Impaired Store-Operated Calcium Entry and STIM1 Loss Lead to Reduced Insulin Secretion and Increased Endoplasmic Reticulum Stress in the Diabetic β-Cell Free

Reductions in β-cell endoplasmic reticulum (ER) calcium (Ca²⁺) levels contribute to the pathophysiology of both type 1 diabetes and type 2 diabetes (T2D) and lead to decreased insulin secretion, activation of intracellular stress pathways, and β-cell death. Steady-state ER Ca²⁺ levels are maintained by the balance of Ca²⁺ transport into the ER lumen by the sarco-ER Ca²⁺ ATPase (SERCA) pumps and Ca²⁺ release via the inositol triphosphate receptors and ryanodine receptors (RyRs) (1–4). ER Ca²⁺ depletion also triggers a tightly regulated rescue mechanism that serves to replenish ER Ca²⁺ stores through a family of channels referred to as store-operated or Ca²⁺ release–activated channels (5–7). This process, known as store-operated Ca²⁺ entry (SOCE), is initiated by the dissociation of Ca²⁺ from the ER Ca²⁺ sensor, Stromal Interaction Molecule 1 (STIM1), followed by STIM1 oligomerization and translocation to the ER/plasmalemmal junctional regions (8). Here, STIM1 complexes with selective Orai Ca²⁺ channels (9) and nonspecific transient receptor potential canonical channel 1 (TRPC1), leading to the activation of Ca²⁺ influx from the extracellular space, with subsequent transfer of Ca²⁺ into the ER lumen (10,11). Although pathologic reductions in SERCA-mediated ER Ca²⁺ uptake and dysregulated RyR-mediated ER Ca²⁺ leakage have been described in the diabetic β-cell (4,12,13), a role for impaired β-cell SOCE in this phenotype remains untested.

In other cell types, SOCE Ca²⁺ transients have been implicated in a number of signaling pathways, including those that regulate proliferation, growth, inflammation, apoptosis, and lipogenesis. In addition, defective SOCE has been associated with several clinical syndromes, including immunodeficiency, myopathy, Alzheimer disease, and vascular disease (14–18). Recently, pharmacologic inhibitors of SOCE or dominant-negative forms of either Orai1 or TRPC1 were shown to decrease insulin secretion in rat islets and clonal β-cell lines (11), while STIM1 was also shown to interact with the sulfonylurea receptor 1 subunit of the K_ATP channel and regulate β-cell K_ATP activity (19).

Given these recent implications of SOCE in the regulation of insulin secretion, we hypothesized that dysfunctional β-cell SOCE may likewise contribute to diabetes pathogenesis. To this end, we profiled SOCE and the expression of SOCE molecular components in multiple diabetic models, including islets from streptozotocin (STZ)-treated mice, human and mouse islets and rat insulinoma (INS-1) cells treated with proinflammatory cytokines, INS-1 cells treated with palmitate, and human islets isolated from donors with T2D. Our data revealed a preferential loss of STIM1 expression but preserved expression of Orai1 across these models. Moreover, β-cell STIM1 loss as well as STIM1 knockdown led to impaired glucose-stimulated Ca²⁺ oscillations and insulin secretion, and increased β-cell susceptibility to ER stress, whereas STIM1 gain of function rescued these defects. Taken together, these data define a novel role for altered SOCE in diabetes and suggest that efforts to restore STIM1 expression and/or SOCE-mediated Ca²⁺ transients have the potential to improve β-cell function and health.

Mouse and human interleukin­-1β (IL-1β), interferon-γ (IFN-γ), and tumor necrosis factor-α (TNF-α) were obtained from Invitrogen (Carlsbad, CA); and 2-aminoethoxydiphenyl borate (2-APB), 1-(5-chloronaphthalene-1-sulphonyl)-¹H-hexahydro-1,4-diazepine (ML-9), and tunicamycin (TM) were obtained from Tocris Bioscience (Bristol, U.K.). Adenoviruses expressing STIM1 and Cre recombinase were from ViraQuest Inc. (North Liberty, IA) (19). Small interfering RNAs (siRNAs) were obtained from GE Healthcare (Lafayette, CO); and all other chemicals were from Sigma-Aldrich (St. Louis, MO). Supplementary Tables 1 and 2 contain a complete list of PCR primers and antibodies.

Male C57BL/6J mice were obtained from The Jackson Laboratory (Bar Harbor, ME). At 8 weeks of age, C57BL/6J were injected intraperitoneally with 50 mg/kg STZ or normal saline daily for 5 days. Mice with loxP sites flanking exon 2 of the Stim1 gene were obtained from The Jackson Laboratory and backcrossed onto a C57BL/6J background for at least 10 generations. Mice were maintained under protocols approved by the Indiana University Institutional Animal Care and Use Committee. Cages were kept in a standard light/dark cycle with ad libitum access to food and water.

Cadaveric human islets from donors without diabetes and donors with T2D were obtained from the Integrated Islet Distribution Program. The data presented include the analysis of islets from 8 donors without diabetes (5 female; 3 male) and 11 donors with T2D (7 female; 4 male). The average age (±SEM) of the donors without diabetes was 45.9 ± 10.3 years; the mean BMI was 25.6 ± 4.7 kg/m². The average age of donors with T2D was 50.9 ± 8.9 years; the average BMI was 33.2 ± 7.1 kg/m².

Rat INS-1 832/13 cells and human and mouse pancreatic islets were cultured as previously described (20,21). CRISPR/Cas9 genomic editing was used to create a STIM1 knockout (KO) INS-1 832/13 cell line in the Genome Engineering and iPSC Center at Washington University (St. Louis, MO). To mimic proinflammatory and diabetic conditions, INS-1 cells were treated with 5 ng/mL IL-1β, and mouse and human islets were treated with 5 ng/mL IL-1β, 100 ng/mL IFN-γ, and 10 ng/mL TNF-α for 24 h as previously published (12). To induce ER stress, INS-1 cells were treated with 10 μmol/L TM for 3–24 h. For viral transduction studies, 50–100 islets were handpicked within 1 h of isolation or receipt, incubated with adenovirus for 16 h, followed by an additional 48–72 h of incubation in fresh medium without adenovirus. To obtain STIM1 KO islets, isolated islets from STIM^flox/flox mice were transduced with an adenovirus encoding Cre recombinase under control of the cytomegalovirus promoter. Glucose-stimulated insulin secretion (GSIS), immunoblot, and immunofluorescence in cultured wild-type (WT) INS-1 cells, STIM1 KO cells, and isolated mouse and human islets were performed as previously described (4,12). Islet GSIS was also measured using the Biorep Perifusion System (Biorep, Miami Lakes, FL) (22). Twenty-four hours after isolation, 50 handpicked islets were loaded into each perifusion chamber; islets were perifused with Krebs buffer containing 2.8 mmol/L glucose for 20 min, followed by 16.7 mmol/L glucose for 40 min at a rate of 120 μL/min. Secreted insulin was measured using an ELISA (Mercodia, Uppsala, Sweden); results were normalized to DNA content. Quantitative real-time (qRT)-PCR was performed using previously published primer sequences (23) or primers outlined in Supplementary Table 1. To generate transmission electron micrographic images, INS-1 cells were fixed in 2% glutaraldehyde and 4% paraformaldehyde in 0.1 mol/L sodium cacodylate buffer and transferred to the Advanced Electron Microscopy Facility at the University of Chicago (Chicago, IL).

Cytosolic Ca²⁺ dynamics in INS-1 cells were measured using the FLIPR Calcium 6 Assay Kit and a FlexStation 3 Plate Reader (Molecular Devices, Sunnyvale, CA). To measure SOCE, INS-1 832/13 β-cells were loaded with Calcium 6 in growth medium containing 11 mmol/L glucose for 2 h. Immediately prior to Ca²⁺ imaging, cells were incubated in Ca²⁺-free Hanks’ balanced salt solution (HBSS) with the following composition: 138 mmol/L NaCl, 5.3 mmol/L KCl, 0.34 mmol/L Na₂HPO₄, 0.44 mmol/L KH₂PO₄, 4.17 mmol/L NaHCO₃, and 5.5 mmol/L glucose for 4 min. Ca²⁺ imaging experiments were performed according to the strategy shown in Fig. 1A. Baseline fluorescence (F0) was measured for a minimum of 10 s under Ca²⁺-free conditions and in the presence of 0.5 mmol/L EGTA (Ca²⁺ chelator), 10 μmol/L verapamil (L-type voltage-dependent Ca²⁺ channel blocker), and 200 μmol/L diazoxide (K_ATP channel opener applied to prevent voltage-dependent Ca²⁺ channel activation). Next, thapsigargin (TG; a SERCA inhibitor) was used to empty ER Ca²⁺ stores, followed by supplementation with 2 mmol/L Ca²⁺ in the media. SOCE was detected as an elevation of Calcium 6 intensity (vertical red arrow; ΔF) in response to Ca²⁺ addition, which was normalized to the basal F0 (red dotted line), according to the formula ΔF/F0. Cytosolic Ca²⁺ imaging was performed in isolated islets incubated in HBSS buffer supplemented with 2 mmol/L Ca²⁺ and the ratiometric Ca²⁺ indicator fura-2-acetoxymethylester (Fura-2 AM) (Life Technologies) using a Zeiss Z1 microscope as previously described (4,24).

To directly image ER Ca²⁺ levels, cells were transfected with an adenovirus encoding the ER-targeted D4ER probe expressed under the control of the rat insulin promoter (23). Fluorescence lifetime imaging microscopy (FLIM) was used to monitor steady-state ER Ca²⁺ levels in accordance with previously published protocols (4,25). For Förster resonance energy transfer (FRET) experiments, confocal images were acquired with a Leica TCS SP8 confocal/multiphoton imaging system (Leica Microsystems, Inc., Buffalo Grove, IL). Imaging was performed using a 448-nm single excitation laser line, and fluorescent emission collected with two hybrid detectors set to 460–500 and 515–550 nm emission slit widths. Time series images (z stacks) of two to three stage-registered fields were acquired over a period of 20 min. Images were analyzed using Leica LAS X software (version 3.3) to calculate the change in FRET ratio over time.

Differences between groups were examined for significance using either a two-tailed Student t test or one-way ANOVA followed by the Tukey-Kramer post-test using GraphPad Prism statistics software (GraphPad Software, Inc., San Diego, CA). Pearson correlations were used to analyze the relationships between STIM1 mRNA levels and donor BMI. Unless indicated, results were displayed as the mean ± SEM; a P value of <0.05 was used to indicate a significant difference between groups.

To image β-cell SOCE, INS-1 cells were loaded with Calcium 6; Ca²⁺ imaging experiments were performed according to the schematic shown in Fig. 1A and described in the research design and methods. As shown previously (26,27), pharmacologic SOCE inhibitors ML-9 and 2-APB reduced the ΔF/F0 activated in response to ER Ca²⁺ depletion (Fig. 1B and C). Moreover, ML-9 and 2-APB reduced GSIS by 61% and 68%, respectively, in INS-1 β-cells (Fig. 1D). To test whether SOCE inhibitors similarly impacted glucose-stimulated Ca²⁺ oscillations, islets from WT C57BL/6J mice were treated with either ML-9 or 2-APB and then loaded with Fura-2 AM for the analysis of glucose-stimulated Ca²⁺ responses (Fig. 1E). Although pharmacologic inhibition of SOCE had no effect on the phase 1 ΔF response to glucose (Fig. 1F), ML-9 and 2-APB reduced the average amplitude of the oscillatory response (ΔF phase 2) and increased the oscillatory period (Fig. 1G and H). In islet perifusion experiments, the addition of 2-APB to the high-glucose buffer significantly reduced both first- and second-phase insulin secretion (Fig. 1I and J). Consistent with the effects observed on glucose-stimulated Ca²⁺ oscillations, a more striking effects was observed on second-phase insulin secretion.

Next, islets were isolated from STZ or saline-treated C57BL/6J mice and glucose-stimulated Ca²⁺ imaging performed. Islets from STZ-treated mice had reduced amplitude of the ΔF phase 1 and 2 responses and a decreased oscillatory period (Fig. 2A–E). To test whether observed changes in calcium signaling may be related to changes in the expression of the molecular components of the β-cell SOCE complex, first, we compared the expression levels of STIM and Orai isoforms in human and mouse islets and INS-1 cells (Supplementary Fig. 1A–C). In human islets, mouse islets, and INS-1 cells, STIM1 was expressed at higher levels compared with STIM2. Orai2 was the most highly expressed Orai isoform in human islets. In mouse islets, Orai1 and Orai3 levels were nearly equivalent, and these were the most abundantly expressed Orai isoforms. Orai3 was the mostly highly expressed isoform in INS-1 cells. Analysis of islets from STZ-treated mice revealed a specific reduction in STIM1 gene and protein expression (Fig. 2F–H), whereas no differences in STIM2, Orai1, Orai2, or Orai3 gene expression were observed between saline and STZ-treated groups (Fig. 2F).

Homozygous STIM1 KO mice experience perinatal lethality, and STIM2 KO mice die shortly after birth (28,29). To test whether KO of STIM1 was sufficient to impair glucose-stimulated Ca²⁺ responses, islets from STIM1^flox/flox mice were transduced with an adenovirus encoding Cre recombinase. STIM1 protein was effectively reduced in Cre-transduced islets from STIM^flox/flox mice (pSTIM1KO) (Fig. 2I). Consistent with the results observed in islets isolated from STZ-treated mice, the amplitude of the phase 1 and 2 responses were reduced in pSTIM1KO islets (Fig. 2J–M). Similar to the results obtained in islets from STZ-treated mice, the oscillatory period was also reduced in pSTIM1KO islets (Fig. 2E and N). However, this was noted to be in contrast to the increased oscillatory period observed in islets treated with ML-9 or 2-APB (Fig. 1H).

Next, we tested whether the expression of SOCE molecular constituents was altered in islets from human cadaveric donors with T2D. This analysis revealed a significant reduction in STIM1 and Orai2 gene expression and a trend toward reduced Orai1 mRNA expression in islets from donors with T2D (P = 0.06) (Fig. 3A). Immunoblot analysis revealed no change in Orai1 or Orai2 protein levels (data not shown), but STIM1 protein expression was reduced by ∼40% in islets from donors with T2D (Fig. 3B and C). Next, Pearson correlation tests were used to assess the relationships between donor BMI and STIM1 mRNA levels. In islets from donors without diabetes, STIM1 mRNA levels were positively correlated with BMI (R² = 0.573; P = 0.048), whereas this correlation was lost in donors with T2D (Fig. 3D).

To define whether STIM1 expression was similarly impacted under in vitro diabetic stress conditions, human and mouse islets were treated with a mixture of proinflammatory cytokines, consisting of 5 ng/mL IL-1β, 100 ng/mL IFN-γ, and 10 ng/mL TNF-α for 24 h Cytokine treatment led to a significant reduction in STIM1 protein levels in human and mouse islets (Fig. 4A and B). Because INS-1 cells are known to be more sensitive to proinflammatory cytokines (30), INS-1 cells were treated with increasing doses of IL-1β alone for 24 h. Consistent with the results obtained in human and mouse islets, STIM1 protein and mRNA levels decreased significantly with IL-1β treatment (Fig. 4C–E), whereas no reductions in STIM2, Orai1, Orai2, or Orai3 mRNA expression were observed (Fig. 4C). Reduced STIM1 expression correlated with a dose-dependent reduction in SOCE (Fig. 4F and G). Quantitation of the ΔF in response to TG was reduced in parallel, suggesting impaired ER Ca²⁺ storage in IL-1β–treated INS-1 cells (Fig. 4H).

To test these findings in another model of diabetic stress, INS-1 cells were treated with 0.5 mmol/L palmitate combined with 25 mmol/L glucose to mimic glucolipotoxicity (GLT). In GLT-treated INS-1 cells, both STIM1 and STIM2 mRNA levels were reduced (Fig. 4I), and STIM1 protein was reduced in parallel (Fig. 4J). No significant change in Orai1–3 mRNA levels was observed. SOCE was significantly decreased by GLT treatment as well as palmitate treatment alone (Fig. 4K and L). Similar to the results observed with IL-1β treatment, the ΔF response to TG was reduced in both palmitate and GLT-treated INS-1 cells (Fig. 4K and Supplementary Fig. 2A and B).

Taken together, our data identified a specific and consistent reduction in STIM1 expression in islets and pancreatic β-cells across a variety of diabetes models. Moreover, pharmacologic inhibition of SOCE reduced GSIS and impaired glucose-stimulated Ca²⁺ oscillations, whereas the effects on calcium oscillations were partially phenocopied by STIM1 knockdown in islets.

Off-target effects of chemical SOCE inhibitors have been reported (31), including an effect of 2-APB on inositol triphosphate receptor activity. To this end, CRISPR/Cas9 genome editing was used to create an INS-1 STIM1 KO cell line (KO cells). Gene expression analysis of KO cells demonstrated the absence of STIM1 mRNA, with no compensatory upregulation of other SOCE constituents including STIM2 and Orai1–3 (Fig. 5A). In addition, no change in Serca2b expression was observed. Similar to the findings observed in our in vitro models of cytokine, GLT, and palmitate-stress, the ΔF response to TG was reduced in STIM1 KO cells (Supplementary Fig. 2C and D). In addition, SOCE measured in response to TG and carbachol treatment was significantly reduced in KO cells (Fig. 5B–D), whereas no further reduction in SOCE was observed when STIM1 KO cells were treated with IL-1β (Supplementary Fig. 3A and B), GLT (Supplementary Fig. 3C and D), ML-9, or 2-APB (Supplementary Fig. 3E).

To more definitively examine the effect of STIM1 KO on ER Ca²⁺ levels, WT and KO cells were transduced with the ER-targeted, FRET-based, D4ER Cameleon probe, and imaging experiments were performed in Ca²⁺ containing HBSS buffer. Under basal conditions, STIM1KO cells displayed a lower FRET/donor ratio, which is indicative of decreased ER Ca²⁺ levels (Fig. 5E). To confirm these findings, FLIM analysis was performed to measure the lifetime of the D4ER cyan fluorescent protein (CFP) donor. As expected, the lifetime of the D4ER donor was increased in KO cells compared with WT cells (Fig. 5F and Supplementary Fig. 3F). Finally, GSIS was measured in WT and STIM1 KO cells. Similar to results obtained with pharmacologic SOCE inhibition (Fig. 1D), STIM1 deletion significantly reduced GSIS, whereas no effect on basal insulin secretion was observed (Fig. 5G).

We hypothesized reductions in ER Ca²⁺ levels arising from impaired SOCE may likewise impact stress responses in STIM1 KO cells. To test this, WT and KO cells were treated with 25 mmol/L glucose + 0.5 mmol/L palmitate (GLT) or 10 μmol/L TM for 3 h. In response to both GLT and TM, an increase in the spliced/unspliced XBP-1 ratio was observed in KO cells (Fig. 6A). Moreover, STIM1 KO cells exhibited increased cleaved caspase-3 activation in response to IL-1β treatment (Fig. 6B and C). Next, to compare acute and chronic STIM1 loss, siRNA-mediated STIM1 knockdown was performed in WT INS-1 cells, resulting in an ∼50% reduction in STIM1 gene and protein expression (Fig. 6D and E). Similar to the results observed in STIM1KO cells, siRNA-treated cells exhibited significantly higher levels of spliced/unspliced XBP-1 in response to TM treatment (Fig. 6F). Finally, to determine whether these functional changes correlated with changes in ER ultrastructure, electron micrographic images of WT and KO cells were analyzed. In contrast to regularly spaced stacks of ER observed in WT cells, the ER was swollen and dilated in STIM1 KO cells (Fig. 6G). Taken together, these data indicate that STIM1 loss is associated with changes in ER morphology and increased β-cell susceptibility to TM-, cytokine-, and GLT-mediated stress.

To test the effects of STIM1 gain of function, WT and KO cells were transduced with a STIM1-expressing adenovirus, and a dose-dependent increase in STIM1 expression was observed (Fig. 7A). Consistent with this, a dose-dependent increase in SOCE was also observed with STIM1 overexpression (Fig. 7B and C). Next, to test the effect of STIM1 overexpression on ER Ca²⁺ levels, STIM1 KO cells were transduced with the D4ER-expressing adenovirus in combination with either an empty vector or STIM1-expressing adenovirus. Under steady-state conditions, STIM1 restoration led to an increase in the FRET/CFP ratio in KO cells (Fig. 7D). This result was confirmed by FLIM analysis, in which the lifetime of the D4ER donor probe was significantly reduced by STIM1 overexpression in KO cells (Fig. 7E).

Next, FRET analysis was performed to dynamically monitor ER Ca²⁺ storage in WT and STIM1 KO cells transduced with an empty virus or STIM1-expressing adenovirus. No differences in the FRET/CFP ratio were observed at baseline under Ca²⁺-free conditions. The initial response to carbachol was similar between the groups. In contrast, STIM1 KO cells demonstrated impaired ER Ca²⁺ restoration after ER Ca²⁺ depletion when Ca²⁺ was added to the buffer. Moreover, ER Ca²⁺ restoration was significantly improved in KO cells by STIM1 overexpression (Fig. 7F–H).

Next, we tested whether STIM1 overexpression was sufficient to protect against IL-1β–mediated activation of cleaved caspase 3. In both WT and STIM1 KO cells, STIM1 overexpression reduced cleaved caspase 3 protein levels (Fig. 7I and J). Consistent with this, STIM1 overexpression rescued SOCE under conditions of cytokine stress (Supplementary Fig. 4A and B). Finally, human cadaveric islets from donors with T2D were transduced with the STIM1-expressing adenovirus or an empty viral control (Fig. 7K). Compared with islets transduced with empty virus, GSIS was significantly improved in STIM1-transduced human islets (Fig. 7L).

Recent findings suggest that ER dysfunction triggers a range of chronic human diseases, including Alzheimer disease, Parkinson disease, atherosclerosis, and diabetes (32–35). In the pancreatic β-cell, a key determinant of ER function is the maintenance of robust levels of intraluminal Ca²⁺, which is required for protein folding and a number of key cellular signaling events. Reductions in β-cell ER Ca²⁺ occur in response to proinflammatory and glucolipotoxic stress and contribute to the pathophysiology of both major forms of diabetes. Under normal conditions, ER Ca²⁺ store depletion triggers a tightly regulated rescue mechanism known as SOCE or Ca²⁺ release-activated Ca²⁺ current. This process was first proposed by Putney (36). However, a complete understanding of the SOCE molecular complex remained elusive until 2005, when STIM1 was identified as the long sought after “ER Ca²⁺ sensor” (8). Since that time, SOCE has been accepted as a predominant pathway of Ca²⁺ entry into nonexcitable cells, where it has been shown to regulate a variety of cellular functions, including proliferation, growth, inflammation, apoptosis, and lipogenesis (14,37). To date, however, a role for SOCE in excitable and secretory cells, including the pancreatic β-cell, has remained incompletely explored.

Here, we show impairment of glucose-stimulated Ca²⁺ oscillations and reduced insulin secretion in response to pharmacologic SOCE inhibition and under conditions of STIM1 loss. Across multiple models of diabetes, our results revealed a preferential reduction in STIM1 mRNA and protein levels in pancreatic islets and β-cells. Furthermore, we show that the genetic loss of STIM1 was sufficient to reduce β-cell SOCE, glucose-stimulated Ca²⁺ oscillations, and insulin secretion, while simultaneously increasing susceptibility to ER stress and cell death in response to glucolipotoxic and proinflammatory conditions. Previously, STIM1, Orai1, and TRPC1 have been identified as key constituents of the β-cell SOCE complex, whereas pharmacologic SOCE inhibition or a dominant-negative form of Orai1 or TRPC1 was shown to reduce insulin secretion in rat islets and clonal β-cell lines (11). Our data highlight an expanded role for STIM1 in nucleating this process and implicate a loss of STIM1 and impaired SOCE in β-cell dysfunction under diabetic conditions.

The impact of altered SOCE on insulin exocytosis is likely to be multifactorial. However, a key finding from our study is that SOCE inhibition and STIM1 loss impaired first-phase glucose-stimulated cytosolic Ca²⁺ responses as well as glucose-induced Ca²⁺ oscillations. These effects were similar to those observed in a mouse model of SERCA2 haploinsufficiency, where reductions in SERCA2 were shown to reduce β-cell ER Ca²⁺ levels and cytosolic Ca²⁺ oscillations (4). Thus, our data offer additional support to the notion that ER Ca²⁺ stores shape the architecture of glucose-induced Ca²⁺ oscillations as well as the β-cell insulin secretory response. Intriguingly, STIM1 was shown recently to interact with the sulfonylurea receptor 1 subunit of the K_ATP channel and regulate β-cell K_ATP activity in MIN6 cells (19). In this report, short hairpin RNA–mediated STIM1 knockdown reduced K_ATP channel activation, whereas STIM1 reconstitution restored K_ATP activity, suggesting that reductions in insulin secretion with STIM1 loss/SOCE inhibition may also result from impaired K_ATP activity (19). A hallmark of advanced human T2D is the development of an impaired response to sulfonylurea medications, which act to close the K_ATP channel to increase insulin secretion (38–40). In our data, we observed a positive correlation between donor BMI and STIM1 expression levels in islets from humans without diabetes, but this correlation was lost in donors with T2D. Moreover, our data revealed that STIM1 reconstitution in islets from donors with T2D improved GSIS. These findings imply that STIM1 upregulation may be a compensatory response observed in islets from obese donors that is responsible for the initial maintenance of insulin secretion in the face of advancing insulin resistance. By corollary, our data suggest that reductions in STIM1 could play an important role in the development of impaired insulin secretion and loss of secretagogue efficacy in T2D.

Pathologic reductions in β-cell ER Ca²⁺ levels have been linked with the development of ER stress. Thus, reduced ER Ca²⁺ levels in STIM1-deficient cells led us to hypothesize an additional role for STIM1 in ER stress signaling. Indeed, we found that genetic loss of STIM1 as well as STIM1 knockdown increased β-cell susceptibility to TM-induced ER stress and proinflammatory cytokine–induced cell death. In contrast, STIM1 overexpression was able to protect against proinflammatory cytokine–induced activation of cleaved caspase 3. Recent reports have implicated SERCA2 loss (4,12,41), as well as RyR-mediated ER Ca²⁺ release, in the activation of ER stress signaling in diabetes (13). Here, we have identified an additional role for SOCE dysfunction and STIM1 loss in this phenotype. These observations are in agreement with findings observed in other cell types, including both excitable cells in the central nervous system as well as nonexcitable mouse embryonic fibroblasts (42,43).

Our data revealed a consistent loss of STIM1 in the β-cell across multiple models of diabetes. However, a potential caveat to our observation of reduced STIM1 expression in islets from human donors with T2D is that loss of islet β-cell number may also accompany T2D (44). In addition, STIM1 did not appear as differentially regulated in a microarray analysis of laser-captured islets from human donors with T2D (45). However, we were able to confirm the β-cell downregulation of STIM1 expression using in vitro and ex vivo diabetogenic stress in isolated islets and INS-1 β-cells. Among these models, we found that STIM1 mRNA and protein levels were reduced by proinflammatory cytokines, which are known to be a prominent component of the pathophysiology of type 1 diabetes (46). Additional studies are needed to define whether β-cell SOCE dysregulation may also be present in human type 1 diabetes.

We have not yet explored the intervening pathways that lead to loss of β-cell STIM1 expression under diabetic conditions. In other cell types, the inhibition of both AKT and mammalian target of rapamycin signaling led to a similar loss of STIM1 (47). Notably, both of these signaling pathways play an essential role in the maintenance of β-cell health and function (48,49). In addition, nuclear factor-κB has been shown to directly bind and modulate STIM1 gene promoter activity in other cell types (50). We and others have previously identified a deleterious effect of nuclear factor-κB–mediated nitric oxide signaling on ER calcium regulation (41,51), and this will be tested in follow-up studies. Here, we focused on the loss of STIM1 expression as the cause of impaired SOCE. However, in other cell types, mislocalization or impaired formation of the SOCE complex has been observed under disease conditions (52). Finally, STIM2 overexpression has been shown to have a strong negative effect on endogenous SOCE in human embryonic kidney 293, PC12, A7r5, and Jurkat T cells (53). Thus, our observation of impaired SOCE in STIM1 KO cells and a reduced Ca²⁺ oscillatory amplitude in STIM1-deleted islets could also be partly explained by alterations in the STIM1/STIM2 ratio.

Notwithstanding these uncertainties, our findings indicate that the loss of STIM1 expression and impaired SOCE in rodent and human models of diabetes resulted in decreased β-cell ER Ca²⁺ levels, increased ER stress, abnormal Ca²⁺ oscillation patterns, and decreased insulin secretion (Fig. 8). The identification of this novel role for STIM1 in the pathogenesis of diabetes suggests that the restoration of STIM1 expression and/or reconstitution of SOCE has the potential to improve β-cell health and function.

Acknowledgments. The authors thank Dr. Richard Day (Indiana University) for helpful advice and technical discussions. The authors also thank Sukrati Kanojia, Preethi Krishnan, Wataru Yamamoto, Morgan Robertson, Gary Considine, Kara Orr, and Daenique Jengelley (Indiana University) for expert technical assistance.

Funding. This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grants R01-DK074966 and R01-DK092616 (to M.W.R.), R01-DK-093954 and UC4-DK-104166 (to C.E.-M.), and P30-DK-097512 (Indiana Diabetes Resarch Center Islet and Physiology, Translation, and Imaging Cores), and gifts from the Sigma Beta Sorority, the Ball Brothers Foundation, the George and Frances Ball Foundation (to C.E.-M.). X.T. was supported by the Diabetes and Obesity DeVault Fellowship at the Indiana University School of Medicine. R.N.B. was supported by National Institutes of Health National Institute Allergy and Infectious Disease Training Grant T32-AI-060519 and a JDRF Postdoctoral Research Award (3-PDF-2017-385-A-N).

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

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

Author Contributions. T.K. directed the conception and design of the study, the data analysis and interpretation, the collection and assembly of data, and the writing of the manuscript. X.T., S.T., R.N.B., H.I., C.-C.L., and P.S. participated in data collection and critical revision of the manuscript. P.G. and M.W.R. contributed to data analysis, provided critical reagents, and provided critical revision of the manuscript. C.E.-M. directed funding acquisition and study conception and design; participated in the collection and assembly of data; contributed to data analysis; directed the writing of the manuscript; and gave final approval of the manuscript. C.E.-M. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 76th Scientific Sessions of the American Diabetes Association, New Orleans, LA, 10–14 June 2016, and at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017.