Proinsulin folding and trafficking defects trigger a common pathological disturbance of endoplasmic reticulum homeostasis

Abstract

Primary defects in folding of mutant proinsulin can cause dominant‐negative proinsulin accumulation in the endoplasmic reticulum (ER), impaired anterograde proinsulin trafficking, perturbed ER homeostasis, diminished insulin production, and β‐cell dysfunction. Conversely, if primary impairment of ER‐to‐Golgi trafficking (which also perturbs ER homeostasis) drives misfolding of nonmutant proinsulin—this might suggest bi‐directional entry into a common pathological phenotype (proinsulin misfolding, perturbed ER homeostasis, and deficient ER export of proinsulin) that can culminate in diminished insulin storage and diabetes. Here, we've challenged β‐cells with conditions that impair ER‐to‐Golgi trafficking, and devised an accurate means to assess the relative abundance of distinct folded/misfolded forms of proinsulin using a novel nonreducing SDS‐PAGE/immunoblotting protocol. We confirm abundant proinsulin misfolding upon introduction of a diabetogenic INS mutation, or in the islets of db/db mice. Whereas blockade of proinsulin trafficking in Golgi/post‐Golgi compartments results in intracellular accumulation of properly‐folded proinsulin (bearing native disulfide bonds), impairment of ER‐to‐Golgi trafficking (regardless whether such impairment is achieved by genetic or pharmacologic means) results in decreased native proinsulin with more misfolded proinsulin. Remarkably, reversible ER‐to‐Golgi transport defects (such as treatment with brefeldin A or cellular energy depletion) upon reversal quickly restore the ER folding environment, resulting in the disappearance of pre‐existing misfolded proinsulin while preserving proinsulin bearing native disulfide bonds. Thus, proper homeostatic balance of ER‐to‐Golgi trafficking is linked to a more favorable proinsulin folding (as well as trafficking) outcome.

1. INTRODUCTION

In the syndrome of Mutant INS Gene‐Induced Diabetes of Youth (MIDY), dominant mutations impairing the folding of proinsulin predispose to insulin‐deficient diabetes (Liu et al., 2010). Even in the islets of mice that express two WT alleles of Ins1 as well as one WT allele of Ins2, proinsulin misfolding from the single mutant allele of Ins2‐Akita is sufficient to bring about diabetes in all male animals, accompanied by dilation of the endoplasmic reticulum (ER) with failed anterograde trafficking of both mutant and WT proinsulin (Liu et al., 2010). Protein cargo conveyed from ER‐to‐Golgi via the classical secretory protein trafficking pathway must first meet the requirements needed to pass ER quality control (Ellgaard & Helenius, 2003). As per the original definition of ER quality control, some of the key factors in meeting those requirements include the speed of substrate protein maturation (Kumari & Brodsky, 2021), substrate solubility versus aggregation/insolubility (Poothong et al., 2021), as well as competing recognition/retention by ER molecular chaperones/oxidoreductases (Adams et al., 2019; Jang et al., ²⁰¹⁹) versus recognition by cargo receptors for ER export (Saegusa et al., 2022). There is evidence that in MIDY, structurally‐defective proinsulin forms an increased amount of aberrant disulfide‐linked complexes in the ER (Haataja et al., 2021) that engage both mutant and WT proinsulin (Sun et al., 2020), limiting its anterograde trafficking. However, most forms of diabetes do not involve INS gene mutations but rather exhibit polygenic susceptibility, with various patients inheriting common as well as rare genetic variants (Dornbos et al., 2022).

A significant fraction of risk alleles linked (to varying degrees) to glycemic control involve genes expressed in islet β‐cells (Vinuela et al., 2020). Beyond the tripartite ER stress sensor proteins, an interesting subset of these genes encode proteins that contribute to ER homeostasis and protein trafficking in the ER‐Golgi system (including genes like HYOU1, FKBP2, WFS1, PREB, TFG, SEC24A, SEC24B, SEC16A, SEC16B, ARAP1, SAR1B, SEC24A, KDELR2, as well as others as reflected in the type 2 diabetes [T2D] genetics portal [https://t2d.hugeamp.org/] and elsewhere [Hoefner et al., ²⁰²³]). We hypothesize that outbound ER protein trafficking is needed not only to convey proinsulin to the distal secretory pathway (where insulin is made [Liu et al., ²⁰²¹]) but also, continuous trafficking represents part of the homeostatic maintenance of the ER itself (Fang et al., 2015), which is responsible for the proinsulin folding environment (Supplemental Figure S1). For example, it has recently been reported that whereas the primary phenotype of Sar1 deficiency in β‐cells (and all cells) is impairment of ER‐to‐Golgi traffic, such deficiency also leads to substantial proinsulin misfolding (accompanied by ER stress; Zhu et al., ²⁰¹⁹).

Although the relationships between proinsulin folding and intracellular protein trafficking have not yet been studied systematically, there are important hints suggesting that proinsulin misfolding may serve not only as a primary driver of dysfunctional β‐cell protein trafficking, but also may occur as a consequence of impaired or imbalanced protein trafficking in the secretory pathway (Preston et al., 2009; Zhu et al., ²⁰¹⁹). It is this relationship that we seek to explore further in the current study (Supplemental Figure S1), using simple modifications of a protocol of proinsulin immunoblotting after nonreducing SDS‐PAGE—a tool that is easily available and can be replicated in any laboratory. Specific immunoblotting of proinsulin (as opposed to insulin) relies upon features of its C‐peptide and/or cleavage sites: antibodies must bind either to a region at or near the B‐C junction, the middle of the C‐peptide, or the C‐A junction. Interestingly, despite primary sequence differences, the C‐peptides of different species exhibit many common (and conserved) features (Landreh et al., 2014) that have co‐evolved with insulin (Wang et al., 2012). Specifically, the N‐terminal acidic portion of the C‐peptide can, under different conditions, alternately exhibit α‐helical or β‐sheet features (Lind et al., 2010), and this region contributes importantly to proinsulin folding (Chen et al., 2002) and proinsulin solubility (Dodson & Steiner, 1998). Additionally, the mid‐portion of C‐peptide includes a conserved nonpolar region, whereas the carboxyl‐terminal sequence of the C‐peptide is able to assume an α‐helical (Yang et al., 2010) or unstructured conformation (Henriksson et al., 2005). Intact proinsulin constrains the N‐ and C‐terminal C‐peptide ends, which are tethered to insulin prior to endoproteolytic cleavage (Landreh et al., 2014). Moreover, a relationship between C‐peptide structure and the disulfide bonds of the insulin moiety has been described (Min et al., 2004). In the current report, recognizing the ability of the proinsulin C‐peptide to assume distinct structures, we have found a means to properly quantify the relative abundance of distinct folded and misfolded forms of proinsulin in the ER by “forcing” all SDS‐denatured proinsulin molecules (after nonreducing SDS‐PAGE) to assume a single, common conformation yielding consistent antibody recognition. This has allowed us to develop a quantitative assessment of the impact of defects in ER homeostasis on the proinsulin folding environment, and the ability to improve folding outcomes upon reversal of such defects.

2. RESULTS

2.1. Quantitative detection of properly and improperly folded proinsulin

Some proinsulin molecules in the ER achieve native intramolecular disulfide bonding, whereas others may exhibit either non‐native intramolecular disulfide bonding (misfolded monomers) or intermolecular disulfide bonded dimers and higher‐order complexes (Arunagiri et al., 2019)—each of which can impact C‐peptide structure (Henriksson et al., 2005; Lind et al., ²⁰¹⁰; Yang et al., ²⁰¹⁰). These differently folded species (bearing different possible disulfide combinations) are resolved by nonreducing SDS‐PAGE and can be detected by immunoblotting with proinsulin‐specific antibodies that react with different regions of the C‐peptide. We discovered that post‐gel disulfide reduction (incubation of the gel with DTT and heating, Figure 1a) “forces” all proinsulin forms to become reduced proinsulin monomers prior to electrotransfer to nitrocellulose for immunoblotting (cartoon in Figure 1b). Whereas anti‐insulin antibodies (partially cross‐reacting with proinsulin) actually lose affinity for insulin or proinsulin when their disulfide bonds are broken (Figure 1c lanes 1 → 3, dashed red arrows), recognition of native disulfide‐bonded proinsulin monomers by proinsulin‐specific antibodies is markedly increased after post‐gel disulfide reduction to convert proinsulin to a fully reduced species (with 100 mM DTT and heating; cartoon in Figure 1b; demonstrated in Figure 1c lanes 5–16, curved blue arrows). Importantly, this was observed for anti‐proinsulin mAbs binding at or near the B‐C junction (Figure 1c lanes 5 + 7), mid‐portion of C‐peptide (lanes 9 + 11), or the C‐A junction (lanes 13 + 15). Thus, post‐gel disulfide reduction does not represent a quirk of a particular antibody but rather reflects a physical change in the accessibility of the C‐peptide within the proinsulin structure.

FIGURE 1.

After nonreducing SDS‐PAGE, post‐gel disulfide reduction improves detection of proinsulin in distinct folded states. (a) Schematic of different immunoblotting methods to detect proinsulin misfolding. At left: Conventional immunoblotting after nonreducing SDS‐PAGE is super‐sensitive to misfolded disulfide‐linked complexes of proinsulin. Red arrow pathway: Treatment of samples with DTT + heating after nonreducing SDS‐PAGE converts all proinsulin contained within the gel to reduced monomers. Using a fixed percentage of acrylamide ensures even transfer efficiency across the gel. This method enhances detection of proinsulin monomers (particularly native monomers) after nonreducing SDS‐PAGE and provides a more quantitative estimate of the distribution of differently‐folded proinsulin species. (b) Cartoon highlighting the impact of the blotting method described in panel A for pancreatic β‐cells after nonreducing SDS‐PAGE. (c) Immunblotting with anti‐insulin or with monoclonal antibodies (generated by Hytest) targeting different regions of rodent proinsulin C‐peptide (A cartoon above shows the location of antibody binding on the proinsulin C‐peptide). Identical samples of INS1E cell lysates were resolved by SDS‐PAGE under nonreducing (lanes 1, 3, 5, 7, 9, 11, 13, and 15) or reducing conditions (lanes 2, 4, 6, 8, 10, 12, 14, and 16). After running, the gels were divided, and lanes 3,4; 7,8; 11,12; and 15,16 were treated with 100 mM DTT + heating for post‐gel disulfide reduction (indicated below as – or +) before electrotransfer and immunoblotting with the indicated antibodies. Immunoblotting with anti‐insulin that cross‐reacts with proinsulin show dramatic signal loss (for both proinsulin and insulin, dotted red lines) upon post‐gel disulfide reduction. The curved blue arrows connecting lanes 5–7, 9–11, or 13–15 bearing samples that were run under nonreduced conditions show dramatically improved detection of proinsulin monomers by post‐gel disulfide reduction. (d) 293 T cells were transfected with proinsulin “keep one bond” constructs—keep‐B19/A20, keep‐B7/A7, or keep‐A6/A11. After 24 h, the cells were lysed and identical aliquots of cell lysate were resolved by SDS‐PAGE (12% NuPAGE) under nonreduced (first 6 lanes) or reduced conditions (last three lanes). After electrophoresis, a portion of the nonreduced gel underwent post‐gel disulfide reduction with 100 mM DTT + heating to 60°C for 15 min (middle three lanes) prior to electrotransfer and immunoblotting with anti‐human proinsulin. The detection of proinsulin monomers by nonreducing SDS‐PAGE increased after post‐gel disulfide reduction with DTT + heating (curved blue arrows). (e) Immunoblots were quantified by densitometry; the relative recovery of proinsulin monomers before and after post‐gel disulfide reduction is shown (n = 4 independent experiments; mean ± SD; *p < 0.05; ns = non‐significant). (f) 293 T cells were either untransfected (“U”) or transfected to express WT hPro‐CpepMyc or that bearing the L(A16)P mutation (here simply labeled as A16P), and culture media was collected overnight. The samples were resolved by 12% NuPAGE under nonreduced (first 5 lanes) or reduced conditions (last 5 lanes) and the nonreduced gel underwent post‐gel disulfide reduction (indicated at bottom) before electrotransfer and immunoblotting with anti‐human proinsulin. Cyclophilin B (CypB) is a loading control. (g) Quantitation of native proinsulin monomers recovered as a fraction of total proinsulin (n = 4 independent experiments; mean ± SD; **p = 0.0022).

To understand which of the native disulfide bonds may limit the proinsulin C‐peptide accessibility under immunoblotting conditions, we performed experiments using three artificially‐engineered proinsulin constructs (keep‐B19/A20; keep‐B7/A7; keep‐A6/A11) that each bear two Cys residues capable of forming only one of the native disulfide bonds of proinsulin (Haataja et al., 2016). An increase in detection of proinsulin monomers was observed upon post‐gel disulfide reduction for keep‐B19/A20 or keep‐B7/A7 (Figure 1d, e) indicating that presence of B‐to‐A‐chain native disulfide bonds limit proinsulin C‐peptide epitope exposure even after denaturation for SDS‐PAGE. In contrast, we could obtain no evidence that the intrachain A6‐A11 disulfide bond significantly affects detection with proinsulin‐specific antibodies (Figure 1d, e).

After nonreducing SDS‐PAGE of purified recombinant human proinsulin expressed in Escherichia coli (N‐terminally 6xHis‐tagged, purchased from Creative Biomart, Shirley, NY; with no attempt for in vitro refolding to the native state) increased detection of monomers, with decreased detection of intermolecular disulfide‐linked forms, was observed upon post‐gel disulfide reduction (Supplemental Figure S2A, quantified in S2B ). Further, human proinsulin epitope‐tagged within the middle of the C‐peptide (hPro‐CpepMyc), when secreted from transfected 293 T cells, was nearly undetectable by conventional immunoblotting after SDS‐PAGE under nonreducing (NR) conditions (despite being detectable under reducing [R] conditions, Supplemental Figure S2C lanes 3 vs. 4). After nonreducing SDS‐PAGE, post‐gel disulfide reduction revealed the secretion of myc‐tagged human proinsulin (Supplemental Figure S2C lane 7); moreover, the intracellular monomers of hPro‐CpepMyc appeared as two distinct bands with the faster‐migrating monomer corresponding to that released into the medium (Supplemental Figure S2C curved blue arrow) and a slower‐migrating component that was neither native proinsulin nor fully reduced proinsulin (Figure S2C lane 5, red arrow), indicating partially oxidized, non‐native hPro‐CpepMyc monomers.

The native disulfide isomer of hPro‐CpepMyc was particularly well‐resolved from non‐native monomers, which were retained intracellularly (Figure 1f). The myc‐tagged wild‐type proinsulin formed native monomers that are secretion‐competent (Figure 1f blue curved arrow) as well as non‐native monomers that are secretion‐incompetent. In contrast, a representative MIDY mutant L(A16)P exhibited exclusively non‐native proinsulin monomers (Figure 1f red arrows) that were incompetent both in forming native disulfide isomers (Figure 1g) and in secretion (Figure 1f). Co‐expression of active Ero1, an ER oxidase that enhances the folding of WT proinsulin (Wright et al., 2013), increased the formation of native hPro‐CpepMyc, leading to its improved secretion (Supplemental Figure S2D solid blue arrow; also seen from the reducing gel, and quantified in Supplemental Figure S2E).

Together, the foregoing data establish that (1) the native B‐A‐chain disulfide bonds of insulin inhibit detection of proinsulin monomers (either rodent, human, or epitope‐tagged) by immunoblotting; (2) intracellular proinsulin monomers include both native and non‐native disulfide isomers that can both be detected by simple methodological modifications; and (3) essentially only the native form can be secreted. This improved detection is apparent for purified recombinant human proinsulin (Supplemental Figure S2A) to endogenous rodent proinsulin expressed and secreted from culture Min6 or INS1 β‐cell lines (e.g., Figure 2a), all the way to human islets (see below). The essential methodological modification involves post‐gel disulfide reduction: by turning all proinsulin into reduced monomers regardless of their initial gel mobility, proinsulin antibodies can then recognize all species of proinsulin equally, after‐the‐fact.

FIGURE 2.

Folding status of endogenous (intracellular and secreted) proinsulin from Min6 cells. (a) Identical aliquots of Min6 cell lysate and 6 h conditioned media were resolved by nonreducing 12% NuPAGE. After electrophoresis, the gel was cut in fourths (dotted lines) and each quarter treated as indicated (below the image) prior to electrotransfer to nitrocellulose for immunoblotting with anti‐rodent proinsulin. Improved detection of secreted proinsulin (blue arrow, as well as intracellular proinsulin monomer bearing native disulfide bonds) is noted after post‐gel disulfide reduction (optimized with DTT + heating). (b) Min6 cells in complete medium were incubated for 4 h at 37 or 29°C, as indicated. Secretion of proinsulin (Proins) from cells to media is shown at both temperatures (curved blue arrows; solid or dashed, respectively). The non‐native proinsulin monomer band is identified with a red arrow. Cyclophilin B (Cyp B) is a loading control. (c) Quantitation of native proinsulin intracellularly from replicate experiments like those shown in panel B as a fraction of total (cells + media) proinsulin (n = 3 independent experiments; mean ± SD; *p = 0.045). Each point represents an independent experiment (mean ± SD). (d) Min6 cells were either untreated or treated with Monensin (15 μM for 90 min) as indicated. Secretion of proinsulin (Proins) from cells to media under both conditions is indicated (curved blue arrows; solid or dashed, respectively). The non‐native proinsulin monomer band is identified with a red arrow. Cyclophilin B (Cyp B) is a loading control. In panels 2a and 2d, nonspecific bands in the media migrating at >50 kDa are largely derived from serum albumin, observed in similar blotting exposures of culture medium that was never incubated with cells. (e) Quantitation of native proinsulin intracellularly from replicate experiments like those shown in panel D as a fraction of total (cells + media) proinsulin (n = 4 independent experiments; mean ± SD; *p = 0.012).

2.2. Impaired intracellular trafficking in the distal secretory pathway does not trigger proinsulin misfolding

Insulin biogenesis requires export of proinsulin from ER‐to‐Golgi (early secretory pathway), followed by Golgi‐to‐secretory granule trafficking (distal secretory pathway) for insulin storage and release (Liu et al., 2021). Some unprocessed proinsulin is also released from β‐cells, and this is comprised essentially exclusively of the native disulfide isomer (Figure 2a). Blockade of anterograde proinsulin trafficking potentially might trigger proinsulin misfolding in the early secretory pathway, the distal secretory pathway, or both. To test this, we first examined the impact of inhibition of protein trafficking in the distal secretory pathway on the folded state of proinsulin intracellularly. Cooling of cells to temperatures greater than 20°C but less than 37°C inhibits Golgi/post‐Golgi protein trafficking and exocytosis (Matlin & Simons, 1983; Renstrom et al., ¹⁹⁹⁶). Lowering the prevailing incubation temperature to 29°C for 4 h decreased proinsulin release from Min6 β‐cells to the medium while increasing by ~50% native proinsulin monomers intracellularly (Figure 2b dashed blue arrow; quantified in Figure 2c). Second, treatment of cells with the carboxylic ionophore, monensin, which selectively impairs anterograde protein trafficking within the Golgi complex (Tartakoff, 1983), also blocked proinsulin secretion while increasing native proinsulin monomers intracellularly (Figure 2d dashed blue arrow; quantified in Figure 2e). (Independently, we noted that adding a DTT‐plus‐heating step enhanced immunofluorescence detection of endogenous proinsulin [Supplemental Figure S3A] in the juxtanuclear Golgi region [Supplemental Figure S3B]; n = 3 experiments). Altogether, these data support that proinsulin residing in the distal secretory pathway is limited to the native disulfide isomer and is not converted to misfolded proinsulin upon impairment of distal secretory pathway protein trafficking.

2.3. Proinsulin folding responds to changes in the ER environment

We previously reported that misfolded proinsulin can enter aberrant disulfide‐linked complexes (Arunagiri et al., 2019). By repeating the antigenic epitope within ‘concatamers’, these higher molecular weight forms can increase antibody avidity and thereby increase immunoreactive signal strength. If all proinsulin is first converted to fully‐reduced monomers, then equal immunoblotting efficiency is achieved for all proinsulin forms (particularly when using a fixed percentage acrylamide for identical electrotransfer efficiency across the entire gel)—thus it is not surprising that post‐gel disulfide reduction actually decreased the intensity of proinsulin‐containing bands at higher molecular weights (Figure 2a lane 7 vs. 1). With this in mind, we explored the fact that ER protein folding and anterograde trafficking are energy‐requiring events that are blocked by inhibitors of mitochondrial respiration such as antimycin A (Jamieson & Palade, 1968)—which inhibits electron transport between cytochromes b and c to impair ATP generation while contributing to ROS production. Antimycin A treatment of Min6 cells not only inhibited proinsulin secretion, it also resulted in a dramatic intracellular decrease of proinsulin bearing native disulfide bonds (Figure 3a lane 6 blue arrow, quantified in 3b). Although it previously appeared that most proinsulin molecules formed very high molecular weight disulfide‐linked complexes (Figure 3a lane 2), new methodology establishes that the major mispaired disulfide‐linked species are actually dimers and misfolded monomers (Figure 3a lane 6 red arrows). FCCP, which uncouples the inner mitochondrial membrane (blocking ATP generation) but contributes less to ROS production, also limited proinsulin export while decreasing intracellular proinsulin bearing native disulfide bonds (Supplemental Figure S4). These data do not exclude a contribution by ROS to proinsulin misfolding, but strongly suggest that mitochondrial energy supply supports proinsulin folding in the ER, in parallel with its known importance in anterograde protein trafficking in the secretory pathway.

FIGURE 3.

Proinsulin folding response to pharmacological perturbation of ER homeostasis. (a) Identical wells of Min6 cells were treated ± Antimycin A (6 μM) for 1.5 h, and equal aliquots of cell lysate or media were resolved by 12% NuPAGE under nonreducing (lanes 1–8) or reducing conditions (lanes 9–12). Identical nonreduced samples were either not exposed (lanes 1–4) or exposed (lanes 5–8) to post‐gel disulfide reductions (as indicated below) prior to electrotransfer and immunoblotting with anti‐rodent proinsulin (Proins). Non‐native proinsulin monomers and disulfide‐linked proinsulin dimers are indicated (red arrows); native proinsulin monomers migrate faster (blue arrow). (b) Quantitation of native proinsulin monomers recovered as a fraction of total proinsulin (n = 3 independent experiments; mean ± SD; *p = 0.0045). (c) Identical wells of Min6 cells were untreated or treated with PERK inhibitor (2 μM GSK2656157, overnight, as indicated above), and equal aliquots of cells (C) or media (M) were resolved by 12% NuPAGE under nonreducing (lanes 1–8) or reducing (lanes 9–16) conditions. Identical nonreduced samples were either exposed (lanes 1–4) or not exposed (lanes 5–8) to post‐gel disulfide reduction (as indicated below) prior to electrotransfer and immunoblotting with anti‐rodent proinsulin (Proins). Proinsulin monomers bearing native (blue arrow) or non‐native (red arrow) disulfide bonds are indicated. (d) Quantitation of native proinsulin monomers recovered as a fraction of total proinsulin (n = 5 independent experiments; mean ± SD; *p = 0.0062). (e) Human islets (nondiabetic donor) were untreated (DMSO) or treated with PERK inhibitor (2 μM GSK2656157, 24 h) and equal aliquots of cells or media were resolved by 12% NuPAGE under nonreducing (lanes 1–8) or reducing (lanes 9–12) conditions. nonreducing or reducing conditions as indicated at bottom. Identical nonreduced samples were either not exposed (lanes 1–4) or exposed (lanes 5–8) to post‐gel disulfide reduction (as indicated below) prior to electrotransfer and immunoblotting with anti‐human proinsulin (Proins). Proinsulin monomers bearing native (blue arrow) or non‐native (red arrow) disulfide bonds are indicated.

It has previously been reported that PERK deficiency in β‐cells (Harding et al., 2012) also impairs ER homeostasis and anterograde trafficking accompanied by accumulation of misfolded proinsulin (Gupta et al., 2010; Sowers et al., ²⁰¹⁸) in high molecular weight disulfide‐linked complexes (Arunagiri et al., 2019) (Figure 3c lane 7). New methodology establishes that in PERK inhibitor‐treated Min6 cells there is a considerable intracellular pool of monomeric proinsulin (Figure 3c lane 3), including two distinct forms (red and blue arrows) with a decreased fraction of proinsulin bearing native disulfide bonds (Figure 3d). Upon PERK inhibitor treatment, both monomeric proinsulin forms could also be observed in the islets of Ins2‐KO mice (indicating that the two bands do not come from distinct Ins genes, Supplemental Figure S5 lane 6) as well as in human islets that only express one INS gene (Figure 3e lane 6 blue and red arrows). Thus, each of these pharmacological perturbations of ER homeostasis are associated with diminished efficiency in the formation of proinsulin bearing native disulfide bonds, with an increase of non‐native proinsulin monomers (a form that has previously been under‐recognized), as well as intermolecular (non‐native) disulfide‐bonded species, especially dimers.

2.4. Proinsulin folding response to impairment of ER‐to‐Golgi trafficking

From the above treatments (energy poisons and PERK inhibition) we wondered if the adverse effects on the ER intraluminal folding environment were entirely distinct from the indirect effects of these agents on ER‐to‐Golgi trafficking. YIP family (YIPF) genes encode polytopic membrane proteins found in protein complexes that are linked to ER‐to‐Golgi trafficking. Patients bearing bi‐allelic mutations of YIPF5 (expressed in pancreatic islets and elsewhere) have been found to exhibit neonatal diabetes in addition to central nervous system defects. In vitro‐differentiated β‐like cells bearing YIPF5‐I98S (or YIPF5‐KO) exhibit defective ER‐to‐Golgi trafficking of proinsulin with diminished insulin biogenesis (De Franco et al., 2020). We examined proinsulin folding status within in vitro‐differentiated β‐like control cells and YIPF5‐I98S cells; in the latter case, proinsulin folding to the native disulfide isomer was diminished (Figure 4a lane 3 blue arrow; quantified in Figure 4b) with an increase of misfolded proinsulin monomers (Figure 4a red arrow) and decreased insulin yield (Figure 4a lane 9; quantified in Figure 4c).

FIGURE 4.

Proinsulin misfolding in the ER of YIPF5‐I98S iPSCs. (a) Control or YIPF5‐I98S (YIPF5*) iPSCs were differentiated to stage 7 β‐like cells prior to analysis. Media containing secretion from human islets was collected as a control to identify the position of native human proinsulin as detected by immunoblotting under nonreduced conditions (lanes 1 and 7). Lysates of β‐like cells were analyzed under nonreduced (lanes 2,3 and 8,9) or reduced conditions (lanes 5,6 and 11,12). Identical nonreduced samples were either exposed (lanes 1–3) or not exposed (lanes 7–9) to post‐gel disulfide reduction (as indicated above) prior to electrotransfer and immunoblotting with anti‐human proinsulin (Proins). Intracellularly, the mutant β‐like cells exhibited a decreased abundance of native proinsulin (blue arrow). Non‐native proinsulin monomers are shown with red arrow. The same samples were analyzed by immunoblotting with guinea pig anti‐insulin, which primarily recognizes disulfide‐linked two‐chain mature insulin (lanes 8,9). Cyclophilin B (CypB) is a loading control. (b) Quantitation of native proinsulin (blue arrow) as a fraction of total cellular proinsulin (recovered under reduced conditions) from at least three independent differentiation experiments like that shown in panel A (mean ± SD; *p = 0.044). (c) Quantitation of mature insulin (normalized to the loading control) from the same samples as those quantified in panel B (mean ± SD; *p = 0.032). In panels D and E, Stage 7 islet‐like clusters differentiated from embryonic stem cells from WT or YIPF5‐I98S were implanted under the kidney capsule of immunocompromised NOD/SCID‐γ mice and then retrieved for processing 3 months thereafter (the WT cell images are above and YIPF5‐I98S cell images are below). (d) Co‐immunostaining for insulin (INS, panel 2), proinsulin (PROINS, panel 3), or merged image (panel 4); counter‐stained with Hoechst (panel 1). (e) Co‐immunostaining for proinsulin (PROINS, panel 2), the ER protein calreticulin (CALR, panel 3) or merged image (panel 4); counter‐stained with Hoechst (panel 1). Note that in YIPF5‐I98S β‐like cells, the accumulated proinsulin colocalized with calreticulin.

When control or YIPF5 mutant embryonic stem cells were in vitro‐differentiated into islet‐like clusters and engrafted into NOD/SCID‐γ murine hosts (as previously described [De Franco et al., ²⁰²⁰]), a decrease in immunostainable insulin (Figure 4d panel 2, compare lower to upper) was accompanied by a dispersal of cytoplasmic proinsulin immunostaining (Figure 4d panel 3, compare lower to upper), which strongly overlapped with that of calreticulin (Figure 4e panel 4, compare lower to upper), indicating an ER localization. Such a change in the steady‐state distribution of proinsulin in this cellular model of a genetic defect linked to neonatal diabetes (De Franco et al., 2020) is curiously similar to that reported in the islets of homozygous LepR ^db/db mice [a T2D model], in which we found an increased abundance of high molecular‐weight disulfide‐linked proinsulin complexes (Arunagiri et al., 2019). Here we re‐examined db/db mouse islets in comparison to db/+ control counterparts. The homozygous db/db mouse islets contain more total proinsulin (Supplemental Figure S6A lane 4 vs. 3) with a particular increase of misfolded monomers and disulfide‐linked dimers (lane 2 red arrows). Proinsulin molecules bearing native disulfide bonds are also present in db/db mouse islets (Supplemental Figure S6A lane 2 blue arrow) but of the total proinsulin, the fraction bearing native disulfide bonds was clearly diminished (Supplemental Figure S6B)—and this correlates with a diminished amount of mature insulin (Supplemental Figure S6A lane 6). Thus, from two wholly unrelated primary genetic defects leading to two forms of diabetes in two entirely different species, we observe a shared β‐cell phenotype of proinsulin misfolding with a shift in proinsulin distribution to the ER, suggesting the possibility that these two phenotypes are related (proposed in Supplemental Figure S1).

Treatment of cells with brefeldin A (BFA), which stimulates the ADP‐ribosylation of CTBP1 (a gene strongly linked to T2D that affects Golgi structure [Colanzi et al., ²⁰¹³]), impairs ER‐to‐Golgi anterograde trafficking and has been reported to concomitantly induce proinsulin misfolding (Jang et al., 2019). To explore this, we first introduced a BFA‐imposed ER exit block for 10 min, which decreased natively disulfide‐bonded proinsulin (Figure 5a blue arrow) and increased non‐native proinsulin, which was not primarily the higher molecular weight disulfide‐linked oligomers that have previously been reported (Jang et al., 2019; Zhu et al., ²⁰¹⁹) but rather mostly included misfolded monomers and disulfide‐linked dimers (Figure 5a red arrows). (Co‐treatment of the live cells with 5 mM DTT blocked the appearance of these forms, yielding only reduced proinsulin monomers [Figure 5a]). We also prepared islets from Ins2‐KO mice so that, once again, we could ensure that only one Ins gene product was expressed. Moreover, islets were pre‐treated with cycloheximide (CHX) for 4 h to allow for turnover of most pre‐existing proinsulin and endoproteolytic conversion intermediates; this (Figure 5b first lane) was then followed by washout of CHX to begin to synthesize new proinsulin in the absence (Figure 5b second lane) or presence of BFA (third lane). With the ER‐to‐Golgi trafficking blockade imposed by BFA there was a decrease of new proinsulin bearing native disulfide bonds (Figure 5b blue arrow; quantified in Figure 5c) and an increased amount of new proinsulin recovered as non‐native species (Figure 5b red arrows) and this same behavior was also observed in human islets (Figure 5d red arrows). These data strongly implicate a connection between impaired ER anterograde trafficking and proinsulin misfolding.

FIGURE 5.

Misfolded proinsulin accumulates upon disruption of ER‐to‐Golgi trafficking. (a) Min6 cells were treated ± Brefeldin A (BFA, 10 μg/mL) for 10 min in the absence or presence of 5 mM DTT (as indicated) before analysis by 12% NuPAGE under nonreducing conditions or reducing (last lane used as a MW marker of fully reduced proinsulin) followed by immunoblotting with anti‐rodent proinsulin (CCI‐17). Secreted proinsulin from untreated Min6 cells (far left) is a marker of endogenous native proinsulin. Native intracellular proinsulin is shown with blue arrow; proinsulin disulfide‐linked dimers and misfolded proinsulin monomers are shown with red arrows. (b) Isolated pancreatic islets from Ins2‐KO mice were treated as outline at the top of panel: first a 4 h treatment with cycloheximide (first lane: CHX) followed by two washes with PBS and a further 45 min culture in complete medium in the absence (Vehicle) or presence of BFA (10 μg/mL). The islets were then lysed and resolved by straight 12% NuPAGE under nonreducing (left panel) or reducing (right panel) conditions. The last lane in each panel shows the position of proinsulin (Proins) secreted overnight from Ins2‐KO islets as an internal control. The nonreducing gel was treated with DTT + heating prior to transfer for immunoblotting with anti‐rodent proinsulin. Native proinsulin is shown (blue arrow); non‐native monomers and disulfide‐linked proinsulin dimers are also observed (red arrows). Cyclophilin B (Cyp B) is a loading control. (c) Quantitation of native proinsulin monomers recovered as a fraction of total proinsulin from experiments performed as in panel B (n = 3 independent experiments; mean ± SD; **p = 0.0099). (d) Isolated human islets were treated as described for rodent islets in panel B; BFA treatment decreased the recovery of native proinsulin (blue arrow) and increased misfolded proinsulin (red arrows).

2.5. Restoration of proinsulin folding upon reversal of induced ER‐to‐Golgi trafficking defects

We reasoned that if proinsulin misfolding occurs in states of deficient ER‐to‐Golgi trafficking, then ameliorating the ER‐to‐Golgi trafficking defect may result in improved proinsulin folding. As a first test we used washout of BFA (as its effects are reversible) in INS1E cells (while imposing a distal secretory pathway block with monensin to capture native proinsulin molecules within the Golgi region; e.g., Figure 2d, e). BFA pre‐treatment induced blockade of ER‐to‐Golgi transport and triggered a decrease of proinsulin monomers bearing native disulfide bonds with an increase of misfolded proinsulin including disulfide‐linked proinsulin dimers and misfolded monomers (Figure 6a first lane red arrows; similar to Figure 5). After BFA pre‐treatment, CHX was added to block any further proinsulin synthesis in order to enable continued examination of the pre‐existing misfolded proinsulin. Remarkably, upon a 90 min BFA washout, misfolded proinsulin nearly vanished (including disappearance of both disulfide‐linked dimers and misfolded monomers) while native disulfide‐bonded proinsulin increased significantly (Figure 6a) lane 2 blue arrow; quantified in Figure 6b; and total proinsulin recovery was 95% (no significant change) from pre‐washout values in n = 4 independent experiments. Thus, when ER‐to‐Golgi trafficking was restored, misfolded proinsulin was refolded to the native disulfide isomer. This could not be explained merely by blocking the synthesis of new (unfolded) proinsulin molecules because when CHX was included during the continued presence of BFA, misfolded proinsulin remained unabated (Figure 6a lane 3; quantified in Figure 6b). Thus, these data highlight that proinsulin folding is dramatically remediated upon restoration of ER‐to‐Golgi trafficking.

FIGURE 6.

Amelioration of proinsulin misfolding in β‐cells. (a) INS1E cells were pre‐treated with brefeldin A for 2 h (“BFA pre‐treat”) followed either by an additional 90 min in the presence of cycloheximide (“BFA con't + CHX”) or BFA washout involving a quick PBS wash followed by addition of complete RPMI medium containing CHX plus Monensin (“BFA → CHX + Mon”). Cell lysates from each group were either resolved by 12% NuPAGE under reduced conditions, or under nonreduced conditions with post‐gel disulfide reduction before electrotransfer and immunoblotting with anti‐rodent proinsulin. Native proinsulin monomers (blue arrow) and disulfide‐linked dimers and non‐native proinsulin monomers (red arrows) are indicated. β‐actin (β‐act, shown above) is a loading control. (b) Quantitation (from 4 independent experiments; mean ± SD) of disulfide‐linked proinsulin dimers; nonnative proinsulin monomers; and native proinsulin monomers are shown with p‐values or nonsignificant changes (ns) indicated on the figure. (c) INS1E cells were pre‐treated for 60 min in phosphate buffered saline plus 20 mM 2‐deoxyglucose (“2DG”), followed either by an additional 30 min in the presence of CHX (“→2DG, 30 min”) or by washout involving replacement with complete RPMI medium containing CHX plus Monensin (“→RPMI, 30 min”). Cell lysates from each group were either resolved by 12% NuPAGE under reduced conditions, or under nonreduced conditions with post‐gel disulfide reduction before electrotransfer and immunoblotting with anti‐rodent proinsulin. Native proinsulin monomers (blue arrow) and disulfide‐linked dimers and non‐native proinsulin monomers (red arrows) are indicated. Hsp90 (shown below) is a loading control. (d) Quantitation (from at least four independent experiments; mean ± SD) of disulfide‐linked proinsulin dimers; nonnative proinsulin monomers; and native proinsulin monomers are shown with p‐values or nonsignificant changes (ns) indicated directly on the figure.

We sought to explore whether proinsulin misfolding in β‐cells can be similarly improved upon rescue from energy depletion (again imposing a distal secretory pathway block with monensin, as above). Because the effects of antimycin A on mitochondrial electron transport are thought to be irreversible, instead INS1E cells were pre‐incubated (60 min) in PBS (i.e., lacking glucose or amino acids) containing 20 mM 2‐deoxyglucose. Energy depleted β‐cells exhibited non‐native proinsulin monomers and disulfide‐linked dimers (Figure 6c red arrows). After pre‐incubation, CHX was added once again to ensure that no further proinsulin synthesis could take place (thus enabling examination of pre‐existing misfolded proinsulin). Upon returning the cells to complete nutrient‐enriched media, disulfide‐linked proinsulin dimers, and misfolded proinsulin monomers were both significantly decreased (although a significant increase in native proinsulin monomers could not be demonstrated, Figure 6d; and total proinsulin recovery [by reducing SDS‐PAGE] was unchanged from values after pre‐incubation). Altogether, the data in Figure 6 establish that the presence/accumulation of misfolded proinsulin is potentially fixable and responsive to improved ER homeostasis that follows restoration of ER‐to‐Golgi trafficking, or restoration of β‐cell energy charge (which supports ER protein folding and trafficking).

3. DISCUSSION

Proinsulin in the ER is susceptible to misfolding (Haataja et al., 2016), and primary defects in proinsulin folding from INS gene mutations account for the development of diabetes in the syndrome known as MIDY (Arunagiri et al., 2018). Recent reports are also consistent with the idea that a partial or subtle predisposition to proinsulin misfolding can be genetically encoded (Alam et al., 2021), and when combined with an environmental/dietary component, this results in defective proinsulin trafficking (Alarcon et al., 2016; Arunagiri et al., ²⁰¹⁹). Molecular crowding as well as the high rate of proinsulin biosynthesis within the ER (Yong et al., 2021) may help to account for proinsulin misfolding in individuals expressing exclusively WT proinsulin (Liu et al., 2018; Liu et al., ²⁰²¹). Additionally, despite the presumptively protective isomerase activity of ER oxidoreductases (Jang et al., 2019; Rajpal et al., ²⁰¹²; Rohli et al., ²⁰²²), unfavorable metabolic or genetic factors may result in misfolding of proinsulin that cannot progress from the ER into the distal secretory pathway (Alam et al., 2021; Arunagiri et al., ²⁰¹⁹). Among the many genetic variants linked to diabetes are genes expressed in β‐cells that impact on ER homeostasis and ER‐to‐Golgi trafficking (see Introduction). Some of these primary genetic deficiencies in β‐cells have already been reported to result in proinsulin misfolding, accompanied by ER stress (Zhu et al., 2019), including a recent report of IER3IP1 deficiency (Yang et al., 2022). To us, this suggests a ‘positive feedback loop’ in which primary proinsulin misfolding perturbs ER homeostasis and thus affects anterograde trafficking in the secretory pathway (Yuan et al., 2012), and primary impairment of anterograde trafficking in the secretory pathway perturbs ER homeostasis and thus affects the folding of proinsulin (Supplemental Figure S1). We believe that such a feedback loop can help to account for ER distension and activated ER stress response as a common phenotype shared between several forms of monogenic diabetes as well as the far more common T2D (Back & Kaufman, 2012; Marchetti et al., ²⁰⁰⁷).

In this report, we've modified the method (that has grown in popularity in recent years) for examining proinsulin misfolding by the use of immunoblotting after nonreducing SDS‐PAGE (Alam et al., 2021; Arunagiri et al., ²⁰¹⁹; Jang et al., ²⁰¹⁹; Ninagawa et al., ²⁰²⁰; Sun et al., ²⁰²⁰; Tran et al., ²⁰²⁰), to develop an unbiased approach that allows for a more quantitative analysis of different disulfide‐bonded forms of proinsulin (Figure 1a). By nonreducing SDS‐PAGE with a fixed percentage acrylamide (for uniform transfer efficiency) and treating the SDS‐gel with high‐dose DTT and heating prior to electrotransfer (thereby converting all of the differently folded forms of proinsulin into reduced monomers sharing equal immunoreactivity) followed by immunoblotting with proinsulin‐specific antibodies, we have enabled a more accurate assessment of the relative abundance of proinsulin bearing native and non‐native disulfide bonds. These tools should help diabetes research laboratories around the world examine proinsulin misfolding in states of health and disease—including (as shown in this report) various pancreatic β‐cell lines, rodent islets, human islets, and human iPSCs differentiated into β‐like cells. Moreover, our results lend support to the conclusion that it is proinsulin molecules bearing native disulfide bonds (the fastest‐migrating immunoreactive band) that are exported into the distal secretory pathway and secreted (Figures 1f and 2a), thus serving as a standard for properly‐folded proinsulin.

Perturbation of ER homeostasis, including energy depletion by inhibition of mitochondrial electron transport, uncouplers, nutrient deprivation—or inhibition of PERK—each lead to significant proinsulin misfolding (Figures 3, 5 and Supplemental Figures S4 and S5). Under these conditions, it is primarily misfolded proinsulin monomers and disulfide‐linked dimers that are abnormally increased at the expense of natively folded proinsulin monomers. Higher‐order aberrantly‐disulfide‐linked proinsulin complexes can also form in pancreatic islets (and these can be ultrasensitively detected using previously published methods [Arunagiri et al., ²⁰¹⁹]) but their steady‐state abundance is not quite as high as previously believed (Supplemental Figure S6). As proinsulin molecules bearing native disulfide bonds have no available cysteine that is free to participate in the formation of higher‐order covalent complexes, it seems most likely that the misfolded proinsulin monomers represent the proinsulin species recruited into these higher‐order disulfide‐linked complexes (Supplemental Figure S5). Importantly, in this report, the term ‘misfolded monomers’ detected after SDS‐PAGE does not exclude proinsulin engaged in complexes that are noncovalently associated within the ER (Tran et al., 2020); similarly, the term “native monomers” also includes proinsulin molecules that normally form noncovalent dimers and go on to assemble into noncovalent hexamers (Huang & Arvan, 1995).

Based on published work, YIPF5 can be added to the list of genes whose mutation triggers a phenotype of impaired in ER‐to‐Golgi proinsulin trafficking (De Franco et al., 2020). Here we find that β‐like cells differentiated from YIPF5 mutant β‐like cells exhibit a diminished fraction of proinsulin bearing native disulfide bonds while proinsulin shifts its steady‐state distribution to the ER (Figure 4); additionally, the cells exhibit a significant diminution of mature insulin (Figure 4). Together, these data support a common pathological endpoint suggested in the diagram of Supplemental Figure S1. It is with this in mind that we note that db/db early‐stage diabetic mice maintain an increase in total proinsulin, which is comprised considerably of misfolded proinsulin monomers and disulfide‐linked dimers (Supplemental Figure S6). The simplest explanation of these findings is that there is a perturbation of β‐cell ER homeostasis (the causes of which include diminished ER‐to‐Golgi proinsulin trafficking) during the development and progression of diabetes in rodent models of T2D, and perhaps in human T2D as well. However, we hasten to add that we have not proven here that the proinsulin misfolding occurring in the absence of INS gene mutations is the cause of diabetes, as it may merely reflect the perturbed ER homeostasis that occurs in β‐cells in diabetic states. However, proinsulin misfolding is known to cause ER stress; thus it is likely that the observed misfolding triggers some of the β‐cell ER stress responses that have been reported in the islets of animals and humans with T2D (Shrestha et al., 2021).

In this study, we show that acute blockade of ER‐to‐Golgi anterograde trafficking (with BFA treatment) brings on near‐immediate proinsulin misfolding in a β‐cell line (Figure 5a), and similar observations were made in rodent islets (Figure 5b, c) and human islets (Figure 5d). Given that the blockade of ER‐to‐Golgi trafficking is reversible upon BFA washout, this report demonstrates for the first time that restoration of normal ER‐to‐Golgi trafficking allows for a rapid improvement in intracellular proinsulin folding status (Figure 6a, b). Moreover, we find that restoration of energy charge in β‐cells also provides notable benefit to the overall folding status of intracellular proinsulin (Figure 6c, d). These interesting observations suggest that treatments designed to improve ER‐to‐Golgi trafficking and energy charge in β‐cells should be explored for their possible contributions to restore ER homeostasis, enhance proinsulin folding to the native state, deliver more proinsulin to the distal secretory pathway, and thus lay the groundwork for increased insulin production, which would be of benefit in T2D and some monogenic forms of diabetes. Moreover, the tools and approaches described herein should enable examination of proinsulin folding status even in whole human pancreas tissue, thus making it possible to study cadaveric and/or surgical specimens from both human T1D and T2D patients. Clearly, studies of such patients and donors will be needed to understand the implications of these findings in the islets of humans with diabetes, and to correlate the magnitude of proinsulin misfolding with the progression of disease.

4. MATERIALS AND METHODS

4.1. Antibodies and reagents

Chemicals were purchased from Sigma‐Aldrich or ThermoFisher. Recombinant human proinsulin (expressed in E. coli) was purchased from Creative Biomart (Catalog # INS‐315H). Antibodies in this study include rabbit anti‐Myc (Immunology Consultants, RRID:AB_2921297); guinea pig anti‐insulin (Covance, RRID:AB_10013624); rabbit anti‐cyclophilin B (ThermoFisher, RRID:AB_2169138); rabbit polyclonal anti‐calreticulin (Abcam RRID:AB_303402); rabbit mAb anti‐Hsp90 (Cell Signaling, RRID:AB_2233307); rabbit anti‐GM130 (Abcam, RRID:AB_880266); mouse mAb anti‐human proinsulin B‐C junction sequence PKTRREAEDLQVGQ (Abmart, RRID:AB_2921300); mouse mAb anti‐rat proinsulin (CCI‐17, Novus Biologicals RRID:AB_1107982); mouse anti‐rat C‐peptide CII‐29 (Advanced Immunochemical Inc., Catalog ID: 1‐CP‐r); mouse anti rat C‐Peptide I 6H1 (Biorad Catalog # MCA2857) and as a back‐up, a custom‐made rabbit polyclonal antibody against mouse C‐peptide‐2 sequence PQVAQLELGGGPGAGDLQT (Biogot, RRID:AB_2921302).

4.2. Cell culture and transfection

HEK293T cells (ATCC, CRL‐3216) were cultured in DMEM supplemented with 10% calf serum, 100 IU/mL penicillin and 100 μg/mL streptomycin. INS1E rat pancreatic β‐cells were cultured in RPMI‐1640 medium (supplemented with 10% FBS), 10 mM HEPES, 1 mM sodium pyruvate, penicillin/streptomycin and 0.05 mM beta‐mercaptoethanol. Min6 (mouse insulinoma) cells (from Dr. D. Stoffers, U. Pennsylvania) were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), 100 IU/mL penicillin and 100 μg/mL streptomycin, and 0.05 mM 2‐mercaptoethanol. The cells were grown in a 37°C incubator with 5% CO₂. For transfection, cells were seeded into 12‐well plates; 1 day later, 1 μg plasmid DNA and 2.5 μL lipofectamine‐2000 (ThermoFisher) diluted into Opti‐MEM were added to each well. At 24–48 h after transfection, samples were analyzed (with a fresh change of media 6 h before harvesting).

4.3. Immunocytochemistry and image analysis

Min6 cells grown in 8‐well chamber slides (Millicell‐EZ) were fixed with 10% neutral‐buffered formalin for 20 min at room temperature and washed with PBS. Where indicated, fixed cells were incubated with 100 mM DTT in PBS at 60°C for 10 min. Cells were washed twice with PBS and permeabilized with 0.4% Triton X‐100 in TBS for 20 min; washed thrice with TBS and blocked in TBS with 3% BSA plus 0.2% Triton X‐100; incubated overnight (4°C) with primary antibodies (mouse mAb anti‐proinsulin and rabbit anti‐GM130) diluted in blocking buffer; and then rinsed and incubated with secondary antibodies conjugated to Alexa Fluor 488 or 555 (Invitrogen). Slides were mounted with Prolong Gold with DAPI and imaged by epifluorescence in a Nikon‐A1 confocal microscope (60× oil objective). ImageJ software was used to analyze images for intracellular proinsulin distribution.

4.4. Murine islets

Isolation of islets from C57BL/6 or LepR^db/db (referred herein as db/db) or Ins2‐KO mice was performed as described previously (Arunagiri et al., 2019). For overnight recovery, islets were incubated in RPMI‐1640 supplemented with 10% FBS, 10 mM HEPES pH 7.35, 1 mM sodium pyruvate and 0.05 mM 2‐mercaptoethanol, in a humidified 5% CO₂ incubator at 37°C.

4.5. Human induced pluripotent stem cell‐derived β‐like cells, and embryonic stem cell‐derived islet‐like clusters

Peripheral blood mononuclear cells were obtained from two sisters who developed diabetes at the age of 8.5 and 15 months as a result of a homozygous missense mutation in YIPF5 p.(Ile98Ser) or I98S. These cells were reprogrammed into iPSCs (2 lines per patient) and then differentiated into β‐like cells as previously described; control iPSCs (HEL115.6 and 1023A) were differentiated using the same 7‐stage protocol (De Franco et al., 2020). At the end of the differentiation in each case, cell clusters contained 41 ± 7% and 50 ± 15% insulin‐positive cells for YIPF5‐I98S and control cells, respectively. Cell clusters were pelleted and frozen at −80°C before lysis and analysis by SDS‐PAGE. For immunostaining, Stage 7 cells from WT and YIPF5‐I98S embryonic stem cells differentiated into islet‐like aggregates were implanted under the kidney capsule of immunocompromised NOD/SCID‐γ mice as previously described (De Franco et al., 2020). The grafts were retrieved at 3 months after implantation and fixed with 4% formaldehyde at room temperature, paraffin embedded, and then cut into 5 μm sections. Thereafter, the sections on slides were deparaffinized, and antigen retrieval was performed by heating slides in 0.1 M citrate buffer (pH 6) using Decloaking chamber (Biocare Medical) at 95°C for 12 min. Immunofluorescence imaging was performed with a Zeiss Axio Observer Z1 microscope.

4.6. Human islets

Human pancreatic islets from COVID‐negative non‐diabetic donors were obtained from Prodo Labs with 90% purity, 95% viability.

4.7. Western blotting

Cells/islets were lysed in RIPA buffer with protease inhibitor cocktail on ice for 15 min and cleared by centrifugation at 12,000× g at 4°C for 15 min. Samples were heated to 95°C in gel sample buffer containing no reducing agents (nonreducing), or 200 mM DTT for 5 min (reducing), electrophoretically resolved on straight 12%‐NuPAGE gels. Thereafter, the gels were either untreated, or treated with 100 mM DTT at 60°C for 10 min, followed by electrotransfer to nitrocellulose. Primary antibodies were diluted 1:1000 (in TBST plus 5% BSA) and incubated at 4°C overnight. HRP‐conjugated‐secondary antibodies were diluted 1:5000 and incubated at room temperature for 1 h. Peroxidase‐ECL used the Clarity Western ECL Substrate. In all figures (except Figure 2c + e), quantification of native proinsulin monomer was normalized to total proinsulin, defined as the band recovered upon immunoblotting in the same gel after reducing SDS‐PAGE. As described in the headers and Figure legends in Figures 2c + e, total proinsulin was defined as the total of that recovered in the cells + medium after reducing SDS‐PAGE.

4.8. Statistical analysis

Statistical analysis was assessed by unpaired t‐test to determine the differences between groups (GraphPad QuickCalcs), except for the panels in Figure 6 in which one‐way ANOVA with Tukey‐s multiple comparison test was used to compare between groups. Data are presented as mean ± SD; a p‐value <0.05 was considered statistically significant.

AUTHOR CONTRIBUTIONS

Peter Arvan: Conceptualization; funding acquisition; methodology; project administration; resources; supervision; writing – original draft; writing – review and editing. Anoop Arunagiri: Conceptualization; data curation; formal analysis; investigation; methodology; supervision; validation; writing – review and editing. Maroof Alam: Investigation; writing – review and editing. Leena Haataja: Investigation; writing – review and editing. Hassan Draz: Investigation; writing – review and editing. Bashiyer Alasad: Investigation; writing – review and editing. Praveen Samy: Investigation; writing – review and editing. Nadeed Sadique: Investigation; writing – review and editing. Yue Tong: Investigation; validation; writing – review and editing. Ying Cai: Investigation; validation; writing – review and editing. Hadis Shakeri: Investigation; validation; writing – review and editing. Federica Fantuzzi: Investigation; validation; writing – review and editing. Hazem Ibrahim: Data curation; formal analysis; investigation; validation; writing – review and editing. Insook Jang: Investigation; validation; writing – review and editing. Vaibhav Sidarala: Investigation; writing – review and editing. Scott A. Soleimanpour: Funding acquisition; project administration; supervision; writing – review and editing. Leslie S. Satin: Supervision; project administration; funding acquisition; writing – review and editing. Timo Otonkoski: Project administration; writing – review and editing; resources. Miriam Cnop: Funding acquisition; project administration; supervision; resources; writing – review and editing. Pamela Itkin‐Ansari: Funding acquisition; project administration; resources; supervision; writing – review and editing. Randal J. Kaufman: Funding acquisition; supervision; project administration; writing – review and editing; resources. Ming Liu: Funding acquisition; project administration; supervision; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Arunagiri A, Alam M, Haataja L, Draz H, Alasad B, Samy P, et al. Proinsulin folding and trafficking defects trigger a common pathological disturbance of endoplasmic reticulum homeostasis. Protein Science. 2024;33(4):e4949. 10.1002/pro.4949

DATA AVAILABILITY STATEMENT

All data are contained within the figures; additional data available upon request (corresponding author, Peter Arvan).