In obesity, increased mitochondrial metabolism with the accumulation of oxidative stress leads to mitochondrial damage and β-cell dysfunction. In particular, β-cells express antioxidant enzymes at relatively low levels and are highly vulnerable to oxidative stress. Early in the development of obesity, β-cells exhibit increased glucose-stimulated insulin secretion in order to compensate for insulin resistance. This increase in β-cell function under the condition of enhanced metabolic stress suggests that β-cells possess a defense mechanism against increased oxidative damage, which may become insufficient or decline at the onset of type 2 diabetes. Here, we show that metabolic stress induces β-cell hypoxia inducible factor 2α (HIF-2α), which stimulates antioxidant gene expression (e.g., Sod2 and Cat) and protects against mitochondrial reactive oxygen species (ROS) and subsequent mitochondrial damage. Knockdown of HIF-2α in Min6 cells exaggerated chronic high glucose–induced mitochondrial damage and β-cell dysfunction by increasing mitochondrial ROS levels. Moreover, inducible β-cell HIF-2α knockout mice developed more severe β-cell dysfunction and glucose intolerance on a high-fat diet, along with increased ROS levels and decreased islet mitochondrial mass. Our results provide a previously unknown mechanism through which β-cells defend against increased metabolic stress to promote β-cell compensation in obesity.
Introduction
β-Cell dysfunction is a critical etiologic component of type 2 diabetes mellitus (T2DM) (1). During the development of obesity, β-cells respond to insulin resistance by adaptively increasing insulin secretion. However, with the onset of T2DM or impaired glucose tolerance, β-cell insulin secretion declines (2–4), with decreased β-cell mass and glucose-stimulated insulin secretion (GSIS) (5–8). Moreover, the incretin effect also declines in patients with T2DM, with increased hepatoporto GLP-1 inactivation, further limiting the postprandial increase in plasma insulin levels (9,10).
The intracellular mechanisms by which β-cell dysfunction develops in T2DM have been studied, and increased oxidative stress has been implicated as a causal mechanism (1,11,12). Mitochondrial metabolism is an essential component of β-cell glucose sensing and energy production, and this is accompanied by production of superoxide free radicals (13,14). In normal islets, glucose induces low levels of reactive oxygen species (ROS), which functions as a signal to enhance GSIS (15,16). However, in obesity, a chronic increase in mitochondrial metabolism enhances electron flux along the mitochondrial transport chain, resulting in premature electron escape to molecular oxygen (ROS production). This causes mitochondrial damage, leading to decreased GSIS and increased β-cell death (glucolipotoxicity) (1,17–21). In addition, accumulation of endoplasmic reticulum stress (22,23) and islet amyloid polypeptide (24) and inflammation (11,25) also stimulate cytosolic pro-oxidant enzyme expression/activity (e.g., NADPH oxidase and inducible nitric oxide synthase [iNOS]) (26,27). Indeed, oxidative stress is increased in T2DM islets and obese rodent islets (28–30), along with abnormal mitochondrial morphology and function (31,32). Moreover, treatment with chemical antioxidants (e.g., N-acetyl cysteine or aminoguanidine) improves β-cell function in obese rodents (32,33) or isolated islets exposed to chronic high glucose levels ex vivo (34).
In mammalian cells, superoxide free radicals are quickly converted to hydrogen peroxide by superoxide dismutases (SODs) and then further metabolized into water by glutathione peroxidase (GPx) and catalase (35). However, in β-cells, the expression of these antioxidant enzymes is relatively low (0–40% of liver) (36,37), and therefore, β-cells are thought to be highly sensitive to oxidative stress. Consistent with this idea, overexpression of various antioxidant enzymes protects from the development of β-cell dysfunction. For example, among genetically obese and insulin resistant db/db mice, those engineered to overexpress GPx1 in β-cells are protected against the development of hyperglycemia (38). Moreover, overexpression of catalase, cytosolic Cu,Zn-SOD, or mitochondrial Mn-SOD protects against streptozotocin-induced β-cell loss in mice (39–42).
Tissue oxygen tension critically affects cell growth, metabolism, and function. Cellular responses to decreased oxygen tension (hypoxia) are largely mediated by hypoxia-inducible factors (HIFs) (43–45). The HIF is a heterodimeric transcription factor consisting of an oxygen-sensitive HIF-α subunit and a constitutively active HIF-1β (also known as aryl hydrocarbon receptor nuclear translocator 1 [Arnt1]) subunit. HIF-αs are ubiquitously expressed, but under normal oxygen concentration, they are rapidly degraded through a mechanism involving prolyl hydroxylase domain enzymes (PHDs) and von Hippel-Lindau (VHL) E3 ubiquitin ligase. In hypoxia, PHDs are inactivated, and stabilized HIF-αs translocate to the nucleus and induce genes necessary for adaptation to decreased oxygen levels. These include genes involved in anaerobic respiration (e.g., Slc2a1 and Pdk1) and/or resistance to oxidative stress (e.g., Gpx1, Cat, Fxn, and SOD) (46). In humans and mice, three isoforms of the HIF-α subunit have been identified (HIF-1α, HIF-2α, and HIF-3α). Among these, HIF-1α and HIF-2α (also known as EPAS1) possess transactivation domains. Of interest, although HIF-1α and HIF-2α share 48% amino acid sequence identity with several common target genes, they mediate distinct physiologic effects. In β-cells, HIF-1α suppresses embryonic β-cell development, and dynamic suppression of HIF-1α expression in embryonic islets is required for islet development (47), whereas HIF-2α is necessary for embryonic pancreatic development in mice (48). HIF-1α also regulates the function of β-cells in adult mice. High levels of glucose or GLP-1 promote HIF-1α expression in β-cell lines (49,50). In T2DM, mRNA expression of Arnt1/Hif1b and Hif1a is decreased in islets compared with normal subjects, and β-cell–specific Arnt1 or Hif1a knockout (KO) mice develop glucose intolerance as a result of decreased insulin secretion (51,52). These results suggest that physiologic HIF-1α expression is necessary to maintain normal β-cell function.
Although HIF-2α plays a crucial role in the regulation of oxygen and metabolic homeostasis, it is not yet known whether HIF-2α plays a role in the regulation of adult β-cell function, particularly during the pathogenesis of β-cell dysfunction in obesity and T2DM. In this article, we investigate the effect of HIF-2α in the regulation of β-cell insulin secretion in normal chow diet (NCD)–fed/lean and high-fat diet (HFD)–fed/obese mice, isolated primary islets, and Min6 cells.
Research Design and Methods
Animals
To generate inducible β-cell–specific HIF-2α KO mice, Hif2afl/fl mice were crossed with mice expressing Cre recombinase–mutated estrogen receptor fusion protein (Cre-ERT) under the control of the mouse insulin promoter (MIP-CreERT+/−:Hif2afl/fl, H2βKOMIP) (53) or the pancreatic and duodenal homeobox 1 (Pdx1) promoter (Pdx1-CreERT+/−:Hif2afl/fl, H2βKOPDX1) (54). Mice were housed in colony cages in 12 h light/12 h dark cycles. For the HFD study, male mice were subjected to 60% HFD (cat. no. D12492; Research Diets, Inc.). After 12 weeks of this HFD, the diet was switched to a tamoxifen-containing HFD for 1 week to induce Cre activation and then switched back to the regular HFD not containing tamoxifen until the termination of the experiments. After 2 weeks of recovery and tamoxifen clearance, mice underwent oral glucose tolerance tests. Briefly, the mice were fasted for 6 h, and basal blood samples were taken, followed by oral glucose gavage (2 g/kg). Blood samples were drawn at 10, 20, 30, 45, 60, 90, and 120 min after gavage, as described previously (55,56). Phenotypes of H2βKOMIP or H2βKOPDX1 mice were compared with those of Cre−/− littermate control mice (Cre−/−:Hif2afl/fl) and/or age-matched MIP-CreERT+:Hif2a+/+ mice (MIP-CreERT). To avoid possible experimental bias resulting from the use of tamoxifen (57), all mice, including Cre−/− control mice, were given tamoxifen before analyses. All animal procedures were performed in accordance with an Institutional Animal Care and Use Committee–approved protocol and the research guidelines for the use of laboratory animals of the University of California San Diego.
Histology
Immunohistochemistry (IHC) and β-cell mass analyses were performed as described previously (58). Briefly, deparaffinized tissue sections were washed with once with Tris-buffered saline with Tween 20 (TBST) and blocked for endogenous peroxidase and background staining by 10-min incubation in Ultravision Hydrogen Peroxidase Block (Thermo Fisher Scientific) solution, followed by another 10 min incubation in Background Punisher (cat. no. BP974M; BioCare). Each step of blocking was followed by an interval in which the slides were washed twice in TBST. Sections were then incubated with control immunoglobulin G (IgG), anti-insulin (cat. no. A0564; Dako), and/or antiglucagon (cat. no. PA1-85465; Pierce Biotechnology, Inc.) antibodies, followed by incubation with secondary antibodies conjugated with horseradish peroxidase (cat. no. 711-036-152; Jackson ImmunoResearch Laboratories) for 30 min. After washing with TBST twice, DAB chromogens (cat. no. 95041-478; VWR) were applied for 5 min and washed twice in double-distilled water, followed by Mayer’s hematoxylin counterstaining for 5 min. For morphometry analyses, images were captured using a NanoZoomer slide scanner system with NanoZoomer Digital Pathology software (Hamamatsu Photonics) and analyzed using ImageJ software (National Institutes of Health, Bethesda, MD). For HIF-2α IHC studies, whole pancreata were excised and snap frozen at an optimum cutting temperature (Fisher Healthcare). Tissue sections measuring 5 μm were fixed with 4% paraformaldehyde for 15 min at room temperature and were subsequently incubated in hydrogen peroxide for 10 min at room temperature for quenching of endogenous peroxidases. Before adding primary antibodies, sections were blocked with 1% bovine serum albumin (BSA) for 30 min at room temperature. Fixed tissue sections were incubated with primary antibodies (guinea pig anti-human insulin [cat. no. IS002; Dako] and rabbit anti–HIF-2α [cat. no. NB100-122; Novus]) (dilution of 1:200), followed by incubation with secondary antibodies. For detecting insulin, anti-rabbit IgG (goat) conjugated with biotin (cat. no. NEF813001EA; Jackson ImmunoResearch Laboratories) and streptavidin conjugated with Alexa Fluor 647 (cat. no. 016-600-084; Jackson ImmunoResearch Laboratories) were used. For detecting HIF-2α, the PerkinElmer TSA Fluorescein Kit (cat. no. NEL701A001KT) was used. Nuclei were stained with DAPI. Images were acquired on a Leica SP8 confocal microscope and processed with ImageJ software.
Plasma Insulin and C-Peptide Concentration and Intraislet Insulin Content
Plasma insulin (cat. no. 80-INSHU-E01.1; Alpco Diagnostics) and C-peptide (cat. no. CPTHUE01.1; Alpco Diagnostics) levels were measured by ELISA in accordance with the manufacturers’ instructions. Pancreatic insulin content was determined as described previously (58).
Quantitative Real-Time PCR
Total RNA was extracted by TRIzol reagent (cat. no. 15596026; Invitrogen) or RNeasy Mini Kit (cat. no. 74104; QIAGEN). Synthesis of cDNA was performed using the High-Capacity cDNA Reverse Transcription Kit (cat. no. 4368813; Applied Biosystems). Quantitative real-time PCR was performed using Power SYBR Green PCR Master Mix (cat. no. 4312704; Thermo Fisher Scientific). Primer sequences are shown in Supplementary Table 1.
Western Blot Analysis
Tissues and cells were lysed in lysis buffer (20 mmol/L Tris-HCl [pH 7.4], 100 mmol/L NaCl, 1.5 mmol/L MgCl2, and 0.1% [vol/vol] Nonidet P-40) containing protease and phosphatase inhibitor cocktail (cat. nos. 04693159001 and 04906845001, respectively; Roche Diagnostics) and then centrifuged at 13,000 rpm for 15 min at 4°C. The supernatants were separated in SDS-PAGE gels (cat. no. 567-1084; Bio-Rad Laboratories) and electrotransferred to polyvinylidene difluoride membranes. The membranes were blocked for 1 h in TBST (10 mmol/L Tris-HCl and 0.1% TritonX-100 [pH 7.4]) containing 5% BSA and then incubated with specific antibodies at 4°C overnight: HIF-1α (cat. no. ab-2185; Abcam), HIF-2α (cat. no. NB100-122; Novus Biologicals), and actin (cat. no. A2228; Sigma). After washing with fresh TBST, the membrane was incubated with secondary antibody conjugated with horseradish peroxidase specific to rabbit or mouse IgG (Jackson ImmunoResearch Laboratories) (dilution of 1:5,000) and visualized using the ECL system (cat. no. WBKLS0050; Merck Millipore) followed by autoradiography or the Bio-Rad ChemiDoc XRS+ imaging system. Intensity of the bands in the autoradiograms was measured using ImageJ software.
HIF-2α Knockdown and GSIS Assay
For HIF-2α knockdown (KD) studies, Min6 cells (passage numbers of ∼12–15) were transfected with mock or Hif2a-specific siRNAs using lipofectamine RNA iMAX reagent (cat. no. 11668030; Invitrogen). Six hours after transfection, media were changed to appropriate media in accordance with each of the experimental conditions. Hif2a-specific siRNA was purchased from Dharmacon (cat. no. L-040635-01-005; ON-TARGETplus siRNA). Static GSIS assays were performed as described previously (58). Briefly, Min6 cells were incubated in 2.8 mmol/L glucose DMEM for 5 h and washed twice with Krebs-Ringer bicarbonate buffer (KRBB) (2.5 mmol/L CaCl2/2H2O, 1.16 mmol/L MgSO4/7H2O, 1.2 mmol/L KH2PO4, 4.7 mmol/L KCl, 114 mmol/L NaCl, 25.5 mmol/L NaHCO3, 20 mmol/L HEPES/Na-HEPES, and 0.2% BSA). Basal insulin release was measured after incubating the cells in KRBB supplemented by 2.8 mmol/L glucose for 30 min. Glucose-stimulated insulin secretion was measured after incubating the cells in KRBB supplemented by 16.7 mmol/L glucose for 30 min.
Citrate Synthase Activity
Citrate synthase (CS) activity was measured using a commercial kit according to the manufacturer’s instructions (cat. no. K318-100; BioVision).
Immunofluorescence Analysis of Mitochondrial Density
A total of 50,000 Min6 cells were seeded in each well of the four-chamber plate (cat. no. 08-774-215; Corning) and grown in DMEM (5 g/L glucose, 4% v/v fetal bovine serum, and 0.1% v/v penicillin/streptomycin) for 24 h. Cells were transfected with 20 nmol/L control or Hif2a-specific siRNA using Lipofectamin RNA iMAX (cat. no. 11668030; Invitrogen). Six hours after transfection, media were changed with fresh media. Media were changed with fresh 2.8 or 16.8 mmol/L glucose media 24 h after transfection. After 24 h, cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100 and 1% BSA phosphate-buffered saline (PBS), followed by blocking with 1% BSA PBS. To stain with anti-Tom20 antibody, cells were incubated with anti-Tom20 antibody (cat. no. sc-17764; Santa Cruz Biotechnology) (dilution of 1:500) overnight at 4°C, followed by incubation with secondary antibodies conjugated with Alexa Fluor 488 (cat. no. A32723; Invitrogen) (dilution of 1:200) for 1 h at room temperature. Nuclei were stained with DAPI (cat. no. ab228549; Abcam). Images were acquired on a Leica SP8 confocal microscope and were processed with ImageJ software.
Intracellular Lactate Concentration
Lactate content in total lysate of Min6 cells was measured by using the Lactate Colorimetric Assay Kit (cat. no. K607-100; BioVision, Inc., Milpitas, CA).
Glucose Uptake
Min6 cells were seeded at 100,000 cells per 24-well plate and grown in DMEM containing 5 g/L glucose, 4% v/v FBS, and 0.1% v/v penicillin/streptomycin. The cells were washed with fresh 2.8 mmol/L glucose KRBB (2.6 mmol/L CaCl2/2H2O, 1.2 mmol/L MgSO4/7H2O, 1.2 mmol/L KH2PO4, 4.9 mmol/L KCl, 98.5 mmol/L NaCl, and 25.9 mmol/L NaHCO3, supplemented with 20 mmol/L HEPES and 0.1% BSA) (cat. no. A8806; Sigma). After incubating with 2.8 mmol/L glucose KRBB for 5 h, the cells were incubated with 2.8 and 16.8 mmol/L glucose KRBB for 1 h with 2-deoxyglucose labeled with tritium (2.8 mmol/L cold glucose with 0.16 μCi/mL [3H]glucose and 16.8 mmol/L cold glucose with 1 μCi/mL of [3H]glucose). After three washings in ice-cold PBS, the cells were lysed in lysis buffer with 0.1% sodium dodecyl sulfate and subjected to scintillation counting to determine their 3H radioactivity, and radioactivity was normalized by total protein content.
Measurement of Intracellular and Mitochondrial ROS
Cytosolic and mitochondrial ROS levels were measured using 5- and 6-chloromethyl-2′, 7′-dichlorodihydro-fluorescein diacetate (cat. no. C6827; Thermo Fisher Scientific) and MitoSOX red (cat. no. M36008; Thermo Fisher Scientific) fluorescent dyes, respectively. Thirty minutes after dye loading, fluorescence intensity in each of the wells was measured using a multiplate reader (FilterMax F5; Molecular Devices).
Isolation of Primary Mouse Islets
Primary mouse islet isolation was performed as described previously with minor modifications (58). Briefly, the bile duct near the ampulla of Vater was ligated, and the common bile duct was cannulated and injected with 3 mL KRBB containing collagenase XI (800 U/mL) (Sigma, Ronkonkoma, NY). The pancreas was dissected from the surrounding tissues, removed, and incubated in a stationary bath for 15 min at 37°C. The digested tissue was washed with KRB without collagenase, and the islets were then purified by a density gradient (Histopaque 1077 and 1119; Sigma) centrifuged at 3,000g for 25 min. Collagen-digested pancreata were filtered through 1,000- and 500-μm sieves, and islets >75 and <250 μm were handpicked under a stereoscope. Islets were cultured in suspension in RPMI 1640 medium, 5 mmol/L glucose, 10% fetal calf serum, 50 units/mL penicillin, 50 μg/mL streptomycin, and 40 μg/mL gentamicin. For Western blots, islets (∼150–200 per well) were kept under normoxic or hypoxic conditions (1% O2) for 6 h. RIPA lysis buffer (25 mmol/L Tris-Cl [pH 7.4], 150 mmol/L NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitor, and phosphatase inhibitor) was added within the hypoxia chamber, and cells were lysed.
Mitochondrial DNA Content
Total DNA was extracted from cells by the proteinase K DNA extraction method. The relative mitochondrial DNA copy number was determined by normalizing mitochondrial DNA copy number to nuclear DNA (18S rRNA) copy number. Primer sequences are shown in Supplementary Table 1.
Intracellular Nitrite Level
Intracellular nitrite content was measured using the Griess Reagent System (cat. no. G2930; Promega) in accordance with the manufacturer’s protocol.
Data and Resource Availability
The data generated during this study are available upon reasonable request. Resources generated during this study are available upon reasonable request.
Results
Obesity Induces β-Cell HIF-2α Expression
In mice, HFD-induced β-cell dysfunction involves progressive deterioration of GSIS, along with increased islet mass and basal insulin secretion, compared with NCD (59,60). To understand whether β-cell HIF-2α expression is changed in obesity, we measured mRNA expression of Hif2a in islets from NCD- and HFD-fed mice. Going from NCD to HFD, Hif1a mRNA expression was moderately increased, whereas the enhanced expression of Hif2a was more robust (Fig. 1A). Because HIF-2α is regulated by posttranslational regulation of protein stability, we assessed β-cell HIF-2α protein expression in NCD- and HFD-fed mice after fasting and refeeding. Hyperoxia and cellular stress during islet isolation can profoundly affect HIF-2α expression; therefore, we chose IHC analysis in snap-frozen pancreatic tissue sections processed and stained side by side and imaged under the same microscopic settings. As seen in Fig. 1B and C, in β-cells from fasted NCD-fed mice, HIF-2α expression was relatively low, and the majority of HIF-2α was located in the cytoplasm. Upon refeeding, the majority of HIF-2α was translocated to the nucleus, without changes in overall HIF-2α expression (Fig. 1B and C). In HFD-fed/obese mice, overall β-cell HIF-2α expression was markedly increased compared with lean mice, and the majority of HIF-2α was observed in the nucleus regardless of feeding state (Fig. 1B and C and Supplementary Fig. 1A). To test whether HFD-induced β-cell Hif2a mRNA and protein expression result from increased metabolic stress, we incubated Min6 cells in high-glucose (16.8 mmol/L) media for 24 and 48 h and measured Hif2a expression. Unlike islets, baseline Hif2a expression was lower compared with Hif1a in Min6 cells (Fig. 1D). Nonetheless, after chronic incubation in high-glucose media, mRNA expression of Hif1a was somewhat increased (∼37%), whereas the increase in Hif2a mRNA expression was considerably greater (∼200%). Moreover, HIF-2α protein expression was markedly increased after 24-h incubation in high-glucose media, which was enhanced by cotreatment with palmitic acid (PA) (Fig. 1E and Supplementary Fig. 1B). HIF-2α expression remained unchanged after acute (30 min) high-glucose challenge (Fig. 1E and Supplementary Fig. 1B). HIF-1α expression was also increased by both acute and chronic high-glucose challenges, and PA treatment increased HIF-1α expression in both low- and high-glucose conditions (Fig. 1E and Supplementary Fig. 1B). These changes in HIF-α expression were associated with decreased intracellular oxygen levels (hypoxia) as measured by pimonidazole adduct formation (Supplementary Fig. 1C and D). Together, these results suggest that chronic metabolic stress in obese β-cells promotes HIF-2α expression.
Generation of Inducible β-Cell–Specific HIF-2α KO Mice
To assess the effect of HIF-2α on β-cell growth and function, we generated β-cell–specific HIF-2α KO mice. Because conventional HIF-2α KO mice develop defective pancreatic development (48), to assess the effect of HIF-2α on adult β-cell function, we generated inducible β-cell–specific HIF-2α KO (H2βKO) mice using the Cre-LoxP system. Specifically, we generated two different inducible H2βKO strains using MIP-CreERT (H2βKOMIP) (53) and Pdx1-CreERT (H2βKOPDX1) (54) mice. To induce Cre activity, mice were fed chow containing tamoxifen for 1 week. After 2 weeks of recovery, mice were subjected to oral glucose tolerance tests as illustrated in Supplementary Fig. 2A. When assessed 3 weeks after final tamoxifen administration, Hif2a mRNA and HIF-2α protein expression were decreased by 67–87% in the primary islets from H2βKOMIP or H2βKOPDX1 mice compared with CreERT−/−:Hif2afl/fl littermate or agematched MIP-CreERT control mice, without changes in Hif1a or HIF-1α expression (Supplementary Fig. 2B–D). Body weight remained comparable between H2βKOMIP, Cre−/−:Hif2afl/fl, and MIP-CreERT mice or between H2βKOPDX1 and Cre−/−:Hif2afl/fl mice before (data not shown) and after tamoxifen treatment (Supplementary Fig. 2E and F).
Inducible Depletion of β-Cell HIF-2α Does Not Impair Glucose Tolerance in Mice Fed NCD
Fasting blood glucose levels and glucose tolerance were comparable in H2βKOMIP and H2βKOPDX1 mice compared with Cre−/−:Hif2afl/fl and/or MIP-CreERT mice on NCD, with comparable levels of plasma insulin and C-peptide (Supplementary Fig. 2G–J). Moreover, the characteristic features of islet morphology, such as predominance of insulin-positive β-cells in the core mantled by glucagon- positive α-cells, were unchanged in NCD-fed H2βKOMIP and H2βKOPDX1 mice compared with NCD Cre−/−:Hif2afl/fl or MIP-CreERT mice (Supplementary Fig. 2K and L). Although β-cell mass was slightly increased in H2βKOMIP mice, with a tendency toward increased islet number (Supplementary Fig. 2M and N), β-cell mass and islet number were unchanged in H2βKOPDX1 mice compared with Cre−/−:Hif2afl/fl mice (Supplementary Fig. 2O and P). Moreover, pancreatic mass was unchanged in the two H2βKO mouse strains (Supplementary Fig. 2Q and R). Interestingly, although glucose tolerance was unchanged, ex vivo GSIS was slightly decreased in the islets from H2βKOMIP and H2βKOPDX mice compared with the islets from Cre−/−:Hif2afl/fl and MIP-CreERT mice (Supplementary Fig. 2S).
β-Cell HIF-2α Depletion Impairs GSIS and Worsens Obesity-Induced Glucose Intolerance
Because β-cell HIF-2α expression was increased in HFD-fed mice, we assessed the effect of β-cell HIF-2α KO in HFD-fed/obese mice. H2βKOMIP, H2βKOPDX1, and wild-type (WT) littermate control mice (Cre−/−:Hif2afl/fl) and age-matched MIP-CreERT mice were fed an HFD and treated with tamoxifen as illustrated in Fig. 2A. After tamoxifen treatment, body weight remain comparable between Cre−/−:Hif2afl/fl, MIP-CreERT, and H2βKOMIP mice and between Cre−/−:Hif2afl/fl and H2βKOPDX1 mice (Fig. 2B and C and Supplementary Fig. 3A). Interestingly, HFD-fed H2βKOPDX1 mice showed impaired glucose tolerance compared with HFD-fed Cre−/−:Hif2afl/fl littermate control mice (Fig. 2D and Supplementary Fig. 3B). Consistent with this, glucose tolerance was worse in HFD-fed H2βKOMIP mice compared with HFD-fed Cre−/−:Hif2afl/fl and MIP-CreERT control mice (Fig. 2E). This change was associated with decreased plasma insulin and C-peptide levels after glucose challenge in fasted HFD-fed H2βKOMIP mice compared with fasted HFD-fed Cre−/−:Hif2afl/fl and HFD-fed MIP-CreERT mice (Fig. 2F and G). Moreover, ex vivo GSIS was markedly decreased in isolated islets from HFD-fed H2βKOMIP and H2βKOPDX1 mice compared with the islets from HFD-fed Cre−/−:Hif2afl/fl littermate and/or MIP-CreERT control mice (Fig. 2H). Intracellular insulin content per islet was not decreased in KO islets (Supplementary Fig. 3C). Furthermore, glucose plus GLP-1–stimulated insulin secretion was lower in the islets from HFD-fed H2βKOMIP mice compared with HFD-fed MIP-CreERT mice (Supplementary Fig. 3D), suggesting that β-cell HIF-2α depletion exaggerates HFD/obesity-induced β-cell dysfunction. These changes were associated with decreased expression of genes associated with β-cell differentiation, maturation, and function, such as Pdx1, Neurod, Hnf4a, Slc2a2, Ins1, Ins2, Uro3, and Cx36, in HFD-fed H2βKOMIP mice compared with HFD-fed Cre−/−:Hif2afl/fl mice (Fig. 2I). Islet morphology was unchanged in the two H2βKO mouse strains compared with Cre−/−:Hif2afl/fl and MIP-CreERT control mice on HFD (Fig. 2J and K). Moreover, pancreatic mass, relative β-cell mass, and number of islets per section area were not decreased in HFD-fed H2βKOMIP or H2βKOPDX1 mice compared with HFD-fed Cre−/−:Hif2afl/fl littermate control mice and/or age-matched MIP-CreERT mice (Fig. 2L–O and Supplementary Fig. 3E and F), suggesting that HIF-2α depletion impairs GSIS without affecting β-cell number or mass. Moreover, the proportion of α-cells and the ratio of α-cells/β-cells within each of the islets were unchanged in both HFD-fed H2βKOMIP and H2βKOPDX1 mice compared with HFD-fed WT control mice (Supplementary Fig. 3G–I). Taken together, these results suggest that inducible β-cell HIF-2α depletion exacerbates obesity-induced impaired GSIS and glucose tolerance without affecting β-cell number or mass or the ratio of α-cells/β-cells.
HIF-2α KD Exacerbates Chronic High Glucose–Induced Impaired GSIS
To understand the mechanism through which HIF-2α regulates β-cell GSIS, we performed HIF-2α KD experiments in Min6 cells. Because β-cell HIF-2α expression was increased by chronic high glucose levels, we determined the effect of HIF-2α on β-cell function after chronic incubation in high- or low-glucose media as illustrated in Fig. 3A. To achieve HIF-2α KD, Min6 cells were transfected with mock or Hif2a-specifc siRNAs, followed by 24- or 48-h incubation in low- or high-glucose media. Hif2a-specific siRNA transfection reduced Hif2a and HIF-2α expression by >80% (Fig. 3B and C) and did not affect Hif1a or HIF-1α expression (Fig. 3B and C and Supplementary Fig. 4A–C). HIF-2α depletion can increase the number of free Arnt proteins that can heterodimerize with HIF-1α and thus indirectly enhance HIF-1α target gene expression. We also tested this hypothesis. When assessed by immunoprecipitation (with anti-Arnt antibodies)/Western blot analyses, Arnt-associated HIF-1α levels were not increased by HIF-2α KD (Supplementary Fig. 4D). Interestingly, chronic incubation in high-glucose media increased Arnt expression (along with HIF-1α and HIF-2α) (Supplementary Fig. 4D), suggesting that increased HIF-α protein levels are backed up by increased Arnt expression under metabolic stress. Taken together, these results suggest that high glucose–induced HIF-2α does not affect HIF-1α expression or limit free Arnt levels to heterodimerize with HIF-1α.
When measured under static incubation conditions, GSIS was decreased after chronic (24 or 48 h) preincubation in high-glucose media in mock siRNA–transfected control cells (Fig. 3D, lane 3 vs. 7), which is a widely recognized phenomenon known as glucotoxicity (61–63). Interestingly, HIF-2α KD exacerbated the decrease in GSIS induced by chronic preincubation in high-glucose media (Fig. 3D, lane 7 vs. 8). In contrast, HIF-2α KD did not affect GSIS in cells maintained in low-glucose media (Fig. 3D, lane 3 vs. 4), consistent with a previous report (49).
HIF-2α KD Enhances Metabolic Stress–Induced Mitochondrial Dysfunction
The process of GSIS is initiated by increased glucose uptake followed by increased mitochondrial metabolism, with an increase in the ATP/ADP ratio. This inhibits ATP-dependent potassium channel activity, leading to membrane depolarization, activation of voltage-dependent calcium channel, and increased intracellular calcium ion (Ca2+) levels. Increased intracellular Ca2+ levels trigger insulin secretion. Therefore, we assessed whether HIF-2α KD lowers glucose-induced intracellular ATP levels after preincubation in low- or high-glucose media. In mock-transfected cells incubated in low-glucose media, intracellular ATP levels were induced by acute (30 min) high-glucose challenge (Fig. 4A, lane 1 vs. 3). Chronic preincubation in high-glucose media lowered the effect on acute high glucose–induced ATP levels (Fig. 4A, lane 3 vs. 7). This effect of preincubation in high-glucose media was associated with increased expression of Pdk1 (Fig. 4B), which blocks mitochondrial pyruvate utilization. Moreover, preincubation in high-glucose media decreased intact mitochondrial mass as measured by CS activity (Fig. 4C). Consistently, immunocytochemical analysis of Tom20-positive mitochondrial mass revealed that HIF-2α KD enhanced chronic high glucose–induced decreased mitochondrial content (Fig. 4D). The expression of glucose transporters, including Slc2a1 (encoding a high-affinity glucose transporter, Glut1) and Slc2a2 (encoding the low-affinity high-capacity glucose transporter, Glut2), was slightly increased by chronic preincubation in high-glucose media (Fig. 4B); however, this did not lead to increased glucose uptake (Fig. 4E, lane 1 vs. 5 and lane 3 vs. 7). Intracellular lactate levels were higher in cells preincubated in high-glucose media for 48 h compared with cells maintained in low-glucose media (Fig. 4F). These results are consistent with the view that chronic high glucose levels impair GSIS by blocking mitochondrial metabolism (17). Interestingly, HIF-2α KD exaggerated the decrease in glucose-stimulated intracellular ATP levels, conferred by chronic preincubation in high-glucose media (Fig. 4A, lane 7 vs. 8). These changes were associated with decreased intact mitochondrial mass (Fig. 4C and D). In contrast, the expression of glycolytic genes, such as Slc2a1, Pdk1, Pgk1, and Ldha, was not increased by HIF-2α KD in Min6 cells incubated in high-glucose media (Fig. 4B). Moreover, glucose uptake (Fig. 4E, lane 7 vs. 8) was not changed by HIF-2α KD, although Slc2a2 expression was slightly decreased (Fig. 4B). Taken together, these results suggest that HIF-2α KD exacerbates chronic high glucose–induced impaired GSIS by enhancing mitochondrial dysfunction.
HIF-2α Induces Antioxidant Gene Expression and Preserves Intact Mitochondria in a Chronic High-Glucose Condition
To test whether the decrease in mitochondrial activity by HIF-2α KD is associated with decreased mitochondrial biogenesis, we measured mitochondrial DNA content and the expression of genes involved in mitochondrial biogenesis. As seen in Fig. 5A, mitochondrial DNA content was not changed by HIF-2α KD in Min6 cells incubated in high-glucose media. Moreover, mRNA expression of Pgc1a, Tfam, Nrf1, and Nrf2, which are involved in mitochondrial biogenesis, was not decreased by HIF-2α KD (Fig. 5B). Therefore, we turned our attention to mitochondrial damage as a possible cause of decreased mitochondrial activity in HIF-2α KD cells. Incubation of mock-transfected control Min6 cells for 24 or 48 h in high-glucose media increased mitochondrial and cytosolic hydrogen peroxide levels (Fig. 5C and D and Supplementary Fig. 4E and F). Of interest, HIF-2α KD exaggerated the chronic high glucose–induced ROS levels (Fig. 5C and D and Supplementary Fig. 4E and F). Treatment with a mitochondria-specific ROS scavenger, MitoTEMPO, abolished the effect of chronic high glucose to decrease mitochondrial mass (CS activity and Tom20 intensity) in both control and HIF-2α KD cells and, consequently, erased the effect of HIF-2α KD (Fig. 5E and F and Supplementary Fig. 4G). In contrast to the effect of HIF-2α KD, HIF-1α KD decreased ROS levels in Min6 cells incubated in high-glucose media (Supplementary Fig. 4H). Overexpression of both HIF-1α and HIF-2α by treatment with a pan PHD inhibitor (64) enhanced high glucose–induced ROS levels, and this effect was completely reversed by HIF-1α KD (Supplementary Fig. 4H), suggesting that, opposite to HIF-2α, HIF-1α increases ROS levels. Interestingly, selective induction of HIF-2α with PHD inhibitor treatment in HIF-1α KD cells reduced high glucose–induced ROS levels compared with untreated mock siRNA control or HIF-2α KD cells (Supplementary Fig. 4H, lane 3 vs. 4). Together, these results suggest that high glucose–induced HIF-2α, but not HIF-1α, preserves mitochondrial activity by reducing oxidative stress and subsequent mitochondrial damage.
To understand the mechanism through which HIF-2α reduces oxidative stress, we measured pro- and antioxidant gene expression in HIF-2α KD and control cells. Of interest, HIF-2α KD decreased chronic high glucose–induced Sod2 expression, which encodes mitochondrial antioxidant Mn-SOD (Fig. 5G). Moreover, HIF-2α KD decreased Cat expression (Fig. 5G), which encodes catalase that detoxifies hydrogen peroxide into water. Furthermore, protein levels of SOD2 and catalase were markedly increased after chronic incubation in high-glucose media in an HIF-2α–dependent manner (Fig. 5H). Of interest, treatment with MitoTEMPO blocked high glucose–induced SOD2 and catalase expression, along with decreased HIF-2α, suggesting that high glucose–induced antioxidant gene expression is mediated by ROS-dependent HIF-2α expression. Consistent with these results, DNA sequence motif analyses revealed that SOD2 and CAT gene promoters contain putative hypoxia-response elements in both humans and mice (Supplementary Fig. 5). In contrast, the expression of pro-oxidant genes such as Nox2, Nox4, and p47phox, which encode components of NADPH oxidase, were not increased by HIF-2α KD (Fig. 5I). Nos2 (encoding iNOS) expression was slightly increased by HIF-2α KD, but this was accompanied by a decrease in in the expression of Arg1 (Fig. 5I). Arginase competes with iNOS for arginine utilization and functionally suppresses iNOS activity. However, nitric oxide levels were not increased by HIF-2α KD (Fig. 5J). Together, these results suggest that HIF-2α suppresses the increase in ROS levels in β-cells after chronic exposure to high glucose levels by stimulating antioxidant gene expression, including Sod2 and Cat.
β-Cell HIF-2α Depletion Enhances Obesity-Induced Oxidative Stress and Mitochondrial Loss in Obese β-Cells
Our results suggest that metabolic stress–induced β-cell HIF-2α preserves mitochondrial activity and GSIS by inducing antioxidant gene expression and decreasing mitochondrial damage. To test this concept in the in vivo setting, we measured ROS levels and intact mitochondrial mass in the islets isolated from HFD-fed H2βKO and WT control mice. As seen in Fig. 6A and B, ROS levels were increased with decreased intact mitochondrial mass (as measured by CS activity) in the islets from HFD-fed H2βKOMIP and H2βKOPDX1 mice compared with HFD-fed Cre−/−:Hif2afl/fl and MIP-CreERT mice. Moreover, mRNA expression of Sod2 and Cat was significantly decreased in the islets from HFD-fed H2βKOMIP mice compared with HFD-fed Cre−/−:Hif2afl/fl mice (Fig. 6C). Consistent with this, ex vivo incubation of WT islets in a low-oxygen condition (1%) markedly increased SOD2 and catalase protein expression, and this effect was substantially attenuated in HIF-2α KO islets (Fig. 6D). In contrast, mRNA expression of glycolytic genes such as Pdk1 and Slc2a1 was not decreased in HIF-2α KO islets (Fig. 6C), consistent with the results in Min6 cells chronically incubated in high-glucose media. Taken together, these results suggest that inducible β-cell HIF-2α depletion exacerbates obesity-induced β-cell dysfunction by enhancing oxidative stress and mitochondrial damage in mice (Supplementary Fig. 6).
Discussion
Here, we show that obesity-induced β-cell HIF-2α stimulates antioxidant gene expression and protects against increased oxidative stress and the development of β-cell dysfunction. Metabolic stress induced by chronic incubation of Min6 cells in high-glucose media or by treatment with PA increased HIF-2α expression. HIF-2α KD in Min6 cells exacerbated the chronic high glucose–induced decrease in intracellular ATP levels and GSIS without affecting glucose uptake, whereas it did not affect GSIS in cells maintained in low-glucose media. The decrease in intracellular ATP levels was due to increased oxidative stress and mitochondrial damage. Treatment with a mitochondria-specific ROS scavenger erased the detrimental effects of HIF-2α KD. At the molecular level, HIF-2α KD reduced the expression of ROS-detoxifying enzymes (e.g., Sod2 and Cat) and enhanced chronic high glucose–induced mitochondrial damage. Consistent with these results, HIF-2α expression was increased in the β-cells of HFD-fed/obese mice compared with NCD-fed/lean mice. Moreover, inducible depletion of β-cell–specific HIF-2α in HFD-fed/obese mice exacerbated β-cell dysfunction and glucose intolerance, with decreased Sod2 and Cat expression, increased ROS levels, and decreased intact mitochondrial mass in the islets. In NCD-fed/lean mice, depletion of HIF-2α did not affect plasma insulin levels or glucose tolerance. Taken together, our results suggest that chronic high glucose–induced β-cell HIF-2α– and HIF-2α–dependent antioxidant enzyme expression constitutes a previously unknown mechanism, allowing β-cells to defend against increased oxidative stress to preserve mitochondrial activity and GSIS in obesity.
Although the mechanism through which obesity induces β-cell dysfunction is relatively well studied, our understanding of how β-cells endure metabolic stress during the development of insulin resistance is limited. Notably, although both β-cell mass and function (GSIS) are decreased in patients with T2DM (5–8), GSIS is greater in the islets from obese individuals without diabetes compared with healthy individuals, and obese individuals without diabetes exhibit relatively normal meal- or glucose-induced plasma insulin levels (65–68). Moreover, individuals with insulin resistance without obesity or diabetes show increased β-cell mass, with a trend toward higher in vivo GSIS compared with individuals with insulin sensitivity without obesity or diabetes (69). Similarly, before the onset of hyperglycemia, Zucker fatty rats (at the age of 10–12 weeks) show hypersulinemic normoglycemia with increased in vivo and ex vivo islet GSIS and increased islet mitochondrial glucose metabolism compared with Zucker lean rats (70). This paradoxic increase in β-cell function, even in the presence of increased metabolic stress, suggests that β-cells possess a defense mechanism against metabolic stress early in the development of insulin resistance, which may decline or become insufficient in the later stages of disease.
One of the best-known mechanisms by which metabolic stress induces β-cell dysfunction is through increasing oxidative stress and mitochondrial dysfunction (20). Although β-cells have highly active mitochondrial metabolism, they express relatively low levels of antioxidant enzymes (encoded by genes such as Sod1, Sod2, Cat, and Gpx1) compared with liver or kidney (36,37). Therefore, it was suggested that β-cells are vulnerable to chronic oxidative stress. Indeed, several lines of evidence indicates that, although expressed at low levels in the basal state, antioxidant enzymes are necessary to maintain normal β-cell function. Thus, global deletion of Sod1, Sod2, or Gpx1 or β-cell–specific deletion of Fxn increases oxidative stress in pancreatic islets and causes β-cell dysfunction in mice (71–73). Interestingly, the expression of antioxidant genes such as Cat, Sod2, and Gpx1 is increased in the islets of obese mice compared with lean mice (74,75). However, the mechanism through which obesity increases antioxidant gene expression was unknown. Our results suggest that HIF-2α mediates the obesity-induced increase in antioxidant gene expression in β-cells. Because suppression of mitochondrial ROS by MitoTEMPO treatment substantially attenuated high glucose–induced SOD2 and catalase expression, along with decreased HIF-2α, it is likely that ROS-induced HIF-2α expression constitutes a negative feedback mechanism to protect β-cells under metabolic stress. Consistent with our results, it was shown that HIF-2α regulates antioxidant gene expression in liver (76,77). That study showed HIF-2α overexpression drives a seven- to eightfold induction of isolated mouse Sod2 (approximately −1,452 to +40 bp) and Cat (approximately −746 to +76 bp) promoter activity. Consistent with this, we found SOD2 and CAT gene promoters contain putative hypoxia-response elements in both humans and mice. Similar to the phenotypes of β-cell–specific HIF-2α KO mice, Sod2 heterozygous mice exhibit normal metabolic profile on NCD, but on HFD, these mice develop more severe glucose intolerance compared with WT mice as a result of decreased β-cell GSIS, without changes in insulin sensitivity (72). On the onset of T2DM in humans, islet SOD2 expression is decreased, whereas CAT and GPx expression is still increased in the islets from individuals with T2DM compared with healthy individuals (78). Future studies are required to elucidate when and how HIF-2α–dependent induction of antioxidant enzyme expression is compromised during the development of β-cell dysfunction in T2DM.
It is interesting to note that HIF-2α expression was only induced after chronic, but not acute, high-glucose challenge, whereas HIF-1α displayed increased expression after both acute and chronic high-glucose challenges. This raises the possibility that HIF-1α and HIF-2α play distinct roles in response to metabolic stress in β-cells. β-Cell HIF-1α stimulates glycolytic gene expression (e.g., Slc2a1 or Pdk1) required for immediate adaptation to oxygen deprivation (i.e., anaerobic respiration) (79). However, prolonged activation of this can compromise β-cell glucose sensing. Therefore, it is reasonable to hypothesize that HIF-1α mediates immediate responses to increased blood glucose levels. In line with this hypothesis, β-cell– specific HIF-1α KO mice develop β-cell dysfunction and glucose intolerance on NCD (52), whereas overexpression of β-cell HIF-1α by β-cell–specific VHL KO also leads to β-cell dysfunction (80,81). In contrast, we found that HIF-2α KO or KD did not or only marginally affected glycolytic gene expression (e.g., Slc2a1, Ldha, Pgk1, or Pdk1) in islets or Min6 cells. Consistently, glucose uptake was unchanged, and intracellular levels of a glycolytic metabolism product, lactate, were increased, but not decreased, by HIF-2α KD. These changes occurred without an increase in HIF-1α expression or activity. Therefore, it is likely that HIF-2α is dispensable for the metabolic switching upon acute high-glucose challenge in normal β-cells. Consistent with this idea, we found HIF-2α KD did not affect GSIS in Min6 cells maintained in low-glucose media. Moreover, β-cell–specific HIF-2α KO mice showed normal glucose tolerance on NCD. However, β-cell HIF-2α KO or KD affected β-cell function in obese mice and in Min6 cells after chronic incubation in high-glucose media. Therefore, we suggest that HIF-2α mediates a delayed response to metabolic stress to protect β-cells from increased oxidative stress. Notably, the glucose-induced acute rise in ROS levels is necessary for full GSIS in normal β-cells by acting as a signaling mediator (15,16). Therefore, it is likely that the HIF-2α–dependent induction of antioxidant enzyme expression is only protective against chronic metabolic stress and does not interfere with acute glucose-induced ROS levels or GSIS in normal β-cells.
Although HIF-1α and HIF-2α share similarities in protein structure, DNA-binding sequence specificity, and regulatory mechanisms (involving VHL and PHDs), several lines of evidence indicate that they mediate distinct metabolic functions. For example, HIF-1α expression is increased in the liver of obese mice (9), whereas postprandial/physiologic increases in liver HIF-2α expression are blunted in obesity (64,82). Obesity-induced hepatocyte HIF-1α enhances first-pass GLP-1 degradation and contributes to the development of glucose intolerance by inducing Dpp4 expression and sinusoidal flow resistance (9). In contrast, postprandial induction of hepatocyte HIF-2α suppresses glucagon signaling and increases insulin sensitivity by inducing cAMP-specific phosphodiesterase and Irs2 expression (64,82,83). In adipocytes, obesity-induced HIF-1α expression stimulates iNOS and expression of various chemokines, triggering adipose tissue inflammation and insulin resistance (55), whereas HIF-2α stimulates UCP1 expression (84) and suppresses weight gain, adipose tissue inflammation, and insulin resistance in HFD-fed/obese mice (55). Our current results suggest that β-cell HIF-2α also plays a protective role against the development of hyperglycemia in obese mice by preserving mitochondrial activity and glucose sensing. Therefore, it seems reasonable to consider a more generalized concept that HIF-2α plays protective roles against the development of hyperglycemia through systemic effects.
Finally, although there are several transgenic mouse strains expressing Cre in β-cells, several problems were raised in all of these mice when we started this study. Thus, most β-cell Cre mice (including the popular rat insulin promoter–Cre and Pdx1-Cre mice) show Cre activity in the brain, as well as in β-cells (85). Although mouse insulin promoter–Cre mice show limited Cre expression/activity in adult β-cells, these mice show abnormally increased β-cell size resulting from aberrant expression of a human growth hormone minigene introduced as a part of the DNA-targeting cassette (57,86,87). Therefore, to avoid possible artifacts generated from off-target effects, in the current study, we generated two different inducible β-cell HIF-2α KO mice using MIP-CreERT and Pdx1-CreERT mice and tried to extract common phenotypes. Interestingly, H2βKOMIP and H2βKOPDX1 mice exhibited similar metabolic phenotypes, such as exaggeration of HFD-induced islet oxidative stress and mitochondrial dysfunction, leading to impaired GSIS and glucose tolerance. These changes were not associated with changes in β-cell mass or the ratio of α-cells/β-cells on HFD in either KO mouse strain. The only discrepancies we observed between βKOMIP and βKOPDX1 mice were the differences in β-cell mass on chow diet. Because the hGH minigene expression in MIP-CreERT mice causes β-cell hypertrophy (57), which is absent in βKOPDX1 mice, we speculate increased β-cell mass in βKOMIP mice could have been due to combinatorial effects of ectopic hGH minigene expression and the depletion of HIF-2α. Consistent with this idea, we did not find any noticeable changes in cell proliferation or growth after HIF-2α KD in Min6 cells, although we did not directly assess cell death. In addition to this, because HIF-2α protects against oxidative stress, and the islet isolation procedure inevitably causes cellular stress, an additional caveat should be taken into consideration; cellular stress caused during islet isolation could have exaggerated the effect of HIF-2α KO.
In summary, we demonstrate that metabolic stress–induced HIF-2α– and HIF-2α–dependent antioxidant gene expression represent a previously unknown defense mechanism against metabolic stress–induced β-cell dysfunction. Additional studies are necessary to understand whether HIF-2α expression or activity is associated with metabolic health in individuals with obesity and when and how this protective mechanism fails during the course of disease progression, leading to β-cell failure.
J.B.S. is currently affiliated with Mokpo National University, Cheonggye-myeon, Republic of Korea.
This article contains supplementary material online at https://doi.org/10.2337/figshare.19607331.
Article Information
Acknowledgments. The authors thank the staff of the San Diego Digestive Disease Research Center and the staff of the La Jolla Institute Microscopy and Histology Core Laboratory for their expert assistance with immunofluorescence staining and slide scanning.
Funding. This study was supported by grant DK124298 from the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health (NIH) and University of California San Diego Health Sciences Research Grant RG084153. M.R. was supported by a postdoctoral fellowship from the American Heart Association (16POST29990015). V.H.-A. was supported by grants FPU16/06190 and EST18/00847 from the Ministrio de Educacion, Cultura Deported of Spain. This publication includes data generated or processed at the University of California San Diego Core Laboratories and Centers funded by the NIH (Microscopy Core Laboratory [NS047101] and Histology Core Laboratories [P30CA23100 and P30DK120515]).
Duality of Interest. This study was funded by grants from the Janssen Pharmaceuticals, Inc., Cymabay Therapeutics, Inc., and pH Pharma, LTD. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. J.-S.M. designed and performed a majority of in vitro experiments and histology and mouse tissue analyses and analyzed the data. M.R. performed primary islet isolation and supported mouse experiments. J.B.S., V.H.-A., and R.I. performed in vitro experiments and tissue analysis. Y.S.L. conceived, designed, and supervised the project, interpreted data, and wrote the manuscript. All authors discussed the results and commented on the manuscript. Y.S.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.