A murine model of Barth syndrome recapitulates human cardiac and skeletal muscle phenotypes Open Access

Barth syndrome is a mitochondrial disorder with hallmarks of cardiac and skeletal muscle weakness. It is caused by pathogenic variants in the X-linked gene tafazzin (TAZ), required for cardiolipin remodeling. Previously described germline and conditional Taz knockout models are not ideal for therapeutic development because they lack the combination of robust survival to adulthood, cardiomyopathy and skeletal muscle weakness. We characterized a cardiac and skeletal muscle-specific Taz knockout model (Taz^mKO) in which Cre recombinase is expressed from the muscle creatine kinase promoter (mCK-Cre). Taz^mKO mice survived normally. Cardiolipin composition was abnormal in both heart and skeletal muscle. Taz^mKO had reduced heart function by 2 months of age, and function progressively declined thereafter. Reduced treadmill endurance and diminished peak oxygen consumption were evident by 3 months of age, suggesting reduced skeletal muscle function. Electron microscopy showed abnormalities in mitochondrial structure and distribution. Overall, Taz^mKO mice display diminished cardiac function and exercise capacity while maintaining normal survival. This model will be useful for studying the effects of TAZ deficiency in striated muscles and for testing potential therapies for Barth syndrome.

Barth syndrome is an X-linked mitochondrial disorder characterized by cardiac and skeletal myopathy (Clarke et al., 2013). It is caused by pathogenic variants in the gene tafazzin (TAZ) (Bione et al., 1996), which encodes a cardiolipin transacylase that catalyzes the exchange of cardiolipin acyl chains. Cardiolipin plays critical roles in maintaining the structural integrity of the inner mitochondrial membrane (Ikon and Ryan, 2017; Joubert and Puff, 2021) and the activity of the electron transport chain (Dudek et al., 2016; Pfeiffer et al., 2003), regulating apoptosis (Kagan et al., 2005) and mitochondrial dynamics (Wang et al., 2023; Zhang et al., 2022).

Currently, there are no targeted therapies for Barth syndrome (Thompson et al., 2022). Development of potential therapies, such as adeno-associated virus-mediated expression of TAZ (Suzuki-Hatano et al., 2019; Wang et al., 2020), require an appropriate animal model (Pu, 2022). In the pure C57BL/6J background, Taz germline knockout (Taz-KO) mice are small and have low survival beyond the neonatal period (Wang et al., 2020, 2023). The mice that do survive develop both cardiac and skeletal myopathy (Wang et al., 2020, 2023). This low survival made this model suboptimal for studies of adult mice, including therapeutic trials. When crossed with other inbred strains, the F1 Taz knockout progeny had normal survival and variable degrees of cardiomyopathy or skeletal myopathy, pointing to an important role for genetic modifiers in the expression of Barth syndrome phenotypes (Wang et al., 2023). However, in different inbred strains, cardiac and skeletal myopathy were often dissociated. For instance, CAST[F1] Taz-KO mice had severe cardiomyopathy but normal treadmill endurance, whereas WSB[F1] Taz-KO mice had significantly impaired treadmill endurance and mild, late-onset cardiomyopathy. Other than C57BL/6J, no tested genetic background yielded normal survival and both severe skeletal and cardiac muscle phenotypes.

An alternative approach to obtaining a Taz knockout mouse model for therapeutic trials is conditional gene inactivation using the Cre-loxP system. Previously, we showed that, in the C57BL/6J background, inactivation of a floxed Taz allele by cardiomyocyte-selective Myh6-Cre resulted in severe systolic dysfunction by 3 months of age, while survival was normal (Wang et al., 2020). However, this model lacked Taz deficiency in skeletal muscles.

To obtain a more suitable mouse model for developing new therapeutics for Barth syndrome, we here characterized mCK-Cre conditional Taz knockout mice, in which the cardiac and skeletal muscle selective mCK-Cre transgene (Brüning et al., 1998) catalyzes inactivation of the floxed Taz allele (Ren et al., 2019; Wang et al., 2020).

We crossed mCK-Cre male and Taz^fl/fl female mice to obtain Taz^fl/Y; mCK-Cre (abbreviated as Taz^mKO) and control (Taz^fl/Y without mCK-Cre) littermates. We performed reverse transcription quantitative PCR (RT-qPCR) and capillary western blotting to assess the timing and extent of Taz knockout at postnatal day (P)7, P14 and P28 (Fig. 1A). In heart, Taz mRNA was significantly depleted already at P7. In skeletal muscle, Taz mRNA became significantly depleted at P14. In contrast, Taz mRNA was not significantly affected in the liver.

We measured depletion of TAZ protein by capillary western blotting (Fig. 1B) and quantified the reduction in TAZ protein (Fig. 1C). TAZ protein was markedly reduced in Taz^mKO heart by P7. In skeletal muscle, TAZ protein ablation was not evident until P28. TAZ protein was unaffected in the liver. Together, these studies confirm the selective inactivation of Taz in Taz^mKO cardiac and skeletal muscle.

Unlike germline Taz-KO mice on the C57BL/6J background (Wang et al., 2020), Taz^mKO mice were born at the expected Mendelian frequency, and survival to 6 months of age, the oldest age examined, was normal. Also unlike germline Taz-KO mice, Taz^mKO mice showed a normal increase in body weight compared to the controls at 4, 8 and 12 weeks of age (Fig. 1D). Taz^mKO mice subsequently maintained a relatively stable weight, whereas littermate controls continued to gain weight so that by 24 weeks of age Taz^mKO mice had 20% lower weight. Tibia length was comparable between genotypes (17.8±0.4 mm control versus 17.9±0.4 mm Taz^mKO, n=8 and 14, respectively), indicating that the weight difference is not due to a difference in linear growth. We performed DEXA scanning to measure body mass composition. Taz^mKO mice had a significantly reduced proportion of fat compared to non-fat mass (Fig. 1B). After factoring in this difference in body fat, the calculated non-fat mass of Taz^mKO mice remained lower than that of controls (21.3 g versus 24.9 g). Even though Taz depletion was limited to striated muscle, Taz^mKO mice also had reduced fat mass, suggesting a non-cell autonomous effect of ablation of striated muscle Taz on the amount of adipose tissue.

Most patients with Barth syndrome have cardiomyopathy, leading to infant mortality and heart transplantation in ∼15% of patients (Chowdhury et al., 2022; Roberts et al., 2012). To assess cardiac function, Taz^mKO and control mice underwent echocardiography monthly from 1 to 6 months of age. Cardiac dysfunction, measured echocardiographically by left ventricular fractional shortening, became statistically significant at 3 months of age (Fig. 2A; Fig. S1). With increasing age, systolic heart function continued to decline, whereas that of controls remained relatively stable. Mutant mice had increased left ventricular internal diameter at end systole starting at the age of 3 months, consistent with contractile dysfunction and a systolic heart failure phenotype (Fig. 2B).

After euthanasia, we measured heart weight and normalized it to body weight (HW/BW ratio). At the age of 2 months, the HW/BW ratio was similar between Taz^mKO mice and littermate controls; however, by 6 months of age, the Taz^mKO HW/BW ratio was significantly higher (Fig. 2C). RNA levels of cardiac stress markers Nppb and Myh7 were significantly elevated in Taz^mKO mice compared to controls (Fig. 2D). Nppa was also higher in Taz^mKO mice than in controls but did not reach statistical significance. Myh6, the major cardiac myosin isoform in mice, was significantly decreased. Together, these data indicate that Taz^mKO mice develop systolic heart failure.

We and others previously demonstrated that Taz deficiency strongly activates the integrated stress response, with signature changes found in both the transcriptome and metabolome (Kutschka et al., 2023; Wang et al., 2023; Zhu et al., 2022). An important driver of this response is post-transcriptional upregulation of ATF4, a transcription factor that mediates many of the gene expression changes observed in the integrated stress response. We observed strong upregulation of ATF4 protein in Taz^mKO myocardium (Fig. 2E), consistent with the prior studies.

We analyzed tissue sections for evidence of cardiac fibrosis. Picrosirus Red staining did not reveal significantly increased fibrosis in Taz^mKO myocardium (Fig. 2F). Consistent with this observation, collagen (Col1a1) and periostin (Postn) RNA levels did not differ significantly between groups (Fig. 2G).

Skeletal muscle weakness and low endurance is a prominent symptom of patients with Barth syndrome (Bowen et al., 2019). Skeletal muscle function can be measured by treadmill exercise testing (Bashir et al., 2017; Spencer et al., 2011). The skeletal muscle performance of Taz^mKO mice was assessed by measuring treadmill endurance monthly on a metabolic treadmill. By the age of 3 months, Taz^mKO mice had lower treadmill endurance than littermate controls, and endurance progressively declined thereafter (Fig. 3A). Using a metabolic treadmill, we monitored oxygen consumption (VO₂) and carbon dioxide production (VCO₂) rates, and the respiratory exchange ratio (RER=VCO₂/VO₂). Peak oxygen consumption was significantly decreased in mutant mice at 4 and 6 months of age (Fig. 3B). The RER at exhaustion did not significantly differ between genotypes (Fig. 3C).

We compared quadricep and gastrocnemius muscle weights at 2 and 6 months of age (Fig. 3D). At 2 months, muscle weights were comparable. At 6 months of age, Taz^mKO quadricep muscles had significantly lower mass than that of controls, whereas gastrocnemius weights did not significantly differ.

To assess skeletal muscle fibrosis, we performed Picrosirius Red staining of quadricep muscles at 6 months of age. We observed no increase in fibrosis in Taz^mKO muscles (Fig. 3E).

To further characterize the Taz^mKO mice, we performed indirect calorimetry on 4- to 5-month-old mice and, at the same time, monitored food intact and activity level. Mice were acclimated to the system for 31 h and then monitored for 84 h. Energy expenditure was higher in control mice than in Taz^mKO mice (Fig. 4A). To account for the higher weight of control mice, we analyzed the relationship between energy expenditure and weight. In control mice, energy expenditure was proportional to body weight, as expected; in Taz^mKO mice, energy expenditure did not vary with body weight (Fig. 4B), perhaps owing to low energy expenditure by mutant muscle tissue. Surprisingly, Taz^mKO mice had decreased RER during the light photoperiod, suggestive of increased fatty acid utilization (Fig. 4C,D), which may contribute to their reduced body fat (Fig. 1E). RER did not significantly differ between genotypes during the dark photoperiod, although RER appeared lower in mutants than in controls in the late-dark photoperiod (Fig. 4C, arrows), when mice have reduced activity and food intake compared to in the early-dark photoperiod. Food intake and activity levels did not significantly differ between genotypes (Fig. 4E,F).

TAZ catalyzes the biogenesis of mature cardiolipin (CL) (Schlame and Greenberg, 2017), which, in cardiac muscle, primarily contains four 18:2 acyl chains (tetralinoleoyl CL). TAZ deficiency disrupts normal CL remodeling, resulting in lower total CL, elevated monolysocardiolipin (MLCL) and atypical acyl chain composition (Schlame and Greenberg, 2017). We used mass spectrometry to measure the CL and MLCL composition of 6-month-old Taz^mKO cardiac and quadricep muscles. Taz^mKO cardiac muscles had the low CL, high MLCL and high MLCL/CL ratio that is the hallmark of TAZ deficiency (Fig. 5A-C). Taz^mKO skeletal muscle had a trend towards lower CL, no significant difference in MLCL and a significantly elevated MLCL/CL ratio (Fig. 5A-C). These changes in CL, MLCL and MLCL/CL in skeletal muscle were more mild than has been reported previously in skeletal muscle of mice with widespread Taz knockdown (Acehan et al., 2011) or in patients with Barth syndrome (Houtkooper et al., 2009). This difference might be due to differences in species, strain, muscles sampled and potentially incomplete Taz inactivation in Taz^mKO mice.

CL and MLCL species differ by their acyl chain composition. We compared the relative abundance of individual CL species in heart (Fig. 5D) and skeletal muscle (Fig. 5E). Control myocardium showed the expected strong enrichment for tetralinoleoyl CL (four 18:2 acyl chains=72:8) as well as 72:7 (Fig. 5D). In Taz^mKO myocardium, these species were depleted. The most abundant CL species in mutant heart was 68:2, which was elevated compared to that in control heart. Control skeletal muscle tissue showed a wider range of CL species, with 72:8 and 72:7 being the most abundant. Taz^mKO skeletal muscle was depleted for these isoforms (Fig. 5E). The most abundant CL species in mutant skeletal muscle were 68:4, 68:3 and 68:2, which were comparable in abundance to in controls.

Each MLCL species was present at low levels in control heart and elevated in Taz^mKO heart (Fig. 5F). In skeletal muscle, the abundance of each MLCL species was comparable between mutant and control (Fig. 5G).

Together, these observations indicate that Taz^mKO skeletal and cardiac muscle have CL abnormalities consistent with Taz deficiency.

CL is a critical lipid of the mitochondrial inner membrane that has been implicated in its folding into cristae (Ikon and Ryan, 2017; Joubert and Puff, 2021), and TAZ deficiency causes abnormal mitochondrial morphology, size and number (Acehan et al., 2011; Soustek et al., 2011; Wang et al., 2020, 2023). We used electron microscopy to analyze mitochondrial morphology in Taz^mKO heart and skeletal muscle. Consistent with prior studies of TAZ-deficient cardiac muscle (Acehan et al., 2011; Soustek et al., 2011; Wang et al., 2020, 2023), Taz^mKO cardiac mitochondria had abnormal morphology and organization. The Taz^mKO cardiac mitochondria had highly simplified internal cristae (Fig. 6A) and lower cross-sectional area than those of controls (Fig. 6A,B). The number density of cardiac mitochondria was higher in mutants than in controls, although this difference did not reach statistical significance (Fig. 6A,C). Control cardiac mitochondria were organized so that each sarcomere neighbored and contacted approximately two mitochondria, whereas in Taz^mKO mice the mitochondria appeared piled up, and a significantly higher fraction was not in contact with a sarcomere (Fig. 6A,D).

In skeletal muscle, Taz^mKO mitochondria also had sparse cristae and lower cross-sectional area than controls (Fig. 6A,B). Unlike in cardiac muscle, in skeletal muscle, the number of mitochondria was lower than that in controls, with the difference being significant in quadriceps and tending towards significance in gastrocnemius (Fig. 6C). Also different from cardiac muscle, mitochondria did not form disorganized clumps in Taz^mKO skeletal muscle (Fig. 6A). Mitochondria remained localized near Z-lines, but the average number of mitochondria per sarcomere was lower in Taz^mKO skeletal muscles than in control skeletal muscles (Fig. 6E).

To further characterize the effect of TAZ depletion on mitochondria, we measured the level of proteins in respiratory complexes I-V in the heart and quadriceps (Fig. 6F,G). In the heart, the level of complex I protein NDUFB8 was significantly lower in Taz^mKO mice than in control mice, consistent with prior studies of hearts from Taz knockdown mice (Le et al., 2020) and patients with Barth syndrome (Chatfield et al., 2022). In the quadriceps, we did not observe a significant difference in the levels of the measured respiratory complex proteins.

We characterized a new Barth syndrome mouse model in which a floxed Taz allele is inactivated by mCK-Cre, a well-characterized Cre allele selectively active in cardiac and skeletal muscle (Brüning et al., 1998). mCK-Cre regulatory elements turn on late in embryogenesis and in the first postnatal week (Brüning et al., 1998). Consistent with this timing, Taz transcript levels declined from P7 to P14, and TAZ protein was significantly decreased by P7 in heart and by P28 in skeletal muscle. Taz^mKO mice survived normally and had normal size through 8 weeks of age, unlike germline Taz-KO mice, which are significantly smaller from birth (Wang et al., 2020). Taz^mKO mice subsequently had lower body weight than their control counterparts. Taz^mKO mice also had significantly reduced fat mass compared to control littermates, indicating that Taz expression in striated muscle has cell non-autonomous effects on the size of the adipose compartment. This might be due to increased fatty acid oxidation during light photoperiods, which we detected by indirect calorimetry. Because Taz is required for efficient fatty acid oxidation, increases in fatty acid oxidation likely occur in Taz-replete, non-muscle tissue such as adipose tissue.

Likely owing to Taz inactivation in the early postnatal period, Taz^mKO mice survived normally to adulthood. Taz^mKO had progressive cardiac dysfunction. The degree of systolic dysfunction was less severe than that of either Taz-KO mice or Taz^fl/Y; Myh6-Cre cardiac-specific knockout mice. Furthermore, we did not detect substantial cardiac fibrosis, unlike in Taz-KO or Taz^fl/Y; Myh6-Cre models (Wang et al., 2020, 2023). This could be due to later timing or possibly lower extent of cardiomyocyte Taz inactivation by mCK-Cre. Taz^mKO mice also had progressive declines in treadmill endurance, with an associated reduction in maximal VO₂. The phenotype was more mild than that of Taz-KO mice in the same C57BL/6J genetic background (Wang et al., 2020, 2023), again possibly owing to later or less complete Taz inactivation. Because we have observed that Taz-KO mice in the CAST-C57BL/6J F1 background have severe cardiac dysfunction yet normal treadmill endurance (Wang et al., 2023), it is most likely that reduced treadmill endurance in Taz^mKO mice is a skeletal muscle phenotype, rather than a result of impaired cardiac function. We previously noted stark differences in the phenotypes of Taz-KO mice in different strain backgrounds, with C57BL/6J mice exhibiting the most severe combination of cardiac and skeletal muscle phenotypes (Wang et al., 2023). This study also used the C57BL/6J strain, and it is likely that the phenotypes observed would be weaker in other strain backgrounds.

Taz^mKO muscles exhibited hallmark biochemical features of Taz deficiency, including upregulation of ATF4, a marker of the integrated stress response (Kutschka et al., 2023; Wang et al., 2023; Zhu et al., 2022) and abnormal cardiolipin composition (Schlame and Greenberg, 2017). Associated with these cardiolipin abnormalities, Taz^mKO muscles contained small, disorganized mitochondria with reduced cristae, consistent with mitochondrial abnormalities previously noted in Taz-KO cardiomyocytes (Acehan et al., 2011; Soustek et al., 2011; Wang et al., 2020, 2023). In addition to these similarities between Taz^mKO heart and skeletal muscle mitochondria, we also observed differences. In the heart, the number of mitochondria per field tended to be higher than in controls (although this did not reach statistical significance), and mitochondria bunched together and lost their characteristic alignment next to sarcomeres. In the quadriceps, the number of mitochondria per field was significantly reduced, but mitochondria retained their characteristic alignment next to sarcomeres. Our analysis of Taz-KO mice in different genetic backgrounds suggested a correlation between cardiac dysfunction and impaired mitophagy (Wang et al., 2023), and we suspect that the abnormal organization and number of cardiac mitochondria reflects altered mitochondrial dynamics and quality control, which may be restricted to cardiomyocytes.

It is important to recognize that Taz^mKO mice do not mimic all aspects of Barth syndrome. For instance, patients with Barth syndrome have considerable heterogeneity in clinical manifestations, with a subset having severe neonatal cardiomyopathy that requires transplantation (Taylor et al., 2022). This aspect of Barth syndrome is not well modeled in Taz^mKO mice or even in Taz-KO [C57BL/6J] mice (Wang et al., 2020), which do have neonatal cardiac dysfunction. Patients with Barth syndrome (Spencer et al., 2011) exhibit markedly elevated RER during exercise, and mice with widespread Taz knockdown also had elevated RER at peak exercise (Powers et al., 2013). In contrast, RER did not significantly differ between control and Taz^mKO mice at peak exercise. Under normal cage activity, RER was even significantly lower in Taz^mKO mice during light photoperiods, unlike mice with widespread Taz knockdown, which had unchanged RER during light photoperiods and elevated RER during dark photoperiods (Goncalves et al., 2021). This difference suggests that Taz-deficient striated muscles stimulate increased fatty acid oxidation in other Taz-replete tissues in the Taz^mKO model. Perhaps related to the divergent RER responses between patients with Barth syndrome and Taz^mKO mice, the exercise intolerance of Taz^mKO mice is comparatively mild, with no difference in activity under normal cage activity and a modest (∼33%), but significant, reduction in endurance with forced treadmill running. Finally, we noted more mild abnormalities of cardiolipin composition in Taz^mKO skeletal muscle than in prior studies of Taz knockdown mice or patients with Barth syndrome, which might be due to differences in species, strain, muscles sampled and potential for incomplete Taz inactivation in the Taz^mKO mice. This difference may contribute to a more mild skeletal muscle phenotype than that observed in Taz^KO [C57BL/6J] mice that survive to adulthood (Wang et al., 2020, 2023) and should be kept in mind when using the Taz^mKO model.

Despite these limitations, the Taz^mKO mouse models aspects of Taz deficiency in striated muscles and is a suitable pre-clinical model to test the efficacy of novel therapies designed to the treat the heart and skeletal muscle manifestations of Taz deficiency in Barth syndrome.

Animal experiments were performed under protocols approved by the Boston Children's Hospital Institutional Animal Care and Use Committee. The mice were housed in groups of two to five per cage in an animal facility with exposure to a standard diet. mCK-Cre transgenic (Brüning et al., 1998) male mice were bred with Taz^fl/fl females, in which exons 5-10 are flanked by loxP sequences (Ren et al., 2019; Wang et al., 2020). Litters were genotyped for mCK-Cre and Taz^fl, and males were used for the experiments described in this paper. Genotyping was performed by PCR analysis of toe clip genomic DNA using the primers listed in Table S1.

We measured the exercise capacity of mice using an incremental exercise test on a metabolic treadmill, which provides gas exchange measurements, including oxygen consumption (VO₂) and carbon dioxide production (VCO₂) rates normalized to mouse body weight (Columbus Instruments). Mice were placed in an enclosed chamber containing a motorized treadmill with adjustable speed and incline. The treadmill was equipped with an electrified metal grid at the end of the moving belt to provide motivation for mice to run rather than rest on the grid. Electric shock intensity was set to 0.08 mA. Animals were all familiarized with the treadmill prior to the initiation of the protocol. In the test, the mice were recorded at rest for 5 min, then started to run at a 5% incline at 5 m/min, increasing by 5 m/min every 5 min until exhaustion. The shock grid was turned off or the treadmill was stopped for exhaustion if the mice stayed on the shock grid for over 5 s or if they stayed on the shock grid for 2 s or more five times. Gas samples were collected and analyzed by the Oxymax system throughout the experiment. The time mice spent running on the treadmill was recorded manually.

Mice were sacrificed at 1 and 6 months. Samples of heart muscle and skeletal muscles (quadriceps and gastrocnemius) were collected. The body weight, absolute weights of collected tissues and tibial lengths were recorded. Samples were collected, fixed and stored as appropriate.

DEXA scanning (Piximus) was performed under isoflurane anesthesia to collect body composition measurements, including fat and non-fat tissue.

Echocardiography was performed monthly in unsedated mice using a Vevo 2100 (VisualSonics). After removal of chest hair, mice were held in a standard hand grip. M-mode short-axis images were acquired for heart function measurements. Measurements included heart rate, fractional shortening, left ventricular internal diameter at end diastole, left ventricular internal diameter at end systole, left ventricular posterior wall at end diastole and left ventricular posterior wall at end systole. Echocardiography was performed unaware of genotype.

Mice were housed at 23°C in a Promethion indirect calorimetry system (Sable Systems International) within a temperature-controlled cabinet. Data collected include VO₂, VCO₂, physical activity beam breaks, food intake and body mass. For the duration of the experiment, the mice had ad libitum access to Labdiet 5008 (3.56 kcal/g) and were maintained on a 12 h/12 h photoperiod with lights on from 06:00 to 18:00. Data were analyzed with CalR2 (Mina et al., 2018).

Cardiac and skeletal muscle samples were collected and fixed overnight in 4% paraformaldehyde (PFA) at 4°C with rotation. Tissue samples were paraffin embedded, sectioned and stained with Picrosirius Red/Fast Green. Images were captured at 20× magnification using a wide-field Keyence microscope.

Samples were fixed in EM fixative (1.25% paraformaldehyde, 2.5% glutaraldehyde and 0.03% picric acid in 0.1 M sodium cacodylate buffer, pH 7.4) overnight at 4°C. After fixation, tissues washed in 0.1 M sodium cacodylate buffer and postfixed with 1% OsO₄/1.5% KFeCN₆ for an hour. Samples were then washed in water twice, once in 50 mM maleate buffer pH 5.15 and incubated in 1% uranyl acetate in maleate buffer for 1 h. Samples were then washed in maleate buffer, then water twice, and dehydrated for 10 min each in 50%, 70%, 90%, 100% and 100% ethanol. The samples were then put in propyleneoxide for 1 h and infiltrated overnight in a 1:1 mixture of propyleneoxide and TAAB Epon. The following day, the samples were embedded in TAAB Epon and polymerized at 60°C for 48 h. Then, 80 nm sections were cut on a Reichert Ultracut S microtome and imaged with a JEOL 1200EX transmission electron microscope at 80 kV. Images were recorded with an AMT 2k CCD camera and analyzed using ImageJ.

Cardiac and skeletal muscles were collected and snap frozen. Tissues were homogenized in RIPA buffer at 25 Hz for 2 min using a Qiagen TissueLyser II, incubated for 10 min, then centrifuged at 10,000 g for 10 min at 4°C, and the supernatant was collected. Protein concentrations were obtained using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Protein quantification was performed using a capillary electrophoresis device (WES, ProteinSimple). Primary antibodies used were as follows: anti-tafazzin mouse monoclonal antibody (mAb), Santa Cruz Biotechnology, sc-365810 (1:40 dilution); anti-Cre-recombinase rabbit mAb, Cell Signaling Technology, 15036S (1:100 dilution); anti-ATF4 rabbit mAb, Cell Signaling Technology, 11815S (1:100 dilution); anti-GAPDH rabbit mAb, Life Technologies, PA116777 (1:100 dilution); and OxPhos Rodent WB Antibody Cocktail (Thermo Fisher Scientific, 45-8099).

Tissue samples were collected and snap frozen. Total RNA was purified using TRIzol (Thermo Fisher Scientific). Reverse transcription was performed using Superscript III reverse transcriptase (Life Technologies). RT-qPCR was performed in duplicate for each sample using SYBR Green Master Mix (Thermo Fisher Scientific) and a Bio-Rad CFX96 or CFX384 Touch instrument. Gene expression values were normalized to Gapdh. The primer sequences used in this study are listed in Table S1.

Tissue was spiked with cardiolipin mix I (Avanti, LM6003; 30 µl internal standard per 10 mg tissue), and lipids were extracted with chloroform/methanol. Extracted lipids were analyzed by matrix-assisted laser desorption/ionization-time of flight mass spectrometry. Cardiolipin and monolysocardiolipin were expressed as pmol per mg tissue.

Statistical analyses were performed using GraphPad Prism 10 software. All values were expressed as the mean±s.d. The statistical test for each comparison and the number of samples are indicated in the figure legends.

Electron microscopy and indirect calorimetry were performed with support from the Beth Israel Deaconess Medical Center Electron Microscopy and Energy Balance Core facilities.