MED19 and MED26 Are Synergistic Functional Targets of the RE1 Silencing Transcription Factor in Epigenetic Silencing of Neuronal Gene Expression^,

Abstract

A key hub for the orchestration of epigenetic modifications necessary to restrict neuronal gene expression to the nervous system is the RE1 silencing transcription factor (REST; also known as neuron restrictive silencer factor, NRSF). REST suppresses the nonspecific and premature expression of neuronal genes in non-neuronal and neural progenitor cells, respectively, via recruitment of enzymatically diverse corepressors, including G9a histone methyltransferase (HMTase) that catalyzes di-methylation of histone 3-lysine 9 (H3K9me2). Recently, we identified the RNA polymerase II transcriptional Mediator to be an essential link between RE1-bound REST and G9a in epigenetic suppression of neuronal genes in non-neuronal cells. However, the means by which REST recruits Mediator to facilitate G9a-dependent extra-neuronal gene silencing remains to be elucidated. Here, we identify the MED19 and MED26 subunits in Mediator as direct physical and synergistic functional targets of REST. We show that although REST independently binds to both MED19 and MED26 in isolation, combined depletion of both subunits is required to disrupt the association of REST with Mediator. Furthermore, combined, but not individual, depletion of MED19/MED26 impairs REST-directed recruitment to RE1 elements of Mediator and G9a, leading to a reversal of G9a-dependent H3K9me2 and de-repression of REST-target gene expression. Together, these findings identify MED19/MED26 as a probable composite REST interface in Mediator and further clarify the mechanistic basis by which Mediator facilitates REST-imposed epigenetic restrictions on neuronal gene expression.

The specification and maintenance of neuronal identity within the developing vertebrate nervous system derives from the influence of both genetic and epigenetic programs that combine to establish unique spatiotemporal patterns of neuronal-specific gene expression. Expressed genes that confer unique and highly specialized morphological, biochemical, and physiological properties on individual neuronal subtypes must be suppressed in non-neuronal tissues, and the regulatory mechanisms that coordinate these processes are fundamentally important for proper nervous system development and function (1–3). A key factor in the orchestration of epigenetic modifications that restrict the expression of neuronal genes to the nervous system is the RE1 silencing transcription factor (REST,2 also known as neuron restrictive silencer factor, NRSF) (4, 5).

REST is a Kruppel-type zinc finger transcription factor that binds to a 21-bp RE1 silencing element present in over 900 human genes, many of which encode proteins with dedicated roles in neuronal determination, identity, and function (4–10). REST occupies a central role in non-neuronal lineage restriction through its ability to suppress the nonspecific and premature expression of neuronal genes in non-neuronal cells and neural progenitor cells, respectively (4, 5, 10, 11). Consistent with such a role, the expression of REST is dominantly constrained in non-neuronal and neural progenitor cells, although low levels of REST protein are maintained in some populations of postmitotic neurons, most notably those of the hippocampus (10, 12–15). Functional inactivation of REST in vertebrates leads to early embryonic lethality and ectopic expression of neuronal genes in non-neuronal tissues (16), whereas its forced overexpression causes axon-pathfinding errors (17). Misregulation of REST-directed repression has been linked with a variety of pathologic conditions in humans, including Huntington's disease, epilepsy, ischemia, dilated cardiomyopathy, X-linked mental retardation, and cancer (18–30). Taken together, these observations reveal fundamental links between REST and vertebrate development and disease, and emphasize the importance of a more comprehensive understanding of REST-mediated gene repression.

In this regard, REST has previously been characterized as a bipartite transcriptional repressor harboring two spatially and functionally distinct repression domains: one spanning its N-terminal 83 amino acids and a second encompassing its C-terminal zinc finger (31–36). Mechanistically, the N- and C-terminal repression domains in REST have been shown to exert repressive activity through recruitment of the SIN3/ HDAC and CoREST/HDAC/LSD1 corepressor complexes, respectively, both of which function to impose restrictive epigenetic modifications on the chromatin structure of REST target genes (31–36).

Recently, we identified a comparably potent, yet previously uncharacterized, internal repression domain in REST (amino acids 141–600) encompassing its DNA-binding domain followed by a lysine-rich region (21). We found that REST-(141–600) directly recruits a distinct corepressor complex comprising Mediator, a multisubunit global coregulator of RNA polymerase II transcription, and G9a HMTase, an enzyme dominantly responsible for transcriptionally repressive histone 3 lysine-9 mono-(H3K9m) and di-methylation (H3K9me2) within mammalian euchromatin (21). In contrast to the well-established role of Mediator as a bridge between DNA-bound activators and the RNA polymerase II general transcription machinery, our findings revealed a critical requirement for Mediator in recruitment of enzymatically active G9a by RE1-bound REST, thus revealing Mediator to be a direct link between REST and G9a-dependent H3K9me2 required for extra-neuronal gene silencing (21). Nonetheless, key elements of this repressive protein interaction network remain to be established, including the identity of the Mediator subunit(s) with which REST directly interfaces to recruit Mediator/G9a onto RE1 elements.

Here, using an unbiased in vitro protein interaction screen to identify REST-binding subunits in Mediator, we identified MED19 and MED26 as candidate REST target subunits. We validated independent association of REST with both MED19 and MED26 in isolation, but nonetheless found that combined depletion of both subunits was required to disrupt the association of REST with Mediator. Furthermore, we found that combined, but not individual, depletion of MED19/MED26 impairs REST-directed recruitment to RE1 elements of Mediator, G9a, and G9a-dependent H3K9me2, leading to de-repression of REST-target genes in vivo. Collectively, these findings identify MED19 and MED26 as synergistic physical and functional targets of REST in Mediator and further clarify the mechanistic basis by which Mediator facilitates REST-imposed epigenetic restrictions on neuronal gene expression.

EXPERIMENTAL PROCEDURES

Plasmids—Plasmids for in vitro transcription/translation and/or mammalian expression of REST, MED1, 6, 7, 8, 9, 10, 11, 12, 14, 15, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, and CDK8L have been described (37–45). pCS3+CDK8-FLAG, pCS3+CycC-FLAG, pCS3+MED31-FLAG, pCS3+MED16-FLAG, and pCS3+MED17-FLAG were constructed by subcloning PCR-amplified corresponding cDNAs into XhoI/ClaI-linearized pCS3+ vectors bearing FLAG epitope tag sequence. pCS2+MED4-His₆-FLAG and pCS2+MED30-His₆-FLAG were constructed by subcloning PCR-amplified cDNAs encoding MED4 and MED30, respectively, into EcoRI/ClaI linearized pCS2+ vectors bearing His₆ and FLAG epitope tag sequences. pCS2+CBP-MED13 was constructed by subcloning a PCR-amplified MED13 cDNA into a pCS2+ vector bearing CBP (calmodulin-binding peptide) epitope tag sequence. REST truncation derivatives used in GST pull-down assays and transient reporter-based transcriptional repression assays have also been described (21).

Antibodies—Antibodies used for immunoprecipitation and Western blot analyses correspond to the following: MED1 (sc-8998 and sc-5334), MED6 (sc-9443), CDK8 (sc-1521), REST (sc-15118), MED16 (sc-5366), MED17 (sc-12453), and MED26 (sc-81237) were purchased from Santa Cruz Biotechnology; MED12 (A300–774A) was purchased from Bethyl Laboratories; MED15 (H00051586-M02) was purchase from Abnova; MED23 (551175) and CCNC (558903) were purchased from BD Pharmingen; CDK8 (RB-018) was purchased from Lab Vision Corp.; G9a (G6919) antibody were purchased from Sigma; H3K9me2 (07-441) were purchased from Upstate Biotechnology. Murine HA monoclonal antibody was purchased from Roche Applied Science. Production and purification of murine G9a and rabbit MED4/30 polyclonal antibodies has been described (21, 45). Rabbit polyclonal anti-MED19 serum has also been described (42, 44).

Cell Culture, Transfections, RNA Interference, and Reporter Assays—HeLa cells were obtained from American Type Culture Collection and cultured in Dulbecco's modified Eagle's medium (Invitrogen) medium with 10% bovine growth serum (Hyclone). DNA transfections were performed using FuGENE 6 (Roche Applied Science) and siRNA transfections using TransIT-siQUEST (Mirus Bio Corp.) transfection reagents following the manufacturer's instructions. For siRNA transfections, cells (∼60% confluent) were transfected with siRNAs at a final concentration of 20 nm for 3 days before further analyses. siRNAs (Dharmacon) correspond to the following: MED19 (J-016056-11); MED26 (J-011948-09); control non-target siRNA (D-001210-01).

GST Pull-Down, Immunoprecipitation, and Chromatin Immunoprecipitation Analyses—For GST pull-down assays using radiolabeled recombinant Mediator subunits or HeLa nuclear lysates, GST derivatives were immobilized on glutathione-Sepharose beads and washed extensively with Lysis 250 buffer (50 mm Tris-HCl, 250 mm NaCl, 5 mm EDTA) containing 0.5% Triton X-100 prior to incubation with either radiolabeled Mediator subunits or HeLa nuclear lysates (dialyzed against 0.1 m KCl D buffer (20 mm HEPES, pH 7.9, 0.2 mm EDTA, 20% glycerol) for no less than 4 h). Beads were washed five times with 0.3 m KCl D buffer containing 0.2% Nonidet P-40 and eluted with Laemmli sample buffer followed by SDS-PAGE and WB or autoradiography analysis. For immunoprecipitation of intact Mediator, nuclear lysates (0.5 mg) prepared as described previously (45) were adjusted to 0.1 m KCl and 0.1% Nonidet P-40 and subjected to overnight immunoprecipitation at 4 °C using protein A-Sepharose conjugated to anti-MED4 antibody. Immunoprecipitates were washed 5 times with 0.3 m KCl D buffer containing 0.2% Nonidet P-40, eluted in Laemmli sample buffer, and processed by SDS-PAGE for Western blot analysis. For chromatin immunoprecipitation assays, cells were cross-linked, and harvested into cell lysis buffer (5 mm HEPES pH 7.9, 85 mm KCl and 0.5% Triton X-100) and then pelleted and resuspended in nuclei lysis buffer (50 mm Tris-HCl, pH 8.0, 10 mm EDTA pH 8.0, and 1% SDS). Chromatin was solubilized and sheared by pulsed sonication (Fisher Scientific, Model 100) and clarified by high-speed centrifugation. Chromatin-containing fractions were diluted 10-fold in dilution buffer (50 mm Tris-HCl pH 8.0, 2 mm EDTA pH 8.0, 150 mm NaCl, and 1% Triton X-100) followed by incubation with primary antibodies as indicated at 4 °C overnight. Immune complexes were precipitated with a pre-blocked mix of protein G-agarose and protein A-Sepharose for 2 h followed by sequential washes with sarcosyl buffer (TE and 0.2% Sarcosyl), low-salt buffer (0.1% SDS, 1% Triton X-100, 2 mm EDTA pH 8.0, 20 mm Tris-HCl pH 8.0, and 150 mm NaCl), high-salt buffer (0.1% SDS, 1% Triton X-100, 2 mm EDTA pH 8.0, 20 mm Tris-HCl pH 8.0, and 500 mm NaCl), LiCL detergent buffer (10 mm Tris-HCl pH 8.0, 1 mm EDTA pH 8.0, 1% deoxycholate, 1% Nonidet P-40, and 250 mm LiCl) and TE. DNA was recovered in elution buffer (1% SDS, 50 mm Tris-HCl, pH 8.0 and 10 mm EDTA, pH 8.0) and subjected to proteinase K treatment and decross-linking at 65 °C overnight. DNA was purified by phenol/chloroform (1:1) extraction and ethanol precipitation and resuspended in double distilled water. RE1 occupancy levels are expressed relative to RE1 occupancy levels in control siRNA-transfected cells. Primer sequences for quantitative chromatin immunoprecipitation assays have been described (21).

Reverse Transcription-qPCR (RT-qPCR) Analyses—RT-qPCR analyses have been described previously (21). Briefly, RNA extracted from HeLa cells transfected 3 days prior with specified siRNAs, was subjected to reverse transcription and real-time PCR analyses Results represent the average of three independent experiments performed in duplicate. mRNA levels are expressed relative to mRNA levels in control siRNA-transfected cells. Sequences of primers used in RT-qPCR analyses are as follows: β-Actin (5′-CAAAGACCTGTACGCCAACACAGT-3′ and 5′-ACTCCTGCTTGCTGATCCACATCT-3′); MED19 (5′-TGGTTCCCATGATAACAGCAGCCT-3′ and 5′-CGGCTCTGTTTGTGCTTGTGCTTA-3′); MED26 (5′-AAACCTCTGACCCAGAAAGAGCCA-3′ and 5′-ACAGCTCCTTCCAGTTCGTCTGTT-3′); SNAP25 (5′-AACTGGAACGCATTGAGGAAG-3′ and 5′-GGTCCGTCAAATTCTTTTCTGC-3′); Syn1 (5′-GGTCTCTGAAGCCGGATTTTG-3′ and 5′-GTCCCCAGTTTCTTATGCAGTC-3′); M4 (5′-TTCATCCAGTTCCTGTCCAACCCA-3′ and 5′-GGCTTCTTGACGCTCTGCTTCATT-3′).

RESULTS

REST Independently Binds to MED19 and MED26 both in Vitro and in Vivo—In an initial attempt to identify the REST target subunit(s) in Mediator, we screened 31 of 33 possible mammalian Mediator subunits for in vitro interaction with a GST-REST derivative expressing REST amino acids 141–600 and therefore encompassing the REST DNA-binding domain followed by a lysine-rich region. Recently, we showed that this domain harbors autonomous repression activity in a manner requiring its direct interaction with Mediator (21). Mediator subunits omitted from this screen included MED12L and MED13L, for which cDNAs are currently unavailable. Among the remaining Mediator subunits tested, only MED19 and MED26 exhibited significant REST binding activity, with MED19 possessing an apparent greater affinity for REST than MED26 (supplemental Figs. S1–S3). Interestingly, comparative genomic analyses indicate that whereas MED19 is broadly represented throughout the eukaryotic kingdom, with identifiable orthologs present in species spanning humans to fungi, MED26 appears to be restricted to higher eukaryotes, as no identifiable ortholog is present in fungi.

To confirm the interaction specificities of MED19 and MED26 for REST amino acids (aa) 141–600, we also tested these two Mediator subunits for their respective abilities to bind to REST N-terminal (aa 1–140) and C-terminal (aa 601–1098) fragments. Notably, MED19 and MED26 bound only to REST (141–600) corresponding precisely to the Mediator-binding domain on REST (Fig. 1A) (21). To determine if REST binds to MED19 and MED26 in vivo, we examined the ability of REST to coimmunoprecipitate along with either MED19 or MED26 following their ectopic expression in HEK 293 human embryonic kidney cells. This analysis revealed specific and efficient precipitation of MYC-REST by FLAG-specific antibodies only in the presence, but not in the absence, of either FLAG-MED19 or FLAG-MED26 (Fig. 1B). Taken together, the results of in vitro and in vivo binding analyses reveal that REST-(141–600) interacts specifically and selectively with both MED19 and MED26, suggesting that either one or both of these subunits might represent a functionally important target(s) of REST in Mediator.

FIGURE 1.

MED19 and MED26 bind specifically to REST. A, recombinant full-length MED19 and MED26 proteins were expressed and radiolabeled with [³⁵S]methionine by translation in vitro prior to incubation with glutathione-Sepharose-immobilized GST or GST-REST derivatives as indicated. Bound proteins were eluted with Laemmli sample buffer, resolved by 15% SDS-PAGE, and visualized by phosphorimager analysis. Input, 10% of each in vitro translated protein used in binding reactions. The amount of MED19 and MED26 bound by GST-REST-(141–600) was quantified and expressed as a percentage of the total input. Results are representative of at least three independent binding experiments. B, Myc-REST was expressed with or without FLAG-MED19 or FLAG-MED26 in HEK 293 cells prior to immunoprecipitation (IP) of whole cell lysates using antibodies specific for the FLAG epitope as indicated. Immunoprecipitates were resolved by SDS-PAGE and processed by Western blot (WB) analysis using FLAG- or Myc-specific antibodies as indicated. Input, 10% of the nuclear lysate used for IP reactions. Asterisk indicates the position of IgG heavy chain.

MED19 and MED26 Synergistically Mediate the Interaction between REST and Mediator—As a precondition for exploring potential functional interactions between REST and MED19/MED26 through the use of RNAi in mammalian cells, we evaluated the impact of depleting either or both of these subunits on the integrity of Mediator. To this end, we immunoprecipitated Mediator from HeLa cells following siRNA-mediated depletion of MED19 and/or MED26, and examined the immunoprecipitates for the presence of Mediator subunits representing each of four structurally apparent Mediator subdomains (Head, Middle, Tail, and Kinase) (46–48). We observed that neither individual nor combined depletion of MED19/MED26 significantly altered the apparent stability of other Mediator subunits examined or their stable incorporation into Mediator (Fig. 2). This finding is consistent with the prior observation that Saccharomyces cerevisiaie Mediator isolated under physiological conditions from a MED19-deficient yeast strain is not structurally compromised (49). Thus, under physiological conditions, neither MED19 nor MED26 are likely to be essential for the structural integrity of Mediator.

FIGURE 2.

MED19 and MED26 are not essential for the apparent integrity of Mediator. Nuclear lysates from HeLa cells transfected with control (CNTL), MED19-, MED26-, or MED19 and 26-specific siRNAs were subjected to immunoprecipitation (IP) using antibodies specific for MED4. Immunoprecipitates were extensively washed prior to resolution by SDS-10% PAGE and processing by Western blot (WB) analysis using the specified antibodies. Input, 10% of the nuclear lysates used for IP reactions. Structural domains to which individual Mediator subunits may be relegated are indicated (Head, Middle, Tail, Kinase, or Unassigned (Unass.)). Asterisks mark MED19 and MED26 Mediator subunits targeted for RNAi-mediated depletion.

Next, we examined the ability of Mediator deficient in MED19 and/or MED26 to associate with REST. To this end, GST-REST-(141–600) was tested for its ability to bind Mediator present in nuclear extracts from HeLa cells depleted of MED19 and/or MED26 by RNAi. Consistent with our recent delineation of REST amino acids 141–600 as Mediator-binding domain (21), we confirmed that GST-REST-(141–600) bound Mediator present in nuclear extracts from HeLa cells transfected with control siRNA (Fig. 3, lanes 1 and 5). Notably, we observed that only combined, but not individual, depletion of MED19/MED26 significantly impaired the interaction between Mediator and REST-(141–600) (Fig. 3, lanes 2–4 and 6–8). These results confirm that REST physically associates with Mediator, likely through both its MED19 and MED26 subunits.

FIGURE 3.

MED19 and MED26 synergistically mediate the interaction between REST and Mediator. Nuclear lysates from HeLa cells transfected with control (CNTL), MED19-, MED26-, or MED19 and 26-specific siRNAs as indicated were incubated with immobilized GST-REST-(141–600). Specifically bound proteins were resolved by SDS, 10% PAGE prior to WB analysis using the specified antibodies. Input, 10% of the nuclear lysates used in binding reactions.

MED19 and MED26 Are Synergistically Required for REST Repressor Activity—Based on our recent finding that REST-dependent recruitment of Mediator is essential to link RE1-bound REST with G9a in epigenetic gene silencing, we hypothesized that the interaction between REST-(141–600) and MED19/MED26 might therefore be critical for REST-directed repression. As a direct test of this hypothesis, we investigated the impact of MED19 and/or MED26 depletion on the repressive activity of REST-(141–600) using a transient reporter-based transcriptional repression assay. To this end, REST-(141–600), tethered to the GAL4 DNA-binding domain, was tested for its ability to repress transcription from a constitutively active GAL4-responsive reporter plasmid in HeLa cells depleted of MED19 and/or MED26 by RNAi. As controls for these experiments, we also monitored the impact of MED19 and/or MED26 depletion on the respective repressive activities of the REST N- and C-terminal repression domains, neither of which binds to Mediator (21). Although individual depletion of MED19 and MED26 had little impact on the repressive activity of REST-(141–600), combined depletion of both Mediator subunits dramatically impaired REST-(141–600) repressor activity (Fig. 4A, left panel). By contrast, neither individual nor combined depletion of MED19/MED26 significantly influenced the repressive activities of the REST N- or C-terminal repression domains (Fig. 4A, right panel). The concordant MED19/MED26 requirement by REST-(141–600) for both Mediator binding and Mediator-dependent transcriptional repression supports the notion that MED19/MED26 are physical and functional targets of REST in Mediator.

FIGURE 4.

MED19 and MED26 are synergistically required for REST-directed transcriptional repression. A, HeLa cells were transfected with control, MED19-, MED26-, or MED19 and 26-specific siRNAs as indicated 48 h prior to co-transfection with pG5TK-Luc along with Gal4 or Gal4-REST derivatives as indicated and subsequent assay of transfected whole cell lysates for normalized luciferase activities. Luciferase activities are expressed relative to the luciferase activity obtained in cells transfected with control siRNA and Gal4. Data represent the mean ± S.E. of at least three independent transfections performed in duplicate. Asterisks denote statistically significant values relative to control siRNA (Student's t test; **, p < 0.05). Gal4-REST-N and -C-terminal derivatives express REST amino acids 1–140 and 999–1098, respectively. Immunoblot analysis of transfected whole cell lysates revealed that Gal4-REST derivatives were expressed at roughly equivalent levels (supplemental Fig. S4). B, RNA from HeLa cells transfected with control, MED19-, MED26-, or MED19 and 26-specific siRNAs as indicated was used for RT-qPCR. mRNA levels are expressed relative to mRNA levels in control siRNA-transfected cells, which was arbitrarily assigned a value of 1. Data represent the mean ± S.E. of at least three independent experiments performed in duplicate. Asterisks denote statistically significant values relative to control siRNA (Student's t test, **, p < 0.05).

To confirm the functional interaction between REST and MED19/MED26 under more physiological conditions, we monitored the impact of MED19 and/or MED26 knockdown on the expression levels of REST-repressed neuronal genes in their natural chromosomal loci. In this regard, we recently identified a Mediator requirement for REST-directed G9a-dependent repression of the SNAP25, Syn1, and M4 genes in HeLa cells (21). Therefore, we used RT-qPCR analyses to monitor the expression levels of these three REST-target genes following RNAi-mediated depletion of MED19 and/or MED26. Consistent with findings from transient repression assays, we observed that only combined, but not individual, depletion of MED19/MED26 disrupted REST repressor function, resulting in de-repression of the SNAP25, Syn1, and M4 genes in HeLa cells (Fig. 4B). Taken together, these findings support the notion that MED19 and MED26 are functionally important targets of REST in Mediator and synergistically modulate REST-directed repression in vivo.

MED19 and MED26 Are Synergistically Required for REST-directed Recruitment of Mediator, G9a, and H3K9me2 to RE1 Elements within REST-repressed Neuronal Genes—Recently, we showed that the REST/Mediator/G9a repressive network is conserved in a broad range of non-neuronal cell types, and chromatin immunoprecipitation (ChIP) analyses further confirmed specific occupancy of RE1 elements within REST-repressed neuronal genes by REST/Mediator/G9a in the absence of markers of gene activation (21). Mechanistically, we showed that Mediator recruited by RE1-bound REST facilitates the deposition of transcriptionally repressive H3K9me2 by G9a, with which Mediator directly interacts through its MED12 interface (21). Therefore, we sought to investigate whether MED19/MED26, as a probable REST interface in Mediator, is required for REST-directed recruitment of Mediator and G9a-dependent H3K9me2 to RE1 silencing elements in vivo. To address this question, we monitored the transcription factor binding and histone methylation profiles of the SNAP25, Syn1, and M4 genes in HeLa cells as a function of MED19 and/or MED26 using RNAi and quantitative chromatin immunoprecipitation (qChIP). This analysis revealed that only combined, but not individual, depletion of MED19/MED26 significantly impaired REST-directed recruitment of Mediator, as well as G9a and G9a-dependent H3K9me2 on all three REST-target genes (Fig. 5). Taken together, our results reveal MED19/MED26 to be a crucial interface in Mediator necessary to link RE1-bound REST with G9a in epigenetic silencing of neuronal gene expression.

FIGURE 5.

MED19 and MED26 are synergistically required for recruitment of Mediator and G9a-dependent H3K9me2 by RE1-bound REST. Soluble chromatin prepared from HeLa cells transfected with control, MED19-, MED26-, or MED19 and 26-specific siRNAs as indicated was subjected to IP using the indicated antibodies. Immunoprecipitated chromatin was analyzed by quantitative PCR using primers flanking RE1 elements within the M4, SNAP25, and Synapsin1 (Syn1) genes. The level of RE1 site occupancy for each protein is expressed relative to its level of occupancy in control siRNA-transfected cells, which was arbitrarily assigned a value of 100%. Data represent the mean ± S.E. of at least three independent experiments performed in triplicate. Asterisks denote statistically significant values relative to control siRNA (Student's t test, **, p < 0.05; ***, p < 0.01).

DISCUSSION

It is now well established that Mediator is a primary conduit of regulatory information conveyed by gene-specific transcription factors to RNA polymerase II. In this capacity, Mediator plays an essential function in regulating the assembly and/or activity of RNA polymerase II transcription complexes on core promoters. However, whether and how Mediator might influence transcription factor-driven chromatin modifications that impact RNA polymerase II transcription has not been clear. In this regard, we recently described a novel role for Mediator in G9a-dependent epigenetic silencing of neuronal gene expression imposed by the RE1 silencing transcription factor REST. We showed that REST, Mediator, and G9a form a trimeric complex in vivo, that G9a binds to Mediator through its MED12 interface, and that the MED12 interface in Mediator is thus essential for REST-directed recruitment of G9a and the imposition of transcriptionally repressive H3K9me2 within REST-targeted neuronal genes (21). Here, we have extended these findings through the identification MED19/MED26 as a direct interface for REST in Mediator, thus providing a more complete description of the physical and functional interactions within the REST/Mediator/G9a network required for extra-neuronal gene silencing. First, we found that although REST binds to both MED19 and MED26 in isolation, combined depletion of both subunits is required to disrupt the association of REST with Mediator. Second, we found that combined, but not individual, depletion of MED19/MED26 impairs REST-directed recruitment of Mediator and G9a to RE1 elements, leading to a reversal of G9a-dependent epigenetic marks and de-repression of REST-target gene expression. Because our results further reveal that MED19 and MED26 do not depend upon one another for incorporation into Mediator and their concomitant loss does not disrupt the apparent integrity of intact Mediator, these finding suggest that MED19/MED26 comprise a probable composite interface in Mediator for synergistic physical and functional association with REST. Combined with our previous findings (21), the data presented here establish a model in which Mediator, much like a molecular clamp, functions to strengthen the interaction between REST and G9a, thus facilitating the imposition of transcriptionally repressive H3K9me2 around RE1 elements within REST-repressed neuronal genes. In this model, REST and G9a interact with Mediator through two distinct interfaces, MED19/MED26 and MED12, respectively, and both interactions are required for recruitment of G9a to RE1-bound REST (Fig. 6).

FIGURE 6.

Schematic model for the network of functional interactions among REST, Mediator and G9a required for epigenetic silencing of neuronal gene expression. RE1-bound REST recruits Mediator through its MED19/MED26 subunits. Mediator, in turn, facilitates recruitment of G9a-dependent H3K9me2 through direct interaction of G9a with the MED12 interface in Mediator. Although REST can bind directly to G9a in vitro, it nonetheless requires Mediator for recruitment of G9a and the imposition of transcriptionally repressive H3K9me2 in vivo (21).

To our knowledge, the identification herein of MED19/MED26 as direct physical and functional targets of REST represents the first instance in which either Mediator subunit has been shown to be a direct interface for gene-specific transcription factors. In this regard, it is well documented that enhancer-bound activators can recruit Mediator through individual subunits to facilitate the assembly and/or stimulation of transcription preinitiation complexes on core promoters (50, 51). For example, members of the nuclear receptor superfamily recruit Mediator via direct interactions with discrete LXXLL motifs within its MED1 subunit (52–56), while another class of activators, including adenovirus E1A, Esx, Elk-1, and C/EBPβ instead target Mediator through its MED23 subunit (57–60). Furthermore, MEDs 12, 14, 15, 16, 17, and 25 have each been implicated as direct physical and functional targets within Mediator of various gene-specific transcriptional activators (38, 45, 61–71). Our discovery that Mediator, through its MED19/MED26 interface, is recruited by RE1-bound REST to silence neuronal genes thus provides a repressive corollary to the well-characterized mechanism of activator-dependent Mediator recruitment.

Recently, a comprehensive comparative genomics analysis of Mediator across the eukaryotic kingdom revealed that MED19 and MED26, along with other MED subunits, arose early on during eukaryotic diversification (72). It has been proposed that the evolutionary addition of these “peripheral” subunits onto an existing core 17-subunit Mediator proto-complex present 1–2 billion years ago could have accommodated unique transcriptional mechanisms that underlie the advanced genetic circuitry of multicellular organisms (72). In this regard, it is perhaps notable that whereas MED19 is broadly represented throughout eukaryotes, with identifiable orthologs present in metazoa, fungi, and plants, amoebae, and diatoms, MED26 is restricted to metazoans and amoebae (72). Within the metazoa, MED26 in fact exhibits a species distribution that closely overlaps that of REST, possibly suggesting that the interaction of REST with a MED19/26 Mediator interface arose as a consequence of co-evolution and contributed to the diversification and development of metazoans.

An unexpected finding to emerge from this study was the identification of MED26 as a physical and functional target of REST. This observation implicates MED26 directly in transcriptional repression and thus challenges a common conception that MED26 functions exclusively in transcriptional activation. A principal activating role for MED26 in transcription control has been invoked largely on the basis of its observed preferential association with an active form of Mediator biochemically isolated from mammalian cells. Thus, elegant structural and functional analyses have previously revealed that a MED26-proficient Mediator species devoid its kinase module is capable of supporting activator-dependent RNA polymerase II transcription in vitro, whereas a MED26-deficient Mediator species that additionally includes the kinase module inhibits this process (57, 73, 74). Nonetheless, how these two biochemically stable isolates relate to the possible range of dynamic Mediator complexes assembled on target gene promoters in vivo remains to be definitively established. It is possible, for example, that the structurally and functionally distinct Mediator species isolated biochemically represent static extremes across a continuum of dynamic Mediator complexes assembled in vivo. This possibility is supported by the results of recent proteomics analyses indicating that the association of MED26 and the kinase module with core Mediator is not, in fact, mutually exclusive (74, 75). This observation suggests the existence of an intermediate or “transition” state of Mediator, one in which incorporation of all possible subunits is accommodated. We speculate that the Mediator complex recruited by REST could reflect this intermediate state, thus providing REST with a Mediator platform of sufficient functional complexity to exert complex regulation of neuronal gene expression. In this regard, our identification of MED26 as a direct target of REST might help to explain the comparably poorly understood role of REST in context-dependent transcriptional activation (12, 16, 76–78).

Our studies revealed that combined depletion of MED19/MED26 impaired REST-directed, G9a-dependent imposition of transcriptionally repressive H3K9me2 to an extent that far exceeded the additive impact of individually depleting either subunit alone, suggesting that MED19/MED26 function synergistically to mediate REST-directed neuronal gene silencing. Previous studies have documented the ability of gene-specific transcriptional activators to target more than one subunit in Mediator. For example, in mammals, the glucocorticoid receptor binds to MED1 and MED14 (79), the retinoid X receptor and the retinoic acid receptor both bind to MED1 and MED25 (71), p53 binds to MED1 and MED17 (80, 81), and VP16 binds to MED17 and MED25 (38, 70, 80). In plants, LEUNIG has been shown to bind to MED14 and CDK8 (82). However, in none of these instances has functional synergy between different Mediator subunits targeted by the same activator been demonstrated. Thus, the identification herein of MED19 and MED26 as dual functional targets of REST represents the first example of functional synergy among Mediator subunits targeted by a common transcriptional regulator. Why does REST require physical interaction with both MED19 and MED26 in Mediator? One possibility is that REST is such an important developmental regulator that sufficient mechanistic redundancy within its regulatory networks must exist for REST to efficiently perform its function throughout the genome. Further studies will be required to elucidate the topological basis for functional synergy within Mediator and further clarify the fundamental logic that drives REST-dependent developmental gene regulation.

In summary, our findings strongly suggest that REST recruits Mediator through a MED19/MED26 interface in order to facilitate epigenetic silencing of neuronal genes in non-neuronal cells. This work thus identifies MED19 and MED26 as critical components of the regulatory apparatus employed by REST to restrict neuronal gene expression to the nervous system and thereby contribute to the specification of neuronal identity.