The Mediator Complex and Transcription Elongation

Abstract

Background

Mediator is an evolutionarily conserved multisubunit RNA polymerase II (Pol II) coregulatory complex. Although Mediator was initially found to play a critical role in regulation of the initiation of Pol II transcription, recent studies have brought to light an expanded role for Mediator at post-initiation stages of transcription.

Scope of review

We provide a brief description of the structure of Mediator and its function in the regulation of Pol II transcription initiation, and we summarize recent findings implicating Mediator in the regulation of various stages of Pol II transcription elongation.

Major conclusions

Emerging evidence is revealing new roles for Mediator in nearly all stages of Pol II transcription, including initiation, promoter escape, elongation, pre-mRNA processing, and termination.

General significance

Mediator plays a central role in the regulation of gene expression by impacting nearly all stages of mRNA synthesis.

1. Introduction

Eukaryotic mRNA synthesis is an elaborate process that proceeds with the synthesis and co-transcriptional processing of a pre-mRNA to form a mature, translatable mRNA. Transcription of pre-mRNA is catalyzed by the evolutionarily conserved, multisubunit enzyme RNA polymerase II (Pol II). Transcription takes place through discrete stages referred to as initiation, elongation, and termination, all of which are now recognized as sites for the regulation of gene expression.

Initiation of transcription by Pol II is a complex biochemical reaction governed by the concerted action of a remarkably large collection of general and gene-specific transcription factors. Fundamental to Pol II initiation are the set of five evolutionarily conserved general transcription factors, which are referred to as TFIIB, TFIID, TFIIE, TFIIF, and TFIIH and which comprise the minimum set of transcription factors needed to support synthesis by Pol II of a basal level of accurately initiated transcripts from its promoters in vitro [21,23,105]. Mechanistic studies have revealed that Pol II assembles together with the general factors at its promoters to form a stable preinitiation complex that is competent to initiate transcription when provided with ribonucleoside triphosphates [22,23,105].

Efforts to understand how transcription initiation by Pol II and the general factors is regulated by the myriad DNA binding transcription factors known to activate or repress pre-mRNA synthesis led to discovery of the Mediator, an enormous multisubunit complex that appears to be present exclusively in eukaryotes and to play an integral role in gene regulation [14,15,109]. The Mediator was first identified in S. cerevisiae transcription extracts and purified chromatographically by Kornberg and coworkers by its ability to support the activation of Pol II transcription by DNA binding transcription factors, in an enzyme system reconstituted with purified general factors [57,93]. Subsequent investigation of the mechanism of action of Mediator from yeast and higher organisms revealed that Mediator promotes activation of Pol II transcription via direct interactions with both DNA binding transcription factors and the Pol II preinitiation complex [5,78,79,93,121]. Further studies have identified an array of Mediator surfaces capable of binding specifically to the transcription activation domains (TADs) of a large number of DNA binding transcription factors [5,13] and to Pol II and several of the general factors [8,9,17,34,48,65]. Through these interactions, Mediator is capable of promoting transcription initiation by Pol II, at least in part by facilitating assembly of functional preinitation complexes [4,18,51,52,130,134].

More recently, experimental evidence implicating Mediator in post-initiation stages of Pol II transcription has emerged. These studies have brought to light roles for Mediator (i) in bypassing or overcoming the activities of factors that negatively regulate elongation [20,50,76], (ii) in recruiting Pol II transcription elongation factors and pre-mRNA processing factors [28,47,90,123] and (iii) in controlling phosphorylation of the heptapeptide repeats in the Pol II C-terminal domain (CTD) [11,28,49,123]. The phosphorylated CTD of elongating Pol II has been shown to function in some steps of chromatin remodeling by acting as a scaffold that recruits histone modifying enzymes to the transcribed region of genes [30,41,68,135]. The phosphorylated CTD also plays a central role in coordinating pre-mRNA processing by recruiting many of the enzymes and proteins critical for proper capping, splicing, and polyadenylation of pre-mRNA, as well as for proper nuclear export and localization of mature mRNAs [16,99,138]. Thus, Mediator might participate indirectly in all of these processes by modulating CTD phosphorylation.

In this review, we begin with brief descriptions of the structure of Mediator, its various functional forms, and their roles in the regulation of transcription initiation by Pol II. We devote the remainder of this review to a discussion of recent evidence supporting roles for Mediator in post-initiation stages of Pol II transcription.

2. Mediator Structure and Function

2.1. Multiple forms of Mediator

Early attempts to purify Mediator and establish its subunit composition led to the discovery that it can be isolated from cell extracts in multiple, chromatographically distinguishable forms [24,93,121]. Of these forms, the least complex is referred to as the Mediator “core” complex and is composed of more than 20 distinct proteins. Based on biochemical and structural studies, the subunits of the Mediator core complex are organized into at least three modules referred to as the “head,” “middle,” and “tail.” In cells, one fraction of the Mediator core complex is associated with a kinase module, which in S. cerevisiae includes the cyclin-dependent kinase CDK8, Cyclin C, and two additional subunits designated MED12 and MED13. Notably, the mammalian kinase module may have evolved a more complex array of functions, since it is composed of Cyclin C and one of two cyclin-dependent kinases CDK8 or CDK19, one of two MED12-like proteins MED12 or MED12L, and one of two MED13-like proteins MED13 or MED13L. Another fraction of the Mediator core complex is free of kinase module, but instead is tightly associated with Pol II in what is sometimes called the Mediator “holoenzyme” complex. The majority of metazoan Mediator core and holoenzyme complexes include an additional subunit, MED26. Finally, a small population of Mediator that contains both kinase module and MED26 can be isolated from mammalian cell extracts [27,109,123].

2.2. Mediator function in transcription activation and repression

A major question prompted by the discovery of multiple forms of Mediator was whether the different forms possessed different transcription activities. Initially, it was proposed that Mediator associated with the kinase module might participate exclusively in transcriptional repression. In yeast, results of genetic experiments supported a role for subunits of the kinase module in repression of Pol II transcription [42,45,108,117,129]. In human cells, evidence suggested that Mediator containing the kinase module contributes to repression of Pol II transcription by the DNA binding transcription factor C/EBPβ, whereas Mediator lacking the kinase module but containing MED26 contributes to activation by the same transcription factor [87]. In a related study, the activation domain of the viral transactivator VP16 was reported to activate Pol II transcription in cells through a form of Mediator that is deficient in kinase module [126].

Consistent with these genetic results, core Mediator or MED26-containing Mediator complexes have been shown to support activation of transcription in vitro far better than Mediator containing the kinase module [1,62,77,92,118,122]. Indeed, activation of Pol II transcription facilitated by the Mediator core complex can be inhibited by addition of purified kinase module to reactions [59]. Evidence suggests that it does so by binding stably to the Mediator core complex and blocking its interaction with Pol II, either by occluding the Pol II binding site or by an allosteric mechanism [33,59].

Though evidence from these early studies argued that Mediator complexes that do or do not include the kinase module have intrinsically different activities, a litany of new findings has revealed that the story is not so simple. CDK8 kinase activity has recently been shown to be required for activation of transcription of a variety of genes, including a reporter gene activated by the model transcriptional activator GAL4-VP16 [37], genes regulated by the thryoid hormone receptor [6], p53 [29], and Smads [2], and genes in the serum response network [28]. CDK8 has also been shown to participate indirectly in activation of Pol II transcription by β-catenin, by phosphorylating and inactivating the transcription factor E2F1, which antagonizes β-catenin-dependent gene activation [88]. In addition, the kinase module subunits MED12 and MED13 have been shown to interact with or be required for transcription activation by a variety of DNA binding transcription factors, including β-catenin [56], its Drosophila ortholog Pygopus [19], Nanog [125], members of the GATA and RUNX families [40] and yeast Pdr3p [112]. Notably in human cells, direct interactions between the β-catenin activation domain and MED12 have been shown to contribute to Mediator recruitment to genes [56]. Finally, although kinase module and MED26 are often thought to have opposing functions, Boyer and coworkers obtained evidence consistent with the idea that Mediator containing both MED26 and kinase module functions in the repression of transcription of some neuronal genes by recruiting the RE1 silencing transcription factor [27].

3. A role for Mediator in Pol II transcription elongation

3.1. Mechanisms governing Pol II transcription elongation

Although the initiation stage of Pol II transcription was long regarded as the primary site for the regulation of pre-mRNA synthesis, over the past decade or so it has become clear that the elongation stage of Pol II transcription is also a major site for gene regulation [53,64,for reviews see 67,110,111]. Early evidence that Pol II transcription could be regulated during elongation came with the identification of a handful of genes whose activation was accomplished at least in part by release of properly initiated, but paused, Pol II into productive elongation. The first examples of genes regulated in this fashion included the Drosophila heat shock, human c-Myc, and HIV-1 early genes [7,39,54,63,75,83,95,96,106,119,124,132]. With the development of genomic-scale methods for chromatin immunoprecipitation (ChIP-chip or ChIP-seq), nascent transcript sequencing, and location of transcriptionally engaged Pol II using nuclear run-on assays (GRO-seq), it has become clear that initiated and promoter-proximally paused Pol II can be found 30–50 nucleotides downstream of the start sites of many active and inactive genes, arguing that transcription of a large fraction of genes can be regulated during elongation [25,43,58,91,94,102,103,140].

Although the exact mechanisms underlying the regulated pausing and release of Pol II are not known, features of these processes have been gleaned from a combination of biochemical and genetic experiments, which suggest that Pol II pausing during early elongation is controlled by multiple transcription elongation factors that either negatively or positively influence Pol II. These studies identified DRB-sensitivity inducing factor (DSIF, composed of SPT4 and SPT5) and negative elongation factor (NELF) as two factors that function together to induce Pol II pausing [38,66,86,103,120,127,133,137]. In addition, results of more recent studies identified an additional factor, Gdown1, as a protein that may cooperate with DSIF and NELF to induce Pol II pausing [20]. The role of Gdown1 in Pol II pausing remains controversial, however, as results of another study led to the proposal that it may instead act upstream of initiation, by preventing formation of an initiation-competent preinitiation complex [50].

A variety of evidence argues that release of Pol II from promoter-proximal pausing requires the cyclin-dependent kinase P-TEFb, which is comprised of CDK9 and Cyclin T [72,81,82,98,101,131]. Among targets of the P-TEFb kinase are the Pol II CTD and the SPT5 subunit of DSIF [82,136]. Although the detailed mechanism by which P-TEFb contributes to the release of Pol II from pausing has not been established, evidence suggests that P-TEFb-dependent phosphorylation events might promote release of Pol II from pausing at least in part by facilitating dissociation of NELF from paused Pol II and by converting DSIF from a negatively acting factor into one that can stimulate the rate of Pol II elongation [36,100,103,128,136,137].

Work carried out recently in several laboratories led to the discovery that P-TEFb functions in release of Pol II from pausing as a subunit of larger, multisubunit complexes referred to as Super Elongation Complexes (SECs), which are created combinatorially from P-TEFb, one of three ELL family members, one of two EAF family members, one of two AF4 family members (AFF1 and AFF4), and either ENL or AF9 [10,44,69,70,116,123,139]. ELL and EAF family members form ELL•EAF heterodimers that interact directly with transcribing Pol II and potently activate the rate of transcript elongation in vitro [60,85,113,114]; ELL and EAF family members are also components of the so-called Little Elongation Complexes (LECs) [115,123], which have been implicated in snRNA gene regulation [115]. Biochemical functions for SEC subunits other than P-TEFb and the ELL and EAF family members have not been clearly defined. The genes encoding ELL and other SEC components, including AFF1, AFF4, ENL, and AF9, have been found as translocation partners of the trithorax-like MLL gene in various leukemias . The resulting MLL fusion proteins can assemble into SECs, and it has been proposed that aberrant Pol II transcription activated by the resulting mutant SECs contributes to the development of leukemias [70,89,139].

3.2. Evidence for post-initiation activities of Mediator

Among the first hints that Mediator might have a role(s) in post-initiation stages of Pol II transcription and perhaps in release of promoter-proximally paused Pol II into productive elongation came from the observation that, although initiated and paused Pol II is consitutively present on Drosophila heat shock genes, Mediator is recruited to these genes only upon heat shock, coincident with release of paused Pol II [97]. Based on these findings, it was proposed that Mediator might function together with elongation factors such as P-TEFb to promote release of paused Pol II [97]. In a subsequent study, deletion of Mediator subunit Med23 from mouse ES cells was found to interfere with activation of expression of the serum-response gene Egr1 and to result both in loss of Mediator recruitment to the Egr1 promoter and in a failure to release paused Pol II [5,130]. Furthermore, core Mediator subunits as well as components of the kinase module have been detected by ChIP not only at promoters but also within the transcribed regions of genes in both yeast and human cells [3,28,29,123,142]. Although the functional significance of these observations has not been firmly established, they are consistent with a role for Mediator in Pol II elongation. Biochemical and molecular genetic studies have suggested multiple mechanisms by which Mediator could contribute to regulation of elongation.

Several lines of evidence suggest that one way Mediator might contribute to elongation control could be to help to bypass or overcome the activities of factors that negatively regulate elongation. Efforts to define biochemical activities that render transcription in vitro more dependent on Mediator led to the observation that purified Mediator complexes could overcome a block to transcription imposed by DSIF, in a reconstituted transcription system that is apparently devoid of P-TEFb [76]. If the results of these experiments faithfully recapitulate an in vivo activity of Mediator, these observations raise the possibility that, at least under some circumstances, Mediator may function independently of, or in parallel with, P-TEFb to allow productive elongation by Pol II. As noted above, the Pol II-associated protein Gdown1 has been proposed to function with DSIF and NELF to induce Pol II pausing [20]. It is intriguing that Mediator has also been shown to be able to overcome an inhibitory activity exerted by Gdown1 during transcription in vitro [46,50], leading to the suggestion that one mechanism by which Mediator might contribute to release of paused Pol II is by helping to overcome a Gdown1-mediated block to elongation [20,35].

3.3. Role for Mediator in recruitment of P-TEFb and the SEC

Recent work from several labs has implicated Mediator in recruitment of Pol II elongation factors to genes. An unexpected twist in this story is evidence suggesting that Mediator can recruit P-TEFb and/or the P-TEFb and ELL•EAF-containing SEC by at least two apparently different mechanisms, one that depends on direct interactions between SEC and Mediator subunit MED26, and a second that depends on the kinase module.

Initial hints that MED26 might afford a physical link between Mediator and the SEC came from proteomic studies revealing that small amounts of SEC subunits consistently copurified with FLAG-MED26 but not with other FLAG-tagged Mediator subunits [123]. Results of an independent high-throughput proteomic study of endogenous protein-protein complexes also suggested that MED26 might link Mediator and SEC [80]. Further investigation identified a MED26 N-terminal domain (NTD) that binds directly to the SEC via its EAF subunit. As a consequence of this interaction, Mediator supports activator-dependent recruitment of ELL•EAF-containing complexes, including SEC, to promoter DNA in vitro [123]. Notably, a domain closely resembling the MED26 NTD has been found in other proteins including TFIIS, Elongin A, and IWS1, which have roles in the regulation of Pol II elongation [12,26,107]. Several lines of evidence support a role in cells for the MED26 NTD in Mediator-dependent recruitment of the SEC and release of paused Pol II [123]. First, MED26 depletion reduces transcription of a collection of genes known to be regulated at elongation by promoter-proximal Pol II pausing; MED26 NTD mutants that do not bind to the SEC fail to rescue the transcription defect of a subset of these genes, including c-MYC. Second, ChIP experiments revealed that MED26 depletion results in loss of recruitment of the SEC to the c-MYC and HSP70 genes and in concomitant loss of Pol II CTD phosphorylation. Intriguingly, the MED26 NTD binds not only to SEC but also to the general transcription factor TFIID, leading to the suggestion that sequential interactions of this domain with TFIID and SEC could contribute to the transition of Pol II from initiation to productive elongation.

Evidence supporting a role for Mediator containing the kinase module in recruitment of P-TEFb to genes was first brought to light in an analysis of the role of CDK8 in activating genes in the serum response network. CDK8 depletion leads to a marked reduction in expression of many serum response genes, including the immediate early genes Fos, Egr1, Egr2, and Egr3 [28]. Results of ChIP experiments showed that CDK8 depletion had little or no effect on the appearance of promoter-proximally paused Pol II at these genes, but strongly reduced both P-TEFb recruitment and Pol II CTD phosphorylation [28]. Similarly, an independent study investigating the role of the kinase module in activation of transcription of the human DIO1 gene by the thryoid receptor found that CKD8 depletion resulted in decreased P-TEFb recruitment [6]. Consistent with these observations, a combination of biochemical and proteomic analyses have provided evidence for an interaction between P-TEFb and the kinase module [28,31].

The relationship, if any, between the mechanisms by which MED26- and kinase module-containing Mediator complexes participate in the regulation of Pol II transcription elongation remains to be determined. Although the contribution of these Mediator components to elongation factor recruitment has not yet been rigorously compared in the same cell types or under the same conditions, it is noteworthy that there is relatively little overlap between the genes found to be most affected by manipulating the expression of MED26 or subunits of the kinase module [28,29,123]. Thus, it is possible that MED26 and the kinase module act via independent mechanisms at different genes and/or in response to different signals (Figure 1, upper panel). On the other hand, it is conceivable that, in some cases, they might collaborate with one another. While the majority of MED26 and kinase module seem to be associated with different populations of Mediator, evidence suggests that a small fraction of Mediator complexes includes both [27,109]. This form of Mediator might be functional and support the simultaneous action of MED26 and kinase module; alternatively, it might represent an intermediate formed during an exchange of MED26 and kinase module. Notably, MED26 depletion has been shown to affect steady state expression of many genes [123], while CDK8 has thus far been shown to play a particularly important role in elongation factor recruitment during rapid activation of genes in response to mitogenic signals or thyroid hormone [6,28]. Biochemical experiments addressing the role of CDK8 in activation of Pol II transcription by the thyroid receptor revealed that the kinase module is released upon one cycle of initiation in vitro, whereas core Mediator subunits remain associated with the promoter [6]. Thus, it is tempting to speculate that kinase module might be especially important during the initial stages of activation of a gene, while MED26 might make a greater contribution at steady state (Figure 1, lower panel).

Figure 1.

Models for Recruitment of Elongation Factors by Mediator. Upper panel, Distinct forms of Mediator containing either MED26 or kinase module act via independent mechanisms at different genes and/or in response to different signals. Lower panel, A single Mediator complex might undergo interconversion between kinase module- and MED26-containing forms at a single gene. For example, kinase module might be especially important for elongation factor recruitment during the initial round(s) of transcription when a gene is first activated, while MED26 might make a greater contribution when the gene is transcribed at steady state.

4. Perspectives

Since its discovery more than 20 years ago, the Mediator of Pol II transcription has been recognized as a central player in eukaryotic gene regulation. Although best known for its functions in Pol II transcription initiation, it has now become clear that Mediator makes significant contributions to the regulation of post-initiation events. Indeed, as we have discussed in this brief review, Mediator can affect the elongation stage of transcription through multiple mechanisms, including by helping to overcome the negative effect of factors that decrease Pol II elongation activity and by acting directly to help recruit and influence the function of positively acting elongation factors. Despite much progress, the molecular mechanisms by which Mediator functions remain enigmatic. A few of the many remaining questions are listed below.

• What is the relationship between MED26- and kinase module-dependent mechanisms of elongation factor recruitment?

• How can one reconcile the observation that the kinase module blocks Mediator interaction with Pol II yet can positively regulate transcription elongation by Pol II? Does kinase module act exclusively in this process as a component of the larger Mediator complex? It is perhaps noteworthy that P-TEFb has been reported to cofractionate with free kinase module [28], raising the possibility that the kinase module could act in some cases by a mechanism that is independent of Mediator.

• To what extent does Mediator help to link the transcription apparatus to the RNA processing machinery? Recruitment of many of the enzymes and proteins crucial for capping, splicing, and polyadenylation of nascent transcripts is regulated by phosphorylation of the Pol II CTD and can be modulated by the rate of transcript elongation by Pol II [32,141]. Hence, it is likely that Mediator will influence all of these processes, at least indirectly, through its ability to modulate CTD phosphorylation. In addition, Mediator subunit MED23 has recently been shown to bind directly to the splicing regulator hnRNP L and as a consequence to regulate a subset of hnRNP L-dependent alternative splicing events [47]. Whether Mediator also acts directly with additional regulators of splicing and other processing events to control RNA processing more generally remains to be determined.

• Does Mediator contribute to transcription termination? Yeast Mediator subunit Med18p has been shown to be highly enriched at the 3′ ends of the INO1 and CHA1 genes, and depletion of Med18p leads to an increase in Pol II readthrough past the termination signal at their 3′-ends [90]. In the future, it will be of considerable interest to determine whether this defect reflects a direct effect of Mediator on transcript termination. Alternatively, it could be secondary to Mediator-dependent Pol II CTD phosphorylation, since as noted above CTD phosphorylation influences the cleavage and polyadenylation of nascent transcripts, a process that is linked to termination [104].

• In addition to evidence that Mediator can influence chromatin structure indirectly by modulating Pol II CTD phosphorylation, evidence for a more direct role of Mediator in chromatin biology has come from studies demonstrating direct interactions between Mediator and nucleosomes and/or chromatin modifying and remodeling enzymes [55,61,71,73,74,84]. Thus, it will be of considerable interest to investigate the extent to which Mediator-dependent regulation of chromatin structure contributes to control of transcription elongation.

In the coming years, we expect that biochemical and genetic studies addressing these issues will not only provide new insights into the mechanism of Mediator action in post-initiation stages of Pol II transcription, but will also unearth many new and tantalizing questions for future research.