Learn more: PMC Disclaimer | PMC Copyright Notice
Structural and Functional Characterization of PC2 and RNA Polymerase II-Associated Subpopulations of Metazoan Mediator
Abstract
The coactivator complexes TRAP/SMCC and PC2 represent two forms of Mediator. To further understand the implications of the heterogeneity of the cellular Mediator populations for regulation of RNA polymerase II (Pol II) transcription, we used a combination of affinity and conventional chromatographic methods. Our analysis revealed a spectrum of complexes, including some containing significant proportions of Pol II. Interestingly, the subunit composition of the Pol II-associated Mediator population resembled that of PC2 more closely than that of the larger TRAP/SMCC complex. In in vitro transcription assays reconstituted from homogeneous preparations of general transcription factors, Mediator-associated Pol II displayed a greater specific activity (relative to that of standard Pol II) in activator-independent (basal) transcription in addition to the previously described effects of Mediator on activator-dependent transcription. Purified PC2 complex also stimulated basal activity under these conditions. Immobilized template assays in which activator-recruited preinitiation complexes were allowed to undergo one cycle of transcription revealed partial disruption of Mediator that resulted in a PC2-like complex being retained in the scaffold. This result implies that PC2 could originate as a result of a normal cellular process. Our results are thus consistent with a dynamic nature of the Mediator complex and further extend the functional similarities between Saccharomyces cerevisiae and metazoan Mediator complexes.
RNA polymerase II (Pol II) and its associated general transcription factors (GTFs) (TFIIA, TFIIB, TFIID, TFIIE,TFIIF, and TFIIH), which are universally required by all class II genes, assemble into a preinitiation complex (PIC) at the core promoters of these genes. The PIC suffices for low (basal) levels of transcription in vitro (38, 39). Gene- and cell type-specific regulators (activators), which transduce developmental and environmental signals to target genes, typically bind to enhancer elements, from which they regulate the formation and function of the Pol II PIC. Although activators can target PIC components directly, coactivators, especially Mediator, play a critical role in the activation process (23, 29, 33).
Mediator is a multiprotein complex that was originally described as the reversibly associating coactivator component of the Pol II holoenzyme in Saccharomyces cerevisiae (23, 33). In metazoans, TRAP (12) and related complexes represent orthologs of the yeast complex, with which they share evolutionarily conserved subunits (5) as well as a similar overall three-dimensional structure (10) and coactivator function (29). Although several laboratories have described metazoan variants of the TRAP complex, they likely reflect the same cellular entity (referred to below as TRAP/Mediator) (6, 29). However, PC2 (20, 27) and CRSP (40) appear to constitute a separate entity that is highly related to TRAP/Mediator and, most likely, a derivative thereof (27, 29). Thus, aside from technical variations, PC2 (like CRSP) lacks several subunits (TRAP240/MED13, TRAP230/MED12, SRB10/CDK8, and SRB11/cyclin C) that are present in the canonical TRAP/Mediator complex; additionally, PC2 (27) and CRSP (40) are enriched in at least one subunit (p78/CRSP70/MED26) relative to the TRAP/Mediator complex.
Since TRAP/Mediator and related complexes (including PC2) function as coactivators to facilitate DNA-templated transcription, it is believed that they act relatively late in the overall activation pathway on natural templates (29). This distinguishes them from other classes of coactivators that, in the main, are believed to be recruited to facilitate chromatin remodeling at the target gene as a prerequisite for PIC formation (13).
Several lines of evidence suggest that Mediator function might be manifested predominantly at the level of the PIC. First, its initial biochemical isolation as a component of the Pol II holoenzyme in yeast (18, 19, 46), together with prior genetic evidence that its SRB subunits interact with the RPB1 subunit of Pol II (35), pointed to a close involvement with Pol II function per se. Second, the ability of the yeast holoenzyme (18, 19, 24) and the metazoan TRAP/Mediator (2, 30, 32) to stimulate both activator-dependent and activator-independent (basal) transcription in in vitro assays—in the latter case, in unfractionated nuclear extracts—is also consistent with Mediator effects on the general transcription machinery. In yeast, recruitment of several GTFs to active promoters also was found to be Mediator dependent (22, 25). Finally, Mediator can both positively (18) and negatively (1) modulate TFIIH activities.
Thus, although the primary role of Mediator is to process signals received from transcriptional activators, via direct physical interactions with distinct target subunits (29), Mediator-dependent transcription may include a significant basal component. However, the precise mechanisms by which the appropriate integrated output is achieved remain unclear, especially for metazoans. At a fundamental level, outstanding issues include uncertainties regarding the actual form of Mediator (TRAP/Mediator versus PC2) that is functional (45) as well as questions about whether TRAP/Mediator stably associates with Pol II.
Here we have used a combination of affinity and conventional chromatographic methods to analyze distinct TRAP/Mediator (and PC2) subpopulations. Our analyses have revealed PC2-like TRAP/Mediator subpopulations that are significantly enriched in Pol II content and coexist with the larger complex in nuclear extracts. We have also functionally characterized PC2 for its effect on basal transcription by Pol II. The results have implications for the function of Mediator in the wider context of activator-dependent transcription.
MATERIALS AND METHODS
Generation of epitope-tagged stable cell lines and purification of Mediator complexes.
For cloning into the pIRES1neo-derived vector VP5, which allows expression of the target proteins as fusions with FLAG and hemagglutinin, cDNAs encoding p36/MED4 and p78/CRSP70/MED26 were amplified by PCR from their corresponding expressed sequence tags. For SOH1/MED31, a full-length cDNA was first obtained by rapid amplification of 5′ cDNA ends from a human cDNA (Marathon; Stratagene) library. The resulting plasmid constructs were used to transfect HeLa S cells, and stable G418-resistant clones expressing the corresponding epitope-tagged protein were selected as described elsewhere (28).
For purification of FLAG-p36/MED4 (f:p36/MED4) complexes from these cells, nuclear extracts were fractionated over phosphocellulose (P11). The 0.85 M KCl fraction was dialyzed against buffer BC100 containing 0.05% NP-40 (28), loaded onto a DE52 column, and step eluted with 0.15 M KCl. This fraction was bound to S-Sepharose (Pharmacia), and the 0.6 M KCl eluate was loaded directly onto a Superose 6 (fast protein liquid chromatography) column that was developed in BC100. Fractions were analyzed for p36/MED4 by immunoblotting, pooled, and subjected to M2-agarose chromatography. The 0.5 M KCl fraction was directly purified over M2-agarose.
For purification of SOH1/MED31 complexes from nuclear pellets of f:SOH1/MED31-expressing cells, a Pol II purification procedure (3) was adapted for extraction of chromatin-associated factors. Briefly, the homogenized pellets were solubilized by addition of 3.8 M (NH4)2SO4 to 0.3 M. After sonication, the extract was diluted to 0.1 M (NH4)2SO4 and precipitated with 0.25% polyethyleneimine (PEI). Following resuspension in a buffer containing 0.25 M (NH4)2SO4, the cleared supernatant was precipitated with solid (NH4)2SO4. The (NH4)2SO4 concentration of the resuspended pellet was adjusted to 0.1 M, and the sample was loaded onto a DE52 column. Flowthrough and gradient [0.1 to 0.5 M (NH4)2SO4]-eluted fractions were analyzed for Pol II and Mediator components. The flowthrough was finally purified over M2-agarose.
Antisera, immunoblotting, and depletion of nuclear extracts.
Most Mediator antibodies, which were either raised in-house or obtained commercially (Santa Cruz), have been described previously (2, 27). Antibodies against LCMR1/MED19, ARC32/MED8, p28b/MED18, Surf5/MED22, and HSPC296/MED11 were a gift of J. W. Conaway and R. Conaway (7, 42). For antibodies against p78/CRSP70/MED26, a partial cDNA encoding the C-terminal half was expressed in bacteria and used to immunize rabbits (Covance). Standard immunoblotting (enhanced chemiluminescence) procedures were used. For probing with multiple antibodies, blots were sequentially stripped and reprobed.
For immunodepletion, an anti-p78/CRSP70/MED26 antiserum was purified by passage over a p78/CRSP70-Sepharose 4B column as described previously (28). The purified anti-p78/CRSP70/MED26 antibodies were cross-linked to protein A-Sepharose and incubated with HeLa nuclear extracts as described previously (28).
In vitro transcription assays.
Detailed procedures for purification of GTFs to homogeneity (including affinity methods for Pol II and TFIIH) have been published recently (28). In vitro transcription reactions with unfractionated nuclear extracts or reconstituted from homogeneous GTFs are also described in that report (28). As specified in figure legends, reactions were reconstituted either with affinity-purified TFIID or with recombinant TATA-binding protein (TBP). In the assay systems used here, standard metal ion and salt conditions were 4 mM MgCl2 and 60 mM KCl and stringent conditions were 10 mM MgCl2 and 90 mM KCl.
Immobilized-template assay.
Recruitment assays utilized a biotinylated restriction fragment from the pA4xMLΔ53 plasmid (30), which carries core promoter elements from an adenovirus major late (ML) promoter and was immobilized to M280-streptavidin Dynabeads (Dynal). The assays were performed with highly purified GTFs, Pol II, PC4, and Mediator preparations [f:p36 (P110.5)] as indicated in figure legends. Conditions for binding, washing, and analysis of template-bound factors were exactly as described elsewhere (30).
Immobilized-template assays utilizing unfractionated extracts were performed with templates that had been adsorbed onto streptavidin-Sepharose (Amersham-Pharmacia). Following a wash in transcription buffer (20 mM HEPES [pH 7.6], 4 mM MgCl2, 60 mM KCl, 0.08 mM EDTA, 8 mM dithiothreitol, 10% glycerol, 0.4 mg of bovine serum albumin/ml, and 0.05% NP-40), beads containing DNA were blocked as described elsewhere (37). After incubation with nuclear extracts (1 h at room temperature), the beads were washed with transcription buffer. Depending on the experiment, nucleoside triphosphates (NTPs) (in transcription buffer) were then added for 2 min. The beads were washed again and digested with EcoRI for 30 min at 37°C. The supernatants were collected through a spin column, precipitated with trichloroacetic acid, and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblotting. Alternatively, the supernatants were immunoprecipitated with anti-TRAP25/MED30 antibodies (by using protein A-Sepharose), eluted with glycine, and visualized by SDS-PAGE and silver staining.
RESULTS
Isolation of PC2 and a Pol II-enriched TRAP/Mediator preparation.
FLAG tagging either of SRB10/CDK8 (14), a subunit that is reversibly associated with TRAP/Mediator, or of NUT2/MED10 (27), an integral (core) subunit, previously allowed isolation of very similar (but not identical) TRAP/SMCC complexes from nuclear extracts of corresponding cell lines. In the latter case, additional chromatographic steps also allowed isolation of PC2. Since complex subunit composition could differ depending on which subunit is epitope tagged, we extended our analysis to include some of the polypeptides that appeared to exhibit variable associations with the TRAP/Mediator or PC2 complexes.
We focused first on the p36 subunit (27), which originally was identified by mass spectrometric analysis as a constituent of the USA-derived PC2 fraction (27). Although a purified preparation of PC2 obtained via f:NUT2/MED10 affinity chromatography seemingly lacked this polypeptide, it has been reported in several metazoan preparations (29) and has been proposed to be an ortholog of yeast MED4 (5). We generated stable HeLa cell lines that expressed f:p36/MED4 and confirmed by M2-agarose chromatography of the derived nuclear extracts that p36/MED4 is associated with TRAP/Mediator subunits (data not shown). To potentially segregate distinct Mediator subpopulations, the nuclear extracts were subjected to chromatography on P11 (Fig. (Fig.1A).1A). The resulting 0.5 and 0.85 M KCl fractions, which together contain the bulk of the cellular TRAP/Mediator, were processed further. The 0.5 M KCl fraction was affinity purified directly on M2-agarose and analyzed by silver staining (Fig. (Fig.1B)1B) and immunoblotting (Fig. (Fig.1C).1C). The 0.85 M KCl fraction was subjected to sequential chromatography on DE52, S-Sepharose, and Superose 6. Superose 6 fractions containing f:p36/MED4 (by immunoblotting [data not shown]) were pooled and subjected to affinity purification on M2-agarose before analysis for subunit composition (Fig. 1B and C).
Isolation of PC2 and a Pol II-enriched Mediator preparation. (A) Chromatographic scheme for the isolation of various Mediator complexes from nuclear extracts derived from a HeLa cell line (M4) that stably expresses f:p36/MED4. (B) Affinity-purified f:p36/MED4 complexes from the P11 0.5 M (lane 1) and P11 0.85 (lane 2) fractions were resolved by SDS-PAGE and visualized by silver staining. “P11 0.85, Sup6” (lane 2) designates the originating P11 fraction and the step (Superose 6) just prior to affinity chromatography. Readily identifiable and prominent polypeptide bands are labeled. Asterisk marks a polypeptide that purifies (nonspecifically) on M2-agarose. (C) The preparations (f:p36/MED4 P11 0.5 [lane 3] and f:p36/MED4 P11 0.85, Sup6 [lane 4]) shown in panel B were analyzed by immunoblotting with the indicated antibodies. HeLa nuclear extract (lane 1) and Pol II (purified from the nuclear pellet) (lane 2) were included for reference.
As predicted from previous data (27), the final preparations from the 0.5 and 0.85 M KCl fractions shared a majority of subunits but were not identical. In agreement with the previously described composition of the P11 0.85 M-derived f:NUT2/MED10 PC2 (27), the P11 0.85 M-derived f:p36/MED4 preparation was deficient in components of the TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C module (Fig. (Fig.1C,1C, lane 4). In contrast, the P11 0.5 M-derived preparation contained readily detectable amounts of all these subunits (except SRB10/CDK8, which shows poor immunoreactivity in this experiment) (Fig. (Fig.1C,1C, lane 3). All core subunits (29), including NUT2/MED10, SOH1/MED31, SRB7/MED21, TRAP25/MED30, TRFP/MED20, MED6, MED7, TRAP80/MED17, and TRAP170/MED14 (as well as p36/MED4), were found to be shared by the P11 0.5 M- and 0.85 M-derived complexes (Fig. (Fig.1C,1C, lane 4 versus lane 3) (data not shown). Of the newly described subunits (42), p28b/MED18, Surf5/MED22, and HSC296/MED11, which likely also are core components (5), were found in both complexes; however, ARC32/MED8 (7) was more prominent in the P11 0.5 M-derived complex (lane 4 versus lane 3), and LCMR1/MED19 was not detectable in either complex at the sensitivity tested. The TRAP220/MED1 subunit, previously not detected in f:NUT2/MED10-selected PC2 (27), was found in both complexes. Other subunits, notably SUR2/MED23, TRAP100/MED24, and TRAP95/MED16, were significantly underrepresented in the P11 0.85 M-derived preparation relative to that from the P11 0.5 M fraction (Fig. (Fig.1C,1C, lane 4 versus lane 3). In the case of the joint loss of SUR2/MED23, TRAP100/MED24, and TRAP95/MED16, our results could further reflect the fact that these subunits constitute a distinct submodule within the TRAP/Mediator complex (15, 44). Further, although no antibodies against TRAP97/MED25 and TRAP93 are available at present, inspection of the silver-stained gel (Fig. (Fig.1B,1B, lane 2) strongly suggested that these polypeptides were also either greatly reduced or absent in the P11 0.85 M-derived complex. As determined by immunoblotting, both preparations also contained p78/CRSP70/MED26.
No GTFs (including TFIIH) were detected in the f:p36/MED4 complexes (Fig. (Fig.1C)1C) (data not shown). However, significant amounts of Pol II subunits (RPB1, RPB4, RPB6, and RPB7) were found in the P11 0.5 M-derived complex but not in the P11 0.85 M-derived complex (Fig. (Fig.1C,1C, lane 4 versus lane 3). Indeed, in the former case, the two largest Pol II subunits (RPB1 and RPB2) could be discerned in the silver-stained gel (Fig. (Fig.1B,1B, lane 1).
We conclude that the f:p36/MED4 preparation derived from the P11 0.5 M fraction contains an essentially intact TRAP/Mediator complex and includes a population that, similar to the yeast Pol II holoenzyme described previously (18, 19), is associated with Pol II. Note, however, that this analysis by itself does not eliminate the possibility that the P11 0.5 M-derived preparation [referred to below as f:p36 (P11 0.5)] is heterogeneous. Indeed, as argued below, this preparation likely contains both a PC2-like subpopulation that is associated with Pol II (PC2 · Pol II) and an apparently complete (Pol II-free) TRAP/Mediator complex. Except for minor differences, the f:p36/MED4 complex derived from the P11 0.85 M fraction is equivalent to the PC2 described previously (27) and lacks detectable Pol II.
Isolation of Pol II-associated PC2 from the nuclear pellet.
SOH1/MED31 had previously been detected in the human SMCC complex (14). While S. cerevisiae possesses a SOH1 protein, it has not been reported in any Mediator-like complexes (33). Nonetheless, genetic analysis has implicated yeast SOH1 (which was isolated as a suppressor of a protein [HPR1] involved in DNA recombination) in interactions with TFIIB and the RPB1 subunit of Pol II (11). Recently, SOH1 was also identified as a component of a network of elongation factors that interact with the yeast histone methyltransferase SET2 by a genetic-proteomic screen (21). This raised the issue of whether SOH1 represents an integral TRAP/Mediator subunit or whether it merely copurifies with TRAP/Mediator. Therefore, we also generated a stable HeLa cell line expressing an epitope-tagged human SOH1 polypeptide in order to purify SOH1-containing complexes.
M2-agarose chromatography of nuclear extracts from the f:SOH1/MED31 cell line yielded a complex that, based on silver staining (Fig. (Fig.2C,2C, lane 1) and immunoblotting, closely resembled the canonical TRAP/Mediator complex. In agreement with recent reports that also describe SOH1/MED31 as a component of other metazoan complexes (36, 42), this result confirmed that SOH1/MED31 is a bona fide TRAP/Mediator subunit. Nonetheless, since the yeast genetic data suggest that SOH1/MED31 may have additional functions, ongoing analysis is aimed at identifying other SOH1-associated (non-Mediator) polypeptides that may also have copurified.
Isolation of Pol II-associated PC2 from the nuclear pellet. (A) Chromatographic scheme for the isolation of Mediator complexes from solubilized nuclear pellets derived from a HeLa cell line (S3-1) that stably expresses f:SOH1/MED31. (B) Immunoblot of DE52 column fractions of nuclear pellet extract. (C) Silver-stained SOH1/MED31 complexes isolated directly from nuclear extract [lane 1, f:SOH1/MED31 (NE)] or from the DE52 flowthrough of nuclear pellet extract [lane 2, f:SOH1/MED31 (NP/DE52 FT)] by affinity chromatography on M2-agarose. Readily identifiable and prominent polypeptide bands are labeled. Although bands marked with an asterisk have not been definitively identified, they may be contaminants. (D) Identification of Pol II in f:SOH1/MED31 (NP/DE52 FT). Immunoblots of the f:SOH1/MED31 (NP/DE52 FT) preparation (lane 3) and the DE52 flowthrough fraction (lane 1) from which it was derived. Pol II (lane 2) was included for reference. (E) Immunoblot of the f:SOH1/MED31 (NP/DE52 FT) preparation (lane 3) with the indicated antibodies. Mediator complexes isolated from nuclear extracts of f:CDK8 (lane 1) and f:SOH1/MED31 (lane 2) cell lines were also included. An extra band in the SOH1/MED31 panel is a result of incomplete stripping of the prior antibody.
We also evaluated whether there might be TRAP/Mediator populations not only in soluble nuclear extracts but also in tight association with chromatin and chromatin-bound Pol II. Presumably, these would reflect populations of the complex that are engaged in transcription. For this purpose we solubilized nuclear pellets from f:SOH1/MED31-expressing cells (see Materials and Methods). The material was further fractionated by PEI precipitation of nucleoprotein complexes and chromatography over DE52 (Fig. 2A and B). As determined by immunoblotting (Fig. (Fig.2B,2B, input), the PEI extract (DE52 input) contained both TRAP/Mediator and Pol II. Analysis of the subsequent DE52 fractions (Fig. (Fig.2B)2B) showed that Pol II, but not TRAP/Mediator, bound efficiently to the column. Nonetheless, a small fraction of the Pol II population was in the flowthrough fraction with TRAP/Mediator. Assuming that this segregation pattern of Pol II reflected free (DE52-bound) and Mediator-complexed (flowthrough) forms of Pol II, an excess of Mediator-free Pol II in the nuclear pellet is consistent with the estimated relative cellular abundances of these proteins in yeast (4, 26).
Subsequent purification of the DE52 flowthrough fraction on M2-agarose resulted in copurification of some of this Pol II with the f:SOH1/MED31 TRAP/Mediator, as assayed by immunoblotting (Fig. (Fig.2D,2D, lane 3) and visualization by silver staining (Fig. (Fig.2C,2C, lane 2). Interestingly, (partial) determination of the subunit composition of the TRAP/Mediator in this preparation revealed a more PC2-like complex (Fig. (Fig.2E).2E). Thus, whereas most core subunits were present, constituents of the TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C subcomplex were undetectable and, consistent with the composition of PC2 (Fig. (Fig.1B),1B), levels of the TRAP100/MED24 and TRAP95/MED16 components of the putative SUR2/MED23-TRAP100/MED24-TRAP95/MED16 subcomplex were severely reduced (Fig. (Fig.2E,2E, lane 3 versus lanes 1 and 2). We conclude that a relatively small percentage of chromatin-engaged Pol II may exist in association with a PC2-like complex (designated PC2 · Pol II).
A p78/CRSP70/MED26-containing PC2 population exists in equilibrium with the bulk Mediator complex in unfractionated nuclear extracts.
Based on the way in which PC2 was originally defined (20), both past (27) and present purification steps have entailed chromatography over ion-exchange columns (Fig. (Fig.1A).1A). Hence, we considered the possibility that the reduced subunit content of PC2 might reflect consequences of these additional chromatographic manipulations. Given further indications that the p78/CRSP70/MED26 subunit is greatly enriched in PC2 (and CRSP) relative to total TRAP/Mediator populations (including the P11 0.5 M-derived preparation) (27, 40), we generated a HeLa cell line that stably expresses FLAG-tagged p78/CRSP70/MED26. Nuclear extracts from this cell line were subjected to M2-agarose chromatography either directly or following passage over P11. In the latter case, we focused on the 0.85 M KCl fraction. Subunit composition analyses of the resulting purified complexes by silver staining and immunoblotting revealed that they were virtually identical to each other and, in turn, to PC2 obtained via other routes (f:NUT2/MED10 [27] or f:p36/MED4 [this study]) (Fig. (Fig.3).3). Thus, while most core subunits (including the tagged p78/CRSP70/MED26) were present, components of the TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C subcomplex were undetectable. In addition, the TRAP100/MED24 component of the putative SUR2/MED23-TRAP100/MED24-TRAP95/MED16 subcomplex was undetectable. Although we could not detect TRAP220/MED1 in either preparation, perhaps due to technical difficulties with immunological detection of limiting amounts of this subunit, a band that corresponded in molecular weight to this subunit was seen in the silver-stained gels (Fig. (Fig.3A).3A). Altogether, our analysis reveals that a PC2-like complex preexists in unfractionated nuclear extracts.
Purification and analysis of p78/CRSP70/MED26-containing PC2 in unfractionated nuclear extracts. (A) Silver staining (following SDS-PAGE) of p78/CRSP70/MED26 complexes isolated directly from nuclear extracts [lane 1, f:p78/CRSP70/MED26 (NE)] from the 78-1 cell line, or from the corresponding P11 0.85 M fraction [lane 2, f:p78/CRSP70/MED26 (P11 0.85)] by affinity chromatography. Readily identifiable and prominent polypeptide bands are labeled. Asterisk marks a polypeptide that purifies (nonspecifically) on M2-agarose. (B) Immunoblot of the f:p78/CRSP70/MED26 complexes shown in panel A [lane 2, f:p78/CRSP70/MED26 (NE); lane 3, f:p78/CRSP70/MED26 (P11 0.85)] with the indicated antibodies. Mediator complexes isolated from nuclear extracts of f:p36/MED4 (lane 1) were also included.
The p78/CRSP70/MED26-containing PC2 population accounts for a majority of the activator-independent activity of Mediator in unfractionated nuclear extracts.
To determine how PC2 in unfractionated nuclear extracts contributes to its transcriptional activity, we immunodepleted the p78/CRSP70/MED26-containing subpopulation of TRAP/Mediator. For depletion, HeLa nuclear extracts were incubated with antigen-purified anti-p78/CRSP70/MED26 antibodies immobilized on protein A-Sepharose beads. Immunoblot analysis of the treated nuclear extracts confirmed that p78/CRSP70/MED26 was quantitatively depleted (Fig. (Fig.4A,4A, lane 4 versus lane 3). However, no diminution in the amounts of any of the other 19 subunits that we tested was apparent. This is in contrast to what was seen when PAQ (32), TRAP25 (2), or NUT2 (30) subunits were targeted for immunodepletion. However, we were able both to affinity purify a p78/CRSP70/MED26-containing PC2-like complex (Fig. (Fig.3)3) and to detect the appropriate subunits in anti-p78/CRSP70/MED26 immunoprecipitates (data not shown). Therefore, we further conclude that the p78/CRSP70-containing population, which preexists in the nuclear extracts, represents a rather small fraction of the total TRAP/Mediator.
Depletion of p78/CRSP70-containing complex from HeLa nuclear extracts. (A) HeLa nuclear extracts were incubated with control beads (lane 3) or with beads containing cross-linked anti-p78/CRSP70/MED26 antibodies (lane 4). Unbound extract was immunoblotted with the indicated antibodies. HeLa nuclear extracts (lane 2) and Mediator complexes isolated from nuclear extracts of f:p36 (lane 1) were also included. (B) In vitro transcription reaction mixtures contained control (lanes 1 and 4) or anti-p78/CRSP70-depleted (lanes 2, 3, and 5) nuclear extracts. Purified f:p78/CRSP70/MED26-selected PC2 was added to the reaction mixture in lane 3. Templates used in reaction mixtures were pMLΔ53 in lanes 1 to 3 and pVA in lanes 4 and 5.
To test the anti-p78/CRSP70/Med26-depleted extract for transcriptional activity (Fig. (Fig.4B),4B), we used a template (pMLΔ53) that contains core elements from the adenovirus ML promoter. In a control extract, basal transcription (lane 1) from this template was readily apparent. However, the anti-p78/CRSP70/MED26-depleted extract failed to support either activator-independent (lane 2) or activator-dependent (data not shown) transcription. Control experiments with a Pol III-dependent template (VA) established that the effect of the depletion was specific to Pol II (lanes 4 and 5). Readdition of purified f:p78/CRSP70/MED26-selected PC2 to the depleted extract restored full activator-independent activity (compare lane 3 with lane 1). Similar restoration of basal activity was seen when a Mediator complex obtained from f:p36 (P11 0.5) was added back to the depleted extract (data not shown). We conclude that the p78/CRSP70/MED26-containing PC2 population is primarily responsible for activator-independent (basal) activity of TRAP/Mediator in nuclear extracts.
Higher specific activity of Mediator-associated Pol II relative to core Pol II in activator-independent transcription.
To functionally characterize the Pol II-associated Mediator in the f:p36 (P11 0.5) preparation, we first normalized (by immunoblotting [Fig. [Fig.5A])5A]) its Pol II content against a homogeneous (“core”) Pol II preparation that had been affinity purified from nuclear extracts of an f:RPB9-expressing cell line (28). Note that because of the reference to Pol II, the normalized f:p36 (P11 0.5) preparation contains a stoichiometric excess of TRAP/Mediator over Pol II.
Higher specific activity of Mediator-associated Pol II relative to core Pol II in activated transcription. (A) Variable amounts of f:RPB9 nuclear extract-derived Pol II (lanes 1 to 3) and affinity-purified f:p36 (P11 0.5) (lane 4) were immunoblotted with the indicated Pol II and Mediator antibodies. Normalization for functional assays was based on RPB1 content. (B) Equivalent amounts of core Pol II (lanes 1 and 2) and f:p36 (P11 0.5) (lane 3) were assayed for HNF-4-dependent activity (lanes 2 and 3) on the pA4xMLΔ53 template in a system reconstituted (28) from f:TBP-derived TFIID and recombinant GTFs (TFIIA, TFIIB, TFIIE, TFIIF, and TFIIH) under our standard in vitro transcription conditions. Reaction mixtures also contained PC4.
Equivalent amounts of core Pol II and f:p36 (P11 0.5) Pol II were compared in our standard in vitro transcription assay (28) reconstituted from homogeneous TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH plus the cofactor PC4 (Fig. (Fig.5B).5B). Under these conditions, the extent of activated transcription displayed by the test activator (orphan nuclear receptor HNF-4 [30]) was significantly higher with f:p36 (P11 0.5) (lane 3) than with core Pol II (lane 2), consistent with the presence in the former reaction of both Pol II and the Mediator moiety. This result is also consistent with our previous functional analyses in which activated transcription was reconstituted with core Pol II and purified TRAP/Mediator (16, 30).
We also compared the basal activities of core Pol II and Mediator-associated Pol II in the f:p36 (P11 0.5) preparations in in vitro transcription reactions that were reconstituted from GTFs (and exclusively dependent on ectopic Pol II for catalytic activity [e.g., Fig. Fig.6D,6D, lane 3]). As expected, core Pol II efficiently transcribed a supercoiled template (containing adenovirus ML core promoter) both in the absence (Fig. (Fig.6A,6A, lane 1) and in the presence (lane 3) of TFIIH. Mediator-associated Pol II in f:p36 (P11 0.5) displayed slightly lower levels of transcription under these conditions (compare lane 2 with lane 1 and lane 4 with lane 3). However, when the assay conditions were made more stringent by elevating the salt and metal ion concentrations of the reaction, the Pol II in the f:p36 (P11 0.5) preparation displayed considerably higher transcription activity than core Pol II both in the absence and in the presence of TFIIH (compare lane 6 with lane 5 and lane 8 with lane 7). When this experiment was repeated with recombinant TBP in place of TFIID, we again observed that the f:p36 (P11 0.5) preparation had significantly higher basal activity than core Pol II (Fig. (Fig.6B;6B; compare lane 2 with lane 1 and lane 4 with lane 3). However, when TBP was used, the activity differential was apparent even under standard assay conditions, probably reflecting some overlap between the mechanism of action of TBP-associated factors (TAFs) in TFIID and TRAP/Mediator. The fact that the effect of Mediator-associated Pol II is manifested with TBP provided a convenient experimental system for follow-up analyses. Note that in these experiments the need for higher-ionic-strength reaction conditions to elicit the functional differences could simply indicate that these conditions more closely mimic cellular conditions or, alternatively, that our “standard” conditions have not been optimized for the activities.
Higher specific activity of Mediator-associated Pol II relative to core Pol II in basal transcription. (A) Equivalent amounts of core Pol II (lanes 1, 3, 5, and 7) and f:p36 (P11 0.5) (lanes 2, 4, 6, and 8) were assayed for basal activity on the pA4xMLΔ53 template in a system reconstituted from f:TBP-derived TFIID and recombinant GTFs (TFIIB, TFIIE, and TFIIF) in the presence (lanes 3, 4, 7, and 8) or absence (lanes 1, 2, 5, and 6) of TFIIH under either standard (lanes 1 to 4) or stringent (lanes 5 to 8) conditions. (B) Comparison of the basal activities on the pA4xMLΔ53 template of core Pol II (lanes 1 and 3) and f:p36 (P11 0.5) (lanes 2 and 4) in TBP-nucleated transcription reaction mixtures (under standard conditions). Reactions were conducted both with (lanes 3 and 4) and without (lanes 1 and 2) TFIIH. (C) Immobilized template assays showing recruitment of PIC components. PICs were formed on M280-strepatavidin beads carrying a biotinylated ML core promoter fragment. After washing, the bound material was probed for selected factors by immunoblotting. PICs were formed with GTFs (TBP, TFIIB, TFIIF) plus either core Pol II (lane 5) or f:p36 (P11 0.5) (lane 6). As a control, the same reactions were carried out but without any TBP (lanes 7 and 8). Inputs (lanes 1 to 4) are shown. Binding reaction mixtures also contained PC4, which facilitates factor recruitment (30). (D) In vitro transcription from the pA4xMLΔ53 template. TBP-nucleated reaction mixtures contained GTFs and either core Pol II (lanes 1 and 4) or f:p36 (P11 0.5) (lane 2). PC2 (f:p36/MED4) was added to reaction mixtures in lanes 3 and 4.
To ascertain that the enhanced basal activity of the Pol II in f:p36 (P11 0.5) was in fact due to associated TRAP/Mediator, we determined if Pol II and TRAP/Mediator were corecruited to the template. An immobilized DNA fragment containing the adenovirus ML promoter was incubated with the components of the reconstituted transcription system (Fig. (Fig.6C).6C). After washing to remove unbound proteins, the factors retained on the beads were probed by immunoblotting. As previously shown (30), core Pol II (RPB6) and the GTFs (TBP and TFIIB) were recruited to the promoter under these conditions (lane 5). This result reflected specific PIC assembly, because no retention of factors on the beads was seen in a control experiment in which TBP was omitted (lane 7). Similarly, the resident Pol II within f:p36 (P11 0.5), which was found to be recruited to the PIC in the presence (lane 6) of TBP, showed significantly reduced binding in its absence (lane 8). Importantly, retention of TRAP/Mediator subunits (p78/CRSP70/MED26 and SRB7/MED21) confirmed that Pol II and Mediator components of f:p36 (P11 0.5) are corecruited to the PIC. However, when results were normalized to levels of SRB7/MED21 (a subunit shared by bulk TRAP/Mediator and PC2) and p78/CRSP70/MED26 (a subunit specific for PC2 [and CRSP]) in the input for the binding reaction (lane 4), preferential retention of p78/CRSP70/MED26 (relative to SRB7/MED21) was apparent (lane 6). As before, omission of TBP led to elimination of specific assembly of a Mediator- and Pol II-containing PIC (lane 8). (A background level of retention of Mediator and Pol II components appears to reflect a potential nonspecific [low-level] DNA binding activity of the complex.) Thus, in addition to showing that a bona fide Pol II- and Mediator-containing complex is responsible for the higher basal activity of the f:p36 (P11 0.5) preparation relative to core Pol II, these results also suggest that that this higher basal activity may reflect the presence and preferential selection of a PC2-like Pol II-associated subpopulation (PC2 · Pol II) for PIC formation.
To further establish that the higher basal activity of Mediator-associated Pol II (relative to core Pol II) is due to its Mediator (PC2) component, we examined the effect of directly adding purified PC2 (derived from the P11 0.85 fraction) to core Pol II in TBP-nucleated transcription reaction mixtures (Fig. (Fig.6D).6D). Consistent with the undetectable levels of Pol II in PC2, a control experiment showed that f:p36/MED4 PC2 (P11 0.85) had negligible activity on its own (lane 3). However, when added to reaction mixtures that contained core Pol II, PC2 boosted basal transcription severalfold (lane 4 versus lane 1), to a level that was obtained with a Pol II-normalized amount of f:p36 (P11 0.5) (lane 2).
Generation of a PC2-like complex upon transcription by activator-recruited PICs assembled on immobilized templates.
Although the immobilized-template experiment using purified PIC components (Fig. (Fig.6C)6C) strongly pointed to the preferential incorporation of a PC2-like form of Mediator, it was biased for a population of Pol II that was likely preassociated with that form of Mediator because it relied on the f:p36 (P11 0.5) preparation as the sole source of Pol II. To obtain a less biased picture of Mediator recruitment into the PIC, as well as insight into its dynamics during the course of transcription, we also performed immobilized-template experiments with unfractionated nuclear extracts (Fig. (Fig.77).
Partial disruption of the Mediator upon transcription generates a PC2-like complex. (A) Schematic outline of the experimental design. Streptavidin-immobilized templates iMLΔ53 (containing ML core promoter sequences) and iG5HML (containing five cognate sites for GAL4 upstream of a hybrid core promoter) were incubated for 1 h with unfractionated HeLa nuclear extracts (NE) either with no activator (iMLΔ53) (panel B, lanes 2 and 3) or in the presence of GAL4-p53 (iGHML) (panel B, lanes 4 and 5; panel C, lanes 2 and 3). After a wash to remove unbound material, NTPs were added to the beads for 2 min (panel B, lanes 3 and 5; panel C, lane 3). Control reaction mixtures (panel B, lanes 2 and 4; panel C, lane 2) received no NTPs. After another round of washing, the DNA-bound complexes were eluted by digestion with EcoRI and processed for immunoblot analysis. Separately, eluates from reactions shown in panel B, lanes 4 and 5, were subjected to immunoprecipitation (ip) with anti-TRAP25/MED30 antibodies (panel C). (B) Immunoblot analysis of eluates (lanes 2 to 4) from the immobilized template experiment (see panel A) with the indicated antibodies. Input HeLa NE were included (lane 1) for reference. (C) The anti-TRAP25/MED30 immunoprecipitates from eluates from reactions in panel B, lanes 4 and 5, were resolved by gradient SDS-PAGE and stained with silver (lanes 2 and 3, respectively). A Mediator preparation isolated from nuclear extracts of f:NUT2/MED10-expressing HeLa cells was run in parallel (lane 1). Top panels shows blow-ups of the upper portions of the gel. Asterisks indicate contaminating immunoglobulin G bands.
We immobilized two templates, MLΔ53 (containing ML core promoter sequences) and G5HML (containing five cognate sites for GAL4 upstream of a hybrid core promoter [31]), onto streptavidin beads and incubated them with unfractionated HeLa nuclear extracts. The G5HML template was incubated in the presence of GAL4-p53, a hybrid activator containing the DNA-binding domain from GAL4 and the activation domain from the tumor suppressor p53. No activator was added to reaction mixtures with MLΔ53. After 1 h of incubation, the beads were washed to remove unbound material, and NTPs were added to the beads for 2 min to allow transcription by any preassembled PICs. For each set, control reactions that received no NTPs were also performed. Additional controls (data not shown) confirmed that during this time, one cycle of transcription is completed. After another round of washing, the DNA-bound complexes were eluted and analyzed by immunoblotting (Fig. 7A and B). For reaction mixtures containing G5HML and GAL4-p53, part of the eluate was additionally subjected to immunoprecipitation with an anti-TRAP25/MED30 antibody that was previously demonstrated to efficiently deplete total Mediator pools from unfractionated nuclear extracts (2), to allow visualization of Mediator by silver staining (Fig. (Fig.7C7C).
As revealed by immunoblotting of their constituent polypeptides, both the MLΔ53 and G5HML templates bound the various transcription factors that are thought to constitute the PIC. These factors include TFIIB, TFIID (TBP, TAF31, TAF100), TFIIE (TFIIEα), TFIIF (RAP30), TFIIH (p62), and Pol II (RPB1) (Fig. (Fig.7B,7B, lanes 2 and 4). Importantly, we also probed for TRAP80/MED17 and TRAP25/MED30 as representatives of the invariant (PC2-like) Mediator subunits and for TRAP230/MED12, TRAP220/MED1, and SRB11/cyclin C as representatives of subunits that tend to be variable, according to analyses in this paper. Interestingly, both templates recruited each of these subunits in approximately the same proportions, suggesting that in the presence of a natural distribution of the factors (as distinct from that of the experiment for which results are shown in Fig. Fig.6C),6C), the intact Mediator is efficiently incorporated into the PICs. This conclusion is consistent with previously published data on recruitment of an SRB10/CDK8-containing complex to immobilized PICs (17) and with the initial observation that liganded thyroid hormone receptor could be isolated together with an essentially complete complement of TRAP/Mediator subunits (12). However, it should be noted that this analysis does not permit conclusions on the effects, per se, of an activator on the recruitment of Mediator, Pol II, and GTFs. (Nonetheless, and as partial evidence that bound factors reflect bona fide PICs, independent experiments in which the effects of an activator were rigorously examined have shown that under the present experimental conditions, the activator does indeed stimulate recruitment of selected PIC components [H. J. Baek and R. G. Roeder, unpublished data]).
Analysis of PICs that had been allowed to begin transcription by exposure to NTPs (Fig. (Fig.7B,7B, lanes 3 and 5) revealed dramatic differences in the responses of the two templates. With MLΔ53, transcription resulted in significant disruption of the initially formed DNA complex. Thus, levels of Pol II and TFIIB, which are known to be removed from the promoter upon transcription in both the metazoan (49) and yeast (48) systems, were greatly reduced (compare lane 3 with lane 2). On the other hand, in agreement with previous data (47-49), TFIID remained associated with DNA. (The reproducible aberration in the mobility of the TBP band likely indicates a posttranslational modification such as phosphorylation.) We further found that whereas most of the TFIIE is lost from the template, the amounts of TFIIF and TFIIH were not significantly reduced. In the case of TFIIF, at least, this finding may reflect its nonspecific DNA binding property. Importantly, both classes of Mediator subunits showed marked reductions in the amounts that remained associated following transcription, suggesting that during the course of activator-independent transcription from the ML promoter, the entire Mediator complex tends to dissociate from the template. (Note further that inasmuch as the immobilized-template-assembled PICs responded to NTPs, this is additional evidence that the assay scores authentic transcription complexes [but see Discussion].)
In contrast to what was observed with the MLΔ53 template, the Mediator assembled on the GAL4-p53-containing G5HML template showed only partial disruption (Fig. (Fig.7B;7B; compare lane 5 with lane 4). The amounts of core subunits TRAP80/MED17 and TRAP25/MED30 were essentially unaffected upon NTP treatment of preassembled PICs, suggesting that they are retained during the course of transcription. However, the amounts of bound TRAP230/MED12, TRAP220/MED1, and SRB11/cyclin C were significantly reduced. The retention pattern of Pol II and GTFs was essentially the same as that on the MLΔ53 template (no activator). We should also note that this effect is likely to be activator specific, since it was not observed when either GAL4-VP16 or GAL4-RXR was used as the activator in a similar experiment (data not shown).
This selective loss of the largest Mediator subunits in the preceding experiment was visualized directly by examining the silver-stained gel of anti-TRAP25/MED30 immunoprecipitates of total eluates obtained from the beads (Fig. (Fig.7C;7C; compare lane 3 and lane 2).
We conclude that in the context of GAL4-p53, the residual PIC contains a partial Mediator complex that has lost subunits that, for the most part, are also deficient in the PC2 complex purified by other (fractionation) procedures. This result thus raises the possibility that PC2 is derived from the larger complexes as part of a natural cellular process.
DISCUSSION
In this work we have used a variety of approaches to identify and functionally characterize distinct TRAP/Mediator subpopulations. The main findings are that the PC2 complex coexists both in equilibrium with the larger TRAP/Mediator complex and in association with Pol II. Indeed, PC2 appears to be generated from the larger complex during the course of transcription. We also found that the ability of PC2 to stimulate the basal transcription activity of Pol II, which could be observed in a reconstituted cell-free system, is an intrinsic property of the metazoan Mediator. Collectively, these new results further reinforce the dynamic nature of Mediator and the similarities between the yeast and metazoan Mediator complexes.
Subunit composition of PC2 and Pol II-containing subpopulations of Mediator.
We have now isolated a PC2-like complex from at least three distinct sources: (i) from a precursor chromatographic fraction that originally (20) defined PC2 (27); (ii) from chromatin, in association with Pol II that is likely to have been engaged in transcription (Fig. (Fig.2);2); and (iii) directly from unfractionated nuclear extracts (Fig. (Fig.3).3). We now can reasonably deduce a “consensus” composition for PC2 (Table (Table1).1). On this basis, PC2 contains a relatively tightly associated core consisting of HSPC296/MED11, NUT2/MED10, SOH1/MED31, Surf5/MED22, SRB7/MED21, TRAP25/MED30, p28b/MED18, TRFP/MED20, MED6, MED7, p36/MED4, p78/CRSP70/MED26, TRAP80/MED17, PAQ/MED15, and TRAP170/MED14. Loosely (variably) associated subunits include TRAP220/MED1, SUR2/MED23, TRAP100/MED24, TRAP95/MED16, and ARC32/MED8. However, the fractional contents of TRAP240/MED13, TRAP230/MED12, SRB10/CDK8, and SRB11/cyclin C are greatly reduced in this complex. These conclusions are generally in good agreement with a recent bioinformatic attribution of metazoan orthologs of yeast Mediator subunits and their putative organization into subcomplexes and modules (5). Additionally, they are broadly consistent with the results of Sato et al. (43), which were published after completion of this work. (Immunoblot analysis for TRAP97/MED25, TRAP93, and p37/MED27 is pending; however, the latter was found in our initial mass spectrometric analysis of conventionally purified PC2 [27].)
TABLE 1.
Updated composition of TRAP/Mediator and the PC2 complexa
TRAP | PC2 |
---|---|
TRAP240/MED13 | |
TRAP230/MED12 | |
TRAP220/MED1 | TRAP220/MED1b |
TRAP170/MED14 | TRAP170/MED14 |
SUR2/MED23 | SUR2/MED23* |
PAQ/MED15 | PAQ/MED15 |
TRAP100/MED24 | TRAP100/MED24* |
TRAP97/ACID/MED25c | |
TRAP95/MED16 | TRAP95/MED16* |
TRAP80/MED17 | TRAP80/MED17 |
CRSP70/p78/MED26 | CRSP70/p78/MED26 |
SRB10/CDK8 | |
p37/MED27 | p37/MED27 |
p36/MED4 | p36/MED4 |
MED7 | MED7 |
SRB11/cyclin C | |
MED6 | MED6 |
ARC32/MED8 | ARC32/MED8 |
TRFP/MED20 | TRFP/MED20 |
p28b/MED18 | p28b/MED18 |
TRAP25/MED30 | TRAP25/MED30 |
SRB7/MED21 | SRB7/MED21 |
Surf5/MED22 | Surf5/MED22 |
SOH1/MED31 | SOH1/MED31 |
NUT2/MED10 | NUT2/MED10 |
HSPC296/MED11 | HSPC296/MED11 |
Association of metazoan Mediator with Pol II.
A novel observation of the present study is the isolation from human cells of functional Pol II-containing Mediator complexes that are analogous to previously described holoenzyme complexes in yeast (18, 23, 33). Evidence presented here suggests that a PC2-like complex is the entity that is associated with Pol II. Even in the case of the f:p36 (P11 0.5) preparation, which contains an apparently complete TRAP/Mediator complex and Pol II, the results of our recruitment experiments (Fig. (Fig.6C)6C) make it highly probable that the preparation reflects a mixed population of complete TRAP/Mediator and PC2, with the Pol II existing in association with the latter (PC2 · Pol II). Furthermore, assuming equivalence between PC2 and CRSP, our data, collectively, extend those of a recent report that showed an interaction between CRSP and the C-terminal domain of RPB1 but that did not address interactions with intact Pol II (34).
Interestingly, significant amounts of Pol II were not detectable in f:SRB10/CDK8-derived isolates of TRAP/SMCC/Mediator (14). As suggested for the Schizosaccharomyces pombe system (41), it is likely that interactions of the core Mediator (PC2) with Pol II and with TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C are mutually exclusive. Thus, epitope tagging of SRB10/CDK8 may have led to enrichment of a TRAP/SMCC/Mediator population that excludes Pol II. On the other hand, tagging of integral subunits (p36/MED4 and SOH1/MED31) of the core complex, as described here, leads to enrichment for a Pol II-containing subpopulation.
Whereas our observations provide evidence for a stable association between Pol II and metazoan Mediator that is functional, they do not allow us to draw any conclusions about which form of Pol II is normally recruited to the promoters. Indeed, recent evidence has questioned the earlier one-step holoenzyme recruitment model. Thus, structural studies of the yeast holoenzyme (9) indicate that threading of template DNA around Pol II would be precluded without dissociation of the Mediator and Pol II moieties. Furthermore, in vivo data have revealed a temporal disjunction in the relative order of recruitment of Mediator and Pol II to target genes (8, 36). Therefore, even though Pol II and PC2 copurify, it is possible that they are not necessarily in tight association with each other, which could allow for stepwise assembly of a PIC. It also remains possible that the isolated PC2 · Pol II reflects an intermediate that is generated after PIC assembly (discussed further below).
Potential role of PC2 in basal and activated transcription.
An immobilized-template experiment (Fig. (Fig.7)7) indicating initial recruitment of an intact TRAP/Mediator that is evidently converted to a PC2-like complex suggests that the two forms are interconvertible. This experiment relies on the ability to assemble, in vitro, specific PICs for which functionality (response to NTPs) can be demonstrated. However, a potential caveat of this experiment is that not all complexes scored in this assay may represent functional PICs. Rather, a subset could represent some kind of moribund complexes. (This caveat applies equally to an earlier study that utilized extracts from yeast and first reported that a scaffold consisting of a subset of GTFs and Mediator stays behind at the promoter after the first round of transcription [48]. Notably, however, this study monitored, at most, only a couple of “core” Mediator subunits and therefore allowed no conclusions about the makeup of the PIC-incorporated Mediator before and after transcription.) Even if the observed partial disruption reflects these nonproductive complexes, it is clear that NTPs are triggering the conversion. Therefore, the results allow us to conclude that there is potential for a regulatory event to effect the interconversion. Thus, while we have identified here one set of conditions under which this conversion can take place, alternative pathways might also exist.
Furthermore, modifications in our assay protocol now clearly establish that the basal activity of TRAP/Mediator, which previously was apparent only in the context of an unfractionated nuclear extract (2, 30, 32), is an intrinsic property of PC2, as described for the purified yeast Mediator complex (18) and the PC2 fraction (20).
Given that the natural function of PC2 is realized in the context of activator-dependent transcription (27, 29), our present data permit us to model the step where PC2 might act in that pathway. A recent structural analysis of CRSP and ARC (equivalent to PC2 and TRAP/Mediator, respectively) hinted at the possibility that only the former is active in activated transcription (45). However, our present data, together with published studies, suggest an alternative mechanism.
As originally isolated in association with liganded thyroid hormone receptor (12), the TRAP complex contains apparently stoichiometric levels of essentially all subunits, including the TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C module. This finding, together with our present results, makes it highly likely that in the context of an activator, the intact TRAP/Mediator is initially recruited to the promoter. This step could be followed by facilitated entry of Pol II into the PIC, with concomitant addition of p78/CRSP70/MED26, loss of the TRAP240/MED13-TRAP230/MED12-SRB10/CDK8-SRB11/cyclin C module, and conversion of TRAP/Mediator to PC2, perhaps in an initiation-coupled manner. The in situ-generated PC2 could then act upon the (preassembled) PIC. Thus, Mediator may function dynamically at two levels. First, in concert with the activator, the Mediator may promote PIC assembly. Second, following a structural or topological conversion, the Mediator may modulate the (basal) function of Pol II in the PIC.
PC2, therefore, could be an intermediate in the activation process. Indeed the relative cellular levels of PC2 and TRAP/Mediator—the p78/CRSP70/MED26-containing PC2 is estimated to constitute less than 10% of the total Mediator population—are consistent with what would be expected for a transient kinetic intermediate. However, it is also possible that PC2 reflects the Mediator remnants of the scaffold that remains at the promoter following the first round of transcription (48).
Acknowledgments
We thank J. Conaway, R. Conaway, E. Golemis, M. Meisterernst, M. Ito, Y. Tao, S. Yamamura, and C.-X. Yuan for antibodies.
This work was supported by institutional funds to the Laboratory of Biochemistry and Molecular Biology and by an NIH grant (RO1 DK060764) to S.M.