A20 protects mice from lethal liver ischemia/reperfusion injury by increasing peroxisome proliferator-activated receptor-α expression

Authors

  • Haley E. Ramsey,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Current affiliation:
    1. Medical University of Vienna, Vienna, Austria
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Cleide G. Da Silva,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Christopher R. Longo,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Current affiliation:
    1. Institute Necker, Hospital Necker, Paris, France
    Search for more papers by this author
  • Eva Csizmadia,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Peter Studer,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Virendra I. Patel,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Scott M. Damrauer,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Jeffrey J. Siracuse,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Soizic Daniel,

    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Christiane Ferran

    Corresponding author
    1. Division of Vascular Surgery and the Center for Vascular Biology Research, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    2. Division of Nephrology, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    3. Transplant Institute, Departments of Surgery and Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
    • Center for Vascular Biology Research, RN #370/G, Beth Israel Deaconess Medical Center, 99 Brookline Avenue, Boston MA 02215
    Search for more papers by this author
    • Telephone: 617-6670440; FAX: 617-667-0445


Abstract

The nuclear factor-κB inhibitory protein A20 demonstrates hepatoprotective abilities through combined antiapoptotic, anti-inflammatory, and pro-proliferative functions. Accordingly, overexpression of A20 in the liver protects mice from toxic hepatitis and lethal radical hepatectomy, whereas A20 knockout mice die prematurely from unfettered liver inflammation. The effect of A20 on oxidative liver damage, as seen in ischemia/reperfusion injury (IRI), is unknown. In this work, we evaluated the effects of A20 upon IRI using a mouse model of total hepatic ischemia. Hepatic overexpression of A20 was achieved by recombinant adenovirus (rAd.)-mediated gene transfer. Although only 10%-25% of control mice injected with saline or the control rAd.β galactosidase survived IRI, the survival rate reached 67% in mice treated with rAd.A20. This significant survival advantage in rAd.A20-treated mice was associated with improved liver function, pathology, and repair potential. A20-treated mice had significantly lower bilirubin and aminotransferase levels, decreased hemorrhagic necrosis and steatosis, and increased hepatocyte proliferation. A20 protected against liver IRI by increasing hepatic expression of peroxisome proliferator-activated receptor alpha (PPARα), a regulator of lipid homeostasis and of oxidative damage. A20-mediated protection of hepatocytes from hypoxia/reoxygenation and H2O2-mediated necrosis was reverted by pretreatment with the PPARα inhibitor MK886. In conclusion, we demonstrate that PPARα is a novel target for A20 in hepatocytes, underscoring its novel protective effect against oxidative necrosis. By combining hepatocyte protection from necrosis and promotion of proliferation, A20-based therapies are well-poised to protect livers from IRI, especially in the context of small-for-size and steatotic liver grafts. Liver Transpl 15:1613–1621, 2009. © 2009 AASLD.

Ischemia/reperfusion injury (IRI) of the liver is a major cause of hepatic failure in the context of liver resection surgery, liver trauma, and liver transplantation.1, 2 The clinical outcome of IRI is largely dependent on the severity of the insult and on the pre-existing condition of the liver. For instance, steatosis, which affects 25% of the western population, is a critical aggravating factor for IRI.3

The precise sequence of events leading to liver IRI has not been completely elucidated. There is, however, substantial evidence that ischemia depletes ATP levels (energy) and results in the activation of Küpffer cells, production of proinflammatory cytokines such as tumor necrosis factor (TNF), generation of reactive oxygen species (ROS), and perturbation of the hepatic microcirculation.4 These events lead to a massive inflammatory response occurring as a result of the activation of nuclear factor kappa B (NF-κB).5 Activation of NF-κB increases the expression of adhesion molecules on sinusoidal endothelial cells and the production of chemokines, which, upon reperfusion, recruit neutrophils to the site of injury. Through releasing ROS and proteases, neutrophils further amplify liver damage by promoting hepatocyte necrosis and apoptosis.6, 7 Accordingly, antioxidant therapy limits IRI.8, 9

We have identified the zinc finger protein A2010 as a critical component of the physiologic “hepatoprotective” armamentarium of the hepatocyte. A20 is induced in hepatocytes as part of the protective response to inflammatory insults or to liver resection.11, 12 Up-regulation of A20 in response to injury serves a dual cytoprotective function. A20 protects hepatocytes from TNF-mediated apoptosis by interrupting the activation of the caspase cascade at the level of the initiator caspase-8, thus safeguarding mitochondrial integrity.11–13 A20 also curtails inflammation by inhibiting NF-κB activation, either through its association with IκB kinase-γ/NF-κB essential modifier (IKKγ/NEMO) within the signalosome or through its ubiquitin-editing functions.11, 12, 14, 15 A20 knockout mice are born cachectic and die within 3 weeks of birth as a result of unfettered liver inflammation, indicating the high rank held by A20 in the hierarchy of physiologic anti-inflammatory responses, particularly in hepatocytes.16 A20 is itself a NF-κB–dependent gene and as such is part of a negative regulatory feedback loop aimed at re-establishing homeostasis.17 Furthermore, induction of A20 following loss of liver mass promotes liver regeneration by decreasing the transcription of the cyclin-dependent kinase inhibitor (CDKI) p21waf.12

The combined antiapoptotic, anti-inflammatory and pro-proliferative functions of A20 in hepatocytes translate into in vivo protection of mice from toxic hepatitis induced by D-galactosamine/lipopolysaccharide11 and from fulminant hepatic failure following lethal radical (87%) hepatectomy.12 Interestingly, transcription of A20 is inhibited in small-for-size liver grafts, indicating that lower expression of A20 may be an important pathogenic contributor to their increased susceptibility to IRI.18 In this work, we directly evaluated the effects of A20 upon IRI and oxidative damage to the liver, by using a mouse model of total hepatic ischemia.

Abbreviations

βgal, βgalactosidase; BrdU, 5 bromo-2-deoxyuridine; CDKI, cyclin-dependent kinase inhibitor; FA, fatty acids; HPF, high-power field; IRI, ischemia reperfusion injury; MOI, multiplicity of infection; NF-κB, nuclear factor kappa B; PPARα, peroxisome proliferator-activated receptor alpha; rAd., recombinant adenovirus; ROS, reactive oxygen species; SEM, standard error of the mean.

MATERIALS AND METHODS

Reagents

All reagents were purchased from Sigma (St. Louis, MO), unless otherwise noted. All culture media were purchased from Invitrogen, Carlsbad, CA.

Cell Culture and Generation of Recombinant Adenoviruses

Mouse hepatocytes (NMuLi, CRL-1638) were purchased from the American Type Culture Collection (ATCC, Manassas, VA). Primary mouse hepatocytes from 6-week-old to 7-week-old BALB/c mice (Taconic, Germantown, NY) were isolated and cultured according to a modified Seglen's protocol.19 Recombinant adenoviruses encoding A20 (rAd.A20) or β-galactosidase (rAd.βgal) were generated and titered using the human embryonic kidney cell line HEK 293, as described.11 NMuLi cells and primary hepatocytes were infected with rAd. at a multiplicity of infection (MOI) of 25-100 plaque-forming units (pfu) per cell to achieve high expression of the transgene in more than 98% of cells 24-48 hours after infection with minimal toxicity.11 All experiments utilized two control groups: nontransduced and rAd.βgal-transduced cells.

Mouse Ischemia Reperfusion Model

Expression of A20 in the livers of 8-week-old BALB/c mice was achieved by rAd-mediated gene transfer through penile vein injections. We had previously demonstrated that intravenous injection of 1 × 109 pfu per mouse leads to the expression of the transgene in hepatocytes, peaking 5 days following gene transfer.11 Nontransduced livers and livers transduced with rAd.βgal were used as controls. Five days after gene transfer, ischemia was achieved by resecting the minor hepatic lobes (right middle, right lateral, pyriform process, and caudate lobes) leaving two-thirds of the original liver mass, followed by cross-clamping the hilum of the remaining lobes (right medial, left medial, and left lateral lobes) for 70 minutes prior to organ reperfusion. Animals received humane care according to the criteria outlined in the “Guide for the Care and Use of Laboratory Animals”. All procedures were approved by the Institutional Animal Care Committee.

Aspartate Aminotransferase and Bilirubin Measurements

Serum aspartate aminotransferase (AST) and bilirubin levels were measured in mouse sera using the Infinity Reagent System (Thermo Scientific, Waltham, MA) and a colorimetric assay, respectively (Sigma).

Histology and Immunochemistry

Liver tissue samples were recovered and fixed in 10% buffered formalin. Five-micrometer sections were stained with hematoxylin and eosin (H&E) for morphologic evaluation. Hemorrhagic necrosis and steatosis were graded as absent, mild, or extensive. Immunohistochemistry analysis was performed using antibodies to 5-bromo-2-deoxyuridine (BrdU; BD Biosciences, San Jose, CA), CDKI p21waf1 (BD, Biosciences), PPARα (N-19), p65 (C-20), and A20 (R-20) from Santa Cruz Biotechnology (Santa Cruz, CA). BrdU and CDKI p21waf1 immunostaining was evaluated by counting the number of positive cells per high-power field (HPF). PPARα and A20 immunostaining were graded as absent, moderate, or intense. Apoptosis was evaluated using the ApopTag assay (Oncor, Gaithersburg, MD). The p65 immunostaining was recorded as cytoplasmic or nuclear. We used appropriate secondary antibodies and isotype-specific negative controls for each antibody. A minimum of three HPF/slide were analyzed in a blinded fashion by C.F.

In Vitro Hypoxia/Reoxygenation and Evaluation of Cell Death

Hepatocytes were subjected to 60 minutes of hypoxia (defined as less than 1% O2, 5% CO2, and 94% of N2) using a controlled oxycycler chamber (Biospherix, Redfield, NY) followed by reoxygenation in 5% CO2 and 95% air for 30 and 120 minutes. We measured cell viability by Trypan blue exclusion.

Western Blot Analysis

Total cell and nuclear protein extracts from nontransduced, rAd.βgal-transduced, and rAd.A20-transduced hepatocytes were analyzed by western blot (WB) for the expression of PPARα using a polyclonal rabbit anti-PPARα antibody (Biovision Incorporated, Mountain View, CA). Immunoblotting for the housekeeping gene GAPDH (glyceraldehyde 3-phosphate dehydrogenase) was used to correct for loading (Chemicon International Inc., Temecula, CA). Appropriate secondary antibodies were purchased from Pierce (Rockford, IL).

Gene Chip Arrays

We performed gene microarray analysis at the Beth Israel Deaconess Medical Center Genomic Center using the Affymetrix GeneChip Mouse Genome 430.2.0 Array from Affymetrix (Santa Clara, CA). This chip provides a comprehensive mouse genome expression by covering more than 39,000 transcripts. The scanned array images were analyzed by dChip.20 Samples were compared using the lower confidence bound of the fold change. If the 90% lower confidence bound of the fold change between samples was above 2, the corresponding gene was considered differentially expressed. We extracted total messenger RNA (mRNA) from livers transduced with rAd.β-gal and rAd.A20 using the RNAeasy extraction kit (Qiagen Inc., Valencia, CA). We included six mice per group and per time-point. Two GeneChip per group and per time-point were blotted with RNA pooled from three animals.

Statistical Analysis

Quantitative data were expressed as mean ± standard error of mean (SEM). Statistical analysis was performed using analysis of variance and the Tukey-Kramer multiple comparisons test. Paired statistics were applied when appropriate. Survival curves were calculated using the Kaplan-Meier method and compared using the log-rank test. P values were adjusted for multiple comparisons using the Bonferroni correction. P < 0.05 was considered to be statistically significant.

RESULTS

Overexpression of A20 in Mouse Livers Protects from Lethality Following Prolonged IRI

To determine whether A20 can affect the outcome of severe liver IRI, we overexpressed A20 by intravenous administration of 1 × 109 pfu of rAd.A20 per mouse weighing 25-30 g. Five days following gene transfer, the mice were subjected to prolonged, total hepatic ischemia, as described above. The rAd.A20-treated mice had a significantly better survival rate (67%) 2 days following this procedure as compared to saline-treated (23%; P < 0.01) and rAd.βgal-treated (9%; P < 0.001) mice (Fig. 1A). All surviving mice were monitored for 6 months and showed no evidence of disease or development of tumors in the liver.

Figure 1.

Expression of A20 in the mouse liver decreases lethality and protects from liver damage following prolonged and lethal IRI. (A) Survival data following prolonged ischemia in saline-treated, rAd.A20-treated, and rAd.βgal-treated mice demonstrate that expression of A20 provides a significant long-term survival advantage in comparison to saline and βgal controls (n = 11-13 mice/group). (B) A20 prevents hepatocellular damage following extended hepatectomy. Mice treated with rAd.A20 have significantly (P < 0.01 and P < 0.001) lower serum AST levels 24 hours following prolonged IRI as compared to saline-treated and rAd.βgal-treated mice, respectively (n = 3 mice/time-point/group). (C) Similarly, A20 preserves bilirubin clearance following IRI. Mice expressing A20 have significantly lower total bilirubin levels 24 hours following IRI, as compared to saline-treated and rAd.β-gal treated mice (P < 0.001; n = 3 mice/time-point/group). Data are presented as mean ± SEM.

Significantly improved survival rate observed in rAd.A20-treated mice was associated with decreased liver damage following IRI. In the saline-treated and rAd.βgal-treated mice, AST levels rose from 174 ± 3 IU/mL and 320 ± 18 IU/mL before IRI to 1343 ± 81 IU/mL and 3291 ± 237 IU/mL, 24 hours after IRI, respectively (n = 3 mice/time-point/group; Fig. 1B). Significantly lower AST levels were noted in mice expressing A20 in their livers; AST levels rose from 190 ± 11 to 491 ± 85 in these mice (n = 3; P < 0.01 versus saline and P < 0.001 versus rAd.βgal; Fig. 1B). The tendency for higher AST levels in rAd.βgal-treated mice as opposed to the saline-treated group, which was especially significant after IRI as tested by paired analysis of the samples (P < 0.001), is a likely reflection of adenoviral toxicity.12 Mice treated with rAd.A20 also had significantly lower total serum bilirubin levels (1.7 ± 0.3 mg/dL) as compared to saline-treated (3.5 ± 0.2 mg/dL) and rAd.βgal-treated mice (4 ± 0.4 mg/dL) 24 hours after IRI, indicating preservation of bilirubin conjugation and excretion by hepatocytes (P < 0.001 versus both saline and rAd.βgal; n = 3 mice/time-point/group; Fig. 1C). Severe hypoglycemia, a feature of fulminant hepatic failure following IRI, was also totally averted in mice treated with rAd.A20 as compared to controls (data not shown). Interestingly, we did not detect any A20 immunostaining in hepatocytes of saline-treated mice prior to IRI, whereas a number of A20-positive hepatocytes were present in surviving saline-treated mice 24 hours after IRI (Fig. 2; n = 5 animals/group). This confirms that A20 is part of the response to injury that may have contributed to the recovery of these mice.

Figure 2.

Immunohistochemistry analysis of A20 expression in mouse livers following IRI. Representative immunohistochemistry sections from saline-treated livers demonstrate no A20 expression prior to IRI as opposed to substantial number of A20-positive hepatocytes 24 hours following IRI (brown staining). A20 expression remains limited to certain hepatocytes as opposed to generalized expression in rAd.A20-transduced livers (original magnification, 200×).

A20-Mediated Protection from IRI Is Associated with Preserved Liver Architecture, Decreased Hemorrhagic Necrosis and Steatosis, Decreased Inflammation, and Increased Hepatocyte Proliferation

We assessed liver architecture and morphology by histopathology before and 48 hours after IRI in saline-treated, rAd.A20-treated, and rAd.βgal-treated mice. All slides were reviewed by E.C. and in a blinded fashion by C.F. There were no significant differences between experimental groups before IRI (Fig. 3; n = six mice/group for A20 and βgal and five mice/group for saline). Extensive hemorrhagic necrosis and steatosis were present in saline-transduced and rAd.βgal-transduced livers 48 hours following IRI (Fig. 3; n = 6 animals for saline and 5 for rAd.βgal). Overexpression of A20 in mouse livers substantially decreased the incidence of hemorrhagic necrosis and microvesicular steatosis (graded as absent to moderate) and maintained liver architecture (Fig. 3; n = 6). Notably, rAd.βgal-treated mice fared worse for all measured parameters in comparison to saline-treated mice, likely as a result of additional adenoviral toxicity.21 This indicates that A20 overexpression not only protected mice from the untoward effects of total and prolonged IRI but also blunted the inflammatory insult secondary to the adenoviral effect.

Figure 3.

A20 preserves liver morphology and limits damage following prolonged and lethal ischemia reperfusion injury (IRI). Saline-treated and rAd.βgal-treated mice demonstrate extensive hemorrhagic necrosis (H/N) and steatosis (S) at 24 hours after IRI. In contrast, mice treated with rAd.A20 display only minimal steatosis with mild sinusoidal congestion and hemorrhage. Liver morphology before IRI remains within normal limits in all groups. However, there was already increased mitosis (M) in rAd.A20-transduced livers, likely reflecting repair in response to rAd. toxicity. Three mice were evaluated in each group, and three HPF/sample/mouse were analyzed. Data shown are representative of all tissue samples analyzed at a magnification of 200×.

Livers overexpressing A20 demonstrated a markedly improved proliferative index, as evidenced by increased staining for BrdU, a synthetic nucleoside analogue to thymidine that incorporates into newly synthesized DNA in replicating cells, 24 hours after IRI. We observed 55.3 ± 9.1 BrdU-positive hepatocytes/HPF in rAd.A20-treated livers as compared to 0.8 ± 0.2/HPF in rAd.βgal-treated livers (P < 0.001; n = 3 mice/group; Fig. 4). Moreover, there was increased staining for the CDKI p21waf1 in livers of rAd.βgal-treated mice (>60/HPF) versus almost no positive cells (0-1/HPF) in livers of Ad.A20-treated mice (P < 0.001; n = 3 animals/group; Fig. 4). The protective effect of A20 against IRI was also related to decreased inflammation. Immunostaining for p65/RelA showed a substantial number of p65-positive nuclei in hepatocytes of control livers 24 hours after IRI, indicating p65 translocation to the nucleus and NF-κB activation. In contrast, p65 immunostaining remained cytoplasmic in hepatocytes of A20-treated mice after IRI (n = 3 mice/group; Fig. 5).

Figure 4.

A20-mediated protection from ischemia reperfusion injury is associated with increased hepatocyte proliferation. Mice treated with rAd.A20 have a significant (P < 0.001) proliferative advantage over rAd.βgal-treated mice 24 hours after extended IRI as determined by BrdU staining. This correlates with a significant decrease of immunostaining for the cyclin-dependent kinase inhibitor (CDKI) p21waf1 in rAd.A20 as opposed to rAd.βgal-treated mice. BrdU-positive and p21waf1-positive nuclei demonstrate the typical brown staining. Three mice were evaluated in each group; at each time-point, three HPF/sample/mouse were analyzed at a magnification of 200×. Data shown are representative of all tissue samples analyzed.

Figure 5.

Expression of A20 in the mouse liver blocks p65 translocation to the nucleus in hepatocytes, following IRI. Immunohistochemistry analysis demonstrates cytoplasmic p65 immunostaining in hepatocytes of saline-treated and A20-treated mice prior to IRI. p65 immunostaining becomes massively nuclear in the hepatocytes of saline-treated mice 24 hours after IRI, indicating NF-κB activation. In contrast, p65 immunostaining remains cytoplasmic in A20-expressing hepatocytes 24 hours following IRI. Three mice were evaluated in each group; at each time point, three to six HPF/sample/mouse were analyzed at a magnification of 200×. Data shown are representative of all tissue samples analyzed.

Expression of A20 in Mouse Livers Increases the Expression of PPARα in Hepatocytes

Based on gene chip analysis that compared liver tissue taken from rAd.A20-treated mice and rAd.βgal-treated mice before and 24 hours following extended hepatectomy, we had observed 2.6-fold higher levels of PPARα mRNA in livers from rAd.A20-treated mice as compared to livers from rAd.βgal-treated mice prior to resection (n = 2 gene microarrays/time-point/group; manuscript in preparation). These data were validated by immunohistochemistry in our model of IRI. Our results clearly demonstrate a drastic increase of PPARα immunostaining, mostly detected in nuclei of hepatocytes, in rAd.A20-treated mice as compared to saline-treated and rAd.βgal-treated mice prior to any IRI (Fig. 6A; n = 3 mice/group). This effect persisted 24 hours after IRI. We confirmed this effect of A20 on PPARα expression in primary mouse hepatocyte cultures. In brief, primary hepatocytes isolated from 4-week-old to 6-week-old C57BL/6 mice were transduced with rAd.A20 or rAd.βgal at a MOI of 100 or left nontransduced, and PPARα protein levels were evaluated 48 hours later in total and nuclear cell extracts by western blot analysis. Our data indicate that mere overexpression of A20 in hepatocytes substantially increases PPARα protein levels relative to that of nontransduced and rAd.βgal-transduced hepatocytes (Fig. 6B; n = 3).

Figure 6.

Expression of A20 in the mouse liver increases the expression of PPARα in hepatocytes. (A) Immunohistochemistry analysis demonstrates intense staining for PPARα in rAd.A20-treated livers as opposed to moderate expression in saline-treated and rAd.βgal-treated mice, before and 24 hours after IRI. Three mice were evaluated in each group; at each time-point, three to six HPF/sample/mouse were analyzed at a magnification of 200×. Data shown are representative of all tissue samples analyzed. (B) Western blot analysis of PPARα in primary hepatocytes using total and nuclear cell lysates demonstrates higher expression in rAd.A20-transduced hepatocytes as compared to nontransduced (NT) and rAd.βgal-transduced hepatocytes. Cells were recovered 48 hours following transduction with rAd. GAPDH was analyzed to correct for sample loading. Data shown are representative of three independent experiments.

A20 Protects Hepatocytes from Hypoxia/Reoxygenation and H2O2-mediated Necrosis in a PPARα-Dependent Fashion

We set up two in vitro systems mimicking the oxidative stress incurred by hepatocytes during IRI in order to further examine the role of A20-induced PPARα up-regulation in mediating hepatocyte survival. These included sequential incubation of hepatocytes in hypoxia followed by incremental periods of reoxygenation and treatment of hepatocytes with H2O2. NMuLi cells transduced with rAd.A20 at a MOI of 100 were significantly protected from hypoxia/reoxygenation-mediated necrosis (Fig. 7A; n = 6). In nontransduced and rAd.βgal-transduced hepatocytes, the percentage of cell death rose from 17.4% ± 0.8% and 17.4% ± 0.9% before reoxygenation to reach 26.2% ± 1.8% and 27.7% ± 1.6% at 30 minutes and 33.9% ± 2.4% and 30.5% ± 3% at 120 minutes following reoxygenation, respectively. In contrast, in A20-expressing hepatocytes, the percentage of cell death insignificantly increased from 17.1% ± 1.3% before reoxygenation to 18% ± 0.8% and 21.7% ± 1.1%, at 30 and 120 minutes after reoxygenation (P < 0.05 and P < 0.001 for A20 versus nontransduced, respectively; P < 0.01 and P < 0.05 for A20 versus βgal at 30 and 120 minutes, respectively). Remarkably, pretreatment of NMuLi cells with 20 μmol/L of the PPARα inhibitor MK88622 totally abolished the protective effect of A20 against hypoxia/reoxygenation-mediated necrosis of cells. The percentage of cell death in NMuLi cells transduced with rAd.A20 and pretreated with MK886 was comparable or even slightly greater than that of controls, rising from 18.2% ± 2.3% before reoxygenation to 31.1% ± 2.3% and 34.4% ± 4% at 30 and 120 minutes after reoxygenation (Fig. 7B; n = 3).

Figure 7.

A20 protects hepatocytes from hypoxia/reoxygenation and H2O2 mediated cell death through a PPARα-dependent mechanism. (A) A20-expressing NMuLi cells are significantly protected from death induced by hypoxia followed by reoxygenation (reO2) as compared to nontransduced (NT) and rAd.βgal-transduced cells. Cell death is determined by Trypan blue staining before, 30 minutes and 120 minutes following reO2 and presented as mean ± SEM of six independent experiments. (B) Pretreatment of NMuLi hepatocytes with 20 μmol/L of the PPARα inhibitor MK886 for 60 minutes prior to hypoxia/reoxygenation totally abrogates the protective effect of A20. Cell death is determined by Trypan blue staining. Data shown are mean ± SEM of three independent experiments. (C) A20 expressing NMuLi cells are also significantly protected from H2O2-induced cell death as compared to nontransduced (NT) and rAd.βgal-transduced cells. Pretreatment of NMuLi cells with 20 μmol/L of the PPARα inhibitor MK886 for 60 minutes prior to treatment with H2O2 totally abrogates the protective effect of A20. Cell death is determined by Trypan blue staining before and 30 minutes following treatment with H2O2 (50 mmol/L). Data shown are mean±SEM of six independent experiments. The same volumes of ethanol, used as the solvent for MK886, were added to all wells that were not treated with MK886 in (B) and (C). P values are as follows: *P < 0.05; **P < 0.01; ***P < 0.001.

Similarly, treatment of NMuLi cells with 50 mmol/L of H2O2 increased the percentage of cell death in nontransduced and rAd.βgal-transduced NMuLi cells from 15.6% ± 4.1% and 18.8% ± 2.9% before treatment to 94.9% ± 1.4% and 91.6% ± 2.4% 30 minutes following treatment with H2O2, respectively (Fig. 7C; n = 6). In contrast, rAd.A20-transduced NMuLi hepatocytes were significantly protected from H2O2-mediated necrotic cell death. The percentage of cell death increased from 19.7% ± 3.7% before treatment with H2O2 to 61% ± 3% at 30 minutes after H2O2 treatment (P < 0.001 A20 versus nontransduced and P < 0.001 versus rAd.βgal-transduced). Here again, pretreatment of NMuLi hepatocytes with 20 μmol/L of MK886 totally abolished the protective effect of A20 against H2O2-mediated necrosis in these cells. The percentage of cell death in rAd.A20-transduced hepatocytes pretreated with MK886 was comparable to that of controls, rising from 21% ± 2.5% before treatment to 85% ± 1.7% at 30 minutes after treatment with H2O2 (Fig. 7C).

Taken together, these data constitute the first demonstration that A20 protects hepatocytes from oxidative necrosis through a novel PPARα-dependent mechanism.

DISCUSSION

IRI is the single most important factor incriminated in the early perioperative morbidity and mortality associated with liver transplantation and major hepatic resection. IRI accounts for >5% of primary nonfunction and 10%-30% of primary graft dysfunction in liver transplantation and is responsible for up to 81% of retransplantation during the first week after surgery.23 The incidence of IRI in orthotopic liver transplantation is further on the rise with the expansion of donor acceptance criteria to include marginal livers with increased susceptibility to IRI, such as those from non–heart beating donors that usually suffer a prolonged ischemia time, steatotic livers or small-for-size grafts from living donors.24, 25

The precise sequence of events leading to IRI is still a matter of debate. However, two major culprits have been identified including uncontrolled oxidative stress and unfettered inflammation, whose effects culminate in the necrotic cell death of hepatocytes, the hallmark of severe IRI.7, 9, 26 Therapies aimed at curbing inflammation and blocking oxidative necrosis of hepatocytes in the setting of IRI have been heavily explored.27–29

We reasoned that the NF-κB inhibitory protein A20, with its ability to restrain inflammation and to protect hepatocytes from TNF-mediated apoptosis would likely down-modulate the inflammatory arm of IRI.11, 12 We clearly achieved this objective by showing that activation of NF-κB following IRI is completely inhibited in A20-expressing livers as opposed to controls. However, the impact of A20 on the most significant pathogenic effector of IRI, i.e., oxidative damage, had not yet been explored. To address this quest, we chose a rather extreme model of IRI in order to mimic the worst clinical scenarios, including prolonged warm ischemia time doubled by the need for regeneration. As such, we expect that any benefit obtained in our model would translate into even better results in less severe cases of IRI. Our present data provide strong evidence that expression of A20 in mouse livers for a few days prior to subjecting them to partial hepatectomy followed by prolonged ischemia significantly protects the livers from IRI by inhibiting necrosis. This amounted to significantly less liver damage as assessed by lower aminotransferases and bilirubin levels and, more importantly, led to a clear survival advantage. Interestingly, endogenous A20 was up-regulated, 24 hours after IRI, in hepatocytes of surviving control mice treated with saline. This suggests that A20 is part of the regulatory response to IRI, because it is part of the regenerative response following hepatectomy.12 It is our hypothesis that the fate of control mice following severe IRI may be, at least in part, determined by the level of expression of A20. Unfortunately, this was difficult to prove in this model given the significant mortality rate of control mice that limited the analysis of A20 hepatic expression to the few surviving mice that are likely to express higher A20 levels. Loss-of-function experiments, using A20+/− heterozygote mice, in this total IRI model as well as in a partial IRI model are better suited to address this issue and are underway in the laboratory.

A20 protected hepatocytes in vitro from necrosis triggered by hypoxia/reoxygenation or by treatment with H2O2, providing us with a direct proof of this novel protective effect of A20 against oxidative necrosis, in hepatocytes. This result stands in contrast to A20-mediated sensitization of Hela cells to oxidative stress–mediated necrosis by terminating NF-κB–dependent survival signals.30 This discrepancy highlights major cell-type–specific differences in the function of the versatile A20 gene. This agrees with the opposite effects of A20 on cell proliferation in hepatocytes (pro-proliferative by decreasing the expression of p21waf1) versus smooth muscle cells (antiproliferative by affecting the same target CDKI p21waf1, but in an opposite manner).12, 31

To identify the mechanism involved in the protective effect of A20 against oxidative necrosis, we made use of gene chip data comparing rAd.A20-transduced to rAd.βgal-transduced livers (manuscript in preparation). Because oxidative necrosis in the setting of IRI is often the consequence of ATP depletion,7 we focused our search on genes that promote energy production. Our data indicated that PPARα, a gene that fulfills the criteria of an energy provider, was affected by A20 expression. We have evidence that PPARα mRNA and protein levels were higher in rAd.A20-transduced livers both before and following severe IRI.

PPARα, a member of the nuclear receptor superfamily, modulates triglyceride and high-density lipoprotein metabolism as well as the import of fatty acids (FAs) into mitochondria by up-regulating carnitine palmitoyl transferase type 1a and expression of genes involved in peroxisomal, microsomal, and mitochondrial β-oxidation systems.32, 33 We have gene microarray–based evidence that CPT1c is increased by 1.7-fold in A20-expressing livers as compared to controls, suggesting that PPARα is functional in our setting (data not shown). PPARα is mostly present in tissues characterized by high rates of FA metabolism such as the liver. Under optimal conditions, FAs, that are produced in response to injury, activate PPARα which then heterodimerizes with retinoid X receptor and liver X receptor, leading to the transcription of a number of genes that are involved in lipid turnover and peroxisomal and mitochondrial β-oxidation, resulting in generation of ATP. By promoting β-oxidation of FAs, PPARα enhances degradation of lipid-derived inflammatory mediators by shunting them away from lipid peroxidation, thus augmenting ATP generation, which is required to “fuel” liver repair and regeneration.34 In contrast, in conditions where PPARα function and/or expression is altered such as in steatotic livers, hepatitis C infection, small-for-size liver grafts, alcoholic hepatitis, or in the presence of an overwhelming liver injury,35–38 FA metabolism is deviated toward the accumulation of inadequately metabolized fat, favoring lipid peroxidation and ROS generation. As a consequence, ATP production is decreased and the demise of hepatocytes via necrotic cell death is increased, halting liver repair.39 Accordingly, mice with targeted disruption of PPARα show increased inflammation and necrosis following IRI and display delayed liver regeneration following partial hepatectomy.40, 41

Our in vitro results showing that the protective effect of A20 against oxidative necrosis triggered by hypoxia/reoxygenation and H2O2 is abrogated by the PPARα antagonist, MK886 is a direct proof linking PPARα to this novel protective function of A20. In spite of this direct link, one needs to recognize that the strong in vivo protection against prolonged IRI is the reflection of the sum of the anti-inflammatory and pro-proliferative effects of A20 in hepatocytes, not only of increased PPARα. Furthermore, increased expression of PPARα provides a molecular basis for the metabolic advantage (improved lipid and glucose metabolism) seen in mice expressing A20 in their livers following IRI. This translates into decreased hepatic steatosis, which as discussed earlier, could only improve the resistance of hepatocytes to IRI.25, 37

The mechanism behind increased PPARα mRNA and protein levels in A20-expressing hepatocytes is unknown. Although there is no evidence that A20 has any direct transcriptional activity, A20 may certainly positively affect the expression or function of transcriptional activators or conversely inhibit transcriptional suppressors of PPARα. The transcriptional regulation of PPARα is still unraveling with the very sparse literature implicating the protein CLOCK in the transcriptional regulation of the circadian variations of PPARα transcription in hepatocytes and high glucose as a possible negative regulator of PPARα levels.42, 43 Alternatively, expression of A20 in hepatocytes may increase PPARα mRNA stability, thereby increasing its levels. Future work in the laboratory is aimed at investigating these two possibilities.

We propose that short-term A20-based therapies should improve the outcome of liver grafts from marginal donors and of extensive surgical liver resections in patients with pre-existing liver disease such as steatosis, ethanol toxicity, and hepatitis C. We recognize that expression of an antiapoptotic gene such as A20 might disturb the regulatory apoptosis required for involution of excessive liver mass or promote tumor formation. These problems can be avoided with limited expression of A20 as achieved by recombinant adenoviruses.11 Notably, we did not observe excessive hepatomegaly or liver neoplasia during the long-term follow-up (up to 6 months) of mice treated with A20.

In summary, our results clearly demonstrate that A20 protects livers from oxidative necrosis through a novel PPARα-dependent mechanism. From a basic science standpoint, these data unravel a novel target by which A20 restores homeostasis. From a clinical standpoint, most of the new knowledge that we have gathered on the multiple “hepatoprotective” functions of A20 including protection from oxidative necrosis is both conceptually important and directly relevant to clinical problems associated with liver transplantation and liver disease.

Acknowledgements

We wish to acknowledge Drs. Hasan Otu and Towia Lieberman from the Genomics Center at Beth Israel Deaconess Medical Center for their help with Gene Array analysis, and Dr. Vishva Dixit from Genentech for providing us with the A20 plasmid that served to generate the adenovirus. We are also grateful to Drs. Elizabeth Maccariello and Elzbieta Kaczmarek for their critical review of the manuscript.

Ancillary