Fig 2: GR and SETD1A/COMPASS co-occupy genomic regions in macrophages(A) Venn diagram of GR (blue, n = 3), SETD1A (yellow, n = 2), and CXXC1 (brown, n = 2) ChIP-seq peak overlap in macrophages treated with Dex + LPS.(B) Differential Centrimo motif enrichment of indicated subsets over the total called 27,127 GR, SETD1A, and CXXC1 peaks, with Bonferroni-adjusted (adj) binomial p values (right).(C) Feature distribution over genomic locations of the different peak subsets (promoters defined as <1 kb from the TSS, intergenic defined as >1 kb from any gene).(D) KEGG pathway over-representation analysis for peak subsets (Benjamini-Hochberg adj hypergeometrical p value). Circle sizes represent gene ratios.

Fig 3: GR-mediated SETD1A/COMPASS recruitment to enhancers shows locus-specific histone methylation(A) Normalized SETD1A, CXXC1, H3K4me1/me2/me3, and H3K27 acetylation (H3K27ac) ChIP-seq signals in LPS- and Dex + LPS-treated macrophages, ±2 kb around the GBS. Loci are sorted by clusters identified in Figure S4C. Invariant: GBSs with unchanged SETD1A (p > 0.1, -1.5 < FC < 1.5) occupancy, 588 bp around the GBS. ? represents the coverage difference between Dex + LPS and LPS treatments.(B) Log2FC in SETD1A, CXXC1, H3K4me1/me2/me3, and H3K27ac ChIP-seq signals at intergenic regions. mRNA expression of the genes closest to the enhancers from cluster 1, cluster 2, and cluster 3 is displayed together with those showing invariant SETD1A occupancy (inv., p > 0.1, -1.5 < FC < 1.5; gray). Black lines indicate the distribution mean. (ChIP-seq: Kruskal-Wallis test with post hoc Dunn’s test; RNA-seq: one-way ANOVA with post hoc pairwise two-sided t test, Benjamini-Hochberg adj). n = 2.(C) Normalized median genome browser tracks for H3K4me1/me2/me3 and H3K27ac ChIP-seq (n = 2, LPS- (red) versus Dex + LPS-treated (blue) BMDMs) plus GR. Arrowheads point at treatment-dependent changes; lines below GR peaks indicate primer locations for (D).(D) Normalized spike-in ChIP-qPCR for selected loci; mean Z scores of total H3 normalized % inputs. Error bars represent standard deviation (n = 3, two-sided Wilcoxon-Mann-Whitney test).(E) Correlation plot of log2FC in SETD1A, CXXC1, H3K4me1/me2/me3, and H3K27ac at intergenic GBSs with differential SETD1A occupancy (-1.5 > FC > 1.5, p < 0.1) and the mRNA expression of the closest gene. Pearson correlation coefficients with p < 0.001 are displayed as numbers.(F) Log2FC in mRNA expression of genes with nearby GBS over changes in SETD1A occupancy and/or H3K27ac levels (+, gain; -, invariant or lost). Each dot reflects one enhancer. Red: mean and 95% confidence intervals (Kruskal-Wallis test with post hoc Dunn’s test, Benjamini-Hochberg adj). Enhancer numbers are in parentheses.

Fig 4: H3K4 methylation dynamics upon SETD1A depletion(A) Heatmaps for mean SETD1A, CXXC1, and H3K4me1/me2/me3 ChIP-seq at ±2 kb around GR-SETD1A common sites in wild-type (WT) and Setd1aDel/+ (Del) RAW264.7 cells treated with LPS or Dex + LPS. GBSs are sorted by descending SETD1A ChIP signal strength (n = 2–3). See legend (B).(B) Example genome browser tracks with median ChIP-seq signals for GR, H3K4me1/me2/me3 in WT and mutant RAW264.7 cells. Arrows point towards GR binding sites or changes observed at the TSS.(C) Violin plots for log2FC in SETD1A, CXXC1, and H3K4me1/me2/me3 ChIP-Seq signals at intergenic regions, comparing Dex + LPS to LPS. GBSs bind GR and SETD1A, while background (bck) sites show H3K4me1/me2/me3, but no GR, SETD1A, or CXXC1 peaks. Signal distributions for constant (light blue, dashed outline) and for GR-SETD1A sites with changed SETD1A binding (full blue, solid outline) are shown (LPS + Dex versus LPS, p < 0.05). Treatment effect was determined by one-sample Wilcoxon-Mann-Whitney test, genotype differences by two-sided Wilcoxon-Mann-Whitney test, and group differences by paired two-sample Wilcoxon-Mann-Whitney test.(D) Spike-in normalized H3K27ac ChIP qPCR in LPS- or Dex + LPS-treated RAW264.7 cells. Mean Z scores of total H3 normalized % inputs are plotted (n = 3). Error bars represent standard deviations. Kruskal-Wallis with post hoc Dunn’s test, with Benjamini-Hochberg adj p values (adj ps). For all graphs, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig 5: GR recruits SETD1A/COMPASS to chromatin in response to ligand(A) Heatmap of mean GR, SETD1A, and CXXC1 ChIP-seq coverage in Dex + LPS- and LPS-treated macrophages (n = 2). Each line is one GR-bound site ±2 kb. Binding sites are sorted by log2FC of SETD1A occupancy between Dex + LPS and LPS. ? represents the difference in normalized coverage between Dex + LPS and LPS.(B) Example genome browser tracks of ChIP-seq for GR, SETD1A, and CXXC1 in macrophages treated with LPS or Dex + LPS. GR occupancy is the filled area under the curve (blue) (n = 1). Lines are medians of two replicates. Arrows point toward signal changes.(C) Correlation of SETD1A and CXXC1 occupancy at GR-SETD1A-CXXC1 co-bound enhancers; scatterplot of log2FC of CXXC1 and SETD1A. The dashed regression line includes the 95% confidence interval (gray shadow). ****p < 0.0001. RS, Spearman correlation coefficient.(D) Changes in SETD1A occupancy (log2FCSETD1A) in response to Dex + LPS over LPS. GBSs with increased SETD1A occupancy (FC > 1.5, p < 0.1) are purple; GBSs with reduced SETD1A binding (FC < -1.5) are green. Selected enhancers are labeled.(E) MEME motif enrichment of sites recruiting SETD1A (p < 0.1, FC > 1.5), with E-values and numbers of positive sites.(F) GO analysis for biological processes of the closest genes, with examples and the -log10 of the hypergeometric false discovery rate (FDR).