A.W. and B.W. contributed equally to the study.
Liver Biology and Pathobiology
Adenoviral gene transfer of ABIN-1 protects mice from TNF/galactosamine-induced acute liver failure and lethality†
Article first published online: 15 JUL 2005
Copyright © 2005 American Association for the Study of Liver Diseases
Volume 42, Issue 2, pages 381–389, August 2005
How to Cite
Wullaert, A., Wielockx, B., Van Huffel, S., Bogaert, V., De Geest, B., Papeleu, P., Schotte, P., El Bakkouri, K., Heyninck, K., Libert, C. and Beyaert, R. (2005), Adenoviral gene transfer of ABIN-1 protects mice from TNF/galactosamine-induced acute liver failure and lethality. Hepatology, 42: 381–389. doi: 10.1002/hep.20785
Potential conflict of interest: Nothing to report.
- Issue published online: 15 JUL 2005
- Article first published online: 15 JUL 2005
- Manuscript Accepted: 16 MAY 2005
- Manuscript Received: 11 NOV 2004
- Fund for Scientific Research-Flanders (FWO-Vlaanderen)
- Belgian Federation against Cancer
- Interuniversity Attraction Poles (IUAP-V)
- Flemish Institute for the Promotion of Scientific-Technological Research in the Industry (IWT)
Tumor necrosis factor (TNF) is a proinflammatory cytokine that plays a central role in acute and chronic hepatitis B and C infection and alcoholic liver disease as well as fulminant liver failure. TNF-induced liver failure is characterized by parenchymal cell apoptosis and inflammation leading to liver cell necrosis. The transcription factor NF-κB is believed to mediate at least part of the proinflammatory effects of TNF, and is therefore a favorite drug target. However, NF-κB also suppresses TNF-mediated hepatocyte apoptosis, implicating a potential cytotoxic effect of NF-κB inhibitors in the liver. This dual function of NF-κB emphasizes the need for therapeutics that can inhibit both TNF-induced NF-κB activation and cell death. Here we describe that adenoviral expression of the NF-κB inhibitory protein ABIN-1, but not an IκBα superrepressor (IκBαs), completely prevents lethality in the TNF/D-(+)-galactosamine–induced model of liver failure. Protection was associated with a significant decrease in TNF-induced leukocyte infiltration as well as hepatocyte apoptosis. The differential effects of ABIN-1 and IκBαs suggest a role for an NF-κB independent function of ABIN-1. Indeed, ABIN-1 was found to prevent not only NF-κB activation, but also apoptosis of cultured hepatocytes in response to TNF, explaining its protective effect against TNF-induced liver failure. In conclusion, ABIN-1 has a dual NF-κB inhibitory and anti-apoptotic activity in the liver, which might be of considerable interest for the treatment of inflammatory liver diseases. (HEPATOLOGY 2005;42:381–389.)
Tumor necrosis factor (TNF)-induced cytotoxicity is a critical event in a variety of liver diseases.1 Indeed, TNF-mediated hepatocyte injury has been observed in acute and chronic hepatitis B and C infections and alcoholic liver disease, as well as in fulminant hepatic failure.2–4 Therefore, TNF-induced signaling pathways leading to hepatocyte injury seem to be an attractive target for therapeutic intervention in liver failure. Although TNF is known to activate several signaling pathways leading to cell death,5 hepatocytes are only sensitive to the cytotoxic effect of TNF in the presence of inhibitors of transcription or translation.6 Also in mice, TNF only exerts clear hepatotoxic effects when it is administered in combination with the hepatocyte specific transcriptional inhibitor D-(+)-galactosamine (GalN), suggesting that GalN sensitizes mice to TNF-mediated liver failure by inhibiting the production of essential hepatoprotective factors.7, 8 In this context, a role for nuclear factor-κB (NF-κB) dependent gene expression has been proposed. Indeed, hepatocytes can also be sensitized to TNF-induced apoptosis by suppressing NF-κB activation,9, 10 and transgenic mice compromised in hepatocyte NF-κB activation display the same symptoms after TNF injection as GalN-sensitized mice.11 In addition, inhibition of NF-κB activation in the liver by adenoviral overexpression of an IκBα superrepressor (IκBαs) enhances TNF hepatotoxicity and makes mice susceptible to otherwise sublethal doses of TNF.12, 13 These studies suggest that the transcription of essential hepatoprotective factors is most likely NF-κB dependent. Because TNF is itself a very potent activator of NF-κB, at least some of these protective proteins might be upregulated by increased TNF expression in the liver. However, NF-κB also positively regulates the expression of a variety of proinflammatory genes in response to TNF, suggesting that increased NF-κB activation because of increased TNF expression in the liver might function as a double-edged sword.5 Ideally, therapeutic intervention of inflammatory liver disease should target TNF-induced apoptosis as well as TNF-induced NF-κB activation. Using the TNF/GalN model of acute hepatitis, which is characterized by massive parenchymal cell death and proinflammatory cell infiltration, we describe in the current study that adenoviral gene transfer of ABIN-1 completely protects mice against TNF/GalN-induced liver toxicity and lethality. ABIN-1 has previously been described as an inhibitor of TNF-induced NF-κB activation, both in vitro and in vivo.14, 15 Interestingly, we now found that ABIN-1 exerts a dual NF-κB and cell death inhibiting effect in hepatocytes, adding another player in the machinery of hepatoprotective proteins.
Materials and Methods
Eight- to 12-week-old wild-type female C57BL/6 mice were purchased from Charles River (Sulzfield, Germany). Male outbred Sprague-Dawley rats (200-300 g) were obtained from Charles River Laboratories (Brussels, Belgium). All animals were maintained under standard conditions and received humane care in accordance with the National Institutes of Health (NIH) guidelines and with the legal requirements in Belgium. All animal experiments were performed in accordance with protocols approved by the Institutional Animal Care and Research Advisory Committee.
Recombinant Adenovirus Production.
A recombinant adenoviral vector AdABIN-1 was generated by cloning the murine ABIN-1 cDNA, N-terminally fused to an E-tag, into the pACpLpA.CMV shuttle vector and co-transfected with the rescue plasmid pJM17 (which encodes the adenovirus dl309 genome, lacking E1 and E3 functions) into HEK293 cells via calcium phosphate coprecipitation.16 Recombinant plaques were isolated, extracted DNA was verified by polymerase chain reaction (PCR), and expression of the correct transgene from the ubiquitously active cytomegalovirus (CMV) promoter was confirmed by means of Western blotting. Control viruses without transgene (AdRR5) or expressing the β-galactosidase gene (AdLacZ), and a virus expressing the IκBα superrepressor (AdIκBαs)17 (gift from Dr. R. Hay, University of St. Andrews, St. Andrews, Scotland) were generated with the same pJM17 adenoviral backbone vector. A virus expressing an NF-κB luciferase reporter gene (AdNFκBLuc)18 was obtained from Dr. B. McGray (Universty of Iowa College of Medicine, Iowa City, IA). High-titer virus stocks were prepared in HEK293 cells and purified via single CsCl banding. Titers were determined via plaque assay in HEK293 cells and calculated as plaque-forming units (pfu) per milliliter virus stock.
Adenoviral Infection of Mice and TNF/GalN Challenge.
For adenovirus infection, mice were intravenously injected with 2.5 × 109 pfu of virus diluted in pyrogen-free phosphate-buffered saline (PBS). Three days after infection, mice were challenged intraperitoneally with 0.3 μg murine TNF (109 IU/mg) in combination with 20 mg GalN (Sigma Chemical Co., St. Louis, MO) (a dose determined in preliminary studies as the lethal dose in all exposed subjects). Liver samples were taken 8 hours after the challenge (unless otherwise stated), the time at which the condition of the control (AdRR5) mice started to deteriorate.
Pieces of liver were homogenized by douncing in lysis buffer (10 mmol/L Tris-HCl pH 7.5, 1% NP-40, 200 mmol/L NaCl, 5 mmol/L EDTA, 10% glycerol, 0.1 mmol/L aprotinin, 1 mmol/L phenylmethyl sulfonyl fluoride, and 1 mmol/L gluthatione). After 20 minutes of incubation on ice, homogenates were centrifuged for 30 minutes at 4°C. Equal amounts of protein (50 μg) were subjected to SDS-PAGE and immunoblotted with polyclonal anti-IκBα (Santa Cruz Biotechnology, Santa Cruz, CA), anti–inducible nitric oxide synthase (iNOS) (Chemicon International Inc., Ternuca, CA), anti-caspase 3 (provided by Dr. P. Vandenabeele, Ghent University, Belgium), or anti–β-actin (MP Biomedicals, Irvine, CA) antibodies. Rabbit polyclonal anti–ABIN-1 antibodies were raised against an ABIN-1–specific peptide (NH2-CTARPTEPESPKNDREGPQ-COOH) coupled with keyhole limpet hemocyanin. BWTG3 lysates were made in Laemmli buffer and immunoblotted with anti-caspase 3, anti-PARP (Biomol, Plymouth, PA) or anti–β-actin. After incubation with appropriate horseradish peroxidase–coupled secondary antibodies (Amersham Biosciences, Roosendaal, The Netherlands), immunoreactivity was revealed via chemiluminescence (Perkin Elmer, Boston, MA).
Electrophoretic Mobility Shift Assay (EMSA).
Pieces of liver were homogenized by douncing in swelling buffer (15 mmol/L HEPES pH 7.5, 300 mmol/L sucrose, 60 mmol/L KCl, 15 mmol/L NaCl, 2 mmol/L EDTA, 0.5 mmol/L EGTA, 0.15 mmol/L spermine, 0.5 mmol/L spermidine, and 14 mmol/L β-mercaptoethanol). Samples were centrifuged at 800g for 5 minutes at 4°C. The pellet was resuspended in 1 mL swelling buffer with 0.5% NP40. Gradient centrifugation was performed over 3 mL swelling buffer with 30% sucrose at 1,600g for 5 minutes at 4°C. Pellets were resuspended in 200 μL nuclear extraction buffer (20 mmol/L HEPES pH 7.5, 1% NP40, 1 mmol/L MgCl2, 400 mmol/L NaCl, 10 mmol/L KCl, 20% glycerol, 0.5 mmol/L EDTA, 0.1 mmol/L EGTA, 2 mmol/L PefaBlock, 0.5 mmol/L dithiothreitol [DTT], and 0.15 IU/mL aprotinin). After vortexing and centrifugation for 1 minute at maximum speed at 4°C, supernatants were taken and stored at −70°C until use. Ten micrograms nuclear extract protein was incubated at room temperature for 30 minutes with a 32P-labeled NF-κB–specific DNA-probe (5′-agctagaggggactttccgagagg-3′) in the following buffer: 20 mmol/L HEPES pH 7.5, 4 % Ficoll 400, 60 mmol/L KCl, 2 mmol/L DTT, 100 μg/mL poly d(I-C), and 1 mg/mL acetylated bovine serum albumin. Extracts were run on a 4% native polyacrylamide gel. Gels were dried, and radioactivity was visualized by exposure to X-ray films.
Assessment of Liver Damage and Caspase Activity.
For measuring the serum concentration of alanine aminotransferase (ALT), blood was withdrawn at the orbital plexus under light ether anesthesia and serum was prepared by clotting and centrifugation. The concentration of ALT was measured using a colorimetric test from Sigma Chemical Co.
Total liver apoptosis was determined using a specific ELISA for histon-bound DNA fragments, as previously described.19 Caspase activity was measured by incubating 30 μg liver homogenate in 200 μL cell-free system buffer (10 mmol/L HEPES pH 7.5, 220 mmol/L mannitol, 68 mmol/L sucrose, 2 mmol/L NaCl, 2 mmol/L MgCl2, 2.5 mmol/L KH2PO4, 10 mmol/L DTT) in the presence of 50 μmol/L Ac-DEVD-AMC (Peptide Institute, Osaka, Japan) at 30°C. Release of 7-amino-4-methyl coumarin (AMC) was monitored over 60 minutes in a fluorometer (CytoFluor; PerSeptive Biosystems, Cambridge, MA) at an excitation wavelength of 360 nm and an emission wavelength of 409 nm.
Myeloperoxidase (MPO) activity was determined as a marker for neutrophil infiltration.20 Briefly, tissue samples were weighed and homogenized by sonication in buffer A (0.5% hexadecyltrimethylammonium bromide in 50 mmol/L potassium phosphate buffer, pH 6.0). Homogenates were subjected to 3 freeze/thaw cycles of 5 minutes each. After centrifugation for 20 minutes, 20 μL supernatant of each sample was mixed with 280 μL buffer B (0.167 mg/mL o-dianisidine dihydrochloride and 0.0005% H2O2 in 50 mmol/L potassium phosphate buffer, pH 6). After 20 minutes, absorbance was measured spectrophotometrically at 460/490 nm. Purified human MPO was used as a standard.
Histology and Immunohistochemistry.
For histological analysis, paraffin sections of 4 μm were stained with hematoxylin-eosin, using standard methods. For analyzing neutrophil and macrophage influx by immunohistochemistry, paraffin sections were incubated with anti-human MPO (Prosan, Merelbeke, Belgium) and anti-mouse Mac-3 (Becton Dickinson, San Diego, CA) antibody, respectively, as previously described.21
Cell Culture and Isolation of Primary Hepatocytes.
The mouse hepatoma cell line BWTG322 was obtained from Dr. C. Szpirer (Free University of Brussels, Brussels, Belgium) and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, nonessential amino acids, and antibiotics. Primary rat hepatocytes (viability > 85%) were isolated by a 2-step collagenase perfusion23 and were plated at a density of 0.4 × 105 cells/cm2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, and antibiotics. Four hours after plating, medium was removed and fresh medium, supplemented with 7.5 μg/mL hydrocortisone hemisuccinate, was added to the cells. For adenoviral infection, hepatocytes were incubated with the indicated multiplicity of infection (moi) adenovirus in a minimal volume of serum-free medium for 6 hours, after which the medium was replaced by fresh serum-containing medium.
Luciferase Reporter Gene Assay for NF-κB Activity.
BWTG3 cells or primary hepatocytes were seeded in 24-well plates at a density of 0.4 × 105 cells/cm2, and infected the next day with adenovirus at a moi of 200 or 75, respectively. The adenovirus mixture comprised 25% AdNFκBLuc, 25% AdLacZ, and 50% AdRR5 or AdABIN-1. One day after infection, cells were either untreated or stimulated with 1,000 IU/mL murine TNF for 6 hours. NF-κB promoter activity was determined by measuring luciferase (Luc) activity in cell extracts as previously described.24 β-Galactosidase (Gal) activity was assayed using the Galactostar reporter gene assay system (Applied Biosystems, Foster City, CA). Luc values were normalized for Gal values to correct for differences in infection efficiency (plotted as Luc/Gal).
Cell Death Assay.
BWTG3 cells or primary hepatocytes were seeded in 96-well plates at a density of 0.4 × 105 cells/cm2 and infected the next day with adenovirus at a moi of 200 or 37, respectively. The next day, cells were either treated with 1 μg/mL actinomycin D (ActD) alone or in combination with 1,000 IU murine TNF/mL in a final volume of 150 μL. Hepatocyte viability was determined by adding 15 μL/well of a 5-mg/mL stock of 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-2H-tetrazolium bromide (MTT) (Sigma) in PBS. After incubation for 4 hours at 37°C, 60 μL of a solubilization-stop solution (10% sodium dodecyl sulfate, 0.01 mol/L HCl) was added to each well. After overnight incubation the absorbance was read at 595/600 nm in a 96-well plate reader (Bio-Rad, Hercules, CA).
All data represent at least 3 independent experiments and are expressed as mean values ± standard deviations, which were compared using an unpaired Student t test, with Welch's correction. Survival curves were compared using a log-rank chi-squared test, and final outcomes using a chi-squared test. The levels of probability are noted (*P < .05, **P < .01, ***P < .0001).
Adenoviral Gene Transfer of ABIN-1 Protects Mice Against TNF/GalN-Induced Lethal Hepatitis.
To analyze the effect of ABIN-1 on TNF/GalN-induced lethality, C57BL/6 mice were injected intravenously with a recombinant adenovirus expressing a murine ABIN-1 transgene (AdABIN-1) or an adenovirus expressing no transgene (AdRR5) as a control. Because expression of ABIN-1 in the liver was maximal 3 days after infection (Fig. 1), mice were challenged with a lethal dose of TNF/GalN 3 days after infection. Interestingly, whereas all control mice died, ABIN-1–expressing mice survived and did not show any signs of illness (Fig. 2). Histologically, TNF/GalN-induced liver injury differed drastically in severity between the 2 groups of animals. In control mice, the hepatic microarchitecture was no longer discernible and was characterized by erythrocyte influx (hemorrhage) at the site of the sinusoids (Fig. 3A). In contrast, TNF/GalN-injected mice that were infected with AdABIN-1 showed a better-preserved liver architecture, with no signs of hemorrhage. The marked difference between the 2 groups of mice could also easily be appreciated on macroscopic examination of the liver. Control mice had a black liver because of massive hemorrhage, whereas the liver of AdABIN-1–infected mice had a normal appearance (Fig. 3B). Furthermore, significant amounts of neutrophils and macrophages could be detected in the liver parenchyma of control mice 8 hours after TNF/GalN injection. In AdABIN-1–infected mice, however, this TNF/GalN-induced infiltration of inflammatory cells was severely diminished (Fig. 3C). In agreement with the inhibitory effect of ABIN-1 on TNF/GalN-induced leukocyte influx into the liver, liver extracts prepared from AdABIN-1–infected mice showed substantially less activity of the neutrophil marker enzyme MPO as compared with control mice (Fig. 3D).
As a parameter for total apoptosis in the liver, we measured the amount of internucleosomal DNA fragmentation by immunochemical detection of histon-complexed DNA fragments (Fig. 4A). This showed that TNF/GalN-induced DNA fragmentation is severely reduced in AdABIN-1–infected mice compared with control AdRR5 mice, indicating that ABIN-1 not only inhibits the inflammatory response, but also the apoptotic response in the liver after injection of TNF/GalN. Because apoptosis typically requires the activation of caspases, we also analyzed caspase activity in liver extracts prepared from these TNF/GalN-treated mice by measuring the hydrolysis of Ac-DEVD-AMC, a substrate of the effector caspase-3 and -7. Again, TNF/GalN-induced caspase activity was significantly reduced upon adenoviral delivery of ABIN-1 (Fig. 4B). Consistent with the increased caspase activity observed in the liver of TNF/GalN-injected mice, also proteolytic cleavage of caspase-3 with the generation of a mature p20 fragment could be detected in liver homogenates prepared from these mice. However, no caspase-3 cleavage could be detected in livers of AdABIN-1–infected mice (Fig. 4C). As a parameter for liver damage, we also measured the TNF/GalN-induced release of ALT in the blood. ALT levels were significantly diminished in the serum of AdABIN-1–infected mice when compared with control mice (Fig. 4D). Altogether, these results clearly show that adenoviral gene transfer of ABIN-1 in the liver protects mice against TNF/GalN-induced lethality by inhibiting the proinflammatory as well as the cytotoxic effects of TNF.
The NF-κB Inhibitory Function of ABIN-1 Is Not Sufficient to Protect Against TNF-Induced Liver Toxicity.
Taking into account the generally accepted anti-apoptotic function of NF-κB in the liver, the observed protective effect of the NF-κB inhibitory protein ABIN-1 against TNF-induced hepatotoxicity is rather surprising. To further investigate the effect of NF-κB inhibition on TNF/GalN-induced liver toxicity, we also tested the effect of NF-κB inhibition by adenoviral gene transfer of an IκBα superrepressor (AdIκBαs). The latter is a mutant of IκBα in which Ser32 and Ser36 have been mutated to Ala, preventing its stimulus-induced phosphorylation and degradation.25 In contrast to adenoviral expression of ABIN-1 in the liver, similar expression of IκBαs did not protect against TNF/GalN-induced lethality (Fig. 5A). Given these different effects of ABIN-1 and IκBαs on TNF/GalN-induced liver failure, we analyzed to what extent TNF-induced NF-κB activation was inhibited in the liver upon ABIN-1 and IκBαs expression. Both IκBαs and ABIN-1 strongly reduced TNF-induced NF-κB activation in the liver, as indicated by electrophoretic mobility shift assay for nuclear NF-κB, or by Western blotting for TNF-induced IκBα degradation, and iNOS expression, which is known to be NF-κB dependent26 (Fig. 6). Because the empty virus AdRR5 already slightly decreased the TNF-induced iNOS expression compared with noninfected cells, the AdRR5-infected cells were used as a control in the other NF-κB assays. Because ABIN-1 was slightly less effective then IκBαs in some of these assays, one could still speculate that gene-specific effects or subtle differences in efficiency of NF-κB inhibition might explain the different effect of ABIN-1 and IκBαs on TNF/GalN-induced liver failure. To investigate these possibilities, we infected mice with a combination of AdABIN-1 and AdIκBαs, which allowed us to analyze the effect of ABIN-1 on TNF/GalN-induced lethality in the setting of complete NF-κB inhibition by IκBαs. Interestingly, similar to AdABIN-1 alone, the combination of AdABIN-1 and AdIκBαs also completely protected mice against TNF/GalN-induced lethality (Fig. 5B). These results suggest that an unknown function of ABIN-1, independent of its NF-κB inhibitory effect, is responsible for the protection against TNF/GalN-induced liver failure.
ABIN-1 Inhibits TNF-Induced Apoptosis in Cultured Hepatocytes.
Because the above results indicate that the protective effect of ABIN-1 against TNF/GalN-induced lethality cannot solely result from its NF-κB inhibitory function, we evaluated the effect of adenoviral expression of ABIN-1 on other TNF-induced responses. In particular, because overexpression of ABIN-1 significantly diminished TNF/GalN-induced apoptosis and necrosis in the liver (Fig. 4), we focused on the effect of ABIN-1 on TNF-induced cell death in hepatocytes. As expected and in line with the observed NF-κB inhibiting effect of ABIN-1 in the liver, ABIN-1 overexpression significantly inhibited the TNF-induced expression of an NF-κB dependent luciferase reporter gene in murine BWTG3 hepatoma cells as well as in primary rat hepatocytes (Fig. 7A). To investigate the effect of ABIN-1 on TNF-induced cell death, these cells were sensitized to TNF by cotreatment with the transcriptional inhibitor actinomycin D (ActD), which is reminiscent of the GalN-induced TNF sensitization of hepatocytes in vivo. Importantly, overexpression of ABIN-1 rescued 50% of the BWTG3 cells and nearly 80% of the primary hepatocytes from TNF/ActD-induced cell death (Fig. 7B). This protective effect of ABIN-1 against TNF/ActD-induced cell death could also easily be seen on microscopic examination. Whereas AdLacZ-infected BWTG3 cells showed membrane blebbing and cell shrinkage after TNF/ActD treatment, AdABIN-1–infected cells kept their normal morphology (Fig. 8A). Similar observations were made in the case of primary hepatocytes (data not shown). Because these morphological features are characteristic of apoptosis, we also investigated the effect of ABIN-1 on the TNF/ActD-induced proteolytic activation of caspase-3 and cleavage of its substrate poly(ADP-ribose)polymerase (PARP) in BWTG3 cells. In AdLacZ-infected cells, caspase-3 processing to its mature p20 subunit, as well as PARP cleavage, could be detected as soon as after 5 hours TNF/ActD treatment. In contrast, no caspase-3 activation or PARP cleavage could be detected in ABIN-1–expressing hepatocytes (Fig. 8B). In conclusion, these data demonstrate that in addition to its NF-κB inhibitory function, ABIN-1 also exerts an anti-apoptotic activity in hepatocytes, which might explain its beneficial effect in TNF/GalN-induced liver failure.
In the present study we demonstrate that ABIN-1, a protein that was previously shown to have an NF-κB inhibitory function,14, 27 exerts an additional anti-apoptotic effect upon overexpression in hepatocytes. This novel function of ABIN-1 came to our attention when we unexpectedly found that adenoviral expression of ABIN-1 in the liver completely protected mice against TNF/GalN-induced liver failure and lethality. This protection was associated with a significant inhibition of leukocyte infiltration as well as an increased survival of liver cells. However, the increased survival of hepatocytes on NF-κB inhibition by ABIN-1 is not in line with the well-established anti-apoptotic role of NF-κB in the liver.11, 13, 28 Indeed, also in the current study NF-κB inhibition by adenoviral expression of IκBαs did not prevent TNF/GalN-induced liver failure. Moreover, our observation that ABIN-1 is still protective when NF-κB–dependent gene expression is inhibited by coexpression of IκBαs, suggested that its hepatoprotective effect is not the result of its NF-κB inhibitory function. In this context, our finding that ABIN-1 exerts a novel anti-apoptotic function when overexpressed in hepatocytes could explain the protective effect of ABIN-1 expression against TNF-induced liver failure. Moreover, as TNF-induced parenchymal apoptosis has been shown to be a critical event triggering neutrophil extravasation in the liver,29, 30 the anti-apoptotic effect of ABIN-1 might also be responsible for the observed reduction of inflammatory cell infiltration. Although this would suggest that the anti-apoptotic function of ABIN-1 is sufficient for its protective effect against TNF/GalN-induced liver failure, we cannot exclude that an additional anti-inflammatory effect of NF-κB inhibition by ABIN-1 also contributes to its hepatoprotective effect. Because previous observations showed that inhibition of neutrophil influx protects mice from TNF-mediated liver injury,21, 31, 32 a reduced influx of leukocytes on NF-κB inhibition by ABIN-1 expression, consistent with the known role of NF-κB in leukocyte recruitment,33 could still contribute to the hepatoprotective effect of ABIN-1. However, because liver sinusoidal endothelial cells are not permissive for adenoviral type 5,34 a direct effect of ABIN-1 on the liver vasculature and neutrophil recruitment is rather unlikely under the conditions used.
Besides the activation of multiple caspases, TNF-induced apoptosis of hepatocytes also involves the mitochondrial death pathway, the activation of c-Jun N-terminal kinase, and the generation of reactive oxygen species.35–37 Although our data indicate that ABIN-1 interferes with the TNF-induced activation of caspase-3, the underlying mechanism for the anti-apoptotic effect of ABIN-1 is not known and beyond the scope of the current study. ABIN-1 expression did not change the binding of TNF to its receptor, because we could not detect any significant difference in 125I-labeled TNF binding to BWTG3 cells on overexpression of ABIN-1. Moreover, ABIN-1 overexpression did not have any effect on the TNF-induced activation of p38 and ERK MAP kinases in BWTG3 cells as well as in liver homogenates, excluding a nonspecific effect of ABIN-1 on TNF/TNFR activation (data not shown). It is also worth mentioning that ABIN-1 is able to bind to the zinc finger protein A20,14 which is known to inhibit TNF-induced NF-κB activation and apoptosis in vitro as well as in vivo.38 However, A20 is not constitutively expressed at detectable levels in most cell types, including hepatocytes.39 Moreover, A20 expression is itself NF-κB dependent, which makes a role for A20 in the hepatoprotective effect of ABIN-1 unlikely because we show that ABIN-1 still protects against TNF/GalN-induced lethality in the setting of complete inhibition of NF-κB by coexpression of IκBαs.
In conclusion, our data illustrate a hepatoprotective function of ABIN-1. Its dual NF-κB and cell death–inhibiting function might confer ABIN-1 the ability to block NF-κB activation without the inherent risk of sensitizing hepatocytes to the cytotoxic effects of TNF. Therefore, strategies that increase the expression or activity of ABIN-1 might have a therapeutic potential for the treatment of inflammatory liver disease.
The authors thank Dr. R. Hay, Dr. B. McGray, and Dr. P. Vandenabeele for generously providing reagents used in this study. B. Coornaert, K. Goethals, F. Delaei, and G. Elaut are acknowledged for technical assistance.
- 14The zinc finger protein A20 inhibits TNF-induced NF-kappaB-dependent gene expression by interfering with an RIP- or TRAF2-mediated transactivation signal and directly binds to a novel NF-kappaB-inhibiting protein ABIN. J Cell Biol 1999; 145: 1471–1482., , , , , , et al.
- 23Isolation of rat hepatocytes. In: PhilipsIR, ShephardEA, eds. Methods in Molecular Biology. Volume 107: Cytochrome P450 Protocols. Totowa, NJ: Humana Press, 1998: 295–301., , , , , .