From: Three-dimensional genome architecture and emerging technologies: looping in disease
Technique | Most applicable scenario and/or advantages | Limitations | Relevant example(s) | Suitable computational pipeline(s) |
---|---|---|---|---|
DNA-centric view of genome architectural methods | ||||
 Chromosome conformation capture (3C) [92] | Interrogating looping interactions between single gene locus to single regulatory locus (one locus to one locus) | Not suitable for high-throughput identification of novel looping interaction | Association of causal variants from GWASs in 16p13 to DEXI gene in type 1 diabetes and multiple sclerosis [93] | Not required |
 Circular chromosome conformation capture on chip (4C) [94] | Exploring all possible interactions with a single clinically relevant locus (one locus to all loci) | Limited throughput | Association of regulatory SNP with target genes [95] | FourCseq [96] |
 Circular chromosome conformation capture combined with sequencing (4C-seq) [32] | Exploring all possible interactions with a single clinically relevant locus (one locus to all loci) | GC content or length of interacting fragment may introduce PCR bias | Chromosomal rearrangement detection [97] | FourCseq [96] |
 Chromosome conformation capture carbon copy (5C) [98] | Studying interactions between many chromosomal loci with many interacting regions across the genome (many loci to all loci) | Complicated primer/probe design can introduce amplification bias. Occasionally misses weak long-range contacts | Determined interaction profiles at pilot promoter regions in ENCODE project [2]. X-chromosome 5C study provided first evidence of topologically associated domains (TADs) [99] | HiFive [100] |
 Genome-wide chromatin conformation capture (Hi-C) [11] or its variant (in situ Hi-C) [6] | Circumstances where extensive chromatin reorganization occurs (i.e., stem cell differentiation), in which it is important to understand interactions between all parts of the genome (all loci to all loci). The most extensively used genome architectural method | Insensitive method for probing local intra-TAD interactions (<40 kb) unless performed at very high resolution | Genome-wide TAD distributions [1, 6]. Three-dimensional (3D) architectural changes during mitosis [101]. Determination of chromosomal translocations [102] | Methods are primarily divided into: (1) quality control and mapping; (2) domain calling; (3) visualization; and (4) 3D modeling. These methods have been extensively reviewed earlier [46, 47] |
 Tethered conformation capture (TCC) [103] | Proximity ligation step performed on solid substrate, thus reducing random intermolecular ligation (all loci to all loci) | Although more specific, proximity ligation occurs outside the cell, and thus some native cell context may be lost. Biotinylation step may require optimization | Originally applied to B-cell line, but intended for direct clinical/diagnostic applications | Accompanied by a novel method for TCC data analysis [103] |
 Genome-wide chromatin conformation capture with DNase I digestion (DNase Hi-C) or targeted DNase Hi-C [36] | Improves on Hi-C restriction-enzyme-mediated resolution limits, and thus is most suitable for higher-resolution architectural studies (all loci to all loci) | DNase I treatment may digest bait-targeted region; thus, tiling probes designed across the region are needed for targeted DNase Hi-C | Targeted DNase Hi-C was used to investigate chromatin architecture at many lncRNA loci | Analysis pipelines are similar to Hi-C methods |
 Genome architectural mapping (GAM) [66] | First ligation-free method for investigating cis-interactions in an unbiased manner. Moreover, it can capture three-way interactions more effectively than Hi-C | Time and specialization required to individually section and dissect out nuclei. Cell asynchrony and heterogeneity affect overall outcome | Uses thin tissue sample slices, which can be applied to frozen clinical tissue samples | GAMTools: specialized automated pipeline [66] |
DNA-centric view of genome architectural methods with target enrichment | ||||
 Chromosome conformation capture (3C) coupled with oligonucleotide capture technology (capture-C) [34] or next generation (NG-capture-C) [104] | Delineating interaction profiles for many chromosomal loci in a single experiment without introduction of PCR bias and without missing weak long-range interactions (many loci to all loci) | Initial capture-C data suffered from insufficient depth and captured some non-specific interactions. NG-capture-C overcomes these limitations and provides higher sensitivity and resolution | Initially found complex patterns of HIF response by defining chromatin architecture at multiple HIF-bound enhancer and promoter sites [105]. Can be applied to SNP-specific chromatin interaction profile generation | Capture-C analyzer and capture-C oligo design tools available on github [104] |
 Targeted locus amplification (TLA) [106] | Little requirement of prior sequence knowledge. Most suitable for studying chromosomal rearrangements, single nucleotide variants (SNVs), transgene integration sites, and haplotyping at large genomic intervals. Entire restriction fragments are sequenced, unlike 4C-seq where only ends of fragments are analyzed (many loci to all loci) | Potential for applying to purified genomic DNA or formalin-fixed paraffin embedded material, but current protocol is limited to cells only | Used for haplotyping at BRCA1 locus. Identified uncommon SNVs and indels. Identification of ApoE transgene locus, viral integration sites, and used to study chromosomal rearrangement for MLL gene | TLA analysis pipeline details in [107] |
 Targeted chromatin capture (T2C) [108] | Provides affordable diagnostic tools with restriction enzyme resolution to understand domain and compartments at clinically relevant site. Can be applied to many regions of the genome simultaneously (many loci to all loci) | Output limited to preselected regions. Does not perform well at repeat regions | Was used to validate architecturally well-characterized mouse β-globin and human H19/IGF loci | No specialized pipeline. Uses mainly well-known tools such as BWA, Samtools, and BEDtools |
 Hi-C coupled with RNA bait capture probes (CHi-C) [82] | Provides high-resolution cis-interactome data at clinically relevant loci such as regulatory elements, single nucleotide polymorphisms (SNPs) from GWASs, TAD boundaries or promoters. Important tool for connecting GWAS outcomes to target genes (many loci to all loci) | Difference in hybridization of RNA probes may introduce enrichment bias. RNA probe location is restricted due to restriction enzyme sites and requires tilling of probes, which increases cost | Identification of three cancer-associated gene deserts in cis-interactome [55]. Cis-interactome at 14 colorectal-cancer-risk-associated loci [83]. Many other clinically relevant examples discussed in the review | CHiCAGO tools [109] |
 Promoter capture-Hi-C (p-CHi-C) [7] | Similar to CHi-C, but RNA enrichment baits target all promoters (many loci to all loci) | Similar limitations to CHi-C | A detailed catalogue of 22,000 promoter interactions where autoimmune- and hematological-disorder-related SNPs are significantly enriched [84] | CHiCAGO tools [109] |
 Promoter-anchored chromatin interaction (HiCap) [35] | Similar approach to CHi-C but uses a 4-bp cutter restriction enzyme for improved resolution (many loci to all loci) | Similar limitations to CHi-C | Promoter-anchored interactions for 15,905 promoters in mouse embryonic stem cells (mESCs) | CHiCAGO tools [109] |
DNA-centric view of single-cell genome architectural methods | ||||
 Single-cell genome-wide chromatin conformation capture (single-cell Hi-C) [41] | Can delineate cellular heterogeneity at architectural level. Overcomes limitation of population ensemble average structure from bulk Hi-C (all loci to all loci at single-cell level) | Can be technically more challenging than bulk Hi-C. Data from multiple, individual cells are likely needed for a useful interpretation | Single-cell Hi-C has been used to understand architectural heterogeneity for Th1 cells, cell cycle transition and during oocyte to zygotic transition [110, 111] | Single cell Hi-C Pipeline (scell_hicpipe) [41] |
 Single-cell combinatorial indexing Hi-C (sciHi-C) [42] | Probes cellular heterogeneity by using combinatorial indexing, thus eliminating requirement of single-cell separation using fluorescence-activated cell sorting. Provides rapid scaling for large number of cells. Technically feasible to use for clinically important tissue samples (all loci to all loci at single-cell level) | Comparatively new method; may require optimization compared to bulk Hi-C | sciHi-C data for more than 10,000 single cells was reported. Yet to be explored clinically, but has potential for application to important diseases such as cancer where cellular heterogeneity plays crucial role | Single-cell combinatorial indexing Hi-C pipeline on github [42] |
Protein-centric view of genome architectural methods | ||||
 Chromatin interaction analysis-end tag sequencing (ChIA-PET) [112] | To understand the protein-specific chromatin interactome. Important in identifying chromatin architectural roles for proteins (many loci to all loci) | Requires known/target protein of interest, similar to chromatin immunoprecipitation followed by sequencing (ChIP-seq). Protein may not bind directly to DNA but bind in complex | Used for studying chromatin architecture mediated by estrogen receptor α binding [112] and CTCF [15]. Applied to diseases such as cancer, can provide an understanding of how changes in the binding of these factors alter 3D genome structure and gene expression | ChIA-PET2 data analysis pipeline [113] |
 Hi-C chromatin immunoprecipitation (HiChIP) [37] | Protein-centric view of genome architecture similar to ChIA-PET but more sensitive and requires fewer cells (many loci to all loci) | As above | Identified genome-wide cohesin-mediated looping interactions [37]. Can be used to determine disease-altering looping structure for specific architectural proteins | Uses Hi-C Pro for data processing; Fit-Hi-C, Mango, and Juicer for contact interaction calls; and MACS2 for peak calls [114,115,116,117,118] |
 Proximity ligation assisted chromatin immunoprecipitation (PLAC-seq) [38] | Protein-centric view of genome architecture similar to ChIA-PET, but more sensitive and requires fewer cells (many loci to all loci) | As above | Generated improved maps of promoter–enhancer interactions in mESCs using H3K4me3 mark. Can be used in place of CHi-C methods and does not require probe design/acquisition | PLAC-seq data analysis pipeline [38] |