d-mannose induces regulatory T cells and suppresses immunopathology

Abstract

d-mannose, a C-2 epimer of glucose, exists naturally in many plants and fruits, and is found in human blood at concentrations less than one-fiftieth of that of glucose. However, although the roles of glucose in T cell metabolism, diabetes and obesity are well characterized, the function of d-mannose in T cell immune responses remains unknown. Here we show that supraphysiological levels of d-mannose safely achievable by drinking-water supplementation suppressed immunopathology in mouse models of autoimmune diabetes and airway inflammation, and increased the proportion of Foxp3⁺ regulatory T cells (T_reg cells) in mice. In vitro, d-mannose stimulated T_reg cell differentiation in human and mouse cells by promoting TGF-β activation, which in turn was mediated by upregulation of integrin α_vβ₈ and reactive oxygen species generated by increased fatty acid oxidation. This previously unrecognized immunoregulatory function of d-mannose may have clinical applications for immunopathology.

d-mannose is a C-2 epimer of glucose and occurs naturally in many plants and fruits, especially cranberries. Glucose is known to have central roles in energy generation, storage, and regulation in the cell, as well as pathogenic roles in diabetes and obesity. However, d-mannose, which has a physiological blood concentration less than one-fiftieth of that of glucose¹, has not received much attention. Nevertheless, d-mannose is important in the glycosylation of certain proteins^2,3. d-mannose has been reported to be mainly beneficial in human disease states. For example, a d-mannose supplement has been shown to be an effective therapy for congenital disorders of glycosylation type Ib^2,4. It has also been used as a non-antibiotic treatment for bacterial urinary tract infection in animals⁵ and humans⁶, with a proposed mechanism of binding to the type 1 pili of enteric bacteria and thereby blocking their adhesion to uroepithelial cells⁷. It is well established that glucose has a vital role in immune cell activity, especially in T cell activation and differentiation^8–10. However, it is unknown whether d-mannose has any effect on T cells and immune responses. Here we show that oral administration of d-mannose suppressed immunopathology in models of autoimmune diabetes and airway inflammation. d-mannose was able to induce the generation of T_reg cells from naive CD4⁺ T cells by promoting activation of the latent form of TGF-β. d-mannose-mediated TGF-β activation required integrin α_vβ₈ and reactive oxygen species (ROS) in T cells. Moreover, we observed that d-mannose decreased glycolysis but increased fatty acid oxidation in T cells. We conclude that collectively, these actions drive immune responses toward an immunoregulatory phenotype and subsequent tolerance.

RESULTS

d-mannose induces T_reg cells

To determine whether d-mannose has a role in T cell activation, we cultured naive mouse CD4⁺CD25⁻ T cells in medium supplemented with mannose or other sugars in the presence of T cell receptor (TCR) stimulation. d-mannose affected neither the T-cell-activation-associated markers nor apoptosis (Supplementary Fig. 1a,b). However, d-mannose suppressed T cell proliferation (Fig. 1a), resulting in lower absolute cell numbers after 48–72 h of culture. In d-mannose-treated T cells, levels of mRNAs associated with type 1 helper T cells (T_H1 cells) (Ifng and Il2), type 2 helper T cells (T_H2 cells) (Il4 and Il13) and Il6 were lower than in control cells, yet amounts of Il17a and Il10 mRNA remained unchanged (Supplementary Fig. 2a–g). d-mannose induced the generation of significantly more Foxp3⁺ T_reg cells and higher amounts of Foxp3 mRNA from naive CD4⁺CD25⁻ T cells compared with the control medium (Fig. 1b–d), whereas the absolute number of CD4⁺Foxp3⁻ non-T_reg cells among d-mannose-treated cells was lower at 72 h after culture (Supplementary Fig. 2h). Similarly, numbers of T_reg cells were higher in d-mannose-treated TCR-stimulated cultures of naive CD4⁺CD25⁻GFP⁻(Foxp3⁻) T cells isolated from transgenic mice expressing Foxp3–GFP (Fig. 1e). The induction of T_reg cells was dependent on the dose of d-mannose, and similar doses of glucose did not increase T_reg cell generation (Supplementary Fig. 3a). Similarly, stimulation of naive CD4⁺ T cells by d-mannose in cultures with soluble anti-CD3 and splenic antigen-presenting cells (APCs) also significantly increased T_reg cell generation (Fig. 1f, Supplementary Fig. 3b). Thus, d-mannose specifically induced Foxp3 expression and T_reg cell fate in naive T cells in vitro.

Figure 1.

d-mannose induces T_reg cell differentiation in vitro and in vivo. (a–d) We cultured CD4⁺CD25⁻ (naive) T cells from spleen and peripheral lymph nodes of C57BL/6 mice with anti-CD3 and anti-CD28 for up to 3 d in 10% FBS glucose-free DMEM complete medium (CTRL) or with added 25 mM mannose (Man) or other sugars (Fru, fructose; Glu, glucose; Gal, galactose). We measured proliferation by labeling with CFSE dye. (a) Representative FACS plots gated on unproliferated cells among CD4⁺ T cells cultured for 24, 48 or 72 h. Values in the upper left corners are mean ± s.d. Bars indicate the gate of unproliferated cells. (b,c) Representative FACS plots (b) and the frequency of CD25⁺Foxp3⁺ T_reg cells among CD4⁺ T cells (c, top) after 3 d in culture. (c, bottom) Absolute numbers of CD25⁺Foxp3⁺ T_reg cells. In the FACS plots, numbers adjacent to outlines indicate the percentage of cells in the gate. (d) Foxp3 mRNA expression at 24 h. (e) The ratio of CD4⁺CD25⁺GFP⁺(Foxp3⁺) T_reg cells among CD4⁺CD25⁻GFP⁻(Foxp3⁻) T cells from spleens and peripheral lymph nodes of transgenic mice expressing Foxp3–GFP cultured with anti-CD3 and anti-CD28 in 25 mM mannose-supplemented DMEM or control medium for 3 d. (f) The frequency of CD25⁺Foxp3⁺ T_reg cells among naive T cells from spleens and lymph nodes of C57BL/6 mice after 3 d of culture with soluble anti-CD3 plus APCs. (g) KJ1–26⁺CD4⁺CD25⁻ naive T cells from spleens and lymph nodes of DO11.10 TCR-transgenic Rag2^−/− mice were cultured with OVAp_323–339 plus APCs. The plot shows the frequency of KJ1–26⁺CD25⁺Foxp3⁺ T_reg cells after 4 d of culture. (h) KJ1–26⁺CD4⁺CD25⁻ naive T cells from spleens and lymph nodes of DO11.10 TCR-transgenic Rag2^−/− mice were adoptively transferred into BALB/cJ mice. The plot shows the frequency of KJ1–26⁺CD4⁺Foxp3⁺ T_reg cells among adoptively transferred cells in the indicated organs after 5 d of ovalbumin gavage (n = 3). PP, Peyer’s patches; MLN, mesenteric lymph nodes; LPL, lamina propria. Data are presented as mean ± s.d. In box plots (c–g), center lines indicate the median, box limits represent the upper and lower quartiles, and whiskers extend to the minimum and maximum values. Data were analyzed by one-way analysis of variance (ANOVA) with Tukey’s post hoc test (c,h) or unpaired two-tailed Student’s t-test (d–g). *P < 0.05, **P < 0.01. Data are pooled from three (e,g) or five (c,d,f) experiments or are representative of three (a,h) or five (b) independent experiments.

Extending this finding to antigen-specific stimulation, d-mannose converted naive CD4⁺CD25⁻KJ1–26⁺ TCR-transgenic T cells into Foxp3⁺ T_reg cells in response to their cognate antigen OVAp_323–339 (Fig. 1g, Supplementary Fig. 3c). Importantly, d-mannose induced T_reg cells in vivo. We administered d-mannose in drinking water to BALB/cJ mice for 2 weeks, and then subjected the mice to adoptive transfer of KJ1–26⁺CD4⁺CD25⁻ naive T cells and ovalbumin to induce T_reg cell generation in a mucosal T_reg cell induction model^11,12 (Supplementary Fig. 4a). d-mannose significantly increased the frequency of T_reg cells in transferred TCR-transgenic T cells in response to ovalbumin in Peyer’s patches, mesenteric lymph nodes, lamina propria and spleen (Fig. 1h) without affecting the host T_reg cell frequency (Supplementary Fig. 4b). However, there was no significant decrease in the number of IFN-γ⁺ and IL-17A⁺ T cells among transferred DO11.10 TCR-transgenic T cells (Supplementary Fig. 4c,d). Taken together, our findings show that d-mannose consistently promoted T_reg cell differentiation.

d-mannose-induced T_reg cells exhibit suppressive capacity

We next investigated the function of T_reg cells induced by d-mannose. In vitro, d-mannose-induced T_reg cells potently suppressed the proliferation of naive CD4⁺ T cells similarly to TGF-β1-induced T_reg cells and freshly isolated CD4⁺CD25⁺GFP⁺ T_reg cells from the spleens of Foxp3–GFP transgenic mice in vitro¹³ (Supplementary Fig. 5a). To investigate the immunosuppressive function of mannose-induced T_reg cells in vivo, we used a T-cell-transfer model of colitis¹⁴. We transferred CD4⁺CD45RB^hi T cells into Rag1^−/− mice, with or without cotransfer of d-mannose-induced T_reg cells, TGF-β1-induced T_reg cells or CD4⁺CD25⁺GFP⁺ T_reg cells from Foxp3–GFP mice. All three types of T_reg cells effectively prevented weight loss and suppressed colon inflammation (Supplementary Fig. 5b,c). In addition, d-mannose-induced T_reg cells exhibited a DNA-methylation pattern similar to that of TGF-β1-induced T_reg cells (data not shown)¹⁵. These results suggest that mannose-induced T_reg cells possess suppressive capacity.

d-mannose induces T_reg cell differentiation by TGF-β activation

As TGF-β signaling is critical for T_reg cell generation^16–18, we investigated the role of TGF-β in this process. We first determined that d-mannose treatment enhanced TGF-β signal transduction, as demonstrated by increased expression of TGF-β-inducible^19,20 Smad7 and Fos mRNAs (Supplementary Fig. 6a,b). Pharmacological or antibody-mediated blockade of TGF-β signaling completely abrogated the increase in Foxp3 mRNA expression and the consequent increase in the number of T_reg cells induced by d-mannose in wild-type naive CD4⁺ T cells (Fig. 2a,b, Supplementary Fig. 6c) and in naive KJ1–26⁺CD4⁺ T cells cultured with OVAp_323–339 (Supplementary Fig. 6d). Interestingly, naive CD4⁺CD25⁻ T cells from mice deficient in TGF-β receptor I (Tgfbr1^f/fCD4-Cre)^18,21 or II (Tgfbr2^f/fER-Cre, plus tamoxifen treatment)²² did not differentiate into T_reg cells at all in response to d-mannose stimulation (Fig. 2c,d). Consistently, CD4⁺ naive T cells deficient in Smad3 (Smad3^−/−)²³, a critical mediator downstream of TGF-β^24,25, showed a significant reduction in amounts of d-mannose-induced T_reg cells (Supplementary Fig. 6e; P < 0.01). TGF-β signaling was also required for d-mannose-induced T_reg cell generation in vivo (Supplementary Fig. 7a), as injection of anti-TGF-β abolished the increase in the number of T_reg cells among adoptively transferred KJ1–26⁺CD4⁺ T cells induced by d-mannose in BALB/cJ mice (Fig. 2e), without affecting the frequency of host T_reg cells (Supplementary Fig. 7b). Thus, we conclude that TGF-β signaling is required for d-mannose induction of T_reg cells in culture and in vivo.

Figure 2.

d-mannose induces T_reg cell differentiation via activation of TGF-β. (a,b) CD4⁺CD25⁻ (naive) T cells from spleens and peripheral lymph nodes of C57BL/6 mice were cultured with anti-CD3 and anti-CD28, as well as latent TGF-β1, anti-TGF-β or SB431542 (an inhibitor of TGF-β receptor). “Med” indicates TCR stimulation only. (a) Foxp3 mRNA expression at 24 h. (b) The frequency of CD25⁺Foxp3⁺ T_reg cells among CD4⁺ T cells after 3 d of culture. (c,d) Naive CD4⁺ T cells deficient in TβRI (c) or TβRII (d) from spleens and lymph nodes of Tgfbr1^f/fCD4-Cre or tamoxifen-treated Tgfbr2^f/fER-Cre (Tamoxifen) mice were cultured with anti-CD3 and anti-CD28, with or without latent TGF-β1. Tamoxifen was dissolved with sunflower oil (Oil). The plots show the frequency of CD25⁺Foxp3⁺ T_reg cells after 3 d of culture. (e) KJ1–26⁺CD4⁺CD25⁻ naive T cells from spleens and lymph nodes of DO11.10 TCR-transgenic Rag2^−/− mice were adoptively transferred into BALB/cJ mice. The plot shows the frequency of KJ1–26⁺CD4⁺Foxp3⁺ T_reg cells among adoptively transferred cells in the indicated organs after 5 d of ovalbumin gavage (n = 5). Data are presented as mean ± s.d. (f) Naive CD4⁺ T cells purified from human peripheral blood mononuclear cells were cultured with anti-CD3, anti-CD28 and IL-2, with or without anti-TGF-β and SB431542. The plot shows the frequency of CD25^hiFoxp3⁺ T_reg cells after 4 d of culture. (g) Naive CD4⁺ T cells from spleens and lymph nodes of C57BL/6 mice were cultured with anti-CD3 and anti-CD28, with or without latent TGF-β1 or anti-TGF-β plus SB431542. The plot shows the frequency of CD25⁺Foxp3⁺ T_reg cells after 3 d of culture. (h) Naive CD4⁺ T cells from spleens and lymph nodes of C57BL/6 mice were cultured with anti-CD3 and anti-CD28 plus different doses of latent TGF-β1. The plot shows the frequency of CD25⁺Foxp3⁺ T_reg cells after 3 d of culture (mean ± s.d.). In box plots (a–d,f,g), center lines indicate the median, limits represent the upper and lower quartiles, and whiskers extend to the minimum and maximum values. Data were analyzed by one-way ANOVA with Tukey’s post hoc test (a–g) or by unpaired two-tailed Student’s t-test (h). **P < 0.01. Data are pooled from five (a–d,f,g) or three (h) experiments or are representative of two (e) independent experiments.

We also investigated whether d-mannose promotes the generation of human T_reg cells. d-mannose treatment indeed significantly upregulated the frequency of human CD4⁺CD25^hiFoxp3⁺ T_reg cells²⁶ compared with the control condition (Fig. 2f). Blockade of TGF-β signaling significantly reduced this effect (Fig. 2f). d-mannose-induced human T_reg cells had suppressive activity toward normal human CD4⁺ T cell proliferation, similarly to TGF-β1-induced T_reg cells in standard suppression assays in vitro¹³ (Supplementary Fig. 8a,b). Thus, as in mouse T cells, d-mannose also drove T_reg cell generation in human naive CD4⁺ T cells via a TGF-β-dependent mechanism.

We next determined which aspects of TGF-β signaling^19,27,28 are influenced by d-mannose. d-mannose did not change the levels of Tgfb1 mRNA or total TGF-β1 protein compared with pre-supplementation levels, which suggested that there was no effect on TGF-β transcription or protein synthesis (Supplementary Fig. 9a). d-mannose treatment in the presence of TCR stimulation slightly upregulated the expression of mRNA for TGF-β receptor I (TβRI) and II (TβRII) compared with TCR stimulation alone (Supplementary Fig. 9b,c).

TGF-β is produced as a latent form in complex with latent associated protein (LAP)^28,29. d-mannose increased T_reg cell numbers only slightly in the presence of bioactive TGF-β1 (Supplementary Fig. 9d), which suggests that increased expression of the TGF-β receptors is not the main mechanism of D-mannose action. In contrast, d-mannose significantly potentiated T_reg cell generation in cultures supplemented with LAP–TGF-β1 (Fig. 2b–d,g). This was further shown by significant increases in amounts of Foxp3 mRNA (Fig. 2a) in T cells at 24 h, and by the dose-dependence of LAP–TGF-β1 in d-mannose-treated cells (Fig. 2h). Similarly, stimulation of wild-type naive CD4⁺ T cells by anti-CD3 and APCs or of TCR-transgenic KJ1–26⁺CD4⁺CD25⁻ naive T cells by OVAp_323–339 and APCs, together with d-mannose and LAP–TGF-β1 treatment, also significantly increased (P < 0.01) numbers of T_reg cells compared with culture without d-mannose (Supplementary Fig. 6c,d). Moreover, when we cultured naive CD4⁺CD25⁻ T cells in X-vivo-20 medium, which does not contain TGF-β, as it is serum-free, d-mannose promoted T_reg cell generation only in the presence of LAP–TGF-β1 (Supplementary Fig. 9e,f).

To confirm that d-mannose-mediated activation of LAP–TGF-β1 enhanced TGF-β signal transduction, we determined that the amount of phospho-Smad3 increased in T cells in response to treatment with d-mannose and LAP–TGF-β1 (Supplementary Fig. 9g). Importantly, the d-mannose- and LAP–TGF-β1-mediated increase in the number of T_reg cells was completely abolished when TGF-β signaling was blocked (Fig. 2g), or when naive CD4⁺ T cells lacked either TβRI or TβRII expression (Fig. 2c,d). Naive Smad3^−/− CD4⁺ T cells also showed a 2.5-fold decrease in numbers of d-mannose/LAP–TGF-β1 induced T_reg cells (Supplementary Fig. 6e). Finally, we investigated whether d-mannose could induce substantial numbers of T_reg cells at physiologically relevant concentrations (~1 mM), by culturing naive CD4⁺ T cells with LAP–TGF-β1 in vitro in the presence of different doses of d-mannose (Supplementary Fig. 9h). Taken together, our findings suggest that d-mannose induces the generation of T_reg cells by augmenting TGF-β signaling via activation of latent TGF-β.

Integrin α_vβ ₈ and ROS are required for d-mannose-induced T_reg cell generation

Several molecules have been suggested to be involved in TGF-β activation in immune cells, including integrin α_vβ₈ (encoded by Itgav and Itgb8 subunits)^30–32 and reactive oxygen species (ROS)^33,34. d-mannose stimulation increased the expression of Itgav and Itgb8 mRNA in naive CD4⁺ T cells (Fig. 3a,b) compared with that in unstimulated T cells. We then activated naive CD4⁺CD25⁻ integrin β₈–deficient T cells³² in vitro in the presence of d-mannose and observed that the loss of integrin β₈ significantly reduced d-mannose-induced T_reg cell generation even in the absence of exogenous LAP–TGF-β1 (Fig. 3c). When we added exogenous LAP–TGF-β1 to the cultures, integrin β₈–deficient T cells showed a significant defect in T_reg cell generation (~50% decrease) in response to d-mannose (Fig. 3c), thus suggesting that integrin β₈ has an important role in mediating TGF-β1 activation induced by d-mannose.

Figure 3.

Integrin α_vβ₈ and ROS are required for d-mannose-mediated TGF-β1 activation and T_reg cell generation. (a,b) CD4⁺CD25⁻ (naive) T cells from spleens and peripheral lymph nodes of C57BL/6 mice were cultured with anti-CD3 and anti-CD28. The plots show relative Itgb8 (a) and Itgav (b) mRNA expression at 24 h. (c) CD4⁺CD25⁻ (naive) T cells from spleens and peripheral lymph nodes of control or Itgb8^f/fCD4-Cre mice were cultured with or without latent TGF-β1 or NAC. The plot shows the frequency of CD25⁺Foxp3⁺ T_reg cells after 3 d of culture. (d) CD4⁺CD25⁻ (naive) T cells from spleens and peripheral lymph nodes of C57BL/6 mice were cultured with anti-CD3 and anti-CD28. The plot shows ROS expression after 24 h of culture. In box plots (a–c), center lines indicate the median, limits represent the upper and lower quartiles, and whiskers extend to the minimum and maximum values. Data were analyzed by one-way ANOVA with Tukey’s post hoc test (c) or by unpaired two-tailed Student’s t-test (a,b). *P < 0.05, **P < 0.01. Data are pooled from three to five (a–c) experiments or are representative of three (d) independent experiments.

As the action of integrin β₈ could not totally account for all TGF-β1 activation, we also determined that ROS participated in TGF-β1 activation by d-mannose. ROS can be produced in T cells after TCR activation^35,36. We found that d-mannose treatment increased ROS production in T cells compared with TCR stimulation alone (Fig. 3d). Importantly, blockade of ROS activity by N-acetyl-L-cysteine (NAC)³⁷ also significantly reduced numbers of d-mannose-induced T_reg cells in the presence of LAP–TGF-β1 (Fig. 3c). Notably, neutralization of ROS by NAC in integrin β₈–deficient naive CD4⁺ T cells further decreased T_reg cell generation induced by d-mannose by 70–80% (Fig. 3c), suggesting a combinatorial function involving the integrin and ROS pathways. Thus, the integrin α_vβ₈ and ROS pathways can be assumed to have independent yet complementary roles in d-mannose-mediated TGF-β1 activation and consequent T_reg cell generation in vitro.

d-mannose increases fatty acid oxidation

We next investigated how d-mannose induces ROS. After TCR activation, naive T cells switch from oxidative phosphorylation to aerobic glycolysis⁸. We hypothesized that as d-mannose and d-glucose increased ROS production to similar levels (Supplementary Fig. 10a), d-mannose might also upregulate glycolysis in T cells. To test this, we cultured naive T cells in sugarless media supplemented with pyruvate and L-glutamine and measured their extracellular acidification rate in response to d-glucose or d-mannose. Unexpectedly, cells supplemented with d-mannose had a markedly lower acidification rate at maximal respiration compared with cells cultured in d-glucose (Supplementary Fig. 10b). This indicated a reduced capacity to utilize glycolysis in response to stress.

As fatty acid oxidation (FAO) can generate ROS^38,39, we hypothesized that d-mannose might induce ROS via that pathway. To investigate this, we measured the oxygen consumption rate (OCR) of naive CD4⁺ T cells cultured in either d-mannose or d-glucose. Although both oxidative phosphorylation and FAO occur in the mitochondria and contribute to the OCR, only FAO is affected by etomoxir, an inhibitor of CPT1, the enzyme responsible for the transport of fatty acids into mitochondria. Cells cultured in d-mannose had significantly higher (P < 0.01) OCRs (~25%) at maximal respiration than cells cultured in d-glucose (Supplementary Fig. 10c). However, simultaneous treatment with etomoxir led to a decrease in the OCR of mannose-cultured cells, whereas no such decrease was observed in etomoxir-treated glucose-cultured cells, indicating that a portion of the OCR in mannose-treated cells can be attributed to active FAO, especially where spare capacity is required (e.g., during proliferation) (Supplementary Fig. 10c). The greater utilization of FAO in mannose-cultured T cells might explain the higher ROS levels observed in these cells, although how d-mannose-induced FAO drives ROS production remains unknown. Nonetheless, the increased amount of ROS contributes to the activation of latent TGF-β during the generation of T_reg cells. Thus, we suggest that T cells cultured with d-mannose preferentially use FAO, which can result in higher ROS levels and thus greater TGF-β activation.

d-mannose suppresses type 1 diabetes in NOD mice

We next determined whether d-mannose has beneficial effects in autoimmunity, using as a type 1 diabetes model non-obese diabetic (NOD) mice, in which pathology involves T_reg cell defects⁴⁰. NOD mice received d-mannose in their drinking water starting at 7.5 weeks of age, when the mice are considered prediabetic, and at which point the inflammatory process has just been initiated but blood glucose levels are still within the normal range⁴⁰ (Supplementary Fig. 11a). As expected, control (non-d-mannose-treated) NOD mice started to develop diabetes at about 12–13 weeks of age, and 80–90% became diabetic by the age of 23 weeks (Fig. 4a). However, most of the NOD mice supplemented with d-mannose were diabetes-free through 23 weeks of age (Fig. 4a). Consistent with protection from diabetes, d-mannose-treated mice showed considerably less insulitis and more preserved islets compared with controls (Fig. 4b,c).

Figure 4.

d-mannose suppresses type 1 diabetes in NOD mice. (a–c) Female NOD mice were supplemented with d-mannose in drinking water from 7.5 weeks of age and were subsequently checked for the development of type 1 diabetes. (a) The frequency of mice without type 1 diabetes in the indicated groups over time (n = 20). (b) Representative histology sections of pancreas from the mice in a. White arrows indicate pancreatic islands. (c) The frequency of islets with grade X insulitis in the indicated groups. The stages refer to diabetes progression (also applies to j). (d–g) Female NOD mice were treated with d-mannose in drinking water daily from 7.5 weeks of age, and were euthanized when the mice were 14 weeks old. The plots in d–f show the frequencies of CD25⁺Foxp3⁺ T_reg cells (d), IFN-γ⁺CD4⁺ T cells (e) and IFN-γ⁺CD8⁺ T cells (f) in the spleens and DLNs of the mice. (g) Splenocytes from the NOD mice were cultured with GAD65 peptide (1 μg/ml). The plot shows the concentration of IFN-γ in the culture medium after 3 d, as determined by ELISA. (h–k) NOD mice with blood glucose levels between 140 and 160 mg/dL were supplemented with d-mannose in drinking water and then monitored for the progression of diabetes. (h) The levels of blood glucose in individual mice from the indicated groups over time (n = 7–8). (i) Representative histology sections of pancreas from the mice in h. White arrows indicate pancreatic islands. (j) The frequency of islets with grade X insulitis in the indicated groups. (k) The frequencies of CD25⁺Foxp3⁺ T_reg cells, CD4⁺IFN-γ⁺ T cells, and CD8⁺IFN-γ⁺ T cells in the pancreases of NOD mice. Summary data are presented as mean ± s.e.m. (d–f,g,k). *P < 0.05, **P < 0.01, Mantel–Cox log-rank test (a) or unpaired two-tailed Student’s t-test (d–h,k). Data are pooled from two (h,j) or four (a,c) experiments or are representative of two (g,i,k) or four (b,d–f) independent experiments.

We examined the T cell responses of these mice at 13–14 weeks of age, when the untreated NOD mice were expected to start to develop hyperglycemia. The frequency of CD4⁺CD25⁺Foxp3⁺ T_reg cells in the spleen and pancreatic draining lymph nodes (DLNs) was significantly higher in d-mannose-treated mice than in untreated mice (Fig. 4d). In contrast, the frequencies of CD4⁺IFN-γ⁺ and CD8⁺IFN-γ⁺ T cells were lower in the spleens of the NOD mice supplemented with d-mannose (Fig. 4e,f). In addition, numbers of CD4⁺IL-4⁺ (T_H2) cells were also lower in the spleens of d-mannose-treated mice (Supplementary Fig. 11b), whereas the frequencies of CD4⁺IL-17A⁺ (T_H17) and CD4⁺IL-10⁺ T cells were unchanged compared with those in controls (Supplementary Fig. 11b). In the pancreas, the frequency of CD4⁺Foxp3⁺ T_reg cells was increased and that of IFN-γ-producing CD4⁺ and CD8⁺ T cells was decreased in d-mannose-treated mice compared with controls (Supplementary Fig. 11c). In addition to the ex vivo analysis of T cells, we also examined autoantigen-specific T cell cytokine production in splenic T cells in response to restimulation with a pancreas-derived peptide, GAD65. GAD65-specific T cell IFN-γ production was significantly reduced in the spleens of d-mannose-treated NOD mice (Fig. 4g), whereas GAD65-specific IL-17 and IL-10 production did not change compared with the levels in control mice (Supplementary Fig. 11d). Moreover, treatment of NOD mice with d-mannose once they had reached prediabetic blood glucose levels of 140–160 mg/dL (Supplementary Fig. 11e) or new-onset diabetic levels of 200–230 mg/dL (refs. 41,42) (Supplementary Fig. 12a) suppressed the progression of diabetes (Fig. 4h–k, Supplementary Fig. 12b–d). Similar to what we observed in the treated prediabetic NOD mice, d-mannose treatment at these stages also significantly increased the frequency of T_reg cells and decreased the frequencies of T_H1 cells and CD8⁺IFN-γ⁺ T cells in the pancreas compared with those in control mice (Fig. 4h–k, Supplementary Fig. 12).

To study whether the increase in the amount of T_reg cells is involved in d-mannose-treatment-mediated suppression of diabetes, we depleted CD4⁺CD25⁺ T_reg cells with anti-CD25 in d-mannose-treated and untreated NOD mice (Supplementary Fig. 13a). We first confirmed that anti-CD25 sufficiently depleted T_reg cell populations in NOD mice (~50% depletion of total CD4⁺Foxp3⁺ T_reg cells; 70–80% depletion of CD4⁺CD25⁺Foxp3⁺ T_reg cells). In NOD mice, anti-CD25 did not significantly change the frequency of CD25⁺Foxp3⁻ effector cells. We found that depletion of CD4⁺CD25⁺ T_reg cells abolished the protective effects of d-mannose on diabetes development, and the same anti-CD25 treatment at that stage slightly affected the development of diabetes in untreated NOD mice (Fig. 5a). Consistently, administration of anti-CD25 reversed the d-mannose-mediated decrease in insulitis and preservation of the total number of islets in the pancreas (Fig. 5b,c). In the spleen and DLNs of d-mannose-treated mice, the increase in the frequency of CD4⁺CD25⁺Foxp3⁺ T_reg cells was abolished by anti-CD25 treatment (Fig. 5d). Consequently, the increased ratios of CD25⁺Foxp3⁺ T_reg cells to CD4⁺IFN-γ⁺ or CD8⁺IFN-γ⁺ T cells in d-mannose-treated mice were eliminated after anti-CD25 treatment (Fig. 5e,f). Furthermore, we validated a role of T_reg cells in d-mannose-mediated suppression of diabetes in NOD-Foxp3^DTR mice treated with diphtheria toxin (data not shown). These data suggest that the increased T_reg cell population is involved in d-mannose-treatment-mediated suppression of diabetes. Lastly, neutralization of endogenous TGF-β with anti-TGF-β (Supplementary Fig. 13b) in d-mannose-treated NOD mice also abrogated the suppressive effects of d-mannose on diabetes, with abolishment of the increase in T_reg cell numbers (Fig. 5g,h, Supplementary Fig. 13c,d). Together these data indicate that d-mannose suppresses the immunopathology of autoimmune diabetes in NOD mice, and that T_reg cells and TGF-β are involved in this process.

Figure 5.

T_reg cells and TGF-β are involved in d-mannose-mediated suppression of autoimmune diabetes in NOD mice. (a–f) Female NOD mice were supplemented with d-mannose in drinking water from 7.5 weeks of age, injected with anti-CD25 or isotype-control antibody twice at 13–14 weeks of age and then checked for the development of type 1 diabetes. (a) The frequency of diabetes-free mice in the indicated groups over time (n = 10). (b) Representative histology sections of pancreas from the mice in a. White arrows indicate pancreatic islands. (c) The frequency of islets with grade X insulitis in the indicated groups. Stages refer to diabetes progression (also applies to h). (d–f) The frequency of CD25⁺Foxp3⁺ T_reg cells (d), the ratio of CD25⁺Foxp3⁺ T_reg cells to IFN-γ⁺CD4⁺ T cells (e), and the ratio of CD25⁺Foxp3⁺ T_reg cells to IFN-γ⁺CD8⁺ T cells (f) in the spleens and DLNs of female NOD mice. Data in d–f are presented as mean ± s.e.m. (g,h) Female NOD mice were supplemented with mannose in drinking water from 7.5 weeks of age, injected with anti-TGF-β or isotype-control antibody once a week for 6 weeks and then checked for the development of type 1 diabetes. (g) The frequency of diabetes-free mice in the indicated groups over time (n = 10). (h) The frequency of islets with grade X insulitis in the indicated groups. *P < 0.05, **P < 0.01, ***P < 0.001, Mantel–Cox log-rank test (a,g) or one-way ANOVA with Tukey’s post hoc test (d–f). Data are pooled from two (a,c,g,h) experiments or are representative of two (b,d–f) independent experiments.

d-mannose prevents and suppresses ovalbumin-induced airway inflammation

The success of d-mannose supplementation in suppressing diabetes development in NOD mice encouraged us to investigate whether mannose-mediated T_reg cell generation has broader effects in immunopathology. We next tested d-mannose function in a model of lung airway inflammation. We used an ovalbumin-induced airway inflammation model in which KJ1–26⁺CD4⁺CD25⁻ naive T cells isolated from DO11.10 TCR-transgenic Rag2^−/− mice were adoptively transferred into BALB/cJ mice^43–45. This system allowed us to assess the conversion of KJ1–26⁺CD4⁺CD25⁻ naive T cells to T_reg cells in vivo in the airway inflammation model in response to d-mannose administration (Supplementary Fig. 14a). Indeed, we found that d-mannose treatment prevented the development of airway inflammation in the lungs, as demonstrated by considerably less infiltration of inflammatory cells and reduced mucus production in the airways compared with that in control mice (Fig. 6a,b). The bronchoalveolar lavage fluid of d-mannose-treated mice had significantly fewer inflammatory leukocytes, particularly eosinophils, compared with that of control mice (Fig. 6c). In accordance with the diminished inflammation in their lungs, d-mannose-treated mice showed significantly lower frequencies of IL-13⁺ T cells and substantially reduced frequencies of IL-4⁺KJ1–26⁺ T cells in the lungs and the peripheral lymphoid tissues compared with control mice (Fig. 6d,e). Intriguingly, there were also lower frequencies of IL-13⁺ and IL-4⁺ T_H2 cells among the nontransgenic CD4⁺ host T cells in the lungs of mice treated with d-mannose compared with untreated control mice, although there were no changes in their peripheral lymphoid tissues (Supplementary Fig. 14b,c). Importantly, d-mannose treatment resulted in a significant increase in the number of KJ1–26⁺CD4⁺Foxp3⁺ T_reg cells in the lungs as well as in the spleen and DLNs compared with those of control mice (Fig. 6f). However, we observed a decrease in the frequency of nontransgenic CD4⁺Foxp3⁺ host T_reg cells in the lungs and peripheral lymphoid tissues in the same d-mannose-treated mice compared with those in controls (Supplementary Fig. 14d).

Figure 6.

d-mannose induces antigen-specific T_reg cells and suppresses ovalbumin-induced airway inflammation in BALB/cJ mice. (a–f) Disease suppression by d-mannose treatment in BALB/cJ mice with ovalbumin-induced airway inflammation. (a) Representative lung histology sections with periodic acid Schiff staining. Insets show magnified views (10× magnification) of the outlined regions. (b) Collated inflammation scores. (c) Absolute numbers of polymorphonuclear neutrophils (PMN), basophils (Bas), eosinophils (Eos), macrophages (Mac) and lymphocytes (Lymph) in bronchoalveolar lavage fluid of BALB/cJ mice after intranasal challenge with ovalbumin. (d–f) Frequencies of IL-4⁺ T cells (d), IL-13⁺ T cells (e) and CD25⁺Foxp3⁺ T_reg cells (f) among KJ1–26⁺CD4⁺ T cells in BALB/cJ mice after induction of airway inflammation. (g–l) Disease amelioration by d-mannose treatment in BALB/cJ mice with ovalbumin-induced airway inflammation. (g) Representative lung histology sections with periodic acid Schiff staining. Insets show magnified views (10× magnification) of the outlined regions. (h) Collated inflammation scores. (i) Absolute numbers of polymorphonuclear neutrophils, basophils, eosinophils, macrophages and lymphocytes (abbreviated as in c) in bronchoalveolar lavage fluid of BALB/cJ mice after intranasal challenge with ovalbumin. (j–l) Frequencies of IL-4⁺ T cells (j), IL-13⁺ T cells (k) and CD25⁺Foxp3⁺ T_reg cells (l) among KJ1–26⁺CD4⁺ T cells in BALB/cJ mice. Summary data (b–f,h–l) are presented as mean ± s.e.m.; data points represent individual mice (n = 4 mice per group). *P < 0.05, **P < 0.01, unpaired two-tailed Student’s t-test. All data are representative of two independent experiments.

To investigate whether d-mannose can be effective in a clinically relevant setting, we first induced airway inflammation in the mice and then treated them with d-mannose in drinking water (Supplementary Fig. 15a). We found that d-mannose treatment significantly ameliorated airway inflammation (Fig. 6g–i). This was accompanied by decreased frequencies of IL-4⁺ and IL-13⁺ T cells and increased frequencies of Foxp3⁺ T_reg cells within the population of transferred KJ1–26⁺CD4⁺ T cells in the lungs, DLNs and spleen compared with those in control mice (Fig. 6j–l). The frequencies of IL-4⁺ and IL-13⁺ T cells among the nontransgenic host CD4⁺ T cells in the lungs were also significantly decreased (P < 0.05), but the frequency of the host T_reg cells was not changed in the same d-mannose-treated mice (Supplementary Fig. 15b–d). These results collectively show that d-mannose supplementation induces antigen-specific T_reg cells and suppresses the immunopathology of airway inflammation in mice, which indicates a broad function of d-mannose-mediated immunoregulation.

DISCUSSION

Here we have outlined a previously unrecognized ability of d-mannose, a hexose sugar, to suppress experimental type 1 diabetes and lung airway inflammation. We have also demonstrated that d-mannose induces the generation of T_reg cells from naive CD4⁺ T cells by enhancing TGF-β signaling via activation of TGF-β from its latent form. Integrin α_vβ₈ and ROS pathways were required for T_reg cell induction by d-mannose in T cells. Molecular studies showed that d-mannose increased integrin α_vβ₈ expression, FAO, and ROS levels in T cells. Importantly, we determined that d-mannose enhances T_reg cell generation in human T cells.

d-mannose induces T_reg cells from naive CD4⁺ T cells. Supporting this conclusion are our in vitro experiments in which naive CD4⁺ T cells were stimulated in culture medium containing different hexose sugars. Only d-mannose was able to induce significant numbers of Foxp3⁺ T_reg cells. This T_reg cell generation was dose-dependent for d-mannose (0–50 mM). We observed that d-mannose induction of T_reg cells required TCR signaling, as d-mannose treatment without TCR stimulation did not induce T_reg cells. This was further shown in an in vivo DO11.10 TCR-transgenic naive T cell adoptive transfer system. d-mannose induced more DO11.10 T_reg cells than did water administration in vivo, but there was no detectable increase in the number of T_reg cells among host BALB/cJ CD4⁺ T cells. This finding is important for potential future applications in human autoimmune diseases, because it would be preferable if d-mannose induced only autoantigen-specific T_reg cells, without affecting other, irrelevant T_reg cell populations. However, it should be noted that the reasons for the lack of an increased frequency of T_reg cells among the polyclonal nontransgenic CD4⁺ T cells remain unknown and require further investigation. Of note, under these culture conditions d-mannose also decreased the levels of multiple effector T cell cytokine mRNAs, including Ifng, Il4, Il6 and Il13 mRNA, but did not induce significant changes in the amount of Il10 and Il17a mRNAs.

TGF-β is essential for the generation of Foxp3⁺ T_reg cells from naive CD4⁺ T cells^16–18, and we show that enhanced TGF-β signaling is an underlying mechanism that promotes d-mannose-induced T_reg cell generation. Indeed, blockade of TGF-β signaling in vitro and in vivo abolished d-mannose-induced T_reg cell generation. Further mechanistic studies showed that the activation of latent TGF-β, rather than TGF-β protein synthesis, was key to the increased TGF-β signaling brought about by d-mannose in T cells. Moreover, our finding that the addition of exogenous LAP–TGF-β1 but not of active TGF-β1 enhanced T_reg cell generation in d-mannose-treated T cells further supports this conclusion, as LAP–TGF-β1 cannot signal without activation. A previous study showed that a high concentration of glucose (for example, >25 mM) can activate TGF-β and may be involved in the growth of epithelial and mesenchymal cells⁴⁶. Our results here show that 5–50 mM glucose did not induce significant T_reg cell generation in T cells in the absence of exogenous LAP–TGF-β1. The detailed mechanism remains unknown, but it is possible that a high concentration of glucose induces T cell activation through glycolysis and increased production of inflammatory cytokines such as IFN-γ, IL-6 and IL-4 (refs. 8,47), which may antagonize Foxp3 induction by TGF-β. Indeed, we noticed that glucose treatment of TCR-stimulated naive CD4⁺ T cells in the presence of anti-IL-4, anti-IL-6 and anti-IFN-γ increased the frequency of T_reg cells compared with control antibody treatment, although the increase was still far less than that observed in d-mannose-induced T_reg cells.

We then determined that d-mannose-mediated TGF-β activation and T_reg cell generation involved integrin α_vβ₈ and ROS^30–32,48. Supporting this conclusion are the findings that d-mannose increased integrin α_vβ₈ expression and ROS production in T cells, and that deletion of Itgb8 and/or blockade of ROS activity abolished the majority of d-mannose-induced T_reg cell generation. The exact mechanisms underlying d-mannose-driven ROS production remain unknown, but they may involve increased FAO that also produces more ROS, as d-mannose increases FAO in T cells. The finding that d-mannose suppressed glycolysis in T cells may provide an explanation for the decreased T cell proliferation and IFN-γ production observed after d-mannose treatment, as these are canonical features of glycolysis induced by TCR stimulation. Notably, although both glucose and mannose can upregulate ROS, only d-mannose, and not the corresponding amount of glucose, induced Foxp3. Nevertheless, further molecular details on this pathway are required. However, one possibility might be that d-mannose but not glucose upregulates integrin α_vβ₈ expression. Although we cannot currently conclude that this is the case, our data strongly suggest that d-mannose upregulation of integrin α_vβ₈ is an important factor in the difference.

Importantly, d-mannose effectively suppressed autoimmune type 1 diabetes and airway inflammation in mice. Oral supplementation with d-mannose in NOD mice before they developed hyperglycemia affected diabetes development in those mice. In addition, oral administration of d-mannose was able to block the progress of diabetes even in new-onset diabetic NOD mice. Similarly, oral administration of d-mannose also prevented and suppressed airway inflammation in the lungs. Of note, long-term supplementation with d-mannose had no obvious side-effects in the NOD mice—a finding that might have implications for the development of a similar clinical therapy for type 1 diabetes in humans⁴⁹. Although a systemic increase in the amount of active TGF-β might have potential effects on the fibrotic response⁵⁰, the fact that d-mannose was well tolerated in mice here suggests that TGF-β activation might have local effects on surrounding T cells, which are less likely to cause a notable fibrotic response. More long-term studies could help rule out this possibility. Moreover, given that d-mannose can activate TGF-β and promote T_reg cell generation, whether it affects other disease conditions such as cancer also remains to be investigated.

d-mannose treatment increased T_reg cell frequencies and decreased numbers of IFN-γ-producing T effector cells in NOD mice compared with untreated controls. Although the beneficial effects of d-mannose in NOD mice were abrogated in the absence of T_reg cells and TGF-β, the central role of this pathway in disease development and the limitations of the NOD experimental model do not allow us to dissect the extent to which T_reg cells and TGF-β mediated d-mannose effects on diabetes in this study, and it is possible that d-mannose may act through additional mechanisms to suppress diabetes in vivo.

Notably, we replicated and confirmed this immunoregulatory effect of d-mannose in an ovalbumin-induced airway inflammation model, which indicates broader efficacy of mannose-mediated therapeutic effects on immunopathology. The notion that d-mannose treatment induces antigen-specific T_reg cells and suppression of immunopathology was supported by this model of ovalbumin-induced airway inflammation.

The physiological level of d-mannose in the blood of humans and mice is approximately 100 μM. However, it has been reported that the amount of circulating d-mannose increases up to ninefold (from 100 to 900 μM) in mice receiving d-mannose in drinking water, with no adverse consequences⁴⁹. Here we used the same amount of d-mannose as used in ref. 49 to supplement the drinking water in our in vivo experiments. In humans, stable serum d-mannose levels of up to 2 mM can be reached and are well tolerated, without signs of liver or renal toxicity⁵¹. We found that even concentrations of d-mannose as low as 1 mM could induce a considerable level of T_reg cells, adding credence to the physiological significance of the findings. Moreover, the possible connection between the consumption of fruits rich in d-mannose (for example, cranberries) and autoimmunity may be an interesting and important issue.

In sum, we have discovered a previously unrecognized immunoregulatory effect of d-mannose on T cells in both preventive and therapeutic models of type 1 diabetes and lung airway inflammation. Our findings warrant further exploration of the basic immunological mechanisms and potential clinical applications of hexose sugars.

ONLINE METHODS

Mice.

C57BL/6, BALB/cJ, Rag1^−/−, CD45.1 (on a C57BL/6 background) and NOD/ShiLtJ mice were obtained from The Jackson Laboratory. DO11.10 TCR-transgenic Rag2^−/− mice (on a BALB/cJ background) were purchased from Taconic. Tgfbr1^f/fCD4-Cre¹⁸, Tgfbr2^f/fER-Cre²², Smad3^−/− (on a C57BL/6 background)²³, and Foxp3–GFP reporter¹⁶ mice (on a C57BL/6 background) were bred in our facility under specific-pathogen-free conditions. Itgb8^f/fCD4-Cre mice³² were obtained from Dr. E.M. Shevach (NIAID, US National Institutes of Health, Bethesda, Maryland, USA), and NOD-Foxp3^DTR mice⁵² were obtained from Drs. C. Benoist and D. Mathis (Harvard Medical School, Boston, Massachusetts, USA). All mice used for experiments were aged 5–12 weeks. All animal studies were performed according to National Institutes of Health (NIH) guidelines for the use and care of live animals and were approved by the Animal Care and Use Committees of the National Institute of Dental and Craniofacial Research (NIDCR).

Antibodies and reagents.

Purified anti-mouse CD3 (no azide and low-endotoxin; 145–2C11), purified anti-mouse CD28 (no azide and low-endotoxin; 37.51), purified anti-human CD3 (OKT3), purified anti-human CD28 (CD28.2), and fluorochrome-conjugated antibodies (anti-mouse CD4 (RM4–5), anti-mouse CD8α (53–6.7), anti-mouse TCR-β (H57–597), anti-mouse CD45 (30-F11), anti-mouse CD25 (PC61.5 and eBio7D4), anti-mouse DO11.10 TCR (KJ1–26), anti-mouse/rat Foxp3 (FJK-16a), anti-mouse IL-4 (11B11), anti-mouse IL-13 (eBio13A), anti-human CD4 (RPA-T4), anti-human CD45RA (HI100), anti-human CD25 (BC96) and anti-human Foxp3 (PCH101)) were from eBioscience. Fluorochrome-conjugated anti-mouse IL-17A (TC11–18H10.1), anti-mouse IFN-γ (XMG1.2), and anti-mouse IL-10 (JES5–16E3) were from BioLegend. Fluorochrome-conjugated anti-human Ki67 (556027) was from BD Pharmingen. Recombinant human IL-2 (202-IL), human latent TGF-β1 (299-LT) and human TGF-β1 (240-B) were from R&D Systems. Anti-TGF-β (1D11.16.8) and isotype control antibody (MOPC-21), and anti-CD25 (PC-61.5.3) and isotype control antibody (HRPN) were from Bio X Cell. SB431542 (TGF-β receptor inhibitor) was from Selleckchem (S1067) and was used at 5 μM. NAC (ROS scavenger) was from Calbiochem (106425) and was used at 10 mM.

Flow cytometry analysis.

Intranuclear staining was carried out with Fixation/Permeabilization buffer solution (eBioscience) according to the manufacturer’s instructions. For intracellular cytokine staining, cells were stimulated with PMA (10 ng/ml), ionomycin (250 ng/ml) and Golgi-Plug (1:1,000 dilution; BD Pharmingen) at 37 °C for 4 h, and then fixed with the Fixation/Permeabilization buffer solution (BD Biosciences) according to the manufacturer’s instructions. Stained cells were analyzed on a FACS-Calibur or LSRFortessa (BD Biosciences), and data were analyzed with FlowJo software.

In vitro differentiation of mouse T_reg cells.

CD4⁺CD25⁻ (naive) T cells or CD4⁺CD25⁻GFP⁻(Foxp3⁻) T cells were purified by magnetic cell sorting (Miltenyi Biotec) or FACS sorting (BD FACSAria II) from mouse spleens and peripheral lymph nodes and cultured at 0.4 × 10⁶ cells per well in 24-well plates with plate-bound anti-CD3 (1.5 μg/mL) and soluble anti-CD28 (1.5 μg/mL), with or without latent TGF-β1 (10 ng/mL) or TGF-β1 (2 ng/mL), at 37 °C. Cells were cultured in ‘complete’ glucose-free DMEM without added exogenous glucose (control medium; this medium contained low levels of endogenous glucose (~0.5–0.6 mM) derived from 10% FBS) or in the same medium supplemented with a high concentration (25 mM) of d-mannose or another sugar monomer such as glucose, fructose, galactose or mannitol (a d-mannose derivative) in the presence of TCR stimulation. Three days later, cells were analyzed by FACS staining.

Real-time RT-PCR.

Total RNA was derived from cultured cells with an RNeasy mini kit (Qiagen), and cDNA was synthesized with a High Capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative real-time PCR was carried out according to the protocol for TaqMan gene expression assay kits (Applied Biosystems). Results were normalized to the expression of Hprt mRNA.

In vivo induction of ovalbumin-specific T_reg cells.

BALB/cJ mice were supplemented with d-mannose in drinking water (1.1 M) or given unsupplemented control water for 10 d before cell transfer (and were maintained with d-mannose-supplemented or control water until death). BALB/cJ mice were then injected with 1 × 10⁶ KJ1–26⁺CD4⁺CD25⁻ naive T cells isolated from DO11.10 TCR-transgenic Rag2^−/− mice, and were supplemented with ovalbumin by gavage (160 mg/d) for another 5 d. Mice were then euthanized, and T_reg cells in the small intestine lamina propria (LPL), spleen, mesenteric lymph nodes and Peyer’s patches were analyzed by FACS staining.

Non-obese diabetic mouse model.

Female NOD/ShiLtJ mice were supplemented with d-mannose in drinking water (1.1 M) from 7.5 weeks of age, and the blood sugar level of these mice was measured every 3 d. NOD/ShiLtJ mice were reared as described⁵³. Once a mouse’s blood sugar concentration reached >200 mg/dL for two consecutive weeks, the mouse was euthanized; all mice were euthanized when more than half of the mice in the control group developed disease. For investigation of therapeutic effects on type 1 diabetes, new-onset diabetic NOD mice^41,53 (blood glucose levels of 200–230 mg/dl) were supplemented with d-mannose in drinking water until the termination of the experiment. The spleens, DLNs and pancreases of mice were harvested for FACS and histology.

Ovalbumin-induced airway inflammation model.

Ovalbumin-induced airway inflammation was generated as described⁴⁵, with some modifications. BALB/cJ mice were supplemented with d-mannose in drinking water (1.1 M) and then subjected to adoptive transfer of 1 × 10⁶ KJ1–26⁺CD4⁺CD25⁻ naive T cells isolated from DO11.10 TCR-transgenic Rag2^−/− mice and injection of ovalbumin to induce ovalbumin-specific T_reg cells. The mice were then challenged with ovalbumin by intranasal injection for four consecutive days (100 μg/d/mouse). For therapeutic experiments on airway inflammation, BALB/cJ mice were induced to develop airway inflammation and then supplemented with d-mannose in drinking water, after which the mice were challenged again with ovalbumin. One day after the last challenge, mice were euthanized and their spleens, DLNs and lungs were harvested for analyses.

In vitro T_reg cell suppression assays.

CD4⁺CD25⁻ (naive) T cells were isolated from spleens and peripheral lymph nodes of congenic CD45.1 mice. CD4⁺CD25⁺GFP⁺(Foxp3⁺) T_reg cells (CD45.2⁺) were isolated via FACS sorting (BD FACSAria II) from the spleens of Foxp3–GFP⁺ mice or from cultured CD4⁺CD25⁻GFP⁻(Foxp3⁻) T cells with plate-bound anti-CD3 (1.5 μg/mL) and soluble anti-CD28 (1.5 μg/mL) in the presence of d-mannose or TGF-β1 for 3 d (all were CD45.2⁺). T_reg cells were cultured at different ratios with CFSE-labeled CD45.1⁺CD4⁺CD25⁻ T cells in the presence of γ-irradiated spleen cells and 0.5 μg/ml anti-CD3. CFSE dilution of CD45.1⁺ effector T cells was analyzed by FACS after 3 d of culture.

T cell–transfer colitis.

CD4⁺CD25⁻CD45RB^hi T cells were isolated from spleens and peripheral lymph nodes of congenic CD45.1 mice via FACS sorting (BD FACSAria II) and injected into Rag1^−/− mice (0.4 × 10⁶ cells per mouse) to induce colitis as described¹⁴. Some Rag1^−/− mice received cotransfers of different populations of T_reg cells (0.1 × 10⁶ cells per mouse) by intravenous injection. The weight of the mice was measured every other day. Tissues were harvested at the end of the experiments for histopathological and immunological analyses.

Cell isolation from small intestine and pancreas.

To obtain LPL from the small intestine, we depleted intraepithelial lymphocytes by mechanical separation from the small intestine as described^54,55. LPL were then separated from gut tissue after vigorous shaking in RPMI medium supplemented with Liberase TL (0.25 mg/ml) and DNase (0.2 mg/ml). After isolation, cell suspensions were passed through 70-μm and 40-μm cell strainers, and cell populations were used for flow cytometry. To obtain pancreatic lymphocytes, we treated pancreatic tissue with collagenase IV (4 mg/ml) and DNase (4 mg/ml) for 30 min, after which we passed cell suspensions consecutively through 70-μm and 40-μm cell strainers and analyzed them by flow cytometry.

In vitro culture of human T cells.

Human peripheral blood mononuclear cells were provided by healthy volunteers and obtained from the NIH Department of Transfusion Medicine (DTM) through their approved protocol number NCT000001846. Blood samples were provided by the DTM on a de-identified basis. Signed informed consent was obtained from all donors. Naive CD4⁺ T cells were purified with Naive T Cell Isolation Kit II (Miltenyi Biotec) or by FACS sorting, and cultured with plate-bound anti-human CD3 (5 μg/mL) and soluble anti-human CD28 (2.5 μg/mL) plus IL-2 (10 ng/ml). Cells were analyzed by FACS staining 4 d later.

Statistics.

Unless otherwise noted, comparison between two different groups was done by unpaired two-tailed Student’s t-test; one-way ANOVA (with Tukey’s multiple-comparison post-tests) was used for comparisons between more than two groups. We used the Kaplan–Meier method and the Mantel–Cox log-rank test for evaluation of type 1 diabetes development in NOD mice. All P values less than 0.05 were considered significant. Statistical analysis was done with GraphPad Prism 6.