Forkhead box p3+ regulatory T cell underlies male resistance to experimental type 2 autoimmune hepatitis


  • Pascal Lapierre,

    1. Division of Gastroenterology, Hepatology and NutritionUniversité de Montréal, Canada
    Search for more papers by this author
  • Kathie Béland,

    1. Division of Gastroenterology, Hepatology and NutritionUniversité de Montréal, Canada
    Search for more papers by this author
  • Caroline Martin,

    1. Division of Gastroenterology, Hepatology and NutritionUniversité de Montréal, Canada
    Search for more papers by this author
  • Fernando Alvarez Jr.,

    1. Division of Gastroenterology, Hepatology and NutritionUniversité de Montréal, Canada
    Search for more papers by this author
  • Fernando Alvarez

    Corresponding author
    1. Division of Gastroenterology, Hepatology and NutritionUniversité de Montréal, Canada
    2. Department of Pediatrics, CHU Sainte-Justine Université de Montréal, Canada
    3. Department of Microbiology and Immunology, Université de Montréal, Canada
    • Division of Gastroenterology, Hepatology and Nutrition, CHU Sainte-Justine, 3175 Côte Ste-Catherine, Montréal (Québec), Canada H3T 1C5
    Search for more papers by this author
    • fax: 514-345-4999

  • Potential conflict of interest: Nothing to report.


Autoimmune hepatitis (AIH), like many autoimmune diseases, is most prevalent in young women. The immunological basis of this age and sex susceptibility bias was investigated in a murine model of AIH. Xenoimmunization of 7-week-old female C57BL/6 mice resulted in more severe AIH with higher levels of liver inflammation, serum alanine aminotransferase, specific T-cell cytotoxicity, and autoantibody than younger and older females. Vaccinated males developed minimal liver inflammation and higher percentages of CD4+CD25+FoxP3+ regulatory T cell in peripheral blood mononuclear cells, spleen, and liver than females. Regulatory T cells (Tregs) were virtually absent in liver-lymphocytes infiltrates of females. Castration of C57BL/6 mice, with or without 17β-estradiol supplementation, did not modify susceptibility in males, nor Treg numbers, suggesting minimal contribution of testosterone and estradiol to autoimmune hepatitis (AIH) susceptibility. Xenoimmunized Aire(+/0) mouse displayed similar AIH susceptibility, sex bias, and Tregs numbers as C57BL/6 mice, suggesting that susceptibility in females is not the result of less stringent thymic central tolerance. Autoreactive B cell response against formiminotransferase-cyclodeaminase correlated with disease activity, possibly linking B-cell autoreactivity and AIH pathogenesis. Conclusion: Peripheral tolerance and development of regulatory T cells after self-mimicking antigen exposure, and not sexual hormone nor central tolerance, are the main factors for susceptibility to AIH in females. HEPATOLOGY 2010

Prevalence of most autoimmune diseases shows a striking sex difference, with women being affected more often than men.1, 2 The ratio of female patients to male ranges from 20:1 in Sjögren's syndrome, to 3:2 in multiple sclerosis.2 Less frequently, the ratio approaches 1:1, as in ulcerative colitis and diabetes.2 In autoimmune hepatitis (AIH), the female to male ratio ranges from 3:1 in type 1 AIH to 9:1 in type 2 AIH.3

Differences in the immune response have also been observed between women and men. Higher level of antibodies and stronger T cell activation are observed in women after vaccination.4 Women also have higher absolute numbers of CD4+ T cells and produce higher levels of Th1 cytokines than men.5 Age also affects immune responses, including incidence of several autoimmune diseases. In AIH, 40% of cases of type 1 are diagnosed before the age of 18 years, with a mean age at onset of 10 years,6, 7 and 80% of cases of type 2 are diagnosed before the age of 18 years, with a mean age at onset of 6.5 years.6, 7 A second peak of incidence of AIH has also been reported in women after menopause.8 These prepubertal and postmenopausal peaks of incidence suggest that the hormonal status could influence susceptibility to AIH.

Research on autoimmune diseases sex bias is scarce. Studies on AIH susceptibility factors, including its sex bias, have been severely limited by the lack of experimental models. Recently, a murine experimental model of AIH has been produced9 in which mice develop a disease very similar to that observed in humans.10 This murine model of type 2 AIH is initiated by xenoimmunization of 6-week-old to 8-week-old female C57BL/6 mice with human type 2 AIH antigens that, by molecular mimicry, triggers an autoreactive immune response against homologous murine liver proteins.9 C57BL/6 mice were found to be more susceptible to developing an AIH than 129S/v or BALB/c mice, showing that this model of AIH is under the influence of both major histocompatibility complex and non–major histocompatibility complex genes.11 The close parallels between this experimental model and AIH in humans10, 11 are such that it is ideally suited for the study of immunological mechanisms of susceptibility to AIH on the basis of sex and age.

Herein, we report that, as in humans, female mice of a specific age were most susceptible to developing an AIH. In these mice, a break of B cell immunological tolerance against liver proteins was detected early on and then paralleled the grade of liver inflammation. Female susceptibility was not the result of a failure in thymic negative selection of autoreactive T cells but of the generation of lower numbers of FoxP3+ regulatory T cells (Tregs) in response to xenoimmunization. Furthermore, male resistance to AIH was not mediated by testosterone nor testes-induced peripheral tolerance to liver antigens, and susceptibility in females was not linked to 17β-estradiol levels.

Materials and Methods


AIH, autoimmune hepatitis; CYP2D6, Cytochrome P450 2D6; FTCD, formiminotransferase cyclodeaminase; IL, interleukin; LC1, liver cytosol type 1; PBMC, peripheral blood mononuclear cells; PCR, polymerase chain reaction; Treg, regulatory T cell.


All experiments with C57BL/6 mice (Charles River, Canada) and B6.129S2-Airetm1.1Doi/J (The Jackson Laboratory, MN) were performed under protocols approved by the institutional committee for animal care and following guidelines published by the Canadian Council on Animal Care. Genotype of B6.129S2-Airetm1.1Doi/J mice was confirmed using two specific polymerase chain reaction (PCR) reactions on tail clippings as previously described.12 Briefly, two PCR reactions were performed with the following primers: Set 1-GTCATGTTGACGGATCCAGGGTAGAAAGT and AGACTAGGTGTTCCCTCCCAACCTCAG; Set 2-ATAGCACCACGACACCCAAG and ATATCATTCTCCAACTCCTGCCTCTTT. In wild-type mice, the first set of primers results in an amplicon of 1150 bp, whereas in knockout mice a fragment of 690 bp is produced. The second set of primers generates a product of 507 bp in wild-type mice, whereas in knockout mice for the Aire gene no amplicon is produced.

Surgical castration of 3-week-old male C57BL/6 mice was performed by Charles River, Canada. A group of castrated C57BL/6 mice received subcutaneous 90-day timed-release pellets (Innovative Research of America, FL) containing E2 (17β-estradiol) (1.5 mg/pellet), which reproduces murine physiological levels.13 These pellets were implanted every 90 days for the duration of the study.

Experimental Type 2 Autoimmune Hepatitis.

Experimental AIH was induced in mice by xenoimmunization as previously described.9, 11 Briefly, C57BL/6 or B6.129S2-Airetm1.1Doi/J mice (male or female at 4, 7, or 14 weeks of age) were injected in the tibialis cranialis muscle with 100 μg (50 μL) of plasmids coding for type 2 AIH human autoantigens and murine interleukin (IL)-12 (pRc/CMV- CTLA-4-CYP2D6-FTCD and pVR-IL12)9, 11 dissolved in saline buffer. Mice were injected three times, at 2-week intervals. Control mice were injected with the pVR-IL12 plasmid only (100 μg, 3 times). All plasmids were propagated in Escherichia coli by standard techniques and purified using QIAGEN Endofree Plasmid Giga Kit (QIAGEN, Santa Clarita, CA), according to the manufacturer's guidelines.

Serum Alanine Aminotransferase Activity.

Serum alanine aminotransferase levels were measured in a Beckman-Synchron CX9 apparatus, from blood samples taken every month after the last plasmid injection.

Histopathology and Immunohistochemistry.

Mice were sacrificed, and their livers were dehydrated, embedded in paraffin, sectioned, and stained with hematoxylin-phloxine-safran.

Enzyme-Linked Immunosorbent Assay.

Enzyme-linked immunosorbent assay was performed as described.9, 11, 14 Briefly, the fusion protein produced by the pMAL-cR1-CYP2D6-FTCD plasmid (human FTCD and CYP2D6) or pEt-30C-mFTCD (murine FTCD) or pEt-30c-CYP2D9 (murine member of the P450 2D subfamily homologous to human CYP2D6) were purified and used as antigen in the enzyme-linked immunosorbent assay (0.2 μg/well). An antiserum was considered positive if its specific OD was at least 2 times higher than the mean optical density of the preimmune mice sera.

Western Blot Analysis.

Proteins expressed from the pMAL-cR1-CYP2D6-FTCD or pEt-30C-mFTCD vector were separated by electrophoresis on 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred onto nitrocellulose filters (Amersham Life Sciences, Oakville, Canada). The western blot technique was carried out as previously described.9, 11

Flow Cytometry Analysis.

Regulatory T cells' frequency in blood, spleen, or amongst liver-infiltrating lymphocytes was assessed by simultaneous surface and intracellular immunofluorescence staining using the mouse regulatory T cell staining kit (eBioscience, CA). Each reaction was performed with 1 × 106 cells, and a minimum of 200,000 events were recorded. Isotypic controls were included for each sample tested. Fluorescence-positive cells were analyzed with a FACScalibur unit (Becton Dickinson, CA).

Cytotoxicity Assay.

EL4 cells, an H-2b lymphoma T cell line (ATCC, VA), served as targets for cytotoxicity assays. Briefly, 1 × 104 target cells were incubated with CYP2D6-FTCD fusion protein and left for 24 hours for antigen processing. Cells were then co-cultured with serial dilutions of 1 × 104 to 5 × 105 effector cells in a final volume of 200 μL. After 5 hours of incubation at 37°C, the release of lactate dehydrogenase was measured at 490 nm using the CytoTox 96 assay kit (Promega, Madison,WI). Lysis percentage was calculated by the formula: 100 × (A − B − C)/(D − C), in which A is experimental value (test release), B is spontaneous background signal value from effector cells, C is spontaneous background signal value from target cells, and D is the target maximum signal value. Maximum release and spontaneous release were determined by incubating cells with lysis solution and culture medium, respectively.

CYP2D9 and mFTCD Expression Level.

RNA was isolated from thymuses of newborn mice (1-2 days old) and livers of newborn and 7-week-old C57BL/6 mice using the RNeasy Micro kit (QIAGEN, CA). Sex of newborns was confirmed by PCR with male-specific Sry primers (TGGGACTGGTGACAATTGTC and GAGTACAGGTGTGCAGCTCT) as previously described.15 Expression of liver autoantigens was studied using specific primers for murine FTCD and CYP2D9 (TGCTGCCTGTTTGGAGGCAA, AAGCAAGGCTTGGGCCACTT and GAGCAGAGGCGATTCTCTGT, CCCAGGTGGTCCTATTCTCA, respectively). PCR was performed using the OneStep RT-PCR Kit (QIAGEN, CA), and murine β-actin expression level was used as internal reference.

Statistical Analysis.

Differences between groups were tested using the Kruskal-Wallis test with Dunn's post test. Correlation coefficients were computed using Pearson's test. In all graphs, error bars represent standard deviations. All statistical analyses were performed using GraphPad Prism version 4 (GraphPad Software, CA).


Sex and Age Influence on Susceptibility to Experimental AIH.

To assess the influence of sex and age on the development of an experimental autoimmune hepatitis, 4-week-old, 7-week-old, and 14-week-old C57BL/6 female mice and 7-week-old male mice were xenoimmunized with pRc/CMV-CTLA-4-CYP2D6-FTCD and pVR-IL12.9 Female C57BL/6 mice immunized at 7 weeks of age were the only group that showed elevated serum levels of alanine aminotransferase, a marker of hepatocyte lysis, from month 6 post-immunization (P < 0.05) (Fig. 1A). Liver histological analysis showed very mild inflammation in male and 14 week-old mice compared with 7-week-old females (P < 0.01). Younger (4-week-old) female mice showed an intermediate grade of liver inflammation (Fig. 1B). Liver-infiltrating T cells and splenocytes isolated from males and 4-week-old or 14-week-old females showed low specific cytotoxicity against type 2 AIH antigens compared with 7-week-old female mice (Fig. 1C). Seven-week-old female C57BL/6 mice were susceptible, whereas 4-week-old and 14-week-old females were less prone to the development of an experimental AIH. Seven-week-old female mice were also more vulnerable than males at the same age. In this animal model, as was observed in humans, female sex and age are susceptibility factors for the onset of AIH.

Figure 1.

C57BL/6 female mice xenoimmunized at 7 weeks of age develop the most severe autoimmune hepatitis. (A) From month 6 post-vaccination, C57BL/6 female mice vaccinated at 7 weeks of age show significantly higher alanine aminotransferase levels compared with males or females vaccinated at 4 or 14 weeks of age (n = 8 for each group, histogram represent the average of two separate experiments) (*P < 0.05). (B) At autopsy, female mice vaccinated at 7 weeks of age have increased liver inflammation compared with males or female mice vaccinated at 14 weeks of age (**P < 0.01, liver inflammation grading is based on the modified Ishak histological activity index). (C) Lymphocytes isolated from female mice vaccinated at 7 weeks of age have higher specific cytotoxicity against hepatic autoantigens (FTCD and CYP2D6) than mice of other sex or age (effector/target ratios [E:T]).

B Cell Immune Response.

Antibodies against the injected xenoantigens were measured in the four groups of mice. Females vaccinated at 7 weeks of age showed the highest titer of antibodies (P < 0.05) (Fig. 2A). Titers reached their maximal level in the first month and remained elevated until the 8th month post-vaccination. When autoantibodies against murine formiminotransferase-cyclodeaminase (FTCD) and CYP2D9 (the murine homolog of human CYP2D6) were measured, female mice vaccinated at 7 weeks of age showed significantly higher levels of anti-mFTCD autoantibodies than mice from other groups (Fig. 2C). No statistically significant differences in reactivity against CYP2D9 were found between sera from all four groups (Fig. 2B). However, autoantibodies level increased over time, a feature evident in 7-week-old female mice sera reactivity against mFTCD.

Figure 2.

Autoreactive B cell response. (A) A statistically significant difference of specific type 2 autoimmune hepatitis antibodies (anti- liver/kidney microsomal type 1/anti-LC1) titer was observed between male and female mice when human antigens were used (*P < 0.05). (B) When mouse autoantigens were used, anti-CYP2D9 autoantibody titer was similar in all groups; however, (C) anti-mFTCD (mouse) autoantibody titers were higher in female mice vaccinated at 7 weeks of age, and (D) this B cell response against murine FTCD on month 8 correlates with liver inflammation levels in mice from each group. (E) Mice that will develop an AIH show a conversion from reactivity against the vaccinated human proteins (CYP2D6-FTCD) to an autoreactivity against the FTCD murine autoantigen.

Interestingly, levels of mFTCD autoantibodies correlated with the histological activity index in mice from all four groups, suggesting a possible role for B cell response against mFTCD in development of the disease (Fig. 2D). To characterize this B cell response, reactivity against xenoantigens (CYP2D6 and FTCD) and mFTCD were compared by western blot to detect high-affinity antibodies targeting linear epitopes. Early on, reactivity against injected antigens was found in all mice (human CYP2D6-FTCD) (Fig. 2E). However, a shift of B-cell reactivity (human FTCD) to autoreactivity (murine FTCD) occurred in female mice (Fig. 2E). This type of B cell shift to autoreactivity was not observed in male mice.

Autoantigen Expression Level in the Target Organ.

The expression level in the liver of the targeted antigens, mFTCD and CYP2D9 could potentially influence the reactivity of specific T cells and development of AIH. Therefore, their expression level was assessed in livers from newborn and 7-week-old C57BL/6 mice. Male and female newborn (data not shown) and 7-week-old C57BL/6 mice showed similar hepatic expression levels of both mFTCD and CYP2D9 (Fig. 3A). Therefore, the amount of autoantigen in hepatocytes is not related to the female's susceptibility to AIH.

Figure 3.

Hepatic expression level of FTCD and CYP2D9 and central tolerance against these autoantigens are not responsible for female C57BL/6 mice susceptibility to AIH. (A) Male and female C57BL/6 mice have similar hepatic expression levels of CYP2D9 and mFTCD. (B) Male and female C57BL/6 mice have similar thymic expression levels of CYP2D9 and mFTCD. However, C57BL/6 mice show low-level thymic expression of FTCD compared with CYP2D9 expression levels. (C) Heterozygous and homozygous B6.129S2-Airetm1.1Doi/J mice (+/0 and 0/0) have reduced FTCD thymic expression (n = 8 per group). (D) Heterozygous B6.129S2-Airetm1.1Doi/J male and female mice (+/0) show the same sex bias in AIH development as C57BL/6 mice after xenoimmunization. The grade of liver inflammation is comparable to that of xenoimmunized wild-type mice (+/+) (n = 8 per group;*P < 0.05, **P < 0.01).

Central Tolerance.

The thymic expression level of FTCD and CYP2D9 was measured in C57BL/6 newborn mice. No differences were observed between males and females (Fig. 3B). FTCD thymic expression level was lower than that of CYP2D9 (Fig. 3B), raising the possibility that a less effective negative selection of autoreactive T-cells against FTCD could occur.

To test for other factors influencing the expression of known liver autoantigens in the thymus and their relationship with the observed sex difference in AIH susceptibility, B6.129S2-Airetm1.1Doi/J transgenic Aire knockout mice were studied. Aire, which stands for Autoimmune Regulator, is a transcription factor responsible for the ectopic expression of peripheral antigens in the thymus to allow deletion of self-reactive T cells. FTCD but not CYP2D9 is, as insulin,16 under control of the Aire transcription factor (Fig. 3C). The invalidation of one copy of the Aire gene in heterozygous mice (+/0) lowers the expression of FTCD in the thymus (Fig. 3C). Therefore, heterozygous Aire mice offers a model in which the importance of partial failure in T cell–negative selection for specific liver autoantigens on AIH development can be studied.

After xenoimmunization, male and female Aire heterozygous mice showed the same sex-bias as observed in C57BL/6 mice (Figs. 1B, 3D). Therefore, the invalidation of one copy of the Aire gene in heterozygous mice (+/0) did not modulate the grade of liver inflammation compared with wild-type mice (+/+) (Fig. 3D).

Peripheral Tolerance.

Peripheral tolerance by regulatory T cells could influence the development of an autoimmune hepatitis in mice. Xenoimmunized 7-week-old C57BL/6 male mice show a statistically significant higher percentage of Tregs in the spleen, blood, and liver than vaccinated females of the same age (Fig. 4A). The same difference is observed in vaccinated heterozygous Aire mice. Male mice show higher levels of regulatory T cells in the spleen, blood, and liver when compared with females (Fig. 4B). Significantly higher levels of regulatory T cells are found in liver infiltrates of male mice compared with female where regulatory T cells were virtually absent (Fig. 4B).

Figure 4.

Female C57BL/6 and B6.129S2-Airetm1.1Doi/J mice have fewer CD4+CD25+Foxp3+ regulatory T cells after xenoimmunization. (A) Male C57BL/6 mice (n = 8) have a higher proportion of CD4+CD25+Foxp3+ regulatory T cells in the spleen, peripheral blood mononuclear cells, and liver compared with females (n = 8) after xenoimmunization. (B) Heterozygous B6.129S2-Airetm1.1Doi/J male mice (n = 8) also show a higher proportion of regulatory T cells after xenoimmunization compared with females (n = 8) (**P < 0.01).

Sex-Related Factors Impact on Immune Tolerance.

Testes are an immunological privileged site, and as such, provide an environment able to suppress and control immune responses. In C57BL/6 mice, ectopic expression of FTCD and CYP2D9 was found in testes (Fig. 5A), and their expression was independent of the Aire transcription factor in this organ (Fig. 5B). This finding suggests that testes could influence susceptibility to AIH through peripheral conversion of autoreactive naïve T cells to FoxP3+ regulatory T cells. Sexual hormones can also directly modulate immune responses locally and systemically, and in doing so, alter the development of an autoimmune disease.

Figure 5.

C57BL/6 and B6.129S2-Airetm1.1Doi/J testes show ectopic expression of CYP2D9 and FTCD. (A) C57BL/6 male mice express FTCD and CYP2D9 ectopically in their testes. (B) Wild-type and heterozygous B6.129S2-Airetm1.1Doi/J mice express FTCD and CYP2D9 at similar levels.

Therefore, to assess the role of testes and sexual hormones on AIH susceptibility, we xenoimmunized castrated male C57BL/6 mice, supplemented or not with physiological levels of 17β-estradiol. After an 8-month follow-up, castrated male C57BL/6, supplemented or not with 17β-estradiol, showed a similar grade of liver inflammation after xenoimmunization than vaccinated male C57BL/6 mice (Fig. 6A). Castrated males' liver inflammation was also significantly lower than that of vaccinated females, showing that the absence of testes and loss of testosterone secretion have little impact on males' resistance to the development of an AIH. In addition, the administration of physiological levels of 17β-estradiol over a 9-month period did not increase male susceptibility to AIH.

Figure 6.

The absence of testosterone, testes, and 17β-estradiol supplementation does not increase male C57BL/6 mice susceptibility to AIH. (A) Castrated male C57BL/6 mice (n = 8) and castrated male C57BL/6 mice supplemented with 17β-estradiol (n = 6) develop similar liver inflammations as C57BL/6 male mice (n = 8) after xenoimmunization. Their liver inflammation is significantly lower than that of xenoimmunized C57BL/6 female mice (**P < 0.01). (B) Castrated C57BL/6 male and castrated male mice supplemented with 17β-estradiol develop higher levels of regulatory T cells compared with C57BL/6 female mice after xenoimmunization in both the liver and spleen.

Regulatory T cells populations were assessed in the spleen and liver of castrated males supplemented, or not, with 17β-estradiol, and no statistically significant differences were found with noncastrated males (Fig. 6B). However, castrated males, with or without 17β-estradiol, developed significantly higher numbers of Tregs in both the spleen and liver after xenoimmunization compared with females (Fig. 6B). These results indicate that the Tregs' population and susceptibility to AIH is not influenced by testes, testosterone, or 17β-estradiol levels.


The gender bias present in AIH has been known since the initial description of the disease by Waldenström and Kunkel, when AIH patients were referred to as “Kunkel-Waldenström girls.”17 Since then, little progress has been made in understanding the fundamental basis of female susceptibility to AIH. Herein, we report that an experimental model of AIH exhibits a similar sex bias as described in humans and is influenced by age at the time of encounter with the triggering agent.

Development of liver/kidney microsomal type 1 and liver cytosol type 1 (LC1) autoantibodies is a hallmark of type 2 AIH.18 As in previous studies,9, 11 xenoimmunized 7-week-old female mice develop high titers of liver/kidney microsomal type 1 and LC1 antibodies, switching to autoantibodies directed against mouse liver autoantigens. In mice of all group, anti-mFTCD (Anti-LC1) reactivity correlated with the grade of liver inflammation found after xenoimmunization. This suggests that in mice that develop an AIH, a loss of B cell immunological tolerance against mFTCD occurs in the first months after immunization, and through exposure to self-antigen, this B cell autoimmune response is perpetuated. Interestingly, anti-LC1 autoantibody titers were found to parallel disease activity in type 2 AIH patients.19 These observations suggest that a loss of B cell tolerance against hepatic autoantigens could be an initial and fundamental step in the development of late-onset liver autoimmunity.

In this model, a break of T cell immunological tolerance also occurs after xenoimmunization, because CD8+ T cell cytotoxicity leading to hepatocyte lysis and subsequent liver inflammation is observed. T cell immunological tolerance results from the thymus-negative selection and is directly influenced by the thymic expression level of autoantigens.16 A reduction in insulin thymic expression level has been shown to result in a proportional increase in the number of insulin-specific autoreactive T cells.20–22 Therefore, the thymus expression levels of the targeted autoantigens, CYP2D9 and FTCD, were used as surrogate markers of its ability to negatively select T cells specific to these antigens. Although no differences in expression levels were found between wild-type males and females, FTCD was found to be expressed at very low levels in the thymus compared with CYP2D9, suggesting a reduced ability in C57BL/6 mice to negatively select FTCD-specific autoreactive T cells.

Because FTCD thymic expression is under Aire control, the invalidation of one copy of the Aire gene in heterozygous B6.129S2-Airetm1.1Doi/J mice induces a thymus-specific reduction of FTCD expression in mice with an otherwise normal phenotype. Xenoimmunization of these B6.129S2-Airetm1.1Doi/J mice induced an AIH with a similar grade of liver inflammation as C57BL/6 mice and exhibited the same sex bias. Therefore, despite lowered expression of FTCD in the thymus, male mice are still resistant to AIH, and female mice develop an AIH of similar intensity. This observation and the similar thymic expression level of mFTCD and CYP2D9 in both sexes suggests that central tolerance is likely not the main factor responsible for the observed sex bias in AIH.

Peripheral tolerance to AIH autoantigens could be involved in susceptibility or resistance to AIH. The main mechanisms of peripheral tolerance are (1) the induction of functional unresponsiveness (anergy) or deletion of autoreactive T cells and (2) suppression by regulatory T cells. Because we have no means to directly measure the frequency of circulating AIH-specific CD8+ T cells, peripheral deletion or induction of anergy cannot be excluded as a possible factor in AIH susceptibility. However, male and female C57BL/6 mice express similar levels of AIH autoantigens, implying similar exposure to autoantigens in both sexes. Therefore, C57BL/6 male resistance to AIH is probably not the result of extensive peripheral deletion or induction of anergy of AIH-specific autoreactive T cells.

Regulatory T cells are key elements in the control of autoimmunity stemming from molecular mimicry,23 because Tregs have the ability to be activated at 10-fold to 100-fold lower antigen concentrations compared with naïve T cells and in the presence of low levels of CD80/86 and self-peptide/major histocompatibility complex on antigen-presenting cells.24 Furthermore, once activated by specific antigens, Tregs can exert a suppressive action on T cells irrespective of their antigen specificity.24 C57BL/6 male mice showed significantly higher numbers of regulatory T cells in the spleen, peripheral blood mononuclear cells, and liver in response to the xenoimmunization compared with females. In vaccinated female mice, Tregs were virtually absent from liver infiltrates. This parallels the observation by Longhi et al.25 that patients with AIH have decreased numbers of circulating regulatory T cells. In B6.129S2-Airetm1.1Doi/J mice, males also show higher levels of regulatory T cells than females in the spleen, peripheral blood mononuclear cells, and liver after vaccination. Interestingly, the percentage of Tregs in peripheral blood mononuclear cells and liver were higher in male B6.129S2-Airetm1.1Doi/J mice than in male C57BL/6 mice. This higher number of regulatory T cells in B6.129S2-Airetm1.1Doi/J mice could be an adaptive response to the presence of higher numbers of autoreactive T cells in these mice, which would explain the development of an AIH of similar intensity in heterozygous Aire knockout mice and C57BL/6 mice despite the reduced negative selection against mFTCD. This type of autoreactive T cells suppression in B6.129S2-Airetm1.1Doi/J mice by Foxp3+ regulatory T cells has been previously observed.26

From these data, we believe that the presence of Tregs in males could have limited the development of an autoreactive B cell response and inhibited the proliferation and cytotoxicity of autoreactive T cells, hence preventing the development of AIH. The lowered requirement of Tregs for co-activating molecules24 and the fact that hepatocytes can serve as antigen-presenting cells, with reduced expression of co-stimulatory molecules27 during an inflammatory response28 raises the possibility that Tregs could have been activated locally in the liver preferentially over naïve autoreactive T cells. Therefore, the ability to induce the proliferation of regulatory T cells after exposure to a triggering agent (xenoimmunization in this model) could be critical in preventing the development of an AIH.

The role of FoxP3 in the development of regulatory T cells and its location on the X chromosome suggests that differential regulation of this gene expression could influence the development of autoimmune diseases. However, there is no evidence that the FoxP3 gene shows a variable pattern of methylation as found in other X-linked genes.29 In addition, heterozygous female carriers of FoxP3 mutations, which in the male leads to the immune dysregulation, polyendocrinopathy, and enteropathy with x-linked inheritance syndrome, are healthy despite expression of the mutated allele in half of circulating CD4+ T cells.30 In our model, no differences in the level of FoxP3 expression in regulatory T cells were found between male and female C57BL/6 mice (data not shown).

Other factors could explain the higher proportion of regulatory T cells found in males after xenoimmunization, such as the hormonal environment and the presence of male-specific sexual organs. Testes are an immunologically privileged site, and as such, immune responses to antigens are reduced at this site. In experimental models of autoimmune diseases, intratesticular antigen injections can induce systemic tolerance and prevent development of the disease.31-33 Testes are also capable of promiscuous expression of autoantigens,34 and their repertoire of ectopic autoantigens expression is different from that of the thymus.34 In C57BL/6 mice, we found that ectopic expression of FTCD and CYP2D9 in testes and their expression was independent of the Aire transcription factor. Herein, castrated males developed the same level of liver inflammation as male C57BL/6, significantly less than females. Castration of C57BL/6 mice did not modify the level of regulatory T cells found in the spleen and liver after vaccination, indicating that induction of regulatory T cells in C57BL/6 males by xenoimmunization is not influenced by ectopic expression of hepatic autoantigen in testes.

Sexual hormones are known to directly modulate immune responses, and, in doing so, alter the development of autoimmune diseases.35 Xenoimmunization of castrated C57BL/6 males and castrated males supplemented with 17β-estradiol resulted in a grade of liver inflammation similar to that observed in noncastrated male C57BL/6 mice. Therefore, the absence of testosterone or the presence of 17β-estradiol in males did not modify the development of AIH. The production of regulatory T cells was also unaffected by the absence of testosterone or presence of 17β-estradiol: castrated males showed significantly more Tregs than females after xenoimmunization. Therefore, in this experimental model of AIH, 17β-estradiol and testosterone levels are not the main factors responsible for the observed sex bias in disease susceptibility.

Recently, using Sry(−)Y and Sry(+)X transgenic mice, Smith-Bouvier et al.36 have shown that the XX sex chromosome complement conferred susceptibility to both experimental autoimmune encephalomyelitis and lupus, irrespective of the type of gonads present.36 This observation and our data, although not excluding that sexual hormones could have some influence on the sex bias observed in autoimmune diseases, suggests that other factors related to the X chromosome could be involved in women's susceptibility to autoimmune diseases.

In summary, susceptibility to experimental AIH is not influenced by testosterone or estradiol levels nor is it the result of reduced central tolerance. Peripheral tolerance and development of regulatory T cells after self-mimicking antigen exposure are the main factor resulting in susceptibility to AIH. This suggests that the immune response raised to an initiating antigenic event could be the deciding factor for the development of an AIH.