Multi-omics characterization of a diet-induced obese model of non-alcoholic steatohepatitis

Abstract

To improve the understanding of the complex biological processes underlying the development of non-alcoholic steatohepatitis (NASH), a multi-omics approach combining bulk RNA-sequencing based transcriptomics, quantitative proteomics and single-cell RNA-sequencing was used to characterize tissue biopsies from histologically validated diet-induced obese (DIO) NASH mice compared to chow-fed controls. Bulk RNA-sequencing and proteomics showed a clear distinction between phenotypes and a good correspondence between mRNA and protein level regulations, apart from specific regulatory events discovered by each technology. Transcriptomics-based gene set enrichment analysis revealed changes associated with key clinical manifestations of NASH, including impaired lipid metabolism, increased extracellular matrix formation/remodeling and pro-inflammatory responses, whereas proteomics-based gene set enrichment analysis pinpointed metabolic pathway perturbations. Integration with single-cell RNA-sequencing data identified key regulated cell types involved in development of NASH demonstrating the cellular heterogeneity and complexity of NASH pathogenesis.

Figure 1

Histological characterization of DIO-NASH model – validation of NASH pathology. (A) Representative images of Hematoxylin & Eosin (H&E) and Pico-Sirius Red (PR) stainings, and immunohistochemical stainings of collagen 1a1, α-SMA and galectin-3. (B) Histological assessment of steatosis, inflammation and NALFD activity score (H&E) and fibrosis stage (PR). (C) Quantitative histochemical assessment of liver lipid content (p < 0.001) (H&E), collagen 1a1 (p = 0.042), α-SMA (p < 0.001), galactin-3 positive cells (p < 0.001) (% of fractional area), whole body weight (p < 0.001) and liver weight (p < 0.001). In panel C, data are expressed as the mean of n = 4-5 ± SEM. A student’s t-test was used. *p < 0.05, **p < 0.01, ***p < 0. 001 vs. CHOW.

Figure 2

Model validation using transcriptomics and proteomics. (A) mRNA expression of genes (prototypical gene nomenclature) associated with NASH development in chow and DIO-NASH mice. (B) mRNA expression and protein abundance of histological markers Col1a1, α-SMA and Lgals3. (C,D) Principal component analysis (PCA) of the 500 most variable genes (C) and proteins (D). (E) Venn diagram showing total number of identified genes (red) and proteins (blue), the total amount of differentially expressed (DE) genes and proteins and the overlap between total number of identified and DE genes and proteins. (F) Correspondence between mRNA and protein expression. Black dots represent common differentially expressed genes and proteins (adjusted p < 0.05). In panel (A), data is presented as standardized relative expression. In panel (B), data is presented as fold change of mean of n = 4-5 ± SEM (DIO-NASH/CHOW) (relative expression). * adjusted p < 0.05, ** adjusted p < 0.01, *** adjusted p < 0. 001 vs. CHOW. In panel (F), data is presented as log2 fold change (DIO-NASH/CHOW). For all panels (A–F), n = 4-5.

Figure 3

Global pathway analysis of transcriptomic and proteomic data. (A) Global transcriptome and proteome perturbations according to enrichment of individual gene sets in the Reactome pathway database indicating perturbed pathways in DIO-NASH vs. CHOW. (B) Transcriptome and proteome perturbations of Immune system sub-pathways. In panels (A,B), perturbed pathways are ranked according to level of statistical significance (mRNA only). The degree of perturbation for each pathway is normalized and presented as the relative regulation of the maximum p-value for each dataset.

Figure 4

Metabolic and fibrotic pathway profiling. (A) Transcriptome and proteome perturbations of Metabolism sub-pathways. (B) mRNA expression and protein abundance of genes in the Metabolism pathway that are differentially expressed only on protein level. (C) Transcriptome and proteome perturbations of Transport of small molecules sub-pathways. (D) mRNA expression and protein abundance of genes in the Transport of small molecules pathway that are differentially expressed only on protein level. (E) Transcriptome and proteome perturbations of Extracellular matrix organisation sub-pathways. (F) mRNA and protein expression of genes in Extracellular matrix organisation pathway that are differentially expressed only on mRNA level. In panels (A,C,E), perturbed pathways are ranked according to level of statistical significance (mRNA only). The degree of perturbation for each pathway is normalized and presented as the relative regulation of the maximum p-value for each dataset. In panels (B,D,E), data is presented as the fold change of the mean of n = 4–5 ± SEM (DIO-NASH/CHOW) (relative expression). * adjusted p < 0.05, ** adjusted p < 0.01, *** adjusted p < 0. 001 vs. CHOW in RNAseq data. ^# adjusted p < 0.05, ^## adjusted p < 0.01, ^### adjusted p < 0. 001 vs. CHOW in proteomics data.

Figure 5

Discovering cell-specific changes in protein and gene levels of DIO-NASH mice. (A) UMAP projection of 6888 cells from CHOW and DIO-NASH livers. Each point represents a single cell. Cells that show similar transcriptomic profile are grouped by colour based on unsupervised clustering. 23 different cell populations were identified. Cell type identity was assigned by matching canonical markers with gene expression profile of each cluster. (B) UMAP projection showing how each experimental group contribute to each cell cluster. (C) UMAP projections showing sub-populations of endothelial cells, macrophages, T cells and dendritic cells. (D) Differentially expressed genes and proteins which was either up- or downregulated in DIO-NASH has been assigned to a specific cell-type as identified by scRNAseq by cross-referencing with marker genes (enriched genes).

Figure 6

Uncovering regulated marker genes for key cell types involved in NASH pathogenesis. Cell specific distribution of gene expression of histological markers, Col1a1, Acta2 and Lgals3, and significantly regulated marker genes for hepatocytes, macrophages and stellate cells. Dot size represents the percentage of cells in a cluster expressing a given gene (detection rate, % cell).