Mitochondrial Energetics in the Heart in Obesity-Related Diabetes: Direct Evidence for Increased Uncoupled Respiration and Activation of Uncoupling Proteins

Male db/db mice (C57BLKS background) and lean C57BLKS controls, obtained from The Jackson Laboratories (Bar Harbor, ME) were studied between the ages of 8 and 9 weeks. At this age, db/db mice are consistently severely hyperglycemic, and this age range corresponds to that at which we previously studied substrate metabolism and cardiac efficiency (9). Mice were maintained with 12-h light/12-h dark photoperiods with free access to water and food and studied in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of Utah. All studies were performed in random-fed animals.

Samples, taken from the endocardium of the left ventricle, were fixed in 2.5% glutaraldehyde and 1% paraformaldehyde, postfixed in 2% osmium, embedded in resin, and sectioned (80–100 nm thick). Morphometric analysis was assessed using blind counting to determine mitochondria number per myofibrillar surface. Three sections from each of three separate hearts per genotype were quantified at magnification of ×8,000.

DNA was extracted and purified using the DNeasy Tissue kit (Qiagen, Valencia, CA). Primer pairs were designed using GenBank reference sequences for cytoplasmic β-actin and mitochondrial cytochrome b (for primer sequences, see supplementary Table S1 in the online appendix [available at http://dx.doi.org/10.2337/db07-0481]). Real-time PCR was performed using an ABI Prism 7900HT instrument (Applied Biosystems, Foster City, CA) in 384-well plate format with SYBR Green I chemistry and ROX internal reference (Invitrogen, Carlsbad, CA). All reactions were performed in triplicate. Relative quantification was performed by interpolating crossing point data on an independent standard curve, thereby accounting for any difference in amplification efficiency. Mitochondrial DNA was expressed relative to nuclear DNA.

Mitochondrial H₂O₂ generation was measured by monitoring H₂O₂-induced fluorescence of homovanillic acid (excitation wavelength 312 nm, emission wavelength 420) under the catalysis of horseradish peroxidase using a spectro-fluorophotometer (RF5301PC; Shimadzu, Columbia, MD) as previously described by Barja (29). Succinate (4 mmol/l) was used to stimulate ROS production after inhibition of the F₁F₀-ATP synthase with oligomycin (1 μg/ml). The inhibitor rotenone (10 μmol/l) was then added to the mixture to stop complex I–mediated superoxide production. Although the readout for this assay is H₂O₂, it is widely accepted that this method assays mitochondrial superoxide production.

Lipid peroxidation was determined by measuring MDA using a Lipid Peroxidation Assay kit (Calbiochem-Novabiochem).

Cardiac function in the presence or absence of fatty acid was studied in paced (360 bpm), isolated hearts from db/db and control mice (perfusion pressure of 60 mmHg for 20 min) using Langendorff retrograde preparations as described previously (7).

Saponin-permeabilized cardiac fibers isolated from the left ventricle endocardium were used to measure respiratory parameters of the total mitochondrial population in situ, using protocols that we previously described (30,31). Studies were performed in hearts perfused with glucose alone (11 mmol/l) or glucose (11 mmol/l) and palmitate (1 mmol/l) to mimic the in vivo milieu of db/db mice (9). Fibers were exposed to three independent substrates: 5 mmol/l glutamate, 10 mmol/l pyruvate, or 0.02 mmol/l palmitoyl-carnitine (all with 2 mmol/l malate). ATP synthesis rates were determined under state 3 conditions by sampling the respiratory buffer every 10 s for 60 s immediately after addition of ADP (7). To determine potential mediators of mitochondrial uncoupling, saponin-permeabilized cardiac fibers under state 4 conditions (1 μg/ml oligomycin) and in the presence of succinate (4 mmol/l), rotenone (10 μmol/l), and either guanosine diphosphate (GDP) (500 μmol/l) alone or combined with atractyloside (ATR) (5 μmol/l) were analyzed.

Mitochondria were prepared from whole hearts by differential centrifugation at 4°C, and proton leak kinetics measured in succinate-stimulated mitochondria as previously reported (25,27). Proton leak measurements were performed in hearts that were preperfused for 20 min with 11 mmol/l glucose and 1 mmol/l palmitate prebound to 3% (wt/vol) BSA. Proton leak was measured in the absence (control) or presence of GDP (500 μmol/l) alone or combined with ATR (5 μmol/l).

Total or mitochondrial proteins were resolved by SDS-PAGE as described previously (7). Antibodies used were mouse anti-Oxphos complex II (the 30-kDa protein) and anti-Oxphos complex V (F1 complex, α-subunit) (Molecular Probes-Invitrogen, Carlsbad, CA), mouse anti–manganese (mitochondrial) superoxide dismutase (BD Biosciences, San Jose, CA), rabbit anti-4-HNE–Michael adducts (Calbiochem, La Jolla, CA), rabbit anti-UCP3 (Affinity Bioreagents, Golden, CO), and rabbit polyclonal anti-ANT1 serum (provided by Douglas Wallace, University of California, Irvine, CA). For loading controls, mouse anti–α-tubulin (Sigma, St. Louis, MO) was used for heart proteins and Coomassie Blue R-250 (Bio-Rad, Hercules, CA) staining for mitochondrial proteins. UCP3KO and ANT1KO heart lysates were used as negative controls for UCP3 and ANT1 Western blots.

Total RNA was extracted from hearts with TRIzol reagent (Invitrogen), purified with the RNEasy kit (Qiagen), and reverse transcribed. Equal amounts of cDNA from the hearts of six mice were subjected to real-time PCR as described for mitochondrial DNA. Data were normalized by expressing them relative to the levels of the invariant transcript Lamin A.

Data are means ± SE. Significance was determined by ANOVA followed by Fisher's least protected squares test using Statview 5.0.1 software (SAS Institute, Cary, NC). Nonparametric Mann-Whitney test was used to compare cardiac efficiency between glucose-perfused db/db and wild-type hearts. Proton leak curves were log transformed so that a general linear model is applicable. Comparisons of the slopes and intercepts were determined by a t test, and P < 0.05 was considered significant.

Heart weights were not different between the two groups (Table 1). In glucose-perfused db/db hearts, left ventricular–developed pressure (LVDP) and rate pressure product (RPP) were significantly reduced (P < 0.05). MVo₂ was unchanged, and thus cardiac efficiency was lower in db/db mice versus controls (P < 0.05). Similar to previous reports (6,7), perfusion with glucose and palmitate reduced contractile function (systolic pressure, LVDP, and RPP) in wild types, and MVo₂ proportionately decreased (Table 1) so that cardiac efficiency was similar in glucose- and glucose-plus-palmitate–perfused control hearts. Fatty acid exposure caused a greater decline in contractile function in db/db hearts (Table 1). Compared with equivalently perfused wild-type hearts, MVo₂ was 1.4 ± 0.1-fold higher in glucose-plus-palmitate–perfused db/db hearts, resulting in a 50.6 ± 7.2% reduction in cardiac efficiency (Table 1).

By electron microscopy, ultrastructure of sarcomeric units in db/db hearts was not grossly different from wild types except for increased intracellular lipid droplets in db/db hearts (Fig. 1A; supplementary Figure S5). Cardiac steatosis was also confirmed by oil-red-O staining (supplementary Figure S5) and by direct measurement of myocardial triglyceride content, which was 2.2-fold higher in db/db hearts (7.4 ± 1 vs. 16.3 ± 1.4 μmol/g wet heart wt, P < 0.005). Evidence for mitochondrial biogenesis was also observed (Fig. 1A). Mitochondrial number was 1.36 ± 0.1-fold and mitochondrial DNA copy 1.4 ± 0.09-fold higher in db/db hearts (Fig. 1B and C). Mitochondrial volume density was also increased by 18% in db/db hearts (45.5 ± 3.6 vs. 37.6 ± 2.1%, P < 0.05). The fold increase in volume density was ∼50% of the increase in mitochondrial number, therefore indicating that the mitochondrial biogenic response in the db/db heart is associated with mitochondria that are somewhat smaller.

In the presence of oligomycin alone, H₂O₂ generation was not increased in db/db mitochondria. Succinate increased H₂O₂ production to a greater extent in db/db mitochondria (0.55 ± 0.11 vs. 0.31 ± 0.06 μmol/l · min⁻¹ · mg⁻¹ mitochondrial proteins, P < 0.01). Rotenone (complex I inhibitor) reduced H₂O₂ production by 52% in wild-type and by 39% in db/db hearts (Fig. 2A), so that H₂O₂ generation remained 2.3-fold greater in db/db mitochondria than similarly treated wild-type mitochondria (P = 0.06) and was significantly increased relative to db/db mitochondria exposed to oligomycin alone (P < 0.03). In contrast, rotenone returned H₂O₂ generation in wild-type hearts to levels that were unchanged relative to mitochondria exposed to oligomycin alone (P = 0.3). Thus complex I is the major source of ROS in succinate-treated mitochondria from wild-type hearts, whereas ROS production in db/db mitochondria arises not only from complex I but from additional complexes. Mitochondrial ROS generation was associated with increased lipid and protein peroxidation. MDA (1.2 ± 0.01-fold, P < 0.005) and 4-HNE protein adducts (P < 0.05) were increased in db/db hearts (Fig. 2B and C) despite a modest increase in mitochondrial superoxide dismutase protein (Fig. 2C).

In saponin-permeabilized fibers isolated from glucose-perfused hearts, state 2 (V₀) and ADP-stimulated respiration rates (state 3; V_ADP) with glutamate, pyruvate, and palmitoyl-carnitine were depressed in db/db hearts (Fig. 3A–C). ATP synthesis rates were proportionately reduced in db/db hearts by 58.1 ± 5.3 and 35.9 ± 4.8% with glutamate and pyruvate, respectively (Fig. 3D and E), so that ATP-to-O ratios were not different from controls for any substrate (Fig . 3D–F). Reduced ATP synthesis is unlikely due to differences in endogenous ATPases, because ATP hydrolysis rates were not increased in db/db fibers (supplementary Figure S2). We then measured protein levels of several subunits of the electron transport chain: complex I (subunit 9), complex II (30-kDa protein), complex III (core I), and complex V (F1 complex, α-subunit). Complex I, II, and III protein levels were unchanged in db/db hearts (data not shown). However, levels of the α-subunit of the F1 ATP synthase complex were reduced by 41 ± 3% in db/db hearts (P < 0.05) (Fig. 3G).

Mitochondria from db/db hearts examined after palmitate plus glucose perfusion showed increased state 4 (V_Oligo) respirations with palmitoyl-carnitine compared with similarly perfused controls (Fig. 4B). State 3 (V_ADP) respirations, which were reduced in glucose perfused db/db hearts, actually increased after perfusion with palmitate plus glucose (13.06 ± 0.9 vs. 16.38 ± 1.27 nmol O₂ · min⁻¹ · mg⁻¹ dry wt, P = 0.05) (Fig. 4A). This contrasts with state 3 (V_ADP) respiration rates from wild-type fibers, which were reduced slightly after palmitate plus glucose perfusion. Despite increased state 3 respirations, ATP production was lower in db/db hearts (Fig. 4C), and ATP-to-O ratios were reduced by 36 ± 7% (Fig. 4D), consistent with fatty acid–mediated mitochondrial uncoupling in db/db mouse hearts.

Cardiac fibers from fatty acid–perfused db/db hearts were studied under state 4 conditions in the presence of succinate and rotenone (an inhibitor of complex I) in the presence or absence of specific inhibitors of UCPs (GDP) and the ANT (ATR). Before the addition of GDP or ATR, oxygen consumption was similar in both groups of mice, suggesting equivalently uncoupled respirations under these conditions. GDP reduced respiration rates in db/db and controls (P < 0.001). Interestingly, in the presence of GDP, O₂ consumption remained 37% higher in db/db compared with controls (8.1 ± 0.8 vs. 5.9 ± 0.6 nmol · min⁻¹ · mg dry wt⁻¹, P = 0.05) (Fig. 5A). Addition of ATR abolished this increase in db/db but had no additional effect in controls, suggesting that ANT may mediate a component of the mitochondrial uncoupling observed in db/db mitochondria.

We next performed proton leak kinetics analyses in mitochondria isolated from glucose-plus-palmitate–perfused hearts to more directly evaluate mitochondrial uncoupling (for means ± SE for all proton leak experiments and log-transformed regression equations, see supplementary Tables S2 and S3). Figure 5B–F illustrates proton leak kinetics before and after the serial addition of GDP and ATR. Addition of GDP did not alter proton leak kinetics in wild-type mitochondria. Intriguingly, a leftward shift in the wild-type proton leak curve (P < 0.03 for intercepts) suggested that ATR increased proton leak in wild-type mitochondria (Fig. 5B). In contrast, treatment with GDP reduced proton leak in db/db mitochondria (rightward shift in proton leak curves, P < 0.002 for slope and intercept; Fig. 5C). In the presence of GDP plus ATR, proton leak remained significantly different from untreated db/db mitochondria (P < 0.04, for intercepts) and tended to be shifted to the right versus GDP alone (P = 0.08 for slope and P = 0.1 for intercept; Fig. 5C). Figure 5D–F compares proton leak kinetics between wild-type and db/db mitochondria under similar treatment conditions. In the absence of inhibitors (GDP and ATR), db/db mitochondria (red circles) exhibit increased proton leak (Fig. 5D), particularly at membrane potentials <150 mV (P < 0.05 and 0.03, respectively, for slopes and intercepts). GDP had no effect in controls but shifted db/db curves to the right (Fig. 5B and C), restoring proton leak to wild-type levels (Fig. 5E). Because ATR had opposite effects in db/db and wild-type mitochondria, in the presence of GDP and ATR, intercepts of the proton leak kinetics curves were different (P < 0.02) (Fig. 5F). Taken together, these data indicate that proton leak is increased in db/db mitochondria and is mediated to a large extent by activation of UCPs, with a small component mediated by ANT.

UCP3 protein was not increased in db/db hearts (Fig. 6A), nor was there any difference in mRNA levels of UCP2 (Fig. 7). ANT proteins were 33% lower in db/db relative to controls (Fig. 6B). Thus mitochondrial uncoupling likely occurs via increased activation that is independent of changes in protein levels. Additional genes were measured to explore transcriptional mechanisms that might further elucidate mechanisms responsible for the fatty acid–mediated uncoupling in db/db mitochondria (Fig. 7). Expression levels of medium-chain acyl CoA dehydrogenase, long-chain acyl CoA dehydrogenase, PPAR-α, hydroxyacyl CoA dehydrogenase α- and β-subunits, mitochondrial acyl-CoA thioesterase 1, and PPAR-γ coactivator 1 α (PGC1-α) were increased in db/db hearts, consistent with roles in activating FAO and increasing mitochondrial biogenesis. Despite increased PPAR-α gene expression, UCP3 expression was reduced (P < 0.05), UCP2 levels were unchanged, and ANT1 expression trended upward in db/db hearts (P = 0.08). The expression of electron transfer flavoprotein α- and β-subunits (ETFA and ETFB, respectively) that deliver reducing equivalents generated in β-oxidation to the electron transport chain were increased in db/db hearts. In contrast, there was no coordinate increase in oxidative phosphorylation subunit gene expression.

We have performed a comprehensive analysis of the mitochondrial phenotype in the hearts of db/db mice, a model of obesity and severe type 2 diabetes. Our study yielded a number of novel observations. Diabetes and obesity are associated with mitochondrial proliferation in the heart. This observation is consistent with a recent study showing increased mitochondrial proliferation in insulin-resistant UCP-DTA mice (32). This mitochondrial proliferation in db/db hearts may be driven in part by increased expression of PGC1-α. Although an increase in PGC1-α and PPAR-α signaling may promote increased expression of FAO genes, and although increased expression of electron transfer flavoprotein genes may increase the delivery of reducing equivalents to the electron transport chain, there is not a uniform upregulation of oxidative phosphorylation gene expression. These findings are consistent with the hypothesis that increased delivery of reducing equivalents from FAO coupled with reduced ability to completely oxidize these equivalents might contribute to increased mitochondrial ROS generation, which leads to mitochondrial uncoupling. In isolated hearts, addition of fatty acid to perfusates increases oxygen consumption in db/db hearts without a proportionate increase in cardiac work. Although db/db mitochondria exhibit reduced oxidative capacity, they become significantly uncoupled in fibers isolated from hearts that were exposed to fatty acid. Uncoupling was further supported by increased proton leak mediated predominantly by activation of UCPs, with minor contributions from ANT that were independent of increased protein levels. Mitochondrial uncoupling may initially represent an adaptative response to increased FAO and fatty acid–mediated ROS generation; however, it does not completely normalize mitochondrial ROS overproduction, as evidenced by accumulation of lipid peroxidation products such as MDA and 4HNE. Moreover, it could also be argued that the negative impact of mitochondrial uncoupling to reduce mitochondrial ATP supply might ultimately prove to be maladaptive in terms of maintaining myocardial high-energy phosphate reserves.

Mitochondrial biogenesis represents an adaptive mechanism by which oxidative capacity can be increased in tissues in response to various stimuli. In hearts, upregulation of PGC-1α–mediated signaling plays an important role in metabolic maturation from infancy to adulthood and in increasing fatty acid oxidative capacity in response to physiological cardiac hypertrophy, as occurs during development or after exercise (33,34) (E.D.A., unpublished observations). In these contexts, PGC-1α increases oxidative capacity and mitochondrial antioxidant defense mechanisms (35). Cardiac mitochondrial biogenesis has been described in a mouse model of type 1 diabetes (36), and we now show that this also occurs in obesity and type 2 diabetes. Whereas these mitochondrial changes might be driven in part by increased PGC1-α, they are not associated with a coordinate increase in oxidative phosphorylation subunit gene expression or increased mitochondrial oxidative capacity. Moreover, increased ROS production coupled with increased protein and lipid peroxidation implies that ROS defenses in these diabetic hearts were not sufficiently increased to circumvent the increased oxidant burden. A number of potential mechanisms might contribute to this maladaptive phenotype. First, increased PPAR-α expression coupled with increased delivery of fatty acid ligands might activate FAO pathways disproportionately in db/db hearts (37,38). Second, it is possible that additional transcriptional targets must be activated to fully execute a coordinated response to increased PGC-1α expression. Support for this comes from studies in which forced overexpression of PGC-1α in the heart leads to mitochondrial biogenesis that is maladaptive (39,40). The specific nature of these additional factors remains to be determined, and we would hypothesize that activation of these pathways are impaired in the diabetic heart.

The present study confirms the existence of oxidative stress in diabetic hearts (41–44). Direct evidence is provided that mitochondria are a likely source of increased ROS, given increased H₂O₂ production in db/db mitochondria. Complex I contributes to succinate-induced mitochondrial ROS generation in wild-type and db/db mitochondria. In contrast, a significant increase in ROS generation persisted in rotenone-treated db/db mitochondria. This would suggest that additional downstream mechanisms contribute to increased ROS generation in db/db mitochondria. Gene expression studies raise the possibility that increased transfer of electrons from reducing equivalents generated by β-oxidation and succinate could contribute to increased mitochondrial ROS generation in db/db mitochondria downstream of complex I. ETFA, ETFB, and ETFQO play an important role in electron transfer between Acyl-CoA substrates to ubiquinone (45), linking the oxidation of fatty acids and some amino acids to the mitochondrial respiratory system. The electron flow can be summarized as follows: Acyl-CoA → Acyl-CoA dehydrogenases → ETF (A/B) → ETFQO → ubiquinone → complex III (46). Thus, increased delivery of electrons to complex III in fatty acid–perfused db/db hearts coupled with reduced activity of ATP synthase could contribute to increased ROS generation at or downstream of complex III. Mitochondria are also targets of oxidative damage in the hearts of diabetic animal models (36,47), which could also contribute to the impaired mitochondrial function observed.

We observed two distinct mitochondrial functional defects in db/db mice. By removing the confounding effects of fatty acid, a significant defect in mitochondrial oxidative capacity was revealed in glucose-perfused db/db mouse hearts in part due to reduced expression of the α-subunit of the F1 complex of ATP synthase. Although protein levels of selected subunits of the electron transport chain were unaltered, we cannot exclude reduction in activity of other complexes. Second, mitochondrial uncoupling was evident only after palmitate perfusion of db/db hearts. Potential mechanisms for fatty acid–mediated mitochondrial uncoupling include upregulation of β-oxidation pathways and increased capacity of db/db hearts to metabolize fatty acid (6,7,9), leading to increased generation of FADH2 (reduced flavin adenine dinucleotide) and NADH and increased expression of electron transfer flavoproteins, which could increase the delivery of electrons from these reducing equivalents to the electron transport chain, thereby increasing ROS production that activates UCPs. Additional changes in complex I and III in db/db hearts likely also increase ROS generation as well.

Importantly, our proton leak studies show that uncoupling mechanisms are activated in db/db hearts despite the lack of a significant increase in the expression levels of UCPs and ANT. Using specific inhibitors, we show that mitochondrial uncoupling in db/db hearts was GDP sensitive, implicating UCPs. Future studies using UCP3- or UCP2-null animals will be required to determine which UCP isoform(s) mediate these effects. ATR had a small additive effect, suggesting that ANT may also play a role in fatty acid–mediated uncoupling in db/db mouse hearts. These data support the existence of activators of proton conductance that were present only in the hearts of diabetic animals, which activate UCPs and ANT, leading to mitochondrial uncoupling. Our data suggest that these activators could be increased ROS and increased levels of oxidative products, such as 4-HNE, which have been shown to contribute to mitochondrial uncoupling via ANT and UCPS (18,25). Others have suggested that UCP expression levels are increased in the hearts of various models of diabetes, including db/db mice (11). Although the basis for the lack of increased UCP content or gene expression levels in our hands are not immediately obvious, our observations that uncoupling mechanisms are clearly activated in these mice would indicate that in those instances where UCP content is also increased, then the combination of increased UCP content and increased activation should have a synergistic effect.

Recent studies in mice and humans have confirmed that cardiac efficiency is reduced in obesity and diabetes. In some studies, this has been associated with decreased cardiac function (6,7,9) and in others with preserved cardiac function (5,8). However, in all of these studies, FAO and MVo₂ were increased. The present studies confirm that cardiac efficiency is reduced in db/db hearts perfused with either glucose or glucose plus fatty acid. The reduction in cardiac efficiency in glucose-perfused db/db hearts might reflect the mobilization of the endogenous triglyceride pool, which is increased in these hearts (48). Addition of palmitate reduces cardiac efficiency even further in db/db mice, which we postulate is due to mitochondrial uncoupling, which would further limit myocardial ATP supply. Our analyses of ex vivo cardiac function revealed impaired contractile function in 9-week-old db/db mouse hearts. Echocardiography revealed no significant differences in in vivo cardiac function of these mice (data not shown). The absence of significant contractile dysfunction in vivo is consistent with a number of studies that have indicated that there is an age-related impairment in function in db/db hearts (9). Systolic and diastolic dysfunction by echocardiography are present in 12-week-old but not in 6-week-old mice (49). Using gated magnetic resonance imaging analysis in db/db mice at 5, 9, 13,17, and 22 weeks of age, increased end diastolic volume developed only at 13 weeks, impaired left ventricle contractility (PV loop analysis) at 15 weeks, and reduced left ventricle ejection fraction at 22 weeks (50). Using micro-tipped conductance catheters, Van den Bergh et al. (51) observed impaired load-independent left ventricle function in 24-week-old but not in 12-week-old db/db mice, whereas conventional load-dependent parameters, such as dP/dt (peak rate of left ventricular pressure change), were normal at both ages because of increased preload and decreased afterload. Thus the present study demonstrating mitochondrial uncoupling in 9-week-old db/db mice and our prior report indicating that reduced cardiac efficiency may occur in isolated hearts as early as 4 weeks of age (9) indicate that altered myocardial energetics are an early characteristic of these hearts, which can be detected using in vitro preparations that clearly precede measurable alterations in in vivo cardiac function.

In conclusion, this study provides important new information regarding the multiple mechanisms that impair mitochondrial energetics in the heart in diabetes and obesity and provides direct evidence that mitochondrial uncoupling does occur in diabetic hearts. We propose that mitochondrial uncoupling may represent an important mechanism by which diabetic hearts may initially adapt to the characteristic increase in myocardial fatty acid metabolism and increased oxidative stress with the untoward consequence of reduced cardiac efficiency and lower ATP generation. Defective energy metabolism in the heart is likely to impair energy requiring processes in the heart such as diastolic relaxation and may increase the susceptibility of diabetic hearts to injury in the context of ischemia and cardiac hypertrophy.

S.B. has received a postdoctoral fellowship from the Juvenile Diabetes Foundation. V.G.Z. has received a postdoctoral fellowship from the American Heart Association. E.D.A. has received National Institutes of Health Grants RO1-HL-73167 and UO1-HL-70525, support from the Ben and Iris Margolis Foundation, and is an Established Investigator of the American Heart Association.

We thank Dr. Bradford B. Lowell for providing us with UCP3 knockout mice. We acknowledge Prof. Douglas C. Wallace for providing hearts from ANT knockout mice and ANT1 antiserum and Adrian Flier for technical support with the ANT Western blot.

Finally, this work is dedicated to the memory of one of the co-authors, Josie I. Johnson, whose life was prematurely cut short in a biking accident before the publication of these studies.