Glucagon-Like Peptide-1 Agonists Protect Pancreatic β-Cells From Lipotoxic Endoplasmic Reticulum Stress Through Upregulation of BiP and JunB

Male Wistar rats (Charles River Laboratories, Brussels, Belgium) were housed and used according to the guidelines of the Belgian Regulations for Animal Care; experiments were approved by the local ethics committee. Islets were hand picked following collagenase digestion. β-Cells were purified from dispersed islet cells by autofluorescence-activated cell sorting (FACSAria; BD Bioscience, San Jose, CA) (30). The preparations, containing 91 ± 2% β-cells, were precultured overnight in Ham's F-10 medium containing 10 mmol/l glucose and 5% heat-inactivated FBS (31).

The rat insulin–producing INS-1E cell line (a kind gift from Prof. C. Wollheim, Centre Medical Universitaire, Geneva, Switzerland) was cultured in RPMI-1640 (with 2 mmol/l GlutaMAX-I) containing 5% FBS (32) and used at passages 56–75.

With the approval of the ethics committee in Pisa, islets of four multiorgan donors (two female and two male subjects, aged 64 ± 10 years, BMI 26 ± 1 kg/m²) were isolated, and cultured as previously described (12). The human islets contained 58 ± 2% β-cells, assessed following islet dispersion (12).

For FFA exposure, β-cells were cultured in medium with 1% BSA and without FBS; INS-1E cells were cultured in medium with 1% FBS and 1% BSA. Oleate and palmitate (Sigma-Aldrich, Schnelldorf, Germany) were dissolved in 90% ethanol and used at a final concentration of 0.5 mmol/l (10,33) and a FFA/BSA molar ratio of 3.4 (33). The sarcoendoplasmic-reticulum pump Ca²⁺-ATPase (SERCA) blockers cyclopiazonic acid (CPA) and thapsigargin (Sigma) were dissolved in DMSO and used at concentrations of 25 and 1 μmol/l, respectively. Salubrinal (ChemBridge, San Diego, CA) was diluted in DMSO and used at 75 μmol/l. Tautomycin was diluted in ethanol and used at 30 nmol/l. Exendin-4 was diluted in RPMI-1640 and used at 10 nmol/l for primary cells and 50 nmol/l for INS-1E cells (29). Forskolin and the PKA inhibitor H89 were diluted in DMSO and used at concentrations of 20 and 15 μmol/l, respectively (Sigma).

Apoptotic and necrotic β-cells and INS-1E cells were counted in fluorescence microscopy following staining with the DNA-binding dyes propidium iodide (5 μg/ml) and Hoechst 33342 (10 μg/ml) (11). The percentage human islet cell death was determined as described (12). The apoptosis index was calculated as [(% apoptosis in ER stressor–treated cells under indicated conditions transfected with negative or target siRNA − % apoptosis in ER stressor–treated negative siRNA-transfected cells)/(100% − % apoptosis in ER stressor–treated negative siRNA-transfected cells)] × 100. Apoptosis was also assessed by the Cell Death Detection ELISA^plus kit (Roche Diagnostics, Mannheim, Germany) (10). Forskolin did not modify INS-1E cell proliferation under the present experimental conditions (assessed by BrdU incorporation, data not shown).

To study cytochrome C release, cells were prepared as described in the online appendix supplementary Methods. After total protein quantification (Bio-Rad Laboratories, Munchen, Germany), equal amounts of protein were blotted using tubulin as cytoplasmic and apoptosis-inducing factor (AIF) as mitochondrial control.

Caspase 3 and 7 activation was measured with the Caspase-Glo 3/7 assay (Promega, Madison, WI) and caspase 12 activation with the CaspGlow Caspase-12 staining kit (Biovision, Mountain View, CA) (12).

Real-time PCR and Western blot were performed as described in the online appendix supplementary Methods.

Proteins were knocked down using small-interfering RNA (siRNA) against CHOP (Ambion, Austin, TX), XBP1, JunB (both from Invitrogen, Carlsbad, CA), GADD34, and BiP (both from Dharmacon, Chicago, IL). A negative control of 21-nucleotide duplex RNA with no known sequence homology was from Qiagen (Hilden, Germany). Lipid-RNA complexes were formed in Optimem1 (1 μl Dharmafect to 150 nmol/l siRNA; Dharmacon) and added at a final concentration of 30 nmol/l siRNA for transfection as described (12). Transfected cells were cultured for 1–4 days and subsequently treated.

Data are presented as means ± SE. Comparisons were performed by ANOVA followed by paired t test with the Bonferroni correction for multiple comparisons. A P value <0.05 was considered statistically significant.

In agreement with Yusta et al. (29), exendin-4 partially protected primary and clonal β-cells against the chemical SERCA2 blockers CPA and thapsigargin, and the adenylate cyclase stimulator forskolin had a similar effect (Fig. 1 A and E).

We next examined whether GLP-1R activation protects β-cells against the physiologically relevant ER stressors oleate and palmitate (10,,,–14). We also asked whether GLP-1 is antiapoptotic following selective activation of the PERK pathway. To this end, we used salubrinal, an inhibitor of eIF2α dephosphorylation (26) that, without causing ER stress, markedly upregulates ATF4 and CHOP and thereby triggers β-cell apoptosis (11). Exendin-4 and forskolin protected primary β- and INS-1E cells against oleate and palmitate after 14–24 h (Fig. 1,B and F), in keeping with previous reports (34,35), but the protection against palmitate was lost following prolonged exposure (Fig. 1,C and G). GLP-1R activation protected human islets against palmitate for 72 h (Fig. 1,D). GLP-1 agonists also prevented salubrinal-induced β-cell death (Fig. 1,B and F). The findings were confirmed using an enzyme-linked immunosorbent assay for DNA strand breaks (Fig. 1 H).

In clonal β-cells exposed to synthetic ER stressors, exendin-4 induces ATF4 and CHOP, thereby upregulating GADD34 (29). The recruitment of PP1 by GADD34 results in eIF2α dephosphorylation, and this was proposed to accelerate recovery from ER stress (29). In FFA-treated INS-1E cells, exendin-4 and forskolin did not modify ATF4 and CHOP mRNA (Fig. 2,A and supplementary Fig. S2A), but after 14 h, ATF4 and CHOP protein expression doubled in palmitate-treated cells (Fig. 2,B), and this induced GADD34 mRNA (Fig. 2,A). When combined with salubrinal, per se, a potent ATF4-CHOP inducer (11), forskolin further upregulated ATF4-CHOP-GADD34 RNA and protein (Fig. 2 and supplementary Fig. S2).

We next examined whether this GADD34 induction causes eIF2α dephosphorylation. Exendin-4 and forskolin effectively decreased salubrinal-induced eIF2α phosphorylation in INS-1E cells (Fig. 2,C and supplementary Fig. S2C) and tended to decrease palmitate-induced eIF2α phosphorylation after 14 h (Fig. 2,C). GADD34 knockdown by siRNA (55% decrease in mRNA) increased eIF2α phosphorylation by 75% in salubrinal plus forskolin–treated INS-1E cells (14 h, n = 3). A detailed time-course analysis revealed that very early on (<1 h), forskolin diminished eIF2α phosphorylation by salubrinal (Fig. 2,D), suggesting that forskolin immediately affects eIF2α phosphorylation, in addition to enhancing its dephosphorylation via ATF4-CHOP-GADD34. The effect of exendin-4 and forskolin on eIF2α was not PERK dependent, as phospho-PERK was unchanged or even increased (Fig. 2,E and supplementary Fig. S2D). To examine whether GADD34-PP1 dephosphorylation of eIF2α mediates the antiapoptotic effect of forskolin, we used the PP1 inhibitor tautomycin (36). In the presence of tautomycin, forskolin did not prevent eIF2α phosphorylation by salubrinal (Fig. 3,A). Despite the high phospho-eIF2α levels, forskolin still protected against salubrinal-induced apoptosis (Fig. 3,B), suggesting that the eIF2α phosphorylation state does not determine forskolin-mediated β-cell survival. To further evaluate the participation of ATF4-CHOP-GADD34 in the forskolin protection, the feedback loop was interrupted by RNAi. CHOP knockdown prevented CHOP protein induction by palmitate/salubrinal plus forskolin (Fig. 3,C and F) and decreased GADD34 mRNA (Fig. 3,D and G), demonstrating efficient disruption of CHOP-GADD34 signaling. Decreasing CHOP expression partially prevented palmitate-induced (12) and salubrinal-induced apoptosis (Fig. 3,E and H). The antiapoptotic effect of forskolin was unaffected by knockdown of CHOP (Fig. 3 E, H, and I) or GADD34 (8 ± 1% apoptosis vs. 6 ± 1% with control siRNA following salubrinal plus forskolin exposure, compared with 18–22% after salubrinal exposure, 14 h, n = 3), strongly arguing against their participation in β-cell protection. CHOP knockdown did not abrogate the forskolin protection against CPA (supplemental Fig. S3), casting doubt onto its proposed importance for GLP-1 protection against ER stress (29).

We next examined the potential involvement of the other ER stress branches in the protective effects of GLP-1 (supplementary Fig. S1). Exendin-4 and forskolin upregulated XBP1s mRNA and protein in INS-1E cells basally (Fig. 4 and supplementary Fig. S4), suggesting that GLP-1R agonists activate IRE1, but they did not enhance the XBP1s induction by FFAs (Fig. 4). Similar changes were observed for ORP150 mRNA, a XBP1s transcriptional target (data not shown).

The role of XBP1s in β-cell death was studied using RNAi, resulting in 74% decreased XBP1 protein expression. XBP1s knockdown increased apoptosis by oleate (Fig. 4,C) but not palmitate or salubrinal (Fig. 4,E and G), illustrating that the signaling response engaged by different ER stressors has context-dependent cell survival outcomes (supplementary Fig. S8A). The antiapoptotic effect of forskolin in oleate-, palmitate-, and salubrinal-treated cells did not depend on XBP1s expression, as XBP1 knockdown did not abrogate its protective effect (Fig. 4 H).

BiP is a key ATF6-dependent ER chaperone. Exendin-4 and forskolin doubled BiP protein levels in FFA-treated INS-1E cells; mRNA induction was observed for oleate only (Fig. 5,A–C and supplementary Fig. S4). Conversely, forskolin induced BiP mRNA but not protein in salubrinal-exposed cells (Fig. 5 and supplementary Fig. S4), pointing to posttranscriptional regulation. Forskolin also increased basal BiP protein (Fig. 5,C and supplementary Fig. S4C). BiP protein induction by forskolin was confirmed in primary β-cells (Fig. 5 D and supplementary Fig. S4D). p58^IPK mRNA levels, another ATF6 target, were comparable to BiP (data not shown).

The role of BiP induction in β-cell survival was examined by RNAi. The relatively modest 50% BiP knockdown (Fig. 5,E and G) is probably due to the marked BiP stability in INS-1E cells (data not shown). BiP knockdown increased apoptosis induced by oleate and palmitate (Fig. 5,F and H and supplementary Fig. S8A), which is in keeping with the reverse experiment showing palmitate resistance in BiP-overexpressing cells (13). The antiapoptotic action of forskolin was not affected by interference with BiP in oleate-exposed cells (Fig. 5,F), but a marked loss of protection against palmitate was observed (Fig. 5 H and I), showing that BiP induction is one of the mechanisms by which GLP-1 protects against ER stress.

To further elucidate the mechanisms of β-cell protection, we examined the apoptosis pathways downstream of ER stress that can be ER specific or employ the classic intrinsic mitochondrial cell death pathway.

To assess the contribution of mitochondria in the apoptotic process, mitochondrial cytochrome C release and the antiapoptotic protein Bcl-2 were measured. Palmitate and salubrinal decreased Bcl-2 and stimulated cytochrome C release, as did CPA (Fig. 6 and supplementary Fig. S5) but not oleate (data not shown). Both exendin-4 and forskolin restored Bcl-2 protein to control levels in salubrinal-treated cells (Fig. 6,C and supplementary Fig. S5B) and thereby largely prevented cytochrome C release (Fig. 6,B and supplementary Fig. S5A). It has been suggested that CHOP mediates Bcl-2 downregulation (37), but we did not observe changes in Bcl-2 expression following CHOP knockdown (data not shown). In palmitate-treated β-cells, no restoration of Bcl-2 expression was observed, while forskolin increased Bcl-2 levels in control and oleate conditions (Fig. 6 and supplementary Fig. S5).

Cytochrome C released from the mitochondria binds to the proapoptotic protease-activation factor 1 (APAF1) and caspase 9, constituting the apoptosome that activates the effector caspase 3 (38). This is counteracted by X-chromosome–linked inhibitor of apoptosis (XIAP), a potent caspase inhibitor that binds to caspase 9 or 3 (39). XIAP mRNA was not modified (data not shown), but as with Bcl-2 XIAP protein expression was reduced by palmitate, salubrinal, and CPA (Fig. 6 C and supplemental Fig. S5C). Again, exendin-4 and forskolin restored XIAP levels in salubrinal- and CPA-treated cells but not in palmitate-treated cells (supplemental Fig. S5C).

In keeping with the restoration of Bcl-2 and XIAP and the inhibition of cytochrome C release, caspase 3 activation was nearly entirely prevented by exendin-4 and forskolin in salubrinal- or CPA-treated β-cells (Fig. 6,D). Exendin-4 and forskolin partially reduced caspase 3 activation by palmitate, but residual activity remained three- to fourfold above control (Fig. 6 D).

During ER stress, caspase 3 can also be activated by caspase 12 (40). FFAs and CPA activate ER membrane–localized caspase 12 in β-cells (12). Exendin-4 or forskolin did not modify caspase 12 activation in FFA-treated cells but abolished its activation by CPA (Fig. 6 E). This represents an additional mechanism through which GLP-1 can protect against synthetic ER stress, but it does not play a role in physiological ER stress.

The transcription factor JunB was recently shown to be antiapoptotic in β-cells exposed to chemical ER stressors (41). JunB protein expression was downregulated by CPA, palmitate, and salubrinal but not by oleate (Fig. 7,A and supplemental Fig. S6). Exendin-4 restored JunB protein to control levels, and forskolin further induced JunB by two- to fivefold (Fig. 7,A and supplemental Fig. S6). This marked induction was not observed at the mRNA level (data not shown). Forskolin also induced JunB protein in primary β-cells by threefold (Fig. 7 B and supplemental Fig. S6B).

An 85% JunB knockdown by siRNA (Fig. 7,C, E, and G) sensitized INS-1E cells to oleate (Fig. 7,D). The protection by forskolin against oleate was abolished in JunB-depleted cells demonstrating that cAMP-stimulated JunB induction is necessary and sufficient to prevent cell death (Fig. 7,D and I). Loss of protection against palmitate was also observed with JunB knockdown, though it was partial only (Fig. 7,F). Salubrinal-induced apoptosis was not modified by JunB knockdown, and neither was the antiapoptotic effect of forskolin (Fig. 7 H and I).

Finally, we examined whether the upregulation of these antiapoptotic proteins by exendin-4 and forskolin was PKA dependent. The PKA inhibitor H89 prevented BiP, JunB, and Bcl-2 protein induction basally and following palmitate treatment (supplemental Fig. S7A) and abrogated the protective effect of forskolin (supplemental Fig. S7B).

Lipotoxic ER stress–mediated β-cell dysfunction and apoptosis may be relevant in the development of type 2 diabetes (9). Therapies that increase β-cell resistance to FFAs and ER stress could have considerable clinical impact, as they might prevent type 2 diabetes or attenuate the progression of the disease. Activation of the GLP-1R by agonists such as exendin-4 induce cAMP-PKA, extracellular signal–regulated kinase 1/2 mitogen-activated protein kinase, and phosphoinositide 3 kinase-Akt signaling, leading to improved β-cell function through increased insulin biosynthesis and secretion, β-cell proliferation (at least in rodents), and decreased β-cell death (28,42). It has recently been shown that GLP-1R activation also stimulates β-cell survival during ER stress. In rodent models, exendin-4 treatment decreased in vivo pancreatic expression of CHOP, XBP1s, and BiP (29,43), but detailed in vitro studies demonstrated that GLP-1R activation leads to a cAMP-PKA–dependent enhancement of ATF4-CHOP-GADD34 signaling, followed by PP1-mediated eIF2α dephosphorylation and accelerated recovery of translational repression (29). Since ER stress signaling and outcomes are stimulus dependent, we examined whether these mechanisms mediate GLP-1R protection from the more pathophysiologically relevant lipotoxic ER stress. We observed that the protective mechanisms previously described for synthetic ER stress (29) are not operative in this ER stress condition and that exendin-4 and forskolin reduce FFA-induced apoptosis by increasing β-cell defense mechanisms.

The PERK-regulated ER stress branch that modulates global protein translation is necessary for β-cell survival. Humans and/or mice with a PERK deletion or Ser51Ala substitution in eIF2α (precluding its phosphorylation by PERK) cannot attenuate insulin translation and have progressive β-cell loss (24,44,45), suggesting that PERK-dependent eIF2α phosphorylation is beneficial to β-cells. Excessive or prolonged eIF2α phosphorylation and downstream signaling is, however, poorly tolerated by β-cells and specifically exacerbates FFA-induced apoptosis (11). We therefore asked whether increased signaling in the PERK pathway by exendin-4 during lipotoxic ER stress would be pro- or antiapoptotic. The present and previous (29) data support the idea that exendin-4 and forskolin enhance ATF4-CHOP-GADD34 expression. This does not, however, explain the antiapoptotic outcome because CHOP or GADD34 knockdown, or chemical PP1 inhibition, do not abrogate the protective effect of forskolin in lipotoxic or synthetic ER stress conditions (Fig. 3 and supplementary Fig. S3). We therefore undertook a search for alternative mechanisms mediating β-cell survival by GLP-1R activation in lipotoxic ER stress.

Activation of IRE1 leads to the expression of XBP1s that induces ER chaperones and ER expansion; for the latter lipid synthesis genes are upregulated. The induction of XBP1s by exendin-4 and forskolin suggests that these agents activate IRE1, either directly or by triggering a mild UPR (through enhanced protein synthesis) with activation of IRE1 but not PERK. GLP-1 and high glucose mildly activate IRE1 (46), and chronic exposure to high glucose causes XBP1s induction in β-cells that is not accompanied by ATF4 or CHOP expression (46,47). Knockdown of XBP1s leads to a small decrease in β-cell viability (present data). XBP1 RNAi potentiated oleate- but not palmitate-induced β-cell apoptosis (Fig. 4 H and supplementary Fig. S8A), suggesting differential activation of pro- and antiapoptotic signals by the FFAs downstream of IRE1; moreover, XBP1s might play a role in oleate handling. The prosurvival effect of forskolin was not altered by XBP1 knockdown, excluding it plays part in the observed protection.

IRE1 activation can directly cleave caspase 12. GLP-1 agonists did not modify caspase 12 cleavage by FFAs but nearly completely prevented its activation by CPA. In parallel, forskolin increased CPA-mediated PERK phosphorylation, suggesting that ER stress is not improved in this condition but that UPR transducers can be differentially regulated by GLP-1. ER stress can trigger apoptosis via the mitochondrial pathway of cytochrome C release, formation of the apoptosome, and caspase 3 activation. eIF2α phosphorylation is essential to proteasome degradation of antiapoptotic Bcl-2 family members during chemical ER stress (48). We currently observed that palmitate, but not oleate, induced degradation of Bcl-2 and cytochrome C release. eIF2α phosphorylation by salubrinal and CPA also decreased Bcl-2 with consequent cytochrome C translocation. The repression of eIF2α phosphorylation by exendin-4 and forskolin may restore Bcl-2 expression and thus prevent cytochrome C release and caspase 3 activation by salubrinal and CPA, but they have little effect on palmitate-induced apoptosis. GLP-1R activation also restored the antiapoptotic protein XIAP in cells exposed to salubrinal or CPA but not to palmitate, delineating context-dependent mechanisms of GLP-1R–mediated antiapoptotic effects (supplementary Fig. S8B).

BiP is an ATF6-dependent ER chaperone and a master regulator of UPR activation. GLP-1 agonists doubled BiP expression in a PKA-dependent manner, while a 50% reduction in BiP expression sensitized β-cells to FFAs, confirming the role of ER stress in lipotoxic β-cell apoptosis. The upregulation of BiP by forskolin protected β-cells against palmitate, as the protection was partially lost following BiP knockdown (Fig. 5 I). Conclusive demonstration for the role of BiP will come from studies examining whether GLP-1 agonists are protective over and above adenovirus-mediated BiP overexpression. GLP-1R activation thus increases β-cell protein folding capacity and defense mechanisms against lipotoxic ER stress, prompting us to examine the potential contribution of other antiapoptotic proteins.

GLP-1 was previously shown to induce several early response genes in INS-1 cells, among which was JunB (49). JunB was recently shown to play a protective role in ER-stressed β-cells (41). Palmitate, salubrinal, and CPA decreased JunB expression, which was restored and further increased by exendin-4 and forskolin (supplementary Fig. S6). This PKA-dependent JunB induction is an important promoter of β-cell survival since JunB RNAi sensitized cells to FFAs and it abrogated the antiapoptotic effect of GLP-1R activation.

GLP-1 analogs provide hope that the natural history of type 2 diabetes can be modified by the preservation of functional β-cell mass (28). We demonstrate here that exendin-4 and cAMP production protect β-cells against ER stress in a context-dependent manner. Lipotoxic β-cell apoptosis is prevented by increasing cellular defense mechanisms through the induction of the ER chaperone BiP and the antiapoptotic protein JunB, whereas protection from synthetic ER stressors occurs via inactivation of caspase 12 and upregulation of Bcl-2 and XIAP to block mitochondrial apoptosis (supplementary Fig. S8). The present identification of the proapoptotic ER stress pathways and downstream signaling modulated by GLP-1 is an important milestone for the development of targeted approaches to alleviate β-cell ER stress in diabetes.

This work was supported by the European Union (Integrated Project EuroDia LSHMCT-2006-518153 and STREP Savebeta contract no. 036903 in the Framework Programme 6 of the European Community), the Belgian Program on Interuniversity Poles of Attraction initiated by the Belgian State (interuniversity attraction pole [IUAP] P6/40), the Fonds National de la Recherche Scientifique, Fonds de la Recherche Scientifique Médicale, and Actions de Recherche Concertées de la Communauté Française, Belgium (to M.C. and D.L.E.).

No potential conflicts of interest relevant to this article were reported.

We thank Marie-Anne Neef, Gilbert Vandenbroeck, Maryse Urbain, Jeanine Schoonheydt, Anyishai Musuaya, Stephanie Mertens, Ruth Leeman, and Rachid Makhnas for expert technical assistance and Dr. Agnes Burniat for kind help with the cell proliferation assay.