Contribution of Topological Domains and Loop Formation to 3D Chromatin Organization

Abstract

Recent investigations on 3D chromatin folding revealed that the eukaryote genomes are both highly compartmentalized and extremely dynamic. This review presents the most recent advances in topological domains' organization of the eukaryote genomes and discusses the relationship to chromatin loop formation. CTCF protein appears as a central factor of these two organization levels having either a strong insulating role at TAD borders, or a weaker architectural role in chromatin loop formation. TAD borders directly impact on chromatin dynamics by restricting contacts within specific genomic portions thus confining chromatin loop formation within TADs. We discuss how sub-TAD chromatin dynamics, constrained into a recently described statistical helix conformation, can produce functional interactions by contact stabilization.

Keywords: CTCF; TAD borders; chromatin dynamics; chromatin loops; statistical helix; topological domains.

Figure 1

Schematic representation of genome organization in mammals. Between the nucleosomal scale (nucleofilament) and the nuclear scale (chromosome territories), 3D organization of the genome at the supranucleosomal scale has been recently explored thanks to 3C-derived methods (see Figure 2). Beyond the transcriptionally active (A) or inactive (B) chromosomal compartments, Topologically Associating Domains (TADs) and chromatin loops are two essential determinants of eukaryotic genome organization.

Figure 2

Principles of 3C-derived methods. All the methods derived from the Chromosome Conformation Capture (3C) protocol involve a formaldehyde cross-linking step followed by an enzymatic digestion and a ligation step. Each method uses different approaches to generate genomic libraries: secondary digestion for 3C-qPCR, circularization and inverse PCR for the 4C, “carbon copy” amplification for the 5C and biotinylation and purification on streptavidine beads for Hi-C. Ligation products are quantified by real-time qPCR in the first method or by high-throughput sequencing in the others (adapted from [14]).

Figure 3

Contact frequencies at the HoxD locus in ESC and NPC. Contact frequencies have been determined at the HoxD locus (mouse chromosome 2) in mouse ESC (blue circles) and neural progenitor cells (NPC, pink rectangles) using 3C-qPCR as described in [54]. The data have been aligned with the directivity index and the CTCF sites previously described for mouse ESC [15]. The viewpoint used in each 3C experiment is indicated by an anchor. Contact frequencies have been measured from two different viewpoints located in (a) the Hoxd4 gene in the most telomeric TAD (colored in red) or (b) the Lnp gene in the most centromeric TAD (colored in green). The horizontal lines represent the mean basal contact level (continuous lines) and the associated background noise (dashed lines) [55]. The position of the TAD border is delimited by thick dashed vertical lines (note that this TAD border is delimited by two conserved CTCF binding sites). Error bars correspond to standard error of the mean of five biological replicates each quantified at least in triplicate. Experimental points for which the extremely low contact frequency could not be quantified in all of the five replicates are depicted in grey. In these cases, the points represented in the figure involve an upper bound of the contact frequency corresponding to the detection limits of our qPCR assays: for these points, a Ct of 45 cycles was assigned to replicates that could not be detected by quantitative PCR because of extremely low contact frequencies. These data points were then analyzed according to the same procedure as all the others experimental points as previously described [55]. Mean contact frequencies (values above white horizontal arrows) were calculated over 20 kb on each side of the TAD border (grey rectangle on the right) or on each side of an equivalent region located on the centromeric side of the same viewpoint (grey rectangle on the left). The fold change (ratio) F between these mean contact frequencies is indicated together with the p-value of their difference (Student t-test).

Figure 4

Contact frequencies at the Ttll7 locus in ESC and NPC. Following the conventions described in Figure 3, contact frequencies have been determined at the Ttll7 gene locus (mouse chromosome 3) in mouse ESC (blue circles) and neural progenitor cells (NPC, pink circles) from (a) an intergenic viewpoint located in the most telomeric TAD (colored in red) or (b) from an intragenic (Ttll7 gene) viewpoint in the most centromeric TAD (colored in green). Error bars correspond to standard error of the mean of five biological replicates each quantified at least in triplicate. The most telomeric CTCF site appears as an interesting element delineating a putative border (dashed red vertical line). The fold change (F) between the mean contact frequencies on each side of this putative border and the p-value of their difference were determined as explained in the legend of Figure 3.

Figure 5

Two different kinds of borders: TAD borders and loop closures. The figure depicts a border separating two adjacent TADs along the genome, sketched as a blue box (Left), and a chromatin loop and its closure by stabilizing factors, sketched as a red box (Right). When performing a quantitative 3C experiment with the viewpoint represented by the anchor, TAD border plays a strong insulating role by preventing contacts between the viewpoint (anchor) and point A, as observed in Figure 3, thus demarcating two adjacent domains. In contrast, the formation of a chromatin loop, while locally increasing the contact frequency between the two stretches of DNA in the stem region (red rectangle), weakly lowers (or does not affect) the contact frequency between the anchor and point B. This weak-border signature may correspond to that seen in Figure 4.

Figure 6

From statistical helix to 2D representation of chromatin loops. As recently shown [36,54], spontaneous conformational fluctuations of the chromatin fiber are constrained at supranucleosomal scale, which produces a helical statistical shape. Specific factors stabilize statistical contacts (i.e., random collisions), turning them into stable interactions and creating 3D topological loops of chromatin [59]. 2D projection of such 3D rosette-like structures (as sketched by the curved arrow) leads to the current plane representation of chromatin loops [5,58]. Green and red arrows depict putative CTCF sites involved in chromatin loop formation and represent the directionality of these sites [5,48].