Fig. 5: Selective binding of Rel and p50 in mouse BMDMs examined by ChIP–seq.

a, ChIP–seq datasets obtained with antibodies against Rel and p50 in BMDMs stimulated with lipid A for 0 and 1 h were analyzed. The 2,891 peaks with the strongest peak scores obtained with each antibody at the 1-h time point were selected and merged, yielding 4,414 peaks (the peak number was smaller than in the RelA–Rel comparison because of the small number of peaks obtained with p50 antibody, probably owing to relatively poor antibody quality). Rel/p50 RPKM ratios were then calculated at each peak and are plotted on the y axis, with the 4,414 peaks along the x axis (displayed as a percentage of total peaks). The top and bottom 300 peaks based on ratio (7%) were then selected for motif analysis. On the right, numbers of peaks with different ranges of RPKM ratios are shown. b, IGV tracks are shown for ChIP–seq peaks in the promoter regions of the Nos2 and Stard4 genes. The Nos2 peaks exhibited moderately preferential p50 binding in comparison with Rel and RelA, and the Stard4 gene exhibited strongly preferential p50 binding. The RPKM scales are shown to the left, and gene annotations and annotated peak numbers are at the bottom. c, The top two most enriched motifs from de novo motif analyses performed with HOMER are shown for three different groups of peaks. Motif analysis was performed with 300 peaks (7%) representing the largest peak ratios, 300 peaks representing the smallest peak ratios and 300 peaks from the middle of the ratio distribution. d, The 4,414 peaks were divided into 15 bins of equal size across the spectrum of Rel/p50 RPKM ratios. The enrichment of the three consensus NF-κB motifs in HOMER, as well as one of the p50-preferential motifs and one of the Rel-preferential motifs identified in b, was then examined across the spectrum. NF-κB half-sites (not shown) were enriched equally across the spectrum, consistent with the presence of half-sites in all of the full consensus motifs examined.