Fig 1.
Epigenome-wide methylation profiles.
Manhattan plots of differentially methylated regions (DMRs) across chromosomes shows that Pcyt2 + /- exhibits (A) 6281 hypermethylated and (B) 3297 hypomethylated DMRs. (C) Volcano plots show the magnitude of change in DMRs from in Pcyt2 + /- where blue and red dots represent hypo- and hypermethylation, respectively. DMRs above the black line are significant (p = 0.05). Data points at x=+/-7 represent genes where fold change = infinity. (D) The distribution of significant hyper- and hypomethylated DMRs in relation to the nearest gene region show that most DMRs are in the genebody, and the promoter region that is most effected is the 3K region.
Fig 2.
Genomic distribution and functional enrichment of differentially methylated genes.
DMRs present in gene-coding regions, termed differentially methylated genes (DMGs), were selected for further analysis. (A) Pcyt2 + /- contains 5226 wherein for hypermethylated DMGs, 7.5% are in promoter regions and 92.5% are in the gene body, and in hypomethylated DMGs, 14% are in promoter regions and 86% in the gene body. (B) The chromosomal distribution of hyper- and hypomethylated DMGs shows that 1, 2 and 5 harbour the greatest number of DMGs. Bars above the black line show chromosomes possessing DMGs that account for over 6% of the total DMGs. Top 5 (C) and top 15 (D) functionally enriched pathways of hyper- and hypomethylated DMGs in Pcyt2 + /- using BioPlanet, Gene Ontology (GO) Biological process and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases. Functional enrichment shows an enrichment in pathways linked well-known NASH development processes such as to hepatic inflammation and fibrosis, and interestingly, a great enrichment in nervous system related pathways suggesting the importance of phospholipid homeostasis in CNS/PNS function. Analysis was completed using a Benjamini-Hochberg adjusted p value (<0.05) as the significance cut off. Combined score was represented using a z-score permutation background correction on Fisher’s exact test p-value to generate an overall score for each pathway; this method of pathway analysis has been shown to recover the most correct terms [31].
Fig 3.
DNA methylation associates with altered mRNA expression of key genes involved in NASH pathogenesis.
(A) Venn diagram showing the overlapping genes from Pcyt2 + /- Me-DNA-seq and 2-month Pcyt2 + /- liver microarray (GEO# GSE55617), showing the set of genes that are both differentially methylated and expressed. (B) Heatmaps show the top 20 hypermethylated and hypomethylated genes based on mRNA change in 2-month-old Pcyt2 + /- mice. Genes that exhibited that greatest changes in methylation and mRNA expression including genes related to mitochondrial function, autophagy, lipid metabolism, glucose homeostasis, MAPK signalling, inflammation, n-glycan biosynthesis, and cell cycle. (C) RT-PCR analysis of insulin, MAPK and JAK/STAT signalling pathways in 8-month-old Pcyt2 + /- mice shows altered DNA methylation influences the expression of genes within key pathways relevant to NASH pathogenesis. Genes that have a methylation increased or decreased of infinity are represented as Log2FC = 5.
Fig 4.
Network analysis of genes that are both differentially methylated and expressed (DMEG).
(A) Protein-protein interaction network of DMEGs with one large subnetwork consisting of 34 interrelated seed genes, 540 nodes and 686 edges. Important genes that show the greatest degree of connectively and thus, the greatest influence on the network include proteins related to mitochondrial function, glucose homeostasis, and regulation of cell cycle and proliferation. Red and blue nodes indicate genes with increased and decreased mRNA expression, respectively, and circle size indicates the grade of connectivity. (B) Functional enrichment analysis of protein-protein interaction network using KEGG database showing and enrichment in processes highly relevant to including Pi3k-Akt signalling pathway, Foxo signalling pathway, insulin signalling, oxidative phosphorylation, multiple inflammation related pathways and NAFLD and previously unexplored pathways such as regulation of actin cytoskeleton, focal adhesion, cell cycle regulation and cellular senescence.
Fig 5.
Epigenome-wide methylation profiles.
Manhattan plots of differentially methylated regions (DMRs) across chromosomes shows that PEA treatment equally reduces both hyper-and hypomethylated DMRs from (A) 6281 hypermethylated DMRs in Pcyt2 + /- to 210 in Pcyt2 + /- + PEA and (B) 3297 hypomethylated in Pcyt2 + /- to 136 hypomethylated in Pcyt2 + /- + PEA. (C) Volcano plots show the magnitude of change in Pcyt2 + /- + PEA DMRs relative to Pcyt2 + / + . Blue and red dots represent hypo- and hypermethylation, respectively. DMRs above the black line are significant (p = 0.05). Data points at x=+/-7 represent genes where fold change = infinity. Both the Manhattan and volcano plots show dramatically reduced DNA methylation in PEA treated liver. (D) The distribution of significant hyper- and hypomethylated DMRs in relation to the nearest gene region in Pcyt2 + /- + PEtn show that most DMRs are in the genebody, and the promoter region that is most effected is the 3K region.
Fig 6.
Genomic distribution and functional enrichment of differentially methylated genes.
(A) PEA treatment reduced Pcyt2 + /- DMGs by 96.7% from 5226 to 175 DMGs in Pcyt2 + /- + PEA with 43% and 57% hyper- and hypomethylated DMGs, respectively, this shows a similar distribution to Pcyt2 + /- showing a coordinated action of methylation and demethylation processes. (B) DMG distribution across chromosomes showing the highest relative proportions of hypermethylated genes PEA at Chromosomes 1, 2, 3, 8, 11 and 13 (6.4-9.2%) and the highest proportions of hypomethylated genes at Chromosomes 2, 5, 7, 8, 11, and 15 (7.1-14%). (C) Enrichment analysis of DMG in Pcyt2 + /- + PEA mice showing nearly all enriched pathways were abolished by PEA treatment (D) PEA treatment reduced genes that are differently methylated and expressed (DMEGs) by 95.2% from 162 to 8. Correlation plot of the remaining 8 DMEGs after PEA treatment. Only one gene, Aida, was differentially methylated in promoter region. (E) PCR validation of key genes that were differentially methylated and differentially expressed (Man2a1, Foxo1 and Ndufv2) showing that the expressions of all genes were attenuated by PEA treatment. Band intensities were measured using ImageJ. * p < 0.05, ***p < 0.001.
Fig 7.
The effect of PEA supplementation on DMGs that were not reversed.
(A) Principal component analysis of the methylation changes of the DMG’s that were not revered by PEA treatment showing 2 central clusters of genes. Genes were assigned grouping variables according to known associations with cellular pathways. Genes involved in processes such as glycerophospholipid flippase activity, DNA binding and regulation of extra cellular matrix activity were regulated similarly. In contrast, genes involved in RNA binding and cell cycle showed a greater difference in epigenetic regulation. (B) 3D plot showing the 114 common DMGs and specific DMGs that exhibit distinct separation and therefor different regulation from 2 central clusters. Data presented as log2 fold change. Colour indicates variation along the y-axis (Pcyt2 + /-). (C) Line plot visualizing the directional change across Pcyt2 + / + , Pcyt2 + /- and Pcyt2 + /- + PEA, with Key genes that exhibit the greatest exacerbation or attenuation of methylation status by PEA treatment are labelled. Most DMGs in this group were unaffected by PEA showing that methylation changes at these loci in Pcyt2 + /- are strong and not reversible. The subset of genes that showed an exaggeration of methylation status should be targets of future studies to delineate the therapeutic potential of PEA.