Hepatic Ketogenesis Regulates Lipid Homeostasis via ACSL1-mediated Fatty Acid Partitioning

Abstract

Background & Aims

Liver-derived ketone bodies play an essential role in energy homeostasis during fasting by supplying fuel to the brain and peripheral tissues. Ketogenesis also helps to remove excess acetyl-CoA generated from fatty acid oxidation, thereby protecting against diet-induced hepatic steatosis. Despite this, the role of ketogenesis in fasting-associated hepatocellular lipid metabolism has not been thoroughly investigated.

Methods

We used mice with liver-specific knockout of HMGCS2 mice to determine how ACSL1-mediated esterification contributes to fasting-induced steatosis and performed biochemical assays, gene expression profiling, Western blotting, and histologic analyses. We further investigated the association between HMGCS2 expression, lipid re-esterification, and steatosis using human primary hepatocytes and liver samples from patients with metabolic dysfunction-associated steatohepatitis.

Results

We show that ketogenic insufficiency, achieved through disrupting hepatic HMGCS2, worsens liver steatosis in both fasted chow-fed and high-fat-fed mice. Our findings indicate that hepatic steatosis arises from increased fatty acid partitioning to the endoplasmic reticulum (ER) for re-esterification, a process mediated by acyl-CoA synthetase long-chain family member 1 (ACSL1). Mechanistically, the accumulation of acetyl-CoA because of impaired hepatic ketogenesis drives the elevated translocation of ACSL1 to the ER. Furthermore, our study reveals heightened ER-localized ACSL1 and lipid re-esterification in human metabolic dysfunction-associated steatohepatitis cases exhibiting impaired hepatic ketogenesis. We also demonstrate that L-carnitine, which buffers excess acetyl-CoA, reduces ER-associated ACSL1 and alleviates hepatic steatosis.

Conclusions

Hepatic ketogenesis plays a crucial role in maintaining intracellular acetyl-CoA balance, regulating lipid partitioning, and preventing the development of fasting-induced hepatic steatosis.

Graphical abstract

Summary.

This study shows that impairment in ketogenesis worsens fasting-induced steatosis by promoting lipid re-esterification partly through endoplasmic reticulum–associated acyl-CoA synthetase long-chain family member 1, highlighting its central role in regulating acetyl-CoA homeostasis and lipid partitioning during fasting.

Hepatic steatosis hinges on the balance between free fatty acid uptake, de novo lipogenesis, fatty acid oxidation (FAO), and very-low-density lipoprotein secretion.^1,2 In this process, lipid partitioning plays a pivotal role in adapting to systemic energy needs.³ For instance, during fasting, free fatty acids entering the hepatocytes are directed toward mitochondrial FAO. Conversely, in nutrient-rich conditions, fatty acids are channeled to the endoplasmic reticulum (ER) for esterification and storage as lipid droplets.³ An imbalance in lipid partitioning results in hepatic steatosis and injury, marking the initial step in metabolic dysfunction-associated steatotic liver disease (MASLD) pathogenesis,^3,4 affecting ∼24% of the population in the United States, with a rapidly rising global incidence.5, 6, 7, 8 Despite the significance of fatty acid partitioning in hepatic lipid homeostasis, the mechanisms governing this process remain underexplored.

Fasting serves as an apt model to dissect lipid partitioning mechanisms, because most fatty acids are partitioned to FAO via a controlled manner.^9,10 Hepatic FAO generates acetyl-CoA, which is then converted into ketone bodies via 3-hydroxymethyllglutaryl-CoA synthase 2 (HMGCS2), a rate-limiting mitochondrial enzyme highly expressed in hepatocytes.¹¹ Fasting-induced elevation in ketone bodies provides fuel for peripheral organs, such as the brain and muscles.^12,13 Recent studies show that neonates with ketogenic insufficiency develop fatty liver, which can be mitigated by early weaning.^14,15 Similarly, feeding a high-fat diet (HFD) to mice with ketogenic insufficiency induces hepatic steatosis and liver injury.^13,15,16 These findings suggest that FAO and ketogenesis are coregulatory mechanisms that eliminate excess dietary lipids, thereby protecting against fatty liver. During fasting, hepatocytes are bombarded with a huge influx of adipose tissue–derived fatty acids.¹⁷ Although ketogenesis is known to be induced during fasting,¹⁷ its precise role in regulating hepatocellular lipid homeostasis remains unclear.

Our study uses conditional liver-specific HMGCS2 knockout mice to demonstrate that impairment in ketogenesis exacerbates fasting-induced hepatic steatosis in chow-fed and high-fat-fed mice. Notably, impairment in ketogenesis did not affect fatty acid uptake or de novo lipogenesis but FAO. Because elevation in free fatty acids from impaired FAO is toxic to cells, the enzyme acyl-CoA synthetase long-chain family member 1 (ACSL1) esterifies the fatty acids into a nontoxic acyl-CoAs.^18,19 Mitochondrial ACSL1 promotes β-oxidation, whereas ER-associated ACSL1 favors lipid storage through re-esterification.²⁰ Our study reveals that the accumulation of acetyl-CoA because of ketogenic insufficiency induces ER translocation of ACSL1, leading to fatty acid re-esterification and steatosis. We report a similar mechanism in human metabolic dysfunction-associated steatohepatitis (MASH) livers, where reduced HMGCS2 is associated with increased ER-associated ACSL1. Additionally, we observe a diminished buffering capacity for excess acetyl-CoA in MASH because of acquired carnitine deficiency. Restoring L-carnitine (LC) attenuates fasting-induced acetyl-CoA accumulation, ER-ACSL1, fatty acid re-esterification, and hepatic steatosis. Overall, our study defines the crucial role of hepatic ketogenesis in lipid homeostasis by regulating fatty acids partitioning via the acetyl-CoA-ER-ACSL1 axis.

Results

Hepatic Ketogenesis Protects Against Fasting-induced Hepatic Steatosis

Genetic studies dissecting the role of ketone bodies were performed in mice with constitutive knockout of hepatic HMGCS2, which are prone to severe hepatic steatosis and mitochondrial dysfunction at the neonatal age.¹⁴ To assess the role of hepatic ketogenesis in metabolic homeostasis without the confounding developmental defects, we generated a mouse model with conditional knockdown of HMGCS2 in the liver (Hmgcs2^ΔLiv) by crossing our Hmgcs2-floxed mice (Hmgcs2^F/F) with mice expressing tamoxifen-inducible Cre under the control of albumin promoter (Figure 1A). Administration of tamoxifen decreased HMGCS2 protein expression in the liver but not in other organs, such as the colon, kidney, and heart (Figure 1B and C), which are minimally involved in ketogenesis.¹² Note that littermate Hmgcs2^F/F mice that do not express Cre recombinase were administered with tamoxifen and used as control animals throughout the studies. Ketone bodies play a crucial role in fasting energetics¹²; therefore, we assessed Hmgcs2^F/F and Hmgcs2^ΔLiv mice under fed and fasted (0–24 hours) conditions. Fasting decreased body weight, fat mass, and lean mass with no difference between the genotypes (Figure 1D and E). As expected, Hmgcs2^F/F mice showed a time-dependent increase in serum ketone body, β-hydroxybutyrate, and hepatic HMGCS2 expression (Figure 1F and G). Fasting-induced ketonemia is completely abrogated in Hmgcs2^ΔLiv mice (Figure 1G), consistent with the recent report that the liver is the major source of ketone bodies.²¹

Figure 1.

Hepatic ketogenic insufficiency does not affect glucose levels. (A) Schematic representation showing generation of conditional liver-specific HMGCS2 knockout (Hmgcs2^ΔLiv) mice. (B) Western blot (WB) analysis showing HMGCS2 protein in the liver and colon of Hmgcs2^ΔLiv mice. (C) WB analysis showing HMGCS2 expression in kidney and heart of Hmgcs2^ΔLiv mice fed or 16-hour fasted. (D) Body weight in fed and 16-hour fasted Hmgcs2^ΔLiv mice. (E) Fat mass and lean mass determined by Nuclear Magnetic Resonance (NMR). (F) WB analysis showing HMGCS2 protein expression in the livers of fasted Hmgcs2^F/F mice. (G) Serum β-hydroxybutyrate (BOH) and (H) blood glucose levels in fasted Hmgcs2^ΔLiv mice. (I) Liver glycogen levels, (J) serum glucagon levels, (K) serum lactate levels. All the data are presented as mean ± standard error of the mean. ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001 analyzed by 2-tailed Student t test and 1-way analysis of variance (Tukey multiple-comparisons test).

Ketogenic insufficiency is shown to increase hepatic glucose production.²² Moreover, ketone body metabolism in peripheral tissues, particularly muscle, competes with glucose utilization, increasing circulating glucose levels.²³ However, conditional ketogenic insufficiency does not affect fasting glucose levels (Figure 1H), consistent with recent report.²⁴ Furthermore, hepatic glycogen, serum glucagon, and serum lactate levels were similar in both genotypes (Figure 1I-K), suggesting that hepatic ketogenesis insufficiency does not affect glucose homeostasis under acute nutrient deprivation.

Although constitutive deletion of hepatic HMGCS2 induces fatty liver in neonatal mice pups (as early as 4 days after birth),^14,15 knockdown of Hmgcs2 did not affect the hepatic lipid content in adult mice assessed under fed conditions.¹⁶ Consistently, temporal disruption of Hmgcs2 in adult mice did not affect hepatic triglyceride (TAG) content and liver-to-body weight ratio under fed state. However, fasting significantly elevated the hepatic triglyceride resulting in increased liver-to-body weight ratio (Figure 2A-C) and paler liver in Hmgcs2^ΔLiv mice (Figure 2D and E). Hematoxylin and eosin (H&E) analysis showed microsteatosis in the livers of fasted Hmgcs2^ΔLiv mice (Figure 1F). Serum triglycerides were similar between the fed and fasted groups (Figure 2G). Serum cholesterol levels trended higher in Hmgcs2^ΔLiv mice; however, fasting did not affect cholesterol levels in both genotypes (Figure 2H). Ketone bodies protect against inflammation,^25,26 a key regulator of hepatic lipid metabolism.^27,28 However, no difference was noted in the macrophage markers, such as F4/80 and CD45 (Figure 2I), and the expression of inflammatory genes, such as Tnfα, Ccl2, Il-6, and Il1-β (Figure 2J). Of note, histologic analysis showed no morphologic difference in the kidneys of Hmgcs2^ΔLiv mice (Figure 2K). Together, our data suggest that impairment in ketogenesis induces hepatic steatosis under nutrient-deprived conditions.

Figure 2.

Hepatic ketogenesis protects against fasting-induced hepatic steatosis. (A) Liver triglycerides (TAG) in fasted Hmgcs2^ΔLiv mice. (B) Liver TAGs normalized to protein levels in fasted Hmgcs2^ΔLiv mice. (C) Liver-to-body weight ratio in fasted Hmgcs2^ΔLiv mice. (D) Gross image of liver from fed and 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (E) Gross morphology of livers from fasted Hmgcs2^F/F mice. (F) H&E of liver from fed and 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (G) Serum TAGs and (H) serum cholesterol from Hmgcs2^F/F and Hmgcs2^ΔLiv mice fed or 16-hour fasted. (I) Immunohistochemistry analysis assessing inflammatory markers (F4/80 and CD45) in the livers (20X magnification) of fed and 16-hour fasted Hmgcs2^ΔLiv mice. (J) Relative mRNA levels inflammatory genes in fed or 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (K) H&E staining of the kidney (20X magnification) from fed or 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001 analyzed by 2-tailed Student t test and 1-way analysis of variance (Tukey multiple-comparisons test).

Ketogenic Insufficiency Decreases Fatty Acid Oxidation Despite Normal Mitochondrial Function

During fasting, nonesterified fatty acids released from the adipose tissue enter the liver to undergo FAO and enhanced adipose tissue lipolysis drives hepatic steatosis.²⁹ Adipose tissue levels of lipolysis-related proteins, such as pHSL and ATGL (Figure 3A), and the circulating nonesterified fatty acids levels (Figure 3B) were similar between the Hmgcs2^F/F and Hmgcs2^ΔLiv mice, suggesting no difference in adipose tissue lipolysis. We then mimicked fasting in isolated primary hepatocytes (purity confirmed by quantitative polymerase chain reaction [qPCR] analysis for hepatocyte markers, such as Alb and Hnf4a [Figure 3C]) by treating with bovine serum albumin (BSA)-conjugated palmitate (PA) or oleate (OA). We found higher lipid accumulation in primary hepatocytes from Hmgcs2^ΔLiv mice (Figure 3D and E), indicating that ketogenic insufficiency induces steatosis likely through a cell-autonomous mechanism.

Figure 3.

Ketogenic insufficiency induces steatosis likely through a cell-autonomous mechanism. (A) WB analysis showing lipolysis-related proteins in the white adipose tissue (WAT) of fed or 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (B) Serum nonesterified free fatty acids (NEFA) levels. (C) Relative mRNA levels of genes encoded for hepatocytes (Hep) and nonparenchymal cells (NPC) of the liver. (D) Oil-red-O staining of primary hepatocytes from Hmgcs2^ΔLiv mice and Hmgcs2^F/F mice incubated with 200 μM BSA-conjugated PA or BSA alone for 16 hours (20X magnification). (E) Oil-red-O staining of primary hepatocytes from Hmgcs2^ΔLiv mice and Hmgcs2^F/F mice incubated with 200 μM BSA-conjugated OA or BSA alone for 16 hours (60x magnification). Relative mRNA levels of genes associated with (F) fatty acid uptake and (G) lipid-droplet-associated proteins in the liver. WB showing (H) lipolysis-related proteins and (I) autophagy-related proteins in the liver. (J) Lipid secretion assay in Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (K) WB analysis showing HMGCS2 expression in hepatocytes from Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (L) De novo lipogenesis assessed in primary hepatocytes treated with insulin (10 nM) or saline for 16 hours. n = 6 replicates per mouse and performed with n = 3 mice per genotype. (M) Relative mRNA levels of genes involved in de novo lipogenesis in the livers of fed and 16-hours fasted Hmgcs2^ΔLiv mice. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 analyzed by the 2-tailed Student t test and 1-way analysis of variance (Tukey multiple-comparisons test).

Hepatic lipid content is controlled by a tight interaction between fatty acid uptake, secretion, synthesis, oxidation, and esterification.³⁰ The fatty acid transporter CD36 is elevated in the liver of mice with constitutive knockdown of hepatic Hmgcs2.³¹ Our qPCR analysis in the liver of Hmgcs2^ΔLiv mice showed a similar trend toward an increase in CD36 but did not reach significance. Moreover, the expression of fatty acid uptake genes, such as Fatp2 and Fatp5, was not different between the genotypes (Figure 3F). We found that the livers of fasted Hmgcs2^ΔLiv mice expressed a significantly elevated level of fat-specific protein 27 (Fsp27) and perilipin (Plin2) (Figure 3G). FSP27 increases lipid storage by inhibiting ATGL-mediated lipolysis, whereas perilipin2 confers resistance of lipid droplets to lipophagy.³² We excluded the involvement of lipolytic³³ and autophagic mechanisms³⁴ by measuring the expression of pHSL, CGI-58, ATGL, ATG5, ATG7, and LC3B, which showed no difference between Hmgcs2^F/F and Hmgcs2^ΔLiv mice (Figure 3H and I). Moreover, very-low-density lipoprotein secretion was not different between Hmgcs2^F/F and Hmgcs2^ΔLiv mice (Figure 3J), suggesting that impairment in lipid secretion does not account for steatosis in fasted Hmgcs2^ΔLiv mice. De novo lipogenesis analysis using primary hepatocytes showed a small increase in lipogenesis in Hmgcs2^ΔLiv mice (Figure 3K and L), but no difference in the mRNA levels of lipogenesis-related genes, such as Srebp1, Fasn, Acc1, and Scd1 (Figure 3M), was noted in the liver.

However, FAO was significantly reduced in the primary hepatocytes of Hmgcs2^ΔLiv mice (Figure 4A). The nuclear transcription factor peroxisome proliferator-activated receptor (PPAR)α is the master regulator of FAO,²⁹ and disruption of hepatic PPARα induces severe steatosis on acute fasting.³⁵ Assessment of hepatic PPARα and its target genes showed no difference between the genotypes, except Acox1 and Cpt2, which were increased in Hmgcs2^ΔLiv mice (Figure 4B). Thus, fasting-induced hepatic steatosis in Hmgcs2^ΔLiv mice is linked with altered lipid metabolism in the hepatocytes, particularly a decrease in FAO, despite no apparent difference in PPARα signaling. ER stress potently inhibits FAO, leading to steatosis.^36,37 Analysis for ER stress markers, such as PDI, PERK, CHOP, and IRE1α, showed no difference in their expression between the genotypes (Figure 4C), suggesting that ER stress may not be involved in the fasting-induced steatosis in the ketogenesis-insufficient mice. Mitochondrial homeostasis, including biogenesis and function, is crucial for lipid metabolism.³⁸ Because ketones induce mitochondrial biogenesis,^39,40 we assessed whether impaired ketogenesis affected mitochondrial homeostasis in Hmgcs2^ΔLiv mice. We found no difference in the mRNA expression of Pgc1α , Nrf2, and various other mitochondria-associated genes, such as Atp6, Cox1, and Nd4, in Hmgcs2^ΔLiv mice (Figure 4D). Moreover, the expression of OXPHOS proteins, such as NDUFB8, SDHB, UQCRC2, and cytochrome C oxidase, was similar between the genotypes (Figure 4E), suggesting that the temporal impairment in ketogenesis does not affect mitochondrial content. Ketogenic insufficiency in neonatal pups increases the acetylation of mitochondrial proteins leading to mitochondrial dysfunction.¹⁴ Our analysis showed increased acetylation of mitochondrial proteins in the livers of fasted Hmgcs2^ΔLiv mice (Figure 4F); therefore, we assessed mitochondrial respiratory capacity in purified liver mitochondria using Oroborus respirometry. In the presence of palmitoylcarnitine, there were no differences in mitochondrial respiration devoted to oxidative phosphorylation or maximum electron transport chain activity in the uncoupled state (Figure 4G and H). Consistent with the previously mentioned data, there was no difference in liver citrate synthase activity (Figure 4I), which serves as a surrogate of mitochondrial content.⁴¹ Together, these data suggest that the hepatic steatosis and impairment in FAO in ketogenic insufficient mice is not associated with ER stress or mitochondrial dysfunction.

Figure 4.

Ketogenic insufficiency decreases FAO without affecting mitochondrial function. (A) ¹⁴C-oleate-oxidation measured in Hmgcs2^ΔLiv primary hepatocytes treated with 200 μM BSA/ Oleate in the presence or absence of 10 μM etomoxir. n = 6 replicates per mouse and performed with n = 3 mice per genotype. (B) Relative mRNA levels of FAO genes in the livers of fed and 16-hour fasted Hmgcs2^ΔLiv mice. (C) WB analysis of ER-stress associated proteins in the liver lysates from Hmgcs2^ΔLiv mice. (D) Relative mRNA levels of mitochondrial biogenesis and content-related genes in the liver of Hmgcs2^ΔLiv mice. (E) WB analysis of Oxphos proteins in the liver of Hmgcs2^ΔLiv mice. (F) Mitochondrial AcK-100 in the liver. (G, H) Mitochondrial respiration devoted to oxidative phosphorylation or maximum electron transport chain activity in uncoupled state. (I) Liver citrate synthase enzymatic activity. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗∗∗P < 0.0001 analyzed by 1-way analysis of variance (Tukey multiple-comparisons test) and 2-tailed Student t test.

Ketogenic Insufficiency Increases Hepatic Esterification via ER-associated ACSL1

Fatty acids are esterified into fatty acyl-CoA before entering the mitochondria and ER for oxidation and triglyceride synthesis, respectively.³ ACSL is a rate-limiting enzyme that converts fatty acids into acyl-CoA.⁴² Inhibiting ACSL abolishes the mitochondrial translocation of fatty acids and FAO.^3,20 Our qPCR analysis showed elevated mRNA levels of Acsl1 in fasted Hmgcs2^ΔLiv livers (Figure 5A). However, other ACSL isoforms (Figure 5A) and esterification-related genes, such as Dgat1, Dgat2, Gpat, and lipin1 (Figure 5B), were similar between the genotypes. Consistent with the mRNA, ACSL1 protein expression was also increased in the livers of fasted Hmgcs2^ΔLiv mice (Figure 5C). ACSL1 plays a crucial role in the subcellular partitioning of fatty acids for catabolism or storage as lipid droplets.¹⁹ Mitochondria-associated ACSL1 generates Acyl-CoA for FAO, whereas ER-associated ACSL1 promotes fatty acid esterification to synthesize triglycerides.^20,43 We assessed whether ketogenesis regulates ACSL1 subcellular localization and found no difference in the mitochondrial ACSL1 protein levels between the fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice (Figure 5D). Interestingly, ER-associated ACSL1 was significantly elevated in fasted-Hmgcs2^ΔLiv livers (Figure 5D). Because ER-associated ACSL1 induces triglyceride storage via fatty acid re-esterification,²⁰ we performed thin-layer chromatography (TLC) to assess whether esterification is enhanced in Hmgcs2^ΔLiv mice. Our data showed a marked increase in the triglyceride fraction from the fasted Hmgcs2^ΔLiv mice, even after normalizing with phospholipids (Figure 5E and F). However, no difference in phospholipids, cholesterol/diacylglycerol, and cholesterol esters was noted (Figure 5G). To further determine whether ACSL1 is responsible for steatosis in Hmgcs2^ΔLiv mice, we incubated Hmgcs2^ΔLiv primary hepatocytes with BSA-conjugated PA or OA in the presence or absence of Triascin C, a specific ACSL1 inhibitor.^44,45 Notably, Triascin C alleviated lipid accumulation in primary hepatocytes of Hmgcs2^ΔLiv mice (Figure 5H-J).

Figure 5.

Ketogenic insufficiency increases ER-associated ACSL1 leading to fatty acid partitioning toward esterification. (A) Relative mRNA levels of ACSL isoforms in the livers of fed and 16-hour fasted Hmgcs2^ΔLiv mice. (B) Relative mRNA levels of genes involved in triacylglycerol synthesis in the liver. (C) WB analysis and quantification of ACSL1 in the liver of Hmgcs2^ΔLiv mice. (D) WB analysis and quantification of ACSL1 in the hepatic mitochondrial (mito) and microsomal fractions (ER) from 16-hour fasted Hmgcs2^F/F mice. (E) TLC showing lipid fractions, including phospholipids (PL), cholesterol/diacylglycerol (Chol/DG), free fatty acids (FFA), triacylglycerol (TAG), and cholesterol esters (CE). (F) Quantified ratio of TAG/PL. (G) Quantified band densities for TAG lipid fractions in the liver of fasted Hmgcs2^ΔLiv mice. (H, I) Oil-red-O staining and quantification of primary hepatocytes from Hmgcs2^ΔLiv and Hmgcs2^F/F mice incubated with 200 μM BSA/PA for 16 hours in the presence or absence of 5 μM Triascin C (20X magnification. (J) Oil-red-O staining of primary hepatocytes from Hmgcs2^ΔLiv and Hmgcs2^F/F mice incubated with 200 μM BSA/OA for 16 hours in the presence or absence of 5 μM Triascin C (60X magnification). All the data are presented as mean ± standard error of the mean. ∗∗P < 0.01, ∗∗∗∗P < 0.0001 analyzed by 1-way analysis of variance (Tukey multiple-comparisons test) and 2-tailed Student t test.

We further evaluated the role of ACSL1 in vivo by injecting Hmgcs2^ΔLiv mice with an AAV8 adenoviral vector carrying shRNA against Acsl1 (Figure 6A). Knockdown of hepatic ACSL1 (Figure 6B-D) did not affect serum β-hydroxybutyrate (Figure 6E) and triglyceride levels (Figure 6F) but fasting-mediated increase in the liver-to-body weight and hepatic steatosis were significantly attenuated in Hmgcs2^ΔLiv mice (Figure 6G and H). H&E analysis showed a reduction in microsteatosis in fasted Hmgcs2^ΔLiv mice injected with Acsl1 shRNA (Figure 6I). Further, TLC showed decreased triglycerides in fasted Hmgcs2^ΔLiv mice injected with Acsl1 shRNA (Figure 6J-L); however, knockdown of ACSL1 did not affect triglyceride fraction in the fasted Hmgcs2^F/F mice (Figure 6M and N). Collectively, our data suggest that fasting-induced hepatic steatosis in ketogenesis-insufficient mice is driven by ACSL1-mediated re-esterification of fatty acids in the ER.

Figure 6.

ACSL1 knockdown alleviates steatosis in ketogenic insufficiency mice. (A) Schematic showing the administration of AAV8-adenoviral particles containing ShScr (control) or ShAcsl1. (B) WB analysis for ACSL1 in the livers of Hmgcs2^ΔLiv and Hmgcs2^F/F mice injected with AAV8-ShAcsl1 and fasted for 16 hours. (C, D) Relative mRNA levels of indicated genes in the livers of Hmgcs2^ΔLiv and Hmgcs2^F/F mice injected with AAV8-ShAcsl1 and control (AAV8-ShScr) adenovirus and fasted for 16 hours. Graphs showing (E) serum BOH, (F) serum TAG, (G) liver to body weight ratio, and (H) liver TAG measured in AAV8-ShAcsl1-injected Hmgcs2^F/F mice fasted for 16 hours. (I) Representative images of H&E-stained livers (10X magnification). (J) TLC showing lipid fractions in AAV8-ShAcsl1-injected Hmgcs2^ΔLiv mice fasted for 16 hours. (K) Quantified ratio of TAG/PL. (L) Quantified band densities of TAG lipid fractions in the liver of Hmgcs2^ΔLiv mice. (M, N) TLC showing lipid fractions and quantified band densities for TAG lipid fractions in the liver of Hmgcs2^F/F mice. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001 analyzed by 1-way analysis of variance (Tukey multiple-comparisons test) and 2-tailed Student t test.

Ketogenic Insufficiency Exacerbates Diet-induced Hepatic Steatosis via Re-esterification

To understand whether hepatic ketogenesis insufficiency aggravates diet-induced hepatic steatosis in response to fasting, 8-week-old Hmgcs2^ΔLiv mice were treated with tamoxifen to disrupt HMGCS2 and then fed with 60% HFD for 4 weeks. We found no difference in the body weight between Hmgcs2^F/F and Hmgcs2^ΔLiv mice on an HFD (Figure 7A). As expected, ketone body levels were significantly lower in the fasted Hmgcs2^ΔLiv mice (Figure 7B). No difference in serum triglyceride and increase in cholesterol levels was noted (Figure 7C and D). Similar to chow diet conditions, the liver-to-body weight ratio was significantly higher in fasted HFD-fed Hmgcs2^ΔLiv mice (Figure 7E). H&E analysis showed microsteatosis in Hmgcs2^F/F mice, whereas Hmgcs2^ΔLiv mice showed macrosteatosis (Figure 7F). Liver triglyceride levels were significantly elevated in HFD-fed fasted Hmgcs2^ΔLiv mice (Figure 7G). Moreover, TLC analysis showed a significant increase in triglyceride fraction in HFD-fed Hmgcs2^ΔLiv mice (Figure 7H and I). But no difference in other lipid moieties was observed (Figure 7J). ER-associated ACSL1 is increased in HFD-fed Hmgcs2^ΔLiv mice (Figure 7K), but no change in mitochondria-associated ACSL1 was observed (Figure 7K). Of note, ER stress markers were not different between Hmgcs2^F/F and Hmgcs2^ΔLiv mice (Figure 7L). However, our analysis showed increased acetylation of mitochondrial proteins in the livers of fasted Hmgcs2^ΔLiv mice (Figure 7M).

Figure 7.

Ketogenic insufficiency exacerbates diet-induced hepatic steatosis via re-esterification. (A) Body weight, (B) serum β-hydroxybutyrate (BOH) levels, (C) serum TAG levels, (D) serum cholesterol, and (E) liver-to-body weight ratio in 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice fed with a 60% HFD for 4 weeks. (F) Representative H&E images (20X magnification) and (G) TAG levels from the liver of HFD-fed fasted Hmgcs2^ΔLiv mice. (H) TLC showing lipid fractions, (I) TAG/PL ratio, and (J) quantified band densities of lipid fractions in the liver of HFD-fed fasted Hmgcs2^ΔLiv mice. (K) WB analysis and quantification of ACSL1 in the hepatic microsomal (ER), mitochondrial (mito) fractions from the liver of HFD-fed fasted Hmgcs2^ΔLiv mice. (L) WB analysis showing proteins associated with ER stress in the livers of 16-hour fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice fed with a 60% HFD for 4 weeks. (M) WB analysis showing mitochondrial AcK-100 in the liver. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001 analyzed by 2-tailed Student t test.

We then knocked down ACSL1 in Hmgcs2^ΔLiv mice fed with a 60% HFD for 8 weeks (Figure 8A) and found no change in body weight and fasting glucose levels (Figure 8B and C). Similar to chow-fed mice, ACSL1 knockdown in Hmgcs2^ΔLiv mice reduced fasting-induced hepatomegaly and hepatic TAG accumulation without affecting serum TAG levels (Figure 8D-F). H&E analysis confirmed the reduction in hepatic steatosis in fasted Hmgcs2^ΔLiv mice with ACSL1 knockdown (Figure 8G). Collectively, our data show that impaired ketogenesis under nutrient-rich conditions exacerbates hepatic steatosis via ACSL1 .

Figure 8.

ACSL1 knockdown attenuates diet-induced steatosis in ketogenic insufficiency mice. (A) WB analysis of the indicated proteins in the livers of Hmgcs2^F/F and Hmgcs2^ΔLiv mice fed with a 60% HFD for 8 weeks injected with adenovirus (AAV8-ShAcsl1) and control (AAV8-ShScr) and fasted for 16 hours. (B) Body weight and (C) fasting blood glucose levels, (D) liver-to-body weight ratio, (E) liver TAG, and (F) serum TAG in Hmgcs2^F/F and Hmgcs2^ΔLiv mice fed with a 60% HFD for 8 weeks and injected with AAV8-ShAcsl1 or control AAV8-ShScr and 16-hours fasted. (G) Representative H&E images of the livers (10X magnification). All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 as analyzed by 1-way analysis of variance (Tukey multiple-comparisons test).

Impaired Ketogenesis and Increased ER-associated ACSL1 is Observed in Human MASH

MASH is the most common liver disease, where excess and sustained lipid accumulation triggers a myriad of pathologic changes culminating in liver inflammation, fibrosis, and cancer.⁷ Hepatic steatosis is majorly driven by a mismatch in the delivery (excessive lipolysis and dietary intake) and handling of lipids (fatty acid synthesis, oxidation, storage, and secretion) by the hepatocytes.¹ When the flux of fatty acids is acute, a compensatory increase in FAO and ketogenesis protects the liver from lipotoxicity.⁴⁶ However, chronic steatotic conditions decrease FAO and ketogenesis resulting in the accumulation of lipid droplets. Indeed, studies show that ketone body levels decline progressively in patients with MASH.^13,22,47 Based on our results, we investigated whether reduced hepatic ketogenesis is associated with elevated fatty acid esterification in deidentified liver tissues from patients with MASH undergoing surgical resection and compared with control human (non-MASH) subjects. As expected, patients with MASH had higher hepatic triglycerides but no difference in serum triglyceride levels was noted and β-hydroxybutyrate levels trended lower (Figure 9A-C). Furthermore, the liver from patients with MASH showed a significant decrease in the protein levels of HMGCS2 and 3-hydroxybutyrate dehydrogenase (BDH1) (Figure 9D and E); the latter catalyzes the interconversion of β-hydroxybutyrate and acetoacetate. A correlation analysis shows that triglyceride levels increase as the hepatic expression of HMGCS2 decreases (Figure 9F). Because ketogenesis is impaired in MASH, we asked whether ACSL1 translocation is modulated in the MASH livers. We found significantly elevated levels of ER-associated ACSL1 in MASH subjects (Figure 9G). Moreover, we observed a negative correlation between the ER-ACLS1 and HMGCS2 protein expression (Figure 9H); however, no difference in the mitochondria-associated ACLS1 was noticed (Figure 9I). TLC analysis showed increased levels of esterified triglyceride compared with other lipid fractions (Figure 9J-L), suggesting that impaired ketogenesis is associated with increased ER-ACSL1 -mediated re-esterification of lipids in patients with MASH. Similar to mouse models, impaired ketogenesis is also associated with elevated levels of acetylated mitochondrial proteins in human MASH livers (Figure 9M). Collectively, our data show that impaired ketogenesis induces lipid accumulation in patients with MASH in part by promoting the ER-ACSL1-mediated fatty acid re-esterification.

Figure 9.

Impaired ketogenesis in human MASH is associated with increased ER-associated ACSL1 and fatty acid esterification. (A) Liver TAG measured in non-MASH and MASH subjects. (B, C) Serum TAG and BOH levels in non-MASH and MASH subjects. (D) WB analysis for HMGCS2 and BDH1 in the livers of non-MASH and MASH subjects. (E) WB quantification for HMGCS2 and BDH1. (F) Correlative analysis between liver TAG and relative HMGCS2 protein in non-MASH and MASH subjects. (G) WB analysis showing liver ER-associated ACSL1 protein expression and their relative quantification (bottom). (H) Correlative analysis between relative liver ER-associated ACSL1 and HMGCS2 protein in non-MASH and MASH subjects. (I) WB assessing ACSL1 expression in the liver mitochondrial fraction. (J) TLC showing lipid fractions from the liver tissues. TLC quantified band densities for (K) TAG/PL ratio and (L) lipid fractions from the liver tissues. (M) WB analysis for hepatic mitochondrial AcK-100 in non-MASH and MASH livers. All the data are presented as mean ± standard error of the mean. Correlative analysis was performed using a nonparametric Spearman test. ∗P < 0.05 analyzed by the 2-tailed Student t test.

Hepatic Acetyl-CoA-L-Carnitine Homeostasis Regulates Lipid Partitioning via ACSL1

We then sought to understand how ketogenesis regulates the ER translocation of ACSL1. A recent study showed that TBK1 regulate the subcellular localization of ACLS1 by acting as a chaperone.²⁰ However, the expression of TBK1 or its phosphorylated form were similar between the fasted Hmgcs2^F/F and Hmgcs2^ΔLiv mice (Figure 10A). Ketogenesis is the major route of eliminating acetyl-CoA generated by FAO, although some amount of acetyl-CoA would be metabolized via the TCA cycle or exported out of mitochondria as citrate.^12,48 In the cytosol, citrate is converted into acetyl-CoA by ATP citrate lyase (ACLY). We found a significant increase in the mitochondrial and cytosolic levels of acetyl-CoA (Figure 10B). However, no difference in the expression of ACLY was noted (Figure 10C). Given the coincidental increase in acetyl-CoA and ER-associated ACSL1, we posited that elevated acetyl-CoA induces the subcellular localization of ACSL1 in Hmgcs2^ΔLiv mice. To test our hypothesis, we incubated mouse primary hepatocytes under lipogenic conditions with or without acetate, which increases intracellular acetyl-CoA levels (Figure 10D).49, 50, 51 Western blot analysis showed that acetate increases ER-associated ACSL1 both in the presence and absence of palmitic acid (Figure 10E), independent of ER stress (Figure 10F). Although acetate alone did not affect the lipid droplet content, cotreatment with PA and OA increased lipid accumulation in primary hepatocytes (Figure 10G and H), suggesting that elevated acetyl-CoA induces ACSL1 translocation to ER and exacerbates steatosis under lipogenic conditions. ACSL1 activity has been reported to be regulated by acetylation of Lys544 in the mitochondria from fasted mice.^52,53 Our analysis by immunoprecipitation using acetyl-lysine antibody did not reveal ACSL1 acetylation in both the genotypes (Figure 10I). β-Hydroxybutyrylation is characterized as 1 of the posttranslational modifications (PTM) that regulate protein function. A recent study identified lysine-β-hydroxybutyrylation sites on ACSL1.⁵⁴ Our effort to assess whether β-hydroxybutyrylation regulate ACSL1 function was hindered by the nonspecificity of the available commercial antibody against lysine-β-hydroxybutyryl residues (Figure 10J). Future investigations are warranted to determine whether PTM govern ACSL1-ER membrane translocation and function.

Figure 10.

Hepatic acetyl-CoA induces ACLS1 ER translocation and steatosis. (A) WB analysis of the indicated proteins in the livers of Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (B) Acetyl-CoA levels in the mitochondria and cytosol of fasted Hmgcs2^ΔLiv mice. (C) WB analysis of the indicated proteins in the livers of Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (D) Acetyl-CoA levels in primary hepatocytes treated with 20 mM sodium acetate for 16 hours. (E) WB analysis for ER-ACSL1 in primary hepatocytes incubated with 200 μM BSA/PA in the presence or absence of 20 mM sodium acetate (Ac) for 16 hours and relative band quantification of ER-ACSL1. (F) WB analysis of the indicated proteins in primary hepatocytes treated with sodium acetate (20 mM) for 16 hours. (G) Oil-red-O staining and quantification in primary hepatocytes incubated with 200 μM BSA-PA in the presence or absence of 20 mM sodium acetate for 16 hours (60X magnification). (H) Oil-red-O staining and TAG quantification in primary hepatocytes incubated with 200 μM BSA-OA in the presence or absence of 20 mM sodium acetate or NaCl for 16 hours (60X magnification). (I) Whole cell liver lysate and primary hepatocyte IP analysis with Ac-lysine K-100 antibody for ACSL1 in fasted in Hmgcs2^F/F and Hmgcs2^ΔLiv mice. (J) WB analysis for pan-β hydroxybutyryllysine (kbhb; # PTM-1201, PTM Bio, Chicago, IL) in whole cell liver lysate. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗or P < 0.0001 analyzed by 2-tailed Student t test.

We then interrogated ways to decrease acetyl-CoA levels as a resort to alleviate hepatic steatosis under conditions of impaired ketogenesis. LC acts as a buffer for acetyl-CoA by forming acetyl-carnitine (ie, excess acetyl-CoA exists in the form of acetyl-carnitine).⁵⁵ Because LC is required for the mitochondrial import of acyl-CoA, excess mitochondrial acetyl-CoA impair mitochondrial FAO by reducing free LC levels.^56,57 Our analysis showed significantly reduced free LC levels in the livers of fasted Hmgcs2^ΔLiv mice (Figure 11A), which partly explains impaired FAO in Hmgcs2^ΔLiv mice. We then tested whether elevating LC would reduce acetyl-CoA levels and restore lipid homeostasis in Hmgcs2^ΔLiv mice. To this end, we provided fasted Hmgcs2^ΔLiv mice with drinking water containing LC at 10 mg/mL. No change in body weight was observed by LC supplementation (Figure 11B). As expected, LC supplementation reduced acetyl-CoA levels in the fasted Hmgcs2^ΔLiv mice (Figure 11C). We also noted a modest decrease in ACLY expression in LC-treated mice (Figure 11D). Remarkably, the livers of LC-treated Hmgcs2^ΔLiv mice were reddish compared with untreated littermates (Figure 11E). LC treatment significantly reduced the hepatomegaly and steatosis in Hmgcs2^ΔLiv mice (Figure 11F and G). H&E analysis confirmed the reduction in microsteatosis in LC-treated fasted Hmgcs2^ΔLiv mice (Figure 11H). Because LC plays a significant role in transferring the long-chain fatty acids for mitochondrial oxidation, we assessed whether LC treatment affects FAO using primary hepatocytes.^56,58 Our analysis showed that carnitine restores FAO in the primary hepatocytes of Hmgcs2^ΔLiv mice (Figure 11I). We further demonstrate that LC supplementation decreased the ER localization of ACSL1 (Figure 11J), resulting in reduced triglyceride esterification in fasted Hmgcs2^ΔLiv mice (Figure 11K-M). Additionally, we noticed that LC decreased the levels of acetylated mitochondrial proteins in Hmgcs2^ΔLiv mice (Figure 11N) but no difference was observed in ER stress markers (Figure 11O). Together, our data indicate that LC recuperates acetyl-CoA homeostasis, restores FAO, and alleviates ER-ACSL1-mediated hepatic steatosis in mice with impaired ketogenesis.

Figure 11.

Hepatic acetyl-CoA-L-carnitine homeostasis regulates lipid partitioning. (A) LC levels were measured in the livers of fasted Hmgcs2^ΔLiv mice. (B) Body weight change. (C) Hepatic acetyl-CoA levels in fasted Hmgcs2^ΔLiv mice treated with or without 10 mg/mL LC. (D) WB assessing ACLY expression in the liver. (E) Gross liver image, (F) liver-to-body weight, (G) liver TAG, and (H) H&E analysis of the livers from fasted Hmgcs2^ΔLiv mice treated with or without LC. (I) ¹⁴C-oleate-oxidation measured in Hmgcs2^ΔLiv primary hepatocytes treated with 200 μM BSA/Oleate in the presence or absence of 1 mM carnitine with 10 μM etomoxir. n = 3 replicates per mouse and performed with n = 3 mice per genotype. (J) WB analysis showing ER-ACSL1 in the livers of fasted Hmgcs2^ΔLiv mice treated with or without LC. (K) TLC showing lipid fractions, (L) TAG/PL ratio, and (M) TLC quantified band densities of lipid fractions in the livers of fasted Hmgcs2^ΔLiv mice treated with or without LC. (N) WB analysis for AcK in the mitochondrial fractions of fasted Hmgcs2^ΔLiv mice treated with or without LC. (O) WB analysis of the indicated proteins in the liver of fasted Hmgcs2^ΔLiv mice treated with or without LC. All the data are presented as mean ± standard error of the mean. ∗P < 0.05, ∗∗P < 0.01 analyzed by 2-tailed Student t test.

LC Alleviates Steatosis in Human Hepatocytes Partly by Reducing Acetyl-CoA and ER-associated ACSL1

We then evaluated the relationship between acetyl-CoA and ACSL1-mediated esterification in the livers of patients with MASH. As shown in previous studies,^22,59 liver acetyl-CoA levels were elevated in MASH subjects (Figure 12A). Correlative analysis showed that ER-associated ACSL1 was in proportion with acetyl-CoA levels in human MASH livers (Figure 12B). Similar to the mouse models, free LC levels were significantly reduced in MASH livers and were positively correlated with hepatic HMGCS2 expression (Figure 12C and D). Previous studies have demonstrated that LC improves steatosis in patients with MASH by increasing mitochondrial function and FAO.^60,61 We tested whether LC affects fatty acid esterification using primary human hepatocytes isolated from patients with MASH. BODIPY staining showed reduced lipid accumulation with LC treatment in primary human hepatocytes cultured under lipogenic conditions (Figure 12E). We demonstrate that LC decreases the ER-associated ACSL1 (Figure 12F) and fatty acid esterification (Figure 12G). In contrast to mouse livers, LC increased ACSL1 levels in the mitochondria and the levels of acetylated mitochondrial proteins were decreased (Figure 12H). Collectively, we demonstrate that the antisteatosis effect of LC is partly associated with decreased ER translocation of ACSL1 in lipid-laden human hepatocytes.

Figure 12.

L-carnitine alleviates steatosis in human MASH partly by reducing ACSL1 -mediated esterification of fatty acids. (A) Acetyl-CoA levels in the livers of non-MASH and MASH subjects. (B) Correlation analysis for liver acetyl-CoA levels and relative protein expression of hepatic ER-localized ACSL1. (C) Hepatic L-carnitine levels. (D) Correlation analysis for liver L-carnitine levels and relative protein levels of HMGCS2 in non-MASH and MASH subjects. (E) Images of BODIPY-stained primary human hepatocytes from MASH subjects pretreated with 1 mM L-carnitine for 6 hours and loaded with 200 μm BSA-PA for 16 hours (20X magnification). (F) WB analysis of ACSL1 in the microsomal fractions from primary human hepatocytes from MASH subjects pretreated with 1 mM L-carnitine for 6 hours and loaded with 200 μm BSA-PA for 16 hours. (G) TLC showing lipid fractions in primary human hepatocytes from subjects with MASH pretreated with 1 mM L-carnitine for 6 hours and loaded with 200 μm BSA-PA for 16 hours. (H) WB analysis showing mitochondrial ACLS1 and AcK-100 in primary human hepatocytes from subjects with MASH pretreated with 1 mM L-carnitine for 6 hours and loaded with 200 μm BSA-PA for 16 hours. (I) Schematic diagram generated using BioRender software. All the data are presented as mean ± standard error of the mean. Correlative analysis was performed using a nonparametric Spearman test. subjects with MASH pretreated ∗∗P < 0.01 analyzed by the 2-tailed Student t test.

Discussion

Perturbations in hepatic fatty acid metabolism emerge as a key risk factor driving various metabolic diseases, including obesity, diabetes, and MASLD. An intricate interplay of increased free fatty acid influx, augmented lipogenesis, and impaired lipid disposal in the liver results in triglyceride accumulation, triggering hepatic steatosis.^46,62,63 Lipid disposal pathways, particularly β-oxidation, plays a crucial role in eliminating excessive hepatic lipids under physiological and pathologic conditions. Despite the liver’s capability to dispose of up to 250 g of lipid per day through β-oxidation and ketogenesis during fasting,⁶⁴ the specific role of hepatic ketogenesis in regulating lipid metabolism remains unclear.

In this study, we present evidence that mice with temporal disruption of hepatic ketogenesis exhibit severe steatosis on fasting. Ketogenesis is identified as a regulator of fatty acid partitioning via ER-associated ACSL1. We demonstrate that insufficient ketogenesis induces the accumulation of acetyl-CoA, leading to the ER translocation of ACSL1 and subsequent fatty acid esterification. Consequently, incoming fatty acids are misdirected for storage rather than undergoing FAO, because elevated acetyl-CoA levels from ketogenic insufficiency are misinterpreted as excessive FAO. Furthermore, LC supplementation in ketogenic-insufficient mice counteracts acetyl-CoA-mediated increase in ER-associated ACSL1 and triglyceride esterification (Figure 12I). Elevated acetyl-CoA, ER-associated ACSL1, and enhanced triglyceride esterification are also observed in patients with MASH with impaired ketogenesis, revealing a novel role of ketogenesis in regulating hepatic lipid metabolism.

Previous studies have demonstrated the causal role of ketogenesis in fatty liver pathogenesis, particularly in neonates and adult mice with ketogenic insufficiency (via disruption of hepatic HMGCS2).^14,15 These mice develop fatty liver when exposed to lipid-rich mothers' milk and an HFD.¹⁵ Our findings show that acute fasting alone is sufficient to induce steatosis in ketogenesis-insufficient mice. Despite the availability of food ad libitum, mice tend to eat at night, undergoing physiological fasting during the day. No difference in liver triglyceride levels in ad libitum–fed ketogenesis-insufficient mice indicates that a small increase in fatty acids from physiological fasting is well-tolerated. Refeeding induces the secretion of hepatic triglycerides as very-low-density lipoprotein, which are taken up by peripheral tissues, including adipose tissue, heart, kidney, and vascular tissue.⁶³ Previous studies showed that ketogenesis regulates the composition of hepatic lipids.^16,65 Investigation into the impact of ketogenic insufficiency on peripheral lipid homeostasis may provide insight into the mechanistic underpinnings of metabolic diseases, including cardiovascular disease.

Fasting-mediated adipose tissue lipolysis and a drop in insulin levels coordinate to induce hepatic ketogenesis. Inhibition of adipose tissue ATGL is shown to decrease circulating ketone levels.²⁹ Under these circumstances, ketogenesis is tightly linked with FAO, evident from PPARα knockout mice eliciting diminished fasting-associated FAO and ketogenesis, despite no difference in circulating fatty acids.³⁵ Despite no impairment in hepatic PPARα signaling or lipogenic pathways our data reveal increased fatty acid storage ketogenetic-insufficient mice, evident from microsteatosis and increased expression of lipid droplet-associated proteins.

Fatty acids entering hepatocytes are immediately esterified into acyl-CoA by ACSL1. Although mitochondrial ACSL1 directs acyl-CoA for β-oxidation,^19,20 ER-localized ACSL1 generates lipid substrates for triglyceride synthesis via re-esterification.⁶² Our study establishes the tight link between impaired ketogenesis and ER localization of ACSL1 in fasted mice, leading to triglyceride esterification and steatosis, a mechanism also observed in human MASH livers. Acetyl-CoA, a metabolic node regulating glucose, lipid, and amino acid metabolism,⁶⁶ increases during fasting and high-fat feeding.⁶⁷ The conduit of acetyl-CoA as ketone bodies is crucial to maintaining cellular acetyl-CoA levels. When FAO exceeds ketogenesis, the increase in acetyl-CoA pauses mitochondrial FAO by acetylating mitochondrial proteins^68,69 and channels the incoming fatty acids toward ER for storage as triglycerides. However, mitochondrial function remains unaffected in fasted ketogenic insufficient mice, suggesting that mitochondrial dysfunction is not the primary cause of lipid accumulation. Excess acetyl-CoA shuttles to the cytosol as citrate, where cytosolic ACLY breaks down citrate into acetyl-CoA and oxaloacetate.⁶⁶ Elevating cytosolic acetyl-CoA using acetate⁷⁰ increases ER-translocation of ACSL1 and lipid-laden hepatocytes, supporting the hypothesis that cytosolic acetyl-CoA promotes fatty liver by inducing the esterification of lipids via ACSL1. Although de novo lipogenesis is not induced during fasting, elevated acetyl-CoA levels may later fuel lipogenesis and contribute to postprandial hepatic steatosis,⁷¹ especially when hepatic lipid clearance is compromised, as seen in MASH.

Mitochondrial membranes being impermeable to fatty acyl-CoA, acyl-CoA is esterified with LC to form acyl-carnitine. No export mechanism exists to excrete cellular acetyl-CoA and LC buffers acetyl-CoA forming acetyl-carnitine, which is excreted in the urine.^72,73 A decrease in carnitine levels is observed in the latter stage of liver disease^74,75 indicating that carnitine deficiency may not trigger hepatic steatosis but could exacerbate disease progression. However, patients with primary carnitine deficiency caused by a lack of organic cation/carnitine transporter develop hepatomegaly and hepatic steatosis, which is alleviated by LC supplementation.76, 77, 78 Moreover, LC is rate-limiting for ketogenesis from endogenous fatty acids and carnitine supplementation during ketosis improves glucose homeostasis, insulin sensitivity, and lipid profile.^79,80 Thus, carnitine is a key determinant of lipid metabolism, ketogenesis, and hepatic steatosis. We demonstrate that carnitine supplementation attenuates hepatic steatosis under ketogenic insufficiency by alleviating ER translocation of ACSL1. We posit that carnitine, via forming acetyl-carnitine, acts as a conduit to remove excess acetyl-CoA, highlighting its potential therapeutic use in a subset of patients with MASLD exhibiting lower carnitine and ketone bodies. Several studies have demonstrated the beneficial anti-inflammatory role of acetyl-carnitine in metabolic diseases.^81,82 Further investigation is required to determine whether carnitine sequestration as acetyl-carnitine mediates the anti-inflammatory and mitochondrial dysfunctional response in patients with MASH exhibiting impaired ketogenesis.

Conclusions

Hepatic ketogenesis traditionally serves a selfless function in providing nutrients to other organs.¹² However, a recent study demonstrated that ketones are dispensable fuel during fasting, and mice with ketogenic haploinsufficiency tolerate fasting potentially through alternate fuel use mechanisms.²⁴ Our data reveal that fasting ketogenesis has a crucial role in regulating lipid partitioning and metabolism in the liver. When ketogenesis is impaired, fasting elevates acetyl-CoA levels resulting in ACSL1-mediated triglyceride esterification, leading to hepatic steatosis.^19,20,83 Notably, fasting ketogenic haploinsufficient mice induces ACOT12, the enzyme involved in converting acetyl-CoA to acetate, elevating serum acetate levels. Further studies are needed to determine the crucial role of acetyl-CoA-acetate balance in hepatic steatosis, particularly when ketogenesis is impaired.²⁴ Under conditions of increased fatty acid inflow and impaired hepatic lipid clearance, which is often observed in MASH, ketogenic insufficiency may significantly contribute to MASH progression. Although pharmacologic inhibition of ACSL1 ameliorates steatosis,^18,20 the risk of liver injury is enhanced by accumulating free fatty acids.⁸⁴ We provide empirical evidence that LC inhibits ER translocation of ACSL1 and hepatic steatosis by buffering acetyl-CoA. However, LC should be recommended with utmost care because carnitine metabolism in the intestine generates TMAO, a metabolite with atherogenic and carcinogenic properties.^85,86 Because carnitine is excreted as acetyl-carnitine,⁸⁷ therapeutic interventions with LC could be monitored noninvasively in patients with MASH by measuring urinary acetyl-carnitine levels.

Study Limitations

Although our study shows that ACSL1 exacerbates hepatic steatosis in ketogenic haploinsufficient mice, the mechanistic underpinnings of ACSL1 association with ER remains unclear. ACSL1 activity has been reported to be regulated by PTM, including phosphorylation and acetylation.⁵² Moreover, a recent study identified lysine-β-hydroxybutyrylation sites on ACSL1.⁵⁴ Future investigations characterizing ACSL1-PTM may reveal the mechanism of ACSL1 membrane translocation and function. Our non-MASH patient samples characterized by pathologic lesions exhibited features of noncirrhotic hypertension, nodular regenerative hyperplasia, and mild portal fibrosis, which may have confounding effects on ketogenesis and steatosis. Therefore, future studies are required to characterize the relationship between ketogenesis and esterification along the spectrum of MASLD including steatosis, inflammation, and fibrosis. Moreover, our studies did not account for any mutations, such as PNPLA3, which is shown to regulate ketogenesis in patients with MASLD.⁸⁸

Materials and Methods

Human MASH Subjects and Characteristics

A total of 20 human subjects (10 MASH and 10 non-MASH) scheduled for liver biopsies to diagnose MASH were recruited by participating physicians at the University of Pittsburgh Medical Center (Pittsburgh, PA). All patients fulfilled the inclusion criteria, such as no positive for viral hepatitis, Wilson disease, or any other causes of liver dysfunction. The patients were excluded if the alcohol consumption exceeded 20 g/week. The body mass index ranged from 19.15 to 38.41 and age from 33 to 68 years. All the patients were fasted at the time of biopsy. The patients with cirrhosis and MASH as the etiology were compared with the groups with no MASH diagnosis. MASH was diagnosed by pathologists based on steatosis [0 (<5%) – 3 (>66)], lobular inflammation [0 (none) - 3 (>4 foci/200x], hepatocyte ballooning [0 (none) - 2 (many/prominent ballooning)], and fibrosis [0 (none)- 4 (cirrhosis)]. Detailed characteristics of the patients with MASH and other diseases are outlined in Supplementary Table 1.

Animal Studies

We generated a Hmgcs2^F/F founder line on a C57BL6 background using CRISPR/Cas-9-mediated genome engineering to flox exon 2. Mice with conditional knockout of HMGCS2 (Hmgcs2^ΔLiv mice) were generated by crossing Hmgcs2^F/F mice with mice carrying tamoxifen-inducible Cre recombinase under the control of albumin promoter (Alb-Cre/ERT2). Littermates not expressing Cre recombinase served as control animals. Tamoxifen (Cayman Chemical, Ann Arbor, MI) was dissolved at 10 mg/mL in corn oil (Sigma-Aldrich, St. Louis, MO) and administered intraperitoneally at a dose of 100 mg/kg body weight for 2 consecutive days. All experiments were carried out at least 1 week after tamoxifen injection. Animals were maintained on a chow diet provided by the Division of Laboratory Animal Resources (DLAR). For high-fat studies, mice were fed an HFD (60% kcal from fat, D12492; Research Diets) for 4 weeks. Mice were euthanized in the fed state or after overnight fast (6:00 pm–10:00 am). For time course fasting experiments, mice were fasted for 0, 4, 8, 16, and 24 hours. For the LC experiment, 6- to 8-week-old Hmgcs2^ΔLiv mice were provided with drinking water containing LC (10 mg/mL; TCI, Portland, OR; C0049) 9 hours before fasting. For knocking down Acsl1, chow-fed and 8-week HFD-fed mice were injected with AAV8 adenoviral vectors carrying (shRNA) EGFP-U6>Scramble (Vector Builder, Chicago, IL; Cat: AAVCP (VB010000-0023jze-f) or pAAV (shRNA) EGFP-U6-mACSL1 (Vector Builder; Cat: AAV8/MP/VB2308141350ptx). AAV8 virus was administered at a dose of 2.5 x 10¹² vector genomes per mouse and maintained for an additional 7 days. Mice were sacrificed after overnight fasting. All mice were maintained on a 12-hour light/dark cycle with free access to food and water. All experiments used age- and sex-matched littermates, unless indicated otherwise. Both males and females were included. All animal experiments were approved by the Animal Care and Use Committee at the University of Pittsburgh.

Human Primary Hepatocyte Isolation

Primary hepatocytes were isolated from explanted human liver segments from patients undergoing orthotopic liver transplantation for decompensated liver cirrhosis caused by MASH. Specimens were obtained by the Human Synthetic Liver Biology Core at the Pittsburgh Liver Research Center under the protocol approved by the Human Research Review Committee and the Institutional Review Board (IRB STUDY20090069) at the University of Pittsburgh. Liver specimens were protected from ischemic injury by flushing with ice-cold University of Wisconsin solution immediately after resection, keeping them on ice, and transporting them immediately to the laboratory. Hepatocytes were isolated by a modified 3-step perfusion technique. Briefly, livers were flushed under a sterile biosafety hood through the hepatic vessels (recirculation technique) with prewarmed calcium-free Hanks' Balanced Salt Solution (Sigma-Aldrich, H6648-1L) supplemented with 0.5 mM EGTA (Thermo Fisher, 50-255-956) and then with collagenase/protease solution (VitaCyte, 007-1010) until fully digested. The digestion time was 45–60 minutes. The digested liver was removed and immediately cooled with ice-cold Leibovitz's L-15 Medium (Invitrogen, 11415114) supplemented with 10% fetal bovine serum (Sigma-Aldrich, F4135). The cell suspension was centrifuged twice at 65 x g for 7 minutes at 4^oC and the medium was aspirated. The yield and viability of freshly isolated hepatocytes were assessed using trypan blue staining.

Mouse Primary Hepatocyte Isolation

Mouse primary hepatocytes were isolated from 6- to 8-week-old Hmgcs2^ΔLiv male mice. Briefly, mice were anesthetized with isoflurane, and the inferior vena cava was cannulated and infused with 10 mL of perfusion buffer (Hanks' Balanced Salt Solution with no Ca²⁺, no Mg²⁺, no phenol red, supplemented with 0.5 mM of EDTA and 25 mM of HEPES, pH 7.4) at 3 mL/min. Then, 10 mL of digestion buffer (Hanks' Balanced Salt Solution with no Ca²⁺, no Mg²⁺, no phenol red, supplemented with 25 mM of HEPES, pH 7.4) with collagenase type 1 (15 mg/50 mL) was infused using a peristaltic pump. The liver was excised, minced, filtered using a 70-μm cell strainer (Falcon), and centrifuged at 50 x g for 2 minutes to pellet hepatocytes. Dead cells were removed by centrifugation at 50 x g for 10 minutes in 90% Percoll solution (Sigma-Aldrich). Hepatocyte were resuspended in Dulbecco's Modified Eagle Medium (Gibco) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. The supernatant was centrifuged at 1200 x g for 10 minutes to remove nonparenchymal cells. Cells were counted and plated at a density of 0.2 x 10⁶cells/mL on a 12-well or 10-cm plates. After 4 hours, the cells were refreshed with Williams E media supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. For Triascin C treatment, hepatocytes were incubated with 200 μM BSA-conjugated PA or OA (BSA-PA or BSA-OA; Cayman Chemical) and treated with 5 μM Triascin C (Enzo Life Sciences, Farmingdale, NY) or dimethyl sulfoxide for 16 hours. For acetate treatment, mouse primary hepatocytes were incubated for 16 hours with 200 μM BSA-PA or BSA-OA or BSA in the presence of 20 mM sodium chloride or sodium acetate (Sigma-Aldrich).

De Novo Lipogenesis

De novo lipogenesis was assessed as described in previous studies.⁸⁹ In brief, primary hepatocytes (0.4 x 10⁶cells/well) were plated for 24 hours and then incubated overnight with experimental media composed of Dulbecco's Modified Eagle Medium with 1 μCi of glucose, D-¹⁴C (PerkinElmer) with or without 100 nM insulin (Sigma-Aldrich). The cells were washed with 1X phosphate-buffered saline (PBS) and pelleted by trypsinization. Total lipids were extracted by Folch’s lipid extraction method with 1 mL of chloroform: methanol and incubation at room temperature with gentle agitation for 2 hours. One hundred microliters of 1M H₂SO₄ was added to the homogenate and centrifuged at 2000 rpm for 10 minutes and the bottom organic layer was transferred to glass scintillation vials. After complete evaporation, the radioactivity in fatty acid fraction was counted and the values were normalized to protein concentrations.

Fatty Acid Oxidation Assay

FAO assay was performed as described in previous studies.⁹⁰ In brief, primary hepatocytes plated overnight (0.4 x 10⁶cells/well) in 6-wells and treated with serum-free FAO media composed of Dulbecco's Modified Eagle Medium, 200 μM BSA-OA or BSA control (Cayman Chemical), 1 μCi of 1-¹⁴C oleic acid (PerkinElmer), and 1 mM Carnitine (Cayman Chemical). FAO was inhibited using 10 μM Etomoxir (Cayman Chemical). After 3-hour incubation at 37^oC, the media was collected and centrifuged at 13,000 rpm for 5 minutes, and 900 μL of media was transferred into acidification vials with 500 μL of 1M NaOH. Two hundred fifty microliters of 1M perchloric acid was injected directly into the media, gently swirled, and incubated overnight. The captured ¹⁴CO₂ was used to measure radioactivity for FAO and values were normalized to total protein concentrations.

Hepatic Glycogen Assay

Hepatic glycogen levels were measured using a glycogen assay kit (Cell Biolabs, Inc, San Diego, CA) as per the kit instructions. The liver (≅10 mg) was homogenized in 1X PBS and centrifuged at 10,000 rpm for 10 minutes at 4^oC. An aliquot (25 μL) of the supernatant was used for the assay and the samples were added with or without amyloglucosidase (to measure endogenous glucose background). The reaction was initiated by adding the reaction mixture provided with the kit and glycogen levels were normalized to tissue weight.

Administration of Poloxamer-407

Six- to 8-week-old Hmgcs2^F/F and Hmgcs2^ΔLiv mice were fasted for 4 hours. Poloxamer-407 (dissolved in saline at 10 mg/mL, Sigma-Aldrich) and saline (vehicle) were injected intraperitoneally at 1 mg/kg body weight.²⁰ Blood was collected at 0, 1, 2, and 3 hours to measure plasma triglyceride levels using triglyceride reagent as per the kit instructions (Thermo Fischer scientific). The triglyceride levels were represented as mg/mL.

Neutral Lipid Staining Using Oil-Red-O and BODIPY

Cells were washed with 1X PBS, fixed with 10% PBS-buffered formalin at room temperature for 30 minutes, washed with ddH₂O, and incubated with 60% isopropyl alcohol for 5 minutes. Cells were stained with 0.5% Oil-red-O (Sigma-Aldrich) for 30–45 minutes, counterstained with hematoxylin for 15 seconds, and washed several times with ddH₂O. The images were captured using an EVOS microscope (Olympus, Tokyo, Japan). Intracellular lipids were quantified by extracting the stain using 250 μL isopropanol and absorbance was measured at 492 nm. BODIPY staining was performed by incubating formalin-fixed cells with 250 μL of BODIPY (1 mg/mL stock; diluted 1:1000; Cayman Chemical) for 15–20 minutes. The cells were washed with PBS, and images were captured using a Nikon microscope.

Blood and Serum Analysis

The tail was snipped to measure blood glucose using a glucometer (Bayer, Parsippany, NJ). Serum β-hydroxybutyrate levels were measured using a calorimetry kit following manufacturers' instructions (Cayman Chemical). Serum triglycerides and cholesterol levels were measured using colorimetric Infinity Triglyceride and Cholesterol kits (Thermo Fisher Scientific, Middletown, VA). Serum nonesterified fatty acid was quantified using a colorimetric assay (Fujifilm Wako Diagnostics, Lexington, MA). Serum L-lactate and glucagon levels were measured using L-lactate assay kit (Eton Bio, San Diego, CA) and Mouse glucagon enzyme-linked immunosorbent assay kit (Crystal Chem, Elk Grove Village, IL), respectively.

Liver Triglyceride Assay

Hepatic triglyceride levels were measured as described previously.⁹¹ Briefly, frozen liver tissues (∼50 mg) were homogenized in 3 mL of chloroform: methanol (2:1) mixture and incubated for 60 minutes at room temperature while rotating on a shaker. We also homogenized the liver tissue in ice-cold PBS followed by chloroform: methanol to normalize TAGs based on protein concentration. The homogenate was acidified with 1 mol/L H₂SO₄ and lipid fractions in the lower organic phase were collected by centrifugation at 1300 rpm for 10 minutes and transferred to clean glass vials. Triglyceride levels were determined using the triglyceride reagent (Thermo Fisher Scientific) and normalized to liver weight or protein concentrations, as indicated in the figure legends.

Thin Layer Chromatography

TLC was performed as described in previous studies.²⁰ Hepatic lipids were extracted from human and mouse livers⁹¹ and an equal amount of lipids (after normalizing with tissue weights) were dried completely under a N-EVAP nitrogen evaporator (Organomation, Berlin, MA) and resuspended with 100 μL chloroform/methanol (2:1). Before running the TLC plate, the running chamber was equilibrated with 200 mL of the solvent system containing petroleum ether/ethyl ether/acetic acid mixture (25:5:1). TLC plate was baked at 75°C for 30 minutes and aliquots of 25 μL resuspended lipid extracts and lipid standard (Nu-Chek-Prep, Elysian, MN) were loaded onto the TLC plates (EMD Millipore). Lipids were separated using the solvent system. The plate was then transferred to an equilibrated iodine tank for about 30–45 minutes for staining, and images were captured with CanoScan LidE 220 imager (Canon). ImageJ software (National Institutes of Health, Bethesda, MD) was used to quantify the band density of individual lipid fractions. The changes in triglyceride fractions were normalized with phospholipid fractions.

Liver Histology

Liver tissue was excised and immediately fixed in 10% PBS-buffered formalin (Thermo Fisher Scientific). H&E staining was performed in 6-μm paraffin-embedded sections and the images were captured using an EVOS microscope. Immunohistochemistry was performed as described in previous studies. Briefly, paraffin-embedded sections were deparaffinized, rehydrated, and the antigen was retrieved using 10 mM sodium citrate buffer, 0.05% Tween20, pH 6.0. After cooling, the slides were permeabilized with TBST and treated with 3% hydrogen peroxide in methanol for 5 minutes. After several washings, the slides were blocked with 5% goat serum containing 1% BSA for 30 minutes. The slides were then incubated overnight at 4°C with primary antibodies against F4/80 (1:250, Cell Signaling) and CD45 (1:250, Cell Signaling). After several washings, the slides were incubated with Dky Rabbit IgG Biotin secondary antibody (1:500, EMD Millipore) and developed using DAB peroxidase substrate kit (Vector Laboratories, Burlingame, CA). The color development time was optimized by monitoring the signal under the microscope. The nuclei were stained with hematoxylin and the slides were mounted using Permount mounting medium (Fischer scientific) and the images were captured using a Nikon microscope.

RNA Isolation, cDNA Synthesis, and qPCR Analysis

Total RNA was extracted using TRIzol reagent (Life Technologies) and 1 μg of RNA was reverse-transcribed using Mu-MLV reverse transcriptase (Promega, Madison, WI). mRNA levels were analyzed using SYBR Green qPCR Master Mix (ApexBio, Houston, TX) with QuantStudio 3 Station qPCR machine (Applied Biosystems, Foster City, CA). The relative expression of target genes was calculated using a comparative delta threshold cycles (ΔCT) method after normalizing to β-actin. The primer sequences are provided in Supplementary Table 2.

Western Blotting

Cells and frozen tissues (∼10 mg) were homogenized using radioimmunoprecipitation assay lysis buffer (0.5% NP-40, 0.1% sodium deoxycholate, 150 mmol/L NaCl, 50 mmol/L Tris-Cl, pH 7.5) containing 1 mmol/L PMSF, protease inhibitor cocktail (Sigma-Aldrich), and 2 mmol/L sodium orthovanadate.⁹² The homogenate was centrifuged at 13,000 x g for 10 minutes at 4°C to collect the supernatant and protein concentration was quantified using a protein assay kit (Bio-Rad, Hercules, CA), and the sample was resolved on. For immunoprecipitation, 500 μg of protein lysate was incubated with anti–acetyl lysine antibody (Cell Signaling, #9814) overnight at 4°C and pulldown was performed with protein A Dynabeads (Thermo Fisher Scientific). Proteins were resolved on a sodium dodecyl sulfate–polyacrylamide gel electrophoresis and the membranes were blocked with 3% skim milk and incubated overnight at 4°C with primary antibodies. Secondary antibodies conjugated with DyLight (Cell Signaling Technology) were added to the membranes (antibody details are in Supplementary Table 3) and visualized using the Odyssey CLx Imaging System (LI-COR, Lincoln, NE).

Subcellular Fractionation

Mitochondrial and microsomal fractions were isolated as described previously.²⁰ In brief, cell pellet and liver tissues were minced in mitochondrial isolation buffer (MSHE buffer; 70 mM sucrose, 210 mM mannitol, 5 mM HEPES, 1 mM EGTA, 2% fatty acid-free BSA, pH 7.2) supplemented with protease and phosphatase inhibitors. The minced tissues were homogenized using a Teflon glass homogenizer and centrifuged at 800 x g for 10 minutes to remove nuclei and cell debris. The supernatant was centrifuged at 8000 x g for 10 minutes and the mitochondrial pellet was washed with MSHE buffer. The supernatant containing the microsomal fraction was transferred to a new tube containing 8 mM of CaCl₂ and incubated on a rotating platform in the cold room for 10 minutes.²⁰ The microsomal pellet was collected by centrifuging at 30000 x g for 30 minutes. The mitochondrial and microsomal pellets were resuspended in RIPA buffer and used for Western blot analysis.

Crude Mitochondria Isolation for Respirometry

A half lobe of the freshly sampled liver was transferred to ice-cold PBS, cut into small pieces, and washed 3 times with PBS. Two hundred milligrams of the liver was weighed and transferred to 2 mL of SMET buffer (10 mM Tris HCl [pH 7.5], 220 mM Mannitol, 70 mM Sucrose, 1 mM EDTA, 0.25% BSA) and homogenized using a glass dounce homogenizer. Homogenates were centrifuged at 800 × g for 10 minutes at 4°C to discard cell debris and nuclei. A fraction of the supernatant was saved for protein assays and citrate synthase activity measurement. The remaining supernatant was centrifuged at 8000 x g for 10 minutes at 4°C to pellet crude mitochondria. Pellets were resuspended in 400 μL SMET buffer and 80 μL was used for respirometry assays. Remaining pellet suspension was centrifuged at 8000 x g for 10 minutes at 4°C and saved for protein quantification. Mitochondrial respiratory capacity was assessed using an Oroboros O₂K High-Resolution Respirometer and MiR05 buffer at 37°C under constant mixing in a sealed 2-mL chamber. Respirometry assays were performed to determine respiration devoted to ATP synthesis, maximum or uncoupled electron transport chain capacity, and nonmitochondrial respiration. Membrane integrity was assessed by adding adenosine diphosphate followed by cytochrome C. The substrate and inhibitor concentrations were as follows: palmitoylcarnitine (40 μM), malate (2 mM); adenosine diphosphate (4 mM); cytochrome c (10 μM), FCCP (1 μL titration), and antimycin A (2.5 μM). The protein concentration of the crude mitochondria was determined by BCA protein assay and used to express respiratory capacities as oxygen consumption per protein mass (pmol s⁻¹ mg⁻¹). Respiration in the presence of antimycin A was subtracted from all respiratory rates to account for nonmitochondrial respiration.

Acetyl-CoA Measurement

Acetyl-CoA levels were measured using a fluorometric PicoProbe Acetyl CoA assay kit (Abcam, Waltham, MA) following manufacturers’ instructions. Liver tissues (∼25 mg) were homogenized in RIPA buffer and centrifuged at 13,000 x g for 10 minutes at 4°C to collect the supernatant as total protein lysate. Mitochondrial and cytosolic fractions were isolated as described previously. Total protein concentration was quantified in all the fractions and then deproteinized using perchloric acid to a final concentration of 1 M and incubated for 5 minutes. The samples were centrifuged at 13,000 x g for 2 minutes and the supernatant was neutralized to pH 7–8 using ice-cold 2 M KOH. The supernatant was collected by centrifuging at 13,000 x g for 15 minutes at 4°C. The assay was carried out per the manual instructions and the final concentration was adjusted based on the dilution factors and protein concentration.

LC Measurement

LC was measured using a colorimetric LC assay kit (Abcam, Waltham, MA) according to manufactures instructions. Frozen liver tissues (∼25 mg) were homogenized with 250 μL assay buffer and the supernatant were collected by centrifuging at 13,000 x g for 10 minutes. The samples were deproteinized using ice-cold perchloric acid to a final concentration of 1 M and centrifuged at 13,000 x g for 2 minutes at 4°C. The supernatant was neutralized to pH 7–8 using ice-cold 2 M KOH. The assay was performed in the supernatants collected by centrifuging the samples at 13,000 x g for 15 minutes at 4°C and the final LC concentration were calculated based on the dilution factors and protein concentration.

Body Composition Analysis

Body composition was assessed in conscious mice using EchoMRI (EchoMRI LLC). Percent fat and lean mass were calculated for individual mice by normalizing to their respective body weight.

Quantification and Statistical Data Analysis

The data are presented as mean ± standard error of the mean. Statistical analyses were performed using Prism software version 9.0.0 (GraphPad Software). Comparisons between 2 groups were analyzed using a Student t tests. One-way analysis of variance (Tukey multiple-comparisons test) was used to compare more than 2 groups. Differences were considered statistically significant if ∗P < .05; ∗∗P < .01 or ∗∗∗P < .001 or ∗∗∗∗P < .0001. Western blot and TLC band intensities were quantified using Image J 1.52q. The statistical methods of each experiment are indicated in the Figure legends. In vitro experiments were replicated at least 3 independent times, except for human MASH hepatocytes.