Impaired central tolerance induces changes in the gut microbiota that exacerbate autoimmune hepatitis

Abstract

Medullary thymic epithelial cells (mTECs) induce T cell tolerance in the thymus through the elimination of self-reactive thymocytes. Commensal bacteria are also critical for shaping T cell responses in the gut and distal organs. We previously showed that mice depleted of mTECs (Traf6ΔTEC) generated autoreactive T cells and developed autoimmune hepatitis (AIH). In this report, we found that Toll-like receptor (TLR)-mediated microbial sensing on liver hematopoietic cells and the gut microbiota contributed to AIH development in Traf6ΔTEC mice. While adoptive transfer of thymic Traf6ΔTEC T cells in immune-deficient mice was sufficient for AIH development, colonization of germ-free mice with Traf6ΔTEC microbiota failed to induce AIH, suggesting that the gut microbiota contributes to but is not sufficient for AIH development. Microbiota-mediated exacerbation of AIH associated with increased numbers of hepatic Foxp3⁺ T cells and their increase was proportional to the degree of inflammation. The contribution of the gut microbiota to AIH development associated with an altered microbial signature whose composition was influenced by the qualitative nature of the thymic T cell compartment. These results suggest that aberrant selection of T cells in the thymus can induce changes in the gut microbiota that lead to exacerbation of organ-specific autoimmunity and AIH. Our results add to our understanding of the mechanisms of AIH development and create a platform towards developing novel therapeutic approaches for treating this disease.

Keywords: Autoimmune hepatitis; Central tolerance; Foxp3 T regulatory Cells; Gut microbiota; Medullary thymic epithelial cells; Toll-like receptors.

Figure 1.. Traf6ΔTEC mice exhibit upregulation of immune responses genes.

(A) H&E-stained sections of formalin-fixed, paraffin-embedded (FFPE) liver sections from WT and cKO mice visualized on a bright field scanner (NDP NanoZoomer). Scale bar represents 50μm. (B) Principal component analysis (PCoA) of all samples indicating that a disease covariate (AIH) has a significant impact on gene expression patterns compared to normal (n=5 WT and 6 cKO mice). (C) Heat map generated via unsupervised clustering of normalized data from WT and cKO mice. (D) Volcano plot showing differentially expressed genes in WT vs. cKO animals. Lines represent adjusted p values as shown. (E) Directed global significance scores showing gene expression change in different pathways. (F) Heat map showing clustering of pathway scores according to genotype and samples with similar pathway score profiles. Data analysis was performed using Nanostring Advanced nSolver software. Shown is one representative of three separate experiments.

Figure 2.. AIH inflammation in Traf6ΔTEC mice associates with upregulation of adaptive and innate/antimicrobial immune gene expression.

(A-C) Heat maps generated via unsupervised clustering of normalized data of gene-sets upregulated in cKO mice compared to WT controls. Box plots show distribution of samples within each group of mice with pathway scores determined from expression levels of gene-sets within specific pathways.

Figure 3.. TLR signaling in hematopoietic cells contributes to AIH development in Traf6ΔTEC mice.

(A) Liver and colon homogenates were prepared from approximately 8-week old WT and Traf6 cKO mice. Bacterial load was quantified by qRT-PCR using pan-bacteria primers (n=5-6 mice per group). (B) Relative expression of TLRs in sorted CD45⁺ hepatic cells analyzed by qRT-PCR. (C-D) Representative H&E-stained sections from the liver of WT and cKO mice transplanted with WT or TRIF/MyD88^−/− bone marrow (n =3-4 mice per group pooled from two experiments). Scale bar represents 50μm. Histological parameters of AIH including portal (black arrows) and lobular (black arrowheads) inflammation, interface hepatitis (white arrows) and total liver damage were determined from H&E stained sections as described in Methods and as we previously described [27]. (E). Absolute numbers of CD3-gated CD4⁺, CD8⁺, CD4⁺CD8⁺ and CD4⁺Foxp3⁺ T cells in the liver of mice in the different groups shown, were determined by flow cytometry. Statistical significance was analyzed with Student’s t test for (A) and (B) and one-way ANOVA with Tukey's Multiple Comparison Test for (D) and (E). Bars represent average ± S.E.M.

Figure 4.. The gut microbiota of Traf6ΔTEC mice contributes to development of AIH.

8-12 week old WT and Traf6ΔTEC mice were either left untreated (-ABX) or treated with antibiotics (+ABX) for 4 weeks (n = 2-15 mice per group, pooled from four separate experiments). (A-B) Representative H&E stained (scale bar represent 50μm) and pathological assessment of FFPE liver sections of untreated and antibiotic-treated cKO mice showing AIH parameters including portal (black arrows) and lobular (black arrowheads) inflammation, interface hepatitis (white arrows) and total liver damage which were determined as described in Methods. (C) Quantification of absolute numbers of liver CD3⁺, CD4⁺, CD4⁺CD8⁺ and CD4⁺Foxp3⁺ T cells in the different animal groups as shown, was determined by flow cytometry by gating on CD3⁺ T cells. (D) Absolute numbers of CD4⁺Foxp3⁺Helios⁺, CD4⁺Foxp3⁺Nrp1⁺, CD4⁺Foxp3⁺Ki67⁺ and CD4⁺Foxp3⁺GITR⁺ in the different animal groups as shown, was determined by flow cytometry by gating on CD3⁺CD4⁺Foxp3⁺ T cells. Statistical significance was analyzed by one-way ANOVA with Tukey’s Multiple Comparison Test for (B-D). Bars represent average ± S.E.M.

Figure 5.. Thymic Traf6ΔTEC CD4⁺ T cells but not the gut microbiota are sufficient for AIH development.

(A-B) Sorted CD4⁺GFP⁻ (4x10⁵) and equal numbers (5x10⁴) of CD4⁺GFP⁺ thymocytes from WT^Foxp3-GFP or Traf6ΔTEC^Foxp3-GFP mice were adoptively transferred into Rag1^−/− mice (WT>Rag1^−/−) and (cKO>Rag1^−/−) respectively. After 8 weeks, liver tissue sections were stained with H&E and scored for AIH inflammation. Total hepatic damage was determined from AIH histological parameters of portal and lobular inflammation (black arrow and arrowhead respectively) and interface hepatitis (white arrow) as described in Methods. Results are compiled from two independent experiments (n=6 mice per group). Scale bar represents 100μm. (C) Frequencies and absolute numbers of liver CD4⁺GFP⁺Foxp3⁺ T cells gated on CD3⁺ T cells from WT>Rag1^−/− and cKO>Rag1^−/− mice 8 weeks after thymocyte transfer (n=6-11 mice per group as shown). (D) Rag1^−/−, WT→Rag1^−/− and cKO→Rag1^−/− recipients were transplanted as in (A) and stool bacteria communities of recipient mice were quantified by qRT-PCR using specific primers and expressed relative to total Eubacteria. n= 6-11 mice per group. Statistical significance was analyzed using student’s t test for (B) and (C) and one-way ANOVA with Sidak test for multiple comparison for (D), Bars represent average ±S.E.M.

Figure 6.. Traf6ΔTEC mice exhibit altered microbiota composition compared to WT mice.

Stool from WT and cKO mice was collected and used for 16s rRNA sequencing analysis using Quantitative Insights into Microbial Ecology (QIIME). (A) Principal Coordinate Analysis (PCoA) plot based on unweighted UniFrac distances estimated from microbial community composition in WT and cKO (n = 10 WT and 16 cKO mice). (B) Relative abundance of bacterial communities at the taxa and genera levels. (C) Top operational taxonomic units (OTUs) as ranked by a Random Forest classifier to distinguish between cKO and WT mice (left panel) and relative abundance of each OTU in each sample (right panel). Bars indicate the average ±S.E.M, classification accuracy decrease in the Random Forest classifier if the OTU is removed. Each row represents an OTU with taxonomy annotated on the left. Results are pooled from two independent experiments.

Figure 7.. The nature of the thymic T cell compartment influences microbial composition in the gut of Traf6ΔTEC mice.

Rag1^−/− mice were adoptively transferred with either total thymocytes from WT or cKO mice (total) or sorted CD4⁺GFP⁻ and equal numbers of CD4⁺GFP⁺ thymocytes (equal) as described in Figure 5. Stool from Rag1^−/− recipient mice was collected and used for 16s rRNA sequencing analysis. (A) Principal Coordinate Analysis (PCoA) plot based on unweighted UniFrac distances from indicated groups. (B-C) Alpha diversity (phylogenetic diversity) in the total or equal thymocyte transfer models. (D-E) Top 10 OTUs as ranked by a Random Forest classifier to distinguish between Rag1^−/− mice transferred with total or equal WT (D) or cKO (E) thymocytes. Each row represents an OTU with taxonomy annotated on the left. Bars indicate the mean (+ S.D.), classification accuracy decrease in the Random Forest classifier if the OTU is removed. n = 3-9 mice per group; results are representative of two independent experiments.

Figure 8.. Working model for thymus-gut-liver-mediated regulation of murine AIH.

Homing of thymic T cells to the liver is sufficient for inducing AIH development in Traf6ΔTEC mice. TLR signaling in hepatic hematopoietic cells and the gut microbiota contribute to AIH development. The aberrant thymic T cell compartment of Traf6ΔTEC mice associates with changes in the gut microbiota that exacerbate AIH. Microbiota-derived production of hepatic Foxp3⁺ T cells is an integral part of the inflammatory response and correlates with disease severity.