Mitochondrial nucleoid in cardiac homeostasis: bidirectional signaling of mitochondria and nucleus in cardiac diseases

The mitochondrion is a lipid bilayer membrane organelle. It is highly enriched in adult cardiomyocytes which occupy a large portion (approximately 20–30%) of the cytoplasmic space and produce 90% of ATP [193]. Mitochondria contain an outer membrane and inner membrane, two aqueous sub-compartments known as the intermembrane space (IMS) between the two membranes, and the matrix within the inner membrane. The outer membrane is a double phospholipid membrane which is similar to the eukaryotic cell membrane. The outer membrane not only separates the mitochondrion from the cytoplasm, but also contains multiple receptors to mediate communication between mitochondria and other organelles. In addition, some special regions of the outer membrane have been shown to interact with other organelles or other mitochondria, including (i) mitochondria-associated endoplasmic reticulum membranes (MAMs) and (ii) intermitochondrial junctions (IMJs). The intermembrane space contains cytochrome c, which is involved in electron transport and apoptosis regulation. The inner membrane is the site of oxidative phosphorylation (OXPHOS) complexes (complexes I–V) which are involved in electron transport and ATP synthesis. The inner membrane also folds and creates a layered structure named cristae that protrude into the matrix to increases the surface area for energy production. The matrix is enriched in the enzymes and chemicals of the Krebs tricarboxylic acid (TCA) and fatty acid cycle, mitochondrial DNA (mtDNA), ribosomes, enzymes, and ions.

During heart development, the mitochondrial system is dynamically regulated to support ATP synthesis, energy transduction, and cell signaling events (Fig. 2). The increase in mitochondrial mass in the growing heart is accompanied by structural differentiation into interfibrillar (IFM) and subsarcolemmal mitochondria (SSM) [43]. These two mitochondrial subpopulations exhibit distinct location, morphology, and functional characteristics. Elongated IFMs are tightly packed into the space between sarcomere Z-lines and lined with sarcoplasmic reticulum, providing ATP for contraction and having higher calcium (Ca²⁺) accumulation. In contrast, the SSMs are located beneath the sarcolemma and provide ATP for active transport of electrolytes and metabolites across the sarcolemma [94, 154]. It has also been reported that the stress response of these two mitochondrial subpopulations are different under pathological conditions, such as cardiac volume [214] and pressure overload [194], hypoxia [87], diabetes mellitus [39, 40], and cardiomyopathy [96]. Metabolic changes also occur during the developing heart. Immature mitochondria can be observed in cardiomyocytes in mice from embryonic day 8.5 (E8.5) to E10.5, and some very immature mitochondria can be observed in cardiomyocytes, as characterized by low cristae density, absence of SSM/IFM specialization, and lactose metabolism. Although OXPHOS complexes can be detected in the inner membrane of mitochondria, cardiomyocytes at these stages primarily acquire ATP through anaerobic glycolysis [180]. Intriguingly, embryonic development will be terminated at E8.5 mouse models lacking essential mitochondrial components [28, 31, 105, 130] and all heart structure [28], suggesting that mitochondria also play a pivotal role in heart development. The number of cristae in cardiomyocytes is increased in mice from E11.5 to E13.5, and many tubular cristae are formed that are connected with the periphery. By E13.5, the functionally matured mitochondria are observed, and cardiomyocytes at this stage acquire ATP through both anaerobic glycolysis and OXPHOS [37, 146]. Noticeably, the mitochondrial permeability transition pore (mPTP) is closed at E13.5, which has been reported to drive mitochondrial elongation, higher mitochondrial membrane potential and cardiomyocyte differentiation in mouse hearts [95]. From E13.5 to prenatal stages, the ATP production pathway shifts from anaerobic glycolysis to aerobic metabolism [69, 146]. At the end of gestation, OXPHOS provides > 50% of ATP in prenatal hearts. Shortly after birth and during the first week of postnatal stages, mitochondrial biogenesis is increased, and the mitochondrial mass of cardiomyocytes is doubled [43, 75]. In addition, the energy metabolism shifts from glucose- and lactate-consuming respiration to the more energy producing β-oxidation of fatty acids [146, 180]. One reason for the change of energy metabolism from glycolysis to OXPHOS in developing heart may be the development of hypoxia inducible factor 1α (HIF-1α) signaling. In the embryonic and fetal (prenatal) heart, HIF-1α is activated in the presence of a hypoxic environment, whereas HIF-1α signaling is decreased in cardiomyocytes after birth [146, 147, 168]. These studies emphasize the importance of mitochondrial homeostasis as essential for embryonic heart development.

Although mitochondria play an important role in heart development, little is known about their function at the cardiac precursor stage as a result of technical limitations. With the development of efficient cardiac differentiation protocols, the process of cardiac differentiation from pluripotent stem cells (PSCs) is known to recapitulate early cardiac development, thereby providing a potential source for studying otherwise inaccessible cells or tissues. To maintain pluripotency, PSCs mainly depend on glucose and glutamine metabolism. When PSCs exit from pluripotency to the mesoderm, metabolism switches away from glycolysis to OXPHOS which is modulated by several key signaling pathways, including MYC/MYCN transcription inhibition, insulin inhibition, GSK3 inhibition, and Yap inhibition. Upon differentiation, PSCs also display increased numbers and maturation of mitochondria [88]. Studies in PSC cardiac differentiation further suggest that mitochondrial maturation, mitochondrial dynamics, and metabolic shift are prerequisites for the differentiation of stem cells into a functional cardiac phenotype (Fig. 2). In mouse and human PSC-derived cardiomyocytes, the level of mitochondrial OXPHOS and cristae maturation is significantly increased compared to those in PSC, whereas glycolysis is decreased [37]. Inhibition of OXHOS during cardiac differentiation prevents mitochondrial organization, causing poor cardiomyocyte differentiation, as characterized by deficient sarcomerogenesis and contractile malfunction [37]. Consistent with mouse developing heart, the fragmented mitochondrial network in PSCs progressively fuses and elongates during differentiation, which further confirms the essential role of mitochondrial fusion in proper cardiomyocyte differentiation [112]. mPTP inhibition by Cyclosporin A (CsA) facilitates the differentiation of functional cardiomyocytes from mouse and human PSCs [35]. Moreover, the differentiated cardiomyocytes display unique metabolic features in contrast to non-cardiomyocytes. PSC-derived non-cardiomyocytes proliferate and mainly rely on glucose and glutamine metabolism; however, differentiated cardiomyocytes rely on lactate oxidation [211, 212]. Understanding the metabolic cues in heart development not only facilitates cardiac differentiation, but also enables the development of nongenetic methods to purify PSC-derived cardiomyocytes for clinical applications [88].

Accumulating evidence has demonstrated the central role of mitochondria in the preservation of cardiac homeostasis. Major theories of mitochondrial dysregulation, including mitochondrial DNA (mtDNA) mutation/damage, impaired OXPHOS, defects in mitochondrial translation, imbalanced mitochondrial dynamics, dysregulated lipid metabolism, and mitochondrial Ca²⁺ overload, are all associated with progressive decline in myocardial structure and function (Fig. 3). The mtDNA has a very high rate of mutation [162, 185], but only limited repair capacity [245]. Therefore, mtDNA mutation and inadequate mtDNA level compromise the integrity of the mitochondrial genome, resulting in defects in electron transport chain (ETC) complexes and ATP supply. Various mtDNA mutations have been detected in heart tissues. Genetic defects in ETC structure subunits of the OXPHOS system (complex I, II or III) [5, 7, 29, 161] and proteins associated with proper assembly of respiratory chain complexes [30], such as mutation in COX10 and COX15 or the assembly factors of complex IV [8], have been described in different types of cardiomyopathy. Recently, pathogenic mutations in components of the mitochondrial translation machinery have also been linked to various progressive human diseases, including neurological disorders, heart/muscle diseases, and endocrine problems [127]. During translation of mitochondrial mRNAs, various translation factors interact with the mitochondrial ribosome to synthesize the polypeptides. Mitochondrial RNA contains two mitochondrial ribosomal RNAs (12S mt-rRNA and 16S mt-rRNA) and 22 mt-tRNAs encoded by mtDNA. Mutation of mt-rRNAs [52, 145, 192], mt-tRNAs [4, 74, 76], mitochondrial ribosomal proteins (MRPL3 and MRPL44), and the translation elongation factor (TSFM) [1, 48, 54, 65] have all been documented in various cardiomyopathies, together with decreased synthesis of mitochondrial polypeptides and impaired energy production. As highly dynamic organelles, homeostasis of mitochondria is tightly controlled by a quality control system consisting of fission, fusion, and degradation of mitochondria, collectively referred to as mitochondrial dynamics. Fusion involves content exchange, allowing complementation of mitochondrial solutes, proteins and DNA. Fission allows segregation of damaged mitochondria eliminated by selective degradation (mitophagy). Mitochondrial dynamics is essential for mitochondrial shape, size, distribution, quality control (mitophagy), and transport of mitochondria within the cell. For example, when cells are challenged by stress, mitochondrial dynamics spring into action to trigger compensatory and adaptive cellular response [49, 176]. Pathogenic mutations in the core machinery and defects of mitochondrial dynamics-related proteins have been documented in various cardiac diseases. Recently, mitochondrial lipid metabolism was found to play an important role in the regulation of mitochondrial dynamics [70]. Although most lipids are synthesized in the endoplasmic reticulum (ER), the mitochondrial membrane also contributes to phospholipid synthesis, including cardiolipin (CL), phosphatidic acid, and phosphatidylethanolamine (PE) [157]. Mitochondrial lipid metabolism has been associated with cardiomyopathy in Barth syndrome (BTHS), an x-linked autosomal recessive disease [103]. The underlying cause of BTHS is the mutation of CL transacylase tafazzin (TAZ), which results in the impaired synthesis of CL and compromises mitochondrial homeostasis. CL is a highly abundant lipid in heart tissues and is important in maintaining mitochondrial homeostasis, ranging from mitochondrial protein transport, mitochondrial dynamics, mitochondrial cristae formation, and respiratory chain to cytochrome c-related apoptosis [51]. CL directly interacts with optic atrophy 1 (Opa1), a core component of mitochondrial fusion machinery [70], and it is also associated with the processing of Opa1 [172]. Impaired Opa1 processing has been reported to cause dilated cardiomyopathy and heart failure [221], indicating an intriguing involvement of CL-regulated Opa1 processing. Interestingly, mtDNA-protein complexes were found to be co-fractionated with mitochondrial lipid components [71]. Further research has shown that mtDNA stability and segregation are associated with mitochondrial lipid metabolism [148]. Recent studies of mitochondrial Ca²⁺ overload have revealed a critical role in the pathophysiology of heart failure [240]. Specifically, excessive Ca²⁺ in mitochondria will increase the generation of ROS and mPTP opening, thereby causing mitochondrial dysfunction and cell death [163]. No direct evidence has suggested that mtDNA polymorphisms/mutations are involved in dysregulated mitochondrial Ca² in heart diseases. Nevertheless, it has been documented that mtDNA polymorphisms alter mitochondrial calcium levels in forebrain cells of mutant mtDNA polymerase (Polg) transgenic mice [114]. These studies put the mitochondrial genome squarely in the center of the cardiac homeostatic network which is responsible for coordinating the involvement of mitochondria in cellular homeostasis.

Mitochondria are endosymbiotic organelles containing their own genome, mtDNA. The mtDNA, as an ‘extranuclear’ genetic molecule, is unique and different from nuclear DNA (nDNA), the regulation of which is independent from nDNA. The mtDNA is generally packaged into nucleoids, the heritable units of mtDNA. The organization of mtDNA is compact without intronic structures in protein-coding genes and often contains overlapping reading frames between neighboring genes, thereby leading to higher gene density in mtDNA than that in nDNA [243]. Although the size of mtDNA is small, the mRNA transcribed from mtDNA can represent a surprisingly large proportion of total cellular mRNA, nearly 30% in the heart [159]. Mammalian mtDNA is a compact circular DNA molecule of 16,569 base pairs with a contour length of 5 µm that possesses coding regions (> 90%, around 93% in humans and mice) and noncoding regions. The coding regions encode 11 mRNAs (translated to critical proteins of the OXPHOS), 2 ribosomal RNAs (rRNAs, 12S and 16S), and 22 transfer RNAs (tRNAs). The noncoding regions include D-loop and the light-strand origin of replication (OriL) that harbors regulatory regions for mtDNA transcription and replication.

Unlike nDNA, mtDNA lacks histones and is packaged into the nucleoid, as noted above, inside the mitochondria matrix, the only known organization unit of the mitochondrial genome containing mtDNA and associated architectural proteins, thus forming a distinct protein-DNA structure. Mammalian mitochondrial nucleoids are slightly ellipsoid with an average shape varying between slightly elongated (80 × 80 × 100 nm) [125] to truly ellipsoid (25 × 45 × 100 nm) [3, 24]. Because of its extranuclear localization, mtDNA was thought to follow a maternal pattern of inheritance with no paternal contribution [99, 222]. However, the discovery of Will Luo and colleagues presents an interesting conceptual breakthrough in that mtDNA can also be occasionally paternally inherited [149]. The number of mtDNA molecules per nucleoid is currently debated among several researchers. Recent studies support the idea of multiple mtDNA copies, ranging from 1.4 to 7.5 mtDNA per nucleoid [133], challenging the idea of a single copy of mtDNA per nucleoid [125]. However, the nucleoid morphology and size are independent of mtDNA copy number [125]. In mouse embryonic fibroblast cells (MEFs), an increase in mtDNA copy number induced by human TFAM overexpression leads to an increase in the total number of nucleoids per cell and cell size when compared with the wild-type MEFs, while the nucleoid size is fundamentally unaffected [125]. As revealed by proteomic analyses, a large number of potential mitochondria nucleoid-associated proteins have been identified and can be grouped into different functional classes. Apart from the key components of mitochondrial transcription (POLRMT, TFAM, TFB2M, TEFM) and replication (POLG, Twinkle, mtSSB) machinery, other important proteins, such as mitochondrial ribosomal proteins, proteases, chaperones, RNA-binding proteins, and RNA-processing proteins, have also been associated with the nucleoid. It was recently suggested that the mitochondrial nucleoid, not the mtDNA, is a unit of mitochondrial genome inheritance [68]. Nucleoids are semiregularly spaced within mitochondria to secure correct mtDNA transmission into the daughter cells at cell division [108, 173]. In addition, nucleoid-associated proteins, such as TFAM, have also been reported to influence mtDNA transmission [2, 113]. Knockdown of TFAM not only results in an abnormal nucleoid, but also causes asymmetric transmission of mtDNA into two daughter cells [113]. Thus, significant interest in nucleoid structure and its regulation has arisen to address the underlying mechanisms of mitochondrial diseases.

As the most abundant architectural protein in the mitochondrial nucleoid, TFAM plays a major role in nucleoid structure. TFAM, a DNA-binding protein with tandem high-mobility group (HMG)-box domains, is highly conserved across species and present at a ratio of 1000 molecules per mtDNA, or 1 subunit per 16–17 bp of mtDNA [21]. When bound specifically to mtDNA promoters or nonspecifically throughout the mtDNA genome, TFAM fully coats and packages mtDNA into nucleoids by creating a stable U-turn with an overall bend of 180° [169]. TFAM is essential for the maintenance of mtDNA integrity, including mtDNA packaging [116, 170], transcription [196], replication [56], mtDNA copy number maintenance [101], and correct transmission [2, 113]. Growing evidence obtained from TFAM knockout mice has established a critical role of TFAM in the regulation of mitochondrial function and cardiac homeostasis. Inactivation of TFAM by Nkx2.5Cre in the mouse embryonic heart at E7.5 induces disrupted mitochondrial biogenesis and morphology, elevates ROS production, and depolarizes mitochondria, thereby causing embryonic lethality at E15.5 with marked myocardial wall thinning [248]. Other cardiomyocyte-specific TFAM knockout mouse models further characterized the effect of TFAM inactivation on mitochondrial cardiomyopathy with the development of dilated cardiomyopathy (DCM) [225]; [134, 202]. A decrease of TFAM expression has also been observed in several cardiac failure models with mitochondrial dysfunction [67, 100, 115, 131]. On the contrary, overexpression of TFAM exhibited therapeutic potential in heart failure, leading to the reduction of protease expression, ROS production, cytoplasmic calcium [128], and attenuated pathological hypertrophy [64] in cardiomyocytes. Cardioprotective effects were further confirmed in transgenic mouse models containing human TFAM gene, as evidenced by ameliorated mitochondrial deficiencies and improved cardiac function after myocardial infarction [102]. Although TFAM inhibition efficiently decreased the levels of mitochondria-encoded transcripts [104, 225], notably, overexpression of TFAM did not affect mtRNA levels, but rather mtDNA copy number [102]. These lines of evidence imply the presence of more complex regulatory mechanisms responsible for the association of TFAM, mtDNA, and mitochondrial homeostasis.

In fact, different mtDNA compaction from fully compacted nucleoids to naked DNA has been observed from the effect of physiological variation in TFAM on mtDNA ratios [60]. At high TFAM:mtDNA ratios, the nucleoid is fully compacted such that TFAM forms large, stable filaments on the DNA in a manner that blocks DNA melting through POLRMT and TWINKLE, thus preventing active mtDNA replication and transcription [60]. In support of this notion, a mild increase in TFAM levels in vivo (∼twofold) leads to a proportional increase in mtDNA copy number [53], whereas forced overexpression of TFAM leads to mtDNA depletion [178]. Interestingly, when mtDNA was depleted in the presence of ethidium bromide (EB), knockdown of Lon (the regulator of TFAM) is unable to degrade TFAM, thereby causing a dramatic increase in the TFAM:mtDNA ratio, in turn resulting in the inhibition of mitochondrial transcription [156]. This indicates that the TFAM:mtDNA ratio may affect the number of compacted mtDNA and open mtDNA molecules.

Considering the role of TFAM as a histone-like protein in mtDNA packaging, it is plausible that TFAM also has modifications equivalent to those that occur on nuclear histones. In the nucleus, histones pack nDNA into nucleosomes forming chromatin. Modification of histones, including phosphorylation, methylation, acetylation, and ubiquitylation, strongly influences gene integrity [117]. Recently, nuclear-encoded transcription factors that regulate nuclear chromatin were found in the mitochondrial nucleoid, including MOF [32], members of the AP1 family (c-Jun and JunD), CEBPB [20] and MEF2D [195]. Most importantly, MOF, a histone lysine acetyltransferase that remodels chromatin, was found to directly bind to mtDNA and inhibit mtDNA transcription [32]. Interestingly, TFAM levels were not obviously affected in these MOF-depleted cells. The discovery of TFAM-DNA-binding affinity, as regulated by acetylation and phosphorylation of TFAM [120], raises the question of whether TFAM is the acetyltransferase target of MOF in mitochondria. Moreover, TFAM was found to facilitate damaged mtDNA degradation by accelerating strand cleavage at abasic (apurinic/apyrimidinic, AP) sites [187], a ubiquitous form of DNA damage [141]. Considering the full coating of mtDNA by TFAM, these results suggest the ubiquitous nature of the TFAM-mediated abasic site containing mtDNA degradation.

For a long time, the higher-order organization of mtDNA in the nucleoid was considered far less regulated and complex owing to the lack of chromatin and histones. Intriguingly, recent analysis of DNase-seq and ATAC-seq experiments from multiple human and mouse samples revealed the formation of a conserved footprint during embryonic development, as an indication that mtDNA sites are increasingly occupied, but have low TFAM occupancy [19, 153]. Moreover, mtDNA sites with low occupancy of TFAM tend to adopt noncanonical nucleic secondary structures known as G-quadruplex sequences (GQs) [19]. GQs are notably conserved across species, and GQ motifs have been increasingly recognized to be associated with epigenetic reprogramming and chromatin remodeling [216]. GQs are widely present in mtDNA, as revealed by recent studies [55, 98], and GQ formation potentially regulates mitochondrial gene transcription and replication [55, 91, 167]. These studies further reflect the existence of the higher-order organization of protein-DNA structure in the mitochondrial genome in which the accessibility of mtDNA is potentially regulated by the TFAM:mtDNA ratio. Development of efficient experimental tools that allow selective detection of mitochondrial GQs or isolation of mitochondrial GQ structures could facilitate the discovery of novel mtDNA-binding proteins. Although important progress has been made in deciphering the molecular mechanisms of mito-nuclear communication in protein-DNA interaction, mitochondrial genome regulation is more complex and more regulated than once thought. Hence, it will be important for future studies to decipher the structure and organization of nucleoids and their specialized functions.

The discovery of noncoding RNAs (ncRNAs) has extended our knowledge of the molecular pathways involved in cardiac homeostasis. NcRNAs, as a large part of the noncoding genome, play various roles in cellular processes, rather than serving as a protein template. NcRNAs are classified into three types based on their localization and genetic origin: (1) cytoplasmic, nuclear-encoded ncRNAs; (2) mitochondria, nuclear-encoded ncRNAs and imported to the mitochondria; (3) mitochondria, mtDNA-encoded ncRNAs. Recently, ncRNAs have been increasingly found in the mitochondria and termed as mito-ncRNAs. These ‘mito-ncRNAs’ are ncRNAs located inside mitochondria, irrespective of their genetic origin [217]. Significant research has been aimed at unveiling how the ncRNAs are linked with their specific subcellular localizations and functions. These mito-ncRNAs can be grouped into microRNAs [miRNAs, 17–23 nucleotides (nt)], long noncoding RNAs (lncRNAs, > 200 nt), and circular RNAs (circRNAs). Cytoplasmic ncRNAs can modulate mitochondrial functions through a cytoplasmic mechanism. Next, we will review the role and function of particular mito-ncRNAs in cardiac homeostasis (see Table 1).

Although mitochondrial-generated ncRNAs (e.g., ASncmtRNA-1 and ASncmtRNA-2) are translocated into the nucleus and associated with perinuclear chromatin [62, 129], their functions in transcription regulation remain largely unknown. Recently, four approaches, termed global RNA Interactions with DNA by deep sequencing (GRID-seq) [137], Chromatin-Associated RNA sequencing (ChAR-seq) [15], HiChIRP [164], and RNA in situ conformation sequencing (RIC-seq) [26], have identified that chromatin-associated RNAs (CARs) are largely involved in long-range chromatin interactions. For example, ncRNA can bind to the MYC enhancer and MYC promoter and interact with the RNA-binding protein hnRNPK, while the oligomerization of hnRNPK can juxtapose the enhancer to MYC promoter and promote MYC transcription [26]. In addition, ncRNA ThymoD facilitates CTCF/cohesin-dependent loop extrusion and repositions Bcl11b enhancer from nuclear periphery proximal to Bcl11b promoter, leading to transcription activation of Bcl11b and T cell commitment [106].

In addition to juxtaposing enhancers and promoters in the 3D genome, CARs can also modulate chromatin topology. For example, YY1 is an emerging architectural protein for orchestrating enhancer-promoter interaction [229], and its localization on DNA is facilitated by direct binding with CARs [198]. Furthermore, insulated neighborhoods are chromatin topological structures (loops) formed by CTCF-CTCF homodimers [50] and the major structural components of TADs, the boundaries of which are enriched by CTCF [92]. CTCF can bind large numbers of ncRNA, and a recent report from Danny Reinberg’s lab indicated that mutation of the RNA-binding region of CTCF could disrupt CTCF loop formation, resulting in transcriptional inactivation [190]. This means that the functions of ncmtRNAs in CTCF formation could be further delineated in the future. In addition, whether nuclear-localized ncmtRNAs regulate genome compartment (e.g., heterochromatin) formation and maintenance through m6A-mediated phase separation [186] will be another intriguing topic for further investigation (Fig. 6A).

Although the vast majority of mitochondrial proteins are encoded in the nucleus, some new classes of peptides were recently discovered with mitochondrial origins. These mitochondrial-derived peptides (MDPs) are the peptides encoded by small open reading frames (ORFs) within mtDNA (mtORF) [223]. At present, three types of MDPs have been identified, including 16S rRNA-encoded Humanin (HN), 12S rRNA-encoded MOTS-c and 16S rRNA-encoded small humanin-like peptides (SHLP1-6). As the first MDP, HN was discovered by Hashimoto and his colleagues in the undamaged brain of a patient with Alzheimer’s disease [86]. HN structure and function has been well studied, and various HN analogs are believed to have protective effects against cardiovascular diseases [78] in addition to neuronal diseases [132, 208]. Administration of an analog of HN (the serine at position 14 replaced by glycine, termed as HNG) prior to, or at the time of, reperfusion has been shown to play a cardioprotective role in a mouse model of myocardial ischemia and reperfusion, which potentially improved cardiomyocyte survival and attenuated apoptosis [166]. Similarly, high-doses of HNG applied during the ischemic period also exerted cardioprotection, leading to significantly decreased cardiac arrhythmia, myocardial infarct size, apoptosis, cardiac mitochondrial dysfunction, and left ventricular dysfunction [209]. Under pathologic conditions, increased ROS is detrimental to cardiomyocytes [179]. Specifically, complexes I and III of the ETC are major sites of ROS production in cardiac mitochondria [203]. Recent studies confirmed the contribution of HNG to reduced ROS production in damaged cardiac mitochondria. Administration of HNG both activated the Abl- and Arg-dependent cellular defense system and decreased complex I activity [210]. Additionally, cardioprotective effects of HNG were recently found in aged hearts through preventing myocardial fibrosis, mPTP opening, and apoptosis [181]. Further studies also show the relationship between MDPs and coronary microvascular dysfunction (CMD), which are both risk factors for cardiac dysfunction. In the cohort of patients with CMD, lower levels of HN were observed in circulating blood [231]. Another independent study also found that the levels of HN and MOTS-c in circulating blood of patients with impaired endothelial function were lower than those of the control groups [181]. These results suggest that HN or MOTS-c has significant potential as a new marker to diagnose CMD or identify potential therapeutic targets.

Recent research has reported that MOTS-c translocates from mitochondria into the nucleus in response to glucose restriction, leading to retrograde signaling [118, 184]. MOTS-c can bind to nuclear genomic DNA and interact with transcription factor Nrf2 to activate gene transcription, resulting in increased cellular resistance to metabolic stress [118]. Reynolds et al. in 2021 found that MOTS-c is an exercise-induced mitochondrial-encoded regulator of muscle homeostasis which can regulate the nuclear genes involved in metabolism and proteostasis, thereby significantly improving physical performance in young, middle-aged, and old mice [184]. Moreover, administration of MTOS-c at late-life (23.5 months) can increase the lifespan of old mice. These landmark studies provided a new paradigm for understanding the impact of MDPs on nuclear gene regulation in metabolic reprogramming.

In addition, a recent study surprisingly found that calpain can cleave full-length junctophilin-2 to covert it from a cardiac structural protein into a transcription factor, leading to its translocation from the cell membrane into the nucleus for cardiac protection [83]. As calpain is not only localized in cytosol, but also in mitochondria [9, 107, 175, 197], future work is needed to delineate the proteins that shuttle from mitochondria into nucleus after mitochondrial-calpain cleavage and subsequently regulate chromatin architecture (Fig. 6B).

The loop extrusion model is the most popular model for CTCF loop, insulated neighborhood formation and TADs [63, 92]. This model postulates that cohesin complex is loaded onto DNA by physical stimulation and reels in DNA to form loops. Loop extrusion can then be stopped when cohesin meets convergent CTCF sites. Recent studies have provided direct evidence in support of this model and have shown conclusively that cohesin extrudes DNA loop in an ATP hydrolysis-dependent manner [44, 119]. Previously, it was well recognized that ATP is generated only from mitochondria. However, this prevailing “dogma” was recently challenged, showing that ATP can also be generated in the nucleus via the hydrolysis of poly (ADP-ribose) to ADP-ribose and that this was shown to be essential for hormone-induced chromatin remodeling in a breast cancer cell line [235]. This raises several questions. First, is the “nuclear pool” only confined to breast cancer cells or present in all mammalian cells? Second, what is the source of ATP for loop extrusion, mitochondria, nucleus or both? Most mitochondrial diseases are caused by mutations of mtDNA, leading to defects in the OXPHOS system and reduction of ATP production [79] and cardiac involvement [233]. Therefore, third, does ATP reduction caused by mutations of mtDNA cause global loss of CTCF loops and transcriptional dysregulation?

Precise mtDNA editing was long deemed as impossible. However, a recent study developed a novel approach based on RNA-free DddA-derived cytosine base editors (DdCBEs), which can catalyse nucleotide conversions in human mtDNA with high target specificity [160]. Based on this technology, human iPSC-derived cardiomyocytes from patients with diseases like hypertrophic cardiomyopathy [135], before and after mtDNA editing (rectification of mtDNA mutation), can be generated and used to study disease mechanisms. For example, by performing high-resolution 3D genome mapping (e.g., in situ Hi-C or CTCF/Cohesin-centric in situ ChIA-PET [17]; HiChIP [165]; or PLAC-seq [59]) in the nucleus from those cells, it is possible to assess the impact of mtDNA mutation on 3D genome reprogramming in the nucleus of cardiomyocytes (Fig. 6C). Such studies would give completely fresh insight into the pathogenesis of mitochondrial diseases with cardiac involvement and reveal the plausible impact of mtDNA mutation on transgenerational inheritance of abnormal nuclear genome topology in patients.