Potential conflict of interest: Nothing to report.
The pathogenesis of autoimmune liver diseases is poorly understood. Animal models are necessary to investigate antigen presentation and priming of T-cells in the context of autoimmunity in the liver. Transgenic mouse models were generated in which the model antigen ovalbumin is expressed in hepatocytes (TF-OVA) or cholangiocytes (ASBT-OVA). Transgenic OT-I (CD8) or OT-II (CD4) T-cells specific for ovalbumin were adoptively transferred into TF-OVA and ASBT-OVA mice to induce in vivo priming of antigen-specific T-cells. T-cell migration and activation, as well as induction of liver inflammation, were studied. OT-I T-cells preferentially located to the liver of both mouse strains whereas no migration of OT-II T-cells to the liver was observed. OT-I T-cells proliferated in the liver of TF-OVA mice and the liver and liver draining lymph nodes of ASBT-OVA mice. OT-II CD4 T-cells were activated in spleen and liver draining lymph node of TF-OVA mice but not in ASBT-OVA mice. Transfer of OT-I T-cells led to histologically distinct inflammatory conditions in the liver of ASBT-OVA and TF-OVA mice and caused liver injury as determined by the elevation of serum alanine aminotransferase. Conclusion: An antigen expressed in hepatocytes is presented to CD8 and CD4 T-cells, whereas the same antigen expressed in cholangiocytes is presented to CD8 but not CD4 T-cells. In both models, activation of CD8 T-cells occurs within the liver and causes liver inflammation. The models presented here are valuable to investigate the priming of T-cells in the liver and their role in the development of autoimmune disease of the liver. (HEPATOLOGY 2007.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Autoimmune hepatitis and cholangitis are triggered by autoreactive T-cells. Animal models are needed to study the early events in their pathogenesis, namely, the priming of autoreactive T-cells. The requirements for an immune-based animal model are restriction of the immune response to the liver, antigen specificity, and the potential to study the priming of CD8 and CD4 T-cells. Several animal models of autoimmune hepatitis have been developed,1 none of which fulfills these criteria. Transgenic expression of foreign major histocompatibility complex (MHC) class I molecules in the liver has been widely used.2–4 However, this model has a significant disadvantage: unlike the pathophysiology of an immune or autoimmune reaction, CD8 T cells recognize their antigen on hepatocytes but the antigen is not presented by professional antigen-presenting cells (APCs) in this model. Injection of the synthetic peptide ova p257–264 (SIINFEKL) into mice followed by transfer of antigen-specific CD8 T-cells has also been used,5 but the possibility of peptide binding to MHC-I molecules on T-cells themselves and their activation by each other6 or by unrelated cells is a possible concern. Infection with virus expressing ovalbumin7 results in temporary expression of ovalbumin but also activation of innate immune mechanisms. No animal model exists that allows the simultaneous study of the antigen-specific role of CD8 and CD4 T-cells.
Likewise, animal models of autoimmune cholangitis are rarely immune based. Most models are established in rats and rely on bacterial or chemical agents to induce cholangitis.8 A murine graft-versus-host model leads to nonsuppurative cholangitis9; however, the triggering minor MHC antigens are unknown. Recently, a mouse model expressing ovalbumin in biliary epithelia under the rat apical sodium-dependent bile transporter (ASBT) promoter was reported by an independent group.10
The liver has the potential to activate T-cells; however, whether activation results in full effector function7, 11 or leads to tolerance and premature death of T-lymphocytes2, 12, 13 is controversial. Several populations of APCs are present in the liver, namely, bone marrow–derived dendritic cells14 and Kupffer cells,15 and non–bone marrow–derived liver sinusoidal endothelial cells.16 In addition, APCs in the liver-draining lymph node present and cross-present antigens.17 In contrast to other organs, naïve T-cells gain access to the organ's parenchyma and interact with hepatocytes through the fenestrated endothelium of liver sinusoids,18 leading to antigen-specific trapping of T-cells.19
We expressed ovalbumin in hepatocytes or cholangiocytes in transgenic mice, affording presentation of the antigen by the target cells and by professional APCs. The usefulness of this approach was proven in RIP-mOVA mice that express ovalbumin under control of the rat insulin promoter,20 leading to presentation by professional APCs.21 We elucidated the differences resulting from expression in the liver parenchyma or in bile ducts and investigated which antigen-presenting site is responsible for the priming of CD8 and CD4 T-cells by an endogenous, liver-restricted antigen.
ALT, alanine aminotransferase; APC, antigen-presenting cells; ASBT, apical sodium-dependent bile transporter; ASBT-OVA, mouse model in which the model antigen ovalbumin is expressed in cholangiocytes; CFSE, carboxyfluorescein diacetate succinimidyl ester; IFN-γ, interferon gamma; MHC, major histocompatibility complex; OT-I,; PCR, polymerase chain reaction; SIINFEKL, synthetic peptide ova p257–264; TF, transferrin; TF-OVA, mouse model in which the model antigen ovalbumin is expressed in hepatocytes.
Animals and Cells.
For the generation of mice in which the model antigen ovalbumin is expressed in cholangiocytes (ASBT-OVA), a 2,900–base pair (bp) fragment from the ASBT promoter region22 was isolated from C57BL/6 genomic deoxyribonucleic acid by polymerase chain reaction (PCR) with primers 5′-CGGGTACCGGAGACGTTTGGAGGATAGGG and 5′-CGCTCGAGCACTGCTTGTGCTGTGCAAATG, introducing a 5′ KpnI and 3′ XhoI site. A 2,980-bp fragment from the transferrin (TF) promoter region23 was amplified using primers 5′-CGGGTACCCGAAGGACGAAGGACCGTCGG and 5′-CGGAGCTCCCTCTCGGTGTGTGTGTGGCG. The correct sequence was verified. Both fragments were cloned into pBluescript II SK+ (Stratagene, La Jolla, CA), followed by introduction of a 2,470-bp HindIII-NotI fragment from pBlueRIP (a gift from Dr. F.R. Carbone), which encodes the first 118 amino acids of the human transferrin-receptor linked to residues 139 though 385 of ovalbumin and a polyA sequence.20 Digestion of the constructs with KpnI and NotI generated a fragment of approximately 5,500 bp, which was microinjected into fertilized C57BL/6 female pronuclei. To detect the transgene, genomic deoxyribonucleic acid was analyzed by PCR to identify a 480-bp fragment using primers 5′-CAAGCACATCGCAACCA and 5′-GCAATTGCCTTGTCAGCAT.
Bone marrow chimeras were generated by lethal irradiation of mice, followed by supplementation with C57BL/6J/β2m−/− bone marrow (Taconic, Hudson, NY). Mice were used for experiments after 6 weeks. For splenectomy, mice were anesthetized with xylazine and ketamine and used for experiments after 2 weeks.
CD8 and CD4 T-cells were isolated from lymph nodes of OT-I, OT-I RAG−/−, OT-II, or C57BL/6 mice using the isolation kits from Miltenyi (Bergisch-Gladbach, Germany). OT-I and OT-II T-cells are transgenic for T-cell receptors specific for the ovalbumin-derived peptides SIINFEKL/H-2Kb24 and ISQAVHAAHAEINEAGR/I-Ab25, respectively. Purity of preparations was above 90%. For some experiments, cells were further selected for CD62L expression using magnetic beads (Miltenyi). Cells were transferred intravenously in the indicated numbers. For proliferation assays, labeling with carboxyfluorescein succinimidyl ester (CFSE) was performed as described.26 All animals received humane care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences.
PCR, Western Blot Analysis, and Immunohistochemistry.
Ribonucleic acid was extracted using the Qiagen RNeasy kit and digested with deoxyribonuclease (Qiagen, Hilden, Germany). Ribonucleic acid was reverse-transcribed using the cloned AMV First-Strand complementary deoxyribonucleic acid kit from Invitrogen (Karlsruhe, Germany) and amplified using Invitrogen's Platinum PCR Supermix. Protein was extracted from mouse tissues by disruption in lysis buffer [0.5% Nonidet-P40, Tris-Cl pH 7.8, 2 mM ethylenediamine tetraacetic acid, 50 mM β-gylcerolphosphate, 1% glycerol]. Equal amounts of protein were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on a 12.5% gel and transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA). Membranes were probed with rabbit polyclonal anti-ovalbumin antibody (1:1,000, USBiological, Swampscott, MA). Reaction products were visualized with goat anti-rabbit antibody (Dianova, Hamburg, Germany), followed by enhanced chemiluminescence (PerkinElmer, Wellesley, MA).
For histologic analysis, livers were perfused and fixed for 24 hours in 4% paraformaldehyde, followed by embedding in paraffin. Four-micrometer sections were stained with hematoxylin-eosin.
For immunohistochemistry, 4 μm sections were deparaffinized and subjected to a heat-induced epitope retrieval step. Sections were immersed in sodium citrate buffer (pH 6.0) and heated in a high-pressure cooker for 2 minutes. Slides were rinsed in cool water, washed in Tris-buffered saline (pH 7.4), treated with peroxidase or biotin blocking agent (Dako, Hamburg, Germany), and incubated with antibodies against ovalbumin (Chemicon, Temecula, CA, 1:200), CD3 (Dako, 1:100), and Ki-67 (Dako, 1:2000). For detection of ovalbumin, an indirect immunoperoxidase method (EnVision peroxidase kit K 4010, Dako) was employed. Peroxidase was developed with diaminobenzidine chromogenic substrate for 10 minutes. Binding of polyclonal anti-CD3 antibody was visualized with biotinylated donkey anti-rabbit antibody (Dianova, Hamburg, Germany). Mouse anti-Ki-67 was labeled with monoclonal rat antibody (clone TEC-3) followed by biotinylated rabbit anti-rat antibody (Dako) and the StreptavidinAP kit (K0391, Dako), using Fast Red as chromogen. Negative controls were performed by omitting the primary antibodies.
Fluorescence-Activated Cell Sorter Analysis.
Liver-infiltrating cells were isolated by perfusion of the liver with digestion media [0.5 mg/ml collagenase (Sigma, Taufkirchen, Germany) in Roswell Park Memorial Institute medium]. Fragmented livers were incubated in digestion media for 20 minutes at 37°C in an agitating incubator, passed through a nylon mesh, and briefly centrifuged at 300 rpm. The supernatant was centrifuged (1,400 rpm, 5 minutes), and cells were purified on a discontinuous 40/70% Percoll gradient (2,000 rpm, 20 minutes). Antibodies anti-CD8α-PerCP, anti-CD4-PerCP, and anti-Vα2-PE were from Pharmingen (Heidelberg, Germany). For intracellular staining, cells were fixed (1% paraformaldehyd) followed by permeabilization (0.5% saponin, Sigma). Staining was performed with antibodies anti-interferon gamma (IFN-γ) APC, and anti-interleukin 2 phycoerythrin (Pharmingen) in the presence of 0.5% saponin. Cells were analyzed on a Becton Dickinson FACSCalibur using the CellQuest software.
T-Cell Restimulation and In Vivo Cytolysis Assay.
For restimulation experiments, 4 Mio OT-I T-cells were transferred into ASBT-OVA or TF-OVA mice. Mice were killed at day 6, nonparenchymal cells were isolated from the liver, cultured in Roswell Park Memorial Institute medium supplemented with 10% fetal bovine serum (Gibco, Karlsruhe, Germany), and activated in 20 nM phorbol 12-myristate 13-acetate/1 μM ionomycin (Sigma, Taufkirchen, Germany). After 2 hours, 2 μg/ml brefeldin-A (Sigma) was added, and cells were analyzed for intracellular cytokines after an additional 4 hours.
For in vivo cytolysis assays, 4 million OT-I T-cells were transferred into ASBT-OVA or TF-OVA mice. Splenocytes from C57BL/6J mice were pulsed with 1 μg/ml SIINFEKL or left untreated for 1 hour. SIINFEKL-pulsed splenocytes and control splenocytes were labeled in 15 μmol/L or 1.5 μmol/L CFSE, respectively. Cells were mixed in equal numbers, and 8 million splenocytes were injected intravenously into ASBT-OVA or TF-OVA mice 6 days after the transfer of OT-I T-cells. As a control, splenocytes were injected into mice that had not received OT-I CD8 T-cells. After 5 hours, cells were isolated from lymphatic organs and liver and analyzed for CFSE staining. Specific lysis was calculated as follows: 100 × [1 − (%CFSElo(control)/%CFSEhigh(control))/(%CFSElo(OT-I)/%CFSEhigh(OT-I))].
Alanine Aminotransferase Measurement.
Serum samples were obtained on the indicated days. Blood was collected into separation tubes (Becton Dickinson, Heidelberg, Germany), and sera were stored at −20°C before automatic analysis on a Roche modular analyser (Grenzach-Wyhlen, Germany).
Liver-Specific Expression of Ovalbumin in Transgenic Mice.
We generated transgenic mice that express ovalbumin in cholangiocytes (ASBT-OVA mice) or hepatocytes (TF-OVA mice). To prevent secretion of the transgene, ovalbumin was fused to the transferrin-receptor transmembrane domain (Fig. 1A). Genomic integration of the transgene was proved by PCR. The pattern of messenger ribonucleic acid expression was determined by reverse transcription polymerase chain reaction, confirming expression in the liver of both strains and minimal expression in lung and testis of TF-OVA mice and ileum of ASBT-OVA mice (Fig. 1B). Western blot analysis detected significant amounts of ovalbumin in the liver of ASBT-OVA and TF-OVA mice but not in any other organ tested or in the liver of C57BL/6J mice (Fig. 1C). Immunohistochemistry showed expression of ovalbumin in hepatocytes of TF-OVA (Fig. 1D) but not C57BL/6J (Fig. 1E) or ASBT-OVA mice (Fig. 1F), whereas ASBT-OVA mice showed expression in the epithelium of medium-sized and large bile ducts. The signal was best discernible in larger bile ducts because of the larger cytosol-to-nucleus ratio (Fig. 1G). Mice were healthy and showed no sign of liver disease by microscopy or elevation of liver enzymes (data not shown).
Differential Migration of CD8 and CD4 T-Cells in ASBT-OVA and TF-OVA Mice.
OT-I CD8 and OT-II CD4 T-cells were purified from lymph nodes of OT-I and OT-II mice, respectively. Purified cells showed a naïve phenotype (CD69lowCD62Lhigh, data not shown). Four million OT-I or OT-II T-cells were injected intravenously into ASBT-OVA or TF-OVA mice. In both strains, preferential migration of CD8 T-cells to the liver was observed. Whereas CD8+Vα2+ T-cells were also retrieved from other lymphatic organs of ASBT-OVA mice in significant numbers, few CD8+Vα2+ T-cells were present in lymph nodes or spleen of TF-OVA mice, indicating efficient trapping of the T-cells in the liver (Fig. 2A). Migration experiments were performed at 20 hours before proliferation of T-cells had occurred (see later discussion) to assure that results were not obscured by proliferation. In the case of OT-II CD4 T-cells, no migration to the liver of ASBT-OVA or TF-OVA mice was observed. There was a slight preference for the spleen in TF-OVA mice, whereas no specific migration pattern was observed in ASBT-OVA mice (Fig. 2B).
In both strains, an increase in the number of CD8+Vα2+ T-cells was observed in the liver as early as day 2 after transfer (Fig. 2C). At day 3, approximately 8-fold higher cell numbers were retrieved from the liver of TF-OVA than ASBT-OVA mice. Numbers returned to normal at approximately day 12 after transfer. A small increase in cell numbers was also observed in the liver draining lymph node at day 3, whereas numbers in spleen and inguinal lymph node started to increase only at day 6, arguing for redistribution of T-cells rather than for activation at these sites. Because CD8 and Vα2 were efficiently down-regulated after homing of transferred T-cells to the liver (data not shown), the real number of cells infiltrating the liver was likely underestimated by enumeration of CD8+Vα2+ cells (Supplementary Fig. 2).
Differential Priming of CD8 T-Cells in ASBT-OVA and TF-OVA Mice.
Next, we investigated in which lymphatic compartment proliferation of T-cells occurred. No proliferation of T-cells was observed at 20 hours. At 44 hours, proliferation was observed in the liver of TF-OVA mice. Infrequently, small numbers of proliferating cells were also observed in the spleen. The small number of cells that had divided severalfold argues against proliferation within the spleen and for early redistribution of OT-I cells activated within the liver. No proliferation was observed in any of the tested lymph nodes, including the liver draining lymph node (Fig. 3A). In ASBT-OVA mice, proliferation was noted in the liver, although with somewhat slower kinetics than in TF-OVA mice. In addition, proliferation was observed in the liver-draining but not the inguinal lymph node of ASBT-OVA mice. Kinetics of proliferation were slower in the liver-draining lymph node than in the liver (Fig. 3B).
Transfer of C57BL/6J CD8 T-cells into TF-OVA or ASBT-OVA mice as well as transfer of OT-I T-cells into C57BL/6J mice did not result in proliferation in the liver (Fig. 3C) or any other lymphatic organ (data not shown) for up to 68 hours after transfer.
Because T-cells were isolated from mice on a RAG-proficient background, some of the transferred T-cells may have been memory cells. Therefore, we repeated the same experiments after selection of OT-I T-cells for CD62Lhigh cells and with OT-I RAG−/− CD8 T-cells, with identical results (Supplementary Fig. 1), excluding the presence of CD8 memory T-cells as the reason for the observed effects.
Differential Priming of CD4 T-Cells in ASBT-OVA and TF-OVA Mice.
In the case of OT-II T-cells transferred into TF-OVA mice, no proliferation was observed at 20 hours. At 44 hours, proliferation occurred in the spleen and the liver-draining lymph node but not the liver (Fig. 4A). Few CD4 T-cells were retrieved from nonrelated lymph nodes. In ASBT-OVA mice, no proliferation of OT-II CD4 T-cells was observed in any of the organs tested for up to 160 hours. The reduction of staining intensity at the 160-hour time point in Fig. 4B is attributable to bleaching of the CFSE dye and not to dilution by cell division. Transfer of C57BL/6J CD4 T-cells into TF-OVA or ASBT-OVA mice as well as transfer of OT-II T-cells into C57BL/6J mice did not result in proliferation in the spleen (Fig. 4C), liver, or any other lymphatic organ for up to 68 hours after transfer (data not shown). We repeated the same experiments after selection of OT-II T-cells for CD62Lhigh cells, with identical results (data not shown). Selection for CD62L, however, does not deplete central memory CD4 T-cells that may contribute to these findings.
Induction of Liver Inflammation by CD8 T-Cells in ASBT-OVA and TF-OVA Mice.
Next, we performed histological analysis of livers from ASBT-OVA and TF-OVA mice 7 days after transfer of OT-I T-cells. Transfer of 8 million OT-I T-cells led to infiltration of the livers of TF-OVA mice. Portal as well as lobular infiltrates were identified, demonstrating severe hepatitis (Fig. 5A,B). Infiltrating cells contained CD3-positive T-cells (Fig. 5C) and a large fraction of proliferating cells (Fig. 5D). In contrast, the infiltrate was restricted to portal tracts in ASBT-OVA mice with condensation of T-cell infiltrates around medium-sized bile ducts (Fig. 5E,F), whereas lobular infiltrations and infiltrations in peripheral portal tracts were largely absent (inset, Fig. 5F). Immunohistochemistry confirmed the presence of T-cells (Fig. 5G) and proliferating cells (Fig. 5H). No significant infiltration in the liver was noted after transfer of 8 million OT-II CD4 T-cells (data not shown).
To determine the time course of hepatitis caused by OT-I T-cells, we analyzed serum alanine aminotransferase (ALT) levels. ALT levels in TF-OVA mice receiving 8 million OT-I T-cells peaked at day 5 and returned to baseline by day 12 to 15. A maximum increase to approximately 9-fold over baseline was observed. In contrast, the increase in ALT levels was mild in ASBT-OVA mice (maximum, 1.5-fold over baseline) and returned to baseline by day 8 (Fig. 6A). Transfer of 8 million OT-II CD4 T-cells did not lead to an increase in serum ALT levels (data not shown).
CD8 T-Cells Activated in ASBT-OVA and TF-OVA Mice Display Effector Function.
We examined whether OT-I T-cells primed in the livers of TF-OVA or ASBT-OVA mice produce effector cytokines. Interferon-gamma (IFN-γ) production was observed in a substantial fraction of liver-infiltrating T-cells from TF-OVA mice 6 days after transfer, whereas only few T-cells isolated from ASBT-OVA livers produced IFN-γ without restimulation (Fig. 6B). After restimulation in vitro for 6 hours, liver-infiltrating CD8+Vα2+T-cells from both mouse strains produced IFN-γ, although to a higher extent in TF-OVA than in ASBT-OVA mice. The capability of in vivo primed OT-I T-cells to lyse target cells was analyzed by in vivo cytolysis assay. In both mouse strains, antigen-specific cytolytic effects were observed (Fig. 6C), although the extent of specific cytolysis was higher in TF-OVA than in ASBT-OVA mice (Fig. 6D). Cytolysis was observed in all investigated organs, namely, peripheral lymph nodes, spleen, and liver.
Because OT-I CD8 T-cells also located to the spleen after transfer and some proliferation was observed at this site, we transferred OT-I CD8 T-cells into splenectomized mice to assure that priming in the spleen was not responsible for activation of T-cells. The degree of hepatitis caused in splenectomized TF-OVA mice was greater than that in nonsplenectomized TF-OVA mice as determined by ALT serum levels, arguing against a role of the spleen in priming of CD8 T-cells in our model (Fig. 7A). In splenectomized ASBT-OVA mice, the same degree of ALT elevation was observed as in nonsplenectomized mice (Fig. 7B).
Bone Marrow–Derived Antigen-Presenting Cells Are Required for Full Activation of CD8 T-Cells.
To address the question of which subset of APCs is responsible for CD8 T-cell activation, we generated bone marrow chimeras, in which bone marrow–derived APCs are unable to present ovalbumin whereas hepatocytes or cholangiocytes and LSECs remain capable of doing so. Release of ALT was greatly reduced in TF-OVA/β2m−/− chimeras, demonstrating the requirement for bone marrow–derived APCs to induce CD8 T-cell–mediated hepatitis (Fig. 8A). In ASBT-OVA/β2m−/− chimeras, no significant reduction of ALT release was observed, suggesting that ALT release is caused by bystander mechanisms rather than cytolytic T-cell effects (Fig. 8B).
Animal models of autoimmune hepatitis expressing a foreign MHC-I molecule2–4 preclude the analysis of the contribution of professional APCs. Likewise, investigation of the role of antigen-specific CD4 T-cells is not possible. To overcome these hurdles, we generated transgenic mice expressing ovalbumin in the liver. Thus, T-cell epitopes are displayed on target cells and on professional APCs that acquire the antigen via apoptosis of dying target cells, allowing for the analysis of the site of activation and of the role of CD8 and CD4 T-cells.
In agreement with earlier studies, presence of the antigen on hepatocytes led to trapping of antigen-specific CD8 T-cells in the liver,19, 27 whereas the presence of the antigen on cholangiocytes had a much weaker effect. This result is not surprising, given the large number of possible contacts between T-cells and hepatocytes and the much more restricted access of T-cells to cholangiocytes. Apparently, the trapping of antigen-specific T-cells in the liver was so efficient that no migration to the liver-draining lymph node was observed in TF-OVA mice, although the antigen likely is presented by dendritic cells in the lymph node and spleen. In ASBT-OVA mice, in which CD8 T-cells were trapped less efficiently, activation in the draining lymph node was observed. Activation of CD8 T-cells was observed in the livers of both mouse lines, although with different kinetics. In TF-OVA mice, T-cells were primed simultaneously and OT-I T-cells divided in a synchronized fashion whereas priming in ASBT-OVA mice was more prolonged with several discernible generations of dividing T-cells. This observation again argues for a more restricted access of T-cells to the antigen expressed on cholangiocytes or APCs that sample the bile ducts. T-cell numbers in the liver declined at day 6, likely because of apoptosis of activated cells. Redistribution of CD8+Vα2+ T-cells also occurred and likely led to the increased numbers in the spleen. Increased numbers in the liver-draining lymph node may reflect influx from the liver or accumulation of locally proliferating cells.
Our results may be explained by priming of CD8 T-cells by professional APCs within the liver, by APCs in spleen or draining lymph node, or by the target cells, that is, hepatocytes or cholangiocytes. In ASBT-OVA mice, priming by professional APCs does occur as observed in the liver-draining lymph node. Priming in the liver-draining lymph node and redistribution of activated CD8 T-cells to the liver as observed by Bowen et al.28 seems unlikely to be the mechanism leading to liver injury here, because CD8 T-cells in the liver-draining lymph node proliferated much more slowly than those within the liver. Our data demonstrating activation of CD8 T-cells primed within the liver are in keeping with earlier results.7, 27 The advance of our approach is the possibility to determine the subset of APCs responsible for the observed effects. Although hepatocytes13, 27 and APCs in the lymph nodes28 were reported to activate CD8 T-cells, our data suggest that priming occurs within the liver and that professional bone marrow–derived APCs are required, at least in TF-OVA mice. Although whether OT-I cells are additionally primed in the spleen is uncertain, we unambiguously demonstrate that the spleen is dispensable for activation by a liver-restricted antigen. Given the higher ALT levels observed in splenectomized TF-OVA mice, the spleen may attenuate T-cell responses in this model.
As expected by the restricted expression of the antigen within the liver, patterns of inflammation caused by infiltrating T-cells were quite different in the 2 models: whereas a lobular infiltrate was observed in TF-OVA mice, infiltrates were restricted to the portal tracts harboring medium-sized bile ducts in ASBT-OVA mice. We did not observe increased alkaline phosphatase levels (data not shown) but found a mild increase in ALT levels in ASBT-OVA mice. Because the inflammatory process is transient, development of signs of cholestasis including the rise of alkaline phosphatase levels should not be expected. The rise in ALT levels in ASBT-OVA mice may be caused by cytokines produced by inflammatory cells in terms of a bystander hepatitis29 rather than by direct cytolytic effects.
As in any transgenic model, expression levels of the transgene depend on the random location of insertion and on the number of inserted copies. Therefore, differences in expression levels between the 2 lines may account to some extent for the observed effects. However, histological expression patterns are clearly distinct between both lines, establishing them as useful models for the investigation of T-cell responses to antigens expressed in the liver parenchyma or the bile ducts, respectively.
The hepatitis or cholangitis induced by transfer of transgenic T-cells in our model was transient, in keeping with earlier results. The transient nature of the inflammatory response may be indicative of incomplete activation of T-cells as proposed by Bertolino et al.13 Alternatively, the contraction of the immune response may involve the generation of a regulatory T-cell response or the induction of activation-induced cell death of fully activated T-cells.
The results obtained in our model with CD4 T-cells are strikingly different from those obtained with CD8 T-cells. Activation of CD4 T-cells requires the presence of MHC-II expressing APCs. Although present in the liver in sufficient numbers, no activation of CD4 T-cells in the liver was detected, indicating that either the amount of antigen presented locally or the activation status of liver-resident APCs precludes efficient antigen presentation. In contrast, professional APCs in draining lymph node and spleen activated CD4 T-cells in TF-OVA mice. Because most naïve T-cells enter lymph nodes through high endothelial venules rather than afferent lymph vessels,30 T-cells activated in the liver draining lymph node need not have traveled through the liver before encountering their antigen. Expression of the transgene was absent from the spleen. Therefore, antigen must have been taken up by APCs in the liver and carried to the spleen, where antigen presentation takes place. Even after their activation in vivo, CD4 T-cells did not cause hepatitis in the absence of CD8 T-cells. No priming of CD4 T-cells was observed in ASBT-OVA mice, indicating that an antigen restricted to cholangiocytes does not gain access to APCs in sufficient amounts to stimulate naïve CD4 T-cells when presented in a noninflammatory setting.
In the OVA-BIL model reported by Buxbaum et al.,10 more pronounced effects were observed than in our ASBT-OVA model. The difference is likely explained by the fact that nonpurified splenocytes were used, which contain a large number of other immune cells and a higher proportion of activated/memory T-cells. Also, the intraperitoneal route of administration may predispose T-cells to bystander activation and to preferential homing to the liver.
In summary, we established animal models that allow the investigation of antigen processing and antigen presentation in the liver in vivo. Although our models serve well for this purpose, they do not resemble the chronic nature of human autoimmune liver disease. Our data have important implications for the understanding of the role of T-cell subsets in the pathogenesis of autoimmune liver disease. They strengthen the notion that CD8 T-cells are activated locally within the liver by antigens expressed under noninflammatory conditions and are indispensable for the triggering of autoimmune processes in the liver. Conversely, they suggest that CD4 T-cells are much less prone to activation by liver antigens under noninflammatory conditions and display no autonomous cytotoxic effect to the liver.
The authors thank Sandra Vierich and Simone Spieckermann for excellent technical assistance. We thank Alf Hamann for helpful discussions, Thomas Blankenstein for the donation of OT-I RAG−/− mice, and Francis R. Carbone for the donation of the RIP-OVA construct.