Microbial antigen mimics activate diabetogenic CD8 T cells in NOD mice

Abstract

Both animal model and human studies indicate that commensal bacteria may modify type 1 diabetes (T1D) development. However, the underlying mechanisms by which gut microbes could trigger or protect from diabetes are not fully understood, especially the interaction of commensal bacteria with pathogenic CD8 T cells. In this study, using islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP)-reactive CD8 T cell receptor NY8.3 transgenic nonobese diabetic mice, we demonstrated that MyD88 strongly modulates CD8(+) T cell-mediated T1D development via the gut microbiota. Some microbial protein peptides share significant homology with IGRP. Both the microbial peptide mimic of Fusobacteria and the bacteria directly activate IGRP-specific NY8.3 T cells and promote diabetes development. Thus, we provide evidence of molecular mimicry between microbial antigens and an islet autoantigen and a novel mechanism by which the diabetogenicity of CD8(+) T cells can be regulated by innate immunity and the gut microbiota.

Figure 1.

MyD88 deficiency has different effect on diabetes development. (A–E) Individual TLR- or MyD88^−/−NY8.3NOD mice were generated by breeding different TLR^−/− or MyD88^−/−NOD mice with NY8.3 NOD mice. Diabetes development was observed, and data were pooled from at least five independent experiments. (A) TLR2^−/−NY8.3. (B) TLR4^−/−NY8.3. (C) TLR5^−/−NY8.3. (D) TLR9^−/−NY8.3. (E) MyD88^−/−NY8.3. Wilcoxon test for survival was used for analysis of diabetes incidence. *, P < 0.05; **, P < 0.01; ***, P < 0.001. F, female. M, male.

Figure 2.

MyD88 deficiency has no effect on T cell development in thymus and regulatory T cell differentiation. (A) Thymocytes from 6–7-wk-old MyD88^−/−NY8.3NOD and WT NY8.3NOD mice (sex matched) were stained with fluorochrome-conjugated anti-TCRβ, anti-CD4, and anti-CD8 antibodies. n = 8–10 mice/group from two independent experiments. (B) CD8⁺ T cells in the spleen of MyD88^−/−NY8.3NOD mice. The experiment was performed in three independent experiments, and two-tailed Mann-Whitney test was used for statistical analysis. *, P < 0.05. (C) The absolute number of infiltrated CD4⁺ and CD8⁺ T cells in the islets of MyD88^−/−NY8.3NOD mice after gating with TCRβ-positive cells is shown. The data were from two independent experiments. (D) Natural regulatory T cells in MyD88^−/−NY8.3NOD mice. n ≥ 6 mice/group/experiment from two independent experiments. PLN, pancreatic LN. Data are mean ± SEM.

Figure 3.

MyD88 deficiency promotes highly pathogenic NY8.3 CD8⁺ T cells. (A) CD69 expressing CD8⁺ T cells in MyD88^−/−NY8.3 mice. n = 4 mice/group/experiment from more than three experiments. PLN, pancreatic LN. (B) Mean fluorescence intensity (MFI) of CD62L and CD44 of gated splenic CD8⁺ T cells from MyD88^−/−NY8.3 and WT NY8.3 mice was determined by flow cytometric analysis from three independent experiments. (C) Splenocytes from MyD88^−/−NY8.3 or WT NY8.3 mice were tested upon antigen stimulation by [³H]thymidine incorporation assay. n = 3 mice/group/experiment from more than three experiments. Two-way ANOVA was used for the comparison with Bonferroni correction. (D) Granzyme B–expressing CD8⁺ T cells in the spleen of MyD88^−/−NY8.3NOD mice stimulated by anti-CD3 and anti-CD28 followed by ICC staining. Data are from five independent experiments. (E) Diabetes incidence after adoptive transfer. 10⁶ purified IGRP-stimulated splenic CD8⁺ T cells were injected i.v. into recipient NOD mice in two independent experiments. Wilcoxon test for survival was used for analysis. (A, B, and D) Two-tailed Student’s t test was used for analysis. Data are mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

MyD88 deficiency results in distinct gut microbiota in NY8.3NOD mice. (A–C) Fecal samples from WT NY8.3NOD and MyD88^−/−NY8.3NOD mice were used for taxonomic analysis by 16S rRNA sequencing. The taxonomic compositions of gut bacteria were shown in phylum (left, pies; right, bars; A) family (B), and genus (C). (A and C) Two-way ANOVA with Bonferroni correction was used. (D) α-Diversity of gut microbiota of WT NY8.3NOD and MyD88^−/−NY8.3NOD mice was shown by Shannon evenness and Chao richness. Two-tailed Student’s t test was used. (E) β-Diversity, shown by principal component analysis (PCoA; unweighted) of taxonomic families of gut microbiota. ANOSIM was used for statistical analysis. Two independent sequencing experiments with different cohorts were performed, and one representative (A, B, C, and E) and the pooled data (D) are shown. Data are mean ± SEM. *, P < 0.05; ***, P < 0.001. M, male.

Figure 5.

Reduced intestinal antimicrobial peptide expression in NY8.3NOD mice in the absence of MyD88. (A and B) Gene expression for the intestinal antimicrobial peptides. Fragments (∼3 cm) of the small intestines from ileum and large intestine adjacent to the cecum were harvested from 6–7-wk-old sex-matched WT NY8.3NOD and MyD88^−/−NY8.3NOD mice. After intensive washing with sterile PBS, RNA was extracted from the fragments using TRIzol, followed by reverse transcription. qPCR was performed to determine the mRNA levels of antimicrobial peptides, which were normalized to the housekeeping gene GAPDH. (A) Small intestine. (B) Large intestine. Five to eight mice/group from two independent experiments are shown. One-tailed Mann-Whitney test was used for statistical analysis. Data are shown as mean ± SEM. *, P < 0.05; **, P < 0.01. CRP, C-reactive protein.

Figure 6.

Manipulation of gut microbiota protects from diabetes development in MyD88^−/−NY8.3 mice. (A) Diabetes incidence in MyD88^−/−NY8.3NOD mice after 4-wk-old MyD88^−/−NY8.3NOD mice were cohoused with an equal number and age- and sex-matched NOD mice. F, female; M, male. (B) Diabetes development of MyD88^−/−NY8.3NOD mice after fecal OT. The fresh gut bacteria (∼10⁸–10⁹ culturable bacteria/mouse) from diabetes-free cohoused MyD88^−/−NY8.3 mice (from A) were fed orally to sex-matched 4-wk-old naive MyD88^−/−NY8.3NOD mice. (C) Insulitis of MyD88^−/−NY8.3NOD mice after cohousing with WT NOD mice or OT (from A and B). More than120 islets from five to six treated mice/group were examined. Representative pancreatic sections are shown on the left, and the summary of the insulitis scores is shown on the right. χ² test was used for analysis. (D) α-Diversity of gut microbiota is shown by Chao richness. Fecal samples from the control, cohoused, and OT MyD88^−/−NY8.3NOD mice were used for taxonomic analysis by 16S rRNA sequencing. One- or two-tailed Mann-Whitney test was used for analysis. Data are shown as individual samples. (E) β-Diversity is shown. Assessment of the differences in the bacterial composition of the gut microbiota in the mice that were treated either by cohousing or OT (red or green symbols) compared with the control mice (blue symbols) is shown by principal component analysis (PCoA). ANOSIM was used for statistical analysis. (F and G) Comparison of the abundance of gut bacteria in the control, cohoused, and OT MyD88^−/−NY8.3NOD mice is shown by phylum (F) and family (G). Two-way ANOVA was performed with Bonferroni correction. (H) Diabetes incidence in the progeny of MyD88^−/−NY8.3NOD mice that were protected from diabetes after fecal OT (shown in B). (A, B, and G) The data were pooled from at least two independent experiments. (D–G) The results are from one of two independent experiments. Data are shown as mean ± SEM. **, P < 0.01; ***, P < 0.001.

Figure 7.

Microbial peptide mimics can activate NY8.3 CD8⁺ T cells in vitro and induce diabetes in vivo. (A) NY8.3 CD8⁺ T cell response to microbial peptide mimics shown in proliferation assay by [³H]thymidine incorporation. The experiment was repeated at least four times. (B) Representative FACS plots for expression of granzyme B (GzmB) and IFN-γ in NY8.3 CD8⁺ T cells stimulated with mimic peptides or IGRP_206–214 peptide (1 µg/ml). Student’s t test was used for statistical analysis to compare individual peptide mimics with IGRP_206–214 peptide. Data are representative of four independent experiments. n ≥ 10. (C) MHC class I blocking assay. Two-tailed Student’s t test was used for statistical analysis. n = 3/group/experiment, and the experiment was performed three times. (D) MHC restriction assay. n = 3/group/experiment, and the experiment was performed twice. Two-way ANOVA with Bonferroni correction was used for analysis. (E) Diabetes development induced by NY8.3 CD8⁺ T cells activated with microbial peptide mimics from two independent experiments. Log-rank test for survival was used. (F) CTL activity of CD8⁺ T cells induced by W15944 peptide. Two-way ANOVA with Bonferroni correction was used for analysis. n = 4/group/experiment, and the experiment was performed three times. (G) Pathogenic effect of W15944-stimulated NY8.3 CD8⁺ T cells. Log-rank test for survival was used for statistical analysis. The results are from two independent experiments (n = 5/group/experiment). (H–J) MyD88^−/−NY8.3 and WT NY8.3NOD CD8 T cells activated by microbial peptide mimics induced diabetes by adoptive transfer into irradiated NOD mice. The experiments were repeated twice. (H) W15946. (I) W15948. (J) W15944. Log-rank test for survival was used for analysis. (K) Incidence of diabetes in WT NY83.NOD mice after immunization with mimic peptide from L. goodfellowii. The data were pooled from two independent experiments. Log-rank test for survival was used for analysis. (L) Response of insulin-specific G9 T cells to the microbial peptides. (Left) Granzyme B. (Right) IFN-γ. n = 2/group/experiment, and the experiment was performed twice. One-way ANOVA was used for statistical analysis. Data are mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; *****, P < 0.00001.

Figure 8.

Increased abundance of Fusobacteria is associated with diabetes development. (A) Abundance of Fusobacteria and L. goodfellowii in WT NY8.3NOD and MyD88^−/−NY8.3NOD mice. Two-tailed Mann-Whitney test (fecal samples) or two-tailed unpaired Student’s t test with Welch’s correction (oral samples) was used. n = 7–16 mice/group from two independent experiments. Data are mean ± SEM. (B) Abundance of Fusobacteria in fecal samples from the control, cohoused, and OT MyD88^−/−NY8.3NOD mice. Repeated measures ANOVA with Tukey’s posthoc test was used for analysis. (C) Abundance of Fusobacteria in diabetic and nondiabetic female NOD mice. Student’s t test was used. (D) Bacterial DNA was extracted from fecal samples of female NOD mice every two weeks from 6 wk old until diabetes onset. Results from three time points are shown. Repeated measures ANOVA with Tukey’s posthoc test was used. (B–D) Data were pooled from at least two independent experiments. *, P < 0.05; ***, P < 0.001.

Figure 9.

L. goodfellowiican activate NY8.3 CD8⁺ T cells in vitro, and supercolonization accelerates diabetes in naive young WT NY8.3NOD mice in vivo. (A) Expression of IFN-γ (ICC staining) in NY8.3 CD8⁺ T cells stimulated with the L. goodfellowii lysate. (B) MIP-1β secretion (ELISA) from NY8.3 CD8⁺ T cells stimulated with the translated rMgt protein. (C) MIP-1β secretion (ELISA) from NY8.3 CD8⁺ T cells stimulated with the purified truncated rMgt protein. (D) Inflammatory cytokine induction (ICC staining) in NY8.3 CD8⁺ T cells stimulated by APCs pulsed with heat-killed bacteria lysates. (E) MIP-1β (ELISA) and IFN-γ production (ICC staining) of NY8.3 CD8⁺ T cells stimulated with the APCs from NOD mice supercolonized with L. goodfellowii or control B. theta. (F) Secreted IFN-γ (ELISA) from NY8.3 CD8⁺ T cells stimulated with APCs from NOD mice supercolonized with L. goodfellowii or control B. theta. (G) 10⁶ splenocytes from NY8.3 mice were stimulated with different heat-killed bacterial strains. (Left) IFN-γ expression (ICC staining). (Right) IFN-γ secretion (ELISA). One-way ANOVA was used for statistical analysis. n = 2/group/experiment, and the experiment was performed twice. (H) Diabetes incidence in NY8.3NOD mice after supercolonization with L. goodfellowii or control B. theta. Log-rank test for survival was used for analysis. The results are combined from two independent experiments (n = 5–8/group/experiment). (A–F) Two-tailed Student’s t test was used. n = 3–6/group/experiment from three independent experiments. Data are shown as mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.