Mouse insulinoma 6 (MIN6) cell lines (passages 15 to 25) were cultured at 37℃ in a humidified incubator with 5% CO₂ in Dulbecco's Modified Eagle's medium (DMEM) containing 4.5 g/L glucose supplemented with 10% heat-inactivated fetal bovine serum, 100 IU/mL penicillin, 50 µg/mL streptomycin, and 50 µM 2-mercaptoethanol (all from Life Technologies, Paisley, UK). At 50% to 70% confluence, the cells were incubated with or without 10 µM pioglitazone (Takara Pharmaceuticals, Kusatsu, Japan) and/or 0.5 mM palmitic acid (R&D Systems Inc., Minneapolis, MN, USA) in different glucose concentrations (1 or 4.5 g/L) for 48 hours. A group of cells was treated with 10 µg/mL lipopolysaccharide (LPS) and/or 10 µM pioglitazone, and another group was treated with 2 µg/mL tunicamycin with or without pioglitazone for 48 hours.

Male KK-Ay mice 8 weeks of age were obtained from Oriental Bio Laboratory (Seoul, Korea) and bred in standard conditions under a 12-hour light/dark cycle. All experiments using animals were performed in accordance with the Ethics Committee for Animal Experiments of the Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine. The experimental mice were placed on a high-fat (HF) diet (fat: 45% of energy content [39.4% from lard and 4.4 from soybean oil]; protein: 20% of energy content; and carbohydrate: 35% of energy content) or a HF diet containing 0.02% (wt/wt) pioglitazone for 6 weeks. Body weight and blood glucose levels were checked every week.

A stock solution was made by dissolving palmitate (R&D Systems Inc.) with 50% ethanol to 100 mM. This was diluted in serum-free DMEM containing 1% immunoglobulin G and fatty acid-free bovine serum albumin (BSA) fraction V (GIBCO, Grand Island, NY, USA) to 500 µM and then mixed at 40℃. Stock solutions were kept at −20℃ and used for treatment until they had been stored for 6 months. BSA solution in serum-free DMEM medium was used as the vehicle control.

At 60% to 70% confluence, MIN6 cells were pre-incubated in glucose-free Krebs-Ringer bicarbonate HEPES (KRPH) buffer for 60 minutes and treated with 1 or 4.5 g/L glucose in DMEM for 60 minutes. The dose of insulin released in the medium was measured with a mouse insulin enzyme-linked immunosorbent assay (Linco Research, St Charles, MO, USA).

Cellular proteins were prepared by a RIPA lysis buffer (Santa Cruz Biotechnology, Santa Cruz, CA, USA) with protease and phosphatase inhibitors. The concentration of protein was measured by the Bradford assay (Amersham Life Science, Arlington Height, IL, USA), and 20 µg of protein from each group was separated in 4% to 12% SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) gels and transferred to polyvinylidene difluoride membranes. Proteins in the membranes were probed with primary antibodies for p-eIF2α (Ser51), IRE1α (14C10), CHOP (L63F7), phosphor-NF-κB p65 (S536), NF-κB, phosphor-stress-activated protein kinase (SAPK)/JNK (T183/Y185), SAPK/JNK, cleaved caspase-3 (D175), total caspase-3, β-actin (Cell Signaling Technology, Danvers, MA, USA), ATF6 (Abcam, Cambridge, MA, USA), and GRP78 (Santa Cruz Biotechnology) for 24 hours. The membranes were then exposed to secondary antibodies conjugated with horseradish peroxidase, and visualized using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, Rockford, IL, USA). ImageJ, a program for quantification, was used to measure the density of the scanned immunoblot bands. The density of phosphorylated NF-κB, SAPK/JNK, and cleaved caspase-3 was normalized to the levels of NF-κB, SAPK/JNK, and caspase-3, respectively.

Pancreatic tissue samples were fixed with 4% paraformaldehyde for 24 hours and embedded within paraffin. Paraffin sections of 7 µm were dewaxed in xylene, and incubated with 0.5% hydrogen peroxide in methanol for 0.5 hours to block endogenous peroxidase activity. After washing with phosphate-buffered saline, tissue samples were blocked using normal goat serum for 0.5 hours, and then incubated with primary antibodies for insulin (Millipore, Billerica, MA, USA), glucagon, Glut2, monocyte chemoattractant protein-1 (MCP1), ATF6α, and GRP78 (Santa Cruz Biotechnology) for 24 hours. The tissue samples were then incubated with secondary antibodies conjugated with fluorescent dye (Thermo Fisher Scientific) for 1 hour. DAPI (4′,6-diamidino-2-phenylindole) solution (Invitrogen, Carlsbad, CA, USA) was used to visualize the nucleus, and fluorescent images were obtained using an Olympus BX52 microscope (Olympus Optical Co., Tokyo, Japan).

At 60% to 70% confluence, total RNA was extracted using TRizol (Life Technologies, Carlsbad, CA, USA). After dose measurement, 2 µg of RNA was used to synthesize the first strand of cDNA using the RT enzyme Mix (Thermo Fisher Scientific), and it was amplified by SYBR green I Master (Roche Diagnostics, Indianapolis, IN, USA) and the following primers: TNFA, sense 5′-GTTCTATGGCCCAGACCCTCAC-3′ and antisense 5′-GGCACCACTAGTTGGTTGTCTTTG-3′ (NM-013693, 144 bp); IL6, sense 5′-CCACTTCACAAGTCGGAGGCTTA-3′ and antisense 5′-GCAAGTGCATCATCGTTGTTCATAC-3′ (NM-031168, 115 bp); IL1B, sense 5′-AGGTCATCACTATTGGCAAC-3′ and antisense 5′-ACTCATCGTACTCCTGCTTG-3′ (NM-008361, 135 bp). Polymerase chain reaction (PCR) was performed in a Light Cycler 480 real-time PCR machine (Roche, Lewis, UK) with a pre-denaturation, denaturation, and annealing step, followed by additional 40 cycles and an extension step. The emission levels of green fluorescence by SYBR were detected and the relative expression levels of RNA were determined using the ΔΔCt method.

Poly (adenosine diphosphate [ADP]-ribose) polymerase (PARP) activity was measured using a universal colorimetric PARP assay kit (Trevigen, Gaithersburg, MD, USA). The proteins extracted from MIN6 cells were loaded onto a histone-coated 96-well plate, and incubated with biotinylated poly (ADP-ribose) for 1 hour. After treatment with horseradish peroxidase and colorimetric substrates, absorbance was read at 450 nm in a spectrophotometer.

PASW version 17 (SPSS Inc., Chicago, IL, USA), a statistical software program, was used for all statistical analyses in the present study, and one-way analysis of variance was used to evaluate the statistical differences among mean values in different experimental groups. Differences with a P value <0.05 were regarded as statistically significant changes.

To study the role of pioglitazone in β-cells, MIN6 cells were incubated with palmitate and/or pioglitazone. Treatment with 0.5 mM palmitate significantly inhibited 25 mM glucose-stimulated insulin secretion (1% BSA vs. 500 µM palmitate/1% BSA, P<0.001), while insulin levels were significantly restored by pioglitazone treatment for 24 hours (500 µM palmitate/1% BSA vs. 10 nM pioglitazone/500 µM palmitate/1% BSA, P<0.001) (Fig. 1A). Pioglitazone treatment also attenuated apoptosis, with a significant decrease of cleaved caspase-3 and PARP activity observed in MIN6 cells treated with high glucose and palmitate (Fig. 1B, C).

As it has been reported that inflammation and ER stress are major causes of apoptosis and dysfunction in pancreatic β-cells [16,17], we measured the inflammatory response and ER stress. To examine whether the protective effect of pioglitazone against free fatty acid-induced β-cell dysfunction was related to the inflammatory response and ER stress, we measured the expression of cytokines and proteins related to ER stress. The extreme increase in the expression of TNFA, IL6, and IL1B induced by palmitate was significantly ameliorated by co-treatment with pioglitazone (Fig. 1D). Treatment with palmitate induced the expression of specific markers of the UPR pathways responding to ER stress, including p-eIF2α, GRP78, ATF6α, and CHOP, and the expression of these proteins was significantly reduced by pioglitazone treatment (Fig. 1E, F). These results suggest that pioglitazone inhibits palmitate-induced inflammatory response and ER stress signaling, contributing to protection from β-cell dysfunction.

As shown in Fig. 1, pioglitazone simultaneously repressed the palmitate-induced inflammatory response and reduced ER stress. Therefore, we evaluated the role of pioglitazone in MIN6 cells in which an inflammatory response or ER stress was induced. First, MIN6 cells were treated with 10 µg/mL LPS for 48 hours to induce an inflammatory response. In LPS-treated cells, increased phosphorylation of NF-κB and JNK was detected, and the expression of TNFA, IL6, and IL1B was elevated, meaning that LPS induced a major inflammatory response in the MIN6 cells. The induction of the inflammatory response by LPS also increased the expression of ER stress markers, such as p-eIF2α, GRP78, and CHOP. These data suggest that the inflammatory response could induce ER stress in MIN6 cells. Treatment with pioglitazone remarkably repressed the inflammatory response induced by LPS, with diminished phosphorylation of NF-κB and JNK, and reduced expression of the mRNA of inflammatory cytokines (Fig. 2A-C). Consistently, markers of ER stress were also significantly decreased by pioglitazone (Fig. 2D, E). These data suggest that pioglitazone can repress the LPS-induced inflammatory response and ER stress in pancreatic β-cells.

We next administered tunicamycin, an inducer of ER stress, to evaluate the effects of pioglitazone on the regulation of ER stress. Pioglitazone significantly repressed the expression of ER stress-related proteins, such as p-eIF2α, GRP78, cleaved-ATF6, and CHOP (Fig. 3A, B). Treatment with tunicamycin resulted in a slight increase in the expression of TNFA and IL1B, but not IL6 (Fig. 3C), while pioglitazone repressed only IL1B expression, indicating that the induction of the inflammatory response might be upstream of ER stress in β-cell dysfunction and that pioglitazone represses both the inflammatory response and ER stress in pancreatic β-cells.

We next employed morphological investigations of pancreatic islet samples from KK-Ay mice, which were fed HF diets with or without pioglitazone, to explore the effects of pioglitazone on inflammation and ER stress. In the pancreas of HF-fed mice without pioglitazone, the expression of ATF6 and GRP78 was strongly detected in the islet, showing co-localization with insulin. However, the expression of ATF6 and GRP78 was remarkably decreased by pioglitazone, providing evidence that pioglitazone repressed ER stress in the pancreatic islets (Fig. 4A). Treatment with pioglitazone also repressed the inflammatory response in the pancreatic islets, as a significant decrease of MCP1 was observed. Additionally, α-cell infiltration into the core of the pancreatic islet was decreased by pioglitazone (Fig. 4B). In contrast, GLUT2, the main sensor of glucose in β-cells, was significantly increased by pioglitazone treatment (Fig. 4C). These results indicated that pioglitazone treatment preserved β-cells against ER stress and the inflammatory response. Treatment with pioglitazone also attenuated the rise of blood glucose levels induced by the HF diet without altered body weight. In the 6th week, the blood glucose levels of the pioglitazone-treated mice were significantly lower than those in the HF group (254.50±3.33 mg/dL vs. 264.75±2.10 mg/dL, respectively). These data suggest that the preservation of pancreatic β-cell function by pioglitazone may play a meaningful role in the maintenance of blood glucose levels.