Pro-inflammatory cytokines involved in the pathogenesis of type 1 diabetes deplete endoplasmic reticulum (ER) Ca2+ stores, leading to ER-stress and beta cell apoptosis. However, the cytokine-induced ER-stress response in beta cells is atypical and characterised by induction of the pro-apoptotic PKR-like ER kinase (PERK)–C/EBP homologous protein (CHOP) branch of the unfolded protein response, but defective X-box binding protein 1 (XBP1) splicing and activating transcription factor 6 activation. The purpose of this study was to overexpress spliced/active Xbp1 (XBP1s) to increase beta cell resistance to cytokine-induced ER-stress and apoptosis.
Xbp1s was overexpressed using adenoviruses and knocked down using small interference RNA in rat islet cells. In selected experiments, Xbp1 was also knocked down in FACS-purified rat beta cells and rat fibroblasts. Expression and production of XBP1s and key downstream genes and proteins was measured and beta cell function and viability were evaluated.
Adenoviral-mediated overproduction of Xbp1s resulted in increased XBP1 activity and induction of several XBP1s target genes. Surprisingly, XBP1s overexpression impaired glucose-stimulated insulin secretion and increased beta cell apoptosis, whereas it protected fibroblasts against cell death induced by ER-stress. mRNA expression of Pdx1 and Mafa was inhibited in cells overproducing XBP1s, leading to decreased insulin expression. XBP1s knockdown partially restored cytokine/ER-stress-driven insulin and Pdx1 inhibition but had no effect on cytokine-induced ER-stress and apoptosis.
XBP1 has a distinct inhibitory role in beta cell as compared with other cell types. Prolonged XBP1s production hampers beta cell function via inhibition of insulin, Pdx1 and Mafa expression, eventually leading to beta cell apoptosis.
KeywordsApoptosis Beta cell ER-stress Insulin Secretion PDX-1 UPR Unfolded protein response XBP-1
Adenoviruses expressing the enhanced green fluorescent protein
Adenoviruses expressing the luciferase protein
Activating transcription factor
BCL2-associated X protein
Binding immunoglobulin protein
C/EBP homologous protein
Musculoaponeurotic fibrosarcoma oncogene homologue A
Multiplicity of infection
Mouse X-box binding protein 1
Neurogenic differentiation 1
Pancreatic and duodenal homeobox 1
PKR-like ER kinase
Small interference RNA
Unfolded protein response
X-box binding protein 1
Spliced/active X-box binding protein 1
Type 1 diabetes is an autoimmune disease triggered by progressive destruction of beta cells during insulitis. Pro-inflammatory cytokines released by immune cells invading the islets contribute to beta cell dysfunction and apoptosis during insulitis . We have previously shown that cytokines deplete endoplasmic reticulum (ER) Ca2+ stores, leading to activation of the unfolded protein response (UPR) . The UPR seeks to restore normal ER function by attenuating translation, increasing folding capacity through upregulation of ER chaperones and promoting degradation of misfolded proteins via the ER-associated degradation (ERAD) pathway. If the UPR is unable to restore cell function, it eventually triggers the apoptosis cascade . The UPR is mediated by the activation of three ER transmembrane sensor proteins: the activating transcription factor (ATF)6α, the PKR-like ER kinase (PERK) and the endoribonuclease inositol-requiring enzyme (IRE)1α .
IRE1α splices Xbp1 mRNA, leading to translation of the spliced/active form of the X-box binding protein-1 (XBP1s) transcription factor . XBP1s induces production of chaperones and ERAD components, which respectively promote folding and degradation of misfolded proteins, thereby contributing to cellular recovery following ER-stress [5, 6].
Although it is well established that cytokines induce ER-stress in beta cells , the role of ER-stress in cytokine-induced beta cell apoptosis remains controversial. Thus while some studies suggest that C/EBP homologous protein (CHOP)  and binding immunoglobulin protein (BIP)  play pro- and anti-apoptotic roles in this process respectively, others failed to observe any association between ER-stress markers and cytokine-induced beta cell apoptosis [9–11]. We observed that the cytokine-induced ER-stress response is atypical and characterised by strong induction of the pro-apoptotic PERK–ATF4–CHOP branch of the UPR, but with defective XBP1 splicing and ATF6 activation [2, 12]. We also showed that IFN-γ alone decreases expression of Xbp1s and several ER chaperones . We hypothesised that the defective ATF6 and XBP1s activation, and subsequent deficient production of ER proteins may deprive beta cells of a key mechanism for survival, increasing their susceptibility to cytokine/nitric oxide-mediated ER-stress and apoptosis.
The aim of this study was to increase XBP1s levels in order to improve protein-folding in the ER, and therefore protect beta cells against ER-stress-induced apoptosis. We generated adenoviruses encoding the spliced/active form of XBP1 (XBP1s) in rat dispersed islets and FACS-purified beta cells, and studied the molecular consequences of this overproduction, as well as its effects on beta cell function and viability. Surprisingly, XBP1s overproduction inhibited beta cell function and insulin expression, and triggered beta cell apoptosis, whereas it protected rat fibroblasts from cyclopiazonic acid (CPA)-induced apoptosis. XBP1s knockdown experiments suggest that this could be due to downregulation of the beta cell-specific transcription factors pancreatic and duodenal homeobox 1 (PDX1) and musculoaponeurotic fibrosarcoma oncogene homologue A (MAFA). Altogether our data suggest that XBP1s is not crucial for primary beta cell survival during ER-stress induced by cytokines or CPA, and that it plays a distinct inhibitory role in beta cells compared with other cell types.
Male Wistar rats (Charles River Laboratories, Brussels, Belgium) were housed and manipulated according to the guidelines of the Belgian Regulations for Animal Care; all performed experiments were approved by the local Ethical Committee. Rat islets were isolated by collagenase digestion followed by hand-picking under a stereomicroscope. Islets were then dispersed  and primary beta cells purified by autofluorescence-activated cell sorting (FACS Aria; BD Bioscience, San Jose, CA, USA) . The preparations used in the present study contained 88.1 ± 4.5% beta cells (n = 5). Dispersed islets or FACS-purified beta cells were pre-cultured in complete beta cell medium supplemented with 5% heat-inactivated FBS from 20 to 48 h for recovery . Experiments were then conducted in medium without serum [2, 13]. Rat fibroblast 208F cells (ECACC, Salisbury, UK) were cultured in DMEM supplemented with 10% FCS .
Cytokines and ER-stressors
The following cytokine concentrations were used: recombinant human IL-1β (a kind gift from C. W. Reinolds, National Cancer Institute, Bethesda, MD, USA) at 50 U/ml; and recombinant rat IFN-γ (R&D systems, Abingdon, UK) at 500 U/ml (0.036 µg/ml) [17, 18]. The sarco/endoplasmic reticulum calcium ATPase-2 inhibitor, cyclopiazonic acid (CPA) (Sigma-Aldrich, Bornem, Belgium), which induces rapid depletion of ER Ca2+ thus leading to ER-stress was dissolved in DMSO and used at a concentration of 25 μmol/l in beta cells and 50 μmol/l in 208F [2, 19]
Generation of recombinant adenoviruses and cell infection
To produce XBP1 in insulin-secreting cells, we generated recombinant adenoviruses comprising fragments of the mouse Xbp1 (mXbp1) mRNA (gi|18139942|gb|AF443192.1) . The following primers (Electronic supplementary material [ESM] Fig. 1) containing the SalI (forward) and HindIII (reverse) cloning sites (indicated by underlining) were used to amplify the complete spliced mXBP1 (1,363 bp): forward 5'-ATCGTCGACCCGGGACTACAGGACCAATA-3′; reverse 5'-GCGCAAGCTT ATGTGATGGTCAGGGAAAGG-3′. Using an alternate reverse primer 5'-CGCAAGCTTGGCAAGAAGGTGGTCTCAGA-3′, we were able to amplify either 1,129 bp corresponding to a truncated fragment of the spliced mXbp1 or the unspliced mXbp1 (ESM Fig. 1). These PCR fragments were then inserted in the pcDNA3 vector. Truncated spliced mXbp1 and unspliced mXbp1 were distinguished by sequencing. The PCR fragment encoding the unspliced Xbp1 mRNA is 5′-truncated and its splicing by endogenous IRE1α activity results in translation of a truncated and less active mXBP1 (ESM Fig. 1). Adenoviruses were generated using the pShuttle/pAdEasy-1 vector system [17, 21]. Adenoviruses expressing the enhanced green fluorescent protein (Ad-eGFP) or adenoviruses expressing the luciferase protein (Ad-LUC) were used as controls . Rat dispersed islets/FACS-purified beta cells were infected as previously described .
Transient transfection and luciferase assays
The reporter plasmid containing the luciferase gene under the control of five UPR elements was kindly provided by R. Prywes (Columbia University Biological Sciences, New York, USA). At 24 h after infection, dispersed islets cells were co-transfected with the UPR elements vector and the internal control pRL-CMV plasmid. Luciferase activities in the cell lysates were determined 48 h later in a luminometer (TD-20/20; Turner Designs, Sunnyvale, CA, USA) using the dual-luciferase reporter assay system according to the manufacturer’s instructions (Promega, Madison, WI, USA) . Luciferase activities were normalised by renilla activities of the co-transfected plasmid (pRL-CMV).
Insulin secretion measurements
At 48 h after infection, dispersed islet cells were washed in KRBH–HEPES buffer and static insulin secretion was assessed for 2 h in KRBH supplemented with either 1.7 mmol/l glucose or 16.7 mmol/l glucose + 10 µmol/l forskolin. Insulin release was measured by radioimmunoassay .
Small interference RNA (siRNA) transfection in rat dispersed beta cells was conducted according to a protocol recently developed in our laboratory  using DharmaFECT 1 (Termo Scientific, Chicago, IL, USA) with a final concentration of 30 nmol/l siRNA. The efficiency of transfection was >90% . A control siRNA (Allstars Negative Control; Qiagen, Venlo, the Netherlands) and two different BLOCK-iT Stealth Select siRNA (Invitrogen, Paisley, UK) designed against rat Xbp1 were used (sequences, see ESM Table 1). Cells were then cultured for a 48 h recovery period before being collected or exposed to CPA/cytokines as indicated.
Assessment of cell viability and caspase 3 activity
The percentage of viable, apoptotic and necrotic cells was determined using the DNA-binding dyes propidium iodide (5 µg/ml; Sigma-Aldrich) and Hoechst 33342 (5 µg/ml; Sigma-Aldrich) [2, 25]. The cells were examined by inverted fluorescence microscopy (Axiovert 200; Carl Zeiss, Zaventem, Belgium). A kit for live cells (NunView 488 Caspase-3 Kit; Biotium, Hayward, CA, USA) was used to determine caspase-3 activation . A minimum of 500 cells were counted in each experimental condition and viability or caspase activity was evaluated by two independent observers, one of them unaware of sample identity.
mRNA extraction and quantitative RT-PCR
Poly(A) + mRNA was isolated from dispersed islet cells using a kit (Dynabead mRNA DIRECT; Invitrogen) and reverse-transcribed as previously described . Quantitative PCR was performed using IQ SYBR Green Supermix (BioRad, Nazareth Eke, Belgium) in an IQ5 instrument (BioRad). Expression values were corrected to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase. Cytokine or CPA treatment does not modify Gapdh expression in insulin-producing cells under the present experimental conditions [18, 27]. The primers used in this study are described in the ESM Table 1.
Western blot analysis
Cells were washed with cold PBS and lysed with Laemmli buffer. Lysates were resolved by 12% SDS-PAGE and transferred to polyvinylidene fluoride membranes. Immunoblot analyses were performed using an antibody directed against XBP1 (sc-346; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and normalised using anti-β-actin (Cell Signaling, Boston, MA, USA). Horseradish peroxidase-conjugated goat anti-rabbit IgG was used as a secondary antibody (Santa Cruz Biotechnology). Immunoreactive bands were revealed using a chemiluminescent substrate (SuperSignal West Femto; Termo Scientific, Chicago, IL, USA), detected using a LAS-3000 CCD camera (Fujifilm Europe-GE Healthcare Europe, Diegem, Belgium).
Dispersed islet cells were grown on glass culture slides (BD Biosciences Europe, Erembodegem, Belgium), fixed for 15 min in fresh 4% paraformaldehyde, rinsed in PBS and permeabilised for 5 min in PBS-Triton X-100 0.1%. Slides were then blocked using PBS-goat serum 5% and incubated overnight at 4°C in the presence of polyclonal rabbit anti-BCL2-associated X protein (BAX) (1/1000; Santa Cruz Biotechnology) and monoclonal mouse anti-cytochrome c or anti-ATP Synthase β (1/1000; BD Biosciences). Cells were washed and further exposed for 1 h to appropriate Alexa fluor 488 or 555-conjugated antibodies (1/1000; N. V. Invitrogen, Merelbeke, Belgium). After washing, cells were stained with Hoechst 33342, mounted, analysed and photographed using fluorescence microscopy (Axio Imager; Carl Zeiss).
Data are presented as means ± SEM. Comparisons were performed by two-tailed paired Student’s t test or by ANOVA followed by t tests with Bonferroni correction for multiple comparison. A p value of ≤ 0.05 was considered statistically significant.
Adenovirus-mediated Xbp1s overexpression induces XBP1 activity
To assess the role of XBP1s on beta cell function and survival, we generated several adenovirus vectors leading to expression of the following Xbp1 mRNAs (ESM Fig. 1): (1) a full-length active (spliced form) of mouse Xbp1 mRNA (Ad-XBP1s: 371 amino acids); (2) a 3′ truncated fragment of the spliced Xbp1 mRNA (Ad-tXBP1s: 326 amino acids); and (3) the unspliced Xbp1 mRNA (Ad-XBP1u: 267 amino acids). As additional controls, Ad-eGFP or Ad-LUC were also used.
To check whether the viral XBP1 protein drove XBP1 activity in dispersed islet cells, we studied the expression of the reporter gene luciferase under the control of five UPR elements, known to be the target binding site for XBP1s . Infection using Ad-XBP1s dose-dependently increased XBP1 activity, while infection with Ad-eGFP had no effect (Fig. 1d). Interestingly, infection with Ad-tXBP1s induced a two- to threefold activation of the reporter, indicating that the truncated form of XBP1 still conserved a residual transactivating capacity. Ad-XBP1u infection was unable to increase XBP1 activity. Importantly, XBP1s overproduction significantly increased mRNA expression of XBP1 target genes  Edem (also known as Edem1), Sec61α, Pdi-P5 and P58 IPK (Fig. 1e).
XBP1s overproduction induces beta cell apoptosis
In contrast to beta cells and in accordance with previous studies [29, 30], XBP1s overproduction did not induce apoptosis in the 208F rat fibroblast cell line. In fact, it protected these cells against apoptosis induced by 48 h of CPA treatment (Fig. 2g).
To exclude the possibility that XBP1s overproduction triggered apoptosis through non-specific ER-stress, we studied mRNA expression of Bip (also known as Hspa5), Chop (also known as Ddit3) and endogenous rat Xbp1 in infected cells. Ad-XBP1 infection (MOI 15) did not modify Chop or Bip expression, contrasting with the effect of 24 h exposure to CPA (ESM Fig. 2). Interestingly, expression of the endogenous rat Xbp1s mRNA was significantly decreased following Ad-XBP1s infection, suggesting a negative feedback action by the exogenous mXBP1s. Together, these data indicate that XBP1s does not induce apoptosis through increased ER-stress.
XBP1s overproduction impairs insulin secretion and expression
XBP1 knockdown partially prevents cytokine/CPA-induced insulin mRNA decrease
In an attempt to better characterise the role of XBP1 in pancreatic beta cells, we used an adenoviral approach to overexpress the active/spliced form of XBP1 (Xbp1s) in rat islet cells, leading to increased XBP1 activity and expression of several XBP1s target genes. Surprisingly, prolonged XBP1s overproduction impaired insulin expression and glucose-stimulated insulin secretion, and led to BAX translocation, cytochrome c release, activation of caspase 3 and beta cell apoptosis. Neither Xbp1s overexpression nor knockdown modified cytokine- or CPA-induced ER-stress and apoptosis. In contrast, XBP1s overproduction had no effect on basal rat fibroblasts survival and protected these cells against CPA-induced apoptosis. XBP1s overproduction strongly decreased insulin, Pdx1 and Mafa expression in beta cells and XBP1 knockdown partially restored cytokine- or CPA-driven insulin and Pdx1 inhibition. Taken together, our results suggest that XBP1s does not play a key role in beta cell survival during ER-stress induced by CPA or cytokines. However, sustained production of XBP1s inhibits beta cell function and eventually triggers the mitochondrial pathway of apoptosis, probably secondary to inhibition of Pdx1 and Mafa expression.
Based on studies performed in other cell types, the IRE1α–XBP1 pathway is usually considered to be a protective pathway of the UPR [35–37]. We observed that XBP1s overproduction protects rat fibroblasts against CPA-induced apoptosis. In contrast, prolonged XBP1s overproduction induces beta cell apoptosis. This is not due to generation of non-specific ER-stress, since expression of the ER-stress markers Bip and Chop were not modified. In addition, viruses encoding truncated forms of XBP1 induced little (Ad-tXBP1s) or no apoptosis (Ad-XBP1u), and there was a close correlation between XBP1s activity and apoptosis, indicating that the increased apoptosis is specifically triggered by XBP1s. It is likely that the discrepancy between fibroblasts and beta cells is due to the role, discovered here, of XBP1s in inhibiting the expression of Pdx1 and Mafa, two key beta cell-specific transcription factors that are crucial for maintaining function  and, at least for Pdx1, survival [38, 39]. Xbp1 knockdown partially reversed Pdx1 downregulation, but failed to prevent cytokine- or CPA-induced apoptosis, suggesting that Pdx1 downregulation is not a major determinant of beta cell apoptosis in response to cytokines or CPA. Our data indicate that XBP1s overproduction triggers the intrinsic pathway of apoptosis, but further studies need to be done to elucidate the exact mechanism underlying the pro-apoptotic effect.
XBP1 knockdown had no effect on beta cell viability under control conditions or when the cells were treated with cytokines or CPA, suggesting that XBP1 does not affect the primary beta cell capacity to face some forms of ER-stress. This hypothesis is supported by the fact that Xbp1 knockdown had no effect on the expression of the ER-stress markers Bip and Chop, the latter a transcription factor shown to be increased in beta cells under pro-apoptotic ER-stress [15, 19]. This is in line with previous studies showing that XBP1s is not required for normal development of islets  while it is essential to development of liver  and plasma cells . A recent report also suggests that XBP1s is dispensable in neurons , which share many features with beta cells [43, 44].
We and others have previously shown that the IRE1α–XBP1s pathway of the UPR controls insulin expression, mainly through IRE1α-mediated degradation of insulin mRNA [27, 45, 46]. In the present study we demonstrate another level of regulation of insulin expression by the UPR, via XBP1s-dependent downregulation of Pdx1 and Mafa. XBP1s did not affect stability of insulin 1 or 2 mRNA (data not shown), suggesting a transcriptional event. XBP1s strongly decreased Mafa and Pdx1 expression, thereby suppressing the cooperative action between these transcription factors and Neurod1 in stimulating insulin transcription . Moreover, Xbp1 knockdown partially prevented cytokines and CPA-mediated decrease in insulin and Pdx1 mRNA expression [18, 27], supporting the hypothesis that XBP1s inhibits Pdx1 expression. XBP1s dose-dependently inhibited stimulated but not basal insulin secretion, suggesting that XBP1s interferes with the glucose signalling pathways leading to insulin release. XBP1s also downregulates expression of the glucose transporter Glut2. This finding is in accordance with a role of XBP1s in inhibiting Pdx1 and Mafa expression , and may also account for the effect of XBP1s on beta cell function, as GLUT2 plays a critical role in glucose sensing in beta cells .
While XBP1 knockdown partially restored insulin and Pdx1 expression in presence of CPA or cytokines, Mafa and Glut2 expression remained suppressed. Further studies are required to clarify the mechanisms involved in this apparent dissociation between the roles of PDX1 and MAFA in the regulation of key beta cell genes.
The classical view on the role of the IRE1α–XBP1s pathway in the UPR is that this pathway augments the folding capacity of the ER through increased expression of ER chaperones and stimulates degradation of misfolded proteins by the ERAD pathway [5, 6]. However, overexpression of Xbp1s in primary beta cells induced only a moderate (≤ twofold) increase in the expression of its target genes. A similar pattern was observed in fibroblasts , suggesting that XBP1s is necessary, but not sufficient to fully stimulate expression of its target genes. Accumulating evidence suggests that a cross-talk exists between the ATF6α and IRE1α–XBP1s pathways to achieve maximal induction of their downstream target genes [5, 6, 48]. A recent study revealed deleterious effects of ATF6α overexpression on beta cell function and expression of insulin, PDX1 and MAFA in INS-1 cells . Together with our data in primary cells, this suggests that the ATF6α and XBP1 pathways may have redundant roles in the regulation of insulin expression and secretion.
Under stimulatory conditions, production of insulin may represent up to 50% of total protein production, placing an enormous burden on the ER . As a consequence, beta cells are in a constant state of mild ER-stress, characterised by high basal expression of UPR transducers such as IRE1α, allowing them to adapt efficiently to the ever changing insulin demand . In other cell types, activation of the IRE1α-XBP1s pathway increases production of ER-proteins aiming to fold and/or degrade misfolded proteins. Together with previous studies [27, 46], our data suggest that the IRE1α–XBP1s pathway has a distinct role in beta cells, relative to the fine-tuning of insulin mRNA expression through transcriptional and post-transcriptional mechanisms, allowing for rapid adjustment of insulin levels under ER-stress conditions . However, due to the deleterious effects of prolonged increased XBP1s level, this mechanism has to be closely regulated in beta cells, resulting in a limited XPB1 splicing as shown by previous studies [2, 12] and the present data. These observations, together with other studies showing detrimental effects of prolonged eukaryotic translation initiation factor 2 alpha (eIF2α) or ATF6α activation in beta cells [24, 49], suggest that tight control of ER homeostasis is crucial to maintain beta cell function and survival.
This work was supported by grants from the Juvenile Diabetes Research Foundation International (1-2008-536), Fonds National de la Recherche Scientifique (FNRS-FRSM) Belgium, the Communauté Française de Belgique—Actions de Recherche Concertées (ARC), the European Union (Integrated Projects EuroDia in the Framework Programme 6 [FP6] and Novel Immunotherapies for Type 1 Diabetes [NAIMIT] in the FP7 of the European Community) and the Belgium Program on Interuniversity Poles of Attraction initiated by the Belgium State (IUAP P6/40). A. K. Cardozo is a Research Associate of the FNRS. We thank M. A. Neef, G. Vandenbroeck, M. Urbain, R. Leeman, A. M. Musuaya, R. Makhnas, J. Schoonheydt and S. Mertens from the Laboratory of Experimental Medicine, ULB, for excellent technical support.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.