Learn more: PMC Disclaimer | PMC Copyright Notice
Signal-Induced Transcriptional Activation by Dif Requires the dTRAP80 Mediator Module
Abstract
The Mediator complex is the major multiprotein transcriptional coactivator complex in Drosophila melanogaster. Mediator components interact with diverse sets of transcriptional activator proteins to elicit the sophisticated regulation of gene expression. The distinct phenotypes associated with certain mutations in some of the Mediator genes and the specific in vitro interactions of Mediator gene products with transcriptional activator proteins suggest the presence of activator-specific binding subunits within the Mediator complex. However, the physiological relevance of these selective in vitro interactions has not been addressed. Therefore, we analyzed dTRAP80, one of the putative activator-binding subunits of the Mediator, for specificity of binding to a number of natural transcriptional activators from Drosophila. Among the group of activator proteins that requires the Mediator complex for transcriptional activation, only a subset of these proteins interacted with dTRAP80 in vitro and only these dTRAP80-interacting activators were defective for activation under dTRAP80-deficient in vivo conditions. In particular, activation of Drosophila antimicrobial peptide drosomycin gene expression by the NF-κB-like transcription factor Dif during induction of the Toll signaling pathway was dependent on the dTRAP80 module. These results, and the indirect support from the dTRAP80 artificial recruitment assay, indicate that dTRAP80 serves as a genuine activator-binding target responsible for a distinct group of activators.
RNA polymerase II (Pol II) and its general transcription factors comprise a minimal set of proteins required for the accurate initiation of mRNA synthesis in a reconstituted in vitro transcription system (17). In addition to this basal transcription apparatus, several classes of cofactor proteins also participate in Pol II transcription. These transcriptional cofactor proteins help to appropriately regulate the expression of individual genes according to the complicated physiological demands of living cells. Some of these cofactors reconfigure the local transcription environment by covalently modifying or repositioning nucleosomes, while others interact intimately with the basal transcription machinery and modulate its assembly and/or functional activities (34).
Both positive and negative cofactors (called coactivators and corepressors, respectively) are, in many cases, composed of multiple subunits. Some of these subunits communicate directly with sequence-specific, DNA-binding transcription factors, thus serving as built-in activator- and/or repressor-interacting modules or targets. Presumably, the binding specificities of these cofactor subunits target the entire cofactor complex to particular promoters.
One of these multiprotein cofactor complexes, referred to as Mediator, was first identified in and isolated from Saccharomyces cerevisiae (27, 28). Mediator homologs were subsequently detected in several metazoan species (5, 13, 16, 25, 33, 39, 41, 50) and shown to function as cofactors that confer activator responsiveness in a minimal-in vitro-transcription system (38). In certain Mediator preparations, a protein kinase activity that phosphorylated the C-terminal domain of Pol II was detected; however, no other transcription-related enzymatic activity has been assigned conclusively to the Mediator complex (30). Mediator was initially thought to have a key function in recruiting Pol II to activator-bound target promoters mainly because the yeast Mediator complex was isolated in tight association with Pol II (29). However, this model does not fit well with more recent findings that the recruitment of Pol II to natural promoters does not coincide with that of Mediator (3, 9, 37, 44). Therefore, it remains unclear how Mediator regulates activated transcription once it is recruited to a target promoter via interaction with DNA-bound, gene-specific transcriptional regulatory proteins.
Although the requirement for Mediator in transcriptional regulation appears to be universal among eukaryotic species, a number of Mediator proteins appear to be diverged to accommodate species-specific developmental regulation (4a). The highly conserved Mediator subunits are essential for viability in yeast (32) and flies (15), indicating that they have more fundamental and widespread roles in regulating Pol II transcription. However, genetic screens for specific physiological or developmental defects in yeast, nematodes, and fruit flies have detected highly diverged Mediator subunits of yeast (gal11, hrs1, sin4, rox3 [24, 42, 46, 51]) or metazoans (sur-2/CeTRAP150, sop-1/CeTRAP230, kto/dTRAP230, pap/bli/dTRAP240 [4, 45, 53, 56]).
Yeast Mediator proteins with similar genetic properties are grouped together to form structural and functional modules (18, 26, 29) which have been visualized by the electron density mapping of two-dimensional crystal images (1). One of the Mediator modules, the Gal11-containing module, was shown to serve as a physiologically relevant binding interface for many transcriptional activators, such as VP16, Gal4, Gcn4, and Swi5 (3, 29, 36). With respect to metazoans, gene knockout studies in mice showed that the metazoan-specific mediator proteins TRAP220 (thyroid hormone receptor-associated protein) and Sur-2 interact with ligand-bound nuclear receptors and E1A/Elk-1, respectively, and mediate their transcriptional activation functions in vivo (23, 47). Another metazoan Mediator protein, TRAP80, was shown to interact with several acidic transcriptional activators in vitro (22, 37), but it has not yet been shown to be required for transcription in vivo.
Previously, the Drosophila melanogaster Mediator complex was isolated and showed that it is critical for transcriptional activation in response to diverse activator proteins in vitro (35). More importantly, we found that mutant flies deficient in dMED6, one of the conserved Mediator components, exhibit transcriptional defects for a wide range of developmentally regulated genes (15), validating the requirement for dMED6 in gene transcription in vivo. In addition, for transcription of the Drosophila heat shock genes, Mediator is recruited to the promoter region by heat shock transcription factor (HSF) under heat shock conditions and Mediator binding is independent of the recruitment of the general transcription machinery (37). This indicates that Mediator is the primary in vivo activator target that relays regulatory signals from enhancer-bound activators to the basal transcription machinery. However, it is necessary to explore this idea further by examining the interaction of Mediator with many other activator proteins before we accept any generalized view of the mechanism by which Mediator regulates Pol II transcription.
In this study, we tested a number of natural transcriptional activator proteins for their ability to interact with Mediator proteins in vitro and examined the role of dTRAP80, one of the metazoan-specific Mediator components, in gene-specific transcriptional activation in vivo. dTRAP80 interacted specifically with many, but not all, of the transcriptional activators tested, and the dTRAP80-containing Mediator module exhibited an activator-specific requirement for transcriptional activation in vivo. Hence, each activator appears to depend on distinct binding targets for Mediator recruitment and transcriptional activation. These findings will aid us in our understanding of the mechanisms by which gene-specific cofactors are recruited to promoters for selective transcriptional activation in higher eukaryotes.
MATERIALS AND METHODS
Plasmids.
The copper-inducible Gal4 expression vector pMT-G4 was constructed by inserting a PCR-amplified DNA fragment encoding the Gal4 DNA-binding domain (G4DBD) (amino acids 1 to 93) into the EcoRI-BamHI sites of pRmHa-3 (8). pMT-G4 was used in the construction of expression vectors encoding PCR-generated activator protein fragments or full-length Mediator proteins; these genes or gene fragments were inserted into the BamHI-SalI sites of pMT-G4. pG5-E1b-luc and pG2-hsp70-luc were constructed by replacing the chloramphenicol acetyltransferase cassette of pG5-E1b-CAT (a gift from Michael Green) and the malE reporter sequences of phsp70-M (31) with the luciferase gene. The glutathione S-transferase (GST) fusion protein expression vectors were constructed by cloning the BamHI-SalI fragment from pMT-G4 into the same restriction sites in pGEX-4T1.
Transfection and reporter gene analysis.
Transfection of cultured Schneider line 2 (SL2) cells was performed as previously described (55) by using Lipofectin (GIBCO BRL, Rockville, Md.) in a six-well plate format. Firefly luciferase and β-galactosidase reporter gene expression was analyzed by using the Luciferase assay system (Promega, Madison, Wis.) and the Galacto-Light Plus system (Tropix, Bedford, Mass.), respectively, according to the manufacturers' instructions.
GST pulldown and ChIP assays.
GST pulldown experiments were carried out as previously described (35). For the chromatin immunoprecipitation (ChIP) assay (37), SL2 cells were treated with formaldehyde and chromatin was processed as described previously. The promoter occupancy of Gal4-Mediator fusion proteins on the G2-hsp70-luc reporter gene was analyzed by quantitative PCR with primers that corresponded to positions −123 to −99 (complementary to the hsp70 promoter region) and +131 to +158 (complementary to the luciferase coding sequence).
RNA interference (RNAi) analysis.
cDNA fragments of dTRAP80 (nucleotide positions +262 to +983 relative to the translation initiation codon +1) and luciferase (nucleotide positions from the initiation codon +1 to +1305) were subcloned into pBluescript II KS(+) (Stratagene, La Jolla, Calif.) to make pdTrap80-5′ and pluc1.3, respectively. To produce single-stranded RNAs (ssRNAs) of the dTRAP80 and luciferase cDNA in both directions, pdTrap80-5′ and pluc1.3 were digested with BssHII and the small fragments that contained the T7 and T3 promoter sequences, respectively, at the ends of each template were purified. By using the purified fragments as templates, ssRNA products were prepared by using the MEGAscript T7 and T3 transcription kits (Ambion, Austin, Tex.) and purified by using Micro Bio-Spin 6 chromatography columns (Bio-Rad, Hercules, Calif.). The ssRNAs of each gene were mixed separately and annealed to make double-stranded RNAs (dsRNAs) by incubation at 65°C for 30 min, followed by slow cooling to room temperature. dsRNAs were analyzed by agarose (1%) gel electrophoresis to ensure that the majority of the dsRNAs existed as a single band of the expected size.
Drosophila SL2 cells were diluted to a final concentration of 106 cells/ml in serum-free HyQ-CCM3 medium (HyClone, Logan, Utah), and 4 ml of cells were plated in tightly closed 35-mm flasks (Nunc, Naperville, Ill.). Transfection of dsRNA (6.25 μg) with Cellfectin reagent (GIBCO BRL) was performed according to the manufacturer's protocol. The transfected cells were incubated for 3 days at 25°C before analysis of the dsRNA interference (dsRNAi) effect.
RNA analysis.
Total RNA (4 μg) isolated from SL2 cells was used in cDNA synthesis reactions performed with the SuperScript II reverse transcriptase system (Invitrogen, Carlsbad, Calif.). The amounts of drosomycin, dTRAP80, and ubiquitously expressed Drosophila ribosomal protein gene rp49 mRNAs were measured by real-time PCR analysis with the following primers: drosomycin forward, 5′-CTCCGTGAGAACCTTTTCCA-3′; drosomycin reverse, 5′-AGCATCAGGACAGCGAAGAG-3′; dTRAP80 forward, 5′-CGACCTCAAGAAGGAGGAGA-3′; dTRAP80 reverse, 5′-AACTGAGTGGCGTCCTTGTC-3′; rp49 forward, 5′-ATCGGTTACGGATCGAACAA-3′; and rp49 reverse, 5′-GACAATCTCCTTGCGCTTCT-3′. The PCRs contained 1× SYBR Green mix (PE Applied Biosystems, Foster City, Calif.), 25 pmol of the forward and reverse primers, and cDNA corresponding to 0.1 μg of total RNA, and the reaction mixtures were subjected to 40 PCR cycles (95°C for 15 s and 60°C for 1 min) in the ABI Prism 7700 sequence detection system (PE Applied Biosystems).
RESULTS
Various types of transcriptional activation domains have a distinct activation potential in Drosophila cells.
Activation domains of naturally occurring transcriptional activator proteins show diverse size and sequence variations, suggesting that each class of transcription activator proteins might interact with a different coactivator complex. In order to assess systematically the importance of the interaction between activation domains and transcriptional coactivator complexes, we sought to measure the transcriptional activation potential of various activator domains in the same structural context. To this end, we generated fusion proteins that contained the G4DBD (Gal4 amino acids 1 to 93) fused to one of several transcriptional activation domains, including Bicoid N amino acids 1 to 255, Bicoid C amino acids 256 to 494 (49), Armadillo amino acids 694 to 844 (54), Notch amino acids 2156 to 2703 (48), Dorsal amino acids 435 to 678 (21), Dif amino acids 373 to 667 (20), Relish amino acids 458 to 597 (11), Dmp53 amino acids 1 to 83 (6), D-Stat amino acids 659 to 761 (19), and HSF amino acids 582 to 691 (55) (Table (Table1).1). Activation domains were chosen from these various Drosophila transcriptional activators on the basis of either previous domain mapping studies or sequence similarity to mammalian homologs.
TABLE 1.
Transcriptional activator proteins tested in this study
Transcriptional activator | Developmental and/or physiological function | Signaling pathway | Target promoter |
---|---|---|---|
Bicoid | Embryonic A-P axis formation | hunchback | |
Armadillo | Segment polarity determination | Wingless, Frizzled2 | Ultrabithorax |
Notch | Cell type specification | Delta/Serrate, Presenilin | E(Spl) |
Relish | Immune response | 18-weeler, dIKK | attacin |
Dif | Immune response | Toll, Tube, Pelle, cactus | drosomycin |
Dorsal | Embryonic D-V axis formation | Toll, Tube, Pelle, cactus | twist |
Marelle (D-Stat) | Sexual identity determination | Unpaired Hopscotch | Sex-lethal |
Dmp53 (D-p53) | Cell growth regulation | reaper |
These synthetic G4 activators were cotransfected into SL2 cells together with a luciferase reporter plasmid bearing five Gal4-binding sites in front of the adenoviral E1b core promoter. Using the luciferase assay, we measured the ability of these activator fusion proteins to stimulate reporter gene transcription (Fig. (Fig.1).1). The activation domains from Armadillo, Dif, Notch, and HSF showed strong activation potency in this assay and resulted in a 50-, 200-, 460-, and 730-fold activation of reporter gene expression, respectively. However, the activation domains from Bicoid, Dorsal, Relish, Dmp53, and D-Stat did not elicit a significant transcriptional activation response under these conditions (Fig. (Fig.1A).1A). In addition to these autonomous transcriptional activation domains, the ligand-binding domain (LBD) of retinoid X receptor (RXR) fused to the G4DBD stimulated luciferase expression about 40-fold in a 9CRA-dependent manner (Fig. (Fig.1B).1B). However, the LBD of the retinoic acid receptor (RAR) showed poor transcriptional activation potency in Drosophila cells; the RAR LBD fused to the G4DBD activated luciferase expression less than fourfold in the presence of its specific ligand, ATRA. Immunoblot analysis of the transfected cell extracts revealed similar levels of the G4 fusion protein expression in these experiments (data not shown). Therefore, the fact that we observed huge differences in transcriptional activation potencies among the activators tested might indicate that each activation domain has a distinct cofactor target and that the various domains bind to their targets with varying affinities. In other words, different activation domains may be designed to operate on different core promoters and on the distinct features of chromatin architecture associated with these promoters via interactions with different sets of cofactor complexes.
Transcriptional activation by various types of activation domains in Drosophila cells. (A) Transcriptional activation by constitutive transcriptional activation domains (TADs). SL2 cells were transiently cotransfected with the G4 fusion constructs indicated at the bottom of the graph and the G5-E1b-luc reporter plasmid. In order to normalize transfection efficiency, the hsp70-lacZ plasmid was cotransfected and the basal expression level of lacZ was measured by the β-galactosidase assay. Transcription activation shows the normalized luciferase values obtained by dividing the luciferase activity by the corresponding β-galactosidase activity. BCD-N, Bicoid N; BCD-C, Bicoid C. (B) Inducible transcriptional activation by LBDs. SL2 cells were transiently cotransfected with the G4 fusion constructs indicated at the bottom of the graph and the G5-E1b-luc reporter plasmid. The luciferase activities were normalized as indicated in panel A. The ligands were 10−7 M 9CRA (for RXR) and 10−7 M ATRA (for RAR).
Transcriptional activation domains with a strong activation potential interact with the Mediator complex.
Because it has been suggested that Mediator is a key transcriptional coactivator complex and the primary target for transcriptional activator proteins, we examined the correlation between the transcriptional activation potency of each activation domain and its ability to bind to Mediator. When the naturally occurring Drosophila activation domains were tested by a GST pulldown assay, the GST-fused activation domains of Armadillo, Dif, HSF, Notch, and Dorsal coprecipitated with the Mediator complex from Drosophila cell nuclear extracts (Fig. (Fig.2A).2A). In a similar experiment, Mediator interacted with the ligand-inducible activation domain of RXR only in the presence of 9CRA (Fig. (Fig.2A).2A). However, there was no detectable interaction between Mediator and the activation domain of RAR, even in the presence of ATRA. These observations are in agreement with our results from the transfected cell assays described above, where only the activation domains that bind Mediator, that is, Armadillo, Dif, HSF, and Notch, functioned well as transcriptional activators in Drosophila cells. The Dorsal activation domain sequence that we chose on the basis of its homology to that of mammalian NF-κB appears to retain its ability to bind Mediator, but it lost the potency to activate transcription in SL2 cells. It is also noteworthy that Notch activation domain interacts more strongly with Mediator than the more potent transcriptional activator HSF activation domain. Therefore, the poor transactivation potency of some of these fusion proteins might result from their strong dependence on additional activation signals or other domains within the original protein.
Direct interaction of the Mediator complex with transcriptional activation domains that have strong activation potentials. (A) Interactions between the Mediator complex and various transcriptional activation domains. GST or the GST-fused activation domains indicated at the top of the gels were immobilized on GST-agarose beads and incubated with a Drosophila embryo soluble nuclear fraction (SNF) that contained Mediator proteins. Immunoblotting was carried out with the antibodies indicated at the left. The cognate ligands (9CRA and ATRA) were used at a concentration of 10−6 M each. (B) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of expression levels for GST and the GST fusion activator proteins. Proteins were visualized by Coomassie staining, and the full-length GST fusion proteins are marked with arrows.
dTRAP80 is a target of some, but not all, transcriptional activation domains.
Previously, different types of yeast transcriptional activator proteins were shown to interact with distinct parts of the Mediator complex (36), and this activator-specific interaction appears to be conserved in metazoans as well: TRAP220/DRIP205 and Sur-2 mediate the nuclear receptor and the E1A activation signal, respectively (7, 14, 23, 47). Human TRAP80 was also shown to interact with VP16 and p53 in vitro, indicating its potential as an additional activator-specific binding site within the Mediator complex (22). Among the activation domains tested in this study, the HSF activation domain was previously shown to bind dTRAP80 (37). Nonetheless, knowledge about the activator-binding spectrum of TRAP80 protein is still rudimentary. Moreover, the importance of activator-TRAP80 interaction in gene transcription in vivo has not been studied in detail. Therefore, we examined whether the activation domains we tested share dTRAP80 as a binding site within the Mediator complex. In vitro-translated dTRAP80 protein was incubated with the same set of GST fusion proteins shown in Fig. Fig.2A,2A, and the specific interactions of these proteins were analyzed by GST pulldown assays. Like human TRAP80, dTRAP80 specifically interacted with the functional wild-type version of the VP16 activation domain (Fig. (Fig.3A,3A, lane 4), but not with the transcriptionally defective mutant activation domain (lane 3). dTRAP80 also interacted with the activation domains of Dif and HSF (lanes 6 and 7). However, under the same binding conditions, the activation domains of Armadillo and Notch (lanes 5 and 8) failed to bind dTRAP80 in vitro. Intriguingly, dTRAP80 interacted with the RXR LBD GST fusion protein, but only in the presence of 9CRA (lanes 9 and 10). However, as was the case with the complete Mediator complex, dTRAP80 did not interact with the RAR LBD even in the presence of ATRA (lanes 11 and 12). Therefore, dTRAP80 appears to be a binding target for some, but not all, transcription activators that interact with Mediator. Armadillo and Notch may interact with Mediator by binding to a different Mediator subunit or require some other intermediary factors which are absent in the binding reaction for their interaction with dTRAP80.
Specific interaction of dTRAP80 with various transcriptional activation domains. (A) Detections of interactions between dTRAP80 and various transcriptional activation domains. GST and the GST-fused activation domains indicated above the gel were immobilized on GST-agarose beads and incubated with 35S-labeled dTRAP80 prepared by in vitro translation. For RXR and RAR, the ligands were added at a concentration of 10−6 M (lanes 10 and 12). Proteins were analyzed by autoradiography on a sodium dodecyl sulfate-polyacrylamide gel. m, mutant; w, wild type; TnT, in vitro transcription and translation. (B) GST pulldown assay. Five overlapping 35S-labeled dTRAP80 fragments (in vitro translated; schematics shown to the left of the figure) were incubated with GST-fused activation domains (indicated at the top of the figure) immobilized on GST-agarose beads. For GST-RXR, 10−6 M 9CRA was added before incubation. Proteins were analyzed by autoradiography on a sodium dodecyl sulfate-polyacrylamide electrophoresis gel.
In order to map the activator-binding regions of dTRAP80, five overlapping fragments for the entire dTRAP80 coding region were in vitro translated and analyzed by GST pulldown assay for the ability to interact with the GST-fused activation domains of Dif, HSF, and RXR (Fig. (Fig.3B).3B). Two regions of dTRAP80, designated B and D, interacted strongly with all three activator proteins tested. Binding of region B to the RXR activation domain was ligand dependent, while binding to the D domain was not. Because full-length dTRAP80 did not interact with the RXR activation domain in the absence of ligand (Fig. (Fig.3A,3A, lane 9), the interaction between region D and the various activation domains appears to be nonspecific. Binding to the D fragment might be caused by a sticky region that is exposed in the small fusion fragment but hidden in properly folded, full-length dTRAP80. Further fragmentation of region B resulted in the loss of dTRAP80 binding, indicating that a structural feature produced by the entire B region may be required for interaction with activation domains (data not shown). Therefore, the amino acid sequence contained in region B likely serves as a specific binding site for the Dif, HSF, and RXR activators.
Artificial recruitment of Mediator via dTRAP80 stimulates transcription from the hsp70 promoter.
In order to test whether Mediator recruitment to the promoter via an activator-binding domain is sufficient for transcriptional activation in Drosophila cells, we utilized an artificial recruitment assay. Earlier experiments have shown that Mediator components can be recruited to composite Pol II promoters by being tethered to a heterologous DBD. This artificial recruitment of individual Mediator subunits to a promoter by fusion with a DBD causes robust transcriptional activation in yeast, where Mediator is tightly associated with the holoenzyme form of Pol II (2, 12). However, several attempts at the artificial recruitment of human Mediator components did not result in the activation of transcription in mammalian cells (10). To test whether the artificial recruitment of Drosophila Mediator proteins to promoters can activate transcription, we tested the G4DBD alone or G4 derivatives fused to several Mediator components (dTRAP80, dMED6, dSOH1, and dSRB7) for the ability to activate a luciferase reporter gene containing five Gal4-binding sites upstream of the adenovirus E1b core promoter (G5-E1b-luc). The basal expression level of G5-E1b-luc induced by G4DBD was very low but markedly elevated (approximately 5,000-fold) by the G4-HSF (Fig. (Fig.4A).4A). However, none of the G4-Mediator fusion constructs was able to activate G5-E1b-luc. Therefore, the artificial recruitment of Mediator alone to the E1b promoter was not sufficient for transcriptional activation to occur. Because Mediator and Pol II were recruited independently to certain metazoan promoters, we reasoned that the failure of artificial recruitment resulted from the absence of independently recruited basal transcription machinery on the E1b promoter.
Artificial recruitment of Mediator proteins to two different promoters in SL2 cells. (A and B) Effects of Mediator recruitment to the promoters. SL2 cells were transiently transfected with the G5-E1b-luc reporter (A) or G2-hsp70-luc reporter (B) plasmid, and the G4 fusion constructs are indicated at the bottom of the graphs. For normalization of the assay, an actin-lacZ plasmid was cotransfected and their β-galactosidase activities were used for normalization as described in the legend to Fig. Fig.1.1. (C) Immunoblot analysis (IB) of the expression levels of the G4-Mediator fusion proteins. Cells transfected with the G4 fusion construct indicated at the top of each panel were analyzed with the antibodies indicated below the gels. The specific proteins are marked to the right of each panel. (D) ChIP of SL2 cells that have chromosomally integrated copies of G2-hsp70-luc. The integrated reporter cell lines were subjected to transient transfection with the indicated G4 fusion constructs and then analyzed by a ChIP assay with anti-dSOH1 antibody. Chromatin-immunoprecipitated DNAs and 1% of the cross-linked chromatin used in the immunoprecipitation (Input) were amplified with primers specific to the integrated reporter. The amplified products were analyzed by polyacrylamide gel electrophoresis and autoradiography.
We next performed a similar experiment with a native heat shock promoter (G2-hsp70-luc), where the basal transcription machinery is known to be preassembled in vivo (39a). G2-hsp70-luc showed a considerable amount of basal transcription, even with G4DBD alone (Fig. (Fig.4B).4B). These basal transcription levels were further activated 5.4-fold by G4-HSF but not by the dMED6, dSOH1, or dSRB7 fusion constructs. However, G4-dTRAP80 increased G2-hsp70-luc expression 4.8-fold (Fig. (Fig.4A).4A). Therefore, transcriptional activation requires not only the recruitment per se but also proper positioning of Mediator on the promoter. The recruitment of Mediator via an activator-binding component, G4-dTRAP80, might increase the efficiency of productive preinitiation or elongation complex formation.
To rule out the possibility of a variation in the expression of individual G4-Mediator fusion proteins in transfected cells, we examined the concentrations of G4-Mediator derivatives in the transfected SL2 cells. Immunoblot analysis with anti-Mediator antibodies showed that the G4-Mediator proteins were expressed in amounts comparable to those of the corresponding endogenous Mediator proteins (Fig. (Fig.4C).4C). To further demonstrate that similar amounts of each of the G4-Mediator derivatives were recruited to the promoter, we analyzed promoter-bound proteins by the ChIP assay performed with antibody for one of the endogenous Mediator subunits, dSOH1. The amount of Mediator, or at least another Mediator protein, dSOH1, bound to the promoter increased significantly with the addition of G4-HSF, reflecting the transcriptional activation activity shown in Fig. 4A and B. Despite the differences in transcriptional activation observed with the various G4-Mediator fusions, artificial recruitment of Mediator by G4-dMED6, -dSRB7, and -dTRAP80 was equally successful (Fig. (Fig.4D).4D). Therefore, the transcriptional activation by only the dTRAP80 fusion protein appears to reflect its role as an activator-binding protein suitable for the proper positioning of Mediator between the activator protein and the basal transcription machinery when fused to the DBD.
Activator-specific requirement of dTRAP80 for transcriptional activation.
The previous series of experiments strongly suggest that dTRAP80 is a genuine activator-binding target in cells. In order to verify the in vivo significance of the activator-specific interaction of dTRAP80, we examined in cells the effect of depletion of dTRAP80 on transcriptional activation by the various transcriptional activator proteins. To this end, we generated dTRAP80-deficient cells by using dsRNAi. Immunoblot analysis revealed that the amount of dTRAP80 protein was not diminished in cells transfected with luciferase dsRNA, while the dTRAP80 protein was lost completely from the cells transfected with dTRAP80 dsRNA (Fig. (Fig.5A).5A). Immunoblot analysis of dTRAP80-deficient cells revealed that some of the dTRAP80-interacting Mediator proteins (dMED6 and Trfp) were diminished, but most other Mediator proteins tested were unaffected by the dTRAP80 deficiency (15a). When the dTRAP80 RNAi-treated cell extracts were analyzed by Superose 6 gel filtration, most of the Mediator complex still behaved as a megadalton size complex (data not shown). Therefore, the dTRAP80 RNAi appears to cause loss of the dTRAP80 module (dTRAP80 along with some other dTRAP80-interacting Mediator proteins) from the complex.
Activator-specific requirement for dTRAP80 in transcriptional activation. (A) Depletion of dTRAP80 protein in SL2 cells treated with dTRAP80 dsRNA. SL2 cells were transfected with dsRNA from the coding regions of either luciferase or dTRAP80. The concentrations of the dTRAP80 and dTBP proteins were analyzed by immunoblot analysis with antibodies directed against the proteins indicated to the left. (B) dsRNAi scheme. After dsRNA treatment, induction of activator proteins with CuSO4, addition of ligand (where appropriate) or luciferase, and green fluorescent protein (GFP) analyses were carried out within the indicated time frame. (C) SL2 cells treated with mock or dTRAP80 dsRNA were transfected with various G4 fusion protein expression constructs indicated at the bottom of the graph and G5-E1b-luciferase reporter DNA. Twelve to 24 h after induction of activator activity, the cells were harvested and assayed for luciferase activity. Results from three independent experiments were averaged. Their mean luciferase activity values and the sizes of the standard deviations are shown. The luciferase activities were normalized against the lacZ activity from a cotransfected actin-LacZ plasmid. Fold activation caused by each transcriptional activator compared to that by G4DBD in mock- or dTRAP80 RNAi-treated cells is shown. Fold reductions of transcriptional activation activity caused by dTRAP80 RNAi are shown at the bottom.
To examine the effect of dTRAP80 depletion on transcriptional activation in vivo, a G5-E1b-luciferase reporter was transfected along with one of the G4 activator derivatives into dTRAP80 dsRNA-treated and mock-treated cells and their luciferase activities were examined in triplicate after inducing the expression of the synthetic activator proteins (Fig. (Fig.5B).5B). To confirm that all cells used in the assay had the same transfection and activator induction efficiencies, the amount of the synthetic activators induced in each assay was examined. Immunoblot analysis with anti-Gal4 and anti-TATA-binding protein (TBP) antibodies revealed that the amounts of the G4 activation domain (G4AD) derivatives induced per cell were consistent (data not shown). In contrast to the lack of luciferase activation by G4DBD, induction of G4 Armadillo (G4Arm), G4Dif, and G4Notch in the mock-treated cells led to a nearly 100-fold activation of luciferase expression. G4RXR also activated luciferase 40-fold in a ligand-dependent manner (Fig. (Fig.5C).5C). dTRAP80 dsRNAi caused a reduction (about 1/3) in luciferase activity in all of the assays. However, there were quantitative differences in the effects of dTRAP80 dsRNAi on transcription. For example, transcriptional activation by G4Dif, G4Hsf, and G4RXR (in the presence of specific ligand) was significantly reduced by the addition of dTRAP80 dsRNA (23, 39, and 31% of the levels in the mock-treated samples, respectively) whereas the effect of dTRAP80 dsRNA on G4Arm- and G4Notch-induced luciferase expression (84 and 79% of the levels in the mock-treated samples, respectively) was relatively small (Fig. (Fig.5C).5C). Thus, transcriptional activation by dTRAP80-binding activators was affected most severely by the presence of dTRAP80 dsRNAi. Taken together, these results indicate that the dTRAP80 module is required for activator-specific transcriptional activation in vivo.
dTRAP80 is required for Dif-induced transcriptional activation in the endogenous Toll signaling pathway.
Among the activator proteins whose activation domains depend on dTRAP80 for transcriptional activation, Dif was further analyzed to examine the dTRAP80 requirement for transcriptional activation under physiological conditions. To this end, we analyzed expression of the drosomycin gene, which is induced by the Toll signaling cascade. It was previously shown that stimulation of the Toll receptor by Spaetzle activates a downstream signaling cascade that culminates in the nuclear translocation of Dif and the subsequent activation of drosomycin transcription (40). First, we cotransfected a copper-dependent Dif expression vector and a drosomycin promoter-luciferase reporter gene (droso-luciferase) to the SL2 cells that had been treated with either control or dTRAP80 dsRNA. After Dif overexpression was induced by the addition of cupric sulfate, activation of the reporter gene by Dif was analyzed with the luciferase assay. The overexpression of Dif in these cells led to a 38-fold activation of droso-luciferase. However, dTRAP80 dsRNAi reduced Dif activation of the drosomycin reporter to less than twofold (Fig. (Fig.6A).6A). This result indicates that the Dif activation domain requires the dTRAP80 module, not only in the context of the G4DBD fusion but within the natural Dif context.
Requirement for dTRAP80 in drosomycin gene induction by the Toll signaling pathway. (A) Effect of dTRAP80 dsRNAi on the droso-luciferase gene activation by Dif. SL2 cells treated with control or dTRAP80 dsRNA were cotransfected with a Dif expression vector (solid bars) or an empty vector (open bars) and a droso-luciferase reporter gene and assayed for luciferase activity in triplicate. (B) Effect of dTRAP80 RNAi on the induction of drosomycin transcription by the Toll signaling pathway. SL2 cells treated with control or dTRAP80 dsRNA were transfected with a TollΔLRR expression vector (solid bars) or an empty vector (open bars). Total RNA isolated from each sample was analyzed by real-time PCR in order to measure the relative amounts of dTRAP80 mRNA (upper graph) and drosomycin mRNA (lower graph). RNA concentrations were normalization according to the amounts of rp49 mRNA in the samples.
Next, we examined whether Dif depends on the dTRAP80 module for the transcriptional activation of endogenous target genes. We activated the Toll signaling pathway by introducing a constitutively active Toll derivative into cells; this derivative, termed TollΔLRR, lacks an extracellular leucine-rich repeat region (52). We then measured the induction of endogenous drosomycin gene transcription by TollΔLRR in SL2 cells treated with control or dTRAP80 dsRNA. The amounts of mRNA from the dTRAP80 and drosomycin genes were analyzed by the real-time PCR method. The expression of TollΔLRR in SL2 cells had no effect on the concentration of dTRAP80 RNA. In contrast, treatment of SL2 cells with dTRAP80 dsRNA caused a fourfold reduction in the amount of endogenous dTRAP80 RNA. Expression of TollΔLRR in SL2 cells caused a fourfold increase in endogenous drosomycin expression (Fig. (Fig.6B).6B). However, dTRAP80 dsRNA treatment abolished TollΔLRR-induced drosomycin transcription from the chromosomal loci to nearly background amounts. Therefore, the dTRAP80 module appears to act as a physiological binding target for the Dif activation domain and to regulate drosomycin gene induction via the Toll signaling pathway.
DISCUSSION
Functional analyses of yeast Mediator proteins have shown that different Mediator subunits are required for the transcriptional activation of different genes (18). In other words, not all genes require the same Mediator components for communication between their gene-specific activator proteins and the basal transcription machinery. In an effort to identify metazoan activation domains that depend on a specific subunit of the Mediator complex for transcriptional activation, we examined several distinct activation domains from different transcriptional activators for their ability to interact with dTRAP80 in vitro. This approach permitted us to identify a subset of activation domains that interact physically with dTRAP80 and thus to recognize dTRAP80 as an activator-specific binding target within the metazoan Mediator complex. Intriguingly, the dTRAP80 Mediator module is essential not only for transcriptional activation by G4AD synthetic activator proteins but for transcriptional activation by the naturally occurring Dif protein. This requirement was also observed under conditions that involve the activation of endogenous Dif by Toll receptor signaling. Although dTRAP80 RNAi caused a concurrent loss of several dTRAP80-interacting Mediator proteins from the complex, the integrity of the entire Mediator complex does not appear to have been affected. This is reminiscent of the concurrent loss of the Gal11, Hrs1, and Med2 proteins from the activator-binding module of the yeast gal11 mutant Mediator (29). Therefore, the transcriptional activation defects of the dTRAP80 dsRNA-treated cells may be a result of the compounded effect of several Mediator protein deficiencies. However, the activator-specific interactions in vitro and the selective transcriptional defects on Dif-derived transcription over Armadillo- and Notch-derived transcription strongly indicates that dTRAP80 acts as a physiological mediator of the signal from the Drosophila Toll receptor.
The Drosophila Dif protein shares overall sequence homology with the NF-κB p65 subunit found in mammalian cells (20). As p65 is the only NF-κB family member that possesses a strong transcriptional activation domain (it is functional when transferred to a heterologous DBD) (43), Dif appears to be a functional as well as structural homolog of mammalian p65. In this regard, it would be interesting to test whether TRAP80 mediates transcriptional regulation by NF-κB p65 in mammalian cells.
All the transcriptional activators that possess an activation domain rich in acidic amino acid residues, including Dif and HSF in this study, have been shown to bind TRAP80 homologs. TRAP80 was suggested as the likely metazoan homolog of yeast Srb4 (4). The yeast Srb4 protein was shown to interact with Gal4 transcriptional activator protein, and mutation in the Gal4-interacting domain of Srb4 crippled the Gal4-mediated activation in vivo (28a). On the other hand, other classes of activator proteins that do not contain acidic activation domains utilize other metazoan-specific Mediator subunits as their binding targets; for example, TRAP220/DRIP205 binds nuclear hormone receptors and Sur-2 recognizes E1A and Elk-1 (7, 14, 23, 47). This activator specificity may have evolved in order to accommodate the demand for Mediator proteins to interact with diverse transcriptional activators. Presently, we do not know what the Mediator subunit targets for Armadillo and Notch are. However, it is noteworthy that mutations in the Drosophila and Caenorhabditis elegans homologs of TRAP230 and TRAP240 cause specific developmental phenotypes that share some features with phenotypes that occur along with defects in gene expression regulated by the Wnt/Wingless and Notch signaling pathways (53, 56). Because Armadillo/β-catenin and Notch are the nuclear end points of these signaling pathways, respectively, TRAP230 and TRAP240 may serve as direct binding targets for and control transcriptional activation by these activators. Therefore, TRAP80 and other activator-specific binding targets in the yeast and metazoan Mediator complexes (23, 36, 47) appear to regulate the promoter-specific Mediator recruitment via interaction with distinct enhancer-bound transcriptional activators. Lastly, we observed that G4-dTRAP80 that is recruited to hsp70 via the G4DBD can trigger subsequent transcriptional activation steps whereas other G4-Mediator fusion proteins tested here cannot (Fig. 4A and B). The differential effect in the artificial recruitment studies is consistent with the notion that dTRAP80 is a physiological activator-binding site.
Acknowledgments
Jin Mo Park and Jung Mo Kim contributed equally to this work.
We thank Won-Jae Lee, Young Chul Lee, John T. Lis, Tony Ip, Jaeseob Kim, Michael Young, Mark Peifer, Michael Green, Masamitsu Yamaguchi, Arnold Levine, and Jean-Luc Imler for providing immunocompetent SL2 cells and cloned materials for the Notch, Armadillo, D-Stat, Dif, Dmp53, and Drosomycin genes. We thank John T. Lis and Soyoun Kim for sharing artificial recruitment results prior to publication. We are indebted to Kelly LaMarco for careful reading of the manuscript.
This work was supported by a Creative Research Initiatives grant from the Ministry of Science and Technology, Korea, and Human Frontier Science Program grants to Y.J.K.