Mitochondrial NCLX couples glucagon- and PDE2A-Dependent signals to regulate hepatic lipid metabolism

Abstract

Mitochondrial calcium signaling, particularly its glucagon-mediated oscillatory dynamics, plays a pivotal role in regulating hepatic metabolism and is known to be disrupted in steatotic liver disease. We recently identified the mitochondrial Na⁺/Ca²⁺exchanger NCLX as a key mediator of glucagon-induced mitochondrial calcium oscillations, essential for proper gluconeogenic function. Here, using hepatocyte-specific NCLX knockout (cKO) mice, we demonstrate that NCLX is critical for intrahepatic lipolysis and fatty acid oxidation (FAO); its loss impairs glucagon-stimulated lipid droplet catabolism and blunts FAO. Mechanistically, we find that NCLX deficiency disrupts allosteric activation of lipolytic enzymes and increases CPT1 sensitivity to malonyl-CoA-mediated inhibition, resulting in defective lipolysis and FAO. We further show that glucagon regulates hepatic NCLX via cAMP/PKA-dependent phosphorylation at NCLX Ser258. Notably, PDE2A acts as a negative regulator of this pathway by degrading mitochondrial cAMP. Hepatic mitochondrial PDE2A abundance and cAMP-degrading activity are elevated in HFD, and in vivo BAY 60-7550 treatment suppresses mitochondrial cAMP degradation and augments PKA signaling in steatotic livers. Pharmacologic inhibition of PDE2A with BAY 60-7550 enhances NCLX phosphorylation, restores mitochondrial calcium efflux and oscillations, and stimulates FAO in an NCLX-dependent manner. Importantly, we uncover that cAMP/PKA-dependent phosphorylation of NCLX at Ser258 is suppressed in human steatotic livers, and that pharmacologic inhibition of PDE2A ameliorates hepatic FAO and steatosis in both dietary and genetic MASLD models. Collectively, our findings establish the glucagon–PDE2A–PKA–NCLX signaling axis as a key metabolic rheostat integrating mitochondrial calcium dynamics with lipid homeostasis, providing a promising therapeutic target for MASLD.

Graphical abstract

Highlights

• •NCLX is required for glucagon-induced hepatic lipolysis and fatty acid oxidation.
• •NCLX S258 phosphorylation falls in human steatosis as hepatic PDE2A rises in MASLD.
• •PDE2A inhibition restores NCLX phosphorylation and mitochondrial Ca²⁺efflux.
• •NCLX activation restores mitochondrial β-oxidation in steatotic hepatocytes.
• •Targeting the PDE2A–NCLX axis mitigates steatosis in dietary and genetic MASLD.

1. Introduction

Calcium signaling is central to hepatic physiology, governing diverse processes through spatially and temporally regulated signaling modalities, from hepatocyte proliferation to bile secretion [[1], [2], [3]]. In hepatocytes, mitochondrial Ca²⁺signaling acts as a localized second messenger that integrates hormonal cues and cellular energy status to modulate enzymatic activities and signaling nodes involved in gluconeogenesis, β-oxidation, and lipid synthesis [[3], [4], [5], [6]]. However, the contribution of mitochondrial Ca²⁺to metabolic fuel preference and oxidation remains poorly understood, making it critical to define how this signaling directs hepatocytes toward fatty acid or carbohydrate utilization under the influence of hormones such as insulin and glucagon.

In the liver, glucagon stimulation activates canonical pathways of cyclic adenosine monophosphate (cAMP) signaling and downstream PKA activation [7]. However, it has also been shown to induce Ca²⁺signaling via inositol 1,4,5-trisphosphate receptor (IP3R)-mediated release of endoplasmic reticulum (ER) calcium, operating in parallel with the cAMP waves [[8], [9], [10]]. These cytosolic Ca²⁺signals are then transmitted to mitochondria, where they are sensed and modulated in a tightly regulated fashion [3,6,[9], [10], [11]].

The initial mitochondrial Ca²⁺surge is driven by influx through the mitochondrial calcium uniporter (MCU), followed within seconds by a rapid decline due to Ca²⁺extrusion [[12], [13], [14], [15]]. Although the identity of the mitochondrial Ca²⁺efflux mechanism in hepatocytes has long been debated, we recently identified the Na⁺/Ca²⁺exchanger NCLX as the key mediator of this process in a sodium-dependent manner [16]. Our findings further revealed that NCLX is essential for glucagon-induced, transcription-independent gluconeogenesis both in vivo and in vitro. In hepatocyte-specific NCLX knockout (KO) models, deletion of NCLX led to obliteration of mitochondrial Ca²⁺oscillations and to disruption of cytosolic-to-mitochondrial Ca²⁺transmission, thereby impairing both gluconeogenic output and overall metabolic activity [16].

Mitochondrial Ca²⁺has also been shown to stimulate soluble adenylate cyclase, enhancing the mitochondrial cAMP pool and intramitochondrial PKA signaling [[17], [18], [19], [20]]. Several mitochondrial proteins, including NCLX itself, are known PKA targets [[21], [22], [23], [24], [25], [26], [27]]. Functioning as the efflux arm of mitochondrial Ca²⁺signaling, NCLX plays a critical protective role by safeguarding against Ca²⁺overload and ensuing cell death. NCLX deletion has been shown to cause sustained elevation of matrix Ca²⁺levels following activation, which become cytotoxic upon further stimulation [[26], [27], [28]]. Recently, NCLX activity in neurons was shown to be negatively regulated by PDE2A, a mitochondrial phosphodiesterase that reduces PKA-mediated phosphorylation of NCLX, thereby decreasing its activity [25].

While mitochondrial Ca²⁺oscillations, and hepatic NCLX activity in particular, are critical for maintaining glucagon-induced oxidative phosphorylation and efficient electron transport chain (ETC) function during high-energy demanding gluconeogenesis [3,6,9], the broader role of hepatic mitochondrial Ca²⁺in the regulation of nutrient oxidation and fuel preference transition between the fed and fasted states remains poorly defined. The requirement of hepatic mitochondrial function to oxidize fat, generate ketones, and drive hepatic glucose production during fasting— while conversely promoting lipid storage and synthesis in the fed state— reflects a tightly regulated and complex metabolic balance [7,29]. Free fatty acids (FFAs), derived from intrahepatic lipid droplets and those released to circulation from adipose tissue, are predominantly oxidized during fasting and re-esterified or newly synthesized via de novo lipogenesis (DNL) under nutrient-excess conditions. FFAs contribute at least one-third of the energy required for gluconeogenesis [30], and the dynamic partitioning between β-oxidation and FFA esterification, particularly in response to hormonal cues, is a key determinant of metabolic health.

This balance is particularly relevant in the context of obesity and insulin resistance, where dysregulated-FFA metabolism contributes to the development of metabolic dysfunction-associated steatotic liver disease (MASLD), formerly termed non-alcoholic fatty liver disease (NAFLD) [31,32]. Mitochondrial Ca²⁺overload has been implicated in mitochondrial dysfunction in both alcoholic liver disease (ALD) and MASLD, as well as in the progression to metabolic dysfunction-associated steatohepatitis (MASH) [[33], [34], [35], [36]]. Notably, while experimental modulation of MCU activity has been shown to alleviate features of MASLD, the molecular mechanism linking mitochondrial Ca²⁺ signaling to lipid metabolism in hepatocytes —and its contribution to fatty liver pathologies—remains poorly understood. A significant barrier in investigating and targeting mitochondrial Ca²⁺ signaling pathways in hepatic lipid metabolism has been the unresolved and controversial nature of hepatic mitochondrial Ca²⁺ efflux, coupled with the absence of molecular identification of the responsible efflux machinery in hepatocytes. Recent findings have identified NCLX as the principal mediator of Ca²⁺ efflux mechanism [16]. This challenge is further compounded by the complexity of merging glucagon-stimulated downstream signaling, which integrates both allosteric and transcriptional regulatory mechanisms [7].

In this study, we used a conditional, hepatocyte-specific NCLX knockout (cKO) mouse model to dissect the role of NCLX in fatty acid metabolism. We demonstrate that NCLX activation is essential for shifting FFA fate toward β-oxidation over DNL.

Mechanistically, NCLX reduces malonyl-CoA-mediated inhibition of carnitine palmitoyltransferase 1 (CPT1), the rate-limiting enzyme for mitochondrial FFA import. Conversely, in the absence of NCLX, CPT1 becomes more susceptible and sensitive to malonyl-CoA, favoring lipid accumulation. We further demonstrate that glucagon promotes mitochondrial calcium oscillations via NCLX phosphorylation and activation, and that pharmacological inhibition of PDE2A enhances this effect, stimulating mitochondrial Ca²⁺ efflux and oscillatory signaling in an NCLX-dependent manner. Interestingly, liver specimens from MASLD patients with pronounced steatosis showed a marked reduction in phosphorylated NCLX compared to non-steatotic control individuals. Treatment of primary hepatocytes and obese mouse models with a PDE2A inhibitor restored fatty acid utilization and resolved hepatic steatosis. Together, these findings establish hepatic NCLX as a critical regulator of fatty acid oxidation and a potential therapeutic target in MASLD.

2. Methods

2.1. Patients

2.1.1. Cohort criteria

Patients with suspected hepatic steatosis and fibrosis were recruited between 2019 and 2021 at the Liver Unit, Hadassah Medical Center, Israel. All patients were females over 18 years of age. Diagnosis and underlying etiology were confirmed by liver biopsy, as previously described [37]. Exclusion criteria included steroid administration; recent or prolonged use (≥1 month or any use within the prior 2 months) of warfarin, metformin, thiazolidinediones, or insulin; and daily alcohol intake exceeding 10 g/day. Patients with alternative liver disease etiologies—such as viral hepatitis, autoimmune liver disease, hemochromatosis, Wilson's disease, alpha-1 antitrypsin deficiency, alcohol-related liver disease, or confirmed drug-induced hepatotoxicity were also excluded.

The study was approved by the institutional ethics committee, and written informed consent was obtained from all participants.

2.1.2. Patient and public involvement

Patients or the public were not involved in the design, conduct, reporting, or dissemination of this research.

2.1.3. Human liver biopsies

Ultrasound-guided percutaneous liver biopsies were performed using an 18 G, 15 mm disposable Tru-Cut needle, as previously described [37]. Biopsy samples from patients with varying steatosis grades (see Supplemental Tables 1 and 2) were snap-frozen in liquid nitrogen. Protein extracts were prepared in homogenization buffer (50 mmol/L Tris-HCl, pH 7.6, 0.25% Triton X-100, 0.15 M NaCl, 10 mM CaCl₂) supplemented with protease inhibitor cocktail (Roche Diagnostics) and phosphatase inhibitors (Roche Diagnostics).

2.2. Animal models

2.2.1. Generation of conditional liver-specific NCLX knockout mice

Conditional liver-specific NCLX knockout mice were generated using NCLX-floxed mice carrying loxP sites flanking exons 5–7 of the Slc8b1 gene (chr12:113,298,759–113,359,493), provided by the Elrod lab [38]. Floxed mice were injected via the tail vein with an AAV8-Cre under the control of a hepatocyte-specific thyroxine-binding globulin (TBG) promoter (1.3 × 10¹¹ GC/mouse; Addgene #107787-AAV8). Control floxed littermates received AAV8-TBG-Null (Addgene #105536-AAV8). Unless otherwise specified, adult male NCLX^fl/fl^ mice were used for experiments five weeks post-injection, at 10–12 weeks of age, and were age-matched with littermate controls for all in vivo studies and primary hepatocyte experiments.

2.2.2. Global NCLX knockout mice

Wild-type (WT) C57BL/6NJ mice and NCLX-null mice (C57BL/6NJ-Slc8b1^em1(IMPC)J/J) were obtained from The Jackson Laboratory and bred in-house. Due to global loss of NCLX from birth, these mice were used exclusively for specified and selected in vitro validation and mechanistic experiments. Global NCLX-null and WT C57BL/6NJ mice used for in vitro validation experiments were male, 10–14 weeks of age at the time of tissue collection. Genotyping was performed on tail clips or ear tissue collected at weaning, following Jackson Laboratory protocols using real-time PCR. Primers used were as follows:

• Forward: 5′-GGCTCCTGTCTTCCTCTGTG-3′

• Reverse: 5′-GTGTCCATGGGCTTTTGTG-3′

2.2.3. MASLD models

To model metabolic dysfunction-associated steatotic liver disease (MASLD), we used a genetic model and diet-induced models. For the genetic model, male ob/ob mice (B6. Cg-Lep^ob/J, Jackson Laboratory) aged 10–12 weeks at the start of treatment, were treated intraperitoneally with the PDE2A inhibitor BAY 60-7550 (1 mg/kg/day; Sigma-Aldrich, #SML2311) once daily for 3 weeks [25,39,40]. For the diet-induced models, male C57BL/6 J mice were assigned to one of two dietary regimens:(i) classic high-fat diet (HFD), in which mice were fed a HFD containing 45% kcal from fat ad libitum (#D12451, Research Diets) starting after weaning (3–4 weeks of age) for 12 weeks, with age-matched control mice fed standard chow; or (ii) CDAA-HFD, in which mice (6 weeks old at baseline) were fed a choline-deficient, l-amino acid-defined, high-fat diet (CDAA-HFD; 60% of calories from fat, 0.1% methionine by weight) for 4 weeks, as previously described, with littermate controls maintained on standard chow [41,42]. Throughout the text and figure legends this diet is referred to exclusively as ‘CDAA-HFD’ to distinguish it from conventional high-fat diets. For all models, livers were harvested and either digested for in vitro functional assays or formalin-fixed for histological analyses.

2.2.4. Experimental procedures and ethics

All animal procedures were conducted in accordance with institutional and national ethical guidelines. Protocols were approved by the Institutional Animal Care and Use Committees (IACUC) of Ben-Gurion University and the University of California, Los Angeles (UCLA). All mice were housed at 22 °C under a 12:12 h light–dark cycle. Except where otherwise indicated (e.g., MASLD model), mice were maintained on standard chow.

For selected experiments, mice were fasted with free access to water. Unless otherwise specified, “fasted” refers to animals subjected to a 30-hour fast, initiated at the onset of the light phase and terminated the following day, and compared to ad libitum-fed controls. Experimental groups were randomized and blinded using ear-tagging and numerical coding, which were decoded after experimental endpoints.

2.3. Hepatocyte isolation

Primary hepatocytes were isolated from adult male mice using a two-step collagenase perfusion method, as previously described [9,16,43,44]. Briefly, the liver was perfused via the abdominal inferior vena cava, while the hepatic portal vein was simultaneously cut to allow outflow. Initially, the liver was flushed with calcium- and magnesium-free Hanks' balanced salt solution (HBSS) for 6 min to remove blood. This was followed by perfusion with liver digestion medium (Thermo Fisher Scientific, #17703–034) containing collagenase for 7 min. Throughout the procedure, a consistent perfusion rate of 5 ml/min and a temperature of 37 °C were maintained.

Following perfusion, the gallbladder and diaphragm remnants were removed, and the liver was excised from the body cavity and transferred to a sterile Petri dish containing plating medium. The tissue was gently dispersed and passed through a 70 μm cell strainer. The resulting cell suspension was subjected to low-speed centrifugation at 50×g for 5 min at 4 °C to wash the cells. Hepatocytes were then resuspended in Percoll solution (1:1 ratio; Sigma-Aldrich, #P1644), centrifuged for 10 min, and the resulting pellet collected. Cell viability and yield were assessed via trypan blue exclusion, and only preparations with >70% viability were used for downstream experiments.

Hepatocytes were resuspended and plated in “plating medium” consisting of Dulbecco's Modified Eagle Medium (DMEM; Thermo #11965092) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich #F2442), 100 U/ml penicillin and 100 μg/ml streptomycin (Biological Industries, #03-031-1 B), 2 mM sodium pyruvate (Sigma-Aldrich, #113-24-6), 1 μM dexamethasone (Santa Cruz, #SC-29059), and 100 nM insulin (Actrapid, Novo Nordisk).

For imaging experiments, hepatocytes were seeded onto collagen-precoated coverslips at a density of 1 million cells per 20 cm². For intact-cell Seahorse respirometry assays, 8,000 hepatocytes were plated per well. For biochemical assays, 1 million hepatocytes were plated per well in six-well plates. After 2 h of incubation, the medium was replaced to remove unattached cells with a “maintenance medium” consisting of DMEM supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin, 2 mM sodium pyruvate, 100 nM dexamethasone, 1 nM insulin, and 0.2% fatty acid–free bovine serum albumin (Fraction V BSA, VWR #E588). Cells were maintained for 24–36 h prior to experimental use.

Glucagon was used at 20 nM for metabolic assays (lipid droplet remodeling, glycerol release, FAO/Seahorse) and at 20–100 nM for acute signaling experiments (mitochondrial Ca²⁺ imaging, NCLX phosphorylation, permeabilized FAO); the exact concentration employed in each case is detailed in the corresponding panel and relevant experimental methods.

2.4. Cell lines

Clonal HepG2 cells were cultured in low-glucose (5 mM) DMEM (Sigma-Aldrich #MFCD00217342) supplemented with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin, and 5 mM HEPES buffer. For NCLX phosphorylation studies, cells were treated with glucagon (100 nM, Sigma-Aldrich #16941-32-5), BAY 60-7550 (5 μM, Santa Cruz #sc-396772 A), or co-treated with BAY 60-7550 and the PKA inhibitor H-89 (10 μM, Cayman Chemical #10010556) for 1 h.

2.5. Hybrid mouse diversity panel (HMDP) analyses

Snap-frozen liver tissues from HMDP mice maintained on regular chow (healthy) or a high-fat/high-sucrose (HF/HS; Research Diets D12266B) diet (steatosis) [45] were used for RNA isolation and global gene expression analysis using Affymetrix HT_MG430A PM microarrays. NCLX (Slc8b1) mRNA levels were quantified using probe 1,417,811_at. Filtering criteria for the microarray data were applied as previously described [46]. The bicor (biweight midcorrelation) function from the WGCNA R package was used to assess hepatic NCLX transcript-trait correlations in the Western diet group and compute nominal p-values. Serum and liver triglyceride levels were measured as previously described [47].

2.6. Immunoblotting

For immunoblotting, whole-cell lysates and tissue homogenates were prepared as follows: cells were washed with ice-cold PBS and then lysed on ice for 30 min in RIPA buffer supplemented with protease inhibitor (1:50 dilution, Sigma-Aldrich, #P2714) and PhosSTOP phosphatase inhibitor (1:10 dilution, Roche #04906845001). For tissue samples, mechanical homogenization was performed with 10–15 strokes using a Teflon-glass homogenizer in the lysis buffer. All lysates were centrifuged at 1,000×g for 10 min at 4 °C, and the supernatant was collected for protein analysis.

Protein concentrations were determined using the Bradford assay (Bio-Rad). Twenty micrograms of protein per sample were mixed with 4 × LDS sample buffer and subjected to SDS-PAGE. Proteins were transferred to either nitrocellulose or PVDF membranes using a wet transfer system (Bio-Rad). Membranes were blocked with 5% BSA for 1 h and incubated overnight with the following primary antibodies and used according to the manufacturer's instructions.

• •Phospho-(Ser/Thr) PKA Substrate Antibody (1:1000, Cell Signaling Technology, #9621 S)
• •CPT1A (1:1000, Abcam #ab128568)
• •ACSL1 (1:1000, Cell Signaling Technology #4047 S)
• •Total OXPHOS Rodent WB Antibody Cocktail (1:500, Abcam #ab110413)
• •PDE2A (1:1000, Santa Cruz #sc-271,394 and Proteintech #53306-1-AP)
• •VDAC1 (1:500, Santa Cruz #sc-390,996)
• •HSL (1:1000, Cell Signaling #4107)
• •Phospho-HSL Ser660 (1:1000, Cell Signaling #4126)
• •ATGL 30A4 (1:1000, Cell Signaling #2439)
• •Phospho-ATGL Ser406 (1:1000, Abcam #ab135093)
• •Vinculin (1:1000, Sigma-Aldrich #V9131 and #CP74-100)
• •SDHA (1:1000, Thermo Fisher #459200)
• •NCLX (1:1000, Santa Cruz #sc-161,921 and #sc-161,922)
• •Rabbit polyclonal anti-pSer258 NCLX antibody (custom-made by GenScript, raised against the CTWIYQRQRRG{pSer}LFCP).

After primary antibody incubation, membranes were washed in Tris-buffered saline with 0.1% Tween-20 (TBS-T) and incubated for 1 h with appropriate secondary antibodies. For chemiluminescent detection, membranes were incubated with HRP-conjugated secondary antibodies: anti-rabbit IgG (1:10,000, Santa Cruz #sc-2357), anti-mouse IgG (1:5,000, Santa Cruz #sc-2005), or anti-goat IgG (1:5,000, Santa Cruz #sc-2020).

For fluorescent detection, membranes were incubated with Alexa Fluor-conjugated secondary antibodies (all 1:10,000; Thermo Fisher): donkey anti-mouse Alexa Fluor 488 (#A-21202), donkey anti-mouse Alexa Fluor 647 (#A-31571), donkey anti-rabbit Alexa Fluor 488 (#A-21206), donkey anti-rabbit Alexa Fluor 647 (#A-31573), rabbit anti-mouse Alexa Fluor 647 (#A-21239). Following secondary antibody incubation, membranes were washed in TBS-T. Chemiluminescent blots were developed using the EZ-ECL Chemiluminescence Detection Kit (Sartorius #20-500-120) and imaged on the Fusion SOLO X system (Vilber). Fluorescent blots were imaged directly on the ChemiDoc Molecular Imager (Bio-Rad).

2.7. Live fluorescence imaging

Kinetic live-cell fluorescence imaging was conducted to monitor Ca²⁺ transients using an IX73 inverted microscope (Olympus), equipped with a pE-4000 LED light source and a Retiga 600 CCD camera (QImaging/Teledyne). All images were acquired using a × 20/0.5 Zeiss Epiplan Neofluar objective and captured via Olympus cellSens Dimension software.

For mitochondrial Ca²⁺ measurements ([Ca²⁺]_m), cells were loaded with 1 μM of the Ca²⁺-sensitive dye Rhod2-AM (50,024, Biotium, Fremont, CA), which preferentially localizes to mitochondria. Loading was performed for 30 min at 37 °C using a modified Krebs-Ringer solution containing (in mM): 126 NaCl, 5.4 KCl, 0.8 MgCl₂, 20 HEPES, 1.8 CaCl₂, 15 glucose, 10 lactate, and 1 pyruvate, with pH adjusted to 7.4 using NaOH and supplemented with 0.1% BSA. Following dye loading, cells were washed three times with dye-free Krebs-Ringer solution and incubated for an additional 30 min to allow de-esterification of residual dye. Rhod2-AM was excited at 552 nm, and emission was collected using a 570 nm long-pass filter, as previously described [16,28].

At the start of each experiment, cells were perfused with Ca²⁺-containing Krebs-Ringer solution supplemented with 0.1% fatty-acid-free BSA (E588, VWR). To trigger ionic responses, cells were perfused with supplemented Krebs-Ringer solution containing either ATP (5 μM; 0220, Amresco) or glucagon (100 nM; 16,941-32-5, Sigma-Aldrich). In selected experiments, cells were pretreated with BAY 60-7550 (5 μM; Santa Cruz, Cat #sc-396772 A) for 1 h.

Traces of Ca²⁺ responses were analyzed and plotted using KaleidaGraph software. Oscillatory wave frequencies were quantified using GraphPad Prism 10. The rate of ion transport was calculated from each trace as the linear fit of the change in fluorescence intensity over time (ΔF/dt) for both Ca²⁺ influx and efflux. Rates from n experiments (as indicated in the figure legends) were averaged and displayed as bar graphs.

2.8. Fluorescent lipid imaging

To evaluate lipid content, primary hepatocytes subjected to different treatments were incubated with 1 μM BODIPY493/503 (BD493; D3922, Thermo) for 1 h. Following incubation, cells were washed three times with PBS. Fluorescence intensity was measured using a Tecan Spark 10M multimode plate reader. BD493 was excited at 488 nm, and emission was collected at 504 nm.

To assess mitochondrial mass 200 nM MitoTracker™ Deep Red FM (MTDR; M22426, Thermo) was used similar to above methodology. MTDR was excited at 633 nm, and emission was collected at 665 nm.

2.9. Histology and immunohistochemistry

Liver tissues fixed in formalin and embedded in paraffin were sectioned at 4 μm and stained with hematoxylin and eosin (H&E) for histological evaluation. Frozen liver tissues were cryosectioned and stained with Oil Red O to assess hepatic lipid accumulation, followed by hematoxylin counterstaining, as previously described [48].

Perilipin 3 (TIP47) expression in liver biopsies was evaluated using 3,3′-Diaminobenzidine (DAB)-based immunohistochemistry. Sections were incubated with a TIP47 antibody (Thermo) at a 1:500 dilution, with hematoxylin used for counterstaining, as previously described [49,50].

H&E and DAB-stained slides were digitized using an Aperio AT high-throughput scanner with a 20 × air objective. Oil Red O-stained sections were visualized using an Olympus IX83 microscope equipped with an automated stage and a 60 × oil-immersion objective. Lipid content was quantified using QuPath software [51].

2.10. Free glycerol assessment

Free glycerol content in liver tissue was measured using a Free Glycerol Reagent (F6428, Sigma-Aldrich) according to the manufacturer's protocol. For cellular experiments, primary mouse hepatocytes were seeded at a density of 1 × 10⁶ cells per well in six-well plates and incubated in 1 ml glucose-free DMEM lacking phenol red, supplemented with 20 mM sodium lactate, 2 mM sodium pyruvate, 4 mM glutamine, 100 nM dexamethasone, and 0.1% fatty-acid-free BSA. Cells were treated for 8 h under the following conditions: glucagon (20 nM), glucagon and insulin (20 nM and 100 nM, respectively), or vehicle.

2.11. Respirometry assays

2.11.1. Respirometry in intact hepatocytes

Mitochondrial respiration in primary mouse hepatocytes was assessed using the Seahorse XFe96 Analyzer (Agilent Technologies, Santa Clara, CA), following previously established protocols [52]. In brief, primary hepatocytes were seeded at 8,000 cells/well in collagen-coated XFe96 plates and cultured overnight in plating medium. The following day, cells were switched to glucose-free media and pretreated with glucagon (20 nM) ± BAY 60-7550 (5 μM) for 1 h. Media was then replaced with Seahorse XF DMEM (pH 7.4). Palmitate-BSA complexes were prepared as previously described [53,54]. In brief, palmitic acid was initially dissolved in DMSO to a final concentration of 0.4 M and then dissolved at 56 °C into Seahorse assay medium containing 5% fatty-acid-free BSA (Roche) to create a 10 × (4 mM) stock solution.

Mitochondrial stress tests were performed via sequential injections of 3 μM oligomycin, 3 μM carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), and 2 μM rotenone/antimycin A. Oxygen consumption rate (OCR) was normalized to cell count, as determined by DAPI staining (1 μg/ml; D1306, Thermo) following 4% paraformaldehyde fixation (J61899, Thermo). Fluorescence was measured on a Tecan Spark 10 M plate reader at 358/461 nm (ex/em).

2.11.2. Respirometry in permeabilized hepatocytes

To assess fatty acid oxidation capacity, Seahorse XF respirometry was performed in permeabilized hepatocytes following previously published protocols [52,55,56]. Briefly, 10,000 hepatocytes were plated in collagen-coated XFe96 plates and cultured as described above. On the experiment day, cells were treated with glucagon (100 nM) for 6 h. Following treatment, cells were washed twice with unsupplemented mitochondrial assay solution (MAS) buffer (220 mM mannitol, 70 mM sucrose, 10 mM KH₂PO₄, 5 mM MgCl₂, 1 mM EGTA, 2 mM HEPES; pH adjusted to 7.2 with KOH), then incubated for 10 min at 37 °C in MAS buffer supplemented with substrates for mitochondrial respiration. These included either 40 μM palmitoyl-CoA with 0.5 mM carnitine, 40 μM palmitoyl-carnitine, or 5 mM pyruvate and 5 mM malate. All conditions also contained 5 mM ADP and 8 nM recombinant perfringolysin O (rPFO; XF PMP reagent, 102,504-100, Agilent) to permeabilize the plasma membrane. Where indicated, etomoxir (10 μM) was added to inhibit CPT1-dependent fatty acid import. Sequential injections included: 6 μM oligomycin, 3 μM FCCP, and 4 μM rotenone/antimycin A. Fatty acid oxidation capacity was defined as maximal OCR fueled by 40 μM palmitoyl-CoA and 5 mM carnitine in the presence of ADP. Final OCR values were corrected by subtracting non-mitochondrial respiration measured after rotenone/antimycin A treatment and normalized to cell count, as described above.

2.12. CPT1 activity assays in isolated mitochondria

Fresh livers were washed with ice-cold PBS and homogenized in mitochondria isolation buffer (MIB: 200 mM mannitol, 70 mM sucrose, 1 mM EGTA, 10 mM HEPES, pH 7.4), supplemented with protease and phosphatase inhibitors. Tissue was homogenized with 20 strokes of a Teflon-glass homogenizer. Nuclei and debris were removed by centrifugation at 1,000 g for 5 min. Crude mitochondria were isolated, washed twice and resuspended in MIB buffer for CPT1 activity measurement. CPT1 activity was assessed using a colorimetric assay based on established protocols [52,[57], [58], [59]]. Briefly, 20 μg of mitochondria-enriched heavy membrane (HM) fraction was added to 200 μL of reaction buffer (2 mM DTNB, 116 mM Tris-HCl, pH 8.0, 2.5 mM EDTA, 0.2% Triton X-100) in 96-well plates and incubated at room temperature for 20 min to quench background thiol signals. The enzymatic reaction was initiated by adding 100 μM palmitoyl-CoA and 1 mM carnitine, CPT1 substrate and cofactor, respectively. Free CoA-SH release from palmitoyl-CoA was spectrophotometrically determined as a readout of CPT1 activity and quantified by measuring absorbance at 412 nm using a Tecan Spark 10 M reader in kinetic mode for 1 h total [52].

2.13. Electron microscopy sample preparation and imaging

Liver samples were fixed in 2.5% glutaraldehyde, washed in 0.1 M cacodylate buffer, and post-fixed in 1% osmium tetroxide for 1 h. Samples were again washed in cacodylate buffer and incubated in 2% uranyl acetate in double-distilled water for 1 h. Dehydration was performed through a graded ethanol series (30%, 50%, 70%, 90%, 100%), followed by treatment with acetone and acetone/epon and embedding in EMbed 812 resin (EMS #14120). Polymerization was carried out at 60 °C for 48 h. Ultrathin sections were prepared with a Leica UC6 ultramicrotome, collected on copper grids, and post-stained with 1% uranyl acetate. Imaging was performed using a Tecnai T12 G2 TWIN TEM (FEI) at 120 kV or a Talos F200C (Thermo Fisher Scientific) at 200 kV. Images were captured using Gatan 794 MultiScan CCD or Ceta 16 M CMOS cameras.

2.14. PDE activity assay and cAMP measurements

Mitochondrial PDE activity and intraluminal cAMP were quantified using the Abcam competitive fluorometric cAMP ELISA kit (ab138880) according to the manufacturer's instructions, adapting the general approach described previously for mitochondrial PDE measurements [21]. In this competitive format, samples and a fixed amount of labeled cAMP tracer compete for a limited number of anti-cAMP antibody binding sites, generating an inverse, sigmoidal standard curve that was fit using four-parameter logistic regression. Mitochondria were isolated from livers of ob/ob mice pretreated with vehicle or BAY 60-7550, as well as from chow-fed and 12-week HFD-fed mice in separate experiments. Mitochondrial protein concentration was determined, and 50 μg of mitochondrial protein per well was subjected to the assay. All samples were verified to fall within the assay's dynamic range (IC₁₀-IC₉₀), and cAMP concentrations were interpolated from the fitted curve. PDE activity was calculated from the rate of cAMP degradation and initially expressed per amount of mitochondrial protein. For presentation, PDE activity in each experimental group was normalized to the mean activity of the corresponding control group (chow-fed or vehicle-treated), which was set to 100%, and results are reported as percent of control (cAMP-hydrolyzing activity relative to control). Each biological condition was measured using independent animals (biological replicates; n ≥ 3 per group), with each animal providing a single mitochondrial preparation.

2.15. Quantification and statistical analysis

Statistical analyses were performed using GraphPad Prism 10. BioRender was used to generate schematic illustrations, and Adobe Illustrator was utilized for the graphic design of the manuscript figures. For two-group comparisons, two-tailed unpaired Student's t-tests were used. For comparisons involving more than two groups, one-way or two-way ANOVA followed by Tukey's post hoc test was applied. All bar graphs represent means ± SEM from n biological replicates as indicated. Statistical significance was defined as p < 0.05. Significance symbols used: ns (not significant), p > 0.05, (∗) p ≤ 0.05, (∗∗) p ≤ 0.01, (∗∗∗) p ≤ 0.001, (∗∗∗∗) p ≤ 0.0001.

3. Results

3.1. Hepatic NCLX expression is negatively associated with steatosis and insulin resistance

During fasting, glucagon stimulation drives a metabolic shift from DNL toward β-oxidation and triggers tightly coordinated oscillations in cytosolic and mitochondrial calcium. In hormone-resistant states, such as those associated with obesity and type 2 diabetes, this finely-tuned balance is disrupted, resulting in redox stress, persistently elevated mitochondrial Ca²⁺ levels, and aberrant Ca²⁺ oscillatory patterns [34,[60], [61], [62]].

Mitochondrial calcium homeostasis is maintained by two major mechanisms: Ca²⁺ influx through the mitochondrial calcium uniporter (MCU) and efflux via the Na⁺/Ca²⁺ exchanger NCLX (Fig. 1A). NCLX-mediated efflux is the rate-limiting step due to its slower kinetics and dependence on mitochondrial membrane potential [26,27]. This regulatory bottleneck becomes particularly relevant under conditions of increased metabolic demand or stress [28]. We recently demonstrated that hepatic deletion of NCLX resulted in selective loss of mitochondrial, but not cytosolic, Ca²⁺ oscillations, emphasizing NCLX's central role in regulating mitochondrial Ca²⁺ dynamics in the liver. In vivo, liver-specific NCLX knockout (KO) mice failed to maintain normoglycemia during fasting and exhibited impaired glucagon-stimulated hepatic glucose production, underscoring NCLX's critical role in sustaining gluconeogenesis. However, the precise mechanisms by which mitochondrial Ca²⁺ handling governs hepatic gluconeogenesis are not yet fully understood. Furthermore, how NCLX activity is modulated in conditions of metabolic dysfunction and pervasive hormone resistance, as observed in obesity and MASLD, remains poorly characterized.

Figure 1.

NCLX pathway is negatively associated with insulin resistance and steatosis. (A) Schematic overview of mitochondrial calcium transport pathways. (B) Hepatic Slc8b1 mRNA content measured in 74 strains (x-axis) from the Hybrid Mouse Diversity Panel (HMDP), fed either chow (blue) or Western diet (red). C57BL/6 J is indicated by a dashed line. (C) Biweight midcorrelation values between hepatic NCLX transcript content with the metabolic trait measured in the Western diet-fed HMDP mice (n = 102 strains); nominal p-values are shown. (D) Strategy for hepatocyte-specific NCLX knockout generation. Top: Genomic map of Slc8b1 (NCLX) showing exons 5–7 flanked by loxP sites. Bottom: Mice were injected via the tail vein with AAV8-Cre under the control of the hepatocyte-specific thyroxine-binding globulin (TBG) promoter. AAV8-TBG-Null-injected littermates served as controls. (E) Western blot of NCLX expression in various tissues from mice injected with a hepatic viral Cre vector (TBG-CRE) or TBG-Null (CTRL). VDAC1 was used as the loading control. (F) Representative Oil Red O (ORO) staining of hepatic lipid droplets in CRE and CTRL livers from ad libitum-fed or 30 h fasted mice. Hematoxylin was used to counterstain the nuclei. Scale bar = 500 μm; inset scale = 50 μm. (G) Quantification of mean ORO staining density per cross-sectional area (n = 8–9 per group). (H) Percentage of ORO-positive cells per histological section (n = 8–9 per group). (I) Quantification of lipid droplet accumulation assessed by Bodipy fluorescence intensity (F·I.) in plated primary hepatocytes from CRE and CTRL mice following glucagon or vehicle treatment. Normalized to cell number using DAPI (glucagon: n = 10 CTRL, n = 12 CRE; vehicle: n = 10 each). (J) Quantification of intrahepatic lipolysis assessed by glycerol content in liver homogenates from CRE and CTRL mice under fed and fasted conditions (n = 3 each). (K) Quantification of intrahepatic lipolysis assessed by glycerol release assay in primary hepatocytes isolated from CRE and CTRL mice and stimulated with glucagon, co-stimulation with glucagon and insulin (For basal n = 4 CTRL and n = 3 CRE; For glucagon n = 4 each; For glucagon and insulin n = 4 each). (L) Representative Transmission electron microscopy (TEM) images of livers from ad libitum-fed and 30 h fasted CRE and CTRL mice. Lipid droplet borders are marked by red dashed lines. Scale bar 1 μm. (M) Quantification of average lipid droplet diameter from (L) (n = 4 per genotype). (N) Lipid droplet size distribution analysis from TEM images in (L). Data are presented as mean ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ns – non-significant. Two-tailed unpaired t-tests were used for two-group comparisons; one-way ANOVA with Tukey's post hoc test was used for three or more groups.

To address NCLX role and activity under metabolic stress, we first analyzed hepatic NCLX (Slc8b1) expression across the high fat/high sucrose (HF/HS) diet fed Hybrid Mouse Diversity Panel (HMDP) [45], a collection of ∼100 genetically diverse mouse strains exhibiting varying degrees of obesity, insulin resistance, and hepatic steatosis [46,[63], [64], [65]]. Interestingly, in 69 of the 74 strains with matched chow-fed controls (n = 2–4 per strain), HF/HS diet feeding elevated hepatic NCLX mRNA expression, with fold increases ranging from 1.01 to 2.9 (Fig. 1B). To determine whether NCLX upregulation represents a pathogenic disease-driving factor or a compensatory response to fatty liver progression, we assessed the associations between hepatic NCLX expression and metabolic phenotypes across all HF/HS diet fed strains. Notably, we found a significant negative correlation between NCLX expression and both the severity of hepatic steatosis and systemic insulin resistance (Fig. 1C). Specifically, NCLX expression was negatively correlated with liver triglyceride (TG) content (bicor = −0.284, p = 2.38E-03), HOMA-IR (bicor = −0.481, p = 4.68E-12), fasting insulin levels (bicor = −0.498, p = 6.34E-13), and gonadal fat mass (bicor = −0.294, p = 6.69E-06) (Fig. 1C). Correlations were determined using biweight midcorrelation analysis, as previously described [66]. Taken together, these findings suggest that hepatic NCLX upregulation may serve as a protective response mechanism aimed at mitigating steatosis and insulin resistance in the context of diet-induced fatty liver disease.

During fasting, glucagon promotes the mobilization of lipid stores by stimulating lipolysis of intrahepatic lipid droplets and, more substantially, by inducing adipose tissue lipolysis, in concert with other adrenergic stimuli. This results in the release of free fatty acids (FFAs) into the circulation, which are subsequently taken up by hepatocytes and oxidized through β-oxidation. The resulting fatty acid oxidation (FAO) supplies the ATP, reducing equivalents, and carbon skeletons required to sustain gluconeogenesis [7,29,67]. Building on our previous findings that NCLX is critical for glucagon-stimulated gluconeogenesis and mitochondrial Ca²⁺ handling, we hypothesized that NCLX links glucagon signaling and the energetic demands of fasting by facilitating intrahepatic lipid mobilization and free fatty acid oxidation.

First, to assess whether NCLX plays a role in regulating intrahepatic lipid metabolism, we generated hepatocyte-specific NCLX knockout (cKO) mice by delivering AAV8 vectors encoding Cre recombinase under the control of the hepatocyte-specific TBG promoter, into NCLX^fl/fl mice via tail vein injection (Fig. 1D,E). Western blotting confirmed a liver-specific deletion of NCLX, with its expression preserved in extrahepatic tissues, including skeletal muscle and adipose depots (Fig. 1E). To investigate the impact of hepatic NCLX deletion on lipid handling during fasting, we subjected both NCLX cKO and their control littermates to fasting conditions and liver sections from both genotypes were stained with Oil Red O to highlight lipid droplets (Fig. 1F). Notably, whereas control mice displayed minimal lipid droplet accumulation under both fasted and fed conditions, fasted NCLX cKO mice exhibited a marked increase in hepatic lipid compared to both controls and their own fed state (Fig. 1G,H). To validate this finding in vitro, primary hepatocytes were isolated from NCLX cKO and control mice, plated, and subsequently stimulated with either glucagon or vehicle. Lipid droplet accumulation was then assessed using a BODIPY staining. As expected, glucagon stimulation markedly reduced hepatic lipid droplet content in control hepatocytes (Fig. 1I). In contrast, glucagon failed to elicit a comparable reduction in NCLX-deficient hepatocytes, further supporting a defect in lipid mobilization (Fig. 1I). To determine whether this defect extended to triglyceride hydrolysis, we measured hepatic glycerol content, a byproduct of lipolysis and an indicator of lipid breakdown. In liver homogenates from control mice, fasting significantly increased glycerol levels; however, this response was blunted in NCLX cKO livers (Fig. 1J). Consistently, in vitro, glucagon treatment increased free glycerol levels in control hepatocytes compared to vehicle or insulin/glucagon co-treatment. However, hepatocytes derived from NCLX cKO mice failed to mount a comparable increase in glycerol production, remaining near baseline levels irrespective of glucagon stimulation (Fig. 1K).

In order to gain further insight into lipid droplet morphology and better assess their content, we then conducted high-resolution imaging using electron microscopy to examine the mitochondria and lipid droplets in liver preparations from NCLX cKO mice and their control littermates under fasted and fed conditions (Fig. 1L). Our analysis revealed that in fasted control mice, intrahepatic lipid droplets were disintegrated into minute lipid droplets, resulting in almost complete depletion of the total lipid content (Fig. 1M,N). In contrast, NCLX cKO mice exhibited an aggregation of overall lipid droplets, as well as larger individual lipid droplets, as evidenced by the distribution analysis of lipid droplets by size (Fig. 1M,N).

Importantly, analysis of white adipose tissue (WAT) from fasted liver-specific NCLX cKO mice revealed an intact lipolytic response, comparable to controls. Upon fasting, these mice exhibited elevated WAT free glycerol, increased plasma glycerol levels, enhanced lipid droplet turnover by BODIPY staining, and reduced adipose tissue weight (Supplementary Fig. 1). These findings indicated that peripheral lipolysis remains functional and that the accumulation of intrahepatic FFAs in NCLX-deficient livers stems from impaired hepatic utilization while maintaining an intact FFA mobilization from adipose tissue.

Collectively, our data identify NCLX as a central regulator of intrahepatic lipolysis, and its loss disrupts glucagon-mediated lipid droplet catabolism leading to hepatic lipid droplet accumulation.

3.2. NCLX is essential for glucagon-induced fatty acids oxidation

Building on our current findings, we establish that NCLX is essential not only for glucagon-mediated hepatic glucose production [16], but also for intrahepatic lipid breakdown in response to glucagon. However, the signaling pathways and discrete biochemical steps through which NCLX and mitochondrial Ca²⁺ handling may influence hepatic lipid metabolism remain incompletely understood.

Notably, glucagon stimulation is known to trigger phosphorylation of liver Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) in an IP3R-dependent manner [8]. Inhibition or deletion of CaMKII reduces expression of gluconeogenic proteins in murine livers and compromises hepatic glucose production [8,[67], [68], [69]]. Remarkably, CaMKII activation has been implicated in calcium-dependent regulation of key lipolytic enzymes, including hormone-sensitive lipase (HSL) and adipose triglyceride lipase (ATGL) [8,70].

To investigate the relationship between the observed phenotype of NCLX cKO mice and hepatic lipid metabolism, we first examined key lipolytic enzymes, ATGL and HSL. Total hepatic ATGL and HSL protein levels were comparable between cKO and their control littermates, with both proteins appropriately upregulated during fasting. However, immunoblot analysis revealed markedly reduced phosphorylation at ATGL Ser406 and HSL Ser660, which are sites known to regulate their enzymatic activity, in NCLX cKO livers compared to controls (Supplementary Fig. 2). This finding indicates that NCLX and mitochondrial calcium signaling are critical for the activating phosphorylation and thus proper function of these enzymes. Nonetheless, this finding alone does not fully explain the phenotype, since during fasting, the majority of hepatic FFAs are sourced from circulating free fatty acids derived from white adipose tissue, which remain unaltered in NCLX cKO mice (Supplementary Fig. 1). Therefore, we hypothesize that loss of NCLX signaling impairs not only intrahepatic lipid mobilization but also downstream mechanisms responsible for the effective utilization of FFAs by hepatocytes.

To address this, we conducted Seahorse respirometry on isolated primary hepatocytes from NCLX cKO mice and their control littermates. Cells were pre-treated with glucagon and subsequently stimulated with palmitate as a fuel source or with BSA as a control (Fig. 2A). Our results show that hepatocytes from NCLX cKO mice exhibited a substantial impairment in palmitate-induced respiration and maximal respiration with palmitate serving as a fuel (Fig. 2A–E), indicating defective β-oxidation. Importantly, mitochondrial respiratory integrity was overall maintained under BSA stimulation, and overall basal respiration was comparable (Fig. 2A,B). Thus, these bioenergetic findings point to an impairment in mitochondrial function and suggest a potential link between NCLX and hepatic metabolic regulation in response to glucagon.

Figure 2.

NCLX deletion impairs fatty acid oxidation by increasing CPT1 sensitivity to malonyl-CoA. (A–E) Oxygen consumption rate (OCR) in intact primary hepatocytes from CTRL and NCLX cKO (CRE) mice, pretreated with glucagon and stimulated with palmitate or BSA (vehicle control). (A) Representative real-time respirometry trace showing sequential injections of palmitate, oligomycin, FCCP, and rotenone/antimycin A. (B–E) Quantification of (B) basal respiration, (C) substrate-driven oxygen consumption (D) Maximal respiration capacity (FCCP-induced), and (E) relative increase in maximal respiration (n = 10–19 per group). (F–I) Fatty acid oxidation (FAO) assessed in permeabilized hepatocytes stimulated with palmitoyl-CoA and carnitine, with or without etomoxir. (F) Representative respirometry trace of permeabilized primary hepatocytes driven with palmitoyl-CoA and carnitine. (G–I) Quantification of (G) state III respiration (ADP-driven), (H) oligomycin-sensitive phosphorylating respiration, and (I) FCCP-induced maximal respiration. CTRL vehicle (n = 26), CTRL + etomoxir (n = 16), CRE vehicle (n = 24), CRE + etomoxir (n = 14). (J–M) Pyruvate-malate-driven respiration in permeabilized hepatocytes from CTRL and CRE mice in the presence of ADP. Cells were pretreated with vehicle or etomoxir. (J) Representative respirometry trace of permeabilized primary hepatocytes driven by pyruvate + malate. (K–M) Quantification of (K) state III respiration, (L) oligomycin-sensitive phosphorylating respiration, and (M) FCCP-induced maximal respiration. CTRL vehicle (n = 55), CTRL + etomoxir (n = 26), CRE vehicle (n = 47), CRE + etomoxir (n = 34). (N–O) Western blot analysis of OXPHOS complex subunits I–V (CI–CV) in liver lysates from fed and fasted CTRL and NCLX cKO mice (n = 3 per group). (P–Q) Western blot analysis of ACSL and CPT1 in liver lysates from fed (n = 4) and fasted (n = 3) CTRL and NCLX cKO mice. (R) CPT1 activity in the presence of increasing concentrations of malonyl-CoA (M-CoA), measured in mitochondria-enriched heavy membrane fractions freshly isolated from fed (left) and fasted (right panel) NCLX cKO and their CTRL mice (n = 3–8 per condition). All values are represented as mean ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ns – non-significant. Two-tailed unpaired t-test for comparisons between two groups was used and one-way ANOVA with Tukey's multiple comparison test was used for three or more groups.

The capacity of hepatocyte mitochondria to adapt to shifts in fuel availability vis-à-vis fuel preference, while sustaining fatty acid oxidation (FAO), is critical for metabolic flexibility. To investigate how NCLX loss affects nutrient oxidation and mitochondrial fuel preference, we performed respirometry using permeabilized adherent cell monolayers of freshly isolated hepatocytes from NCLX cKO and their control mice, supplying either palmitoyl-carnitine or pyruvate/malate, respectively (Fig. 2F,J). We initially focused on fatty acid oxidation by assessing the function of carnitine palmitoyltransferase-1 (CPT1), the rate-limiting enzyme for long-chain FAO. In this system, cofactors including ADP, palmitoyl-CoA, and carnitine were provided in excess, with etomoxir used as a control inhibitor of CPT1. FAO was quantified as the etomoxir-sensitive component of oxygen consumption rate (OCR). Specifically, this was defined as the difference between respiration induced by palmitoyl-CoA + carnitine + ADP and that induced by the same mixture with etomoxir (Fig. 2F,G). Remarkably, glucagon-stimulated control hepatocytes exhibited a significant increase in FAO followed by maximal respiration, whereas this response was markedly blunted in NCLX cKO hepatocytes (Fig. 2F–I).

FAO can be regulated at multiple levels, including fatty acid import, mitochondrial metabolism, as well as the capacity for oxidative phosphorylation. If the impairment in FAO in NCLX cKO hepatocytes were due to a general defect in OXPHOS, one would expect pyruvate/malate-fueled respiration to be similarly affected. However, in permeabilized hepatocytes, OCR values with pyruvate/malate were comparable between genotypes (Fig. 2J–M). These results suggest that oxidative phosphorylation is not the limiting factor in FAO, and that glucagon stimulation specifically enhances mitochondrial FAO capacity in an NCLX-dependent mechanism. Furthermore, immunoblotting for OXPHOS complexes, CPT1, and acyl-CoA synthetase long-chain family member 1 (ACSL1) under both fed and fasted conditions between genotypes (Fig. 2N–Q) indicates that the observed metabolic defects are not due to transcriptional or translational regulation. To further investigate the mechanism by which NCLX influences fatty acid oxidation, we employed a high-resolution functional assay to assess CPT1 activity and its sensitivity to malonyl-CoA, a known allosteric inhibitor that decreases CPT1's affinity for carnitine [52,71,72]. Using isolated liver mitochondria from NCLX cKO mice and their control littermates under both fed and fasted conditions, we observed no significant differences in CPT1 activity in response to increasing malonyl-CoA concentrations in the fed state (Fig. 2R). However, under fasting conditions, NCLX-deficient mitochondria showed heightened sensitivity to malonyl-CoA-mediated inhibition, consistent with the reduced FAO observed in the same experimental context (Fig. 2R). Importantly, CPT1 protein expression was not significantly altered in the livers of NCLX cKO mice, ruling out changes in protein abundance as a confounding factor for the altered CPT1 activity (Fig. 2P). Hepatic mitochondria adapt to fasting by relieving malonyl-CoA-mediated inhibition of CPT1, thereby increasing CPT1 activity and shifting substrate use toward mitochondrial FAO to meet energy demands [52]. This well-established adaptive response enables efficient switching to lipid fuels during fasting. Our data show that this adaptive regulation is blunted in NCLX-deficient hepatocytes, where CPT1 remains abnormally sensitive to malonyl-CoA, preventing full activation of FAO in response to glucagon. Consequently, loss of NCLX attenuates the glucagon-stimulated bioenergetic response and reduces hepatic FAO under fasting and lipid-preferential conditions. Together, these findings indicate that NCLX-dependent mitochondrial Ca²⁺ handling is essential for proper CPT1 regulation and for supporting glucagon-driven hepatic lipid oxidation during high energy demand states such as fasting.

3.3. Activation of NCLX by inhibition of PDE2A affects mitochondrial calcium oscillations and FFA oxidation

Hepatic hormones such as glucagon, when applied at physiological concentrations, elicit intracellular Ca²⁺ oscillations that are subsequently propagated into the mitochondrial matrix [3,6,9,10,16]. We recently demonstrated that NCLX plays a crucial role in regulating mitochondrial calcium oscillations [16]. In hepatocytes derived from NCLX cKO and control mice, we observed that, as expected, control hepatocytes responded to glucagon with low-frequency mitochondrial Ca²⁺ oscillations. Remarkably, however, NCLX cKO hepatocytes exhibited a complete cessation of glucagon-induced mitochondrial Ca²⁺ oscillations (Fig. 3A). Upon glucagon binding to its receptor on hepatocytes, a signaling cascade activates adenylyl cyclase and elevates intracellular cyclic AMP (cAMP). This rise in cAMP activates protein kinase A (PKA), which phosphorylates downstream targets involved in glycogen metabolism and gluconeogenesis. Whether this PKA-dependent signaling also contributes to non-transcriptional regulation of mitochondrial calcium handling remains unclear. We have previously shown that PKA-mediated phosphorylation of NCLX at serine 258 (Ser258) enhances mitochondrial Ca²⁺ efflux and promotes survival in a Parkinson's disease model of PINK1-deficient dopaminergic neurons [26,27]. Similarly, norepinephrine-stimulated NCLX activity in brown adipose tissue (BAT) is controlled by PKA activity [[25], [26], [27]]. To investigate whether this regulatory mechanism applies to hepatocytes, we used the human liver-derived HepG2 cell line to assess NCLX phosphorylation status following glucagon stimulation. This model was chosen because of the species-specific selectivity of our phospho-Ser258 NCLX antibody, which recognizes the human epitope. HepG2 cells were treated with either glucagon or vehicle, and lysates were analyzed by phospho-specific and total NCLX immunoblotting. Our results show that HepG2 cells treated with glucagon exhibited a significant increase in phospho-NCLX levels compared to vehicle (Fig. 3B,C); however, total NCLX protein remained unchanged (Fig. 3B,C).

Figure 3.

Activation of NCLX by PDE2A inhibition regulates mitochondrial calcium oscillations and fatty acid oxidation. (A) Mitochondrial Ca²⁺oscillations in primary hepatocytes isolated from TBG-Cre mice (CRE, red) and littermate controls (CTRL, black), triggered by glucagon (100 nM). Mitochondrial Ca²⁺levels ([Ca²⁺]_m) were monitored using Rhod2-AM (1 μM). (B) Western blot analysis of HepG2 cells treated with vehicle or glucagon and probed with pSer258-NCLX (pNCLX) antibody showing NCLX phosphorylation at its PKA-dependent S258 NCLX phosphorylation site. Total NCLX fraction was quantified using a non-phosphorylation-specific NCLX antibody. VDAC1 served as a loading control. (C) Densitometric quantification of pNCLX and total NCLX from groups shown in (B) (n = 4 per condition). (D)(Top) Expression levels of phosphodiesterase 2A (PDE2A) across various tissues. (Bottom) Schematic illustrating the mechanism of PDE2A activity and its pharmacological inhibitor BAY 60-7550 (BAY60). (E) Western blot analysis of HepG2 cells treated with vehicle, BAY60 (5 μM, 1 h), or BAY60 plus PKA inhibitor H-89 (10 μM, 1 h). Samples were probed with a phospho-specific antibody against PKA substrate motifs (phospho-PKA substrate). Vinculin served as a loading control. (F) Densitometric quantification of phospho-PKA substrate signal normalized to Vinculin (n = 8 per condition). (G) Western blot analysis of HepG2 cells pre-treated with vehicle or BAY60 (5 μM, 1 h), with or without H-89 (10 μM, 1 h). Samples were probed for pNCLX and total NCLX. VDAC1 served as a loading control. (H) Densitometric quantification of groups shown in G probed for pNCLX and total NCLX (n = 4 per condition). (I,J) Mitochondrial Ca²⁺oscillations triggered by physiological concentrations of glucagon in cultured WT primary hepatocytes in presence or absence of the PDE2A blocker BAY60 (5 μM). [Ca²⁺]m was monitored by Rhod2-AM. (I) Representative traces of glucagon-induced [Ca²⁺]_m oscillations in WT hepatocytes preloaded with Rhod2-AM, in the presence or absence of BAY60. (J) Quantification of [Ca²⁺]_m oscillation frequency in cells treated with vehicle (n = 183) or BAY60 (n = 134). (K–M) Mitochondrial Ca²⁺kinetics induced by ATP (5 μM) stimulus in cultured primary hepatocytes from NCLX KO mice and WT controls, pre-treated with or without BAY60 (5 μM). (K) Representative traces of [Ca²⁺]_m transients in WT and NCLX KO hepatocytes following stimulation with ATP, in the presence or absence of BAY60. (L) Quantification of [Ca²⁺]ₘ influx rates in WT and KO hepatocytes treated with vehicle or BAY60 (WT-vehicle: n = 55; WT-BAY60: n = 97; KO-vehicle: n = 72; KO-BAY60: n = 69). (M) Quantification of [Ca²⁺]ₘ efflux rates in WT and KO hepatocytes treated with vehicle or BAY60 (WT-vehicle: n = 55; WT-BAY60: n = 97; KO-vehicle: n = 72; KO-BAY60: n = 69). (N–S) Oxygen consumption rate (OCR) analysis of CRE and CTRL primary hepatocytes assessing fatty acid (FA)-driven respiration in the presence or absence of BAY60. (N) Representative respirometry traces of CRE and CTRL intact primary hepatocytes in the presence or absence of BAY60. Palmitate-induced oxygen consumption was measured in real-time. Palmitate, Oligomycin, FCCP, and rotenone/antimycin A were sequentially injected. (O–S) Quantification of (O) Basal OCR, (P) FA-driven OCR, (Q) fold increase in FA-stimulated respiration, (R) maximal OCR induced by FCCP, and (S) fold increase in maximal respiration (n = 23–36 per condition). All values are represented as mean ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001; ns – non-significant. Two-tailed unpaired t-test for comparisons between two groups was used and one-way ANOVA with Tukey's multiple comparison test was used for three or more groups.

Because cAMP signaling is dampened by phosphodiesterases (PDEs), which hydrolyze cAMP to AMP and thereby attenuate PKA activity, we considered PDEs as potential negative regulators of the cAMP-PKA-NCLX signaling axis. Notably, recent work in neurons identified mitochondrial PDE2A as an inhibitory mechanism that lowers mitochondrial matrix cAMP, reduces PKA-dependent NCLX phosphorylation, and attenuates NCLX activation [25]. We therefore hypothesized that PDE2A may similarly constrain glucagon-induced NCLX activation in hepatocytes, thereby modulating mitochondrial calcium efflux and oscillations. Notably, among PDE isoforms, PDE2A is unique in its exclusive localization to the mitochondrial matrix [21,[73], [74], [75]], positioning it to directly influence mitochondrial cAMP levels and NCLX phosphorylation. To explore this, we first analyzed PDE2A expression across multiple mouse tissues. Our tissue expression analysis revealed that PDE2A is abundantly expressed in liver and adipose tissue mitochondria relative to other tissues (Fig. 3D). To directly validate PKA activation downstream of PDE2A inhibition, we next examined global PKA substrate phosphorylation in HepG2 cells. Treatment with the PDE2A-specific inhibitor BAY 60-7550 (BAY60), robustly increased phospho-PKA substrate levels, confirming enhanced PKA signaling upon PDE2A inhibition. Importantly, co-treatment with the PKA inhibitor H-89 completely abolished the BAY60-induced increase in PKA substrate phosphorylation, demonstrating that BAY60's effects are strictly PKA-dependent (Fig. 3E,F). These results provide direct biochemical evidence that PDE2A inhibition elevates cAMP-PKA signaling in hepatocytes and establish the functional link between PDE2A activity and PKA-dependent phosphorylation events. To further define PDE2A's regulatory role over NCLX, we next examined whether the BAY60-induced rise in PKA activity translates into enhanced phosphorylation of NCLX at its PKA-dependent Ser258 site. Accordingly, we pretreated HepG2 cells with vehicle, the BAY60, or BAY60 together with H-89 [[25], [26], [27]]. Inhibition of PDE2A by BAY60 markedly enhanced NCLX phosphorylation at the PKA-dependent Ser258 site compared with vehicle or combined H-89 and BAY60 treatment, indicating that PKA and PDE2A act within the same signaling axis, with PDE2A functioning as a negative regulator of NCLX phosphorylation (Fig. 3G,H). Importantly, total NCLX protein levels remained unchanged across all treatment conditions (Fig. 3G,H).

To further investigate the impact of PDE2A inhibition on mitochondrial Ca²⁺ handling, we pretreated hepatocytes with BAY60 and monitored the mitochondrial Ca²⁺ dynamics using Rhod-2 AM fluorescence imaging. We initially monitored mitochondrial Ca²⁺ oscillations in response to glucagon in the presence or absence of BAY60. BAY60 significantly increased the frequency of glucagon-induced mitochondrial Ca²⁺ oscillations in WT hepatocytes, showing an increase of over 50% (Fig. 3I,J), thereby confirming that PDE2A modulates dynamic mitochondrial Ca²⁺ signaling. We then evaluated mitochondrial Ca²⁺ responses evoked by ATP stimulation. BAY60 treatment significantly enhanced mitochondrial Ca²⁺ efflux in WT hepatocytes, but not in NCLX KO cells (Fig. 3K–M). Notably, BAY60 led to a robust and accelerated Ca²⁺ efflux, resulting in a pronounced drop below basal levels compared to vehicle-treated controls (Fig. 3K,M). Importantly, mitochondrial Ca²⁺ influx remained unaffected by BAY60 in both genotypes (Fig. 3K,L). These findings suggest that PDE2A inhibition enhances mitochondrial Ca²⁺ extrusion through a mechanism dependent on NCLX.

Taken together, our findings demonstrate that the cAMP/PKA-dependent phosphorylation of NCLX at S258 fine-tunes mitochondrial Ca²⁺ responses to hormonal stimulation. These results also indicate that PDE2A-mediated inhibition of PKA dampens NCLX activity and consequently reduces mitochondrial Ca²⁺ extrusion and oscillatory capacity.

So far, our findings demonstrate that fatty acid utilization is dependent on intact NCLX activity, and that mitochondrial Ca²⁺ oscillations and efflux are modulated by PDE2A-regulated phosphorylation of NCLX. We therefore hypothesized that inhibiting PDE2A could enhance mitochondrial FAO. To test this, we performed respirometry assays in hepatocytes pretreated with glucagon and either the selective PDE2A inhibitor BAY60 or vehicle, following the experimental design in Fig. 2. As expected, using palmitate as a substrate, mitochondrial respiration was robustly stimulated in CTRL hepatocytes, whereas NCLX cKO hepatocytes showed a blunted response (Fig. 3N–S). Strikingly, BAY60 treatment further augmented palmitate-driven respiration in WT hepatocytes beyond the vehicle-induced increase (Fig. 3N,P). In contrast, BAY60 failed to restore or enhance OCR in NCLX cKO hepatocytes, underscoring that PDE2A inhibition amplifies fatty acid oxidation in an NCLX-dependent manner and is contingent on functional NCLX activity.

Overall, these results demonstrate that BAY60 enhances NCLX phosphorylation, promotes mitochondrial Ca²⁺ efflux, increases the frequency of mitochondrial Ca²⁺ oscillations, and improves fatty acid oxidation in hepatocytes. These findings support that PDE2A acts as a critical negative regulator of NCLX activity and mitochondrial metabolic function, and modulating its activity can directly alter NCLX activity and overall mitochondrial bioenergetics.

3.4. PDE2 inhibition promotes FFA oxidation and reduces hepatic steatosis

The above findings implicate NCLX and its regulatory mechanisms as critical for hepatic lipid homeostasis. Conversely, impaired NCLX activity disrupts mitochondrial Ca²⁺ handling, impairs FAO, and promotes hepatocellular lipid-droplet accumulation.

Metabolic dysfunction-associated steatotic liver disease (MASLD), along with its more advanced inflammatory form, metabolic dysfunction-associated steatohepatitis (MASH), is characterized by excessive fat accumulation in the liver driven by systemic metabolic stress [31]. Notably, mitochondrial dysfunction is a well-established early and central driver of MASLD [60,[76], [77], [78]]. A cardinal feature of the disease is impaired hepatic mitochondrial bioenergetics and FAO, which tip the balance between fatty-acid influx and clearance toward hepatic lipid accumulation and steatosis [31,32,60,77,78]. Because NCLX governs mitochondrial Ca²⁺ signaling and thereby modulates respiration, bioenergetics, and β-oxidation, we hypothesized that its altered activity represents a plausible mechanistic link to the mitochondrial defects observed in MASLD. Accordingly, we investigated whether dysregulation of NCLX contributes to MASLD pathogenesis.

To assess clinical relevance of NCLX in MASLD, we analyzed liver biopsies from human individuals with advanced steatosis (S3) compared to those with no evidence of histological steatosis (S0) [31]. Individuals with significant fibrosis (≥F2) were excluded to isolate the steatotic component and avoid confounding effects from fibrotic remodeling. All samples were obtained from women of a single, relatively genetically homogeneous ethnic background, which reduces inter-individual variability (Supplemental Tables 1 and 2). In these samples we quantified total NCLX protein using a pan-NCLX antibody and assessed its activation by measuring phosphorylation at Ser258, a critical PKA-targeted residue, with a phospho-Ser258-specific antibody. Surprisingly, our results revealed that while total NCLX abundance was comparable between the patient groups, Ser258 phosphorylation was markedly reduced in individuals with S3 steatosis relative to S0 controls (Fig. 4A,B). This selective loss of Ser258 phosphorylation in patients' data indicates a functional impairment in NCLX activity, which may disrupt mitochondrial Ca²⁺ transients, compromise mitochondrial bioenergetics, and thereby contribute to defects in FAO and hepatocellular lipid accumulation in MASLD.

Figure 4.

PDE2 inhibition promotes fatty acid oxidation and reduces hepatic steatosis. (A) Western blot analysis of liver sample replicates from two patient groups with either advanced steatosis (S3) or no steatosis (S0). Samples were probed with a phospho-specific antibody against Ser258 on NCLX (pNCLX), a PKA-dependent phosphorylation site, and total NCLX using a non-phospho-specific antibody. VDAC served as a loading control. (B) Densitometric quantification of conditions shown in A probed for pNCLX and total NCLX signal (n = 3 patients per group). (C–E) Hepatic mitochondrial PDE2A is upregulated in diet-induced MASLD. (C) Representative immunoblot of PDE2A in mitochondria isolated from livers of chow-fed and 12-week HFD-fed mice. (D) Quantification of PDE2A protein abundance normalized to loading control (n = 5 mice per group). (E) Mitochondrial cAMP-hydrolyzing (PDE) activity in livers from chow and 12-week HFD mice, expressed as percent of chow (n = 6 mice per group). (F–G) Oxygen consumption rate (OCR) in primary hepatocytes isolated from mice fed chow or CDAA-HFD (choline-deficient, l-amino acid-defined, high-fat diet, with 60% of the calorie content as fat and 0.1% methionine by weight) for 4 weeks; fatty-acid (FA)-driven respiration was induced with palmitate in the presence or absence of BAY60. (F) Representative Seahorse respirometry traces of intact primary hepatocytes from chow- and CDAA-HFD fed mice, treated ± BAY60. Palmitate-induced respiration was measured in real-time. Sequential injections of palmitate, oligomycin, FCCP, and rotenone/antimycin A were applied. (G) Quantification of basal OCR, FA-driven respiration, and fold increase in FA-stimulated OCR (n = 20–28 per condition). (H–Q) Functional and histological assessment of hepatic steatosis in ob/ob mice treated with BAY60 (1 mg/kg) for 3 weeks. (H) Western blot analysis of liver mitochondrial samples from ob/ob mice treated with BAY60 or vehicle. Samples were probed with an antibody against PDE2A. SDHA served as a loading control. (I) Densitometric quantification of PDE2A signal normalized to SDHA (n = 5 mice per group). (J) Mitochondrial cAMP-hydrolyzing (PDE) activity in ob/ob mice treated with vehicle or BAY60, expressed as percent of vehicle (n = 12 mice per group). (K) Western blot analysis of liver mitochondrial samples from ob/ob mice treated with BAY60 or vehicle. Samples were probed with a phospho-specific antibody against PKA substrate motifs (phospho-PKA substrate). SDHA served as a loading control. (L) Densitometric quantification of phospho-PKA substrate signal normalized to SDHA (n = 5 mice per group). (M) H&E staining of liver sections from BAY60- or vehicle-treated ob/ob mice. Scale bars: 1000 μm (low magnification) and 100 μm (zoomed-in view). (N) Quantification of lipid vacuole content (% area per field; n = 5–6 per group). (O) High-resolution classification of hepatocytes into non-steatotic, microvesicular, or macrovesicular categories across both treatment groups (n = 5–6). (P) Perilipin-3 (Plin3) immunostaining of liver sections from ob/ob mice treated with BAY60 or vehicle. Scale bars: 1000 μm and 100 μm (zoomed-in). (Q) Quantification of Plin3 signal density across conditions (n = 5–6 per group). All values are represented as mean ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001; ns – non-significant. Two-tailed unpaired t-test for comparisons between two groups was used and one-way ANOVA with Tukey's multiple comparison test was used for three or more groups.

Because PDE2A regulates cAMP availability for PKA-dependent NCLX activation and our human data indicate impaired NCLX phosphorylation in steatosis, we next asked whether hepatic PDE2A is itself altered under metabolic stress. To address this, we examined PDE2A expression and activity in a diet-induced model of MASLD. PDE2A protein abundance was significantly increased in livers from 12-week HFD mice compared to chow-fed controls (Fig. 4C,D). Because PDE2A is the predominant cAMP-degrading phosphodiesterase within the mitochondrial compartment [21,74], we quantified mitochondrial cAMP-degrading activity as a functional readout of intramitochondrial PDE2A. Mitochondria isolated from HFD livers exhibited significantly higher cAMP-degrading activity than chow controls, indicating elevated mitochondrial PDE2A activity under diet-induced MASLD (Fig. 4E).

These data provide mechanistic context for the reduced NCLX phosphorylation observed in human MASLD and suggest that increased PDE2A activity may contribute to impaired NCLX signaling in steatotic liver.

Recent studies show that even short-term high-fat diet feeding dampens hepatic hormone-induced intracellular Ca²⁺ oscillations, a perturbation associated with MASLD progression [62]. However, a molecular mediator linking altered Ca²⁺ dynamics to metabolic dysfunction has not been established. Because NCLX shapes mitochondrial Ca²⁺ transients and oscillations, and is functionally coupled to FAO, we hypothesized that restoring NCLX activity in the setting of steatosis would enhance mitochondrial bioenergetics and promote FAO. To test this, we pharmacologically activated NCLX via PDE2A inhibition in primary hepatocytes isolated from choline-deficient l-amino acid defined high-fat diet (CDAA-HFD), a model that recapitulates key features of human MASLD and the metabolic dysfunction that precedes MASH [42]. Littermate controls maintained on a standard chow diet served as the comparison group. Consistent with previous reports, hepatocytes from CDAA-HFD-fed mice exhibited significantly reduced basal respiration and impaired palmitate-induced FAO relative to chow-fed controls [60,61] (Fig. 4F,G). Remarkably, treatment with the PDE2A-specific inhibitor, BAY60, significantly improved both basal respiration and FAO capacity in hepatocytes derived from CDAA-HFD mice. Furthermore, maximal respiratory capacity was partially restored following BAY60 treatment, suggesting that PDE2A inhibition may counteract parts of the metabolic deficits induced by lipid overload (Fig. 4G).

These findings demonstrate that PDE2A inhibition enhances mitochondrial respiration and FAO in steatotic hepatocytes, providing a mechanistic platform to investigate whether these mitochondrial bioenergetic improvements translate into therapeutic benefit for hepatic steatosis in vivo. To extend these observations in an in vivo context, we selected the ob/ob mouse as our model. PDE2A has previously been linked to obesity-associated changes in expression and signaling, supporting its relevance under metabolic stress conditions. [79]. Given that the ob/ob mouse is a well-characterized model of severe hepatic steatosis it offers a robust and physiologically relevant system to evaluate the effects of PDE2A inhibition on hepatic lipid accumulation in vivo. Ob/ob mice were treated with BAY60 or vehicle control for three weeks [25,39,40]. To validate that BAY60 engages its mitochondrial target in vivo, we first assessed hepatic mitochondrial PDE2A expression and activity. Western blot analysis showed that total PDE2A protein levels were not significantly changed in BAY60-treated ob/ob mice compared with vehicle treatment, indicating that BAY60 does not meaningfully alter PDE2A abundance (Fig. 4H,I). Furthermore mitochondrial cAMP-degrading activity, used as a functional measure of intramitochondrial PDE2A [21,73], was significantly reduced in BAY60-treated ob/ob mice relative to vehicle controls (Fig. 4J), demonstrating effective inhibition of mitochondrial PDE2A in vivo. In parallel, immunoblotting for cAMP-dependent PKA substrate phosphorylation demonstrated higher phospho-PKA substrate levels in BAY60-treated ob/ob mitochondria (Fig. 4K,L), indicating enhanced PKA signaling downstream of PDE2A inhibition. Together, these studies confirm that BAY60 effectively inhibits hepatic mitochondrial PDE2A activity and augments cAMP–PKA signaling in ob/ob mice, providing a mechanistic foundation for evaluating whether these upstream signaling changes translate into improved hepatic lipid handling at the tissue level. Histological analysis revealed a substantial reduction in hepatic steatosis in BAY60-treated mice, with over a one-third decrease in total lipid-droplet area compared with controls, as measured by the area occupied by lipid vacuoles in tissue sections (Fig. 4M,N). Notably, the resolution of steatosis was particularly pronounced in hepatocytes with microvesicular steatosis, whereas macrovesicular lipid droplets showed relative resistance to treatment (Fig. 4M,O), suggesting preferential mobilization of the more metabolically active lipid pools. To further examine the subcellular lipid remodeling, we immunostained for perilipin-3 (Plin3), a marker associated with lipid droplet coating and turnover [80,81]. In vehicle-treated ob/ob livers, hepatocytes exhibited predominantly macrovesicular steatosis with large, Plin3-negative lipid droplets, indicative of metabolically inert stores (Fig. 4P,Q). In contrast, BAY60-treated livers showed a marked increase in Plin3-positive droplets, consistent with enhanced droplet turnover and active lipolysis (Fig. 4P,Q). Together, these results support a model wherein PDE2A inhibition not only restores mitochondrial oxidative capacity but also facilitates the mobilization and breakdown of intracellular lipid stores.

In summary, our data demonstrate that NCLX phosphorylation is significantly reduced in patients with advanced steatosis, and that impaired NCLX activation may contribute to mitochondrial dysfunction and defective fatty acid clearance in MASLD.

Pharmacological activation of the PDE2A–PKA–NCLX axis using BAY60 promotes fatty acid utilization and mitigates hepatic steatosis in both dietary CDAA-HFD and genetic (ob/ob) models of MASLD, respectively. These results support further investigation of PDE2A inhibitors as potential therapeutic agents for restoring mitochondrial function and resolving steatosis in MASLD.

4. Discussion

Previous studies underscored the role of hepatic Ca²⁺signaling in gluconeogenesis. Our previous work established that mitochondrial Ca²⁺signaling is crucial for glucagon-dependent gluconeogenesis. In the current study, we focused on the role of mitochondrial Ca²⁺efflux via NCLX in hepatic lipid metabolism. This line of study is of particular importance because metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as NAFLD, is a growing global health concern, with increasing prevalence driven by obesity, insulin resistance, and type 2 diabetes [32,39]. The liver, a central organ in metabolic regulation, counters lipid overload through tightly coordinated hormonal and mitochondrial responses [7,29,67]. However, failure of these adaptive processes leads to redox imbalance, mitochondrial dysfunction, and impaired FAO; All are key features underlying steatosis and disease progression [60]. While several studies have implicated mitochondrial impairment in the pathogenesis of MASLD, the mechanisms linking mitochondrial calcium dynamics to hepatic lipid metabolism remain poorly understood.

In this study, we identify the mitochondrial Na⁺/Ca²⁺exchanger, NCLX, as a critical regulator of intrahepatic lipolysis and fatty acid oxidation during fasting and glucagon stimulation. Our findings suggest that NCLX functions as an essential metabolic effector linking hormonal signaling to mitochondrial calcium oscillations, which regulate hepatic lipid utilization.

We show that hepatic NCLX expression is significantly upregulated in response to a high-fat/high-sucrose (HF/HS) diet across a genetically diverse panel of mice. Interestingly, rather than being pathogenic, this upregulation correlates negatively with hepatic steatosis and systemic insulin resistance, suggesting that NCLX induction may represent a compensatory, protective response during early metabolic stress. Using hepatocyte-specific NCLX KO mice, we demonstrate that NCLX is indispensable for glucagon-stimulated hepatic lipolysis. Under fasting conditions, NCLX-deficient livers accumulate intrahepatic lipid droplets and exhibit reduced hepatic glycerol production, indicating impaired lipid droplet catabolism. Notably, peripheral lipolysis remained intact in these mice, pointing toward a liver-autonomous defect in fatty acid utilization rather than systemic lipolytic failure.

Mechanistically, at the level of mitochondrial bioenergetics, we show that NCLX deficiency impairs glucagon-stimulated FAO. Seahorse respirometry of primary hepatocytes revealed a marked reduction in palmitate-induced respiration in NCLX cKO hepatocytes, despite preserved mitochondrial OXPHOS function with alternative substrates (pyruvate/malate). These data suggest a selective defect in lipid oxidation, rather than a global collapse of mitochondrial respiration. We recognize that Seahorse-based palmitate-driven OCR is an indirect readout of long-chain FAO and does not replace classic ¹⁴C/¹³C fatty-acid tracing; however, combined with CPT1-sensitive respiration and activity assays, substrate selectivity, and our in vivo lipid and glycerol measurements, it provides a robust functional assessment of NCLX-dependent FAO, while future tracer-based studies will be important to confirm absolute flux changes. These findings align with previous studies linking mitochondrial Ca²⁺dynamics to hepatic lipid metabolism via calcium-sensitive signaling. For example, a prior study using liver-specific MCU KO mice demonstrated that impaired mitochondrial Ca²⁺uptake leads to reduced FAO-coupled OCR, along with delayed cytosolic Ca²⁺clearance, calcium-dependent PP4 activation, and AMPK inactivation [82]. At the molecular level, we find that NCLX regulates the phosphorylation of key lipolytic enzymes, ATGL and HSL, which remain hypophosphorylated under fasting conditions; consistent with disrupted calcium efflux and impaired activation of CaMKII and downstream effectors.

Notably, a later liver-specific MCU KO mouse model from an independent group reported a phenotype that diverged from the earlier MCU KO study [70]. In that report, hepatic MCU deletion reduced liver lipid content and enhanced intrahepatic lipolysis, whereas the earlier study described impaired hepatic lipid clearance following MCU loss [82]. While both the recent MCU KO and our NCLX cKO studies highlight the essential role of mitochondrial Ca²⁺signaling in the phosphorylation and ultimately activation of the lipolytic enzymes ATGL and HSL, they operate via distinct mechanistic routes. In the MCU KO, increased ATGL and HSL phosphorylation was interpreted as a consequence of diminished mitochondrial Ca²⁺uptake, leading to elevated cytosolic Ca²⁺and thereby activation of CaMKII. By contrast, in the NCLX cKO, loss of mitochondrial Ca²⁺efflux abolishes glucagon-evoked, NCLX-dependent mitochondrial Ca²⁺oscillations, leaving ATGL and HSL hypophosphorylated under fasting conditions and blunting glucagon-stimulated phosphorylation dynamics, while concurrently impairing fatty-acid oxidation. Importantly, although NCLX-mediated Ca²⁺extrusion can contribute to activation of cytosolic lipolytic pathways, our data challenge the view that cytosolic Ca²⁺ levels alone dictate such regulation. This is particularly relevant given the relatively modest contribution of mitochondrial Ca²⁺uptake compared with ER stores under physiological conditions, as supported by multiple studies on hepatic calcium dynamics [3,6,9]. Instead, our data suggest that glucagon stimulates NCLX-dependent mitochondrial Ca²⁺oscillations which are abolished in the absence of NCLX. Thus, not only is mitochondrial calcium uptake via the MCU required, but subsequent efflux through NCLX is also essential to sustain these oscillations for effective fatty-acid oxidation during glucagon stimulation [16]. Collectively, these findings support a model in which directional calcium flux through mitochondria—rather than steady-state calcium levels per se— couples mitochondrial energetics to lipid oxidation through NCLX-regulated calcium oscillations [15]. Therefore, mitochondrial calcium cycling, driven by NCLX, sustains ATGL and HSL phosphorylation and tunes CPT1 sensitivity to malonyl-CoA, thereby preserving lipolytic activity and maintaining FAO capacity, providing mechanistic and bioenergetic insights absent in prior MCU deletion studies.

Furthermore, in this study, we investigated upstream regulation of NCLX and identified that it is phosphorylated at the PKA-dependent site Ser258 in response to glucagon, serving as a key determinant of NCLX activity. Under basal conditions, the mitochondrial phosphodiesterase PDE2A constrains this phosphorylation; selective inhibition of PDE2A with BAY60 markedly amplified p-Ser258, promoted NCLX-dependent mitochondrial Ca²⁺efflux and oscillatory signaling, increased FAO, and rescued bioenergetic impairment in steatotic hepatocytes. Reflecting pathological support of this inhibitory mechanism in disease, both PDE2A protein abundance and mitochondrial cAMP-hydrolyzing activity were elevated in livers of HFD mice, while in vivo BAY60 treatment of ob/ob mice reduced hepatic mitochondrial cAMP degradation and increased mitochondrial phospho-PKA substrate levels, confirming target engagement. Importantly, and of clinical relevance, phosphorylation of NCLX at Ser258 was significantly reduced in human livers with advanced steatosis (S3) despite unchanged total NCLX protein, which indicates a conserved post-translational defect that may underlie impaired FAO in MASLD and positions NCLX as a functional node downstream of dysregulated hormonal signaling in metabolic disease.

In the present study, our in vivo analysis focused on a single, well defined relatively prolonged fasting state (30 h) rather than performing a classical overnight fasting (12–16 h). By this time point, mice have largely exhausted hepatic glycogen stores and display a metabolic profile characterized by elevated lipolysis, increased hepatic fatty acid oxidation and ketogenesis, and a shift from glycogenolysis to gluconeogenesis as the dominant source of endogenous glucose. This late fasting state is associated with low insulin, relatively high glucagon, and a strong reliance on lipid-based fuels, which we considered the most relevant in vivo context in which to interrogate the hepatic glucagon–PDE2A–NCLX axis [83]. Notably, α-cell glucagon secretion into the portal vein is pulsatile, exposing the fasted liver to oscillatory rather than steady hormone levels [[84], [85], [86], [87]]. An important consideration in this context is that the metabolic consequences of glucagon depend not only on its concentration but also on the pattern and chronicity of exposure. In humans, acute bolus or short-term glucagon infusions robustly increase energy expenditure, whereas a 72-h continuous glucagon infusion fails to raise energy expenditure, consistent with tachyphylaxis to sustained hyperglucagonemia [88,89]. In contrast, chronic low-dose glucagon delivered continuously via osmotic minipumps in high-fat-fed rodents enhances intrahepatic lipolysis and mitochondrial oxidation, lowers hepatic triglycerides, and improves hepatic insulin sensitivity in an InsP₃R1-dependent manner ([8,90]). Together, these data indicate that both the tonicity and pulsatility of glucagon delivery sculpt hepatic substrate selection. Our study leverages a physiologic prolonged-fasting state and acute low-nanomolar glucagon pulses in isolated hepatocytes in vitro to position NCLX downstream of this axis, but we did not directly manipulate portal glucagon patterns in vivo. A rigorous comparison of continuous versus pulsatile glucagon delivery in wild-type versus hepatocyte-NCLX-deficient mice, using pump-based infusion protocols analogous to those described by Perry and colleagues [8], will therefore be an important goal for future work.

Lastly, we evaluated the in vivo relevance of our findings in complementary MASLD models spanning obesity-driven (HFD, ob/ob) and hepatocyte-intrinsic, non-obese (CDAA-HFD) steatosis. We used a conventional high-fat diet (45% kcal fat, 12 weeks) to model diet-induced obesity with steatosis, and a 4-week CDAA-HFD diet protocol, the second regimen has been extensively characterized as producing robust macrovesicular steatosis and steatohepatitis with only early fibrotic remodeling, whereas advanced bridging fibrosis and cirrhosis typically require 12 weeks of continuous feeding [42,91,92]; the latter was chosen to approximate a steatosis-dominant, early MASH-like stage. In parallel, we studied ob/ob mice at 10–12 weeks of age, a time window in which this genetic model also predominantly manifests MASLD prior to pharmacologic intervention. Across these settings, pharmacologic activation of the PKA–NCLX axis via PDE2A inhibition restored FAO in hepatocytes from CDAA-HFD-fed mice and reduced hepatic steatosis in ob/ob mice, with BAY60 treatment preferentially clearing microvesicular lipid droplets and enhancing overall lipid droplet turnover. These data suggest that modulating PDE2A–NCLX signaling can reverse key features of MASLD by restoring mitochondrial Ca²⁺-dependent control of FAO. Notably, this pathway has also been implicated in neuroprotection and cognition, as well as in regulating mitochondrial interactions with lipid droplets in brown adipose tissue, highlighting potential therapeutic relevance beyond the liver [25,93].

A further limitation of our study is sex as a biological variable and human-mouse comparability. Our human biopsy cohort consists exclusively of women with advanced steatosis but minimal fibrosis, reflecting clinical referral patterns rather than a deliberate exclusion of men, whereas our mechanistic mouse experiments (NCLX cKO, ob/ob, CDAA-HFD, and primary hepatocytes) were performed predominantly in adult males. MASLD is known to be sexually dimorphic, with estrogens generally protecting premenopausal females from severe steatosis and fibrosis and males and postmenopausal women being at higher risk. Therefore, our current design does not fully capture how sex hormones may tune the magnitude of glucagon–PDE2A–NCLX effects. Future studies will therefore need to test this axis explicitly in female mice and to compare male and female cohorts across diet-induced and genetic MASLD models. Nevertheless, this study supports that the glucagon–PDE2A–PKA–NCLX pathway as a conserved hepatocellular mechanism coupling mitochondrial Ca²⁺dynamics to lipid utilization, which is likely operative in both sexes even if quantitatively modulated by sex hormones. Taken together, our findings establish NCLX as a central metabolic hub that integrates hormonal signals with mitochondrial calcium handling to orchestrate lipid metabolism in the liver. Loss of NCLX activity impairs intrahepatic lipolysis and FAO, thereby contributing to steatosis. Conversely, pharmacological inhibition of PDE2A using BAY60 restores NCLX phosphorylation, enhances mitochondrial calcium efflux and oscillatory signaling, promotes FAO, reinstates metabolic flexibility, and facilitates hepatic lipid clearance in both cellular and animal models of MASLD. Although resmetirom and semaglutide have recently received accelerated FDA approval for noncirrhotic MASH with F2–F3 fibrosis, response rates remain incomplete, and no approved therapy targets hepatocellular mitochondrial oxidative capacity directly [94,95]; targeting the PDE2A–PKA–NCLX axis therefore represents a mechanistically distinct strategy. Future studies should aim to define the long-term efficacy, safety, and translational potential of PDE2A inhibitors in preclinical models, as well as in human clinical trials.

CRediT authorship contribution statement

Essam A. Assali: Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Mahmoud Taha: Methodology, Investigation, Data curation. Ahmad Salhab: Resources. Johnny Amer: Methodology. Karthickeyan Chella Krishnan: Formal analysis, Data curation. Marko Kostic: Methodology, Data curation. Michaela Veliova: Data curation. Alexandra J. Brownstein: Data curation. Orian S. Shirihai: Supervision, Resources. Rifaat Safadi: Resources. Israel Sekler: Writing – original draft, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study included organ or tissue donors. This study includes human biological material and consent was obtained by donors, or their next of kin or legal representatives, for use in this study and for publication of the article. The samples used in this research were not sourced from executed prisoners or prisoners of conscience.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was conducted in accordance with Helsinki Declaration; Israeli national regulations for human research. This study was approved by the Hadassah Medical Center Institutional Ethics Committee (Jerusalem, Israel) under protocol numbers HMO-0605-16 and HMO-57398-AR.

This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. This study was approved by the Institutional Animal Care and Use Committee (IACUC), Ben-Gurion University of the Negev; and Institutional Animal Care and Use Committee (IACUC), University of California, Los Angeles (UCLA). (Approval No. 2015-121-01, 2017-019-03 and IL-165-01-2023A).

Funding sources

This work was supported by Israel Science Foundation ISF (1424/17, 1763/21) and DIP (SE 2372/1-1) grants (I·S.) and National Institutes of Health grants R01 DK099618-05 (O·S·S.), R01 CA232056-01 (O·S·S.), R21AG060456-01 (O·S·S.), R21 AG063373-01 (O·S·S.), and ADA Grant No. 1-19-IBS-049 (O·S·S.).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Data availability

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. High-resolution Oil Red O histology images of liver from control and hepatocyte-specific NCLX knockout mice under fed and fasted conditions are deposited in the Dryad Digital Repository (https://datadryad.org/share/LINK_NOT_FOR_PUBLICATION/hkCl5joTIfRUXcAES0vfXkQK5Nuc2w0F7NyzITUAeVc). The hybrid mouse diversity panel (HMDP) data is available at https://systems.genetics.ucla.edu.