The formation of chromatin domains: a new model

In spite of the recent advances in the field of chromatin architecture1,2, the formation mechanism of chromatin domains, TADs, the topologically associating domains, and LADs, the lamina associated domains, is still an open problem. While previous models only dealt with TADs and essentially relied on the architectural proteins CTCF and cohesin, the model presented here concerns both TADs and LADs and is primarily based on the corresponding DNA sequences, the GC-rich and GC-poor isochores, more specifically on their newly discovered 3-D structures. Indeed, the compositionally homogeneous GC-poor isochores were shown to be locally stiff because of the presence of interspersed oligo- Adenines4,5, whereas the compositionally heterogeneous GC-rich isochores were found to be peak-shaped and characterized by increasing gradients of GC and of interspersed oligo- Guanines. In LADs, oligo-Adenines induce local nucleosome depletions4,5 that are responsible for a wavy structure well adapted for interaction with the lamina. In TADs, the increasing GC levels and increasing oligo-Guanines of the isochore peaks are responsible for a decreasing nucleosome density5,6, a decreasing supercoiling7 and an increasing accessibility8. These factors mould the loops of “primary TADs”, that lack self-interactions since they are CTCF/cohesin-free, yet transcriptionally functional structures9-11. This “moulding step” is followed by a second step, in which the cohesin rings bind to the tips of the “primary TADs” and slide down the loops. This process is very likely due to Scc2/Nipbl, an essential factor not only for loading cohesin, but also for stimulating its translocation12 and its ATPase activity13. This “sliding step” creates self-interactions in the loops and stops at the CTCF binding sites located at the base of the loops that are thus closed and insulated.