1. Cell Biology
Download icon

PDIA1/P4HB is required for efficient proinsulin maturation and ß cell health in response to diet induced obesity

  1. Insook Jang
  2. Anita Pottekat
  3. Juthakorn Poothong
  4. Jing Yong
  5. Jacqueline Lagunas-Acosta
  6. Adriana Charbono
  7. Zhouji Chen
  8. Donalyn L Scheuner
  9. Ming Liu
  10. Pamela Itkin-Ansari
  11. Peter Arvan  Is a corresponding author
  12. Randal J Kaufman  Is a corresponding author
  1. SBP Medical Discovery Institute, United States
  2. DLS Consulting, United States
  3. University of Michigan Medical School, United States
  4. University of California, San Diego, United States
Research Article
  • Cited 1
  • Views 511
  • Annotations
Cite this article as: eLife 2019;8:e44528 doi: 10.7554/eLife.44528

Abstract

Regulated proinsulin biosynthesis, disulfide bond formation and ER redox homeostasis are essential to prevent Type two diabetes. In ß cells, protein disulfide isomerase A1 (PDIA1/P4HB), the most abundant ER oxidoreductase of over 17 members, can interact with proinsulin to influence disulfide maturation. Here we find Pdia1 is required for optimal insulin production under metabolic stress in vivo. ß cell-specific Pdia1 deletion in young high-fat diet fed mice or aged mice exacerbated glucose intolerance with inadequate insulinemia and increased the proinsulin/insulin ratio in both serum and islets compared to wildtype mice. Ultrastructural abnormalities in Pdia1-null ß cells include diminished insulin granule content, ER vesiculation and distention, mitochondrial swelling and nuclear condensation. Furthermore, Pdia1 deletion increased accumulation of disulfide-linked high molecular weight proinsulin complexes and islet vulnerability to oxidative stress. These findings demonstrate that PDIA1 contributes to oxidative maturation of proinsulin in the ER to support insulin production and ß cell health.

https://doi.org/10.7554/eLife.44528.001

Introduction

Type two diabetes mellitus (T2D) is a complex disease caused by multiple genetic and environmental factors with an overarching problem of insufficient insulin to meet the level of insulin resistance (Mokdad et al., 2001; Sladek et al., 2007; Narayan et al., 2007; Støy et al., 2007; Kaul and Ali, 2016). Significant evidence supports the notion that pancreatic ß cell failure is fundamental in the etiology of T2D (Alejandro et al., 2015). Insulin resistance increases the burden placed on pancreatic ß cells to increase insulin synthesis and secretion that leads to associated defects including decreased ß cell number (Clark et al., 1988; Butler et al., 2003; Yoon et al., 2003), chronic ER stress and/or oxidative stress (Robertson et al., 2007; Scheuner and Kaufman, 2008; Back and Kaufman, 2012; Han et al., 2015), and a loss of ß cell identity (Talchai et al., 2012; Swisa et al., 2017).

Insulin biosynthesis is a complex and dynamically regulated process. Insulin production is dependent upon differentiation-specific gene expression, translation and translocation of preproinsulin into the ER, ER chaperone-facilitated protein folding and disulfide bond formation, vesicular transport of proinsulin from the ER through the Golgi compartments, and stimulus-coupled granule release (Liu et al., 2018). Protein sorting and assembly of proinsulin into nascent granules in the trans Golgi network sets the framework for proteolytic processing of proinsulin and condensation of insulin with zinc to create mature secretory granules that are staged for secretion in response to stimuli (Liu et al., 2014). Well characterized bottlenecks in protein secretion occur at the stages of correct protein folding in the ER, anterograde trafficking through the secretory pathway and defective stimulus-coupled granule exocytosis. A more thorough understanding of insulin biogenesis should facilitate development of new and highly efficacious treatments for T2D that are based upon enhancing insulin output while preventing the loss of functional ß cells.

Proper oxidative protein folding by the formation of disulfide bonds in the ER is important for protein stability. Misfolded proteins in the ER can be retro-translocated to the cytosol and degraded via the ubiquitin proteasome system (Wu and Rapoport, 2018) and/or autophagy (Loi et al., 2018). Accumulation of misfolded proteins in the ER activates the unfolded protein response (UPR) through the ER stress transducers PERK, IRE1, and ATF6 to alleviate and adapt to the cellular stress. However, chronic stress from an inability to resolve protein misfolding can compromise cell health (Wang and Kaufman, 2016). The vitality of ER homeostasis for ß cell health is underscored by the development of diabetes in rodents and humans with defects that either cause ER protein misfolding or fail to respond when misfolding occurs (Back and Kaufman, 2012).

Proinsulin forms a native folded structure in the ER by disulfide bond formation comprised of two linkages between the A and B polypeptide chains (A7-B7 and A20-B19) and one in the A chain (A6-A11) (Liu et al., 2018; Dai and Tang, 1996; Jia et al., 2003; Yan et al., 2003). Mutations within proinsulin that impact disulfide bond formation cause neonatal diabetes in humans and Akita mice, serving as a model of proinsulin misfolding-induced ß cell failure (Støy et al., 2007; Colombo et al., 2008; Riahi et al., 2018). In general, disulfide bond formation within secretory proteins occurs during the early stages of protein folding as cysteine residues establish proximity to one another; however, enzymes can assist catalyzing this process (Bulleid, 2012). The specific complement of cellular redox machinery required for normal insulin output or to maintenance of insulin secretion under conditions of nutrient excess, obesity, or genetic predisposition to diabetes is undefined.

During disulfide bond formation, ER oxidoreductin 1 (ERO1) transfers oxidizing equivalents from O2 to form disulfide bonds in a large family of ER oxidoreductases (Hudson et al., 2015); this process ultimately culminates in the transfer of electrons from sulfhydryls to molecular oxygen. This oxidoreductase family localized to the ER comprises over 17 members in mammals and each one interacts with specific substrates in a different manner (Jessop et al., 2009; Braakman and Bulleid, 2011). Among them, PDIA1, also known as prolyl 4-hydroxylase beta (P4HB), is the major oxidoreductase located in the ER lumen (Ramming and Appenzeller-Herzog, 2012; Benham, 2012).

Protein disulfide isomerase (PDI) can catalyze oxidation, reduction, and isomerization of disulfide bonds in the ER (Freedman et al., 1994). PDI has four domains (a, a’, b, and b’) with a linker (x) and acidic C-terminal region (c) where a KDEL ER retention signal resides (Hatahet and Ruddock, 2007). Similar to thioredoxin, the two catalytic domains (a, a’) are separated by non-catalytic domains (b, b’) to form a structure of a-b-b’-x-a’-c where each catalytic domain has an active site consisting of Cys-Gly-His-Cys residues (Hatahet and Ruddock, 2007). Thiol groups on cysteine residues in the active motif are responsible for the oxidoreductase activity, enabling PDI to catalyze both disulfide bond formation as well as disulfide bond isomerization for selective substrates (Hudson et al., 2015; Hatahet and Ruddock, 2007; Freedman, 1995; Wang et al., 2015). In addition, PDI also exhibits chaperone activity, independent of its disulfide isomerase-activity as it binds to misfolded proteins to prevent their aggregation (Puig and Gilbert, 1994; Wilson et al., 1998).

Although PDIA1 catalyzes both disulfide bond formation and isomerization of proinsulin in vitro (Winter et al., 2011), it remains unknown how PDIA1 influences proinsulin folding and insulin production in vivo. Here, we generated ß cell specific Pdia1-deleted mice and established that PDIA1 is required for optimal insulin production needed to maintain glucose homeostasis in the face of metabolic challenge. The results support the conclusion that therapeutics directed to promote native disulfide bond formation within proinsulin is an attractive strategy to prevent ß cell failure in T2D.

Results

Generation of ß cell-specific Pdia1 deleted mice

As PDIA1 is highly expressed in islets (Cras-Méneur et al., 2004; Ahmed and Bergsten, 2005), we pursued analysis of ß cell-specific conditional Pdia1-null mice using tamoxifen (Tam)-regulated deletion of floxed Pdia1 alleles (Kim et al., 2013) through rat insulin promoter driven Cre-recombinase (RIP-CreERT) (Figure 1A) (Dor et al., 2004). Quantitative RT-PCR (qRT-PCR) demonstrated an ~75% decrease in Pdia1 mRNA in isolated islets from the ß cell-specific Pdia1-knock out mice (Pdia1 fl/fl;CreERT herein, KO, but genotypes are defined in the figures) with no effects on Insulin 2, Pdia6 or Pdia3 mRNAs (Figure 1B). Early studies demonstrated that mice with or without the RIP-CreERT allele did not show significant differences in glucose homeostasis after a 14 wk HFD (Figure 1—figure supplement 1A). Therefore, we compared mice with two floxed alleles (fl/fl) and mice with one floxed and one wildtype (WT) allele (fl/+) with littermates that also harbor the RIP-CreERT transgene, both before and/or after Tam injection. Western blotting of isolated islets from Tam-treated mice with the RIP-CreERT allele demonstrated significantly reduced PDIA1 protein with increased expression of the UPR genes BiP, PDIA6 and GRP94 (Figure 1C–D), suggesting Pdia1 deletion may cause ER stress in ß cells of the KO mice. Immunohistochemistry (IHC) of pancreas tissue sections with antibodies specific for proinsulin/insulin, glucagon and PDIA1 confirmed the absence of PDIA1 in a ß cell-specific manner in the KO mice (Figure 2A–B, compare middle panels). Interestingly, analysis of glucagon staining in pancreas sections detected very low PDIA1 expression in islet α cells (Figure 2C, middle panels), and we note that proglucagon contains no disulfide bonds.

Figure 1 with 1 supplement see all
Conditional ß cell-specific Pdia1 deleted mice were generated with Tamoxifen (Tam) induction.

(A) Diagram depicts the generation of Pdia1:RIP-CreERT mice. Mice with floxed Pdia1 alleles (Hahm et al., 2013) were crossed with RIP-CreERT transgenic mice (Dor et al., 2004) and progeny were injected IP with Tam to induce CreERT function and Pdia1 deletion. Control littermate mice with one or two floxed Pdia1 alleles, but not harboring the RIP-CreERT transgene, were injected in parallel with Tam. (B–D) Pdia1 deletion is specific. (B) Total RNA was extracted from islets isolated from female mice at eight wks after Tam injection. mRNA levels were measured by qRT-PCR. Mean ± SEM, n = 3 for each group (p<0.01**). (C) Western blot illustrates expression of Vinculin, PDIA1, BiP, PDIA6, GRP94, Proinsulin, and Insulin in islets isolated from female mice at 14 wks after Tam injection. (D) Quantification of indicated proteins by Western blotting (from C) is shown. Each value was normalized to vinculin except for the proinsulin to insulin ratio. Pdia1 fl/fl (n = 2), Pdia1 fl/+ (n = 2), Pdia1 fl/fl;Cre (n = 2).

https://doi.org/10.7554/eLife.44528.002
Pdia1 is specifically and persistently deleted in murine ß cells.

(A–C) Pancreas tissue sections were prepared from female mice at 49 wks after Tam injection and immuno-stained with anti- proinsulin, insulin, PDIA1, and glucagon antibodies. Images were merged with DAPI stain. Scale bar, 20 µm. (D) Old KO mice developed glucose intolerance compared to control genotypes measured by glucose tolerance testing (GTT) at nine wks after Tam injection. Male mice at 9 mon of age were injected with Tam and fed a regular chow. Mice were fasted (4 hr) prior to IP glucose injection (2 g/Kg body weight) and glucose levels were measured by tail bleeding at each time point (0; non-injected, 15, 30, 60, 90 min). control genotypes: Pdia1 fl/+ (n = 2), Pdia1 fl/+;Cre (n = 5), KO; n = 5. (E) Area under the GTT curve (Δ-AUC) of (D) is indicated in graph.

https://doi.org/10.7554/eLife.44528.004

Although the RIP-CreERT allele was reported to be expressed in the hypothalamus (Wicksteed et al., 2010), Western blotting of PDIA1 in hypothalamic tissue did not detect reduced PDIA1 expression (Figure 1—figure supplement 1B). In addition, we previously reported that this RIP-CreERT allele did not affect serum dopamine, which is synthesized in the arcuate nucleus of the hypothalamus (Hahm et al., 2013). Although we detected Cre positive staining in the hypothalamus, there was no difference in expression of growth hormone-releasing hormone (GHRH) in the KO mice compared to the control genotypes with the RIP-CreERT allele.

Young male and female mice did not exhibit any defects in glucose homeostasis. However, on a regular chow diet, the challenge of aging in KO males caused statistically significant increase in glucose intolerance (Figure 2D–E).

KO mice fed a high fat diet (HFD) become glucose intolerant with defective insulin production

To determine the role of ß cell PDIA1 in the face of metabolic stress, male mice were fed a 45% HFD. Genetic controls (Pdia1 fl/fl and fl/+) and KO mice fed HFD up to 32 wks showed no significant differences in body weight or weight gain (Figure 3A). However, fasting blood glucose (4 hr) in the HFD mice was significantly elevated in the KO versus genetic controls (Figure 3B). In addition, glucose intolerance was observed in HFD-fed KO mice as measured by the difference in the area under the glucose excursion curve (Delta-AUC) (Figure 3C). Serum insulin levels were decreased in HFD-fed KO mice, and a fasting-refeeding challenge revealed hypoinsulinemia with an increased proinsulin/insulin ratio (Figure 3D). Pdia1 deletion did not affect insulin sensitivity measured by insulin tolerance tests (Figure 3E) and no significant difference was observed in the percent of ß cell area to total pancreas or ß cell area per islet in HFD-fed KO mice (Figure 3—figure supplement 1). These results indicate that in the setting of metabolic challenge, Pdia1 is required for adequate insulin production to maintain systemic glucose homeostasis.

Figure 3 with 1 supplement see all
ß cell-specific Pdia1 deleted male mice are glucose intolerant with defective insulin secretion when fed a 45% High Fat Diet (HFD).

All mice were Tam injected at three wks after HFD was started. (A) No difference was observed in body weight (g) between control genotypes and KO mice. Mean ± SEM, controls; n = 17, KO; n = 12. (B) Fasting (4 hr) blood glucose levels were elevated in KO mice at 11, 16 and 20 wks after HFD. Glucose levels were measured by tail bleeding. Mean ± SEM, control genotypes; n = 17, KO; n = 12. (C) KO mice displayed higher blood glucose levels and area under the GTT curve (Δ-AUC) compared to control genotypes during glucose tolerance testing (GTT) after HFD for 25 wks. GTT were performed at multiple time points after HFD in two independent cohorts and representative results are shown. Mice were fasted (4 hr) prior to IP glucose injection (1 g/Kg body weight) and glucose levels were measured by tail bleeding at each time point (0; non-injected, 15, 30, 60, 90 min). control genotypes; n = 17, KO; n = 6. (D) KO mice exhibited decreased serum insulin levels and an increased serum proinsulin/insulin ratio compared to control genotypes. Insulin and proinsulin ELISAs were performed with the serum obtained from mice after fasting (overnight) and re-feeding (4 r) after HFD for 17 wks. control genotypes; n = 8, KO; n = 12. (E) No difference was observed in insulin tolerance tests performed after HFD for 20 wks. Mice were fasted for 4 hr before IP injection of insulin (1.5units/Kg). Glucose levels were measured by tail bleeding at each time point (0; non-injection, 15, 30, 60, 90, 120 min). control genotypes; n = 17, KO; n = 12. (F–I) Total RNA was extracted from islets isolated from mice after HFD for 30 wks. mRNA expression was measured by qRT-PCR. control genotypes; n = 3, KO; n = 3 mice. (F). ß cell-, α cell- and insulin processing genes. (G) PDI family and SERCA genes. (H) UPR genes. (I) Antioxidant response- and cell death- related genes. All data are shown as Mean ± SEM. p<0.05*, p<0.01**, p<0.001***.

https://doi.org/10.7554/eLife.44528.005

Pdia1 is not required for expression of ß cell-specific genes, antioxidant response genes or cell death genes

To reveal whether Pdia1 deletion impacts gene expression to cause ß cell dysfunction, we isolated islets from male mice after 30 wks of HFD and analyzed mRNA levels by qRT-PCR. The results demonstrated no significant decrease in ß cell- and α cell- specific mRNAs, mRNAs encoding the insulin processing enzymes PC1/3, PC2, and CPE, or mRNAs encoding ß cell transcription factors PDX1 or MAFA (Figure 3F). Therefore, the reduced insulin content in KO serum and islets (Figure 3D) was not due to reduced ß cell-specific gene expression. There was also no significant change in expression of other PDI family members and Serca2b, except for Pdia4 (Figure 3G). In addition, UPR genes were not significantly elevated at this point in time in the Pdia1 KO islets, other than Grp94 (Figure 3H), which correlated with a slight increase in protein (Figure 1C). Lastly, there were no significant differences in expression of a panel of genes representing the antioxidant response and cell death (Figure 3I). These results show that ß cell-specific Pdia1 deletion does not alter expression of ß cell specific genes, antioxidant response genes, or cell death genes.

β cell Pdia1 KO mice fed HFD exhibit ß cell failure with distinct morphological aberrations

The impact of Pdia1 deletion on ß cell ultrastructure was analyzed by transmission electron microscopy (TEM) (Figure 4). TEM did not detect any significant morphological alterations in the cohorts of KO mice fed a regular diet for 10 wks after Tam injection (data not shown). Strikingly however, ß cells from HFD-fed KO mice showed significant abnormalities including ER vesiculation and distension, mitochondrial swelling, and nuclear condensation, that were not observed in genetic control mice (Pdia1 fl/fl and fl/+) (Figure 4). The red asterisks represent significantly distended ER, reflecting ER stress (Figure 4D,H, expanded). To determine whether Pdia1 deletion also affects insulin granule content, we quantified the number of mature (dense dark core, yellow arrows in enlarged image) and immature (inner gray core, orange filled open arrowheads in enlarged image) granules and discovered that KO mice had fewer mature (~15%) and immature (~50%) granules compared to the control genotypes fl/fl and fl/+ (Figure 4I). Analysis of the cross-sectioned mature granule (MG) vesicle area and dense core size demonstrated that the average cross-sectional area of MGs in KO mice was 20% larger than control genotypes (Figure 4J), however, there was no difference in MG dense core size between genotypes (Figure 4J), suggesting that insulin packaging efficiency within granules was decreased in Pdia1 deleted-ß cells. This is consistent with decreased serum insulin levels in metabolically-challenged KO mice (Figure 3D). Taken together, the findings indicate that Pdia1 is essential to maintain ß cell ultrastructure upon metabolic stress, indicative of suboptimal ß cell function.

Pdia1 deletion induces morphological abnormalities including decreased insulin granule numbers, ER vesiculation and distention, mitochondrial swelling and nuclear condensation in ß cells.

(A–H). Transmission electron microscopy was performed on pancreata obtained from genetic controls (A, B, E, F) and KO (C, D, G, H) male mice after 40 wks of HFD. Images were obtained at 1900X (A, C, E, G) or 4800X (B, D, F, H) magnification. Scale bar represents 5 µm or 1 µm as indicated. Marked area is two times enlarged on the right side: N, nucleus; M, mitochondria; *, distended endoplasmic reticulum; GA, Golgi apparatus. Yellow arrows, mature granules. Orange filled open arrowheads, immature granules. (I) Pdia1 KO mice had reduced numbers of mature and immature granules. For each genotype 80–110 images were quantified. genetic controls; n = 3, KO; n = 2. Mean value is indicated. (J) Mature granule sizes were larger in Pdia1 KO mice compared to genetic controls without differences in mature granule dense core size. For each genotype, 40 images were quantified. genetic controls; n = 3, KO; n = 2.

https://doi.org/10.7554/eLife.44528.008

KO islets have an increased intracellular proinsulin/insulin ratio with accumulation of high molecular weight (HMW) proinsulin complexes

The impact of Pdia1 deletion on proinsulin folding was investigated by analysis of proinsulin and insulin steady state levels in islets isolated from male mice after 30 wks of HFD using Western blotting under reducing and non-reducing conditions. The level of PDIA1 protein was decreased in islets isolated from KO mice, as expected (Figure 5A–B). Insulin levels were lower in KO islets compared to control genotypes while the proinsulin/insulin ratio was significantly elevated (Figure 5A–B). BiP/Hspa5 was induced with increased PDIA4 and PDIA6 in Pdia1-deleted versus Pdia1-sufficient islets (Figure 5A–B). The increased levels of BiP are consistent with the notion that the absence of Pdia1 causes a mild ER stress, suggestive of protein misfolding, which is also consistent with ER distension (Figure 4).

Figure 5 with 3 supplements see all
Pdia1 deletion in HFD fed mice increases islet steady state proinsulin to insulin ratio with accumulation of high molecular weight (HMW) proinsulin complexes.

(A, C) Western blotting was performed for murine islets isolated after HFD for 30 wks. After overnight recovery, islets were lysed and analyzed under reducing (A) or non-reducing (C, D) conditions by SDS-PAGE and Western blotting. Image exposed for a different time for reduced proinsulin (CCI-17) in (A) was used to represent the total proinsulin levels in (D). Six independent mice were analyzed and technical duplicates are indicated as’ on the top of gel. (B). Quantification of indicated proteins was performed under reducing conditions (A). Each value was normalized to vinculin. Proinsulin/insulin ratios were calculated based on the quantification of proinsulin and insulin species under reducing conditions (A). (C) Five samples from A were analyzed under non-reducing conditions on a 4–12% Bis-Tris SDS gel. The proinsulin blot under reducing conditions is under the red line. Left side. Quantification of HMW proinsulin complexes under non-reducing conditions is indicated. (C), left side, upper) The ratio of 49–198 kDa (a) to 14–49 kDa (b) proinsulin complexes is shown. (C), left side, lower) The ratio of 14–49 kDa (b) proinsulin complexes to proinsulin monomer under reducing conditions is shown. Mean value is indicated in graph. (D) The ten samples in A were analyzed under non-reducing conditions after the gel was incubated in 25 mM DTT for 10 min at RT prior to transfer. To control for variable transfer from a gradient gel, we used a fixed percentage gel (12% Bis-Tris SDS). Right side. Quantification is shown for HMW proinsulin complexes under non-reducing conditions. (D), right side, upper) The ratio of 49–198 kDa (a) to 14–49 kDa (b) proinsulin complexes is shown. (D), right side, lower) The ratio of 14–49 kDa (b) proinsulin complexes to proinsulin monomer under reducing conditions is shown. (A–D) genetic controls; n = 3, KO; n = 3 mice. Mean ± SEM, p<0.05*, p<0.01**, p<0.001***. (E) WT murine islets were treated with or without NEM and lysates were analyzed under non-reducing conditions. Equal numbers of islets were divided into two tubes. Left side islets were rinsed with cold-PBS and lysed on ice. Right side islets were rinsed with cold-PBS containing 20 mM NEM and lysed in lysis buffer containing 2 mM NEM. Samples were prepared alongside and lysates were boiled for 5 min. Equal amounts of lysates were loaded and analyzed with or without gel incubation in 25 mM DTT for 10 min at RT prior to transfer. Two different exposure time images (short, long) are shown after DTT incubation. (F) WT murine islets were treated with increasing concentrations of DTT for 20 min in culture at room temperature and then analyzed by non-reducing and reducing SDS-PAGE and Western blotting with proinsulin antibody (CCI-17). The range of oligomeric proinsulin species are identified by an open arrowhead and monomeric proinsulin is indicated by black arrowhead.

https://doi.org/10.7554/eLife.44528.009

The potential role of PDIA1 in disulfide maturation encouraged us to look for the presence of proinsulin disulfide-linked complexes by non-reducing SDS-PAGE (Figure 5C). Western blot analysis under non-reducing conditions using the CCI-17 antibody demonstrated the appearance of multiple proinsulin bands including a monomer at ~6 kDa (black arrowhead, Figure 5C) and a ladder of molecular masses that could correspond to proinsulin dimers up to pentamers (b, 14–49 kDa) as well as a cluster of HMW complexes (a, 49–198 kDa) in Pdia1 fl/fl and KO islets (Figure 5C). The nature of the disulfide-linked oligomeric proinsulin species was described elsewhere (Arunagiri et al., 2019). Pdia1 deletion increased the amount of HMW complexes relative to the oligomeric forms and decreased the oligomers relative to monomeric proinsulin compared to the genetic controls without the RIP-CreERT allele (Figure 5C, quantified in graph).

During the course of our studies we discovered that the epitope reactivity of the CCI-17 monoclonal antibody is very dependent the status of proinsulin sulfhydryls and disulfide bonds. Two major findings support this conclusion. First, when the non-reducing SDS-PAGE gel was treated with dithiothreitol (DTT) prior to transfer to nitrocellulose, we observed a great increase in antibody reactivity with monomeric proinsulin (Figure 5C–D). This suggests that opening the proinsulin molecule significantly enhances epitope exposure to the CCI-17 antibody. It is also important to note that Western blotting analysis of the non-reducing SDS-PAGE gel (with or without subsequent DTT treatment of the gel prior to transfer) demonstrated readily detectable proinsulin disulfide-linked complexes in WT murine islets, nondiabetic human islets and in prediabetic db/db islets prior to onset of hyperglycemia (Arunagiri et al., 2019). The second finding is that treatment with N-ethylmaleimide (NEM) to alkylate free sulfhydryls slightly increased the HMW proinsulin-containing complexes and slightly reduced the disulfide-linked oligomers (Figure 5E). We suggest that PDIA1 may be required to reduce the HMW complexes and NEM treatment inactivates PDIA1, thereby stabilizing the HMW complexes. Similar results were obtained with a selective PDIA1 inhibitor (see below, Figure 6—figure supplement 1).

To gain further insight into the nature of the disulfide-linked oligomers, islets isolated from WT male mice fed a regular diet were treated in culture with increasing concentrations of dithiothreitol (DTT) to increase the ER reduction potential. This treatment produced increasing amounts of the proinsulin monomer and residual disulfide-linked dimer (Figure 5F), consistent with findings of Arunagiri et al. (2019). In addition, female KO mice fed a regular diet for 14 wks after Tam also showed increased HMW complexes relative to the disulfide-linked proinsulin oligomers and reduced oligomers relative to monomeric proinsulin compared to the genetic controls (Figure 5—figure supplement 1).

Oxidant treatment of Pdia1 KO islets increases accumulation of HMW proinsulin complexes

Because ER stress is linked with oxidative stress (Han et al., 2015; Malhotra et al., 2008), we tested whether Pdia1 deletion confers increased sensitivity to oxidants by treating islets (isolated after 30 wks of HFD) with the vitamin K analog menadione (Criddle et al., 2006; Loor et al., 2010), which can render cells susceptible to apoptosis by increasing cytosolic calcium (Gerasimenko et al., 2002) with nuclear condensation (Wyllie et al., 1984). Menadione treatment significantly increased ROS as observed by CellROX Deep Red stain (Figure 6A) and ß cells with Pdia1 deletion showed greater ROS accumulation than those of genetic control fl/+ islets (Figure 6B). Furthermore, both the average size and the histogram-analyzed nuclear size after Hoechst 33342 staining demonstrated that menadione promoted nuclear condensation (Figure 6C) and this effect appeared to be greater in KO islets compared to the genetic controls (Figure 6D). To uncover how Pdia1 deletion may affect the sensitivity of proinsulin maturation to perturbation by oxidants, we performed Western blotting of islet lysates under reducing and non-reducing conditions. Consistently, in the absence of menadione treatment, Pdia1 deletion increased the HMW proinsulin complexes (a, 49–198 kDa) compared to WT islets, and this effect was even greater in islets treated with menadione (Figure 6E). Importantly, DTT treatment of the gel prior to nitrocellulose transfer demonstrated no difference between untreated and NEM-treated islets (Figure 6F). Taken together, these results show that ß cells lacking Pdia1 are more sensitive to oxidizing conditions that promote the formation of HMW proinsulin complexes.

Figure 6 with 1 supplement see all
Pdia1 deletion increases sensitivity to menadione oxidant: Increased ROS, nuclear condensation, and HMW proinsulin complexes were observed in menadione-treated KO islets.

A. Islets isolated from mice after 30 wks HFD were treated with or without Menadione (10 µM, 3 hr) and co-stained with CellROX Deep Red (red) and Hoechst 33342 (blue). Live islet images were obtained by an Opera Phenix high content screening system (63X objective lens) and seven z-stack images (1 µm interval) were combined. Scale bar, 20 μm. genetic controls; n = 3, KO; n = 3. (B) Quantification of ROS mean intensity is shown. CellROX Deep Red mean intensity (divided by area) was measured by image J software. Mean ± SEM, p<0.001***. (C). Quantification of nuclear mean area (µm2) measured in Hoechst 33342 stained images by ImageJ software is shown. Mean ± SEM, p<0.001***. (D). Histogram analysis of nuclear sizes is shown. Percent frequencies are indicated in the graph. (E) Western blot of islets isolated from mice after 37 wks of HFD by SDS-PAGE under reducing and non-reducing conditions is shown. After overnight recovery, islets were treated with menadione (100 µM) for 1 hr. Islet preparations from five independent control and KO mice were performed and representative images are shown. Quantification of the ratio of HMW proinsulin complexes (a) to monomeric proinsulin under reducing conditions is shown in graph (lower). Mean ± SEM, p<0.05*, p<0.01**, p<0.001***. Controls; n = 5, KO; n = 5 mice. (F) WT murine islets were treated with Menadione (100 µM) for 1 hr, treated with or without NEM as in Figure 5E, and lysates were prepared and analyzed under non-reducing conditions.

https://doi.org/10.7554/eLife.44528.013

Proinsulin accumulation in the ER increases oligomeric and HMW complexes

Our studies suggest that PDIA1 is required to prevent formation of HMW proinsulin complexes. Our complementary study (Arunagiri et al., 2019) demonstrated that increased proinsulin accumulation predisposes to oligomer and HMW complex formation. To test the notion that increased proinsulin expression, as in T2D, may exacerbate abnormal disulfide formation, we tested the effect of brefeldin A (BFA), which promotes retrograde COP1 trafficking from the cis-Golgi to the ER to prevent export of secretory proteins to the Golgi, and thus increasing their concentration in the ER. BFA treatment significantly increased proinsulin complex formation, thus supporting the notion that the aberrant multimers/HMW complexes are a consequence of increased proinsulin content in the ER (Figure 5—figure supplement 2). Treatment with the translation elongation inhibitor cycloheximide (CHX), did not significantly affect the results, other than an expected reduced proinsulin content, indicating the presynthesized proinsulin is subject to multimer/HMW complex formation when its abundance in the ER is increased.

PDIA1 is a reductase that facilitates proper proinsulin folding

To study the impact of PDIA1 in the absence of C peptide processing in a more manipulatable system, we analyzed human proinsulin expression delivered by adenovirus to murine embryonic fibroblasts (MEFs) that were co-infected with WT PDIA1 or PDIA1 with 4 Cys to Ser mutations in the two vicinal catalytic PDI sites (Wang et al., 2012). Infection with Ad-hProins produced significant amounts of reduced and oxidized proinsulin upon analysis by non-reducing SDS-PAGE with DTT treatment prior to transfer to nitrocellulose (Figure 5—figure supplement 3). Treatment of cells with increasing concentrations of DTT increased the amount of reduced proinsulin (lanes 3,4), as expected. Significantly, forced expression of PDIA1, but not catalytically inactive PDIA1, increased the amount of reduced proinsulin (lanes 5, 6, 8, 10 and 12). Therefore, the results support the hypothesis that PDIA1 acts as a reductase to prevent aberrant proinsulin oxidation, consistent with previous studies in vitro (Rajpal et al., 2012).

Pharmacological inhibition of PDIA1 recapitulates effects of Pdia1 gene deletion

Recently, Cole et al. (2018) described a selective PDIA1 inhibitor that covalently interacts with the A catalytic motif in PDIA1. Therefore, we tested the effect of this inhibitor in murine WT islets. We also studied the impact of oxidant menadione treatment. The results show that the KSC-34 inhibitor alone slightly increased HMW complexes. However, when combined with menadione, KSC-34 treatment recapitulated the effect of Pdia1 deletion upon menadione treatment (Figure 6—figure supplement 1). Importantly, the findings demonstrate that either a PDIA1 chemical inhibitor or gene deletion exhibit similar effects on the generation of HMW complexes, especially upon oxidant treatment.

Discussion

PDIA1 is the major ER oxidoreductase in the majority of mammalian cells, including ß cells. Although numerous in vitro studies demonstrated that PDI actively engages proinsulin to catalyze disulfide bond formation (Winter et al., 2011; Rajpal et al., 2012; Winter et al., 2002; Wright et al., 2013), there is little information regarding the significance of PDI action in vivo. Here, using ß cell specific Pdia1 deletion we show that PDIA1 is increasingly important for insulin production in the face of either age or metabolic stress imposed by a HFD. Specifically, when compromised by HFD feeding, mice with ß cell-specific Pdia1 deletion displayed exaggerated glucose intolerance with significant ß cell abnormalities including diminished islet and serum insulin accompanied by an increased proinsulin/insulin ratio in islets and serum (Figure 3), with diminished insulin packaging and storage in secretory granules and a reduced number of insulin secretory granules (Figure 4). In addition, ß cell Pdia1 deletion caused abnormal ultrastructural changes including ER distension and vesiculation, mitochondrial swelling, and nuclear condensation (Figure 4). Pdia1-deleted islets were also sensitive to oxidant challenge (Figure 6), which is significant because PDIA1 is a highly abundant ER protein (~mM concentration [Freedman et al., 1994]) that is primarily in a reduced form (Hudson et al., 2015) and although it cycles, it may significantly contribute to redox homeostasis in the ER.

PDIA1 might assist proinsulin folding by facilitating proper intramolecular disulfide bond formation, yet to date, there is no direct evidence to support this notion (Rajpal et al., 2012). Alternatively, PDIA1 may reduce improper proinsulin disulfide bonds as demonstrated during infection with pathogens that require reduction for retro-translocation from the ER to the cytosol (Tsai et al., 2001; Inoue et al., 2015) and from evidence that supports a role for PDI as a reductase important for degradation of mutant Akita proinsulin (He et al., 2015) and mutant thyroglobulin (Forster et al., 2006).

Proinsulin disulfide maturation in the ER is absolutely required for proinsulin export to the Golgi complex for delivery to immature granules (Haataja et al., 2016). The identification of mutations in the INS gene coding sequence increased our understanding how proinsulin misfolding contributes to the development of ß cell failure and diabetes (Liu et al., 2010). If PDIA1 assists in disulfide isomerization to facilitate correct disulfide bonding in proinsulin, its absence could increase aberrant disulfide-linked proinsulin complexes, which we observed in Pdia1-deleted islets (Figure 5). Although the role of PDIA1 in vivo is complicated by the presence of other ER oxidoreductases and glutathione, based on the greater accumulation of HMW disulfide-linked proinsulin complexes in Pdia1-null islets, the data strongly suggest that PDIA1 participates in the resolution/dissolution of these inappropriate disulfide-linked complexes (Figure 7). This could include both PDI chaperone function as well as oxidoreductase function. It is important to note that healthy murine WT islets also exhibit a lower level of these HMW complexes in addition to smaller oligomeric disulfide-linked proinsulin species (Arunagiri et al., 2019). Future studies will characterize the fate of these complexes, and how PDIA1 plays a role in limiting their accumulation and their potential contributions to relative insulin deficiency and ß cell failure, which are phenotypes known to be associated with T2D. Importantly, the increased BiP expression in the islets bearing Pdia1-deleted ß cells suggests that increased accumulation of inappropriate disulfide-linked proinsulin complexes may induce ER stress, supporting a link between aberrant disulfide-linked proinsulin complexes and a compromise in ß cell health and function.

The role of PDIA1 in proinsulin disulfide bond formation.

Here, we show that PDIA1 is not required but increases the efficiency of proinsulin maturation, possibly by reducing HMW proinsulin complexes. In the absence of PDIA1, disulfide bond formation, reduction and/or isomerization in proinsulin are inadequate during increased demand so HMW proinsulin complexes accumulate in the ER upon metabolic pressure, such as a HFD. The figure depicts the formation of proinsulin HMW complexes with cellular proteins and the role of PDIA1 in interconversions to reduced proinsulin with subsequent oxidation to dimer/oligomeric proinsulin and oxidized folded proinsulin.

https://doi.org/10.7554/eLife.44528.015

Based on our results, we conclude that PDIA1 optimizes proinsulin maturation needed for insulin biosynthesis and for the maintenance of normoglycemia under conditions of metabolic stress in vivo. Our findings should open a new field of investigation to elucidate how the PDI family members impact oxidative protein folding and redox homeostasis in the ER of pancreatic ß cells, as well as provide a fundamental basis to understand how protein folding is essential to protect ß cells from collapse, contributing to the onset of T2D.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Gene (M. musculus)Pdia1/P4hbNAMGI:97464
Strain, strain background (M. musculus)Pdia1 fl/flHahm et al., 2013
Strain, strain background (M. musculus)RIP-CreERTDor et al., 2004
Strain, strain background (M. musculus)Pdia1 fl/fl;RIP-CreERTthis paperC57BL/6 mice with Pdia1 floxed alleles were obtained from Dr. J Cho (Univ. of Illinois-Chicago) and crossed with Rat Insulin Promoter (RIP-CreERT) transgenic mice. Congenic Pdia1 gene floxed littermates with or without the CreERT transgene were used for in vivo experiments.
Cell line (M. musculus)WT MEFthis paperWT MEFFreshly prepared primary WT MEFs prior to passage #5 were used for experiments.
Adenovirushuman PDIWang et al., 2012
Adenovirushuman PDImutWang et al., 2012
Adenovirushuman Proinsthis paperHuman proinsulin was cloned into the pAd Easy system.
AntibodyMouse monoclonal anti-VinculinProteintechCat. #: 66305–1-lgWB (1:2000)
AntibodyRabbit monoclonal anti-GRP94Cell signaling TechnologyCat. #: 20292PWB (1:1000)
AntibodyMouse monoclonal anti-BiPBD BiosciencesCat. #: 610979WB (1:1000)
AntibodyRabbit polyclonal anti-PDIA1ProteintechCat. #: 11245–1-APWB (1:1000), IHC (1:200)
AntibodyRabbit polyclonal anti-PDIA4ProteintechCat. #:14712–1-APWB (1:1000)
AntibodyRabbit
polyclonal anti-PDIA6
ProteintechCat. #:18233–1-APWB (1:1000)
AntibodyMouse monoclonal anti-ProinsulinHyTest Ltd.Cat. #: 2PR8 (Mabs: CCI-17)WB (1:10000), IHC (1:200)
AntibodyGuinea pig polyclonal anti-Insulinthis paperProduced in house
WB (1:2000), IHC (1:200)
AntibodyMouse monoclonal anti-GlucagonAbcamCat. #: K79bB10IHC (1:200)
AntibodyIRDye 800CW Goat anti-Mouse IgG (H + L)Li-CorP/N: 926–32210WB (1:5000)
AntibodyIRDye 680RD Goat anti-Mouse IgG (H + L)Li-CorP/N: 926–68070WB (1:5000)
AntibodyIRDye 800CW Goat anti-Rabbit IgG (H + L)Li-CorP/N: 926–32211WB (1:5000)
AntibodyIRDye 680RD Goat anti-Rabbit IgG (H + L)Li-CorP/N: 926–68071WB (1:5000)
AntibodyIRDye 800CW Donkey anti-Guinea Pig IgG (H + L)Li-CorP/N: 926–32411WB (1:5000)
AntibodyIRDye 680RD Donkey anti-Guinea Pig IgG (H + L)Li-CorP/N: 926–68077WB (1:5000)
AntibodyAlexa Fluor 488 goat anti-rabbit IgGInvitrogenCat. #: A-11008IHC (1:500)
AntibodyAlexa Fluor 488 goat anti-mouse IgGInvitrogenCat. #: A-11001IHC (1:500)
AntibodyAlexa Fluor 594 goat anti-mouse IgGInvitrogenCat. #: A-11005IHC (1:500)
AntibodyAlexa Fluor 594 goat anti-guinea pig IgGInvitrogenCat. #: A-11076IHC (1:500)
Commercial assay or kitMercodia Rat/Mouse Proinsulin ELISAMercodia10-1232-01
Commercial assay or kitMercodia Mouse Insulin ELISAMercodia10-1247-01
Commercial assay or kitCellROX Deep Red reagentMolecular ProbesC10422
Chemical compound, drugDTTRoche diagnosticsProduct #.3117006001
Chemical compound, drugMenadioneAdipoGen Life ScienceAG-CR1-3631-G001
Chemical compound, drugCycloheximideSigma-AldrichC7698
Chemical compound, drugBrefeldin ACell signaling TechnologyCat. #: 9972
Chemical compound, drugKSC-34Cole et al., 2018Obtained from Dr. RL Wiseman (The Scripps Research Institute, La Jolla, CA)

Mice

C57BL/6 mice with Pdia1 floxed alleles were obtained from Dr. J Cho (Univ. of Illinois-Chicago) (Hahm et al., 2013) and crossed with Rat Insulin Promoter (RIP-CreERT) transgenic mice (Dor et al., 2004). Congenic Pdia1 gene floxed littermates with or without the CreERT transgene were used for in vivo experiments. Pdia1 deletion was performed by IP injection of the estrogen receptor antagonist Tamoxifen (Tam) (4 mg/mouse) three times a week. Male mice were pair-housed for the high fat diet (HFD) study. All procedures were performed by protocols and guidelines reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the SBP Medical Discovery Institute (AUF #. 17–066).

Generation and culture of primary mouse embryonic fibroblast (MEF)

Request a detailed protocol

WT C57BL/6 d14 embryos were isolated under sterile conditions and placed in 100 mm cell culture dish containing PBS. Placental and other maternal tissues were removed and the embryos were washed three times with PBS. Heads and visceral organs were removed. The heads were used for genotyping. Embryos were finely minced with a sterile razor blade and treated with 1 ml of 0.25% trypsin/EDTA (Corning) for 40 min at 37°C. DMEM medium (Corning) supplemented with 10% FBS, 1% penicillin/streptomycin, 100 µg/ml primocin and 1 mM sodium pyruvate was added to quench trypsin activity. Tissue homogenate was subsequently disaggregated by repeated pipetting, and further centrifuged to collect the fibroblast cell pallet. Cells were plated in 100 mm culture dish in DMEM medium at 37°C in a cell culture incubator under 5% CO2. To prevent mycoplasma, bacterial, and fungal contamination, primary MEFs were cultured in medium containing 100 µg/ml primocin (InvivoGen) and used for experiments prior to passage #5.

Glucose and insulin tolerance tests

Request a detailed protocol

Glucose tolerance tests were performed by IP injection of glucose (1 g/Kg body weight) into mice after fasting for 4 hr. For insulin tolerance tests, 1.5units/Kg of insulin was injected IP into mice after a 4 hr fast. Blood glucose levels were measured by tail bleeding at each time point indicated.

Measurement of serum proinsulin and insulin

Request a detailed protocol

Mice were fasted O/N and re-fed for 4 hr. Blood was collected by retro-orbital bleeding and serum was prepared by centrifugation. Serum proinsulin and insulin levels were measured by ELISA (Mercodia, 10-1232-01, 10-1247-01) according to the manufacturer’s protocol.

Islet isolation

Request a detailed protocol

Islets were isolated by collagenase P (Roche) perfusion as described (Sutton et al., 1986) following by histopaque-1077 (Sigma-Aldrich, Inc St. Louis) gradient purification. Islets were handpicked and studied directly or after overnight culture in RPMI 1640 medium (Corning 10–040-CV) supplemented with 10% FBS, 1% penicillin/streptomycin, 100 µg/ml primocin, 10 mM Hepes, and 1 mM sodium pyruvate.

Islet RNA isolation and qRT-PCR

Request a detailed protocol

Total RNAs were extracted from isolated islets by RiboZol Extraction reagent (VWR Life Science). cDNA was synthesized by iScript cDNA Synthesis kit (Bio-Rad Laboratories, Inc). The relative mRNA levels were measured by qRT-PCR with iTaq Universal SYBR green Supermix (Bio-Rad Laboratories, Inc). All primers are listed in Figure 3—source data 1.

Islet Western blotting

Request a detailed protocol

Isolated islets were lysed in RIPA buffer (10 mM Tris pH 7.4, 150 mM NaCl, 0.1% SDS, 1% NP-40, 2 mM EDTA) with protease and phosphatase inhibitors (Fisher Scientific) on ice for 10 min and lysates were collected after centrifugation at 4°C for 10 min at 12000 g. Samples were prepared in Laemmli sample buffer without (non-reducing) or with (reducing) 5% β-mercaptoethanol. After boiling for 5 min, samples were analyzed by SDS-PAGE (4–12% Bis-Tris gel, Bio-Rad Laboratories, Inc) and transferred to nitrocellulose membranes (Bio-Rad Laboratories, Inc). For DTT incubation prior to transfer, non-reduced samples were electrophoresed on a 12% Bis-Tris SDS gel and then incubated in 25 mM DTT for 10 min at RT. Primary antibodies were as follows: α-vinculin (Proteintech, 66305–1-lg), α-GRP94 (Cell Signaling, 20292P), α-BiP (BD Biosciences, 610979), α-PDIA1 (Proteintech, 11245–1-AP), α-PDIA4 (Proteintech, 14712–1-AP) α-PDIA6 (Proteintech, 18233–1-AP), α-proinsulin (HyTest Ltd., 2PR8, CCI-17). Guinea pig polyclonal α-insulin antibody was produced in-house. For secondary antibodies, goat α-mouse, goat α-rabbit, and donkey α-guinea pig antibodies were used in 1:5000 (Li-Cor, IRDye−800CW or IRDye−680RD).

Pancreas tissue Transmission Electron Microscopy

Request a detailed protocol

Samples were prepared according to the UCSD Cellular and Molecular Medicine Electron Microscopy Facility protocols. Mouse pancreata were perfused in modified Karnovsky’s fixative (2.5% glutaraldehyde and 2% paraformaldehyde (PFA) in 0.15M sodium cacodylate buffer, pH 7.4) and fixed for at least 4 hr, post-fixed in 1% osmium tetroxide in 0.15M cacodylate buffer for 1 hr and stained en bloc in 2% uranyl acetate for 1 hr. Samples were dehydrated in ethanol, embedded in Durcupan epoxy resin (Sigma-Aldrich, Inc St. Louis), sectioned at 50 to 60 nm on a Leica UCT ultramicrotome, and delivered to Formvar and carbon-coated copper grids. Sections were stained with 2% uranyl acetate for 5 min and Sato's lead stain (Sato, 1968) for 1 min. Images were obtained using a Tecnai G2 Spirit BioTWIN transmission electron microscope equipped with an Eagle 4 k HS digital camera (FEI, Hilsboro, OR) with indicated magnifications. Insulin granule numbers were counted manually on images taken at 4800X magnification and divided by islet area measured by image J software. Insulin mature granule vesicles and dense core sizes were also measured by image J software.

Pancreas immunohistochemistry

Request a detailed protocol

Pancreata were harvested and fixed in 4% PFA. Paraffin embedding, sectioning, and slide preparations were done in the SBP Histopathology Core Facility. Sections were stained with the following antibodies; α-glucagon (Abcam, K79bB10), α-PDIA1 (Proteintech, 11245–1-AP), α-proinsulin (HyTest Ltd., 2PR8, CCI-17), and DAPI (Fisher Scientific). Guinea pig polyclonal α-insulin antibody was produced in-house. For secondary antibodies, Alexa Fluor 488 goat α-rabbit IgG, Alexa Fluor 488 goat α-mouse IgG, Alexa Fluor 594 goat α-mouse IgG, and Alexa Fluor 594 goat α-guinea pig IgG antibodies were used (Invitrogen). Images were taken by Zeiss LSM 710 confocal microscope with a 40X objective lens. Scale bar, 20 µm.

For ß cell area measurement, pancreata were harvested, fixed in 4% PFA and embedded in paraffin. Three sections were prepared at 200 µm intervals for each pancreas and stained with guinea pig polyclonal insulin antibody and DAPI. Alexa Fluor 594 goat α-guinea pig IgG was used as a secondary antibody. Images were taken by an Aperio FL Scanner (Leica). Insulin stained ß cell area, islet area, and pancreas area were measured by Aperio Imagescope software.

Analysis of oxidative stress

Request a detailed protocol

Islets were plated onto CellCarrier-96 ultra microplates (Perkin Elmer) in phenol red-free RPMI 1640 medium supplemented with 10% FBS, 1% penicillin/streptomycin, 100 µg/ml primocin, 10 mM Hepes, and 1 mM sodium pyruvate one day prior to staining. Islets were treated with menadione (AdipoGen Life Science, 10 µM) for 3 hr at 37°C. Islets were stained with CellROX Deep Red reagent (Molecular Probes, C10422, 100 µM) and Hoechst 33342 (Invitrogen, 10 µg/ml) for 1 hr at the same time. Islets were washed three times with HBSS and incubated in HBSS for imaging while temperature and CO2 were controlled. Images were obtained by an Opera Phenix high content screening system (63X objective lens) in the SBP High Content Screening (HCS) Facility and seven z-stack images (1 µm interval) were combined.

Statistical analysis

Request a detailed protocol

Data are indicated as Mean ± SEM. Statistical significance was evaluated by unpaired two-tailed Student’s t test. P-values are presented as *p<0.05, **p<0.01, ***p<0.001.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
    Islet amyloid, increased A-cells, reduced B-cells and exocrine fibrosis: quantitative changes in the pancreas in type 2 diabetes
    1. A Clark
    2. CA Wells
    3. ID Buley
    4. JK Cruickshank
    5. RI Vanhegan
    6. DR Matthews
    7. GJ Cooper
    8. RR Holman
    9. RC Turner
    (1988)
    Diabetes Research 9:151–159.
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
    Characteristic, activity and conformational studies of [A6-Ser, A11 -Ser]-insulin
    1. Y Dai
    2. J-G Tang
    (1996)
    Biochimica Et Biophysica Acta (BBA) - Protein Structure and Molecular Enzymology 1296:63–68.
    https://doi.org/10.1016/0167-4838(96)00054-4
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
    Menadione-induced apoptosis: roles of cytosolic ca(2+) elevations and the mitochondrial permeability transition pore
    1. JV Gerasimenko
    2. OV Gerasimenko
    3. A Palejwala
    4. AV Tepikin
    5. OH Petersen
    6. AJ Watson
    (2002)
    Journal of Cell Science 115:485–497.
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
    Protein disulfide isomerase exhibits chaperone and anti-chaperone activity in the oxidative refolding of lysozyme
    1. A Puig
    2. HF Gilbert
    (1994)
    The Journal of Biological Chemistry 269:7764–7771.
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
    A modified method for lead staining of thin sections
    1. T Sato
    (1968)
    Journal of Electron Microscopy 17:158–159.
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
    Metabolic stress and compromised identity of pancreatic beta cells
    1. A Swisa
    2. B Glaser
    3. Y Dor
    (2017)
    Frontiers in Genetics, 8, 10.3389/fgene.2017.00021, 28270834.
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63

Decision letter

  1. Vivek Malhotra
    Senior Editor; The Barcelona Institute of Science and Technology, Spain
  2. Reid Gilmore
    Reviewing Editor; University of Massachusetts Medical School, United States
  3. Raghu Mirmira
    Reviewer; Indiana University, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "PDIA1/P4HB is required for efficient proinsulin maturation and ß cell health in response to diet induced obesity" for consideration by eLife. Your article has been reviewed by a Senior Editor (Vivek Malhotra), a Reviewing Editor, and three reviewers.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The role of protein disulfide isomerases in the folding of insulin is of great interest. Jang et al., generated mice that have a β cell-specific knockout of one of several PDIs that are expressed in islets. The impact of the knockout on blood glucose levels was relatively subtle in the absence of a high fat diet. The reviewers raised major concerns about the manuscript that are unlikely to be addressed in the time period allowed for revisions. The three reviews are included below.

Reviewer #1:

The manuscript from Jang and colleagues investigates the impact of an islet-specific knockout of the PDIA1 gene on the biosynthesis of insulin in mice. PDIA1 is the most abundant PDI family member expressed in the β cells of islets. The central finding of the manuscript is that a deficiency in PDIA1 results in a reduction in insulin secretion and an increase in the proinsulin/insulin ratio that is manifested in mice that are fed a high fat diet. The reduction in insulin secretion in the knockout mice causes an elevation of blood glucose and glucose intolerance. β cells show abnormal morphology including nuclear condensation, ER engorgement, mitochondrial swelling and decreased granule formation. A more severe impact of Pdia1 knockout is probably not observed due to the expression of other PDI family members in the knockout cells. While this manuscript is of potential interest to the readership of eLife there are several major weaknesses.

Essential revisions:

1) The major weakness of this manuscript is the gel analysis of the disulfide linked proinsulin complexes. The gel patterns shown in Figure 5C, Figure 5D, Figure S2A and Figure 6E show highly variable extents of disulfide-linked complexes, particularly in wild type islets. It seems possible that disulfide scrambling has occurred during sample preparation. A standard method to prevent disulfide scrambling is to treat cells with N-ethyl maleimide prior to cell lysis, to block free thiols. In Fig, 5C monomeric proinsulin can't account for more than a few% of total proinsulin in either wild type or KO cells. The proinsulin bands migrating between 14-49 kDa (dimers to pentamers?) appear to be similar in abundance to larger aggregates (49-198 kDa). In Figure 6E, monomeric insulin appears to be a higher% of total in both wild type and KO cells in the absence of Menadione. The dimer to pentamer category is minor in Figure 6E relative to the large aggregates. Quantification of relative protein levels (graphs in Figure 5C, etc.) may be useful for WT and KO comparisons, but it obscures the more serious problem that disulfide linked complexes are predominant in wild type islets. Disulfide-linked proinsulin aggregates are predominant even when wild type mice are fed a standard diet (Figure 5D). These results are inconsistent with the results/conclusions presented in the accompanying manuscript.

2) The X-axis in Figure 6D is confusing since the labels do not point at a tick mark. We realize that nuclei sizes have been binned into 10-integer ranges, but the offset between the labels and tick marks makes it unclear whether the bin with the highest frequency in the upper graph covers 35 to 45, or 40-50. It is also unclear how the mean value for nuclear size can be so severely left-shifted relative to the frequency distribution. For example, the mean value for vehicle treated cells is shown to fall in the bin labeled 30, but 90% of nuclei fall in bins with higher nuclear sizes. How can the mean of a sample not have a more obvious relationship to the frequency distribution? Is this due to an error in calculation of the mean? Perhaps this graph would be improved by using a smaller bin size.

3) The legend to Figure 7A contains excess information. This figure legend reads as if it was copied from another document that the authors have written (early draft of the introduction?), as it contains two citations and is overburdened with information that does not correlate with the figure. For example, why does the legend indicate that β cells express two Ero1 proteins, when a single ERO1 is shown in panel A? Why is there text about Akita diabetic mice in the legend to Figure 7A? The diagram for the disulfide-linked PI complexes is vague. Are the blue circles supposed to be other proinsulin molecules, or are these supposed to be other ER proteins with a free thiol? The impression that Figure 7 was tacked onto the manuscript as an afterthought is also apparent from the in text citations in the Discussion section. Figure 7 is mentioned twice, but the text does not indicate which panel. The citations to Figure 7 that are in the discussion could just as well cite the experimental figures as the diagram in Figure 7B isn't that informative.

Reviewer #2:

Jang et al., studied the role of protein disulfide isomerase A1 (PDIA1) in the regulation of β cell function. They generated β cell-specific Pdia1 knockout mice and found that feeding a high-fat diet (HFD) induced glucose intolerance along with hypoinsulinemia and increased proinsulin/insulin ratio compared to similarly treated wildtype mice. PDIA1 deficiency in β cells induced morphological alterations in the ER and mitochondria and impaired the maturation of secretory granules; this was associated with accumulation of misfolded proinsulin aggregates and islet vulnerability to oxidative stress. Collectively, these findings suggest that PDIA1, which is required for proinsulin maturation and insulin production, is involved in the regulation of ER redox state. This manuscript together with the accompanying paper by the Arvan group support the hypothesis that ER dysfunction with subsequent generation of misfolded proinsulin aggregates plays an important role in the pathophysiology of diabetes. However, there are several major issues that should be addressed.

Specific comments:

1) Oxidoreductases are important for disulfide bonds formation; however, the impact of PD1A1 knockout is relatively small. Thus, glucose homeostasis in young animals was normal; mice developed β cell dysfunction only after prolonged metabolic challenge and probably aging. KO mice developed modest fasting hyperglycemia even on a HFD for 5 months. In the re-feeding experiment (Figure 3D), while proinsulin/insulin ratio was increased, blood glucose was still not increased. In addition, there was no effect on the expression of proinsulin, β cell transcription factors, pancreatic convertases, oxidative stress and UPR genes (except for GRP94). What is the relative importance of PDIA1 vs all other disulphide isomerases and redox enzymes? This question should be addressed.

2) The conclusion that KO mice develop hyperglycemia during aging is based on two KO mice (Figure 2D and E). It is not possible to perform statistical comparison with such a small number of animals. The number of animals in each group should be increased. What were blood glucose levels in aged male Pdia1 KO mice compared to controls? This part of the figure should be either extended or deleted.

3) The quantifications in Figure 5B and 5C were performed on 2 WT and 3 KO mice only. Statistical comparison with such a small number of animals is not possible. Note that in Figure 5C (lane #1, WT) the amount of proinsulin complexes is much higher than the monomeric form of proinsulin and is even greater than that in KO mice. In Figure 5C, the total proinsulin level in KO islets was higher than in WT controls (reducing gel). Is there greater affinity of the proinsulin antibody to misfolded aggregates compared to native proinsulin? These findings should be explained.

4) The pulse-chase experiments showed that PDIA1 KO did not affect proinsulin conversion to insulin (Supplementary Figure 3); this is inconsistent with the authors' suggestion that PDIA1 deficiency impairs proinsulin folding and maturation.

5) It would be expected that impairment of ER redox state by deleting PDIA1 generates oxidative stress in response to metabolic challenge; however, ROS production was similar in wildtype and KO mice on HFD (Figure 7A). ROS production along with nuclear condensation was increased in KO mice only after treatment with a potent pro-oxidant. This raises concerns regarding the importance of PDIA1 in the regulation of ER homeostasis in β cells.

Reviewer #3:

The manuscript by Jang et al., describes studies on the β cell-specific knockout of Pdia1 gene and its effects on metabolic homeostasis and proinsulin processing. The authors used an Ins-CreER transgene to inducibly delete Pdia1 in β cells, then fed mice and their littermate controls a 45% HFD to increase cellular demand on insulin production. They observed glucose intolerance/diabetes, elevated proinsulin/insulin ratios, and evidence of dysmorphic insulin granules in KO mice. The authors conclude that the absence of a critical oxidoreductase to properly guide disulfide bond formation results in an ER stress-like state that leads to β cell dysfunction. The study is of clear importance to the β cell field and emphasizes the critical role of the ER and proinsulin folding in β cell function and glucose homeostasis; it furthers the narrative advanced by the seminal work of this group over the years. However, there are two major issues, and two moderate ones that require further experimentation and consideration:

Major comments:

1) Perhaps the biggest concern with the study is the controls against which the KOs are compared. The legend to Figure 1 states that the "WT" mice are control littermates with one or two floxed Pdia1 alleles. Technically, these are not WT mice, but that is a minor concern that can be easily corrected. The more significant concern is that no Cre+ controls were used. As the authors are aware, particularly with the RIP transgene, Cre+ mice can exhibit glucose intolerance due to β cell stress induced by the transgene. In the absence of this control, the overall conclusions of the study remain uninterpretable, since only the KO mice bear this transgene. I am not arguing that the results of this study are invalid in the absence of the controls, but the magnitude of the effects might not be as great as the authors purport.

2) The study is that it takes a somewhat biased perspective from start to finish. The authors chose this gene for study based on their understanding of its role in ER protein folding. As such, many of the studies presented in the manuscript are leveraged to support this bias. It would have been more convincing if the authors had performed unbiased RNA-Seq studies from the islets, rather than targeted gene RT-PCR. What might be happening to β cell differentiation genes? Or to a more comprehensive list of other oxidoreductases? Or to genes involved in Ca regulation (e.g. SERCA, Ryanodyne R, etc.)? Oxidative stress pathways? A more comprehensive bioinformatics pathway analysis (e.g. Gene Ontology) could address these questions and provide more unbiased evidence to support the authors' hypothesis. Of course, the comparator would have to be done against Cre+ controls, as noted above.

3) The RIP-Cre transgene is known to be expressed in the brain (hypothalamus), and this could affect glucose homeostasis, as has been noted previously. The authors should probably test the mRNA levels of Pdia1 in hypothalamus to confirm or refute this contention. Therefore, the use of Cre+ controls as noted above becomes especially more important.

4) The authors state that UPR genes are not significantly altered, but that there is ER stress occurring in the β cells (based both on the increase in BiP and the morphology in the EM images). Granted that the UPR is a response to ER stress, but does this disconnect refer to a defective UPR or simply that the UPR has failed at this stage of analysis and ER stress predominates?

https://doi.org/10.7554/eLife.44528.018

Author response

Reviewer #1:

The manuscript from Jang and colleagues investigates the impact of an islet-specific knockout of the PDIA1 gene on the biosynthesis of insulin in mice. PDIA1 is the most abundant PDI family member expressed in the β cells of islets. The central finding of the manuscript is that a deficiency in PDIA1 results in a reduction in insulin secretion and an increase in the proinsulin/insulin ratio that is manifested in mice that are fed a high fat diet. The reduction in insulin secretion in the knockout mice causes an elevation of blood glucose and glucose intolerance. β cells show abnormal morphology including nuclear condensation, ER engorgement, mitochondrial swelling and decreased granule formation. A more severe impact of Pdia1 knockout is probably not observed due to the expression of other PDI family members in the knockout cells. While this manuscript is of potential interest to the readership of eLife there are several major weaknesses.

Essential revisions:

1) The major weakness of this manuscript is the gel analysis of the disulfide linked proinsulin complexes. The gel patterns shown in Figure 5C, Figure 5D, Figure S2A and Figure 6E show highly variable extents of disulfide-linked complexes, particularly in wild type islets. It seems possible that disulfide scrambling has occurred during sample preparation. A standard method to prevent disulfide scrambling is to treat cells with N-ethyl maleimide prior to cell lysis, to block free thiols. In Fig, 5C monomeric proinsulin can't account for more than a few% of total proinsulin in either wild type or KO cells. The proinsulin bands migrating between 14-49 kDa (dimers to pentamers?) appear to be similar in abundance to larger aggregates (49-198 kDa). In Figure 6E, monomeric insulin appears to be a higher% of total in both wild type and KO cells in the absence of Menadione. The dimer to pentamer category is minor in Figure 6E relative to the large aggregates. Quantification of relative protein levels (graphs in 5C, etc.) may be useful for WT and KO comparisons, but it obscures the more serious problem that disulfide linked complexes are predominant in wild type islets. Disulfide-linked proinsulin aggregates are predominant even when wild type mice are fed a standard diet (Figure 5D). These results are inconsistent with the results/conclusions presented in the accompanying manuscript.

First, we were aware of this variability and devoted considerable effort to optimize the analysis to obtain more reproducible non-reducing gels. As the reviewer suggested we have treated islets with NEM prior to lysis and demonstrated that alkylation did not significantly alter the pattern of oligomers and HMW complexes, although NEM treatment did result in a relative decrease in homo-oligomers of proinsulin (mw 14-49 kDa) with a corresponding increase in HMW complexes (shown in New Figure 5E). One possible interpretation of the findings is that PDIA1 is required to reduce HMW proinsulin complexes and NEM treatment inactivates PDIA1, thereby stabilizing the HMW complexes. We also provide independent evidence in support of this conclusion, using a PDI-specific inhibitor (see below).

The reviewer was also concerned that we detected HMW complexes in wildtype islets. Indeed, we do detect disulfide-linked oligomers in wildtype islets. The major conclusion from the accompanying paper by Arungagiri et al., is that oligomers are detected in healthy human islets (Figure 3A,B), in wildtype murine islets and increase with increased proinsulin synthesis (Figure 3C) and are detected in db/db islets prior to onset of hyperglycemia (Figure 5C). The ability to detect these disulfide-linked complexes in wildtype and healthy islets is due to epitope exposure that is not detected with our polyclonal anti-insulin antibodies. As we previously demonstrated, addition of DTT to islets in vitro collapsed the multimers down to monomeric proinsulin and an apparently resistant dimer migrating at 16 kDa (New Figure 5F). Importantly, treating the non-reducing gel with DTT prior to transfer greatly increases detection of the monomeric proinsulin indicating that epitope exposure to this antibody increases upon proinsulin reduction (New Figure 5D,E). The increase in HMW complexes is also evident in PDIA1-deficient β cells and even more so when the islets are exposed to the oxidant menadione (New Figure 6F). Analysis of these results also demonstrates significant reproducibility of the data from islet preparations from 6 individual mice (Figure 5C,D).

The corresponding description is in subsection “KO islets have an increased intracellular proinsulin/insulin ratio with accumulation of high molecular weight (HMW) proinsulin complexes” and subsection “Oxidant treatment of Pdia1 KO islets increases accumulation of HMW proinsulin complexes”.

2) The X-axis in Figure 6D is confusing since the labels do not point at a tick mark. We realize that nuclei sizes have been binned into 10-integer ranges, but the offset between the labels and tick marks makes it unclear whether the bin with the highest frequency in the upper graph covers 35 to 45, or 40-50. It is also unclear how the mean value for nuclear size can be so severely left-shifted relative to the frequency distribution. For example, the mean value for vehicle treated cells is shown to fall in the bin labeled 30, but 90% of nuclei fall in bins with higher nuclear sizes. How can the mean of a sample not have a more obvious relationship to the frequency distribution? Is this due to an error in calculation of the mean? Perhaps this graph would be improved by using a smaller bin size.

We thank the reviewer for this constructive suggestion. We have redrawn the Figure and placed the ‘tik’ marks to more accurately reflect the different bins. We agree that inclusion of the mean nuclear areas from Figure 6C into the graphs in Figure 6D generated confusion and they have now been removed.

3) The legend to Figure 7A contains excess information. This figure legend reads as if it was copied from another document that the authors have written (early draft of the introduction?), as it contains two citations and is overburdened with information that does not correlate with the figure. For example, why does the legend indicate that β cells express two Ero1 proteins, when a single ERO1 is shown in panel A? Why is there text about Akita diabetic mice in the legend to Figure 7A? The diagram for the disulfide-linked π complexes is vague. Are the blue circles supposed to be other proinsulin molecules, or are these supposed to be other ER proteins with a free thiol? The impression that Figure 7 was tacked onto the manuscript as an afterthought is also apparent from the in text citations in the discussion. Figure 7 is mentioned twice, but the text does not indicate which panel. The citations to Figure 7 that are in the discussion could just as well cite the experimental figures as the diagram in Figure 7B isn't that informative.

Thank you for picking up on this! Indeed, the legend to Figure 7 was inadvertently included from an earlier draft of the manuscript and we have now deleted it and revised the model. Specifically, we have omitted Figure 7B and only focus on the points in Figure 7A that are most relevant to our findings in this manuscript regarding the pathway of proinsulin oxidation, disulfide-linked oligomer and HMW complex formation. Our results provide new information suggesting a role of PDIA1 in the inter-conversion of oligomeric, HMW proinsulin complexes and native folded proinsulin.

Reviewer #2:

Jang et al., studied the role of protein disulfide isomerase A1 (PDIA1) in the regulation of β cell function. They generated β cell-specific Pdia1 knockout mice and found that feeding a high-fat diet (HFD) induced glucose intolerance along with hypoinsulinemia and increased proinsulin/insulin ratio compared to similarly treated wildtype mice. PDIA1 deficiency in β cells induced morphological alterations in the ER and mitochondria and impaired the maturation of secretory granules; this was associated with accumulation of misfolded proinsulin aggregates and islet vulnerability to oxidative stress. Collectively, these findings suggest that PDIA1, which is required for proinsulin maturation and insulin production, is involved in the regulation of ER redox state. This manuscript together with the accompanying paper by the Arvan group support the hypothesis that ER dysfunction with subsequent generation of misfolded proinsulin aggregates plays an important role in the pathophysiology of diabetes. However, there are several major issues that should be addressed.

Specific comments:

1) Oxidoreductases are important for disulfide bonds formation; however, the impact of PD1A1 knockout is relatively small. Thus, glucose homeostasis in young animals was normal; mice developed β cell dysfunction only after prolonged metabolic challenge and probably aging. KO mice developed modest fasting hyperglycemia even on a HFD for 5 months. In the re-feeding experiment (Figure 3D), while proinsulin/insulin ratio was increased, blood glucose was still not increased. In addition, there was no effect on the expression of proinsulin, β cell transcription factors, pancreatic convertases, oxidative stress and UPR genes (except for GRP94). What is the relative importance of PDIA1 vs all other disulphide isomerases and redox enzymes? This question should be addressed.

This reviewer brings up an important point as to what is the role of other oxidoreductases. We are performing such experiments as part of a future study, but to bolster our conclusions in the current study of PDIA1, we have added new data with KSC-34, an inhibitor that selectively interacts with the A catalytic site of PDIA1 (Cole et al., 2018. We show that KSC-34 increases disulfide-linked HMW proinsulin complexes in islets and also renders the islets susceptible to the oxidant Menadione, with a marked further increase in HMW disulfide-linked proinsulin-containing complexes (New Figure 6—figure supplement 1). These behaviors phenocopy the effect of Pdia1 deletion. These data provide both pharmacological and genetic support for our findings which indicate that PDIA1 facilitates proinsulin maturation by minimizing the accumulation of HMW disulfide linked proinsulin-containing complexes. This is especially important upon metabolic challenge, such as high-fat diet. The corresponding description is in subsection “Pharmacological inhibition of PDIA1 recapitulates effects of Pdia1 gene deletion”.

Regarding the fast-refed blood glucose, it is not surprising there is no change in refed glucose levels. These young mice are likely more insulin sensitive and there are many additional factors that maintain blood glucose within a narrow range. This was pointed out by Jim Johnson in studies of Proinsulin gene deletion that show after 12 days of proinsulin gene deletion there was no change in blood glucose (Szabat et al., 2016). To eliminate confusion, we have deleted these data because the findings are not relevant to our conclusions.

2) The conclusion that KO mice develop hyperglycemia during aging is based on two KO mice (Figure 2D and E). It is not possible to perform statistical comparison with such a small number of animals. The number of animals in each group should be increased. What were blood glucose levels in aged male Pdia1 KO mice compared to controls? This part of the figure should be either extended or deleted.

We agree with this reviewer that an n=2 (~350 day old females) cannot be used for statistics. Therefore, we removed Figure 2D,E. We have now analyzed an additional n=12 ~250 day old male mice fed normal chow and observed a statistically significant worsening of glucose tolerance in the old Pdia1-null mice supporting our original observation (New Figure 2D,E). At this point we cannot comment on whether the phenotype is associated with aging or due to metabolic consequences of age, which is consistent with the HFD fed mice. The corresponding description is in subsection “Generation of ß cell-specific Pdia1 deleted mice.”

3) The quantifications in Figure 5B and 5C were performed on 2 WT and 3 KO mice only. Statistical comparison with such a small number of animals is not possible. Note that in Figure 5C (lane #1, WT) the amount of proinsulin complexes is much higher than the monomeric form of proinsulin and is even greater than that in KO mice. In Figure 5C, the total proinsulin level in KO islets was higher than in WT controls (reducing gel). Is there greater affinity of the proinsulin antibody to misfolded aggregates compared to native proinsulin? These findings should be explained.

The original data in Figure 5A and C from n=2 and 3 mice has now been increased to 6 mice (3 WT and 3 KO) as biological replicates in Figures 5A (reduced), 5C (no DDT added to gel) and 5D (DTT added to gel). The results provide for significant statistical analysis (Figure 5B and 5D) and demonstrate reproducible trends between murine genotypes. The corresponding description is in subsection “KO islets have an increased intracellular proinsulin/insulin ratio with accumulation of high molecular weight (HMW) proinsulin complexes”.

Indeed, as this insightful reviewer suggested, the source for variation between monomeric and multimeric proinsulin, is that reduction of the sample greatly increases epitope exposure of monomeric proinsulin to the antibody. We didn't emphasize this issue in our study because the 2D gel analysis in the companion Arunagiri et al., paper shows that all bands in these complexes contain proinsulin. This key point is most evident by an experiment where we incubated the gel in 25mM DTT before transfer to nitrocellulose. (In addition, we now run 12% polyacrylamide gels as opposed to gradient gels to as much as possible equalize transfer conditions throughout the gel in order to obtain Westerns with greatly increased the signal for the proinsulin monomer relative to the multimers, in a more reproducible manner, as is evident in Figure 1C and Figures 5A,C,D).

Second, this reviewer was concerned about the detection of proinsulin complexes in WT islets. The major conclusion from the companion paper by Arunagiri et al., is that oligomers are detected in healthy human islets (Figure 3A,B), in wildtype murine islets, and further increase with increased proinsulin synthesis (Figure 3C) and are detected in db/db islets prior to onset of hyperglycemia (Figure 5C).

4) The pulse-chase experiments showed that PDIA1 KO did not affect proinsulin conversion to insulin (Supplementary Figure 3); this is inconsistent with the authors' suggestion that PDIA1 deficiency impairs proinsulin folding and maturation.

We agree with this reviewer that the pulse-chase studies did not provide significant insight into the defect in the Pdia1-null islets; thus, we have omitted this figure. Instead, we now include analysis of brefeldin A treatment which increases retrograde trafficking of proinsulin from the cis-Golgi back to the ER. BFA-treated WT islets significantly increased the multimeric forms of proinsulin (New Figure 6—figure supplement 1), suggesting proinsulin accumulation in the ER increases HMW complex formation. The corresponding description is on page 9 under “Proinsulin accumulation in the ER increases oligomeric and HMW complexes”.

5) It would be expected that impairment of ER redox state by deleting PDIA1 generates oxidative stress in response to metabolic challenge; however, ROS production was similar in wildtype and KO mice on HFD (Figure 7A). ROS production along with nuclear condensation was increased in KO mice only after treatment with a potent pro-oxidant. This raises concerns regarding the importance of PDIA1 in the regulation of ER homeostasis in β cells.

The reviewer is correct that we did not see increased ROS production, and thus the reviewer is also correct that our data indicate that the role of PDIA1 in the ER homeostasis of β cells does not appear to be primarily through the consumption of ROS. In our view, the central finding of our paper is that Pdia1 deletion renders β cells less resilient to metabolic challenge because this perturbation of ER homeostasis renders proinsulin more susceptible to formation/accumulation of high molecular weight disulfide-linked complexes, resulting in downstream consequences that include diminished islet insulin. This more aptly describes the importance of PDIA1 in the regulation of ER homeostasis in β cells.

Reviewer #3:

The manuscript by Jang et al., describes studies on the β cell-specific knockout of Pdia1 gene and its effects on metabolic homeostasis and proinsulin processing. The authors used an Ins-CreER transgene to inducibly delete Pdia1 in β cells, then fed mice and their littermate controls a 45% HFD to increase cellular demand on insulin production. They observed glucose intolerance/diabetes, elevated proinsulin/insulin ratios, and evidence of dysmorphic insulin granules in KO mice. The authors conclude that the absence of a critical oxidoreductase to properly guide disulfide bond formation results in an ER stress-like state that leads to β cell dysfunction. The study is of clear importance to the β cell field and emphasizes the critical role of the ER and proinsulin folding in β cell function and glucose homeostasis; it furthers the narrative advanced by the seminal work of this group over the years. However, there are two major issues, and two moderate ones that require further experimentation and consideration:

Major comments:

1) Perhaps the biggest concern with the study is the controls against which the KOs are compared. The legend to Figure 1 states that the "WT" mice are control littermates with one or two floxed Pdia1 alleles. Technically, these are not WT mice, but that is a minor concern that can be easily corrected. The more significant concern is that no Cre+ controls were used. As the authors are aware, particularly with the RIP transgene, Cre+ mice can exhibit glucose intolerance due to β cell stress induced by the transgene. In the absence of this control, the overall conclusions of the study remain uninterpretable, since only the KO mice bear this transgene. I am not arguing that the results of this study are invalid in the absence of the controls, but the magnitude of the effects might not be as great as the authors purport.

This reviewer makes the important point that we have not studied the impact of the RIP-CreERT allele in β cells. We actually did perform these studies which demonstrate the RIP-CreERT allele does not impact glucose homeostasis in mice with one floxed and one wt Pdia1 allele, but neglected to include this data in the original manuscript. We now include the control without RIP-CreERTwhich is similar to RIP-CreERT-containing mice (New Figure 1—figure supplement 1). The corresponding description is insubsection “Generation of ß cell-specific Pdia1 deleted mice”. Therefore, although the RIP-CreERT allele may impact β cell function, it does not significantly contribute to the phenotype we have characterized.

Finally, we respect the reviewer’s concern that characterizing the genotypes simply as WT and KO did not provide sufficient information regarding specifics of the genotype. Therefore, we have changed all WT and KO labelings to reflect genotypes in the figures.

We now label all mice with respect to the genotype of WT Pdia1, floxed Pdia1 and CreERTstatus as follows:

fl/fl:CreERT with Tam, equivalent to KO in the previous draft.

fl/+ and fe/fe without CreERT, equivalent to WT.

fl/+ and fl/- with CreERT and Tam, equivalent to het or WT.

2) The study is that it takes a somewhat biased perspective from start to finish. The authors chose this gene for study based on their understanding of its role in ER protein folding. As such, many of the studies presented in the manuscript are leveraged to support this bias. It would have been more convincing if the authors had performed unbiased RNA-Seq studies from the islets, rather than targeted gene RT-PCR. What might be happening to β cell differentiation genes? Or to a more comprehensive list of other oxidoreductases? Or to genes involved in Ca regulation (e.g. SERCA, Ryanodyne R, etc.)? Oxidative stress pathways? A more comprehensive bioinformatics pathway analysis (e.g. Gene Ontology) could address these questions and provide more unbiased evidence to support the authors' hypothesis. Of course, the comparator would have to be done against Cre+ controls, as noted above.

Thank you for the suggestion. As the reviewer may have surmised, we are currently working on an unbiased proteomics characteriziation of both ER oxidoreductases and proinsulin interactor proteins in human islets, so we very much intend to follow up on the reviewer’s point involving omics-based screening. However, that is quite a major undertaking and an entire project (and manuscript) unto itself. In the current manuscript, we think it is unlikely that mRNA-Seq would change our primary conclusion that Pdia1 deletion renders β cells less resilient to metabolic challenge because this perturbation renders proinsulin more susceptible to formation/accumulation of high molecular weight disulfide-linked complexes, resulting in downstream consequences that include diminished islet insulin.

However, in deference to the reviewer’s concern, we have now included additional qRT-PCR results for PDIA1, PDIA3, PDIA4, PDIA6 and Serca2B in the New Figure 3G. The corresponding description is on page 6 under “Pdia1 is not required for expression of ß cell-specific genes, antioxidant response genes or cell death genes”.

3) The RIP-Cre transgene is known to be expressed in the brain (hypothalamus), and this could affect glucose homeostasis, as has been noted previously. The authors should probably test the mRNA levels of Pdia1 in hypothalamus to confirm or refute this contention. Therefore, the use of Cre+ controls as noted above

becomes especially more important.

Indeed, another issue regarding RIP-CreERT is potential hypothalamic expression. We now show that WT and Pdia1 KO mice (n=6/genotype) express similar levels of PDIA1 by Western blotting of hypothalamic tissue (New Figure 1—figure supplement 1B). The animals also show similar body weights (Figure 1A). Therefore, although the RIP-CreERT allele may impact β cell function, it does not significantly contribute to the phenotype we have characterized. The corresponding description is in subsection “Generation of ß cell-specific Pdia1 deleted mice.”

In addition, we previously reported (Hassler et al., 2015) there was no difference between IRE1α WT and KO mice in serum dopamine, which is synthesized in the arcuate nucleus of the hypothalamus. Although we did detect Cre positive staining in brain sections in the Hassler study, there was little difference in growth hormone-releasing hormone (GHRH) expression between the KO mice and the controls.

4) The authors state that UPR genes are not significantly altered, but that there is ER stress occurring in the β cells (based both on the increase in BiP and the morphology in the EM images). Granted that the UPR is a response to ER stress, but does this disconnect refer to a defective UPR or simply that the UPR has failed at this stage of analysis and ER stress predominates?

We thank the reviewer for raising this point. Our new analysis of a greater number of islet preps from 3 KO and 3 WT mice on HFD now demonstrate significant increases in several UPR-induced proteins; BiP, PDIA4 and PDIA6, which is more consistent with the ER morphology. We have now stated this more clearly in the manuscript – thanks! The corresponding description is in subsection “KO islets have an increased intracellular proinsulin/insulin ratio with accumulation of high molecular weight (HMW) proinsulin complexes”.

https://doi.org/10.7554/eLife.44528.019

Article and author information

Author details

  1. Insook Jang

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7567-3849
  2. Anita Pottekat

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Juthakorn Poothong

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Jing Yong

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Formal analysis, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4970-408X
  5. Jacqueline Lagunas-Acosta

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  6. Adriana Charbono

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Zhouji Chen

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Conceptualization, Writing—review and editing
    Competing interests
    No competing interests declared
  8. Donalyn L Scheuner

    DLS Consulting, Greenfield, United States
    Contribution
    Conceptualization, Supervision, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Ming Liu

    Division of Metabolism Endocrinology and Diabetes, University of Michigan Medical School, Ann Arbor, United States
    Contribution
    Conceptualization, Writing—review and editing
    Competing interests
    No competing interests declared
  10. Pamela Itkin-Ansari

    Department of Pediatrics, University of California, San Diego, San Diego, United States
    Contribution
    Writing—review and editing
    Competing interests
    No competing interests declared
  11. Peter Arvan

    Division of Metabolism Endocrinology and Diabetes, University of Michigan Medical School, Ann Arbor, United States
    Contribution
    Conceptualization, Formal analysis, Writing—review and editing
    For correspondence
    parvan@umich.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4007-8799
  12. Randal J Kaufman

    Degenerative Diseases Program, SBP Medical Discovery Institute, La Jolla, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing—review and editing
    For correspondence
    rkaufman@sbpdiscovery.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4277-316X

Funding

National Institutes of Health (R01DK113171)

  • Randal J Kaufman

National Institutes of Health (R01DK103185)

  • Randal J Kaufman

National Institutes of Health (R24DK110973)

  • Pamela Itkin-Ansari
  • Peter Arvan
  • Randal J Kaufman

National Institutes of Health (P30 CA030199)

  • Randal J Kaufman

National Institutes of Health (P30 DK063491)

  • Randal J Kaufman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the UCSD EM and microscopy cores and the SBP Histopathology Core. Dr. J Cho graciously provided the floxed Pdia1 mice (U Illinois, Chicago, IL). Dr. RL Wiseman (The Scripps Research Institute, La Jolla, CA) kindly provided the PDIA1 inhibitor KSC-34. We thank Kimberly Nagle for assistance with preparation of the manuscript. Portions of this work were supported by NIH/NCI Grants R01DK113171 (RJK), R01DK103185 (RJK), R24DK110973 (RJK, PA, PIA) and the SBP NCI Cancer Center Grant P30 CA030199. RJK is a member of the UCSD DRC (P30 DK063491) and Adjunct Professor in the Department of Pharmacology, UCSD.

Ethics

Animal experimentation: All procedures were performed by protocols and guidelines reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the SBP Medical Discovery Institute (AUF#. 17-066).

Senior Editor

  1. Vivek Malhotra, The Barcelona Institute of Science and Technology, Spain

Reviewing Editor

  1. Reid Gilmore, University of Massachusetts Medical School, United States

Reviewer

  1. Raghu Mirmira, Indiana University, United States

Publication history

  1. Received: December 19, 2018
  2. Accepted: May 15, 2019
  3. Version of Record published: June 11, 2019 (version 1)

Copyright

© 2019, Jang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 511
    Page views
  • 90
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    Craig A Nash et al.
    Research Article
    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Travis J Morgenstern et al.
    Research Article Updated