Preserving Insulin Secretion in Diabetes by Inhibiting VDAC1 Overexpression and Surface Translocation in β Cells

1Type 2 diabetes (T2D) is steadily increasing and represents a worldwide health problem. The majority of cases occur secondary to obesity with its associated insulin resistance. T2D develops after years of prediabetes, or impaired glucose tolerance (IGT). The disease becomes manifest when insulin resistance is no longer compensated by augmented insulin secretion. Accordingly, monitoring the insulin secretory capacity of pancreatic β cells best predicts future diabetes. During prediabetes, blood glucose values gradually increase, exerting harmful effects on a wide variety of organs, including the cardiovascular system and the β cells in the pancreatic islets, so-called glucotoxicity. Lifestyle modifications can prevent the outbreak of T2D and revert early stages of the disease but are difficult to implement. No current medication can efficiently preserve β cell function or prevent T2D. Among the commonly used T2D drugs, metformin acts mainly, but not exclusively, by suppressing glucose production in the liver (reviewed in ), while thiazolidinediones enhance peripheral insulin sensitivity and sulfonylureas force the secretion of more insulin from the already stressed β cells. Additional antidiabetic drugs have been introduced. Among these, glucagon-like peptide-1 (GLP-1) analogs and dipeptidyl peptidase-4 inhibitors enhance insulin secretion, while sodium-glucose cotransporter-2 inhibitors lower blood glucose by preventing renal glucose reabsorption.

2Dysfunction of the β cells with impaired glucose-stimulated insulin secretion (GSIS) is the main defect in T2D, since the reduction in β cell mass varies greatly between different studies (for review, see, ). The β cell is particularly sensitive to caloric overload, as GSIS is attenuated after glucose infusion in healthy subjects and after culture of human islets under glucotoxic conditions. Genetic factors predispose for β cell decompensation, and more than 100 T2D risk genes have been described, most of which encode β cell proteins. Nonetheless, even combined, these risk alleles only explain the disease predisposition in 10%–15% of cases. Therefore, attention has focused on epigenetic changes conferred by high caloric intake and other diabetes risk factors. In particular, the stress-induced gene thioredoxin-interacting protein (TXNIP) is upregulated in islets of T2D organ donors and in β cells exposed to glucotoxic conditions. Its deletion prevents diabetes in rodents. Glucose induces TXNIP expression by increasing the epigenetic histone acetylation of the TXNIP gene. However, the mechanism underlying the harmful effects of TXNIP induction in the β cell remains to be clarified.

3ATP generated by glucose oxidation in β cell mitochondria couples metabolism to plasma membrane depolarization, which increases cytosolic Ca2+ and insulin exocytosis. This signaling cascade is impaired in T2D, mainly due to defective mitochondrial metabolism. The voltage-dependent anion channel (VDAC) is the most abundant protein of the outer mitochondrial membrane. VDAC1 and VDAC2 determine cell life and death by regulating flux of metabolites, nucleotides, including ADP and ATP, as well as ions between the mitochondria and the cytosol, while the VDAC3 isoform is less well characterized. There is a striking comorbidity between T2D and Alzheimer's disease (AD). In AD, VDAC1 is induced early in the disease, associated with its overexpression in the neurolemma. Moreover, VDAC1 antibodies protect cells from amyloid β (Aβ) peptide-induced neurotoxicity. Such effects have not been reported in T2D. Therefore, we investigated the involvement of VDAC in β cell glucotoxicity. In particular, we studied the transcriptional program induced by glucose in insulinoma cells and human pancreatic islets. The role of VDAC1 in the development of hyperglycemia was also examined in the db/db mouse, a commonly used diabetes model. We report that VDAC1 overexpression and mistargeting to the β cell plasma membrane in T2D causes ATP loss. Direct inhibition of VDAC1 in human T2D β cells restores GSIS and prevents development of diabetes in db/db mice. Metformin also acutely improves GSIS by directly blocking VDAC1 channel function, a hitherto not appreciated mode of action of the antidiabetic drug.

4Islets from T2D organ donors (Table S1 for donor characteristics) display upregulated VDAC1 mRNA, while VDAC2 mRNA is repressed, compared with islets from non-diabetic (ND) donors (Figure 1A). These results were substantiated at the protein level (Figures S1A and S1B). VDAC1 mRNA is strikingly correlated with average blood glucose during the months preceding the demise (glycated A1c, HbA1c) in ND islets (Figure 1B). When the results obtained in T2D donors are included, the correlation, albeit significant, is less marked (Figure 1B, insert).

5Metformin is the most frequently used antidiabetic medication. We could document four donors with metformin therapy. The correlation between HbA1c and VDAC1 expression was more significant when the metformin-treated donors were excluded (Figure S1C). Accordingly, the islets from the metformin-treated donors did not display increased VDAC1 mRNA (Figure 1C). Conversely, there was a negative correlation between HbA1c and islet VDAC2 mRNA (Figure 1D), which was only marginally changed by removal of the metformin-treated donors (Figures 1D insert, S1D, and 1E for islet mRNA levels). We identified two abundantly expressed VDAC1 transcript variants (Figures S1E–S1G). Of note, one of these VDAC1 transcript variants appears to be induced by hyperglycemia, as the elevated expression in T2D donors is already seen in part in islets from IGT donors (Figure S1G). This suggests that impaired blood glucose control drives VDAC1 overexpression, while VDAC2 downregulation, albeit significant, is less pronounced. To evaluate if the changes observed in T2D islets can be linked to glucotoxicity, we cultured human islets under glucotoxic conditions (20 mM glucose for 72 hr), known to attenuate GSIS in human islets. Exposure to high glucose reproduces the T2D profile, increasing VDAC1 and decreasing VDAC2 mRNA (Figures 1F and 1G). However, culture of the islets at 10 mM glucose does not induce VDAC1, whereas VDAC2 mRNA is suppressed (Figure S1F). It may be speculated that increased VDAC1 requires longer time culture at high physiological glucose concentrations; while VDAC2 suppression is more sensitive to elevated glucose. The changes in VDAC1 and VDAC2 under glucotoxic conditions were substantiated at the protein level for both human islets and rat insulinoma INS-1 cells (Figure S1I). To study the effect of metformin, we cultured ND human islets under glucotoxic conditions in the presence of 20 μM metformin, corresponding to plasma concentrations during drug treatment. As shown in Figures 1F and 1G, the inclusion of metformin in the culture medium prevents the alterations in VDAC1 and VDAC2 mRNA levels. These results confirm the observations in metformin-treated islet donors.

6In most investigated tissues and cells, VDAC1 is much more abundant than VDAC2, while VDAC2 mRNA is higher than VDAC1 mRNA in human β cells. Moreover, VDAC2 protein is more abundant than VDAC1 in INS-1 cells (Figure S1J). This may suggest an important role for VDAC2 in β cells. It is noteworthy that manipulations of VDAC1 and VDAC2 expression in β cells led to reciprocal changes, overexpression of VDAC1 suppresses VDAC2, and silencing of VDAC2 increases VDAC1 (Figures S2A and S2B for INS-1 cells; Figures S2G and S2H for human islets). Therefore, altered expression of either isoform could impact on β cell function.

7Metabolic and oxidative stress are known to upregulate VDAC1 and cause its oligomerization with ensuing apoptosis. To address this question, we measured apoptosis in INS-1 cells after overexpression of Vdac1 or knockdown of Vdac2, reaching changes in VDAC protein levels similar to those of T2D islets and islets or INS-1 cells cultured under glucotoxic conditions (cf. Figure S1G). Evaluation of cell death by determining cytoplasmic nucleosomes in INS-1 cells after overexpression of Vdac1 or Vdac2 downregulation at 5 mM glucose shows no significant increase in apoptosis. In contrast, altered VDAC1 expression combined with 20 mM glucose culture, results in marked cell death (Figure S2C). Thus, the diabetes-associated alteration in VDAC expression profile does not per se cause β cell death, suggesting that additional insults trigger apoptosis. In this context, it should be emphasized that while loss of functional β cell mass in T2D is well documented, the contribution of β cell death is debated (; reviewed in ). Therefore, we focused on the role of the two VDAC isoforms in β cell function.

8We first examined insulin secretion in INS-1 cells. GSIS is markedly inhibited after Vdac1 overexpression or Vdac2 knockdown (Figures 2A and 2B). In T2D islets, mitochondrial dysfunction underlies the blunted GSIS. We therefore assessed mitochondrial oxygen consumption rate (OCR). OCR is inhibited both at non-stimulatory (2.8 mM) and stimulatory (16.7 mM) glucose concentrations in cells with altered Vdac isoform expression (Figures 2C and 2D; see S2A and S2B for expression levels). The same phenotype with reduced OCR is reproduced by cell culture under glucotoxic conditions (Figures 2E and 2F), further implicating altered VDAC function in the β cell defect.

9Glucose oxidation in the β cell mitochondria increases the crucial intracellular signaling molecules ATP and Ca2+, which couple metabolism to insulin exocytosis. Therefore, we next monitored glucose-induced increases in cytosolic and mitochondrial Ca2+ during acute stimulation. As expected from the attenuated GSIS, the increases in both cytosolic and mitochondrial Ca2+ evoked by glucose are markedly blunted in cells with upregulated VDAC1 or downregulated VDAC2, as well as after glucotoxic treatment (Figures S2D–S2F). Thus, these results strongly support the involvement of altered VDAC expression in the impaired metabolism-secretion coupling evoked by glucotoxicity. VDAC1 is essential in the function of endoplasmic reticulum-mitochondria contact sites. Decreased VDAC1 protein in these contact sites was observed in human T2D β cells. Thus, not only altered VDAC gene expression but also protein localization could contribute to the mitochondrial dysfunction.

10Subsequently, we investigated whether human islets can be protected from glucotoxicity by silencing VDAC1. Knockdown of VDAC1 results in upregulation of VDAC2 mRNA, while VDAC2 silencing increases VDAC1 expression as in INS-1 cells (Figures S2G and S2H). Remarkably, VDAC1 knockdown not only prevents the suppression of VDAC2 induced by culture in 20 mM glucose, but VDAC2 mRNA is considerably higher than in islets maintained at 5 mM glucose (Figures S2G and S2H). To evaluate the impact of altered VDAC expression, we first measured overall cellular reductive capacity, which is conveniently monitored by formazone production from tetrazolium, reflecting β cell mitochondrial metabolism. Reductive capacity in 5 mM glucose-cultured islets is not altered by VDAC1 knockdown, which, however, completely prevents the marked impairment under the glucotoxic condition. In contrast, VDAC2 silencing evokes similar low overall cellular reductive capacity as 20 mM glucose culture (Figure 2G). These results are substantiated by measurements of ATP content and insulin secretion after 1 hr incubation of islets after prior culture at either 5 or 20 mM glucose. In the glucotoxic islets, 16.7 mM glucose-induced increases in ATP and insulin secretion are abrogated. Like reductive capacity, silencing of VDAC1 preserves basal ATP levels and rescues glucose-evoked increases in ATP and insulin secretion. In contrast, VDAC2 suppression reduces basal islet ATP and blunts both ATP elevation and stimulated insulin secretion (Figures 2H and 2I). Taken together, the altered VDAC expression profile in glucotoxic cells impairs mitochondrial metabolism and, as a consequence, GSIS, while the impact on β cell viability is less pronounced.

11Next, we investigated the transcriptional program induced by glucotoxicity. It is well documented that glucotoxicity in the β cell, like oxidative stress, upregulates TXNIP by the induction and nuclear transfer of the carbohydrate response element-binding protein (ChREBP). This genetic program is also activated by the non-metabolizable glucose analog 2-deoxyglucose, which promotes metabolism-independent nuclear transfer of ChREBP. T2D islets display increased transcripts of both ChREBP and TXNIP (Figure 3A), confirming published results. Vdac1 induction by 2-deoxyglucose in INS-1 cells (Figure S3A) substantiates the involvement of ChREBP and TXNIP in VDAC1 gene regulation by glucotoxicity. Moreover, knockdown of either Chrebp or Txnip in INS-1 cells prevents glucotoxicity-induced Vdac1 upregulation (Figure 3B; for silencing efficacy, see Figures S3B and S3C). The transcriptional activation of TXNIP by glucose requires not only ChREBP dimerization with MLX, but also the recruitment of the histone acetyltransferase p300 (Ep300) for efficient transcription. Ep300 activity is necessary for glucose-induced TXNIP upregulation in human islets and INS-1 cells by altering histone acetylation marks in the TIXNIP gene. To further characterize the roles of ChREBP and TXNIP in VDAC expression, we used INS-1 cells with Ep300 knockout (KO) established by CRISPR/Cas9. As expected, INS-1 (Ep300−/−) cells exhibit extremely low Txnip mRNA. The repressed Txnip evoked significantly reduced Chrebp mRNA after culture at 5 mM glucose compared with wild-type (WT) INS-1 cells (Figures 3C and 3D). After 72 hr exposure to 20 mM glucose, the Chrebp transcript is further decreased, while Txnip levels are upregulated, reaching those of the 5-mM-cultured WT-INS-1 cells. Such an attenuated induction of Txnip was already described for the Ep300 KO cells. INS-1 (Ep300−/−) cells are completely protected from the glucotoxicity-induced increase in VDAC1 mRNA (Figure 3E). Conversely, VDAC2 mRNA in INS-1 (Ep300−/−) cells is already elevated under physiological glucose and, rather than decreasing, further increases in the high glucose condition (Figure 3F). Based on the foregoing results, low VDAC1 combined with high VDAC2 expression should preserve GSIS. Hence, INS-1 (Ep300−/−) cells display similar GSIS after 5 mM glucose culture as the WT cells. The Ep300-KO cells are substantially, but not completely, protected from glucotoxicity-mediated abrogation of GSIS (Figure 3G). The incomplete restoration is not due to changes in cellular insulin content (Figure S3D). Epigenetic mechanisms have been implemented in the development of β cell dysfunction in T2D through environmental stress. Our results support this link, as glucotoxicity-mediated VDAC1 overexpression requires the histone acetyltransferase Ep300 for the transcriptional activation of TXNIP.

12The impaired metabolism-secretion coupling prompted us to examine the cellular localization of VDAC1 and VDAC2. Extra-mitochondrial plasma membrane VDAC1 participates in volume regulation, ATP and metabolite transport, and intrinsic mitochondrial apoptosis. Remarkably, confocal microscopy reveals that VDAC1, but not VDAC2, surface expression occurs in β cells from T2D donors (Figures 4A and 4B). In ND and in the single donor of four with documented metformin therapy we could examine (cf. Figure 1C), VDAC1 remains intracellular (Figures 4A and 4B). VDAC1 surface expression correlates positively with the islet donor HbA1c when ND and T2D islets were pooled (Figure 4C). The surface localization may thus be a consequence of the organ donors' high plasma glucose concentrations. To further substantiate the surface localization of VDAC1, we performed triple immunofluorescence staining in pancreas sections of ND and T2D donors. VDAC1 is clearly overexpressed at the T2D β cell surface, as shown by the co-localization with the plasma membrane-associated SNARE protein SNAP-25 (Figures 4D and 4E). Similar to the observations in pancreas sections, β cells in islets isolated from a single T2D donor also show intense VDAC1 overexpression (Figure 4F). Similar results were obtained in β cell of ND islets kept in culture for 72 hr at 20 mM glucose relative to control conditions (Figure 4G) or INS-1 cells (Figures S4A and S4B). Thus, VDAC1 mistargeting to the plasma membrane may cause the impaired GSIS in glucotoxicity and T2D. It is of interest, in this context, that mitochondrial VDAC1 protein is reduced in T2D β cells. Of note, in AD, VDAC1 is also upregulated in neurons of affected brain areas early in the condition, and associated with translocation of the protein to the neurolemma.

13Subsequently, we studied the functional consequences of aberrant VDAC1 subcellular localization. The mouse Vdac1 gene is alternatively transcribed, yielding an exon1 splice variant encoding a plasma membrane-targeted protein (plVDAC1). Such splicing is not occurring in the human and rat VDAC1 genes (see Figures S1E and S1F; see also ). We postulate that VDAC1 overexpression leads to its targeting to extra-mitochondrial locations. Plasma membrane resident VDAC1 has been documented in various mouse and human tissues with the mitochondrial surface residues facing the extracellular space. In neurons, plVDAC1 activation initiated apoptosis, prevented by antibodies directed against the extracellular N terminus of VDAC1. Therefore, we investigated whether VDAC1 antibody, VBIT-4 and AKOS022075291 (AKOS), two specific small-molecule VDAC inhibitors, affect glucotoxicity-evoked VDAC1 induction in INS-1 cells, like metformin in human islets (Figure 1F). Indeed, not only metformin but also all three VDAC inhibitors prevent glucotoxicity-mediated overexpression of VDAC1 mRNA (Figures S4C and S4D). Moreover, inclusion of VBIT-4 in the 20 mM glucose culture of ND islets also decreases VDAC1 protein expression (Figure 4G). Thus, VDAC1 overexpression is prevented by metformin and inhibitors of VDAC1, suggesting that, as in neurons, plVDAC1 conductance is harmful and notably affecting gene transcription in β cells. Of note, in the limited number of T2D donors with documented metformin therapy, there was no VDAC1 overexpression (cf. Figure 1C). The putative protective action of metformin was substantiated by our in vitro experiments. Metformin at a low therapeutic concentration prevented the high glucose-evoked changes in VDAC1 and VDAC2 expression in islets and INS-1 cells. Metformin inhibits glucotoxicity-evoked TXNIP induction in both insulin-secreting and endothelial cells, which are important targets for insulin action. In diabetic mice, VDAC1 upregulation causes apoptosis in heart coronary endothelial cells, whose malfunction is conducive in T2D cardiovascular complications. Moreover, VDAC1 is upregulated in a model of diabetic cardiomyopathy and in cardiomyocytes cultured under glucotoxic conditions. The action of metformin was reported to be due to stimulation of AMPK, which prevents ChREBP nuclear translocation and TXNIP transcriptional activation.

14To further study the consequence of VDAC1 overexpression, we monitored ATP levels during 1-hr experiments in plVdac1-expressing INS-1 cells. Overexpression of WT Vdac1 (mt Vdac1) leads to a 3-fold increase in ATP release from the cells (Figure 5A), suggesting mistargeting of the ATP-conducting Vdac1 to the plasma membrane. This is substantiated by plVdac1 expression, which causes a 10-fold ATP loss (Figure 5A). The robust GSIS in cells transfected with control plasmid is markedly reduced in mtVdac1-transfected cells and completely abolished in plVdac1-expressing cells (Figure 5B). Moreover, glucose (20 mM) aggravates the marginal cell death in mtVdac1 cells, while plVdac1 alone is more harmful (Figures S5A and S5B). In plVdac1 cells, the loss of ATP is rapidly (1 hr) inhibited not only by VDAC1 antibody, the VDAC inhibitors AKOS and VBIT-4, but also by metformin (Figure 5C). VDAC1 overexpression thus leads to ATP loss as a consequence of its translocation to the plasma membrane.

15Altered membrane permeability is further substantiated by patch-clamp experiments. Indeed, plVdac1-expressing INS-1 cells show 30% higher membrane conductance than mtVdac1-transfected cells (Figure S5C). The increased conductance caused by plVdac1 relative to control INS-1 cells is abolished by the acute addition of either VDAC1 antibody (Figure S5D) or metformin (Figure 5D). Superfusion with metformin does not affect membrane currents in control INS-1 cells while abrogating the elevated conductance in plVdac1-transfected cells (Figure 5D).

16As mtVDAC1 and pl-VDAC1 display identical amino acid sequences, we next performed channel conductance recordings with purified mitochondrial VDAC1 protein reconstituted into planar lipid bilayers. Metformin (30 μM) decreases channel conductance to a low conducting state imposed (Figures 5E, S5E, and S5F) like the specific VDAC inhibitor VBIT-4 (Figures 5F, S5G, and S5H). Nonetheless, metformin binding to VDAC1 might be voltage dependent, as 20 min preincubation with 100 μM metformin causes stronger current inhibition (at ±60 mV) in the isolated channel preparation (results not shown). These results demonstrate that metformin at low, therapeutic concentrations, directly inhibits VDAC1 conductance. To achieve quantitative analysis of the interaction of metformin with VDAC1 and extract a binding affinity coefficient (Kd) for the interaction we used microscale thermophoresis (MST). MST is based on changes in the diffusion of the complex protein-ligand. In contrast to VBIT-4, there is no apparent binding to VDAC1 of metformin (concentrations from 0.625 to 200 μM) (Figure 5G). Metformin inhibits the bilayer reconstituted-VDAC1 conductance but shows no apparent interaction with soluble protein. This may be due to failure of inducing changes in protein conformation detectable by thermophoresis. Furthermore, the acute effect of metformin on VDAC1 solute permeation in intact cells is not mediated via activation of AMP kinase or through an antioxidant effect (Figure S5I). VDAC1 is a target in cancer therapy and the chemopreventive action of metformin may in part be exerted through the here reported VDAC1 inhibition.

17Cell surface mistargeting of VDAC1 in T2D may involve post-translational modification of its two cysteine residues (cys127/232), although they are not important for VDAC1-induced apoptosis. Cysteine-depleted Vdac1 is much more efficiently overexpressed in INS-1 cells than mtVdac1, while the reciprocal suppression of Vdac2 observed after mtVDAC1 overexpression (cf. Figures S2A and S2B) is absent (Figures 6A and 6B). Despite the much higher expression, cysteine-depleted Vdac1 shows 50% less plasma membrane near localization than mtVdac1, as revealed by total internal reflection fluorescence microscopy (see ) (Figures 6C and 6D). Moreover, in contrast to mtVDAC1, the cellular ATP/ADP ratio and its increase by glucose stimulation are largely preserved (Figures 6E, 6F, and S6), as is cellular ATP content (Figure 6G). Furthermore, cysteine-depleted Vdac1-expressing cells do not display increased ATP release, which is very pronounced in mtVdac1 cells (Figure 6H). The preserved cellular ATP content and ATP generation by glucose explain the near normal GSIS in the cells transfected with the mutant VDAC1 (Figure 6I). These results are compatible with VDAC1 targeting to the β cell plasma membrane by post-translational cysteine modification, leading to ATP loss from the cells and impaired GSIS.

18Next, we investigated islet cells from diabetic db/db mice. Like islets from T2D donors, β cells from hyperglycemic db/db, but not from normoglycemic C57/bl, mice show surface expression of VDAC1 (Figures S7A and S7B). This is associated with increased Vdac1 mRNA (Figure S7C). ATP levels in islets from hyperglycemic db/db mice are not raised by 16.7 mM glucose and there is increased ATP release (Figures 7A and 7B). As expected from the VDAC1 surface localization, acute addition (1 hr) of VDAC1 antibody or metformin reduces ATP release and increases its cellular levels (Figures 7A and 7B). Consequently, ATP generation and GSIS are markedly enhanced when VDAC1 is inhibited during the high glucose stimulation (Figures 7A and 7C). Neither ATP content nor the stimulated GSIS is affected by VDAC1 inhibitors, including VBIT-4, in islets from normoglycemic C57/bl mice (Figures S7D and S7E), confirming that VDAC1 mistargeting is restricted to diabetic β cells.

19Subsequently, we investigated whether inhibiting cell membrane VDAC1 conductivity also could ameliorate β cell function in human islets. Inhibition of VDAC1 using metformin, VDAC1 inhibitors, or antibodies does not alter ATP content and cellular reductive capacity after culture at 5 mM glucose, but markedly improves metabolism during glucotoxic conditions or in T2D islets (Figures S7F and S7G). Inclusion of VDAC1 antibody or metformin in the culture medium, while not affecting secretion after culture at 5 mM glucose, prevents the attenuation of GSIS observed in the islets cultured at 20 mM glucose (Figure 7D). Similar results were previously reported with metformin in human islets under glucotoxic conditions and the drug improved the function of T2D islets. To probe for membrane effects of VDAC1 antibody, metformin, or VBIT-4, we pooled results of islets from five T2D donors and one with IGT. After only 1 hr exposure to the compounds, VDAC1 inhibition increases islet ATP content both at 1 and 16.7 mM glucose. Notably, the ATP-increasing effect of glucose is 5-fold enhanced by VDAC1 inhibition (Figure 7E). The T2D islets display severe blunting of GSIS. Remarkably, GSIS is increased nearly 4-fold in parallel with the improved ATP generation by each of the four VDAC1 inhibitors (Figure 7F; cf. 2I and 7D for GSIS in ND islets). Of note, acute addition of VBIT-4 concentrations up to 100 μM does not affect GSIS in ND islets (Figure S7H). Thus, GSIS is markedly improved in T2D islets by the acute inhibition of plVDAC1, suggesting that VDAC1 mistargeting rather than the decreased VDAC2 gene expression (cf. Figures 1A–1E) underlies the impaired β cell function. It cannot be excluded that the decreased VDAC2 expression or reduced β cell mass in the T2D islets underlies the incomplete restoration of GSIS.

20The encouraging results in T2D islets prompted us to study the effects of the VDAC1 inhibitor VBIT-4 in vivo. To this end, we subjected diabetes-prone young db/db mice to daily intraperitoneal injections of VBIT-4 for 5 weeks, starting at age 42 days. This treatment is not affecting blood glucose in control C57/bl mice but prevents the development of the severe hyperglycemia, seen in vehicle-injected db/db mice. Upon VBIT-4 cessation, blood glucose concentrations increase gradually over several weeks, reaching those of vehicle-treated animals (Figure 7G). There is no effect on body weight (Figure S7I). The drug markedly improves glucose tolerance and GSIS in vivo (Figures 7H–7J and S7J). Insulin sensitivity was also improved in the treated db/db mice as calculated by the homeostasis model assessment of insulin resistance (HOMA-IR) (Figure S7K), which may reflect involvement of altered VDAC1 expression in insulin target tissues. GSIS was also normalized in islets isolated from VBIT-4-treated db/db mice (Figure S7L). The early administration of VBIT-4 also prevents increases in water consumption and polyuria in the db/db mice, but only marginally lowers blood glucose in 14-week-old animals with manifest hyperglycemia (data not shown). It is noteworthy, that VBIT-4 appears to be more efficient than metformin administered before the outbreak of diabetes in the same animal model. A possible explanation is that metformin inhibits VDAC1 conductance less efficiently than VBIT-4. Nonetheless, both metformin and VBIT-4 also counteracted VDAC1 overexpression, preventing mistargeting to the β cell surface under glucotoxic conditions (see scheme in Figure S7M). Collectively, early initiation of therapy with a VDAC1 inhibitor blocks the development of diabetes in this T2D mouse model. Through the prevention of VDAC1 gene expression, VBIT-4 may have disease-modifying actions. We base this on the remarkable delay (at least 3 weeks) in the establishment of severe hyperglycemia after discontinuation of the compound.

21We demonstrate that the harmful effects of glucose cause VDAC1 induction in T2D. This novel diabetes executer protein impairs β cell metabolism-secretion coupling through its mistargeting to the β cell membrane with consequent ATP depletion (Figure S7M). Our results suggest VDAC1 as a novel target in diabetes therapy with the potential to prevent progression of prediabetes to overt T2D. We also report that metformin inhibits VDAC1 function, a novel mechanism of action for the antidiabetic drug. The β cell protective action of metformin with ameliorated insulin secretion rather than improved insulin effectiveness was recently reported in T2D subjects. It is likely that this beneficial effect targets VDAC1 gene expression and function in the β cells.

22The main limitations of the current study are the limited experiments in diabetic animal models. In particular, in-depth analysis of the relative importance of VDAC1 inhibition for the amelioration of insulin secretion from the β cells relative to action on insulin target tissues, such as liver, muscle, and fat, needs to be performed. It would also be important to crossbreed diabetic mice with Vdac1-KO mice. This is, however, difficult due to the reduced fertility of Vdac1-KO mice. The reciprocal regulation of gene expression between VDAC1 and VDAC2, which we have reported, requires further investigation both in vitro and in vivo. However, such in vivo studies would require β cell-targeted Vdac1 and Vdac2 deletion, as global Vdac2-KO mice are embryonic lethal.

23Further information and requests for resources and reagents should be directed to the Lead Contact, Albert Salehi (s_albert.salehi@med.lu.se).

24INS-1 832/13 cells derived from male rat insulinoma were kindly donated by Dr. C. B. Newgaard, Duke University, (USA).

25Human pancreatic islets were obtained through collaboration between Human Tissue Laboratory within Lund University Diabetes Centre (LUDC) and the Nordic Network for Clinical Islet Transplantation (Prof. Olle Korsgren, Uppsala University, Sweden). Donors were grouped according to HbA1c i.e. less than 6% (ND), between 6 % and 6.5 % (IGT), higher than 6.5% or history of diabetes (T2D) (See Table S1). The human islets (70-90 % purity) had been cultured in CMRL 1066 (ICN Biomedicals, Costa Mesa, CA) supplemented with 10 mM HEPES, 2 mM L-glutamine, 50 μg/ml gentamicin, 0.25 μg/ml fungizone (Gibco, BRL, Gaithersburg, MD), 20 μg/ml ciprofloxacin (Bayer Healthcare, Leverkusen, Germany) and 10 mM nicotinamide at 37°C (5% CO2) for 1 to 5 days prior to the arrival in the laboratory. The islets were then hand-picked under stereomicroscope prior to use. All procedures using human islets were approved by the ethical committees at Uppsala and Lund Universities, Sweden.

265 weeks old Female db/db (on the B6/KS genetic background) mice and control (C57/bl) weighing 18-25 g, were from Janvier Laboratory, France). The experimental procedures were approved by the Ethics Committee for Animal Research at Lund University. The animals were given a standard pellet diet (B&K) and tap water ad libitum. Isolation of pancreatic islets was performed by retrograde injection of a collagenase solution via the pancreatic duct and islets were then collected under a stereomicroscope at room temperature.

27INS-1 832/13 cells were cultured in RPMI-1640 containing 11.1 mM D-glucose and supplemented with 10% fetal bovine serum, 100 U/ml penicillin (Gibco), 100 μg/ml streptomycin (Gibco), 10 mM N-2 hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), 2 mM glutamine, 1 mM sodium pyruvate, and 50 μM β-mercaptoethanol (Sigma), at 37Ί C in a humidified atmosphere containing 95% air and 5% CO2. For VDAC1 over-expression by VDAC1 plasmid (1 μg/ml) and VDAC2 down-regulation by and siRNA, INS1 cells were cultured to 60% confluence. To study the long-term effects (24-72h) of high glucose (20 mM) (glucotoxicity) compared to basal glucose (5 mM), INS-1 cells were cultured with RPMI 1640 complete media in the presence or absence of indicated agents and thereafter VDAC1 expression levels, cell viability and function were measured.

28INS1 cells were seeded in six-well plates at a density of ∼5 x 105 cells in culture medium without antibiotics and transfected with full-length cDNA encoding prevalidated VDAC1 construct (1 μg/ml plasmid) encoding VDAC1 or control plasmid (non-coding) using Effectene Transfection Reagent (Qiagen) according to the manufacturer’s instructions. Tween four post transfection, medium was replaced with fresh medium containing antibiotics. If not indicated otherwise, at 72h post-transfection, the cells were harvested and analyzed by immunoblotting and qPCR for the relative level of various proteins and mRNA.

29Histone acetyltransferase p300 (Ep300) was knocked out in the rat pancreatic β-cell line (INS1 832/13) by CRISPR/Cas9 as recently described.

30Cells with VDAC1 over-expressed or VDAC2 down-regulated were used to measure the insulin secretion. To this end, INS-1 cells were kept in HEPES balanced salt solution (HBSS; 114 mM NaCl; 4.7 mM KCl; 1.2 mM KH2PO4; 1.16 mM MgSO4; 20 mM HEPES; 2.5 mM CaCl2; 25.5 mM NaHCO3; 0.2% BSA, pH 7.2) supplemented with 2.8 mM glucose for 2 h at 37°C. Thereafter the cells were incubated for 1 h in the fresh medium with the denoted glucose concentrations and test agents. After incubation an aliquot of medium was removed for analysis of secreted insulin and cellular insulin content was measured as for islets (see below).

31For electrophysiological experiments, INS-1 cells were transiently co-transfected with plVDAC1 and eGFP. Both intensely and less Fluorescent cells were selected for patch clamping. Whole-cell currents were evoked and recorded by EPC10 amplifier and Pulse software (HEKA, Lambrecht/Pfalz, Germany) as previously described with the temperature maintained at 32°C. The cells were continuously perfused with extracellular solution containing 118 mM NaCl, 20 mM tetraethylammonium chloride, 5.6 mM KCl, 2.6 mM CaCl2, 1.2 mM MgCl2, 5 mM HEPES and 5 mM glucose (pH 7.4 with NaOH) in the presence of 100 uM tolbutamide (T0891, Sigma-Aldrich). The intracellular solution consisted of 125 mM Cs-glutamate, 10 mM CsCl, 10 mM NaCl, 1 mM MgCl2, 5 mM HEPES, 3mM Mg-ATP, 0.1 mM cAMP and 0.05 mM EGTA (pH 7.2 with CsOH). 20 uM metformin was used for perfusion, and approximately 8 mM metformin or 130 nM VDAC1 antibody (both diluted approximately 100 fold) for acute addition as indicated in figure legends or text. Conductance was measured by applying 200-ms voltage ramps from -90mV to -50mV. In the experiments studying the effects of acute addition of VDAC1 ab or metformin (Figures S5C and S5D), conductance was measured continuously in the same single cell before, during and after the acute addition of the respective compounds (anti-VDAC1 antibodies or metformin). Data presented are from 15 s after the addition when steady-state was reached.

32For experiments in Figure 5D, the cell dish in the experimental chamber was continuously exposed to 20 μM metformin. Conductance measurements were performed after at least 5 minutes exposure, and the cell dish was replaced no later than after 1 hour of perfusion. No wash-out experiments were performed in this experiment.

33VDAC1 was purified from rat liver mitochondria as described previously. To measure single and multiple channel current, a planar lipid bilayer (PLB) was prepared from soybean asolectin dissolved in n-decane (30 mg/ml). Purified VDAC1 (10–100 ng) was added to the chamber defined as the cis side containing 0.5 M NaCl. After one or a few channels were inserted into the PLB, excess protein was removed by perfusing the cis chamber with ∼10 volumes of solution to prevent further channel incorporation. Following several recordings of channel activity at different voltages, metformin or VBIT-4 was added to the cis chamber, and currents through the channel were again recorded. Currents were recorded by voltage-clamping using a Bilayer Clamp BC-535B amplifier (Warner Instruments, Hamden, CT). Currents were measured with respect to the trans side of the membrane (ground). The currents were low pass-filtered at 1 kHz and digitized online using a Digidata 1440-interface board and Clampex software (Axon Instruments, Union City, CA). Analysis was done using pClamp 10.2 software (Axon Instruments, Union City, CA), or excel (Microsoft).

34The binding study on the purified VDAC1 channel protein performed using a Nano-Temper Monolith NT.115 apparatus as described previously. Briefly, purified VDAC1 was fluorescently labeled using NanoTemper Protein labeling kit BLUE (L001, NanoTemper Technologies). A constant concentration of the protein was incubated with different concentrations of the tested inhibitor in PBS. Afterward, 3–5 μl of the samples were loaded into a glass capillary (Monolith NT Capillaries), and thermophoresis analysis was performed (LED 20%, IR laser 20%). The results are presented as % of the bound fraction calculated as follows: fraction bound100x(F - F min)/(F max-F min).

35IPGTTs were performed in db/db and c57/bl mice after treatment with VDAC1 blocker (daily intraperitoneal injection with VBIT-4, 25 mg/kg body weight) for 5 weeks. Prior to IPGTT test, the mice were fasted for 4 h. Glucose was dissolved in 0.9% NaCl and 2.0 g glucose/kg body weight was injected intraperitoneally (total volume load was 0.3 ml). Serial blood sampling thereafter from vena saphena was performed at 0, 5, 15, 30 and 90 min as previously described elsewhere. Blood glucose was analyzed using glucose oxidase method and plasma insulin was analyzed by ELISA. The Cumulative (area under the curve) changes in plasma glucose or insulin were calculated by subtracting the recorded values from basal (time 0). The homeostasis model assessment of insulin resistance (HOMA-IR) (insulin (mU/l)∗glucose (mmol/l)/22.5) was used to calculate insulin sensitivity during the IPGTT as described.

36Human pancreatic islets were collected under a stereomicroscope at room temperature and cultured at 5 or 20 mM glucose in the absence or presence of test agents for 72 h. Thereafter the islets were washed and preincubated for 30 min at 37°C in Krebs Ringer bicarbonate buffer (KRB), pH 7.4, supplemented with HEPES (10 mM), 0.1% bovine serum albumin, and 1 mM glucose. Each incubation vial contained 12 islets in 1.0 ml KRB buffer solution and treated with 95% O2 and 5% CO2 to obtain constant pH and oxygenation. After preincubation, the buffer was changed to a medium containing either 1 mM or 16.7 mM glucose. The islets were then incubated for 1 h at 37°C in a metabolic shaker (30 cycles per min). Immediately after incubation an aliquot of the medium was removed for analysis of insulin, and the islets were incubated in acid-ethanol for insulin content determination by radioimmunoassay.

37Total RNA from handpicked mouse islets, human donor islets (Diabetic and non-diabetic) or INS1 cells were extracted using RNAeasy (Qiagen, Hilden, Germany). RNA (0.5 μg) was used for cDNA synthesis with SuperScript (Invitrogen, Carlsbad, CA, USA). Concentration and purity of total RNA was measured with a NanoDrop ND-1000 spectrophotometer (A260/A280>1.9 and A260/A230>1.4) (NanoDrop Technologies, Wilmington, DE) and RNA Quality Indicator (RQI) higher than 8.0 (Experion Automated Electrophoresis, Bio-Rad, USA) was considered to be high-quality total RNA preparations. A 10 μl of reaction mixture with 20 ng cDNA, 5 μl TaqMan mastermix (Applied Biosystems, Foster City, CA, USA), and 100 nM TagMan gene expression assay were run in a 7900HT Fast Real-Time System (Applied Biosystems). The qPCR was carried out as follows: 50°C for 2 minutes, 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds, and 60°C for 1 minute. The amount of mRNA was calculated relative to the amount of housekeeping genes (GAPDH, PPIA or HPRT) mRNA in the same sample by the formula X0/R0 = 2CtR-CtX, where X0 is the original amount of mRNA for the gene of interest, R0 is the original amount of HPRT mRNA, CtR the Ct value for HPRT, and CtX the Ct value for the gene of interest. qPCR results were normalized to housekeeping genes. Primer sequences used in the qPCR assays are provided in Table S2.

38To measure the expression level of VDAC1 and VDAC2 under glucotoxic (20 mM glucose) compared to normal condition (5 mM glucose), human islets (300 islets/30 mm Dish) were cultured in RPMI 1640 medium containing either 5 or 20 mM glucose with or without indicated test agents in a humidified incubator (37°C, 5% CO2) for 24 to 72 h. VDAC1 and VDAC2 mRNA and protein levels were investigated by both qPCR (relative to GAPDH, HPRT or PPIA) and Western blotting (relative to the expression of and β-actin), respectively.

39Human islet or INS1 832/13 cells were suspended in 100 μl of SDS-buffer (50 mM Tris-HCl, 1mM EDTA) supplemented with complete protease inhibitor cocktail (Roche, Basel Switzerland), frozen and sonicated on ice on the day of analysis. The protein content of the homogenates was determined by commercially available kit according to the manufacturer’s recommendation (Thermo Scientific, USA). Homogenate samples of islets and INS1 cells representing 30 μg of total protein were subjected to 7.5% SDS-polyacrylamide gels (Bio-Rad, Hercules, CA, USA). After electrophoresis, proteins were transferred into nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). The membranes were blocked in LS-buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 0.1% Tween-20) containing 5% non-fat dry milk powder for 40 min at 37°C. Subsequently the membranes were incubated over night with rabbit-polyclonal anti-VDAC1 and goat-polyclonal anti-VDAC2 antibodies (1:500) at room temperature. After washing (three times) in LS-buffer the membranes were incubated with horseradish peroxidase-conjugated anti-goat or anti-rabbit antibodies (1:1000). Immunoreactivity was detected using an enhanced chemiluminescence reaction (Pierce, Rockford, IL, USA). The band-intensity was related to β-actin.

40Isolated human or mouse islets as well as INS-1 cells were seeded on glass-bottom dishes cultured overnight. Cells were then washed twice and fixed with 3% paraformaldehyde for 10 min, followed by permeabilization with 0.1% Triton-X 100 for 15 min. The blocking solution contained 5% normal donkey serum in PBS and was used for 15 min. Primary antibodies against mouse VDAC1 (Abcam), goat VDAC2 (Abcam) and Guinea pig insulin (Eurodiagnostica) were diluted in blocking buffer and incubated overnight at 4°C. Immunoreactivity was quantified using fluorescently labeled secondary antibodies (1:200) and visualized by confocal microscopy (Carl Zeiss, Germany). The ratio is calculated by mean intensity of plasma membrane to mean intensity in cytosol, according to the formula: Where i1, i2 and i3 represent the intensities of whole cell, cytosol and nucleus, a1, a2 and a3 represent the area of whole cell, cytosol and nucleus respectively.

41Immunohistochemistry was performed following standard protocol. Briefly, the human pancreases were paraffin embedded, cut into 10 um thick sections and rehydrated for staining with primary antibodies of guinea pig anti-insulin (Eurodiagnostica), rabbit monoclonal anti-VDAC1 (Abcam) and mouse monoclonal anti-SNAP25 (SYSY) were used to detect insulin, VDAC1 and SNAP-25 expression, respectively. The nuclei were stained by Hoechst 33258. The human pancreatic sections from ND and T2D donors were kindly provided by the Nordic Network for Islet transplantation (Prof. Olle Korsgren, Uppsala, Sweden). The images were acquired by confocal microscopy and the colocalization analysis of VDAC1 and SNAP25 in insulin-staining cells was performed with software ZEN2012.

42For single cell ATP/ADP ratio measurements, INS-1 cells were co-transfected as above with either wtVDAC1 or des-Cys(127/232)VDAC1 plasmid together with PercevalHR (1 μg/ml) each. Single cell imaging was performed by confocal microscopy as described and applied to INS-1 cells. Expression levels were determined by qPCR.

43INS-1 cells were seeded on glass-bottom dishes and transfected with wild-type VDAC1-EGFP and des-Cys(127/232)VDAC1-EGFP for 48 hours. The membrane expression of wild type and des-Cys(127/232)VDAC1 was measured by TIRF imaging, which detects the VDAC1 signal about 150 nm close to the glass surface. The analysis of VDAC1 spots was performed by ImageJ Plugin and ZEN2012 software. The experiments were repeated 3 times with 30 cells in each group of wild type and des-Cys(127/232)VDAC1transfected cells.

44For VDAC1 and VDAC2 small interfering RNA (siRNA) experiments, 20-25 nucleotide stealth prevalidated siRNA duplex designed for rat Vdac1 and Vdac2 (Applied Biosystem) were used. INS1 cells were seeded in six-well plates at a density of ∼5 x 105 cells in culture media without antibiotics and transfected with DharmaFECT1 (Dharmacon; Lafayette, CO, USA) according to the manufacturer’s instructions. Cells were transfected for 24 h with the Vdac1 and Vdac2 siRNA at a final concentration of 50 nM or with control siRNA (non-targeting siRNA) at the same concentration before changing to fresh media including antibiotics. At 72 h after transfection, cells were lysed to extract total RNA or protein to measure the knockdown efficacy.

45For silencing Chrebp and Txnip expression in INS1 832/13 cells, 20-25 nucleotide stealth prevalidated siRNA duplex designed for rat Chrebp and Txnip (Applied Biosystem) were used, following the same protocol as for VDAC silencing. At 72h after transfection, cells were lysed to extract total RNA or protein to measure Chrebp and Txnip knockdown efficiency as well as VDAC1 expression.

46Specific silencing of endogenous human hVDAC1 or hVDAC2 was achieved using lentiviral based shRNA-silencing technique (Santa Cruz, CA, USA). Isolated human islets were incubated at 2.8 mM glucose plus Polybree for 90 min. Thereafter the medium was removed and the islets were washed before addition of culture medium with lentiviral particle containing VDAC1-shRNA or VDAC2-shRNA (5 μl/ml) and the islets were cultured for 72 h at 5 or 20 mM glucose. For comparison, scramble (lentiviral particles without targeting any specific region) served as control. After the culture period the medium was removed and the islets were dispersed into single cells and subjected to cell viability assay using MTS reagent kit.

47The reductive capacity of cells was measured either on INS-1 cells or dispersed human islet cells when the INS-1 cells or islets were subjected to 5 or 20 mM glucose for 72 h in the absence or presence of test agents or after down-regulation of VDAC1 and VDAC2 as described elsewhere. Measurement of reductive capacity was performed using the MTS reagent kit according to the manufacturer’s instructions (Promega) and. Apoptosis was measured with the Cell Death Kit (Roche Diagnostics), which quantifies the appearance of cytosolic nucleosomes.

48To measure the cell viability in living cells, we used EthD1 (Ethidium homodimer-1) and calcein to indicate death and live cell in INS-1 cells, respectively according to manufacturer (Thermo Fisher, USA). The plasma membrane targeted VDAC1 (plVDAC1) and mitochondrial VDAC1 (mtVDAC1) were overexpressed in INS-1 cells cultured with either 5 mM glucose (5G) or 20 mM glucose (20G). The mean intensity of Ethidium and Calcein were calculated to indicate live and death cells, respectively.

49OCR was measured in INS-1 832/13 cells using the XF (extracellular flux) analyzer XF24 (Seahorse Bioscience), as previously described in detail. An assay medium composed of 114 mMNaCl, 4.7 mM KCl, 1.2 mM KH2PO4, 1.16 mM MgSO4, 20 HEPES, 2.5 CaCl2, 0.2% bovine serum albumin, pH 7.2, and supplemented with 2.8 mM glucose was used in the XF analysis. The cells were seeded in an XF24 24-well cell culture microplate at 2.5×105 cells/well (0.32 cm2 growth area) in 500 μl of growth medium and incubated overnight at 37°C in a humidified atmosphere of 95% air and 5% CO2. Prior to assay, RPMI 1640 medium was removed and replaced with 750 μl of assay medium. The cells were preincubated under these conditions for 2 h at 37°C in air. The experiments were designed to determine respiration in low (2.8 mM) glucose and for 60 min following the transition to high (16.7 mM) glucose. The proportions of respiration driving ATP synthesis and proton leak were determined by the addition of oligomycin (4 μg/ml). After a further 30 min, 4 μM of dinitrophenol was added to determine maximal respiratory capacity. After a further 10 min, 1 μM rotenone was added to block transfer of electrons from complex I to ubiquinone.

50ATP content and release from isolated mouse or human islets or from INS-1 cells after transfection with mitochondrial targeted VDAC1 (mtVDAC1) or plasma membrane targeted VDAC1 (plVDAC1) plasmids was determined using a luminometric assay kit according to manufacturer's recommendation (Biovision). After incubation of islets (50/vial) or INS-1 cells for 60 min, an aliquot of the media was removed for subsequent measurements of released ATP. Then the islets or INS-1 cells were washed 3 times and the lysates were used for measurements of ATP and protein contents following the protocol provided by the vendor.

51INS-1 cells (6×105 cells) were seeded on 24-well plates and subjected to the following treatments: Control with 5 mM glucose, VDAC1 overexpressed (OE) and silencing VDAC2 (KD) by siRNA in 5 mM glucose culture, and 3 days treated with 20 mM glucose. 24 hours prior to Ca2+ imaging, the cells were transferred to glass-bottom dishes while diluted 1:6 (∼1×105 cells). The cells were loaded 1 h with Rhod-2 (0.75 uM) and Fluo-5F (0.5 uM) dissolved in the perfusion buffer (KRB). Time lapse ROi images were acquired by confocal microscopy and the mean intensity of ROIs was analyzed by ZEN 2009 software. The data calculation was performed with Excel and normalized ratio was calculated by Fi/F0.

52The results are expressed as means ± SEM for the indicated number of individual experiments (as given in figure legends) or illustrated by an observation representative of a result obtained from different experiments (Western blots). The significance of random differences were analyzed by Student's t-test or where indicated the analysis of variance followed by Tukey-Kramers’ multiple comparisons test. P value <0.05 was considered significant. All data were assessed to ensure normal distribution and equal variance between different groups.

53The accession number for the deposited data is GEO: GSE108072.