Using DNase Hi-C techniques to map global and local three-dimensional genome architecture at high resolution

Abstract

The folding and three-dimensional (3D) organization of chromatin in the nucleus critically impacts genome function. The past decade has witnessed rapid advances in genomic tools for delineating 3D genome architecture. Among them, chromosome conformation capture (3C)-based methods such as Hi-C are the most widely used techniques for mapping chromatin interactions. However, traditional Hi-C protocols rely on restriction enzymes (REs) to fragment chromatin and are therefore limited in resolution. We recently developed DNase Hi-C for mapping 3D genome organization, which uses DNase I for chromatin fragmentation. DNase Hi-C overcomes RE-related limitations associated with traditional Hi-C methods, leading to improved methodological resolution. Furthermore, combining this method with DNA capture technology provides a high-throughput approach (targeted DNase Hi-C) that allows for mapping fine-scale chromatin architecture at exceptionally high resolution. Hence, targeted DNase Hi-C will be valuable for delineating the physical landscapes of cis-regulatory networks that control gene expression and for characterizing phenotype-associated chromatin 3D signatures. Here, we provide a detailed description of method design and step-by-step working protocols for these two methods.

Keywords: Chromatin; Chromosome; Chromosome conformation capture (3C); DNase Hi-C; Hi-C; Three-dimensional (3D) genome architecture.

Fig. 1. Size distribution of the projected restriction fragments in mouse and human genomes after the in silico digestion with DpnII and HindIII, two REs frequently used in Hi-C assays

Note that for the 4 bp-cutter DpnII, about 8.73% (559,888 of 6,415,145) of fragments (RE-fragment counts) generated in the mouse genome (NCBI/mm9) and 8.58% (611,434 of 7,127,561) in the human genome (GRCH37/hg19) are larger than 1 kb. Importantly, the DpnII fragments with a size > 1kb cover about 31.25% of the mouse genome and 33.28% of the human genome, respectively. For the 6 bp-cutter HindIII, about 20.38% (167,820 of 823,331) of fragments generated in the mouse genome and 22.74% (190,456 of 837,599) in the human genome are larger than 5 kb. Notably, the HindIII fragments with a size > 5 kb cover about 52.22% of the mouse genome and 58.33% of the human genome, respectively.

Fig. 2

The workflow of whole-genome and targeted DNase Hi-C assays.

Fig. 3. Quality control (QC) of experimental steps in the DNase Hi-C protocol

(A) Example showing titration of DNase I digestion of crosslinked chromatin (see protocol step 25). About 5 × 10⁶ crosslinked K562 cells (fixed in 1% formaldehyde) were lysed and divided into 6 equal aliquots, and each aliquot was digested with 2–10 units of DNase I at RT for 5 min. Genomic DNA was then isolated and 1/10 of the purified genomic DNA from each sample was subjected to gel electrophoresis in a 2% agarose gel. M, DNA size ladder. (B) Example showing QC of the enzymatic reactions in the DNase Hi-C protocol. About 2 × 10⁶ crosslinked K562 cells (fixed in 1% formaldehyde) were subjected to cell lysis, DNase I digestion, end-repair, and dA-tailing according to the DNase Hi-C protocol. After dA-tailing, the sample was divided into 3 equal aliquots, and ligated to the Illumina Y adaptor in solution (S), in a 0.4% low melting gel (G), or not ligated as a control (Control). Upper panel, about 300 ng purified genomic DNA from each sample (Input) was resolved in a 2% agarose gel; Lower panel, the corresponding PCR products of each input (1 ng DNA template) amplified using Illumina PCR primers (11 PCR cycles). M, 100 bp DNA ladder. (C) Example of the BamHI QC experiment (see protocol steps 135–137) performed on a K562 whole-genome DNase Hi-C library. Electrophoresis was carried out on a 2% DNA agarose gel. M, 100 bp DNA ladder.

Fig. 4. Enrichment of a subset of chromatin interactions of interest from whole-genome DNase Hi-C libraries by DNA capture

(A) Cartoon to illustrate capture of target-associated chromatin interactions using probes designed against a specific target. The structure of the chimeric DNA fragments (templates) in DNase Hi-C libraries is shown, including sequencing adaptors and bridge adapter. (B) Example of using real-time PCR to assess enrichment efficiency of target-associated chromatin interactions in targeted DNase Hi-C assays. Enrichment is comparable between a whole-genome DNase Hi-C library (chromatin interaction) and a control genomic library (gDNA) at four loci (HS2, HS3, Nanog, Sox2) using real time PCR.

Fig. 5. Visualization of targeted DNase Hi-C results using Circos diagram and domainograms

(A) Circos plots showing representative targeted DNase Hi-C-identify high confidence cell-type specific inter-chromosomal contacts in H1 and K562 cells. Each arc represents a contact; different colors are used to distinguish individual chromosomes. Targets are HS2-HS3 on chr11, Nanog on chr12, and Sox2 on chr3. (B) Cell type-specific 3D organization of the RUNX1 promoter in K562 and H1 cells. Domainographs showing contact profiles and significant contacts (red or green boxes) within 500 kb (upstream or downstream, chr21:35,921,595-36,923,595) of the target (RUNX1 promoter) are shown. Red arrows indicate the position of the target. The tracks of H3K27ac and DNase hypersensitive sites (DHSs) for both K562 and H1 ESCs are snapshots taken from the UCSC Genome Browser using data from the ENCODE Project Consortium. UCSC Genome Browser tracks for TFBS (CTCF, GATA-1, GATA-2, NF-E2, NF-YA, NF-YB, SPI1, ATF1, CCNT2, PML and EZH2 for K562cells; CTCF, EZH2, SUZ12, POU5F1, and NANOG for H1 ESCs) are also shown. The RefSeq genes are shown in the bottom. Chromosomal position is based on GRCh37/hg19.

Fig. 5. Visualization of targeted DNase Hi-C results using Circos diagram and domainograms

(A) Circos plots showing representative targeted DNase Hi-C-identify high confidence cell-type specific inter-chromosomal contacts in H1 and K562 cells. Each arc represents a contact; different colors are used to distinguish individual chromosomes. Targets are HS2-HS3 on chr11, Nanog on chr12, and Sox2 on chr3. (B) Cell type-specific 3D organization of the RUNX1 promoter in K562 and H1 cells. Domainographs showing contact profiles and significant contacts (red or green boxes) within 500 kb (upstream or downstream, chr21:35,921,595-36,923,595) of the target (RUNX1 promoter) are shown. Red arrows indicate the position of the target. The tracks of H3K27ac and DNase hypersensitive sites (DHSs) for both K562 and H1 ESCs are snapshots taken from the UCSC Genome Browser using data from the ENCODE Project Consortium. UCSC Genome Browser tracks for TFBS (CTCF, GATA-1, GATA-2, NF-E2, NF-YA, NF-YB, SPI1, ATF1, CCNT2, PML and EZH2 for K562cells; CTCF, EZH2, SUZ12, POU5F1, and NANOG for H1 ESCs) are also shown. The RefSeq genes are shown in the bottom. Chromosomal position is based on GRCh37/hg19.