Genetic predisposition and environmental danger signals initiate chronic autoimmune hepatitis driven by CD4+ T cells


  • Potential conflict of interest: Nothing to report.

  • Supported by grants from the German Research Foundation (KFO250 project 7) and the cluster of excellence REBIRTH.


Autoimmune hepatitis (AIH) is defined as a chronic liver disease with loss of tolerance against liver tissue eventually leading to cirrhosis if left untreated. 80%-90% of patients can be treated with a life-long immunosuppression. Unfortunately, there are strong drug-related side effects and steroid-refractory patients. Therefore, there is a need for a model system to investigate the complex immunopathogenesis of this chronic disease and subsequently to develop new therapeutic interventions. We developed a new model of experimental murine AIH (emAIH) by a self-limited adenoviral infection with the hepatic autoantigen formiminotransferase cyclodeaminase (FTCD). After an initial transient hepatitis there was a chronic evolving AIH, finally leading to portal and lobular fibrosis. We could show that the genetic predisposition provided by the NOD background was essential for creating a fertile field for the development of liver-specific autoimmunity. However, a strong environmental trigger was additionally necessary to initiate the disease. Besides the break of humoral tolerance, T-cell tolerance against hepatic self-antigens was also broken and CD4+ T cells were identified as essential drivers of the disease. As the disease was successfully treated with prednisolone and budesonide, the model will be helpful to develop and test new therapeutic interventions. Conclusion: We developed a new murine AIH model closely resembling AIH in patients that explains the mechanisms of AIH pathophysiology. In addition, emAIH provides options to test therapeutic alternatives for patients not achieving remission, with reduced side effects of chronic nonspecific immunosuppression. (Hepatology 2013;58:718–728)




autoimmune hepatitis


experimental autoimmune encephalitis


experimental murine autoimmune hepatitis


formiminotransferase cyclodeaminase


insulin-dependent diabetes mellitus.

Autoimmune hepatitis (AIH) is a chronic autoimmune inflammation directed against liver tissue. It requires life-long immunosuppression in the majority of patients, accompanied by severe side effects due to chronic use of steroids and antimetabolites. Besides this, up to 20% of patients are nonresponders to steroid-based therapies, requiring more intense immunosuppression or eventually orthotopic liver transplantation due to advanced liver cirrhosis. In addition, patients with normal liver transaminases but remaining inflammation in control biopsies were recently reported to have worse clinical outcomes with increased mortality that cannot be influenced by enhanced immunosuppression.[1-3] The pathophysiology of AIH is still poorly understood and there are no reliable animal models resembling the disease which could be used to develop or test new therapeutic interventions.[4]

In terms of pathophysiology, genetic links were reported with MHC II alleles[5] and with immune regulatory genes such as cytotoxic T lymphocyte antigen 4 (CTLA-4).[6] Most of these findings usually followed reports of genetic predispositions in other autoimmune diseases and the contribution to liver-related autoimmunity remained unclear. Regarding the involvement of environmental triggers, several infections[7-9] and drug exposures were reported to precede the development of AIH, but so far no single causative agent has been identified. In addition, most patients are discovered late in the disease course and therefore infections preceding the diagnosis might not to be causative for the initiation of autoimmunity. It rather seems as if there is a long lag-period between initiation of autoimmunity and diagnosis of symptomatic disease, as almost 40% present with liver cirrhosis at time of diagnosis. Therefore, the role of environmental triggers also remains unclear. However, the search for causative agents was supported by the idea of molecular mimicry between environmental agents and self-antigens, usually searching for molecular identity.[10]

Animal models of AIH are usually restricted to short liver-specific immune responses usually ending in liver-specific tolerance rather than chronic autoimmunity. Although these models were very helpful in studying determinants of liver-specific immune responses, they were insufficient to explain AIH or to develop new therapeutic interventions.

We developed a model of chronic AIH by infecting NOD mice with replication deficient adenoviruses expressing the human liver autoantigen formiminotransferase cyclodeaminase (FTCD), formally known as anti-liver cytosol type 1 (LC-1), which is one of the key antigens in AIH Type 2. The mice were developing a chronic hepatitis that closely resembled the human AIH for the first time. Using this model we could demonstrate that experimental murine AIH (emAIH) is just initiated by a strong inflammatory danger signal. This is just occurring in genetically predisposed individuals, explaining the low prevalence of AIH in the general population. In addition, we could show that molecular similarity of the autoantigen is as efficient as molecular identity to lose tolerance against endogenous self-antigens. We could demonstrate that a break of humoral and cellular tolerance is required for the development of liver-specific autoimmunity and that CD4+ T cells act as drivers of the disease. Finally, classic immunosuppressive intervention with prednisolone and budesonide were successful in treating the disease. Taken together, we identified several key elements in the initiation of AIH. The model will be helpful to develop and test new therapeutic interventions in the future.

Materials and Methods

Ethics Statement

Animal care and experiments were done in accordance with institutional and national guidelines. All animal experiments were performed according to protocols approved by the Animal Welfare Commission of the Hannover Medical School and local Ethics Animal Review Board (Niedersaechsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit / LAVES, Oldenburg, Germany). The grant numbers covering the experiments are 06/1137 and 11/0342.


All animal experiments were performed under S2-conditions in the central animal facility of Hannover Medical School. NOD/Ltj (NOD) and NOD.CB17-Prkdcscid/J (NOD scid) mice were bred in specific pathogen-free facilities at Hannover Medical School. For hydrodynamic gene transfer experiments 15-20 μg of the plasmid were diluted in a total volume of 2 mL phosphate-buffered saline (PBS) and injected rapidly intravenously into NOD mice.


The unpaired Student 2-tailed t test or one-way analysis of variance (ANOVA) with Tukey's posttest was performed using the GraphPad Prism program: *difference significant with P  ≤  0.05; **very significant, P  ≤  0.01; ***extremely significant, P  ≤  0.001; P  >  0.05 was considered to be not significant (ns).

Additional methods are described in the Supporting Material.


Infection With Ad-hFTCD and Ad-eGFP (Enhanced Green Fluorescent Protein) Led to Transient Virus-Mediated Hepatitis

The new induced emAIH model described here is based on an immunological danger signal combined with expression of a hepatic autoantigen.

We used self-limited, hepatotropic adenovirus as a danger signal to trigger the immune response and the human and murine FTCD as common autoantigen in AIH. To this end, FTCD (Supporting Fig. 1A) and eGFP as control were cloned in a replication-deficient adenoviral vector. Cell lines were eGFP-positive under the fluorescence microscope after transducing with Ad-eGFP (data not shown). Western blots were stained with human anti-LC1-positive serum that recognized human FTCD. The human anti-LC-1 serum bound to hFTCD-transduced cells, but not to control infected cells (Supporting Fig. 1B). Therefore, FTCD was expressed in hepatocytes after Ad-FTCD infection in a tertiary structure which could recognize by human autoimmune sera.

As described by others,[11] adenoviral infection per se can be used to induce a transient hepatic inflammation in animal models. Therefore, NOD mice were infected with 1 × 10[10] particles of Ad-hFTCD and Ad-eGFP.

The levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were elevated at week 1 and week 2 in both groups. Values reached up to 900 and 1200 U/L of serum for AST and ALT, respectively, but returned to serum levels around the upper normal limit at later timepoints (Fig. 1A). This ALT and AST elevation was transient and not significantly different between mice infected with Ad-hFTCD or Ad-eGFP, which was in line with all hepatitis-like animal models up to date.[12-14]

Figure 1.

Adenoviral infection led to transient virus-mediated hepatitis and Ad-hFTCD-induced break of humoral tolerance. (A) Serum ALT and AST was determined at the indicated timepoints. (B) Liver cryosections at days 0, 5, and 20 days after Ad-eGFP infection. eGFP (green) and DAPI (blue) are shown. (C) Indirect immunofluorescence of rat liver sections with sera of Ad-eGFP and Ad-hFTCD-infected mice. (D) Quantification of sera from Ad-eGFP (n = 20) or Ad-hFTCD (n = 42)-infected mice tested in (C) for cytosolic autoantibodies or ANAs. (E) Western blot of Ad-hFTCD transduced HepG2 cells with representative sera from Ad-hFTCD, Ad-eGFP infected, and wild-type mice. (F) Serum protein electrophoresis of Ad-eGFP or Ad-hFTCD-infected NOD mice.

Ad-eGFP transduced hepatic cells, therefore eGFP-expressing cells were detectable in the liver transiently at days 1 and 5, but not at day 20 (Fig. 1B). Likewise, adenoviral DNA could not be identified by polymerase chain reaction (PCR) from liver tissue at week 12 after infection, demonstrating the self-limiting course of acute adenoviral infection (Supporting Fig. 2). In line with the molecular results, no remnants of the acute infection were observed in liver histology at 4 weeks after infection (data not shown). Therefore, acute infection with replication deficient adenoviruses resulted in acute but transient hepatitis independent of the expressed antigen.

Ad-hFTCD Infection Induced a Break of Humoral Tolerance

In AIH patients autoantibodies and their characterization are helpful tools to diagnose the disease. Therefore, we utilized rat liver sections and slides with HepG2 cells (Fig. 1C). As just high titer autoantibodies are helpful in the diagnostics of clinical AIH, we used a minimum dilution of 1:160 for all of our tests. Using these stringent criteria 92.3% of NOD mice infected with Ad-hFTCD developed autoantibodies directed against hepatocyte specific cytosolic antigens by immunofluorescence and 95% of mice developed antibodies against FTCD (Fig. 1C,D). 12.8% of sera were additionally positive for antinuclear antibodies (ANAs). This showed that immune reactions primed with a liver-specific antigen could lead to the development of antinuclear humoral reactivity. In contrast, just one animal infected with control adenovirus showed relevant autoantibodies. The autoantibody titer did not correlate with disease activity as measured by infiltrate size or Ishak score.

The specificity of the humoral immune responses was tested against recombinant antigens. Neither sera of wild-type nor of the Ad-eGFP controls reacted against human FTCD, while more than 90% of the sera from Ad-hFTCD infected animals recognized FTCD (Fig. 1E).

Another diagnostic criterion for human AIH is the hypergammaglobulinemia seen in 80% of patients. Comparable to human disease, a significant increase (P < 0.001) of the gamma globulin level from 2.5 mg/mL to 3.9 mg/mL in Ad-hFTCD-infected animals (Fig. 1F). Taken together, this demonstrated a break of humoral tolerance and development of antigen-specific autoantibodies in our emAIH model.

Massive Hepatic Infiltrations and Fibrosis in Ad-hFTCD-Infected Mice

After an acute phase of self-limited adenoviral hepatitis, mice were followed for development of autoimmune hepatitis. Pathological analysis of Ad-hFTCD-treated NOD mice after 12 weeks showed portal and periportal lymphoplasmacellular infiltrates, interface hepatitis, intralobular microgranulomas, and spotty single cell necrosis within lobules (Fig. 2A). In Ad-eGFP-infected NOD mice no relevant pathological signs of hepatitis were observed and liver sections of wild-type mice were lacking any inflammation. Grading of hepatitis activity was performed employing the Ishak score in a blinded manner. Ad-hFTCD mice showed a significantly increased grading compared to Ad-eGFP-infected control mice (Fig. 2A,B). In addition, the infiltrate size was significantly increased in Ad-hFTCD mice as compared to controls (Supporting Fig. 4A).

Figure 2.

Chronic hepatic infiltrations and fibrosis in Ad-hFTCD infected mice. (A) Hematoxylin and eosin (H&E) staining of liver sections from mice 12 weeks after Ad-hFTCD and Ad-eGFP infection with lobular inflammation with microgranuloma and single-cell necrosis. (B) Liver sections were analyzed at week 12 using the approved blinded Ishak scoring for chronic hepatitis. (C) Silver staining 12 and 30 weeks after Ad-hFTCD infection with periportal reticular fibers, bridging fibrosis, and dissociation of hepatocytes (▴). (D) Liver sections of NOD/Ltj at week 30 after Ad-hFTCD or Ad-eGFP injection. (E) Ishak grading NOD/Ltj at week 12 and week 30 after Ad-hFTCD injection. (F) Quantitative infiltrate analysis of FVB/N, C57Bl/6, and NOD/Ltj mice at 12 weeks after Ad-hFTCD infection.

Even more important, emAIH was just induced in NOD/Ltj mice and not in FVB/N or C57Bl/6J mice, establishing a role for genetic predisposition for the development of an organ-specific autoimmune disease (Fig. 2F; Supporting Fig. 4C).

Long-term follow-up at 30 weeks postinfection revealed an even advanced hepatitis, demonstrating the progressive course of the disease. Pathological analysis of the liver and Ishak score grading showed a significant (P = 0.0004) increase in hepatitis grade of Ad-hFTCD at 30 weeks compared to the 12-week timepoints (Fig. 2E; Supporting Fig. 4B). In this late stage of disease we observed liver fibrosis by silver staining of liver sections. Reticular fibers of connective tissue (Gomori), periportal fibrosis bridging to neighboring portal tracts, and reticular fibers leading to the dissociation of hepatocytes were observed in 4 of 8 Ad-hFTCD-infected NOD mice (50%) up to fibrosis score 3. In contrast, no meaningful fibrosis was seen in Ad-GFP-infected (Fig. 2C).

Immunofluorescence analysis 12 weeks after infection revealed that the cellular infiltrates consisted predominantly of CD4+ T cells and B cells. In contrast, only a few CD8+ T cells were found (Fig. 3A). In addition, flow cytometry analyses of intrahepatic leukocytes (IHLs) showed no differences in the gdT and abT cell compartment, including CD4+ and CD8+ subpopulations, and the natural killer T (NKT) cells (Fig. 3B-D; Supporting Fig. 5). Only NK cell numbers were significantly elevated in Ad-FTCD animals with emAIH compared to their Ad-eGFP controls (P = 0.0072). Total IHL numbers and absolute and relative numbers of the above subsets were not different between groups. In our model the average portal infiltrate size represents just 1%-2% of the liver area. As even the healthy liver is very rich in intrahepatic lymphocytes, the portal inflammation seen in our model was not sufficient to lead to a significant increase of total IHLs, which has so far just been reported in transgenic models or models with fulminant or fatal AIH due to ablation of several tolerance mechanisms.

Figure 3.

Characterization of IHLs. (A) Immunofluorescence staining of liver and spleen from Ad-hFTCD-infected NOD mice. CD4+(red), CD8+ (blue) T cells, and B cells (green) are shown. (B) Representative multilineage characterization of IHL from Ad-eGFP (n = 5) and Ad-hFTCD (n = 9)-infected mice by way of flow cytometry is shown. Indicated are the populations of the NK, NKT, and CD3+ T cells (left panels) and the CD4+ and CD8+ T-cell subpopulations gated on CD3+ cells (right panels). (C) Bar graph of all FACS blots gated on all IHLs. (D) As (C) gated on CD3+ cells.

Ad-hFTCD-Induced emAIH Is Driven by CD4+ T Cells

The break of humoral tolerance was demonstrated in various animal models for AIH, but T-cell responses with a break of cellular tolerance were not reported outside of transgenic systems.

Therefore, we attempted to adoptively transfer the emAIH by different immune cells of Ad-hFTCD-infected, autoimmune hepatitis-bearing mice. Purified CD4+ and CD8+ T-cell splenocytes as well as total splenocytes were activated with ConA and transferred into NODscid mice. Eight weeks after transfer all mice receiving activated cells from Ad-FTCD mice developed hepatitis by histopathological analysis (Fig. 4A) while no hepatitis was seen after transfer of activated cells from Ad-eGFP mice (data not shown). Encouraged by these results, naïve T-cell subpopulations were sorted from Ad-hFTCD infected NOD mice and transferred without in vitro activation. Even under these conditions animals that received CD4+ T cells developed hepatitis characterized by periportal infiltrates, which was not observed after transfer of CD8+ T (Fig. 4B). We also transferred sera of emAIH-bearing mice into lymphopenic NODscid or NOD mice without any signs of hepatitis in the recipients, which was a hint against antibody-mediated cytotoxicity (Fig. 4A).

Figure 4.

emAIH was driven by CD4+ T cells and was danger signal-dependent. H&E liver stainings 8 weeks after adoptive cell transfer from emAIH-bearing mice into NODscid mice. (A) Adoptive transfer of ConA-activated total splenocytes, CD4+ T cells or CD8+ T cells or serum alone as indicated (B) and of naïve CD4+ or CD8+ T cells. (C) ELISPOT analysis of cells from NOD mice infected with Ad-eGFP or Ad-hFTCD. Cells were stimulated as indicated. (D) Sera cytokine analysis from Ad-eGFP or Ad-hFTCD-infected mice at week 12 (n = 10). (E) Organ cryosections 8 days after hydrodynamic transfer of pIRES-eGFP expression vector into NOD mice. Shown are eGFP (green) and DAPI (blue). Scale bars = 100 μm. (B) H&E liver staining from NOD mice 12 weeks after transfer of pShuttle-CMV-hFTCD.

We next tested the antigen-specificity of the T cells in Ad-FTCD mice. To this end murine FTCD (mFTCD) was produced and purified to be used in enzyme-linked immunospot (ELISPOT) assays (Supporting Fig. 5A). Significantly more T cells from Ad-hFTCD-infected animals produced interferon-gamma (IFN-γ) in response to murine FTCD compared to controls (P = 0.0008), while responses to irrelevant proteins were at background levels (Fig. 4C). Besides the increase of IFN-γ on a cellular level increased concentrations of interleukin (IL)-12p70 and IL-17 (Fig. 4D) were found in sera of animals with chronic emAIH compared to controls, while tumor necrosis factor alpha (TNF-α), IL-10, and T-helper 2 (TH2) cytokines like IL-4 and IL-5 were not changed (Supporting Fig. 5). This highlights that emAIH involves prominent TH1 and TH17 responses, and therefore defines clearer targets for immunointerventions.

Ad-hFTCD-Induced Hepatitis Model Is Danger Signal-Dependent

It remained questionable if the initial adenoviral infection was indeed required to initiate a chronic evolving autoimmune hepatitis. To omit these strong proinflammatory signals, hydrodynamic transfection was used to express the orthologous protein in a large proportion of hepatocytes. Reports have described this delivery method to be even more effective than direct injection into the target organ.[15]

We used a vector containing the gene for hFTCD (CMV-hFTCD) or eGFP (CMV-eGFP) under control of the CMV-promoter and performed a hydrodynamic transfer as described. Organs of recipients were analyzed 8 days after gene transfer. While there were almost no eGFP-expressing cells in lung or kidney, almost all hepatocytes were eGFP-positive. In fact, the number of eGFP-positive cells was similar on day 5 after hydrodynamic transfection compared to Ad-eGFP-infected animals (Fig. 4E). When we analyzed these mice after 12 weeks, neither the CMV-eGFP nor the CMV-hFTCD group (Fig. 4F) developed any signs of chronic hepatitis. This supports the notion that an inflammation amplified by danger signals was necessary to break tolerance against liver tissue.

Molecular Similarity Was as Effective in Inducing emAIH as Molecular Identity

Reports of environmental agents or infections preceding AIH so far concentrated on molecular identity searching for highest sequence homology to endogenous self-antigens. We therefore wondered whether molecular similarity is as efficient as identity in initiating emAIH.

To this end, we tested molecular identity by an adenoviral construct coding for murine FTCD (Ad-mFTCD) and the results were compared to emAIH induced by Ad-hFTCD.

The break of humoral tolerance was comparable in both settings. Even if the reactivity of autoantibodies recognizing FTCD was lower in the sera of Ad-mFTCD recipients (Fig. 5A), the overall humoral autoimmunity and the amount of gamma globulins was unchanged (Fig. 5B,C). The humoral tolerance is easier to break than the cellular tolerance. It was therefore interesting to note that the livers of Ad-mFTCD-infected NOD mice were as severely inflamed by lymphocytes as the mice infected with the orthologous Ad-hFTCD (Fig. 5D). This resulted in an equal staging by blinded Ishak scoring (Fig. 5E) and in comparable sizes of hepatic infiltrates (data not shown). Moreover, T-cell responses against mFTCD were comparable in Ad-mFTCD mice (Fig. 5F) as compared to Ad-hFTCD mice (Fig. 4C). Taken together, molecular similarity was as effective as molecular identity in triggering emAIH.

Figure 5.

Molecular similarity was as effective in inducing emAIH as molecular identity. (A) Western blot of Ad-hFTCD transduced HepG2 cells with sera from Ad-hFTCD, Ad-mFTCD-infected, or wild-type NOD mice. (B) Indirect immunofluorescence staining of rat liver sections with sera of Ad-eGFP and Ad-mFTCD-infected NOD mice with DAPI counterstained. (C) Gamma globulin levels in sera from NOD mice infected with Ad-eGFP, Ad-hFTCD, and Ad-mFTCD. (D) H&E staining of NOD mice 12 weeks after Ad-mFTCD or Ad-hFTCD infection. (E) Liver sections were histologically graded using the Ishak score. (F) ELISPOT analysis of cells from Ad-mFTCD or Ad-eGFP-infected NOD mice that were stimulated as indicated.

Classic Therapeutic Intervention Can Cure emAIH

We next wanted to test if the newly developed model of emAIH was not only suited to investigate the pathophysiology of the disease but if it could also be used to develop new therapeutic strategies. To this end it was important to show that the disease can be successfully treated with our current immunosuppressive standard therapies. Indeed, a single 8-week course of oral therapy with prednisolone was sufficient to substantially improve the hepatitis and 70% of investigated mice showed complete histological remission (Fig. 6). Recently, a prospective randomized trial has established budesonide as an alternative to prednisolone for induction of remission.[16] In fact budesonide was superior to prednisolone in inducing remission and patients treated with budesonide had fewer steroid-related side effects. It was therefore reassuring that budesonide was as effective as prednisolone in treating murine hepatitis. As both standard immunosuppressive therapies were also effective in treating emAIH, we suggest that our model might also be suited to develop and test new therapeutic regimens for treatment of AIH.

Figure 6.

Classic therapeutic interventions can cure emAIH. Quantitative analysis of the average size of intrahepatic infiltrates at week 16 after Ad-hFTCD infection. The groups received no therapy (n = 11), prednisolone (n = 13), or budesonide (n = 10).


 We developed a model of experimental murine autoimmune hepatitis closely resembling the human disease. Semi-blinded pathological analysis showed features suggestive of AIH such as lymphoplasmacellular infiltrates, interface hepatitis, and lobular inflammation. The disease was chronically evolving after initial priming, which is in sharp contrast to all animal models using liver-specific expression of a model antigen paired with T cells from T-cell receptor transgenic animals.[14, 17-19] In all such models hepatitis is rather short-lived and the outcome of the initial immune attack is usually tolerance. Even in double transgenic models suggesting a chronic hepatitis the liver-specific CD8+ T cells were rather hyporeactive and anergic.[14, 18] Nonetheless, some approaches result in massive lymphoid infiltrations as reasoned by the high precursor frequency of transgenic effector cells.[14, 19] Although these models were helpful in understanding tolerance mediated by the liver, they were not very helpful to study the pathophysiology of AIH or to develop new therapeutic concepts. Models of human disease with a polyspecific T-cell repertoire are, on the other hand, limited to identify exact cellular and molecular mechanisms, as the precursor frequency and affinity of autoreactive is usually low in this setting.

AST and ALT were not elevated during the chronic phase of the emAIH. However, this is well in line with other models of chronic hepatic inflammations in mice, where no elevated transaminases were seen despite heavy inflammation in histological analysis.[12-14] It could well be that murine hepatocytes are not as prone to damage caused by chronic inflammation. Besides this, the normal values for AST and ALT are not well defined in mice and seemed to be higher than the ones reported for humans (e.g., normal values for ALT were 70-120 U/L for our NOD/Ltj strain as compared to below 50 U/L for human samples).

Therefore, transaminase levels are probably not the best parameter to monitor the disease. In this respect it is interesting to note that recently reports were published on AIH patients with complete biochemical remission but still significant inflammation on histology.[1-3] The total intrahepatic number was not changed in our model. This was not expected, given the fact that the portal infiltrates just represent 1%-2% of the analyzed hepatic area. The only occasions in which the number of IHLs were increased were T-cell receptor transgenic models with very high precursor frequencies or models of fulminant and fatal hepatitis caused by simultaneous ablation of several tolerance mechanisms.[14, 19, 20]

In addition to the chronic evolving nature of emAIH, we also detected portal and lobular and advanced bridging fibrosis up to F3 within just 30 weeks as seen in patients with AIH. This is the first time that such a development of fibrosis was seen in an animal model. Christen and coworkers[12] also reported on the development of subcapsular fibrosis in their AIH models, but the fibrosis in their model was not typical for AIH. In fact, the development of fibrosis in that model was completely dependent on intraperitoneal application of adenovirus. The strong intraperitoneal immune response could potentially be responsible for the development of subcapsular fibrosis and not the intrahepatic inflammation itself.

Despite these criticisms the model of Christen and colleagues comes closest to our model in that hepatic infiltrates were caused by transient adenovirus-mediated hepatitis. But the study was just studying the break of humoral tolerance. T-cell responses and drivers of autoimmunity were not identified and therapeutic interventions not tested. The same holds true for the studies of Alvarez and coworkers[13] in which hepatic infiltrates developed rather late after priming with an artificial fusion protein containing parts of liver autoantigens.

In addition to the striking similarity of emAIH with AIH in humans,[21] we could also demonstrate that the disease can be successfully treated with classical immunosuppressive therapy used in patients with AIH. This also opens the opportunity to develop and test new therapeutic interventions in the future. Such therapies are desperately needed to reduce the side effects of chronic unspecific immunosuppression on the one hand and to offer new therapeutic alternatives for patients not reaching a complete histological remission.

As the new model of emAIH closely resembled human disease it was suited to investigate the pathophysiology of disease initiation and perpetuation.

Only NOD mice were susceptible to induction of emAIH, while C57BL/6 and FVB/N mice did not develop chronic AIH after an initial Ad-FTCD infection. This highlights the importance of genetic predisposition for the development of autoimmunity. A strain specificity for development of hepatitis had been reported earlier,[22] but this was usually not linked to genetic predispositions also seen in AIH patients. In fact, the NOD mice are not just developing type I diabetes, but are also developing sialitis, autoimmune neuropathy, and experimental autoimmune encephalitis (EAE) after induction.[23] The clinical coincidence of several autoimmune diseases in patients supports the idea of common genetic differences predisposing for the development of autoimmunity. The basis for this is susceptible genes shared by many patients with autoimmune diseases, while other genes are rather linked to organ-specific manifestations (e.g., insulin gene with type I diabetes).[24] To this end it is interesting to note that the strongest genetic link for the development of type 1 diabetes is the unique MHC haplotype I-Ag7 (idd1). This correlates well with the association of HLA DR3/4 in human type 1 diabetes (IDDM1). Interestingly, an HLA DR3/4 identical association was also reported for patients with AIH. In detail, a predisposition of extended HLA-DRB1 alleles like DRB1*0301, DRB1*0401, DRB1*0404, and DRB1*0405 can be found in patients with AIH[25] as well as patients with type 1 diabetes.[26] In addition, non-HLA susceptibility genes like CTLA-4 (IDDM12), CD45, TNF-α, or vitamin D receptor (IDDM34)[27-30] are associated with the development of type I diabetes and were also associated with the development of AIH.[6, 30-34] Taken together, the use of the NOD strain for successful induction of emAIH underlines the importance of these genetic associations for the induction of AIH. Congenic NOD mice will therefore serve as valuable tools to delineate the importance of individual genes for the development of AIH in the future. The genetic predisposition of NOD mice might also explain the rather rapid development of emAIH at 12 weeks with fibrosis at 30 weeks. Alvarez and coworkers[13] did not see any AIH in C57BL/6 animals at 10 weeks after induction. In fact, they could just observe infiltrates at 8 months.

While genetic predisposition is creating a fertile field for the development of autoimmunity, it is clearly not sufficient for disease induction, as NOD mice do not develop emAIH spontaneously. This highlights the importance of environmental factors triggering autoimmunity. In fact, many environmental and especially infectious agents were suggested to induce AIH in humans.[7-9] However, the lag phase between initiation of autoimmunity and diagnosis of the disease makes it difficult to causally relate environmental triggers to the disease initiation. Therefore, the proof that a short, self-limiting, liver-specific infection can cause chronic evolving emAIH is of major importance, as it highlights the need for a strong inflammatory stimulus to break hepatic tolerance. In fact, we could prove that widespread expression of the autoantigen by hydrodynamic transfection was not sufficient to prime emAIH. The absence of detectable virus in the chronic course of disease highlights the short-lived nature of the infection and supports a hit-and-run hypothesis for the development of AIH in which an initial time-limited strong stimulus is sufficient to trigger autoimmunity. To this end, it is interesting to note that attempts to treat autoimmunity with antibiotics or antiviral therapy have largely failed, supporting that a constant trigger is not necessary.[35]

While studies trying to identify environmental triggers have largely focused on molecular identity, we show with our experiments that molecular similarity is as efficient in triggering chronic autoimmunity.[36] This broadens the spectrum of potential environmental agents which could lead to a loss of tolerance against tissue-specific self-antigens.

Although innate and adaptive immune responses are usually involved in autoimmune tissue destruction, drivers of the autoimmune disease were so far not identified for AIH. The break of humoral tolerance against hepatic antigens was also reported by other groups[12, 13]; this is probably not sufficient to lead to hepatitis, as serum transfer did not lead to hepatitis in our model. Instead, we could demonstrate the break of T-cell tolerance with evolving TH1/TH17 cytokine profile. In addition we could demonstrate that the disease could be transferred by CD4+ T cells, thereby identifying antigen-specific CD4+ T cells as the potential drivers of emAIH. This would be well in line with the described genetic association with MHC II alleles in AIH.[5]

In summary, we have developed a model of experimental murine AIH which closely resembles the human disease. The model was used to explain fundamental aspects of the pathophysiology of initiation and perpetuation of AIH. In addition, standard immunosuppressive therapy could successfully treat the disease, thereby opening the possibility for the development of new therapeutic interventions in the future. These therapies should avoid the side effects of chronic unspecific immunosuppression and offer an alternative for patients not achieving histological remission with standard therapy.


We thank Maren Sievers and Konstantinos Iordanidis for technical assistance in performing the experiments and the laboratory for detection of liver-specific autoantibodies, Dept. of Gastroenterology, Hepatology & Endocrinology, and Prof. Ralf Lichtinghagen from the Inst. of Clinical Chemistry for technical assistance.

Author Contributions: study concept and design (MHW, EJ); acquisition of data (MHW, KF, FN, JS, CSF, MS, NW, FK, RT); analysis and interpretation of data (MHW, KF, FN, RT, EJ); drafting of the article (MHW, EJ); critical revision of the article for important intellectual content (KF, FN, JS, CSF, FK, RT, MPM); statistical analysis (MHW, KF); obtained funding (MPM, EJ); administrative, technical, or material support (JS, CSF, NW, FK, MPM); study supervision (MHW, EJ).