Blockade of MCU-Mediated Ca²⁺ Uptake Perturbs Lipid Metabolism via PP4-Dependent AMPK Dephosphorylation

SUMMARY

Mitochondrial Ca²⁺ uniporter (MCU)-mediated Ca²⁺ uptake promotes the buildup of reducing equivalents that fuel oxidative phosphorylation for cellular metabolism. Although MCU modulates mitochondrial bioenergetics, its function in energy homeostasis in vivo remains elusive. Here we demonstrate that deletion of the Mcu gene in mouse liver (MCU^Δhep) and in Danio rerio by CRISPR/Cas9 inhibits mitochondrial Ca²⁺ (_mCa²⁺) uptake, delays cytosolic Ca²⁺ (_cCa²⁺) clearance, reduces oxidative phosphorylation, and leads to increased lipid accumulation. Elevated hepatic lipids in MCU^Δhep were a direct result of extramitochondrial Ca²⁺-dependent protein phosphatase-4 (PP4) activity, which dephosphorylates AMPK. Loss of AMPK recapitulates hepatic lipid accumulation without changes in MCU-mediated Ca²⁺ uptake. Furthermore, reconstitution of active AMPK, or PP4 knockdown, enhances lipid clearance in MCU^Δhep hepatocytes. Conversely, gain-of-function MCU promotes rapid _mCa²⁺ uptake, decreases PP4 levels, and reduces hepatic lipid accumulation. Thus, our work uncovers an MCU/PP4/AMPK molecular cascade that links Ca²⁺ dynamics to hepatic lipid metabolism.

Graphical Abstract

In Brief

Hepatic mitochondrial Ca²⁺ shapes bioenergetics and lipid homeostasis. Tomar et al. demonstrate that MCU-mediated _cCa²⁺ buffering serves as a crucial step in controlling hepatic fuel metabolism through an MCU/PP4/AMPK molecular cascade. Identification of these molecular signaling events aids in understanding how perturbation of mitochondrial ion homeostasis may contribute to the etiology of metabolic disorders.

INTRODUCTION

Mitochondrial Ca²⁺ (_mCa²⁺) uptake plays a fundamental role in the spatiotemporal regulation of cytosolic Ca²⁺ (_cCa²⁺) and cellular bioenergetics in most cell types (Balaban, 2009; Baughman et al., 2011; Cárdenas et al., 2010; De Stefani et al., 2011; Denton and McCormack, 1980; Drago et al., 2011; Hajnóczky et al., 2000; Mallilankaraman et al., 2012a, 2012b). Because of the highly selective permeability of the inner mitochondrial membrane (IMM), _mCa²⁺ uptake is tightly regulated by the mitochondrial electrochemical gradient (ΔΨ_m), generated by the electron transport chain and facilitated by the mitochondrial Ca²⁺ uni-porter (MCU) complex (Baughman et al., 2011; De Stefani et al., 2011; Kirichok et al., 2004). The core component of the complex, the MCU, and its regulators MICU1, MICU2, MCUR1, essential MCU regulator (EMRE), and MCUb have been identified and actively studied in recent years (Baughman et al., 2011; De Stefani et al., 2011, 2016; Kamer and Mootha, 2014; Mallilankaraman et al., 2012a, 2012b; Patron et al., 2014; Payne et al., 2017; Perocchi et al., 2010; Plovanich et al., 2013; Raffaello et al., 2013; Sancak et al., 2013; Tomar et al., 2016). The MCU-regulated _mCa²⁺ fine-tunes mitochondrial bioenergetics through activation of Ca²⁺-dependent dehydrogenases and promotion of ATP synthesis (Balaban, 2009; Glancy and Balaban, 2012; Glancy et al., 2013). However, MCU-mediated _mCa²⁺ overload favors opening of the mitochondrial permeability transition pore (mPTP), leading to a cascade of cell death events (Bernardi, 1999; Biasutto et al., 2016; Izzo et al., 2016; Shanmughapriya et al., 2015a). Despite evidence that MCU functions as a multi-faceted oligomeric complex, the molecular mechanisms underlying the requirement of this complex for cellular metabolism in vivo remain highly debated (Kwong et al., 2015; Liu et al., 2017; Luongo et al., 2015; Pan et al., 2013; Tomar et al., 2016). The liver plays a central role in carbohydrate and lipid metabolism, and metabolic dysfunction in this organ has a significant effect on whole-body energy homeostasis. An imbalance between lipid catabolism and anabolism, resulting in hepatic lipid accumulation, is observed in several metabolic disorders, including diabetes, obesity, and nonalcoholic fatty liver disease (Fu et al., 2011; Mayes and Felts, 1967; Perry et al., 2014; van den Berghe, 1991). Mitochondria comprise around 18% of the hepatocyte volume and form a very efficient _cCa²⁺ regulatory system in hepatocytes (Weibel et al., 1969). Hepatic mitochondria have a high Ca²⁺ threshold for the opening of MCU channels and are directly coupled to oxidative phosphorylation (OXPHOS) (Paillard et al., 2017). Therefore, in hepatic tissue, _mCa²⁺ may play a critical role in shaping metabolic outcomes. However, the mechanistic link between Ca²⁺ dynamics and lipid metabolism, if any, remains unknown.

Because the MCU is the primary driver of changes in _mCa²⁺ and is responsible for Ca²⁺-induced remodeling of mitochondrial function, we generated loss- and gain-of-function mouse models for this protein. Genetic ablation of Mcu from hepatocytes results in depletion of matrix Ca²⁺ and impaired OXPHOS activity. Hepatic lipid accumulation occurs as a result of delayed _cCa²⁺ clearance and protein phosphatase 4-dependent AMP-activated protein kinase (AMPK) dephosphorylation. In parallel, acute deletion of hepatic AMPKα1α2 isoforms was sufficient to raise hepatic triglyceride levels, supporting a model where impaired AMPK activity is the mechanistic link between _mCa²⁺ buffering and steatosis. Importantly, _cCa²⁺ chelation or metformin treatment restored AMPK phosphorylation and lipid clearance in MCU^Δhep hepatocytes. These results demonstrate that MCU-mediated _cCa²⁺ buffering serves as a crucial step in controlling hepatic fuel metabolism.

RESULTS

Loss of Mcu Lowers Fatty Acid Oxidation-Coupled Mitochondrial Oxygen Consumption in Hepatocytes

To decipher the role of _mCa²⁺ in liver metabolism, we generated a liver-specific knockout mouse model of Mcu (MCU^fl/flAlb-Cre⁺; hereafter referred to as MCU^Δhep) by crossing the MCU^fl/fl C57BL/6 mouse with the hepatocyte-specific Cre-recombinase line (Figures 1A and S1A). MCU^Δhep mitochondria showed a striking deficit in their MCU current (I_MCU), which further establishes that MCU is the core component of the uniporter complex necessary for hepatic _mCa²⁺ uptake (Figures 1B–1D). Consistently, permeabilized hepatocytes from MCU^fl/fl but not MCU^Δhep exposed to a bolus of 10 03B1μM Ca²⁺ exhibited rapid _mCa²⁺ uptake (Figures 1E and 1F) without any effect on ΔΨ_m(Figures 1E, S1B, and S1C). Although mitochondria are known to sequester inositol 1,4,5-trisphosphate receptor (IP₃R) and calcium-release activated channel (CRAC)-mediated _cCa²⁺ (Cárdenas et al., 2010; Gilabert and Parekh, 2000; Hajnóczky et al., 1999, 2000; Hawkins et al., 2010; Pacher et al., 2000), it was unknown whether the basal mitochondrial matrix Ca²⁺ is altered in MCU^Δhep mitochondria. Using permeabilized primary hepatocytes (Mallilankaraman et al., 2012a, 2012b; Shanmughapriya et al., 2015b; Tomar et al., 2016), we found that MCU^Δhep hepatocytes have 0.5 μM mitochondrial matrix calcium, which is ~8-fold less than Mcu^fl/fl hepatocytes (4.0 μM) (Figures 1G and 1H).To further validate the specificity of Mcu loss in MCU^Δhep mice, we reconstituted the FLAG-tagged MCU wild-type (WT) in MCU^Δhep mice through adenoviral delivery (Figure 1I). The adenoviral expression of MCU-FLAG restored both MCU-mediated _mCa²⁺ uptake and basal matrix Ca²⁺ (Figures 1J–1M). We next sought to determine whether deletion of MCU lowers mitochondrial reactive oxygen species (mROS) production in hepatocytes. Primary hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were incubated with MitoSOX Red to visualize mROS levels, and basal mROS is greatly suppressed in MCU^Δhep (Figures S1D and S1E), indicating a role for MCU-dependent _mCa²⁺ uptake in mROS production.

Figure 1. Hepatocyte MCU Activity Regulates FAO-Coupled Mitochondrial Respiration.

(A) Generation and confirmation of MCU^Δhep mice by PCR genotyping (left) and western blotting (right).

(B) Schematic representation of the patch-clamp technique for measuring MCU channel activity.

(C) Representative I_MCU traces derived from MCU^fl/fl and MCU^Δhep mitoplasts.

(D) I_MCU densities (picoamperes/picofarads) in mitoplasts. n = 6.

(E) Representative traces of extramitochondrial Ca²⁺ ([Ca²⁺]_out) clearance and DJm in permeabilized hepatocytes from MCU^fl/fl and MCU^Δhep. n = 3.

(F) _mCa²⁺ uptake rate calculated from (E). n = 3.

(G) Representative traces of [Ca²⁺]_out rise from MCU^fl/fl and MCU^Δhep. n = 3.

(H) Quantification of _mCa²⁺ after addition of carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) from (G). n = 3.

(I) Immunoblot for the reconstitution of MCU in MCU^Δhep mice using adenovirus-mediated delivery. n = 2.

(J) Reconstitution of MCU restores [Ca²⁺]_out clearance ability. n = 4.

(K) _mCa²⁺ uptake rate calculated from (J). n = 4.

(L) MCU reconstitution restores the matrix Ca²⁺ in MCU^Δhep. Shown are representative traces of [Ca²⁺]_out rise in response to FCCP. n = 4.

(M) Quantification of _mCa²⁺ calculated after addition of FCCP from (L). n = 4.

(N) MCU reconstitution restores the OCR in MCU KO hepatocytes. n = 3.

(O) Quantification of basal, ATP-coupled, and maximal OCRs in hepatocytes from (N). n = 3.

(P) The FAO OCR was measured in MCU^fl/fl and MCU^Δhep hepatocytes using palmitate as a substrate with or without etomoxir (40 mM). n = 3.

(Q) Quantification of basal, ATP-coupled, and maximal FAO-coupled OCRs from (P). n = 3.

(R) Measurement of hepatocyte ATP. n = 3.

(S) Measurement of AMP:ATP ratio in hepatocytes. n = 3. Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001; ns, non-significant.

(T) Schematic of hepatocyte deletion of MCU, showing reduction in _mCa²⁺ and bioenergetic parameters.

Next, we found no changes in OXPHOS protein complex expression (Figure S1H), despite a greatly diminished basal mitochondrial oxygen consumption rate (OCR) in MCU^Δhep hepatocytes (Figures S1F and 1G). Maximal uncoupled and ATP-coupled OCRs were also found to be significantly decreased (Figures S1F and 1G), suggesting a lack of optimal production of Ca²⁺-dependent reducing equivalents like nicotinamide adenine dinucleotide (NADH) in the matrix (Glancy and Balaban, 2012; Glancy et al., 2013; White et al., 2005). Adenoviral reconstitution of MCU-FLAG in MCU^Δhep mice restored the basal, maximal, and ATP-coupled OCRs in primary hepatocytes (Figures 1N and 1O). Because _mCa²⁺ occupies a central role in mitochondrial bioenergetics, and alterations in MCU-mediated _mCa²⁺ uptake may account for the differences observed in glucose and fat oxidation, we speculated that the fatty acid oxidation (FAO)-coupled OCR was compromised in MCU^Δhep hepatocytes. To examine the FAO-dependent OCR, hepatocytes were supplemented with exogenous palmitate in glucose-free medium. To confirm endogenous FAO involvement, etomoxir was used as an inhibitor of carnitine palmitoyltransferase-1 (CPT-1). Basal and maximal FAO-coupled OCRs were significantly reduced in MCU^Δhep hepatocytes compared with the control (Figures 1P and 1Q). CPT-1 expression was unchanged, indicating that the reduction in FAO-coupled OCR was not due to the alteration in mitochondrial fatty acid transport (Figures S1I and S1J). MCU^Δhep hepatocytes did not display changes in the mitochondrial biogenesis-related transcription factors PGC1a, mitochondrial transcription factor A (mtTFA), and NRF1 (Figures S1K and S1L) but did show significant changes in the AMP:ATP ratio, consistent with the decreased OXPHOS activity (Figures 1R and 1S). These data clearly indicate that both glucose- and FAO-dependent OCRs and ATP production in hepatocytes are Ca²⁺-dependent (Figure 1T).

Loss of Mcu Promotes Hepatic Lipid Accumulation and Lowers Ketone Body Production

We next investigated the broader metabolic effects of hepatocyte-specific Mcu ablation in mice. Although the body weight was approximately equivalent, the total body fat content of MCU^Δhep mice was significantly increased during both the fed and fasting states (Figures 2A and S3A). The ratios of liver:body weight and white adipose tissue (WAT):body weight were also significantly increased in MCU^Δhep mice (Figures S3B and S3C). Biochemical analysis of livers from MCU^Δhep mouse livers revealed significant increases in fed and fasted triglycerides (TAGs) despite an increase in plasma TAGs being observed only in the fasting state (Figures 2B and 2C). During fasting, plasma ketones were significantly decreased in MCU^Δhep mice (Figure 2D), suggesting a link between MCU-dependent _mCa²⁺ and acetyl-coenzyme A (CoA) oxidation. Ten-week-old mice were subjected to three different states (fed, fasted, and re-fed) for assessment of food intake, locomotor activity, heat, fat utilization, O₂ consumption, and respiratory exchange ratio (RER). These whole-body metabolic parameters were unchanged between control and MCU^Δhep mice (Figures S2A–S2F).

Figure 2. Loss of MCU Promotes Hepatic Lipid Accumulation and Increases Total Body Fat.

(A) Body mass was monitored using NMR and DEXA before and after 24 h fasting. n = 11–12 mice per group.

(B) Liver triglycerides were measured using an enzymatic assay. n = 4–6 mice per group.

(C) Hepatic deletion of MCU results in increased plasma TAG in the fasting state, as measured by enzymatic assay. n = 5–6 mice per group.

(D) MCU^Δhep mice showed decreased plasma ketones under the fasting state, as measured by enzymatic assay. n = 5 mice per group.

(E) Primary hepatocytes were isolated from MCU^fl/fl and MCU^Δhep mice and cultured in low glucose, high glucose, and high glucose followed by starvation. Lipid droplets were visualized by confocal microscopy. n = 3.

(F) Quantification of the number of lipid droplets from (E). n = 3.

(G) Quantification of the size of lipid droplets from (E). n = 15–30 cells from each isolation. n = 3.

(H) Representative electron micrograph showing a large number of lipid droplets in MCU KO hepatocytes. n = 3 mice per group.

(I) Representative histological section image showing massive oil red O staining in MCU KO liver. n 40 images. n = 3 mice.

(J) Schematic of hepatic ablation of MCU, showing increased lipid deposition in the liver, which subsequently increased the whole-body fat content. Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

Hepatic lipid metabolism is tightly linked with glucose metabolism. Therefore, we assessed glucose homeostasis in MCU^Δhep mice. Fasting and fed blood glucose levels, as well as glucose tolerance, were unchanged between MCU^fl/fl and MCU^Δhep mice (Figures S3D and S3E). Stored hepatic glycogen is the primary energy source during early fasting, and hepatocytes generate glucose from glycogen through glycogenolysis (Petersen et al., 2017). Interestingly, MCU^Δhep mice showed increased glycogen during the fed state, with the same level of clearance in the fasting state (Figure S3F). MCU^fl/fl and MCU^Δhep hepatocytes had comparable glucose output, which suggests that deletion of hepatic Mcu does not compromise glycogenolysis (Figure S3G). MCU^Δhep hepatocytes had normal glucose up-take (Figures S3H and S3I). This suggests that the elevated TAG levels and reduced ketone bodies are possibly due to lower mitochondrial matrix Ca²⁺ in MCU^Δhep mice. We next investigated whether deletion of Mcu in hepatocytes led to lipid accumulation in the liver. Hepatocytes isolated from MCU^fl/fl and MCU^Δhep animals were cultured in the presence of low (5 mM) and high (25 mM) glucose, followed by starvation. MCU^Δhep hepatocytes retained significantly higher levels of lipid vesicles compared with the control (Figures 2E–2G). Upon starvation, lipids were depleted in control but not in MCU^Δhep hepatocytes, suggesting a lack of fat mobilization (Figures 2F and 2G). In addition to the number of lipid droplets, the size of the lipid droplets was significantly larger in MCU^Δhep hepatocytes (Figure 2G). Consistent with confocal imaging analysis, ultrastructural studies revealed increased accumulation of lipid droplets in MCU^Δhep mice under fed and fasting conditions (Figure 2H). Oil red O staining of liver sections recapitulated this and revealed extensive lipid staining in MCU^Δhep liver sections (Figure 2I). Hepatic lipid accumulation has been linked to increased endoplasmic reticulum (ER) stress; however, MCU^Δhep hepatocytes have an intact ER with no sign of stress (Figures S3J and S3K).

We asked whether Mcu deletion induces acute hepatic inflammation. Plasma alanine aminotransferase (ALT) levels were found to be unchanged between MCU^fl/fl and MCU^Δhep mice (Figure S3L). H&E staining of both MCU^fl/fl and MCU^Δhep mouse livers revealed no infiltration of inflammatory cells in MCU^Δhep tissue (Figure S3M). We next investigated whether loss of Mcu has any effect on liver TAG secretion and de novo lipogenesis. Analysis of liver TAG secretion and hepatic and white adipose lipo-genesis were found to be unchanged between MCU^fl/fl and MCU^Δhep mice (Figures S3N–S3P), indicating that hepatic lipid accumulation in MCU^Δhep mice could be due to decreased fatty acid utilization as a result of MCU deletion (Figure 2J).

Genetic Ablation of Mcu Limits _mCa²⁺ Uptake and Elevates Total Body Fat in Zebrafish

Because liver-specific deletion of Mcu in mice with the C57BL/6 genetic background caused aberrant lipid accumulation, the zebrafish model system could address the conservation of this mechanism across species. Adaptation of the CRISPR/Cas9 gene editing strategy eliminates morpholino-induced false positive and negative genetic mutant phenotypes (Figure 3A; Clark et al., 2011). We targeted a region proximal to the Mcu translation initiation codon and confirmed the 268-bp deletion allele by DNA sequencing, qRT-PCR, and western blotting (Figures 3B–3D). Isolated primary zebrafish cells were used to assess the _mCa²⁺ dynamics for both control and MCU knockout (KO) zebrafish. As expected, _mCa²⁺ uptake was significantly reduced in MCU KO zebrafish without altering ΔΨ_m (Figures 3E, 3F, S4A, and S4B). Similar to MCU^Δhep hepatocytes, cellular ATP levels were lower in MCU KO zebrafish and, to a lesser extent, in MCU^+/−cells (Figure 3G). During zebrafish cell isolation, we found that supernatants collected from MCU KO zebrafish cells appeared to have a strikingly higher fat content that was visually apparent (Figure 3H). We therefore extended the in vitro observation to the in vivo imaging approach of fat deposition in intact live zebrafish. The fluorescence intensity of lipids in the dorsal fin and tail fin were considerably increased (Figures 3I–3K), further supporting our results in MCU^Δhep mice showing that MCU deletion results in higher lipid accumulation.

Figure 3. Loss of MCU Limits _mCa²⁺ Uptake and Elevates Total Body Fat in Zebrafish.

(A) Schematic of the generation of global MCU KO zebrafish using CRISPR/Cas9.

(B) Genotyping for MCU deletion.

(C) Bar graph showing MCU mRNA abundance in WT, MCU^+/−, and MCU^−/− zebrafish.

(D) Western blot for MCU expression. n = 3.

(E) Measurement of _mCa²⁺ uptake. n = 5–10.

(F) Quantification of ionomycin-induced peak _mCa²⁺ levels from (E). n = 5–10.

(G) Bar graph representing cellular ATP levels in cells isolated from MCU^+/+, MCU^+/−, and MCU^−/− zebrafish. n = 4.

(H) Adult WT and MCU^−/−zebrafish were homogenized and centrifuged. MCU^−/−zebrafish samples show a clear yellow color lipid layer on top of the protein lysate. n = 6.

(I) Adult WT and MCU^−/−zebrafish were stained with Lipid Green to monitor the distribution of lipids in the whole body. n = 3. (J and K) Bar graphs represent the quantification of Lipid Green staining from the dorsal (J) or tail (K) fin. n = 3.

Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

MCU Deletion in Hepatocytes Restrains Nutrient Sensing through Altered _cCa²⁺ Signaling

MCU-mediated _mCa²⁺ plays an essential role in mitochondrial bioenergetics and is required for maintenance of hepatic ATP content (Figure 1). Having observed elevated TAGs and decreased ketone body levels in MCU^Δhep mice, we speculated that the decreased ATP levels might activate signaling via the energy-sensing AMPK (Hardie, 2015; Hardie et al., 2012; Kahn et al., 2005; Mihaylova and Shaw, 2011). In contrast to our expectations, MCU deletion resulted in a large reduction in T172-phosphorylated (active) AMPK-α, whereas total AMPK protein and mRNA levels remained unchanged (Figures 4A, 4B, and S5A). Expression of the upstream kinases LKB1 and calcium-regulated kinase calcium/calmodulin dependent protein kinase kinase-II (CaMKK-II) remained unchanged (Figures 4A and 4C). However, downstream inhibitory phosphorylation of acetyl-CoA carboxylase (ACC) was decreased (Figure 4A). Expression of two other key proteins involved in lipid synthesis, fatty acid synthase and SREBP1, remained unchanged (Figure 4A). Adenoviral reconstitution of MCU in MCU^Δhep hepatocytes restored phosphorylation of AMPK (Figure 4A). To rule out a global dephosphorylation event, we also monitored extracellular signal-regulated kinase (ERK) phosphorylation and found normal ERK phosphorylation in MCU KO hepatocytes (Figure S5B). To confirm the potential of impaired AMPK signaling to drive hepatic lipid accumulation in MCU^Δhep mice, we next generated hepatic AMPK-KO (AMPK^Δhep) mice by injecting adenovirus-associated virus 8-Cre (AAV8-Cre) into AMPKα1/α2^fl/fl mice, which have LoxP sites flanking both the AMPKα1 and AMPKα2 loci. The loss of AMPK protein was validated by western blotting (Figure 4D). Biochemical analysis of hepatic tissue collected from overnight-fasted mice showed significant hepatic lipid accumulation in AMPK^Δhep mice (Figure 4E). This observation was further validated by oil red O staining of liver tissue section and boron-dipyrromethene (BODIPY) staining of primary hepatocytes (Figures 4F, S5C, and S5D). However, the loss of AMPK did not affect _mCa²⁺ uptake or ΔΨ_m (Figures S5E and S5F), indicating that AMPK is downstream of MCU in lipid accumulation. The involvement of AMPK was further confirmed by using consti tutively active AMPK (AMPKα1 T172D, hereafter called CAAMPK) (Hardie, 2011; King et al., 2009). Primary hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were infected with an CA-AMPK-expressing adenovirus and monitored for lipid accumulation (Figures 4G–4J and S5G).

Figure 4. Ablation of _mCa²⁺ Uptake Augments the _cCa²⁺ Rise That Enhances AMPK Dephosphorylation.

(A) Hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were lysed in radioimmunoprecipitation assay (RIPA) buffer and immunoblotted for the indicated antibodies. n = 6.

(B) Bar graph representing the pAMPK:AMPK ratio from (A). n = 6.

(C) Immunoblot analysis for phospho and total CaMKK-II. n = 6.(D) Hepatocyte lysates from AMPKα1/α2^fl/fl and AMPKα1/α2^fl/flCre+ mice were probed for AMPKa protein abundance. n = 3.

(E) Hepatic TAG levels. n = 5.

(F) Assessment of hepatic lipid accumulation by Oil-Red-O staining. n = 5.

(G) Primary hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were infected with a CA-AMPK-expressing adenovirus. Hepatocytes were stained with oil red O and measured. n = 6–12 replicates from 3 mice.

(H) Visualization of lipid droplets from CA-AMPK reconstituted MCU^Δhep hepatocytes. n = 3.

(I) Quantification of the number of lipid droplets from (H). n = 20–30 cells. n = 3.

(J) Schematic depicting a link between MCU and AMPK phosphorylation.

(K) GCaMP6-expressing hepatocytes from MCU^fl/fl and MCU^Δhep were stimulated with thapsigargin (Tg) or Vasopressin, and _cCa²⁺ dynamics were monitored. n = 3.

(L) Quantification of the _cCa²⁺ clearance rate from (K). n = 15–25 cells. n = 3.

(M) GCaMP6-expressing hepatocytes from MCU^fl/fl and MCU^Δhep were stimulated with Vasopressin and _cCa²⁺ dynamics were monitored. n = 3.

(N) Quantification of the _cCa²⁺ clearance rate from (M). n = 15–25 cells. n = 3.

(O) Hepatocytes isolated from MCU^fl/fl and MCU^Δhep adult mice were transduced with an adenovirus expressing the _cCa²⁺ sensor GCaMP6, and basal _cCa²⁺ fluorescence was quantified. n = 74–91 cells from 3 mice in each group.

(P) Hepatocytes were treated with the intracellular Ca²⁺ chelator BAPTA-AM for various times. Cell lysates were immunoblotted for the indicated antibodies. n = 4.

(Q) Hepatocytes were treated with BAPTA-AM overnight, and oil red O stain was quantified. 12 replicates from 3 mice per group.

(R) Visualization of lipid droplets from control and BAPTA-AM-treated hepatocytes. n = 3 mice per group.

(S) Quantification of lipid droplets from control and BAPTA-AM-treated hepatocytes. n = 20–30 cells per group. n = 3 mice per group.

(T) Schematic depicting how _cCa²⁺ determines lipid clearance, possibly through AMPK phosphorylation.

Statistical analysis: mean ± SEM. **p < 0.01, ***p < 0.001.

Blockade of _mCa²⁺ Uptake Augments a _cCa²⁺ Rise that Enhances AMPK Dephosphorylation

To decipher the underlying mechanism of AMPK dephosphorylation in MCU^Δhep mice, we investigated whether loss of MCU has any effect on _cCa²⁺ dynamics. Hepatocytes were infected with an adenovirus expressing the _cCa²⁺-reporter GCaMP6, and _cCa²⁺ dynamics were monitored in response to thapsigargin and vasopressin. The MCU^Δhep hepatocytes exhibited delayed _cCa²⁺ clearance after stimulation with thapsigargin and vasopressin (Figures 4K–4N). Remarkably, basal _cCa²⁺ was elevated in MCU^Δhep hepatocytes (Figure 4O). Because loss of MCU caused persistent elevation of _cCa²⁺, we investigated whether chelation of _cCa²⁺ restores the phosphorylated form of AMPK in MCU^Δhep hepatocytes. When hepatocytes were pretreated with 1,2-Bis (2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid tetrakis (acetoxymethyl ester) (BAPTA-AM; 25 μM), a significant amount of phosphorylated AMPK was detected, even as early as 30 min (Figure 4P). Next, we investigated whether BAPTA-AM-mediated chelation of Ca²⁺ c improves lipid clearance in MCU^Δhep hepatocytes. Interestingly, chelation of _cCa²⁺ reduced the lipid content in MCU^Δhep hepatocytes, as indicated by oil red O and BODIPY staining (Figures 4Q–4S). Collectively, these data indicate that blockade of _mCa²⁺ buffering results in AMPK dephosphorylation through aberrant _cCa²⁺ elevation, leading to hepatic lipid accumulation in MCU^Δhep mice (Figure 4T).

Cytosolic Ca²⁺ Rise Augments AMPK Dephosphorylation through Protein Phosphatase-4

Having determined that AMPK dephosphorylation is a _cCa²⁺-dependent signal, we used pharmacologic and genetic tools to identify the phosphatase responsible for AMPK dephosphorylation. The pan protein phosphatase inhibitor (PP1, PP2A, PP2B, and PP4 inhibitor) okadaic acid restored AMPK phosphorylation, suggesting a phosphatase dependent-event (Figure 5A). Because AMPK dephosphorylation is a _cCa²⁺-dependent protein phosphatase and calcineurin is a Ca²⁺-calmodulin-activated protein phosphatase, we tested whether FK506 would prevent AMPK dephosphorylation. FK506 pre-treatment did not restore AMPK phosphorylation in MCU^Δhep mice (Figure 5B). We asked whether the Ca²⁺-dependent, ubiquitously expressed PP2A subfamily phosphatase PP4 could be responsible for AMPK dephosphorylation (Shui et al., 2007). Pre-treating MCU^Δhep hepatocytes with the PP4 inhibitor cantharidin restored the phosphorylated form of AMPK (Figure 5C). We also found that PP4 protein expression was increased in MCU^Δhep hepatocytes (Figures 5D and 5E), indicating a _cCa²⁺-regulated increased PP4 protein abundance. To verify the involvement of PP4 in AMPK dephosphorylation, we silenced PP4 expression in MCU^Δhep hepatocytes using small interfering RNA (siRNA) (Figure 5F). PP4 knockdown resulted in restoration of AMPK phosphorylation (Figure 5F). Next we used the reverse approach to confirm the involvement of PP4 in AMPK dephosphorylation by ectopic expression of PP4. Overexpression of PP4-FLAG resulted in dephosphorylation of AMPK in MCU^fl/fl hepatocytes that was further enhanced in MCU^Δhep hepatocytes, demonstrating PP4 to be a major phosphatase of AMPK in the liver (Figure 5G). To identify whether AMPK and PP4 exist as a Ca²⁺-dependent protein complex, we performed protein-protein interaction studies, which demonstrated that AMPK and PP4 bind one another (Figure 5H) and that this binding is regulated by _cCa²⁺ (Figures 5I and 5J). Silencing of PP4 in MCU^Δhep hepatocytes decreased the lipid content, as indicated by oil red O and BODIPY staining (Figures 5K–5M). These findings reveal that the blockade of rapid sequestration of Ca²⁺ into the mitochondrial matrix by MCU deletion results in activation of Ca²⁺-dependent PP4, which dephosphorylates AMPK (Figure 5N).

Figure 5. AMPK Dephosphorylation in MCU KO Hepatocytes Is Likely Due to PP4.

(A) Hepatocytes were treated with okadaic acid (OA) (50 nM) for 2 h. Cell lysates were immunoblotted for the indicated antibodies. n = 4.

(B) MCU^fl/fl and MCU^Δhep hepatocytes were treated with FK506 (2 μM) for 2 h and immunoblotted for the indicated antibodies. n = 4.

(C) MCU^fl/fl and MCU^Δhep hepatocytes were treated with cantharidin (50 μM) for 2 h and immunoblotted for the indicated antibodies. n = 4.

(D) Hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were lysed and immunoblotted for the indicated antibodies. n = 4.

(E) Densitometric analysis of PP4 protein abundance. n = 4.

(F) Hepatocytes were transfected with PP4 siRNA for 72 h. Cell lysates were immunoblotted for the indicated antibodies. n = 4.

(G) Hepatocytes were transfected with PP4-FLAG for 48 h. Cell lysates were immunoblotted for the indicated antibodies. n = 4.

(H–J) HepG2 cells were transfected with AMPK-hemagglutinin (HA) and PP4-FLAG plasmids for 48 h. Cell lysates were immunoprecipitated with an HA antibody and probed with the indicated antibodies (H). Under a similar condition (I and J), cells were treated with BAPTA-AM in the presence or absence of ionomycin (50 nM) stimulation. The reciprocal immunoprecipitation was performed with HA or FLAG antibodies and probed with the indicated antibodies. n = 3.

(K) Visualization of lipid droplets from scrambled (Scr) siRNA- and PP4 siRNA-treated hepatocytes. n = 3.

(L) Quantification of lipid droplets from Scr siRNA- and PP4 siRNA-treated hepatocytes. n = 20–30 cells. n = 3.

(M) Hepatocytes isolated from MCU^fl/fl and MCU^Δhep adult mice were transfected with PP4 siRNA for 72 h. Cells were stained with oil red O and quantified. n = 3.

(N) Schematic depicting aberrant _cCa²⁺ clearance in MCU KO hepatocytes exhibiting AMPK dephosphorylation through elevated PP4 activity.

Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

The Gain-of-Function MCU Controls PP4, which Enhances Hepatic Lipid Clearance during Fasting

We sought to understand whether MCU hyperactivation has the potential for higher lipid mobilization and increased FAO. A recent report revealed that the human MCU C97A mutation results in hyperactivation of MCU channel activity (Figure S6A; Dong et al., 2017). Therefore, we generated a gain-of-function MCU mouse model (C96A-knockin [KI]) with a global MCU C96A (corresponding to human MCU C97) KI using the CRISPR/Cas9 approach (Figures 6A and S6B), which was confirmed by restriction fragment length polymorphism (RFLP) and Sanger sequencing (Figures 6A and S6C). The C96A-KI mice were viable (Figure 6B) but had reduced body weight compared with the WT control (Figures 6C and S6D). We investigated the _mCa²⁺ dynamics and noted that C96A-KI hepatocytes exhibited an increased _mCa²⁺ uptake rate as well as higher basal matrix _mCa²⁺ under normal ΔΨ_m (Figures 6D–6G, S6E, and S6F). Although western blot analysis of OXPHOS proteins revealed no differences between WT and C96A-KI hepatocytes (Figure 6H), C96A-KI hepatocytes exhibited higher basal and maximal OCRs (Figures 6I and 6J).

Figure 6. Hepatocytes Harboring MCU-C96A Display Reduced PP4 and Augmented Hepatic Lipid Clearance during Fasting.

(A) Schematic depicting the strategy used to generate MCU-C96A KI mice. Top right: genotyping of MCU-C96A KI mice. Bottom: MCU protein abundance.

(B) 12-week-old MCU-C96A KI mice are viable.

(C) MCUC96A-KI mice have reduced body weight at 12 weeks. n = 5.

(D) MCU-mediated extramitochondrial Ca²⁺ clearance and _mCa²⁺ uptake. n = 3.

(E) MCU-mediated _mCa²⁺ uptake rate was calculated from (D). n = 3.

(F) Measurement of basal _mCa²⁺ after addition of FCCP. n = 3.

(G) Quantification of basal _mCa²⁺ after addition of FCCP from (F). n = 3.

(H) Western blot analysis of mitochondrial respiratory chain subunits of complex I (NDUFB8), complex II (succinate dehydrogenase complex iron sulfur subunit B [SDHB]), complex III (UQCRC2), complex IV (MTCO1), and ATP synthase subunit ATP5A. MCU and TOM20 were used as loading controls. n = 3.

(I) Measurement of mitochondrial OCR. n = 3.

(J) Measurement of basal, ATP coupled, and maximal OCR from (I). n = 3.

(K) Hepatocytes isolated from WT and C96A-KI mice were lysed and probed with the indicated antibodies. n = 3.

(L) Densitometric analysis of pAMPK/AMPK ratio from (K). n = 3.

(M) Densitometric analysis of PP4 protein expression from (K). n = 3.

(N) Measurement of liver and triglycerides (TAGs) under fasting conditions. Measurement of these parameters is described in Figure 2. n = 5.

(O) Measurement of plasma triglycerides (TAG) under fasting conditions. n = 5.

(P) Measurement of plasma ketone bodies under fasting conditions. n = 5.

(Q) Visualization of lipid droplets from WT and C96A KI hepatocytes. n = 3.

(R) Quantification of lipid droplets from (Q). n = 20–30 cells. n = 3.

(S) Representative histological section image shows reduced lipid accumulation in C96A-KI mice. n = 3 mice per group. Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

Next we monitored the AMPK and ACC phosphorylation status and the expression of PP4 in C96A-KI hepatocytes. C96A-KI he patocytes had higher levels of phosphorylated AMPK and ACC and a significant reduction in PP4 protein abundance (Figures 6K–6M). Biochemical analysis of hepatic tissue from C96A-KI mice clearly showed a reduction in liver and plasma TAG levels along with a significant increase in plasma ketone levels (Figures 6N–6P). We next investigated the effect of C96A-KI on high-glucose-induced lipid accumulation in primary hepatocytes. C96A-KI hepatocytes showed a clear reduction in lipid droplet accumulation under high-glucose conditions (Figures 6Q and 6R). In support of these data, oil red O staining confirmed enhanced lipid clearance by C96A-KI mice during fasting (Figure 6S). This newly generated mouse model further validates the involvement of MCU in the regulation of hepatic lipid homeostasis.

Restoration of AMPK Phosphorylation in MCU^Δhep Mice Improves Lipid Clearance

It has been shown that cells treated with metformin exhibit increased AMPK phosphorylation (Hawley et al., 2002; Madiraju et al., 2014; Shaw et al., 2005; Zhou et al., 2001). Metformin treatment did not alter _mCa²⁺ uptake (Figure S7A). We next tested whether supplementing metformin would promote AMPK phosphorylation and lipid clearance in MCU^Δhep. MCU^Δhep hepatocytes treated with metformin had significantly increased AMPK phosphorylation, similar to the effect of 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) (Figures 7A and 7B). Given the effect of metformin on AMPK phosphorylation and fatty acid metabolism, we examined whether metformin administration could deplete lipid accumulation in MCU^Δhep mice. After 3 weeks of oral metformin administration (1.25 mg/mL), we noted a marked reduction in lipid content in MCU^Δhep livers, as evidenced by oil red O staining (Figure 7C). Similarly, acute metformin treatment markedly reduced hepatocyte lipid content, as evidenced by BODIPY-positive lipid vesicle numbers (Figures 7D and 7E), but did not normalize the remaining lipid droplet size (Figure S7B).

Figure 7. Activation of AMPK by Metformin Corrects Hepatic Lipidome Remodeling in MCU KO Mice.

(A) Hepatocytes isolated from MCU^fl/fl and MCU^Δhep treated with metformin (1 mM), AICAR (100 μM) for 16 hours. After treatment, cell lysates were probed with indicated antibodies. n = 3.

(B) Bar graph showing the quantification of pAMPK in Figure 7A. n = 3.

(C) Representative histological section image shows clearance of lipids in MCU KO metformin-administered mice. n = 3.

(D) Visualization of lipid droplets from MCU^fl/fl and MCU^Δhep mice with or without metformin treatment. n = 3.

(E) Quantification of lipid droplets from (D). n = 15–30 cells per mice. n = 3.

(F) Lipids were analyzed as described in STAR Methods, and the identified features were subjected to a hierarchical cluster analysis. n = 4.

(G) Significant top 15 lipids (false discovery rate [FDR] = 0.001) representing the relative lipid modulation among the groups. The y axis represents normalized values. *, significantly different from MCU^Δhep; #, significantly different from metformin-administrated MCU^Δhep. n = 4 per group.

Statistical analysis: mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

Finally, we conducted a comprehensive lipidomics analysis and characterized the lipid species in control, MCU^Δhep, and metformin-treated MCU^Δhep mice. Lipidomics analysis of MCU^Δhep liver tissue showed a significant accumulation of TAG and diacylglycerol (DAG) lipid species (Figures 7F, 7G, S7C, and S7D). Notably, administration of metformin not only restored AMPK phosphorylation but also significantly lowered several TAG and DAG species (Figures 7F and 7G). Collectively, these data indicate that loss of MCU results in altered hepatic lipid metabolism, leading to hepatic lipid accumulation and lipidomic remodeling.

DISCUSSION

We report that hepatic deletion of MCU lowers hepatic _mCa²⁺, reduces FAO-coupled mitochondrial respiration, and increases hepatic lipid accumulation, linking _mCa²⁺ handling to hepatic lipid metabolism. Our characterization of this mouse model reveals that, following MCU ablation, there is an accumulation of _cCa²⁺ because of a lack of _mCa²⁺ sequestration via MCU into the matrix. This observation is in contrast to earlier reports heart- and endothelium-specific MCU KO models, in which no apparent changes were observed in _mCa²⁺ (Luongo et al., 2015; Pan et al., 2013; Tomar et al., 2016). Depletion of _mCa²⁺ and reduced _cCa²⁺ clearance ability in MCU^Δhep hepatocytes indicates that MCU could be a major player in shaping metabolic outcomes in the liver. Although _mCa²⁺ overload is necessary for PTP opening and mitochondrial dysfunction, MCU KO failed to protect against cell death under ionomycin and oxidative stress conditions, signifying the physiological importance of _mCa²⁺ buffering and cell survival (Nemani et al., 2018; Pan et al., 2013). Recently, it has been demonstrated that prevention of MCU-mediated Ca²⁺ up-take exhibited enhanced mitochondrial shape transition through Miro-1 and triggered cell death in a PTP-independent pathway (Nemani et al., 2018) Alternatively, MCU-dependent Ca²⁺ enhances mitochondrial constriction, which is Drp1-independent, suggesting that Ca²⁺ dynamics may be an initiator of preconstruction (Chakrabarti et al., 2018). Nevertheless, altering _mCa²⁺ may contribute significantly to changes in the mitochondrial phenotype and cellular metabolism.

It has been established that the _cCa²⁺ clearance mechanisms operate differentially between tissues (Fieni et al., 2012; Paillard et al., 2017; Pan et al., 2013). For example, there are several reasons why one could speculate that _mCa²⁺ remained unchanged in myocytes after MCU deletion. The extracellularly and intracellularly mediated _cCa²⁺ rise is rapidly cleared by the high-Ca²⁺ affinity sarcoplasmic or endoplasmic reticulum Ca²⁺ ATPase (SERCA) and plasma membrane pumps (plasma membrane Ca²⁺ ATPase [PMCA]) and sodium Ca²⁺ exchanger (NCX) in cardiomyocytes (Fearnley et al., 2011), where ΔΨ_m and mitochondrial buffering capacity is weaker because of excitation-contraction (E-C) coupling. In non-excitable cell types like hepatocytes, MCU clears _cCa²⁺ faster than SERCA and PMCA pumps. Because _mCa²⁺ uptake is primarily driven by the electro-chemical gradient across the mitochondrial inner membrane, it is possible that hepatocyte mitochondria are maintained in a hyperpolarized state specifically to facilitate _cCa²⁺ clearance and maintain heterogeneity (Collins et al., 2002; Kuznetsov et al., 2004). Another striking difference we observed in hepatocytes was the reduction of both glucose and fatty acid-dependent mitochondrial OXPHOS. This further strengthens our finding of lower basal _mCa²⁺ in MCU KO hepatocytes.

The most striking phenotype of MCU^Δhep mice is dyslipidemia and ectopic accumulation of lipid in the liver, which was associated with reduced fasting plasma ketones. These data suggest an impairment in FAO and, consequently, ketone body production in MCU KO hepatocytes. The lipid accumulation phenotype was observed in the mouse as well as zebrafish model systems, indicating conservation of MCU-regulated lipid metabolism across species. During starvation, AMPK promotes phosphorylation and inactivation of ACC, which results in fatty acid uptake into mitochondria for β-oxidation (Hardie et al., 2012). Remarkably, loss of MCU in the liver caused impairment of FAO and massive accumulation of total lipids under fasting conditions, suggesting the possibility that AMPK activity is impaired by the dysregulated cellular Ca²⁺ dynamics (Figures 4K–4O). As predicted, deletion of MCU in hepatocytes caused delayed _cCa²⁺ clearance under basal and hormonal stimulation. Activation of AMPK by increased AMP:ATP is well-established in the literature, but our findings demonstrate that, despite greatly decreased ATP and increased AMP in MCU KO hepatocytes, there was decreased AMPK activation (Figures 1S and 4A). In follow-up studies, we tested whether loss of AMPK alone was sufficient to confer a fatty liver phenotype. Interestingly, liver-specific AMPK KO mice displayed similar hepatic lipid accumulation, confirming the central role of AMPK in MCU KO-induced hepatic lipid accumulation. Chelation of _cCa²⁺ restored phospho-AMPK levels in MCU KO hepatocytes, confirming the effect of Ca²⁺-dependent regulation (Figure 4P).

It has been demonstrated that, in addition to its regulation by AMP, AMPK is activated by the upstream kinases LKB1 and CaMKK-II, which phosphorylate a conserved Thr172 in the a subunit (Hawley et al., 2005; Hurley et al., 2005; Woods et al., 2003). CaMKK-II activates AMPK via phosphorylation of the a subunit in the presence of increased _cCa²⁺ (Hawley et al., 2005; Hurley et al., 2005). Because we propose a model of Ca²⁺-mediated regulation of AMPK, it is pertinent to question whether CaMKK-II plays a role in this mechanism because of its known Ca²⁺-induced activation.

However, we did not observe any significant alteration in CaMKK-II phosphorylation in MCU^Δhep mice. This finding is also consistent with studies that have demonstrated that CaMKK-II has limited distribution in hepatic tissue in comparison with neurons and T cells (Anderson et al., 1998; Mihaylova and Shaw, 2011). Surprisingly, our biochemical analysis found significant levels of AMPK dephosphorylation without any effect on the upstream kinases LKB1 and CaMKK-II in MCU KO hepatocytes, suggesting a mechanism of AMPK regulation independent of the canonical signaling cascade.

Although ATP antagonizes AMPK activation, dephosphorylation-dependent inactivation of AMPK is mediated by serine-threonine phosphatase proteins, including PP2A and PP2Cα (Hardie, 2011). Our search for a Ca²⁺-dependent AMPK dephosphorylation mechanism revealed that pharmacologic inhibition of _cCa²⁺ regulated PP4 by cantharidin and knockdown of PP4 restored AMPK phosphorylation and reduced hepatic lipid accumulation in MCU KO hepatocytes (Figure 5). Our study demonstrates direct regulation of AMPK through Ca²⁺-induced PP4 activity. To further test the physiological role of MCU, we created a unique gain-of-function MCU mouse model that displays higher _mCa²⁺ uptake and rapidly clears _cCa²⁺ (Figure 6). Interestingly, MCU C96A-KI mice not only exhibited higher MCU-mediated Ca²⁺ uptake but also reduced PP4 levels and lower circulating and hepatic TAGs in the fasted state.

The CA form of double ACC KI mice exhibits elevated lipogenesis and lower FAO, implying that AMPK activation is crucial for lipid clearance (Fullerton et al., 2013). Our data provide evidence that sustained _cCa²⁺ elevation in MCU^Δhep mice results in activation of ACC because of AMPK dephosphorylation and, consequently, lipid accumulation. Strikingly, our MCU C96A-KI mice restored AMPK and ACC phosphorylation states and rapid clearance of hepatic lipid during fasting. The insulin-sensitizing role of metformin has been linked to AMPK phosphorylation in the setting of obesity and insulin resistance (Howell et al., 2017; Li et al., 2011). Administration of metformin in MCU^Δhep mice improved lipid clearance through MCU-independent AMPK phosphorylation, supporting a therapeutic intervention for disorders of hepatic insulin resistance and lipid accumulation.

STAR⋆METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for reagents may be directed to and will be fulfilled by the Lead Contact, Muniswamy Madesh (muniswamy@uthscsa.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse models

Hepatocyte-specific Mcu knockouts (MCU^Δhep) were generated by crossing MCU^fl/fl mice (Luongo et al., 2015) with hepatocyte-specific Cre recombinase transgenic mice (B6.Cg-Tg(Alb-cre)21Mgn/J, The Jackson Laboratory, USA). Hepatic AMPK-KO (AMPK^Δhep) mice were generated by retro orbital injection of the 1×10¹¹ viral particles of adenovirus associated virus 8-Cre (AAV8-Cre under the control of the TGB promoter) in the AMPKα1/α2^fl/fl mice which have the Lox-P site on both AMPKα1 and AMPKα2 loci. Mice were allowed to rest for 2 weeks before the experiment. To generate MCU-C96A knock in mice, gRNA1, gRNA2, ssODN containing BssHII site along with Cas9 protein was microinjected in single-celled embryo and transferred to surrogate mother. Details of oligo’s sequence used for C96A-KI are given in Figure S5B. The C96A-KI is confirmed by restriction fragment analysis using BssHII restriction enzyme. All mice were grouped according to same sex and age for all the experiments. For studies utilizing primary hepatocytes in vitro, both male and female mice were used. However, male mice were used for all in vivo experiments. All animal experiments were approved by Temple University’s IACUC and followed AAALAC guidelines.

Zebrafish

To generate zebrafish mcu mutant, three different sgRNA target sequences were designed (with PAM sequences in bold, #1: GCC GGGTTTCACTTCAGAGATGG (+strand), #2: GGCTGCGAAAGTGTGTAGATCGG (+ strand), and #3: TGTGCCCTGATGCCTCTGT GAGG (−strand). sgRNAs were in vitro transcribed and co-injected with mRNA coding for nCas9n into 1-cell zebrafish embryos. Activity of sgRNAs was tested by loss of underlined restriction enzyme sites Hpy118I (TCNGA), Sau3AI (GATC) and EcoNI (#3, CCTN₅AGG). Primers mcu-F1 (TAGAGACCGTGGTATTTCAGTCTAC) and mcu-R1 (TGCACTGCACAGTGATGATCAGCA) were used for genotyping. A deletion allele tpl124, generated using sgRNA#2, containing a 268 base pair deletion encompassing parts of exon 1 and intron 1, was selected for further study. All Zebrafish experiments were approved by Temple University’s IACUC and followed AAALAC guidelines.

Cell culture

Primary hepatocytes were grown in Williams E medium containing 1% (v/v) antibiotic-antimycotic solution (GIBCO), 1% (v/v) 200mM L-glutamine, and 1% (v/v) non-essential amino acids under standard culture condition. HepG2 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/10% FBS, supplemented with 1% (v/v) antibiotic-antimycotic solution (GIBCO) at 5% CO₂ and 37 C.

METHOD DETAILS

Isolation of mouse primary hepatocytes

Primary adult mouse hepatocytes were isolated from 10–12 week old male and female animals using a two-step collagenase perfusion technique with slight modifications (Li et al., 2010). In brief, mice liver was sequentially perfused with perfusion medium-I (DPBS containing 10mM HEPES, 0.05% w/v KCl, 5mM Glucose, 200 μM EDTA, pH 7.4) and perfusion medium-II (DPBS containing 30mM HEPES, 0.05% w/v KCL, 5mM Glucose, 1mM CaCl₂, pH 7.4) containing collagenase D (400 μg/ml). Liver lobes were dissected, dissociated, and crude hepatocyte preparation passed through the gauze mesh filter (100 μM diameter). The crude hepatocyte preparation was centrifuged at the speed of 50 g for 2 min to pellet down parenchymal hepatocytes. The hepatocytes were washed five times with perfusion medium-II using the above-mentioned centrifuge conditions. Following the wash, cells were plated in culture dishes in hepatocyte attachment medium (Williams E medium containing 1% (v/v) antibiotic-antimycotic solution (GIBCO), 1% (v/v) 200mM L-glutamine, 1% (v/v) non-essential amino acids, and 10% heat inactivated fetal bovine serum). The next day, attachment media was replaced with hepatocyte culture medium (Williams E medium containing 1% (v/v) antibiotic-antimycotic solution (GIBCO), 1% (v/v) 200mM L-glutamine, and 1% (v/v) non-essential amino acids).

Mitoplast I_MCU patch-clamp recording

Mitoplast patch-clamp recordings for I_MCU current were performed at 30 C as reported earlier (Chaudhuri et al., 2013; Hoffman et al., 2013; Joiner et al., 2012; Kirichok et al., 2004; Tomar et al., 2016) with slight modifications. In brief, mitoplasts isolated from hepatocytes of MCU^fl/fl and MCU^Δhep mice were bathed in a solution containing 150mM sodium gluconate, 5.4mM KCl, 5mM CaCl₂, and 10mM HEPES (pH 7.2). The pipette solution contained 150mM sodium gluconate, 5mM NaCl, 135mM sucrose, 10mM HEPES, and 1.5mM EGTA (pH 7.2). I_MCU currents were recorded using an Axon200B patch-clamp amplifier with a Digidata 1320A acquisition board (pCLAMP 10.0 software; Axon Instruments). The external/bath solution (5 mM Ca²⁺) was chosen on the basis of previous measurements.

Measurement of mitochondrial Ca²⁺ uptake, ΔΨ_m, and mitochondrial matrix Ca²⁺

Mitochondrial Ca²⁺ uptake and ΔΨ_m were measured as described earlier using dual-wavelength emission fluorimeter with slight modifications (Mallilankaraman et al., 2012a; Mallilankaraman et al., 2012b; Shanmughapriya et al., 2015a; Tomar et al., 2016). Freshly isolated mouse primary hepatocytes were washed with Ca²⁺/Mg²⁺ free DPBS (GIBCO). An equal number of cells (5×10⁶ cells) were permeabilized with 40 μg/ml digitonin in 1.5 mL of intracellular medium (ICM-120 mM KCl, 10 mM NaCl, 1 mM KH₂PO₄, 20 mM HEPES-Tris, pH 7.2) supplemented with 2 μM thapsigargin to block the SERCA pump and 5mM succinate to energize the mitochondria. Fura-FF (0.5 mM) was loaded at 0 s time point for the measurement of extramitochondrial Ca²⁺ ([Ca²⁺]_out). Fluorescence emissions were monitored in a multi-wavelength excitation dual-wavelength emission fluorimeter (Delta RAM, PTI). The [Ca²⁺]_out is shown as the ratio of Fura-FF/FA fluorescence (excitation 340 nm/380 nm) and ΔΨ_m as the ratio of the JC-1 fluorescence of J-aggregate (570 nm excitation/595 nm emission) and monomer (490 nm excitation/535 nm emission) forms of JC-1. The ΔΨ_m indicator, JC-1 (800 nM), a bolus of 10 μM Ca²⁺, and the mitochondrial uncoupler, FCCP (3 μM), were added at the indicated time points with constant stirring at 37 C. To measure the mitochondrial matrix Ca²⁺, hepatocytes were permeabilized in 40 mg/ml digitonin in 1.5 mL of intracellular medium supplemented with thapsigargin, Ru360 (1 μM, to block mitochondrial Ca²⁺ uptake), CGP37157 (1 μM, to block mitochondrial Ca²⁺ efflux), and Fura-FF (0.5 μM). Mitochondrial uncoupler, FCCP (10 μM) was added at indicated time points without adding any Ca²⁺ bolus to release the mitochondrial Ca²⁺ store.

Mitochondrial and cytosolic Ca²⁺ measurements using confocal live cell imaging system

Primary zebrafish cells were grown on Poly-L-lysine coated glass coverslips and loaded with _mCa²⁺ sensor Rhod-2 (2 μM) for 45 min in extracellular media (ECM-120mM NaCl, 5mM KCl, 1mM KH₂PO₄, 0.2mM MgCl₂, 0.1mM EGTA, 20mM HEPES, pH 7.4). Cells were washed and imaged in ECM using Carl Zeiss LSM510 confocal live cell imaging system using the 560 nm excitation (Ex) laser and 580 nm emission (Em) spectra is collected. Ionomycin was added at indicated time points to increase _cCa²⁺ levels. For _cCa²⁺ measurements, primary hepatocytes were grown on 25-mm glass coverslips and transduced with GCaMP6 encoding adenovirus. Coverslips were mounted in an open perfusion micro-incubator (PDMI-2; Harvard Apparatus) at 37 C and imaged using the Carl Zeiss LSM510 confocal live cell imaging system using the Ex-488 nm and Em-510 nm. After 1 min of baseline recording, vasopressin (100nM) or thapsigargin (2 μM) was added, and confocal images were obtained every 3 s at 488-nm excitation using a 40x oil objective. Images were analyzed and quantitated using ZEN 2010 and ImageJ software.

Mitochondrial respiration measurements

Primary hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were plated in Poly-L-lysine coated 96 well Seahorse culture plates. Oxygen consumption rate (OCR) was measured at 37 C in an XF96 extracellular flux analyzer (Seahorse Bioscience). Respiratory chain inhibitors were added sequentially at indicated time points. For the measurement of FAO coupled OCR, cells were glucose depleted overnight and treated with carnitine palmitoyl transferase-1 inhibitor Etomoxir (40 μM) for 1 hour to monitor endogenous fatty acid utilization. Exogenous palmitate is used as fatty acid substrate for the OCR measurement using XF96 extracellular flux analyzer (Seahorse Bioscience).

Cellular AMP/ATP measurement

Mouse hepatocyte and zebrafish total cellular ATP levels were measured using CellTiter-Glo luminescent assay kit (Promega, Madison, WI, USA) as per the manufacturer’s protocol. AMP levels were measured using the AMP-Glo luminescent assay kit (Promega, Madison, WI, USA). Luminescence was measured using a microplate reader (Infinite M1000 PRO, Tecan).

Mitochondrial reactive oxygen species measurement

Primary hepatocytes isolated from MCU^fl/fl and MCU^Δhep mice were plated on Poly-L-lysine coated glass coverslips. Next day, cells were loaded with the mitochondrial superoxide sensitive fluorophore MitoSOX Red (Life Technologies; 2 μM) and Dihydrorhodamine 123 (Rhod123; 2.5 μg/ml) in William’s E medium (without serum) at 37 C for 30 min. Cells were then washed and imaged using a Carl Zeiss 510 confocal microscope with a 40 3 oil immersion objective at 561 nm as described earlier (Mukhopadhyay et al., 2007). MitoSOX fluorescence was quantified using ImageJ software and plotted as arbitrary units in Graphpad Prism 6 software.

TMRM staining for ΔΨ_m

Mouse primary hepatocytes and zebrafish primary cells were placed on Poly-L-lysine coated glass coverslips. The next day, cells were stained with Tetramethylrhodamine, methyl ester (TMRM; 50 nM) and Dihydrorhodamine 123 (Rhod123; 2.5 μg/ml) for 30 min at 37 C. Images were acquired using a Carl Zeiss 510 confocal microscope using a 40 × oil objective at excitations of 561 nm and 488 nm, respectively. Images were quantified for TMRM fluorescence using ImageJ software and plotted as arb unit in Graphpad Prism 6 software.

Metabolic assessment of mice

The metabolic status of MCU^fl/fl and MCU^Δhep male mice were assessed using a Comprehensive Laboratory Animal Monitoring System (CLAMS; Columbus Instruments, Columbus, OH) as described earlier (Tomar et al., 2016). Mice were single housed in 12 hour light/dark cycles with a normal chow diet and water ad libitum. Mice were allowed to acclimate for 3 days and then oxygen consumption, carbon dioxide production, food intake and locomotor activity were measured continuously for 3 days (24 hr Fed, 24 hr Fasted and 24hrs Re-fed state) at an ambient temperature of 22 C. During the fasted state, the chow diet is restricted for 24 hours. The respiratory exchange ratio is the ratio of carbon dioxide production and oxygen consumption. For the oxidation of carbohydrate six molecules of oxygen is utilized with generation of six molecules of CO₂, resulting a RER of 1.0. However, for the oxidation of one fatty acid molecule 23 oxygen molecules are utilized with a generation of 16 CO₂ molecules, resulting in a RER of 0.7. Therefore, we can calculate the percentage of fat utilization by using following equation:

$$\text{\% Fat Utilized = }((1.00-\text{RER)/(1.00–0.70))}\ast \text{100}$$

Animal body composition, liver/plasma biochemistry, glucose tolerance test and glucose output assay

Normal chow diet fed 10 weeks old male mice used for body fat composition analysis by nuclear magnetic resonance (NMR) and dual emission X-ray absorptiometry (DEXA) (Echo Medical Systems, Houston, TX). Enzymatic colorimetric assay was performed to monitor liver/plasma triglyceride, plasma ketones, liver glycogen, and blood glucose as described earlier (Alenghat et al., 2008; Tomar et al., 2016). For the glucose tolerance test, 12 week old male mice were fasted overnight for approximately 16 hours by transferring mice to clean cages. Next day, mouse body weight (wt) was measured and 20% (w/v) glucose solution (2g of glucose/kg body mass) was injected intraperitoneally. The blood glucose levels were measured at the indicated time points through the tail puncture using the OneTouch Ultra Glucose Meter. The glucose output assay was performed on primary hepatocytes isolated from 12-weeks old MCU^fl/fl and MCU^Δhep mice. The freshly isolated hepatocytes were maintained in Williams E medium containing 1% (v/v) antibiotic-antimycotic solution (GIBCO), 1% (v/v) 200mM L-glutamine, 1% (v/v) non-essential amino acids, and 10% heat inactivated fetal bovine serum overnight. The following day, the media was changed to glucose output media (GOM: 118 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 1.2mM CaCl₂, 20mM NaHCO₃, 20mM HEPES pH 7.4, and BSA 0.025%) for 2h before induction of glycogenolysis. The cells were stimulated with 5nM Glucagon in GOM for 2h, and culture supernatants were collected and glucose content was measured by Glucose (HK) Assay Kit.

Lipid staining in primary hepatocytes, liver sections and whole zebrafish

Primary mice hepatocytes were isolated from MCU^fl/fl and MCU^Δhep and cultured in Poly-L-lysine coated glass coverslips with Williams E medium containing 5mM Glucose (low glucose condition) or 25mM Glucose (high glucose condition) for 24 hours. The next day, the medium for the hepatocytes was replaced with Earle’s Balanced Salt Solution (EBSS) to induce starvation for 2 hours. After incubation, cells were loaded with BODIPY^® 493/503 (1 μg/ml) in serum-free medium for 30 min. Cells were washed and images were acquired using a Carl Zeiss 510 confocal microscope using a 40 × oil objective at 488 nm excitation. Images were quantified for the size and number of lipid droplets using ImageJ software. For the histological evaluation of hepatic lipid accumulation, mice liver was perfused with PBS and subsequently with 4% paraformaldehyde to fix the tissue. Liver tissue was embedded in paraffin to cut the 5 μM thin tissue sections and subsequently stained for Oil Red O and hematoxylin and eosin (H&E). Histological images were acquired in bright field light microscope (Nikon, USA). Zebrafish lipid content was visualized by staining with Lipid Green. Wild-type and MCU-KO zebrafish were incubated in 100 μM of Lipid Green for 15 min (Lee et al., 2011). After incubation, zebrafish were anesthetized in tricaine containing water and in vivo confocal microscopy was performed using the Carl Zeiss LSM710 two-photon confocal system.

Electron microscopy to visualize hepatic ultrastructure

MCU^fl/fl and MCU^Δhep mice were anesthetized, and the liver was perfused with PBS for 2 minutes at 35 C followed by 0.15 M cacodylate buffer (Ted Pella Inc., Redding, CA) pH7.4 containing 2.5% glutaraldehyde (Electron Microscopy Sciences, Hartfield, PA), 2% formaldehyde (fresh from paraformaldehyde (EMS)) with 2mM CaCl₂ and 20 mM KCl at 35 C for 5 minutes. Liver lobes from the animal were removed and fixed for an additional 2–3 h on ice in the same tissue fixing solution. Liver tissue was sectioned, and images were acquired using the Zeiss LIBRA120 TEM equipped with Gatan UltraScan, 1000 2k × 2k CCD EFTEM, and energy filtering.

Immunoblotting and co-immunoprecipitation assay

Immunoblotting for the detection of protein expression and phosphorylation was performed using the equal amount of total cell lysate prepared from primary mice hepatocytes or zebrafish. Proteins were extracted from primary hepatocytes isolated from mice liver or zebrafish using RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.25% deoxycholic acid, 1 mM EDTA, 1% NP-40 supplemented with protease inhibitor cocktail (Roche) and PhosSTOP Phosphatase Inhibitor Cocktail (Roche)). Protein concentrations were quantified using Pierce™ 660nm Protein Assay (Thermo Scientific Inc.) and equal amounts of protein were separated on 4%–12% Bis-Tris polyacrylamide gel. Proteins were transferred to a PVDF membrane, and probed with indicated antibodies. For co-immunoprecipitation, proteins were extracted from HepG2 cells transiently transfected with AMPK-HA and/or PP4-Flag using RIPA buffer. Indicated antibodies were used to pull-down the specific protein. The input (10% of total cell lysates) and immunoprecipitates were probed with specified antibody.

Global lipidomic profiling

Preparation of samples for lipidomic analysis

To monitor global lipidomics alterations in MCU^Δhep mice, 12-week-old mice were used. MCU^Δhep mice were fed with Metformin (1.25mg/ml metformin in drinking water) for three weeks. Mice liver were harvested and frozen in liquid nitrogen until the lipids were extracted. Liver sections were subjected to homogenization to uniformity in ice cold phosphate buffered saline (PBS) such that a final suspension of 10% w/v liver homogenate in PBS were obtained. 50 μL of this suspension was used in the extraction and analysis of liver lipids.

Extraction of lipids from liver tissues

50 μL of liver homogenate was diluted with 150 μL of PBS. To the mixture thus obtained, a mixture of internal standards were added followed by vortexing and incubating on ice for 10 minutes for equilibration of the IS with the sample. Following incubation, 1 mL of methanol was added to the sample followed by sonication to mix the samples. Thereafter 0.5 mL of chloroform was added and the samples were sonicated again to mix the solvents to obtain a well dispersed monophasic extraction mixture. This mixture was then incubated at 48 C for one hour to optimize the extraction of highly non polar lipids. Following incubation, a phase break implemented out by the addition of 1 mL of chloroform followed by 2mL of water. The resultant mixture was vortexed for 30 s and then centrifuged at 3500 g to enable phase separation. Following centrifugation, the bottom organic layer containing the lipids was transferred to a fresh tube. The top aqueous layer was washed with an additional 1 mL of CHCl₃, vortexed, centrifuged and the separated organic layer was combined with the previous organic fraction. The lipid extracts thus obtained was dried via vacuum centrifugation and re-suspended in 200 mL of methanol. The re-suspended solution was incubated at 48 C for 10 minutes vortexed, and 50 μL transferred to 200 μL autosampler vials for analysis via ultra-performance liquid chromatography high resolution mass spectrometry.

Analysis of lipids by untargeted mass spectrometry

5 μL of the lipid extract prepared as detailed above was analyzed following separation on a Acquity CSH C18 reverse phase column (100Å_, 3.5 μm, 3 mm × 150 mm, Waters) using a Shimadzu Nexera UPLC chromatographic system coupled to Sciex TripleTOF 5600+ quadrupole time of flight mass analyzer. Data was acquired in the positive mode in an untargeted data dependent manner in the mass range of 50–1200 Da. A 28 minute chromatographic separation utilizing a linear gradient of 60:40 acetonitrile:water with 10mM ammonium formate and 0.1% formic acid as mobile phase A and 90:10 isopropanol:acetonitrile with 10mM ammonium formate and 0.1% formic acid as mobile phase B.

Data Analysis

The acquired data was subjected to post processing via MarkerView software (Sciex LLC). Following normalization to total ion current and alignment of retention times, features were extracted as m/z retention time pairs. The extracted features were subjected statistical analysis using Metaboanalyst. The data was first subjected to log transformation followed by centering using Pareto scaling (mean-centered and divided by the square of the standard deviation of each variable). Cluster analysis was performed using Euclidean distance measure and Ward clustering algorithm. ANOVA and Tukey’s HSD post hoc analysis with an FDR = 0.001 threshold was used to identify the top feature differences between the three groups. Those features identified as being significantly different were then identified via comparison against Lipidmaps database as well as the elution retention time window for that lipid class.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data were expressed as the mean ± SE and statistical significance was evaluated via Student’s unpaired t test, one-way and twoway ANOVA with Tukey’s HSD post hoc analysis, where appropriate. p < 0.05 was considered statistically significant. All experiments were conducted at least three times unless specified. Replicates and statistical analysis information is reported in the figure legends. Data were plotted either with Sigma Plot 11.0 software or GraphPad Prism version 6 software.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit Polyclonal anti- ACC	Cell Signaling Technology	Cat# 3662
Rabbit Polyclonal anti- AMPKα	Cell Signaling Technology	Cat# 2532
Rabbit Polyclonal anti- ATF-6α	Santa Cruz Biotechnology Inc.	Cat# sc-166659
Mouse monoclonal anti- ATP5A	Abeam	Cat# ab14748
Rabbit monoclonal anti- CaMKII-α (D10C11)	Cell Signaling Technology	Cat# 11945
Rabbit Polyclonal anti- Carnitine Palmitoyl Transferase-1	Alpha Diagnostic Intl. Inc.	Cat# CPT1M11-A
Mouse monoclonal anti- ERK1/ERK2	Thermo Scientific Pierce	Cat# MA5–15605
Rabbit monoclonal anti- Fatty Acid Synthase (C20G5)	Cell Signaling Technology	Cat# 3180
Mouse monoclonal anti- FLAG M2-Peroxidase (HRP)	Sigma-Aid rich	Cat# A8592–1MG
Mouse monoclonal anti- FLAG M2	Sigma-Aid rich	Cat# F1804–1MG
Rat monoclonal anti- HA-Peroxidase	Sigma-Aid rich	Cat# 12013819001
Rabbit monoclonal anti- LKB1 (D60C5)	Cell Signaling Technology	Cat# 3047
Rabbit Polyclonal anti- MCU	Sigma-Aid rich	Cat# HPA016480
Rabbit Polyclonal anti- MCUR1	Tomar et al., 2016	N/A
Rabbit Polyclonal anti- MICU1	Tomar et al., 2016	N/A
Rabbit Polyclonal anti- mtTFA	Abeam	Cat# ab138351
Rabbit Polyclonal anti- NRF1	Cell Signaling Technology	Cat# 12381
Mouse monoclonal Total OXPHOS Rodent WB Antibody Cocktail	Abeam	Cat# ab110413
Rabbit Polyclonal anti- Phospho- ACC (Ser79)	Cell Signaling Technology	Cat# 3661
Rabbit Polyclonal anti- Phospho-AMPKα (Thr172)	Cell Signaling Technology	Cat# 2531
Rabbit monoclonal anti- Phospho-CaMKII (Thr286) (D21E4)	Cell Signaling Technology	Cat# 12716S
Rabbit Polyclonal anti- Phospho-p44/42 MAPK (Erk½) (Thr202/Tyr204)	Cell Signaling Technology	Cat# 9101
Rabbit monoclonal anti- Phospho-PERK (Thr980) (16F8)	Cell Signaling Technology	Cat# 3179
Rabbit Polyclonal anti- PGC1 alpha	Abeam	Cat# ab54481
Mouse monoclonal anti- PPX (C-6)	Santa Cruz Biotechnology Inc.	Cat# sc-374106
Rabbit Polyclonal anti- SREBP-1 (C-20)	Santa Cruz Biotechnology Inc.	Cat# sc-366
Rabbit Polyclonal anti- Tom20 (FL-145)	Santa Cruz Biotechnology Inc.	Cat# sc11415
Rabbit monoclonal anti- XBP-1 s (D2C1 F)	Cell Signaling Technology	Cat# 12782
Mouse monoclonal anti- β-Actin (C4)	Santa Cruz Biotechnology Inc.	Cat# sc47778
Goat anti-Mouse IgG (H+L)-HRP Conjugate	Bio-Rad Laboratories	Cat# 1721011
Goat anti-rabbit IgG, HRP-linked	Cell Signaling Technology	Cat# 7074S
Chemicals, Peptides, and Recombinant Proteins
Acetonitrile	Sigma-Aid rich	Cat# 271004
Ammonium formate	Sigma-Aid rich	Cat# 516961
Antibiotic-antimycotic solution	Thermo Fisher Scientific	Cat# 15240112
Antimycin A	Sigma-Aid rich	Cat# A8674
BAPTA-AM	Thermo Fisher Scientific	Cat# B6769
Bis-Tris polyacrylamide gel (4–12%)	Thermo Fisher Scientific	Cat # WG1402BOX
BODIPY® 493/503	Thermo Fisher Scientific	Cat# D3922
CaCI2	Sigma-Aldrich	Cat# 21115
Cacodylate buffer	Electron Microscopy Sciences	Cat# 11652
CGP37157	Sigma-Aldrich	Cat# C8874
Chloroform	Sigma-Aldrich	Cat# CX1055
Collagenase D	Sigma-Aldrich	Cat# 11088866001
Digitonin	Sigma-Aldrich	Cat# D141
Dihydrorhodamine 123	Thermo Fisher Scientific	Cat# D632
DPBS Ca2+/Mg2+ free	Thermo Fisher Scientific	Cat# 14190136
DPBS with Ca2+/Mg2+	Thermo Fisher Scientific	Cat# 14040117
Dulbecco’s modified Eagle’s medium (DMEM)	Thermo Fisher Scientific	Cat# 11885092
Earie’s Balanced Salt Solution (EBSS)	Thermo Fisher Scientific	Cat# 24010043
EDTA	Thermo Fisher Scientific	Cat# AM9260G
EGTA	Sigma-Aldrich	Cat# 03777
Etomoxir	Sigma-Aldrich	Cat# 236020
FCCP	Sigma-Aldrich	Cat# C2920
Fetal bovine serum	HyClone-GE	Cat# SH30071.04
Formic acid	Sigma-Aldrich	Cat# 33015
Fura-FF	Sigma-Aldrich	Cat# 17085
Glucose	Sigma-Aldrich	Cat# G8644–100ML
Glucagon	Sigma-Aldrich	Cat# G2044
Glutaraldehyde	Electron Microscopy Sciences	Cat# 16020
HEPES	Thermo Fisher Scientific	Cat# 15630080
lonomycin	Thermo Fisher Scientific	Cat# I24222
JC-1	Thermo Fisher Scientific	Cat# T3168
KCl	Sigma-Aldrich	Cat# P9333
KH2PO4	Sigma-Aldrich	Cat# P5655
L-glutamine	Thermo Fisher Scientific	Cat# 25030164
MgCl2	Sigma-Aldrich	Cat# 449172
MitoSOX Red	Thermo Fisher Scientific	Cat# M 36008
NaCI	Sigma-Aldrich	Cat# 746398
Non-essential amino acids	Thermo Fisher Scientific	Cat# 11140050
Oil Red O	Sigma-Aldrich	Cat# O0625
Oligomycin	Sigma-Aldrich	Cat# 75351
Paraformaldehyde	Electron Microscopy Sciences	Cat# 15700
PhosSTOP Phosphatase Inhibitor Cocktail	Sigma-Aldrich	Cat# 4906845001
Poly-L-lysine	Sigma-Aldrich	Cat# P4707–50ML
complete Protease Inhibitor Cocktail	Sigma-Aldrich	Cat# 11697498001
Rhod-2-AM	Thermo Fisher Scientific	Cat# R1245MP
RIPA buffer	EMD-Millipore	Cat# 20–188
Rotenone	Sigma-Aldrich	Cat# 557368
Ru360	Sigma-Aldrich	Cat# 557440
Sodium gluconate	Sigma-Aldrich	Cat# S2054
Sucrose	Sigma-Aldrich	Cat# 84097
Tetramethylrhodamine methyl ester	Thermo Fisher Scientific	Cat# T668
Thapsigargin	Thermo Fisher Scientific	Cat# T7458
Tricaine	Sigma-Aldrich	Cat# E10521
Vasopressin	Sigma-Aldrich	Cat# 1711100
William’s E medium	Thermo Fisher Scientific	Cat# 32551020
Critical Commercial Assays
Pierce™ 660nm Protein Assay	Thermo Fisher Scientific	Cat# 22662
CellTiter-Glo Assay	Promega	Cat# G7571
AMP Glo Assay	Promega	Cat# V5011
ALT Activity Assay	Sigma-Aid rich	Cat# MAK052
β-hydroxybutyrate Assay	Cayman	Cat# 700190
Glucose (HK) Assay Kit	Sigma-Aid rich	Cat# GAHK20
Triglyceride Assay	Cayman	Cat# 10010303
Experimental Models: Cell Lines
HepG2	ATCC	Cat# HB-8065
Primary Mouse Hepatocytes	This Study	N/A
Experimental Models: Organisms/Strains
C57BL/6J Mice	The Jackson Laboratory, USA	Stock# 000664
MCUfl/fl mice	Luongo et al., 2015	N/A
B6.Cg-Tg(Alb-cre)21 Mgn/J	The Jackson Laboratory, USA	Stock# 003574
MCUΔhep mice	This Study	N/A
AMPKα1/α2fl/fl mice	Foretz et al., 2010	N/A
AMPK α1/α2Δhepmice	This Study	N/A
MCU-C96A-KI mice	This Study	N/A
MCU-KO zebrafish	This Study	N/A
Oligonucleotides
PP4-siRNA (Smart Pool)	Dharm aeon	Cat# L-040058–02-0005
Recombinant DNA
Ad-T172DAMPKα1 (CA-AMPK)	This Study	N/A
Ad-MCU-Flag	This Study	N/A
Ad-GCaMP6.m	This Study	N/A
pECE HA AMPKα1 AS	Addgene	Plasmid # 69831
pEBG-T172D AMPKaα1	King et al., 2009	N/A
PP4-Flag	OriGene Technologies Inc.	Cat# MR2 04327
Software and Algorithms
Fiji ImageJ software	Fiji ImageJ	https://fiji.sc/
Sigma Plot 11.0	Systat Software Inc.	https://systatsoftware.com/products/sigmaplot/
Zen 2010	Carl Zeiss	https://www.zeiss.com/microscopy/us/products/microscope-software/zen.html
Graphpad Prism 6	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
Metaboanalyst	Xia Lab at McGill Univeristy	https://www.metaboanalyst.ca/
MarkerView software	Sciex LLC	https://sciex.com/products/software/markerview-software
Canvas 11.0	ACD Systems	https://www.canvasgfx.com/en/support/canvas-11/

Highlights.

• Mitochondrial Ca²⁺ powers FAO-dependent hepatocyte mitochondrial respiration

• Hepatic MCU deletion promotes lipid accumulation and lowers ketone bodies

• Blockade of _mCa²⁺ buffering enhances AMPK dephosphorylation through PP4

• Restoration of AMPK activity in MCU^Δhep model improves lipid clearance