Molecular and Metabolic Evidence for Mitochondrial Defects Associated With β-Cell Dysfunction in a Mouse Model of Type 2 Diabetes

Fluorescent dyes Rhod-2 and Rh123 were from Molecular Probes (Eugene, Oregon). Dispase II was from Roche Diagnostics (Germany). Other reagents were from Sigma-Aldrich Canada.

Mice were maintained in a 12-h light/dark cycle and had free access to water and food (diet no. 8664; Harlan Tekland, Madison, WI). MKR and FVB male mice were used as described previously (19). Animal care was conducted according to protocols and standards of the Canadian Council on Animal Care and approved by the Animal Care and Use Committee at the University of Toronto.

Pancreatic islets were isolated as described previously (21).

Electron microscopic analysis was completed on islets isolated from five to seven mice per genotype essentially as previously described (22). The samples were observed under a Philips CM100 electron microscope operating at 75 kV. Images were recorded digitally using a Kodak 1.6 Megaplus camera system operated using AMT software (Advanced Microscopy Techniques Corporation). β-Cells were recognized by the typical ultrastructural appearance (13). The area and number of β-cell mitochondrion and dense core insulin granules were quantified by analyzing 50–100 10,000× magnification electron microscopic images from random areas of islets isolated from each mouse. Area and number were determined by using the threshold setting and the particle analysis tool in Image J software.

Islet insulin secretion and ATP/ADP measurements were performed as described previously (23). Total islet ATP/ADP ratio was determined in the islets after GSIS using the ApoSENSOR ATP/ADP Ratio Assay Kit (BioVision, Mountain View, CA).

Mitochondrial membrane potential (ΔΨ_m) was measured using Rh123 in dispersed β-cells as reported previously (21). Fluorescent responses after the addition of nutrient substrates (11 mmol/l glucose or 10 mmol/l ketoisocaproic acid [KIC]) and the respiration inhibitor (5 mmol/l sodium azide) were compared with baseline (1 mmol/l glucose) and used to characterize mitochondrial hyperpolarization and depolarization, respectively. A decrease in fluorescence corresponded to an increase in ΔΨ_m. The identification of β-cells was based on size and mitochondrial hyperpolarization (21).

Dispersed islet cells were loaded with Rhod-2 (1 μmol/l) during a 45-min pretreatment at 37°C in Krebs-Ringer buffer. The loaded cells were transferred to a perifusion chamber on the thermo platform of an Olympus fluorescent BX51W1 microscope. Images were collected using 540 nm excitation and tetramethylrhodamine methyl ester emission filter set.

Oxidative stress was estimated in dispersed β-cells using the fluorescence emission of 2′,7′-dichlorodihydro-fluorescein-diacetate (DCF) as reported previously (24).

O₂ consumption was measured using a Clark-type electrode coupled to an Oxygraph unit (Hansatech, Pentney, U.K.). Freshly isolated, dispersed, and permeabilized (with saponin, 80 μg/ml, 5 min) islets were suspended at a concentration of 0.6–0.9 mg protein/ml in incubation medium containing 0.25 mol/l sucrose, 10 mmol/l HEPES, 1 mmol/l MgCl₂, 20 μmol/l EGTA, and 0.1% BSA, pH 7.3. Ascorbic acid (10 mmol/l) and N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) (0.4 mmol/l) were added as substrate and used to estimate cytochrome c oxidase activity (25). Oxygen kinetic traces were analyzed by measuring the slopes of the oxygen consumption curves minus background. Respiration rates were converted into molar oxygen units using O₂ solubility in sucrose medium, as previously reported (21).

Sample processing, microarray experiments, and quantitative PCR were performed as described previously (26), and primers are listed in supplemental Table S4 in the online appendix available at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0129/DC1.

Total DNA was extracted from islets using a QIAamp DNA Mini Kit (Qiagen, Germany). The content of mtDNA was calculated using quantitative PCR by measuring the ratio of mitochondrially encoded Cox1, Cox2, and Cox3 versus a nuclear-encoded gene (β-actin).

The global protein expression profiles of freshly isolated islets from 10-week-old MKR and control mice were determined using the isobaric tags for relative and absolute quantification (iTRAQ) quantitative proteomic approach combined with high-performance liquid chromatography mass spectrometry/mass spectrometry (HPLC-MS/MS). Three independent iTRAQ analyses using islets from 8–10 mice were performed. The detailed experimental design and results were reported elsewhere (26). Cluster analysis of detected mitochondrial proteins was performed using GoMiner (27) and GeneMAPP (28) programs.

All experiments were performed with islets pooled from at least three to five mice of each genotype and three to six independent preparations. Results are expressed as means ± SE. Statistical significance was assessed using either a Student's t test or one-way or two-way ANOVA for repeated measures followed by multiple Bonferroni comparisons. P < 0.05 was considered statistically significant.

At 3 weeks of age, MKR mice are normoglycemic, but have significantly increased fasting plasma insulin levels (P < 0.001) compared with WT (Fig. 1,A and B). Five weeks of age represents the time point just before the onset of hyperglycemia in MKR mice, whereas at 6 weeks, their blood glucose is ∼20 mmol/l (19). In contrast, 10-week-old MKR mice were clearly hyperglycemic (P < 0.001) and hyperinsulinemic (P < 0.001), and this was consistent with previous results (Fig. 1,A and B) (18,26). MKR mice are hyperlipidemic by 3 weeks of age, and their plasma triglycerides remain elevated throughout their lifespan (18,29). Ex vivo islet characterization revealed no significant difference in GSIS or ATP/ADP ratio at 3 weeks of age (Fig. 1,C and E). Islets isolated from 5- and 10-week-old MKR mice had significantly higher basal insulin secretion (data not shown) but did not stimulate secretion to the same extent as WT islets in the presence of high glucose (Fig. S1 in the online appendix and Fig. 1,D). Islets from 10-week-old WT mice showed increased total intracellular ATP/ADP ratio after glucose stimulation, whereas islets from diabetic MKR mice showed a blunted response (Fig. 1 F).

Oxidative phosphorylation produces a proton gradient across the inner mitochondrial membrane, which hyperpolarizes the mitochondrial membrane and drives ATP synthesis. Therefore, we measured changes in ΔΨ_m in β-cells using Rhodamine 123 (Rh123) under nutrient stimulation (30). At all ages, the addition of 11 mmol/l glucose hyperpolarized (decreased Rh123 fluorescence) and 5 mmol/l of the respiratory inhibitor NaN₃ completely depolarized ΔΨ_m (Fig. 2,A–C and Fig. S2A in the online appendix). There was no difference in glucose-induced hyperpolarization of ΔΨ_m in β-cells from 3-week-old MKR and control mice. However, β-cells from 5- and 10-week-old MKR mice showed a 46 and 41% decrease in hyperpolarization of ΔΨ_m, respectively, suggesting defective glucose metabolism and mitochondrial function (Fig. 2,C). Defective mitochondrial function was further substantiated using KIC, which is a direct substrate for the mitochondrial TCA cycle and bypasses glycolysis (31). β-Cells from 3-week-old MKR and control mice (Fig. 2,D and F) exhibited a similar response to KIC, whereas β-cells from 5- and 10-week-old MKR mice displayed 41 and 56% lower hyperpolarization of ΔΨ_m, respectively (Fig. 2 E and F and Fig. S2B in the online appendix).

The uptake of Ca²⁺ into the mitochondria reflects mitochondrial metabolic capacity and potentially activates key dehydrogenases of the TCA cycle and OxPhos (32). At 3 weeks of age, 11 mmol/l glucose caused a similar increase in Rhod2 fluorescence in islets from MKR and WT mice (Fig. 3,A and B), indicating similar mitochondrial calcium [Ca²⁺]_m accumulation. However, islets from 5-week-old mice showed a significant 40% reduction (P < 0.05) (Fig. 3,C and D) and 10-week-old diabetic islets showed an even greater 60% attenuation in [Ca²⁺]_m uptake in response to high glucose stimulation, compared with WT (P < 0.001) (Fig. 3 E and F). Direct confirmation of reduced [Ca²⁺]_m loading capacity in MKR islets was obtained using permeabilized β-cells (Fig. S3) and a slight but significant reduction in cytosolic Ca²⁺ uptake was observed in whole islets from 6- and 10-week-old MKR mice (P < 0.05) (Fig. S4).

Maximal respiratory capacity of islet mitochondria was estimated by measuring the activity of complex IV (cytochrome c oxidase) while using ascorbic acid as substrate (25). The rate of decrease in oxygen tension in the chamber, reflecting respiration by mitochondria, was significantly lower in 10-week-old MKR diabetic islet cells compared with WT (P < 0.01), indicating reduced mitochondrial oxidative capacity (Fig. 3 G and H).

We documented mitochondrial morphology changes in pancreatic β-cells of hyperinsulinemic (3 weeks old), just before hyperglycemia (5 weeks old), and diabetic (10 weeks old) MKR mice using electron microscopy. β-Cells in 3-week-old MKR and WT mice had similar mitochondrial morphology (Fig. 4,A and B). Further quantitative analysis revealed no significant difference in mitochondrial number and area or insulin granule number (Fig. 4,C and J and Fig. S6A–C in the online appendix). β-Cells from 5-week-old WT mice showed abundant insulin granules and mitochondria (Fig. 4,D and Fig S6D), whereas MKR β-cells contained significantly less granules (P < 0.05) (Fig. 4,J and Fig S6E and F), 22% fewer mitochondria (P = 0.1), and 30% larger mitochondria (P < 0.05) (Fig. 4,E and F). By 10 weeks of age, MKR β-cells contained significantly less granules (P < 0.001) (Fig. 4,J and Fig S6G–I) and 43% fewer mitochondria (P < 0.001); however, each mitochondria was ∼75% larger (P < 0.01) (Fig. 4 H and I) compared with WT. Notably, the mitochondria in these cells were often swollen, and the inner mitochondrial membranes had a disrupted structure with abnormal cristae.

To investigate the molecular defects responsible for the altered mitochondrial morphology and metabolic coupling in MKR β-cells, we performed a simultaneous analysis of transcript and protein expression profiles in freshly isolated islets from 10-week-old MKR and WT mice (Table S1 and S2; Fig. 5 and Fig. S7). Using the GoMiner software (27), we determined that 55 of 854 differentially expressed genes were located in mitochondria based on Gene Ontology nomenclature. Approximately 50% (29 genes) of those genes were reduced in MKR diabetic islets. Cluster analysis revealed that 17 genes in the mitochondrial inner membrane were dysregulated in diabetic islets, with 65% of them being reduced. Several genes related to cell apoptosis, including CYCS, LGALS12, and SGPP1, were upregulated in MKR diabetic islets, whereas genes involved in glutamate metabolism were decreased (Fig. 5,D and Fig. S7D). Genes in the OxPhos pathway, including Atp5o, Atp5g1, Ndufa1, and Ndufa3, were increased in MKR diabetic islets (Fig. 5 F). Interestingly, neither the microarray nor qPCR results revealed a significant difference in the expression of transcription factors and co-regulators such as nuclear respiratory factor 1 (NRF1), PGC1a, ESRRA, GABPA, GABPB, Tfam, Mfn2, Polg1, and Ucp2 between the two genotypes (Fig. S7C). These genes have previously been implicated in regulation of the mitochondrial electron transport chain and mitochondrial biogenesis (16,17).

Protein expression ultimately affects cellular function, so we performed global protein profiling of islets to complement the microarray data. Using an iTRAQ proteomic strategy combined with HPLC-MS/MS, ∼590 unique proteins were detected at 95% confidence in islets from 10-week-old MKR and WT mice (26). Cluster analysis based on subcellular location revealed 107 proteins belonging to the mitochondrial compartment, and 36 of these proteins were differentially expressed in MKR diabetic islets versus control (Fig. 5,B, Table S3). Remarkably, ∼61% (22 proteins) of changed mitochondrial proteins were located in the inner mitochondrial membrane, and all but one (CYCS) of these proteins were decreased (Fig. 5,B and Fig S8B). Those inner membrane proteins are mainly involved in mitochondrial OxPhos and include key enzymes in the TCA cycle, β-oxidation, glutamate metabolism, and electron transport chain (Fig. 5,B–F). A comparison of the 55 differentially expressed genes and 36 proteins revealed only 10 gene-protein pairs that changed similarly at both the mRNA and protein level (Fig. 5 A and B), implying that posttranslational modifications must contribute significantly to the decrease in protein and altered mitochondrial phenotype.

We quantified the amount of DCF fluorescence and measured the expression of antioxidant genes by qRT-PCR (24,33). At 3 weeks of age, MKR and WT dispersed islets showed similar oxidation of DCF and no major differences in antioxidant enzyme expression (Fig. 6,A, B, and E and Fig. S9A). However, at 10 weeks of age, MKR islet cells exhibited a marked twofold increase in the fluorescent signal from oxidized DCF (P < 0.001) and significantly upregulated antioxidant gene expression compared with WT (Fig. 6,C, D, and F and Fig. S9B). Chronic exposure to pro-oxidants leads to mtDNA damage in a variety of experimental models (16,34). The ratio of mtDNA to nuclear DNA in islets isolated from 3-week-old mice was similar (Fig. 6,G); however, mtDNA content was significantly lower in 10-week-old MKR islets (Fig. 6 H).

Our study reveals that mitochondrial defects do not appear during the early stages of insulin resistance (3-week-old MKR mice); however, just before the onset of hyperglycemia, mitochondrial defects are already apparent, and when the mice are diabetic (hyperglycemic, hyperinsulinemic, and hyperlipidemic [35]), there are clear metabolic and morphological defects in β-cell mitochondria. Consistent with abnormalities in mitochondrial morphology and decreased OxPhos, 22 of 36 differentially expressed mitochondrial proteins were located in the inner membrane, and 95% of them were decreased in MKR diabetic islets. Complexes I–IV of the electron transport chain are located at the inner mitochondrial membrane, and the flux of electrons along this respiratory chain establishes the proton gradient, which in turn generates the ΔΨ_m and drives the production of ATP. In β-cells from MKR mice, just before and after the onset of hyperglycemia, glucose-induced hyperpolarization of ΔΨ_m was reduced compared with control. Our finding is in agreement with a study that showed lower glucose-induced ΔΨ_m and ATP/ADP ratio in islets from human type 2 diabetic patients (12). These changes could be mediated by reduced glucose sensing, reduced mitochondrial metabolism, or a combination of both. However, the reduced ΔΨ_m hyperpolarization in response to KIC (which is a direct substrate for the TCA cycle, bypassing glycolysis) suggests that impaired glucose sensing is not the primary defect. No alteration in glucokinase expression (data not shown) corroborates this idea. Despite having significantly less insulin granules at 5 and 10 weeks of age, the MKR mice maintain the ability to secrete insulin in vivo in response to potent secretagogues such as arginine (19). Therefore, the inability of glucose to hyperpolarize ΔΨ_m is likely a key defect, resulting in a reduced ATP/ADP ratio and GSIS.

The proteomic scan of MKR diabetic islets revealed decreased expression of multiple proteins involved in oxidative metabolism, including several components of the mitochondrial respiratory chain. The reduction of three components in complex IV, which regulates COX activity in response to ATP binding (36), may be responsible for the lower rate of cytochrome c oxidase–dependent oxygen consumption in MKR diabetic islets. Decreased cytochrome c oxidase activity has previously been demonstrated in islets from a diabetic patient (37), and a critical role of mitochondrial substrate oxidation in the β-cell has been demonstrated in patients with mutations in the mitochondrial genome (38). Similarly, we observed that proteins related to oxidative pathways were decreased in 10-week-old diabetic islets. Electron transfer flavoproteins are necessary electron acceptors for many dehydrogenases in the mitochondria, which then transfer electrons to the mitochondrial respiratory chain via electron transfer flavoprotein–ubiquinone oxidoreductase (39). The reduction in electron transfer flavoproteins likely reduces flux through the respiratory chain in MKR β-cells. Furthermore, downregulation of the adenine nucleotide translocator (SLC25A4) was observed in MKR islets. Adenine nucleotide translocators transfer ATP to the cytosol in exchange for ADP (40). Collectively, the downregulation of these numerous mitochondrial proteins in MKR islets are likely associated with the decreased oxidative function, ATP production and transport, and consequently impaired GSIS.

Mitochondria couple cellular metabolism with Ca²⁺ homeostasis and signaling. Several studies have suggested a role for enhanced mitochondrial Ca²⁺ in modulating ATP levels and insulin secretion from cells (rev. in 32); however, this idea is somewhat controversial (30,41). In turn, mitochondrial Ca²⁺ uptake is highly dependent on the potential gradient of the inner mitochondrial membrane. MKR pre-diabetic and diabetic β-cells exhibited significantly attenuated mitochondrial Ca²⁺ uptake in response to glucose, which could not be accounted for by reduced cytosolic Ca²⁺ uptake. It is difficult to distinguish whether the lower mitochondrial Ca²⁺ capacity has reduced the activity and hence expression of enzymes regulating the TCA cycle (such as IDH2 and OGDH) to further augment the mitochondrial respiratory defect or if the reduced Ca²⁺ uptake is a consequence of impaired hyperpolarization of ΔΨ_m due to a reduced flow of substrate through the electron transport chain. Regardless, the data support the idea of reduced mitochondrial metabolic capacity in MKR islets just before and after the onset of hyperglycemia.

A potential consequence of mitochondrial dysfunction is increased production of ROS. Indeed, we observed that 10-week-old diabetic islets exhibited a twofold increase in the production of oxidized DCF but enhanced expression of antioxidant genes compared with control. Generally, increased ROS is a function both of the efficiency of transfer of electrons through the respiratory chain and the level of antioxidant defense in the cell (normally low in β-cells) (42). The enhanced antioxidant gene expression in diabetic MKR islets suggests a compensatory response for the increased oxidative stress (33). Therefore, it appears that the oxidative stress is probably a consequence not a cause of the mitochondrial metabolic dysfunction we observe in 10-week-old MKR islets. The proteomics study revealed decreased expression of a major mitochondrial antioxidant protein, manganese superoxide dismutase (SOD2), in MKR diabetic islet mitochondria. Interestingly, gene expression of Sod2 was not changed, again showing the importance of posttranslational modification. SOD2 converts superoxide to oxygen plus hydrogen peroxide and serves as the primary defense against mitochondrial superoxide accumulation (43). Impaired SOD2 activity has been found in islets from type 1 and 2 diabetic patients (44,45) and type 1 diabetic animals (46). Collectively, the increased oxidative stress and attenuated ΔΨ_m in MKR islets potentially cause further deterioration of mitochondrial function.

The mitochondria of pre- and posthyperglycemic MKR β-cells were swollen and distorted with fewer dense core insulin granules. These data lend further support for the concept of altered mitochondrial function in the development of diabetes in MKR mice. The inner mitochondrial membrane protein, mitofilin (IMMT), which was recently reported to control cristae morphology and facilitate correct mitochondrial function (47), was decreased in MKR diabetic islets at the protein but not mRNA level (Fig. 5 B), confirming again the importance of posttranslational modification in the regulation of the mitochondrial inner membrane. Downregulation of mitofilin in Hela cells caused a drastic change in the organization of the inner membrane and was associated with high ROS production (47). Thus, mitofilin may be a critical organizer of the mitochondrial cristae morphology and indispensable for normal mitochondrial function; however, this requires further validation. These changes in mitochondrial structure may be a biomarker for altered mitochondrial function. Several studies support that mitochondrial morphology and metabolism play a linked role in β-cell function and survival in vitro (9,–11). In addition, hyperglycemia-induced ROS production caused dynamic changes in mitochondrial morphology (48), and hyperglycemia was associated with enlarged mitochondria in Zucker diabetic fatty (ZDF) rat islets (49). Exposure of human pancreatic islets to cytokines induced β-cell damage, caused mitochondrial swelling and enlargement, and ultimately reduced cell survival (50). A distorted morphology similar to that in MKR mouse islets has been reported in β-cell mitochondria from diabetic mice with impaired respiratory chain function (7) and human type 2 diabetic patients (12). Whether the decreased mitochondrial function seen in MKR diabetic islets was caused by reduced mitochondrial content or reduced functional capacity of mitochondria is unclear. Compared with control mice, we showed a significant decrease in mtDNA in MKR diabetic islets, which indicates that decreased mitochondrial content at least partly explains reduced mitochondrial metabolic coupling. However, when cytochrome c oxidase–dependent oxygen consumption rate was normalized to mtDNA, a significant decrease was still observed. This result suggests that functional impairments in mitochondrial oxidative function are not fully accounted for by reduced mtDNA, but by the interplay of decreased mitochondrial function and content.

In conclusion, based on our analysis of the proteomic/transcriptomic/metabolic phenotype of MKR mice, we propose the following sequence of events for the progression of insulin resistance to β-cell dysfunction (Fig. 7): the reduction of multiple proteins in the mitochondrial inner membrane impairs OxPhos, initiating a cycle of reduced redox-stimulated mitochondrial metabolic coupling factors, causing decreased ΔΨ_m and mitochondrial Ca²⁺ capacity, a decline in ATP generation, and impaired GSIS. Concomitantly increased islet oxidative stress might further impair OxPhos and transduction of the glucose-sensing signal. Moreover, reduced expression of the adenine nucleotide translocator would inhibit the translocation of ATP to the cytosol, further inhibiting normal electron transport until the rate of ATP production falls below that of ATP demand, resulting in metabolic failure and impaired insulin secretion. Interestingly, many of the observed mitochondrial impairments and reduced GSIS occurred just before overt hyperglycemia and therefore suggest that improving mitochondrial function may improve defective β-cell secretion seen in type 2 diabetes.

This work was supported by an operating grant to M.B.W. and a postdoctoral fellowship to E.M.A. from the Canadian Diabetes Association.

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

We are thankful for the technical assistance of Dr. Ying Yang for mass spectrometry and Dr. Alpana Bhattacharjee for qPCR. We thank Dr. Derek LeRoith for continued collaboration and for providing the MKR mice.