MPE-seq, a new method for the genome-wide analysis of chromatin structure

Abstract

The analysis of chromatin structure is essential for the understanding of transcriptional regulation in eukaryotes. Here we describe methidiumpropyl-EDTA sequencing (MPE-seq), a method for the genome-wide characterization of chromatin that involves the digestion of nuclei withMPE-Fe(II) followed by massively parallel sequencing. Like micrococcal nuclease (MNase), MPE-Fe(II) preferentially cleaves the linker DNA between nucleosomes. However, there are differences in the cleavage of nuclear chromatin by MPE-Fe(II) relative to MNase. Most notably, immediately upstream of the transcription start site of active promoters, we frequently observed nucleosome-sized (141-190 bp) and subnucleosome-sized (such as 101-140 bp) peaks of digested chromatin fragments with MPE-seq but not with MNase-seq. These peaks also correlate with the presence of core histones and could thus be due, at least in part, to noncanonical chromatin structures such as labile nucleosome-like particles that have been observed in other contexts. The subnucleosome-sized MPE-seq peaks exhibit a particularly distinct association with active promoters. In addition, unlike MNase, MPE-Fe(II) cleaves nuclear DNA with little sequence bias. In this regard, we found that DNA sequences at RNA splice sites are hypersensitive to digestion by MNase but not by MPE-Fe(II). This phenomenon may have affected the analysis of nucleosome occupancy over exons. These findings collectively indicate that MPE-seq provides a unique and straightforward means for the genome-wide analysis of chromatin structure with minimal DNA sequence bias. In particular, the combined use of MPE-seq and MNase-seq enables the identification of noncanonical chromatin structures that are likely to be important for the regulation of gene expression.

Keywords: MPE-Fe(II); chromatin; genome-wide analysis; micrococcal nuclease; promoter.

Fig. 1.

Comparison of fragments generated by MPE-Fe(II) and MNase. (A) DNA fragments generated by digestion of J1 mouse embryonic stem cell nuclei with MPE-Fe(II) or MNase. For MPE-Fe(II) digestion, the nuclei were incubated with 50 μM of MPE-Fe(II) for the indicated times before quenching the reaction. For MNase digestion, the nuclei were incubated with the indicated concentrations of MNase for 10 min. The DNA was isolated from the digested samples and then analyzed by agarose gel electrophoresis. (B) The size distributions of sequenced DNA fragments. Sequencing libraries prepared from MPE-Fe(II)– or MNase-generated fragments were subjected to paired-end sequencing, and the sizes of the fragments were inferred from the positions of the mapped ends. (C–F) Cumulative base compositions of residues at positions relative to the mapped ends.

Fig. S1.

DNA fragments generated by different MPE-Fe(II) concentrations. Nuclei were digested for 20 min by using different MPE-Fe(II) concentrations as indicated. DNA was isolated and analyzed by agarose gel electrophoresis.

Fig. S2.

Cumulative base compositions of residues at positions relative to the mapped ends. (A) Data from samples digested with the low (1 U/mL) concentration of MNase (this study). (B) Data from MNase digestion of human granulocyte nuclei by Valouev et al. (10). (C) Data from MNase digestion of human granulocyte genomic DNA by Valouev et al. (10). (D) Data from MNase digestion of mouse embryonic stem cell nuclei by Teif et al. (12). (E) Data from MNase digestion of mouse embryonic stem cell nuclei by Li et al. (13). (F) Data from DNase I digestion by Vierstra et al. (22). (G) Data from ATAC-seq by Buenrostro et al. (27).

Fig. S3.

Occurrence of different dinucleotides at positions relative to the 5′ ends of mapped reads. (A) Data from samples digested with MPE-Fe(II). (B) Data from samples digested with MNase.

Fig. 2.

Analysis of chromatin structure with MPE-seq. (A) Diagram of nucleosome positioning and cutting site analyses. Fragments generated by MPE-Fe(II) or MNase digestion are indicated by the gray lines. The cutting sites were inferred from the positions of the DNA sequences at the ends of the fragments, which are denoted by squares. In the cutting site analyses, we used DNA fragments of all size ranges. The ends of forward and reverse reads were analyzed separately and then normalized to the genome-wide average. Nucleosome positioning analysis was performed as follows: We used fragments that are 141–190 bp in length because this size range is comparable to the amount of DNA that is typically associated with a single nucleosome. We assigned a value of 1 to each of the middle 60 bp of such fragments, as depicted by the yellow bars. For fragments with an odd-numbered length, we assigned a value of 1 to each of the middle 59 bp and a value of 0.5 to the bp on each end, as shown. The Nucleosome Positioning Index for each position is defined to be the sum of the values normalized to the genome-wide average. (B) Comparison of MNase-seq and MPE-seq data. Representative screenshots are shown for data obtained from J1 mouse embryonic stem cell nuclei that were digested by MNase under standard conditions (MNase), a low concentration of MNase (MNase Low), or MPE-Fe(II). The MPE-Fe(II) peaks labeled a, b, c, and d are discussed in the text.

Fig. 3.

Genome-wide analysis of the chromatin structure of promoters with MPE-Fe(II). (A–C) Averaged Nucleosome Positioning Index of chromatin digested with (A) MPE-Fe(II), (B) the standard concentration of MNase, or (C) the low concentration of MNase (MNase Low) at positions within 1 kb of the TSS with 20,195 promoters. The estimated average positions of nucleosomes are indicated by ovals. (D) Comparison of nucleosome positioning data with MPE-Fe(II), MNase, and MNase Low. (E–G) Averaged cutting site analysis of chromatin digested with (E) MPE-Fe(II), (F) the standard concentration of MNase, or (G) MNase Low for each strand at positions within 1 kb of the TSS with 20,195 promoters. (H) Cutting site analysis of forward reads. (I) Cutting site analysis of reverse reads. (J–L) Analysis of fragments of the indicated size ranges in chromatin digested with (J) MPE-Fe(II), (K) the standard concentration of MNase, or (L) the low concentration of MNase. The Fragment Positioning Index was calculated in a manner that is analogous to the Nucleosome Positioning Index (Fig. 2A) (SI Materials and Methods).

Fig. S4.

Fragment midpoint versus length plots of chromatin digested with (A) MPE-Fe(II), (B) the standard concentration of MNase, or (C) the low concentration of MNase. The fragments were classified according to their lengths and the positions of their midpoints relative to the TSS. The densities of fragments of specific lengths at their midpoint positions were then plotted.

Fig. 4.

MPE-Fe(II) and MNase digestion patterns at 16,800 promoters. The promoters were ranked according to their transcript levels [fragments per kilobase of exon per million fragments mapped (FPKM)] from RNA-seq data. The asterisks denote the position of the signal that is immediately upstream of the TSS. (A) Heat maps from the cutting site and nucleosome positioning analyses (140–191 bp DNA fragments). (B) Heat maps of the Fragment Positioning Index of subnucleosome-sized particles (101–140 bp) generated by digestion with MPE-Fe(II), MNase, or MNase Low conditions.

Fig. 5.

ChIP-seq analysis of histones H2B and H3 with chromatin digested with MPE-Fe(II) versus MNase Low conditions. The ChIP-seq experiments were performed with soluble chromatin that was generated by digestion with MPE-Fe(II) or MNase Low conditions; hence, the ChIP-seq input DNA samples are not identical to the DNA fragments in the MPE-seq and MNase Low-seq experiments, which were performed with total (soluble and insoluble) chromatin. To distinguish between nucleosome- and subnucleosome-sized particles, we analyzed the data in separate groups that correspond to 141–190 (nucleosome-sized) bp, 101–140 bp, and 50–100 bp DNA fragments. The peak of histone H2B and H3 localization in the upstream promoter region is indicated by the arrow. This analysis was performed with 20,195 RefSeq TSSs. (A, C, and E) MPE-ChIP-seq (A) 140–191 bp, (C) 101–140 bp, and (E) 50–100 bp DNA fragments. (B, D, and F) MNase Low-ChIP-seq analysis with (B) 140–191 bp, (D) 101–140 bp, and (F) 50–100 bp DNA fragments. (G) MPE-seq (101–140 bp DNA fragments) analysis from –10 kb to +10 kb relative to the TSS. (H) MPE-ChIP-seq (101–140 bp DNA fragments) of histones H2B and H3 from –10 kb to +10 kb relative to the TSS.

Fig. S5.

MPE-ChIP-seq and MNase Low-ChIP-seq with histones H2B and H3 at 16,800 promoters. The promoters were ranked according to their transcript levels (FPKM values) from RNA-seq data. The red dots denote the position of the signal that is immediately upstream of the TSS. (A) Heat maps of the nucleosome positioning analyses of 140–191 bp DNA fragments. (B) Heat maps of the Fragment Positioning Index of subnucleosome-sized particles (101–140 bp).

Fig. 6.

MPE-seq versus MNase-seq analyses of 3′ splice sites of internal exons. (A) The average AT content at positions relative to the 3′ splice sites of internal exons was plotted. (B) Cutting site analysis of MNase-digested chromatin around 3′ splice sites. (C) Cutting site analysis of MNase-digested genomic DNA around 3′ splice sites. (D) Cutting site analysis of MPE-Fe(II)–cleaved chromatin around 3′ splice sites. (E) Cutting site analysis of MPE-Fe(II)–cleaved genomic DNA around 3′ splice sites. (F) Averaged Nucleosome Positioning Index around 3′ splice sites.

Fig. S6.

MPE-seq versus MNase-seq analyses of 5′ splice sites of internal exons. (A) The average AT content at positions relative to the 5′ splice sites of internal exons was plotted. (B) Cutting site analysis of MNase-digested chromatin around 5′ splice sites. (C) Cutting site analysis of MNase-digested genomic DNA around 5′ splice sites. (D) Cutting site analysis of MPE-Fe(II)–cleaved chromatin around 5′ splice sites. (E) Cutting site analysis of MPE-Fe(II)–cleaved genomic DNA around 5′ splice sites. (F) Averaged Nucleosome Positioning Index around 5′ splice sites.

Fig. 7.

MPE-Fe(II) and MNase digestion patterns in the vicinity of CTCF binding sites. (A–C) Cutting site analysis. A total of 21,470 CTCF ChIP-seq peaks with a single CTCF motif within 1 kb were aligned according to the positions and orientations of the CTCF motif. Of these 21,470 CTCF peaks, 1,018 are within 1 kb of an annotated RefSeq TSS. The number of cuts for each strand at each position relative to the CTCF motif was averaged, normalized, and then plotted. (D–F) Fragment midpoint versus length plots. MPE-Fe(II)–, MNase-, or MNase Low-generated fragments were classified according to the lengths and the positions of their midpoints relative to the midpoints of the CTCF motif. The densities of fragments of specific lengths and relative midpoint positions were then plotted. (G and H) Detection of sequence-specific binding of CTCF to chromatin via cutting site analysis with MPE-seq. The CTCF binding sites were classified into five groups based on the relative enrichment of CTCF ChIP-seq signals, with the first quintile having the highest CTCF enrichment. For each quintile, the averaged profile of cuts generated by MPE-Fe(II) (G) or MNase (H) at positions within 50 bp of the middle of CTCF motif was plotted for the chromatin sample and the genomic DNA (control) sample. The CTCF motif used for this analysis and the average AT content at positions relative to the middle of CTCF motif are shown at the bottom of the figure.

Fig. S7.

Cutting site and nucleosome positioning analyses of MPE-Fe(II) and MNase digestion patterns in the vicinity of CTCF binding sites. The heat maps display 21,470 CTCF binding sites that were ranked based on their relative enrichment of CTCF ChIP-seq signals. (A) MPE-seq data. (B) MNase-seq data.

Fig. S8.

MPE-Fe(II) and MNase digestion patterns in the vicinity of REST binding sites. (A and B) Cutting site analysis. A total of 1,478 REST ChIP-seq peaks with a single REST motif within 1 kb were aligned according to the positions and orientations of the REST motif. The number of cuts for each strand at each position relative to the REST motif was averaged, normalized, and then plotted. (C and D) Fragment midpoint versus length plots. MPE-Fe(II)–, or MNase-generated fragments were classified according to the lengths and the positions of their midpoints relative to the midpoints of the REST motif. The densities of fragments of specific lengths and relative midpoint positions were then plotted. (E and F) Detection of sequence-specific binding of REST to chromatin via cutting site analysis with MPE-seq. The REST binding sites were classified into five groups based on the relative enrichment of REST ChIP-seq signals, with the first quintile having the highest REST enrichment. For each quintile, the averaged profile of cuts generated by MPE-Fe(II) (E) or MNase (F) at positions within 50 bp of the middle of REST motif was plotted for the chromatin sample and the genomic DNA control sample. The REST motif used for this analysis and the average AT content at positions relative to the middle of REST motif are shown at the bottom of the figure.