TYK2 signaling promotes the development of autoreactive CD8+ cytotoxic T lymphocytes and type 1 diabetes

Abstract

Tyrosine kinase 2 (TYK2), a member of the JAK family, has attracted attention as a potential therapeutic target for autoimmune diseases. However, the role of TYK2 in CD8⁺ T cells and autoimmune type 1 diabetes (T1D) is poorly understood. In this study, we generate Tyk2 gene knockout non-obese diabetes (NOD) mice and demonstrate that the loss of Tyk2 inhibits the development of autoreactive CD8⁺ T-BET⁺ cytotoxic T lymphocytes (CTLs) by impairing IL-12 signaling in CD8⁺ T cells and the CD8⁺ resident dendritic cell-driven cross-priming of CTLs in the pancreatic lymph node (PLN). Tyk2-deficient CTLs display reduced cytotoxicity. Increased inflammatory responses in β-cells with aging are dampened by Tyk2 deficiency. Furthermore, treatment with BMS-986165, a selective TYK2 inhibitor, inhibits the expansion of T-BET⁺ CTLs, inflammation in β-cells and the onset of autoimmune T1D in NOD mice. Thus, our study reveals the diverse roles of TYK2 in driving the pathogenesis of T1D.

Fig. 1. Tyk2 deficiency suppresses autoimmune diabetes in NOD mice.

a Incidence of spontaneous diabetes in female Tyk2^+/+ (n = 14), Tyk2^+/− (n = 10), and Tyk2^−/− NOD mice (n = 13). Diabetes was defined by a non-fasting blood glucose level exceeding 250 mg/dL. b Insulitis scores of normoglycemic mice at 14w (Tyk2^+/+, n = 8; Tyk2^+/−, n = 8; Tyk2^−/−, n = 8) and 24w (Tyk2^+/+, n = 10; Tyk2^+/−, n = 8; Tyk2^−/−, n = 10). Score 0, no insulitis; Score 1, peri-insulitis; Score 2, infiltrative insulitis less than 50% of the islet area; and Score 3, infiltrative insulitis more than 50% of the islet area (Supplementary Fig. 1d). c Insulin autoantibody levels in the serum of normoglycemic mice at 14w (Tyk2^+/+, n = 17; Tyk2^+/−, n = 22; Tyk2^−/−, n = 21) and 24w (Tyk2^+/+, n = 10; Tyk2^+/−, n = 8; Tyk2^−/−, n = 16). d Incidence of diabetes in female wild type NOD.SCID mice adoptively transferred with splenocytes from normoglycemic female 14w Tyk2^+/− NOD mice (n = 7), 14w Tyk2^−/− NOD mice (n = 7), or 24w Tyk2^−/− NOD mice (n = 10). Splenocytes (1 ×10⁷) were intravenously transferred into 6–8w recipient mice. Diabetes was defined by a non-fasting blood glucose level exceeding 250 mg/dL. e Incidence of diabetes in female Tyk2^−/− NOD.SCID mice adoptively transferred with splenocytes (1 ×10⁷) from normoglycemic female 14w Tyk2^+/− NOD mice (n = 8), 14w Tyk2^−/− NOD mice (n = 6), or 24w Tyk2^−/− NOD mice (n = 8). Data represent the mean ± SEM. P-values were calculated using the log-rank test (a, d, e), and one-way ANOVA with Tukey’s posttest (b, c).

Fig. 2. Increased inflammatory signature in β-cells with age is attenuated by Tyk2 deficiency.

a Principal component analysis (PCA) of transcriptome data of purified β-cells pooled from 6w female Tyk2^+/+ (n = 6), 6w female Tyk2^−/− (n = 6), 11w female Tyk2^+/+ (n = 8), and 11w female Tyk2^−/− NOD mice (n = 9). β-cells were purified using FluoZin-3-AM and TMRE, and analyzed for gene expression profiles. b Gene ontology (GO) biological process enrichment analysis of differentially expressed genes between β-cells from 11w mice and 6w mice (11w Tyk2^+/+ vs 6w Tyk2^+/+, or 11w Tyk2^−/− vs 6w Tyk2^−/−). Selected GO terms are shown. The two-sided adjusted p-values were calculated using Benjamini-Yekutieli method (FDR p < 0.05) by GeneTrail. c GO biological process enrichment analysis of differentially expressed genes in β-cells from 11w mice (11w Tyk2^+/+ vs 11w Tyk2^−/−). Selected GO terms are shown. The two-sided adjusted p-values were calculated using Benjamini-Yekutieli method (FDR p < 0.05) by GeneTrail. d, e Gene expression profiles in purified β-cells. Data are normalized by Z score. Selected genes are shown. Asterisks (*) indicate differentially expressed genes between 11w wild type vs 11w KO mice. Daggers (†) indicate differentially expressed genes between 6w wild type vs 11w wild type mice. Double daggers (‡) indicate differentially expressed genes between 6w KO vs 11w KO mice. Criteria used for identifying differentially expressed genes were as follows: upregulated genes, Z score >2.0 and ratio <1.5; downregulated genes, Z score <−2.0 and ratio >0.66. f Protein expression levels of MHC I (H-2K^d (Tyk2^+/+, n = 9; Tyk2^+/−, n = 12; Tyk2^−/−, n = 9) and H-2D^b (Tyk2^+/+, n = 7; Tyk2^+/−, n = 9; Tyk2^−/−, n = 9)) in β-cells from 11–12w female mice. Data represent the mean ± SEM. P-values were calculated using Kruskal-Wallis test with Dunn’s posttest (f).

Fig. 3. Increased naïve CD8⁺ T cells and reduced IFN-γ-producing CD8⁺ T cells in the PLN of Tyk2-deficient NOD mice.

a Cell number (#) or frequency among CD45⁺ cells (%) of CD8⁺ T cells in the PLN and pancreas measured by flow cytometry. Pancreas; 6w (Tyk2^+/−, n = 6; Tyk2^−/−, n = 6), 11w (Tyk2^+/−, n = 5; Tyk2^−/−, n = 9), 14w (Tyk2^+/−, n = 11; Tyk2^−/−, n = 15), and normoglycemic 24w (Tyk2^−/−, n = 8) female NOD mice. PLN; 6w (Tyk2^+/−, n = 9; Tyk2^−/−, n = 7), 11w (Tyk2^+/−, n = 11; Tyk2^−/−, n = 11), 14w (Tyk2^+/−, n = 11; Tyk2^−/−, n = 11), and normoglycemic 24w (Tyk2^−/−, n = 3) female NOD mice. b Frequency of naïve (CD44^lo CD62L⁺), effector memory (CD44^hi CD62L^-), and central memory (CD44^hi CD62L⁺) CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/−, n = 6; Tyk2^−/−, n = 6). c Frequency of Ki67⁺ cells among the indicated immune cells in the PLN of 6w female mice (Tyk2^+/−, n = 11; Tyk2^−/−, n = 10). d (Left) Representative flow cytometry plots of IFN-γ and IL-17-producing CD8⁺ T cells upon PMA/iono stimulation in the PLN of 6w female mice. (Right) The graphs indicate the frequency of IFN-γ and IL-17-producing CD8⁺ T cells among CD8⁺ T cells in the PLN at the indicated ages (6w (Tyk2^+/−, n = 9; Tyk2^−/−, n = 7), 11w (Tyk2^+/−, n = 9; Tyk2^−/−, n = 10), and 14w (Tyk2^+/−, n = 14; Tyk2^−/−, n = 14)). NT not tested. Data represent the mean ± SEM. P-values were calculated using two-tailed Student’s t tests.

Fig. 4. Defective APC function of CD8⁺ rDC in Tyk2-deficient NOD mice.

a Frequency of proliferating CD44^hi Tyk2^+/− or Tyk2^−/− CTV-labeled 8.3 CD8⁺ T cells in the PLN of the indicated female recipient mice after 5d or 10d post transfer: 4 ×10⁶ labeled naïve 8.3 CD8⁺ T cells were transferred into 6–8w female recipient mice. “24w” indicates the age of the recipient mice. (Tyk2^+/− → Tyk2^+/−, n = 6; Tyk2^−/− → Tyk2^+/−, n = 8; Tyk2^+/− → Tyk2^−/−, n = 4; Tyk2^−/− → Tyk2^−/−, n = 4; Tyk2^+/− → 24w Tyk2^−/−, n = 3; Tyk2^−/− → 24w Tyk2^+/−, n = 5; Tyk2^−/− → Tyk2^+/− 10d, n = 6; Tyk2^+/− → Tyk2^−/− 10d, n = 6.) b Frequency of IGRP peptide-specific CD8⁺ T cells in the PLN of female mice (6w (Tyk2^+/−, n = 14; Tyk2^−/−, n = 11), 11w (Tyk2^+/−, n = 10; Tyk2^−/−, n = 12). c (Left) Representative flow cytometry plots of resident DC (rDC) in the PLN of 6w mice. rDCs are gated on MHC II^mid CD11c^hi cells (Supplementary Fig. 5a, b). (Right) The frequencies of CD8⁺ rDC and CD11b⁺ rDC among rDC are shown (Tyk2^+/+, n = 4; Tyk2^+/−, n = 6; Tyk2^−/−, n = 4). d Protein expression levels of H-2K^d and H-2D^b in CD8⁺ rDC in the PLN of 6w female mice (Tyk2^+/+, n = 4; Tyk2^+/−, n = 6; Tyk2^−/−, n = 4). e Volcano plot of transcriptome data in CD8⁺ rDC in the PLN of 6w female mice (Tyk2^+/+, n = 4; Tyk2^−/−, n = 4). The CD8⁺ rDC were sorted and RNA-sequencing analysis was performed. Pairwise comparisons of differentially expressed genes were performed using DESeq2 and used to create the plot (FDR <  0.1, logFC > |0.58|). ISGs and genes associated with APC function are highlighted. f (Left) Representative histogram and (right) proliferation of CTV labeled naïve 8.3 CD8⁺ T cells after co-culture with CD8⁺ rDC and 1 ng/mL of IGRP peptide for 3d (Tyk2^+/+, n = 4; Tyk2^−/−, n = 5). Data represent the mean ± SEM. P-values were calculated using two-tailed Student’s t test (a, b, and f) and one-way ANOVA with Dunnett’s posttest (c, d).

Fig. 5. Defective development of CD8⁺ T-BET⁺ CTLs in the PLN and effector function of CD8⁺ CTLs in the pancreas of Tyk2-deficient NOD mice.

a (Left) Representative flow cytometry plots and (right) graph indicate the frequency of CD44^hi T-BET⁺ polyclonal cells among CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/+, n = 6; Tyk2^+/−, n = 12; Tyk2^−/−, n = 7). b Frequency of CD44^hi T-BET⁺ cells among IGRP-specific CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/−, n = 8; Tyk2^−/−, n = 6). c Protein expression levels of T-BET in CD44^hi T-BET⁺ IGRP-specific CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/−, n = 9; Tyk2^−/−, n = 6). Frequency of CD44^hi T-BET⁺ cells (d) and protein expression levels of T-BET (e) in unstimulated or ex vivo-activated Tyk2^+/− or Tyk2^−/− CD8⁺ T cells with anti-CD3/CD28 beads and IL-2 in the presence or absence of IL-12 for 3d (Tyk2^+/−, n = 4; Tyk2^−/−, n = 4). f (Left) Representative flow cytometry plots and (right) graph indicate the frequency of CD44^hi CXCR3⁺ cells among polyclonal CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/+, n = 7; Tyk2^+/−, n = 5; Tyk2^−/−, n = 7). g Frequency of CD44^hi CXCR3⁺ cells among IGRP-specific CD8⁺ T cells in the PLN of 6w female mice (Tyk2^+/−, n = 7; Tyk2^−/−, n = 5). h PCA of transcriptome data of purified CD44^hi IGRP-specific CD8⁺ T cells in the PLN (Tyk2^+/+, n = 3; Tyk2^−/−, n = 3) and IGRP-specific CD8⁺ T cells in the pancreas (Tyk2^+/+, n = 3; Tyk2^−/−, n = 3). Gene expression profiles in the purified cells were analyzed by RNA-seq analysis. i Box plots show the pattern of differentially expressed genes in IGRP-specific CD8⁺ T cells in the PLN (Tyk2^+/+, n = 3; Tyk2^−/−, n = 3) and pancreas (Tyk2^+/+, n = 3; Tyk2^−/−, n = 3). Likelihood ratio tests were used to determine differentially expressed genes between the genotypes (FDR < 0.1, fold change >1.5). Selected genes are highlighted on the right side of each box plots. Boxes and whiskers represent the median and the lower and upper hinges correspond to the 25th and 75th percentiles. The whiskers extend from the hinges to the values no further than 1.5 times the inter-quartile ranges from each box hinge. Data represent the mean ± SEM. P-values were calculated using one-way ANOVA with Dunnett’s posttest (a, f) and two-tailed Student’s t test (b–e, and g).

Fig. 6. Loss of Tyk2 impairs the cytotoxic activity of CD8⁺ CTLs.

a Protein expression levels of GZMB in the unstimulated 8.3 CD8⁺ T cells, ex vivo-activated 8.3 CD8⁺ T cells, or ex vivo-activated 8.3 CD8⁺ T cells cocultured with NIT-1 cells (Tyk2^+/−, n = 6, Tyk2^−/−, n = 6). Purified 8.3 CD8⁺ T cells were activated with IL-2, IL-12, and anti-CD3/28 beads for 3d. For coculture, ex vivo-activated 8.3 CD8⁺ T cells were cultured with NIT-1 cells for 24 h (CTLs:NIT-1 = 1:1). b Protein expression levels of GZMB in Tyk2^+/− 8.3 CTLs after 24 h of coculture with NIT-1 cells in the presence of IFNAR1 blocking antibody (n = 7), IFN-β neutralizing antibody (n = 7), or IFN-α neutralizing antibody (n = 7). Purified Tyk2^+/− 8.3 CD8⁺ T cells were activated with IL-2, IL-12, and anti-CD3/28 beads for 3d, and cultured with NIT-1 cells. c Gene expression levels of Ifnb1 in NIT-1 cells after coculture with ex vivo-activated 8.3 CD8⁺ T cells (0 h, n = 4; 1 h, n = 5; 2 h, n = 5; 6 h, n = 4; 12 h, n = 3). d The killing capacity of ex vivo-activated 8.3 CTLs against NIT-1 cells at the indicated ratios. Purified 8.3 CD8⁺ T cells were activated with IL-2 and CD3/28 beads in the presence or absence of IL-12 for 3d. NIT-1 cells were cultured with activated 8.3 CD8⁺ T cells for 24 h, and then the survival of NIT-1 cells was analyzed (1:0.1, n = 4; 1:1, n = 3; 1:2, n = 3). Data represent the mean ± SEM. P-values were calculated using one-way ANOVA with Dunnett’s posttest (c) and two-tailed Student’s t test (a, b, and d).

Fig. 7. BMS-986165 prevents the development of autoimmune T1D.

a (Left) Representative histogram and (right) protein expression levels of T-BET in stimulated Tyk2^+/+ CD8⁺ T cells. Purified Tyk2^+/+ CD8⁺ T cells were activated with IL-2, IL-12, and anti-CD3/28 beads for 3d in the presence or absence of BMS-986165 (unstimulated control, n = 3; stimulated cells, n = 4). b The β-cell killing capacity of ex vivo-activated Tyk2^+/+ 8.3 CD8⁺ T cells. Purified 8.3 CD8⁺ T cells were primed with IL-2, IL-12, and anti-CD3/28 beads for 3d in the presence or absence of BMS-986165. NIT-1 cells were cultured with activated 8.3 CD8⁺ T cells for 24 h, and then the survival of NIT-1 cells was analyzed (n = 5). c Protein expression levels of CXCR3 in stimulated Tyk2^+/+ CD8⁺ T cells. Ex vivo activated CD8⁺ T cells with IL-2, IL-12, and anti-CD3/28 beads for 3d in the presence or absence of BMS-986165 were re-cultured without BMS-986165 for 48 h and the induction of CXCR3 was analyzed (unstimulated control and stimulated cells without BMS-986165, n = 3; stimulated cells with BMS-986165, n = 4). Day 0 data indicate the CXCR3 expression levels just after ex vivo stimulation. Gene expression levels of Cxcl10 (d) and Isg15 (e) in NIT-1 cells stimulated with 100 U/mL of IFN-β or 0.1 ng/mL of IFN-γ in the presence or absence of BMS-986165 for 3 h (n = 3). f Incidence of diabetes in vehicle (n = 6) or BMS-986165-treated (n = 6) wild type female NOD mice for 4w from 6w of age. Mice were dosed by oral gavage once daily with BMS-986165 30 mg/kg in vehicle (EtOH:TPGS:PEG300, 5:5:90) or vehicle alone. Diabetes was defined by a non-fasting blood glucose level exceeding 250 mg/dL. g Incidence of diabetes in vehicle (n = 6) or BMS-986165-treated (n = 4) wild type female NOD mice for 4w from 12w of age. Data represent the mean ± SEM. P-values were calculated using one-way ANOVA with Dunnett’s posttest (a–e) and log-rank test (f, g).