Exendin-4 Protects β-Cells From Interleukin-1β–Induced Apoptosis by Interfering With the c-Jun NH₂-Terminal Kinase Pathway

Isolated human islets were from the Cell Isolation and Transplantation Center (islets for research distribution program) of the Geneva University Hospital. Islets were cultured in CMRL-1066 supplemented with 10% fetal bovine serum (Mediatech, Herndon, VA) in 5% CO₂ humidified atmosphere at 37°C. The rat insulin-secreting cell line INS-1E was cultured as previously described (28). Rat and mouse islets were isolated by hand-picking after collagenase digestion of pancreas as described previously (29) and were maintained overnight in RPMI-1640 supplemented with 10% FCS, 10 mmol/l HEPES, pH 7.4, 1 mmol/l sodium pyruvate, 100 units/ml penicillin-streptomycin, 50 μmol/l β-mercaptoethanol, and 11 mmol/l glucose.

The cells were scrapped into the PBS and lysed using a passive lysis buffer (Promega). The cell protein extracts (20 μg) were separated on 10% SDS–polyacrylamide gel and electrically blotted to nitrocellulose membrane. The proteins were detected using a buffer containing 0.1% Tween 20 and 5% milk and incubated overnight at 4°C specific with primary antibodies (mouse IB1 antibodies [BD Transduction Laboratories] or mouse β-tubulin antibodies [Sigma]) and were visualized with IRDye 800 (Rockland) as secondary antibodies and quantified in an Odyssey Infrared Imaging System (Li-Cor). For quantification of IB1, we took the sum of the two band intensities and normalized against β-tubulin intensity.

Total RNA was extracted using Aurum Total RNA mini kit (Bio-Rad) according to the manufacturer's protocol. The reverse transcription reaction was performed as previously described (30). Real-time quantitative PCR assays were carried out on the Bio-Rad MyiQ Real-Time PCR Detection System using iQ SyBr Green Supermix (Bio-Rad) as the amplification system with 100 nmol/l primers and 1.5 μl template (RT product) in 20-μl PCR volume and annealing temperature of 59°C. Primers sequences were as follows. Rat and mouse ib1: forward, 5′-ATGTCTTCATGAGTGGCCG-3′, and reverse, 5′-GATTTCAAGGACACAGCTGG-3; human IB1: forward, 5′-ATCAGCCTGGAGTTTGA-3′, and reverse, 5′-AGGTCCATCTGCAGCATCTC-3′; L10E (60S acidic ribosomal protein P0): forward, 5′-ACCTCCTTCTTCCAGGCTTT-3′, and reverse, 5′-ACCTCTTTCTTCCAAGCTTT-3; and rat c-Jun: forward, 5′-AGT CTC AGG AGC GGA TCA AG-3′, and reverse, 5′-CTC TGT CGC AAC CAG TCA AG-3′.

The regulatory regions and promoter analysis of the human MAPK8IP1 gene were subjected to a computer-assisted search (http://genomatix.de). To design target-specific short hairpin RNA (shRNA) directed against IB1 (shIB1), we selected sequences of the type AA(N₁₉) (N, any nucleotide) from the coding sequence of the rat IB1 mRNA. The selected small interfering RNA (siRNA) sequence was also submitted to BLAST search to ensure that it was specific to the target mRNA. Oligonucleotides contained both the 19-nucleotide sense and 19-nucleotide antisense strands separated with a short spacer from the reverse complement of the same sequence and five thymidines as termination signal. The primers used were the following: sense, 5′-GATCCCC C̅A̅G̅C̅G̅A̅C̅T̅G̅G̅A̅T̅T̅G̅A̅C̅C̅A̅G̅ TTCAAGAGA C̅T̅G̅G̅T̅C̅A̅A̅T̅C̅C̅A̅G̅T̅C̅G̅C̅T̅G̅ TTTTTGGAAA-3′, and antisense, 5′-AGCTTTTCCAAAAA C̅A̅G̅C̅G̅A̅C̅T̅G̅G̅A̅T̅T̅G̅A̅C̅C̅A̅G̅ TCTCTTGAA C̅T̅G̅G̅T̅C̅A̅A̅T̅C̅C̅A̅G̅T̅C̅G̅C̅T̅G̅ GGG-3′. The complementary target sequences of IB1 and thymidines are underlined and in bold, respectively. These primers are hybridized and ligated downstream of the H1-RNA promoter by HindIII/BglII sites of the pSUPER vector (31). A 19-nucleotide prevalidated siRNA duplex (siIB1) that corresponds to the shIB1 sequence was designed as recommended and was chemically synthesized by Mycrosynth (Balgach, Switzerland).

The INS-1E cells (10⁵) plated in 24-well dishes were transiently transfected using Effectene transfection reagent (Qiagen). After transfection (24 h), the cells were incubated with fresh medium supplemented or not with 10 μmol/l forskolin and 100 μmol/l isobutylmethylxanthine (IBMX) for 16 h. Then the cells were washed with PBS and lysed using passive lysis buffer (Promega). Luciferase activities were measured with 25-μl protein extracts solution using the Dual-Luciferase reporter assay system (Promega). The siRNA duplex was introduced using the lipofectamine 2000 (Invitrogen) exactly as described by the manufacturer's protocol.

Nuclear protein extracts from the cells were prepared exactly as described previously (32). The electromobility shift assay (EMSA) procedure was conducted exactly as previously reported (30). The sequences of oligonucleotides used are as follows. CreIB1: forward, 5′-TGTGTTACGTTACATT-3′, and reverse, 5′-AATGTAACGTAACACA-3′; CreL2: forward, 5′-GCTCTGAAGTCACTAA-3′, and reverse, TTAGTGACTTCAGAGC-3′; CreL3: forward, 5′-CCACTGACATCCTCTC-3′, and reverse, 5′-GAGAGGATGTCAGTGG-3′; and CreCons: forward, 5′-GGACGTAGTCTGACGTCAGCGGA-3′, and reverse, 5′-CATCAGACTGCAGTCGCCTCCGA-3′. The sequences for specificity protein-1 (Sp-1) and neuron restrictive silencer element (NRSE) were those previously described (30).

The preparation of whole-cell protein extracts and the kinase assays were conducted as previously described (21). Briefly, cell extracts were incubated for 1 h at room temperature with 1 μg glutathione S-transferase (GST)-Jun (amino acids 1–89) and 10 μl glutathione-agarose beads (Sigma-Aldrich, St-Gallen, Switzerland). Phosphorylation of substrate proteins was examined after overnight exposure of polyacrylamide gels to autoradiography; gel quantifications were accomplished by Phosphor-Imager analysis (Molecular Imager FX; Bio-Rad Laboratories, Basel, Switzerland).

Chromatin immunoprecipitation (ChIP) assays were conducted exactly as previously described (33) with the following modifications. The anti-CREB1 and anti-CREB2 antibodies were used for immunoprecipitation. After the reverse cross-linkage, the DNA was precipitated by phenol-chloroform extraction method. Next, we performed the quantitative PCR to measure the enrichment of CreIB1 sequence by amplifying 5 μl DNA using the followings primers: forward, 5′-AGGTCTGCAGGGTTTGTCAT-3′, and reverse, 5′-CTGTGGTCTGCTGGGGTTAT-3′. Amplification of the sytl4 promoter was used as an internal control: the primers used were forward, 5′-TGGGGGAGGGTATGGTAAAT-3′, and reverse, 5′-CCTTCTAGCACTCTGGAAGCA-3′.

Apoptosis was determined by scoring cells displaying pycnotic nuclei visualized with Hoechst 33342 (Invitrogen, Basel, Switzerland) (21).

Data are shown as means ± SE. Statistical significance of differences was calculated either by ANOVA or two-tailed t test for single comparisons.

Exposure of β-cells to IL-1β mediates activation of JNK, which in turn induces phosphorylation of its target transcription factor c-Jun (26). This is observable between 1 to 16 h after incubation with IL-1β (22). To determine whether ex-4 could interfere with the JNK pathway, INS-1E cells were pretreated with ex-4 or the vehicle for 8 h, and thereafter, the cells were incubated with IL-1β for 16 h. Whole protein was then extracted from treated cells and incubated with the c-Jun recombinant. As expected, in vitro kinase assays showed an increase in c-Jun phosphorylation with extracts of cells incubated with IL-1β for 16 h (Fig. 1A). To validate activation of the JNK signaling cascade by IL-1β, expression of the c-Jun gene was then quantified. The transcriptional activity of the promoter of this gene is positively regulated by JNK (21). Real-time PCR analysis showed a statistically significant augmentation by two- and threefold in c-Jun mRNA in INS-1E and rat isolated islet cells, respectively, cultured with IL-1β for 16 h (Fig. 1B and C). Both phosphorylation of c-Jun and its expression induced by IL-1β were efficiently diminished by ex-4 (Fig. 1A–C). The effect of ex-4 started at 10 nmol/l, but the maximum efficiency was obtained with 50–100 nmol/l concentration (Fig. 1A and B). To determine the pathways by which ex-4 inhibits the JNK signaling, INS-1E cells were co-incubated with wortmannin or H89, two pharmacological inhibitors of the phosphatidylinositol 3-kinase (PI 3-kinase)/PKB and PKA pathways, respectively. Quantitative PCR showed that inhibition of the PI 3-kinase/PKB signaling did not counteract the inhibitory effects of Ex-4 on the c-jun expression induced by IL-1β (Fig. 1D). In contrast, treatment of the cells with H89 prevented the inhibitory effects of ex-4 (Fig. 1D), indicating the requirement of the PKA activity for inhibition of the JNK signaling achieved in response to ex-4.

The scaffold protein IB1 has been described as a direct regulator of the JNK signaling. Overexpressing IB1 or its JNK-binding domain in β-cells prevents activation of c-Jun stimulated by IL-1β (25,26,34). We then assessed the possibility that ex-4 might modulate the levels of IB1 to block JNK-mediated activation of c-Jun. Western blotting experiments showed higher levels in IB1 contents in INS-1E cells incubated with 100 nmol/l ex-4 (Fig. 2A). The rise in IB1 protein levels elicited by ex-4 occurred in a time-dependent manner reaching a peak at 12–16 h (Fig. 2A) and was confirmed in isolated pancreatic islets from rat (Fig. 2C). Induction in the IB1 levels was observable with 10 nmol/l ex-4 but was further elevated with 100 nmol/l ex-4 for 12 h (Fig. 2B). In line with previous reports (22,34), we found that activation of the JNK signaling by IL-1β was associated with a reduction in the IB1 contents (Fig. 2D). Western blotting experiments showed that ex-4 efficiently counteracted the diminution of IB1 elicited by IL-1β (Fig. 2D). Taken together, these results indicate that ex-4 prevents activity of JNK by increasing the IB1 levels.

In view of the findings described above, we sought for a mechanistic explanation of the augmentation of IB1 elicited by ex-4. Because the latter regulates the expression of many genes through the cAMP/PKA pathway, we assessed whether the effect of ex-4 could be mimicked with cAMP-raising agents. Exposure of INS-1E cells with forskolin and IBMX led to an increase in IB1 protein levels (Fig. 3A). The increase in IB1 expression was apparent after a 2-h incubation and was maximal after 12–16 h (Fig. 3A). This result was consistent with an effect of IBMX and forskolin on JNK activity. Treatment of the isolated rat islets with IBMX and forskolin prevented induction of c-jun mRNA mediated by IL-1β (Fig. 3B). Incubation of INS-1E cells with H89 abolished the action of ex-4 and IBMX and forskolin on the IB1 expression, confirming that the effect of ex-4 involves the cAMP/PKA pathway (Fig. 4A and B). Induction in the IB1 levels evoked by ex4 was however unaltered in the cells cotreated with the PI 3-kinase/PKB pathway inhibitor wortmannin (Fig. 4B).

We then explored whether ex-4 modulates IB1 content by enhancing mRNA levels. Real-time PCR analysis showed an increase in IB1 mRNA in INS-1E and in isolated pancreatic islets cells from rat and human cultured for 12 h with ex-4 or IBMX and forskolin (Fig. 5A and B). Treatment of INS-1E cells with actinomycin D, a RNA synthesis inhibitor, prevented IBMX- and forskolin-mediated induction of IB1 (Fig. 5C), suggesting an effect of the cAMP pathway on IB1 at the transcriptional level. To test this hypothesis, a 731-bp fragment of the human IB1 proximal promoter linked to a luciferase gene reporter (IB1luc) (35) was transfected in INS-1E cells. A high luciferase activity from the IB1luc construct was detected in INS-1E cells as previously described (35). We found that incubation of the cells with IBMX and forskolin or ex-4 generated a twofold increase in the transcriptional activity of the IB1 promoter (Fig. 5D). To determine whether the transcriptional induction of IB1 requires the PKA activity, INS-1E cells were transfected with IB1luc and were treated with IBMX and forskolin plus or minus the H89 or wortmannin. Addition of the PKA inhibitor significantly diminished stimulation of the IB1 promoter activity mediated by IBMX and forskolin or ex-4 (Fig. 5E). Inspection of the human promoter revealed the presence of a CRE (CreIB1), which is homologous to the consensus sequence and conserved between rat and mouse (Table 1). EMSAs were performed to validate the ability of the CreIB1 sequence to interact with the CREB transcription factors. The CreIB1 sequence was used as the labeled probe and incubated with nuclear extracts of INS-1E cells treated with IBMX and forskolin for 16 h. The binding pattern obtained was efficiently competed using a 100-fold excess of the unlabeled CreIB1 and the consensus oligonucleotide (Fig. 6A). Neither the cold sequences of the IB1 promoter similar to the CreIB1 sequence (CreL2 and CreL3) nor the unrelated sequences, including Sp-1 and NRSE, were able to compete with the binding activity of CreIB1, confirming the high specificity of the probe/protein interaction (Fig. 6A). Preincubation of nuclear protein extracts with antibodies that recognize CREB1 or CREB2 factors resulted in a disruption of the CreIB1 pattern only with the CREB1 but not with the CREB2 or hairy and enhancer of split-1 (HES-1) antibodies (negative control) (Fig. 6B). After a rise in cAMP concentration, CREB transcription factors are rapidly phosphorylated by PKA (36). Phosphorylation of CREB increases the ability of the factors to bind to the target element, thereby converting CREB to a powerful activator (37). In line with the induction of the CREB binding activity, we observed an increase in the intensity of the CreIB1 binding pattern in cells cultured with IBMX and forskolin or ex-4 for 12 h (Fig. 6C). ChIP followed by real-time PCR experiments confirmed the in vivo interaction of CREB1 to the endogenous IB1 promoter in insulin-secreting cells. Enrichment of the PCR product for IB1 promoter was only detected with the use of immunoprecipitated DNA with the CREB1 antibody from the cells exposed to IBMX and forskolin or ex-4 (Fig. 6D). As observed in Fig. 6B, CREB2 did not associate with IB1 promoter (Fig. 6D). If CREB1 regulates the activity of the IB1 promoter, the transcriptional activity of IB1luc construct should be silenced by the inducible cAMP early repressor (ICER). ICER is a member of the CRE modulator family of basic leucine zipper transcription factors that is thought to serve as a dominant-negative repressor of cAMP-dependent gene expression (38). Overexpression of ICER-Iγ (28) reduced the basal and the IBMX and forskolin- or ex-4–stimulated promoter activity of IB1luc (Fig. 6E). In summary, these data reveal that ex-4 upregulates the expression of IB1 at the promoter levels in a mechanism involving the PKA and the CREB1 transcription factor.

IB1 is an important component of pro-survival signaling pathway in β-cells. In this regard, we examined the possibility that the protective effect of ex-4 on β-cells might require IB1. To test this hypothesis, we used a RNAi strategy to selectively suppress the expression of IB1. First, we designed a plasmid expressing a specific and efficient sequence shRNA directed against IB1. INS-1E cells were then cotransfected with the HA-tagged IB1 (HA-IB1) and the plasmid encoding the shIB1. The specificity of the shRNA-IB1 was tested using a vector encoding HA-tagged IB1 resistant to the siRNA (HA-IB1-Si^r). Mutation was introduced within the target sequence by changing one nucleotide in each codon without affecting the primary sequence of the amino acids. The shRNA-IB1 efficiently reduced the expression of the exogenous HA-tagged IB1 (Fig. 7A). However neither the mutated IB1 construct, nor the green fluorescent protein (GFP) (cotransiently expressed) had their expression modified in the presence of the shRNA, confirming the specificity of the shRNA-IB1 (Fig. 7A). The effect of the siRNA on the endogenous levels of IB1 was assessed with a duplex siRNA sequence (siIB1) corresponding to the shIB1. Western blotting analysis of extracts from transfected INS-1E cells revealed the efficiency of the siIB1 to diminish the IB1 contents (Fig. 7B). Reduction in the IB1 levels led to an increase in basal c-Jun mRNA levels and potentiated the effects of IL-1β (Fig. 7C). However, ex-4 was unable to abolish induction of c-Jun achieved by IL-1β (Fig. 7C). To establish the involvement of IB1 in the anti-apoptotic action of ex-4, we therefore measured the viability of INS-1E cells in which IB1 was suppressed. As expected, incubation of the cells with 10 ng/ml IL-1β for 48 h led to a twofold increase in the rate of apoptosis, whereas the viability of the cells co-incubated with ex-4 was unchanged compared with control (Fig. 7D). However, reduction in IB1 contents potentiated IL-1β–induced apoptosis and abolished the anti-apoptotic action of ex-4, confirming the requirement of IB1 in the mechanism of action of ex-4 (Fig. 7D).

The long-acting GLP-1R agonist ex-4 represents a new available class of drugs for treatment of type 2 diabetes. This disease is thought to occur when pancreatic β-cells fail to release sufficient insulin levels to compensate for insulin resistance in target tissues. Morphometric studies on postmortem pancreases of people with type 2 diabetes provide convincing evidence that β-cell mass is reduced, thus explaining in part the β-cell failure (39). Genetics and changes in environmental factors contribute to the β-cell destruction over time. These changes include an increase in concentration of IL-1β, a pro-inflammatory cytokine implicated in the pathogenesis of type 1 diabetes (14,15,40). IL-1β has detrimental effects in β-cells mainly through activation of the JNK signaling, which lead to increased phosphorylation and expression of its target c-Jun (26,41). In line with this statement, we showed a rise in JNK activity in the cells exposed to IL-1β, which was concomitant with increased β-cells apoptosis. In vitro and in vivo studies show that one of the major beneficial properties of ex-4 is the enhancement of β-cell mass at least through inhibition of apoptosis (8–13). In support of this function, we confirmed that pretreatment of the cells with ex-4 for several hours completely prevented apoptosis induced by IL-1β. Herein, we report that this effect of ex-4 is associated with diminution in phosphorylation and expression of c-Jun in isolated pancreatic islets and INS-1E cells challenged with IL-1β for 16 h. Inhibition of either the PKA or PI 3-kinase/PKB pathway, the two major pathways through which ex-4 triggers its anti-apoptotic effects, showed that only PKA is responsible for ex-4–mediated blocking of the JNK signaling. Incubation of the cells with IBMX and forskolin, two strong activators of the cAMP/PKA pathway, mimicked the effects of ex-4 on the c-Jun expression. Inhibition of PKB did not modulate c-Jun expression, whereas the use of H89, a pharmacological inhibitor of the PKA, also efficiently antagonized the effects of IBMX and forskolin on the JNK activity. We and others have previously shown that selective inhibition of the JNK pathway using the TAT-linked peptide derived from the JBD of IB1 (JNKi) reduces >50% of the β-cells apoptosis elicited by IL-1β, indicating that activation of the JNK pathway is one of the causes of IL-1β–induced apoptosis (41). In this report, we showed that the full protection mediated by ex-4 against apoptosis provoked by IL-1β was associated with an inhibition of the JNK pathway. Treatment of β-cells with a cocktail of cytokines (IL-1β, tumor necrosis factor-α, and interferon-γ) also leads to apoptosis through activation of JNK (11,42). However, in that case, ex-4 only confers a partial protection against apoptosis (11). This situation therefore assumes that inhibition of the JNK pathway is not sufficient for ex-4 to completely achieve its protective effect. A previous study reports that β-cells challenged with IL-1β alone had unaltered PKB activation (43) whereas phosphorylation of PKB is markedly reduced in response to cytokines. In the latter condition, ex-4 fails to recover the loss of PKB activation induced by cytokines (11). Thus it can be assumed that both inhibition of the PKA-dependent JNK signaling and adequate activation of PKB are needed for ex-4 to exert a full cytoprotective effect of β-cells against the cocktail of cytokines.

IB1, a scaffold protein that tethers components of the JNK pathway, is thought to facilitate the activation of JNK in acute response to various stimuli (44). Data obtained from different laboratories establish an important function for IB1 in the control of β-cell survival in response to stressful stimuli. Decreased IB1 expression caused by sustained exposure of β-cells to IL-1β or cocktail of cytokines, including IL-1β, tumor necrosis factor-α, and interferon-γ, is systematically accompanied by an increase in JNK-mediated phosphorylation of c-Jun (22,25,34). Consistent with this, we found that reduction in IB1 expression in the cells exposed to IL-1β was associated with more phosphorylated c-Jun and apoptosis. In addition, the latter phenotype was exacerbated in the cells where the resting IB1 levels were silenced with the use of the RNAi approach. Both apoptosis and the JNK activity can be efficiently overcome by compensating the reduced levels of IB1 in overexpressing IB1 or overloading β-cells with a JNKi (25,26,41). In this report, we show that the relative recovering of the IB1 levels after treatment of the cells with ex-4 for 12–16 h is also sufficient to reduce phosphorylation/expression of c-Jun and to abolish apoptosis elicited in response to IL-1β.

The mechanism leading to restored IB1 levels was deciphered. In fact, incubation of the cells with the PKA inhibitor H89 prevented the stimulation of the IB1 expression levels achieved by ex-4. Our data showed that inhibition of PKB activation does not interfere with the capacity of ex-4 to stimulate IB1 expression, indicating that PKB is not required for the effects of ex-4 on the JNK signaling. Using a combination of approaches, we demonstrated that ex-4 compensates IL-1β–mediated reduction of IB1 by stimulating the promoter activity of IB1 through the CREB1 transcription factor. The peak for IB1 induction was reached at 16 h. After that time, IB1 expression returned to the level comparable with that observed in untreated cells (data not shown). This mechanism could be attributed to the transcription factor ICER. After a sustained stimulation of the cAMP/PKA pathway, the transcriptional activation of the CREB factors is followed by the activity of ICER, a powerful repressor that negatively regulates expression of genes containing a CRE element (21). In agreement with a regulation of IB1 by ICER, we found that overexpression of ICER-Iγ efficiently silenced the promoter activity of IB1. Suppression of IB1 expression by RNAi impaired the anti-apoptotic effect of ex-4. These data provide evidence that IB1 plays a key role in mediating the effect of ex-4 against apoptosis in a JNK-dependent manner. According to these data, defects in the GLP-1 signaling are expected to impair induction of IB1. Interestingly mice with targeted disruption of the GLP-1R have reduction in pancreatic β-cells mass caused by increased stress-induced apoptosis (12). Low plasma levels in GLP-1 have been observed in human type 2 diabetes (45). Therefore impaired induction of IB1 could partly account for elevated β-cell apoptosis in diabetes.

Activation of the JNK signaling is partly responsible for the loss of functional pancreatic islet β-cell mass after grafting. This loss of viability is caused by the pro-inflammatory cytokines, including IL-1β, and the islet isolation procedure itself with a major loss occurring shortly after the final purification steps (41,46–48). A recent report shows that improvement of β-cell viability is associated with a decreased JNK activity both in cultured rat and human isolated islets (46). Thus, direct inhibition of the JNK signaling cascade has been proposed as a strategy to improve the cell viability for transplantation purpose (46). To this end, we believe that ex-4 or long-acting analogs mimetics may be used as efficient available therapeutic drugs. Like IL-1β, prolonged exposure of β-cells to high glucose and nonesterified fatty acid concentration have devastating effects on β-cells and thereby participate in progression and development of type 2 diabetes. The mechanism responsible for β-cell failure include increased apoptosis that is associated with a rise in JNK activity (21,49,50). Ex-4 can also protect β-cells against apoptosis provoked by elevated concentrations of glucose and lipids (8). Thus it is possible that cytoprotective action of ex-4 against chronic hyperglycemia and hyperlipidemia involves inhibition of the JNK signaling.

This study highlights the potent inhibitory effect of ex-4–cAMP–PKA system on the JNK activity as a major mechanism for preventing β-cell apoptosis induced by IL-1β. This mechanism largely contributes to the efficiency of ex-4 and its mimetics analogs in the long-time preservation of β-cells against apoptosis in the treatment of type 2 diabetes.

C.W. has received Swiss National Foundation Grant 3100A0-107819. R.R. has received Swiss National Foundation Grant 3200B0-101746. G.W. has received Swiss National Foundation Grant 310000-109281/1. A.A. has received Swiss National Foundation Grant 3100A0-105425. This work was also supported by the Placide Nicod and Octav Botnar Foundations.

Human islets of Langerhans were provided by the Cell Isolation and Transplantation Center at the University of Geneva School of Medicine. We thank the European Consortium for Islet Transplantation islets for research distribution program sponsored by the Juvenile Diabetes Research Foundation. We are grateful to Eric Bernardi for editorial assistance.