The Putative Cofactor TIF1α Is a Protein Kinase That Is Hyperphosphorylated upon Interaction with Liganded Nuclear Receptors*

Ligand-induced gene activation by nuclear receptors (NRs) is a complex process requiring dissociation of corepressors and recruitment of coactivators. The putative transcriptional intermediary factor TIF1α has been previously characterized as a nuclear protein that interacts directly with the AF-2 ligand-dependent activating domain present in the ligand-binding domain of numerous steroid and nonsteroid receptors, including the estrogen (ERα) and retinoid X (RXRα) receptors. We report here that TIF1α is both a phosphoprotein and a protein kinase. TIF1α coexpressed in COS-1 cells with RXRα or ERα is phosphorylated and becomes hyperphosphorylated upon ligand treatment. This hyperphosphorylation requires the binding of TIF1α to transcriptionally active NRs since it is prevented by mutations either in the core (α-helix 12 of the ligand-binding domain) of the AF-2 activating domains of RXRα and ERα or in the NR box of TIF1α that are known to prevent TIF1α-NR interactions. Thus, TIF1α is a phosphoprotein that undergoes ligand-dependent hyperphosphorylation as a consequence of nuclear receptor binding. We further show that purified recombinant TIF1α possesses intrinsic kinase activity and that, in addition to autophosphorylation, TIF1α selectively phosphorylates the transcription factors TFIIEα, TAFII28, and TAFII55 in vitro. These latter results raise the possibility that TIF1α may act, at least in part, by phosphorylating and modifying the activity of components of the transcriptional machinery.

Ligand-induced gene activation by nuclear receptors (NRs) is a complex process requiring dissociation of corepressors and recruitment of coactivators. The putative transcriptional intermediary factor TIF1␣ has been previously characterized as a nuclear protein that interacts directly with the AF-2 ligand-dependent activating domain present in the ligand-binding domain of numerous steroid and nonsteroid receptors, including the estrogen (ER␣) and retinoid X (RXR␣) receptors. We report here that TIF1␣ is both a phosphoprotein and a protein kinase. TIF1␣ coexpressed in COS-1 cells with RXR␣ or ER␣ is phosphorylated and becomes hyperphosphorylated upon ligand treatment. This hyperphosphorylation requires the binding of TIF1␣ to transcriptionally active NRs since it is prevented by mutations either in the core (␣-helix 12 of the ligand-binding domain) of the AF-2 activating domains of RXR␣ and ER␣ or in the NR box of TIF1␣ that are known to prevent TIF1␣-NR interactions. Thus, TIF1␣ is a phosphoprotein that undergoes ligand-dependent hyperphosphorylation as a consequence of nuclear receptor binding. We further show that purified recombinant TIF1␣ possesses intrinsic kinase activity and that, in addition to autophosphorylation, TIF1␣ selectively phosphorylates the transcription factors TFIIE␣, TAF II 28, and TAF II 55 in vitro. These latter results raise the possibility that TIF1␣ may act, at least in part, by phosphorylating and modifying the activity of components of the transcriptional machinery.
Nuclear receptors (NRs) 1 represent a large family of ligand-inducible transcription factors that play numerous roles in the control of cell growth and differentiation, development, and homeostasis in response to small hydrophobic ligands, such as steroid and thyroid hormones, vitamin D, and retinoids (for reviews, see Refs. [1][2][3][4]. Like other transcription factors, NRs display a modular structure with three main conserved regions: an N-terminal A/B region harboring an autonomous activation function (AF-1); a highly conserved C region encompassing most of the DNA-binding domain; and a C-terminal E region, which, in addition to a ligand-binding domain (LBD), contains a dimerization interface and a ligand-dependent activation function (AF-2). A well conserved amphipathic ␣-helix (helix 12) referred to as the AF-2 AD core has been identified in the C-terminal part of the E region and shown to be an essential element of the ligand-inducible AF-2 function (see Refs. [1][2][3][4]. Upon ligand binding, this helix is thought to fold back over the LBD to generate transcriptionally active receptors (3,5).
For controlling gene expression at the transcription level, NRs must interact with components required for the formation of stable preinitiation complexes, either directly or indirectly via transcriptional intermediary factors (mediators). While both types of interactions have been found for a number of NRs (for reviews, see Refs. 3 and 6), direct interactions are not sufficient to account for transcriptional interference/squelching between receptors (7-9) and for the role of receptor-associated cofactors that have been recently identified on the basis of their ability to interact with receptors in an agonist-and AF-2 AD core-dependent manner. These putative mediator proteins include TIF1␣, RIP-140, SUG1, SRC-1 (and the related proteins TIF2/GRIP1, pCIP/ACTR/AIB1/RAC3), and CBP/p300 (3, 10 -13). Some of these putative coactivators (e.g. SRC-1, TIF2/ GRIP1, pCIP/ACTR/AIB1/RAC3, and CBP) contain an intrinsic activation function and can both relieve NR-induced squelching and stimulate the AF-2 activity of the NRs with which they interact (3, 10 -13). Although CBP has been recently shown to be associated with RNA polymerase II via RNA helicase A (14), the molecular mechanisms by which cofactors function have not yet been identified. That enzymatic activities may be involved is strongly suggested by the recent findings that SUG1 is a DNA helicase (15), whereas SRC-1, pCIP/ACTR/AIB1/ RAC3, and CBP/p300 possess histone acetyltransferase activities in vitro (16 -20) and p300 can, in addition, acetylate basal transcription factors such as TFIIE␤ and TFIIF (21).
TIF1␣ belongs to a new family of nuclear proteins that also includes TIF1␤ and TIF1␥ (22). 2 Based on amino acid homologies, a typical TIF1 domain structure has been ascribed to members of this family, comprising an N-terminal region that contains a RBCC (RING finger-B boxes-coiled coil) motif, a poorly conserved central region, and a C-terminal region that contains a PHD finger and a bromodomain (22,23). TIF1␣ was originally identified in a yeast genetic screen for mammalian proteins that can enhance the AF-2 activity of the retinoid X receptor (RXR) (23) and was subsequently found to interact with the AF-2 ADs of NRs in vivo as well as in vitro (23)(24)(25).
Additionally, TIF1␣ interacts with the heterochromatin-associated proteins HP1␣, MOD1 (HP1␤), and MOD2 (HP1␥) (22) and with the so-called transcriptional repression domain KRAB, which is present in about one-third of the vertebrate Krü ppel-type (C2H2) zinc finger proteins (26,27). These results raise the possibility that TIF1␣ may play a dual role in the control of transcription at the chromatin level, being involved both in repression through the formation of transcriptionally inactive heterochromatin and in hormone-dependent activation through reversion to transcriptionally active euchromatin (22,27).
In this report, we demonstrate that TIF1␣ is a phosphoprotein that undergoes hyperphosphorylation upon interaction with liganded nuclear receptors in transfected cells. Moreover, TIF1␣ possesses intrinsic kinase activity responsible for autophosphorylation and phosphorylation of the transcription factors TFIIE␣, TAF II 28, and TAF II 55 in vitro. These results suggest that hyperphosphorylation of TIF1␣ upon interaction with nuclear receptors, as well as phosphorylation of TFIIE␣, TAF II 28, and TAF II 55 by TIF1␣, might be instrumental in the control of transcription by nuclear receptors.

EXPERIMENTAL PROCEDURES
Plasmids-All plasmids used in transfection studies in COS-1 cells were pSG5 (28) derivatives already described (22). The His-TIF1␣ construct described by Le Douarin et al. (23) was modified by replacing the BamHI/XhoI fragment of the pAcSGHisNT-B baculovirus expression vector (Pharmingen) with the BglII/XhoI fragment of pET15b, thus eliminating the protein kinase A site, but preserving the reading frame and the His 6 tag.
SDS-PAGE and Western Blot Analysis-Nuclear extracts were subjected to SDS-8% PAGE and electrotransferred to nitrocellulose filters. The filters were incubated with mAb4T followed by a peroxidase-conjugated anti-mouse IgG secondary antibody and developed using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech).
Purification of Recombinant His-TIF1␣-Plated Sf9 cells infected with baculovirus encoding a His-tagged TIF1␣ cDNA in a modified pAcSGHisNTB-B vector were collected in growth medium, pelleted (10 min at 3000 rpm and 4°C), and resuspended in phosphate-buffered saline containing 30% glycerol. The cells were pelleted, homogenized in a Dounce homogenizer in NEB containing 400 mM KCl and protease inhibitor mixture, and centrifuged as described (15). The supernatant was dialyzed against load buffer (NEB containing 50 mM KCl) and loaded onto an equilibrated DEAE column, which was washed and eluted with load buffer containing 400 mM KCl. The eluate fraction was applied directly onto an equilibrated nickel chelate (Ni 2ϩ -nitrilotriacetic acid) column (QIAGEN Inc.). The column was washed extensively with NEB containing 400 mM KCl and then 400 mM NaCl, followed by wash buffer (50 mM phosphate buffer, pH 6, containing 400 mM NaCl and 50 mM imidazole), and then eluted with 500 mM imidazole in the same buffer. The collected fractions were dialyzed against dialysis buffer (50 mM Tris-HCl, pH 7.9, 20% glycerol, and 0.5 mM dithiothreitol) containing 50 mM KCl, subjected to SDS-PAGE, and either Coomassie Blue-or silver nitrate-stained or analyzed by Western blotting as described above.
Immunopurification of the Ni 2ϩ -nitrilotriacetic acid elution fractions was performed essentially as described (15). Briefly, the fraction was precleared with protein G-Sepharose for 1 h at 4°C, and the supernatant was incubated with mAb bound to protein G-Sepharose in immunoprecipitation buffer (50 mM Tris-HCl, pH 7.9, 10% glycerol, 5 mM MgCl 2 , and 0.5 mM dithiothreitol) containing 500 mM KCl. Bound protein was washed with a minimum of 50 bed volumes of immunoprecipitation buffer containing 1 M KCl before re-equilibration and peptide elution in immunoprecipitation buffer. Eluted fractions were dialyzed, analyzed by silver nitrate staining, and tested for kinase activity.
Expression and Purification of Transcription Factors-Expression in E. coli and purification of recombinant human TBP, His-TFIIA␣␤, His-TFIIA␥, His-TFIIE␣, and His-TFIIE␤ were as described (31). Histagged full-length TAF II 28 was overexpressed in E. coli and purified by Ni 2ϩ -nitrilotriacetic acid affinity chromatography (32). GST-hTAF II 55 (where "h" is human) expressed in E. coli was immobilized on a glutathione-Sepharose column and digested with thrombin in the presence of 5 mM CaCl 2 in order to release full-length human TAF II 55. GST-hTAF II 135 (amino acids 372-1083), GST-hTAF II 20, and GST-hTAF II 18 (32) were purified by immobilization on glutathione-Sepharose and elution with reduced glutathione.
In Vitro and in Vivo Phosphorylation-In vitro phosphorylation reactions were performed with 5 ng of purified His-TIF1␣ in 20 mM HEPES, pH 7.9, 15 mM MgCl 2 , 100 M ATP, and 5 Ci of [␥-32 P]ATP with or without substrate (0.5 g) as indicated. Reactions were performed at 30°C for 30 min and stopped by the addition of Laemmli buffer and boiling. The phosphorylated proteins were subjected to 8% SDS-PAGE, Coomassie Blue-stained, and autoradiographed.
For in vivo phosphorylation, transfected COS-1 cells were starved overnight in phosphate-deficient medium and then incubated in the same medium for 4 h with 250 Ci/ml [ 32 P]orthophosphate as described (33). Whole cell or nuclear extracts were subjected to immunoprecipitation and SDS-PAGE and analyzed by autoradiography and immunoblotting (33).
In-gel Kinase Assay-Immunopurified His-TIF1␣ (50 -500 ng) was subjected to 8% SDS-PAGE. After removing SDS from the gel by two washes in buffer A (50 mM Tris-HCl, pH 8, and 5 mM ␤-mercaptoethanol) containing 20% isopropyl alcohol, proteins were denatured in buffer A containing 6 M guanidine and gradually renatured by a series of five incubations over a 16-h period at room temperature in buffer A containing 0.04% Tween 40. The gel was equilibrated for 1 h in kinase buffer (40 mM HEPES, pH 7.6, 2 mM magnesium acetate, and 30 mM KCl), and the phosphorylation reaction was performed in the same buffer with 25 Ci of [␥-32 P]ATP for 1 h at room temperature. The gel was then extensively washed with 1% sodium pyrophosphate and 5% trichloroacetic acid, dried, and autoradiographed.

RESULTS AND DISCUSSION
Ligand-and Nuclear Receptor-dependent Modification of TIF1␣ by Phosphorylation-The observation that TIF1␣ interacts with liganded NRs in vitro, but does not stimulate the AF-2 activity of NRs in transfected cultured cells (23), prompted us to examine the mechanism through which TIF1␣ might act. COS-1 cells were transiently transfected with TIF1␣ in the presence or absence of RXR␣ or ER␣ and their respective ligands (10 Ϫ7 M 9C-RA and E 2 , respectively). The nuclear ex-tracts from transfected cells were subjected to SDS-PAGE and immunoblotting with a TIF1␣-specific monoclonal antibody (mAb4T). A single immunoreactive species with a molecular mass of ϳ120 kDa (Fig. 1A, lanes 1, 2, 5, and 6), which is in agreement with the expected molecular mass of the TIF1␣ protein (23), was detected in cells transfected with TIF1␣ alone in the absence or presence of ligand. Interestingly, a fraction of TIF1␣ showed a pronounced decrease in its electrophoretic mobility in ligand-treated cells cotransfected with either ER␣ (Fig. 1A, compare lanes 3 and 4) or RXR␣ (compare lanes 7 and 8). Thus, TIF1␣ could undergo a post-translational modification upon ligand-induced interaction with ER␣ or RXR␣.
Since phosphorylation often alters protein mobility during SDS-PAGE, we examined whether TIF1␣, a serine-rich protein (23), could be phosphorylated. The above COS-1 cells nuclear extracts were treated with calf intestinal alkaline phosphatase in the absence or presence of the phosphatase inhibitor sodium vanadate. Treatment with calf intestinal alkaline phosphatase alone abolished both the RXR␣/9C-RA-and ER␣/E 2 -induced mobility shift of TIF1␣ (Fig. 1, B and C, compare lanes 1-3), whereas in the presence of vanadate, the slower migrating TIF1␣ species was unaffected (Fig. 1, B and C, lane 4). Thus, the interaction of TIF1␣ with the liganded nuclear receptors RXR␣ and ER␣ appears to promote phosphorylation of TIF1␣.
To support this possibility, transfected COS-1 cells coexpressing TIF1␣ and RXR␣ were labeled with [ 32 P]orthophosphate in the absence or presence of 9C-RA, and nuclear extracts were immunoprecipitated with mAb4T. Immunoprecipitates were subjected to SDS-PAGE, autoradiography, and immunoblotting (Fig. 1D). In the absence of RXR␣ and 9C-RA,  4 and 8). A nonspecific band (ns), which was previously undetected in A and B, is visible on Western blots, probably resulting from an enrichment by immunoprecipitation and subsequent immunoblotting with the same mAb (lanes 5-8). E, COS-1 cells were cotransfected with TIF1␣ and ER␣ expression vectors in the absence (lane 1) or presence of 10 Ϫ7 M E 2 (lanes 2 and 4) and/or 10 Ϫ7 M OHT (lanes 3 and 4). Immunoblotting analysis of the corresponding nuclear extracts indicated that OHT inhibited the E 2 induction of the upper immunoreactive hyperphosphorylated TIF1␣ species (lane 4). F, COS-1 cells were transfected with TIF1␣ either alone (lanes 1 and 2) or in combination with wild-type RXR␣ (lanes 5 and 6) or RXR␣ mutated in the AF-2 AD core (RXR␣-(⌬455-467) (lanes 7 and 8), RXR␣ F455A/L456A (lanes 9 and 10), and RXR␣ M459A/L460A (lanes 11 and 12)). Cells were either untreated (lanes 1, 3, 5, 7, 9, and 11) or treated (lanes 2, 4, 6, 8, 10, and 12) with 10 Ϫ7 M 9C-RA. Nuclear extracts were prepared and immunoblotted as described for A. Note that the RXR␣ AF-2 AD core mutants did not induce the appearance of the upper immunoreactive hyperphosphorylated TIF1␣ species. Lanes 3 and 4 correspond to cells transfected with RXR␣ only. TIF1␣ was weakly phosphorylated (Fig. 1D, lanes 1, 2, 5, and  6). However, cotransfection of RXR␣ and treatment of the transfected cells with 9C-RA clearly induced hyperphosphorylation of TIF1␣, resulting in a reduction of its electrophoretic mobility (Fig. 1D, lanes 4 and 8). Altogether, these results confirm that the liganded NR-induced TIF1␣ modification occurs through phosphorylation, as it is abolished by phosphatase treatment and is accompanied by an increase in 32 P labeling.
To investigate whether the hyperphosphorylation of TIF1␣ was specific for ligands that induce AF-2 activity, similar experiments were performed with ER␣ in the presence of the anti-estrogen OHT. OHT is known to prevent ER␣ AF-2 activity in mammalian and yeast cells (34,35) and to antagonize the E 2 -dependent interaction between the LBD of ER␣ and TIF1␣ (23). In COS cells cotransfected with TIF1␣ and ER␣, OHT failed to induce any upward shift of TIF1␣ (Fig. 1E, compare  lanes 1 and 3) and antagonized the shift induced by E 2 (com- pare lanes 2 and 4). This indicates that hyperphosphorylation of TIF1␣ requires the transcriptionally active conformation of the ER␣ holo-LBD that is required for interaction with TIF1␣.
It has been previously demonstrated in vitro and in yeast (23) that the ligand-dependent interaction of NRs with TIF1␣ requires the integrity of the core of the AF-2 AD (␣-helix 12). Therefore, we investigated whether this integrity was required for TIF1␣ hyperphosphorylation. Cotransfection experiments were performed with TIF1␣ and RXR␣ either lacking the AF-2 AD core (RXR␣-(⌬455-467)) or bearing amino acid substitution in the conserved hydrophobic residues (RXR␣ F455A/L456A and RXR␣ M459A/L460A) (23,24,36). In contrast to wild-type RXR␣ (Fig. 1F, lanes 5 and 6), the RXR␣ deletion mutant (lanes 7 and 8) and the AF-2 AD core point mutants (lanes 9 -12) failed to induce the characteristic mobility shift seen upon hyperphosphorylation. Similar amino acid mutations in the AF-2 AD core of ER␣ (ER␣ M543A/L544A) that prevent the interaction with TIF1␣ (22) also eliminated the ER␣/E 2 -induced modification of TIF1␣ (Fig. 1C, compare lanes 5 and 6  with lanes 1 and 2). These results, which are in agreement with the functional interactions detected both in yeast and in vitro (23,24), demonstrate that TIF1␣ hyperphosphorylation induced by liganded RXR␣ or ER␣ requires the integrity of the AF-2 AD core (␣-helix 12 of the LBD).
The ligand-and AF-2 AD core-dependent TIF1␣-NR interaction is also known to require the TIF1␣ NR box (amino acids 726 to 735) (22), and point mutations in this NR box (TIF1␣ L730A/L731A) abolish the binding of TIF1␣ to nuclear receptors (22). To determine whether the TIF1␣ NR box was required for NR-induced modification of TIF1␣, wild-type and mutant (L730A/L731A) TIF1␣ and RXR␣ expression vectors were cotransfected with RXR␣ in COS-1 cells with and without 9C-RA. Mutation of the NR box abolished TIF1␣ hyperphosphorylation (Fig. 1B, compare lanes 2 and 5). The same results were obtained when COS cells were cotransfected with NR box mutant TIF1␣ and ER␣ and treated with E 2 (data not shown). Thus, TIF1␣ hyperphosphorylation requires an intact NR box. Altogether, these results demonstrate that the phosphorylation of TIF1␣, which occurs in the presence of liganded RXR␣ and ER␣, exhibits the same requirements as those previously shown to be indispensable for the binding of TIF1␣ to the LBD of NRs.
Autophosphorylation of Recombinant His-TIF1␣-With the aim of identifying the kinase responsible for TIF1␣ phosphorylation, we produced recombinant His-TIF1␣ using a baculovirus expression vector in infected insect Sf9 cells and purified it to near homogeneity by DEAE-Sephadex followed by nickel chelate (Ni 2ϩ -nitrilotriacetic acid) affinity chromatography (see "Experimental Procedures"). After a series of high concentration salt washes, elution, and dialysis, purified TIF1␣ was tested in an in vitro kinase assay. TIF1␣ was phosphorylated without the addition of any other factors or kinase, suggesting that TIF1␣ might have an autokinase activity ( Fig. 2A, lanes 4  and 11).
To eliminate the possibility that this kinase activity might be due to a copurifying contaminating kinase, an additional immunopurification step was performed. Purified His-TIF1␣ was immunoprecipitated using three different monoclonal antibodies (mAb1T, mAb2T, and mAb6T), each recognizing a different TIF1␣ epitope. His-TIF1␣ bound to the antibodies was washed extensively with 1 M KCl before elution with an excess of the cognate epitope peptide. The remaining supernatant ( Fig. 2A,  Sn), the peptide eluate (E) containing TIF1␣, and the eluted beads (Bd) were subjected to an in vitro kinase reaction and analyzed by SDS-PAGE and autoradiography. Both the immunoprecipitation and elution steps were efficient ( Fig. 2A, lanes  1, 2, 5, 6, 8, and 9). Strikingly, TIF1␣ eluted from the three immunoprecipitates and incubated in the kinase reaction was phosphorylated ( Fig. 2A, lanes 3, 7, and 10). Thus, after three purification steps (DEAE-Sephadex, nickel chelate, and immunoaffinity chromatography), His-TIF1␣ retained the ability to autophosphorylate.

FIG. 2. Recombinant His-TIF1␣ is an autokinase. A, recombinant
His-TIF1␣ expressed in insect Sf9 cells and purified by DEAE-Sephadex followed by nickel chelate affinity chromatography was immunoprecipitated with mAb1T (lanes 1-3), mAb2T (lanes 5-7), and mAb6T (lanes 8 -10) as indicated. TIF1␣ was specifically eluted with the respective epitope peptides and dialyzed. The initial nickel chelate affinity chromatography fraction (Ni-Ch; lanes 4 and 11), the peptide eluate (E; lanes 3, 7, and 10), the remaining eluted beads (Bd; lanes 1, 5, and 8), and the supernatant (Sn) from the immunoprecipitates (lanes 2, 6, and 9) were each tested by an in vitro kinase assay. The phosphorylated proteins were resolved by SDS-PAGE and analyzed by autoradiography. The species detected at the top of the gel corresponds to TIF1␣ that was insufficiently reduced by the dithiothreitol present in the Laemmli buffer, as it disappeared upon addition of 5 mM ␤-mercaptoethanol (data not shown). B, the silver-stained gel of purified His-TIF1␣ eluted from mAb1T immunoprecipitates was subjected to SDS-PAGE. This affinity-purified fraction consists primarily of a single protein species migrating at the expected position of TIF1␣. C, His-TIF1␣ eluted from mAb1T immunoprecipitates (ϳ50 and 500 ng) was tested for autokinase activity in an in-gel kinase assay. The 32 P autoradiographic signal was coincidental with the position of the His-TIF1␣ protein.
To eliminate the possibility that a copurifying kinase remained associated with TIF1␣ after these three purification steps, His-TIF1␣ eluted from mAb1T was resolved by SDS-PAGE and silver-stained. No contaminating polypeptides could be detected (Fig. 2B). The resolved His-TIF1␣ was tested for autokinase activity in an in-gel kinase assay. Only one protein corresponding in size to His-TIF1␣ was phosphorylated (Fig.  2C), thus ruling out the possibility that the phosphorylation of TIF1␣ could be due to a contaminating kinase. No signal was revealed in controls performed with uninfected or wild-type baculovirus-infected Sf9 cells (data not shown), thus corroborating the specificity of our results.
TIF1␣ does not possess any obvious well conserved domains typical of serine kinases. However, the Rossmann kinase ATPbinding motif GXGXXG, might be loosely conserved at amino acids 659 -669 (GSRGSSGSSSK) of the TIF1␣ amino acid sequence (23). As TIF1␣ is very rich in serine and threonine residues (23), further mutational analysis will be required to define the phosphorylated amino acids and to identify the kinase ATP-binding motif.
TIF1␣ Phosphorylates Certain Transcription Factors in Vitro-We next investigated whether TIF1␣ could phosphorylate substrates other than itself. The first proteins to be tested were those known to interact directly with TIF1␣, such as RXR␣, ER␣, HP1␣, and MOD1 (HP1␤) (22,23). None of the corresponding recombinant proteins were phosphorylated by purified TIF1␣ in an in vitro kinase assay (data not shown). To investigate whether TIF1␣ could phosphorylate ER␣ and RXR␣ in a cellular environment, COS-1 cells cotransfected with TIF1␣ and either ER␣ or RXR␣ were labeled with [ 32 P]orthophosphate and, after disruption, were immunoprecipitated with ER or RXR antibodies. The phosphorylation of ER␣ was induced by estradiol (Fig. 3A, compare lanes 5 and 6). However, this phosphorylation was not affected by overexpres-sion of either wild-type TIF1␣ (lanes 1 and 2) or TIF1␣ mutated in the NR box (lanes 3 and 4). Similarly, TIF1␣ did not affect the level of phosphorylation of RXR␣, even in the presence of 9C-RA (Fig. 3B).
Conclusion-TIF1␣ was originally identified as a transcriptional intermediary factor on the basis of its direct ligand-dependent interaction with NRs in vitro and in vivo (23,24). Results from our laboratory have provided evidence supporting the idea that TIF1␣ may control, at least in part, transcription by NRs through chromatin remodeling (22). We demonstrate here that TIF1␣ undergoes hyperphosphorylation that is dependent on direct interaction with a nuclear receptor upon binding of the cognate ligand. We also demonstrate that TIF1␣ possesses an intrinsic kinase activity as it autophosphorylates and phosphorylates transcription factors such as TFIIE␣, TAF II 28, and TAF II 55 in vitro. Thus, TIF1␣ is both a phosphoprotein and a protein kinase.
TIF1␣ contains several conserved domains, including a bromodomain that is also present in a number of transcriptional regulatory proteins (22,23,37). In most members of the bromodomain family, this domain was found in association with other conserved domains, some of which possess an enzymatic activity (helicase, ATPase, or histone acetyltransferase activity) (37). Interestingly, the 250-kDa TATA-binding proteinassociated factor of RNA polymerase II, TAF II 250, which contains a bromodomain, has been shown to possess a histone acetyltransferase activity and two kinase domains (38), whereas the Brahma proteins (BRM or SNF2␣ and BRG1 or SNF2␤) were found to be phosphoproteins (39). Thus, TIF1␣ can be added to the growing number of bromodomain-containing proteins that possess a kinase activity and are themselves phosphorylated.