Genome-wide transcriptome analysis identifies novel gene signatures implicated in human chronic liver disease

Abstract

The molecular mechanisms behind human liver disease progression to cirrhosis remain elusive. Nuclear receptor small heterodimer partner (SHP/Nr0b2) is a hepatic tumor suppressor and a critical regulator of liver function. SHP expression is diminished in human cirrhotic livers, suggesting a regulatory role in human liver diseases. The goal of this study was to identify novel SHP-regulated genes that are involved in the development and progression of chronic liver disease. To achieve this, we conducted the first comprehensive RNA sequencing (RNA-seq) analysis of Shp(-/-) mice, compared the results with human hepatitis C cirrhosis RNA-seq and nonalcoholic steatohepatitis (NASH) microarray datasets, and verified novel results in human liver biospecimens. This approach revealed new gene signatures associated with chronic liver disease and regulated by SHP. Several genes were selected for validation of physiological relevance based on their marked upregulation, novelty with regard to liver function, and involvement in gene pathways related to liver disease. These genes include peptidoglycan recognition protein 2, dual specific phosphatase-4, tetraspanin 4, thrombospondin 1, and SPARC-related modular calcium binding protein-2, which were validated by qPCR analysis of 126 human liver specimens, including steatosis, fibrosis, and NASH, alcohol and hepatitis C cirrhosis, and in mouse models of liver inflammation and injury. This RNA-seq analysis identifies new genes that are regulated by the nuclear receptor SHP and implicated in the molecular pathogenesis of human chronic liver diseases. The results provide valuable transcriptome information for characterizing mechanisms of these diseases.

Keywords: chronic hepatitis C virus; gene expression; human chronic liver diseases; knockout mice; ribonucleic acid sequencing; small heterodimer partner.

Figures

Fig. 1.

Identification of novel candidate small heterodimer partner (SHP)-regulated genes by RNA sequencing (RNA-seq) analysis of wild-type (WT) and Shp−/− liver. A: Integrated Genome Browser visualization tracks from RNA-seq reads depict complete loss of SHP [nuclear receptor subfamily 0, group B, member 2 (Nr0b2)] exon 1 expression in Shp−/− mice. B: known SHP target early growth response 1 (Egr-1) is increased 9.1-fold in Shp−/− mice compared with WT. C: potential SHP target monocyte-to-macrophage differentiation-associated 2 (Mmd2) shows a dramatic 8.0-fold increase above WT Mmd2 levels. D: cholesterol 7 α-hydroxylase (Cyp7a1) shows 4.2-fold increased expression in Shp−/− mice compared with WT mice. E: expression of Serpina12 is decreased 3.8-fold in Shp−/− mice compared with WT across exons 3, 4, and 5. F: cytochrome P-450, family 2, subfamily a, polypeptide 5 (Cyp2a5) is decreased 2.6-fold in Shp−/− compared with WT mice.

Fig. 2.

Unique and common differentially expressed genes (DEGs) in SHP−/− mice, nonalcoholic steatohepatitis (NASH), and cirrhotic livers. A: Venn diagram shows total numbers and overlapping DEGs between SHP−/− mice and hepatitis C cirrhosis (HCV cirrhosis) (left). The histograms show the most highly upregulated (middle) and downregulated (right) genes common in Shp−/− and HCV cirrhosis (P = 0.0011, hypergeometric distribution = 0.0002). B: Venn diagram shows total numbers and common DEGs in both Shp−/− mice and NASH (left) (P = 1.919E-006, hypergeometric distribution = 4.288E-007). The histograms show differentially upregulated (middle) and commonly downregulated (right) genes in Shp−/− and NASH. C: left, Venn diagram showing total numbers of genes in the three groups and amount commonly downregulated in all of Shp−/−, HCV cirrhosis, and NASH. Right, histogram identifying those DEGs common across all three datasets, both upregulated and downregulated in HCV cirrhosis, Shp−/−, and NASH.

Fig. 3.

Hierarchical clustering of DEGs in Shp−/− mice, chronic hepatitis C cirrhosis, and NASH. A: unsupervised hierarchical clustering of genes common to the NASH microarray and chronic hepatitis C cirrhosis and Shp−/− RNA-seq gene expression arrays. B: heat map showing supervised hierarchical clustering of genes with fold changes in expression greater than or equal to ±5-fold in Shp−/− compared with WT mice depicting potential SHP-activated genes in group 1 and SHP-repressed genes in group 2. C: heat map showing supervised hierarchical clustering of selected genes with fold changes greater than or equal to ±1.5 in both cirrhotic compared with normal livers and in Shp−/− compared with WT livers, demonstrating correlation between the disease state and Shp−/− gene expression, those repressed in chronic hepatitis C cirrhosis in group 3 and activated in group 4. D: similar to chronic hepatitis C cirrhosis, this heat map shows selected genes with fold changes greater than or equal to ±1.5-fold in NASH compared with normal livers common in Shp−/− compared with WT livers and has a similar pattern.

Fig. 4.

qPCR validation of new genes that are differentially expressed in Shp−/− mice, human liver steatosis, fibrosis, NASH, and alcohol and hepatitis C cirrhosis. A: from left to right, peptidoglycan recognition protein 2 (PGLYRP2), dual specific phosphatase-4 (DUSP4), and tetraspanin 4 (TSPAN4). B: from left to right, thrombospondin 1 (THBS1), SPARC-related modular calcium binding protein-2 (SMOC2), and SHP. Expression levels of each gene for each sample were normalized to the hypoxanthine phosphoribosyltransferase 1 expression level of that sample as an internal control. ¥P <0.001, ‡P <0.01, and *P <0.05. The no. of specimens in each group is as follows: HCV cirrhosis (n = 39), alcohol cirrhosis (n = 28), NASH cirrhosis (n = 13), fibrosis (n = 7), steatosis (n = 15), and control (n = 24). C: immunohistochemistry analysis of DUSP4 protein in human alcohol and hepatitis C cirrhosis compared with the normal liver. Two representative results from each group are shown. N, normal; AC, alcohol cirrhosis; HC, HCV cirrhosis. Top: DUSP4 protein expression. Bottom: DUSP4 overlay with DAPI staining. Magnification ×20.

Fig. 5.

Functional analysis of selected genes in mouse liver inflammatory and injury models. A: qPCR of gene expression in hepatocytes isolated from 2-mo-old male mice that were cultured for 6 and 24 h. *P <0.01, 24 h vs. 6 h. B: qPCR of gene expression from WT and Shp−/− hepatocytes treated with lipopolysaccharide (LPS) for 6 and 24 h. *P <0.01, LPS 10 μg vs. LPS 1 μg or control in WT cells at 6 h; ‡P <0.01, LPS 10 μg vs. control in Shp−/− cells at 6 h; §P <0.01, Shp−/− vs. WT cells at 6 h; ¶P <0.01, Shp−/− vs. WT cells at 24 h; δP <0.01, LPS vs. control at 24 h. C: qPCR of gene expression in mice fed with CCl4 or 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) to induce liver fibrosis. *P < 0.01, CCl4 or DDC vs. respective control.

Fig. 6.

Alteration in expression of selected genes in stellate cells from alcohol-induced liver injury. A and B: qPCR of gene expression in Huh7 cells with SHP overexpression (A) or knockdown (B). C: qPCR of gene expression in stellate cells [hepatic stellate cells (HSC)] isolated from alcohol-treated rats. *Significance was determined at P < 0.01.

Fig. 7.

Gene Ontology (GO) pathways enriched in Shp−/− mice, HCV cirrhosis, and NASH. DEGs among Shp−/− mice, HCV cirrhosis, and NASH datasets were fed into the Database for Annotation, Visualization, and Integrated Discovery (DAVID) version 6.7 functional annotation software and used to generate GO pathways enriched in our gene sets. Pathways enriched in genes upregulated (A) and downregulated (B) in Shp−/− mice and altered in HCV cirrhosis and NASH are shown. Pie chart slices represent the numbers of DEGs listed in each pathway. Significance was determined at P < 0.05. C: GO pathway analysis of DEGs in both human hepatitis C cirrhosis and Shp−/− mouse showing several highly enriched pathways and the percentage enrichment of these pathways (significance of P < 0.05). Two notable genes in each pathway are shown in parentheses.