Metformin prevents diabetes development in type 1 diabetes models via suppression of mTOR and STAT3 signaling in immune cells

Abstract

Type 1 diabetes (T1D) is an organ-specific autoimmune disease caused by T cell-mediated pancreatic β cell destruction. To evaluate the effects of metformin on immune cells in autoimmune diabetes, we administered metformin intraperitoneally to two T1D mouse models and analyzed autoimmune diabetes progression. In a cyclophosphamide (CY)-induced T1D model in male non-obese diabetic (NOD) mice, intraperitoneal administration of metformin significantly prevented autoimmune diabetes. Treatment with metformin showed a decrease in activated T cells, CD44^hiCD62L^lo effector memory cells, macrophages, and dendritic cells (DCs), and an increase in CD44^hiCD62L^hi central memory cells, B cells, and regulatory T cells (Tregs) in splenocytes. Interestingly, metformin treatment showed a decrease in activated T cells, CD4⁺ effector memory T cells and Th1-type antigen-specific cells in PLN cells. IL-17 production was significantly suppressed in metformin-treated mice. TNF-α production from DCs in vitro was dose-dependently suppressed by metformin. Activity of mTOR signaling was significantly reduced in CD4⁺ T cells, CD8⁺ T cells, and B220⁺ B cells. In addition, activities of mTOR and STAT3 signaling in DCs were also reduced significantly. Furthermore, metformin treatment in female NOD mice, a spontaneous T1D model, significantly suppressed autoimmune diabetes onset as well and an increase in Tregs was observed. Our results suggest that metformin may suppress autoimmunity and have therapeutic potential in T1D progression as an immunomodulator.

Keywords: AMPK; Immunomodulator; Metformin; STAT3; Type 1 diabetes; mTOR.

Fig. 1

Protection against development of autoimmune diabetes in CY-given male NOD mice (a CY-induced T1D model) treated with metformin. (A) Flowchart of the mouse experiment. Administration of cyclophosphamide (CY) at 7 weeks of age and/or second CY 14 days later to male NOD mice accelerates diabetes onset (a CY-induced T1D model). 100 mg/kg of metformin or vehicle (saline) was given intraperitoneally for 10 days (5 days before and after the initial CY injection), and diabetes incidence was followed up for 8 weeks after the initial CY injection. (B) The majority of the CY-given NOD mice treated with metformin (n = 19) did not develop diabetes (2/19), whereas approximately 70% of the CY-given male control NOD mice untreated with metformin (n = 16) developed overt diabetes (11/16). The development of diabetes was significantly suppressed in the metformin-treated NOD mice compared with metformin-untreated controls (***p < 0.001). ***p < 0.001 vs. no metformin by a log-rank test.

Fig. 2

Suppression of autoimmune insulitis by metformin treatment. (A) Representative histological appearances are shown for CY-given male NOD mice treated or untreated with metformin. (B) The severity of islet inflammation was quantified by the degree of insulitis at approximately 13 weeks of age. Islets were categorized as: 0 (within normal limits, absent), 1 (lymphocyte infiltration around islets, peri-insulitis), 2 (the percentage of islet lymphocyte infiltration, < 25%), 3 (25–50%), or 4 (50–100%). Insulitis score was calculated for each mouse (n = 3 mice/group, approximately 100 islets were blind scored). The score was significantly suppressed in metformin-treated mice, compared with metformin-untreated mice (***p < 0.001). (C) Immunohistochemical analysis in CY-given diabetic control NOD mice untreated with metformin revealed that CD3⁺ T cells predominantly infiltrated to the islets but not macrophage, and that the infiltrating CD3⁺ T cells were both CD4⁺ T cells and CD8⁺ T cells. ***p < 0.001 vs. no metformin by Mann–Whitney U test.

Fig. 3

Flow cytometric analysis of the phenotypes of splenocytes and PLN cells. (A) A gating strategy and some representative flow cytometric figures. The phenotype of splenocytes (B–G) and PLN cells (at day14) (H–M) isolated from CY-given male NOD mice treated or untreated with metformin and diabetic female NOD mice were analyzed by flow cytometry. (B-G) (B)The absolute number of splenocytes at diabetes onset in metformin-untreated mice was significantly decreased compared with in metformin-treated “tolerized” nondiabetic mice at day 56 after the initial CY injection (**p < 0.01). In metformin-treated mice, the percentages of CD4⁺ T cells (B), B220⁺ cells (B), CD4⁺CD69⁺ T cells (C) and CD44^hiCD62L^hi central memory cells (D) were significantly increased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 56, *p < 0.05  at day 14 and *p < 0.05 at day 14, respectively), and the percentages of CD11b⁺ macrophages (B), CD11c⁺ dendritic cells (B), CD4⁺CD44^hi T cells (C), CD8⁺CD44^hi T cells (C) and CD44^hiCD62L^lo effector memory cells (whole, CD4⁺, CD8⁺) (D) were significantly decreased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 14, ***p < 0.001 at day 56, *p < 0.05 at day 14 and *p < 0.05 at day 56, and whole; **p < 0.01, CD4⁺, CD8⁺;*p < 0.05 at day 14, respectively). The percentage of CD4⁺Foxp3⁺ Tregs (C) was significantly increased in metformin-treated “tolerized” mice at day 56, compared with in metformin-untreated diabetic mice (**p < 0.01). (E) The increased levels of CD127 expression on CD4⁺ T cells and CD8⁺ T cells at day 56 (*p < 0.05 and *p < 0.05, respectively) and the decreased levels of PD-1 expression on CD8⁺ T cells at day 56 (*p < 0.05) were significantly shown in metformin-treated “tolerized” mice, compared with metformin-untreated diabetic mice. No differences were observed in the percentages of PD-1⁺CD44^hi cells (F) and CD11a^hiCD49d^hi cells (G). (H–M) (H) No differences were observed in the absolute cell number and the frequencies of CD4⁺ T cells and CD8⁺ T cells. The percentages of CD4⁺CD44^hi T cells (I) and CD44^hiCD62L^lo CD4⁺ effector memory cells (J) were significantly decreased compared with in metformin-untreated mice, respectively (**p < 0.01). The decreased level of PD-1 expression on CD4⁺ T cells (K) was significantly shown in metformin-treated mice, compared with metformin-untreated mice (*p < 0.05). PD-1⁺CD44^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, **p < 0.01 and *p < 0.05, respectively) (L) and CD11a^hiCD49d^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, *p < 0.05 and **p < 0.01, respectively) (M) were significantly decreased compared with in metformin-untreated mice, respectively. Data are represented as mean ± SEM for at least five mice per group in one of the three independent experiments. *p < 0.05, **p < 0.01 and ***p < 0.001 vs. no metformin by Mann–Whitney U test.

Fig. 3

Flow cytometric analysis of the phenotypes of splenocytes and PLN cells. (A) A gating strategy and some representative flow cytometric figures. The phenotype of splenocytes (B–G) and PLN cells (at day14) (H–M) isolated from CY-given male NOD mice treated or untreated with metformin and diabetic female NOD mice were analyzed by flow cytometry. (B-G) (B)The absolute number of splenocytes at diabetes onset in metformin-untreated mice was significantly decreased compared with in metformin-treated “tolerized” nondiabetic mice at day 56 after the initial CY injection (**p < 0.01). In metformin-treated mice, the percentages of CD4⁺ T cells (B), B220⁺ cells (B), CD4⁺CD69⁺ T cells (C) and CD44^hiCD62L^hi central memory cells (D) were significantly increased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 56, *p < 0.05  at day 14 and *p < 0.05 at day 14, respectively), and the percentages of CD11b⁺ macrophages (B), CD11c⁺ dendritic cells (B), CD4⁺CD44^hi T cells (C), CD8⁺CD44^hi T cells (C) and CD44^hiCD62L^lo effector memory cells (whole, CD4⁺, CD8⁺) (D) were significantly decreased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 14, ***p < 0.001 at day 56, *p < 0.05 at day 14 and *p < 0.05 at day 56, and whole; **p < 0.01, CD4⁺, CD8⁺;*p < 0.05 at day 14, respectively). The percentage of CD4⁺Foxp3⁺ Tregs (C) was significantly increased in metformin-treated “tolerized” mice at day 56, compared with in metformin-untreated diabetic mice (**p < 0.01). (E) The increased levels of CD127 expression on CD4⁺ T cells and CD8⁺ T cells at day 56 (*p < 0.05 and *p < 0.05, respectively) and the decreased levels of PD-1 expression on CD8⁺ T cells at day 56 (*p < 0.05) were significantly shown in metformin-treated “tolerized” mice, compared with metformin-untreated diabetic mice. No differences were observed in the percentages of PD-1⁺CD44^hi cells (F) and CD11a^hiCD49d^hi cells (G). (H–M) (H) No differences were observed in the absolute cell number and the frequencies of CD4⁺ T cells and CD8⁺ T cells. The percentages of CD4⁺CD44^hi T cells (I) and CD44^hiCD62L^lo CD4⁺ effector memory cells (J) were significantly decreased compared with in metformin-untreated mice, respectively (**p < 0.01). The decreased level of PD-1 expression on CD4⁺ T cells (K) was significantly shown in metformin-treated mice, compared with metformin-untreated mice (*p < 0.05). PD-1⁺CD44^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, **p < 0.01 and *p < 0.05, respectively) (L) and CD11a^hiCD49d^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, *p < 0.05 and **p < 0.01, respectively) (M) were significantly decreased compared with in metformin-untreated mice, respectively. Data are represented as mean ± SEM for at least five mice per group in one of the three independent experiments. *p < 0.05, **p < 0.01 and ***p < 0.001 vs. no metformin by Mann–Whitney U test.

Fig. 3

Flow cytometric analysis of the phenotypes of splenocytes and PLN cells. (A) A gating strategy and some representative flow cytometric figures. The phenotype of splenocytes (B–G) and PLN cells (at day14) (H–M) isolated from CY-given male NOD mice treated or untreated with metformin and diabetic female NOD mice were analyzed by flow cytometry. (B-G) (B)The absolute number of splenocytes at diabetes onset in metformin-untreated mice was significantly decreased compared with in metformin-treated “tolerized” nondiabetic mice at day 56 after the initial CY injection (**p < 0.01). In metformin-treated mice, the percentages of CD4⁺ T cells (B), B220⁺ cells (B), CD4⁺CD69⁺ T cells (C) and CD44^hiCD62L^hi central memory cells (D) were significantly increased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 56, *p < 0.05  at day 14 and *p < 0.05 at day 14, respectively), and the percentages of CD11b⁺ macrophages (B), CD11c⁺ dendritic cells (B), CD4⁺CD44^hi T cells (C), CD8⁺CD44^hi T cells (C) and CD44^hiCD62L^lo effector memory cells (whole, CD4⁺, CD8⁺) (D) were significantly decreased compared with in metformin-untreated mice, respectively (***p < 0.001 at day 56, **p < 0.01 at day 14, ***p < 0.001 at day 56, *p < 0.05 at day 14 and *p < 0.05 at day 56, and whole; **p < 0.01, CD4⁺, CD8⁺;*p < 0.05 at day 14, respectively). The percentage of CD4⁺Foxp3⁺ Tregs (C) was significantly increased in metformin-treated “tolerized” mice at day 56, compared with in metformin-untreated diabetic mice (**p < 0.01). (E) The increased levels of CD127 expression on CD4⁺ T cells and CD8⁺ T cells at day 56 (*p < 0.05 and *p < 0.05, respectively) and the decreased levels of PD-1 expression on CD8⁺ T cells at day 56 (*p < 0.05) were significantly shown in metformin-treated “tolerized” mice, compared with metformin-untreated diabetic mice. No differences were observed in the percentages of PD-1⁺CD44^hi cells (F) and CD11a^hiCD49d^hi cells (G). (H–M) (H) No differences were observed in the absolute cell number and the frequencies of CD4⁺ T cells and CD8⁺ T cells. The percentages of CD4⁺CD44^hi T cells (I) and CD44^hiCD62L^lo CD4⁺ effector memory cells (J) were significantly decreased compared with in metformin-untreated mice, respectively (**p < 0.01). The decreased level of PD-1 expression on CD4⁺ T cells (K) was significantly shown in metformin-treated mice, compared with metformin-untreated mice (*p < 0.05). PD-1⁺CD44^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, **p < 0.01 and *p < 0.05, respectively) (L) and CD11a^hiCD49d^hi cells (whole, CD4⁺, CD8⁺) (**p < 0.01, *p < 0.05 and **p < 0.01, respectively) (M) were significantly decreased compared with in metformin-untreated mice, respectively. Data are represented as mean ± SEM for at least five mice per group in one of the three independent experiments. *p < 0.05, **p < 0.01 and ***p < 0.001 vs. no metformin by Mann–Whitney U test.

Fig. 4

Cytokine production from splenocytes in mice treated with metformin in vivo. Cytokine production from splenocytes stimulated with anti-CD3ɛAb in CY-given male NOD mice treated or untreated with metformin were analyzed. In metformin-treated mice, (A) IFN-γ production from splenocytes was similar at day 14 after the initial CY injection and was significantly increased at day 56 compared with in metformin-untreated diabetic mice (n = 4–9/group) (*p < 0.05). (B) IL-17 production from splenocytes was significantly increased at day 14 and was significantly decreased at day 56, compared with in metformin-untreated mice at day 14 and at diabetes onset, respectively (n = 4–5/group) (*p < 0.05 at day 14 and day 56, respectively). (C) IL-10 production from splenocytes was similar to metformin-untreated mice at day 14 and at diabetes onset, respectively (n = 4–9/group). Data are represented as mean ± SEM for at least four mice per group in one of the three independent experiments. *p < 0.05 vs. no metformin by Mann–Whitney U test.

Fig. 5

Suppression of mTOR signaling in T and B cells upon metformin treatment in vivo. Intracellular staining to detect p-S6^235/236 activity in splenocytes from NOD mice treated or untreated with metformin was performed and analyzed for the activity of mTOR signaling by flow cytometry at day 5 after the initial CY injection. In metformin-treated mice, p-S6^235/236 activity of splenocytes was significantly decreased in CD4⁺ T cells, CD8⁺ T cells, and B220⁺ B cells (n = 9–10) (*p < 0.05, *p < 0.05 and *p < 0.05, respectively). Data are represented as mean ± SEM for at least nine mice per group. Three independent experiments were performed. *p < 0.05 vs. no metformin by Mann–Whitney U test.

Fig. 6

Cytokine production from DCs in the presence of metformin in vitro. Bone marrow-derived dendritic cells (BMDCs) from 7–9-week-old NOD mice were stimulated with LPS after 16 h exposure to metformin (0–16 mM). After 24 h stimulation with LPS, the concentration of TNF-α and IL-10 in the culture supernatant were measured by ELISA. (A) TNF-α production from BMDCs was dose-dependently suppressed by metformin (n = 3/group). (B, C) BMDCs exposed to metformin showed a significant decrease in both TNF-α (B) and IL-10 (C) production compared with those unexposed to metformin (**p < 0.01 for TNF-α and IL-10, respectively). Data are represented as mean ± SEM (n = 5 for TNF-α and n = 6 for IL-10 in one of the three independent experiments). **p < 0.01 vs. no metformin by Mann–Whitney U test.

Fig. 7

Suppression of mTOR and STAT3 signaling in DCs in the presence of metformin in vitro. (A, B) BMDCs from 7–9-week-old NOD mice were stimulated with LPS after 16 h exposure to metformin (0 or 16mM). After 24 h stimulation with LPS, intracellular staining to detect p-S6^235/236 (A) and p-STAT3^Y705 (B) activities in DCs were performed and analyzed for the activity of mTOR and STAT3 signaling by flow cytometry. In BMDCs stimulated with LPS after exposure to metformin in vitro, p-S6^235/236 and p-STAT3^Y705 activities of BMDCs were significantly decreased compared with in BMDCs stimulated with LPS alone, respectively (n = 5–8 for p-S6^235/236 activity and n = 4–6 for p-STAT3^Y705 activity. Three independent experiments were performed.) (**p < 0.01 for LPS alone vs. control, ***p < 0.001 for LPS alone vs. LPS + metformin in p-S6^235/236 and *p < 0.05 for LPS alone vs. control, *p < 0.05 for LPS alone vs. LPS + metformin in p-STAT3^Y705, respectively). Data are represented as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by Mann–Whitney U test.

Fig. 8

Protection against development of autoimmune diabetes in female NOD mice (a spontaneous T1D model) treated with metformin. 100 mg/kg of metformin (n = 19) or vehicle (saline) (n = 16) was given intraperitoneally three times a week from 4 to 15 weeks of age in female NOD mice, and diabetes incidence was followed until 40 weeks of age. (A) The development of diabetes was significantly suppressed in the metformin-treated NOD mice compared with metformin-untreated controls (*p < 0.05). *p < 0.05 vs. no metformin by a log-rank test. (B) Representative flow cytometry plots show increased Tregs in metformin-treated mice. (C) In metformin-treated mice, the percentages of Tregs in splenocytes significantly increased compared with in metformin-untreated mice (15–20 weeks of age)(n = 6/group) (*p < 0.05). Data are represented as mean ± SEM. *p < 0.05 vs. no metformin by Mann–Whitney U test. (D) Splenocytes (1.5 × 10⁷) isolated from newly diabetic NOD mice with splenocytes (1.5 × 10⁷) isolated from metformin-treated non-diabetic NOD mice (n = 5) or metformin-untreated non-diabetic NOD mice (n = 5) were injected into 5 to 6-week-old NOD.Scid recipients. Splenocytes from newly diabetic NOD mice with splenocytes from metformin-treated non-diabetic NOD mice significantly suppressed diabetes development in NOD.Scid recipients (*p < 0.05). *p < 0.05 vs. splenocytes from newly diabetic NOD mice with splenocytes from metformin-untreated non-diabetic NOD mice by a log-rank test.