Impairment of the DNA Binding Activity of the TATA-binding Protein Renders the Transcriptional Function of Rvb2p/Tih2p, the Yeast RuvB-like Protein, Essential for Cell Growth*

In Saccharomyces cerevisiae , two highly conserved proteins, Rvb1p/Tih1p and Rvb2p/Tih2p, have been demonstrated to be major components of the chromatin-remodeling INO80 complex. The mammalian orthologues of these two proteins have been shown to physically associate with the TATA-binding protein (TBP) in vitro but not clearly in vivo . Here we show that yeast proteins interact with TBP under both conditions. To assess the functional importance of these interactions, we examined the effect of mutating both TIH2/ RVB2 and SPT15 , which encodes TBP, on yeast cell growth. Intriguingly, only those spt15 mutations that affected the ability of TBP to bind to the TATA box caused synthetic growth defects in a tih2-ts160 background. This suggests that Tih2p might be important in recruiting TBP to the promoter. A DNA microarray tech-nique was used to identify genes differentially expressed in the tih2-ts160 strain grown

Nucleosomes are the primary components of chromatin, and their structure and position can play a crucial role in determining eukaryotic transcription (1,2). Certain structural configurations of chromatin have been shown to repress transcription by blocking the access of the transcriptional apparatus to its target promoters (1)(2)(3)(4). Conversely, DNA unwinding and reconfiguration of nucleosomal structures are thought to allow transcription factors to bind to their sites, thereby facilitating transcription initiation (1)(2)(3)(4). Various factors and multiprotein complexes have been identified that regulate transcription by remodeling the structure of the nucleosome, and some of these have been shown to participate crucially in the transcriptional regulation of particular sets of genes (3,4). These factors and complexes require ATPase activity and/or function by covalently modifying histones through acetylation, methylation, or phosphorylation (4 -6). Alteration of chromatin structure is an important step not only for transcriptional regulation but also for DNA repair, recombination, and replication (7,8). Recent studies show that individual chromatin-modifying complexes are apparently involved in multiple DNA processing reactions (9 -13). This includes the INO80 chromatin remodeling complex in Saccharomyces cerevisiae, which appears to be involved in both transcription and DNA repair (9). This complex contains the Ino80p ATPase, which belongs to the SWI2/ SNF2 superfamily, and two ATP-dependent DNA helicases called RuvB-like protein 1 (Rvb1p/Tih1p) and RuvB-like protein 2 (Rvb2p/Tih2p), which share homology with the prokaryotic DNA helicase RuvB. RuvB acts on the process of branch migration at Holliday junctions at the late stages of homologous recombination in DNA repair (14,15). Consistent with this, ino80 mutants are susceptible to agents that cause DNA damage (9). The Tip60-containing complex has been shown to contain the human orthologues of Rvb1p/Tih1p and Rvb2p/ Tih2p, which have also been designated as Tip60-associated protein 54␣ (TAP54␣) 1 and TAP54␤, respectively (10). This human complex can bind to structural DNA that mimics Holliday junctions. Importantly, ectopic expression of a mutant Tip60 lacking its histone acetyltransferase activity causes human cells to become defective in double-strand DNA break repair (10). These observations indicate that the multiple roles of individual chromatin-modifying complexes in various DNA processing reactions may be evolutionarily conserved from yeast to man.
A number of studies have demonstrated that Rvb1p/Tih1p and Rvb2p/Tih2p play a crucial role in transcription. First, the rat orthologue of Rvb1p/Tih1p was originally identified as a TATA-binding protein (TBP)-interacting protein and was therefore denoted as TIP49a (16). Second, the human orthologue of Rvb1p/Tih1p was shown to bind to the RNA polymerase II holoenzyme complex (17). In this study, the orthologue was denoted as RUVBL1 (RuvB-like protein 1). Notably, the mammalian orthologues of Rvb1p/Tih1p and Rvb2p/Tih2p have been discovered independently many times and consequently have a variety of designations, including Pontin52/TIP49/ TIP49a/RUVBL1/ECP-51/TAP54␣ and Reptin52/TIP48/ TIP49b/RUVBL2/ECP-54/TAP54␤, respectively. Third, both of these proteins can bind to transcriptional regulatory factors, e.g. ␤-catenin and c-Myc, and influence their function either positively or negatively (18 -21). Furthermore, certain TIP49-TIP48-BAF53-containing complexes were shown to be essential for c-Myc-mediated transformation (21,22). Although these three proteins were originally identified as components of the Tip60-containing complex described earlier (10), it is possible they can also function distinct from Tip60, for example, as part of a p400 complex that is an essential E1A transformation target (22,23). Notably, BAF53 was also found to be a component of the mammalian SWI/SNF-related chromatin-remodeling complex. Fourth, TIP49 was recently shown to bind to the E2F1 transactivation domain and to modulate its apoptotic activities (24). Finally, we and others (25,26) have demonstrated that yeast Rvb1p/Tih1p and Rvb2p/Tih2p are required for the transcription of at least a subset of genes.
A number of studies have suggested that Rvb1p/Tih1p and Rvb2p/Tih2p and their mammalian orthologues possess diverse functions, in addition to transcriptional regulation (we hereafter will refer to Rvb1p/Tih1p and Rvb2p/Tih2p as Tih1p and Tih2p and their mammalian orthologues as TIP49a and TIP49b, respectively). Indeed, these proteins appear to be involved in various apparently distinct biological events, although their precise functions are not completely understood (27)(28)(29)(30)(31)(32)(33). As a first step to clarify the precise role of Tih1p and Tih2p in transcription, we studied their functional interaction with TBP using biochemical and genetic approaches. Here we show for the first time that these proteins clearly form a complex with TBP in vivo. Significantly, the combination of a temperature-sensitive mutation in the TIH2 gene (tih2-ts160) with mutations in the gene encoding TBP (SPT15) defective for TATA binding resulted in synthetic growth defect. Some of the Tih2p target genes that we identified by DNA microarray analyses using a tih2-ts160 strains were also specifically affected in these spt15 mutants. These observations strongly argue that the cooperative and functional interactions between Tih1p/ Tih2p and TBP are crucial for the transcription of at least a subset of yeast genes.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Plasmid Constructions-Yeast strains used in this report are listed in Table I. Rich medium (YPD) and synthetic complete (SD) medium with 5-fluoro-orotic acid (5-FOA) and appropriate nutrients were prepared as described previously (34). Standard methods were used to genetically manipulate yeast cells as described previously (34,35). Details of primers used in this study are available on request. The TIH1 locus of genomic DNA was amplified by PCR with primers that introduce XhoI-NotI sites at each end. The XhoI/NotI-digested 2-kb DNA fragment was then ligated into the pRS316 and pRS314 vectors to obtain pRS316-TIH1 and pRS314-TIH1, respectively. Similarly, the EcoRI-NotI genomic DNA fragment (2.3 kb) containing the TIH2 locus was amplified by PCR with primers that create XhoI-NotI sites at each end. The XhoI/NotI-digested 2.3-kb DNA fragment was then ligated into pRS314 and pRS315 vectors to obtain pRS314-TIH2 and pRS315-TIH2, respectively. Likewise, pRS315-tih2-ts160 was constructed by PCR with primers that create XhoI-NotI sites at each end.
Tih1p and Tih2p constructs with the native promoters were tagged with a triple hemagglutinin (HA) epitope tag at their amino-terminal and carboxyl-terminal ends, respectively, using fusion PCR methods (36). Primers were designed to introduce EcoRI and BamHI sites at the 5Ј-and 3Ј-ends of the triple HA-tag, respectively. The final PCR products were subcloned as ϳ2.1and 2.4-kb XhoI/NotI-digested fragments into the pRS314 vector to generate pRS314-HA-TIH1 and pRS314-TIH2-HA, respectively. Similarly, TBP with a triple FLAG epitope tag at its amino-terminal end was also constructed by the fusion PCR method. The final PCR product was subcloned as a ϳ2.4-kb EcoRI/NotI-digested fragment into the pRS313 vector to generate pRS313-FLAG-TBP. The TRP1-marked plasmid carrying TBP or its derivatives and the hexahistidine-tagged TBP expression vector for bacterial cells were described previously (37). pGEX-3X-TIH1 and pGEX-3X-TIH2 plasmids were constructed by inserting the TIH1 or TIH2 ORFs amplified by PCR into the BamHI-EcoRI sites of pGEX-3X (Amersham Biosciences).
The TIH1 knockout plasmid, pDisTIH1, was constructed by PCR amplification of the 5Ј-and 3Ј-flanking DNA (ϳ1 and 0.8 kb, respectively) of the TIH1 ORF, with the addition of the appropriate restriction sites at each end, and insertion into the pRS305 vector. To replace the entire TIH1 ORF with the LEU2 gene on the genome, pDisTIH1 was linearized with BamHI and transformed into a diploid strain (FY strain) (25). Disruption was confirmed by Southern blotting. After transformation of pRS316-TIH1 into this strain, the Ura ϩ Leu ϩ haploid strain was selected by tetrad analysis as the parental strain for plasmid shuffling. The YHO1 and YHO2 strains were subsequently obtained from this parental strain by plasmid shuffling (Table I). The YHO4 and YHO5 strains were obtained as follows. A Ura ϩ Leu ϩ His Ϫ haploid was selected by tetrad analysis, and this strain was transformed with pRS314-TIH1 or pRS314-HA-TIH1 and then grown on 5-FOA plates to remove pRS316-TIH1. These strains, expressing TIH1 or HA-TIH1, were further successively transformed with pRS313-FLAG-SPT15 and the hisG cassette plasmid for SPT15 gene disruption (37). The URA3 gene sandwiched by the two hisG sequences was removed by growth on 5-FOA plates to generate the YHO4 and YHO5 strains, respectively. Disruption of the SPT15 gene was confirmed by Southern blotting. The YHO3 strain, which was deleted for the TIH2 gene, was obtained from the parental CRPA1 strain (25) by plasmid shuffling in a manner similar to YHO1 and YHO2.
To obtain the parental strain containing double deletions of the chromosomal TIH2 and SPT15 genes, for subsequent plasmid shuffling, a diploid strain was created that had a heterozygous disruption of the TIH2 gene by insertion of the HIS3 gene. This strain was transformed with the hisG cassette plasmid for SPT15 gene disruption (37). Disruption was confirmed by Southern blotting after removing the URA3 gene situated in between the hisG sequences by growth on 5-FOA plates. This strain therefore bears heterozygous disruptions of the TIH2 and SPT15 genes and was further transformed with URA3 and LYS2marked plasmids carrying the SPT15 and TIH2 genes, respectively, and subsequently sporulated and dissected to obtain haploid strains containing double deletions of the chromosomal TIH2 and SPT15 genes. Most of the other strains listed in Table I were obtained from this parental strain by plasmid shuffling using 5-FOA and ␣-aminoadipic acid to counterselected against the URA3 and LYS2 genes, respectively. The ⌬ino80 strain is a generous gift from Carl Wu (National Institutes of Health). All strains are of the S288C background.
Whole Cell Extract Preparation and Immunoprecipitation-Procedures were based on methods described previously (25) with several modifications. Yeast strains incubated in 20 ml YPD at 30°C (A 600 ϭ 1) were washed once with lysis buffer (50 mM Tris-HCl, pH 8.0, 5 mM EDTA) and resuspended in 200 l of lysis buffer with protease inhibitors (10 g/ml each of aprotinin, leupeptin, pepstatin A, and 1 mM phenylmethylsulfonyl fluoride), and cell lysates were extracted by glass beads. The lysates (175 l) were preincubated for 30 min at 4°C in four volumes of immunoprecipitation buffer (1.25% Triton X-100, 180 mM NaCl, 6 mM EDTA, 60 mM Tris-HCl, pH 8.0, 6% skim milk) containing 10 l of protein A-Sepharose beads. The supernatants were collected, and 0.5 l of anti-TBP polyclonal antibodies or preimmune antibodies were added and incubated on a rotator overnight at 4°C. 15 l of protein A-Sepharose was then added, followed by incubation for 1 h at 4°C. The beads were then washed three times with wash buffer (50 mM Tris-HCl, pH 8.0, 1% Triton X-100, 150 mM NaCl, 5 mM EDTA), and 25 l of SDS sampling buffer was added. The samples were boiled for 5 min and fractionated on 10% (for HA-Tih1p and HA-Tih2p) or 12% (for TBP) SDS-polyacrylamide gels and then blotted onto nitrocellulose membranes. HA-Tih1p/Tih2p and TBP were detected with the monoclonal anti-HA 12CA5 antibody (Roche Molecular Biochemicals) and the polyclonal anti-TBP antibody, respectively. For the reciprocal immunoprecipitation analysis, the monoclonal anti-HA 12CA5 antibody was used, and detection was performed with the monoclonal anti-HA 12CA5 and anti-FLAG antibodies (Sigma).

Physical and Genetic Interaction between TBP and TIH2/RVB2
A, and 1% Triton X-100. After sonication and centrifugation, cell debris was removed by filtration. The cell lysate was subjected to TALON metal affinity resin (Clontech) to purify TBP. Control GST, GST-Tih1p, and GST-Tih2p were expressed in E. coli BL21-codonPlus, (DE3)-RIL. After incubation with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h at 25°C in 500 ml of 2ϫ YT medium, the cells were collected and resuspended in 40 ml of phosphate-buffered saline with protease inhibitors and detergent and sonicated to collect the cell extracts as described above. GST and GST fusion proteins were purified on glutathione-Sepharose 4B (Amersham Biosciences). The GST pulldown assay was performed as follows. 3 g each of GST, GST-Tih1p, or GST-Tih2p and 3 g of TBP were incubated with a 10-l bed volume of glutathione-Sepharose in 1 ml of binding buffer (50 mM Tris-HCl, pH 8.0, 1% Triton X-100, 150 mM NaCl, 5 mM EDTA) for 1.5 h at 4°C. Beads were washed extensively with binding buffer and then boiled in 20 l of SDS sampling buffer. The eluted proteins were fractionated on 12% SDS-polyacrylamide gel, blotted onto a nitrocellulose membrane, and probed with the polyclonal anti-TBP antibody. Recombinant proteins were quantified by the BCA protein assay kit (Pierce) and/or Coomassie Brilliant Blue staining on the gel. Synthetic Lethal Assay-YHOTM1-13 and their wild type strains (Table I) were grown in liquid YPD medium at 23°C to the early log phase and then spotted on 5-FOA plates and incubated at 29°C for an additional 5 days. Likewise, YHOTW and YHOTD1-YHOTD3 strains were grown in liquid YPD medium to the early log phase at 23°C and then spotted on YPD plates and incubated at 32°C for an additional 4 days.
Microarray Analysis-Wild type (CRH6) and thi2-ts160 (CRH8) strains were grown in YPD medium at 23°C to an A 600 of ϳ0.4. Preheat shock treatment was performed by incubating the cells at 34°C for 15 min (38). Cells were incubated at 23°C for 45 min and then shifted to 34°C for an additional 45 min of incubation before being harvested. Total RNA was extracted by the hot phenol method (25), and poly(A) ϩ RNA was prepared by using an oligo(dT) column (Amersham Biosciences). Microarray analysis was performed as described in Ref. 39 using Yeast Chip, version 1.0 (Hitachi Software Engineering, Yokohama, Japan). 2 Briefly, fluorescently labeled cDNA was made from 1 g of poly(A) ϩ RNA in a reaction volume of 20 l containing 0.5 g of oligo-(dT) 18-mer, 2 l of a dNTP mixture mix (consisted of 5 mM each of dATP, dCTP, and dGTP and 2 mM dTTP), 2 l of 1 mM Cy3-or Cy5-conjugated dUTP, and 400 units of Superscript II reverse transcriptase (Invitrogen). The two labeled cDNA pools were combined and competitively hybridized to a microarray under a coverslip for 16 h at 65°C. Experiments were done in duplicate using different batches of template mRNA prepared under the same experimental condition. Fluorescent array images were obtained for Cy3 and Cy5 emissions by using a ScanArray Lite (PerkinElmer Life Sciences) scanner. Image intensity data were analyzed by using QuantArray 3.0 (PerkinElmer Life Sciences) software. Statistical analyses were carried out using Microsoft EXCEL. Spots that have mean intensity values of less than 1000 arbitrary units were discarded. Expression ratio (Cy3 (or Cy5) signal intensity/Cy5 (or Cy3) signal intensity) was calculated and expressed as a logarithmic value. The average from duplicated experiments was taken as the final expression ratio. Highly irreproducible data, as judged from S.D., were discarded.
RNA Preparation and Northern Blot Analysis-RNA preparation and Northern blot analysis were performed as described (25). Strains grown in YPD medium at 23°C to the early exponential phase (A 600 , ϳ0.2) were shifted to 34°C and harvested at the indicated time. 10 g of each RNA was subjected to electrophoresis in a 1% formaldehyde/ agarose gel, followed by transfer to a nylon membrane. All probes used for hybridization were generated by PCR amplification using genomic DNA as template and labeled using a random primer labeling kit (TaKaRa).

Tih1p and Tih2p
Interact with TBP in Vivo-TIP49a was originally purified as a TBP-interacting protein from rat nuclear extract in vitro (16). More recently, GST pulldown experiments showed that both TIP49a and TIP49b proteins could directly bind to TBP (18,19). Nonetheless, the in vivo interaction between TIP49a/TIP49b and TBP had not been clearly established, as their interaction in the rat nuclear extracts appears to be quite weak (16), and TBP has consistently failed to be identified as a component in several TIP49a/TIP49bcontaining complexes (10,21,23). To examine whether Tih1p and Tih2p, the yeast orthologues of TIP49a and TIP49b, respectively, could complex with TBP in vivo, immunoprecipitation (IP) analyses were performed using cell lysates prepared from yeast strains expressing Tih1p tagged at the amino terminus or Tih2p tagged at the carboxyl terminus. Immunoprecipitates obtained with the polyclonal anti-TBP antibody or preimmune antiserum were subsequently probed with the anti-HA antibody (Fig. 1A, upper panel). Both Tih1p and Tih2p proteins were found to be specifically coprecipitated with endogenous TBP (Fig. 1A, compare upper and lower panels). The same results were obtained irrespective of the position or type of epitope tag used (data not shown). To further confirm the interaction between Tih1p/Tih2p and TBP, we conducted reciprocal IP experiments (Fig. 1B). Lysates prepared from yeast strains expressing FLAG-tagged TBP and either HA-tagged Tih1p or Tih2p were precipitated with anti-HA antibody and subsequently probed with anti-FLAG antibody (Fig. 1B, lower  panel). A single specific immunoreactive band corresponding to FLAG-tagged TBP was detected on the blot of samples from 2 C. R. Lim, A. Fukakusa, and K. Matsubara, submitted for publication.
Precipitates with anti-HA or preimmune antibodies were fractionated, blotted, and probed with the antibodies indicated on the right side of the blot. 5.7 and 0.1% input proteins were loaded in the upper and lower panels, respectively (lanes 1 and 2 and lanes 5 and 6). cells coexpressing HA-tagged Tih1p or Tih2p (Fig. 1B, lanes 4  and 8). These observations indicate that at least some of the Tih1p/Tih2p and TBP molecules form a complex in yeast cells.
Tih1p and Tih2p Interact Directly with TBP in Vitro-Previous studies showed that mammalian orthologues of Tih1p/ Tih2p could bind to TBP directly (18,19). We therefore conducted GST pulldown experiments to examine whether yeast Tih1p/Tih2p could also bind directly to TBP in vitro (Fig. 2). Recombinant TBP was mixed with the same amounts of GST-Tih1p, GST-Tih2p, or GST alone (Fig. 2, lower panel). After incubation, complexes bound to the glutathione-Sepharose 4B beads were analyzed by Western blotting with anti-TBP antibody. The bands corresponding to TBP were detected only when recombinant TBP was mixed with GST-Tih1p or Tih2p (Fig. 2, upper panel). Thus, yeast Tih1p and Tih2p can bind to TBP directly.
TIH2 and SPT15, Encoding TBP, Interact Genetically-Our observation that Tih1p and Tih2p can interact with TBP both in vitro and in vivo implies that these proteins may function in transcriptional regulation as a single entity. In this respect, it is intriguing that the temperature-sensitive tih2-ts160 mutant we isolated previously (25) exhibited phenotypes similar to those of taf1 mutants (25). The TAF1 gene encodes a TFIID subunit that directly associates with TBP (40). Both mutants were defective in the transcription of genes encoding the ribosomal proteins and G 1 cyclins and exhibited a G 1 cell cycle arrest phenotype (25,(41)(42)(43). In addition, an allele of taf1 that lacks the TBP binding domain acts as a synthetic lethal in the presence of various mutants of spt15, the gene encoding TBP (37). Interestingly, these TBP mutants were defective specifically in the post-recruitment step of TBP to the core promoter (37). Hence, we reasoned that the tih2-ts160 allele might also exhibit synthetic growth defects when combined with various mutants of spt15.
We constructed yeast strains lacking both the TIH2 and SPT15 genes and containing a URA3-marked plasmid encoding wild type TBP and a LEU2-marked plasmid encoding either the wild type TIH2 gene or the tih2-ts160 gene. These strains were subsequently transfected with TRP1-marked plasmids encoding wild type TBP or with mutant TBPs that are defective in various TBP functions (37, 44 -47). The location of the mutation site of each TBP derivative is indicated on TBP crystal structure shown in Fig. 3A (48). The synthetic growth defects resulting from the combined expression of these mutant TBPs and the tih2-ts160 allele was assessed by examining growth rates on 5-FOA plates at 29°C (Fig. 3B). We observed severe synthetic growth defects with the TBP-V161A and TBP-N159D mutants, whereas TBP-N159L and TBP-S118L showed weak but discernible defects (Fig. 3B). Intriguingly, all of these TBP mutants are impaired in their ability to bind to the TATA element (44,47). We observed no synthetic growth defects with any of the remaining TBP mutants (Fig. 3B), which are apparently impaired in other functions (37,47). This suggests that the TATA binding activity of TBP is absolutely essential for yeast cell growth when TIH2 function is impaired. Note that these synthetic growth defects were not observed at lower temperatures (e.g. at 23°C; data not shown), possibly because

FIG. 3. Genetic interaction between the TIH2 and SPT15
genes. Mapping of the each TBP derivatives on the crystal structure of carboxyl-terminal domain of yeast TBP is shown. TBP has four ␣ helices (H1, H2, H1 Ј , and H2 Ј ) and ten ␤ strands (S1 to S5 and S1 Ј to S5 Ј ). DNA interacts with ␤ strands on the concave surface of TBP (DNA surface). B, TRP1-marked plasmids encoding TBP or its derivatives as indicated on the left were individually introduced into a strain lacking both the TIH2 and SPT15 genes and containing a URA3-marked plasmid encoding wild type (WT) TBP, as well as with a LEU2-marked plasmid encoding either the wild type TIH2 gene or the tih2-ts160 gene, as indicated on top. The resulting transformants were grown on 5-FOA plates at 29°C for 5 days. C, the transformants described for A were grown on 5-FOA plates at 23°C for several days to remove the URA3marked plasmid expressing wild type TBP. The strains expressing wild type or mutant TBP from TRP1-marked plasmids were then grown on YPD plates at 32°C for 4 days. TATA binding activity is partially restored. In addition, on YPD plates, only those spt15 alleles having the strongest phenotypes, such as spt15-V161A and spt15-N159D, exhibited synthetic growth defects (Fig. 3C). This suggests that rich medium may lower the requirement for TATA binding activity of TBP in cell growth and transcriptional regulation. Similar defects were observed when yeast strains were cultured in liquid YPD media (Fig. 3D).
Thus, different synthetic phenotypes are observed when spt15 mutations are combined with either taf1 mutant alleles lacking the TBP binding function (37) or with tih2 mutants. This suggests that distinct functions of TBP are essential for cell growth depending on which gene, i.e. TAF1 or TIH2, is mutated.
Genome-wide Analysis of Gene Expression in the tih2-ts160 Mutant-Both Tih1p and Tih2p participate in a chromatinremodeling complex composed of about 12 distinct polypeptides, including Ino80p, a SWI2/SNF2 superfamily protein (9). However, like their mammalian orthologues (10,22,23), Tih1p and Tih2p may also be part of other distinct multiprotein complexes, as suggested by Wu and co-workers (9). Supporting this idea is the observation that the TIH1 and TIH2 genes are essential for yeast cell growth, whereas the INO80 gene is not (9). Furthermore, in a study of genome-wide gene expression, Dutta and co-workers (26) found that depletion of Tih1p or Tih2p using the temperature-inducible N-degron system affected expression of about 5% of yeast genes in cells grown in galactose-containing media. This system should result in the total degradation of the target proteins and thus should disrupt any complex containing Tih1p or Tih2p.
The tih2-ts160 allele affects the function but not the stability of the TIH2 protein (25). In good agreement with this notion, the G 1 cell cycle arrest phenotype observed with the tih2-ts160 allele is completely reversible (25). Thus, we reasoned that if this mutant is shifted to the restrictive temperature, we might observe a different genome-wide pattern of gene expression profile from that obtained by Dutta and co-workers (26). We chose to cultivate the yeast strains in YPD medium, containing glucose, as this would allow us to compare our results with those of other genome-wide studies (49 -51), which have mostly been performed in YPD. We hoped that identification of the target genes for Tih2p by DNA microarray experiments would allow the comparison of their expression profiles in various mutants and help to elucidate some Tih2p functions. For instance, we might be able to determine why synthetic growth defects arise from a particular set of TBP mutants when combined with mutant Tih2p. It would also be valuable to investigate how broadly the target genes of Tih2p overlap with those of Ino80p.
To address these fundamental questions, we conducted DNA microarray experiments using RNA prepared from wild type and tih2-ts160 mutant strains 45 min after shifting the temperature from 23 to 34°C. Each RNA preparation was labeled with Cy3 or Cy5 fluorescent dye and then hybridized to a slide glass on which partial DNA fragments corresponding to about 6000 different genes had been spotted (see "Experimental Procedures"). Only 34 genes were significantly and reproducibly affected (i.e. changed more than 2.5-fold) in the tih2-ts160 mutant at the restrictive temperature (see Table II and Supplemental Material). Of these, 20 genes showed a decrease in expression, and 14 genes showed an increase. It is noteworthy that transcription of the PHO and VTC genes, which are involved in phosphate metabolism, decreased markedly in this mutant. Some of these genes had been identified previously (9, 49 -52) to be targets of the SAGA, TFIID, SWI/SNF, NuA4, and INO80 complexes. Thus, the expression of at least some of these PHO and VTC genes appear to require the integral function of multiple histone acetyltransferase and chromatin remodeling complexes. In addition, only two of the target genes we identified here were also identified by Jonsson et al. (26), viz. MEP2 and DLD3 (26). Among the most significantly affected genes, no other genes were represented on both lists. Although the exact number of overlapping target genes between the two studies is not certain, they presumably correspond to target genes that are regulated by Tih2p irrespective of the carbon source in the media.
Comparison of the Transcriptional Defects of the tih2-ts160 and ino80 Null Mutants-As mentioned, although Tih1p/Tih2p and Ino80p are components of the INO80 complex, it is conceivable that these proteins also function in other multiprotein complexes. If this is the case, the transcriptional defects resulting from mutations in these factors might differ. To explore this possibility, the transcription of several Tih2p target genes in wild type, tih2-ts160, and ino80 null mutant strains were compared by Northern blotting (Fig. 4). PHM2, SPL2, PHO5, and MAE1 were selected as representatives of genes showing de-creased transcription in the tih2-ts160 mutant (Fig. 4A), whereas MEP2, GDH1, and DLD3 were selected as genes showing increased transcription (Fig. 4B). As a control, the transcription of the ADH1 and ACT1 genes, which are not affected by the tih2-ts160 allele, were also examined. Quantification of each band is shown below the blot as the ratio relative to the amount of mRNA recovered from the wild type strain at the time of the temperature shift. Intriguingly, all of the genes found to decrease in the tih2-ts160 background also decreased in the ino80 null mutant. In contrast, MEP2 and GDH1 did not increase at all in the ino80 null mutant, although DLD3 increased slightly. These results indicate that the target genes of Tih2p and Ino80p overlap but are not entirely identical. However, further analysis is needed to determine whether the differential effects of Tih2p and Ino80p on the transcription of the MEP2 and GDH1 genes are because of functionally different defects of a single complex (e.g. the INO80 complex) or defects in two or more separate but as yet unidentified complexes.
Transcription of Tih2p Target Genes Was Specifically Affected in the spt15 Mutants That Display Severe Synthetic Growth Defects When Combined with tih2-ts160 -To investigate why the TBP mutants that lack TATA binding activity show synthetic growth defects when combined with the tih2-ts160 allele, we used Northern blotting to measure the transcription of several Tih2p target genes in various spt15 mu- FIG. 4. Comparison of transcription of several Tih2p target genes in the wild type, tih2-ts160, and ino80 null mutant strains. A, total RNA was isolated from wild type (WT), tih2-ts160, and ⌬ino80 strains at the indicated time after shifting the temperature from 23 to 34°C. 10 g of total RNA prepared from each strain was subjected to Northern blot analysis. DNAs corresponding to the down-regulated genes in the tih2-ts160 mutant indicated on the left were labeled as probes. The amount of each transcript was quantified in duplicate. The average amount is shown below the blot relative to the amount of mRNA obtained from the wild type strain just at the point of the temperature shift. One of the duplicate experiments is shown here as a blot. B, the up-regulated genes in the tih2-ts160 mutant were examined as described for A.
tants after shifting the temperature from 23 to 34°C (Fig. 5). We found earlier that although temperature shift (i.e. 34°C) restricted the growth of the tih2-ts160 mutant, they did not affect the growth of any of the TBP mutants we tested here. However, combination of some of the TBP mutants with tih2-ts160 were lethal even at the permissive temperature (Fig. 3). FIG. 5. Comparison of transcription of several Tih2p target genes in the wild type, spt15, tih2-ts160, and spt15/ tih2-ts160 double mutant strains. A, total RNA was isolated from the wild type (WT), spt15-V161A, tih2-ts160, and spt15-V161A/tih2-ts160 double mutant strains at the indicated time after shifting the temperature from 23 to 34°C. Northern blot analysis was conducted as described for Fig. 4 with the probes indicated on the left. The amount of each transcript was quantified as described for Fig. 4. B, total RNA was isolated from strains whose SPT15 and TIH2 alleles are indicated on the left and the top, respectively, 45 min after shifting the temperature from 23 to 34°C. Northern blot analysis was conducted as described for Fig. 4 with the probes indicated on the bottom. C, summary of the results of mRNA quantification obtained in B. The numbers on the x axis represent the spt15 alleles as indicated in B. Black and white bars represent the wild type and mutant (ts160) alleles of the TIH2 gene, respectively.
It is possible that these TBP mutants are impaired in functions that are shared with Tih2p and defective in tih2-ts160. To investigate this possibility we examined transcription of the SPL2 and PHO84 genes, both of which are Tih2p target genes whose transcription is severely reduced in the tih2-ts160 strain. We found that transcription of both genes were almost abolished in the spt15-V161A mutant, as well (Fig. 5A). In addition, transcription of the PHO5 gene, which was decreased in the tih2-ts160 mutant, albeit to a lesser extent than SPL2 and PHO84, was also reduced in the spt15-V161A mutant (Fig.  5A). In contrast, the MEP2 and GDH1 genes, whose expression increased in tih2-ts160, was unchanged in the spt15-V161A mutant (Fig. 5A). Additional transcription defects were not observed when these two mutations were combined (V161A x ts160; see Fig. 5A). Taken together, these observations suggest that Tih2p can have both stimulatory and inhibitory effects on transcription, both of which are affected by the ts160 mutation, and that the stimulatory function overlaps with that of the TBP and is impaired in the spt15-V161A mutant.
Other TBP mutations that affect TATA binding also induced synthetic growth defects when combined with the tih2-ts160 allele (Fig. 3). To examine the relationships among TATA binding, transcription, and the consequent growth phenotypes of these mutant alleles, we examined the transcription of five genes (PHM2, SPL2, PHO5, GDH1, and ADH1) in wild type strains versus strains harboring one of the eleven spt15 mutant alleles, the tih2-ts160 allele, or both. Gene expression was analyzed 45 min after shifting to the restrictive temperature (Fig. 5B). Results of the mRNA quantification are summarized in Fig. 5C. The spt15-V161A and spt15-N159D alleles exhibited the most severe synthetic growth defects when combined with the tih2-ts160 allele (Fig. 3). In parallel, transcription of the PHM2, SPL2, and PHO5 genes, which are decreased in tih2-ts160 mutants, were almost abolished in these two TBP mutants (Fig. 5, A and B). In contrast, increased expression of the GDH1 gene was observed only in the tih2-ts160 mutant (Fig.  5B). In the spt15-N159D mutant, transcription of the GDH1 gene was considerably weaker even when the tih2-ts160 mutation was present. This suggests that N159D might impair TBP function more strongly than other TBP mutations. Consistent with this is the observation that even transcription of the ADH1 gene, which showed no change in tih2-ts160 and was used as a control, was reproducibly decreased in the spt15-N159D mutant (Fig. 5B).
Somewhat unexpectedly, the effect of the N159L mutation on transcription was rather different from that of N159D (Fig.  5B). Although both showed synthetic growth defects when combined with tih2-ts160 mutation and grown on synthetic media (Fig. 3A), the N159L mutation alone uniquely increased the transcription of genes like PHM2, SPL2, and PHO5, whose transcription is decreased in tih2-ts160 (Fig. 5C). This increase in transcription was also observed in the double spt15-N159L, tih2-ts160 mutant. Thus, the mechanisms causing synthetic growth defects in the spt15-N159L and spt15-N159D double mutants may differ even though they harbor substitutions at the same residue. This may be responsible for the different growth phenotypes of these two mutants on YPD media (Fig. 3B).
Another TBP mutant, S118L, which showed the weakest synthetic growth defects (Fig. 3), also exhibited impaired transcription of the PHM2 and SPL2 genes. Other TBP mutants that did not display synthetic growth defects were almost normal in the transcription of these genes, except for the increased expression of GDH1 in the spt15-P65S mutant.
Thus, the extent to which TBP mutations cause a decrease in yeast growth in the context of the tih2-ts160 mutation appears to correlate with the decreased expression of several genes that are independently affected by both mutations. We conclude that there are functions of TBP and Tih2p that overlap and appear to be involved in recruiting TBP to the promoter and that disrupting both of these functions at the same time has a synthetic effect on yeast cell growth. DISCUSSION In this study, we verified that Tih2p and TBP physically and functionally interact in vivo. Although it is unclear why previous studies failed to show such a physical association in vivo, it may be partly because of the relative weakness of their interaction. In fact, we estimate that less than 1% of the total Tih2p pool can be coprecipitated with TBP and vice versa. Considering that other TBP-interacting partners, such as TBP-associated factors, can be coprecipitated stoichiometrically with TBP, it is likely that the interaction between Tih2p and TBP is transient and/or consists of only a small fraction of these two proteins that can form a stable complex in vivo.
The most important findings presented in this report are that only some spt15 alleles show a synthetic growth defect with the tih2-ts160 allele. Furthermore, all mutation sites of these alleles locate on the surface side of DNA binding region of TBP, and these TBP mutants all had defects in TATA binding. This suggests that Tih2p might be involved in recruiting TBP and/or other TBP-related complexes such as TFIID to the promoter. Intriguingly, the integrity of the TATA binding activity of TBP has been shown previously (44,45) to be important for the response to acidic activators like GAL4, and Tih1p and Tih2p are also required for GAL4-dependent transcriptional activation (26). These observations are consistent with a model in which TBP or related complexes are recruited by Tih1p and Tih2p.
We identified the target genes of Tih2p by DNA microarray analysis. Comparison of severely affected genes in the tih2-ts160 strain with those identified previously (26) by depleting Tih1p and Tih2p levels revealed that they did not perfectly overlap with each other. The difference between these two studies might be because of different experimental conditions, including the use of different carbon sources, the application of the protein depletion method in the Dutta study (26), and variations in the array-making technology used. Importantly, increased expression of some target genes (e.g. MEP2, GDH1, and DLD3) was observed in the tih2-ts160 mutant but not in the ino80 null mutant. This clearly indicates that Tih2p and Ino80p have distinct functions in transcription, even though they are components of the same chromatin remodeling complex.
Notably, we found a correlation between the synthetic growth defects and the expression profile of several genes (namely, PHM2, SPL2, and PHO5); transcription of these genes was abolished not only in the tih2-ts160 mutant but also in the spt15-V161A and spt15-N159D mutants that displayed the most severe synthetic growth defects. Thus, the same or overlapping steps operating during the course of transcription, such as the recruitment of TBP to the promoter, may be partially impaired by both of these mutations, and combining them may result in transcriptional defects in a much broader range of genes, thereby leading to synthetic growth defects. Significantly, the effect of these mutations on transcription seems to be independent of the physical interaction between TBP and Tih2p, because immunoprecipitation assay revealed that none of the TBP mutations appear to be defective in this interaction. 3 Consistent with this is the observation that overexpression of TBP did not rescue the temperature-sensitive pheno-type of the tih2-ts160 mutant. 3 Therefore we speculate that the complex containing both Tih proteins might change the neighboring chromatin environment in some specific genes rather than by simply associating physically with TBP. One of our possible working models of the function of Tih2p is that the complex containing Tih1p and Tih2p might cause the chromatin remodeling in some specific genes through the interaction with acidic activators and efficiently assist the recruitment of TBP to the TATA box of these genes. According to this scenario, tih2-ts160 mutant might fail the activity of either chromatin remodeling of target genes or interactions with acidic activators under the restrictive temperature. We are currently conducting additional studies to examine this possibility.
Our previous study demonstrated that at least some of the genes affected in the taf1 mutant, e.g. cyclin and ribosomal protein genes, were also affected in the tih2-ts160 mutant. However, a recent study (53) using proteomic approaches revealed that Taf14p is also a component of the INO80 complex. This protein has also been shown to be a component of multiple transcription complexes, including the TFIID, TFIIF, Swi-Snf, NuA3, and mediator complexes (54 -56). Furthermore, genes involved in phosphate metabolism appeared to be a common target of the TFIID, SAGA, Swi-Snf, NuA4, and INO80 complexes. These observations support the notion that the INO80 complex may carry certain core promoter functions, in addition to its chromatin-remodeling activity, such as TFIID, TFIIF, and Mediater.
Tih1p and Tih2p both belong to the AAA ϩ family, a superfamily of proteins designated the ATPases associated with various cellular activities family (57,58), and may form hetero-or homohexameric ring-like structures (9). Other members of this family are also known to play an important role in transcription. For instance, a ring-like structure that corresponds to part of the 19 S proteasome subcomplex can be recruited to the GAL1 promoter by the GAL4 activator. This complex, which is apparently composed of six AAA proteins (i.e. Rpt1p-Rpt6p), was designated recently (59) as the APIS (AAA proteins independent of 20 S) complex. Biochemical and genetic studies demonstrated that the APIS complex is essential for activation by GAL4 (59). However, it is unclear how the AAA proteins in the APIS complex and Tih1p/Tih2p are involved in transcription. Elucidation of the molecular defects occurring in the tih2-ts160 mutant may bring us new insights into the mechanisms of transcriptional regulation in eukaryotes.