(a) Pulsed field gel electrophoresis of DNase I–digested nuclear DNA. The concentrations of DNase I used for DNase-chip are labeled as A, B and C. (b) Outline of DNase-chip protocol. (c) Histogram of signal ratios of DNase I–treated versus random-sheared DNA. Tiled oligos that displayed the top 5% ratios are located to the right of the red bar. (d) Identification of regions with significant P values. The raw ratio data are plotted in gray, with the y-axis label on the right; the top 5% cutoff is displayed as a dotted horizontal gray line. The P value data for sliding 500-bp windows are plotted in red, with the y-axis label on the left.

(a–f) DNase-chip–identified regions were confirmed by real-time PCR for CD4⁺ T cells (a,c,e) and GM06990 cells (b,d,f). ΔCt values represent the number of additional cycles to achieve threshold levels of amplification between DNase I–treated and non-digested nuclear DNA. Higher ΔCt values represent elevated levels of DNase I–digestion. Control primer sets were designed around random regions of the genome, as well as known DNase I hypersensitive sites generated from a separate study8 (MPSS cluster). Real-time PCR using primer sets flanking DNase-chip peaks that are present with all three DNase I concentrations (a,b). Real-time PCR using primers sets flanking DNase-chip peaks that are present with two out of three DNase I concentrations (c,d). Real-time PCR using primer sets flanking DNase-chip peaks that are present with only a single DNase I concentration (e,f).

(a,b) DNase-chip P value data from two different ENCODE regions shows peaks that are detected using multiple concentrations of DNase I. MPSS clusters (in red) show DNase I hypersensitive sites that were previously identified using a sequence-based approach and clustering of tags. Note that there are regions detected by DNase-chip that have not been detected by MPSS, as expected because of limitations in the number of sequence tags. Many DNase-chip peaks that are only detected with a single DNase I concentration appear to be false positives. (c) Venn diagram shows the number of DNase I hypersensitive sites identified by DNase-chip (black; the number of DNase-chip regions are normalized using positive predictive values) and validated MPSS clusters (red).

(a,b) Real-time PCR was performed on both CD4⁺ T cells and GM06990 cells. Real-time PCR using primer sets that flank random regions of the genome or DNase-chip peaks that are present for both cell types (a). Real-time PCR using primer sets that flank DNase-chip peaks that are present in only CD4⁺ T cells (CD4) or GM06990 cells (GM; b).

(a,b) DNase-chip peaks were mapped to ENCODE regions stratified by gene density and human-mouse sequence conservation for both CD4⁺ T cells (a) and the GM06990 lymphoblastoid cell line (b). (c) The genomic locations of DNase I hypersensitive sites (detected with at least two concentrations of DNase I) and computationally generated random controls (n = 1,000) were compared to Gencode transcription start and end sites (within a 2-kb window), CpG islands, first introns, non-first introns, first exons, non-first exons, conserved sequences (MCS), conserved sequences minus coding exons (MCS-no-CDS). The number of DNase I hypersensitive sites at different distances (0 kb, 2 kb, 10 kb, and 25 kb) from Gencode genes was also determined. Error bars represent the entire range of values seen randomly generated mock datasets (n = 1,000). Compared to the random controls, the locations of the DNase-chip peaks are significantly (Monte Carlo P < 0.001) over- or under-represented at all positions except transcription end and > 0 kb from gene.

(a,b) The distance of each transcription start site (blue dots) to the nearest DNase I hypersensitive site was compared to the gene expression values of each transcript for both CD4⁺ T cells (a) and the GM06990 lymphoblastoid cell line (b). Horizontal red lines mark the expression level that separates most genes that have a DNase I hypersensitive site nearby (<1,000 bp) versus those that do not (>1,000 bp). (c) Average expression values of genes that have a DNase I hypersensitive site within 1 kb versus those that do not were determined for both cell types. Average expression values of genes that have a DNase I hypersensitive site are significantly different from genes that do not (P < 1 × 10⁻³⁸). CD4⁺ T cells, CD4; or GM06990 cells, GM.