Potential conflict of interest: Nothing to report.
Previously, we demonstrated that intrahepatic upregulation of the immunoactivating molecules CD40 and CD40 ligand (CD40L) are early mechanisms for liver cell damage in human and murine fulminant hepatic failure (FHF). In the present study, we investigated the functional effects of intrahepatic overexpression of CD40L by adenoviral-mediated gene transfer (AdCD40L) in mice. AdCD40L injection induced severe liver cell damage, which was associated with increased alanine aminotransferase (ALT) levels peaking at day 5 after vector administration (AdCD40L, 1,707 ± 279 U/L; AdLacZ, 213 ± 25 U/L) and with lethality in half of the mice. Except for mild splenomegaly, no organs other than the liver were involved in inflammatory reactions. CD40–CD40L interaction was mandatory for liver damage, because CD40−/− mice were completely protected. Furthermore, CD40L-induced FHF depended on competent lymphocytes, because inflammatory reactions were strongly decreased in SCID and Rag1−/− mice. In contrast, neither natural killer T (NKT) cells nor Kupffer cells relevantly influenced histology as shown in NKT cell–deficient CD1d−/− mice and by gadolinium depletion of Kupffer cells. Furthermore, immunosuppression by dexamethasone and cyclosporin A was not sufficient to block CD40L damage. In conclusion, we present a model of FHF with strong similarities to human FHF with respect to time course and histological changes. This model suggests involvement of the CD40–CD40L system in FHF and might have important implications for future pathophysiological studies of this condition. (HEPATOLOGY 2006;44:430–439.)
Fulminant hepatic failure (FHF) is a life-threatening condition induced by pathogens such as hepatitis B virus infection and drug toxicity. The underlying pathophysiological mechanisms are not fully understood. This renders a causal therapeutic approach difficult.1, 2 Uncontrolled hepatic immunoactivation is discussed as a pivotal pathomechanism of FHF.3–5 Recently, we described significant intrahepatic upregulation of the immunoactivating molecules CD40 and its ligand (CD40L/CD154) in patients with FHF.3 In human FHF, CD40 was upregulated in hepatocytes, sinusoidal and vascular endothelial cells, and macrophages/Kupffer cells, whereas CD40L was expressed in an increased number of intrahepatic lymphocytes. Taking into account our observation of strongly enhanced hepatic CD40 and CD40L expression in FHF and the potent immunoactivating properties of the CD40–CD40L interaction, we hypothesized that the interaction of CD40 with CD40L is a pivotal trigger of immune cascade–inducing FHF.
CD40L is a member of the tumor necrosis factor receptor superfamily and is expressed on activated CD4+ T cells and platelets.6, 7 The CD40–CD40L interaction induces strong immunoactivation at different levels of the immune system, including T cells, B cells, macrophages, and dendritic cells. Interaction of CD40–CD40L on dendritic cells and other antigen-presenting cells such as monocytes/macrophages and B lymphocytes upregulates costimulatory molecules CD80 and CD86 and thus enhances T cell activation. Deletion of either the CD40 or CD40L gene resulted in severe immunodeficiency in mice.8, 9 The clinical impact of CD40L–CD40 interactions becomes obvious in the X chromosome–linked hyper–immunoglobulin M syndrome characterized by high susceptibility to opportunistic infections.10 Furthermore, CD40–CD40L interaction plays an essential role in autoimmunity and transplant rejection.11 Blocking of CD40L may become a promising therapeutic approach in autoimmune processes and in transplantation medicine.12, 13 Recently, the potency of blocking the CD40–CD40L interaction by CD40 antisense oligonucleotides was elucidated to treat experimental colitis.14 On the other hand, CD40L overexpression has been demonstrated to overcome tumor tolerance in several malignant diseases. Enhanced expression of CD40L was associated with upregulation of interleukin-12, interferon-γ, and chemokines such as macrophage inflammatory proteins 1α, 1β, and 2; RANTES; and eotaxin.15, 16
To study the role of CD40–CD40L interactions with respect to the induction of FHF, we overexpressed CD40L in murine livers via systemic application of a CD40L-encoding adenoviral vector (AdCD40L) and evaluated the resultant biochemical and histological effects.
Male Balb/c, SCID (on Balb/c background), and Rag1−/− (on C57BL/6 background) mice (8-12 weeks old) were purchased from Charles River (Sulzfeld, Germany). CD40−/− (on C57BL/6 background) and CD1d−/− (on C57BL/6 background) mice were bred in the central animal facility of the University Hospital Bonn. Immunodeficient mice were kept under pathogen-free conditions with single-cage ventilation. All procedures were performed according to approved protocols and followed the recommendations for the proper use of laboratory animals of a governmental review board.
The 293 package cell line (adenoviral E1-transformed human embryonic kidney cell line) and the human lung cancer cell line (A549; gene expression experiments) were purchased from the American Type Culture Collection and cultured with Dulbecco's modified Eagle medium supplemented with 10% heat-inactivated fetal bovine serum, 2 mmol/L L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin at 37°C and 5% CO2.
Vector Construction and Transgene Expression.
The control vector AdLacZ and AdCD40L were both constructed as previously described.16, 17 Briefly, for AdCD40L, mouse CD40L complementary DNA had been inserted into an expression cassette resulting in pMV60/CD40L. pMV60/CD40L and pJM17 (first-generation adenoviral construction vector) were cotransfected into the adenoviral package cell line (293 cells) to generate infectious recombinant adenovirus encoding CD40L. A single virus plaque was selected for further virus expansion and purification. Virus stocks were stored in glycerol aliquots at −80°C until use. Titration was performed via plaque assay on 293 cells (given as plaque-forming units per milliliter [pfu/mL]). Contamination with wild-type virus was excluded via plaque assay in A549 cells. After purification, transgene expression of virus stocks was tested as previously described.16 To verify CD40L expression, A549 cells, which had been infected with AdCD40L and AdLacZ 48 hours earlier (multiplicity of infection 250), were harvested and were allowed to react with R-phycoerythrin-conjugated anti-mouse CD154 antibody (CD40L; PharMingen, San Diego, CA) for 30 minutes at 4°C. Samples were washed twice in FACS buffer (phosphate-buffered saline with 3% fetal bovine serum and 0.02% sodium azide). In vitro transfection efficacies of AdCD40L were approximately 98% as determined via FACScan (Beckman Coulter, Krefeld, Germany) analysis (data not shown).
In Vivo Expression of CD40L After Systemic Administration of AdCD40L.
To evaluate hepatic CD40L in vivo expression via Western blot analysis, recombinant adenoviruses were intravenously injected via a dilated tail vein, and mice were sacrificed at days 1, 2, 3, 4, 5, and 6 after vector injection. Liver samples were cryopreserved and stored at −80°C until examination. Proteins were extracted from homogenized liver samples. Total protein content was determined using the DC Protein Assay (Bio-Rad, Munich, Germany). A total amount of 10 μg protein was transferred to an SDS-PAGE gel (4% stacking gel and 10% running gel) followed by immunoblotting. Membranes were hybridized with a monoclonal antibody against murine CD40L (hamster anti-mouse antibody, clone MR1; Pharmingen, Heidelberg, Germany) and a secondary goat anti-hamster antibody against immunoglobulin G (SC-2443; Santa Cruz Biotechnology, Heidelberg, Germany) and were developed according to the manufacturer's protocol (ECL-Plus; Amersham Pharmacia Biotech, Little Chalfont, UK).
Effect of Systemic Administration of AdCD40L on Survival and Liver Histology.
To study the biological effects of CD40L overexpression in the liver, AdLacZ (n = 6) and AdCD40L (n = 6) were injected intravenously at concentrations of 5 × 109 pfu per mouse on day 0. Balb/c mice were closely observed for changes in behavior, appearance, and food intake for an overall observation period of 3 months.
To analyze the time course of histological changes induced by ADCD40L in Balb/c mice, 2 more groups of mice were treated with AdCD40L (n = 24) and with the control vector AdLacZ (n = 24) as described in the previous paragraph. During the following week, 4 mice in each group were sacrificed every day (day 1 to day 6) for preservation of livers and spleens. Livers and spleens were weighed after organ resection and were embedded in paraffin. Standard hematoxylin-eosin staining was performed in 5-μm sections to study histological changes in liver tissue. Alanine aminotransferase (ALT) levels were determined in the corresponding sera. In an additional subgroup of 5 animals out of each group, organ samples of brain, thymus, heart, lungs, gut, pancreas, spleen, and kidney were evaluated for histological changes at day 5 after vector administration.
To further differentiate inflammatory infiltrates, cryopreserved liver sections were immunostained against CD4+ (rat anti-mouse antibody; abcam, Göttingen, Germany) and CD8+ (rat anti-mouse antibody; abcam) T cells using a biotinylated rabbit anti-rat antibody (Dako) as secondary antibody. Enzymatic activity was developed using AEC (Dako) followed by Mayer hematoxylin counterstaining. B cells were stained using a rat anti-mouse/human CD45R/B220 antibody (Becton Dickinson, Bedford, MA), a fluorescent secondary antibody (Alexa Fluor 546, Molecular Probes, Karlsruhe, Germany) was employed, and nuclear staining was performed with Hoechst stain (Sigma Aldrich, Munich, Germany). Images were generated via an overlay technique for liver green fluorescent protein auto-, nuclear, and CD45R/B220 fluorescence microscopy under 200-fold magnification.
CD40L Effects in CD40−/− Mice.
To confirm the central role of the CD40–CD40L interaction for the CD40L-induced liver damage in this model, AdCD40L (n = 5) and AdLacZ (n = 6) were injected intravenously at concentrations of 5 × 109 pfu/mouse into CD40L knockout mice on day 0. Mice were sacrificed on day 5 after vector injection, and liver and serum samples were harvested for further histological evaluation and ALT determination, respectively.
CD40L Effects Under Kupffer Cell Depletion and Immunodeficient SCID, Rag1−/− and CD1d−/− Mice.
To investigate the possible involvement of Kupffer cells in CD40L-induced liver damage, gadolinium-chloride (Sigma, Taufkirchen, Germany) was administered intraperitoneally daily from day 1 to day 5 after AdCD40L injection at a dosage of 40 mg/kg/d. Mice were sacrificed and organ samples were taken at day 5 after AdCD40L administration. Samples were prepared as described in the “Effect of Systemic Administration of AdCD40L on Survival and Liver Histology” section.
Dexamethasone and cyclosporin A were given intraperitoneally at a dosage of
500 μg/kg/d and 50 mg/kg/d, respectively, from day 1 to day 5. Mice were sacrificed and organ samples were taken at day 6 after AdCD40L administration.
To further elucidate B cell and T cell involvement, CD40L effects were investigated in SCID and Rag1−/− mice. The role of NKT cells was evaluated in the CD1d−/− mouse strain. All mice were sacrificed at day 5 after injection of AdCD40L or AdLacZ, and serum and liver samples were taken for further examination.
Data are given as the mean and SEM or SD as appropriate. Differences between groups were compared via a 2-tailed, nonparametric Mann-Whitney U test. A P value of less than .05 was considered statistically significant. All calculations were performed using the SPSS software package.
CD40L Expression After Adenoviral-Mediated Gene Transfer.
To evaluate hepatic in vivo expression of CD40L after adenoviral gene transfer, liver tissues of mice that had received either AdCD40L or AdLacZ were homogenized and analyzed for CD40L expression via Western blot analysis on 6 subsequent days. Only low basal expression levels of CD40L were detected in the AdLacZ control group throughout the observation time from day 1 to day 7 (Fig. 1). In contrast, CD40L expression continuously increased in mice that were injected with AdCD40L, with a maximum level seen on days 5 and 6 after vector administration (Fig. 1). Vector activities of the 2 vector constructs AdLacZ and AdCD40L were also demonstrated in vitro via X-Gal staining and FACScan analysis for CD40L/CD154 in AdCD40L-infected A549 cells, respectively (data not shown).
Biological Effects of Hepatic CD40L Expression
After confirming in vivo and in vitro expression of CD40L, we focused on the biological effects of hepatic CD40L expression in mice. Intravenous injection of AdCD40L and AdLacZ led to behavioral changes such as slightly reduced mobility, impaired social contact, and neglect of fur care in both groups of mice within 24 hours after vector application. Animals of the AdCD40L group deteriorated continuously over the next 5 days and had a significant increase in mortality, whereas control animals (n = 6) recovered completely within 48 hours after vector administration. Fifty percent of the AdCD40L-treated mice (n = 3) died on day 5; the remaining mice (n = 3) recovered within the second week and survived the 3-month observation period. In contrast to the AdCD40L-treated animals, no mortality was observed in the AdLacZ control group (Fig. 2).
We then studied the time course of disease in a larger number of mice (AdCD40L, n = 24; AdLacZ, n = 24). We sacrificed 4 mice a day from each group from day 1 to day 6 after administration of AdCD40L or AdLacZ to harvest organ samples for further analysis. First, we analyzed serum ALT levels as an indicator of hepatic inflammatory reaction in both groups of mice (Fig. 3). AdLacZ-treated mice had stable ALT levels throughout the observation period, whereas AdCD40L-treated mice showed a significant increase in ALT levels peaking at days 4 and 5 after vector injection (AdCD40L, 1,707 ± 279 U/L; AdLacZ, 213 ± 25 U/L).
ALT elevation correlated with the presence of hepatic inflammatory reactions. On day 1 after vector application, liver samples of AdCD40L-treated mice had a completely normal histology (Fig. 4). On day 2 and day 3, only minimal periportal lymphocytic infiltrates could be anticipated with completely preserved liver architecture. On day 4, lymphocytic infiltrates became dense, spreading into liver parenchyma (Fig. 4). Furthermore, on day 4, some hepatocytes presented typical morphological characteristics of apoptosis. These changes strongly aggravated and resulted in severe hepatitis with extended areas of parenchymal liver cell damage in all mice at days 5 and 6 (Fig. 4). In contrast, no hepatitis was observed in the control group during the complete observation period. On autopsy, livers showed greater size and increased weight only in the CD40L group (Fig. 5).
To exclude CD40L effects on other organs and to prove liver specificity in our model, we additionally analyzed hematoxylin-eosin–stained cross-sections of spleen, brain, heart, lung, gut, pancreas, and kidney throughout the observation period. With the exception of a mild increase in spleen weight due to moderate mononuclear inflammatory reaction in this location, no inflammatory changes were observed in any other organ. Interestingly, no long-term liver damage was obvious in surviving AdCD40L-treated mice (n = 3) when examined 3 months after vector administration (data not shown).
CD40L Effects in CD40−/− Mice.
To underline the role of the CD40–CD40L interaction in this model, we challenged CD40−/− mice with AdCD40L or AdLacZ. All of the mice survived, and histologically no liver damage was induced by CD40L overexpression in this CD40 knockout mouse strain, with no differences between AdCD40L and AdLacZ treatment. These experiments clearly demonstrate that AdCD40L-induced liver damage is specifically induced by CD40–CD40L interaction.
CD40L Effects Under Kupffer Cell Depletion and in Immunodeficient SCID, Rag1−/−, and CD1d−/− Mice.
According to the histology with dense mononuclear infiltrates, we proposed an immunomediated mechanism involving cellular immunity. T lymphocytes and NKT cells in particular could be involved in this cellular immunity. Therefore, we evaluated the effect of CD40L on B cell and T cell deficiency in SCID mice and in Rag1−/− mice, as well as CD1d−/− mice, which are deficient in invariant NKT cells. In SCID mice—and moreso in Rag−/− mice—inflammatory infiltration and hepatocellular damage were markedly decreased (Fig. 6). In contrast, NKT cell deficiency did not ameliorate liver inflammation or hepatocellular damage (Fig. 6). These data imply the importance of competent T cells for the mediation of CD40L-induced liver damage in this model.
Because Kupffer cells can be activated by AdCD40L and might potentially mediate CD40L-induced liver damage, we analyzed whether blockade of Kupffer cells by gadolinium-chloride can ameliorate AdCD40L-induced liver damage. However, ALT levels were not significantly reduced by treatment of the animals with gadolinium-chloride (Fig. 7). Furthermore, administration of dexamethasone and cyclosporin A did not improve survival in this model when administered 1 day after AdCD40L injection.
Immunostaining against CD4+ and CD8+ T and B cells in CD40L-damaged livers revealed that the lymphocytic infiltrates varied over the time course: liver injury was initiated by periportal infiltrates starting at day 3. These infiltrates were dominated by CD8+ cytotoxic T cells and, to a lesser extent, CD4+ T cells; only a few B cells were detectable at this time point. On subsequent days, the number of CD8+ T cells increased, which was paralleled by a strong gain of B lymphocytes, shifting the T cell–B cell ratio toward the B cell component (Fig. 8). These data support the initiatory role of T cells and underline that full liver damage is the result of a sequential interplay of T and B cell infiltrates. Liver sections of control animals had only single T- and B-lymphocytic infiltrates throughout the observation period.
In this study, hepatic overexpression of CD40L caused severe hepatitis and mortality in mice, confirming our hypothesis that the interaction of CD40 with CD40L is a pivotal trigger of the immune cascade that induces FHF. We thus used a recombinant adenovirus encoding CD40L to target transgene expression to the liver.18, 19 Systemic gene delivery of CD40L resulted in hepatic overexpression of CD40L as confirmed via Western blot analysis. Of note, half of the treated mice died within the first week after gene delivery. Surviving wild-type mice, however, fully recovered during the subsequent 3 weeks. The histological characteristics of murine livers closely resembled those of lesions seen in human viral-induced FHF. The similarity of this experimentally induced hepatitis to human FHF was reflected by (1) a delayed onset of FHF almost 1 week after vector administration, (2) the potential recovery of the liver from severe hepatitis, (3) the lack of any survival benefit of corticosteroids, and (4) the involvement of CD40L, which is also overexpressed in human FHF.3 These observations are in line with the hypothesis that an imbalance of costimulatory molecules is a pivotal early step in pathogenesis of FHF. Importantly, CD40−/− mice were fully protected against CD40L-induced liver failure. These experiments clearly demonstrate that CD40L-induced liver damage is specifically induced by CD40–CD40L interaction.
Dense mononuclear infiltrates account for a mediation of the CD40L-induced liver damage by cellular immunity. Our data on a protection in SCID and Rag1−/− mice strongly suggest that T lymphocytes were the major mediators of liver damage in our model. In contrast, no major role could be demonstrated for NKT cells or Kupffer cells.
The absence of severe liver damage in SCID and Rag1−/− mice argues against a hypothetical direct toxicity of the adenoviral-mediated gene delivery of CD40L, because toxic effects can be vastly independent of the immune response. Furthermore, the complete abrogation of any liver damage in CD40 knockout mice convincingly excludes adenoviral toxicity as the initiating pathophysiological factor. Additionally, the direct cytopathic effects of adenoviruses usually develop within the first 2 days after application,20 whereas we observed the onset of liver damage as late as 5 days.
Considering the protective effects of T cell and B cell deficiency and the particular role of T cells in initiating CD40L-induced liver injury, we considered whether immunosuppressive treatment after vector application was capable of ameliorating CD40L-induced liver damage. Contrary to our expectations, neither dexamethasone nor cyclosporin A improved survival when treatment was initiated after AdCD40L administration.
Our observations in mice are in line with a clinical study in humans indicating that systemic CD40L treatment can cause severe side effects.21 In the human study, recombinant human CD40L protein had been administered subcutaneously to patients with advanced solid tumors and non-Hodgkin's lymphomas. Transient, dose-dependent ALT elevations peaking at day 6 with subsequent ALT normalization within 2 weeks had been observed in this trial. The authors proposed a cytotoxic effect of CD40L on CD40-positive hepatocytes as a putative cause of these ALT elevations. However, our data on CD40L-induced liver failure suggest that the ALT flare in CD40L-treated patients might have been caused by an enhanced immune response. In support of the latter hypothesis, it has been reported that transgenic mice expressing CD40L in keratinocytes spontaneously developed autoimmune dermatitis.22 Consistent with our finding that CD8+ T cells dominated early periportal infiltrates, CD40L-induced skin inflammation was also attributed to activation of antigen-presenting cells (Langerhans cells) mounting a specific CD8+ T cell response against keratinocytes.
Beyond supporting a crucial role of CD40L in the pathogenesis of FHF, our data provide the opportunity to establish a new animal model for the study of immune-triggered FHF. FHF has been studied in several experimental animal models, including concanavalin A–, galactosamine–lipopolysaccharide-, galactosamine–tumor necrosis factor–, and Fas–antibody-induced FHF. However, the rapid onset of FHF and the short survival times in these models hamper the experimental follow-up of the inflicted damage and thus are more suited to investigate effector mechanisms of FHF rather than the induction phase. Zhou and coworkers,23 in a concanavalin A model, reported that the CD40L–CD40 interaction plays a major role in triggering hepatocellular apoptosis by demonstrating that hepatitis, tumor necrosis factor α levels, and hepatocyte death was significantly attenuated in CD40L−/− mice.
Of note, the development of liver failure was prolonged over 5 days in our model, which involved detailed studies of the initiating pathogenic steps during the early time course of liver damage. Furthermore, the delayed appearance of FHF in our model more closely resembled the human counterpart, which is usually characterized by a disease course of several days.
In conclusion, our present observations in mice and our previous findings in patients with FHF suggest that the CD40–CD40L system plays a pivotal pathogenic role in the development of FHF. Furthermore, AdCD40L-induced liver failure represents a method to further study the underlying mechanisms leading to damage and regeneration with high resolution in time.