Mediator Influences Schizosaccharomyces pombe RNA Polymerase II-dependent Transcription in Vitro*

The fission yeast Schizosaccharomyces pombe has proved an important model system for cross-species comparative studies of many fundamental processes in the eukaryotic cell, such as cell cycle control and DNA replication. The RNA polymerase II transcription machinery is, however, still relatively poorly understood in S. pombe, partially due to the absence of a reconstituted in vitro transcription system. We have now purified S. pombe RNA polymerase II and its general initiation factors TFIIB, TFIIF, TFIIE, and TFIIH to near homogeneity. These factors enable RNA polymerase II to initiate transcription from the S. pombe alcohol dehydrogenase promoter (adh1p) when combined with Saccharomyces cerevisiae TATA-binding protein. We use our reconstituted system to examine effects of Mediator on basal transcription in vitro. S. pombe Mediator exists in two distinct forms, a free form, which contains the spSrb8, spTrap240, spSrb10, and spSrb11 subunits, and a smaller form, which lacks these four subunits and associates with RNA polymerase II to form a holoenzyme. We find that spSrb8/spTrap240/spSrb10/spSrb11 containing Mediator repress basal transcription, whereas Mediator lacking these subunits has a stimulatory effect on transcription. Our findings thus demonstrate that the spSrb8/spTrap240/spSrb10/spSrb11 subcomplex governs the ability of Mediator to stimulate or repress basal transcription in vitro.

The minimal set of general transcription factors required for basal transcription in vitro include, RNA polymerase II (pol II), 1 the TATA-binding protein (TBP), and transcription factors (TFs) IIB, IIE, IIF, and IIH. These factors are both necessary and sufficient for proper initiation of transcription in transcription systems reconstituted from Saccharomyces cerevisiae and mammalian cells (1).
The structure and function of Schizosaccharomyces pombe pol II has been investigated in great detail. The enzyme is composed of 12 subunits, Rpb1 to Rpb12, similar to what has been described for the budding yeast S. cerevisiae and mammalian cells (2,3). Biochemical analysis of the enzyme has revealed the presence of several distinct subcomplexes and has proved important in the analysis of the recently published x-ray crystallography structure of S. cerevisiae pol II (4,5). Previous characterization of S. pombe TFIIH has only concerned a specific subcomplex of this transcription factor, containing the Mcs6, Mcs2, and Pmh1 gene products (6). These proteins correspond to the cyclin-dependent kinase 7 (Cdk7), cyclin H, and Mat1, in mammalian TFIIH. Cdk7 phosphorylates the C-terminal domain (CTD) of the largest subunit of pol II upon initiation of transcription (7)(8)(9). Interestingly, Cdk7 is also the mammalian cdk-activating kinase (Cak): the kinase needed for the activating phosphorylation of other cyclin-dependent kinases (10). A similar situation also appears to exist in S. pombe, where the trimeric Mcs6-Mcs2-Pmh1 complex also possesses CAK activity (6).
We have reported previously on the purification and characterization of the S. pombe Mediator (11)(12)(13). The Mediator complex is essential for basal and regulated expression of nearly all pol II-dependent genes in S. cerevisiae (14), and depletion of human Mediator from nuclear extracts abolishes transcription by pol II (15). The S. pombe Mediator complex exists in at least two specific forms (13): one smaller core Mediator in complex with pol II and one larger form of Mediator, devoid of pol II, but containing the spSrb8, spTrap240, spSrb10, and spSrb11 proteins. Homologues to these four proteins are found in certain S. cerevisiae and human Mediator preparations and they are collectively referred to as the Srb8 -11 module (13,16). The SRB11 and SRB10 genes encode cyclin C and the cyclin C-dependent kinase, respectively (17), and genetic analysis indicates that the Srb8 -11 module is involved in the negative regulation of a small subset of genes during exponential growth (13,18). Why Mediator exists in two different forms and what differential roles these complexes have in transcriptional regulation are still unclear.
Protein Purification and Identification-RNA pol II holoenzyme was purified from a TAP-spMed7 strain in which spTrap240 ϩ had been deleted. spSrb8 -11/Mediator and the pol II holoenzyme⌬spTrap240 were purified as described previously (13).
We purified core TFIIH from a strain containing a C-terminal TAPtag on sptfb2 ϩ , following previously published protocols (13), but with the following modifications. We grew 20 liters of S. pombe to mid-log phase in YES medium (20) supplemented with 0.2 g/liter adenine. Cells were collected by centrifugation (JA-10, 6,000 rpm, 5 min, ϩ4°C), washed once with ice-cold water, and suspended in 0.5 ml of 3 ϫ lysis buffer (0.3 M Tris acetate (pH 7.8), 0.15 M potassium acetate, 30% (v/v) glycerol, 3 mM EDTA, 1.5 mM dithiothreitol, and 3ϫ protease inhibitors) per g of cell pellet. Cells were lysed by bead beating (BeadBeater, Stratech Scientific Ltd.) using 25 cycles, 30 s of beating, and 90 s of rest. We cleared the supernatant by centrifugation (JA-10, 9,000 rpm, 25 min, ϩ4°C), added 1 ⁄9 volume of 5 M KAc and stirred for 15 min. We next added 0.2% (v/v) polyethyleneimine and stirred for an additional 30 min. After ultracentrifugation (Ti45, 42,000 rpm, 90 min, ϩ4°C), the supernatant was dialyzed against IPP150 overnight, added to 1 ml of IgG-Sepharose beads (Amersham Biosciences), and incubated by rotation for 2 h at ϩ4°C. The beads were collected by sedimentation, loaded into a column, and washed with 30 ml of IPP150 followed by 10 ml of TEV cleavage buffer (10 mM Tris chloride (pH 8.0), 150 mM sodium chloride, 0.1% Nonidet P-40, 0.5 mM EDTA, and 1 mM dithiothreitol). We eluted TFIIH by addition of 500 units of TEV protease in 1.5 ml TEV cleavage buffer and incubation for 2 h at ϩ16°C. We added 3 ml of calmodulin binding buffer and 3 l of 1 M CaCl/ml of eluate and loaded the mixture on a 1-ml calmodulin column (Amersham Biosciences). The column was rotated for 2 h at ϩ4°C and washed with 30 ml of IPP150 Calmodulin binding buffer. Core TFIIH was eluted by rotation at ϩ4°C for 1 h in IPP150 calmodulin elution buffer with 500 mM NaCl.
TFIIK (TAP-Pmh1) was purified as core TFIIH but eluted with calmodulin elution buffer containing 150 mM NaCl. We also used an alternative, less stringent protocol to monitor for additional proteins, which interacted only weakly with TFIIK. In the alternative protocol we decreased the Nonidet P-40 concentration from 0.1% to 0.01% throughout the purification procedure and furthermore decreased the buffer volumes used to wash the IgG-Sepharose and calmodulin-Sepharose columns to 5 column volumes.
For RNA pol II purification, 972h Ϫ cells were grown and collected as described for purification of core TFIIH. Cells were dissolved in 3ϫ pol lysis buffer (300 mM Hepes-KOH (pH 7.9), 150 mM KCl, 3 mM EDTA, 10 mM DTT, 30% glycerol, 3ϫ protease inhibitors). We lysed the cells by bead beating (25 cycles, 30 s of beating, and 90 s of rest) and cleared the supernatant by centrifugation (JA-10, 9,000 rpm, 25 min, ϩ4°C). We added 0.05 volume of 4 M ammonium sulfate to the supernatant and stirred for 15 min. We next added 0.2% (v/v) polyethyleneimine and stirred for an additional 30 min. After ultracentrifugation (Ti45, 25,000 rpm, 15 min, ϩ4°C), an equal volume of 4 M ammonium sulfate was added to the supernatant at a speed of 10 ml/min, and the mixture was stirred for an additional 30 min. After ultracentrifugation (Ti45, 25,000 rpm, 15 min, ϩ4°C) the pellet was dissolved in buffer D-0 (50 mM Hepes (pH 7.5), 0.5 mM EDTA, 1 mM DTT, 1ϫ protein inhibitors; the number after the hyphen indicates the ammonium sulfate concentration in molar units) until a 1:200 dilution gave below 400 microsiemens on the conductivity meter. The supernatant was clarified by centrifugation (JA-17, 13,000 rpm, 30 min, ϩ4°C) and loaded onto an 8WG16 antibody-Sepharose column (CNBr-activated Sepharose 4B (Amersham Bioscience) coupled to antibody at 2 mg/ml), which had been equilibrated with Buffer D-0.4. The column was washed with 25 column volumes of buffer D-0.25 and subsequently moved to room temperature.
Half a column volume of elution buffer was added (25 mM Hepes (pH 7.5), 500 mM ammonium sulfate, 50% glycerol, 0.5 mM EDTA, 1 mM DTT, 1ϫ protein inhibitors) and left to incubate for 15 min. More elution buffer was added, and 0.5-ml fractions were collected until no more pol II appeared to be eluting as judged by SDS-PAGE and Coomassie Brilliant Blue staining.
Recombinant His 6 -TFIIB was expressed in BL21-CodonPlus (DE3)-RP cells (Stratagene) and purified over Ni 2ϩ -NTA fast flow (Qiagen) according to the manufacturer's recommendations. The eluate was dialyzed against buffer A-0.1 (25 mM Hepes-KOH (pH 7.6), 10% (v/v) glycerol, 1 mM EDTA, 1 mM dithiothreitol, and 1ϫ protease inhibitors; the number after the hyphen indicates the potassium acetate concentration in molar units) overnight. The dialyzed material was applied to a Mono S 5/5 column (Amersham Biosciences), which had been equilibrated in buffer A-0.1. The column was washed with 5 ml of buffer A-0.1, and proteins were eluted with a linear gradient (15 ml) of buffer A-0.1 to A-1.0. A peak of essentially pure protein eluted around 0.5 M KAc.
For purification of TFIIF we co-infected Sf9 cells with recombinant baculoviruses expressing spTfg1, spTfg2, and spTfg3 at a multiplicity of infection of 5 for each virus. Cells were collected 72 h after infection by centrifugation (JA-10, 3,500 rpm 10 min, 4°C), washed with ice-cold PBS, and centrifuged as before. Cells were dissolved in 10 ml of buffer B-0 (25 mM Tris-HCl (pH 8.0), 10 mM ␤-mercaptoethanol, and 1ϫ protease inhibitors; the number after the hyphen indicates the sodium chloride concentration in molar units) and incubated on ice for 20 min followed by lysis using a Dounce homogenizer (15 ml) 15 times. After rotation in 10 ml of buffer B-1.6 for 45 min at ϩ4°C broken cells were ultracentrifuged (TLA 100.3, 45,000 rpm, 45 min, 4°C). The supernatant was loaded on 1 ml Ni 2ϩ -NTA fast flow (Qiagen) and purified according the manufacturer's recommendations. The peak of TFIIF was dialyzed against buffer Q-0.1 (20 mM Tris acetate (pH 7.9), 0.2 mM EDTA, 10% glycerol, 1 mM dithiothreitol, and 1ϫ protease inhibitors; the number after the hyphen indicates the potassium acetate concentration in molar units) overnight. The dialyzed material was loaded on a Mono Q 5/5 FPLC column (Amersham Biosciences) which had been equilibrated in buffer Q-0.1. The column was washed with 5 ml of buffer Q-0.1 and eluted by a linear gradient (15 ml) of buffer Q-0.1 to Q-1.5. Essentially pure TFIIF eluted at about 900 mM KAc.
For purification of TFIIE we co-infected Sf9 cells with recombinant baculoviruses expressing Toa1 and Toa2 at a multiplicity of infection of 5 for each virus. We purified TFIIF following the same protocol as for TFIIE but with the following modifications. The peak of TFIIE from Ni 2ϩ -NTA fast flow column was dialyzed against buffer C-0.1 (20 mM Tris acetate (pH 7.9), 10% glycerol, 1 mM dithiothreitol, and 1ϫ protease inhibitors; the number after the hyphen indicates the potassium acetate concentration in molar units) overnight. The dialyzed material was loaded on a 1-ml DEAE-Sepharose column (Amersham Biosciences), which had been equilibrated in buffer C-0.1. The column was washed with 5 ml of buffer C-0.1 and eluted by a linear gradient (15 ml) of buffer C-0.1 to C-1.5. Essentially pure TFIIE eluted at about 0.8 M KAc.
Purification of S. cerevisiae TBP was as described previously (21). The identification of proteins in this report, by MALDI-TOF analysis of in-gel digested proteins was essentially as described previously (12).
CTD Phosphorylation-S. pombe pol II (100 ng) was incubated with TFIIK and core-TFIIH as indicated in the legend to

RESULTS
We wanted to examine the influence of the spSrb8 -11 complex on basal transcription in vitro. To this end we needed to purify a defined Mediator complex, lacking the spSrb10 cyklindependent kinase. We used a tandem affinity purification (TAP) tag on spMed7 and purified Mediator from a yeast strain carrying a deletion of sptrap240 ϩ . The ⌬sptrap240 Mediator was purified in complex with pol II, forming a holoenzyme (holoenzyme⌬sptrap240). The absence of Srb10 from the holoenzyme⌬trap240 was verified with silver staining and a CTD kinase assay (Fig 1, A and B). For comparison we used the defined spSrb8 -11/Mediator complex, which we had isolated previously (13). Evidently the spTrap240 subunit is necessary for the association of the spSrb10 cyklin-dependent kinase with the Mediator complex.
We next wanted to examine the influence of the spSrb8 -11 complex on basal transcription in vitro. Since no in vitro system had been established for S. pombe, we first set out to purify the necessary transcription factors. TFIIF from different eukaryotes contains two highly conserved subunits, Tfg1 and Tfg2. In S. cerevisiae a third subunit, Tfg3, has been shown to associate with TFIIF but was lost during later steps of purification (24). Homologues to these three proteins are also found in S. pombe, and upon co-expression in insect cells they generate a stable, heterotrimeric TFIIF, which could be purified to homogeneity (Fig. 2, A and B). We noted that large amounts of spTfg1 and spTfg2 precipitated during purification, suggesting that spTfg3 may be needed to generate a soluble complex ( Fig.  2A). In fact, we failed to obtain any soluble TFIIF, when only spTfg1 and spTfg2 was co-expressed in insect cells (data not shown).
We next purified S. pombe RNA pol II using an anti-CTD monoclonal antibody column and generated a basically homogenous material (Fig. 2C). We identified the subunits of pol II using MALDI-TOF mass fingerprinting, and our results were in perfect agreement with the subunit analysis reported previously (data not shown) (2).
In S. cerevisiae and higher eukaryotic cells, TFIIH contains nine subunits (25). To purify S. pombe TFIIH, we introduced a TAP-tag on the C terminus of Pmh1. Purification on IgG-Sepharose and Ca 2ϩ /calmodulin-Sepharose generated a trimeric complex (Fig. 3A). The individual subunits were identified with MALDI-TOF mass fingerprinting as Mcs2, Mcs6, and Pmh1 (Table I). A similar subcomplex of TFIIH has also been identified in S. cerevisiae and denoted TFIIK (7). Notably absent in our purified complex were the subunits of core TFIIH.
In an attempt to isolate the holo-form form of TFIIH (core ϩ TFIIK), we used an alternative protocol, in which we decreased the Nonidet P-40 detergent concentrations during the TAP purification (see "Materials and Methods"). In these preparations we could identify substoichiometric amounts of Rad15 the S. pombe homologue to the S. cerevisiae Rad3 helicase (Fig. 3B) (26). No other components of TFIIH was, however, found in the TAP-Pmh1 preparations, suggesting that TFIIH may be a less stable complex in S. pombe than what has been described in S. cerevisiae and human cells (25). To purify the S. pombe core-TFIIH complex, we fused a TAP-tag to the C-terminal part of spTfb2. TAP purification generated a five-subunit complex, which was characterized with MALDI-TOF mass fingerprinting as Ercc3sp, spTfb1, spTfb2, spSsl1, and spTfb4 (Fig. 3B). The MALDI-TOF fingerprinting data of the identified TFIIH subunits are summarized in Table I. We tested the ability of TFIIK to phosphorylate the C-terminal domain of pol II in the presence and absence of core TFIIH (Fig. 3C). Phosphorylation was detected in both cases, although the level was slightly lower in the presence of core TFIIH.
We identified the SPAC16E8.16 gene as a putative S. pombe TFIIB homologue. The gene product was fused in-frame with a C-terminal His 6 -tag, expressed in E. coli, and purified to near homogeneity (Fig. 4A). TFIIE contains two subunits in S. cerevisiae, which are encoded by the TOA1 and TOA2 genes. We identified two highly conserved homologues to these genes in S. pombe, which upon co-expression in insect cells generated dimeric TFIIE, which could be purified to homogeneity over Ni 2ϩ -agarose and DEAE-Sepharose (Fig. 4B). We next set out to reconstitute transcription in vitro. We used negatively supercoiled templates with either the adenovirus major late promoter or the S. pombe adh1 ϩ promoter (adh1p) followed by a G-less cassette. Addition of all the factors did indeed generate strong, promoter specific transcription from the endogenous S. pombe promoter (Fig. 5A). Lower levels of in vitro transcription could also be observed from the adenovirus major late promoter construct (data not shown). To test the activity of each individual transcription factor we performed a dropout transcription assay where one factor at the time was left out (Fig. 5A). We found that transcription was completely abolished if TBP, TFIIB, pol II, or TFIIE were omitted. In the absence of TFIIF we could observe a low level of transcription (2% relative the complete reaction), probably due to small amounts of TFIIF contaminating the purified pol II. Our reactions were less dependent on TFIIH, which had a 3-fold stimulatory effect on basal transcription. In agreement with our observation, TFIIH is not needed for transcription of certain promoters on negatively supercoiled templates in mammalian in vitro transcription systems (27)(28)(29). In contrast, our S. pombe in vitro transcription system appears to be more dependent on TFIIE than the human system (27,28), since we can observe no transcription in the absence of this factor. We next tested the effects of holoenzyme⌬spTrap240 and spSrb8 -11/Mediator on basal transcription (Fig. 5B). We found that Mediator devoid of spSrb8 -11 had a significant stimulatory effect on basal transcription. In contrast, Mediator containing spSrb8 -11 had a strong negative effect, diminishing the basal levels of transcription up to 5-fold. The stimulation of basal transcription observed for holoenzyme⌬spTrap240 is not merely an effect of adding more pol II, since the addition of free pol II to the same concentrations as found in the holoenzyme⌬spTrap240 preparation at best had a modest (1.2fold) stimulatory effect on the transcription reaction (data not shown). Neither do we believe that the inhibitory effect associated with spSrb8 -11/Mediator is due to any contaminating factor. When we do repeated elutions of Srb8 -11/Mediator at the Ca 2ϩ /calmodulin-Sepharose purification step, we find that the levels of inhibition observed in our in vitro transcription system correlates well with the relative amounts of Mediator and spSrb10 associated CTD-kinase activity found in the individual eluted fractions (Fig. 5C).

FIG. 3. Characterization of TFIIH.
A, TFIIK (TAP-Pmh1) was purified according to the high stringency procedure described under "Materials and Methods." The proteins were separated on a 10% SDS-PAGE gel, revealed by Coomassie Brilliant Blue, and identified with MALDI-TOF mass fingerprinting. Arrows on the left indicate identified proteins, and molecular masses according to standard are indicated on the right. The band below spTfb4 (*) is a contaminant. B, Rad3 associates with TFIIK. TFIIK (TAP-Pmh1) was purified according to the low stringency procedure described under "Materials and Methods." The eluate was analyzed as described in A. The major contaminants Actin and Ef1a are indicated. C, core TFIIH (TAP-spTfb2) was purified and analyzed as described in the legend to A. D, RNA polymerase II (core) was phosphorylated by TFIIK. The presence of core TFIIH did not significantly influence the CTD kinase activity.

DISCUSSION
Cross-species comparisons between S. pombe and S. cerevisiae have been successfully used in the characterization of the eukaryotic cell cycle. We have initiated a project aimed at a comparative analyzes of regulated pol II transcription in these two yeast species. Here we report on the reconstitution of a highly purified in vitro system for S. pombe pol II transcription. The purified S. pombe TFIIF contains three subunits, in contrast to metazoan TFIIF, which only contains two subunits. In S. cerevisiae this additional third subunit, Tfg3/Taf30/Anc1, appears only weakly associated with TFIIF, since it does not stay associated with the protein complex during gel filtration (24). Interestingly, the Tfg3/Taf30/Anc1 protein has also been identified as a member of the yeast SWI/SNF complex (30) as well as TFIID (24). The phenotypes of the tfg3 Ϫ S. cerevisiae strain, however, demonstrate that Tfg3 is not essential for many of the tasks associated with these complexes (24). Our finding that spTfg3 is absolutely required for reconstitution of S. pombe TFIIF in recombinant form could suggest that the protein rather play a role as an assembly factor of multiprotein complexes.
We recently reported that Mediator in S. cerevisiae and S. pombe exists in at least two specific forms (13). The smaller of these forms is the core Mediator, which may interact with pol II and form a holoenzyme. Another larger form, which contains four additional Mediator subunits, spSrb8, spTrap240, spSrb10, spSrb11, is always isolated devoid of pol II. These proteins collectively form the Srb8 -11 module, which was recently isolated in free form (31). We here demonstrate that Mediator isolated from a sptrap240 Ϫ deletion strain also lacks spSrb10. Apparently spTrap240 acts as an anchor for the other components of the module. Our isolation of these two well defined forms of Mediator also allows us to compare their effects on basal transcription in vitro. We find that core Medi- , and core TFIIH (60 ng). We used the S. pombe adh1 ϩ promoter followed by a G-less casette (200 ng) as our reporter construct. Individual transcription factors were omitted as indicated. Transcription levels relative to the complete transcription reaction are indicated below each lane. B, the effect of holoenzyme⌬spTrap240 and spSrb8 -11/Mediator on basal transcription. Transcription assay was performed as described in A except that basal factors and Mediator were preincubated for 15 min before the template was added. Reactions contained no Mediator (lanes 1 and 6), only holoenzyme⌬spTrap240 (lane 2), only spSrb8 -11/Mediator (lane 6), or combinations of holoenzyme⌬spTrap240 and spSrb8 -11/Mediator in a molar ratio of 4:1 (lane 3), 1:1 (lane 4), or 1:4 (lane 5). Transcription levels relative the transcription reaction without Mediator are indicated below each lane. C, the relative amounts of Mediator-and spSrb10associated kinase activity correlate well with the inhibitory effect observed in the in vitro transcription system. Immunoblotting was with the spSrb4 antibody. The CTD-kinase assays were done in the presence or absence of pol II as indicated. The transcription assays were performed as described in B, and individual Srb8 -11/Mediator fractions (1 l) were added as indicated. Transcription levels relative to the transcription reaction without spSrb8 -11/Mediator are indicated below each lane.
ator has a stimulatory effect on basal transcription, whereas the spSrb8 -11-containing Mediator has an inhibitory effect. This observation could lend support to our previous speculation that the key role of spSrb8 -11 may be to repress Mediator function prior to transcriptional activation (13).
Another possible role for the spSrb8 -11 submodule is in global down-regulation of transcription upon entry into stationary phase. Specific growth factors and the availability of essential nutrients, control eukaryotic cell proliferation. If either of these signals is lacking, cells may enter into a specialized non-dividing resting state, known as stationary phase or G 0 , characterized by a considerable down-regulation of transcription and protein synthesis (32). Interestingly, subunits of the Srb8 -11 have been shown essential for establishing stationary phase in S. cerevisiae (33). It is thus possible that the main role of the spSrb8 -11 module is to down-regulate transcription upon entry into stationary phase and that the in vitro data presented here reconstitute this specific event.