|Home | About | Journals | Submit | Contact Us | Français|
TRAF6, an E3 ubiquitin protein ligase, plays a critical role in T cell tolerance by regulating medullary thymic epithelial cell (mTEC) development. mTECs regulate T cell tolerance by ectopically expressing self-antigens and eliminating autoreactive T cells in the thymus. Here we show that mice with mTEC depletion due to conditional deletion of Traf6 expression in murine thymic epithelial cells (Traf6ΔTEC mice) showed a surprisingly narrow spectrum of autoimmunity affecting the liver. The liver inflammation in Traf6ΔTEC mice exhibited all the histological and immunological characteristics of human autoimmune hepatitis (AIH). The role of T cells in AIH establishment was supported by intrahepatic T cell population changes and AIH development after transfer of liver T cells into immunodeficient mice. Despite a 50% reduction in natural Treg thymic output, peripheral tolerance in Traf6ΔTEC mice was normal, whereas compensatory T regulatory mechanisms were evident in the liver of these animals. These data indicate that mTECs exert a cell-autonomous role in central T cell tolerance and organ-specific autoimmunity, but play a redundant role in peripheral tolerance. These findings also demonstrate that Traf6ΔTEC mice are a relevant model with which to study the pathophysiology of AIH, as well as autoantigen-specific T cell responses and regulatory mechanisms underlying this disease.
T cell development and establishment of tolerance to self-antigens occurs in the thymus and depends on cross-talk between thymocytes and the thymic epithelium (1). This cellular compartment is composed of 2 major populations: cortical and medullary thymic epithelial cells (cTECs and mTECs, respectively). Whereas cTECs regulate positive selection of thymocytes and the diversity of the T cell repertoire, mTECs ectopically express tissue-restricted antigens (TRAs) for the purpose of deleting autoreactive T cells that recognize these antigens with high affinity (2–6). In addition, mTECs also affect peripheral tolerance through the production of natural Tregs (nTregs), which suppress aberrant T cell activation in peripheral tissues (7–9). The importance of mTECs in preventing autoimmunity has been revealed through the analysis of genetic mutations that control mTEC development and function in mice and humans (5, 10). In mutant mice, the absence or disruption of the 3-dimensional medullary architecture and defective TRA expression correlates with development of autoimmunity of varied severity, manifested as lymphocyte infiltrates within and autoantibody production against peripheral tissues (11–15). In humans, mutations in the AIRE protein (which controls TRA expression in mTECs) result in autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED; also known as autoimmune polyglandular syndrome type 1) (16, 17). This syndrome is characterized by chronic mucocutaneous candidiasis, hypoparathyroidism, and adrenal insufficiency as well as by a higher incidence of diabetes, inflammatory bowel disease, and autoimmune hepatitis (AIH). Thus, proper development of — and antigen presentation by — mTECs plays a key role in the establishment of central tolerance and prevention of autoimmunity.
Existing animal models with mutations that disrupt mTEC development and T cell selection have advanced our understanding of the role of mTECs in induction of central tolerance. However, as most of these mutations also affect the hematopoietic compartment, it is difficult to ascertain the individual contributions of the epithelial versus the hematopoietic compartments to abnormal T cell selection and development of autoimmunity. For example, while members of the TNF receptor superfamily and their corresponding signaling partners have been shown to inhibit TEC development (11, 12, 14, 15), these molecules are also expressed in hematopoietic cells, where they exert their effects in mediating immune responses. In this report, we describe the generation and characterization of conditional knockout mice with deletion of Traf6 specifically in TECs (Traf6ΔTEC mice). These mice developed a T cell–mediated inflammatory response against the liver that exhibited histopathological features of human AIH, representing a novel animal model for the study and treatment of this disease.
Conditional deletion of Traf6 in TECs was achieved by crossing floxed Traf6 mice (18) to animals expressing the Cre recombinase under the control of the foxn1 promoter (ref. 19 and Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI65414DS1). This gene encodes the transcription factor FOXN1, which regulates thymic epithelial cell development (20). Deletion of Traf6 expression was confirmed by staining thymic sections with an anti-TRAF6 antibody and the mTEC-specific marker lectin ulex europeous agglutinin–1 (UEA) (21), showing a drastic reduction of TRAF6+UEA+ mTECs in Traf6ΔTEC mice (Supplemental Figure 1B). H&E staining of thymic sections revealed diminished medullary areas in Traf6ΔTEC mice compared with WT controls (Supplemental Figure 1C). Further, staining of thymic sections with the cortical and medullary markers keratin-8 (K8) and K5, respectively, revealed a marked reduction in K5+ mTECs in Traf6ΔTEC mice (Supplemental Figure 2A). The diminished medulla and lack of K5+ mTECs also correlated with markedly reduced UEA+AIRE+ mature mTECs (Supplemental Figure 2A). Quantitative analysis of mTEC subpopulations by flow cytometry showed drastically decreased production of immature and mature mTECs (Supplemental Figure 2B), previously described as UEA+MHCIIint and UEA+MHCIIhi, or intermediate and mature MHCIIhiAIRE– and MHCIIhiAIRE+ cells (22–24). In contrast, cTECs identified as UEA–MHCII+ cells were unaffected by Traf6 deficiency (Supplemental Figure 2B), consistent with a specific role for TRAF6 in mTEC development. Depletion of mTECs in Traf6ΔTEC mice had no apparent effect on thymocyte development, as the total numbers of the different thymocyte populations were comparable to those in WT controls (Supplemental Figure 3A). In addition, mature T cells in the spleen and lymph nodes were unaffected in Traf6ΔTEC mice compared with controls (Supplemental Figure 3, B and C). These results showed that TRAF6 is required for, and plays an intrinsic role in, mTEC differentiation, whereas Traf6 deletion and mTEC depletion has no apparent effect on cTEC and T cell development.
Defects in mTEC development correlate with the presence of peripheral autoimmune manifestations, such as autoantibody production against and presence of inflammatory infiltrates in multiple organs (5, 10). While young Traf6ΔTEC mice (6–12 weeks old) appeared healthy, older Traf6ΔTEC mice (~6 months old) were smaller compared with WT controls and appeared hunched with ruffled fur (Figure (Figure1A).1A). Despite looking healthy, approximately 80% of young Traf6ΔTEC mice developed autoantibodies with reactivity against the liver, lung, kidney, and small and large intestines (Figure (Figure1B1B and Table Table1).1). Inflammatory infiltrates were also present, but exhibited a more narrow distribution, found predominantly in the liver (100%), and to a lesser extent in the lung (50%) and kidney (33%), of Traf6ΔTEC mice (Figure (Figure1C1C and Table Table1).1). The small intestine and colon appeared normal, despite the presence of autoantibodies against these tissues in the majority of Traf6ΔTEC mice (Figure (Figure1C1C and Table Table1).1). In older Traf6ΔTEC animals (6–9 months), the penetrance and distribution of the infiltrates and autoantibodies was similar to that of younger mice, although infiltrates and autoantibodies were also found in some WT mice analyzed (Table (Table1).1). Together, these results suggest that the incidence and distribution of autoimmune manifestations in Traf6ΔTEC mice appears relatively constant over time, but the phenotype may be compounded as the mice age, which is due to age rather than mTEC depletion–related effects.
Because 100% of the Traf6ΔTEC animals examined exhibited liver infiltrates (Table (Table1),1), we targeted this organ for further analysis. The hepatic inflammation in older Traf6ΔTEC animals was diagnosed as AIH based on several histopathological features typically observed in humans with AIH, including portal inflammation (defined by the presence of inflammatory infiltrates surrounding hepatic portal tracts), interface hepatitis (characterized by infiltrates crossing the limiting plate of inflammation and invading the liver parenchyma), and lobular inflammation (consisting of infiltrates within the liver parenchyma) (Figure (Figure2,2, A–C). The consequences of chronic hepatic inflammation and tissue damage in Traf6ΔTEC mice were reflected by increased collagen deposition and fibrosis surrounding the hepatic vessels and elevated serum alanine aminotransferase (ALT) levels, respectively (Figure (Figure2D).2D). Collagen deposition and fibrosis was quantified by morphometry of picrosirius red–stained, paraffin-embedded liver sections (see Methods and Figure Figure2D).2D). Analysis of young mice also showed significant differences in AIH histopathology between Traf6ΔTEC and WT mice; however, the scores were lower than those of older animals, in which AIH histopathological features peaked at 6 months of age (Supplemental Figure 4).
In addition to the histological manifestations discussed above, development of AIH in humans correlates with increased presence of plasma cells in intrahepatic inflammatory lesions as well as elevated total serum Ig levels and organ-specific and non–organ-specific autoantibodies (25–27). To further characterize the AIH phenotype of Traf6ΔTEC mice, we assayed for the presence of plasma cells in the liver and the nature of autoantibodies in the serum of these mice. Staining of paraffin-embedded liver sections from Traf6ΔTEC and WT mice with the plasma cell–specific dye methyl green-pyronin revealed an increased presence of plasma cells in the portal infiltrates of Traf6ΔTEC livers (Figure (Figure3A).3A). A marked increase in the number of CD138+ plasma cells in Traf6ΔTEC mice was also found when intrahepatic lymphocyte populations were analyzed by flow cytometry (Figure (Figure3B).3B). This increase in plasma cells, in turn, correlated with ANA presence in Traf6ΔTEC mice, as assayed by staining HEp-2 cells with sera from Traf6ΔTEC or control animals and determining fluorescence intensity based on arbitrary scores of 1–5 (Figure (Figure3C).3C). The majority of Traf6ΔTEC mice (75%) stained positive for ANAs, whereas 20% had autoantibodies against perinuclear proteins and 5% had autoantibodies against both nuclear and cytoplasmic antigens (Figure (Figure33D).
The presence of autoantibodies in Traf6ΔTEC mice correlated with a significant increase in total serum Ig and, more specifically, IgG1 and IgM isotype antibody levels (Figure (Figure4,4, A and B). Western blot analysis of liver extracts showed that autoantibodies in the sera of different Traf6ΔTEC animals exhibited similar binding specificities against several liver proteins, whereas sera from WT mice showed limited reactivity (Figure (Figure4C).4C). Both non–organ-specific and organ-specific autoantibodies were present in the sera of Traf6ΔTEC mice: some protein bands were present in all tissue extracts, whereas others were unique to the liver (Figure (Figure4D).4D). Together, the presence of plasma cells in the hepatic infiltrates, the increased level of total serum Ig, the presence of ANA, and the organ-specific and non–organ-specific autoantibodies in Traf6ΔTEC animals are consistent with the classic features of AIH described in human patients (25–27).
Based on the nature of the autoantibodies present, 2 types of AIH have been identified: type 1, characterized by ANA and/or anti-SMA antibodies; and type 2, characterized by anti–liver and kidney microsomal type 1 (anti-LKM) or anti–liver cytosol type I (anti-LC1) antibodies (26, 28, 29). In addition, anti–soluble liver antigen (anti-SLA) antibodies serve as highly specific serological markers for AIH and exhibit 99% specificity for diagnosing this disease (30–33). Anti-SLA antibodies have been found in 11%–22% of white North American and northern European adult patients with type 1 AIH and are common (18%–44% incidence) in children with type 1 and type 2 AIH (30, 34–36). To further characterize the nature of the autoantibodies in Traf6ΔTEC mice, we tested for the presence of anti-SLA, anti-SMA, and anti-LKM autoantibodies. While there was no difference in the levels of anti-SMA and anti-LKM antibodies between Traf6ΔTEC and control mice, levels of anti-SLA autoantibodies were significantly increased in Traf6ΔTEC mice compared with controls (Figure (Figure4E).4E). We therefore concluded that the presence of ANA and anti-SLA antibodies together in the serum of Traf6ΔTEC mice support an AIH phenotype in these animals similar to human type 1 AIH.
Since mTECs regulate negative selection of T cells in the thymic medulla (3, 4, 37), and given the observed inflammatory lesions in the liver of Traf6ΔTEC mice, we examined whether mTEC depletion correlates with abnormal T cell presence and function in these animals. Costaining frozen liver sections with anti-CD3 and anti-CD4 or anti-CD8 antibodies revealed that the portal infiltrates primarily consisted of CD3+CD4+ T cells, whereas CD3+CD8+ cells were more prevalent within the liver parenchyma (Figure (Figure5A5A and data not shown). Further analysis of resident liver T cell populations by flow cytometry confirmed an increase in the total number of CD4+ and CD8+ T cells in Traf6ΔTEC mice compared with WT controls (Figure (Figure5B).5B). In contrast to the liver, we found no significant quantitative or qualitative differences in T cell populations in the spleen and lymph nodes of Traf6ΔTEC and control mice (Supplemental Figure 3, B and C, and Supplemental Figure 5B), suggestive of organ-specific, rather than systemic, autoimmunity in Traf6ΔTEC animals. Furthermore, analysis of T cells subsets in tissues such as the lung and kidney showed that T cell accumulation was restricted to the liver of Traf6ΔTEC mice, as T cell numbers in these tissues were similar to those of controls (Supplemental Figure 5, A, C, and D). In addition to increased numbers of CD4+ and CD8+ T cell subsets, an aberrant CD4+CD8+ double-positive population of T cells was markedly increased in the liver of Traf6ΔTEC mice, but not in other tissues (Figure (Figure5B5B and Supplemental Figure 5). CD4+CD8+ cells have been identified in peripheral blood of patients with autoimmune diseases and are thought to have escaped thymic selection (38, 39). Therefore, it is possible that the presence of such cells in the liver of Traf6ΔTEC mice corresponds to abnormal thymic selection of T cells due to the observed mTEC depletion. Alternatively, as CD4+CD8+ T cells have been shown to represent peripherally hyperactivated T cells in animals and humans (40, 41), their presence in the liver may result from the inflammation induced in this organ by mTEC depletion.
Further analysis of Traf6ΔTEC and WT liver T cells revealed that CD44 and CD69 expression increased 3-fold in the Traf6ΔTEC CD4+ T cell subset and 6-fold in the Traf6ΔTEC CD8+ T cell subset compared with controls (Supplemental Figure 6, A and B), suggesting an activated phenotype for these cells. In contrast, and although activated T cells subsets were found in the spleen and, to a lesser extent, the lung and kidney of Traf6ΔTEC mice, the increase was <1-fold compared with WT controls (Supplemental Figure 6, A and B). These results were consistent with the low prevalence and distribution of infiltrates in the lung and kidney of Traf6ΔTEC animals compared with widespread infiltration in the liver. These data also suggest targeted homing of T cells to the liver of Traf6ΔTEC mice, as opposed to sequestration of activated T cells from the periphery. In support of targeted sequestration, perfusion of the liver prior to T cell isolation and the finding that T cell subsets in the blood of Traf6ΔTEC mice were composed entirely of naive (CD62+CD44–) T cells (Supplemental Figure 6C) argue against recruitment of activated T cells into the liver from circulating blood.
Although the trigger for human AIH is currently unknown, existing evidence suggests that disrupted immune homeostasis plays a permissive role in disease development. Liver damage is thought to be mediated by autoreactive CD4+ T cells that recognize and respond to liver-specific peptide(s) (26, 42). Given that mTECs regulate elimination of autoreactive T cells in the thymus and the presence of activated T cells in the liver of Traf6ΔTEC mice, we examined whether liver T cells respond to liver APCs in syngeneic mixed lymphocyte reactions. We found significantly increased proliferation of Traf6ΔTEC liver T cells compared with controls in response to syngeneic APC stimulation (Figure (Figure5C);5C); in contrast, these cells responded to the same extent when stimulated with allogeneic APCs or anti-CD3/CD28 antibodies (Figure (Figure5C5C and data not shown), which suggests that there were no inherent defects in the ability of WT cells to proliferate. In addition, consistent with normal peripheral responses in Traf6ΔTEC animals, WT and Traf6ΔTEC splenic T cells responded similarly to syngeneic or allogeneic stimulation (Supplemental Figure 7A). The specificity of liver Traf6ΔTEC T cells for liver APC/antigen complexes was further assessed by incubating liver T cells from WT and Traf6ΔTEC animals with syngeneic APCs isolated from different organs. Whereas Traf6ΔTEC liver T cells responded robustly to liver APCs compared with controls, these cells failed to proliferate when stimulated with APCs isolated from the lung or kidney of the same animals and responded similarly to WT T cells when stimulated with splenic APCs (Figure (Figure5D).5D). Furthermore, when the specificity of Traf6ΔTEC liver T cells for liver antigens was tested using in vitro–differentiated APCs loaded with cell extracts from different tissues, we found a significant increase in Traf6ΔTEC liver T cell proliferation in response to stimulation with liver extracts, but not lung or kidney extracts (Figure (Figure5E).5E). The proliferation of Traf6ΔTEC liver T cells under these conditions exhibited longer kinetics as well as a lower level of activation than those observed with liver-derived APCs (Figure (Figure5D),5D), likely reflecting the rarity of autoreactive T cells responding to liver antigens. In contrast, similar to the results with splenic APCs (Figure (Figure5D),5D), there was no significant difference in the proliferation of WT versus Traf6ΔTEC liver T cells when these were stimulated with unloaded APCs (Figure (Figure5E),5E), which suggests that both types of cells respond to endogenous antigens to a similar extent. Collectively, these results suggest that mTEC depletion in the thymus results in the production of autoreactive T cell clones that home predominantly to the liver of Traf6ΔTEC mice, where they bind and respond to liver antigens and promote the development of inflammatory lesions and AIH.
As mentioned above, liver damage associating with AIH development is thought to be mediated by autoreactive CD4+ T cells that recognize and respond to liver-specific peptides. Once activated, Th0 cells differentiate into Th1 cells secreting IFN-γ, an important intermediary in tissue destruction, and/or into Th2 cells, which produce IL-4, IL-10, and IL-13 cytokines and promote antibody production by plasma cells (26, 42). Consistent with this model, we found significantly increased numbers of IFN-γ– and IL-4–producing cells and, to a lesser extent, IL-17–producing cells in the liver of Traf6ΔTEC versus WT mice (Figure (Figure5F5F and data not shown), whereas normal numbers of cells producing these cytokines were found in the spleens of the same animals (Supplemental Figure 7B). Similarly, we found no difference in serum Th1, Th2, Th17, or inflammatory cytokine levels between Traf6ΔTEC and control mice (Supplemental Figure 7C), consistent with normal systemic immune responses in Traf6ΔTEC animals. Thus, the T cell–containing infiltrates in the liver of Traf6ΔTEC mice exhibit an effector T cell phenotype that recapitulates that observed in human AIH and may be driving chronic liver damage and fibrosis in these mice.
One hallmark of autoimmune disease is the ability of immune cells to adoptively transfer the disease to another host. To determine whether the T cells present in the liver of Traf6ΔTEC animals were able to promote AIH development, sorted liver CD3+ T cells were adoptively transferred into Rag1–/– recipient mice. Rag1–/– recipients of Traf6ΔTEC liver T cells developed AIH of increased severity compared with the donor mice (Supplemental Figure 8) and with the mice analyzed in Figure Figure2.2. Portal and lobular inflammation and interface hepatitis were all significantly increased in Rag1–/– recipients of Traf6ΔTEC versus WT liver T cells and compared with donor mice (Figure (Figure6,6, A and B, and Supplemental Figure 8, A and B). The hepatic inflammation correlated with significantly increased levels of ALT in recipients of Traf6ΔTEC liver T cells compared with WT controls and donor mice (Figure (Figure6C6C and Supplemental Figure 8C). Liver fibrosis was also significantly increased in Rag1–/– recipients of Traf6ΔTEC liver T cells, with evident tissue damage shown as hepatocyte dropout areas around fibrotic vessels (Figure (Figure6,6, D and E). The fibrosis was more severe than in Traf6ΔTEC mice (compare Figure Figure6D6D and Figure Figure2D),2D), whereas minimal fibrosis was observed in Rag1–/– mice receiving WT liver T cells (Figure (Figure6,6, D and E). Some Rag1–/– recipient animals developed colitis regardless of the nature of the donor cells (data not shown), which suggests that liver T cells exhibit colitogenic properties. In contrast, transfer of splenic T cells or total splenocytes isolated from the same mice failed to promote disease development in Rag1–/– recipients (data not shown). Together, these data suggest that Traf6 deletion in TECs and mTEC depletion associates with organ-specific rather than systemic autoimmunity and that production of aberrant liver T cell populations is responsible for AIH development in Traf6ΔTEC mice.
To further determine whether a specific T cell subpopulation is responsible for AIH transfer in Rag1–/– recipients, we performed adoptive transfers of sorted T cell subsets from WT and Traf6ΔTEC mice. Total CD3+ T cell transfers from different mice were included in these experiments as positive controls (Figure (Figure7,7, A and B). Transfer of Traf6ΔTEC CD4+ T cells led to accumulation of infiltrates in hepatic portal tracks in the majority of recipient Rag1–/– mice (Figure (Figure7A).7A). However, the histopathological scores of AIH in these recipients were not significantly different from those of mice adoptively transferred with WT cells (Figure (Figure7C).7C). On the other hand, transfer of CD8+ cells sorted from the same mice failed to promote infiltrate formation or significant differences in AIH scores in the majority of the recipient animals (Figure (Figure7,7, A and D). Similar to CD4+ T cells, transfer of serum from Traf6ΔTEC and WT mice into Rag1–/– recipients led to some infiltrate formation, but again, no significant differences were observed in the histopathological AIH scores of the different recipients (Figure (Figure7,7, A and E). In these experiments, we also attempted to transfer CD4+CD8+ T cells; however, we failed to purify enough cells from several mice, as these cells were exceedingly rare, especially in WT mice. Therefore, the ability of the CD4+CD8+ cells to transfer the disease could not be assessed. Finally, analysis of ALT levels of the different recipient mice was consistent with their histopathological scores: increased ALT levels were observed only in recipients of total Traf6ΔTEC CD3+ T cells, not in recipients of T cell subsets (Figure (Figure7F).7F). These results suggest that although liver T cells are important for disease development in Traf6ΔTEC mice, a complex interplay among different T cell subsets, and likely B cells, is required for disease development. Thus, CD4+ T cells may be responsible for initiating the inflammatory response in an antigen-specific manner, whereas other T cell subsets and/or B cells play a secondary/complimentary role in disease development.
In addition to regulating the negative selection of autoreactive T cells, mTECs have also been shown to regulate production of nTregs in the thymus (7–9). Because a reduction in the numbers and/or functional impairment of Tregs is thought to correlate with AIH development (43, 44) and because Tregs are potent suppressors of inflammation and autoimmunity, we next examined whether Treg production and function is impaired in Traf6ΔTEC mice. Staining of thymocyte cell suspensions with the Treg-specific marker FOXP3 revealed a 50% decrease in the total number of FOXP3+ thymocytes (Figure (Figure8A),8A), consistent with a role for mTECs in nTreg production. However, there was no difference in the number of splenic FOXP3+ T cells between Traf6ΔTEC and WT mice, and Traf6ΔTEC CD25+FOXP3+ splenic T cells were equally able to suppress T effector function in vitro compared with WT controls (Supplemental Figure 9, A and B). On the other hand, we found a marked increase in the number of FOXP3+ T cells in Traf6ΔTEC versus WT liver (Figure (Figure8B).8B). The increase in cell numbers correlated with increased production of induced CD25+FOXP3+ Tregs (iTregs) when Traf6ΔTEC CD4+CD25– liver T cells were incubated with TGF-β in vitro (Figure (Figure8C).8C). In contrast, and consistent with the normal numbers of CD25+FOXP3+ Tregs in the spleen of Traf6ΔTEC mice (Supplemental Figure 9A), we observed no difference in the percentage of iTregs when splenic Traf6ΔTEC or WT CD4+CD25– T cells were similarly stimulated in vitro (Figure (Figure8C).8C). Consistent with the in vitro production of iTregs, we found significantly increased protein and RNA levels of TGF-β and IL-10 in liver extracts from Traf6ΔTEC animals compared with controls (Figure (Figure8,8, D and E). Thus, despite the decreased thymic output of nTregs, peripheral mechanisms such as homeostatic expansion of nTregs in the spleen and/or inflammation-driven production of iTregs in the liver of Traf6ΔTEC mice may compensate for the reduced nTreg thymic output in these animals.
In this report, we examined the role of impaired mTEC development on T cell–mediated autoimmunity within the context of a normal hematopoietic cell compartment, which provided a unique opportunity to decipher the mechanisms of mTEC differentiation and central tolerance. Peripheral T cell subsets and T cell immune responses were normal in Traf6ΔTEC mice, supporting organ-specific (rather than systemic) development of autoimmunity. Consistent with this, and although the majority of Traf6ΔTEC mice developed autoantibodies against most of the tissues tested, the distribution of inflammatory lesions was largely restricted to the liver — and, to a lesser extent, the lung and kidney — of these mice. The reduced prevalence or absence of infiltrates in nonhepatic tissues was not due to different kinetics of autoimmunity development, as the incidence of inflammatory infiltrates in the different organs was similar between young and older animal cohorts. In addition, the reduction or lack of T cell–containing infiltrates in tissues other than the liver was not due to altered (patchy or diffuse) distribution of immune cells, as quantification of total T cells from the spleen, lung, and kidney of several mice revealed similar numbers of T cells between WT and Traf6ΔTEC animals. This was in contrast to the liver, where we found large increases in the total numbers of both CD4+ and CD8+ T cells. Given the presence of autoantibodies in the serum and inflammatory infiltrates in the lung and kidney of some Traf6ΔTEC mice, we currently cannot exclude the possibility that these effects also contribute to the overall disease phenotype in Traf6ΔTEC animals. Nonetheless, our results showing (a) complete penetrance of liver inflammation in all Traf6ΔTEC animals tested; (b) increased numbers of total and activated/autoreactive T cells in Traf6ΔTEC liver compared with other tissues; (c) stimulation of Traf6ΔTEC liver T cell proliferation by liver, but not by other APCs, and by APCs loaded with liver cell extracts, but not lung and kidney extracts, and (d) the presence of anti-SLA and other liver-specific autoantibodies in the serum of Traf6ΔTEC mice suggest that the liver is targeted in these mice by autoreactive T cells that escape deletion in the thymus due to mTEC depletion.
It is currently unclear why the liver is the organ predominantly affected in Traf6ΔTEC mice, as expression of all peripheral antigens in mTECs should be similarly affected by mTEC depletion. As different TRAs are stochastically expressed in mTECs (45), it is possible that autoimmune manifestations vary in a manner dependent on the genetic background of the animals, or the nature of the genetic mutation resulting in mTEC depletion. In terms of genetic background, whereas inflammation was mostly contained within the liver in C57BL/6 mice, we observed a wider range of phenotypes affecting different tissues, including the skin, eye, joints, and muscle, in a mixed (129/SvJ/C57BL/6) genetic background. It is possible that mTEC depletion by Traf6 deletion in a more permissive, autoimmune-prone background may reveal a wider spectrum of autoimmune manifestations than in the current mouse model. In terms of the genetic mutations used to deplete mTECs, the phenotype of Traf6ΔTEC mice was in sharp contrast to previous experiments with straight Traf6 knockout mice and other models with defective mTEC development. In Traf6ΔTEC mice, TEC-specific deletion of Traf6 correlates with chronic inflammation, and despite looking unhealthy, the mice survive for at least 1 year after birth. In contrast, in straight knockout mouse models with mutations affecting proteins that regulate the noncanonical NF-κB pathway and mTEC development (including TRAF6), mTEC depletion in homozygous mice correlated not only with peripheral autoimmune manifestations, but also early lethality (12). In those animal models, the hematopoietic compartment was also shown to contribute to autoimmunity development and reduced survival, as the affected proteins were expressed in both the epithelial and the hematopoietic compartments. Autoimmune manifestations in these mice were more pronounced and led to early death due to systemic effects and lymphoproliferation, effects that are absent in Traf6ΔTEC mice, in which the hematopoietic compartment is intact. Consistent with this, T cell–specific deletion of Traf6 (Traf6ΔT mice) led to multiorgan inflammatory disease and lymphadenopathy due to defective Treg-mediated suppression of Traf6ΔT T cell responses (46). Despite the differences in survival between existing models and our Traf6ΔTEC mouse, liver inflammation has been found in all animals with mTEC defects; however, a connection to AIH had not previously been established. Therefore, the extended lifespan of the Traf6ΔTEC mice, together with the chronic, organ-specific (instead of systemic) nature of the inflammation, makes these animals an attractive and convenient model with which to study AIH.
The chronic nature of liver inflammation in Traf6ΔTEC mice correlated with increased presence of FOXP3+ T cells in the liver of these animals. Therefore, despite the reduced thymic nTreg output in Traf6ΔTEC animals, the normal presence and function of the peripheral Treg compartment, the increased frequency of iTregs, and the increased levels of TGF-β and the immunosuppressive cytokine IL-10 in the liver of Traf6ΔTEC mice may all be responsible for their lack of overt peripheral T cell responses and the chronic nature of their hepatic inflammation. A correlation between reduced Treg numbers and/or functional impairment of Tregs and development of AIH has been observed in human patients (26). However, a recent report showed that the numbers and suppressive function of FOXP3+ T cells isolated from AIH patients were similar to those of normal controls. In addition, the intrahepatic frequency of these cells was higher and correlated with the degree of inflammation in the liver of AIH patients (47). Our results support the latter observations and suggest that in Traf6ΔTEC mice, inflammation-driven production of iTregs in the liver may compensate for the reduced nTreg thymic output and drive the chronic nature of the disease. Consistent with this, existing evidence supports a protective role for iTregs in established or induced autoimmunity in several mouse models (48, 49).
The liver inflammation and AIH development in Traf6ΔTEC mice is quite intriguing, given that the liver is considered a privileged organ from an immunological standpoint. It is well known that transplanted livers are less frequently rejected than other solid organ transplants (50–53). In addition, liver autoimmune diseases like AIH develop with lower frequency compared with other autoimmune conditions, for example, type 1 diabetes and multiple sclerosis (54). Although the causes of AIH remain unknown, current models of AIH development propose that hepatitis development and ensuing tissue damage is orchestrated by naive CD4+ T lymphocytes recognizing liver self-peptides. Following antigen binding, activated CD4+ Th0 cells differentiate into IFN-γ–secreting Th1 cells that activate macrophages and increase MHCI expression on hepatocytes, rendering liver cells susceptible to CD8+ T cell–mediated cytotoxicity. Th0 cells can also differentiate into Th2 effector cells that secrete IL-4, IL-10, and IL-13 cytokines, favoring autoantibody production by plasma cells (26, 42). The major components of this model were recapitulated in the Traf6ΔTEC mice, shown by increased total numbers of CD4+ and CD8+ T cells, increased production of both Th1 and Th2 cytokines, and increased presence of plasma cells and autoantibody production. In addition, our results showing that individual T cell subsets failed to recapitulate AIH development in recipient mice further supported the idea that complex interplay among lymphocyte cell subsets is required for disease development, with self-antigen recognition by autoreactive CD4+ T cells playing a key role in disease initiation. Consistent with this, susceptibility of human patients to AIH has been linked to deviant DRB1 alleles within the human leukocyte antigen (HLA) region (55–57). As these molecules present peptide antigens to CD4+ T cells, compromised antigen presentation by susceptibility alleles and recognition of liver self-antigens by naive T lymphocytes is thought to contribute to AIH pathogenesis. A similar mechanism may be in effect in Traf6ΔTEC mice, as antigen presentation was indeed compromised by mTEC depletion in these animals. In addition, our observation of anti-SLA autoantibodies present in the sera of Traf6ΔTEC mice supports faulty antigen presentation in these mice, as anti-SLA autoantibody production has been linked to HLA DRB1*0301 alleles (33, 58), which, together with DRB1*0401 alleles, are the principal genetic risk factors for development of type 1 AIH in certain patients (30, 34–36).
Finally, impaired antigen presentation due to mTEC depletion and elimination of autoreactive T cells in Traf6ΔTEC mice may mimic or replicate the conditions that promote pathogenesis of AIH in humans in different disease settings. The observation that 10%–15% of patients with APECED (the result of impaired AIRE function and antigen expression in mTECs) develop AIH (59, 60) provides further support for impaired antigen presentation and mTEC function in the pathogenesis of AIH. Therefore, despite the existence of several mouse models of AIH involving chemically induced or liver-intrinsic perturbations of hepatic immune responses (25, 54, 61), we believe that the Traf6ΔTEC mouse represents a relevant model by which to study the pathogenic mechanisms leading to development of this disease and explore new therapeutic approaches. Further characterization of the Traf6ΔTEC mouse model can not only improve our understanding of the pathophysiology of AIH, but may also lead to a broader understanding of how liver tolerance is established and maintained, with wider impact on other liver autoimmune diseases and liver transplantation.
Disruption of Traf6 expression in TECs was achieved by crossing floxed Traf6 mice (18) to animals in which a cDNA encoding for the Cre recombinase was knocked into the 3′ untranslated region (3′UTR) of the foxn1 locus (19). The efficiency of deletion through foxn1-driven, Cre-mediated recombination in both cTECs and mTECs was previously shown to be approximately 80% in the Rosa-26-YFP reporter mice (19). Traf6fl/fl/foxn1-Cre (Traf6ΔTEC) mice were backcrossed for 8 generations to the C57BL/6 background. Heterozygous mouse breeding allowed for the use of littermates as WT controls. Rag1–/– mice were obtained from the Jackson Laboratories.
Thymi from Traf6ΔTEC and WT mice were dispersed, and single–epithelial cell suspensions were enriched using a Percoll gradient as described previously (24). TEC fractions were treated with anti-CD16/CD32 blocking antibody (BD) and incubated with anti-CD45–PE (30-F11; eBioscience), anti-MHCII–allophycocyanin (M5/114.15.2; eBioscience), biotinylated UEA (Vector Laboratories), streptavidin-PerCP (BD), and/or AIRE-FITC (5H12; provided by H.S. Scott, Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia) antibodies. Multiparameter analysis of stained cell suspensions were performed on an LSR II flow cytometer (BD) at Icahn School of Medicine at Mount Sinai Flow Cytometry Shared Resource Facility and analyzed with FlowJo software (TreeStar Inc.).
For assessment of liver inflammation, the left liver lobe from Traf6ΔTEC and WT mice was excised, fixed in 10% formalin, and embedded in paraffin. Liver sections were stained with H&E. AIH was evaluated by a liver pathologist and scored using light microscopy and a modified Scheuer scoring scale (62, 63), assigning scores for portal inflammation (0, normal; 1, mild — sprinkling of cells in portal tracts; 2, moderate — lymphoid aggregates in some portal tracts; 3, severe — lymphoid aggregates in most portal tracts), interface hepatitis (0, none; 1, mild — focal, present in few portal tracts; 2, moderate — focal in most portal tracts or continuous in few portal tracts; 3, severe — continuous in most portal tracts), and lobular inflammation (0, none; 1, mild — scattered foci of lobular-infiltrating lymphocytes; 2, moderate — numerous foci of lobular-infiltrating lymphocytes; 3, severe — extensive pan–lobular-infiltrating lymphocytes).
ALT levels were detected by reflectance spectrophotometry using the VITROS ALT Slides and the VITROS Chemistry Products Calibrator Kit 3 on VITROS 250/350/950 and 5,1 FS Chemistry Systems and the VITROS 5600 Integrated System at Icahn School of Medicine at Mount Sinai Chemistry, ENDO and STAT Shared Resource Facility.
Paraffin-embedded liver sections were stained with picrosirius red to measure collagen content as described previously (64), and a Bioquant computerized morphometry program was used to quantify staining intensity. Assessment of collagen staining was obtained by birefringence microscopy using an Axioplan II microscope (Carl Zeiss Inc.) at Icahn School of Medicine at Mount Sinai Microscopy Shared Resource Facility. Each staging of fibrosis was assessed in 10–30 fields at ×100 magnification, and the average was calculated for each mouse.
For immunofluorescence staining, frozen thymic or liver sections were fixed in ice-cold acetone. Sections were incubated in 1% H2O2 in PBS for 5 minutes, blocked in 10% egg white and 0.5% BSA in PBS, washed in 0.05% Tween in PBS, and incubated in 2 μg/ml UEA conjugated to biotin (Vector Laboratories) or AIRE (M-300; Santa Cruz Biotechnology Inc.), K5 (AF138; Covance), K8 (TROMA-1; provided by Developmental Studies Hybridoma Bank, University of Iowa), TRAF6 (H-274; Santa Cruz Biotechnnology Inc.), CD3 (Dako Cytomation), CD4 (H129.19; BD), and CD8 (53-6.7; BD) antibodies for 2 hours at room temperature. Imaging was performed using anti-rat Alexa Fluor 488, anti-rabbit Alexa Fluor 488, or Alexa Fluor 594 (Invitrogen), and streptavidin-rhodamine (Jackson Immunoresearch Laboratories) secondary fluorescent antibodies. All images were acquired with an Axioplan II microscope (Carl Zeiss Inc).
Single-cell suspensions of thymi and peripheral lymph nodes were prepared by mechanical dispersion. Various tissues from perfused mice were homogenized, and lymphocyte populations were isolated on a 40%/70% Percoll (GE Healthcare) gradient as previously described (33). Cell suspensions were incubated with anti-CD16/CD32 blocking antibody (BD) for 10 minutes and stained with the corresponding antibody mixtures on ice. The following monoclonal antibodies were used (from eBioscience, unless otherwise indicated): CD138-biotin (281.2; BD); NK1.1-biotin (PK136; BD); CD25-FITC (7D4; BD); CD3ε-FITC (145-2C11); CD69-FITC (H1.2F3; BD); CD44-PE (IM7; BD); PE-streptavidin; CD4-PerCP (LT34); CD8a-PerCP (53-6.7; BD); CD4–PE-Cy7 (GK1.5); steptavidin–PE-Cy7; CD3ε–PerCP-Cy5.5 (145-2C11); CD4-allophycocyanin (L3T4; BD); CD62L-allophycocyanin (MEL-14); MHCII-allophycocyanin (M5/114/15.2); CD8a–Pacific Blue (53-6.7); CD45–Alexa Fluor 750 (30-F11); CD19–eFluor 450 (eBio1D3); FOXP3-PE (FJK-16s). Intracellular staining for FOXP3 was performed with the anti-mouse/rat FOXP3 staining antibody set (eBioscience) and cytofix/cytoperm kit (BD). Intracellular cytokine staining was performed with anti–IL-4–PE (BD), anti–IL-17A–PE-Cy7 (BD), and anti–IFN-γ–allophycocyanin (BD) after cells had been stimulated in vitro for 6 hours with Leukocyte Activation Cocktail and GolgiPlug (BD) according to the manufacturer’s instructions. Multiparameter analysis of stained cell suspensions was performed on an LSR II flow cytometer (BD) and analyzed with FlowJo software (TreeStar Inc.).
The presence of autoantibodies in the sera of Traf6ΔTEC and control mice were analyzed by incubating frozen sections of tissues derived from Rag1–/– mice with sera (1:40 dilution) from different animals, followed by FITC-labeled anti–mouse IgG secondary antibody and fluorescent microscopy. Serum ANAs were detected using HEp-2 cells fixed on glass slides (MBL BION) according to the manufacturer’s protocol. Anti-SMA, anti-LKM, and anti-SLA autoantibodies were analyzed in sera of Traf6ΔTEC and WT mice by mouse-specific ELISA kits (Glory Science Co.; BlueGene Biotech). All samples were run in duplicates. ELISA plates were read on a μQuant Microplate Reader (Bio-Tek Instruments Inc.). Plasma cells were detected by staining paraffin-embedded liver sections with methyl green-pyronin (Sigma-Aldrich). The severity of plasma cell infiltration was based on an estimate of the proportion of plasma cells in the inflammatory infiltrate, as follows: 0, none; 1, occasional plasma cells; 2, moderate plasma cells (<30%); 3, severe plasma cells (>30%, or occurring in sheets or clusters). All images were acquired with an Axioplan II microscope (Carl Zeiss Inc.).
Total Ig and serum Ig isotypes were determined with a commercial ELISA kit (SouthernBiotech) using 1:500 serum dilutions. All samples were run in triplicates. ELISA plates were read on a μQuant Microplate Reader (Bio-Tek Inc.). Serum cytokine levels were determined using a Cytometric Bead Array (CBA; BD) cytokine kit according to the manufacturer’s instructions. Samples were collected on an LSR Fortessa cytometer (BD) and analyzed with FCAP Array software (Soft Flow Inc.). TGF-β and IL-10 protein and RNA levels were quantified by ELISA and real-time quantitative PCR kits, respectively. All samples were run in triplicates. ELISA plates were read on a μQuant Microplate Reader (Bio-Tek Instruments Inc.). RNA was isolated with RNeasy Mini Kit (Qiagen) as described by the manufacturer. RNA (1 μg) was used to generate cDNA using a Clontech cDNA Synthesis Kit. mRNA was quantified with a LightCycler 480 thermocycler (Roche). Transcript abundance was normalized to that of mRNA encoding GAPDH. IL-10 and TGF-β were quantified in liver lysates with mouse-specific ELISA kits (BD; Roche).
Tissue samples were homogenized on ice in lysis buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, 1% NP-40, 10% glycerol, and a protease/phosphatase inhibitor cocktail (Roche). Lysates were clarified by centrifugation, and supernatants were mixed with SDS Laemmli buffer, separated by SDS-PAGE, and visualized by Western blot. Protein blots were blocked with 5% nonfat dried milk in PBS for 1 hour, washed with PBS with 0.05% Tween20, and incubated for 1 hour in a 1:100 dilution of sera from individual Traf6ΔTEC or WT mice. After washing with PBS with Tween20, blots were incubated with anti–mouse IgG–horseradish peroxidase (1:2,000; GE Healthcare) for 1 hour, and protein bands were revealed with a chemiluminescence reagent (Western Lightning; PerkinElmer).
Sorted CD4+CD8–CD25– T cells (5 × 104) were cultured with irradiated splenocytes (5 × 104), stimulated with 1 μg/ml plate-bound anti-CD3ε (eBioscience), and labeled with CFSE (5 μM; Molecular Probes) in the presence or absence of CD4+CD8–CD25+ Tregs at the indicated ratios. After 72 hours, cultured cells were stained with anti-CD4–PE-Cy7 (GK1.5; eBioscience) antibody, and proliferating CD4+CFSE+ T cells were assessed by flow cytometry. For induction of iTregs, sorted CD4+CD8–CD25– T cells (5 × 104) were cultured in the presence of anti-CD3 antibody (1 μg/ml), 10 ng/ml mouse recombinant IL-2 (BD), and 5 ng/ml human recombinant TGF-β (eBioscience) for 4 days. Analysis of iTregs was performed by staining with anti-CD4–PE-Cy7 (GK1.5; eBioscience), anti-CD25–FITC (7D4; BD), and anti-FOXP3–PE (FJK-16s; eBioscience) antibodies and flow cytometry. All samples were run in triplicates, collected on an LSR II flow cytometer (BD), and analyzed with FlowJo software (TreeStar Inc.).
Hematopoietic cells were isolated from various tissues of Traf6ΔTEC or WT mice using a Percoll gradient. Total cells were then incubated with CD11c-coupled MicroBeads (Miltenyi Biotec) and passed through magnetic columns for positive selection of APCs. The flow-through was collected, incubated with a biotin cocktail followed by anti-biotin-coupled MicroBeads (Miltenyi Biotec), and once again passed through magnetic columns for negative selection of T cells. Total CD4+ T cells were counted and stained with CFSE. Allogeneic or syngeneic liver and spleen, and syngeneic lung and kidney, CD11c+ cells were plated with CD4+ T cells (1:1 ratio) in a 96-well flat-bottom plate for 72 hours. DCs (1 × 105, unloaded or loaded with lysates) were cocultured with CD4+ T cells (1 × 105) in a U-bottom 96-well tissue plate for 5 days. 10 ng/ml mouse recombinant IL-2 (BD) was added to the culture after 3 days. All samples were run in triplicates. Proliferation of CD4+ T cells was analyzed by CFSE dilution on an LSR II flow cytometer (BD).
Bone marrow–derived DCs were generated after 5 days of culture with GM-CSF supernatant purified from J558 cells (ATCC) grown in RPMI1640 (Invitrogen) supplemented with 10% FBS (HyClone) and 1% antibiotic-antimycotic solution (Invitrogen) at 37°C in a 5% CO2 atmosphere.
Whole-cell extracts were resuspended in serum-free medium and subjected to 4 freeze-thaw cycles by alternating between dry ice and 37°C. The lysate was centrifuged for 10 minutes at 300 g to remove debris, and the supernatant was collected. The protein concentration of the lysate was determined by a commercial assay (Bio-Rad). The lysate was added to DCs at a final concentration of 100 μg protein/ml.
2.5 × 105 sorted liver or spleen NK1.1–CD3+ T cells, or liver NK1.1–CD3+CD4+ or NK1.1–CD3+CD8+ T cells, isolated from Traf6ΔTEC or control mice were intravenously injected into Rag1–/– recipient mice. Sera isolated from Traf6ΔTEC or WT mice were intravenously injected into Rag1–/– recipient mice. Mice were sacrificed approximately 6 weeks after adoptive transfer, and liver samples were fixed in 10% buffered formalin. Paraffin-embedded sections were stained with H&E and analyzed by light microscopy. Histopathological scoring was performed by a liver pathologist as described above.
Traf6ΔTEC and WT samples were compared by 2-tailed Student’s t test (Excel; Microsoft Inc.). A P value less than 0.05 was considered significant.
Animals were housed in specific pathogen–free conditions and were used and maintained in accordance with institutional guidelines. Animal protocols were approved by the IACUC of the Icahn School of Medicine at Mount Sinai.
The authors thank Andrea D. Branch and Arielle Klepper for protocols and advice, Scott L. Friedman and Andrea D. Branch for critically reading the manuscript, Nancy R. Manley for providing foxn1-Cre knockin mice, and Marie E. Grace for advice and technical support with serum biochemical analysis. A.J. Bonito was supported by Icahn School of Medicine at Mount Sinai’s Immunology Institute training T32 AI007605-09 grant and by R01 grant AI068963-01. C. Aloman was supported by grant 1K08DK088954. E.G. Weinstein was supported by R01 grants AI088106-01 and AI068963-01. This work was supported by NIAID grants R01 AI49387-01, R56 AI049387-05, and R01 AI068963-01.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2013;123(8):3510–3524. doi:10.1172/JCI65414.