Dysfunctional cardiac mitochondrial bioenergetic, lipidomic, and signaling in a murine model of Barth syndrome.

Barth syndrome is a complex metabolic disorder caused by mutations in the mitochondrial transacylase tafazzin. Recently, an inducible tafazzin shRNA knockdown mouse model was generated to deconvolute the complex bioenergetic phenotype of this disease. To investigate the underlying cause of hemodynamic dysfunction in Barth syndrome, we interrogated the cardiac structural and signaling lipidome of this mouse model as well as its myocardial bioenergetic phenotype. A decrease in the distribution of cardiolipin molecular species and robust increases in monolysocardiolipin and dilysocardiolipin were demonstrated. Additionally, the contents of choline and ethanolamine glycerophospholipid molecular species containing precursors for lipid signaling at the sn-2 position were altered. Lipidomic analyses revealed specific dysregulation of HETEs and prostanoids, as well as oxidized linoleic and docosahexaenoic metabolites. Bioenergetic interrogation uncovered differential substrate utilization as well as decreases in Complex III and V activities. Transgenic expression of cardiolipin synthase or iPLA2γ ablation in tafazzin-deficient mice did not rescue the observed phenotype. These results underscore the complex nature of alterations in cardiolipin metabolism mediated by tafazzin loss of function. Collectively, we identified specific lipidomic, bioenergetic, and signaling alterations in a murine model that parallel those of Barth syndrome thereby providing novel insights into the pathophysiology of this debilitating disease.

penetrating mechanistic insights into the role of tafazzin in regulating mitochondrial lipidomics, signaling, and bioenergetic function have been defi ned, thereby identifying the complexity of alterations resulting from tafazzin loss of function and the multiple pathologies manifest in Barth syndrome patients.

Materials
Synthetic phospholipids used as internal standards in mass spectrometric analyses were purchased from Avanti Polar Lipids (Alabaster, AL). Solvents for sample preparation and mass spectrometric analysis were purchased from Burdick and Jackson (Muskegon, MI) as well as Sigma Aldrich (St. Louis, MO).

Induction of the doxycycline inducible Taz KD mouse model of Barth syndrome
Developmental doxycycline induction of the Taz KD mouse model was performed in utero and maintained postnatally as previously described in detail ( 24 ). A syngeneic transgenic colony was generated by breeding several generations onto a C57BL/6J mouse background, which were used for all studies. Briefl y, dams were fed a 625 mg/kg doxycycline diet (Harlan Teklan) for 5 days prior to mating. Upon initiation of mating with a shRNA tafazzin-positive heterozygote male, the diet was removed for 3 days until a confi rmation of insemination was obtained at which time the male was removed from the cage and the doxycycline diet was returned to the breeding cage and maintained until the pups were weaned. The genotype of the mice was confi rmed by PCR as previously described ( 24 ) and male wild-type littermates and Taz KD mice were maintained on the doxycyline diet until two months of age at which time lipidomics and biochemical experiments were performed for developmental characterization. Additional experiments were performed utilizing double genetic crossed mice [Taz KD×cardiolipin synthase transgenic (CLS-TG) and Taz KD×iPLA 2 ␥ knockout (KO)]. In these experiments, male mice were raised until 2 months of age without doxycycline induction and at 2 months of age the mice were induced with doxycyline until 4 months of age at which time the mice were sacrifi ced and experiments were performed. All wild-type mice were also maintained on a 625 mg/kg doxycycline diet as control. All animal procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the Animal Studies Committee at Washington University School of Medicine.

Multidimensional mass spectrometry-based shotgun lipidomic analysis of the cardiac lipidome
Briefl y, myocardial tissue was removed, washed with 10× diluted PBS and freeze clamped in liquid nitrogen for lipidomic analysis. Lipidomic analyses were performed as previously described, using a modifi ed Bligh and Dyer extraction protocol (26)(27)(28). Individual lipid extracts were reconstituted with 1:1 (v/v) CHCl 3 /CH 3 OH, fl ushed with nitrogen, and stored at Ϫ 20°C prior to electrospray ionization-MS using a TSQ Quantum Ultra Plus triple-quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with an automated nanospray apparatus (Advion Biosciences Ltd., Ithaca, NY) and customized sequence subroutine operated under Xcalibur software. Enhanced MDMS-SL analysis of cardiolipins was performed with a mass resolution setting of 0.3 Thomson as described previously in detail ( 29 ). novel therapeutic approaches for treatment of this lethal disease.
Regulation and maintenance of the mitochondrial lipidome is critical for bioenergetic effi ciency, cellular signaling, and multiple other mitochondrial processes (e.g., fusion and fi ssion) (10)(11)(12)(13). Mitochondria are comprised of a unique double bilayer membrane structure that facilitates the compartmentalization of multiple processes to effi ciently integrate mitochondrial function with cellular energy needs ( 11,14,15 ). A prominent lipid regulator of mitochondrial inner membrane surface charge, molecular dynamics, and membrane curvature is cardiolipin, which contains a unique tetra-acyl structure ( 12,15,16 ). Cardiolipin is a doubly charged mitochondrial phospholipid comprised of two phosphates, three glycerol groups, and four acyl chains (17)(18)(19). Regulation of the content and molecular species composition of cardiolipin is critical for electron transport chain efficiency, adenine nucleotide translocase activities, mitochondrial protein import, and uncoupling, as well as TCA cycle fl ux ( 20,21 ). The molecular species composition of cardiolipin is dynamically regulated by integrated cellular control of cardiolipin de novo synthesis, phospholipasemediated deacylation, and membrane remodeling by the subsequent actions of either transacylase or acyltransferase activities that are coordinately regulated to lead to a mature cardiolipin molecular species distribution ( 22 ). Additionally, because the remodeling of cardiolipin through transacylation harvests acyl chains from choline and ethanolamine glycerophospholipids, the dynamic balance of cardiolipin remodeling by transacylation versus acyltransferase activity is critical for the maintenance of mitochondrial membrane architecture, surface charge, and molecular dynamics. Thus, the precisely regulated balance of cardiolipin synthesis, remodeling, and degradation exerts tight regulatory control of mitochondrial membrane structure and function.
Herein, we examined the bioenergetic, lipidomic, and signaling mechanisms that were altered in a tafazzin lossof-function mouse model (23)(24)(25) that was predicted to recapitulate the pathology of Barth syndrome in an animal model thereby facilitating a greater understanding of the multiple processes contributing to hemodynamic dysfunction in Barth syndrome. Through utilization of integrated molecular, chemical, and lipidomic approaches in conjunction with high-resolution respirometry, multiple novel mechanistic roles of tafazzin in regulating cardiolipin and lysocardiolipin homeostasis and myocardial signaling have been identifi ed and their resultant effects on mitochondrial electron transport chain function, bioenergetics, and cardiac transcriptomic networks delineated. Additionally, we employed double crosses of genetic models of key enzymes involved in the cardiolipin remodeling process, namely cardiac myocyte-specifi c transgenic expression of cardiolipin synthase as well as the ablation of iPLA 2 ␥ in the Taz KD mouse model of Barth syndrome to investigate potential therapeutic strategies to attenuate maladaptive cardiolipin remodeling. Thus, through the utilization of complementary transgenic approaches, BSA, pH 7.4) and homogenized using 12-15 passes with a Tefl on homogenizer using a rotation speed of 120 rpm. Next, the homogenate was centrifuged for 5 min at 850 g , and the supernatant was collected and centrifuged at 7,200 g for 10 min. The pellet was collected and resuspended in MIB without BSA. Mitochondrial protein content was determined using a BCA protein assay (Thermo Fisher Scientifi c, San Jose, CA). High-resolution respirometry was performed using 50 µg of mitochondrial protein per 2 ml chamber with the substrate and inhibitor addition protocol previously described ( 27,31 ).

Enzymatic characterization of electron transport chain and functional adenine nucleotide translocase activities
Complex I. Complex I (NADH-ubiquinone oxidoreductase) activity was determined by measuring the decrease in the concentration of NADH at 340 nm and 37°C as previously described ( 32,33 ). The assay was performed in buffer containing 50 mM potassium phosphate (pH 7.4), 2 mM KCN, 5 mM MgCl 2 , 2.5 mg/ml BSA, 2 M antimycin, 100 M decylubiquinone, and 0.3 mM K 2 NADH. The reaction was initiated by adding purifi ed mitochondria (5 g). Enzyme activity was measured for 5 min and values were recorded 30 s after the initiation of the reaction. Specifi c activities were determined by calculating the slope of the reaction in the linear range in the presence or absence of 1 M rotenone (Complex I inhibitor).
Complex II. Complex II (succinate decylubiquinone 2,6dichloroindophenol (DCIP) oxidoreductase) activity was determined by measuring the reduction of DCIP at 600 nm as previously described ( 33,34 ). The Complex II assay was performed in buffer containing 25 mM potassium phosphate (pH 7.4), 20 mM succinate, 2 mM KCN, 50 M DCIP, 2 g/ml rotenone, and 2 g/ml antimycin. Purifi ed mitochondria (5 g) were added prior to initiation of the reaction. The reaction was initiated by adding 56 M decylubiquinone. Specifi c activities were determined by calculating the slope of the reaction in the linear range in the presence or absence of 0.5 mM thenoyltrifl uoroacetone (Complex II inhibitor).
Complex III. Complex III (ubiquinol-cytochrome c reductase) activity was determined by measuring the reduction of cytochrome c at 550 nm and 30°C. The Complex III assay was performed in buffer containing [25 mM potassium phosphate (pH 7.4), 1 mM EDTA, 1 mM KCN, 0.6 mM dodecyl maltoside, and 32 M oxidized cytochome c] using purifi ed mitochondria (1 g). The reaction was initiated by adding 35 M decylubiquinol. The reaction was measured following the linear slope for 1 min in the presence or absence of 2 M antimycin (Complex III inhibitor). Decylubiquinol was made by dissolving decylubiquinone (10 mg) in 2 ml acidifi ed ethanol (pH 2) and using sodium dithionite as a reducing agent. Decylubiquinol was further purifi ed with cyclohexane ( 32,33,35 ).
Complex IV. Complex IV (cytochrome c oxidase) activity was determined by measuring the oxidation of ferrocytochrome c at 550 nm and 25°C. The Complex IV assay was performed in buffer containing [10 mM Tris-HCl and 120 mM KCl (pH 7.0)] using purifi ed mitochondria (2.5 g). The reaction was initiated by adding 11 M reduced ferrocytochrome c and monitoring the slope for 30 s in the presence or absence of 2.2 mM KCN (Complex IV inhibitor) ( 33,36 ).
Complex V. Complex V (F1 ATPase) activity was determined using a coupled reaction measuring the decrease in

Oxidized lipid metabolite analysis
Tissues ( ‫ف‬ 100 mg) were quickly washed with cold PBS (pH 7.4) solution, blotted, snap-frozen in liquid nitrogen, and stored at Ϫ 80°C until extraction. For extraction, 2 ml of icecold methanol/CHCl 3 (1:1 v/v with 1% HAc) and 2 l of antioxidant mixture (0.2 mg/ml BHT, 0.2 mg/ml EDTA, 2 mg/ml triphenylphosphine, and 2 mg/ml indomethacin in a solution of 2:1:1 methanol/ethanol/water) were added to the tissue samples. Internal standards (250 pg each of TXB2-d4, PGE2-d4, LTB4-d4, 12-HETE-d8, 13-HODE-d4, and 9,10-DiHOME-d4 in 5 ul acetonitrile) were also added at this step . The samples were immediately homogenized and subsequently vortexed several times during a 15 min incubation on ice. Next, 1 ml of ice-cold water was added to the sample which was briefl y vortexed and centrifuged at 1,500 g for 15 min. The CHCl 3 layer was transferred to a new tube and the remaining methanol/water layer was reextracted with 1 ml of CHCl 3 and centrifuged at 1,500 g for 15 min. The combined CHCl 3 layers were dried down with N 2 and reconstituted in 1 ml of 10% methanol solution.
The reconstituted solution was immediately applied to a Strata-X solid phase extraction cartridge that had been preconditioned with 1 ml of methanol followed by 1 ml of 10% methanol. The cartridge was then washed with 2× 1 ml of 5% methanol and additional solvent was fl ushed out with N 2 at a pressure of 5 psi. Eicosanoids were eluted with 1 ml of methanol containing 0.1% HAc. All cartridge steps were carried out using a vacuum manifold attached to a house vacuum line. After the organic solvent was evaporated with a SpeedVac, the residues were derivatized with N -(4-aminomethylphenyl)pyridinium (AMPP).
The derivatization with AMPP was performed as previously described in detail ( 30 ). Briefl y, 12.5 l of ice-cold acetonitrile/ N , Ndimethylformamide (4:1, v:v) was added to the residue in the sample vial. Then 12.5 l of ice-cold 640 mM [3-(dimethylamino) propyl]ethyl carbodiimide hydrochloride in HPLC grade water was added. The vial was briefl y vortexed and 25 l of 5 mM N -hydroxybenzotriazole/15 mM AMPP in acetonitrile was added. The vials were vortexed briefl y and placed in a 60°C water bath for 30 min.
Metabolites were analyzed using a hybrid tandem mass spectrometer (LTQ-Orbitrap, Thermo Scientifi c) via selected reaction monitoring in positive ion mode with sheath, auxiliary, and sweep gas fl ows of 30, 5, and 1, respectively. The capillary temperature was set to 275°C and the electrospray voltage was 4.1 kV. Capillary voltage and tube lens were set to 2 and 100 V, respectively. Instrument control and data acquisition were performed using the Thermo Xcalibur V2.1 software.

Mitochondrial high-resolution respirometry
Mice used for experiments were sacrifi ced and the hearts were immediately removed and dissected on ice (4°C ambient temperature). Briefl y, the dissected heart was placed in mitochondrial isolation buffer (

Tafazzin defi ciency results in altered choline and ethanolamine glycerophospholipid molecular species
Cardiolipin molecular species remodeling involves the coordinated regulation of various phospholipase, acyltransferase, and transacylase activities. Mutants in tafazzin have previously been associated with defective transacylation of specifi c acyl chains from choline and ethanolamine glycerophospholipid molecular species ( 3,8 ). In the present study, loss of tafazzin enzymatic activity in the Barth syndrome mouse model results in the accumulation of specifi c choline diacyl (D) glycerophospholipid molecular species containing linoleic acid in the sn -2 position, specifi cally D16:0-18:2 and D18:0-18:2 ( Fig. 2A ). Due to the increased linoleic acid content in choline glycerophospholipid molecular species, the utilization of linoleic acid to synthesize arachidonic acid by acyl chain elongation is also altered as evidenced by an increased content of 20:3-(an intermediate in the synthesis of 20:4 from 18:2) and 20:4-containing molecular species (e.g., D18:0-20:3 and D18:0-20:4). Furthermore, molecular species containing docosahexaenoic acid at their sn -2 positions are decreased including D16:0-22:6 and D18:2-22:6, thus demonstrating an imbalance in the architectural restructuring of choline glycerophospholipid molecular species. The increased presence of -6 polyunsaturated fatty acids (i.e., linoleic acid and arachidonic acid) may partially account for the defi ciency of docosahexaenoic acid because the biosynthesis of both -6 and -3 polyunsaturated fatty acids compete for the same enzyme systems. Interestingly, analysis of ethanolamine glycerophospholipid molecular species also displayed an overall increase in molecular species containing NADH concentration at 340 nm and 37°C as previously described (37)(38)(39). The Complex V assay was performed in buffer containing (50 mM Tris-HCl, 25 mM KCl, 5 mM MgCl 2 , 4 mM Mg-ATP, 200 µM K 2 NADH, 1.5 mM phosphoenolpyruvate, 5 units pyruvate kinase, 5 units lactate dehydrogenase, 2.5 M rotenone, and 2 mM KCN) using purifi ed mitochondria (10 g). The reaction was initiated by the addition of mitochondria and the reaction was monitored for 6 min. The slope in the linear range was used to calculate the reaction rate. Oligomycin (2.5 mg/ml) (Complex V inhibitor) was added to designated cuvettes to calculate the specifi c Complex V activity.

Microarray analysis of the cardiac transcriptome
RNA was extracted from 2-month-old male WT and Taz KD mice using Trizol and the RNeasy extraction kit (Qiagen). RNA integrity was calculated and transcriptome analysis was performed using an Illumina BeadArray. Quantile analysis was utilized for postprocessing expression analysis.

Statistical analysis
Data were analyzed using a two-tailed unpaired Student's t -test. Differences were regarded as signifi cant at the * P < 0.05, **/ # P < 0.01. All data are reported as the means ± SEM unless otherwise indicated.

Identifi cation of the cardiac cardiolipin phenotype of the developmental inducible Taz KD mouse model of Barth syndrome
We used a multidisciplinary approach to investigate the biochemical and biophysical mechanisms leading to mitochondrial dysfunction resulting from tafazzin loss of function in mice. Mass spectrometric analysis of myocardial cardiolipin molecular species was performed by MDMS-SL analysis using the M + 1/2 isotopologue approach we previously developed ( 29 ). The results revealed dramatic alterations in cardiolipin content and molecular species distribution induced by tafazzin loss of function ( Fig. 1A ). Quantitative analysis of cardiolipin molecular species revealed a dramatic decrease in linoleic (18:2)-enriched molecular species, most notably tetra-18:2 (18:2-18:2-18:2-18:2), which is the major molecular species of cardiolipin in myocardium ( Fig. 1B ). Importantly, the decrease in tetra-18:2 molecular species is a hallmark characteristic of Barth syndrome ( 46 ). Selective cardiolipin molecular species containing dihomo-␥ -linolenic acid (20:3) or docosahexaenoic acid (22:6) Molecular species below 0.2 nmol/mg protein were omitted from the fi gure for visual clarity. C: Analysis of lysocardiolipins revealed an increase in both dilysocardiolipin and monolysocardiolipin molecular species. Molecular species below 0.1 nmol/mg protein were omitted from the fi gure for visual clarity. Values represent the mean quantitative value of molecular species ± S.E. (N = 3 hearts per group; black bars, wild-type littermates; white bars, Taz KD mice). * P < 0.05 level, ** P < 0.01 level .

Decreased tafazzin activity results in altered myocardial generation of biologically potent oxidized signaling metabolites
Signaling metabolites generated from the oxidation of linoleic, arachidonic, and docosahexaenoic acids are potent mediators of calcium homeostasis, infl ammation, and vascular regulation (47)(48)(49)(50)(51). Examination of Taz KD myocardium revealed the complex dysregulation of oxidized 18:2, 20:4, and 22:6 fatty acyl molecular species. Analysis of multiple eicosanoids revealed increases in 5-HETE and 11-HETE as well as a decrease in 15-HETE content in Taz KD compared with the wild-type littermate myocardium ( Fig. 3A ). Interestingly, cardioprotective EETs were unchanged in myocardium. Analysis of prostanoids revealed an increase in PGE 2 , PGF 2 ␣ , TXB 2 , 6keto-PGF 1 ␣ , and PGF 1 ␣ metabolites in the Barth syndrome mouse model that are likely to result in multiple pathologic alterations in infl ammation, ion channel function, and cellular signaling cascades.
The  glutamate-stimulated oxidation was increased by 25% during state 3 respiration in cardiac mitochondria isolated from the Taz KD mice compared with wild-type littermates, which suggests a dramatic shift toward the selection of amino acids for preferential substrate oxidation ( Fig. 4C ). In order to test the adaptability of mitochondria to the utilization of multiple substrates entering the TCA cycle, pyruvate and glutamate were employed to determine the dynamic flux of these TCA cycle substrates used in the wild-type and the Taz KD mouse model . Utilization of pyruvate and glutamate as substrates demonstrated a 15% decrease in state 3 respiration, suggesting that the redox capacity and metabolic fl exibility of the TCA cycle in isolated cardiac mitochondria from the Taz KD mice is defi cient relative to wild-type littermates ( Fig. 4D ). Comparison of multiple substrate combinations to drive state 3 respiration by measuring substrate control ratios demonstrated that fatty acid oxidation is markedly impaired in the Taz KD mouse model, yet amino acid fermentation utilizing glutamate appears to predominate as the preferential fuel to meet energetic demands ( Fig. 4E ). This selective shift in substrate oxidation will lead to multiple downstream bioenergetic repercussions, because normal myocardium generally utilizes fatty acids and glucose under physiological conditions and not amino acids as a primary fuel substrate.

Inhibition of tafazzin expression precipitates alterations in Complex III, Complex V, and glutamate-stimulated adenine nucleotide translocase activities
The effi ciency and enzymatic activity of the electron transport chain has been closely associated with alterations acid release by phospholipases. Analysis of the biologically potent oxidized metabolites of 18:2 fatty acid revealed decreases in 9-HODE, 9-oxoODE, and 9(10)-EpOME, but not other oxidized 18:2 derivatives, demonstrating selective metabolic channeling of the 18:2 fatty acyl chains present in phospholipids due to decreased tafazzin-mediated transacylation ( Fig. 3B ). Investigation of 22:6 oxidized aliphatic chain content, which would be prone to oxidation due to its high degree of unsaturation, reveals a selective decrease in the anti-infl ammatory metabolites RVD1 and RVD2 in the Taz KD compared with wild-type control myocardium ( Fig. 3C ). In contrast, DiHDoHE, DiHDPA, and HDoHE were unchanged.

Tafazzin defi ciency leads to altered myocardial substrate utilization for respiration
Alterations in the mitochondrial membrane lipidome precipitate bioenergetic ineffi ciency and impair adaptive alterations in substrate utilization during metabolic transitions. In the present study, redistribution of acyl chains in cardiolipin and mitochondrial glycerophospholipids in the Taz KD model resulted in a shift in mitochondrial metabolism in a substrate-specifi c manner. Pyruvate oxidation was unaltered in isolated mitochondria from Taz KD cardiac mitochondria compared with wild-type littermates ( Fig. 4A ). However, fatty acid oxidation utilizing palmitoyl-L -carnitine as substrate was decreased by 25% during state 3 respiration in Taz KD cardiac mitochondria compared with wild type. Moreover, this defi ciency was maintained upon addition of succinate, which would combine both Complex I and Complex II electron and proton donation through the respiratory chain ( Fig. 4B ). Surprisingly, to modulate ADP/ATP exchange in a substrate-specifi c manner to direct bioenergetic metabolite oxidation ( 52,53 ). To investigate the differences between state 3 substrate utilization in myocardium from the murine model of Barth syndrome, we measured functional ANT activity driven by pyruvate, glutamate, palmitoyl-L -carnitine, and succinate . Analysis of functional ANT activity revealed a selective 6-fold increase in glutamate-stimulated activity in isolated cardiac mitochondria from Taz KD mice compared with wild-type littermates ( Fig. 5B ). This suggests that altering the mitochondrial lipidome infl uences the substrate selectivity of the ANT leading to downstream changes in electron transport chain fl ux and coupling effi ciency.

Inducible Taz KD results in compensatory alterations in the myocardial transcriptome
To gain further molecular insight into the compensatory mechanisms that result from alterations in the mitochondrial lipidome and myocardial membrane remodeling, we examined the cardiac transcriptome in the Taz KD mouse model of Barth syndrome. Gene Set Enrichment Analysis (GSEA) ( 54, 55 ) revealed dramatic increases in various processes involved in amino acid synthesis, protein translation, and amino acid metabolism in addition to increases in nucleotide metabolism, GTP hydrolysis, and folate metabolism, all of which suggest dramatic compensatory metabolic alterations in response to changes in the mitochondrial lipidome and the accumulation of lysocardiolipin ( Table 1 ). Pathways that were transcriptionally downregulated included branched-chain amino acid catabolism as well as valine, leucine, and isoleucine degradation. Thus, the unexpected effects of increased amino acid synthesis and intraconversion in combination with the decreased catabolism of amino acids revealed dramatic alterations in amino acid and protein metabolism in response to altered lipid remodeling in the mitochondrial membrane which collectively precipitated alterations in substrate utilization.

Removal of doxycycline from the diet for 2 months attenuates bioenergetic and lipidomic dysfunction in the inducible Taz KD mouse model
Utilizing the inherent genetic malleability of the inducible Taz KD mouse model, we examined the effect of removal of doxycycline from the diet following treatment to determine if the distinctive bioenergetic and lipidomic phenotype observed in the Taz KD model was restored to wild-type levels after removal of doxycycline. Following removal of the doxycycline from the diet for 2 months, analysis of glutamate stimulated adenine nucleotide translocase activity in Taz KD cardiac mitochondria, which were 6-fold increased during knockdown ( Fig. 5B ), were attenuated to wild-type level (supplementary Fig. IA). Additionally, high-resolution respirometry analysis of state 3 respiration under various substrates revealed an attenuation of palmitoylcarnitine, glutamate, and pyruvate/ glutamate-stimulated state 3 respiration in Taz KD mice compared with wild-type mice which were removed from doxycycline treatment for 2 months (supplementary Fig. IB).
in the lipid composition of mitochondrial membranes and in the content and molecular species composition of cardiolipin molecular species in particular ( 11,12 ). Due to extensive alterations in cardiolipin molecular species composition and the accumulation of lysocardiolipin in cardiac mitochondria isolated from Taz KD mice, we measured the activities of the electron transport chain complexes in wild-type mice and the mouse model of Barth syndrome. Examination of electron transport chain activities revealed a 45% decrease in Complex III activity and a 25% decrease in Complex V activity in cardiac mitochondria isolated from Taz KD mice compared with wild-type littermates ( Fig. 5A ). These results demonstrate the essential biophysical role of alterations in mitochondrial membrane lipid composition in the Barth syndrome mouse model.
Regulation of ANT activity is partially regulated by the molecular composition of cardiolipin which has been shown

Expression of cardiolipin synthase and inhibition of iPLA 2 ␥ in conjunction with Tafazzin defi ciency leads to altered cardiolipin and lysocardiolipin molecular species
Regulation of cardiolipin remodeling involves several enzymatic steps that could be modulated by pharmacologic intervention to decrease maladaptive cardiolipin remodeling as observed in Barth syndrome due to Tafazzin defi ciency. To determine the potential therapeutic efficacy of increasing cardiolipin synthase (CLS) expression or blocking iPLA 2 ␥ expression as a possible treatment for Barth syndrome, we generated a doubly transgenic mouse strain crossing the inducible Taz KD mice with a transgenic mouse strain that expresses human CLS in a cardiac myocyte-specifi c manner which was previously demonstrated to increase cardiolipin remodeling ( 40 ). Additionally, we crossed the inducible Taz KD mice with a strain that was null for iPLA 2 ␥ , which is involved in the generation of monolysocardiolipin for the transacylation of acyl chains for cardiolipin remodeling ( 56 ). These genetic models were used to interpret the potential to restore alterations in mitochondrial cardiolipin and lysocardiolipin composition and function due to tafazzin downregulation in the murine model of Barth syndrome. Mass spectrometric analysis of the phospholipids of wild-type, Taz KD, Taz KD crossed with CLS-TG, and Taz KD crossed with iPLA 2 ␥ KO male mice at 4 months of age (2 months of doxycycline treatment) revealed distinct alterations in cardiolipin and lysocardiolipin molecular species. Analysis of immature cardiolipin molecular species enriched in 16:0, 16:1, and 18:1 acyl chains revealed that increased expression of cardiolipin synthase or ablation of iPLA 2 ␥ under conditions of Taz KD increased the content of immature cardiolipin molecular species demonstrating that CLS and iPLA 2 ␥ maintain critical roles in the initial stages of cardiolipin remodeling and synthesis that is partially independent of the presence of tafazzin ( Fig. 6A ) The corresponding analysis of the cardiac lipidome also demonstrated a return toward wild-type levels after removal of doxycycline from the diet. More specifi cally, highresolution MDMS-SL analysis of cardiolipin revealed that after 2 months of removal of doxycycline, cardiolipin, monolysocardiolipin, and dilysocardiolipin levels returned to wild-type levels in the inducible Taz KD mouse model (supplementary Fig. IC).  5. Regulation of electron transport chain and adenine nucleotide translocase activities in Taz KD mice. A: Electron transport chain activities in isolated cardiac mitochondria revealed a selective decrease in Complex III and Complex V activities in the Taz KD mouse model compared with wild-type littermates. B: Analysis of the functional ANT activities driven by various substrates revealed a dramatic increase in glutamate stimulated ANT activity in the Taz KD mouse model compared with wild-type littermates. Values represent the mean enzyme activity (nmol/min/mg mitochondrial protein, ETC) or (nmol cATR/mg protein, ANT) ± S.E. (N = 5 isolated cardiac mitochondria per group; black bars, wild-type littermates; white bars, Taz KD mice). ** P < 0.01 level. Data analyzed from GSE33452 using GSEA. and 18:1-18:1 DLCL compared with Taz KD alone, suggesting that iPLA 2 ␥ plays a critical role in the initial and rapid production of DLCL for cardiolipin remodeling in an acyl chain-specifi c manner ( Fig. 6B ). DLCL molecular species in CLS-TG×Taz KD mice were similar compared with Taz   provide further mechanistic insight into the role of tafazzin in the temporal lifecycle of cardiolipin as well as identify potential phospholipid substrate donors used for the transacylation of lysocardiolipin acceptors. Tafazzin deficiency results in the dynamic redistribution of unsaturated acyl chains in the mitochondrial lipidome (primarily in choline and ethanolamine glycerophospholipids) thereby impacting membrane biophysical properties and signaling through alterations as a reservoir of linoleic, arachidonic, and docosahexaenoic fatty acids for release by phospholipases and subsequent oxidation.
Interestingly, loss of tafazzin function in myocardium leads to changes in the mitochondrial lipidome resulting in the dysregulated generation of potent oxidized derivatives of polyunsaturated fatty acids. Thus, tafazzin serves as a previously unrecognized regulator of multiple processes leading to changes in the vasoresponsive and infl ammatory capacity of myocardium in Barth syndrome presumably through its ability to infl uence acyl chain location in phospholipids, the activity of distinct phospholipases, and/or channeling of polyunsaturated fatty acid substrates to a variety of lipoxygenases, cyclooxygenases, and cytochrome P450 enzymes. We specifi cally note that oxidized lipid metabolites also serve as key regulators of ion channel function as well as calcium homeostasis which likely modulate myocardial function in complex metabolic disease such as Barth syndrome ( 50,51,68 ). Additionally, because these oxidized molecules originate from the mitochondria, it would appear that mitochondria in Barth syndrome may also impact mitokine signaling precipitating maladaptive alterations in lipid metabolism, signaling, and bioenergetics.
A comprehensive interrogation of mitochondrial bioenergetics in Barth syndrome myocardium has previously been hindered due to lack of suffi cient appropriate specimens to adequately investigate the full spectrum of bioenergetic capacity. The inducible Taz KD mouse model described in the present study represents a valuable tool for investigation of mitochondrial function as an experimental model of Barth syndrome. Herein, we demonstrate that cardiac mitochondria isolated from tafazzin-defi cient mice are capable of undergoing effective coupled respiration even with a severe defi ciency of mature tetra-acyl cardiolipin as well as the accompanied accumulation of lysocardiolipin species (MLCL and DLCL). High-resolution respirometric analysis of isolated cardiac mitochondria revealed defi ciencies in fatty acid oxidation, which was compensated by increased glutamate-stimulated metabolism, thus demonstrating a characteristic shift in the fl ux of the TCA cycle and substrate preference of cardiac mitochondria to select amino acid fermentation over the normally preferred fatty acid metabolism. This phenomenon was further supported by alterations in key transcriptional pathways indicating that tafazzin defi ciency precipitates altered cardiolipin remodeling thereby resulting in a preferential substrate shift toward de novo amino acid biosynthesis as well as increased amino acid utilization by the TCA cycle. These data appear to support the previous fi nding of increased whole-body protein 18:2-18:2-18:1 MLCL molecular species compared with Taz KD mice, thus demonstrating a critical role of iPLA 2 ␥ in the temporal and sequential remodeling of DLCL and MLCL toward a mature cardiolipin molecular species distribution.

DISCUSSION
Deconvolution of the biophysical, temporal, and integrated roles of cardiolipin, its metabolic intermediates (MLCL and DLCL), as well as the integrated processes by which cardiolipin is remodeled represents a paramount goal to understanding the mechanisms by which the mitochondrial membrane regulates bioenergetic homeostasis ( 57 ). Alterations in cardiolipin molecular speciation are evident in a variety of metabolically complex diseases such as diabetes, heart failure, Tangiers disease, cancer, hyperthyroidism, and neurodegeneration, as well as Barth syndrome ( 21,33,(58)(59)(60)(61)(62)(63)(64)(65)(66). Thus, associating the specifi c roles of cardiolipin molecular species with their causative effects on bioenergetic capacity and metabolic fl ux is a critical objective to develop therapeutic strategies targeting the mitochondrial lipidome to reestablish bioenergetic homeostasis in a variety of complex metabolic diseases. The results of the present study investigating cardiac bioenergetic, lipidomic, and signaling mechanisms in the Taz KD mouse model of Barth syndrome demonstrate: i ) clear resemblance of the mouse model to the human condition resulting in the accumulation of MLCL with the unexpected accumulation of DLCL; ii ) altered distribution of acyl chains in choline and ethanolamine glycerophospholipids; iii ) dysregulated generation of potent oxidized lipid metabolites critical for hemodynamic function; iv ) a shift in preference for glutamate-stimulated oxidation; and v ) an inability of the regulation of cardiolipin synthetic or mitochondrial phospholipase activities to attenuate altered cardiolipin remodeling in the tafazzin shRNA Barth syndrome mouse model.
The phenotype associated with Barth syndrome is intricately intertwined with the loss of tafazzin function, which sculpts and maintains the optimal cardiolipin molecular species distributions to coordinate metabolic homeostasis ( 3 ). The primary cause of death in those affl icted with Barth syndrome is heart failure; however, tools to experimentally dissect the complex molecular pathophysiology of this phenotype did not exist until the generation of the experimental mouse model of Barth syndrome which mimics the pathophysiologic condition in humans ( 23,24 ). Importantly, the inducible Taz KD mouse model presents with cardiomyopathy as well as several other traits characteristic of Barth syndrome ( 25 ). A hallmark of Barth syndrome is the characteristic accumulation of MLCL, which is primarily quantifi ed to confi rm a diagnosis ( 67 ). Untargeted MDMS-SL analyses of cardiolipin species in myocardium from the Taz KD mouse model revealed a dramatic depletion of tetra-acyl CL species as well as a signifi cant increase in MLCL species in addition to an unexpected increase in DLCL molecular species. These results components of the initial remodeling machinery to maintain a homeostatic balance of cardiolipin molecular species.
Although the doxycycline-inducible knockdown construct provides a malleable genetic tool for the investigation of bioenergetic and lipidomic remodeling associated with tafazzin defi ciency, several additional caveats should be considered in regard to the use of tetracycline-inducible promoters used in numerous genetic models. Tetracyclines, including doxycycline, have previously been associated with modulation of secretory phospholipases ( 84,85 ) as well as the inhibition of metalloproteinases, downregulation of cytokines, and cell proliferation ( 86 ), all of which should be considered in regard to the phenotypic characterization of the Barth syndrome mouse model. Because the inducible Taz KD mice were compared with wild-type age-matched littermates also fed a doxycycline-enriched diet, the differential phenotype displayed represents the pathological changes induced by cardiac myocyte tafazzin defi ciency. Furthermore, it was previously reported that the level at which doxycycline or tetracyclines inhibited these biological processes in vitro far exceeds the pharmacological dose that would be administered in vivo ( 86,87 ), thus demonstrating the strength of the pathological fi ndings manifest during tafazzin defi ciency that are reversible upon its reexpression.
In summary, the inducible Taz KD mouse represents an effi cacious model system that recapitulates many of the underlying myocardial lipidomic and bioenergetic phenotypes present in Barth syndrome. Moreover, the use of this model in conjunction with integrated analytic technologies has allowed increased understanding of the complexity of molecular alterations resulting from tafazzin loss of function that likely exist in Barth syndrome patients. Our results demonstrate that tafazzin loss of function results in profound alterations in the myocardial lipidome, deleterious changes in bioenergetic fl ux, and altered signaling processes that collectively contribute to the pathology of Barth syndrome. catabolism in Barth syndrome patients ( 69 ). Pyruvate metabolism was unchanged in tafazzin-defi cient mouse mitochondria indicating that cardiolipin is not obligatory for pyruvate utilization, but does appear essential for fatty acid oxidation. This is predominantly due to the unique role of cardiolipin in maintaining the mitochondrial trifunctional complex, which is essential for effi cient fatty acid oxidation ( 70 ) and cannot be compensated with lysocardiolipin species (i.e., MLCL and DLCL).
Alterations in cardiolipin have previously been associated with the regulation of electron transport chain complex components as well as several other pivotal metabolic enzymes in the mitochondrial membrane (71)(72)(73)(74)(75)(76)(77). Previously, a decrease in Complex III activity in Barth syndrome fibroblasts from two patients was reported in addition to several other minor metabolic deficiencies ( 78 ). In addition, whole-body and oxidative capacity, indirect measurements of mitochondrial function, were signifi cantly decreased during exercise in humans with Barth syndrome ( 5 ). However, mitochondrial function in highly metabolically active tissues such as myocardium is closely integrated with physiologic demands and likely determines the underlying alterations in bioenergetic capacity in vivo in Barth syndrome patients. To identify the upstream mechanism underlying the shift in preference toward glutamate oxidation in the Taz KD mouse, we investigated the adenine nucleotide translocase which exhibits differences in substrate selectivity between various tissues as well as a dependence on cardiolipin for its catalytic activity ( 53,74,79,80 ). The reorganization of cardiolipin and lysocardiolipin molecular species in this Barth syndrome mouse model likely precipitates a dramatic shift in glutamate preference driving ADP/ATP exchange in the mitochondria, thus linking cardiolipin to the substrate-specifi c regulation of respiration.
A distinct advantage of utilizing an inducible shRNA knockdown mouse model of Barth syndrome is its ability to be combined with other genetic tools to investigate therapeutic strategies to target cardiolipin metabolism or other aspects of the disease. Recently, genetic models either expressing human CLS in myocardium or eliminating iPLA 2 ␥ expression have been investigated regarding their participation in cardiolipin biosynthesis and remodeling in the heart ( 40,56 ). Characterization of the cardiac-specifi c human CLS-transgenic mouse revealed a signifi cant increase in CL remodeling and tetra-18:2 cardiolipin content compared with wild-type mice, thus identifying a compensatory mechanism to ameliorate the defi ciency in tetra-18:2 cardiolipin found in Barth syndrome ( 40 ). Furthermore, phospholipases have been suggested as pharmacologic targets to prevent deacylation of cardiolipin, thereby preventing monolysocardiolipin accumulation which is a hallmark characteristic of Barth syndrome (81)(82)(83). In the current study, transgenic expression of human CLS in myocardium or ablation of iPLA 2 ␥ in conjunction with tafazzin defi ciency did not prevent the decreased cardiolipin content (predominantly tetra-18:2 CL) present in the Taz KD mouse, thus demonstrating that CLS and/or iPLA 2 ␥ are likely independent of tafazzin as