Rescue experiment identifies Runx1-target genes that mediate cell adhesion. (A) Experimental design of Runx1 rescue experiments used to define direct target genes. 4-OHT, 4-hydroxytamoxifen. (B) Heat maps depicting microarrays expression results from untreated GMP from Runx1^Δ/Δ and Runx1^+/+ mice, as well as rescue experiments analyzing transduced GMP. Left panel shows the relative expression levels for 42 genes downregulated in Runx1^Δ/Δ GMPs and upregulated after tamoxifen treatment of RUNX1-transduced, but not control-transduced (ERt2), GMPs. Right panel shows 94 genes that are upregulated in Runx1^Δ/Δ GMPs and downregulated after activation of RUNX1. Genes deregulated by a factor >2 (P ≤ .05) and that could be rescued by a mean factor of >1.5 are shown. Roman numerals indicate genes belonging to the GO Biological Processes indicated in panel C. (C) Biological processes (bar graph) that are enriched in the 94 upregulated genes were identified by GO analysis using the STRING database. The number of genes per class and the statistical significance is indicated. (D) Protein interactions clustered by k-means (n = 92; k = 5) to highlight interaction groups (http://string-db.org). (E) Adhesion assay of Runx1^+/+ or Runx1^Δ/Δ BM progenitors. Lin⁻ cells were plated on MS-5 stromal cells and assessed for adherence after 48 hours. Shown are the results of 1 out of 2 independent experiments. Each dot represents an independent well. Horizontal line, median; error bars, ±SE. P values were calculated by a unpaired Student t test; ***P < .001. (F) Experimental design to test candidate genes that impact on Meg-skewing through adhesion interactions (left). The relative number of Megs was assessed by CD41 expression and confirmed microscopically (right). Expression levels in BFP expressing cultures were set as 1. Error bars, ±SE. P values were calculated by a unpaired Student t test; **P < .01. HPSC, hematopoietic stem and progenitor cells.