Understanding the molecular basis of cardiomyopathy

Abstract

Inherited cardiomyopathies are a major cause of mortality and morbidity worldwide and can be caused by mutations in a wide range of proteins located in different cellular compartments. The present review is based on Dr. Ju Chen’s 2021 Robert M. Berne Distinguished Lectureship of the American Physiological Society Cardiovascular Section, in which he provided an overview of the current knowledge on the cardiomyopathy-associated proteins that have been studied in his laboratory. The review provides a general summary of the proteins in different compartments of cardiomyocytes associated with cardiomyopathies, with specific focus on the proteins that have been studied in Dr. Chen’s laboratory.

INTRODUCTION

The heart is a complex organ composed of cardiomyocytes as well as other cell types, such as endothelial cells, vascular smooth muscle cells, fibroblasts, pericytes, immune-related cells, and others (1, 2). Cardiomyocytes are the contractile cells of the myocardium, allowing the heart to pump. The structural organization of the cardiomyocyte is depicted in Fig. 1. The basic contractile unit is the sarcomere, which is made up of thick and thin filaments sliding over each other during contraction (3, 4). Sarcomeres are organized into myofilaments that are functionally connected at intercalated disks, which join individual cardiomyocytes together, allowing the cardiomyocytes to contract as a single coordinated unit and permitting mechanical and electrical coupling between cells (5–7). The sarcolemma, i.e., the cardiomyocyte plasma membrane, contains deep invaginations called t-tubes (transverse tubules), which allow for rapid transmission of the action potential to the interior of the cell to induce muscle contraction in a process called excitation-contraction coupling (8–10). The action potential is induced by pacemaker cells in the sinoatrial and atrioventricular nodes, which causes an influx of calcium through calcium channels in the t-tubules, triggering synchronized calcium release from the nearby sarcoplasmic reticulum (SR), the calcium storage unit of the cell, to induce muscle contraction. Cytosolic calcium is subsequently pumped back into the SR, causing relaxation. To meet the high energy demand of the continuously contracting cardiomyocytes, they contain a high number of mitochondria (11). A network of microtubules, intermediate filaments, and nonsarcomeric actin laterally links adjacent myofibrils, associates nuclei to the sarcomere, and connects myofibrils to the sarcolemma and the extracellular matrix at the costamere, providing a structural framework for the cell (12, 13). Furthermore, the cytoskeletal network mediates the transport of proteins and organelles and plays an important role in mechanosensing and signal transduction.

Figure 1.

Schematic representation of cardiomyocyte structure, including key proteins associated with human cardiomyopathy. Cardiomyopathy-associated proteins studied in Dr. Ju Chen’s laboratory (listed in Table 1) are located in different compartments of the cardiomyocyte, including the sarcomere, intercalated disk (ICD), sarcoplasmic reticulum (SR), costamere, nucleus, and cytosol. ALP, actinin-associated lim protein; BAG3, BCL2-associated athanogene 3; CAR, coxsackievirus and adenovirus receptor; CARP, cardiac ankyrin repeat; CASQ2, calsequestrin 2; ECM, extracellular matrix; eEF1A2, eukaryotic elongation factor 1 alpha 2; ENH, Enigma homolog protein; ER, endoplasmic reticulum; FAK, focal adhesion kinase; ILK, integrin linked kinase; FHLs, four and a half LIM proteins; ILK, integrin-linked kinase; JPH2, junctophilin 2; LAP1/2, lamina-associated polypeptide 1 and 2; LBR, lamin B receptor; LEMD2, LEM domain nuclear envelope protein 2; LIMP-2, lysosomal integral membrane protein-2; LTCC, L-type calcium channel; MLP, muscle LIM protein; MYL2/3, myosin light chain 2/3; MYBP-C, myosin binding protein-C; NCX, sodium/calcium exchanger; PLN, phosholamban; PRDM16, PR/SET domain 16; RyR2, ryanodine receptor 2; SERCA, SR calcium ATPase; SUN1/2, Sad1 and UNC84 domain-containing 1 and 2; TBX20, T-box transcription factor 20; T-cap, titin cap; XIRP, xin actin-binding repeat containing; ZO-1, zonula occludens-1.

Mutations in structural or regulatory proteins of the cardiomyocyte can lead to cardiomyopathy, which is a heterogeneous disease defined as “a myocardial disorder in which the heart is structurally and functionally abnormal in the absence of coronary artery disease, hypertension, valvular disease and congenital heart disease sufficient to cause the observed myocardial abnormality” (14). Cardiomyopathies can be divided into hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), restrictive cardiomyopathy (RCM), arrhythmogenic cardiomyopathy (ACM), and left ventricular noncompaction (LVNC), which can be further classified into genetic and acquired forms (15).

HCM affects ∼1 in 500 and is the leading cause of sudden cardiac death in young athletes (16, 17). HCM is characterized by left ventricular (LV) hypertrophy in the absence of hemodynamic stresses (e.g., hypertension, aortic valve stenosis) or systemic diseases, such as amyloidosis and glycogen storage disease (reviewed in Refs. 18, 19). Patients typically show preserved systolic function but impaired LV relaxation (i.e., diastolic dysfunction), which may eventually develop into LV dilation, wall thinning, and systolic dysfunction. However, the clinical manifestation is very variable, ranging from asymptomatic LV hypertrophy to severe heart failure or sudden cardiac death. At the histological level, cardiomyocyte hypertrophy, myofibrillar disarray, and interstitial fibrosis are typical features. HCM is in most cases caused by autosomal dominant mutations in genes encoding sarcomeric proteins, which can be identified in ∼60% of clinical cases (18). A common molecular mechanism leading to HCM is enhanced calcium sensitivity and affinity of the myofilament as well as inefficient ATP utilization for tension generation, resulting in a higher energy demand and consequent energetic inefficiency (reviewed in Ref. 20).

DCM is characterized by progressive ventricular dilation and systolic dysfunction in the absence of abnormal loading conditions (e.g., hypertension, coronary artery disease, valvular disease) (reviewed in Refs. 21–24) and has an estimated prevalence of 1 in 250 (24). The age of onset ranges from newborn to old age, although most patients are diagnosed between 20 and 50 yr of age. The clinical manifestation varies from asymptomatic to severe and may include heart failure, arrhythmia, thromboembolism, conduction defects, and sudden cardiac death. DCM can be caused by a wide range of conditions, such as inflammatory conditions (e.g., myocarditis, autoimmune disease), metabolic disorders (e.g., hyperthyroidism), and toxins (e.g., alcohol, chemotoxins, drugs) (22). Furthermore, 30–50% of DCM cases are familial (25–27), and causative gene mutations have been identified in >100 genes encoding proteins of various cellular compartments, including the sarcomere, intercalated disk (ICD), costamere, SR, mitochondria, sarcolemma, and nuclear envelope (reviewed in Ref. 21). The inheritance is mostly autosomal dominant, but autosomal recessive, X-linked, and mitochondrial inheritance have also been described. In contrast to HCM, reduced myofilament calcium sensitivity is commonly associated with DCM. Furthermore, molecular mechanisms associated with DCM include impaired force generation and transmission, altered myofilament calcium handling, ion channel dysfunction, defective mechanosensing, myocardial energy deficit, and structural changes in the sarcomere, cytoskeleton, and/or nucleus (21).

RCM is a rare form of cardiomyopathy characterized by increased ventricular stiffness, resulting in impaired ventricular filling and consequent abnormal relaxation of the ventricles and diastolic dysfunction (reviewed in Ref. 28). Moreover, the atria are often enlarged because of increased end-diastolic pressure in the ventricles, whereas wall thickness and ventricular volume typically remain normal until advanced stages of the disease. The prevalence of RCM is unknown, but it has been estimated to account for <5% of cardiomyopathies. Among the cardiomyopathies, RCM has the poorest prognosis, especially in children, where the mortality is as high as 50% within the first 2 yr after diagnosis (29). In particular, RCM is associated with an increased risk of arrhythmias and sudden cardiac death. Common causes of RCM are infiltrative disease (e.g., amyloidosis, sarcoidosis), lysosomal and glycogen storage disorders (e.g., Fabry disease, hemochromatosis, glycogen storage disease), endomyocardial fibrosis, and cancer treatments (anthracycline, radiation) (30). Furthermore, RCM can be familial, mostly caused by autosomal dominant mutations in genes encoding sarcomeric and sarcomere-associated proteins (28). Many RCM-associated mutations have been associated with increased myofilament calcium sensitivity, resulting in delayed relaxation and increased energy consumption, whereas others cause protein aggregation, likely as a result of impaired protein quality control.

LVNC is characterized by prominent LV trabeculations, deep intratrabecular recesses, and a two-layered LV wall composed of a thin compacted outer (epicardial) layer and a thicker noncompacted inner (endothelial) layer, giving the LV a spongy appearance (reviewed in Refs. 31–33). Pathological LVNC is thought to result from an arrest in compaction during early myocardial development, but a higher prevalence of LVNC in pregnant women and athletes suggests that LVNC can also occur as a physiological adaptation to pressure overload in healthy adults (34). The prevalence of LVNC remains unclear, as there is a wide variation in the reported prevalence, partly due to the absence of specific diagnostic criteria or “gold standard” and the use of different imaging techniques for diagnosis. In particular, in a meta-analysis of studies reporting LVNC prevalence in adults, the overall prevalence was estimated to 1.28% when diagnosed by echocardiography and 14.79% when based on cardiac magnetic resonance imaging (34). The clinical manifestation of LVNC is highly heterogeneous, ranging from asymptomatic to severe heart failure, arrhythmias, thromboembolism, LV dysfunction, and sudden cardiac death. Both sporadic and familial forms have been described, and LVNC can occur both in an isolated form and associated with other cardiomyopathies. Up to 48% of cases are familial (35) and are most frequently caused by mutations in sarcomeric proteins but can also be caused by mutations in nuclear envelope, cytoskeletal, and mitochondrial proteins, transcription factors, and ion channels (32, 35). The inheritance of LVNC is mostly autosomal dominant or X-linked recessive, although autosomal recessive and mitochondrial inheritance also occur.

ACM is a progressive disease defined by cardiomyocyte death and fibrofatty replacement of the myocardium, leading to increased susceptibility to ventricular arrhythmias and sudden cardiac death (reviewed in Refs. 36–38). Furthermore, advanced stages of ACM are characterized by systolic dysfunction, biventricular dilation, and heart failure. ACM principally affects the right ventricle (RV), but cases with biventricular or principal LV involvement have become increasingly reported (36, 39, 40). For that reason, the disease was renamed from arrhythmogenic right ventricular cardiomyopathy (ARVC) or dysplasia (ARVD) to the broader term ACM in a consensus statement from the Heart Rhythm Society (HRS) and the European Heart Rhythm Association (EHRA) (41). ACM affects ∼1 in 1,000–5,000 but is likely underdiagnosed, as the initial manifestation may be sudden cardiac death. In particular, ACM is a leading cause of sudden death in athletes and young adults and has been estimated to be responsible for 10% of cardiovascular deaths in people under 65 yr of age (42, 43). Genetic mutations are responsible for up to 60% of cases and have been identified in 16 genes, mostly with autosomal dominant inheritance with variable penetrance (37), although autosomal recessive inheritance has also been reported. Mutations in genes encoding desmosomal proteins account for ∼80% of cases with confirmed pathogenic mutations, while causative mutations have less frequently been identified in genes encoding sarcomeric, ICD, SR, and nuclear envelope proteins or growth factors. A common molecular mechanism underlying ACM is thought to be ICD remodeling due to mutations in desmosomal proteins, leading to disruption of cardiomyocyte adhesion and consequent apoptosis or necrotic cardiomyocyte death, resulting in cardiomyocyte loss and fibrofatty replacement (37, 38).

The major focus of the laboratory of Dr. Ju Chen, the senior author of this article, in the last two and half decades, has been to provide insights into the molecular basis of cardiomyopathy by using genetically engineered mouse models, physiological measurements, and a range of molecular and cell biological techniques. The present review is based on his 2021 Robert M. Berne Distinguished Lectureship of the American Physiological Society (APS) Cardiovascular Section. The lecture summarized the advancements in the understanding of the molecular mechanisms underlying human cardiomyopathies obtained in his laboratory. Thus, although this review aims to provide a general summary of the current understanding of the molecular basis of cardiomyopathy, it mainly focuses on the cardiomyopathy-associated proteins that have been studied in his laboratory (listed in Table 1).

Table 1.

Association of selected cardiac genes with human cardiomyopathies

Gene	Protein Name	Location	Type of Cardiomyopathy	Other Diseases	References
LDB3	Cypher/ZASP/LIM domain binding 3	Z-line	DCM, HCM, LVNC, ACM	MFM	(58–60, 62–63)
CSRP3	Muscle LIM protein (MLP)/cysteine and glycine-rich protein 3 (CSRP3)	Z-line, M-line, ICD, costamere, sarcolemma, nucleus	DCM, HCM		(95–103)
NEBL	Nebulette	Z-line	DCM, EFE, HCM, LVNC		(138–140)
MYPN	Myopalladin	Z-line, I-band, nucleus	DCM, HCM, RCM	NEM, CM, CAPM	(157–160)
FHL1	Four and a half LIM protein 1 (FHL1)	I-band, M-line	HCM, ACM, DCM	EDMD, RBM, SPM, XMPMA	(177–186)
FHL2	Four and a half LIM protein 2 (FHL2)	I-band, M-line	HCM, DCM		(194, 195)
ANKRD1	Cardiac ankyrin repeat protein (CARP)	I-band, nucleus	DCM, HCM		(212–214)
MYL2	Myosin light chain 2 (MYL2)/ventricular myosin light chain 2 (MLC-2v)	A-band	HCM, DCM	MFM	(265–285)
OBSCN	Obscurin	M-line, Z-line, A/I junction, costamere, ICD, SR	HCM, DCM, LVNC, ACM		(328–332)
JUP	Junction plakoglobin/γ-catenin	ICD	ACM	NXD	(373–388)
CXADR	Coxsackievirus and adenovirus receptor (CAR)	ICD	Ischemia-induced ventricular fibrillation, elevated blood pressure (SNPs)		(414–416)
TJP1	Zonula occludens-1 (ZO-1)	ICD	ACM		(433)
VCL	Vinculin, metavinculin	Costamere, ICD	DCM, HCM		(485–489)
ILK	Integrin-linked kinase (ILK)	Costamere	ACM		(483, 533–535)
FLNC	Filamin C	Costamere, Z-line, ICD	HCM, RCM DCM, ACM.	MPD, MFM	(482, 572–586)
JPH2	Junctophilin-2	Junctional SR	HCM, DCM		(627–631)
NEXN	Nexilin	Junctional SR	HCM, DCM		(654–656)
SYNE1/ SYNE2	Nesprin-1/nesprin-2	Nuclear envelope	DCM	EDMD, SCARSCAR, AMC	(680, 682–689)
TMEM43	Luma	INM	ACM	EDMD	(715–722)
TBX20	T-box transcription factor 20 (TBX20)	Nucleus	Congenital heart disease, DCM, LVNC		(730–734)
PRDM16	PR domain-containing protein 16 (PRDM16)	Nucleus	DCM, LVNC		(755–761)
BAG3	BCL2-associated athanogene 3 (BAG3)	Cytosol, Z-line	DCM, RCM, HCM	MFM	(572, 807–819)
EEF1A2	Eukaryotic translation elongation factor 1 alpha 2 (eEF1A2)	Cytosol	DCM	DEE, MRD	(804–806)

ACM, arrhythmogenic cardiomyopathy; AMC, arthrogryposis multiplex congenita spinocerebellar; CAPM, cap myopathy; CM, congenital myopathy; DCM, dilated cardiomyopathy; DEE, developmental and epileptic encephalopathy; EDMD, Emery–Dreifuss muscular dystrophy; EFE, endocardial fibroelastosis; HCM, hypertrophic cardiomyopathy; ICD, intercalated disk; INM, inner nuclear membrane; LVNC, left ventricular noncompaction; MFM, myofibrillar myopathy; MPD, distal myopathy; MRD, mental retardation, autosomal dominant; NEM, nemaline myopathy; NXD, Naxos disease; RBM, reducing body myopathy; RCM, restrictive cardiomyopathy; SCAR, spinocerebellar ataxia, autosomal recessive; SNP, single-nucleotide polymorphism; SPM, scapuloperoneal myopathy; XMPMA, X-linked myopathy with postural muscle atrophy.

THE SARCOMERE

The sarcomere is the smallest contractile unit of striated muscle in which interdigitating thick myosin filaments and thin actin filaments slide past each other during muscle contraction and relaxation (3, 4). A third filament is formed by titin, the largest known vertebrate protein (3–3.7 MDa), which extends over 1 µm from the Z-line to the M-line. Titin acts as a molecular spring responsible for the passive stiffness of striated muscle through the mechanical properties of its multiple domains (44, 45). Furthermore, it is important for a variety of processes, including sarcomere assembly and organization, force transmission, and mechanosensing (reviewed in Refs. 46, 47). Mutations in sarcomeric proteins have been associated with a wide range of congenital cardiomyopathies, including HCM, DCM, RCM, and LVNC (48). Below the different parts of the sarcomere are described and proteins associated with congenital cardiomyopathies are discussed.

Z-Line

The Z-line is a highly organized multiprotein complex at the boundary between sarcomeres in which actin and titin filaments from adjacent sarcomeres are anchored and cross-linked by α-actinin. α-Actinin plays a central role in connecting proteins at the Z-line and also interacts with a wealth of other Z-line proteins, including CapZ, nebulette, PDZ-LIM proteins, and members of the palladin/myopalladin/myotilin family (reviewed in Refs. 49–52). Z-lines are laterally aligned through a link to the intermediate filament desmin (53, 54) and connect myofibrils to the sarcolemma and extracellular matrix via the costamere (reviewed in Refs. 55, 56). Thus, in addition to a structural role in providing structural stability, the Z-line is essential for efficient force production and transmission. Furthermore, the Z-line plays a key role in mechanosensing and transduction of biomechanical signals. Consequently, mutations in a large number of Z-line and Z-line-associated proteins have been linked to cardiomyopathies (49–51).

Genes encoding Z-line proteins associated with DCM include LDB3 (cypher/ZASP), CSRP3 (muscle LIM protein, MLP), TCAP (titin cap/telethonin), ACTN2 (α-actinin 2), MYPN (myopalladin), NEBL (nebulette), and PDLIM3 (PDZ and LIM domain 3/actinin-associated lim protein, ALP) (reviewed in Ref. 49). Furthermore, many Z-line proteins have been linked to HCM, including MYOZ2 (myozenin-2/calsarcin-1), LDB3, CSRP3, TCAP, ACTN2, and MYPN. MYPN mutations have also been associated with RCM, LDB3 and ACTN2 mutations with LVNC, and LDB3 mutations with ACM (listed in Ref. 49). Mutations located within the NH₂-terminal Z-line portion of titin have also been linked to DCM and in rare cases HCM and LVNC (46, 57). A major focus of our laboratory has been to dissect the role of Z-line proteins in cardiac function and disease, including cypher/ZASP, MLP, nebulette, and myopalladin, which are reviewed in detail below.

Cypher/ZASP.

Heterozygous missense mutations in the LIM domain binding 3 (LDB3) gene, encoding cypher/ZASP, are causative for various human cardiomyopathies, including DCM (58, 59), HCM (60–62), LVNC (59, 62, 63), and ACM (63). In addition, LDB3 mutations can cause skeletal muscle myopathies, referred to as zaspomyopathies (64–67). Cypher was first identified from mouse (68), whereas the human ortholog of cypher, Z-band alternatively spliced PDZ-motif protein (ZASP), was independently identified (69). In addition, cypher was independently cloned from mouse as Oracle (70). Cypher/ZASP is a striated muscle-specific PDZ-LIM protein located in the Z-line of the sarcomere. In mouse, six cypher isoforms have been identified, comprising four long isoforms and two short isoforms. All isoforms share an NH₂-terminal PDZ domain, whereas only long isoforms contain three COOH-terminal LIM domains (68, 71). Additionally, cypher contains a ZASP-like motif (ZM), also present in the other PDZ-LIM proteins ALP and CLP36, encoded by either exon 4 or exon 5–7, responsible for its cardiac or skeletal muscle specific expression, respectively (71, 72). One short and two long isoforms are specifically expressed in heart and skeletal muscle, respectively. In human, eight ZASP isoforms have been identified, which are similar in structure to murine cypher isoforms, with some minor differences (59, 69, 72, 73). However, as opposed to mouse, human isoforms containing exon 4 are expressed in both heart and skeletal muscle and isoforms containing exon 6 are highly expressed in both cardiac and skeletal muscle (73). This is important, as mutations both in exons 4 and 6 have been linked to cardiomyopathies (59).

Within the Z-line, cypher/ZASP directly interacts with α-actinin (68, 69, 74), myozenin-2/calsarcin-1/FATZ (75–77), and myotilin (76, 77) through its PDZ domain. Furthermore, the cypher/ZASP PDZ domain binds to ANKRD2 (73), tumor protein P53 (TP53) (73), and the L-type calcium channel (LTCC) (78). Long cypher/ZASP isoforms bind to protein kinase C (PKC) through their LIM domains, and cypher/ZASP itself is a phosphorylation target of PKC (68, 79). Importantly, cypher/ZASP binds to the regulatory subunit RIIα of protein kinase A (PKA), making it a PKA anchoring protein (AKAP) (78). In particular, cypher/ZASP is itself a target of PKA and was found to promote PKA-mediated phosphorylation of the LTCC, regulating channel activity (78, 80). Furthermore, cypher/ZASP interacts with glycolytic enzyme phosphoglucomutase 1 (PGM1) (81) as well as binds to and inhibits the Ser/Thr phosphatase calcineurin, which also targets the LTCC (78). Thus, cypher/ZASP is important for striated muscle structure and signaling, and its requirement for normal cardiac function has been confirmed in studies of various mouse models as described below.

Conventional knockout (KO) of cypher in mouse resulted in postnatal death within 1–5 days after birth, associated with reduced milk intake, muscle weakness, cardiac dilation, and severely disorganized Z-lines (74). In addition, double KO of cypher and the PDZ-LIM protein Enigma homolog protein (ENH), both of which are dispensable for cardiac development, resulted in embryonic lethality by embryonic day 11.5 (E11.5), associated with cardiac dilation and Z-line disorganization, demonstrating that the two proteins play redundant roles in maintaining normal cardiac function and structure during embryonic development (82). Cardiomyocyte-specific deletion of cypher both postnatally and in adult heart resulted in severe DCM and premature death, accompanied by disorganization of the Z-line and altered cardiac signaling (77). Furthermore, although deletion of short cypher isoforms in mouse did not cause any detectable phenotypic abnormalities, ablation of long cypher isoforms resulted in partial neonatal lethality, with surviving mice showing growth retardation and late-onset DCM associated with Z-line abnormalities, cardiac fibrosis, calcification, and altered cardiac signaling (83). Moreover, cypher long isoform-specific KO mice showed an exaggerated pathological response to mechanical pressure overload or chronic β-adrenergic stress, characterized by cardiac dilation and cardiac systolic dysfunction (83). A recent in vitro study also suggested that cypher deficiency induces apoptosis (84), consistent with the finding that cypher acts as a negative regulator of the proapoptotic protein TP53 (73). To study the effect of the ZASP4 p.S196L mutation located in the cardiac-specific exon 4 and associated with human DCM with or without LVNC (59, 61, 62), Li et al. (85) generated transgenic (Tg) mice with cardiomyocyte-specific overexpression of wild-type (WT) or mutant ZASP isoform 4 (the longest cardiac-specific isoform). ZASP-S196L Tg mice developed late-onset DCM associated with mild fibrosis and ultrastructural abnormalities in the Z-line and sarcomere, recapitulating the clinical phenotype, whereas ZASP-WT Tg mice showed no detectable changes. This was preceded by electrocardiogram (ECG) abnormalities and altered L-type Ca²⁺ and Na⁺ currents, suggesting that cardiac conduction defects and atrioventricular block precede the development of DCM. Similarly, in cellular studies the ZASP1 p.D117N mutation was shown to cause loss of function of the Na_v1.5 cardiac voltage-gated sodium channel, predicted to cause cardiac conduction defects (86). Other in vitro studies showed that the ZASP p.D626N mutation located in the third LIM domain increases the binding affinity to PKC (58), whereas mutations located in exons 4 (ZASP p.S189L and ZASP p.T206I) and 10 (ZASP p.I345M) reduce the binding affinity to PGM1 (81).

Also, the PDZ-LIM proteins enigma homolog protein (ENH/PDLIM5) and actinin-associated LIM protein (ALP/PDLIM3) (87) may be involved in cardiac disease, as polymorphisms in the corresponding PDLIM5 and PDLIM3 genes have been associated with an increased risk of idiopathic DCM (88, 89) and PDLIM5 mutations have been proposed as possible disease modifiers (90). Consistent with their possible role in cardiac pathology, global and cardiomyocyte-specific deletion of ENH in mouse resulted in DCM (91, 92), whereas ALP KO mice showed right ventricular chamber dilation and dysfunction (93, 94).

CSRP3/muscle LIM protein.

Heterozygous missense and nonsense mutations in the cysteine and glycine-rich protein 3 (CSRP3) gene, encoding muscle LIM protein (MLP), have been associated with DCM (95–97) and HCM (98–103). Furthermore, reduced MLP protein levels have been reported in patients with DCM or ischemic cardiomyopathy (104). MLP is a 21-kDa striated muscle-specific protein present in the Z-line, M-line, ICD, costamere, plasma membrane, and nucleus (reviewed in Ref. 105). It contains two LIM domains and belongs to the LIM-only domain family (106). MLP can bind and bundle filamentous actin (F-actin) (107) and within the Z-line it binds to telethonin/T-cap (97, 108), α-actinin (109), cofilin-2 (110), calcineurin (111), and HDAC4 (112) as well as itself and the MLP-b isoform, containing the NH₂-terminal half LIM domain followed by 22 amino acids (113). At the ICD, MLP interacts with nebulin-related anchoring protein (NRAP) (114), whereas in the costamere it binds to zyxin (109), integrin-linked kinase (ILK) (115), and β1-spectrin (116). Furthermore, MLP has been shown to contain a nuclear localization signal (NLS) and shuttle between the cytoplasm and the nucleus (117), where it interacts with MyoD, myogenin, and myogenic regulatory factor 4 (MRF4) (118). Within the nucleus, MLP is found in its monomeric form, whereas in the cytoskeleton and membrane it is present in its polymeric form (119). Accumulation of monomeric MLP in the nucleus and a decline in MLP oligomerization were reported in human failing hearts, suggesting that reduced nonnuclear MLP may contribute to heart failure (119).

Ablation of MLP in mouse resulted in death of 45–65% of homozygous KO mice within the first 2 wk after birth, whereas the rest survived to adulthood and developed eccentric hypertrophy, characterized by LV dilation, wall thinning, and systolic dysfunction (120). MLP KO mice exhibited interstitial fibrosis and severe disruption of myofibrillar organization (120), including misalignment of Z-lines (120, 121), altered costamere organization (114), abnormal ICD morphology and composition (114, 122), as well as mitochondrial disorganization and loss (123). Furthermore, before the development of DCM, MLP KO mice showed abnormal intracellular calcium handling (112, 121, 124, 125), alterations in passive myocardial properties (126, 127), and defective passive stretch sensing (97). Similarly, human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) deficient for MLP showed typical features of HCM, such as increased cardiomyocyte size, multinucleation, and disorganized sarcomere structure, which progressed to mitochondrial dysfunction, increased reactive oxygen species (ROS) production, and impaired calcium handling, recapitulating heart failure. Importantly, restoration of defective calcium handling with an L-type calcium channel blocker prevented the development of HCM and heart failure (128). While KO of MLP has detrimental effects, cardiac-specific overexpression of MLP in mouse did not affect cardiac morphology and function either under basal conditions or in response to mechanical pressure overload of chronic angiotensin II (ANG II) stimulation (129).

MLP KO mice have been used as a model of DCM in numerous studies aimed at reversing the cardiac pathological phenotype to provide insights into the molecular mechanisms leading to DCM and identify novel therapeutic targets. In particular, ablation of the SR calcium ATPase (SERCA2A) inhibitor phospholamban in MLP KO mice completely rescued the morphological, ultrastructural, molecular, and functional abnormalities of MLP KO mice and prevented the development of DCM (130), suggesting that defective SR calcium cycling plays a critical role in the progression toward heart failure in the MLP model. Transgenic overexpression of a peptide inhibitor of β-adrenergic receptor (β-AR) kinase 1 (βARK1) also prevented the development of DCM and reversed the β-AR desensitization observed in MLP KO mice, whereas overexpression of β₂-AR did not improve the phenotype (131). In a later study, β₂-AR ablation was found to rescue the DCM phenotype of MLP KO mice, whereas β₁-AR ablation worsened the phenotype (132). Notably, β₂-AR ablation improved myocyte shortening and reversed pathological alterations in calcium handling (132). As inhibition of the calcium/calmodulin-dependent protein phosphatase calcineurin has been shown to have beneficial effects on cardiac hypertrophy, the effect of calcineurin inhibition in the MLP KO model of DCM was tested by generation of double-KO mice for the calcineurin Aβ catalytic subunit (111). This resulted in reduced cardiac function and enhanced apoptosis, cardiomyocyte death, and fibrosis, leading to early lethality. In contrast, modest overexpression of activated calcineurin was sufficient to improve systolic function and fibrosis (111). Also, ablation of the ANG II type 1a (AT1a) receptor in MLP KO mice was found to reduce ventricular dilation and improve systolic function but not the impaired calcium handling (133). More recently, ablation of cardiac ankyrin repeat protein (CARP/ANKRD1) or ankyrin repeat domain 2 (ANKRD2) in MLP KO mice, both upregulated in MLP KO mice, was found to rescue the DCM phenotype through a novel mechanism (134). More specifically, MLP was found to directly inhibit protein kinase Cα (PKCα), and its ablation results in upregulation of CARP and Ankrd2, which accumulate at the ICD, leading to the formation of a multiprotein complex composed of CARP/ANKRD2, PKCα, and phospholipase-C β1 (PLCβ1), resulting in detrimental chronic PKCα activation and consequent maladaptive remodeling and DCM. Ablation of CARP or ANKRD2 prevented the formation of the maladaptive signaling complex at the ICD, thus averting the development of DCM in MLP KO mice. Consistent with this mechanism, PKCα deletion or ablation reduced ventricular dilation and restored cardiac function in MLP KO mice (135, 136). It is possible that the other successful strategies for rescuing the DCM phenotype in MLP KO mice may also have affected maladaptive PKCα activation through other pathways.

The MLP p.W4R polymorphism, which has a frequency of ∼0.00222 in the Genome Aggregation Database (gnomAD) exomes, has been associated with both DCM and HCM (96–99, 102, 108), suggesting that the outcome may be affected by the effect of modifier genes, epigenetic factors, different environmental conditions, and/or incomplete penetrance (108). In biochemical studies, the variant was found to reduce the binding of MLP to telethonin, resulting in mislocalization of telethonin from the Z-line in a patient carrying the variant (97, 108). Knockin (KI) mice harboring the MLP p.W4R variant developed late-onset HCM as evidenced by increased LV wall and septal thickness, heart weight-to-body weight ratio, and fractional shortening as well as myofibrillar disarray and interstitial fibrosis (108). MLP was significantly downregulated in both heterozygous and homozygous MLP-W4R KI mice, and although MLP was less abundant in the Z-line, it was increased in the nucleus (108). Other MLP mutations have been studied in in vitro studies. The MLP p.K69L mutation, identified in a patient with DCM and endocardial fibroelastosis and located within the predicted NLS adjacent to the LIM1 domain, was found to affect the subcellular localization of MLP and abrogate the binding to α-actinin (96). Similarly, the MLP p.C58G HCM mutation, located in the LIM1 domain, reduced the binding affinity to both α-actinin and N-RAP (100, 137). In subsequent studies, the MLP p.C58G mutant was found to show reduced stability, suggesting that reduced levels of functional MLP may contribute to the development of cardiomyopathy. This is consistent with the finding that heterozygous MLP KO mice, which have reduced MLP levels, develop heart failure in response to myocardial infarction (111).

Nebulette.

Heterozygous missense mutations in the NEBL gene, encoding nebulette, have been associated with DCM with or without endocardial fibroelastosis, characterized by a thickening of the ventricular endocardium due to an increase in the amount of supporting fibrous and elastic tissue (138). Furthermore, rare heterozygous NEBL missense mutations have been identified in HCM and LVNC patients, without mutations in other known cardiomyopathy genes, suggesting that NEBL mutations may also be associated with HCM and LVNC (139). In addition, a NEBL polymorphism has been linked to idiopathic DCM in the Japanese population (140). Nebulette is a 107-kDa cardiac-specific protein located in the Z-line of the sarcomere. Like its larger skeletal muscle homolog nebulin, it is highly modular in structure, containing an NH₂-terminal glutamine-rich region, 23 35-residue nebulin-like repeats, a serine-rich linker region, and a COOH-terminal Src homology-3 (SH3) domain (141, 142). The nebulette SH3 domain binds to proline-rich regions in myopalladin (143), palladin (143), α-actinin (144), titin (145–147), zyxin (148), N-WASP/WASL (149), and XIRP1/2 (150). In addition, the NH₂-terminal acidic region of nebulette binds to filamin C (151), and the nebulin-like repeats bind to actin (144), α-actinin (144), troponin T (144), tropomyosin (144), and desmin (53). In particular, nebulette was found to be downregulated and show altered organization in desmin KO mice, and a role of nebulette in regulating the interaction between actin and desmin intermediate filaments has been proposed (53). Furthermore, nebulette expression studies in cardiomyocytes suggested a role of nebulette in myofibrillar organization and stabilization of the thin filament (152, 153).

Surprisingly, nebulette KO mice showed normal cardiac morphology and function both under basal conditions and in response to biomechanical stress despite upregulation of cardiac pathological markers and progressive Z-line widening (154). In contrast, a mouse model in which exon 3 has been deleted, resulting in ablation of the first three nebulin-like repeats, developed diastolic dysfunction at 6 mo of age (155). Furthermore, nebulette exon 3-deleted mice showed exercise intolerance and developed eccentric hypertrophy and LV chamber dilation after chronic exercise, associated with disruption of ICDs and mitochondria (155). Tg mice overexpressing four different DCM-related NEBL mutations (p.K60N, p.Q128R, p.G202R, and p.A592E) in cardiomyocytes have also been reported (138). The NEBL p.K60N and p.Q128R mutations caused embryonic lethality, whereas the NEBL p.G202R, and p.A592E mutations led to DCM and systolic dysfunction at 6 mo of age, preceded by altered mechanical behavior by magnetic resonance imaging (decreased torsion in NEBL-G202R Tg hearts and increased twist and untwisting rate in NEBL-A592E Tg hearts), altered calcium handling, changes in the expression and localization of sarcomeric proteins, and ultrastructural abnormalities (138, 156). In particular, mitochondrial abnormalities and increased I-band width were observed in both models, whereas t-tubular enlargement and desmosomal separation at the ICD was observed in NEBL-G202R Tg mice (138, 156). For more insights into the effect of NEBL mutations, KI mice mimicking specific human mutations are needed.

Myopalladin.

Heterozygous missense and nonsense mutations in the MYPN gene, encoding myopalladin, have been associated with DCM, HCM, and RCM (157–160). Furthermore, biallelic MYPN loss-of-function mutations have been linked to nemaline myopathy (161), cap myopathy (162), and congenital myopathy with hanging big toe (163). Myopalladin is a 145-kDa striated muscle-specific protein present in both the nucleus and the sarcomere, where it has a dual localization in the Z-line and I-band (143). Myopalladin contains five immunoglobulin (Ig) domains and a proline-rich region and belongs to the palladin/myopalladin/myotilin family of actin-associated Ig-containing proteins in the Z-line (143, 164, 165). Within the cardiac Z-line, myopalladin interacts with α-actinin (143), nebulette (143), titin (166), and PDZ-LIM family members, including cypher/ZASP, CLP36, ALP, and RIL (76). Furthermore, it binds to the stress-inducible transcriptional cofactor cardiac ankyrin repeat protein (CARP/ANKRD1), which, like myopalladin, is present in both the nucleus and the I-band, where it binds to the titin N2A region (143, 167). Like the other family members, myopalladin also binds and bundles F-actin, preventing actin depolymerization (168). In addition, it interacts with myocardin-associated transcription factors (MRTF-A and MRTF-B), which act as cofactors for serum response factor (SRF), controlling its activity in response to alterations in actin dynamics (168).

Myopalladin KO mice developed a mild form of DCM but showed a maladaptive response to mechanical pressure overload induced by transaortic constriction (TAC) as characterized by LV dilation and severely impaired systolic function, accompanied by fibrosis, increased fetal gene expression, altered calcium handling, and increased ICD fold amplitude (166). Furthermore, Tg mice with cardiomyocyte-specific overexpression of the MYPN p.Y20C polymorphism, which has been linked to both DCM and HCM, have been reported (159). MYPN-Y20C Tg mice developed HCM, associated with overexpression of fetal genes and severely disrupted ICDs together with alterations in the expression and localization of interaction partners and ICD proteins (159). KI mice for the nonsense MYPN p.Q529X mutation associated with RCM in patients have also been described (159, 169). Whereas homozygous MYPN-Q529X KI mice showed no pathological cardiac phenotype, supposedly due to very low expression of the mutant protein, heterozygous MYPN-Q529 KI mice showed diastolic dysfunction with preserved systolic function, cardiac arrhythmias, and interstitial fibrosis, consistent with RCM (169). Furthermore, enlarged t-tubules and widened, more convoluted ICDs were observed together with alterations in myopalladin-interacting proteins and cardiac signaling pathways.

Thin Filament

Actin is the main component of the thin filaments of the sarcomere, which consist of two strands of polymerized globular actin monomers (G-actin) twisted around each other to form double helical filaments (F-actin) (reviewed in Refs. 170, 171). F-actin polymerization is accompanied by ATP hydrolysis, and actin filaments undergo continuous treadmilling by F-actin polymerization and depolymerization. The barbed end of actin is anchored at the Z-line and capped by CapZ, whereas its pointed end reaches the A-band, where it is capped by tropomodulin. Along the actin filament, actin binds to tropomyosin dimers, each spanning seven actin subunits and associated with one troponin complex comprised of the three subunits T, I, and C, named after their tropomyosin binding, inhibitory, and calcium binding activities during muscle contraction. In the absence of calcium, tropomyosin sterically hinders the actomyosin interaction. Upon release of calcium from the SR following excitation, calcium binds to troponin, causing tropomyosin to move over the surface of actin, exposing myosin-binding sites on the actin filament. Myosin heads of the thick filament can then bind to actin subunits and form myosin cross bridges, allowing for the production of force and muscle contraction (172).

Several genes encoding proteins associated with the thin filament have been linked to cardiomyopathies. Mutations in TNNT2 (troponin T2), TNNI3 (troponin I3), TPM1 (tropomyosin 1), TNNC1 (troponin C1), and ACTC1 (α-cardiac actin) have been associated with HCM, DCM, and, less frequently, RCM (28, 173, 174). Furthermore, mutations in ACTC1, TNNT2, TPM1, and TNNI3 have been linked to LVNC (31), and ACTC1 mutations have been associated with atrial septal defects (175). Mutations in the I-band region of titin have also been associated with various forms of cardiomyopathy, such as DCM and, less frequently, ACM, HCM, and RCM (46, 176). Four and a half LIM proteins (FHLs) and cardiac ankyrin repeat proteins (CARPs) bind to the titin N2B and N2A regions in the I-band, respectively and have been associated with various forms of cardiomyopathies. These proteins have been extensively studied in our laboratory and are described in detail below.

Four and a half LIM proteins.

Mutations in the four and a half LIM proteins FHL1 and FHL2, also known as skeletal muscle LIM (SLIM) 1 and 3, have been associated with human cardiomyopathies. In particular, a number of studies have associated mutations in FHL1, positioned on the X chromosome, with HCM (177–184). Furthermore, FHL1 mutations have been linked to various myopathies with cardiac involvement, including Emery–Dreifuss muscular dystrophy (EDMD) with HCM (179) or ACM (185), X-linked recessive distal myopathy with HCM (181), and reducing body myopathy with DCM (186). In addition, FHL1 is upregulated during hypertrophy in both human (187–190) and mouse (191–193). Missense mutations in FHL2 have been associated with HCM (194) and DCM (195). Furthermore, reduced FHL2 transcript levels have been found in heart failure patients and during cardiac hypertrophy in both human (194) and mouse (187, 194). FHL proteins have a molecular mass of 32 kDa and contain four LIM domains and an NH₂-terminal half LIM domain. FHL1 is highly expressed in skeletal muscle and to a lesser extent in heart, whereas FHL2 is nearly exclusively expressed in heart (191). Both proteins are localized at the I-band and M-line of the sarcomere (196, 197) and interact with a wealth of proteins, including structural proteins, ion channels, receptors, and signaling proteins (reviewed in Refs. 198, 199). In particular, both proteins interact with the elastic titin N2B region in the I-band of the heart (193, 196), and FHL1 was shown to be part of a biomechanical stress sensor complex targeting MAPKs (Raf1, MEK2, ERK2) to the titin N2B region and function as a positive regulator of ERK-mediated signaling in response to hypertrophic stimuli (193).

FHL1-deficient mice exhibited normal cardiac morphology and function under basal conditions (193) but developed age-dependent skeletal muscle myopathy, associated with structural changes and decreased life span (200). In response to mechanical pressure overload induced by TAC, FHL1 KO mice showed a blunted hypertrophic response but preserved LV function, as characterized by reduced LV weight-to-body weight ratio, cardiomyocyte cross-sectional area, LV wall thickness, and fetal gene expression compared with WT mice (193). In particular, although WT mice showed reduced systolic function after TAC, the systolic function of FHL1 KO mice was comparable to that of sham-operated mice. Similarly, FHL1 ablation was sufficient to prevent pathological hypertrophy, LV dilation, and systolic dysfunction in Tg mice overexpressing constitutively active Gα_q in cardiomyocytes (201), indicating that FHL1 plays an essential role in Gα_q-mediated hypertrophic signaling (193). Consistent with a role of FHL1 in biomechanical stress sensing, FHL1 KO mice showed reduced ERK1/2 signaling following TAC, loss of stretch-induced hypertrophic signaling, and increased muscle compliance (193, 202). Surprisingly, ablation of FHL1 in a myosin heavy chain (MHC) p.R403Q KI mouse model of cardiac hypertrophy (203), showing upregulation of a long FHL1 isoform and downregulation of FHL2, resulted in the development of a more severe HCM phenotype (187). The authors hypothesized that the detrimental effect could be due to downregulation of FHL2, which is not altered in FHL1 KO mice and is a negative regulator of hypertrophy (204). KI mice for the FHL1 p.W122S mutation associated with X-linked scapuloperoneal myopathy were recently described (205, 206). In contrast to male mice, which developed late-onset, slowly progressive myopathy without cardiac involvement (207), female mice showed no skeletal myopathy but developed late-onset cardiac systolic dysfunction associated with enlarged rectangular nuclei (205). Thus, consistent with the location of FHL1 on the X chromosome, FHL1 mutations can have sex-specific effects.

FHL2 KO mice were initially reported to show normal cardiac function both under basal conditions and in response to mechanical pressure overload by TAC (208). However, in subsequent studies, FHL2 KO mice were demonstrated to show worsened cardiac hypertrophy in response to chronic infusion of the β-adrenergic agonist isoproterenol (ISO) (204, 209). Consequently, FHL2 has been proposed to play a protective role in the heart by inhibiting β-adrenergic signaling and hypertrophy through inhibition of calcineurin (209) and ERK signaling (210), both promoting cardiac hypertrophy. A beneficial effect of FHL2 on the hypertrophic response was demonstrated in cardiomyocyte-specific Rho-associated coiled-coil containing kinase 2 (ROCK2) KO mice, which showed a blunted hypertrophic response to TAC or chronic ANG II infusion associated with upregulation of FHL2 and reduced ERK2 phosphorylation (211). Double heterozygous KO mice for ROCK2 and FHL2 showed a restored hypertrophic response to ANG II and ERK phosphorylation levels, demonstrating that the antihypertrophic response of ROCK2 KO mice is dependent on FHL2 (211). Thus, FHL1 promotes stress-induced cardiac pathological hypertrophy, whereas FHL2 plays a protective role by negatively regulating cardiac hypertrophy.

ANKRD1/cardiac ankyrin repeat protein.

Heterozygous missense mutations in ANKRD1, encoding cardiac ankyrin repeat protein (CARP), have been associated with DCM (212, 213) and HCM (214), with frequencies of ∼2% and 0.8%, respectively. Furthermore, increased CARP levels have been reported in patients with various forms of cardiomyopathy, including DCM, HCM, ACM, and ischemic cardiomyopathy (215–219). CARP is a 36-kDa protein belonging to the muscle ankyrin repeat protein family of stress-inducible proteins, also including ankyrin repeat domain protein 2 (ANKRD2/ARPP) and diabetes-related ankyrin repeat protein (DARP/ANKRD23) (167). All family members contain a NLS, a coiled-coil region, four ankyrin repeats, and a PEST sequence, responsible for their rapid turnover by the ubiquitin-proteasome pathway (167, 220, 221). CARP is highly expressed in the early embryonic heart but downregulated in adult heart (222–224). However, it is highly induced in response to various stress conditions, such as biomechanical stress or stimulation with adrenergic agonists or ANG II, and is part of the fetal gene program that is induced during cardiac remodeling (120, 215, 223, 225–228). Like the other family members, CARP has a dual localization in the nucleus and the sarcomeric I-band, where it binds to the titin N2A region within the elastic I-band region of titin (167, 229, 230). In particular, CARP was found to affect PKA-dependent phosphorylation of the titin N2A region (229, 231, 232) and induce cross-linking of titin and actin myofilaments (233, 234), suggesting a role of CARP in increasing sarcomere stiffness and stability to preserve muscle mechanical performance under conditions of stress, possibly also influencing mechano- and phosphorylation-dependent signaling. A role of CARP in cardiac signaling was suggested by biochemical and cellular studies, where CARP was found to bind to YB-1 (224) and function as a nuclear transcriptional cofactor by negatively regulating the expression of cardiac genes such as Nppa, Myl2, and Tnnc1 (222–224, 235, 236). Furthermore, CARP is able to dimerize (237) and bind to myopalladin (143, 167), desmin (237), calsequestrin 2 (238), talin-1 (213), FHL2 (213), MuRF1 (239), MuRF2 (239), TP53 (240), nucleolin (241), PLCβ1 (242), and 14-3-3 proteins (236). In addition, CARP coprecipitates with GATA4 (243–245) and the p50 subunit of NF-κB (246) as well as affects their activity, but it is unknown whether it directly binds to these proteins.

Surprisingly, CARP KO mice exhibit no basal phenotype and show a normal hypertrophic response to mechanical pressure overload induced by TAC (245, 247). However, they show complete attenuation of phenylephrine-induced cardiac hypertrophy (245). Furthermore, CARP ablation rescued the DCM phenotype of MLP KO mice by preventing PKCα accumulation at the ICD (see CSRP3/muscle LIM protein) (242). In-line with this, CARP was reported to enhance ANG II-induced cardiomyocyte hypertrophy and apoptosis in vitro, and myocardial injection of adenovirus expressing CARP after TAC led to increased cardiac hypertrophy and apoptosis, associated with exacerbated cardiac dysfunction, whereas CARP knockdown inhibited TAC-induced hypertrophy and apoptosis (248, 249). Similarly, the selective ANG II inhibitor olmesartan prevented ANG II-stimulated upregulation of CARP in vitro and likewise reduced CARP upregulation and cardiac hypertrophy in mice subjected to TAC (249). In contrast, Tg mice overexpressing CARP in cardiomyocytes were reported to exhibit no basal cardiac pathological phenotype and show attenuated TAC- and isoproterenol (ISO)-induced hypertrophy through inhibition of Erk1/2 and transforming growth factor (TGF)-β signaling (250). Thus, the role of CARP in cardiac hypertrophy remains controversial, although the discrepancy may be explained by the difference between permanent versus transient CARP overexpression. More recently, CARP Tg mice generated by another group were reported to be born with a sinus venosus defect as a result of impaired remodeling of the early embryonic heart (251). Consequently, adult CARP Tg mice showed diastolic dysfunction with preserved ejection fraction, which developed into heart failure at ∼10 mo of age, associated with impaired relaxation, left atrial enlargement, sarcomeric disorganization and loss, mitochondrial damage, and accumulation of lipid droplets (251). The development of sinus venosus defect in CARP Tg mice is consistent with the identification of ANKRD1 as a candidate gene for total anomalous pulmonary venous return (TAPVR), which has been associated with increased ANKRD1 transcript levels (252). Because of the late development of gross cardiac abnormalities in CARP Tg mice, it is possible that the basal phenotype may have been missed in the previously published study (250).

As of yet no mouse models for specific ANKRD1 mutations have been generated. However, various ANKRD1 mutations associated with DCM and HCM have been studied in vitro. ANKRD1 DCM mutations were found to have various effects dependent on their location, including reduced binding to talin-1 and/or FHL2, blunted Mlc2v repressor activity, decreased CARP-mediated inhibition of phenylephrine-induced cardiomyocyte hypertrophy, and altered stretch-induced signaling, whereas CARP localization was unaffected (212, 213). On the other hand, ANKRD1 HCM mutations were reported to increase the binding affinity of CARP for titin and MYPN (titin mutations at the CARP-binding site likewise increased binding affinity), cause nuclear translocation of CARP, and affect cardiac contractile parameters (214, 253). This is consistent with the general proposed mechanism, which seems to apply to many sarcomeric genes, that gain-of-function mutations result in HCM, whereas loss-of-function mutations cause reduced contractility and DCM (254).

Thick Filament

The bipolar thick filament is formed by myosin molecules positioned with their globular heads facing outward and their rod regions facing inward (reviewed in Refs, 255, 256). Myosin is a hexamer consisting of two myosin heavy chains, two essential light chains, and two regulatory light chains. The heavy chains are organized into three structural domains: the globular head domain containing actin and ATP binding sites, the lever arm domain containing essential and regulatory light chains, and the tail region containing an α-helical coiled-coil region required for dimerization to form rods. During muscle contraction the myosin heads attach to opposite oriented thin filaments and pull them toward the center of the sarcomere. Within the central region (C zone) of the A-band, myosin-binding protein-C (MyBP-C3) binds to myosin along the filament, while simultaneously interacting with titin and actin, forming 7–11 transverse stripes (257) (reviewed in Ref. 258). MyBP-C3 plays an important role in stabilizing the thick filament and modulating the actomyosin interaction.

Mutations in MYBPC3 (cardiac myosin-binding protein C) are responsible for 40–50% of familial HCM cases and have also been associated with DCM, LVNC, and RCM (reviewed and summarized in Refs. 173, 256, 259). Mutations in MYH7 (myosin heavy chain 7, also known as β-myosin heavy chain) account for ∼40% of congenital HCM cases and have also been linked to DCM and LVNC. Also, mutations in MYH6 (myosin heavy chain 6, also known as α-myosin heavy chain) have been associated with HCM, DCM, and atrial septal defects. Mutations in MYL2 and MYL3 (myosin light chain 2 and 3) are less frequent causes of HCM, and MYL2 mutations have also in rare cases been associated with DCM (reviewed in Ref. 260). Furthermore, a truncating mutation in MYBPHL (myosin-binding protein-H like) was recently associated with DCM and arrhythmia (261). A high number of mutations in the A-band region of titin have been associated with DCM (listed in Refs. 46, 256), and titin-truncating mutations in the A- and M-line regions have been estimated to be responsible for 25–30% of familial cases of DCM (262, 263). Additionally, a small number of missense mutations have been linked to ACM (46, 264). MYL2, which has been studied in our laboratory, is reviewed in detail below.

Myosin light chain 2.

Heterozygous mutations in MYL2, encoding myosin light chain 2 (MYL2), also known as ventricular myosin light chain 2 (MLC2v), have been associated with HCM (265–281) and, more rarely, DCM (282, 283). Furthermore, a recessive MYL2 mutation caused severe DCM, resulting in infantile death (284), whereas a homozygous frameshift variant, resulting in loss of function, caused HCM, resulting in death before 1 yr of age (285). MYL2 is a 19-kDa protein expressed in adult ventricular and slow-twitch skeletal muscle (286, 287). In the heart, myosin is composed of two heavy chains, two essential light chains in the NH₂-terminal half of the neck, and two regulatory light chains (MYL2) at the neck/tail junction (288).

Ablation of MYL2 in mouse resulted in embryonic lethality by E12.5, associated with myofibrillar disorganization and sarcomere misalignment, cardiac dilation, and severe systolic dysfunction despite compensatory upregulation of the cardiac atrial myosin light chain 2 (MYL7/MLC2a) isoform, which is mainly expressed during embryonic development and normally present only in the atria of the adult heart (71, 287). Interestingly, heterozygous MYL2 KO mice showed normal MYL2 protein expression levels despite a 50% reduction at the mRNA level and consequently showed no pathological phenotype either under basal conditions or after TAC (289). The function of MYL2 is regulated by phosphorylation by cardiac myosin light chain kinase 3 (MYLK3/cMLCK) at Ser15 in human and Ser14/Ser15 in mouse (reviewed in Refs. 290, 291). In particular, through analysis of a nonphosphorylatable MYL2-S14A/S15A KI mouse model and computational modeling (292), MYL2 phosphorylation was found to regulate cross-bridge cycling kinetics and fine-tune the myofilament calcium sensitivity to force by increasing myosin lever arm stiffness and myosin head diffusion, thereby slowing down myosin cycling kinetics and prolonging the cross-bridge duty cycle. MYL2-S14A/S15A KI mice showed an early increase in the rate of cardiac muscle twitch relaxation and reduction in ventricular peak torsion preceding the development of a DCM phenotype and premature death (292). Furthermore, MYL2-S14A/S15A KI mice showed a maladaptive response to TAC-induced pressure overload, to which they showed an eccentric rather than concentric response, resulting in severe systolic dysfunction (292). MYLK3 KO and hypomorphic mice, showing loss or reduction of MYL2 phosphorylation, developed similar phenotypes (293, 294), whereas Tg mice overexpressing MYLK3 specifically in cardiomyocytes showed an attenuated response to TAC (294). Similarly, Tg mice overexpressing myosin phosphatase 2 in cardiomyocytes, resulting in reduced MYL2 phosphorylation, likewise developed a DCM phenotype (295). On the other hand, Tg mice with cardiomyocyte-specific overexpression of the MYL2-S14A/S15A mutant showed atrial defects, but no DCM phenotype, as well as a blunted contractile response to β-adrenergic stimulation (296, 297), likely due to the different strategies used and different levels of MYL2 dephosphorylation obtained. Consistent with a role of MYL2 dephosphorylation in DCM, MYL2 dephosphorylation has been reported in human patients with DCM and heart failure (298–302), and reduced MYL2 protein levels as a result of proteinase-mediated cleavage have been found in DCM patients (303). Furthermore, MYL2 dephosphorylation has been reported in patients with a rare form of familial HCM carrying specific MYL2 mutations (281, 304).

A number of Tg mouse-lines overexpressing specific MYL2 disease mutants in cardiomyocytes have been extensively characterized. Tg mice overexpressing the human MYL2 p.E22K mutant associated with HCM were reported to show mildly enlarged interventricular septa and papillary muscles (305), whereas Tg mice overexpressing the corresponding mouse mutant did now show any sign of hypertrophy (306). Functional analyses of MYL2-E22K Tg mice showed increased calcium sensitivity of myofibrillar ATPase activity and force (305). The mutation had no effect on cross-bridge kinetics but caused a reduction in maximal force and ATPase in skinned fibers as well as decreased magnitude and duration of force and calcium transients in electrically stimulated muscle fibers (307, 308). On the other hand, prolonged calcium transients were reported in electrically stimulated muscle fibers from Tg mice expressing the human MYL2 p.N47K and p.R58Q HCM mutants (309, 310). In addition, MYL2-R58Q Tg mice showed a prolonged force transient, decreased cross-bridge kinetics, as well as increased calcium sensitivity of ATPase and steady-state force. MYL2-N47K and MYL2-R58Q Tg mice did not develop HCM, but MYL2-R58Q Tg mice showed reduced MYL2 phosphorylation and impaired relaxation and diastolic dysfunction (265, 274). In a subsequent study, adeno-associated virus 9 (AAV9)-mediated specific RNA silencing in MYL2-N47K Tg mice, reducing expression of the mutated allele, was shown to ameliorate the disease phenotype by reducing fetal gene expression and partially restore contraction, relaxation, and calcium kinetics (311). As in MYL2-R58Q Tg mice, Tg mice overexpressing the human MYL2 p.D166V HCM mutant showed decreased MYL2 phosphorylation, delayed muscle relaxation, reduced maximal myofibrillar ATPase and force, and increased calcium sensitivity of contractile force (312), which could be reversed by MYLK-induced phosphorylation of myofibrils (313). Constitutive MYL2 phosphorylation, obtained by generation of Tg mice expressing the MYL2 p.D166V mutant with constitutive phosphorylated Ser15 (MYL2-S15D-D166V), was subsequently shown to reverse the functional effects of the D166V mutation and prevent the development of HCM (314). Similarly, AAV9-mediated delivery of MYL2-S15D to MYL2-R58Q Tg mice improved cardiac performance, suggesting its therapeutic potential (315). A Tg mouse model overexpressing the human MYL2 p.K104E HCM mutant showed diastolic dysfunction before the development of late-onset HCM and fibrosis, associated with reduced MYL2 phosphorylation, reduced maximum tension, impaired muscle relaxation, and inefficient energy use (316). A subsequent gene expression profiling study on MYL2-R58Q, MYL2-D166V, and MYL2-K104E Tg mice showed distinct gene expression patterns, suggesting that the three HCM mutations lead to HCM through different mechanisms (317). A Tg mouse overexpressing the human MYL2 p.D94A mutant associated with DCM was also recently reported (318). MYL2-D94A Tg mice developed DCM, mimicking the clinical phenotype. This was associated with structural alterations in the myosin head affecting its interaction with the actin filament, resulting in aberrant cross-bridge cycling and reduced calcium sensitivity of force, ultimately leading to systolic dysfunction (318). The effects of the MYL2 p.R633H (319) and p.R58Q (320) HCM mutations have also been studied in patient-specific hiPSC-CMs. HiPSC-CMs carrying the MYL2 p.R58Q mutation, which has been extensively studied in Tg mice, showed cardiomyocyte hypertrophy, myofibrillar disarray, irregular beating, decreased calcium transients, and reduced LTCC peak current, recapitulating the human HCM phenotype (320). Similarly, hiPSC-CMs carrying the MYL2 p.R633H mutation showed increased cardiomyocyte size, upregulation of hypertrophic marker genes, arrhythmia at the single-cell level, dysregulation of calcium cycling, and elevation of intracellular calcium, recapitulating common features of HCM (319). Importantly, pharmacological restoration of calcium homeostasis prevented the HCM phenotype, suggesting that patient-derived hiPSC-CMs may provide a useful system for testing of novel therapies. Overall, although the extensive characterization of Tg mice overexpressing human MYL2 mutants associated with disease has provided important insights into the disease mechanisms leading from MYL2 mutations to cardiomyopathy, the majority of the models do not recapitulate the human HCM phenotypes. In the future, the generation of KI models, which more closely mimic the human condition, would be helpful to better understand the disease mechanisms and provide more appropriate models for the testing of potential therapies.

M-Line

The M-line is located at the center of the A-band, where the rods of antiparallel myosin arrays and the COOH termini of titin from each half-sarcomere overlap (reviewed in Refs. 268, 321, 322). Furthermore, the M-line contains the three structurally related proteins, myomesin 1, myomesin 2/M-protein, and myomesin 3, that form antiparallel dimers interacting with myosin and titin, thereby cross-linking adjacent myosin and titin filaments (M bridges) to maintain thick filament alignment and stabilizing the sarcomere during contraction. The M-line also contains other proteins, such as muscle creatine kinase (CKM), obscurin, obscurin-like 1, MyBP-C1, FHLs, muscle-specific RING finger proteins (MURFs), ankyrins, and spectrins, implicating it in multiple cellular processes, including signal transduction, mechanosensing, metabolism, and protein turnover.

Patients homozygous or compound heterozygous for truncating mutations in the COOH-terminal M-line portion of titin have been reported to develop skeletal myopathy with DCM and/or other cardiac disorders, such as LVNC, septal defects, and cardiac arrhythmia, resulting in early cardiac death (323, 324), whereas heterozygous missense mutations in the COOH-terminal region of titin have been associated with various forms of skeletal myopathy with no cardiac involvement. Furthermore, a mutation in MYOM1 (myomesin 1) has been associated with HCM (325), and a potential link of a MYOM3 (myomesin 3)-truncating mutation to DCM was proposed (326). Additionally, increased levels of the embryonic heart myomesin (EH-myomesin) isoform were reported in myocardial biopsies from DCM patients (327). Mutations in OBSCN (obscurin) have been linked to various cardiomyopathies (summarized in Ref. 256), and the current knowledge on obscurin is summarized below.

Obscurin.

Missense and frameshift mutations in the OBSCN gene, encoding obscurin, have been associated with HCM (328, 329), DCM (330, 331), LVNC (331), and ACM (332). Furthermore, obscurin isoform switching was found in the LV of DCM patients (333), and increased obscurin transcript levels were reported in a canine model of tachycardia-induced DCM (334) as well as in mice subjected to mechanical pressure overload by TAC (335). Obscurin is a myosin light chain kinase belonging to a family of three related proteins, including striated preferentially expressed gene (SPEG) and obscurin-like 1 (OBSL1). In striated muscle, several obscurin isoforms are expressed, including two large isoforms designated obscurin-A (720 kDa) and obscurin-B (870–970 kDa), which have different COOH termini. Obscurin-A and obscurin-B share ∼70 Ig-like domains, 2–3 fibronectin III (FN3)-like domains, a calmodulin IQ-binding motif, a SRC homology 3 (SH3) domain, a rho-guanine nucleotide exchange factor (RhoGEF) domain, and a pleckstrin homology (PH) domain. However, whereas the COOH-terminal region of obscurin-A, the dominant isoform in the heart, comprises a nonmodular region and several potential phosphorylation sites, the COOH-terminal region of obscurin-B contains two Ser/Thr kinase-like domains, two additional Ig domains, and a FN3 domain (336–339). In addition, two shorter isoforms with predominant expression in the heart have been characterized, corresponding to the COOH-terminal region of obscurin-B and containing one [“single” obscurin-myosin light chain kinase (s-MLCK), 55–70 kDa] or two [“tandem” obscurin-myosin light chain kinase (t-MLCK), 120–150 kDa] kinase-like domains (336, 339, 340). Two small isoforms, obscurin-40 (40 kDa) and obscurin-80 (80 kDa), containing the PH domain and localized at the ICD, were also recently described and shown to be reduced in mouse models of pressure overload and myocardial infarction (341). Obscurin is present in different locations of the cell, including the M-line, Z-line, and A/I junction of the sarcomere as well as the costamere, ICD, and SR (reviewed in Ref. 342). At the M-line, obscurin interacts with titin (M10) (343), myomesin (343), RhoA (344), and slow skeletal myosin-binding protein C (MyBPC1) (345), whereas at the Z-line it interacts with titin (Z9-Z10 and the novex-3 titin isoform) (338, 346) and RAN binding protein 9 (RanBP9) (347). Furthermore, the obscurin-A isoform binds to small ankyrins (ANK1, ANK2) at the M-line and SR (348–351), whereas the kinase domains of the obscurin-B isoform can bind to Na⁺-K⁺-ATPase (ATP1A1) and phosphorylate N-cadherin at the ICD (352). Additionally, the PH domain present in the two recently identified obscurin-40 and obscurin-80 isoforms binds to phosphatidylinositol phosphates [phosphatidylinositol 4,5-bisphosphate (PIP₂)] (341). Although the exact function of obscurin is still not well understood, it is thought to be involved in myofibril assembly and provide structural stability by connecting sarcomeric complexes to the SR and sarcolemma, potentially also implicating it in mechanosensing and signaling (reviewed in Refs. 321, 342, 353).

Obscurin KO mice showed normal sarcomeric organization and cardiac and skeletal muscle function but exhibited altered longitudinal SR architecture in skeletal muscle as well as disrupted sANK1.5 expression and localization in both cardiac and skeletal muscle (354, 355). Furthermore, obscurin KO mice showed reduced exercise performance depending on age and exercise intensity, which was associated with ultrastructural abnormalities in the diaphragm (349, 356). The mild phenotype of obscurin KO mice was thought to be due to the presence of nontargeted COOH-terminal isoforms and/or compensation by its homolog obscurin-like 1. Characterization of double skeletal muscle KO mice for obscurin and obscurin-like 1 mice confirmed the redundant role of the two proteins in sarcolemmal integrity, SR organization, and muscle metabolism (357), although it remains to be determined whether they play redundant roles also in the heart. More recently, homozygous KI mice for the OBSCN p.R4344Q variant in obscurin Ig58 associated with HCM were described (358). OBSCN-R4344Q KI mice showed no obvious morphological and functional abnormalities in the heart, but at 1 yr of age OBSCN-R4344Q KI mice displayed increased SR calcium content, faster contraction and relaxation kinetics, and spontaneous ventricular arrhythmia, associated with upregulation and overactivation of SERCA2. Furthermore, OBSCN-R4344Q KI mice showed a maladaptive response to TAC-induced pressure overload, after which they developed DCM and interstitial fibrosis (358). The phenotype was thought to be attributed to increased affinity of obscurin Ig58 for its newly identified binding partner phospholamban (SERCA2 inhibitor), which was found to be reduced in OBSCN-R4344Q KI hearts. This was subsequently challenged by Fukuzawa et al. (359), who confirmed the binding in vitro but found no interaction or colocalization between the two proteins in cotransfected cells and thus suggested that the interaction is an in vitro artifact. They also did not confirm the previously reported impaired binding of the OBSCN p.R4344W mutant to titin Z9-Z10 (328). Consistent with these considerations, the pathogenicity of the OBSCN p.R4344Q variant has been questioned, as it has been identified in up to 15% of Black Americans and found to have an overall allele frequency of 0.01165 (330, 360). Nevertheless, the importance of the obscurin Ig58/59 region was recently demonstrated in a mutant mouse model expressing obscurin lacking Ig58/59 (OBSCN-ΔIg58/59) (361). Young OBSCN-ΔIg58/59 mice exhibited no basal phenotype but showed cardiac arrhythmia after acute β-adrenergic stimulation. At 6 mo of age, OBSCN-ΔIg58/59 mice showed LV hypertrophy, which developed into DCM with atrial enlargement, systolic dysfunction, and severe arrhythmia by 12 mo of age. This was associated with abnormal calcium handling and altered expression and phosphorylation of SR proteins, suggesting a role of the obscurin Ig58/59 region in calcium handling. The stronger pathological phenotype compared with obscurin KO mice suggests that the Ig58/59 deletion has a dominant negative effect, while proteins with redundant functions, such as obscurin-like 1, may compensate for the complete absence of obscurin. A recently identified frameshift mutation in OBSCN (L5218fs) associated with ACM was recently studied in patient-derived hiPSC-CMs and found to cause lipid accumulation, increased pleomorphism, Z-line irregularities, increased L-type calcium currents, mislocalization and reduced expression of ANK1.5, downregulation of desmosomal genes, and upregulation of N-cadherin and adipogenesis pathway-related genes (332). This is consistent with fibrofatty replacement in the myocardium of ACM patients and may provide a molecular basis for the development of arrhythmia in ACM patients carrying the frameshift mutation.

INTERCALATED DISK

The intercalated disk (ICD) is a highly organized structure connecting adjacent cardiomyocytes to each other, essential for synchronized contraction and the maintenance of structural integrity (reviewed in Refs. 5–7). The ICD is composed of three major junctional complexes: desmosomes, which link the cell membrane to the intermediate filament desmin; fascia adherens junctions, which connect the cell membrane to the actin filament, facilitating the transmission of contractile force between cardiomyocytes; and gap junctions, which allow for the passage of ions and small molecules between neighboring cardiomyocytes, enabling electrical and metabolic coupling between cells. Rather than functioning as distinct units, the three junctions collaborate through interaction between their components in mixed junctions termed “area composita,” facilitating mechanical and electrical coupling as well as providing structural stability.

Mutations in many components of the ICD have been associated with human disease. Mutations in genes encoding desmosomal proteins are responsible for 85–90% of familial cases of ACM, including PKP2 (plakophilin-2), DSP (desmoplakin), DSG2 (desmoglein-2), DSC2 (desmocollin-2), and JUP (plakoglobin/γ-catenin) (reviewed in Ref. 37). Furthermore, DSP mutations have been linked to DCM with woolly hair and keratoderma (362), PKP2 mutations have been associated with catecholaminergic polymorphic ventricular tachycardia (363), and a polymorphism in DSG2 has been associated with increased risk of DCM (364). Also mutations in genes encoding other ICD-associated proteins have been linked to ACM, including CTNNA3 (α-T-catenin), CDH2 (N-cadherin), DES (desmin), TJP1 (tight junction protein-1/zonula occludens-1), SCN5A (voltage-gated sodium channel alpha subunit 5), and TMEM43 (transmembrane protein 43) (reviewed in Ref. 37). Furthermore, DES mutations have been associated with DCM and, less frequently, HCM, RCM, and LVNC (reviewed in Ref. 365). In animal models, genetic modification of the fascia adherens junction proteins N-cadherin (366–368) and α-E-catenin (366, 369) resulted in DCM, whereas ablation of xin actin binding repeat containing 1 (XIRP1) caused HCM (370). However, as of yet, mutations in these proteins have not been associated with human DCM or HCM. No direct link between mutations in gap junction proteins and human disease has been established, although a potential link between mutations in the gap junction protein connexin 43 (GLA1) and sudden cardiac death has been reported (371, 372). The ICD proteins plakoglobin, coxsackievirus and adenovirus receptor (CAR), and zonula occludens-1 (ZO-1) have been studied in our laboratory and are reviewed in detail below.

Plakoglobin

Recessive missense mutations or deletions in the JUP gene, encoding junction plakoglobin (hereafter referred to as plakoglobin), also known as γ-catenin, have been shown to be causative for Naxos disease (373–381), a recessive form of ACM with palmoplantar keratoderma and woolly hair, whereas dominant missense mutations or insertions/deletions have mostly been associated with ACM alone (382–388). Furthermore, reduced plakoglobin localization at the ICD in ACM patients regardless of the causative mutation has been reported (389, 390). Plakoglobin is an 82-kDa protein belonging to the armadillo protein family and a close homolog of β-catenin. Although β-catenin is present only in the fascia adherens junction, plakoglobin is present both in the desmosome and the fascia adherens junction of the ICD (reviewed in Ref. 391). At the fascia adherens junction, plakoglobin binds to N-cadherin (392, 393) and α-catenin (393–395), indirectly linking cadherin to the actin cytoskeleton, whereas at the desmosome it binds to desmoglein (396), desmocollin (397), and desmoplakin (398).

Global KO of plakoglobin resulted in embryonic lethality from E10.5 due to severe cardiac dysfunction, associated with the absence of desmosomes and alterations in the fascia adherens junctions, often resulting in cardiac rupture (399, 400). Furthermore, some animals with a different genetic background, surviving until around the time of birth, showed skin blistering and subcorneal acantholysis (400). Myofibers from KO embryos were found to be less compliant, suggesting that decreased passive compliance may be responsible for the ICD alterations (401). Heterozygous plakoglobin KO mice developed late-onset ACM, characterized by right ventricular dilation and spontaneous ventricular tachycardia, which was accelerated by endurance exercise (402). In particular, heterozygous mice showed right ventricular dilation and spontaneous ventricular tachycardia but no fibrofatty replacement or ICD abnormalities. Plakoglobin hypomorphic mice with an ∼60% reduction in plakoglobin expression have also been described and demonstrated to show increased β-catenin expression but otherwise normal ICD structure and cardiac function after the postnatal period, when ∼50% of the hypomorphic mice died (403). Conditional ablation of plakoglobin in the heart resulted in a more severe ACM phenotype, characterized by dilation of the right atrium and ventricles, ventricular aneurisms, severe cardiac fibrosis, cardiomyocyte death, cardiomyocyte hypertrophy, loss of desmosomes, systolic dysfunction, and spontaneous ventricular arrhythmias, resulting in sudden death starting from 1 mo of age with an average life span of 4.6 mo (404). However, in contrast to human ACM patients, no adipose tissue deposition was found. Consistent with the loss of desmosomes, desmoglein-2, an interaction partner of plakoglobin, was downregulated, whereas upregulation of β-catenin was thought to compensate for the loss of plakoglobin at the fascia adherens junction without affecting Wnt/β-catenin-mediated signaling. A similar mouse model generated by another group showed a comparable phenotype and was used in a study demonstrating that adrenergic stimulation increases cardiac myocyte cohesion, referred to as positive adhesiotropy, via PKA-mediated phosphorylation of plakoglobin at Ser665 (405). Furthermore, positive adhesiotropy was found to be associated with ultrastructural strengthening of the ICD via plakoglobin (406). Consequently, in response to β-adrenergic stimulation, plakoglobin-deficient mice showed defective myocyte cohesion and ultrastructural reorganization as well as a blunted inotropic and chronotropic response (405, 406). The same group recently demonstrated that stabilization of desmoglein-2 interactions by a specific linking peptide is sufficient to acutely rescue cardiac arrhythmia in plakoglobin KO mice as well as restore dysfunctional conduction of excitation as a result of impaired desmosome integrity (407). Inducible cardiac-specific KO of plakoglobin in adult mice resulted in a milder cardiac phenotype, starting from ∼5 mo after induction, characterized by ventricular dilation, progressive loss of cardiomyocytes, inflammatory infiltration, interstitial fibrosis, cardiomyocyte hypertrophy, upregulation of hypertrophic genes, and cardiac dysfunction (408). However, no cardiac arrhythmia could be induced despite a decreased number of desmosomes, which also showed structural alterations as well as reduced expression of connexin 43 at the gap junction. As β-catenin and β-catenin-mediated signaling were upregulated and an increased association between β-catenin and connexin 43 was found, it was hypothesized that upregulation of β-catenin can partially compensate for the absence of plakoglobin and protects from cardiac arrhythmia and sudden cardiac death (408, 409). This was confirmed in cardiac-specific double-KO mice deficient for plakoglobin and β-catenin induced at adult stage, which showed severe conduction abnormalities and spontaneous arrhythmia, resulting in sudden cardiac death between 3 and 5 mo after deletion (409) [β-catenin KO mice show no cardiac abnormalities presumably due to upregulation of plakoglobin (410)]. Consistently, double-KO mice showed loss of ICD structures, associated with a dramatic reduction in fascia adherens junction and desmosomal proteins as well as gap junction remodeling, occurring before the onset of arrhythmia.

Whereas there were no signs of adipose deposition in any of the reported plakoglobin KO mice, Tg mice overexpressing plakoglobin in cardiomyocytes showed fibrofatty infiltration and fibrosis but otherwise normal desmosome structure and cardiac function (411). The plakoglobin transgene was targeted to the nucleus, where it was found to bind to the TCF712 transcription factor and suppress Wnt/β-catenin signaling, resulting in upregulation of adipogenic factors normally inhibited by canonical Wnt signaling. Tg mice with cardiomyocyte-specific overexpression of truncated plakoglobin, corresponding to a homozygous 2-base pair deletion, causative for Naxos disease (377), showed an amount of fibrofatty infiltration and fibrosis similar to plakoglobin WT Tg mice but demonstrated increased LV weight-to-body weight ratio, LV dilation, and systolic dysfunction (412). Furthermore, although the WT plakoglobin transgene was detected both in the nucleus and the desmosome, mutant plakoglobin was absent from the desmosome and showed reduced binding to the desmosomal proteins desmoplakin and desmoglein-2. To exactly mimic the human Naxos disease mutation, KI mice for the 2-bp deletion were generated (413). KI mice died at postnatal day 1 and showed significant downregulation of plakoglobin due to nonsense-mediated RNA decay, associated with downregulation of desmosomal proteins and upregulation of β-catenin, but otherwise no cardiac abnormalities, similar to the phenotype of plakoglobin hypomorphic mice (403, 413). In contrast, an engineered mouse model expressing mutant plakoglobin at WT levels showed no cardiac morphological or functional alterations, indicating that the clinical phenotype of patients with Naxos disease is due to loss of function and suggesting that increasing levels of truncated or WT plakoglobin may be used as a potential therapeutic approach for Naxos disease (413). No evidence of truncated plakoglobin was found, suggesting that the nuclear localization of plakoglobin in Tg mice was a result of the nonphysiological overexpression.

CAR

CXADR, encoding the coxsackievirus and adenovirus receptor (CAR), has not been directly linked to cardiac disease, but in a genome-wide association study a single-nucleotide polymorphism (SNP) at locus 21q21 (rs2824292) in the vicinity of CXADR has been strongly associated with ventricular fibrillation after acute myocardial infarction, a leading cause of sudden cardiac death (414). The risk allele was associated with lower CXADR mRNA levels in human LV biopsies, suggesting that reduced CAR levels predispose to ischemia-induced ventricular fibrillation (415). In another study, a nonsynonymous SNP (rs437440) in the CXADR gene was associated with elevated systolic and diastolic blood pressure (416). CAR is a 46-kDa transmembrane cell adhesion protein with a well-known role as a viral receptor involved in the pathogenesis of viral myocarditis (417, 418). CAR is strongly expressed during development but downregulated in the adult heart, where it is present only in interstitial cells. However, it was reported to be reexpressed in the ICD and sarcolemma in human DCM (419). Furthermore, increased CAR transcript levels were reported in patients with DCM, ischemic cardiomyopathy, and mitral valve disease (420), and CAR was found to be upregulated in rat models of cardiac injury (421) and inflammation (422). Consistent with its location at the ICD, CAR has been reported to interact with ZO-1, β-catenin, and connexin 45 (413, 423, 424).

Global KO of CAR resulted in embryonic lethality at ∼E11.5–E13.5, as independently demonstrated by three different groups using slightly different targeting approaches (425–427). In all three mouse-lines, cardiac embryonic development was delayed. Asher et al. (427) reported myocardial apoptosis and wall thinning, resulting in myocardial rupture and thoracic hemorrhage. In contract, the other two groups did not observe apoptosis or hemorrhaging (425, 426), although Dorner et al. (426) reported more frequent hemorrhaging during embryo preparation. Both Dorner et al. (426) and Chen et al. (425) reported reduced myofibrillar organization, thickness, and density as well as dilation of cardiac veins. However, although Dorner et al. found no alterations in proliferation (426), Chen et al. reported increased cardiomyocyte proliferation, resulting in hyperplasia of the LV (425). Furthermore, abnormal junctions between LV cardiomyocytes were observed. Chen et al. also found absence of the sinoatrial valve (425), whereas Dorner et al. reported enlarged endocardial cushions and abnormal atrioventricular canal formation, consistent with a developmental delay (426). The different phenotypes of the mice were suggested to be related to different genetic backgrounds and targeting strategies (425). Chen et al. (425) also generated two cardiomyocyte-specific CAR KO mouse-lines by breeding Car floxed mice with Tnnt2-Cre mice and Myh6-Cre mice. Cardiomyocyte-specific CAR ablation by E9.5 (using Tnnt2-Cre mice) (425, 428) resulted in embryonic lethality and a phenotype similar to global CAR KO mice, whereas mice with cardiomyocyte-specific CAR ablation from ∼E11 (using Myh6-Cre mice) (425) survived to adulthood and showed no obvious cardiac abnormalities. This suggests that CAR is essential for normal cardiac development at a specific temporal window, after which it is no longer required for survival. Both global CAR KO mice and cardiomyocyte-specific KO mice (using Myh6-Cre mice) were later generated by Lim et al. (429). Consistent with previous reports, global KO mice died embryonically, often with hemorrhage and pericardial effusion, whereas no evidence of structural defects, hypertrophy, or ventricular wall thinning was found (429). Cardiomyocyte-specific KO mice were viable but showed late-onset cardiomyopathy starting from ∼5 mo of age, characterized by reduced systolic function, fibrosis, and ICD alterations but otherwise normal cardiac dimensions. This was preceded by a first-degree or complete block of atrioventricular (AV) conduction as well as loss of connexin 45 from the cell-cell junctions of the AV node, where CAR is normally present. Furthermore, reduced expression levels and localization of β-catenin and ZO-1 at the ICD were observed. Similarly, prolonged PR intervals were observed in the embryonic heart of global CAR KO mice, consistent with a first-degree AV block. In another study, cardiomyocyte-specific KO mice generated with Myh6-Cre mice were found to be embryonic lethal, so to circumvent the embryonic lethality inducible cardiomyocyte-specific CAR KO mice were generated (430). Similar to observations by Lim et al., increasing prolongation of AV conduction was observed from 2 wk after induction, resulting in a complete AV block by 6 wk after induction (430). This was often associated with altered sinus node function, as indicated by accelerated junctional rhythms and the presence of sinus node tachycardia and bradycardia in some mice. Furthermore, reduced protein levels of connexin 43 and 45 were found. Similarly, a complete AV block was reported in global inducible CAR KO mice 24 wk after induction, which also showed ICD abnormalities, including a disconnection between myofilaments, the presence of big vacuoles, and wider fascia adherens junctions (431). Since elevated CAR expression has been reported in heart failure patients (419, 420), Tg mice overexpressing CAR in cardiomyocytes were studied (432). Tg mice died before 4 wk after birth with enlarged, misshaped, and disorganized cardiomyocytes; increased heart size; dilated ventricles; as well as ICD alterations, including disrupted fascia adherens junctions. This was associated with altered N-cadherin expression as well as translocation of β-catenin to the nucleus and consequent activation of c-Myc, a major downstream target of β-catenin involved in cardiac hypertrophy.

ZO-1

Missense mutations in the TJP1 gene, encoding zonula occludens-1 (ZO-1), also known as tight junction protein 1 (TJP1), have recently been linked to ACM (433), although further studies are required to establish whether TJP1 is a novel disease gene. ZO-1 is a member of the membrane-associated guanylate kinase (MAGUK) family and is expressed at the tight junction in different cell types and tissues (434). In the heart, ZO-1 is localized at the adherens and gap junctions of the ICD (7, 435–437). ZO-1 is a 220-kDa protein, containing three PDZ domains, a proline-rich SH3-domain binding region, and a catalytically inactive guanylate kinase domain. In the heart, ZO-1 interacts with F-actin (438–440) as well as several proteins at the ICD, including α-catenin (438), connexin 43 (435, 441, 442), connexin 45 (443, 444), CAR (423), and vinculin (445). In particular, through its link to connexin 43, ZO-1 has been shown to play a role in regulating the localization, size, number, and distribution of gap junctions (437, 446–448). ZO-1 was found to be upregulated in heart failure patients and to show stronger colocalization and increased interaction with connexin 43 (436), based on which a role of ZO-1 in the connexin 43 downregulation and decreased size of gap junctions observed in failing human heart was proposed. In contrast, other studies reported reduced ZO-1 expression and loss of ZO-1 from the ICD in failing human hearts (449, 450), warranting further studies on larger patient groups.

Global KO of ZO-1 in mouse resulted in embryonic lethality by E11.5, associated with developmental delay and severe growth defects from E8.5. Furthermore, KO embryos exhibited extensive apoptosis in the notochord, neural tube area, and allantois as well as extraembryonic defects, including impaired yolk sac angiogenesis and defective chorioallantoic fusion. To circumvent the embryonic lethality, conditional (451) and inducible (452) cardiomyocyte-specific ZO-1 KO mice were generated. Conditional KO of ZO-1 in the heart did not affect ventricular structure, morphology, or function either under basal conditions or in response to mechanical pressure overload. However, atrial mass was increased and cardiomyocyte-specific ZO-1 KO mice exhibited various degrees of high-grade AV block, including complete block. Upregulation of ZO-2 appeared to compensate for ZO-1 deficiency in the ventricles, whereas its expression was unaltered in the AV node. While the expression and localization of ZO-1-associated proteins were not affected in the ventricles, reduced connexin 45 and CAR expression were found in atrial tissue and expression levels of connexin 45 and the cardiac sodium channel Na_v1.5 were reduced in AV nodal cells. As in conditional cardiomyocyte-specific ZO-1 KO mice (451), induction of ZO-1 KO in adult mice resulted in AV block. This was associated with reduced expression and localization of connexin 40 and CAR at the ICD of the AV node (452). Furthermore, in contrast to observations in conditional ZO-1 mice, the localization of connexin 43 and CAR at the ICD in ventricular cardiomyocytes was decreased and a modest reduction in systolic function in the absence of histological abnormalities was observed from 10 days after induction. Analysis of inducible conduction system-specific (using Hcn4-Cre) and AV bundle-His-Purkinje-specific (using Kcne1-Cre) ZO-1 KO mice revealed that ZO-1 expression in the conduction system proximal to the His bundle is required for conduction and that myocardial dysfunction is unrelated to the conduction system defects in cardiomyocyte-specific ZO-1 KO mice (452). The relevance of ZO-1 in the human AV node was demonstrated by the colocalization of ZO-1 with connexin 40 in ICDs of the AV node and atria but not ventricles in human biopsies.

COSTAMERE

Costameres are striated muscle-specific structures connecting the contractile apparatus to the sarcolemma, structurally and functionally resembling focal adhesions (56, 453, 454). Costameres physically link Z-lines to the extracellular matrix (ECM) and act as major mechanotransduction hubs, bidirectionally transmitting force signals between the sarcomere and the ECM. The dystrophin-glycoprotein (DGC) and vinculin-talin-integrin complexes are the two major components of costameres. Core components of the DGC include extracellular (α-dystroglycan), transmembrane (sarcoglycans, β-dystroglycan, sarcospan), and cytosolic (dystrophin, dystrobrevin, syntrophins) compartments (reviewed in Refs. 455–457). Dystrophin is a key component, physically connecting sarcomeres to the DGC via its bivalent interaction with β-dystroglycan (458) and actin filaments (459, 460). Moreover, through interaction with ankyrins, dystrophin connects costameres to the spectrin-based filament network (461). The vinculin-talin-kindlin-integrin complex tethers membrane-spanning integrins to the actin cytoskeleton via several interconnected mechanisms (see Fig. 1) (462–465). Talin connects integrins to the actin cytoskeleton via vinculin-mediated interaction with α-actinin and recruits focal adhesion kinase (FAK) to mediate downstream integrin signaling. Moreover, integrins are connected to the actin cytoskeleton via the ILK-PINCH-parvin complex (see Fig. 1) (466–470). In addition to the core components, numerous other proteins associate with the DGC and vinculin-talin-integrin complex at the costamere (reviewed in Refs. 55, 56). One of them, filamin C, has been shown to associate both with β₁-integrin and the DGC via interaction with sarcoglycans (reviewed in Ref. 471), proving a link to the Z-line, where it interacts with several proteins.

The clinical relevance of mutations in components of the DGC is exemplified by mutations in DMD (dystrophin), which are causative for X-linked Duchenne and Becker muscular dystrophy, also associated with DCM, which is a main cause of death in these patients (472, 473). A significant prevalence of DCM in sarcoglycanopathy patients, primarily manifesting as skeletal muscle myopathy, has also been reported (474, 475), and mutations in SGCD (δ-sarcoglycan) have also been associated with isolated DCM (476, 477). Furthermore, mutations in DTNA (α-dystrobrevin) have been associated with LVNC (62, 478, 479), and a STNA1 (α1-syntrophin) mutation has been linked to long QT syndrome (480). Also, mutations in genes encoding other components of the costamere have been associated with disease, including VCL (vinculin), associated with DCM and HCM (481), ILK (integrin-linked kinase), linked to DCM and ACM (482, 483), and FLNC (filamin C) (484), associated with various types of cardiomyopathy. Our laboratory has contributed to the understanding of the cardiac functions of vinculin, the ILK-PINCH-parvin complex, and filamin C as reviewed in detail below.

Vinculin

Mutations in the VCL gene, encoding vinculin and metavinculin, have been associated with both DCM (485, 486) and HCM (486–489). Despite low penetrance, VCL loss-of-function variants were enriched in probands with primarily pediatric-onset DCM, indicating that heterozygous loss of function of VCL alone is insufficient to cause cardiomyopathy but that these variants contribute to disease risk (481). Indeed, it was reported that cosegregation of heterozygous variants in VCL and sarcomeric genes, such as TPM1 and MYBPC3, causes or modulates the severity of cardiomyopathy (490, 491). Vinculin is a ubiquitously expressed 117-kDa membrane-associated scaffolding protein present at the costamere and fascia adherens junction of the ICD (492, 493). Vinculin contains a globular head, a flexible proline-rich linker, and a tail and interacts with numerous interaction partners through its three regions. The head domain binds to talin (494–496), α-actinin (497, 498), α-catenin (499, 500), β-catenin (501, 502), and ZO-1 (445); the proline-rich region interacts with vasodilator-stimulated phosphoprotein (VASP) (503), vinexin/SORBS3 (504), ponsin/CAP (505), and the Arp2/3 complex (506); and the tail region binds to phosphatidylinositol 4,5-bisphosphate (PIP₂) (507, 508), paxillin (509, 510), raver1 (511), α-synemin (512), PKCα (513), and F-actin (514–517). Through autoinhibitory interaction between the head and tail domains vinculin is kept in an inactive state, occluding ligand binding, and is believed to get activated through a conformational change induced by synergistic ligand binding, phosphorylation, and/or force (reviewed in Ref. 518). In cardiomyocytes, vinculin links integrins in costameres and cadherins in the fascia adherens junction of the ICD to the actin cytoskeleton through its interactions with talin, catenins, and α-actinin (519–522). Moreover, via interaction with ZO-1, vinculin stabilizes connexin 43-containing gap junctions at the ICD (445). A larger splice isoform of vinculin, named metavinculin (145 kDa), containing a 68-amino acid insertion within the actin-binding tail domain, is specifically expressed in cardiac and smooth muscle (493, 523–526). Both vinculin and metavinculin bind to F-actin, but whereas vinculin is able to bundle F-actin through dimerization, metavinculin inhibits vinculin-mediated F-actin bundling (527, 528) and mutations within the 68-residue insertion in the metavinculin tail domain have consequently been found to promote the formation of large disorganized actin assemblies in vitro (487). Functionally, metavinculin is capable of sustaining higher mechanical forces, but a smaller fraction of metavinculin molecules is engaged in mechanotransduction compared with vinculin (529).

Global loss of vinculin caused embryonically lethality by E8–E10 with severe cardiac hypoplasia evident at E9.5, whereas heterozygous vinculin KO animals were viable and demonstrated haploinsufficiency (530). Heterozygous vinculin KO mice had normal cardiac contractility but showed widened QRS complexes and ICD abnormalities. Furthermore, when subjected to mechanical pressure overload by TAC, they showed increased lethality and systolic dysfunction associated with Z-line misalignment and abnormal myofibril anchorage at the ICD (531). Cardiomyocyte-specific KO of vinculin led to sudden arrhythmic death in ∼49% of KO mice before 3 mo of age due to conduction abnormalities and ventricular tachycardia, whereas surviving mice developed DCM and died by 6 mo of age (532). Ultrastructural analyses showed abnormal fascia adherens junctions and ICD structure, associated with reduced cadherin and β₁-integrin levels as well as laterization of connexin 43 (532). In contrast, global metavinculin-specific KO mice were born at expected Mendelian ratios, exhibited unaltered cardiac structure, and showed a normal hypertrophic response to TAC (529), indicating that metavinculin is not essential for proper cardiac function.

Integrin-Linked Kinase

Missense mutations in the ILK gene, encoding integrin-linked kinase (ILK), a central component of the costameric ILK-PINCH-parvin complex, have been associated with ACM (533). Furthermore, rare ILK variants with unknown pathogenicity have been reported in DCM patients (483, 534, 535). In addition, ILK was reported to be upregulated in hypertrophic ventricles from patients with outflow tract obstruction (536). The ILK-PINCH-parvin complex connects β-integrins to the actin cytoskeleton, acting as a regulator of microtubule dynamics, gene transcription, and cell-cell adhesion (537, 538). ILK is a 55-kDa protein, consisting of four NH₂-terminal ankyrin repeats, a central phosphoinositide-binding pleckstrin homology domain, and a COOH-terminal atypical kinase domain (539). The ILK ankyrin repeats bind to PINCH-1/2/LIMS1/2 (540, 541), thymosin β4 (542), EPHA1 (543), and ILKAP (544, 545), whereas the ILK kinase domain interacts with β₁-integrin (546), β₃-integrin (547), α-parvin/CH-ILKBP/actopaxin (548, 549), β-parvin/affixin (550), paxillin (551–553), kindlin-2 (554, 555), rictor (556), AKT1(557), and SRC (558). ILK was initially thought to act as an active protein kinase capable of phosphotransfer (546, 559) However, follow-up studies demonstrated that ILK is a catalytically inactive pseudokinase, although this has been a matter of debate (560, 561). In-line with this, effects of supposably “kinase-dead” ILK mutants in mouse models were shown to be caused by destabilization of the ILK pseudokinase domain structure and impaired protein-protein interactions, and not the loss of its putative catalytic activity (561–563).

Global KO of ILK led to death at the preimplantation stage before heart formation (564). Striated muscle-specific KO of ILK caused severe DCM, resulting in death within 5–18 wk of age. This was accompanied by structural remodeling, fibrosis, downregulation of ion channels, and strong arrhythmogenicity as well as reduced AKT and β₁-integrin/FAK signaling (565, 566). Homozygous KI mice carrying putative constitutive active (p.S343A) or activation-resistant (p.R211A) ILK mutations in the presumed ILK kinase domain were normal and did not show changes in AKT or GSK-3b phosphorylation (567), consistent with the notion that ILK is a pseudokinase. Tg mice overexpressing WT ILK or the ILK p.S343A mutant in cardiomyocytes developed a compensated form of cardiac hypertrophy, characterized by preserved systolic and diastolic function and the absence of fibrosis (536). In contrast, Tg mice with cardiomyocyte-specific overexpression of the ILK p.R211A mutant, incapable of binding to phosphoinositol (557) and α-parvin (548, 568), did not develop cardiac hypertrophy and showed a blunted hypertrophic response to ANG II (536). Furthermore, ILK-R211A Tg mice showed HSPA-dependent cardioprotection against myocardial infarction-induced injury, reducing infarct size and cardiac dysfunction (569). Similarly, transgenic expression of ILK p.R211A protected against doxorubicin-induced cardiotoxicity, limiting apoptosis, maintaining sarcomeric structure, and preserving cardiac function through modulation of SERCA2 and phospholamban function via a scaffolding mechanism (570). Noteworthily, the progenitor cell marker islet-1 was expressed in the heart of ILK-R211A Tg mice, supporting a cardiomyogenic role of ILK (571). Taken together, biochemical and mouse studies indicate that ILK is essential for normal cardiac function and development, which is not mediated via phosphorylation. Further studies are required to identify through which noncatalytic mechanisms ILK governs cardiac homeostasis.

Filamin C

Mutations in the FLNC gene have been associated with all major types of human cardiomyopathy, including HCM, DCM, RCM, and ACM. Clinically, filamin C-truncating mutations have been associated with DCM and ACM with an increased risk of sudden cardiac death (482, 572–581), whereas missense mutations have predominantly been linked to HCM and RCM (573, 582–586). Moreover, cardiac conduction defects and hypertrophy are observed in one-third of filamin C-related skeletal myopathy cases (587, 588). Although protein aggregates are typical pathological findings in skeletal muscle biopsies in patients with filamin C-related skeletal myopathy, no aggregates have been found in hearts of DCM patients carrying FLNC mutations (579, 589). Because of the high prevalence of rare FLNC variants in the general population, variable penetrance of HCM mutations, and lack of mechanistic insights, more studies are required to identify the true pathogenicity of FLNC variants (590, 591). Filamins are large, homodimeric actin-binding and cross-linking proteins (592–594). Filamin C is a 290-kDa protein specifically expressed in striated muscle (595, 596) and composed of an NH₂-terminal paired actin-binding calponin-homology (CH) domain, followed by 24 Ig-like domains, structured into two rod regions (594, 596). The relative flexibility of the two rod regions regulates the spatial characteristics of F-actin cross linking via multiple molecular interactions (597). The last COOH-terminal Ig domain, Ig24, mediates dimerization of filamin C, which is required for its actin cross-linking activity (598, 599). In cardiomyocytes, filamin C is localized at the Z-line, costamere, and ICD (595, 600). In the Z-line, filamin C interacts with myotilin (601, 602), members of the calsarcin/FATZ/myozenin family (75, 602–604), myopodin (605), nebulette (151), and aciculin/phosphoglucomutase 5 (PGM5) (606); at the costamere, it binds to sarcoglycans (596), β₁-integrin (607), and ponsin/CAP/SORBS1 (608); and at the ICD, it interacts with NRAP (609), Xin/XIRP1/CMYA1 (610), and XIRP2/CMYA3 (610). These multifaceted protein associations underlie the role of filamin C in the organization and stabilization of the cardiomyocyte cytoskeleton (471, 611, 612).

Deletion of the last 8 exons (exons 41–48) of Flnc in mouse resulted in perinatal lethality due to respiratory failure and severe defects in skeletal myogenesis but no overt cardiac phenotype (613). The introduced deletion resulted in a hypomorphic allele, as truncated filamin C was still expressed at low levels in mutant mice. A true Flnc null allele was generated by deletion of exons 9–13, causing a frameshift in the Flnc coding sequence and subsequent loss of the protein (614). Both global and cardiomyocyte-specific loss of filamin C led to embryonic lethality at E10.5 (614), accompanied by severe chest edema (unpublished data). Cardiomyocyte-specific ablation of filamin C in adult animals led to rapidly progressing DCM leading to death within 2 wk, accompanied by upregulation of multiple proteins, including direct filamin C interaction partners at the costamere and ICD (614).

T-TUBULES AND SARCOPLASMIC RETICULUM

Deep invaginations in the sarcolemma, termed t-tubules, and their interaction with the sarcoplasmic reticulum (SR) in microdomains, named dyads, are essential for calcium handling and excitation-contraction coupling, governing heart contraction (reviewed in Ref. 9). At the dyads, depolymerization of the plasma membrane causes an influx of calcium through voltage-gated L-type calcium channels (LTCCs), triggering calcium release from the SR through the ryanodine receptor (RyR2) in a process called calcium-induced calcium release (CICR). The rapid increase in intracellular calcium initiates sarcomere contraction, which is subsequently terminated by the removal of cytosolic calcium, primarily through its reuptake into the SR via the cardiac SR calcium ATPase (SERCA2), regulated by phospholamban (615) but also to a lesser extent by extrusion through the sodium/calcium exchanger (NCX) in the plasma membrane. Calsequestrin 2 (CASQ2), a calcium-binding protein located in the lumen of the SR, acts as a calcium buffer and negative regulator of RyR2 in complex with triadin and junctin to prevent spontaneous SR calcium release (616). The key interaction between LTCCs localized in the t-tubules and RyR2 on the cytosolic side of the dyad is stabilized and maintained through numerous protein-protein and protein-lipid interactions (reviewed in Refs. 8–10, 617).

Mutations in genes encoding cardiac dyad proteins directly involved in calcium handling, such as RYR2, TRDN (triadin), and CASQ2 (calsequestrin 2), most frequently lead to an inherited arrhythmia known as catecholaminergic polymorphic ventricular tachycardia (616), which in some cases is accompanied by LVNC (618). Furthermore, RYR2 mutations have been identified in ACM (619, 620) and DCM (621) patients, and RYR2 loss of function mutations were recently associated with sudden cardiac death with normal exercise stress test, a novel syndrome that was named RyR2 calcium release deficiency syndrome (CRDS) (622). In addition, mutations in PLN (phospholamban) have been associated with DCM, HCM, and ACM (623). Mutations in genes encoding LTCC subunits have been linked to different arrhythmogenic diseases, including Brugada syndrome (CACNA1C, CACNB2, CACNA2D1), Timothy syndrome (CACNA1C), long QT syndrome (CACNA1C, CACNA2D1), and short QT syndrome (624). Moreover, mutations in LRRC10, encoding the LTCC auxiliary protein leucine-rich repeat containing protein 10, have been associated with DCM (625, 626). Our laboratory has contributed to the understanding of cardiac dyads by identifying the human cardiomyopathy gene NEXN (nexilin) as a cardiac dyad protein and creating a junctophilin-2 based murine model to study the composition of cardiac dyads, as discussed in detail below.

Junctophilin-2

Mutations in the JPH2 gene, encoding junctophilin-2, have been found in patients with HCM (627–630) and autosomal recessive pediatric DCM (631). Junctophilin-2 is a 74-kDa protein belonging to a four-member family of human junctophilins (632). All junctophilins contain eight NH₂-terminal membrane occupation and recognition nexus (MORN) motifs, a central α-helical domain, and a COOH-terminal transmembrane region (633). Junctophilin-2 is predominantly expressed in the heart and tethers transverse tubules to the junctional SR (634), forming cardiac dyads (9, 635). The MORN motifs mediate the interaction with the plasma membrane (636), the central α-helical domain is thought to determine the 12-nm distance between the plasma membrane and the SR (637, 638), and the transmembrane region anchors junctophilin-2 to the SR membrane (638). Junctophilin-2 plays an important role in cardiomyocyte calcium homeostasis, which is partially mediated via interaction with RyR2 (639, 640), the CACNA1C subunit of the LTCC (641–643), the voltage-gated potassium channel subunit K_v7.1 (KCNQ1), and the small-conductance calcium-activated potassium channel subtype 2 (KCNN2) (644) as well as the nonchannel proteins nexilin (645), caveolin-3 (646), and striated muscle enriched protein kinase (SPEG) (647). Moreover, under cardiac stress, the proteolytic fragment of junctophilin-2 translocates to the nucleus, where it acts as a transcriptional regulator, attenuating pathological remodeling in response to cardiac stress (634).

Global KO of junctophilin-2 resulted in embryonic lethality from E10.5 due to developmental cardiac failure (638). Cardiomyocytes from junctophilin-2 KO embryos isolated at E9.5 showed defective calcium handling due to physical LTCC and RyR2 uncoupling, causing impaired CICR and consequent calcium overload in the SR, leading to vacuolization. Constitutive cardiomyocyte-specific shRNA-mediated downregulation of Jph2 led to mouse strain-specific development of DCM, accompanied by abnormal calcium transients and impaired maturation of t-tubules (648, 649). Depending on the extent of inducible junctophilin-2 downregulation, adult mice developed acute heart failure under basal conditions or exhibited enhanced cardiac vulnerability to pressure overload (640, 650). In all reported studies, adult junctophilin-2 downregulation in heart led to abnormal calcium handling (640, 650, 651). Murine KI models of JPH2 mutations recapitulated certain features of human disease. A murine pseudo-KI (PKI) model of the JPH2 p.E169K mutation found in a patient with early-onset paroxysmal atrial fibrillation (AF) in the context of HCM has been reported (639). The JPH2-E169K PKI model was generated by intercrossing cardiomyocyte-specific mutant Tg mice with a mouse-line with inducible cardiac-specific shRNA-mediated knockdown of JPH2 (640). JPH2-E169K PKI mice were more susceptible to AF induction and had increased spontaneous SR calcium leak attributed to weaker binding of mutant protein to RyR2 (639). However, the HCM phenotype was not recapitulated in the JPH2-E169K PKI model. Using the same principle, a murine PKI model for the JPH2 p.A399S mutation, corresponding to the human JPH2 p.A405S mutation found in an HCM patient, was generated (628). In accordance with the clinical history of the patient, no overt calcium handling abnormalities were observed in this PKI model. However, asymmetric septal hypertrophy was observed, closely recapitulating the clinical findings. Unaltered calcium handling yet pathological septal hypertrophy in the JPH2-A399S PKI mouse model suggest that junctophilin-2 is involved in cardiac homeostasis independent of calcium handling. To advance the understanding of cardiac dyad molecular organization in the native context, a new mouse model was created by fusing BioID2 to the endogenous JPH2 coding sequence (652). BioID2 is a promiscuous biotin ligase mutant used for proximity-dependent biotinylation for characterization of protein complexes (653). This approach has been proven to be more sensitive and specific than other methods applied to study the interactome of cardiac dyads and led to the identification of potential novel junctophilin-2-associated proteins that may play essential roles in cardiac dyad formation, maintenance, and function (652).

Nexilin

Mutations in the NEXN gene, encoding nexilin, have been associated with HCM (654) and DCM (655, 656). Nexilin is expressed in two isoforms of ∼75 and 100 kDa and predominantly expressed in cardiac and skeletal muscle as well as in smooth muscle cells (655, 657, 658). Initially identified as a cardiac Z-disk protein (655), nexilin was later found to be a component of junctional membrane complexes in cardiac dyads, where it interacts with junctophilin-2 and RyR2 (645). Furthermore, it binds to F-actin (657) and has been shown to promote actin polymerization and cell migration in smooth muscle cells (658). Nexilin was also recently reported to be reduced in human atherosclerotic plaques as well as in the serum of patients with coronary artery disease (659). Reduced nexilin levels were found to be associated with increased atherosclerosis and inflammation in atherosclerotic lesions, suggesting nexilin as a potential therapeutic target in atherosclerosis-related diseases.

Global nexilin KO mice developed rapidly progressive postnatal DCM with wall thinning, leading to death within 12 days after birth (645, 660). Furthermore, nexilin KO mice showed endocardial deposits, which were initially attributed to endomyocardial fibroelastosis (660) but later suggested to be intracardiac mural thrombi (645). Cardiomyocyte-specific nexilin KO recapitulated the phenotype, demonstrating that the phenotype of global nexilin KO mice is a direct consequence of its loss in cardiomyocytes (645). Further analyses showed failure of cardiac-specific nexilin KO mice to initiate and form t-tubules, resulting in impaired calcium handling, altered expression of calcium-related proteins, and consequent DCM (645). Ablation of nexilin in adult murine hearts, in which the t-tubular network is already formed, led to DCM accompanied by disorganization of the t-tubular network, defective calcium homeostasis, and reduced sarcomere shortening, indicating that nexilin is required for the maintenance of the t-tubular network in adult heart (661). Homozygous KI mice carrying the NEXN p.G645del mutation equivalent to the human NEXN p.G650del mutation associated with DCM developed progressive DCM, characterized by reduced t-tubular formation and disorganization of the transverse-axial tubular system (662). Levels of mutant protein in homozygous animal hearts were only 30% of WT levels, suggesting that the NEXN p.G645del mutation affects nexilin stability. However, heterozygous global nexilin KO mice or mice carrying a deletion of Nexn exons 3 and 4 showed normal heart function, despite expression of 50% and 20% of nexilin, respectively, compared with WT control animals (662). These findings indicate that not pure haploinsufficiency but altered functionality of mutant nexilin is driving cardiomyopathy in the NEXN-G645del KI model (662).

NUCLEUS

Nuclei in cardiomyocytes, highly specialized, mechanically active, mainly postmitotic cells, face a unique environment that requires special molecular mechanisms to govern nuclear integrity and tissue functionality. The nucleus is separated from the cytoplasm by the nuclear envelope (NE), which is composed of two lipid bilayers, the outer nuclear membrane (ONM) and the inner nuclear membrane (INM), enclosing the perinuclear space (PNS), which also contains nuclear pore complexes (NPCs) allowing bidirectional transport of macromolecules across the NE (reviewed in Refs. 663–665). The ONM is connected to the peripheral SR and the cytoskeleton, whereas the INM is supported by the nuclear lamina, a meshwork of intermediate filaments composed of A-type (A and C) and B-type (B1 and B2) lamins as well as lamina-associated proteins such as the lamin B receptor (LBR), lamina-associated polypeptide 1 and 2 (LAP1 and LAP2), emerin, and LEM domain containing 3 (LEMD3)/MAN1. The linker of nucleoskeleton and cytoskeleton (LINC) complex spans the NE and is formed by the association of nesprins at the ONM with Sad1 and UNC84 domain-containing (SUN) proteins at the INM within the PNS. SUN1/2 proteins at the INM interact with laminA/C, emerin, luma, components of the nuclear core complex, and heterochromatin in the nucleoplasm, whereas nesprins at the ONM associate with actin filaments, microtubules, and intermediate filaments, tethering the LINC complex to the cytoskeleton. Thus, the LINC complex provides structural support to the nucleus and mechanically couples the cytoskeleton to the nuclear lamina and associated chromatin. Furthermore, increasing evidence indicates a role of the LINC as a mechanosensor, translating external mechanical forces into biochemical signals, triggering changes in NE structure and composition, chromatin organization, gene expression, and nuclear calcium handling (reviewed in Ref. 666). In addition, cardiac development, maintenance, and response to stress are dependent on cardiac transcription factors directly interacting with genomic DNA (667–670).

Because of the specialized environment of cardiac nuclei, mutations in nuclear and nucleus-associated proteins have been associated with a wide range of clinical syndromes, including cardiomyopathy. Mutations in several genes encoding proteins of the NE can cause Emery–Dreifuss Muscular Dystrophy (EDMD), associated with DCM, HCM, cardiac conduction defects, and/or arrhythmias, including EMD (emerin), LMNA (lamin A), TMEM43 (luma), and SYNE1/2 (nesprin-1/2) (reviewed in Refs. 664, 671). LMNA mutations have also been associated with DCM and congenital dystrophy with cardiac involvement (672, 673), TMEM43 mutations with ACM (664), and SYNE1/2 mutations with isolated DCM and ataxia with DCM (672). Mutations in SUN proteins do not appear to be a primary cause of cardiomyopathy, but SUN variants can act as disease modifiers in combination with mutations in other EDMD-associated genes (674). Mutations in LEMD2, encoding LEM domain nuclear envelope protein 2 located in the INM, were found to be associated with autosomal recessive juvenile‐onset cataracts and arrhythmic cardiomyopathy with mild impairment of LV systolic function (675, 676). Furthermore, mutations in TOR1AIP1 (LAP1) have been linked to dystonia and cerebellar atrophy with DCM as well as LGMD with cardiac involvement (677, 678), whereas a mutation in TMPO2 (LAP2α) has been associated with DCM (679). Mutations in genes encoding nucleoporins (NUPs) have also been linked to cardiovascular disease, including NUP155 and NUP37, which have been associated with atrial fibrillation and sudden cardiac death (680, 681). Furthermore, mutations in proteins enclosed in the nucleoplasm, such as transcription factors, can cause cardiomyopathies. For example, biallelic truncating mutations in ALPK3 (α-protein kinase 3), exclusively expressed in striated muscle and localized in the nucleus, where it acts as a transcriptional regulator through phosphorylation of cardiac transcriptional factors, have been linked to severe pediatric cardiomyopathy, mostly manifesting as neonatal DCM transitioning to HCM, whereas heterozygous ALPK3 mutations have been associated with HCM (reviewed in Ref. 623). Key components of the LINC complex, nesprins and luma, as well as the transcriptional regulators of cardiac development and function, TBX20 and PRDM16, have been extensively studied in our laboratory and are reviewed in detail below.

Nesprin-1 and Nesprin-2

Gene mutations in SYNE1 and SYNE2, encoding nuclear envelope spectrin repeat proteins Nesprin-1 and Nesprin-2, respectively, lead to nesprinopathies (682). Nesprinopathies manifest as isolated DCM (680, 683–685), EDMD with DCM (680, 682, 686, 687) or HCM (688), or ataxia with DCM (689). Nesprins are components of the LINC complex connecting the nuclear lamina to the cytoskeleton and belong to a ubiquitously expressed four-member family of mammalian spectrin repeat proteins of which Nesprin-1, -2, and -3 are expressed in the heart (689). Nesprin-1 and -2 exist in numerous isoforms due to variable RNA processing, resulting in alternative transcription initiation, termination, and splicing (690). The largest nesprin-1 and -2 isoforms, named “giant,” are 1.01 MDa and 796 kDa in size, respectively, and located at the ONM (691, 692). Giant nesprins contain two NH₂-terminal calponin-homology (CH) domains that bind to F-actin (693), a central spectrin repeat-containing rod domain of variable length, and a COOH-terminal transmembrane Klarsicht/Anc/Syne homology (KASH) domain, which anchors nesprins to the NE via interaction with the INM proteins SUN1 and SUN2 across the PNS (694, 695), forming the LINC complex. The smaller nesprin-1/2 isoforms lack either the NH₂-terminal CH domains, the COOH-terminal KASH domain, or both, and include a variable number of spectrin repeats (690, 696, 697). Although giant nesprin-1/2 isoforms are ubiquitously expressed, many smaller isoforms have tissue-specific expression and are located at distinct cellular locations, with some specifically expressed in cardiac and/or skeletal muscle (e.g., nesprin-1α2, -2α1, and -2ε2) (696). Importantly, smaller nesprin-1/2 isoforms are found at the INM, where they directly interact with emerin, lamin A/C, and SUN1/2 through their COOH-terminal spectrin repeats (698–702). Furthermore, nesprin-1α was found to dimerize and bind to muscle A-kinase anchoring protein (mAKAP/AKAP6) through its NH₂-terminal spectrin domains, targeting mAKAP to the NE (701, 703). The COOH-terminal region of nesprin-1/2 was also shown to bind to α-N-catenin (704) as well as kinesin light chains (KLC)-1/2 in skeletal muscle cells, implicating nesprins in nuclear positioning (705–707).

Genetic mouse models have revealed that the cardiac functions of nesprin-1 and nesprin-2 largely overlap and that the two proteins to a great extent can compensate for the loss of each other. Mice in which the nesprin-1 KASH domain has been replaced by an unrelated sequence of 61 amino acids (nesprin-1^rKASH mice), showed 50% perinatal mortality (683, 708), and the surviving mice developed an EDMD-like phenotype. At ∼1 yr of age, the mice developed cardiac systolic dysfunction and cardiac conduction defects, preceded by the presence of elongated nuclei with irregular shape and reduced heterochromatin. However, although the interaction with SUN proteins was disturbed, the location of LINC proteins at the NE was unaffected. In contrast to the nesprin-1^rKASH model, deletion of the KASH domains of either nesprin-1 (nesprin-1^ΔKASH) or nesprin-2 (nesprin-2^ΔKASH) did not affect viability (709), possibly because of the expression of nesprin-1/2 KASH-less isoforms (710). On the other hand, nesprin-1^ΔKASH and nesprin-2^ΔKASH double mutants died of respiratory failure within 20 min of birth (709). Similar to the nesprin-1^rKASH model, global KO of nesprin-1 resulted in 60% perinatal lethality and severe skeletal muscle defects, including defects in nuclear positioning and anchorage (710). However, no cardiac pathological phenotype was observed up to 12 mo of age. Global double-KO mice for nesprin-1 and the intermediate filament protein desmin, which like nesprins has been demonstrated to play a role in nuclear anchorage and positioning, showed increased mortality and a more severe dystrophic phenotype compared with single-KO mice (711). This was associated with more severe nuclear positioning and anchorage defects as well as decreased nuclear deformation under biomechanical stretch, i.e., cytoskeletal-nuclear strain transmission, compared with single-KO mice, suggesting that nesprin-1 and desmin play partially redundant roles in skeletal muscle nuclei anchorage in skeletal muscle. Through deletion of the nesprin-1 CH domain in mouse it was later shown that the nesprin-1 giant isoform is dispensable for postnatal viability and nuclear positioning, whereas nesprin-1α2 KO mice showed a phenotype similar to nesprin-1 KO mice, demonstrating that this isoform is essential for postnatal viability and skeletal muscle function (712). In contrast to the severe effect of nesprin-1 ablation, global nesprin-2 KO mice showed no defects in either cardiac or skeletal muscle (713). As early lethality of nesprin-1^ΔKASH and nesprin-2^ΔKASH double mutants (709) had previously precluded the analysis of cardiac function, nesprin-1 and nesprin-2 double-KO mice with cardiac-specific nesprin-1 deletion were generated (713). At 10 wk of age, double-KO mice showed reduced wall thickness and systolic disfunction, associated with fibrosis, apoptosis, reexpression of fetal genes, and mislocalization of emerin and lamin A/C from the NE. Furthermore, changes in nuclear positioning, morphology, and heterochromatin localization were observed, which was also found to a lesser extent in single-KO mice (710, 713). Noteworthily, double-KO mice showed an impaired biomechanical gene response upon application of strain, which was also partially blunted in single-KO mice (713). This demonstrated that nesprin-1 and -2 show partial functional redundancy in the heart. Interestingly, in contrast to nesprin-2 KO mice, showing no obviously pathological phenotype, RNA interference-mediated knockdown of nesprin-2 targeting both the NH₂- and COOH-terminal-encoding parts in mice resulted in early embryonic lethality, with no knockdown embryos found by E13 (714). The authors speculated that certain isoforms may not have been targeted in the conventional nesprin-2 KO mice, whereas loss of all isoforms was achieved by the shRNA-based approach, leading to the severe embryonic phenotype not observed in other models. However, this remains to be proven.

Luma

Mutations in TMEM43, encoding luma, an evolutionarily conserved, ubiquitously expressed NE protein, have been linked to ACM (715–721) and EDMD with cardiac arrhythmias (722). Notably, a recent international evidence-based reappraisal of genes associated with ACM classified the luma p.S358L mutation as a definite cause of cardiomyopathy (723). Luma is a 45-kDa protein, containing four transmembrane helices and a “domain of unknown function,” DUF1625, located between the third and fourth helix, which comprises about two-thirds of the protein and might act as a tetraspanin-like membrane organizer (724, 725). Luma is located in the INM, where it interacts with emerin, lamin A/C, lamin B1, and SUN2 (722, 724). Furthermore, several potential direct or indirect binding partners were identified by coimmunoprecipitation-coupled mass spectrometry, including β-actin (726).

Unexpectedly, global luma KO mice in the C57BL/6J background were indistinguishable from WT control mice both under basal conditions and in response to mechanical pressure overload (727). In contrast, heterozygous cardiomyocyte-specific luma KO mice in a mixed 129/C57BL/6N background developed late-onset cardiomyopathy, characterized by LV dilation, systolic dysfunction, increased cardiomyocyte size, fibrofatty infiltration, and apoptosis (728). This was associated with activation of the DNA damage response (DDR) pathway and consequent activation of TP53 and upregulation of senescence-associated secretary proteins, including transforming growth factor-β1 (TGF-β1). KI mice for the luma p.S358L mutation were generated by two different laboratories. In one study, homozygous luma-S358L KI mice in the C57BL/6J genetic background showed no changes in cardiac function under basal conditions up to 12 mo of age (727), whereas in another study 8-wk-old heterozygous luma-S358L KI mice in a mixed 129/C57BL/6 genetic background were found to show an ACM-like phenotype, characterized by LV dilation, posterior wall thinning, and fibrofatty infiltration, although electrocardiographic abnormalities were not detected (729). In particular, activation of the nuclear factor-κB (NF-κB)-TGF-β signaling cascade was found to promote cardiac fibrosis. A Tg mouse model overexpressing the luma p.S358L mutant in cardiomyocytes more closely recapitulated human ACM, developing progressive ventricular dilation, systolic dysfunction, cardiac conduction defects, apoptosis, cardiomyocyte death, and fibrofatty replacement, leading to premature death within 6 mo after birth (726). The mutant protein was found to display partial delocalization from the NE to the cytoplasm as well as reduced interaction with emerin and β-actin. In addition, luma-S358L Tg mice showed reduced AKT activity and consequent glycogen synthase kinase-3β (GSK3β) activation and inhibition of β-catenin-dependent transcription. Importantly, although inhibition of fibrosis had no effect on cardiac function in luma-S358L Tg mice, inhibition of GSK3β improved cardiac function and survival, suggesting GSK3β as a novel therapeutic target for ACM (726).

TBX20

Mutations in the TBX20 gene, encoding a member of the T-box family of transcription factors, have been shown to be responsible for congenital heart diseases including septation and cardiac valve defects (730–732). Furthermore, TBX20 mutations have been associated with DCM (731, 733, 734) and LVNC (735). In addition, increased TBX20 RNA levels were found in myocardial biopsies from patients with idiopathic DCM and shown to negatively correlate with LV function (736). TBX20 is a 49-kDa highly evolutionarily conserved protein consisting of a single centrally located T-box domain and expressed in almost all cardiac cell-lineages (737). TBX20 plays critical roles during early heart development, and numerous studies have identified TBX20 as a key modulator of cardiac gene expression programs mechanistically linked to other cardiac transcription factors, including Nkx2.5, GATA4, GATA5 (738), ISL1, MEF2C (739), COUP-TFII (740), TBX5 (741), CASZ1 (742), P21, MEIS1, and BTG2 (743). In cardiomyocytes TBX20 regulates the expression of key potassium and calcium channels, other ion transporters, gap junction components, and genes involved in cell cycle and growth (743–745).

Global TBX20 KO in mice resulted in embryonic lethality around E10.5, accompanied by severe hypoplastic ventricles and defects in cardiac looping (746–748). Similarly, complete Tbx20 knockdown by RNA interference in mouse embryos resulted in defective cardiac morphogenesis, whereas partial TBX20 inhibition led to congenital heart defects, including impaired outflow tract septation, hypoplasia of the right ventricle, and defective valve formation (739). Ablation of TBX20 specifically in the atrioventricular canal (AVC) and outflow tract (OFT) of the early heart tube through crossing of Tbx20 floxed mice with Tbx2-Cre mice resulted in embryonic lethality by E10.5 (749). The mutant embryos showed failure to form the AVC constriction and loss of AVC cushion mesenchymal cells, demonstrating an essential role of TBX20 for early AVC patterning and epithelial-mesenchymal transition (EMT). This effect was partially mediated through its downstream target BMP2, as reexpression of BMP2, which was dramatically decreased in mutant embryos, rescued the EMT defects. Specific deletion of TBX20 in the endocardium and cushion mesenchyme through Tie2-Cre-mediated targeting did not affect the number of AVC cushion mesenchymal cells, indicating that TBX20 modulates the EMT during early AVC development (749). Similarly, Nfatc1-Cre-directed deletion, likewise ablating TBX20 in the cushion endocardium and mesenchyme, did not affect EMT initiation but led to embryonic lethality between E14.5 and E16.5 due to defective valve elongation and maturation as well as atrioventricular cushion formation, associated with aberrant WNT/β-catenin signaling and reduced ECM gene expression (750). Tie2-Cre-mediated TBX20 deletion was subsequently reported to cause cardiac cushion abnormalities and septal defects, including outflow tract (OFT) septation and atrioventricular septal defects (AVSDs) as well as lack of the dorsal mesenchymal protrusion (DMP), resulting in embryonic lethality by E14.5 (751). This was associated with reduced proliferation in OFT cushions and defective endocardium-derived cell migration as a result of reduced expression of ECM and cell migration genes critical for cardiac septation. Like Tie2-Cre-targeted TBX20 KO mice, Nfatc1-Cre-targeted TBX20 KO mice exhibited OFT septation defects but did not show AVSDs, which could be explained by the targeting of Tie2-Cre but not Nfatc1-Cre to the endothelium adjacent to the DMP (pulmonary venous endothelium, sinus venosus, and common atrium) (751). Cardiomyocyte-specific KO of TBX20 from around E9.5 (using Tnnt2-Cre mice) did not affect initiation of heart chamber formation but led to failure of cardiac chamber expansion and septal defects, associated with reduced cardiomyocyte proliferation, resulting in embryonic lethality around E14.5 (740). Cardiomyocyte-specific ablation of TBX20 in adult mice led to cardiac death within 5–16 days after induction, associated with chamber dilation, wall thinning, contractile dysfunction, and arrhythmias (745). The observed phenotype could to a large extent be explained by dysregulation of a transcriptional network consisting of MEF2A, TEAD, CREB1, and ESRRA, regulating calcium homeostasis and ion transport in adult heart (745). Combined genome-wide chromatin immunoprecipitation and transcriptomic analyses on this model subsequently demonstrated a role of TBX20 as a transcriptional activator and repressor depending on its association with different cofactors (744). More specifically, TBX20 was found to activate genes involved in cardiac contraction and energy metabolism, whereas it represses genes involved in development and specification of noncardiac tissues and systems repressed in the heart.

Transgenic overexpression of TBX20 under the control of the Myh6-Cre promoter, known to mediate overexpression by ∼E11 (425), resulted in partial pre- and postnatal lethality (752). Surviving mice showed reduced body size and developed early-onset DCM, associated with systolic dysfunction, septum abnormalities, and ventricular hypertrabeculation similar to the phenotype of Tg mice for bone morphogenetic protein 10 (BMP10), an upstream regulator of TBX20. Another group overexpressed TBX20 in embryonic (E9.5) and fetal (E12.5–17.5) cardiomyocytes using Nkx2.5-Cre and Myh7-Cre, respectively (753). Overexpression of TBX20 in embryonic heart resulted in growth retardation due to reduced cardiomyocyte proliferation, resulting in embryonic lethality by E10.5. In contrast, Myh7-Cre-mediated TBX20 overexpression in fetal cardiomyocytes resulted in increased thickness of the developing myocardium from E14.5 as well as induction of cardiomyocyte proliferation. Myh7-Cre-mediated TBX20 overexpression promoted cardiomyocyte proliferation also in adult heart, resulting in an increased number of small, proliferating, mononucleated cardiomyocytes with fetal-like characteristics associated with induction of BMP2/SMAD1/5/8 and PI3K/AKT/GSK3β/β-catenin signaling (754). Induction of cardiomyocyte-specific TBX20 overexpression in adult mice likewise promoted cardiomyocyte proliferation through activation of multiple proliferative pathways and repression of negative cell cycle regulators without affecting cardiac morphology and function (743). Notably, overexpression of TBX20 in a model of myocardial infarction enhanced cardiac repair, reducing infarct size and cardiac dysfunction, resulting in significantly increased survival (743).

PRDM16

Mutations in the PRDM16 gene, encoding PRDI-BF1 and RIZ homology (PR) domain-containing 16 (PRDM16), have been identified in patients with DCM and LVNC (755–761). Moreover, PRDM16 variants have been linked to QRS duration (762). PRDM16, also known as MEL1, belongs to a 19-member family (in human) of PR domain-containing (PRDM) transcription regulators and histone methyltransferases, containing a conserved NH₂-terminal PR domain and a variable number of zinc fingers (52, 763–765). Human PRDM16 is expressed in four isoforms ranging from 170 to 150 kDa, a full-length isoform, a smaller isoform lacking the PR domain, and two isoforms resulting from alternative splicing of full-length PRDM16 (766, 767). All isoforms share zinc finger repeats, a proline-rich domain, a repressor domain, and a COOH-terminal acidic domain (767). Through its zinc finger domains, PRMD16 directly binds to DNA as well as interacts with various transcription factors and chromatin regulators to promote or repress gene expression depending on the biological context (766, 768, 769). In particular, PRMD16 has been found to function as a transcriptional cofactor for the transcription factors EHMT1 (770), MED1 (771, 772), PGC-1α (773), C/EBPβ (774), PPARγ (775), SMAD (776), and SKI (777). The intrinsic histone methyltransferase activity of PRMD16 is dependent on its NH₂-terminal PR domain (778), and together with PRDM3, PRDM16 catalyzes monomethylation of lysine 9 of histone 3 (H3K9me1), ensuring mammalian heterochromatin integrity (779).

Global PRDM16 KO mice died shortly before or after birth and showed multisystemic defects, including gross cardiac ventricular hypoplasia (780–782). Cardiomyocyte-specific PRDM16 KO mice generated by crossing of Prdm16 floxed mice with Myh6-Cre mice displayed increased heart weight-to-body weight ratio, longitudinal elongation of the heart, and cardiomyocyte hypertrophy, associated with ventricular fibrosis, cardiac conduction defects, and increased expression of hypertrophic markers at 5 mo of age (783). Another conditional PRDM16 KO mouse model generated using Mesp1-Cre, which drives Cre expression in cardiogenic progenitors (784–786), developed late-onset HCM starting from 9 mo of age, progressing to heart failure (768). This was associated with fibrosis, increased cardiomyocyte size, increased expression of hypertrophic markers, mitochondrial dysfunction, and metabolic defects. Young mice subjected to metabolic stress induced by high-fat diet developed a similar phenotype. In contrast to the two aforementioned conditional PRDM16 KO models, Xmlc2-Cre- and Tnnt2-Cre-mediated targeting, resulting in efficient PRDM16 deletion from as early as E7.5 (425, 787), resulted in biventricular noncompaction, LV-specific dilation, and cardiac contractile dysfunction, leading to death within 7 days after birth, recapitulating the LVNC phenotype of patients carrying PRDM16 mutations (769). This was associated with reduced cardiomyocyte proliferation specifically in the LV, likely contributing to the development of LV dilation and systolic dysfunction. The failure of Myh6-Cre-and Mesp1-Cre-targeted PRDM16 KO mice to recapitulate the LVNC phenotype may be due to later developmental deletion and variable recombination efficiencies. Cardiomyocyte-specific KO of PRDM16 in adult mice did not result in any pathological cardiac phenotype, consistent with strongly decreased PRDM16 protein levels after birth, indicating that PRDM16 does not play an important role in adult cardiomyocytes. This is in accordance with the developmental origin of LVNC, which is thought to be caused by an arrest of the normal compaction of the endomyocardial layer of the heart during early embryogenesis (769). Mechanistically, PRDM16 was found to function as a transcriptional cofactor, maintaining compact myocardial cardiomyocyte identity by activating a subset of myocardial genes required for compaction while repressing trabecular genes in LV compact myocardium, at least partly through cooperation with the LV-enriched transcription factors TBX5 and HAND1. Consequently, PRDM16 ablation in cardiomyocytes of the developing heart led to a shift in the transcriptomic identity of cardiomyocytes associated with compact myocardium to a gene signature resembling that of trabecular cardiomyocytes and/or neurons, suggesting that misspecification of compact or trabecular cardiomyocytes may be a common pathomechanism of LVNC.

CYTOSOL

The cytosol is a semifluid solution filling the interior of the cell and surrounding the various organelles and subcellular compartments (788). The cytosol is enclosed by the cell membrane and membranes of the organelles, constituting a separate cellular compartment (789). The cytosol is composed of ∼70% water as well as ions (e.g., sodium, potassium, calcium, chloride, bicarbonate, magnesium, and amino acids), smaller organic molecules (e.g., glucose and other simple sugars, polysaccharides, amino acids, nucleic acids, and fatty acids), and macromolecules (e.g., proteins) (790). The cytosol also contains protein complexes, for example, of enzymes involved in the same metabolic pathway (791) as well as protein compartments, such as the proteasome, which forms an enclosed compartment containing proteases degrading cytosolic proteins (792, 793). The cytosol allows for free movement of ions and molecules across the cell, and a major function of the cytosol is to transport metabolites from their site of production to where they are needed (794). Furthermore, molecules taken up by the cell or being secreted are transported through the cytosol in membrane-bound vesicles moved along the cytoskeleton by microtubule-based motor proteins (795, 796). Other important processes occurring in the cytosol include metabolic processes such as protein synthesis and glycolysis, cell division through mitosis, as well as signal transduction from the sarcolemma to the different sites within the cell (797–799). Furthermore, differences in concentrations between ions in the cytosol and the organelles or extracellular fluid are essential for many cellular processes, including cell signaling, osmosis, and cellular excitability, such as CICR triggering muscle contraction (800, 801).

Several cytosolic proteins have been associated with cardiomyopathy (21). Mutations in BAG3, involved in protein quality control, have been associated with DCM, HCM, and RCM (802, 803), whereas mutations in the eukaryotic elongation factor 1 alpha 2 (eEF1A2) have been linked to neurological disease associated with DCM (804–806). Both proteins have recently been studied in our laboratory and are reviewed in detail below.

BAG3

Mutations in the BAG3 gene, encoding BCL2-associated athanogene 3 (BAG3), have been found to be causative for isolated adult-onset DCM (572, 807–816) or early-onset skeletal myopathy with or without RCM or HCM (817–819). Moreover, several studies have identified BAG3 as a genome-wide significant locus for idiopathic DCM (811, 820, 821), all-cause heart failure development (821, 822), or reduced LV ejection fraction after adjusting for preexisting cardiac conditions (823, 824). In particular, the SNP rs2234962, causing a p.C151R substitution in the BAG3 protein, has been associated with a decreased risk of DCM (811, 816, 820–822). However, why or how the BAG3 C151R allele provides cardioprotection is not known. Several studies have also explored BAG3 as a potential biomarker. In particular, increased BAG3 serum levels have been associated with adverse outcome in heart failure patients (825) and anti-BAG3 antibodies were detected in the serum of patients with chronic heart failure (826). Consistently, decreased BAG3 protein levels have been reported in the heart of patients with end-stage heart failure (815), suggesting that extracellular BAG3 is released by stressed cardiomyocytes, resulting in the production of autoantibodies. BAG3 is a 75-kDa cochaperone belonging to a six-member family of BAG proteins, which share a common conserved COOH-terminal BAG domain that mediates interactions with the HSPA family of chaperones involved in protein quality control (827–830). Furthermore, BAG proteins bind to the antiapoptotic protein BCL2 through their COOH-terminal region (831, 832). In addition to its COOH-terminal BAG domain, BAG3 contains an NH₂-terminal WW domain, binding to proline-rich regions, a PxxP repeat, associated with SH3 domain-containing proteins [e.g., phospholipase Cγ, PLC-γ (827) and Src (833)], and two conserved IPV (Ile-Pro-Val) motifs, binding to small HSPs (sHSPs), including HSPB5 (αB-crystallin), HSPB6, and HSPB8 (834–838). In particular, BAG3 physically links ATPase-dependent HSPA family members, which are themselves ATPases, with ATPase-independent sHSPs without enzymatic activity in large multichaperone complexes (836, 838), implicating it in protein folding, chaperone-assisted autophagy to remove misfolded and damaged proteins, inhibition of protein aggregation, and maintenance of mitochondrial stability (reviewed in Refs. 802, 839). Furthermore, BAG3 inhibits apoptosis through its binding to BCL2 (831). BAG3 is strongly expressed in cardiac and skeletal muscle and is localized both in the cytosol and at the Z-line (840), where it interacts with the F-actin capping protein CapZβ1 (841), thereby maintaining the structural integrity of the Z-line. In particular, BAG3 was found to promote the interaction between CapZβ1 and HSPA/HSP70, and in the absence of BAG3 CapZβ1 was lost from the Z-line and degraded by the ubiquitin-proteasome system, resulting in myofibrillar degeneration in response to mechanical stress (841). BAG3 has also been shown to localize to the sarcolemma and t-tubules and coimmunoprecipitate with the β1-adrenergic receptor and LTCC (842). In-line with this, BAG3 was found to modulate myocyte contractility and excitation-contraction coupling in ventricular cardiomyocytes in response to β-adrenergic stimulation (842). Consistent with a role in cardiac contractility, reduced sarcomeric BAG3 expression was shown to correlate with decreased myofilament contractile function in DCM patients, which was found to be associated with impaired protein turnover, resulting in incorporation of ubiquitinated misfolded sarcomeric proteins in the sarcomere (843).

Global BAG3 KO mice showed impaired postnatal growth and developed DCM, leading to early death within 3–4 wk after birth (840, 844, 845). Cardiomyocyte-specific KO of BAG3 caused progressive DCM, accompanied by reduced levels of sHSPs, suppressed autophagosome formation, and accumulation of insoluble protein aggregates, resulting in premature death (840). It should be pointed out that among the four cardiac sHSPs HSPB5, HSPB6, and HSBP8, but not HSPB7, were downregulated in cardiomyocyte-specific BAG3 KO mice (835, 840). In agreement with this observation, it has been shown that HSPB5, HSPB6, and HSBP8, but not HSPB7, interact with BAG3, which is critical for their stability, whereas HSPB7 exhibits unique BAG3-independent functions (835, 846–848). Heterozygous cardiac-specific BAG3 KO mice with haploinsufficiency developed progressive systolic dysfunction and LV dilation from 10 wk of age (849). Furthermore, analyses on adult cardiomyocytes showed decreased autophagy flux and increased apoptosis as well as a blunted contractile response to adrenergic stimulation. At 6 wk of age, before the development of LV dysfunction or structural alterations, significantly increased protein ubiquitination and reduced myofibrillar force generation were observed, suggesting that impaired BAG3-mediated turnover of sarcomeric proteins, resulting in integration of ubiquitinated proteins in the sarcomere, is responsible for the contractile dysfunction (843). It was further demonstrated that BAG3 exerts its function via stress-induced chaperone-assisted selective autophagy (CASA) at the Z-line, and several sarcomeric proteins were identified as candidates for BAG3/CASA-mediated turnover. Notably, AAV9-mediated delivery of BAG3 to mice subjected to myocardial infarction, known to be associated with reduced BAG3 levels, restored normal myofibrillar contractile function and sarcomere protein turnover, resulting in improved cardiac systolic function (843), whereas no effect was observed in mice subjected to sham surgery. Thus, BAG3-mediated sarcomere turnover is critical for normal contractile function, and BAG3 may represent a potential therapeutic target. In a recent study, the effect of homozygous and heterozygous cardiomyocyte-specific BAG3 KO in adult mice was studied (850). Under basal conditions, homozygous KO mice showed decreased systolic function, LV dilation, and interstitial fibrosis 4 wk after tamoxifen induction, whereas heterozygous KO mice showed normal cardiac morphology and function. Since BAG3 has been found to be upregulated during physiological and pathological hypertrophic remodeling, homozygous and heterozygous mice were subjected to swimming exercise or chronic phenylephrine infusion to induce cardiac physiological and pathological hypertrophy, respectively. Although both homozygous and heterozygous KO mice showed a blunted hypertrophic response to exercise (physiological hypertrophy), they displayed aggravated pathological maladaptive remodeling in response to phenylephrine treatment. In combination with in vitro studies, it was revealed that BAG3 promotes physiological hypertrophy and inhibits pathological remodeling through activation of the AKT/mTOR pathway and inhibition of the Calcineurin/NFATC2 pathway, respectively (850). Thus, several-lines of evidence suggest that BAG3 overexpression protects against deterioration of cardiac function under conditions of cardiac stress or injury. It should, however, be noted that transgenic overexpression of BAG3 in the heart caused mildly reduced cardiac systolic function associated with reduced levels of sHSPs due to increased protein turnover via activation of autophagy (851), demonstrating that permanent high-level BAG3 overexpression is not beneficial.

Several KI and Tg mouse models of BAG3 mutations associated with DCM have been studied. A mouse KI model carrying the BAG3 p.E455K mutation, found to be an unequivocal cause of DCM in a large multigenerational family (10), developed impaired postnatal growth and premature lethality by 4 wk of age, similar to BAG3 global KO mice (840). Cardiomyocyte-specific KI mice for the BAG3 p.E455K mutation largely recapitulated the cardiac phenotype of cardiomyocyte-specific BAG3 KO mice, demonstrating that the BAG3 p.E455K mutation results in loss of function. The BAG3 p.E455K mutation was found to disrupt the binding of BAG3 to HSPA, implying that the interaction is essential for the role of BAG3 in maintaining cardiomyocyte protein homeostasis and cardiac function (840). Several murine models of the BAG3 p.P209L mutation found in patients with severe early-onset skeletal myopathy combined with restrictive cardiomyopathy (817–819) have been reported. A murine KI model carrying the BAG3 p.P215L mutation (equivalent to the BAG3 p.P209L mutation in human) did not display any cardiac abnormalities up to 16 mo of age (852), whereas Tg mice overexpressing the human BAG3 p.P209L mutant either specifically in cardiomyocytes or globally showed distinct pathological phenotypes. Tg mice generated by the use of a cardiomyocyte-specific promoter developed progressive late-onset systolic and diastolic dysfunction associated with cardiomyocyte hypertrophy and increased anterior wall thickness (853), whereas ubiquitous transgenic expression resulted in growth retardation and the development of early-onset restrictive cardiomyopathy, associated with fibrosis, disintegration of sarcomeres, and formation of protein aggregates, recapitulating the human disease in model (854). Notably, the phenotype could be mitigated by AAV-mediated delivery of shRNA against the transgene.

Eukaryotic Elongation Factor 1 Alpha 2

Mutations in eukaryotic elongation factor 1 alpha 2 (eEF1A2), a 50-kDa protein encoded by the EEF1A2 gene, have been found to cause a neurodevelopmental disorder, in some cases associated with DCM (804–806). eEF1A2 is one of two isoforms of the alpha subunit of the eEF1 complex, which facilitates the GTP-dependent recruitment of aminoacyl-tRNAs to the acceptor site of the ribosomal complex during protein translation (855). Although eEF1A1 is ubiquitously expressed, eEF1A2 expression is restricted to skeletal muscle, heart, brain, and spinal cord (42, 856–858), explaining the clinical manifestations of eEF1A2 mutations. Despite 92% sequence identity between eEF1A1 and eEF1A2, computational models and recent crystal structure have revealed subtle, but significant, differences between the two isoforms, potentially resulting in differences in aminoacyl-tRNA selectivity, actin binding, and oligomerization patterns (859, 860). A distinct pattern of posttranslational modifications, including lipidation, is proposed to be a specific adaptation of eEF1A2, explaining its enrichment in excitable cell types such as neurons and myocytes (860). These subtle changes could provide a structural basis for the diverse functions of eEF1A2 besides delivering aminoacyl-tRNAs to the ribosome, including cell shape regulation via its interaction with phosphatidylinositol-4 kinase IIIβ (861, 862) and carcinogenesis through its interaction with a set of tyrosine kinases and phosphatases via its proline-rich motif (863).

An autosomal recessive mutation in the EEF1A2 gene, resulting in eEF1A2 ablation, has been shown to be responsible for the spontaneously arising wasted (Wst) mutation in an inbred mouse colony (864). Homozygous Wst mice develop profound neuromuscular defects, including muscle wasting, tremors, ataxia, and motor neuron degeneration, starting from ∼21 days of age, resulting in death within 28 days after birth (864). This onset of the phenotype coincides with the decline in eEF1A1 expression to undetectable levels by 21 days of age and a concomitant increase in eEF1A2 expression, so that Wst mice are deficient for both eEF1A isoforms by 21 days of age (856). Tg expression of eEF1A2 in muscle did not ameliorate the atrophic phenotype of Wst mice, indicating that the muscle wasting resulted from the loss of eEF1A2 in neurons (865). Cardiomyocyte-specific KO of eEF1A2 was recently shown to result in the development of rapidly progressing LV dilation and systolic dysfunction from ∼6 wk of age, accompanied by cardiac fibrosis, upregulation of markers of cardiac remodeling, and reduced expression of components of the mitochondrial electron transport chain (866). This led to death within 22 wk of age, starting already from 7 wk of age. At the molecular level, compensatory upregulation of the eEF1A1 isoform was observed and no defect in global protein translation was found (866). This suggests that loss of eEF1A2 may only affect the translation of a specific subset of proteins and/or that eEF1A2 may have aminoacyl-tRNA transport-independent functions in the heart. Global KI mice homozygous for the eEF1A2 p.P333L mutation, which has been associated with neurodevelopmental disorder with DCM in patients homozygous for the mutation, showed a phenotype similar to Wst mice and died at ∼4 wk of age. On the other hand, cardiac-specific eEF1A2-P333L KI mice exhibited essentially the same DCM phenotype as cardiomyocyte-specific eEF1A2 KO mice and showed reduced eEF1A2 protein expression levels, although RNA levels were unaffected, indicating that the mutation constitutes a loss-of-function mutation affecting protein stability.

DISCUSSION

Studies by others and us on sarcomeric and nonsarcomeric proteins, including intermediate filament proteins, SR proteins, costameric proteins, NE proteins, transcription factors, cochaperones, etc., are developing a comprehensive understanding of the complex cytoskeletal and noncytoskeletal network required for cardiac muscle development, structure, and optimal function as well as how mutations in these proteins lead to cardiomyopathies. This advancement in knowledge would not have been possible without genetic mouse models. The CRISPR/Cas9 gene-editing technology has significantly simplified and streamlined the generation of genetic mouse models, and our laboratory has actively used this technology (867). Intensified use of murine models in cardiac research has revealed specific limitations and unexpected advantages as well as stimulating new-lines of future research as discussed below. However, genetic mouse models do not always recapitulate human disease. The work on BAG3 by our laboratory has revealed that the utility of mice as a genetic model not only varies from protein to protein but can even depend on the location and nature of the mutation in the same protein. The BAG3 p.P209L (818, 819, 868) and p.E455K (816) mutations both have a solid clinical indication of their pathogenicity. Still, only BAG3 p.E455K KI recapitulated the human disease, whereas artificial overexpression of the BAG3 p.P209L mutant protein was required for the mutant to elicit pathology in mice (852–854, 869). Moreover, mouse genetic background can significantly affect the phenotype, implying the existence of powerful modifier genes (870, 871). For example, our group did not observe any overt cardiac phenotype in luma KI and KO models, whereas other groups using different genetic backgrounds partly recapitulated the ACM phenotype observed in humans (727–729). Conversely, differences between phenotypes in a mouse model and human patients might be beneficial and aid translational research. For example, KO of LAMP2, a protein essential for autophagy and responsible for Danon disease, a multisystem disorder characterized by cardiomyopathy (872, 873), results in only mild impairment of cardiac function that does not cause premature death, allowing for the use of LAMP2 KO mice as a preclinical model for testing of potential therapies (874, 875). Similarly, mice can sustain otherwise potentially fatal ventricular tachycardia in humans, making it an ideal model to study and screen for new treatments for arrhythmogenic diseases, such as polymorphic catecholaminergic ventricular arrhythmia (876, 877). However, it should be pointed out that maladaptive responses leading to heart failure in murine hearts differ from those in humans (reviewed in Ref. 878). Therefore, findings from murine studies might not always directly apply to humans. Several drug candidates were shown to be effective in murine models but failed in human trials (reviewed in Ref. 879), indicating species-specific molecular pathways leading to heart failure. Therefore, caution should be taken when pathomechanistically linking genetic defects to the development of cardiomyopathy and heart failure in murine models. As observed in numerous instances in this review, genetic background plays a significant role in whether the murine model will recapitulate the human disease or not. Natural genetic variability of mouse strains can be successfully applied to uncover the genetic bases of cardiovascular pathophysiological manifestations in otherwise hard-to-study diseases, such as nonfamilial forms of heart failure (reviewed in Ref. 880). Moreover, changing genetic background can be a successful strategy to bring a genetic mouse model close to the human disease. For example, the mdx mouse model of human dystrophinopathy recapitulates the human disease better in the DBA/2 than the C57BL/10 background (881). Finally, modern genome editing technologies have lifted species limitations and catalyzed the generation and use of nonmurine models for human cardiomyopathies (882, 883).

With their strengths and weaknesses, genetic mouse models will continue to be critical for understanding the pathomechanisms of human cardiomyopathies. The function of many cardiomyopathy-related proteins, such as ALPK3 (884–888), LEMD2 (675, 676), and TAX1BP3 (tax1 binding protein 3) (889) are still elusive, and the molecular mechanisms leading from mutations in certain proteins to cardiomyopathy are not well understood. Furthermore, it is well known that mutations even within the same gene can result in different forms of cardiomyopathy (see Table. 1), which may depend on the position of the mutation, affecting different functions of the protein. However, the exact mechanisms still remain elusive. Intriguingly, even the same mutation may lead to distinct forms of cardiomyopathy in different patients, such as the MYPN p.Y20C (159) and MLP p.W4R (96–99, 102, 108) polymorphisms, which have been associated with both DCM and HCM while being present also in the healthy population. The different outcomes may be explained by the effect of genetic modifier(s) and/or environmental effects. However, future studies on mouse models mimicking distinct human mutations in different mouse strains and in various environmental conditions are needed to determine this. Furthermore, accumulating clinical data have demonstrated the oligogenic nature of specific human cardiomyopathies, and mouse models for these diseases carrying mutations in more than one gene have been proven to be valuable models by several groups, including ours (491, 890–893). Thus, research on oligogenic cardiomyopathies using genetic mouse models will continue to expand. Moreover, despite their clinical relevance, cardiomyopathy-linked mutations in the noncoding regions of the genome remain poorly understood and understudied, opening new research areas (894). Thus, although there has been significant progress in understanding the molecular mechanisms underlying human cardiomyopathies, mouse models will continue to be an essential tool for uncovering still unanswered research questions as well as for testing of potential therapies. Insight into the molecular basis of cardiomyopathy is essential for the identification of new therapeutic strategies specific for each of the distinct forms of cardiomyopathy to be translated to the clinical setting.

GRANTS

This work was funded by grants from the National Heart, Lung, and Blood Institute, American Heart Association, and California Tobacco-Related Disease Research Program. J.C. holds an American Heart Association-endowed chair in cardiovascular research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.-L.B., J.B., and J.C. prepared figures; M.-L.B. and J.B. drafted manuscript; M.-L.B., J.B., and J.C. edited and revised manuscript; M.-L.B., J.B., and J.C. approved final version of manuscript.