Mitofusin2 Promotes β Cell Maturation from Mouse Embryonic Stem Cells via Sirt3/Idh2 Activation

β cell dysfunction is the leading cause of diabetes. Adult β cells have matured glucose-stimulated insulin secretion (GSIS), whereas fetal and neonatal β cells are insensitive to glucose and are functionally immature. However, how β cells mature and acquire robust GSIS is not fully understood. Here, we explored the potential regulatory proteins of β cell maturation process and the capacity for GSIS. Combined with the data from public databases, we found that the gene expression of Mitofusin2 (Mfn2) showed an increasing trend from mouse neonatal β cells to mature β cells. Moreover, its protein expression increased during mouse embryonic pancreas development and β cell differentiation from mouse embryonic stem cells. Knocking down Mfn2 reduced Urocortin3 (Ucn3) expression, GSIS, and ATP production in induced β cells, while overexpressing it had the opposite effect. However, neither Mfn2 knockdown nor overexpression affected the differentiation rate of insulin-positive cells. In immature and mature β cells, Mfn2 and its correlated genes were enriched in tricarboxylic acid (TCA) cycle-related pathways. The expressions of Sirtuin 3 (Sirt3) and isocitrate dehydrogenase 2 (NADP+) and mitochondrial (Idh2) were Mfn2-regulated during β cell differentiation. Inhibiting Idh2 or Sirt3 reduced cellular ATP content and insulin secretion levels that increased by Mfn2 overexpression. Thus, Mfn2 modulated the induced β cell GSIS by influencing the TCA cycle through Sirt3/Idh2 activation. We demonstrated that Mfn2 promoted embryonic stem cell-derived β cell maturation via the Sirt3/Idh2 pathway, providing new insights into β cell development. Our data contribute to understanding diabetes pathogenesis and offer potential new targets for β cell regeneration therapies.


Introduction
Diabetes is a common metabolic disorder characterized by hyperglycemia. The International Diabetes Federation reported that 463 million people were living with diabetes in 2019 (https://www.diabetesatlas.org). The pathogenesis of diabetes is complicated, and β cell dysfunction resulting in impaired insulin secretion function plays a vital role in both type 1 and type 2 diabetes [1]. Available medications treat but do not cure diabetes; thus, finding effective therapies is paramount. Restoring β cell insulin secretion in diabetic patients is one possible strategy.
Pancreatic β cell maturation occurs gradually during postnatal development [2,3]. During β cell development, immature fetal and neonatal β cells are poorly responsive to glucose and have low glucose-stimulated insulin secretion (GSIS) [4][5][6]. As they mature, β cells acquire highly sensitive and robust GSIS. Dedifferentiation of mature β cells into immature β cells or progenitor cells contributes to the development of type 2 diabetes [7][8][9][10]. Moreover, differentiation of functional mature β cells from embryonic stem (ES) or induced pluripotent stem cells is possible approaches to replenishing the β cell pools [11]. However, β cells derived from stem cells possess limited GSIS, resembling fetal β cells [12,13]. Therefore, exploring the underlying mechanisms of β cell maturation and their GSIS capacity would contribute to understanding the pathogenesis of diabetes and finding new targets for β cell regeneration therapy.
GSIS is driven by glucose stimulated ATP generation, which relies on the tricarboxylic acid cycle (TCA) cycle and oxidative phosphorylation [1]. Glycolysis predominates in fetal and neonatal β cells, while the mature β cells have an increased rate of oxidative glucose metabolism [14]. Hence, switching from anaerobic glycolysis to mitochondrial oxidative phosphorylation is key to enhancing β cell GSIS function [15]. Mitofusin2 (Mfn2) participates in mitochondrial fusion, contributing to the maintenance and operation of the mitochondrial network. We previously demonstrated that Mfn2 knockdown in adult mouse islets causes impaired GSIS function [16]. Mfn2 deficiency promoted the conversion of somatic cells to a pluripotent state and maintained it through restructuring mitochondrial dynamics and bioenergetics [17]. However, the role of Mfn2 in β cell differentiation and maturation from stem cells remains unclear.
Here, we demonstrated that Mfn2 promoted the maturation of mouse ES cell-derived β cells. It increased their ATP generation and GSIS through Sirtuin3 (Sirt3)/isocitrate dehydrogenase 2 (NADP+) and mitochondrial (Idh2) activation. Our results showed crucial aspects of the Mfn2/ Sirt3/Idh2 pathway in the functional maturation of β cells and provide new insights into the mechanisms of β cell differentiation and regeneration.

2.2.
Knocking Down or Overexpressing Mfn2 in Differentiated Cells. The lentivirus vector expressing short hairpin RNA (shRNA) targeting Mfn2 (shRNA: GGAAGA GCACCGTGATCAA), the vector for overexpressing Mfn2, and their control were designed and synthesized by Gene-Chem (Shanghai, China). Differentiated cells on day 5 + 20 were cultured in a culture medium containing 5 μg/mL polybrene (Sigma-Aldrich). They were infected with lentivirus at a multiplicity of infection (MOI) of 100 for 16 h and further cultured until day 5 + 28.

Immunocytochemistry Analysis.
Mouse ES cell-derived cells were fixed with methanol and blocked with 10% FBS. Cells were then incubated with the primary antibodies: anti-insulin (1 : 200; Cell Signaling Technology, Danvers, MA, USA) and anti-Mfn2 (1 : 100; Abcam, Cambridge, MA, USA) at 4°C overnight. The cells were treated with the appropriate secondary antibodies and DAPI (Sigma-Aldrich). The overlay images were merged using the Image-Pro Plus software.
2.5. Flow Cytometry Analysis. Mouse ES cell-derived cells on different differentiation days were digested into single cells with Accutase (Invitrogen, Waltham, MA, USA). The cells were fixed in 4% paraformaldehyde and then blocked with 3% BSA. After blocking, cells were incubated overnight with anti-insulin (1 : 500) at 4°C. The cells were treated with the appropriate secondary antibody and collected using a flow cytometer (Beckman Coulter, Carlsbad, CA, USA). The results were expressed as the percentage of fluorescence intensity.
2.6. Glucose Stimulated Insulin Secretion Analysis. The differentiated cells on different differentiation days were cultured without insulin for 3 h and were preincubated in Krebs-Ringer Bicarbonate HEPES buffer for 1 h. The medium was replaced with buffer containing 27.7 mM glucose or 5.5 mM glucose and incubated for 1 h. The supernatant was analyzed using the Rat/Mouse Insulin ELISA kit (Sigma-Aldrich). Released insulin was normalized to the total protein content. Due to the discrepancy in insulinpositive cell populations across different treatment groups, the insulin secretion level was defined as the ratio of the insulin secretion value to insulin-positive cell rates.
2.7. ATP and Lactate Production of Differentiated Cells. Differentiated cells were incubated with 2.5 mmol/L glucose for 1 h and analyzed using an ATP Bioluminescent Assay Kit (Sigma-Aldrich) or lactate assay kit (Sigma-Aldrich) according to the manufacturer's protocol.  [22], a Cytoscape plugin, to elucidate the dynamic molecular interaction networks of Mfn2-correlated genes during β cell maturation. DyNet dynamic rewiring score (Dn-score) performs an advanced analysis of the interactions associated with the nodes of multiple networks. A Dn − score > the first quartile of all correlated genes was identified as Mfn2 highly correlated genes during β cell maturation. Gene ontology (GO) biological process and KEGG pathway annotations of highly correlated genes were analyzed using the ClusterProfiler package [23]. An adjusted P value < 0.05 was set as the cutoff criterion. Differentially expressed genes (DEGs) of GSE54374 between P0 β cells and 8-12-week β cells were analyzed using the GEO2R web tool (https://www.ncbi.nlm.nih.gov/geo/geo2r/). Genes were considered differentially expressed when the adjusted P value < 0.05 and |fold change | >1:5.
2.12. Activity of Idh2 in Differentiated Cells. Idh2 activity was quantified using the Idh assay kit (Sigma-Aldrich), following the manufacturer's instructions. Mitochondria were extracted using a mitochondrial extraction kit (Solarbio, Beijing, China), and the extracts were used to evaluate the Idh2 activity. One unit of Idh2 is the amount of enzyme that generates 1.0 μmol of NADP per min at pH 8.0 and 37°C.
2.13. Molecular Docking. The protein 3D structures of Mfn2 and Sirt3 were obtained from the Protein Data Bank. They were optimized using UCSF Chimera (v.1.14) [24], and their structures were prepared for molecular docking with the Dock prep plugin. The docking was performed using the InterEvDock3 online server (http://bioserv.rpbs.univ-parisdiderot.fr/services/InterEvDock3/) [25], and the best model was selected based on the InterEvDock3 scoring function. ViewDock tool of UCSF Chimera was used to analyze the docking results.
2.14. Statistical Analysis. Data are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism 8.3.0 (San Diego, CA, USA). Differences between two groups were assessed for significance using a two-tailed Student's t-test, whereas those between multiple groups were assessed by a one-way or two-way ANOVA.
Next, we assessed GSIS as a functional readout of the induced cells on different days (Figure 1(d)). The insulin levels of the induced cells on day 5 + 20 were not detected. The cells on day 5 + 24 seemed to be glucose responsive, but the insulin levels were low and not glucose concentration dependent. On day 5 + 26, the cells secreted insulin after glucose stimulation. Although the released insulin level   Stem Cells International increased numerically after high-concentration glucose incubation compared with that after low concentration, we observed no significant difference between the two groups. Nonetheless, the released insulin of differentiated cells on day 5 + 28 was significantly higher after stimulation with high concentration glucose compared with low concentration. These results indicated that the insulin-positive cells on day 5 + 24 or day 5 + 26 resemble immature β cells, which had deficient responses to glucose, while those on day 5 + 28 represent mature β cells. Insulin secretion from β cells could be caused by glucose stimulated ATP production. Therefore, we analyzed the production of lactate and ATP at the immature (day 5 + 24) and mature β cell stage (day 5 + 28). Stimulating differentiated cells with 2.5 mmol/L glucose for 1 h revealed that the lactate level decreased and ATP production considerably rose on day 5 + 28 ( Figure 1(e)). Thus, we speculated that glycolysis might be the dominant metabolic pathway in differentiated cells on day 5 + 24, while mitochondrial oxidative phosphorylation played a key role in differentiated cell on day 5 + 28.

Mfn2
Expression during β Cell Differentiation. We first compared Mfn2 expression in the mouse embryonic pancreas in vivo to evaluate its expression during β cell development (Figure 2(a)). It remained almost at the same level from embryonic day E13 to E17 pancreas and displayed an increasing trend from E17 to the newborn pancreas stage. Next, we analyzed Mfn2 expression during the third β cell differentiation stage from days 5 + 9 to 5 + 28 ( Figure 2(b)). We also demonstrated by immunofluorescence staining that almost all of the insulin-positive cells expressed Mfn2 on day 5 + 28 ( Figure 2(c)).

Roles of Mfn2 in β Cell Differentiation and Maturation.
Insulin-positive β cells appeared on day 5 + 20; therefore, we explored the effects of Mfn2 on the β cell generation and maturation. We used shRNA to knockdown Mfn2 on day 5 + 20 and found that Mfn2 protein expression decreased to 35% upon silencing (Figure 3(a)). The ratio of insulin-positive cells after Mfn2 knockdown (20:53% ± 2:87%) was similar to that of the control group (21:41% ± 1:43%) (Figure 3(b)). Ucn3 was also decreased after Mfn2 knockdown (Figure 3(c)). Next, we evaluated GSIS of the induced insulin-positive cells. Because of the discrepancy in the ratios of insulin-positive cells, the insulin secretion level was calculated as the ratio of the released insulin to insulin-positive cell rates. Insulin secretion levels of the induced cells were significantly reduced after Mfn2 knockdown than that of the control group. In addition, they were similar after low-or high-concentration glucose stimulation (Figure 3(d)). Thus, Mfn2 knockdown rendered the differentiated cells unresponsive to glucose and conferred immature β cells.

Stem Cells International
We also used a lentivirus to overexpress Mfn2 on day 5 + 20. Expression of Mfn2 increased to 2.8-fold compared with that of the control group on day 5 + 28 ( Figure 3(e)). The ratio of induced insulin positive cells after Mfn2 overexpression (21:74% ± 4:38%) did not significantly changed compared with the control group (19:67% ± 2:69%) (Figure 3(f)). Expression of Ucn3 was also enhanced following Mfn2 overexpression (Figure 3(g)), indicating an increase of mature β cells. Moreover, ELISA showed that the insulin secretion levels of differentiated β cells increased after Mfn2 overexpression (Figure 3(h)). These results implied that Mfn2 did not promote the differentiation but the maturation of induced insulin-positive cells.
Furthermore, we quantified lactate and ATP levels after Mfn2 knockdown or overexpression. After glucose stimulation on day 5 + 28, Mfn2 knockdown increased the cellular lactate level (Figure 3(i)) and decreased the ATP content (Figure 3(j)). The ATP content on day 5 + 24 was numerical increased without significance after Mfn2 overexpression (supplementary Figure 1). Nonetheless, the overexpression reduced the lactate level (Figure 3(k)) and enhanced the ATP content (Figure 3(l)) in the differentiated cells on day 5 + 28. Hence, Mfn2 could alter the glucose metabolic pathways, modulating mature β cell insulin secretion.

Mfn2-Related Dynamic Networks in Mouse Immature
and Mature β Cells. Since Mfn2 promoted β cell maturation during mouse embryonic stem cell differentiation, we analyzed its expression during the maturation process using the NCBI GEO public database. The transition from immature to mature β cells occurred between P9 and P15 [21]. In a microarray dataset GSE54374, expression of Mfn2 was significantly higher in mature β cells (P15 and 8-12 weeks) than in the immature β cells (P0) (Figure 4(a)). It was also higher in mature β cells (P15 and P60) versus the immature (P0) at the bulk cell level in the single-cell RNA-seq dataset GSE87375 (Figure 4(b)). Correlated genes with Mfn2 in GSE54374 dataset were then analyzed at different time points. Altogether, 1357 correlated genes at P0, 470 correlated genes at P15, and 1257 correlated genes at 8-12 weeks were identified. The PPI networks of correlated genes at P0, P15, and 8-12 weeks were constructed and further analyzed with DyNet plugin of Cytoscape. The DyNet central reference network contained 2102 nodes and 7155 edges. Consequently, 476 correlated genes with a Dn − score > 3 (the first quartile of all correlated genes) were identified as Mfn2 highly correlated genes during β cell maturation (supplementary Table 1). Next, functional pathway enrichment of highly correlated genes with Mfn2 was then analyzed. The top 10 GO biological process and KEGG pathways, as well as the enriched correlated genes, were listed (supplementary Table 2). We found that the KEGG pathway Citrate cycle (TCA cycle) (Figure 4(c)) and GO biological process pathways tricarboxylic acid metabolic process and tricarboxylic acid cycle (Figure 4(d)) were enriched in highly correlated genes. Altogether, 10 highly correlated genes were enriched in these three pathways. Their and Mfn2 expression levels in the GSE54374 were presented as a heat map (Figure 4(e)). Mfn2, Dhtkd1, and Ihd2 were upregulated DEGs between immature P0 and mature 8-12-week β cells, while Glud1 was the downregulated DEG.
We then quantified the mRNA and protein expression of the three Mfn2 correlated DEGs. We demonstrated that the mRNA level of Idh2 was higher on day 5 + 28 than on day 5 + 24, while Dhtkd1 and Glud1 were constant during β cell differentiation (Figure 4(f)). Similarly, the protein expression of Idh2 also increased on the last day of β cell differentiation,

12
Stem Cells International while Dhtkd1 or Glud1 was unchanged (Figure 4(g)). Furthermore, we assessed the activity of Idh2 on different days and found that it was higher on day 5 + 28 than on day 5 + 24 (Figure 4(h)). The results showed that Idh2 expression and activation increase during the maturation of differentiated β cells.

Effects of Mfn2 Knockdown or Overexpression on Idh2 in
Differentiated β Cells. Idh2 plays a role in intermediary metabolism and energy production. We next analyzed the role of Mfn2 on Idh2 expression and activation. We analyzed how changes in Mfn2 expression affect Idh2 expression and protein activity. The Idh2 mRNA level ( Figure 5(a)) and protein expression level ( Figure 5(b)) were both declined after Mfn2 knockdown. Moreover, the activity of Idh2 decreased in Mfn2 knockdown cells ( Figure 5(c)). Conversely, upon Mfn2 overexpression, both the mRNA ( Figure 5(d)) and protein levels ( Figure 5(e)) increased. In agreement, the activity of Idh2 was also enhanced ( Figure 5(f)).
To further explore the role of Idh2 in differentiated β cells, the Idh2 inhibitor AGI6780 which inhibits both mutated and wild-type protein [26] was applied to Mfn2 overexpression cells from day 5 + 20. On day 5 + 28, GSIS was evaluated in the differentiated cells. AGI6780 significantly reduced Mfn2 overexpression-induced high insulin levels after low-or high-concentration glucose stimulation ( Figure 5(g)). It also reversed the Mfn2 overexpression- (c, f) Idh2 activity after Mfn2 knockdown (c) or Mfn2 overexpression (f) detected by Idh activity assay kit. (g) Insulin secretion level of β cells was measured at termination day of differentiation after Mfn2 overexpression with or without AGI6780 treatment. The insulin secretion level in each group was normalized to the insulin secretion level of WT cells stimulated with 5.5 mmol/L glucose. (h) ATP production on day 5 + 28 after one-hour 2.5 mmol/L glucose stimulation. The ATP levels were compared to the total protein amount. n = 3. Values represent mean ± S:D. Statistical significance was set as * P < 0:05 and * * P < 0:01. Student's t-test for (a-f); one-way ANOVA for (h); two-way ANOVA for (g).

13
Stem Cells International enhanced ATP production caused by glucose stimulation (Figure 5(h)). These results demonstrated that Mfn2 may promote the β cell maturation via Idh2 activation.

Mfn2 Regulated β Cell Maturation through Sirt3/Idh2
Pathway. To explore how Mfn2 regulates Idh2 activation, we used the STRING database (https://string-db.org/) to identify potential interacting proteins. Indeed, the database predicted the protein-protein interactions among Mfn2, Sirt3, and Idh2 (Figure 6(a)). Therefore, we analyzed the effects of Sirt3 on β cell differentiation. Sirt3 is an NADdependent protein deacetylase that contributes to energy metabolism. We found that its expression reduced after Mfn2 knockdown, whereas it increased after Mfn2 overexpression ( Figure 6(b)). To identify whether Mfn2 regulates Idh2 through Sirt3, we inhibited Sirt3 by a Sirt3 inhibitor 3-TYP in Mfn2 overexpressing cells from day 5 + 20 during β cell differentiation. The increased Idh2 activity and ATP level caused by Mfn2 overexpression was significantly suppressed by 3-TYP treatment (Figures 6(c) and 6(d)). Sirt3 inhibition by 3-TYP could also decrease the insulin secretion in Mfn2 overexpressing cells after stimulation with low or high glucose (Figure 6(e)). Moreover, we analyzed the molecular docking of Mfn2 and Sirt3. We found that Mfn2 might directly interacted with Sirt3. The Mfn2 residues F331, Q332, and G319 and the Sirt3 residues H305, P326, and E323 were predicted to mediate the contact between the two proteins ( Figure 6(f)). In addition, Sirt3 coimmunoprecipitated with Mfn2, indicating the potential interaction in differentiated cells (supplementary Figure 3). These data suggest that Sirt3/Idh2 regulates the insulin secretory ability in Mfn2-promoted β cell maturation.

Discussion
Loss of β cell GSIS is the main cause of diabetes. Considerable differences exist in the glucose metabolic pathways and GSIS between immature and mature β cells. Our study demonstrated that Mfn2 played an important role in the functional maturation of β cells differentiated from mouse ES cells. Exploring gene profiling datasets from public databases, we found that Mfn2 regulated GSIS and ATP production of the differentiated β cells through Sirt3 and Idh2 activation.
During β cell differentiation from mouse ES cells, insulin-positive cells first appeared on day 5 + 20. We found that the induced β cells on different differentiation days (days 5 + 24, 5 + 26, and 5 + 28) had inconsistent insulin responses to glucose. We inferred that this disparity was because of distinct glucose metabolic manners. Mitochondrial TCA cycle produces more ATP than glycolysis. In addition, glucose-derived ATP directly regulates the K ATP channel activity, promoting insulin release [27]. Diabetes causes marked inhibition of mitochondrial metabolism in β cells, resulting in impaired GSIS [28]. Moreover, blunted mitochondrial function contributes to the immaturity of fetal and neonatal β cells [5]. Immature fetal and neonatal β cells favor anaerobic glycolysis, which occurs in the cytoplasm and results in increased lactate production, while mitochondrial TCA cycle and oxidative phosphorylation predominate in mature adult β cells [14,29,30]. Cells on day 5 + 24 had higher lactate levels and lower ATP levels after glucose stimulation, while differentiated cells on day 5 + 28 had lower glucose-induced lactate levels and higher ATP production. Additionally, Ucn3 is a confident marker for distinguishing between immature insulin-positive cells from functional mature β cells [31]. We found that Ucn3 expressed at a fairly low level on days 5 + 24 and 5 + 26 but high on day 5 + 28. These results indicated that differentiated β cells resembled immature fetal or neonatal β cells on day 5 + 24 and were more like mature β cells on day 5 + 28.
The present study elucidated the important role of Mfn2 in functional β cell differentiation. Mfn2 localizes at the mitochondrial membrane and in the cytosol and drives mitochondrial fusion [32]. We found that Mfn2 expression showed an increasing trend from mouse neonatal β cells (P0) to mature β cells (P15, P60, or 8-12 weeks) in the public datasets GSE54374 [20] and GSE87375 [21]. We demonstrated that Mfn2 protein expression increased during mouse embryo pancreas development in vivo and during β cell differentiation from mouse ES cells in vitro. Mfn2 was also highly expressed in insulin-positive cells. Interestingly, Mfn2 knockdown or overexpression from the day insulinpositive cells first appeared did not change the differentiation rate of insulin-positive cells but promoted the maturation of induced β cells. Overexpressing Mfn2 enhanced Ucn3 expression and glucose-induced insulin release, accompanied by decreased lactate level and increased ATP content. By contrast, Mfn2 knockdown elevated lactate level and declined the Ucn3 expression, ATP production, and insulin secretion level stimulated by different glucose concentrations. Furthermore, Mfn2-knockdown β cells were not glucose-responsive, and their patterns of insulin release were closer to that of immature β cells. This result agrees with our previous study, in which Mfn2 was knocked down in isolated adult islets [16].
Generally, pluripotent cells rely heavily on anaerobic glycolysis, whereas differentiated mature cells rely on mitochondrial oxidative phosphorylation [33,34]. Knockdown of Mfn2 induced impaired mitochondrial fusion, decreasing mitochondrial oxidative phosphorylation [35]. Functional Mfn2 is crucial for embryonic development. Mouse ES cells maintain the pluripotency in the presence of leukemia inhibitory factor, and its withdrawal induced spontaneous differentiation of ES cells, accompanied by elevated Mfn2 expression [36]. Mfn2 mutant embryos exhibit specific and severe disruption of the mice placental trophoblast giant cell layer [37]. Low Mfn2 expression in vitro, by contrast, attenuated the blastocyst formation and preimplantation embryo development [38]. Moreover, Mfn2 deficiency dramatically enhances major enzymes involved in glycolysis and intracellular lactate production, reverting somatic cells into pluripotent cells [17]. Lactate augmentation suggested that glucose-derived pyruvate shifted away from the mitochondria [39]. In differentiated cells, Mfn2 knockdown-induced lactate accumulation and ATP weakening imply mitochondrial dysfunction. The drop in anaerobic glycolysis enzymes, such as lactate dehydrogenase A, in mature β cells is believed 14 Stem Cells International  activity was detected by Idh activity assay kit at termination day of differentiation after Mfn2 overexpression with or without 3-TYP treatment. (d) Insulin secretion level of β cells was measured at termination day of differentiation after Mfn2 overexpression with or without 3-TYP treatment. The insulin secretion level in each group was normalized to the insulin secretion level of WT cells stimulated with 5.5 mmol/L glucose. (e) ATP production on day 5 + 28 after one-hour 2.5 mmol/L glucose stimulation. The ATP levels were compared to the total protein amount. n = 3. (f) Molecular docking analysis of Mfn2 and Sirt3. The green model represents Sirt3, and the yellow model represents Mfn2. Values represent mean ± S:D. Statistical significance was set as * P < 0:05 and * * P < 0:01. Student's t-test for (b, c); one-way ANOVA for (e); two-way ANOVA for (d).
to be critical for coupling glucose metabolism to insulin secretion [14]. We suggested that Mfn2 switched glucose anaerobic glycolysis to aerobic oxidation, promoting functional β cell maturation.
We demonstrated that Mfn2 and its correlated genes in immature and mature β cells were enriched in TCA cyclerelated pathways. Oxidative phosphorylation depends on the activity of dehydrogenase enzyme in the TCA cycle [32]. The Mfn2-correlated gene Idh2 was involved in Mfn2regulated GSIS in differentiated β cells. Reductive TCA cycle flux via Idh2 is required for GSIS, and Idh2 inhibition impairs GSIS in mice [26]. The role of Idh2 in pancreatic islet development has not been fully explored. We found that Mfn2 regulated Idh2 expression and activity. Inhibiting Idh2 reduced cellular ATP content and insulin secretion levels enhanced by Mfn2 overexpression. We suggested that Mfn2 modulated GSIS of induced β cells by influencing the TCA cycle via Idh2.
Our bioinformatic analysis revealed interactions among Mfn2, Idh2, and Sirt3. Overexpressing Mfn2 increased ATP production and activated Sirt3 in mouse neuroblastoma N2a cells [40]. Furthermore, Sirt3 could enhance Idh2 activity [41][42][43]. Sirt3 expression was markedly decreased in islets of patients with type 2 diabetes [44]. Knocking out Sirt3 in β cells caused blunted insulin secretion [45], while overexpressing it reversed palmitate-induced pancreatic β cell apoptosis and enhanced GSIS [46,47]. Until recently, little was known about the role of Sirt3 in pancreatic development. A recent study showed that prenatal testosterone exposure impaired insulin secretion in elderly female offspring via suppressing Sirt3 [48]. These finding implies an underlying association between Sirt3 and β cell function during the embryonic period. We demonstrated that Mfn2 regulated Sirt3 during β cell differentiation. Inhibiting Sirt3 abolished the elevated Idh2 activity, enhanced GSIS, and increased ATP content caused by Mfn2 overexpression.

Conclusion
We demonstrated that Mfn2 plays a key role in promoting β cell maturation from mouse ES cells via Sirt3/Idh2 activation. These results shed light on the potential mechanisms that influence the β cell maturation process and the capacity for GSIS. Our findings will improve our understanding of diabetes pathogenesis and suggest potential targets for β cell regeneration therapies.

Data Availability
The datasets GSE54374 and GSE87375 can be retrieved from Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih .gov/geo/) database in the National Center for Biotechnology Information (NCBI).