While several molecular targets are under consideration, mechanistic underpinnings of the transition from uncomplicated nonalcoholic fatty liver disease (NAFLD) to nonalcoholic steatohepatitis (NASH) remain unresolved. Here we apply multiscale chemical profiling technologies to mouse models of deranged hepatic ketogenesis to uncover potential NAFLD driver signatures. Use of stable-isotope tracers, quantitatively tracked by nuclear magnetic resonance (NMR) spectroscopy, supported previous observations that livers of wild-type mice maintained long term on a high-fat diet (HFD) exhibit a marked increase in hepatic energy charge. Fed-state ketogenesis rates increased nearly 3-fold in livers of HFD-fed mice, a greater proportionate increase than that observed for tricarboxylic acid (TCA) cycle flux, but both of these contributors to overall hepatic energy homeostasis fueled markedly increased hepatic glucose production (HGP). Thus, to selectively determine the role of the ketogenic conduit on HGP and oxidative hepatic fluxes, we studied a ketogenesis-insufficient mouse model generated by knockdown of the mitochondrial isoform of 3-hydroxymethylglutaryl-CoA synthase (HMGCS2). In response to ketogenic insufficiency, TCA cycle flux in the fed state doubled and HGP increased more than 60%, sourced by a 3-fold increase in glycogenolysis. Finally, high-resolution untargeted metabolomics and shotgun lipidomics performed using ketogenesis-insufficient livers in the fed state revealed accumulation of bis(monoacylglycero)phosphates, which also accumulated in livers of other models commonly used to study NAFLD. In summary, natural and interventional variations in ketogenesis in the fed state strongly influence hepatic energy homeostasis, glucose metabolism, and the lipidome. Importantly, HGP remains tightly linked to overall hepatic energy charge, which includes both terminal fat oxidation through the TCA cycle and partial oxidation via ketogenesis.
Keywords: Fatty acid oxidation; Glucose metabolism; Hepatology; Metabolism; Mouse models.