Stock solutions of stearic acid (Sigma, St. Louis, MO, USA) were prepared as described previously [29]. Briefly, after the sodium salt was dried, it was resuspended in saline and then heated to 80°C until it completely dissolved. Then, 20% (w/v) bovine serum albumin (BSA) was added, and the mixture was stirred at 50°C for 4 hours. The complex was then sterilized and aliquoted. Melatonin (M5250-1G, Sigma-Aldrich, St. Louis, MO, USA) was dissolved in anhydrous ethanol to form a stock solution of 200 mmol/L. The working concentration was 100 μmol/L after dilution in culture medium based on previous studies [30] and our dose-dependent test of melatonin, in which the final ethanol content was less than 0.05% (v/v).

Seven-week-old male C57BL/6 mice were obtained from Beijing Vital River Laboratory Animal Technology Company (Beijing, China). The mice were randomly allocated to normal diet and high-stearic-acid diet (HSD) groups (n=10 per group). The compositions of the normal diet and HSD were provided by the Beijing HFK Bioscience (Beijing, China) ( Supplemental Table S1). All animal procedures were approved by Institutional Animal Care and Use Committee of Harbin Medical University, and the animals were handled following the guidelines of the Animal Experimental Center of Harbin Medical University. Pancreatic tissues and blood samples were collected after 12-week feeding.

Fasting serum nonesterified fatty acids were transformed into fatty acid methyl esters as described previously [5,31]. Heptadecanoic acid (C17:0) was used as an internal standard. Fasting serum glucose, total cholesterol, triacylglycerol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol levels were calculated with an automatic analyzer (Hitachi-7100, Hitachi, Tokyo, Japan). All kits were purchased from Biosino Biotechnology (Beijing, China). Serum insulin concentrations were measured using a mouse/rat insulin enzyme-linked immunosorbent assay kit (Cat. No. EZRMI-13K, Millipore, Burlington, MA, USA).

Mouse pancreatic β-TC6 cells were obtained from the Chinese Sciences Cell Library (Shanghai, China) and cultured in Dulbecco’s modified Eagle’s medium (Gibco/Life Technologies, Carlsbad, CA, USA), supplemented with 15% (v/v) fetal bovine serum (Gibco/Life Technologies), 1.5 g/L NaHCO3, and 100 IU/mL penicillin-streptomycin (Gibco/Life Technologies). The β-TC6 cells were treated with 400 μmol/L stearic acid for 24 hours. The mouse islets were provided by Procell Life Science & Technology Co., Ltd. (Wuhan, China) and were isolated after digestion through ductal injection of collagenase P (Cat. No. 11213873001, Roche Molecular Biochemicals, Mannheim, Germany) and purified using Ficoll 400 (Cat. No. 17-0300-10, Pharmacia, Stockholm, Sweden), as described previously [32].

Cell viability was determined using a cell counting kit 8 (CCK-8) assay (C0038, Beyotime Biotechnology, Shanghai, China). The β-TC6 cells were seeded into 96-well plates at a density of 3×104 cells/well. Then, 10 μL of CCK-8 was added to each well. After incubation at 37°C for 2 hours, optical density was evaluated at 450 nm using a microplate reader (SpectraMax M2, Molecular Devices, San Jose, CA, USA).

Secretion buffer was prepared as described previously [3]. β-TC6 cells were incubated in secretion buffer (2.8 or 20 mmol/L) for 60 minutes. The culture medium was then collected for insulin measurements, and the cells were simultaneously lysed for protein concentration measurements to normalize the insulin level. Insulin levels were detected using a mouse/rat insulin enzyme-linked immunosorbent assay kit (Cat. No. EZRMI-13K).

Senescence-associated β-galactosidase (β-gal) activity was measured using a senescence β-gal staining kit (C0602, Beyotime Biotechnology), as described previously [33,34]. β-TC6 cells in 24-well plates were washed with phosphate-buffered saline (PBS) three times. Then, 250 μL of fixative solution was added to each well and incubated for 15 minutes at room temperature. After three washes, the cells were incubated with X-Gal staining buffer at 37°C without CO₂. We calculated the ratio between the number of positive cells (blue stained) and the total number of cells present in five different fields in each well. The total number of cells in each field was determined by Hoechst 33342 staining (C1022, Beyotime Biotechnology). We independently repeated the experiment three times.

EdU staining was carried out using a BeyoClick EdU-488 cell proliferation kit (C0071S, Beyotime Biotechnology). The β-TC6 cells were seeded onto a 24-well plate at a density of 1×10⁵ cells/well. They were fixed with 4% paraformaldehyde for 10 minutes and washed with PBS buffer 3% BSA, followed by incubation with 0.3% Triton X-100 (P0096, Beyotime Biotechnology) for 15 minutes at room temperature. After washing, EdU reaction buffer was then added and incubated for 30 minutes at room temperature in the absence of light. Finally, Hoechst 33342 was used for cell counting.

A RiboFect CP Transfection Kit (C10511-05, RiboBio Co. Ltd., Guangzhou, China) and Lipofectamine 2000 (Cat.11668019, Invitrogen, Carlsbad, CA, USA) were used to transfect cells with oligonucleotides and luciferase reporter vectors, according to the manufacturer’s instructions. Cells were collected for detection 24 hours after transfection. miR-146a-5p mimic, anti-miR-146a oligonucleotide (AMO-146a), miR-8114 mimic, anti-miR-8114 oligonucleotide (AMO-8114), overexpression plasmid pEX-RB-Mam carrying Mafa, and their negative controls were synthesized and constructed by Guangzhou RiboBio Co. Ltd.

The total RNA from β-TC6 cells and mouse islets was extracted using TRIzol reagent (Invitrogen) [35], according to the manufacturer’s protocol. A miRVana miRNA isolation kit (Ambion, Austin, TX, USA) was used to isolate miRNAs. Quantitative polymerase chain reaction (PCR) was performed using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). β-actin and U6 were used as internal controls for mRNA and miRNA, respectively. All primers were synthesized by Sangon Biotech Co. Ltd. (Shanghai, China), and their sequences are shown in Table 1.

Cells were gently washed three times in PBS, and 50 μL of intermediate radio-immunoprecipitation assay (RIPA) lysis buffer (Beyotime) was then added. Protein concentrations were detected using bicinchoninic acid (BCA) protein assay kits (Cat. No. P0010, Beyotime). Protein samples (50 μg) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes. The primary antibodies used were as follows: MAFA (DF13429, rabbit polyclonal, 1:1,500, Affinity Biosciences, Cincinnati, OH, USA) and β-actin (ab8266, rabbit polyclonal, 1:1,000, Cell Signaling Technology, Beverly, MA, USA). The secondary antibody was an anti-rabbit alkaline phosphatase-conjugated antibody (S373B, 1:7,500, Promega, Madison, WI, USA). Protein detection was performed using stabilized substrate for alkaline phosphatase (S3841, Promega) and the FluorChem R system (ProteinSimple, San Jose, CA, USA).

Luciferase reporter plasmids carrying the 3′-untranslated region (3′-UTR) of Mafa containing wild-type or mutant binding sites for miR-146a-5p or miR-8114 were constructed by RiboBio Co. Ltd. HEK293 cells were co-transfected with 200 ng of vectors and miR-146a-5p or miR-8114 mimics or their negative controls. Cell lysates were collected 24 hours after transfection, and the luciferase activity was measured using a luciferase reporter assay kit (E1910, Promega) and a GloMax 20/20 luminescence detector.

All data were statistically analyzed using SPSS version 21.0 (IBM Corp., Armonk, NY, USA) and are presented as mean±standard deviation. The two-tailed Student t test was used to analyze differences between two groups, and one-way analysis of variance followed by the Student-Newman-Keuls test was used to test differences among multiple groups.

An HSD in the mouse model that mimicked increased proportions of stearic acid was successfully established, as evidenced by a significant elevation of circulating stearic acid levels ( Supplemental Table S2). The metabolic characteristics of these animals are summarized in Supplemental Table S3. After 12-week feeding with the HSD, the expression of markers of aging (Igf1r and Bambi), senescence (Cdkn2a and Trp53bp1), senescence-associated secretory phenotype factors (Ccl, Il1a, Il6, Tnf, and Cd99), and disallowed genes (Cat and Ldha) was significantly upregulated, and the expression of β-cell identity genes (Ins1 and Mafa) was robustly downregulated in mouse islets (Fig. 1).

The mouse β-TC6 cell line is derived from insulinoma cells arising in transgenic mice expressing the SV40 T antigen and can maintain stable in terms of differentiation in cell culture. It has remarkably similar characteristics to those of normal islets and maintains normal glucose responsiveness, as described previously [36-38]. We also observed a low level of cell proliferation after EdU staining (Supplemental Fig. S1A) and significant upregulation of p21 expression (Fig. 2, Supplemental Fig. S1B) in stearic acid-treated β-TC6 cells. Therefore, the β-TC6 cell line was selected as an appropriate model for this study.

The incubation of β-TC6 cells with different concentrations (12.5, 25, 50, 100, 200, 400, 800, 1,600, 3,200, 6,400, and 12,800 μmol/L) of melatonin showed that cell survival rate decreased in a dose-dependent manner at concentrations above 400 μmol/L (Fig. 2A). We chose a 100 μmol/L dose of melatonin to treat β-TC6 cells for 6, 12, 24, and 48 hours, and observed a significant reduction in cell viability at the 48-hour time-point (Fig. 2B). To confirm that the ethanol content in the melatonin solution did not affect cell viability, we also treated β-TC6 cells with different concentrations of ethanol in parallel with different concentrations of melatonin. We found no significant effect on cell viability when the ethanol concentration was under 0.4% (v/v), which equates to a melatonin concentration of 800 μmol/L (Supplemental Fig. S2). Next, treatment with 100 μmol/L melatonin for 24 hours significantly ameliorated the stearic acid-induced decrease in β-TC6 cell viability (Fig. 2C) and impairment on the glucose-stimulated insulin secretion assay (GSIS) (Fig. 2D). Moreover, stearic acid significantly increased the expression of markers of aging (Igf1r and Bambi), senescence (Cdkn2a and Trp53bp1), senescence-associated secretory phenotype factors (Ccl, Il1a, Il6, Tnf, and Cd99), and disallowed genes (Cat and Ldha) in β-TC6 cells, and decreased the expression of β-cell identity genes (Ins1 and Mafa). The effects on all these genes were robustly reversed after melatonin treatment (Fig. 2E). Meanwhile, stearic acid-increased senescence-associated β-gal activity in β-cells was also blocked after melatonin exposure (Fig. 2F).

In our previous study [28], we found that 13 miRNAs were upregulated and three miRNAs were downregulated after chronic exposure to stearic acid in β-TC6 cells using high-throughput sequencing. Among them, the expression of miR-146a-5p and miR-8114 was markedly elevated. Here, we verified this result in both HSD mice islets and stearic acid-treated β-TC6 cells. As shown in Fig. 3A, B, stearic acid significantly increased the expression of miR-146a-5p and miR-8114 in HSD-fed mice islets, compared with the control group. Similar increases in the levels of these two miRNAs were also observed in β-TC6 cells (Fig. 3C, D) and in mouse islets treated with stearic acid (Fig. 3E, F). Interestingly, these increases were effectively reversed by the addition of melatonin in β-TC6 cells (Fig. 3C, D) and in isolated islet cells (Fig. 3E, F).

Transfection of AMO-146a significantly suppressed endogenous levels of miR-146a-5p in the presence and absence of stearic acid (Fig. 4A). After treatment of β-TC6 cells with AMO-146a-5p, the stearic acid-induced reduction of cell viability and impaired GSIS significantly recovered (Fig. 4B, C). Meanwhile, inhibition of miR-146a-5p alleviated stearic acid-promoted β-cell senescence by suppressing the mRNA levels of senescence-related genes (Fig. 4D) and stearic acid-promoted β-gal activity (Fig. 4E). In contrast, overexpression of miR-146a-5p significantly impaired GSIS (Supplemental Fig. S3A) in β-TC6 cells in the absence of stearic acid, and this impairment was largely rescued by AMO-146a. Moreover, transfection of miR-146a-5p mimics resulted in the dysregulation of senescence-related gene expression (Supplemental Fig. S3B) and β-gal activity (Supplemental Fig. S3C). However, transfection of miR146a-5p alone did not affect cell viability (Supplemental Fig. S3D). The transfection efficiency of the miR-146a mimic is shown in Supplemental Fig. S3E. In parallel with miR-146a-5p, similar results were observed with transfection of miR-8114 AMO or its mimic with or without stearic acid treatment (Fig. 5, Supplemental Fig. S4), indicating that miR-8114 also participates in stearic acid-induced β-TC6 cell senescence and dysfunction. In addition, knockdown of miR-146a-5p and miR-8114 also blocked stearic acid-induced dysregulation of senescence-related gene expression in mouse islet β-cells (Supplemental Fig. S5).

Stearic acid significantly decreased Mafa expression, and melatonin reversed this inhibition in both β-TC6 cells (Fig. 6A) and islet β-cells (Supplemental Fig. S6A). The predicted Mafa binding sites of miR-146a and miR-8114 are shown in Fig. 6B, C, respectively. Mafa mRNA and protein levels were significantly decreased after the incubation of β-TC6 cells with stearic acid. This was significantly reversed by knockdown of miR-146a-5p or miR-8114 (Fig. 6D-G). Similar changes in Mafa protein expression were also observed in stearic acid-treated islet cells after inhibition of miR-146-5p and miR-8114 with AMO-146a-5p and AMO-8114 (Supplemental Fig. S6B, C). In the absence of stearic acid, transfection of miR-146a-5p or miR-8114 mimics led to downregulation of Mafa expression (Fig. 6H-K). In addition, the luciferase activity of the vector encoding the wild-type 3’-UTR of Mafa was significantly suppressed after co-transfection with miR-146a or miR-8114 mimics; however, this suppression was not observed in the luciferase reporter vectors carrying mutant binding sites for miR-146a or miR-8114 (Fig. 6L).

Melatonin blocked the upregulation of miR-146a-5p and miR-8114 levels after transfection of their mimics alone (Figs. 7A, 8A). Melatonin also obviously alleviated miR-146a-5p-induced GSIS impairment (Fig. 7B) and β-cell senescence (Fig. 7C, D). Meanwhile, the decreases in Mafa mRNA and Mafa protein levels induced by the miR-146a-5p mimic were significantly reversed after melatonin treatment (Fig. 7E, F). Melatonin also significantly inhibited miR-8114-induced GSIS impairment (Fig. 8B), β-cell senescence (Fig. 8C, D) and downregulation of Mafa mRNA (Fig. 8E) and Mafa protein levels (Fig. 8F).