Glucolipotoxicity-Inhibited miR-299-5p Regulates Pancreatic β-Cell Function and Survival Free

Pancreatic β-cells tightly regulate insulin synthesis and secretion, thereby playing a pivotal role in systemic glucose homeostasis. A direct or indirect insufficiency of insulin results in diabetes (1,2). The occurrence of hyperglycemia and hyperlipidemia, which are common disorders in diabetes, gradually diminishes β-cell function and mass (3), which in turn further exacerbate the disease (4). The concept of glucolipotoxicity was proposed initially to explain the damaging influences of chronically excessive levels of glucose and lipids on β-cells. One critical issue that currently limits attempts to prevent or slow the progression of diabetes is an insufficient understanding of the molecular mechanisms that lead to the destruction of β-cells. Certain canonical mechanisms, including endoplasmic reticulum (ER) stress, the production of reactive oxygen species, and the induction of proinflammatory cytokines, have been implicated in diabetes biology and are undergoing intense study (5,6).

Diabetes is well accepted to have a strong genetic predisposition. Nevertheless, the current explosive epidemic of diabetes cannot be explained by genetic variability alone (7,8). Epigenetic mechanisms also play a crucial role in the development of diabetes by modulating gene expression through direct DNA methylation, chromatin (histone) modifications, and noncoding RNAs (ncRNAs) (8,9).

miRNAs are small (21–23 nucleotides) ncRNA molecules that act predominantly by inducing the decay of mRNAs or repressing their translation by binding to their 3′untranslated regions (3′UTRs) (10). A β-cell–specific Dicer deletion that occurs during embryonic progression disrupts the maturation of miRNAs and causes obvious diabetes (11,12). Moreover, DICER inactivation in adult β-cells impairs glucose-stimulated insulin secretion (GSIS), insulin content, and β-cell mass (13), highlighting the indispensable functions of miRNAs in the biology of β-cells. One example is miR-375, which is enriched in pancreatic islets. The genetic deletion of miR-375 in mice leads to hyperglycemia, reduced β-cell mass, and increased α-cell mass (14).

miRNAs also have been proposed to influence the pathogenesis of diabetes separately from their roles in the regulation of β-cell biology. We previously reported an association between elevated levels of miR-24 and impaired β-cell function in the islets of db/db mice and mice fed a high-fat diet (HFD) (15). Similarly, increased levels of miR-34a in the islets of diabetic mice had a deleterious effect on insulin secretion and β-cell survival (16). Other studies also have indicated an involvement of specific miRNAs in β-cell failure during the development of diabetes. However, we are still far from a full understanding of how these small, but powerful regulators function during the natural history of the β-cell in diabetes (16,17).

In this study, we analyzed the global variations in the expression profile of miRNAs in human islets incubated with palmitate and identified a series of differentially expressed miRNAs, some of which are in line with other studies (16,18). After assessing the effect of 17 differentially expressed miRNAs of interest on β-cell function and survival, we selected miR-299-5p for further investigations. These investigations revealed that the inhibition of miR-299-5p led to β-cell dysfunction and eventually β-cell apoptosis by upregulating a Trp53 effector, Perp. Our findings indicated that miR-299-5p is an essential regulator of β-cell biology, the downregulation of which may link PERP enhancement to β-cell dysfunction and apoptosis in glucolipotoxic settings. These observations provide new insights into the mechanisms underpinning glucolipotoxicity-induced β-cell failure and expand our understanding of the roles miRNAs play in the pathophysiology of diabetes.

The mouse pancreatic β-cell lines MIN6 and β-TC-6 (passage 16–30) were grown in DMEM (Invitrogen, Grand Island, NY) containing 15% FBS (Gibco, Burlington, Ontario, Canada). The rat pancreatic β-cell line INS-1 (passage 16–30) was cultured in RPMI medium (Invitrogen) containing 10% FBS. Both media were supplemented with 100 μg/mL streptomycin, 100 units/mL penicillin, 10 mmol/L HEPES, and 50 μmol/L β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO). All cells were cultured at 37°C in a humidified 5% CO₂ atmosphere. For high-glucose and palmitate treatment, cells were incubated in modified medium with 0.5% (weight for volume) BSA, various concentrations of glucose (low glucose: 5.5 mmol/L glucose + 19.5 mmol/L mannitol for INS-1 cells, 5.5 mmol/L glucose + 27.8 mmol/L mannitol for MIN6 cells; high glucose: 25 mmol/L for INS-1 cells, 33.3 mmol/L for MIN6 cells) and palmitate (0.2 mmol/L for INS-1 cells, 0.4 mmol/L for MIN6 cells). Mannitol (Sigma-Aldrich) served as an osmotic control. Palmitate (Sigma-Aldrich) was prepared as previously described (19).

The use of human islets was approved by the research ethics committee of Nanjing Medical University. The use of human islets was approved by the research ethics committee of the First Affiliated Hospital of Nanjing Medical University. All animal experiments were approved by the institutional animal care and use committee at Nanjing Medical University. Animals for experiments were purchased from the National Resource Center for Mutant Mice Model Animal Research Center of Nanjing University. The HFD was purchased from Research Diets (New Brunswick, NJ). Islets were isolated and cultured as described previously (20). Isolated islets were gathered and cultured together in RPMI medium (glucose: 11.1 mmol/L) containing 10% FBS, 100 units/mL penicillin, and 100 mg/mL streptomycin at 37°C in a humidified 5% CO₂ atmosphere. After equilibrating for 3 h, the islets were counted and replanted into 6- or 48-well plates and cultured overnight for further experiments. For high-glucose and palmitate treatment, islets were incubated in modified medium with 0.5% (weight for volume) BSA, various concentrations of glucose (low glucose: 5.5 mmol/L; high glucose: 25 mmol/L for primary Sprague Dawley [SD] rat islets, 33.3 mmol/L for primary mouse islets) and palmitate (0.5 mmol/L).

For GSIS and potassium-stimulated insulin secretion (KSIS) assays, cells and isolated primary islets were seeded in 48-well plates. After incubation for 1 h in HEPES-balanced Krebs-Ringer bicarbonate buffer (KRBH) (115 mmol/L NaCl + 4.7 mmol/L KCl + 1.2 mmol/L MgSO₄ · 7 H₂O + 1.2 mmol/L KH₂PO₄ + 20 mmol/L NaHCO₃ + 16 mmol/L HEPES + 2.56 mmol/L CaCl₂ + 0.2% BSA), the islets and MIN6 cells were incubated for 1 h in KRBH with low glucose (3.3 mmol/L for islets, 2 mmol/L for MIN6 cells, 0.2 mmol/L for β-TC-6 cells) or stimulatory glucose (16.7 mmol/L for islets, 20 mmol/L for MIN6 cells, 11.2 mmol/L for β-TC-6 cells). The INS-1 cells were incubated for 1 h in KRBH with low glucose (3.3 mmol/L) and KCl (50 mmol/L).

For human islet perfusion analysis, islets of equal number (80 per group) and size were placed on a nylon filter in a plastic perfusion chamber (Millex-GP; Millipore) and perfused with KRBH at a flow rate of 125 μL/min. The insulin levels of supernatants and extractions were measured using radioimmunoassay. Both insulin secretion and insulin content were normalized to total protein amount.

Total RNA was isolated using TRIzol reagent (Invitrogen) from human islets treated with or without palmitate. Global miRNA expression profiling was performed at Gene Tech Biotechnology (Shanghai, China) using the GeneChip miRNA Galaxy Array (Affymetrix, Santa Clara, CA).

All sequences of miR-299-5p response elements (MREs) were from public sources (miRanda, TargetScan, and PicTar). Short sequences of the Perp (NM_022032.4), Ras association domain family member 6 (Rassf6) (NM_001025671.1), siah E3 ubiquitin protein ligase 1 (Siah1) (NM_009172.2), and serum/glucocorticoid regulated kinase 1 (Sgk1) (NM_001161845.2) genes were obtained by annealing and then cloned into pMIR-REPORT Luciferase miRNA Expression Reporter Vector (Ambion, Foster City, CA) between the SpeI and HindIII sites. To construct Perp expression plasmids, the coding sequence for Perp was amplified by PCR from the mouse full-length cDNA and then cloned into pCMV5 vector between the BglII and HindIII sites and pEGFP-C1 vector between the BglII and SmaI sites. The NPY-mOrange plasmid was provided by X. Tao (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China). Plasmids were confirmed to be correct by sequencing. Primer sequences for PCR are shown in Supplementary Table 1.

miRNA duplex mimics, 2′-O-methylated single-strand miRNA antisense oligonucleotides, and negative controls were obtained from GenePharma (Shanghai, China). Small interfering RNAs (siRNAs) were designed and synthesized by RiboBio (Guangzhou, Guangdong, China). For transient transfection, Lipofectamine 2000 reagent (Invitrogen) was mixed with miRNA mimics/inhibitors, siRNAs, or overexpression/reporter plasmids as previously described (15). Luciferase activities were measured using the Dual-Glo Luciferase Assay System (Promega, Madison, WI) on a TD-20/20 Luminometer (Turner BioSystems, Sunnyvale, CA) according to the manufacturer’s protocols as previously described (15).

MIN6 cells were washed with ice-cold PBS, collected, and dissolved in lysis buffer (7 mol/L urea, 1% CHAPS). Protein digestion, tandem mass tag labeling, and mass spectrometry analysis then were conducted at the Analysis Center of Nanjing Medical University as previously described (21).

Total RNA was extracted using TRIzol reagent, and both mRNA and miRNA quantification were performed by using the SYBR Green PCR Master Mix and LightCycler480 II Sequence Detection System (Roche, Basel, Switzerland) as previously described (15). U6 and actin were used as internal standards for miRNAs and mRNAs, respectively. For miR-299-5p and U6, TaqMan probes (Ambion) were used to confirm the results. The primers used are shown in Supplementary Table 2.

Western blotting was performed as previously described (20). The antibodies used are listed in Supplementary Table 3.

Cells were seeded in 48-well plates (4 × 10⁴ cells/well) in 200 μL culture medium. Cell Counting Kit-8 (CCK-8) assay was performed according to the manufacturer’s protocol (Vazyme Biotech, Nanjing, Jiangsu, China).

Cellular DNA isolation and DNA ladder detection were performed according to the manufacturer’s instructions (Roche) as previously described (20).

TUNEL staining was performed using the In Situ Cell Apoptosis Detection Kit III (fluorescein isothiocyanate) (Boster Biological Technology, Wuhan, Hubei, China), according to the manufacturer’s protocol (22). Islets and MIN6 cells were observed under laser scanning confocal microscope (FV1200; Olympus, Tokyo, Japan).

MIN6 cells were plated in 3.5-cm Greiner dishes and cotransfected with enhanced green fluorescent protein (EGFP)–tagged PERP (EGFP-PERP) and NPY-mOrange. During imaging, cells were cultured in a heated, gas-perfused chamber (Tokai Hit, Shizuoka, Japan) at 37°C with 5% CO₂ and visualized with a laser scanning microscope (FV1200). Cells were incubated with low glucose (2 mmol/L), high glucose (20 mmol/L), and KCl (50 mmol/L). For each treatment, cells were imaged for 5 min.

At least three independent experiments were performed. Comparisons were performed using the Student t test between two groups or ANOVA in multiple groups. Results are presented as mean ± SEM. P <0.05 is considered statistically significant.

To investigate the contribution of miRNAs to β-cell dysfunction and apoptosis during the development of type 2 diabetes, we performed global miRNA expression profiling in human pancreatic islets incubated with 0.25 and 0.5 mmol/L palmitate for 48 h in the presence of high glucose (11.1 mmol/L). The damaged phenotypes of the treated human islets are presented in Supplementary Fig. 1. The miRNAs that showed a ≥1.5-fold difference between the control and palmitate-exposed islets in our microarray experiments are displayed in Supplementary Tables 4 and 5. Using miR-34a as a positive control, we confirmed the upregulation of miR-199a-3p, miR-483-5p, and miR-933 as well as the downregulation of miR-299-5p, miR-204, and miR-362-3p by quantitative RT-PCR (qRT-PCR) in isolated human islets incubated with palmitate (Fig. 1A). Similar results were discovered in isolated islets from db/db mice (Fig. 1B) as well as from INS-1 cells incubated with palmitate (Fig. 1C). Because miR-933 does not exist in mouse and rat, we did not quantify miR-933 in mouse islets and INS-1 cells.

After identifying 17 differentially expressed miRNAs of interest from miRNA expression profiling in human islets (Fig. 2A), we mimicked their alterations by transiently transfecting the oligonucleotide duplexes corresponding to the upregulated miRNAs or specific inhibitors of the downregulated miRNAs into β-TC-6 cells and INS-1 cells. We then assessed GSIS in β-TC-6 cells, KSIS in INS-1 cells, and insulin content in both cell lines at 48 h after transfection (Fig. 2B–E and Supplementary Fig. 2). The viability of β-TC-6 cells was tested 72 h after transfection (Fig. 2F and G). The miRNAs that influenced cell viability were further assessed for their effects on apoptosis in β-TC-6 cells (Fig. 2H and I). Finally, miR-299-5p, which is highly conserved among humans, mice, and rats, was chosen for further investigations because it showed the strongest comprehensive influences on β-cell function and survival among all the miRNAs we tested.

The downregulation of miR-299-5p was detected in the islets of db/db mice and Goto-Kakizaki (GK) rats (Fig. 3A and B). Moreover, it was observed in isolated mouse and rat islets, MIN6 cells, and INS-1 cells treated with high glucose and palmitate (Fig. 3C–F). However, other canonical in vitro diabetic milieus created by interleukin-1β, tumor necrosis factor-α, interferon-γ, thapsigargin, and tunicamycin did not significantly influence miR-299-5p expression in MIN6 and INS-1 cells (Supplementary Fig. 3), suggesting that glucolipotoxicity may specifically regulate miR-299-5p expression in β-cells.

We further confirmed the role of miR-299-5p downregulation by inhibiting endogenous miR-299-5p expression in isolated C57BL/6 mouse islets, MIN6 cells, and INS-1 cells (Supplementary Fig. 4A–C). Defective GSIS and KSIS as well as decreased insulin content were observed as early as 24 h after transfection for isolated islets and 12 h after transfection for β-cell lines (Fig. 4A–D and Supplementary Fig. 4D and E). In particular, both the GSIS and the insulin content of MIN6 cells were reduced in a dose-dependent manner by the inhibition of miR-299-5p (Fig. 4C and D). However, neither the mRNA levels of insulin 1 (Ins1) and Ins2 nor the protein levels of critical transcription factors of Insulin, including MAF bZIP transcription factor A (MAFA), pancreatic and duodenal homeobox 1 (PDX1), and neuronal differentiation 1 (NEUROD1), were affected by miR-299-5p inhibition (Supplementary Fig. 4F–H). These results suggest that events responsible for the decrease in insulin content may not take place at the transcriptional level. In addition, the overexpression of miR-299-5p (Supplementary Fig. 4I) attenuated the detrimental effects of glucolipotoxicity on GSIS and insulin content in MIN6 cells (Fig. 4E and F). These findings imply that downregulation of miR-299-5p expression was at least partly responsible for the glucolipotoxicity-induced β-cell dysfunction.

Further investigations demonstrated that the inhibition of miR-299-5p expression in MIN6 cells and INS-1 cells resulted in significantly decreased cell viability, a markedly increased number of pyknotic nuclei, and typical fragmentations of DNA (Fig. 5A and Supplementary Fig. 5A–E). Of note, the knockdown of miR-299-5p increased the cleavage of poly (ADP-ribose) polymerase 1 (PARP1), which is a downstream substrate of caspase-3, in a time-dependent manner. This response was observed as early as 24 h after transfection in MIN6 cells (Fig. 5B and Supplementary Fig. 5F). Furthermore, the inhibition of miR-299-5p in primary SD rat islets (Supplementary Fig. 5G) led to a similar time-dependent increase in the number of TUNEL-positive β-cells (Fig. 5C and D). These findings indicate that the knockdown of miR-299-5p alone is sufficient to induce β-cell apoptosis after the impairment of β-cell function. Likewise, TUNEL assays revealed the resistance of miR-299-5p–overexpressing MIN6 cells against apoptosis induced by high-glucose and palmitate incubation (Fig. 5E and Supplementary Fig. 5H), further suggesting that miR-299-5p downregulation acts as a mediator of the glucolipotoxicity-induced apoptosis of β-cells.

Transfection with anti-nc or anti-miR-299-5p into MIN6 cells for 24 h resulted in a differential expression of proteins, as detected by proteomics analysis (Supplementary Table 6). The changes in endoplasmic reticulum to nucleus signaling 1 (ERN1), GINS complex subunit 2 (GINS2), and pyruvate kinase (PKLR) were confirmed by Western blotting (Supplementary Fig. 6A and B), verifying the reliability of the proteomics analysis. We noticed that INS2 was one of the downregulated proteins (Supplementary Table 6), which agrees with the reduced insulin content resulting from the inhibition of miR-299-5p. We used ingenuity pathway analysis to interrogate the altered proteins indicated by the proteomics analysis and revealed enrichments in the pathways of β-cell dysfunction and apoptosis (Supplementary Fig. 6C and D), which further confirmed the β-cell phenotypes elicited by miR-299-5p downregulation.

We included the top 100 predicted target genes of miR-299-5p obtained from miRanda, TargetScan, and PicTar in gene ontology analysis using DAVID Bioinformatics Resources 6.7, and revealed significant correlations between the predicted target genes and several biological processes, including regulation of the protein modification process and apoptosis (Supplementary Fig. 7A). The gene ontology analysis identified several apoptosis-related predicted target genes, and we assessed the conservation of their miR-299-5p binding sites among humans, mice, and rats (Supplementary Fig. 7B). Finally, Perp, Siah1, Rassf6, and Sgk1 were selected as candidates.

As expected, the regulatory role of miR-299-5p in the expression of the four predicted target genes was confirmed by luciferase assay (Fig. 6A and B). Moreover, the protein levels of Perp, Siah1, Rassf6, and Sgk1 were decreased by the overexpression of miR-299-5p and conversely, were increased by the knockdown of miR-299-5p in MIN6 cells (Fig. 6C and D) and INS-1 cells (Supplementary Fig. 7C). The mRNA levels of the four target genes in both cell lines are shown in Fig. 6E and F and Supplementary Fig. 7D and E. We also investigated the protein and mRNA levels of these four target genes under glucolipotoxic conditions (Fig. 6G and H) and demonstrated the induction of increased protein levels of the four targets in MIN6 cells by high-glucose and palmitate treatment.

The protein levels of Perp were significantly increased in the islets from db/db mice as well as HFD-fed SD rats (Fig. 7A and B), which showed increased body weight and blood glucose levels (Supplementary Fig. 8A and B). In addition, the mRNA level of Perp was significantly increased in isolated human islets incubated with palmitate (Fig. 7C). Given that the binding sites for miR-299-5p within the 3′UTR of Perp were highly conserved among 11 species (Supplementary Fig. 8C), Perp was chosen for further study.

Endogenous Perp was silenced by siRNAs (Supplementary Fig. 8D and E), which resulted in a partial restoration of GSIS, insulin content, and cell viability as well as a reduction in cell apoptosis in MIN6 cells transfected with anti-miR-299-5p (Fig. 7D–H). Human islet perfusion analysis also showed that disrupted biphasic GSIS and KSIS of human islets induced by miR-299-5p downregulation were partly reversed by silencing Perp (Fig. 7I–L and Supplementary Fig. 8F and G).

Moreover, the deletion of Perp partly preserved β-cell function and mass in glucolipotoxic settings (Fig. 7M–O). These findings establish Perp as a main target of miR-299-5p and implicate this gene as a mediator of the β-cell dysfunction and apoptosis induced by glucolipotoxicity.

The results of the GSIS assay in MIN6 cells overexpressing Perp revealed that both GSIS and insulin content were decreased by Perp overexpression in MIN6 cells (Fig. 8A and B), further verifying the roles of PERP in regulating insulin secretion and content in β-cells. Moreover, Perp overexpression alone was insufficient to cause β-cell apoptosis but was able to exacerbate the glucolipotoxicity-induced apoptosis in MIN6 cells (Supplementary Fig. 9A).

PERP originally was reported to be localized to the plasma membrane and Golgi apparatus (23). Here, we also identified much localization of PERP to ERs, early endosomes, and late endosomes as well as slight localization to lysosomes and mitochondria in MIN6 cells overexpressing EGFP-PERP (Supplementary Fig. 9B). Next, we cotransfected MIN6 cells with EGFP-PERP and NPY-mOrange, the latter of which is used widely to visualize insulin granules in pancreatic β-cells (24,25), and found that EGFP-PERP was abundantly localized on the membrane of insulin granules during the formation, trafficking, docking, fusion, and retrieval of insulin granules (Fig. 8C and Supplementary Video 1). Of note, stimulations by high glucose or KCl led to a significant accumulation of PERP on the membrane of insulin granules near the nuclei in MIN6 cells (Fig. 8D and Supplementary Videos 2–4). Taken together, our findings strongly suggest that despite being a well-known apoptosis effector, PERP also plays a critical role in modulating insulin secretion.

Our study identified a group of differentially expressed miRNAs from human islets incubated with palmitate and assessed their roles in β-cell function and survival. Among them, miR-299-5p stood out because of its remarkable influences on β-cell function and survival. Further investigations supported Perp as a main functional target of miR-299-5p in β-cells.

miR-299-5p is located in the imprinted Dlkl-Dio3 region of chromosome 14q32.31, which is an evolutionarily conserved region that has a strong correlation with cancer (26,27). Previously, miR-299-5p was discovered to have a proapoptotic function in a human prostate cancer cell line (28), whereas in neurons of mice with Alzheimer disease, miR-299-5p was reported to attenuate autophagy-associated apoptosis by targeting Atg5 (29). We attribute these discrepancies in the effects of miR-299-5p, compared with the current results, to differences in miR-299-5p expression patterns and its main functional targets among various diseases and tissues.

We preliminarily ruled out the possible effects of other diabetic milieus, including ER stress and inflammation (or cytokines), on the regulation of miR-299-5p expression in β-cells, implying that miR-299-5p is very likely to act as a specific effector of imbalanced energy metabolism in β-cells exposed to excessive glucose and lipids. Given the elevated levels of reactive oxygen species generated by nutritional stress (30) and the low levels of antioxidant enzymes in β-cells (5,31), future studies are needed to identify whether glucolipotoxicity-induced miR-299-5p downregulation is mediated by oxidative stress. To investigate the molecular mechanisms that regulate miR-299-5p expression, we analyzed the 20-kb upstream region of the miR-299-5p precursor by using the Matrix Search Tool (www.bioinformatics.org/grn/npb3) and found several putative p53 binding sites. It has been reported that the transcriptional activity of p53 increases in the context of glucolipotoxicity (32,33) and that p53 represses the transcription of several miRNAs in other cell lines (34). Of note, resveratrol, which is a well-known activator of p53 (35), aggravated the downregulation of miR-299-5p induced by glucolipotoxicity in MIN6 cells, whereas silencing p53 reversed the decreases in miR-299-5p expression caused by resveratrol and glucolipotoxicity (data not shown). These findings raise the possibility that the transcription of miR-299-5p is repressed by p53 under the condition of glucolipotoxicity, which needs further exploration.

We observed a significant downregulation of ERN1 (also named IRE1) in MIN6 cells after transfection with anti-miR-299-5p for reasons that still require further exploration. ERN1 is an ER transmembrane protein that is pivotal to the unfolded protein response and ER stress (36) and is essential for insulin biosynthesis and secretion (37). Indeed, β-cell–specific Ire1 knockout mice exhibit hyperglycemia, hypoinsulinemia, and glucose intolerance. Moreover, the primary islets of these Ire1 knockout mice has a significantly decreased insulin content and reduced GSIS (38). Of note, Hassler et al. (39) reported decreased insulin biosynthesis at the posttranscriptional level as well as β-cell apoptosis through oxidative stress in adult β-cell–specific Ire1 knockout mice. These findings suggest that ERN1 plays a dual role in β-cell function and survival, with mechanisms yet to be determined. This finding is similar to another serine/threonine protein kinase, macrophage stimulating 1 (MST1), which is viewed as a critical regulator of apoptotic β-cell death and function (40). Therefore, future studies will be directed toward ascertaining the mechanisms by which miR-299-5p inhibition downregulates ERN1 expression and ERN1 downregulation influence insulin biosynthesis as well as insulin secretion.

As a widely accepted effector of p53-dependent apoptosis (23,41,42) and a target of miR-299-5p that is putatively repressed by p53, PERP may be upregulated both directly and indirectly by p53 under glucolipotoxic conditions (Supplementary Fig. 10). Previously, Ladiges et al. (43) demonstrated an upregulation of Perp in the islets of P58^IPK-null mice, which displayed hyperglycemia, hypoinsulinemia, and β-cell depletion. In the current study, Perp was increased in the islets from both db/db mice and HFD-fed rats as well as the human islets incubated with palmitate. Perp deficiency partly reversed the defective insulin secretion and insulin content as well as β-cell apoptosis induced by miR-299-5p inhibition and glucolipotoxicity. By contrast, Perp overexpression decreased the insulin secretion and insulin content of β-cells without influencing their apoptosis and aggravated the proapoptotic effect of glucolipotoxicity on β-cells. These findings strongly argue that the β-cell dysfunction induced by PERP is not a secondary effect of cell apoptosis and point to a new miR-299-5p/Perp regulatory pathway in β-cells under the conditions of glucolipotoxicity for the first time. As a transmembrane protein, PERP originally was found to be localized to the plasma membrane and Golgi apparatus (23), and we identified its localization also to the ERs, early endosomes, late endosomes, and lysosomes, all of which are typical organelles involved in the membrane trafficking system and closely related to insulin secretion in β-cells (44). Thus, we wondered whether PERP might influence insulin secretion through the membrane trafficking system either directly or indirectly. After ruling out the possibility that PERP may indirectly affect insulin secretion through perturbing F-actin remodeling (Supplementary Fig. 9C and D), which is essential for insulin secretion (45), we discovered two lines of evidence to support the direct engagement of PERP in insulin secretion. First, PERP was found to be localized to the membrane of insulin granules during their formation, trafficking, docking, fusion, and retrieval, providing a material basis for the modulatory effect of PERP on insulin secretion. Second, we captured a dramatic accumulation of PERP near the nuclei of MIN6 cells stimulated by high glucose or potassium by live-cell imaging, which indicates that PERP is very likely to be involved in mediating the response of insulin granules to glucose or potassium stimulation. The results of live-cell imaging also implied that PERP may take part in exocytosis-endocytosis coupling of insulin granules in β-cells. Because PERP was much more strongly localized to the ERs than to the lysosomes, we propose that in β-cells, PERP may promote endocytosis followed by its accumulation in the ERs, which in turn perturbs the exocytosis-endocytosis coupling of insulin granules as well as ER homeostasis. When such perturbations are mild, they can be compensated by moderate ER stress, which results in only a modest decrease of insulin secretion and insulin content. However, when such perturbations and impairment are too intense to be compensated, the induction of excessive ER stress may aggravate β-cell dysfunction and even lead to β-cell apoptosis through the PERK and ATF6 pathways, which might be further exacerbated by decreased ERN1 induced by miR-299-5p downregulation. These speculations all need further experimental validation, but they open up new opportunities for future advances. We cannot exclude the possible contributions of the other confirmed targets of miR-299-5p, including Siah1, Rassf6, and Sgk1, which also should be explored in further studies.

In conclusion, this study discovered miR-299-5p among a large group of differentially expressed miRNAs in human islets incubated with palmitate. To our knowledge, this study is the first to illustrate that miR-299-5p is essential to β-cell biology and that its downregulation specifically is induced by glucolipotoxicity-aggravated β-cell dysfunction and apoptosis partly by targeting Perp. Because an appropriate expression of miRNAs is required for optimal β-cell biology, whereas alterations in the levels of certain miRNAs have deleterious effects on β-cells that may facilitate the development of diabetes, obtaining a comprehensive understanding of their roles and mechanisms is necessary to broaden our strategies for the prevention and treatment of diabetes.

Acknowledgments. The authors thank Tao Xu, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China, for gifting the NPY-mOrange plasmid and Xiaosu Li, a freelance translator in Beijing, China, for editorial assistance.

Funding. This study was supported by research grants from the National Key Research and Development Program of China (2016YFC1304804) to X.H. and the National Natural Science Foundation of China (81420108007 to X.H. and 81670703 to Y. Zhu). X.H. is a fellow at the Collaborative Innovation Center for Cardiovascular Disease Translational Medicine.

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

Author Contributions. Q.H., W.Y., Y.L., Y.S., Y.Zho., and Y.Zha. performed the research. Q.H., W.Y., Y.L., and Y. Zhu analyzed the data. Q.H. and Y. Zhu wrote the manuscript. Q.H., Y. Zhu, and X.H. designed the research. D.L. and S.Z. contributed new reagents and analytic tools. Y. Zhu and X.H. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.