DUX4 Binding to Retroelements Creates Promoters That Are Active in FSHD Muscle and Testis
Figure 3
Some DUX4-bound repetitive elements are transcriptionally activated by DUX4.
(A) ∼1% of DUX4-bound regions show statistically-significant activation in response to DUX4 in our conservative analysis. We show normalized RNA-seq read counts within an arbitrary distance of 1 kb from DUX4-bound regions (peaks), comparing counts averaged over two myoblast lines that do not express DUX4 (x-axis) to counts averaged over two myoblast lines transduced with lentiviral DUX4 (y-axis). Only regions with at least 10 reads summed over all four samples are shown, and counts are plotted on a log scale. The 738 regions shown as green points show statistically significant activation in response to DUX4 (≥2-fold activation, FDR-adjusted p-value≤0.1) and the 6 orange points show significant repression (same thresholds). Regions surrounding DUX4-bound repeats are shown as “+” symbols, with dots representing DUX4-bound unique regions. Some regions (“x” symbols) had normalized counts of 0 in one condition and are plotted at an arbitrary low value so they appear on a log-scale. (B) DUX4-bound regions with more predicted DUX4 binding motifs are more likely to be transcriptionally activated. The y-axis value gives the percentage of regions that are transcriptionally activated by DUX4 (≥2-fold, FDR-adjusted p-value≤0.1). (C) DUX4-bound regions with more predicted DUX4 binding motifs have greater DUX4 occupancy, using ChIP-seq peak height as a proxy for DUX4 occupancy (shown on a log-scale).