Endoplasmic Reticulum Protein Quality Control in β cells

Abstract

Type 1 and type 2 diabetes are associated with loss of β cell function. Optimal β cell function is linked to protein homeostasis in the endoplasmic reticulum (ER). Here, we review the roles of ER protein quality-control mechanisms, including the unfolded protein response (UPR), autophagy (specifically ER-phagy) and ER-associated degradation (ERAD), in β cells. We propose that different quality control mechanisms may control different aspects of β cell biology (i.e. function, survival and identity), thereby contributing to disease pathogenesis.

1. INTRODUCTION

Diabetes mellitus encompasses a spectrum of diseases with dysregulated secretion of insulin and glucagon from β and α cells of the endocrine pancreas, respectively, which result in hyperglycemia. Studies of rare monogenic diabetes such as mutant INS-gene induced diabetes of youth (MIDY), Wolfram syndrome, and Wolcott-Rallison syndrome, have revealed disrupted endoplasmic reticulum (ER) homeostasis in the pathogenesis of β cell dysfunction [1, 2]. The ER is responsible for synthesis, folding, and maturation of secretory and transmembrane proteins. In pancreatic β cells, insulin is initially synthesized as preproinsulin, which undergoes cotranslational translocation across the ER membrane. Following signal peptide cleavage, proinsulin folds to form three disulfide bonds between its A and B chains (B7:A7, B19:A20, and A6:A11). Once folded, proinsulin molecules are transported through the Golgi complex, undergo proteolysis by prohormone convertase(s), and are stored in secretory granules as mature insulin [3].

Insulin release in response to physiologic stimuli such as hyperglycemia occurs through a sequence of events termed stimulus-secretion coupling. Glucose enters β cells through the low-affinity glucose transporter GLUT2 and subsequently is metabolized to generate ATP. The increased intracellular ATP/ADP ratio promotes closure of ATP-sensitive K+ channels, leading to plasma membrane depolarization, calcium influx through voltage-dependent calcium channels, and subsequent fusion of insulin granules with the membrane. Additionally, amino acids, lipids, hormones and neurotransmitters each influence insulin secretion [4–6] (Figure 1).

Figure 1. Stimulus-coupled secretion in β cells.

Glucose is the primary stimulus for insulin secretion, though other macronutrients, hormones, and nervous system inputs (as detected by membrane receptors) modulate the β cell response. Glucose enters β cells via glucose transporter GLUT2 and is metabolized. The increased ATP/ADP ratio leads to closure of ATP-sensitive potassium channels, subsequent membrane depolarization, and opening of voltage-gated sodium and calcium channels. Increased intracellular calcium triggers fusion of insulin-containing granules with the cell surface membrane and release of insulin into the bloodstream. Fusion of insulin vesicles with the plasma membrane is coordinated by SNARE proteins and synaptotagmin calcium sensors. Abbreviations: ACh, acetylcholine; ADP, adenosine diphosphate; ATP, adenosine triphosphate; CCK, cholecystokinin; ER, estrogen receptor, FFA, free fatty acids; GIP, gastric inhibitory polypeptide; GLP1, glucagon-like peptide 1; GLUT2, glucose transporter 2; IP3R, inositol 1,4,5-trisphosphate receptor; RyR, Ryanodine Receptor K_ATP, potassium-gated ATP channel; VDCC, voltage-dependent calcium channel; NaV, voltage gated sodium channel.

Disrupted ER homeostasis, often broadly referred to as “ER stress”, has been associated with β cell dysfunction in diabetes [2, 7–9]. One prominent example of ER dysfunction causing β cell pathology is found in individuals with MIDY, in which mutations in proinsulin lead to defective protein folding and impaired β cell function. The cysteine-to-tyrosine mutation at residue A7 [C(A7)Y] of the insulin gene, found in some MIDY patients and in the Akita mouse model of diabetes, prevents formation of the critical B7:A7 disulfide bond. This misfolded mutant proinsulin exerts a dominant-negative effect, causing misfolding and aggregation of wild-type, bystander proinsulin, thus impairing β cell function [3, 10]. Similarly, in type 1 diabetes (T1D) models such as the nonobese diabetic (NOD) mouse, ER dysfunction is evident before β cell destruction [11], suggesting that differential post-translational modifications of proteins in β cells during ER stress may contribute to the generation and presentation of neo-antigens that lead to autoimmunity [12]. Whether causative of cellular dysfunction or reflecting compensatory changes, the ER volume is increased in β cells from individuals with T2D [8].

Pancreatic β cells employ three distinct quality control pathways to maintain normal ER homeostasis and protein secretion (Figure 2A). These systems include the unfolded protein response (UPR), autophagy/ER-phagy, and ER-associated degradation (ERAD) (Figure 2B–D).

Figure 2. ER protein quality control pathways.

(A) Properly folded ER proteins are trafficked from the ER lumen through the Golgi apparatus, prior to reaching their final destination in the plasma membrane or secretory granules. (B) Misfolded proteins inside the ER may trigger the UPR via three sensors (IRE1α, PERK, and ATF6) to halt new protein synthesis while enhancing ER folding and degradation capacity. (C) Insoluble protein aggregates may be targeted to the lysosome for degradation by autophagy/ER-phagy. (D) Unfolded, terminally misfolded, or unassembled proteins are retrotranslocated to the cytosol by the ERAD complex for proteasomal degradation. The Sel1L-Hrd1 protein complex represents the most conserved ERAD components in mammalian cells. Together, these three protein quality control systems work in coordination to maintain ER homeostasis.

Below, we review recent literature evaluating the function of each pathway in β cells, with a focus on studies using β cell-specific genetic mouse models (Table 1).

Table 1.

Mouse models of UPR, autophagy, and ERAD in pancreatic β cells.

Gene	Model (Cre used)	Phenotype	Reference
UPR
Ern1 (IRE1α)	RIP-Cre	Hyperglycemia and hypoinsulinemia on NCD; impaired compensatory β cell proliferation following HFD due to reduced Xbp1s-Ccnd1 axis signaling	[25]
Ern1 (IRE1α)	Rat Ins2 promoter-driven CreERT (inducible)	Hyperglycemia and hypoinsulinemia; reduced proinsulin translation but no effect on insulin mRNA (or PCs, CPE), reduced total proinsulin and insulin content; no cell death	[23]
Ern1 (IRE1α)	Ins2-Cre and adenovirus-Cre	Hyperglycemia and hypoinsulinemia, decreased insulin biosynthesis, reduced proinsulin and insulin content; defective proinsulin folding due to reduced expression of five PDI family proteins	[24]
Xbp1	RIP-Cre	Modest hyperglycemia and hypoinsulinemia, impaired proinsulin processing, increased proinsulin/insulin in serum, reduced proliferation, no cell death	[26]
Eif2ak3 (PERK)	Global KO	Progressive hyperglycemia, increased cell death, Increased insulin biosynthesis at high glucose but no impairment in proinsulin to insulin conversion	[40]
Eif2ak3 (PERK)	Global KO, Neurog3-Cre, Pdx1-Cre, RIP-Cre	Hyperglycemia and β cell loss in all models except RIP-Cre; Reduced β cell mass due to reduced proliferation and differentiation during perinatal and postnatal periods but no cell death; Defective proinsulin trafficking and reduced stimulated insulin secretion	[42]
Eif2ak3 (PERK)	Global KO	Increased proinsulin accumulation in ER due to impaired ER-Golgi trafficking and reduced chaperones; impaired ERAD and reduced proteasomal activity	[45, 47]
Eif2ak3 (PERK)	Global inducible Cre-ERT2	Hyperglycemia, loss of β cell architecture, intracellular accumulation of proinsulin and Glut2, increased proliferation and cell death, no decrease in insulin mRNA, increased Xbp1s and cleaved ATF6	[44]
Eif2ak3 (PERK)	Pdx1-CreER	Hyperglycemia, reduced islet number and cell size, increased cell death and proliferation	[44]
Eif2a	Ser51Ala mutant	Hyperglycemia, increased proinsulin translation, defective cargo trafficking, increased oxidative damage, reduced expression of stress response genes and apoptosis	[46]
Atf4	Global KO	Normal glucose homeostasis and islet morphology	[46]
Ddit3 (CHOP)	Global KO crossed with mouse models of diabetes	Improved β cell function, maintenance of insulin content, increase proliferation and decreased apoptosis	[49, 50]
ATF6	Global KO	Impaired glucose tolerance, reduced insulin secretion and content upon HFD feeding but improved insulin sensitivity; worsened phenotype of Akita mice	[55]
Autophagy
Atg7	RIP-Cre	Hyperglycemia, impaired insulin secretion, islet degeneration, lack of compensatory increase of β cell mass upon high fat feeding	[56]
Atg7	RIP-Cre	Hyperglycemia, glucose intolerance, reduced serum insulin, reduced β cell mass and pancreatic insulin content due to increased apoptosis and reduced proliferation of β cells	[57]
Becn1 (Beclin-1)	Becn1F121A (Global knock-in)	Glucose intolerance with reduced insulin content upon HFD but improved insulin sensitivity	[102]
ERAD
Sel1l (Sel1L)	RIP-Cre	Reduced GSIS, proinsulin accumulation	[75]
Sel1l (Sel1L)	Ins1-Cre	Hyperglycemia and hypoinsulinemia, postnatal reduction in markers of mature β cells with increased expression of markers of immaturity; no proinsulin accumulation	[76]
Derl3 (Derlin 3)	Global KO	No effect in glucose homeostasis; increased expression of Hrd1 and GRP78	[106]
Herpud1	Global KO	Impaired glucose tolerance (basal glucose was normal); increased expression of Hrd1 and GRP78	[106]

Abbreviations: CPE, carboxypeptidase E; GSIS, glucose stimulated insulin secretion; HFD, high fat diet; KO, knockout; NCD, normal chow diet; PDI, protein disulfide isomerase

2. UPR

The UPR is an adaptive signaling pathway that responds to accumulation of misfolded proteins within the ER. In mammals, the UPR comprises signaling branches orchestrated by three ER-localized transmembrane sensors (Figure 2B), namely inositol requiring enzyme 1α (IRE1α), protein kinase RNA-activated (PKR)-like ER kinase (PERK) and activating transcription factor 6 (ATF6). In the basal state, these sensors are maintained in inactive monomers, bound to the ER lumenal chaperone BiP. Misfolded proteins within the ER lumen engage BiP, releasing it from the UPR sensors. Once released from BiP, the sensors induce distinct signaling cascades to activate downstream effectors in an effort to maximize protein folding and degradation capacity within the ER.

Increased UPR signaling has been reported in islets obtained from T2D patients [13] and rodent models of diabetes (high fat-fed or ob/ob mice) [13, 14], possibly triggered by inflammatory stress and/or glucolipotoxicity associated with increased ER protein folding demands [15]. As such, UPR activation has been traditionally associated with β cell pathology. Alternatively, UPR signaling directly affects proinsulin biosynthesis and folding, and may be important for β cell proliferation during conditions such as high fat-diet feeding. Supporting this concept, recent single-cell sequencing data suggested that healthy β cells constantly undergo cycles of high and low UPR signaling [16], consistent with data demonstrating that cells with active UPR signaling are more likely to proliferate [16–18]. Furthermore, high fat-fed rodents and ob/ob mice show increased expression of UPR markers and ER chaperones during their β cell expansion phase, highlighting the need for increased UPR signaling in β cell proliferation/compensation [14]. This raises an important question of whether the increased UPR signaling and chaperone expression in islets merely indicates an adaptive response instead of pathology [17], a distinction that warrants further investigation. Further insights into the role of individual UPR pathways in β cell function have been obtained through rodent models with targeted deletion of UPR proteins, as summarized below and in Table 1.

2.1. The IRE1α-Xbp1 pathway

IRE1α is a type I transmembrane protein that is largely in an inactive state during basal conditions. Upon activation, IRE1α undergoes dimerization and trans-autophosphorylation, leading to non-conventional splicing of Xbp1 mRNA [19]. Spliced Xbp1 (Xbp1s) is a potent transcription factor that upregulates expression of ER proteins that promote protein synthesis, maturation, and turnover [20, 21].

In β cells, IRE1α signaling has been implicated in promoting insulin biosynthesis during high glucose stimulation in vitro [22]. Indeed, genes required for proinsulin biosynthesis, including those mediating ribosome recruitment to the ER, cotranslational translocation, signal peptide cleavage, protein folding, and trafficking, were significantly reduced in IRE1α-deficient islets treated with high glucose [23]. Proinsulin folding may further be affected in IRE1α-deficient β cells as a result of reduced levels of ER chaperones [24]. In vivo, both constitutive (rat insulin promoter [RIP]-Cre-mediated) [25] and inducible (Ins2-driven CreERT-mediated) [23] β cell-specific IRE1α knockout mice exhibited reduced proinsulin synthesis, insulin secretion, and increased oxidative stress on a normal chow diet. Similarly, β cell-specific Xbp1 null mice (RIP-Cre-mediated) displayed impaired proinsulin maturation and defective ER to Golgi trafficking accompanied by peri-islet fibrosis [26]. In addition to ER protein biosynthesis and maturation, the IRE1α-Xbp1 pathway may also be important for compensatory proliferation of β cells during high fat feeding. High fat fed IRE1α null mice exhibited defective islet expansion [25]. Moreover, β cell-specific Xbp1 null mice had a 50% reduction in β cell mass even on a normal chow diet [26] likely due to reduced cyclin D1 expression, a downstream target of Xbp1s [25] (Table 1).

Upon prolonged activation, IRE1α signaling is hypothesized to induce apoptosis by multiple mechanisms, such as regulated IRE1α-dependent degradation (RIDD), which causes promiscuous degradation of ER-localized mRNA [27, 28], and the activation of pro-apoptotic c-Jun kinase [29], B cell lymphoma 2 (Bcl-2) [30] and thioredoxin-interacting protein (TXNIP) signaling [31–33]. However, the physiological relevance of these pathways in β cells remains unclear.

2.2. The PERK-eukaryotic initiation factor-2α (eIF2α) Pathway

Similar to IRE1α, the UPR sensor PERK undergoes dimerization and trans-autophosphorylation upon activation and phosphorylates its substrate eIF2α at serine residue 51 (Ser51) to attenuate global protein translation. However, translation of Atf4 is preferentially increased [34], which in turn induces the expression of ER chaperones [35], genes implicated in redox and amino acid metabolism [36] and autophagy [37], and the well-known proapoptotic gene C/EBP-homologous protein (CHOP) [38].

Exemplifying the importance of the PERK pathway in β cells is Wolcott-Rallison syndrome, in which a loss-of-function PERK mutation (resulting in premature termination due to a nucleotide insertion at residue 1103, 1103insT) leads to progressive pancreas degeneration, including loss of islet cells [39]. This phenotype has been recapitulated in global PERK knockout mice [40, 41]. Global, pancreas-, and islet-targeted (using Pdx1-Cre and Neurog3-Cre) PERK deletion models exhibited severe defects in β cell proliferation and differentiation, resulting in decreased β cell mass, accompanied by defective proinsulin trafficking and reduced insulin secretion [42]. Similarly, loss-of-function eIF2α phospho-mutant (S51A) mice exhibit hyperglycemia with increased proinsulin translation, defective cargo trafficking, increased oxidative damage, reduced expression of stress response genes, and increased apoptosis [8, 43]. One of the most striking observations in PERK-deficient islets is the distinct accumulation of proinsulin in the ER [35, 41, 42, 44]. However, the precise mechanism by which PERK affects proinsulin maturation and trafficking remains debated [45]. PERK deletion was initially proposed to derepress translation of proinsulin, thereby increasing the ER burden [40, 46], while other studies point to the direct role of PERK in proinsulin maturation and trafficking, by promoting the expression of chaperones [47] or by enhancing ER-Golgi trafficking and ERAD-mediated proteasomal degradation [45] (Table 1).

While PERK-ATF4 signaling has a primary pro-survival effect mediated by reduced global translation and selective induction of autophagic proteins, it also promotes transcriptional activation of a key pro-apoptotic player, CHOP [48]. CHOP may induce apoptosis through several mechanisms, including activation of proapoptotic proteins caspases, DNA damage-inducible 34 (GADD34), and downregulation of anti-apoptotic proteins. Islets from diabetic mouse models and patients with T2D have increased cytosolic and especially nuclear CHOP expression [49, 50]. Supporting the detrimental role of CHOP expression, global deletion of CHOP in genetic (db/db mice) or induced (high fat-fed or streptozotocin-treated mice) models of diabetes significantly improved glycemic status and β cell function, proliferation, and survival [49]. Global deletion of CHOP also delayed onset of diabetes in heterozygous Akita mice by 8–10 weeks [50] (Table 1).

2.3. The ATF6 Pathway

ATF6α and ATF6β are ubiquitously expressed ER membrane-bound transcription factors. Following stress, they exit the ER and undergo proteolytic cleavage by site-1 and −2 proteases in the Golgi apparatus. This proteolysis generates a N-terminal cytosolic fragment encoding a transcription factor, ATF6(N), which translocates to the nucleus and induce the expression of ER chaperones and foldases [51], calcium transport proteins, and redox genes [18, 52]. Missense mutations in ATF6α in Dutch [53] and Pima Indian [54] cohorts have been linked to T2D, although causality remains unclear. Mice with global ATF6α deletion exhibited no discernible impairment in glucose homeostasis except after high fat-diet feeding, which led to impaired glucose tolerance and failed β cell mass expansion [55] (Table 1). More recently, using small molecule inhibitors, ATF6 was shown to be necessary and sufficient for glucose-induced proliferation of mouse β cells [18]. However, no mouse model targeting ATF6 in β cells has been reported.

3. AUTOPHAGY/ER-PHAGY

Autophagy, or cellular “self-eating,” is an evolutionarily conserved mechanism for recycling cellular components via lysosomal degradation. This term refers to a general process (macroautophagy), or to degradation of specific intracellular targets, including secretory vesicles (crinophagy or vesicophagy), mitochondria (mitophagy), or endoplasmic reticulum (ER-phagy). In macroautophagy, a double-membraned structure called the autophagosome wraps around the cellular material destined for degradation and eventually fuses with the lysosome. Macroautophagy plays a prominent role in promoting β cell survival and aggregate clearance [56–58] (Table 1). As other comprehensive reviews have covered this topic in β cells [59–61], here we will primarily focus on the emerging role of autophagy on ER homeostasis, a process known as ER-phagy.

ER-phagy is a mechanism to selectively degrade the ER and its lumenal content in a receptor-dependent manner [62]. Some of the putative ER-phagy receptors are family with sequence similarity 134 member B (FAM134B), reticulon-3 (RTN3), Sec62, cell cycle progression-1 (CCPG1) and atlastin3 (ATL3). The key feature of these receptors is one or more LC3/GABARAP-interacting regions in the cytosolic domain that mediate the interaction of ER with the autophagosome membrane. Once the receptor binds to LC3 of the autophagosomal precursors, the membrane expands and seals around the ER to form a discrete vesicle, and the autophagosome fuses with the lysosome to be degraded (Figure 2C).

A recent study showed that insoluble ER aggregates of misfolded Akita mutant proinsulin are selectively removed via RTN3-dependent lysosomal degradation [63]. Using a cell culture model, they demonstrated that ER lumenal chaperone Grp170 selectively associates with soluble, high molecular weight complexes of Akita insulin and prevents them from entering the detergent-insoluble fraction of aggregated proteins [63, 64]. Depletion of Grp170 led to Akita aggregates colocalized with lysosomal marker LAMP1, suggesting lysosomal clearance. However, concurrent depletion of RTN3 prevented Akita insulin-LAMP1 colocalization, suggesting that lysosomal delivery of insoluble aggregates is RTN3-dependent. As prior in vivo studies on autophagy in β cells employed disruption of broad processes via deletion of Atg7 or use of nonspecific chemical inhibitors, the extent to which defects in autophagy specifically impair ER function remains unclear. Future experiments with animal models targeting ER-phagy receptors will be important to appreciate the effects and significance of β cell ER-phagy in diabetes pathogenesis.

4. ERAD

ERAD is a principal quality-control checkpoint that targets lumenal or membrane-bound ER proteins for cytosolic proteasomal degradation (Figure 2D). The best-characterized core ERAD components in mammals include the E3 ubiquitin ligase Hrd1 and its adaptor protein Sel1L, both of which are ubiquitously expressed and evolutionarily conserved from yeast to humans. The major function of Hrd1 is to serve as a retrotranslocon for extraction and ubiquitination of ER proteins [65, 66]. Sel1L is essential for maintaining the stability of Hrd1 [67], and in yeast, is also involved in substrate recruitment [68]. Recent studies in vivo have demonstrated a pivotal role of ERAD in health and disease, as Sel1L-Hrd1 ERAD controls organismal growth and body weight, systemic water balance, and lipid metabolism, as reviewed in [69, 70].

Islets from young Akita mice exhibit increased Sel1L expression [71, 72], suggesting an attempt to improve ERAD function that may ultimately be insufficient for clearance of misfolded insulin aggregates. Interestingly, female Akita mice are resistant to the development of hyperglycemia. A study investigating the protective role of estrogen in β cells of Akita mice found that signaling through the estrogen receptor α suppresses UBC6e, an ER-resident E2 ubiquitin-conjugating enzyme that targets ERAD-associated proteins for degradation [73]. The estrogen receptor α-mediated suppression of UBC6e thus stabilized the expression of Sel1L and Hrd1, leading to the clearance of mutant but not wild-type proinsulin [74].

Recently, ERAD function was studied in vivo using β cell-specific deletion of Sel1L using RIP-Cre [75] and Ins1-Cre mice [76] (Table 1). Both models exhibited impaired glucose tolerance and had reduced β cell insulin content. Sel1L-deficient β cells from the RIP-Cre model demonstrated increased proinsulin accumulation in the ER [75]. In contrast, Sel1L-deficient islets from our Ins1-Cre model had reduced proinsulin content without accumulation within the ER. Instead, we observed loss of β cell identity in young adult mice, with reduced expression of mature β cell proteins GLUT2, Ucn3, and MafA, and increased expression of markers associated with immaturity or dedifferentiation, such as Ngn3 and Aldh1a3 [76]. Using single-cell RNA sequencing, we found that ERAD deficiency altered TGFβ pathway signaling, with elevated TGFβ receptor 1 (TGFβR1) expression. Treatment of isolated Sel1L-deficient islets with a TGFβR1-specific antagonist restored expression of MafA and insulin. Supporting this role of ERAD in maintaining β cell identity, an independent study showed that the Sel1L-Hrd1 pathway is induced in β cells of streptozotocin-treated mice following treatment with GLP1-estrogen conjugate and PEG-insulin, a combination that reversed β cell dedifferentiation and improved β cell function [77].

Whether native proinsulin is an ERAD substrate has implications for T1D, as proinsulin-derived peptides presented by MHC class I molecules on β cells are targets of the autoimmune process. Antigen presentation by MHC class I is increased in stressed β cells, and increased ERAD function during stress may contribute to presentation of abnormal self-peptides [78, 79]. Thus far, this specific consideration has only been addressed using in vitro studies, in which shRNA-mediated silencing of ERAD-associated proteins Derlin-2, p97, and Hrd1 led to accumulation of proinsulin in transfected K562 (MHC class I-expressing leukemia) cells [80]. It remains to be demonstrated whether and how ERAD affects the generation of autoantigen and the pathogenesis of T1D in vivo.

Proinsulin is not the only putative peptide substrate of Sel1L-Hrd1 ERAD. In β cells, ERAD may degrade tomosyn (syntaxin binding protein 5), a protein that limits formation of the SNARE complex involved in insulin exocytosis. In response to insulin secretagogues, tomosyn-2 was phosphorylated, allowing for direct binding to Hrd1, which targeted its degradation in insulinoma cells [81]. Another study recently reported β cell transcription factor MafA as a putative Hrd1 substrate in β cells of diabetic mice, and showed that overexpression of Hrd1 in β cells contributed to impaired insulin secretion, partly through reduced nuclear MafA expression [82]. Whether MafA is an endogenous ERAD substrate under physiological conditions, and how ERAD recognizes and degrades MafA remain open questions. Thus, further investigation is warranted to investigate how ERAD regulates β cell function in a substrate-dependent manner under these various settings in vivo.

5. CROSSTALK BETWEEN ER QUALITY CONTROL MECHANISMS

Protein quality control pathways are often studied in isolation for experimental simplicity, but act in a synergistic and/or compensatory fashion to maintain ER homeostasis in vivo. ERAD and autophagy are evolutionarily conserved degradative mechanisms in eukaryotic cells (Figure 2C–D). In principal, ERAD is the first line of defense against misfolded ER proteins by clearing monomeric and soluble substrates; however, terminally misfolded or insoluble protein aggregates may exhibit structural constraints for dislocation via the retrotranslocon [83]. In that case, a direct ER-to-lysosomal degradation pathway may be activated to clear these proteins [84, 85]. A notable example of this phenomenon is the Z-variant of human alpha-1-antitrypsin, whose degradation pathway depends on its expression and aggregation level [86]. Normally, this mutant is a substrate of ERAD, but when overexpressed it is removed by autophagy. Indeed, soluble Akita mutants were principally cleared by ERAD whereas insoluble larger aggregates were cleared by ER-phagy in vitro [63, 87].

Following ER stress, UPR signaling may drive autophagy; however, a detailed mechanism for this interplay remains elusive and may involve both transcriptional and post-transcriptional regulation. Both IRE1α-Xbp1 and PERK-eIF2α pathways can directly increase the transcription of autophagic components by binding to their promoter elements [37, 84, 88, 89]. Moreover, IRE1α may also activate autophagosome formation by phosphorylating Beclin2 via JNK activation [90, 91], and PERK may activate autophagy by inhibiting mTORC1 through the CHOP-Trib3 axis [92]. Parallel induction of ER stress and autophagy was reported in Akita β cells, although it is unclear if a specific UPR pathway is involved. Improvement in ER stress following treatment with chemical chaperones led to decreased autophagic flux [93], suggesting coordination between the two pathways. Additionally, the autophagy enhancer rapamycin dramatically improved hyperglycemia, increased pancreatic insulin content, and prevented β cell apoptosis in Akita mice [93]. Conversely, inhibition of autophagy in β cells by Atg7 deletion has been associated with ER stress, ER dilation and cell death [56, 57] suggesting that ER-autophagy crosstalk may coordinate to eliminate irreparably misfolded proinsulin and protect β cell survival and function.

Crosstalk also exists between ERAD and UPR. During ER stress, IRE1-Xbp1 activation induces the expression of ERAD components to promote degradation of unfolded/misfolded proteins and restore ER homeostasis [94–97]. Conversely, ERAD directly controls the expression of IRE1α by targeting it for proteasomal degradation in several cell types [98], suggesting a unique feedback mechanism to fine-tune IRE1α signaling.

In summary, maintenance of ER homeostasis in β cells involves complex interplay of multiple quality-control pathways, each with specific roles (Figure 3). Whether these pathways act sequentially or synergistically in vivo remains a subject of intense investigation.

Figure 3. ER quality control pathways differentially regulate β cell processes.

Experimental evidence to date has revealed specific functions of each pathway in maintaining β cell health (left), or contributing to β cell pathology when the pathways are overactive or otherwise dysfunctional (right). Our understanding of crosstalk between these pathways is limited; thus, investigating interactions between these pathways under various physiologic conditions will advance our understanding of the coordinated β cell response to cellular stress.

6. FUTURE DIRECTIONS

Modulating ER function is a promising pharmaceutical target for diabetes, and several antihyperglycemic agents in clinical use alleviate ER protein folding burden, either through reducing endogenous insulin demand or promoting ER chaperone expression (summarized in [99]). Chemical chaperones, including taurine-conjugated ursodeoxycholic acid (TUDCA) and 4-phenylbutyric acid (4-PBA), increase ER folding capacity and were shown to alleviate ER stress [100, 101]. Autophagy enhancers such as rapamycin are beneficial in delaying diabetes in mouse models of diabetes [93, 100, 101]. However, each of these agents has broad effects in vivo, and may be difficult to titrate to a specific response. As demonstrated with global knock-in Becn1^F121A mice, which harbor a mutation in the Beclin-1 protein to promote constitutively active autophagy and have improved insulin sensitivity at the expense of β cell function [102] (Table 1), enhancement of a pathway in one cell type may be less helpful or even detrimental in another. Recently, studies on the role of Sel1L-Hrd1 ERAD revealed its importance in β cell identity [76]. Indeed, one of two top pathways induced in pharmacologic reversal of dedifferentiated β cells is Sel1L-Hrd1 ERAD [77], pointing to an exciting potential new target for diabetes therapy. However, these experimental models suggest that ERAD protein levels must be precisely titrated for optimal β cell function, and further investigation is needed to determine whether changes in ERAD protein expression are causative mechanisms or adaptive β cell responses in diabetes pathogenesis.

Work over the past decades has uncovered a remarkable role of ER protein quality-control mechanisms in β cell fate (Figure 3). UPR activation has been suggested to have a binary effect, beneficial to cells in early stages but leading to cell death when stress is prolonged [103, 104]. Similarly, loss of autophagy leads to increased protein aggregates and β cell death. Loss of ERAD, on the other hand, results in loss of β cell identity, thus demonstrating an emerging role for protein quality control mechanisms in cell fate determination. In β cells, it is unclear if a dedifferentiated state reflects a temporary transition state while the cell is responding to stress instead of a terminal process [105]. While genetic deletion models provide important insights on the importance of individual pathways, cells in vivo integrate multitude of signals to adapt to stressors. Future work should thus focus on investigating the complex cross-communication and interdependence of protein quality control pathways in vivo, and their potential impact on β cell fate.