VDAC1 is a target for pharmacologically induced insulin hypersecretion in β cells

Abstract

β cells are dysfunctional in type 2 diabetes (T2D) and congenital hyperinsulinism (HI), but the mechanisms linking hypersecretion to β cell failure are poorly understood. Here, we use proteomics and functional assays in human and mouse β cell lines to identify VDAC1 as a target for the small molecule hypersecretion inducer SW016789. By enhancing membrane depolarization, SW016789 acutely increases Ca²⁺ influx, eventually driving β cell dysfunction. Time-course transcriptomics analysis reveals a distinct hypersecretory response signature compared to classical endoplasmic reticulum (ER) stress, highlighting ER-associated degradation (ERAD) as a key adaptive pathway. While SW016789 reduces ERAD substrate OS-9 levels, broader ERAD component changes are limited in cell lines. However, immunostaining of the T2D human pancreas shows altered distributions of the ratios of the core ERAD components SEL1L, HRD1, and DERL3 in β cells. This work provides a detailed mechanistic characterization of a hypersecretion-specific stress response, revealing potential therapeutic targets, including VDAC1 and ERAD, for modulating β cell function and survival in disease.

Keywords: CP: Cell biology; CP: Metabolism; ER stress; ER-associated degradation; congenital hyperinsulinism; diabetes; hypersecretion; insulin secretion; pancreatic islets; target identification; transcriptomics; β cells.

Figure 1.. VDAC1 is implicated in hypersecretion induced by SW016789 in human and mouse β cells

(A) SW016789-treated (10 μM) and Tg-treated (3 μM) EndoC-β1 cells expressing a fluorescent XBP1 splicing-based ER stress sensor. Data are the mean ± SE of n = 3. Differences are assessed by two-way repeated measures ANOVA with Dunnett’s multiple comparisons test. Data points with p < 0.05 vs. DMSO are indicated by a darker fill color (SW016789 [SW], dark red; Tg, dark green). (B) Area under the curve (AUC) analysis for (A). Data are the mean ± SE of n = 3. *p < 0.05 vs. DMSO by one-way ANOVA with Dunnett’s multiple comparisons test. (C) Gene expression in human islets treated for 6 or 24 h with SW (10 μM) or thapsigargin (Tg; 1 μM) or DMSO (0.1%) for 24 h. Data are the mean ± SD of n = 5. *p < 0.05 vs. DMSO by one-way ANOVA with Dunnett’s multiple comparisons test. (D) Chemical structure of SW and the photoaffinity probe Z629. (E) InsGLuc secretion assay in β cells treated for 24 h with DMSO, SW (5 μM), or Z629 (5 μM) in the presence or absence of nifedipine (10 μM). Data are the mean ± SD of n = 5. Differences (p < 0.05 by two-way ANOVA with Tukey’s multiple comparison test) are indicated by labeling with different letters. (F) Streptavidin blot indicating biotin-conjugated proteins in MIN6 cells. (G) Candidate targets of SW overlapping between MIN6 and EndoC-βH1 cells. Candidates among the 37 hits in Table S2 include DYNC1H1, PCSK1, VDAC1, VDAC2, VDAC3, CKAP4, SLC3A2, and SEC22B. (H) Resting membrane potential measurements in MIN6 cells treated with DMSO (0.1%) or SW (5 μM). Data are the mean ± SD of n = 3–4. *p < 0.05 by unpaired t test. See also Figure S1.

Figure 2.. VDAC1 is stabilized by and required for SW activity

(A) Depletion of Vdac1 in InsGLuc-MIN6 β cells. Data are the mean ± SD of n = 3. *p < 0.05 by unpaired t test. (B) Glucose (20 mM) stimulated insulin secretion in Vdac1-depleted InsGLuc-MIN6 cells co-treated with 5 μM SW016789 (SW). Data are the mean ± SD of n = 3. Differences (p < 0.05 by two-way ANOVA with Tukey’s multiple comparison test) are indicated by different letters. (C) KCl-induced secretion in Vdac1-depleted InsGLuc-MIN6 cells. Data are the mean ± SD of n = 3. Differences (p < 0.05 by two-way ANOVA with Tukey’s multiple comparison test) are indicated by labeling with different letters. (D) 24 h treatment with SW (5 μM) in the presence or absence of the VDAC1 inhibitor VBIT4 (10 μM) or nifedipine (10 μM). Data are the mean ± SD of n = 6. Differences (p < 0.05 by two-way ANOVA with Tukey’s multiple comparison test) are indicated by labeling with different letters. (E) VDAC1 immunoblot from a cellular thermal shift assay (CETSA) in MIN6 β cells treated with SW. Data are the mean ± SD of n = 3. *p < 0.05 by two-way ANOVA with Sidak’s multiple comparisons test. (F) Isothermal dose-response CETSA for SW in MIN6 β cells. Data are the mean ± SD of n = 3. *p < 0.05 by one-way ANOVA with Dunnett’s multiple comparisons test. (G) Working model of SW actions through VDAC1. See also Figure S2.

Figure 3.. Ca²⁺ influx mediates the loss of β cell function in response to small-molecule hypersecretion inducers

(A) Glucose-stimulated insulin secretion in InsGLuc-MIN6 cells after 24 h treatment with DMSO (0.2%), SW (5 μM), dextrorphan (DXO; 10 μM), or glimepiride (10 μM) in the presence or absence of nifedipine (10 μM). Data are the mean ± SD of n = 8. *p < 0.05 by two-way ANOVA with Dunnett’s multiple comparisons test. (B) InsGLuc-MIN6 cells treated as in (A), except testing different structurally distinct hypersecretion-inducing compounds (5 μM) and Tg (100 nM). Data are the mean ± SD of n = 3. *p < 0.05 by two-way ANOVA with Dunnett’s multiple comparisons test. See also Figure S3.

Figure 4.. Time-course temporal transcriptomics uncovers differential effects of hypersecretion and ER stress

(A) MIN6 β cells treated with DMSO (0.1%), SW (5 μM), or Tg (100 nM) for 1, 2, 6, and 24 h. n = 3. (B) Time-course edgeR analysis results showing overlap of differentially expressed genes (DEGs) between Tg and SW. (C) Heatmaps showing selected pathways containing DEGs that were significant in SW, Tg, or both. Asterisks inside heatmap cells indicate FDR < 0.05 for treatment vs. DMSO for that time point. Asterisks to the right of each gene row indicate significance (FDR < 0.05) across the entire time course vs. baseline. See also Figure S4.

Figure 5.. Clustering analyses unveil a distinct β cell hypersecretory response signature

(A) Weighted gene co-expression network analysis (WGCNA) of the unfiltered transcriptomics dataset. (B) Module eigengene (ME) values for modules that have similar kinetics between SW and Tg (black), different kinetics (yellow), or opposite directional effects (royalblue and turquoise). Data are the mean ± SD of n = 3. (C) Directed network analysis of MEs with their top enriched ontology shown and colored by their kinetics and effect directions. Edges connecting the nodes are colored by positive (red) or negative (blue) correlation, and thickness indicates the magnitude of the correlation, with thicker edges showing stronger relationships. (D) Dirichlet process-Gaussian process (DPGP) clustering analysis. (E) UpSet plot comparison of all WGCNA modules and DPGP clusters. The overlap between SW_DPGP_cluster_3 and the royalblue module is highlighted in light blue. These two sets were merged and analyzed by GO biological process (inset). (F) Select genes for the ERAD pathway show differential expression dynamics between SW and Tg. Values shown are transcripts per million (TPMs) from RNA-seq data (N = 3). See also Figure S5.

Figure 6.. Effects of hypersecretion and ER stress on the expression of ERAD, ER stress, and immediate-early response proteins

(A) OS-9 immunoblots from MIN6 cells treated for 24 h with 1 μM eeyarestatin I (ES1), 5 μM SW, ES1 + SW, or 100 nM Tg. Data are the mean ± SD of n = 3. *p < 0.05 by one-way ANOVA with Dunnett’s multiple comparisons test. (B) Caspase-Glo assay in MIN6 cells exposed to the indicated treatments for 24 h. Data are the mean ± SD of n = 3. Statistical differences (p < 0.05 by one-way ANOVA with Tukey’s multiple comparisons test) are labeled by different lettering. (C) InsGLuc-MIN6 secretion assay in cells treated for 24 h with the indicated drugs, followed by stimulating using diazoxide (250 μM), KCl (35 mM), and glucose (20 mM). Data are the mean ± SD of n = 3. Differences (p < 0.05 by two-way ANOVA with Tukey’s multiple comparisons test) are labeled by different lettering. (D) MIN6 β cells were treated with SW (5 μM), Tg (100 nM), or DMSO (0.1%) for the indicated times, and samples were analyzed by immunoblotting. Data are the mean ± SD of n = 3. *p < 0.05 vs. respective DMSO control by two-way ANOVA with Dunnett’s multiple comparisons test. (E) Human EndoC-βH1 cells were treated as in (D), except Tg was used at 1 μM. Data are the mean ± SD of n = 3. *p < 0.05 vs. respective DMSO control by two-way ANOVA with Dunnett’s multiple comparisons test. Vertical white spaces between blots in (A) and (E) indicate that superfluous lanes were eliminated. See also Figure S6.

Figure 7.. Expression of ERAD components in non-diabetic and T2D human pancreas

(A) Immunohistochemical staining of insulin (INS; red), HRD1 (green), and SEL1L (violet) and overlaid images showing INS+HRD1, INS+SEL1L, and HRD1+SEL1L. Donor identifiers are shown on the right. (B) CellProfiler quantification of the ratio of β cell intensities for SEL1L and HRD1. Axes are pixels (1,024 × 1,024), 120 pixels = 50 μm (C) Histogram plot of all quantified β cell SEL1L/HRD1 ratios. *p < 0.05 by Kolmogorov-Smirnov test. (D) Immunohistochemical staining for INS (red), DERL3 (green), and SEL1L (violet) and overlaid images showing INS+DERL3 and SEL1L+DERL3. All images shown are representative of three independent non-diabetic (ND) and T2D donors. (E) CellProfiler quantification of the ratio of β cell intensities for SEL1L and DERL3. *p < 0.05 by Kolmogorov-Smirnov test. (F) Histogram plot of all quantified β cell SEL1L/DERL3 ratios. Histogram plots in (C) and (F) also show the mean ± SD of n = 3 donors (3–5 imaged regions per donor). All scale bars represent 50 μm. See also Figure S7.