Stress-Induced MicroRNA-708 Impairs β-Cell Function and Growth

Pancreatic β-cells play a central role in the control of glucose homeostasis. In situations of increased insulin demand, β-cells undergo a compensatory response to maintain normoglycemia. This process relies on an increase in both the secretory capacity and the mass of β-cells. Understanding how β-cell mass is regulated and the mechanisms by which physiological mitogens drive the expansion of β-cells is a major focus in diabetes research. Besides its role as an insulin secretagogue, glucose promotes long-term effects in β-cells. β-Cells need to be stimulated by glucose to maintain their secretory capacity (1,2). Glucose is also known to promote the survival of β-cells by suppressing a constitutive apoptotic program (3). For this reason, a low glucose concentration is a highly stressful setting for pancreatic β-cells. Glucose is also a potent mitogen in rodent and human β-cells (4–6), and it has been suggested to be one of the systemic signals that allow β-cells to sense insulin demands (7). How glucose stimulation drives β-cell proliferation is still not fully understood and is under debate. Several mechanisms have been proposed to underlie the mitogenic action of glucose on β-cells, involving signals derived from the glycolytic flux and components of the insulin signaling pathway (7–9), which may be induced by the autocrine action of insulin and islet amyloid polypeptide (8,10,11).

The β-cell transcriptome is highly responsive to glucose concentration oscillations (12–15). In mouse pancreatic islets, stimulatory glucose concentrations allow the activation of genes related to β-cell function and proliferation. Concomitantly, stimulatory glucose levels repress the expression of many genes associated with stress responses. For instance, we and others have previously demonstrated (12,14,16) by global gene expression profiling that glucose represses the expression of Chop and many other genes that are coregulated or under the control of this transcription factor. Chop is induced by an array of stress settings, including endoplasmic reticulum (ER) stress. Upon ER stress, cells activate a succession of signal transduction cascades known as the unfolded protein response. Pancreatic β-cells are particularly sensitive to alterations of ER homeostasis, and prolonged or excessive unfolded protein response activation can lead to β-cell dysfunction (17).

MicroRNAs (miRNAs) are small, noncoding, and single-stranded RNA molecules that are ∼22 nucleotides in length and act mainly as post-transcriptional regulators of gene expression by binding the 3′ untranslated region (UTR) of their target mRNAs, leading to either a translational repression or mRNA degradation (18). Although some studies clearly established a link between miRNA expression dynamics and pancreatic β-cell maturation and physiology (19–24), little is known about the dynamics of miRNA expression in response to external cues such as glucose and the role of miRNAs in mediating the actions of glucose on pancreatic β-cell function and mass.

In this study, we aimed to identify the miRNAs that are differentially expressed in pancreatic islets exposed to different glucose concentrations. Through an miRNA panel screening, we identified miR-708 as the most upregulated miRNA in pancreatic islets under low glucose conditions. We show that amelioration of ER stress reduces miR-708 induction. Moreover, increased expression of miR-708 in pancreatic β-cells reduced the insulin secretory capacity and the proliferation and survival rates of β-cells, recapitulating the effects of low-glucose conditions. Thus, we have uncovered a novel mechanism of gene regulation by glucose in pancreatic β-cells linking ER stress with β-cell dysfunction and reduced proliferative capacity.

Mouse pancreatic islets were isolated from 8- to 10-week-old C57BL/6J male mice after perfusion with collagenase as described previously (10). Islets were allowed to recover for 24 h at 37°C and 5% CO₂ in RPMI 1640 medium (11.1 mmol/L glucose [G11]) supplemented with 10% FBS (v/v), 2 mmol/L glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. Islets were then generally cultured for 2.5 days in RPMI at different glucose concentrations (2.8 mmol/L [G3], 5.5 mmol/L [G5], G11, and 25.5 mmol/L [G25]). B6.V-Lep^ob/ob (ob/ob) male mice and heterozygous male littermates (ob/+) were purchased from Janvier Laboratories. Experimental procedures were approved by the Animal Ethics Committee of the University of Barcelona according to the Principles of Laboratory Animal Care.

Dissociated islets cells (DICs) obtained by treatment with 0.05% trypsin-EDTA for 4–5 min were seeded in 384-well plates (15,000 cells/well) with RPMI 1640 medium. The mouse pancreatic β-cell line MIN6 was maintained at 37°C and 5% CO₂ in DMEM (25 mmol/L glucose) supplemented with 10% FBS (v/v), 2 mmol/L glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, and 50 µmol/L β-mercaptoethanol. The rat pancreatic β-cell line INS1E was maintained in RPMI 1640 medium (G11) supplemented with 10% FBS, 2 mmol/L glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, and 50 µmol/L β-mercaptoethanol at 37°C and 5% CO₂.

Knockdown of Chop in INS1E cells was performed using 20 µmol/L small interfering RNA (siRNA) for Chop (Invitrogen) or 20 µmol/L its respective scrambled siRNA (Applied Biosystems) using Metafecten Pro (Biontex Laboratories GmbH) as a transfection reagent, according to the manufacturer instructions. To study miR-708 inhibition in islets and DICs, an LNA miRNA Power Inhibitor of miR-708 or a scrambled control miRNA (500 nmol/L; Exiqon) were added directly to the culture medium.

Adenovirus encoding GFP (AdGFP), GFP and miRNA-708 (AdmiR-708), and the open reading frame of human NNAT (transcript variant 1, AdNnat) under the cytomegalovirus promoter were obtained from Amsbio (Oxfordshire, U.K.). Recombinant adenoviral particles were amplified by transducing cells of the human embryonic kidney 293 cell line. Adenoviral titers were determined by quantifying the amount of virus required to produce a cytopathic effect in the inoculated cells. Freshly isolated pancreatic islets or DICs were transduced at a multiplicity of infection (MOI) of 5–20, as indicated.

Total RNA was extracted with the miRNeasy kit (Qiagen), and cDNA synthesis was conducted using the Universal cDNA Synthesis Kit (Exiqon). Mouse and Rat Panel I V3 384-well PCR plates containing LNA primers (Exiqon) were used to perform a screening of miRNA expression using biological triplicates. miRCURY LNA Universal RT miRNA PCR Kit and ExiLENT SYBR Green master mix (Exiqon) were used to perform real-time quantitative PCR (qPCR) in a 7900HT Fast Real-Time PCR system (Applied Biosystems). Primers for individual miRNA PCR assays were also obtained from Exiqon. Let7i, RNU1A1, and RNU5G were used to normalize individual miRNA PCR assays because they were identified as the most stably expressed small RNAs present in the panel using the GenEx software (version 6). Individual qPCR assays confirmed that their expression was unchanged in islets exposed to different glucose concentrations. RNU1A1 was used to normalize the miRNA panels. miRNAs with two or more Ct values >37 were excluded from the analysis. A fold change was calculated for each sample with respect to the mean expression in islets cultured at G11. Statistical significance was determined by multiple t tests corrected for multiple comparisons using the Holm-Sidak method.

RNA was extracted with the miRNeasy kit (Qiagen), and cDNA was synthesized using the SuperScript Reverse Transcriptase kit (Life Technologies). Gene expression was examined by qPCR using SYBR Green (Life Technologies) in a 7900HT Fast Real-Time PCR system (Applied Biosystems). Primer sequences used for gene expression analysis are listed in Supplementary Table 1. Tbp1 and Hprt1 were used to normalize mRNA expression of genes of interest.

MIN6 cells were collected in lysis buffer (5 mmol/L EDTA, 50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1% Triton X-100, 10 mmol/L Na₃PO₄, 10 mmol/L NaF, and 10% protease inhibitor mixture; all from Sigma-Aldrich). Protein lysates (5 µg) were separated by 14% SDS-PAGE and transferred to polyvinylidene fluoride membranes (PerkinElmer Life Sciences). Membranes were blocked with 5% nonfat dry milk and were incubated overnight at 4°C with primary rabbit anti–β-actin (1:1,000; Sigma-Aldrich) and rabbit anti-Neuronatin (Nnat) (1:500; Abcam). Horseradish peroxidase–conjugated antibodies (GE Healthcare Life Sciences) were used as secondary antibodies.

Islets were preincubated with Krebs-Ringer bicarbonate HEPES buffer solution (115 mmol/L NaCl, 24 mmol/L NaHCO₃, 5 mmol/L KCl, 1 mmol/L MgCl₂·6H₂O, 1 mmol/L CaCl₂·2H₂O, 20 mmol/L HEPES, and 0.5% BSA, pH 7.4) containing 2.8 mmol/L glucose for 90 min at 37°C. Eight islets per assay were then incubated with either 2.8 mmol/L glucose or 16.7 mmol/L glucose. After 1 h, supernatants were collected and cellular insulin contents were recovered in acid-ethanol solution. Insulin concentration was determined by Insulin Mouse ELISA (Crystal Chem).

DICs were fixed for 15 min with 10% formalin and then were permeabilized with 0.2% Triton X-100 and blocked with 1% BSA. Cells were incubated overnight at 4°C with primary antibodies against guinea pig anti-insulin (1:500; Dako) and mouse anti-Ki67 (1:50; BD Biosciences, San Jose, CA). Then, secondary antibodies were added for 1 h at room temperature (Alexa Fluor conjugate anti-mouse 1:500; GE Healthcare; or anti–guinea pig-Cy2 1:500; Jackson ImmunoResearch). Hoechst 33258 stain (Sigma-Aldrich) was used to stain nuclei. Fluorescence images were obtained using Leica LAS Image Analysis software. ImageJ version 1.49 software (National Institutes of Health) was used to determine the area that was positive for insulin staining by setting the scale and defining a threshold to remove nonspecific signal, and Ki67⁺/insulin⁺ cells were counted. The total number of β-cells per well (∼25,000 cells) was estimated by counting the number of insulin-positive cells in a fraction of the insulin-positive area. Cell death was assessed in fixed islets by TUNEL assay using the DeadEnd Fluorometric TUNEL system (Promega, Madison, WI). TUNEL⁺/insulin⁺ cells from at least 20 islets/condition (sections taken every 30 μm) were counted (analyzing ∼200 β-cells/islet).

Data are expressed as the mean ± SEM, and statistical significance was determined by two-tailed Student t test and one-way or two-way ANOVA with post hoc Tukey test as appropriate. Results were considered significant at P < 0.05.

The islet transcriptome is highly dependent on extracellular glucose concentrations. Here we aimed to assess how glucose regulates the expression of miRNAs in mouse pancreatic islets. We first compared the miRNA expression profiles of mouse pancreatic islets cultured at low glucose levels (G3) and at a stimulatory glucose concentration (G11) for 2.5 days using qPCR panels interrogating almost 400 murine miRNAs. As rodent islets are usually maintained ex vivo at stimulatory glucose concentrations, which result in better survival outcomes (3,14), the expression levels at G11 were taken as reference values. As shown in Table 1, seven miRNAs were significantly upregulated at G3 compared with G11, four of them with a fold change >2, whereas five miRNAs were downregulated.

The three most upregulated miRNAs at G3 were miR-708-5p, miR-451a, and miR-339-5p, and their glucose-dependent expression changes were validated by qPCR (Fig. 1A–C). miR-708 expression was inversely correlated with glucose concentrations. Remarkably, moderate low levels of glucose (G5) already induced a pronounced increase in its expression (Fig. 1A). These results show that increasing glucose concentrations gradually represses miR-708 expression and that its expression is regulated in the transition between G5 and G11. This pattern of regulation resembles that for coding genes related to stress responses such as Chop (2,12,14).

miR-708 is a highly evolutionarily conserved miRNA, which resides in the first intron of the Chop-regulated gene Odz4. Since intronic miRNAs are generally coexpressed with their protein encoding host genes (25,26) and miR-708 was previously shown to be transcriptionally activated by Chop upon ER stress induction (27), we next evaluated the expression pattern of Odz4 and Chop in islets cultured at four different glucose concentrations (G3, G5, G11, and G25). Interestingly, miR-708, Odz4, and Chop followed the same pattern of regulation by glucose, being highly expressed at G3 and G5 compared with G11, suggesting a common mechanism of regulation of the three gene entities in β-cells (Fig. 1A, D, and E). Of note, Chop, Odz4, and miR-708 were not induced at G25, illustrating that, at least for short culture periods (2.5 days), only low-glucose conditions are stressful for β-cells.

In order to gain insights into the transcriptional hierarchy of Chop, Odz4, and miR-708, we next investigated the time course of their expression upon a low-glucose challenge. Chop induction reached a maximum at 6 h (Fig. 1F), whereas Odz4 and miR-708 induction was much more delayed because, although expression started to increase after 24 h, the highest expression was observed at 48 h (Fig. 1G and H). This is consistent with Chop being a transcriptional activator of both miR-708 and Odz4 in pancreatic islets under stress.

The premise that miR-708/Odz4 are transcriptional targets of Chop prompted us to investigate whether miR-708 expression is linked to ER stress in β-cells. For this, we treated pancreatic islets with the chemical ER stress inducer thapsigargin (Thp; 1 µmol/L) for 24 h. Thp treatment not only induced the expression of Chop, as expected, but also fostered a strong induction of miR-708 and Odz4 (Fig. 2A–C). In addition, we used siRNA siChop to knock down the expression of Chop in INS1E β-cells treated with Thp for 24 h. As expected, Chop induction by Thp was decreased in siChop-treated cells (Fig. 2D). Remarkably, Chop silencing also reduced miR-708 expression in Thp-treated cells, confirming that ER-induced expression of this miRNA is mediated by Chop (Fig. 2E).

We also evaluated the effects of high glucose levels (G25) during a longer period ex vivo (7 days). Chop, Odz4, and miR-708 were slightly upregulated (Fig. 2F–H), although to a much lower extent than at low glucose levels in 2.5-day cultures. These results show, in another physiological setting, that the expression of the ER stress marker Chop and miR-708 show a quantitative correlation.

To explore whether the induction of miR-708 by low glucose levels was mediated by ER stress, islets were treated with the chemical ER stress reliever 4-phenylbutyrate (PBA; 2.5 mmol/L). PBA treatment dramatically reduced the expression of Chop in islets cultured at low glucose concentrations (Fig. 2I), uncovering a role of ER stress in this response. Remarkably, PBA treatment also decreased the induction of miR-708 and Odz4 at G3 and G5 (Fig. 2J and K). Taken together, these data indicate that ER stress activates the Chop-Odz4/miR-708 axis in β-cells when exposed to nonstimulatory glucose levels.

In order to identify targets of miR-708, we used miRNA target prediction algorithms (TargetScan and TargetRank) based on the presence of binding sites for miRNAs in the 3′UTR of mRNAs. Thirty-three genes were identified with both algorithms and were considered for further analysis. Because miR-708 was upregulated at G3 and G5, we examined whether any of these candidate genes was present in a set of 190 genes downregulated more than twofold in islets cultured at low glucose levels that we had previously identified in a global gene expression analysis (12). This integrative analysis identified Nnat as the only potential miR-708 target that was at the same time downregulated at low glucose levels. Moreover, previous studies in prostate and metastatic breast cancer cells demonstrated that Nnat is a target of miR-708 (28,29). Indeed, the 3′UTR sequence of Nnat contains one highly evolutionary conserved binding site with perfect matching in both the seed and flanking sequences of miR-708 (Supplementary Fig. 1). Of interest, Nnat is expressed in the ER membrane and has been shown to regulate intracellular Ca²⁺ levels and in insulin secretion in pancreatic β-cells (30,31).

To have a direct functional evidence of Nnat being a target of miR-708, we overexpressed miR-708 using an adenovirus in islets and DICs cultured at G11. This forced expression of miR-708 resulted in 10.2 ± 1.2-fold and 14.4 ± 1.97-fold increases in the levels of this miRNA in islets and DICs, respectively, which is similar to the induction observed in cultures at G3 (20 MOI AdmiR-708) (Supplementary Fig. 2). Remarkably, Nnat expression was reduced by 25 ± 2.5% in islets and 34.3 ± 8.7% in DICs after miR-708 overexpression (Fig. 3A and B). miR-708 overexpression also induced a potent reduction of Nnat protein levels in MIN6 cells (50.2 ± 9.3% of reduction) (Fig. 3C), further supporting that Nnat is a target of miR-708.

In line with the previous findings, Nnat expression was higher in islets under stimulatory glucose concentrations and was also induced at all glucose concentrations when ER stress was inhibited by PBA treatment (Fig. 3D). Moreover, Nnat expression decreased in islets treated with Thp (Fig. 3E), and siRNA-mediated knockdown of Chop led to a recovery of Nnat levels in β-cells exposed to Thp (Fig. 3F). Altogether, these results show that Nnat expression is inversely correlated with those of miR-708 and Chop, which is consistent with Nnat being a direct target of miR-708.

To explore whether miR-708 levels were increased in pathophysiological settings in vivo, we determined the expression of miR-708 in islets of animal models such as ob/ob mice, which exhibit ER stress (32). Strikingly, the expression of both miR-708 and Odz4 were highly induced in islets of these mice, whereas the β-cell–enriched miR-375 was unaltered (Fig. 4A–C). Consistent with these findings, Nnat levels were largely reduced in ob/ob mice, whereas the β-cell–specific gene Pdx1 was unchanged (Fig. 4D and E). Despite the increase of miR-708 in ob/ob islets, we did not detect increased levels of Chop and other genes related to ER stress, with the exception of Atf3 (Fig. 4F–I), suggesting that other mechanisms may induce miR-708 expression in this model. This increase in miR-708 levels and decrease in Nnat expression may be relevant for the reduced secretory capacity of β-cells observed in ob/ob mice over time (32,33).

As Nnat plays a role in insulin secretion (30,31), we hypothesized that miR-708 may impair insulin secretion. To test this, we performed glucose-stimulated insulin secretion (GSIS) assays in mouse pancreatic islets transduced with AdmiR-708. The capacity to secrete insulin was reduced by 49.4% in miR-708–overexpressing pancreatic islets in response to a 16.7 mmol/L glucose concentration with respect to islets transduced with a control adenovirus expressing GFP (Fig. 5A). Insulin content did not change (Fig. 5B), indicating that miR-708 affects the secretory mechanisms in β-cells, which is consistent with a reduction of Nnat levels in miR-708–treated cells.

We next investigated whether inhibition of endogenous miR-708 may recover the insulin secretory capacity in β-cells exposed to low glucose levels. When islets were cultured at G5 for 2.5 days, GSIS was remarkably blunted (Fig. 5C) and the total insulin content was slightly reduced (Fig. 5D). Interestingly, when islets cultured at G5 were treated with the miR-708 inhibitor, insulin secretion was notably recovered (Fig. 5C), which was consistent with an increase of Nnat levels (Supplementary Fig. 3A). To further elucidate the mechanistic relationship between miR-708 and Nnat, islets were cotransduced with AdmiR-708 and an adenovirus encoding the open reading frame of human NNAT (5 MOI AdNnat) (Supplementary Fig. 3B–D). NNAT overexpression in control islets induced a slight reduction of GSIS, indicating that excessive levels may also be detrimental for β-cells. But importantly, NNAT overexpression rescued insulin secretion in islets overexpressing miR-708 (Fig. 5E). These results highlight that miR-708 induction and Nnat repression play a key role in the loss of the secretory capacity of β-cells when cultured at low glucose concentrations.

Since miR-708 has been found to be downregulated in several tumors and metastases, illustrating its role as a growth suppressor (28,29,34–37), we next explored whether miR-708 may also inhibit β-cell growth. Cultures of DICs treated with AdmiR-708 presented a 52.0 ± 4.8% reduction in the number of β-cells (Fig. 6A and B), phenocopying the effects observed either at G3 (37.1 ± 2.1% reduction) or in the presence of Thp (32.3 ± 4.7% reduction) (Fig. 6C), two conditions that highly induce miR-708 expression.

To elucidate the biological process underlying this reduction of β-cell growth, we first analyzed the impact of miR-708 in β-cell proliferation. miR-708 overexpression decreased the number of Ki67-positive β-cells from 1.53 ± 0.06% to 0.77 ± 0.07% in DICs cultured at G11 (Fig. 6D). This reduction was similar to that observed in DICs cultured at G3 (0.52 ± 0.17%). In contrast to the effects observed in insulin secretion, miR-708 inhibition was not able to recover β-cell growth when DICs were cultured at G5 (Fig. 6E), and NNAT overexpression was also not able to recover β-cell growth in DICs transduced with AdmiR-708 (Fig. 6F).

Finally, we assessed whether miR-708 also induces cell apoptosis, as observed in other cell types (28,34,36). TUNEL-positive cells increased from 0.82 ± 0.43% in GFP-transduced islets to 8.21 ± 1.08% in transduced islets with AdmiR-708, highlighting a potent induction of β-cell apoptosis by miR-708 (Fig. 6G and H).

These results indicate that increased expression of miR-708 decreases β-cell growth by repressing β-cell proliferation and promoting apoptosis, recapitulating the effects triggered by low glucose levels in pancreatic islets. However, inhibition of miR-708 is not sufficient to restore β-cell growth when β-cells are cultured at low glucose concentrations.

β-Cell mass and function are regulated in accordance with the insulin demand of the organisms. The islet transcriptome is deeply modulated by glucose (12,14). Glucose serves as a main survival factor for pancreatic β-cells (3) and β-cell function is correlated with glucose regulation of many β-cell genes (1,2). It is in this regard that glucose activates and represses the expression of genes involved in a large variety of biological functions (12). Importantly, glucose is also a potent mitogen in rodent and human β-cells (4–7). In the current study, we identified miR-708 as the most upregulated miRNA in pancreatic islets cultured at low glucose concentrations, a setting known to trigger a stress response. Interestingly, the expression pattern of this glucose-regulated miRNA paralleled that of its host gene Odz4 and their transcriptional activator Chop. Indeed, this pattern of gene regulation has been demonstrated to be a general feature of a set of genes related to stress responses (12,14,16,38,39). These genes are upregulated in mouse pancreatic islets cultured at low glucose concentrations (G3 and G5) and repressed when islets are cultured at G11. For this reason, mouse islets cultured ex vivo at a constant glucose concentration require stimulatory glucose levels (i.e., 10–11 mmol/L) to maintain β-cell function and survival as well as the capacity of β-cells to proliferate. Remarkably, the highest relative expression changes in miR-708 and Chop occur between G5 and G11. This illustrates that glucose-dependent changes in the expression of these genes occur within a range of usual glucose concentrations and may be instrumental to the adaptation of β-cell function and mass to insulin demands.

miRNAs are dynamic and fine-tuning regulators of gene expression and are responsive to different stress stimuli, including ER stress (40). The fact that Thp treatment also potentiated the expression of miR-708 in pancreatic islets suggested that ER stress may contribute to the induction of Odz4/miR-708 in islets under low-glucose conditions. Indeed, miR-708 was previously shown to be induced by ER stress in mouse embryonic fibroblasts in a Chop-dependent manner (27). Interestingly, miR-708 induction in islets cultured at low glucose levels was abrogated by relieving ER stress upon PBA treatment. Indeed, PBA treatment also reduced the expression of Chop, illustrating that the induction of stress genes in islets cultured at nonstimulatory glucose levels is mediated, at least in part, by ER stress. The repression of miR-708 and Chop by glucose is in line with the amelioration of ER stress-induced apoptosis in pancreatic β-cells when the glycolytic flux is potentiated by glucokinase activation (41). Thus, our results support a model in which glucose represses the expression of Chop, and in turn its targets Odz4/miR-708, by reducing ER stress.

miRNA target prediction algorithms identified Nnat as a potential target of miR-708. We focused on this gene because Nnat was the only potential target of miR-708 among a set of glucose-induced genes identified in a previous transcriptomics analysis (12). Moreover, previous studies (28,29) in prostate and metastatic breast cancer cells demonstrated that Nnat is a target of miR-708. Nnat protein is present at the ER membrane and triggers Ca²⁺ release from the ER. Consistently, the inhibition of Nnat by miR-708 has been associated with decreased intracellular calcium levels (28,29). Here, we found that Nnat expression was reduced after miR-708 overexpression in pancreatic islets. Accordingly, Nnat expression was inversely correlated with miR-708 in islets exposed to different glucose concentrations or in islets treated with ER stress activators and inhibitors. Nnat was discovered as a Neurod1 target by exploring the transcriptional changes in Neurod1 knockout mice islets (30) and plays a key role in the secretory capacity of β-cells (30,31). Consistent with this role of Nnat in insulin secretion, miR-708 overexpression reduced the secretory capacity of pancreatic islets. Importantly, the inhibition of miR-708 partially rescued GSIS in islets cultured at G5 and importantly NNAT overexpression restored the inhibitory effect of miR-708. These results uncover miR-708 induction and its suppressive action on Nnat as key events underlying the reduction of the secretory capacity of β-cells exposed to stress settings, such as low-glucose conditions.

To date, the role of miR-708 has been studied mainly as a therapeutic target against metastatic cancers, as it has been shown to suppress cell migration and proliferation and to induce apoptosis (28,29,34–37). Despite all of this body of evidence as a tumor suppressor, there are so far no studies investigating the role of miR-708 in pancreatic β-cells. We have now found that miR-708 overexpression significantly reduces both β-cell proliferation and survival. Thus, the reduction of miR-708 levels may be a novel mechanism underlying the proliferative and survival action of glucose on β-cells. However, the inhibition of miR-708 was not sufficient to restore β-cell growth when islet cells were cultured at G5, suggesting that other events occurring under these conditions maintain the inhibition of β-cell growth. On the other hand, although Nnat has been shown to promote cell proliferation in tumor cell lines (29,42), NNAT overexpression did not restore β-cell proliferation in cells overexpressing miR-708. This suggests that the inhibition of other targets not identified so far maintains the suppressive effects of miR-708 on β-cell growth after restoring Nnat levels.

Several miRNAs that are expressed at lower levels in β-cells under normal physiological conditions have been reported to negatively regulate insulin synthesis and GSIS and may thus contribute to β-cell dysfunction under stressful metabolic conditions (43). In this study, we have identified miR-708 as the most differentially expressed miRNA in islets cultured at different glucose concentrations. This result suggests that the regulation of miR-708 expression is highly dynamic in islets. Importantly, we found that miR-708 was also increased in islets from 14-week-old ob/ob mice, demonstrating that the induction of this miRNA is also present in stress settings in vivo and therefore could be central for β-cell dysfunction. In line with our results, a close inspection of RNA-sequencing data from another study (19) revealed that miR-708 was the most upregulated miRNA in ob/ob mice islets. miR-708 upregulation in ob/ob mice islets may be linked to the defective secretory capacity of β-cells observed in this model over time (32,33) and could be an early manifestation of functional defects that lead to GSIS failure in the progression to diabetes. In contrast, miR-708 was shown to be downregulated in islets from Goto-Kakizaki rats, a model of nonobese spontaneous type 2 diabetes (44). These reduced miR-708 levels may be explained by the long-term exposure of pancreatic islets to hyperglycemia, which would be consistent with the reduced expression of miR-708 in islets cultured at high glucose levels. Collectively, these results show that miR-708 expression is perturbed in pancreatic islets from diabetic and obese mice, although to fully understand whether changes in the expression of this miRNA are the cause or the consequence of β-cell dysfunction in these models still requires further investigation.

In summary, we have identified miR-708 as a glucose-repressed miRNA in pancreatic islets. Because the overexpression of miR-708 leads to decreased insulin secretion and β-cell growth, the repression of this miRNA may underlie the beneficial actions of glucose on β-cell function, proliferation, and viability. Our results also highlight that stimulatory glucose levels repress the induction of miR-708 by relieving ER stress observed under low-glucose conditions. Because miR-708 is also induced by general ER stress inducers and in islets from ob/ob mice, we can anticipate that miR-708 may act as a mediator of ER stress responses in pancreatic β-cells and therefore is a potential target with which to design strategies to recover β-cell function and mass under stress conditions.

Funding. This work was funded by grants from the Spanish Ministerio de Ciencia e Innovación (grant BFU2010-17639 to J.-M.S.), Instituto de Salud Carlos III (grant PI14/00447 to A.N. and J.-M.S.), and MINECO (grant BIO2014-57716-C2-2-R to C.F.) within the framework of the Plan Estatal I+D+I 2013-2016 and cofunded by the ISCIII-Subdirección General de Evaluación y Fomento de la Investigación el Fondo Europeo de Desarrollo Regional (FEDER, Unión Europea, Una manera de hacer Europa), the European Foundation for the Study of the Diabetes (EFSD/Lilly Fellowship 2010 Programme to J.-M.S.), the CIBERDEM, the CERCA Programme (Generalitat de Catalunya), and the Department of Economy and Knowledge of Generalitat de Catalunya (grant 2014_SGR_520). A.M.-A. was the recipient of an Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS) Predoctoral Fellowship. The Diabetes and Obesity Research Laboratory at IDIBAPS is supported by the Sardà Farriol Research Programme. This work was developed at the Centre Esther Koplowitz, Barcelona, Spain.

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

Author Contributions. J.R.-C. contributed to the performance of experiments, data analysis, and writing and review of the manuscript. A.M.-A. contributed to the performance of experiments, data analysis, and review of the manuscript. J.M.-V., M.M., C.C., A.M.-F., X.B.-D.R., J.M.-C., and J.M. contributed to the performance of experiments and review of the manuscript. C.F., R.G., and A.N. contributed to the study design and review of the manuscript. J.-M.S. contributed to the study design, data analysis, and writing and review of the manuscript. All authors approved the final version of the manuscript. J.-M.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 52nd European Association for the Study of Diabetes Annual Meeting, Munich, Germany, 12–16 September 2016.