Restoring mitochondrial calcium uniporter expression in diabetic mouse heart improves mitochondrial calcium handling and cardiac function

Diabetes mellitus is a growing health care problem, resulting in significant cardiovascular morbidity and mortality. Diabetes also increases the risk for heart failure (HF) and decreased cardiac myocyte function, which are linked to changes in cardiac mitochondrial energy metabolism. The free mitochondrial calcium level ([Ca2+]m) is fundamental in activating the mitochondrial respiratory chain complexes and ATP production and is also known to regulate pyruvate dehydrogenase complex (PDC) activity. The mitochondrial calcium uniporter (MCU) complex (MCUC) plays a major role in mediating mitochondrial Ca2+ import, and its expression and function therefore have a marked impact on cardiac myocyte metabolism and function. Here, we investigated MCU's role in mitochondrial Ca2+ handling, mitochondrial function, glucose oxidation, and cardiac function in the heart of diabetic mice. We found that diabetic mouse hearts exhibit altered expression of MCU and MCUC members and a resulting decrease in [Ca2+]m, mitochondrial Ca2+ uptake, mitochondrial energetic function, and cardiac function. Adeno-associated virus-based normalization of MCU levels in these hearts restored mitochondrial Ca2+ handling, reduced PDC phosphorylation levels, and increased PDC activity. These changes were associated with cardiac metabolic reprogramming toward normal physiological glucose oxidation. This reprogramming likely contributed to the restoration of both cardiac myocyte and heart function to nondiabetic levels without any observed detrimental effects. These findings support the hypothesis that abnormal mitochondrial Ca2+ handling and its negative consequences can be ameliorated in diabetes by restoring MCU levels via adeno-associated virus–based MCU transgene expression.

Diabetes-related cardiac dysfunction is caused by multiple factors. Mitochondrial dysfunction has been proposed as a principal pathophysiological mechanism in the development of diabetic heart disease (1). Abnormal mitochondrial Ca 2ϩ handling has been identified as contributing to mitochondrial dysfunction, although the mechanisms causing this alteration are not well understood (1). Numerous studies have shown that mitochondrial Ca 2ϩ uptake is decreased in cardiac myocytes (CM) 3 from the diabetic heart (2)(3)(4), yet the degree to which the decreased CM free mitochondrial Ca 2ϩ concentration ([Ca 2ϩ ] m ) modulates decreased mitochondrial and cardiac function is currently unclear.
[Ca 2ϩ ] m is an important signaling mechanism for mitochondrial energetic activity. [Ca 2ϩ ] m enhances the activity of oxidative phosphorylation, especially mitochondrial complexes I, III, IV, and the V max of complex V, leading to enhanced ATP formation (5). In addition, several dehydrogenases in the mitochondrial matrix (MM) are activated by [Ca 2ϩ ] m including the pyruvate dehydrogenase complex (PDC), which is key to glucose oxidation. [Ca 2ϩ ] m can either directly exert allosteric activation of PDC or increase PDC kinetics, reducing PDC phosphorylation via activation of mitochondrial PDC phosphatases (PDPs) (6,7).
The heart is able to metabolize exogenous substrates such as glucose and free fatty acids (FA) to produce ATP, although FA are preferred (60 -70%) (8,9). However, the diabetic heart is energetically almost completely dependent on mitochondrial FA oxidation as a consequence of elevated levels of circulating FA and decreased intracellular glucose availability (10,11). Maintaining dynamic glucose utilization in the presence of FA is essential for optimal cardiac function for the following reasons. 1) ATP produced from glycolysis could be preferentially used by the sarcoplasmic reticulum (SR) to fuel Ca 2ϩ uptake and by the sarcolemma to maintain ion homeostasis (12)(13)(14) and hence a deficiency in glycolytic ATP could damage the integrity of cellular membranes. 2) Increasing flux through the PDC will prevent the accumulation of potentially toxic glycolytic end products such as lactate. 3) FA are known to have an "oxygen-wasting" effect when compared with carbohydrates, which results in a higher ratio between myocardial oxygen consumption and cardiac work. 4) Glucose utilization is necessary to supply ATP during increased cardiac output following ␤-adrenergic stimulation (10). Glucose and FA oxidation have a reciprocal relationship, described in a process known as the glucose/FA cycle (15). Increased levels of the FA oxidation intermediates acetyl-CoA, citrate, and NADH inhibit phosphofructokinase 1, the first and rate-limiting step in glycolysis (16). PDC is the second important key level of control regarding the use of glucose relative to FA for energy homeostasis. In addition to Ca 2ϩ , PDC is regulated by various isoforms of pyruvate dehydrogenase kinase (PDK1, -2, -3, and -4) and phosphatase (PDP1 and -2), with phosphorylation inhibiting enzyme activity. Furthermore, products of FA oxidation (NADH and acetyl-CoA) activate PDKs resulting in PDC phosphorylation and inhibition (10).
Mitochondria accumulate Ca 2ϩ from the cytosol in a tightly regulated way. In CM only 1-2% of cytosolic Ca 2ϩ enters the MM during systole (17). [Ca 2ϩ ] m is regulated by a complex set of mechanisms influencing MM Ca 2ϩ uptake and release, which have been reviewed (18,19). Briefly, the outer MM is Ca 2ϩ permeable (20), but import across the inner MM is highly regulated. The most important contributor to mitochondrial Ca 2ϩ uptake is the mitochondrial Ca 2ϩ uniporter (MCU) complex (MCUC) with the MCU serving as a highly selective channel that moves Ca 2ϩ ions across the IMM dependent on the mitochondrial membrane potential (⌬ m ) (21). Integrative genomics methods enabled the discovery of the molecular nature of the uniporter pore of MCU and its regulatory subunits: MCUb, EMRE, MICU1, MICU2, and MCUR1 (21)(22)(23)(24)(25)(26). MCU is the pore-forming subunit of the MCU complex (21,22). MCUb is the MCU paralog and behaves as a dominantnegative subunit affecting the Ca 2ϩ permeation properties of the uniporter (25). High-affinity interaction of the MICU1-MICU2 complex with Ca 2ϩ serves as an on-off switch, leading to a tightly controlled channel, capable of responding directly to cytosolic Ca 2ϩ signals (23,24,27). EMRE is essential for the uniporter current and brackets the MCU pore and bridges the Ca 2ϩ -sensing role of MICU1 and MICU2 with the Ca 2ϩ -conducting role of MCU (26,27). In the absence of EMRE, MCUC Ca 2ϩ conductance does not occur (26). MCUR1 binds to MCU and EMRE and functions as a scaffold factor, thus being an important member of the MCUC (28).
Other mitochondrial Ca 2ϩ import-linked proteins in the IMM have been identified (29,30) although MCUC Ca 2ϩ conductance is the dominant mechanism. CM mitochondrial Ca 2ϩ export is mediated by the mitochondrial Na ϩ /Ca 2ϩ /Li ϩ exchanger (mNCLX) preventing mitochondrial Ca 2ϩ overload (31). Short-term opening of the mitochondrial permeability transition pore also contributes to mitochondrial Ca 2ϩ release (32).
During the systolic and diastolic phase of a heartbeat, intermyofibrillar mitochondria, which are in close proximity to the SR, are exposed to the changing cytosolic Ca 2ϩ concentration ( Here we sought to investigate whether normalizing MCU levels in vivo in streptozotocin (STZ)-induced diabetic mice (annotated "DM" in the figures and table) restores mitochondrial Ca 2ϩ handling, mitochondrial function, glucose oxidation, and cardiac function. In addition, we investigated deleterious effects of MCU expression in diabetic hearts. Our results show that restoring MCU toward the normal levels in murine diabetic hearts markedly improved mitochondrial Ca 2ϩ handling, mitochondrial function, cardiac energetic metabolism, and subsequently both cardiac myocyte and heart function in the absence of adverse effects.

Animal model
To ensure the efficacy of the STZ treatment, 8 weeks postinjection, and before experimental evaluation, mouse biometrics were analyzed and recorded (Table 1). We observed significantly lower heart (HW) and body weight (BW), whereas tibia length (TL) was not found different. Heart weight/body weight (HW/BW) ratios and heart weight/tibia length (HW/TL) ratios did significantly differ between control (CTR) and diabetic mice. Glucose tolerance measurements showed significantly higher levels in diabetic mice 1 h post-challenge (1.5 mg of glucose/g of BW) compared with controls, and fasting blood glucose levels were markedly increased in diabetic mice. Consistent with this model, insulin levels were below normal in diabetic mice. Neither thyroid hormone (triiodothyronine) dysregulation nor ketones were observed in diabetic mice. Finally, significantly elevated triglyceride levels were measured

MCU improves cardiac metabolism and function in diabetes
in plasma from diabetic versus CTR mice, whereas cholesterol levels were unchanged (Table 1).

MCU and MCUC member levels are altered in diabetic hearts
MCU and MCUC member protein levels were determined in hearts of diabetic and control mice. As shown in Fig. 1, MCU protein levels were decreased by 50% in diabetic hearts 8 weeks post-STZ, compared with control hearts (Fig. 1A). Moreover, at 26 weeks we observed that MCU protein levels had decreased by 70% from control levels (Fig. 1A). We also determined MCU mRNA levels in 8-week-old diabetic mice by RT-qPCR and observed a 50% decrease, in line with the protein decay (Fig.  1B). Therefore, further studies were conducted using 8-weekold diabetic mice. Western blot analysis of other MCUC members in the 8-week diabetic heart also showed that EMRE was significantly decreased by 36%. In contrast, MCUb protein levels were increased by 31% with no significant change for MICU1, MICU2, and MCUR1 (Fig. 1C).

Adeno-associated virus-MCU (AAV-MCU) expression
To restore MCU protein levels toward normal, an adenoassociated viral vector carrying the complete coding sequence of the murine MCU mRNA (AAV-MCU) was generated in the liver-detargeted, cardiac-preferred serotype 9 variant, 9.45 (35). Cardiac-preferred MCU expression was confirmed by Western blotting in several organs from control, 3-month-old mice (Fig.  2, A and B). To ensure correct evaluation of tissue specificity mice were injected with higher AAV doses (1 ϫ 10 12 genome copies). Even though intramuscular injection (IM) of AAV-MCU was found to induce MCU expression in skeletal muscle, the jugular vein delivery approach used for this study was able to induce AAV accumulation and consequently expression only in heart. Subsequently, MCU restoration was attempted in diabetic mice. AAV-MCU was injected through the jugular vein and after 4 weeks we observed a complete MCU restoration (Fig. 2C) instead of MCU overexpression as in Fig. 2, A and B, due to the lower AAV titer used in these experiments (3 ϫ 10 11 versus 1 ϫ 10 12 genome copies). As control an empty AAV (AAV-empty) was injected. No effect of AAV-empty was observed, as reported by others (36). In the hearts of AAV-MCU-treated mice we measured 8.3 Ϯ 1.1 ϫ 10 5 AAV vector genome copy numbers per g of genomic DNA.

MCU expression restores mitochondrial Ca 2؉ handling, mitochondrial bioenergetics, and reverses energetic metabolism in diabetic mice
The effects of MCU restoration on mitochondrial Ca 2ϩ handling were evaluated in isolated CM from CTR, diabetic, and from hearts of diabetic mice receiving AAV-MCU. All mice received AAV expressing the mitochondrial Ca 2ϩ sensors Mitycam or MitoPericam (Fig. 3). Mitochondrial Ca 2ϩ uptake and release rates, and [Ca 2ϩ ] m were assessed in isolated, fluorescent, rod-shaped cells responding to electric stimulation (pacing frequency ϭ 0.3 Hz). Mitochondrial Ca 2ϩ uptake and release were assessed using Mitycam. CM from hearts of diabetic mice showed markedly reduced mitochondrial Ca 2ϩ transients with significantly lower uptake rate and slightly lower release rate compared with control, whereas MCU restoration rectified these measurements (Fig. 3A). Furthermore, [Ca 2ϩ ] m in paced-contracting CM (Fig. 3B) and the mitochondrial matrix-free Ca 2ϩ content in permeabilized CM (Fig. 3C) were consistently decreased in diabetic CM compared with controls.
[Ca 2ϩ ] m and the mitochondrial matrix-free Ca 2ϩ content were restored toward control values after MCU expression (Fig. 3, B and C).
[Ca 2ϩ ] m plays a central role in controlling the rate of activation of the tricarboxylic acid cycle dehydrogenases and is an important regulator of electron transport chain-mediated ATP production by stimulating different mitochondrial complexes and dehydrogenases during oxidative phosphorylation (5). Therefore, we evaluated the potential beneficial effects arising from improved mitochondrial Ca 2ϩ handling following MCU restoration in diabetic hearts.
The PDC activity was dramatically reduced in hearts from diabetic mice. MCU expression in diabetic mice significantly improved PDC activity (Fig. 4A). These results were consistent with a higher level of phosphorylated PDC protein in diabetic hearts that was reduced after MCU expression (Fig. 4B). Due to the importance of PDC as a key regulator of glucose and FA oxidation, we investigated whether the reduced PDC phosphorylation and increased PDC activity following MCU expression could have an impact on cardiac metabolism. To this end glucose and FA oxidation were measured in working heart preparations (Fig. 4C). We observed in diabetic hearts that the rates of mitochondria-mediated glucose oxidation and palmitate oxidation were significantly reduced and increased, respectively. Interestingly, diabetic hearts expressing transgenic MCU showed completely reversed profiles where glucose oxidation increased and FA oxidation showed a reciprocal decrease, both to control levels.
A metabolic shift was also observed in metabolomics studies. Plasma and hearts samples from CTR, diabetic, and diabetic mice transduced with AAV-MCU were analyzed. Metabolites that showed significant differences between CTR and diabetic mice were plotted across all three groups for both plasma and heart samples (Fig. 5A). In plasma, widespread changes with diabetes were observed, and MCU transgene expression had little effect as demonstrated by hierarchical clustering. In con- trast to plasma, hierarchical clustering of metabolites in heart samples demonstrated that MCU transgene expression reverted the diabetes-induced changes in a large number of metabolites. We then measured the levels of glucose in the plasma and heart from the same group of mice by LC-MS/MS analysis (Fig. 5B).
Consistent with the decreased glucose oxidation measured in working heart preparations and with the diabetic mouse model, higher levels of glucose were measured in both plasma and heart from diabetic mice. MCU transgene expression in the hearts of these mice significantly reduced glucose levels in the

MCU improves cardiac metabolism and function in diabetes
heart, but not in plasma, confirming the cardiac specificity of the transgenic therapeutic intervention and suggesting a direct link between restored mitochondrial Ca 2ϩ handling and normalized oxidative metabolism. MCU protein restoration also positively influenced mitochondrial respiration and functionality. The respiratory control ratio (RCR) (state 3/state 4) was found to be lower in isolated mitochondria from diabetic hearts compared with control ( Fig.  6A), and MCU expression rectified RCR toward control values (Fig. 6A). Furthermore, ⌬ m measured in isolated CM from diabetic mice was decreased by 40% and was returned toward normal levels in CM from diabetic mice after MCU expression (Fig. 6B). In addition, ATP production in isolated mitochondria assessed by the 2-deoxy-D-glucose (2DG) ATP energy clamp method (37) demonstrated that reduced ATP production in mitochondria from diabetic hearts was increased after MCU expression, despite persistent diabetes (Fig. 6C).

Cardiac function in diabetes is improved by MCU expression
The effect of MCU rectification on CM contractility was analyzed with the edge detection technique. A decreased rate of cell shortening/contraction (ϩdL/dt), rate of re-lengthening (ϪdL/ dt), and fractional cell shortening was observed in CM from diabetic mice, compared with control CM, and an improvement in CM from diabetic mice treated with AAV-MCU is shown in Fig. 6. ϩdL/dt and ϪdL/dt were increased by 28 and 74% (Fig. 7, A and B), respectively, in CM from diabetic mice treated with AAV-MCU, compared with CM from diabetic mice. Moreover, rectification of MCU levels improved fractional cell shortening by 86% (Fig. 7C). Subsequently, cardiac contractility was assessed ex vivo in isolated-perfused hearts. Diminished rates of contraction (ϩdP/dt) and relaxation (ϪdP/ dt) in diabetic hearts were observed and MCU transgene expression returned these parameters toward normal (Fig. 7, D and E). In vivo cardiac performance was ultimately evaluated by echocardiography and AAV-MCU-treated diabetic mice showed fractional shortening returned toward control values, despite persistent diabetes (Fig. 7F).

Restoration of normal MCU levels reduces apoptosis, protein oxidation, and cardiac infarct size
Given the role of mitochondrial Ca 2ϩ overload on a variety of processes leading to oxidative stress and cell death, we explored potential, unintended negative effects related to MCU restoration in diabetes (Fig. 8). Apoptosis was assessed by Western blot analysis of activated cleaved caspase 3 (Fig. 8A). The results showed that increased cleavage of caspase 3 in diabetic hearts was reduced toward control levels in the hearts of diabetic mice treated with AAV-MCU. Moreover, oxidation of total proteins extracted from whole hearts clearly showed diminished oxidation in AAV-MCU-treated diabetic samples (Fig. 8B). In addition, myocardial infarct size was evaluated in isolated-perfused hearts in which global ischemia/reperfusion was induced. Diabetic hearts showed an infarcted area 3.2-fold larger than in control mice. In contrast, hearts from diabetic mice treated with AAV-MCU showed a significant reduction in the infarcted area (2.2-fold versus CTR) (Fig. 8C).

Discussion
We report here for the first time that in vivo AAV-based MCU transgene expression in the CM of diabetic mouse hearts had beneficial effects on mitochondrial Ca 2ϩ handling, PDC activity, oxidative phosphorylation substrate utilization, mitochondrial respiration, ⌬ m , and ATP production, despite per-

MCU improves cardiac metabolism and function in diabetes
sistent diabetes. These beneficial effects were associated with improved cardiac performance of the heart both ex vivo, in the isolated CM and Langendorff-perfused heart, and in vivo. No detrimental effects from MCU restoration in diabetic hearts were observed. This experimental evidence was obtained using an STZ-induced diabetic mouse model that shares clinically important features with human type 1 diabetes. Indeed, STZtreated animals have been used to study diabetogenic mechanisms and for preclinical evaluation of novel anti-diabetic therapies. Although the etiopathology of diabetes in humans is largely dependent by an autoimmune mechanism the diabetes in our mouse model is chemically induced by STZ. However, diabetic cardiomyopathy can successively be achieved in STZinduced diabetic mice so our results provide the basis for further studies that could be performed in the human heart.
Mitochondrial Ca 2ϩ uptake is achieved primarily through the MCUC, which consists of a pore-forming protein MCU. The MCU pore is a highly selective Ca 2ϩ channel that moves Ca 2ϩ ions across the IMM, and interacts with its regulatory and structural subunits: MCUb, MICU1, MICU2, MCUR1, and EMRE.
Recently, the pathophysiological role of MCUC in diabetes has become a topic of investigation. We previously reported that MCU protein levels and [Ca 2ϩ ] m were decreased in diabetic mice (34), and have confirmed these results here. In addi-tion, we observed diminished mitochondrial Ca 2ϩ uptake and mitochondrial Ca 2ϩ content in diabetic CM. Regulatory members of the MCUC such as MCUb and EMRE were also influenced by diabetes. Surprisingly, we did not observe significant changes in MICU1, which was found down-regulated in the hearts of the Leptin receptor-deficient (db/db) C57BLKS diabetic mouse model (38). The same report showed no changes to MCU levels in 12-week-old db/db mouse hearts although lower MCU protein levels can be observed in their Western blotting results at 18 weeks of age (38). Differences between our diabetic model and the db/db diabetic mouse model may be expected because STZ-induced diabetic mice have very low insulin levels in contrast to the db/db diabetic model mice, which display hyperinsulinemia (39). In addition, in db/db mice signaling by the leptin receptor is abnormal (40). It is interesting to note that different diabetic models result in specific changes of MCUC members. This necessitated varying rescue efforts focusing on different MCUC members.
Our findings show decreased MCU protein levels, corresponding to reduced MCU mRNA levels (Ϫ50%). Hyperglycemia may be a contributing factor to MCU down-regulation in a STZ-induced diabetic model because we have previously demonstrated that exposing cultured neonatal CM to hyperglycemia leads to markedly decreased MCU levels (34).

MCU improves cardiac metabolism and function in diabetes
Our results and prior findings (2-4, 34, 38) strongly suggest a link between dysfunctional mitochondrial Ca 2ϩ handling and impaired cardiac function in diabetes. In this report we further investigated whether restoring mitochondrial Ca 2ϩ handling in diabetic hearts could have beneficial effects on cardiac metabolism and function, despite persistent diabetes.
In vivo AAV-mediated MCU expression in the hearts of diabetic mice restored MCU protein levels and Ca 2ϩ handling. [Ca 2ϩ ] m enhances oxidative phosphorylation leading to enhanced ATP formation (5). In addition, several dehydrogenases in the MM are activated by [Ca 2ϩ ] m , including PDC (6, 7). Accordingly, our results show that restoring mitochondrial Ca 2ϩ handling in diabetic hearts leads to positive downstream Ca 2ϩ -dependent effects. By rectifying [Ca 2ϩ ] m , normalized allosteric regulation by Ca 2ϩ of several mitochondrial respiratory chain complexes is likely the reason why we observed restoration of mitochondrial respiration, which enhances ATP production and ⌬ m . Furthermore, restored [Ca 2ϩ ] m might either directly stimulate PDC activity or influence it by activation of Ca 2ϩ -dependent mitochondrial phosphatases (6, 7). Consequently, the shift from FA to glucose oxidation observed both in the working heart preparation and with metabolomics studies is possibly mediated by the higher PDC activity in diabetic mice with MCU restoration.
Improving mitochondrial function also resulted in improved CM contractility, ex vivo cardiac contraction, and in vivo fractional shortening. Based on the data presented in this report, and consistent with a model in which cardiac metabolic substrates utilization is normalized, the observed increase in contractile function can be viewed as a consequence of a more efficient metabolism resulting from improved mitochondrial Ca 2ϩ handling. Our results are in agreement with findings from Ji et al. (38). These authors overexpressed MICU1 in a db/db diabetic mouse model resulting in alleviated diabetic cardiomyopathy by an antioxidant mechanism. MICU1 regulates the MCUC and potentially could improve mitochondrial Ca 2ϩ handling, but detailed analysis of mitochondrial Ca 2ϩ handling was not provided (38).
In heart failure that is linked to myocardial ischemia mitochondrial Ca 2ϩ overload can occur, triggering mitochondrial permeability transition pore opening and leading to CM death (41). To assure that no detrimental consequences result from MCU restoration in diabetes, where mitochondrial [Ca 2ϩ ] is restored toward normal, we determined CM death and myocardial infarct size in diabetic mice and in diabetic mice with AAV-MCU expression. Our findings show that MCU expression did not increase the infarcted area after global ischemia/ reperfusion in the diabetic heart. On the contrary, the infarcted area was reduced in diabetic heart when MCU was expressed. Improved mitochondrial respiratory function and ATP formation as a consequence of increased glucose utilization may explain the better recovery of cardiac myocytes upon ischemia/ reperfusion injury (42). These results also correlate with the reduction in cleaved caspase 3 in diabetic heart with MCU expression. In addition, increased protein oxidation in the diabetic heart was reduced by MCU expression. All together these data indicate that MCU restoration in the diabetic heart does not trigger detrimental effects nor worsen Ca 2ϩ overload-induced damage.
Only recently have the molecular components of the MCUC been identified (21)(22)(23)(24)(25)(26). The physiological and pathological relevance of the MCUC are actively investigated. The physiological role of MCU is currently debated, in response to data obtained in MCU KO models (43). For instance, early results from MCU KO models showed a very mild cardiac phenotype unless acutely challenged (43). However, recent data from new KO models for MCU produced by the International Mouse Phenotype Consortium (IMPC) showed that ablation of MCU (that completely blocks mitochondrial Ca 2ϩ uptake) leads to embryonic lethality, and MCUϩ/Ϫ heterozygous mice show several impaired physiological functions including decreased cardiac stroke volume (reviewed in Ref. 44). The paradoxical function of MCU observed in the various MCU KO models may be due to compensatory effects, alternative Ca 2ϩ import mechanisms, genetic background, or the techniques used to make the models. Currently no reports have addressed the influence of MCU in chronically stressed and challenged models such as the STZ-induced diabetic mouse. Our data demonstrate for the first time that restoring MCU protein levels in the diabetic heart promotes beneficial effects, suggesting a key role of

. MCU transgene expression increases PDC activity by reducing PDC phosphorylation thereby inducing a metabolic oxidation switch. A,
PDC activity was measured in total heart lysates using a microplate assay kit following the reduction of NAD ϩ , coupled to the reduction of a reporter dye (450 nm). Data were normalized over total protein content and are representative of 5 total mouse heart lysates. B, Western blot analysis of PDC phosphorylation (pPDC) levels (top) representative of 4 animals. Total PDC protein levels were not changed and used for normalization. Total PDC levels were normalized over VDAC levels. Summarized densitometric band analysis are shown (bottom). C, glucose and palmitic acid were used to measure glucose and fatty acid oxidation in working heart preparations. Substrate oxidation is reported as micromole/min/g dry weight. All data are presented as mean Ϯ S.D. One-way ANOVA with Tukey's multiple comparisons test was used. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. Figure 6. MCU expression improves mitochondrial respiratory dysfunction in diabetic hearts. A, mitochondrial respiration was measured by Clark electrode and the RCR was calculated according to procedures described under "Experimental procedures." Data are representative of cardiac mitochondria isolated from 4 mouse hearts per group. B, mitochondrial transmembrane potential (⌬ m ) was evaluated as the ratio of JC-1 fluorescence at 584/531 nm. Fluorescence emission was collected from isolated CM. Data are representative of at least 30 individual cell recordings from 3 mice for each group. C, rates of mitochondrial ATP production measured in mitochondria isolated from 4 mice per group. ATP production is reported as percentage of phosphorylated 2DG (2DGP) over total 2DG obtained in a 20-min reaction. Data are relative to rates registered in mitochondria from control mice and are normalized per g of mitochondria. All data are presented as mean Ϯ S.D. One-way ANOVA with Tukey's multiple comparisons test was used. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

Figure 5. Metabolomics analysis confirms the global cardiac-specific effect of AAV-MCU in diabetic mice.
A, metabolomics analysis was conducted in plasma and heart, and metabolites that differed significantly were used to generate heat maps. Hierarchical clustering was obtained using Spearman rank correlation. Each row represents one metabolite characterized by specific molecular mass. A color gradient was introduced to visualize relative metabolite levels (blue ϭ low levels, red ϭ high levels). MCU transgene expression shifted the metabolomics profile toward CTR only in the heart as evidenced by hierarchical clustering. B, glucose levels in heart and plasma samples (glucose is reported in arbitrary units, AU). All data are presented as mean Ϯ S.D. One-way ANOVA with Tukey's multiple comparisons test was used. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

MCU improves cardiac metabolism and function in diabetes
MCUC in the pathophysiology leading to diabetes-mediated decreased cardiac function.
In conclusion, restoring MCU levels in diabetic hearts improves both CM and heart metabolism and function by rectifying the abnormal mitochondrial Ca 2ϩ handling associated with diabetic cardiac disease without any detrimental effects. Therefore, mitochondrial Ca 2ϩ handling can be considered a suitable therapeutic target in diabetes-related cardiac disease.

Animals
All investigations conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication No. 85-23, revised 1985). This study was conducted in accordance with the guidelines established by the Institutional Animal Care and Use Committee at the University of California, San Diego.

Diabetic mouse model
Diabetes was induced in C57BL/6NHsd mice (25 g, 3 months old) by giving a daily intraperitoneal injection of STZ (40 mg/kg) for five consecutive days, as previously described (45). Diabetes onset was documented by measuring blood glucose above 250 mg/dl. These mice were nonketotic and euthyroid, as previously reported (46). Total cholesterol and triglycerides were measured by the UCSD Mouse Phenotyping Service according to standard procedures.

Determination of AAV vector genome copy number
Approximately 15 mg of tissue from the apex of the heart was processed using the PureLink Genomic DNA Mini Kit (Life Technologies). Vector genome copy numbers were determined by qPCR using DyNAmo ColorFlash SYBR Green (Thermo) and an Agilent Mx3005P cycler. Primers specific for the FLAGtagged MCU transgene were used for amplification (forward, 5Ј-AGGAGCCAAAAAGTCACGTTTC; reverse 5Ј-CTTATC GTCGTCATCCTTGTAATC). A standard curve was created using known copy numbers of plasmid DNA containing the transgene construct. Results are expressed as the vector copy number per microgram of genomic DNA as described (48). Hearts were paced at 400 bpm and the resulting pressure waves were recorded and analyzed. Summarized data are related to D, ϩdP/dt max and E, ϪdP/dt min are shown and demonstrate a significant improvement in DM ϩ AAV-MCU versus DM hearts. F, in vivo heart function was assessed by M-mode echocardiography and a significantly increased % fractional shortening is shown in DM ϩ AAV-MCU versus DM hearts. All data are representative of 4 animals and are presented as mean Ϯ S.D. One-way ANOVA with Tukey's multiple comparisons test was used. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

Isolation of CM
Ca 2ϩ tolerant adult CM were isolated from ventricular tissue by a standard enzymatic digestion procedure described previously (49). CM were cultured on glass coverslips coated with laminin.

Mitochondria preparation
Mitochondria were isolated as previously described (45). To ensure mitochondrial viability for ATP production and O 2 consumption assays all steps were performed at 4°C.

Langendorff-perfused hearts and global ischemia/reperfusion protocol
Heart function was assessed ex vivo as previously described by us (50,51). Pressure development was recorded digitally (1 kHz) by connecting the intraventricular balloon to a 2-F Millar pressure transducer. The hearts were paced at 400 bpm and the resulting pressure waves were analyzed for pressure derivatives (ϩdP/dt, ϪdP/dt) and pressure development. Global ischemia/ reperfusion experiments were performed as described previously (52). Hearts were allowed to stabilize for 30 min and subsequently, perfusion flow was stopped for 15 min followed by 60 min reperfusion. Hearts were frozen and cut into 6 -8 transverse slices from apex to base of equal thickness (1 mm). Heart slices were stained with 10% tetrazolium chloride for 30 min and infarct size was determined by computer morphometry using ImageJ software.

Transthoracic M-mode echocardiography
Cardiac echography was performed on lightly anesthetized mice using a FUJIFILM Visualsonics Vevo 2100 ultrasound system, as described (53).

Measurement of CM contractility by edge detection
The edge detection of cultured adult CM was performed according to the method described (49).

Figure 8. MCU expression reverts caspase 3 activation and protein oxidation in diabetic hearts and does not increase ischemia/reperfusion damage.
A, Western blot analysis of cleaved (activated) caspase 3 as a general marker of apoptosis (left). VDAC was used as protein loading control. Summarized densitometric band analysis is shown (right). Data are representative of total heart protein lysates from 4 animals per group. B, Western blot analysis of total protein oxidation (left) as marker of oxidative stress in heart lysates. Polyvinylidene difluoride membranes were stained for total protein as loading control. Summarized densitometric band analysis is shown (right). Data are representative of 4 animals per group. C, influence of MCU expression on mouse heart infarct size following global ischemia/reperfusion. Following ischemia/reperfusion, hearts were sectioned and photographed. The infarcted area is highlighted in bright red around the perimeter of the section using ImageJ (top). Dark red represents viable tissue. Infarct size (% infarcted area over total area) was calculated and is reported (bottom). The experiment was performed on 3 hearts per group. All data are presented as mean Ϯ S.D. One-way ANOVA with Tukey's multiple comparisons test was used. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

Protein oxidation
Protein oxidation assays from whole heart lysates were carried out using the OxyBlot Protein Oxidation Detection Kit (Millipore S7150) following the manufacturer's instructions.

[Ca 2؉ ] m in intact CM with MitoPericam
Data were collected from paced-contracting individual CM from the emission channel at a rate of 20 Hz and the ratio between the intensities of the two excitations (410/485) was calculated providing for relative comparisons of the [Ca 2ϩ ] m between experimental treatments. [Ca 2ϩ ] m was determined as described (56). Calibration of the MitoPericam signal to molar [Ca 2ϩ ] m was done as described (56). Basal [Ca 2ϩ ] m was obtained from paced myocytes averaging diastolic and systolic [Ca 2ϩ ] m , which are the minimum and maximum pericam ratios, respectively, that were measured during contraction.

Mitochondrial Ca 2؉ transients in intact CM with Mitycam
Fluorescence was measured with excitation at 488 nm (emission between 545 Ϯ 50 nm) using the system described for MitoPericam. Mitycam measurement was performed as described by Lu et al. (57). Analysis of mitochondrial Ca 2ϩ transients was performed as described (34,58). Ca 2ϩ uptake is the maximum slope of the upstroke. Ca 2ϩ release is the maximum slope during Ca 2ϩ decline.

Ca 2؉ content in permeabilized CM
Ca 2ϩ content in permeabilized CM was performed as described (59). Isolated adult CM were transferred to a Ca 2ϩfree medium containing thapsigargin (sarcoplasmic reticulum Ca 2ϩ -APTase inhibitor), digitonin, protease inhibitors, and succinate, pH 7.2. Digitonin-permeabilized CM were treated with Ru360 (MCU inhibitor) and CGP-37157 (mNCLX inhibitor) to inhibit mitochondrial Ca 2ϩ exchange. FCCP was added to disrupt the ⌬ m and release all free mitochondrial matrix Ca 2ϩ and detected by Fura-2.

⌬ m in isolated CM
⌬ m was measured deploying the cationic carbocyanine JC-1 dye, as described previously (60).

PDC activity
Pyruvate dehydrogenase complex activity was assessed from whole heart homogenates using the PDC Enzyme Activity Microplate Assay Kit from Abcam (ab109902) following the manufacturer's instructions.

Mitochondrial O 2 respiration
Mitochondrial respiration was measured with a Clark-type electrode (YSI) as described (61). Basal respiration was measured with succinate (10 mM) as substrate, 2 M rotenone was present in the basal condition (state 4). 4 mM ADP was added to produce state 3 respiration, as previously described (62). RCR was calculated as ratio of state 3:state 4 respiration.

NMR analysis of mitochondrial ATP production
ATP production was monitored using the 2DG ATP energy clamp method as described (37).

Energy substrate metabolism in the Neely working heart model
Glucose and fatty acid metabolism were measured in isolated working hearts as described previously (8). Briefly, hearts were perfused with a modified Krebs-Henseleit buffer containing 11 mM glucose and 0.8 mM palmitate bound to 3% BSA (fatty acid free). [5-3 H]Glucose was used to measure glucose oxidation and [9,10-3 H]palmitate to measure fatty acid oxidation.

Metabolomics analysis
LC-MS/MS analysis was used to measure metabolites. Blood was harvested from heparinized mice before heart excision and plasma was extracted using standard protocols. Excised hearts were rinsed in ice-cold PBS and snap frozen. Frozen heart samples were transferred to a 2.0-ml impact-resistant tube containing 200 mg of 1-mm zirconium beads. To each tube, 50 l/mg of ice-cold 80:20 methanol:water was added. This was followed by two 3 ϫ 15-s homogenization cycles at 6,400 Hz in a Precellys 24 tissue homogenizer. For the plasma, 40 l was added to 160 l of ice-cold 80:20 methanol:water to extract metabolites and precipitate protein. The heart and plasma samples were then placed in the Ϫ20°C freezer for 30 min to allow for precipitation of protein. Samples were thereafter vortexed for 30 s, centrifuged at 14,000 ϫ g for 10 min at 4°C, and supernatants were transferred to LC-MS vials containing a 200-l glass inserts. All samples were kept at 4°C in the autosampler compartment until 2 l was injected for analysis. LC-MS-based metabolomics analysis was performed using a Thermo QExactive orbitrap mass spectrometer coupled to a Thermo Vanquish UPLC system. Chromatographic separation of metabolites was achieved using a Millipore (Sequant) Zic-pHILIC 2.1 ϫ 150mm, 5-m column maintained at 25°C using a flow rate of 0.3 ml/min. Compounds were eluted via a 19-min linear gradient starting from 90:10 acetonitrile: 20 mM ammonium bicarbonate to 45:55 acetonitrile: 20 mM ammonium bicarbonate. A Thermo Q-Exactive orbitrap mass spectrometer was operated in positive and negative ion modes using a heated electrospray ionization (HESI) source at 35,000 resolution, 100-ms ion trap time for MS1 and 17,500 resolution, 50-ms ion trap time for MS2 collection. Data were collected over a mass range of m/z 67-1000, using a sheath gas flow rate of 40 units, auxillary gas flow rate of 20 units, sweep gas flow rate of 2 units, spray voltage of 3.5 and 2.5 kV for positive and negative ion modes, respectively, capillary inlet temperature of 275°C, auxillary gas heater temperature of 350°C, and an S-lens RF level of 45. For the MS2 collection, MS1 ions were isolated using a 1.0 m/z window and fragmented using a normalized collision energy of 35. Fragmented ions were placed on dynamic exclusion for 30 s before being allowed to be fragmented again. Collected data were imported into the mzMine 2.20 software suite for analysis. A Log2-fold change adjustment followed by a z-score adjustment was done. The metabolites plotted were filtered by doing a t test between the control and diabetic groups and the metabolites that had a p value below 0.05 were further analyzed. From this heat maps for plasma and heart were generated using Morpheus by the Broad Institute (https://software.broadinstitute.org/morpheus/). 4 Hierarchical clustering between samples was done by a Ϫ1 Spearman rank correlation. Distance between clusters was measured by average. Glucose was identified using comparison of accurate mass, retention time, and MS2 features with pure standards, with relative quantitation of levels using MS1 intensity values.

Statistical analysis
Results are presented as mean Ϯ S.D. One-way ANOVA with appropriate post hoc test, or unpaired Student's t test for comparison between two groups, were used. Gaussian distribution was always assumed. p Ͻ 0.05 was considered to be statistically significant.