Potential conflict of interest: Nothing to report.
Autoimmune hepatitis (AIH) is characterized by a loss of immunological tolerance to hepatocytes. Patients respond well to immunosuppression but progression to endstage liver disease occurs in 10%-20% of cases, leading to liver transplantation. Using a murine model of type 2 AIH, we identified susceptibility factors for autoimmune hepatitis and attempted to restore immunological tolerance to liver autoantigens. An increased ectopic expression of a liver autoantigen (FTCD) in the thymus leading to reduced numbers of circulating autoreactive T cells was sufficient to prevent development of AIH in mice. However, in the presence of a reduced central tolerance to FTCD, a strong regulatory T-cell response was able to inhibit proliferation of liver-specific autoreactive T cells and prevent AIH. Development of a severe AIH stemmed from reduced numbers of functional regulatory T cell (Tregs) leading to an increased proliferation of FTCD-specific autoreactive T and B cells. Adoptive transfer of ex vivo expanded CXCR3+ Tregs in mice with AIH efficiently targeted the inflamed liver, restored peripheral tolerance to FTCD, and induced remission of AIH. Conclusion: Peripheral tolerance to liver autoantigens in AIH is paramount. Autologous infusion of ex vivo expanded CXCR3+ Tregs in AIH patients could be an effective therapeutic approach to restore peripheral tolerance and induce remission of AIH. (HEPATOLOGY 2013)
Autoimmune hepatitis (AIH) is a disorder of unknown etiology characterized by an immune-mediated injury that gradually destroys the hepatic parenchyma. Most patients respond well to immunosuppressive therapy; however, progression and endstage liver disease occurs in 10%-20% of cases and liver transplantation may be necessary.1 The observation that patients with AIH have a significant decrease of circulating CD4+ regulatory T cells (Tregs) compared to healthy controls2-4 and that these cells could be expanded ex vivo to generate functional Tregs5, 6 has generated great interest in restoring immune tolerance to liver autoantigens through autologous regulatory T-cell infusion.7 However, progress has been hindered by the lack of experimental evidence showing that Treg infusion could effectively restore immunological tolerance in AIH.
In mice, DNA vaccination can break immune tolerance to a liver neo-self antigen8 and immunization with xenoantigens can, by molecular mimicry, trigger the development of AIH in C57BL/6 mice.9-11 Furthermore, DNA immunization against human type 2 AIH autoantigens: cytochrome P450 2D6 (CYP2D6) and formiminotransferase cyclodeaminase (FTCD), breaks the immune tolerance to the murine homologous autoantigens and induces an autoimmune inflammation of the liver.11 As in humans, mice display a Th1 phenotype of autoimmune response and show elevated serum alanine aminotransferase (ALT) levels with a characteristic liver inflammatory infiltrate composed of CD4+, CD8+, and B lymphocytes.11 They also show high immunoglobulin levels, anti-LKM1, and anti-LC1 autoantibodies, hallmarks of type 2 AIH.9, 11 As in humans, females are more susceptible to AIH10 and development of the disease is influenced by the genetic background.9
Using this model of type 2 AIH, we determined whether central or peripheral tolerance was the main factor for AIH susceptibility and if restoration of immunological tolerance was achievable. We found that low thymic expression of a targeted liver autoantigen was necessary but not sufficient to allow the development of AIH and that decreased peripheral tolerance to a liver autoantigen was the main factor responsible for the development of an autoimmune liver inflammation. Adoptive transfer of ex vivo expanded CXCR3+ Tregs from xenoimmunized mice to mice with AIH restored peripheral tolerance and induced remission of the disease.
Experimental AIH was induced in mice by xenoimmunization as described.9-11 Briefly, C57BL/6, 129S/v, or BALB/c female mice (6 to 8 weeks of age) (Charles River, St-Constant, Canada) were injected in the tibialis cranialis muscle with 100 μg (50 μL) of plasmids coding for type 2 AIH human autoantigens and murine IL-12 (pRc/CMV-CTLA-4-CYP2D6-FTCD and pVR-IL12)11 dissolved in phosphate-buffered saline (Invitrogen, Carlsbad, CA). Mice were injected three times at 2-week intervals. All experiments were performed under protocols approved by the CHU Sainte-Justine institutional committee for animal care and following guidelines published by the Canadian Council on Animal Care. All plasmids were purified using the Qiagen Endofree Plasmid Giga Kit (Santa Clarita, CA), according to the manufacturer's guidelines.
Ex Vivo Treg Expansion.
Isolated CD4+CD25+ Tregs were expanded for 14 days with anti-CD3-CD28 microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany). Briefly, isolated CD4+CD25+ cells were cultured in the presence of anti-CD3-CD28 microbeads at a bead-to-cell ratio of 2:1 in RPMI 1640 supplemented with 10% fetal calf serum (FCS) and 2,000 U/mL of murine rIL-2 (Invitrogen, Carlsbad, CA). On days 3 and 5, 50% of medium was replaced with fresh RPMI 1640 supplemented with 10% FCS and 2,000 U/mL of murine rIL-2. At day 7, cells were washed and beads were magnetically removed and cells were put back in culture for an additional 7 days with anti-CD3-CD28 microbeads at a bead-to-cell ratio of 1:1 in RPMI 1640 supplemented with 10% FCS and 2,000 U/mL of murine rIL-2. At day 14, cells were washed and beads magnetically removed and analyzed by flow cytometry (BD Biosciences, Mississauga, Canada).
Regulatory T-Cell Adoptive Transfer.
Ex vivo expanded Tregs were used for all adoptive transfer. Cells were CFSE labeled (5 μM) (Vybrant CFDA SE Cell Tracer Kit, Molecular Probes, Eugene, OR) and washed in phosphate-buffered saline (PBS) / 5% FCS. Cells (1 × 106 or 2.5 × 106) were then suspended in sterile PBS (Invitrogen) and IV-transferred into xenoimmunized C57BL/6 mice. Frequency of CFSE+ cells in peripheral blood mononuclear cells (PBMCs) and CXCR3 expression was monitored by flow cytometry (BD Biosciences).
The detailed methodology is described in the Supporting Methods.
Susceptibility to Experimental Type 2 AIH Correlates with Autoreactivity to FTCD.
C57BL/6 (H2-b), 129S/v (H2-b), and BALB/c (H2-d) were DNA vaccinated with pRc/CMV-CTLA-4-CYP2D6-FTCD encoding for human CYP2D6 and FTCD12-15 and pVR-IL12, expressing murine IL-12, to induce a Th1 immune response. Consistent with a previous report,9 5 months after xenoimmunization C57BL/6 (H2-b) mice showed elevated levels of serum ALT, indicative of hepatocyte lysis, whereas 129S/v (H2-b) and BALB/c (H2-d) mice maintained normal levels (Fig. 1A). Liver histology revealed that C57BL/6 mice had significantly higher grades of liver inflammation than 129S/v and BALB/c (Fig. 1B). C57BL/6 mice developed higher titers of anti-mouse FTCD autoantibodies than 129S/v or BALB/c mice (Fig. 1C). Titers of anti-mouse FTCD autoantibodies correlated with the degree of inflammation in all three strains (Fig. 1D). This result extends our previous observation that B-cell autoreactivity toward FTCD correlates with AIH disease activity in male and female C57BL/6 mice.10
Xenoimmunized Mice with AIH Show an Autoreactive T-Cell Proliferative Response to FTCD.
The T-cell proliferative response to murine FTCD was used to detect the presence of autoreactive FTCD-specific T cells in xenoimmunized animals. T cells from xenoimmunized BALB/c mice showed only modest proliferation in the presence of FTCD, and removal of CD8− cells (purified CD8+) did not significantly change that level (Fig. 2A). Splenocytes from 129S/v showed low FTCD-induced CD8+ T-cell proliferation, but it increased when CD8+ T cells were purified, suggesting that CD8− T cells had an inhibitory effect on proliferation (Fig. 2A). Splenocytes from xenoimmunized C57BL/6 mice showed robust CD8+ T-cell proliferation in the presence of FTCD, which increased significantly when CD8+ T cells were purified, suggesting that, as in 129S/v mice, the CD8− compartment contains cells able to inhibit autoantigen-induced proliferation (Fig. 2A). CD8+ T cell purified from liver-infiltrating lymphocytes (LILs) showed a higher level of FTCD-induced proliferation than those isolated from spleen, except for LIL from BALB/c mice (Fig. 2B,C). The percentage of FTCD-specific T cell that divided (precursor frequency) in the liver after xenoimmunization (Fig. 2C) correlates with the grade of inflammation observed (r2 = 0.8463, P < 0.0001) and with serum ALT levels (r2 = 0.4439, P = 0.0198) (data not shown). These results show that FTCD is targeted by autoreactive CD8+ T cells, and in animals that develop liver inflammation (C57BL/6 and 129S/v), CD8− cells can inhibit the proliferation of autoreactive CD8+ T cells.
Central Tolerance to AIH Autoantigen.
The thymic expression level of an autoantigen is known to inversely correlate with the number of autoreactive T cell specific to that autoantigen.16-18 To explore this possibility, the expression level of liver autoantigens in the thymus was measured. All strains expressed similar levels of CYP2D9, the murine equivalent of human CYP2D6, but low levels of mouse FTCD (Fig. 3A). Moreover, BALB/c mice expressed significantly higher levels of FTCD compared to C57BL/6 and 129S/v mice (Fig. 3A). Therefore, this increased central tolerance to a liver autoantigen in BALB/c mice could result in low numbers of circulating FTCD-specific autoreactive T cells. The expression level of the AIRE (AutoImmune REgulator) transcription factor, a factor necessary for the ectopic expression of several peripheral autoantigens in the thymus19 including FTCD,10 was similar in all three strains (Fig. 3B). Expression of CYP2D9 in the liver was similar in all three strains but the expression of FTCD was significantly lower in C57BL/6 mice than in 129S/v or BALB/c (Fig. 3C). This lower peripheral expression of autoantigen could influence the development of peripheral tolerance to FTCD in C57BL/6 mice.
Xenoimmunized C57BL/6 Mice Show Reduced Levels of Functional CD4+CD25+FoxP3+ Tregs.
Xenoimmunized C57BL/6 and 129S/v mice showed an increased autoimmune T-cell proliferation when CD8− cells were removed (Fig. 2), suggesting that peripheral tolerance by Tregs could be involved in AIH susceptibility. Levels of CD4+CD25highFoxp3+ Tregs were measured in xenoimmunized mice of all three strains. C57BL/6 showed a lower frequency of CD4+CD25+FoxP3+ cells in the spleen and liver than 129S/v or BALB/c mice (Fig. 4A, left panel). The level of FoxP3 expression by CD4+ Tregs was similar in all three strains (Fig. 4A, right panel). Tregs present in xenoimmunized mice of each strain were able to suppress their own FTCD-induced CD8+ T-cells proliferation (Fig. 4B). Furthermore, Tregs from xenoimmunized C57BL/6 (H2-b) or 129S/v (H2-b) mice suppressed equally well proliferation of C57BL/6 CD8+ T cells in the presence of FTCD (Fig. 4C). However, at the 0.1:1 ratio, Tregs from 129S/v mice more efficiently controlled proliferation of CD8+ T cell than C57BL/6 mice (Fig. 4C). This suppression was mostly mediated through secretion of IL-10 (Fig. 4D,E), no TGF-β was detected (data not shown). Tregs from C57BL/6 secreted similar amounts of IL-10 than those from 129S/v mice (Fig. 4D). CD8+ T-cell proliferation in the presence of Tregs and neutralizing IL-10 antibody was 90% that of CD8+ T cell without Tregs (Fig. 4E), suggesting that the inhibition of proliferation by CD4+ Tregs was mainly mediated through IL-10 secretion. Altogether, these data show that even if present in lower number, Tregs in xenoimmunized C57BL/6 mice are functional and able to suppress the proliferation of FTCD-specific autoreactive T cells, mostly through IL-10 secretion.
Ex Vivo Expansion of Tregs and Their Recruitment by the Liver.
Xenoimmunization of C57BL/6 mice induced the expression of chemokines MIG (CXCL9) and IP10 (CXCL10) in the liver at significantly higher levels than in 129S/v (Fig. 5A), correlating with the grade of inflammation present in that organ (r2 = 0.6825, P = 0.0115). CD4+CD25+FoxP3+ Tregs from xenoimmunized C57BL/6 mice expressed high levels of CXCR3 (Fig. 5B), the cognate receptor of MIG and IP10, especially Tregs isolated from the liver (Fig. 5C). This observation suggests that the recruitment of Tregs by the liver in C57BL/6 mice could be mediated by CXCR3. To test this hypothesis, CD4+CD25+ Tregs from xenoimmunized mice were isolated and expanded in vitro for 2 weeks. Ex vivo expanded cells maintained their phenotype, including FoxP3 expression (Fig. 6A), and were able to suppress proliferation of T cells in the presence of FTCD (data not shown). Tregs were not specifically enriched for CXCR3 but its expression among Tregs after ex vivo expansion suggests that polyclonal expansion combined with the high levels of IL-2 were sufficient to maintain expression of CXCR3 (35.8 ± 3.7% of Tregs were CXCR3+ initially versus 28.0 ± 4.7% after expansion). CFSE-labeled Tregs (1 × 106) were transferred IV into xenoimmunized C57BL/6 mice. CXCR3-expressing Tregs were detectable after transfer among PBMC (Fig. 6B) but rapidly migrated to the liver where the majority of transferred cells were recovered (Fig. 6C). These data show that Tregs from xenoimmunized C57BL/6 mice can be expanded ex vivo, maintain their functionality, and migrate efficiently to the liver when adoptively transferred.
Adoptive Transfer of Ex Vivo Expanded Tregs.
Our results suggest that a lack in numbers, not functionality, of CD4+ Tregs in the liver of xenoimmunized C57BL/6 mice could be responsible for the chronic autoimmune hepatitis. Thus, 2.5 × 106ex vivo expanded CD4+CD25+ T cells from xenoimmunized C57BL/6 mice were IV-transferred to xenoimmunized C57BL/6 mice with AIH (6 months postvaccination). On month 7 postvaccination, LILs were isolated and proliferation to FTCD was measured. Mice that received Tregs showed significantly reduced T cell proliferation to FTCD (Fig. 7A). The percentage of CD4+CD25+FoxP3+ Tregs in the liver of C57BL/6 mice that received Tregs did not increase significantly compared to xenoimmunized control C57BL/6 mice (Fig. 7B). Liver histology revealed a significant decrease in the degree of liver inflammation in Tregs-treated mice (Fig. 7C).
Susceptibility to autoimmune hepatitis results from the interaction of several factors including age, sex, genetic background, and environment.20 The study of its pathogenesis has proven to be particularly challenging, thus great efforts have been made to develop animal models of the disease. Herein, we show that in the type 2 AIH experimental model, host factors have a definite influence on liver injury and that they encompass both central and peripheral tolerance mechanisms and that adoptive transfer of autologous ex vivo expanded CXCR3+ Tregs can restore peripheral tolerance and significantly reduce liver inflammation.
Xenoimmunization with DNA coding for human type 2 AIH autoantigens (FTCD and CYP2D6) induces an autoimmune hepatitis in female10 C57BL/6 mice but not in BALB/c mice and only a mild liver inflammation in 129S/v mice. The AIH in C57BL/6 mice is characterized by the development of autoantibodies directed against mouse FTCD at titers that directly correlates with the grade of liver inflammation.10 Interestingly, in patients with type 2 AIH titers of anti-LC1 autoantibodies (anti-FTCD) also correlate with disease activity.21 In contrast, titers of anti-CYP2D9 (the murine homolog of CYP2D6) did not correlate with disease activity in this model (data not shown and Lapierre et al.10), as was observed with anti-LKM1 autoantibodies (anti-CYP2D6) in patients with type 2 AIH.21 CD8+ autoreactive T cells against mouse FTCD were found in the spleen and liver of xenoimmunized mice and their proliferation correlated with the grade of inflammation and serum ALT levels, suggesting that FTCD-specific cytotoxic T lymphocytes (CTL) are involved in hepatocyte lysis. Therefore, B- and T-cell responses against FTCD are both linked with the development of AIH. Interestingly, xenoimmunization of C57BL/6 mice with DNA encoding only for human FTCD can break immune tolerance to mouse FTCD and induce AIH (data not shown), suggesting that an immune response targeted exclusively against murine FTCD is sufficient to produce AIH.
Thymic expression of autoantigens is a key factor for the negative selection of autoreactive T cells; a reduction in insulin thymic expression level has been shown to result in a proportional increase in the number of circulating insulin-specific autoreactive T cells and increased susceptibility to diabetes.16-18 Thymic expression of FTCD was significantly higher in BALB/c mice and showed a very low frequency of FTCD-specific T cells. The FTCD gene is identical between 129S/v, BALB/c, and C57BL/6 except for one A>C polymorphism in intron 2 of C57BL/6 mice (Mouse Genomes Project, Wellcome Trust Sanger Institute, accession numbers ERP000035, ERP000039, ERP000041). Therefore, the increased FTCD expression in BALB/c mice is likely not the result of a mutated FTCD gene. FTCD expression in the thymus is under control of the AIRE transcription factor,10 a factor responsible for the ectopic expression of peripheral autoantigens in the thymus.19 There were no significant differences in the expression level of AIRE between the three strains. In humans with type 1 or 2 AIH, AIRE mutations have not been associated with the disease22 despite early reports.23
BALB/c mice express different MHC genes than 129S/v or C57BL/6 mice (H2-d versus H2-b). Therefore, in addition to the increased thymic expression of FTCD, the repertoire of FTCD-specific T cells is likely different from that of 129S/v or C57BL/6, a factor which could influence the ability of the xenoimmunization to activate crossreactive T cells. The thymic expression level of FTCD, the lack of autoreactive CD8+ T cell, and the absence of a significant Treg response allow us to postulate that in BALB/c mice autoreactive FTCD-specific T cell are more efficiently deleted in the thymus. Therefore, in this model, central tolerance and/or MHC genes contribute to make the BALB/c mice resistant to AIH.
129S/v mice develop a B- and T-cell reactivity against murine FTCD which does not lead to a significant degree of inflammation of the liver. Purified CD8+ T cells from these mice showed increased proliferation, suggesting that xenoimmunization induced a Treg response in these mice. 129S/v mice have significantly higher numbers of Treg after xenoimmunization than C57BL/6 mice. IL-10-secreting Tregs from 129S/v and C57BL/6 mice were equally effective in controlling the proliferation of CD8+ T cell induced by FTCD. These results suggest that the lack of peripheral tolerance to FTCD in C57BL/6 mice does not stem from a dysfunction, but a lack of Tregs. In patients with AIH, as in those with multiple sclerosis24 or lupus,25 low numbers of functional Tregs have been reported.2 Despite the presence of FTCD-specific autoreactive T cells, 129S/v mice are likely able to prevent the development of AIH by controlling their proliferation by way of an increased Treg response. Interestingly, CD4+ Tregs isolated from xenoimmunized 129S/v mice were more effective at inhibiting proliferation at lower ratios (0.1:1) than Tregs from C57BL/6 mice, possibly contributing to the observed mild autoimmune disease in these mice. Previously, we found that xenoimmunized B6.129S2-AIREtm1.1Doi/J (+/0) mice did not develop AIH, despite reduced central tolerance to FTCD and related increased in FTCD-specific autoreactive T cells, owing to an increased number of Tregs.10 Therefore, we believe that peripheral tolerance to liver autoantigens is paramount in AIH susceptibility.
The lower expression of FTCD in the liver of C57BL/6 mice could reduce the exposure of APCs to murine FTCD, diminishing their ability to induce anergy in FTCD-specific autoreactive T cells. This could lead to higher numbers of FTCD-specific autoreactive T cells available for activation by the xenoimmunization. This reduced expression of FTCD by the liver could also influence the number of FTCD-specific Tregs generated through peripheral conversion of naïve CD4+ autoreactive T cells.26 Peripheral conversion is believed to be responsible for a small fraction of Tregs,26 but in a model of autoimmune colitis, peripheral conversion to Tregs diminished the severity of the disease27 highlighting the importance of this pathway in the control of autoimmune inflammation.
CXCR3+ Tregs express T-bet, the Th1-specifying transcription factor, and specifically accumulate at sites of Th1 cell-mediated inflammation.28 CXCR3 was found to mediate recruitment of Tregs to the liver, through expression of CXCL9 and CXCL10 by liver sinusoidal endothelium cells in a proinflammatory microenvironment, such as the one observed in the liver of patients with chronic hepatitis.29 In mice, a lack of CXCR3 signaling leads to reduced recruitment of Tregs to the liver and an exacerbated liver disease.30 A lack of Treg recruitment in the central nervous system (CNS) in the absence of CXCR3 has also been observed in a model of experimental autoimmune encephalitis.31 The lack of Tregs in the liver of our xenoimmunized C57BL/6 mice was not the result of a deficiency in their recruitment; CXCL9 and CXCL10 are abundantly expressed in the liver of those animals and their cognate receptor, CXCR3, is expressed by C57BL/6 Tregs. Furthermore, in C57BL/6 mice the majority of Tregs expressing high levels of CXCR3 are found in the liver. In addition, ex vivo expanded CFSE-labeled CXCR3+ Tregs from xenoimmunized C57BL/6 mice IV-transferred into C57BL/6 mice with AIH were rapidly recruited by the liver showing that, in C57BL/6 mice, CXCR3-mediated Treg recruitment was functional.
Remarkable efforts have been made to develop adoptive transfer methods of ex vivo expanded Tregs as a treatment for patients with autoimmune diseases.32 In AIH, CD4+ Tregs are present in fewer numbers in patients than in healthy controls2-4 and functional Tregs can be expanded ex vivo.5, 6 Therefore, the idea that Tregs could be used to treat AIH patients has generated great enthusiasm.7 However, the effectiveness of Treg infusions, as a means to restore immunological tolerance to liver autoantigens, has never been demonstrated. Herein, we found that Tregs from xenoimmunized C57BL/6 mice expanded ex vivo maintained their functionality and CXCR3 expression. Adoptively transferred ex vivo expanded CXCR3+ Tregs were recruited by the liver of C57BL/6 mice with AIH, similar to an autologous transfer in AIH patients. They restored peripheral tolerance to a liver autoantigen (FTCD) and induced remission of AIH. One month after transfer, the numbers of Tregs in the liver of IV-transferred B6 mice were comparable to those of control xenoimmunized B6 mice. This observation suggests that upon the initial resolution of the inflammation and restoration of immunological tolerance to FTCD, high levels of Tregs were not necessary to maintain remission. However, the long-term maintenance of this immunological tolerance is likely conditional on the absence of a triggering factor (xenoimmunization). Therefore, despite an initial break of tolerance, long-term restoration or “resetting” of immunological tolerance in AIH can be achieved.
Based on these observations, we believe that infusion of autologous ex vivo expanded Tregs could be an effective therapeutic approach for the treatment of patients with autoimmune hepatitis. Because the CXCR3 pathway of Tregs recruitment in AIH is functional,29 CXCR3+ Tregs could target the inflamed liver hence potentiating the effectiveness of autologous Treg transfers.