Research ArticleStress Response

IRE1 prevents endoplasmic reticulum membrane permeabilization and cell death under pathological conditions

See allHide authors and affiliations

Sci. Signal.  23 Jun 2015:
Vol. 8, Issue 382, pp. ra62
DOI: 10.1126/scisignal.aaa0341

Abstract

The endoplasmic reticulum (ER) has emerged as a critical regulator of cell survival. IRE1 is a transmembrane protein with kinase and RNase activities that is localized to the ER and that promotes resistance to ER stress. We showed a mechanism by which IRE1 conferred protection against ER stress–mediated cell death. IRE1 signaling prevented ER membrane permeabilization mediated by Bax and Bak and cell death in cells experiencing ER stress. Suppression of IRE1 signaling triggered by its kinase activity led to the accumulation of the BH3 domain–containing protein Bnip3, which in turn triggered the oligomerization of Bax and Bak in the ER membrane and ER membrane permeabilization. Consequently, in response to ER stress, cells deficient in IRE1 were susceptible to leakage of ER contents, which was associated with the accumulation of calcium in mitochondria, oxidative stress in the cytosol, and ultimately cell death. Our results reveal a role for IRE1 in preventing a cell death–initializing step that emanates from the ER and provide a potential target for treating diseases characterized by ER stress, including diabetes and Wolfram syndrome.

INTRODUCTION

The endoplasmic reticulum (ER) is a membranous network in cells that is involved in multiple functions including production of secretory proteins, calcium storage, and regulation of cellular redox state (1). Homeostatic alterations in the ER play roles in the pathogenesis of chronic human disorders, such as type 1 and type 2 diabetes, myocardial infarction, stroke, and neurodegeneration, as well as inherited disorders including Wolfram syndrome, which is characterized by β cell death and neurodegeneration (2, 3). Under ER stress conditions, cell fate is controlled by the three major regulators of the unfolded protein response (UPR): inositol-requiring enzyme 1α (IRE1α), protein kinase R–like ER kinase (PERK), and activating transcription factor 6α (ATF6α) (4, 5). The opposing effects of IRE1α and PERK determine whether ER-stressed cells live or die. IRE1α activation confers protection against cell death through the regulated IRE1-dependent decay (RIDD) of death receptor 5 (DR5), whereas prolonged activation of PERK induces cell death mediated by CCAAT/enhancer-binding protein homologous protein (CHOP) and DR5 under pathological ER stress (6, 7). Bax- and Bak-dependent ER membrane permeabilization plays a role in ER stress–mediated cell death (8), which prompted us to study the relationship between the UPR and ER membrane permeabilization.

RESULTS

IRE1 signaling suppresses ER membrane permeabilization

ER luminal proteins distribute to the cytosol by Bax- and Bak-dependent ER membrane permeabilization under ER stress conditions (8). To confirm this finding, we monitored the redistribution of ER luminal proteins in wild-type and Bax/Bak double knockout (DKO) mouse embryonic fibroblasts (MEFs) treated with tunicamycin and thapsigargin. As expected, the redistribution of GRP78 and GRP94 to the cytosol was attenuated in DKO MEFs (Fig. 1A). However, we observed some leakage of ER contents in DKO MEFs treated with thapsigargin, suggesting that there could be a minor pathway mediating the leakage of ER contents independently of Bax and Bak. We also found that ectopic expression of Bak caused the redistribution of GRP78 and GRP94 to the cytosol in DKO MEFs treated with tunicamycin (Fig. 1B). Electron microscopic imaging revealed dilated ER under ER stress conditions; however, pores in ER membranes were not obvious (fig. S1A).

Fig. 1 ER stress induces ER membrane permeabilization.

(A) Immunoblot analysis of GRP94 and GRP78 (ER luminal), VAPB (ER membrane), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (cytosolic) in cytosolic and membrane fractions of wild-type (WT) and DKO MEFs treated with tunicamycin (TM) or thapsigargin (TG) or untreated (Untx). (B) Left: Immunoblot analysis of GRP94 and GRP78 (ER luminal), VAPB (ER membrane), and GAPDH (cytosolic) in cytosolic and membrane fractions of WT, DKO MEFs, and DKO MEFs rescued with WT-Bak (DKO + Bak) treated with tunicamycin or untreated. Right: Quantification of cytosolic GRP78 in WT, DKO MEFs, and DKO MEFs rescued with Bak (DKO + Bak) treated with tunicamycin or untreated. (C) Immunoblot analysis of GRP94, GRP78, calreticulin (CRT) and protein disulfide isomerase (PDI) (ER lumen), IRE1α and VAPB (ER membrane), and GAPDH in cytosolic and membrane fractions of WT MEFs treated with or without tunicamycin. (D) Upper: Immunoblot analysis of GAPDH and hemagglutinin (HA)–tagged A1AT-NHK mutant expressed in NSC34 cells cultured with or without kifunensine (Kif.) in the presence of cycloheximide (CHX). Lower: Quantitation of A1AT-NHK in immunoblots. (E) Immunoblot analysis of GRP94, GRP78, calreticulin, VAPB, and GAPDH in cytosolic or membrane fraction of WT MEFs treated with tunicamycin with or without kifunensine. White arrowhead, nonspecific signal. (F) Immunoblot analysis of ERDj5 and GAPDH in INS1 832/13 cells transfected with small interfering RNA (siRNA) scramble control (si-Cont) or siRNA against ERDj5 (si-ERDj5). (G) Immunoblot analysis of A1AT-NHK and GAPDH in INS1 832/13 cells transfected with siRNA scramble control or siRNA against ERDj5, treated with cycloheximide or untreated. (H) Quantitation of A1AT-NHK shown in (G). Statistical significance was calculated by Student’s t test. (I) Immunoblot analysis of GRP78, calreticulin, and GAPDH in cytosol of INS1 832/13 cells transfected with siRNA scramble control or siRNA against ERDj5, treated with thapsigargin or untreated. (J) Quantification of cytosolic GRP78 and calreticulin shown in (I). (K) Immunoblot analysis of A1AT-NHK, IRE1α, and GAPDH in whole-cell lysates (WCL), cytosol, or membrane fractions of NSC34 cells overexpressing A1AT-NHK mutant. An arrow shows glycosylated A1AT-NHK, and an arrowhead shows A1AT-NHK deglycosylated by endoglycosidase H (Endo-H). (L) Immunoprecipitation (IP) of cytosolic and membrane fractions of WT MEFs treated with tunicamycin or untreated with anti-GRP78 antibody, followed by immunoblot analysis with anti-ubiquitin (Ub) and anti-GRP78 antibodies. Left panels show input for immunoprecipitation. n = at least 3 biological replicates for (A) to (L). Representative blots are shown. Unless otherwise stated, statistical significance was calculated by one-way analysis of variance (ANOVA) followed by Tukey’s test. *P < 0.05; **P < 0.01; n.s., not significant. Error bars show SD.

We examined the leakage of various ER luminal and membrane proteins. All the examined ER luminal proteins, including GRP94, GRP78, calreticulin, and protein disulfide isomerase, leaked from the ER, whereas ER membrane proteins, including IRE1α and VAPB, did not leak to the cytosol (Fig. 1C). We therefore speculated that there was no selectivity for the leakage of ER content. To exclude the possibility that ER luminal contents were retrotranslocated from the ER lumen through ER-associated degradation (ERAD) pathway, we tested whether a small-molecule inhibitor for ERAD or knockdown of an ERAD component affected the leakage of ER contents to the cytosol. As reported previously, kifunensine inhibited the degradation of the ERAD substrate A1AT-NHK (α1-antitrypsin null Hong Kong) mutant, an ERAD substrate (9) (Fig. 1D), but did not affect the leakage of ER contents (Fig. 1E). We next examined the effect of RNA interference (RNAi)–mediated knockdown of ERDj5 also known as DNAJC10 [DNAJ (heat shock protein 40) homolog, subfamily C, member 10)], which is required for ERAD (10). Although knockdown of ERDj5 inhibited the degradation of A1AT-NHK (Fig. 1, G and H), it did not affect the leakage of ER contents (Fig. 1, I and J). To exclude the possibility that the leakage of ER luminal proteins was due to the abnormal translation of ER luminal proteins in cytosol, we overexpressed the A1AT-NHK mutant, which is misfolded in the ER lumen and causes ER stress (9). A1AT-NHK is also glycosylated in the ER (11), which occurs only in the ER; thus, if the A1AT-NHK in the cytosol is glycosylated, the protein is translated in the ER and then leaked from the ER. We detected glycosylated A1AT-NHK in the cytosol, indicating that this protein was not translated in the cytosol (Fig. 1K). Finally, we determined that cytosolic GRP78 was not polyubiquitinated (Fig. 1L), indicating that it was not a substrate for ERAD. These results indicate that the Bax- and Bak-dependent ER membrane permeabilization, but not the ERAD pathway, causes the redistribution of ER luminal proteins to the cytosol.

Dysregulation of ER homeostasis is linked to the activation of the UPR, raising the possibility that the ER membrane permeabilization might be modified by the regulators of the UPR, including IRE1α, PERK, and ATF6α (1). To test this notion, we analyzed ER membrane permeabilization in wild-type, Ire1α, Perk, and Atf6α knockout MEFs treated with thapsigargin for 6 hours, which induces mild leakage of ER contents in wild-type MEFs (Fig. 2A and fig. S1B). Cell fractionation analysis revealed that Ire1α knockout MEFs were more susceptible than wild-type, Perk knockout, or Atf6α knockout MEFs to thapsigargin-induced leakage of ER luminal proteins (Fig. 2A). Ectopic expression of IRE1α in Ire1α knockout MEFs decreased the leakage of the ER contents to the cytosol (Fig. 2B). As further confirmation of these observations, we found that thapsigargin exposure enhanced leakage of ER content in rodent INS1 832/13 β cells with transient knockdown of endogenous Ire1α by RNAi (Fig. 2C). Conversely, overexpression of wild-type IRE1α in INS1 832/13 cells significantly suppressed ER content leakage in response to thapsigargin (Fig. 2, D and E), suggesting that IRE1α protects against ER membrane permeabilization.

Fig. 2 IRE1 signaling suppresses ER membrane permeabilization.

(A) Left: Immunoblot analysis of GRP94, GRP78, calreticulin, VAPB, and GAPDH in cytosolic and membrane fractions of WT, Atf6α knockout (Atf6−/−), Ire1α knockout (Ire1α−/−), and Perk knockout MEFs (Perk−/−) treated with thapsigargin or untreated. Right: Quantification of cytosolic GRP78 in WT, Atf6−/−, Ire1α−/−, and Perk−/− MEFs treated with thapsigargin or untreated. (B) Immunoblot analysis of GRP94 and GRP78 (ER luminal), VAPB and IRE1α (ER membrane), and GAPDH (cytosolic) in cytosolic and membrane fractions of WT, Ire1α−/−, and Ire1α−/− rescued with WT-IRE1α (Ire1α−/−res.), treated with thapsigargin or untreated. (C) Left: Immunoblot analysis of GRP94, GRP78 and calreticulin (ER luminal), VAPB and IRE1α (ER membrane), and GAPDH (cytosolic) in cytosolic and membrane fractions of INS1 832/13 cells transiently transfected with scrambled control siRNA or siRNA against IRE1α (si-IRE1α). Right: Quantification of cytosolic GRP78 and calreticulin. (D) Immunoblot analysis of GRP94, GRP78, calreticulin, VAPB, GAPDH, and IRE1α in the cytosolic or membrane fractions of INS1 832/13 cells transduced with lacZ or WT IRE1α, treated with thapsigargin or untreated. (E) Quantitation of cytosolic GRP78 and calreticulin. (F) Immunoblot analysis of the proapoptotic proteins Bax and Bak, the antiapoptotic proteins Bcl-2 and Bcl-xL, and the proapoptotic BH3-only proteins PUMA, BimEL, Bad, Bid, and Bnip3 in Ire1α+/+ and Ire1α−/− MEFs treated with tunicamycin, thapsigargin, or untreated. n = at least 3 biological replicates for (A) to (F). Representative blots are shown. Statistical significance was calculated by one-way ANOVA followed by Tukey’s test. *P < 0.05; **P < 0.01. Error bars show SD.

Transcriptional regulation and/or posttranslational modification of Bcl-2 homology 3 (BH3)–only proteins, including BAD, Bim, Puma, Bid, and Bnip3 (Bcl-2/adenovirus E1B 19-kD protein–interacting protein 3), results in the activation of Bax and Bak (12, 13). We hypothesized that this group of proteins might play a role in the ER membrane permeabilization mediated by Bax and Bak and that such a mechanism might be enhanced in Ire1α knockout MEFs as compared to control MEFs. We found that the abundance of Bnip3 was increased in Ire1α knockout MEFs treated with thapsigargin and tunicamycin (Fig. 2F). Bnip3 is a BH3-only proapoptotic protein localized to the mitochondria and ER through its C-terminal transmembrane domain and induces apoptotic cell death by interfering with the antiapoptotic proteins Bcl-2 and Bcl-xL (14) after ischemia/reperfusion and under oxidative stress conditions (1518). Immunoblot analysis revealed multiple bands corresponding to monomers, dimers, and oligomers of Bnip3 (fig. S2A), as reported previously (19, 20). Treatment of the cell lysates with λ protein phosphatase to dephosphorylate proteins resulted in a single band corresponding to the monomeric form of Bnip3, suggesting that Bnip3 was a highly phosphorylated protein (fig. S2, A and B). Tunicamycin induced accumulation of highly phosphorylated Bnip3 (fig. S2C).

Bnip3 induces ER membrane permeabilization

We next sought to verify that ectopic expression of Bnip3 could induce the ER membrane permeabilization. Ectopic expression of Bnip3 in wild-type and Ire1α knockout MEFs induced the redistribution of GRP94 to the cytosol under normal conditions, which was enhanced by thapsigargin treatment (Fig. 3A). Fractionation analysis further confirmed that ectopic expression of Bnip3 induced the redistribution of GRP94 to the cytosol in Ire1α knockout MEFs in a dose-dependent manner (Fig. 3B). Furthermore, ectopic expression of Bnip3 was associated with calcium depletion of the ER (fig. S3, A and B). Because Bnip3 is localized to the ER and mitochondria (21), we wanted to determine whether Bnip3 was involved in the activation of the ER membrane permeabilization mediated by Bax and Bak. The interaction of Bnip3 with Bcl-2 induces the activation of Bax and Bak and apoptosis (14, 22). As expected from previous studies, the interaction between Bnip3 and Bcl-2 was enhanced by tunicamycin treatment in wild-type and Ire1α knockout MEFs (Fig. 3C). Bnip3 formed a heterodimer with Bcl-2 even in untreated Ire1α knockout MEFs, suggesting that this mechanism could underlie the susceptibility of these cells to ER membrane permeabilization (Fig. 3C). To further confirm that Bnip3 forms a heteromer with Bcl-2 on the ER membrane, we first performed subcellular fractionation to obtain pure ER fractions without mitochondrial contamination (Fig. 3D). Multimeric and presumably highly phosphorylated forms of Bnip3 were present in the ER fraction (fig. S4A). Immunoprecipitation analysis of the purified ER fraction also showed that phosphorylated Bnip3 interacted with Bcl-2, an interaction that was enhanced by tunicamycin treatment (fig. S4B). In addition, ectopic expression of Bnip3 induced the oligomerization of ER-localized Bak, but not of GRP94, in Ire1α knockout MEFs (Fig. 3E and fig. S5A), and oligomerization was further enhanced by thapsigargin treatment (Fig. 3E). To further confirm that ER-localized Bak was activated by Bnip3, we performed immunoprecipitations with an antibody that recognizes the active conformation of Bak (23) using ER fractions from Ire1α knockout MEFs. The amount of ER-localized active Bak was modestly increased by ectopic expression of Bnip3 and enhanced by thapsigargin treatment (Fig. 3F). Ectopic expression of Bnip3 did not increase the redistribution of GRP78 and calreticulin to the cytosol in DKO MEFs, whereas it increased the distribution of GRP78 and calreticulin to the cytosol in wild-type MEFs or DKO MEFs also expressing wild-type Bak, indicating that the ability of Bnip3 to promote ER membrane permeabilization was Bax- and Bak-dependent (Fig. 3G). Finally, we found that short hairpin RNA (shRNA)–mediated knockdown of Bnip3 reduced the redistribution of GRP94 and GRP78 into the cytosol in Ire1α knockout MEFs treated with thapsigargin, suggesting that Bnip3 knockdown prevented ER membrane permeabilization (Fig. 3H). Collectively, these results indicate that Bnip3 promotes the ER membrane permeabilization mediated by Bax and Bak.

Fig. 3 Bnip3 induces ER membrane permeabilization.

(A) Left: Immunoblot analysis of GRP94, VAPB, and GAPDH in cytosolic and membrane fractions of Ire1α+/+ and Ire1α knockout (Ire1α−/−) MEFs transduced with adenovirus expressing lacZ or Bnip3 and then treated with thapsigargin or untreated. Right: Quantitation of cytosolic GRP94. (B) Immunoblot analysis of GRP94, VAPB, and GAPDH in cytosolic and membrane fractions of Ire1α−/− MEFs transduced with adenovirus expressing lacZ or Bnip3 at indicated multiplicity of infection (MOI). (C) Coimmunoprecipitation analysis of endogenous Bnip3 and Bcl-2 in Ire1α+/+ and Ire1α−/− MEFs treated with tunicamycin or untreated. Arrows indicate monomer, black arrowheads indicate dimer, and white arrowheads indicate oligomer. (D) Examples of mitochondria and ER fractions of Ire1α−/− MEFs treated with thapsigargin or untreated. Tomm20 and protein disulfide isomerase were used as mitochondrial and ER markers. (E) Gel filtration analysis of endogenous Bak in the ER fractions of Ire1α−/− MEFs transduced with adenovirus expressing lacZ or Bnip3 treated with or without thapsigargin. M.W., molecular weight. (F) Immunoprecipitation of activated Bak using anti–Bak-NT (N-terminus) antibody from the ER fraction of Ire1α−/− MEFs transduced with adenovirus expressing lacZ or Bnip3 treated with thapsigargin or untreated. (G) Immunoblot analysis of GRP78, calreticulin, VAPB, and GAPDH in cytosolic and membrane fractions of WT, DKO MEFs (DKO), and DKO MEFs rescued with Bak (DKO + Bak), transduced with adenovirus expressing lacZ or Bnip3, and then treated with tunicamycin. (H) Left: Immunoblot analysis of GRP94, calreticulin, and GAPDH in cytosolic fractions of Ire1α−/− MEFs stably expressing shRNA directed against luciferase (shLuc) or Bnip3 treated with thapsigargin or untreated. Right: Quantitation of cytosolic GRP94 and calreticulin. n = at least 3 biological replicates for (A) to (H). Representative blots are shown. n = 3 biological replicates. Statistical significance was calculated by one-way ANOVA followed by Tukey’s test. *P < 0.05. Error bars show SD.

Impaired IRE1 signaling leads to accumulation of Bnip3

In response to ER stress, IRE1α activates multiple signals through its endonuclease and kinase domains. The endonuclease domain of IRE1α promotes splicing of the mRNA encoding X-box binding protein 1 (XBP1) and regulates IRE1α-dependent decay of mRNAs including that encoding DR5 (7). The kinase domain of IRE1α promotes the activation of c-Jun N-terminal kinase (JNK) by recruiting tumor necrosis factor α receptor–associated factor-2 (TRAF2) (24), which prompted us to test the relationship between the IRE1α-TRAF2-JNK cascade and ER membrane permeabilization. To determine the enzymatic activity of IRE1α that contributes to the suppression of ER membrane permeabilization, we monitored the leakage of ER luminal proteins into the cytosol in INS1 832/13 cells treated with tunicamycin and STF-083010, an IRE1α endonuclease-specific inhibitor, or SP600125, a JNK inhibitor (25). SP600125, but not STF-083010, increased the leakage of ER luminal contents to the cytosol, suggesting that the kinase domain of IRE1α played a role in preventing the ER membrane permeabilization through JNK activation (Fig. 4A and fig. S6A). To further confirm this observation, we used an endonuclease-dead K907A mutant of IRE1α and a kinase-dead K599A mutant that can still interact with TRAF2 (2628). Wild-type and K907A-mutant IRE1α, but not K599A-mutant IRE1α, suppressed ER membrane permeabilization in thapsigargin-treated Ire1α knockout MEFs, indicating that the kinase activity of IRE1α was essential to inhibit ER membrane permeabilization (Fig. 4B). We also monitored ER membrane permeabilization in wild-type MEFs (Xbp1+/+) and XBP1 knockout MEFs (Xbp1−/−) treated with thapsigargin (fig. S7A). Consistent with the observations above, ablation of Xbp1 did not affect ER membrane permeabilization. Next, we monitored the leakage of ER luminal proteins to the cytosol in Traf2 knockout MEFs, which were more susceptible to tunicamycin-induced leakage of ER luminal proteins as compared to control cells (Fig. 4C). Finally, we noted that the phosphorylation of JNK induced by ER stress preceded ER membrane permeabilization (fig. S8A). These results indicate that the IRE1-TRAF2-JNK pathway prevents the ER membrane permeabilization.

Fig. 4 Impaired IRE1 signaling leads to accumulation of Bnip3.

(A) Left: Immunoblot analysis of GRP94, GRP78, VAPB, phosphorylated JNK (p-JNK), JNK, and GAPDH in cytosolic and membrane fractions of INS1 832/13 cells untreated and/or treated with tunicamycin, STF-083010 (STF), or SP600125 (SP) as indicated. Right: Quantitation of GRP78 in the cytosolic fractions of INS1 832/13 cells untreated and/or treated with tunicamycin, STF-083010, or SP600125 as indicated. (B) Left: Immunoblot analysis of GRP94, calreticulin, GAPDH, and IRE1α in cytosolic and membrane fractions of Ire1α−/− MEFs transduced with adenovirus encoding lacZ, WT IRE1α, K599A-IRE1α, or K907A-IRE1α, treated with thapsigargin or untreated. Right: Quantification of cytosolic calreticulin. (C) Immunoblot analysis of GRP94, GRP78, calreticulin, VAPB, and GAPDH in cytosolic and membrane fractions of MEFs containing floxed TRAF2 alleles, transduced with adenovirus encoding lacZ or Cre recombinase (Cre), treated with tunicamycin or untreated. Statistical significance was determined by one-way ANOVA followed by Tukey’s test. (D) Quantitative PCR of Bnip3 of Ire1α+/+ and Ire1α−/− MEFs treated with tunicamycin or untreated. Gene expression was normalized to β-actin mRNA. (E) Left: Immunogold labeling of Bnip3 in human embryonic kidney (HEK) 293 cells. mito, mitochondria. Right: Immunogold double labeling of Bnip3 and LC3 in HEK293 cells. Bnip3 was labeled with 18-nm gold particles, and LC3 was labeled with 12-nm gold particles. Imaging was performed in two independent experiments. (F) Ire1α+/+ and Ire1α−/− MEFs transduced with adenovirus expressing GFP-LC3 treated with tunicamycin or untreated. Scale bars, 20 μm. Imaging was performed in three independent experiments. (G) Immunoblot analysis of LC3 and GAPDH of Ire1α+/+ and Ire1α−/− MEFs, treated with tunicamycin or untreated. (H) Cycloheximide chase of endogenous Bnip3 in WT, Ire1α knockout (Ire1α−/−), and Atg5 knockout (Atg5−/−) MEFs treated with tunicamycin and cycloheximide. (I) Mean intensities of Bnip3 in WT, Ire1α−/−, and Atg5−/− MEFs treated with tunicamycin and cycloheximide were plotted on a semilog graph. Statistical significance was calculated by one-way ANOVA followed by Dunnett’s test. (J) Immunoblot analysis of Bnip3 and GAPDH in WT MEFs in the absence or presence of rapamycin (Rapa). n = at least 3 biological replicates for (A) to (D) and (G) to (J). Representative blots and images are shown. Unless otherwise indicated, statistical significance was calculated by one-way ANOVA followed by Tukey’s test. *P < 0.05; **P < 0.01. Error bars show SD. The scale bar in the electron microscopic images indicates 100 nm and the scale bars in the confocal images indicate 20 μm.

We next investigated the mechanisms that mediate the increase in Bnip3 abundance in the absence of IRE1 signaling. The basal expression of Bnip3 mRNA was higher in Ire1α knockout MEFs than in wild-type MEFs, but the induction of Bnip3 mRNA by tunicamycin was higher in wild-type MEFs. As a result, the tunicamycin-induced expression of Bnip3 mRNA was comparable in both cell types (Fig. 4D), raising the possibility that the accumulation of Bnip3 protein in Ire1α knockout MEFs might be due to delayed protein degradation. Bnip3 protein turnover can be regulated by autophagy (29, 30). IRE1α-mediated activation of JNK induces cytoprotective autophagy under ER stress conditions, and suppression of autophagy can render cells susceptible to death triggered by ER stress (31, 32). We therefore used LY294002, which inhibits phosphatidylinositol 3-kinase signaling and autophagy, or MG132, an inhibitor of the ubiquitin-proteasome system, and found that both inhibitors increased the amount of Bnip3 in HEK293 cells ectopically expressing this protein (fig. S9A). The proteasomal activity of wild-type MEFs and Ire1α knockout MEFs did not significantly differ (fig. S9B), suggesting that proteasomal degradation of Bnip3 proceeds normally in Ire1α knockout MEFs. Immunogold electron microscopy revealed that Bnip3 was internalized in autophagosomes (Fig. 4E), prompting us to examine the role of IRE1α-mediated autophagy in Bnip3 degradation.

Punctate light chain 3 (LC3)–green fluorescent protein (GFP) dots, a marker for autophagy induction, were detected in wild-type MEFs treated with tunicamycin, but not in Ire1α knockout MEFs (Fig. 4F), as previously reported (31, 32). Furthermore, the conversion of LC3-I to LC3-II was increased by tunicamycin in wild-type MEFs but was attenuated in Ire1α knockout MEFs (Fig. 4G), an effect that was rescued by ectopic expression of wild-type IRE1α (fig. S9C), indicating that the induction of autophagy in response to ER stress was impaired in Ire1α knockout MEFs. We next monitored the degradation of endogenous Bnip3 under ER stress conditions using cycloheximide treatment. Bnip3 protein was more stable in Ire1α knockout or Atg5 knockout MEFs (in which autophagy is impaired) than in wild-type MEFs (Fig. 4, H and I). Furthermore, the steady state abundance of Bnip3 protein was decreased by rapamycin, which induces autophagy by inhibiting mammalian target of rapamycin C1 (mTORC1) (Fig. 4J).

Bax and Bak directly interact with IRE1α and regulate its activity (33). In DKO MEFs, IRE1α activation, as well as JNK phosphorylation, was suppressed under ER stress conditions. Therefore, in DKO MEFs, Bnip3 could accumulate because of decreased autophagy. The phosphorylation of JNK and induction of autophagy in response to tunicamycin-induced ER stress were suppressed in DKO MEFs, and as a result, the abundance of Bnip3 was increased (fig. S10, A and B). Collectively, these results indicate that Bnip3 accumulation in Ire1α knockout MEFs was mediated by impaired autophagy induction under ER stress conditions.

IRE1 inhibits the initial step of cell death

To further study the role of IRE1 in inhibiting ER membrane permeabilization and cell death, we used Ire1α knockout MEFs stably expressing mammalian ER-targeted redox-sensitive GFP (MERO-GFP). MERO-GFP changes its excitation spectrum depending on the oxidation status of two engineered cysteines. The reduction of the cellular environment leads to a change in the MERO-GFP reporter to the reduced state, which increases the MERO-GFP ratio (34). We found that tunicamycin treatment induced the leakage of MERO-GFP from the ER to the cytosol at 6 hours and that the MERO-GFP ratio was increased by more than 2 by the reducing conditions of the cytosol (Fig. 5, A and B, and fig. S11, A to D). Because the MERO-GFP ratio in cells with ER membrane permeabilization indicated by the leakage of MERO-GFP to the cytosol was increased by more than 2, whereas the MERO-GFP ratio in cells with intact ER was below 2, we defined cells with a MERO-GFP ratio greater than 2 as cells experiencing ER membrane permeabilization (Fig. 5, A and B). To study the relationship between Bax- and Bak-mediated ER membrane permeabilization and cell death, we next performed a flow cytometry analysis of the MERO-GFP ratio and the mitochondrial activity using MitoProbe DilC1(5) dye, which is sequestered by active mitochondria. We found that cells undergoing the ER membrane permeabilization still maintained mitochondrial activity (Fig. 5C). Dual live-cell imaging with tetramethylrhodamine methyl ester (TMRM) also showed that the MERO-GFP leaked from ER and that the MERO-GFP ratio was increased before mitochondrial activity decreased, indicating that ER membrane permeabilization preceded cell death (Fig. 5, D to F, and fig. S12A). To biochemically monitor the permeabilization of ER membranes and mitochondrial outer membranes under ER stress conditions, we performed subcellular fractionation of Ire1α knockout MEFs treated with tunicamycin. Consistent with the results obtained with flow cytometry, leakage of ER contents preceded the release of cytochrome C from mitochondria (Fig. 5G). As further confirmation, we performed dual live-cell imaging of MERO-GFP and propidium iodide in Ire1α knockout MEFs treated with tunicamycin. We found that cells with highly reduced MERO-GFP eventually underwent cell death (Fig. 5, H and I, and fig. S12B). Furthermore, MERO-GFP that leaked to the cytosol in dying cells was eventually oxidized (Fig. 5, E and I, and fig. S12B), because the cytosol in apoptotic cells was oxidized (fig. S13, A and B). We also verified that Ire1α knockout MEFs underwent cell death after ER membrane permeabilization (Fig. 5J).

Fig. 5 IRE1 inhibits the initial step of cell death.

(A) Live-cell imaging of MERO-GFP (excitation, 488 nm) in Ire1α knockout (Ire1α−/−) MEFs treated with tunicamycin (upper panels). The ratio images of 488/405 are displayed in false colors (lower panels). (B) Ratio traces of Ire1α knockout (Ire1α−/−) MEFs treated with tunicamycin. n = 12 cells imaged over three independent experiments. (C) Mitochondrial membrane potential (ΔΨm) as measured by MitoProbe dye in WT MEFs expressing MERO-GFP treated with thapsigargin or untreated. Statistical significance was calculated by one-way ANOVA followed by Tukey’s test. (D) Dual time-lapse imaging of TMRM and MERO-GFP (for example, 488 nm) in Ire1α−/− MEFs treated with tunicamycin for indicated times. Yellow dashed lines indicate the shape of the cell. (E) Time-lapse tracing of the MERO-GFP ratio and TMRM in Ire1α−/− MEFs treated with tunicamycin for the indicated times. (F) Traces of TMRM intensity in Ire1α−/− MEFs treated with tunicamycin during ER membrane permeabilization (EMP). The time point when the MERO-GFP started leaking from ER was set as T = 0. n = 10 cells from three independent experiments. The Wilcoxon signed-rank test was performed to compare the signal intensities at 0, 20, and 40 min to those at −20 min. None of these comparisons were significantly different (P > 0.05). (G) Left: Immunoblot analysis of GRP78, calreticulin, cytochrome C (cyto. C), and GAPDH in cytosol fractions of Ire1α−/− MEFs treated with tunicamycin for the indicated periods. Middle and right: Quantification of cytosolic GRP78 and cytochrome C. Statistical significance was calculated by one-way ANOVA followed by Dunnett’s test. (H) Dual live-cell imaging of MERO-GFP ratio and propidium iodide (PI) in Ire1α−/− MEFs treated with tunicamycin for the indicated times. White dashed lines indicate the shape of the cell experiencing ER membrane permeabilization. (I) The MERO-GFP ratio of each cell shown in (H) at the indicated times. (J) Fates of Ire1α−/− MEFs treated with tunicamycin. n = 30 cells imaged over three independent experiments. (K) Mitochondrial calcium monitored by Rhod-2 and MERO-GFP ratio in WT MEFs treated with thapsigargin. Statistical significance was calculated by Student’s t test. (L) ROS and MERO-GFP ratios in WT MEFs treated with thapsigargin. Statistical significance was calculated by Student’s t test. (M) ROS generation and MERO-GFP ratios in Perk knockout (Perk−/−) MEFs treated with thapsigargin. Statistical significance was calculated by Student’s t test. (N) Induction of mCherry fluorescence driven by the human Chop promoter and MERO-GFP ratios in HEK293 cells treated with thapsigargin. Statistical significance was calculated by Student’s t test. n = at least three biological replicates. Representative blots and images are shown. *P < 0.05; **P < 0.01. Error bars show SD. Scale bars, 20 μm.

To understand the basis of cell death mediated by the leakage of ER contents, we first analyzed the accumulation of calcium in the mitochondria. Because the ER is the largest store for intracellular calcium, ER membrane permeabilization might release calcium from the ER to the mitochondria and trigger apoptotic signals (35). The ER is also highly oxidized with hydrogen peroxide for efficient formation of disulfide bonds (36), and ER membrane permeabilization might induce cytosolic reactive oxygen species (ROS) accumulation. Flow cytometric analysis of MERO-GFP and Rhod-2 dye to measure mitochondrial calcium indicated that wild-type MEFs with ER membrane permeabilization had higher calcium concentrations in the mitochondria (Fig. 5K). Overexpression of Bnip3 induced ER membrane permeabilization, leading to the leakage of calcium (fig. S3), suggesting that the source of mitochondrial calcium might be cytosolic calcium originally contained in the ER. Flow cytometric analysis of MERO-GFP and CellRox dye (which detects cytosolic ROS) revealed that ROS concentrations were increased in cell populations with higher MERO-GFP ratios (Fig. 5L). Attenuation of oxidative stress under ER stress conditions is mediated by the PERK-ATF4 pathway (37). In Perk−/− MEFs stably expressing MERO-GFP, ROS concentrations were increased in cells with higher MERO-GFP ratios (Fig. 5M), suggesting that the increased ROS concentrations were due to ER membrane permeabilization rather than the activation of the PERK-ATF4 pathway. Furthermore, ER stress as measured by a human Chop promoter driving mCherry expression (fig. S14, A and B) was increased in the cells undergoing ER membrane permeabilization (Fig. 5N). These results suggest that IRE1-deficient cells were susceptible to alterations in cellular calcium concentrations, redox homeostasis, and ER stress. Accumulation of calcium in mitochondria causes release of cytochrome C (Fig. 5G), which is an inducer of the intrinsic apoptotic cascade (38), and as a result, apoptotic cell death could be induced by ER membrane permeabilization.

We next investigated the involvement of ER membrane permeabilization in various pathophysiological conditions. Hypoxia in cerebral tissues, cardiomyocytes, and cancer cells can lead to ER stress–mediated cell death (3941), prompting us to test if ER contents may be redistributed to the cytosol in mouse models of cerebral and myocardial infarction. Transient middle cerebral artery occlusion (TMCAO) in mice induced the distribution of ER luminal proteins to the cytosol (Fig. 6A). Furthermore, ischemia/reperfusion modeling in mice caused the distribution of the ER luminal proteins to the cytosol (Fig. 6B). In addition to these disease models, we assessed ER permeabilization in WFS1 knockout mice, which are a model of Wolfram syndrome, an inherited condition characterized by diabetes and neurodegeneration and associated with ER dysfunction (4245). Brain tissues from WFS1 knockout mice had more ER proteins in the cytosol than those from control mice (Fig. 6C). These results suggest that ER membrane permeabilization might play important roles in the pathogenesis of various diseases.

Fig. 6 ER membrane permeabilization in pathophysiological conditions.

(A) Immunoblot analysis of calreticulin, GRP94, VAPB, and GAPDH in cytosolic fractions of cerebrums from four mice with TMCAO. # shows the pair of control cerebrum and infarcted cerebrum from the same sample. (B) Left: Immunoblot analysis of GRP78, VAPB, COXIV, and GAPDH in cytosolic and membrane fractions of cardiac tissue with ischemia/reperfusion (I/R) or sham operation. Right: Ratios of cytosolic GRP78 (cyto)/membrane GRP78 of three independent pairs of mice with ischemia/reperfusion or sham operation. Statistical significance was calculated by unequal variance t test. n = 3 mice per each condition. (C) Immunoblot analysis of GRP94, VAPB, and GAPDH in cytosolic and membrane fractions of cerebrums from floxed-WFS1 mice crossed with Nestin-Cre transgenic mice. Each lane represents a separate animal. n = 3 mice per each condition. Representative blots are shown.

DISCUSSION

We have demonstrated that IRE1 prevented ER membrane permeabilization mediated by the proapoptotic Bcl-2 family proteins Bax and Bak. IRE1-mediated cytoprotective autophagy suppressed the amount of Bnip3, its subsequent heterodimerization with Bcl-2, which in turn led to the oligomerization of Bax and Bak in the ER membrane. IRE1 played a role in suppressing Bnip3-mediated ER membrane permeabilization, which resulted in changes in cellular redox states, intracellular calcium concentrations, ER stress, and cell death (fig. S15). Because the mitochondrial membrane potential was still maintained at the initiation of ER membrane permeabilization, our data suggest that IRE1 signaling prevented one of the earliest steps in ER stress–induced cell death. We discovered that the ER membrane permeabilization occurred in acquired pathological states associated with ER dysfunction, namely, ischemia/reperfusion injury and stroke, as well as Wolfram syndrome, an inherited disease state. Thus, our results provide a potential target for ER stress–related diseases.

MATERIALS AND METHODS

Animal experiments

All animal experiments were performed according to procedures approved by the Institutional Animal Care and Use Committee at the Washington University School of Medicine (A-3381-01).

Reagents

Thapsigargin, tunicamycin, CHAPS, Endoplasmic Reticulum Isolation Kit, and TMRM were purchased from Sigma. Anti-GAPDH antibody, anti-calreticulin antibody, anti-GRP94 antibody, anti-Bim antibody, anti-Bid antibody, anti-Puma antibody, anti-Bad antibody, anti-Bax antibody, anti-Bak antibody, anti–Bcl-2 antibody, anti–Bcl-xL antibody, anti-IRE1α antibody, anti-LC3A/B antibody, anti–caspase 3 antibody, and anti–cleaved caspase 3 antibody were obtained from Cell Signaling Technology. Anti-VAPB antibody, anti-Tomm20 antibody, anti–protein disulfide isomerase antibody, and anti-Bnip3 antibody were bought from Bethyl, Abcam, ENZO, and Genetex, respectively. Anti-GRP78 antibody and anti-WFS1 antibody were obtained from Proteintech. Dulbecco’s modified Eagle’s medium (DMEM), Rhod-2, Fura-2, CellRox dye, MitoProbe DilC1(5) Kit, propidium iodide solution, and protein G–Sepharose beads were obtained from Invitrogen. Anti-GFP antibody, kifunensine, and agarose-conjugated anti–Bcl-2 antibody were purchased from Santa Cruz Biotechnology. Anti Bak-NT antibody was from Millipore. Chroma spin columns were obtained from Clontech. An siRNA targeting rat ERDj5 (rArGrCrArArArUrArArArCrUrArGrArGrGrArUrCrGrUrUTG) and a scramble control were obtained from OriGene. An siRNA targeting rat IRE1α was reported previously (46). A plasmid encoding mCherry driven by human Chop promoter was a gift from Q. Lu. A plasmid encoding A1AT-NHK was provided by R. Kopito. Adenoviruses encoding wild-type IRE1α, K599A-IRE1α, and K907A-IRE1α were from R. Kaufman and K. Zhang.

Plasmids and viral vectors

The construction of MERO-GFP was reported previously (34). Lentivirus constructs expressing shRNA were obtained from the Genome Institute of Washington University in St. Louis. The shRNA target sequences are as follows: shLuc, TCTACTGGTCTGCCTAAAGGT; mouse shBnip3-1, CCATCTCTGTTACTGTCTCAT; and mouse shBnip3-2, AGAAGTTGAAAGTATCCTGAA. Adenovirus vectors expressing lacZ and Bnip3 were prepared as previously reported (47) and purified with Adenovirus Purification ViraKit (Virapur). Adenovirus vectors encoding wild-type IRE1, K599A-IRE1, and K907A-IRE1 were provided by R. Kaufman and K. Zhang. Titration of adenovirus was done with Adeno-X Rapid Titer Kit following the manufacturer’s instructions (Clontech).

Cell culture and live-cell imaging

Wild-type, Ire1 knockout, and Perk knockout MEFs were from D. Ron (University of Cambridge). Atf6 knockout MEFs were provided by R. Kaufman (University of California, San Diego). Bax/Bak DKO and Bax/Bak DKO rescued by wild-type Bak MEFs were from N. Chandel (Northwestern University). Xbp1+/+ and Xbp1−/− MEFs were from L. Glimcher (Cornell University). These MEFs were cultured in DMEM supplemented with 10% fetal bovine serum (FBS)–DMEM and antibiotics. HEK293 cells were cultured in DMEM supplemented with 10% FBS-DMEM and antibiotics. For live-cell imaging, HEK293 or Ire1α knockout MEF cells were plated on a poly-l-lysine–coated glass-bottom 35-mm dish (MatTek) the day before the analysis. The cells were then cultured with 10% FBS-DMEM containing tunicamycin (1 or 5 μg/ml) in association with propidium iodide (0.5 μg/ml) or 100 nM TMRM in 5% CO2 at 37°C for 12 to 24 hours. The images were captured every 20 min using Yokogawa spinning disc confocal on a Nikon TE-2000E2 inverted microscope with a 40× objective lens. Analyses were done on 16- or 32-bit images using ImageJ 1.45.

Cell fractionation

Fractionation of cytosol and membranous organelles from MEFs treated with tunicamycin (0.5 μg/ml for 16 hours for MEFs and 1 μg/ml for 16 hours for INS1 832/13 cells, unless indicated) or thapsigargin (0.5 μM for 16 hours for MEFs and 2 μM for 4 hours for INS1 832/13 cells, unless indicated) was done by ProteoExtract Subcellular Proteome Extraction Kit (Millipore) following the manufacturer’s protocol. If necessary, the cells were treated with 0.5 μM kifunensine. After fractionation, equal volume of the lysates was subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblot analyses.

Cycloheximide chase

Wild-type MEFs, Ire1 knockout MEFs, or Atg5 knockout MEFs were treated with cycloheximide (100 μg/ml) in combination with tunicamycin (2 μg/ml) for the indicated periods, followed by immunoblot analysis.

Coimmunoprecipitation with Bcl-2

Wild-type MEFs or Ire1 knockout MEFs, untreated or treated with tunicamycin (5 μg/ml) for 6 hours, were lysed in cell lysis buffer [1% CHAPS, 150 mM NaCl, 10 mM Hepes (pH 7.4), and protease inhibitors] followed by centrifugation at 15,000 rpm to remove cellular debris. The supernatants were mixed with agarose-conjugated anti–Bcl-2 antibody and incubated at 4°C for 4 hours. The beads were washed four times with the lysis buffer and analyzed by SDS-PAGE and immunoblot analysis. Coimmunoprecipitation with anti–Bcl-2 antibody was also performed with the ER fraction obtained from Ire1α knockout MEFs, untreated or treated with tunicamycin (5 μg/ml) for 6 hours using the ER Isolation Kit (Sigma) following the manufacturer’s protocol.

Immunoprecipitation of active Bak

Immunoprecipitation with anti–active Bak antibody (Bak-NT, Millipore) was performed with the ER fraction obtained from Ire1α knockout MEFs, untreated or treated with 1 μM thapsigargin for 6 hours using the ER Isolation Kit (Sigma) following the manufacturer’s protocol.

Immunoprecipitation of cytosolic GRP78

Immunoprecipitation with anti-GRP78 antibody (Proteintech) was performed with the cytosolic fractions from wild-type MEFs untreated or treated with tunicamycin (0.5 μg/ml) for 16 hours using the ProteoExtract Subcellular Proteome Extraction Kit (Millipore).

Gel filtration analysis

To detect oligomerized Bak in the ER, spin column–based gel filtration analysis was performed according to a previous report (48). Briefly, Ire1 knockout MEFs infected with adenovirus expressing lacZ or Bnip3 at MOI 200, untreated or treated with 1 μM thapsigargin for 6 hours, were fractionated into ER or mitochondrial fraction using Endoplasmic Reticulum Isolation Kit (Sigma) following the manufacture’s protocol, and the ER fraction was dissolved by adding CHAPS buffer [1% CHAPS, 150 mM NaCl, 10 mM Hepes (pH 7.4), and protease inhibitors] and applied to a Chroma spin column, followed by sequential gel filtration by centrifugation.

Fluorescence-activated cell sorting analyses

For flow cytometry analyses, HEK293 cells or MEFs expressing MERO-GFP were plated onto 12-well plates, treated with each compound for indicated times, and then harvested by trypsinization. Flow cytometry analyses were performed with LSR II (BD Biosciences). For measuring mitochondrial membrane potential, mitochondrial Ca2+ and cytosolic ROS, MitoProbe DilC1(5) Kit (Invitrogen), Rhod-2 (Invitrogen), and CellRox (Invitrogen) were used according to the manufacturer’s protocols. The results were analyzed by FlowJo version 7.6.3.

Transduction of cells with adenovirus

INS1 832/13 cells were transduced with adenovirus encoding wild-type and mutant IRE1α at MOI 10. MEFs were transduced with adenovirus encoding Bnip3 or lacZ at MOI 200 if not mentioned.

TMCAO procedure

Adult C57Bl/6J male mice (4 months of age) were anesthetized, and left MCAO was achieved using a nylon suture for 60 min under an operating microscope. Briefly, the left external carotid artery (ECA) was exposed and ligated with a 5-0 silk suture through a midline neck incision. A 14-mm 6-0 surgical monofilament nylon suture was introduced into the ECA stump and advanced 9 to 10 mm from the bifurcation of the common carotid artery into the circle of Willis to occlude the ostium of the MCA. After the desired time of MCAO, mice were reanesthetized, and ischemia was terminated by removal of the intraluminal suture. The cerebrums were isolated 24 hours after the procedure, followed by subcellular fractionation. Healthy right hemispheres were used as negative controls.

In vivo ischemia/reperfusion model

Adult mice (8 to 10 weeks of age) were subjected to transient left anterior descending artery (LAD) coronary ligation as described (49). Briefly, mice anesthetized were surgically prepared and ventilated. After thoracotomy, the LAD was ligated with a 9-0 polypropylene suture. Ischemia was confirmed by the presence of ST elevation on electrocardiogram as well as the absence of blood flow visually confirmed under the microscope. Thirty minutes after the occlusion, reperfusion was induced by cutting the knot of the suture. A sham operation was performed similarly without LAD ligation. One and a half hours after the induction of reperfusion, the mice were euthanized, and the apical third of the heart (ischemia/reperfusion–injured segment or sham) was subjected to subcellular fractionation.

Neuron-specific WFS1 conditional knockout mice

Mice with floxed WFS1 alleles have been previously described (50). For conditional knockout of WFS1 in neurons, WFS1-floxed mice were crossed with mice expressing Cre recombinase under the control of rat nestin promoter and enhancer (Jackson Laboratory). At the age of 8 months, the mice were sacrificed and cerebrums were isolated, followed by subcellular fractionation.

Electron microscopy of immunogold-stained Bnip3

HEK293 cells transfected with HA-Bnip3 and GFP-LC3, treated with tunicamycin (1 μg/ml) for 3 hours, were fixed in 4% paraformaldehyde/0.05% glutaraldehyde (Polysciences) in 10% gelatin Pipes/0.5 mM MgCl2 (pH7.2) for 1 hour at 4°C. Samples were then embedded in 10% gelatin and infiltrated overnight with 2.3 M sucrose/20% polyvinyl pyrrolidone in Pipes/MgCl2 at 4°C. Samples were trimmed, frozen in liquid nitrogen, and sectioned with a Leica Ultracut cryo-ultramicrotome (Leica Microsystems). Sections (50 nm) were blocked with 5% FBS/5% normal goat serum for 30 min and subsequently incubated with rabbit anti-HA antibody (Sigma) and goat anti-GFP antibody (ab5450, Abcam), followed by secondary donkey anti-rabbit antibody conjugated to 18-nm colloidal gold and donkey anti-goat antibody conjugated with 12-nm colloidal gold (Jackson ImmunoResearch Laboratories). Sections were washed in Pipes buffer, followed by a water rinse, and stained with 0.3% uranyl acetate/2% methyl cellulose. Samples were viewed with a JEOL 1200EX transmission electron microscope (JEOL USA) equipped with an AMT 8-megapixel digital camera (Advanced Microscopy Techniques). All labeling experiments were conducted in parallel with controls omitting the primary antibody. These controls were consistently negative at the concentration of colloidal gold–conjugated secondary antibodies used in these studies.

Endoglycosidase H assay

The NSC34 cell lysates overexpressing A1AT-NHK was treated with endoglycosidase H (1500 U per sample) (New England Biolabs) for 1 hour at 37°C following the manufacturer’s protocol.

Measurement of cytosolic calcium

Cytosolic calcium concentration of wild-type MEFs transduced with adenovirus expressing lacZ or Bnip3 at MOI 200 (30,000 cells per well, n = 6 wells per condition) was measured with Fura-2 dye. After measurement of basal calcium, the cells were treated with 1 μM thapsigargin.

Proteasomal activity

Proteasomal activity of wild-type MEFs and Ire1α knockout (Ire1α−/−) MEFs was determined by Proteasome-Glo assay (Promega) following the manufacturer’s instructions.

Dephosphorylation of Bnip3

The proteins were dephosphorylated by λ protein phosphatase (New England Biolabs). Samples were reacted in the buffer containing 50 mM Hepes (pH 7.5), 10 mM NaCl, 2 mM dithiothreitol, 0.01% Brij 35, 1 mM MnCl2, and λ protein phosphatase (40 U per sample) at 30°C for 1 hour, followed by immunoblot analysis.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/8/382/ra62/DC1

Fig. S1. Leakage of ER contents to the cytosol precedes the apoptotic cascade.

Fig. S2. Phosphorylation of Bnip3.

Fig. S3. Role of Bnip3 in the leakage of ER calcium to the cytosol.

Fig. S4. Subcellular localization of Bnip3.

Fig. S5. Gel filtration of GRP94.

Fig. S6. Confirmation of attenuation of IRE1 signaling by small-molecule inhibitors.

Fig. S7. Leakage of ER contents to the cytosol in Xbp1+/+ and Xbp1−/− cells.

Fig. S8. Time-course analysis of the phosphorylation of JNK and leakage of ER contents.

Fig. S9. Degradation mechanism of Bnip3.

Fig. S10. Abundance of Bnip3 and phosphorylation of JNK and LC3-II in wild-type cells and Bax/Bak DKO cells.

Fig. S11. Live-cell imaging of MERO-GFP–expressing IRE1 knockout cells under ER stress conditions.

Fig. S12. Time-lapse ratio imaging of ER membrane permeabilization in IRE1 knockout cells.

Fig. S13. Time-lapse ratio imaging of HEK293 cells expressing cytosolic roGFP under ER stress conditions.

Fig. S14. Fluorescent imaging of HEK293 cells expressing the mCherry-Chop reporter under ER stress conditions.

Fig. S15. Scheme of ER membrane permeabilization by ER stress.

REFERENCES AND NOTES

Acknowledgments: We thank P. Furcinitti, C. Brown, and M. Kanekura for technical support. We thank D. Ron, R. Kaufman, L. Glimcher, K. Zhang, R. Davis, D. Mann, R. Brink, H. Virgin, N. Chandel, R. Kopito, S. Oakes, and F. Papa for providing us with various reagents. We also thank W. Beatty for technical and scientific assistance with the electron microscopy studies. Funding: K. K. was partly supported by fellowships from the Japan Society for the Promotion of Science and the Uehara Memorial Foundation. A. D. is supported by grants from NIH (HL107594) and Department of Veterans Affairs (I01BX000448 and I01BX001969). F. U. is supported by grants from the NIH (DK067493 and DK020579), Juvenile Diabetes Research Foundation (17-2013-512), American Diabetes Association (1-12-CT-61), the Team Ian, the Team Alejandro, the Ellie White Foundation for Rare Genetic Disorders, and the Jack and JT Snow Scientific Research Foundation. Author contributions: K.K., A.D., and F.U. designed the experiments. K.K., X.M., and J.T.M. performed the experiments. L.J.Z. performed statistical analyses of the data. K.K., A. D., and F.U. wrote the article. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article