Structures and Compositional Dynamics of Mediator in Transcription Regulation

Abstract

The eukaryotic Mediator, comprising a large Core (cMED) and a dissociable CDK8 kinase module (CKM), functions as a critical coregulator during RNA polymerase II (RNAPII) transcription. cMED recruits RNAPII and facilitates the assembly of the pre-initiation complex (PIC) at promoters. In contrast, CKM prevents RNAPII binding to cMED while simultaneously exerting positive or negative influence on gene transcription through its kinase function. Recent structural studies on cMED and CKM have revealed their intricate architectures and subunit interactions. Here, we explore these structures, providing a comprehensive insight into Mediator (cMED-CKM) architecture and its potential mechanism in regulating RNAPII transcription. Additionally, we discuss the remaining puzzles that require further investigation to fully understand how cMED coordinates with CKM to regulate transcription in various events.

Introduction

In eukaryotes, basal transcription of RNA polymerase II (RNAPII) is a fundamental process that initiates the synthesis of mRNA from DNA templates, laying the foundation for gene expression and cellular function[1]. This process entails the assembly of general transcription factors (GTFs) and RNAPII at promoters, forming the pre-initiation complex (PIC). Notably, Mediator plays a crucial role in functionally and structurally bridging transcription activators to RNAPII, thereby conveying regulatory signals from enhancers to RNAPII machinery to activate transcription[2,3].

The complete Mediator complex, consisting of a large Core (cMED) and a dissociable CDK8 kinase module (CKM), is an evolutionally conserved multi-subunit transcriptional coregulator. The cMED, composed of 21 subunits in yeast and 26 subunits in humans, is structurally organized into Head, Middle, and Tail modules. The CKM, which contains 4 subunits, can reversibly associate with the cMED (Figure 1)[4–6]. The Mediator complex is primarily involved in regulation of the PIC assembly at promoters during transcription initiation[4,7,8]. Functionally, cMED stimulates basal transcription by recruiting RNAPII to promoters, promoting PIC assembly, and stimulating phosphorylation of the RNAPII C-terminal domain (CTD)[3,4,7]; On the other hand, the dissociable CKM, comprising MED12, MED13, Cyclin C (CycC), and CDK8, also plays a significant role in transcription initiation[9]. In vertebrates, three sequence-distinct paralogous subunit pairs within the CKM have emerged, including MED12/12L, MED13/13L, and CDK8/CDK19 (also known as CDK8-like)[10,11]; together with CycC, the paralogous subunits assemble into the CKM in a mutually exclusive manner[12]. Given that current structural studies of CKM primarily focus on CDK8 rather than CDK19 [13–15], hereafter CKM will be referred to as the CDK8 kinase module. CKM can reversibly interact with cMED, hindering the binding of RNAPII to cMED, thereby repressively inhibiting the formation of the cMED-PIC complex[13,16]. Recent functional studies, however, have provided definitive evidence that the mammalian CKM also plays a positive role in gene expression by participating in the regulation of RNAPII during post-initiation events[17–19]. CDK8 phosphorylates specific pausing-related factors, and inhibition of its kinase activity results in increased RNAPII pausing[17,20,21]. Additionally, CDK8 within the CKM serves as a crucial regulator of genes particularly in human cancer cells, promoting activator-driven transcription elongation under various conditions, such as the p53 transcriptional network, the serum response network, hypoxia signaling pathway, and IFN-γ pathway[17,18,22,23].

Figure 1. Subunit composition of the eukaryotic Mediator complex.

(a) and (b) Schematic representations of the yeast and human Mediator complexes with modular composition[4]. The Mediator complex comprises a CKM and a large cMED, which includes Head, Middle and Tail modules. Yeast Med5, Med3 and Med2 (a) are orthologous to MED24, MED27 and MED29 in Metazoans (b). Metazoan-specific subunits are indicated by asterisks in (b).

Given the critical roles of the Mediator complex in transcription and its dynamic compositions, we have reviewed recent structural studies of both cMED and CKM, along with early investigations into the Mediator complex (cMED-CKM), providing valuable insights into the mechanism by which CKM regulates cMED-dependent transcription.

1. Structures of the cMED and CKM

Although the Mediator has been studied for over 30 years, a comprehensive structural understanding of the entire cMED-CKM complex remains elusive. Nonetheless, recent advances have provided detailed structural information on both cMED and CKM. Furthermore, the determination of structures of cMED in complex with PIC or PIC-nucleosome has significantly enriched our understanding of cMED-dependent transcriptional processes [14,24–29].

1.1. Conserved architecture of the cMED across different species

After extensive efforts from multiple laboratories over the past decade, the near-atomic resolution structures of cMED from various species, including yeast, human, mouse, and C. thermophilum, have been reported. These structures collectively enhance our comprehension of the subunit organization and interactions of cMED[2,25,27–32]. The overall architecture of cMED is conserved across eukaryotic organisms, particularly in the Head and Middle modules[33,34]. Furthermore, the mammalian cMED incorporates four additional structurally metazoan-specific subunits (MED23, MED25, MED28 and MED30) connecting Head and Tail modules, along with another metazoan-specific MED26 subunit associating with the Middle module (Figure 1). This MED26 subunit serves as a docking site facilitating the transition of RNAPII into productive elongation[35].

While the division of the cMED into three modules is widely accepted (Figure 1), the assignment of several subunits remains contentious due to their structural presence across different modules. Here, we primarily review the most well-defined and complete structure of human cMED based on recent reported studies[27–29]. The clearly visualized human cMED structure has facilitated accurate subunit assignment (Figure 2a)[28]. The three modules of cMED can be distinctly subdivided into detailed components. The Head module, previously divided into three parts (Neck, Fixed jaw, and Movable jaw) in yeast S. cerevisiae[36], can now be further divided into Shoulder, Arm, Spine, Joint, Movable jaw, and Fixed jaw (Figure 2b–c). The Fixed jaw can also be subdivided into Finger, Tooth and Nose, which connects the Head module with the Middle module within cMED[27]. The Middle module has been further categorized into five parts, including Hook, Knob, Connector, Beam and Plank (Figure 2b–c). Between Hook and Connector is a conserved Hinge region formed by MED7/MED21 subcomplex, which is required for cMED-RNAPII interaction[37,38]. The Tail module is subdivided into proximal and distal parts (Figure 2b–c)[27]. In this configuration, MED14 acts as a backbone linking the three modules and stabilizing the complex. MED17 also serves as a scaffold subunit, interacting with 15 subunits within the Head and Middle modules[28,29]. Certain subunits, including MED27, MED28 and MED30, establish connections between the proximal Tail and the Head modules, enhancing the structural integrity of cMED (Figure 2a–c).

Figure 2. Overall structure of the human cMED.

(a) Schematic subunit organization of the human cMED and CKM. The Head, Middle, and Tail modules of the cMED and the dissociable CKM are indicated with corresponding colors. (b-c) The structure of human cMED (PDB code: 7EMF)[28] is shown in surface and cartoon representations.

1.2. The structure of the dissociable CKM

Early structural studies have provided initial insights into the overall architecture and subunit organization of human and S. cerevisiae CKM[13,39]. However, incorrect assignments of subunit within the CKM were made for Med13 and Med12. The accurate subunit organization was obtained through our recently determined cryo-EM structure of the S. cerevisiae CKM[14]. The overall architecture of CKM adopts an extended conformation, which is divided into three lobes: Kinase-, Central- and H-lobes (Figure 3a). The Kinase- and H-lobes extend from the Central-lobe. Within the Kinase-lobe, Cdk8 and CycC form a tightly-bound CDK/Cyclin-pair, anchored by the N-terminal portion of Med12 (Med12N) (Figure 3b). The Central-lobe comprises Med13 along with the C-terminal region and HEAT-1 of Med12, while the H-lobe is formed solely by the HEAT 2–5 domains of Med12 (Figure 3a and c)[14].

Figure 3. Structure of the yeast CKM.

(a) Overall structure of the CKM (PDB code: 7KPX)[14], color-coded by subunits. Each lobe of CKM is indicated. (b) The Kinase-lobe of CKM, indicating an active conformation of the Cdk8 T-loop (red) bound by the N-terminal region of Med12 (Med12N). The RHYT segment of Cdk8 is colored in magenta. The substrate binding site is indicated by a dashed black oval. (c) Structural organization of Med12. Domains are colored and indicated, respectively. (d) Structural comparison of Med13 and the guide RNA-bound human Ago2 (hAgo2) (PDB code: 4W5N)[42]. Each structure is color-coded according to its functional domains. The IDR missing in the structure of Med13 is indicated by a dashed line. The guide RNA in hAgo2 is colored in red.

Cdk8 and its paralog CDK19 in vertebrates are unique among all CDK family proteins in lacking a phosphorylation residue within the T-loop. Unlike classical CDK proteins, which become active upon cyclin binding, CDK8 requires CycC and MED12 to attain its kinase activity both in yeast and human, suggesting a non-canonical CDK8 activation mechanism through MED12[14,40]. The structure of yeast CKM provides molecular understanding of Med12-dependent Cdk8 activation (Figure 3b). The Kinase lobe adopts a typical CDK/Cyclin architecture, with the additional Med12N region in contact with Cdk8 and CycC. CycC, featuring two cyclin-box fold domains (N- and C-CBFs), binds to the N-lobe of Cdk8 through its N-terminal CBF (Figure 3b). The N-terminal segment (residue 35–56) of Med12, forming an H1 helix followed by a coil, establishes extensive contacts with the T-loop and the RHYT segment, thereby stabilizing the T-loop (Figure 3b)[14,41]. Notably, compared to the reported crystal structures of human CDK8/Cyclin C, where their T-loops are either inactive or partially disordered, the T-loop of yeast Cdk8 bound by Med12 within the CKM exhibits a well-organized active conformation suitable for substrate recognition and phosphorylation[41]. This underlines the significance of Med12 in Cdk8 activation.

An intriguing structural feature within CKM involves Med13, one of the largest subunits of the entire Mediator complex (Figure 3a). Despite exhibiting low sequence homology to classical Argonaute (Ago) proteins, the Med13 possesses an Ago-like architecture, comprising four globular domains (N, PAZ, MID and PIWI) and two linker domains (L1 and L2) (Figure 3d)[42]. Notably, the PIWI domain lacks catalytic residues, suggesting that Med13 may not have nuclease function. Further investigation is required to ascertain whether Med13 exhibits nucleic acid-binding capability. Another significant difference distinguishing Med13 from Ago proteins is the presence of a large intrinsically disordered region (IDR) inserted between the PAZ and L2 domain (Figure 3d)[14]. Although the Med13 IDR (~500 amino acids) is absent in the current CKM structure, a comprehensive understanding of its role and the rationale behind the Ago-like architecture of Med13 necessitates further investigations through structural and functional studies.

1.3. The architecture of the entire Mediator complex (cMED-CKM)

To date, although the high-resolution structure of the entire Mediator complex remains lacking, several laboratories have reported low-resolution electron microscopy (EM) reconstructions of human and yeast cMED bound by CKM[13,39,43]. In the human Mediator complex, the CKM adopts a relatively fixed orientation to bind to cMED[39]. A subsequent study on the S. cerevisiae Mediator complex revealed distinct binding modes of CKM to cMed. Notably, around 60% of the S. cerevisiae cMed-CKM particles displayed an extended contact surface between CKM and cMed[13]. Both structural studies indicated that the Hook domain of cMED is bound by CKM (Figure 4a). Several biochemical and functional studies have demonstrated the crucial role of MED13 in mediating the CKM interaction with cMED[13,39,44,45]. However, the detailed molecular interactions between CKM and cMED remain elusive. Therefore, obtaining a comprehensive structural understanding of the entire Mediator complex is essential. This will enable us to elucidate the intricate interactions between CKM and cMED and further decipher how CKM inhibits the binding of RNAPII to cMED.

Figure 4. Potential mechanisms by which CKM inhibits the interaction of RNAPII with cMED and the formation of cMED-PIC.

(a) Schematic diagram showing the distinct binding modes of yeast CKM (green) to cMED (gray). The positions of CKM subunits and the Hook domain of cMED are indicated. The CKM shown with dashed lines indicates a proposed binding position based on XL-MS analysis[13,39,44]. (b) Schematic diagram of the formation of cMED-PIC at a promoter. (c) CKM binding may impede RNAPII interaction with cMED via (1) steric hindrance, overlapping with the main body of RNAPII, (2) inducing a conformational change in cMED unfavorable for RNAPII binding, or (3) employing another mechanism.

2. Dynamics of the Mediator complex during transcription

2.1. Dynamic interaction between CKM and cMED affects the initiation of transcription

The Mediator complex exists distinct compositions in cells, including cMED-CKM, cMED, and CKM, with dynamic association or dissociation between CKM and cMED[15,43]. CKM has been reported to function at promoter or enhancer regions. In yeast, it primarily presents at the upstream activation sequences (UAS) rather than gene promoters[46]. In humans, it binds to promoter regions alongside with cMED[47], while also showing enhancer occupancy in specific cell types[48]. In yeast, the cMed-CKM complex can be recruited by transcription activators to bind at UAS[49]. In this regard, the CKM prevents binding of RNAPII to cMed, which is supported by both low-resolution EM and cross-linking mass spectrometry (XL-MS) analysis, indicating that RNAPII and CKM have a shared binding region on cMED, thereby obstructing the RNAPII binding (Figure 4a) [13,44]. Subsequently, the CKM dissociates from the cMed. As a result, cMed can physically bridge activators at UAS with RNAPII at core promoters[46,49], facilitating the assembly of RNAPII with GTFs into a PIC[50]. This process results in the formation of a larger complex known as the cMED-PIC complex, consisting of cMED, GTFs and RNAPII (Figure 4b).

2.2. CKM functions positively during post-initiation processes

CKM not only inhibits the interaction between RNAPII and cMED, leading to the repression of transcription, but also exerts a positive influence on transcription under specific contexts[9]. For instance, cMED-CKM was indicated to associate with the positive transcription elongation factor b (P-TEFb), which facilitates its recruitment to the promoter regions of serum response genes[18]. Additionally, CKM regulates transcription through its CDK8 kinase, which phosphorylates multiple transcriptional regulators involved in RNAPII pausing or elongation[20]. Furthermore, CKM, particularly MED12, contributes to mediating transcriptional activation through its association with ncRNA-a, thereby enhancing chromatin architecture[51]. Collectively, the intrinsic kinase activity and unique structure of CKM provide a more intricate mechanism for Mediator to regulate transcription during different transcription steps.

Conclusion and perspectives

Recent structural studies of cMED and its complexes with PIC have yielded invaluable insights into the complicated molecular mechanisms governing cMED-dependent transcription(Figure 4b)[24,27–29,52,53]. However, significant gaps persist in our understanding of how the association between cMED and CKM occurs and how CKM modulates the functionality of cMED. The mechanism by which CKM prevents RNAPII binding to cMED remains particularly intriguing. It is plausible that steric hindrance, resulting from the overlap of main body of CKM with RNAPII, contributes to this inhibition. Yet, it’s equally plausible that CKM binding induces conformational changes or another barrier in cMED that is unfavorable for RNAPII interaction (Figure 4c). Further structural investigations into the cMED-CKM complex are poised to elucidate how CKM precisely regulates cMED-dependent transcription and influences gene expression via its CDK8 function. Moreover, the regulatory role of CKM extends beyond initiation to postinitiation processes. Understanding how CKM governs RNAPII pausing and release, and whether it does so in a cMED-dependent or -independent manner, remains a challenge. Furthermore, the Ago-like architecture of Med13 suggests its potential involvement in nucleic acid interactions. Exploring the function of Ago-like structure of Med13 and its large IDR may unveil its participation in specific cellular processes, warranting further investigation.

Data availability

No data was used for the research described in the article.