Chromatin, the combination of DNA and protein that resides within our nuclei, takes on a variety of forms. These forms vary depending on the type of cell, the stage of a cell’s cycle and environmental conditions. Techniques to pinpoint regions of chromatin contacts within a nucleus have provided great insight into nuclear organisation and the consequences this has on gene expression. However, our mechanistic understanding of these dynamic chromatin changes and the role different proteins play is still incomplete. Recently published in Cell (1) Rao et al. have established how loss of one protein, cohesin, affects nuclear architecture and transcription.
To fit DNA into a nucleus it needs to be packaged. This is achieved with the help of proteins that not only aid compaction, but may bind specific DNA sequences bringing regions of DNA kilobases away to interact. Consequently, this forms loops. Loops are a common structural motif that commonly group to form loop domains (see my previous blog (4) on the hierarchal nature of chromatin packaging for more detail).
One protein tightly connected with loops is cohesin. Most famously known for its role in regulating sister chromatid separation during mitosis, cohesin is an integral member of the cellular proteome. But separating chromosomes is not its only function. In the nucleus, a ring of cohesin, typically associated with CTCF (a CCCTC- binding factor), together coordinate chromatin loops and topologically associated domains (TADs) (Figure 1) (2), where DNA contact are frequent. The extent to which cohesin participates in this process and why this happens is still unclear.
Arguably the best way to determine what a protein does is to see what happens (or doesn’t happen) without it. Cohesin is a complex of proteins – by destroying its core, RAD21, with an auxin-inducible degron system*, the complex could no longer associate with DNA. The experiments were performed in human colon cancer cells.
*This is the name of a technique to target proteins for destruction exploited from plants. Rapid degradation occurs when auxin is present; auxin binds the degron that is tagged to RAD21 allowing it to be recognised by TIR1, stimulating protein destruction. See (3) for more.
Where did they go?
To see the changes that occurred during cohesin removal, Rao’s team used Hi-C, a technique that examines the frequency of physical contact between loci in the genome. For example, this technique would identify the presence of loops where pairs of loci have greater contact frequency than random pairs of loci that are at a similar distance along the genome sequence.
Before auxin treatment, the Hi-C algorithms identified 2,140 loop domains, all of which disappeared when cohesin was removed.
A time course of cohesin removal and recovery was conducted to get a better understanding of the reversibility of loop formation. When auxin was removed, loops reformed by 1 hour, however there was much variability in rates between single loops.
Fast-forming loops typically contained NIPBL binding sites at loop anchors. NIPBL is a component of the cohesin loading complex, so I guess that kind of makes sense. Promoter and enhancer* interactions also quickly reformed. The strongest correlation though was with superenhancers (basically a DNA sequence with lots of enhancers) – they were 159x more likely to span a fast than slow loop domain.
*A brief into to enhancers (http://www.nature.com/scitable/definition/enhancer-163)
What are the consequences?
As well as locating DNA interactions, Rao’s team also studied the epigenetic and transcriptional consequences of cohesin knock-down. Histone proteins, that DNA wraps around (4), are modified with various tags such as methyl and acetyl groups that can affect chromatin compaction – without cohesin, ChIP-seq data confirmed that little changed.
Regions of chromatin form compartments where similarly modified histones are grouped together. The compartments themselves showed little change, but by observing interaction frequencies at compartment boundaries before and after auxin addition, it was seen that there was an increase during cohesin’s absence. Cohesin seems not to be involved in compartment formation, but may interfere by forming domains of differently modified histones.
Transcription also appears to be minimally affected by cohesin loss. It was thought that without cohesin rogue enhancers could ectopically express previously unexpressed genes. However, this occurred for around 1% non-expressed genes examined. More genes were discovered to be down-regulated suggesting it facilitates enhancer interactions.
These results bring us closer to looping the loop and crossing the T’s and reinforce previous roles of cohesin and gene regulation.
If you want to see the loop kinetics in action, check out this video-https://www.youtube.com/watch?v=Q_KdrtsmYoE – much better than my dodgy sketches!
(1) S. Rao. Cohesin loss eliminates all loop domains Cell 2, 305-320 (2017) – should check out
(2) G.Busslinger Cohesin is positioned in mammalian genomes by transcription, CTCF and WapI. Nature 544 503-507 (2017)
(3) K. Nishimura. An auxin based degron system for the rapid depletion of proteins in nonplant cells Nature Methods 6, 917-922 (2009)