LIVER INJURY/REGENERATION

Increased Susceptibility of Natural Killer T-Cell-DeficientMice to Acetaminophen-Induced Liver Injury

Brittany V. Martin-Murphy,1 Douglas J. Kominsky,2 David J. Orlicky,3 Terrence M. Donohue, Jr.,5

and Cynthia Ju1,4

Acetaminophen (APAP) overdose causes severe, fulminant liver injury. The underlyingmechanism of APAP-induced liver injury (AILI), studied by a murine model, displays simi-lar characteristics of injury as those observed in patients. Previous studies suggest thataside from APAP-induced direct damage to hepatocytes, the hepatic innate immune systemis activated and may contribute to the overall pathogenesis of AILI. The current studyemployed the use of two murine natural killer (NK) cells with T-cell receptor (NKT) cellknockout models (CD1d2/2 and Ja182/2) to elucidate the specific role of NKT cells inAILI. Compared to wild-type (WT) mice, NKT cell-deficient mice were more susceptibleto AILI, as indicated by higher serum alanine transaminase levels and mortality. Increasedlevels of cytochrome P450 2E1 (CYP2E1) protein expression and activities, which resultedin increased APAP protein adduct formation, were observed in livers of APAP-treatedNKT cell-deficient mice, compared to WT mice. Compared to WT mice, starvation ofNKT cell-deficient mice induced a higher increase of ketone bodies, which up-regulateCYP2E1 through protein stabilization. Conclusion: Our data revealed a novel role of NKTcells in regulating responses to starvation-induced metabolic stress. Elevated ketone bodyproduction in NKT cell-deficient mice resulted in increased CYP2E1-mediated APAPbiotransformation and susceptibility to AILI. (HEPATOLOGY 2013;57:1575-1584)

Acetaminophen (APAP) is a commonly usedantipyretic and analgesic known to be safe andeffective at therapeutic doses (1-4 g/day).1

However, severe liver injuries have been observed afteran acute or cumulative overdose of APAP (10-15g/day).1

APAP-induced hepatocyte damage is initiated byformation of the reactive metabolite, N-acetyl-p-benzo-quinone imine (NAPQI).2 NAPQI rapidly depletesglutathione (GSH) within the liver and covalentlybinds to cellular macromolecules. Impairment of mac-romolecules results in mitochondrial dysfunction, lossof adenosine triphosphate (ATP), and centrilobularnecrosis.3 In addition to APAP-induced direct hepato-toxicity, activation of innate immune cells and theirproduction of pro- and anti-inflammatory mediators

may further influence the severity of APAP-inducedliver injury (AILI).4-8

Natural killer (NK) T (NKT) cells are a uniquesubset of T lymphocytes that express NK cell markersand represent one subset of innate immune cellswithin the liver (30%-50% of liver lymphocytes).9

NKT cells possess an invariant T-cell receptor (Va14-Ja18) and recognize glycosphingolipids.9 NKT cellactivation by glycolipid antigens occurs through themajor histocompatibility complex class I–like molecule,CD1d, which presents glycolipid antigens to the T-cellreceptor.10 Although the role of NKT cells in AILI hasnot been examined directly, an earlier study using anti-NK1.1 antibody (Ab), which depletes both NK andNKT cells, demonstrated a protoxicant role of thecombination.11 The data suggested that interferon

Abbreviations: Ab, antibody; AILI, APAP-induced liver injury; ALT, alanine transaminase; APAP, acetaminophen; ATP, adenosine triphosphate; BOH, 3-hydroxybutyrate; CT-L, chymotrypsin-like activity; CYP2E1, cytochrome P450 2E1; DMSO, dimethyl sulfoxide; FA, fatty acid; GSH, glutathione; H&E,hematoxylin and eosin; HPLC, high-performance liquid chromatography; IFN, interferon; IHC, immunohistochemical; IL, interleukin; IP, intraperitoneal; KO,knockout; MMP, mitochondrial membrane potential; mRNA, messenger RNA; NAPQI, N-acetyl-p-benzoquinone imine; NK, natural killer; NKT, NK cells withT-cell receptor; PCR, polymerase chain reaction; ROS, reactive oxygen species; T-L, trypsin-like activity; WT, wild type.From the 1Skaggs School of Pharmacy and Pharmaceutical Sciences, 2Department of Anesthesiology and Perioperative Medicine and Mucosal Inflammation Program,

3Department of Pathology, 4Integrated Department of Immunology, University of Colorado Anschutz Medical Campus, Aurora, CO; and 5Department of VeteransAffairs, VA Nebraska-Western Iowa Health Care System and Department of Internal Medicine, University of Nebraska, Omaha, NE.Received May 14, 2012; Revised October 18, 2012; accepted October 23, 2012.This work was supported by F31DK082269 (to B.V.M.) and RO1ES012914 (to C.J.). The authors thank Dr. Lance Pohl for the generous gift of anti-APAP

Ab and Dr. Laurent Gapin for the generous gift of Ja18�/� mice.

1575

(IFN)-c secretion from NK/NKT cells was responsiblefor induction of inflammatory mediators, enhancedleukocyte recruitment, and Fas ligand expression. Amore-recent study revealed that this pathogenic role ofNK and NKT cells in AILI was likely the result of di-methyl sulfoxide (DMSO) that was used as a solventfor APAP.12 It was found that DMSO increased thenumber and activation of hepatic NK and NKT cells.As such, the role of NK and/or NKT cells in AILIremains to be elucidated. In the present study, weaimed to investigate the specific role of NKT cells inAILI by the use of mouse models of genetic deletionof NKT cells (CD1d�/� and Ja18�/�).Our data showed that both CD1d�/� and Ja18�/�

mice developed higher degrees of liver injury thanwild-type (WT) mice after APAP challenge. Thisincreased susceptibility in NKT cell-deficient mice wasthe result of their increased expression and activity ofcytochrome P450 2E1 (CYP2E1), resulting in enhancedAPAP metabolism and protein adduct formation.

Materials and Methods

Mice Treatment. Female and male Balb/cJ WT,CD1d�/�, C57Bl/6J WT (Jackson Laboratories, BarHarbor, ME), and Ja18�/� mice13 (gift from Dr. Lau-rent Gapin, National Jewish Health, Denver, CO)were maintained in the Center for Animal Care. Mice(7-10 weeks old) were allowed food and water ad libi-tum until experimental use. Before treatment, food waswithheld overnight (16 hours). APAP (Sigma-Aldrich,St. Louis, MO), dissolved in warm phosphate-bufferedsaline, was administered by intraperitoneal (IP) injec-tion and food restored. After various time points,blood and liver tissues were collected.GSH Measurement. Livers were sonicated in 0.1 N

of perchloric acid (1:20, w/v). Glutathione (GSH) wasmeasured by high-performance liquid chromatography(HPLC) equipped with electrochemical detection,using a CoulArray system (ESA, Chelmsford, MA).Measurement of Mitochondrial Membrane Poten-

tial and Reactive Oxygen Species. Mitochondria wereisolated by homogenization of liver tissue (0.5 g), fol-lowed by two centrifugation steps at 650�g and5,400�g. JC-1 dye (5 lM; Molecular Probes, Grand

Island, NY) or MitoSOX dye (10 lM; Invitrogen, GrandIsland, NY) was added to mitochondrial pellets (1 mg/mL). Membrane potential and reactive oxygen species(ROS) were detected by fluorescence excitation/emissionspectra of 490/590 and 485/520 nm, respectively.CYP2E1 and Proteasomal Activity. CYP2E1 activ-

ity of microsomal protein was measured by hydroxyla-tion of p-nitrophenol, as previously described.14 Pro-teasomal activity of liver homogenates were assayed forchymotrypsin-like (CT-L) and trypsin-like (T-L) activ-ity, as previously described.15

Ketone Body Measurement. Serum 3-hydroxybuty-rate (BOH) was measured using the EnzyChromKetone body assay kit (BioAssay Systems, Hayward,CA). Absorbance was measured at 340 nm.Statistics Analysis. Statistical analysis was per-

formed using the Student t test. Differences in valueswere considered significant at P < 0.05.

Results

CD1d2/2 Mice Are Significantly More Susceptibleto AILI Than WT Mice. Female WT and CD1d�/�

mice were IP injected with APAP (385 mg/kg).CD1d�/� mice displayed significantly greater serumalanine aminotransferase (ALT) levels than WTmice at8 and 24 hours post-APAP challenge (Supporting Fig.2). Moreover, a significant decrease in survival was alsoobserved in CD1d�/� mice, compared to WT mice,starting at 8 hours post-APAP challenge. Only 25% ofCD1d�/� mice survived at 24 hours, whereas all theWT mice survived (Fig. 1A). When a lower dose ofAPAP (350 mg/kg) was administered, marked increasesin serum ALT levels were observed in CD1d�/� mice,compared to WTmice, at 24 and 48 hours post-APAPchallenge (Fig. 1B). Blinded histopathological evalua-tion of hematoxylin and eosin (H&E)-stained livertissue samples was performed. Histological analysisrevealed more-dramatic liver injury in CD1d�/� mice,compared to WT mice, 48 hours post-APAP challenge(Fig. 1E, F). To determine whether increased suscepti-bility of CD1d�/� mice to AILI is gender specific, wefurther compared susceptibilities of male WT andCD1d�/� mice to AILI. Similar to female mice, adecrease in survival was observed in male CD1d�/�

Address reprint requests to: Cynthia Ju, Ph.D., Skaggs School of Pharmacy, University of Colorado Anschutz Medical Campus, C238, 12850 East MontviewBoulevard, Aurora, CO 80045. E-mail: Cynthia.ju@ucdenver.edu; fax: 303-724-7266.CopyrightVC 2013 by the American Association for the Study of Liver Diseases.View this article online at wileyonlinelibrary.com.DOI 10.1002/hep.26134Potential conflict of interest: Nothing to report.Additional Supporting Information may be found in the online version of this article.

1576 MARTIN-MURPHY ET AL. HEPATOLOGY, April 2013

mice, compared to WT mice, starting at 8 hours withno mice surviving at 48 hours post-APAP challenge(235 mg/kg; Fig. 1C). At this dose, CD1d�/� micedisplayed significantly greater serum ALT levels at 8hours, compared to WT mice; however, there were notenough mice that survived to measure ALT at 24 and48 hours post-APAP challenge (Supporting Fig. 2). At

a lower dose, in which both groups of mice survive,ALT levels were significantly higher in CD1d�/� mice,compared to WTmice, at 24 and 48 hours post-APAPchallenge (230 mg/kg; Fig. 1D).Increased APAP-Protein Adduct Formation in

CD1d2/2 Mice, Compared to WT Mice. The mecha-nism of AILI involves APAP biotransformation into

Fig. 1. Increased susceptibilityof CD1d�/� mice to AILI. (A)Decreased survival of femaleCD1d�/� mice, compared to WTmice, after APAP challenge (385mg/kg). (B) Serum ALT levels,measured by using a colorimetricassay (Teco Diagnostics, Anaheim,CA), were increased in femaleCD1d�/� mice, compared to WTmice, at 24 and 48 hours post-APAP challenge (350 mg/kg). (C)Decreased survival of male CD1d�/

� mice, compared to WT mice, af-ter APAP challenge (235 mg/kg ofAPAP). (D) Increased serum ALTlevels in male CD1d�/� mice,compared to WT mice, at 24 and48 hours post-APAP challenge (230mg/kg of APAP). (E) Photomicro-graph (�100 and �400 final mag-nification) of H&E-stained liversections from female WT andCD1d�/� mice after 48-hour APAP.CV, central vein. (F) Necrosis,inflammatory infiltration, and sinu-soidal dilation scores from WT andCD1d�/� mice after 48-hour APAP.Individual necrosis (A) and cellularinfiltration (B) sinusoidal dilation(C) scores from each mouse strain(3 mice per group) were deter-mined. Results in (A-D) representmean 6 standard error of themean of 10 mice per group.Results in (F) represent average 6standard error of 3 mice per group.*P < 0.05 versus WT mice. Datashown are representative of two in-dependent experiments.

HEPATOLOGY, Vol. 57, No. 4, 2013 MARTIN-MURPHY ET AL. 1577

NAPQI, which depletes GSH in the liver. Upon GSHdepletion, NAPQI binds to hepatocellular proteins,forming APAP protein adducts.16 To assess whetherdifferential amounts of APAP protein adducts areformed in WT and CD1d�/� mice after APAP chal-lenge, female WT and CD1d�/� mice were treated for2 hours with APAP (350 mg/kg). Levels of hepaticprotein adducts were significantly increased in CD1d�/�

mice, compared to WT mice (Fig. 2A,C). Studies haveshown that adduct formation in the mitochondria isessential in APAP toxicity, because this leads to induc-tion of mitochondrial ROS formation and mitochon-drial permeability transition.17 Therefore, mitochondriawere isolated after 2-hour APAP challenge to measureAPAP protein adducts. Levels of mitochondrial proteinadducts were significantly higher in CD1d�/� than WTmice (Fig. 2B,D).Increased Mitochondrial ROS and Dysfunction in

CD1d2/2 Mice Than WT Mice. Mitochondrial ROSinduction has been demonstrated after APAP chal-lenge.17 In agreement with these findings, we observed

a significant increase in mitochondrial superoxide after1-hour APAP challenge in WT mice. Importantly,CD1d�/� mice exhibited significantly higher superox-ide levels in mitochondria, compared to WT mice(Fig. 3A). Interestingly, we also observed a significantincrease in superoxide levels after 16-hour starvation ofCD1d�/� mice, but not in WT mice (Fig. 3A). Asso-ciated with the increase in mitochondrial ROS, therewas a significant decrease in mitochondrial membranepotential (MMP) in CD1d�/� mice, compared to WTmice, after starvation as well as 1 and 2 hours afterAPAP challenge (Fig. 3B). These data indicate thatCD1d�/� mice are uniquely susceptible to mitochon-drial oxidative stress and dysfunction after starvationand APAP challenge.Similar GSH Levels in WT and CD1d2/2 Mice

Before and After APAP Challenge. Covalent bindingof NAPQI by GSH represents an important defensemechanism against APAP toxicity. To assess whetherGSH levels were innately different between WT andCD1d�/� mice, liver GSH levels in naı̈ve mice were

Fig. 2. Increased APAP protein adducts in CD1d�/� mice, compared to WT mice, after APAP challenge. Female CD1d�/� and WT mice werestarved overnight before being IP injected with APAP (350 mg/kg). After 2 hours, mice were sacrificed and livers harvested. (A) APAP proteinadducts in liver homogenates and (B) mitochondrial fractions were probed using anti-APAP antisera (1:5,000; gift from Dr. Lance Pohl, NationalInstitutes of Health, Bethesda, MD). Quantification of APAP protein adducts in (C) liver homogenates and (D) mitochondrial fractions by densito-metric analysis. Results represent mean 6 standard error of the mean of 3 mice per group. *P < 0.05 versus WT mice. Data shown are repre-sentative of three independent experiments.

Fig. 3. Increased mitochondrial dysfunction and ROS generation in CD1d�/� mice, compared to WT mice, after starvation and APAP chal-lenge. Female WT and CD1d�/� mice were starved overnight for 16 hours before IP injected with APAP (350 mg/kg). After 0, 1, or 2 hours ofAPAP treatment, mice were sacrificed and mitochondria isolated. (A) Mitochondrial polarization was detected by fluorometric analysis using JC-1cationic dye (Molecular Probes, Grand Island, NY). (B) Mitochondrial ROS was detected by using MitoSOX (Invitrogen, Grand Island, NY). Resultsrepresent mean 6 standard error of the mean of 3 mice per group. *P < 0.05 versus WT mice. Data shown are representative of two independ-ent experiments.

1578 MARTIN-MURPHY ET AL. HEPATOLOGY, April 2013

measured, and data showed similar levels in WT andCD1d�/� mice. Starvation of mice for 16 hourscaused a similar reduction in GSH levels (approxi-mately 50%) in WT and CD1d�/� mice (Fig. 4).After APAP challenge, GSH levels in WT andCD1d�/� mice decreased to the lowest level at 2 hoursand began to rebound at 8 and 19 hours. GSH levelswere lower in CD1d�/� than WT mice at 8 hourspost-APAP challenge, perhaps the result of enhancedhepatotoxicity in these mice.CYP2E1 Protein Expression and Activity Are

Higher in CD1d2/2 Mice, Compared to WT Mice,After Starvation. CYP2E1 is the major metabolizingenzyme in the biotransformation of APAP intoNAPQI.3 Therefore, we compared expression levels ofCYP2E1 in WT and CD1d�/� mice. No difference inCYP2E1 protein levels between naı̈ve WT andCD1d�/� mice was observed. However, after 16-hourstarvation, CYP2E1 protein levels were significantlyhigher in CD1d�/� mice, compared to WT mice (Fig.5A,B,D). Although immunoblotting analysis did notshow changes in CYP2E1 protein levels in WT miceupon starvation, immunohistochemical (IHC) stainingrevealed a significant increase in CYP2E1 expression inWT mice; however, starvation-induced up-regulationof CYP2E1 was markedly greater in CD1d�/� mice.CYP2E1 activity was higher in CD1d�/� mice thanWT mice after 16-hour starvation (Fig. 5C). To con-firm that increased susceptibility of CD1d�/� micewas the result of starvation-induced up-regulation ofCYP2E1, we treated naı̈ve nonstarved female WT andCD1d�/� mice with APAP (700 mg/kg). The resultsshowed no difference in serum ALT levels betweenWT and CD1d�/� mice (Supporting Fig. 1).

Similar Cyp2e1 Messenger RNA and ProteasomalDegradation, but Significantly Higher Ketone BodyLevels, in CD1d2/2 Mice, Compared to WT Mice. -CYP2E1 regulation occurs at both transcriptional andpost-translational levels.18,19 To explore the mechanismof increased CYP2E1 protein expression in CD1d�/�

mice after starvation, we examined hepatic Cyp2e1messenger RNA (mRNA) levels. Real-time polymerasechain reaction (PCR) analysis of naı̈ve and starvedWT and CD1d�/� mice demonstrated that starvationdid not up-regulate Cyp2e1 mRNA and that mRNAlevels were similar between WT and CD1d�/� mice(Fig. 6A). Furthermore, proteasome peptidase activ-ities, measured by CT-L and T-L activity analyses,were similar between WT and CD1d�/� mice afterstarvation (Fig. 6B,C), suggesting that differential pro-teasomal degradation cannot explain the increasedexpression level in CD1d�/� mice. Post-translationalstabilization mediated by substrate binding is anotherpossible mechanism accounting for the increased pro-tein expression of CYP2E1.20 BOH represents a mainketone body produced in the liver, which is freely con-verted into acetoacetate and broken down into ace-tone, two molecules reported to stabilize CYP2E1post-translationally.18,21 Our data demonstrated thatnaı̈ve WT and CD1d�/� mice exhibited similar levelsof BOH in serum. Starvation of mice increased BOHlevels in WT and CD1d�/� mice. A significantlygreater elevation of BOH levels was observed inCD1d�/� mice, compared to WTmice (Fig. 6D).Increased Susceptibility of Ja182/2 Mice to

AILI. To determine whether increased susceptibility ofCD1d�/� mice to AILI was the result of NKT celldepletion, but not an unexpected effect of CD1d dele-tion, we examined another strain of NKT cell-deficientmice (i.e., Ja18�/� mice). Female WT and Ja18�/�

mice were injected with APAP (350 mg/kg). A markedincrease in serum ALT levels was observed in Ja18�/�

mice, compared to WT mice, at 8, 24, and 48 hourspost-APAP challenge (Fig. 7A). Similar to femalemice, male Ja18�/� mice developed a greater degreeof injury, compared to WT mice (Fig. 7B). Signifi-cantly higher APAP protein adducts and a significantdecrease in MMP were observed in Ja18�/� mice after1-hour APAP challenge, compared to WT mice (Fig. 7C,D). A trending decrease in membrane potential wasobserved after starvation in Ja18�/�, compared to WTmice (data not shown). Similar to CD1d�/� mice, star-vation resulted in a significant increase in CYP2E1 pro-tein levels in Ja18�/� mice than WT mice (Fig. 7E).Last, starvation caused greater elevations of serum BOHlevels in Ja18�/� mice, compared to WT mice (Fig.

Fig. 4. Comparison of GSH levels between WT and CD1d�/� mice.GSH levels in liver homogenates were determined at 0 (before treat-ment), 2, 8, and 19 hours after treatment of WT and CD1d�/� mice withAPAP (350 mg/kg) by HPLC analysis. Results represent mean 6 stand-ard error of the mean of 10 mice per group. *P < 0.05 versus WT mice.Data shown are representative of two independent experiments.

HEPATOLOGY, Vol. 57, No. 4, 2013 MARTIN-MURPHY ET AL. 1579

7F). These data confirm that it was the deletion ofNKT cells that rendered mice more susceptible to AILI.

Discussion

Our data demonstrate that NKT cell-deficient miceare more susceptible to AILI than WT mice. This isthe result, in part, to starvation-induced up-regulationof CYP2E1 protein expression and activity, which isassociated with marked increases in hepatic APAP pro-tein adduct formation. Starvation also caused greaterelevations of ketone bodies in NKT cell-deficientmice, which may account for the increase in CYP2E1protein levels.Upon activation, NKT cells rapidly produce cyto-

kines, such as interleukin (IL)-4 and IFN-c.9 Manystudies have shown both protective and pathologicalfunctions of these cytokines in liver disease mod-els.22,23 Based on these findings, we examined whetherdifferential production of these cytokines between WTand CD1d�/� mice may explain the increased suscep-tibility of NKT cell-deficient mice to AILI. However,message levels of a number of cytokines were similarin liver tissues and isolated liver mononuclear cells (inwhich NKT cells are enriched) from APAP-treatedWT and CD1d�/� mice (data not shown). These

results suggest that APAP treatment does not triggerNKT cells to produce protective cytokines.It is established that APAP metabolism to NAPQI

and its covalent modification of liver proteins areessential in triggering hepatocyte damage.16 Because aseries of downstream events, such as mitochondrialdysfunction, ATP depletion, and DNA damage, takeplace before ALT release, there is a delay betweenNAPQI generation and increase of serum ALT levels.Compared to WT mice, CD1d�/� mice had signifi-cantly higher levels of APAP protein adducts as earlyas 2 hours post-APAP (Fig. 2); however, depending onthe dose of APAP, a significantly higher ALT level wasnot observed until 8 (Supporting Fig. 2) or 24 hoursafter APAP challenge (Fig. 1).GSH plays a pivotal role in AILI through scaveng-

ing of NAPQI. It has been demonstrated that mice de-ficient in both IL-10 and IL-4 (IL-10/4�/� mice) aremore susceptible to AILI, compared to WT mice. Thisappears to be the result of lower GSH levels in IL-I0/4�/� mice before, and more dramatically after, APAPchallenge.24 We observed no differences in total GSHlevels in livers of naı̈ve or starved WT and CD1d�/�

mice (Fig. 4). Although there appears to be a slightdelay in GSH rebound in the CD1d�/� mice at 8hours, GSH levels were similar in WT and CD1d�/�

Fig. 5. Increased CYP2E1 protein expression and activity in CD1d�/� mice, compared to WT mice, upon starvation. (A) CYP2E1 protein levelsin female naı̈ve and starved WT and CD1d�/� mice were detected in liver homogenates by immunoblotting analysis with anti-CYP2E1 Ab(1:3,000; Millipore, Bedford, MA). (B) Quantification of CYP2E1 protein levels by densitometric analysis. (C) CYP2E1 activity levels in femalestarved WT and CD1d�/� mice. (D) Photomicrograph (�400 final magnification) of CYP2E1 (1:1,000) IHC staining of liver sections from naı̈veand starved female WT and CD1d�/� mice. CV, central vein. Results represent mean 6 standard error of the mean of 3 mice per group.*P < 0.05 versus WT mice. Data shown are representative of three independent experiments.

1580 MARTIN-MURPHY ET AL. HEPATOLOGY, April 2013

mice at 19 hours after APAP treatment. Furthermore,we did not observe significant differences in expressionand holoenzyme formation of glutamate cysteineligase, the rate-limiting enzyme for GSH synthesis(data not shown). These data suggest that increasedsusceptibility of NKT-deficient mice to AILI, com-pared to WT mice, is not caused by differential levelsof GSH stores or synthesis.Although CYP2E1 protein expression is similar in

naı̈ve WT and NKT cell-deficient mice, it is signifi-

cantly higher in CD1d�/� and Ja18�/� mice thanWT mice upon starvation (Figs. 5A,B,D and 7E).CYP2E1 activity is induced under a variety of physio-logical and pathological conditions, including chronicalcohol consumption, nonalcoholic steatohepatitis, anddiabetes.20 CYP2E1 protein levels, but not mRNA lev-els, have been shown to increase 2- to 8-fold aftertreatment with ethanol, acetone, pyrazole, and isonia-zid. In pathological conditions, such as diabetes, andobesity, CYP2E1 levels have been observed to increase

Fig. 6. Transcriptional and post-translational regulation of CYP2E1 in WT and CD1d�/� mice. (A) Quantification of liver Cyp2e1 mRNA expres-sion by real-time PCR. (B) Quantification of CT-L proteasomal degradation activities in WT and CD1d�/� mice after 16-h starvation. (C) Quantifi-cation of T-L proteasomal degradation activities in WT and CD1d�/� mice after 16-hour starvation. (D) Quantification of serum BOH in naı̈ve andstarved WT and CD1d�/� mice. Results represent mean 6 standard error of the mean of 5-10 mice per group. Data shown are representativeof two independent experiments.

Fig. 7. Increased susceptibility of Ja18�/�

mice to AILI. (A) Increased serum ALT levelsin female Ja18�/� mice, compared to WTmice, at 8, 24, and 48 h post-APAP challenge(350 mg/kg of APAP). (B) Increased serumALT levels in male Ja18�/� mice, comparedto WT mice, at 24-hour post-APAP challenge(235 mg/kg of APAP). Female WT andJa18�/� mice were starved overnight beforeIP injection with APAP (350 mg/kg of APAP).After 1 hour, mice were sacrificed and liversharvested. (C) APAP protein adducts weredetected in liver homogenates. After 1 hour ofAPAP treatment, mice were sacrificed and mi-tochondria isolated. (D) Mitochondrial polar-ization was detected by fluorometric analysisusing JC-1 cationic dye (Molecular Probes,Grand Island, NY). (E) CYP2E1 protein levelsin female WT and Ja18�/� mice after 16-hour starvation were detected in liver homoge-nates by immunoblotting analysis with anti-CYP2E1 Ab. (F) Quantification of serum BOHin starved WT and Ja18�/� mice. Results in(A, B, and F) represent mean 6 standarderror of the mean of 10 mice per group. *P< 0.05. Results in (D) represent 3 mice pergroup. *P < 0.05 versus WT mice. Datashown are representative of two independentexperiments.

HEPATOLOGY, Vol. 57, No. 4, 2013 MARTIN-MURPHY ET AL. 1581

3- to 8-fold at both mRNA and protein levels.20 Theelevation of CYP2E1 after these conditions has beenattributed to changes in metabolism, specifically, theincrease of ketone bodies during these states.25 Ourdata demonstrated no significant difference in tran-scriptional activation in starved WT and CD1d�/�

mice (Fig. 6A). WT and CD1d�/� mice displayedsimilar amounts of proteasomal activity after starvation(Fig. 6B,C), indicating that a change in overall protea-somal function was not responsible for increasedCYP2E1 protein and activity. CYP2E1 substrates, suchas acetone, pyrazole, and ethanol, have been reportedto enhance CYP2E1 protein expression throughincreasing of protein stability.26 Studies of in vivo pro-tein labeling in rats revealed a biphasic turnover ofCYP2E1 at 7 and 32 hours. Acetone treatmentresulted in loss of the 7-hour degradation of CYP2E1,a process termed ‘‘substrate-induced stabilization.’’18

Computational modeling of a predicted cytosolic do-main of CYP2E1 identified a potential ubiquitylationsite, which may also serve as a site for substrate inter-action. This finding provides a possible mechanism forthe ability of substrate to bind and shield the enzymefrom proteasomal degradation.27 Additional CYPenzymes have been shown to be regulated by substrate-induced stabilization. For example, CYP3A protein isstabilized by troleandomycin.28

Ketone bodies are produced primarily in the liverand serve as a source of energy during starvation. Ourdata demonstrated that, after 16-hour starvation, NKTcell-deficient mice produced significantly higheramounts of BOH than WT mice (Figs. 6D and 7F).The correlation of increased ketone bodies to induc-tion of CYP2E1 is supported by many reports. In arat model of streptozocin-induced hyperketonemia,increased CYP2E1 protein expression and activity wereobserved.29 Diabetic rats with severe ketosis, consistingof high BOH in plasma, were found to have signifi-cantly higher CYP2E1 than nondiabetic controlmice.25 Furthermore, treatment of cultured mouse he-patocytes with acetoacetate stabilizes CYP2E1 proteinexpression in vitro.21 Acetone has also been implicatedin the induction of CYP2E1 activity. When adminis-tered to rats in drinking water, acetone inducedCYP2E1 2-fold higher, compared to control.18 Asidefrom the up-regulation of CYP2E1, starvation ofCD1d�/� mice significantly affected mitochondria, asevident by increased ROS production and decreasedmitochondrial viability (Fig. 3). This may be attribut-able to starvation-induced elevation in ketone bodies,because they are known to be produced in the mito-chondrial matrix of hepatocytes and have been shown

to induce mitochondrial ROS and dysfunction.30 Ele-vation of ketone bodies (acetoacetate) has been associ-ated with decreased GSH levels in diabetic patients aswell as in vitro cell-culture models.31 Because GSH isa potent ROS scavenger, reduction in GSH levels isimportant in causing mitochondrial dysfunction. Mito-chondrial impairment was dramatically worsened inCD1d�/� and Ja18�/� than WT mice upon APAPchallenge, which likely contributes to increased suscep-tibility of CD1d�/� mice to AILI (Figs. 3B and 8D).Increased ketone body production in NKT cell-defi-

cient mice suggests an underlying role of NKT cells inmetabolism. Several lines of evidence support a linkbetween NKT cells and metabolism. Patients with abe-talipoproteinemia, a rare Mendelian disorder character-ized by a lack of functional microsomal triglyceridetransfer protein, also exhibits reduced number of NKTcells and impaired functionality of these cells.32 Inmurine models of obesity (ob/ob mice), NKT cells aredecreased in number.33 Upon adoptive transfer ofNKT to ob/ob mice, a significant reduction in liver ste-atosis was observed, coinciding with marked improve-ment in glucose sensitivity.34 Furthermore, stimulationand expansion of NKT cell populations by means ofnorepinephrine or glucocerebroside injection has beenshown to decrease size and fat accumulation in theliver and decrease overall hepatic injury.35

The mechanisms by which NKT cells regulate me-tabolism during conditions of energy deficit or over-supply remain largely unknown, despite several recentstudies on this topic.36,37 We hypothesize that intrinsicIL-4 production by NKT cells may be critical in main-taining metabolic homeostasis. A recent report suggeststhat IL-4 activation of signal transducer and activatorof transcription 6 in hepatocytes can regulate fatty acid(FA) oxidation by suppression of peroxisome prolifera-tor-activated receptor alpha.38 It is also reported thatIL-4 increases thermogenic gene expression, FA mobi-lization, and energy expenditure by means of stimulat-ing alternatively activated macrophages.39 Anotherstudy demonstrated that IL-4 produced by eosinophilsin adipose tissue is important in protecting mice fromhigh-fat-diet–induced obesity.40 It is our plan forfuture studies to examine the role of endogenous IL-4production by NKT cells in metabolic regulation,which will require the use of IL-4-reporter mice.In conclusion, our data demonstrate that NKT cells

protect mice from AILI because genetic deletion ofthese cells causes significantly higher ketone body pro-duction upon starvation. Increased ketone bodies leadto mitochondrial stress even before APAP challenge,rendering mice more susceptible to another insult.

1582 MARTIN-MURPHY ET AL. HEPATOLOGY, April 2013

Increased ketone bodies also stabilize CYP2E1 protein,resulting in a marked increase of APAP bioactivationto generate the hepatotoxic metabolite, which causesliver injury (Fig. 8). We found that message levels of anumber of cytokines were similar in liver tissues andliver mononuclear cells (in which NKT cells areenriched) isolated from APAP-treated WT andCD1d�/� mice (data not shown). These results suggestthat APAP treatment does not trigger NKT cells toproduce protective cytokines. Our data do not supportan active protective role for NKT cells, but rather thatthe lack of NKT cells renders mice more susceptibleto AILI. This is the first study to examine the specificrole of NKT cells in AILI. The findings provide fur-ther insights into the underlying mechanisms of drug-induced liver injury, as well as other liver conditions inwhich CYP2E1-mediated ROS generation plays an im-portant pathological role.41 Aside from genetic condi-tions, such as abetalipoproteinemia, lipid antigens, bac-terial, and viral pathogens have been demonstrated toactivate NKT cells, which leads to decreased cell num-ber.42 Under such situations, NKT cell deficiency mayresult in increased susceptibility to metabolic stress, aswell as hepatotoxin-induced liver injury.

Acknowledgment: The authors thank Drs. ChrisFranklin and Don Backos for their assistance withglutathione cysteine ligase western blotting analysis. The

authors thank Casey Trambly for conducting the protea-some and CYP2E1 activity assays and Dr. James Galli-gan for assistance in CYP2E1 IHC. Special thanks toDr. Sean Colgan for the generous use of HPLC instru-mentation and Brittelle Bowers and Adrianne Burgessfor their technical assistance with HPLC setup.

References1. Kaplowitz N. Drug-induced liver disorders: implications for drug devel-

opment and regulation. Drug Saf 2001;24:483-490.2. Raucy JL, Lasker JM, Lieber CS, Black M. Acetaminophen activation

by human liver cytochromes P450IIE1 and P450IA2. Arch BiochemBiophys 1989;271:270-283.

3. Lee SS, Buters JT, Pineau T, Fernandez-Salguero P, Gonzalez FJ. Roleof CYP2E1 in the hepatotoxicity of acetaminophen. J Biol Chem1996;271:12063-12067.

4. Masubuchi Y, Bourdi M, Reilly TP, Graf ML, George JW, Pohl LR.Role of interleukin-6 in hepatic heat shock protein expression and pro-tection against acetaminophen-induced liver disease. Biochem BiophysRes Commun 2003;304:207-212.

5. Bourdi M, Masubuchi Y, Reilly TP, Amouzadeh HR, Martin JL,George JW, et al. Protection against acetaminophen-induced liverinjury and lethality by interleukin 10: role of inducible nitric oxidesynthase. HEPATOLOGY 2002;35:289-298.

6. Ju C, Reilly TP, Bourdi M, Radonovich MF, Brady JN, George JW,et al. Protective role of Kupffer cells in acetaminophen-induced hepaticinjury in mice. Chem Res Toxicol 2002;15:1504-1513.

7. Ishida Y, Kondo T, Kimura A, Tsuneyama K, Takayasu T, Mukaida N.Opposite roles of neutrophils and macrophages in the pathogenesis ofacetaminophen-induced acute liver injury. Eur J Immunol 2006;36:1028-1038.

8. Ishida Y, Kondo T, Tsuneyama K, Lu P, Takayasu T, Mukaida N. Thepathogenic roles of tumor necrosis factor receptor p55 in acetamino-phen-induced liver injury in mice. J Leukoc Biol 2004;75:59-67.

9. Godfrey DI, Hammond KJ, Poulton LD, Smyth MJ, Baxter AG. NKTcells: facts, functions, and fallacies. Immunol Today 2000;21:573-583.

10. Van Kaer L. NKT cells: T lymphocytes with innate effector functions.Curr Opin Immunol 2007;19:354-364.

11. Liu ZX, Govindarajan S, Kaplowitz N. Innate immune system plays acritical role in determining the progression and severity of acetamino-phen hepatotoxicity. Gastroenterology 2004;127:1760-1774.

12. Masson MJ, Carpenter LD, Graf ML, Pohl LR. Pathogenic role of nat-ural killer T and natural killer cells in acetaminophen-induced liverinjury in mice is dependent on the presence of dimethyl sulfoxide. HE-

PATOLOGY 2008;48:889-897.13. Cui J, Shin T, Kawano T, Sato H, Kondo E, Toura I, et al. Require-

ment for Valpha14 NKT cells in IL-12-mediated rejection of tumors.Science 1997;278:1623-1626.

14. Chen Q, Cederbaum AI. Cytotoxicity and apoptosis produced by cyto-chrome P450 2E1 in Hep G2 cells. Mol Pharmacol 1998;53:638-648.

15. Osna NA, White RL, Todero S, McVicker BL, Thiele GM, ClemensDL, et al. Ethanol-induced oxidative stress suppresses generation ofpeptides for antigen presentation by hepatoma cells. HEPATOLOGY 2007;45:53-61.

16. Pumford NR, Halmes NC, Hinson JA. Covalent binding of xenobiot-ics to specific proteins in the liver. Drug Metab Rev 1997;29:39-57.

17. Hanawa N, Shinohara M, Saberi B, Gaarde WA, Han D, KaplowitzN. Role of JNK translocation to mitochondria leading to inhibition ofmitochondria bioenergetics in acetaminophen-induced liver injury. JBiol Chem 2008;283:13565-13577.

18. Gonzalez FJ. The 2006 Bernard B. Brodie Award Lecture. Cyp2e1.Drug Metab Dispos 2007;35:1-8.

19. Lieber CS. Cytochrome P-4502E1: its physiological and pathologicalrole. Physiol Rev 1997;77:517-544.

Fig. 8. Proposed mechanism for the increased susceptibility of NKTcell-deficient mice to APAP-induced liver injury. We hypothesize thatthe intrinsic IL-4 production by NKT cells may be critical in maintainingmetabolic homeostasis, thus inhibiting the elevation of ketone bodiesthat are known modulators of CYP2E1 stability. Elevated CYP2E1 canlead to elevated NAPQI metabolite and overall increased susceptibilityto AILI. [Color figure can be viewed in the online issue, which is avail-able at wileyonlinelibrary.com.]

HEPATOLOGY, Vol. 57, No. 4, 2013 MARTIN-MURPHY ET AL. 1583

20. Novak RF, Woodcroft KJ. The alcohol-inducible form of cytochromeP450 (CYP 2E1): role in toxicology and regulation of expression. ArchPharm Res 2000;23:267-282.

21. Abdelmegeed MA, Carruthers NJ, Woodcroft KJ, Kim SK, Novak RF. Ace-toacetate induces CYP2E1 protein and suppresses CYP2E1 mRNA in pri-mary cultured rat hepatocytes. J Pharmacol Exp Ther 2005;315:203-213.

22. Ishida Y, Kondo T, Ohshima T, Fujiwara H, Iwakura Y, Mukaida N. Apivotal involvement of IFN-gamma in the pathogenesis of acetamino-phen-induced acute liver injury. FASEB J 2002;16:1227-1236.

23. Ryan PM, Bourdi M, Korrapati MC, Proctor WR, Vasquez RA, YeeSB, et al. Endogenous interleukin-4 regulates glutathione synthesis fol-lowing acetaminophen-induced liver injury in mice. Chem Res Toxicol2012;25:83-93.

24. Bourdi M, Eiras DP, Holt MP, Webster MR, Reilly TP, Welch KD,et al. Role of IL-6 in an IL-10 and IL-4 double knockout mouse modeluniquely susceptible to acetaminophen-induced liver injury. Chem ResToxicol 2007;20:208-216.

25. Bellward GD, Chang T, Rodrigues B, McNeill JH, Maines S, RyanDE, et al. Hepatic cytochrome P-450j induction in the spontaneouslydiabetic BB rat. Mol Pharmacol 1988;33:140-143.

26. Song BJ, Veech RL, Park SS, Gelboin HV, Gonzalez FJ. Induction ofrat hepatic N-nitrosodimethylamine demethylase by acetone is due toprotein stabilization. J Biol Chem 1989;264:3568-3572.

27. Banerjee A, Kocarek TA, Novak RF. Identification of a ubiquitination-Target/Substrate-interaction domain of cytochrome P-450 (CYP) 2E1.Drug Metab Dispos 2000;28:118-124.

28. Watkins PB, Wrighton SA, Schuetz EG, Molowa DT, Guzelian PS.Identification of glucocorticoid-inducible cytochromes P-450 in the in-testinal mucosa of rats and man. J Clin Invest 1987;80:1029-1036.

29. Shimojo N, Ishizaki T, Imaoka S, Funae Y, Fujii S, Okuda K. Changesin amounts of cytochrome P450 isozymes and levels of catalytic activ-ities in hepatic and renal microsomes of rats with streptozocin-induceddiabetes. Biochem Pharmacol 1993;46:621-627.

30. Rains JL, Jain SK. Hyperketonemia decreases mitochondrial membranepotential and its normalization with chromium (III) supplementationin monocytes. Mol Cell Biochem 2011;349:77-82.

31. Jain SK, McVie R. Hyperketonemia can increase lipid peroxidation andlower glutathione levels in human erythrocytes in vitro and in type 1diabetic patients. Diabetes 1999;48:1850-1855.

32. Zeissig S, Dougan SK, Barral DC, Junker Y, Chen Z, Kaser A, et al.Primary deficiency of microsomal triglyceride transfer protein in humanabetalipoproteinemia is associated with loss of CD1 function. J ClinInvest 2010;120:2889-2899.

33. Guebre-Xabier M, Yang S, Lin HZ, Schwenk R, Krzych U, Diehl AM.Altered hepatic lymphocyte subpopulations in obesity-related murinefatty livers: potential mechanism for sensitization to liver damage. HE-

PATOLOGY 2000;31:633-640.34. Elinav E, Pappo O, Sklair-Levy M, Margalit M, Shibolet O, Gomori

M, et al. Adoptive transfer of regulatory NKT lymphocytes amelioratesnon-alcoholic steatohepatitis and glucose intolerance in ob/ob mice andis associated with intrahepatic CD8 trapping. J Pathol 2006;209:121-128.

35. Li Z, Oben JA, Yang S, Lin H, Stafford EA, Soloski MJ, et al. Norepi-nephrine regulates hepatic innate immune system in leptin-deficientmice with nonalcoholic steatohepatitis. HEPATOLOGY 2004;40:434-441.

36. Miyagi T, Takehara T, Uemura A, Nishio K, Shimizu S, Kodama T,et al. Absence of invariant natural killer T cells deteriorates liver inflam-mation and fibrosis in mice fed high-fat diet. J Gastroenterol 2010;45:1247-1254.

37. Kotas ME, Lee HY, Gillum MP, Annicelli C, Guigni BA, Shulman GI,et al. Impact of CD1d deficiency on metabolism. PLoS One 2011;6:e25478.

38. Ricardo-Gonzalez RR, Red EA, Odegaard JI, Jouihan H, Morel CR,Heredia JE, et al. IL-4/STAT6 immune axis regulates peripheral nutri-ent metabolism and insulin sensitivity. Proc Natl Acad Sci U S A2010;107:22617-22622.

39. Nguyen KD, Qiu Y, Cui X, Goh YP, Mwangi J, David T, et al. Alter-natively activated macrophages produce catecholamines to sustainadaptive thermogenesis. Nature 2011;480:104-108.

40. Wu D, Molofsky AB, Liang HE, Ricardo-Gonzalez RR, Jouihan HA,Bando JK, et al. Eosinophils sustain adipose alternatively activated mac-rophages associated with glucose homeostasis. Science 2011;332:243-247.

41. Cederbaum AI, Lu Y, Wu D. Role of oxidative stress in alcohol-induced liver injury. Arch Toxicol 2009;83:519-548.

42. Kinjo Y, Tupin E, Wu D, Fujio M, Garcia-Navarro R, Benhnia MR,et al. Natural killer T cells recognize diacylglycerol antigens from path-ogenic bacteria. Nat Immunol 2006;7:978-986.

1584 MARTIN-MURPHY ET AL. HEPATOLOGY, April 2013