The Role of Inflammation in β-cell Dedifferentiation

Abstract

Chronic inflammation impairs insulin secretion and sensitivity. β-cell dedifferentiation has recently been proposed as a mechanism underlying β-cell failure in T2D. Yet the effect of inflammation on β-cell identity in T2D has not been studied. Therefore, we investigated whether pro-inflammatory cytokines induce β-cell dedifferentiation and whether anti-inflammatory treatments improve insulin secretion via β-cell redifferentiation. We observed that IL-1β, IL-6 and TNFα promote β-cell dedifferentiation in cultured human and mouse islets, with IL-1β being the most potent one of them. In particular, β-cell identity maintaining transcription factor Foxo1 was downregulated upon IL-1β exposure. In vivo, anti-IL-1β, anti-TNFα or NF-kB inhibiting sodium salicylate treatment improved insulin secretion of isolated islets. However, only TNFα antagonism partially prevented the loss of β-cell identity gene expression. Finally, the combination of IL-1β and TNFα antagonism improved insulin secretion of ex vivo isolated islets in a synergistic manner. Thus, while inflammation triggered β-cell dedifferentiation and dysfunction in vitro, this mechanism seems to be only partly responsible for the observed in vivo improvements in insulin secretion.

Figure 1

Pancreatic islets of patients with type 2 diabetes contain more CD45 + immune cells than islets of non-diabetic subjects. (a) Representative islet histology of type 2 diabetic (T2D) and non-diabetic (ND) subjects. Peri-islet and intra-islet CD45 + immune cell count (b and c, respectively) and distribution (d and e, respectively) in human pancreatic tissue sections of T2D and ND control subjects. Islet area (f) and islet-area-corrected average CD45 + immune cell count (g) in T2D and ND control subjects. (b–e) n = 17, 16 for T2D and ND subjects, respectively, (f,g) n = 14, 10 for T2D and ND subjects. n = 30–35 islets/subject were counted. ***P < 0.001. Statistical significance (P) was determined using the two-tailed Mann-Whitney U test. All error bars denote s.e.m.

Figure 2

Cytokine induced β-cell dedifferentiation. (a) mRNA expression levels of Insulin (Ins2), GLUT2 (Slc2a2), Pdx1 and Nkx6-1, Foxo1, glucokinase (Gck) and GLUT1 (Slc2a1) in mouse islets after 24 hours of treatment with 1ng/ml of IL-1β, 100ng/ml of IL-6 and 10 ng/ml of TNFα relative to solvent (dashed line). (b) mRNA expression levels in mouse islets after 24 hours of exposure to various concentrations of IL-1β relative to solvent (dashed line). (c) Protein quantification and a representative Western blot of FoxO1 in mouse islets after 24 hours treatment with IL-1β (1 ng/ml) or saline. (d) mRNA expression levels in mouse islets after 24 hours treatment with 0.25 mM of stearate in the absence (black bars) or presence of the IL-1 receptor antagonist IL-1Ra (1 μg/ml; blue bars) compared to solvent (BSA). (e) mRNA expression levels in human islets after 24 hours of IL-1β (1 ng/ml) treatment relative to solvent (dashed line). (f) Protein quantification and a representative Western blot of FoxO1 in human islets after 24 hours of IL-1β (1 ng/ml). (a) n = 12. (b) n = 9. (c) n = 3–5. (d) n = 14–16. (e) n = 9–10. (f) n = 6–8. All sample sizes represent the sum of 3 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 of treatment group vs. untreated control. Statistical significance (P) was determined using the two-tailed Mann-Whitney U (Fig. a,c–f) and one-way ANOVA (Fig. b). All error bars denote s.e.m.

Figure 3

In vivo anti-inflammatory treatment ameliorates diabetic phenotype and improves β-cell insulin secretion capacity. (a) Glucose levels and (b) corresponding insulin levels following an ipGTT in DIO and DIO/STZ mice. (c,d) GSIS, corresponding fold insulin secretion and insulin content of isolated islets from DIO and DIO/STZ mice. (e) Glucose levels following an ipGTT in DIO/STZ mice ± anti-IL-1β (aIL-1β). (f,g) GSIS, corresponding fold insulin secretion and insulin content of isolated islets from DIO/STZ mice ± aIL-1β. (h–k) GSIS, corresponding fold insulin secretion and insulin content of isolated islets from (h,i) db/db mice ± aIL-1β treatment for 2 weeks and (j,k) DIO mice ± aIL-1β treatment for 8 weeks. (l) Glucose levels following an ipGTT in DIO/STZ mice ± sodium salicylate. (m,n) GSIS, corresponding fold insulin secretion and insulin content of isolated islets from DIO/STZ mice ± sodium salicylate treatment. (o) Glucose levels following an ipGTT in DIO/STZ mice ± anti TNFα (aTNFα) treatment. (p,q) GSIS, corresponding fold insulin secretion and insulin content of isolated islets from DIO/STZ mice ± aTNFα treatment. (r,s) GSIS and corresponding fold insulin secretion of isolated islets from DIO mice ± aTNFα treatment for 8 weeks. (t,u) GSIS, corresponding fold insulin secretion and insulin content from DIO/STZ mice ± aTNFα/aIL-1β treatment. (a,b) n = 19 each, 3 experiments. (c,d) n = 18 each, 3 experiments. (e) n = 19 each, 3 experiments. (f,g) n = 18 each, 3 experiments. (h,i) n = 6 each, 1 experiment. (j,k) n = 5 each, 1 experiment. (l) n = 14 each, 2 experiments. (m,n) n = 12 each, 2 experiments. (o) n = 11 each, 2 experiments. (p,q) n = 12 each, 2 experiments. (r,s) n = 6 each, 1 experiment. (t,u) n = 12 each, 2 experiments. n represents the number of mice in in vivo experiments and biological replicates in experiments with isolated islets. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance (P) was determined using the two-tailed Mann-Whitney U (Fig. a-s) and one-way ANOVA (Fig. t,u). All error bars denote s.e.m. DIO, diet-induced obese; DIO/STZ, diet-induced obese/streptozotocin; ipGTT, intraperitoneal glucose-tolerance test.

Figure 4

Impact of in vivo anti-inflammatory treatment on β-cell dedifferentiation. (a) Relative mRNA expression levels of isolated islets of DIO/STZ compared to DIO mice. (b) Relative mRNA expression levels of isolated islets from DIO/STZ mice ± aIL-1β treatment. (c) Relative mRNA expression levels of isolated islets from DIO/STZ mice ± salicylate treatment. (d) Relative mRNA expression levels of isolated islets from DIO/STZ mice ± aTNFα treatment. (e) Relative mRNA expression levels of isolated islets from DIO/STZ mice ± aTNFα/aIL-1β treatment. (a) n = 20 (DIO/STZ), 29 (DIO), 3 experiments, except for IL-1β and TNFα n = 5. (b) 18–20 each, 3 experiments. (c) n = 15–16 each, 2 experiments. (d) n = 7–10, 2 experiments. (e) n = 7–9, 2 experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance (P) was determined using the two-tailed Mann-Whitney U test. All error bars denote s.e.m. DIO, diet-induced obese; DIO/STZ, diet-induced obese/streptozotocin.