Mutations in the TATA-binding Protein, Affecting Transcriptional Activation, Show Synthetic Lethality with the TAF145Gene Lacking the TAF N-terminal Domain in Saccharomyces cerevisiae *

The general transcription factor TFIID, which is composed of the TATA box-binding protein (TBP) and a set of TBP-associated factors (TAFs), is crucial for both basal and regulated transcription by RNA polymerase II. The N-terminal small segment of yeast TAF145 (yTAF145) binds to TBP and thereby inhibits TBP function. To understand the physiological role of this inhibitory domain, which is designated as TAND (TAF N-terminaldomain), we screened mutations, synthetically lethal with the TAF145 gene lacking TAND (taf145ΔTAND), in Saccharomyces cerevisiae by exploiting a red/white colony-sectoring assay. Our screen yielded several recessive nsl (ΔTANDsynthetic lethal) mutations, two of which,nsl1-1 and nsl1-2, define the same complementation group. The NSL1 gene was found to be identical to the SPT15 gene encoding TBP. Interestingly, both temperature-sensitive nsl1/spt15 alleles, which harbor the single amino acid substitutions, S118L and P65S, respectively, were defective in transcriptional activation in vivo. Several other previously characterized activation-deficient spt15alleles also displayed synthetic lethal interactions withtaf145ΔTAND, indicating that TAND and TBP carry an overlapping but as yet unidentified function that is specifically required for transcriptional regulation.

In eukaryotes, transcriptional initiation of protein-coding genes is precisely regulated by the concerted action of a large number of proteins, e.g. general transcription factors (TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH), negative and positive cofactors, coactivators, and chromosome-modifying factors, in addition to RNA polymerase II (reviewed in Refs. [1][2][3][4][5]. The general transcription factor TFIID, a multiprotein complex composed of the TATA-binding protein (TBP) 1 and more than 10 TBP-associated factors (TAFs), can recognize specifically a number of core promoter elements such as the TATA box, initiator element, and downstream promoter element (reviewed in Refs. 6 -9). It nucleates the assembly of the preinitiation complex around the transcriptional initiation site by recruiting several other general transcription factors and RNA polymerase II, either in a stepwise manner or as a preassembled unit (i.e. RNA polymerase II holoenzyme) (reviewed in Refs. 10 -12). The importance of the binding of TFIID (or TBP) to the core promoter in transcriptional regulation has been extensively studied by various approaches (13)(14)(15)(16)(17)(18)(19). Biochemical studies have demonstrated that suboptimal core promoter recognition by TFIID might generate a substantial energetic barrier for initiating transcription, however, gene-specific activators should overcome this rate-limiting step by inducing conformational changes of TFIID (16 -19). Activator-bypass experiments, mostly performed in yeast, in which TBP or TAFs were physically connected to the heterologous DNA binding domain of gene-specific activators, showed that binding of TFIID (or TBP) to the core promoter was indeed a rate-limiting step in vivo that could be alleviated by artificial recruitment of TFIID (reviewed in Refs. 12 and 20). The recently developed DNA-cross-linking chromatin immunoprecipitation assay has made it possible to test directly whether binding of TFIID (or TBP) to the core promoter could be a bona fide regulatory step for activators in living cells (14,15,21,22). Results of assays for transcriptional activity using more than 30 promoters correlates well with the degree of TBP occupancy on the core promoter (14), arguing that TBP-TATA interactions should be the most critical step for gene regulation. These observations imply the presence of cofactors that modulate TBP-TATA interactions negatively and/or positively.
Several proteins have been reported to date that inhibit TBP-TATA interactions (reviewed in Refs. [23][24][25]. TAND (TAF N-terminal domain), originally isolated from the N terminus of Drosophila TAF230 (dTAF230), interacts with TBP directly so that TAND prevents TBP from binding to the TATA element (26 -29). Functionally equivalent domains are conserved at the N terminus of orthologous TAFs among various species, 2 (30,31) suggesting that TAND should be involved in certain principal functions of TFIID (24,32,33). TAND consists of two subdomains, TAND1 and TAND2, each of which binds to the concave and convex surface of TBP in a competitive fashion with acidic activation domains and TFIIA, respectively (28,30,34). This structural configuration of the complex formed between TAND and TBP was recently confirmed by NMR spectroscopic studies (29,35). Together with previous observations that activators and TFIIA are essential factors for inducing conformational change of TFIID during the course of activation (16,17,19), we assume that TAND-TBP interactions should be one of the most important regulatory targets for activators (32).
SAGA, a Gcn5-containing histone acetyltransferase coactivator complex, plays a key role in the regulation of transcription of several genes, e.g. HIS3 and TRP3 in Saccharomyces cerevisiae (reviewed in Refs. 36 and 37). Whereas several yeast TAFs (e.g. TAF90, TAF68, TAF60, TAF25, and TAF17) as well as TBP are shared by TFIID and SAGA, histone acetyltransferase activities involved in coactivator function are encoded by distinct subunits, TAF145 and Gcn5, that are specific to TFIID and SAGA, respectively (36,37). Interestingly, Spt3 and Spt8, which are SAGA-specific components, bind to TBP and inhibit TBP-TATA interactions as observed for TAND in TFIID (38). Although it remains to be determined whether Spt3-or Spt8-TBP interactions are regulated by activators or other factors, Spt8 appears to be dissociated from the SAGA complex under conditions of activation so that such inhibitory interactions can be alleviated (38).
Mot1, a member of the SWI2/SNF2 helicase family, is the third intriguing factor that inhibits TBP-TATA interactions (39). It was originally identified as the mot1-1 allele that leads to increased basal expression of many genes in S. cerevisiae (40). Several lines of investigation uncovered it to be a 170-kDa component of the TBP-TAF complex distinct from TFIID (41,42) or as an ADI (ATP-dependent inhibitor of TBP binding to DNA) factor that dissociates TBP from the TATA element in an ATP-dependent manner (39,43). Recent studies provide evidence that Mot1 can not only repress but also stimulate transcription of certain genes presumably by regulating the distribution of a limiting pool of TBP between promoter and nonpromoter sites (44,45). Interestingly, it was demonstrated that MOT1, SPT3, and TOA1, which encodes the larger subunit of TFIIA, interacted genetically with each other, i.e. toa1 and spt3 are synthetically lethal with mot1 and the ⌬spt3 phenotype is partially suppressed by overexpression of Toa1 (45). These observations strongly indicate that TBP-TATA interactions are intricately regulated in vivo by a wide variety of factors such as TFIID, SAGA, Mot1, and TFIIA.
Here, to identify factors that are functionally related to the TAF N-terminal domain (TAND) of TAF145, we have screened for nsl (⌬TAND synthetic lethal) mutations that cause lethality in combination with the TAF145 gene that lacks TAND (taf145⌬TAND) in S. cerevisiae. Our screen identified two distinct temperature-sensitive (TS) alleles of the NSL1 gene, as alleles of the SPT15 gene which encodes TBP. Further characterization suggests that activation-defective TBP mutants tend to display synthetic lethal interactions with the taf145⌬TAND gene. This is in accordance with a hypothetical role of TAND in transcriptional regulation that we have recently proposed. Taken together with previous observations that the TS phenotype of TAND-lacking strains can be suppressed by overexpressing TBP or TFIIA, TAND is likely to assist TBP function rather than simply inhibit it, at least in vivo.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Genetic Analyses-Standard techniques were used for yeast growth and transformation (46 -48). Yeast extract/ peptone/dextrose (YEPD) and selective media have been described (46). Transformation was done using the lithium acetate procedure (49). Yeast strains used in this study are listed in Table I. The host strain, TMY4-2, used for the synthetic lethal screen was constructed by targeted integration of the taf145⌬TAND allele into the original TAF145 locus of the CH1305 strain (kindly provided by Dr. Connie Holm) (50) as follows. pTM6 plasmid was constructed by subcloning a 1.2-kb HindIII fragment of pGT5 (51) carrying the URA3 gene and a 2.4-kb NotI-BglII (partial) fragment of pM10 (28) carrying a 3Ј-truncated taf145⌬TAND allele into the HindIII and NotI-BamHI sites, respectively, of pBluescript II (Stratagene). TMY4-2 was generated by a recombination-mediated two-step gene replacement procedure (52), first by transforming CH1305 to Ura ϩ with pTM6 that had been linearized at the unique BglII site, and then by screening for a Ura Ϫ Ts ϩ segregant bearing the taf145⌬TAND allele on 5-FOA plates. The gene replacement was confirmed by PCR.
TMY17-2c and TMY16-2c were made by mating-type interconversion from TMY4-2 and CH1305, respectively. The YEp13-HO plasmid, expressing the homothallic switching endonuclease endonuclease (53), was transformed into TMY4-2 and CH1305, both of which are MATa, to generate MATa/MAT␣ diploid cells. After segregating the YEp13-HO plasmid away, cells were sporulated and dissected to isolate MAT␣ haploid cells, i.e. TMY17-2c and TMY16-2c. MAT␣ ade2 ade3 leu2 ura3 lys2 can1 spt15-S118L This study The YTK271 strain was generated from H2440 (kindly provided by Dr. A. G. Hinnebusch) (30) by targeted disruption of the SPT15 gene using a marker cassette that has a URA3 gene between duplicated copies of a Salmonella hisG gene segment (54). The cassette plasmid has the 5Ј-flanking sequence (ϳ500 bp upstream of the initiation codon) and 3Ј-flanking sequence (ϳ500 bp downstream of the termination codon) of the SPT15 gene on each side of URA3 marker. These flanking sequences were amplified by PCR with primers to create EcoRI-BglII and SalI-BamHI sites, respectively. The linear fragment digested with EcoRI and SalI was used to transform H2440. The structure of the disrupted gene was confirmed by Southern blotting. Since the SPT15 gene is essential for viability, we dissected heterozygously disrupted strains that had been transformed with the TRP1-marked plasmid carrying the SPT15 gene. Ura ϩ Trp ϩ haploid strains obtained from tetrad analysis were grown on 5-FOA plates to excise the URA3 marker from the chromosome. The resulting Ura Ϫ Trp ϩ strains were subsequently transformed with the URA3-marked plasmid carrying the SPT15 gene (pYN11). The YTK271 strain (Ura ϩ Trp Ϫ ) was selected by plasmid segregation. YAK289, 293, 493, 495, 582, 584, 586, 588, 620, 622, 633, 636, 938 strains were then generated from YTK271 by a plasmid shuffling technique. YTK271 (⌬spt15 strain) was crossed with YKII1 (28) (⌬taf145 strain) and then dissected to obtain the haploid strain, YAK284, carrying double deletions of TAF145 and SPT15 genes. As TAF145 and SPT15 are both essential genes, the growth of YAK284 is supported by pYN11/ SPT15 (URA3 marker) and pM11/TAF145 (TRP1 marker). YAK303 andYAK307 were constructed by replacing pM11 of YAK284 with pM3217 and pTM26, respectively.
Synthetic Lethal Screen-To perform an ade2 ade3 colony-sectoring assay for synthetic lethality (50), the TMY4-2 haploid strain was transformed with pTM17. The pTM17 plasmid was constructed by subcloning a 5.2-kb SalI-BamHI (partial) fragment of pYN2 (30) carrying the TAF145 gene and a 3.7-kb NheI-BamHI fragment of pDK255 (55) carrying the ADE3 gene into the restriction enzyme sites between SalI and XbaI of pRS316 (56). TMY4-2, bearing pTM17, was grown in liquid SD medium lacking uracil (SD-Ura) to an absorbance at 600 nm (A 600 ) of around 0.7 and plated on YPD plates at a density of ϳ4000 cells/plate. The plates were then UV-irradiated, resulting in ϳ5-7.5% survival (200 -300 cells/plate), and incubated for 5 days at 25°C. Red colonies were restreaked three times on YPD plates. Colonies that remained red during this cultivation process were subsequently counter-selected on 5-FOA-containing plates. To exclude the possibility that nonsectoring and 5-FOA-sensitive phenotypes might be due to genomic integration of the pTM17 plasmid, mutant candidates were transformed with the plasmids bearing the TAF145 or taf145⌬TAND genes. True synthetic lethal mutants should show restored sectoring and growth on 5-FOA plates only upon transformation with the plasmid bearing the wild type TAF145 gene. From a total of ϳ40000 colonies screened, 14 strains (A1, A6, A22, A38, B9, B16, C2, C40, C60, C72, D7, D16, E4, and E10) had the synthetic lethal phenotype with the taf145⌬TAND gene (i.e. nsl phenotype). C40 and D7 strains were further characterized in this study.

Cloning of a Gene That Complements C40 and D7 nsl Mutants-C40
and D7 mutants were crossed to TMY17-2c bearing the taf145⌬TAND allele, and in both cases red to white sectoring was observed, indicating that these mutations were recessive. Diploids were sporulated, and 18 tetrads were dissected and analyzed phenotypically. When four spores were recovered, the sectoring phenotype and the 5-FOA lethality segregated 2:2, indicating that the synthetic lethality was presumably caused by mutation at a single locus. Crosses between mutant segregants from C40 and D7 assigned them to a complementation group.
TMY4-2 is unable to grow at 35°C due to the presence of the taf145⌬TAND allele but forms normal sized colonies at 25°C. C40 and D7 strains also show the TS phenotype even though they contain the pTM17 plasmid expressing the wild type TAF145 gene. Multiple backcrosses with TMY17-2c allowed determination of whether the TS phenotype was linked to synthetic lethality. As both mutations are linked to the TS phenotype, mutant segregants derived from C40 and D7 were transformed with a low copy number plasmid library (ATCC77162) yielding ϳ100,000 transformants on SD-Leu plates when grown at 25°C for 12-18 h and then shifted to 35°C and incubated for 7 days. Plasmids containing complementing genomic DNA fragments were recovered from the positive colonies and were amplified in Escherichia coli DH5␣. These plasmids were retransformed into C40 and D7 strains to confirm the complementation of the red/white sectoring, 5-FOA lethality, and the TS phenotype. Insert DNA boundaries were sequenced and compared with the yeast genome data base. Overlapping regions from chromosome V were obtained in all cases, and subcloning indicated that the presence of the SPT15 ORF was sufficient to complement all mutant phenotypes shown by C40 and D7.
Identification of Amino Acid Substitutions in SPT15 Gene of C40 and D7-The mutations in the SPT15 gene of C40 and D7 strains responsible for synthetic lethality were identified by sequencing. The 1.2-kb DNA fragment, including the entire SPT15 gene, was amplified by PCR using the primer pairs TK127 and TK128 from the isolated genomic DNA of these mutants. Direct sequencing of amplified DNA fragments using TK222, TK223, TK224, and TK225 as sequence primers revealed single C 3 T point mutations at 353 bp in C40 and at 193 bp in D7, which result in the amino acid substitutions S118L and P65S, respectively. Proof that these mutations conferred synthetic lethality was obtained by testing the nsl phenotype of taf145 alleles bearing these mutations produced by site-specific mutagenesis (57) as described below. Oligonucleotides used in this study are listed in Table II.
Phenotypic Analyses-To confirm the presence of the synthetic lethal phenotype in different general genetic backgrounds, YAK303 and YAK307 were transformed with pRS314-based plasmids encoding TBP derivatives as described above and then incubated on 5-FOA plates at 30°C for 5 days. To test the complementing activities of various plasmids by a red/white sectoring assay, colonies transformed with these plasmids were streaked onto YPD plates and then incubated at 25°C for 8 -10 days. Recovery of the TS phenotype was assayed by comparing the growth rates at 25 and at 35°C of yeast transformants incubated on YPD plates for 3-4 days.
In Vivo Activation Measured by ␤-Galactosidase Activity-For the artificial recruitment experiments, plasmids encoding GAL4-TBP derivatives were introduced into the CH1305 strain containing pB20, a multicopy URA3 plasmid with the GAL1 promoter upstream of the lacZ structural gene (kindly provided by Dr. A. G. Hinnebusch). The resulting strains were grown to an A 600 of 0.7 in YPD medium and then treated with repeated freeze/thaw cycles to measure the ␤-galactosidase activity as described previously (32). To measure activation by classical activation domains, yeast strains bearing mutant TBP derivatives (Table I) were transformed with pB20 and plasmids expressing various activators.
Preparation of Recombinant Proteins and GST Pulldown Assay-Hexahistidine-tagged TBP derivatives were expressed in E. coli BL21(DE3) (Novagen) by induction with 0.4 mM isopropyl-1-thio-␤-Dgalactopyranoside for 6 h at 16°C in M9 medium, and the cells were then resuspended in 10 ml of 0.5 M KCl, buffer C (25 mM Hepes, pH 7.6, 0.1 mM EDTA, pH 8.0, 12.5 mM MgCl 2 , 10% (v/v) glycerol, 0.1% Nonidet P-40, 1 mM dithiothreitol) per liter of culture volume. After sonication, cell debris was removed by centrifugation, and the supernatant was stored at Ϫ30°C. For electrophoretic mobility shift assays, cell lysates were subjected to Ni 2ϩ -nitrilotriacetic acid resin (Qiagen) to purify TBP derivatives. TBP, in the cleared lysate or purified through the Ni 2ϩresin, was quantitated by SDS-PAGE and Coomassie Brilliant Blue staining.
To study interactions between GST-TAND and TBP derivatives, E. coli extracts containing GST-TAND (30 pmol) and TBP derivatives (30 pmol) were mixed in 100 l of 0.2 M KCl/buffer D at 4°C for 30 min, incubated with 10 l of glutathione-Sepharose 4B (Amersham Pharmacia Biotech) for another 30 min, and washed three times with 500 l of 0.2 M KCl/buffer D. The complexes on the beads were eluted by boiling in SDS sample buffer and analyzed by Western blotting with polyclonal antibodies to TBP. Interactions between GST-VP16 and TBP derivatives were examined similarly except that two different concentrations of KCl (0.1 M and 0.2 M) were used in the binding and washing buffers.
Gel Retardation Assays-The gel retardation assay was performed as described previously (26,59) except that 2 g/l BSA was added to the reaction mixture. For TBP-TATA element interactions, 20 ng of TBP was used, and the complex was analyzed in a 4% polyacrylamide gel (59:1) containing TGMg buffer (25 mM Tris, 192 mM glycine, 2 mM MgCl 2 ) and 5% (v/v) glycerol using TGMg as a running buffer. When TFIIA binding to the TBP-TATA complex was tested, 10 ng of TBP and12 ng of TFIIA were added to the reaction mixture, and the complex was analyzed in a 4% polyacrylamide gel (59:1) containing 0.5ϫ TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA, pH 8.0) and 5% (v/v) glycerol using 0.5ϫ TBE as a running buffer.

RESULTS
Screen for nsl Mutants That Are Synthetically Lethal with the Loss of TAND Function-Recently we proposed that TAND, located at the N terminus of TAF145, might be involved in transcriptional regulation by acidic activators (32). However, as target gene(s), whose expression depends on the TAND-TBP interactions, have not been identified despite our extensive efforts, it seems likely that some other regulatory systems may exist in yeast cells that compensate for the loss of TAND-TBP interactions. To identify genes involved in such a parallel pathway, we screened nsl genes displaying synthetic lethal interactions with the taf145 gene that lacks TAND (taf145⌬TAND), by exploiting an ade2/ade3-based red/white colony sectoring assay (50,55). The principle of a synthetic lethal screen is that, although a single mutation is tolerable for the cell, a combination of mutations in functionally related pathways results in severe growth inhibition or cell death. This type of screen was often used for the identification of unknown component(s) regulating a common process in a wide variety of biological phenomena (60 -62).
The host strain, TMY4-2, used for our screen was constructed to harbor a taf145⌬TAND allele at the original chromosomal locus by homologous recombination and was then transformed with pTM17, a centromere-based plasmid that contains the URA3 and ADE3 nutritional markers and the wild type TAF145 gene. This host strain, TMY4-2, is red on YPD media but forms white sectors when the plasmid pTM17 is lost under nonselective conditions. Thus, we can expect that nsl mutant strains should show no white sectors in colonies since the TAND function on pTM17 becomes essential for growth. TMY4-2 was mutagenized by UV irradiation, and of approximately ϳ40000 surviving colonies screened, 30 failed to show red/white sectoring at 25°C. Of these, 14 regained a sectoring phenotype when transformed with a plasmid carrying the wild type TAF145 gene, but not with the taf145⌬TAND gene, and failed to grow on 5-FOA media (i.e. pTM17 plasmid carrying the URA3 and TAF145 genes is essential for cell growth). These observations strongly suggest that they contain mutations that are synthetically lethal with the taf145⌬TAND gene. Two of these nsl mutants, C40 and D7, were further characterized in this study.
Two nsl1 Mutants Carry Different Amino Acid Substitutions in the SPT15 Gene Encoding TBP-C40 and D7 mutants segregated into the plasmid-dependent red/white sectoring phenotype 2:2 when backcrossed to an isogenic parent strain harboring the chromosomal taf145⌬TAND gene, indicating that their nsl phenotypes resulted from single gene mutations. A complementation test revealed that they were recessive and allelic, thus hereafter this gene is designated as NSL1 (⌬TAND syn-thetic lethal 1). We also refer to mutant alleles of C40 and D7 as nsl1-1 and nsl1-2. Both nsl1 alleles seemed to exhibit TS growth phenotypes since all nsl1 spores obtained through backcross experiments with a parental strain harboring the chromosomal TAF145 or taf145⌬TAND gene were TS even in the presence of the plasmid, pTM17, expressing the wild type TAF145 gene (data not shown). Thus we tried to isolate the wild type NSL1 gene by complementing the TS phenotypes of these nsl1 mutants. nsl1-1 and nsl1-2 mutants were transformed with a partial Sau3A yeast genomic library, and several complementing colonies were isolated from both mutants. Retransformation analysis established that growth recovery at 35°C was dependent on the presence of the plasmids that were isolated from the original colonies (data not shown). Consistent with the allelismic test, restriction enzyme digestion and sequencing analysis revealed that genomic inserts included in plasmids, derived from either mutant (i.e. nsl1-1 or nsl1-2), overlapped with each other (data not shown). Importantly, only the SPT15 gene encoding TBP was found in all inserts. Thus we subcloned an ϳ2.4-kb EcoRI-BamHI fragment, from which only the SPT15 gene could be expressed, into a centromeric vector and tested it using the complementation assay. The plasmid carrying the subcloned fragment rescued the TS growth defects of both mutant strains (Fig. 1), suggesting that NSL1 might be allelic to SPT15.
To identify possible mutation(s) in the SPT15 gene of both mutants, we sequenced PCR-amplified genomic fragments encompassing the entire open reading frame plus 5Ј-and 3Јadjacent DNA regions (ϳ500 bp each) of the SPT15 gene. We found the single amino acid substitutions, S118L and P65S, in the coding region of the SPT15 gene of C40 and D7 mutants, respectively. Thus, we next asked whether these TBP mutations were sufficient to reproduce the nsl phenotype and whether such a phenotype depended on a particular genetic background. To address these questions, we constructed yeast strains containing either the wild type TAF145 gene (YAK303) or taf145⌬TAND gene (YAK307) on a centromeric LEU2 plasmid as well as the wild type SPT15 gene on a centromeric URA3 plasmid in combination with double deletions of chromosomal TAF145 and SPT15 genes. These strains have different general genetic backgrounds from the one used in the original genetic screen for the nsl mutants. The spt15-S118L and spt15-P65S alleles were reconstructed on a centromeric TRP1 plasmid by site-directed mutagenesis to exclude any other possible mutations. These plasmids were transformed into YAK303 and YAK307 strains described above and tested for their growth on 5-FOA plates. We reasoned that if spt15-S118L and spt15-P65S are responsible for the nsl phenotype, only strains carrying the wild type TAF145 gene (i.e. derived from YAK303) would be viable on 5-FOA plates that select for cells that had lost the URA3 and SPT15 containing plasmid. Consistent with this expectation, yeast strains carrying the taf145⌬TAND gene (i.e. derived from YAK307) grew well on 5-FOA plates only when they expressed the wild type SPT15 gene but not when they expressed the spt15-P65S or spt15-S118L alleles (Fig. 3A). We also confirmed that these mutant alleles were recessive and TS in the genetic background of YAK303 strains as observed for the original mutant strains. These observations support the notion that these TBP mutations, S118L and P65S, are synthetically lethal with the taf145⌬TAND gene even under the different general genetic background. Similar conclusions were confirmed by genetic experiments in which spores harboring the combination of taf145⌬TAND and spt15-S118L or taf145⌬TAND and spt15-P65S alleles were never recovered, as more than 20 asci were dissected for each diploid (data not shown).
Transcriptional Activation Is Impaired in spt15/nsl1 Mutants-The spt15-S118L allele was previously isolated as one of the TBP mutants that was specifically impaired in its response to acidic activators (63). Yeast strains containing the S118L mutant as the sole source of TBP are deficient for activation by Gcn4, Gal4, and Ace1, whereas transcription from pol I (rDNA), pol III (tRNA-I), TATA-less pol II (TRP3), and constitutive pol II (DED1 and RPS4) promoters is not impaired (63). On the other hand, the P65S mutant was reported to be defective for transcription in vitro from pol II (CYC1) as well as pol III (5 S rDNA and tRNA-L) promoters (64). These TBP mutants, S118L and P65S, showed poor growth at higher temperatures (63,64), and the TS phenotype of the latter was exploited to isolate the BRF1 gene (65) (also called as TDS4 (66) and PCF4 (67)) encoding a component of the pol III-specific general transcription factor, TFIIIB. As the overexpression of BRF1 suppressed the TS phenotype of the P65S mutant, transcription by pol III should be more severely affected than that by pol II in this mutant (65).
We proposed that TAND may play an important role in transcriptional activation by acidic activators (32). In this regard, it is intriguing that the activation-defective TBP mutant, S118L, was isolated in our screen as one of the nsl alleles. We reasoned that the P65S mutant also might be deficient in the response to acidic activators. Thus, we examined activation efficiencies in the P65S mutant that had been backcrossed to an isogenic wild type strain more than three times so as to avoid the effect of other unrelated mutations. The ␤-galactosidase activity from the Gal4 upstream activating sequence-dependent reporter plasmid was measured when various activation domains fused to Gal4 DNA binding domain were coexpressed in the cell ( Fig. 2A). As we expected, activation efficiencies were constantly lower in the P65S mutant than that in the wild type strains. TAND1, which binds the concave surface of TBP, can function as a strong activation domain when recruited onto the promoter by the Gal4 DNA binding domain (32). Interestingly, activation by TAND1 was most severely affected in the P65S mutant (4.2% of the wild type), whereas activation by TADIV of ADR1, which requires the presence of intact TFIID for its function (68), was least affected (50.7% of the wild type).
Weakened interactions of TBP, either with the TATA ele-

FIG. 1. Growth comparison of nsl1-1 and nsl1-2 mutants transformed with the single copy plasmids containing SPT15 (؉TBP) or no insert (؉vector).
These strains were streaked on SD plates and incubated at 25 or 35°C for 3 days. ment, TFIIA, or activation domains, were reported to lower the activation efficiencies by acidic activators (63, 69 -72). Thus, to identify which defects were most relevant to the nsl phenotype, we examined the abilities of these TBP mutants to bind to the TATA element and interactions with TFIIA, TAND, and VP16. TBP-P65S bound to the TATA element at normal levels, whereas binding of TBP-S118L was reduced to 50% or less of the wild type (Fig. 2B, upper panel). Although previous studies demonstrated that these TBP mutants lacked TATA binding activity (63,64), we found that they could bind to the TATA element when they were produced in bacterial cells incubated at 16°C. In contrast to TATA binding, TBP-S118L interacted with TFIIA much less weakly than the wild type form (a very faint signal was detected in Fig. 2B, lower panel, lane 4), whereas TBP-P65S did not form any detectable amount of the TFIIA-TBP-TATA complex under the same conditions (Fig. 2B,  lower panel, lane 6). Such affected interactions with TFIIA were previously observed for the other TBP mutant, N2-1 (K138T/Y139A), that also is defective in response to acidic activators (70). The interaction of TBP-S118L with TAND was weaker than that of TBP-P65S or the wild type form (Fig. 2C,  upper panel). Thus, serine 118 is important for the interaction with TFIIA and TAND (Fig. 2, B and C), whereas proline 65 appears to be specifically required for the interaction with TFIIA (Fig. 2, B and C). Consistently, biochemical and structural studies argue that TFIIA and TAND share, at least in part, the interaction surface of TBP (30,35). Interestingly, both TBP mutants bound to the acidic activation domain of VP16 more avidly than the wild type form (Fig. 2C, lower panel). Strong interactions became more evident when the complex was washed with higher concentrations of salt in the buffer. Contrary to this, the other type of activation-defective TBP mutant, L114K, was reported to be impaired in the interaction with the activation domain of VP16 (72). These two opposite traits of TBP mutants could lead to the same outcome (i.e. activation deficiency) according to our "two-step hand off model" where interactions, either too weak or too strong, between the activation domain and the concave surface of TBP may prevent the activation process (32).
Although the molecular defect found in TBP-S118L and TBP-P65S that contributes most significantly to the nsl phenotype has remained elusive, the results described above prompted us to test whether other activation-deficient TBP mutants also show synthetic lethal interactions with the taf145⌬TAND gene.
Activation-defective TBP Mutants Show Synthetic Lethality with the taf145 Gene Lacking TAND-A large number of TBP mutants have been isolated so far, and most have been characterized at the molecular level (reviewed in Refs. 1, 73, and 74). Special attention has been paid to a class of TBP mutants displaying activation-specific defects. Earlier genetic screens seeking such yeast TBP mutants identified several amino acid substitutions such as V71A, P109A, F116Y, S118L, F148L, N159D, N159L, and V161A (63,69). Most of these residues are located on the DNA-binding surface of TBP. These mutants are defective in TATA binding, suggesting that activators may facilitate the TBP-TATA complex formation or stabilize it. Other studies report the isolation of K138T/Y139A, F148H, T153I, E236P, and F237D substitutions as other types of activation-defective TBP mutants (70, 71). As described above, the K138T/Y139A mutant associated with the TATA element normally, but it had specifically lost the ability to bind TFIIA (70). The F237D mutant associated with the TATA element with a similar affinity as the wild type form but with an altered conformation so that it interacted quite poorly with TFIIA and TFIIB (71). On the other hand, the F148H, T153I, and E236P mutants interacted normally with the TATA element, TFIIA, TFIIB as well as the acidic activation domains, implying that it should be impaired in binding to unknown factors that are important for activation (71).
To see the correlation between the activation defects and the nsl phenotype, we selected different types of TBP mutants like N159D, N159L, and V161A (defective for TATA binding) (63,69), K138/139A and F237D (defective for TFIIA binding) (70,71), F148H, T153I, and E236P (impaired interaction with unknown factors) (71), and R220H and Y231A (specifically impaired in pol III-driven transcription) (75), and we determined their nsl phenotypes (Fig. 3A). As described for S118L and P65S mutants, we transformed the yeast strains, YAK303 and YAK307, with a centromeric TRP1 plasmid expressing each TBP mutant, and we examined their growth on 5-FOA plates at 30°C. Like the S118L and P65S mutants, other activationdefective TBP mutants also exhibited synthetic growth defects with the taf145⌬TAND gene but not with the wild type TAF145 gene. In contrast, the mutants, R220H and Y231A, which are specifically defective for transcription by pol III, did not cause any deleterious effect on growth with either alleles of the TAF145 gene. Despite all of the spt15/nsl1 alleles tested here that exhibit the TS phenotype on their own (data not shown), only the activation-defective TBP mutants had the nsl phenotype. Therefore, we assumed that the P65S mutant was isolated as a nsl1-2 mutant in our original screen due to its inefficient activation by pol II rather than to its impairment in pol III transcription.
We next investigated how these TBP mutants interacted with the TATA element, TFIIA, TAND, and VP16 (Fig. 3, B and  C), although similar analyses had been conducted for some of these mutants previously. Consistent with these studies (63,69), N159D, N159L, and V161A mutants were affected in their binding to the TATA element (Fig. 3B, upper panel; lanes 7, 14  and 15), and such defects were not rescued by the addition of TFIIA (Fig. 3B, lower panel, lanes 3, 10, and 11). K138/139A and F237D mutants were impaired in TFIIA binding (lower panel, lanes 2 and 8), but they could bind to the TATA element (upper panel, lanes 9 and 12). In this regard, they resemble S118L and P65S mutants. Note that the TATA binding activity of the K138T/Y139A mutant was sensitive to the presence of BSA in the reaction buffer. Much stronger TATA binding could be seen in the absence of BSA (upper panel, compare lane 6 and 9). F148H, T153I, E236P, R220H, and Y231A mutants interacted with both the TATA element and TFIIA almost normally. On the other hand, as shown in Fig. 3C, decreased interactions with TAND were observed for F237D, F148H, R220H, and Y231A mutants. Considering that the latter two mutants, i.e. R220H and Y231A, were viable when combined with the taf145⌬TAND gene, we believe that such deficiency may not be directly related to the nsl phenotype. With regard to the interactions with VP16, all TBP mutants retained the binding activity although some of these mutants, i.e. N159D, N159L, V161A, F237D, and E236P, bound to VP16 more strongly than the wild type in the buffer containing 0.2 M KCl (Fig. 3C, lower  panel) which was similar to the result obtained with the S118L and P65S mutants (Fig. 2C, lower panel). Interestingly, F148H and T153I mutants, which appeared to bind to VP16 with normal affinities, showed a weaker nsl phenotype (Fig. 3A). Collectively, the nsl phenotype can be achieved by various types of activation-defective TBP mutants but not by pol III transcription-defective TBP mutants. It is notable that stronger interactions with VP16 seem to be most closely correlated to this phenotype.

TBP Mutants Defective for the Post-recruitment
Step Exhibit a Stronger nsl Phenotype-To see the correlation between the degree of the nsl phenotype and that of activation defects, we compared activation efficiencies of these TBP mutants, described above, under the same conditions where VP16, GAL4, and GCN4 activation domains fused with GAL4DNA binding domain were used as activators (Fig. 4A). Consistent with previous studies (see also Fig. 2), all TBP mutants, except FIG. 3. A, activation-defective TBP mutants show the nsl phenotype. The TRP1-marked plasmids encoding TBP derivatives, as indicated on the left, were individually introduced into the strains with double deletions of TAF145 and SPT15 genes but which instead contained either the LEU2-marked plasmid encoding the wild type gene or the taf145⌬TAND gene, as indicated at the top, in addition to the URA3-marked plasmid encoding wild type TBP. The resulting transformants were grown on 5-FOA plates at 30°C for 5 days. B, TBP-TATA and TFIIA-TBP-DNA interactions were analyzed as described in Fig. 2. Note that TBP-K138T, Y139A binds weakly to the TATA element in the presence of 2 g/l BSA (lane 6), whereas it binds as well as the wild type in the absence of BSA (lanes 8 and 9). C, TBP-TAND and TBP-VP16 interactions were analyzed as described in Fig. 2. R220H and Y231A that are supposed to be specifically impaired in pol III transcription, were found to be more or less affected in activation by these activators (Fig. 4A). Nonetheless, we did not see any distinct correlations among the degree of the nsl phenotypes and the activation defects or the sort of activation domains present and the molecular defects observed. It was noteworthy that F148H and T153I mutants were affected in activation to a similar level as other TBP mutants (Fig. 4A) although they showed a weaker nsl phenotype than the others (Fig. 3A).
Previous studies demonstrated that transcription could be activated in the absence of activators by artificial recruitment of TBP that is physically connected to a heterologous DNA binding domain (76 -78). This simple in vivo recruitment assay could predict roughly which step(s) is impaired in each activation-defective TBP mutant (71). The rationale of how to interpret the result was originally provided by Stargell and Struhl (71), namely if an activation-defective TBP can activate transcription when it is recruited to the promoter, its defect must be involved in the step(s) before recruitment to the TATA element. Conversely, if the TBP mutant fails to activate transcription under the same conditions, it should lack the ability to proceed to post-recruitment steps. We transformed wild type strains with the reporter plasmid harboring the GAL1 promoterdriven lacZ gene as well as the effector plasmid expressing each TBP mutant fused to the GAL4 DNA binding domain (Fig.  4B). Consistent with previous studies, K138T/Y139A, E236P, and F237D mutants activated transcription less efficiently than the wild type gene (71,79). Besides these mutants, P65S, S118L, N159D, N159L, and V161A mutants also showed lower activities than the wild type gene in this assay. In contrast, F148H, T153I, R220H, and Y231A mutants activated transcription at similar levels to the wild type gene. Thus the defects in the post-recruitment step tend to represent a stronger nsl phenotype.
Integrity of the TFIID Complex Containing TAF145 Lacking TAND and the Activation-defective TBP Mutant-We previously demonstrated that the same amount of TAF145 protein was coprecipitated with TBP from cell lysates prepared either from wild type or ⌬TAND strains (28). The reason differences between these two strains were not observed was probably because other regions of TAF145 or other TAF components contributed to the stability of the TAF145-TBP interactions and/or supported the integrity of TFIID. Therefore, it is likely that the nsl phenotype may occur due to the instability of TFIID enhanced by combining two different mutations within the same complex. To explore this possibility, we constructed yeast strains in which the chromosomal SPT15 gene was deleted and instead carried two plasmids, one expressing the HA-tagged taf145⌬TAND gene (LEU2 marked plasmid) and the other expressing either wild type or activation-defective TBP (TRP1 marked plasmid). All strains were viable due to the presence of the wild type TAF145 gene on the chromosome. We tried to examine the integrity of TFIID in these strains by FIG. 4. A, activation by acidic activators in TBP mutants. ␤-Galactosidase activities of a GAL4-dependent reporter system were measured in strains containing the indicated TBP derivatives and either of three activators, i.e. GAL4DBD-VP16AD, GAL4DBD-GAL4AD, or GAL4DBD-GCN4AD. The values are represented as a percentage of the value obtained in the wild type strain containing the corresponding activators. Note that the host strain used here is different from that in Fig. 2A. B, artificial recruitment assay of GAL4DBD-TBP derivatives. The relative ␤-galactosidase activities of a GAL4-dependent reporter plasmid were measured in the wild type strain expressing the indicated GAL4DBD-TBP derivatives. Each value represents the average of three different isolates of each strain (A and B).
measuring the amounts of TAF145⌬TAND and TAF61 proteins that were coprecipitated with TBP (Fig. 5). Precipitates were blotted onto a membrane and probed with anti-HA, TAF145, TAF61, and TBP antibodies, respectively. The signal detected by the anti-HA antibody represents the amount of TAF145⌬TAND proteins, whereas the signal detected by anti-TAF145 antibody is derived from the sum of wild type TAF145 and TAF145⌬TAND proteins. Note that nontagged wild type TAF145 proteins and HA-tagged TAF145⌬TAND proteins migrated at the same position on this gel. When the signals were normalized with the amount of precipitated TBP (lowest panel), it appeared that similar amounts of TAF145 and TAF61 proteins were included in the TFIID complex in all yeast strains. In contrast, smaller amounts of TAF145⌬TAND proteins were coprecipitated with TBP in the N159D mutant. Such differences were not observed for other TBP mutants. However, it is still possible that even the N159D mutant can make a stable TFIID complex with the TAF145⌬TAND protein if the wild type TAF145 protein is not present in the cell, although we cannot test this possibility directly due to its synthetic lethality. Collectively, the instability of the TFIID complex does not simply explain why these TBP mutants display the nsl phenotypes.

DISCUSSION
In this study, we screened nsl genes that have genetic interaction with TAND of TAF145. The NSL1 gene isolated in our screen was found to be allelic to the SPT15 gene encoding TBP. We also identified the amino acid substitutions, S118L and P65S, for the two spt15/nsl1 alleles. They are recessive in both the nsl and TS phenotypes in different general genetic backgrounds. Previous studies demonstrated that the S118L mutant was deficient in activation by acidic activators (63); however, it is unknown whether the P65S mutant has similar defects or not. On the other hand, the P65S mutant was reported to be defective in pol III transcription (64) although the S118L mutant seems to suffer only pol II-specific defects. Since TAF145 is a pol II-specific general transcription factor, it is likely that the P65S mutant is also damaged in activation by acidic activators like the S118L mutant; we found that this is the case. Intriguingly, all of the other activation-defective TBP mutants that we tested also showed synthetic lethal interactions with the taf145⌬TAND gene. In contrast, similar effects were not observed for TBP mutants that were specifically impaired in pol III transcription. Further characterization of the biochemical properties of these activation-defective TBP mutants revealed that enhanced interaction with VP16 appeared to be closely related to the nsl phenotype (Table III). In addition, an artificial recruitment assay produced further evidence for the good correlation between the defect in the post-recruitment step and the nsl phenotype (Table III).
We recently proposed that TAND might be involved in the initial stage of activation by acidic activators (32). Some functional similarities have been found between TAND1 (that is the N-terminal subdomain of TAND that binds to the concave surface of TBP) and acidic activation domains like VP16, GAL4, and EBNA2 (32). Such unexpected functional similarities prompted us to build a two-step hand off model in which TAND1, bound to the concave surface of TBP, could first be displaced by an acidic activation domain (AD) and the AD could be successively displaced by the TATA element (32). We believe that such transfer of TBP from TAND to the TATA element may trigger the isomerization process of TFIID that leads to increased stimulation of transcription. According to this model, TAND must play two different roles, i.e. when activators are absent near the promoter, TAND inhibits TBP-TATA interactions to prevent leaky transcription, whereas it must be able to release TBP once activators come close to the promoter. Taken that TFIID can recognize the core promoter, not only by TBP-TATA interactions but also by TAF-DNA interactions (reviewed in Refs. 8 and 7), TBP might be converted into an active form at a much closer position to the TATA element when it is liberated from TAND by the action of activators. If this is the case, TAND performs dual functions as a negative and positive regulator. Previous studies demonstrated that the TS phenotype of taf145⌬TAND strains could be suppressed by overexpression of TBP and TFIIA (30,31). The assumption described above may explain why TFIID that lacks TAND needs higher concentrations of TBP or TFIIA to perform its normal function.
Our studies demonstrate that activation-defective TBP mutants confer severe damage to yeast strains carrying the taf145⌬TAND gene. It is especially intriguing that TBP mutants defective in the post-recruitment step produce a stronger nsl phenotype (Table III). As described above, it is believed that TAND is involved in the initial step during activation, i.e. stable binding of TFIID to the promoter (pre-recruitment step). The combined defects of the pre-and post-recruitment steps can be expected to yield the lowest activation and thereby inhibit cell growth. In this regard, the results obtained here support our previous model that TAND is involved in prerecruitment steps.
Another issue to be discussed is how these TBP mutants are impaired in the post-recruitment step. TBP may function together with several other transcription machineries besides TFIID, e.g. SAGA and RNA pol II holoenzyme (reviewed in Ref. 23). Thus, it is possible that these TBP mutants affect the function of other machineries involved in post-recruitment FIG. 5. Coimmunoprecipitation analysis to test the integrity of the TFIID complex containing both TAF145⌬TAND protein and TBP derivatives. Whole cell extracts were prepared, using glass beads, from yeast strains containing the wild type gene or the indicated TBP derivatives and the HA-tagged TAF145⌬TAND proteins. Note that these strains have the nontagged wild type TAF145 gene on the chromosome to support viability. Aliquots of whole cell extract proteins were immunoprecipitated with anti-TBP polyclonal antibodies (even-numbered lanes) or preimmune antibodies (odd-numbered lanes). Proteins coprecipitating with TBP were fractionated on SDS-PAGE, transferred to nitrocellulose membranes, and probed with the indicated antibodies.
steps. Although we have not yet examined the integrity of those machineries, most of these TBP mutants appear to be normal at least in the integrity of TFIID. However, it is still likely that TFIID may be impaired in its interactions between TBP-TAF or TBP-TFIIA that are required specifically for the post-recruitment steps. To determine which complex is most severely damaged in these TBP mutants, we need to separate each complex and test its activity individually using in vitro transcription experiments.
Recently, TBP has been shown to exist, at least in part, as an independent form of TAFs (21,22). TAFs are recruited in much smaller amounts to TAF-independent (TAFind) promoters when compared with TAF-dependent (TAFdep) promoters. Consistently, TBP is recruited, in a manner apparently independent of TAF function, to the TAFind promoters. Conversely, TAFs are recruited independently of TBP function to TAFdep promoters. These observations strongly suggest that there are at least two transcriptionally active forms of TBP, i.e. associated form and nonassociated form with TAFs (21,22). Thus it might be possible that activation-defective TBP mutations affect the TAF-nonassociated form of TBP predominantly, whereas TAF145⌬TAND affects only the TAF-associated form of TBP, so that double mutants can decrease the expression of a much broader range of genes. It has yet to be determined whether these alternative forms of TBP, other than TFIID, correspond to the free TBP molecule or any other known or unknown TBP-containing complexes.
It is still unclear whether TAND is involved in the activation of most genes or just a particular set of genes in vivo. Earlier studies demonstrated that activation may occur even in the absence of TAFs in vitro as well as in vivo (80,81). More recent genome-wide expression analyses have shown that the requirements for various TAFs are not the same among different genes probably because some TAFs are shared by TFIID and SAGA, which are functionally redundant (reviewed in Refs. [82][83][84]. Consistently shared TAFs such as TAF17, TAF25, and TAF60 appear to be more generally required for gene expression than TFIID-specific TAFs such as TAF145 and TAF150 (84). However, there are exceptions like TAF40, which is a TFIID-specific TAF but nonetheless required by most promoters (85). Since our preliminary genome-wide expression analysis showed that TAF145⌬TAND produced much smaller effects, 3 we believe that the possible detrimental effect of TAF145⌬TAND on gene expression might be concealed in vivo by compensatory interactions between TFIID and other machineries like SAGA, as just recently observed for TAF145 and Gcn5 (84). TAF145 and Gcn5, both of which encode catalytic subunits of histone acetyl-transferase in TFIID and SAGA, appear to regulate the expression of a large fraction of genes through the function of either complex. In this regard, it would be intriguing to examine the redundancy between TAND and Spt3/Spt8 that are analogous subunits, which negatively regulate TBP function in each complex. Isolation of other NSL genes could solve the functional redundancies in vivo that are accomplished by multiple TBPinteracting factors.