LEGENDS TO THE SUPPLEMENTAL FIGURES
Supplemental Figure 1. Isolation, and properties of blood cell lineages. (A) Flow cytometry analysis of isolated immature (Imm, middle row) and mature (Mat, bottom row) specified blood cells. The immunophenotypes verify lineages. (B) Reanalysis of isolated immature cell populations by flow cytometry verified their purity. (C) Heatmap showing robust differential expression of the 200 most erythroid cell-specific and 163 most granulocyte-specific transcripts, based on hybridization to Affymetrix microarrays. (D) RT-PCR analysis of representative erythroid-specific genes confirmed absent or negligible expression in immature granulocytes and MK. Integrated Genome Viewer (IGV) traces at the genes marked with * are shown as examples in other figures. (E) Isolation of blood progenitors from cultures of mouse fetal liver cells for 1 day in IL-3 and SCF. After depletion of cells expressing lineage markers (Lin-), cells were separated by sequential flow cytometry, first into multipotent HSC (KithiSca1hi) and KithiSca1lo progenitors (top panel). The latter cells were re-sorted to isolate CMP (CD34hiFcRint), GMP (CD34hiFcRhi) and MEP (CD34intFcRlo). Boxes denote gates applied to isolate each population.
Supplemental Figure 2. Overview of ChIP-Seq experiments. (A) ChIP-Seq sample sizes and identified peak counts using MACS. (B) Distribution of ChIP-Seq MACS peaks in the genome.
Supplemental Figure 3. Enhancer H3K4me2 in blood cell progenitors (A) Robust ChIP-seq detection of H3K4me2-marked areas in blood progenitors. ChIP-seq profiles have different morphologies in progenitor and committed cells because chromatin was prepared differently: sonication vs. MNase digestion. Marked histones, readily mapped in both cases, allow reliable enhancer comparisons across cell populations. (B) 3-dimensional representation of the quantitative map of H3K4me2-marked enhancer chromatin dynamics.
Supplemental Figure 4. Cis-element analyses in hematopoietic progenitors. (A) Heatmaps of promoter H3K4me2 (normalized read count, RPKM) at the 200 most erythroid-specific (top) and 163 most granulocyte-specific genes (bottom). (B-C) Examples of loci expressed strongly and selectively in young granulocytes (Fpr1) or erythroid cells (Hbb-b2), illustrating a dearth of H3K4me2 enhancer marks in multipotent HSC and CMP or bipotential GMP and MEP.
Supplemental Figure 5. Breadth and specificity of H3K4me2 and DHS at blood-cell enhancers. (A) Data traces at the erythroid-specific Prokr1 locus, showing similar H3K4me2 marking in all 3 specified immature blood cell types. (B) Venn diagram showing negligible binding of NF-E2, PU.1, and GATA1 to the same sites in different cells. (C) H3K4me2 marking in primary mouse blood lineages 500 bp on either side of GATA1 binding sites in MEL cells. (D) Representative erythroid- (Tspo2) and granulocyte- (Ctsg) expressed genes show H3K4me2-marked nucleosomes in various blood cell types, but not in thymocytes (Thy) or intestinal cells (Int). (E) Heatmaps show that most well-positioned, H3K4me2-marked nucleosomes flanking NF-E2-binding sites in immature erythroid cells are not marked in thymocytes or intestinal villi.
Supplemental Figure 6. Persistence of primed enhancers in terminally differentiated blood cells. (A) Composite H3K4me2 profiles in immature and terminally mature erythroid cells and granulocytes at the same ~1,800 NF-E2 binding sites and ~7,900 PU.1-binding sites examined and shown in Figures 5 and 6. The represented region spans -500 to +500 bp from the summits of TF binding.
Supplemental Figure 7. Erythroid enhancers primed in granulocytes show H3K27ac and DHS. (A) Nucleosomal H3K27ac marks at representative H3K4me2-marked ERY- (Ube2o) and GRAN- (Mpo) specific loci in immature cells of each lineage. (B) Distributions of H3K27ac peak widths in erythroid cells (top) and granulocytes (bottom) in relation to lineage-favored expression of nearby genes, grouped and ranked in 200-gene bins, as in Fig. 5C. The widths of MACS-identified H3K27ac-marked enhancers in each bin are represented as boxplots. Even genes highly specific to each lineage showed similar H3K27ac peak distributions in the other cell type. (C) Composite plots of H3K4me2 ChIP-seq tag density in erythroid cells (red), MK (green) and granulocytes (blue) near the top 1,000 H3K4me2-marked GATA1-binding sites in MEL cells. (D) Composite DHS profile in immature granulocytes at the same 1,000 sites, with data from the + and - strands represented with filled and dashed lines, respectively. (E) Heatmaps of DHS in MEL cells between -500 bp and +500 bp from the summits of NF-E2 binding sites in primary erythroid cells (left) and PU.1 binding sites in primary granulocytes (right).