The β-TC6 mouse insulinoma cell line was purchased from the National Collection of Authenticated Cell Cultures. Cells were maintained in DMEM (Sigma-Aldrich; Merck KGaA) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.) at 37˚C and 5% CO₂. A stock solution of PA (Sigma-Aldrich; Merck KGaA) was prepared in 100% ethanol at 55˚C until it was completely dissolved at a concentration of 100 mM. The final ethanol concentration in culture did not exceed 0.1% (v/v), and vehicle controls were included accordingly. Prior to PA treatment, the stock solution was added to DMEM with 10% FBS at the indicated concentrations.

SC79 (Selleck Chemicals) was dissolved in DMSO to prepare a 50-mM stock solution stored at -20˚C. For experiments, the stock was diluted in DMEM to a final concentration of 5 or 10 µM, ensuring the DMSO concentration did not exceed 0.1%. β-TC6 cells at 70-80% confluency were treated with SC79 for 4 h at 37˚C, 5% CO₂. DMSO-treated cells were used as controls. Following treatment, cells were harvested for protein extraction or insulin secretion assays.

RNA-seq data (GSE53949) was downloaded from the GEO database (32) and analyzed in RStudio (version 1.4.1717; Posit PBC) using the DESeq2 package (version 1.30.1; https://bioconductor.org/packages/release/bioc/html/DESeq2.html). GSE53949 included five control and five PA-treated human pancreatic islets samples; all ten samples were used for differential and correlation analyses. Following initial quality control, normalization was performed to adjust for library size variations. Differential expression analysis identified genes with |log2FC|>1 and P<0.05 as significant, and P-values were adjusted using the Benjamini-Hochberg method to control for false discovery rate. Volcano plots were generated with ggplot2 (version 3.3.5; http://ggplot2.tidyverse.org), where upregulated genes were highlighted in red and downregulated genes in green. Functional enrichment, including Gene Ontology (http://geneontology.org) and Kyoto Encyclopedia of genes and genomes (KEGG; https://www.genome.jp/kegg/) pathway analysis, was conducted using the clusterProfiler package (version 4.0.5; http://bioconductor.org/packages/release/bioc/html/clusterProfiler.html), and results were visualized through bar plots. To assess the relationship between thioredoxin domain-containing protein 5 [TXNDC5 (ERp46)] and solute carrier family 2 member 2 [SLC2A2 (GLUT2)], Pearson's correlation analysis was conducted using normalized RNA-seq expression data. A linear regression model was applied, and the coefficient of determination (R²) was calculated to quantify the strength of the correlation. Scatter plots with fitted regression lines were generated to visualize the results.

Total RNA was extracted from β-TC6 cells using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). First-strand complementary DNA was synthesized using the Hifair^® II 1st Strand cDNA Synthesis Kit (Shanghai Yeasen Biotechnology Co., Ltd.), according to the manufacturer's instructions. Subsequently, RT-qPCR was performed using the Hieff^® qPCR SYBR Green Master Mix (Shanghai Yeasen Biotechnology Co., Ltd.) according to the standard method. The thermocycling conditions were as follows: Initial denaturation at 95˚C for 2 min, followed by 40 cycles of denaturation at 95˚C for 10 sec and annealing/extension at 60˚C for 30 sec. A melting curve analysis was performed to verify amplification specificity. The relative expression (fold) was calculated using the comparative method 2^-ΔΔCq method (33). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used for normalization. The primer sequences are shown in Table I.

The culture dish in which β-TC6 cells were seeded was washed twice with cold PBS to completely remove the culture medium, and the cells were lysed using RIPA lysis buffer (Beijing Solarbio Science & Technology Co., Ltd.) supplemented with protease inhibitors. After gently scraping the cells with a cell scraper, they were transferred to a 1.5 ml Eppendorf tube and incubated on ice for 30 min for lysis. The cells were vortexed every 5 min to enhance protein extraction. The lysate was centrifuged at 12,000 x g for 15 min at 4˚C to precipitate cell debris. The supernatant containing the extracted protein was carefully transferred to a new tube for subsequent analysis.

Protein concentration was determined using the bicinchoninic acid (BCA) assay (Thermo Fisher Scientific, Inc.). A standard curve was prepared using bovine serum albumin (Thermo Fisher Scientific, Inc.) as a standard. Protein samples were added to different wells of a 96-well plate. Subsequently, 200 µl of BCA reagent mixture (comprised of reagent A and reagent B mixed at a ratio of 50:1) was added to each well. The plate was incubated at 37˚C for 30 min, and absorbance was measured at 562 nm using a microplate reader. Protein concentration was calculated by comparing the absorbance of the sample with the standard curve. Depending on the concentration, protein samples were diluted to 1 µg/µl using RIPA and then added to protein loading buffer, boiled at 100˚C for 5 min, aliquoted and stored in a freezer at -80˚C.

Protein samples (50 µg) were separated by 10% SDS-PAGE and transferred to PVDF membranes (Wuhan Servicebio Technology Co., Ltd.). Following blocking with 5% skim milk in TBST for 1 h at room temperature, the membranes were incubated with primary antibodies overnight at 4˚C: Anti-ERp46 (1:1,000; cat. no. sc-271667; Santa Cruz Biotechnology, Inc.), anti-GLUT2 (1:1,000; cat. no. K006592P; Beijing Solarbio Science & Technology Co., Ltd.), anti-p-Akt (1:1,000; cat. no. GB150002), anti-Akt (1:1,000; cat. no. GB111114), anti-pancreatic and duodenal homeobox 1 (PDX1; 1:1,000; cat. no. GB11917), and anti-tubulin (1:2,000; cat. no. GB11017; all from Wuhan Servicebio Technology Co., Ltd.), anti-binding immunoglobulin protein (BiP; 1:1,000; cat. no. 3177T), C/EBP homologous protein (CHOP; 1:1,000; cat. no. 2895T), and anti-Na⁺/K⁺-ATPase [1:1,000; cat. no. 3010S; all from Cell Signaling Technology, Inc.]. Following washing three times with TBST (TBS containing 0.1% Tween-20) for 10 min each time, the membranes were incubated with secondary antibodies: Anti-rabbit IgG, HRP-linked antibody (cat. no. 7074) and anti-mouse IgG, HRP-linked antibody (cat. no. 7076) (both 1:5,000; Cell Signaling Technology, Inc.), for 1 h at room temperature. Protein bands were visualized using enhanced chemiluminescence (Thermo Fisher Scientific, Inc.), and band intensities were quantified using ImageJ software (version 1.53t; National Institutes of Health).

Supernatants of β-TC6 cells exposed to various concentrations (0, 0.1, 0.25 and 0.5 mM) of PA for 24 h at 37˚C were collected, and insulin levels were measured using a mouse insulin ELISA kit [cat. no. 634-01481 (formerly AKRIN-011T) FUJIFILM Wako Pure Chemical Corporation], according to the manufacturer's instructions. Absorbance was measured at 450 nm, and insulin concentrations were calculated using a standard curve.

For the knockdown of ERp46, 1x10⁶ cells were seeded into a 6-well plate. A total of 50 pmol ERp46 small interfering (siRNA) (sense, 5'-GUACUCGGUACGAGGUUAUTT-3' and antisense, 5'-AUAACCUCGUACCGAGUACTT-3') or negative control siRNA (sense, 5'-UUCUCCGAACGUGUCACGUTT-3' and antisense, 5'-ACGUGACACGUUCGGAGAATT-3') were transfected using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions (at 3˚C for 6 h; typically 5 µl reagent per well in Opti-MEM). The transfection medium was then replaced with DMEM containing 12% FBS, and cells were cultured for 24 h at 37˚C. The expression of ERp46 was detected using western blotting and RT-qPCR.

Statistical analysis was performed using GraphPad Prism 8.0 (GraphPad Software, Inc.). Data are presented as the mean ± standard error of the mean from three to five independent experiments, with each experiment including n=6 biological replicates per group. Statistical comparisons between two groups were performed using an unpaired two-tailed Student's t-test, and multiple-group comparisons were analyzed using ANOVA followed by Sidak's multiple-comparisons test. P<0.05 was considered to indicate a statistically significant difference.

As aforementioned, FFAs can cause lipotoxicity in pancreatic β-cells (10,15,19), and it was found that following PA stimulation, pancreatic β-cells secreted less insulin, which was proportionate to the dose of PA (Fig. 1A), consistent with previous studies (16,17). When stimulated with 0.5 µM PA, insulin secretion decreased over time; however, there was no significant difference between 24 and 48 h, suggesting that the effect reached saturation (Fig. 1A). To further investigate the cause of the reduced insulin secretion, RNA-seq analysis of a public dataset (GSE53949) was performed. By analyzing the RNA-seq results of PA-stimulated β-cells compared with the control group, TXNDC5 (gene name of ERp46) was significantly upregulated (Fig. 1B). This upregulation was supported by enrichment analysis, which highlighted the activation of ER stress-related pathways in β-cells following PA stimulation (Fig. 1C). These findings were consistent with previous reports, in which the upregulation of TXNDC5 was associated with its role in promoting correct protein folding and maintaining redox homeostasis under ER stress conditions (25-27).

By contrast, the key glucose transporter SLC2A2 (gene name of GLUT2) in β-cells was significantly downregulated following PA stimulation. Enrichment analysis identified the ‘negative regulation of transport’ pathway (Fig. 1D), highlighting the importance of GLUT2 in β-cell function. The downregulation of GLUT2 suggested a reduced glucose-transport capacity and was compatible with the decreased insulin secretion observed in Fig. 1A. Previous studies have demonstrated that GLUT2 plays a critical role in glucose sensing and insulin secretion in β-cells (8,9), although the present data did not directly establish causality. Consistently, RT-qPCR confirmed that SLC2A2 mRNA was decreased while TXNDC5 mRNA was increased in PA-treated cells (Fig. 1E). Western blotting further demonstrated decreased GLUT2 protein and increased ERp46 protein levels following PA treatment (Fig. 1F). In conclusion, PA stimulation was associated with the activation of ER stress-related pathways and reduced GLUT2 expression and insulin secretion in β-cells.

ERp46 knockdown efficiency was first verified. Western blotting confirmed a robust reduction of ERp46 protein following siERp46 transfection (Fig. 2A). This result validated the effectiveness of the knockdown system for subsequent experiments. Next, GLUT2 protein was examined across the four conditions (Control, PA, siERp46 and PA + siERp46). As shown in Fig. 2B, PA alone decreased GLUT2, whereas siERp46 alone had no detectable effect. Of note, the combination of siERp46 with PA further reduced GLUT2 compared with PA alone. These findings indicated that ERp46 is not essential for basal GLUT2 expression, but contributes to the maintenance of GLUT2 under lipotoxic stress.

Functionally, insulin secretion followed the same pattern (Fig. 2C): PA reduced insulin release; siERp46 alone did not alter secretion; and siERp46 plus PA led to a further decrease relative to PA alone. This indicates that the protective role of ERp46 becomes evident only in the presence of PA-induced stress. Since ERp46 is an ER oxidoreductase, ER stress markers were assessed. PA increased BiP and CHOP, and siERp46 further increased their levels under PA, whereas siERp46 alone showed no significant change (Fig. 2D). These results are consistent with the interpretation that ERp46 depletion aggravates ER stress in β-cells exposed to PA. PDX1, a β-cell identity/functional factor (12,30), was also evaluated. PA decreased PDX1 mRNA and protein levels, while siERp46 did not produce an additional significant reduction beyond PA in our conditions (Fig. 2E and F). This suggested that ERp46 depletion primarily impacted GLUT2 and insulin secretion rather than β-cell identity markers under PA stress.

Collectively, these data indicated that ERp46 depletion exacerbates PA-induced ER stress and is associated with a greater loss of GLUT2 and insulin secretion, whereas ERp46 knockdown alone has minimal effects at baseline. Thus, ERp46 appears to play a stress-dependent protective role in sustaining GLUT2 expression and β-cell functionality.

The relationship between TXNDC5 (ERp46) and SLC2A2 (GLUT2) was first examined using the public RNA-seq dataset (GSE53949). In the correlation plot, each dot represents one RNA-seq sample from GSE53949 (PA-stimulated or control β-cell samples). A modest positive correlation was observed (R²=0.4001, P=0.0497; Fig. 3A), supporting an association between ERp46 and GLUT2 under lipotoxic conditions. Next, AKT activation was assessed. Total AKT (t-AKT) remained unchanged across groups, whereas PA decreased phosphorylated AKT (p-AKT). Of note, siERp46 alone had little effect, but siERp46 in the presence of PA further reduced p-AKT (Fig. 3B). These findings indicated that ERp46 depletion exacerbated the PA-induced suppression of AKT phosphorylation, while having a minimal impact under basal conditions.

To determine whether AKT activation is sufficient to counteract the PA/siERp46 effects, cells were treated with the AKT activator SC79. SC79 increased p-AKT and partially restored GLUT2 protein in PA-exposed cells under both control siRNA and siERp46 conditions (Fig. 3C). Functionally, SC79 also improved insulin secretion that was reduced by PA and further diminished by siERp46 (Fig. 3D). To evaluate specificity, an additional membrane protein and ER-stress markers were probed. Na⁺/K⁺-ATPase was selected as a representative plasma membrane housekeeping protein whose expression remains stable under metabolic or ER stress conditions, serving as a control for nonspecific changes in membrane protein abundance. Na⁺/K⁺-ATPase levels were unchanged across treatments, and SC79 did not reduce PA-induced increases of BiP or CHOP (Fig. 3E), indicating that AKT activation does not broadly elevate membrane proteins or alleviate ER stress but can still restore GLUT2 and insulin secretion.

In conclusion, ERp46 depletion was demonstrated to enhance the PA-induced loss of p-AKT, GLUT2 and insulin secretion, while pharmacological AKT activation rescued GLUT2 and insulin without reducing ER-stress markers. These findings supported a model in which AKT acts downstream or in parallel to ER-stress pathways, with ERp46 helping to preserve AKT activity and β-cell function specifically under lipotoxic stress.