3D disorganization and rearrangement of genome provide insights into pathogenesis of NAFLD by integrated Hi-C, Nanopore, and RNA sequencing

Abstract

The three-dimensional (3D) conformation of chromatin is integral to the precise regulation of gene expression. The 3D genome and genomic variations in non-alcoholic fatty liver disease (NAFLD) are largely unknown, despite their key roles in cellular function and physiological processes. High-throughput chromosome conformation capture (Hi-C), Nanopore sequencing, and RNA-sequencing (RNA-seq) assays were performed on the liver of normal and NAFLD mice. A high-resolution 3D chromatin interaction map was generated to examine different 3D genome hierarchies including A/B compartments, topologically associated domains (TADs), and chromatin loops by Hi-C, and whole genome sequencing identifying structural variations (SVs) and copy number variations (CNVs) by Nanopore sequencing. We identified variations in thousands of regions across the genome with respect to 3D chromatin organization and genomic rearrangements, between normal and NAFLD mice, and revealed gene dysregulation frequently accompanied by these variations. Candidate target genes were identified in NAFLD, impacted by genetic rearrangements and spatial organization disruption. Our data provide a high-resolution 3D genome interaction resource for NAFLD investigations, revealed the relationship among genetic rearrangements, spatial organization disruption, and gene regulation, and identified candidate genes associated with these variations implicated in the pathogenesis of NAFLD. The newly findings offer insights into novel mechanisms of NAFLD pathogenesis and can provide a new conceptual framework for NAFLD therapy.

Keywords: 3C, chromosome conformation capture; 3D genome; 3D, three-dimensional; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Abcg5, ATP-binding cassette sub-family G member 5; BWA, Burrows-Wheeler Aligner; CNV, copy number variation; Camk1d, calcium/calmodulin-dependent protein kinase type 1D; Chr, chromosome; Chromatin looping; DEG, differentially expressed gene; DEL, deletion; DI, directionality index; DUP, duplication; Elovl6, elongation of very long chain fatty acids protein 6; FDR, false discovery rate; FFA, free fatty acid; Fgfr2, fibroblast growth factor receptor 2; GCKR, glucokinase regulator; GO, gene ontology; GSH, glutathione; Gadd45g, growth arrest and DNA damage-inducible protein GADD45 gamma; Grm8, metabotropic glutamate receptor 8; Gsta1, glutathione S-transferase A1; H&E, hematoxylin-eosin; HFD, high-fat diet; HSD17B13, hydroxysteroid 17-beta dehydrogenase 13; Hi-C, high-throughput chromosome conformation capture; IDE, interaction decay exponent; INS, insertion; INV, inversion; IR, inclusion ratio; IRGM, immunity related GTPase M; IRS4, insulin receptor substrate 4; KEGG, Kyoto Encyclopedia of Genes and Genomes; Kcnma1, calcium-activated potassium channel subunit alpha-1; LPIN1, lipin 1; MBOAT7, membrane bound O-acyltransferase domain containing 7; MDA, malondialdehyde; NAFLD, non-alcoholic fatty liver disease; NF1, neurofibromin 1; NGS, next-generation sequencing; NOTCH1, notch receptor 1; ONT, Oxford Nanopore Technologies; PCA, principal component analysis; PNPLA3, patatin like phospholipase domain containing 3; PPP1R3B, protein phosphatase 1 regulatory subunit 3B; PTEN, phosphatase and tensin homolog; Pde4b, phosphodiesterase 4B; Plce1, 1-phosphat-idylinositol 4,5-bisphosphate phosphodiesterase epsilon-1; Plxnb1, Plexin-B1; RB1, RB transcriptional corepressor 1; RNA-seq, RNA-sequencing; SD, standard deviation; SOD, superoxide dismutase; SV, structural variation; Scd1, acyl-CoA desaturase 1; Sugct, succinate-hydroxymethylglutarate CoA-transferase; TAD, topologically associated domain; TC, total cholesterol; TG, triglyceride; TM6SF2, transmembrane 6 superfamily member 2; TP53, tumor protein p53; TRA, translocation; Topologically associated domain; Transcriptome; WGS, whole-genome sequencing; Whole-genome sequencing.

Graphical abstract

Figure 1

The reprogrammed liver transcriptome in C57BL/6 mice with NAFLD. (A) The levels of serum marker of mice in normal and NAFLD group, including ALT, AST, MDA, SOD, GSH, TC, TG, and FFA (mean ± SD, n = 10/group, ∗∗P < 0.01 vs. control). (B) The morphological changes in the livers of NAFLD mice assessed by H&E and oil red O staining. Scale bar: 50 μm. (C) Heatmap showing the Pearson correlation of gene expressions in RNA-seq across liver samples from control (C1, C2, and C3) and NAFLD model (M1, M2, and M3) groups. (D) The MA plot of differentially expressed genes (DEGs) identified by RNA-seq between control and NAFLD groups, with absolute value of log₂ (fold change) > 1 and adjusted P < 0.01 vs. control (mean ± SD, n = 3). Green dots represent significantly downregulated genes in model group, red dots represent significantly upregulated genes in model group, and black dots represent genes with on significant difference in expression between two groups. (E) The top 10 most significant biological process terms in GO enrichment within DEGs. Abscissa represents the number of DEGs enriched in the corresponding biological process, while color of the column represents the adjusted P value by hypergeometric test. (F) The top 20 enriched KEGG pathways within DEGs. Abscissa represents the rich factor, which is the ratio of the proportion of DEGs annotated to the pathway to all genes annotated to the pathway. The ordinate is −log₁₀ (Q value) that represents the enrichment significance of DEGs in this pathway, where Q is the adjusted P value after multiple hypothesis test.

Figure 2

Mapping 3D chromatin conformation of the liver tissues of mice in control and NAFLD group by Hi-C. (A) The distributing proportion of cis- and trans-interaction in the Hi-C libraries obtained from control and NAFLD mice. Liver tissues from 3 mice in each group were mixed equally, and two Hi-C libraries were established for each group. Cis-close, the cis-interactions with the linear distance between two terminals was less than 20 kb; cis-far, the cis-interaction with the linear distance between two terminals was more than 20 kb; Trans, the ends of the read aligned to different chromosomes. (B) The cis/trans-interaction ratio in each chromosome in the established Hi-C libraries. (C) The interaction frequency heatmaps between chromosomes of the livers from mice in control and model groups. Red color in heatmap suggests that the distance between the two chromosomes is relatively close and there is a stronger interaction; while blue color indicates a closer distance between the two chromosomes and a weaker interaction. Some obvious differences between control and model groups are outlined in yellow. (D) Hi-C heatmap of genome-wide interaction pattern of liver samples from control and NAFLD mice. (E) The representative Hi-C interaction matrices of Chr 2, Chr 3, and Chr 4.

Figure 3

The changes in A/B compartment and the related DEGs in the livers of mice with NAFLD by Hi-C and RNA-seq. (A) The distributions of A/B compartment on Chr 4, Chr 9, and Chr 13. Red represents A compartment and blue represents B compartment. (B) The distribution ratio of A/B compartment on each chromosome in control and model group. Red represents A compartment and blue represents B compartment of control group, whereas green represents A compartment and orange represents B compartment of NAFLD model group. (C) The gene number distributed on A and B compartments in the livers of mice in control (C) and NAFLD (M) groups. ∗∗P < 0.01 vs. A compartment (mean ± SD, n = 3). (D) The gene expressions distributed on A and B compartments in RNA-seq. ∗∗P < 0.01 vs. A compartment (mean ± SD, n = 3). (E) The changes of A/B compartment in the livers of mice with NAFLD. Undefined means chromosome segment that impossible to be identified as A or B compartment due to the insufficient coverage or the inconsistency in biological duplication of the Hi-C data. (F) The distributions of A/B switching compartment in each chromosome. “--” represents undefined compartment.

Figure 4

Changes in TADs and the TAD-impacted DEGs in the livers of mice with NAFLD identified by Hi-C and RNA-seq. (A) The PC1, IR, and DI scores of Chr 6: 32217279–35110067 bp in control and NAFLD groups generated by Hi-C. (B) Hi-C contact matrix and TADs (black triangle) in Chr 3 in control and NAFLD groups. (C) Number of TAD boundaries identified by Hi-C in control and NAFLD groups. (D) The distribution number of TAD boundaries in each chromosome in control (C) and NAFLD (M) groups. (E) The DEGs and unchanged-genes ratios associated with the unique TAD boundary in each group and the unchanged boundary in both samples. (F) Example depicting relationship between gene expression and TAD on Chr 6: 32217279–35110067 bp. Black triangles in triangular contact matrix indicate the TADs. Red shadow displays changes in TAD boundary affected gene expression in NAFLD, and blue shadow shows that gene expression was also related to the alternation of domains in addition to TAD boundary.

Figure 5

Changes in chromatin looping and the related DEGs in NAFLD identified by Hi-C and RNA-seq. (A) Number of chromatin loop identified by Hi-C in control and NAFLD groups. (B) The distribution number of TAD boundaries in each chromosome in control (C) and NAFLD (M) groups. (C) Distribution of chromatin loop in each chromosome in control (green) and NAFLD (red) groups. (D) Number of different types of chromatin looping. (E) Number of chromatin looping with anchor located on enhancer (E) and promoter (P).

Figure 6

Changes in structure variants (SV) in NAFLD identified by Nanopore sequencing. (A) The representative distribution (C1 group) of sequencing depth. Distributions of the other samples are shown in Supporting Information Fig. S7B. (B) The representative distribution (C1 group) of read coverage. Distributions of the other samples are shown in Supporting Information Fig. S8. (C) Number of different type of SV identified by Nanopore sequencing, including deletion (DEL), insertion (INS), translocation (TRA), inversion (INV), and duplication (DUP). (D) Changes in the SV number in each chromosome. (E) Distributions of different type of SV in each chromosome.

Figure 7

Gsta1 and Camk1d as the candidate pathogenic genes of NAFLD caused by reorganization of chromatin structure identified by the comprehensive analysis of Hi-C, Nanopore, and RNA sequencing. (A) Summary of the 100 highest chromatin interactions in Hi-C and SV data in Nanopore sequencing in control and NAFLD groups: (a) The chromosome number; (b) The distribution of SVs (DEL is shown in green, INS is shown in orange, and DUP is shown in red); (c) Copy number data from Nanopore sequencing (gain and amplification are shown in red dot, loss and deletion are shown in blue dot, and the normal CVs are shown in gray dot). In the innermost layer, lines represent the highest 100 chromatin contact identified by Hi-C. (B) Mapping of local neighborhood interactions of Gsta1. (C) Mapping of local neighborhood interactions of Camk1d. The results of contact matrix, triangular, domain, and loop reflect the changes in spatial organization generated by Hi-C. SV and CNV were obtained from Nanopore sequencing. The abundance of gene expressions was generated by RNA-seq.