Skip to main content
. 2018 May 14;7:e34077. doi: 10.7554/eLife.34077

Figure 6. Alb 4C-seq exemplifies intra-TAD insulation and super-enhancer interaction.

(A) The Alb promoter makes multiple directional contacts with the adjacent super-enhancer region in both male and female mouse liver, as determined by 4C-seq with a viewpoint at the Alb promoter. All reproducible interactions occur within the TAD loop containing the Alb TSS and its super-enhancer (red bar beneath H3K27ac track), and all but two contacts in male liver occur within the predicted intra-TAD loops (pink). 4C-seq interaction scores are shown as –log10(pval) values across replicates, as calculated by R3C-seq (see Materials and methods). Also see Figure 6—figure supplement 1. (B) The 4C-seq interaction signal within the Alb TAD is orders of magnitude above the background signal and generally decays with distance. Far-cis and trans interactions are represented on a per TAD basis, expressed as RPKM per TAD, to control for sequencing depth and TAD length. The overall background within mouse chromosome five is significantly higher than all trans chromosomes; immediately adjacent TADs also show higher 4C-seq signal than the overall cis background. The 4C-seq signal decayed to background levels within ~3 TADs of the Alb viewpoint TAD. Each data point represents a single TAD and each color represents a 4C-seq replicate. (C) Background model used for distal cis interactions, showing a rapid decay in per TAD signal intensity. Each data point represents a single TAD along chromosome 5. (D) Distal cis and trans TADs that highly interact with the Alb promoter tend to be active TADs, while a majority of the TADs that interact less than the background model are predicted to be inactive. A simple inverse logarithmic decay of signal per TAD was used to determine the background signal along the cis chromosome, while the 4Cker package was used to determine high, medium, low, and non-interacting TADs in trans based on a hidden markov model with adaptive windows better suited for low signal regions.

Figure 6.

Figure 6—figure supplement 1. Alb 4C-seq replicates and cis/trans 4C-seq signal distribution.

Figure 6—figure supplement 1.

(A) UCSC genome browser screenshots for all biological replicates. The proximal view shows a 150 kb window around the Alb TSS, while the distal view shows a 400 kb view around the Alb TSS. For each sample, the colored track indicates normalized interactions for that region with the viewpoint (Alb promoter) in reads per million. The black bars below the colored signal track are interactions identified by the R3C-seq pipeline, expressed as a –10log of the p-value. (B) Alb promoter interactions are highly reproducible, with an average R2 of 0.87 between pairs of biological replicates between individual male (M) and female (F) mouse livers, as indicated on the x-axis. (C) Interactions originating from the Alb promoter viewpoint show no major apparent sex differences between male and female liver samples. This is visualized both in terms of strength of interaction (−10*log(pval), as above) as well as normalized read depth (reads per million, below). (D) The expression of protein-coding genes within non-, low, medium, or high interacting TADs in cis or trans, expressed as log2(FPKM +1) values. Genes in highly interacting TADs are more highly expressed compared to non-interacting TADs or the genome wide/cis control (p<0.001; KS t-test). This fits with the transcription factory model of compartment organization, as one would expect highly active regions to weakly associate in hubs of active transcription. Genes in TADs that are not interacting with the Alb promoter show the lowest expression of all groups. Shown for comparison is the expression of all genes on chromosome 5 (cis) and also genome-wide (trans). (E) Median eigenvalues within cis or trans non-, low, medium, or high interacting TADs. A positive value indicates a more open/active chromatin compartment (‘A type’), while a negative value indicates a more closed/inactive chromatin compartment (‘B type’). The highest interacting TADs show higher eigenvalues than the background or non-interacting TADs (p<0.001; KS test). Only trans non-interacting TADs show a consistent negative median eigenvalue, indicating a repressed compartment.