Isolation of Mouse TFIID and Functional Characterization of TBP and TFIID in Mediating Estrogen Receptor and Chromatin Transcription*

TFIID is a general transcription factor required for the assembly of the transcription machinery on most eukaryotic promoters transcribed by RNA polymerase II. Although the TATA-binding subunit (TBP) of TFIID is able to support core promoter and activator-dependent transcription under some circumstances, the roles of TBP-associated factors (TAFIIs) in TFIID-mediated activation remain unclear. To define the evolutionarily conserved function of TFIID and to elucidate the roles of TAFIIs in gene activation, we have cloned the mouse TAFII55 subunit of TFIID and further isolated mouse TFIID from a murine FM3A-derived cell line that constitutively expresses FLAG-tagged mouse TAFII55. Both mouse and human TFIIDs are capable of mediating transcriptional activation by Gal4 fusions containing different activation domains in a highly purified human cell-free transcription system devoid of TFIIA and Mediator. Although TAFII-independent activation by Gal4-VP16 can also be observed in this highly purified human transcription system with either mouse or yeast TBP, TAFIIs are strictly required for estrogen receptor-mediated activation independently of the core promoter sequence. In addition, TAFIIs are necessary for transcription from a preassembled chromatin template. These findings clearly demonstrate an essential role of TAFIIs as a transcriptional coactivator for estrogen receptor and in chromatin transcription.

Regulation of eukaryotic transcription by gene-specific transcription factors often requires protein cofactors, in addition to the general transcription machinery. Currently, there are three classes of general cofactors commonly thought to be essential for activator-dependent transcription. The first class is RNA polymerase II-specific TBP-associated factors (TAF II s) 1 initially defined as components of TFIID (1)(2)(3)(4)(5)(6). TAF II s are highly conserved through evolution and exhibit many properties accounting for the functional activities of TFIID. In general, TFIID is a core promoter-binding factor that has intrinsic activity to recognize the TATA box, initiator and downstream promoter elements, and initiates preinitiation complex assembly on both TATA-containing and TATA-less promoters (1,7,8). The nucleation pathway for preinitiation complex formation usually begins with TFIID binding to the core promoter region, followed either by sequential assembly of other general transcription factors (GTFs) and RNA polymerase II (pol II) or by recruitment of a preassembled pol II holoenzyme complex (9 -11). In addition to the core promoter-binding activity, TFIID has also been implicated as a general coactivator or corepressor in transducing the regulatory signals to the general transcription machinery, as exemplified by many protein-protein interactions occurring between gene-specific regulatory factors and components of TFIID (1)(2)(3)(12)(13)(14). A universal coactivator function of TFIID has recently been challenged by both in vivo yeast studies (15)(16)(17) and in vitro mammalian cell-free transcription assays (18 -21), which suggest that TAF II s are not ubiquitously required for activated transcription. This viewpoint is further substantiated by the observation that the requirement of the largest subunit of TFIID for transcription of G 1 /S cyclin genes is mainly determined by the sequence context of the core promoter region (22,23). Although mutations in some TAF II components of TFIID seem to affect activator-dependent transcription in vivo (24 -29), a direct demonstration of the coactivator function of TFIID in cell-free transcription systems devoid of many inhibitory activities that lead to the requirement for TAF II s is still lacking.
TFIID also possesses protein kinase activity that phosphorylates the RAP74 component of TFIIF (30) and the positive cofactor PC4 (31,32), suggesting that TFIID may post-translationally modulate the activities of other transcriptional components. The observations that the largest subunit of TFIID has both protein kinase and histone acetyltransferase activities (30,33) and that some TAF II components can form a histonelike structure (34,35) further indicate that TFIID, as a multiprotein complex, may play a role in transcribing chromatin templates. These diverse features have implicated TFIID as a central factor in eukaryotic transcription. However, the recent identifications of a TBP-free TAF II -containing complex as well as SPT-ADA-GCN5-acetyltransferase and p300/CBP-associated factor acetyltransferase complexes (36 -39) argue that some TAF II s can also associate with proteins other than TBP and may have distinct biological activities. Indeed, the findings that TBP-free TAF II -containing complex can functionally substitute for TFIID in mediating basal and activated transcription from both TATA-containing and TATA-less promoters (39) and that yeast TAF II 61/68 is required for SPT-ADA-GCN5acetyltransferase-dependent nucleosomal histone acetyltrans-ferase activity and transcriptional activation from chromatin templates in vitro (36) further extend the novel properties of TAF II s beyond those activities originally defined in TFIID.
The second class of general cofactors is upstream stimulatory activity-derived components found in the phosphocellulose P11, 0.85 M KCl fraction of HeLa nuclear extracts (40). Further fractionation of upstream stimulatory activity led to the identification of several positive cofactors (PCs) including PC1, PC2, PC3, and PC4, and negative cofactors (41). Recombinant human PC4 can substitute for the crude upstream stimulatory activity fraction in supporting activator function (19,20,32,(42)(43)(44)(45) and is indispensable for both TBP-and TFIID-mediated activation in vitro (19,20). Interestingly, TFIIA and TFIIH, working in conjunction with PC4, TBP, TFIIB, TFIIE, TFIIF, and pol II are able to support Gal4-VP16-mediated activation in vitro in the absence of TAF II s (20), suggesting that TFIIA and TFIIH may potentially function as coactivators in TAF IIindependent activation. The third class of general cofactors is Mediator which joins the initiation complex via its interaction with the nonphosphorylated form of pol II and is also found as a component of pol II holoenzyme (11, 46 -48). Mediator can stimulate both basal and activated transcription, as well as phosphorylation of the largest subunit of pol II by TFIIH (46). Although components of yeast Mediator are differentially required for gene activities (25, 49 -52), human Mediator appears dispensable for transcriptional activation mediated by Gal4-VP16 in a highly purified mammalian cell-free transcription system reconstituted with only recombinant GTFs, PC4, and epitope-tagged multiprotein complexes (20). Therefore, the role of mammalian Mediator (53), including potential human factors such as SMCC (54), TRAP/DRIP (54,55), CRSP (56), and NAT (57), in activator-dependent transcription remains to be elucidated.
To define the evolutionarily conserved function of TFIID and to elucidate the roles of TAF II s in gene activation, we have cloned the mouse TAF II 55 (mTAF II 55) subunit of TFIID and further isolated mouse TFIID from a murine FM3A-derived cell line that expresses FLAG-tagged mTAF II 55. The protein composition of mouse TFIID is similar to that of human TFIID, as judged by silver staining and Western blotting with both antihuman TBP and anti-human TAF II antibodies. The ability of mouse TFIID in mediating basal and activated transcription was then examined in a highly purified human cell-free transcription system reconstituted with recombinant TFIIB, TFIIE, TFIIF, PC4, Gal4 fusions containing different activation domains, and epitope-tagged TFIID, TFIIH, and pol II complexes (20). The evolutionarily conserved functions between human and mouse TFIIDs were also evidenced by the finding that both mouse and yeast TBPs can support transcriptional activation mediated by Gal4-VP16 in this highly purified human transcription system in the absence of TAF II s. To further define the role of TBP and TAF II s in activator-dependent transcription, we extend our studies toward the use of more physiologically relevant activators and transcriptional templates. We found that TAF II s become essential when the transcriptional activity of human estrogen receptor ␣ (ER) was investigated in this highly purified in vitro transcription system. The TAF II -dependent ER-mediated activation can be observed with a DNA template containing four estrogen response elements (EREs) linked to the adenovirus major late promoter, which is the same core promoter also linked to five Gal4-binding sites in a DNA template used for TAF II -independent Gal4-VP16-mediated activation, suggesting that TAF II s are collectively required as a coactivator, not as a core promoter-binding factor, for ER-mediated activation. Furthermore, by using the Drosophila S190 chromatin assembly extract and purified core histones for chromatin assembly (58) in conjunction with an in vitro transcription system comprising a preassembled TFIIDdeficient pol II holoenzyme complex and either TFIID or TBP (19), we demonstrate that TAF II s are necessary for transcription from a preassembled chromatin template. These findings clearly illustrate an essential role of TAF II s as a transcriptional coactivator for ER-mediated activation and in chromatin transcription.

EXPERIMENTAL PROCEDURES
Isolation of mTAF II 55 cDNA-A DNA fragment encoding human TAF II 55 was excised from pF:55(ORF)-11d (59) with NdeI and BamHI, and used as probe to screen a 16-day mouse embryo cDNA library (Novagen, catalog number 69640-1). The probe was made by using the multiprimer DNA labeling kit (Amersham Pharmacia Biotech) and purified by passing through a Nick column containing Sephadex G-50 (Amersham Pharmacia Biotech). Plating and screening of the cDNA library were performed as described (60) except that 25% of formamide was also included in the prehybridization and hybridization solutions. After hybridization at 42°C overnight, the filters were washed with 4X SSC plus 0.1% SDS (60) four times at room temperature and three times at 42°C (10 min for each wash), and finally one time at room temperature for 60 min. Autoradiography of filters was done at Ϫ80°C. Eleven clones that hybridized with the hTAF II 55 probe were picked up and replated for a secondary screening. Five clones were then further analyzed by DNA sequencing after excision and plasmid rescue according to the manufacturer's protocols. The sequence of a full-length mouse TAF II 55 cDNA, determined from a single phage isolate, was deposited to GenBank with accession number AF144562.
Plasmid Constructions-A DNA fragment with an N-terminal NdeI site and a C-terminal BamHI site flanking the mTAF II 55-coding region was isolated by polymerase chain reaction amplification from pEXlox-mTAF II 55 (number 28), which is the mTAF II 55 cDNA-containing plasmid rescued from a mouse cDNA library. The polymerase chain reaction fragment was then cloned into pBn-F:55 (59) and pF:TBP-11d (61), after replacing the original insert with the mTAF II 55-coding region between NdeI and BamHI sites, to generate pBn-F:mTAF II 55 and pF: mTAF II 55-11d, respectively. A bacterial expression plasmid, pF:yTBP-11d, which contains FLAG-tagged yeast TBP, was similarly made by cloning the yeast TBP insert from pGEMIID8 (obtained from A. Hoffmann and R. G. Roeder) into pF:TBP-11d between NdeI and BamHI sites.
The p4ERE⌬53 plasmid, containing 4 copies of the Xenopus vitellogenin A2 gene ERE (62) linked to the adenovirus major late core promoter (AdMLP) in front of a G-less cassette of approximately 280 nucleotides, was created by polymerase chain reaction amplification with a pair of the EcoRI site-containing primers (sense primer: 5Ј-AAGGATCCGAA-TTCAAGCTTGCATGCCTGCAG-3Ј; antisense primer: 5Ј-AAGGATCC-GAATTCATAGGACTGGGGATCCTC-3Ј) that bind to the sequences flanking the 4 EREs of pERE (63), and cloned into the EcoRI-linearized pML⌬53 plasmid (40). The pHMC 2 AT-200 plasmid with a G-less cassette of approximately 200 nucleotides preceded by the HIV-1 TATA and AdMLP initiator elements was also constructed by polymerase chain reaction with an EcoRI site-containing sense primer (5Ј-AGT-GAATTCGAGCTCGGTAC-3Ј) that anneals to the 5Ј-flanking sequence of the HIV-1 TATA box and an XbaI site-containing antisense primer (5Ј-GGATAAGATTTCTAGAGGGGAGG-3Ј) that anneals to ϳ200 nucleotides downstream of the transcription start site of pMHIVTATA (40), and cloned into pML(C 2 AT) (64) between EcoRI and XbaI sites.
To construct a baculovirus-expressing plasmid for the expression of FLAG-tagged human papillomavirus type 11 (HPV-11) E2 protein, we first subcloned the HPV-11 E2 cDNA from p6HisF:E2-11d (59) into pFLAG o (S)-7 (65) between NdeI and BamHI sites to create pF o :E2-7. The FLAG-tagged HPV-11 E2-coding region was then isolated from pF o :E2-7 between EcoRI and BamHI sites, and cloned into pVL1392 (Invitrogen) at the same enzyme-cutting sites to generate pVL-F o :E2. The baculovirus-expressing plasmid pVL-F:Sp1(FL), containing the full-length open reading frame of human Sp1, was constructed initially by cloning the NdeI (created at the initiation codon) and SmaI-digested fragment from pBS-Sp1-f1 (obtained from J.-L. Chen and R. Tjian) into pF:TBP-11d after removing the TBP insert between NdeI and Klenow enzyme-treated BamHI sites to generate pF:Sp1(FL)-11d, in which the Sp1 insert was subsequently cloned into pFLAG(AS)-7 (66) between NdeI and EcoRI sites to create pF:Sp1(FL)-7. The FLAG-tagged Sp1coding sequence was then cleaved from pF:Sp1(FL)-7 between NsiI and XbaI sites and cloned into pVL1392 between PstI and XbaI sites to produce pVL-F:Sp1(FL).
Protein Purification-FLAG-tagged mouse TFIID was purified from a mouse FM3A-derived cell line, FM55-3, that constitutively expresses the FLAG-tagged mTAF II 55 subunit of mouse TFIID following P11 chromatography and immunoaffinity purification. The FM55-3 cell line was cloned by limiting dilution from pooled G418-resistant colonies, initially obtained by retrovirus-mediated gene transfer with pBn-F:m55 as described (66). Both FM55-3 and FM3A cell lines were maintained in RPMI 1640 supplemented with either 10% fetal bovine serum for monolayer culture or 5% calf serum for suspension culture. Mouse TFIID was then purified from the P11, 0.85 M KCl fraction of FM55-3 nuclear extracts following the same procedure for purification of FLAG-tagged human TFIID (66), except that BC100 was used for the final wash before peptide elution. FLAG-tagged TFIIH and FLAG-tagged pol II were purified from the P11, 0.5 M KCl fraction of F:62-8(H) nuclear extracts and hRPB9-3 S100, respectively (31). Purification of recombinant TFIIB, TFIIE, TFIIF, PC4, and various Gal4 fusion proteins was conducted as described previously (19). Bacterially expressed FLAGtagged mouse and yeast TBPs were purified following the same procedure for purification of FLAG-tagged human TBP (61). Drosophila S190 chromatin assembly extracts and core histones were prepared according to the published protocol (58).
The TFIID-deficient pol II holoenzyme complex (f:pol II) was purified from hRPB9-3 cells that conditionally express the FLAG-tagged RPB9 subunit of human pol II as described previously (19). Further purification of f:pol II was conducted by applying 1 ml of the immunoaffinitypurified f:pol II complex (isolated from S100) onto a Mono-Q HR5/5 column (Amersham Pharmacia Biotech) at 100 mM KCl-containing BC buffer (66) with 10% glycerol. Proteins were fractionated with a 10-ml linear gradient from 0.1 to 1.2 M KCl-containing BC buffer with 10% glycerol and collected at 0.5 ml/tube at a flow rate of 0.5 ml/min. The number 11 fraction elutes at a KCl concentration around 0.65 M.
FLAG-tagged human ER (or L540Q) was purified from insect Sf9 cells infected by recombinant baculoviruses harboring the FLAG-tagged ER-coding sequence derived from pPKERf (or pPKERf(L540Q), Ref. 63). Briefly, 1 g of pPKERf (or pPKERf(L540Q)) was incubated with 0.25 g of BaculoGold linear DNA (PharMingen), 10 l of cationic liposome solution (Invitrogen), and 0.5 ml of TC-100 medium. The mixture was vortexed vigorously, left at room temperature for 15 min, and then added to a 60-mm plate containing ϳ2 ϫ 10 6 Sf9 cells. After a 4-h incubation, 1.5 ml of TC-100 was added to the plate. Incubation was continued in a 27°C humidified chamber for 4 days. The supernatant, collected after pelleting cells at 3,000 rpm for 5 min, was designated as the P 0 virus stock. 0.5 ml of P 0 was incubated with ϳ4 ϫ 10 5 Sf9 cells in a 60-mm plate containing 3 ml of TC-100. After 5 days, the supernatant (P 1 virus, 0.5 ml) was used to infect 6 ϫ 10 6 Sf9 cells in a 150-mm plate containing 25 ml of TC-100. The supernatant (P 2 virus, 5 ml), collected after 5 days of incubation, was used to infect 250 ml (ϳ0.6 ϫ 10 6 cells/ml) of Sf9 cells in suspension. Fifty ml of the final P 3 virus stock, collected after a 5-day incubation, was then used to infect 500 ml (1 ϫ 10 6 cells/ml) of Sf9 cells for protein production which was conducted for 48 h. In the interim, estradiol, if included, was added to the medium at a final concentration of 10 nM, 16 h before cells were harvested. To purify FLAG-tagged ER, the infected Sf9 cells were spun down at 1,000 rpm for 5 min, washed with cold phosphate-buffered saline, and resuspended in 30 ml of Bacterial Lysis Buffer (61). After sonication (3 ϫ 30 s) and centrifugation (14,000 rpm, 30 min), 14 ml of the supernatant was incubated with 0.1 ml of anti-FLAG M2-agarose (Sigma) at 4°C for 6 -12 h. The immobilized proteins were then washed sequentially with 10 ml of Bacterial Lysis Buffer plus 0.1% Nonidet P-40, BC100 (5 times for each wash), and finally eluted with protein buffer (BC100 for the first two elutions, and BC300 for the next three elutions) containing 0.2 mg/ml FLAG peptide and 0.03% Nonidet P-40 as described (66). ER and L540Q mainly eluted at BC300 with FLAG peptide. FLAG-tagged mouse CBP was similarly purified from Sf9 cells infected with recombinant baculoviruses harboring the FLAG-tagged CBP-coding sequence (Ref. 67, the plasmid used for transfection was obtained from A. M. Nää r and R. Tjian) as described above. Six histidine-tagged human p300 was also purified from Sf9 cells following the published protocols (63).
Western Blotting-Components of mouse TFIID were detected by Western blotting with antibodies against individual human TFIID subunits as described previously (59,65). All the anti-TFIID antibodies were used at 1000-fold dilution. For the detection of CBP, p300, and SRC-1 in the highly purified reconstituted transcription system, and RPB1, RPB2, RPB6, cyclin C, CDK8, BRG1, and RAP74 in immunoaffinity-purified pol II holoenzyme and the Mono-Q fractions, 1000-fold dilution of the anti-mouse CBP antibodies (obtained from R. G. Roeder), anti-human p300 antibodies (purchased from Santa Cruz Biotechnology, Inc.), anti-SRC-1 antibodies (obtained from M.-J. Tsai), anti-RPB1 and anti-RPB2 antibodies (obtained from N. Thompson and R. Burgess), anti-RPB6 and anti-RAP74 antibodies (obtained from R. G. Roeder), anti-cyclin C and anti-CDK8 antibodies (obtained from P. Rickert and E. Lees), and anti-BRG1 antibodies (obtained from W. Wang and G. R. Crabtree) were used in the assays.
In Vitro Transcription-A highly purified in vitro transcription system reconstituted with recombinant human TFIIB, TFIIE, TFIIF, PC4, Gal4 fusions, TBP or FLAG-tagged TFIID (f:TFIID), FLAG-tagged TFIIH, and FLAG-tagged pol II has been described (20). Unless otherwise specified, the condition for in vitro transcription is the same as described previously (20). For ER-dependent transcription, 35 ng of each pG 5 MLT (20) and p4ERE⌬53 templates were preincubated with 33.3 ng of FLAG-tagged ER, 150 ng of PC4, 10 ng of TFIIB, 10 ng of TFIIE␣, 5 ng of TFIIE␤, 28 ng of renatured TFIIF (20 ng of RAP74 and 8 ng of RAP30), 25 ng of FLAG-tagged TFIIH (purified directly from the P11, 0.5 M KCl fraction, see Ref. 31), 40 ng of pol II, and equivalent amounts (ϳ1 ng of TBP in the complex) of human or mouse TFIID at 30°C for 30 min. Transcription reactions were then initiated by providing ribonucleoside triphosphates and processed as described previously (20). For chromatin transcription, 80 ng of core histones and 6 l of the Drosophila S190 chromatin assembly extract were first incubated on ice for 30 min, and then added to the protein/DNA mixture in a final volume of 15 l containing 100 ng of pG 5 HMC 2 AT, 5 mM dithiothreitol, 30 mM creatine phosphate, 3 mM ATP, 4.1 mM MgCl 2 , and 1 g/ml creatine kinase, in the presence or absence of 3 l of TFIID-deficient pol II holoenzyme (f:pol II, Ref. 19), 50 ng of Gal4-VP16, and either 2 ng of human TBP or an equivalent amount of FLAG-tagged human TFIID. After incubation at 27°C for 4.5 h, the assembled chromatin template was added to the transcription mixture (19) containing the remaining transcriptional components, 10 units of RNase T1, and 160 ng of pHMC 2 AT-200 in a final volume of 30 l. Transcription was conducted at 30°C for 30 min and terminated by adding proteinase K and SDS, to final 0.33 mg/ml and 0.1%, respectively. Samples were then processed as described previously (20). Transcription signals were quantitated by PhosphorImager (Molecular Dynamics). II 55 and Mouse TFIID-To define the evolutionarily conserved function of TFIID, we first set out to isolate the mouse homologue of human TAF II 55 (hTAF II 55), which is an intrinsic TFIID subunit that has been shown to interact with many transcriptional activators and with other components of TFIID (59,68). Using hTAF II 55 cDNA as probe for low-stringency screening of a mouse cDNA library, we obtained a mouse cDNA clone encoding a protein with 341 amino acids (GenBank accession number AF144562). The predicted mouse TAF II 55 (mTAF II 55) protein shows 95% identity and 97% similarity to its human counterpart. 2 To isolate mouse TFIID, we linked the FLAG epitope sequence to the N terminus of the mTAF II 55-coding region, and introduced the FLAG-tagged mTAF II 55 (f:mTAF II 55) construct into a mouse mammary carcinoma FM3A cell line by retrovirus-mediated gene transfer (66). A stable mouse cell line, FM55-3, that constitutively expresses f:mTAF II 55 was isolated and further cloned by limiting dilution. Nuclear extracts, prepared from FM55-3 cells, were fractionated over a P11 phosphocellulose ion-exchange column. Mouse TFIID was then purified from the P11, 0.85 M KCl fraction by anti-FLAG immunoaffinity purification and peptide elution methods (66). As shown in Fig. 1A, anti-hTAF II 55 antibodies cross-react with FM3A mTAF II 55 which comigrates with hTAF II 55 present in a HeLa-derived 3-10 cell line that constitutively expresses FLAG-tagged human TBP (lanes 1 versus 11). The apparent molecular masses (ϳ55 kDa) of mTAF II 55 and hTAF II 55 in the gel differ significantly from their predicted sizes (ϳ40 kDa), presumably due to an unusual charge distribution (ϳ20% positive and ϳ20% negative residues) of these two proteins (59). In FM55-3 nuclear extracts, two proteins corresponding to endog-enous mTAF II 55 and exogenous f:mTAF II 55 were recognized by anti-hTAF II 55 antibodies (lane 2). Most of these two mouse proteins eluted at the P11, 0.85 M KCl fraction (lanes 3-6), similar to the fractionation profile of hTAF II 55 (59). In this purification, over 50% of mouse TFIID bound to the anti-FLAG monoclonal antibody-conjugated agarose beads was recovered in the first peptide elution (lanes 7-10).

Isolation and Characterization of mTAF
The protein composition of mouse TFIID is similar to that of human TFIID as revealed by silver staining (Fig. 1B). To examine if any of these mouse TFIID subunits correspond to those defined in human TFIID (59,66), we performed Western blotting with antibodies against various components of human TFIID (Fig. 1C). Interestingly, all these anti-human TFIID antibodies cross-react with their mouse homologues. In the cases of TAF II 250, TAF II 135, TAF II 95, and TBP, the mouse proteins are smaller than the human proteins, whereas TAF II 150/CIF150 (69, 70), TAF II 80, TAF II 55, TAF II 43,   TAF II 31, TAF II 30, TAF II 28, and TAF II 20/TAF II 15 show similar electrophoretic mobilities in both species (Fig. 1, A and C). This high degree of similarity between human and mouse TFIID TAF II s prompted us to investigate whether mouse TFIID can substitute for human TFIID and functionally interact with the human pol II transcription machinery. Thus, we performed an in vitro transcription assay using a highly purified human cell-free transcription system reconstituted with recombinant (r) human TFIIB, rTFIIE, rTFIIF, FLAG-tagged TFIIH, and FLAG-tagged pol II (20,31), in conjunction with either FLAGtagged human TFIID or FLAG-tagged mouse TFIID. For activator-dependent transcription, we also included recombinant Gal4 fusions as activators and recombinant PC4 as a coactivator. The transcription template pG 5 MLT contains five Gal4binding sites preceding the AdMLP TATA and initiator elements in front of a G-less cassette of approximately 380 nucleotides (20), whereas pML⌬53 has a shorter G-less cassette  (TAF II 250, TAF II 150, TAF II 135, TAF II 95, and TAF II 80), 10% (TAF II 55,  TAF II 43, and TBP), or 15% (TAF II 31, TAF II 30, TAF II 28, and TAF II 20/TAF II 15) SDS-polyacrylamide gels and detected by Western blotting with anti-human TFIID antibodies as described (59,69). The asterisk (*) on the left of the TAF II 55 panel points to the position of FLAG-tagged human TBP which was also recognized by the anti-hTAF II 55 antisera number 3095 (59).
(ϳ280 nucleotides) driven by the same AdMLP core promoter elements as in pG 5 MLT but lacks the activator-binding sites (40). As shown in Fig. 2, mouse TFIID is able to mediate basal transcription to a level similar to that achieved by an equivalent amount of human TFIID (lane 1 versus 8), as normalized by the content of TBP by Western blotting. A slightly lower level of basal transcription observed with mouse TFIID may reflect suboptimal interactions between mouse TFIID and the human transcription machinery. Nevertheless, mouse TFIID can still support transcriptional activation mediated by Gal4 fusions with different activation domains (Fig. 2), as observed previously with human TFIID (19,20).
TAF II -independent Activation Mediated by Mouse and Yeast TBPs-Since mouse TFIID is capable of mediating both basal and activator-dependent transcription in conjunction with the human pol II transcription machinery, we wondered whether the evolutionarily conserved function between human and mouse TFIIDs can be further extended to TAF II -independent activation (20). As shown in Fig. 3, mouse TBP can indeed support Gal4-VP16-mediated activation in the human pol II transcription machinery in the absence of TAF II s (lanes 1-4 versus [5][6][7][8]. As in the case of human TBP, TAF II -independent activation mediated by mouse TBP can only occur in the presence of PC4 (lanes 2 versus 4, and lanes 6 versus 8). This result prompted us to explore an interesting possibility that yeast TBP may function in a highly purified human transcription system to mediate TAF II -independent activation by Gal4-VP16. Intriguingly, yeast TBP can indeed support Gal4-VP16mediated activation without TAF II s to a comparable level as achieved by human TBP (lanes 1-4 versus 9 -12). This functional equivalence between human and yeast TBPs in mediating TAF II -independent activation is consistent with previous observations that yeast TBP is able to support Gal4-VP16mediated activation in heat-treated HeLa nuclear extracts (71) and human TBP can in general substitute for yeast TBP in supporting pol II transcription in vivo (72). Since TBPs isolated from different species differ significantly at their N-terminal regions, both in size and sequence, but are highly conserved (at least 80% identical) in the C-terminal sequences (73), our data suggest that the C-terminal region of TBP may play an important role in TAF II -independent activation.
TAF II s Are Essential for Estrogen Receptor-mediated Activation-To further define whether TAF II s are indeed necessary for activator function in our cell-free transcription system where TAF II s appear dispensable for transcriptional activation mediated by various Gal4 fusions, we began to test the transcriptional activity of a natural activator in our highly purified in vitro transcription system. Since estrogen receptor ␣ (ER) has been shown to functionally interact with several components of TFIID, including TBP (74,75), TAF II 30 (76), and TAF II 28 (77), we speculated that ER-mediated activation may show a greater dependence on TAF II s as also reflected by previous in vivo transfection assays (76,77). To establish an ER-dependent in vitro transcription system, we first expressed the full-length FLAG-tagged human ER protein in insect cells using a baculovirus expression system (63) and purified ER to near homogeneity by immunoaffinity purification and peptide elution methods (Fig. 4A). The identity of the purified FLAGtagged ER protein was confirmed by Western blotting with both anti-FLAG and anti-ER antibodies. 2 A DNA template, p4ERE⌬53, which contains 4 EREs linked to the AdMLP TATA and initiator elements preceding a G-less cassette of approximately 280 nucleotides, was also constructed and used for the transcriptional assay (Fig. 4B). When tested in our highly purified in vitro transcription system reconstituted with rT-FIIB, rTFIIE, rTFIIF, FLAG-tagged TFIIH, FLAG-tagged pol II, and either FLAG-tagged human TFIID or FLAG-tagged mouse TFIID, we could clearly detect ER-mediated activation on p4ERE⌬53, but not on an internal control template (pG 5 MLT) containing 5 Gal4-binding sites linked to the same AdMLP core promoter elements (Fig. 4C, lanes 1-3 and 5-7). In contrast, Gal4-VP16 only activated transcription from pG 5 MLT but not p4ERE⌬53 (Fig. 4C). This is the first documentation that ER-mediated activation can occur in a highly purified cell-free transcription system in a sequence-specific manner. Interestingly, ER-mediated activation was completely abolished when TBP was used in place of TFIID (Fig. 4D, lanes 2  versus 3, and lanes 6 versus 7). Under the same experimental condition, TAF II -independent activation could still be observed by Gal4-VP16 with either mouse or human TBP (Fig. 4D). These results indicate that TAF II s, although dispensable for Gal4-VP16-mediated activation, are critical for ER-mediated activation. Furthermore, the role of TAF II s is likely to serve as a transcriptional coactivator, not a core promoter-binding factor, as the requirement for TAF II s in ER-mediated activation is not dictated by specific core promoter elements (Fig. 4B).
Since ER contains an N-terminal ligand-independent activation domain (AF1) and a C-terminal ligand-dependent activation domain (AF2) that can functionally regulate transcription in vivo (78), we wondered whether ER-mediated activation observed in our highly purified transcription system is liganddependent. To address this important issue, we conducted a parallel purification of ER, in the absence or presence of 17␤estradiol (see "Experimental Procedures"). An ER mutant, L540Q, that impairs the AF2 activation function but retains the ligand-binding activity (63, 79) was also concurrently pu- rified in the presence of 17␤-estradiol (Fig. 5A). As shown in Fig. 5B, ER purified in the absence of ligand activates transcription to a level similar to that observed with the one purified in the presence of ligand (Fig. 5B, lanes 1-8). Interestingly, the L540Q mutant, although its AF2 activation function is impaired, still activates transcription through the ERE-containing template as efficiently as the wild-type protein (Fig. 5B,  lanes 3-11). This experiment, irrespective of whether human or mouse TFIID was used, 2 clearly demonstrates that ER-mediated activation in our highly purified in vitro transcription system occurs in a ligand-independent manner. Further evidence was supported by the observation that the addition of 1 M anti-estrogen, trans-hydroxytamoxifen, did not affect ERmediated activation in our reconstituted transcription system. 2 Our result is consistent with a previous report that the insect cellular lysate containing mouse ER can activate transcription in a ligand-independent manner through the ERE-containing template in a cell-free transcription assay performed with HeLa nuclear extracts (80). This ER-mediated activation in our highly purified transcription system was not caused by nonspecific insect proteins copurified with ER, as the Sp1 and human papillomavirus type 11 (HPV-11) E2 proteins, also purified from insect cells using the same procedure, could not activate transcription through the ER-binding sites (Fig. 5B, lanes 13   and 14). The activation was also not facilitated by CBP, p300, or SRC-1, since these nuclear hormone receptor coactivators were not detected in our in vitro transcription system (Fig. 5C). Our data therefore suggest that ER has an intrinsic ability to activate transcription in vitro in a ligand-and nuclear hormone receptor coactivator-independent manner, presumably through functional interactions with components of the general transcription machinery (see "Discussion").
TAF II s Are Necessary for Chromatin Transcription-Since TAF II s are essential for the function of a natural activator like ER, we wondered whether transcription from a more physiologically relevant DNA template also requires TAF II s. The observations that the largest subunit of TFIID has both histone acetyltransferase and protein kinase activities (30,33) and that some components of TFIID also form a histone-like structure (34,35) suggest that TAF II s may play a critical role in chromatin transcription. To explore this, we first employed the Drosophila S190 chromatin assembly extract and purified core histones for chromatin assembly (58). However, no chromatin transcription could be detected in our highly purified cell-free transcription system where TAF II -independent and TAF II -dependent activation could occur on deproteinized DNA templates. 2 Therefore, we resorted to the use of an in vitro transcription system containing a preassembled TFIID-deficient FIG. 3. TAF II -independent activation mediated by TBPs isolated from various species. In vitro transcription was conducted with recombinant human TFIIB, TFIIE, TFIIF, FLAG-tagged TFIIH, FLAG-tagged pol II, and human (h), mouse (m), or yeast (y) TBP, in the presence (ϩ) or absence (Ϫ) of recombinant PC4 and Gal4-VP16, as indicated. Fold activation is the same as defined in the legend to Fig. 2.

FIG. 4. TAF II s are required for ER-mediated activation in vitro.
A, purified FLAG-tagged human ER. FLAG-tagged ER (ER) was purified from insect cells infected with recombinant baculoviruses in the presence of 17␤-estradiol as described under "Experimental Procedures," and visualized by Coomassie Blue staining. The molecular masses (in kDa) of prestained protein size markers (Life Technologies, Inc.) are indicated on the left. The arrow points to the position of FLAG-tagged human ER. B, DNA templates used for transcriptional assays. Both pG 5 MLT and p4ERE⌬53 contain the TATA (TATA) and initiator (Inr) elements derived from the adenovirus major late promoter (MLP) in front of a G-less cassette of approximately 380 or 280 nucleotides, and preceded by 5 Gal4-binding sites or 4 EREs, respectively. C, both human and mouse TFIIDs can support ER-mediated activation. In vitro transcription was conducted with recombinant human TFIIB, TFIIE, TFIIF, FLAG-tagged TFIIH, FLAG-tagged pol II, and either human (h) or mouse (m) TFIID, in the presence (ϩ) or absence (Ϫ) of recombinant PC4 and ER or Gal4-VP16, as indicated. D, TAF II -independent activation can only be observed by Gal4-VP16 but not by ER. In vitro transcription was performed as described in C, except either human (h) or mouse (m) TBP was substituted for TFIID. Fold activation in each set of reaction conditions is defined as the signal intensity quantitated by PhosphorImager (Molecular Dynamics) from either pG 5 MLT or p4ERE⌬53 relative to that from the same DNA template performed in the absence of an activator and PC4 (i.e. the first lane of each reaction set).
pol II holoenzyme complex (f:pol II, Ref. 19) and a TATAbinding factor provided by either TFIID or TBP. The utilization of pol II holoenzyme offers a unique opportunity for transcribing chromatin templates, since our f:pol II contains not only a subset of GTFs as described previously (19) but also components of human SRBs (suppressors of RNA polymerase B mutations) such as cyclin C and CDK8 (81), and BRG1 (82,83), a component of human SWI/SNF chromatin remodeling complex (Fig. 6A, bottom panels). Moreover, f:pol II, isolated from either nuclear extracts or S100, contains only the non-phosphorylated (IIa) form of RPB1 (Fig. 6A, top panels, lanes 1-8) and other components of pol II (Fig. 6A, RPB2 and RPB6 panels), 2 suggesting that f:pol II represents an initiation form of pol II holoenzyme (46 -48, 81, 84). The presence of BRG1, cyclin C, and CDK8 in f:pol II was substantiated by further fractionation of the immunoaffinity-purified f:pol II by Mono-Q ion-exchange chromatography. All these proteins cofractionated with a subset of GTFs found in f:pol II as well as the transcriptional activity of f:pol II (Fig. 6B). 2 Since f:pol II contains all GTFs, except TFIID, and components of SRBs and SWI/SNF complexes, it was used in conjunction with either TBP or TFIID to address the role of TAF II s in chromatin transcription. Furthermore, by performing an orderof-addition experiment in this pol II holoenzyme transcription system, we will be able to define the role of TAF II s during and after chromatin assembly. In this assay, transcriptional components (Gal4-VP16, f:pol II, and a TATA-binding factor) were incubated, individually or together, either during or after chromatin assembly. Inclusion of transcriptional components during chromatin assembly may prevent nucleosome formation on the promoter region, thereby alleviating nucleosome-mediated repression. Transcription was then initiated by adding the remaining transcriptional components and ribonucleoside triphosphates (see the outline in Fig. 7). The DNA template pG 5 HMC 2 AT containing 5 Gal4-binding sites (66) was used for chromatin assembly. A similar plasmid that contains the same core promoter elements as found in pG 5 HMC 2 AT but without Gal4-binding sites was constructed on a shorter G-less cassette (ϳ200 nucleotides). This plasmid, pHMC 2 AT-200, was added after chromatin assembly and used as an internal control for naked DNA transcription. As shown in Fig. 7, in the absence of Gal4-VP16, preincubation of the transcriptional components during chromatin assembly indeed alleviates nucleosome repression (lanes 1 versus 4, and lanes 5 versus 8). This derepression is more significant when TAF II s are present during chromatin assembly (lanes 4 versus 8). Although preincubation of either TFIID or TBP, but not f:pol II, during chromatin assembly also prevents nucleosome repression, the presence of f:pol II during chromatin assembly further enhances the ability of the TATA-binding factor to compete with nucleosomes for binding to the promoter region (lanes 2 versus 4, and lanes 6 versus 8).
The alleviation of nucleosome repression by the transcriptional components is generally enhanced when Gal4-VP16 is also included during chromatin assembly (lanes 1-4 versus 10 -13,  lanes 5-8 versus 15-18). Without a transcriptional activator, f:pol II and TFIID (or TBP) cannot transcribe a preassembled chromatin template (lanes 1 and 5). Interestingly, in the presence of Gal4-VP16, only TFIID but not TBP can work in conjunction with f:pol II in transcribing a preassembled chromatin template (lanes 9 versus 14), indicating that both TAF II s and a FIG. 5. ER-mediated activation is independent of ligands and nuclear hormone receptor coactivators in a highly purified reconstituted transcription system. A, Coomassie Blue-stained gel of purified FLAG-tagged wild-type and mutant ER proteins. The FLAGtagged wild-type (WT) and mutant (L540Q) ER proteins were purified from insect cells infected with individual recombinant baculoviruses, in the absence (Ϫ) or presence (ϩ) of 17␤-estradiol (estradiol), as described under "Experimental Procedures." The charcoal/dextran-treated fetal bovine serum was used in the culture medium for the purification of ER in the absence of ligand. The molecular masses (in kDa) of prestained protein size markers (Life Technologies, Inc.) are indicated on the left. The arrow points to the position of FLAG-tagged human ER. Several bands around the 68-kDa protein marker are nonspecific polypeptides also found in purified Sp1 and HPV-11 E2 proteins (see footnote 2). A band slightly above the 97.4-kDa marker, which does not contribute to ER-mediated activation, is a unique polypeptide associated with ER presumably in a ligandsensitive or AF2 conformation-specific manner, whose identity remains to be further characterized. B, ligand-independent activation mediated by wild-type and mutated ER proteins. In vitro transcription was conducted with recombinant human TFIIB, TFIIE, TFIIF, FLAG-tagged TFIIH, FLAG-tagged pol II, and FLAG-tagged human TFIID, in the presence (ϩ) or absence (Ϫ) of recombinant PC4 and either increasing amounts of ER (16.6, 33.3, and 66.6 ng), Gal4-VP16 (50 ng), Sp1 (33.3 ng), or human papillomavirus type 11 E2 (100 ng), as indicated. Fold activation is the same as defined in the legend to Fig. 4. C, nuclear hormone receptor coactivators are not present in the highly purified reconstituted transcription system. Quantitative Western blotting was performed by loading differential amounts of purified FLAG-tagged CBP and 6 histidine-tagged p300 (rProtein, in ng) or HeLa nuclear extract (NE, in g), along with immunoaffinity-purified TFIID, TFIIH, pol II, wild-type ER (purified in the presence of 17␤-estradiol), recombinant TFIIB/TFIIE/TFIIF (rB/E/F), and mouse TFIID. The amounts of loaded protein factors are either the same (1x) or 4-fold (4x) of the concentrations used in the transcriptional assays. Antibodies used for each filter are indicated on the left. transcriptional activator are essential for chromatin transcription. Consistent with previous results (85), an equivalent level of chromatin transcription could be observed by including Gal4-VP16 either during or after chromatin assembly (lanes 9 versus 10). Throughout the experiment, the naked DNA template (pHMC 2 AT-200) was actively transcribed and showed little variations, suggesting that repression of chromatin transcription is not due to inactivation of the transcriptional components during chromatin assembly (lanes 1-18). Furthermore, this experiment clearly demonstrates that although TAF II s are crit-ical for chromatin transcription, they are indeed dispensable for basal transcription from naked DNA templates. This study not only stipulates the concept that nucleosome formation indeed prevents the assembly of the general transcription machinery on the promoter region and thereby inhibits transcription, but also demonstrates an important role of TAF II s in chromatin transcription.

DISCUSSION
The Evolutionarily Conserved Function of TFIID-In yeast, TBP only weakly associates with TAF II s to form a multisubunit TFIID complex (86,87). In contrast, in Drosophila and mammalian cells, TFIID is a tightly associated complex which is resistant to high salts and mild chaotropic denaturants (88,89). Therefore, the requirement of TAF II s for activator-dependent transcription may vary between lower and higher eukaryotes. The isolation of a cell type-specific TAF II in humans (90) further indicates that some TAF II s may perform specialized functions uniquely required in higher eukaryotes. Since no homologues of human TAF II 55 were found in the databases when it was first reported (59), we wondered if TAF II 55 represents a human-specific TFIID subunit or it may also be found in other organisms. The cloning of mouse TAF II 55 (GenBank accession number AF144562) and the identification of Saccharomyces cerevisiae and Caenorhabditis elegans homologues in the recent database (accession numbers for ScTAF II 55 and CeTAF II 55 in the EMBL data library are Z49939 and Z67755, respectively) indicate that TAF II 55 is not unique to human cells. Instead, TAF II 55 is an evolutionarily conserved TFIID subunit which has been identified thus far in humans, mice, nematode, and yeast. Interestingly, mTAF II 55 and hTAF II 55 are both found in several chromatographic fractions (Ref. 59 and Fig. 1A), suggesting that TAF II 55 may be present in multiple complexes similar to what is observed for the histone-like TAF II s. However, this possibility remains to be further investigated.
Purified mouse TFIID has a protein composition similar to that of previously characterized human TFIID (Fig. 1B). Although additional polypeptides are also found in purified mouse TFIID, we are not certain whether these proteins are indeed TAF II s uniquely found in mouse TFIID or represent impurities copurified with the mouse sample. This issue remains to be further investigated. Nevertheless, since all antihuman TAF II antibodies cross-react with mouse TAF II s (Fig.  1C), it suggests that mouse and human TFIIDs may play a comparable role in the transcriptional process. Indeed, mouse TFIID is able to functionally interact with components of the human pol II transcription machinery to mediate both basal and activator-dependent transcription (Fig. 2). Interestingly, a chimeric TFIID comprising human TBP and mouse TAF II s, which was isolated from an FM3A-derived cell line (FM-hTBP-2) that constitutively expresses FLAG-tagged human TBP, is capable of mediating transcription at a level similar to that observed by using mouse or human TFIID. 3 This suggests that, although the apparent molecular weight of mouse TBP is relatively smaller than that of human TBP due to a shorter stretch of the glutamine residues in the mouse protein (Ref. 91, also see Fig. 1C), human TBP can still interact with mouse TAF II s to assemble into a functional TFIID complex. Thus, mammalian TFIIDs are highly conserved as evidenced by sequence comparison (91), 2 antibody cross-reactivity (Fig. 1C), and transcriptional assays (Fig. 2). The evolutionarily conserved function of TFIID is further supported by previous findings that the C-terminal domain of human TBP is sufficient to FIG. 6. A TFIID-deficient human RNA polymerase II holoenzyme complex (f:pol II) also contains components of SRBs and SWI/SNF chromatin remodeling factor. A, components of pol II (RPB1, RPB2, and RPB6), SRBs (cyclin C and CDK8), and SWI/SNF (BRG1) are present in the immunoaffinity-purified FLAG-tagged pol II holoenzyme complex. Nuclear extracts (N.E.) and cytoplasmic fractions (S100) isolated from hRPB9-3 cells (19) were incubated with anti-FLAG M2-agarose (Sigma). Bound proteins were step-eluted with a synthetic FLAG peptide (Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys). The presence of RPB1 in the input (In), flow-through (FT) and the first (1 o ) elution was then detected by Western blotting with anti-RPB1 CTD (C-terminal domain) antibodies (top panels, lanes [1][2][3][4][5][6][7][8]. Core-pol II (lane 8 in the top panel), purified from the hRPB9-3 nuclear pellet (31), was used for comparison to indicate the positions of the IIo and IIa forms of RPB1. Detection of other pol II subunits, SRBs, and BRG1 in the f:pol II complex (bottom panels) was similarly performed by Western blotting with anti-RPB2, -RPB6, -cyclin C, -CDK8, and -BRG1 antibodies, respectively, using HeLa nuclear extracts as a positive (ϩ) control. S100 from hRPB9-3 cells was used for immunoaffinity purification of the f:pol II complex as described previously (19). In parallel, S100 from HeLa cells was also used for immunoaffinity purification and served as a negative (Ϫ) control. B, BRG1, cyclin C, and CDK8 cofractionate with general transcription factors in f:pol II through Mono-Q ion-exchange chromatography. The immunoaffinity-purified f:pol II complex was used as the input (In) and separated by a Mono-Q HR5/5 column (Amersham Pharmacia Biotech) as described under "Experimental Procedures." The fractions used for Western blotting analysis are indicated above the lanes. Western blotting was conducted with different antiprotein antibodies as indicated on the left of each panel.
form a functional TFIID complex (92) and human TBP is able to interact with yeast TAF II s and vice versa (86,92).
TAF II -independent and TAF II -dependent Activation-Although TFIID clearly plays a central role in eukaryotic transcription, the mechanisms by which TAF II s are involved in gene regulation remain unclear. In many cases, interaction studies and functional characterizations performed on individually isolated TAF II s have not yet been demonstrated in the entire TFIID complex, which is normally the functional entity in the living cell. It is likely that protein domains characterized to be important for protein-protein interactions on individual subunits are in fact masked in the protein complex. Therefore, it is important to functionally characterize the activities of the entire TFIID complex. When compared with the properties of TBP in parallel, the role of TAF II s in the transcriptional process can be deduced by the functional differences between TFIID and TBP. It is then possible to define whether TAF II s collectively function as a core promoter-binding factor, an antirepressor, or mainly as a transcriptional coactivator (or corepressor) for individual gene expression.
Using a highly purified human cell-free transcription system reconstituted with only recombinant proteins (TFIIB, TFIIE, TFIIF, PC4, and Gal4-VP16) and nearly homogeneous preparations of epitope-tagged multiprotein complexes (TFIID, TFIIH, and pol II), we have demonstrated that TAF II s are dispensable for Gal4-VP16-mediated activation (Ref. 20, and Figs. 3 and 4D). This TAF II -independent activation mediated by TBP in the presence of PC4 is not potentiated by Mediator, since no components of human Mediator are present in our in vitro transcription system (20). In contrast, we have illustrated that TFIIA and TFIIH can support Gal4-VP16-mediated activation in the absence of TAF II s (20). Thus, the requirement of TAF II s in previous transcription systems reconstituted with relatively impure protein factors such as upstream stimulatory activity and E/F/H fractions appears to antagonize the inhibi-tory activities commonly present in the crude systems. The antirepressor function of TAF II s was also used to overcome PC4-mediated inhibition of basal transcription when a transcriptional activator is not present (19,20,31,32). It is likely that the protein kinase activity of TFIID leads to PC4 inactivation as a repressor (31,32), although this hypothesis remains to be tested.
Whether TAF II s mainly function as an antirepressor or a genuine transcriptional coactivator could not be easily distinguished in vivo, nor in previous in vitro transcription systems due to the presence of many potential coactivators and corepressors. To address this important issue, we begin to investigate the requirement of TAF II s for activator function in our highly purified cell-free transcription system devoid of many cofactor activities. Under the same condition where TAF IIindependent activation mediated by Gal4-VP16 can be observed by different species of TBPs in the presence of PC4 (Figs. 3 and 4D), we are able to demonstrate that TAF II s are essential for ER-mediated activation (Fig. 4C). This TAF II -dependent activation by ER clearly illustrates a collective role of TAF II s as a transcriptional coactivator, as no stimulation of transcription could be detected in the absence of TAF II s. Furthermore, since TAF II -dependent ER-mediated activation can be observed with a DNA template containing estrogen response elements linked to the adenovirus major late promoter, which is the same core promoter also linked to 5 Gal4-binding sites used for TAF II -independent Gal4-VP16-mediated activation, this result suggests that TAF II s do not act as a core promoterbinding factor in ER-mediated activation. Obviously, the requirement of TAF II s in ER-mediated activation is activator-dependent and also relies on the presence of cognate activatorbinding sites. This analysis demonstrates that both TAF IIindependent and TAF II -dependent activation can be recapitulated in vitro in a highly purified reconstituted transcription system, thereby providing an invaluable assay to FIG. 7. TAF II s are necessary for chromatin transcription. In vitro chromatin transcription was performed with TFIID-deficient pol II holoenzyme (f:pol II) and a human TATA-binding factor (either TFIID or TBP), in the absence (Ϫ) or presence of Gal4-VP16, as outlined at the bottom. Transcriptional components (Gal4-VP16, f:pol II, and a TATA-binding factor) were added either during (Dur) or post (Post) chromatin assembly which was conducted on pG 5 HMC 2 AT by using Drosophila S190 chromatin assembly extract (S190) and purified core histones. The DNA template pHMC 2 AT-200, added after chromatin assembly, was used as an internal control for naked DNA transcription. Initiation of transcription was then started by adding ribonucleoside triphosphates (NTPs) and the remaining transcription components not included during chromatin assembly. The transcripts are finally analyzed on a polyacrylamide/urea gel and visualized by autoradiography. Fold activation is defined as the signal intensity, quantitated by PhosphorImager (Molecular Dynamics), of the pG 5 HMC 2 AT/pHMC 2 AT-200 ratio in each reaction relative to that performed with TFIID and f:pol II added after chromatin assembly in the absence of Gal4-VP16 (i.e. the first lane of the reaction set). further dissect the molecular mechanisms of TAF II -independent and TAF II -dependent activation by various transcriptional activators. Interestingly, the ER-mediated activation in our highly purified transcription system occurs in a ligand-independent manner (Fig. 5B). 2 This is distinct from the ligand-dependent ER-mediated activation previously observed on an ERE-containing chromatin template with HeLa nuclear extracts as the source of general transcription factors and cofactors (63). Since CBP, p300, and SRC-1 are not present in our highly purified in vitro transcription system (Fig. 5C), our data suggest that ER may functionally target components of the general transcription machinery to activate transcription. This viewpoint is further supported by the observations that ER can interact directly with TBP (74,75), TAF II 30 (76), TAF II 28 (77), and TFIIB (75,93). It therefore seems likely that the primary function of many nuclear hormone receptor coactivators are to help antagonize the inhibitory activities such as nucleosomes and negative cofactors commonly present in a cellular environment. When these inhibitory activities are not present, ER has the intrinsic activity to enhance the level of transcription. It is of great interest to build up the complexity of protein factors and natural DNA templates to our highly purified transcription system in order to address the role of other general cofactors and many nuclear hormone receptor-specific coactivators in ER-mediated activation. Whether other transcriptional cofactors can functionally replace or work synergistically with TAF II s in ER-mediated activation can now be readily tested in our highly purified in vitro transcription system, which will eventually uncover the roles of general cofactors and genespecific cofactors in eukaryotic transcription.
Chromatin Transcription-Another role of TAF II s in gene regulation is likely to occur at the chromatin level, given the observations that the largest subunit of TFIID has both protein kinase and histone acetyltransferase activities (30,33) and that some TAF II components can form a histone-like structure (34,35). Since chromatin transcription could not be observed in our highly purified cell-free transcription system where TAF IIindependent and TAF II -dependent activation could occur on deproteinized DNA templates, 2 we resorted to the use of a transcription system reconstituted with a TATA-binding factor and TFIID-deficient pol II holoenzyme which contains pol II, a subset of GTFs (TFIIB, TFIIE, TFIIF, and TFIIH), SRBs, the SWI/SNF chromatin remodeling factor, and many undefined polypeptides that may play a critical role in chromatin transcription (Ref. 19, and Fig. 6). Whether the active components in f:pol II required for chromatin transcription correspond to any of the previously described chromatin remodeling factors and cofactors such as ACF, NURFs, CHRAC, RSC, SWI/SNF, E-RC1, RSF, and FACT (94 -98) remains to be established. Nevertheless, the use of TFIID-deficient pol II holoenzyme, in conjunction with either TFIID or TBP, allows us to define the role of TAF II s in chromatin transcription. The observation that preincubation of either TFIID or TBP during chromatin assembly helps overcome nucleosome-mediated repression in our chromatin transcription system (Fig. 7) is consistent with previous results performed in relatively crude transcription systems (99). Although TAF II s are necessary for transcribing chromatin templates (Ref. 67 and Fig. 7), it is unclear which activity of TFIID is responsible for chromatin transcription. Since TFIID has been implicated as a core promoter-binding factor, an antirepressor, a transcriptional coactivator (or corepressor), a histone acetyltransferase, and a protein kinase, any of these activities may confer upon TFIID the ability to work in conjunction with the pol II transcription machinery as well as chromatin remodeling factors and cofactors in transcribing a preassembled chromatin template. Clearly, the molecular mechanism by which TAF II s are collectively required for chromatin transcription remains to be further investigated.