Sirtuin 3 (SIRT3) Regulates α-Smooth Muscle Actin (α-SMA) Production through the Succinate Dehydrogenase-G Protein-coupled Receptor 91 (GPR91) Pathway in Hepatic Stellate Cells

Abstract

Sirtuin 3 (SIRT3) is an NAD⁺-dependent protein deacetylase. Recent studies have shown that SIRT3 expression is decreased in nonalcoholic fatty liver disease (NAFLD). Moreover, SIRT3 is a key regulator of succinate dehydrogenase (SDH), which catalyzes the oxidation of succinate to fumarate. Increased succinate concentrations and the specific G protein-coupled receptor 91 (GPR91) are involved in the activation of hepatic stellate cells (HSCs). In this study, we aimed to establish whether SIRT3 regulated the SDH activity, succinate, and GPR91 expression in HSCs and an animal model of NAFLD. Our goal was also to determine whether succinate released from hepatocytes regulated HSC activation. Inhibiting SIRT3 using SIRT3 siRNA exacerbated HSC activation via the SDH-succinate-GPR91 pathway, and SIRT3 overexpression or honokiol treatment attenuated HSC activation in vitro. In isolated liver and HSCs from methionine- and choline-deficient (MCD) diet-induced NAFLD, the expression of SIRT3 and SDH activity was decreased, and the succinate concentrations and GPR91 expression were increased. Moreover, we found that GPR91 knockdown or resveratrol treatment improved the steatosis in MCD diet-fed mice. This investigation revealed a novel mechanism of the SIRT3-SDH-GPR91 cascade in MCD diet-induced HSC activation in NAFLD. These findings highlight the biological significance of novel strategies aimed at targeting SIRT3 and GPR91 in HSCs with the goal of improving NAFLD treatment.

Introduction

Nonalcoholic fatty liver disease (NAFLD)² is the most common chronic liver disease in many developed countries (1), and nonalcoholic steatohepatitis (NASH), the more severe histological form of NAFLD, is associated with an increased risk for the progression to cirrhosis in 20% of these patients (2). NAFLD also increases the cardiometabolic risk (3–5) and all-cause mortality (6, 7) in humans. It is presently regarded as the main cause of cryptogenic liver cirrhosis in the United States (8). During liver injury, quiescent hepatic stellate cells (HSCs) transdifferentiate into activated myofibroblasts, which produce α-smooth muscle actin (α-SMA) and become a major cell type in hepatic fibrogenesis (9, 10).

Sirtuin 3 (SIRT3) is an NAD⁺-dependent protein deacetylase predominantly localized in the mitochondrial matrix (11–13). SIRT3 is up-regulated during prolonged fasting or a calorie-restricted diet and is thus involved in the metabolic regulation of obesity and diabetes (14–16). Based on several recent studies, SIRT3 is a primary regulator of the acetylation of mitochondrial proteins and their biological activity (16–19) and is associated with NAFLD (20–22).

Two studies yielded findings showing that SIRT 3 is a major physiological regulator of succinate dehydrogenase (SDH) activity (23, 24). SDH catalyzes the oxidation of succinate to fumarate, thereby decreasing SDH activity, resulting in increased succinate levels (25, 26). The succinate receptor (also known as GPR91) is a G protein-coupled receptor expressed in various tissues, including the retina, liver, and kidneys (27–31). Locally increased succinate levels and GPR91 activation have recently emerged as novel signaling molecules in local stress situations (25).

In a previous study, we showed that decreased SDH activity led to increased cellular succinate levels and succinate receptor (GPR91) overexpression with increased α-SMA production in the isolated HSCs of MCD diet-induced NASH mice (32). These observations led us to question whether SIRT3 expression could modulate HSC activation through SIRT3-SDH-GPR91 signaling in NASH. To the best of our knowledge, the role of SIRT3 in the regulation of HSC activation has not been fully characterized. In this study, we evaluated the effects of SIRT3 on GPR91 regulation through SDH to mitigate the progression of NASH in HSCs and an animal model, and we determined whether succinate secreted from hepatocytes regulated HSC activation.

Experimental Procedures

Materials

Overexpression of α-SMA, a hallmark of myofibroblastic trans differentiation, was used as a marker for HSC activation (33, 34). DMEM completely deficient in methionine and choline (MCD medium) and a methionine and choline supplement (MCS medium, control medium) were purchased from Welgene (Kyeongsan, Korea). Palmitate was purchased from Sigma. AAV-GPR91 shRNA (Vector Biolabs, Philadelphia, PA) or AAV6-GFP shRNA (Vector Biolabs) was used for viral production.

Cell Culture

LX2 cells are immortalized human stellate cells and were provided by Prof. Ja June Jang (Seoul National University). The cells were cultured in DMEM with 10% FBS supplemented with 1% penicillin/streptomycin antibiotic solution. AML12 cells were cultured in DMEM F12 medium (Welgene) supplemented with 10% FBS and 1% penicillin/streptomycin antibiotic solution. Cells were maintained in a humidified 37 °C incubator with 5% CO₂.

Western Blotting Analysis

Whole cells were lysed in radioimmunoprecipitation assay buffer containing 25 mm Tris-HCL (pH 7.6), 150 mm NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, 1% SDS, and protease inhibitor mixture (Roche Diagnostics) on ice. Equal amounts of proteins were resolved on SDS/PAGE and then electrotransferred onto PVDF membranes and blocked with 5% nonfat dry milk for 30 min at room temperature. Levels of proteins were determined by incubation with primary antibodies at appropriate dilutions. Primary antibodies included those specific to GPR91 (sc-50466, Santa Cruz Biotechnology, Santa Cruz, CA), ERK1/ERK2 (MAB1576, R&D Systems, Minneapolis, MN), phospho-ERK1/ERK2 (AF1018, R&D Systems), SIRT3 (2627, Cell Signaling, Danvers, MA), α-SMA (GTX112861, GeneTex), and GAPDH (GTX627408, GeneTex). The bound antibodies were further incubated with secondary antibodies conjugated to HRP, detected with the Westsave Star detection reagent system (AbFrontier, Seoul, Korea).

SIRT3 siRNA Transfection and Adenoviral Transfection of SIRT3

LX2 cells grown in the exponential phase were seeded in a 6-well plate and then transfected with 100 nm SIRT3-targeted siRNA (sc-61555, Santa Cruz Biotechnology) or nontargeting RNA for 6 h using Lipofectamine RNAiMax (Invitrogen) according to the instructions of the manufacturer. Adenovirus transfection was performed when human LX2 cells reached 30–40% confluence in a 6-well plate. Before transfection, LX2 cells were changed using either control or MCD medium and then infected with the human SIRT3 adenovirus (Ad-SIRT3, 1499, Vector Biolabs) or a control adenovirus expressing LacZ (Ad-CMV-β-gal, 1080, Vector Biolabs) at an MOI of 30. 24 h post-infection, cells were lysed and subjected to Western blotting. Some LX2 cells were infected with human Ad-SIRT3 or control adenovirus expressing LacZ (Ad-LacZ) at an MOI of 30. 8 h post-transfection, cells were stimulated with palmitate for 20 h, after which total proteins were extracted.

Description of Animals and Isolation of HSCs and Hepatocytes

Male C57BJ6 mice, 6–8 weeks old and weighing 18–20 g, were purchased from Central Animal Laboratory. All mice were housed at ambient temperature (22 ± 1 °C) with a 12/12-h light/dark cycle and free access to water and food. The mice were fed a methionine- and choline-deficient diet (MCD diet group) as an animal model of NAFLD or control chow diet (control group) for 4 weeks. All mice were fed their assigned MCD diet for 4 weeks, and the AAV6-GFP shRNA (4 × 10¹¹ pfu, n = 4) or AAV-GPR91 shRNA (4 × 10¹¹ pfu, n = 8) were injected via the tail vein on the first day of MCD diet feeding. In another group, mice received 100 mg/kg/day of resveratrol daily with the MCD diet for 4 weeks (n = 8), and C57BL/6 mice on a regular chow diet were used as the control group (WT control, n = 4). Primary mouse HSCs and hepatocytes were isolated from the livers of mice (10–12 weeks old) by in situ Pronase E and collagenase B perfusion, followed by density gradient centrifugation. Primary cells were >95% pure. Cells were grown in standard tissue culture plastic dishes in DMEM with 10% FBS and antibiotics. Primary cells were incubated at 37 °C and used 3 days after plating.

Succinate Dehydrogenase Assay and Succinate Assay

The SDH assay was performed using the ab109908-Complex II enzyme activity microplate assay kit (Biovision, Milpitas, CA). Cell lysate (5 μl) was added to a mixture containing SDH assay buffer, SDH substrate mix, and SDH probe. Absorbance readings at 600 nm were taken every 20 s for a total of 60 min. The data are expressed as mean optical density (mOD)·min⁻¹. The level of cellular succinate was determined with a succinate colorimetric assay kit (BioVision). Succinate levels were read at 450 nm, with each measurement performed in triplicate.

Deacetylation Assay

SDH subunit A (SDHA) was immunoprecipitated from the total cell lysate with SDHA (sc-166909, Santa Cruz Biotechnology) antibody. Then Western blots were probed with anti-acetylated lysine antibody (9441S, Cell Signaling Technology).

RT-PCR

Total RNA was extracted with the RNeasy mini kit (Qiagen, Hilden, Germany). Primers were designed as follows: SIRT3, 5-CGT CAC TCA CTA CTT TCT CC-3 and (5-ACC ACA TGC AGC AAG AAC CT-3; GPR91, 5-GCA TGT GTC TAA CAC TGT TG-3 and 5-CTT CTG TCC CAA CTA CTG TG-3; α-SMA, 5-CCA CCG CAA ATG CTT CTA AGT-3 and 5-GGC AGG AAT GAT TTG GAA AGG-3; TGF-β1, 5-TCG ACA TGG AGC TGG TGA AA-3 and 5-GAG CCT TAG TTT GGA AGA TCT G-3; collagen a1, 5-GAA CGC GTG TCA TCC CTT GT-3 and 5-GAA CGA GGT AGT CTT TCA GCA ACA-3; and GAPDH, 5-GGC ATG GAC TGT GGT CAT GAG-3 and 5-TGC ACC ACC AAC TGC TTA GC-3. cDNA was synthesized by reverse transcription with the PrimeScript first strand cDNA synthesis kit (Takara Bio, Shiga, Japan) and amplified by PCR with Maxima SYBR Green/ROX qPCR Master Mix (2×) (Thermo) using standard protocols. All amplified products were assessed with 2% agarose gel electrophoresis and photographed using ultraviolet illumination.

Hepatic Triglyceride Measurement

Triglyceride contents in the liver were determined using the triglyceride reagent kit (T2449, Sigma) as described by the manufacturer.

Histological Analysis and Immunohistochemistry

Samples of mouse liver were fixed in 10% (w/v) phosphate-buffered formalin for 18–20 h. After dehydration through a graded series of ethanol solutions, the tissues were embedded in paraffin wax. Serial frontal sections were cut at 5-μm intervals and stained with H&E and Masson trichrome. The obtained 5-μm sections were deparaffinized in xylene and rehydrated through a graded ethanol series into water. Endogenous peroxidase activity was blocked in 3% H₂O₂. The slides were subsequently placed on a Dako Autostainer immunostaining system for use in immunohistochemistry analyses using polyclonal antibodies against GPR91(1:100, sc-50466, Santa Cruz Biotechnology), SIRT3 (1:200, 2627, Cell Signaling Technology), and α-SMA (1:500, GTX112861, GeneTex) and incubated for 5 h. In the next step, the slides were blocked for endogenous biotin with an avidin-biotin blocking system (Dako, X0590). Labeled streptavidin-biotin complex plus (Dako, K0675) served as the detection system, and hematoxylin was used for counterstaining.

Immunofluorescence Microscopy

To locate the mitochondria, heat shock protein 60 (HSP 60) was used to co-label the samples. For double staining of SIRT3 and HSP 60, sections were stained with primary antibodies (both 1:200, sc-49744 and sc-13966, respectively, Santa Cruz Biotechnology) and then with goat anti-rabbit FITC-conjugated secondary antibody (1:200, 31635, Invitrogen) and donkey anti-goat Cy3-conjugated secondary antibody (1:200, ab6949, Abcam, Cambridge MA) for 1 h. Hoechst 33342 staining for 10 min was used to evaluate chromosomes. After three washes, liver samples were mounted in anti-fade medium (Vectashield, Burlingame, CA) and examined under a laser-scanning confocal microscope (LSM 710, Zeiss).

Statistical Analyses

All data are expressed as mean ± S.E. from at least three independent experiments, and GAPDH was used as a loading control. Data analyses for the two groups were performed using a t test, with p < 0.05 indicating statistical significance.

Results

Succinate, SDH Inhibitor and Fumarase Inhibitor Activates HSCs

To investigate the role of succinate in HSC activation, we treated HSCs with succinate and demonstrated that succinate itself increased the expression of GPR91, ERK phosphorylation, and α-SMA production in HSCs (Fig. 1A). However, pretreatment with 10 μm U0126 (ERK inhibitor) significantly blocked the succinate-induced up-regulation of GPR91 and α-SMA expression in LX2 cells. These findings suggest that the ERK pathway is downstream of the succinate pathway. Treatment with malonate, a known SDH inhibitor, showed increased expression of GPR91 and α-SMA production in HSCs (Fig. 1B) and decreased SDH activity and increased succinate concentrations (Fig. 1, C and D). Treatment with D-malate, a fumarase inhibitor, increased GPR91 and α-SMA production in HSCs (Fig. 1E), decreased fumarase and SDH activity (Fig. 1, F and G), and increased succinate concentrations (Fig. 1H).

FIGURE 1.

Succinate, succinate dehydrogenase activity, and expression of GPR91 and α-SMA in LX2 cells. A, LX2 cells were cultured in the presence or absence of succinate (400 μm) for 24 h, and U1026 was added 2 h before harvest. Whole-cell lysates were subjected to Western blotting analyses with the indicated antibodies (left panel). Band intensities were calculated using ImageJ software (National Institutes of Health) (right panel). ***, p < 0.001 versus control. B, LX2 cells were cultured in the presence or absence of malonate (3 mm) for 24 h, and GPR91 and a-SMA protein levels were analyzed using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. C, LX2 cells were cultured in the control or malonate (3 mm) for 24 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control. D, LX2 cells were cultured in the presence or absence of malonate (3 mm) for 24 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control. E, LX2 cells were cultured in the presence or absence of D-malate (1 mm) for 24 h, and GPR91 and a-SMA protein levels were analyzed using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. F, LX2 cells were cultured in the presence or absence of D-malate (1 mm) for 24 h, and fumarase activity was measured in whole-cell lysates. ***, p < 0.001 versus control. G, LX2 cells were cultured in the presence or absence of D-malate (1 mm) for 24 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control. H, LX2 cells were cultured in the presence or absence of D-malate (1 mm) for 24 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control.

SIRT3 siRNA Transfection Activates HSCs

To investigate the role of SIRT3 in HSC activation, we used siRNA to deplete SIRT3 in HSCs. SIRT3 siRNA transfection increased the mRNA levels of GPR91, α-SMA, TGF-β1, and collagen type 1 (Fig. 2, A and B). SIRT3 siRNA transfection increased the expression of the mRNA levels of GPR91, suggesting a negative association of SIRT3 and GPR91 in HSCs activation.

FIGURE 2.

Up-regulation of GPR91, α-SMA, TGF β1, and collagen type 1 in LX2 cells by siRNA-mediated knockdown of SIRT3. A, RT-PCR analysis of SIRT3, GPR91, α-SMA, TGF-β1, and collagen type I mRNA expression in HSCs transfected with SIRT3 siRNA or control siRNA. B, quantitative analysis of the band density. Each column represents mean ± S.E. **, p < 0.01; ***, p < 0.001 (versus control).

SIRT3 Regulates SDH Activity and GPR91 Expression in HSCs

LX2 cells treated with SIRT3 siRNA for 24 h demonstrated decreased expression of SIRT3 and increased protein expression of GPR91 and α-SMA compared with control siRNA treatment (Fig. 3A). Moreover, LX2 cells treated with SIRT3 siRNA for 24 h demonstrated decreased SDH activity and increased succinate concentrations (Fig. 3, B and C). SIRT3 siRNA treatment in LX2 cells induced decreased SIRT3 expression and increased acetylation of SDHA in immunoprecipitations (Fig. 3D). LX2 cells treated with adenoviral transfection of SIRT3 overexpression for 24 h showed increased expression of SIRT3 and decreased expression of GPR91 and α-SMA compared with a control adenoviral transfection (Fig. 3E). Additionally, LX2 cells transfected with adenoviral SIRT3 overexpression showed increased SDH activity and decreased succinate concentrations in HSC lysates (Fig. 3, F and G). SIRT3 overexpression in LX2 cells increased SIRT3 production and decreased acetylation of SDHA in immunoprecipitation (Fig. 3H). Therefore, SIRT3 regulates SDH activity and cellular succinate levels, leading to HSC deactivation through inhibition of GPR91. Both SIRT3 silencing and overexpression modulate MAP kinase, as evidenced by the SIRT3 siRNA-induced increase in phosphorylation of ERK (Fig. 3A) or the Ad-SIRT3-induced decrease in ERK phosphorylation (Fig. 3E).

FIGURE 3.

SIRT3 regulation of the SDH-GPR91-α-SMA signaling pathway. A, Western blotting analysis of SIRT3, GPR91, α-SMA, TGF-β1, and collagen type I protein expression in LX2 cells with or without SIRT3 knockdown. The level was normalized to the expression of GAPDH (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control siRNA. B, LX2 cells were transfected with SIRT3 siRNA or control siRNA for 24 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control siRNA. C, LX2 cells were transfected with SIRT3 siRNA or control siRNA for 24 h, and succinate concentration (conc.) was measured in whole-cell lysates. ***, p < 0.001 versus control siRNA. D, LX2 cells were transfected with SIRT3 siRNA or control siRNA for 24 h. Then the LX2 cells were immunoprecipitated (IP) with monoclonal antibodies against SDHA. SDHA acetylation was examined using immunoblotting (left panel). Band intensities were calculated using ImageJ software (right panel). ***, p < 0.001 versus control siRNA. E, overexpression of SIRT3 by transfection with a human SIRT3 adenovirus (Ad-SIRT3) or a control adenovirus expressing LacZ (Ad-CMV-β-gal) at an MOI of 30. After 24 h, infected LX2 cells were lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control adenovirus. F, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. After 24 h, infected cells were lysed, and SDH activity was measured. ***, p < 0.001 versus control adenovirus. G, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at a MOI of 30. After 24 h, infected cells were lysed and succinate concentrations were measured. ***p < 0.001, versus control adenovirus. H, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. After 24 h, infected cells were immunoprecipitated with monoclonal antibodies against SDHA. SDHA acetylation was examined using immunoblotting (left panel). Band intensities were calculated using ImageJ software (right). ***, p < 0.001 versus control adenovirus.

Palmitate and MCD Medium Treatment on SIRT3, Succinate, and GPR91 on HSCs

We tested whether SIRT3 expression was affected by palmitate or MCD medium treatment in HSCs. When LX2 cells were incubated with palmitate for 20 h, this led to decreased protein expression of SIRT3 and increased expression of GPR91 and α-SMA (Fig. 4A) as well as decreased mRNA expression of SIRT3 in cell lysates compared with control treatment (Fig. 4B). We used MCD medium instead of the MCD diet, which is a widely used technique to create an animal model of NASH, to investigate the influence of MCD medium on the SIRT3, SDH, and GPR91 pathway in HSCs. We cultured LX2 cells in MCD medium for 24 h and monitored SIRT3, GPR91, and α-SMA expression in LX2 cells after treatment. LX2 cells incubated with MCD medium for 24 h showed decreased protein expression of SIRT3 and increased expression of GPR91 and α-SMA (Fig. 4C). Additionally, LX2 cells incubated with MCD medium for 24 h demonstrated decreased mRNA expression of SIRT3 in HSC lysates compared with control treatment (Fig. 4D). These results indicate that palmitate and MCD medium can induce activation of HSCs through SIRT3 deactivation and GPR91 activation.

FIGURE 4.

Effects of palmitate-deficient and MCD medium on SIRT3 and GPR91 expression in LX2 cells. A, LX2 cells were treated with or without palmitate for 20 h. Subsequently, SIRT3, GPR91, and α-SMA were detected using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. B, LX2 cells were treated with or without palmitate for 20 h, and SIRT3 mRNA was detected using RT-PCR analysis (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. C, LX2 cells were cultured in control or MCD medium for 24 h. Cell lysates were prepared, and SIRT3, GPR91, and α-SMA protein levels were analyzed using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). *, p < 0.05; ***, p < 0.001 (versus control medium). D, LX2 cells were cultured in control or MCD medium for 24 h, and SIRT3 mRNA was detected using RT-PCR analysis (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control medium.

SIRT3 Overexpression Attenuates Palmitate- and MCD Medium-induced HSC Activation

To test whether SIRT3 could improve palmitate or MCD medium-induced HSC activation, LX2 cells were infected with Ad-SIRT3 or Ad-control and subsequently treated with palmitate (300 μm) for 20 h. Palmitate treatment significantly decreased SIRT3 expression (Fig. 5A) and increased GPR91 and α-SMA protein expression in LX2 cells infected with Ad-control (Fig. 5A). However, GPR91 and α-SMA protein expression was attenuated in LX2 cells infected with Ad-SIRT3 in the presence of palmitate (Fig. 5A). We also found that SIRT3 adenoviral overexpression ameliorated the palmitate-induced decrease in SDH activity and the palmitate-induced increase in succinate concentrations (Fig. 5, B and C). MCD medium treatment significantly increased GPR91 and α-SMA protein expression in LX2 cells infected with Ad-control (Fig. 5D). However, GPR91 and α-SMA protein expression was decreased in LX2 cells infected with Ad-SIRT3 in the presence of MCD medium (Fig. 5D). Overexpression of SIRT3 ameliorated the reduction of SDH activity and attenuated increased concentrations of succinate by MCD medium (Fig. 5, E and F). We further found that, when LX2 cells were incubated with Ad-SIRT3, phosphorylation of ERK was attenuated in the presence of palmitate or MCD medium (Fig. 5, A and D).

FIGURE 5.

SIRT3 overexpression attenuates palmitate- and MCD medium-induced HSC activation. A, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. After 8 h, infected cells were incubated with or without palmitate (300 μm) for 20 h, and then cells were lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). *, p < 0.05; **, p < 0.01; ***, p < 0.001 (versus control). B, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. After 8 h, infected cells were incubated with or without palmitate (300 μm) for 20 h, and SDH activity was measured in whole-cell lysates. **, p < 0.01; ***, p < 0.001 (versus control). C, LX2 cells were transfected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. After 8 h, infected cells were incubated with or without palmitate (300 μm) for 20 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control. D, before transfection, LX2 cells were changed to control or MCD medium. Then LX2 cells were infected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. 24 h post-infection, cells were lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). *, p < 0.05; **, p < 0.01; ***, p < 0.001 (versus control). E, before transfection, LX2 cells were changed to control or MCD medium. Then LX2 cells were infected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. 24 h post-infection, SDH activity was measured in whole-cell lysates. **, p < 0.01 versus control medium. F, before transfection, LX2 cells were changed to control or MCD medium. Then LX2 cells were infected with Ad-SIRT3 or Ad-CMV-β-gal at an MOI of 30. 24 h post-infection, succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control medium.

Honokiol Treatment Attenuates Palmitate- and MCD Medium-induced HSC Activation

To test whether honokiol, a natural biphenolic compound, could improve palmitate- or MCD medium-induced HSC activation, LX2 cells were treated with or without honokiol (10 μm). After 4 h, the cells were incubated with or without palmitate (300 μm) for 20 h. Palmitate treatment significantly decreased SIRT3 expression (Fig. 6A) and increased GPR91 and α-SMA protein expression in LX2 cells treated with the control (Fig. 6A). However, GPR91 and α-SMA protein expression was attenuated in LX2 cells treated with honokiol in the presence of palmitate (Fig. 6A). We also found that honokiol treatment increased SIRT3 expression (Fig. 6A) and ameliorated the palmitate-induced decrease in SDH activity and the palmitate-induced increase of succinate concentrations (Fig. 6, B and C). MCD medium treatment significantly increased GPR91 and α-SMA protein expression in LX2 cells (Fig. 6D). However, GPR91 and α-SMA protein expression was decreased in the LX2 cells treated with honokiol in the presence of MCD medium (Fig. 6D). Honokiol treatment attenuated the MCD medium-induced decrease in SIRT3 expression (Fig. 6D) and ameliorated the reduction of SDH activity and attenuated increased concentrations of succinate by MCD medium (Fig. 6, E and F). We further found that, when LX2 cells were incubated with honokiol, phosphorylation of ERK was attenuated in the presence of palmitate or MCD medium (Fig. 6, A and D).

FIGURE 6.

Honokiol attenuates palmitate- and MCD medium-induced HSC activation through the SIRT3-SDH-succinate pathway. A, LX2 cells were treated with or without honokiol (10 μm). After 4 h, the cells were incubated with or without palmitate (300 μm) for 20 h, and then cells were lysed and subjected to Western blotting (top). Band intensities were calculated using the ImageJ software (NIH) (bottom). ***p < 0.001, versus control. B, LX2 cells were treated with or without honokiol (10 μm). After 4 h, the cells were incubated with or without palmitate (300 μm) for 20 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control. C, LX2 cells were treated with or without honokiol (10 μm). After 4 h, the cells were incubated with or without palmitate (300 μm) for 20 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control. D, LX2 cells were changed to control or MCD medium. Then LX2 cells were treated with or without honokiol (10 μm) for 24 h, and cells were lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom). ***, p < 0.001 versus control. E, LX2 cells were changed to control or MCD medium. Then LX2 cells were treated with or without honokiol (10 μm) for 24 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control medium. F, LX2 cells were changed to control or MCD medium. Then LX2 cells were treated with or without honokiol (10 μm) for 24 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control medium.

AAV-GPR91 shRNA Knockdown and SIRT3, GPR91, and α-SMA Expression in Isolated HSCs and Liver from MCD Diet-fed Mice as a Model of NAFLD

In a previous study, we demonstrated overexpression of GPR91 protein, decreased SDH activity, and increased succinate concentrations in isolated HSCs of MCD diet-fed mice as a model of NAFLD (32). In this study, we applied a novel gene therapeutic approach using recombinant adeno-associated virus (AAV)-mediated RNA knockdown of GPR91 gene expression in a mouse model of NAFLD.

All mice were fed their assigned MCD diet for 4 weeks, and AAV6-GFP shRNA (control shRNA) or AAV-GPR91 shRNA was injected into C57BL/6 mice via the tail vein on the first day of MCD diet feeding. In MCD diet-fed mice treated with AAV-GPR91 shRNA, we noted a decrease in steatohepatosis in H&E staining (Fig. 7A) and decreased fibrosis in Masson trichrome staining (Fig. 7B) compared with AAV-GFP shRNA treatment. Immunostaining for GPR91 and α-SMA confirmed the Western blotting results, which revealed that AAV-GPR91 shRNA knockdown decreased the amount of GPR91- and α-SMA-positive cells (Fig. 7, C and D).

FIGURE 7.

AAV-GPR91 shRNA-mediated knockdown of GPR91 alters α-SMA expression in isolated HSCs and livers from the MCD diet-fed mouse model of NAFLD. A, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. Hepatic steatosis was evaluated with H&E staining. B, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. Masson trichrome staining was also performed. C, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. The expression of GPR91 was evaluated by immunohistochemistry. D, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. The expression of α-SMA was evaluated by immunohistochemistry. E, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. The livers were lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus chow diet. F, MCD diet-fed mice were treated with AAV-GPR91 shRNA or AAV6-GFP (control shRNA) on the first day of MCD diet feeding. HSCs were isolated and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). **, p < 0.01; ***, p < 0.001 versus chow diet.

In the livers and isolated HSCs from mice that were fed the MCD diet, which was injected with control shRNA (AAV6-GFP shRNA), the expression of SIRT3 was not changed, and the expression of GPR91 and α-SMA protein was increased compared with mice that were fed the MCD diet, which was injected with AAV6-GPR91 shRNA. On the other hand, the data pertaining to the AAV-GPR91 shRNA treated mice showed decreased GPR91 and α-SMA protein expression (Fig. 7, E and F). To explore a possible mechanism by which GPR91 ameliorated hepatic steatosis, we conducted a Western blotting analysis, which revealed that GPR91 knockdown with AAV-GPR91 shRNA treatment decreased ERK phosphorylation in the liver and isolated HSCs from the MCD diet-fed mice (Fig. 7, E and F). Based on these results, we suggest that GPR91 knockdown improved steatohepatitis and decreased HSC activation through decreased ERK phosphorylation.

Resveratrol Treatment and SIRT3, GPR91, and α-SMA Expression in Isolated HSCs and Livers from MCD Diet-fed Mice as a Model of NAFLD

To examine its physiological relevance in vivo, we tested whether SIRT3 expression with resveratrol was attenuated in the livers and isolated HSCs of MCD diet-fed mice as a model of NAFLD. Resveratrol treatment prevented hepatic steatosis and fibrosis induced by an MCD diet (Fig. 8, A and B). Immunostaining for GPR91 and α-SMA showed that the MCD diet increased the amount of GPR91- and α-SMA positive cells, and resveratrol treatment in mice that were fed the MCD diet decreased the amount of GPR91- and α-SMA-positive cells (Fig. 8, C and D). Resveratrol treatment increased SRT3 protein expression and decreased GPR91 and α-SMA protein expression in the livers and isolated HSCs of MCD diet-fed mice (Fig. 8, E and F). Immunostaining for SIRT3 showed that the MCD diet decreased the number of SIRT3-positive cells (Fig. 9A), and resveratrol treatment of MCD diet-fed mice showed increased numbers of SIRT3-positive cells (Fig. 9A). We co-stained a mitochondrion-specific HSP 60 (green) with SIRT3 (red) to determine the subcellular localization of SIRT3 in cells, and the merged image showed a complete overlap of the two signals (Fig. 9B), indicating that SIRT3 exclusively localized to mitochondria.

FIGURE 8.

Effect of resveratrol on GPR91 and α-SMA expression in isolated HSCs and livers from the MCD diet-fed mouse model of NAFLD. A, MCD diet-fed mice were treated with or without resveratrol. Hepatic steatosis was evaluated with H&E staining. B, MCD diet-fed mice were treated with or without resveratrol. Masson trichrome staining was performed. C, MCD diet-fed mice were treated with or without resveratrol. The expression of GPR91 was evaluated by immunohistochemistry. D, MCD diet-fed mice were treated with or without resveratrol. The expression of α-SMA was evaluated by immunohistochemistry. E, MCD diet-fed mice were treated with or without resveratrol. The liver was lysed and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). *, p < 0.05; **, p < 0.01; ***, p < 0.001 (versus chow diet). F, MCD diet-fed mice were treated with or without resveratrol. HSCs were isolated and subjected to Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). *, p < 0.05; ***, p < 0.001 (versus chow diet).

FIGURE 9.

Effect of resveratrol on SIRT3 expression in liver from the MCD diet-fed mouse model of NAFLD. A, MCD diet-fed mice were treated with or without resveratrol. The expression of SIRT3 was evaluated by immunohistochemistry. B, representative immunofluorescent double staining for HSP60 (green) and SIRT3 (red). Merged images are shown on the right. C, MCD diet-fed mice were treated with AAV-GPR91 shRNA or resveratrol, and triglyceride (TG) contents in the livers were determined. **, p < 0.01; ***, p < 0.001 (versus chow diet).

Hepatic triglycerides were increased in mice fed the MCD diet compared with mice fed the chow diet but were decreased significantly in MCD diet-fed mice with AAV-GPR91 shRNA knockdown and MCD diet-fed mice with resveratrol treatment (Fig. 9C). Thus, resveratrol treatment decreased the accumulation of lipid droplets by increasing SIRT3 activity and decreasing GPR91 and α-SMA protein levels.

Palmitate and MCD Medium Treatment on Hepatocytes Isolated from Chow Diet-fed Mice

Palmitate treatment of isolated hepatocytes from chow diet-fed mice decreased SIRT3 expression in primary hepatocytes (Fig. 10A). Palmitate treatment of isolated hepatocytes from chow diet-fed mice decreased SDH activity and increased succinate concentrations in cell lysates (Fig. 10, B and C). Primary hepatocytes from chow diet-fed mice exposed to MCD medium also showed decreased SIRT3 protein expression (Fig. 10D). Additionally, MCD medium treatment of primary hepatocytes showed decreased SDH activity and increased succinate concentrations in cell lysates (Fig. 10, E and F). These findings indicate that, when primary hepatocytes are subjected to palmitate or MCD medium, this results in decreased SDH activity, increasing the cellular succinate concentrations.

FIGURE 10.

Effect of palmitate and MCD medium treatment on hepatocytes isolated from chow diet-fed mice. A, primary hepatocytes isolated from the livers of chow diet-fed mice were treated with or without palmitate (300 μm) for 20 h, and SIRT3 protein levels were determined using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. B, primary hepatocytes isolated from the livers of chow diet-fed mice were treated with or without palmitate (300 μm) for 20 h, and SDH activity was measured in whole-cell lysates. **, p < 0.01, versus control. C, primary hepatocytes isolated from the livers of chow diet-fed mice were treated with or without palmitate (300 μm) for 20 h, and succinate concentrations were measured in whole-cell lysates. ***, p < 0.001 versus control. D, primary hepatocytes isolated from the livers of chow diet-fed mice were cultured in control or MCD medium for 24 h, and SIRT3 protein levels were determined using Western blotting (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control medium. E, primary hepatocytes isolated from the livers of chow diet-fed mice were cultured in control or MCD medium for 24 h, and SDH activity was measured in whole-cell lysates. ***, p < 0.001 versus control medium. F, primary hepatocytes isolated from the livers of chow diet-fed mice were cultured in control or MCD medium for 24 h, and succinate concentration was measured in whole cell lysates. ***, p < 0.001 versus control medium.

Conditioned Medium Treatment from Hepatocytes Exposed to Palmitate or MCD Medium on HSC Activation In Vitro

We previously demonstrated that palmitate or MCD medium directly activated HSCs via decreased SDH activity, thus increasing succinate levels and inducing GPR91 activation (32). To establish whether the activation of HSCs was induced indirectly through the paracrine action of hepatocytes, we treated mouse hepatocytes (AML12 cells) with palmitate (300 μm) for 20 h. The conditioned medium (CM) from palmitate-treated hepatocytes was transferred to LX2 cells (Fig. 11A). We first measured SDH activity and succinate concentrations of the CM from palmitate-treated hepatocytes and demonstrated that SDH activity decreased significantly, whereas succinate concentrations increased in CM from palmitate-treated hepatocytes compared with control-treated hepatocytes (Fig. 11, B and C). In addition, the CM from palmitate-treated hepatocytes decreased the protein expression of SIRT3 and increased the expression of GPR91 and α-SMA proteins in LX2 cells (Fig. 11D). The CM from palmitate-treated hepatocytes was shown to decrease SDH activity and increase succinate concentrations in whole-cell lysates of LX2 cells (Fig. 11, E and F).

FIGURE 11.

Effect of CM from hepatocytes exposed to palmitate on HSC activation in vitro. A, schematic of cell-to-cell experiments with mouse hepatocytes and hepatic stellate cells. B, AML12 cells were treated with or without palmitate (300 μm) for 20 h, and SDH activity was measured. ***, p < 0.001 versus control. C, AML12 cells were treated with or without palmitate (300 μm) for 20 h, and succinate concentrations were measured. ***, p < 0.001 versus control. D, CM from palmitate-treated AML12 cells was transferred to LX2 cells for 20 h, and Western blotting analysis was performed with the indicated antibodies in LX2 cells (top panel). Band intensities were calculated using ImageJ software (bottom panel). ***, p < 0.001 versus control. E, CM from palmitate-treated AML12 cells was transferred to LX2 cells for 20 h, and SDH activity was measured in whole-cell lysates of LX2 cells. ***, p < 0.001 versus control. F, CM from palmitate-treated AML12 cells was transferred to LX2 cells for 20 h, and succinate concentrations were measured in whole-cell lysates of LX2 cells. ***, p < 0.001 versus control.

Discussion

In this study, we provided several lines of evidence that implicate SIRT3 in SDH activity and GPR91 expression in HSC activation in NAFLD. First, succinate and decreased SDH activity activate HSCs. Second, inhibiting SIRT3 with SIRT3 siRNA induced decreased SDH activity, increased succinate concentrations in whole-cell lysates, and increased protein expression of GPR 91 and α-SMA in LX2 cells, demonstrating HSCs activation through the SIRT3-SDH-GPR91 pathway. Third, SIRT3 overexpression with adenoviral SIRT3 transfection or honokiol treatment ameliorated palmitate-induced and MCD medium-induced HSC activation through the SIRT3-SDH-GPR91 pathway. Fourth, we demonstrated that GPR91 knockdown with AAV-GPR91 shRNA or nonspecific SIRT3 activation with resveratrol improved steatosis and decreased α-SMA production in MCD diet-fed mice. Fifth, increased concentrations of succinate in the cell lysates of primary hepatocytes exposed to palmitate or MCD medium and CM of palmitate-treated hepatocytes may indirectly enhance the activation of HSCs, suggesting that succinate acts as a paracrine modulator between hepatocytes and HSCs.

To the best of our knowledge, this is the first reported study to examine the association of SIRT3 and HSC at the cellular level. SIRT3 is the primary mitochondrially enriched deacetylase (35).

SIRT3 KO high-fat diet-fed mice were reported to exhibit increased insulin resistance because of defects in skeletal muscle glucose uptake (36). Chronic high-fat feeding for 12 weeks was found to result in decreased SIRT3 mRNA levels and SIRT3 protein expression compared with chow-fed controls (37). Our study demonstrated for the first time that SIRT3 siRNA transfection in HSCs induced decreased SDH activity, thereby increasing succinate-GPR91 signaling, ultimately resulting in the activation of HSCs.

In a previous study, we demonstrated that the succinate receptor GPR91 is overexpressed in HSC activation, and GPR91 siRNA treatment of LX2 cells showed decreased α-SMA protein expression (32). In this study, we demonstrated that GPR91 knockdown by GPR91 shRNA tail vein injection decreased α-SMA expression in the liver, isolated HSCs, and ameliorated the NASH phenotype in MCD diet-fed mice, suggesting that GPR91 could be a therapeutic target for the treatment of NAFLD.

Consistent with previous studies, we demonstrated that SDH, complex II of the electron transport chain, is regulated by SIRT3 (23, 24). Thus, although SIRT3 increases SDH activity and decreases succinate concentrations, these findings differ from those reported in other studies (35, 37). The conflicting results regarding the relationship of SDH and SIRT3 are currently not well understood, and this warrants further investigation. Nonetheless, the data reported here demonstrate that SIRT3 regulates SDH activity, thereby decreasing cellular succinate concentrations. Moreover, decreased succinate concentrations deactivate GPR91 expression, resulting in HSC deactivation, suggesting SIRT3 as a drug target for an HSC deactivator.

Succinate is an important metabolic intermediate that constitutes one of the intermediates of the citric acid cycle and is regarded as a signaling molecule (25, 27, 38), acting by binding to the succinate receptor, GPR91. There have been several reports showing that succinate and GPR91 are involved in renin release (39–42) in the kidney. However, few studies have explored any association between succinate metabolism and metabolic syndromes, including NAFLD. Correa et al. (43) demonstrated that succinate levels in the perfusate effluent of rat liver were rapidly increased up to 14-fold after ischemia was induced by interrupting portal flow. They also reported that succinate accelerated HSC activation. Our study demonstrated that succinate and GPR91 activation induced HSC activation and GPR91 knockdown using AAV-GRP91 shRNA and ameliorated HSC activation and the NASH phenotype of MCD diet-fed mice.

A pharmaceutical or nutriceutical activator of SIRT3 is of great interest. However, to date, no known SIRT3-specific agonist or antagonist has been developed. Resveratrol is a natural polyphenol, found mainly in grape skin and berries, that serves as a calorie restriction mimetic. In 2003, Howitz et al. (44) reported for the first time that resveratrol, which is also a polyphenol in red wine, was a small molecule activator of sirtuin 1 (SIRT1). Resveratrol regulates 5′-AMP-activated protein kinase (AMPK) and SIRT1 directly and indirectly in concert to promote protection against metabolic disease, mimicking calorie restriction. Resveratrol also inhibits phosphodiesterases and the NF-κB family of transcription factors (45). Some papers have shown that resveratrol did not activate SIRT1 in vitro in the presence of p53 and PGC1α, calling into question its ability to directly activate SIRT1 (46, 47). Based on these findings, the possibility has been raised that resveratrol could have health benefits through other unknown mechanism(s) apart from SIRT1 activation.

Resveratrol prevented high-fat diet-induced hepatic steatosis and endoplasmic reticulum stress in rats (48). In a randomized double-blind controlled study, resveratrol for 30 days in obese humans demonstrated decreased intrahepatic lipid contents and increased intramyocellular lipid levels. This resulted in activated 5′-AMP-activated protein kinase, increased SIRT1, and PGC-1α protein expression, as determined by muscle biopsies (49). In another study, it was shown that resveratrol acted as a direct stimulator of complex 1-stimulated SDH activity and citrate synthase activity by activation of SIRT3, rather than SIRT1 activity, in HepG2 cells (50). Intriguingly, resveratrol has been reported to be an activator of sirtuins, such as SIRT3. Our findings indicated that resveratrol treatment increased SIRT3 expression and decreased GPR91 and α-SMA expression in the liver and isolated HSCs while improving the NASH phenotype. Honokiol (3′,5-di-(2-propenyl)-1,1′-biphenyl-2,4′-diol) is a bioactive natural biphenolic compound derived from magnolia trees. Recently, Gupta and co-workers (51) demonstrated that honokiol bound to SIRT3 physically and increased the deacetylase activity of SIRT3 as a SIRT3 activator in the myocardium. Our study showed that honokiol treatment enhanced SIRT3 expression and ameliorated the palmitate- or MCD medium-induced HSC activation via the SIRT3-SDH-GPR91 pathway in LX2 cells.

Several recent studies reported cross-talk between hepatocytes and HSCs as a means of mediating the progression of hepatic fibrosis (52, 53). Qian et al. (52) demonstrated that hepatocyte nuclear factor 1α (HNF1-α) suppression in hepatocytes enhanced the activation of HSCs, suggesting the presence of an HNF1-α-modulated feedback circuit between hepatocytes and HSCs. Another study showed that free fatty acid-induced HSC activation was significantly enhanced in a simultaneous co-culture system with hepatocytes and LX2 cells in the same plate, whereas this effect was absent in both single cells and a Transwell model, suggesting cell-to-cell interaction (53).

Our study showed that the decreased activity of SDH with concurrent increased succinate levels in the CM of palmitate-treated hepatocytes may enhance the activation of HSCs, suggesting that succinate released from hepatocytes modulates HSC activation by paracrine action. In chronic liver injury, hepatocytes are impaired extensively, which could result in the release of many cytokines, including succinate, and may influence HSC activation through cell-to-cell interactions. Thus, restoration of hepatocyte dysfunction or deactivation of HSC, particularly through SIRT3 activity, may be an effective strategy to prevent NAFLD.

Members of the MAPK family and ERK1 and ERK2 activity are linked to cell growth, proliferation, differentiation, motility, and survival (54, 55). We found that succinate activated ERK1/2 phosphorylation signaling in LX2 cells. Additionally, pretreatment with an ERK inhibitor (U0126) significantly blocked the succinate-induced up-regulation of GPR91 and α-SMA expression in LX2 cells, suggesting that the ERK pathway is involved in GPR91 signaling. There have been many reports suggesting that restoration of SIRT3 in SIRT3-depleted animals and tissue cultures has a protective effect on mitochondrial metabolism, including fatty acid oxidation and NAFLD development. However, there is presently little direct evidence suggesting that tissue-specific SIRT3 overexpression in animal models of NAFLD is protective against insulin resistance. These and other related experiments will be critical for understanding the possible therapeutic value of a SIRT3 agonist. Presently, many questions pivotal to a complete understanding of the potential benefits and pathway of SIRT3-SDH-GPR91 signaling for the treatment of NAFLD remain unanswered and require further investigation. However, our study suggests that palmitate treatment of hepatocytes or HSCs decreases SIRT3 activity, thereby decreasing SDH activity and increasing succinate accumulations because of decreased SDH activity in hepatocytes or to HSCs moving out from the cytosol and activating GPR91 in HSCs, causing hepatic fibrogenesis (Fig. 12).

FIGURE 12.

The SIRT3, SDH, and GPR91 signaling pathway in HSCs and hepatocytes. Each solid line and arrow denotes a step in an activating pathway, and the dashed lines and arrows denote a step in an inhibiting pathway. Palmitate decreases SIR3 activity, thereby decreasing SDH activity in hepatocytes and HSCs. Increased succinate accumulation from decreased SDH activity in hepatocytes or HSCs moves out from the cytosol and activates GPR91 in HSCs, resulting in the production of α-SMA, TGF-β1, and collagen type I proteins.

In conclusion, this investigation showed a novel molecular and cellular mechanism of a SIRT3/SDH/GPR91 cascade in MCD-induced HSC activation in NAFLD. These findings highlight the biological significance of novel strategies targeting SIRT3 and GPR91 in HSCs with the goal of improving NAFLD.

Author Contributions

E. H. C. designed the study. E. H. C. and Y. H. L. co-wrote the paper. Y. H. L. performed most of the experiments. S. R. S. and E .H. L. performed the immunofluorescent staining of SIRT3 and mitochondrial co-staining. S. L. performed the special staining of liver tissue using SIRT3, GPR91, and α-SMA. E. H. C., D. H. C., and Y. H. L. analyzed all of the experiments. All authors reviewed the results and approved the final version of the manuscript.