Endoplasmic Reticulum Stress Links Oxidative Stress to Impaired Pancreatic Beta-Cell Function Caused by Human Oxidized LDL

Elevated plasma concentration of the pro-atherogenic oxidized low density lipoprotein cholesterol (LDL) triggers adverse effects in pancreatic beta-cells and is associated with type 2 diabetes. Here, we investigated whether the endoplasmic reticulum (ER) stress is a key player coupling oxidative stress to beta-cell dysfunction and death elicited by human oxidized LDL. We found that human oxidized LDL activates ER stress as evidenced by the activation of the inositol requiring 1α, and the elevated expression of both DDIT3 (also called CHOP) and DNAJC3 (also called P58IPK) ER stress markers in isolated human islets and the mouse insulin secreting MIN6 cells. Silencing of Chop and inhibition of ER stress markers by the chemical chaperone phenyl butyric acid (PBA) prevented cell death caused by oxidized LDL. Finally, we found that oxidative stress accounts for activation of ER stress markers induced by oxidized LDL. Induction of Chop/CHOP and p58IPK/P58IPK by oxidized LDL was mimicked by hydrogen peroxide and was blocked by co-treatment with the N-acetylcystein antioxidant. As a conclusion, the harmful effects of oxidized LDL in beta-cells requires ER stress activation in a manner that involves oxidative stress. This mechanism may account for impaired beta-cell function in diabetes and can be reversed by antioxidant treatment.


Introduction
The progressive dysfunction and destruction of pancreatic beta-cells is a key feature of the onset and progression of type 2 diabetes (T2D) [1][2][3][4]. The resulting decline in beta cell function is characterized by a loss in cell number caused by an increased apoptosis rate and defective insulin production and secretion from the remaining beta cells [1][2][3][4]. It has been suggested that in the context of systemic insulin-resistance, low grade inflammation, chronic excess of cholesterol and of metabolic fuels including the non-esterified fatty acid palmitate and glucose, trigger beta-cell damage over time, especially in genetically predisposed individuals [1][2][3][4]. Furthermore, elevated plasma levels of oxidized low density lipoprotein cholesterol (LDL) act as additional potential diabetogenic stressor and increase the risk for associated cardiovascular diseases [5]. Indeed, specific antibodies against oxidized LDL have been reported in patients with T2D [6]. High oxidized LDL levels are commonly found in the obesity-associated metabolic syndrome [7] and further increase throughout the development of T2D [8]. Importantly, several studies have reported the presence of receptors for oxidized LDL in both human and rodent islet beta-cells [9][10][11][12]. The deleterious effects of human oxidized LDL on beta-cell function have been evidenced by in vitro experiments. The copper-mediated oxidation of LDL provokes similar modification within the particles to those occurring in human [13]. This oxidation is therefore commonly used to mimic the effects of oxidized LDL [11,[14][15][16]. The administration of mildly oxidized LDL (2 mmol/l) to isolated human and rat pancreatic islets, as well as into insulin producing beta-cells decreases both production and secretion of insulin, and ultimately kills beta-cells by inducing apoptosis [11,[14][15][16]. The adverse effects of oxidized LDL rely on mechanisms that involve both oxidative stress and induction of cAMP responsive element modulator (CREM, also called ICER) [16]. However native LDL at similar cholesterol concentration (2 mmol/l) does not trigger harmful effects on beta cells [15,16].
The endoplasmic reticulum (ER) might play a key role in mediating adverse effects of oxidized LDL on beta-cells. First, ER stress is involved in beta-cell dysfunction and death caused by several diabetogenic stressors including chronic hyperglycemia and hyperlipidemia [17][18][19][20]. Second, the ER stress is closely linked to oxidative stress [21,22]. We have previously shown that oxidative stress is induced by oxidized LDL, contributing to beta-cell death and impaired insulin expression [16]. In contrast, treatment of beta-cells with the antioxidant N-acetylcystein (NAC) prevented beta-cell decline caused by oxidized LDL. In addition, high density lipoprotein cholesterol (HDL) has antioxidant property and antagonizes the harmful effects of oxidized LDL [11,[14][15][16]. Last, oxidized LDL triggers ER stress signaling transducers that includes the eukaryotic translation initiation factor 2 alpha kinase 3 (EIF2AK3, also called PERK), endoplasmic reticulum to nucleus signaling 1 (ERN1, also called IRE1α), and activating transcription factor 6 (ATF6) in vascular cells [23]. Thus, we postulated that the ER stress may act as a possible mediator for the deleterious effects of oxidized LDL on pancreatic betacells.

Lipoprotein preparation
Blood collection from human healthy donors, plasma LDL (LDL density, 1.063) and HDL fractions were prepared exactly as described [15,16]. Oxidation of LDL particles was done by incubation of 1 mg LDL protein/ml PBS with 5 μmol/l CuSO 4 at 37°C for 6-8 h [15,16]. Oxidation was verified by determining the lipid peroxide content [15,16], and the oxidation degree was evaluated by measuring the relative electromobility rate on agarose gels as previously described [15,16].

Cell culture and preparation of isolated islets
The insulin-secreting cell line (MIN6) was maintained as previously described [15,16]. Isolated human islets were obtained from the biotherapies for diabetes unit from the «Centre Hospitalier Régional et Universitaire de Lille». Human pancreatic tissue was harvested from braindead adult human donors in the context of the traceability requirements for the clinical islet transplantation program of the laboratory INSERM UMR 1190, Lille, France (clinicaltrials.gov, NCT01123187, NCT00446264, NCT01148680). The experimental design was approved in agreement with French regulations, our Institutional Ethical Committee of the University of Lille and the Centre Hospitalier Régional Universitaire de Lille. Informed consent from the next of kin, is obtained on the behalf of the deceased by the National French Procurement Agency «Agence de la BioMedecine». Consent was verbal. In France, it is presumed that you agree to organ donation unless you have expressed your refusal in the National Registry for Refusal of organ donation (NRR). The Agence de la BioMedecine Coordinators ask for verbal consent and then consult the NRR (ie written refusal). Therefore the written refusal is based on the NRR. None of the transplant donors were from a vulnerable population and all donors or next of kin provided both informed verbal consent and written consent after consulting the NRR that was freely given. As a direct result of informed consent for scientific research, we check a box for each donor on our human islet data base autorization for Research obtained YES or NO. Human islets that did not receive prior informed consent for Science are never used for research. Isolation and preparation of islets in Lille were conducted as described previously [24]. Experiments were carried out on fresh human islets preparation with >80% viability and >80% purity. After isolation, human islets were cultured in CMRL-1066 medium culture containing 5 mmol/l glucose, 100 U/ml penicillin, 100 μg/ml streptomycin and 10% fetal bovine serum (Mediatech Herndon, VA) in 5% CO 2 humidified atmosphere at 37°C. For measuring gene expression and insulin secretion in response to lipoproteins, 200 and 40 islets equivalent number (IE) were plated in 24-well plate, respectively.

Apoptosis assay
Apoptosis was determined by scoring cells displaying pyknotic nuclei (visualized with Hoechst 33342) [15,16]. The counting was performed blind by three different experimenters.

Statistical analyses
The experiments including more than two groups were analyzed by ANOVA or with the nonparametric equivalent Kruskal-Wallis test. Appropriate corrections such as Tukey's and Bonferroni's post hoc tests were used for multiple comparisons. Statistical analyses were performed using GraphPad PRISM, 5.0 (GraphPad Software, La Jolla, California, USA)

Human oxidized LDL particles activate the Ire1α and ATF6 pathways in MIN6 cells and isolated human islets
To investigate whether the induction of ER stress contributes to the adverse effects of human oxidized LDL, MIN6 cells were incubated with human mildly oxidized LDL at a 2 mmol/l cholesterol concentration and at different incubation times. As a positive control for ER stress, MIN6 cells were cultured for 6 h with Thapsigargin (thaps), a sarcoendoplasmic-reticulum Ca 2+ -ATPase pump inhibitor [25]. As expected, we found that this chemical compound activated PERK and increased the phosphorylation of its direct substrate, eukaryotic translation initiation factor 2 subunit alpha (eIF2α) (Fig 1a). The ER stress sensor Ire1α was phosphorylated by thaps in MIN6 cells (Fig 1a). Phosphorylation of Ire1α also occurred in response to oxidized LDL, whereas the induction of the PERK pathway was not detectable in this context (Fig 1a). In support of Ire1α activation, splicing of Xbp1 increased in MIN6 cells cultured with oxidized LDL for 48 h (S1a Fig). When the unfolded protein response (UPR) is activated, the expression of heat shock protein family A (Hsp70) member 5 (HSPA5, also known as BiP) and Protein Disulfide Isomerase (PDI) is known to be increased [19,26]. In this regard, we found that UPR induced by thaps was associated with an increase in BiP and PDI expression in pancreatic betacells (Fig 1a), but this was not seen when the cells were exposed to oxLDL (Fig 1a and S1b Fig).
If the ER stress is severe and chronic, the DNA damage inducible transcription factor 3 (DDIT3; also known as CHOP) expression is known to be increased, leading to apoptosis [27,28]. Change in the expression of CHOP was found to be associated with diabetes [20,21] and the genetic ablation of Chop in mice can delay the development of diabetes [28]. Here, consistent with the activation of Ire1α, Chop/CHOP expression was increased in isolated human islets and MIN6 cells exposed to oxidized LDL (Fig 1b and 1c). As expected, induction of Chop mRNA by oxidized LDL was followed by the increase of the protein (S1c Fig). In contrast, the expression of Chop was not stimulated by 2 mmol/l native LDL-cholesterol (Fig 1b and 1c), which is in support of the harmless effect of the native LDL at this concentration that we have previously reported [15,16]. Activation of Atf6 is known to lead to an increase in DnaJ (Hsp40) homolog, subfamily C, member 3 (Dnajc3, also called p58IPK) mRNA level in response to ER stress inducers [29]. Here, the oxidized LDL elicited a three-four -fold elevation in p58IPK/P58IPK mRNA in MIN6 and isolated human islets cells (Fig 1b and 1c), respectively, thus mirroring induction of the ATF6 pathway. However, the human modified LDL did not induce the expression of Atf4/ATF4, suggesting that the PERK pathway is not activated by oxidized LDL (Fig 1b and 1c). The chemical chaperone 4-phenyl butyrate (PBA) improves the folding capacity and trafficking of mutant proteins out of the ER and thereby reduces the load of unfolded proteins in the ER [30]. As anticipated, the induction of Chop/CHOP and p58IPK/P58IPK mRNA by oxidized LDL was efficiently blocked in MIN6 and isolated human islet cells that were co-cultured with PBA (Fig 1b and 1c). HDL has been shown to protect against ER stress [26,31]. Co-culture of MIN6 and human islet cells with HDL and oxidized LDL abolished the elevation of Chop/CHOP and p58IPK/P58IPK elicited by the oxidized LDL (Fig 1d and 1e). This result suggests that the induction of ER stress plays a key role in the deleterious effects caused by oxidized LDL.

PBA partially prevents decreased insulin expression and apoptosis caused by oxidized LDL
We next investigated whether the activation of ER stress links oxidized LDL to beta cell dysfunction and death. We found a significant reduction in the number of pyknotic nuclei in MIN6 and isolated human islet cells that were co-cultured with the chemical chaperone 4-phenyl butyrate (PBA) (Fig 2a and 2b). Oxidized LDL-mediated apoptosis results in part from  Oxidized LDL and ER Stress in Pancreatic Beta-Cells decreased anti-apoptotic Mapk8ip1 (also called Ib1) and Bcl2 genes expression [12,13]. Here, the protective effect triggered by PBA was accompanied by a partial restoration in the expression of Ib1 and Bcl2 (Fig 2c and 2d), confirming the anti-apoptotic action of PBA against oxidized LDL. Apoptosis caused by oxidized LDL is known to involve Chop/CHOP [32]. Indeed, we found that the silencing of Chop by siRNAs ( S2 Fig) reduced pancreatic beta-cell death induced by oxidized LDL (Fig 3).
Beside apoptosis, oxidized LDL is known to impair insulin gene expression. Indeed, the loss of insulin expression appears earlier than both insulin secretion deficiency and cell death [15,16]. To determine whether the induction of ER stress by oxidized LDL contributes to the impaired insulin expression, MIN6 and isolated human islet cells were cultured with lipoproteins with or without PBA. Inhibition of the ER stress markers by PBA were accompanied by a partial restoration of insulin mRNA levels, indicating a role for ER stress in the deleterious effect of oxidized LDL (Fig 4a and 4b). Oxidized LDL have been shown to hamper glucoseinduced insulin secretion [16]. However, in our study, insulin secretion was not rescued by PBA co-treatment (data not shown), suggesting that induction of ER stress by the modified lipoproteins is not involved in the impaired insulin secretion. In addition to ER stress, the induction of CREM (also called as inducible cAMP early repressor; ICER) also accounts for the loss of insulin production, impaired glucose-induced insulin secretion and decline in beta-cell survival provoked by oxidized LDL [16]. However, in our study PBA efficiently alleviated the increase of ER stress markers and apoptosis, thus the chemical chaperone was unable to antagonize the oxidized LDL-induced augmentation of Icer/ICER in MIN6 and isolated human islets (Fig 5a and 5b). These results indicate that induction of Icer by oxidized LDL relies on mechanisms that do not involve ER stress.
ER stress links oxidative stress to beta-cell dysfunction caused by human oxidized LDL Beta cell exposed to oxidized LDL produces hydrogen peroxide [16]. Treatment with the antioxidant N-acetylcystein (NAC) antagonizes the loss of insulin production, insulin secretion and beta-cell death provoked by oxidized LDL [16]. Therefore, we sought to determine whether ER stress couples oxidative stress to beta-cell dysfunction and death caused by oxidized LDL. This link is further supported by our observation that exposure of MIN6 cells and isolated human islets to hydrogen peroxide rapidly increased Chop/CHOP and p58IPK/P58IPK mRNA levels (Fig 6a and 6b). In fact, the co-culture of MIN6 cells and isolated human islets with NAC prevented the increase in Chop/CHOP and p58IPK/P58IPK mRNA by oxidized LDL (Fig 6c  and 6d).

Discussion
In this study, we evaluated the contribution of human oxidized lipoproteins, at the concentrations observed in sera of atherogenic dyslipidemic patients on ER stress signaling. We found that mildly oxidized LDL induced IRE1α signaling. Typically the IRE1α branch elicits mitogen-activated protein kinase 8 MAPK8 (also called JNK1) activity and Chop/CHOP expression [21,33,34]. Further, induction of IRE1α is consistent with the previously described activation of MAPK8 in beta-cells exposed to oxidized LDL [13]. The rise of CHOP content favors betacell apoptosis [28]. In our study, Chop silencing indeed attenuated beta-cell death caused by the modified LDL. Thus, the activation of IRE1α signalling may contribute to apoptosis evoked by oxidized LDL. The level of Chop can also be stimulated by the PERK pathway. Once activated PERK phosphorylates eIF2α, thus activating ATF4 [34]. In turn, ATF4 stimulates the expression of Chop [34]. The contribution of PERK pathway to the induction of Chop is unlikely as the PERK pathway was not activated by oxidized LDL. The activation of ATF6 triggers the expression of P58IPK. The latter is a cytosolic inhibitor of PERK activity, which is thought to be critical for regulating the latter phase of the ER stress response [35]. The expression of P58IPK increased in response to oxidized LDL. Our result suggests that the induction of P58IPK by oxidized LDL inhibits the induction of PERK.
Beside the ER stress activation, we have previously shown an increase in the expression of ICER passive transcriptional repressor in isolated islets and insulin producing cells cultured Oxidized LDL and ER Stress in Pancreatic Beta-Cells with oxidized LDL [16]. Although increased level of ICER correlated with this of Chop, our data rule out a link between ER stress and ICER. While the elevation of the ER stress marker was attenuated by PBA, induction of ICER by oxidized LDL remained elevated in cells cultured with the chaperone. A major target of ICER is IB1 [36]. ICER competes with cAMP-dependent transcriptional activators including cAMP response element (CRE) binding protein 1 (CREB1) for binding the CRE within target genes [37]. As the result of ICER binding, CREB1 target gene expression is silenced [16]. The loss of the IB1 transcript by Icer is one of the mechanisms that couples oxidized LDL to beta-cell dysfunction and death [15,16]. Treatment of cells with the PBA partially prevented the loss of Ib1 mRNA. This result suggests that the loss of IB1 expression, and thereby the reduced insulin expression and apoptosis induced by oxidized LDL, rely on two independent mechanisms, one dependent of Icer and another one that involves the ER stress (Fig 7).
ER stress is a hallmark of beta-cell damage in T2D [30]. In this study, we unveiled that NAC efficiently antagonized the elevation of ER stress markers in response to oxidized LDL. We have previously shown activation of oxidative stress in beta-cells exposed to oxidized LDL, leading to peroxide formation [16]. Production of reactive oxygen species is known to cause damages to DNA, proteins, sugars, and lipids [38]. Modification of proteins by reactive species may lead to misfolded proteins and thereby, induce ER stress in beta cell exposed to oxidized LDL. Our findings strengthen the hypothesis that the supplementation with antioxidants may help in combating beta-cell failure in diabetes and its secondary cardiovascular complications. In a meta-analysis study, vitamin E and carotenoids antioxidant intake was associated with a reduced T2D risk [39]. Our data highlight a link between oxidative stress and ER stress, which could be efficiently antagonized by antioxidant treatments.
Supporting Information S1 Fig. mRNA level of ER stress markers in response to oxidized LDL. Quantification of (a) Xbp1 mRNA splicing and (b) Bip in MIN6 cells exposed to oxidized LDL. The mRNA level was quantified by quantitative real-time PCR in MIN6 cells cultured for 48 h with vehicle (V), 2 mmol/l cholesterol native (nLDL) or oxidized LDL (oxLDL). Xbp1 cDNA was amplified by PCR and digested with PstI enzyme. Spliced Xbp1 cDNA corresponds to the activated form. This form lacks the restriction site and consequently remains intact. Spliced and unprocessed Xbp1 was quantified by densitometry. The value obtained for processed Xbp1 was expressed as a ratio of the total Xbp1 mRNA levels for each sample. The expression of Bip was normalized against Rplp0 and the expression levels from cells cultured with vehicle were set to 100%. The ratio from cells cultured with vehicle was set to 100%. Data are the mean ± SEM of at least 3 independent experiments measured in triplicate ( Ã , P<0.05).

Acknowledgments
This work was supported by the "European Genomic Institute for Diabetes" (E.G.I.D., ANR-10-LABX-46), European Commission, the Regional Council Nord Pas de Calais and the European Regional Development Fund. We are grateful to Denis Fahmi, Mr. Gilles Dubuis, Mr Guy Niederhauser, Mr Raphael Boutry and Ms Laure Rolland for technical assistance.

Author Contributions
Conceptualization: GW PF AA.