Research Resource: Monitoring Endoplasmic Reticulum Membrane Integrity in (cid:2) -Cells at the Single-Cell Level

Endoplasmic reticulum (ER) membrane integrity is an emerging target for human chronic diseases associated with ER stress. Despite the underlying importance of compromised ER membrane integrity in disease states, the entire process leading to ER membrane permeabilization and cell death is still not clear due to technical limitations. Here we describe a novel method for monitoring ER membrane integrity at the single-cell level in real time. Using a (cid:2) -cell line expressing ER-targeted redox sensitive green fluorescent protein, we could identify a (cid:2) -cell population undergoing ER membrane permeabilization induced by palmitate and could monitor cell fate and ER stress of these cells at the single-cell level. Our method could be used to develop a novel therapeutic modality targeting the ER membrane for ER-associated disorders, including (cid:2) -cell death in diabetes, neurodegeneration, and Wolfram syndrome. ( Molecular Endocrinology 29: 473–480, 2015)

T ype 2 diabetes is defined by hyperglycemia caused by a relative deficiency of insulin. It has been proposed that ␤-cell death is an important pathogenic component of type 2 diabetes (1-3). Type 2 diabetes is the most common form of diabetes and is associated with obesity and hyperlipidemia. It has been established that prolonged exposure to elevated fatty acids causes impaired glucosestimulated insulin secretion, suppression of insulin gene expression, and apoptosis of ␤-cells. Multiple models have been proposed to explain the underlying mechanisms by which elevated free fatty acids trigger dysfunction and death of ␤-cells (4 -12).
Endoplasmic reticulum (ER) plays a range of vital roles including synthesis and oxidative folding of secretory proteins, calcium homeostasis, and regulation of cell death. Dysregulation of ER homeostasis activates the unfolded protein response, which controls the cell fate either to adapt to stress or to initiate apoptosis (13). Recently it has been shown that saturated fatty acids target the ER and trigger ER stress in ␤-cells (4,5,12,14). The lipid profile of ER membrane and ER morphology are also affected by overloading of free fatty acids, and palmitate, a saturated fatty acid, induces the disruption of ER membrane integrity (15). Impairment of the integrity of ER membrane followed by the leakage of ER contents, including calcium, has been shown under various chemical or pathological ER stress conditions (15)(16)(17)(18). Although the leakage of ER contents possibly contributes to malfunction of ER and initiation of apoptosis, the underlying mechanisms have been poorly investigated due to the lack of methods for detecting the loss of integrity of the ER membrane. Here we report a novel method for evaluating the ER membrane integrity in palmitate-treated ␤-cells at the single-cell level.

Cell culture
INS-1 832/13 cells were cultured in the RPMI 1640 supplemented with 10% fetal bovine serum (FBS), penicillin and streptomycin, sodium pyruvate, and 0.1% ␤-mercaptoethanol. Mouse embryonic fibroblasts were cultured in DMEM containing 10% FBS and antibiotics. Primary humans islets, obtained from Prodo, were plated onto a six-well plate (1500 islets/well) precoated with laminin V produced by 804G cells and cultured in CMRL medium (medium developed at the Connaugh Medical Research Laboratories) supplemented with FBS, nonessential amino acids, sodium pyruvate, and antibiotics. For establishing stable cell lines, cells were transduced with lentiviruses expressing MERO-GFP (pLenti-CMV-puro-MERO-GFP) and selected with puromycin. The lentivirus vectors including pLenti-CMV-puro-dest vector was obtained from Addgene.

Preparation of fatty acids
Palmitate, oleate, and linoleate were conjugated with BSA by soaping each fatty acid to sodium hydroxide and mixing with BSA as described previously (19). Twenty millimolar of the solution of each fatty acid in 0.02 M NaOH was incubated at 70°C for 30 minutes and then mixed with 5% BSA in PBS in a 1:3 vol ratio. Each solution was diluted with 10% FBS-RPMI 1640 medium to the designated concentration.

Gene expression profiling
INS1-MERO-GFP cells treated with 0.5 mM PA for 24 hours were sorted by FACS-AriaIII (BD Biosciences), and total RNA was extracted with a RNeasy kit (QIAGEN). The purified RNA was applied to a GeneChip Rat Gene 1.0 ST array (Affymetrix) according to the manufacturer's protocols. The obtained data were analyzed by getting the difference from each pair of cells with the MERO to GFP ratio greater than 2 and less than 2, and we set the value of the mean of each pair as 0. The heat map was generated from the data reflecting the differences of gene expressions in each pair in log2 scale.

Transmission electron microscopy
INS-1 832/13 cells and human primary islets plated onto a six-well plate were fixed by adding 1 mL of 2.5% glutaraldehyde (vol/vol) in 0.1 M Na cacodylate-HCl buffer (pH 7.2) to each of the wells containing 1 mL of media. The cells were then fully fixed with 2.5% glutaraldehyde (vol/vol) in 0.1 M Na cacodylate-HCl buffer (pH 7.2) overnight at 4°C. The fixed samples, washed three times in 0.1 M Na cacodylate-HCl buffer (pH 7.2), were pelleted and postfixed for 1 hour in 1% osmium tetroxide (wt/vol) in distilled water. The fixed cells, dehydrated through a series of ethanol, were transferred through two changes of propylene oxide and then into a mixture of 50% propylene oxide/50% epon resin. The epoxy blocks were trimmed, and ultrathin sections of 70 nm were cut on a Reichart-Jung ultramicrotome. The sections, contrasted with lead citrate and uranyl acetate, were examined using a FEI Tecani 12 BT with 80-Kv accelerating voltage, and images were captured using a Gatan transmission electron microscopy charge-coupled device camera.

Live imaging
INS1-MERO-GFP cells were plated on a poly-L-lysinecoated, glass-bottom, 35-mm dish (MatTek) the day before the analysis. The cells were then cultured with 10% FBS-RPMI 1640 medium containing 0.4 mM PA and 0.5 g/mL of PI in 5% CO 2 at 37°C for 15 hours. The images were captured every 20 minutes using a Yokogawa spinning disk confocal on a Nikon TE-2000E2 inverted microscope with a ϫ40 objective lens. All analyses were done on 16-bit or 32-bit images using ImageJ 1.45 (National Institutes of Health, Bethesda, Maryland).

Cell fractionation
The fractionation of cytosol and membranous organelles from INS1 832/13 cells treated with or without PA were done by a Qproteome cell compartment kit (QIAGEN) following the manufacture's protocol. After fractionation, an equal volume of the lysates was subjected to SDS-PAGE followed by immunoblot analyses.

Mass spectrometry and liquid chromatography and mass spectrometry analysis
Liquid chromatography and mass spectrometry analysis of lipid was conducted on a Thermo Scientific Vantage TSQ mass spectrometer with Thermo Accela UPLC operated by Xcalibur software. Mass spectra of the eluted peaks containing the phospholipids in the same family were averaged, and the lipid profiles among samples were compared. Structural identification of the lipid species was performed on a Thermo Scientific LTQ Orbitrap Velos mass spectrometer with an Xcalibur operating system.

Palmitate alters lipid composition of the ER membrane and disrupts its structure
To test whether palmitate alters the lipid composition of the ER membrane in pancreatic ␤-cells, ER fractions from a rat insulinoma cell line, INS1 832/13 cells, treated with 0.5 mM plamitate or 0.5 mM oleate for 24 hours were analyzed with liquid chromatography-mass spectrometry ( Figure 1A). Both palmitate and oleate increased the quantity of lipid species in the ER membrane (20). The degree of saturation of fatty acids was also increased by treatment with palmitate (Supplemental Figure 1).
It has been shown that increased saturation of lipid species in biological membrane is associated with stiffness and less flexibility of membrane (21)(22)(23). To test whether palmitate damages ER membrane structure in ␤-cells, we treated INS1 832/13 cells and human primary islets with palmitate. To minimize the effects of apoptosis on the membrane structure, we treated INS1 832/13 cells and human primary islets with 0.5 mM palmitate for 4 hours and then examined the morphology of the ER using transmission electron microscopy. In untreated cells, the morphology of the ER appeared normal (Figure 1, B-E), whereas palmitate-treated cells showed dilated ER with discontinuous ER membrane and dissociation of ribosomes from the rough ER (Figure 1, F-I). Furthermore, treatment with palmitate induced the mislocalization of ER-DsRed, a red fluorescent protein expressed in the lumen of the ER, in human embryonic kidney-293 cells (Supplemental Figure 2). Collectively these results indicate that palmitate directly impairs the ER membrane integrity and induces ER stress in ␤-cells (5, 15, 24, 25). Disruption of ER membrane by palmitate causes the leakage of ER contents to the cytosol in cardiomyocytes and pancreatic ␤-cells (15,17,26). To date, the evaluation of the leakage of ER contents depends on cell fractionation of bulk-prepared cells and tissues, followed by immunoblot, and it has been difficult to analyze at the single-cell level. To overcome this challenge, we adopted the roGFP localizing to mammalian ER (mammalian ER localized roGFP, MERO-GFP) (25,27,28). We have previously shown that MERO-GFP is specifically localized to the ER (28). Depending on the status of an artificially inserted cysteine pair, roGFP can indicate precise redox change by calculating the ratio of the signal from excitation at 488 nm and signal from excitation at 405 nm ( Figure 2A). Because the ER lumen has a highly oxidative environment, the ratio between reduced and oxidized roGFP expressed in the ER (MERO-GFP ratio) is much lower than the ratio between reduced and oxidize roGFP expressed in the cytosol (cytosolic roGFP ratio) ( Figure  2B) (28). We therefore considered the possibility that the dramatic shift of the roGFP ratio can be used to detect the leakage of MERO-GFP to the cytosol (Figure 2, B and C). To test this, INS1 832/13 cells stably expressing MERO-  (Figure 3, A-C). Interestingly, the combination of palmitate and oleate or linoleate did not create the cell population harboring MERO-GFP ratio greater than 2, raising the possibility that oleate and linoleate may prevent the disruption of ER membrane by palmitate ( Figure 3C). To test this idea, we treated the INS1 832/13 cells with BSA, 0.5 mM palmitate, or 0.5 mM palmitate in combination with 0.5 mM oleate or 0.5 mM linoleate for 16 hours and then performed cell fractionation analysis ( Figure 3D). Consistent with the observations from the flow cytometry analysis, both oleate and linoleate prevented the leakage of ER luminal proteins to the cytosol mediated by palmitate. Collectively these findings indicate that oleate and linoleate can prevent the disruption of ER membrane by palmitate Bcl-2 family proteins, Bcl-2-associated X protein (Bax) and Bcl-2-antagonist/killer (Bak), have been reported to play a role in the ER membrane permeabilization under ER stress conditions, raising the possibility that palmitate-induced ER membrane disruption might be mediated by Bax and Bak (16). To test this idea, wild-type MEFs (wt-MEFs) or Bax/Bak double-knockout MEFs stably expressing MERO-GFP were treated with BSA, 0.2 mM palmitate, or 0.5 mM palmitate. To minimize the effects of apoptosis on the membrane integrity, we treated wt-MEFs and double-knockout MEFs for 6 hours, which did not induce apparent cell death, and then monitored MERO-GFP distribution to the cytosol by flow cytometry. We found that deletion of Bax/Bak did not prevent the creation of a cell population harboring MERO-GFP ratio greater than 2 (Supplemental Figure 3A). Cell fractionation followed by immunoblot assay showed that deletion of Bax/Bak prevented the leakage of ER luminal proteins to the cytosol mediated by a canonical ER stress inducer, tunicamycin, but not by palmitate, suggesting that palmitate-induced ER membrane disruption is not Bax/Bak dependent (Supplemental Figure 3B). AICAR, an activator for AMPdependent protein kinase, has been reported to confer protection against palmitate-mediated cell death (29,30). We were therefore interested in determining whether AICAR could prevent the ER membrane permeabilization by palmitate. We treated INS1-MERO-GFP cells with 0.25 mM palmitate in combination with or without 0.5 mM AICAR for 1 hour and analyzed the distribution of MERO-GFP with flowcytometry. As expected, AICAR significantly decreased the creation of a cell population harboring the MERO-GFP ratio greater than 2 (Supplemental Figure 3C).
Loss of ER membrane integrity destroys the oxidative environment in the ER lumen. We have previously reported that palmitate treatment depletes ER calcium store, which impairs the functions of calcium-dependent chaperones localized to ER, resulting in ER stress (31). Therefore, it is possible that the cells experiencing the loss of ER membrane integrity have higher ER stress levels due to the inefficient protein folding. To test this possibility, INS1-MERO-GFP cells were treated with 0.5 mM palmitate and then sorted by flow cytometry based on MERO-GFP ratio ( Figure 4A). Gene expression profiles of sorted cells showed that genes involved in ER stress were upregulated in a cell population harboring an MERO-GFP ratio greater than 2 as compared with a cell population with the ratio less than 2 ( Figure 4B). Because the ER stress response triggers apoptotic signaling in response to intolerable ER stress, we examined whether ER stressrelated apoptotic cascades were activated in cells experiencing the ER membrane permeabilization (32). Quantitative PCR analysis showed that ER stress markers, BiP and spliced XBP1, as well as proapoptotic genes in the ER stress response, including Chop and Puma, were significantly up-regulated in cells undergoing the ER membrane permeabilization ( Figure 4C) (33)(34)(35). Collectively these results indicate that the loss of ER membrane integrity induces ER stress. We next asked whether the cells harboring an MERO-GFP ratio greater than 2 were susceptible to apoptosis and performed the dual live-cell imaging of MERO-GFP and PI in INS1-MERO-GFP cells treated with palmitate. Both the MERO-GFP ratio and the intensity of PI staining in these cells were measured every 20 minutes. We found that cells harboring highly reduced MERO-GFP eventually underwent cell death ( Figure 4D and Supplemental Movie 1). MERO-GFP leaked to the cytosol in dying cells was eventually oxidized ( Figure 4D and Supplemental Figure 4).

Discussion
By using mammalian ER-targeted roGFP (MERO-GFP), we have established a method for evaluating the ER membrane integrity in palmitate-treated ␤-cells at the singlecell level. Palmitate impairs ER membrane integrity, resulting in the leakage of MERO-GFP to the cytosol. The MERO-GFP distributed to the cytosol is immediately reduced, and the ratio between reduced and oxidized MERO-GFP significantly increases upon the leakage. The change in the MERO-GFP ratio can be monitored at the single-cell level in real time. Our method provides new insights into how palmitate induces ER stress and ␤-cell death in real time, which can be used to develop a novel therapy for protecting ␤-cells from free fatty acids in diabetes.
Previous studies have shown that palmitate and saturated fatty acids impair ER membrane integrity, leading to apoptosis (15, 17, 26, 36 -38). Despite the underlying importance of ER membrane permeabilization in disease states, the mechanisms of loss of ER membrane integrity are still not clear due to methodological limitations. Thus, there is an urgent need to accurately monitor ER mem- brane integrity in real time, and this has been successfully accomplished in this study. The lumen of the ER maintains an oxidative environment with high concentrations of calcium, which is optimized for efficient protein folding. Using the method described in this article, we can potentially identify environmental and pathophysiological factors that disrupt the oxidative environment in the ER. In addition, a flow cytometic analysis and time-lapse imaging identified a previously uncharacterized population of cells generated by palmitate treatment. These cells were isolated by the FACS and characterized. We found that genes involved in ER stress, especially proapoptotic ER stress genes including Chop and Puma, were up-regulated in this cell population undergoing ER membrane permeabilization as compared with a cell population not experiencing it (Figure 4, A-C) (35,39). We therefore conclude that pancreatic ␤-cells harboring compromised ER membrane integrity by palmitate undergo cell death due to intolerable ER stress.
Palmitate treatment increased the incorporation of saturated fatty acid into the ER membrane in ␤-cells, leading to the ER membrane permeabilization and leakage of ER contents to the cytosol (Figures 1 and 3 and Supplemental Figure 1) (20). The mechanisms by which palmitate impairs the ER membrane integrity remain undetermined. The results shown above and a report from others indicate that canonical ER stress inducers, such as tunicamy-  cin, induce ER membrane permeabilization in a Bax/ Bak-dependent manner (Supplemental Figure 3B) (16). However, palmitate-mediated ER membrane permeabilization is not mediated by Bax and Bak (Supplemental Figure 3B). One possible mechanism is that palmitate reduces the fluidity of the ER, resulting in the stiff and fragile ER membrane (22).
It is noteworthy that unsaturated fatty acids, such as oleate and linoleate, could prevent ER membrane permeabilization mediated by palmitate (Figure 3, C and D). It has been reported oleate protects pancreatic ␤-cells from palmitate-mediated ␤-cell death by suppressing ER stress (40). It is possible that unsaturated fatty acids prevent excess incorporation of palmitate into lipid species and protects the ER membrane from palmitate, which in turn suppresses ER stress.
We have shown here a method for evaluating the ER membrane integrity in palmitate-treated ␤-cells at the single-cell level. Our method based on ER-targeted roGFP provides new insights into the mechanisms of cell death mediated by compromised ER membrane integrity, which can be targeted to develop novel therapeutic modalities for ER diseases, including ␤-cell death in diabetes, neurodegeneration, and Wolfram syndrome, a prototype of human ER disease (41,42).