Endoplasmic reticulum stress and eIF2α phosphorylation: The Achilles heel of pancreatic β cells

Abstract

Background: Pancreatic β cell dysfunction and death are central in the pathogenesis of most if not all forms of diabetes. Understanding the molecular mechanisms underlying β cell failure is important to develop β cell protective approaches.

Scope of review: Here we review the role of endoplasmic reticulum stress and dysregulated endoplasmic reticulum stress signaling in β cell failure in monogenic and polygenic forms of diabetes. There is substantial evidence for the presence of endoplasmic reticulum stress in β cells in type 1 and type 2 diabetes. Direct evidence for the importance of this stress response is provided by an increasing number of monogenic forms of diabetes. In particular, mutations in the PERK branch of the unfolded protein response provide insight into its importance for human β cell function and survival. The knowledge gained from different rodent models is reviewed. More disease- and patient-relevant models, using human induced pluripotent stem cells differentiated into β cells, will further advance our understanding of pathogenic mechanisms. Finally, we review the therapeutic modulation of endoplasmic reticulum stress and signaling in β cells.

Major conclusions: Pancreatic β cells are sensitive to excessive endoplasmic reticulum stress and dysregulated eIF2α phosphorylation, as indicated by transcriptome data, monogenic forms of diabetes and pharmacological studies. This should be taken into consideration when devising new therapeutic approaches for diabetes.

Keywords: ATF, activating transcription factor; CHOP, C/EBP homologous protein; CRISPR, clustered regularly interspaced short palindromic repeats; CReP, constitutive repressor of eIF2α phosphorylation; Diabetes; ER, endoplasmic reticulum; ERAD, ER-associated degradation; Endoplasmic reticulum stress; GCN2, general control non-derepressible-2; GIP, glucose-dependent insulinotropic polypeptide; GLP-1, glucagon-like peptide 1; GWAS, genome-wide association study; HNF1A, hepatocyte nuclear factor 1-α; HRI, heme-regulated inhibitor kinase; IAPP, islet amyloid polypeptide; IER3IP1, immediate early response-3 interacting protein-1; IRE1, inositol-requiring protein-1; ISR, integrated stress response; Insulin; Islet; MEHMO, mental retardation, epilepsy, hypogonadism and -genitalism, microcephaly and obesity; MODY, maturity-onset diabetes of the young; NRF2, nuclear factor, erythroid 2 like 2; PBA, 4-phenyl butyric acid; PERK, PKR-like ER kinase; PKR, protein kinase RNA; PP1, protein phosphatase 1; PPA, phenylpropenoic acid glucoside; Pancreatic β cell; Pdx1, pancreatic duodenal homeobox 1; RIDD, regulated IRE1-dependent decay; RyR2, type 2 ryanodine receptor/Ca2+ release channel; SERCA, sarcoendoplasmic reticulum Ca2+ ATPase; TUDCA, taurine-conjugated ursodeoxycholic acid derivative; UPR, unfolded protein response; WFS, Wolfram syndrome; XBP1, X-box binding protein 1; eIF2, eukaryotic translation initiation factor 2; eIF2α; hESC, human embryonic stem cell; hPSC, human pluripotent stem cell; hiPSC, human induced pluripotent stem cell; uORF, upstream open reading frame.

Figure 1

Endoplasmic reticulum stress signaling. ER stress leads to increased binding of the ER chaperone BiP to misfolded proteins in the ER lumen, causing the dissociation of BiP from the ER stress transducers PERK, IRE1, and ATF6, resulting in their activation. Activated (phosphorylated) PERK phosphorylates eIF2α and thereby attenuates general protein translation to relieve the ER workload during stress. In parallel, eIF2α phosphorylation enhances ATF4 translation. ATF4 induces transcription of chaperones and CHOP. CHOP induces expression of GADD34, which targets PP1 to eIF2α for dephosphorylation and relief of translational inhibition. IRE1 activation (phosphorylation) causes the splicing of XBP1 mRNA, generating the transcription factor sXBP1. sXBP1 upregulates the expression of chaperones, folding enzymes and components of the ERAD machinery. Activated ATF6 translocates to the Golgi where it is cleaved to a mature transcription factor that will drive chaperone expression. WFS1 inhibits ATF6 through ubiquitination and proteasomal degradation, by targeting HRD1 (E3 ubiquitin ligase) to ATF6. CREB3 and CREB3L1 through CREB3L4 are cell type- and context-specific ER stress transducers that are cleaved/activated in a similar manner as ATF6.

Figure 2

Regulation of eIF2α phosphorylation. ER stress leads to PERK phosphorylation and activation, and phosphorylation of eIF2α and NRF2, an antioxidant response transcription factor. The eIF2 protein consists of three subunits, eIF2α, -β, and -γ. Active eIF2 has a non-phosphorylated eIF2α and low affinity to the guanine nucleotide exchange factor eIF2B. In that state eIF2B exchanges GDP to GTP from the eIF2γ subunit, ensuring its active state. Non-phosphorylated eIF2α, eIF2β, GTP-loaded eIF2γ, methionyl tRNA (Met-tRNA_i), and eIF5 form the ternary complex. Upon start site recognition eIF2 and eIF5 dissociate from the complex, and translation initiation and elongation ensue. eIF2α phosphorylation at Ser51 increases its affinity for eIF2B and eIF5, reduces the guanine nucleotide exchange that slows down the formation of the ternary complex, and thereby attenuates translation initiation. In parallel, this initiates ATF4 translation and downstream expression of chaperones, antioxidant response genes, CHOP and GADD34. GADD34-bound PP1 dephosphorylates eIF2α and ends translational inhibition and ATF4 expression/signaling. CReP is another constitutively expressed non-enzymatic cofactor for PP1 that tonically keeps eIF2α phosphorylation down. The BiP co-chaperone p58^IPK inhibits PERK and downstream signaling. eIF2α can also be phosphorylated by the non-ER stress-related kinases HRI, PKR, and GCN2.

Figure 3

Monogenic diabetes due to excessive or dysregulated endoplasmic reticulum stress signaling. Four monogenic forms of diabetes pertain to the PERK branch of the UPR. Inactivating mutations in EIF2AK3, encoding PERK, cause Wolcott-Rallison syndrome (left). In these patients, PERK is unable to phosphorylate eIF2α, leading to absent PERK signaling, loss of translational control, ER stress and β cell loss. In the three other monogenic forms (middle), eIF2α phosphorylation/inactivation and downstream signaling are enhanced. Loss-of-function mutations in DNAJC3, encoding p58^IPK, cause diabetes and neurodegenerative features. The p58^IPK inactivation results in increased PERK activity and eIF2α phosphorylation. Loss-of-function mutations in PPP1R15B, encoding CReP, causes a syndrome comprising diabetes, short stature, intellectual disability, and microcephaly. The PPP1R15B mutation destabilizes the CReP-PP1 holophosphatase complex and thereby enhances eIF2α phosphorylation. Mutations in EIF2S3, encoding eIF2γ, cause MEHMO syndrome (mental retardation, epilepsy, hypogonadism and hypogenitalism, microcephaly, and obesity). These damaging EIF2S3 mutations impair eIF2 function and enhance downstream signaling. Missense mutations in IER3IP1 lead to a Wolcott-Rallison-like syndrome of microcephaly, epilepsy, and neonatal diabetes. Recessive mutations in WFS1 and WFS2 lead to Wolfram syndrome. INS mutations that impair proinsulin folding cause β cell demise and neonatal diabetes.

Figure 4

Heatmap of genes related to the endoplasmic reticulum stress response. RNA-seq gene expression levels (in RPKM) from human tissues were obtained from GTEx (v4.p1) . RNA-seq data of FACS-purified human β cells were from Nica et al. and human islet RNA-seq data from Eizirik et al. and Cnop et al. . Genes and tissue/cell types are arranged according to cluster analysis. Protein names are indicated in parenthesis when different from gene names. Panel A shows more abundantly expressed genes (100–400 RPKM) and panel B less abundant genes (0–100 RPKM). Some MODY genes are shown for comparison.