Acetaminophen-induced Liver Injury Is Attenuated in Male Glutamate-cysteine Ligase Transgenic Mice*

Acetaminophen overdose is a leading cause of drug-related acute liver failure in the United States. Glutathione, a tripeptide antioxidant protects cells against oxidative damage from reactive oxygen species and plays a crucial role in the detoxification of xenobiotics, including acetaminophen. Glutathione is synthesized in a two-step enzymatic reaction. Glutamate-cysteine ligase carries out the rate-limiting and first step in glutathione synthesis. We have generated C57Bl/6 mice that conditionally overexpress glutamate-cysteine ligase, and report here their resistance to acetaminophen-induced liver injury. Indices of liver injury included histopathology and serum alanine aminotransferase activity. Male transgenic mice induced to overexpress glutamate-cysteine ligase exhibited resistance to acetaminophen-induced liver injury when compared with acetaminophen-treated male mice carrying, but not expressing glutamate-cysteine ligase transgenes, or to female glutamate-cysteine ligase transgenic mice. We conclude that glutamate-cysteine ligase activity is an important factor in determining acetaminophen-induced liver injury in C57Bl/6 male mice. Because people are known to vary in their glutamate-cysteine ligase activity, this enzyme may also be an important determinant of sensitivity to acetaminophen-induced liver injury in humans.

GSH plays a major role in detoxifying many hepatotoxicants including acetaminophen (APAP), 3 an over-the-counter analgesic and antipyretic (2)(3)(4)(5). APAP overdose is responsible for nearly 50% of the acute liver failure cases in the United States (6) and is thus of high public health concern. APAP metabolism has been well defined, making it a good model for drug-induced liver toxicity. APAP is primarily metabolized through sulfation and glucuronidation pathways (7)(8)(9). However, a fraction of APAP is bioactivated by cytochrome P-450s to n-acetyl-p-benzoquinoneimine (NAPQI), which can bind to cellular proteins (3, 7, 10 -12). NAPQI also covalently binds to GSH and is either converted back to APAP, or forms the non-toxic APAP-GSH conjugate. APAP overdose results in depletion of hepatic GSH (by as much as 90%) (2). As GSH stores are depleted, increased levels of NAPQI-protein adducts form, and such adducts are thought to be an important contributor to APAP hepatotoxicity (3,7,8,10,11).
Pretreatment with N-acetylcysteine (7,9,10), a source of cysteine for GSH biosynthesis, attenuates APAP-induced hepatotoxicity. N-Acetylcysteine given soon after APAP (within 1-2 h) is highly protective against liver injury. However, this protection rapidly diminishes with time. Nonetheless, even when given as much as 2 h post-APAP administration, N-acetylcysteine affords some protection against hepatotoxicity (7).
In most tissues, the two-step biosynthesis of GSH is primarily limited by the activity of glutamate-cysteine ligase that carries out the first and rate-limiting step in GSH synthesis (13). Glutamate-cysteine ligase is a heterodimeric enzyme composed of catalytic (Gclc) and modifier (Gclm) subunits. All catalytic functions of the holoenzyme are carried out by Gclc, whereas Gclm influences the catalytic efficiency of Gclc by lowering the K m for glutamate and increasing the K i for GSH feedback inhibition.
To investigate the influence of GSH biosynthesis capacity on the extent of hepatotoxicant-induced injury we developed transgenic mice that conditionally overexpress Gclc, Gclm, or both subunits. Our initial attempts to generate constitutively expressing glutamate-cysteine ligase transgenic mice using either the cytomegalovirus or metallothionein promoters to drive increased glutamate-cysteine ligase subunit mRNA expression yielded small litter sizes and no transgene positive progeny. These results suggested that glutamate-cysteine ligase overexpression was incompatible with normal fetal development. Utilizing the liver-specific transactivator GLVP (14) we have successfully generated transgenic mice that conditionally overexpress Gclc, Gclm, or both transgenes. In this model system, mifepristone (also known as RU486), a progesterone antagonist, binds to the transgene-derived chimeric GLVP transactivator protein, causing it to translocate to the nucleus, where it binds to and transactivates target glutamate-cysteine ligase transgenes.
Glutamate-cysteine ligase is variably expressed in humans (15). Because GSH plays such an important role in detoxifying APAP, variation in glutamate-cysteine ligase expression may play a role in idiosyncratic reactions to APAP. This transgenic mouse model of inducible glutamate-cysteine ligase overexpression allowed us to directly test the hypothesis that glutamate-cysteine ligase expression influences susceptibility to APAP-induced liver injury.
Glutamate-cysteine Ligase Transgenic Mice-All animal procedures were carried out following protocols approved by the University of Washington Institutional Animal Care and Use Committee (IACUC). All mice were housed under specific pathogen-free conditions in microisolator caging. C57Bl/6 (75%) X C3H (25%) pronuclear embryos were collected on E0.5, following superovulation. E0.5 was taken as the morning of the day that copulation plugs were observed. Pregnant mare serum gonadotropin for superovulation was purchased from the National Hormone and Peptide Program (Torrance, CA). Constructs were injected into the pronucleus, and embryos were transferred into the oviducts of E0.5 pseudopregnant dams either on the day of injection or following overnight culture to the 2-cell stage and allowed to progress to term.
Genotyping of Gclc and Gclm Transgenic Mice-At 3-4 weeks of age tail biopsies were taken from pups and gluta-mate-cysteine ligase transgene status was determined by PCR using reverse primers designed to hybridize internally to either the Gclc (GAAGTAGCCTCCTTCCGGCG) or Gclm (CTGTGCAACTCCAAGGACGGA) cDNA sequences, and a forward primer (GAACACCGAGCGACCCTGCA) designed to hybridize to the thymidine kinase promoter region in the integrated plasmid fragment (Fig. 1).
Genotyping of GLVP Transgenic Mice-The same tail biopsies that were used to determine glutamate-cysteine ligase transgene status were used to identify the GLVP transgene status by PCR using GLVP forward (GACGCGCTA-GACGATTTC) and reverse (AGCAAAGAACTGGAG-GTG) primers (14).
Mifepristone Administration-In initial experiments, a single intraperitoneal injection of mifepristone (5 mg/kg in sesame oil; Sigma) was administered to mice 8 -9 h prior to sacrifice, to determine which founder strains were carrying responsive glutamate-cysteine ligase transgenes. In later experiments, animals from responsive strains were injected intraperitoneally 3 times with mifepristone (5 mg/kg each time at 24, 16, and 8 h prior to sacrifice). Control animals were administered vehicle only (sesame oil) at the same times.
Tissue Collection-Mice were sacrificed by CO 2 narcosis/ cervical dislocation. Livers were removed for biochemical, histological, and gene expression analyses.
Northern Blot Analysis-Total RNA was isolated from liver tissue using TRIzol reagent (Invitrogen) and subjected to Northern blot analysis using standard methods (19). Blots were hybridized with either Gclc or Gclm 32 P-labeled murine cDNA, corresponding to the amino acid coding regions (16,17) or ␤-actin cDNA probes, and relative mRNA levels were determined by phosphorimager analysis (GelDoc, Bio-Rad). Glutamate-cysteine ligase transcripts originating from the transgenes were distinguished from the endogenous glutamate-cysteine ligase transcripts based on their differential migration in agarose gels (different 3Ј non-coding region lengths).
Quantitative Real Time PCR (qRT-PCR)-Total liver RNA was extracted in TRIzol reagent and used to generate cDNA by reverse transcription using Superscript reverse transcriptase (Invitrogen). The cDNA was then used to quantitate the mRNA levels of Gclc and Gclm endogenous and transgenes using an ABI 7700 Sequence Analyzer (Applied Biosystems, Foster City, CA). The technique involved the use of sequence-specific fluorogenic probes that border exon/exon boundaries in the processed mRNAs. PCR primers and probes were selected using Primer Express 1.5 TM software (Applied Biosystems). A reference standard of normal mouse kidney RNA was serially diluted to derive a linear regression formula over 4 orders of magnitude that was then used to calculate and quantitate expression. Glyceraldehyde-phosphate dehydrogenase mRNA expression was used to normalize glutamate-cysteine ligase mRNA expression.
Endogenous transcripts were differentiated from transgenic sequences using primers specific for Gclc and Gclm and the bovine growth hormone polyadenylation region of the target gene (Table 1 and Fig. 1).
GCL Protein Expression-Proteins were extracted from mouse livers by sonication in 20 mM Tris, pH 7.4, 1 mM EDTA, 250 mM sucrose, 1 mM L-serine, 20 mM boric acid supplemented with protease inhibitors as previously described (20,21). Fifty g of soluble protein were subjected to Western analysis using anti-glutamate-cysteine ligase antisera generated in rabbits (22), with enhanced chemiluminescence detection (Amersham Biosciences). Exposed x-ray films were then subjected to image analysis (GelDoc) to determine the levels of Gclc and Gclm proteins.
Glutathione and Glutamate-cysteine Ligase Activity-Baseline and inducible glutamate-cysteine ligase activity and GSH levels in liver homogenates were assayed by HPLC or by fluorescence biochemical assay essentially as previously described (23)(24)(25).
APAP Administration-Prior to the administration of APAP (Sigma) or vehicle (normal saline), some mice were dosed 3 times with mifepristone dissolved in sesame oil as stated above and other mice received sesame oil alone. All animals in the study were fasted for 12 h prior to APAP injections. Mice from both the trigenic (Gclc/Gclm/GLVP) and control (Gclc/Gclm only) groups received APAP (300 mg/kg, intraperitoneal) 8 h after the last mifepristone (or sesame oil) injection. A second set of mice with similar genotypes received vehicle (saline) instead of APAP. Immediately after APAP (or saline) injections, food was returned to all cages. Six hours later, the mice were sacrificed by CO 2 narcosis/cervical dislocation, and livers were harvested.
Alanine Aminotransferase Assay-Immediately after sacrifice, a blood sample was taken via cardiac puncture. Serum alanine aminotransferase activity was evaluated spectrophotometrically using a commercially available kit (Sigma).
Histopathology-Hematoxylin and eosin (H&E)-stained liver sections were prepared using routine methods (Anatomical Pathology Services, University of Washington Medical Center, Seattle, WA). Scoring of tissue injury (on a scale of 0 to 5, with 0 ϭ no injury; 5 ϭ severe hepatocellular necrosis) was conducted by three independent observers blinded with respect to transgene status and treatment. Further evaluation of the liver sections was made by Dr. Robert Pierce, University of Rochester, a board certified pathologist.
Acetaminophen Protein and Glutathione Adducts-We modified the method of Muldrew et al. (26) to measure the levels of APAP protein and APAP glutathione adducts using HPLC with electrochemical detection. For glutathione APAP adducts, liver tissue was homogenized (1:5, w/v) in 10 mM sodium acetate, pH 6.5, and then centrifuged at 16,000 ϫ g for 20 min at 4°C. One hundred l of supernatant was combined with 100 l of 20% ice-cold trichloroacetic acid for 15 min, and then centrifuged for 5 min at 16,000 ϫ g. Fifty l of supernatant was then added to 950 l of 10 mM sodium acetate, pH 6.5, and aliquots were then assayed using reverse-phase HPLC with electrochemical detection. For protein APAP adducts, livers were homogenized as above in 10 mM sodium acetate and centrifuged at 16,000 ϫ g. Five hundred l of supernatant was then mixed 1:1 with 10 mM sodium acetate and subjected to dialysis (Slide-A-Lyser, Pierce; 3,500 MWCO) against 10 mM sodium acetate, pH 6.5, for 19 -22 h at 4°C. Samples were then diluted 1:3 in 10 mM sodium acetate buffer. Ten l of protease solution (200 mg/ml, Sigma) was then added to 1 ml of dialyzed sample and incubated at 50°C for 20 -24 h. One-hundred fifty l of digested sample was then mixed with 150 l of 20% trichloroacetic acid, placed on ice for 15 min, and then centrifuged at 16,000 ϫ g for 5 min at 4°C. Fifty l of supernatant was then added to 450 l of 10 mM sodium acetate buffer, and samples were analyzed by HPLC with electrochemical detection as above. APAP-glutathione adduct standards and APAP-cysteine adduct standards were generated by incubating 250 M GSH or 250 M cysteine with 250 M APAP for 4 h in the presence of NADPH (2 mM) and microsomes (400 g of protein/ml) con-FIGURE 1. Transgene system used to construct glutamate-cysteine ligase inducible transgenic mice. The GLVP transactivator is composed of a mutated progesterone receptor ligand binding domain, the yeast Gal4 protein DNA binding domain, and the herpes simplex virus VP16 transactivation domain, and is driven by the transthyretin gene promoter, giving liver-specific expression. Addition of mifepristone (RU486) induces dimerization and nuclear localization of GLVP, whereupon the transactivator binds to quadruplicated Gal4 DNA consensus sequences (17 X 4) residing 5Ј of the target transgene (Gclc or Gclm) and effects transcription using the thymidine kinase minimal promoter. Primers used to determine genotype of the mice are represented as F, whereas primers used for quantitative real-time PCR are designated ࡗ.

Mouse primers and probe sequences T m°C
taining recombinant human cytochrome P450 2E1 (BD Biosciences). Statistical Analyses-The results for GSH levels, glutamatecysteine ligase activity, glutathione and protein APAP adducts, alanine aminotransferase activity, and histopathology scores were analyzed by analysis of variance and Student's t test. Regression analyses and F tests were also performed on selected comparisons of these data. Differences yielding a p value of less than 0.05 were considered statistically significant.

Generation of Glutamate-cysteine Ligase Transgenic Mice-
Using standard techniques, we generated mice possessing Gclc or Gclm cDNA transgenes flanked by the 17 ϫ 4 Gal4 recognition elements and the minimal tk promoter (Fig. 1). Four Gclc and 7 Gclm transgenic founder lines were identified by PCR using reverse primers specific for the glutamate-cysteine ligase subunit cDNA portion of each plasmid construct, as well as a forward primer homologous to the tk promoter (present in both glutamatecysteine ligase subunit plasmid constructs; Fig. 1). These transgenes were inherited in a simple Mendelian manner, and there were neither gender bias of heredity in the progeny, nor fertility problems associated with the presence of any of the transgenes (data not shown).
Gclc and Gclm transgenic mice were crossed with mice carrying the liver-specific and mifepristone responsive transactivator GLVP. Subsequently, these two lines were intercrossed (Gclc/GLVP X Gclm/ GLVP) to generate trigenic mice (Gclc/Gclm/GLVP). All mice were backcrossed onto a C57Bl/6 genetic background for at least 6 generations prior to the initiation of the APAP exposures.
Characterization of Glutamatecysteine Ligase Overexpressing Transgenic Mice-The 4 Gclc/ GLVP and the 7 Gclm/GLVP founder lines were analyzed for inducible RNA transgene expression by administration of mifepristone. Initially, both Northern blot and qRT-PCR analyses were applied to detect both the endogenous and the transgenic Gclc and Gclm transcripts. After several concurring experiments qRT-PCR became the preferred method of detection and quantification of these transcripts (Fig. 2). Utilizing a primer sequence from the bovine growth hormone poly(A) 3Ј end of the target gene construct (Fig. 1), we were able to distinguish the transgenic transcripts from the endogenous glutamate-cysteine ligase transcripts. Several of our founder lines showed no detectable transgene mRNA expression.
Protein expression and glutamate-cysteine ligase activity were next evaluated in Gclc/GLVP bigenic, Gclm/GLVP bigenic, and Gclc/Gclm/GLVP trigenic mice that showed inducible transgene mRNA expression. There was an increase in Gclm and Gclc protein expression that was predicted by the presence of Gclc and Gclm transgene mRNA expression (Fig.  3). Interestingly, Gclm/GLVP bigenic mice treated with mifepristone exhibited an increase not only in Gclm protein but also appeared to have increased levels of Gclc protein 9 h after mifepristone treatment (even though Gclc transgene mRNA was not increased in these mice; Fig. 2). This may be due to a transient stabilizing effect of Gclm on Gclc protein in this strain of transgenic mice. Importantly, there was an increase in glutamate-cysteine ligase activity associated with mifepristone-induced increases in glutamate-cysteine ligase protein expression (Fig. 3D).
The results of early experiments with this transgenic model indicated that maximal glutamate-cysteine ligase protein expression in the transgenic mice was between 8 and 9 h after a single mifepristone injection. To achieve more consistency in glutamate-cysteine ligase induction and expression, we administered 3 consecutive 5 mg/kg doses of mifepristone given 8 h apart. With repeated mifepristone dosing there was again a notable increase in Gclm protein expression in Gclc/Gclm/ GLVP trigenic mice (but not in Gclc/Gclm bigenic mice) 8 h after the last mifepristone injection (Fig. 4). However, under these conditions Gclc protein expression was not elevated. We  Fig. 2. A, chemiluminescent image of Gclc and Gclm protein detected with anti-glutamate-cysteine ligase antisera 9 h after a single injection of 5 mg/kg mifepristone. B, mifepristone-mediated induction of Gclc protein expression. Gclc/Gclm/GLVP trigenic mice had significantly higher protein expression than controls (GLVP only; *, p Ͻ 0.05). The number of mice in genotype groups is indicated within each bar. C, mifepristone-mediated induction of Gclm protein expression. Both Gclm/GLVP bigenic and Gclc/Gclm/GLVP trigenic mice had significantly higher Gclm protein expression than controls (GLVP only; *, p Ͻ 0.05). The number of mice in genotype groups is indicated within each bar. D, mifepristone-mediated induction of liver glutamate-cysteine ligase activity. With a single mifepristone treatment of 5 mg/kg, Gclc/Gclm/GLVP trigenic mice show significantly higher levels of liver glutamate-cysteine ligase activity than control (GLVP only) mice 9 h after treatment (*, p Ͻ 0.05). The number of mice in genotype groups is indicated within each bar.
also noted that glutamate-cysteine ligase activity was elevated in female Gclc/Gclm/GLVP trigenic mice (but not males) with repeated doses of mifepristone (Fig. 5A). No significant increases in glutathione were noted. We found a similar effect in glutamate-cysteine ligase-transfected Hepa-1 cells (24), and this may be related to feedback inhibition of glutamate-cysteine ligase by GSH, preventing further increases in GSH content (13).
Effects of APAP on Glutamate-cysteine Ligase Activity and GSH Levels-APAP overdose is known to deplete liver GSH stores (2,7,8). We anticipated that mice overexpressing glutamate-cysteine ligase would either be resistant to APAP-induced GSH depletion or may replete GSH stores at a faster rate subsequent to APAP treatment. In a preliminary experiment we found that the kinetics of GSH depletion and repletion in mifepristone pretreated Gclc/Gclm bigenic (control) mice and Gclc/Gclm/GLVP trigenic mice after administration of APAP was similar in these two strains, although the trigenic mice were slightly more able to restore their GSH by 6 h (84%) relative to that seen in Gclc/Gclm mice (72%) (data not shown). Furthermore, although the number of mice was small (two male and two female mice of each genotype per time point) we saw no gender dependence on the kinetics of GSH depletion or repletion.
We next evaluated the effects of mifepristone induction of glutamate-cysteine ligase on the extent of APAP-induced liver injury. Because livers from the APAP-treated mice were not harvested until 14 h after the 3 rd dose of mifepristone, we wanted to determine whether glutamate-cysteine ligase activity was still elevated at this time. In fact, it was significantly higher in both male and female trigenic mice receiving mifepristone and APAP than in all the other groups, including Gclc/Gclm mice receiving both mifepristone and APAP (Fig. 6).
Induction of Glutamate-cysteine Ligase Activity and Protection against APAP-induced Liver Injury-Treatment with 300 mg/kg APAP alone resulted in a large increase in serum alanine aminotransferase levels (greater than 10-fold) in both Gclc/ Gclm mice and Gclc/Gclm/GLVP trigenic mice relative to control mice administered vehicle alone (Fig. 7). Male Gclc/Gclm mice receiving both mifepristone and APAP also had high serum alanine aminotransferase levels. Importantly, when APAP was administered to male Gclc/Gclm/GLVP trigenic mice that had been pretreated with mifepristone, serum alanine aminotransferase activity was significantly less than that seen in   male trigenic mice receiving APAP without mifepristone pretreatment, or in male Gclc/Gclm mice receiving APAP, irrespective of mifepristone pretreatment. Interestingly, APAP-treated female mice experienced only a moderate increase in alanine aminotransferase (relative to male mice, despite their relatively high glutamate-cysteine ligase activity). This gender-dependent sensitivity to APAP-induced liver injury has been recently noted by others (27). Moreover, female Gclc/Gclm/GLVP trigenic mice pretreated with mifepristone were not protected from APAP-induced liver injury (Fig. 7B). We saw no effect of mifepristone treatment alone on serum alanine aminotransferase activity in either Gclc/ Gclm mice or Gclc/Gclm/GLVP trigenic mice (data not shown).
Further support for the protective effect of high glutamate-cysteine ligase expression in male mice comes from a direct comparison of serum alanine aminotransferase activity with glutamate-cysteine ligase activity in APAP-exposed trigenic mice pretreated with mifepristone. There is a clear inverse relationship between glutamate-cysteine ligase activity and serum alanine aminotransferase levels after APAP treatment in male mice, and those mice with the highest glutamate-cysteine ligase activity showed the best protection from APAP-induced increases in serum alanine aminotransferase activity (Fig. 7C). No such relationship was observed in female trigenic mice (Fig. 7D). In addition, there was no correlation between glutamate-cysteine ligase and serum alanine aminotransferase activities in APAP-treated male or female trigenic mice that had not been pretreated with mifepristone, or in APAP-treated male or female Gclc/Gclm mice irrespective of mifepristone pretreatment (data not shown).
A spectrum of histopathological patterns of centrilobular injury was observed in the livers of mice treated with 300 mg/kg APAP, including hepatocellular glycogen depletion, Mice were treated with mifepristone (or sesame oil vehicle) and APAP (or saline) as described in the legend to Fig. 6. A, APAP treatment resulted in an increase in serum alanine aminotransferase in Gclc/Gclm bigenic male mice (irrespective of mifepristone pretreatment) and in Gclc/Gclm/GLVP trigenic male mice not pretreated with mifepristone. However, in APAP-treated Gclc/Gclm/GLVP trigenic male mice, pre-treatment with mifepristone resulted in significantly lower serum alanine aminotransferase activity. B, female mice exhibited less APAP-induced alanine aminotransferase activity than males, and mifepristone pretreatment did not result in a lowering of alanine aminotransferase activity in female Gclc/Gclm/GLVP trigenic mice. C, there exists an inverse relationship between glutamate-cysteine ligase activity and serum alanine aminotransferase activity in APAP-treated Gclc/Gclm/GLVP trigenic male mice, indicating more protection in mice with high glutamate-cysteine ligase activity. D, there was no significant correlation between glutamate-cysteine ligase activity and serum alanine aminotransferase activity in APAP-treated trigenic female mice. *, significantly different from control (saline only) mice (p Ͻ 0.01). #, significantly different from APAP-treated trigenic mice not pretreated with mifepristone, and from APAP-treated Gclc/Gclm bigenic mice, irrespective of mifepristone treatment (p Ͻ 0.01). The number of mice in each genotype is indicated above (controls) or within each bar. sinusoidal congestion and hemorrhage, hepatocyte microand macrovesicular steatosis, and necrosis. Representative photomicrographs of H&E-stained liver sections highlighting these patterns of injury are shown in Fig. 8. When analyzed according to transgene status, glutamate-cysteine ligase activity, and overall histopathological changes, it was evident that APAP-treated male trigenic mice that had been pre-treated with mifepristone experienced less injury than APAP-treated male trigenic mice not receiving mifepristone, or Gclc/Gclm male mice receiving both APAP and mifepristone (Fig. 9A). APAP-treated Gclc/Gclm female mice and Gclc/Gclm/GLVP trigenic female mice experienced a similar degree of histopatholigical change, which was overall less than that seen in males (Fig. 9B). We saw no effect of mifepristone treatment alone on histopathology in either Gclc/Gclm mice or Gclc/Gclm/GLVP trigenic mice (data not shown).
Glycogen depletion and microvascular injury/hemorrhage have been described as early histopathological lesions that are evident prior to any significant increase in serum transaminases (28,29), whereas steatosis and necrosis emerge later in the time course of injury and are associated with increases in serum transaminases (3,7,8,10,11). Most mice demonstrated a mixed pattern of injury after APAP exposure. Thus, each H&E-stained liver section was scored separately for glycogen depletion, hemorrhage, steatosis, and necrosis (scale 0 -3). When examined as a function of histopathological changes, it was again evident that mifepristone-pretreated male trigenic mice tended more toward early and less severe pathological changes (e.g. glycogen depletion), and significantly less necrosis than APAP-treated Gclc/Gclm male mice and APAP-treated trigenic male mice not pretreated with mifepristone (Fig. 9C). There was no apparent effect of transgene status or mifepristone pretreatment on the spectrum of histological features in APAPtreated female mice (Fig. 9D).
The level of APAP protein adducts after APAP overdose has been correlated with liver injury (8). Thus, we determined the level of residual GSH and APAP protein adducts in the livers of mifepristonepretreated Gclc/Gclm/GLVP male trigenic and Gclc/Gclm male bigenic mice, 2 h after intraperitoneal injection of 300 mg/kg APAP (Fig.  10). There was a modest correlation between glutamate-cysteine ligase activity and residual GSH in the liver, and there were significantly fewer APAP protein adducts in the trigenic mice than in the Gclc/Gclm mice, which indicates a more efficient scavenging of NAPQI in mice overexpressing glutamate-cysteine ligase. There was no correlation between glutamate-cysteine ligase activity and glutathione-APAP adducts, which may reflect the rapid disposal of these adducts at 2 h after treatment (data not shown).

DISCUSSION
The role of GSH as a cellular chemoprotectant and antioxidant is well established (1,7,30,31). Treatment with many xenobiotics results in GSH depletion due either to direct binding of toxic metabolites to GSH or through glutathione S-transferase-mediated pathways (1,31). When GSH is depleted, cells become more susceptible to damage from oxidative and nitrosative species sometimes leading to cell death (10,32). Moreover, various cytokines, environmental chemicals, and drugs have been shown to influence the levels of glutamate-cysteine ligase mRNA and protein expression, or to have direct effects on its activity (5,(33)(34)(35)(36)(37).
In this study, we report that high expression of glutamatecysteine ligase is protective against APAP-induced liver injury in male mice. Presumably this is due to the increased ability of these glutamate-cysteine ligase transgenic mice to resynthesize GSH subsequent to or during APAP-induced GSH depletion. Male glutamate-cysteine ligase trigenic mice showed diminished serum ALT activity, and less evidence of liver injury (as indicated by histopathological evaluation) after APAP treatment. Because GSH can limit glutamate-cysteine ligase activity via non-allosteric feedback inhibition (38,39), GSH is usually restored via increased glutamate-cysteine ligase activity released from the suppressive effects of high levels of GSH (13). In addition, GSH resynthesis can be enhanced through increased glutamate-cysteine ligase gene transcription, which results in increased mRNA and protein expression, and increased activity. We have previously shown in cultured mouse hepatocytes that increasing glutamate-cysteine ligase expression can increase GSH synthesis and attenuate the apoptotic effects of tumor necrosis factor-␣ (24). Using the inducible glutamate-cysteine ligase transgenic mouse model reported here, we were able to investigate protection afforded by increasing glutamate-cysteine ligase levels in vivo.
Utilizing the conditional transactivator GLVP, Wang et al. (14) developed a system in which expression of target genes in transgenic mice can be controlled via the administration of mifepristone. We generated a plasmid vector with either the Gclc or Gclm cDNA as targets for GLVP transactivation. We crossed mice carrying these target glutamate-cysteine ligase transgenes to GLVP transgenic mice in which the GLVP gene was driven by a liver-specific (transthyretin) promoter. We then determined which founder lines had increased levels of glutamate-cysteine ligase mRNA, protein, and activity following mifepristone induction. Administration of a single dose of mifepristone resulted in highly variable glutamate-cysteine ligase transgene mRNA expression (Fig. 2). Importantly, endogenous Gclc or Gclm mRNA expression was apparently unaffected by mifepristone administration (data not shown). Gclc protein levels were only slightly increased in Gclc/GLVP bigenic mice, whereas Gclm protein was significantly induced in Gclm/GLVP bigenic mice after receiving mifepristone. This stronger effect of the Gclm transgene remained when these bigenic mice were intercrossed to generate Gclc/Gclm/GLVP trigenic mice (Fig. 3). After administration of mifepristone, we found glutamate-cysteine ligase activity to be significantly higher in trigenic mice when compared with mice carrying just the GLVP transactivator gene (Fig. 3).
Kitteringham et al. (5) showed that APAP (530 mg/kg) decreased glutamate-cysteine ligase activity in the liver of CD-1 mice at 1 and 24 h even though both the mRNA and protein levels had increased. In our studies, livers were harvested 6 h after APAP administration, and at this time we found that Gclc/ Gclm control and Gclc/Gclm/GLVP trigenic mice receiving APAP alone had glutamate-cysteine ligase activity similar to control mice (Fig. 6). In addition, the mice used here are on the C57Bl/6 genetic background, whereas the studies of Kitteringham et al. (5), were performed in a different mouse strain (CD-1). When trigenic mice received mifepristone in addition to APAP there was a significant increase in glutamate-cysteine ligase activity (Fig. 6). FIGURE 9. Histopathology of APAP-induced liver injury. Mice were treated with mifepristone and APAP as described in the legend to Fig. 6. A, total liver injury scores (0 ϭ no injury, 5 ϭ significant centrilobular necrosis) in control (saline-treated) male mice, APAP-treated Gclc/Gclm male mice Ϯ mifepristone pretreatment, and APAP-treated Gclc/Gclm/GLVP male trigenic mice Ϯ mifepristone pretreatment. B, total liver injury scores in control (salinetreated) female mice, APAP-treated Gclc/Gclm female mice Ϯ mifepristone pretreatment, and APAP-treated Gclc/Gclm/GLVP trigenic female mice Ϯ mifepristone pretreatment. C, prevalence of early and late markers of liver injury among APAP-treated male mice. Histopathological features associated with mild injury (glycogen depletion), intermediate injury (microvascular injury/hemorrhage and steatosis), and marked injury (necrosis). Each slide was scored (scale 0 -3) for these aspects of injury, and averaged for each treatment group. Male Gclc/Gclm/GLVP trigenic mice pretreated with mifepristone display predominantly early and mild forms of liver injury after APAP treatment. **, significantly different from the necrosis score of trigenic mice not pretreated with mifepristone (p Ͻ 0.01) and of Gclc/Gclm mice not pretreated with mifepristone (p Ͻ 0.05). D, prevalence of early and late markers of liver injury among APAP-treated female mice. No significant effect of glutamate-cysteine ligase transgene induction on the distribution of APAPinduced histopathological changes was observed in female mice. The number of mice in each genotype and treatment is indicated above (controls) or within each bar in A and B, which are also the same numbers for APAP-treated mice represented in C and D, respectively.
When mice are administered high doses of APAP, GSH is rapidly depleted due to covalent binding to the APAP metabolite NAPQI (2,7,11,12). Subsequently, NAPQI is able to bind to cysteine on proteins. We found that residual GSH levels were correlated with glutamate-cysteine ligase activity, and an inverse relationship between glutamate-cysteine ligase activity and APAP-cysteine protein adducts 2 h after APAP dosing (Fig.  10). This suggests that the increased glutamate-cysteine ligase activity present in male trigenic mice may have resulted in less initial protein damage and therefore protected them from APAP-induced necrotic injury.
Surprisingly, Rzucidlo and colleagues (40) have shown that glutathione synthetase transgenic mice are more susceptible to APAP than their wild type littermates, despite the fact that they have increased GSH in their livers. The increased sensitivity to APAP was attributed to higher levels of kidney injury, possibly because of increasing amounts of APAP-glutathione conjugates being formed in the liver, which were then transported into the kidney causing injury there. We have not yet investigated whether kidney injury also occurs in glutamate-cysteine ligase trigenic mice after APAP exposure, but such studies are planned for the near future.
Both oxidative and nitrosative stress have been implicated in APAP-induced hepatotoxicity (9,10,(41)(42)(43). The metabolism of APAP by cytochrome P-450 enzymes to NAPQI is an oxidative event but there appears to be little evidence that this contributes to hepatic necrosis per se (8,10). Once NAPQI is formed it readily binds to thiols and depletes the cell of GSH by up to 90%. The remaining NAPQI forms adducts with cellular proteins especially in centrilobular hepatocytes leading to necrosis (44,45). Because most adduct formation occurs by 2 h after APAP administration (44,46,47), glutamate-cysteine ligase likely protects against NAPQI protein adduct formation via an increased supply of GSH resulting in less steatosis and necrosis.
The mechanisms by which APAP induces hepatocellular necrosis is still uncertain, but many causative factors have been rejected including NADPH oxidase (48), lipid peroxidation (49), and Kupffer cell activation (50). In addition, NAPQI protein adduct levels per se do not necessarily lead to hepatotoxicity (10). Recent studies support the view that oxidative stress may not be the only factor responsible for APAP toxicity. For example, several studies support the hypotheses that peroxynitrite formation (50 -53) and mitochondrial permeability transition (10,43) are critical for APAP-induced hepatocellular necrosis.
The reasons for the increased APAP sensitivity of male mice relative to female mice are not clear. Recently, Dai and colleagues (27) also reported that female mice were relatively resistant to APAP-induced liver injury, despite the fact that APAP-protein adducts were similar in male and female mice. Differences in the amount of GST present in their livers might be one attributing factor in the differential sensitivity of male and female mice. Our data suggests that GSH synthesis by glutamate-cysteine ligase may also be a gender-dependent protective factor, whereby males are much more dependent upon GSH synthesis for protection from APAP than female mice.
Because glutamate-cysteine ligase activity can be limiting for APAP-induced liver injury in mice, and because its expression is known to vary among humans (15), it may also be an important determinant of the extent of APAP-induced liver injury in humans. Recently, it has been shown that polymorphisms in both GCLC and GCLM are associated with impaired glutamate-cysteine ligase induction, lower levels of GSH, and FIGURE 10. Liver residual GSH and APAP-cysteine protein adducts in APAP-treated male Gclc/Gclm bigenic and male Gclc/Gclm/GLVP trigenic mice. Mice received 3 doses of mifepristone (5 mg/kg) at 8-h intervals. APAP (300 mg/kg intraperitoneal) was administered 8 h after the final mifepristone dose. Mice were sacrificed at 2 h post-APAP administration. A, residual GSH levels in APAP-treated mice. Although GSH appeared to be lower in Gclc/Gclm bigenic mice, this was not statistically different from Gclc/Gclm/GLVP trigenic mice (p ϭ 0.08; n ϭ 6 for both groups). B, liver APAP-cysteine protein adducts were lower in Gclc/Gclm/ GLVP trigenic mice than those in Gclc/Gclm mice (**, p Ͻ 0.01; n ϭ 6 for both groups). C, there was a positive correlation between glutamate-cysteine ligase activity and residual GSH level after APAP treatment (p Ͻ 0.05, f-test; filled circle, APAP-treated Gclc/Gclm mice; filled square, APAP-treated Gclc/Gclm/GLVP trigenic mice). D, there was a negative correlation between glutamate-cysteine ligase activity and APAP-protein adducts (p Ͻ 0.01, f-test; filled circle, APAP-treated Gclc/Gclm mice; filled square, APAP-treated Gclc/Gclm/GLVP trigenic mice).
increased risk of cardiovascular disease (54 -56). These polymorphisms may also be important determinants of APAP-induced liver injury. If so, this suggests that administration of GSH ethyl ester, or other GSH pro-drugs that by-pass the need for de novo GSH synthesis may be a more effective therapy in APAP overdose than N-acetylcysteine, especially for those people carrying glutamate-cysteine ligase alleles associated with lower GSH synthetic capacity.