TRAF6 integrates innate immune signals to regulate glucose homeostasis via Parkin-dependent and Parkin-independent mitophagy

Abstract

Innate immune signaling is activated in immunometabolic diseases, including type 2 diabetes, yet its impact on glucose homeostasis is controversial. Here, we report that the E3 ubiquitin ligase TRAF6 integrates innate immune signals following diet-induced obesity to promote glucose homeostasis through the induction of mitophagy. Whereas TRAF6 was dispensable for pancreatic β cell function at baseline, TRAF6 was pivotal for insulin secretion, mitochondrial respiration, and mitophagy following metabolic stress in mouse and human islets. TRAF6 was critical for the recruitment and function of the ubiquitin-mediated (Parkin-dependent) mitophagy machinery. Glucose intolerance induced by TRAF6 deficiency following metabolic stress was reversed by concomitant Parkin deficiency by relieving obstructions in receptor-mediated (Parkin-independent) mitophagy. Our results establish that TRAF6 is vital for traffic through Parkin-mediated mitophagy and implicates TRAF6 in the cross-regulation of ubiquitin- and receptor-mediated mitophagy. Together, we illustrate that β cells engage innate immune signaling to adaptively respond to a diabetogenic environment.

Fig. 1.. Palmitate activates innate immune pathways in β cells.

(A) Luciferase activity, normalized to GFP intensity, of Min6 β cells transfected with empty vector (EV), NFκB, RELB, IRF3, STAT3, STAT1, type I interferon, type II Interferon, IFNB1, MAPK, TFEB, C/EBPβ, ATF4, and p53 reporter plasmids following exposure to vehicle [dimethyl sulfoxide (DMSO) + BSA], LPS (5 nM for 2 hours), or palmitate (0.5 mM for 48 hours). n = 3 per group. *P < 0.05 and ****P < 0.0001 by two-way analysis of variance (ANOVA). RLU, relative light units. (B) Quantitative polymerase chain reaction (qPCR) analysis of Traf6 (left), normalized to Hprt expression, and TRAF6 protein expression by Western blot (WB), quantified by densitometry (right), in CRISPR-generated sgScram or sgTraf6 Min6 β cells. Cyclophilin B (PPIB) served as a loading control. n = 4 per group. ****P < 0.0001 by Student’s unpaired t test. (C) Luciferase activity, normalized to GFP intensity, of sgScram and sgTraf6 Min6 cells transfected with reporter plasmids following exposure to vehicle or 5 nM LPS for 2 hours. n = 3 per group. *P < 0.05 and ****P < 0.0001 by two-way ANOVA. (D) Luciferase activity, normalized to GFP intensity, of sgScram and sgTraf6 Min6 cells transfected with reporter plasmids following exposure to vehicle or 0.5 mM palmitate for 48 hours. n = 3 per group. ****P < 0.0001 by two-way ANOVA.

Fig. 2.. TRAF6 deficiency elicits glucose intolerance following DIO.

(A) Schematic diagram illustrating Cre-mediated recombination at the Traf6 locus. (B) WB for TRAF6 with densitometry normalized to PPIB in islets from mice fed 15 weeks RFD or HFD. PPIB and vinculin (VCL) were loading controls. n = 4 per group. ***P < 0.001 and ****P < 0.0001 by one-way ANOVA. (C) Representative immunofluorescence images from pancreatic sections of mice stained for TRAF6 (red), insulin (green), and DNA [4′,6-diamidino-2-phenylindole (DAPI); blue]. n = 3 per group. Scale bars, 50 μm. (D) Total body mass of male mice following RFD or HFD. n = 9 to 14 per group. (E) Blood glucose concentrations measured during intraperitoneal glucose tolerance test (IPGTT; top) and area under the curve (AUC; bottom) of male mice fed RFD (dashed lines) or HFD (solid lines) for 15 weeks (left) or 30 weeks (right). n = 9 to 12 per group. **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA (IPGTT) or one-way ANOVA (AUC).

Fig. 3.. TRAF6 preserves insulin secretion following metabolic stress in mouse and human β cells.

(A) Serum insulin levels measured during in vivo glucose-stimulated insulin release assays from 15-week or 30-week RFD or HFD-fed mice. n = 7 to 10 per group. *P < 0.05 and ****P < 0.0001 by one-way ANOVA. (B) GSIS from islets from mice fed 15-week RFD or HFD (left) or 30 weeks (right). n = 6 to 9 per group. **P < 0.01 and ****P < 0.0001 by two-way ANOVA. (C) KCl-stimulated insulin secretion in islets from 15-week RFD or HFD-fed mice. n = 4 per group. (D) Insulin content in islets from 15-week RFD or HFD-fed mice. n = 7 to 8 per group. (E) β Cell mass from 15-week or 30-week RFD or HFD-fed mice. n = 5 per group. ***P < 0.001 and ****P < 0.0001 by one-way ANOVA. (F) GSIS in human islets treated with DMSO or 1 μM C25-140 for 24 hours and exposed to glucolipotoxicity (GLT; 25 mM glucose + 0.5 mM palmitate) or control (5 mM glucose + BSA) for 48 hours. n = 4 to 5 per group. *P < 0.05 and ****P < 0.0001 by two-way ANOVA. (G) Insulin content in human islets treated with DMSO or 1 μM C25-140 for 24 hours and exposed to GLT or control for 48 hours. n = 4 to 5 per group.

Fig. 4.. TRAF6 deficiency results in accumulation of dysfunctional mitochondria following DIO.

(A) OCR following exposure to 3 mM glucose, 20 mM glucose, 10 μM oligomycin, and 3 mM KCN in islets from mice fed 15-week RFD or HFD. LG, low glucose; HG, high glucose. n = 4 to 6 per group. ****P < 0.0001 by two-way ANOVA. (B) ATP/ADP ratio in islets from mice fed 15-week RFD or HFD. n = 4 per group. ***P < 0.001 and ****P < 0.0001 by two-way ANOVA. (C) mtDNA/nuclear DNA ratio in islets from mice fed 15-week RFD or HFD by qPCR. n = 5 per group. *P < 0.05 by one-way ANOVA. (D) Citrate synthase activity in islets from mice fed 15-week RFD or HFD. n = 3 per group. *P < 0.05 by one-way ANOVA. (E) WB for OXPHOS subunits, TOM20, and MFN2 with densitometry (relative to control RFD) in islets from mice fed 15-week RFD or HFD. VCL or PPIB served as loading controls. n = 3 to 4 per group. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA. (F) Representative transmission electron micrograph of β cells from mice fed 15-week RFD or HFD with quantification of mitochondrial morphology (~100 mitochondria scored per animal). n = 3 per group. *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA. Scale bars, 500 nm.

Fig. 5.. TRAF6 promotes mitophagy in mouse and human β cells following metabolic stress.

(A) Flow cytometry quantification of mitophagy flux in β cells and non–β cells from mice fed 15-week RFD or HFD. n = 4 to 6 per group. ***P < 0.001 by one-way ANOVA. (B) Flow cytometry quantification of mitophagy flux in β cells from mt-Keima mice fed 15-week RFD or HFD, exposed to DMSO or 1 μM C25-140 for 24 hours. n = 3 per group. **P < 0.01 by one-way ANOVA. (C) Flow cytometry quantification of mitophagy flux in human β cells exposed to DMSO or 1 μM C25-140 for 24 hours and exposed to GLT or control for 48 hours. n = 4 per group. **P < 0.01 by one-way ANOVA. (D) WB with densitometry of phospho-Ser⁶⁵-Ub (pUb^S65) in Min6 cells exposed to BSA or 0.5 mM palmitate for 48 hours and 250 nM valinomycin for the final 3 hours. PPIB was a loading control. n = 3 per group. **P < 0.01 by one-way ANOVA. (E) WB with densitometry of MFN2 in islets from mice fed 15-week RFD or HFD following 250 nM valinomycin exposure for up to 3 hours. Densitometry represents change in MFN2 at 3 hours following valinomycin relative to 0-hour baseline. PPIB was a loading control. n = 3 per group. ****P < 0.0001 by one-way ANOVA.

Fig. 6.. TRAF6 regulates recruitment of mitophagy and ubiquitin receptors following metabolic stress.

(A) Schematic of proteomic studies. (B) Volcano plot of differentially expressed proteins in mitochondrial fractions identified by P < 0.05 and |log₂ fold change (FC)| > 0.25. n = 4 per group. (C) Workflow following proteomic analysis of Min6 cells. (D) Differential expression heatmap (integrated mean log₂FC) from Min6 proteomic studies. n = 4 per group. (E) WB with densitometry of TAX1BP1, p62, and BNIP3 in mitochondrial fractions from Min6 cells. TOM20 was a loading control. n = 3 per group. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA. (F) Representative deconvolution immunofluorescence images from pancreatic sections of 15-week RFD or HFD-fed mice displaying BNIP3 (green), mitochondria (SDHA; red), insulin (cyan), and DNA (DAPI; blue). BNIP3-SDHA colocalization in insulin⁺ areas quantified by Pearson’s correlation coefficient. n = 3 per group. **P < 0.01 and ***P < 0.001 by one-way ANOVA. Scale bars, 20 μm. (G) WB with densitometry of TAX1BP1, p62, and BNIP3 in whole-cell lysates (WCL) from Min6 cells. PPIB was a loading control. n = 3 per group. ****P < 0.0001 by two-way ANOVA. (H) WB with densitometry of TAX1BP1, p62, and BNIP3 in islets from 15-week RFD or HFD-fed mice. PPIB was a loading control. n = 3 per group. ****P < 0.0001 by two-way ANOVA.

Fig. 7.. TRAF6 deficiency affects the mitochondrial localization of Parkin following metabolic stress.

(A) Representative WB of FLAG, HA, and Myc following immunoprecipitation (IP) in Min6 cells transfected with plasmids expressing FLAG-TRAF6, HA-NRDP1, and Myc-Parkin and exposed to BSA or 0.5 mM palmitate for 48 hours. VCL was a loading control. n = 3 per group. IgG, immunoglobulin G. (B) WB of NRDP1 and Parkin with densitometry (relative to sgScram BSA) in mitochondrial fractions from Min6 cells exposed to BSA or 0.5 mM palmitate for 48 hours. TOM20 was a loading control. n = 3 per group. ***P < 0.001 and ****P < 0.0001 by two-way ANOVA. (C) WB with densitometry of NRDP1 and Parkin (relative to sgScram BSA) in whole-cell lysates of Min6 cells exposed to BSA or 0.5 mM palmitate for 48 hours. VCL was a loading control. n = 3 per group. ****P < 0.0001 by two-way ANOVA. (D) WB with densitometry of NRDP1 and Parkin (relative to Control RFD) in islets from 15-week RFD or HFD-fed mice. VCL was a loading control. n = 3 per group. **P < 0.01 by two-way ANOVA.

Fig. 8.. Loss of Parkin rescues glucose homeostasis and β cell function following TRAF6 deficiency.

(A) WB with densitometry for TRAF6 and Parkin in 15-week HFD-fed mouse islets. VCL was a loading control. n = 3 per group. ****P < 0.0001 by one-way ANOVA. (B) Representative immunofluorescence images from pancreatic sections stained for TRAF6 (red), insulin (green), and DAPI (blue). n = 3 per group. Scale bars, 50 μm. (C) Blood glucose measured during IPGTT in 15- or 30-week HFD-fed male mice. n = 8 to 12 per group. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA (IPGTT) or one-way ANOVA (AUC). (D) Serum insulin measured during in vivo glucose-stimulated insulin release assays in 15-week HFD-fed male mice. n = 6 to 10 per group. **P < 0.01 by two-way ANOVA. (E) GSIS from islets of 15-week HFD-fed mice. n = 6 to 7 per group. *P < 0.05 and **P < 0.01 by two-way ANOVA. (F) KCl-stimulated insulin secretion from islets from 15-week HFD-fed mice. n = 4 per group. (G) Insulin content in islets from 15-week HFD-fed mice. n = 6 to 7 per group. (H) β Cell mass from in 15- or 30-week HFD-fed mice. n = 4 to 5 per group. Note that all studies in HFD-fed Ctrl and Traf6^Δβ mice were performed together alongside Parkin^Δβ and Traf6/Parkin^Δβ littermates and may appear twice for purposes of relevant comparisons [in Figs. 2E and 3 (A to E)].

Fig. 9.. Loss of Parkin restores β cell mitochondrial mass and function following TRAF6 deficiency.

(A) OCR following exposure to 3 mM glucose, 20 mM glucose, 10 μM oligomycin, and 3 mM KCN in islets from 15-week HFD-fed mice. n = 3 to 6 per group. **P < 0.01 by two-way ANOVA. (B) ATP/ADP ratio in islets from 15-week HFD-fed mice. n = 4 per group. **P < 0.01 and ***P < 0.001 by two-way ANOVA. (C) mtDNA/nucDNA ratio in islets from 15-week HFD-fed mice. n = 3 to 5 per group. *P < 0.05 and **P < 0.01 by one-way ANOVA. (D) Citrate synthase activity in islets from 15-week HFD-fed mice. n = 3 per group. *P < 0.05 and **P < 0.01 by one-way ANOVA. (E) WB for OXPHOS subunits and TOM20 with densitometry in islets from 15-week HFD-fed mice. VCL was a loading control. n = 3 to 4 per group. **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA. (F) Representative TEM of β cells from 15-week HFD-fed mice with mitochondrial morphology quantification (~100 mitochondria scored per animal). n = 3 per group. *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA. Scale bars, 500 nm. Note that all studies in HFD-fed Ctrl and Traf6^Δβ mice were performed together alongside Parkin^Δβ and Traf6/Parkin^Δβ littermates and may appear twice for purposes of relevant comparisons [in Fig. 4 (A to F)].

Fig. 10.. Loss of Parkin restores β cell mitophagy following TRAF6 deficiency.

(A) Flow cytometry quantification of mitophagy flux by MtPhagy approach in β cells from 15-week HFD-fed mice. n = 3 to 6 per group. *P < 0.05 and **P < 0.01 by one-way ANOVA. (B) Representative deconvolution immunofluorescence images (left) from pancreatic sections from 15-week HFD-fed mice stained for BNIP3 (green), mitochondria (SDHA; red), insulin (cyan), and DNA (DAPI; blue). Insulin-positive areas selected and BNIP3-SDHA colocalization quantified by Pearson’s correlation coefficient (right). n = 3 per group. *P < 0.05 and **P < 0.01 by one-way ANOVA. Scale bars, 20 μm. (C) WB with densitometry of TAX1BP1, p62, and BNIP3 (relative to sgScram siNT BSA) in mitochondrial fractions from Min6 cells 72 hours after transfection with nontargeting (NT) or Parkin-specific siRNA with exposure to BSA or 0.5 mM palmitate for the final 48 hours. TOM20 was a loading control. n = 3 per group. **P < 0.01, ***P < 0.001, and ****P < 0.0001 by two-way ANOVA. Note that all studies in HFD-fed Ctrl and Traf6^Δβ mice were performed together alongside Parkin^Δβ and Traf6/Parkin^Δβ littermates and may appear twice for purposes of relevant comparisons (in Figs. 5A and 6F).