Functional Connections between Mediator Components and General Transcription Factors of Saccharomyces cerevisiae *

The yeast Gal11 protein is an important component of the Mediator complex in RNA polymerase II-directed transcription. Gal11 and the general transcription factor (TF) IIE are involved in regulation of the protein kinase activity of TFIIH that phosphorylates the carboxyl-terminal domain of RNA polymerase II. We have previously shown that Gal11 binds the small and large subunits of TFIIE at two Gal11 domains, A and B, respectively, which are important for normal function of Gal11 in vivo. Here we demonstrate that Gal11 binds directly to TFIIH through domain A in vitro. A null mutation in GAL11 caused lethality of cells when combined with temperature-sensitive mutations in the genes encoding TFIIE or the carboxyl-terminal domain kinase, indicating the presence of genetic interactions between Gal11 and these proteins. Mutational depletion of Gal11 or TFIIE caused inefficient opening of the transcription initiation region, but had no significant effect on TATA-binding protein occupancy of the TATA sequence in vivo. These results suggest that the functions of Gal11 and TFIIE are necessary after recruitment of TATA-binding protein to the TATA box presumably at the step of stable preinitiation complex formation and/or promoter melting. We illustrate genetic interactions between Gal11 and other Mediator components such as Med2 and Pgd1/Hrs1/Med3.

The Mediator complex was isolated as a factor that enables gene-specific activators to stimulate transcription in a cell-free reconstituted system of the yeast Saccharomyces cerevisiae (1). The complex consists of 20 global transcription regulators including Srb proteins, Med proteins, Gal11, Rgr1, Sin4, Pgd1, Rox3, Nut1, and Nut2. Genetic depletion of any of these Mediator components causes either a lethal or conditional growth phenotype in yeast (1)(2)(3)(4)(5)(6). Mediator binds to the carboxyl-terminal domain (CTD) 1 of the largest subunit of RNA polymerase II, thereby forming an initiation subcomplex termed RNA polymerase II holoenzyme (1,7). Successive isolations of mammalian Mediator containing homologues of the yeast Mediator subunits disclosed that Mediator is an evolutionarily conserved critical complex of the eukaryotic transcription machinery (6, 8 -14). However, the role of each component protein in the function of Mediator remains totally unknown. A priori, the function can be divided at least into two groups according to their roles in Mediator: the receiver/conveyer of an activator signal(s) and the modulator of the activity of RNA polymerase II. Several reports have revealed physical interactions between Mediator components and natural or artificial activators, suggesting that such components play the former role (4,15), whereas none has dealt with the latter role, except for those describing that Srb proteins bind to the CTD of RNA polymerase II (16).
Mutational depletion of Mediator subunit Gal11 results in inefficient transcription of various genes (17). Purified or recombinant Gal11 stimulates basal transcription in a cell-free system reconstituted with recombinant or highly purified general transcription factors and RNA polymerase II (18). The following observations suggested that Gal11 and the general transcription factor (TF) IIE, while interacting with each other, function in a common regulatory pathway. First, two domains of Gal11, A and B, which are essential for its in vivo function, make contacts with the small (Tfa2) and large (Tfa1) subunits of TFIIE, respectively (17,18). Second, yeast cells expressing a mutant form of TFIIE that fails to bind Gal11 show phenotypes similar to those of gal11 null mutants (17). Third, Gal11 cooperates with TFIIE to stimulate CTD phosphorylation catalyzed by TFIIH (19). Phosphorylation of the CTD associates with transition from initiation to elongation in the synthesis of transcript (20). Therefore, it may be in this process that Gal11 and TFIIE function to regulate transcription.
In this study, we present biochemical and genetic evidence indicating the existence of physical as well as functional connections among Gal11, TFIIE, and TFIIH. We also show that the functions of Gal11 and TFIIE are necessary after the binding of TATA-binding protein (TBP) to the TATA box using an in vivo potassium permanganate footprint technique. Furthermore, we describe analyses of yeast cells containing mutations in GAL11, MED2, PGD1, or SIN4, which encode proteins that form a subcomplex in Mediator (3,4,21,22). These results have revealed functional interactions among these proteins and thereby shed light on the question of how Mediator regulates RNA polymerase II activity.
Plasmids-The wild-type TFA1 and tfa1-21 genes cloned into a TRP1-marked low-copy number vector (pRS314) were described previously (23). The tfa1-21 gene was subcloned into a HIS3-marked lowcopy number vector (pRS313). The GAL11 gene was cloned into a low-copy number vector marked with TRP1 (pRS314) or URA3 (pRS316) and a high-copy number vector marked with TRP1 (pTV3) as described (17,18). To clone MED2, PGD1, and SIN4, the following DNA segments were amplified by PCR from genomic DNA of strain W303-1a: Ϫ705 to ϩ1527 for MED2 (2), Ϫ484 to ϩ1505 for PGD1 (24), and Ϫ1123 to ϩ3619 for SIN4 (25,26). (Numbers are relative to the translation initiation site of the respective gene.) To disrupt these genes, the isolated segments were subcloned into pBluescript, and the following regions were replaced with the LEU2 gene: the 1.2-kilobase pair HindIII-SalI fragment of MED2, the 0.6-kilobase pair EcoRI-SalI fragment of PGD1, and the 2.9-kilobase pair HpaI fragment of SIN4.
Each of the GST fusion proteins was immobilized on glutathioneagarose (Sigma) as described (17). The resin was equilibrated with buffer A (17) containing 0.1 M potassium acetate (buffer A-0.1) and incubated with purified proteins for 1 h on ice. After washing the resin with buffer A-0.1, bound proteins were eluted with buffer containing 20 mM glutathione, 0.1 M Hepes-KOH, pH 7.6, 0.1 M potassium acetate, 1 mM EDTA, and 20% glycerol. Bound and unbound proteins were subjected to SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting by the use of an enhanced chemiluminescent detection system (ECL, Amersham Pharmacia Biotech). For the experiment in Fig. 1C, bound proteins were eluted with buffer A-0.5. Bound and unbound proteins were precipitated by 10% trichloroacetic acid, separated on an SDS-polyacrylamide gel, and visualized by silver staining.

RESULTS
Physical Interaction between Gal11 and TFIIH-Direct binding of Gal11 to TFIIH was tested by protein affinity chromatography. The GST/Gal11 fusion protein (Fig. 1A) immobilized on glutathione-agarose was incubated with TFIIH purified from yeast. After extensive washing, bound and unbound pro-teins were subjected to immunoblotting and probed with an antibody raised against the 73-kDa subunit (Tfb1) of TFIIH (30). As shown in Fig. 1B (middle panel, lane 2), most of Tfb1 was detected in the bound fraction when TFIIH was incubated with GST/Gal11-immobilized resin. When TFIIH was incubated with control resin with GST alone, Tfb1 was exclusively detected in the unbound fraction (lane 1). This observation suggests that Gal11 specifically binds to TFIIH in vitro. To dissect the Gal11 region involved in binding to TFIIH, various deletion derivatives of Gal11 were constructed (Fig. 1A). The results show that the Gal11 region at amino acids 866 -929 is necessary for binding ( Fig. 1B, lanes 3-7). The proteins retained on the GST/Gal11-(866 -929) resin were also analyzed by silver staining. As shown in Fig. 1C (lane 1), the TFIIH sample used for the assay contained the TFIIH subunits and small amounts of contaminating proteins. (Although bands corresponding to Ssl1, Ccl1, and Kin28 may not be seen in this figure, we could unambiguously recognize these proteins in the original gel. We were unable to resolve Tfb3 and Tfb4 in our electrophoresis system because the two proteins have a similar molecular mass (5,27).) The TFIIH subunits were significantly enriched in the bound fraction, whereas a major part of each contaminating protein was contained in the unbound fraction (compare lanes 2 and 3). A band with a molecular size similar to that of Rad3 was seen in the unbound fraction, however. We did not ascertain whether this band represents Rad3 or a contaminating protein. We also could not identify Kin28 on the gel because it comigrated with GST/Gal11-(866 -929), which was dissociated from glutathione-agarose during the binding reaction. These results strongly suggest that Gal11 directly binds to TFIIH via its genuine subunit(s), but not via the contaminating proteins, even though we were unable to identify definitively all of the TFIIH subunits in the Gal11-bound fraction for technical reasons. It remains to be determined which subunit(s) of TFIIH binds to Gal11.
The Gal11 region from amino acids 866 to 929 designated domain A is necessary for normal function in vivo as well as for binding to the small subunit (Tfa2) of TFIIE in vitro (17,18). It was therefore interesting to analyze the effect of Tfa2 on the interaction between domain A and TFIIH. As shown in Fig. 2A (lanes 3-5), the addition of increasing amounts of Tfa2 to the mixture containing GST/domain A and TFIIH resulted in decreases in the Tfb1 subunit in the bound fractions. However, a Tfa2 derivative, Tfa2-N302, which lacks the domain A-binding region in Tfa2 (17), showed a marginal effect on TFIIH binding (lane 6). The domain A-TFIIH interaction was inhibited specifically with Tfa2, suggesting that both TFIIH and Tfa2 bind the identical or an overlapping region in domain A of Gal11.
To further analyze the effect of TFIIE on the Gal11-TFIIH interaction, GST/Gal11-immobilized resin was incubated with TFIIH in the presence or absence of TFIIE subunits (Fig. 2B). In this experiment, the amount of immobilized GST/Gal11 was reduced such that ϳ30% of input Tfb1 was retained on the resin (lane 5). In accordance with the experiment in Fig. 2A, the  6 and 8). Perhaps, conformation of Tfa2 changes through binding to Tfa1, which would alleviate the inhibitory effect on the Gal11-TFIIH interaction. Genetic Interactions between Tfa1 and Mediator Components-Previously, we have shown that cells producing a TFIIE mutant lacking the Gal11-binding regions (TFIIE-⌬C) show phenotypes similar to those of gal11 (gal11 ⌬ ) null strains (17). The combination of TFIIE-⌬C with gal11 ⌬ did not enhance the phenotypes of either single mutation, suggesting that TFIIE and Gal11 function through a common pathway in transcription (17). Here we analyzed phenotypes of gal11 ⌬ cells when combined with a temperature-sensitive mutation in the TFA1 gene (tfa1-21) (23). A TRP1-marked plasmid harboring the tfa1-21 gene was introduced into tfa1 ⌬ gal11 ⌬ cells containing a wild-type TFA1 plasmid marked with URA3. To shuffle out the TFA1-bearing plasmid, transformants were replica-plated onto medium containing 5-FOA. As shown in Fig. 3A, transformants failed to grow on a 5-FOA-containing plate, indicating that cells containing both the tfa1-21 and gal11 null mutations are nonviable. We next analyzed the effect of the combination of tfa1-21 with a null mutation in MED2, PGD1, or SIN4, encoding Mediator subunits that are known to form a subcomplex with Gal11 (3,4,21,22). A null mutation in either MED2 or PGD1 caused lethality in a tfa1-21 background, whereas sin4 ⌬ tfa1-21 was viable (Fig. 3A).
We further studied interactions between Tfa1 and the Mediator subunits by examining whether the lethal phenotype of cells containing double mutations, i.e. tfa1-21 and a null mutation in one of the Mediator components, could be suppressed by overexpressing a heterogeneous Mediator component (Fig.  3B). The synthetic lethality of tfa1-21 gal11 ⌬ was not sup-pressed by introducing MED2, PGD1, or SIN4 with a high-copy plasmid. Interestingly, the lethality of tfa1-21 med2 ⌬ and tfa1-21 pgd1 ⌬ was suppressed by introducing GAL11 with a high-copy plasmid. These results suggest that Gal11 becomes indispensable for growth in tfa1-21 cells and that Gal11 is functionally impaired in tfa1-21 med2 ⌬ and tfa1-21 pgd1 ⌬ cells despite the fact that Gal11 should be produced in these cells (see "Discussion").

FIG. 3. Phenotypes of cells containing a null mutation in Mediator components in
Gal11 is involved in the regulation of CTD phosphorylation by TFIIH (19), we suspected that GAL11 could genetically interact with KIN28 as well. We then constructed the kin28-ts3 gal11 ⌬ strain containing GAL11 on a URA3-marked plasmid. Cells were transformed with either a plasmid harboring GAL11 or a vacant vector, and the requirement of GAL11 for growth was tested by a plasmid shuffling experiment. As shown in Fig. 3C, transformants harboring the vacant vector failed to grow on a 5-FOA-containing plate, indicating that cells containing both kin28-ts3 and gal11 ⌬ mutations are non-viable. This synthetic lethal phenotype was not suppressed by overexpressing MED2, PGD1, or SIN4.
Effects of TFIIE and Gal11 on Promoter Opening-To explore which step of the transcription process is impaired by the tfa1-21 and gal11 ⌬ mutations, we employed the in vivo potassium permanganate footprint technique (Fig. 4). Base modification by KMnO 4 occurs preferentially in T residues in singlestranded DNA; and therefore, this reagent has been used for analysis of the open polymerase complex and polymerase paused complex (28,29,35,36). Since both TFIIE and Gal11 are necessary for full activation of GAL7 by Gal4 in response to galactose (18,23), we analyzed the effects of the tfa1-21 and gal11 ⌬ mutations on structural change of the GAL7 promoter. When wild-type TFA1 cells were grown in glucose medium, KMnO 4 -reactive T residues were observed at nucleotides Ϫ7 and Ϫ9 on the bottom strand of GAL7 (Fig. 4A, lane 1). Under the galactose-induced condition, every T residue from nucleotides Ϫ33 to ϩ26 became hypersensitive to KMnO 4 (lane 2). On the top strand, reactive T residues were detected at nucleotides Ϫ31 and Ϫ40 in glucose-grown cells, and the addition of galactose induced reactivity of T residues from nucleotides Ϫ42 to ϩ17 (lanes 7 and 8). As summarized in Fig. 5, activation of GAL7 caused the creation of a hypersensitive region ϳ70 bases in length initiated at 21 base pairs downstream of the TATA box (at nucleotide Ϫ63). This observation is consistent with the previous findings that DNA melting begins ϳ20 base pairs downstream of the TATA box and extends to the 3Ј region ϳ60 bases in length in some yeast genes tested (29,35,36).
When tfa1-21 cells were grown at 28°C, the addition of galactose induced KMnO 4 hypersensitivity of the GAL7 promoter, as in the case of wild-type TFA1 cells (Fig. 4A, lanes 5  and 11). To inactivate Tfa1-21 protein function, galactosegrown cells were shifted to the restrictive temperature (37°C) for 30 min before KMnO 4 treatment. In tfa1-21 cells, growth at the restrictive temperature results in degradation of both subunits of TFIIE, and thereby in the cessation of mRNA synthesis in most genes (23). Preincubation at 37°C caused almost complete loss of the hypersensitive region on both strands, and the base modification changed to yield patterns similar to those of glucose-grown cells (lanes 6 and 12). A null mutation in GAL11 also inhibited the creation of the hypersensitive region in galactose-grown cells (Fig. 4B). These results suggest that the galactose-inducible melting of the GAL7 promoter requires the functions of both TFIIE and Gal11.
Occupancy of the GAL7 TATA Region in tfa1-21 and gal11 ⌬ Cells-It has been shown that the TATA sequence is also sensitive to KMnO 4 treatment because of its non-B form structure. The binding of TBP protects the TATA region from KMnO 4 modification (28,35,36). Chromatin immunoprecipitation experiments have indicated that the addition of galactose induces binding of TBP to the TATA box of galactose-inducible genes (37,38). This observation agreed with the results from the KMnO 4 footprint experiments in the GAL7 promoter as follows.
In the presence of glucose, T residues at nucleotides Ϫ60, Ϫ62, and Ϫ71 on the bottom strand were hypersensitive to KMnO 4 treatment (Fig. 4C, lane 1). The addition of galactose caused protection of two T residues at nucleotides Ϫ60 and Ϫ62 in the TATA box (lane 2), indicating that the GAL7 TATA box is occupied with TBP. To determine the requirement of TFIIE for the occupancy of the TATA region, tfa1-21 cells were grown in the presence of galactose at 28°C (lane 5) and then shifted to 37°C (lane 6). As shown here, protection of the two T residues was maintained even after the temperature shift. In gal11 ⌬ cells, the addition of galactose induced protection of the TATA sequence as well as in the wild-type cells (compare lanes 8 and 10). These results indicate that the binding of TBP to the GAL7 TATA box is not significantly affected by loss of function of TFIIE or Gal11. DISCUSSION Here we have demonstrated the direct binding of Gal11 to TFIIH. Thus, physical linkages among Gal11, TFIIE, and TFIIH have been completed, together with the previous findings for Gal11-TFIIE (17,18) and TFIIE-TFIIH (31) interactions. We have further shown that the gal11 ⌬ mutation causes synthetic lethality in combination with the tfa1-21 or kin28-ts3 mutation. These and previous findings (17)(18)(19) strongly suggest that Gal11, together with TFIIE, is directly involved in the CTD phosphorylation by the Kin28 kinase. In other words, these proteins function in a common regulatory process in transcription.
Using the in vivo KMnO 4 footprint technique, we found that galactose-inducible melting of the GAL7 promoter region is inhibited by a gal11 null mutation. This effect may be due to an inefficient recruitment of TFIIE to the preinitiation complex in gal11 ⌬ cells (see below) (18). Although Gal11 is necessary for full activation of GAL7, occupancy of the GAL7 TATA sequence by TBP was unaffected by depletion of Gal11. By contrast, Li et al. (37) reported that inactivation of Srb4, another essential Mediator component, causes inhibition of the binding of TBP to the TATA box in Gal4-activated genes. Mediator is composed of at least two stable subcomplexes, one containing Rgr1 and the other Med6, of which Gal11 and Srb4 are constituents, respectively (3). It is also known that Srb4 is a target of the transcription activation domain of Gal4 (15). These observations lead us to hypothesize that Srb4 and Gal11 have distinct roles in Mediator; Srb4 is necessary for activator (Gal4)-mediated stable binding of TBP to the TATA box, whereas Gal11 is required for a process after the FIG. 5. Potassium permanganatereactive T residues of the GAL7 gene. The sequence from nucleotides Ϫ74 to ϩ30 relative to the transcription initiation site of the GAL7 gene is shown. Boldface letters indicate the TATA box and the transcription and translation initiation sites. Open and closed arrowheads indicate T bases weakly and strongly reactive to KMnO 4 , respectively, in glucose (dextrose (Dex))-and galactose (Gal)-grown cells.
TBP recruitment. It should be noted here that Lee et al. (4) suggested the role of Gal11 as that of an activator-binding module based on its binding to acidic activators.
Inactivation of TFIIE caused inefficient opening of the GAL7 promoter, but had no significant effect on TBP binding to the TATA region. It was suggested that both RNA polymerase II and Rad25/Ssl2 protein, a DNA helicase subunit of TFIIH, are necessary for promoter melting (29,36). One might argue that the inefficient promoter melting after inactivation of TFIIE is an indirect consequence of cessation of mRNA synthesis. Instead, we favor that TFIIE is directly involved in open complex formation for the following reasons. First, TFIIE is located near the active center of RNA polymerase II in TFIIE-polymerase cocrystals and, therefore, may be involved in the regulation of the conformational change of the polymerase-DNA interaction (39,40). Second, TFIIE has an ability to bind single-stranded DNA (41). Third, mammalian TFIIE regulates DNA helicase activity in TFIIH (42,43). Fourth, in the absence of TFIIH, TFIIE stimulates transcription depending on the helical stability of promoter DNA in vitro (44). We assume therefore that inactivation of TFIIE leads to failure in the association of polymerase with DNA and/or in the generation or maintenance of promoter melting. On the other hand, TFIIE is dispensable for occupancy of the GAL7 TATA region. Similarly, inactivation of TFIIH subunits, such as Rad25 and Kin28, does not affect the binding of TBP to the TATA box (36,38). In light of these findings, we propose that recruitment of TBP to the TATA box and a reaction that requires TFIIE and TFIIH occur at distinctive steps in transcription.
A null mutation in GAL11, MED2, or PGD1 caused lethality of cells when combined with a tfa1-21 mutation. It is known that gal11 ⌬ , med2 ⌬ , and pgd1 ⌬ cells show similar phenotypes, such as temperature-sensitive growth, defect in galactose utilization, and inefficient production of ␣-pheromone (2,17,21,45). These genetic observations may be interpreted to imply that the Gal11, Med2, and Pgd1 proteins interact functionally with each other as well as with TFIIE. This idea is supported by biochemical evidence that Gal11, Med2, and Pgd1 form a subcomplex in Mediator (3,4,21,22). We have shown that the lethality of tfa1-21 med2 ⌬ and tfa1-21 pgd1 ⌬ cells was suppressed by overexpression of GAL11. However, overexpression of MED2 or PGD1 failed to suppress the lethality of tfa1-21 gal11 ⌬ cells. We assume that the integrity of the Mediator complex is distorted in the absence of Med2 or Pgd1, resulting in functional loss of Gal11, which in turn makes med2 ⌬ or pgd1 ⌬ cells non-viable in a tfa1-21 background. Overexpressed Gal11 would suppress the lethal phenotype of tfa1-21 med2 ⌬ or tfa1-21 pgd1 ⌬ cells by restoring the integrity of the Mediator complex.
We have found that sin4 ⌬ , unlike gal11 ⌬ , med2 ⌬ , or pgd1 ⌬ , did not cause lethality of the cells when combined with tfa1-21. Yeast cells bearing sin4 ⌬ share several phenotypes with gal11 ⌬ , med2 ⌬ , and pgd1 ⌬ , but not one for galactose utilization (25,26,46,47). In addition, SUC2 (encoding invertase) is de-repressed in sin4 ⌬ cells, indicating that Sin4 acts as a negative factor as other Ssn proteins, whereas SUC2 is repressed in gal11 ⌬ , indicating that Gal11 acts as a positive factor as other Snf proteins (48,49). The gal11 ⌬ and sin4 ⌬ mutations show phenotypes distinct from each other when combined with mutations in SNF and SWI, which encode components of chromatin-remodeling factors (50). All these findings suggest that Sin4 has a function(s) distinct from those of Gal11, Med2, and Pgd1 in the cell, despite the fact that Sin4 is included in the Rgr1 subcomplex, as are Gal11, Med2, and Pgd1.
Our present data have furnished information as to how Mediator controls the activity of RNA polymerase II. We have thus illustrated through genetic and biochemical approaches that functional interactions exist among Mediator components (such as Gal11, Med2, and Pgd1), general transcription factors (such as TFIIE and TFIIH), and RNA polymerase II. We have further demonstrated that the interactions are necessary at a late step in transcription initiation.