Human Mediator Enhances Basal Transcription by Facilitating Recruitment of Transcription Factor IIB during Preinitiation Complex Assembly*

The multisubunit Mediator is a well established transcription coactivator for gene-specific activators. However, recent studies have shown that, although not essential for basal transcription by purified RNA polymerase II (pol II) and general initiation factors, Mediator is essential for basal transcription in nuclear extracts that contain a more physiological complement of factors (Mittler, G., Kremmer, E., Timmers, H. T., and Meisterernst, M. (2001) EMBO Rep. 2, 808–813; Baek, H. J., Malik, S., Qin, J., and Roeder, R. G. (2002) Mol. Cell. Biol. 22, 2842–2852). Here, mechanistic studies with immobilized DNA templates, purified factors, and factor-depleted HeLa extracts have shown (i) that Mediator enhancement of basal transcription correlates with Mediator-dependent recruitment of pol II and general initiation factors (transcription factor (TF) IIB and TFIIE) to the promoter; (ii) that Mediator and TFIIB, which both interact with pol II, are jointly required for pol II recruitment to the promoter and that TFIIB recruitment is Mediator-dependent, whereas Mediator recruitment is TFIIB-independent; (iii) that a high level of TFIIB can bypass the Mediator requirement for basal transcription and pol II recruitment in nuclear extract, thus indicating a conditional restriction of TFIIB function and a key role of Mediator in overcoming this restriction; and (iv) that an earlier rate-limiting step involves formation of a TFIID-Mediator-promoter complex. These results support a stepwise assembly model, rather than a preformed holoenzyme model, for Mediator-dependent assembly of a basal preinitiation complex and, more important, identify a step involving TFIIB as a key site of action of Mediator.

The transcription of eukaryotic genes is mediated by functional interactions of the general transcription machinery with common core promoter (e.g. TATA) elements and further regulated by gene-specific factors bound to distal regulatory elements (1). Studies with purified factors have demonstrated that pol II 4 and cognate initiation factors TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH (general transcription machinery) are necessary and sufficient for robust basal (activator-independent) transcription from TATA-containing DNA templates and that the formation of a functional PIC involves the ordered stepwise assembly of these factors (2,3). In contrast, activator-dependent transcription from DNA templates was found to require additional cofactors, the most prominent of which is the multisubunit Mediator complex, which is conserved from yeast to human (4,5). Early demonstrations of its interaction with pol II (6,7), along with later demonstrations of direct interactions with activators (4,8), led to the notion that Mediator serves as a bridge between DNA-binding regulatory factors and the general transcription machinery to facilitate formation and/or function of the PIC (4,5). Various biochemical and genetic studies have substantiated this view.
Despite its paramount role in activator-dependent transcription, Mediator also has been implicated in basal transcription events by both biochemical and genetic assays. Thus, consistent with the ability of yeast Mediator to enhance basal transcription in a purified system (7), mutations in Mediator subunits were shown to eliminate or reduce basal transcription in crude nuclear extracts (9,10). This is consistent with genetic studies implicating yeast Mediator in transcription of essentially all genes in yeast (11). Similarly, human Mediator has also been shown to be essential for basal transcription in nuclear extracts (12,13). As nuclear extracts contain a more natural complement of cellular factors, these assays likely give a more accurate reflection of Mediator requirements in living cells and support the emerging view that Mediator functions as a general transcription factor under more physiological assay conditions.
The mechanism(s) by which Mediator facilitates basal transcription remain to be fully understood and could include direct or indirect functions at any of the steps (PIC formation, promoter melting, initiation, promoter clearance, etc.) leading to productive transcription (3). The isolation of stable Mediator-pol II (holoenzyme) complexes, initially in yeast (6,7) and later in mammalian cells (4, 14 -16), suggested the possibility of Mediator recruitment to, and possible stabilization of, the PIC through Mediator-pol II interactions. A role for Mediator in basal PIC formation or function was also suggested by the ability of yeast Mediator to enhance TFIIH-mediated phosphorylation of the pol II CTD (7) and an associated CTD requirement both for basal transcription in yeast and mammalian nuclear extracts (17) and for formation (activator-dependent) of the PIC in yeast nuclear extracts (10). Consistent with the demonstration of cooperativity between human Mediator and TAF components of TFIID in basal transcription in nuclear extracts (13), human TFIID and Mediator have been reported to act cooperatively in activatordependent PIC assembly (18). These observations, along with structural studies of the transcription machinery (19), are consistent with the view that Mediator interactions with pol II and general initiation factors may facilitate formation and possibly function (initiation, promoter clearance) of the PIC. A recent study has demonstrated activator-independent (basal) functions of Mediator in both PIC formation and re-initiation in yeast extracts (20).
Still unexplained when this study initiated was why basal transcrip-tion from DNA templates absolutely requires Mediator in assays with nuclear extracts, but not in assays with purified factors. A direct positive role has been suggested by the recently demonstrated ability of Mediator to stimulate basal transcription (4,7,21) or to compensate for limiting amounts of general initiation factors (17) in basal transcription assays reconstituted with purified factors. In contrast, an indirect antirepression effect was suggested from yeast genetic studies in which negative cofactors (NC2 and Not1) that inhibit the general transcription machinery were isolated as suppressors of a mutant Mediator subunit (22,23). Although a recent study with yeast nuclear extract failed to show Mediator effects through Not1 and NC2 complexes (24), it did not eliminate possible effects through other known (25,26) or unknown factors that may place constraints on PIC formation or function. Another question of importance in relation to Mediator function and potential targets is whether the poorly characterized basal PIC assembly pathway in nuclear extracts is identical to the pathway established with purified factors. The latter involves TATA element recognition by TBP, generally in the context of TFIID and with possible stabilization by TFIIA, with subsequent sequential binding of TFIIB (to TBP), TFIIF-pol II (to TFIIB), TFIIE (to pol II), and TFIIH (to TFIIE) (2,3). In this pathway, pol II recruitment is critically dependent upon prior binding of TFIIB to the TBP/TFIID-promoter complex, whereas TFIID-Mediator and Mediator-pol II interactions (see above) might possibly result in less of a dependence on TFIIB for pol II recruitment in nuclear extracts. Thus, and relevant to this study, there could be some redundancy or cooperativity between TFIIB and Mediator in PIC formation. Related studies of activator-driven PIC assembly in yeast extracts have argued in favor of a holoenzyme model (6) involving joint interdependent recruitment of pol II, Mediator, and a subset of general initiation factors (10), although cell-based chromatin immunoprecipitation assays have clearly shown that Mediator and pol II are not simultaneously recruited on at least some genes (27)(28)(29).
In this study, we used immobilized template assays in conjunction with purified factors and factor-depleted nuclear extracts to investigate the PIC assembly pathway and basis of the Mediator requirement for basal transcription in extracts from human cells. Our results demonstrate a Mediator function in basal PIC formation that surprisingly involves facilitated recruitment of TFIIB. They further demonstrate a rate-limiting step in basal transcription that appears to involve formation of a TFIID-Mediator-promoter complex.
Immobilized Templates-Streptavidin-Sepharose beads (Amersham Biosciences) were concentrated by centrifugation and washed twice with 300 l of Buffer T (10 mM Tris-HCl (pH 7.5), 1 mM EDTA, and 1 M NaCl) supplemented with 0.05% Nonidet P-40 (10). The beads were resuspended in Buffer T (0.14 ml of beads/ml of buffer) and incubated with 1.5 pmol of biotinylated template/l of beads in Buffer T for 30 min at room temperature with constant agitation. The immobilized templates were concentrated by centrifugation and washed three times with 300 l of Buffer T. The immobilized templates were suspended in block buffer (10 mM Hepes-KOH, pH 7.6, 100 mM potassium glutamate, 10 mM magnesium acetate, 5 mM EGTA 3.5% glycerol, 60 mg/ml casein, and 5 mg/ml polyvinylpyrrolidone) (10 ml/ml of beads) for 15 min at room temperature with constant agitation (10). The beads were concentrated by centrifugation, washed three times with transcription buffer (20 mM Hepes-KOH (pH 7.6), 4 mM MgCl 2 , 60 mM KCl, 0.08 mM EDTA, 8 mM dithiothreitol, 0.4 mg/ml bovine serum albumin, 0.05% Nonidet P-40, and 10% glycerol), and resuspended in BC100 (30) at 0.5 ml/ml. Immobilized templates were prepared fresh before each experiment.
In Vitro Transcription with Immobilized Templates-Transcription reactions were carried out in two steps. First, preincubation reactions were set up and contained (in a final volume of 25 l) 2 l of immobilized template, 500 ng of sonicated Escherichia coli DNA, 4 l of HeLa nuclear extract (at ϳ10 mg/ml), 20 mM Hepes-KOH (pH 8.2), 11-16% glycerol, 4 mM MgCl 2 , 60 mM KCl, 8 mM dithiothreitol, 0.4 mg/ml bovine serum albumin, and 20 units of RNasin (Promega). Reactions were incubated at 30°C for 1 h, and immobilized templates were washed three times with 300 l of transcription buffer. Second, these immobilized templates were further incubated in 25-l reactions containing 20 mM Hepes-KOH (pH 8. Analysis of Proteins on Immobilized Templates-For analysis of proteins bound to immobilized templates, scaled-up (100 l) transcription reactions were used. After incubation of extract with immobilized templates, the templates were concentrated by centrifugation and washed three times with 300 l of transcription buffer. Templates were resuspended in 50 l of Buffer 3 (New England Biolabs) with 60 units of EcoRI. After incubation for 30 min at 37°C with constant agitation, the supernatants were collected on compact reaction columns (USB Corp.). Proteins in the supernatants were concentrated by precipitation with trichloroacetic acid and then redissolved in 1ϫ SDS loading buffer. After boiling, samples were resolved on 4 -16% polyacrylamide gels (Invitrogen). Proteins were analyzed by immunoblotting.
Time Course Assays-Immobilized DNA templates were prepared as described above with minor modifications for convenience of handling. Streptavidin-coupled M-280 Dynabeads (Dynal) were concentrated by Dynal MPC-S and washed twice with 300 l of Buffer T supplemented with 0.05% Nonidet P-40. Beads were resuspended in Buffer T at 5 g/ml and incubated with 10 fmol of biotinylated template/g of beads in Buffer T for 30 min at room temperature with constant agitation. The immobilized templates were concentrated by MPC-S and washed three times with 300 l of Buffer T. The immobilized templates were blocked in block buffer (0.5 l/g of beads) for 15 min at room temperature with constant agitation. The beads were concentrated by MPC-S, washed three times with transcription buffer, and resuspended in BC100 at 0.1 ml/g. This immobilized template was incubated with HeLa nuclear extract for time t. After washing the immobilized template twice with transcription buffer, either transcription was initiated by addition of NTPs, or bound proteins were analyzed by immunoblotting.
Immunodepletion of Nuclear Extracts-Nuclear extracts depleted of Mediator (referred to as NE(⌬MED)) or TFIID and Mediator (NE(⌬IID/ ⌬MED)) were prepared as described (13). For preparation of nuclear extract lacking TFIIB (NE(⌬IIB)), 200 l of anti-TFIIB antisera (provided by Dr. Sohail Malik) were purified on 100 l of protein A-Sepharose (Amersham Biosciences) and then cross-linked with dimethyl pimelimidate FIGURE 1. TATA-and Mediator-dependent recruitment of pol II, TFIIB, and TFIIE. A, experimental scheme. Two different bead-immobilized templates with intact (iML⌬53TATAϩ) and mutated (iML⌬53TATAϪ) TATA boxes were prepared from an adenovirus major late core promoter (TATA and Inr) construct with a downstream G-less cassette (13). Templates were incubated with 50 g of NE(⌬IID/⌬MED) supplemented with amounts of TBP and Mediator(f:TRAP220AB) comparable with the levels in the control extract as indicated. For transcription reactions, immobilized templates were washed and analyzed as described under "Materials and Methods." For immunoblot analyses, reactions were scaled up by 3-fold; and after incubation, immobilized templates were washed, digested with EcoRI, and subjected to immunoblotting. B, purified Mediator preparations analyzed by SDS-PAGE and silver staining. Mediator(f:NUT2) was purified from cells expressing a FLAG-tagged MED10/NUT2 subunit (31), and Mediator(f:TRAP220AB) was purified from cells expressing a FLAGtagged N-terminal (AB) fragment of MED1/TRAP220 (32). SM, size markers. C, immobilized template recruitment assay with NE(⌬IID/⌬MED) supplemented with recombinant TBP and purified Mediator(f:TRAP220AB) at levels equivalent to those of the corresponding endogenous factors in the control extract. Antibodies to proteins indicated on the right were used for immunoblot analyses. D, transcription assay with NE(⌬IID/⌬MED) supplemented with TBP and purified Mediator(f:TRAP220AB) as indicated. Relative transcription (txn) levels were determined by PhosphorImager analysis.

RESULTS
pol II, TFIIB, and TFIIE Are Recruited to a TATA-containing Promoter in a Mediator-dependent Manner-Whereas pol II and general transcription factors suffice for robust basal transcription in purified systems (2,3), basal transcription in a crude HeLa nuclear extract-based transcription system requires, in addition, human Mediator, as shown in other studies (12,13). Because PIC assembly is an essential and generally rate-limiting step in the overall transcription reaction (33,34), we first investigated the possibility that Mediator enhances basal transcription in the nuclear extract assay by contributing to PIC assembly.
To this end, we employed an immobilized template assay (10) that allows, after template incubation with nuclear extract and subsequent washing, assessment of both bound factors (by immunoblotting) and transcription activity (by incubation with nucleoside triphosphates). To avoid the problem of substantial TATA-independent binding of endogenous TFIID and Mediator under basal transcription assay conditions (data not shown) and to assess the requirement for Mediator in PIC formation, we employed NE(⌬IID/⌬MED) supplemented with TBP and variably with highly purified Mediator. The TFIID-and Mediator-depleted extract has been described previously (13), and the affinity-purified Mediator(f:TRAP220AB) (32) and Mediator(f:NUT2) (31) preparations used in this study are shown in Fig. 1B.
As shown in Fig This likely reflects the nonspecific DNA binding activities of these factors and has precluded assessment in this study of the possible Mediator dependence of their recruitment to the PIC at the TATA element. Nonetheless, the results clearly indicate that, in the nuclear extract context, Mediator plays an important role in recruitment to the basal promoter of key components (minimally TFIIB, TFIIE, and pol II) of the PIC.

Mediator-dependent Basal Transcription Activity in Nuclear Extracts Correlates with Mediator-dependent Recruitment of Factors to the PIC-A
corresponding analysis of the transcription activity of the PIC assembled on the immobilized template revealed a close correlation with the factor binding results. Thus, basal transcription showed an almost complete (49-fold) dependence upon an intact TATA element (Fig. 1D, lane 3 versus lane 2) and a Ͼ5-fold dependence upon Mediator (lane 3 versus lane 4), clearly reflecting the TATA-and Mediator-dependent recruitment of pol II and the GTFs observed in Fig. 1C. However, because these assays contained TBP in place of TFIID for technical reasons (see above) and because TAFs contribute to basal transcription in nuclear extracts (13), these results may underestimate the true contribution of Mediator to basal transcription.
To confirm and extend the single time point correlation between transcription factor recruitment and transcription activity, we compared the temporal recruitment of transcription factors with the temporal acquisition of transcription activity according to the scheme in Fig. 2A. This analysis employed a Mediator-depleted nuclear extract supplemented with purified Mediator(f:TRAP220AB) to avoid the nonspecific binding activity of endogenous Mediator. As shown in Fig. 2 (B and C), the transcription activity of the immobilized template complex increased linearly (up to 8-fold in 32 min) with the time t of preincubation (prior to washing) with the nuclear extract. As shown in Fig. 2D, the recruitment of TFIIB, TFIIE, pol II, and Mediator showed comparable increases (4 -8-fold) over the same time period. In contrast and for reasons (nonspecific DNA binding) described above, TFIIH and TBP/ TFIID showed constitutively high binding. The close correlation between the temporal formation of functional PICs (exhibiting transcription activity) and the temporal recruitment of Mediator and key PIC components again supports the hypothesis that Mediator-dependent recruitment of PIC components governs the level of basal transcription in a nuclear extract.
Mediator Contributes to Activator-dependent Transcription through Enhanced Recruitment of GTFs-The above experiments investigated how Mediator contributes to basal transcription. If this is attributed, as suggested, to Mediator-dependent recruitment of GTFs such as pol II, TFIIB, and TFIIE, the same molecular mechanism might also be relevant to the previously reported role of Mediator in transcription activation by DNA-binding regulatory factors (4). Indeed, previous studies have documented activator-and Mediator-dependent recruitment of pol II and GTFs (Refs. 35 and 36 and references therein).
To explore this possibility in our system, the effects of the transcription activator Gal4-VP16 were analyzed in the immobilized template assay outlined in Fig. 3A. In this case, an immobilized template (iG5HML) containing five Gal4-binding sites upstream of a core promoter was incubated in the Mediator-free nuclear extract supple-  Thus, without Mediator and the possible contribution of a previously described direct interaction between Gal4-VP16 and Mediator (37), Gal4-VP16 does not effectively recruit pol II and other GTFs. Therefore, Mediator appears to provide an important link that is necessary for the transcription activator to recruit pol II and GTFs to the promoter in the context of a nuclear extract.
The cooperative effects of Gal4-VP16 and Mediator in transcription factor recruitment were also reflected in the transcription activity of the isolated immobilized templates (Fig. 3C). The 44-fold elevation in transcription observed in the presence of Gal4-VP16 and Mediator (lane 4 versus lane 1) was reduced by ϳ10-fold in the absence of either Mediator (lane 2) or activator (lane 3). Thus, a transcription-enhancing mechanism involving Mediator-dependent recruitment of GTFs and pol II seems to be relevant to both basal transcription and activator-dependent transcription in nuclear extracts.

Elevated Levels of TFIIB Can Bypass the Normal Mediator Requirement for PIC Assembly and Basal Transcription in Nuclear Extracts-Studies
with purified factors established a key and direct role for TFIIB in pol II recruitment to the PIC (2, 3), whereas demonstrations of direct Mediatorpol II interactions are also consistent with a direct role for Mediator in pol II recruitment (4,5). These results thus suggest similar or overlapping functions for TFIIB and Mediator in pol II recruitment. However, Mediator is essential for basal transcription and PIC assembly, including TFIIB recruit-ment, only in nuclear extracts and not in purified systems. This suggests some basic difference in the PIC assembly pathways or in factors that regulate them in the two systems. One possibility is that nuclear extracts contain a factor(s) that constrains recruitment of TFIIB to the PIC and that Mediator counteracts, either directly or indirectly, the inhibitory activity of this factor. In this case, a higher than normal level of TFIIB might be expected to overcome the inhibitory effect without the aid of Mediator in the nuclear extract assay.  . Transcription factor recruitment contributes to transcription activation by Gal4-VP16. A, experimental scheme. An immobilized template (iG5HML) containing five copies of the Gal4-binding site, a hybrid major late promoter, and a G-less cassette (13) was prepared. The iG5HML template was incubated for 1 h at 30°C with NE(⌬MED) alone or supplemented with Mediator(f:TRAP220AB) in the presence or absence of Gal4-VP16. After washing, the template was subjected to in vitro transcription and immunoblot assays as indicated in B and C. B, transcription factor recruitment. iG5HML was incubated with NE(⌬MED) supplemented with Mediator(f:TRAP220AB) or Gal4-VP16 (40 ng) as indicated, and bound factors were assayed with antibodies to proteins indicated on the right. C, single round transcription with or without Gal4-VP16. iG5HML was incubated with NE(⌬MED) supplemented with Mediator and 10 ng of Gal4-VP16 as indicated. After washing, immobilized templates were subjected to standard transcription assays, except that Sarkosyl (0.08% final concentration) was added 30 s after transcription initiation to prevent transcription re-initiation. Relative transcription (txn) levels determined by PhosphorImager analysis are indicated.
To test this possibility, basal transcription was measured in the Mediator-depleted nuclear extract supplemented with recombinant TFIIB. The analysis in Fig. 4A shows a dose-dependent increase in transcription activity, with a normal level of transcription (equivalent to that in a control nuclear extract) recovered at a 25-fold molar excess of recombinant TFIIB over endogenous TFIIB. In a reciprocal assay involving complementation of the TFIIB-depleted extract (see Fig. 5 below) with excess Mediator, up to a 10-fold molar excess of Mediator effected only a modest increase in transcription (equivalent to 17% of the control level) in undepleted nuclear extracts (Fig. 4D). Given the essential role for TFIIB, the activity observed almost certainly reflects some residual TFIIB in the depleted extract. As a control and emphasizing the specificity of the Mediator bypass for TFIIB, a 25-fold excess of TFIIF had essentially no effect on the residual basal activity in the Mediator-depleted extract (Fig. 4C).
To investigate the molecular basis for the ability of excess TFIIB to bypass the Mediator dependence, an immobilized template factor recruitment assay (as outlined in Fig. 1A) was employed. As shown in Fig. 4B, addition of a 25-fold excess of TFIIB to the Mediator-depleted extract resulted in significantly enhanced recruitment of TFIIE and pol II to the immobilized template (lane 4 versus lane 3). This result indicates that the recovery of transcription in a Mediator-deficient extract following addition of excess TFIIB is attributed primarily to enhanced PIC assembly and that this mechanism mimics the normal contribution of Mediator to basal transcription enhancement. Thus, excess TFIIB can bypass the Mediator requirement for the recruitment of pol II and TFIIE and likely other general initiation factors during PIC assembly in nuclear extracts.
The above results indicate that the direct effect of Mediator in a crude nuclear extract is related to facilitation of TFIIB recruitment during PIC assembly. However, given indications of TAF-dependent synergy between TFIID and Mediator in basal transcription (13) and the possibility of TAF-dependent interactions when these factors are promoterbound (18), it was of interest to determine whether the ability of excess TFIIB to bypass the Mediator requirement for basal transcription in nuclear extracts is TAF-dependent. As shown in Fig. 4E, a 25-fold excess of recombinant TFIIB alone largely restored basal transcription to the TFIID-and Mediator-depleted nuclear extract supplemented with TBP (lane 4 versus lane 1). As a control, a comparable 25-fold excess of TFIIF failed to restore any transcription to the TFIID-and Mediator-depleted extract in the presence of TBP (Fig. 4F). Hence, TAFs are not essential either for Mediator or for excess TFIIB in the absence of Mediator to effect basal transcription in nuclear extracts. The TAF-Mediator synergy in basal transcription (13) must therefore reflect effects that are superimposed on Mediator or high TFIIB effects observed with TBP. Overall, the results with either TBP or TFIID are consistent with the function of Mediator and TFIIB in basal transcription through a common or linked mechanism.
An Essential Role for TFIIB in Recruitment of pol II and TFIIE to the Promoter Is Maintained during Mediator-dependent Basal Transcription-TFIIB is established as an essential GTF both in systems reconstituted with purified factors (2, 3) and in crude nuclear extract systems (38). However, its mechanism of action in crude nuclear extracts that contain more physiological concentrations of the various general initiation factors and other factors such as Mediator has not been investigated in any detail. This is particularly important because documented interactions of Mediator with pol II (4, 5, 14 -16, 19) offer a possible alternative to a strictly TFIIB-based mechanism for pol II recruitment.
For further analysis, nuclear extracts were depleted of TFIIB with anti-TFIIB antibody. The analysis in Fig. 5A shows that TFIIB was selectively depleted, with no detectable effects on the levels of other factors (lane 3 versus lane 2) and no significant association of other factors with the anti-TFIIB immunoprecipitate relative to the preimmune serum immunoprecipitate (lane 5 versus lane 4). As expected, TFIIB depletion effected a significant reduction of basal transcription that was fully restored by addition of recombinant TFIIB at a level equivalent to that in a normal nuclear extract (Fig. 5B). Thus, reduction of basal transcription in the TFIIB-depleted nuclear extract must be due primarily to a deficiency of endogenous TFIIB.
To assess the role of TFIIB in pol II recruitment in the nuclear extract system, in which basal transcription and pol II recruitment are dependent upon Mediator, the TFIIB-depleted extract was used for immobilized template factor recruitment and transcription assays as outlined in Fig. 5C. The analysis in Fig. 5D shows that TFIIE and pol II recruitment was compromised in the TFIIB-depleted extract relative to the control extract (lane 2 versus lane 1), but that this recruitment was significantly enhanced (5-and 3-fold, respectively) by inclusion of a normal nuclear extract equivalent of recombinant TFIIB (lane 3 versus lane 2). The recruitment of other factors was constitutively high for reasons (nonspecific DNA binding) mentioned above. This effect of TFIIB on PIC assembly was reflected in basal transcription from the isolated immobilized DNA template. Transcription was enhanced 16-fold by addition of TFIIB (Fig. 5E), in agreement with the previous result with supercoiled DNA in the standard assay (Fig. 5B). Strictly comparable results were obtained when TFIID-, TFIIB-, and Mediator-depleted extracts were complemented variably with TBP, TFIIB, and Mediator(f: TRAP220AB) (to preclude any nonspecific binding of endogenous TFIID and Mediator) (Fig. 5, F and G).
These results demonstrate that, despite the potential for Mediatorbased pol II recruitment, TFIIB is still essential for basal transcription and PIC formation (including pol II recruitment) in a crude nuclear extract just as in a purified reconstituted system. However, these results do not exclude cooperative functions between Mediator and TFIIB in pol II recruitment. Moreover and significantly, the results show that, although recruitment of TFIIB is dependent upon Mediator, specific (TBP-and TATA-dependent; see Fig. 1C) recruitment of Mediator is not dependent upon TFIIB.

Formation of a TFIID-Mediator-Promoter Complex as a Rate-limiting
Step in Basal Transcription-The results described above indicate TATA-and Mediator-dependent recruitment (minimally TFIIB, TFIIE, and pol II) to a promoter for basal transcription. This in turn suggests assembly of TFIID and Mediator on the promoter as a prerequisite for subsequent recruitment of other factors to the PIC and as a possible rate-limiting step in transcription. Support for this idea is provided by a previous study of TFIID-Mediator cooperativity in basal transcription in nuclear extracts (13) and by a study showing that TFIID and Mediator assemble cooperatively on promoter DNA in response to transcription activators (18). We investigated this issue in our nuclear extract-based system using an immobilized template assay conducted according to the protocol in Fig. 6A. In this assay, a 30-min preincubation of the template with TFIID in the presence or absence of Mediator was followed by preincubation for time t with the TFIID-and Mediator-depleted nuclear extract and, if omitted in the first step, with added Mediator. The immobilized template was washed after each preincubation step to exclude the effect of unbound transcription factors on subsequent steps and to ensure that transcription resulted from events occurring during the allocated incubation time. Moreover, to prevent transcription reinitiation, Sarkosyl was added 2 min after addition of NTPs.
As shown in Fig. 6 (B and D), transcription activity from the immobilized template that was preincubated with only TFIID in the first step increased almost linearly with incubation times up to 32 min in the second step. In contrast, transcription from the immobilized template that was preincubated with both TFIID and Mediator in the first step plateaued (at a high level) following only a short incubation (Ͻ10 min) in the second step (Fig. 6,  C and D). From these results, it can be inferred that the transcription activity of the TFIID-pretreated template was roughly determined by the time spent to recruit Mediator to the promoter. Thus, assembly of a TFIID-and Mediator-containing complex on the promoter appears to be a rate-limit- FIGURE 5. TFIIB is essential for pol II and TFIIE recruitment and enhanced basal transcription in nuclear extracts. A, depletion of TFIIB from HeLa nuclear extract. HeLa nuclear extract was incubated with anti-TFIIB antibody-coupled resin at 4°C for 4 h in BC300 (30). The final flow-through, NE(⌬IIB), was collected and dialyzed against BC100. The bound protein on the resin was eluted with 100 mM glycine (pH 2.5) after extensive washing of the resin with BC300. The control nuclear extract (NE), preimmune antibody-depleted extract (NE(preimmune)), NE(⌬IIB), immobilized preimmune antibody eluate (␣-PRE eluate), and immobilized anti-TFIIB antibody eluate (␣-TFIIB eluate) were analyzed by immunoblotting with antibodies to proteins indicated on the right. B, TFIIB-dependent basal transcription in NE(⌬IIB). Standard transcription assays were performed with G5HML and adenovirus major late ⌬53 plasmid templates (13) in the control nuclear extract or in NE(⌬IIB) supplemented with various amounts of recombinant TFIIB (rTFIIB). The reaction in lane 5 contained an amount of recombinant TFIIB comparable with that of endogenous TFIIB in the control nuclear extract. Relative transcription (txn) levels were measured by PhosphorImager analysis. C, experimental scheme for immobilized template assays with NE(⌬IIB). An immobilized template (iML⌬53TATAϩ) was incubated with NE(⌬IIB) alone or supplemented with recombinant TFIIB for 1 h at 30°C. After washing, transcription and immunoblot analyses were performed as described under "Materials and Methods." D, immobilized template factor recruitment analysis showing that TFIIB is required for the normal recruitment of pol II and TFIIE in nuclear extracts. An immobilized template was incubated with 200 g of NE(⌬IIB) with or without an equimolar amount (relative to the control nuclear extract) of recombinant TFIIB in a 200-l reaction. After washing, bound proteins were analyzed by immunoblotting. The mobility of recombinant TFIIB in lane 3 is higher than that of endogenous TFIIB in lane 1 due to the hexahistidine (H6) tag. E, basal transcription from an immobilized template requires TFIIB. Transcription was performed in standard assays with NE(⌬IIB) and an amount of recombinant TFIIB equivalent to that in the control nuclear extract. The result was visualized by autoradiography, and relative transcription levels were measured by PhosphorImager analysis. F, immobilized template assay showing that TFIIB is not required for the recruitment of Mediator in nuclear extracts. The analysis was conducted as described for D, except that reactions contained NE(⌬IIB/⌬IID/⌬MED), recombinant TBP, and Mediator(f:TRAP220AB) in place of NE(⌬IIB), resulting in TATA-and TBP-dependent recruitment of Mediator (ectopic) (see Fig. 1C). G, basal transcription from an immobilized template requires TFIIB. The analysis was conducted as described for E, except that reactions contained NE(⌬IIB/⌬IID/⌬MED), recombinant TBP, and Mediator(f:TRAP220AB) in place of NE(⌬IIB).
ing step in transcription and a prerequisite for recruitment of other transcription factors and formation of a functional PIC.
Mediator has not been reported previously to have an intrinsic capability for sequence-specific DNA binding, and DNase I footprinting studies failed to show any independent interactions on the adenovirus major late promoter (data not shown). Therefore, the above results and those of Johnson et al. (18) indicate the likelihood of a direct interaction between TFIID and Mediator as a basis for subsequent recruitment of pol II and other general initiation factors in nuclear extracts. To explore this issue, we employed an immobilized template assay involving template incubation with Mediator in the presence or absence of TFIID, followed by immunoblotting to score bound proteins. As shown in Fig. 6E, Mediator recruitment to the template was increased by 3-4-fold in a TFIID-dependent manner (Fig. 6E). Therefore, it seems likely that Mediator and TFIID directly interact in the context of a DNA template. A similar analysis in which TBP replaced TFIID showed no TBP-enhanced recruitment of Mediator (Fig. 6F), thus indicating a role for TAFs in TFIID-dependent recruitment of the Mediator.

DISCUSSION
The function of Mediator in activator-dependent transcription is well established, and most studies of its mechanism of action have been in this context. These studies have established Mediator recruitment through direct interactions with promoter-bound activators and subsequent interactions with the basal transcription machinery (4,5,19) that facilitate both PIC formation (10,35,36) and post-recruitment functions (20,21,36). However, the demonstration of Mediator function in basal transcription and mechanistic studies in this context offer the promise of revealing novel aspects of Mediator function that are likely to be relevant to activator-dependent transcription as well. In following up our earlier demonstration (13) of Mediator-dependent basal transcription in HeLa nuclear extracts, as well as a strong synergy between Mediator and TAFs, we have demonstrated (i) that Mediator enhances basal transcription by facilitating recruitment of pol II and corresponding general initiation factors (minimally TFIIB and TFIIE), (ii) that elevated levels of ectopic TFIIB can bypass the Mediator requirement for both basal transcription and pol II recruitment, and (iii) that recruitment of FIGURE 6. Mediator recruitment to a TFIID-bound template is a rate-limiting step in basal transcription. A, experimental scheme. An immobilized template (Imm. Temp.; iML⌬53TATAϩ) was preincubated with TFIID alone or with Mediator(f:NUT2) for 1 h at 30°C. After washing, the immobilized template was further incubated with NE(⌬IID/⌬MED) and the missing transcription factor (Mediator) during time t. After washing, standard transcription assays with added NTPs were performed. To prevent transcription re-initiation, Sarkosyl was added to a final concentration of 0.08% at 2 min after transcription initiation with NTPs. B, transcription time course for an immobilized template preincubated with TFIID. The immobilized template was preincubated with purified TFIID and, after washing, was further incubated with NE(⌬IID/⌬MED) supplemented with Mediator for 4, 8, 16, or 32 min. After washing, transcription was performed by addition of NTPs containing radiolabeled UTP. Relative transcription (txn) levels were determined by PhosphorImager analysis and are plotted in D. C, transcription time course for an immobilized template preincubated with TFIID and Mediator. The immobilized template was preincubated with purified TFIID and purified Mediator (MED) for 30 min and, after washing, was further incubated with NE(⌬IID/⌬MED) for 4, 8, 16, or 32 min. After washing, transcription was assayed following addition of NTPs containing radiolabeled UTP. Relative transcription levels were determined by PhosphorImager analysis and are plotted in D. D, comparisons of transcription time courses following template preincubation with TFIID versus TFIID plus Mediator. E, immobilized template assay for TFIID-dependent recruitment of Mediator. The immobilized template was incubated with amounts of purified TFIID and purified Mediator(f:TRAP220AB) approximating the relative ratios in a nuclear extract. After washing, bound proteins were analyzed by immunoblotting with antibodies to proteins indicated on the right. Lane 1 shows 10% input of Mediator. F, immobilized template assay for TBP-dependent recruitment of Mediator. The immobilized template was incubated with amounts of purified TBP and purified Mediator approximating those in a standard nuclear extract. After washing, bound proteins were analyzed by immunoblotting with antibodies to proteins indicated on the right.
Mediator to the promoter is a limiting step in PIC assembly and may involve direct interactions with TFIID. The implications of these findings are discussed below. (13) and others (12) have shown previously that basal transcription in nuclear extracts is strictly dependent upon Mediator. Using an immobilized template assay in conjunction with Mediator-depleted HeLa nuclear extract, we now have shown that recruitment of pol II and general initiation factors (minimally TFIIB and TFIIE) is also completely dependent upon Mediator. The ability of ectopic, highly purified Mediator to restore basal transcription and PIC formation argues for a direct role of Mediator in these processes. The relevance of the Mediator-dependent pol II and GTF recruitment to Mediator-dependent basal transcription is further underscored by their close temporal correlation in a kinetic analysis and by their mutual dependence upon an intact TATA element. Given the present and previous (35,36) demonstrations of activator-and Mediator-dependent recruitment of pol II and GTFs in HeLa nuclear extract, the observed effects of Mediator on basal transcription and PIC formation must underlie at least some of the effects of Mediator on activator-dependent transcription.

Mediator Effects Basal Transcription in Nuclear Extracts by Enhancing PIC Formation-We
Our results with human Mediator parallel those of a recent study (20) showing strong activator-independent effects of yeast Mediator on PIC formation, as well as effects on re-initiation, thus arguing for their generality. However, distinct Mediator effects through TFIIH were also suggested by the ability of yeast Mediator to stimulate CTD phosphorylation by TFIIH (7) and the CTD requirement for basal transcription in nuclear extracts (17). A recent study has provided evidence for both CTD-dependent and CTD-independent effects of Mediator on basal transcription in yeast extracts (17), although these could well relate to TFIIH effects (e.g. promoter melting, initiation, and promoter clearance) through intrinsic helicase and CTD kinase activities subsequent to PIC formation. In this regard, high levels of TFIIH were found to stimulate basal transcription in the absence of Mediator, but most of this activity was dependent upon the CTD (17). Another recent study has demonstrated that the CTD kinase activity of the Srb10 subunit of yeast Mediator is partially redundant with the CTD kinase activity of the Kin28 subunit of TFIIH (39).
Although effects of Mediator on basal transcription have recently been observed in purified systems (7,15,17), the absolute dependence on Mediator has been most apparent in crude extracts. It has been suggested that this may reflect the presence of constraints to basal transcription (which would include PIC formation in our analysis) that are unique to the crude systems. Formal possibilities include negative cofactors that directly block one or more steps in the formation or function of the PIC or negative cofactors that effectively lower the concentration of one or more PIC components. Biochemical studies in yeast extracts (24) failed to relate Mediator-dependent basal transcription to Not1 and NC2 complexes, known negative regulators that can interfere with the function of general initiation factors through interactions with TBP or TFIID (reviewed in Ref. 25). However, current studies do not exclude the possible involvement of other negative regulators (25,26) or merely a naturally limiting concentration of a general initiation factor in eliciting Mediator-dependent basal transcription.
The Mediator Effect on Basal Transcription and PIC Formation Is Linked to Restriction of TFIIB Function-Genetic and biochemical studies have established that TFIIB is an essential general initiation factor in vivo and that it acts directly to recruit pol II during PIC formation in systems reconstituted with purified factors (2, 3). Consistent with these results and despite the possibility of pol II recruitment through Media-tor (6, 7, 14 -16, 32), the present study demonstrates a TFIIB requirement for basal transcription and PIC assembly in nuclear extracts that cannot be overcome by excess Mediator. In contrast and most significant, we have found that high levels of ectopic TFIIB can bypass the Mediator requirement for both basal transcription and PIC assembly (monitored by pol II and TFIIE recruitment). The specificity of this effect is evident from the inability of a comparably high level (25-fold molar excess) of TFIIF to bypass the Mediator requirement. Thus, in the absence of Mediator in nuclear extracts, TFIIB recruitment and subsequent function in PIC assembly (minimally pol II and TFIIE recruitment) are limiting. This could reflect either a naturally limiting concentration of TFIIB or the presence of negative cofactors (see above) that interfere directly or indirectly with TFIIB function and whose effects can be overcome by cooperative functions of TFIIB and Mediator. The latter might include joint-stabilizing interactions with pol II and possibly with TFIID or other GTFs, as well as direct interactions between Mediator and TFIIB (40). Potentially relevant is the recent observation that yeast Mediator can stimulate basal transcription in a purified system at limiting concentrations of TFIIB, TFIIE, or TFIIH (17).
In relation to the apparent function of Mediator in facilitating TFIIB recruitment and function, it is interesting to note that a number of early studies of activation mechanisms implicated TFIIB as an activator target (reviewed in Ref. 41). In apparent contradiction, analysis of TFIIB mutants in yeast suggested that TFIIB recruitment is not generally limiting for gene activation (41). However, because the latter study analyzed TFIIB mutants that were shown to be defective only for TBP-dependent recruitment in vitro, it did not eliminate the possibility (as suggested) that alternative TFIIB recruitment pathways (e.g. through TAF subunits of TFIID and/or Mediator) were intact and that TFIIB recruitment might indeed be limiting. Specifically, it was not shown that TFIIB recruitment to the promoter becomes limiting in the absence of Mediator, as indicated in the present study. A recent study has shown that TFIIB recruitment to the activated yeast GAL1 promoter is dependent upon Mediator, but this could reflect the associated deficiency of TBP/TFIID recruitment (42). More relevant to the present study, an in vitro study with yeast nuclear extracts has shown interdependent recruitment of Mediator, TFIIB, and pol II subsequent to recruitment of TFIID and TFIIA (10). Although interpreted in terms of a "holoenzyme" containing GTFs in addition to Mediator and pol II, the results showed only cooperative binding to the promoter and not entry as a preformed complex. Although this study analyzed PIC formation in response to an activator, the observation of cooperative binding of TFIIB and Mediator is consistent with our demonstration of Mediatordependent TFIIB recruitment and an ability of excess TFIIB to overcome a constraint that is otherwise reversed by Mediator. One minor difference is that our analysis with TFIIB-and Mediator-depleted extracts has shown Mediator-dependent TFIIB recruitment but TFIIBindependent Mediator recruitment, whereas Ranish et al. (10) observed interdependent Mediator and TFIIB recruitment. This difference may be attributed to the analysis of basal versus activator-dependent PIC formation or to yeast versus human nuclear extracts. In addition, chromatin immunoprecipitation studies have clearly shown a temporal disjunction between Mediator and pol II recruitment to both yeast (27,28) and metazoan (29) promoters. Hence, our studies favor a model of stepwise assembly of the PIC, at least in basal transcription, rather than the holoenzyme model of recruitment of a preformed Mediator-pol II complex (with or without a subset of GTFs).
It should also be emphasized that, although our studies indicate a role for Mediator in basal transcription through effects on PIC formation, they do not exclude additional effects at subsequent stages such as pro-moter melting, initiation, and promoter clearance/early elongation. In this regard, post-recruitment functions have been documented not only for Mediator (20,21,36), but also for TFIIB (10,43,44). In accord with this possibility, studies in yeast have established genetic interactions of Mediator and TFIIB with transcription elongation factors and with each other (45)(46)(47). Of special note with respect to our demonstration of a functional link between TFIIB and Mediator, including the ability of excess TFIIB to bypass the Mediator requirement, the gene encoding TFIIB was identified as a high copy suppressor of yeast growth defects resulting from mutations in the Soh1 subunit of Mediator and elongation factor TFIIS (47). Limiting Step in Basal Transcription Involves Functional TFIID-Mediator Interactions-Our demonstration of Mediator-and TATA-dependent basal transcription and the corresponding recruitment of TFIIB, TFIIE, and pol II suggested the possibility of a preassembled TFIID-Mediatorpromoter complex as a prerequisite for subsequent recruitment of other factors to the basal PIC and as a possible rate-limiting step in basal transcription. In agreement with this idea, our immobilized template assays showed that preincubation of the template with TFIID and Mediator, but not with TFIID alone, greatly accelerated transcription when the washed template was added to extracts lacking only the component(s) present in the preincubation reaction. The further demonstration of facilitated Mediator recruitment by TFIID, but not by TBP, suggests a role for TAFs in formation of a TFIID-Mediatorpromoter complex. Support for the presence of such a complex involving direct TFIID-Mediator interactions is provided by previous studies showing human TAF-Mediator cooperativity in both basal (13) and activated (48) transcription, cooperative binding of purified human TFIID and Mediator to promoter DNA in the presence of a transcription activator (18), and Mediator-dependent binding of TBP/TFIID to yeast promoters (49,50). In the present study, the alternative possibility of an effect of Mediator through prior formation of a holoenzyme complex that facilitates concerted binding of pol II and GTFs (including TFIIB) to a TFIID-promoter complex, as proposed for PIC assembly in yeast extracts (10), was eliminated by our demonstration of TFIIB-independent recruitment of Mediator.
Overall, our studies suggest that, for basal transcription, an initial limiting step may involve formation of a stable TFIID-Mediator-promoter complex, followed by Mediator (and TFIID)-dependent recruitment of pol II and the other general initiation factors. The ability to recruit Mediator independently of TFIIB (and hence pol II and other GTFs) argues in favor of a stepwise assembly pathway for PIC formation, but does not exclude the possibility of cooperative interactions and essentially concomitant binding to the promoter of Mediator, pol II, and various GTFs. Moreover, although our studies indicate conditions under which TFIIB function is restricted, the ability of Mediator to overcome this restriction apparently relieves a potential limitation in PIC formation and basal transcription at this step. Key questions that remain are how the basal transcription functions of Mediator relate to Mediator functions in activator-dependent transcription and especially the mechanism underlying the restriction of TFIIB function and the alleviation of this restriction by Mediator.