Hierarchy of transcription factor gene expression during ES cell differentiation. (A) Experimental strategy. Embryoid bodies were cultured for up to 3.75 days, and epiblast-like cells expressing neither marker (Bry⁻Flk1⁻), mesodermal cells expressing GFP only (Bry⁺Flk1⁻), and hemangioblast-enriched cell population expressing GFP and Flk-1 (Bry⁺Flk1⁺) were purified using fluorescence-activated cell sorting (FACS; supplemental Figure 1A). To obtain hematopoietic cells, we further cultured the (Bry⁺Flk1⁺) cells for 4 days under conditions supporting the generation of blast colonies and from these cultures isolated CD41⁺ cells, which have been shown to represent the first fully committed hematopoietic cells.³⁹ (B) Timing of mRNA expression of hematopoietic regulator genes (Runx1, Pu.1, and Csf1r) in differentiating ES cells. ES indicates ES cells; blast, 4-day blast cell culture derived from +/+ cells; LSK, Lin⁻Sca-1⁺c-kit^hi cells; CMP, common myeloid precursor cell; and Mac, primary macrophages. Numbers in panel B represent mean values of 2 independent experiments analyzed in duplicate.

Chromatin at Pu.1 cis-regulatory elements unfolds at the onset of hemangioblast formation and before the Csf1r promoter. Assay examining relative DNaseI accessibility of the Csf1r promoter (A) and the Pu.1 promoter (B,C) in ES cells, ES-derived cells, 3T3 fibroblast cells, CD41⁺ cells sorted from day-4 blast cell cultures, and macrophages (Mac). (C) LM-PCR looking at Pu.1 with a primer set hybridizing at greater distance from the transcription start. Naked DNA (DNA) served as control and sequences were annotated by an LM-PCR amplifying genomic DNA modified by a G-specific Maxam-Gilbert G reaction (G). For each experiment, equal DNaseI digestion of samples was confirmed by amplification with primers specific for rDNA genes. Regions of localized increases/decreases in DNaseI accessibility at Csf1r and Pu.1 are indicated as gray or white bars, respectively. Signal intensities were measured across the indicated regions and were calculated relative to the rDNA control, with ES cells set as 1. The nature of transcription factor binding sites and their position are indicated at the left. Note that chromatin opening was established in the absence of gene expression. This is further illustrated by the absence of increased DNaseI accessibility in the coding region compared with macrophages (C).

Selective DNA demethylation at the Pu.1 promoter. DNA from the indicated cell types was subjected to bisulfite sequencing. (A) Map of the Pu.1 promoter indicating the position of CpGs and transcription factor binding sites. (B) Average DNA methylation levels of 20 independently sequenced clones.

Induction of RUNX1 expression in RUNX1^−/− ES cells carrying an inducible RUNX1 allele rapidly induces chromatin unfolding at Pu.1. (A) Experimental strategy. (B) Expression analyses of Runx1, Pu.1, and Fli1 mRNA in the indicated cell types and with/without doxycycline (DOX) induction (concentration 0.1 μg/mL). (C) DNaseI in vivo footprinting assays examining the Pu.1 promoter in the indicated cell types. Signal intensities were measured across the displayed regions and were calculated relative to the rDNA control, with ES cells set as 1. (D) iRUNX1 cells were differentiated into embryoid bodies and induced with 0.1 μg/mL DOX for 4 hours or left uninduced. Flk1⁺ cells were purified by magnetic cell sorting. ChIP assays were performed in duplicate with 10⁷ cells for each assay and were normalized against the signal with an amplicon located on chromosome 2 (Chr2 control) representing a nonexpressed GAPDH pseudogene. Numbers in panels B and D are mean values between 2 measurements.

RUNX1 binds transiently to Pu.1 in hemangioblasts. DamId assay analyzing GATC methylation levels at the indicated genes with hemangioblasts expressing Dam-methylase alone (iDam control) or RUNX1-Dam (iRUNX1-Dam) after a 4-hour induction with DOX. Differential methylation was detected by cleaving first with Dpn1, which cleaves only methylated DNA, and then with DpnII, which cleaves only unmethylated DNA.³² The Chr2 control represents a nonexpressed GAPDH pseudogene. The values represent the mean values of 3 independent experiments.

A 12-hour pulse of RUNX1 expression is sufficient to rescue continuous Pu.1 mRNA expression, induce Csf1r mRNA expression, and rescue subsequent hematopoietic development. ES cells were differentiated as indicated. mRNA of the indicated cell populations (iRUNX1 and wt) was measured by real-time PCR (A) and RUNX1 protein was measured by Western blotting (B) using an anti-RUNX1 antibody after 4 days of blast culture (except for the +DOXd2 samples, which were measured before or after 8 hours of withdrawal, as indicated). (C) Number of hematopoietic colonies before and after DOX withdrawal; (D) Giemsa-stained cytospin of representative colonies. mRNA of iRUNX1 Flk1⁺ cells was prepared straight after cell sorting. Numbers in panels A and C represent mean values of 2 independent experiments analyzed in duplicate.

Active chromatin is maintained after RUNX1 withdrawal. (A) In vivo DMS footprinting assay examining the Pu.1 3′ URE conducted with 3T3 cells, macrophages (Mac), and CD41⁺ cells sorted after 4 days of a blast culture that had been continuously induced with 1 μg/mL doxycycline at day 2 (+DOX) and where the inducer was removed after day 3 (−DOX). Transcription factor binding sites are indicated on the right. Guanines displaying hyperreactivity to DMS-methylation are marked by •; protections from methylation are indicated by white circles; the weak protection over the RUNX site is indicated by a gray circle. Only reproducible changes are marked. Remaining band intensity (with protected bands) or fold enhancement (with hyperreactive bands) is indicated next to the bands with the G reaction set as 1. (B) ChIP assay demonstrating binding of C/EBPβ to its binding sites at the 3′ URE and the Pu.1 promoter in blast cultures from wt cells (■) and induced iRUNX1 cells (iRUNX-DOX) where DOX has been withdrawn as described in panel A. (C) Network diagram illustrating the initiating role of RUNX1 in establishing open chromatin and the expression of hematopoietic master regulators such as PU.1 in pluripotent hematopoietic precursor cells as well as the maintenance of the active state in differentiating myeloid cells. Once Runx1 is activated in pluripotent hematopoietic cells,^60,61 a regulatory circuit is established capable of maintaining open chromatin in the absence of RUNX1 and driving myelopoiesis. The induction of secondary factors driving macrophage-specific genes has been described in Krysinska et al⁹ and Laslo et al.⁶² The relative importance of the different transcription factors is illustrated by thin (less important) and bold (more important) lines. Numbers in panel B represent mean values of 2 independent experiments analyzed in duplicate.