Melatonin protects hepatocytes against bile acid-induced mitochondrial oxidative stress via the AMPK-SIRT3-SOD2 pathway.

Abstract Mitochondrial oxidative damage is hypothesized to contribute to the pathogenesis of chronic cholestatic liver diseases. Melatonin, an indolamine synthesized in the pineal gland, shows a wide range of physiological functions, and is under clinical investigation for expanded applications. Melatonin has demonstrated efficient protective effects against various types of oxidative damage in the liver system. This study investigates the protective effects of melatonin pretreatment on glycochenodeoxycholic acid (GCDCA)-induced hepatotoxicity and elucidates the potential mechanism of melatonin-mediated protection. Melatonin markedly decreased mitochondrial ROS (mROS) production in L02 cells treated with 100 μM GCDCA, and inhibited GCDCA-stimulated cytotoxicity. Notably, melatonin exerted its hepatoprotective effects by upregulating sirtuin 3 (SIRT3) activity and its expression level, thus regulating superoxide dismutase 2 (SOD2) acetylation and inhibiting the production of mROS induced by GCDCA. Moreover, siRNA targeting SIRT3 blocked the melatonin-mediated elevation in mitochondrial function by inhibiting SIRT3/SOD2 signaling. Importantly, melatonin-activated SIRT3 activity was completely abolished by AMP-activated, alpha 1 catalytic subunit (AMPK) siRNA transfection. Similar results were obtained in rat with bile duct ligation or BDL. In summary, our findings indicate that melatonin is a novel hepatoprotective small molecule that functions by elevating SIRT3, stimulating SOD2 activity, and suppressing mitochondrial oxidative stress at least through AMPK, and that SIRT3 may be of therapeutic value in liver cell protection for GCDCA-induced hepatotoxicity.


Introduction
Cholestasis, impairment in bile formation, occurs in a wide variety of human liver diseases [1]. High concentrations of certain bile acids induce rapid hepatocellular injury, infl ammation, bile duct proliferation, and fi brosis, and may also promote liver tumorigenesis [2]. Mitochondrial oxidative stress, which can be defi ned as an imbalance between the production of mitochondrial reactive oxygen species (mROS) and the presence of antioxidant molecules, is one of the most important mechanisms contributing to the progression of cholestatic liver diseases [3,4]. For example, our previous study showed that excess mROS induce oxidative damage to unsaturated fatty acids, proteins, and mitochondrial DNA (mtDNA) in the mitochondria of extrahepatic cholestatic patients [5]. Glycochenodeoxycholic acid (GCDCA) is the main toxic component of bile acid in patients with extrahepatic cholestasis [6], and the toxicity of GCDCA is thought to be mediated through the stimulation of lipid peroxidation and the induction of mitochondrial oxidative damage in a with various subcellular localizations and enzymatic activity, thus infl uencing their in vivo substrates and cellular functions [19]. Among them, SIRT3 is highly expressed in metabolically active tissues including the liver, is located to mitochondria, and regulates mitochondrial function via the deacetylation of key energy metabolic enzymes as well as antioxidant enzymes, such as superoxide dismutase 2 (SOD2) [20,21]. More importantly, SIRT3 physically interacts with and deacetylates SOD2, increasing SOD2 activity and thereby, inhibiting mitochondrial oxidative damage and maintaining mitochondrial function [22]. Recently, changes in SIRT3 expression and activity have been linked to various human liver diseases, including acute liver injury and hepatic lipid homeostasis [21,23].
Considering the profound impact of SIRT3 on mitochondrial function, we hypothesized that melatonin might attenuate oxidative injury in liver cells via mROS homeostasis regulation through the AMPK/SIRT3/SOD2 signaling pathway. Our results indicate that SIRT3 is required for melatonin ' s hepatoprotection, as SIRT3 inhibition abolished its observed protection in GCDCA-induced hepatotoxicity.

Cell culture
The human normal liver cell line, L02, was purchased from the cell bank of the Institute of Biochemistry and Cell Biology (Shanghai, China). The L02 cells were cultured in 1640 medium (HyClone) supplemented with 10% heat-inactivated fetal calf serum or FCS (HyClone) and 1% (v/v) penicillin/streptomycin (Sigma, St Louis, MO, USA) in a 5% CO 2 humidifi ed atmosphere at 37 ° C. At 80% confl uence, the cells were treated with 1 μ M melatonin for 2 h and then exposed to 100 μ M GCDCA (Sigma, St Louis, MO, USA) for 6 h for the various experiments.

Animal studies
Two-month-old rats were maintained in a 12:12-h light -dark phase and fed ad libitum. They were adapted for 2 weeks to the above conditions before experiments. The rats receiving BDL were randomized into two groups: fi rst group rats underwent BDL and were i.p. injected with melatonin (10 mg/kg), and second group rats underwent BDL and received an equal volume of normal saline. In addition, rats that received sham ligation of the bile duct were included as the sham group. Animals were euthanized under anesthesia at 7 days after surgery [24]. All animal experiments were approved by the Fuzhou General Hospital for accreditation of laboratory animal care.

Mitochondrial oxidative stress determination
Oxidative stress within the mitochondria was determined using MitoSOX Red (Invitrogen Corp., Carlsbad, CA, USA), a mitochondria-targeted fl uorescent probe for the highly selective detection of mitochondrial superoxide [25]. After treatment with melatonin and GCDCA, L02 cells were incubated with culture medium containing 5 μ M Mito-SOX for 10 min at 37 ° C. Changes in fl uorescence intensity were measured using microplate reader at excitation and emission wavelengths of 485 and 530 nm, respectively.

Measurement of mitochondrial membrane potential ( △ Ψ m)
A mitochondrial membrane potential assay kit including JC-1 (Beyotime Company, Shanghai, China) was used to measure the △ Ψ m in L02 cells. Briefl y, 1 ϫ 10 4 cells were seeded in a 96-well plate and incubated with 1xJC-1 in growth medium at 37 ° C for 20 min. Monomeric JC-1 green fl uorescence emission and aggregate JC-1 red fl uorescence emission were measured using an Infi nite ™ M200 Microplate Reader (Tecan, Mannedorf, Switzerland). The △ Ψ m in each group was calculated as the fl uorescence ratio of red to green and expressed as a multiple of the level in the control groups.

ATP content determination
ATP was measured with an ATP Determination Kit (Beyotime Company, Shanghai, China). Briefl y, cell or liver tissue lysates were mixed with a reaction buff er from the kit containing 1 mM dithiothreitol, 0.5 mM luciferin, and 12.5 μ g/ml of luciferase. After the solutions were mixed gently, intensity readings for the mixtures were obtained using an Infi nite ™ M200 microplate reader (Tecan, Mannedorf, Switzerland). The ATP concentrations in the samples were calculated using an ATP standard curve.

Cell viability assay
Cell viability was analyzed using a Cell Counting Kit-8 according to the manufacturer ' s instructions (Dojindo Molecular Technologies, Kumamoto, Japan). Briefl y, 1 ϫ 10 4 cells were seeded into 96-well plates. After treatment, 90 μ l of medium and 10 μ l of CCK-8 solution were added to each well. The cells were then incubated at 37 ° C for 2 h. After incubation, the absorption at 450 nm was measured using an Infi nite ™ M200 microplate reader.

SOD2 activity determination
Manganese SOD (SOD2) activity was determined with the SOD1 and SOD2 Assay Kit (Beyotime Company, China). The types of SOD were identifi ed by adding SOD1 inhibitor A and B to inhibit SOD1 activity (i.e., to detect SOD2). The absorption at 450 nm was measured using an Infi nite ™ M200 Microplate Reader.

SIRT3 activity
SIRT3 enzymatic activity was assayed using a fl uorometric kit (Enzo Life Sciences Inc.) following the manufacturer ' s instructions with modifi cations described below. A 40-μ g sample of protein was incubated at 37 ° C for 45 min with specifi c substrates. Then, 25 μ l of developer was added and the samples were incubated for an additional 45 min. Fluorescence intensity was measured with excitation at 350 nm and emission at 450 nm using an Infi nite ™ M200 Microplate Reader (Tecan, Mannedorf, Switzerland), as previously described [26].

RNA interference of SIRT3 or AMPK
L02 cells (1 ϫ 10 6 ) were transfected with either 100 nmol/l SIRT3 or AMPK-targeting small siRNA (Santa Cruz) or a control nonspecifi c siRNA (Santa Cruz). At 24 h after transfection, the cells were exposed to melatonin and GCDCA. The cells were then collected and processed for immunoblotting and other assays.

Western blot analysis
L02 cell or liver tissue lysates were centrifuged for 15 min at 12,000 ϫ g, and the resulting supernatant was transferred to a new tube. The protein concentrations were determined using a Bradford protein assay kit (Beyotime Company, Shanghai, China). The protein samples were separated by SDS-PAGE. Following protein transfer to PVDF membranes, the membranes were blocked and then incubated overnight at 4 ° C with antibodies against SIRT3 (Abcam, MA, USA), SOD2 (acetyl K68) (Abcam, MA, USA), SOD2 (Santa Cruz, CA, USA), pAMPK (Santa Cruz, CA, USA), AMPK (Santa Cruz, CA, USA), and β -actin (sigma, St Louis, MO, USA). The protein signals were visualized using an ECL detection system (Thermo Scientifi c) [27].

Assays of serum enzymes
Serum aspartate aminotransferase (AST) was assayed using a commercial test kit (Uscn, SEB214Ra) according to the manufacturer ' s instructions. The activity of serum enzyme determined is expressed as an international unit (IU/L).

Statistical analysis
Data were analyzed using GraphPad Prism-5 software. All experimental data are expressed as the mean Ϯ SEM, and each experiment was performed a minimum of three times. One-way ANOVA was used to determine statistical signifi cance, and P Ͻ 0.05 was considered to be statistically signifi cant.

Melatonin reduces GCDCA-induced mitochondrial oxidative damage in L02 cells
As GCDCA causes mitochondrial oxidative stress and cytotoxicity in L02 cells, we tested whether melatonin could interfere with this GCDCA-induced hepatotoxicity in vitro. Exposure to 100 μ M GCDCA increased mROS production to 2.2-fold of that in control L02 cells. However, pretreatment with 1 μ M melatonin successfully reduced this increase in mROS production ( Figure 1A). Excess mROS exposure could result in the oxidative damage to mitochondrial function and cell viability. Thus, we analyzed the mitochondrial membrane potential ( △ Ψ m), the ATP content and cell viability to explore the protective eff ects of melatonin against GCDCA. As expected, melatonin administration signifi cantly reversed observed reductions in mitochondrial function in GCDCA-treated cells ( Figure 1B -D).
Melatonin suppresses GCDCA-induced mitochondrial oxidative damage not by upregulating SOD2 expression but by downregulating acetylated SOD2 expression SOD2, the primary mitochondrial oxidative scavenger, plays a crucial role in the regulation of mROS [28]. In an attempt to elucidate the mechanism by which melatonin inhibits mitochondrial oxidative stress, the eff ects of melatonin on SOD2 activity were investigated. As expected, SOD2 activity was reduced after L02 cells were exposed to 100 μ M GCDCA, and melatonin pretreatment reversed this decrease (Figure 2A). SOD2 is regarded as a scavenging enzyme, and its activity is thought to be dependent on its mitochondrial level [29]. However, neither GCDCA nor melatonin treatment had a signifi cant eff ect on SOD2 protein levels ( Figure 2B). Moreover, SOD2 activity is also tightly regulated by acetylation at its lysine residues [30], with K68 being an important acetylation site. We found the GCDCA-induced SOD2(acetyl K68) was reversed by the addition of melatonin ( Figure 2C).

The mitochondrial-protective action of melatonin is SIRT3/SOD2 dependent on GCDCA-induced hepatotoxicity
Since SIRT3 function as a key regulator of SOD2 expression via epigenetic regulation of SOD2 gene, we sought to investigate the eff ect of melatonin on SIRT3. As shown in Figure 3B and C, GCDCA treatment resulted in a signifi cant decrease in SIRT3 expression and activity in L02 cells. Notably, melatonin pretreatment resulted in a signifi cant increase in both SIRT3 expression and activity. To confi rm whether melatonin is involved in GCDCAinduced mitochondrial oxidative damage regulated by SIRT3, we used an siRNA targeting SIRT3. SIRT3 siRNA successfully inhibited SIRT3 protein activity and expression (Supplementary Figure 1 to be found online at http://informahealthcare.com/doi/abs/ 10.3109/107157 62.2015.1067806). Moreover, SIRT3 siRNA reversed the protective eff ects of melatonin on GCDCA-induced mROS production, △ Ψ m decreases, ATP decline, and cytotoxicity ( Figure 4A -D). Most importantly, as shown in Figure 4E, the melatonin-induced decrease in acetylated-SOD2 expression was signifi cantly attenuated by SIRT3 siRNA in L02 cells exposed to GCDCA. Together, these data suggest a SIRT3-dependent eff ect of melatonin on acetylated-SOD2 expression with GCDCA exposure.

AMPK accounts for the key role of the SIRT3/SOD2 pathway in the hepatoprotective eff ects of melatonin
It was recently reported that AMPK could act upstream of the SIRT3 pathway [31,32]. We assessed AMPK signaling using Western blotting for the total and phosphorylated form of the protein. As shown in Figure 5A, treatment with melatonin increased the expression of p-AMPK. However, AMPK siRNA completely prevented the induction of SIRT3 activity and but did not change SIRT3 expression in L02 cells ( Figure 5B and C), indicating that AMPK is required for the activation of SIRT3 by melatonin. As expected, the melatonin-induced decrease in the expression of acetylated-SOD2 as well as the increase of cell viability was also blocked by AMPK siRNA ( Figure 5D and Supplementary Figure 2 to be found online at http:// informahealthcare.com/doi/abs. 10.3109/10715762.2015. 1067806).

Melatonin suppresses mitochondrial oxidative damage in the livers of rats with BDL
To determine whether melatonin suppressed GCDCAinduced mitochondrial oxidative damage through a SIRT3/ SOD2-dependent pathway in vivo, we examined the eff ects of melatonin in a BDL rat model. BDL resulted in  liver damage, which was confi rmed by increases in plasma AST levels and elevated AST level was partly attenuated by melatonin therapy ( Figure 6A). As expected, in the liver, melatonin signifi cantly increased p-AMPK expression ( Figure 6B and Supplementary Figure 3A to be found online at http://informahealthcare.com/doi/abs/10.3109/  10715762.2015.1067806). Moreover, SIRT3 protein levels and activity was decreased in BDL groups compared with shams. Interestingly, melatonin only restored SIRT3 activity without signifi cantly aff ecting SIRT3 protein levels in livers of BDL rats ( Figure 6B, C and Supplementary Figure 3B to be found online at http://informahealthcare. com/doi/abs/ 10.3109/10715762.2015.1067806). Next, the level of SOD2 (acetyl K68), a downstream target of the SIRT3 pathway, was also measured. Melatonin significantly attenuated the upregulation of acetylated SOD2, restored the SOD2 activity, and increased ATP synthesis in the livers of BDL rats ( Figure 6B, D, E and Supplementary Figure 3C to be found online at http:// informahealthcare.com/doi/abs/10.3109/10715762. 2015.1067806).

Discussion
A major advance of this study is that it is the fi rst to demonstrate that melatonin, a novel small molecule drug, protects liver cells against GDCDA-induced hepatotoxicity by activating the AMPK/SIRT3/SOD2 signaling pathway and reducing mitochondrial oxidative damage. More importantly, our fi ndings demonstrate a key role for SIRT3 in the inhibitory eff ects of melatonin on oxidative injury in liver cells, thus providing new insight into the hepatoprotective eff ects of melatonin.
Chronic liver cholestasis is responsible for the rapid development of progressive liver failure, for which there is still no eff ective therapy. GCDCA is the main toxic component of bile acid in patients with extrahepatic cholestasis. Experimental evidence indicates that increased mitochondrial oxidative stress plays an important role in the pathogenesis of extrahepatic cholestasis [25,33]. In cholestatic patients, the accumulation of mROS levels would decrease the synthesis of key subunits of respiratory complexes, reduce mtDNA copy number, and aff ect mitochondrial homeostasis [5,25]. Moreover, toxic bile acids have been found to increase mROS production, disrupt the mitochondrial membrane potential ( Δ Ψ m), and inhibit ATP synthesis in liver cells [34]. Consistent with previous research, we found that GCDCA increased mitochondrial oxidative damage in vitro. Taken together, GCDCAinduced hepatotoxicity may be the consequence of mROS.
Evidence suggests that mitochondria are the primary source of ROS, and the homeostatic regulation of mROS is infl uenced by various enzymes involved in the tricarboxylic acid or TCA, ETC, OXPHOS, and free-radicalscavenging systems in mitochondria, especially SOD2 [35,36]. Indeed, SOD2 activity is crucial for maintaining mROS balance. In the present study, cadmium treatment resulted in a decrease in the activity of SOD2, an enzyme that is required for scavenging excessive mROS in mitochondria. SOD2 activity is thought to be dependent on its mitochondrial levels; however, SOD2 protein expression was unaff ected by GCDCA treatment (Figures 2B, 6B and Supplementary Figure 3D to be found online at http:// informahealthcare.com/doi/abs/ 10.3109/10715762.2015. 1067806). Moreover, SOD2 is also regulated at the posttranslational level, as modifi cations, such as acetylation, signifi cantly infl uence its activity and the activity of SOD2 is inversely proportional to its acetylation [29,37]. Here, we observed that GCDCA increased the expression of acetylated SOD2 and inhibited SOD2 activity. These data suggest that GCDCA induces mROS production through the upregulation of acetylated SOD2 without aff ecting SOD2 protein levels.
Sirtuins are NAD ϩ -dependent enzymes that have been implicated in a wide range of physiological and pathophysiological conditions, largely through their deacetylation of numerous substrates. Recently, three mitochondrial deacetylation enzymes have been identifi ed, including SIRT3, SIRT4, and SIRT5 [38]. SIRT3 is the most robust mitochondrial deacetylase, and increases in SIRT3 expression were found to be accompanied by decreases in SOD2 deacetylation and increased SOD2 activity, limiting the accumulation of mROS [39]. Importantly, mice lacking SIRT3 showed striking hyperacetylation of mitochondrial proteins, which was associated with accelerated development of metabolic syndrome, whereas a lack of SIRT4 or SIRT4 caused no change [40 -42]. Our study shows that GCDCA decreased SIRT3 protein expression and activity, providing evidence for a detailed mechanism by which GCDCA increases both acetylation of SOD2 and mitochondrial oxidative damage in L02 cells.
Melatonin is an eff ective free radical scavenger and antioxidant, and it can prevent GCDCA-induced liver damage in young rats [24,43]. However, the hepatoprotective eff ects of melatonin are not simply related to the direct scavenging of free radicals and reduced infl ammation. Considering the profound impact of SIRT3 on mitochondrial function, an important fi nding is that melatonin can activate SIRT3 and thus downregulate mROS production in a dose-and time-dependent manner ( Figure 1A and Supplementary Figure 4 to be found online at http:// informahealthcare.com/doi/abs/ 10.3109/10715762.2015. 1067806). In the present experiments, we clearly show that melatonin can upregulate SIRT3 expression and its activity in liver cells, and in turn protect against GCDCA-induced mitochondrial oxidative damage. Nonetheless, blocking SIRT3 by siRNA could only partly abolish the liver cell protection conferred by melatonin suggesting that melatonin possibly acts via other pathways. One possibility is that melatonin readily enters mitochondria and has a direct role in maintaining SOD2 or other free-radical-scavenging enzymes to counteract the action of GDCCA; however, future studies are required to examine this possibility. Moreover, seven mammalian SIRTs (SIRT1-7) have been now identifi ed with distinct subcellular localization, enzymatic activities, and substrates [44]. In our study, among the seven mammalian sirtuins, only SIRT1 and SIRT3 were inhibited by GCDCA, whereas melatonin increased SIRT3 levels but failed to restore SIRT1 levels (Supplementary Figure 5 to be found online at http://informahealthcare.com/doi/abs/ 10.3109/10715762.2015.106780 6), which further confi rmed the important role of SIRT3 in melatonin-mediated protective eff ect.
Mechanistically, it is unclear how melatonin increases SIRT3 levels and activities. We can consider two distinct alternatives. First, we found that melatonin-mediated increases in SIRT3 mRNA levels may contribute to changes in SIRT3 levels or activity ( Figure 3A). Second, another new fi nding indicates that AMPK plays a key role in intracellular metabolism and is an attractive therapeutic target, and that switching on/off of AMPK leads to alteration in SIRT3 expression and activity [45]. Our studies demonstrated that melatonin triggers AMPK phosphorylation and activation, which is required for SIRT3 activity but not its expression in GCDCA-induced hepatotoxicity. Moreover, it is noteworthy that AMPK was phosphorylated by LKB1 CaMkkB,etc. [46,47]. Further study is needed to identify mechanisms by which melatonin exerts eff ects on the AMPK pathway.
In summary, our study clearly demonstrates that melatonin, as mediated by the AMPK-SIRT3-SOD2 axis, protects liver cells from GCDCA-induced mitochondrial oxidative damage. Taken together, our data illustrate a new molecular mechanism underlying the capabilities of melatonin, which should be explored for future clinical treatment of GCDCA-induced hepatotoxicity.