Supplementary Figure 1 Strategy for parallel detection of DHSs and adjacent nucleosomes

Transcription

1 Supplementary Figure 1 Strategy for parallel detection of DHSs and adjacent nucleosomes DNase I cleavage DNase I DNase I digestion Sucrose gradient enrichment Small Large F1 F F9 F1 F1 F2 F3 F4 F5 F6 F7 F8 F9 F1 * * * 1-kb Purification of hypersensitive sites Sequencing library construction Massively parallel paired-end sequencing of two size ranges bp <2-bp Sequencing of both small and large Map to reference genome and infer actual DNase I In silico reconstruction of regulatory DNA architecture In silico 1- to 125-bp separation of based on length 126- to 185-bp Transcription factors Nucleosomes A standard DNase I digestion of intact human nuclei (John 29, Thurman 212, Neph 212) preferentially releases small pertaining to sites of transcription factor binding and nucleosome occupancy. DNase I hypersensitive sites (DHSs) are purified via ultracentrifugation of DNA layered on top of a 9% sucrose cushion. Small remained near the top of tube, while larger migrated to the bottom. The separation was monitored on a 1.5% agarose gel and <1-kb (red dashed line) were used to construct a DNase I library. Canonical DHS are small (<1-bp) while nucleosomal are larger (>125-bp), therefore two size ranges were selected from the same DNase I library via agarose gel extraction and sequenced independently in paired-end mode. The paired sequencing reads were mapped to the genome and used to reconstruct DNase I in silico. Stratification of the inferred on the basis on length revealed sites of transcription factor or nucleosome occupancy. Nature Methods: doi:1.138/nmeth.2713

2 Supplementary Figure 2 Size selection of DNase I 1-bp 5-bp 3-bp 2-bp 1-bp Sequenced libraries DNase I were purified and ligated with adapters compatible with the Illumina sequencing platform. The ligated were amplified with 8 cycles of PCR and loaded onto a 2% agarose gel and run for 1V for 1 hr in Tris-acetate/EDTA (TAE) buffer. DNA was purified from two independent regions of the gel and sequenced. Dotted lines indicate the excised gel slices. Nature Methods: doi:1.138/nmeth.2713

3 Supplementary Figure 3 Distribution of DNase I fragment lengths a bp bp Frequency (in millions) Class DHS Class 1.4 bp periodicity 1 Class Nucleosome ,861,593 DNase I DNase I fragment length (bp) b 1.4 bp Power ( 1 6 ) Frequency (bp) (a) A histogram of inferred DNase I fragment lengths from the >25 million sequenced. The shaded curves highlight the regions on the bimodal distribution that correspond to derived from cleavages in a DHS (orange) or around nucleosomes (blue). Grey dashed lines demarcate the size ranges used for the stratification of. (b) The periodicity of the histogram in Fig. 1c was analyzed by Fourier transformation and plotted as a power spectrum (x-axis, frequency in base-pairs; y-axis, magnitude of signal). Nature Methods: doi:1.138/nmeth.2713

4 Supplementary Figure 4 DNase I cleavage on the nucleosome core particle NCP DNase I DNase I cleavage (per-nt) n = 54,173 nucleosomes Distance to nucleoseome dyad (bp) Per nucleotide DNase I cleavage profile around the 1% (n core particle.top, localization of theoretical DNase I released from digestion of the nucleosome. Within Nature Methods: doi:1.138/nmeth.2713

5 Supplementary Figure 5 Detection of nucleosomes flanking DHSs a DHS vs. nucleosome fragment signal intensity log 2 (Normalized DHS density) Linear model r log 2 (Normalized nucleosome density) b DNase I density profiles surrounding DHS peaks (i) 1- to 125-bp (ii) 126- to 185-bp Mean DNase I fragment density Distance from DHS peak (bp) 8 1%-ile 9%-ile 8%-ile 7%-ile 6%-ile 5%-ile 4%-ile 2%-ile 1%-ile Distance from DHS peak (bp) Mean DNase I fragment density (a) Scatterplot of small (y-axis) vs. large fragment (x-axis) reveals a strong correlation (linear model; r 2 P < 2.2 x 1-6 ). (b-c) ( bp). Nature Methods: doi:1.138/nmeth.2713

6 9% Mean length of DNase I covering >5% of motif (bp) High DNase I cleavage Relative position from motif (bp) Low Low High 1% 5,92 NF-Y motif instances in DHS 4% 5% 6% 7% 8% 9% NF-Y binding site NF-Y binding site Relative position from motif (bp) Relative position from motif (bp) Low High DNase I cleavage Low High 1- to 125-bp fragment density Low 5 +5 Mean length of DNase I covering >5% of motif (bp) Relative position from motif (bp) 5 1% +5 5 Nucleosome derived 3% Least accessible TF derived 25 2 Mean length of all DNase I 2% h 5,92 Nf-Y motif instances in DHS ordered as in (a) Most accessible g 5 NF-Y f 1 Binding sites grouped by decile e High 126- to 185-bp fragment density 1- to 125-bp fragment density % 8% +5 7% 6% 5% Low Relative position from motif (bp) 4% Relative position from motif (bp) NRF1 binding site 2 3% Least accessible/ occupied NRF1 binding site 5 4,716 NRF1 motif instances in DHSs Mean length of all DNase I 2% Nucleosome derived +5 1% TF derived 4,716 NRF1 motif instances in DHSs ordered as in (a) Most accessible/ occupied 1 5 NRF1 Binding sites grouped by decile Supplementary Figure 6 DNase I fragment length parallels TF occupancy a b c d High 126- to 185-bp fragment density (a) Aggregate profile (top) and heatmap (bottom) of the per-nucleotide cleavages surrounding ±5-bp surrounding 5 4,716 predicted NRF1 binding sites in accessible chromatin. Each row of the heatmap corresponds to one predicted NRF1 binding site and the rows are sorted by the cleavage density ±25-bp surrounding the TF recongition sequence. (b) Distribution of the mean fragment lengths overlapping each NRF1 motif. Each box corresponds to 1% of the predicted NRF1 binding sites ordered as in (a). A dashed red line represents the mean fragment length of all detected irrespective of binding site motif. (c-d) Heatmaps of the density of TF-derived (c) and nucleosome derived (d) ±5-bp surrounding predicted NRF1 binding sites from (a). (e-h)same as (a-d) for the 5,92 predicted binding sites for the transcriptional regulator NF-Y in accessible chromatin. Nature Methods: doi:1.138/nmeth.2713

7 Supplementary Figure 7 Nucleotide conservation surrounding nucleosomes flanking DHSs NCP Linker Regulatory DNA Motif density (TRANSFAC per-nt) Conservation (phastcons 46-way vertebrate per-nt) Distance to nucleoseome dyad (bp) 5 Nucleotide conservation surrounding the 1% most accessible nucleosomes. Grey, density of predicted transcription factor binding sites (TRANSFAC, P < 1 5 ). Magenta, the average (mean) phastcons (46-way Nature Methods: doi:1.138/nmeth.2713

8 Supplementary Figure 8 Sequence composition surrounding nucleosomes flanking DHSs a Dinucleotide pattern associated with well-positioned nucleosomes (Valouev et al., 211) Motif density GC dinucleotide GG dinucleotide AA dinucleotide NCP Linker Regulatory DNA Motif density (TRANSFAC per-nt) Dinucleotide frequency ( 1 2 ) (21-bp avg.) Distance to nucleoseome dyad (bp) 5 b 1 2 Shuffle sequence maintaining dinucleotide composition within region bp from nucleosome dyad Scan for TRANSFAC motifs (P < 1 5 ) Motif density (TRANSFAC per-nt) Not Dinucleotides shuffled shuffled (a) Sequence features surrounding the 1% most accessible nucleosomes. Grey, density of predicted transcription factor binding sites (TRANSFAC, P < 1 5 ). Colored curves show the dinucletide frequency for GC dinucleotides linker DNA. (b) Analysis of TRANSFAC motif density controlling for sequence dinucleotide composition. Nature Methods: doi:1.138/nmeth.2713

9 Supplementary Figure 9 Nucleosome organization surrounding TF binding sites b USF Nucleosome dyad distance to motif (bp).12 Nucleosome dyad distance to motif (bp).25 Density Density Density c AP-1 2 CTCF 2 a Nucleosome dyad distance to motif (bp) (a-c) The nucleosome profiles surrounding the top 1% most highly occupied transcription factor binding sites of (a) CTCF, (b) AP-1 and (c) USF-1. Nature Methods: doi:1.138/nmeth.2713

10 Supplementary Figure 1 DNase I reveals a diversity of promoter types Total unidirectional promoter DHSs (n = 18,454) Promoter with 1 TSS 12 No. DHS No. +1 nucleosomes in promoter DHSs Promoter with 2 TSSs utilizing distinct +1 nucleosomes 18,454 DHS scorresponding to unidirectional promoters were scanned for the presence of one or more +1 nucleosomes. Nature Methods: doi:1.138/nmeth.2713

11 Supplementary Figure 11 Fig. 5a Fig. 5a Fig. 5a Nature Methods: doi:1.138/nmeth.2713

12 Supplementary Table 1 Summary of sequencing statistics corresponding to the two size fractions gel purifed. DNase I library Total mapped (Q > 1) Proper mapping (-chrm) % unique mapping No. cleavages detected <2 bp 119,623,788 89,763, % 179,527, bp 29,192,86 162,97, % 324,195,288 Combined total 328,8, ,861, % 53,723,186 Nature Methods: doi:1.138/nmeth.2713