The 3D chromatin structure is essential for gene regulation and cellular function. Studying the three-dimensional structure of chromatin can help us understand the spatial relationship between distal regulatory elements and target genes (Ling et al., 2022), including how the protein factors associated with enhancers [Enhancers are non-coding sequences in the genome that can activate the expression of target genes transcribed by RNA polymerase II (RNAPII) (Panigrahi and O'Malley, 2021)] promote chromatin interactions, the effect of enhancer-promoter interaction on gene expression, and the importance of histone modifications and transcription factors in the regulation of chromatin accessibility. Therefore, in-depth research on the three-dimensional structure of chromatin can reveal the mechanism of gene expression regulation, thereby pinpointing disease genes that are candidates for novel therapeutics at the genome spatial level.

The 3D chromatin structure refers to the spatial organization formed by the DNA and proteins in the nucleus of the genome. The genome is divided into chromosomal territories, A/B compartments, topologically associating domains (TADs) and chromatin loops through the three-dimensional folding of chromosomes (Figure 1) (Deng et al., 2022). Chromosomal territories refer to the distinct regions occupied by different chromosomes within the cell nucleus, which are often independent and non-overlapping (Cremer et al., 1993). According to the level of transcriptional activity, chromatin is divided into A/B compartments. The A compartment mainly consists of open chromatin, with significantly higher levels of activating histone modifications and transcriptional activity compared to the B compartment (Mohanta et al., 2021). Additionally, the I compartment (Vilarrasa-Blasi et al., 2021; Cuartero et al., 2023) is a highly dynamic intermediate compartment independent of the A and B compartments, and is rich in chromatin structures with stable promoters and repressive chromatin, capable of interacting with the A and B compartments. The compartments contain chromatin interaction domains formed by loop extrusion, known as TADs (Dixon et al., 2012; Rajderkar et al., 2023), and TADs are conserved across species and cell types (Winick-Ng et al., 2021). The boundaries of TADs are primarily maintained by CCCTC-binding factor (CTCF), condensin, cohesin, and histone modifications to ensure stability (Rajderkar et al., 2023). CTCF is a transcriptional regulatory protein that is widely present in eukaryotes. It plays an important role in maintaining the insulation of TAD boundaries. Cohesin can cooperate with CTCF to participate in the formation and regulation of chromatin loops, and it can also regulate the binding and functional effects of CTCF on chromatin (Li et al., 2020). TADs are the fundamental structural units of chromatin and the basic regulatory units of the genome, and they are enriched with various regulatory elements and their target genes internally (Lupianez et al., 2016; Barrington et al., 2019; Wang et al., 2022). New research has found that DNA double-strand breaks damage certain TADs, and these damaged TADs aggregate to form a new chromatin compartment-D compartment, which is targeted by active transcription protein markers within other undamaged TADs (Arnould et al., 2023). Chromatin loops are ring-like structures formed by proteins and other molecules mediating, folding, and encapsulating chromatin. Chromatin loops can facilitate the spatial binding [This binding may be dynamic (Xiao et al., 2021; Zuin et al., 2022)] of promoters and distal regulatory elements, such as enhancers, to regulate gene expression levels (Figure 1) (Mohanta et al., 2021). In addition, long-range loop interactions between promoters and enhancers are common means of regulating gene expression levels, and their activity is associated with a specific combination of colocalization of post-translational histone modifications, also known as “chromatin state.” Current research indicates that active promoters are enriched with H3K27ac and H3K4me2/3, while active enhancers are enriched with H3K27ac and H3K4me1/2 (Miko et al., 2021; Guo and Wang, 2022). In addition, Schoenfelder and Fraser proposed the “selecting–facilitating–specifying” model of chromatin in 2019, which provides a more dynamic explanation for the interactions between enhancers and promoters. This model involves two layers of regulatory processes. The first layer involves the activity of molecules such as transcription factors, which select cell-specific regulatory elements by increasing the accessibility of chromatin regions and depositing marks. The second layer involves compartments, TADs, and loops that bring spatial proximity to regulatory elements with similar activity states and facilitate the contact between enhancers and promoters. These two layers of regulatory mechanisms work together to ensure the specificity and stability of gene expression within the cell (Schoenfelder and Fraser, 2019).

The above classifications of chromatin 3D structure were identified using techniques such as Hi-C. In addition, researchers have also used DamID (Dam identification) or chromatin immunoprecipitation (ChIP) techniques to identify genomic regions that closely interact with the nuclear lamina (NL), known as lamina-associated domains (LADs) (van Steensel and Belmont, 2017). Through comparison and identification, LADs correspond to the B compartment, while inter-LADs correspond to the A compartment. Furthermore, LADs and TADs are very similar, and both contribute to the organization of chromatin’s topological structure (Briand and Collas, 2020). Recent studies have redefined chromatin functional units, namely RAMs, through chromatin immunoprecipitation assays. RAMs reflect the modular organization of the 3D genome and are the functional units of chromatin spatial organization (Zheng and Wang, 2022). Compared to the TADs defined by Hi-C technology, RAMs have a higher proportion of promoter-enhancer clusters, loop anchors, and super-enhancer [Super-enhancers are regulatory regions on chromatin with unusually strong enrichment for the binding of transcriptional co-activators (Pott and Lieb, 2015)] clusters. In addition, the RAMs boundary has high inhibitory histone modifications, such as H3K9me3 and H3K27me3, and the disruption of the RAMs boundary may be closely related to the occurrence of cancer (Zheng and Wang, 2022).

The 3D chromatin structure plays an important role in the regulation of gene expression, cell differentiation and disease progression (Zheng and Xie, 2019). In terms of transcriptional regulation, the formation of chromatin loops and interactions within chromatin regions ensure the proximity of regulatory elements to genes. The formation of compartments and TADs ensures the precision of gene regulation. In terms of genetic stability, the spatial organization of chromatin, particularly the formation of TAD boundaries, to some extent can prevent the occurrence of genetic abnormalities (Hsieh et al., 2020; Zhao et al., 2023). The deletion, inversion, duplication, or misalignment of TAD boundaries will affect its insulation degree and connectivity (Barrington et al., 2019; Watt et al., 2021), and may change the genome structure, so that the enhancers adjacent to TAD will ectopically activate genes, leading to gene misexpression and disease. Congenital limb malformations such as syndactyly, brachydactyly, polydactyly, Liebenberg syndrome, and autosomal dominant adult-onset demyelinating leukodystrophy (ADLD) are all associated with the loss or misregulation of TAD boundaries (Lupianez et al., 2016). Research on the pathogenesis of dilated cardiomyopathy (DCM) indicates that the overexpression of the transcription factor HAND1 leads to a significant increase in enhancer-promoter interactions, causing widespread chromatin reprogramming, ultimately resulting in transcriptional dysregulation and the onset of DCM (Feng et al., 2022). Additionally, in acute myeloid leukemia (AML), the loss of CTCF binding sites induced by high methylation leads to the loss of TAD insulation, enhancing chromatin interactions in AML, which may be associated with AML induction (Xu et al., 2022). Research on pancreatic tumors has found that the chromatin accessibility is increased near the active genes of mutated kras cells in malignant tumors, significantly increasing their epigenetic plasticity and driving tumorigenesis (Burdziak et al., 2023). Researchers have also observed the reorganization of the A, B, and I compartment in samples of colorectal cancer. These topological changes suppress the tumor’s stemness and invasive programs while inducing the expression of anti-tumor immune genes, thus inhibiting malignant tumor progression (Johnstone et al., 2020).

At present, there are a variety of methods to study the 3D chromatin structure, and they are also developing and improving. The earliest technology used by researchers was DNA FISH (Price, 1993), which allows for the direct detection of interactions between a small number of loci within individual cells. This technology is currently developing rapidly, and when combined with super-resolution microscopy, the resolution is greatly improved. Additionally, the probe design and signal intensity have been enhanced. However, the standardization of this technology is relatively poor, and image analysis requires experienced and highly skilled personnel (Wagner and Haider, 2012; Prudent and Raoult, 2019). Later, with the continuous advancement of technology, 3C (Chromosome Conformation Capture), Capture-C, 4C (Circular Chromosome Conformation Capture) and 5C (Chromosome Conformation Capture Carbon Copy) technologies have emerged one after another.

3C (Dekker, 2006) is a technique used to detect the interaction frequency between different DNA fragments in chromatin. But the 3C technology requires rigorous operation to obtain correct 3C data. The three quality controls in the middle include: PCR efficiency, assessing the level of background random collisions, and data normalization (Dekker, 2006). Moreover, when the 3C technology is applied, the detection of the interaction fragment is limited to specific genomic intervals. In order to screen candidate fragments for interaction with target fragments without preference on a genome-wide scale, the 4C technique was developed by researchers (Simonis et al., 2006). 4C technique uses specific primers to enrich DNA fragments that interact with the target locus, enabling the study of interactions between a specific locus and the rest of the genome (Simonis et al., 2006; Göndör et al., 2008). Subsequently, researchers developed the 5C technique based on 3C, which expanded the scope of 3C applications and can be used to map the cis and trans interactions of large-scale genomic elements, as well as for deeper studies of chromosomal structures (Dostie et al., 2006). In this context, Hi-C technology came into being in 2009 (Lieberman-Aiden et al., 2009) and has become an advanced technology for studying the 3D chromatin structure, and has been widely used. Although the techniques in the 3C family are continually being upgraded and improved, their core methodological steps remain the same: crosslinking, restriction digest, and proximity ligation (McCord et al., 2020).

Furthermore, the interaction between enhancers and promoters is crucial for gene regulation. In addition to chromatin contacts, open chromatin regions can also facilitate this interaction (Bendl et al., 2022). Chromatin accessibility techniques and methods to study protein-chromatin interactions are of great significance for studying the spatial 3D structure of chromatin. High-throughput chromatin accessibility sequencing technology has become an important tool for studying open chromatin regions, such as ATAC-seq, DNase-seq, MNase-seq, scATAC-seq, etc. Among them, ATAC-seq can efficiently label and sequence open regions on chromatin by utilizing the action of Tn5 transposase (Grandi et al., 2022). DNase-seq uses DNase I, an endonuclease, to cleave DNA sequences and obtain information about open regions of chromatin through sequencing analysis (Lv et al., 2023). MNase-seq, similar to DNase-seq, uses MNase, an enzyme, to obtain chromatin accessibility information. In addition, ATAC-seq combined with single-cell sequencing technology gave birth to scATAC-seq, which is characterized by the determination of chromatin accessibility and genomic sequences within a single cell. The biggest difference between this technique and conventional ATAC-seq is that individual cells need to be labeled to ensure specificity for each cell (Wang et al., 2023). Techniques for identifying specific protein binding sites include ChIP-seq and CUT&Tag. ChIP-seq is a genome-wide high-throughput sequencing technique to study the binding of proteins to target DNA sequences, providing a method for epigenetic studies such as transcription factors and histone modifications (Nakato and Sakata, 2021). CUT&Tag is a novel technology that has been improved based on ChIP-seq, significantly enhancing the signal-to-noise ratio of histone marks by utilizing the activity of Tn5 transposase, while also offering the advantages of low cost and high efficiency (Kaya-Okur et al., 2020). Of particular note, CUT&Tag technology has the capability to handle low-cell-number samples, thus holding broad potential for applications in single-cell chromatin analysis, offering a new avenue for studying chromatin organization. In addition, the ABC computational model can be used to calculate the contact frequency between cell-specific gene promoters and enhancers, and can predict the connections between enhancers and genes (Fulco et al., 2019; Nasser et al., 2021). The above-mentioned tools can be used to reveal mechanisms of changes in gene expression levels. Additionally, direct techniques for detecting gene expression levels include RNA-seq (Wang et al., 2009), among others.

Furthermore, recently developed single-cell multi-omics detection methods, such as Paired-Tag and scREG (Zhu et al., 2021; Duren et al., 2022), have emerged. Both of these techniques enable integrated analysis of the transcriptome and chromatin accessibility within single cells (Figure 2).

Hi-C is a technology based on 3C and utilizes high-throughput sequencing to detect chromatin interactions across the entire genome (Lafontaine et al., 2021). Hi-C technology can unveil the intricate folding structure of chromatin, which is closely linked to the proportion of cells in a population that experience specific contacts. Additionally, this method allows for the study of interactions within chromatin at a nanometer-scale distance. However, Hi-C is limited in its ability to determine only the relative distance and interaction frequency between different regions on the chromosome. As a result, the data can only represent the relative contact frequency between these regions. Furthermore, the lack of standardization within Hi-C hinders the comparison of data across different conditions and cell types. At the same time, the standard Hi-C method can only be used to detect the interaction between two specific regions on the chromosome, and cannot reveal whether multiple regions simultaneously or mutually repel each other (McCord et al., 2020).

Hi-C includes a biotinylation step between digestion and ligation, giving Hi-C a better ability to enrich for ligated junctions compared to 3C, making it more favorable for the analysis of chromosomal spatial structure. In 2012, the DNase Hi-C technology was developed (Duan, 2021). Compared to Hi-C, the major improvement of this technique lies in the use of DNase I for chromatin cleavage, overcoming the limitations of low digestion efficiency and accuracy associated with the traditional Hi-C method (Ma et al., 2015). In 2013, Micro-C was proposed as a complementary method to Hi-C. This technique uses micrococcal nuclease instead of the restriction enzyme used in Hi-C to fragment chromatin, and is suitable for short-range analysis of nucleosome fiber folding (Hsieh et al., 2015). In 2013, single-cell Hi-C was used, which is able to visualize the three-dimensional structure of chromatin within a single cell with specificity (Nagano et al., 2013). However, this technique cannot reveal the interactions between each restriction fragment and all of its spatially proximal loci, as bulk Hi-C does (Nagano et al., 2013). This problem can be addressed by the Multiplex-GAM technology mentioned later in the text (Beagrie et al., 2023). In 2014, the concept of In situ Hi-C was proposed, which reduces false positive results, improves the signal-to-noise ratio, and shortens the experimental cycle compared with Hi-C (Rao et al., 2014). In 2017, DLO Hi-C technology was developed. Compared to Hi-C, this technique offers the advantages of lower cost, simplified library construction, shorter processing time, and higher yield of effective data (Zhang et al., 2020a). Also in 2017, Hi-C 3.0 was utilized (Akgol Oksuz et al., 2021). Compared to Hi-C, this technique offers higher resolution and improved reproducibility, allowing for the detection of thousands of loops. In addition, Capture-C is also a high-resolution technique based on 3C for detecting chromatin interactions, which allows for the extraction of restriction fragments of interest from in situ 3C material, such as promoters, enhancers, and super-enhancers (Downes et al., 2022). The emergence of these emerging technologies enables scientists to identify interaction patterns between different genes, transcription factors, and regulatory elements, uncover key genes or transcription factors, and further interpret gene function (Zhang et al., 2013).

In recent years, many new studies have demonstrated further optimized techniques related to chromatin conformation capture. Micro-Capture-C (MCC) is a technique that combines Micro-C with Capture-C, with high resolution that allows precise localization to individual base pairs (Crump et al., 2023). Region Capture Micro-C (RCMC) is a method that combines region capture with Micro-C, which resolves highly nested and focal interactions, most of which are nested interactions between promoter and enhancer regions, and labels these interactions as “microcompartments.” Additionally, RCMC technology can mine sequencing-limited data to generate high-depth 3D genome maps (Goel et al., 2023). Multiplex-GAM is a technology used for genome-wide interaction analysis, capable of detecting multiple genomic regions missed by Hi-C (Beagrie et al., 2023). Moreover, this technology can provide a comprehensive understanding of the 3D chromatin structure at the single-cell level (Beagrie et al., 2023). Pore-C, a chromatin conformation-capturing technique combined with nanopore sequencing of linkers, enables faster analysis of proximal higher-order chromatin contacts at the genome scale (Deshpande et al., 2022). Compared to Hi-C, which can only detect interactions between paired genomic loci, this technique can detect interactions between two or more DNA loci by using long-read sequencing (Deshpande et al., 2022).

HiChiP is a technique that combines Hi-C and chromatin immunoprecipitation (ChIP), allowing for the simultaneous detection of chromatin interactions and protein binding sites on chromatin. Specifically, H3K27ac HiChIP can detect interactions at higher resolution within promoters and enhancers (Mumbach et al., 2017). Promoter capture Hi-C (PCHi-C) is a derivative technique primarily used to identify the interaction regions between enhancers and promoters. It is of significant importance for studying the relationship between genome regulatory mechanisms and disease occurrence. When this technology is used in conjunction with expression quantitative trait loci (eQTL) analysis, it can better reveal the relationship between disease gene expression and chromosomal spatial structure (Javierre et al., 2016; Baxter et al., 2018).

The analysis process of Hi-C data typically includes the following steps (Ay and Noble, 2015): first, the established Hi-C library is subjected to quality control filtering to obtain high-quality data. During the data processing, the data is aligned to the desired genome following the principle of paired-end unique mapping, and an interaction matrix is generated. Subsequently, the matrix is subjected to normalization, followed by the detection of chromatin interactions. Finally, the data is visualized for further analysis (Figure 2). In constructing the Hi-C matrix, we typically use two tools, Juicer (Durand et al., 2016) and HiC-Pro (Servant et al., 2015). Both Juicer and HiC-Pro can directly convert raw data into normalized contact maps (Servant et al., 2015; Durand et al., 2016). Juicer can automatically annotate the three-dimensional structure of chromatin, while HiC-Pro can construct allele-specific contact maps using phased genotype data (Servant et al., 2015; Durand et al., 2016). When analyzing chromatin structure, a series of tools are typically used to detect structures at different levels. For compartment detection, tools such as HiCSeg (Lévy-Leduc et al., 2014) and Fit-Hi-C2 (Kaul et al., 2020) are employed. For TAD detection, we use tools such as TADbit (Serra et al., 2017) and TADpole (Soler-Vila et al., 2020). The most significant feature of TADbit is its ability to complete the entire process of Hi-C data processing and analysis. It is important to note that when analyzing interaction matrices, this method uses a breakpoint detection algorithm to calculate the position of TADs along chromosome boundaries and employs the Needleman-Wunsch algorithm to align TADs (Serra et al., 2017). TADpole’s advantage lies in its ability to identify hierarchical structural domains in interaction matrices and robustness in terms of data resolution, normalization strategy, and sequencing depth (Soler-Vila et al., 2020). For the detection of chromatin loops, we often use tools such as HiCCUPS (Durand et al., 2016), GOTHiC (Mifsud et al., 2017), and HiCExplorer (Wolff et al., 2022). In fact, the scope of these tools is not absolute; some tools can detect multiple chromatin substructures. Table 1 provides a comparison of the prominent features of these analysis tools.

In addition, deep learning techniques have been applied to computational 3D genomics. Deep learning is a machine learning method that simulates the structure and function of the human brain’s neural network, and its greatest feature is “learning from what is seen” (Li et al., 2023). Based on the amount of labeled data used for training, deep learning can be divided into three modes: supervised learning, unsupervised learning, and semi-supervised learning. Supervised learning uses all labeled data as the training set, and the trained model can predict the remaining unlabeled data. Convolutional Neural Network (CNN) is an example of this learning mode. Unsupervised learning uses unlabeled data as the training set and aims to discover hidden information in the data, as well as to perform functions such as dimensionality reduction, clustering, and annotation. Autoencoders, Generative Adversarial Networks (GAN), and self-supervised learning are examples of unsupervised learning. Semi-supervised learning involves training with a small amount of labeled data and a large amount of unlabeled data, resulting in a model with enhanced performance and generalization capabilities (Srinidhi et al., 2021; Chen et al., 2022). Currently, deep learning has been applied in various fields, and in the study of chromatin three-dimensional structure, it has primarily been used to enhance the resolution of Hi-C matrices and to predict 3D spatial structures, including compartments, TADs, and chromatin loops, using high-resolution matrices. The deep learning techniques for enhancing the resolution of Hi-C data include HiCPlus (Zhang et al., 2018), HiCNN [Liu and Wang (2019a)], HiCNN2 [Liu and Wang (2019b)], SRHiC [Li and Dai (2020)], VEHiCLE [Highsmith and Cheng (2021)]. The high-resolution matrix prediction models based on GAN include hicGAN (Liu et al., 2019), DeepHiC (Hong et al., 2020), EnHiC [Hu and Ma (2021)], HiCARN [Hicks and Oluwadare (2022)]. Other deep learning models include iEnhance (Li et al., 2023) and Orca (Zhou, 2022). The specific content is as shown in Table 2.

The Hi-C technology is a method for studying chromatin structure, which can be used to investigate the relationship between chromatin structure changes and gene expression. It can be applied to study gene regulation and the mechanisms of disease occurrence (van Berkum et al., 2010). In terms of cell differentiation, scientists have studied the interaction patterns between enhancers and promoters across the entire genome during the differentiation process of human primary keratinocytes (Gong et al., 2021). In the context of cancer, researchers have compared the TAD boundaries and quantities between myeloma cells and normal cells (Gong et al., 2021). In mammalian embryonic development, scientists have investigated the strength of chromatin interactions between mammalian sperm and egg cells at various stages of division (Ke et al., 2017). Furthermore, scientists have used Hi-C to accurately locate the centromeres in the yeast genome (Varoquaux et al., 2015). With the continuous development of Hi-C technology, it is gradually gaining importance in various fields such as agricultural science, life science, and medical science.

In addition, researchers can gain in-depth understanding of the spatial organization and interaction patterns of the genome in immune cells through the use of Hi-C and other high-throughput sequencing technologies, thereby revealing the complex network of gene regulation. Hi-C can reveal interactions between elements on the chromatin loop in immune cells, interactions of genes within TADs, as well as interactions between compartments, among others. Research on the spatial organization and interactions of the genome is crucial for unraveling the functionality of the immune system and the mechanisms underlying related diseases. Next, we will further explore the chromatin 3D structure of immune cells and important findings related to immune system functionality and disease mechanisms.