Loss of NOX-Derived Superoxide Exacerbates Diabetogenic CD4 T-Cell Effector Responses in Type 1 Diabetes

Type 1 diabetes (T1D) is characterized by a loss of self-tolerance, in which T-cells destroy insulin-secreting pancreatic β-cells. Currently, no cure exists for this disease; meanwhile, the worldwide incidence is steadily climbing by 4.7% per year in developed countries (1,2). CD4 T cells have classically been considered to be of chief importance in T1D, as CD4 T cell–deficient nonobese diabetic (NOD) mice are protected from T1D onset (3,4).

Superoxide synthesis is mediated by NADPH oxidase (NOX), a heme-containing multisubunit enzyme composed of cytosolic (p67^phox, p47^phox, p40^phox, and Rac1/2) and membrane-associated components (p22^phox and gp91^phox) (5). Normally localized in the cytosol in a quiescent state, the p47^phox subunit of NOX migrates to the membrane upon activation and associates with other components of the NOX machinery, inducing production of superoxide, which is quickly dismutated to form hydrogen peroxide (6). Reactive oxygen species (ROS) are classically known to possess potent antimicrobial properties, but ROS are also powerful modulators of the immune response (7).

Studies by our laboratory and others have demonstrated ROS function as a proinflammatory-derived third signal to synergize innate with adaptive immune responses (8–13). ROS promote proinflammatory cytokine and type I interferon synthesis via the redox-sensitive mitogen-activated protein kinase, nuclear factor-κB, and activator protein-1 signaling pathways (8,9,14,15). We previously reported that NOD mice possessing a spontaneous point mutation (Ncf1^m1J), resulting in a premature truncation of the p47^phox NOX subunit (16), were protected against T1D onset due, in part, to attenuated antiviral innate immune responses (17), dysregulated CD4 and CD8 T-cell responses (10,18,19), and an enhancement in alternatively activated M2 macrophages (19).

We sought to further examine the role of NOX-derived ROS in diabetogenic CD4 T-cell effector responses using the NOD.BDC-2.5 (BDC-2.5) mouse strain. The BDC-2.5 mouse is an invaluable tool for dissecting the role of a single CD4 T-cell clone in T1D (20,21). Recognizing chromogranin A (22), a protein component of the islet secretory granules, BDC-2.5 CD4 T cells are intrinsically activated to a Th1-like phenotype and destroy pancreatic β-cells by recruiting classically activated M1 macrophages (23,24). Here, we report that stimulated BDC-2.5.Ncf1^m1J splenocytes and CD4 T cells exhibited exacerbated proinflammatory cytokine and chemokine production, with a concomitant increase in spontaneous T1D and enhanced diabetogenicity upon transfer into NOD.Rag recipients. This heightened diabetogenicity was due in part to less suppressive T-regulatory (Treg) cells. Addition of an exogenous superoxide generator blunted Th1 autoreactive immune effector responses within BDC-2.5.Ncf1^m1J CD4 T cells by curbing interleukin-12 receptor β2 (IL-12Rβ2) signaling and attenuating P-signal transducer and activator of transcription 4 (STAT4) (Y693) activation. These results demonstrate the dual roles of superoxide in functioning as a proinflammatory third signal for efficient adaptive immune maturation but also in restricting autoreactive CD4 T-cell responses.

NOD.Cg-Ncf1m1J/Mx (NOD.Ncf1^m1J), NOD.Cg-Tg(TcraBDC2.5,TcrbBDC2.5)/DoiJ (BDC-2.5), and NOD.129S7(B6)-Rag1^tm1Mom/J (NOD.Rag) mice were bred and housed under pathogen-free conditions in the Research Support Building animal facility at the University of Alabama at Birmingham. BDC-2.5 mice were originally obtained from Kathryn Haskins at the National Jewish Hospital (Denver, CO), and NOD.Rag mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were maintained on a light/dark (12/12 h) cycle at 23°C and received continuous access to standard laboratory chow and acidified water. Age- and sex-matched BDC-2.5 and BDC-2.5.Ncf1^m1J mice were used for all experiments and in accordance with the University of Alabama at Birmingham and University of Florida Institutional Animal Care and Use Committee.

Anti–γ-interferon (IFN-γ), –interleukin (IL)-2, –IL-10, and –IL-17A antibody pairs for ELISA; fluorochrome-conjugated anti-CD4, -CD8, -B220, -Vβ4, -Vβ5, -Vβ8, -CD11b, and -CD11c antibodies for fluorescence-activated cell sorter (FACS); and anti-CD3ε and anti-CD28 antibodies were purchased from BD Biosciences. IL-1β, CCL5, and tumor necrosis factor (TNF)-α DuoSet ELISA kits, CXCL10 antibody pairs, and fluorochrome-conjugated anti–IL-12Rβ2 were purchased from R&D Systems. Fluorochrome-conjugated anti-CD19, -CD25, -CD44, and -CD69 antibodies were purchased from eBioscience, while biotin anti-mouse CD4, in addition to anti-F4/80 fluorochrome-conjugated and live/dead fluorochrome-conjugated antibodies, was purchased from Invitrogen. Anti–P-STAT4 (Y693) antibody was purchased from Cell Signaling, while anti-STAT4 antibody was obtained from Biolegend. The BDC-2.5 mimotope (EKAHRPIWARMDAKK) was synthesized by the University of Alabama at Birmingham peptide synthesis core facility. Xanthine oxidase and β-actin were obtained from Sigma-Aldrich, and r-hIL-2 was purchased from Peprotech.

NOD.BDC-2.5.Ncf1^m1J (BDC-2.5.Ncf1^m1J) mice were generated by crossing BDC-2.5 (21) to NOD.Ncf1^m1J (10) mice followed by intercrossing of the F1 progeny. F2 progeny homozygous for the Ncf1^m1J mutation (10) and the BDC-2.5 transgene (21) were used to establish this strain. The inability to synthesize superoxide did not affect lymphocyte development and differentiation in BDC-2.5.Ncf1^m1J mice, as lymphocyte population percentages revealed equivalent expression of Vβ4, CD4 T cells (Supplementary Fig. 2A) within superoxide-sufficient and -deficient strains according to the gating strategy depicted in Supplementary Fig. 1. No differences were observed in the percentages of B cells (Supplementary Fig. 2B), macrophages (Supplementary Fig. 2C), dendritic cells (Supplementary Fig. 2D), and regulatory CD4 T cells (Supplementary Fig. 2E).

The redox-sensitive substrate luminol was used to detect superoxide synthesis (17). BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes and CD4 T cells were seeded onto 96-well polystyrene round-bottom plates with 200 µmol/L luminol (Sigma-Aldrich), 0.32 units/mL horseradish peroxidase (Sigma-Aldrich), 100 ng/mL phorbol 12-myristate 13-acetate (PMA), and 1 μg/mL ionomycin or 1 µmol/L BDC-2.5 Mimotope (9). Luminescence was quantified using a SpectraMax L Luminescence microplate reader with readings every 2 min for 60 min.

Macromolecular-centered free radicals were detected upon stimulating splenocytes with 1 µmol/L BDC-2.5 mimotope in the presence of 1 mmol/L 5,5-dimethyl-1-pyrroline N-oxide (DMPO; Dojindo) in tissue culture–treated chamber slides. Cells were fixed in 4% paraformaldehyde in PBS, permeabilized with 0.5% Triton X-100 in PBS, blocked with 5% BSA in PBS, and incubated with 20 μg/mL of chicken IgY anti-DMPO as previously described (25,26). DMPO adducts were detected with Alexa Fluor 488–conjugated goat anti-chicken IgY secondary antibody (1:500; Invitrogen). P-STAT4 (Y693) and STAT4 were detected with Alexa Fluor 488–conjugated donkey anti-rabbit IgG (H+L) secondary antibody (1:500) and Cy3-conjugated Donkey anti-mouse IgG (H+L) (1:500) (Jackson Immunoresearch), respectively. CD4 T cells were identified with Alexa Fluor 647–conjugated streptavidin secondary antibody (1:500; Life Technologies). Images were obtained with an Olympus IX81 Inverted Microscope at a 40× objective and analyzed with cellSens Dimension imaging software, version 1.12. For quantitation of fluorescence intensity, three to six images were obtained for each data point. Each image was collected at the same exposure time, adjusted to the same intensity level for standardization, and the fluorescence intensity was measured using ImageJ software (National Institutes of Health).

According to the gating strategies shown in Supplementary Figs. 1 and 5, BDC-2.5 and BDC-2.5.Ncf1^m1J splenic single cell suspensions were surface stained in FACS buffer (1% BSA in PBS) with appropriately diluted fluorochrome-conjugated antibodies and isotype controls (BD Biosciences, eBioscience, and R&D) (Supplementary Figs. 6 and 10) (17). One hundred thousand cell events were collected with the FACSCalibur, and data were analyzed with FlowJo, version 9.6.1, software (Tree Star Incorporated).

Splenic CD4 T cells from BDC-2.5 and BDC-2.5.Ncf1^m1J mice were purified by negative selection according to the manufacturer’s protocol using the EasySep CD4 T-cell enrichment kit (STEMCELL Technologies). CD4 T-cell purity was routinely assessed by flow cytometry and found to be >90% (data not shown). T cells were stimulated with plate-bound anti-CD3ε (0.1 µg/mL) and anti-CD28 (1 µg/mL), and exogenous superoxide was added via 1 mU/mL xanthine oxidase (XO) (17). BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes were incubated with 1 µmol/L BDC-2.5 mimotope or 10⁴ dispersed NOD.Rag islet cells in the presence or absence of 1 mU/mL XO (9,15). Greiss assay and ELISAs were read on a Synergy2 microplate reader with Gen5, version 1.10, software (BioTek) as previously described (17). 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay to assess T-cell proliferation was performed according to the manufacturer’s protocol (Sigma-Aldrich).

Adoptive transfer of purified CD4 T cells into NOD.Rag recipients was performed as described previously (10,27). BDC-2.5 and BDC-2.5.Ncf1^m1J Tregs were purified via negative selection, stained with fluorescein isothiocyanate–labeled anti–glucocorticoid-induced tumor necrosis factor receptor–related protein and phycoerythrin-labeled anti-CD25. The CD25/GITR double-positive Treg population was sorted with a BD FACSAria III (BD Biosciences). The percentage of Foxp3⁺ Treg cells isolated after sorting was similar between BDC-2.5 and BDC-2.5.Ncf1^m1J (Supplementary Fig. 8). Polyclonally stimulated BDC-2.5 CD4 T cells were used as effector cells. Stimulated-BDC-2.5 CD4 effector T cells (10⁵) were transferred intraperitoneally, either alone or cotransferred with BDC-2.5 or BDC-2.5.Ncf1^m1J Tregs (2 × 10⁴), into NOD.Rag mice and monitored for T1D as described.

Data were analyzed via GraphPad Prism, version 5.0, statistical software. The difference between mean values for each experimental group was assessed using the two-tailed Student t test with P < 0.05 considered significant.

For exploration of the role of NOX-derived superoxide on diabetogenic CD4 T-cell effector responses, the BDC-2.5.Ncf1^m1J mouse strain was generated, and luminol oxidation by BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes was used to corroborate the deficiency in NOX-derived superoxide (Fig. 1A). Stimulated BDC-2.5.Ncf1^m1J splenocytes displayed a 21-fold reduction in chemiluminescence compared with BDC-2.5 splenocytes with luminol and horseradish peroxidase (HRP). To validate blunted free radical synthesis within the CD4 T-cell compartment, we stimulated BDC-2.5 and BDC-2.5.Ncf1^m1J CD4 T cells with PMA/ionomycin in the presence of luminol and HRP (Fig. 1B). Primary human and mouse CD4 T cells have been shown to express a functional phagocyte type NOX, and T-cell receptor (TCR) stimulation elicits rapid production of superoxide and hydrogen peroxide (28). Similar to splenocytes, luminol oxidation by stimulated BDC-2.5.Ncf1^m1J CD4 T cells was decreased 1.4-fold compared with BDC-2.5 CD4 T cells, confirming the blunted respiratory burst capacity from our previous studies using CD4 or CD8 T lymphocytes (10,18).

For further demonstration that NOX enzymatic activity was curtailed in the absence of NOX-derived superoxide, immuno-spin trapping using the free radical spin trap DMPO was performed to quantify the self-oxidation of BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes (Fig. 2). The formation of free radicals in macromolecules of the cells results in the formation of stable DMPO adducts that were imaged and quantified. The BDC-2.5 mimotope elicited an increase in DMPO-adduct formation by BDC-2.5 CD4 T cells and MHC-II (OX-6) cells; however, BDC-2.5.Ncf1^m1J splenocytes displayed a severe reduction in DMPO adducts within CD4 and MHC-II compartments compared with BDC-2.5 (Fig. 2A). In the absence of DMPO addition to antigen-stimulated BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes, immunofluorescence detection of free radical adducts was absent (Supplementary Fig. 3). The diminution in DMPO-adduct synthesis by BDC-2.5.Ncf1^m1J CD4- and MHC-II–expressing cells was significant upon quantification, further corroborating reduced respiratory burst activity by NOX-deficient BDC-2.5 CD4 T cells upon stimulation with their cognate antigen (Fig. 2B).

For investigation of the role of NOX-derived superoxide in diabetogenic CD4 T cells, BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes were stimulated with dispersed NOD.Rag islet cells and the BDC-2.5 mimotope. Contrary to results from our laboratory and others implicating ROS as a proinflammatory-derived third signal essential for effective adaptive immunity (8–10,14), BDC-2.5.Ncf1^m1J splenocytes displayed significant elevations in nitrite (NO₂⁻) (2.7-fold), IFN-γ (4.3- and 2.0-fold), TNF-α (4- and 1.7-fold), IL-1β (1.5- and 2.5-fold), IL-17A (3.0- and 2.0-fold), and IL-2 (2- and 1.5-fold) upon islet and mimotope stimulation, respectively, compared with BDC-2.5 (Fig. 3A–F). Besides an influx in pathogenic proinflammatory cytokines that contribute to β-cell demise and T1D autoreactivity in T1D, BDC-2.5.Ncf1^m1J splenocytes displayed enhanced synthesis of proinflammatory chemokines implicated in T1D pathogenesis, including CCL5 (chemokine [C-C] motif ligand 5) (3- and 2.0-fold) and CXCL10 (chemokine [C-X-C motif] ligand 10) (332- and 2.0-fold) upon incubation with dispersed islet cells or mimotope, respectively (Fig. 3G–H) (29–31). In addition to splenocytes, purified BDC-2.5.Ncf1^m1J CD4 T cells exhibited significant elevations in nitrite (3.8-fold), IFN-γ (2.8-fold), TNF-α (twofold), IL-1β (4.4-fold), IL-17A (4.5-fold), IL-2 (1.5-fold), CCL5 (2.9-fold), and CXCL10 (twofold) (Supplementary Fig. 4A–H). Thus, despite possessing deficiencies in NOX-derived superoxide, BDC-2.5.Ncf1^m1J splenocytes and CD4 T cells displayed elevated levels of proinflammatory mediators implicated in T1D pathogenesis.

For further demonstration that BDC-2.5.Ncf1^m1J CD4 T cells display an exacerbated effector response in contrast to BDC-2.5, expression of the T-cell activation markers CD69, CD25, and CD44 was assessed by flow cytometry after antigenic stimulation (Fig. 4). Superoxide-deficient BDC-2.5 CD4 T cells exhibited an increase in surface expression of CD69 (early activation marker) (Fig. 4A), a 1.6-fold elevation in the IL-2 receptor α-chain (CD25) (Fig. 4B), and a 1.3-fold elevation in percentage of CD44 (Fig. 4C), a cell-surface glycoprotein that marks effector memory CD4 T cells (32). Pooled data from activated BDC-2.5.Ncf1^m1J CD4 T cells further demonstrated a significant (P < 0.05) increase in CD69 (A), CD25 (B), and CD44 (C) T-cell activation markers in contrast to BDC-2.5 (Supplementary Fig. 7). In agreement with increased proinflammatory cytokine and chemokine synthesis by stimulated BDC-2.5.Ncf1^m1J T cells (Fig. 3 and Supplementary Fig. 4), the inherent absence of NOX-derived superoxide synthesis elicited an elevated T-cell activation phenotype.

For further demonstration that loss of NOX-derived superoxide can augment CD4 T-cell effector responses, spontaneous T1D incidence was compared between BDC-2.5 and BDC-2.5.Ncf1^m1J mice. While BDC-2.5 mice do not spontaneously develop autoimmune diabetes, 20% of BDC-2.5.Ncf1^m1J female mice were hyperglycemic at 30 weeks of age (Fig. 5A). For examination of the influence of NOX-derived superoxide on CD4 T-cell diabetogenicity, adoptive transfers of purified, preactivated BDC-2.5 and BDC-2.5.Ncf1^m1J CD4 T cells into NOD.Rag recipients were performed (Fig. 5B). Because of the enhanced synthesis of proinflammatory cytokines and chemokines by BDC-2.5.Ncf1^m1J splenocytes (Fig. 3) and CD4 T cells (Supplementary Fig. 4), we hypothesized that adoptive transfer of superoxide-deficient CD4 T cells would elicit enhanced diabetogenicity compared with BDC-2.5 (21,33). Indeed, transfer of BDC-2.5.Ncf1^m1J CD4 T cells was more diabetogenic. Only 26% of BDC-2.5.Ncf1^m1J-transferred NOD.Rag (n = 45) recipients remained diabetes free at 11 days posttransfer. In contrast, 56% of BDC-2.5–transferred NOD.Rag mice (n = 24) (Fig. 5B) did not develop T1D. However, at 31 days posttransfer, no significant difference in diabetogenicity was observed among the groups of NOD.Rag mice receiving BDC-2.5 or BDC-2.5.Ncf1^m1J CD4 T cells (Fig. 5B).

For determination of whether the observed enhanced diabetogenicity could be due to an intrinsic defect in Treg suppressive function, splenic BDC-2.5 and BDC-2.5.Ncf1^m1J Treg cells were purified and used in an in vivo suppression assay with effector CD4 T cells isolated from BDC-2.5 mice (Fig. 5C). Coadoptive transfers into NOD.Rag recipients demonstrated that, as expected (34), BDC-2.5 Treg cells were functional and capable of significantly delaying T1D. The majority of the NOD.Rag mice transferred with BDC-2.5 Treg cells remained diabetes free (7 of 10 at 30 days posttransfer). However, BDC-2.5.Ncf1^m1J Treg cells were compromised in their ability to delay T1D. This in vivo suppression assay resulted in only 33% of the NOD.Rag recipients remaining diabetes free (n = 9) at 30 days postcotransfer (Fig. 5C) even though the percentage of Foxp3⁺ CD4⁺ CD25⁺ GITR⁺ T cells was similar from purified BDC-2.5 and BDC-2.5.Ncf1^m1J splenic Treg cells (Supplementary Fig. 8).

For determination of whether elevated proinflammatory cytokine and chemokine synthesis was directly linked to deficiencies in NOX-derived superoxide, the superoxide-generating enzyme XO was added to BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes stimulated with the BDC-2.5 mimotope (Fig. 6A–E) or to purified, splenic BDC-2.5 and BDC-2.5.Ncf1^m1J CD4 T cells stimulated with α-CD3ε and α-CD28 (Fig. 6F–H). At 24 h poststimulation with the BDC-2.5 mimotope, BDC-2.5.Ncf1^m1J splenocytes exhibited a 2.3-fold elevation in IFN-γ synthesis that was blunted by this superoxide donor. XO-mediated superoxide production reduced IFN-γ synthesis levels similar to BDC-2.5 CD4 T cells (Fig. 6A). The enhanced levels of TNF-α by stimulated BDC-2.5.Ncf1^m1J splenocytes were also reduced to wild-type levels with XO addition (Fig. 6B). Synthesis of IL-17A was threefold enhanced with mimotope stimulation, but XO addition only slightly attenuated IL-17A production by BDC-2.5.Ncf1^m1J in comparison with BDC-2.5 (Fig. 6C). Further demonstrating the selective decreases in cytokine and chemokine responses by XO addition to stimulated BDC-2.5.Ncf1^m1J splenocytes, synthesis of the T-cell growth factor IL-2 remained unchanged with this superoxide donor (Fig. 6D). CCL5 synthesis was regulated by redox status, as XO induced profound reductions in CCL5 by both stimulated BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes (Fig. 6E). Similar results were obtained with the chemokine CXCL10, with levels below the limit of detection after XO addition (data not shown). In addition to an elevation in effector cytokine and chemokine synthesis, antigen-stimulated BDC-2.5.Ncf1^m1J splenocytes also displayed a twofold increase in T-cell proliferation as demonstrated by formazan release in an MTT assay, in contrast to BDC-2.5 splenocytes (Supplementary Fig. 9). Interestingly, unlike effector cytokine synthesis, addition of XO did not decrease T-cell proliferation of mimotope-stimulated BDC-2.5 and BDC-2.5.Ncf1^m1J splenocytes.

For ascertainment of whether exacerbations in IFN-γ by purified BDC-2.5.Ncf1^m1J CD4 T cells were similarly attributed to defects in NOX-derived superoxide, XO was added to α-CD3ε– and α-CD28–stimulated BDC-2.5 and BDC-2.5.Ncf1^m1J CD4 T cells (Fig. 6F–H). As anticipated, polyclonal stimulation of BDC-2.5.Ncf1^m1J CD4 T cells elicited a twofold elevation in IFN-γ, but exogenous XO addition dampened IFN-γ production by 1.4-fold to approximately BDC-2.5 levels (Fig. 6F). Similar to BDC-2.5.Ncf1^m1J splenocytes, modulation of redox status did not affect Th17 cytokine responses with BDC-2.5.Ncf1^m1J CD4 T cells, as production of IL-17A was unchanged with XO addition (Fig. 6G). Interestingly, XO addition enhanced IL-2 synthesis with BDC-2.5 and BDC-2.5.Ncf1^m1J CD4 T cells (Fig. 6H). XO did not negatively impact the viability of BDC-2.5 or BDC-2.5.Ncf1^m1J splenocytes or CD4 T cells, as determined by an MTT assay (data not shown).

For determination of the redox-dependent mechanism of elevated Th1 cytokine responses by BDC-2.5.Ncf1^m1J (Fig. 3B; Supplementary Fig. 4), expression of IL-12Rβ2, a Th1 signal transducer, (Fig. 7 and Supplementary Fig. 10), and the phosphorylation status of STAT4, a transcription factor essential for Th1 responses (Fig. 8 and Supplementary Fig. 11), were examined via flow cytometry and immunofluorescence, respectively. Mimotope stimulation elicited a 1.5-fold increase in IL-12Rβ2⁺, CD4⁺ T cells by superoxide-deficient splenocytes and a 1.35-fold elevation in gMFI (geometric mean fluorescence intensity) (Fig. 7). Upon addition of xanthine oxidase, IL-12Rβ2 expression was attenuated by mimotope-stimulated superoxide-deficient CD4 T cells (Fig. 7). Similar to IL-12Rβ2, P-STAT4 (Y693), an essential Th1 transcription factor, was significantly enhanced by BDC-2.5.Ncf1^m1J (Fig. 8 and Supplementary Fig. 11). With exogenous superoxide addition via XO, P-STAT4 (Y693) was severely diminished to levels similar to NOX-intact BDC-2.5 cells (Fig. 8). Thus, in addition to functioning as a proinflammatory-derived third signal, ROS can attenuate Th1 proinflammatory cytokine responses by lessening IL-12Rβ2 signaling and STAT4 activation.

In T1D, ROS directly destroy β-cells and induce proinflammatory cytokines that further propagate β-cell destruction by maturing autoreactive adaptive immune effector responses (35,36). Reports by our laboratory and others have demonstrated that systemic ablation of innate immune-derived ROS results in antigen-specific hyporesponsiveness (8,15). We previously demonstrated that NOD mice defective in NOX-derived superoxide via the spontaneous mutation Ncf1^m1J were protected against T1D onset due to skewed CD4 T-cell responses (10), reduced CD8 T-cell cytotoxic capacity (18), lessened antiviral proinflammatory responses (17,27), and deviation to an alternatively-activated M2 macrophage phenotype (27). To further define the role of ROS in diabetogenic CD4 T-cell responses, we generated the BDC-2.5.Ncf1^m1J mouse strain, possessing autoreactive and superoxide-deficient CD4 T cells. Because of the unique T1D resistance displayed by NOD.Ncf1^m1J mice, we hypothesized that BDC-2.5.Ncf1^m1J CD4 T cells would exhibit decreased effector cytokine synthesis and diabetogenicity. Surprisingly, unlike the reduced Th1 cytokine profile displayed by NOD.Ncf1^m1J, activated BDC-2.5.Ncf1^m1J splenocytes and CD4 T cells displayed elevated synthesis of proinflammatory chemokines, Th1 and Th17 cytokines; enhanced diabetogenicity; and reduced Treg suppressive function.

Although initially apparently contradictory, these results are supported by other lines of evidence in additional autoimmune diseases. Deficiencies in NOX-derived superoxide have been shown to enhance autoimmunity in animal models of autoimmune arthritis (37,38), multiple sclerosis (10,37), Crohn disease (39), systemic lupus erythematosus (40), and psoriasis (41). With the aid of the NOD.BDC-2.5.Ncf1^m1J high-affinity TCR transgenic murine model, we can now include NOX-derived superoxide as a novel immunoregulatory molecule of diabetogenic CD4 T cell responses and pancreatic β-cell destruction in T1D. In addition to enhanced proinflammatory cytokine and chemokine synthesis, we report that BDC-2.5.Ncf1^m1J CD4 T cells were more diabetogenic during spontaneous and adoptive transfer of T1D, due to trending reductions in Treg-mediated immunosuppression. Our data corroborate a previous report of defective immunoregulation by Tregs possessing the Ncf1^m1J mutation (42). Additionally, rodents bearing a spontaneous mutation in the Ncf1^m1J gene were more susceptible to autoimmune arthritis (37), and administration of the ROS donor phytol curbed arthritogenic T-cell responses (43). Similarly, we demonstrated that T1D-resistant NOD.Ncf1^m1J mice exhibited enhanced sensitivity to myelin oligodendrocyte glycoprotein (MOG_35–55)-induced experimental autoimmune encephalomyelitis (10). Analogous to murine models, chronic granulomatous disease patients, harboring mutations in the phagocyte NOX complex, possessed an elevated type I interferon gene signature, increased synthesis of proinflammatory cytokines (i.e., IL-17A) and chemokines (40), macrophages that were profoundly less efficient in inducing Tregs compared with healthy controls (44), and an elevated risk of developing autoimmunity (45–47). These results provide evidence that NOX-derived superoxide can negatively regulate effector T-cell and B-cell responses in various autoimmune diseases.

Recent studies by our laboratory demonstrated that NOD.Ncf1^m1J CD4 T cells were equally pathogenic as NOD CD4 T cells, as cotransfers of purified NOD.Ncf1^m1J CD4 T cells with NOD CD8 T cells resulted in rapid T1D-onset kinetics (18). These results provide evidence that NOX-derived superoxide is involved in peripheral tolerance and elicits a redox-dependent signaling cue to halt diabetogenic CD4 T cells. ROS synthesis negatively impacted IL-12Rβ2 signaling and STAT4 activation, providing mechanistic insight to the process whereby ROS can blunt CD4 T-cell activation. Our results provide evidence that moderate increases in oxidative stress may be beneficial as an essential checkpoint to curtail T cell–mediated diseases dominated by Th1 cytokine responses, such as T1D.

Here, we demonstrate that NOX-deficient CD4 T cells bearing a high-affinity autoreactive T-cell receptor exhibited elevated IFN-γ production. This is in conflict with our previous observation that IFN-γ was blunted in the absence of NOX-derived superoxide by polyclonal NOD.Ncf1^m1J CD4 T cells (10). This difference in Th1 responses could be attributed to inherent disparities in T-cell phenotype. NOD CD4 T-cell responses are polyclonal, characterized by distinct avidities to different autoantigens, while BDC-2.5 CD4 T cells are inherently skewed to a Th1-like phenotype and recognize a single autoantigen, chromogranin A (21,22). NOD mice spontaneously develop T1D, whereas BDC-2.5 mice exhibit a low incidence due to immunosuppressive Treg cells (34). Enhanced Th1 cytokine responses have not only been observed within BDC-2.5.Ncf1^m1J CD4 T cells; our preliminary data also demonstrate that BDC-6.9, another diabetogenic CD4 T cell clone that recognizes islet amyloid polypeptide (IAPP) (48), exhibits an enhanced Th1 and Th17 cytokine response in the absence of NOX-derived superoxide (L.E. Padgett, N.N. Morgan, and H.M. Tse, unpublished data). These monoclonal TCR transgenic murine models of T1D have enabled us to uncover the importance of ROS in T-cell effector inhibition, and thus, similar to rheumatoid arthritis (RA) (37,38), multiple sclerosis (10,37), Crohn disease, psoriasis (41), and systemic lupus erythematosus (40), increasing the production of free radicals by CD4 T cells may be efficacious in T1D.

The disparate in vitro Th1 responses by NOD.Ncf1^m1J and BDC-2.5.Ncf1^m1J could additionally be attributed to differences in inherent levels of ROS and sensitivity of immune cells to redox-dependent signaling. Illustrative of the inherent dissimilarity in redox status, BDC-2.5.Ncf1^m1J CD4 T cells displayed elevated nitrite, but no difference was observed in NOD and NOD.Ncf1^m1J CD4 T cells (10). A profound mechanism by which ROS influence T-cell effector responses at the immunological synapse is by reversibly oxidizing thiol residues located within TCR adaptor molecules. TCR adaptor molecules are regulated by redox status and, more importantly, T-bet, the master transcription factor involved in Th1 cytokine responses, is also influenced by ROS synthesis (10). Future studies will dissect differences in redox-dependent signaling by exploring TCR adaptor molecule oxidation status within NOD.Ncf1^m1J and BDC-2.5.Ncf1^m1J CD4 T cells, as unraveling these may reveal novel targets for T1D protection and prevention.

By using T1D-resistant NOD.Ncf1^m1J mice lacking the ability to generate NOX-derived superoxide (10), our laboratory has defined an influential role for free radicals on the maturation of autoimmune responses in T1D. We have mechanistically determined that the induction of innate immune signaling pathways and differentiation of islet-resident and islet-infiltrating proinflammatory M1 macrophages are dependent on NOX-derived superoxide during spontaneous T1D (10,17,18,27). The absence of superoxide can skew macrophage responses toward an alternatively activated M2 phenotype and elicit β-cell protection. However, the role of NOX-derived superoxide on T cells is more complex. Free radicals may have distinct effects on T-cell subsets, in addition to expansion, contraction, and memory formation. CD8 T cells from NOD.Ncf1^m1J mice are impaired and unable to transfer T1D (10,18). These observations would suggest that CD8 T cells are highly dependent on NOX-derived superoxide as a signaling mediator for cytotoxic lymphocyte maturation. CD4 T cells from NOD.Ncf1^m1J mice also require superoxide as a third signal to efficiently generate Th1 cytokine-producing cells (10), but within a subset of high-affinity autoreactive T cells (BDC-2.5, BDC-6.9), NOX-derived superoxide may also be necessary as a checkpoint to quell T-cell effector responses, facilitate T-cell contraction, and/or restore peripheral tolerance in T1D. Our studies further define the synergism of redox biology in autoimmune dysregulation in T1D and may lead to the development of novel immunotherapies to delay disease progression.

Acknowledgments. The authors are grateful to Stew New from the Department of Microbiology at University of Alabama at Birmingham (UAB) for technical assistance with ImageJ and immunofluorescence assays. The authors thank Ashley Burg (Department of Microbiology) and Shaonli Das (Department of Microbiology), Chad Hunter (Department of Medicine), and Sasanka Ramanadham (Department of Cell, Developmental and Integrative Biology) from UAB for critical reading of the manuscript.

Funding. This work was supported by a National Institutes of Health (NIH)/National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) R01 award (DK-099550), an American Diabetes Association Career Development Award (7-12-CD-11), a P30 Pilot Feasibility award from the UAB Comprehensive Diabetes Center/Diabetes Research Training Center, an NIH National Institute of Allergy and Infectious Diseases (5T32AI007051-35) Immunologic Diseases and Basic Immunology T32 training grant (to L.E.P.), and a Travel for Techniques Program Award from the American Association for Immunologists, Inc. (to H.M.T.). Further support was provided by an NIH/NIDDK RO1 award (DK-074656) and a grant from the JDRF (to C.E.M.). This research was partially supported by the Intramural Research Program of the National Institute of Environmental Health Sciences/NIH. The following core facilities of UAB were used to generate data for the manuscript: Animal Resources Program (G20RR025858, Sam Cartner); the Comprehensive Arthritis, Musculoskeletal, and Autoimmunity Center: Analytic and Preparative Cytometry Facility (P30 AR48311, John D. Mountz); and the Comprehensive Arthritis, Musculoskeletal, and Autoimmunity Center: Epitope Recognition Immunoreagent Core (P30 AR48311, Mary Ann Accavitti-Loper).

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

Author Contributions. L.E.P. researched data, wrote the manuscript, and reviewed and edited the manuscript. B.A. researched data and reviewed and edited the manuscript. C.L. researched data. D.G. researched data and reviewed and edited the manuscript. R.P.M. and J.D.P. reviewed and edited the manuscript. C.E.M. reviewed and edited the manuscript and contributed to discussion. H.M.T. researched data, wrote the manuscript, and reviewed and edited the manuscript. H.M.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.