Ketogenesis mitigates metabolic dysfunction-associated steatotic liver disease through mechanisms that extend beyond fat oxidation

Abstract

The progression of metabolic dysfunction-associated steatotic liver disease (MASLD) to metabolic dysfunction-associated steatohepatitis (MASH) involves alterations in both liver-autonomous and systemic metabolism that influence the liver's balance of fat accretion and disposal. Here, we quantify the contributions of hepatic oxidative pathways to liver injury in MASLD-MASH. Using NMR spectroscopy, UHPLC-MS, and GC-MS, we performed stable isotope tracing and formal flux modeling to quantify hepatic oxidative fluxes in humans across the spectrum of MASLD-MASH, and in mouse models of impaired ketogenesis. In humans with MASH, liver injury correlated positively with ketogenesis and total fat oxidation, but not with turnover of the tricarboxylic acid cycle. Loss-of-function mouse models demonstrated that disruption of mitochondrial HMG-CoA synthase (HMGCS2), the rate-limiting step of ketogenesis, impairs overall hepatic fat oxidation and induces an MASLD-MASH-like phenotype. Disruption of mitochondrial β-hydroxybutyrate dehydrogenase (BDH1), the terminal step of ketogenesis, also impaired fat oxidation, but surprisingly did not exacerbate steatotic liver injury. Taken together, these findings suggest that quantifiable variations in overall hepatic fat oxidation may not be a primary determinant of MASLD-to-MASH progression, but rather that maintenance of ketogenesis could serve a protective role through additional mechanisms that extend beyond overall rates of fat oxidation.

Keywords: Fatty acid oxidation; Hepatology; Intermediary metabolism; Metabolism; Obesity.

Figure 1. Metabolic characteristics of MASLD-MASH.

(A) The initial stages of metabolic dysfunction–associated steatotic liver disease (MASLD) begin with hepatic steatosis, linked to accelerations in: de novo lipogenesis (DNL), turnover of the tricarboxylic acid (TCA) cycle, and phosphoenolpyruvate-derived (PEP-derived) gluconeogenesis (GNG). Metabolic shifts during the progression of MASLD to metabolic dysfunction–associated steatohepatitis (MASH) remain poorly understood. HCC, hepatocellular carcinoma. (B) Study design. Patients with BMI ≥ 35 and liver proton density fat fraction (PDFF) greater than 5% underwent a liver biopsy, FSIVGTT, and DXA imaging. After an overnight fast, 8 hepatic intermediary metabolic fluxes were quantified using ²H/¹³C stable isotope tracing. (C) Distributions of the NAFLD activity scores (NASs) in all 16 participants (15 female, 1 male). (D) Correlation matrix of NAS with histological scores for steatosis, ballooning, inflammation, fibrosis, PDFF, liver enzymes, and bilirubin. Pearson’s correlation coefficients (r) are shown in heatmap format with the magnitude of the correlation given by the right-hand legend and displayed in each square. Correlations were accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001. ALP, alkaline phosphatase; AST, aspartate aminotransferase; ALT, alanine transaminase.

Figure 2. Liver injury does not correlate with endogenous glucose production (EGP) or TCA cycle turnover.

Fluxes through hepatic intermediary metabolic pathways were quantified in humans after oral administration of heavy water (²H₂O) and [U-¹³C₃]propionate. [3,4-¹³C₂]Glucose and d-[U-¹³C₄]βOHB were intravenously infused, allowing whole-body glucose (V_EGP) and βOHB turnover (V_RaβOHB) to be quantified at metabolic and isotopic steady state. (A) Fractional sourcing of glucose can be quantified from the ²H enrichment pattern of plasma glucose using ²H NMR, which, by multiplying by V_EGP, allows absolute reaction velocities (V) (i.e., flux) for hepatic glucose sourcing pathways to be quantified. (B) Administration of [U-¹³C₃]propionate ¹³C-enriches TCA cycle intermediates, which sources phosphoenolpyruvate (PEP). Using ¹³C NMR, and the multiplet arising from the C2 resonance of plasma glucose, the resulting metabolic network models oxidative and anaplerotic nodes of the TCA cycle in parallel to glucose production. By normalizing of fluxes to V_PEP, the absolute reaction velocities of the TCA cycle, anaplerosis/cataplerosis, and pyruvate cycling can be quantified. (C) Average percentage of V_EGP derived from glycogen, glycerol, and PEP, highlighting that TCA cycle–sourced PEP is the major contributor to V_EGP in the fasted state in humans. (D) Average reducing equivalents (REs) derived from GNG, β-oxidation, and the TCA cycle. (E and F) The correlations of NAS with V_EGP (E) and TCA cycle turnover (V_CS) (F). Data are either expressed as mean ± SD or shown as correlations. Pearson’s correlation coefficients (r) are shown on each group along with a line of best fit and 95% confidence intervals calculated using linear regression. Correlations were accepted as significant if P < 0.05. P values are shown on each graph.

Figure 3. Liver injury correlates with endogenous ketogenesis and hepatic fat oxidation.

(A–C) The correlation of NAS with endogenous ketogenesis (V_RaβOHB) (A), total fat oxidation (B), and total RE turnover (C). (D–F) Correlation of serum non-esterified fatty acid (NEFA) concentrations with endogenous ketogenesis (V_RaβOHB) (D), total plasma ketone bodies (AcAc + βOHB) (E), and NAS (F). Pearson’s correlation coefficients (r) are given on each graph along with a line of best fit calculated using linear regression. Correlations were accepted as significant if P < 0.05. P values are shown on each graph.

Figure 4. Loss of hepatocyte HMGCS2 impairs fasting ketosis.

(A) Ketogenesis-null mice were generated by deletion of the gene encoding 3-hydroxymethylglutaryl-CoA synthase 2 (HMGCS2) in hepatocytes. Littermate control (WT) and HMGCS2 hepatocyte-specific knockout (HMGCS2-Liver-KO) male and female mice were maintained on standard chow diet, switched to a high-fat carbohydrate-restricted (HFCR) diet for 1 week, and studied in the random-fed or fasted state. (B and C) Functional loss of HMGCS2 was confirmed in vivo in chow-fed mice by demonstration that HMGCS2-Liver-KO mice failed to increase fasting total ketone bodies (TKBs) (B), marked by a decrease in both AcAc and βOHB (4-hour-fasted) (n = 4–8 per group) (C). (D) Ratio of βOHB to AcAc in 4-hour-fasted WT and HMGCS2-Liver-KO mice fed chow diet (n = 4–8 per group). (E) Serum TKB analysis of WT and HMGCS2-Liver-KO mice fed HFCR diet (n = 4–6 per group). (F) The ratio of βOHB to AcAc in WT and HMGCS2-Liver-KO mice fed HFCR diet (n = 4–6 per group). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5. Ketogenic insufficiency induces hepatomegaly and steatosis.

Littermate control (WT) and HMGCS2-Liver-KO male and female mice maintained on standard chow diet were switched to HFCR diet for 1 week. (A) Net change in body weight after switching of diets (n = 7 per group). (B and C) Total liver weight (grams) (B) and relative liver weight (percentage of body weight) (C) (n = 3–8 per group). (D and E) Total liver triacylglycerols (TAGs) in random-fed livers quantified colorimetrically (D) and shown histologically with H&E stain (E). Scale bars: 25 μm. (F) Total liver TAGs in HFCR diet–fed mice after 1 week of refeeding of standard chow (n = 4 per group). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests and accepted as significant if P < 0.05. **P < 0.01, ****P < 0.0001.

Figure 6. Ketogenic insufficiency induces relative hypoglycemia in HFCR diet–fed mice.

Systemic physiological markers of whole-body metabolism in random-fed (A–D) and 4-hour-fasted (E–H) male and female littermate control (WT) and HMGCS2-Liver-KO mice, including blood non-esterified fatty acids (NEFAs), TAGs, and glucose. Also included is total liver glycogen content (n = 4–11 per group). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01.

Figure 7. Ketogenic insufficiency diminishes gluconeogenesis in mice.

(A) Overview of in vivo hepatic flux modeling study design. Top: Fluxes were measured in female littermate control (WT) and HMGCS2-Liver-KO mice. Five days after surgical implantation of indwelling catheters, mice were switched from standard chow to an HFCR diet, and then fluxes were measured after a 2-day acclimation period. Middle: Stable isotope tracers were administered, including [6,6-²H₂]glucose, [U-¹³C₃]propionate, and heavy water (²H₂O). Red blood cells from donor mice were infused to maintain hematocrit. Samples were collected for analysis during the final 30 minutes after metabolic and isotopic steady state had been reached. Bottom: Diagram of reaction velocities (V) (i.e., fluxes) modeled using a mass isotopomer distribution analysis of derivatized plasma glucose acquired via GC-MS. (B) Rates of whole-body endogenous glucose production (V_EGP) as measured via tracer dilution of [6,6-²H₂]glucose (n = 4–6 per group). (C) Absolute rates of glucose sourcing pathways, including glycogenolysis (V_PYGL), total gluconeogenesis (V_Aldo), and rates of glycerol (V_GK) and phosphoenolpyruvate (PEP) (V_Enol) flux to glucose (n = 4–6 per group). (D) Absolute rates for reactions in the oxidative flux network shown in A, including lactate dehydrogenase (LDH; V_LDH), PEP carboxykinase (PCK; V_PCK, i.e., total TCA cycle cataplerosis), anaplerosis into the TCA cycle via pyruvate carboxylase (PC; V_PC) or via propionyl-CoA carboxylase (PCC; V_PCC), pyruvate cycling as the sum of pyruvate kinase (PK) and malic enzyme (ME) (V_PK+ME), citrate synthase (CS) flux (V_CS; i.e., TCA cycle turnover), and succinate dehydrogenase (SDH) flux (V_SDH) (n = 4–6 per group). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 8. Ketogenic insufficiency impairs fat oxidation in perfused livers.

(A) Prior studies have demonstrated that ketogenesis insufficiency induced by loss of HMGCS2 causes accumulation of mitochondrial acetyl-CoA and acceleration of the TCA cycle. Livers of male mice treated with scrambled control antisense oligonucleotide (ASO) or mouse Hmgcs2 ASO were perfused with octanoate (C8) and oxidative fluxes quantified using ²H/¹³C stable isotope tracing. (B and C) Total fat oxidation quantified as the summation of (2 × ketogenesis) + TCA cycle turnover (B), and reducing equivalent (RE) turnover (NADH + FADH₂) broken down into REs from gluconeogenesis (GNG), β-oxidation, and the TCA cycle (C), in perfused livers from control and Hmgcs2 ASO mice on chow diet (n = 10–11 per group). Fat oxidation was also studied in livers perfused ex vivo with C8 from control and Hmgcs2 ASO–treated mice maintained on a 42% high-fat high-sucrose Western diet (WD) for 8 weeks. (D–F) TCA cycle turnover (D), total fat oxidation (E), and RE production rate (NADH + FADH₂) (F) in perfused livers from control and Hmgcs2 ASO–treated mice on WD (n = 5–6 per group). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests or 2-way ANOVA and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 9. Ketogenic insufficiency impairs fat oxidation in isolated mitochondria.

(A) Mitochondrial diagnostics workflow. Liver mitochondria isolated by differential centrifugation and fueled by palmitoyl-l-carnitine plus α-ketoglutarate were used to assess the effects of hymeglusin (HG; HMGCS inhibitor) versus vehicle control (VCTRL) on βOHB production, respiratory kinetics, and redox potential during a creatine kinase (CK) energetic clamp, and on metabolite abundance after CK clamp. (B and C) Relative abundance of HMG-CoA, AcAc, and βOHB in mitochondria (B) and βOHB production measured in supernatant collected after CK clamp (C) (n = 4 per group). (D) Respiration (JO₂) plotted as a function of energy demand (ΔG_ATP [kcal/mol]), with respiratory sensitivity (i.e., JO₂ conductance) measured as the slope of the curve (n = 4 per group). Results of Student’s t test of slopes between VCTRL and HG are shown on the graph. (E and F) Relative abundance of acyl-CoA species (E) and free coenzyme A (F) in mitochondria (n = 4 per group). (G) Redox potential [NAD(P)H percentage reduction] plotted as a function of energy demand (ΔG_ATP [kcal/mol], n = 4 per group). Pool sizes are from mitochondria after incubation at a fixed energy demand (ΔG_ATP = –13.94 kcal/mol). Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests or 2-way ANOVA and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ^#significant by 2-way ANOVA (P < 0.05).

Figure 10. Loss of hepatocyte BDH1 impairs total hepatic fat oxidation but does not exacerbate HFD-induced liver fibrosis.

(A) Male hepatocyte-specific BDH1-null (BDH1-Liver-KO) and littermate control (WT) mice were maintained on standard chow diet, then were switched to a 42% high-fat high-sucrose WD for 16 weeks, after which oxidative fluxes were studied in livers perfused ex vivo with long-chain fatty acids. (B and C) Ketone body production (B) and AcAc and βOHB production (C) in livers of WT and BDH1-Liver-KO mice (n = 10–11 per group). (D) Absolute rates of oxidative pathways, including pyruvate cycling (V_PK+ME), total anaplerosis and cataplerosis (V_PEPCK), and TCA cycle turnover (V_CS) (n = 10–11 per group). (E and F) Total fat oxidation calculated at (2 × ketogenesis) + acetogenesis + TCA cycle turnover (E), and reducing equivalent (RE) turnover (NADH + FADH₂) broken down into REs generated from gluconeogenesis (GNG), β-oxidation, and the TCA cycle (F) (n = 10–11 per group). (G–I) Liver H&E and Picrosirius red histological staining (G), gene expression for fibrotic gene markers (H), and absolute levels of lipid peroxide 4-hydroxyalkenal species (I) in livers of WT and BDH1-Liver-KO mice maintained on 42% kcal fat WD (n = 10–12 per group). Scale bars: 25 μm in H&E images, 100 μm in Picrosirius red images. HHE, 4-hydroxy-2-hexenal; HNDE, 4-hydroxynona-2E, 6Z-dienal; HNE, 4-hydroxy-2-nonenal; HDTE, 4-hydroxy-2E, 6Z, 8Z-decatrienal. Data are expressed as mean ± SD. Statistical differences were determined by Student’s t tests and accepted as significant if P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001.