Acetaminophen/paracetamol‐induced liver failure—which is induced by the binding of reactive metabolites to mitochondrial proteins and their disruption—is exacerbated by fasting. As fasting promotes SIRT3‐mediated mitochondrial‐protein deacetylation and acetaminophen metabolites bind to lysine residues, we investigated whether deacetylation predisposes mice to toxic metabolite‐mediated disruption of mitochondrial proteins. We show that mitochondrial deacetylase SIRT3−/− mice are protected from acetaminophen hepatotoxicity, that mitochondrial aldehyde dehydrogenase 2 is a direct SIRT3 substrate, and that its deacetylation increases acetaminophen toxic‐metabolite binding and enzyme inactivation. Thus, protein deacetylation enhances xenobiotic liver injury by modulating the binding of a toxic metabolite to mitochondrial proteins.
There is a Hot off the Press (August 2011) associated with this Scientific Report.
The acetylation of protein lysine residues is a common post‐translational modification (Kim et al, 2006; Zhao et al, 2010). In the liver, mitochondrial proteins show nutrient‐level‐dependent lysine acetylation and deacetylation (Kim et al, 2006; Schwer et al, 2009), and the functional consequences of this being explored. The mitochondrial‐enriched deacetylase SIRT3 promotes protection against redox and nutrient‐excess stressors (Yang et al, 2007; Sundaresan et al, 2008; Bao et al, 2010; Tao et al, 2010), and SIRT3‐dependent reduction in redox‐stress is augmented by fasting or caloric restriction (Someya et al, 2010; Tao et al, 2010).
In the light of these nutrient‐restricted SIRT3 antioxidant effects, an interesting paradox exists: fasting or caloric restriction exacerbate redox‐stress‐dependent toxicity of the analgesic agent acetaminophen (N‐acetyl‐p‐aminophenol or APAP; Whitcomb & Block, 1994; Fernando & Ariyananda, 2009). APAP is the most common drug‐induced cause of acute liver failure (Hutson, 2010); the APAP toxic metabolite, N‐acetyl‐p‐benzoquinoneimine (NAPQI) depletes hepatic glutathione in a dose‐dependent manner (James et al, 2003; Jaeschke & Bajt, 2006). Unconjugated NAPQI then binds to cysteine residues causing oxidative stress, mitochondrial disruption and hepatotoxicity (Ruepp et al, 2002; Kon et al, 2004). Although there is N‐acetylcysteine therapy, mortality from APAP‐induced liver failure is not uncommon (Lai et al, 2006) and alternative therapies are being investigated (North et al, 2010). Although NAPQI also binds to hepatic‐protein lysine residues (Zhou et al, 1996), the functional significance of this interaction is unknown. As lysine acetylation might partly function by competing with alternative post‐translational modifications, targeting the same residue (Yang & Seto, 2008), the paradox regarding fasting and redox stress in the context of acetaminophen‐induced liver injury (AILI) might be due, in part, to protein lysine residue deacetylation‐mediated susceptibility to NAPQI binding.
We hypothesized that if protein acetylation allosterically inhibits NAPQI binding, SIRT3‐mediated deacetylation might exacerbate hepatotoxicity, despite the protective role of SIRT3 against redox‐stress.
The absence of SIRT3 protects against AILI
We administered a toxic dose of APAP to SIRT3+/+ and SIRT3−/− mice in fed and fasted conditions. Baseline serum alanine transaminase (ALT) levels were indistinguishable between genotypes, and similar ALT release was seen in fed mice, irrespective of their genotype (Fig 1A). As expected, fasting increased susceptibility to AILI (Fig 1A). However, in contrast to the fed mice, the fasted SIRT3−/− mice showed less hepatotoxicity, as shown by ALT release and less centrilobular coagulative necrosis—compared with the SIRT3 competent mice—in response to APAP administration (Fig 1A–C). This difference in fasting hepatotoxic susceptibility between the SIRT3+/+ and SIRT3−/− mice is not because of the differential rate of depletion of glutathione, as the levels of glutathione are similar following the overnight fast in the two genotypes, and are similarly depleted 4 h after APAP administration (Fig 1D).
Identification of putative targets for SIRT3‐mediated AILI
To identify new mitochondrial‐protein acetylation targets that might orchestrate the protective effects against AILI, we screened for candidate proteins that could potentially mediate this effect, by comparing hepatic mitochondrial‐protein acetylation profiles between fasted SIRT3−/− and SIRT3+/+ mice. Two‐dimensional gel and immunoblot analysis with an antibody recognizing acetylated lysine were used. Differentially acetylated spots were isolated, and sequenced by tandem mass spectrometry (MS/MS). We identified 17 proteins that reproducibly showed increased lysine acetylation in the SIRT3−/− compared with SIRT3+/+ mice (supplementary Table S1 online; supplementary Fig S1 online).
We selected aldehyde dehydrogenase 2 (ALDH2) as the index protein, because ALDH2 is a known target of NAPQI (Landin et al, 1996; Qiu et al, 1998), NAPQI binding is associated with reduced ALDH2 activity (Landin et al, 1996) and this dehydrogenase functions to oxidize and detoxify aldehydes, including lipid peroxidation products such as trans‐4‐hydroxy‐2‐nonenal (4‐HNE; Doorn et al, 2006). Increased acetylation of ALDH2 in SIRT3−/− liver mitochondria was evident (Fig 2A). To confirm that ALDH2 was differentially acetylated, immunoprecipitation studies were performed using liver mitochondrial‐protein lysates. Immunoprecipitation with antibodies against ALDH2 and acetylated‐lysine residues with reciprocal immunoblot analysis showed that ALDH2 was more‐robustly acetylated in SIRT3−/− mice (Fig 2B). This change in acetylation was associated with a direct interaction between SIRT3 and ALDH2 by bidirectional immunoprecipitation and immunoblot analysis using antibodies against SIRT3 and ALDH2 (lysates from SIRT+/+ mice; Fig 2C). We also demonstrate in vitro deacetylation of ALDH2 in the presence of SIRT3 and NAD+ and the preservation of acetylation in the presence of the sirtuin deacetylase inhibitor nicotinamide (Fig 2D). Furthermore, we confirmed that fasting increased hepatic ALDH2 deacetylation in SIRT3‐competent mice (Fig 2E), and that the SIRT5 mitochondrial deacetylase does not target ALDH2 for deacetylation (Fig 2F).
ALDH2 activity in SIRT3−/− mice after APAP administration
As baseline ALDH2 activity was indistinguishable between genotypes (Fig 3A), we evaluated whether functional effects of ALDH2 acetylation could be revealed by measuring enzyme activity and consequences of enzymatic action in response to APAP toxicity in fasted mice. A reduction in mitochondrial ALDH2 activity at 1 and 4 h after APAP administration was evident in both genotypes, although the relative activity was sustained at approximately 40% higher levels in SIRT3−/− mitochondria at both of these time points (Fig 3A). This difference in activity was probably not due to changes in protein degradation, as steady‐state ALDH2 levels were not diminished in response to APAP (Fig 3B). To evaluate the functional consequences of altered ALDH2 activity, we measured levels of lipid peroxidation adduct‐binding to mitochondrial proteins. The baseline levels of 4‐HNE in liver mitochondria were similar (supplementary Fig S2 online). By contrast, 1 h after APAP exposure, the SIRT3−/− mitochondrial proteins showed lower 4‐HNE‐adduct levels compared with SIRT3+/+ mice (Fig 3C). As disruption of mitochondrial respiration is an early indicator of APAP‐induced hepatotoxicity (Donnelly et al, 1994), we quantified mitochondrial respiration in response to APAP and showed that oxidative metabolism was preserved to a greater extent in the SIRT3−/− compared with SIRT3+/+ mice (supplementary Fig S3 online).
The depletion of ALDH2 counters AILI resistance
The functional requirement of ALDH2 activity in resisting APAP toxicity in SIRT3−/− mice was then interrogated using lentiviral short‐hairpin RNA (shRNA). Different shRNA viral particles targeting ALDH2 or scrambled controls were tested in murine liver cells and construct sh5–ALDH2 showed the most robust knockdown (supplementary Fig S4A online). Five days after tail‐vein injection with this shALDH2 and scrambled control lentiviral particles, liver extracts showed an approximately 50% reduction in ALDH2 protein levels compared with scrambled shRNA‐infected mice (Fig 3D). APAP toxicity studies were performed in control and ALDH2 shRNA‐targeted SIRT3+/+ and SIRT3−/− mice. The introduction of lentiviral particles increased liver injury in both strains compared with toxicity in the absence of virus (compare Fig 3E with Fig 1A). The specific knockdown of ALDH2 in SIRT3−/− mice increased liver injury in response to APAP, with greater elevations in ALT levels at 8 and 24 h (Fig 3E). AILI susceptibility was not, however, changed in SIRT3+/+ mice (Fig 3E), indicating that the disruption of ALDH2 function by AILI in wild‐type mice cannot be perturbed further by the shRNA depletion of ALDH2. The enhanced liver injury following ALDH2 knockdown in the SIRT3−/− mice was reflected by increased levels of 4‐HNE adducts binding to mitochondrial proteins (Fig 3F) and by exacerbation of centrilobular necrosis (supplementary Fig S4B online).
ALDH2 acetylation modulates NAPQI binding
The possibility that a reduction in ALDH2 deacetylation could maintain enzymatic activity in response to toxic doses of acetaminophen has not been previously recognized. As binding of the APAP reactive metabolite, NAPQI, to ALDH2 has previously been associated with lower ALDH2 activity (Landin et al, 1996), we investigated whether inhibition of sirtuin activity could modulate NAPQI binding. Exogenous ALDH2 was expressed in murine liver cells and exposed to NAPQI in the presence or absence of the sirtuin inhibitor nicotinamide. The amount of NAPQI bound to ALDH2 was reduced by inhibition of sirtuin deacetylase functioning, with a concomitant increase in ALDH2 protein acetylation (Fig 4A). In parallel, binding of NAPQI to ALDH2 was increased after overexpression of deacetylase‐competent SIRT3, and attenuated in the presence of the deacetylase‐mutant SIRT3, with reduced ALDH2 acetylation correlating to the degree of NAPQI binding (Fig 4B). In parallel, the small‐interfering RNA knockdown of SIRT3 in these hepatocytes diminished NAPQI binding to ALDH2 (Fig 4C) and the overexpression of SIRT5 did not alter NAPQI binding to ALDH2 (Fig 4D).
Deacetylation of ALDH2 Lys377 enhances NAPQI binding
To confirm that the acetyl status of lysine residues modifies NAPQI binding to ALDH2, we identified ALDH2 lysine residues that were deacetylated by SIRT3. Liquid chromatography‐MS/MS sequencing of acetylated lysine antibody immuno‐captured trypsin‐digested peptides (Kim et al, 2006) from SIRT3+/+ and SIRT3−/− mice mitochondria identified Lys370 and Lys377 as acetylation sites that might be operational in this allosteric modification of NAPQI binding (Fig 4E). Mutational studies were performed to mimic deacetylation (Lys to Arg) and acetylation (Lys to Gln), and showed that the ALDH2 Lys377 is a principal lysine residue modifying NAPQI binding (Fig 4F).
The predominant approach to ameliorate toxic effects of AILI has focused on antioxidant pathways on the basis of NAPQI binding to cysteine residues (Nelson, 1990; Qiu et al, 1998). Even though NAPQI binds to lysine residues (Zhou et al, 1996), the functional consequences of this interaction have, to our knowledge, not been explored. This study investigates whether occupancy of lysine by acetyl groups modifies NAPQI binding, and assesses whether lysine acetylation ameliorates APAP hepatotoxic pathophysiology. Our findings support the hypothesis that maintenance of lysine acetylation attenuates competing metabolites targeting the same residue (Yang & Seto, 2008), and advances this to include retardation of NAPQI binding.
The temporal sequence of APAP‐induced toxicity is well established; adduct binding and glutathione depletion occur within the first few hours, followed by the release of liver enzymes, and cell death over the next 8–48 h (Roberts et al, 1991; Hinson et al, 2004). The role of fasting‐induced SIRT3 lysine residue deacetylation in increasing NAPQI binding to ALDH2 in the early phase of hepatotoxicity after an acute toxic dose of APAP is shown in this study. The consequences of this toxicity are illustrated by the exacerbation of liver injury 24 h later in wild‐type compared with SIRT3‐null mice.
These data do not dismiss the role of glutathione depletion in APAP toxicity, as evidenced by greater injury in both genotypes when comparing fasted with fed mice. Instead this study expands our understanding of the pathophysiology of APAP toxicity by uncovering a new functional role for ALDH2 in attenuating acetaminophen hepatotoxicity, and identifying the role of acetylation in altering NAPQI binding to mitochondrial proteins. Also, although the knockdown of ALDH2 nullifies the ameliorative effects of SIRT3 deficiency in this pathophysiology, our data cannot exclude the role of the other SIRT3 mitochondrial targets identified in this (supplementary Table S1 online) and other studies. Characterization of these additional SIRT3 targets in this toxicity warrants further investigation. Interestingly, reactive metabolites of halothane, which are also hepatotoxic, similarly bind to lysine residues (Zhou et al, 1996; Bourdi et al, 2001). Whether the concept of toxic metabolite binding to lysine has broader implications in xenobiotic‐mediated hepatic injury should also be explored.
Finally, this study has not only identified ALDH2 as a new SIRT3 target, but also shows that the allosteric binding of the acetaminophen toxic metabolite to ALDH2 is controlled by the acetylation status of this dehydrogenase. This nutrient‐dependent function of SIRT3 identifies a unique mediator of stress‐susceptibility in response to a xenobiotic metabolite, that is distinct from the antioxidant effect of SIRT3, previously shown in the activation of manganese superoxide dismutase by deacetylation (Tao et al, 2010).
In conclusion, the SIRT3‐dependent acetylation status of mitochondrial proteins modulates susceptibility to AILI. Mitochondrial ALDH2 is identified as a new functional substrate of SIRT3 deacetylation that has an important detoxifying role in modulating APAP toxicity, and the acetylation of ALDH2 at position Lys377 modifies the amount of NAPQI binding. The manipulation of mitochondrial protein acetylation is a new concept to investigate in the search for strategies to reduce acetaminophen hepatotoxicity.
Detailed experimental procedures are described in the supplementary information online.
Animal studies. The inbred C57BL/6 SIRT3+/− mice were studied and a heterozygous breeding scheme was used. APAP was administered in a single toxic dose (350 mg/kg, by intraperitoneal injection) to either fed or overnight‐fasted mice. All animal experiments were approved by the NHLBI Animal Care and Use Committee.
Lentiviral shRNA preparation and mice injection. Lentiviruses were produced encoding a non‐targeting control sequence or constructs encoding mouse ALDH2 shRNA sequences. Mice were subjected to tail‐vein injections with control or shRNA lentivirus (1.2 × 1011 particles). ALDH2 knockdown was confirmed and APAP was subsequently administered to fasting mice. ALT measurement and liver histology were then analysed.
NAPQI‐binding studies. pCMV6–ALDH2–Flag plasmid was transfected into Hepa‐1c1c7 cells. To evaluate the role of SIRT3 deacetylase, the wild type (hSIRT3–Flag), the deacetylase catalytically inactive constructs (hSIRT3–H248Y–HA) or mSIRT5–3xFlag were overexpressed. Cells were then treated with NAPQI (200 μM) and nicotinamide (5 mM) or vehicle control. Immunoprecipitation using a Flag antibody to extract the ALDH2 was performed and run on SDS–polyacrylamide gel electrophoresis gels. APAP metabolite binding to ALDH2 was detected by immunoblot analysis with an antibody to NAPQI protein adducts.
Statistical analysis. Differences between data groups were evaluated for significance using the two‐tailed Student's t‐test. Multiple comparison analysis was performed using analysis of variance. Statistical analysis was performed using Graphpad Prism, and data are expressed as mean±s.e.m. P<0.05 is considered to be statistically significant.
Supplementary information is available at EMBO reports online (http://www.emboreports.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
We thank F.W. Alt and H.‐L. Cheng from Harvard Medical School, USA, for their generous gift of the C57BL/6 strain of SIRT3+/− mice. We thank T. Finkel (NHLBI) and B.J. Song (National Institute of Alcohol Abuse and Alcoholism) for helpful discussions, and the NHLBI Laboratory of Animal Medicine and Surgery core facility for assistance with murine tail‐vein injection of lentivirus shRNA. This research was funded by NHLBI Division of Intramural Research.
Author contributions: Z.L., M.B., M.G., L.R.P. and M.N.S. designed the study. Z.L., M.B., J.H.L., A.M.A. and Y.C. performed the experiments. D.B.L. contributed key reagents. Z.L., M.B., A.M.A. and Y.C. analysed the data. Z.L., M.B., D.B.L., M.G., L.R.P. and M.N.S. wrote the manuscript.
- Copyright © 2011 European Molecular Biology Organization