Failure of thymic deletion and instability of autoreactive Tregs drive autoimmunity in immune-privileged liver

Abstract

The liver is an immune-privileged organ that can deactivate autoreactive T cells. Yet in autoimmune hepatitis (AIH), autoreactive T cells can defy hepatic control and attack the liver. To elucidate how tolerance to self-antigens is lost during AIH pathogenesis, we generated a spontaneous mouse model of AIH, based on recognition of an MHC class II-restricted model peptide in hepatocytes by autoreactive CD4+ T cells. We found that the hepatic peptide was not expressed in the thymus, leading to deficient thymic deletion and resulting in peripheral abundance of autoreactive CD4+ T cells. In the liver, autoreactive CD4+ effector T cells accumulated within portal ectopic lymphoid structures and maturated toward pathogenic IFN-γ and TNF coproducing cells. Expansion and pathogenic maturation of autoreactive effector T cells was enabled by a selective increase of plasticity and instability of autoantigen-specific Tregs but not of nonspecific Tregs. Indeed, antigen-specific Tregs were reduced in frequency and manifested increased IL-17 production, reduced epigenetic demethylation, and reduced expression of Foxp3. As a consequence, autoantigen-specific Tregs had a reduced suppressive capacity, as compared with that of nonspecific Tregs. In conclusion, loss of tolerance and the pathogenesis of AIH were enabled by combined failure of thymic deletion and peripheral regulation.

Keywords: Autoimmune diseases; Autoimmunity; Hepatitis; Hepatology; T cells.

Figure 1. Generation of Alb-iGP_Smarta mice and characterization of thymic and peripheral autoreactive CD4⁺ T cells.

(A) Scheme of the Alb-iGP system. The CLIP sequence in the CD74 gene was replaced by the GP61–80 sequence, facilitating high occupancy of MHC II molecules and GP61–80 presentation. (B) Cre-mediated removal of a STOP cassette facilitates expression of the mutant CD74 molecule under Rosa26 promoter control. (C) GP61–80 expression levels in thymus and liver of Itgax-iGP_Smarta and Alb-iGP_Smarta mice, as determined by quantitative RT-PCR relative to the HPRT housekeeper gene expression. (D) Representative flow cytometry dot plots of antigen-specific I-A(b) GP66–77 tetramer–binding CD4⁺ T cells in thymus (top) or spleen (bottom) in C57BL/6 mice, Alb-iGP mice expressing the mutant CD74 molecule, Smarta mice expressing a transgenic T cell receptor recognizing the cognate GP61–80 peptide, Itgax-iGP_Smarta mice, or Alb-iGP_Smarta mice, featuring both presentation of the GP61–80 peptide and the cognate Smarta T cell receptor. (E) Thymic (top) and splenic (bottom) frequencies of I-A(b) GP66–77 tetramer–specific CD4⁺ T cells in C57BL/6, Alb-iGP, Smarta, Itgax-iGP_Smarta, or Alb-iGP_Smarta mice. Data are shown as the mean ± SEM (n = 4–12). *P < 0.05; ****P < 0.0001 (C, Mann-Whitney; E, ANOVA).

Figure 2. Spontaneous development of AIH features in Alb-iGP_Smarta mice.

(A) Spontaneous development of sickness symptoms, shown as age-dependently reduced percentage of symptom-free Alb-iGP_Smarta mice (red line; n = 12), compared with Alb-iGP control mice (blue line; n = 13) and Itgax-iGP_Smarta control mice (pink line; n = 10). (B) Serum ALT and AST (each U/l; n = 8–9) in Alb-iGP_Smarta mice at early or late disease stage, and Alb-iGP control mice. (C) Representative histology of 2 Alb-iGP_Smarta livers in early disease stage, showing periportal infiltrates (top), or late disease stage (bottom), showing periportal and interface hepatitis with mainly lymphocytic infiltrates (original magnification, ×100). (D) The mHAI score (n = 4–13) of Alb-iGP_Smarta livers (early and late disease) and Alb-iGP control livers. (E) Serum IgG levels (g/ml; n = 5–9). (F) Hepatic CD19⁺CD138⁺ plasma cells (n = 4–5). (G) Antinuclear antibody (ANA) titers (n = 4–11) and representative fluorographs. Data are shown as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (A, Mantel-Cox; B, D, and E, ANOVA; F, Fisher’s exact test).

Figure 3. Characterization of liver-infiltrating T cells in Alb-iGP_Smarta mice.

(A) Immunofluorescence of liver sections revealing infiltration of CD4 (red) and CD8 cells (green) in Alb-iGP_Smarta (top, early stage; middle, late stage) or Alb-iGP control mice (bottom). Nuclei are stained in blue (original magnification, ×100). (B) Percentage of liver-infiltrating CD4⁺ T cells in Alb-iGP Smarta and Alb-iGP control mice (n = 4–5). (C) Percentage of GP tetramer-binding CD4⁺ T cells in livers of Alb-iGP_Smarta and Alb-iGP control mice (n = 4–5). (D) Percentage of liver-infiltrating CD4⁺ T cells with activated/memory phenotype in Alb-iGP_Smarta and Smarta control mice (n = 7–13). (E) Frequencies of IFN-γ–producing CD4⁺ T cells, (F) TNF and IFN-γ coproducing CD4⁺ T cells, and (G) IL-17–producing CD4⁺ T cells in livers of Alb-iGP_Smarta and Alb-iGP control mice (n = 4–8). (H) Hepatic gene expression of IL12A and IL12B and (I) IL12rb1 and IL12rb2 in Alb-iGP_Smarta mice in early or late stage and Alb-iGP control mice (n = 4–12). Data are shown as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (B–E, Mann-Whitney; F–I, ANOVA).

Figure 4. DCs as pathogenic drivers in Alb-iGP_Smarta mice and human AIH.

Hepatocellular gene expression of GP (A) and I-A^b (B) in Alb-iGP_Smarta, Smarta, and Alb-iGP mice (n = 6–8). (C) Immunostaining of MHC II (red) and CD45 (green) in livers of Alb-iGP Smarta mice. Nuclei are stained in blue. The yellow arrow indicates a CD45⁺MHC II⁺ APC; the white arrows indicate CD45^–MHC II⁺ hepatocytes (original magnification, ×400). (D) CD11c⁺ DCs, (E) Ly6C^hiCD11b⁺ monocytes, and (F) Ly6G⁺CD11b⁺ neutrophils in livers of Alb-iGP Smarta mice at early stage as compared with Alb-iGP controls. (G) Histological analysis of liver biopsies from newly diagnosed, untreated patients with AIH, as compared with pseudo-healthy control tissue from subjects undergoing bariatric surgery (original magnification, ×100). CD11c⁺ DCs are stained in pink. Data are shown as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 (A and B, ANOVA; D–F, Mann-Whitney).

Figure 5. Histological characterization of portal ectopic lymphoid tissue in early disease livers of Alb-iGP_Smarta mice.

Segregation of CD4⁺ T cells (A) and B220⁺ B cells (B) in separate periportal zones in consecutive sections of Alb-iGP_Smarta livers. (C) CD11c⁺ DC network in periportal infiltrate of Alb-iGP_Smarta liver. (D) PNAd⁺ high endothelial venules in periportal infiltrate of Alb-iGP_Smarta liver (original magnification, ×200).

Figure 6. Quantification of Tregs in Alb-iGP_Smarta mice.

(A) Representative flow cytometry dot plots of thymic and splenic Foxp3⁺ Tregs in C57BL/6, Alb-iGP, Smarta and Alb-iGP_Smarta mice. (B) Frequencies of thymic (left) and splenic (right) CD4⁺Foxp3⁺ Tregs in the respective mouse strains (n = 4–8). (C) Representative Foxp3 immunohistochemistry of Tregs in periportal infiltrates of Alb-iGP_Smarta liver in early disease (left) or late disease (right) (original magnification, ×200). (D) Flow cytometry of antigen-specific (tetramer⁺) and nonspecific (tetramer^–) CD4⁺ Foxp3⁺ Tregs in Alb-iGP_Smarta liver in early disease (left) or late disease (right) and quantification of Treg numbers in early and late stage (n = 5). Data are shown as the mean ± SEM. **P < 0.01; ****P < 0.0001 (B, ANOVA; D, Mann-Whitney).

Figure 7. Functional characterization of Tregs in Alb-iGP_Smarta mice.

(A) Selectively reduced Foxp3 MFI in tetramer-specific vs. nonspecific CD4⁺ Foxp3⁺ Tregs in livers and spleens of Alb-iGP_Smarta mice (n = 4). (B) Reduced CNS2 element demethylation at the Foxp3 gene locus in tetramer-specific vs. nonspecific splenic CD4⁺ Foxp3⁺ Tregs purified and pooled from 10 Alb-iGP_Smarta mice. (C) Age-dependent decline of tetramer-specific vs. nonspecific CD4⁺ Foxp3⁺ Tregs (n = 5–7). (D) No age-dependent change in overall Foxp3⁺ Treg frequency among CD4⁺ T cells (n = 5–7). (E) Selectively reduced percentage of CD39-expressing cells, and (F) selectively increased percentage of IL-17 producers among tetramer-specific vs. nonspecific CD4⁺Foxp3⁺ Tregs in livers and spleens of Alb-iGP_Smarta mice (n = 4). (G) Selectively reduced suppressive function of tetramer-specific (blue) vs. nonspecific (red) splenic CD4⁺Foxp3⁺ Tregs of Alb-iGP_Smarta mice (n = 5). Data are shown as the mean ± SEM. *P < 0.05; **P < 0.01; ****P < 0.0001 (A, C, and D, Mann-Whitney; E, ANOVA).