|Home | About | Journals | Submit | Contact Us | Français|
The tail module subunits of Mediator complex are targets of activators both in yeast and metazoans. Here we discuss recent evidence from studies in yeast for tail module specificity for SAGA-dependent, TATA-containing genes including highly regulated stress response genes, and for independent recruitment and function of the tail module.
The Mediator complex is a central player in the transcriptional regulation of protein-coding genes, being required for nearly all transcription by RNA polymerase II (RNAP II), and is highly conserved from yeast to human.1 Structural and biochemical studies, together with yeast genetic studies showing similar phenotypes resulting from deletion of specific non-essential subunits, indicate that the core Mediator complex comprises three distinct modules: head, middle, and tail. In addition, a fourth module that includes a cyclin-CDK pair of subunits and hence is often referred to as the kinase module, associates transiently with the core Mediator complex. Yeast Mediator comprises 25 subunits, all of which have homologs in mammals.2,3 Five additional metazoan specific subunits make mammalian Mediator a 30 subunit complex. The yeast Mediator tail module comprises Med2, Med3, Med15/Gal11 and Med16/Sin4 subunits, and microarray experiments together with biochemical data indicate that Med2, Med3, and Med15/Gal11 belong to a functionally discrete submodule.4,5 (A new nomenclature for Mediator subunits was adopted in 2004,6 so some subunits have more than one name.)
Direct interactions between gene-specific transcriptional activators and tail module subunits have been reported in several previous studies. For example, Gcn4 interacts with Med15/Gal11, Med2 and Med3 tail subunits, and Oaf1 and Gal4 interact directly with Med15,7-10 while in mammals, SREBP-1 interacts with Med15.11 Subunits of the head and middle modules, on the other hand, interact with general transcription factors and RNAP II subunits, although interactions have also been reported with gene-specific activators, particularly in mammalian systems.1,12-14 Based on these studies, the general model of Mediator complex function that emerges is that the tail module subunits, by interacting with activators, signal through the head and middle modules to the basal transcription machinery to regulate transcription initiation (Fig. 1).
Several Mediator middle and head module subunits are essential, and inactivation of essential subunits from the head module affects expression of over 90% of all RNAP II transcribed genes in yeast.9 In contrast, the tail module contains only non-essential subunits, and deletion of any individual tail module subunit affects the expression of only a fraction of the genome,4,5 indicating that the tail module has a gene selective rather than a general role in transcription. The question then arises as to which genes require the tail module for their expression. To gain new insight into the role of the tail module in genome wide transcription, we assessed the effect of both single and double tail subunit deletions.5 We found that the tail module subunits Med2, Med3 and Med15/Gal11 function redundantly at many target genes.
Yeast genes can be divided into two principal types: (i) stress regulated, TATA-containing, SAGA dependent genes (10–15%) and (ii) constitutive, TATA-less, TFIID-dependent genes (85–90%).15,16 Strikingly, our analysis of genome-wide expression changes in a med3 med15/gal11 mutant revealed that the tail module preferentially regulates genes that depend on SAGA and Swi/Snf, and not TFIID, for their activation. Many factors that regulate chromatin, TBP, and RNAP II play a greater role at SAGA dominated genes than at TFIID dominated genes.15,16 Our observation that the tail module preferentially regulates SAGA dominated genes suggests that the tail module might function along with specific chromatin regulators to help pre-initiation complex (PIC) formation on these highly regulated genes (Fig. 1).
The remodeling of nucleosomes at the promoter is an important step in transcriptional activation, and can involve multiple mechanisms. ATP-dependent chromatin remodelers such as the yeast Swi/Snf complex can increase the accessibility of the transcriptional machinery to nucleosomal DNA,17 while cofactors such as the SAGA complex can modify histone N-terminal tails, thus altering affinity of transcriptional regulatory factors to promoter chromatin.18 It is not difficult to imagine that chromatin remodeling at active gene promoters would be required for Mediator to efficiently and productively recruit RNAP II and, indeed, previous studies have established genetic interactions among Mediator, Swi/Snf, and SAGA. For example, loss of the Spt20 subunit of the SAGA complex leaves yeast cells inviable if Mediator subunits, including tail subunits Med15/Gal11 or Med16/Sin4, or the Snf2 subunit of Swi/Snf complex, are also absent.19
Studies on the interrelationships among Mediator, Swi/Snf and SAGA at individual active promoters in yeast indicate that there is no single mechanism for their recruitment.20 For example, at the Gcn4 dependent ARG1 and SNZ1 promoters, recruitment of Mediator and SAGA is interdependent, while recruitment of Mediator and SAGA to promoters controlled by the activator Met4 occurs independently.21,22 These and other differences in the requirement of Mediator for the recruitment of SAGA or Swi/Snf complex at different promoters suggest that complex interactions between specific activators and Mediator tail subunits, Swi/Snf and SAGA subunits, as well as between Mediator subunits and Swi/Snf and SAGA subunits, are required to provide transcriptional specificity at particular promoters.20 In contrast, TFIID-dependent genes, many of which may not require Swi/Snf, could depend on Mediator subunits other than those in the tail module to interact with activators or TFIID components, thus leading to the interactions with other GTFs and formation of PIC.23-25
In summary, although connections between the Mediator tail, Swi/Snf and SAGA have been established, the detailed mechanisms behind these interconnections likely differ at different promoters, and an understanding of the precise molecular basis underlying these connections remains to be achieved.
Genes that are commonly upregulated during general environmental stress are strongly biased toward being SAGA dominated.15,16 Consistent with the strong correlation that we found between SAGA dependent and tail module dependent genes, we observed significant enrichment of general stress response genes among genes affected by loss of tail module subunits.5 The tail module therefore seems to act as a global transcriptional co-activator of stress response genes. In contrast, deletion of kinase module subunits Cdk8/Srb10 and Med12/Srb8 causes upregulation of many of these same stress response genes.
Previous work has pointed to antagonistic roles for the tail and kinase modules in regulating gene expression. Genetic analysis indicates that this module is involved in the negative regulation of a subset of genes in yeast, and these genes are enriched among those showing decreased expression upon tail subunit deletion.4,9 These and our genetic analyses used yeast grown in rich medium or complete synthetic medium, conditions in which many genes are expressed at low levels or not at all. Additional experiments using other growth conditions have also revealed an antagonistic relationship between the tail and kinase modules. For example, osmotic stress genes are repressed by the Tup1/Ssn6 complex, which appears to function in part via the kinase module26,27; these same genes show diminished activation in tail subunit deletion mutants.28 Furthermore, SAGA and Swi/Snf are required for overcoming Tup1-mediated repression of osmotic stress inducible genes,27 consistent with these co-activators being preferentially involved in activation of genes that require the Mediator tail module for their expression.5 Similarly, the general stress response genes, which are activated by Msn2/Msn4,29 are upregulated upon loss of kinase module subunits,9 and require Med15/Gal11 for activation.30
Structural and biochemical studies have shown that the Med13 subunit of the kinase module binds the Mediator tail module, and this interaction blocks Mediator interaction with RNAP II.31,32 This suggests a scenario in which, under inducing conditions, the interaction of Mediator tail subunits with activators or co-activators such as SAGA and Swi/Snf complex causes transcriptional activation, whereas under non-inducing conditions, Mediator tail rather interacts with the kinase module subunit, blocking its association with RNAP II and leading to transcription repression. Consistent with this scenario, a dynamic interaction between the core Mediator complex and the kinase module is suggested by a ChIP-chip analysis showing that the kinase module is associated with the core module at all sites but with lower occupancy.33 Such a dynamic interaction could be stabilized or destabilized by gene-specific repressors and activators, thereby modulating gene expression. Alternative mechanisms for repression by the kinase module are also possible. For example, Msn2 is targeted for degradation by phosphorylation by Srb10/Cdk8,30,34 suggesting a mechanism involving activator destruction. Further experiments are needed to fully understand the mechanistic details of Mediator tail regulated activation and kinase module dependent repression of stress response genes.
The specific requirement for the tail module also includes the activation of metabolic genes. A significant percentage of genes involved in metabolic pathways such as drug metabolism and detoxification, carbohydrate metabolism and amino acid metabolism are downregulated or upregulated by Mediator tail deletion.5 Mediator complex, and particularly the tail module, has been implicated in the transcription of specific metabolic pathway genes both in yeast and metazoans. Although many metabolic pathway genes are also stress-regulated, dependence on tail subunits has been observed for activation of some such genes via non-stress pathways. For example, the Med15 subunit of the tail module has been shown to regulate the expression of genes involved in fatty acid metabolism by directly interacting with activator SREBPα through its KIX-domain, leading to the transcriptional activation of target genes in human.11 Similarly, SBP-1 in C. elegans and Oaf1 in yeast interact directly with Med15 through the conserved KIX domain to regulate fatty acid metabolism.7,11,35 The Med15 subunit has also been shown to regulate galactose and amino acid metabolism in yeast by directly interacting with activators Gal4 and Gcn4.10,36-38
The requirement of Med15 in the metabolism of xenobiotics has also been reported in yeast and C. elegans. In yeast, Med15 directly binds to Pdr1 orthologs, which regulate the expression of genes involved in drug-efflux pumps.39 Similarly, in C. elegans, genome-wide expression analysis shows that loss of the Med15 subunit causes significant downregulation of drug metabolic genes and genes required for energy homeostasis such as glucose and lipid metabolism.40 These studies suggest that the tail module might act as co-sensor in the nutrient/metabolic signaling pathway: tail subunits could directly or indirectly interact with upstream signaling components and transfer the signaling information to the promoter by either binding to activators or by interacting with chromatin regulators, which will then lead to chromatin remodeling at the promoter and PIC assembly. The tail module thus helps to provide a platform for PIC assembly. Our finding that tail subunits act redundantly is consistent with the idea, also proposed by others, that multiple contacts between tail subunits and various transcription regulators contribute to transcriptional specificity.5,8,37 Thus, environmental perturbations are perceived by tail subunits and the information is further relayed to the general transcription machinery to cause the required transcriptional outcomes.
Biochemical fractionation and purification studies suggest that the head, middle, and tail modules of Mediator are present as a single complex.1 A genome-wide localization study also shows the presence of all modules of Mediator together on promoters.33 However, several reports show that under some circumstances, the Med15/Med3/Med2 triad can be recruited and may be able to exert some functions independently of the remainder of the Mediator complex.5,8,41 Previous studies in our lab have shown that the loss of Mediator head and middle modules from several constitutive as well as inducible gene promoters in med17/srb4 ts mutant yeast leaves the tail module associated with these promoters.23,41 However, the loss of head and middle modules in this mutant resulted in greatly reduced binding of RNAP II, TBP and other GTFs to all RNAP II transcribed genes examined,23 consistent with loss of transcription of almost all RNAP II transcribed genes in this mutant.9 The tail module, however, might be able to function independently in early steps of PIC formation such as recruitment of SAGA or Swi/Snf complex and chromatin remodeling. In support of this possibility, He et al. showed that the recruitment of the Swi/Snf complex and chromatin remodeling remains unaffected at the CHA1 promoter in med17/srb4 ts mutant yeast at the non-permissive temperature, although CHA1 induction is nearly eliminated (Fig. 2A).41 However, although this could mean that the tail module was able under these circumstances to contribute to Swi/Snf recruitment and chromatin remodeling, these events could also have occurred independently of Mediator.
A related result was observed by Zhang et al., who reported that the Med15/Med3/Med2 triad is recruited as a stable subcomplex in yeast lacking the Med16/Sin4 subunit of the tail.8 In spite of the apparent lack of recruitment of the middle and head modules of Mediator, the activator Gcn4 was still able to recruit TBP and RNAP II to the ARG1 promoter, causing wild type transcription of the ARG1 gene. How can this finding be reconciled with the inability of the Med15/Med3/Med2 triad to activate transcription in med17/srb4 yeast? One possibility is that Mediator functions differently at the genes studied in these reports (Fig. 2B). Another possibility is that recruitment of the head/middle modules is reduced, but not entirely eliminated, in med16/sin4 yeast (Fig. 2C). This would be consistent with a recent study showing that loss of Med16/Sin4 causes reduced recruitment of the head module compared with the tail module at the induced FLR1 gene.42 Further work is needed to whether and how the Med15/Med3/Med2 triad can function as an independent entity in transcriptional activation.
In conclusion, the tail module of Mediator in yeast, and possibly in higher eukaryotes as well, is largely dedicated to the activation of genes involved in cellular responses to environmental signals. In yeast, these genes tend to contain TATA elements and depend on the SAGA and Swi/Snf complexes. The signals for which the tail subunit serves as a co-sensor may be caused by various stresses or by nutrient/metabolite limitations, or maybe signals involved in developmental pathways.43 One important topic for future research is to determine the relationship between recruitment and functions of Mediator, Swi/Snf and SAGA, particularly in metazoan cells. Other open questions are whether a partitioning of the role of the Mediator tail module among different categories of genes occurs in mammalian cells as it does in yeast, how Mediator is recruited to TFIID-dependent genes in yeast, and whether the tail Med15/Med3/Med2 subcomplex functions independently from the rest of Mediator. Clearly, we are a long way from an exhaustive understanding of this fundamental player in transcriptional activation, and additional interesting surprises are sure to lie ahead.
We thank Joe Wade for a critical reading of the manuscript. This work was supported by NSF grant MCB0949722.
Previously published online: www.landesbioscience.com/journals/transcription/article/19840