Chromatin Compartmentalization Analysis And Tool Development

The three-dimensional structure of chromatin is closely and intrinsically linked to gene regulation and cell function.Chromosome Conformation Capture(3C)-based experiments show that eukaryotic chromosomes are divided into A/B compartments in threedimensional structure and coincide with euchromatin/heterochromatin,respectively.Thus,A/B compartments are associated with gene expression and are also cell-specific.Traditionally,A/B compartments are determined by the first principal component obtained by Principal Component Analysis(PCA)of the contact matrix.However,this A/B compartments scheme,which simply divides the chromosome in two,is hardly a complete and accurate representation of chromatin function.To address this issue,we first developed a new chromatin compartmentalization analysis tool,MOSAIC(MOdularity-and Singular v Alue decomposition-based Identification of Compartments),targeting chromatin compartmentalization structures,and then applied it to different cell lines to analyse the relationship between chromatin structure and function.In the first study,Singular Value Decomposition(SVD)was performed on the intrachromosomal interaction matrix obtained by High-throughput Chromosome Conformation Capture(Hi-C).The obtained Eigen Vectors(EVs)were analysed and the two EVs with the highest modularity scores,EV1 and EV2,were found to be highly correlated with the structure and function of chromatin.Based on this,MOSAIC used the information of EV1 and EV2 to perform more accurate and fine compartmentalization of chromosomes.In addition to identifying compartmental states A1 and B1,which correspond to the conventional A/B compartments,MOSAIC also identified two additional compartmental states A2 and B2,corresponding to chromatin with low transcriptional activity and transcriptional repressed chromatin enriched for H3K27me3 marks,respectively.Genome-wide analysis revealed that compartmental states A2 and B2 typically correspond to genomic regions of 100-300 Kb in length,which are much shorter than conventional A1 and B1,and are therefore referred to as micro-compartments.Although micro-compartments cover only about 30% of the genomic region,they are highly dynamic across different cell types.More importantly,distinguishing the microcompartments underpins accurate characterization of chromatin structure-function relationship compared to A/B compartment schemes.In contrast,existing subcompartment schemes based on inter-chromosomal interactions more reflect the position of different subcompartments within the nucleus and are therefore less accurate in both the structure and function of individual chromosome than the compartmental states described by MOSAIC.In the second study,the ability of MOSAIC to analyse chromatin compartmentalization was verified on several cell lines,and it was found that chromosomes in most of the cell lines studied could be consistently divided into four compartmentalized states,and five compartmentalized states were only identified on some chromosomes of human colon cancer cell HCT116.In the comparison of two similar cell lines,GM12878 and K562,MOSAIC was able to identify more chromatin structure-function associations compared to the A/B compartments scheme.In particular,the expression of cell-specific genes such as CD86,ILDR1 and GATA2 exhibited consistency with compartmental states,suggesting an important role of chromatin high-order structure behind disease and transcriptional regulation.For the HCT116 cell line,MOSAIC identified a novel compartmental state C on some of its chromosomes corresponding to heterochromatin highly enriched in H3K9me2 and H2 A.Z.Furthermore,based on the analysis of MOSAIC,we identifies two opposite changes in chromatin compartmentalization in HCT116 cells with knockdown of cohesin: chromatin compartmentalization is enhanced in H3K9me2-enriched chromatin;chromatin compartmentalization is diminished in H3K9me3-enriched chromatin.Therefore,based on the liquid-liquid phase separation(LLPS)mechanism,we proposes that the main cause of the chromatin compartmentalization changes in this experiment may be directly caused by the decrease in the level of H3K9me2 modification,rather than the dominant view of a competitive relationship between chromatin loops and compartmentalization.The MOSAIC developed in this paper is a multi-compartmental states analysis tool developed based on intrachromosomal interaction data,which is able to identify multiple function-specific compartmental states on an individual chromosome and is an accurate description of chromatin compartmentalization.Compared to the coarse splitting of chromosomes into two by the A/B compartments scheme,the precise compartmentalization description can reveal more chromatin structure-function associations in cell comparisons.In addition,for HCT116 cells with knocked out cohesin,accurate compartmentalization descriptions can also fully capture the different changes in chromatin compartmentalization that occur in different compartmental states.Thus,accurate compartmentalization descriptions are important for clarifying the relationship between chromatin structure and function.