Identification of two novel TAF subunits of the yeast Saccharomyces cerevisiae TFIID complex.

Using a combination of ion exchange and immunoaffinity chromatography we have purified the general transcription initiation factor TFIID to near homogeneity from Saccharomyces cerevisiae. Yeast TFIID is composed of TBP, the TATA box binding protein, and 14 distinct TBP-associated factors (TAFs), which range in size from 17 to 150 kDa. Twelve of the TAF subunits have been previously identified, but two, TAF48p and TAF65p, are novel. TAF48p exhibits significant sequence similarity to the conserved C-terminal region of Drosophila TAF110p, human TAF130p, and human TAF105p and is encoded by a previously identified gene MPT1. TAF65p shows no significant sequence homology to any previously identified TAFp. The genes encoding TAF48p and TAF65p are single copy and essential for normal yeast cell growth. Furthermore, neither TAF48p nor TAF65p are associated with the histone acetylase Spt-Ada-Gcn5 complex or other non-TFIID TBF.TAF complexes. The significance of these results in terms of TFIID structure, function, and organization is discussed.

Using a combination of ion exchange and immunoaffinity chromatography we have purified the general transcription initiation factor TFIID to near homogeneity from Saccharomyces cerevisiae. Yeast TFIID is composed of TBP, the TATA box binding protein, and 14 distinct TBP-associated factors (TAFs), which range in size from 17 to 150 kDa. Twelve of the TAF subunits have been previously identified, but two, TAF48p and TAF65p, are novel. TAF48p exhibits significant sequence similarity to the conserved C-terminal region of Drosophila TAF110p, human TAF130p, and human TAF105p and is encoded by a previously identified gene MPT1. TAF65p shows no significant sequence homology to any previously identified TAFp. The genes encoding TAF48p and TAF65p are single copy and essential for normal yeast cell growth. Furthermore, neither TAF48p nor TAF65p are associated with the histone acetylase Spt-Ada-Gcn5 complex or other non-TFIID TBF⅐TAF complexes. The significance of these results in terms of TFIID structure, function, and organization is discussed.
The general transcription initiation factor TFIID 1 plays a central role in the initiation of DNA-dependent RNA polymerase II (RNAP II) transcription. TFIID is the only general transcription initiation factor (GTF) with a specific TATA box binding activity, and binding of TFIID to the TATA box is the first and rate-limiting step in formation of a complex competent to initiate transcription in vitro (1). A TATA box element is found within the promoter of many mRNA encoding genes and is required for specific transcription initiation both in vitro and in vivo (2).
Biochemical studies published to date from yeast, human, and Drosophila systems have revealed TFIID to be a multisubunit complex comprised of TBP, the TATA box binding protein, and 10 -12 TBP-associated factors (TAFs) (2)(3)(4). TBP is responsible for the TATA box binding activity of TFIID and is conserved throughout eukaryotes. Like TBP, TAFs are highly conserved throughout eukaryotes, but their exact role in TFIID function has been an area of considerable debate (see Ref. 5 for review). Studies in yeast have clearly demonstrated a critical role for TAFs in mediating RNAP II transcription in vivo (6 -10). Whether this role can be directly ascribed to TFIID function has been difficult to interpret as a subset of TFIID TAF subunits, specifically TAF90p, TAF61p, TAF60p, TAF25p, and TAF17p, have been identified as integral subunits of the yeast histone acetylase Spt-Ada-Gcn5 (SAGA) complex (11). However, a recent study has shown that loss of TAF40p function (TAF40p is believed to be specific to TFIID) severely impairs ongoing high level RNAP II-mediated transcription in vivo (12). This result strongly argues that TFIID is required for the transcription of many RNAP II-dependent genes in vivo.
Studies in the yeast system have proven invaluable in understanding the role of TAFs, and thus TFIID function, in mediating RNAP II transcription in vivo. It is somewhat surprising then that yeast TFIID is rather poorly defined biochemically. Independently, we and others have used a combination of biochemical purification and protein sequence comparison to identify 12 yeast TFIID TAF subunits (13)(14)(15)(16)(17). Although these studies clearly suggest the existence of a yeast TFIID complex composed of TBP plus 12 TAFs, this fact has not been directly shown biochemically. Indeed, it is not clear at present whether this is the minimal or core TFIID assembly. Of the 12 known yTAF TFIID subunits, only one, TAF47p, does not have a known metazoan homologue (see Table I). Conversely, metazoan TAFs dTAF110p and hTAF130p, which have been shown to mediate transactivation by the glutamine-rich transactivator Sp1 (18,19), do not have a known yeast homologue. Thus it has been suggested that yeast may be evolutionarily distinct from metazoans and lacking the co-activator(s) necessary to mediate transactivation by glutamine-rich activators, as it was initially reported that Sp1 could not stimulate transcription in yeast (20). However, this conclusion seems contradictory because native yeast activators such as Hap2p and Mcm1p both contain glutamine-rich domains, which by themselves are capable of transactivation (21). Furthermore, Sp1 can stimulate transcription in yeast when the reporter gene of interest is carried on a high copy plasmid, but not when integrated into the genome (21). Additionally, a recent report has identified a potentially novel histone-like pair in hTAF20p and hTAF130p (22). hTAF20p has a yeast homologue in yeast TAF61p. Because other histone-like pairs are highly conserved between yeast and humans, it would be surprising if yeast did not have an equivalent interacting pair to hTAF20p-hTAF130p. Together, these results suggest the possibility of an unidentified yeast equivalent to metazoan dTAF110p and hTAF130p.
As a first step toward a concise biochemical analysis of yeast TFIID, we have purified TFIID from Saccharomyces cerevisiae to near homogeneity. We find that yeast TFIID is reproducibly composed of TBP plus 14 distinct TAFs. Two of the 14 TAFs, TAF48p and TAF65p, are novel. Immunoprecipitation experiments indicate that TAF48p and TAF65p are TFIID-specific TAFs and that these TAFp are not components of the yeast SAGA complex. The genes encoding TAF48p and TAF65p are both single copy and essential for vegetative growth. TAF65p does not resemble any previously identified metazoan TAF, but TAF48p is similar to the conserved C-terminal region of dTAF110p, hTAF130p, and hTAF105p. The significance of these results is discussed below.

MATERIALS AND METHODS
Strains and Plasmids-Escherichia coli strain XL-1 Blue (23) was used for routine plasmid propagation. Plasmid pDP15-HATBP was derived from pTBP p (24) by insertion of an oligo encoding one copy of the influenza hemagglutinin (HA) epitope (25) into the SalI site at the N terminus of the TBP open reading frame (ORF). pRS313-HATAF130 was derived from pRS313-HA 3 TAF130 (24) by deleting the second and third HA repeats using site-directed mutagenesis. E. coli expression plasmids for TAF48p and TAF65p were created by polymerase chain reaction amplification of the appropriate ORF and ligation into pR-SET-A (Invitrogen). An E. coli expression plasmid for Dr1p was made in a similar fashion except the ORF was ligated into pET15b (Novagen). Standard laboratory protocols and techniques for DNA manipulations were used (26), and all plasmids were verified by DNA sequencing.
Purification of TFIID-Buffer A (BA) is 20 mM HEPES-KOH, pH 7.6, 10% glycerol, 1 mM DTT, 0.1 mM PMSF, 1 mM benzamidine, plus the indicated millimolar amount of KOAc (e.g. BA300 has 300 mM KOAc). All manipulations were performed at 4°C unless otherwise stated. The 1 M Bio-Rex 70 fraction was prepared from YSLS14 or YSLS18 essentially as described (33) except loading and washing of the Bio-Rex 70 column was performed with BA300. Immunoaffinity chromatography was performed as follows. Typically, to 40 mL of 1 M Bio-Rex 70 faction (15-20 mg/mL), 10% Nonidet P-40, and 10 mg/mL ethidium bromide were added so that, after dilution with BA0 to a conductivity equal to that of BA300, the final concentrations were 0.1% and 0.1 mg/mL, respectively. 10 mg of anti-HA (mAb 12CA5, Roche) cross-linked to 2.5 mL of protein A-Sepharose (Sigma) was added and the slurry was incubated 12 to 14 h on a tilted board. The resin was collected in a 10-mL column and washed extensively with BA300 plus 0.1% Nonidet P-40, then with BA200 plus 0.001% Nonidet P-40. Elution was performed by adding 1 volume of BA200 containing 0.001% Nonidet P-40 with 2 mg/mL 3 ϫ HA peptide (14). The slurry was incubated 30 to 40 min at 24°C, the elute was collected by a brief spin at 1000 rpm, and a second elution was performed as above. Elutes were pooled, and bovine serum albumin was added to a final concentration of 50 g/mL. The TFIID fraction was applied to a HR 5/5 Mono S column (Amersham Pharmacia Biotech) equilibrated in BA200 plus 0.001% Nonidet P-40. Neither the 3 ϫ HA peptide nor major protein contaminants bound to the column. Bound proteins were resolved with a 15-mL linear gradient to BA1000 plus 0.001% Nonidet P-40. TFIID eluted in a single peak at ϳ375 mM KOAc. Peak fractions (3-4 mL) were pooled, dialyzed against BA200 with 30% glycerol, aliquoted, and stored at Ϫ80°C.
Protein Identification-The identification of previously known TFIID subunits was confirmed in the Mono S fraction by immunoblotting. The identification of unknown proteins was performed as follows. Mono S-purified HATBP TFIID was resolved using SDS-polyacrylamide gel electrophoresis (PAGE) on a 10% NuPAGE (Novex) gel and stained with Coomassie Brilliant Blue. Each of the unknown bands was excised and subjected to in-gel tryptic digestion and MALDI-TOF mass spectroscopy (Borealis Biosciences Inc.). The identified peptides covered 40 -50% of the total protein sequence for each identified protein. Protein sequence searches were performed with BLAST 2.0 software (NCBI), and Clust-alW alignments were done with MacVector 6.5 software (Oxford Molecular).
Antibodies and Immunoprecipitations-Recombinant protein production and generation of affinity-purified polyclonal antibodies for Dr1p, TAF48p, and TAF65p was performed as described (10). Polyclonal antibodies against other TFIID subunits and Gcn5p have been described (10). Horseradish peroxidase-conjugated anti-HA (mAb 3F10, Roche) was utilized for immunoblotting HA-tagged proteins. For immunoprecipitations, 50 A 600 units aliquots of exponentially growing yeast cells were harvested, washed with distilled H 2 O, transferred to a 1.5-mL microcentrifuge tube, frozen on dry ice, and stored at Ϫ80°C. All subsequent steps were performed at 4°C. Cells were resuspended in 0.6 mL of lysis buffer (20 mM HEPES-KOH (pH 7.6), 10% glycerol, 300 mM KOAc, 0.1% Nonidet P-40, 1 mM dithiothreitol, 1 mM EDTA, plus protease inhibitors), ϳ0.7 mL of glass beads were added, and cells were lysed by a 30-s burst with a Mini-Beadbeater-8 (BioSpec). Cell debris was pelleted by centrifugation (10 min at 15,000 ϫ g). Typically 0.2-0.3 mL of whole cell extract (ϳ7 mg/mL) was recovered. The extract was pre-cleared by addition of a one-half volume of protein A-Sepharose, which had been washed and slurried to 1:1 in lysis buffer. The resinextract slurry was incubated for 5 min, the beads were pelleted by a brief spin in a microcentrifuge, and an aliquot of the cleared extract was taken as input. To 50 L of the cleared extract was added: 5-10 g of anti-HA (mAb 12CA5) or anti-Flag (mAb M2, IBI) antibody cross-linked to protein A-Sepharose resin (2 mg/mL); 2 L of 10 mg/mL ethidium bromide; and lysis buffer to 200 L. Reactions were incubated 12-14 h with agitation. Antigen-antibody complexes bound to the beads were recovered by a brief spin in a microcentrifuge, beads were washed three times with 0.5 mL of lysis buffer, and 20 L of sample buffer was added. The indicated amount of input and precipitate (see Figure legends) was subjected to SDS-PAGE on a 4 -12% NuPAGE gel and transferred to Immobilon-P (Millipore), and immunoblotting was performed as described (10).

RESULTS
Purification of TFIID-Having previously used a combination of ion exchange and immunoaffinity chromatography to first identify TAFs in S. cerevisiae (14,33), we took a similar approach to purify TFIID. Strains that harbored a single copy of the HA epitope tag at the N terminus of either SPT15 (TBP) or TAF130 were subjected to the purification scheme outlined in Fig. 1A. It is important to note that the HA-tagged alleles were the sole source of TBP or TAF130p in the cell. Furthermore, tagged alleles were under the control of the appropriate native promoter such that protein levels in HA-tagged and wt untagged strains were identical (data not shown). SDS-PAGE and Coomassie staining of pooled peak fractions from the final Mono S column indicated that these TFIID preparations contained 17 polypeptides, which were consistently copurified (Fig.  1B). Immunoblotting confirmed the identity of TBP and all 12 previously identified TAF subunits believed to comprise yeast TFIID ( Fig. 1B and data not shown). Four additional unknown polypeptides with apparent molecular masses of ϳ65 kDa and a triplet of proteins of ϳ48 kDa also were consistently copurified in the TFIID fraction (see TAF48p* and TAF65p*, in Fig.  1B and below). Importantly, the composition of TFIID was essentially identical whether purified from the HATBP or HATAF130 strain. These results, together with the fact that immunoblotting indicated that the major fraction (70 -100%) of those TFIID TAF subunits previously believed to be specific to TFIID (TAF130p, TAF67p, TAF47p, TAF40p, and TAF19p) were recovered in the 1 M Bio-Rex 70 fraction (data not shown), suggest that the major form of TFIID in S. cerevisiae is composed of TBP plus 14 distinct TAFs.
Identification of Novel TAFs-As described above, four unknown polypeptides were consistently copurified in the TFIID fractions. Each of the unknown polypeptide bands were excised from a Coomassie-stained gel and subjected to in-gel trypsin digestion and MALDI-TOF mass spectrometry for identification. Peptide sequence from the unknown 65-kDa protein iden-tified a previously predicted ORF, YML114c, which encodes a protein with a deduced molecular mass of 58 kDa with no known function. We refer to the protein encoded by YML114c as TAF65p. BLAST search analysis revealed no significant homology to any protein of known function. However, TAF65p does contain a potential coiled-coiled domain similar to that of myosin (data not show). Gene disruption in a diploid strain followed by tetrad analysis indicated that TAF65 is essential for yeast cell viability (data not shown).
Peptide sequence from the three ϳ48 kDa polypeptides indicated that they represented differentially phosphorylated forms of the same protein (data not shown) encoded by the previously identified ORF MPT1. MPT1 is an essential yeast gene (data not shown) 2 previously predicted to encode a protein with a deduced molecular mass of 42 kDa required for protein synthesis. 2 We refer to the protein encoded by MPT1 as TAF48p. Interestingly, BLAST search analysis revealed that the C-terminal region of TAF48p bears significant similarity to the conserved C termini of metazoan dTAF110p, hTAF130p, and hTAF105p (Fig. 2). This result was quite unexpected, because previous BLAST search analysis of the complete S. cerevisiae genome with dTAF110p and hTAF130p sequence had not revealed an obvious sequence homologue. In retrospect, however, this result may not be surprising, because the region of similarity between TAF48p and its metazoan counterparts is quite small.
To confirm the identity of TAF48p and TAF65p in TFIID, recombinant protein and affinity-purified polyclonal antibodies directed against either TAF48p or TAF65p were generated. Immunoblotting of the HATBP and HATAF130p TFIID fractions shown in Fig. 1B indicated that the antibodies specifically recognized the appropriate polypeptides identified by mass spectroscopy (data not shown).
To further test whether TAF65p and TAF48p where indeed TFIID TAF subunits, immunoprecipitations were performed with either anti-Flag or anti-HA antibodies utilizing whole cell extracts prepared from yeast strains with either wt (untagged) or HA 3 -tagged alleles of MPT1 (TAF48) or TAF65. It is important to note that the genomic copy of each allele harbors the HA 3 tag, thus the epitope-tagged allele is the sole source of either TAF48p or TAF65p in the cell. A HA 3 TAF25 strain was utilized as a control. Immunoblotting with anti-HA indicated that each HA 3 -tagged protein was specifically precipitated with anti-HA antibody but not with the control anti-Flag antibody (data not shown). Immunoblotting of precipitates indicated that TFIID subunits TBP, TAF130p, TAF90p, TAF67p, TAF61p, TAF60p, TAF47p, TAF40p, TAF25p, TAF19p, and TAF17p were specifically co-immunoprecipitated with anti-HA antibody from each HA 3 -tagged strain ( Fig. 3 and data not  shown). Importantly, no TFIID subunits were precipitated by anti-HA antibody from the untagged wt strain (Fig. 3). Together, these results strongly argue that both TAF48p and TAF65p are indeed integral TAF subunits of TFIID. Additionally, Superose-6 size fractionation of the crude whole cell extract used for TFIID purification revealed that TAF48p and TAF65p co-eluted with other TFIID TAF subunits in a single peak of ϳ1.7 MDa (data not shown). In this size fractionation assay no significant amount of either TAF48p or TAF65p was detected in the range of free TAFp, suggesting that TAF48p and TAF65p function only in the context of TFIID.
TAF48p and TAF65p Are TFIID-specific TAFs-In addition to the TFIID TBF⅐TAF complex, several other distinct TBF⅐TAF complexes have been identified (see Ref. 34 for review). To test whether or not TAF48p and TAF65p are specific to TFIID, immunoprecipitations from wt, HA 3 TAF48, HA 3 TAF65, and HA 3 TAF25 strains were probed for Dr1p, a component of the NC2 TBF⅐TAF complex (35,36). Although present in the input, Dr1p was not detected in the precipitate from any strain (Fig. 3). To further test the specificity of TAF48p and TAF65p, immunoprecipitations were performed from HA 3 TAF25, HA 3 MOT1, and HA 3 BRF1 strains as described above. Mot1p is in a TBF⅐TAF complex distinct from TFIID (37), and Brf1p is a TAF subunit of the RNAP III-specific GTF TFIIIB (33, 38). As expected, TBP was specifically coimmunoprecipitated from each HA 3 -tagged strain with anti-HA antibody (Fig. 4A). However, TAF48p and TAF65p were specifically co-immunoprecipitated with only HA 3 TAF25p (Fig.  4A). These results indicate that TAF48p and TAF65p are specific to the TFIID TBF⅐TAF complex.
Because a subset of TAFs (TAF90p, TAF61p, TAF60p, TAF25p, and TAF17p) have been identified as subunits of the yeast histone acetylase SAGA complex (11), we tested whether or not TAF48p and TAF65p were SAGA components. Immunoprecipitations from wt, HA 3 TAF48, HA 3 TAF65, and HA 3 TAF25 strains were probed for Gcn5p, the histone acetylase component of SAGA (39). Whereas Gcn5p specifically coimmunoprecipitated with HA 3 TAF25p, an integral subunit of both TFIID and SAGA, Gcn5p was not associated with either HA 3 TAF48p or HA 3 TAF65p (Fig. 3). Additionally, both TAF90p and TAF60p, but not TAF48p or TAF65p, were specifically co-immunoprecipitated by anti-HA antibody from a strain that harbors an HA 3 ADA3 allele as the sole source of Ada3p in the cell (Fig. 4B). Ada3p is an integral subunit of the SAGA complex (39). Together these results strongly argue that TAF48p and TAF65p are not components of the yeast histone acetylase SAGA complex. DISCUSSION Here we describe the use of a combination of ion exchange and immunoaffinity chromatography to purify the GTF TFIID from S. cerevisiae. Yeast TFIID is composed of TBP plus 14 distinct TAF subunits. A comparison of yeast TFIID with known metazoan TFIID subunits indicates that two yeast TAF subunits, TAF47p and TAF65p, currently do not have known  Fig. 3. A, TAF48p and TAF65p co-immunoprecipitate with HA 3 TAF25p but not HA 3 Mot1p or HA 3 Brf1p. Input equals 0.5% and precipitate equals 12.5%. B, TAF90p and TAF60p, but not TAF48p or TAF65p, co-immunoprecipitate with HA 3 Ada3p. Input equals 0.5% and precipitate equal 25%. metazoan homologues (Table I). Given the high conservation of TAFs in general, it is likely that metazoan homologues to yeast TAF47p and TAF65p will be identified. Two of the TAF subunits of yeast TFIID, TAF48p and TAF65p, are novel, and experiments presented here clearly demonstrate that both TAF48p and TAF65p are indeed bona fide TAFs. Moreover, these TAFs are specific to the TFIID TBF⅐TAF complex and are not shared subunits of the SAGA complex. Together these results represent a concise attempt to biochemically define TFIID from S. cerevisiae.
Several lines of evidence argue that the TFIID presented here is the major form of TFIID in S. cerevisiae. First, purification of TFIID from either HATBP-or HATAF130p-expressing yeast strains yielded TFIID preparations of identical polypeptide composition (Fig. 1). Second, immunoblotting indicated that TFIID eluted in a single peak from the Mono S column and TFIID TAF subunits eluted only in the TFIID peak (data not shown). Third, the major fraction (70 -100%) of those TFIID TAF subunits previously believed to be specific to TFIID (TAF130p, TAF67p, TAF47p, TAF40p, and TAF19p) were recovered in the 1 M Bio-Rex 70 fraction used for immunopurification (data not shown). Finally, immunoprecipitation of any one TFIID subunit from the 1 M Bio-Rex 70 fraction co-immunoprecipitated all other TFIID subunits in similar relative amounts (data not shown). It is important to note however, that our data does not rule out the possibility of other TFIID subunits, which are either loosely associated or substoichiometric. Additionally, the experiments presented here do not fully address the possibility of another, as yet, unidentified TFIID TAF-containing complex(es). Further biochemical studies will be required to address these possibilities.
The identification of a yeast equivalent to metazoan dTAF110p and hTAF130p, TAF48p, has important implications in terms of TFIID structure and function. It is interesting to speculate that, like its metazoan counterparts, yTAF48p may be a key cofactor in mediating transactivation by glutamine-rich activators. However, it is important to note that the N-terminal regions of dTAF110p and hTAF130p, which have been shown to be critical for mediating Sp1 activation (18,19), are not present in yTAF48p. Additionally, the glutaminerich domain of the native yeast activator Hap2p can stimulate transcription in yeast equally well whether the reporter gene of interest is on a high copy plasmid or integrated into the genome. Stimulation of transcription by metazoan Sp1 requires the reporter gene to be carried on a high copy plasmid (21). These results suggest that, if TAF48p is a co-activator for glutamine-rich activators in yeast, the mechanism of action is not highly conserved.
With the recent identification of a potentially novel histonelike pair between hTAF20p (similar to yTAF61p) and hTAF130p (22), it is easy to speculate that TAF48p is the heterodimeric partner of TAF61p in yeast TFIID. This hypothesis is supported by the fact that the region of highest conservation between yTAF48p and its metazoan counterparts is in the region of the putative histone-fold motif (Fig. 2). Furthermore, overexpression of TAF48p, but not other TFIID subunits, specifically suppresses the temperature-sensitive phenotype of a taf61 strain. 3.4 Additional studies will be required to formally prove this hypothesis.
It is intriguing that the potential dimeric partner of TAF61p is different in TFIID (TAF48p) and SAGA (Ada1p; see Ref. 22). One possible explanation for this could be that interactions between the histone-folds in these molecules could provide a key determinant in the formation of either TFIID or SAGA. Shared TAF subunits (TAF90p, TAF61p, TAF60p, TAF25p, and TAF17p) could form a structural scaffold upon which either TFIID or SAGA could be formed. Interactions between TAF61p and its TFIID-specific (TAF48p) or SAGA-specific (Ada1p) dimeric partner could further direct formation of the complete TFIID or SAGA complex. This possibility is supported by in vivo studies that suggest inactivation of a shared TAF subunit disrupts the integrity of both the TFIID and SAGA complexes (7,10,11). This possibility is not limited to yeast, because sharing of TAFp subunits has also been reported between human TFIID and the PCAF histone acetylase complex (40). Further biochemical and genetic studies will be crucial in investigating these possibilities as well as other aspects of TFIID structure and function to dissect the complex process of TFIID-mediated RNAP II gene transcription.  TBP  TBP  TBP  TAF150  TAF150  TAF150  TAF130(145)  TAF250  TAF230  TAF90  TAF100  TAF80  TAF67  TAF55  TAF55  TAF65 a  TAF61(68)  TAF20  TAF30  TAF60  TAF80(70)  TAF62  TAF48 a  TAF130/105  TAF110  TAF47  TAF40  TAF28  TAF30  TAF30 AF