Partial restoration of mitochondrial dysfunction by AAV-Ant1 protects from dilated cardiomyopathy in Ant1^-/- plus mtDNA mutant mice

Abstract

Primary mitochondrial disease (PMD) patients manifesting cardiomyopathy are twice as likely to die as other PMD patients. One PMD with cardiomyopathy is caused by null mutations in the heart-muscle isoform of the adenine nucleotide translocator (SLC25A4, ANT1) gene, with the severity of cardiomyopathy mediated by mitochondrial DNA. To optimize strategies for addressing mitochondrial cardiomyopathy, we generated an Ant1 null mouse and combined it with the ND6^P25L mitochondrial DNA mutation to mimic the hypertrophic versus dilated cardiomyopathies observed in patients. Here, we transduce the neonatal Ant1^-/- and Ant1^-/-+ND6^P25L mouse hearts with an AAV2/9-pDes-Gfp-mAnt1 cDNA vector. We show that restoration of just 10% of Ant1 gene expression was sufficient to ameliorate the cardiomyopathies in these mice. Proteomics and single-nucleus RNA sequencing reveal the reversal of dysregulated mitochondrial metabolic genes, including PGC1α, as well as cardiac contractile and extracellular matrix proteins. Hence, a modest increase in cardiac mitochondrial energetics can have profound benefits on cardiac function and is effective in treating mitochondrial cardiomyopathy.

Patients with primary mitochondrial disease manifesting cardiomyopathy are twice as likely to die compared to those without cardiomyopathy. Here, the authors show that a modest increase in cardiac mitochondrial energetics via gene therapy can significantly improve cardiac function and is effective in treating mitochondrial cardiomyopathy.

Introduction

Primary mitochondrial disease (PMD) is one of the most common inherited metabolic disorders affecting both children and adults, with an incidence of ~1 in 5000. PMD can result from mutations encoded in over 350 nuclear DNA (nDNA) genes¹ or in the 37 mitochondrial DNA (mtDNA) genes². Pediatric mitochondrial diseases are more often the result of nDNA mutations, while adult mitochondrial diseases are more likely the result of mtDNA mutations^1,3,4.

20–40% of PMD patients develop cardiomyopathy between birth and 27 years of age^4–8, with the ten-year mortality being 67-71% for PMD children with cardiomyopathy versus 18–26% for PMD children without cardiomyopathy⁶. Therefore, it is imperative that we develop an approach to treat mitochondrial cardiomyopathy if we are to save the lives of the most vulnerable PMD patients.

To address this critical clinical need, we have focused on a PMD resulting from an nDNA gene mutation whose severity is modulated by the mtDNA genetic background, the combination of the nDNA and mtDNA variant resulting in life-threatening pediatric dilated cardiomyopathy. These patients harbor homozygous null mutations in the nuclear-coded adenine nucleotide translocator-1 (ANT1) gene (SLC25A4), and ANT1^−/− patients present with cardiomyopathy and myopathy^9–11.

Starting with a thirteen-generation consanguineous pedigree harboring a SLC25A4 frameshift mutation (c.523delC, p.Q175RfsX38), the homozygous null patients paired with the common European mtDNA haplogroup H (mtDNA^H) manifest progressive hypertrophic cardiomyopathy associated with myocardial thickening, hyperalaninemia, lactic acidosis, exercise intolerance, and persistent adrenergic activation. Echocardiographic studies revealed abnormal contractile mechanics, myocardial repolarization abnormalities, and impaired left ventricular relaxation. Their end-stage heart disease is characterized by symmetric, concentric cardiac hypertrophy, widespread cardiomyocyte degeneration, and extensive subendocardial interstitial fibrosis, which is compatible with life into middle age. However, family members with the same ANT1^−/− nDNA mutation but who inherit the rarer haplogroup “U2” mtDNA (mtDNA^U2) from their mothers develop pediatric dilated cardiomyopathy. Haplogroup “U2” differs from haplogroup “H” in harboring an MT-RNR2 (16S rRNA) m.1811A>G and an MT-TL2 (tRNA^Leu(CUN)) m.12308A>G mutation, which may diminish the translation of the seven mtDNA-coded oxidative phosphorylation (OXPHOS) complex I polypeptides⁹.

To model this pedigree, we inactivated the mouse heart-muscle ANT1 gene (Ant1), resulting in autosomal recessive myopathy and hypertrophic cardiomyopathy^12,13. We then combined this nDNA mitochondrial gene mutation with our mtDNA complex I missense mutation MT-ND6 m.13997G>A (P25L, ND6^P25L)¹⁴, the same amino acid change that results in maternally inherited PMD in humans [ND6 m.14600G>A (P25L, ND6^P25)]¹⁵. The Ant1^−/− plus ND6^P25L combination results in early onset dilated cardiomyopathy, affecting echocardiogram measurements and leading to an 80% reduction in cardiac contractility and dramatic ventricular dilation¹⁶, thus modeling the severe dilated cardiomyopathy manifest by the ANT1^−/−+mtDNA^U2 children¹⁶.

There are four ANT isoforms in humans^17–27 and three isoforms in mouse^12,28–31. ANT1 is the predominant ANT in muscle, while ANT1 and ANT2 are present in the heart, raising the possibility that ANT2 may partially compensate for the ATP export deficiency in the ANT1 null heart²⁹.

ANT1/Ant1 has been found to have four functions: ATP/ADP exchange, modulation of the permeability transition pore (mtPTP)^30,32, generation of a lipid-modulated proton channel³³, and regulation of mitophagy³⁴. Ant1 deficiency in muscle is associated with a massive accumulation of mitochondria with abnormal morphology^34,35.

There is no FDA-approved cure for PMD-associated cardiomyopathy. Current treatment is primarily palliative^36–38, prompting us to develop a gene therapy. Using the Ant1^−/− and Ant1^−/−+mtDNA ND6^P25L mice as a model system, we show that AAV-Ant1 mediated gene therapy can mitigate the mouse dilated cardiomyopathy. This could save the lives of the ANT1^−/−+mtDNA^U2 children who typically require a heart transplant at a young age. To treat the nDNA Ant1^−/− defect, we use an adeno-associated virus (AAV), which permits the transduction of the Ant1 cDNA into the Ant1^−/− and Ant1^−/−+mtDNA ND6^P25L postnatal mouse hearts. Our results show that even a partial rescue of Ant1 expression significantly ameliorates both the Ant1^−/− hypertrophic and the Ant1^−/−+mtDNA ND6^P25L dilated cardiomyopathies.

Results

AAV-mediated transduction of the mitochondrial Ant1 gene

To transduce the hearts of Ant1^−/− mice, we designed a heart- and muscle-directed vector using the mouse Desmin promoter (pDes), Gfp reporter gene, and mouse Ant1 cDNA, the cDNAs separated by a 2A translation interruption sequence^39,40, encapsulated within the AAV2/9 coat (AAV2/9-pDes-Gfp-mAnt1) and referred to as AAV-Ant1 hereafter (Fig. 1A). As one control, we generated a second AAV2/9 construct without the Ant1 gene (AAV2/9-pDes-Gfp).

Fig. 1. AAV-ANT1 transduction of Ant1^−/− mouse hearts.

A Representation of the AAV2/9-pDes-Gfp-mAnt1 construct (generated in SnapGene); B Scheme of the experimental design (Ant1^+/+ = WT; Ant1^−/−; Ant1^−/−+ND6^P25L) created via BioRender, https://BioRender.com/ycg4fqw; C Representative image of Gfp mRNA detection in liver, diaphragm, and heart from an 1-month-old mouse injected with the AAV2/9-pDes-Gfp-mAnt1 vector (up) (n = 2); Representative image of GFP IHC staining in an isolated heart from a 1-month-old mouse injected with the AAV2/9-pDes-Gfp-mAnt1 vector (bottom) (n = 3); D Western blot of ANT1 and ANT2 proteins in isolated mitochondria from Ant1^+/+ = WT, Ant1^−/−, and Ant1^−/−+ND6^P25L 1-month-old mice injected with AAV2/9-pDes-Gfp-mAnt1 (AAV-Ant1) or PBS (WT n = 3, PBS-treated Ant1^−/− n = 2, AAV-Ant1-treated Ant1^−/− n = 2, PBS-treated Ant1^−/−+ND6^P25L n = 1, and AAV-Ant1-treated Ant1^−/−+ND6^P25L n = 2).

We injected the AAV-Ant1 vector into the pericardium of male and female Ant1^+/+ (wild-type, WT); Ant1^−/− (Ant1-null); and Ant1^−/−+ND6^P25L (Ant1-null plus partial mtDNA complex I defect) mice at three days of age. Littermates were injected with PBS as a control for the procedure or with AAV2/9-pDes-Gfp to confirm that the cardiac therapeutic benefit was not due to the vector or GFP expression. AAV-Ant1 and PBS-injected animals were continuously monitored, and their cardiac function was evaluated quarterly for 12–14 months post-injection (Fig. 1B).

At one month post-pericardial injection, Gfp mRNA was detected in the hearts of AAV-Ant1 transduced mice, as well as in the diaphragm and liver. Hence, the heart was transduced but some vectors diffused from the pericardium to other tissues (Fig. 1C, top). To further investigate the distribution of the vector following pericardial delivery, we evaluated the presence of the vector genome by performing digital PCR (dPCR) in the heart, liver, lung, gastrocnemius, and diaphragm tissues isolated from Ant1^+/+ AAV-Ant1 transduced mice (Supplementary Fig. 1). This revealed that the AAV vector is present in the liver at similar level to the heart and present at the level ~3 fold less in the lung, diaphragm and gastrocnemius than the heart. The expression of the GFP protein was confirmed by immunohistochemical staining in coronal, anterior, and posterior cardiac sections, with GFP protein being observed in cells dispersed throughout the myocardium (Fig. 1C, bottom).

To confirm the expression of the Ant1 cDNA, mitochondria were isolated from the heart of the AAV-Ant1 transduced mice and tested for ANT1 protein by western blot. ANT1 protein was found in transduced Ant1^−/− and Ant1^−/−+ND6^P25L mice but at a much lower level than found in Ant1^+/+ mice, about 10% (Fig. 1D, and Supplementary Fig. 2). While the ANT2 protein was present in the Ant1^−/− and Ant1^−/−+ND6^P25L transduced mice, ANT2 levels were not increased relative to WT indicating that the Ant2 gene was not induced to compensate for the cardiac Ant1 defect (Fig. 1D).

To determine if the ~10% production of Ant1 was due to low expression in all the cells or high expression in a small percentage of cells, we determined the percentage of GFP-positive cardiomyocytes. On average, 20% of the cardiomyocytes were positive for GFP (Fig. 2A, B). Hence, a small percentage of cardiomyocytes were transduced but they expressed high levels of transgene products.

Fig. 2. Evaluation of the percentage of AAV-Ant1 transduced cells.

A Representative section of one Ant1^+/+ AAV-Ant1 transduced heart stained with Hoechst for nuclei (blue), WGA to outline cells (red), and GFP (green). Successfully transduced cardiomyocytes are surrounded by red, with the cytoplasm stained green and nuclei blue. B Quantification of the percentage of GFP-positive cardiomyocytes, with each dot representing data from one microscopic field. Three sections from each heart, representing different anatomical planes (coronal, anterior, and posterior), were analyzed, with five random fields examined per section across three hearts, totaling 5 × 3 × 3 = 45 random fields. Center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range. An average of ~20% of cardiomyocytes are GFP-positive and thus transduced. C Representative images of GFP immunohistochemical staining intensity, indicating overall ventricular GFP expression. D Quantification of the relative ventricular area with GFP expression in AAV-Ant1 transduced hearts at 1, 3, 6, 9, and 12 months. Each point represents a mouse (WT = black dots, Ant1^−/− = red dots, and Ant1^−/−+ND6^P25L = blue dots; Median is indicated; n = 2 per condition), with individual values reported in the table.

Immunohistochemical staining for GFP in cardiac cells was performed on hearts at 3, 6, 9, and 12–14 months post-transduction. This revealed cardiac GFP at all time points, confirming that the AAV-Ant1 transgenes were stably expressed for over one year, though some variability in expression was noted by 12–14 months (Fig. 2C, D, Fig. 3A, and Supplementary Fig. 3). We found no significant tissue damage generated by the injection procedure, as determined by H&E and Trichrome staining, nor any increase in cell death, as shown via TUNEL staining (Fig. 3B and Supplementary Fig. 4) and cleaved caspase 3 (Supplementary Fig. 5), nor inflammation, as shown via interleukin-18 (IL-18) staining (Supplementary Fig. 6). Pericardial transduction with AAV2/9-pDes-Gfp confirmed GFP histochemical expression without altered cardiac function or toxicity (Supplementary Fig. 3B and Supplementary Table 1).

Fig. 3. Immunohistochemistry of AAV-ANT1 transduction of Ant1^+/+, Ant1^−/−, Ant1^−/−+ND6^P25L mice.

A Representative images of GFP IHC staining in isolated hearts from 6 and 12 months old wild-type (WT) = Ant1^+/+, Ant1^−/−, and Ant1^−/−+ND6^P25L mice injected with the AAV2/9-pDes-Gfp-mAnt1 vector or PBS; B Representative images of H&E, Trichrome and TUNEL staining in isolated hearts from 6-month-old WT = Ant1^+/+, Ant1^−/− and Ant1^−/−+ND6^P25L mice injected with the AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) or PBS. For the histological evaluation, n = 3 mice per condition were analyzed.

Amelioration of Ant1-deficient cardiomyopathy by AAV-Ant1 transduction

To determine the effect of AAV-Ant1 transduction on cardiac function, we performed echocardiographic analysis on male and female Ant1^+/+; Ant1^−/−; and Ant1^−/−+ND6^P25L mice without and with AAV-Ant1 viral transduction at 3, 6, 9, and 12 months post-pericardial injection. Using the M-Mode echocardiographic measurements of the left ventricle in short and long views, we observed marked alterations in the Ant1^−/− and Ant1^−/−+ND6^P25L hearts relative to Ant1^+/+ hearts in the stroke volume, ejection fraction, fractional shortening, cardiac output, left ventricle mass, and the thickness of the anterior and posterior walls. The heart rate of the Ant1^−/− and Ant1^−/−+ND6^P25L mice was slightly higher than that of the Ant1^+/+ mice but not statistically significant (Figs. 4A, D and Supplementary Table 1). Differences were found between sexes, with the cardiomyopathy in females being less severe than in males (Fig. 4).

Fig. 4. Ultrasound and heart/body weight analysis of AAV-ANT1 transduction of Ant1^+/+, Ant1^−/−, Ant1^−/−+ND6^P25L mouse hearts.

A Echocardiograph analysis of 6-month-old WT = Ant1^+/+ (black dots), Ant1^−/− (red dots), and Ant1^−/−+ND6^P25L (blue dots) male mice injected with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) (empty circles) or PBS (filled circles). Stroke Volume Ant1^−/− PBS vs AAV-Ant1 **p = 0.006 and WT PBS vs mutant ***p < 0.001, Ant1^−/−+ND6^P25L PBS vs AAV-Ant1 **p = 0.01 and WT PBS vs mutant ****p < 0.0001; Ejection Fraction Ant1^−/− PBS vs AAV-Ant1 *p = 0.019 and WT PBS vs mutant ****p < 0.0001, Ant1^−/−h + ND6^P25L PBS vs AAV-Ant1 *p = 0.026 and WT vs mutant ****p < 0.0001; Fractional Shortening Ant1^−/− PBS vs AAV-Ant1 *p = 0.015 and WT PBS vs mutant ****p < 0.0001, Ant1^−/−+ND6^P25L PBS vs AAV-Ant1 *p = 0.023 and WT vs mutant *^***p < 0.0001; Cardiac Output Ant1^−/− PBS vs AAV-Ant1 **p = 0.004 and WT PBS vs mutant ***p = 0.0002, Ant1^−/−+ND6^P25L PBS vs AAV-Ant1 **p = 0.008 and WT PBS vs mutant ****p < 0.0001; Left Ventricles mass Ant1^−/− PBS vs AAV-Ant1 *p = 0.016 and WT vs mutant ***p = 0.0003 and ****p < 0.0001, Ant1^−/−+ND6^P25L PBS vs AAV-Ant1 *p = 0.020 and WT vs mutant ****p < 0.0001. B Male mice heart/body weight ratio measured in 3, 6, 9, and 12 months WT = Ant1^+/+ (black), Ant1^−/− (red), and Ant1^−/−+ND6^P25L (blue) injected with AAV2/9-pDes-Gfp-mAnt1 vector (empty circles) or PBS (filled circles). *p = 0.013, **p = 0.006; C Male mice body weight measured in 3, 6, 9, and 12 months WT = Ant1^+/+ (black), Ant1^−/− (red), and Ant1^−/−+ND6^P25L (blue) mice injected with AAV2/9-pDes-Gfp-mAnt1 vector (empty circles, right) or PBS (filled circles, left). Echocardiography and heart/body weight ratio data were obtained by analyzing WT PBS   = 8, Ant1^−/− PBS   = 11, Ant1^−/−+ND6^P25L PBS   = 8, WT AAV-Ant1   = 13, Ant1^−/− AAV-Ant1  = 12, Ant1^−/−+ND6^P25L AAV-Ant1   = 11 male mice. D Echocardiograph analysis of 6-month-old WT = Ant1^+/+ (black dots), Ant1^−/− (red dots), and Ant1^−/−+ND6^P25L (blue dots) female mice transduced with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) (empty circles) or PBS (filled circles). Stroke Volume WT PBS vs Ant1^−/−+ND6^P25L PBS ***p = 0.0005; Ejection Fraction WT PBS vs Ant1^−/− PBS *p = 0.0105 and WT AAV-Ant1 vs mutant ***p < 0.0001, WT PBS vs Ant1^−/−+ND6^P25L PBS ****p < 0.0001 and WT AAV-Ant1 vs mutant ***p = 0.0007; Fractional Shortening WT PBS vs Ant1^−/− PBS **p = 0.0098 and WT AAV-Ant1 vs mutant ***p = 0.0002, WT PBS vs Ant1^−/−+ND6^P25L PBS ****p < 0.0001 and WT AAV-Ant1 vs mutant ***p = 0.0008; Cardiac Output WT PBS vs Ant1^−/−+ND6^P25L PBS **p = 0.0011 and WT AAV-Ant1 vs mutant *p = 0.0132; Left Ventricles mass WT vs Ant1^−/− ****p < 0.0001, WT vs Ant1^−/−+ND6^P25L ****p < 0.0001, Ant1^−/− PBS vs AAV-Ant1 **p = 0.0084. E Heart/body weight ratio measured in 3, 6, 9, and 12 months old Ant1^+/+ (WT) (black), Ant1^−/− (red), and Ant1^−/−+ND6^P25L (blue) female mice transduced with AAV-Ant1 (empty circles) or PBS (filled circles). Echocardiography and heart/body weight ratio data were obtained by analyzing WT PBS n = 10, Ant1^−/− PBS n = 6, Ant1^−/−+ND6^P25L PBS n = 5, WT AAV-Ant1 n = 10, Ant1^−/− AAV-Ant1 n = 11, Ant1^−/−+ND6^P25L AAV-Ant1 n = 9 mice. Each dot in the echocardiography data represents a mouse, as a biological replicate. Mutant mice were compared with WT mice, and AAV-Ant1-treated mice versus PBS-treated mice. Data are presented as Mean ± SEM.

Postnatal periventricular AAV-Ant1 transduction markedly improved the echocardiogram parameters of the age-matched Ant1^−/− and Ant1^−/−+ND6^P25L mouse hearts. The transduced Ant1^−/− and Ant1^−/−+ND6^P25L hearts had increased stroke volume, ejection fraction, fractional shortening, and cardiac output, and decreased left ventricular mass (Fig. 4A), and these beneficial effects were maintained for 12 months (Supplementary Table 1). By 12 months of age, the left ventricular mass of the untransduced Ant1^−/−+ND6^P25L hearts had increased dramatically (Supplementary Table 1) which was reflected in the heart/body weight ratio which increased linearly with age (Fig. 4B). AAV-Ant1 transduction of the Ant1^−/−+ND6^P25L hearts returned the 12-month heart mass to that of the transduced Ant1^−/− hearts in male mice (Fig. 4B). AAV-Ant1 stabilization of the Ant1^−/−+ND6^P25L heart/body weight was attributed to changes in heart size since the body weight of all mice in the study increased comparably (Fig. 4C).

The injection process proved safe since the buffer-injected hearts did not exhibit cardiac damage or body weight changes, and the efficacious benefit was due to Ant1 expression since AAV2/9-pDes-Gfp transduction had no apparent benefit (Supplementary Table 1).

Amelioration of the metabolic effects of Ant1 deficiency by AAV-Ant1 transduction

To determine the level of Ant1 gene expression following AAV-Ant1 transduction of the Ant1^−/− and Ant1^−/−+ND6^P25L mouse hearts, we performed bulk RNA-seq analysis on the hearts of Ant1^+/+ and non-transduced and AAV-Ant1 transduced Ant1^−/− and Ant1^−/−+ND6^P25L littermates at 8–9 months post-transduction. This revealed that AAV-Ant1 transduction of both Ant1^−/− and Ant1^−/−+ND6^P25L mice produced approximately 10% of the cardiac Ant1^+/+ Slc25A4 mRNA (Fig. 5A). Further transcriptome Gene Ontology (GO) pathway analysis revealed that the non-transduced Ant1^−/− mouse hearts manifested altered cardiac muscle differentiation, development, and contraction pathways (Fig. 5B) and that the Ant1^−/− and Ant1^−/−+ND6^P25L hearts experienced the downregulation of extracellular matrix (ECM) and upregulation of sarcomere gene mRNAs, indicating changes in all components of the contractile machinery (Fig. 5C, D). These deficiencies were partially mitigated by the restoration of 10% Ant1^+/+ of the cardiac Ant1 mRNA levels, showing that relatively little OXPHOS gene expression can have a major beneficial functional effect on cardiac function.

Fig. 5. Gene expression and proteomic analysis of AAV-ANT1 transduction of Ant1^+/+, Ant1^−/−, Ant1^−/−+ND6^P25L mouse hearts.

A The Ant1 mRNA level in WT = Ant1^+/+, Ant1^−/−, and Ant1^−/−+ND6^P25L heart samples injected with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) or PBS (n = 3, mean); B fold change (AAV-Ant1) (purple column) compared with PBS-injected mice (turquoise column); C and D Normalized Enrichment Score analysis of extracellular matrix (ECM) and sarcomere proteins expression in Ant1^−/− and Ant1^−/−+ND6^P25L heart samples injected with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) compared with PBS-injected mice (Ant1^−/−+AAV-Ant1 virus-injected, purple column; Ant1^−/− PBS injected, turquoise column; Ant1^−/−+ND6^P25L AAV-Ant1 virus injected, blue column; Ant1^−/−+ND6^P25L PBS injected, orange column); E and F ANT1 and ANT2 levels assessed by mass spectrometry in isolated hearts from WT = Ant1^+/+ and Ant1^−/− mice transduced with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) or non-transduced Ant1^−/−. E ANT1 single peptides (gray lines) and ANT1 protein (brown dash line), F, ANT2 single peptides (gray lines) and ANT2 protein (brown dash line); G–I Volcano plot of the mass spectrometry data obtained comparing Ant1^−/− versus Ant1^+/+ mice (G), Ant1^−/− AAV-Ant1 vector injected versus Ant1^−/− mice (H), Ant1 peptides were observed but fell below Log₂ fold change scale range shown), and Ant1^−/− AAV-Ant1 vector transduced versus Ant1^+/+ mice (I). Three animals per condition were evaluated for RNA-seq and mass spectrometry studies. Statistic methods used are: DESeq2 - negative binomial distribution and Wald test for significance, corrected for multiple testing with the Benjamini–Hochberg (BH) procedure; GO Enrichment - hypergeometric test and the BH procedure for multiple testing; GSEA - Kolmogorov–Smirnov-like statistic, significance assessed via permutation testing, and the standard cutoff is FDR < 0.25; mass spectrometry - Student’s t-test, two-tailed (p < 0.05).

These RNA-seq observations were complemented by mass spectrometry proteomic analysis performed on the hearts of 6-month-old Ant1^+/+ and Ant1^−/− mice and AAV-Ant1 transduced littermates. The inactivation of Ant1 had profound consequences on the cardiac proteome which were partially mitigated by AAV-Ant1 transduction (Supplementary Fig. 7). Tryptic peptides from ANT1 protein were absent in the Ant1^−/− mice but present in Ant1^−/− transduced hearts (Fig. 5E). ANT2 (Slc25a5) tryptic peptides were detected in all samples but were slightly downregulated in the Ant1^−/− hearts (Fig. 5F). Hence, ANT2 is not induced to compensate for ANT1 deficiency in the heart.

Volcano plot quantification of heart proteomes confirmed the absence of ANT1 protein (Slc25a4) in the Ant1^−/− hearts (Fig. 5G) and the restoration of ANT1 protein following AAV-Ant1 heart transduction, but at low levels (Fig. 5I).

Relative to Ant1^+/+ hearts, Ant1^−/− hearts (Fig. 5G) exhibited decreased levels of many mitochondrial complex I proteins (NDUFA6, NDUFA10, NDUFA2, NDUFB8, NDUFA9, NDUFB1, MT-ND5, MT-ND1, and MT-ND4), consistent with ANT1 deficiency being associated with the downregulation of complex I proteins¹⁶. Additional downregulated proteins in the Ant1^−/− hearts included developmental and myogenic proteins (PPP1CA/B, NYX, FHL2, and SERPINA1D) and immunoglobulin proteins (IGHG1 and IGKC). Upregulated proteins in Ant1^−/− hearts included proteins of fatty acid metabolism (ECHDC2 and ACSF2) and some mitochondria-related proteins (DGUOK, ANKRD29, AIFM, ACO2, AK2, and ATAD3), as well as collagen VI proteins (COL6A1, COL6A2, and COL6A3), though most of the ECM genes were downregulated (Fig. 5C and G).

AAV-Ant1 transduction of Ant1^−/− hearts resulted in a striking contraction in the variation of protein concentrations relative to Ant1^+/+ hearts (compare the Log₂ fold change in Fig. 5G versus Fig. 5H and I). AAV-Ant1 partially restored the levels of a variety of mitochondria and functionally related proteins (ATP5F1D, ANKRD29, CAND2, UQCRQ, H1.4, MT-ND5, HNRNPU, DOCK10, ATP5MG, ECHDC2, PSMD8, NYX, GLUD1, HSP90B1, GRIN2B, TMEM143, MYOZ2, ATAD3, AIFM1, LRPPRC, and others), including some complex I protein subunits such as MT-ND5. However, the induction of mitochondrial gene expression by AAV-Ant1 transduction was insufficient to fully reverse the Ant1^−/−-associated complex I deficiency¹⁶. Moreover, upregulation of the collagen VI (COL6A1, COL6A2, and COL6A3) proteins was reversed in the AAV-Ant1 transduced hearts (Fig. 5G versus H). We didn’t observe full rescue of age-related accumulation of somatic mtDNA deletion mutations in AAV-Ant1 transduced Ant1^−/− heart either¹⁶.

Finally, AAV-Ant1 transduced hearts exhibited no induction of innate immune genes or proteins, even in hearts already expressing normal levels of ANT1 (Supplementary Fig. 7). Hence, the current level of transgene expression is not associated with an adverse immune reaction to the therapy.

Mitigation of ventricular cardiomyocyte defects by AAV-Ant1 transduction

To further define mechanistically how different Ant1^−/− cardiac cell types respond following AAV-Ant1 transduction, the Ant1^+/+, Ant1^−/−, and AAV-Ant1 transduced Ant1^−/− (Ant1^−/− + AAV-Ant1) hearts were subjected to single-nucleus RNA sequencing (snRNA-seq). This was performed using hearts 1-month post AAV transduction to assess the early effects of Ant1 transduction on mitochondrial OXPHOS and cardiac structure and function, and to avoid secondary outcomes. snRNA-seq clearly annotated the various cardiac cell types by their distinct gene expression profiles, with all three groups of mouse hearts showing concordant cell-type identities and composition (Fig. 6A, Supplementary Fig. 8A, B). Because of its low expression, snRNA-seq could not capture the Ant1 gene expression in the cell nuclei of AAV-Ant1 transduced hearts. Rather, qRT-PCR analysis of the same heart samples confirmed the cardiac Ant1 transgene expression in these mice (Supplementary Fig. 8C).

Fig. 6. snRNA-seq analysis of AAV-ANT1 transduction of Ant1^−/− mouse hearts.

A Uniform Manifold Approximation Projection (UMAP) visualization of cellular distribution from pooled snRNA-seq data of WT = Ant1^+/+, Ant1^−/−, and Ant1^−/− AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) transduced mice with grouped cardiac tissues. Eight cell types were annotated: ventricular cardiomyocyte (vCM), endothelial cells (Endo), fibroblasts (Fb), macrophage, pericytes, lymphatic endothelial (LEC), atrial cardiomyocyte (aCM), and neuron-like cell (NLC); B Volcano plot showing differentially expressed genes in vCM of Ant1^−/− and WT = Ant1^+/+ mice. Numbers of upregulated (red) and downregulated (blue) genes in the Ant1^−/− mice heart are labeled. Plot was constructed for differentially expressed genes that showed a log₂ fold change of 0.25 or greater in their transcript levels; C Functional enrichment GO analysis of differential gene expression pathways of genes significantly downregulated in vCM of Ant1^−/− mice heart compare to WT = Ant1^+/+ mice; D Volcano plot showing differentially expressed genes in vCM of Ant1^−/− mice transduced with AAV2/9-pDes-Gfp-mAnt1 vector (AAV-Ant1) versus Ant1^−/− mice. Numbers of upregulated (red) and downregulated (blue) genes in the Ant1^−/− hearts transduced with AAV-Ant1; E Functional enrichment GO analysis of differential gene enrichment pathways of genes significantly upregulated in vCM of Ant1^−/− transduced with AAV-Ant1 mouse hearts compared to Ant1^−/− mice; F Violin plots showing representative genes significantly [P-adj < 0.05 and log₂(fold change) > 0.25] downregulated in vCM in Ant1^−/− mice hearts compared to WT = Ant1^+/+ and Ant1^−/− AAV-ANT1 transduced hearts. Differential gene expression tests for individual cell types were identified using the FindMarkers function in Seurat with the LR (likelihood ratio) test.

The impact of the Ant1^−/− mutation on different cardiac cell types was determined by comparing the cell-type-specific gene expression between Ant1^−/− and Ant1^+/+ hearts and between Ant1^−/−+AAV-Ant1 and Ant1^−/− hearts. Ant1 was highly expressed in all cardiac cell types, as was Ant2, though at a lower level than Ant1 (Supplementary Fig. 9A). Overall, Ant2 expression was comparable between Ant1^+/+ and Ant1^−/− hearts, confirming that its expression was not increased to compensate for the Ant1 deficiency (Supplementary Fig. 9B).

Changes in gene expression for the eight cardiac cell types revealed the biggest alterations in gene expression between Ant1^−/− and Ant1^+/+ hearts in ventricular cardiomyocytes (vCM), followed by fibroblasts (Fb), endothelial cells (Endo), pericytes (PC), and macrophages (Mac). Transduction of Ant1^−/− cells with AAV-Ant1 had a significant effect on the gene expression of these same cell types (Supplementary Fig. 10). Focusing on vCM, the expression of 395 genes were significantly downregulated, and 33 genes were upregulated in Ant1^−/− vCMs compared to Ant1^+/+ vCMs (Fig. 6B). Pathway analysis revealed that downregulated genes in Ant1^−/− vCMs are all related to mitochondrial bioenergetics and cardiac muscle function (Fig. 6C). These included mitochondrial OXPHOS genes such as ATP synthetase (complex V) Atp5b and Atp5o, and cardiac contractile genes Tpm1 and Myl3. Upregulated genes in Ant1^−/− vCMs included Nppa and Nppb, markers of cardiomyopathy (Fig. 6B).

The AAV-Ant1 transgene had a substantial impact on Ant1^−/− hearts, particularly in vCMs. Ninety-two genes were significantly upregulated in AAV-Ant1 transduced Ant1^−/− vCMs compared to Ant1^−/− vCMs (Fig. 6D, Supplementary Fig. 10). Strikingly, pathway analysis revealed that the upregulated genes in transduced Ant1^−/− vCMs are virtually all related to mitochondrial bioenergetics and cardiac muscle function (Fig. 6E), demonstrating that AAV-Ant1 was able to reverse the mitochondrial bioenergetic and cardiac contractile defects in Ant1^−/− vCMs at the transcriptome level. Importantly, the downregulation of 22 genes in the Ant1^−/− vCMs was reversed by AAV-Ant1 transgene (Supplementary Fig. 11). These include the master OXPHOS transcriptional coactivator peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α) (Pgargc1a)^41–43, the complex I N-module structural subunit Ndufa6⁴⁴, as well as the muscle function genes calcium/calmodulin-dependent protein kinase ID (Camk1d), cardiac Z-disc protein (Nexn), and α1-adrenergic receptor (Adra1a) in vCMs (Fig. 6F, Supplementary Fig. 11).

Significant gene expression changes in Ant1^−/− mice that were reversed by AAV-Ant1 transduction were also observed in other cardiac cell types, though to a lesser extent (Supplementary Fig. 10 and 12–15). For example, 141 genes were downregulated in Ant1^−/− fibroblasts and 17 downregulated genes were reversed by AAV-Ant1 transduction (Supplementary Fig. 12). Comparable degrees of change were seen for endothelial cells (Supplementary Fig. 13), pericytes (Supplementary Fig. 14), and macrophages (Supplementary Fig. 15). All cells lacking ANT1 showed induction of Nppa, and all, but macrophages, showed induction of skeletal muscle α-actin (Acta1). Since the Desmin promoter is expected to limit AAV9-Ant1 transduction to cardiomyocytes only, the fact that other cardiac cell types showed significant gene expression changes suggests cell-cell crosstalk that impacts non-cardiomyocytes. Such cell-cell crosstalk could be mediated by secreted signals or cell surface ligand-receptor signaling.

Discussion

We have demonstrated that in a mouse model of PMD with cardiomyopathy due to Ant1 deficiency, low-level AAV-mediated pericardial transduction of the normal Ant1 cDNA into the heart is both safe and efficacious in mitigating severe cardiomyopathy. While our ultimate objective is to produce a systemic therapy that mitigates PMD defects in both heart and muscle, it was possible that a therapy that was beneficial for the muscle might be less so for the heart and vice versa. Accordingly, we first demonstrated the safety and efficacy of muscle AAV-Ant1 transduction via intramuscular injection into the Ant1^−/− mouse⁴⁵. Here we present the safety and effectiveness of pericardial AAV-Ant1 injection in the Ant1^−/− and Ant1^−/−+ND6^P25L mouse hearts. We chose pericardial injection instead of systemic delivery to specifically evaluate whether AAV-Ant1 transduction is both effective and safe for treating PMD cardiomyopathy. Because myopathy and cardiomyopathy due to a frameshift null mutation in the SLC25A4 (ANT1) gene is an autosomal recessive disease, it is crucial to develop a therapy with broader tissue targeting, including both the heart and skeletal muscle, likely through systemic delivery, for greater therapeutic efficacy. Furthermore, we have favored AAV-gene therapy over base editing since base editing changes only one of the multiple possible mutation sites in a PMD gene, while AAV-transduction can mitigate the biochemical defect of any recessive loss-of-function mutation within the gene.

When we injected the AAV-Ant1 vector in the Ant1^−/− mouse skeletal muscle, we were able to generate ~20% of the muscle ANT1 protein, resulting in the restoration of approximately 40% of the muscle ATP level and the rapid removal of the massive accumulation of abnormal muscle mitochondria, without overt toxicity³⁴. The removal of the abnormal mitochondria was mediated by the restoration of the ANT1 requirement for successful PINK-Parkin pathway mitophagy⁴⁵.

Neonatal pericardial injection of AAV-Ant1 into Ant1^−/− mouse produced ~10% of the normal ANT1 protein in the heart. However, this was the product of transducing only ~20% of cardiomyocytes, indicating that AAV-Ant1 produces ~50% of normal ANT1 protein in individual cardiomyocytes. Given that the AAV-Ant1 transduced Ant1^+/+ cells would express this level of induced ANT1 plus the endogenous ANT1, this indicates that there is no overt toxicity to the expression of ANT1, even at high levels. Hence, AAV-Ant1 transduction is both efficacious and safe. Accordingly, we can move on to systemic (retro-orbital) AAV-Ant1 injection of the Ant1^−/− and Ant1^−/−+ND6^P25L mice¹⁶ in preparation for treating ANT1^−/− patients⁹.

For both Ant1^−/− muscle and heart, the partial restoration of Ant1 by AAV-Ant1 transduction had a disproportional beneficial effect. This sharp threshold effect is commonly seen in mitochondrial bioenergetic diseases³. For example, in Leber Hereditary Optic Neuropathy (LHON), the mtDNA MT-ND4 m.11778G>A mutation must be essentially pure mutant (homoplasmic) before the patient will undergo vision loss⁴⁶. In Mitochondrial Encephalopathy, Lactic Acidosis, and Stroke-like Episodes (MELAS), subtle changes in the percentage of the mutant MT-TL1 m.3242A>G mtDNA have substantial effects on phenotype⁴⁷. Thus, clinically transformative therapeutic intervention can be achieved by transduction of only a small proportion of affected cells, minimizing the potential of a toxic response.

In our model system, AAV-Ant1 transduction was able to reverse the downregulation of OXPHOS and cardiac gene expression in Ant1^−/− hearts, ventricular cardiomyocytes, and other cardiac cells in both the milder Ant1^−/− mice and the more severe Ant1^−/−+ND6^P25L mice. Thus, an AAV-mediated ANT1 transduction of the heart might save the lives of the ANT1^−/−+mtDNA^U2 children with fulminant dilated cardiomyopathy^9,16.

ANT1 deficiency is associated with reduced expression of OXPHOS, including a 50% reduction in cardiac complex I protein levels¹⁶, which we confirmed by mass spectrometry analysis (Fig. 5G). While AAV-Ant1 transduction did not restore full complex I activity, it did upregulate critical complex I genes, including Ndufa6 and MT-ND5, as well as the master mitochondrial biogenesis regulatory gene, Pgc-1α. Thus, large phenotypic effects can be elicited by subtle changes in OXPHOS function.

Ant1 and Ant2 transcripts were observed in every cell type of the heart, suggesting that the reason Ant1 deficiency is not lethal is the partial compensation by Ant2. Yet in the Ant1-null mice, Ant2 was not induced, perhaps due to differences in transcriptional regulation of the two isoforms²². While Ant1^−/− mouse skeletal muscle experiences a massive accumulation of abnormal mitochondria⁴⁵, the accumulation of abnormal mitochondria is not apparent in the Ant1^−/− mouse heart¹⁶, presumably because the heart also expresses Ant2, which is able to sustain PINK-Parkin-mediated mitophagy³⁴.

Inactivation of Ant1 in the heart resulted in the downregulation of the ECM genes. We have observed that OXPHOS deficiency is linked to altered ECM gene expression in a human cell model of mtDNA diseases⁴⁸. By contrast, collagen VI genes were upregulated in the Ant1^−/− heart. The genetic inactivation of COL6A1 causes Bethlem myopathy and Ullrich Congenital Myopathy and predisposes the skeletal muscle to activation of the mtPTP and apoptosis^49–51. The upregulation of the collagen VI genes in the Ant1^−/− mouse heart may then be a compensatory response to stabilize the mtPTP and inhibit cardiomyocyte apoptosis to mitigate fibrosis and preserve cardiac function⁵².

Our model system demonstrates the feasibility of AAV-mediated mitochondrial gene therapy to treat life-threatening PMD cardiomyopathy. The neonatal pericardial injection of Desmin promoter-driven Ant1 has proven safe and efficacious in restoring cardiac function of Ant1^−/− mice, even when only 10% of the normal ANT1 level is produced. Since we have shown that AAV-Ant1 transduction of Ant1^−/− muscle resulting in ~20% production of muscle ANT1 can restore near normal muscle histology without pathology⁴⁵, we can conclude that AAV-Ant1 transduction should be both safe and efficacious for the systemic treatment of combined myopathy and cardiomyopathy of both the mouse Ant1^−/− and Ant1^−/−+ND6^P25L models¹⁶ and potentially for human ANT1^−/−+mtDNA^H and ANT1^−/−+mtDNA^U2 myopathy and cardiomyopathy⁹. Thus, AAV-mediated gene therapy is a promising approach to saving the lives of individuals suffering from mitochondrial myopathy and cardiomyopathy.

Methods

Reagents and resource sharing

Further information and requests for resources and reagents should be directed to Douglas C. Wallace (wallaced1@chop.edu)

Experimental model - mouse genetics

All mouse studies are approved by CHOP IACUC. Mice were maintained on the C57Bl/6JEiJ background, which was derived from the JAX C56Bl/6J line into which a functional nicotinamide nucleotide transhydrogenase (Nnt) gene was reintroduced⁵³. Mice without specific nDNA or mtDNA mutations were designated as WT (Ant1^+/+). These mice were compared to mice harboring the nDNA Ant1 (Slc25a4)-null mutation gene (Ant1^−/− mice). The phenotype of these mice was previously reported on the C56Bl/6J Nnt^−/− strain^12,13. We transferred the Ant1^−/− mutation into the C57Bl/6JEiJ strain (Nnt^+/+)^54,55 and the C57Bl/6JEiJ Ant1^−/− males were crossed with female C57Bl/6JEiJ ND6^P25L mice to generate Ant1^−/−+ND6^P25L mice¹⁴. Our Ant1^+/+ and Ant1^−/− mouse mtDNAs differ from the published C57BL/6J mtDNA⁵⁶ by the addition of a mtDNA MT-ND5 m.12352C>T (ND5^S204F) missense mutation. The phenotype of these mice has been described¹⁶. The Ant1^+/+ and Ant1^−/− mice were maintained by brother-sister mating. The Ant1^−/−+ND6^P25L mutant mice were maintained by crossing the Ant1^−/−+ND6^P25L females with Ant1^−/− males. All mouse experimental procedures were approved by the Children’s Hospital of Philadelphia Institutional Animal Care and Use Committee (IACUC, IAC 25-000925). Mice were pooled at weaning to attain 3 to 5, genotype-, sex-, and age-matched mice per cage with ad libitum access to food (5LOD diet from PicoLab) and water, on a 12:12 h light-dark cycle. The room temperature range is 68–74 degrees Fahrenheit with a relative humidity of 30–70%. A minimum of 3 subjects per strain was used for all experiments. Additional animals were included, as noted, based on the availability of age- and strain-matched mice and the nature of the assay¹³. Subjects were randomly numbered during sample preparation to blind the experimenter for subsequent tests and analysis. Besides cage changes and daily health checks, these mice were left undisturbed. When mice were determined moribund by the veterinary staff, they were humanely euthanized, and the date of death was recorded.

Study Design and Pericardial Injections

All measurements were taken from distinct biological samples. The number of biological samples is labeled in the figure legends. Three-day-old female and male mice were pericardially injected with AAV2/9-pDes-Gfp-mAnt1 or AAV2/9-pDes-Gfp vectors at a dose of 2.5 × 10¹¹ vg per mouse, diluted in PBS to a final volume of 20 μl. Control littermates were injected with PBS. Pericardial injection was performed without using ultrasound guidance^57,58. Briefly, three-day-old pups were anesthetized by cooling on paper on top of ice, then carefully injected with the virus or PBS using a Hamilton Gastight 1710 syringe with the needle entering right beneath the edge of the sternum. The needle was inserted in the trans-diaphragm direction towards the heart at a ∼30° angle to the handling table surface, and through the chest wall on the left border of the sternum at the axillary level⁵⁷. A plastic tube was put on the needle such that only ~2–3 mm of the needle enters the chest. The pups were allowed to recover from anesthesia and monitored at 1-, 3-, 6-, 9-, and 12-month post-injection.

Production of adeno-associated virus (AAV)

The Gfp reporter cDNA and mouse Ant1 cDNA were cloned into a bicistronic vector separated by the 2A translational interruption sequence^39,40, transcribed from the mouse Desmin promoter. The plasmid was produced by VectorBuilder Inc., USA (https://en.vectorbuilder.com/) and sequenced to confirm the correct insertion of Ant1. The AAV (serotype 2/9) plasmids were packaged at the Children’s Hospital Vector Core with titers from 1.5 × 10¹³ vg/ml to 4.9 × 10¹³ vg/ml.

Western blotting

Mice were sacrificed by cervical dislocation, and heart tissues for western blot analysis were isolated, rinsed, and homogenized in ice-cold medium (isolation buffer) containing 250 mM sucrose, 10 mM HEPES, 1 mM EGTA, and 0.1% bovine serum albumin (BSA) (pH adjusted to 7.2 with Trizma® base) using a glass homogenizer with six slow strokes of a Teflon pestle rotating at 600 rotations per minute. The homogenate was centrifuged at 700 × g for 5 to 10 min to pellet nuclei and unbroken cells. The resulting supernatant was centrifuged at 8500 × g for 10 min to isolate mitochondria. The final pellet was resuspended in the isolation buffer devoid of BSA, and protein concentrations were quantified using the Bradford method. Lysates containing 20 µg of protein were subjected to electrophoresis on 4–12% Bis-Tris SDS-polyacrylamide NuPAGE™ gels (Invitrogen). The separated proteins were transferred onto a nitrocellulose membrane using the iBlot Gel Transfer System (Invitrogen). Following this, the membranes were blocked for one hour in a 5% nonfat milk solution in TBST buffer (150 mM NaCl, 25 mM Tris, pH 7.5, 0.1% Tween® 20). The membranes were incubated overnight at 4 °C with shaking in primary isoform-specific antibodies (anti-Ant1 or Ant2)¹², diluted to 1:1000 in TBS. Following three washes with TBST, the membranes were exposed to secondary peroxidase-labeled antibodies for 2 h at room temperature. Finally, immune-reactive bands were visualized using an Enhanced Chemiluminescence reagent and detected via autoradiography. To estimate the relative level of ANT1 protein generated by transduction (Supplementary Fig. 2), the membranes were incubated overnight at 4 °C, shaking in primary isoform-specific antibodies (anti-Ant1)¹², diluted to 1:500 in TBS. For porin detection, anti-VDAC1/Porin primary antibody (Abcam, ab14734; 1:5000) was used. Following three washes with TBST, the membranes were exposed to secondary antibody for 2 h at room temperature. The relative level of protein fluorescence was quantified using the Licor system.

RNA isolation, PCR, RNA library, and sequencing

For RNA isolation and gene expression analyses^58,59, mice were sacrificed by cervical dislocation, and tissues for RNA sequencing analysis were isolated and flash-frozen in liquid nitrogen. Then, total RNA was extracted in TRIzol™ Reagent (Thermo Fisher Scientific, 15596026). Purified RNA was assessed by NanoDrop spectrophotometer (Thermo Fisher Scientific) for a 260/280 ratio and quantified by Qubit Flex (Thermo Fisher Scientific) using the Qubit RNA HS Assay Kit (Thermo Fisher Scientific, Q32855).

PCR Protocol: cDNA synthesis was generated from total RNA using the SuperScript IV Reverse Transcription kit (Thermo Fisher Scientific, 18090010). PCR amplification was conducted on a PCR Thermocycler (Eppendorf Mastercycler) with specific primers: EGFP-N-Reverse (5′- CGTCGCCGTCCAGCTCGACCA) with an annealing temperature of 66 °C and EGFP-C-Forward (5′ – CATGGTCCTGCTGGAGTTCGTG) with an annealing temperature of 61 °C. The PCR cycling conditions comprised initial activation at 50 °C for 2 min, polymerase activation at 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing and extension at 61 °C for 30 s. The resulting cDNA was separated and visualized via DNA gel electrophoresis.

RNA sequencing: The RNA was diluted to 1.43 ng/μl and reverse transcribed into cDNA using the Ion Torrent™ NGS Reverse Transcription Kit (Cat. No. A45003). The RNA library was prepared using the Ion Torrent S5 Chef and Ion AmpliSeq Kit for Chef DL8 (Thermo Fisher Scientific, A29024) and sequenced using the Ion Torrent S5 Sequencer and Ion 540 Kit (Thermo Fisher Scientific, A30011). The raw data was normalized in R (version 4.2.2) using the “DESeq2” (version 1.38.3) package. Volcano plots were generated in R using the “EnhancedVolcano” (version 1.16.0) package. GO data was generated using “clusterProfiler” (version 4.6.2), “AnnotationDbi” (version 1.60.1), and “org.Mm.eg.db” (version 3.16.0) packages in R. GO plots were generated in GraphPad Prism 9.

Total DNA Isolation and cellular transgene levels determined by digital PCR

Total DNA was isolated from tissue samples collected from various organs using the Qiagen DNeasy Blood & Tissue Kit. Tissues were homogenized in lysis buffer and digested with Proteinase K, followed by DNA purification using silica-based spin columns. In addition to the organ-derived samples, C2C12, a non-transduced cell line, served as the negative control. Genomic DNA from the C2C12 cell line was extracted using the DNeasy Blood & Tissue Kit. DNA concentration and purity for all samples were measured using a Qubit fluorometer (Thermo Fisher Scientific), and samples were diluted to a uniform concentration prior to downstream analysis. To quantify the viral genomes in the transduced mouse tissues, Gfp in the viral vector genome and Gapdh in the cell nuclear genome were quantified by digital PCR (dPCR) performed using the QuantStudio™ Absolute Q™ dPCR System. Each 10 μL reaction mixture consisted of 2 μL of 5× Absolute Q Digital PCR Master Mix, 0.5 μL of FAM- or VIC-labeled Applied Biosystems TaqMan probe for eGFP and GAPDH, respectively. 1 μL of 0.05 and 0.5 ng of genomic DNA template for eGFP and GAPDH, respectively, with nuclease-free water added to reach the final volume. Reactions were loaded onto Absolute Q dPCR plates and placed into the Absolute Q instrument. Following amplification, the Absolute Q Analysis Software was used to obtain absolute quantification results for each target. Copy number per microliter (copies/μL) for GAPDH and eGFP was directly exported from the software output and analyzed in Microsoft Excel. Vector genomes per diploid genome equivalent were calculated as GFP/GAPDH x 2.

Mass spectrometry

Mice were sacrificed by cervical dislocation, and tissues for RNA sequencing analysis were isolated and flash-frozen in liquid nitrogen. Flash-frozen mouse hearts were resuspended in 100 mM ammonium bicarbonate and 8 M urea at a ratio of 1:5 g:mL. Hearts were homogenized using an Omni TH, centrifuged at 20,000 × g for 5 min at 4 °C, and then a protein assay using BC. Fifty µg of protein were diluted to 50 µL in 100 mM ammonium bicarbonate, and dithiothreitol (DTT) was added to a concentration of 10 mM DTT. Proteins were incubated in DTT at 37 °C for 45 min, and then cysteines were alkylated by adding 5.5 µL of 0.5 M iodoacetamide in the dark and incubating at RT. Ammonium bicarbonate (50 mM) was used to dilute samples to 150 µL, and proteins were digested by adding 2 µg of Sequencing Grade Modified Trypsin (Promega) and incubating overnight at 37 °C. Samples were acidified to 0.1% trifluoroacetic acid (TFA) and desalted using C18 stagetips: tips were conditioned using acetonitrile (ACN) and then 0.1% formic acid (FA). Samples were then loaded and washed with 0.1% FA and eluted in 0.1% FA in 60% ACN. Peptides were dried in a Savant SpeedVac and then resuspended in 20 µL of 0.1% TFA. A NanoDrop (Thermo Fisher Scientific) was used to measure absorbance at 280 nm to normalize injection volumes. Proteomics was performed using a Dionex UltiMate 3000 nanoLC and a Q Exactive HF (Thermo Fisher Scientific). The solvent system was comprised of buffer A (0.1% FA) and buffer B (0.1% FA in 80% ACN). A C18 trap column (Thermo Fisher Scientific) and an in-house packed C18 (Dr. Maisch, GMBH) analytical column were used. The analytical gradient was run at 300 nL/min with 5–25% buffer B over 90 min and then 25–45% buffer B over 30 min. Data independent acquisition (DIA) was used with 24 m/z MS2 windows. Data were searched with DIA-NN using a complete UniProt FASTA digest library-free search without heuristic protein inference and processed using MSstats. Mass spec data and additional experimental details are deposited in PRIDE.

Heart/body weight ratio and histology

Mice were sacrificed by cervical dislocation, and heart tissues were collected for histological analysis and rinsed, and fixed in 10% formalin-neutral buffer for 48 h and processed by Histowiz Inc. (https://home.histowiz.com/). Before proceeding with the fixation process, each mouse and the excised mouse heart were weighed, and the heart/body weight ratio was determined.

Evaluation of the number of transduced cells

To assess the number of GFP-positive cells, we analyzed heart sections from 3 Ant1^+/+ mice 1-month post AAV-Ant1 injection. Three sections from each heart representing different anatomical planes (coronal, anterior, and posterior) were analyzed, with 5 random fields analyzed per section for each of 3 hearts, a total of 5 × 3 × 3 = 45 random fields. We then performed Wheat Germ Agglutinin (WGA, conjugated with red fluorescence) staining to label cell plasma membranes to locate each cell, and GFP-positive cells were visible on the same section/field. We set the minimal area (size) threshold to exclude non-cardiomyocytes. We then counted the total number of cardiomyocytes and GFP-positive cardiomyocytes in the field and calculated the % of GFP-positive cells manually. A representative picture and quantification of %GFP-positive cells are represented in Fig. 2A, B. We found that, on average, 20% of the cardiomyocytes express GFP.

Evaluation of the time course of cardiac Gfp expression

To establish the time course of expression of the GFP transgene, we used immunofluorescence of GFP (antibody: Abcam #183734) to determine the extent of GFP-positive area relative to the total area of heart sections. Whole-slide images were captured using identical scanner settings. We used FIJI to quantify the images. Background brightness was stable at 240–244 (8-bit), and GFP-negative tissue appeared at 214–230 across all slides. Ventricular myocardium regions of interest were manually outlined (Fig. 2C, D), and pixels were classified as DAB-positive if saturation was ≥5 and brightness was ≤201. Multiple representative slides were analyzed to distinguish valid DAB signals from GFP-negative areas, thus minimizing false positives in controls. Automatic thresholding was avoided because it recalculates the cutoff based on each image’s histogram, causing the thresholds to shift with variations in stain amount and background, leading to inconsistent performance across sample types in our tests. Fixed thresholds were applied uniformly to all slides in batch mode, and the percentage of DAB-positive area was calculated for each region of interest. This allowed us to evaluate samples injected with AAV2/9-pDes-Gfp-mAnt1 for each genotype (WT = Ant1^+/+, Ant1^−/−, and Ant1^−/−+ND6^P25L) at each time point (1, 3, 6, 9, and 12 months). The analysis confirmed the stability of the transduction until 12 months after injection, at which point appreciable downregulation was observed.

Evaluation of cardiac pathology and inflammation

To detect and quantify any AAV-Ant1 transduction-induced pathology of the hearts, coronal longitudinal sections of paraffin-embedded tissues were sliced (samples were trisected to have three coronal views per sample) and stained for H&E, Trichrome, and TUNEL. At least two animals per genotype (Ant1^+/+, Ant1^−/−, and Ant1^−/−+ND6^P25L) were histologically evaluated at each time point (3, 6, 9, and 12 months) for each of the three genotypes. There was no sign of tissue damage or increased cell death (Fig. 3B and Supplementary Figs. 3–6). Approximately 30 mice were studied. IHC for GFP was performed using the Abcam # 183734 antibody. Heart sections from 3-month-old PBS-injected or AAV-Ant1 transduced mice (n = 12 for all treatments and genotypes) were analyzed and stained for cleaved caspase 3 (antibody: Cell Signaling #9661S) and IL-18 (antibody: Abcam #71495).

Echocardiography

Echocardiography was performed using a VEVO3100 and an MS550D transducer (Visual Sonics)⁶⁰. Mice were anesthetized with 2% isoflurane in 80% O₂, which was lowered to 1.5% during the procedure. The heart rate was targeted to 400-450 bpm for the imaging. The average anesthesia length was ∼10 min per mouse. First, B-mode images of the parasternal long-axis view were taken, followed by M-mode images, and by B- and M-mode of the parasternal short-axis view. For analysis, the LV trace function of VEVO LAB 3.2.6 (Visual Sonics) was used on the M-mode images of the parasternal short and long-axis views. Analysis of short and long views provided comparable results, and only short-view data have been included in the manuscript.

snRNA-seq and data analysis

snRNA-seq and data analysis^61,62 involved Isolated nuclei processed using the Chromium Single Cell 3ʹ Reagent V3.1 kit from 10X Genomics. A total of 8000 nuclei per sample were loaded into a Chip G. Barcoding, cDNA amplification, and library construction were performed according to the Chromium 3ʹ V3.1 protocol. Sequencing was performed on a DNBSEQ-T7 platform targeting 20,000 reads per nucleus. Raw data was aligned to the reference GRCm38 genome using the CellRanger V7.2.0 software (10X Genomics) according to the 10X Genomics’ instructions. The snRNA-seq data of each sample were processed using Seurat (v5). Nuclei were filtered for mitochondrial reads <10%, nFeature_RNA > 800, and nCount_RNA < 10,000, and three datasets were integrated using Harmony⁶³. The cluster UMAP, violin plots, feature plots, and volcano plots were generated^61,62. We used clusterProfiler⁶⁴ to perform GO pathway analysis.

Statistical analysis

The ultrasound data and the heart/body weight ratio were analyzed using Prism 6.0 (GraphPad, La Jolla, CA, USA), and samples were compared between two conditions at a time using the two-tailed t-test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

A.A. and D.C.W. designed the research; A.A., K.K., J.W.G., G.A.W., K.J., and Z.C. performed the experiments; V.C. and M.B. maintained the mouse colony; L.P. and P.P. contributed to methodology and reagents/analytic tools; A.A., J.W.G., G.A.W., K.J., and Z.C. analyzed the data; P.L. and R.S. performed snRNA-seq experiments and analyzed the results. A.A., L.P., and D.C.W. wrote the paper; A.A., K.K., P.L., J.W.G., K.J., A.O., D.M., L.P., P.P., and D.C.W. reviewed and edited the manuscript; D.C.W. and L.P. generated the funding.

Peer review

Peer review information

Nature Communications thanks Eddy Kizana, Nils-Göran Larsson, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The published article includes all datasets generated and analyzed during this study. The RNA-seq and snRNA-seq data are deposited in the Gene Expression Omnibus (GEO) under accession GSE280991. The proteomics datasets are deposited in PRIDE with accession number PXD058090. The AAV2/9-pDes-Gfp and AAV2/9-pDes-Gfp-mAnt1 construct sequences and maps are available in the Source data files. Source data are provided with this paper.

Competing interests

D.C.W. is a consultant for Pano Pharmaceuticals and Medical Excellence Capital. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-67134-4.