Rlapped by PU.1 ChIP-seq peaks. g Relationship between total PU.1 ChIP signal found in vehicle controls and SAHA-treated K562 at all PU.1 peaks. ChIP-seq signal is normalized by total number of mapped reads in each condition (Spearman’s correlation = 0.811)local deposition of the histone enhancer mark H3K4me1 [16]. We detected H3K4me1 enrichment at each of the six sites prior to HDACi SCR7 cost treatment (Additional file 6: Fig. S3C) that modestly decreased at five of the six sites following treatment (Fig. 3d), possibly reflecting nucleosome repositioning. To characterize genome-wide PU.1-binding changes with HDACi treatment, ChIP-seq was performed on vehicle and SAHA-treated K562 cells in triplicate, which identified 31977 PU.1-binding sites in total. We observed hundreds of sites with increased PU.1 binding following HDACi treatment similar to those flanking the CDKN1A gene (p21) (Fig. 3e), a well-studied tumor suppressor that mediates p53-dependent G1 growth arrest and becomes up-regulated following HDACi treatment [20]. As predicted, PU.1 binding was highly enriched in opened DHS sites relative to all DHS sites in K562 cells or closed DHS sites (Fig. 3f). Interestingly, PU.1 ChIP-seq signal increased globally at the vast majority of peaks following SAHA treatment (Fig. 3g) while maintaining a similar distribution of total binding sites. This result was detected in each of the three replicates processed as pairs of HDACi-treated and vehicle control cells. The global increase in PU.1 binding also matches our ChIP-qPCR results (Fig. 3b, c).PU.1 overexpression modestly increases accessibilityIn addition to our ChIP-seq showing global increases in PU.1 occupancy following SAHA treatment, we also detected that PU.1 expression levels increase during HDACi exposure in our RNA-seq data (Additional file 7: Fig. S4A). PU.1 transcription is known to be tightly regulated during normal hematopoietic differentiation with different expression levels facilitating key transition points in cell lineage [21, 22]. To test whether HDACi-induced up-regulation of PU.1 was responsible for the chromatin accessibility changes, we transfected K562 cells with either a PU.1 cDNA under control of a viral promoter or an empty vector control. We selected for transformed cells by G418 resistance and performed DNase-seq. On the day of cell harvest, we confirmed PU.1 overexpression by qPCR (Additional file 7: Fig. S4B) and western blot (Fig. 4a).To characterize the impact of overexpression of PU.1 on chromatin accessibility, we analyzed PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/26024392 the original set of 7962 SAHA-opened DHS sites and divided them by those bound by PU.1 (n = 2137) versus the remaining that did not bind PU.1 (n = 5825). For the 2137 sites bound by PU.1, we observed a reproducible increase in mean accessibility in cells that overexpress PU.1 (Fig. 4b). This increase in accessibility was not detected at the 5825 SAHA-opened DHS sites that do not bind PU.1 (Fig. 4b). Furthermore, the distribution of fold-changes in accessibility found at SAHA-opened DHS sites with PU.1 binding was significantly greater than that of the SAHAopened DHS sites without PU.1 binding (Mann hitney test, P < 2.9 ?10-59) or PU.1-bound DHS sites that did not open further with SAHA treatment (P < 6.7 ?10-39) (Fig. 4c). These patterns are exemplified by two DHS sites found in the same intron of the TMEM51 gene; both sites open with SAHA treatment, but only the site with PU.1 bound displays increased hypersensitivity following PU.1 overex.