The cohesin complex — best known for its role in cell division — does not rest between divisions, and instead participates in regulating gene expression. How it does this is only now becoming clear.
Cohesin is one of the large, ring-shaped protein complexes that constitute a substantial fraction of the chromosomal proteins in nucleated cells. It acts as the chromosomal 'glue', and is thought to encircle and so trap pairs of replicated chromosomes, known as sister chromatids. This allows recognition of these chromatids for segregation during the nuclear-division phase of the cell cycle1. But accumulating evidence suggests that cohesin is also involved in regulating gene expression, and that mutations of human cohesin or its regulators cause severe developmental disorders, including Cornelia de Lange syndrome and Roberts syndrome. Six studies2,3,4,5,6,7, including one by Wendt et al.4 on page 796 of this issue, provide insight into how cohesin contributes to gene regulation.
The first indication that cohesin does more than pairing sister chromatids came from gene-expression studies8 in the budding yeast Saccharomyces cerevisiae. Cohesin was identified as part of a chromosomal boundary — which consists of molecular elements that separate a region of gene repression from one of active expression. In the fruitfly Drosophila, both cohesin and a protein complex that loads cohesin rings onto chromosomes, Nipped-B, have been implicated9 in gene regulation. Active gene expression often depends on enhancer sequences that are located some distance away from their target genes along the chromosome. In such a setting, cohesin prevents enhancers from activating genes, an effect known as insulation, and Nipped-B alleviates this insulation. Although suggestive, these studies left open the question of whether cohesin directly functions as an insulator, or whether its role in sister-chromatid cohesion has indirect knock-on effects on gene expression.
Two teams2,3 now show that Drosophila neurons — cells that no longer proliferate and so have no sister chromatids in their nuclei — contain cohesin. They also find that inactivation of cohesin specifically in a certain type of neuron in the developing Drosophila brain has drastic consequences: expression of the ecdysone hormone receptor is reduced; the neurons fail to eliminate unwanted neuronal projections; and larval development is aborted. These observations clearly indicate that cohesin acts independently of sister-chromatid cohesion. Cohesin is also found in mouse neurons and many other non-dividing cell types4, opening up the possibility that, in many cases, it contributes to gene regulation.
To determine how cohesin controls gene expression, important questions include where along chromosomes cohesin binds, and what we can learn from its binding pattern. For analysis of such protein–DNA interactions, the technique of chromatin immunoprecipitation (ChIP) is used. The protocol involves chemically crosslinking cohesin to chromosomal DNA, followed by preparing chromatin from the cells, shearing it into small fragments, and purifying cohesin from the mixture. Because of the crosslinking step, the purified cohesin also contains DNA sequences that it had bound to. The DNA-binding sites of cohesin can then be identified by hybridization to DNA microarrays.
First performed10 in the budding yeast, ChIP analysis revealed that cohesin is loaded onto chromosomes at sites occupied by its loader protein (known as Scc2/4 in S. cerevisiae), but afterwards translocates away from the loading sites and accumulates at sites where genes transcribed in opposite directions converge (Fig. 1a). When they reach this final destination, cohesin rings seem to be irreversibly trapped on DNA, as is necessary for stable sister-chromatid cohesion. Similar analyses in Drosophila (Fig. 1b) have yielded6 a seemingly different result, whereby most cohesin remains close to its loader. Thus Drosophila cohesin could be remobilized after loading, when insulation of genes from their enhancers is established or eliminated — processes that do not occur to the same extent in the yeast.
The latest ChIP analyses4,5,7 of mouse and human cohesin reveal another exciting pattern (Fig. 1c). Although the relationship between cohesin and its loader was not investigated in these studies, a striking co-localization between cohesin and a well-known gene regulator, CTCF, is reported. CTCF is a chromosomal-boundary and insulator element, and its close co-localization with cohesin suggests that cohesin might play a part in this protein's function. Indeed, these studies4,5,7 show that CTCF recruits cohesin, and that, in several instances, cohesin is required for CTCF's insulator function.
How proteins such as CTCF set up chromosome boundaries and insulate enhancers from their genes is still largely mysterious. But now that cohesin has firmly taken the stage, new possibilities emerge. If cohesin rings can embrace two sister chromatids, they might also be able to link a CTCF-binding site to other sequences, or to a second CTCF-binding site, some distance away along the same chromosome. This idea, put forward by Wendt et al.4, is aptly illustrated in Figure 2. Boundary and insulator elements are known to interact with each other, forming clusters in the nucleus, and CTCF contributes to this clustering11. Whether cohesin is indeed responsible for tethering CTCF-binding sites can now be tested. Less clear is how clustering of these sites creates domain boundaries and insulators. Nevertheless, cohesin's role as an insulator could explain its contribution to gene regulation and human developmental disorders.
The next question, then, is how its ability to tether CTCF sites relates to cohesin's role in sister-chromatid cohesion. Most human cohesin complexes dynamically associate with chromosomes12, which could aid the establishment and disassembly of gene-regulatory interactions. After sister-chromatid formation, a part of cohesin becomes stably trapped on chromosomes, probably to provide enduring cohesion between sister chromatids. Are the same sites used for both gene regulation and sister-chromatid cohesion? And would stable binding at such sites be compatible with the possible need for changes in the gene-expression programme? Did the genome-wide analyses4,5,6 detect all sites of cohesin association, or could more cohesin be spread out over large intergenic regions? Although we have gained exciting insight into how architectural components of chromosomes contribute to gene regulation, further fascinating questions remain to be addressed.
Nasmyth, K. & Schleiffer, A. Phil. Trans. R Soc. B 359, 99–108 (2004).
Pauli, A. et al. Dev. Cell 14, 239–251 (2008).| Article |
Schuldiner, O. et al. Dev. Cell 14, 227–238 (2008).| Article |
Wendt, K. S. et al. Nature 451, 796–801 (2008).
Parelho, V. et al. Cell 132, 422–433 (2008).
Misulovin, Z. et al. Chromosoma 117, 89–102 (2008).
Stedman, W. et al. EMBO J. doi:10.1038/emboj.2008.1 (2008).
Donze, D., Adams, C. R., Rine, J. & Kamakaka, R. T. Genes Dev. 13, 698–708 (1999).
Rollins, R. A., Korom, M., Aulner, N., Martens, A. & Dorsett, D. Mol. Cell. Biol. 24, 3100–3111 (2004).
Lengronne, A. et al. Nature 430, 573–578 (2004).
Wallace, J. A. & Felsenfeld, G. Curr. Opin. Genet. Dev. 17, 400–407 (2007).
Gerlich, D., Koch, B., Dupeux, F., Peters, J.-M. & Ellenberg, J. Curr. Biol. 16, 1571–1578 (2006).
About this article
Seminars in Immunopathology (2020)
Cohesin gene defects may impair sister chromatid alignment and genome stability in Arabidopsis thaliana
EMBO reports (2008)