6.7 Molecular mechanisms driving genome folding

A strong focus of current research in the field of three-dimensional chromatin architecture is to understand the molecular mechanism behind the formation of chromatin loops and TADs. It is not clear how molecular interactions between proteins at the nanometer scale organize micron-long protein-bound DNA molecules and how TAD formation can lead to insulating or facilitating of interactions between specific genomic elements (Fudenberg et al. 2018). Enrichment of CTCF at TAD boundaries, the convergent orientation of CTCF motifs in chromatin loops, as well as, extensive polymer modeling result in the loop extrusion model (Section 1.4.6). According to this model, loops and TADs are formed by active extrusion of chromatin through a loop-extruding molecule, which was suggested to be cohesin. Thereby, barrier elements, such as CTCF can arrest the extrusion and lead to stable loops. Multiple further extrusion processes within such loops can explain TAD like structures in Hi-C (Fudenberg et al. 2016). It is however not entirely clear which molecular forces could drive this extrusion process. Interestingly, active transcription was suggested to be involved in chromatin organization and TAD formation (Ulianov et al. 2015; Rowley et al. 2017). However, transcription and passive diffusion of cohesin might are likely to slow (Barrington et al. 2017). Active translocations of cohesin by some motor forces are more likely to explain the somewhat rapid formation of loops by extrusion (Rao et al. 2017).

While cohesin and CTCF are the best studied architectural proteins, it is not clear which other factors might drive loop formation by extrusion or other processes. Using our loop prediction tool 7C, we could screen ChIP-seq datasets from 124 different TF to quantify the ability of the TF binding signals to predict chromatin looping events. Interestingly, we could confirm the implication of many known architectural proteins, such as CTCF, ZNF143, and the subunits of cohesin RAD21 and SMC3. These findings are largely consistent with recent perturbation studies that analyze the effect of cohesin or CTCF depletion or knock-out on genome folding. Removal of cohesin and deletion of the cohesin-loading factor Nipbl lead to global loss of loops and TADs (Rao et al. 2017; Schwarzer et al. 2017). Also, depletion of CTCF in embryonal stem cells leads to loss of loops and TADs and impacts transcriptional regulation (Nora et al. 2017). Interestingly, the formation of A/B compartments seems not to be affected by CTCF depletion (Nora et al. 2017) and even increased upon cohesin knock out (Schwarzer et al. 2017), indicating an independent mechanism for loop formation and compartmentalization. Since A/B compartments correlation strongly with epigenetic marks (Fortin and Hansen 2015; Di Pierro et al. 2017; Prakash and Fournier 2018), a mechanism similar to phase separation might drive compartmentalization. Also, imaging-based data from Drosophila suggest the formation of heterochromatin domains by phase separation (Strom et al. 2017). Polymer modeling and Hi-C data in rod photoreceptor neurons, which have inverted nuclei, confirm that heterochromatin drives the compartmental organization in both, inverted and conventional, nuclei (Falk et al. 2018).

Together, these studies point to an essential role of architectural proteins, such as CTCF and cohesin, in forming loops and domains, while clustering of epigenetic marks by a process similar to phase-separation might case the A/B compartment formation. Future studies are needed to confirm the role of other architectural proteins as well as the causal relationship between heterochromatin and compartmentalization.

Multidisciplinary studies combing imaging and genome-wide proximity-ligation methods with computational modeling are needed to better understand the complexity of three-dimensional genome folding and gene regulation. Large collaborative projects, such as the 4D Nucleosome projects (Dekker et al. 2017), will provide new standards and improved methods. These efforts, as well as many independent studies, are promising to improve our understanding of three-dimensional genome folding in different species as well as its regulation and dynamics during cell-cycle, evolution, in development and disease.