Correlation of Runx2 occupancy and Runx2-reponsive genes identifies novel targets. (A) Expression levels of genes responsive to Runx2 silencing in differentiated MC3T3-E1 cells. Values are plotted as the log2(expression level) from shRunx2-expressing cells (vertical axis) versus control shRNA expression (horizontal axis). Each point represents mean mRNA expression level from three independent biological replicates. Several representative genes are labeled. Diagonal lines demarcate the threshold for significant increase or decrease (≥1.5-fold) in expression. (B-D) Runx2 peaks associated with upregulated (Up), downregulated (Down), or unchanged (Non-responsive) gene expression upon Runx2 knockdown were compared by: average peak number per gene (B), and peak distribution (C) and fold change of peak signals (d9 versus d0) across genomic locations (D) with shRunx2 non-responsive genes as a control. In (B), three groups were compared to non-responsive genes: all peaks in shRunx2-regulated genes (All peaks), all peaks present at day 9 in shRunx2-regulated genes (D9 peaks), and all peaks present at day 0 in shRunx2 regulated genes (D0 peaks). Values are mean ± SEM (B,D) and statistical significance (*P < 0.01, **P < 0.05) determined by Mann-Whitney test (B,D) or Fisher’s exact test (C). (E) Runx2 enrichment across gene bodies (±10 kb) of genes downregulated (Down), upregulated (Up), and unchanged (Non-responsive) by shRunx2 treatment at day 9. Mean signal ratios (IP/Input) at each genomic region were determined using PeaksToGenes. Error bars represent SEM. (F) Mean phyloP conservation scores of Runx2 motifs associated with genes significantly (fold change ≥1.5, FDR <0.05) downregulated (Down), upregulated (Up), or unchanged (Non-responsive) upon shRunx2-treatment. Conservation of Runx2 motifs was compared between shRunx2 downregulated and upregulated genes and length-matched non-responsive genes (Non-responsive, Down and Up, respectively) and statistical significance determined (**P < 0.05) by Kolmogorov-Smirnov test.
Wu et al. Genome Biology 2014 15:R52 doi:10.1186/gb-2014-15-3-r52