IER3IP1 Mutations Cause Neonatal Diabetes Due to Impaired Proinsulin Trafficking

Insulin, the vital hormone responsible for tightly regulating glucose homeostasis, is mainly synthesized in pancreatic β-cells (1). In response to glucose stimulation, β-cells swiftly elevate the translation and subsequent processing of the gene encoding the insulin precursor preproinsulin (PPI) (2). After translation, preproinsulin is directed by its signal peptide to the endoplasmic reticulum (ER) (3), where signal peptidase cleaves the signal peptide forming proinsulin. ER-resident chaperones aid in the folding of proinsulin within the oxidative environment of the ER. Subsequently, correctly folded proinsulin is transported through the ER-to-Golgi trafficking machinery to the Golgi apparatus, where it is packaged into secretory granules (4). Once stored in the secretory granules, proinsulin undergoes gradual processing by prohormone convertase (PC1/3) and carboxypeptidase E to form mature insulin and C-peptide in equimolar amounts (5). Thus, the proper function of the ER-to-Golgi trafficking machinery is crucial for maintaining ER homeostasis and the dynamic insulin production and secretion capabilities of β-cells.

Recent studies have elucidated that impaired ER export disrupts proinsulin folding and induces ER stress, subsequently leading to β-cell failure and diabetes (6–9). For instance, perturbations in the coat protein complex II (COPII)-dependent ER export pathway in β-cells have been shown to cause intracellular accumulation of proinsulin in the ER and the initiation of ER stress-induced apoptosis (10). Furthermore, inhibition of SURF4, an ER cargo receptor that mediates proinsulin export, leads to retention and misfolding of proinsulin within the ER, thereby compromising the normal functionality of β-cells (11). Moreover, we recently reported that mutations in YIPF5, which facilitates cargo export from the ER, cause early-onset diabetes due to exacerbated ER stress and apoptosis in β-cells (12).

Monogenic mutations in immediate early response 3 interacting protein 1 (IER3IP1) have been reported to cause microcephaly, epilepsy, and early-onset diabetes syndrome 1 (Online Mendelian Inheritance in Man: 614231) (summarized in Supplementary Table 1) (13–18). IER3IP1 is highly expressed in the developing brain cortex and pancreatic β-cells (17). It localizes to the ER, ER exit sites, and Golgi complex and has been postulated to mediate cargo transport between these compartments (19,20). A CRISPR-based screen performed on human cerebral organoids identified IER3IP1 as a pivotal regulator of ER homeostasis and extracellular matrix protein secretion (21). Furthermore, a recent study has shed light on the essential role of IER3IP1 in preserving β-cell function and survival in mice (22). Decreased IER3IP1 levels have been observed in the β-cells of patients with type 2 diabetes (22,23). However, the precise mechanism through which IER3IP1 functions in human β-cells remains to be fully elucidated.

In this study, we used genome-edited stem cells and directed differentiation into pancreatic stem cell-derived islets (SC-islets) to decipher the role of IER3IP1 in human β-cell development, function, and survival. Our results show that IER3IP1 mutations lead to reduced β-cell number and altered trafficking of proinsulin, resulting in impaired β-cell functionality. Moreover, we implanted IER3IP1-deficient stem cell islets into immunocompromised mice to further reveal the phenotypic defects in human β-cells resulting from IER3IP1 mutations.

Human embryonic stem cells (hESCs) (WA01/H1 line, Wicell) were differentiated toward functional pancreatic endocrine cells as previously described (24).

IER3IP1-knockout (KO) hESCs were generated using two guides targeting the first exon. To introduce the patient mutation IER3IP1^V21G, a guide RNA targeting the mutation site and 100-b single-stranded oligodeoxynucleotides mutation template were designed using Benchling. Details are provided in the Supplementary Material.

hESC-derived cells were dissociated and incubated with the corresponding antibodies overnight and then analyzed with FACSCalibur cytometer. Details are described in the Supplementary Methods.

Total RNA was extracted using NucleoSpin Plus RNA kit (Macherey-Nagel), and cDNA was generated using GoScript Reverse Transcriptase kit (Promega) according to the manufacturer’s instruction. Quantitative (q)RT-PCR was performed using 5x HOT FIREPol EvaGreen qPCR Mix Plus in a Rotor-Gene Q (Qiagen) thermocycler. Details are provided in the Supplementary Methods.

Fifty SC-islets were incubated in 1 mL full S7 media supplemented with 1 µmol/L thapsigargin (TG; T9033, Sigma-Aldrich) or 5 µg/mL tunicamycin (TM; NT7765, Sigma-Aldrich). After 48 h, aggregates were collected and processed for immunohistochemistry.

Dispersed SC-islets were transduced with the lentivirus containing the retention using selective hooks (RUSH) construct and plated on Matrigel-coated glass-bottom dishes (P35G-1.5-14-C, Mattek). After 72 h, 200 μmol/L biotin was added to initiate the proCpepRUSH trafficking. Details of the plasmid cloning and virus generation are provided in the Supplementary Methods.

SC-islets and graft-containing kidneys were fixed in 4% paraformaldehyde and processed for paraffin embedding. Following antigen retrieval of 5-µm deparaffinized sections and blocking, primary antibodies were added at 4°C overnight, followed by 1-h incubation with the corresponding secondary antibody at room temperature. Details are provided in the Supplementary Methods.

Stained paraffin sections were imaged using Zeiss Axio Observer Z1 with Apotome. Confocal analysis was conducted using Leica Stellaris 8 FALCON/DLS inverted confocal microscope. For electron microscopy, a Jeol JEM-1400 transmission electron microscope was used. Sample preparation and image processing are detailed in the Supplementary Methods.

For static tests, 50 SC-islets were sequentially treated with different glucose concentrations and KCL. Dynamic tests were conducted using a perifusion apparatus, with samples collected every 4 min. Insulin, proinsulin, and glucagon levels from in vitro samples, and human-specific C-peptide from plasma samples, were measured using respective ELISA kits from Mercodia, Uppsala, Sweden. Details are provided in the Supplementary Methods.

Approximately 500 SC-islets were implanted under the kidney capsule of immunocompromised female NSG mice, as previously described (26). All experiments were approved by the Animal Welfare Committee of Southern Finland (ESAVI/9734/2021). Detailed steps are provided in the Supplementary Methods.

Engrafted mice were fasted for 4 h and then injected with 2 g glucose/kg intraperitoneally. Blood glucose levels were measured using a Contour XT glucometer, and blood samples were collected from the saphenous vein for measuring human C-peptide levels.

All data are presented as individual points with their mean ± SEM. Statistical significance evaluation was conducted by Welch and Brown-Forsythe ANOVA (with unpaired t test) using GraphPad Prism 9 software (GraphPad Software, La Jolla, CA).

The data generated and analyzed during this study are included in the published article and its online supplementary files. Additional data and resources are available from the corresponding authors upon request.

Our previous RNA sequencing data of pancreatic differentiation of the hESC line H1 indicated an abundant expression of IER3IP1 throughout the differentiation, with the highest expression at the final stage of functional SC-islets (S7) (Fig. 1A) (27). This coincides with the peak of expression of insulin, in line with its putative role in insulin biosynthesis. To study the role of IER3IP1 in human β-cell development, we generated two cell lines carrying different IER3IP1 mutations using CRISPR/Cas9-mediated genome editing (Fig. 1B). In the first line, we introduced the patient mutation IER3IP1^V21G (Supplementary Table 1) by targeting the first exon of IER3IP1 to create an isogenic homozygous knock-in IER3IP1 mutant model (IER3IP1^V21G). In the second line, we created a deletion in the first exon of IER3IP1, generating a homozygous KO model (IER3IP1^−/−). We confirmed the successful generation of the targeted mutations in the mutant lines by Sanger sequencing (Supplementary Fig. 1A, C, and D). The mutant clones expressed hallmark pluripotency markers, as shown by qPCR and immunocytochemistry, and showed no chromosomal abnormalities (Supplementary Fig. 1B and E).

To investigate the impact of IER3IP1 mutations on human β-cell development, we differentiated the knock-in and KO cell lines, alongside their wild-type (WT) counterpart, toward pancreatic islets using an optimized seven-stage differentiation protocol (Fig. 1C) (28). The three cell lines differentiated with similar high efficiency into definitive endoderm, assayed by the presence of the surface marker CXCR4 (∼95% CXCR4⁺ cells) (Supplementary Fig. 2A). Similarly, the three cell lines produced similar proportions of pancreatic progenitor cells coexpressing PDX1 and NKX6.1 (∼80% PDX1⁺/NKX6.1⁺ cells) at stage 4 (Supplementary Fig. 2B). At the final stage (S7, SC-islets), IER3IP1^−/− SC-islets contained a notably lower percentage of insulin-positive (INS⁺) cells compared with WT (28% vs. 43%) and higher percentage of glucagon-positive (GCG⁺) cells (49% vs. 34%), while IER3IP1^V21G SC-islets showed a similar but milder trend (40% INS⁺ cells and 40% GCG⁺ cells), as shown by immunohistochemistry (Fig. 1D) and flow cytometry (Fig. 1E and Supplementary Fig. 2C and D). Taken together, these data demonstrate that while IER3IP1 mutations did not affect the early development of endodermal and pancreatic progenitors, they altered the development of the endocrine populations at the final stage of differentiation, decreasing β-cell numbers.

To evaluate the consequences of IER3IP1 mutations on β-cell functionality, we measured dynamic insulin secretion using a perifusion assay. The three cell lines exhibited expected responses to high glucose, as well as the GLP-1 analog exendin-4, and KCl depolarization by increasing their insulin secretion (Fig. 1F). However, the absolute amount of insulin secreted by the IER3IP1^−/− SC-islets was severely reduced compared with their WT counterparts (Fig. 1F). This defect was further confirmed by a sequential static insulin secretion assay (Supplementary Fig. 3A and B). IER3IP1^−/− SC-islets showed drastically diminished insulin content (116 vs. 671 ng INS/µg DNA) (Fig. 1G), and significantly elevated GCG content (1.5-fold increase) (Fig. 1H) compared with WT. Comparably, IER3IP1^V21G SC-islets displayed a trend for decreased insulin and increased glucagon content (466 ng INS/µg DNA and 1.2-fold increase in GCG content) (Fig. 1G and H). Interestingly, IER3IP1^−/− SC-islets demonstrated a significantly higher ratio of proinsulin to insulin content (5 vs. 3.5), implying impaired insulin-processing capacity (Fig. 1I). Altogether, these data indicate that IER3IP1 mutations exert a deleterious influence on the insulin-processing capacity and intracellular content of β-cells in vitro, without abrogating their stimulus-secretion coupling.

To investigate the role of IER3IP1 in regulating ER homeostasis and trafficking in human β-cells (Fig. 2A), we quantified the proinsulin levels in SC-islets of the three genotypes. In agreement with the increased ratio of proinsulin to insulin content (Fig. 1I), IER3IP1^−/− β-cells showed significantly higher accumulation of proinsulin (21% vs. 7% PROINS^{HI +} INS⁺/INS⁺), while IER3IP1^V21G β-cells showed no difference compared with WT (Fig. 2B and C). Additionally, we examined the subcellular localization of proinsulin by confocal microscopy. In WT β-cells, proinsulin was mainly concentrated in the juxtanuclear space, while in the IER3IP1^−/− β-cells, proinsulin was diffusely present throughout the cell cytoplasm with significantly higher median intensity (Fig. 2D and Supplementary Fig. 3C). Interestingly, proinsulin exhibited a higher degree of localization to the ER in IER3IP1^−/− β-cells, as shown by protein disulfide isomerase (PDI) colocalization (an ER marker) (55% vs. 32% of proinsulin-PDI colocalized volume/total proinsulin volume) (Fig. 2D and E). Immunostaining of Golgi structures using the marker GM130 showed less localization of proinsulin to the Golgi apparatus of IER3IP1^−/− β-cells compared with WT (4% vs. 12% of proinsulin-GM130 colocalized volume/total proinsulin volume) (Fig. 2F and G). Thus, these results suggest an impaired trafficking of proinsulin in IER3IP1^−/− β-cells, whereby proinsulin is trapped in the ER.

To directly assess the role of IER3IP1 in proinsulin trafficking, we performed the RUSH assay by transducing SC-islets with lentivirus carrying a RUSH-proinsulin construct (Fig. 3A). The construct was designed to be specifically expressed in β-cells using a human insulin promoter driving the transcription of KDEL-tagged streptavidin and human preproinsulin C-peptide RUSH (ProCpepRUSH), containing a streptavidin-binding protein (SBP) and superfolder green fluorescent protein (sfGFP) inserted within the C-peptide region (25). Addition of biotin to the β-cells expressing this construct will mediate the release of GFP-labeled proinsulin (ProCpepRUSH) from the ER and allow its trafficking into the Golgi. Specific expression of the construct in β-cells was confirmed by immunostaining for the β-cell marker NKX6.1 (Supplementary Fig. 3D). Additionally, immunostaining for PDI showed localization of ProCpepRUSH to the ER network (Supplementary Fig. 3E). Live cell confocal imaging revealed the presence of ProCpepRUSH in the Golgi of WT β-cells within 12 min after biotin addition, with increasing GFP signal in the Golgi region over time, indicating the continuous ER-to-Golgi trafficking of ProCpepRUSH (Video 1). At 20 min postbiotin addition, ProCpepRUSH condensed into the Golgi, followed by the appearance of numerous ProCpepRUSH puncta, which remained proximal to the Golgi for additional 30 min. After 60 min of biotin addition, several ProCpepRush puncta were observed across the cell cytoplasm, with an increasing number of ProCpepRUSH puncta distinct from the Golgi by time. This indicates that ProCpepRUSH-containing vesicles had been released from the Golgi to the pool of insulin storage granules. In stark contrast, IER3IP1^−/− β-cells showed ProCpepRUSH in the Golgi only after 30 min postbiotin addition, with distinctly less accumulation of GFP signal in the Golgi region compared with WT. Even after 90 min of biotin addition, ProCpepRUSH was still clearly visible in the ER of IER3IP1^−/− β-cells with few ProCpepRUSH puncta proximal to the Golgi, demonstrating a severe trafficking defect in IER3IP1^−/− β-cells (Video 2 and Supplementary Videos 1 and 2).

To further validate these findings, we performed a biotin chase assay where we fixed the cells at six different time points after biotin addition (0, 15, 22, 30, 45, and 60 min) and immunoassayed for GM130 (Fig. 3B). Colocalization analysis confirmed the pronounced difference in the trafficking dynamics between WT, IER3IP1^V21G, and IER3IP1^−/− β-cells (Fig. 3C). At 15 min after biotin addition, >60% of ProCpepRUSH colocalized with GM130 in WT β-cells, indicating efficient ER-to-Golgi trafficking. In contrast, only 38% colocalization was observed in IER3IP1^V21G β-cells (P = 0.0056) exhibiting aberrant trafficking dynamics, and a mere 17% in IER3IP1^−/−β-cells (P = 0.0012) with an evident trafficking defect at all time points. Altogether, these findings demonstrate the profound impact of IER3IP1 mutations on ER-to-Golgi trafficking of proinsulin.

To understand the overall impact of IER3IP1 mutations on the transcriptome of human pancreatic cells, we performed RT-qPCR analysis on the generated SC-islets. The significant reduction of insulin transcript (INS) and elevation of glucagon transcript (GCG) are in line with the altered composition of the endocrine populations seen in IER3IP1^−/− SC-islets (Fig. 1E and Fig. 4A). Interestingly, IER3IP1^−/− SC-islets showed significant upregulation of several ER stress markers, such as HSPA5 (BiP), sXBP1, ATF6, and death protein 5 (DP5), while IER3IP1^V21G SC-islets did not exhibit such an upregulation (Fig. 4A). This indicates an activation of the IRE1α and ATF6 arms of the unfolded protein response (UPR), but not the PERK arm. We validated these findings using immunohistochemistry, which showed a considerable increase in the number of β-cells, but not α-cells (Supplementary Fig. 4A), expressing high levels of BiP in IER3IP1^−/− SC-islets (31% vs. 7% BiP^HI+ INS⁺/INS⁺) (Fig. 4B–C and Supplementary Fig. 4B). As an additional sign of ER stress, immunostaining for mesencephalic astrocyte–derived neurotrophic factor (MANF) showed increased expression in IER3IP1^−/− β-cells compared with WT (Supplementary Fig. 4C).

Close inspection of the ER ultrastructure using electron microscopy showed a markedly dilated and distorted ER structure indicating elevated ER stress in IER3IP1^−/− β-cells (Fig. 4D). A morphometric analysis of ER volume confirmed that in comparison with WT and IER3IP1^V21G β-cells, the mean cytoplasmic volume fraction occupied by ER was increased by 85% in IER3IP^−/− β-cells (Fig. 4E). There is no generally accepted quantitative criterion for ER stress in electron microscopy sections; however, when the sections from individual β-cells were scanned for ER width of >100 nm, such ER elements were found in 33% of IER3IP^−/− β-cells (n = 190 cells). In contrast, ER profiles of this width were found in 12% of WT β-cells and 15% of IER3IP1^V21G β-cells (n = 200 and 240 cells, respectively) (Supplementary Fig. 4D).

To further investigate whether reducing ER stress could alleviate the observed trafficking defects (Fig. 3B), we treated IER3IP1^−/− β-cells with tauroursodeoxycholic acid (TUDCA) for 48 h before performing the RUSH assay. TUDCA is a chemical chaperone that has been shown to facilitate protein folding and alleviate ER stress (29). Despite this treatment, no improvement in the colocalization of ProCpepRUSH with GM130 was observed at either 15 or 30 min postbiotin addition (Supplementary Fig. 3F). This suggests that the defect in IER3IP1^−/− β-cells primarily arises from impaired proinsulin trafficking.

Next, we treated SC-islets with chemical ER stress inducers and quantified the percentage of apoptotic β-cells using TUNEL assays. In the vehicle-treated control condition, both IER3IP1^−/− and IER3IP1^V21G β-cells showed a trend of increased β-cell death compared with WT (Supplementary Fig. 4F). Upon the induction of ER stress using TM or TG, the percentage of apoptotic β-cells increased in IER3IP1^−/− SC-islets (6% vs. 3% with TM, and 5% vs. 1% with TG), while no marked change was observed in IER3IP^V21G SC-islets (2% with TM, and 1% with TG) (Supplementary Fig. 4E and F). These findings indicate that loss of IER3IP1 leads to elevated ER stress, altered ER ultrastructure, and heightened β-cell susceptibility to ER stress-induced apoptosis.

To further examine the phenotypic defects caused by IER3IP1 mutations in vivo, we implanted WT, IER3IP^−/−, and IER3IP^V21G SC-islets under the kidney capsule of immunocompromised mice (Fig. 5A). We measured human C-peptide as a proxy for human insulin secretion from the implanted SC-islets. After 2 weeks of implantation, we observed an evident reduction in the amount of circulating human C-peptide in mice implanted with IER3IP1^−/− cells compared with WT (27 ± 14 vs. 136 ± 29 pmol/L). This difference became even more pronounced during the following 2 months, as IER3IP1^−/− grafts were unable to increase their C-peptide output to a magnitude similar to that of their WT counterparts (286 ± 134 vs. 1,801 ± 198 pmol/L at 3 months). Interestingly, IER3IP1^V21G grafts also presented diminished function already at 1 month postimplantation (300 ± 73 vs. 688 ± 147 pmol/L) (Fig. 5B). At 3 months, the difference in the human C-peptide secretion was evident between IER3IP1^V21G and WT grafts (892 ± 125 vs. 1,801 ± 199 pmol/L). These differences in graft function across the three genotypes were also reflected on blood glucose levels. Only the mice engrafted with WT cells showed blood glucose values corresponding to the human normoglycemic level of 4 mmol/L, while the mice implanted with IER3IP1^−/− or IER3IP1^V21G SC-islets remained at the murine level of 8 mmol/L (Fig. 5C).

To confirm the functional defects observed in the IER3IP1 mutant grafts, we subjected the mice to an intraperitoneal glucose tolerance test 3 months postimplantation. In response to the injected high dose of glucose, the WT grafts transiently increased their human C-peptide secretion, while IER3IP1^V21G grafts presented a drastically attenuated response and IER3IP1^−/− grafts displayed no response (Fig. 5D and E). This was coupled with a rapid blood glucose clearance in the mice engrafted with WT cells, whereas a significantly lower glucose clearance rate was observed in the IER3IP1^−/− or IER3IP1^V21G grafted mice (Fig. 5F and G).

At 3 months postimplantation, immunohistochemistry analyses of the retrieved grafts revealed a significant difference in the proportions of β- and α-cells between WT and IER3IP1^−/− grafts. In WT grafts, β-cells were the dominant cell type, and α-cells accounted for a smaller proportion (70% INS⁺ cells and 29% GCG⁺ cells). This was in striking contrast to the IER3IP1^−/− grafts, where α-cells made up the largest proportion of hormone-positive cells (28% INS⁺ cells and 71% GCG⁺ cells) (Fig. 6A and B). Although IER3IP1^V21G grafts displayed hormone-cell composition similar to that of the WT grafts (65% INS⁺ cells and 34% GCG⁺ cells) (Fig. 6A and B), their β-cells showed substantially higher levels of proinsulin accumulation (30% PROINS^HI+ INS⁺/INS⁺ vs. 3% in WT) (Fig. 6C and D). Similarly, IER3IP1^−/− grafts showed considerably higher accumulation of proinsulin in their β-cells compared with WT (10% PROINS^HI+ INS⁺/INS⁺) (Fig. 6C and D). Immunostaining for the ER stress marker BiP revealed markedly higher levels of ER stress in IER3IP1^V21G and IER3IP1^−/− β-cells compared with their WT counterparts (24% vs. 51% vs. 2% BiP^HI+ INS⁺/INS⁺, respectively) (Fig. 6E and F). Collectively, these data confirm the deleterious impact of IER3IP1 mutations on the functionality of human β-cells and demonstrate the efficiency of the in vivo model to reveal the disease phenotype of IER3IP1^V21G β-cells.

In the current study, we established a functional human β-cell model to determine the precise role of IER3IP1 in β-cell development and physiology. Our results demonstrate the crucial need of IER3IP1 for the proper ER-to-Golgi trafficking of proinsulin and for human β-cell function and survival. Impaired proinsulin trafficking in IER3IP1 mutant β-cells results in distorted ER microstructure and elevated ER stress, leading to altered functionality, especially upon implantation into mice.

Biallelic mutations in IER3IP1 have been reported in 10 patients from 8 families, all diagnosed with microcephaly, epilepsy, and early-onset permanent diabetes syndrome 1 (Supplementary Table 1). These clinical characteristics overlap with those reported in patients with mutations in other genes impacting ER homeostasis and UPR response (30,31). Notably, early-onset diabetes and neurological disorders have been associated with pathogenic mutations in MANF (32), YIPF5 (12), EIF2AK3 (6), WFS1 (33,34), DNAJC3 (35,36), and PPP1R15B (37), highlighting the similarities between β-cells and neurons in their indispensable mechanisms for coping with their high protein synthesis demand while ensuring survival.

Patients with IER3IP1 mutations develop insulin-dependent diabetes (15). Postmortem examination of their pancreata showed reduced islet size, accompanied by fewer insulin-positive cells (17). Furthermore, β-cell–specific deletion of IER3IP1 in mice resulted in severe insulin-deficient diabetes, associated with decreased islet size and β-cell mass and reduced insulin content (22). These findings align with our results, demonstrating a markedly reduced β-cell number and insulin content in IER3IP1^−/− SC-islets, despite preserving their stimulatory index to glucose stimulation in vitro. These results indicate that lack of IER3IP1 does not completely abolish insulin secretion, but rather plays a crucial role in insulin production.

Secretory proteins are transferred from the ER to the Golgi via COPII-coated vesicles, either through receptor-mediated sorting at ER exit sites or nonselective bulk flow (38). Recent studies have shown that COPII-mediated transport is indispensable for proinsulin trafficking, the predominant soluble cargo in β-cells (10). Additionally, SURF4, which localizes to the ER exit sites, was identified as a sorting receptor essential for the efficient export of proinsulin from the ER (11). Alteration of COPII or SURF4 function in β-cells impairs proinsulin trafficking, causing its accumulation in the ER (10,11). Similarly, in our human β-cell model, lack of IER3IP1, which also localizes to the ER exit sites (20), leads to increased retention of proinsulin in the ER. Using the RUSH assay, we demonstrated that the proinsulin accumulation is due to defective ER-to-Golgi trafficking in IER3IP1 mutant β-cells. This demonstrates that IER3IP1 is essential for the ER-to-Golgi trafficking of proinsulin and might therefore play a role in proinsulin sorting at the ER exit sites, with further studies needed to confirm its interaction with proinsulin.

Numerous studies have linked defective ER-to-Golgi trafficking to induced ER stress and β-cell failure (6–9). Impaired ER-to-Golgi transport leads to accumulation of unfolded protein in the ER, rendering β-cells more susceptible to ER stress-induced failure. Moreover, retention of unfolded proinsulin in the ER disrupts the trafficking of properly folded proinsulin leading to decreased insulin synthesis (39), a key factor in the pathogenesis of both type 1 and type 2 diabetes (40,41). In our model, IER3IP1 deletion led to increased proinsulin accumulation in the ER of β-cells and elevated ER stress. Furthermore, IER3IP1^−/− β-cells displayed severely dilated and altered ER structures, which highlights the essential role of IER3IP1 in preserving ER morphology and homeostasis in functional β-cells.

To recapitulate the patient genotype, we generated SC-islets harboring the pathogenic variant Val21Gly, located in one of the IER3IP1 transmembrane domains, which might disturb its ER localization. Compared with IER3IP1^−/−, IER3IP1^V21G β-cells exhibited less impaired ER-to-Golgi trafficking and milder ER stress, suggesting that the V21G mutation results in a hypomorphic IER3IP1 that does not impair β-cell functionality to the same degree as the complete absence of IER3IP1 in vitro. However, the failure of IER3IP1^V21G β-cells after implantation indicates that the mutation significantly compromises β-cell functionality in vivo. IER3IP1^V21G grafts exhibited a noteworthy, albeit less pronounced, functional impairment compared with IER3IP1^−/− grafts. This demonstrates the advantages of using stem cell models to capture the wide spectrum of ER stress-induced defects caused by various degrees of mutation severity, where increased demand on the β-cells upon implantations enables the recapitulation of the disease phenotype, as it has previously been shown for INS (42), YIPF5 (12), and MANF (32) mutations.

It is worth noting the apparent absence of ER stress in α-cells, despite the disruption of IER3IP1 function in all cells. Immunostaining of the human pancreas showed that IER3IP1 is expressed only in the β-cells (22). This might indicate a cargo-specific function of IER3IP1 toward insulin trafficking. Additionally, it may infer a heightened sensitivity of β-cells to defects impacting protein synthesis and trafficking compared with α-cells. In comparison with β-cells, α-cells have indeed been reported to express higher levels of HSPA5, encoding BiP, and the antiapoptotic gene BCL2L, together with decreased expression of the proapoptotic gene CHOP (43). This could render α-cells more resistant to ER stress-induced cell death.

In conclusion, our study illustrates the crucial role of IER3IP1 in proinsulin trafficking and ER homeostasis, emphasizing the necessity of efficient ER-to-Golgi trafficking in sustaining β-cell function and survival. We further demonstrate that SC-islet modeling, combined with CRISPR-Cas9 technology, is a powerful tool to dissect diabetes pathologies stemming from trafficking defects. The RUSH assay, combined with high-resolution imaging, offers new avenues to better understand the ER-to-Golgi trafficking dynamics. Such models offer a useful platform for developing therapeutic interventions aimed at enhancing proinsulin trafficking, alleviating ER stress, and preventing β-cell failure, thereby improving diabetes management.

Acknowledgments. The authors thank Anna Ahmala, Nea Asumaa, Heli Grym, and Jarkko Ustinov, Stem Cells and Metabolism Research Program, Faculty of Medicine, University of Helsinki, Helsinki, Finland, for their expert technical assistance and also thank Mirabai Cuenca, Centre for Genomic Regulation (CRG), The Barcelona Institute of Science and Technology, Barcelona, Spain, for sharing her expertise in lentivirus production and transduction of SC-islets. The authors thank the Institute of Biotechnology Electron Microscopy Unit, University of Helsinki, for providing laboratory facilities. Confocal imaging was performed at the Biomedicum Imaging Unit, University of Helsinki, supported by the Helsinki Institute of Life Science (HiLIFE) and Biocenter Finland.

Funding. T.O. received funding provided by The Academy of Finland (grant 297466) and as its Center of Excellence (MetaStem, grant 312437), the Novo Nordisk Foundation, and the Sigrid Juselius Foundation. H.M. received funding provided by the Doctoral Program in Integrative Life Science at University of Helsinki, Biomedicum Helsinki Foundation, Orion Research Foundation, Kyllikki and Uolevi Lehikoinen Foundation, the Diabetes Research Foundation, and the Maud Kuistila Memorial Foundation. N.B. received funding provided by Finska Läkaresällskapet and the Perklén Foundation.

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

Author Contributions. H.M. conceived and conceptualized the study, established the H1 hESCs lines, performed the differentiation experiments, acquired and analyzed the data, and wrote the original manuscript. H.M. and J.S.-V. performed and analyzed animal experiments. S.L. assisted in performing the differentiation experiments and in data acquisition and analysis. E.V. and N.B. acquired and analyzed the electron microscopic images. A.G. assisted in the electron microscopic analysis. S.E. assisted in the genome editing of the hESCs, performed RNA extraction, cDNA synthesis, and qPCR experiments, and performed ELISA. H.I., V.L., T.B., and D.B. participated in manuscript writing. S.B.S. designed and provided the ProCpepRUSH construct. T.O. supervised the study, provided resources, acquired funding, and participated in manuscript writing. T.O is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Preliminary results from this study were presented as an abstract at the 58th Annual Meeting of the European Association for the Study of Diabetes, Stockholm, Sweden, 19–23 September 2022.