The Mediator complex: co-transcriptional regulation and beyond

ABSTRACT

The Mediator complex functions as a central multi-subunit coactivator essential for RNA polymerase II-dependent transcription in eukaryotes, integrating signals from enhancers to promoters across the transcription cycle. This review summarizes its modular composition and structural dynamics revealed by cryo-EM, and details its mechanistic roles in pre-initiation complex assembly, promoter-proximal pausing release, elongation progression, co-transcriptional splicing, termination via 3’-end processing, and chromatin changes. Recent progress elucidates allosteric signal transmission, TF multivalency, cohesin-mediated looping, and phase separation, while dysregulation implicates Mediator in cancer and developmental disorders. Future directions emphasize in vivo kinetics and therapeutic targeting of kinase modules to decode context-specific gene control.

1. Overview of the Mediator complex

Mediator was first identified in yeast as a multi‑subunit factor required for activator‑dependent transcription in reconstituted Pol II systems [1,2], a discovery that established it as an essential component of the general transcription machinery in eukaryotes. Subsequent biochemical and genetic studies in metazoans showed that Mediator is recruited to a vast majority of Pol II‑transcribed genes and acts as a central integrator of signalling pathways, coupling environmental and developmental cues to genome‑wide transcriptional programmes [3–6].

As an evolutionarily conserved transcriptional co‑regulator, the mammalian Mediator complex consists of ~ 30 subunits and exists as a dynamic ensemble of core and kinase-containing species in recent structural work. High‑resolution cryo‑EM and integrative studies have converged on a core Mediator (cMED) composed of Head, Middle, and Tail modules, with a reversibly associated CDK8 kinase module (CKM) [7,8] (Table 1). The human cMED is organized around MED14, which spans almost the entire assembly and functions as the structural backbone linking Head and Middle while anchoring the Tail [7,9,10]. The Head module engages the Pol II stalk and dock regions and contacts general transcription factors (GTFs), while the Middle module forms an elongated wedge beneath the Head that buttresses these interactions and provides additional surfaces for preinitiation complex (PIC) components [7,11]. The Head and Middle modules form a relatively rigid scaffold, whereas the Tail is conformationally flexible and enriched for transcription factor – interacting subunits, consistent with its role in integrating regulatory inputs [7,9,10]. High‑resolution maps now reveal that the distal Tail is extensively interwoven with cMED, and that its conformation can shift depending on bound activators, providing a structural route by which transcription factors remodel cMED to modulate PIC function [7,12]. The CKM (CDK8 or CDK19, Cyclin C, MED12/MED12L, MED13/MED13L) associates dynamically with one end of cMED. Structural as well as ChIP‑based analyses indicate that human CKM binding remodels the overall Mediator complex and occupies surfaces that otherwise engage Pol II and MED26 [13], thereby creating a mutually exclusive relationship between CKM-containing and PIC‑bound Mediator. Recent high‑resolution structural work further clarifies this mutual exclusivity, revealing that the intrinsically disordered region (IDR) of MED13 within the CKM masks the Pol II‑binding surface of cMED, directly blocking productive Mediator–Pol II interaction [14]. This IDR‑dependent masking mechanism provides a key structural basis for the mutually exclusive assembly of CKM‑bound versus Pol II‑bound Mediator, enabling compositional switching between enhancer‑bound and promoter‑bound states [13–17].

Table 1.

Components of the human Mediator complex.

Mediator orchestrates Pol II transcription by physically coupling activators to promoter PIC assembly, modulating kinase access to the Pol II C-terminal domain (CTD), and facilitating transitions from initiation to elongation, termination and RNA maturation. Mechanistically, TF binding to the Tail induces Tail module rearrangements that promote PIC assembly and promoter DNA commitment [18,19]. Besides, Mediator positions TFIIH/CDK7 near the Pol II CTD to facilitate CTD phosphorylation, as visualized in Mediator-PIC structures [7,9]. Mediator also interfaces with super-enhancers, where high TF density and phase-separated condensates amplify signal integration, driving robust expression of lineage-specifying genes [20–24]. Beyond basal roles, Mediator couples transcription to co-transcriptional processes: CTD phosphorylation recruits capping enzyme, splicing factors, and 3’-end processors, while Mediator-bound hnRNPs influence alternative splicing [4,25–27]. In vivo perturbation studies demonstrate Mediator’s necessity for stimulus-inducible transcription waves, with rapid eviction collapsing enhancer-promoter loops and gene expression dysregulation [3,9,28]. Thus, Mediator functions not merely as a transcriptional adaptor but as a central processor decoding regulatory landscapes into spatiotemporal gene expression patterns.

2. Transcriptional and co-transcriptional regulation

2.1. Pre-initiation and initiation

Mediator promotes transcription pre‑initiation and initiation by recruiting transcription machinery to the core promoter, organizing the PIC, and positioning the RNA Pol II CTD for TFIIH‑dependent phosphorylation that launches RNA synthesis. Distinct Head, Middle, Tail, and Kinase‑module subunits carry out separable mechanistic roles in activator recruitment, PIC nucleation, stabilization of Pol II and TFIIE/TFIIH, and regulation of the transition into a productively initiating complex (Figure 1(A)).

Figure 1.

Functional roles of the Mediator complex across the Pol II transcription cycle. Schematic models illustrate Mediator’s involvement in key transcription stages: (A) transcription initiation; (B) promoter escape and re-initiation; (C) pause release and elongation; (D) co‑transcriptional mRNA processing; and (E) transcription termination. Each panel highlights the relevant Mediator subunits and their associated co-regulators.

2.1.1. Recruitment of Mediator to regulatory DNA

At most protein‑coding genes, activator proteins bound to enhancers or upstream promoter elements nucleate Mediator recruitment via direct contacts with Tail‑module subunits (for example, human MED15, MED16, MED23, MED24) or with the CKM [5,29,30] (Figure 1(A)). Specifically, multiple DNA-bound activators synergize by bridging multiple binding sites through a single Mediator complex rather than by recruiting multiple Mediator molecules [30]. These interactions are often multivalent and dynamic, reflecting intrinsically disordered activation domains in transcription factors and flexible regions in Mediator that support the formation of transcriptional condensates at regulatory loci [5,29]. Once tethered, Mediator can bridge long‑range enhancer–promoter communication by simultaneously engaging activators and components of the basal transcription machinery, thereby increasing the local concentration of Pol II and GTFs at the target promoter [3,5,21,31]. Single‑molecule tracking in yeast indicates that Mediator molecules explore the nucleoplasm in a constrained fashion and dwell transiently on chromatin, guiding Pol II and TBP to accessible promoters to seed productive PIC assembly [32].

2.1.2. Assembly of the pre‑initiation complex

Key Mediator subunits act as structural scaffolds and functional hubs that stabilize the Pol II PIC and promote GTF recruitment [7,33] (Figure 1(A)). MED14 forms the central backbone of Mediator, maintaining overall complex integrity required for efficient PIC assembly at promoters. It interacts extensively with subunits such as MED17 and MED10, positioning the Head and Middle modules to contact Pol II and TFIIB, and its perturbation reduces stable recruitment of Mediator, Pol II, and TFIIB to promoters, indicating a key role in TFIIB‑dependent PIC formation [7,33,34]. MED17, a core Head‑module subunit, is crucial for organizing the Head and for anchoring Mediator to Pol II and promoter‑proximal GTFs; conditional MED17 mutants show a genome‑wide loss of PIC components and transcription, highlighting its central role in Head‑module – driven PIC assembly [7,33,34]. MED10 in the Middle module helps couple Mediator to TFIIB and other GTFs; mutations that weaken MED10–MED14 contacts selectively impair TFIIB and Pol II occupancy and destabilize later GTFs (TFIIE, TFIIH), underscoring how these ‘architectural’ subunits collectively coordinate the stepwise assembly and stability of initiation‑competent PICs [33,34].

2.1.3. Positioning and phosphorylation of the Pol II CTD

A hallmark of initiation is the phosphorylation of the heptad‑repeat CTD of Pol II by CDK7 within TFIIH [35–37]. Structural and biochemical work shows that human Mediator sandwiches the CTD between itself and TFIIH, stabilizing CDK7 and positioning the CTD for efficient multi‑site phosphorylation during early initiation [7]. In human cells, CDK7‑dependent phosphorylation of Ser5 and Ser7 at promoter regions is largely Mediator‑dependent, and Mediator–Pol II subcomplexes can independently recruit TFIIB and CDK7 to core promoters, highlighting Mediator as a platform for CTD modification [7,35,38]. This CTD phosphorylation leads to Mediator release and transitions Pol II towards promoter escape, thereby mechanistically linking Mediator‑guided PIC architecture to the onset of productive RNA synthesis [38–41].

Overall, Mediator drives transcription initiation by recruiting and organizing the PIC, stabilizing Pol II and GTFs at promoters, and positioning the Pol II CTD for TFIIH-dependent phosphorylation that enables promoter escape and early RNA synthesis.

2.2. Promoter escape and re-initiation

Promoter escape and transcription re‑initiation represent two mechanistically linked, post‑recruitment stages at which Mediator exerts crucial control over RNA Pol II transcription. After assembly of a Pol II–GTF PIC at the core promoter, Pol II must clear the promoter and transition into a stable early elongation complex; subsequently, promoters often undergo repeated rounds of initiation, producing transcriptional ‘bursts’ [5]. Biochemical, structural, and genomic studies indicate that Mediator influences both the efficiency and timing of promoter escape, as well as the formation and use of scaffold complexes that underlie re‑initiation and bursting [5,41] (Figure 1(B)).

2.2.1. Mediator dissociation and promoter escape

Functional data in budding yeast demonstrate that depletion or inactivation of Kin28 (the CDK7 ortholog) causes accumulation of Mediator at core promoters, consistent with a model in which CTD phosphorylation triggers Mediator release and thereby permits rapid promoter escape [41]. Genome‑wide occupancy profiling reinforces this view. Acute loss of Head or Middle subunits in yeast causes a downstream shift of Pol II signal relative to transcription start sites, suggesting that Mediator physically or functionally restrains Pol II at promoters and that its timely dissociation is required for efficient escape [42]. Together, these observations support a mechanistic sequence in which activator‑bound Mediator assembles and stabilizes the PIC, positions TFIIH to phosphorylate the CTD, and then disengages as Pol II transitions from an initially transcribing complex to an early elongation complex.

2.2.2. Mediator as a scaffold for re‑initiation

In defined in vitro systems, transcription re‑initiation is often faster than the first round of initiation, and this kinetic advantage depends on the persistence of a scaffold PIC that remains at the promoter after Pol II escape [3,43–45]. Mediator functions as a dynamic scaffold that stabilizes transcription complexes at promoters and promotes efficient RNA Pol II re-initiation rather than acting as a static coactivator [5,46]. Classically, this scaffold was thought to retain TFIID/TFIIA, TFIIH and TFIIE after the first round of initiation, providing a pre-assembled platform that rapidly recruits new Pol II, TFIIB and TFIIF for subsequent rounds of transcription, thereby supporting transcriptional bursting [5,45,47]. However, recent structural work revises this view, revealing stepwise dissociation of initiation factors during promoter escape [48]: short nascent RNAs (6 nt) displace the TFIIB reader domain without releasing full TFIIB; at 13–14 nt, upstream transcription bubble rewinding triggers dissociation of the TBP–TFIIA–TFIIB upstream promoter complex; TFIIF, TFIIE, and TFIIH persist in early elongation complexes until the RNA reaches ≥22 nt, when DSIF/NELF recruitment promotes TFIIE/TFIIH dissociation and elongation commitment [48]. This refined model suggests that DNA-binding factors (TFIID, TFIIA, TFIIB) may be more stably retained at promoters, whereas it is less likely for TFIIE, TFIIF and TFIIH to persist as stable scaffold components.

2.2.3. CKM with re-initiation

The CKM modulates Mediator’s interaction with Pol II and thereby influences re‑initiation [49,50]. Structural analyses in fission yeast demonstrate that the CDK8 module binds to the Pol II‑binding cleft of Mediator, sterically blocking Pol II association independently of CDK8 catalytic activity [49]. Specifically, the CDK8 subcomplex associates with the Mediator tail region through Med13, and this association prevents Mediator from productively interacting with RNA polymerase II [50]. More broadly, conceptual models posit a cycle in which cMED supports PIC assembly and initial initiation, after which CDK8 module binding to Mediator prevents immediate re‑recruitment of a second Pol II to the same scaffold [3,49,50]. Because efficient re-initiation depends on rapid recruitment of additional Pol II molecules to promoter-bound scaffold complexes, CDK8 provides a way to fine-tune transcript output by limiting repeated rounds of initiation.

Overall, Mediator promotes promoter escape by enabling Pol II to clear the promoter after CTD phosphorylation, and it supports rapid transcription re-initiation by acting as a scaffold that preserves promoter-bound factors for repeated rounds of bursting transcription.

2.3. Promoter-proximal pausing and elongation

Promoter-proximal pausing is now recognized as a widespread and regulated checkpoint in metazoan transcription, in which Pol II is paused approximately 30–60 nt downstream of the transcription start site [51–53]. Pausing is enforced primarily by DSIF (SPT4/SPT5) and NELF, which interact with the early elongation complex to stabilize a paused conformation of Pol II [54–57]. Release from this paused state requires recruitment of P‑TEFb (CDK9–Cyclin T) and associated elongation factors [54,56–61] (Figure 1(C)).

At the chromatin level, paused Pol II accumulates at nucleosome-depleted promoter regions upstream of the +1 nucleosome, and efficient transit through this first nucleosome represents an additional kinetic barrier [62]. Mediator, which is present at many active promoters, is well placed to couple pause establishment and release to upstream signals and local chromatin architecture. This section focuses on how Mediator or its subunits contribute mechanistically to pausing, pause release, and productive elongation.

2.3.1. Core Mediator in early elongation and +1 nucleosome transit

Biochemical reconstitution has shown that Mediator can directly promote Pol II passage through the +1 nucleosome in a defined chromatin template system [62]. In this assay, Mediator stimulated synthesis of RNAs and enhanced Pol II transit across nucleosomal DNA, indicating a direct role beyond pre-initiation and initiation [62]. Mechanistically, Mediator appears to facilitate the formation of an elongation-competent complex by coordinating Pol II, GTFs, and elongation factors such as SII(TFIIS) and DSIF, which are known to cooperate with Mediator in yeast and mammalian systems [62,63]. These data support a model in which Mediator, initially assembled at promoters, can remain associated with early elongation complexes, helping Pol II overcome the +1 nucleosome and transition from a paused to a processive state.

2.3.2. Mediator–SEC–p‑TEFb networks in pause release and elongation

Several studies converge on a model in which Mediator, through MED26 and other subunits, helps assemble P‑TEFb-containing complexes at paused promoters to drive pause release [58,64,65]. The super‑elongation complex (SEC), built around AFF family scaffold proteins, collaborates with P‑TEFb to phosphorylate Pol II Ser2, DSIF SPT5, and NELF subunits, events that promote NELF dissociation and convert DSIF into a positive elongation factor [54,58,64–67]. Mechanistically, MED26 serves as a docking site that recruits AFF1/4‑SEC–P‑TEFb complexes to paused Pol II, forming a coordinated recruitment pathway with Brd4 and PAF1C that positions multiple P‑TEFb species on DSIF and NELF [54,58,65]. Loss of Mediator subunits produces phenotypes similar to P‑TEFb or SEC depletion in early Drosophila embryos, where a subset of rapidly transcribed, non-paused genes is highly sensitive to these factors, supporting in vivo collaboration between Mediator and SEC/P‑TEFb, in elongation control [68]. Together, these findings support the idea that Mediator scaffolds an integrated pausing – elongation module by bridging activators, pausing factors, and SEC/P‑TEFb complexes (Figure 1(C)).

The Tail module subunit MED23 promotes RNA Pol II transcription beyond initiation. MED23 maintains basal transcription by recruiting P-TEFb to target genes via direct binding to its CDK9 subunit [69]. In Med23 knockout embryonic stem cells, Egr1 mRNA is reduced despite unchanged promoter-bound Pol II, indicating a defect after initiation [69]. ChIP and ChIP‑seq reveal decreased P-TEFb and Ser2‑phosphorylated Pol II over gene bodies, while DSIF and NELF occupancy are unaffected [69]. Biochemical assays demonstrate MED23–CDK9 interaction both in vivo and in vitro, supporting a model in which MED23-containing Mediator promotes pause release and productive elongation under basal conditions [69].

2.3.3. Kinase module regulation of elongation

The Mediator kinase module reversibly associates with cMED to form CDK‑MED, adding another regulatory layer to pausing and elongation. Historically viewed as a repressive module that blocks Pol II–MED interaction, the kinase module is now known to exert positive and negative effects on elongation, depending on gene context and signalling state [13,15,70–72]. CDK8 can phosphorylate transcription factors, Pol II CTD, and chromatin regulators, thereby tuning their capacity to recruit or activate elongation factors [15,73]. Within the serum‑response network, CDK8 acts as a positive regulator of post-recruitment steps, promoting rapid induction of immediate-early genes where paused Pol II is a common feature [73]. At these genes, CDK8 is required for proper recruitment or activation of elongation factors and for productive passage of Pol II into gene bodies, consistent with a role in pause release and early elongation [73].

Notably, the kinase subunit paralogs CDK8 and CDK19 are functionally non-redundant and exert distinct regulatory roles in stimulus-specific transcription programmes [70,72]. In the hypoxia response, CDK8–but not CDK19–functions as an essential coactivator. Under hypoxic conditions, the transcription factor HIF1A directly interacts with CDK8-containing Mediator via its C-terminal transactivation domain, recruiting CDK8-MED independently of HIF1A-driven histone acetylation. Hypoxia-inducible genes are already transcriptionally engaged and paused at promoters in normoxia, with pause release as the rate-limiting step for activation [70]. CDK8 is critical for recruiting the SEC, thereby triggering Pol II pause release and productive elongation [70]. In the IFN‑γ response, CDK8 acts in a kinase‑dependent manner, phosphorylating STAT1 and promoting pause release, while CDK19 functions through a kinase‑independent scaffolding role to support antiviral gene expression [72].

More broadly, CDK‑MED influences enhancer activity, eRNA transcription, and Pol II pausing at regulatory elements, suggesting that Mediator can tune pausing and elongation not only at promoters but also at enhancers that feed into promoter-proximal checkpoints [15,72].

Overall, Mediator helps convert promoter-proximal pausing into productive elongation by coordinating +1 nucleosome transit, recruiting P-TEFb/SEC and other elongation factors, and tuning Pol II escape and gene-body transcription in a context-dependent, subunit-specific manner.

2.4. Co-transcriptional RNA processing and termination

Mediator and specific subunits help couple transcription by RNA Pol II to RNA splicing, 3′ end formation, and termination by physically and functionally bridging Pol II with RNA processing machineries and termination factors (Figure 1(D,E)).

2.4.1. Coupling transcription to RNA processing

Mediator subunit MED23 links transcription to alternative mRNA processing by directly interacting with splicing factors, especially hnRNP L [25]. MED23 recruits these factors to chromatin, partially colocalizing with U1/U2 snRNP-associated machinery, and thereby controls a subset of hnRNP L – dependent alternative splicing and alternative cleavage/polyadenylation events genome-wide [25]. Exon arrays, minigene reporters, and ChIP-seq show that knockdown of MED23 alters hnRNP L occupancy and disrupts specific transcript isoforms, establishing Mediator as a platform coordinating transcription with co-transcriptional mRNA processing [25].

Mediator kinase module is shown to regulate pre‑mRNA splicing as one of its downstream mechanisms for controlling gene expression. Functionally, the kinase module helps maintain pathway‑ and cell‑type‑specific splicing patterns, particularly for interferon‑responsive genes in trisomy‑21 (T21) cells, where its hyperactivity contributes to excessive interferon signalling [74]. Inhibition of CDK8/CDK19 with a selective inhibitor (cortistatin A, CA) alters alternative splicing events – such as exon skipping/inclusion – across IFN‑related and metabolic pathways in a context‑dependent manner [74]. The research suggests that Mediator kinases may influence splicing both directly, by phosphorylating splicing‑linked proteins, and indirectly, by modulating transcription factors and expression levels of splicing regulators [74]. In Down Syndrome (DS)‑derived cells, this kinase‑dependent splicing control helps tune the interferon response, and inhibiting the Mediator kinase module reduces aberrant splicing and dampens hyperactive interferon signalling [74].

2.4.2. Mediator and 3′‑end processing/termination crosstalk

Recent work shows that Mediator directly participates in 3′‑end processing and transcription termination by interacting with the cleavage and polyadenylation specificity factor (CPSF) complex [27] (Figure 1(E)). Mediator forms a complex with CPSF through a MED23–FIP1 interaction, linking Mediator to the pre‑mRNA 3′‑end processing machinery at gene termini [27]. MED23 functions as an RNA‑binding protein at 3′ regions; its knockdown causes a subset of transcriptional readthrough and the formation of gene–gene fusion transcripts, demonstrating an active role in enforcing termination [27]. MED23‑dependent termination defects correlate with readthrough events in breast cancer, implicating Mediator‑CPSF crosstalk in tumorigenesis [27].

Mediator also participates in promoter–terminator communication, likely through gene looping and higher‑order chromatin interactions. In budding yeast, Srb5/Med18 is required for efficient termination [75,76]; Med18 mutants show Pol II accumulation downstream of poly(A) sites and defective recruitment of CF1 3′‑end processing factors near the 3′ end [75]. Srb5/Med18 promotes promoter–terminator gene looping, and termination defects in Med18 mutants correlate with loss of such loops, suggesting that Mediator‑dependent chromatin architecture stabilizes CPA factor binding and termination efficiency [76].

The MED26–little-elongation complex (LEC) axis illustrates how Mediator specifies termination pathways for particular transcript classes. MED26 recruits LEC to snRNA and replication‑dependent histone genes, often localized within Cajal bodies and histone locus bodies, where LEC associates with CBC-ARS2, NELF/DSIF, and Integrator or histone‑specific 3′‑processing factors [77]. Recent work further shows that MED26–LEC signalling drives TES-proximal pausing at replication‑dependent histone genes, which is essential for accurate 3’-end processing [78]. Mechanistically, MED26-LEC maintains physical association between Cajal bodies and histone locus bodies, enabling efficient delivery of 3’-processing factors. Disrupting this interaction abolishes TES-proximal pausing and impairs 3’ maturation [78]. Thus, the MED26–LEC axis enforces proper termination and promotes proper 3′‑end processing, preventing readthrough and aberrant extension of non‑polyadenylated transcripts [77,78].

Collectively, Mediator coordinates co-transcriptional RNA processing by linking RNA Pol II with splicing and 3′-end machinery, while also helping enforce proper transcription termination and prevent readthrough (Figure 1(E)).

2.5. Chromatin modification and organization

Mediator modulates chromatin both by recruiting chromatin modifiers/remodellers and by influencing higher-order genome organization, especially enhancer – promoter communication and domain architecture.

2.5.1. Direct links to chromatin modifiers

Mediator can interact directly with p300/CBP, promoting cooperative assembly of complexes that increase local histone acetylation, thereby enhancing promoter accessibility and transcriptional activation [79,80].

CDK-MED associates with PRMT5–WDR77 and exhibits histone H4-specific arginine‑methyltransferase activity, catalysing H4R3 symmetric demethylation (H4R3me2s) and contributing to transcriptional repression at C/EBPβ-inducible promoters [81]. H4R3me2s can recruit DNMT3A, linking Mediator-dependent arginine methylation to DNA methylation and durable transcriptional silencing [81]. MED12 physically interacts with the repressive histone methyltransferase G9a, forming a CDK8-module – associated pathway that represses neuronal genes via H3K9 methylation [82]. CDK-containing modules can form distinct Mediator subcomplexes with chromatin-modifying activity, such as T/G-Mediator containing TRRAP–GCN5L, which can directly acetylate chromatin and influence transcriptional programmes [83,84].

Tail subunits such as MED16, MED23 and MED25 integrate signals from sequence-specific activators and are genetically connected with chromatin modifiers and HDACs, positioning them to coordinate upstream signalling with local chromatin state [85,86]. For example, Mediator MED23 couples monoubiquitination of histone H2B (H2Bub) to transcriptional control [87]. Mechanistically, MED23 recruits or stabilizes the H2B ubiquitination machinery at active genes by physically associates with the E3 ligase complex RNF20/40 and the PAF complex, and in cell‑free chromatin systems Mediator enhances H2Bub1 when MED23 is present [87]. MED23 knockout alters H2Bub levels at a subset of target genes and balancing cell fate decisions between myogenesis and carcinogenesis [87].

In plants, Mediator subunits modulate recruitment of hormonal signalling components and histone-modifying activities to promoters, illustrating conserved principles of Mediator-guided chromatin changes across eukaryotes [88–91].

2.5.2. Enhancer – promoter communication and 3D genome organization

Mediator complex emerges as a central organizer of enhancer–promoter communication by partnering with cohesin to shape 3D chromatin structure and transcription. In detail, Mediator and cohesin jointly bridge enhancers and promoters to control active gene expression programmes in mouse embryonic stem cells [21]. Genome-wide ChIP-seq reveals extensive co-occupancy of Mediator and cohesin at promoters and pluripotency enhancers, where they physically interact in a shared complex [21]. The cohesin loader NIPBL associates with this complex, enabling cohesin loading at regulatory sites. Chromosome conformation assays demonstrate that Mediator – cohesin – occupied sites engage in enhancer–promoter DNA loops whose disruption reduces target gene transcription [21]. Distinct sets of genes show co-occupancy in different cell types, indicating cell-type-specific looping architectures linked to transcriptional programmes [21].

Acute depletion of cMED subunits by rapid protein-depletion strategies reduces enhancer–promoter contact frequencies and strongly decreases expression of target genes, indicating a direct role in promoting productive enhancer–promoter interactions [22]. Loss of Mediator causes a rapid, genome wide reduction in enhancer–promoter interaction frequencies that closely tracks transcriptional downregulation. At the same time, cohesin is redistributed away from enhancers towards CTCF sites, increasing CTCF–CTCF contacts [22]. These results suggest Mediator helps focus cohesin at enhancers and promoters to stabilize regulatory loops and that, in its absence, cohesin is ‘captured’ by CTCF-anchored architectural loops [22].

MED23 helps bridge enhancers and promoters; an intellectual-disability–linked MED23 R617Q mutation weakens enhancer–promoter looping without disrupting Mediator complex assembly [92]. This impaired looping selectively deregulates chromatin architecture and gene expression programmes, highlighting MED23’s role in regulating higher-order chromatin topology [92].

In yeast, Mediator binds the boundaries of chromosomal interaction domains and associates with complexes such as RSC (remodeller), CPF/CF1A (RNA processing complex), and actin-related complexes, suggesting that Mediator helps couple transcription, chromatin remodelling, and nuclear architecture at domain borders [93].

Research also dissects how the PIC and Mediator cooperate in promoter–enhancer communication. Using acute depletion of TAF12 (TFIID) or MED4 (Mediator middle subunit) in mouse embryonic stem cells, researchers show that both factors are required for optimal Mediator association with promoters and for transcription, but their loss does not disrupt promoter–enhancer looping measured by 3C-based assays [94]. Thus, PIC integrity strongly influences Mediator binding at regulatory elements and transcriptional output, yet stable promoter–enhancer physical contacts seem to persist without a fully assembled PIC or normal Mediator occupancy, functionally uncoupling looping from PIC-dependent activation [94].

A growing body of work links Mediator’s intrinsic disorder and flexible regions to phase separation and transcriptional condensate formation at enhancers and promoters [20,23,95,96]. Early studies have uncovered the function of ncRNA/eRNA in transcription regulation, in part through interactions with Mediator to facilitate enhancer communication [97,98]. More-recent research revealed that Mediator integrates ncRNA/eRNA signals into a broader network, facilitating phase-separated transcriptional hubs and positive feedback in which eRNA production reinforces Mediator-driven hubs [95]. This positions Mediator as essential for RNA-mediated gene regulation, with implications for development and disease.

Overall, Mediator acts as a chromatin organizer that couples activator signalling to histone modification, cohesin-dependent enhancer–promoter looping, and higher-order genome architecture to coordinate transcriptional activation or repression across cell types.

3. Future perspectives

3.1. Integrative role in transcription

Mediator acts as an interface between sequence-specific transcription factors, enhancers, and the Pol II pre‑initiation complex, enabling enhancer-dependent and stimulus-responsive control of gene expression. Recent work emphasizes its roles not only in initiation, but also in elongation, reinitiation, transcriptional bursting, and coordination with chromatin regulators. Going forward, an important perspective is to treat Mediator less as a passive scaffold and more as an active regulatory machine whose conformational states encode regulatory information. Dissecting how individual subunits and modules tune transcriptional outputs across different genomic and cellular contexts remains a key challenge.

3.2. Structural dynamics

Cryo‑EM has delivered near‑complete structures of yeast and human Mediator, as well as high‑resolution views of Mediator within the Pol II pre‑initiation complex [7,9,10,99–103]. These studies reveal a modular architecture and extensive conformational plasticity that underlie its regulatory versatility. Integrating high‑resolution structural snapshots with dynamic, single‑molecule imaging and functional assays will be critical to unravel how Mediator’s conformational changes are precisely coupled to the diverse and dynamic transcription factor signalling. Future structural studies, combined with in vivo structural biology and computational modelling, will further clarify how disease‑associated mutations alter Mediator dynamics and transcriptional output, offering new opportunities for targeted therapeutic interventions.

3.3. In vivo function and disease relevance

Genetic models, including conditional mouse knockouts, have begun to define essential roles for Mediator subunits in development, metabolism, and organ homoeostasis. Specific subunits such as MED13, MED13L, MED15, MED23, and components of the CDK8 kinase module have emerged as regulators of neural system development, cardiovascular function, metabolic homoeostasis, and stress responses, highlighting the complex as a potential precision‑medicine target [104–110].

Parallel studies implicate subunits such as MED19, MED23 and kinase‑module components (e.g. CDK8) in tumour progression, viral responses, and control of protein synthesis, suggesting opportunities for targeted therapies and combination regimens [111–116]. A key perspective is that pathological phenotypes often arise from subunit‑specific, tissue‑restricted dysfunction rather than global loss of the entire complex, opening a window for selective intervention.

3.4. Therapeutic and technological opportunities

The therapeutic outlook includes small‑molecule inhibitors of CDK8/19, modulators of specific Mediator–transcription factor interfaces, and approaches that tune the expression or stability of disease‑linked subunits. For example, CDK8/19 inhibition with CA rewires metabolism, especially under hyperactive interferon signalling in Down Syndrome (DS) cells [74]. Metabolomics show altered core metabolic pathways in a cell type – specific manner, with broad upregulation of anti-inflammatory lipid mediators during IFNγ stimulation [74]. These lipids act as ligands for nuclear receptors PPAR and LXR, whose activation is selectively enhanced in kinase-inhibited, IFN-stimulated T21 cells, linking Mediator kinase activity to lipid metabolism and nuclear‑receptor signalling [74]. RNA-seq further indicates increased expression of fatty‑acid and cholesterol‑metabolism genes after CDK8/19 inhibition, helping explain elevated anti-inflammatory lipid production [74]. Since JAK inhibition also dampens DS interferonopathy in vivo [117,118], CDK8/19 inhibitors might offer an orthogonal or combinable strategy acting downstream at the transcriptional level. Integration with emerging genome‑editing and RNA‑based technologies could enable correction of pathogenic variants in Mediator subunits or fine‑tuning of their dosage in affected tissues.

Conceptually, future work will benefit from systems‑level approaches that combine genome‑wide binding and perturbation data with structural and single‑molecule information to build predictive models of Mediator-controlled gene networks. Understanding how Mediator integrates environmental cues, developmental programmes, and cellular stress into precise transcriptional outputs will be central to translating basic mechanistic insight into clinical applications.

Funding Statement

The work was partially supported by the Ministry of Science and Technology of the People’s Republic of China (SQ2024YFE0200827 to G.W.), and the Mainland-Hong Kong Technology Cooperation Funding Scheme of Innovation and Technology Commission of Hong Kong SAR (MHP/372/24 to Q.Z.).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing is not applicable to this article, as no data were created or analysed in this research.