Enhancer sequences increase gene transcription with the help of a co-activator complex, the Mediator. Another protein complex — cohesin — seems to work with Mediator to bring together enhancers and promoters. See Article p. 430
The multiprotein Mediator complex is a general regulator of gene transcription. In eukaryotic organisms (such as animals, plants and fungi), it allows high levels of activator-dependent transcription by bringing together diffusible trans-acting factors, such as proteins that regulate gene expression, and the basal transcriptional machinery at the promoter sequences1. Trans-acting transcription factors bind to cis-regulatory DNA sequences, which regulate the expression of genes on the same DNA molecule. It has long been suspected that Mediator also links up proximal or distal cis-regulatory elements, such as enhancers, with the promoters. But how does Mediator do this? A paper2 in this issue describes the discovery that the Mediator complex recruits another protein complex, cohesin, to provide coherence between enhancer and promoter sequences (Kagey et al., page 430).
Mediator function — effected by more than two dozen different proteins in the complex — is evolutionarily strongly conserved and is similar in both yeast and mammals. Proximal and distal cis-regulatory elements facilitate the recruitment of this complex to promoter regions1. The Mediator–promoter interaction subsequently stabilizes the formation of the pre-initiation protein complex, which, among other functions, helps to position the RNA polymerase II enzyme complex over gene transcription start sites1. The formation of the pre-initiation complex is likely to be facilitated by the direct interaction between Mediator and the carboxy-terminal domain of the largest subunit of RNA polymerase II. Subsequently, phosphorylation of this domain, and the consequent release of RNA polymerase II from Mediator, may be necessary to trigger the elongation step of transcription3. Although this sketch outlines well-established features of Mediator function, the inner workings of this complex are poorly understood.
Kagey et al.2 provide insight into this enigma. The authors started out by screening for regulators of pluripotency — the ability of stem cells to differentiate into various cell types. Specifically, they searched for genes that are essential for maintaining the expression of the transcription factor Oct4 in mouse embryonic stem (ES) cells. As anticipated, they identified genes encoding members of the Mediator complex. But the screen also revealed a few surprises: among the regulators of Oct4 expression were genes encoding key members of the cohesin complex (Smc1a, Smc3 and Stag2) as well as the gene for Nipbl, a protein that loads cohesin onto chromatin4,5 (DNA–protein complexes).
The cohesin complex is mainly known for its function in keeping together sister chromatids of a chromosome in the period from the S phase of the cell cycle to the M phase. More recently, however, it has also been linked to regulation of gene transcription4,5,6. In particular, the observation that cohesin co-localizes to binding sites for the CTCF 'insulator' protein has generated substantial interest4,5. Although the underlying mechanism remains elusive, it is generally assumed that cohesin stabilizes repressive chromatin loops organized by CTCF binding, thereby aiding CTCF to 'insulate' promoters from enhancer communications4,5.
Kagey et al. convincingly show that, in mouse ES cells, a significant fraction of chromatin-bound cohesin also localizes to genomic regions that are devoid of CTCF-binding sites. This finding agrees well with another observation7 that, in human cells, cohesin localizes to regions that do not bind to CTCF but contain binding sites for known regulators of tissue-specific gene expression. Similarly, genetic evidence indicates6 that, in the fruitfly Drosophila, enhancer–promoter communications might require a cohesin function.
What makes Kagey and colleagues' report particularly significant is the robust demonstration that the cohesin and Mediator complexes not only co-localize to regions devoid of CTCF-binding sites, but also are simultaneously present in close physical proximity to each other in chromatin. Moreover, the authors show that the Mediator–cohesin complexes promote and/or stabilize the physical proximity between enhancers and promoters of active genes only (Fig. 1).
That the Mediator–cohesin complexes occupy about 60% of all active promoters in ES cells hints that any tampering with this complex will probably affect most of the pivotal features of ES cells, including pluripotency2. Although it remains to be determined, active promoters that do not interact with Mediator may still depend on cohesin function — perhaps in combination with transcriptional co-activators other than Mediator.
The conclusion that a significant part of cohesin's function is to facilitate crosstalk between enhancers and the basal transcriptional machinery bound to the promoter region throws new light on several human diseases. Investigations of disorders with mutations in members of the Mediator complex — such as Opitz–Kaveggia and Lujan syndromes, and schizophrenia8 — should now also consider a NIPBL/cohesin component. Conversely, studies of Cornelia de Lange syndrome, which features mutations in NIPBL (ref. 9), should take into account a possible role for the Mediator complex.
But although these diseases all involve misregulation of gene expression8,9, they do not seem to have other comparable features. One explanation for this could be cell-type specificity of Mediator–cohesin function. Indeed, as Kagey et al.2 demonstrate, there is an extensive change in the patterns of cohesin and Mediator co-localization on chromatin between ES cells and cells derived from them — mouse embryonic fibroblasts. In all likelihood, this reflects epigenetic reprogramming events that control the availability of lineage-specific cis-regulatory elements.
Kagey and co-workers' report opens a Pandora's box, raising more questions than it answers. Pertinent issues include the need for a better understanding of how the cohesin complex can be differentially recruited to the chromatin regions displaying enhancer and insulator functions, and how the associated cohesin functions are executed. Among the possible ways in which this complex might facilitate activation of transcription are stabilization of an open chromatin conformation6 and/or the anchoring of the transcriptional units to transcription 'factories' (Fig. 1) — nuclear compartments that can be visualized when several genes, even from different chromosomes, gather together in the interchromosomal space to be transcribed simultaneously10.
Given that cohesin is the sole known common denominator between both activation and inhibition — insulation — of transcription, it is not far-fetched to assume a mechanistic relationship between these two processes. It is of interest that, on the evolutionary scale, cohesin's role in activation of transcription may have pre-dated its involvement in CTCF-mediated insulation of transcription6. Whether this means that a CTCF–cohesin function acting at long range has emerged in vertebrates to counteract the enhancer–cohesin function, thereby preventing unscheduled communications between enhancers and promoters from neighbouring expression domains, is another intriguing puzzle.