Fig. 3

NRF1 and endogenous retroviruses establish active enhancers in the absence of DNA methylation or repressive histone modifications. a Left: Pie chart depicting the number of NRF1 motifs at C1 and C2 STARR-seq peaks. Right: Boxplots of CpG methylation at C1 and C2 STARR-seq peaks with (dark color) or without (light color) NRF1 motif. CpG methylation is significantly higher at C2 STARR-seq peaks that harbor a NRF1 motif, compared to those without a NRF1 motif. ***p < 2e−16, Wilcoxon rank-sum test [PD]. b NRF1 ChIP-seq occupancy at the n = 562 C2 STARR-peaks in WT and DNMT TKO ESCs cultured in SL. Top: Barplot that depicts that 61 of the 562 C2 STARR-peaks significantly gain NRF1 occupancy (p < 0.05, DESeq2, Wald test) in DNMT TKO cells. Bottom: Scatterplot that depict NRF1 occupancy in WT and DNMT3 TKO ESCs cultured in SL. NRF1-bound loci with differential occupancy are shown in blue. Four selected examples (red) are shown in c [PD]. c CpG methylation, STARR-seq, and NRF1 occupancy for the four example regions highlighted in b. Top: MethylC tracks [58] with blue bars depicting the fraction of methylated CpGs and black dots denoting the coverage depth. Purple: STARR-seq enrichment over input in 2iL and SL-ESCs. Middle: NRF1 occupancy in WT and DNMT TKO ESCs cultured in SL (RPKM). Bottom: NRF1 motif locations and CpG-methylation for the annotated CG-dinucleotides in 2iL and SL. [PD]. d DNA motifs enriched (pink) or depleted (blue) at C2 STARR-seq peaks that overlap a H3K9me3 peak relative to C2 STARR-seq peaks that do not overlap a H3K9me3 peak. Enriched motifs frequently co-occur (see e). e Co-occurrence of TF motifs enriched in d. P values were derived by the chi-squared test and corrected for testing multiple motif pairs (Benjamini-Hochberg). f STARR-seq peaks overlapping different ERV categories. The IAP-I ERV retroviruses overlap 60% of the C2-loci where the TBX20, NRF1, and KLF/SP (TNK) motifs co-occur. ***p < 2e−16, hypergeometric test. g Examples of IAPs with a STARR-seq peak (yellow highlight) covered by H3.3, TRIM28, and H3K9me3. STARR-seq tracks depict the enrichment over input, all other tracks depict RPKM [PD]. Some of the panels in these figures contain public data. These panels are annotated with [PD]. The accession numbers of public data and their corresponding panels are annotated in Additional file 2: Table S1