Functional Conservation of the Glutamine-Rich Domains of Yeast Gal11 and Human SRC-1 in the Transactivation of Glucocorticoid Receptor Tau 1 in Saccharomyces cerevisiae^▿

Yeast strains and plasmids.

The yeast strains used in this study are listed in Table 1. Strain EGY48 was generated according to the manufacturer's protocol (Clontech). Strains derived from EGY48 or YPH500 were generated by specific gene disruption according to standard procedures. To construct the gene disruption plasmids, the flanking regions of MED2 (nucleotides −353 to −12 and +1251 to +1501, where the first nucleotide of the starting ATG codon is represented by +1), GAL11 (−300 to +253 and +1851 to +2450), HRS1 (−284 to −11 and +1140 to +1411), SIN4 (−347 to −12 and +2831 to +3076) and SWI3 (−315 to −16 and +2088 to +2438) were amplified from yeast genomic DNA by PCR using Vent polymerase (New England Biolabs). Each pair of 5′ upstream and 3′ downstream regions was cloned into the BamHI and HindIII sites of pBluescript II SK (Stratagene) by three-piece ligation. TRP1 was then inserted between the cloned 5′ and 3′ regions by EcoRI restriction digestion and ligation. DNA fragments consisting of TRP1 and the flanking regions of each target gene were obtained by BamHI/HindIII digestion and transformed into an appropriate strain to create the corresponding deletion strains.

The cDNAs for hGR τ1 (residues 77 to 262) and its mutant derivatives (H1ala, D196Y) in pGAL4(1-100) and pGEX4T were kindly provided by Annika Wallberg (Karolinska Institute, Sweden) (52). For in vitro translation, hGR τ1 and its derivatives were excised from pGAL4(1-100) and subcloned into pcDNA3. Wild-type and mutant τ1c fragments of hGR (residues 187 to 244) were amplified by PCR from their corresponding pGEX-τ1 templates and subcloned into pCL313 to assay transactivation activity (17). To construct the episomal 8XlexAo plasmid for chromatin immunoprecipitation (ChIP) analysis, a DNA fragment corresponding to the 8XlexAo region was amplified by PCR from pSH18-34 (Invitrogen) and subcloned into pRS316 using XhoI restriction digestion and ligation.

The glutathione S-transferase (GST) fusion proteins of Gal11p and Gal11p derivatives were described previously (40). The amino acid substitution mutants of the Gal11 Qr domain (residues 116 to 277), Am (R145A and Q147A), and Bm (Q198 and V199A), were constructed by PCR-based site-directed mutagenesis with overlap extension, and subcloned into pGEX for bacterial expression. For the expression of the mini-Gal11p derivatives M-WT, M-Am, M-Bm, and M-ABm in yeast, wild-type and mutant Gal11p Qr domains (Qr-Am, Qr-Bm, and Qr-ABm) were amplified by PCR using the corresponding pGEX derivatives as templates and subcloned into the BamHI and PstI sites of pRS315, which carried the Mediator association domain of Gal11p (residues 799 to 1081) located between the promoter and terminator regions of GAL11 (pRS315-GAL11MAD). To make expression vectors for F-WT, F-Am, and F-Bm, the internal region of Gal11p (residues 278 to 798) was inserted into the PstI site of mini-Gal11p derivatives (pRS315 M-WT, M-Am, and M-Bm) using the in vivo gap repair system. Then, the N-terminal region of Gal11p (residues 1 to 115) was amplified by PCR and subcloned into the BamHI sites of the clones, resulting in the construction of full-length Gal11 derivatives. The DNA inserts were confirmed by sequencing analysis after clone isolation. To generate the GST fusion protein of the Qr domain of human SRC-1 (residues 1020 to 1185), the corresponding gene sequence was amplified by PCR from a cDNA library and subcloned into pGEX4T-1. Mutated variants of the human SRC-1 Qr domain [Am (R1053A, Q1055A), Bm (Q1105A, M1106A), and the ABm double mutant] were also generated by PCR-based site-directed mutagenesis and subcloned into pGEX4T-1 for GST pull-down assays. Chimeric mini-Gal11p derivatives (SQ-WT and SQ-ABm) were constructed by subcloning the corresponding SRC-1 Qr domain fragments into the BamHI and PstI sites of pRS315-GAL11MAD. For the expression of the LexA fusion protein of the AF-1 region of mouse AR (residues 1 to 556) in yeast, the corresponding gene sequence was isolated from a mammalian expression plasmid and subcloned into pCL313. Full-length clones for GAL11, SIN4, HRS1, and MED2 were subcloned with appropriate restriction enzyme sites into pcDNA3 for in vitro translation.

Assays for GR τ1c transactivity in yeast.

pCL313 derivatives and a pLGZ-2XlexA reporter plasmid (52) were transformed into wild-type and mutant yeast strains by a lithium acetate method. Several colonies from each transformation were patched on minimal medium plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) to verify the homogeneity of their transactivation phenotype. Three or more transformants were grown in minimal medium to an optical density at 600 nm (OD₆₀₀) of 0.2 and then treated for 4 to 6 h with 0.5 mM CuSO₄ to induce the expression of LexA fusion proteins. Whole-cell extracts were prepared and assayed for β-galactosidase activity as described previously (40). To monitor chromosomal expression of the LEU2 reporter gene, a spot assay was performed using EGY48-derived yeast strains expressing LexA-GR τ1c. Cells at the same concentration were spotted in 10-fold serial dilutions onto duplicate minimal medium plates lacking or containing leucine (as a positive control for spotting) and then incubated for 2 to 3 days.

Western blot analysis.

Yeast protein extracts were prepared by growing cells to mid-log phase in the appropriate medium, followed by glass bead disruption in lysis buffer (150 mM Tris-OAc [pH 7.9], 50 mM KOAc, 5 mM MgSO₄, 1 mM EDTA, 10% glycerol). Unless otherwise noted, manipulation of all proteins was carried out at 4°C and in the presence of the following protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 mM benzamidine, 5 μg/ml of leupeptin, 5 μg/ml pepstatin). The protein concentrations of the whole-cell extracts were normalized using the Bradford protein assay, according to the manufacturer's instructions (Bio-Rad). Purified proteins or immune complexes were separated on either 10% or 12% sodium dodecyl sulfate (SDS)-polyacrylamide gels and then transferred to a Hybond-ECL enhanced chemiluminescence nitrocellulose membrane (Amersham). Membranes were incubated with a monoclonal antibody for anti-LexA (sc-7544; Santa Cruz Biotech), a polyclonal antibody to anti-GST (sc-459; Santa Cruz Biotech), or polyclonal antibodies to Mediator subunits. The antisera against Mediator subunits Gal11, Rgr1, Hrs1, and Srb5 were generated in rats using bacterially expressed recombinant polypeptides as antigens, as described previously (30). The antisera against Med1 were kindly provided by Stefan Björklund (Umea University, Sweden). Immunoreactive proteins were visualized using an Amersham ECL kit, according to the manufacturer's instructions.

Protein preparations and GST pull-down assay.

pGEX4T derivatives were introduced into Escherichia coli strain BL21 or DH5α, and the resultant transformants were grown to an OD₆₀₀ of 0.4. Overexpression of GST alone or GST proteins was induced by incubation for 6 h at 30°C in the presence of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and proteins were purified using glutathione-agarose beads (Promega), according to the manufacturer's instructions. All buffers used for the purification of GST fusion proteins and GST pull-down assays contained protease inhibitors, as described above. Purified GST fusion proteins were stored at −80°C until use. The Pol II holoenzyme fractions or whole-cell extracts were prepared from wild-type, rgr1Δ2, hrs1Δ, gal11Δ, srb5Δ, and med1Δ mutant cells, as described previously (31). Radiolabeled proteins were synthesized by in vitro-translation using pcDNA3-based expression constructs and the TNT transcription-coupled translation system (Promega). Radiolabeled proteins or Pol II holoenzyme fractions were added to equivalent amounts of GST or GST fusion protein (2 to 4 μg) bound to glutathione-agarose beads that had been preequilibrated with buffer A (150 mM Tris-HCl [pH 7.9], 5% glycerol, 1 mM EDTA, 1 mM dithiothreitol, 1× protease inhibitor, 0.01% NP-40, 150 to 200 mM KCl) in a final volume of 250 μl. Following incubation, the beads were washed three times in equilibration buffer and bound proteins were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and autoradiography.

RT and quantitative PCR.

Yeast total RNA was isolated from wild-type yeast cells bearing the pRS316-8xlexAop plasmid and overexpressing LexA-GR τ1c. Total RNA was treated with RNase-free DNase to remove residual genomic DNA. The reverse transcription (RT) reaction was carried out using the thermoscript RT kit (Invitrogen) in a 50-μl volume of reaction mixture containing 5 μg of total RNA and random hexaoligonuclotide primers. The synthesized cDNA was then subjected to PCR by using three primer sets that were specific for URA3 (U-F, 5′-CCTAGTCCTGTTGCTGCCA-3′; U-R, 5′-ATAATGCCTTTAGCGGCTT-3′) and the two flanking regions of the lexAo region (A-F, 5′-TCACTGCCCGCTTTCCAGT-3′; A-R, 5′-CCGCCTTTGAGTGAG CTGA-3′; B-F, 5′-CACACCCGCCGCGCTTAAT-3′; B-R, 5′-ATCGCCTTGCAGCACATCC-3′) to determine the levels of transcription from these regions within the pRS316-8xlexAop plasmid. As a positive control, genomic DNA was prepared from an identical strain and subjected to genomic PCR using the same primer sets.

ChIP assay.

ChIP assays were performed essentially as described previously (26), with minor modifications. In brief, cells were grown in minimal medium to an OD₆₀₀ of 0.2, and then treated for 4 h with 0.5 mM CuSO₄ to induce the expression of LexA fusion proteins. Cells were harvested, washed, and treated with formaldehyde (1% [vol/vol]) to cross-link DNA and proteins. Total cell extracts were prepared by bead beating and sonication, and equal volumes of extract were incubated overnight at 4°C with the indicated antibodies. Protein G- or A-agarose beads (Invitrogen) were added to the samples to isolate specific protein-bound DNA, and then the beads were washed three times using wash buffer. Elution buffer was added and the beads were incubated at 65°C overnight to separate protein-DNA complexes. DNA was purified and then amplified by PCR for 25 cycles to detect the chromosomal lexA operator region and the HIS4 promoter or for 20 cycles to detect the episomal lexA operator region. The following primer sets were used for PCR: T set (5′-ATGGATTCATTAGATCCGT-3′ and 5′-AATAGGTCCTTAAATAATC-3′) for the chromosomal lexA operator region, N set (5′-TAATACGACTCACTATAGG-3′ and 5′-AATTAACCCTCACTAAAGG-3′) for the episomal lexA operator region, and H set (5′-GAAGACAACAGTGGTGTGA-3′ and 5′-CAACACACATCGGAGGTGA-3′) for the TATA region of HIS4. The amplified PCR products were separated by 2% agarose gel electrophoresis and visualized by ethidium bromide staining.

Sulfo-SBED biotin label transfer assays.

Sulfo-SBED {sulfosuccinimidyl [2-6-(biotinamido)-2-(p-azidobenzamido)-hexanoamido]-ethyl-1,3′-dithiopropionate} biotin label transfer assays were essentially carried out according to the manufacturer's protocol (Pierce). In brief, a recombinant form of GST (300 nM) or GST-GR τ1 (500 nM) was incubated with a 5 M excess of sulfo-SBED in the dark to conjugate the biotin label to the GST derivatives. The excess cross-linking reagent was removed from the GST proteins by dialysis overnight at 4°C. The protein conjugated to sulfo-SBED was incubated with purified Mediator in binding buffer (50 mM HEPES [pH 7.5], 150 mM NaCl, 10% glycerol, 0.1% Tween 20) for 1 h. The reaction mixture was then exposed to UV light to cross-link GST-GR τ1 and Mediator, after which 4× SDS sample buffer was added to the reaction mixture and incubated for 10 min. Dithiothreitol (100 mM) was added to reaction mixtures in order to transfer the biotin label from GST proteins to the interacting partner(s) of Mediator. The mixtures were applied to 12% SDS-PAGE and subjected to Western blot analysis using streptavidin-horseradish peroxidase and anti-Gal11 antibody.

The Gal11 module of Mediator is specifically required for the transactivity of hGR τ1 in yeast.

Previous reports have demonstrated that yeast whole-cell extracts deficient in histone acetyltransferase activity can activate GR τ1-mediated transcription of naked DNA templates (51), which suggests that a novel coactivator protein other than a chromatin remodeling/modifying factor is involved in mediating the transactivity of GR τ1 in vitro. To determine the functional requirement of Mediator for GR τ1 transcriptional activity in yeast, we conditionally expressed a LexA-DNA binding domain (DBD) fusion of GR τ1c (LexA-GR τ1c) under the control of the CUP1 promoter, whose activation was previously shown to be independent of Mediator (28, 36). An expression plasmid for LexA-GR τ1c was transformed along with the episomal 2XlexAo-LacZ reporter construct into wild-type and Mediator mutant strains. Transformants were grown in minimal medium containing copper to induce the expression of LexA-GR τ1c and then subjected to a liquid β-galactosidase assay. We initially tested various coactivator mutant strains that were deficient in representative subunits of the tail (rgr1Δ2, a C-terminal truncation of Rgr1), middle (med1Δ and med9Δ), and head (srb2Δ and srb5Δ) domains of Mediator along with mutant strains of SWI/SNF (swi2Δ) and SAGA (gcn5Δ) complexes. The transactivity of GR τ1c was significantly reduced in the rgr1Δ2 strain to the levels of the swi2Δ and gcn5Δ mutants but not in the other Mediator mutant strains (Fig. 1A). These results indicated that there is a specific requirement for the tail domain of the Mediator complex in GR τ1c transactivation in yeast. It was previously demonstrated that a C-terminal truncation of Rgr1 or deletion of any single member of the Gal11 module (Gal11, Med2, Sin4, and Hrs1) results in the loss of the entire Gal11 module from the Mediator complex (31). Therefore, we examined the transactivity of GR τ1c in yeast strains deficient in each of the genes encoding components of the Gal11 module. As shown in Fig. 1A, GR τ1c transactivity was significantly reduced in all Gal11 module mutant strains examined, similar to the rgr1Δ2 strain, which confirmed that the Gal11 module of the Mediator complex is required for GR τ1c transactivity in yeast. The expression levels of LexA-GR τ1c in all of the Gal11 module mutant strains were similar to wild type; thus, the reduced activity of GR τ1c in the mutant strains was not due to changes in its expression level (Fig. 1A, lower panel). Interestingly, a double-deletion mutant of the Gal11 module and SAGA (gcn5Δ sin4Δ) exhibited a more severe defect in GR τ1 transactivity than either of the single-deletion mutants alone (Fig. 1A). These results indicate that Mediator and chromatin-related complexes have nonredundant roles and can potentiate GR τ1 transactivity in yeast via independent pathways, which would explain the residual transactivity of GR τ1 in the single-deletion mutants of these coactivator complexes.

We also examined GR τ1c transactivity in wild-type or Gal11 module mutant strains carrying the chromosomal LEU2 reporter gene under the control of the LEU2 TATA box and six copies of the lexA operator. Consistent with the previous results, GR τ1c-dependent LEU2 expression was severely reduced in all of the Gal11 module mutant strains, as assessed by growth on minimal medium plates lacking leucine (Fig. 1B). These results demonstrated that the Gal11 module of the Mediator complex is essential for the transcriptional activity of GR τ1 in yeast.

GR τ1 directly interacts with Mediator complex in a Gal11 module-dependent manner.

We next examined whether the Gal11 module physically interacts with GR τ1 and whether this interaction is functionally relevant. To examine the direct interaction between GR τ1 and Mediator, we performed a GST pull-down assay using GST fusion proteins of GR τ1 and GR τ1 mutants (wild type, D196Y, and H1ala), and Pol II holoenzyme fraction containing Mediator. Mediator interacted with GR τ1 but not with GST alone (Fig. 2A, lanes 2 and 3). Interestingly, Pol II was not associated with GR τ1 under these conditions, which indicated that the binding activity between Mediator and GR τ1 is relatively stronger than that between Mediator and Pol II. GR τ1 D196Y is a gain-of-function mutant that exhibits enhanced transactivity (two- to threefold) compared to wild-type GR τ1, whereas the transactivity of GR τ1 H1ala is less than 10% of that of the wild type. Consistent with this, the interaction of GR τ1 D196Y with Mediator was two- to threefold stronger than that of wild-type GR τ1, and the H1ala mutant exhibited severely reduced binding to Mediator (Fig. 2A, lanes 3 to 5). These results demonstrated that there was a good correlation between the ability of GR τ1 mutants to bind to Mediator and their transactivities in yeast, although it has been shown that their transcriptional activities are also affected by altered interactions with other coactivators, such as SWI/SNF and SAGA complexes (51, 52). To determine the role of the Gal11 module in the Mediator-GR τ1 interaction, we measured the binding strength of GR τ1 for Mediator complexes prepared from Gal11 module mutant strains (rgr1Δ2, hrs1Δ, and gal11Δ) using the GST pull-down assay. GR τ1 failed to bind to mutant Mediator complexes lacking the Gal11 module (Fig. 2B and C), which was in good agreement with the reduced GR τ1 transactivity in these strains. In contrast, deletion of the middle (med1Δ) and head (srb5Δ) subunits of Mediator had no effect on their interactions with GR τ1 (Fig. 2C). These results indicated that GR τ1 directly interacts with the Mediator complex via the Gal11 module.

Mediator recruitment by GR τ1c to target promoters occurs in the absence of transcription in vivo.

To confirm the direct role of the Gal11 module in transcriptional activation by GR τ1 in vivo, we carried out ChIP assays using two different DNA templates harboring LexA binding sites (Fig. 3A). The first contained the chromosomal LEU2 reporter gene, which can be actively transcribed by LexA-GR τ1c (TATA-containing, transcribed template), and the other contained only the lexA operator region (TATA-less, nontranscribed template). We confirmed that there were no transcripts from the regions neighboring the LexA binding sites on the nontranscribed template by RT-PCR (Fig. 3B). First, we sought to demonstrate the activator-dependent recruitment of the Mediator complex to the lexA operator region by GR τ1c. As shown in Fig. 3C, Mediator (Rgr1 ChIP) was recruited to LexA binding sites of the templates only when LexA-GR τ1c expression was induced by adding copper to the medium. Next, we examined the in vivo occupancy of the lexA operator region of the two templates by Mediator and TATA-binding protein (TBP) to determine whether the association of GR τ1 with Mediator at promoter regions is transcription-dependent. We also examined the occupancy of the TATA region of the HIS4 gene as a control for normalizing the ChIP signals (Fig. 3A). We first investigated the binding of LexA-GR τ1c to the two templates and found that GR τ1c associated with the lexA operators of transcribed and nontranscribed templates with similar levels of efficiency but not with the TATA region of HIS4 (Fig. 3D, lanes 5 and 6). Second, we observed that Mediator (Rgr1 ChIP) is recruited by GR τ1c the to lexA operators of the transcribed and nontranscribed templates with equal levels of efficiency, which indicated that the recruitment of Mediator by GR τ1c onto target promoters occurs in the absence of transcription in vivo (Fig. 3D, lanes 9 and 10). Third, GR τ1c was able to recruit Mediator to lexA operator regions in the absence of Swi3p (Fig. 3D, lane 11), which is an essential component of the SWI/SNF chromatin remodeling complex. This result suggested that GR τ1c-dependent recruitment of Mediator onto target DNA sequences occurs in the absence of chromatin remodeling. The association of Mediator at LexA-GR τ1c-bound regions of transcribed and nontranscribed templates required the Gal11 module of Mediator (Fig. 3D, lane 12), indicating that the direct interaction of GR τ1c with Mediator is dependent on the Gal11 module in vivo. As a control, GR τ1c-dependent TBP occupancy was observed on the transcribed template in wild-type and swi3Δ strains, but not in gal11Δ strains, and it was not observed on the nontranscribed template in any of the strains examined (Fig. 3D, lanes 14 to 16). In addition, there were no significant changes in the levels of LexA-GR τ1c bound to the lexA operator, in swi3Δ or gal11Δ strains (Fig. 3D, lanes 6 to 8). Taken together, these results indicated that Mediator interacts directly with GR τ1 bound to target DNA elements in vivo and that this interaction is dependent on the Gal11 module but not on transcription per se.

Gal11p directly interacts with GR τ1.

To determine which Mediator subunit(s) interacts directly with GR τ1, we analyzed the interaction of GST-GR τ1 and individual proteins of the Gal11 module that were synthesized by in vitro translation (Fig. 4A). A significant amount of Gal11p bound specifically to GST-GR τ1 but not to GST alone. Sin4p also interacted specifically with GR τ1, but this interaction was much weaker than that of Gal11p. This result indicated that Gal11p is the major binding partner of GR τ1 in the Gal11 module. Since Gal11 module proteins form a subcomplex independently of the rest of the Mediator complex, we investigated the binding of reconstituted Gal11 modules to GR τ1. Gal11, Sin4, Hrs1, and Med2 proteins were synthesized together by in vitro translation and incubated with GST-GR τ1. As shown in Fig. 4B, in contrast to the results using individual proteins (Fig. 4A), all of the Gal11 module proteins associated with GR τ1 under these conditions, which indicated that reconstructed Gal11 modules can directly interact with GR τ1 via Gal11p. Next, we determined the GR τ1 target subunit of Mediator in the context of the whole complex. We employed a sulfo-SBED label transfer method to cross-link interacting protein in a protein-protein interaction assay. Briefly, bacterially purified GST or GST-GR τ1 was labeled with sulfo-SBED (biotin) and incubated with the purified Mediator complex. The biotin label on GR τ1 was then cross-linked and transferred to its target subunit in Mediator by UV irradiation and dithiothreitol treatment, respectively. The transferred biotin label was detected by Western analysis using horseradish peroxidase conjugated to streptavidin. We found that the biotin label of GST-GR τ1, but not of GST alone, was exclusively transferred to the Gal11 subunit of Mediator, which was confirmed by performing immunoblot analysis of the same blot with anti-Gal11 antibody (Fig. 4C). This result clearly showed that Gal11p is a direct and exclusive target of GR τ1 for the Mediator interaction.

To begin to understand the physiological significance of the direct interaction of Gal11p with GR τ1, we first mapped the GR τ1-binding domain of Gal11p using a set of GST fusion proteins containing individual fragments of Gal11p (Fig. 4D). The Gal11 derivatives were incubated individually with ³⁵S-labeled GR τ1, and their ability to have a physical interaction with GR τ1 was monitored by GST pull-down assay. We identified amino acid residues 116 to 255 of Gal11p as the exclusive GR τ1-binding domain (Fig. 4D). Since we had previously observed that Mediator directly interacts with GR τ1 (Fig. 2A), we investigated the interaction of GST-Gal11 116-255 with GR τ1 D196Y and GR τ1 H1ala. The binding of Gal11 116-255 to GR τ1 D196Y was two- to threefold stronger than that of wild-type GR τ1, and there was negligible binding of Gal11 116-255 to GR τ1 H1ala (Fig. 4E). This binding pattern was consistent with the binding of GR τ1 to Mediator complex (compare Fig. 2A and 4E) and indicated that amino acids 116 to 255 of Gal11p represent the physiological target for GR τ1 binding.

Differential requirements for the Qr domain motifs in the interaction of Gal11 and SRC-1 with GR τ1 or AR AF-1.

We were interested in identifying the motifs within amino acids 116 to 255 of Gal11p that were important for the interaction of Gal11p with GR τ1. We found that amino acids 116 to 255 of Gal11p contain a region that is rich in glutamine, termed the Qr domain, which has similarity to the Qr domain of SRC-1. Interestingly, the Qr domain of SRC-1 has previously been shown to functionally interact with the AF-1 region of AR (5). As shown in Fig. 5A, multiple sequence alignments of the Qr domains of yeast Gal11p and SRC family members revealed a high degree of conservation between yeast Gal11p and SRC proteins. Intriguingly, some of the conserved residues (A-box and B-box in Fig. 5A) were previously shown to be important for the function of SRC-1 as a coactivator of AR AF-1 transactivity in mammalian cells (9, 41). Therefore, we generated two double-substitution mutants of the Gal11 Qr domain, Qr-Am (R145A, Q147A) and Qr-Bm (Q198A, V199A) (Fig. 5A), based on the presence of conserved sequence motifs in Gal11p that corresponded to A-box and B-box motifs, respectively, in the Qr domains of SRC proteins. We first examined the effect of the Am and Bm mutations on the ability of the Gal11 Qr domain to bind to GR τ1 in vitro. While Gal11 Qr-Am interacted with GR τ1 similarly to the wild type, the Qr-Bm mutant failed to interact with GR τ1 (Fig. 5B). To determine whether there was functional similarity between the Qr domains of yeast Gal11 and human SRC-1 proteins, we examined the ability of AR AF-1 to bind directly to the Qr domain or Qr domain mutants of Gal11p. Gal11 Qr-Am bound less well to AR AF-1 than both the wild type and Qr-Bm, which indicated that residues in the A-box are more important than those in the B-box for the interaction of Gal11p with AR AF-1 (Fig. 5B). This result was in contrast to the binding pattern of the Gal11p Qr domain for GR AF-1.

Next, we tried to identify the determinants of the interaction between the SRC-1 Qr domain and the AF-1 region of GR or AR. To this end, we constructed mutant derivatives of the SRC-1 Qr domain (residues 1020 to 1185) containing mutations in either the A-box or the B-box [Am (R1053A, Q1055A), Bm (Q1105A, M1106A)] or a double mutant containing substitution mutations in both motifs (ABm) (Fig. 5A) and examined their ability to interact with GR τ1 and AR AF-1. Similar to the binding pattern of the Gal11 Qr domain (Fig. 5B), the Bm and ABm mutants, but not the Am mutant, were defective in their interactions with GR τ1, whereas only the Am and ABm mutants of the SRC-1 Qr domain were defective in their interaction with AR AF-1 (Fig. 5C). These results demonstrated that there are differential requirements of A-box and B-box residues in the interaction of the Qr domains of Gal11 and SRC-1 with the AF-1 regions of AR and GR.

Functional conservation between the Qr domains of yeast Gal11 and human SRC-1 in the potentiation of GR τ1 and AR AF-1 transactivities in yeast.

The similar binding patterns of the Qr domain mutants of Gal11 and SRC-1 with the AF-1 regions of GR and AR suggested the functional conservation of these domains for the potentiation of GR and AR transactivities in yeast. To address this possibility, we first examined whether the Am and Bm mutations affected GR τ1 transactivity in vivo. We constructed a low-copy plasmid expressing the full-length versions of the Am and Bm mutants from the GAL11 promoter (Fig. 6A) in a gal11Δ strain. As shown in Fig. 6B, the wild-type (F-WT) and Am mutant (F-Am), but not the Bm mutant (F-Bm), of Gal11p could potentiate GR τ1 transactivity, as suggested by the requirement for the Bm residues in Gal11 and SRC-1 for their in vitro binding to GR τ1 (Fig. 5B). Next, we determined whether the interaction between the Gal11 Qr domain and GR τ1 was sufficient for GR τ1 transactivation in yeast. To this end, we constructed a Gal11p variant, mini-Gal11p, that contained the Gal11 Qr domain fused directly to the Mediator association domain of Gal11p (residues 799 to 1081), which has been shown to be both necessary and sufficient for the incorporation of Gal11p into Mediator complexes (3). Mini-Gal11p could potentiate GR τ1 transactivation in a gal11Δ strain like the full-length Gal11p did (Fig. 6B), which indicated that the interaction of GR τ1 with the Gal11 Qr domain is both necessary and sufficient for Mediator-dependent transactivity of GR τ1. We also generated mini-Gal11p proteins that contained the Am or Bm mutations (M-Am and M-Bm) and examined their ability to function as coactivators of GR τ1 transactivity in a gal11Δ strain. The level of GR τ1 transactivity in cells expressing M-Am was approximately 90% that in cells expressing wild-type mini-Gal11p (M-WT). In contrast, Gal11 M-Bm failed to potentiate transcriptional activation by GR τ1. These results were consistent with the in vitro binding patterns of the Am and Bm Qr domain mutants (Fig. 5B) and indicated that amino acids Q198 and V199 within the B-box of the Gal11 Qr domain are essential for the GR τ1-Gal11 interaction and the coactivator function of Mediator in GR τ1 transactivity. Notably, when we tested the abilities of Gal11 derivatives to potentiate AR AF-1 transactivity, all of the mutant versions of full-length Gal11p could mediate AR AF-1 function (Fig. 6C). Moreover, mini-Gal11p was unable to rescue AR AF-1 transactivity in the absence of Gal11p (Fig. 6C). Taken together, these results indicate that the in vitro interaction between the Gal11 Qr domain and AR AF-1 (Fig. 5B) is not sufficient to mediate AR-AF-1 transactivity in yeast. Interestingly, we detected a strong physical interaction between AR AF-1 and another Qr region located in the middle region of Gal11 that is essential for AR AF-1 transactivity in yeast (D. H. Kim and Y. C. Lee, unpublished data).

Finally, we investigated whether the SRC-1 Qr domain could functionally substitute for the Gal11 Qr domain in the potentiation of AR AF-1 and GR τ1 transactivities in yeast. We constructed chimeric mini-Gal11 proteins consisting of the wild-type SRC-1 Qr domain (SQ-WT) or its ABm mutant derivative (SQ-ABm) fused directly to the Mediator-association domain of Gal11p (residues 799 to 1081) (Fig. 6A). As shown in Fig. 6B, the reduced GR τ1 transactivity in a gal11Δ strain was restored by SQ-WT (to approximately 50% that of full-length Gal11p) and this effect was completely abolished by the ABm mutations. Similarly, SQ-WT, but not SQ-ABm, potentiated AR AF-1 transactivity in a gal11Δ strain (Fig. 6C).

In conclusion, our results suggest that the Qr domains of yeast Gal11 and human SRC-1 proteins might be evolutionally conserved and serve as common targets for certain activators (e.g., AF-1 of steroid receptors) and that they play comparable roles in the mediation of the transcriptional activities of these activator proteins in yeast.

Identification of Mediator complex as a novel coactivator that mediates GR τ1 transactivity in yeast.

In general, most eukaryotic coactivator proteins (or complexes) are directly or indirectly recruited to activator-bound regulatory regions and enhance the transcriptional efficiency of target genes by either antagonizing the repressive effect of nucleosome structure and/or directly interacting with the transcriptional machinery (29). The former group of coactivator proteins includes a wide range of protein complexes with chromatin remodeling and/or histone modification activities, such as the SAGA, NuA4, and SWI/SNF complexes of yeast, whereas the latter group includes yeast TFIID and Mediator complexes, which directly interact with components of the general transcription machinery, such as TBP and Pol II. In the current model of transcriptional activation, the concerted or combined action of the two classes of coactivator proteins results in the relief of nucleosomal repression, followed by the assembly of the Pol II transcriptional machinery to achieve full activation of target genes.

It has been reported that GR τ1 directly and functionally interacts with SAGA and Ada-independent NuA4 histone acetyltransferase complexes, as well as the SWI/SNF histone remodeling complex, all of which are essential for GR τ1 transactivity in yeast (51, 52). Interestingly, Wallberg et al. showed that yeast cell extracts deficient in histone acetyltransferase activity can also mediate GR τ1-dependent transcriptional activation from naked DNA templates in vitro (51). Although the human TFIID complex has been shown to be involved in GR τ1 transactivation in vitro (14), this observation strongly suggested that a novel transcriptional coactivator that functions independently of chromatin state is involved in GR τ1 transactivation in yeast.

Our study demonstrated that in addition to previously identified chromatin remodeling/modifying factors, yeast Mediator complex functions as a novel coactivator of the AF-1 region of the GR in yeast. This finding fits well in the current model of transcriptional activation, in which two classes of coactivators act cooperatively to maximize transcription efficiency. Our observation that a double-deletion mutant of the Gal11 module protein Sin4 and the Gcn5 subunit of SAGA exhibited a more severe defect in GR τ1 transactivity than single-deletion mutants also supports the model (Fig. 1A). This result also indicate that different types of coactivator complexes are required for the transactivity of GR τ1 and that Mediator and chromatin-related complexes have nonredundant, specialized roles in transcriptional activation by GR τ1 in yeast.

Requirement of Gal11p of the Mediator complex for GR τ1 transactivity in yeast.

The Gal11 subunit of the Mediator complex is required for transcriptional activation by many activators in vitro and in vivo (16, 31) and functions as a direct, physiological target of various activator proteins, as well as artificial nonacidic activator peptide (22, 27, 31, 32, 40). In the present study, we showed that the mammalian activators GR τ1 and AR AF-1 are among the activator proteins that functionally interact with Gal11p via direct binding and that this interaction potentiates their transactivities in yeast. In previous studies, mammalian Med14 (TRAP170/ARC150/DRIP150), the proposed homologue of yeast Rgr1, was identified as a GR AF-1-interacting protein based on a yeast two-hybrid screen (20). GR τ1 transactivity was impaired in a strain carrying a C-terminal deletion of Rgr1 (Fig. 1A, rgr1Δ2); thus, we first examined the physical interaction between the Rgr1 C-terminal region and GR τ1. We found that there was no physical interaction between these two proteins in a GST pull-down assay (data not shown). Also, although the intact Rgr1 existed within the purified Mediator from the strains lacking the Gal11 module, we could not observe the interaction between these mutant Mediators with GR τ1 (Fig. 2B and C). Rather, we found that GR τ1 directly interacts with the Gal11 module, whose association with the Mediator complex is completely dependent on the C-terminal region of Rgr1p (31). This result clearly demonstrated that the impaired transactivity of GR τ1 in an rgr1Δ2 strain was due to the lack of the Gal11 module rather than to the lack of the C-terminal region of Rgr1p.

The Gal11p subunit of the Gal11 module interacted directly with GR τ1 and was required for the recruitment of Mediator onto promoter-bound GR τ1 in vivo, even in the absence of transcription (Fig. 3 and 4). The Sin4 subunit of the Gal11 module also interacted with GR τ1. However, this interaction was much weaker than that of Gal11p, and we found no evidence of a functional interaction between Sin4 and GR τ1 mutants (data not shown), which ruled out the possibility that Sin4p was the physiological target of GR τ1. Based on the specialized role of Gal11p in the function of various activator proteins, several mammalian homologues of yeast Gal11p have been proposed. Bioinformatics-based approaches using data from functional and structural studies have identified human Med23 (hSur2/TRAP150/DRIP130) as a Gal11 orthologue (6, 7). In support of this, Med23, Med24 (TRAP100/DRIP100), and Med16 (TRAP95/DRIP92) form a submodule of the mammalian Mediator complex that is regarded as the functional counterpart of the yeast Gal11 module (21). Other studies have identified human Med15 (PCQAP/ARC1051) as the mammalian homologue of yeast Gal11p, based on its homology to the Gal11 N-terminal region and its function as a coactivator for the acidic activator VP16 (39). However, rigorous experimental data are needed to confirm the identity of these mammalian homologues, as these predictions were based on preliminary data and weak homology between short fragments. We found that the N-terminal KIX homology domain of Gal11p was not essential for GR τ1 transactivity in yeast and were unable to detect a physical interaction between mammalian Med15 and Med23 and GR τ1 in vitro (data not shown). These results indicate that the mammalian orthologues of Gal11p that have been proposed to date are not the functional counterparts of yeast Gal11p in the potentiation of GR τ1 transactivity in yeast.

Functional conservation between the Qr domains of Gal11p and SRC as a direct targets for the AF-1 regions of steroid receptors in yeast.

We showed that the Qr domain of Gal11p (residues 116 to 255) was the exclusive target for GR τ1 binding and was both necessary and sufficient for GR τ1 transactivity in yeast, as shown by the analysis of mini-Gal11p (Fig. 6). To identify a conserved motif(s) within the Gal11 Qr domain, we carried out a sequence alignment of the functional orthologues of yeast Gal11p and other known GR AF-1 coactivators. We first aligned Gal11p 116-255 with the reported mammalian homologues of yeast Gal11p (hMed15 and hMed23) and the GR AF-1-interacting subunit of mammalian Mediator (hMed14) and failed to identify any significant homologies between them. Next, we aligned the Gal11p Qr domain with the Qr regions of ySwi1p (residues 286 to 536) and ySnf5p (residues 28 to 273), which are known to serve as the targets of various activator proteins (37, 42). However, this search revealed again that there were no meaningful similarities between these domains except for certain stretches of conserved Q residues. In contrast, amino acids 116 to 255 of Gal11p showed a high degree of sequence similarity to the Qr domains of SRC proteins (Fig. 5A). The A-box and B-box motifs of SRC-1 Qr domain have been shown to be required for the function of SRC-1 as a coactivator of AR AF-1 transactivity (5). Interestingly, only mutations in the B-box (Qr-Bm) of the Gal11 Qr domain resulted in defects in the physical association of Gal11p with GR τ1, whereas double- or triple-alanine substitutions in the A-box (Q143/L144/R145, R145/Q147) did not affect the interaction (Fig. 5B and data not shown). The B-box motif was also important in the interaction of GR τ1 with the SRC-1 Qr domain (Fig. 5C), indicating the B-box motif of Gal11p and SRC is the binding interface for GR τ1. Additional experiments in a relevant biological system are needed to determine the functional importance of the B-box motif in the stimulation of GR τ1 transactivity by SRC-1.

In contrast to GR τ1, the interaction of the Qr domains of Gal11 and SRC-1 with AR AF-1 required the A-box motif, suggesting that there are different requirements for the two motifs in the interaction of Gal11p with the AF-1 regions of nuclear receptors. In support of this hypothesis, the AF-1 regions of other nuclear receptors (including estrogen receptor and Nurr1) also showed differential requirements for conserved A- and B-box motifs in their interaction with the Gal11p Qr domain and in the activation of their transactivities in yeast (D. H. Kim, G. S. Kim, and Y. C. Lee, unpublished data). It was shown using a mammalian system that both the A- and B-box motifs of the Qr domain were essential for the stimulation of AR AF-1 transactivity by SRC-1 (9, 41). These previous experiments also showed that the A-box motif is more important than the B-box motif, which is somewhat consistent with our in vitro and in vivo interaction data (Fig. 5 and 6).

The group working with M. Ptashne has demonstrated that the Gal11 region (residues 261 to 351) flanking the Qr domain is both necessary and sufficient for mediating a weak transactivity of AH (for amphipathic helix) activator in yeast (3), suggesting this region of Gal11p as the physiological target of the artificial activator AH. However, their in vitro binding assay also revealed that AH protein does not significantly bind to the region from 261 to 351 of Gal11p (3). These controversial results raised the intriguing possibility that a weak transactivity of AH activator shown in yeast might have resulted from the artifactual interaction between AH and Gal11 residues 261 to 351, which is too weak to be reproduced in an in vitro condition. In support of this assumption, we could observe that this region of Gal11p was not required for the interactions of Gal11p with many natural activator proteins we tested and in the potentiation of their transactivities in yeast (D. H. Kim, G. S. Kim, and Y. C. Lee, unpublished data).

In summary, our results provide evidence that the Qr domains of Gal11p and SRC family proteins have conserved binding sites for the AF-1 regions of some steroid receptors and that they have comparable functions in mediating their transactivities in yeast. Furthermore, based on our unpublished observations showing that the Qr domain of Gal11p is a general target for various activator proteins, we suggest that the Qr domain of yeast Gal11p is functionally equivalent to the Qr domain of mammalian SRC proteins and that it constitutes the direct, physiological target of transcriptional activators.