The Structural And Functional Studies Of Topologically Associating Domains

The impact of three-dimensional genome organization on transcriptional regulation and thereby on cellular development and diseases is now widely accepted.In the past,the research of nuclear organization mainly relied upon microscopy.Recently,the chromosome conformation capture(3C)technique and its derivatives,especially Hi-C,which is able to capture interactions on a genome-wide scale,have enabled the analysis of nuclear organization at an unprecedented resolution and throughput,and gradually unveiled the hierarchies of the spatial organization of chromosomes.Topologically associating domain(TAD)is an important structural unit in this hierarchy.TADs have been identified in a wide range of species.Comparative studies have shown that the genomic positions of TADs remain stable among different cell types and even species.As for biological functions,TADs provide the structural contraints for enhancer-promoter interactions,and act as the basic units of transcription regulation,DNA recombination and replication-timing regulation.However,people know little about the inner structures of TADs.In this work,we explored the chromatin interaction patterns inside TADs,and proposed a quantitative parameter for measuring the overall structural characteristics of TADs and a novel algorithm for detecting hierarchical TADs,to investigate the structures and functions of TADs in a systematic manner.First,we analyzed the aggregation degree of statistically significant interactions within TADs and defined a theoretical parameter called aggregation preference(AP)which was able to measure the overall structural characteristics of TADs.Then we further explored the associations between TAD structures and biological functions based on this parameter.Specifically,we analyzed 11 data sets on 9 human and mouse cell types generated by traditional Hi-C and in-situ Hi-C,and found that the AP values calculated from each data set were widely distributed between 0 and 1,suggesting that TADs were highly heterogeneous in all cell lines.By integrating data of DNA sequences,epigenomic signals and gene expressions,we found that this structural heterogeneity was significantly correlated to functional activities.Finally,comparisons between cell types showed that the structural rearrangements within TADs were tightly correlated to transcription remodeling during cell changes.Second,the TAD itself is organized hierarchically.Therefore,to study different roles of domain hierarchies in higher-order chromosomal organizations and biological functions,it's pivotal to develop the algorithm for hierarchical TAD detection.To solve this problem,we refined the traditional TAD definition and further constrained TADs to the optimal domains separating global intra-chromosomal interactions under given objective function.Sub-TADs were defined recursively as optimal domains separating intra-domain interactions in a similar way.Both local and long-range chromatin interaction information were considered.Based on above definitions,we developed a novel algorithm called Hi TAD to detect hierarchical TADs.Then we cross-validated Hi TAD by using traditional and in-situ Hi-C data sets on 7 human and mouse cell types from different perspectives.First,Hi TAD outperformed 2 existing softwares Arrowhead and TADtree in domain sensitivity,replicate reproducibility and inter cell-type conservation;second,boundaries of all hierarchies detected by Hi TAD shared similar properties with traditional TAD boundaries,such as insulation effects,signal enrichments and CTCF motif directions.To facilitate systematic comparisons between two domain sets,we developed a novel alignment algorithm for hierarchical domains.Based on this algorithm,we defined several types of “boundary-level changes” and “domain-level changes”.As for boundary-level analysis,we found it was TADs but not Sub-TADs that mainly separated higher-order chromosomal compartments and replication timing domains.As for domain-level analysis,we found TADs and Sub-TADs differed significantly in transcription regulations.In conclusion,we proposed the parameter AP to represent the overall aggregation degree of chromatin interactions,and the Hi TAD algorithm to resolve the hierarchical details of domains within TADs.Based on these two computational pipelines,we explored the relationship between TAD structures and biological functions from different perspectives.