A critical appraisal of the tafazzin knockdown mouse model of Barth syndrome: what have we learned about pathogenesis and potential treatments?

Abstract

Pediatric heart failure remains poorly understood, distinct in many aspects from adult heart failure. Limited data point to roles of altered mitochondrial functioning and, in particular, changes in mitochondrial lipids, especially cardiolipin. Barth syndrome is a mitochondrial disorder caused by tafazzin mutations that lead to abnormal cardiolipin profiles. Patients are afflicted by cardiomyopathy, skeletal myopathy, neutropenia, and growth delay. A mouse model of Barth syndrome was developed a decade ago, which relies on a doxycycline-inducible short hairpin RNA to knock down expression of tafazzin mRNA (TAZKD). Our objective was to review published data from the TAZKD mouse to determine its contributions to our pathogenetic understanding of, and potential treatment strategies for, Barth syndrome. In regard to the clinical syndrome, the reported physiological, biochemical, and ultrastructural abnormalities of the mouse model mirror those in Barth patients. Using this model, the peroxisome proliferator-activated receptor pan-agonist bezafibrate has been suggested as potential therapy because it ameliorated the cardiomyopathy in TAZKD mice, while increasing mitochondrial biogenesis. A clinical trial is now underway to test bezafibrate in Barth syndrome patients. Thus the TAZKD mouse model of Barth syndrome has led to important insights into disease pathogenesis and therapeutic targets, which can potentially translate to pediatric heart failure.

INTRODUCTION

Pediatric heart failure is a poorly understood condition with significant morbidity and mortality. Pathophysiological etiologies of heart failure have been difficult to define, but there is significant overlap between pediatric heart failure, congenital heart disease, cardiomyopathy, rhythm abnormalities, and acquired conditions, such as myocarditis (19, 60, 67, 68, 91). Etiology and symptoms differ by age group, and cellular mechanisms of the young heart are unique from adult hearts, thus requiring a better understanding of the pathophysiology of heart failure in children (60, 100). Therapeutic options for pediatric heart failure remain poorly developed due to the challenges of extrapolating adult data to children (19, 66, 100).

Impaired mitochondrial function is an important contributor to the impaired cardiac function in heart failure (28, 55, 57). Alterations in mitochondrial membrane lipid composition may contribute to abnormal mitochondrial function (25). Cardiolipin (CL) is a major cardiac phospholipid found in the inner mitochondrial membrane with essential roles in respiratory chain, metabolite carriers, and mitochondrial metabolism (22). Loss of CL leads to the generation of excessive reactive oxygen species (ROS) as a result of inefficient electron transport chain complexes (20). CL content has been shown to be significantly altered in pediatric heart failure, indicating a role of CL biosynthesis and regulation in the pathogenesis of heart failure (14, 80). Most famously, alterations in CL biochemistry are implicated in the pathogenesis of Barth syndrome (BTHS) and Barth cardiomyopathy (25, 46). Barth syndrome (MIM 302060) is a rare X-linked mitochondrial disorder in which patients are afflicted by cardiomyopathy, skeletal myopathy, neutropenia, and growth delay. Mutations in a nuclear-encoded mitochondrial phospholipid transacylase (tafazzin) lead to CL deficiency and mitochondrial dysfunction. Tafazzin catalyzes CL remodeling reactions at the final stage of CL biosynthesis, producing “mature” CL, which is critical to the structural integrity and diverse functions of mitochondria (27, 34, 61, 72, 77). A mouse model of Barth syndrome was developed almost 10 yr ago. This model relies on a doxycycline-inducible short hairpin RNA (shRNA) to knock down expression of tafazzin mRNA (TAZKD).

The goals of any model include a deeper understanding of disease pathogenesis and translational insights into potential therapies. The purpose of this review is to summarize the published data from the TAZKD mouse to determine its contributions to our pathogenetic understanding of, and potential treatment strategies toward, Barth syndrome. We focused on the following questions: 1) Does the mouse mimic the clinical syndrome? 2) Has this model led to a deeper understanding of disease pathogenesis? 3) Have potential therapies been suggested by this model?

METHODS

We searched PubMed and Medline using the terms “mouse,” “tafazzin,” and “knockdown,” as well as “TAZKD.” We selected all published articles that used the TAZKD mouse model to generate data. Unpublished data, data from abstracts only, or data presented at conferences but not published were not included in this analysis. We also reviewed selected literature of clinical data on Barth syndrome patients.

RESULTS

TAZKD Model: Genetic Engineering, Knockdown Induction, and Differences in TAZKD Induction

Transgenic mice were originally generated at TaconicArtemis (Köln, Germany) under contract from the Barth Syndrome Foundation; these mice are now available through The Jackson Laboratory (stock no. 014648; Bar Harbor, ME). These TAZKD mice have been previously described in detail (1, 79) on the basis of a tight inducible shRNA expression system (74) (Fig. 1). TAZKD transgenic mice used in studies are the heterozygous offspring of heterozygous (male)-wild-type (female) C57Bl/6 crosses, as per TaconicArtemis’ protocol [Phoon et al. (58)]. Doxycycline was administered in drinking water or in chow to induce knockdown of tafazzin. An advantage of this system is that tafazzin knockdown can be “turned on” as desired.

Fig. 1.

A: inserted by recombinase-mediated cassette exchange, the F3-FRT cassette contains the tafazzin (taz)-specific short hairpin RNA (shRNA) H1-tet-On promoter, a tet repressor (tetR), and a neomycin selection gene. B: the tetO-shRNA promoter is inactivated by tetR until doxycycline (DOXY) is added to the system, inactivating tetR, and allowing transcription of taz-specific shRNA. The system is reversible. Different DOXY tafazzin knockdown (TAZKD) induction protocols are listed. Knockdown of tafazzin will lead to rapid decreases in taz mRNA, but the time course of decreases in tafazzin protein, changes in the cardiolipin (CL) profiles, changes in mitochondrial ultrastructure and function, and changes at the cellular, tissue, and organ levels have not been precisely determined. L4CL, tetralinoleoyl cardiolipin; MLCL, monolyocardiolipin; SA, Splicing acceptor signal. [Adapted from Phoon et al. (58). Used with permission.]

The initial characterization studies notably used three different doxycycline induction, taz-knockdown protocols (1, 58, 79) (Fig. 1). In the Phoon et al. (58) study, doxycycline was administered in the drinking water [2 mg/mL and sucrose 10% (for palatability)] to wild-type female mice on a continuous basis, to make sure doxycycline remained in the maternal system and the embryonic transgene would, therefore, be induced throughout gestation. This was the same induction protocol as in the original TaconicArtemis protocol (74). However, because the males carry the transgene as heterozygotes, mating pairs were fed regular water without doxycycline to reduce the possibility of male sterility, which is a known adverse effect in a Drosophila model of tafazzin deficiency (50) and a more recently developed tafazzin knockout (TAZKO) mouse model (62). Doxycycline was reintroduced into the drinking water (2 mg/mL and sucrose 10%) of the females on the day following overnight mating (day of plug). Of note, this doxycycline dosing differed from the first two published investigations on the TAZKD mouse model [625 mg/kg chow (1); 200 mg/kg chow (79)]. It was estimated in the Phoon study (58) that TAZKD mice ingested 3–10 times the doxycycline used in studies by Acehan et al. (1) and Soustek et al. (79), because mice ingest, on average, 15 g/100 g body wt in food and 15 mL/100 g body wt in water daily (https://www.lvma.org/mouse.html).

Administration of doxycycline induces the shRNA expression system of the TAZKD model to decrease tafazzin mRNA and protein levels (Fig. 1). Doxycycline dosage has been shown to influence the speed at which the knockdown achieves steady-state levels (74). Various studies have differed in the timing of doxycycline administration as opposed to starting administration at the start of gestation [in drinking water from 3 mo of age (44); 625 mg/kg in chow from 2 mo of age (90)]. Still other studies combined induction methods, starting with 625 mg/kg chow from the start of gestation and switching to drinking water [1 mg/mL (32); 0.5 mg/mL (71)] at 3 mo of age. Data show that taz knockdown is highly efficient, with >90–95% knockdown of mRNA and protein at 2 mo of age when induced prenatally (1, 79) in heart and to somewhat varying degrees in other organs, such as liver, skeletal muscle, and brain. In embryos, 80% knockdown of tafazzin mRNA could be achieved within 3 days (58).

As will be seen, the different doxycycline dosing regimens and taz-knockdown induction protocols are important in the interpretation of published data in this mouse model. We note that, while mRNA may be knocked down rapidly, tafazzin levels and/or CL may take additional time to decrease, particularly since the half-life of CL is quite long (48, 98, 102); however, the timeline required to achieve tafazzin protein deficiency (to very low, steady-state levels) has not been carefully determined due to the lack of a good antibody. Once CL profiles have been altered, it may take time for a cellular and tissue/organ phenotype to develop (Fig. 1). The kinetics of sufficient protein knockdown indicate that the results of the model are based on the minimum time that knockdown is induced, which limits conclusions from some studies that did not induce knockdown for at least 3 mo. This may partially explain the difference in results across investigations based on the dosage and timing of induction. Some studies may not have allowed enough time for tafazzin protein depletion or for significant CL abnormalities to occur and, therefore, did not observe the same abnormalities that have been described in other investigations of this model (43, 75, 90).

A recent study by Moullan et al. (54) issued a caution on the potential confounding effects of doxycycline on experimental results in biomedical research. In addition, the high concentration of sucrose in the drinking water for some TAZKD protocols may further confound the interpretation of the results.

Phenotype

Cardiac abnormalities.

In the TAZKD mouse, two distinct cardiac phenotypes created by two different doxycycline induction protocols have been reported: 1) perinatal lethality, due to severe diastolic dysfunction associated with left ventricular (LV) noncompaction (LVNC) (58), and 2) a mild, adult-onset dilated cardiomyopathy (DCM) (1, 79). With tafazzin knockdown induced by the higher dose doxycycline-in-drinking-water protocol, the prenatal TAZKD model demonstrated complete lethality by the neonatal period and significant loss of TAZKD embryos at embryonic days 12.5–14.5 (58). Echocardiography revealed diastolic cardiac dysfunction, but with preserved systolic function, in embryonic hearts. Abnormalities in mitochondrial ultrastructure and sarcomeric organization were seen, including vacuolated cristae, a reduction in total mitochondrial area density, a reduction in mitochondrial size, and lower cristae density. Also, newborn TAZKD mice demonstrated a disruption of the normal parallel orientation between mitochondria and sarcomeres. Overall, these abnormalities indicate abnormal mitochondria and are consistent with delayed cardiomyocyte differentiation. Histology of the embryonic and newborn TAZKD hearts demonstrate altered cellular proliferation in association with the myocardial hypertrabeculation noncompaction, suggesting abnormalities in myocardial formation and patterning (58).

In contrast, in TAZKD mice induced with the lower doses of doxycycline in chow, there has not been evidence of significant defects in cardiac function in 2-mo-old TAZ knockdown mice under basal conditions. Cardiac dysfunction was not reported to develop until 8 mo of age (1, 21), although a more recent report suggested cardiac dysfunction by 5 mo of age (88). The dramatic differences in the survival and cardiac phenotype have been attributed to the differences in doxycycline dosing (74). In an adrenergic stress model of the TAZKD mouse, induced by chronic isoproterenol infusion, the cardiac phenotype was rapidly exacerbated with DCM and LV systolic dysfunction observed by 4.5 mo of age, compared with 7 mo of age without β-adrenergic stress (32). Thus the cardiomyopathy induced by this doxycycline protocol becomes apparent by echocardiography between 5 and 8 mo of age.

The cardiac dysfunction observed in the TAZKD mouse includes LV dilation, loss of LV muscle mass, and decrease in LV ejection force (1, 79). In yet another TAZKD mouse study, hypertrophic cardiomyopathy (HCM) instead of DCM was seen (37). Cardiac muscle of the TAZKD model demonstrates disturbed sarcomere organization, mitochondrial proliferation, myofibrillar disarray, mitophagy, and mitochondria-associated membrane abnormalities (1, 79).

“Acquired” tafazzin deficiency can be studied as well, to circumvent any developmental abnormalities that might arise and to mimic adult-onset cardiac diseases. Kim and coworkers (43) induced tafazzin knockdown at weaning and studied 8- to 12-wk-old TAZKD mice; therefore, the period of doxycycline induction spanned 4–8 wk only. Total CL was reduced by 50%, with a dramatic, 25-fold increase in monolysocardiolipin (MLCL). The investigators isolated subsarcolemmal and interfibrillar cardiac mitochondria and identified significant mitochondrial abnormalities, including changes in mitochondrial ultrastructure and decomposition of respiratory supercomplexes (with a concomitant increase in supercomplex I), but oxidative phosphorylation (OXPHOS) was unchanged in TAZKD mice. The investigators speculated that OXPHOS was compensated by an increase in supercomplex I (43). Because altered CL profiles have been implicated in ischemia-reperfusion injury, Szczepanek and coworkers (90) investigated how acquired tafazzin deficiency would impact mitochondrial function and response to cardiac injury (Langendorff model: 25 min ischemia + 60 min reperfusion). The TAZKD mouse was induced postnatally, from 2 to 4 mo of age, such that tafazzin deficiency would be “acquired” in young adult mice. Although TAZKD hearts showed increased ROS and increased susceptibility to mitochondrial permeability transition pore opening at baseline, the rate of oxygen consumption was unchanged, as was the response (LV infarct size) to ischemia-reperfusion injury. It should be noted that TAZKD mice were induced with doxycycline for only 2 mo, as reflected by the only 40% decrease in CL.

Clearly, the TAZKD mouse exhibits a cardiomyopathy, as does the human Barth syndrome. It is not clear at this time why some investigators have reported a DCM while others have found an HCM in the TAZKD mouse. The prenatal lethality of the TAZKD mouse model (58) also mirrors the male fetal loss and stillbirth that results from the fetal cardiomyopathy associated with Barth syndrome (86). However, because cardiomyopathy typically presents very early in Barth patients (most often within the first year), the delayed presentation of cardiac defects in the TAZKD mouse model has led some to criticize the model’s representation of the human condition (43, 90). Since human Barth patients will exhibit stabilization of cardiac function at older ages, an alternative explanation is that the TAZKD mouse does not exhibit an infantile cardiomyopathy/heart failure phenotype. Endocardial fibroelastosis has not been described in the TAZKD mouse. Moreover, TAZKD mice have not demonstrated any significant conduction abnormalities or ventricular arrhythmias, in contrast to findings in human Barth patients (83) (summarized in Tables 1 and 2).

Table 1.

Cardiac abnormalities

	Human Barth Patients	TAZKD Mouse Model
	Diagnosed in young patients (<5 yr of age and infants):  Fetal cardiomyopathy and heart failure  Dilated cardiomyopathy (DCM)  Left ventricular noncompaction cardiomyopathy (LVNC)  Hypertrophic cardiomyopathy (HCM)	Diagnosed in young mice (prenatal or <2 mo of age):  Pre-/perinatal LVNC cardiomyopathy
Variable course and severity	Ventricular arrhythmia Endocardial fibroelastosis Sudden cardiac death	4 mo of age  No change in heart contractile recovery after ischemia-reperfusion Chronic isoproterenol treatment (age 2.5–3 mo) exacerbated the cardiac dysfunction at age 4.5 mo  LVFS↓ LVEF↓ IVSd↓ 6–7 mo of age  Hypertrophic cardiomyopathy 7–8 mo of age  Dilated cardiomyopathy  LVFS↓ LVEF↓  LV dilation
Cardiac response to exercise	Blunted contractile function Impaired cardiac reserve ↓Peak O2 consumption and work rate ↓PCr recovery Response to endurance training:  ↑Peak O2 consumption (modest)  No changes in mean heart rate, cardiac function  Improved quality of life scores, symptoms	↓Peak O2 consumption Response to endurance training:  Cardiac hypertrophy  ↓Complex I, III, and IV enzymatic activities at baseline, with ↑complex III activity with endurance training
Cardiomyocytes	Irregular sarcomere organization Abnormal mitochondrial structure	Abnormal sarcomere organization Abnormal mitochondrial ultrastructure
References	5, 8, 13, 16, 21, 26, 38, 56, 64, 81–83, 86, 99	1, 21, 32, 37, 58, 59, 71, 78, 79, 90

↑, Increase; ↓, decrease; IVSd, interventricular septum thickness at end-diastole; LVEF, left ventricular ejection fraction; LVFS, left ventricular fractional shortening; TAZKD, knock down expression of tafazzin mRNA.

Table 2.

Phenotype and pathophysiology

	Human Barth Patients	TAZKD Mouse Model
	Fetal loss	100% pre-/perinatal lethality
Skeletal	Delayed motor milestones Lipid storage myopathy Proximal myopathy Abnormal fatigability Exercise intolerance Impaired balance and motion reaction time ↓O2 utilization and extraction ↓Phosphocreatine/ATP ratio ↓PCr recovery rate Blunted fat oxidation	Soleus muscle weakness Decreased exercise capacity Impaired O2 utilization
Growth	Abnormal growth trajectories Constitutional growth delay with delayed bone age Failure to thrive Need for assistive feeding or nutritional support	Lagging growth Decreased body weight Lower fat and lean mass
Hematological	Variable neutropenia	Low neutrophil count
Neurological	Cognitive difficulties Psychological and sensory abnormalities	Significant memory deficiency
References	5, 8, 12, 16, 29, 36, 52, 63–65, 81, 82, 85–87	1, 17, 18, 33, 37, 58, 59, 79, 82

↓, Decrease; TAZKD, knock down expression of tafazzin mRNA.

Skeletal myopathy.

The TAZKD mouse also exhibits significant reduction in skeletal muscle function (79). There is an observed reduction in muscle strength, with significant decreased soleus muscle contractility and diminished exercise capacity in the mouse model (59, 79). The TAZKD mouse also exhibits impaired oxygen utilization at high workload, consistent with findings in human Barth patients (59, 82). These functional deficiencies are accompanied by ultrastructural abnormalities, including aggregation, disrupted cristae, size, and shape variations, in mitochondrial and muscle fiber (1).

The skeletal and mitochondrial myopathy associated with exercise intolerance present in the TAZKD mouse is representative of the human condition of Barth syndrome (29, 82) (Table 2).

Growth delay.

The initial characterization of the TAZKD mouse showed normal growth through the first 2 mo of age, but lagging growth by 8 mo (1). TAZKD mice typically weigh less than their wild-type littermates due to hypermetabolism (18). On average, TAZKD mice weigh 16% less than their wild-type littermates (33). This weight difference has been attributed to lower fat and lean mass in TAZKD mice due to elevated energy expenditure and activity levels (37). Growth delay appears similar to human BTHS (Table 2).

Additional characteristics.

In a single study, low neutrophil count was observed in 50% of TAZKD mice, although the data were not explicitly reported (79). However, the investigators noted that this result is complicated by the use of doxycycline, which exerts reductive effects on neutrophil counts. Significant memory deficiencies have also recently been identified in the TAZKD mouse (17). However, the TAZKD data are scant for a thorough comparison with human BTHS (Table 2).

Disease Pathogenesis

See Table 3.

Table 3.

Biochemistry

Human Barth Patients	TAZKD Mouse Model
Abnormal molecular composition of cardiolipin, phosphatidylcholine, and phosphatidylethanolamine	Taz mRNA↓ Taz protein↓ Differential expression of TMEM65
TLCL↓	TLCL↓
↑MLCL:CL ratio  >0.3 is threshold for diagnosis	MLCL/CL↑↑ Plasmalogen↓ OCR↓ Reduced components of mitochondrial respiratory chain (especially I, III, IV)
↓Cytochromes
Decreased mitochondrial respiratory chain complex activity in skeletal muscle, cultured fibroblasts, and lymphoblasts
Decrease in supercomplexes	Decrease in supercomplexes
Cardiac-specific loss of succinate dehydrogenase	Cardiac-specific loss of succinate dehydrogenase
3-Methylglutaconic aciduria ↓Arginine level	↓Mitochondria bound myoglobin
↓Prealbumin level
↓Serum cholesterol
ROS↑	ROS↑
Lactic acidemia	↓Fatty acid oxidation enzymes
Blunted fat oxidation and elevated glucose metabolism upon exercise	↑Cytosolic glycolysis Abnormal mitochondria  Mitophagy  Paracrystalline structures  Vacuolated cristae
Structural and functional abnormalities of mitochondria  Inner membrane adhesions  Disrupted cristae
Refs. 2, 3, 5–7, 9, 12, 15, 21, 24, 30, 39, 40, 47, 53, 64, 70, 73, 81, 93, 95, 97, 99, 101, 103	Refs. 1, 21, 31, 37, 42, 44, 58, 59, 71, 78, 79, 89, 90, 101

↑, Increase; ↓, decrease; CL, cardiolipin; MLCL; monolysocardiolipin; TAZKD, knock down expression of tafazzin mRNA; TLCL, tetralinoleoyl cardiolipin; TMEM65, transmembrane protein 65.

Phospholipid composition, profiles, and alterations.

On induction, the knock down of tafazzin has significant effects on the overall lipid composition of the mitochondrial membrane. The primary finding is that tafazzin knockdown causes tetralinoleoyl CL deficiency with MLCL accumulation, specifically in cardiac and skeletal muscle (1, 21, 90). Notably, different tissues respond differently to tafazzin deficiency, because they exhibit different CL profiles, with less tetralinoleoyl CL in organs such as kidney, liver, and brain. Thus levels of tetralinoleoyl CL in the liver, brain, and white (but not brown) adipose tissue of TAZKD mice are similar to wild-type control levels (17, 18, 21). However, MLCL increases in all studied tissues in the TAZKD mouse, thereby increasing the MLCL-to-CL ratio to varying degrees. The increased MLCL-to-CL ratio has been observed in both adult and embryonic mouse hearts (1, 58, 79). The mouse demonstrates accumulation of MLCL and dilysocardiolipin, altered distribution of acyl chains in choline and ethanolamine glycerophospholipids, and dysregulated generation of potent oxidized lipid metabolites critical for hemodynamic function (42). It has also been found that the TAZKD mouse exhibits decreased plasmalogen in the heart (44). Xu and coworkers (101) found that CL is tightly bound to proteins, while MLCL is not, in the TAZKD mouse tissues as well as human lymphoblasts.

Respiratory chain complexes.

The disruption of the composition of the inner mitochondrial membrane as a result of CL abnormalities destabilizes respiratory chain complexes by disrupting supercomplex assembly (21). Organization of both respiratory chain complexes and supercomplexes is disrupted in TAZKD mitochondria, affecting electron flux, proton gradient, and ATP synthesis (31). The disruption of supercomplex formation is also tissue specific and mirrors the abnormalities in CL profiles; therefore, while heart, skeletal muscle, and brown adipose tissue exhibit destabilization of supercomplexes (17, 18, 21, 31, 101), liver, brain, and white adipose tissue are not substantially affected (17, 18). Once again, it is apparent that heart (and skeletal) muscle is most severely affected. Seahorse extracellular flux analysis shows that oxygen consumption rate in TAZKD mitochondria is significantly reduced compared with controls (21). There is a decrease in the abundance of large protein assemblies in the TAZKD mouse, including respiratory supercomplexes, ATP synthase oligomers, and complexes containing the ADP-ATP carrier (101). TAZKD mouse hearts have reduced components of the mitochondrial respiratory chain (31). While some results have shown a reduction in complex III activity with complex IV activity not affected in cardiac mitochondria (21, 42, 59), another study found that there is a reduction in complex I, III, and IV activity in cardiac tissue, and that exercise training improves complex III activity, suggesting different metabolic abnormalities under different conditions of exercise (78). Interestingly, while respiratory chain complexes are significantly affected in TAZKD mouse mitochondria, the ATPase complex is not (21).

Metabolism.

Proteomic analysis of TAZKD mouse hearts demonstrated marked metabolic remodeling, including disruptions in fatty acid oxidation enzymes and reduced mitochondria-bound myoglobin, potentially disrupting intracellular oxygen delivery to the OXPHOS system (31). Maximal capacity of OXPHOS is lowered in cardiac mitochondria from the TAZKD mouse (37); consequently, CL-deficient cardiomyocytes have an increased reliance on cytosolic glycolysis (59). The mouse demonstrates a downregulation of genes related to O₂/CO₂ exchange and aerobic metabolism (71), and a shift in preference for glutamate (amino acid) has been found to stimulate oxidation over fatty acid metabolism (42, 71). Additionally, the formation of ROS, specifically hydrogen peroxide (H₂O₂), is significantly increased in the TAZKD mouse (21, 59, 90).

Mitochondrial ultrastructural abnormalities.

Adult TAZKD mouse muscle shows abnormal mitochondrial structures with signs of mitophagy in addition to paracrystalline and honeycomb-like structures that are associated with altered lipid and phospholipid metabolism (1). In embryos, abnormal mitochondrial ultrastructure includes disrupted sarcomeres, vacuolated cristae, reduction in total mitochondrial area density, reduction in mitochondrial size, and lower cristae density (58). Ventricular and atrial muscle of TAZKD hearts show evidence of significant mitochondrial deterioration and mitophagy (1). Mitochondria also have large central vacuoles and abnormalities in mitochondria-associated membrane size and morphology (1). The mouse exhibits increased mtDNA content in heart and skeletal muscle, in accordance with mitochondrial proliferation (1).

Potential Therapies

See Table 4.

Table 4.

Therapeutic interventions

	Human Barth Patients	TAZKD Mouse Model
Gene replacement	ModRNA (full-length human tafazzin cDNA) in Barth iPSC-derived cardiomyocytes AAV-mediated gene therapy tested in dermal fibroblasts	Cardiomyopathy reversible when DOX inducer is removed Des-TAZ vector gene therapy AAV9-TAZ gene therapy CL-Nanodisk as lipid replacement therapy
Pharmacological	Bezafibrate tested in lymphocytes and current clinical trials Resveratrol tested in lymphocytes Elamipretide clinical trials MitoTempo (antioxidant) tested in iPSC-derived cardiomyocytes Linoleic acid tested in fibroblasts and iPSC-derived cardiomyocytes	Bezafibrate ROS scavengers
Cardiovascular	Angiotensin converting enzyme inhibitors (ACEi), angiotensin receptor blockers (ARB), β-blockers Berlin EXCOR device Heart transplant Implantable cardioverter defibrillators
Skeletal myopathy	Exercise training	Endurance training
Neutropenia	Granulocyte colony stimulating factor (G-CSF)
References	13, 36, 65, 84, 85, 88, 92, 96, 99, 101	1, 32, 33, 71, 78, 88, 89, 92, 99

AAV, adeno-associated virus; CL, cardiolipin; DOX, doxycycline; iPSC, induced pluripotent stem cells; TAZ, tafazzin; TAZKD, knock down expression of tafazzin mRNA.

Based on the clinical and experimental data, potential therapeutic targets for Barth syndrome have included TAZ deficiency, altered CL profiles, and ROS, among others. As such, strategies such as gene therapy, CL protection, and antioxidants have been implemented in both the mouse model and clinical trials (Fig. 2). Unsurprisingly, the mouse “trials” have focused on the cardiac and muscular manifestations of the disease.

Fig. 2.

Gene replacement or gene editing therapy may be possible in the future, particularly with the refinement of CRISPR/cas9 approaches. Enzyme replacement therapy is another possible therapeutic approach. *Tested in TAZKD mouse model. †Tested in human Barth cells or in human clinical trials. AAV, adeno-associated virus; HF, heart failure; ND-CL, nanodisk-cardiolipin; OXPHOS, oxidative phosphorylation; PL, phospholipids; ROS, reactive oxygen species; TAZ, tafazzin.

Gene or lipid replacement therapy.

Conceptually, replacing the deficient enzyme of the abnormal lipid would “cure” Barth syndrome. Tafazzin is a nuclear-encoded enzyme that remodels CL to its mature, tetralinoleoyl form; abnormal tafazzin, therefore, leads to abnormal CL profiles. In this X-linked (Xq28) disorder, many different tafazzin splice variants and mutations have now been described. Pathogenic variants include missense, nonsense, splicing defects, full or partial deletions, and frame shifts (45). The human tafazzin gene produces four major mRNA splice variants, with two known to be functional: TAZ lacking exon 5 (Δ5) and full-length TAZ (45). Further evidence has suggested that the TAZ-Δ5 isoform is likely to be predominant and the most functional (49, 94, 104). Indeed, no disease-causing variant has been attributed to exon 5 to date (Barth Syndrome Foundation Human Tafazzin Gene Variants Database, accessed August 9, 2019: https://www.barthsyndrome.org/research/tafazzindatabase.html). A small percentage of patients may exhibit a milder clinical phenotype, correlating with milder biochemical abnormalities (11). However, no genotype-phenotype correlation has been demonstrated to date (38).

Indeed, withdrawal of doxycycline in the inducible TAZKD mouse can reverse tafazzin mRNA knockdown at least partially within 4 wk (1). Proof-of-concept studies of human Barth cell lines show that gene replacement therapy has great potential to mitigate or cure the disease.

Adeno-associated virus (AAV)-mediated gene replacement therapy has been proposed as a mechanism to compensate for the native dysfunctional tafazzin gene and restore mitochondrial and cardioskeletal function (88). The AAV vectors package the genetic promoter and gene of interest to provide stable gene transfer. Studies in the TAZKD model using AAV9-TAZ gene therapy have shown promise for clinical translation in the future (89). Suzuki-Hatano and coworkers (88, 89) reported both phenotypic rescue and normalization of proteomics profiles in TAZKD mice treated with AAV-mediated gene replacement. Phenotypic and biochemical rescue included several parameters of activity and exercise, cardiac function, mitochondrial ultrastructural changes, and mitochondrial functioning (88); adults seemed to respond very similarly to neonates. However, it should be noted that TAZKD mice were injected with AAV vectors as neonates and as 3-mo-old adult mice, which would be before the manifestations of any disease phenotype. Therefore, we believe what this group showed was prevention (not necessarily treatment) of Barth syndrome symptoms, biochemical abnormalities, and proteomic abnormalities in the TAZKD mouse model.

Nanodisk (ND) delivery particles encasing exogenous CL have also been attempted in the TAZKD mouse, based on in vitro cell lines studies (33). However, the 10-wk administration of ND-CL failed to induce any improvement in the MLCL-to-CL ratio. Therefore, AAV-mediated gene replacement to restore normal tafazzin (and therefore CL) appears to be a more viable approach.

Pharmacological therapies targeting consequences of gene mutation.

Peroxisome proliferator-activated receptors (PPARs) are members of the nuclear hormone receptor superfamily, act as ligand-activated transcription factors, play a central role in energy metabolism and mitochondrial bioenergetics, and have been studied as a target for Taz associated cardiac dysfunction. Bezafibrate is a pan-PPAR agonist (41). The rationale to study bezafibrate in Barth syndrome stemmed from 1) known effects on mitochondrial biogenesis (4); 2) effects on fatty acid metabolism in a disease in which fatty acid metabolism is dysregulated (12, 70); 3) a clinical trial of bezafibrate in human patients with a different mitochondrial disorder, CPT2 deficiency (10); and 4) salient effects on human Barth lymphocytes, including an improvement/amelioration of the MLCL-to-CL ratio (101). Treatment with the pan-PPAR agonist bezafibrate prevents development of LV dilation and LV systolic dysfunction in TAZKD mice (32, 71), and the PPAR pan-agonist bezafibrate ameliorated the cardiomyopathy in TAZKD mice, while increasing mitochondrial biogenesis (32). It has also been suggested that voluntary exercise potentiates the bezafibrate action in skeletal muscle (71), and that endurance training provides beneficial effects, such as increased mitochondrial content, decreased ROS production, and restoration of complex II activity in cardiac muscle (78). Surprisingly, the improvement in systolic function in TAZKD mice was accompanied by a decrease in CL in the heart, a finding attributed to the increased number of mitochondria with reduced content of normal CL. Nevertheless, it remains unclear why in vivo there was no improvement in the MLCL-to-CL ratio as had been seen in Barth lymphocytes; we speculate there may have been an improvement in β-oxidation and fatty acid metabolism.

ROS are another favored target for potential therapies. Therapies targeting ROS have been investigated in the TAZKD mouse by targeting catalase expression to mitochondria to relieve oxidative stress, by crossing the so-called MCAT mouse with the TAZKD mouse (37). Catalase reduces levels of H₂O₂ by converting this to water and oxygen. While this genetic approach improved oxidative stress, there were no significant effects on the development of cardiomyopathy in the TAZKD mouse. The results suggested a need for further investigation of ROS, both as a pathogenetic pathway and as a potential therapeutic target.

DISCUSSION

Altered profiles in the signature mitochondrial phospholipid CL have now been implicated in heart failure, including pediatric heart failure. Therefore, insights into the pathogenesis of Barth syndrome, which is the only disease in which CL is primarily affected, may offer insights into heart failure. Although many assume that limitations of energy (ATP) or bioenergetics arise from mitochondrial disorders such as Barth syndrome, especially in a highly metabolic organ such as the heart, the TAZKD mouse model has shed insights into mitochondrial functioning and pathogenesis of this rare disease that go beyond simple ATP.

Modern biomedical research relies on animal models of human disease to study disease pathogenesis and insights into potential targets for therapeutic intervention. For the past several years, the doxycycline-inducible, shRNA-mediated, tafazzin-knockdown mouse has represented the best mammalian model of human Barth syndrome. It is apparent from the review of the published data above that this mouse exhibits many of the physiological, cellular, and biochemical features of the human Barth syndrome (1, 42, 58, 59, 61, 79) and offers us some insights into several aspects of pediatric heart failure and cardiomyopathy, including prenatal-lethal cardiomyopathies. This mouse model has demonstrated, unsurprisingly, the importance of mitochondria and mitochondrial phospholipids in cardiac function, but there are also suggestions that normal mitochondrial functioning during development is essential for normal myocardial patterning, formation, and functioning. Altered bioenergetics does not seem to explain the entire picture. Nevertheless, the TAZKD model has not recapitulated the human disease perfectly. Moreover, biochemical and metabolic data have not been particularly insightful into the pathogenesis of the Barth cardiomyopathy, specifically, why the abnormal biochemistry, omics, and mitochondria lead to DCM, HCM, and/or LVNC.

Alterations in lipid metabolism are now well-known to coexist with, and contribute to, the pathogenesis of cardiac and skeletal muscle disorders (69). During heart failure, cardiac CL content is reduced (76) and the CL pool is altered such that a progressive decrease of mitochondrial function and energy metabolism is observed (22). Deficient CL biosynthesis and remodeling has been examined in both pediatric and adult heart failure patients and has demonstrated evidence of different mechanisms of disease between children and adults (14). Thus using an animal model with a similarly altered CL profile has the potential to lead to new therapeutic approaches for heart failure in children that differ from those used in adults.

Limitations of TAZKD Model

This is a knockdown—not a knockout—mouse model. Therefore, a small percentage of functional tafazzin is still present. Additionally, the inducible knockdown model requires an inducer, which may have confounding effects on the tissues. However, with the TAZKD model, the controls have also been fed doxycycline to account for potential confounders. Doxycycline itself may also be problematic, particularly in mitochondrial diseases, since mitochondria are bacteria that, eons ago, made their way into the cell in a symbiotic relationship. Lastly, the details of the animal model must be taken into account when analyzing results: in this case, there must be adequate time allowed for the knockdown of mRNA, protein, and finally CL, all of which precede disease manifestations or progression.

Future Directions

ROS is postulated by many to be an important contributing factor in the etiology and progression of many disease processes, including Barth syndrome (see above), but this view has been challenged in part by the failure of antioxidant therapy to consistently yield beneficial effects. The MCAT × TAZKD mouse crosses experiments outlined above provide one such example. Also, superoxide and H₂O₂ exert deleterious effects inside a cell by different mechanisms (35), so this may explain inconsistent results in “antioxidant” trials. The results indicate a need for further study in ROS and antioxidant therapy in relation to pediatric cardiomyopathies and heart failure. Downstream of metabolic changes, we really do not understand the link to cardiomyopathy and skeletal myopathy and other clinical features of Barth syndrome. Mouse models can be used to further analyze downstream consequences (and, therefore, potential therapeutic targets) of altered metabolism, including ROS.

Data based on the TAZKD model have shown that therapeutics using AAV vectors may be useful in the future for the subset of those pediatric cardiomyopathies with heart failure that are genetic in origin. It is interesting that trials have utilized the full-length human tafazzin cDNA, when it seems as if the Δ5 tafazzin isoform would be perhaps more appropriate. Similar considerations for the packaging of specific protein isoforms may need to be taken into account when applying AAV vector therapy to other pediatric genetic cardiomyopathies.

A new TAZKO mouse model has now been developed, and data are beginning to emerge on the pathogenesis of Barth syndrome, with an accumulation of unpublished data on its representativeness of the human condition. As such, an appraisal of the TAZKD mouse model will provide further context for the utility of this new model going forward. The movement toward a knockout model to represent the human condition of Barth syndrome could present differences in phenotype compared with the knockdown due to possible compensatory pathways in the knockout model not present in the TAZKD mouse (23). The inducible TAZKD model also has its own advantages that allow for time-dependent modeling of different diseases and studying of reversibility, neither of which is possible with a knockout. Especially when considering gene therapy, which will occur late in the disease, it has to be proven that the damage caused by the disease is reversible; therefore, requiring a knockdown model over a knockout. Thus the findings from the TAZKD mouse phenotype are significant in our understanding of altered CL profiles due to tafazzin deficiency and potential therapeutic approaches.

Conclusions

Despite some variations from the clinical human syndrome, the TAZKD mouse model of Barth syndrome has led to important insights into disease pathogenesis, especially the Barth cardiomyopathy. Given heart failure in children is most often due to cardiomyopathies (51), new insights into pediatric heart failure can be gleaned from our understanding of the TAZKD mammalian model that exhibits a primary cardiomyopathy. Such insights may bring us closer to identifying further therapeutic targets, including CL manipulation, which could translate to therapies for pediatric heart failure.

GRANTS

This study was supported by grants from the Barth Syndrome Foundation (to C. K. L. Phoon), a Dean’s Undergraduate Research Fund (to P. C. Miller), and the National Institutes of Health (to M. Schlame).

DISCLOSURES

M. Ren and M. Schlame are on the Scientific and Medical Advisory Board of the Barth Syndrome Foundation.

AUTHOR CONTRIBUTIONS

M.R. and C.K.P. conceived and designed research; M.R., P.C.M., M.S., and C.K.P. analyzed data; C.K.P. interpreted results of experiments; C.K.P. prepared figures; M.R., P.C.M., M.S., and C.K.P. drafted manuscript; M.R., P.C.M., M.S., and C.K.P. edited and revised manuscript; M.R., P.C.M., M.S., and C.K.P. approved final version of manuscript.