genes whose homeodomains are homologous to the Drosophila NK-2 gene (38,39). In addition to Nkx2.2, at least nine other family members have been identified and shown to be essential regulators of tissue development in vertebrates. In the central nervous system, Nkx2.2 has a primary role in ventral neuronal patterning and functions as a repressor to inhibit motor neuron generation (40,41). Together with Olig2, Nkx2.2 establishes the boundary to specify oligodendrocyte precursor cells (40,42,43). These findings indicate that Nkx2.2 is involved in the initiation and specification of oligodendrocyte lineage cells. There is also evidence that Nkx2.2 plays a role in the transition of oligodendrocytes from progenitor cells to mature oligodendrocytes (14). However, expression of Nkx2.2 alone is insufficient to promote oligodendrocyte development in the dorsal spinal cord (43).
Nkx2.2 Is Down-regulated in Terminally Differentiated Cells-Expression of Nkx2.2 begins well before the expression of myelin genes in oligodendrocytes. It is extensively expressed even before oligodendrocyte specification of the neuroepithelial cells of the ventricular zone (12). In our study, we found that Nkx2.2 was expressed in proliferating CG4 cells and was downregulated as these cells underwent terminal differentiation. Other reports have also shown that Nkx2.2 is down-regulated in vivo in mature oligodendrocytes. Toward the end of the embryonic stage in the chick spinal cord, when mature oligodendrocytes appear, the expression of Nkx2.2 is down-regulated (44). A recent report also shows that there is a transient up-regulation of Nkx2.2 expression in oligodendrocyte precursor cells and immature oligodendrocytes during remyelination, while the expression of Nkx2.2 is dramatically decreased in terminally differentiated oligodendrocytes (23). These observations are consistent with our findings that Nkx2.2 was ex- Ϫ105Luc plus either empty vector, an Nkx2.2 expression plasmid, or increasing amounts of an Sp1 expression plasmid as indicated. Transfection and assays were performed as described in Fig. 2 pressed in proliferating CG4 cells and that its expression decreased as the cells differentiated.
Nkx2.2 Represses MBP Expression-The targets of Nkx2.2 regulation are largely unknown in the central nervous system. Our findings indicated that one role of Nkx2.2 is to repress the expression of MBP in proliferating oligodendrocytes and to prevent its precocious expression prior to the onset of terminal differentiation. The expression of genes encoding myelin proteins, including MBP, increase dramatically as oligodendrocytes differentiate. It has been proposed that the transcription of these genes is repressed through the binding of transcriptional repressors to the negative regulatory regions in the promoters in immature oligodendrocytes (45)(46)(47)(48). We found that expression of Nkx2.2 repressed MBP promoter activity in both proliferating and differentiating oligodendrocytes in culture. By DNase I footprinting and EMSA analysis, two ciselements were identified in the proximal region of the MBP promoter. These include an Nkx consensus binding site and a novel binding site. Further experiments demonstrated that the novel binding site, despite its divergence from the core homeodomain consensus binding motif, was also involved in Nkx2.2 repression of the MBP gene. Other transcription factors, including homeodomain-containing proteins, have also been reported to modulate gene expression through interaction with regulatory elements that do not conform to their consensus binding motifs (49 -51).
However, Nkx2.2 can also act as a transcriptional activator to stimulate target genes. Both activation and repression domains have been found in the Nkx family of proteins (26,52,53). In the pancreas, Nkx2.2 is required for the final differentiation of the beta cells (54 -56). Additionally the insulin gene is directly activated by Nkx2.2 in differentiated pancreatic beta cells (54). In migrating and differentiating oligodendrocytes, expression of Nkx2.2 spatially overlaps the expression of several early oligodendrocyte lineage markers including plateletderived growth factor receptor ␣, Sox10, and galactocerebroside (43,44,57). Thus it has also been proposed that Nkx2.2 might act as an activator of the expression of myelin genes. Expression of Nkx2.2 in fibroblasts was indeed found to activate proteolipid protein (PLP) expression (58). Conversely differentiated oligodendrocytes that strongly express proteolipid protein have an undetectable level of Nkx2.2 expression, and Nkx2.2 ϩ cells show very weak expression of PLP (44). One possible explanation is that the promoter activity in the fibroblasts may not be an accurate reflection of promoter response in an oligodendrocyte context as shown by others (59) and in our previous findings (60). While oligodendrocyte precursors can express PLP and an Nkx2.2 consensus binding site is also found in the PLP promoter, the role of Nkx2.2 in the regulation of PLP expression in oligodendrocyte lineage cells needs to be further investigated.
Molecular Mechanisms of Nkx2.2 in Repression of MBP Expression-Our studies revealed that there are at least two mechanisms involved in Nkx2.2-mediated repression of MBP promoter activity. First binding of Nkx2.2 to the MBP promoter occludes the binding of the transcriptional activator Pur␣. Pur␣ is a single-stranded DNA-binding protein that interacts with the purine-rich strand in a sequence-specific manner. It also binds to double-stranded DNA. Pur␣ has been shown to be able to transactivate both viral and mammalian promoters (61)(62)(63). In the central nervous system, Pur␣ was found to bind to the proximal region of the MBP promoter and stimulate MBP transcription (32). Sp1 transiently interacts with Pur␣, and this interaction may result in the stable association of Pur␣ with the promoter to activate the MBP promoter (4, 64). In our study, we showed that Pur␣ bound a sequence that partially overlaps with Nkx2.2 binding site B. Further competition experiments based on EMSAs showed that binding of Nkx2.2 was able to compete off the binding of Pur␣ to this site. These studies indicated that Nkx2.2 was able to repress MBP promoter activity by preventing the binding of at least one other transcriptional activator.
A second mechanism involved in Nkx2.2 repression of MBP promoter activity involves the recruitment of a corepressor complex containing HDAC1 to the MBP promoter. The N termini of core histones can be modified by acetylation. Acetylation of histones increases the accessibility of ATP-dependent chromatin-remodeling complexes, such as SWI/SNF and ISWI complexes, and facilitates the disruption of nucleosome structure (65). Therefore, acetylated histones are generally associated with actively transcribed genes. Deacetylation of histones by HDACs such as the HDAC1-mSin3A complex can establish repression. During oligodendrocyte development, the expression of many genes, such as those for PLP and MBP, are maintained at low levels in progenitor cells. It has been proposed that the transcription of these genes is repressed through binding of the transcriptional repressors to the negative regulatory regions in the promoter region (31,(45)(46)(47). Using coimmunoprecipitation, we found that Nkx2.2 could form a complex with mSin3-HDAC1 complexes. Additionally inhibition of HDAC activity by TSA blocked the ability of Nkx2.2 to repress MBP promoter activity. These experiments indicated that Nkx2.2 regulated the stage-specific expression of MBP in oligodendrocytes by recruiting mSin3-HDAC complexes to the MBP promoter.
Similar mechanisms have been reported in the regulation of gene expression in other cells. In non-endothelial cells, the NFY transcription factor recruits HDAC1 and -2 to inhibit von Willebrand factor promoter activity (51). In premitotic neurons, the expression of the transcriptional repressor element 1-silencing transcription factor recruits an HDAC complex to neuron-specific genes to repress their expression, while in postmitotic neurons, decreased expression of repressor element 1-silencing transcription factor results in the disruption of these repressive complexes and activation of target genes (66). Interestingly, in primary oligodendrocyte cultures, histone deacetylation has been observed during the window between exit from the cell cycle and onset of differentiation (47). The idea that HDAC activity is important for oligodendrocyte lineage progression is consistent with our findings that Nkx2.2 helped to prevent precocious expression of MBP in immature oligodendrocytes.
Regulation of MBP Expression by Nkx2.2 and Sp1-The transition of MBP expression from basal levels in progenitor or early differentiating cells to high levels in mature oligodendrocytes is not completely understood. Various transcription factors have been found to be involved in this process (67,68). In our previous studies, we found that Sp1, together with Sox10, plays an essential role in the regulation of oligodendrocytespecific expression of MBP (5,6). In the present study we showed, using an in vitro binding assay, that Sp1 was able to compete off the binding of Nkx2.2 to its consensus binding site. Expression of Sp1 could also reverse the repression of the MBP promoter mediated by Nkx2.2. Thus we propose that, in oligodendrocyte progenitors and immature oligodendrocytes, the MBP promoter is mainly recognized and bound by transcriptional repressors such as Nkx2.2. Nkx2.2 recruits the HDAC1-mSin3A complex to the promoter, keeping the core histones deacetylated. Deacetylated histones promote the condensation of chromatin and inhibit transcription initiation (35,36). When the cells are induced to differentiate, expression of Nkx2.2 is down-regulated. The absence of Nkx2.2 in mature oligodendro-cytes results in the release of repressive complexes from the promoter. At the same time the levels of other transcriptional activators, such as Sp1 and Sox10, are increasing (5,6). The removal of Nkx2.2 from the promoter would also allow access of these factors to their sites on the MBP promoter and lead to activation of gene expression.