Figure 1.TT‐seq analysis of immediate response to T‐cell stimulation
A. Experimental design. RNA in cells was labeled with 4‐thiouridine (4sU) for consecutive 5‐min intervals. Total and 4sU‐labeled RNA was extracted before T‐cell stimulation and 5, 10, and 15 min after T‐cell stimulation and subjected to deep‐sequencing.
B, C Exemplary genome browser views for (B) an upregulated mRNA (JUN) and (C) a downregulated mRNA (NKX3‐1). Blue coverage: TT‐seq data for 0, 5, 10, and 15 min after stimulation; gray coverage: total RNA‐seq data for 0, 5, 10 and 15 min after stimulation. TSS: transcription start site, TTS: transcription termination site, pA: poly‐adenylation site.
Segmentation workflow. The Watson and Crick strands are in dark blue and green, respectively. The top 8 tracks show antisense‐corrected TT‐seq data tracks (log2 scale) that were used as input for GenoSTAN. The other tracks indicate the stepwise annotation of transcripts. From the GENCODE annotation, only full transcripts with transcript_support_level 1 are depicted.
Jaccard index (compared to GENCODE annotation) for different choices of thresholds (x‐axis: reads per kilobase (RPK)). The red line indicates the selected RPK value where the Jaccard index reaches the maximal value.
Number of transcripts per transcript class.
Distribution of transcript lengths per transcript class. Box limits are the first and third quartiles, the band inside the box is the median. The ends of the whiskers extend the box by 1.5 times the interquartile range.
Figure 2.TT‐seq captures transcriptional changes after T‐cell stimulation
TT‐seq signal for statistically significant differentially expressed transcripts after 5, 10, and 15 min compared to time point 0 min, before cell stimulation. Significantly differentially expressed transcripts are indicated by black points. The numbers in the plots correspond to the numbers of significantly up/downregulated transcripts at each time point. The vertical orange lines indicate the fold change cutoff of 2, and the horizontal orange line shows the P‐value cutoff of 0.05. P‐values were derived via DESeq2 (Wald‐Test).
Sense and antisense transcription at the FOS gene and its annotated upstream enhancer. The rectangles with arrows indicate transcripts annotated in this study.
Figure EV3.Example of eRNA identification using GenoSTAN
Shown are the DNAse hypersensitivity signal (DHS) peaks from ENCODE (top), the GenoSTAN enhancer states (violet colored) obtained from 14 Roadmap Epigenomics T‐cell lines (middle), and the obtained transcript annotation (bottom) for a region on chromosome 11.
Distribution of identified eRNAs among ncRNAs annotated based on TT‐seq signal. The outer circle segments show the number of actively transcribed eRNAs (RPK ≥ 16.5) at each time point.
Length distribution of eRNAs and other ncRNAs. The P‐value was derived by two‐sided Mann–Whitney U‐test. Box limits are the first and third quartiles, the band inside the box is the median. The ends of the whiskers extend the box by 1.5 times the interquartile range.
Half‐life distribution of eRNAs and other ncRNAs. The P‐value was derived by two‐sided Mann–Whitney U‐test. Box limits are the first and third quartiles, the band inside the box is the median. The ends of the whiskers extend the box by 1.5 times the interquartile range.
Average DNase hypersensitivity sites (DHS) signal at the TSS of eRNAs and other ncRNAs.
Motif enrichment in the 250 bp upstream sequences of eRNA versus other ncRNAs. Displayed are only motifs of upregulated TFs with odds ratio > 1.2 and P‐value < 0.05. The stars indicate TFs with statistical significant enrichment upon eRNAs after multiple testing correction (Benjamini–Hochberg method, FDR < 0.05).
Figure 4.Pairings of transcribed enhancers with promoters
Schematic of transcribed enhancer–promoter pairing based on CTCF‐insulated neighborhoods.
Distance distribution between eRNA and mRNA TSS. The lower histogram depicts the full distance range in 100‐kb steps. The upper histogram shows a zoom‐in of the region [TSS − 100 kb, TSS + 100 kb] in 2‐kb steps. The position of the paired mRNA is indicated together with its median length.
Number of transcribed enhancers per paired promoter and promoters per paired transcribed enhancer.
Correlation of TT‐seq signal over time between proximal (left, dark violet) or distal (right, light violet) transcribed enhancers and promoters by change in promoter TT‐seq signal (from left to right: downregulated, unchanged, upregulated promoters). The Pearson correlation coefficient was calculated between read counts across the time series (replicates averaged per time point) for each transcribed enhancer–promoter pair. The P‐values were derived by two‐sided Mann–Whitney U‐tests. Box limits are the first and third quartiles, the band inside the box is the median. The ends of the whiskers extend the box by 1.5 times the interquartile range.
Figure EV4.Details on transcribed enhancer–promoter pairs
Number of transcribed enhancers and transcribed promoters per ChIA‐PET‐defined insulated neighborhood. Count values were jittered for visualization purposes.
Location and orientation of transcribed enhancers with respect to their paired target gene TSS. Negative values indicate the transcribed enhancer location upstream of the promoter TSS, positive values downstream of the promoter TSS. Upper histogram: Distance distribution for transcribed enhancers on the same (“sense”) strand as their target promoter. Lower histogram: Distance distribution for transcribed enhancers on the opposite, antisense strand.
Distribution of the Pearson correlation between observed TT‐seq signal at transcribed enhancers and promoters for the enhancer–promoter pairs derived here (where both eRNA and mRNA change significantly between 0 min and 15 min time points; red line) and for 1,000 randomly shuffled enhancer–promoter pairs (gray lines). The colored profile indicates the quantiles (5–95% of the data) with the black median line and the gray 25% and 75% quantile lines. Observed correlations are enriched for positive correlations (right peak) and depleted for negative correlations (left peak).
Distribution of insulated neighborhood size by cohesin‐ChIA‐PET and CTCF‐ChIP‐seq. The median size of an insulated neighborhood was 255 kb.
Distance of transcribed enhancer–promoter pairs versus the size of the corresponding insulated neighborhood. The lines indicate the size of the insulated neighborhood (by which the distance is bound by definition), and a third of the size of the insulated neighborhood (which is the expected distance between two randomly drawn positions in a neighborhood). The P‐value was derived by a one‐sided Mann–Whitney U‐test. For visualization purposes, 127 points with insulated neighborhood size > 2,000 kb are not shown.
Figure 5.Temporal changes in enhancer and promoter transcription
Development of TT‐seq signal over time after T‐cell stimulation for paired promoters and enhancers (n =131) that are both significantly upregulated (FC ≥ 2, FDR ≤ 0.05) 15 min after stimulation (over the whole eRNA/the first 2,200 bp of the mRNA). The y‐axis shows the normalized read counts over the whole transcribed enhancer region (violet) and the first 2,200 bp (average length of eRNA) of the paired mRNA (red). The black line indicates the median.
As in panel (A) but using RNA‐seq read counts.
TT‐seq signal change as in panel (A) but for paired promoters and enhancers (n = 128) that are both significantly downregulated (FC ≤ 1/2, FDR ≤ 0.05) 15 min after stimulation.
As in panel (C) but using RNA‐seq read counts.
Data information: Box limits are the first and third quartiles, the band inside the box is the median. The ends of the whiskers extend the box by 1.5 times the interquartile range.