Tissue-specific pathways and networks underlying sexual dimorphism in non-alcoholic fatty liver disease

Abstract

Background: Non-alcoholic fatty liver disease (NAFLD) encompasses benign steatosis and more severe conditions such as non-alcoholic steatohepatitis (NASH), cirrhosis, and liver cancer. This chronic liver disease has a poorly understood etiology and demonstrates sexual dimorphisms. We aim to examine the molecular mechanisms underlying sexual dimorphisms in NAFLD pathogenesis through a comprehensive multi-omics study. We integrated genomics (DNA variations), transcriptomics of liver and adipose tissue, and phenotypic data of NAFLD derived from female mice of ~ 100 strains included in the hybrid mouse diversity panel (HMDP) and compared the NAFLD molecular pathways and gene networks between sexes.

Results: We identified both shared and sex-specific biological processes for NAFLD. Adaptive immunity, branched chain amino acid metabolism, oxidative phosphorylation, and cell cycle/apoptosis were shared between sexes. Among the sex-specific pathways were vitamins and cofactors metabolism and ion channel transport for females, and phospholipid, lysophospholipid, and phosphatidylinositol metabolism and insulin signaling for males. Additionally, numerous lipid and insulin-related pathways and inflammatory processes in the adipose and liver tissue appeared to show more prominent association with NAFLD in male HMDP. Using data-driven network modeling, we identified plausible sex-specific and tissue-specific regulatory genes as well as those that are shared between sexes. These key regulators orchestrate the NAFLD pathways in a sex- and tissue-specific manner. Gonadectomy experiments support that sex hormones may partially underlie the sexually dimorphic genes and pathways involved in NAFLD.

Conclusions: Our multi-omics integrative study reveals sex- and tissue-specific genes, processes, and networks underlying sexual dimorphism in NAFLD and may facilitate sex-specific precision medicine.

Keywords: Bayesian networks; Coexpression networks; Hybrid mouse diversity panel; Key regulator genes; Multi-omics integration; Non-alcoholic fatty liver disease (NAFLD); Sexual dimorphism.

Fig. 1

Schematic representation of the methodology. Genotype, liver and adipose tissue gene expression data, and hepatic triglyceride phenotypic data from both sexes of the hybrid mouse diversity panel (HMDP) mice were first integrated using Marker Set Enrichment Analysis in the Mergeomics pipeline to predict sex- and tissue-specific pathways perturbed in NAFLD. Then, potential regulatory genes (key drivers) for male-specific, female-specific, and shared pathways were identified using the Key Driver Analysis in Mergeomics. TG triglyceride, eQTL expression quantitative trait loci, GWAS genome-wide association studies

Fig. 2

Comparison between NAFLD processes perturbed in the liver and adipose tissue for females. Putative causal pathways that are common to both tissues and unique to each tissue are listed. Co-expression modules are annotated with the most over-represented gene ontology terms. “NA” indicates no over-represented terms were found for a given module. BCAA branched-chain amino acid, BCR B cell receptor, ECM extracellular matrix, TCR T cell receptor

Fig. 3

Comparison between NAFLD processes perturbed in the liver and adipose tissue of both sexes. Putative causal pathways that are a shared between sexes in one or both tissues and b unique to each sex and each tissue are listed. TCA the citric acid

Fig. 4

Bayesian gene network representations of NAFLD pathways and their key driver genes. a Liver Bayesian subnetwork comprised of shared liver NAFLD supersets between sexes and their top key drivers. b Adipose tissue Bayesian subnetwork comprised of shared adipose NAFLD supersets between sexes and their top key drivers. Female-specific c liver Bayesian subnetworks and d adipose tissue Bayesian subnetworks and their corresponding top key drivers. Male-specific e liver Bayesian subnetworks and f adipose tissue Bayesian subnetworks and their corresponding top key drivers. Key driver genes are shown with larger node sizes, human GWAS candidate genes are represented in hexagon shapes, and the rest of the genes are represented by smaller node sizes. Each NAFLD-associated superset is indicated with a distinct color in each network. Network genes that are not members of the NAFLD supersets are represented in gray. See also Additional file 4