Assembly of Partial TFIID Complexes in Mammalian Cells Reveals Distinct Activities Associated with Individual TATA Box-binding Protein-associated Factors*

The TATA box-binding protein (TBP) and TBP-associated factors (TAFIIs) compose the general transcription factor TFIID. The TAFII subunits mediate activated transcription by RNA polymerase II by interacting directly with site-specific transcriptional regulators. TAFIIs also participate in promoter recognition by contacting core promoter elements in the context of TFIID. To further dissect the contribution of individual TAFII subunits to mammalian TFIID function, we employed a vaccinia virus-based protein expression system to study protein-protein interactions and complex assembly. We identified the domains of human (h) TAFII130 required for TAFII-TAFII interactions and formation of a complex with hTBP, hTAFII100, and hTAFII250. Functional analysis of partial TFIID complexes formed in vivo indicated that hTAFII130 was required for transcriptional activation by Sp1 in vitro. DNase I footprinting experiments demonstrated that purified hTBP/hTAFII250 complex reconstituted with or without additional TAFIIs was significantly reduced for TATA box binding (as much as 9-fold) compared with free hTBP. By contrast, hTAFII130 stabilized binding of hTBP to the TATA box, whereas hTAFII100 had little effect. Thus, our biochemical analysis supports the notion that TAFIIs possess distinct functions to regulate the activity of TFIID.

The TATA box-binding protein (TBP) and TBP-associated factors (TAF II s) compose the general transcription factor TFIID. The TAF II subunits mediate activated transcription by RNA polymerase II by interacting directly with site-specific transcriptional regulators. TAF II s also participate in promoter recognition by contacting core promoter elements in the context of TFIID. To further dissect the contribution of individual TAF II subunits to mammalian TFIID function, we employed a vaccinia virus-based protein expression system to study protein-protein interactions and complex assembly. We identified the domains of human (h) TAF II 130 required for TAF II -TAF II interactions and formation of a complex with hTBP, hTAF II 100, and hTAF II 250. Functional analysis of partial TFIID complexes formed in vivo indicated that hTAF II 130 was required for transcriptional activation by Sp1 in vitro. DNase I footprinting experiments demonstrated that purified hTBP/hTAF II 250 complex reconstituted with or without additional TAF II s was significantly reduced for TATA box binding (as much as 9-fold) compared with free hTBP. By contrast, hTAF II 130 stabilized binding of hTBP to the TATA box, whereas hTAF II 100 had little effect. Thus, our biochemical analysis supports the notion that TAF II s possess distinct functions to regulate the activity of TFIID.
Regulation of transcription in eukaryotes requires the participation of a number of transcription factors, many of which exist as multiprotein complexes. The general transcription factor TFIID is one such complex composed of the TATA boxbinding protein (TBP) 1 and multiple TBP-associated factors (TAF II s) and is required at many gene promoters to initiate transcription by RNA polymerase II. In the context of TFIID, TAF II s have been shown to interact with 1) specific transcriptional activators to mediate activation, 2) basal transcription factors, 3) other TAF II s, and 4) specific DNA sequences, such as the downstream promoter element or gene-specific core pro-moter sequence, thereby contributing to promoter selectivity (reviewed in Refs. [1][2][3][4][5][6][7]. Recent discovery of a subset of TAF II s shared by TFIID and the chromatin remodeling complexes PCAF and SAGA (8,9), as well as TBP-free TAF II -containing complex (10) further points to multiple functions of TAF II s in regulating gene expression in eukaryotes (reviewed in Refs. [11][12][13]. Although many of the TAF II functions described above have been demonstrated in vitro, the role of TAF II s in vivo has been controversial. Genetic experiments in Drosophila point to a role for TAF II s in activated transcription during development (14); however, genetic studies in yeast have challenged the essential role of TAF II s in regulated transcription suggested from the biochemical studies (15)(16)(17)(18). More recent studies suggest that the requirement for specific TAF II s may be gene-specific. Indeed, genome-wide expression profiling in yeast demonstrates that 16% of 6000 genes in the yeast genome are affected by the inactivation of a temperature-sensitive allele of yTAF II 145, the homolog of the largest metazoan TAF II 250 (19). By contrast, the effects of mutations in the yTAF II 17 gene as well as yTAF II 60 and yTAF II 61 indicate that mutations in these TAF II genes lead to generalized reduction in RNA polymerase II transcription (20 -23). Significantly, all three yTAF II s contain histone-like folds, which may be essential for maintaining the integrity of the TFIID complex.
Inactivation of certain yeast and mammalian TAF II s results in cell cycle phenotypes; for example, cells arrest in G 1 /S when yTAF II 145 or its mammalian homolog TAF II 250 is inactivated (15,24). Examination of the promoters of genes affected by TAF II 250 inactivation has demonstrated that responsiveness to TAF II 250 resides in both upstream and core promoter sequences, suggesting that TAF II s can regulate transcription through the core promoter (25)(26)(27). By contrast, promoter mapping studies of yTAF II 145-dependent genes has led to the conclusion that yTAF II 145 functions to recognize the core promoter (28). In the case of yTAF II 17, the upstream activating sequence appears to render some genes yTAF II 17-dependent (20). These studies collectively suggest that TAF II s have distinct properties and that a subset of TAF II s may be required for appropriate expression of each gene. The molecular basis for gene-specific requirement for TAF II s remains to be determined. A gene may be dependent on a subset of TAF II s for expression because it is regulated by site-specific transcription factors that require certain TAF II s for activation or repression. On the other hand, some TAF II s have been shown to contact core promoter directly, suggesting that TAF II s can modulate the association of TFIID with promoter DNA in a manner dependent on the specific sequence of the gene promoter.
To further characterize the function of individual TAF II s and the relationship among TAF II s and TBP within the context of TFIID, we have carried out biochemical analyses utilizing the vaccinia virus protein expression system (29). We have investigated TAF II -TAF II interactions between hTAF II 250 (30, 31), hTAF II 130 (32,33), and hTAF II 100 (32,34,35) in pairwise combinations as well as in the partial complexes containing two or more subunits. The vaccinia virus protein expression system offers a number of advantages: 1) The recombinant proteins are expressed in mammalian cells and thus receive appropriate post-translational modifications; 2) the vaccinia virus/T7 RNA polymerase hybrid system permits coordinated expression of multiple genes by the T7 promoter (36); 3) transient transfection of T7 promoter-regulated genes following infection with a virus expressing the T7 RNA polymerase leads to robust expression permitting the analysis of protein-protein interactions and complex assembly; and 4) coinfection with recombinant viruses carrying T7 promoter-regulated genes facilitates large scale production of proteins in a stable complex that can be purified and used in biochemical analyses. The transient transfection protocol allowed us to test a series of mutant constructs of hTAF II 130 for TAF II -TAF II interactions and their ability to assemble into a partial complex with other subunits without the need to make recombinant viruses for each mutant. The recombinant viruses generated were used for purification of dimeric, trimeric, and tetrameric complexes.
Using these vaccinia virus-based assays, we examined the assembly among TBP and different TAF II subunits and mapped the "surfaces" required for these interactions. We show the formation of TFIID complexes containing two, three, or four subunits and their role in transcriptional activation and TATA box binding. The functional analysis of individual TAF II subunits both separately and in the context of the partial TFIID complexes has permitted us to assess the role of each subunit in TFIID function.

Transient Transfection and TFIID Complex Assembly Assay in 293T
Cells-The procedure was carried out as described (32,37) except LipofectAMINE (Life Technologies, Inc.) was used to transfect the plasmid DNA into 293T cells. HA-tagged hTAF II 130 (amino acids 1-947) and its deletion derivatives in a cytomegalovirus-based expression vector have been reported (32,38). A rabbit polyclonal antiserum against hTBP (39) was used to immunopurify the TFIID complex from crude nuclear extracts following the transfection. The products were separated by SDS-PAGE and immunoblotted with ␣-HA or ␣-hTAF II 250 monoclonal antibody (6B3; Ref. 30 Institutes of Health, and S. Shuman, Memorial Sloan-Kettering Institute) following the protocols described (40). Crude nuclear extracts were prepared from these cells 24 h after infection/transfection according to Ref. 41. The overexpressed TBP, TAF II s, and partial complexes were immunopurified in Buffer C (20 mM Hepes-KOH, pH 7.8, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 25% glycerol) containing 0.3-0.4 M KCl using the antiserum against the epitope tag, TBP, or TAF II . The immunoprecipitated products were separated by SDS-PAGE, and the interacting proteins were detected by immunoblotting using the antiserum specific for each TFIID subunit. In addition to the antibodies described above, we used the following monoclonal antibodies: ␣-T7 To detect assembly of the partial TFIID complexes in transient assays, protein complexes immunopurified with ␣-T7 antibody-conjugated agarose beads (Novagen), or ␣-hTBP antibody and protein A-Sepharose (Amer-sham Pharmacia Biotech) were separated by SDS-PAGE and stained with silver. The cDNA and antiserum for CIF150 were gifts of S. Smale (UCLA).

Generation of Recombinant Vaccinia Viruses and Large Scale Production of the Partial Complexes Containing TBP and TAF II s for in Vitro Transcription and DNase I Footprinting Assays-
The same recombinant vaccinia virus plasmids regulated by the T7 promoter were used to generate recombinant vaccinia viruses by homologous recombination as described (40). The viruses expressing Sp1 and the following TFIID subunits were made for this study: hTBP, T7-hTAF II 250, FLAG-hTAF II 130, FLAG-hTAF II 130(⌬Q2-Q4), HA-hTAF II 100, and Myc-CIF150. Coinfection of HeLa cells with the recombinant viruses and vTF7-3 permitted large scale production of TBP and TAF II proteins. Crude extracts prepared from the infected cells were used as before to immunopurify partial complexes that were used in subsequent in vitro transcription experiments. In some cases, proteins or protein complexes were eluted from the antibody using a peptide corresponding to the epitope (e.g. FLAG, HA, TBP, and Myc). Elution with the peptide corresponding to the T7 epitope was not successful under nondenaturing conditions. Immunopurified and eluted proteins were always separated by SDS-PAGE and detected by silver staining to normalize for the amount of TBP protein added to each transcription reaction.
In vitro transcription reactions using a heat-treated HeLa nuclear extract (43) were carried out as described (39). Sp1 was partially purified by wheat germ affinity chromatography from HeLa cells (44) infected with a recombinant vaccinia virus overexpressing Sp1. The Gless cassette templates TI and G 6 TI have been described (45). Fold activation was quantitated by a PhosphorImager (Molecular Dynamics).
The DNase I footprinting assays were carried out as described (46) using end-labeled fragments containing the adenovirus major late promoter sequence (Ϫ116 to ϩ61). hTBP and the partial hTBP-containing complexes were immunopurified from the recombinant vaccinia virusinfected HeLa cells with ␣-hTBP antibody SL39 and eluted with a peptide containing the epitope. The amount of purified protein added to each reaction was normalized to the amount of hTBP present in each sample preparation. Purified recombinant human TFIIA was a gift of Carla Inouye (University of California, Berkeley). hTAF II 250 was immunopurified from an insect cell lysate overexpressing HA-hTAF II 250 (gift of Edith Wang, University of Washington) and eluted with HA peptide.

The Conserved C-terminal Domain of hTAF II 130 Is Required for Stable Association with the Endogenous TFIID Complex in
Transfected 293T Cells-We previously demonstrated that recombinant hTAF II 130 (residues 1-947) transiently expressed in 293 cells can stably associate with the endogenous TBP and TAF II s to form a TFIID complex (32). To further define the domain of hTAF II 130 that is required for TFIID complex assembly, we expressed HA-tagged deletion constructs of hTAF II 130 ( Fig. 1B) in 293T cells and examined their incorporation into the TFIID complex by immunoprecipitation with a polyclonal ␣-hTBP antibody (Fig. 1A). Wild-type hTAF II 130 and ⌬CI lacking conserved region I (conserved between human and Drosophila proteins (32, 33)) associated with TFIID (lanes 1-6), whereas deletion of the C-terminal 247 amino acids (N700) that included conserved region II (CII; Ref. 32) abolished the ability of hTAF II 130 to associate with TFIID (lanes 7-9). Two additional polypeptides spanning the N-terminal (N334, lanes 10 -12) or the central (hTAF II 130N/C, lanes [13][14][15] domains of hTAF II 130 failed to be incorporated into the TFIID complex. By contrast, C301 that contained the C-terminal 301 residues of hTAF II 130 was capable of complex assembly (lanes 16 -18). The presence of endogenous TAF II s in immunopurified TFIID was detected by probing the same immunoblots with ␣-hTAF II 250 antibody and found hTAF II 250 to be present in all lanes immunoprecipitated with ␣-hTBP antibody (data not shown). The data (summarized in Fig. 1B) suggest that the C-terminal conserved domain CII of hTAF II 130 is necessary for stable incorporation into the TFIID complex.
Transient Expression of TFIID Subunits in HeLa Cells Using Vaccinia Virus Permits Rapid Analysis of TAF II -TAF II Interac-tions and Partial TFIID Complex Assembly-To further characterize specific protein-protein interactions occurring between hTAF II 130 and other subunits of TFIID, we employed a transient vaccinia virus/T7 RNA polymerase hybrid system developed by Moss and colleagues (29). We generated recombinant plasmids carrying cDNAs for hTAF II 250, hTAF II 130, hTAF II 100, and hTBP under the control of the bacteriophage T7 promoter. The coding sequence of the TAF II subunits was tagged with different epitopes at their N terminus to facilitate identification and purification: hTAF II 250 was tagged with a T7 epitope, hTAF II 130 was tagged with a FLAG epitope, and hTAF II 100 was HA-tagged. The recombinant plasmids were introduced into HeLa cells by liposome-mediated transfection and infected with a recombinant vaccinia virus (vTF7-3) carrying the gene for T7 RNA polymerase. In this way, we were able to simply and efficiently express epitope-tagged TAF II s and their derivatives individually and in combination to characterize TAF II -TAF II interactions in vivo.
We examined the ability of hTAF II 130 to interact with hTAF II 100 and hTAF II 250 using the infection/transient transfection protocol. To determine the region of hTAF II 130 necessary for interaction with hTAF II 100, HeLa cells were transfected with constructs expressing HA-hTAF II 100 in the absence or presence of various hTAF II 130 derivatives. Nuclear extracts were prepared and a monoclonal antibody directed against the N terminus of hTAF II 130 was added to immunoprecipitate hTAF II 130 and associated proteins. The presence of hTAF II 100 was determined by blotting the hTAF II 130 immunoprecipitates with a monoclonal antibody against hTAF II 100. As shown in Fig. 2A and summarized in Fig. 2D, hTAF II 100 was efficiently co-immunoprecipitated by wild-type hTAF II 130 (lanes 4 -6) as well as hTAF II 130 derivatives deleted of the central conserved region I (CI, lanes 7-9), a portion of the C-terminal conserved region II (N841, lanes 10 -12), or the central glutamine-rich regions (⌬Q2-Q4, lanes 16 -18). These interactions were confirmed in reciprocal immunoprecipitations using antibodies against the opposite subunit (data not shown). In contrast, N334, a derivative lacking both the central and C-terminal domains, showed no association with hTAF II 100 (lanes [13][14][15]. The interactions detected in these assays are most likely to be direct because endogenous TAF II s are not present in high enough concentrations to be detected in the assay (data not shown). Examination of the interaction between hTAF II 250 and hTAF II 130 derivatives showed similar pattern of interaction (summarized in Fig. 2D). These results suggest that the N-terminal portion of the conserved region II of hTAF II 130, containing residues 627-841, participates in the interaction with hTAF II 100 and hTAF II 250. The results are consistent with those shown in Fig. 1, in which a C-terminal fragment (C301) containing residues 646 -947 of hTAF II 130 was sufficient to associate with the endogenous TAF II s to form a TFIID complex.
We next examined the domain necessary for hTAF II 130 selfassociation using the infection/transfection protocol. (Fig. 2B; summarized in Fig. 2D). HeLa cells were transfected with constructs expressing FLAG-hTAF II 130 in the presence or absence of HA-hTAF II 130 (lanes 1-6). Addition of ␣-FLAG antibody to the nuclear extracts specifically immunoprecipitated HA-hTAF II 130 as detected by immunoblotting with ␣-HA antibody, demonstrating self-association of hTAF II 130. Deletion of the conserved region CI had no effect on self-association (lanes 7-15). However, deletion of the C-terminal 106 amino acids (N841, lanes 16 -21) and a construct containing only the N-terminal 334 amino acids (N334, lanes [22][23][24][25][26][27] failed to in-FIG. 1. The C-terminal domain of hTAF II 130 is required for TFIID complex assembly with endogenous TBP and TAF II s. A, 293T cells were transiently transfected with HA-tagged hTAF II 130 derivatives (shown schematically in B) using LipofectAMINE. Nuclear extracts were prepared and TFIID complex immunopurified with rabbit polyclonal ␣-hTBP antibody (lanes 2, 5, 8, 11, 14, and 17). Mock precipitation of the nuclear extracts were carried out with protein A-Sepharose alone (lanes 3, 6, 9, 12, 15, and 18). Input lanes (lanes 1, 4, 7, 10, 13, and 16) represent 5% of each reaction. Immunoblots were incubated with ␣-HA monoclonal antibody (12CA5) and detected by ECL (Amersham Pharmacia Biotech). The bands in the input lanes corresponding to the transfected HA-tagged hTAF II 130 derivatives are marked with arrows. Nonspecific cross-reacting bands are indicated by asterisks. Although the input signal corresponding to hTAF II 130N/C is weak (lane 13), the corresponding band was never visible in lane 14 even after a long exposure, suggesting that hTAF II 130N/C does not associate detectably with the endogenous TFIID. The same immunoblots were reprobed with a monoclonal antibody against hTAF II 250 to confirm the presence of an endogenous TAF II in the immunoprecipitated TFIID complex. B, schematic diagram of HA-tagged hTAF II 130 derivatives used in the transient transfection experiment. The numbering of the amino acids is according to Ref. 32. CI and CII are highly conserved regions between hTAF II 130 and dTAF II 110 (32). Glutamine-rich domains Q1-Q4 (38) are shaded. The results of the experiment shown in A are summarized to the right of each construct. teract with wild-type FLAG-hTAF II 130. Thus, C-terminal 106 residues are required for the self-association of hTAF II 130.
To test the ability of hTAF II 130 to form a partial TFIID complex in the presence of TBP and other TAF II s, we transfected each FLAG-tagged hTAF II 130 derivative along with plasmids encoding wild-type hTBP, T7-tagged hTAF II 250, and HA-tagged hTAF II 100 into HeLa cells following infection with vTF7-3 vaccinia virus. Proteins associated with the transfected FL-hTAF II 130 (see for example, ⌬Q2-Q4 derivative in Fig. 2C antibody and identified on a silver-stained gel. All hTAF II 130 derivatives that showed interactions with hTAF II 250 and hTAF II 100 were capable of forming a partial TFIID complex with hTBP, hTAF II 250, and hTAF II 100 (summarized in Fig.  2D), suggesting that the interactions seen in the binary complex are significant in the partial TFIID. This included derivative N841, which failed to dimerize with hTAF II 130, indicating that dimerization of hTAF II 130 is not required for partial TFIID complex assembly. Thus, the vaccinia virus-based transient transfection protocol has been most useful for the identi-FIG. 2. Transient transfection assays using the vaccinia virus/T7 RNA polymerase hybrid system reveal the domains of hTAF II 130 required for TAF II -TAF II interactions and complex assembly. A, hTAF II 130-hTAF II 100 interactions were studied by transfecting HA-hTAF II 100 and FLAG-hTAF II 130 derivatives (schematically shown in D), in a vaccinia virus vector under the control of the T7 promoter, into HeLa cells that had been infected with a recombinant vaccinia virus (vTF7-3) expressing the T7 RNA polymerase gene. FLAG-hTAF II 130 and its derivatives were immunoprecipitated from the nuclear extract of the transfected cells using ␣-hTAF II 130 monoclonal antibody (lanes 2, 5, 8, 11, and 14) or ␣-FLAG antibody (lane 17), separated by SDS-PAGE, and analyzed by Western immunoblotting with ␣-hTAF II 100 monoclonal antibody to detect the presence of hTAF II 100. Mock precipitations were performed with protein A-Sepharose and the cell lysate. The same blots were reprobed with ␣-FLAG antibody to confirm the presence of hTAF II 130 (data not shown). Results of the reciprocal experiments (data not shown) in which ␣-HA antibody was used to immunoprecipitate hTAF II 100 and ␣-FLAG antibody was used to detect hTAF II 130 derivatives were consistent with the results shown here. Asterisks indicate nonspecific cross-reacting bands. B, hTAF II 130 and its derivatives tagged with FLAG or HA were used to detect hTAF II 130-hTAF II 130 interactions in an assay similar to A. ␣-FLAG antibody was used to immunoprecipitate proteins associated with FLAG-hTAF II 130 (lanes 5, 11, 20, and 26) or FLAG-hTAF II 130(⌬CI) (lane 14) and ␣-HA antibody was used to detect the HA-tagged hTAF II 130 derivatives in immunoblots. The same blots were reprobed with ␣-FLAG antibody to confirm the presence of the FLAG-tagged derivatives (data not shown). Nonspecific cross-reacting bands are indicated with asterisks that include the ϳ115-kDa band visible in lanes 17 A, B, and C. The ability of an hTAF II 130 derivative to form a partial complex with hTBP, hTAF II 250, and hTAF II 100 was determined by expressing all four subunits in HeLa cells, followed by the immunopurification of the complex with ␣-T7 antibody, and detection of the subunits on a silver-stained gel. To introduce the subunits into HeLa cells, we used the vaccinia virus transient transfection protocol to transfect all subunits along with the vTF7-3 virus. In some cases HeLa cells were coinfected with the recombinant viruses for three subunits and transfected with a construct expressing the fourth subunit, for which a recombinant virus was not made. Because a recombinant virus was made for hTAF II 130(⌬Q2-Q4) construct, complex assembly assay was carried out by coinfection with all four recombinant viruses and vTF7-3 as described in C. ND, not determined. fication of domains in hTAF II 130 necessary for TAF II -TAF II interactions and TFIID complex assembly in vivo.
Assembly and Purification of TFIID Complexes from TBP and TAF II Subunits-To express large amounts of each recombinant TFIID subunit for functional studies, cDNAs encoding hTBP, T7-hTAF II 250, FLAG-hTAF II 130, and HA-hTAF II 100 were introduced into vaccinia virus by homologous recombination. These viruses were used to coinfect HeLa cells in different combinations along with the vTF7-3 virus carrying the gene for T7 RNA polymerase. Coinfection of HeLa cells with the recombinant viruses afforded a greater degree of protein overexpression as compared with the transient infection/transfection protocol.
Coinfection with two viruses permitted the isolation of dimeric complexes composed of hTAF II 250/hTBP, hTAF II 250/ hTAF II 130, hTAF II 250/hTAF II 100, and hTAF II 130/hTAF II 100 (Fig. 3A). These interactions were stable as the dimeric complexes were formed in the presence of 0.3-0.4M KCl. Interactions between hTBP and hTAF II 130 or between hTBP and hTAF II 100 were found to be weaker because they were detected in 0.1 M KCl but not in high salt concentrations (data not shown). Because T7-hTAF II 250 is limiting because of its large size and its known function as a scaffold in the TFIID complex assembly (47), we used ␣-T7 antibody to purify the partial complexes assembled in vivo. The trimeric complexes containing hTAF II 250/hTAF II 130/hTAF II 100, hTAF II 250/hTAF II 130/ hTBP, and hTAF II 250/hTAF II 100/hTBP were isolated using the ␣-T7 antibody (Fig. 3B, lanes 1-3). Finally, a TFIID complex containing four subunits was purified from the cells coexpressing hTAF II 250, hTAF II 130, hTAF II 100, and hTBP with ␣-T7 or ␣-hTBP antibodies (lanes 4 and 5). The integrity of the immunopurified partial complexes was confirmed by eluting the ␣-hTBP antibody-precipitated complexes with a peptide containing the hTBP epitope and reprecipitating the eluate with ␣-T7 antibody. The same pattern of polypeptides was observed after the second immunoprecipitation with ␣-T7 antibody, indicating that the preformed complex was stable (data not shown).
A Partial TFIID Complex Containing hTAF II 130 Supports Sp1-dependent Activation in Vitro-hTAF II 130 was previously shown to function as a coactivator for Sp1 by directly interacting with its activation domains (32,38). To examine the functional role of hTAF II 130 in the partial TFIID complex, we performed in vitro transcription reactions using a HeLa nuclear extract that had been heat-treated to inactivate the endogenous TFIID (43). Partial complexes immunopurified with the antibody against the T7 tag on hTAF II 250 (similar to those shown in Fig. 3, A and B) and normalized to the amount of hTBP present were used in the experiment. The partial complexes bound to agarose beads (Fig. 4, lanes 5-10) were incubated with a heat-treated HeLa nuclear extract and recombinant Sp1 (purified from vaccinia virus-infected HeLa cells) together with the G-less cassette template TI, containing the TATA box from the adenovirus major late promoter (AdMLP) and an initiator sequence, or G 6 TI, containing six GC boxes linked to the same core promoter (45). The fold activation by Sp1 is expressed as the ratio of transcripts derived from G 6 9 and 10), approaching the level found in non-treated nuclear extract (lanes 11 and 12). hTBP alone slightly inhibited transcription in the presence of Sp1 (lanes 1-4) as observed previously (48). Note that 2-3-fold reduction in basal transcription was seen in the presence of hTAF II 250 (compare lane 3 with lanes 5, 7, and 9), consistent with the finding that hTAF II 250 reduced binding of hTBP to the promoter in a footprinting assay (see below). In the absence of the partial complexes, heat-treated nuclear extract showed virtually no transcriptional activity with either template, and hTAF II 250 alone showed no transcriptional activity (data not shown). A complex containing hTBP/hTAF II 250/hTAF II 130 displayed an intermediate level of stimulation by Sp1, between that of the dimeric hTBP/hTAF II 250 complex and the complex containing all four subunits (data not shown). These results indicate that hTAF II 130 participates in the activation of transcription by Sp1 in vitro.
Partial Complexes Containing hTAF II 250 Reduce Binding of hTBP to the TATA Box-We next examined the DNA binding properties of the partial complexes using DNase I footprinting assay on the AdMLP. As in the previous transcription assays, we used partial complexes containing comparable amounts of hTBP. Because ␣-T7 immunopurified complexes could not be eluted with the T7 peptide, hTBP and hTBP-containing complexes were immunopurified using ␣-hTBP monoclonal antibody and eluted with a peptide containing the epitope sequence. The DNase I footprinting experiments were performed with noncoding and coding strands of AdMLP in the presence of increasing amounts of hTBP or hTBP/hTAF II 250 complex (Fig.  5A). Purified hTBP generated a defined footprint over the TATA box on AdMLP on the noncoding strand at the lowest concentration tested (lane 3). Interestingly, binding of the dimeric hTBP/hTAF II 250 complex to the TATA box was less efficient than hTBP alone and was reduced as much as 9-fold (lanes 6 -9). Similar results were obtained for the coding strand of the template DNA (lanes 10 -17). Furthermore, a partial complex containing four TFIID subunits (hTBP/hTAF II 250/ hTAF II 130/hTAF II 100) also demonstrated reduced affinity of hTBP for the TATA box similar to that observed with hTBP/ hTAF II 250 complex (Fig. 5, compare A and B). The partial complex bound to both strands of the template with at least 9-fold reduced affinity compared with the purified hTBP alone (Fig. 5B). Thus, we concluded that hTBP in a complex with hTAF II 250 binds to the AdMLP TATA box less efficiently than  3 and 4). This reflects differences in the level of expression of each component; thus, FL-hTAF II 130 is present in excess over T7-hTAF II 250, and two proteins appear more stoichiometric in a complex immunoprecipitated with ␣-T7 antibody (lane 3) but not in a complex immunoprecipitated with ␣-FLAG antibody against the more abundant FL-hTAF II 130 subunit (lane 4). For these reasons, immunoprecipitation with ␣-T7 antibody was routinely performed to isolate partial complexes for biochemical studies since T7-hTAF II 250 was the least well expressed subunit that bound all other subunits, permitting isolation of a homogeneous complex. In some experiments, ␣-hTBP monoclonal antibody SL39 was used to immunoprecipitate TBP-containing complexes. B, similar to the experiment in A, except trimeric and tetrameric complexes were isolated using the antibodies indicated below each lane of the silver-stained SDS-polyacrylamide gel.

FIG. 4. In vitro transcription reactions carried out in a heattreated HeLa nuclear extract demonstrate a requirement for hTAF II 130 in mediating activated transcription by Sp1.
Heattreated nuclear extract was incubated with hTBP immunopurified with the monoclonal antibody SL39 and eluted with an epitope-containing peptide (lanes 1 and 2) or bound to protein A-Sepharose beads (lanes 3 and 4). Partial complexes immunopurified with ␣-T7 antibody and bound to protein A-Sepharose were added to the reaction mixture in lanes 5-10. The G-less cassette template TI was used in the reactions shown in lanes 1, 3, 5, 7, 9, and 11, and G 6 TI template containing six Sp1 binding sites was used in the reactions shown in lanes 2, 4, 6, 8, 10, and 12. Recombinant Sp1 purified from HeLa cells infected with a recombinant Sp1 vaccinia virus was added to each reaction. Fold activation was calculated by quantitation of the radioactive bands with a PhosphorImager. In the control experiments, heat-treated extract showed very little activity with either template in the absence of TBP or a partial complex. hTBP by itself. The presence of hTAF II 130 or hTAF II 100 in the partial complex or the addition of CIF150/hTAF II 150 to the reaction did not alter the inhibition of hTBP binding by hTAF II 250 ( Fig. 5B and data not shown). Similar results were obtained with synthetic promoters such as G 6 TI (45), E 4 T (49), and G 5 E1bCAT (50), as well as the natural p21 promoter (data not shown). Furthermore, addition of selected activators to the footprinting reactions with templates that contained the activator binding sites did not have any detectable effect on the affinity or pattern of the footprint by the partial complex (data not shown). Interestingly, footprinting of AdMLP with a partial complex containing the four subunits plus CIF150/hTAF II 150 failed to generate an extended footprint reported for the endogenous TFIID (43) (data not shown). These results suggest that additional component(s) missing from the partial complex but present in the endogenous TFIID is likely responsible for the characteristic extended footprint observed on AdMLP.
The Effects of Different hTAF II Subunits on hTBP Binding to the TATA Box-To examine how each hTAF II subunit might singly influence hTBP binding to the TATA box, we carried out similar DNase I footprinting assays with increasing amounts of purified hTBP in the absence or presence of hTAF II 130 (Fig.  6A), hTAF II 100 (Fig. 6B), or hTAF II 250 (Fig. 6C). Increasing the amount of hTBP alone in the reaction resulted in nonspecific protection of AdMLP DNA (Fig. 6A, lanes 3-5). Interestingly, hTAF II 130 facilitated specific binding of hTBP to the TATA box at low hTBP concentrations (compare lanes 7, 11, and 15 with lane 3) and prevented nonspecific binding by hTBP at high hTBP concentrations (compare lanes 9, 13, and 17 with lane 5), suggesting that hTAF II 130 stabilizes the binding of hTBP on the promoter. hTAF II 130 alone had no DNA binding activity even at high concentrations (lanes 6, 10, and 14). Interestingly, hTAF II 130 suppressed the binding of hTBP to a non-TATA sequence (an A-rich sequence between Ϫ66 and Ϫ81) bound by hTBP alone (compare lanes 14 -17 with lanes  2-5). The effect of hTAF II 130 on hTBP binding to the TATA box was specific because hTAF II 130 had no effect on the binding of another site-specific transcription factor, the POU DNA binding domain of Oct 1, to its target DNA response element, the TAATGARAT sequence (51) (data not shown). Furthermore, unlike hTAF II 130, hTAF II 100 or bovine serum albumin did not alter the footprint of hTBP under similar experimental conditions ( Fig. 6B and data not shown). By contrast, hTAF II 250 inhibited the association of hTBP with the TATA box (Fig. 6C), consistent with the result of the previous footprinting experiment (Fig. 5).
We next tested the effects of hTAF II s on hTBP binding in the presence of recombinant hTFIIA (Fig. 7). TFIIA interacts with TBP and stabilizes its association with the promoter DNA (reviewed in Refs. 5 and 52). Because dTAF II 110 was also shown to interact with dTFIIA (53), we sought to determine the potential combined effects of hTAF II 130 and hTFIIA on hTBP binding. In the DNase I footprinting experiment, recombinant hTFIIA stabilized hTBP binding on the promoter (Fig. 7A). The presence of hTFIIA could be detected by the previously observed hypersensitive sites at Ϫ41 and Ϫ50 (54). The addition of hTAF II 130 to hTBP/hTFIIA slightly facilitated hTBP binding to the TATA box (Fig. 7B, lanes 2-9), similar to its effect in the absence of hTFIIA (Fig. 6A). By contrast, hTAF II 250 inhibited hTBP binding as before even in the presence of hTFIIA (Fig. 7B, lanes 10 -13). hTAF II 100 had a small stimulatory effect on the binding of hTBP with hTFIIA. Collectively, these experiments suggest different effects of the hTAF II subunits on DNA binding by hTBP. hTAF II 130, but not hTAF II 100, facilitated specific binding by hTBP in a manner similar to hTFIIA, whereas hTAF II 250 blocked hTBP binding.

DISCUSSION
Functional Analysis of the Partial TFIID Complexes-We have described successful use of the vaccinia virus protein expression system to characterize the interactions between subunits of TFIID in mammalian cells using wild-type and mutant forms of TAF II s and TBP. High levels of protein expression achieved by the infection/transfection protocol allowed  lanes 1-9) and coding (lanes 10 -18) strand of the AdMLP sequence. Increasing amounts (normalized to hTBP) of purified hTBP (lanes 3-5 and 11-13) or hTBP/hTAF II 250 dimeric complex (lanes 7-9 and 15-17) were added to the reaction (3-fold difference in the amount of protein added between the neighboring lanes). The GϩA lane represents the position of guanine and adenine residues in the promoter sequence. B, DNase I footprinting experiment similar to A, except a 4-protein complex containing hTBP/hTAF II 250/hTAF II 130/hTAF II -100 (lanes 7-9 and 15-17) was used in the reaction compared with hTBP alone (lanes 3-5 and 11-13).
analysis of subunit interactions and TFIID complex assembly by introducing multiple recombinant plasmids encoding TBP and TAF II s in a single transfection experiment. The method also allowed us to assess the stoichiometry of TBP and TAF II s using wild-type and mutant plasmid constructs. 2 Analysis of the binary interactions involving some of the TAF II subunits described in our study has also been reported by other researchers using different methods (34,35,55).
To obtain enough protein for functional studies, the same recombinant plasmids were used to generate recombinant viruses, which were then used to coinfect the recipient cells and isolate partial TFIID complexes containing two, three, or four subunits. We have used purified partial complexes in transcription assays to determine the contribution of each subunit to TFIID function. In vitro transcription assays using a heattreated HeLa nuclear extract demonstrated the requirement for hTAF II 130 in mediating transcriptional activation by Sp1. This is consistent with the previous demonstration that dTAF II 110, the Drosophila homolog of hTAF II 130, was required in a reconstituted TFIID complex to mediate Sp1-dependent transcription in vitro (47,56). In our study, a complex containing hTBP, hTAF II 250, hTAF II 130, and hTAF II 100 was capable of activating transcription by Sp1. A trimeric complex composed of hTBP, hTAF II 250, and hTAF II 130 was less active compared with the tetrameric complex; hTAF II 100 may stabilize the association of hTAF II 130 with other components of the complex by interacting directly with hTAF II 130 ( Fig. 2A). Consistent with this, we noticed that hTAF II 130 tends to dissociate from the complex in the absence of hTAF II 100.
The interaction of hTAF II 130 with itself was perhaps not 2 T. Furukawa and N. Tanese, manuscript in preparation.  Fig. 6 were carried out using the noncoding strand of the AdMLP and increasing amounts of purified hTBP (0.6, 2.5, and 10 ng) in the presence of purified recombinant hTFIIA (10 ng, lanes 6 -9; 40 ng, lanes 10 -13; 160 ng, lanes 14 -17). The hypersensitive sites generated by the addition of hTFIIA are indicated by the asterisks. B, the footprinting reactions were carried out in the presence of 40 ng of recombinant hTFIIA and 80 ng of hTAF II 130 (lanes 6 -9) or 20 ng of hTAF II 250 (lanes 10 -13) or 80 ng of hTAF II 100 (lanes 14 -17).
surprising as the report on the cloning of a closely related TAF II , hTAF II 105, has implicated the presence of both hTAF II 130 and hTAF II 105 in the same TFIID complex (57). Based on our deletion study, the C-terminal half of the conserved region II (CII) is involved in the dimerization of hTAF II 130 (Fig. 2, B and D). However, the dimerization domain of hTAF II 130 appears to be distinct from the domain involved in its interaction with hTAF II 250 and hTAF II 100, because the C-terminal deletion derivative N841 that does not associate with the wild-type hTAF II 130 can interact with hTAF II 250 and hTAF II 100 and form a partial complex. The CII domain of hTAF II 130 was recently reported to heterodimerize with hTAF II 20 to form a histone-like pair in TFIID (58). The minimal domain of hTAF II 130 CII required for interaction with hTAF II 20 was mapped to residues 734 -775, corresponding to the N-terminal half of CII (residues 870 -911 according to the nomenclature used in Ref. 58). It remains to be established whether hTAF II 250 and hTAF II 100 target residues in CII distinct from hTAF II 20. In the same study, the authors failed to detect homodimerization of the C-terminal 700 residues of hTAF II 130 in a yeast two-hybrid assay (58). It is possible that the dimerization of hTAF II 130 detected in our study requires the N-terminal sequence in addition to the C terminus, or alternatively, the difference may be due to proper protein folding and/or modifications occurring in mammalian cells but not in yeast. It should also be noted that we have detected an interaction between hTAF II 130 and hTAF II 105 in an in vitro protein binding assay, 3 and hTAF II 105 and dTAF II 110 have been reported to interact directly with each other (57).
The Effects of hTAF II s on hTBP Binding to Promoter DNA-Multiple footprinting experiments with hTBP and the partial complexes containing different combinations of hTAF II s used in this study produced the same result; a complex containing hTBP and hTAF II 250 with or without additional TAF II s did not alter the pattern of the footprint from that of hTBP alone, which was confined to the TATA box (Fig. 5). However, the partial complex bound with significantly reduced (as much as 9-fold) affinity to the TATA box compared with hTBP alone. The presence of hTAF II 100, hTAF II 130, and CIF150 did not alter the pattern of the footprint or affinity of hTBP in the complex for the TATA box. On a variety of synthetic and natural promoters we observed similarly reduced binding and a footprint restricted to the TATA box by the partial complex. Thus, hTBP bound the TATA box DNA with reduced affinity when complexed with hTAF II 250. Indeed we observed reduced levels of basal transcription in vitro in the presence of hTAF II 250 (Fig. 4). Our finding is consistent with previous reports in which recombinant TAF II 230 from Drosophila was shown to inhibit dTBP binding (59,60); however, we show for the first time that purified hTBP-hTAF II 250 complex as well as a reconstituted TFIID complex containing four subunits are inhibited for TATA box binding compared with free TBP. This phenomenon has been termed "autoinhibition of TFIID" (61), and a recent structural study suggests that the N-terminal region of dTAF II 230 that binds dTBP has the same structure as the unwound TATA box DNA (62). It has been proposed that DNA mimicry of protein is responsible for the reduced binding of dTBP by dTAF II 230.
The photocross-linking experiment carried out by Roeder and colleagues (63) demonstrated relatively proximal contacts between hTAF II s and specific DNA sequences within AdMLP. They showed hTAF II 130 (TAF II 135) contacting specific regions flanking the TATA box that were further enhanced in the presence of TFIIA. TFIIA has been shown to bind dTAF II 110, the Drosophila homolog of hTAF II 130 (53). We interpreted the results of their photocross-linking experiment to suggest that hTAF II 130 may play a direct role in promoter recognition in the context of TFIID. It should be noted that recent reportings of cDNAs encoding CIF150 (64)/hTAF II 150 (65) suggest, at least in one study, a tight association of hTAF II 150 with the TFIID complex and its comigration with hTAF II 130 in SDS-polyacrylamide gels (65). Thus, it is possible that the photocross-linked hTAF II 130 reported in the earlier study (63) may have been hTAF II 150/CIF150 and not hTAF II 130, or both. The histonelike dTAF II 60/dTAF II 42 were found to cross-link to the downstream promoter elements on some Drosophila promoters and play a role in promoter recognition (66). yTAF II 145, the yeast homolog of hTAF II 250, was found to contribute to promoter selectivity of selected genes whose activity was down-regulated upon inactivation of yTAF II 145 (28). In the case of hTAF II 130, its potential role in promoter recognition has not been reported. In our DNase I footprinting assays, the addition of high concentrations of hTAF II 130 to the footprinting reactions did not result in specific DNA binding; however, we did find hTAF II 130 to stabilize hTBP binding (Fig. 6A). We also detected formation of a stable hTBP-DNA complex with increasing concentrations of hTAF II 130 in electrophoretic mobility shift assays of the same samples used in the footprinting reactions (data not shown). Although we did not observe a higher order DNAprotein complex containing hTBP and hTAF II 130 in this assay, an interaction between hTBP and hTAF II 130 was detected in a protein binding assay in a buffer containing 0.1 M KCl (data not shown). Enhanced binding of hTBP to promoter DNA in the presence of hTAF II 130 is consistent with our observation that overexpression of hTAF II 130 stimulated expression of a reporter gene linked to a minimial AdMLP sequence by 9-fold or greater in transient transfection assays. Significantly, hTAF II 130-mediated stimulation was dependent on the presence of a wild-type TATA box because a reporter gene linked to the same promoter with a mutated TATA box failed to respond to overexpression of hTAF II 130. Similar findings were made using several different promoter constructs, suggesting that increased levels of hTAF II 130 affect basal transcription. 4 The effect of hTAF II 130 on hTBP binding to the TATA box was similar to that of hTFIIA. Increasing concentrations of hTAF II 130 or hTFIIA both resulted in the stable association of hTBP on promoter DNA (compare Figs. 6A and 7A). Interestingly, hTAF II 130 at high concentrations suppressed hTBP binding to a nonspecific upstream A-rich sequence (Ϫ66 to Ϫ81) (Fig. 6A, compare lanes 2-5 with lanes 14 -17). By contrast, hTFIIA did not block hTBP binding to this secondary site (Fig. 7A, compare lanes 2-5 with lanes 14 -17); however, unlike hTAF II 130, the effect of hTFIIA on hTBP footprinting was evident as indicated by the two hypersensitive sites positioned near Ϫ40 and Ϫ50 (indicated by an asterisk in Fig. 7) (54). The combined effect of hTAF II 130 and hTFIIA on hTBP binding appeared to be additive (Fig. 7B). The exact mechanism by which hTAF II 130 stabilizes hTBP binding remains to be defined. Stabilization of the TBP-TFIIA complex on the TATA box was also reported for the SV40 large T antigen (67), which was previously shown to associate with TFIID and perform a TAFlike function (68).
We found hTAF II 250 to inhibit hTBP binding both in the absence (Fig. 6C) or in the presence (Fig. 7B) of hTFIIA. Ozer et al. (69) similarly reported the inhibition of recombinant hTAF II 250 on hTBP-DNA binding. Interestingly, they found that preincubation of hTFIIA with hTBP prevented such inhibition by hTAF II 250, indicating a role for hTFIIA in overcoming repression by TFIID. Coleman et al. (70) on the other hand proposed that TFIIA regulates TBP/TFIID binding by promoting the dissociation of TBP dimers, a slow and rate-limiting step in DNA binding by TBP/TFIID. It remains to be determined whether the mechanism by which TFIIA stimulates loading of TBP/TFIID onto promoter DNA differs depending on the sequence and context of the target promoter.
The binding of TFIID to the initiator sequence has been shown to be mediated by dTAF II 150/CIF150. The recombinant human CIF150 was shown to bind to a specific sequence (71); however, we were unable to detect footprints extending downstream to the initiator region by the partial TFIID complex together with CIF150. A complex containing hTAF II 250 and dTAF II 150 with or without hTBP was successfully used to identify target sequences that matched the initiator consensus (72). Because CIF150 appears to be less tightly associated with TFIID than dTAF II 150 (64,71), it is possible that additional factors are necessary to stabilize the association of CIF150 with the components of the partial human TFIID complex to generate an extended footprint. The biochemical analysis of individual TAF subunits described in this study and by others have revealed distinct activities associated with each TAF. These findings are consistent with the genetic studies in which genespecific functions of TAFs have been observed. The assembly and characterization of partial TFIID complexes formed with defined components should reveal the functional contributions of individual TAF subunits and facilitate further characterization of the role of TFIID in transcription of specific target genes.