Sorcin Links Pancreatic β-Cell Lipotoxicity to ER Ca²⁺ Stores

Pancreatic β-cell dysfunction is central to the pathogenesis of type 2 diabetes. During the progression of obesity and insulin resistance, pancreatic islets of Langerhans initially increase β-cell mass and overproduce insulin (1). The increase in biosynthetic demand induced by chronic hyperglycemia activates the unfolded protein response (UPR), while increases in circulating free fatty acids and cytokines lower endoplasmic reticulum (ER) calcium (Ca²⁺) stores (2,3), triggering ER stress and apoptosis if prolonged (4). The molecular mechanisms linking lipotoxicity and associated inflammation (5) to ER Ca²⁺ stores are largely unknown. Sorcin (gene name SRI) is a 22-kDa member of the penta-EF-hand family of calcium binding proteins that undergoes Ca²⁺-dependent conformational changes (6–10). Sorcin is highly conserved among mammals and strongly expressed in primary mouse islets (11). In extrapancreatic cells, notably cardiac myocytes, sorcin associates with the ryanodine receptor (RyR) (12), the pore-forming α₁ subunit of voltage-dependent L-type Ca²⁺ channels (L-type VDCC) (13), and sarcoendoplasmic reticulum Ca²⁺ ATPase (SERCA) pumps (14) to modulate excitation-contraction coupling through changes in intracellular Ca²⁺ homeostasis (12). Sorcin inhibits RyR activity (15) and plays a role in terminating Ca²⁺-induced Ca²⁺ release (12), an inherently self-sustaining mechanism that, if unchecked, may deplete intracellular Ca²⁺ stores (16).

We have recently shown that siRNA-mediated knockdown of sorcin in MIN6 insulinoma β-cells resulted in an apparent reduction in ER Ca²⁺ stores, as judged by stimulation with an inositol 1,4,5-triphosphate (IP3) mobilizing agonist, and an inhibition of glucose-stimulated insulin secretion (GSIS) (17). These data indicated that sorcin may be required to maintain intracellular Ca²⁺ stores, possibly through its known capacity to inhibit RyRs (15) and activate SERCA pumps (14).

Since we and others have demonstrated that elevated palmitate (2) and cytokine levels (18,19) cause ER stress and apoptosis in pancreatic β-cells at least in part by decreasing ER Ca²⁺ stores (3), it follows that sorcin overexpression might protect against ER stress induced by inflammation and lipotoxicity. Indeed, recent data regarding human islets showed a decrease in sorcin expression induced by TNFα (20).

In the present report, we test this hypothesis using 1) mice bearing null alleles of the sorcin gene (Sri^−/−), 2) transgenic mouse lines overexpressing sorcin in the pancreatic β-cell, and 3) adenovirus-mediated overexpression of sorcin in isolated human and murine islets.

Murine sorcin cDNA (17) was inserted in pBI-L vector (PvuII-NotI sites), which contains a bidirectional Tet-responsive promoter driving the expression of both mSRI and firefly luciferase (21). After injection into the pronucleus of 0.5-day-old pure C57BL/6 fertilized oocytes (Embryonic Stem Cell and Transgenic Facility, Medical Research Council, London, U.K.), two founders, bearing 1 and 10 copies of the transgene (TetOn-Sri-1 and TetOn-Sri-10), were identified by PCR screening. β-Cell selectivity was achieved using the Tet-on system and RIP7-rtTA mice, which express the reverse tetracycline transactivator (rtTA) under the control of the rat insulin 2 promoter (21). Hemizygous TetOn-Sri-1 or -10 mice were crossed with homozygous RIP7-rtTA mice to generate double hemizygous TetOn-Sri/RIP7-rtTA (hereafter named SRI-tg1 and SRI-tg10) and single hemizygous RIP7-rtTA as littermate controls. Doxycycline (0.5 g/L in drinking water) and a high-fat diet (HFD; 60% kcal as fat, mainly saturated) were administered from 4 weeks of age unless specified otherwise.

Homozygous sorcin-null mice (Sri^−/−) on a 129S1/SvImJ genetic background were generated by homologous recombination as previously described (22). In brief, the targeting construct was generated by flanking exon 3, present in both sorcin isoforms (accession no. NM_001080974.2 and NM_025618.3) with loxP sites for the Cre recombinase and inserting a phosphoglycerol kinase promoter–driven neomycin selection cassette flanked by an additional loxP site in the intron between exons 3 and 4.

Mice were fasted overnight for 14 h. Glucose solution (20% d-glucose/water, weight for volume, 1–3 g/kg body weight) or human regular insulin solution (0.5 or 1 units/kg, catalog no. 19278; Sigma-Aldrich) was administrated intraperitoneally and blood glucose was measured from the tail vein at 0, 15, 30, 60, 90, and 120 min using an ACCU-CHECK Aviva glucometer (Roche). Plasma insulin levels were measured using an ultrasensitive mouse insulin ELISA kit (Crystal Chem, Downers Grove, IL), and plasma glucose was assessed by Glucose Assay Kit (catalog no. 65333; Abcam) when above the glucometer detection limit.

Plasmid pGL3-hG6PC2(−1075+124), containing the proximal promoter of the human glucose-6-phosphatase catalytic subunit-2 (G6PC2) gene upstream of luciferase reporter gene, was generated by PCR using human genomic DNA and the following primers: forward 5′-ACACGGTACCATCCTAGACACAATCCAGCTCTCT-3′ and reverse 5′-ACACAAGCTTTAAATGAAAAAGATATTCCTGGGG-3′. The resulting 1,220-bp fragment was subcloned into pCR2.1 by TA cloning, digested by KpnI-XhoI, and subcloned into pGL3basic. A nuclear factor of activated T-cells (NFAT) luciferase reporter containing three tandem repeats of a 30-bp fragment of the IL-2 promoter for analysis of NFAT activity was a gift from Dr. Toren Finkel (Laboratory of Molecular Biology, National Heart, Lung, and Blood Institute, National Institutes of Health) (23). p5xATF6-GL3, containing five tandem repeats of ATF6 binding sites, was a gift from Ron Prywes (Department of Biological Sciences, Columbia University) (24). Plasmids pLKO.1-shSc (scrambled), pLKO.1-shSRI144, and pLKO.1-shSRI457 were constructed using the pLKO.1-TRC cloning vector from Addgene (plasmid no. 10878, protocol http://www.addgene.org/tools/protocols/plko/) and the oligonucleotides presented in Supplementary Table 1. Plasmid pAd-hSRI was generated by subcloning the human sorcin cDNA (pDNR-LIB-SRI) in pAdTrackCMV (BglIII-HindIII sites). All the constructs were verified by DNA sequencing. The adenovirus Ad-hSRI-GFP was subsequently produced as in Noordeen et al. (17). Ad-mSRI-GFP encoding the murine sorcin cDNA and Ad-Null-GFP (empty vector) were described in Noordeen et al. (17).

MIN6 β-cells were used between passages 24 and 39 as in Noordeen et al. (17). Human and rat β-cell lines 1.1B4 and INS1(832/13) and HEK293 cells were cultured as in Noordeen et al. (25). Cells were transfected using Lipofectamine 2000 and Opti-Mem (Invitrogen), and luciferase assays were performed using a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions.

Human islets from normoglycaemic donors were cultured as in Hodson et al. (26). Donor characteristics are presented in Supplementary Table 2. Pancreatic mouse islets were isolated and cultured as previously described (27). Transgenic islets were cultured in the presence of 0.5 µg/mL doxycycline to sustain transgene expression. Human and WT mouse islets (10-week-old C57BL/6 mice) were transduced with Ad-hSRI-GFP, Ad-mSRI-GFP, or Ad-Null-GFP adenovirus at a multiplicity of infection of 100 for 48 h prior to total RNA extraction, GSIS, or intracellular Ca²⁺ imaging.

Total RNA was extracted, reversed transcribed, and analyzed as described in Sun et al. (28). Primers for SYBR Green assays are presented in Supplementary Table 3. Expression of each gene was normalized to β-actin and fold change in mRNA expression versus controls calculated using the 2^−ΔCT method.

Total RNA isolated from islets from 8-week-old, HFD-fed six SRI-tg10 male mice and six RIP7rtTA (littermate controls) was sent to the High Throughput Genomics Facility, Wellcome Trust Centre for Human Genetics, University of Oxford, and analyzed on Illumina MouseWG-6 v2 Expression BeadChips.

Insulin secretion assays on murine and human islets were performed as previously described in Leclerc et al. (29). Secreted and total insulin contents were quantified using a homogenous time-resolved fluorescence kit (HTRF) insulin kit (Cisbio).

Human islets were cultured for 72 h with either BSA or 0.5 mmol/L BSA-palmitate in 5.5 mmol/L glucose RPMI, and MIN6 cells were transduced with Ad-mSRI-GFP or Ad-Null-GFP adenoviruses for 24 h, prior to 48 h treatment with BSA or 0.5 mmol/L BSA-palmitate in 25 mmol/L glucose DMEM before total RNA extraction and qRT-PCR analysis.

Western blotting was performed as in Leclerc et al. (29) using mouse monoclonal anti-sorcin (1:300, 25B3; Invitrogen) and mouse monoclonal anti–α-tubulin (1:10,000–20,000, catalog no. T5168; Sigma-Aldrich) antibodies.

Imaging of free cytosolic Ca²⁺ concentration ([Ca²⁺]_cyt) in isolated islets was performed using the trappable intracellular fluorescent Ca²⁺ dye Fura-2-AM (Invitrogen) (27). Imaging data were analyzed with ImageJ software using an in-house macro, and the fluorescence emission ratios were derived after subtracting background fluorescence.

Clusters of isolated islets were transduced for 48 h with an adenovirus encoding the low-Ca²⁺-affinity sensor D4 addressed to the ER, Ad-RIP-D4ER, as in Ravier et al. (30). Prior to acquisitions, cells were preincubated in Krebs-Ringer bicarbonate HEPES (KRBH) media (17) containing 3 mmol/L glucose for 1 h at 37°C and were perifused with KRBH supplemented with 17 mmol/L glucose and 250 μmol/L diazoxide (Diaz) (catalog no. D9035; Sigma-Aldrich) with the subsequent consecutive additions of 100 μmol/L acetyl-choline (catalog no. A2661; Sigma-Aldrich) and 1 μmol/L thapsigargin (catalog no. 586005; Calbiochem). Image analysis was performed as above.

Analyses were performed as described in Sun et al. (28).

Data are presented as means ± SEM. Significance was assessed by appropriate unpaired or paired two-tailed Student t tests or one- or two-way ANOVA as indicated, using GraphPad Prism 6.0 or Microsoft Excel. P < 0.05 was considered significant.

Studies involving human islets were approved by the National Research Ethics Committee London as detailed in Hodson et al. (26). All procedures involving animals received ethical approval and were compliant with the U.K. Animals (Scientific Procedures) Act 1986 or approved by the University Committee on Use and Care of Animals (University of Michigan, Ann Arbor, MI). Animals were housed two to five per individually ventilated cage in a pathogen-free facility with a 12-h light-dark cycle and had free access to food and water.

We previously reported that sorcin silencing in MIN6 cells leads to a complete abolition of ATP-evoked Ca²⁺ release from intracellular stores and an inhibition of GSIS (17). These findings prompted us to investigate the roles of sorcin in β-cell pathophysiology provoked by lipotoxicity, a condition known to trigger ER stress and β-cell failure (2).

In line with our findings in cell lines (17), sorcin-null mice (Sri^−/−, standard chow [SD] fed) displayed decreased glucose tolerance compared with sex-, weight-, and age-matched wild-type (WT) controls during intraperitoneal glucose tolerance tests (IPGTTs) (area under the curve [AUC], arbitrary units, WT vs. Sri ^−/−, 2 months old: 43.5 ± 1.64 vs. 48.0 ± 1.1, n = 6–10, P < 0.05; 9 months old: 39.2 ± 2.5 vs. 49.1 ± 1.9, n = 4–7, P < 0.01) (Fig. 1A and B).

To determine whether sorcin overexpression might be protective against β-cell stress, we generated transgenic mice overexpressing sorcin in the pancreatic β-cell on the C57BL/6 genetic background, since males of this strain become glucose intolerant and insulin resistant under an HFD (31). Sorcin mRNA and protein levels were increased by at least twofold in isolated islets from SRI-tg1 and SRI-tg10 mice compared with those from littermate controls (Supplementary Fig. 1).

Glucose tolerance was improved in HFD-fed SRI-tg1 and SRI-tg10 male mice compared with their littermate controls during IPGTTs (AUC, arbitrary units, controls vs. SRI-tg1: 128.7 ± 6.1 vs. 101.2 ± 8.1, n = 7–8, P < 0.05; controls vs. SRI-tg10: 95.8 ± 5.4 vs. 73.0 ± 2.4, n = 9–13, P < 0.001) (Fig. 1C and D), despite similar insulin sensitivity as assessed by intraperitoneal insulin tolerance tests (IPITTs) (Fig. 1E and F, left panels) and body weights (Fig. 1E and F, right panels). Remarkably, the above phenotype was not apparent in vivo in the absence of β-cell stress, i.e., during normal chow feeding (Supplementary Fig. 2).

We next investigated whether the changes in glucose tolerance observed in SRI-tg and Sri^−/− mice were secondary to changes in insulin secretion. In vivo glucose-stimulated insulin release was assessed in SRI-tg10 and Sri^−/− mice by IPGTTs (3 g glucose/kg body weight). As shown in Fig. 2A (top panel), plasma insulin concentrations were significantly higher at 30 min in SRI-tg10 compared with controls (plasma insulin, ng/mL, SRI-tg10 vs. controls, 30 min: 0.60 ± 0.06 vs. 0.43 ± 0.05, P < 0.05, n = 5–7), despite similar concomitant blood glucose values (Fig. 2A, bottom panel). Conversely, Sri^−/− males showed a marked impairment of GSIS compared with WT controls, with plasma insulin concentrations barely rising during IPGTT (AUC insulin, WT vs. Sri^−/−, arbitrary units: 1.01 ± 0.23 vs. 0.53 ± 0.10, n = 4–6, P < 0.05) (Fig. 2B, top panel), despite robust increases in associated blood glucose values (Fig. 2B bottom panel).

We next explored whether the enhanced GSIS observed in SRI-tg10 islets might be secondary to an increase in β-cell mass. As shown in Fig. 2C–E, there were no significant changes in mean pancreas surface, islet size, individual β-cell or α-cell mass, although there was a significant decrease in β-cell to α-cell ratio.

The sorcin-induced improvement in in vivo GSIS was also observed ex vivo in islets isolated from SRI-tg1/10 mice and from human and mouse islets transduced for 48 h with an adenovirus encoding sorcin. Indeed, when stimulated with 17 mmol/L glucose, SRI-tg10 islets secreted 55% more insulin compared with control islets (insulin, percent of total, controls vs. SRI-tg10: 0.372 ± 0.02 vs. 0.577 ± 0.07, n = 3, P < 0.05), without any changes in insulin secretion at 3 mmol/L glucose (Fig. 2F), whereas SRI-tg1 islets secreted 68% more insulin at high glucose than controls (Supplementary Fig. 3). Adenovirus-mediated overexpression of sorcin consistently increased GSIS in human (Fig. 2G) and mouse (Fig. 2H) islets at 17 mmol/L glucose compared with islets transduced with a null-GFP virus.

We next assessed whether the enhanced GSIS observed after sorcin overexpression was associated with changes in intracellular Ca²⁺ dynamics. Islets isolated from HFD-fed SRI-tg1 and SRI-tg10 mice were loaded with Fura-2 and perifused sequentially with low (3 mmol/L) and elevated (17 mmol/L) glucose concentrations. High glucose elicited a greater [Ca²⁺]_cyt response in sorcin-overexpressing islets compared with controls (AUC, arbitrary units, controls vs. SRI-tg1: 100.0 ± 5.0 vs. 120.1 ± 5.0, n = 3, P < 0.05 [Fig. 3A]; controls vs. SRI-tg10: 100.0 ± 17.80 vs. 149.9 ± 16.2, n = 3, P < 0.05 [Fig. 3B]). Likewise, transduction of dissociated human (Supplementary Fig. 4) or mouse (not shown) islets in vitro with sorcin-encoding adenoviruses significantly increased the number of islets displaying strong and high-amplitude [Ca²⁺]_cyt oscillations in response to 17 mmol/L glucose.

Free Ca²⁺ in the ER ([Ca²⁺]_ER) was measured in clusters of isolated islets from HFD-fed SRI-tg1, SRI-tg10, and their littermate controls, transduced for 48 h with an adenovirus encoding the low-Ca²⁺-affinity sensor D4 addressed to the ER under the control of the insulin promoter Ad-RIP-D4ER (30), and incubated in 17 mmol/L glucose with the addition of 250 μmol/L Diaz to fully open ATP-sensitive K⁺ channels and prevent extracellular Ca²⁺ influx (30). After acetylcholine-induced ER Ca²⁺ release, transgenic islets experienced a larger fall in [Ca²⁺]_ER than control islets, indicating a higher initial [Ca²⁺]_ER content (Fig. 3C and D). We next fully depleted the ER Ca²⁺ stores in islets isolated from HFD-fed SRI-tg10 male mice and littermate controls using the SERCA pump inhibitor cyclopiazonic acid (32,33) before measuring [Ca²⁺]_cyt in response to 17 mmol/L glucose. Under these conditions, the subsequent increase in [Ca²⁺]_cyt induced by high glucose was no longer significantly different between sorcin-overexpressing islets and control islets (AUC [Ca²⁺]_cyt, arbitrary units, controls vs. SRI-tg10: 100.0 ± 13.5 vs. 104.6 ± 10.3, n = 3, NS) (Supplementary Fig. 5). Taken together, these results are consistent with a positive role for sorcin in GSIS and intracellular Ca²⁺ homeostasis, corroborating our in vitro data in MIN6 insulinoma cells (17).

To further explore the underlying mechanisms behind sorcin’s actions, we performed a transcriptomic analysis of islets from HFD-fed SRI-tg10 mice and controls using oligonucleotide microarrays (GEO accession no. GSE72719) (Ingenuity Pathway Analysis presented in Supplementary Table 4). Interestingly, G6pc2, one of the most highly expressed genes in β-cells (9), was strongly repressed in islets from SRI-tg10 mice. Subsequent qRT-PCR analysis in isolated islets from SRI-tg10, SRI-tg1, and Sri^−/− mice confirmed the inverse relationship between Sri and G6pc2 expression levels (Fig. 4A–F). Indeed, islets from SRI-tg1 and SRI-tg10 mice displayed 35 and 56% decreases, respectively, in G6pc2 mRNA levels (Fig. 4D and E), whereas Sri mRNA levels were increased 20- and 42-fold, respectively (Fig. 4A and B). In islets from Sri^−/− mice, sorcin expression was reduced by >90% and there was a 3.6-fold increase in G6pc2 expression (Fig. 4C and F). G6PC2 is an islet-specific isoform of glucose-6-phosphatase, which negatively regulates basal GSIS (9). G6PC2 is also a major determinant of fasting blood glucose in humans as revealed by genome-wide association studies (34). Accordingly, HFD-fed SRI-tg10 mice displayed lower fasting blood glucose throughout the study compared with controls (Fig. 4G).

We next overexpressed sorcin in human islets with an adenoviral vector and likewise observed a reduction in G6PC2 mRNA levels, and in the levels of mRNA encoding the ER stress markers C/EBP homologous protein (CHOP) and glucose-regulated protein 78/binding immunoglobulin protein (GRP78/BiP) (Fig. 5A).

Given the apparent protection conferred by sorcin overexpression under lipotoxic conditions, we next investigated the regulation of endogenous sorcin during HFD in vivo and in human islets and MIN6 cells cultured in the presence of palmitate. Sorcin expression was downregulated in islets from HFD-fed C57BL/6 (Fig. 5B) and DBA2J mice (Supplementary Fig. 6) compared with chow-fed mice. Furthermore, human islets incubated for 72 h in the presence of palmitate and 5.5 mmol/L glucose showed a significant reduction in sorcin expression and an increase in G6PC2 and CHOP expression (Fig. 5C). Similarly, lipotoxicity experiments in MIN6 cells also demonstrated a profound suppression of sorcin expression accompanied by a robust increase in G6pc2, Chop, and Grp78/BiP mRNA after exposure to palmitate (Fig. 5D). Remarkably, adenovirus-mediated overexpression of sorcin during lipotoxic conditions in MIN6 cells prevented the increase in G6pc2, Chop, and Grp78/BiP expression (Fig. 5E).

We next determined whether the repressive effect of sorcin on G6PC2 expression was transcriptionally mediated. As shown in Fig. 6A (DMSO, gray bar), overexpressed sorcin repressed the activity of a 1.2-kb proximal fragment of the human G6PC2 promoter (hG6PC2p) in MIN6 cells. Conversely, the activity of a reporter containing three tandem repeats of an NFAT binding site (3xNFATr) was significantly stimulated by sorcin (Fig. 6B, DMSO, gray bar), consistent with the increase in cytosolic [Ca²⁺] induced by sorcin (35). In order to confirm the roles of NFAT and [Ca²⁺]_cyt in mediating the inhibitory effect of sorcin on G6PC2 expression, we showed that NFAT cDNA cotransfection repressed hG6PC2p activity while robustly stimulating 3xNFATr (Fig. 6A and B, DMSO, black bars). In silico TRANSFAC analysis revealed three putative NFAT binding sites on G6PC2 promoter (not shown). We next added Diaz (100 μmol/L, inhibiting Ca²⁺ influx) and cyclosporine A (CsA, 0.2 μmol/L, inhibiting NFAT nuclear translocation) in the culture medium. As expected, both agents prevented the inhibitory and stimulatory effects of sorcin on hG6PC2p and 3xNFATr, respectively (Fig. 6A and B, Diaz and CsA, gray bars). Moreover, the addition of CsA stimulated the activity of hG6PC2p compared with DMSO, whereas Diaz inhibited it (Supplementary Fig. 7), confirming the repressive contribution of the NFAT signaling pathway. The suppressive effect of Diaz, however, implies additional [Ca²⁺]_cyt-dependent stimulatory pathways. Intriguingly, the addition of Diaz and CsA (up 1 μmol/L, not shown) did not prevent the inhibitory effects of NFAT on hG6PC2p but prevented the stimulatory effect of NFAT on 3xNFATr (Fig. 6A and B, Diaz and CsA, black bars).

The inverse relationship between sorcin expression and the ER stress markers CHOP and GRP78/BiP prompted us to investigate the effect of sorcin on the UPR. To study the ATF6 branch (36), we chose a reporter assay containing five tandem repeats of ATF6 binding sites (24), since ATF6 is cleaved in the Golgi in response to ER stress before translocating to the nucleus (37). The activity of the ATF6 luciferase reporter was reproducibly stimulated by tunicamycin and thapsigargin, two ER stress inducers, and inhibited by the chemical chaperone 4-PBA, confirming its sensitivity to ER stress (Fig. 7A–D). Sorcin cotransfection increased the activity of the ATF6 luciferase reporter in three β-cell lines, i.e., MIN6, 1.1B4, and INS1 (832/13), as well as in HEK293 cells, in basal (DMSO) and stimulated (thapsigargin and tunicamycin) conditions, but not in the presence of 4-PBA, compared with cotransfection with GFP (Fig. 7A–D). Conversely, sorcin silencing in MIN6 cells by short hairpin RNA decreased the activity of the ATF6 reporter (Fig. 7E and F). Sorcin overexpression did not affect XBP-1 splicing, used as a surrogate of the IRE-1 branch, either in basal conditions or after thapsigargin treatment of MIN6 (Fig. 7G and H) or HEK293 (not shown) cells, whereas CHOP repression indicated repression of the PERK branch.

The findings herein indicate that sorcin lies on a pathway linking β-cell lipotoxicity to ER calcium and ER stress, representing a mechanism for dysregulation of β-cell function under conditions of metabolic stress (Fig. 8). Thus, we show that sorcin is downregulated in pancreatic β-cells under conditions of lipotoxic stress (Fig. 5), whereas overexpression of sorcin is sufficient to protect against β-cell failure and glucose intolerance during HFD (Fig. 1).

Interestingly, in the absence of β-cell stress, i.e., during normal chow feeding, the role of sorcin in pancreatic β-cells was less prominent. Nonetheless, forced sorcin expression enhanced GSIS and [Ca²⁺]_cyt oscillations in human islets from normoglycemic donors as well as in islets from young chow-fed mice (Fig. 2 and Supplementary Fig. 4). Importantly, the observed increase in GSIS in our transgenic models was not due to an increase in β-cell mass. Rather, we observed a decrease in the β-cell to α-cell ratio (Fig. 2), suggesting that the sorcin-overexpressing islets display enhanced function, and a possible resistance to HFD-induced hyperplasia (38). The improved GSIS observed in our HFD-fed SRI-tg1/10 mice is most likely secondary to the increases in glucose-induced intracellular Ca²⁺ fluxes (Fig. 3 and Supplementary Fig. 4). Although not tested here, the combined effects of RyR inhibition and SERCA activation by sorcin described in cardiomyocytes (14,15) might thus explain the increased capacity of ER Ca²⁺ stores in SRI-tg1/10 islets (Fig. 3). In the rat heart, sorcin overexpression is associated with an increase in Ca²⁺ transients and enhanced cardiac contractility, rescuing diabetic contractile dysfunction (39). In β-cells, a role for RyR, in particular RyR2, has recently been supported by studies using “leaky” mutants, both in humans and in mice, which display glucose intolerance, decreased insulin secretion, and islet ER stress (40,41).

One intriguing finding of the current study was the inverse relationship between sorcin and G6PC2 expression in islets and β-cells. G6PC2 acts by hydrolyzing glucose-6-phosphate (G6P) in the ER, thus opposing the action of the glucokinase (9,42). Islets from G6pc2^−/− mice display increased cytosolic [Ca²⁺] and enhanced GSIS (9). This suggests that an important mechanism of action of sorcin is to regulate G6PC2 expression to influence calcium homeostasis, GSIS, and ER stress. Interestingly, others have found that glucose cycling and G6Pase activity were markedly enhanced in pancreatic islets of HFD-fed obese hyperglycemic mice, impairing GSIS (43). However, it is possible that G6PC2 also exerts effects beyond glucose cycling and glycolytic flux (44) and earlier studies have linked G6pase activity and G6P levels to cytosolic and ER Ca²⁺ concentrations, both in the liver and in pancreatic β-cells (45,46).

Our study highlights several beneficial effects of sorcin in the β-cell. For example, sorcin-induced increase in intracellular Ca²⁺ activates NFAT signaling, which is fundamental for maintaining the islet β-cell phenotype (47), and whose inhibition is responsible for posttransplantation diabetes caused by calcineurin inhibitors (48). Furthermore, sorcin overexpression under lipotoxic conditions prevented the induction of the ER stress markers CHOP and GRP78/BiP. Interestingly, ATF6 signaling was stimulated by sorcin, a change that, unlike the other two branches of the UPR, i.e., PERK and IRE-1, is not usually associated with apoptosis but with favorable outcomes (36,49). We note that our data may also have direct relevance for β-cell failure in humans. Thus, analysis of unpublished results from the IMIDIA consortium (M. Solimena, A.M. Schulte, L. Marselli, F. Ehehalt, D. Richter, M. Rösler, H. Mziaut, K.-P. Knoch, J.P., M. Bugliani, A. Siddiq, A. Jörns, F. Burdet, R. Liechti, M. Suleiman, D. Margerie, F. Syed, M. Distler, R. Grützmann, E. Petretto, A. Moreno, C. Wegbrod, A. Sönmez, K. Pfriem, A. Friedrich, J. Meinel, C. Wollheim, G. Baretton, R. Scharfmann, E. Nogoceke, E. Bonifacio, D. Sturm, U. Boggi, H.-D. Saeger, F. Filipponi, M. Lesche, P. Meda, A. Dahl, L. Wigger, I. Xenarios, M. Falchi, B.T., J. Weitz, K. Bokvist, S. Lenzen, G.A.R., P. Froguel, M. von Bülow, M.I., P.M., unpublished data) from large sets of human donor islets indicates a significant positive correlation between SRI mRNA levels and GSIS in both diabetic and nondiabetic islets, and a tendency toward lower sorcin levels in islets from patients with type 2 diabetes versus healthy islets. Thus, agents that increase sorcin expression or activity may increase insulin secretion while protecting against β-cell exhaustion.

Acknowledgments. The authors thank Anke Schulte (Sanofi Aventis, Frankfurt, Germany) and Michel Solimena (Technical University, Dresden, Germany) for sharing expression data on sorcin in human islets ahead of publication. The authors thank Pauline Chabosseau (Imperial College London) for help in designing macros for Ca²⁺ imaging and β-cell mass measurements. The authors thank the High-Throughput Genomics Group at the Wellcome Trust Centre for Human Genetics (University of Oxford) for the generation of the gene expression data.

Funding. This study was supported by the European Foundation for the Study of Diabetes (Albert Renold Travel Fellowship/94386 to A.M.), the National Institutes of Health (grants R01-HL120108 and R01-HL055438 to H.H.V.), and Diabetes UK (BDA:12/0004535 to I.L.). G.A.R. was supported by grants from the Wellcome Trust (WT098424AIA), the Medical Research Council (Programme MR/J0003042/1), and the Biotechnology and Biological Sciences Research Council (BB/J015873/1). The work leading to this publication also received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement no. 155005 (IMIDIA) (P.M., C.M., B.T., and G.A.R.), resources of which are composed of a financial contribution from the European Union's Seventh Framework Programme (FP7/2007-2013) and European Federation of Pharmaceutical Industries and Associations companies’ in-kind contribution. G.A.R. is a Royal Society Wolfson Research Merit Award holder.

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

Author Contributions. A.M. designed research studies, conducted experiments, acquired and analyzed data, and wrote the manuscript. J.P. conducted experiments and acquired and analyzed data. X.C., J.H., and N.M. conducted experiments and acquired data. L.C. conducted experiments, acquired data, and wrote the manuscript. P.M., L.P., D.B., P.J., and J.A.M.S. provided human islets. C.C.-G., C.M., and B.T. provided IMIDIA data. M.I. performed a bioinformatics analysis of IMIDIA data. H.H.V. provided Sri^−/− mice and polyclonal anti-SRI antibody. G.A.R. designed research studies, analyzed data, provided reagents, and wrote the manuscript. I.L. designed research studies, conducted experiments, acquired and analyzed data, provided reagents, and wrote the manuscript. I.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.

Prior Presentation. This study was presented at the 2014 Diabetes UK Professional Conference, 5–7 March 2014, Liverpool, U.K.; the 2015 Diabetes UK Professional Conference, 11–13 March 2015, London, U.K.; the 50th Annual Meeting of the European Association for the Study of Diabetes (EASD), 15–19 September 2014, Vienna, Austria; the 51st Annual Meeting of the EASD, 14–18 September 2015, Stockholm, Sweden; and the 31st Congress of the French Society of Endocrinology, 5–8 November 2014, Lyon, France.