Removing the protein complex cohesin from chromosomes destroys one layer of the genome's 3D structure but leaves another intact. Genome structure is therefore built by independent processes that work together. See Letter p.51
For decades, biologists have studied how a metre-long genome folds to fit into a microscopic cell nucleus, and how such folding affects genome function. On page 51, Schwarzer et al.1 report a leap in our understanding of how this process works in mammals. They demonstrate that removing the ring-shaped protein complex cohesin from chromosomes in the mouse liver drastically alters three-dimensional genome structure. In so doing, they uncover key principles underlying chromosome organization.
The 3D interactions of chromosomal regions across the genome can be analysed using a technique called Hi-C, which combines molecular tools that capture interacting chromosome regions with high-throughput DNA sequencing. Detailed Hi-C interaction maps2,3 show that mammalian chromosomes are partitioned into structural units called topologically associating domains (TADs), which help to control the timing of gene activation. If a boundary between two TADs is lost, pieces of the genome that should be kept separate can interact, leading to gene activation at the wrong time or in the wrong tissue. Such a loss of TAD boundaries can lead to cancer or to atypical development, including the growth of extra fingers or toes4,5.
Groups of TADs associate to form larger domains called compartments, in which transcriptionally active (A compartment) and inactive (B compartment) chromosomal regions are spatially separated (Fig. 1a). The pattern of TAD segregation into compartments changes during cell differentiation and is disrupted in some diseases6. Despite this evidence that TADs and compartments are important for genome function, however, basic information about how these structures are built and maintained has been lacking. For instance, researchers have debated whether compartmentalization depends on the integrity of smaller-scale TADs. It has also been unclear whether compartments and TADs are separate types of structure or just the same type of phenomenon at different scales.
Various DNA-binding proteins and complexes, including cohesin and the protein CTCF, cluster at TAD borders7. The available data are consistent with a model8 in which the extrusion of DNA through the cohesin ring creates loops, helping to form TADs. However, attempts to determine whether removing cohesin causes TADs to fall apart have been frustratingly inconclusive. Previous studies9,10,11 found that TADs can form even when cohesin levels are reduced. But perhaps the cell-culture systems used or the particular subunit of cohesin deleted in these experiments left sufficient levels of chromosome-bound cohesin for TADs to form. Removing cohesin is inherently tricky: it is essential for proper cell division, so must be removed only in mature, non-dividing cells.
Despite the correlative evidence and models, there has been no conclusive proof that cohesin helps to fold chromosomes into functional domains. The fact that chromosome structure seems to be hard to disrupt is probably a good thing for the cell, but has been vexing for researchers.
Schwarzer et al. have finally succeeded in breaking TAD structure by generating mice whose liver cells lack the gene Nipbl. The NIPBL protein loads cohesin onto DNA, so chromosomal cohesin levels are drastically reduced in these cells. The authors' work provides the first convincing evidence that cohesin is essential to building TADs. The results also reveal key properties of those genome structures that are not directly attributable to cohesin. For instance, the researchers generated Hi-C maps to demonstrate that, even though TADs are lost in Nipbl mutants, A/B compartments are maintained. Thus, TADs and compartments are formed by independent mechanisms.
Strikingly, when cohesin was absent, some chromosome regions that had been in the same TAD split between A and B compartments. This segregation depended on the molecular modifications to local histone proteins, around which DNA is packaged (different histone modifications are associated with different levels of gene activation or suppression). These observations imply that cohesin is needed to help key regulatory sequences associate with target genes in the same TAD. In its absence, interacting sequences might become separated in the nucleus, causing a loss of communication between them (Fig. 1b). Consistent with this idea, the authors find that the expression of many genes is decreased in Nipbl-deficient cells, particularly when those genes are located in regions of the genome distant from regulatory sequences called enhancers, which promote the genes' expression. Meanwhile, transcription close to enhancers is increased, even if the transcribed regions do not code for protein.
Interestingly, as TADs fragment into different A/B compartments without cohesin, the structure of the mouse genome starts to resemble that of the fruit-fly genome. Fruit flies seem to lack TAD loops defined by cohesin and CTCF, and instead have a strong and frequently alternating A/B-compartment structure12. This suggests that similar forces govern genome organization across diverse organisms, and that differences result from the evolution of new processes that add to existing ones.
The current study complements other recently published work13, which demonstrated that removing CTCF also disrupts TADs without destroying compartments. But Schwarzer and colleagues' results demonstrate that CTCF alone cannot fold chromosomes normally. Taking these results together, a model emerges in which histone modifications promote spatial segregation of active and inactive chromosomal regions, but cohesin loop extrusion can counteract that force, pulling segments of chromosomes into the same TAD. The size of the TAD is limited by CTCF, which blocks further extrusion8,13, delineating TAD boundaries. These balancing forces ensure proper communication while preventing unwanted crosstalk between chromosomal regions.
This newly clarified role of cohesin in mammalian genome folding may shed light on cohesin-related conditions such as Cornelia de Lange syndrome, which affects the development of many body parts. The observation that some but not all layers of genome structure are disrupted by cohesin loss might help to explain the complex set of traits associated with such syndromes.
One question that remains is whether cohesin removal might have this marked effect only in certain cell types. Encouragingly, a recent study14 shows a similar effect when cohesin is removed in a different system: a human colon-cancer cell line. So it is likely that cohesin has a similar role across mammalian tissues.
Another important question is how cohesin performs its function: does it actually extrude loops through its ring? Much remains to be discovered, but researchers can now move forward with confidence that cohesin's role in genome folding is not only real, but also crucial to 3D genome structure.
Schwarzer, W. et al. Nature 551, 51–56 (2017).
Nora, E. P. et al. Nature 485, 381–385 (2012).
Dixon, J. R. et al. Nature 485, 376–380 (2012).
Lupiáñez, D. G. et al. Cell 161, 1012–1025 (2015).
Hnisz, D. et al. Science 351, 1454–1458 (2016).
Schmitt, A. D. et al. Cell Rep. 17, 2042–2059 (2016).
Merkenschlager, M. & Nora, E. P. Annu. Rev. Genom. Hum. Genet. 17, 17–43 (2016).
Fudenberg, G. et al. Cell Rep. 15, 2038–2049 (2016).
Zuin, J. et al. Proc. Natl Acad. Sci. USA 111, 996–1001 (2014).
Seitan, V. C. et al. Genome Res. 23, 2066–2077 (2013).
Sofueva, S. et al. EMBO J. 32, 3119–3129 (2013).
Rowley, M. J. et al. Mol. Cell 67, 837–852 e837 (2017).
Nora, E. P. et al. Cell 169, 930–944 (2017).
Rao, S. et al. Preprint available at bioRxiv http://doi.org/cdhr; Cell 171, 305–320 (2017).