ER Stress in Pancreatic β Cells: The Thin Red Line Between Adaptation and Failure

See allHide authors and affiliations

Science Signaling  23 Feb 2010:
Vol. 3, Issue 110, pp. pe7
DOI: 10.1126/scisignal.3110pe7


Secretory cells, such as pancreatic β cells, face the challenge of increasing protein synthesis severalfold during acute or chronic stimulation. This poses a burden on the endoplasmic reticulum (ER), the organelle where proinsulin synthesis and folding takes place. Thus, β cells use various adaptive mechanisms to adjust the functional capacity of the ER to the prevailing demand. These check-and-balance mechanisms are collectively known as the unfolded protein response (UPR). It remains unclear how UPR signaling is ultimately regulated and what delineates the boundaries between a physiological and a pathological response. New discoveries point to the divergent effects of acute and chronic metabolic fluxes and chemical ER stressors on the formation of complexes among UPR transducers, scaffold proteins, and phosphatases. These and other findings provide a first glimpse on how different signals trigger diverging UPR outcomes.

Samuel Beckett once remarked: “Ever tried. Ever failed. No matter. Try again. Fail again. Fail better.” Secretory cells, however, have a limited margin for failure. Take, for instance, the case of pancreatic β cells, the source of the hormone insulin. After stimulation by high glucose concentrations, β cells increase by more than 10-fold the synthesis of proinsulin, with hormone synthesis approaching 50% of the total protein production (1). Translation of proinsulin and other secretory proteins occurs on ribosomes located on the cytosolic surface of the endoplasmic reticulum (ER) (2). The newly synthesized proinsulin is directed into the ER, where it forms disulfide bonds and folds into its correct three-dimensional structure (2). This nutrient-induced protein synthesis poses both acute and long-term burdens on the ER of the β cell. First, β cells need to respond to varying demands for insulin biosynthesis and release, as determined by food ingestion. Second, human pancreatic β cells are very long-lived cells: Recent data suggest that the β cell population is largely established in childhood (3). Thus, β cells in obese individuals will need to compensate for the increased requirement for insulin production over decades (obesity is a major cause of insulin resistance). This has forced β cells, as well as other secretory cells, to develop checks and balances to adapt ER function to short- and long-term prevailing demands. These adaptive responses are collectively known as the unfolded protein response (UPR) and are triggered by ER-located transmembrane sensors with luminal domains that detect ER stress in the form of unfolded proteins in the ER and cytosolic domains that convey the information to nuclear and cytosolic effectors (4). The UPR is mediated by three main pathways, namely the inositol-requiring enzyme 1α (IRE1α), eIF2α (eukaryotic initiation factor 2α)−PERK (RNA-activated protein kinase–like ER-localized eIF2α kinase)–ATF4 (activating transcription factor 4), and ATF6 signaling networks (4, 5). Activation of these pathways can decrease the arrival of newly synthesized proteins in the ER, increase the folding capacity of the organelle by increasing synthesis of ER chaperones, and augment the extrusion of irreversibly misfolded proteins. In the event that these steps fail to restore ER homeostasis, β cells will undergo apoptosis through activation of the mitochondrial pathway of cell death (5, 6). How and why this transition from “physiological” to “pathological” UPR takes place in β cells remains to be determined. This is an important question, because accumulating evidence suggest that ER stress plays a role in β cell demise in type 2 and perhaps also type 1 diabetes and may contribute to insulin resistance (5). A paper by Qiu and co-workers (7) provides new and interesting insights into the differential regulation of IRE1α, a key transducer of the UPR, during early β cell adaptation to high glucose concentrations or in the course of severe ER stress.

IRE1p was the first transmembrane ER stress transducer to be discovered in yeast (8, 9), and its crystal structure was recently clarified (10, 11). The ubiquitously expressed mammalian homolog of IRE1p is IRE1α, a protein with kinase and endonuclease activities (4). The only known phosphorylation substrate of IRE1α is IRE1α itself, and autophosphorylation of IRE1α triggers its effector functions. One such function is the activation of the transcription factor X-box binding protein 1 (XBP-1) through alternative splicing. XBP-1spliced (XBP-1s) acts as a transcriptional transactivator of genes that encode factors involved in ER expansion, protein maturation, and degradation of misfolded proteins (1214), and it may also contribute to the initiation of apoptosis (15). Another effector function of IRE1α is the cleavage of mRNAs targeted to the ER to alleviate the translation load on the ER. This phenomenon was originally discovered in Drosophila cells (16) but also operates in pancreatic β cells, where activated IRE1α contributes to Insulin mRNA degradation (17, 18). Yet another effector function of IRE1α is to enhance proinsulin biosynthesis after acute exposure (on the order of 1 to 3 hours) to high glucose concentrations, a phenomenon that is dissociated from Xbp1 splicing (19). In contrast, long-term exposure (on the order of 1 to 3 days) of rodent β cells to high glucose concentrations causes ER stress, excess IRE1α activation, Xbp1 splicing, and reduced Insulin mRNA abundance (18). But how is IRE1α signaling regulated? Furthermore, how can a secretory cell determine whether IRE1α activation indicates that the cell should adapt, mount a stress response, or even undergo apoptosis?

To address these questions, Qiu et al. (7) searched for IRE1α-interacting partners that may regulate its signaling. For this purpose, they performed a yeast two-hybrid screen by using the cytoplasmic portion of human IRE1α as bait and isolated the scaffold protein RACK1 (receptor for activated C-kinase 1). RACK1 binds to membrane receptors and protein kinases, integrating signals originating from diverse cellular processes (20). In a series of elegant experiments, they showed that acute exposure (3 hours) of β cells to high glucose concentrations increased the association of IRE1α with RACK1, which was followed by increased phosphorylation of Ser724 at the activation site of IRE1α. This effect depended on glucose metabolism and the consequent rise in cytosolic calcium concentrations and was not accompanied by activation of other markers of ER stress. Deletion studies then identified the linker region of IRE1α as the site for RACK1 interaction and indicated that IRE1α dimerization is required for RACK1 binding. RACK1 is constitutively associated with protein phosphatase 2A (PP2A), and the formation of the glucose-induced ternary complex between IRE1α-RACK1-PP2A provides negative feedback on glucose stimulation of IRE1α through dephosphorylation of IRE1α by PP2A. This would be expected to attenuate IRE1α-dependent up-regulation of proinsulin biosynthesis, preventing excessive production of the hormone. A similar regulatory pattern has been described for another branch of the UPR, namely the phosphorylation state of eIF2α, which controls protein translation (4, 21). Here, early proinsulin biosynthesis is stimulated by a rapid (15 min) glucose-induced dephosphorylation of eIF2α by protein phosphatase 1 (PP1) (22). It is noteworthy that prolonged eIF2α phosphorylation (lasting over 12 hours), as induced by the chemical agent salubrinal, causes β cell dysfunction and eventually apoptosis (23), emphasizing the need for tight and temporally adequate control of the UPR pathways in β cells.

A different IRE1α-RACK1-PP2A response was observed when β cells were exposed to the chemical ER stressors thapsigargin or tunicamycin, to high glucose concentrations for a prolonged period (72 hours), or to the saturated free fatty acid palmitate, which causes ER stress at least in part by depletion of ER calcium stores (24, 25). Under these conditions, several markers of ER stress were induced, including Xbp1 splicing, IRE1α phosphorylation, and increased interaction between IRE1α and RACK1 (7). In contrast to short-term exposure to high glucose concentrations, PP2A dissociated from RACK1, resulting in prolonged and increased phosphorylation of IRE1α in the RACK1 complex. This IRE1α activation was paralleled by decreased Insulin mRNA abundance, but it is unclear whether there is a cause-and-effect relation between these two phenomena. Recent findings suggest that phosphotransfer activation of IRE1α is required to trigger the degradation of ER-targeted mRNAs and that activated IRE1α transmits proapoptotic signals that are independent of Xbp1 splicing (26). It will be important to evaluate whether and how RACK1 and PP2A contribute to the selective induction of these alternate modes of IRE1α activation.

It might be expected that RACK1 dissociation from PP2A during ER stress would actually favor proinsulin synthesis, in a mirror image of what happens during acute glucose exposure. This was not the case, perhaps because IRE1α-mediated degradation of Insulin mRNA and ER stress–induced inhibition of translation prevailed. Palmitate triggered a more complex response, with induction of several ER stress markers, including phosphorylation of eIF2α and induction of the transcription factor CHOP (C/EBP homologous protein), but not phosphorylation of IRE1α (7). Acute exposure of β cells to palmitate increases insulin secretion and current through L-type calcium channels (27), whereas a more prolonged palmitate treatment (over 6 hours) induces Xbp1 splicing and inhibits insulin release (5). This suggests that glucose may lead to formation of the IRE1α-RACK1-PP2A complex by additional calcium-independent mechanisms, which remain to be clarified. In this context, and in order to definitively assign a specific action to glucose, it will be important to test the effects of other nutrient secretagogues, such as the amino acids leucine and glutamine and the unsaturated fatty acid oleate, on the IRE1α-RACK1-PP2A complex. Of note, the above described engagement of diverse branches of the UPR following acute or chronic metabolic flux, or palmitate- or chemical-induced ER stress, reminds us that we are probably dealing with qualitatively distinct UPR responses, specific to given metabolic and ER conditions prevailing in the cell, instead of just up- or down-scaled versions of the same UPR (28).

The discovery of the IRE1α-RACK1-PP2A protein interaction (7) adds to the list of proteins known to directly interact with IRE1α (29). In addition to a series of adaptor molecules mediating downstream IRE1α effects, other modulator proteins regulating IRE1α activation have been described. Among these, only BAX inhibitor-1 is known so far to attenuate IRE1α signaling by inhibiting the endoribonuclease activity of the transducer (30, 31).

In pancreatic islets isolated from the leptin receptor–deficient db/db mice early in the progression to diabetes, RACK1 mRNA and protein abundance was decreased, and this was accompanied by enhanced IRE1α phosphorylation, induction of the ER stress markers BiP and Xbp1, and increased islet insulin content (7). The reasons for this in vivo decrease in RACK1 abundance remain to be clarified, but data obtained in another mouse model of chronic β cell ER stress, the Akita mouse, suggest that some key mRNAs encoding proteins involved in ER-related activities start to undergo active degradation (26), which may herald progressive β cell dysfunction and apoptosis. To answer this question, it will be necessary to perform detailed time course analyses in β cells from this and other rodent models of type 2 diabetes to determine whether RACK1 depletion actually precedes degradation of mRNAs encoding BiP and other key ER proteins.

How can we integrate the present findings with the available knowledge on the regulation of the UPR in β cells during adaptation to ER stress? It is conceivable that short-term exposure to high glucose concentrations causes a mild UPR (Fig. 1A), at least in part because PP1 action in another branch of the UPR (that mediated by PERK) and RACK1-PP2A inhibition of IRE1α constrain signal transduction. Under chronic or severe ER stress, the dissociation of PP2A from RACK1 relieves the brake on IRE1α phosphorylation and contributes to full-fledged IRE1α activation, with activation of XBP-1 and c-Jun N-terminal kinase (JNK) and degradation of the mRNA encoding proinsulin. PERK-induced phosphorylation of eIF2α and the above-mentioned IRE1α-dependent signals inhibit proinsulin biosynthesis and impair β cell function. If ER stress persists, JNK and other signals downstream of IRE1α, together with the transcriptional activation of CHOP, activate the proapoptotic Bcl-2 family member death protein 5 (DP5) (6) and other mitochondrial proteins to trigger β cell apoptosis (5) (Fig. 1B).

Fig. 1

The transition from mild physiologic UPR signaling in pancreatic β cells to a severe ER stress response that culminates in β cell failure. (A) The acute metabolic flux triggered by short-term exposure to high glucose concentrations increases proinsulin biosynthesis in the ER and mild UPR signaling. Signaling in the PERK branch of the ER stress response is restrained by glucose metabolism and calcium influx–dependent activation of PP1. Opposing the effect of the kinase PERK, the phosphatase PP1 dephosphorylates eIF2α and promotes protein translation. In the IRE1α branch, the glucose metabolism and calcium influx-dependent formation of the IRE1α-RACK1-PP2A complex limits IRE1α phosphorylation and attenuates proinsulin biosynthesis. (B) Chronic or severe ER stress can be induced in β cells by chronic glucolipotoxicity, decreased ER calcium content, proinflammatory cytokines, mutations in genes encoding insulin or UPR transducers, or decreased chaperone function. Under these conditions, the dissociation of PP2A from RACK1 contributes to sustained IRE1α phosphorylation and activation, leading to Xbp1 splicing, degradation of ER-associated mRNAs (including that encoding proinsulin), and activation of JNK. These signals, together with the decreased translation initiation because of eIF2α phosphorylation, inhibit proinsulin biosynthesis in the ER and cause β cell dysfunction. IRE1α-mediated activation of JNK and of other signals that remain to be defined, and the induction of the proapoptotic transcription factor CHOP downstream of PERK, activate DP5 and other mitochondrial proteins to trigger the apoptosis program if ER stress persists.

The study by Qiu and colleagues (7) and previous studies (18, 19, 22, 25, 26, 32) indicate temporal differences between the early stimulatory effects of glucose on some branches of the UPR and later effects, which are accompanied by severe ER stress. This would suggest the inability of β cells to cope with prolonged stimulation (on the order of 3 to 7 days). How can this be reconciled with the fact that β cells from obese individuals compensate for insulin resistance for decades, in most cases without failing? The present (7) and previous (18, 19, 26) data are mostly based on insulin-producing cell lines, which are susceptible to ER stress induced by high glucose concentrations. On the other hand, human islets or primary rat β cells exposed for several days to 28 mM glucose (a high glucose concentration) maintain increased insulin production with minimal signs of ER stress but show severe ER stress in the presence of palmitate (25). This suggests that primary β cell ER stress is not simply the consequence of β cell “overwork” but rather that excessive demand must be accompanied by specific impairments of ER function, for instance as induced by saturated fatty acids that both increase β cell demand and deplete ER calcium stores (25, 33).

The discovery of the differential role for RACK1 and PP2A on the early “functional” and late “pathological” effects of IRE1α (7) opens new avenues for research in the field. The challenge ahead of us is to further understand the different players and levels of checks and balances implicated in the dynamic control of the diverse branches of the UPR and then to integrate this knowledge into a coherent model that explains the “thin red line” between β cell adaptation to demand and failure. We are not there yet, but there are good clues on where to start the search.


Supported by grants from the European Union (projects Eurodia and Naimit in framework programs 6 and 7 of the European Community), the FNRS (Fonds National de la Recherche Scientifique), ARC (Actions de Recherche Concerteé de la Communauté Française), Belgium, and the Belgium Program on Interuniversity Poles of Attraction initiated by the Belgian State (IUAP P5/17 and P6/40).

References and Notes

View Abstract

Stay Connected to Science Signaling

Navigate This Article