Adaptation of Myocardial Substrate Metabolism to a Ketogenic Nutrient Environment*

Heart muscle is metabolically versatile, converting energy stored in fatty acids, glucose, lactate, amino acids, and ketone bodies. Here, we use mouse models in ketotic nutritional states (24 h of fasting and a very low carbohydrate ketogenic diet) to demonstrate that heart muscle engages a metabolic response that limits ketone body utilization. Pathway reconstruction from microarray data sets, gene expression analysis, protein immunoblotting, and immunohistochemical analysis of myocardial tissue from nutritionally modified mouse models reveal that ketotic states promote transcriptional suppression of the key ketolytic enzyme, succinyl-CoA:3-oxoacid CoA transferase (SCOT; encoded by Oxct1), as well as peroxisome proliferator-activated receptor α-dependent induction of the key ketogenic enzyme HMGCS2. Consistent with reduction of SCOT, NMR profiling demonstrates that maintenance on a ketogenic diet causes a 25% reduction of myocardial 13C enrichment of glutamate when 13C-labeled ketone bodies are delivered in vivo or ex vivo, indicating reduced procession of ketones through oxidative metabolism. Accordingly, unmetabolized substrate concentrations are higher within the hearts of ketogenic diet-fed mice challenged with ketones compared with those of chow-fed controls. Furthermore, reduced ketone body oxidation correlates with failure of ketone bodies to inhibit fatty acid oxidation. These results indicate that ketotic nutrient environments engage mechanisms that curtail ketolytic capacity, controlling the utilization of ketone bodies in ketotic states.

acids for contribution to tricarboxylic acid cycle flux. In the short term, ketone bodies may serve as an energetically favorable carbon source to the myocardium (27), but data on long term myocardial utilization of ketone bodies are lacking. Here, we measure the influence of a ketogenic milieu on in vivo and ex vivo myocardial metabolism.

EXPERIMENTAL PROCEDURES
Animals-Specific pathogen-free C57Bl/6J wild-type and PPAR␣ Ϫ/Ϫ mice (Jackson Laboratory) were maintained on ␥-irradiated standard polysaccharide-rich chow (LabDiet number 5053; 62% of calories from carbohydrate, 13% from fat, 25% from protein), autoclaved water ad libitum, and standard housing conditions, including sawdust bedding and a 12-h lightdark cycle (lights on at 0600 h). Bedding was changed at the onset of fasting for mice fasted for 24 h (initiated at 0830 h); water was provided ad libitum to fasting animals. ␥-Irradiated ketogenic diet (Bio-Serv AIN-76A; 95% of calories from fat, 4.5% from protein, and 0.5% from carbohydrate; diet is fortified with essential micronutrients) was initiated in mice no younger than 7 weeks old and was maintained for 5 weeks. See "Neonatal Rat Cardiomyocyte Culture" below for a description of the use of rats. All experimental methods were approved by the Animal Studies Committee at Washington University.
Serum Metabolite Measurement-Glucose, total ketone bodies, ␤OHB, and free fatty acid concentrations were determined from serum using biochemical assay kits according to the manufacturer's instructions (Wako). AcAc concentrations are derived from total ketones and ␤OHB. Triglyceride levels were determined using a biochemical assay (ThermoFisher), and insulin concentrations were determined by an enzyme-linked immunosorbent assay (Millipore).
Gene Expression Profiling-Myocardial RNA isolation, target preparation, and hybridization to Affymetrix MOE 430 2.0 mouse GeneChip arrays was performed as described previously (28). All data sets were generated by our laboratory (samples from chow-fed control and 24-h fasting conditions were deposited with GEO submission GSE14929 (28)). All data sets passed quality control tests using the SimpleAffy package (Bioconductor). Unbiased statistical methods were applied to generate lists of genes that (i) met statistical significance for differences between the standard chow-fed control group versus each of the two treatment groups and (ii) passed false discovery rate testing, eliminating lists of genes that generated false positives. Gene expression summarization values were generated using the Affymetrix MAS5.0 algorithm implemented in SimpleAffy for the R Statistical Computing Environment (also see R Project for Statistical Computing Web site) (29,30). After normalization, differentially expressed genes were calculated using permutation-adjusted p values for the step-down maxT multiple testing procedure, implemented in the Multtest Bioconductor package. The maxT procedure provides strong control of the family-wise Type I error rate (31)(32)(33). Significant genes were detected using a cut-off of maxT of Ͻ0.05. Overrepresentation analysis was used to functionally annotate differentially expressed lists of genes. The hypergeometric distribution was applied to detect functionally defined Kyoto Encyclopedia of Genes and Genomes pathway groups represented in gene lists more than expected by chance (34).
Reverse transcriptase-coupled real-time quantitative PCR (RT-qPCR) was performed, using primer sets, SYBR Green, and the 2 Ϫ⌬⌬Ct method of -fold change as described previously (28).
Neonatal Rat Cardiomyocyte Culture-Cardiomyocytes were isolated from CO 2 -asphyxiated 1-day-old Sprague-Dawley rats using an isolation kit from Worthington, following the manufacturer's instructions (36). Purified cardiomyocytes were plated onto gelatin-coated 12-well plates in high glucose (4.5 g/liter) Dulbecco's modified Eagle's medium supplemented with 10% horse serum, 5% fetal bovine serum, 100 M bromodeoxyuridine, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin and maintained in humidified tissue culture incubators at 37°C and 5% CO 2 . The morning after an overnight incubation, cells were washed in Hanks' balanced salt solution and fed low glucose (1 g/liter) Dulbecco's modified Eagle's medium supplemented with 100 g/ml transferrin, 10 ng/ml (ϭ1.7 nM) bovine insulin, 100 M bromodeoxyuridine, 2 mM glutamine, 0.5 mg/ml fatty acid-free BSA, 100 units/ml penicillin, and 100 g/ml streptomycin. Later that day, cells were refed with the latter maintenance medium, and stimulations were initiated using 0.2 mM oleic acid (18:1) (complexed to BSA in a 2:1 molar ratio) or fatty acid-free BSA (vehicle control, 0.1 mM). Stimulations with 100 nM dexamethasone or 100 nM Wy-14643 (Sigma) were normalized to ethanol (EtOH) vehicle control; final EtOH concentrations in the culture medium were 0.1%. For conditions that contained both 18:1 and dexamethasone, vehicle control-treated cells received both fatty acid-free BSA and EtOH. Cells were harvested for RNA isolation (Qiagen RNeasy) 24 h after stimulation.
RNA interference was performed using the HiPerfect transient transfection system (Qiagen) according to the manufacturer's instructions using 8 nmol of siRNA/well of a 12-well plate; siRNA used for PPAR␣ was Rn_Ppara_1 HP siRNA (SI01963451), which was compared with transfection of the "AllStars" negative control siRNA. Cells were harvested for RNA isolation 24 h after transfection. For all experiments using cultured cardiomyocytes, at least three independent experiments (in triplicate wells) were performed.
Working Heart-Hearts were dissected, prepared, cannulated, and perfused in the working mode as described previously (28). Briefly, mice received 100 units of heparin by intraperitoneal injection and 10 min later were anesthetized with an intraperitoneal injection of 400 mg/kg sodium pentobarbital. Hearts were excised and placed in ice-cold Krebs-Henseleit bicarbonate solution (118 mM NaCl, 25 mM NaHCO 3 , 4.7 mM KCl, 0.4 mM KH 2 PO 4 , 2.5 mM CaCl 2 , pH 7.4) supplemented with 5 mM glucose, 1.2 mM palmitic acid (bound to 3% fatty acid-free BSA), and 1 microunit/ml human insulin. Hearts were cannulated via the aorta and temporarily perfused, in a retrograde fashion, using the Langendorff mode, with continuous bubbling of a 95% O 2 , 5% CO 2 gas mixture into the buffer reservoir in order to maintain tissue oxygenation during subsequent left atrial cannulation. Following cannulation, the perfusion circuit was changed to the antegrade working mode at 37°C. Samples of the perfusate were collected every 10 min for measurement of (i) 14 (37). For perfusions coupled to NMR using 13 C-labeled substrates, see "NMR Substrate-Metabolite Profiling." Measurements of cardiac output, aortic flow, peak systolic pressure, and heart rate were acquired every 10 min for 10 s using inline flow probes (Transonic Systems, Inc.), the MP100 system from AcqKnowledge (BioPac Systems, Inc.), and a pressure transducer (TSD 104A, BIOPAC Systems, Inc.). Cardiac work was calculated as the product of peak systolic pressure and cardiac output. At the end of each perfusion, the dry weight of the ventricles was measured after desiccation in a vacuum oven.
NMR Substrate-Metabolite Profiling-The myocardium metabolizes ␤-hydroxy[2,4-13 C 2 ]butyrate and 2,4-13 C 2 -acetoacetate to [2-13 C]acetyl-CoA (i.e. 13 C label on the methyl carbon), which forms [4-13 C]citrate when combined with oxaloacetate and eventually ␣-keto-[4-13 C]glutarate. ␣-Keto-[4-13 C]glutarate is in equilibrium with [4-13 C]glutamate via transamination. Subsequent turns of the tricarboxylic acid cycle ultimately distribute 13 C atoms to glutamate, and the extents of accumulation of these carbons can be used as surrogates for rates of oxidative flux of a 13 C-labeled substrate, enriching 13 C metabolites in mouse heart over natural abundance 13 C (1.1%) (38 -40). We used 13 C-edited proton NMR (gHSQC) to measure the accumulation of 13 C in glutamate (individually on carbon positions 3 and 4) and normalized 13 Cassociated proton resonances to those of carbon 2 of the organic acid taurine, which does not become labeled by metabolism of these substrates but which is present at high myocardial concentrations that did not differ among conditions (supplemental Fig. S1A). Therefore, quantifying the resonances derived from 13 C-bound protons present within glutamate (whose total myocardial concentrations also did not differ among conditions; supplemental Fig. S1B), the parent substrate, and taurine provides signatures that reflect the capacity for a given substrate to be metabolized within a tissue (38,40,41).
For ex vivo perfusions, hearts were perfused for 15-20 min prior to freeze clamp (using liquid N 2 -chilled tongs) with 1 mM [2,4-13 C 2 ]␤OHB, 1 mM [2,4-13 C 2 ]AcAc, or 1.2 mM [U-13 C]palmitic acid that replaced the unlabeled substrate in the perfusion buffer (see full buffer composition under "Working Heart"). [2,4-13 C 2 ]AcAc was prepared from ethyl-[2,4-13 C 2 ]AcAc by hydrolysis using 3 M NaOH, pH 12, at room temperature for 2 h, followed by neutralization with HCl. Procession of this reaction was followed by proton NMR, which confirmed that Ͼ90% of the substrate had been hydrolyzed to [2,4-13 C 2 ]AcAc. Freeze-clamped biospecimens were freezedried in a vacuum freezer at Ϫ35°C for at least 30 h and homogenized in 3.6% perchloric acid, followed by removal of precipitated debris and neutralization of the supernatant with KOH. The resulting extracts were lyophilized to dryness and maintained at Ϫ12°C prior to study by NMR. Extracts were dissolved in 580 l of D 2 O (Cambridge Isotope Laboratories) spiked with 1 mM TSP (Sigma), and charged into a 5 mm NMR tube. The use of TSP provides both a chemical shift reference and an internal concentration standard permitting determination of metabolite concentrations in the extracts.
For measurements of in vivo substrate utilization, mice were injected with 10 mol/g body weight of D-[2,4-13 C 2 ]␤OHB (Cambridge Isotope Laboratories) intraperitoneally. Fifteen minutes later, mice were rapidly euthanized by CO 2 asphyxiation, the chest was opened within 30 s of introduction into the CO 2 chamber, and hearts were freeze-clamped.
Two NMR measurements were made at 25°C on each extract sample at a field strength of 11.75 teslas using an Inova-500 spectrometer (Varian). The first measurement was an 8-transient 1 H{ 13 C} (where proton doublets bound to 13 C are collapsed to singlets at a chemical shift similar to protons bound to 12 C) collection made under quantitative equilibrium conditions (90°pulse width of 9 s, sweep width ϭ 4395 Hz, selective inversion null on residual H 2 O, 15-s preacquisition delay, globally optimized alternating phase rectangular pulse carbon decoupling centered at 50 ppm, 25°C). Measurements of substrate and metabolite concentrations relative to TSP are established by signal integration. The second NMR measurement is a first increment gHSQC. This selects only for 1 H signals that are bound to 13 C, highlighting signals from molecules enriched in 13 C. Depending on available signal to noise, this acquisition is collected from 400 -900 transients under quantitative conditions (proton 90°pulse width ϭ 9 s, carbon 90°p ulse width of 14 s, spectral width ϭ 4295 Hz, 15-s preacquisition delay, globally optimized alternating phase rectangular pulse carbon decoupling centered at 50 ppm, 25°C). The time domain signal from the gHSQC is filtered with a Gaussian apodization function prior to Fourier transformation, resulting in a spectrum represented by 4,096 complex data points. Within myocardial extracts, the protons on the C-4 position, and one of two C-3 position protons, of glutamate are unobscured and therefore easily integrated and yielded internally validated results in our experiments. For ␤OHB, the protons on C-4 (the methyl carbon) and one on C-2 (the CH 2 ) are also unobscured. N-fold enrichment of glutamate can be estimated through comparing the values of glutamate to taurine obtained from the 1 H{ 13 C} experiment with that from the 13 C-edited experiment that detects only 13 C-1 H spin pairs.
Isolated Mitochondrial Extract HMGCS2 Assay-Myocardial mitochondria were isolated from CO 2 -euthanized mice by sucrose gradient centrifugation (42). Briefly, heart (biventricle) and liver specimens (100 -150 mg each) were excised, weighed, and rinsed in ice-cold mitochondrial isolation medium (MIM; 10 mM Na-HEPES, pH 7.2, 300 mM sucrose, 0.2 mM EDTA) and minced with fine scissors in a dry Petri dish (maintained on ice). Samples were resuspended in 2 volumes of ice-cold MIM, pH 7.4, plus 1 mg/ml BSA (MIM ϩ BSA) and homogenized on ice using a Dounce homogenizer. The pestle was washed with two additional volumes of ice-cold MIM ϩ BSA, which was pooled with the homogenate, and centrifuged at 600 ϫ g for 10 min at 4°C. The resulting supernatant, which contained mitochondria, was spun at 8,000 ϫ g for 15 min at 4°C, the supernatant was discarded, the mitochondrial pellet was resuspended in 5 ml of ice-cold MIM ϩ BSA, and the sample was centrifuged again at 8,000 ϫ g for 15 min at 4°C. The pellet was briefly washed in ice-cold MIM and resuspended in 1 volume (per mass of original tissue) of ice-cold MIM ϩ 4 mg/ml BSA (pH 7.4). Confirmation of mitochondrial integrity was confirmed by independent respiration assays (data not shown) (28).
Isolated mitochondrial HMGCS2 activity assays were performed by an adaptation of previously published protocols (43,44). Mitochondrial suspensions were sonicated using a Branson Sonifier 250 at a duty cycle of 30%, output control of 4, for 30 s on ice. Suspensions were centrifuged at 30,000 ϫ g for 30 min at 4°C, and supernatants were collected. 50 l of this enzymatic extract were added to 100 l of a reaction mixture that achieved final concentrations of 50 mM Tris, pH 8.0, 0.1 mM acetoacetyl-CoA, and either 2 or 10 mM acetyl-CoA (Sigma) in a total of 150 l. Reactions proceeded at room temperature for 5, 10, 15, and 20 min, after which 27.5-l aliquots of the reaction were removed and added to a tube containing 21 l of 3% perchloric acid to terminate the reaction. Terminated reaction tubes were vortexed, and 6.5 l of 2 M KOH was rapidly added for neutralization. Time 0 was obtained by acidification of an aliquot of enzymatic extract prior to the addition of substrate. Finally, precipitated debris was briefly centrifuged, and 10 l of each of supernatant was used for total ketone body (AcAc ϩ ␤OHB) quantification using a colorimetric assay kit (Wako). Rates of ketogenesis within extracts of mouse liver mitochondria were linear for the first 10 min of reaction time.
Statistical Analyses-Analyses were performed using GraphPad software (Prism), using tests as described throughout. Analyses of microarray data sets and NMR data are described above.

Effects of Ketogenic Nutrient Milieu on Circulating Metabolites and Myocardial
Transcriptome-Nutrient deprivation and a very low carbohydrate diet each induce shifts from car-bohydrate to fat utilization (45,46). Although the effects of nutrient deprivation on the myocardial transcriptome have been described (28,47), the functional, transcriptional, and metabolic consequences of a very low carbohydrate ketogenic diet on myocardium are not well characterized. Therefore, we maintained 8 -12-week-old wild-type C57Bl/6 male mice on either (i) standard polysaccharide-rich, low fat (13% of calories) chow (control mice), (ii) standard chow, followed by a 24-h fast, or (iii) a very low carbohydrate (0.5% of calories), high fat (95% of calories) ketogenic diet for 5 weeks (n ϭ 5/group). We first determined the effects of these three nutrient states on body weight, serum metabolites, and serum insulin. Body weight of mice maintained on standard chow over a 5-week interval increased from 19.0 Ϯ 0.4 to 22.3 Ϯ 0.7 g; weight of mice fasted for 24 h decreased from 21.7 Ϯ 0.4 to 19.0 Ϯ 0.4 g; and mice maintained on the ketogenic diet for 5 weeks fell from 20.7 Ϯ 0.7g to 19.3 Ϯ 1.2 g after 5 weeks of the ketogenic diet, consistent with the mild weight loss previously reported for ketogenic diet-fed mice (46,48). Serum metabolites reflected expected changes; after a 24-h fast, serum ketone concentrations increased markedly, with more modest changes in serum glucose, free fatty acids, triglycerides, and insulin (Table 1). Compared with control mice maintained on standard chow, those maintained on a ketogenic diet exhibited increased serum ketone body concentrations, higher free fatty acids and triglycerides, and significantly reduced insulin. Therefore, these conditions provided provocative states to study myocardial sensing and response to its nutrient environment.
To explore the effects of ketogenic nutrient environments on the myocardial transcriptome, we generated GeneChip microarray data sets using targets that were created from myocardial RNAs prepared from standard chow-fed mice, 24-h fasted mice, or ketogenic diet-fed mice (n ϭ 4 -5/group). We generated unbiased lists of genes significantly up-or downregulated compared with normal chow and found that a subset of 28 genes was shared between the 24-h fasting and ketogenic diet-fed states (see supplemental Tables 1 and 2 for complete lists of discovered genes). Among these overlapping genes was Hmgcs2 (up-regulated both by fasting and ketogenic diet) and Oxct1 (down-regulated by fasting and ketogenic diet; Fig. 1A).
Pathways analysis using the Kyoto Encyclopedia of Genes and Genomes database confirmed that the process of ketone body metabolism was significant in both adult 24-h fasting and ketogenic diet-fed hearts compared with chow-fed control hearts (see supplemental Tables 3 and 4). These microarray findings were confirmed using RT-qPCR (Fig. 1, B and C), pro- tein immunoblotting (Fig. 1D), and immunohistochemistry ( Fig. 1E) of biological replicate samples of myocardial ventricular tissue.

Mechanisms of Myocardial Hmgcs2 and Oxct1
Regulation-PPAR␣, a fatty acid ligand-activated nuclear receptor transcription factor required for utilization of fatty acids during fasting in liver and heart, activates the HMGCS2 promoter in liver (49 -53). Little is known of the regulation of Oxct1 (15,18). To determine if PPAR␣ is a key regulator of Hmgcs2 and Oxct1 expression in the myocardium, we used RT-qPCR to measure their expression in chow-fed, 24-h fasting, and ketogenic dietfed PPAR␣ Ϫ/Ϫ mice. For this experiment, ketogenic diet-fed mice only received the diet for 2 weeks, because PPAR␣ Ϫ/Ϫ mice (compared with wild-type mice) lost weight at a greater rate when maintained on the ketogenic diet, a phenotype probably due to the hepatic defect in fatty acid oxidation and lipid metabolism in PPAR␣ Ϫ/Ϫ mice (50,51). The ability of a 24-h fast or a ketogenic diet to induce myocardial Hmgcs2 expression was abrogated in mice lacking systemic PPAR␣ ( Fig. 2A). The suppression of Oxct1 expression was maintained in PPAR␣ Ϫ/Ϫ mice fed a ketogenic diet, although it was not in the 24-h fasting state.
To further dissect the roles of fatty acid signals and PPAR␣ in the regulation of myocardial Hmgcs2 and Oxct1, we measured their dynamic expression in primary cultures of neonatal rat cardiomyocytes. The free fatty acid 18:1 induced Hmgcs2 (Fig.  2B) and suppressed Oxct1 expression (Fig. 2C). Hmgcs2 was also induced by the PPAR␣ isoform-specific synthetic ligand Wy-14643 (100 nM), but 100 nM Wy-14643 had no effect on Oxct1 expression (Fig. 2, B and C). In addition to PPAR␣, the glucocorticoid receptor regulates hepatic Hmgcs2 expression (54). Dexamethasone also induced Hmgcs2 in cardiomyocytes, an effect that was synergistic with the addition of 18:1, but dexamethasone had no effect on Oxct1 expression (Fig. 2, B and C).
To further confirm the requirement of cardiomyocyte PPAR␣ for the fatty acid-mediated induction of myocardial Hmgcs2, we transfected siRNAs against either PPAR␣ mRNA or a negative control sequence into vehicle control-or 18:1treated cardiomyocytes. Maximal reduction of PPAR␣ expression was ϳ60% (Fig. 2B), but this was sufficient for a nearly 60% attenuation of 18:1-mediated induction of Hmgcs2. However, PPAR␣ siRNA transfection had no effect on the ability of 18:1 to suppress Oxct1 expression (Fig. 2C), indicating that full expression of PPAR␣ is not required for 18:1-mediated suppression of Oxct1 expression.
In the liver, insulin suppresses Hmgcs2 expression (55). To determine if the effect of insulin is preserved in cardiomyocytes, we incubated 18:1 (versus vehicle-treated) cultured cardiomyocytes with increasing concentrations of insulin and found that insulin blunts the ability of 18:1 to induce Hmgcs2 expression (Fig. 2D). Moreover, increasing concentrations of insulin blunted the ability of 18:1 to suppress Oxct1 expression.
Attenuated Myocardial Ketolytic Capacity in Ketogenic Dietfed Mice-To determine if these modifications of the key mediators of ketone body metabolism are associated with measurable metabolic changes, we profiled substrate fate using perfused hearts ex vivo and in vivo, by acquiring myocardial extracts and resolving 13 C-edited proton NMR spectra. Glutamate exhibits high steady-state concentrations in the myocardium (supplemental Fig. S1B), and because it emanates from the tricarboxylic acid cycle intermediate ␣-ketoglutarate, it is an ideal molecule to exploit as a reporter of the rate of carbon flow through the tricarboxylic acid cycle when 13 C-labeled substrates are administered to a metabolizing system (38,40,41).
To determine if suppression of myocardial ketolysis in ketogenic diet-fed mice is preserved in vivo, we administered D-[2,4-13 C 2 ]␤OHB (bolus of 10 mol/g body weight intraperitoneally) to wild-type chow-fed and ketogenic diet-fed mice. Fifteen minutes after administration of the substrate, myocardial extracts were collected from freeze-clamped specimens. Consistent with our ex vivo observations, 13 C enrichment of glutamate was 27.3 Ϯ 2.0% in hearts of chow-fed mice versus 20.5 Ϯ 1.8% in hearts of ketogenic diet-fed mice (n ϭ 6 -7/group, p Ͻ 0.05; Fig. 3D). In hearts of substrate-treated animals that had been fasting for 24 h, glutamate enrichment was not significantly changed from that of chow-fed animals (data not shown). Fifteen minutes postinjection of D-[2,4-13 C 2 ]␤OHB, total myocardial ␤OHB concentrations were 84% higher in the hearts of ketogenic diet-fed mice (11.6 Ϯ 1.6 nmol/g dry tissue) than in chow-fed mice (6.3 Ϯ 1.0 nmol/g dry tissue; n ϭ 7/group, p Ͻ 0.05; Fig. 3E). Serum ␤OHB concentrations were also elevated 15 min postbolus in these ketogenic diet-fed mice (14.5 Ϯ 1.5 mM) compared with chowfed mice (9.4 Ϯ 1.4 mM, n ϭ 6 -7/group, p Ͻ 0.05). In addition, a proton resonance corresponding to 13 C-enriched acetone was observed in myocardial extracts prepared from five of six keto- genic diet-fed mice injected with [2,4-13 C 2 ]␤OHB but in none of those from chow-fed or 24-h fasting mice (data not shown). Given that AcAc is the only known substrate for acetone that exists in mammalian metabolism (via non-enzymatic decarboxylation), acetone serves as a reporter of AcAc, generated from unoxidized [ 13 C]AcAc during extract preparation, present only in extracts derived from ketogenic diet-fed myocardial samples (p Ͻ 0.01, Mann-Whitney U test).
A Chronic Ketogenic State Does Not Cause Net de Novo Myocardial Ketogenesis-The presence of HMGCS2 protein in the hearts of ketogenic diet-fed mice raises the possibility that these hearts are able to perform de novo ketogenesis. To test this possibility, we performed two separate experiments. First, hearts from chow-fed, and ketogenic diet-fed mice were perfused for 20 min with a buffer that contained 1 microunit/ml insulin, 1.2 mM [U-13 C]palmitate, and 5 mM glucose. Through ␤-oxidation, [U-13 C]palmitate generates [U-13 C]acetoacetyl-CoA and [1,2-13 C 2 ]acetyl-CoA, providing labeled substrates for potential HMGCS2 activity. To limit the detection of 13 C-labeled ketones that emanate from isotopic exchange via pseudoketogenesis, unlabeled ketones were not included in the perfusion buffer. NMR profiling revealed robust 13 C enrichment of glutamate (supplemental Fig. S2A; 53 and 64% in hearts from chow-fed and ketogenic diet-fed mice, respectively (n ϭ 5/group; p ϭ 0.45)), confirming the procession of 13 C-labeled substrate through FAO. However, hearts from both chow-fed and ketogenic diet-fed mice did not yield [ 13 C]acetone, [ 13 C]AcAc, or [ 13 C]␤OHB (supplemental Fig. S2A).
Second, we isolated mitochondria from hearts of chow-fed and ketogenic diet-fed mice to perform in vitro ketogenesis assays. Using mitochondrial extracts from ketogenic diet-fed mouse liver as a positive control, minute quantities of ketone bodies were measured in extracts of hearts from ketogenic dietfed mice, but no evidence of net de novo ketogenesis was observed (supplemental Fig. S2B).
A Chronic Ketogenic State Prevents Inhibition of Fatty Acid Oxidation by Ketone Bodies-Ketone bodies compete with fatty acids for contribution to the myocardial tricarboxylic acid cycle (22, 24 -26). To test the ability of supplemented ketone bodies to inhibit fatty acid or glucose oxidative metabolism, we perfused hearts ex vivo first in the complete absence of ketone bodies, followed by the presence of unlabeled ketone bodies, each for 30 min. Using 1 microunit/ml insulin, 1.2 mM palmitate, and 5 mM glucose in a buffer supplied with radioactive tracers that report rates of fatty acid and glucose oxidation, we observed that FAO rates were reduced 63 and 43% by the addition of 2 mM ␤OHB to the hearts of chow-fed and 24-h fasted mice, respectively (Fig. 4A). However, this effect of ␤OHB was abrogated in hearts of ketogenic diet-fed mice, indicating a loss of metabolic flexibility in this nutrient state. We observed no effects of 2 mM ␤OHB on myocardial glucose metabolism, although, as expected, fasted hearts exhibited slightly reduced glucose oxidation (56) (Fig. 4B). Increased myocardial power (J/s) was observed in hearts from both fasted and ketogenic diet-fed animals (Fig. 4C). Collectively, these findings indicate that ketogenic nutrient environments, particularly those that are chronic, initiate a myocardial response that reduces ketone body utilization and preserves inflexibly high rates of fatty acid oxidation.

DISCUSSION
As an omnivore, the heart adapts to its nutrient environment by maximizing uptake, transport, and utilization of substrates to which it has access. For ketone body utilization, this paradigm has been supported by ex vivo perfusion experiments, which have demonstrated that the rate of myocardial utilization of  ketone bodies proceeds with first order kinetics until a saturating utilization rate is achieved (21,23). Our results indicate that hearts chronically exposed to a ketogenic milieu engage a transcriptional response that offsets the delivery of ketone bodies by decreasing ketolytic capacity, which in turn correlates with resistance to the inhibitory effects that ketone bodies typically exert over fatty acid oxidation.
In ketotic states, two notable transcriptional changes were observed in mouse hearts: suppression of Oxct1 and induction of Hmgcs2. The former yields lower abundance of the ketolytic enzyme SCOT, required for activation of AcAc to AcAc-CoA. The latter change leads to protein expression of HMGCS2, typically excluded from the myocardium but whose role in heart remains to be determined. In the chronic ketotic state, mitochondria to which HMGCS2 protein cargo has been addressed could be programmed to facilitate a futile cycle, in which a proportion of AcAc that is activated to acetoacetyl-CoA by SCOT is restored to AcAc through HMGCS2. Using two methods, however, we did not observe evidence of HMGCS2 activity in the hearts of ketogenic dietfed mice. Several potential explanations emerge. First, myocardial ketogenesis through the HMGCS2 pathway may fall below the limit of detection for each of these approaches. 13 C-Edited proton NMR detection of 13 C-enriched metabolites is quite sensitive but is unlikely to detect metabolites present in nanomolar concentrations. A second prospective explanation is that hydroxymethylglutaryl-CoA lyase is not as abundant in heart as it is in liver, although our microarray data sets indicate that it is expressed in hearts of chow-fed and ketogenic diet-fed mice. Third, thiolase activity may compete against HMGCS2 for AcAc-CoA. Finally, low quantities of myocardial HMGCS2-derived AcAc may ultimately be oxidized, even in ketogenic myocardium. Experimentally, myocardium can be rendered a net producer of ketones, when acetyl-CoA derived from the carnitine palmitoyl transferase-independent substrate octanoate overwhelms the tricarboxylic acid cycle, and acetyl-CoA is diverted to the pseudoketogenic pathway (24), but future experiments, using genetic models, will be needed to definitively ascertain whether myocardial HMGCS2 plays a role in myocardial ketone metabolism.
Reduction of intrinsic myocardial ketolytic capacity has been observed in pathological states. In rodent models of type 1 diabetes using streptozotocin and in an animal model of lipopolysaccharide-induced sepsis, non-enzymatic tyrosine nitration is associated with decreased SCOT activity and, consequently, reduced ketolysis (57)(58)(59). These models demonstrate non-enzymatic consequences of oxidative stress. The data presented in this study provide evidence for nutritionally triggered transcriptional events that attenuate ketone body oxidation. This response could serve a key protective energetic role in myocardial metabolic homeostasis. Previous work from the Taegtmeyer group (60 -63) demonstrated that hearts perfused exclusively with AcAc developed impaired tricarboxylic acid flux, due to reduction of free CoA-SH pools. This defect was rescued by supplementing CoA synthetic precursors or by the provision of anaplerotic substrates. Thus, the adaptive ability to attenuate ketone body oxidation in a ketotic but mixed nutrient environment may serve as a mechanism that helps preserve myocardial tricarboxylic acid flux. The observation that a systemic high dose ketone body bolus promotes relatively higher serum ketone concentrations in ketogenic diet-fed mice suggests the possibility that other ketolytic tissues may enlist similar adaptive mechanisms.
Mechanisms that govern transcriptional regulation of myocardial Hmgcs2 and Oxct1 expression remain to be elucidated. Hmgcs2 induction is dependent on PPAR␣, inhibited by insulin, and enhanced by glucocorticoids. Little is known of the regulation of Oxct1 (15,18). Additional studies will be required to determine whether Oxct1 is regulated through PPAR isoformdependent mechanisms. A transgenic mouse model that overexpresses Glut1 in the myocardium results in increased glucose uptake and utilization. This is accompanied by reductions in Oxct1 expression and, consistent with our findings, decreased myocardial ketone body utilization ex vivo (64). Cumulatively, these results suggest that high delivery/uptake of either fatty acids or glucose each suppress Oxct1 and, consequently, myocardial ketolysis.
The nature of the diets we employed confers an inherent limitation to our study. The unpurified chow diet harbors macro-and micronutritional components that could contribute to the transcriptional and metabolic changes we observed. Future studies, using semipurified control diets, will be required to comprehensively measure nutrient-responsive characteristics of the transcriptome and metabolome.
The use of NMR to profile the procession of 13 C-labeled substrates through oxidative metabolism confers several advantages over radioactive tracer-based methods. Methods that rely on capture of 14 CO 2 or 3 H 2 O can be limited by condition-dependent extents of label dilution with endogenous and exogenous substrate pools and by the inability to assess label exchange with other intermediate pools. As a result, the quantitative rates of substrate oxidation can vary among analytical approaches. The 50 -60% 13 C enrichment of glutamate observed when [U-13 C]palmitate was used as a substrate in our perfusion studies suggests that the radioactive tracer-based method could overrepresent the absolute magnitude of FAO. Nonetheless, the relative reduction of fatty acid oxidative flux that we observed in hearts of chow-fed animals exposed to ␤OHB replicates previous observations. Furthermore, a novel finding, that ␤OHB is unable to suppress FAO in hearts from ketogenic diet-fed mice, compares relative FAO rates within the same hearts. Therefore, the complementary use of both radioactive tracer and NMR-based methods in this study demonstrates the metabolic adaptation of the heart to a ketotic environment.
Insulin resistance and diabetes are states in which systemic ketone body metabolism is perturbed (17,(65)(66)(67). Because altered substrate metabolism is one of the hallmark abnormalities of insulin-resistant and diabetic myocardium (5), regulation of myocardial ketone body metabolism may prove to be an important diagnostic biomarker of potential naturally occurring variations of ketolytic capacity. These studies indicate that autonomous mechanisms in the heart modulate myocardial ketolytic capacity, a potentially clinically relevant metabolic adaptation.