T-Cell–Specific PTPN2 Deficiency in NOD Mice Accelerates the Development of Type 1 Diabetes and Autoimmune Comorbidities

Autoimmune diseases encompass a broad range of heterogenous and complex disorders. They include systemic disorders, such as Sjögren syndrome, systemic lupus erythematosus, and rheumatoid arthritis, and organ-specific diseases, such as Graves disease, Hashimoto thyroiditis, Crohn disease, and type 1 diabetes (1). Although considerable progress has been made in understanding the cellular basis of immune tolerance and the development of autoimmunity, the molecular mechanisms that underpin most autoimmune disorders remain incompletely understood. Most autoimmune diseases are associated with specific alleles of the MHC locus, but genome-wide association studies (GWAS) have identified >200 non-MHC loci for which alleles are associated with one or more autoimmune diseases (1). Most polymorphisms identified by GWAS are noncoding variants that are predicted to modulate gene expression, with several implicated in functional pathways that influence T- and B-cell activation (1).

GWAS have identified PTPN2 (encoding protein tyrosine phosphatase nonreceptor type 2), also known as T-cell protein tyrosine phosphatase, as an important non-MHC locus gene for autoimmunity (2–8). Noncoding PTPN2 single nucleotide polymorphisms (SNPs) have been associated with autoimmune disorders, including Crohn disease, rheumatoid arthritis, and type 1 diabetes (2–8). For example, the type 1 diabetes–associated PTPN2 SNP rs1893217 has been linked with a 40% decrease in PTPN2 expression in CD4⁺CD25⁺ regulatory T cells (T_regs) and CD4⁺CD45RO⁺ memory T cells (9). More recently, coding region PTPN2 variants that affect mRNA stability or alter protein structure have been associated with early-onset type 1 diabetes (10), whereas noncoding PTPN2 SNPs linked with Crohn disease, ulcerative colitis, rheumatoid arthritis, and type 1 diabetes have been reported to decrease PTPN2 in colonic lamina propria fibroblasts in patients with Crohn disease (11,12). Thus, PTPN2 SNPs resulting in decreased PTPN2 expression might determine the course of autoimmunity through influencing both immune and nonimmune cell populations.

Studies in mice lacking PTPN2 have identified critical roles for this regulatory enzyme in T-cell–driven immunity, inflammation, and metabolism (13,14). Changes in these processes are also frequently seen in autoimmune and inflammatory diseases where PTPN2 polymorphisms are associated with disruption of T-cell tolerance (15). PTPN2 dephosphorylates and inactivates the SRC family kinases LCK and FYN to tune T-cell receptor (TCR) signaling (15–17) and influence thymocyte development and peripheral T-cell responses to low-affinity self-antigens to prevent inappropriate T-cell activation (15,18,19). In addition, PTPN2 dephosphorylates and inactivates Janus-activated kinases (JAKs) 1 and 3 (20) and STAT1, 3, and 5 to influence T-cell responsiveness to various cytokines, including interferon (IFN)γ and interleukin (IL)-2, -6, -7, -15, and -21 (15,17–19,21). Beyond its role in T cells, PTPN2 might affect the activation of innate and adaptive immune cells, including B cells, dendritic cells, and myeloid cells (20,22,23), as well as stromal cells (24) to influence disease progression. Consistent with this, Ptpn2^−/− mice that develop wasting disease succumb within weeks of birth (13,14). Similarly, the inducible deletion of PTPN2 in the hematopoietic compartment of adult C57BL/6 mice (Mx1-Cre;Ptpn2^fl/fl) promotes systemic inflammation and overt autoimmunity within 4 weeks of PTPN2 deletion, with mice exhibiting dermatitis, glomerulonephritis, pancreatitis, and overt liver disease (22). Beyond the effector/memory T-cell phenotype, hematopoietic PTPN2 deficiency results in the accumulation of inflammatory monocytes and B cells and germinal center formation linked with increased T-follicular-helper (T_FH) cells, germinal center B cells, and IgG-secreting B cells (22). Therefore, PTPN2 deficiency might not only affect T-cell function but also drive the pathogenic accumulation of T_FH cells and high-affinity antigen-specific IgG-secreting B cells to promote autoimmunity.

Mice lacking PTPN2 in the hematopoietic compartment develop insulitis (infiltration of lymphocytes in pancreatic islets of Langerhans) (22) reminiscent of early-stage disease in humans with type 1 diabetes. However, this phenotype does not result in type 1 diabetes in C57BL/6 mice, even when PTPN2 is additionally deleted in pancreatic β-cells (22). Similarly, these mice do not develop inflammatory bowel disease or inflammatory arthritis (22), diseases that are often associated with PTPN2 loss-of-function SNPs (3,5–8). Therefore, although deficiencies in T cells or the hematopoietic system may result in a loss of tolerance and the development of autoimmunity in C57BL/6 mice, the extent to which PTPN2 deficiency in T cells and/or other immune cells contributes to the onset, maintenance, and severity of human autoimmunity remains unclear.

To explore the influence of PTPN2 deficiency in T cells on autoimmunity, especially type 1 diabetes, we backcrossed the Lck-Cre and Ptpn2^fl/fl loci, originally generated in C57BL/6 mice, onto the NOD background. Unlike other mouse models, NOD mice develop spontaneous autoimmune type 1 diabetes similar to that seen in humans, sharing many of the diabetes susceptibility loci and autoantigens that contribute to the development of type 1 diabetes (25). PTPN2 deficiency in T cells in NOD mice was shown to markedly accelerate the onset and increase the cumulative incidence of type 1 diabetes. The development of inflammatory/autoimmune comorbidities was also accelerated. Disease progression is associated with the infiltration of cytotoxic CD8⁺ T cells, along with CD4⁺ T-helper type 1 (T_H1) and T_FH polarization, as detected in human type 1 diabetes (26). These findings suggest that PTPN2 deficiency in T cells might be sufficient to promote the development of type 1 diabetes and autoimmunity in genetically susceptible individuals.

Lck-Cre;Ptpn2^fl/fl.C57BL/6J mice (15) were backcrossed onto the NOD/Lt background for 11 generations. A genome-wide screen was performed by the Australian Genome Research Facility using the iPLEX GOLD chemistry and the Sequenom MassARRAY spectrometer for SNP genotyping. Data were analyzed using GeneChip Targeted Genotyping Analysis Software. Lck-Cre;Ptpn2^fl/fl;NOD and Ptpn2^fl/fl;NOD littermate mice were maintained on a 12-h light-dark cycle in a temperature-controlled specific pathogen–free high-barrier facility (Monash University Animal Research Laboratory) with free access to food and water. Age- and sex-matched littermates were fed a standard chow diet (8.5% fat, Barastoc; Ridley AgriProducts, Melbourne, Victoria, Australia).

Recombinant mouse IL-2, IL-6, IL-12, IFNγ, and tumor growth factor-β were purchased from PeproTech. Hamster α-mouse CD3ε (145-2C11) and hamster α-mouse CD28 (37.51) were purchased from BD Biosciences. Mouse α-tubulin (Ab-5) and mouse α-GAPDH were purchased from Sigma-Aldrich. Rabbit α-phospho-Stat1 (Tyr701, clone 58D6) and polyclonal rabbit α-Stat1 were purchased from Cell Signaling Technology. Polyclonal rabbit α-phospho-Stat4 (Tyr693) was from Thermo Fisher Scientific.

Percoll was purchased from GE Healthcare and DNase I from Sigma-Aldrich. Complete T-cell medium (RPMI medium supplemented with 10% [v/v] FBS, l-glutamine [2 mmol/L], penicillin [100 units/mL]/streptomycin [100 μg/mL], nonessential amino acids, sodium pyruvate [1 mmol/L], HEPES [10 mmol/L], and 2-mercaptoethanol [50 μmol/L]) was purchased from Thermo Fisher Scientific. Dulbecco PBS (D-PBS), DMEM, and Hanks’ balanced salt solution were from Thermo Fisher Scientific, and collagenase IV was from Worthington.

Single-cell suspensions from thymus, spleen, inguinal lymph nodes, pancreatic lymph nodes, and mesenteric lymph nodes were obtained by gently pressing the tissue through a 40-μm Falcon cell strainer (BD Biosciences). Erythrocytes were removed by incubating cells with Red Blood Cell Lysing Buffer Hybri-Max for 7 min at room temperature. Cell counts were determined with the Z2 Coulter Counter (Beckman Coulter). For surface staining, cells (1 × 10⁶/10 μL) were resuspended in D-PBS supplemented with 2% (v/v) FBS and stained in 96-well microtiter plates (BD Falcon) for 20 min on ice. Cells were washed twice and resuspended in D-PBS/2% FBS and analyzed using an LSR II Fortessa Symphony (BD Biosciences) or CyAn Advance Digital Processing analyzer (Beckman Coulter) or purified using an Influx sorter (BD Biosciences). Data were analyzed using FlowJo version 8.7, version 9.4, or version 10.2 (Tree Star Inc.) software.

For detection of intracellular FoxP3, the Foxp3/Transcription Factor Staining Buffer Set (eBioscience) was used according to the manufacturer’s instructions. For the detection of intracellular cytokines, the BD Cytofix/Cytoperm Fixation/Permeabilization Kit was used according to the manufacturer’s instructions. Serum cytokines were quantified with the BD Cytometric Bead Array Mouse Inflammation Kit according to the manufacturer’s instructions.

Glycosuria was determined using Diastix Reagent Strips for Urinalysis (Bayer). Mice were scored as diabetic after two positive readings (urine glucose ≥55 mmol/L) 48 h apart. This was confirmed by measuring blood glucose levels that were found to be >18 mmol/L in each case. Pancreata were fixed in formalin and processed for histological analysis (hematoxylin and eosin) and scored for the degree of insulitis (grades 0–4). Grade 0 represents no infiltrate, grade 1 periductal accumulation of mononuclear cells, grade 2 circumferential accumulation of mononuclear cells, grade 3 intraislet infiltration, and grade 4 severe structural derangement and complete β-cell loss.

Pancreata from 5-week-old prediabetic Lck-Cre;Ptpn2^fl/fl. NOD and Ptpn2^fl/fl.NOD littermate mice were harvested and pancreatic lymph nodes removed. Pancreata were digested in DMEM containing 1 mg/mL collagenase IV (Worthington) at 37°C for 15 min, and single-cell suspensions were processed for flow cytometry.

CD8⁺CD44^loCD62L^hi were purified from a single-cell suspension obtained from pooled splenocytes and lymph node cells from 5-week-old prediabetic Lck-Cre;Ptpn2^fl/fl.NOD and Ptpn2^fl/fl.NOD littermate control mice using the naive CD8⁺ T Cell Isolation Kit (Miltenyi Biotec) according to the manufacturer’s instructions with the autoMACS Separator (Miltenyi Biotec). Naive CD8⁺CD44^loCD62L^hi (1 × 10⁷) were adoptively transferred into 28-day-old female NOD mice, and diabetes incidence was monitored.

Serum IgM was quantified with the mouse IgM ELISA Ready-SET-Go! kit (eBioscience) according to the manufacturer’s instructions. To detect serum IgG subsets, MaxiSorp ELISA Plates (Nunc) were coated with goat α-mouse IgG (625 ng/mL) in ELISA coating buffer overnight at 4°C. Plates were washed five times and blocked in D-PBS supplemented with 3% (w/v) BSA for 2 h at room temperature. Serum (IgM 1:5 serial dilutions; IgG1, IgG2a, IgG2b, and IgG3 1:10 serial dilutions) was added, and plates were incubated for 1 h at room temperature. Plates were washed five times, and biotinylated α-IgM, α-IgG1, α-IgG2a (to detect α-IgG2c), α-IgG2b, or α-IgG3 (1:1,000) diluted in D-PBS/1% (w/v) BSA was added. Samples were incubated for 1 h at room temperature. Plates were washed five times and incubated with streptavidin-horseradish peroxidase (1:10,000 in D-PBS/1% [w/v] BSA) for 30 min followed by seven washes. 3,3′,5,5′-tetramethylbenzidine was added, and samples were incubated for 15 min at room temperature in the dark. The reaction was stopped with 0.5 mol/L sulfuric acid, and absorbance was read at 450 nm.

ELISA plates (Costar) were coated with human insulin (10 μg/mL) (Actrapid; Novo Nordisk) overnight at 4°C. Plates were blocked with PBS supplemented with 2% BSA for 2 h at room temperature and then incubated with sera from Ptpn2^fl/fl.NOD, Lck-Cre;Ptpn2^fl/fl.NOD, and Lck-Cre;Ptpn2^fl/+.NOD or C57BL/6 mice (1:10 dilution) for 2 h at room temperature. Plates were washed four times and further incubated with biotinylated anti-mouse IgG1 (1:10,000 dilution) (Abcam) for 30 min at room temperature. After four washes, streptavidin-horseradish peroxidase (BioLegend) was added, and plates were incubated for 15 min at room temperature. Plates were washed five times, 3,3′,5,5′-tetramethylbenzidine was added, and absorbance was measured at 450 nm using a POLARstar (BMG Labtech) microplate reader. Each sample was run in duplicate, and absorbance (450 nm) of the negative control sample (C57BL/6 serum) was subtracted from absorbance of the test sample to calculate the true absorbance value for each test sample.

Statistical analysis was performed using the nonparametric two-tailed Mann-Whitney U test or the Student t test. Statistical analyses on Kaplan-Meier estimates were performed using a log-rank (Mantel-Cox) test with 1 df. For all tests, P < 0.05 was considered significant.

All experiments were performed in accordance with the National Health and Medical Research Council Australian code of practice for the care and use of animals. All protocols were approved by the Monash University School of Biomedical Sciences animal ethics committee (ethics number MARP/2012/124) and the U.K. home office (project license PB3E4EE13).

To determine the extent to which PTPN2 deficiency in T cells might contribute to the development of type 1 diabetes, we introduced the Lck-Cre and Ptpn2^fl/fl loci from C57BL/6 mice (15) onto the NOD/Lt (NOD) genetic background by backcrossing for 11 generations. Lck-Cre;Ptpn2^fl/+.NOD mice were thereon bred with NOD wild-type mice to produce the Ptpn2^fl/+.NOD and Lck-Cre;Ptpn2^fl/+.NOD offspring that were interbred. DNA from the 11th generation was genotyped for 597 SNPs by the Australian Genome Research Facility using iPLEX GOLD chemistry and the Sequenom MassARRAY spectrometer for SNP genotyping and analyzed by the GeneChip Targeted Genotyping System Software. Ptpn2^fl/+.NOD mice had a contaminating C57BL/6-derived interval between and including rs13483413 and rs3656892 (∼2 Mb) that encompasses the floxed Ptpn2 locus on chromosome 18, whereas Lck-Cre;Ptpn2^fl/+.NOD mice had an additional contaminating C57BL/6-derived interval on chromosome 15 between and including rs13482618 and rs13482719 (∼28 Mb) that encompasses the Lck-Cre transgene locus. These C57BL/6 intervals are unavoidable, but previous NOD × C57BL/6 outcrosses have not identified Idd loci within these intervals (27).

Diabetes pathogenesis in NOD mice is characterized by severe insulitis followed by the CD4⁺ and CD8⁺ T-cell–mediated destruction of insulin-producing β-cells in the pancreatic islets of Langerhans. Previous studies have reported that female NOD mice first develop insulitis at 5 weeks of age and diabetes at ∼90 days after birth, with 46% of mice having diabetes by 150 days of age (28), although this can be influenced by diet and the gut microbiome (29). By contrast, male mice are significantly less susceptible, with a much lower frequency of diabetes (28). We monitored for diabetes in female and male Ptpn2^fl/fl.NOD and Lck-Cre;Ptpn2^fl/fl.NOD mice (Fig. 1A and B). The onset of diabetes (urine glucose >55 mmol/L, blood glucose >18 mmol/L) was markedly accelerated in both female and male Lck-Cre;Ptpn2^fl/fl.NOD mice. Female Lck-Cre;Ptpn2^fl/fl.NOD mice started to develop diabetes at 36 days of age, with all Ptpn2-deficient mice succumbing by 87 days (Fig. 1A). Similarly, type 1 diabetes onset in male Lck-Cre;Ptpn2^fl/fl.NOD mice occurred at 29 days, with 100% of mice developing diabetes by 108 days (Fig. 1B). By contrast, neither male nor female Ptpn2^fl/fl.NOD littermate mice or parental NOD mice developed diabetes by 100 days of age. Histological analysis in 30- to 36-day-old Lck-Cre;Ptpn2^fl/fl.NOD mice before the development of diabetes revealed advanced invasive insulitis (combined histological grades 3 [intraislet infiltration] and 4 [structural derangement and complete β-cell loss]) in 91% of islets with obliterative lesions and marked immune cell infiltrates (Fig. 1C and D). The immune infiltrates contained not only CD3⁺ T cells but also B220⁺ B cells (Fig. 1E), which are believed to contribute to invasive insulitis and diabetes (30,31). By contrast, invasive insulitis was only observed in ∼2% of islets from 30- to 36-day-old Ptpn2^fl/fl.NOD littermate controls, and islets exhibited minimal peri-insulitis (grade 1) and immune cell infiltration (grade 0) (Fig. 1C and D). Thus, PTPN2 deficiency in T cells in NOD mice markedly accelerates pancreatic islet destruction and the onset of diabetes.

The type 1 diabetes–associated noncoding PTPN2 SNP rs1893217 is accompanied by a reduction in PTPN2 mRNA in human CD4⁺CD45RO memory T cells (9). Similarly, the Crohn disease–associated noncoding PTPN2 SNP rs2542151 that is also associated with early-onset type 1 diabetes (4) is accompanied by a reduction in PTPN2 protein in primary human colonic lamina propria fibroblasts (11,12). Thus, PTPN2 might be haploinsufficient in the pathogenesis of autoimmune disorders such as type 1 diabetes and Crohn disease. We have previously shown that PTPN2 heterozygosity in Lck-Cre;Ptpn2^fl/+ mice results in an ∼50% reduction of PTPN2 protein in T cells (18). We therefore monitored for the incidence of diabetes in Ptpn2^fl/fl.NOD and Lck-Cre;Ptpn2^fl/+.NOD mice (Fig. 1A and B). Although delayed relative to homozygous mice, the onset of diabetes was accelerated in female Lck-Cre;Ptpn2^fl/+.NOD mice compared with their Ptpn2^fl/fl.NOD littermate controls. Moreover, although male Ptpn2^fl/fl.NOD control mice did not develop diabetes, even at 250 days of age, male Lck-Cre;Ptpn2^fl/+.NOD heterozygous mice began to develop diabetes by 113 days of age. Histological analysis in 84- to 105-day-old Lck-Cre;Ptpn2^fl/+.NOD mice revealed invasive insulitis in 50% of islets (Fig. 1F and G), whereas in Ptpn2 ^fl/fl.NOD mice invasive insulitis was only evident in 8% of islets. These results emphasize the importance of PTPN2 in T-cell tolerance and suggest that even partial PTPN2 deficiency in T cells might be sufficient to accelerate the onset of diabetes in autoimmune-prone individuals.

T-cell responses to proinsulin occur early in the pathogenesis of type 1 diabetes in NOD mice and are required for the development of immune responses to other antigens (32). Dendritic cells and insulin-specific CD4⁺ T cells cross-prime CD8⁺ T cells to elicit cytotoxic responses and the destruction of β-cells (33). We have shown previously that PTPN2 deficiency in CD8⁺ T cells negates the need for cross-priming and drives the differentiation and activation of effector T cells and overt responses to self-antigen (15,18,19). C57BL/6 mice deficient for PTPN2 in T cells progressively develop an effector/memory (CD44^hiCD62L^lo) T-cell phenotype and autoimmunity with age (15). In Lck-Cre;Ptpn2^fl/fl.NOD mice, we noted a significant increase in effector/memory (CD44^hiCD62L^lo) CD4⁺ and CD8⁺ T cells in the spleens and lymph nodes of 36-day-old Lck-Cre;Ptpn2^fl/fl.NOD mice before the development of diabetes (Fig. 2A and B). We also noted a significant increase in effector/memory CD4⁺ and CD8⁺ T cells in the pancreatic draining lymph nodes of prediabetic Lck-Cre;Ptpn2^fl/fl.NOD mice (Fig. 2C). Notably, although hardly any CD4⁺ or CD8⁺ T cells were in the pancreata of age-matched Ptpn2^fl/fl.NOD littermates, the early-onset diabetes in Lck-Cre;Ptpn2^fl/fl.NOD mice was preceded by the marked infiltration of CD44^hiCD62L^lo CD4⁺ and CD8⁺ T cells (Fig. 2D). Previous studies have established that the initiation of diabetes in NOD mice relies on the presence of activated insulin-specific CD4⁺ T_H1 cells that produce IFNγ and allow for the licensing of antigen-specific CD8⁺ T cells (34,35). We found that the CD44^hiCD62L^lo CD4⁺ T cells infiltrating the pancreata of Lck-Cre;Ptpn2^fl/fl.NOD mice were enriched for CD4⁺ T_H1 (IFNγ^hi) cells (Fig. 2E). Consistent with previous studies (36), we found that PTPN2-deficient naive CD4⁺ T cells differentiated more efficiently into T_H1 cells ex vivo, even in the absence of IL-12, consistent with this being a cell-intrinsic effect (Supplementary Fig. 1A). At least in part, the enhanced T_H1 differentiation may be due to the promotion of STAT1 signaling, which is required for T_H1 commitment and differentiation (37). STAT1 is a bona fide PTPN2 substrate (38), and PTPN2 deficiency was accompanied by heightened basal and IFNγ–induced STAT1 Y701 phosphorylation (Supplementary Fig. 1B). By contrast, IL-12–induced STAT4 Y693 phosphorylation in activated CD4⁺ T cells was unaltered by PTPN2 deficiency (Supplementary Fig. 1C). In addition to the increased presence of T_H1 cells, we found that infiltrating CD44^hiCD62L^lo CD8⁺ T cells exhibited a terminally differentiated and antigen-experienced CD49d^hiKLRG1^hiCD127^– effector phenotype and expressed high levels of IFNγ, a marker of cytotoxic CD8⁺ T cells (Fig. 2F). Taken together, these results are consistent with PTPN2 deficiency increasing the activation and pathogenic conversion of CD4⁺ and CD8⁺ T cells into T_H1 and cytotoxic CD8⁺ T cells, respectively, to mediate destruction of pancreatic β-cells.

To explore directly whether PTPN2 deficiency in CD8⁺ T cells might subvert T-cell tolerance to promote the CD8⁺ T-cell–mediated destruction of β-cells, we adoptively transferred CD8⁺-naive (CD44^loCD62L^hi) T cells from prediabetic Ptpn2^fl/fl.NOD versus Lck-Cre;Ptpn2^fl/fl.NOD mice into immunoreplete NOD hosts and monitored for diabetes (Fig. 3). Albeit delayed, diabetes was evident by 44 days in mice receiving PTPN2-deficient T cells, with 86% of mice succumbing by 172 days post–adoptive transfer (Fig. 3A). By contrast, the onset of diabetes was not altered in control mice, and diabetes was not evident until 117 days of age (Fig. 3A). Histological analysis at 120 days post–adoptive transfer revealed that Lck-Cre;Ptpn2^fl/fl CD8⁺ T-cell recipients exhibited increased invasive insulitis accompanied by marked lymphocytic infiltrates in islets (Fig. 3B and C). These results are consistent with the accelerated onset of diabetes in Lck-Cre;Ptpn2^fl/fl.NOD mice being largely, albeit not completely, due to perturbations in peripheral CD8⁺ T-cell tolerance to self-antigens.

CD4⁺FoxP3⁺ T_regs prevent autoimmunity by suppressing the activity of autoreactive CD8⁺ T cells that escape negative selection in the thymus (39). SNPs in genes encoding IL-2 and the IL-2 receptor (CD25) that are required for T_reg generation have been linked with various autoimmune diseases, including type 1 diabetes (2). We therefore assessed whether the accelerated onset of type 1 diabetes evident in Lck-Cre;Ptpn2^fl/fl.NOD mice might arise through the suppression of T_reg development or function (Fig. 4). Consistent with our previous studies in C57BL/6 mice (15,17), we found that PTPN2 deficiency in prediabetic 5-week-old Lck-Cre;Ptpn2^fl/fl.NOD mice was accompanied by increased thymic (CD25⁻FoxP3⁺ and CD25⁺FoxP3⁺) T_regs (Fig. 4A). Moreover, PTPN2 deficiency was accompanied by an increased proportion of total (Fig. 4B), resting (CD44^hiCD62L^hi), and effector (CD44^hiCD62L^lo) CD25⁺FoxP3⁺ T_regs in the pancreata and corresponding draining lymph nodes (Fig. 4C and D) with the increase in T_regs in pancreata approximating the increase in effector/memory CD4⁺ T cells (Fig. 4E); total and effector T_regs trended higher also in the spleen and inguinal lymphoid tissues (Supplementary Fig. 2A). T_regs expand and acquire an activated effector phenotype in response to cognate antigens in the context of infection and autoimmunity (40). PTPN2 deficiency not only increased the proportion of effector T_regs but also increased the number of resting and effector T_regs in pancreata and pancreatic lymph nodes. By contrast, antigen-independent tumor growth factor-β/IL-2–induced CD25⁺FoxP3⁺ T_reg generation ex vivo was not altered by PTPN2 deficiency (Supplementary Fig. 2B). These findings are consistent with PTPN2 deficiency increasing the antigen-induced expansion and differentiation of T_regs in vivo. To determine whether PTPN2 deficiency might abrogate T_reg function, we isolated splenic CD25^hi T_regs from Ptpn2^fl/fl.NOD versus Lck-Cre;Ptpn2^fl/fl.NOD mice and assessed their capacity to repress the TCR-mediated expansion of CD4⁺CD25^lo T cells ex vivo (Fig. 4F). We found that Lck-Cre;Ptpn2^fl/fl T_regs were just as efficient as Ptpn2^fl/fl T_regs in repressing the α-CD3ε–induced expansion of CD4⁺ T cells. Therefore, the accelerated onset of diabetes in Lck-Cre;Ptpn2^fl/fl.NOD mice cannot be attributed to diminished T_reg development or function.

Our studies indicate that PTPN2 deficiency in T cells promote not only the infiltration of cytotoxic CD8⁺ and CD4⁺ T cells into the pancreata of NOD mice but also the recruitment of B220⁺ B cells, which promote the expansion and activation of T cells targeting β-cell antigens (41). Therefore, another mechanism by which PTPN2 deficiency might influence disease progression in NOD mice is through effects on the B-cell compartment. CD4⁺ T_FH cells are required for the promotion of B-cell maturation and the production of antigen-specific antibodies (42). Previously, we reported that the inducible deletion of Ptpn2 in the hematopoietic compartment of adult C57BL/6 mice was accompanied by the marked expansion of T_FH cells ex vivo (22). To explore the mechanisms by which PTPN2 deficiency in T cells might result in increased B-cell pancreatic infiltrates, we investigated the presence of T_FH cells and germinal center B cells in the pancreatic draining lymph nodes of 5-week-old Lck-Cre;Ptpn2^fl/fl.NOD mice. T-cell PTPN2 deficiency in NOD mice resulted in a striking increase in CD4⁺CD44^hiCXCR5^hiPD-1^hi T_FH cells in the spleens, inguinal lymph nodes, and pancreatic draining lymph nodes (Fig. 5A). In keeping with T_FH cell expansion, we found that B220⁺GL-7^hiFas^hi germinal center B cells were also markedly increased in spleens, inguinal lymph nodes, and pancreatic draining lymph nodes (Fig. 5B). Germinal center B-cell expansion and class switching from IgM to IgG rely on interactions with germinal center T_FH cells. Consistent with this, we found that circulating IgM and IgG (IgG1, IgG2c, IgG2b, IgG3) were increased in prediabetic Lck-Cre;Ptpn2^fl/fl.NOD mice (Fig. 5C). Importantly, we found that circulating insulin autoantibodies were elevated in 5-week-old Lck-Cre;Ptpn2^fl/fl.NOD mice as well as in 12- to 15-week-old Lck-Cre;Ptpn2^fl/+.NOD mice (Fig. 5D). These results indicate that T-cell PTPN2 deficiency in NOD mice might influence disease progression not only through effects on CD8⁺ T-cell tolerance but also through the CD4⁺ T_FH-mediated expansion and maturation of B cells and the production of autoantibodies.

In addition to type 1 diabetes, NOD mice spontaneously develop autoimmune comorbidities, including sporadic lymphoid infiltrates in the thyroid gland, mimicking Hashimoto thyroiditis (43), and sporadic lymphoid infiltrates in the salivary and lacrimal glands, resembling the human autoimmune disorder known as Sjögren syndrome (44). Moreover, young NOD mice can develop subclinical colitis after weaning as a result of defective tolerance to commensal gut bacteria (45). Such autoimmune comorbidities are also seen in a subset of patients with type 1 diabetes (46). Previous studies have shown that PTPN2 deficiency in T cells or the hematopoietic compartment of C57BL/6 mice promotes systemic inflammation and autoimmunity accompanied by lymphocytic infiltrates in nonlymphoid tissues, such as liver, lung, skin, and kidney (15,22). We therefore assessed the impact of PTPN2 deficiency on the development of systemic inflammation and autoimmunity in prediabetic 5- to 6-week-old Lck-Cre;Ptpn2^fl/fl.NOD mice. PTPN2 deficiency in T cells was accompanied by systemic inflammation as reflected by the increase in circulating proinflammatory cytokines tumor necrosis factor-α and IL-6 (Fig. 6A) and the infiltration of lymphocytes, including B220⁺ B cells and CD3⁺ T cells, into the submandibular salivary glands (Fig. 6B). In these young prediabetic mice, lymphocytic infiltrates were not evident in liver, lung, kidney, or joints, and there was no evidence of inflammatory arthritis (Supplementary Fig. 3A and B). Strikingly, however, we found marked lymphocytic infiltrates in the colons of prediabetic Lck-Cre;Ptpn2^fl/fl.NOD mice and the development of overt colitis as reflected by colon shortening (Fig. 6C). The development of colitis was accompanied by an increased number of CD4⁺ and CD8⁺ effector/memory (CD44^hiCD62L^lo) T cells, T_FH cells (CD4⁺CD44^hiCXCR5^hiPD-1^hi), and germinal center B cells (B220⁺GL-7^hiFas^hi) in the mesenteric lymph nodes; no alterations were evident in CD4⁺FoxP3⁺ T_regs (Fig. 7A). Moreover, colitis was accompanied by the accumulation of lamina propria CD4⁺ and CD8⁺ effector/memory T cells with T_H1 and cytotoxic phenotypes, respectively (Fig. 7B and C), as assessed by their ability to produce IFNγ ex vivo (Fig. 7C). No significant differences were evident in the proportion of lamina propria CD4⁺ T_H17 cells (Fig. 7C), and T_H17 generation ex vivo was not affected by PTPN2 deficiency (Supplementary Fig. 4A). Similarly, intraepithelial and lamina propria TCRγδ⁺ T cells in the colons of Lck-Cre;Ptpn2^fl/fl. NOD mice were not altered (Supplementary Fig. 4B). These results are consistent with T-cell PTPN2 deficiency in NOD mice, driving the development of not only type 1 diabetes but also other autoimmune/inflammatory disorders.

Many overlapping tolerance mechanisms exist that normally prevent self-reactive B and T cells from attacking the body during homeostatic processes and antimicrobial defense. Molecules that negatively regulate T-cell and B-cell signaling are fundamentally important for tuning T-cell and B-cell responses to prevent overt autoreactivity. PTPN2 negatively regulates TCR signaling to prevent overt autoreactivity to low-affinity self-peptide-MHC complexes (15,18,19). Here, we demonstrate that PTPN2 deficiency in T cells alone is sufficient to markedly exacerbate disease onset and severity of diabetes in autoimmune-prone NOD mice.

The development of type 1 diabetes in NOD mice is characterized by the infiltration of antigen-specific T_H1-polarized CD4⁺ and cytotoxic CD8⁺ T cells into pancreatic islets, resulting in the lymphocyte-driven destruction of insulin-producing β-cells (25,47). In this study, we found that the transfer of naive PTPN2-deficient CD8⁺ T cells alone into immunoreplete NOD hosts accelerated the onset and increased the incidence of type 1 diabetes. These findings are consistent with PTPN2 deficiency in CD8⁺ T cells, enhancing TCR-instigated responses to self to promote autoimmunity. However, because the onset of diabetes in Lck-Cre;Ptpn2^fl/fl.NOD mice exceeded that in which PTPN2-deficient CD8⁺ T cells alone had been transferred, it is probable that perturbations in other T-cell subsets might also contribute to disease progression. We have shown previously that PTPN2 deficiency enhances the TCR-induced activation and differentiation of CD4⁺ T cells into effector/memory T cells (15,19). Consistent with this, effector/memory CD4⁺ T cells were increased in the pancreatic draining lymph nodes and the pancreata of Lck-Cre;Ptpn2^fl/fl.NOD mice. The recruitment and activation of CD4⁺ T cells can contribute to the licensing and activation of CD8⁺ T cells. However, because PTPN2 deficiency in CD8⁺ T cells negates the need for CD4⁺ T-cell help during antigen cross presentation and permits the helper-independent acquisition of cytotoxic activity to self-antigens (19), we surmise that other CD4⁺ T-cell–dependent mechanisms may be more pertinent.

Spalinger et al. (36) reported that PTPN2 deficiency in CD4⁺ T cells can enhance T_H1 polarization but impair T_reg generation in dextran sodium sulfate or T-cell transfer colitis models. Consistent with this, we also noted an overt T_H1 polarization in prediabetic Lck-Cre;Ptpn2^fl/fl.NOD mice. Pioneering work by Katz et al. (35) has shown that T_H1 cells, but not T_H2 cells, are required for autoimmune diabetes, whereas a large number of studies have defined the importance of the T_H1-cell signature cytokine IFNγ in promoting the development of insulitis and type 1 diabetes in NOD mice (48). Therefore, the enhanced T_H1 polarization in Lck-Cre;Ptpn2^fl/fl.NOD mice is likely to be an important contributor to the accelerated onset of type 1 diabetes. However, in contrast to the findings of Spalinger et al., we found that thymic T_regs and effector-like T_regs in the pancreata and draining lymph nodes in Lck-Cre;Ptpn2^fl/fl.NOD mice were increased rather than decreased. Consistent with this, previous studies have shown that PTPN2 dephosphorylates JAK1/3 and STAT5 and attenuates IL-2–induced STAT5 signaling in thymocytes to suppress the generation of FoxP3⁺ T_regs (49). Moreover, we have shown previously that Ptpn2 deletion in Lck-Cre;Ptpn2^fl/fl.C57BL/6 mice or Mx1-Cre;Ptpn2^fl/fl.C57BL/6 mice in which Ptpn2 has been inducibly deleted increases the thymic generation of CD25⁺FoxP3⁺ T_regs and does not compromise T_reg function ex vivo (15,17). In this study, we reaffirmed that PTPN2 deficiency does not compromise the suppressor properties of T_regs isolated from NOD mice. Therefore, defective T_reg generation/function is unlikely to contribute to the loss of tolerance in Lck-Cre;Ptpn2^fl/fl.NOD mice.

An additional mechanism by which PTPN2-deficient CD4⁺ T cells may facilitate the development of type 1 diabetes is by promoting the role of B cells in progression toward invasive insulitis and diabetes. Although B cells are not essential for the generation or effector function of islet-reactive T cells, they can contribute to the development of autoreactivity (41). Indeed, anti-CD20–mediated B-cell depletion (50), or the arrest of B-cell maturation by disabling IgM production (30), protects against type 1 diabetes in NOD mice, whereas CD19 deletion in NOD B cells diminishes the expansion of β-cell antigen-specific T cells (31). In humans, B-cell depletion with anti-CD20 can delay disease progression in patients with newly diagnosed type 1 diabetes (51). In our studies, we found a marked infiltration of B220⁺ B cells into the pancreatic islets of Lck-Cre;Ptpn2^fl/fl.NOD mice. This was accompanied by a significant increase in the number of germinal center B220⁺ B cells in lymphoid organs, including in the draining lymph nodes of the pancreas. The increase in germinal center B cells was in turn accompanied by the expansion of CD4⁺ T_FH cells in lymphoid organs. T_FH cells are required for the formation of germinal centers, the promotion of B-cell proliferation, and the production of antigen-specific antibodies (42). In this regard, we identified an increased humoral immunity in Lck-Cre;Ptpn2^fl/fl.NOD mice that was reflected by an increase in circulating IgM and IgG and circulating anti-insulin autoantibodies. Recent gene expression profiling in DO11;RIP-mOVA mice (a model of type 1 diabetes) has identified a T_FH-cell gene signature in islet-specific T cells and has shown that transgenic DO11 T cells with a T_FH signature from diabetic animals transfer diabetes to RIP-OVA recipients (26). Importantly, increased T_FH cells are linked with the development of autoimmunity, and T_FH cells are overrepresented in patients with type 1 diabetes (26,52). The production of T_FH cells and their homing to germinal centers depend on autocrine IL-21 that signals through JAK1/3 and STAT3 (53). STAT3 is required for both T_FH and germinal center B-cell differentiation, and STAT3 gain-of-function mutations that increase T_FH cells and autoantibody production may be associated with type 1 diabetes in humans (42,54). Moreover, along with Il2, Il21 is a candidate gene for the diabetes susceptibility locus Idd3, and IL-21 signaling is required for diabetes development in NOD mice (55). JAK1/3 and STAT3 are bona fide substrates of PTPN2 (21), and the inducible hematopoietic deletion of Ptpn2 in adult C57BL/6 mice is accompanied by heightened IL-21–induced STAT3 signaling in T_FH cells ex vivo (22). Therefore, in addition to lowering the threshold for TCR-instigated responses to self-antigen by enhancing LCK signaling, PTPN2 deficiency may drive STAT3 signaling in CD4⁺ T cells to enhance the formation of T_FH cells and the maturation of B cells to facilitate disease progression in Lck-Cre;Ptpn2^fl/fl.NOD mice.

Beyond spontaneously developing type 1 diabetes, NOD mice also exhibit T-cell–mediated autoimmunity against other tissues, including the thyroid gland and especially the lacrimal and salivary glands, mimicking Sjögren syndrome (43,44). Moreover, young NOD mice also develop low levels of colitis as a result of disturbed tolerance toward autologous commensal gut antigens (45). This is accompanied by an increase in activated CD4⁺ T cells and T_H17 cells in the colonic lamina propria (45). In addition to promoting type 1 diabetes, T-cell–specific PTPN2 deficiency in NOD mice was accompanied by marked lymphocytic infiltrates in salivary glands and the formation of what resembled ectopic lymphoid-like structures that included T cells and B cells. Ectopic lymphoid-like structures that develop at sites of inflammation can form functional germinal centers and contribute to disease pathogenesis. Importantly, ectopic lymphoid-like structures are seen in the salivary glands of patients with Sjögren syndrome (56). Similarly, PTPN2 deficiency resulted in widespread immune infiltrates and the formation of lymphoid clusters, probably ectopic lymphoid-like structures that are a pathological hallmark of inflammatory bowel disease, including Crohn disease (57). Consistent with this, lymphocytic infiltrates in mesenteric lymph nodes included not only CD44^hiCD62L^lo CD4⁺ T cells but also T_FH cells and germinal center B cells. In addition, there was a marked infiltration of activated/cytotoxic CD8⁺ T cells and CD4⁺ T cells with a predominant T_H1 phenotype in the lamina propria of Lck-Cre;Ptpn2^fl/fl. NOD mice. Although previous studies have found that in dextran sodium sulfate or T-cell transfer-induced colitis models, PTPN2 deficiency resulted in the induction of both T_H1 and T_H17 cells and the impaired induction of T_regs (36), we found that T_H17 cells and T_regs were unaltered in the colons of 5-week-old prediabetic mice, and the generation of T_H17 cells and T_regs ex vivo was not affected by PTPN2 deficiency. Nonetheless, we cannot exclude the possibility that such alterations may occur as disease progresses.

The findings of our study underscore the importance of PTPN2 in tuning T-cell responses for the maintenance of T-cell tolerance and prevention of autoimmunity. Moreover, because PTPN2 deficiency in T cells alone exacerbated the development of both type 1 diabetes and colitis in NOD mice, our results argue for perturbations in T-cell function being causally involved in early-onset type 1 diabetes and Crohn disease associated with PTPN2 loss-of-function SNPs in humans. Importantly, the findings of this study suggest that beyond driving type 1 diabetes or Crohn disease, PTPN2 loss-of-function SNPs in humans might also contribute to the development of associated comorbidities.

Acknowledgments. The authors thank Alexandra Ziegler and Lauren Stanton (Monash University) for technical support, Rhana Kostoulias (Monash ARL) for animal husbandry, and Robyn Slattery (Monash University) for providing the NOD/Lt mice.

Funding. This work was supported by the National Health and Medical Research Council (NHMRC) of Australia (1047055 to T.T.), Versus Arthritis (19796, 20770 to S.A.J. and G.W.J), and the Victorian Operational Infrastructure Support Program (to T.C.B. and T.W.H.K). T.T is an NHMRC Principal Research Fellow (1103037).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.W., T.C.B., P.K.G., Y.A.L., G.W.J., D.Y., A.G.B., S.A.J., T.W.H.K., and T.T. contributed to the methodology and writing, review, and editing of the manuscript. F.W., T.C.B., P.K.G., Y.A.L., and G.W.J. contributed to the investigation. F.W., T.C.B., and T.T. wrote the original draft of the manuscript. F.W. and T.T. conceptualized the study. T.T. acquired the funding. T.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.