β-Cell Succinate Dehydrogenase Deficiency Triggers Metabolic Dysfunction and Insulinopenic Diabetes

Type 2 diabetes (T2D) is a chronic disease of altered glucose homeostasis, characterized by a progressive decrease in β-cell function and mass, also termed β-cell failure (1–3). Emerging evidence implicates mitochondrial dysfunction as a central contributor to β-cell failure and T2D pathogenesis (4–6); however, the pathophysiological mechanisms of β-cell mitochondrial dysfunction remain to be established. In β-cells, mitochondria play a fundamental role in coupling glucose metabolism to insulin secretion, ensuring strict regulation of glucose-stimulated insulin secretion (GSIS) and compensatory β-cell mass expansion (7). While the pivotal role of the mitochondria in coupling glucose metabolism to insulin secretion is well-characterized, less is known about the molecular mechanisms that link mitochondrial dysfunction to progressive β-cell dysfunction.

Succinate dehydrogenase (SDH), or complex II (CII), is one of five mitochondrial complexes that participates in the electron transport chain (ETC). SDH is composed of nuclear-encoded subunits (SDHA–D) that form a heterotetrameric complex in the inner mitochondrial membrane (8). SDH also functions in the tricarboxylic acid (TCA) cycle by catalyzing the oxidation of succinate to fumarate (9). This enzymatic reaction is accompanied by the generation of FADH₂, which donates electrons to ubiquinone via CII for oxidative phosphorylation (9). The dual role of SDH in the ETC and TCA cycle places it at the nexus of mitochondrial metabolism and ATP generation. Accordingly, impaired SDH/CII activity is linked to severe human metabolic disorders, including Leigh syndrome and cardiomyopathy (reviewed in Refs. 10,11). Notably, the SDH complex is not required for oxidative phosphorylation as it can be bypassed through alternative metabolic pathways (11). Consequently, it is not fully understood how the loss of SDH profoundly impairs mitochondrial function. Proposed disease-causing consequences of SDH disruption are the intracellular accumulation of succinate, increased reactive oxygen species production, and impaired ATP generation (10–12).

SDH deficiency most severely impacts metabolically active cells with continuous energy requirements, such as cardiomyocytes, skeletal muscle cells, and neural cells (10,11); hence, β-cells, a highly metabolic cell type that functionally reports circulating glucose levels via ATP generation and insulin release, is an ideal cell type to investigate the biological function of SDH. Recent studies in rodent islets revealed that SDH/CII inhibition with 3-nitroprorionic acid impaired insulin secretion (13,14), uncovering a role for SDH in regulating β-cell function. Indeed, Wojtovich et al. (15) suggested that reduced SDH activity and consequent succinate accumulation might be causally linked to diabetes. In this study, we test the hypothesis that SDH deficiency is a driver of mitochondrial dysfunction in β-cell failure and diabetes pathogenesis.

All human pancreatic sections from age-, sex-, and BMI-matched healthy donors without diabetes (ND) and donors with T2D were obtained from the Network for Pancreatic Organ Donors with Diabetes (nPOD). Donor information is presented in Supplementary Table 1.

Animal experiments were performed in compliance with the Institutional Animal Care and Use Committee and the Stanford University Administrative Panel on Laboratory Animal Care. Mice were housed in ventilated cages with access to water and normal chow ad libitum. Both male and female mice were used as noted. Sdhb exon 3–targeted mice were generated from the International Knockout Mouse Consortium clone Sdhb^{tm1b(EUCOMM)Hmgu} and injected into C57BL6/J blastocytes (12). To delete Sdhb in pancreatic β-cells (referred to as SDHB^βKO), SDHB^fl/fl mice were crossed with the rat insulin promoter-Ins2-Cre mice (MGI 2387567) (Supplementary Fig. 1A). To account for the known glucose homeostasis phenotypic effects of the Ins2-Cre transgene, all experiments used Ins2-Cre SDHB^fl/wt (referred to as control) (Supplementary Fig. 1B–E). To sort β-cells, SDHB^βKO were crossed with Cre-reporter mice (ROSA^mT/mG; JAX 007676) (Supplementary Fig. 1F).

Mouse islets were isolated by pancreatic perifusion of Cizyme (VitaCyte) and digestion at 37°C for 13 min. Islets were purified by Histopaque (Sigma-Aldrich) density gradient centrifugation for 10 min at 850g without brake. Islets were collected from the interface, filtered through a 70 µm cell strainer, and cultured overnight in islet medium (5.6 mmol/L DMEM low glucose containing 4 mmol/L l-glutamine and 1 mmol/L sodium pyruvate, with 10% FBS and 1% penicillin/streptomycin) before selection for experiments.

All glucose physiology experiments were performed on age- and sex-matched cohorts. To measure glucose tolerance, mice were fasted for 6 h and blood glucose was measured following 2 g/kg intraperitoneal glucose injection. To measure insulin sensitivity, mice were fasted for 4 h and blood glucose was measured following intraperitoneal insulin (0.65 U/kg; Humalog) injection. HOMA of insulin resistance (HOMA-IR) = insulin (pmol/L) × glucose (mmol/L)/22.5; HOMA β-cell function as a percentage (HOMA-β%) = insulin (pmol/L) × 20/glucose (mmol/L) − 3.5.

To assess GSIS in vivo, mice were fasted for 6 h and blood was collected at indicated times following 2 g/kg intraperitoneal glucose injection. To acquire dynamic insulin secretion profile ex vivo, ∼100 islets/mice were sent to the Vanderbilt Islet Procurement and Analysis Core, and 51 islets/mice at 62.7 islet equivalents equaling 150 μm in diameter were used for islet perifusion. To measure static insulin secretion ex vivo, islets were incubated in 2.8 mmol/L Krebs buffer for 4 h and sequentially stimulated with 5.6 mmol/L and 16.7 mmol/L glucose for 1 h. Islet lysate and medium were collected for insulin measurement (STELLUX Chemi Rodent Insulin ELISA; Alpco).

Respirometry measurements were performed using a Seahorse XFe24 Analyzer (Agilent Technologies) according to the manufacturer’s instructions. Briefly, ∼50 islets/mice were seeded in Islet Capture Microplates and incubated in Seahorse XF Media (3 mmol/L glucose and 1% FBS) for 1 h in a CO₂-free 37°C incubator, and oxygen consumption rate (OCR) was measured upon sequential injections of 16.7 mmol/L glucose, 10 μmol/L oligomycin, 10 µm carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), and 50 µm rotenone/antimycin A (Agilent Technologies). For a comprehensive measurement of ETC complex activity, islets were permeabilized using Seahorse XF Plasma Membrane Permeabilizer, and OCR was measured as previously described (16). OCR was normalized to protein and presented as relative to control or percent of basal OCR.

Intracellular adenine nucleotides were extracted from ∼50 islets/mice and measured using the ATP/ADP/AMP Assay Kit (catalog number A-125; Biomedical Research Service) and a SpectraMax iD5 Microplate Reader.

To measure mitochondrial activity in mouse β-cells, isolated islets from Ins-Cre2 ROSA^mT/mG, SDHB^f/f, or SDHB^f/wt mice were incubated in 5.6 mmol/L and 16.7 mmol/L glucose for 4 h, dissociated, resuspended in FACS buffer (5% FBS/PBS), and loaded with 50 nmol/L tetramethylrhodamine, ethyl ester, perchlorate (TMRE; Thermo Fisher Scientific) for 1 h at 37°C. GFP-positive cells were gated, and TMRE signal was analyzed on a BD LSR II flow cytometer. Data were analyzed using FlowJo.

Metabolic profiling was performed at Northwest Metabolomics Research Center. Briefly, isolated islets were homogenized in MeOH/H₂O (80:20). Supernatant was dried in a vacufuge and reconstituted in HILIC solvent (30% mobile phase A/70% mobile phase B) before analysis on an LC-QQQ system coupled to a SCIEX 6500+ triple quadrupole mass spectrometer operating in multiple reaction monitoring mode through the Analyst 1.6.3 software. Metabolite concentrations were quantified in a relative manner using Multiquant 3.0 software and normalized to total peak sum.

Total RNA was extracted from ∼200 islets/mice by TRIzol extraction followed by RNeasy Plus Micro Kit (Qiagen). A cDNA library was prepared with the TruSeq Stranded Total RNA Kit and sequenced using an Illumina HiSeq 4000. Gene-level quantification was performed on all samples’ sorted BAM files using FeatureCounts, counted by Gencode-defined exons, and aggregated to the gene level. Differential expression analyses were performed with edgeR to generate reads per kilobase of transcript per million mapped reads values. Gene set enrichment analysis (GSEA) was performed on MSigDB annotated gene sets with more than one read per kilobase of transcript per million mapped reads in all replicates. Pathway enrichment analyses of genes differentially expressed (P < 0.05) were performed with the integrated knowledge database software MetaCore (Clarivate Analytics).

Isolated islets were lysed in protease and phosphatase inhibitor–supplemented radioimmunoprecipitation assay buffer and immunoblotted as previously described (17). The following primary antibodies were used: phosphorylated ribosomal protein S6 (p-S6; Catalog no. 4858), phosphorylated AMPKa (Catalog no. 2537), SDHA (Catalog no. 11998) (Cell Signaling Technology); OxPhos Cocktail (ab110413); and b-actin (A5316; Sigma-Aldrich). Antibodies were detected with IRDyes and scanned on the Odyssey CLx (LI-COR Biosciences). Relative band intensity was quantified using Odyssey Image Studio 2.0.

Immunofluorescence staining was performed as previously described (17). Processed tissue sections were incubated in primary antibody at 4°C overnight: SDHB (ab14714; sc-271548), sirtuin 5 (Sirt5; Catalog number 8779), lysine succinylation (K-Succ; PTM-401), and p-S6. For each antigen, immunostaining of pancreas sections from experimental groups was performed in parallel and imaged with fixed settings on the Leica DM IL LED microscope to reflect differences in protein. For morphometric analysis, area measurements and signal intensities were quantified in 5–10 islet images/mouse from n = 3 mice/genotype using Volocity 6.3 by an observer blinded to experimental groups.

β-Cell replication was assessed in pancreas sections and isolated islets as previously described (18). Replication was analyzed using a Cellomics ArrayScan VTi.

Excised pancreata were harvested, fixed in 2.5% glutaraldehyde in 0.1 mol/L cacodylate buffer (pH 7.4), processed, and imaged using a JEOL JEM1400 Digital Capture transmission electron microscope at the Stanford University Cell Sciences Imaging Facility.

Rapamycin (LC Laboratories) was dissolved in DMSO (100 mg/mL stock). For in vitro experiments, isolated islets were incubated with 50 nmol/L rapamycin or vehicle for 24 h. For in vivo studies, rapamycin stock was diluted to 1 mg/mL in 10% polyethylene glycol 400/10% Tween-80 (vehicle). Following intraperitoneal injection of 5 mg/kg rapamycin or vehicle, glucose tolerance test (GTT) and in vivo GSIS assays were performed 1 h postinjection.

Results are presented as the mean ± SD or SEM. Statistical comparisons were performed using Student t test or two-way ANOVA where appropriate (GraphPad), and significance was defined by P < 0.05. Details of sample size and independent experimental repeats are provided in the figure legends.

All data from this study are presented in the published article and the supplementary materials. Additional information is available from the corresponding author upon request.

Loss of SDH activity and expression occurs in the peripheral tissues of diabetic rodents and patients with T2D (19–21). Moreover, SDHB is downregulated in prediabetic islets from high-fat diet mice and obese ob/ob mice (22). However, it is unknown whether SDH expression and/or activity is altered in T2D β-cells. To address this question, we evaluated SDHB expression in human pancreatic sections from healthy ND and T2D donors (Supplementary Table 1). While SDHB was expressed in both insulin-positive β-cells and insulin-negative acinar tissue of ND donors (Supplementary Fig. 2), SDHB expression was significantly lower in β-cells of T2D donors (Fig. 1A and B). By contrast, expression of NDUFB8, a complex I–associated mitochondrial protein, was not significantly changed in T2D β-cells (Fig. 1C), consistent with a relatively selective reduction of SDH. Recently, nonenzymatic K-Succ was identified as a consequence of excess succinate accumulation in the context of SDH loss (23), and the accumulation of K-Succ is counterregulated by the desuccinylase enzyme SIRT5 (24). Although K-Succ was not increased in T2D donor islets (Fig. 1D), a robust increase in SIRT5 expression was observed (Fig. 1E), potentially indicating a compensatory upregulation of desuccinylation activity. The findings of reduced SDHB and increased SIRT5 protein in human T2D β-cells raised the possibility that reduced SDH enzyme activity causally contributes to T2D development.

To investigate the function of SDH in β-cells, we conditionally disrupted SDHB in the β-cells of mice (SDHB^βKO) (Fig. 2A). β-Cell–selective loss of SDHB expression within the islet was confirmed by immunostaining of pancreatic tissue (Fig. 2B and Supplementary Fig. 1G) and Western blotting of isolated islets (Fig. 2C). Because the SDH complex participates in both the ETC and TCA cycle, the functional impact of SDHB disruption on these metabolic pathways was evaluated. First, we evaluated mitochondrial ETC complex (CI–CIV) activity using a Seahorse bioanalyzer. Notably, SDHB^βKO islets demonstrated nearly undetectable CII activity, but no significant change in other ETC complex activities (Fig. 2D). Hence, disruption of CII was not functionally associated with collateral disruption of or compensation by other ETC complexes. Next, we evaluated SDH activity by measuring intracellular succinate levels by mass spectrometry. Indeed, SDHB^βKO islets had significantly elevated succinate levels (Fig. 2E). Consistent with the robust increase in intracellular succinate, SDHB^βKO β-cells exhibited significantly increased protein K-Succ (Fig. 2F) and a robust increase in expression of the desuccinylation enzyme SIRT5 (Fig. 2G). Together, these data confirm β-cell–specific deletion of SDH catalytic activity in our mouse model and uncover excess succinate accumulation and protein hypersuccinylation as functional consequences of SDH deficiency in β-cells (25).

To study the in vivo effects of β-cell SDHB deficiency on glucose homeostasis, we measured blood glucose levels in control and SDHB^βKO littermates from weaning (3 weeks) to 20 weeks of age. Loss of β-cell SDHB expression led to a progressive rise in fed glucose levels, with no sex-specific or body weight effects (Fig. 3A and Supplementary Fig. 1H and I). While young SDHB^βKO mice were normoglycemic and normoinsulinemic until 6 weeks, hyperglycemia was evident by 10 weeks, and overt insulinopenic diabetes (403.8 ± 92.87 mg/dL) was established by 20 weeks (Fig. 3B and C). SDHB^βKO mice at 20 weeks also demonstrated increased fasting glucose levels and inappropriately low fasting serum insulin levels, confirming a diabetic phenotype (Fig. 3D and E). Consistent with a β-cell–selective defect, HOMA-IR was unchanged, while HOMA-β% was significantly reduced in diabetic SDHB^βKO mice (Fig. 3F and G). Similar to human and other mouse models of diabetes (26,27), SDHB^βKO mice demonstrated a progressive reduction in β-cell mass without alterations in α-cell mass (Fig. 3H), which further supported a pure β-cell defect. Next, we examined the impact on β-cell ultrastructure by transmission electron microscopy. At 5 weeks, SDHB^βKO and control littermates demonstrated no apparent differences in mature insulin granules with an electron-dense core or mitochondrial morphology (Fig. 3I and J). However, by 20 weeks, SDHB^βKO islets exhibited large membraned vacuoles containing engulfed organelles, including insulin granules and damaged mitochondria (Fig. 3J), a T2D-associated finding related to autophagy dysregulation (28). Collectively, these data indicate that β-cell–specific disruption of SDHB resulted in insulinopenic diabetes in pubertal-age mice.

To identify the mechanism that drives SDHB-deficient β-cell dysfunction and diabetes, we focused our analysis on prediabetic 5-week-old mice. First, we interrogated the ability of SDHB^βKO mice to handle glucose challenges by intraperitoneal GTT. Compared with control littermates, young SDHB^βKO mice were mildly glucose intolerant (Fig. 4A) with unchanged insulin sensitivity (Fig. 4B). SDHB^βKO mice exhibited reduced insulin release following glucose administration (Fig. 4C and D). Next, we directly evaluated the impact of SDHB disruption on insulin secretion by performing dynamic islet perifusion assays. SDHB^βKO islets failed to secrete insulin in response to leucine (consistent with impaired TCA cycle metabolism) and demonstrated reduced glucose-stimulated first- and second-phase insulin secretion (Fig. 4E). Importantly, exendin-4–augmented insulin secretion was slightly impaired, while potassium chloride–induced insulin secretion by SDHB^βKO was not significantly altered, indicating an intact insulin secretion mechanism downstream of mitochondrial metabolism (Fig. 4E and Supplementary Fig. 3A). Furthermore, control and SDHB^βKO islets demonstrated similar insulin content (Supplementary Fig. 3B), indicating an intact insulin expression and storage in prediabetic islets.

Given the role of mitochondrial metabolism in regulating β-cell replication (4) and our observation of reduced β-cell mass in prediabetic SDHB^βKO mice, we predicted that impaired glucose-stimulated β-cell expansion contributed to the diabetic phenotype of SDHB^βKO mice. In fact, while basal β-cell replication was unchanged, glucose-stimulated β-cell replication was significantly impaired in prediabetic SDHB^βKO islets compared with controls (Fig. 4F). These results indicate a failure of compensatory nutrient-stimulated β-cell expansion with the loss of SDHB (26).

Knowing that GSIS and replication were disrupted, we next investigated the impact of SDH dysfunction on mitochondrial regulation of stimulus-secretion coupling. To assess mitochondrial glucose metabolism, we measured OCR in response to glucose stimulation. In control islets, injection of 16.7 mmol/L glucose significantly increased OCR by 4-fold (Fig. 4G), and mitochondrial uncoupling with FCCP increased OCR by 2.2-fold above the basal rate, resulting in a spare reserve capacity of ∼225% (Fig. 4H). By contrast, SDHB^βKO islets exhibited reduced basal and glucose-stimulated respiration (Fig. 4G) and reduced maximal respiration and spare reserve capacity (Fig. 4H). The lack of spare reserve capacity confirms disruption of SDH activity (29), while altered OCR measurements of SDH-independent parameters implicate a collateral impact on ATP generation and glucose metabolism, despite intact ETC CI, CIII, and CIV (interrogated above). Indeed, SDHB^βKO islets demonstrate a >50% reduction in ATP synthase–related OCR following oligomycin injection (Fig. 4G). To directly assess ATP generation capacity, we measured ATP and ADP levels following glucose stimulation. Whereas exposure of control islets to glucose elevation increased the ATP to ADP ratio by twofold, glucose-exposed SDHB^βKO islets failed to increase the ATP to ADP ratio (Fig. 4I). These data indicate that SDHB^βKO β-cells have respiratory deficits that contribute to reduced ATP generation and, consequently, reduced GSIS.

To evaluate the β-cell–specific mitochondrial consequences of SDH complex disruption, we crossed Ins2-Cre SDHB^βKO mice to Cre-reporter mice (ROSA^mT/mG) and specifically analyzed GFP-positive β-cells (Fig. 4J). Isolated islets from ROSA^mT/mG control and SDHB^βKO mice were loaded with a mitochondrial membrane potential (ΔΨm)–dependent probe, TMRE, to assess mitochondrial activity in GFP-gated β-cells by flow cytometry. As anticipated, control β-cells demonstrated increased ΔΨm in response to high glucose exposure (Fig. 4K). By contrast, SDHB^βKO β-cells had elevated ΔΨm (hyperpolarization) under basal conditions that collapsed upon elevated glucose exposure (Fig. 4K). This paradoxical loss of ΔΨm indicates an inability to maintain the mitochondrial electron gradient under high glucose, which is consistent with the reduced glucose-stimulated ATP generation of SDHB^βKO β-cells and, potentially, accumulation of succinate within the mitochondria (30). Together, these results demonstrate that β-cell disruption of SDH robustly impairs mitochondrial bioenergetics, stimulus-coupled insulin secretion, and compensatory β-cell replication, thereby culminating in diabetes development.

The abnormal ΔΨm of SDHB^βKO β-cells is indicative of altered cellular metabolism (31). Specifically, aberrant mitochondrial hyperpolarization has been associated with metabolic substrate overload (32), leading us to hypothesize that the basal hyperpolarization of SDHB^βKO β-cells reflected significant metabolic perturbation. To investigate this, we performed comparative metabolomics (liquid chromatography-mass spectrometry [LC-MS]) and high-throughput transcriptome (RNA-sequencing) analysis in islets (Fig. 5A). Notably, SDHB^βKO islets demonstrated prominent accumulation of succinate (as anticipated) with no change in fumarate levels, fatty acid intermediates, nucleic acid building blocks, and precursors of protein synthesis despite a deficit of free amino acid pools (Supplementary Fig. 4A and B). Calculated differential abundance scores demonstrated an upregulation of several anabolic pathway intermediates, including fatty acid/lipid, sugar, and nucleotide metabolites (Supplementary Fig. 4C). Similarly, MetaCore analysis indicated that lipid metabolism and amino acid metabolism were the most significantly altered pathways in SDHB^βKO islets (Fig. 5B). These data suggest that loss of SDH triggered an unexpected shift toward cellular anabolism in the setting of reduced cellular energetics.

Accordingly, transcriptomic analysis identified gene signatures of metabolic perturbation in SDHB^βKO islets. Comparable levels of “housekeeping” and islet cell-type identity (α, β, δ, and ɣ cells) islet genes demonstrated intact cellular identity (Supplementary Fig. 5). Consistent with metabolomics data, MetaCore transcriptional analysis identified fatty acid, lipid, and cholesterol metabolism pathways, including SREBP-regulated cholesterol and fatty acid biosynthesis, as the most significantly altered pathways in SDHB^βKO islets (Fig. 5C). In line with this finding, GSEA revealed that SREBP target genes were enriched in SDHB^βKO islets (Fig. 5D), with Hmgcs1, Insig1, Psk9, Stard4, and Fasn among the induced genes (Fig. 5E). Upregulation of these lipogenic genes highlights a potential effect of SDH disruption on SREBP-regulated lipid synthesis. Indeed, examination of lipid content with a lipophilic dye, Nile Red, revealed strong and diffuse staining in insulin-expressing β-cells from SDHB^βKO that was distinct from the punctate lipid droplet staining in control β-cells (Fig. 5F). These observations demonstrate an inappropriate increase in lipid accumulation in SDHB^βKO islets under basal conditions (33) without major effects on alternative metabolic pathways, such as glycerol-3-phosphate and malate-aspartate shuttles (Supplementary Fig. 6). Together, these data further support our hypothesis that loss of SDH triggers a shift toward lipid anabolism.

Prior work has shown the mammalian target of rapamycin complex 1 (mTORC1) pathway is hyperactivated in SDH-deficient tumors (34) and that mTOR activates SREBP, a key regulator of rate-limiting lipogenic gene expression, by inducing SREBP cleavage and nuclear localization (35). Therefore, we hypothesized that mTORC1, a master regulator of cell growth and metabolism that promotes anabolic processes (36), was hyperactivated in SDHB-deficient β-cells (Fig. 5G). Consistent with this view, p-S6, a target of mTORC1 signaling, was significantly increased in SDHB^βKO islets (Fig. 5H). By contrast, AMPK functions as an opposing nutrient sensor that promotes catabolism in response to nutrient deficiency (36). Despite the low energetic state of SDHB^βKO islets, where AMPK pathway activation was anticipated, SDHB^βKO islets demonstrated persistent hyperactivation of the mTORC1 relative to AMPK pathway (Fig. 5I). Additionally, mTORC1 hyperactivation (p-S6) was confined to the β-cell population of the islet (Fig. 5J). Taken together, these data indicate mTORC1 pathway hyperactivation in SDHB^βKO β-cells.

The observed mTORC1 hyperactivation and anabolic metabolite excess in SDHB^βKO islets led us to test whether the functional defects of SDHB^βKO islets would be reversed by the mTORC1 inhibitor rapamycin, known to reprogram mitochondrial metabolism (37). First, we assessed the effects of rapamycin on the glucose-stimulated ΔΨm of GFP⁺ control and SDHB^βKO β-cells (Fig. 6A and B). In basal glucose conditions, acute treatment of rapamycin had limited effects on control β-cells but significantly mitigated the hyperpolarization phenotype of SDHB^βKO β-cells (Fig. 6A and B). Additionally, the paradoxical loss of ΔΨm with high glucose exposure was rescued by rapamycin treatment in SDHB^βKO β-cells. This led us to hypothesize that under basal conditions, SDH disruption elevates succinate, which drives mTOR hyperactivation, and that rapamycin rescues this phenotype by reducing succinate levels. Indeed, our studies in vitro demonstrate that rapamycin reduces succinate levels and lowers ΔΨm in R7T1 β-cells treated with either the irreversible SDH inhibitor 3-nitroprorionic acid or cell-permeable succinate dimethyl succinate (Supplementary Fig. 7). Therefore, excess mTOR pathway activation substantially contributed to the abnormal SDHB^βKO β-cell mitochondrial phenotype.

The rapamycin rescue of SDHB^βKO β-cell mitochondrial activity raised the possibility that mTOR inhibition might also improve the defective stimulus-secretion coupling of these β-cells. To test this hypothesis, we performed static islet GSIS assay after treatment with rapamycin. Both vehicle- and rapamycin-treated control islets demonstrated an approximate fourfold increase in glucose-induced insulin secretion (Fig. 6C). Vehicle-treated SDHB^βKO islets demonstrate impaired GSIS; however, rapamycin treatment significantly enhanced insulin secretion under high glucose with no significant changes at resting glucose levels (Fig. 6C), resulting in an approximate threefold improvement of the stimulation index (Fig. 6D). These data confirmed a rescue of the SDHB^βKO β-cell metabolic phenotype by mTOR inhibition.

Next, we tested the in vivo effects of rapamycin on β-cell function by performing GTTs and GSIS tests on control and SDHB^βKO mice. To avoid the detrimental impacts of chronic rapamycin treatment on β-cell function and peripheral insulin resistance (38), we examined the acute effect of rapamycin (1 h) on glucose homeostasis. We conducted a crossover experiment in which control and SDHB^βKO were subjected to GTT and GSIS assays 1 h after intraperitoneal vehicle or rapamycin (5 mg/kg) injection (Fig. 6E). A crossover design was used to control for the phenotypic variability of individual mice. Consistent with in vitro data, acute rapamycin treatment had no effect on glucose tolerance (Fig. 6F) or insulin secretion in control animals (Supplementary Fig. 8A). By contrast, a single dose of rapamycin treatment marginally improved the glucose intolerance of SDHB^βKO mice, and this modest glucose homeostasis improvement was accompanied by increased insulin secretion following glucose administration (Fig. 6F and Supplementary Fig. 8B). Specifically, the insulin secretion index of rapamycin-treated SDHB^βKO mice was comparable to that of vehicle- or rapamycin-treated control mice (Fig. 6G). Together, these data demonstrate that rapamycin acutely improves mitochondrial function, glucose tolerance, and GSIS in the context of SDH deficiency–related mTORC1 hyperactivation.

SDH/CII complex is at the nexus of mitochondrial bioenergetics and cellular metabolism, but its function in β-cells has not been extensively studied. In the current study, we demonstrate a central role for SDH in regulating β-cell mitochondrial metabolism and identify SDH deficiency as a potential contributing factor to progressive β-cell failure in T2D.

Protein expression of SDHB, a surrogate marker for the SDH complex, is downregulated in the β-cells of human patients with T2D. This observation is consistent with reduced transcriptomic expression of oxidative phosphorylation-related genes in tissue biopsies from patients with T2D (19,20). Moreover, we observed increased expression of the desuccinylase enzyme SIRT5 in T2D β-cells. This finding implicates SIRT5 upregulation in T2D (39) as a counterregulatory mechanism for excess protein succinylation that occurs in the context of SDH enzyme dysfunction (23,24). This mechanism was mirrored in SDHB^βKO mice. Hence, reduced SDH activity is likely to be an underappreciated contributing mechanism to human diabetes (Fig. 7A).

SDHB^βKO mice are a new model of early-onset diabetes caused by β-cell metabolic dysfunction occurring in the absence of dietary and/or obesity-related challenge (peripheral resistance), a phenotype that parallels age-related insulinopenic diabetes (40). A unique aspect of our mouse model, in contrast to mitochondrial diabetes models based upon Pgc1α or Tfam mutation (41,42), is the prominent impairment of mitochondrial function despite retention of mitochondrial mass and integrity. Interestingly, the metabolic phenotype of SDHB^βKO mice is both overlapping and distinct from mice with β-cell–targeted disruption of fumarate hydratase (FH1βKO), the TCA cycle enzyme immediately downstream of SDH (43). Similar to SDHB^βKO mice, FH1βKO mice exhibit a progressive age-dependent diabetes that begins with glucose intolerance at 9–12 weeks of age. However, FH1βKO demonstrated normal glucose-stimulated ATP generation in the prediabetic state that deteriorated in parallel with the development of dysglycemia, indicating an acquired mitochondrial defect. Our extensive characterization of prediabetic SDHB-deficient β-cells revealed defects characteristic of T2D, such as respiratory deficiency (44), compromised bioenergetics (45), and lipid accumulation (6,46) that preceded dysglycemia. Therefore, SDHB^βKO mice provide a valuable model to study the mitochondrial metabolic phenotype of β-cells in T2D.

The direct consequences of SDH deficiency in β-cell are twofold (Fig. 7B): 1) abrogation of CII activity in the ETC reduces basal mitochondrial respiration (↓OCR); and 2) loss of succinate oxidation to fumarate in the TCA cycle results in excess succinate accumulation (>20-fold ↑). These respiratory and metabolic deficits affect the ΔΨm and proton gradient (ΔpH), which are essential for generating the bioenergetic force to synthesize ATP and maintain GSIS (47). In fact, SDHB^βKO islets demonstrated elevated ΔΨm under basal glucose conditions and a paradoxical loss of ΔΨm upon glucose exposure (Fig. 7C). ΔΨm is established primarily by CI, CIII, and CIV activity, which display intact function in prediabetic SDHB^βKO islets. Additionally, ΔΨm may be increased by high levels of metabolic substrates (32), such as succinate and glycerol phosphate (48) that accumulate in SDHB^βKO islets (Supplementary Figs. 4A and 6), and mTORC1 hyperactivation (31). Hence, we suspect mitochondrial hyperpolarization of SDHB^βKO β-cells is due to a combination of reduced basal oxygen consumption, accumulation of intermediary metabolites (including succinate), and mTORC1 hyperactivity. Additionally, the glucose-dependent ΔΨm collapse in SDHB^βKO β-cells is likely a consequence of increased mitochondrial matrix acidification that occurs via excessive glucose-induced succinate accumulation. Succinate is a diprotic acid with similar physiochemical properties to fumarate, which, upon accumulation, acidifies the mitochondrial matrix and reduces ΔΨm (43), compromising β-cell mitochondrial bioenergetics and ATP generation. These findings highlight the essential role of SDH/CII activity in maintaining β-cell ΔΨm and responding to metabolic demands as well as the detrimental effects of succinate accumulation and mTORC1 hyperactivity (29,30).

A central finding of our study is that prediabetic SDHB^βKO islets exhibit succinate-dependent mTORC1 hyperactivation. This is supported by increased S6 phosphorylation as well as an mTORC1-dependent transcription and metabolic alterations (enhanced SREBP-regulated lipid synthesis). Although mTORC1 plays a role in the regulation of β-cell proliferation and survival under physiological conditions (49), sustained overactivation of the mTORC1-S6K1 pathway is observed in islets of patients with T2D and diabetic rodent models (50,51) and is deleterious to β-cell function in T2D (33,52). In our study, acute mTORC1 inhibition with rapamycin in vitro and in vivo partially rescued SDHB^βKO islet secretion, consistent with the beneficial effects of rapamycin treatment in a β-cell–specific mouse model of chronic mTORC1 hyperactivation (β-TSC2^−/−) (50). The rapamycin-induced reversal of ΔΨm loss with glucose exposure in SDHB^βKO islets suggests that rapamycin may improve β-cell function in part by decreasing succinate levels and restoring normal ΔΨm (Fig. 7C) (53). While acknowledging the undesirable effect of subacute/chronic systemic rapamycin treatment (38,49), our data support the potential utility of rapamycin to restore dysregulated mitochondrial function in T2D. Moreover, our findings support mTORC1 inhibition as an alleviating strategy for metabolic disturbances associated with diabetes, highlights the antidiabetogenic effects of rapamycin (54,55) and accentuates the need for β-cell–targeted therapeutic delivery (56).

In summary, we propose SDH deficiency as a pathogenic driver of β-cell metabolic dysregulation and mitochondrial dysfunction. Importantly, we provide a new mechanistic perspective on β-cell dysfunction, suggesting that succinate accumulation induces an inappropriate mTORC1 hyperactivation that can be mitigated by mTORC1 inhibition. Beyond its role in diabetes pathogenesis, the loss of SDH/CII activity may be related to human aging as an age-associated decline in SDHB expression and SDH activity has been observed in human fibroblasts (57), and model organisms demonstrate an indispensable role of SDH in longevity (58). Importantly, our findings provide a testable mechanistic explanation of age-dependent decline in mitochondrial function, insulin secretion (44), and islet lipid accumulation (59). More broadly, our studies suggest that SDH deficiency is relevant to metabolic disorders characterized by mitochondrial dysfunction and mTORC1 hyperactivation.

Acknowledgments. The authors thank the Vanderbilt Islet Procurement and Analysis Core for the islet perifusion studies (National Institutes of Health grant P30 DK020593) and the Northwest Metabolomics Research Center for performing metabolomics analysis. Human tissue sections were provided by nPOD, a collaborative research project with JDRF. Organ Procurement Organizations partnering with nPOD to provide research resources are listed at https://www.jdrfnpod.org/for-partners/npod-partners/. The authors also thank Yang Li (Jiangbin Ye Lab at Stanford University) for assistance with LC-MS.

Funding. This work was supported by National Institutes of Health (NIH)/National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grants (R01 DK101530, R01 DK119955, and P30 DK116074) and JDRF (2-SRA-2019-800-S-B) to J.P.A. S.L. received funding from an NIDDK Endocrinology Training grant (T32 DK007217) and Stanford Child Health Research Institute (UL1TR001085). H.X. was supported by an NIH/NIDDK grant (R01 EB025867) to J.P.A. M.O.H. received support from the NIDDK (NIDDK-110276), JDRF (2-SRA-2019-700-SB), The Vanderbilt Islet Procurement and Analysis Core is funded by NIH/NIDDK (P30 DK020593), and the American Diabetes Association (1-19-IBS-078). A.M.L. and J.Y. received support from the American Cancer Society (RSG-20-036-01).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.L. designed and performed experiments, maintained mice, performed data analysis, and wrote the manuscript. H.X. performed replication experiments. A.V.V. conducted the blinded histochemical analysis. A.M.M. and M.O.H. analyzed RNA-sequencing data. A.M.L. and J.Y. measured R7T1 metabolite levels. J.P.A. supervised the study and revised the manuscript. J.P.A. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.