Bax Inhibitor-1 preserves pancreatic β-cell proteostasis by limiting proinsulin misfolding and programmed cell death

Abstract

The prevalence of diabetes steadily increases worldwide mirroring the prevalence of obesity. Endoplasmic reticulum (ER) stress is activated in diabetes and contributes to β-cell dysfunction and apoptosis through the activation of a terminal unfolded protein response (UPR). Our results uncover a new role for Bax Inhibitor-One (BI-1), a negative regulator of inositol-requiring enzyme 1 (IRE1α) in preserving β-cell health against terminal UPR-induced apoptosis and pyroptosis in the context of supraphysiological loads of insulin production. BI-1-deficient mice experience a decline in endocrine pancreatic function in physiological and pathophysiological conditions, namely obesity induced by high-fat diet (HFD). We observed early-onset diabetes characterized by hyperglycemia, reduced serum insulin levels, β-cell loss, increased pancreatic lipases and pro-inflammatory cytokines, and the progression of metabolic dysfunction. Pancreatic section analysis revealed that BI-1 deletion overburdens unfolded proinsulin in the ER of β-cells, confirmed by ultrastructural signs of ER stress with overwhelmed IRE1α endoribonuclease (RNase) activity in freshly isolated islets. ER stress led to β-cell dysfunction and islet loss, due to an increase in immature proinsulin granules and defects in insulin crystallization with the presence of Rod-like granules. These results correlated with the induction of autophagy, ER phagy, and crinophagy quality control mechanisms, likely to alleviate the atypical accumulation of misfolded proinsulin in the ER. In fine, BI-1 in β-cells limited IRE1α RNase activity from triggering programmed β-cell death through apoptosis and pyroptosis (caspase-1, IL-1β) via NLRP3 inflammasome activation and metabolic dysfunction. Pharmaceutical IRE1α inhibition with STF-083010 reversed β-cell failure and normalized the metabolic phenotype. These results uncover a new protective role for BI-1 in pancreatic β-cell physiology as a stress integrator to modulate the UPR triggered by accumulating unfolded proinsulin in the ER, as well as autophagy and programmed cell death, with consequences on β-cell function and insulin secretion. In pancreatic β-cells, BI-1^-/- deficiency perturbs proteostasis with proinsulin misfolding, ER stress, terminal UPR with overwhelmed IRE1α/XBP1s/CHOP activation, inflammation, β-cell programmed cell death, and diabetes.

In pancreatic β-cells, BI-1^–/– deficiency perturbs proteostasis with proinsulin misfolding, ER stress, terminal UPR with overwhelmed IRE1α/XBP1s/CHOP activation, inflammation, β-cell programmed cell death, and diabetes.

Fig. 1. BI-1-deficient mice exhibit ER stress in pancreatic islets, resulting in hyperglycemia and reduced serum insulin levels in basal conditions.

BI-1^+/+ and BI-1^–/– mice were fed a normal diet (ND) for 6 months or a 3-month High-Fat Diet (HFD) starting at 3 months old. A Fed-state blood glucose levels at endpoint. n = 25–30 mice per group. B Serum insulin levels from respective genotype and diet. n = 8 mice per group. C Fed-state blood glucose levels above time. n = 15 mice per group. D Glucose-stimulated insulin secretion (GSIS) and Area Under the Curve (AUC) were performed. n = 3–4 mice per group. E Representative immunoblotting analysis of ER stress markers from total pancreatic protein lysates. n = 3 (out of 6) are shown per genotype. F Representative TEM images of pancreatic sections from respective genotype after ND. The nucleus is marked “N” and Mitochondria “M” in red. The yellow arrows point to the ER. Note the swelled and fragmented ER in BI-1^–/– compared to normal BI-1^+/+ and the increased density of ER content in BI-1^–/– mice [Scale bar, 2 µm]. G Representative immunoblotting analysis of KDEL from total pancreatic protein lysates. n = 3 (out of 6) per genotype. H RT-qPCR analysis of pancreatic ER stress markers from BI-1^+/+ and BI-1^–/– isolated islets. n = 8–9 mice per group. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. ****P ≤ 0.0001. $ represents differences in the same genotype with their own control.

Fig. 2. BI-1 deletion impairs pancreatic β-cell mass.

BI-1^+/+ and BI-1^–/– mice were treated as in Fig. 1. A Representative hematoxylin and eosin, insulin, and glucagon staining of pancreatic sections from BI-1^+/+ and BI-1^–/– mice fed ND or HFD as in Fig. 1. [Scale bar, 100 µm]. n = 8–10 mice per group. B Quantification of pancreatic islet size was obtained from insulin staining of pancreatic sections from BI-1^+/+ and BI-1^–/– mice fed ND or HFD. n = 8–10 mice per genotype. C Quantification of the number of islets per mm² were obtained from BI-1^+/+ and BI-1^–/– mice fed ND or HFD. n = 8–10 mice per group. D Pancreas weight expressed as % body weight from BI-1^+/+ and BI-1^–/– mice under ND and HFD conditions was measured. n = 6–8 mice per group. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. ****P ≤ 0.0001. $ represents differences in the same genotype with their own control.

Fig. 3. BI-1 loss increases pancreatic β-cell death through apoptosis and NLRP3 inflammasome activation.

BI-1^+/+ and BI-1^–/– mice were fed a 6-month ND. A Representative immunoblotting analysis of NLRP3 inflammasome markers and substrate namely NLRP3, active-caspase-1 and IL-1β from total pancreatic protein lysates. n = 3 (from 6 mice per genotype) are presented. B Representative myeloperoxidase (MPO) staining for neutrophil infiltration (arrows) of pancreatic sections are shown [Scale bar, 100 µm]. n = 3–5 mice per group. C Serum pancreatic lipase levels are presented. n = 10–12 mice per group. D Relative serum cytokine levels evaluated by flow cytometry analysis. n = 7–8 mice per group. E Immunoblotting analysis of apoptotic markers namely Caspase-3, Bcl2, and Puma (α and β), in total pancreatic protein lysates. Representative n = 3 out of 6 mice per genotype are shown. F Representative pictures of TUNEL staining from BI-1^+/+ and BI-1^–/– pancreatic sections under steady state and HFD. Number of apoptotic TUNEL-positive β-cells. n = 3 mice. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001, ****P ≤ 0.0001. $ represents differences in the same genotype with their own control.

Fig. 4. BI-1 deletion causes defects in insulin maturation.

BI-1^+/+ and BI-1^–/– mice were fed a 6-month ND. A Representative TEM images of pancreatic sections with β-cells and insulin granule morphology are shown [Scale bar, 10 µm, 2 µm, 1 µm]. Suffering pancreatic β-cells are represented in a white rectangle. N: nucleus. Lipofuscin granules and altered mitochondria are in black rectangle. Increased magnification on different insulin granules state shows immature (red asterisks), mature and rod-like granules. n = 4 mice. Respective quantification of B total mature insulin granules, C % of immature insulin granules, D total rod-like granules. Quantification was performed from 10 to 15 images per n; n = 4 mice. E The immunogold staining looks stronger in BI-1 deficient mice insulin granules (b, d) than in WT (a, c), particularly in immature granules (IG) which clearly appear more numerous (surrounded by a line in BI-1 deficient mice). F Representative immunoblotting analysis of proinsulin pancreatic protein lysates from WT and BI-1^−/−. n = 3 (out of 6) mice are shown per genotype. G Immunoblotting quantification of proinsulin. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. ****P ≤ 0.0001. $ represents differences in the same genotype with their own control.

Fig. 5. BI-1 deletion increases cell death and leads to ER stress and activated inflammasome markers in human β-cells.

A Cell death was quantified in the human β−cell line EndoC-βH1 cells transfected with control (siCrtl) or BI-1 siRNA (siBI-1) in response to chemical ER stress (Thapsigargin 1 μM; Tunicamycin 5 μg/ml) compared to normal media. n = 4. $P ≤ 0.05; $$P ≤ 0.01; $$$P ≤ 0.001. $$$$P ≤ 0.0001. $ represents differences with control. *Represents differences with indicated treated conditions, **P ≤ 0.01; ****P ≤ 0.0001. B Cell death was quantified in EndoC-βH1. The concentrations of chemicals were used as following: z-VAD-FMK (50 μmol/L, 30 min pre-incubation), Necrostatin-1 (Nec-1) (10 μmol/L, thapsigargin (1 μmol/L), or an equal volume of DMSO (Sigma-Aldrich). $ represents differences with control. *Represents differences with indicated treated conditions, **P ≤ 0.01; ****P ≤ 0.0001. n = 4–7. C Western blotting analysis of phospho-IRE1α, sXBP1, active-caspase-1, and pro-IL-1β protein levels assessed from EndoC-βH1 cells transfected with control or BI-1 siRNA prior to treatment (n = 4–7).

Fig. 6. IRE1α inhibition corrects pancreatic injury and associated metabolic disorders in BI-1 WT and KO mice.

A Protocol timeline for vehicle (kolliphor 16%) or STF-083010 injection (30 mg/kg) in BI-1^+/+ and BI-1^–/– mice twice a week for 2 weeks before sacrifice during a 3-month HFD starting at 3 months old. B Blood glucose levels in HFD-fed-BI-1^+/+ and BI-1^–/– mice injected with STF-083010 or vehicle. n = 15–25 mice per group. C Relative insulin levels in the sera. n = 8 mice. D Serum pancreatic lipases levels from HFD-BI-1^+/+ and BI-1^–/– mice injected with STF-083010 or not (vehicle) at the end of the HFD-diet. n = 3–6 per group. E Representative images of hematoxylin and eosin staining from HFD-BI-1^+/+ and BI-1^–/– pancreatic sections [Scale bar, 100 µm]. n = 6–8 mice per group. F Islet size was quantified. n = 6–8 mice per group. Quantification of number of islets per mm². n = 6–8 mice. G Representative TEM images of pancreatic sections from HFD-fed BI-1^+/+ and BI-1^–/– injected with STF-083010 or vehicle. n = 3–4 mice per group. [Scale bar, 10 µm]. H Quantification of the total granules per µm² is measured from TEM images from BI-1^+/+ and BI-1^–/– mice pancreatic sections treated with STF-083010 or vehicle. n = 3–4 mice per group. I Insulin granule size was measured from representative TEM images in HFD-fed mice treated with STF-083010 or vehicle (10-15 pictures per n, n = 3–4 mice per group). *P ≤ 005; **P ≤ 0.01; ***P ≤ 0.001. ****P ≤ 0.0001. $ represents differences in the same genotype with their own control.