Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 503 (7475), 290-4

A High-Resolution Map of the Three-Dimensional Chromatin Interactome in Human Cells


A High-Resolution Map of the Three-Dimensional Chromatin Interactome in Human Cells

Fulai Jin et al. Nature.


A large number of cis-regulatory sequences have been annotated in the human genome, but defining their target genes remains a challenge. One strategy is to identify the long-range looping interactions at these elements with the use of chromosome conformation capture (3C)-based techniques. However, previous studies lack either the resolution or coverage to permit a whole-genome, unbiased view of chromatin interactions. Here we report a comprehensive chromatin interaction map generated in human fibroblasts using a genome-wide 3C analysis method (Hi-C). We determined over one million long-range chromatin interactions at 5-10-kb resolution, and uncovered general principles of chromatin organization at different types of genomic features. We also characterized the dynamics of promoter-enhancer contacts after TNF-α signalling in these cells. Unexpectedly, we found that TNF-α-responsive enhancers are already in contact with their target promoters before signalling. Such pre-existing chromatin looping, which also exists in other cell types with different extracellular signalling, is a strong predictor of gene induction. Our observations suggest that the three-dimensional chromatin landscape, once established in a particular cell type, is relatively stable and could influence the selection or activation of target genes by a ubiquitous transcription activator in a cell-specific manner.


Figure 1
Figure 1. Fine mapping of chromatin interactions in IMR90 cells
a, An illustration of the Hi-C data analysis procedure to identify regions interacting with a selected genomic region, such as the CCL2 locus as highlighted in yellow (Supplementary Method). b, Genome browser shot of the CCL2 locus showing the results from Hi-C, ChIP-seq and Gro-seq experiments. Each bar in the top 2 tracks are either Hi-C reads count (dark brown) or expected frequency (light brown) from a fragment to CCL2 locus (highlighted in yellow and orange filled box). Black filled boxes are regions interacting with CCL2 called by the peak calling algorithm same as the black filled boxes in (a). Light blue shadows highlight the enhancer/CTCF locations from ChIP-seq data. CCL2 is induced by TNF-α (shown in the GRO-Seq tracks). c, Validation of the DNA looping interactions with CCL2 using 3C assays. Yellow: anchor fragments in Hi-C or 3C (with star). Error bar: s.d. from 3 PCR replicates.
Figure 2
Figure 2. Characterization of the IMR90 chromatin interactome
a, Histogram showing the size distribution of all HindIII restriction fragments in human genome. The red curve plots the size distribution of all anchors used in this study. b, Size distribution of the genomic regions identified as interacting with all anchors. c, Distribution of genomic spans of all identified chromatin interactions. Histogram: frequency distribution; Red curve: cumulative fraction with increasing distances. d, Fraction of chromatin looping interactions that fall within the same topological domains is plotted as a function of genomic distance between the two interacting loci (red curve). Black curve: expected fraction calculated from random shuffling locations of topological domains (100 iterations, dashed error lines: s.d.). e, Average number of peaks identified for anchors with different cis-elements. f, Percentages of all interactions involving various types of cis-elements (either anchor or target peak has the elements). In e–f, Z-scores were calculated comparing the actual values to simulation by randomly shuffling the locations of cis-elements (100 iterations, two-side Z-test). * Z > 50, ** Z > 100, *** Z > 150. g, Box plot showing the distance distribution from different types of anchors to their targets. Median distances are also labeled. t-statistics are computed comparing log-transformed distance between each type of anchors to all anchors as control (dash horizontal line). * t > 20, ** t > 40, *** t > 80 (two-side t-test). h, Preferential interactions between different types of cis-elements. Heatmap shows the fold enrichment of different type of pair-wise combinations. p-values are computed using hypergeometric test and denoted in each cell.
Figure 3
Figure 3. Identification and characterization of promoter-enhancer interactions in IMR90 cells
a, A schematic of promoter tethered regions (PTRs). b, Distribution of distances between the promoters and enhancers found within the PTRs. c, Bar charts show the degrees of interactions. Left: Percentage of enhancers that are looped to active promoters found in PTRs. The first value of μ is the mean degree of promoter interactions for all enhancers; the second value is the mean degree of promoter interactions for enhancers interacting with at least one promoter. Right: distribution of degrees of enhancer interactions for active promoters. d. Genes sharing enhancers are co-regulated. Among the target genes of TNF-α responsive hub enhancers, proportion of co-induced gene pairs (> 2-fold) are plotted and compared to random simulation (100 iterations, two-side Z-test, * p < 0.05, ** p < 0.01, error bar: s.d.) e, Genome browser snapshots showing the virtual 3C plots of the CKAP2L and IL1A promoters. f, Compare the induction of different gene groups based on the location of p65 induced enhancers.
Figure 4
Figure 4. The higher order chromatin structure in IMR90 cells is stable during transient TNF-α signaling
a, A scatter plot comparing reads count at PTRs before and after TNF-α treatment. Grey dots are data for PTRs involving all enhancers, red or blue dots are PTRs involving the top 500 induced or repressed enhancers, respectively. r: Pearson’s Correlation calculated from all data points. b, Scatter plots compare normalized contact frequencies (Supplementary Methods) of all enhancers before and after 1 hr TNF-α stimulation. c, Scatter plots compare contact frequencies of enhancers in IMR90 and hESC cells. Colored points represent the top 2,000 hESC- (red) or IMR90-specific (blue) enhancers defined by H3K27ac mark. d, The relative change of contact frequency (comparing to untreated IMR90 cells) at hESC-specific or TNF-α induced enhancers are plotted in line graphs. e, Box plots of H3K4me1 and H3K27ac signals on enhancers that show increased or decreased H3K27ac signals after TNF-α treatment. f, Boxplots of H3K4me1 and H3K27ac ChIP-seq signals at the IMR90 or hESC specific enhancers. g, Bar charts showing that NF-κB binding sites within PTRs are more likely to activate target genes than those outside PTRs. In this figure, PTRs are identified using the Hi-C data from untreated IMR90 cells. The p-values are calculated using hypergeometric test.

Similar articles

See all similar articles

Cited by 455 PubMed Central articles

See all "Cited by" articles


    1. Bernstein BE, et al. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489:57–74. - PMC - PubMed
    1. Maurano MT, et al. Systematic localization of common disease-associated variation in regulatory DNA. Science (New York, N.Y. 2012;337:1190–1195. - PMC - PubMed
    1. Smallwood A, Ren B. Genome organization and long-range regulation of gene expression by enhancers. Current opinion in cell biology. 2013;25:387–394. - PMC - PubMed
    1. Dekker J, Marti-Renom MA, Mirny LA. Exploring the three-dimensional organization of genomes: interpreting chromatin interaction data. Nature reviews. 2013;14:390–403. - PMC - PubMed
    1. Lieberman-Aiden E, et al. Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science (New York, N.Y. 2009;326:289–293. - PMC - PubMed

Publication types

Associated data