Mapping 3 D genome architecture through in situ DNase

2104 | VOL.11 NO.11 | 2016 | nature protocols IntroDuctIon The manner in which an incredibly long DNA polymer toplogically organizes itself within a cell or nucleus is crucially linked to higher-order cellular function1,2. This form–function relationship, first discovered through early light microscopic studies of higher-order structures such as mitotic chromosomes3, the inactive X Barr body4, and polytene chromosomes5, has only become clearer in the face of advancing technologies. Techniques such as fluorescence in situ hybridization (FISH) of chromatin6–8 have provided clear evidence that chromosomes occupy compartments within the nucleus, ultimately leading to the development of correlative models associating biological function (i.e., transcription, splicing, silencing) with particular nuclear locales9,10. With the advent of genome-scale technologies, high-throughput assays have been developed to characterize nuclear architecture at both increasing scale and increasing resolution. Techniques such as DNA adenine methyltransferase identication11,12, typically used to map protein–DNA interactions13–15, have been modified to map genome-wide associations between primary sequences and the nuclear lamina16 (i.e., lamina-associated domains or LADs), where silenced domains typically reside. Methods involving the ‘proximity ligation’ of chromatin, now termed 3C (ref. 17), have also gained popularity. 3C techniques represent matured versions of early methods that used T4 DNA ligase to quantify the physical proximity of DNA sequences brought together by proteins18,19, and all share a common paradigm: fixation of chromatin within the nucleus via formaldehyde, endonucleolytic digestion of chromatin (normally via restriction enzyme (RE) digestion), and religation of physically proximal fragments. The first 3C variants (e.g., 4C, 5C) used specific primers or sets of primers to determine contact frequencies between predefined sites in the genome20,21. Later, massively parallel versions of 3C, generally termed ‘Hi-C’, were developed22–24 that leverage paired-end sequencing to generate contact-frequency estimates between sequence windows across entire genomes. Since the advent of 3C techniques, much work has gone into characterizing 3D genome architecture in a wide variety of biological contexts25–29, including mitotic cell division30, the life cycle of a parasite31, and mammalian dosage compensation32–35. The vast amount of available Hi-C data has also enabled the discovery of novel ‘units’ of genome topology, including topologically associating domains (TADs)33,36 and chromosomal interacting domains27,37, genomic domains that predominantly self-associate in 3D space. Although the ultimate significance of these domains remains unknown, strong correlations between 1D epigenomic features (e.g., histone marks, DNA methylation and transcription factor binding) and sequence, both within and at the borders of these domains, suggest that they may have a gene regulatory role. Although current Hi-C techniques generally allow us to visualize genome-scale chromosome architecture at a resolution of 100 kb to 1 Mb, methodological resolution limitations imposed by incomplete sequencing depth and genome-wide restriction site density have typically precluded the identification of topological units at smaller scales, in which local interactions may have crucial gene regulatory roles. The need for fine-scale resolution of these higher-order interactions has only become clearer in the wake of the immense amount of high-resolution, 1D epigenomic data generated by consortia such as ENCODE38 and Roadmap Epigenomics39. Given the availability of such data, one crucial interest of the gene regulatory field is the potential link between complex gene Mapping 3D genome architecture through in situ DNase Hi-C