Endoplasmic Reticulum–Associated Degradation (ERAD) Has a Critical Role in Supporting Glucose-Stimulated Insulin Secretion in Pancreatic β-Cells

Pancreatic β-cells are essential to glucose homeostasis and metabolism, and their dysfunction or demise leads to insufficient insulin secretion and to diabetes. Hence, understanding the molecular mechanisms underlying the embryonic development, postnatal function, and pathological death of β-cells has been the focus of numerous past and ongoing studies (1,2).

β-Cells are particularly dependent on proper function of the endoplasmic reticulum (ER) and Golgi apparatus. Preproinsulin precursors are cotranslationally secreted in the ER, where they undergo signal peptide removal and disulfide bond formation (3). This is followed by the transport of proinsulin to the Golgi apparatus where the C-peptide is proteolytically cleaved (4,5). Mature insulin is then packaged into granules and secreted in a Ca²⁺-dependent manner upon glucose stimulation (glucose-stimulated insulin secretion [GSIS]) (6–8).

Even under normal conditions, a subset of proinsulin molecules does not fold properly (9) and must be retrotranslocated from the ER lumen to the cytosol for proteasome-mediated degradation (10,11). This process, known as ER-associated degradation (ERAD), is an evolutionarily conserved protein quality control mechanism required to prevent the accumulation of unfolded or misfolded proteins in the ER (12). Although mammalian cells can overcome temporary stressful conditions by activating the unfolded protein reaction (13), chronic and irresolvable ER burden leads to pathological consequences. Indeed, ER stress has been implicated in the pathogenesis of both type 1 and type 2 diabetes (14,15).

Genetic mutations enhancing proinsulin misfolding have been shown to have a detrimental effect on pancreatic β-cells. For instance, insulin gene mutations that prevent proper folding and processing of proinsulin induce diabetes in both mice (16) and humans (17). To date, however, little effort has been made to elucidate the effect of functional deficiency of the ER-Golgi protein processing machinery on β-cell function and fate. In this context, we recently uncovered a critical role for ERAD in the post-translational processing and maturation of the neuroendocrine hormone AVP (arginine vasopressin) (18), prompting us to ask whether ERAD is similarly essential in pancreatic β-cells. Here we report impaired GSIS in cultured rat insulinoma cell line INS-1 and primary mouse islets lacking ERAD as a consequence of genetic ablation of the core ERAD protein Sel1L. Similarly, mice with β-cell–specific Sel1L ablation are viable but show elevated blood glucose levels and glucose intolerance during postnatal life. Mechanistically, we demonstrate that the Sel1L-mediated ERAD deficiency causes GSIS impairment in β-cells through a combined effect of altered intracellular Ca²⁺ concentration, inability of the ER to process proinsulin, and reduced mitochondrial membrane potential (MMP). Together, our findings provide the first direct evidence that ERAD plays an important role in the regulation of β-cell function.

All chemicals were purchased from Sigma-Aldrich unless otherwise stated. The following reagents were used: FBS (Thermo Fisher Scientific), Lipofectamine 3000, Fluo4/AM, Rhod2 AM, and Mitosox Red (Invitrogen), rat insulin ELISA (Mibio), BCA Kit (Thermo Fisher Scientific), ATP Assay Kit, propidium iodide (PI), JC-1 (Beyotime), and Ru360 (Millipore).

Sel1l^flox/flox mice were previously described (19) and crossed to RIP-Cre transgenic mice (20) on the C57BL/6J background. The resulting Sel1l^flox/+;RIP-Cre mice were then backcrossed with Sel1l^flox/flox mice to generate β-cell–specific knockout (βKO) Sel1l^flox/flox;RIP-Cre and littermate control Sel1l^flox/+;RIP-Cre or Sel1l^flox/+ mice. All mice were fed on a regular diet consisting of 13% fat, 67% carbohydrate, and 20% protein and genotyped by PCR analysis of tail or toe genomic DNA using the following PCR primers: forward, 5′-TGGGACAGAGCGGGCTTGGAAT-3′; reverse, 5′-CACCAGGAGTCAAAGGCATCACTG-3′; R-Geo, 5′-ATTCAGGCTGCGCAACTGTTGGG-3′. All animal procedures have been approved by the Soochow University Animal Care and Use Committee.

Male mice were used in all described physiological experiments unless otherwise indicated. Body weights were recorded on a weekly basis starting 1 week after birth. Blood glucose levels were taken at 8:00 a.m. under nonfasting conditions or after an 8-h fast. For glucose tolerance testing (GTT), mice were fasted for 8 h and then given an intraperitoneal injection of a sterile solution of glucose in PBS at a final concentration of 1 g/kg body weight. Blood glucose levels were measured at 0, 15, 30, 60, and 120 min postinjection using an Ascensia ELITE XL Glucometer. For insulin tolerance testing (ITT), mice were fasted overnight for 8 h and given an intraperitoneal injection of recombinant human insulin in PBS (0.75 international units [IU]/kg body weight) (Eli Lilly), and subsequently, blood glucose levels were measured at 0, 30, 60, 90, and 120 min postadministration.

Pancreatic islet isolation was performed as previously described with minor modifications (21). Briefly, after sacrifice, the pancreas was perfused via the common bile duct with 2 mg/mL collagenase (Sigma-Aldrich) in Hanks’ balanced salt solution (HBSS). The inflated pancreas was excised and incubated at 37°C for 5–8 min. The collagenase-digested pancreas was vigorously shaken, washed three times with 10 mL of ice-cold HBSS, and resuspended in 1 mL of 27% Ficoll 400 (Sigma-Aldrich) in HBSS. Islets were handpicked and used in an ex vivo GSIS assay, as described previously (22). Briefly, isolated islets were cultured overnight in RPMI 1640 medium supplemented with 10% FBS, antibiotics, and 10 mmol/L glucose. Approximately 10 islets per well were selected in a 24-well dish with Krebs-Ringer bicarbonate buffer consisting of the following (in mmol/L): glucose 2.8, NaCl 130, KCl 4.8, MgSO₄ 1.2, KH₂PO₄ 1.2, CaCl₂ 2.5, NaHCO₃ 5.0, and HEPES 10, titrated to pH 7.4 with NaOH and supplemented with 1% BSA (Sigma-Aldrich). Islets were cultured for 1 h at 37°C and 5% CO₂. Islets were then incubated for 30 min in Krebs-Ringer bicarbonate buffer containing either 2.8 or 17.8 mmol/L glucose, after which the supernatants were collected and stored at −80°C until assayed for insulin concentration using a Mouse Insulin Ultrasensitive ELISA (ALPCO). GSIS assay for INS-1 cells was essentially performed as a GSIS assay for islets. The reading value from ELISA for each sample was adjusted using total protein content of islet or INS-1 cells.

Dissected pancreata were fixed in 4% paraformaldehyde at 4°C overnight, embedded in paraffin blocks, and sectioned at 5–8 μm. For insulin immunostaining, sections were rehydrated, blocked for endogenous peroxidase activity (3% H₂O₂), and permeabilized with 0.1% Triton X-100 in PBS. After blocking with 10% normal donkey serum, mouse monoclonal proinsulin (1:100; GS-9A8, Beta Cell Consortium) was applied with rabbit anti-dog calnexin (1:100) (Enzo Life Sciences) overnight at 4°C. Alexa Fluor 488 donkey anti-mouse and 555 donkey anti-rabbit secondary antibodies (1:500) (Life Technologies) were applied and nuclei were counterstained with DAPI.

β-Cell mass was determined as previously described (23). Briefly, total pancreatic and β-cell areas were measured using AxioVision software (version 4.1). β-Cell mass (in mg) was calculated by multiplying the percentage of β-cell area identified by insulin immunostaining (1:1,000) (Linco) with the total pancreatic weight. TUNEL staining was performed with the Apoptog Peroxidase In Situ Apoptosis Detection Kit as described by the manufacturer (Millipore Corporation). The ImmPACT DAB peroxidase kit (Vector Technologies) was used as the substrate, and Gill’s hematoxylin counterstain (diluted 1:10) was applied. Images were acquired using either a Zeiss Axiovert 40 Microscope or a Zeiss LSM 510 Meta Confocal Microscope.

INS-1 cells (ATCC) were cultured in RPMI 1640 medium supplemented with 10% FBS, 100 units/mL penicilin, 100 μg/mL streptomycin, 1 mmol/L sodium pyruvate, and 50 μmol/L β-mercaptoethanol at 37°C in an humidified CO₂ incubator.

The single guide RNA (sgRNA) expression vector pGL3-U6-2sgRNA was constructed to perform targeted deletion of exon 6 of Sel1l in INS-1 cells. sgRNA flanking exon 6 were designed with the Optimized CRISPR Design tool (http://crispr.mit.edu/). A U6 promoter fragment flanked by two sgRNAs was generated by PCR using pUC57 (24) as DNA template. This fragment was digested with ESP3I and subsequently cloned into the vector pGL3-U6-sgRNA (25). The sgRNA expression vector pGL3-U6-2sgRNA and the Cas9-expression plasmid pST1374-NLS-3xflag-linker-Cas9 (25) were cotransfected into INS-1 cells using Lipofectamine 3000 as described above. After transfection, puromycin (5 μg/mL) was added to the culture medium to select for puromycin-resistant cells. A week later, puromycin-resistant INS-1 colonies were picked, expanded, and analyzed by PCR using two pairs of primers:

Forward, 5′-TTGAAAAGTCACCATTACTAACAC-3′.

R1: 5′-CCAATCAGATGGCCTCTTAAAC-3′.

R2: 5′-CAGGGCTTTGGTGTGATTCAT-3′.

A fusion gene containing an ER-targeting DsRed reporter, “self-cleaving” 2A linker and the human proinsulin cDNA with GFP-inserted in the C-peptide coding region was synthesized by Genewiz (Suzhou, People’s Republic of China). The synthetic fusion gene was excised from pUC57 vector by NheI and NotI double digestion and cloned into the same sites in pCDNA 3.1 expression vector to generate pER-DsRed-2A-Proins-GFP. For immunofluorescent staining, HepG2 or INS-1 cells (ATCC) were transfected with plasmids expressing DsRed-2A-ProINS-GFP, or Flag-tagged proINS using Lipofectamine 3000. Living green or red fluorescent images were captured with a FluoviewFV1000 Olympus Confocal Microscope, and images were analyzed by the software ImageJ. For immunofluorescent staining, INS-1 or HepG2 cells transfected with relevant expression plasmids were briefly fixed in 4% paraformaldehyde at 4°C and processed for incubation with primary and fluorescence-tagged secondary antibodies as described in the immunofluorescence section. For monitoring of ER stress, cells were transfected with the ER stress reporter plasmid pCAX-F-XBPDBD-venus, as previously described (26).

INS-1 cells (Sel1l^+/+ or Sel1l^−/−) were seeded into a six-well plate at an initial density of 1 × 10⁵/well and allowed to recover for 12 h. Subsequently, PI was added into the culture medium (10 ng/mL), and the cells were cultured further for 72 h. During this period, cell death was monitored in real time by an Incucyte Zoom system. To assess cell viability, an MTT assay was performed according to the manufacturer’s instruction.

INS-1 cells were seeded into four-well glass-bottomed chambers (Nunc) at a density of 1 × 10⁵/well and cultured in RPMI medium for 24 h. To measure cytosolic and mitochondrial Ca²⁺, freshly prepared Fluo4/AM and Rhod2 AM (Invitrogen) fluorescent dye solutions were added into the Ca²⁺-free PBS, and the cells were stimulated by 16.7 mmol/L glucose for 0.5 h. The cells were subsequently examined and imaged with a FluoviewFV1000 Olympus Confocal Microscope.

INS-1 cells were cultured with or without the ERAD inhibitor EerI in RPMI medium for 48 h and then proceeded for mitochondrial superoxide production (Invitrogen), ATP generation, and MMP (Beyotime) analyses using specific assay kits. For FACS, cells were washed twice with PBS and analyzed by a Beckman Counter Flow Cytometer.

RNA was isolated using the TRIzol-chloroform method. RNA quality and concentration was measured using an Agilent 2100 Bioanalyzer (Agilent Technologies). First-strand cDNA synthesis was performed using RT Buffer, dNTPs (deoxynucleotides), DTT, RNAout, and SuperScript III Reverse Transcriptase (all Invitrogen). Quantitative PCR was carried out using the Power SYBR Green PCR Master Mix or iTaq Universal SYBR Green Supermix (Bio-Rad) on an ABI Prism 7000 Sequence Detection System or Step One Plus Real-Time PCR System (Applied Biosystems). PCR primers were designed using the Primer Select program of Lasergene 7.1 Sequence Analysis Software (DNASTAR).

Cell lysate preparation and Western blotting were performed as previously described (27). Briefly, for denaturing SDS-PAGE, cells were resuspended in radioimmunoprecipitation assay buffer (150 mmol/L NaCl, 1% Triton X-100, 1 mmol/L EDTA, and 50 mmol/L Tris HCl pH 7.5) supplemented with protease inhibitor phenylmethylsulfonyl fluoride and incubated on ice for 25 min. Then, denaturing buffer (250 mmol/L Tris HCl pH 6.8, 1% SDS, 50% glycerol, 1.44 mol/L β-mercaptoethanol, and 0.05% bromophenyl blue) was added to cell lysates and boiled for 5 min prior to separation on an SDS-PAGE gel. For nonreducing SDS-PAGE, lysates were prepared in 5× nondenaturing sample buffer (250 mmol/L Tris HCl pH 6.8, 1% SDS, 50% glycerol, and 0.05% bromophenyl blue) without boiling prior to being separated on an SDS-PAGE gel. The following antibodies were used in this study: Sel1L (rabbit 1:1,000) from Abcam, β-actin (mouse 1:1,000) and Flag (F1804 1:8,000) from Sigma-Aldrich, and GFP (rabbit 1:500) from Santa Cruz Biotechnology. Band density was quantitated using the Image Lab Software on the ChemiDOC XRS + system (Bio-Rad).

Immunoprecipitation (IP) studies were performed essentially as previously described (28). Briefly, cells transfected with expression plasmids were lysed in lysis buffer (150 mmol/L NaCl, 1 mmol/L EDTA, 50 mmol/L Tris-HCl, pH 7.5 or 8.0, protease and protein phosphatase inhibitors, and 10 mmol/L N-ethymaleimide) supplemented with 1% Triton X-100. A total of ∼2 mg protein lysates was incubated with 2 μL of primary or irrelevant antibody overnight at 4°C with gentle rocking. Antibodies used for IP or co-IP were as follows: Anti-Flag and hemagglutinin from Sigma-Aldrich; and Anti-Myc from Abcam. Immunocomplexes were precipitated with protein A-agarose beads (GE) for 8 h at 4°C, washed and eluted by boiling for 5 min in SDS sample buffer and analyzed by Western blotting.

Differences between compared groups were evaluated by performing Student t test or two-way repeated ANOVA using SPSS (version 16.0) software, and P < 0.05 was considered to be significant.

We wished to examine the physiological significance of ERAD functionality in the process of GSIS; therefore, we performed three sets of experiments. First, we treated rat insulinoma (INS-1) cells with the known ERAD inhibitor EerI (29) and assessed insulin secretion in response to glucose stimulation. EerI treatment markedly inhibited basal level insulin release and nearly completely inhibited insulin secretion in response to 16.7 mmol/L glucose (Fig. 1A). EerI also showed strong inhibitory effect on intracellular insulin content in response to 16.7 mmol/L glucose (Fig. 1B).

To further investigate the need for ERAD in insulin secretion, we used CRISPR/Cas9 to generate stable INS-1 cell lines lacking Sel1L (Fig. 1C and D). SEL1L is a core component of the ERAD machinery in mammalian cells (30), and its absence results in a functional ERAD deficiency (31). As expected, the inactivation of SEL1L resulted in elevated ER stress in INS-1 cells, as shown by the marked increase of Xbp-GFP reporter expression (Fig. 1E). Nevertheless, SEL1L-deficient INS-1 cells were viable and showed a comparable cell death rate to that of wild-type (WT) cells (Fig. 1F).

Upon the stimulation of 2.5 mmol/L glucose, SEL1L-deficient cells exhibited similar levels of insulin secretion as WT cells, but exhibited markedly suppressed insulin secretion in response to 16.7 mmol/L glucose stimulation (Fig. 1G). Interestingly, the addition of the Ca²⁺ channel inhibitors 2-APB (2-aminoethoxydiphenyl borate) or Ru360 partially restored insulin secretion in response to 16.7 mmol/L glucose in SEL1L-deficient INS-1 cells (Fig. 1H and I). Together, these in vitro data suggest that the ERAD machinery has an important role in GSIS in pancreatic β-cells.

We next studied mice with βKO of Sel1l to assess the in vivo consequences of ERAD loss of function (Supplementary Fig. 1A). Prior to this, we examined the cell lineage specificity of the Cre transgene by crossing the line to a Cre-dependent tdTomato reporter line. As shown in Supplementary Fig. 1B, red fluorescence was detected only in cell clusters within the boundary of islets, indicating that the Cre transgene is specifically expressed in islet cells. Quantitative RT-PCR–based Sel1l mRNA expression analysis indicated that the Cre-mediated Sel1l deletion in islets is ∼65%. Nevertheless, partial inactivation of Sel1l resulted in the activation of several ER stress–inducible genes: Erdj3, Chop, Xbp1s, and Bip mRNA (Supplementary Fig. 1C).

The βKO mice were born at the expected frequency and were viable (data not shown). Over the first 12 weeks of age, βKO mice exhibited markedly elevated blood glucose levels under both fed and fasted conditions (Fig. 2A and B), despite normal appetite and body weight (Fig. 2C and D). At 6 weeks of age, βKO mice showed impaired glucose tolerance during a GTT (Fig. 2E) but normal glucose disposal during an ITT (Fig. 2F), suggesting a deficit in insulin production and/or secretion. The total pancreatic insulin content of βKO mice was markedly lower in comparison with WT mice (Fig. 2G), and this was not due to a decreased β-cell mass (Fig. 2H). We isolated primary pancreatic islets from WT and βKO mice and performed GSIS experiments using cultured islets. As shown in Fig. 2I, SEL1L-deficient islets showed markedly lower insulin secretion at elevated glucose concentration. Together, these in vivo data support the notion that ERAD function is critical for maintaining pancreatic β-cell function and that impaired ERAD contributes to the development of diabetes.

Next, we investigated the molecular mechanism underlying the impaired GSIS in ERAD-deficient pancreatic β-cells. We have recently shown that ERAD plays a critical role in the maturation of proAVP in hypothalamic neurons (18). Thus, we asked whether ERAD plays a similar role in the processing and maturation of proinsulin in pancreatic β-cells. To this end, we analyzed pancreatic tissue sections from WT and βKO mice by coimmunostaining for proinsulin and the ER resident protein calnexin. Proinsulin was significantly more abundant in βKO islets than WT islets, whereas calnexin was unaffected (Fig. 3A–G), suggesting defective proinsulin maturation in βKO islets. To test this possibility, a human proinsulin with a GFP-tagged C-peptide was coexpressed in Sel1l^−/− INS-1 cells with an ER-targeting red fluorescent protein (ER-DsRed) (32). A substantially higher level of GFP was observed in SEL1L-deficient cells than in WT INS-1 cells (Fig. 3L–O), consistent with insulin maturation depending on a normal ERAD function.

To further elucidate the molecular mechanisms by which ERAD regulates proinsulin processing and maturation, we asked whether the ERAD complex is directly involved in degrading proinsulin. To this end, we first ectopically coexpressed Myc-tagged HRD1, the E3 ubiquitin ligase subunit of the ERAD complex, with Flag-tagged WT or folding-deficient mutant (C72Y) proinsulin in HepG2 cells and performed co-IP studies. As shown in Fig. 4A and B, proinsulin can physically interact with HRD1. Next, we performed cycloheximide (CHX)-based pulse-chase analysis for Flag-tagged WT and folding-deficient mutant proinsulin in SEL1L-normal and SEL1L-deficient HepG2 cells. As shown in Fig. 4C and D, whereas both WT and folding-deficient mutant proinsulin showed efficient degradation in normal SEL1L HepG2 cells, but not in SEl1L-depleted HepG2 cells, the degradation rate of mutant proinsulin was markedly higher than that of WT proinsulin. These results strongly suggest that proinsulin (WT or folding deficient) directly interacts with HRD1 and is a substrate of the SEL1L-HRD1/ERAD complex on the ER membrane.

HRD1 is the E3 ubiquitin ligase subunit of the ERAD complex and has been implicated in polyubiquitination of the proteins destined for the degradation pathway. We therefore next asked whether proinsulin would be polyubiquitinated by the ERAD complex. For this study, we coexpressed hemagglutinin-tagged ubiquitin and Flag-tagged proinsulin in SEL1L-normal and SEL1L-deficient HepG2 cells and performed IP experiments. As shown in Fig. 4E, both WT and folding-deficient proinsulins were efficiently ubiquitinated in SEL1L-normal cells, but not SEL1L-deficient cells. Western blotting using a nonreducing gel showed that the majority of the folding-deficient proinsulin exists as high-molecular-weight protein aggregates (Fig. 4F) (>180 kDa). Our recent work (18) has shown that ERAD prevents protein misfolding in a protein-disulfide isomerase (PDI)-dependent manner. To test whether PDI plays a similar role in the processing and maturation of proinsulin, we coexpressed GFP-tagged proinsulin with Flag-tagged WT or C56A mutant PDI in SEL1L-intact and SEL1L-deficient HepG2 cells. As shown in Fig. 4G, the overexpression of WT PDI significantly lowered GFP expression in Sel1l^−/− INS-1 cells. This effect was not seen with the inactive mutant PDI^C56A, suggesting that post-translational proinsulin processing requires the participation of PDI. Altogether, these data strongly suggest that proinsulin directly interacts with and is a substrate of the ERAD complex on the ER membrane.

Given the finding that SEL1L-deficient INS-1 cells show markedly reduced insulin secretion in response to glucose (Fig. 1G and I), we next assessed the abundance of insulin granules in these cells by immunostaining with an anti–C-peptide antibody. As expected, significantly less C-peptide was detected in SEL1L-deficient cells than in SEL1L-normal cells (Fig. 5D–F vs. Fig. 5A–C and S). This result, which is in line with the coimmunofluorescent data obtained with mouse pancreatic sections, raises the possibility that intracellular trafficking of proinsulin might be impaired in SEL1L-deficient cells. To test this, we ectopically expressed a Flag-tagged proinsulin in HepG2 cells and subsequently analyzed the colocalization of Flag-proinsulin with markers of cis-Golgi (GM130) and early endosome (Rab5). Significantly less colocalization signal of Flag-proinsulin and GM130 or Rab5 was detectable in SEL1L-deficient cells (Fig. 5I vs. Fig. 5L and O vs. Fig. 5R, T, and U). These results suggest that SEL1L/ERAD deficiency inhibits insulin secretion in part by inhibiting intracellular trafficking of proinsulin in the ER-Golgi compartment.

It is well established that intracellular Ca²⁺ concentration is a critical determinant in mediating GSIS (6,7). Given our recent findings in mouse and human hepatic cells that ERAD deficiency causes changes of intracellular Ca²⁺ levels (Q.Lo., Y.H., unpublished observations), we tested whether an altered intracellular Ca²⁺ level might contribute to the defective GSIS in ERAD-defective β-cells. To this end, we used the fluorescent dyes Fluo-4 and Rhod-2 to monitor cytosolic and mitochondrial Ca²⁺ level in INS-1 cells treated with the ERAD inhibitor EerI. EerI treatment caused a dose-dependent decrease of cytosolic Ca²⁺ levels (Fig. 6A) and the exact reciprocal effect on mitochondrial Ca²⁺ levels (Fig. 6B). We tested the effect of the ER Ca²⁺ channel inhibitor 2APB or the mitochondrial calcium uniporter inhibitor Ru360 and found that both 2APB and Ru360 were able to restore cytoplasmic Ca²⁺ levels (Fig. 6C). Cotreatment of INS-1 cells with either 2APB or Ru360 markedly attenuated the effect of EerI on cytosolic and mitochondrial Ca²⁺ (Fig. 6D and E). Interestingly, EerI and high glucose cotreatment did not show a detectable synergistic effect on the cytosolic Ca²⁺ level, and 2APB was able to partially neutralize the effect of EerI (Fig. 6F). Importantly, the inhibitory effect of EerI on GSIS was partially reversed by the cotreatment of INS-1 cells with 2-APB and Ru360 (Fig. 1H).

Using the same techniques, we also measured the cytosolic and mitochondrial Ca²⁺ levels in WT and SEL1L-deficient INS-1 cells. As shown in Fig. 6G and Supplementary Fig. 2, Sel1l-deficient INS-1 cells had a significantly lower level of cytosolic Ca²⁺ than WT INS-1 cells and the exact reciprocal effect on mitochondrial Ca²⁺ level (Fig. 6G). As expected, Ru360 treatment markedly reversed the cytosol/mitochondria Ca²⁺ concentration ratio (Fig. 6G). Finally, we assessed the effect of EerI on ATP production in INS-1 cells. As shown in Fig. 6H and I, EerI induced a dose-dependent decrease of ATP, and this inhibitory effect was partially reversed by cotreatment with 2APB, Ru360, or N-acetyl cysteine (NAC). Together, these data support the notion that the depletion of cytosolic Ca²⁺ pools and the decrease of ATP production as a result of ERAD deficiency contributed at least partly to the GSIS defect in ERAD-deficient β-cells.

A direct correlation between mitochondrial Ca²⁺ level and intracellular reactive oxygen species (ROS) generation has been reported by numerous previous studies (33,34). Given the markedly elevated Ca²⁺ level in the mitochondria of ERAD-deficient cells, we next examined whether these cells have elevated levels of ROS. For this study, we performed fluorescent microscopy on EerI-treated INS-1 cells stained with the ROS-sensitive fluorescent dye Mitosox Red. EerI treatment induced a time- and dose-dependent increase of intracellular ROS (Fig. 7A and B), and this enhanced production of ROS was blocked by cotreatments with 2-APB and Ru360 (Fig. 7C), suggesting that the observed ROS overproduction was due to Ca²⁺ overloading in mitochondria. In addition to its stimulation of ROS overproduction, EerI treatment increased the proportion of cells experiencing low MMP, and this effect could also be reversed by cotreatment with 2-APB and Ru360 cotreatments (Fig. 7D). MTT-based cell viability analysis revealed that EerI treatment induced a time-dependent (6 μmol/L) and dose-dependent change of INS-1 cell viability (Fig. 7E and F). Note that blocking ERAD with low concentrations of EerI does not affect INS-1 cell viability. Similarly, FACS analysis revealed that SEL1L-deficient INS-1 cells had significantly higher intracellular ROS and lower MMP than WT INS-1 cells, and, again, this effect was reversed by treatment with 2-APB and Ru360 (Fig. 7G and H). Together, these data suggest that ERAD deficiency in pancreatic β-cells leads to lower cell viability, likely as a result of ROS-induced lower MMP.

GSIS is a tightly regulated process that requires the coordination of transcriptional, translational, and post-translational processing machineries in the β-cell. In the current study, we asked whether the protein quality control system is also involved. We showed that chemical or genetic disruption of the ERAD system leads to impaired GSIS in cell culture. Moreover, mice with targeted disruption of ERAD function in β-cells are viable but exhibit persistent hyperglycemia and glucose intolerance. Finally, we traced these negative effects of ERAD deficiency on GSIS to the following three molecular mechanisms: reduced intracellular Ca²⁺ level, low MMP, and a defective proinsulin processing and maturation at an early point of the secretory pathway.

The physiological importance of ERAD in GSIS in β-cells was first demonstrated in a set of in vitro experiments using the rat insulinoma cell line INS-1 (Fig. 1A–I). INS-1 cells treated with low doses of EerI, a chemical inhibitor of ERAD, or molecularly depleted of the ERAD key component SEL1L showed markedly reduced insulin secretion in response to 16.7 mmol/L glucose stimulation (Fig. 1A and G, respectively). A similar reduction of GSIS was also seen in primary pancreatic islets derived from mice with targeted disruption of the key ERAD protein SEL1L (Fig. 2G). Together, these in vitro and in vivo data strongly support the view that ERAD has a pivotal role in facilitating stimulus-induced insulin secretion in pancreatic β-cells. This, to the best of our knowledge, is the first direct experimental evidence linking ERAD deficiency to defective insulin biogenesis and release.

The physiological importance of ERAD was supported by in vivo studies. Mice with βKO of the key ERAD component SEL1L had elevated plasma glucose levels during postnatal life (Fig. 2A and B). In addition, these mutant mice suffered from glucose intolerance (Fig. 2E). These findings are consistent with the view that ERAD has a critical role in facilitating stimulus-coupled insulin secretion.

A central focus of the current study is to understand how ERAD deficiency mechanistically impairs insulin secretion in response to glucose stimulation of β-cells. To this end, we first tested whether βKO islets harbor any defect in intracellular trafficking and processing of proinsulin. Coimmunofluorescent staining showed that proinsulin is barely detectable in WT pancreatic islets (Fig. 3A), but is abundant in ERAD-deficient islets (Fig. 3D). Proinsulin colocalization with calnexin (Fig. 3F and G) suggests entrapment in the ER compartment. We then experimentally demonstrated this defect by transiently expressing in SEL1L-deficient INS-1 cells a genetically engineered form of human proinsulin protein with a GFP tag inserted in the C-peptide. Strong GFP was detected in SEL1L-deficient INS-1 cells (Fig. 3M), but not in SEL1L-intact INS-1 cells (Fig. 3I). Together, these findings are supportive of the notion that ERAD is critically required for proinsulin processing in pancreatic β-cells. It is important to point out that overexpression of PDI in SEL1L-deficient INS-1 cells partially attenuates their intracellular GFP signal (Fig. 3P and S). Thus, it is likely that the accumulation of proinsulin in the ER of ERAD-deficient β-cells is due to defective intracellular processing and/or trafficking of proinsulin.

Through IP studies, we have demonstrated that both folding-competent and folding-deficient proinsulins can physically interact with HRD1, the E3 ubiquitin ligase subunit of the ERAD complex (Fig. 4A and B). More importantly, folding-competent and folding-deficient proinsulins showed a time-dependent degradation in SEL1L-intact, but not in SEL1L-deleted, HepG2 cells (Fig. 4C and D). These results strongly suggest that both proinsulin and its folding-deficient mutant are target substrates of the SEL1L/ERAD complex and that the loss of SEL1L/ERAD function likely will result in the accumulation of folded or unfolded proinsulins in the ER compartment. Our data are consistent with the notion that perhaps the major function of ERAD in β-cells is to target unfolded or misfolded proteins for proteosomal degradation, thus maintaining ER homeostasis (10). However, our current data also point to the possibility that the SEL1L/ERAD complex may have a direct role in regulating insulin production by degrading “overproduced” proinsulin as previously suggested (35).

Our coimmunofluorescence analysis further revealed that SEL1L/ERAD-deficient INS-1 cells contain significantly fewer insulin granules (Fig. 5N–R and U). This observation is consistent with the in vivo data that βKO islets had significantly lower pancreatic insulin content (Fig. 2G). Unexpectedly, however, coimmunofluorescence analysis revealed that SEL1L deficiency resulted in a marked reduction of the cis-Golgi compartment in HepG2 cells (Fig. 5J–L vs. Fig. 5G–I). Although the exact molecular mechanisms underlining this marked change of the Golgi apparatus remain to be elucidated, it is very likely that this impairment in Golgi volume may have contributed at least partly to the observed GSIS defect in SEL1L/ERAD-deficient β-cells.

We assessed whether changes in intracellular Ca²⁺ play any role in altering glucose-coupled insulin secretion in ERAD-deficient β-cells. This attempt was prompted by the following: 1) it was found recently in our laboratory that ERAD-defective human and mouse hepatocytes exhibit abnormal intracellular levels; and 2) intracellular Ca²⁺ has been shown by many studies in the past to act as a critical signal-mediating GSIS. Indeed, we found that INS-1 cells treated with the ERAD blocker EerI show a dose-dependent decrease of cytosolic Ca²⁺ level (Fig. 6A) while showing a dose-dependent increase in the mitochondrial Ca²⁺ level (Fig. 6B). These EerI-induced changes of cytosolic and mitochondrial Ca²⁺ levels could be prevented by the Ca²⁺ channel blockers 2-APB and Ru360 (Fig. 6C–F). More importantly, cotreatment of INS-1 cells with EerI and the Ca²⁺ channel blockers 2-APB and Ru360 could reverse the inhibitory effect of EerI on insulin secretion (Fig. 1H). These findings suggest that ERAD deficiency impairs GSIS in part through its ability to alter the intracellular Ca²⁺ level.

It is necessary to point out here that one consequence of the elevated mitochondrial Ca²⁺ levels in ERAD-deficient β-cells is the overproduction of ROS (Fig. 7A–C and G), perhaps because of the overactivation of the mitochondrial oxidative phosphorylation pathway. We have demonstrated that this increase of intracellular ROS is directly responsible for the decline of β-cell MMP (Fig. 7H). β-Cells with a reduced MMP and cell viability likely will show a reduced insulin secretion capacity in response to glucose stimulation. Moreover, these β-cells may have normal viability in normal conditions, but may be more susceptible when facing physiological stress such as insulin resistance. Support for this view comes from our previous observation that high-fat diet feeding, which is widely regarded to promote the development of insulin resistance in rodents (36,37), markedly accelerated the hyperglycemic and glucose-intolerant phenotype of ERAD-depleted mice (38).

In summary, through pharmacological and genetic studies we have demonstrated that ERAD plays a critical role in regulating GSIS in pancreatic β-cells. The regulation of GSIS by ERAD occurs at both the proinsulin processing and insulin granule release steps. Our results point to defective ERAD as a possible contributor to type 1 and type 2 diabetes and as a target for therapeutic intervention.

Acknowledgments. The authors thank Adam Francisco and Dr. Wenyan Ren for technical assistance; Drs. Qi Ling (University of Michigan), Wensheng Zhang, Ying Xu, and Han Wang (Soochow University) for expression plasmids and other reagents; Dr. Masayuki Miura (RIKEN Brain Science Institute, Japan) for expression plasmid; and Drs. Xingen Lei and Bruce Currie (Cornell University) for critical comments on the manuscript.

Funding. This work was supported by National Natural Science Foundation of China grant 31501154 and Natural Science Foundation of Jiangsu Province grant BK20150322 (Z.P.) and by National Natural Science Foundation of China grant 31571489 and the Soochow University Faculty Startup Fund (Q.Lo.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.H. generated Sel1l-deficient INS-1 cell lines, performed GSIS and other in vitro experiments, and made the initial observation of INS-1 cell lines and β-cell–specific Sel1l knockout mice. Y.Ga., M.Z., Q.T., and Y.Go. performed coimmunoprecipitation experiments. K.-Y.D. performed pronuclear injection for the generation of Sel1l^F/F mice. R.S. made the initial observation of β-cell–specific Sel1l knockout mice and performed mouse metabolic experiments. Z.P. and Q.Li. generated and performed mouse metabolic experiments. Y.R.B. provided key reagents and helpful discussions. Q.Lo. designed the experiments and wrote the manuscript. All authors reviewed and approved the manuscript. Q.Lo. 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.