Loss of β‐cell identity in human islets treated with glibenclamide

Abstract

Aims

Loss of β‐cell identity can contribute to the reduction of functional β‐cell mass in type 2 diabetes. Sulfonylureas show shorter durability of antihyperglycaemic action and higher rates of secondary failure compared to other antihyperglycaemic agents, suggesting that they could accelerate the decline of β‐cell functional mass in type 2 diabetes. We aimed to investigate the impact of chronic exposure to sulfonylureas on β‐cell identity.

Materials and Methods

Islets from human multi‐organ donors were cultured for 4–7 days at 5.6 mM glucose with or without glibenclamide. β‐cell function (glucose stimulated insulin secretion, GSIS), apoptosis (TUNEL) and gene (RT‐qPCR) and protein expression (immunofluorescence, genetic β‐cell tracing and Western Blot) were determined.

Results

Human islets exposed to glibenclamide showed increased insulin secretion at low glucose, reduced GSIS, increased apoptosis, endoplasmic reticulum (ER) stress, and loss of β‐cell identity indicated by reduced gene and protein expression of key β‐cell identity markers and insulin. There were no changes in the expression of disallowed or progenitor‐related genes. Genetic β‐cell tracing showed a similar percentage of insulin‐expressing cells in control and sulfonylurea‐treated islets. Addition of the chemical chaperone 4‐phenylbutyrate (PBA) to the culture medium prevented glibenclamide‐induced ER stress and the downregulation of key β‐cell transcription factors, indicating that ER stress mediates, at least partially, the negative effects of glibenclamide on β‐cell identity.

Conclusions

Chronic exposure of human islets to glibenclamide induced the loss of β‐cell identity, which was mediated by ER stress, impaired β‐cell function, and increased β‐cell apoptosis. These negative effects of glibenclamide may contribute to the secondary failure of sulfonylureas and accelerate the decline of functional β‐cell mass in patients with type 2 diabetes.

1. INTRODUCTION

Inadequate β‐cell function and mass to meet the demands of increased insulin resistance lead to the development of type 2 diabetes. ¹ Increased β‐cell apoptosis was considered the main driver of β‐cell loss in type 2 diabetes, ² but more recently reduced β‐cell identity, dedifferentiation, and transdifferentiation, all of them potentially reversible processes, have been pointed out as contributing to reducing the functional β‐cell mass. ³ Loss of beta cell identity, characterised by the reduced expression of genes and proteins essential to β‐cells, has been identified in islets from patients with type 2 diabetes. ³ , ⁴ , ⁵ , ⁶ Beta cell dedifferentiation, involving also the expression of disallowed genes and the conversion to a progenitor‐like state, ³ , ⁴ has been clearly identified in rodent models ⁷ and has been described in human islets, ⁸ , ⁹ although its role in human type 2 diabetes is less well established. ³ , ¹⁰

Sulfonylureas have been used in the treatment of type 2 diabetes since the early 1950s and are highly efficacious in reducing blood glucose. ¹¹ Although the newer antidiabetic agents can be a preferable option for many patients, sulfonylureas remain widely prescribed worldwide. ¹² Sulfonylureas stimulate insulin secretion through the interaction with the sulfonylurea receptor subunit of the ATP‐sensitive potassium (K_ATP) channels of β‐cells. ¹³ The closure of the K_ATP channels causes β‐cell depolarization, which triggers the opening of the voltage‐dependent Ca²⁺ channels and the rise in intracellular calcium [Ca²⁺]_i that stimulates the exocytosis of insulin granules. ¹⁴ However, sulfonylureas show shorter durability of glycemic control and higher rates of secondary failure than other antihyperglycemic agents. ¹⁵ , ¹⁶ , ¹⁷ Several mechanisms may contribute to the failure of sulfonylurea treatment, among them the depletion of insulin stores ¹⁸ and increased β‐cell death. ¹⁹ , ²⁰ , ²¹ Recently, Cyb5r3, a component of the mitochondrial electron transfer system, was identified as a key factor in the secondary failure to sulfonylureas. ²² , ²³ The role of ER stress in the failure to sulfonylurea treatment has been investigated in rodent cell lines with conflicting results. ²⁴ , ²⁵ Chronic administration of glibenclamide was recently suggested to induce partial β‐cell dedifferentiation in mice, ²³ but the impact of sulfonylureas on β‐cell identity has not been explored in human islets. In this study, we have investigated the effects of chronic exposure to the sulfonylurea glibenclamide on β‐cell identity and ER stress in human islets.

2. MATERIALS AND METHODS

A detailed description of materials and methods used in the study is provided as Supporting Information.

2.1. Human islet isolation and culture

Human pancreatic islets were isolated from non‐diabetic adult deceased organ donors, 42.3% female, 56.9 ± 1.80 years (range: 33–73) as previously described. ²⁶ Donor and islet isolation characteristics are provided in Table S1. Use of human islets was approved by our local Ethics Committee (identification number: PR037/17). Signed consent was obtained from relatives of all organ donors.

2.2. Experimental culture conditions

Human islets were cultured for 4 or 7 days in complete CMRL‐1066 medium at 5.6 mM glucose (control condition) with Glibenclamide (Sigma, G0639) (1 μM), or Glibenclamide (1 μM) and the chemical chaperone 4‐phenylbutyrate (PBA) (Sigma, SML0309) (2.5 mM). Culture medium was changed every 48 h. To avoid the confounding effects of high glucose on the assessment of glibenclamide effects, we used a glucose concentration of 5.6 mM known to be optimal for human islet cell survival in vitro. At the end of the culture period, islets were collected to determine glucose‐stimulated insulin secretion (GSIS) or processed for immunohistochemistry, RNA extraction, or protein extraction.

2.3. Glucose‐stimulated insulin secretion

Groups of 20 islets were preincubated in fresh KRBH buffer containing 2.8 mM glucose for 1 h at 37°C and then sequentially incubated with KRBH buffer containing 2.8 and 20 mM glucose for an additional hour. Supernatants were collected and stored at −80°C until assayed for insulin quantification by ELISA (Mercodia AB, Uppsala, Sweden). Insulin secretion was expressed as a percentage of insulin content normalized to DNA. The insulin stimulation index was calculated as the secreted insulin ratio between stimulated (20 mM) and basal (2.8 mM) glucose concentrations.

2.4. Insulin and DNA content

After the GSIS assay, islets were sonicated in 1 mL PBS. For insulin content, an aliquot of the homogenate was extracted with acid‐ethanol solution and stored at −80°C until insulin quantification. For DNA content, the remaining homogenate was stored at −80°C until measured by fluorimetry with Hoechst 33258 (Sigma‐Aldrich) on a fluorescence spectrophotometer (FLUOstar Omega, BMG Labtech, Ortenberg, Germany).

2.5. β‐cell apoptosis

β‐cell apoptosis was determined with the TUNEL technique (ApopTag Plus Fluorescein In Situ Apoptosis Detection Kit; Merck Millipore, Spain). Sections were immunostained with insulin antibody and nuclei labelled with DAPI. β‐cell apoptosis was expressed as a percentage of nuclear TUNEL‐positive β‐cells. A minimum of 1000 cells per sample was counted.

2.6. Quantitative PCR

Reactions were performed using TaqMan Gene Expression Assays and TaqMan Gene Expression Master Mix (Applied Biosystems, Foster, CA, USA) following the manufacturer's protocol. The full listing of assays, gene names, and assay identification numbers is shown in Table S2. Relative quantities (RQ) were calculated using the software Gene Expression Suite v1.0.3 (Applied Biosystems) using the 2^−ΔΔCT analysis method with 60S acidic ribosomal protein P0 (RPLP0) and TATA‐box binding protein (TBP) as endogenous control genes.

2.7. Immunofluorescence intensity

Sections of paraffin embedded islets were double stained for insulin and NKX6.1, NKX2.2 or MAFA. DAPI staining was used to identify cell nuclei. Images of triple‐labelled sections were acquired using a Carl Zeiss LSM 880 spectral confocal laser scanning microscope (Carl Zeiss Microscopy GmbH, Jena, Germany). Fiji ImageJ software (version 2.1.0) was used to analyse the percentage of β‐cells with nuclear labelling of NKX6.1, NKX2.2, and MAFA, and for fluorescence intensity quantification. Cell counts for positive markers were performed manually. Regions of Interest (ROIs) were defined based on insulin staining and mean fluorescence intensity of target molecules was measured within each ROI. For high‐throughput analysis, Fiji macros were used to automate intensity extraction across multiple samples, ensuring reproducibility. Results were expressed in arbitrary units.

2.8. Western blot assay

Human islets were disaggregated by gentle continuous pipetting, subjected to SDS‐PAGE and transferred onto PVDF membranes. Blots were blocked and incubated overnight at 4°C with specific primary antibodies, and HRP‐conjugated goat anti‐mouse/rabbit secondary antibodies were used for developing blots (full listing of antibodies is provided in Table S3). Images were acquired using the Amersham Imager 680 system. Target protein levels (normalized to ACTIN) were quantified by densitometric analysis using ImageJ software.

2.9. β‐cell tracing

Human islets were infected with two lentiviral vectors, pTrip‐RIP 400‐nlsCre (provided by Dr. Shimon Efrat, Sackler Faculty of Medicine, Tel Aviv University, Israel) in which the expression of Cre recombinase was driven by the rat insulin promoter, and pTrip‐CMV‐loxP‐Puro‐loxP‐eGFP, in which the expression of eGFP was conditional on the removal of the floxed puromycin gene by the action of the Cre recombinase. The expression of the whole construct was constitutively driven by the CMV promoter. Infected islets were cultured for 4 days to allow the expression of GFP so β‐cells became permanently labelled. Then, islets were exposed to 1 μM glibenclamide for 7 days, fixed overnight in 4% PFA‐PBS (Merck KgaA) and processed for paraffin embedding. Sections were double labelled for GFP and for insulin, and the preparations were visualised by immunofluorescence. β‐cell dedifferentiation was expressed as a percentage of GFP‐positive insulin‐negative cells. A minimum of 1000 cells per sample was counted.

2.10. Statistical analysis

Data are expressed as mean ± SEM. Statistical analysis were performed using GraphPad Prism 9 software. Normality for all variables was tested using the Kolmogorov–Smirnov test or the Shapiro–Wilk test as appropriate. Statistical significance was determined with the unpaired Student t test or multiple t test with Bonferroni–Dunn correction as appropriate. A p value <0.05 (Student t test) or <0.025 (multiple t test with Bonferroni–Dunn correction) was considered significant.

3. RESULTS

3.1. Impaired β‐cell function and increased apoptosis in human islets cultured with glibenclamide

Islets exposed to glibenclamide for 1 week showed impaired GSIS, as indicated by the lower stimulation index (control: 6.39 ± 1.31; glibenclamide: 1.80 ± 0.22, p = 0.004) (Figure 1A–C). The lower stimulation index was accounted for by the combination of a significantly increased basal insulin secretion at 2.8 mM glucose (control: 1.13% ± 0.21%; glibenclamide: 2.71% ± 0.66%, p = 0.04) and a non‐significant reduction in stimulated insulin secretion at 20 mM glucose (control: 8.34% ± 2.86%; glibenclamide: 5.19% ± 1.8%). Insulin content was lower in glibenclamide‐exposed islets than in control islets (31.7% reduction vs. control islets, p < 0.01) (Figure 1D). DNA content was similar in glibenclamide and control islets (control: 0.61 ± 0.1 μg DNA; glibenclamide: 0.54 ± 0.09 μg DNA).

FIGURE 1.

β‐cell function and β‐cell apoptosis in islets exposed to glibenclamide. After 7 days in culture at 5.6 mM glucose with or without glibenclamide (Glib) islets were sequentially incubated at 2.8 and 20 mM glucose for 1 h (n = 8). (A) Stimulation index expressed as ratio between stimulated (20 mM glucose) and basal (2.8 mM glucose) insulin secretion. (B) Insulin secretion in control and glibenclamide‐exposed islets at 2.8 mM glucose and 20 mM glucose normalized by DNA content (C) Insulin secretion in control and glibenclamide‐exposed islets at 2.8 mM glucose and 20 mM glucose normalized by insulin content. (D) Insulin content normalized to DNA content. (E, F) Beta cell apoptosis after 7 days in culture with or without (control) glibenclamide. (E) Representative image of triple immunofluorescence for TUNEL (green), insulin (red) and DAPI (blue). Scale bar = 100 μm. White arrows indicate insulin⁺TUNEL⁺ cells. (F) Quantification of β‐cell apoptosis (n = 4). Each dot represents an individual experiment, with each experiment conducted using human islets from a different donor. Data are mean ± SEM. *p < 0.05; **p < 0.01 versus control group by Student's t test.

Glibenclamide had a detrimental effect on β‐cell viability. β‐cell apoptosis was significantly increased after 1 week of exposure to glibenclamide (control: 0.75% ± 0.05%; glibenclamide: 1.49% ± 0.24%, p = 0.02) (Figure 1E,F).

3.2. Loss of β‐cell identity in glibenclamide‐treated islets

The expression of insulin (INS), β‐cell specific transcription factors NKX6.1, NKX2.2, PAX6, MAFA, MAFB, PDX1, FOXO1, prohormone convertase 2 (PC2; gene PCSK2) involved in the processing of mature insulin, KCNJ11 encoding the Kir6.2 subunit of the K_ATP channel and β‐cell marker GLP1R was significantly reduced after 7 days of treatment with glibenclamide (Figure 2). Gene expression of prohormone convertase 1/3 (PC1/3; gene PCSK1) was significantly increased. The expression of INS, NKX6.1, MAFA, PDX1, FOXO1, GLP1R and KCNJ11 was already significantly reduced after 4 days of exposure to glibenclamide (Figure 5E). No differences were observed in the expression of the disallowed genes HK1, HK3 and LDHA or the endocrine progenitor‐related genes ALDH1A3, NGN3 and SOX9. Gene expression of the α‐cell‐specific marker ARX was significantly reduced in glibenclamide‐treated islets and the glucagon gene (GCG) showed a tendency to a reduced expression.

FIGURE 2.

Gene expression in islets exposed to glibenclamide. mRNA expression is shown as ratio between islets cultured with 1 μM glibenclamide for 7 days over control islets on a log2 scale. ND, non‐detectable. Each dot represents an individual experiment, with each experiment conducted using human islets from a different donor. Data are mean ± SEM (n = 5–6), *p < 0.05; **p < 0.01; ***p < 0.001 vs control group by Student's t test.

FIGURE 5.

Endoplasmic reticulum (ER) stress and β‐cell identity in human islets exposed to glibenclamide. (A) Expression of ER stress genes in human islets after 4 and 7 days in culture with 1 μM glibenclamide (Glib) or glibenclamide and the chaperone 4‐phenybutyrate (PBA) (Glib + PBA). Data are shown as ratio between the expression of each gene in islets cultured with glibenclamide or glibenclamide+PBA over control islets on a log2 scale (n = 7–10). (B) Representative Western Blot of XBP1s and CHOP in human islets cultured with glibenclamide, PBA or glibenclamide + PBA for 4 days. (C, D) Quantification of protein expression of CHOP (C) and XBP1s (D) in human islets cultured with glibenclamide, PBA or glibenclamide+PBA over control islets on a log2 scale (n = 5–6). (E) Expression of β‐cell identity genes in human islets after 4 and 7 days in culture (n = 6–8). Data are displayed as ratio between the expression of each gene in islets cultured with glibenclamide (Glib) or glibenclamide + PBA (Glib + PBA) over control islets on a log2 scale. (F) Fluorescence intensity of insulin, nuclear NKX6.1 and nuclear MAFA in insulin⁺ β‐cells (n = 5). Data are shown as ratio between the fluorescence intensity for each immunolabelled protein in islets cultured for 7 days with glibenclamide (Glib) or glibenclamide + PBA (Glib + PBA) over control islets on a log2 scale. Each dot represents an individual experiment, with each experiment conducted using human islets from a different donor. Data are mean ± SEM. *p < 0.025; **p < 0.01; ***p < 0.001 versus control group; #p < 0.01 Glib versus Glip + PBA.

To further explore the loss of β‐cell identity, we analysed the expression of insulin and several key β‐cell transcription factors at the protein level. The percentage of INS⁺, NKX6.1⁺, MAFA⁺ and NKX2.2⁺ cells was similar in glibenclamide‐treated and control islets (Table S4). In agreement with the reduced expression of the insulin gene and insulin content, the fluorescence intensity of insulin was significantly lower in glibenclamide‐treated islets (Figure 3). The fluorescence intensity of nuclear NKX6.1 and MAFA in INS⁺ β‐cells was also significantly reduced in glibenclamide‐treated islets, confirming the loss of β‐cell identity. No differences were observed in the nuclear intensity of NKX2.2 in INS⁺ β‐cells. The percentage of INS⁺NKX6.1⁻, INS⁺MAFA⁻ and INS⁺NKX2.2⁻ cells was not statistically different in control and glibenclamide‐treated islets (Table S5). However, there was a numerical increment in the percentage of INS⁺NKX6.1⁻ and INS⁺MAFA⁻ cells in line with the reduced intensity of NKX6.1 and MAFA staining in glibenclamide‐treated islets.

FIGURE 3.

Protein expression of insulin, NKX6.1, MAFA and NKX2.2 in β‐cells of islets exposed to glibenclamide. (A) Representative image of triple immunofluorescence for insulin (white), transcription factors NKX6.1 MAFA or NKX2.2 (red) and DAPI (blue) in human islets cultured with 1 μM glibenclamide for 7 days. Scale bar = 100 μm. (B) Fluorescence intensity of insulin, nuclear NKX6.1, MAFA and NKX2.2 in insulin⁺ β‐cells shown as ratio between islets cultured with glibenclamide over control islets on a log2 scale. Each dot represents an individual experiment, with each experiment conducted using human islets from a different donor. Data are mean ± SEM (n = 4–5), *p < 0.05; **p < 0.01; ***p < 0.001 vs control group by Student's t test.

Genetic β‐cell tracing was used to further characterise the loss of insulin expression in islets exposed to glibenclamide (Figure 4A). The efficiency of the lineage tracing was similar in control and glibenclamide‐treated islets (Figure 4B). The percentage of traced cells expressing insulin (INS⁺GFP⁺/GFP⁺) was similar in control (78.9% ± 4.11%) and glibenclamide (84.7% ± 1.15%) groups (Figure 4C). Thus, the results indicate that the lower expression of insulin in glibenclamide‐treated islets was due to an overall reduction of insulin expression in β‐cells rather than to a complete loss of insulin in a dedifferentiated β‐cell subpopulation.

FIGURE 4.

Genetic β‐cell tracing of control and glibenclamide‐treated islets. (A) Representative image of triple immunofluorescence for insulin (white), GFP (green) and DAPI (blue). Scale bar = 100 μm. (B) Quantification of traced insulin⁺/GFP⁺ cells over total insulin⁺ β‐cells (C) Quantification of traced insulin⁺/GFP⁺ cells over total GFP⁺ cells. Each dot represents an individual experiment, with each experiment conducted using human islets from a different donor. Data are mean ± SEM (n = 5–6).

3.3. Glibenclamide induces ER stress

Islets exposed to glibenclamide showed increased ER stress. The expression of spliced XBP‐1 (XBP1s), an indicator of the unfolded protein sensor IRE1 activation, and CHOP (DDIT3) a downstream gene known to be induced by UPR, was increased in islets cultured with glibenclamide, and the expression of WFS1, whose absence leads to upregulated ER stress, was decreased (Figure 5A and Table S6). The results were confirmed at the protein level, which showed an increased expression of XBP1s and CHOP proteins in glibenclamide‐cultured islets (Figure 5B,C). No differences were found in ATF6 or EIF2AK3 mRNA levels (Figure S1). The expression of two well‐established markers of oxidative stress, catalase (CAT) and superoxide dismutase 2 (SOD2), was not modified in glibenclamide‐cultured islets (Figure S2).

3.4. Alleviation of ER stress prevents glibenclamide‐induced loss of β‐cell identity

ER‐stress has been associated with impaired β‐cell identity. ⁵ , ²⁷ To determine whether glibenclamide‐induced ER stress contributed to the loss of β‐cell identity, the chemical chaperone PBA was used to alleviate ER stress in glibenclamide‐exposed human islets. The addition of PBA to the culture medium prevented the upregulation of XBP1s and DDIT3 and the downregulation of WFS1 in glibenclamide‐cultured islets (Figure 5A and Table S6). This finding was confirmed at the protein level by the similar expression of XBP1s and CHOP in control and glibenclamide‐cultured islets incubated with PBA (Figure 5B–D).

The addition of PBA to the culture medium prevented the downregulation of the β‐cell identity markers NKX6.1, NKX2.2, PAX6, MAFA, PDX1, FOXO1 and KCNJ11 (Figure 5E and Table S6) and attenuated the downregulation of INS and GLP1R expression. The expression of NKX6.1, NKX2.2, FOXO1 and KCNJ11 genes was significantly reduced in the glibenclamide group compared to glibenclamide + PBA cultured islets. Thus, the alleviation of ER stress prevented β‐cell dedifferentiation in islets exposed to glibenclamide. In line with the effects of PBA on gene expression, the reduction of nuclear NKX6.1 and MAFA fluorescence intensity in β‐cells exposed to glibenclamide was also prevented by PBA, and that of insulin was attenuated (Figure 5F). Overall, these results indicate that ER stress contributes to the loss of β‐cell identity in glibenclamide‐cultured human islets.

4. DISCUSSION

In this study, we have found that human islets chronically exposed to glibenclamide show increased ER stress and loss of β‐cell identity. Prevention of ER stress attenuated the loss of β‐cell identity, indicating that ER stress mediates, at least partially, the negative effects of glibenclamide on β‐cell identity. We have also found that human islets exposed to glibenclamide show impaired β‐cell function and increased β‐cell apoptosis, confirming previous reports in rodent islets. The results provide a basis to better understand the mechanisms of secondary failure to sulfonylureas and their potential contribution to type 2 diabetes progression.

β‐cell function was impaired in glibenclamide‐treated human islets which showed a reduction in glucose‐stimulated insulin secretion. The impaired insulin secretion resulted from the combination of a significant increment in basal insulin secretion at low glucose (2.8 mM) and a tendency towards a reduced response to high glucose (20 mM), and is concordant with the characteristics of impaired β‐cell function previously identified in mouse islets exposed to sulfonylureas in vitro. ²⁸ Mice implanted with slow‐release glibenclamide pellets ²⁹ and mice with genetic suppression of K_ATP channels ³⁰ , ³¹ also show reduced insulin secretion in response to glucose. Thus, our results indicate that chronic exposure to glibenclamide leads to secretory dysfunction in human β‐cells and are consistent with previous data in rodent models.

β‐cell apoptosis has been reported in rodent and human β‐cells exposed to sulfonylureas ¹⁹ , ²⁰ , ²¹ although with some exceptions. ²⁴ Moreover, an upregulation of gene networks related to β‐cell death was identified in INS‐1 cells treated with glibenclamide or glimepiride. ³² β‐cell proliferation in adult human islets is extremely low and unlikely to be able to compensate for the doubling in apoptotic rate that we have found in islets exposed to glibenclamide. The results suggest that treatment with sulfonylureas could accelerate the decline of β‐cell mass in patients with type 2 diabetes.

The expression of key β‐cell identity markers was reduced in human islets exposed to glibenclamide, indicating the loss of β‐cell identity. ³ , ⁴ Gene expression of insulin, β‐cell transcription factors NKX6.1, MAFA, PDX1, and FOXO1, β‐cell genes GLP‐1R and KCNCJ1, encoding the Kir6.2 protein of K_ATP channel, was reduced after 4 days of exposure to glibenclamide. After 7 days, the reduction was more profound, and there were additional differences in the expression of transcription factors NKX2.2 and PAX6, suggesting that the deleterious effects of glibenclamide on β‐cell identity were time dependent. The expression of PCSK1 and PCSK2 genes, encoding the prohormone convertases PC1/3 and PC2 respectively and involved in the processing of mature islet hormones, showed opposite changes. Whether these changes translate into alterations in the processing of pro‐insulin or pro‐glucagon in islets exposed to glibenclamide remains to be addressed. The reduced expression of FOXO1 agrees with recent data showing the involvement of FOXO1 protein in secondary failure to glibenclamide treatment. ²² The reduced expression of the α‐cell genes ARX and GCG suggests that glibenclamide did not induce β‐ to α‐cell transdifferentiation. It may indicate, however, that chronic exposure to glibenclamide could have also a negative effect on α‐cell identity.

The expression of insulin, MAFA and NKX6.1 proteins was reduced in β‐cells of islets cultured with glibenclamide, confirming the loss of β‐cell identity. The reduced expression of transcription factors MAFA and NKX6.1 is of particular interest since they may play an important role in compromising β‐cell function in type 2 diabetes. It has been considered that the loss of MAFA represents an early indicator of β‐cell inactivity, and the subsequent deficit of NKX6.1 results in the overt β‐cell dysfunction associated with type 2 diabetes. ⁶ Similar to our results in glibenclamide‐treated islets, the expression of NKX2.2 was not reduced in patients with type 2 diabetes. ⁶ We propose that the reduced expression of β‐cell transcription factors regulating insulin gene expression can accelerate the decline of β‐cell function in patients treated with glibenclamide.

PERK, IRE1 and ATF6 pathways of the unfolded protein response (UPR) were activated in human islets exposed to glibenclamide, indicating ER stress. The expression of XBP1s and DDIT3 genes, and XBP1s and CHOP proteins was increased, and the expression of WFS1, the negative regulator of the ATF6 pathway, ³³ was reduced. WFS1 plays a critical role in maintaining ER and cytosolic Ca²⁺ homeostasis, and ablation of WFS1 depletes ER calcium, increases [(Ca²⁺)]_i and induces cell death in INS‐1 β‐cells. ³⁴ Activation of the IRE1α‐XBP1 pathway has a protective effect on β‐cells, but the sustained increment in XBP1s is associated with impaired glucose‐stimulated insulin secretion and insulin expression, and with increased β‐cell apoptosis. ³⁵ , ³⁶ The increased expression of pro‐apoptotic CHOP protein can also contribute to the higher β‐cell apoptosis that we have identified in glibenclamide‐treated islets.

Single‐cell transcriptomic analysis has recently identified an association between the activation of ER stress and the loss of key β‐cell transcription factors in human islets. ²⁷ To determine whether ER stress plays a role in glibenclamide‐induced loss of β‐cell identity we used the PBA chaperone to prevent ER stress. In the presence of PBA, the expression of UPR genes DDIT3, XBP1 and WFS1, as well as XBP1s and CHOP proteins, was similar in glibenclamide‐treated and non‐treated islets confirming the prevention of ER stress. The loss of β‐cell identity was also prevented in PBA‐cultured islets treated with glibenclamide which showed similar expression of key β‐cell transcription factors and β‐cell genes than control islets. However, insulin content remained low in glibenclamide‐PBA cultured islets, probably reflecting the β‐cell degranulation induced by the continuous stimulation of insulin secretion by glibenclamide. ¹⁸ Thus, our results indicate that the loss of β‐cell identity in human islets chronically exposed to glibenclamide is mediated, at least partly, by ER stress.

β‐cell dedifferentiation, defined by the regression to a progenitor state, ³ , ⁴ was not found in glibenclamide‐cultured islets that did not show upregulation of β‐cell disallowed genes nor progenitor genes. Moreover, we did not identify the presence of β‐cell dedifferentiation in genetically traced human islets treated with glibenclamide, which showed a similar percentage of GFP⁺/insulin⁻ cells to non‐treated islets. β‐cell dedifferentiation has been convincingly shown in rodent models, ³ , ⁷ but its relevance to human type 2 diabetes has not yet been so well established. β‐cell dedifferentiation was described in islets from patients with type 2 diabetes, ⁸ but its contribution to β‐cell mass reduction in type 2 diabetes was considered to be minimal. ¹⁰ Single cell transcriptomic analysis of human islets did not produce clear evidence of β‐cell dedifferentiation ³⁷ , ³⁸ and few studies have shown the conversion of human β‐cells to a progenitor‐like state. ⁸ , ⁹

This study has some limitations. First, the effects that we have observed in vitro cannot be directly extrapolated to the more complex in vivo situation in patients with type 2 diabetes treated with glibenclamide. Second, the concentration of glibenclamide that we used is similar to previous studies ²⁰ , ²¹ and close to the range found in the plasma of treated patients. ³⁹ However, sulfonylureas bind to plasma proteins and therefore the free drug concentration in the plasma could be lower. Third, changes in β‐cell characteristics of cultured human islets ⁴⁰ could have limited the identification of more profound effects of glibenclamide on β‐cell identity. Fourth, we have not explored the effects of glibenclamide in islets from donors with type 2 diabetes. However, the use of islets from donors with type 2 diabetes also faces limitations, such as reduced function and loss of β‐cell identity that can be regained once the islets are removed from the in vivo microenvironment and cultured in a “non‐diabetic” milieu. Fifth, the effects of glibenclamide on β‐cells may be different from those of other sulfonylureas. However, a study of gene regulation induced by glibenclamide, glimepiride, and gliclazide in β‐cells showed that differences in transcriptional regulation among the three drugs were small. ³⁹

In summary, we have found that human islets chronically exposed to the sulfonylurea glibenclamide show loss of β‐cell identity which is mediated, at least partially, by ER stress. Moreover, the sustained exposure to glibenclamide resulted in impaired β‐cell function and increased apoptosis in human islets. The results increase our mechanistic understanding of secondary failure to sulfonylureas. We suggest that the loss of β‐cell identity, as well as the negative effects on β‐cell function and apoptosis, contribute to the secondary failure of sulfonylureas and may accelerate the decline of functional β‐cell mass that characterises the progression of type 2 diabetes.

AUTHORS CONTRIBUTIONS

CF conducted the experiments and performed data analysis, MN, KR and SMC and EE conducted experiments and contributed to data analysis. MN, MS, GMG, and PSJ performed and contributed to pancreas procurement. NT contributed to the design of the study, supervised the project, analysed the data, and edited the manuscript. EM directed the funding acquisition, designed the study, supervised the project, analysed the data and wrote the manuscript. All authors revised the manuscript critically and approved the final version of the manuscript. EM 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.

FUNDING INFORMATION

This study was supported by Agencia de Gestió d'Ajuts Universitaris i de Recerca, CERCA Program, Generalitat de Catalunya. Universitat de Barcelona: APIF grant. Centro de Investigación Biomédica en Red Diabetes y Enfermedades Metabólicas Asociadas. European Regional Develpment Fund, European Commission. Instituto de Salut Carlos III: PI19/00246 and PI22/00334.

CONFLICT OF INTEREST STATEMENT

Eduard Montanya has been on advisory boards, received consulting fees or speaker honoraria from, Medcom, Merck Sharp & Dohme, Novo Nordisk, Roche and Sanofi. Noèlia Téllez has received speaker honoraria from Sanofi.

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/dom.16632.

Supporting information

Data S1. Supporting Information.

Data S2. Supporting Information.

Fernández C, Nacher M, Rivera K, et al. Loss of β‐cell identity in human islets treated with glibenclamide. Diabetes Obes Metab. 2025;27(10):5782‐5792. doi: 10.1111/dom.16632

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.