A Candidate X-linked Mental Retardation Gene Is a Component of a New Family of Histone Deacetylase-containing Complexes*

Eukaryotic genes are under the control of regulatory complexes acting through chromatin structure to control gene expression. Here we report the identification of a family of multiprotein corepressor complexes that function through modifying chromatin structure to keep genes silent. The polypeptide composition of these complexes has in common a core of two subunits, HDAC1,2 and BHC110, an FAD-binding protein. A candidate X-linked mental retardation gene and the transcription initiation factor II-I (TFII-I) are components of a novel member of this family of complexes. Other subunits of these complexes include polypeptides associated with cancer causing chromosomal translocations. These findings not only delineate a novel class of multiprotein complexes involved in transcriptional repression but also reveal an unanticipated role for TFII-I in transcriptional repression.

The genome of eukaryotes is packaged into chromatin, the fundamental unit of which is the nucleosome. The higher order chromatin structure is formed by arrangement of nucleosomes into an array. Such a higher order chromatin structure presents a barrier to cellular processes such as transcription, DNA replication, and DNA repair. Therefore, controlling accessibility to the nucleosomal DNA provides an important regulatory point in these processes (1). One way to modulate nucleosomal structure is through enzymatic modification of histones by acetylation, phosphorylation, or methylation.
A number of transcriptional regulatory complexes have been identified that contain histone acetylation or deacetylation activities. It was previously shown that the hyperacetylated chromatin correlates with active genes whereas the repressed genes exhibit a pattern of hypoacetylation (2,3). This contention was strengthened by the discovery of the association of a number of transcriptional corepressors with histone deacetylation activity. The HDACs 1 identified in mammalian cells can be divided into three classes. Homologs of the yeast protein Rpd3 are members of the Class I HDACs (4,5). Included in this class are HDAC1, HDAC2, HDAC3, and HDAC8. Members of the Class II HDACs include HDAC4, HDAC5, HDAC6, HDAC7, HDAC9, and HDAC10. These HDACs appear to be more similar to yeast protein Hda1 (6 -9). A third class of HDACs exists, which unlike the other classes of HDACs, requires an NAD cofactor for activity. The members of this class are homologs of the yeast Sir2 protein (10 -12).
Previous biochemical analysis revealed the association of transcriptional corepressor Sin3 with a multiprotein complex containing histone deacetylase activity (13)(14)(15). This complex was shown to contain HDAC1,2 and act as a transcriptional corepressor for a number of DNA-binding repressors including Mad, the nuclear hormone receptors, and the RE1-binding silencer protein, REST (also called NRSF) (14, 16 -19). In addition, a number of groups reported the isolation and characterization of a complex termed NuRD (also NURD and NRD) that not only contains histone deacetylases 1 and 2 but also a DNA-dependent ATPase subunit (20 -22).
Here, we report the isolation of a new family of HDAC1,2 complexes that also contain the FAD-binding protein BHC110 (23). Unique to this family of corepressor complexes is the presence of a distinct structural DNA-binding subunit defining different HDAC1/2-containing complexes.

MATERIALS AND METHODS
Immunoaffinity Purification of the BHC110/HDAC2-containing Complex-HeLa nuclear extract was fractionated according to the protocol described above using P11. Anti-BHC110 and anti-HDAC2 antibodies (500 g each) were cross-linked to Protein A-Sepharose (1 ml, Repligen) using standard techniques for affinity purification. The P11 0.3 M KCl fraction was incubated with 1 ml of antibody-Protein A beads for 4 -5 h at 4°C. The beads were washed first with 1 M KCl in buffer A (20 mM Tris-HCl, pH 7.9, 0.2 mM EDTA, 10 mM ␤-mercaptoethanol, 10% glycerol, 0.2 mM PMSF) followed by a wash with 0.5 M KCl in buffer A with 0.5% Tween 20. The beads were then washed with 100 mM KCl in buffer A, and the proteins were eluted with 0.1 M glycine, pH 2.5, and neutralized with 0.1 volume of 1.0 M Tris-HCl, pH 8.0.
Affinity Purification of FLAG-XFIM-FLAG-XFIM and a selectable marker for puromycin resistance were co-transfected into 293 human embryonic kidney cells by calcium phosphate co-precipitation. Transfected cells were grown in the presence of 10 g/ml puromycin, and individual colonies were isolated and analyzed for FLAG-XFIM expression. A cell line expressing FLAG-tagged XFIM, F-XFIM, was used for the affinity purification of the XFIM-containing complex as previously described for the FLAG-BRAF35 cell line (23).
Chromatographic Purification of TFII-I Complex from HeLa Nuclear Extract-HeLa nuclear extract (3 g) was loaded on a 500-ml column of phosphocellulose (P11, Whatman) and fractionated stepwise by the indicated KCl concentration in buffer A. The P11 0.3 M KCl fraction (700 mg) was loaded on a 80-ml DEAE-Sephacel column (Amersham Bio-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  . The column was resolved by using a linear 10-column volume gradient of 50 -500 mM KxPO 4 . A pool of fractions 28 -30 was dialyzed to 700 mM NH 4 SO 4 in Buffer HB (20 mM HEPES, pH 7.6, 4 mM dithiothreitol, 0.5 mM EDTA, 10% glycerol, 0.5 mM PMSF) and loaded on a butyl-Sepharose column (Amersham Biosciences). The column was resolved using a linear 10-column volume gradient of 700 to 0 mM NH 4 SO 4 in Buffer HB. TFII-I-containing fractions 12-16 were dialyzed to 100 mM KCl in Buffer A and loaded on Heparine-5PW (TosoHaas). The column was resolved using a linear 20-column volume gradient of 100 -500 mM KCl in Buffer A. TFII-I containing fractions 12-16 was fractionated on a Superose 6 HR 10/30 column (Amersham Biosciences) equilibrated in 0.5 M KCl in buffer A. Superose 6 was calibrated using molecular weight standards from Amersham Biosciences. The void was determined according to the manufacturer's guidelines (one-third of column volume or 7 ml). Fractions 16 -20 and 24 -28 were used for immunoaffinity purification of the TFII-I-containing complexes.
Mass Spectrometric Peptide Sequencing-Excised bands were subjected to in-gel reduction, carboxyamidomethylation, and tryptic digestion (Promega). Multiple peptide sequences were determined in a single run by microcapillary reverse-phase chromatography (a custom New Objective 50-m column terminating in a nanospray 15-m tip), directly coupled to a Finnigan LCQ Deca quadrupole ion trap mass spectrometer. The ion trap was programmed to acquire successive sets of three scan modes consisting of: full scan MS over alternating ranges of 395-800 m/z or 800 -1300 m/z, followed by two data-dependent scans on the most abundant ion in those full scans. These dependent scans allowed the automatic acquisition of a high resolution (zoom) scan to determine charge state and exact mass and MS/MS spectra for peptide sequence information. MS/MS spectra were acquired with a relative collision energy of 30%, an isolation width of 2.5 daltons, and dynamic exclusion of ions from repeat analysis. Interpretation of the resulting MS/MS spectra of the peptides was facilitated by programs developed in the Harvard Microchemistry Facility (24) and by data base correlation with the algorithm SEQUEST (25).
Immunoblot Analysis-Anti-BHC110, anti-BHC80, and anti-BRAF35 antibodies were described previously (23). Anti-HDAC2 anti-bodies were obtained from Zymed Laboratories Inc.. Anti-TFII-I and XFIM antibodies were developed to a peptide corresponding to the amino acids IKETDGSSQIKQEPDPTW and DPLTLPEKPLAGDLP for TFII-I and XFIM, respectively. Immunoblotting was performed with alkaline phosphatase.
RNAi and Transfections-The small interfering RNA (RNAi) sequence targeting TFII-I (AA GUU ACU CAG CCA AGA ACG A) or the control RNAi was purchased from Dharmacon. Transfections were performed on 2 ϫ 10 6 HeLa cells with a final concentration of 200 mM small interfering RNA duplex using Oligofectamine reagent (Invitrogen) according to the manufacturer's guidelines. After two rounds of RNAi treatment followed by 12 h of starvation in 1% serum medium, cells were subsequently stimulated with 50 nM epidermal growth factor (EGF) (Sigma) and left for the times indicated before harvesting (see Fig. 6b). containing complexes following the scheme in Fig. 1. The anti-BHC110 affinity eluate was subjected to ion trap mass spectrometric sequencing. In addition to other components of the BHC complex (23), this analysis revealed the stable association of BHC110 with ZNF261/XFIM, a candidate gene for X-linked mental retardation in Xq13.1 (28,29), ZNF198/FIM, a gene related to XFIM that is associated with myeloproliferative disorder that involves myeloid hyperplasia and eosinophilia (29,30), KIAA0182, a proline-rich protein of unknown function, and TFII-I, the initiator binding protein and a transcriptional coactivator (31, 32) (Fig. 1, lane 2, and Fig. 2). The association of these polypeptides and BHC110 is specific as the affinity eluate from a control antibody column was devoid of their presence (Fig. 1, lane 3).

BHC110 Defines a New Family of HDAC-containing Com
To determine whether these polypeptides are also stable components of an HDAC-containing complex, we affinity-purified HeLa HDAC2-containing complexes from the 0.3 M KCl eluate of phosphocellulose chromatography (P11) (Fig. 1). Ion trap mass spectrometric sequencing of the anti-HDAC2 affinity eluate revealed that, in addition to components of the previously described complexes of Mi2 (20 -22), Sin3 (13)(14)(15), and BHC (23,26,27), the anti-HDAC2 eluate contained ZNF261/ XFIM, ZNF198/FIM, KIAA0182, and TFII-I (Fig. 1, lane 1, and  Fig. 2). These results indicate that these novel subunits are associated with both HDAC2 and BHC110, although these polypeptides most likely represent multiple distinct HDAC2/ BHC110-containing complexes. To test this hypothesis we embarked on isolating other BHC110-containing complexes that are distinct from the BHC complex.
TFII-I Is a Component of an XFIM Complex-To isolate other BHC110-containing complexes, we developed a 293-derived cell line stably expressing a FLAG-tagged XFIM (F-XFIM). FLAG-XFIM was affinity-purified from F-XFIM nuclear extract using anti-FLAG antibodies followed by elution of bound material with FLAG peptide (Fig. 3a). A combination of ion trap mass spectrometry and Western blot analysis demonstrated the presence of TFII-I, BHC110, and HDAC1/2 polypeptides (Fig.  3a). These polypeptides were absent in affinity-purified eluate of the parent 293 cell line (Fig. 3a). Further analysis of the XFIM complex by Superose 6 gel filtration confirmed the association of XFIM, TFII-I, BHC110, and HDAC2 as a component of a single complex of about 1 MDa, although a small percentage of HDAC2 and TFII-I eluted at a smaller molecular mass (Fig. 3c). This complex was termed the XFIM complex (Fig. 3a). Moreover the XFIM complex displayed HDAC activity toward core histones (Fig. 3c). It is noteworthy that analysis of the XFIM protein following gel filtration by colloidal staining revealed a higher stoichiometry for XFIM (ϳ4 XFIMs per complex) to other subunits of the complex. Taken together, these results establish BHC110 and HDAC2 as common subunits of at least two distinct (BHC and XFIM) histone deacetylase complexes (23).
To determine the fraction of TFII-I that associates with BHC110 we purified TFII-I by conventional column chromatography (Fig. 4a). Analysis of TFII-I on the gel filtration, the last step of purification, revealed the predominant peak of immunoreactivity at fraction 24, whereas a small portion (5-10%) of TFII-I eluted at a larger molecular mass coincident with BHC110 and HDAC2 (Fig. 4a). Affinity purification using anti-TFII-I antibodies revealed the association of BHC110 and   FIG. 4. A fraction of TFII-I is in a  stable complex with BHC110. a, conventional chromatographic purification of TFII-I followed by anti-TFII-I affinity as discussed under "Materials and Methods." The numbers delineate salt concentrations in molar. NE, nuclear extract. b, Western analysis of the affinity-purified ␣-HDAC2, ␣-BHC110, and ␣-BHC80 eluates using anti-TFII-I and anti-HDAC2 antibodies. c, Western analysis of the affinity-purified ␣-TFII-I or control ␣ϪTRAP220 eluates using anti-BHC110 and anti-HDAC2 antibodies. NE, (nuclear extract).
HDAC2 only with the TFII-I derived from the larger molecular mass fraction (Fig. 4a, see fractions 18 -20). Furthermore, immunoprecipitation using anti-HDAC2, anti-BHC110, and anti-TFII-I antibodies demonstrated the specific association of TFII-I with HDAC2 and BHC110 (Fig. 4, b and c). These results indicate that although TFII-I is predominantly monomeric, a fraction of TFII-I is in a stable complex with BHC110 and HDAC2.
XFIM Complex Is Recruited to the c-fos Promoter-Because TFII-I was reported as a transcriptional coactivator for serum response factor (SRF) at the c-fos promoter (33), we analyzed the c-fos promoter as a target of the XFIM complex. We first confirmed the responsiveness of the c-fos promoter to inhibitors of histone deacetylation. Consistent with previous reports c-fos displayed an increased transcription level following either sodium butyrate or trichostatin A treatment (34) (Fig. 5a). Importantly, ChIP experiments established the presence of the components of the XFIM complex at the c-fos promoter in its basal state (Fig. 5b). In response to mitogens and growth factors such as serum or EGF, c-fos displays classical immediateearly gene activation kinetics, where it is induced within 15 min of stimulation, followed by return to basal levels within 2 h of stimulation. Therefore, three phases of c-fos expression can be identified: an initial repressed state, an activated state, and a return to a repressed state (34) (Fig. 5c). To understand the role of the XFIM complex in each phase of c-fos responsiveness, we used ChIP to examine the c-fos promoter following EGF stimulation. Similar to results in Fig. 5b, analysis of the promoter in the initial repressed state revealed the presence of BHC110, HDAC2, and TFII-I in addition to that of the SRF protein (Fig. 5d). However, although the SRF levels at the promoter were enhanced 30 min following the stimulation of c-fos transcription, HDCA2 and BHC110 were no longer detectable (Fig. 5d). BHC110 and HDAC2 were returned to the promoter as the repressed state was reestablished (Fig. 5d, 60 and 90 min). Interestingly, TFII-I occupancy of the promoter was unchanged following EGF stimulation (Fig. 5d). Moreover, although the histone H3 acetylation state is not affected by EGF stimulation, there is an increase in acetylated histone H4 coincident with the absence of HDAC1,2 complex 30 min following EGF stimulation. These results suggest a role for the XFIM complex as a corepressor at the c-fos promoter and are consistent with the contention that TFII-I may play a dual role in that it participates as a component of a corepression complex in the basal repressed state of the promoter, but once the gene is activated it remains bound to the promoter to form a stable complex with the activator as previously described (33,34). . c, RT-PCR analysis of c-fos and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene transcription following the indicated times after stimulation by EGF in 293 cells. d, ChIP analysis of the c-fos SRE promoter in 293 cells using ␣-SRF, ␣-TFII-I, ␣-BHC110, ␣-HDAC2, ␣-acetylated histone H4, and ␣-acetylated histone H3. Total cell extracts were taken at the indicated times after EGF stimulation. Following immunoprecipitation of formaldehyde cross-linked lysates, PCR of eluted DNA using oligonucleotides specific for the c-fos SRE promoter was performed. Total chromatin extracts were used for input control. As a negative control, protein A-precipitated lysate was used.
Recruitment of TFII-I to the Promoter Results in Transcriptional Repression-To directly assess the role of TFII-I in transcription, we tethered TFII-I to the GAL4 DNA-binding domain and tested its activity using a promoter containing five GAL4binding sites (Fig. 6a). Interestingly, although GAL4-VP16 resulted in a potent activation of transcription from this promoter, GAL4-TFII-I caused a moderate (ϳ50%) repression of transcription (Fig. 6a). However, the transcriptional repression by Gal4-TFII-I was smaller than that obtained with either GAL4-SAP30 or GAL4-HP1␣, two previously characterized transcriptional corepressors (35,36). Taken together, our results point to a role for TFII-I in transcriptional repression.
To further assess the role of TFII-I in transcription of an endogenous c-fos promoter, we utilized small interfering RNAmediated depletion (RNAi) specific for TFII-I to inhibit its synthesis. Two rounds of RNAi treatment were necessary to see a substantial (larger than 80%) decrease in TFII-I mRNA levels (Fig. 6b). Analysis of c-fos transcription following TFII-I RNAi indicated a pronounced and specific de-repression of basal transcription in the absence of TFII-I (Fig. 6b). Moreover, the EGFmediated activation of c-fos promoter still persisted. It is difficult to assess the change in the -fold stimulation following TFII-I RNAi because there is no basal activity in the absence of the RNAi treatment. These results point to a role for TFII-I in the maintenance of the basal repressed state of the c-fos promoter. DISCUSSION We identify a new family of HDAC1,2-associated complexes containing BHC110. Moreover, we define the polypeptide composition of a novel member of this family containing the candidate gene for X-linked mental retardation XFIM and the initiator-binding protein TFII-I. A unique feature of HDAC1,2/ BHC110-containing complexes is the presence of specific DNAbinding subunits as a component of each individual complex. Indeed, the majority of the novel subunits identified are either known DNA-binding proteins such as BRAF35 (23) and TFII-I (this work) or are proteins with a putative role in DNA binding such as ZNF261/XFIM, ZNF198/FIM, and ZNF217 (Fig. 2).
The presence of specific and different DNA-binding subunits in each corepressor complex also allows for specificity in targeting each complex to a unique promoter. For example, the TFII-I-containing corepressor complex is recruited to promoters that either contain TFII-I-binding sites and/or DNA-binding proteins such as SRF, which can form a cooperative interaction with TFII-I (31,34). However, upon activation and a consequent increase in the activator concentration at the promoter, the association of TFII-I with the activator prevails over that of the corepressor leading to a function for TFII-I in the coactivation process. The DNA-binding subunit of the corepressor complexes may also increase the cooperative binding of sequence-specific repressor to their regulatory sites.
Although a number of the new HDAC1,2-associated subunits such as TFII-I and BHC80 are unique to mammalian species, others are evolutionarily conserved. Among these BHC110, ZNF261/XFIM, ZNF198/FIM, and BRAF35 have close homologs in Drosophila melanogaster, indicating that similar corepressor complexes may also be involved in gene-specific repression in D. melanogaster. Finally, the close association of a number of HDAC-associated subunits with specific disease states (Fig. 2) attests to the importance of this family of corepressor complexes in human health.