Histone Deacetylase 3 Binds to and Regulates the Multifunctional Transcription Factor TFII-I*

Histone deacetylase 3 (HDAC3) is one of four members of the human class I histone deacetylases that are implicated in transcriptional repression through deacetylation of acetyllysines in amino-terminal tails of core histones. In an immunoaffinity purification using anti-HDAC3, transcription factor TFII-I copurified with HDAC3. Specificity of the HDAC3-TFII-I interaction was confirmed by coimmunoprecipitation of epitope-tagged proteins, GST pull-down assays, and protein colocalization with indirect immunofluorescence. An anti-TFII-I immunoprecipitate contained histone deacetylase enzymatic activity. Mutational analyses revealed that the carboxyl-terminal of HDAC3 (residues 373–401) and residues 363–606 of TFII-I were required for the HDAC3-TFII-I interaction. Transcriptional activation by TFII-I was severely reduced by overexpression of HDAC3. These results suggest that HDAC3 modulates some of the functions of TFII-I and provides a link between histone deacetylase and a multifunctional transcriptional activator.

The organization of chromatin structure is a fundamental, yet extremely critical, component of gene regulation in all eukaryotic cells. Whether chromatin is transcriptionally active or repressed is determined, at least in part, by the modification of nucleosomal histones. Of the many possible post-translational modifications of histones, the most common are acetylation and deacetylation. In humans, about 20 histone deacetylase (HDAC) 1 enzymes have been identified whose functions are primarily to regulate the acetylation status of histones and maintain regions of chromatin in transcriptionally inactive states.
The first human histone deacetylase, HDAC1, was cloned by Schreiber and colleagues (1) using a deacetylase inhibitor affinity matrix. The second human histone deacetylase, HDAC2, was identified in our laboratory in a yeast two-hybrid experiment with the YY1 transcription factor as bait (2). HDAC1 and HDAC2 bear a high degree of sequence homology to the yeast protein RPD3, and both enzymes typically coexist in the same protein complexes.
The third human histone deacetylase, HDAC3, was cloned based on sequence similarities to HDAC1 and HDAC2 (3)(4)(5). Unlike HDAC1/2, HDAC3 is present in a unique large multisubunit protein complex that contains proteins different from those identified in HDAC1/2 complexes (6 -9). Also, unlike HDAC1 or HDAC2, HDAC3 is essential for cell viability and is localized in both the nucleus and cytoplasm of cells (10 -12).
Although HDAC3 clearly plays a crucial role in the regulation of gene expression, the challenge has been to understand the exact mechanisms by which HDAC3 modulates gene expression and affects the functions of the transcriptional machinery. More importantly, there is a need to address the functional differences between HDAC3 and HDAC1/2 and, possibly, other HDACs. The identification and characterization of HDAC1/2-binding proteins has been tremendously useful in elucidating the mechanisms of action and functions of these two HDACs. Following similar strategies, we isolated an endogenous HDAC3-containing multisubunit complex using anti-HDAC3 immunoaffinity chromatography (9). Besides HDAC3, this complex also contains NCoR/SMRT co-repressors plus several unidentified proteins. In parallel, using anti-NCoR/SMRT combined with conventional chromatography, four laboratories independently isolated complexes containing NCoR/SMRT, HDAC3, the WD-40 repeat-containing protein TBL1, SWI/SNF-related proteins, the co-repressor KAP-1, and multiple not-yet characterized proteins (6 -8, 13). We found that transcriptional repression by NCoR is mediated by its interaction with HDAC3 and that NCoR augments the enzymatic activity of HDAC3. Although these earlier findings confirm and extend the concept that HDAC3 is a major player in transcriptional repression, they raise several important questions. What other proteins are present in the HDAC3 complex? Do these proteins increase or antagonize HDAC3 activity? Does HDAC3 reciprocally modify the function of these HDAC3-binding proteins?
We have now identified an additional component of the HDAC3 complex. We show that the protein originally designated p130 (9) is the transcription factor TFII-I (14,15). TFII-I was earlier shown to be a basal transcription factor that binds to and activates transcription from the initiator element (16). Recent evidence suggests that TFII-I is also an inducible multifunctional factor that selectively regulates gene expression when activated by a variety of extracellular signals (17).
In this report, we present evidence that TFII-I interacts specifically with HDAC3, suggesting that HDAC3 may be a natural mediator of TFII-I functions. Consistent with the ability of these two proteins to interact, HDAC3 can also alter the transcriptional activity of TFII-I.
¶ pcDNA3-HA-TFII-I was constructed by subcloning TFII-I cDNA downstream of the cytomegalovirus promoter and in-frame with the HA sequence in pcDNA3.1 (Invitrogen). Plasmids that express HA-TFII-I deletions were constructed by digestion of pcDNA3-HA-TFII-I with various restriction enzymes, followed by religation. The plasmid expressing an enzymatic-deficient HDAC3 mutant (H134Q/H135A/ A136S) was generated using the Kunkel mutagenesis procedure as outlined in the Muta-Gene system (Bio-Rad). All constructs were verified by DNA sequencing.
Immunochemical Reagents and Techniques-Polyclonal anti-TFII-I antibody has been described (22). Monoclonal anti-FLAG M2, mono-clonal anti-GST, and polyclonal anti-HA antibodies were obtained from Sigma.
Immunoprecipitations were performed in a solution of PBS containing 0.1% Nonidet P-40 and protease inhibitors as previously described (23). Immunocomplexes were washed six times with the same buffer, and immunoprecipitated proteins were removed from protein A beads by either boiling in gel loading buffer or by elution with excess peptide antigens. For Western blot analyses, proteins were resolved on SDSpolyacrylamide gels and transferred to polyvinylidene difluoride membranes. After blocking with nonfat dried milk, the membranes were treated with diluted primary antibodies, followed by diluted alkaline phosphatase-conjugated secondary IgG. The blots were then developed using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium.
For intracellular localization of HDAC3 and TFII-I, COS1 cells were grown on acid-etched coverslips in 100-mm tissue culture plates for about 24 h and transfected with 0.5 g of pFLAG-HDAC plasmid as described below. Two days after transfection, the cells were washed with ice-cold PBS, fixed in 4% paraformaldehyde for 15 min, and permeabilized overnight in PBS containing 1% glycine and 0.1% Triton X-100. Cells were incubated for 30 min at room temperature with anti-FLAG coupled with fluorescein isothiocyanate (Upstate) and anti-TFII-I antibodies. After washing with PBS, the cells were incubated with diluted goat anti-rabbit IgG conjugated to tetramethyl rhodamine (Sigma). The cells were then washed extensively with PBS, and coverslips were applied with 40% glycerol before analyzing under a Carl Zeiss confocal microscope.
GST Pull-down Assays-GST and GST-HDAC3 were expressed and purified as described (3). 35 S-TFII-I was prepared using the coupled  transcription-translation rabbit reticulocyte lysate system (Promega). Equal molar quantities of GST and GST-HDAC3, both conjugated to glutathione-Sepharose beads, were incubated with radiolabeled TFII-I. Binding reactions, washing conditions, and analysis by electrophoresis and subsequent autoradiography were performed as previously described (3).
Histone Deacetylase Assay-Deacetylase activity was determined using hyperacetylated core histones purified from HeLa cells (24,25). Briefly, each immunoprecipitated sample was mixed with 5000 cpm of [ 3 H]acetate-labeled core histones. After incubation at room temperature overnight, the reactions were quenched with 1 M HCl and 0.16 M acetic acid (50 l in each sample). Released [ 3 H]acetic acid was extracted with 600 l of ethyl acetate by vortexing and centrifugation (5 min at 14,000 rpm). The ethyl acetate supernatants (250 l from each sample) were quantified by scintillation counting.
Transfection and Luciferase Assay-COS1 cells were cotransfected with plasmids directing the synthesis of various effector proteins and a luciferase reporter using the FuGene 6 transfection reagent (Roche). Each transfection contained 0.5 g each of effector and reporter DNAs, and all transfections were normalized to equal amounts of DNA with parental expression vectors. For the c-fos-luciferase assay, cells were refed with medium containing 10% serum 24 h after transfection. Twelve hours later, the cells were stimulated with 25 ng/ml recombinant human epidermal growth factor (Sigma). Epidermal growth factor treatment was not necessary for the V␤-luciferase assay. Cells were collected and luciferase activity was determined using the dual luciferase reporter assay system (Promega).

RESULTS
Physical Interaction of TFII-I and HDAC3-We previously purified an endogenous HDAC3 complex from a total extract prepared from HeLa cells using an anti-HDAC3 immunoaffinity column (9). We found that in addition to HDAC3, eight proteins (p215, p205, p195, p130, p125, p54, p52, and p35) specifically co-eluted with histone deacetylase activity. When the 215K, 205K, and 195K polypeptides were subjected to in-gel tryptic digestion followed by sequencing by microcapillary HPLC ion trap mass spectrometry, the resulting peptide sequences were shown to be derivatives of the nuclear receptor co-repressors, NCoR and SMRT (9). We have now shown that protein microsequencing of p130 has identified this HDAC3-associated polypeptide as the transcription factor TFII-I (Table I).
To ensure that the co-purification of TFII-I with anti-HDAC3 was not a result of antibody cross-reactivity, we transfected HeLa cells with plasmids expressing either HA-TFII-I or GST-TFII-I and FLAG-HDAC fusion proteins. Extracts prepared from transfected cells were then immunoprecipitated with an anti-FLAG or anti-GST antibody and immunoblotted with an anti-HA antibody. As shown in Fig. 1A, top panel, FLAG-HDAC3 interacted with HA-TFII-I and GST-TFII-I (lanes 3 and 8), whereas FLAG alone, FLAG-HDAC4, FLAG-HDAC5, and FLAG-HDAC6 did not (lanes 1, 4 -6, 9 -11). Interestingly, HA-TFII-I or GST-TFII-I also bound, to a limited extent, both FLAG-HDAC1 and FLAG-HDAC2 (lanes 2 and 7).
To support the observation that HDAC3 and TFII-I interact, we tested whether a GST-HDAC3 affinity matrix would capture TFII-I. Bacterially expressed GST-HDAC3 was bound to glutathione-Sepharose beads and incubated with 35 S-labeled TFII-I produced by in vitro translation in a reticulocyte lysate. The beads were then washed and boiled in sample buffer, and the proteins released from the beads were analyzed by electrophoresis in a SDS-polyacrylamide gel. TFII-I was captured by the GST-HDAC3 fusion protein (Fig. 1B, lane 3) but not by the GST polypeptide alone (lane 2).

FIG. 3. Mapping of the TFII-I-interacting domain on HDAC3.
A, schematic diagram of full-length (1-428) and various truncated FLAG-HDAC3 fusion proteins. Oligomerization indicates the N-terminal region of HDAC3 that is required for self-association. NES denotes nuclear export sequences. Unique represents a region within HDAC3 that has no similarity to any known proteins. For simplicity, the FLAG portions of the fusion proteins are not shown here. The ability of each FLAG-HDAC3 fusion protein to bind HA-TFII-I is indicated (ϩ or Ϫ). B, anti-FLAG immunoprecipitates (top and middle panels) and total extracts (bottom panel) prepared from transfected cells were immunoblotted with the indicated antibodies.
Colocalization studies were performed to confirm the HDAC3-TFII-I interaction in mammalian cells. COS1 cells were transiently transfected with plasmids expressing FLAG-HDAC3, fixed with paraformaldehyde, and immunostained with an anti-FLAG antibody. Consistent with our previous findings, images obtained with a confocal laser scanning system showed that HDAC3 was present in both the nucleus and cytoplasm (Fig. 2). TFII-I was regionally dispersed throughout the nuclei of COS1 cells. Importantly, numerous distinct nuclear regions in which TFII-I and HDAC3 colocalized were identified (middle panel), in agreement with the observation that the two proteins physically interact in vivo. As expected, TFII-I also colocalized with HDAC1 (left panel). In contrast, HDAC5 did not colocalize with TFII-I, acting as a negative control (right panel).
TFII-I Associates with Histone Deacetylase Activity-To determine whether the HDAC3-TFII-I interaction results in the recruitment of HDAC enzymatic activity by TFII-I, we expressed GST-TFII-I in COS1 cells, prepared immunoprecipitates from extracts using an anti-GST antibody, and assayed for HDAC activity. On average, HDAC activity was nearly 50-fold higher in immunocomplexes containing GST-TFII-I when compared with GST alone (Fig. 5). The TFII-I-associated HDAC activity was greatly reduced by Trichostatin A (TSA), a specific inhibitor of deacetylases. Although we cannot exclude the possibility that HDAC1/2 (in addition to HDAC3) may contribute to the HDAC activity associated with TFII-I, this result unequivocally confirms that TFII-I interacts with HDACs.

FIG. 5. Association of HDAC enzymatic activity with TFII-I.
Histone deacetylase activity was assayed in anti-GST immunoprecipitates prepared from COS1 cells transfected with either pEDG or pEDG-TFII-I. ϩ TSA indicates that the immunoprecipitated proteins were treated with 400 nM TSA prior to being assayed for HDAC activity. Each assay was performed in duplicate from three independent samples, and the values shown are the averages Ϯ S.D. FIG. 6. HDAC3 inhibits the activity of TFII-I. Expression plasmids for TFII-I, HDAC3, various HDAC3 mutants, c-Src, HDAC6, or Gal4-VP16 were transfected into COS1 cells together with a reporter construct as indicated. Luciferase activities are the averages Ϯ S.D. from three separate experiments. An anti-TFII-I Western blot was performed to ensure that HDAC3 did not significantly reduce the expression of TFII-I (A, right panel). HDAC3 Regulates the Transcriptional Activity of TFII-I-To determine whether HDAC3 affects the transcriptional activity of TFII-I, we examined its effect on the activity of the c-fos promoter, which is regulated by TFII-I (20). As predicted, overexpression of TFII-I significantly activated the c-fos promoter (Fig. 6A). More importantly, overexpression of HDAC3 substantially inhibited the activation of the c-fos promoter by TFII-I, consistent with the premise that HDAC3 can modulate the activity of TFII-I. The expression of TFII-I was not affected by HDAC3 (Fig. 6A, right panel), ruling out the possibility that HDAC3 down-regulates the human EF-1␣ promoter used to express TFII-I. Activation of the c-fos promoter by TFII-I was not affected by an HDAC3 mutant (H134Q/H135A/A136S) lacking deacetylase activity or by HDAC3 mutants that do not bind TFII-I (1-373, 1-313, 1-265, 1-180) (Fig. 6, A and B).
Recently, it was found that TFII-I can activate the c-fos promoter through a Src-dependent mechanism (26). To determine whether HDAC3 represses Src-dependent TFII-I activation, we expressed TFII-I and c-Src, together with HDAC3, in the presence of the c-fos reporter. As shown in Fig. 6C, activation of c-fos by TFII-I plus c-Src was completely abolished by HDAC3.
In addition to c-fos, TFII-I is required for the transcription of the naturally TATA-less but initiator-containing V␤ promoter (19,27). In separate experiments, HDAC3 inhibited the ability of TFII-I to activate the V␤ promoter (Fig. 6D). However, neither the HDAC3 mutant (H134Q/H135A/A136S) nor HDAC6 had an effect on the activation of the V␤ promoter by TFII-I. Taken together, these results unambiguously show that the transcriptional activity of TFII-I is modulated by interaction with the HDAC3 protein.
To confirm that the inhibition of activation by the TFII-I protein is not due to a general inhibitory effect of HDAC3 overexpression, we transfected cells with pGal4-VP16 and pGal4-TKLuc. As shown, overexpression of HDAC3 did not affect transcriptional activation of the Gal4-TK promoter by the Gal4-VP16 protein (Fig. 6E). Thus, our data strongly suggest that HDAC3 is not a general cytotoxic protein but rather is a specific cellular inhibitor of TFII-I. DISCUSSION The two best studied human histone deacetylases, HDAC1 and HDAC2, regulate gene expression, at least in part, by forming complexes with other cellular factors. To obtain a greater mechanistic and functional understanding of HDAC3, we previously purified and attempted to identify proteins capable of forming heterologous complexes with HDAC3 (9). Here, we identify one of the previously isolated HDAC3-binding proteins as the transcription factor, TFII-I.
Three lines of evidence confirmed that TFII-I is a genuine HDAC3-associated protein. First, coimmunoprecipitation experiments showed that TFII-I interacts with HDAC3 but not HDAC-4, -5 or -6. Second, using a GST-HDAC3 fusion protein, biochemical evidence was obtained indicating that HDAC3 and TFII-I interact in vitro. Finally, immunocytochemical methods revealed a similar subcellular distribution of HDAC3 and TFII-I in mammalian cells. Although the present experiments rely on overexpression of HDAC3 or TFII-I, we believe that they accurately reflect true in vivo interactions, in part because TFII-I copurifies with HDAC3 in an anti-HDAC3 immunoaffinity column using a whole cell extract without overexpression of either protein (9).
In addition to its interaction with HDAC3, TFII-I also interacts very weakly with HDAC1 and HDAC2 in coprecipitation experiments. Although this observation is not surprising considering that several proteins that bind HDAC1/2 also partner with HDAC3 (3, 9, 28 -31), it is conceivable that there are multiple affinity levels of interaction, with HDAC3-TFII-I forming the most stable complex. Also, at least four isoforms of TFII-I exist (15), and at this time, our protein microsequencing data do not exclude the possibility that HDAC3 binds more than one isoform of TFII-I.
Analysis of HDAC3 deletion mutants revealed that residues 373-401 of HDAC3 are required for the HDAC3-TFII-I interaction. Although the importance of this region with respect to mechanisms and functions of HDAC3 is unknown at this time, this area is known to be outside both the HDAC3 nuclear export sequence and the oligomerization domain and contributes to the deacetylase and transcriptional activity of HDAC3 (12). In future studies, it will be important to determine whether this TFII-I-interacting domain binds other cellular factors or associates with TFII-I exclusively.
An unusual feature of the TFII-I protein is the presence of six highly conserved 90-residue repeats (R1-R6), and a striking feature of these repeats is the presence of HLH motifs, which have been implicated in protein-protein interactions. Our deletion mutational studies indicate that residues 363-606 of TFII-I are essential for the HDAC3-TFII-I interaction. This region encompasses a portion of R2 and nearly the entire R3 and R4 I-repeats. Although hetero-and homo-dimerization mediated through HLH motifs is a well known phenomenon, it is presently unclear how the HLH domain of TFII-I forms a complex with HDAC3, which does not appear to contain an HLH domain. It is possible that HDAC3 and TFII-I interact via a third protein that contains both an HLH motif and an HDAC3 interaction domain. Further experiments using highly purified HDAC3 and TFII-I will resolve this issue.
The biological and functional significance of the HDAC3-TFII-I interaction remains in question. One possibility is that, in addition to its many other functions, TFII-I also recruits HDAC activity to promoters containing TFII-I binding sites. By doing so, it may help modulate the overall transcriptional activity from TFII-I-regulated promoters. If so, TFII-I then joins a rapidly growing list of sequence-specific DNA-binding transcription factors that regulate transcription by recruitment of HDAC activity. In support of this model, we show that TFII-I immunoprecipitates contain HDAC activity and that HDAC3 inhibits the activity of the c-fos and V␤ promoters. The crucial question that needs to be addressed now is whether the recruitment of HDAC3 by TFII-I is a regulated process and, if so, how to identify the signal involved. It is also conceivable that HDAC3 blocks the ability of TFII-I to activate transcription without tethering itself to the transcription complex. Finally, the possibility that TFII-I may be a substrate for HDAC3, and possibly for HDAC1/2 as well, cannot yet be ruled out. Experiments designed to explore each of these possibilities are now in progress.