The Endoplasmic Reticulum and Calcium Homeostasis in Pancreatic Beta Cells

Abstract

The endoplasmic reticulum (ER) mediates the first steps of protein assembly within the secretory pathway and is the site where protein folding and quality control are initiated. The storage and release of Ca²⁺ are critical physiological functions of the ER. Disrupted ER homeostasis activates the unfolded protein response (UPR), a pathway which attempts to restore cellular equilibrium in the face of ER stress. Unremitting ER stress, and insufficient compensation for it results in beta-cell apoptosis, a process that has been linked to both type 1 diabetes (T1D) and type 2 diabetes (T2D). Both types are characterized by progressive beta-cell failure and a loss of beta-cell mass, although the underlying causes are different. The reduction of mass occurs secondary to apoptosis in the case of T2D, while beta cells undergo autoimmune destruction in T1D. In this review, we examine recent findings that link the UPR pathway and ER Ca²⁺ to beta cell dysfunction. We also discuss how UPR activation in beta cells favors cell survival versus apoptosis and death, and how ER protein chaperones are involved in regulating ER Ca²⁺ levels.

Abbreviations: BiP, Binding immunoglobulin Protein ER; endoplasmic reticulum; ERAD, ER-associated protein degradation; IFN, interferon; IL, interleukin; JNK, c-Jun N-terminal kinase; KHE, proton-K⁺ exchanger; MODY, maturity-onset diabetes of young; PERK, PRKR-like ER kinase; SERCA, Sarco/Endoplasmic Reticulum Ca²⁺-ATPases; T1D, type 1 diabetes; T2D, type 2 diabetes; TNF, tumor necrosis factor; UPR, unfolded protein response; WRS, Wolcott–Rallison syndrome.

Diabetes mellitus is currently a global epidemic, with type 2 diabetes (T2D) accounting for over 90% of diabetes cases worldwide (1). T2D is a chronic metabolic disorder characterized by impaired glucose-stimulated insulin secretion from islet beta cells in the setting of increased insulin resistance (2–4). It is now widely accepted that reduced secretion in T2D reflects a loss of beta-cell function and possibly reduced beta-cell mass, collectively referred to sometimes as functional beta-cell mass (2,3,5,6). Understanding the loss of functional mass is critical to develop better treatments for the disease. Immunological and stress-induced factors, in contrast, mediate the loss of beta cells in type 1 diabetes (T1D) (or autoimmune diabetes) (7,8), and the further understanding of such factors is also critical.

The ER is an intracellular organelle that plays important roles in protein folding and quality control, lipid synthesis, and Ca²⁺ storage and release (9). Disrupted endoplasmic reticulum (ER) homeostasis and ER stress due to environmental factors have been proposed as potential causes of T2D (10–12), and there is increasing evidence that the unfolded protein response (UPR) becomes activated in pancreatic islets isolated from T2D patients or animal models of diabetes (12–14). A loss of ER homeostasis has also been observed in beta cells during the induction of autoimmunity in T1D, and the association between T1D and UPR activation and consequent beta-cell death has been reviewed by several groups (12,15,16). Our understanding of the relationship between ER Ca²⁺, ER stress, and beta-cell death, while the subject of considerable attention in the literature, is, however, incomplete and there are many competing hypotheses as to the exact mechanisms involved. In this review, we will discuss recent findings concerning ER Ca²⁺ regulation and its possible role in beta-cell dysfunction and compromised cell survival in the context of diabetes.

The ER and UPR activation in beta cells

A properly functioning ER is needed to maintain the health and survival of beta cells. Single beta cells synthesize ~1 million insulin molecules/minute (17), about half of the total amount of protein they synthesize (18). The beta-cell thus requires a very well developed and highly functional ER to support the production and secretion of insulin in response to a rise in glucose. A number of studies have suggested that a malfunctioning ER contributes to the development of both T2D and T1D (10–12,15,19–21).

Proteins fail to fold correctly, resulting in UPR activation. In the UPR pathway, three canonical ER membrane transducer proteins are activated once they dissociate from BiP (Binding immunoglobulin Protein, also known as HSP70 or GRP78). BiP is a molecular chaperone that binds to unfolded/misfolded proteins in the ER lumen by attaching to their hydrophobic regions and then releasing the client proteins once they are folded properly (22,23). The 3 transducers that are activated by BiP dissociation are PERK (PRKR-like ER kinase), ATF6 (activating transcription factor 6), and IRE1α (inositol-requiring kinase-1) (23) (Fig. 1). In healthy cells, these transducers are occupied by BiP on the luminal side of the ER membrane. However, when ER stress develops in response to the accumulation of misfolded proteins in the ER lumen, these proteins act as a sink for BiP, resulting in its dissociation from the transducers and, subsequently, UPR activation. To avoid excessive overlap with other excellent reviews (18–20,24–28), a schematic overview of the UPR cascade is provided in Figs. 1 and 2, and we will only briefly summarize the UPR, as well as highlight new findings.

Figure 1.

UPR activation mediates beta-cell survival. BiP is sequestered by binding to ER-stress inducing unfolded/misfolded protein in the ER lumen, leading to its dissociation from PERK, IRE1α, and ATF6, which are ER membrane-localized stress transducers. Activated PERK phosphorylates eIF2α, which inhibits global protein translation and activates ATF4. ATF4 mediates the transcription of genes that encode chaperones, oxidoreductases, ERAD, and autophagy. IRE1α cleaves Xbp1 mRNA, and spliced Xbp1 (sXbp1) mRNA in turn mediates gene transcription of chaperones, lipid synthesis, ERAD, and autophagy. In addition, sXbp1 promotes proinsulin folding through various PDI family genes. When ATF6 is released from BiP, it translocates to the Golgi complex, where ATF6 is cleaved to generate ATF6f. ATF6f activates the transcription of genes encoding protein chaperones and ERAD. ATF6f also triggers beta-cell proliferation. When mild or tolerable levels of UPR are activated in beta cells, the UPR transducers PERK, IRE1α, and ATF6 mediate the transcription of genes for chaperones, ERAD components, and autophagy in an attempt to restore ER homeostasis. UPR activation also triggers mRNA decay and attenuates global gene translation to reduce the workload on the ER. Finally, the induction of AKT1 and reduction of TXNIP serve to inhibit cell death. Elements of the UPR pathway are essential to maintaining beta-cell homeostasis under normal conditions, due to the high secretory workload placed on the beta-cell.

Figure 2.

UPR activation can also mediate beta-cell apoptosis. Prolonged and unresolvable ER stress and concomitant UPR activation can trigger the activation of apoptotic pathways like the ASK1-JNK- and CHOP-mediated apoptosis pathways. ER stress also upregulates caspases and TXNIP, proteins that are involved in apoptosis. IRE1α cleaves microRNA that suppresses caspase 2 translation. IRE1α also couples to TRAF2 and triggers apoptosis. PERK and ATF6 activation upregulates CHOP and leads to apoptosis by upregulating GADD34, Bim, and ERO1-α and by downregulating Trb-3 and Bcl-2. ERO1-α causes ER hyperoxidation that disrupts ERp44-IP3R1 interaction, causing IP3R1 hypersensitivity and increases ER Ca²⁺ release. Bcl-2 suppresses IP3Rs and RyRs-mediated ER Ca²⁺ release. ER Ca²⁺ depletion leads to beta-cell apoptosis through the activation of calpain 2, which activates caspase 12 and JNK.

Upon dissociating from BiP, IRE1α is activated by dimerizing and then becoming autophosphorylated. Activated IRE1α splices Xbp1 (X-box binding protein 1) mRNA, removing a premature stop codon. Spliced Xbp1 (sXbp1) in turn induces the transcription of genes encoding chaperones, components of the ER-associated protein degradation (ERAD) pathway, autophagy, and lipid synthesis (29–31) (Fig. 1). Additionally, sXbp1 enhances proinsulin folding by directly binding to promoter regions of 5 Protein disulfide isomerases (PDIs) family genes encoding PDI, PDIR, P5, ERp44, and ERp46 and regulating their expression (32). IRE1α cleaves other mRNAs as well, including insulin mRNA, to alleviate the ER’s synthetic workload (33,34). Mice harboring a null mutation in 1 of the Xbp1 alleles become insulin resistant, demonstrating a key role of UPR responses in insulin action (35).

The transducer of the Xbp1 arm of the pathway, IRE1α is also complex as it has both kinase and RNase activity (31). Transiently exposing beta cells to high glucose enhances insulin biosynthesis in a manner that is dependent on the kinase activity of IRE1α but independent of BiP dissociation or Xbp1 splicing (36). In contrast, chronic high glucose suppresses insulin biosynthesis and induces ER stress (36), as reduced insulin transcript has been observed in INS-1 cells treated chronically with high glucose. The various possible outcomes that can result from activation of the IRE1α arm of the UPR in beta cells require that a higher order of regulation must also be involved, although this regulation is not well understood.

In terms of the PERK arm of the UPR, after BiP dissociates from it, PERK also undergoes oligomerization and autophosphorylation, as for IRE1α, leading to the phosphorylation of eIF2α (eukaryotic initiation factor 2 α subunit) (37). Phosphorylated eIF2α represses the initiation of global protein translation and activates ATF4 (activating transcription factor 4), which in turn increases the expression of chaperones, oxidoreductases, and genes involved in ERAD and autophagy (27,38–40), as described for sXbp1. PERK-deficient mice suffer a loss of beta cells and develop diabetes in their early weeks of life (41).

As mentioned, insulin resistance is an important characteristic of T2D and normal beta cells compensate for it by increasing their insulin secretory capacity and their cell number if they are genetically endowed to do so. Several mechanisms underlie the expansion of beta-cell mass that is needed in order to cope with augmented insulin demand, including changes in the expression of cell cycle proteins and transcription factors (42). UPR activation is required to hasten beta-cell proliferation that occurs in response to glucose (43). Thus, an elevation of glucose in vivo or in vitro increases beta-cell proliferation, especially in rodent models, while chemical agents that reduce ER stress decrease it (43). In human islets exposed to high glucose or in islets from the db/db mouse, a model of T2D, beta-cell proliferation occurs simultaneously with UPR activation (43). Within the UPR pathway, ATF6, rather than PERK or IRE1, has been shown to contribute to the beta-cell proliferation that occurs in response to increased insulin demand (43) (Fig. 1). In addition to high glucose exposure, or the addition of chemical stressors such as thapsigargin or tunicamycin (see below), other physiological challenges can lead to ER stress in beta cells. For example, hyperlipidemia, a common feature of patients with type 2 diabetes that is linked to insulin resistance, or exposure to saturated fatty acids such as palmitate have been shown to induce ER stress by activating the PERK and IRE1a pathways (44). Palmitate also increases the saturated lipid content of the ER, resulting in ER dilation (a marker of ER stress), trafficking of the ER chaperones GRP70 and PDI from the ER to the cytosol, and the depletion of ER Ca²⁺ (45).

Mutations in proinsulin, the protein precursor of insulin that normally accounts for 30% to 50% of the total protein synthesis of the beta cell (46) can also lead to ER stress. Akita and Munich mice carry a mutation in the Ins2 (insulin 2) gene (C96Y in Akita and C95S in Munich) that disrupts disulfide bond formation leading to misfolded proinsulin (47,48). These mouse models show that misfolded proteins in the ER can sometimes lead to ER stress-induced diabetes in the whole animal (48,49). IAPP (islet amyloid polypeptide), a protein that is normally cosecreted with insulin from beta cells can, under certain conditions, form toxic oligomers in the cell that can also trigger ER stress (50). It has been reported that transgenic mice expressing hIAPP (human islet amyloid polypeptide) fed a high-fat diet become glucose intolerant and insulin resistant (50,51). It has also been shown that hIAPP transgenics exhibit hIAPP aggregation and misfolding that in turn triggers ER stress, presumably by making ER membrane leaky to Ca²⁺ (50,52).

Beta-cell survival and apoptosis

The UPR exists in 2 states—an adaptive state of low ER stress, maintained ER homeostasis and pro-beta-cell survival and an apoptotic state where apoptosis is initiated and cell death ensues in response to pathological conditions (53). To maintain or activate the prosurvival state when the cell faces stressful conditions, IRE1, PERK, and ATF6 become activated in order to generate a “stress” signal that is transmitted from the ER lumen to the cytosol and nucleus. Transmission of this stress signal re-establishes normal protein homeostasis by inhibiting global gene transcription, increasing the degradation of misfolded proteins through ERAD, and increasing the degradation of dysfunctional cellular components through autophagy (Fig. 1). Additionally, an increase in the protein-folding capacity of the cell occurs due to increased chaperone expression.

Prolonged and unmitigated ER stress, in contrast, activates cellular pathways that favor apoptosis (12,13,19,20,25,54,55) (Fig. 2). Apoptosis then decreases beta-cell mass, which increases the stress experienced by the remaining beta cells as they try to compensate for the reduced insulin levels and experience increased secretory demand, resulting in additional beta-cell death.

IRE1α triggers cell death by inducing the degradation of many ER-localized mRNAs (33), as well as cleaving miRNAs that normally repress caspase 2 mRNA translation; this increases caspase 2-mediated apoptosis (56) (Fig. 2). Additionally, IRE1 α couples to TRAF2 (tumor necrosis factor receptor-associated factor 2) localized to the ER membrane, which recruits ASK1 (apoptosis signal-regulating kinase 1). ASK1 phosphorylates and concomitantly activates JNK (c-Jun N-terminal kinase) to stimulate apoptosis (20). Moreover, the IRE1 α -TRAF2 interaction with procaspase 12 promotes its cleavage to active caspase 12, which in turn cleaves caspase 3 to induce apoptosis (20,57,58).

Recent studies also implicate the ABL family of tyrosine kinases in the enhancement of ER stress-mediated apoptosis through the hyperactivation of IRE1α RNase activity (59). TXNIP (thioredoxin-interacting protein) is a newly discovered proapoptotic protein that is part of the ER stress pathway (60). The IRE1α and PERK pathways, via TXNIP, lead to an increase in the production of interleukin (IL)-1β, a highly proinflammatory and cell death-inducing cytokine (61,62). Cytokine-treated INS-1 cells thus have upregulated TXNIP and increased cell death. Interestingly, suppressing TXNIP with the diabetes drug sitagliptin prevents TXNIP-mediated cell death (63).

PERK promotes beta-cell survival by attenuating global translation while upregulating AATF (apoptosis antagonizing transcription factor) through eIF2α phosphorylation. AATF activates AKT1 via STAT3, and AKT1 is reported to suppress apoptosis (64) (Fig. 1). Activated eIF2α also upregulates ATF4, which promotes cell death through CHOP (CCAAT/enhancer-binding protein homologous protein) (20) (Fig. 2). In contrast, ATF6 promotes beta-cell survival by increasing the protein-folding capacity of the cell and the components of ERAD (Fig. 1). For example, BiP is a major target of ATF6. On the other hand, ATF6 favors cell death through CHOP (20) (Fig. 2). It is clear that these signaling pathways are complex in the outcomes they promote require a balancing of competing factors.

CHOP induces apoptosis through several different mechanisms. CHOP activates ERO1-α (ER oxidoreductase 1α) leading to ER hyperoxidation, and disruption of the interaction between the chaperone ERp44 and IP3R1 (inositol 1,4,5-trisphosphate receptor type 1), causing IP3R1 hypersensitivity and concomitantly increased ER Ca²⁺ release (65–67). ER Ca²⁺ depletion can trigger in turn activate calpain 2, leading to beta-cell apoptosis (68) by activating caspase 12 and JNK (69,70). Moreover, CHOP downregulates antiapoptotic Bcl-2 and Trb3 levels and induces proapoptotic Bim to favor apoptosis (24,25,65–67). CHOP upregulates DNA damage-inducible 34 (GADD34), a protein involved in cell growth arrest and DNA damage, leading to apoptosis (24,25,65–67). Studies in model systems have shown that Bcl-2 can directly interact with both IP3Rs and RyRs (ryanodine receptors) to suppress ER Ca²⁺ release (71,72). Bcl-2 is known to preserve the integrity of mitochondrial membranes and has been shown to inhibit the release of proapoptotic cytochrome C from the mitochondrion (73) (Fig. 2).

The UPR and loss of ER homeostasis in T1D and T2D

Recent studies provide additional evidence for UPR dysregulation in both T1D and T2D. Dysregulated UPR exhibiting decreased ATF6 and Xbp1 has been observed in 2 different mouse models of T1D (NOD mice, a model of autoimmune diabetes, and mice expressing the LCMV gene (lymphocytic choriomeningitis virus)). ATF6 and Xbp1 are also decreased in islets taken from human T1D patients (74). The chemical chaperone TUDCA (tauroursodeoxycholate) prevents the loss of ATF6 and Xbp1, reduces the incidence of diabetes and elevates plasma insulin levels and islet insulin content without altering the immune cell populations of the pancreases of T1D mouse models (74).

Proinflammatory cytokines such as tumor necrosis factor (TNF)-α, IL-1α and IFN-α have been implicated in T1D and T2D (75,76) and have been shown to alter Ca²⁺ handling, induce UPR dysregulation, and alter glucose-induced insulin release in beta cells (76,77). The proinflammatory cytokines, IL-1β, TNF-α, and IFN-γ in combination, IL-1β and interferon (IFN)-γ in combination, or IL-1β or IFN-γ presented alone increase PERK/eIF2α phosphorylation, JNK phosphorylation, and CHOP expression in beta cells (78). In one study, IL-1β increased Xbp1 splicing and ATF4 and CHOP expression (79). In contrast, in a separate report, IFN-γ alone decreased Xbp1 splicing and BiP expression, while having no effect on CHOP and ATF4 expression (80). Therefore, cytokine may preferentially affect different arms of the UPR pathway when applied individually.

The loss of ER homeostasis also affects autoantigen production in T1D. Post-translational modifications of proteins such as deamidation and citrullination are known to be induced in diabetes, and the modified peptides that result can become targets of immune cell recognition and activation (81). Ca²⁺-dependent enzymes that mediate such post-translational modifications, such as tTG2 (tissue transglutaminade2) are activated after thapsigargin-mediated ER Ca²⁺ depletion (82). Additionally, the cytokine IL-1α has been shown to convert specific arginine residues of BiP, a diabetes-related autoantigen, into citrulline (83,84). Such mechanisms can link ER Ca²⁺ dysregulation and elevated cytokine levels to the generation of autoimmune epitopes in T1D.

In T2D human pancreas, CHOP nuclear localization was found to be 6 times higher than in nondiabetic human pancreas, but nuclear CHOP was not detected in T1D human pancreas (85). In addition, BiP, ATF4, ATF6, eIF2α phosphorylation, and Xbp1 splicing are all upregulated in islets isolate from db/db diabetic mice (86). In contrast, other studies demonstrate reduced ATF6 and Xbp1 levels in T2D (87,88).

UPR regulation has also been observed in studies on Wolcott–Rallison syndrome (WRS) and maturity-onset diabetes of young (MODY). WRS is a rare autosomal recessive syndrome (89). A deficiency of PERK in humans is the cause of WRS, and it is associated with permanent neonatal diabetes (89). PERK-deficient mice exhibit increased expression of BiP, IRE1α, Xbp1 splicing, and processed ATF6 (90,91). MODY is a rare inherited form of diabetes that runs strongly in families (92). There are at least 14 types of MODY caused by mutations in 1 of 14 genes which disrupt insulin production (92). For example, MODY4 is caused by heterozygous variants in the Pdx1 (pancreatic duodenal homeobox 1) gene (93). Heterozygous mutations in Pdx1 are associated with T2D. Pdx1 heterozygous mice exhibit elevated BiP transcripts (93).

ER Ca²⁺ regulation

Glucose-dependent insulin secretion is a Ca²⁺-dependent process. In beta cells, exposure to elevated glucose triggers closure of the adenosine 5′-ATP triphosphate-gated K⁺ channel to inhibit K⁺ efflux and this results in plasma membrane depolarization. The voltage-gated Ca²⁺ channel then opens and allows Ca²⁺ influx (94,95). The ER plays an important role in intracellular Ca²⁺ signaling and it can help coordinate various intracellular signaling pathways that affect beta-cell function and insulin secretion (9,10,96). Ca²⁺ ions are sequestered within the ER due to activity of the SERCA pump (Sarco/Endoplasmic Reticulum Ca²⁺-ATPases, see below) and are released into the cytosol in response to various physiological triggers. For example, Ca²⁺ ions are released from the ER following the activation of RyRs or IP3Rs by increased cytosolic Ca²⁺, a process referred to as CICR (Ca²⁺-induced Ca²⁺ release) mechanism (97,98). RyRs and IP3Rs are Ca²⁺ channels that are resident in the ER membrane to mediate Ca²⁺ efflux from the organelle. Of course, a rise in cytosolic Ca²⁺ resulting from either Ca²⁺ influx or ER release results in the triggered fusion and exocytosis of insulin-containing granules to release insulin into the extracellular space. It is crucial for the cell to maintain a concerted regulation of ER Ca²⁺ uptake and release in order to preserve normal Ca²⁺ signaling. As noted above, unresolved ER stress can result in the depletion of Ca²⁺ from the ER and subsequent apoptosis (Fig. 2). Conversely, there are UPR-independent routes to decrease ER free Ca²⁺ under different pathological conditions, which in turn can result in a loss of ER homeostasis and cell death (99–101).

ER Ca²⁺ is regulated by three classes of handling proteins: (1) SERCAs, which are pumps that transport free Ca²⁺ from the cytosol to the ER lumen in an ATP-requiring manner; (2) ER Ca²⁺-binding proteins, which can bind significant amounts of Ca²⁺ in the ER lumen, and (3) ER Ca²⁺ channels, the RyRs, the IP3Rs and the translocon, which all can potentially release Ca²⁺ into the cytosol (Fig. 3a). Dysfunction manifested in any of the 3 classes of ER proteins can disrupt ER homeostasis.

Figure 3.

ER Ca²⁺ regulation in beta cells in the presence and absence of ER stress. ER Ca²⁺ is regulated by SERCAs, RyRs, IP3Rs, and possibly the translocon. SERCA pumping of Ca²⁺ to the ER is concomitant with a proton ejection and a KHE operates to reclaim these ejected protons. BiP is required for IP3R1 tetrameric assembly and Ca²⁺ release from the ER. ERp44 competes with BiP binding to IP3R1 at the same site and inhibits IP3R1-mediated ER Ca²⁺ release. To maintain ER Ca²⁺ homeostasis, 2 members of the twin-pore K⁺ channel family, TALK-1 and TASK-1, come together to form a K⁺ channel mediating K⁺ influx that can help promote ER Ca²⁺ release. In response to ER stress conditions, BiP is sequestered by misfolded proteins, ERp44 becomes upregulated and competes with BiP binding to IP3R1 which results in decreased IP₃-induced Ca²⁺ release.

The genes encoding the SERCAs (eg, SERCA1, 2, and 3) generate more than 14 different isoforms by alternative splicing (102). SERCA 2b (110 kDa) is the predominant isoform expressed in beta cells (103), and is downregulated in islets from the db/db mouse, cytokine-treated beta cells, and islets from T2D donors (79,104) (Fig. 3b). Interestingly, this loss was alleviated by the application of pioglitazone, an agonist of PPAR-g (peroxisome proliferator-activated receptor) that regulates SERCA2b transcription (104). Increasing SERCA2b expression improves insulin secretion and reduces apoptosis in response to inflammatory cytokines or high glucose in INS-1 cells (104). Other studies also support the hypothesis that altered SERCA expression can contribute to diabetes (105). In addition, islets isolated from whole-body SERCA2 haplo-insufficient mice exhibited lower ER Ca²⁺ levels, reduced insulin secretion, and increased ER stress and beta-cell death in mice that were fed a high-fat diet (103).

PERK and calcineurin have been shown to be important to insulin secretion and glucose homeostasis through regulating SERCA activity and ER Ca²⁺ uptake. In addition to regulating the UPR, PERK is also essential for normal beta-cell function. Similar to human WRS, PERK-deficient mice rapidly become hyperglycemic and exhibit reduced insulin content and beta-cells apoptosis beginning in the fourth postnatal week (41). Calcineurin (protein phosphatase 3), a Ca²⁺ and calmodulin-dependent protein phosphatase, regulates human beta-cell survival, as its inhibition induces apoptosis (106). Cavener and his group members have shown that acute inhibition of PERK or calcineurin impairs glucose-stimulated insulin secretion and Ca²⁺ influx through SOCE (store-operated Ca²⁺ entry, see below) in beta cells (107). PERK inhibition decreases ER Ca²⁺ reuptake via SERCA by reducing the interaction of SERCA and calnexin, which binds to SERCA and suppresses the SERCA activity. Calcineurin interacts with PERK and dephosphorylates calnexin. Therefore, PERK and calcineurin act together to regulate SERCA activity (107).

Ca²⁺ uptake and sequestration into the ER can filter Ca²⁺ signaling in beta cells

ER Ca²⁺ uptake in beta cells during glucose-induced Ca²⁺ oscillations filters and shapes the waveforms of the cytosolic Ca²⁺ oscillations by introducing a slow time constant (3). Thus, as Ca²⁺ initially rises in the cytosol, its slower uptake into the ER via SERCA alters the dynamics of the Ca²⁺ rise (reduced SERCA activity increases the rate of rise of cytosolic Ca²⁺ in response to Ca²⁺ influx into the cell) and seen in the cytosol. In addition, once the ER is filled with Ca²⁺ by the end of the oscillation active phase, the ER will begin to passively release Ca²⁺, giving rise to a secondary (and slower) phase of decay at the end of the oscillation. This filtering of cytosolic Ca²⁺ transients has been modeled by Bertam and Sherman, who showed that it can have a profound effect on the time courses of Ca²⁺-dependent processes in beta cells (108). Experimental studies of the ER Ca²⁺ uptake and release processes failed to establish the identity of the ion channel(s) or pathway(s) mediating slow ER Ca²⁺ leak (109). However, it was clearly not mediated by either IP3Rs or RyRs, as it was insensitive to drugs known to block these channels. As unconventional ER Ca²⁺ channels such as Sec61 (see below) may fulfill this role, more investigations seem warranted.

The ER contains approximately 2 mM total Ca²⁺, and only a fraction of this is free. The majority of ER Ca²⁺ is thought to be bound to various Ca²⁺ binding proteins including, calreticulin, BiP, PDI, and GP94 (110). Calreticulin has a highly acidic C-terminal domain that contains multiple low affinity Ca²⁺ binding sites (111,112). These sites bind Ca²⁺ with a K_D in the ER Ca²⁺ range (113,114), and a number of studies indicate that the overexpression of calreticulin increases the Ca²⁺ storage capacity of the ER (115,116). UPR activation induced by high glucose or high-fat diet treatment in mice shows increased expression of calreticulin (43,117,118). In addition to increasing ER Ca²⁺ storage, calreticulin is a chaperone for nascent ER glycoproteins and contains monoglucosylated glycans, the key structural elements of glycoproteins that are recognized by calreticulin. Via this function, calreticulin could contribute to the enhanced maintenance of ER homeostasis in a stressed ER (119).

BiP is suggested to play a role in ER Ca²⁺ storage in some studies (120), whereas other studies support the hypothesis that Ca²⁺ binding by BiP primarily regulates its interactions with nucleotides and other factors (121). Structural studies of truncated BiP constructs indicate the presence of 2 Ca²⁺ binding sites in the vicinity of the nucleotide binding/catalytic site of BiP (122). Further studies are required to understand whether BiP contains low affinity Ca²⁺ binding sites that contribute to ER Ca²⁺ storage, and, if so, where such sites are located.

While a number of reviews indicate that reduced ER Ca²⁺ levels impair the conformation and function of ER Ca²⁺-binding chaperones, there have been relatively few studies of this. In fact, in 1 study, decreasing the Ca²⁺ concentration in the 1 to 0.1 mM range increased the ATPase activity of recombinant BiP (123). Additionally, the binding of calreticulin to monoglucosylated glycans was shown to be independent of Ca²⁺ concentration (124). On the other hand, low affinity Ca²⁺ binding influences the structure of calreticulin (114,125) and mediates its phospholipid binding (126). Moreover, the ER retention of several Ca²⁺ binding chaperones at least partly depends on the maintenance of normal ER Ca²⁺ levels (127–131) and/or the presence of their Ca²⁺ binding domains (132). Thus, by interfering with their proper localization, ER Ca²⁺ depletion is predicted to interfere with the proper function of Ca²⁺ binding chaperones.

ER Ca²⁺ efflux channels: RyRs, IP3Rs, and the translocon

Ca²⁺ exits the ER down its concentration gradient via RyRs or IP3Rs, or the translocon as mentioned previously. Ca²⁺ release through RyRs is primarily triggered by increased cytosolic Ca²⁺ through the process of CICR. The advantage of CICR is that it can occur rapidly and regeneratively, as Ca²⁺ released into the cytosol then facilitates the further release of more Ca²⁺, a positive feedback process (97,133). RyRs are the largest known ion channel proteins. There are 3 RyR isoforms, each having a molecular mass of 565 kDa (134,135). RyR1 and RyR2 are primarily expressed in skeletal and cardiac muscle, respectively, and RyR3 is expressed at rather low levels in various other tissues (134). In beta cells, RyR2 is the predominant isoform, but RyR1 is also expressed, albeit at a comparatively low level (136). RyR3 appears to be minimally expressed in beta cells (136). However, the existence of RyRs and their significance for beta-cell function has been the subject of considerable controversy given their relatively low level of expression (137) and whether or not there is an important role for cyclic ADP ribose as a stimulator of Ca²⁺ release via RyRs (138,139).

RyR activity can be increased by ATP, CaMKII (Ca²⁺/CaM-dependent protein kinase II) (135,140)) or PKA (135,140,141). The drug ryanodine, a classic blocker of RyRs has a complicated pharmacology as it has both agonist and antagonist actions on RyRs depending on the dose used (142). Thus, RyRs are activated by ryanodine at nanomolar concentrations, where ryanodine binding to the RyR keeps it in an open state that allows ER Ca²⁺ release into the cytosol. In contrast, at >100 µM, ryanodine inhibits RyRs (140). Another well-known RyR blocker is the muscle relaxant dantrolene (143). Dantrolene selectively blocks RyR1 and RyR3 but not RyR2 (143). Evans-Molina and colleagues have proposed that tunicamycin-induced ER stress increases RyR activity and ER Ca²⁺ release without altering RyR2 expression and results in increased spontaneous cytosolic Ca²⁺ transients in response to elevated extracellular Ca²⁺ levels in INS-1 cells and dispersed cadaveric human islets. Islets isolated from Akita mice exhibit cytosolic Ca²⁺ oscillations stimulated by 5 mM glucose, whereas wildtype littermate mice do not have the same sensitivity (136). Ryanodine suppresses the glucose-stimulated Ca²⁺ oscillations in Akita mice and decreases tunicamycin-induced spontaneous Ca²⁺ transients in INS-1 cells and human islets. They also suggest RyR2 as the isoform gains function in response to ER stress. In addition to the fact that RyR2 is the predominant isoform of RyR, knock-in mice with a mutation of chronic activation of RyR2 mimicking constitutive CaMKII phosphorylation develops glucose intolerance and exhibits diminished Ca²⁺ responses and impaired glucose-induced insulin secretion (136,144). However, in this study dantrolene (at 1 μM) failed to restore ER Ca²⁺ when applied to tunicamycin-treated INS-1 cells. In a separate study, 30 μM dantrolene decreased thapsigargin-triggered apoptosis in MIN6 cells, an insulin-secreting mouse cell line (145). These findings indicate that ER stress can cause complex and cell type-dependent modulation of ER Ca²⁺ efflux channels.

Calstabin (a Ca²⁺ channel stabilizing binding protein) binds to RyRs and prevents ER Ca²⁺ leak by stabilizing the closed states of the RyRs (134). Calstabin 2 is a subunit of the RyR2 macromolecular complex and its dissociation from RyR2 causes increased ER Ca²⁺ leak (134). Human patients and mice harboring mutations that alter the binding of calstabin2 to RyR2 (eg, catecholaminergic polymorphic ventricular tachycardia) are glucose intolerant and have decreased glucose-stimulated insulin secretion due to increased ER Ca²⁺ leakage. Islets from T2D donors or diabetic mouse islets have oxidation/nitrosylation of RyR2 and a depletion of calstabin2. Inhibiting the dissociation of calstabin2 from RyR2 using the drug Rycal S107 reverses glucose intolerance and improves insulin secretion in both human and murine islets, indicating that the proper functioning of RyR2 channels is critically important for beta cells (134).

IP3Rs contain an IP3 binding core domain near their N-termini. Upon binding IP3, the binding core rearranges and then dissociates from its associated suppressor domain, resulting in IP3R activation. IP3Rs become less sensitive to IP3 when the ER Ca²⁺ level is low in order to protect the ER from becoming too Ca²⁺ depleted. RyRs and IP3Rs share 40% homology (135), and as for RyRs, IP3Rs are also regulated by cytosolic Ca²⁺, ATP, and protein phosphorylation, and they also interact with other proteins (135,146–148). Three IP3R isoforms (IP3R1, IP3R2, and IP3R3) exist and each has a molecular weight of about 300 kDa (135,149). IP3R2 is most highly expressed in cardiac and skeletal muscle, as well as liver and kidney (150), while IP3R3 is expressed in both endocrine and exocrine pancreas, as well as other tissues (150). While genetic variation within IP3R3 has been linked to T1D (151), IP3R1 is the most abundant isoform in beta cells (152).

Many proteins interact with IP3R1. As noted above (Fig. 2), the chaperone ERp44 inhibits IP3R1 when ER Ca²⁺ levels are low (67), and BiP helps assemble IP3R1 into tetramers and is required for Ca²⁺ release by the receptor (67) (Fig. 3a). ERp44 competes with BiP for the same site on IP3R1 (67). In the face of ER stress, ERp44 upregulation causes BiP to dissociate from IP3R1, leading to decreased IP₃-induced Ca²⁺ release (Fig. 3b).

In contrast to RyRs and IP3Rs, the translocon is a Ca²⁺ “leak channel” (153,154), but it is not regulated by cytosolic Ca²⁺ or by G protein-coupled receptor (GPCR) activation. Sec61, a core component of the translocon, is located on the ER membrane (153), along with other protein components. The Sec61 complex forms a protein-conducting channel to translocate nascent polypeptide chains from the cytosol to the ER during translation. The translocon complex has also been proposed to mediate ER Ca²⁺ leak (154). Mice with Sec61 gene point mutation fed on high-fat diet are hyperglycemic and hypoinsulinemic compared with their wildtype littermates. After feeding these mice high-fat diet for a week, the islets isolated from these mice show upregulation of BiP and CHOP as well as an increase in apoptosis (155).

ER membrane localized K⁺ channels

The ER Ca²⁺ concentration is greater (eg, 300–700 αM) than that of the cytosol (eg, 50–100 nM) and the activity of ATP-fueled SERCA pumps is largely responsible for maintaining this ER Ca²⁺ gradient in the absence of an ER potential difference or a K⁺ concentration gradient (156). SERCA pumping of Ca²⁺ from the cytosol to the ER is associated with a proton ejection and a proton-K⁺ exchanger (KHE) operates to reclaim these ejected protons (156–159). Various K⁺ channels in the ER can in turn mediate K⁺ reuptake by the ER.

TALK-1 and TASK-1 are 2 ER-located K⁺ channels (Fig. 3) that likely help mediate these ER K⁺ fluxes (160). Mouse beta cells lacking TALK-1 have been reported to have elevated ER Ca²⁺ and reduced ER Ca²⁺ release (160,161). Therefore, TALK-1 modulates ER Ca²⁺ homeostasis. TALK-1-deficient mice have decreased expression of BiP after a prolonged high-fat diet, which implicates that maintaining ER Ca²⁺ may prevent UPR activation in islets (161). TALK-1-deficient mice also have decreased mRNA encoding SERCA 2b and 3 after a prolonged high-fat diet, which perhaps suggests they may prevent ER Ca²⁺ overload (161). Much remains to be understood about changes to TALK-1 and TASK-1 expression and their functional roles under ER stress conditions in beta cells.

Store-operated Ca²⁺ entry

A reduction in ER Ca²⁺ is sensed by an ER Ca²⁺ sensor called STIM1 (STromal Interaction Molecule 1) which multimerizes and then redistributes to the ER–plasma membrane junction. Here, STIM1 binds to and subsequently opens ORAI1 ion channels which are Ca²⁺ permeable, triggering Ca²⁺ entry into the cytoplasm as part of a process called SOCE (162). This mechanism was originally proposed by Putney and colleagues based on data from inexcitable cells, and they called the mechanism “capacitative Ca²⁺ entry”. This is in reference to the fact that the release of Ca²⁺ from the ER and subsequent ER Ca²⁺ depletion concomitantly increased the uptake of extracellular Ca²⁺ and ER Ca²⁺ repletion (163), much like a capacitor that can store charge but then discharge it. Recent studies show that treating human and mouse islets with a cocktail of TNF-α, IL-1α, and IFN-α or glucolipotoxicity-inducing agents reduce levels of STIM1 protein (77). STIM1 expression is also reduced in STZ (streptozotocin)-treated mice and islets from T2D donors. Restoration of STIM1 expression in islets from T2D donors was found to improve insulin secretion (77).

Mutations in the components comprising SOCE have been linked to several major diseases, including severe combined immunodeficiency (164). Precise regulation of SOCE is critical for maintaining normal glucose-stimulated insulin secretion in beta cells (165). In these cells, SOCE also helps maintain ER Ca²⁺ and provides an extra Ca²⁺ influx mechanism to help depolarize beta cells in response to GPCR agonists such as GLP-1 and acetylcholine (166). In pancreatic alpha cells, Gylfe and colleagues have proposed that SOCE plays an essential role in the stimulus secretion coupling mechanism that links low glucose to increased glucagon secretion (167), although this remains controversial (168).

Conclusions and Future Perspectives

Overall, there has been remarkable progress in our understanding of how the UPR and Ca²⁺ handling become dysregulated in diabetes. Genetic mutations, and membrane lipid and glucose-induced alterations as well as inflammatory mediators influence the induction of the UPR, as well as the expression or activity of key regulators of ER Ca²⁺. In turn, a loss of ER homeostasis has important consequences for insulin secretion and glucose intolerance in T2D and for the generation of autoimmune epitopes in T1D. However, many gaps remain in our understanding of the detailed molecular mechanisms involved in the loss of ER homeostasis and also Ca²⁺ dysregulation in diabetes, which may be important for identifying new drug targets. Further studies thus are needed to apply our current level of knowledge to design of new therapeutics.

Additional Information

Disclosure Summary: The authors have nothing to disclose.