Hypertrophic cardiomyopathy (HCM) is a prevalent cardiovascular disease characterized by thickening of the left ventricular wall in the absence of abnormal loading conditions [1
]. HCM progresses through a compensatory response relying on changes in gene expression, bioenergetics and cellular morphology of cardiomyocytes [2
], in order to maintain the cardiac function required to meet the metabolic demands of the body [3
]. HCM presents wide clinical variability, with patient outcomes ranging from asymptomatic to sudden cardiac death, when the sustained compensatory response leads to lethal energetic and functional imbalance [4
]. This variation is likely due to the highly complex genetic causation characteristic of this condition, displaying variable penetrance and expressivity [5
]. While approximately half of HCM patients display mutations in one or more of the >20 sarcomeric genes that regulate cardiomyocyte contraction, others do not exhibit any mutation in known heart-related genes [6
]. Disease manifestation is mutation-specific, with patients with different mutations in the same gene showing contrasting clinical outcomes [7
]. Even monozygotic twins bearing the same sarcomeric mutation displayed different clinical presentations [8
], implying that factors beyond the single pathological change (e.g., genetic and epigenetic background and environmental modifiers) contribute to the magnitude of the disease effects [9
]. This intricate clinical and genetic complexity of HCM has impaired the development of efficient therapeutics, with heart transplantation being the only long-term solution [10
While numerous hallmarks of HCM have been identified [2
], disrupted mitochondrial bioenergetics is commonly associated with heart failure, whereby the inefficient ATP usage in the heart leads to energy depletion [4
]. Despite the fact that the majority of the proteins present in the mitochondria are encoded by nuclear genes, all the polypeptides, ribosomal RNAs and tRNAs encoded by mitochondrial DNA (mtDNA) constitute essential subunit complexes of the oxidative phosphorylation (OXPHOS) system [12
]. The heart is a highly energetically-demanding organ that sources 90% of the ATP from mitochondrial OXPHOS [13
]. Consistently, cardiomyocytes have higher mtDNA copy number per diploid nuclear genome, resulting in increased mitochondrial content [14
]. Thus, mitochondrial diseases impact preferentially in the heart [15
], mostly due to pathogenic mtDNA mutations [16
] or reduction in mtDNA content [17
]. Importantly, mitochondrial cardiomyopathies (MIC) constitute phenocopies of HCM, particularly in cases where the genes encoding mtRNAs are mutated [18
]. Altogether, we hypothesize that mtDNA is a likely candidate for harboring gene modifiers of HCM, and it remains unexplored.
We previously developed and characterized parallel isogenic human pluripotent stem cell-derived cardiomyocyte (hPSC-CM) models of HCM, either displaying the p.R453C-β-myosin heavy chain (β-MHC) mutation [19
], or the p.E99K-ACTC1 pathological change [20
]. These two sets have revealed overall similarities and differences in HCM hallmarks and highlighted variations in functional phenotypes and gene expression profiles between the different cell lines carrying the same mutation. Remarkably, both sarcomeric-mutant models exhibited increased mitochondrial respiration rates relative to isogenic healthy controls, characteristic of the compensatory stage of HCM that precedes energy depletion [21
]. Thus, we utilized these sets of hPSC-CMs to investigate the role of mtDNA to harbor potential gene modifiers that could underlie the different HCM phenotypes observed. The identification of novel HCM gene modifiers could unveil new molecular targets for future treatment and support genetic screening of patients, to better inform clinicians on the most efficient therapy.
To achieve a deeper understanding of pathological mechanisms of HCM, several models of disease have been developed, ranging from preparations of scarcely available human heart tissue biopsies to whole animals [39
]. Recent technological advances have resulted in the application of hPSC-CMs for disease modeling, advantageous relative to previous models due to the ability to provide an unlimited source of cardiomyocytes and recapture patient genetic makeup [40
]. Genome editing technologies such as CRISPR/Cas9 have facilitated the generation of high-fidelity HCM models, surpassing the inherent genetic and epigenetic baseline variation between unrelated hPSC-CM lines [41
]. Isogenic sets of hPSC-CM models have therefore addressed the complex genetic etiology characteristic of HCM and corroborated clinical observations, highlighting that disease progression is mutation-specific [21
] and dependent on factors beyond the primary mutation [20
Early approaches aiming to establish genotype-phenotype relationships in HCM relied heavily on high numbers of patients in order to associate genetic variants with disease severity, due to the existence of confounding factors such as age and donor variability [43
]. Remarkably, the use of multiple isogenic cell lines sharing the same sarcomeric mutation in different genetic backgrounds has complemented patient cohort studies by facilitating the determination of the penetrance and expressivity associated with that variant. Importantly, the controlled genetic background of this approach overcomes the need for high numbers of patient cohorts and enables the identification of novel gene modifiers of HCM that may constitute new therapeutic targets. We have contributed to this goal by generating and characterizing two models of HCM bearing different sarcomeric mutations (p.R453C-βMHC or p.E99K-ACTC1) in six unrelated healthy hPSC cell lines [19
]. These two models exhibited the main functional hallmarks of HCM (hypertrophy, energy depletion, arrhythmias and changes in contractility) and showed variation in the magnitude of in vitro cellular phenotypes observed between isogenic sets sharing the same mutation. Due to the fact that energetic imbalances in the heart trigger heart failure in HCM patients [4
] and that genetic abnormalities in mtDNA can recapitulate cardiomyopathy phenotypes [16
], we investigated mtDNA as a likely candidate for harboring gene modifiers of HCM.
While sequencing mtDNA of a large cohort of patients would have been advantageous to provide a higher degree of patient-wide validation, the alternative use of isogenic sets of hPSC-CM lines enables a more refined investigation of the contribution of mtDNA mutations to HCM, a disease characterized by highly complex mutation-specific effects [21
]. Additionally, the customary investigation of patient peripheral blood as an alternative to scarcely available cardiac tissue biopsies is not optimal due to tissue-specific mtDNA sequence heteroplasmy [44
] (which is recapitulated in hPSC-CMs). Notwithstanding, the use of cell lines instead of patient DNA for mitochondrial genetics studies also presents some drawbacks such as unintended mtDNA mutagenesis and changes in heteroplasmy that may arise during reprogramming, long-term culture, and cardiac differentiation [26
]. Nevertheless, these technical artefacts were minimized by: (i) applying the same culture conditions across all the cell lines under study; (ii) directly comparing isogenic cell lines where the genetic background is preserved (impossible in patient samples); and mostly by (iii) sequencing all the lineages from patient-derived fibroblasts, through hPSCs and hPSC-CMs to identify HCM-unspecific mtDNA mutations. Additionally, we extended our findings from hPSC-CMs to three unrelated patient families exhibiting the p.E99K-ACTC1 mutation, providing further evidence that NGS of cell lines may complement that of patient DNA.
NGS analysis of mtDNA derived from isogenic sets of healthy and HCM hPSC-CMs revealed 11 variants in the p.R453C-βMHC model and 10 in the p.E99K-ACTC1 set. However, in silico prediction combined with sequencing all the lineages from fibroblasts to hPSCs and hPSC-CMs resulted in the exclusion of four in the former model and two in the latter. By correlating previously reported functional disease phenotypes [19
] with variant heteroplasmy analyzed herein, we identified potential HCM aggravators and protective mtDNA mutations in both models. Typically, cells can withstand a high proportion of mutant:WT mtDNA ratios, but when that proportion exceeds the disease thresholding effect (specific for each variant), OXPHOS defects occur [45
Moreover, some mtDNA mutations were complete, totally abolishing WT copies. While some of such variants were specific to each model (such as m.12715A > G as potentially cardioprotective), one was shared between them (m.152T > C) as a highly likely disease aggravator. Conversely, four variants showed opposing potential effects in the two models, reinforcing the notion that HCM exhibits complex genetic causation. We speculate that these variants reflect the contrasting changes in phenotype induced by the two primary sarcomeric mutations, e.g., while the p.R453C-βMHC change caused hypo-contractility, the p.E99K-ACTC1 mutation led to hyper-contractility [21
Importantly, some of the variants identified in hPSC-CMs were enriched in specific haplogroups, complementing patient cohort studies that highlighted the haplogroup H as a susceptibility factor due to its higher prevalence in HCM patients [38
]. As the identification of the specific mtDNA variants underlying these differences is lacking [46
], further validation of their role in HCM in hPSC-CMs may contribute to refining this association.
To extend the drawn correlations to a wider patient population, three unrelated families where p.E99K-ACTC1 (and consequently HCM) were common were investigated, unveiling a number of shared or family-exclusive mtDNA mutations that may underlie variation in clinical outcomes. However, when compared to the bioinformatics databases focused on human diseases caused by mtDNA mutations, none of the identified variants were overlapping. Moreover, MIC patients presenting cardiomyopathy phenotypes have not showed mutations in the non-coding control region of mtDNA [18
]. This implies that whilst mtDNA mutations are central to MIC and constitute the main cause of disease, they are not sufficient to induce HCM, unlike sarcomeric mutations. Nonetheless, they can exacerbate or improve clinical outcomes, as was illustrated in more severe hypertrophy phenotypes (~72% higher intraventricular septum thickness) in patients showing both sarcomeric (p.R249Q-βMHC) and mitochondrial (m.4300A > G) mutations relative to those bearing only one of them [47
]. The same applies to changes in mitochondrial content: while MIC patients showed striking phenotypes in terms of maladaptive mitochondrial proliferation (e.g., ~3-fold increase in mtDNA content per cell relative to healthy samples [48
]), no conclusive evidence has shown variations in mtDNA content in explanted failing hearts when compared to healthy counterparts [49
], in line with our data from hPSC-CMs.
To fully demonstrate their role in disease progression, gain- or loss-of-function studies should be performed, whereby the mtDNA mutations are corrected or introduced in hPSC-CMs. This was not done in the scope of this study because precise genomic correction of mtDNA mutations is technically very challenging and has not yet been achieved robustly and efficiently [50
]. CRISPR/Cas9 approaches rely on homologous recombination (HR) of double stranded breaks (DSBs) to introduce SNVs of interest [52
]. However, unlike in the nuclear DNA, DSBs in mammalian mtDNA cannot be efficiently repaired by HR [53
], precluding the use of CRISPR/Cas9 [54
]. mtDNA molecules bearing DSBs are rapidly degraded and the remaining intact molecules replicate to restore a steady mtDNA copy number, shifting the heteroplasmic ratio [55
]. This strategy has been explored by engineering restriction endonucleases directed to the mitochondria [56
] to degrade mutant mtDNA. While this approach may be partially feasible in the case of heteroplasmic mutations, it is ineffective for those where the total mtDNA is mutated (such as m.152T > C and m.16319G > A identified in this study). Alternatively, another possibility is the delivery of healthy plasmid DNA to the mitochondria. However, the efficient transport of nucleic acids into this organelle remains elusive, even when conjugated with mitochondriotropic delivery systems [58
]. Clearly, a technological breakthrough is required for efficient and precise genomic correction of mtDNA mutations in MIC and HCM patients.