The Shh Topological Domain Facilitates the Action of Remote Enhancers by Reducing the Effects of Genomic Distances

Abstract

Gene expression often requires interaction between promoters and distant enhancers, which occur within the context of highly organized topologically associating domains (TADs). Using a series of engineered chromosomal rearrangements at the Shh locus, we carried out an extensive fine-scale characterization of the factors that govern the long-range regulatory interactions controlling Shh expression. We show that Shh enhancers act pervasively, yet not uniformly, throughout the TAD. Importantly, changing intra-TAD distances had no impact on Shh expression. In contrast, inversions disrupting the TAD altered global folding of the region and prevented regulatory contacts in a distance-dependent manner. Our data indicate that the Shh TAD promotes distance-independent contacts between distant regions that would otherwise interact only sporadically, enabling functional communication between them. In large genomes where genomic distances per se can limit regulatory interactions, this function of TADs could be as essential for gene expression as the formation of insulated neighborhoods.

Keywords: Shh; chromatin conformation; gene expression; long-distance enhancers; topological domains.

Graphical abstract

Figure 1

Topological and Regulatory Organization of the Shh Locus (A) Schematic representation of the Shh regulatory domain, as defined by the collection of 60 transposon insertions obtained with GROMIT. The location of representative insertions and their expression patterns is shown. Bars represent regulatory domains (En2-Cnpy1, Shh), as outlined by expression patterns reminiscent of the ones of the associated endogenous genes. White bars indicate that insertions in those regions have no expression. Orange arrowhead indicates the ZPA. (B) The Shh regulatory domain compared with the 3D conformation of the locus. Hi-C map of the locus from CH12 cells (Rao et al., 2014) (red contact maps, image generated with 3Dgenome browser, http://www.3dgenome.org) and TADs identified in ESCs (Dixon et al., 2012) (brown bars) are shown. Position and activity of insertions are indicated by colored lines (orange, Shh-like expression; blue, En2-like expression; black, no expression; gray, other/non-attributed expression). Corresponding regulatory domains are boxed. Shown beneath are 4C-interaction profiles (hit percentage with 10 and 100 count thresholds in light and dark green, respectively) of three viewpoints (Shh, Rnf32, ZRS, red arrowheads and lines) located in the regulatory domain and of two viewpoints (Rbm33, Nom1, blue arrowheads and lines) flanking it. For each viewpoint, we indicate the percentage of reads from regions in the Shh domain or from the 1 Mb flanking regions. (C and D) Cumulative 4C read counts as a function of distance from the Shh viewpoint (C) or the ZRS viewpoint (D). Data from different microdissected limb compartments is shown in different colors (fa, anterior forelimb; fm, medial forelimb; fp, posterior forelimb; h, hindlimb, 1 and 2 indicate biological replicates), the TAD/regulatory domain is highlighted in brown and the black bar indicates the constant slope of the curve.

Figure 2

Organization of the Shh Locus and Responsiveness to ZRS (A) 4C-seq interaction profile (read counts, binned in 11-fragment sliding windows) of the ZRS (viewpoint indicated by red triangle) in the anterior, medial, and posterior compartments of E11.5 forelimbs. A red bar underlines the peak contact region around Shh. For further comparisons, see also Figure S2. (B) ZRS-interaction values at the insertion points of the transposon (in the absence of the transposon). x axis, distance to the ZRS; y axis, 4C-interaction score; dot color represents intensity of LacZ staining in the ZPA. (C) Comparison of interaction scores with responsiveness to the ZRS for positions within the Shh TAD (not expressed versus strongly expressed in ZPA; p = 0.0018, two-sided Mann-Whitney test).

Figure 3

Changing Distances within the Shh TAD (A) Schematic representation of the region, including the different insertion points and loxP sites used. (B) Schematic representation of the rearranged alleles. The distance separating the ZRS (orange oval) from Shh is indicated. The transposon at the junction point (when retained) is indicated, and dashed rectangles mark the duplicated regions. The Z2D allele is a replacement of the ZRS by another limb enhancer (yellow oval, DachEn/hs126; Visel et al., 2007), which appeared to be essentially inactive when inserted at this position (Figure S3G). (C) Gene expression by RT-qPCR in DEL(C1-Z) versus WT E11 forelimb buds (for each gene, reference value in WT set as 1, the error bars correspond to SEM. Statistical significance done with t tests: ^∗p < 0.05; ^∗∗∗p < 0.001. (D) Forelimb skeleton of a DEL(Z-C2)/Shh^del mouse showing monodactyly and fused zeugopod. sc, scapula; hu, humerus; fz, fused zeugopod; vph, vestigial phalanges. (E) Hand skeletons of adult mice with different rearranged alleles. Alleles are in trans of either Shh^del (for DUP(5–8) and DUP(C1-Z)) or of a ZRS replacement (Z2D allele, for CTRL and DEL(5–8)), because DEL(5–8) homozygous or compound mutants with Shh^del die at birth due to holoprosencephaly and cranial defect (not shown). (F) Expression of Shh in E10.5 forelimbs in the different alleles. For each line, in situ hybridization was performed on wild-type control littermates and homozygous mutants (n = 3). (G–I) RT-qPCR data in DEL(5–8) (G), DUP(5–8) (H), and DUP(C1-Z) (I) E11 forelimb buds. Homozygous mutant samples are in red (n = 3), stage-matched wild-type samples from the same litters (n = 3) are used as control, except for (G), where wild-type samples include embryos from separate litters (the arrowheads indicate the expression level in wild-type littermates of the mutants). The error bars correspond to SEM. ^∗p < 0.05 (t test). (J–L) LacZ staining of E11.5 embryos with insertions of the sensor at positions 5.2, 8.2, and in the context of DEL(5–8) and DUP(5–8). On the schematic representation of the alleles, the Shh-ZRS TAD is in orange, and red and blue rectangles label the centromeric and telomeric flanking regions of 5.2, respectively. The sensor showed expression in the ZPA in 8.2 DUP(5–8) and DEL(5–8) embryos (orange arrowheads and insets) but not in 5.2 embryos. Expression domains observed in DUP(5–8) or DEL(5–8), but in none of the starting insertions, are labeled with pink and green arrowheads, respectively. For further gene expression and phenotypic data. See also Figure S3.

Figure 4

Consequences of TAD Disrupting Alleles (A) Representation of the insertions used to produce the inversions and del(-500-33). The Shh-ZRS TAD is in orange, the flanking ones in brown. (B) Representations of the rearranged alleles, with the inverted and deleted regions outlined by dashed green and red boxes, respectively. The linear distance between Shh and the ZRS is indicated. Dashed orange and brown blocs indicate the segment corresponding to former TADs. (C) Forelimb skeletons of E18.5 embryos for Shh^del/+ (control), INV(-500-C1)/Shh^del, and INV(6-C2)/Shh^del, with the latter two showing the typical Shh loss of function limb phenotype. sc, scapula; hu, humerus; ra, radius; ul, ulna; vzg, fused zeugopod; ^∗, single vestigial digit. (D) Expression of Shh in E10 mouse embryos by in situ hybridization. The orange arrowhead indicates expression in the ZPA, the purple one expression in the ventral midbrain (md). (E) LacZ expression in E11.5 autopods of embryos carrying the starting insertions in normal (C1, 6.1) or inverted configurations, INV(-500-C1), INV(6-C2). See also Figure S4.

Figure 5

4C Profiles in INV(6-C2) Alleles (A–D) For each viewpoint, the hit percent profiles (with 10 and 100 count thresholds in light and dark color, respectively) obtained from WT (green) and INV (red) samples are plotted on their respective genomic configurations (i.e., with an inversion of the [6-C2] genomic segment for INV). The inverted region is boxed, and the new position of the genes in the INV allele is depicted. The viewpoints are indicated by black arrowheads. To take into account the loss of Shh expression and monodactyly in INV(6-C2), we compared INV whole forelimbs with WT anterior forelimb compartments. (E) Comparison of the interaction profile of the ZRS between WT and INV in the inverted region (plotted with the same orientation). (F) Same comparison as in (E) for the interaction profile of Shh between WT and INV. The box delimits the intra-TAD segment not affected by the inversion and the percentage of counts contained within it. See also Figure S5.

Figure 6

Distance-Dependent Rescue of ZRS Activity on Shh (A–D) Schematic representation of the series of inversions generated from C2. Red, blue, and white circles indicate putative enhancer elements that are progressively moved away from Shh by the inversions. A green arrow identifies the end of the Shh TAD. (E–H) Skeletons of E18 embryos, including close-up views of the hindlimb (I–L). Scale bar, 2 mm. (E and I) Control embryo (Shh^del/+). (F and J) INV(6-C2)/Shh^del. (G and K) INV(4-C2)/INV(4-C2). (H and L) INV(2-C2)/INV(2-C2). md, mandibule; fl, forelimb; hl, hindlimb, ph, phalanges; mt, metatarsal bones; ts, tarsal bones; ti, tibia; fi, fibula; cdv, caudal vertebrae. Arrowheads and asterisks point to deformed structures (cyclopia (1-e), vtf, vestigial partially fused tibia-fibula; pb, proboscis replacing anterior head structures). Photo in (E) was assembled from two images of the same embryo using Adobe Photoshop Photomerge script. See also Figure S6.

Figure 7

TADs Organize Robustness and Specificity of Long-Distance Interactions (A) TADs show an internal structure that determines the propensity of a region to be contacted by another one, hence defining enhancer responsiveness. Shh gene, orange arrow; ZRS, orange oval; an enhancer cold-spot, brown star; a responsive spot, green hexagon. The region folds into three different TADs (blue, orange, and dark gray bars), each of which likely corresponds to a dynamic ensemble of 3D conformations (below) (Fudenberg and Mirny, 2012, Giorgetti et al., 2014). The light colored area represents the region effectively explored by an element (e.g., the ZRS), i.e., with sufficient contact frequency to elicit a transcriptional response. The cold spot is located outside this zone, whereas the responsive spot can come in proximity of the ZRS. (B) TADs contribute to long-distance regulatory interactions by favoring proximity between otherwise distant regions (the two colored ovals represent the regions explored by Shh and the ZRS, respectively; the extent of overlap indicates frequent interactions). Elements located in distinct TADs do not influence genes located in the adjacent ones, not necessarily because of active insulation but simply because of the absence of a mechanism compensating for the buffering effect of genomic distances. (C) Without TADs, contacts between distant regions are too rare to be functional or lead only to sporadic gene activation producing variable phenotypic outcomes. See also Figure S7.