Enhancer Journal Club: Activating RNAs associate with Mediator to enhance chromatin architecture and transcription

Another paper for the enhancers/eRNAs Journal Club series. This time I've been reading Lai et al. which was published in February from Ramin Shiekhattar's lab. Following on from the back to back Li and Lam papers in June, I've decided to go back to this paper because it is (to my knowledge) the first to show a link between expression of a non-coding RNA and the formation of enhancer/promoter contacts.

I've posted a non-technical summary of this paper over at Science Gist: http://sciencegist.com/doi/10.1038/nature11884

The abstract for the paper is as follows:

Recent advances in genomic research have revealed the existence of a large number of transcripts devoid of protein-coding potential in multiple organisms. Although the functional role for long non-coding RNAs (lncRNAs) has been best defined in epigenetic phenomena such as X-chromosome inactivation and imprinting, different classes of lncRNAs may have varied biological functions. We and others have identified a class of lncRNAs, termed ncRNA-activating (ncRNA-a), that function to activate their neighbouring genes using a cis-mediated mechanism. To define the precise mode by which such enhancer-like RNAs function, we depleted factors with known roles in transcriptional activation and assessed their role in RNA-dependent activation. Here we report that depletion of the components of the co-activator complex, Mediator, specifically and potently diminished the ncRNA-induced activation of transcription in a heterologous reporter assay using human HEK293 cells. In vivo, Mediator is recruited to ncRNA-a target genes and regulates their expression. We show that ncRNA-a interact with Mediator to regulate its chromatin localization and kinase activity towards histone H3 serine 10. The Mediator complex harbouring disease displays diminished ability to associate with activating ncRNAs. Chromosome conformation capture confirmed the presence of DNA looping between the ncRNA-a loci and its targets. Importantly, depletion of Mediator subunits or ncRNA-a reduced the chromatin looping between the two loci. Our results identify the human Mediator complex as the transducer of activating ncRNAs and highlight the importance of Mediator and activating ncRNA association in human disease.

The paper starts with a small siRNA screen of possible factors that may mediate activating non-coding RNA effects on neighbouring gene transcription. The single hit shown is for Med12, a component of mediator complex, but they go on to show that several other components of the complex also have the same effect. Once they have confirmed that knocking out a mediator component also affects the transcription of the genes they propose are under the control of this ncRNA, they also show that knocking down the ncRNA itself reduces the occupancy of mediator at the promotors of this gene. This is a nice indication that the ncRNA might be somehow delivering the mediator complex to the target genes. On top of this, they use a kinase assay to show that two of their non-coding RNAs actually appear to stimulate the kinase activity of mediator towards H3S10, at least in vitro. Here it maybe would have been nice to have a ChIP of H3S10p levels in vivo, just to show that this mechanism may also be relevant in the biological system.

The really interesting part of the paper, at least for me, is figure 4. Here they use siRNAs to knock down either mediator or the ncRNA and show a loss of DNA looping between the ncRNA locus and the neighbouring gene. Could this indicate that the ncRNA is somehow involved in forming these chromatin loops? Unfortunately, it is a little difficult to tease apart the causality in this particular experiment, as they have shown that knocking down either mediator or the ncRNA causes a reduction in transcription of the gene. The looping could therefore be a consequence of this transcriptional activity and not be directly caused by the ncRNA/mediator interaction. Alternatively, this could all be working through a cohesin related pathway since cohesin seems to be involved in stabilising these sorts of looping interactions and can also interact with mediator. This would be interesting as the authors show in figure 1 that knocking down SMC1L or NIPBL doesn't seem to impair the ability of the activating ncRNAs to activate luciferase transcription.

Taken together with the papers from the Rosenfeld and Glass labs, this paper helps to demonstrate that there is some relationship between non-coding/enhancer RNAs and enhancer/promoter looping that we really don't understand yet.

Next in the Journal Club series: Hah et al. 2013 in Genome Research, which very much contradicts the results of these three papers.