ChIP-seq analysis of chromatin-binding proteins

Cells (40 × 10⁶) were fixed in 1% paraformaldehyde (PFA) for 10 min, and ChIP-seq was performed as described previously (Liang et al. 2013 (link); Jacinto et al. 2015 (link)). The following antibodies were used for ChIP: Nup98 (purchased from Cell Signaling Technologies, P671, no.2292), H3K4me3 (purchased from Abcam, ab8580), Set1A (purchased from Abcam, ab70378), and Flag (purchased from Sigma, Flag M2 affinity gel, no. A2220). Reads were aligned to the mouse genome (mm10 and GRCm38) (Figs. 1–5) or human genome (hg19 and GRCh37) (Supplemental Fig. S4D) using bwa (version 0.7.12) (Li and Durbin 2009 (link)). Only reads that aligned uniquely to a single genomic location (MAPQ > 10) were used for downstream analysis. ChIP-seq peaks and normalized bedGraph files were generated using HOMER using a false discovery rate of 0.1% and fold enrichment over input of at least fourfold (Heinz et al. 2010 (link)). Data used to characterize binding of genomic elements by RUNX1 and HOXB4 (Fig. 1C) were obtained through Gene Expression Omnibus from Fan et al. (2012) (link) and Wu et al. (2012) (link).