Yeast Nuclear Extract Contains Two Major Forms of RNA Polymerase II Mediator Complexes*

The yeast Mediator complex is required for transcription by RNA polymerase II (pol II) in vivo andin vitro. This complex of over 20 polypeptides associates with pol II and is recruited to transcription complexes at promoters. Previous isolations of yeast Mediator-containing complexes in different laboratories have identified several distinct complexes. To identify the major forms of Mediator in yeast, Mediator was isolated from nuclear extracts using a two-step chromatographic procedure, avoiding ion exchange chromatography and high salt conditions to prevent dissociation of subunits during purification. Components of the Mediator complexes were identified by mass spectrometry and Western analysis. The major form of Mediator, termed pol II·Med, contained pol II and Mediator, including the Srb8–11 module. A second lower molecular size complex was also identified, termed Mediator core (Medc), which lacked pol II, Srb8–11, Rox3, Nut1, and the Rgr1 module. Both of these complexes were active in transcription in vitro, although the Medc complex had significantly lower activity and could compete with the activity of the pol II·Med complexin vitro.


The yeast Mediator complex is required for transcription by RNA polymerase II (pol II) in vivo and in vitro.
This complex of over 20 polypeptides associates with pol II and is recruited to transcription complexes at promoters. Previous isolations of yeast Mediator-containing complexes in different laboratories have identified several distinct complexes. To identify the major forms of Mediator in yeast, Mediator was isolated from nuclear extracts using a two-step chromatographic procedure, avoiding ion exchange chromatography and high salt conditions to prevent dissociation of subunits during purification. Components of the Mediator complexes were identified by mass spectrometry and Western analysis. The major form of Mediator, termed pol II⅐Med, contained pol II and Mediator, including the Srb8 -11 module. A second lower molecular size complex was also identified, termed Mediator core (Medc), which lacked pol II, Srb8 -11, Rox3, Nut1, and the Rgr1 module. Both of these complexes were active in transcription in vitro, although the Medc complex had significantly lower activity and could compete with the activity of the pol II⅐Med complex in vitro.
In vivo, and in crude nuclear transcription extracts, the yeast Mediator behaves functionally as a general transcription factor. Heat shock of a strain containing a temperature-sensitive Srb4 allele results in a global decrease in pol II transcription equivalent to a mutation in pol II (12,13). Likewise, transcription in vitro is highly dependent on Srb2, -4, and -6 in nuclear extracts (14,15). In this nuclear extract system, mutation of Srb2, -4, or -6 blocks recruitment of factors except TFIIA and TFIID to the preinitiation complex (PIC) (15). After transcription initiation, Mediator along with TFIIA, TFIID, TFIIH, and TFIIE remain at the promoter forming a scaffold for recruitment of pol II and other missing factors required for transcription reinitiation (16). Specific activators can stabilize this scaffold complex (16). In highly purified in vitro systems, Mediator stimulates both basal and activated transcription as well as phosphorylation of the pol II CTD (17)(18)(19)(20). In these highly purified systems, Mediator isolated from strains carrying deletions of nonessential subunits is defective in response to specific transcriptional activators.
In the past several years, human homologs of the yeast Mediator complex have been identified by affinity purification, interaction with specific activators, and on the basis of coactivator activity. At least eight different human and mouse Mediator complexes have been identified (1,3,21). These complexes differ in their exact polypeptide composition but contain a common set of polypeptides, including homologs to yeast Rgr1 and Med7. Interestingly, even though Srb2, -4, -5, and -6 are crucial to yeast Mediator function, no homologs to these factors have yet been identified in any organism other than yeasts. Because both yeast and human Mediator subunits interact with specific transcription activators, a current model for activator function is that interaction between specific activators and specific Mediator subunits aids in recruitment of the transcription machinery to promoters (1,3). Furthermore, transcription reinitiation may be facilitated by interactions between activators and Mediator subunits that may stabilize the scaffold complex (16).
Several groups have reported biochemical fractionation of yeast Mediator-containing complexes from whole cell extracts. Young and coworkers followed purification of Mediator subunits by Western blot assay and isolated a complex termed RNA pol II holoenzyme. This complex contains pol II, Mediator (including the Srb8 -11 subunits), TFIIB, TFIIF, TFIIH, and SWI/SNF (18,22,23). In contrast, fractionation by Kornberg, Kim, and coworkers found Mediator (lacking the Srb8 -11 subunits) associated with pol II and sometimes TFIIF in a complex termed holopolymerase or holoenzyme (10,11,17). The reason for these differences in Mediator complexes is not clear, because similar methods involving ion exchange chromatography were used by all three groups (24,25). Finally, using a different purification method, Jaehning and coworkers (26) have isolated a complex termed Paf1 or alternative holoenzyme, which contains pol II, TFIIB, TFIIF, Cdc73, Paf1, Ccr4, Hpr1, and the Mediator subunits Gal11 and Sin4.
Both biochemical and genetic studies have suggested that the Mediator is composed of stable subcomplexes or modules. One module contains the dominant CTD suppressors Srb2, -4, -5, and -6 as well as Med6 and Rox3 (27,28). This subcomplex is stable in 1 M urea. A second module consists of Rgr1, Gal11, Sin4, Hrs1/Pgd1, and Med2. Evidence for this subcomplex is that pol II⅐Med with a C-terminal deletion of Rgr1 lacks Gal11, Sin4, and Hrs1 (20,29). Also, deletion of Gal11 causes loss of the Hrs1 subunit and vice versa (20). Deletion of the Sin4 subunit causes loss of both Med2 and Hrs1 and deletion of Med2 causes loss of the Hrs1 subunit (19). Another distinct module consists of Srb8 -11, which acts genetically as a repressor of pol II (30). Deletion of any one of these Srb genes suppresses deletions in the pol II CTD, and deletion of the Srb8 subunit gene causes loss of Srb10 and -11 in the holoenzyme (30). Whole genome array analysis showed that deletion of Srb10 derepressed transcription from a subset of yeast genes (13).
In light of the different Mediator-containing complexes isolated by biochemical fractionation and the distinct modules of Mediator subunits identified by biochemical and genetic studies, we have investigated the different forms of Mediator complexes in yeast. To avoid high salt conditions and multiple ion-exchange chromatographic steps used previously that could have caused dissociation of subunits during isolation, affinity chromatography and gel filtration were used to isolate yeast Mediator complexes under mild conditions. Using these methods, two major yeast Mediator-containing complexes were identified and found to be distinct from those complexes previously identified. One complex is similar to the holoenzyme isolated by Kornberg, Kim, and colleagues, but in addition contains the Srb8 -11 module. The other complex consists of a subset of Mediator components, and based on other work, appears to be the core Mediator complex.

Construction of FLAG-tagged Yeast
Strains-A PCR-mediated epitope-tagging method (31) was used to C-terminally tag the yeast Srb5, Srb4, and Rgr1 open reading frames. In brief, plasmid p3FLAG-KanMX (32), which contains a tagging cassette (three copies of the FLAG epitope upstream from the KanMX6 marker) was used as the PCR template. Each pair of PCR primers contained 45 bases homologous to the target gene and 18 bases complementary to the tagging cassette. PCR products were transformed to yeast, and the tagging cassette was introduced into the target gene by homologous recombination. Correct colonies were selected on G418 plates and identified by colony PCR. FLAG-tagged proteins were confirmed by Western blotting.
Extract Preparation and Purification of Protein Complexes-The Cterminal FLAG-tagged strains were grown in YPD media, and nuclear extracts were prepared as described previously (33). 100 mg of nuclear extract (20 -40 mg/ml of protein) was diluted to 30 ml with buffer A (20 mM HEPES-KOH, 10% glycerol, 0.01% Nonidet P-40, 2 mM DTT, and proteinase inhibitors) containing 0.25 M KOAc and incubated with 1 ml of anti-FLAG M2-agarose beads (Sigma) at 4°C for 4 h. After ten 30-ml washes with buffer A containing 0.3 M KOAc, proteins were eluted twice from the beads by incubation at 4°C for 1 h with buffer A containing 0.3 M KOAc and 0.5 mg/ml FLAG peptide (Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys) (synthesized by Research Genetics). The eluted proteins were concentrated by Centricon YM-30 (Millipore) and applied to a TSK-G4000SW XL (TosoHaas) gel filtration column at 0.37 ml/min equilibrated in buffer B (20 mM HEPES-KOH, pH 7.5, 0.25 mM EDTA, 0.3 M KOAc, 20% glycerol, 0.01% Nonidet P-40, 1 mM DTT, and proteinase inhibitors). 135-l fractions were collected, and 5 l was used in Western blot analysis. Fractions containing protein complex were combined and concentrated by Microcon YM-10 (Millipore) for use in the transcription assay and immobilized template assay. For the estimation of molecular size of protein complexes, blue dextran (2000 kDa), thyroglobulin (669 kDa), apoferritin (443 kDa), and bovine albumin (66 kDa) (Sigma) were loaded onto the same column as molecular size markers.
Mass Spectrometry Analysis of the pol II⅐Med and Medc Complexes-Peak fractions corresponding to pol II⅐Med or Medc from the gel filtration column were pooled and concentrated using Microcon YM-10 (Mil- lipore). During concentration, the buffer was exchanged by repeated dilution with 100 mM NH 4 HCO 3 with 8 M urea. Potential disulfide bonds were reduced and alkylated by incubation with 1 mM DTT at 37°C for 20 min followed by 10 mM iodoacetamide for 20 min at room temperature. Protein mixtures were digested with 100 ng of endoprotease Lys-C (Roche Molecular Biochemicals). After 3 h at 37°C, the urea concentration was adjusted to 2 M by dilution with 100 mM NH 4 HCO 3 , and CaCl 2 was added to 5 mM. Digestion was continued by addition of modified trypsin (Promega) to a final substrate-to-enzyme ratio of 25:1 (w/w) and incubated for 16 h at 37°C. After concentrating the samples to 50 l and adjusting the pH to 3 with trifluoroacetic acid, the peptides were desalted using C18 Ziptips (Millipore). Samples were lyophilized and resuspended in 5 l of buffer C (5% acetonitrile, 0.5% acetic acid, 0.005% heptafluorobutyric acid).
Identification of proteins was accomplished by microcolumn high performance liquid chromatography coupled to electrospray ionization tandem mass spectrometry and data base searching (34). A 100-by 365-m fused silica capillary (Polymetrics Inc.) with a tapered tip was packed to a length of 10 cm with a 5-m C18 reverse phase resin (Monitor). The sample was directly loaded onto the microcolumn by helium pressurization of the sample in a stainless-steel bomb. The mobile phase for high performance liquid chromatography (HPLC) consisted of buffer C and buffer D (99.5% acetonitrile, 0.5% acetic acid, and 0.005% heptafluorobutyric acid). A precolumn split was used to deliver a flow rate of 300 nl/min through the column. The HPLC pump was programmed to ramp solvent B from 10 to 38% in ϳ70 min. Electrospray ionization was done at a voltage of 1.8 kV. A Finnigan LCQ ion trap mass spectrometer was used to automatically acquire tandem mass spectra during the entire gradient run. Tandem mass spectra were searched against the Saccharomyces cerevisiae genome data base obtained from Stanford University with the SEQUEST program. Each high scoring peptide sequence and the corresponding tandem mass spectrum was manually inspected to ensure the match was correct.
In Vitro Transcription Assay-In vitro transcription assay and primer extension were performed as described (33). 2 The plasmid pSH515 was used as the template, which contains the HIS4 core promoter and a Gal4-binding site (15). Gal4-VP16 was prepared as described (15).
Immobilized Template Assay-Immobilized template derived from plasmid pSH515 was prepared as described previously (15). Preinitiation complex (PIC) formation experiments were performed essentially as described (15). 2 Briefly, 2.5-l immobilized templates were preincubated with 48 ng of Gal4-VP16 for 10 min at room temperature. At the same time, nuclear extract mix was prepared by adding 180 g of nuclear extract and purified proteins to a transcription mix containing 0.05% Nonidet P-40. Then, the immobilized template and 0.5 g of HaeIII-digested E. coli competitor DNA were added to the nuclear extract mix to a final volume of 50 l. Reactions were incubated at room temperature for 40 min to form PICs. After washing three times with transcription buffer containing 0.05% Nonidet P-40 and 2.5 mM DTT, the immobilized template was digested with 60 units of PstI at 37°C for 30 min. The liberated mixture was fractionated on a 4 -12% NuPAGE gel (Invitrogen), and the composition of PICs was analyzed by quantitative Western blotting.

RESULTS
Purification and Identification of Mediator Complexes-Previous biochemical fractionation of Mediator-containing complexes from yeast whole cell extracts have resulted in isolation of several complexes: the holoenzyme containing pol II, Srb8 -11, and the general transcription factors TFIIB, TFIIF, TFIIH (18), the holoenzyme lacking Srb8 -11 and all general transcription factors (11), the holoenzyme including a single general transcription factor TFIIF (17), and free Mediator (lacking both Srb8 -11 and pol II) (36). These fractionation methods have all used multiple ion exchange chromatography steps and high ionic strength buffers. It is possible that some of the differences in Mediator-containing complexes isolated by different laboratories arise from dissociation of subunits under these conditions. For example, it has been shown that a fraction of Mediator can be separated from pol II by ion exchange 2 Available from the S. Hahn laboratory on the web.  chromatography (36). To isolate Mediator complexes under gentler conditions, we first epitope-tagged the genomic copy of the subunits Srb5, Srb6, or Rgr1 with three copies of the FLAG epitope at the C terminus of these genes. As judged by cellular growth rates compared with wild-type strains, and by in vitro transcription using extracts made from these strains, the FLAG epitopes did not affect the function of any of these Mediator subunits. As a source for Mediator purification, we used yeast nuclear extracts, because these extracts have a much higher specific transcription activity compared with whole cell extracts used previously in holoenzyme and Mediator purification (data not shown). In addition, the nuclear extract system is clearly Mediator-dependent and has been used extensively for biochemical study of transcription initiation, reinitiation, and activation (15,16,37). The first step in Mediator isolation was affinity purification on anti-FLAG M2-agarose beads followed by peptide elution. Fig. 1 shows the small-scale affinity purification results from three yeast strains containing C-terminal FLAG-tagged Srb5, Srb6, and Rgr1, respectively. Recovery was measured by probing Western blots for Mediator subunits, pol II subunit (Rpb1), and several general transcription factors. Similar recoveries were obtained for all the strains: ϳ70% of Mediator subunits, 5-10% of pol II and Ͻ1% of both TFIIB and TFIIF were recovered. TFIIH and TBP were undetectable in the affinity-purified fractions (Table I). These results indicate that our FLAG-tagbased affinity purification is specific and efficient. A large-scale affinity-purified fraction from the FLAG-tagged Srb5 strain was then fractionated using gel filtration chromatography. In this step, we monitored each fraction by Western blotting using antibodies against Mediator subunits and general transcription factors (Fig. 2). The results revealed that the Mediator subunits existed in two major forms: one large complex (fractions 39 -47) corresponding to an apparent size of ϳ1.9 MDa, and one smaller complex (fractions 63-69) corresponding to an apparent size of ϳ0.55 MDa. As shown below, the large complex is a form of pol II⅐Mediator complex (termed pol II⅐Med), and the small complex consists of a core subset of Mediator subunits termed Mediator core (Medc).
The molar amounts of factors in the gel filtration fractions were determined by comparison with recombinant protein standards for Srb2, Srb4, Srb10, Rpb3 (pol II subunit), TFIIB, and Tfg1 (TFIIF subunit) ( Fig. 2; data not shown; Table II). The pol II⅐Med complex principally contains pol II, and Mediator, including the Srb8 -11 complex. As quantitated by Western blot of the pol II subunit Rpb3, pol II is estimated to be present at about one-half the molar ratio as compared with Srb2, -4, and -10. As shown below from comparison of the ratio of Mediator subunits to Rpb3 in preinitiation complexes (PICs), we believe that pol II is substoichiometric in the pol II⅐Med fractions compared with PICs. In contrast, a significantly lower amount of the general factors TFIIB and TFIIF was present in these fractions compared with Mediator subunits and pol II. We estimate that these factors are present at a level of 0.15-0.05 compared with other Mediator subunits. From these results, we conclude that the pol II⅐Med fractions contain primarily the pol II⅐Mediator complex, including Srb8 -11 and a much lower amount of larger complexes containing TFIIB and/or TFIIF. In contrast to the holoenzyme complex isolated by Young and colleagues (18, 23), we do not detect TFIIH or SWI/SNF in these fractions (not shown).
In contrast to the pol II⅐Med fractions, the Medc fractions lack a subset of Mediator subunits ( Fig. 2; Table II). Both the Gal11 and Srb10 Mediator subunits were missing from these fractions. In addition, Western blot assay of the Medc fraction with anti-Rgr1 and Sin4 showed the absence of these subunits (not shown).These results suggest the absence of both the Rgr1 and Srb8 -11 modules. These modules have been implicated in transcription regulation but may not be essential for transcription, because most of these subunits are not encoded by essential genes. Therefore, the Medc fraction likely represents a core subcomplex of yeast Mediator. Probing of the gel filtration fractions for the Rpb3 subunit of pol II showed that a fraction of pol II dissociated from either the pol II⅐Med or Medc complexes during the purification and eluted at the position expected for pol II enzyme.
The gel filtration chromatography of affinity-purified Mediator was repeated using buffer containing 0.5 M potassium acetate, compared with the results shown in Fig. 2    To more fully characterize the composition of pol II⅐Med and Medc, we employed microcolumn HPLC-electrospray ionization tandem mass spectrometry (LC/MS/MS) (38) (Table III). This method uses liquid chromatography and tandem mass spectrometry to separate and fragment peptides in a complex mixture, thus offering the ability to identify Ͼ100 proteins in a single run without use of SDS-PAGE. Analysis of the pol II⅐Med fractions identified all 20 Mediator subunits, 9 of 12 pol II subunits, and 3 of 4 subunits of the Srb8 -11 module (Table  IIIA). For the Medc complex, we identified 13 Mediator subunits, with 7 Mediator subunits (Rgr1, Gal11, Hrs1, Med2, Sin4, Rox3, and Nut1) not detected (Table IIIB). Specifically, the first five undetected subunits in Medc are all subunits of Rgr1 module, which was previously implicated in both positive and negative control of yeast transcription. Mass spectrometry analysis of the two fractions also identified ribosomal subunits as contaminants in the pol II⅐Med and Medc fractions. Specifically, many of the proteins of the large ribosomal subunit were identified in the pol II⅐Med fraction. Probing the FLAG affinitypurified fractions for TCM1 (RPL3), a protein in the large ribosomal subunit, indicated that Ͻ0.2% of total TCM1 in the starting extract contaminated the affinity-purified material (not shown). However, because ribosomal subunits are present at levels at least 500-fold over Mediator subunits (39), they represent a significant source of contamination. Probing of the gel filtration fractions with the TCM1 antisera showed near coelution of the large ribosomal subunit and pol II⅐Med, indi-cating the close similarity in molecular size of these two complexes (not shown).
Although the more sensitive Western blot assays showed detectable amounts of TFIIB and TFIIF in the pol II⅐Med fractions, these factors were not detectable in the mass spectrometry analysis, probably because they are quite substoichiometric compared with the other polypeptide subunits. A human homolog to the yeast Soh1 protein has been found in the thyroid hormone receptor-associated proteins-SRB/MED cofactor complex and PC2 Mediator complexes (3). However, yeast Soh1 was not found in either the pol II⅐Med or Medc complexes. It is possible that some subunits were not detected because they were substoichiometric or simply missed due to the complexity of the samples.
In Vitro Transcription Assay of Mediator-containing Complexes-To investigate the role of the two Mediator-containing complexes in transcription, in vitro transcription was performed using the activator Gal4-VP16 and the yeast HIS4 promoter. Transcription was assayed by complementation of yeast nuclear extracts made from strains containing deletions in either the SRB2 or MED9/CSE2 genes (Fig. 3). Both extracts give very low levels of transcription in the absence of any other factors (lanes 1 and 11). Addition of the pol II⅐Med fraction to the ⌬Srb2 extract increased transcription up to 18-fold (lanes 3-6) compared with 14-fold with addition of recombinant Srb2 (lane 2). In contrast, addition of the Medc complex increased transcription by up to 4-fold (lanes 8 -10). Similar results were seen upon addition of these complexes to the ⌬Med9 extract (lanes 12-15). When both the pol II⅐Med and Medc fractions were added to the ⌬Srb2 extract, the Medc complex inhibited the activity of pol II⅐Med (lane 7). From this result, we conclude that these complexes compete for binding to some common transcription factor, or compete for binding to the promoter, and that the Medc complex has inherently lower transcription activity. As discussed below, one possibility is that the Medc complex may function efficiently at only a subset of yeast promoters.
Composition Analysis of PICs and the pol II⅐Med Complex-The results above show that the pol II⅐Med fraction is nearly equivalent to the holoenzyme purified by Kornberg and Kim with the exception that our fraction contains Srb8 -11. Srb8 -11 has been shown to negatively regulate transcription under some conditions, and it was proposed that the repressive activity of some human Mediator complexes is due to the presence of Srb10 and Srb11 homologs (Cdk8 and cyclin C) (13,40,41). Because our pol II⅐Med fraction was shown to be active in transcription, we tested whether Srb8 -11 was recruited to PICs under these conditions. For this assay, we used the HIS4 promoter immobilized to magnetic beads (15,16). In this system, PICs are formed on the immobilized template using nuclear extracts, washed, and then eluted from the beads by digestion of the DNA with PstI which cuts just upstream of the single Gal4 binding site at the promoter. Factors bound to the template were assayed by Western blotting. As shown in Fig. 4, both the ⌬Med9 and ⌬Srb2 extracts have a severe defect in recruitment of pol II, TFIIB, and Mediator subunits to the promoter compared with wild-type extracts (Fig. 4; compare  lanes 1 and 4 with lanes 8 and 9). PIC formation was restored to wild-type levels by addition of the pol II⅐Med complex (lanes 2 and 6). Srb10 was clearly a component of the PICs formed in vitro both with wild-type extracts and extracts supplemented with the pol II⅐Med fraction.
When both the ⌬Med9 and ⌬Srb2 extracts were supplemented with the pol II⅐Med fraction, equivalent levels of Srb5-FLAG were seen in PICs compared with the levels seen when nuclear extracts made from the Srb5-flag strain were used to form PICs (compare lanes 2, 6, and 8). This suggests that the purified pol II⅐Med complex is stable when added to the extracts and that subunits do not exchange between this complex and other Mediator subunits preexisting in the extracts. This nonexchange likely explains why the Medc complex has low activity when added to nuclear extracts containing the remaining Mediator subunits.
To compare the relative amounts of subunits in the PICs to the ones in the purified pol II⅐Med fraction, we also analyzed the pol II⅐Med fraction in the same Western blot (Fig. 4, lane 7). This analysis shows that about 2-fold more Rpb3 was present on PICs than in the pol II⅐Med fraction. This suggests that pol II is somewhat unstable in the Mediator complex and dissociates upon fractionation by either affinity chromatography and/or gel filtration. In the complementation experiment, pol II⅐Med fractions containing pol II may be selectively recruited to the promoter, or the Mediator complex lacking pol II may associate with free pol II in the extract. In contrast to this behavior, Srb10 was substoichiometric in PICs compared with the pol II⅐Med complex (about a 2-fold difference between PICs and pol II⅐Med); (Fig. 4, compare lane 7 with lanes 2, 6, 8, 9). One possible explanation for this result is that the Srb8 -11 module may be unstable in the PIC and a fraction of it dissociates upon washing and isolation of the PICs. DISCUSSION The yeast Mediator complex is critical for both basal and activated transcription in vitro using a crude transcription system. In vivo, Mediator is necessary for nearly all pol II transcription with only a few reported exceptions (42,43). Related Mediator complexes have been found in mammalian cells where they may be central to integration of gene control signals. In both yeast and mammalian cells, Mediator com- FIG. 4. Composition of PICs and the pol II⅐Med. PICs were formed using the Immobilized Template Method. PICs and the pol II⅐Med complex were resolved by SDS-PAGE and analyzed by Western blotting using antibodies directed against the proteins indicated. Nuclear extracts were made from ⌬Med9, ⌬Srb2, Srb5-FLAG, and wildtype (WT) strains. For each reaction, 180 g of nuclear extract was incubated with immobilized templates for 40 min, along with recombinant rSrb2 (100 ng) or pol II⅐Med (2 l) where indicated, to form PICs. The PICs were then isolated and analyzed as described under "Materials and Methods." plexes have been isolated in multiple forms. Whether these multiple forms are pre-existing distinct complexes in vivo or whether they arise during purification is not known.
Here, we attempted to address the nature of the yeast Mediator-containing complexes using affinity purification and gel filtration chromatography, which avoids the high salt and ion exchange methods, used previously. As a starting material, we used yeast nuclear extracts, which have a higher specific activity in transcription than the previously used whole cell extracts. Using these methods, we found two major Mediatorcontaining complexes in nuclear extracts. The first complex, termed pol II⅐Med, migrates in gel filtration similarly to the large ribosomal subunit and contains principally pol II and Mediator, including the Srb8 -11 module. This fraction also contains detectable amounts of TFIIB and TFIIF, but at levels 1/7 to 1/20 the levels of other Mediator subunits analyzed. Because the gel filtration cannot separate large complexes of this size differing by only a few polypeptides, we propose that this pol II⅐Med fraction contains mainly a complex of pol II and Mediator (Fig. 5). Additionally, a much smaller amount of complexes containing either TFIIB and/or TFIIF coexist in this fraction. Therefore, this complex is nearly identical to the holoenzyme isolated by Kornberg, Kim, and coworkers with the exception that it includes the Srb8 -11 module. It is likely that this latter module was lost during their ion exchange purification. However, unlike the results reported by Young and colleagues, we find no detectable TFIIH or SWI/SNF in this fraction.
A novel subcomplex of Mediator was also identified in our work, termed Medc. This complex contains no pol II subunits and also lacks subunits of the Rgr1 module (Hrs1/Pgd1, Med2, Gal11, Sin4, Nut1, and Rgr1) as well as Rox3, Nut1, and the Srb8 -11 module. We estimate the molar amount of Medc to be 20 -30% of pol II⅐Med. The composition of the Medc complex is generally consistent with previous biochemical and genetic analysis of Mediator core subunits. These core subunits contains most essential Mediator genes and have global functions in transcription (3). They lack the Rgr1 regulatory module, which can have a selective effect on regulated transcription (3). The Medc complex is apparently pre-existing in the extract, because high salt buffers, which dissociate pol II from pol II⅐Med, do not increase the relative amount of this complex. Also, the Medc complex appears stable and does not appear to exchange subunits when added to extracts lacking either Srb2 or Med9. Although this Medc complex promotes transcription to a weaker extent than the pol II⅐Med complex, it is possible that this complex functions on only a subset of yeast genes in vivo. For example, the missing Mediator subunits with the exception of Rgr1 are all nonessential subunits (33). Another possibility is that this complex functions specifically in repression of certain genes, similarly to a proposed function for mammalian NAT complex (40). Further genetic and biochemical studies will be required to determine the function of this core Mediator complex.