Modeling Hypertrophic Cardiomyopathy: Mechanistic Insights and Pharmacological Intervention

Hypertrophic cardiomyopathy (HCM) is a prevalent and complex cardiovascular disease where cardiac dysfunction often associates with mutations in sarcomeric genes. Various models based on tissue explants, isolated cardiomyocytes, skinned myo ﬁ brils, and puri ﬁ ed actin/myosin preparations have uncovered disease hallmarks, enabling the development of putative therapeutics, with some reaching clinical trials. Newly developed human pluripotent stem cell (hPSC)-based models could be complementary by overcoming some of the in-consistencies of earlier systems, whilst challenging and/or clarifying previous ﬁ ndings. In this article we compare recent progress in unveiling multiple HCM mechanisms in different models, highlighting similarities and discrepancies. We explore how insight is facilitating the design of new HCM therapeutics, including those that regulate metabolism, contraction and heart rhythm, providing a future perspective for treatment of HCM. the medium and/or the incorporation of a Matrigel ™ overlay step resultedinhigher cardiomyocyte purities(~90%). Todecreasethe variability and costs associatedwith growthfactor-based protocols, several small molecules have been explored for driving cardiac differentiation, such as GSK3 β inhibitor ChiR and the WNT inhibitors IWR-1-4, KY0211, and XAV939, resulting in cost-effective production of cardiomyocytes [121 – 123]. Although cardiac monolayer protocols typically result in high numbers of hPSC-CMs, cardiomyocytes suffer inherent heterogeneity (mixed subtype populations), batch variation, limited scalability, and interlaboratory variability [124] because small-scale 2D cultures are sensitive to ﬂ uctuations in physicochemical parameters. To greatly increase the scalability of differentiation protocols, several 3D suspension bioreactor systems have produced CMs from hPSCs by using anchorage-dependent (i.e., microcarrier-based) and -independent systems [125 – 127]. These recent protocols produce high purities of abundant LV cardiomyocytes that are essential for modeling HCM. reduce the intracellular calcium build-up that underlies arrhythmogenic events associated with HCM. Metabolic modulators such as perhexiline and trimetazidine inhibit fatty acid oxidation, thereby promoting the use of glucose as an energy substrate and fostering improved energy ef ﬁ ciency by reversing the energy depletion phenotype. Blebbistatin and epigallocatechin-3-gallate (EGCG) alter myo ﬁ lament calcium sensitivity. Mevacamten reduces contractility by decreasing the ATPase activity of sarcomeric myosin heavy chain. Omecamtiv mecarbil increases actin – myosin binding, which results in enhanced cardiomyocyte contraction. Abbreviations: β -AR, β -adrenoreceptor; FA, fatty acid; LTCC, L-type calcium channel; NCX, sodium/calcium exchange pump; SR, sarcoplasmic reticulum; RYR2, ryanodine receptor.

Initial studies to dissect the mechanisms underlying HF were performed in cardiac muscle strips from explanted human hearts [17], unveiling hallmarks of HCM. Post-mortem histological analysis of cardiac tissue revealed extensive areas of interstitial fibrosis, myocyte enlargement, and chaotic spatial arrangement in HCM patients [18]. Further investigations showed 2-2.5-fold increased polyploidy in HCM, with inconsistent changes in multinucleation [19][20][21]. Isolated mitochondria from human hypertrophied hearts exhibited about twofold higher oxygen consumption relative to healthy controls [22].
Preservation of an intact sarcolemma in whole tissue preparations is advantageous for the recapitulation of cardiac tissue architecture and pharmacology in vitro. This led to characterization of HCM pathophysiology by the identification of abnormal drug responses, including an attenuated increase in contraction force induced by β-adrenergic agonists [23], and perturbed calcium handling leading to prolonged relaxation during diastole [24]. However, cardiac tissue biopsies are often derived from HF patients who have undergone myectomy to reduce muscle thickening, and therefore require demanding logistics of sample handling (e.g., immediate processing).
Mechanical and enzymatic dissociation of human endocardial tissue generates viable cardiomyocytes that have been used for electrophysiology studies and pharmacological responses [25], revealing arrhythmias (sixfold higher early after-depolarizations) and~50% higher diastolic Ca 2+ concentrations. Single-cell transcriptome investigations revealed highly variable mutant versus wild-type (WT) sarcomeric gene expression in heterozygous HCM patients, underlying heterogeneous cell contractility and Ca 2+ sensitivity [26].
However, isolated human cardiomyocytes dedifferentiate almost immediately after explant and do not proliferate in culture [27]. Thus, human cardiac tissue and its derivatives offer an important, but rarely available, biological source, and this greatly reduces the scope of the physiological parameters that can be investigated.

Myofibrils from Skinned Heart Tissue/Cells: Direct Myofilament Evaluation
Subcellular structures have also been used to model HCM by placing tissue/cell-derived skinned myofibril preparations between a force transducer and length motor, and immersing in solutions with different Ca 2+ concentrations to stimulate contraction/relaxation [28].

Glossary
Cardiac troponin I/T (cTnI/T): cardiac regulatory sarcomeric proteins that control the calcium-mediated interaction between actin and myosin. CRISPR/Cas9: genome-editing technology that enables the introduction or correction of SNPs, for example in sarcomeric genes involved in HCM. The technology facilitates the generation of isogenic sets of cardiac cells for disease modeling. Expressivity: the severity of the phenotype that develops in a patient with a pathogenic mutation. Heterologous expression: expressing a gene in a host organism that normally does not produce it, by recombinant DNA technology. Human pluripotent stem cells (hPSCs): these encompass (i) human embryonic stem cells (hESCs) isolated from the preimplantation blastocyst stage embryo, and (ii) human induced pluripotent stem cells (hiPSCs) reprogrammed from somatic cells donated by HCM patients and healthy individuals. Isogenic sets: cell lines that differ only in the mutation being studied and retain the same genetic background. This allows the effect of the mutation in the disease progression to be isolated (providing a clearer understanding of genetic causation) by directly comparing healthy versus diseased isogenic counterparts. Myosin-binding protein C (MYBPC): a sarcomeric protein involved in regulating the positions of myosin and actin for interaction, acting as a brake on cardiac contraction. Myosin heavy chain (MHC): one of the several sarcomeric proteins involved in regulating cardiomyocyte contraction. Can be expressed in two isoforms, α and β, that are encoded by the MYH6 and MYH7 genes, respectively. Species differences underlie changes in compartmental expression of αversus β-MHC expression in the heart (β is the main form expressed in human ventricles) Myosin light chain 2 (MLC2): also known as regulatory light chain of myosin, MLC2 is a sarcomeric protein involved in modulating cardiac myosin cross-bridge kinetics (its phosphorylation enhances myofilament sensitivity to calcium). Penetrance: the proportion of individuals carrying a pathogenic mutation who display a phenotype.

Trends in Molecular Medicine
This method enables direct access to myofilament function to quantify isometric tension, Ca 2+ sensitivity, and ATP consumption ( Figure 2). Most reports using this methodology show decreased contractile force in human HCM samples. For instance, cardiac explants from patients bearing different mutations in MYH7 and MYBPC3 genes [encoding β-myosin heavy chain (β-MHC), and myosin-binding protein C, respectively] have consistently revealed lower tension forces relative to healthy controls (21 vs 36 kN/m 2 ), with MYH7-mutant samples showing the lowest values when normalized to myofibril density (73 vs 113 kN/m 2 ) [29]. In addition, Kraft et al. showed a modest increase in Ca 2+ sensitivity of skinned myofibers from human cardiac explants bearing the R723G-β-MHC mutation, relative to healthy controls, that was dependent on the hyperphosphorylation state of several sarcomeric proteins [cardiac troponin I/T (cTnI/T), MYBPC, and myosin light chain 2 (MLC2] [30]. Simultaneous measurement of force development and ATPase activity in tissue-extracted myofibrils can be used to quantify the energy cost of contraction. This revealed significant increases for tissues from MYBPC3-mutant and MYH7-mutant patients, compared with sarcomere mutation-negative (SMN) HCM patients, at saturating Ca 2+ concentrations [31]. This was corroborated in multicellular cardiac myofibrils of human R403Q-β-MHC, showing 50% lower tension generation relative to SMN-HCM patients, as well as maximum ATPase activity [32]. This results in a higher cost of contraction, indicating inefficient ATP utilization that causes higher cardiac workload, often leading to HF (termed the 'energy depletion model' [4]).
Sarcomeric mutation-negative (SMN): HCM patients who do not exhibit known mutations in sarcomeric genes, constituting approximately 50% of all cases. Single-nucleotide polymorphisms (SNPs): point mutations that cause a change in the DNA sequence leading to abnormal or truncated proteins with perturbed structure and function, a common cause of HCM.  However, this difference was not observed in single myofibrils derived from the same tissues, indicating that sample preparation affects the endpoint assay [33].

Trends Trends in in Molecular
Nevertheless, myofibrils derived from skinned cardiac muscle offer simpler handling logistics because samples can be frozen in a relaxation solution that preserves their functionality for several months. Remarkably, higher energy cost in HCM muscle strips appears to be a feature shared by different mutations. This was recently corroborated in homozygous K280N-troponin T samples, which showed 24-72% higher values than three different control groups, that was ascribed to faster cross-bridge detachment [34].

Purified Actin/Myosin Proteins: Sarcomeric Interactions
Addressing the low sample availability limitations of whole-cell/tissue preparations, in vitro motility assays were developed by recording the movement of fluorescently labeled actin filaments over a layer of randomly oriented myosin molecules immunoadsorbed to an antibody-coated surface [35]. Tethering an ultracompliant microneedle to actin filaments enables measurement of average force per crossbridge, facilitating direct assessment of sarcomeric mutations and interactions at the molecular level.

Trends in Molecular Medicine
However, studies performed using this technique have produced results conflicting with wholetissue/cell analysis. For instance, rat tissues with heterologous expression of R403Q-α-MHC showed a fourfold decrease in ATPase activity and a fivefold reduction in motility compared with controls [36], whereas the same mutation in myosin isolated from mouse explants led to 2.3-fold higher ATPase activity and 60% higher velocity than WT myosin tissue [37]. These discrepancies extended to human biopsies. Although tissues from patients harboring β-MHC mutations displayed consistently lower sliding velocities (0.11-0.29 μm/s) than healthy controls (0.48 μm/s) [38], another report showed the opposite in R403Q-and L908V-βMHC, which had 30% higher velocities [39].
These differences highlight the main drawback of this method: technical artefacts derived from protein purification procedures. Tissue biopsies contain limited concentrations of myosins that are further degraded by freezing and tissue handling procedures. Alternatively, heterologous production of recombinant proteins often result in changes in structure and expression levels relative to endogenous systems, with functional consequences for the endpoint assay [14].
These limitations are evidenced when analyzing the same protein interactions by different techniques. Laser-trap assays can be used for direct molecular analysis of actin-myosin interactions by measuring the force and displacement resulting from the interaction of a single myosin molecule with an optically trapped actin filament [40]. The same group has shown significant differences between myosin extracted from mice bearing the R403Q-β-MHC mutation vs WT when performing in vitro motility assays (2.3-fold higher ATPase activity, 2.2-fold greater force generation, and 1.6-fold faster actin filament sliding), but no changes in force and displacement in the same samples when using the optical trap assay [37].
Taken together, these differences suggest that more reductionist models, where isolated proteins are investigated under unloaded conditions, tend to generate less robust conclusions due to technical constraints and the absence of physiological complexity that is characteristic of highly organized sarcomeres.

Animal Models: Transgenics and Species Differences
Animal models of heart disease have been crucial in advancing knowledge of pathophysiology towards new therapeutics because the basic principles of cardiac excitation and contraction in the species used are relatively conserved [41]. Although some naturally occurring cardiomyopathies have been detected in animals (e.g., Portuguese waterdogs), transgenic animal models enable detailed physiological and molecular analysis of disease [11,42]. Rodents (e.g., Syrian hamsters [43]) in particular have been extensively employed to model HCM because they overcome cell source limitations and facilitate whole-organism investigation of disease progression over time.
Primary cardiomyocytes isolated from transgenic rats and mice have been integrated into fibrin-based engineered heart tissues (EHTs) that enable assessment of contractility by measuring the displacement of silicon posts to which they are attached [44]. Analysis of the contractile force of rat cardiomyocytes transduced with adeno-associated virus expressing FLH1 variants containing single-nucleotide polymorphisms (SNPs) identified in HCM patients revealed hyper-or hypocontractile phenotypes depending on the mutation (K455fs-Fhl1, 27% higher force; C276S-Fhl1, 23% lower force vs WT controls). Evaluation of the beating kinetics of these tissues showed prolonged contraction and relaxation times in both variants (~18% and~30% longer, respectively) [45]. The same approach was applied to Ankrd1 in rat, showing~50% higher contractile force as well as contraction and relaxation velocities in T123M-Ankrd1 EHTs versus controls, but no discernible phenotypes in P52A and I280V variants [46]. This reinforces the notion that HCM phenotypes are mutation-specific because different mutations in the same locus elicit variable effects on contraction.
Furthermore, data from transgenic mouse models have clearly linked mutations in sarcomeric genes with impaired Ca 2+ handling. Knollmann and colleagues studied isolated cardiomyocytes, perfused hearts, and whole mice bearing the human I79N-cTnT mutation, showing shortened ventricular action potentials at 70% repolarization (14 ms in I79N-cTnT, vs 23 ms in control). Ca 2+ transients of electrically stimulated ventricular I79N-cTnT myocytes were measured using a fluorescent Ca 2+ indicator dye (Fura-2-AM) and showed reduced intensity (half the fluorescent amplitude in I79N-cTnT vs control) and twofold slower decay kinetics, consistent with increased Ca 2+ sensitivity of I79N-cTnT mutant fibers [47]. Moreover, EHTs made from Mybpc3-mutant mice displayed higher sensitivity to Ca 2+ , as evidenced by lower Ca 2+ EC 50 values for force generation relative to the WT (0.34 mM for homozygous, 0.48 mM for heterozygous, vs 0.66 mM for WT) [48,49]. This methodology has also shown differences in Ca 2+ EC 50 values for contraction between species (0.15, 0.39, and 1.05 mM Ca 2+ for rat, mouse, and human EHTs, respectively [50]).
The variation in Ca 2+ EC 50 values for contraction between species highlights the main drawback of animal modelsthe existence of striking dissimilarities in cardiovascular physiology relative to humans. These are particularly prominent in the mouse: mice havẽ 10-fold faster beat rates (500 bpm vs 60 bpm) and 5-10-fold shorter electrocardiogram duration (50-100 ms vs 450 ms) relative to humans [41]. Changes in gene expression are also abundant, such as those pertaining to α/β-MHC expression: whereas in humans the α isoform is mainly located to the atria and the β to the ventricles [51], in the mouse the α-MHC is highly expressed in both compartments [52]. Despite sharing N90% sequence homology, this discrepancy in α versus β-MHC expression is reflected in the animal models generated.
Knock-in R403Q-α-MHC mice exhibited a significant enhancement in ATPase activity and transient kinetics (e.g., 20% increase in ADP release rate) relative to WT littermates, whereas R403Q-β-MHC animals displayed opposite or nonsignificant changes [53,54]. These inconsistencies are further exacerbated when comparing the same mutation in different animals. A recent report characterizing a transgenic rabbit model of the same mutation R403Q-β-MHC showed 20% lower force generation in comparison with WT littermates in single myofibril analysis by atomic force microscopy [55], whereas transgenic mouse models show the opposite in actinmyosin assays [37]. These incongruities between models make it challenging to translate these findings to the human disease phenotype.

hPSC-CMs: An Unlimited Cell Source
Addressing the limitations of previous samples, hPSC-CMs have been harnessed for disease modeling because they constitute an unlimited cell source [56] that enables multiparametric and detailed studies in patient-relevant genomes ( Figure 2). Cardiac differentiation protocols recapitulating heart development can be applied to hPSCs, thereby generating highly pure cardiomyocyte populations [57] (Box 1). To interrogate different features of HCM pathology in hPSC-CMs, several medium-and high-throughput phenotypic assays that measure cardiomyocyte structure, metabolism, electrophysiology, calcium handling, and contractility have been developed [58,59].

Trends in Molecular Medicine
In the past 6 years, hPSC-CMs have been extensively used for modeling HCM [15,16]. Lan et al. pioneered the field by reporting in vitro phenotypes of R663H-β-MHC cardiomyocytes, namely increased multinucleation (48.3%) relative to healthy controls (22.1%), with stark abnormalities in calcium handling (about threefold higher percentage of cells displaying arrhythmias) [60]. Accordingly, R442G-β-MHC hPSC-CMs exhibited irregular Ca 2+ transients (20% of the cells vs virtually none in healthy controls). Mutant cardiomyocytes showed a 21% increase in resting Ca 2+ concentration and 50% lower Ca 2+ release from the sarcoplasmic reticulum (SR) upon caffeine treatment relative to WT controls [61]. Tanaka et al. coupled high-speed video microscopy with a motion vector prediction algorithm to investigate myofibrillar disarray in hPSC-CMs derived from three HCM patients, showing an increased percentage of cells exhibiting disorganized sarcomeres (~9%) relative to controls (~5%) [62]. Moreover, myofibrils derived from hPSC-CMs bearing the E848G-β-MHC mutation showed~56% lower maximal isometric tension forces relative to controls, with faster kinetics of actin-myosin cross-bridge detachment [63]. Finally, R58Q-MYL2 hPSC-CMs recently reported significantly lower I Ca,L current density relative to a control line, which led to irregular beating, suggesting impaired diastolic relaxation [64].
However, although hPSC-CMs enable detailed studies at the molecular, myofibrillar, cellular, and tissue levels (when integrated into EHTs), they are immature in comparison to adult CMs, with differences ranging from morphological and gene expression up to structural and functional properties [65], namely an absence of T tubules [66]. This was evidenced in [67] where induced pluripotent stem cell (iPSC)-derived CMs (iPSC-CMs) bearing truncating mutations in TTN displayed striking hypocontractility, TTN haploinsufficiency, and decreased sarcomere length, which were not observed in adult myofibril samples.

Box 1. Methods for Differentiating hPSCs into Cardiomyocytes
To generate LV hPSC-CMs needed for modeling HCM, several protocols were developed based on recapitulation of cardiac development. hPSCs undergo sequential differentiation steps (mesoderm, cardiac mesoderm, cardiac progenitors) that are spatiotemporally regulated by different growth factor combinations to form cardiomyocytes [118]. These protocols can generally be divided into four main classes: (i) inductive coculture, (ii) embryoid body (EB) formation, (iii) monolayer culture, and (iv) suspension culture [57].
Early differentiation protocols relied on coculture of hPSCs with a murine visceral endoderm-like cell line (END-2) which provided the growth factors [activin A, bone morphogenetic protein (BMP)] necessary to control cardiogenesis [119,120]. Although coculture requires few cells, it is a time-consuming and labor-intensive method that results in impure cardiomyocyte populations surrounded by a murine cell line, rendering this protocol unsuitable for HCM modeling.
Methods for generating EBs are based on suspension cultures of hPSCs as single cells or small clumps in a medium that drives cardiogenesis. EBs can be formed in many ways, allowing variable control of EB size and number (e.g., in ultra-low attachment static conditions, hanging-drop method, and forced aggregation in V-shaped wells). Despite being successful, these methods are not amenable to scale-up and require very strict control of parameters such as the initial number of cells, resulting in heterogeneous cell types. EB cardiomyocytes often do not survive the dissociation into single cells that is required for phenotypic analyses [57].
Unlike EB methods where complex diffusional barriers confound the concentration of growth factors hPSCs are exposed to, uniform monolayer 2D cultures provide a more controlled and reproducible differentiation environment. Additional improvements such as temporal exclusion of insulin from the medium and/or the incorporation of a Matrigel™ overlay step resulted in higher cardiomyocyte purities (~90%). To decrease the variability and costs associated with growth factor-based protocols, several small molecules have been explored for driving cardiac differentiation, such as GSK3β inhibitor ChiR and the WNT inhibitors IWR-1-4, KY0211, and XAV939, resulting in cost-effective production of cardiomyocytes [121][122][123].
Although cardiac monolayer protocols typically result in high numbers of hPSC-CMs, cardiomyocytes suffer inherent heterogeneity (mixed subtype populations), batch variation, limited scalability, and interlaboratory variability [124] because small-scale 2D cultures are sensitive to fluctuations in physicochemical parameters. To greatly increase the scalability of differentiation protocols, several 3D suspension bioreactor systems have produced CMs from hPSCs by using anchorage-dependent (i.e., microcarrier-based) and -independent systems [125][126][127]. These recent protocols produce high purities of abundant LV cardiomyocytes that are essential for modeling HCM.

Trends in Molecular Medicine
Moreover, hPSC-CM disease models fail to recapitulate the physiological complexity of a multicellular intact heart system that encompasses fluid dynamics inherent to the circulatory system as well as neurohormonal control (i.e., metabolic changes) and extracellular matrix alterations (e.g., fibrosis). Recent efforts have included additional cell types such as cardiac fibroblasts or endothelial cells, although these were derived from different sources with multiple genetic backgrounds [68]. Thus, HCM modeling studies using hPSC-CMs should also be interpreted carefully because hallmarks of disease may be under-or overestimated [16]. Nevertheless, these cells offer a direct approach to address the highly complex genetic causation of HCM.

Isogenic Sets to Comprehend Genetic Causation
A limitation of earlier studies employing hPSC-CMs was that the impact of (epi)genetic background on phenotype can exceed that caused by the pathogenic mutation, as evidenced by variations in action potential durations between 100-700 ms among different lines described as 'healthy controls' [69]. This was also verified in patients, as recently reported for monozygotic twins carrying the same mutation (G768-β-MHC) but showing differences in myocardial fibrosis [70].
Gene-editing technologies such as CRISPR/Cas9 enable the generation of isogenic sets wherein precise changes can be made to the genome in an otherwise constant genetic background. Changes can range from large insertions or deletions (e.g., the~65 kb Dip2a gene [71]) to single bases leading to point mutations [72,73]. This means that, by coupling hPSCs with genome-editing technology, sets of disease-related and healthy isogenic hPSC-CMs can be compared to isolate the effect of the mutation on disease progression [74].
This 'isogenic' strategy has only very recently been used to model HCM. Isogenic sets of R302Q-PRKAG2 versus CRISPR-corrected hPSC-CMs were used to evaluate beating parameters by microelectrode array analysis [75]. This model identified arrhythmogenic events in diseased hPSC-CMs which were abolished upon CRISPR-mediated correction. Beat rate variability was much higher in diseased lines (coefficient of variation~30%), although there were differences between the two healthy controls (unrelated~5% vs CRISPR-corrected~12%), reinforcing the need for isogenic sets.
Genome editing enables a wider range of genotypes to be generated. This was explored in [76], wherein hPSCs from an asymptomatic patient carrying a heterozygous A57N-MYL3 mutation were further engineered to generate WT (corrected), homozygous mutant, and heterozygous frameshift mutant. Despite analyzing several parameters (morphology, transcriptome, sarcomeric structure, contractility, action potentials, and calcium handling), none of these lines exhibited a phenotype, unlike another known heterozygous pathogenic mutation (A57G-MYL3).
Wang et al. characterized isogenic sets of I79N-cTnT hPSC-CMs, and these recapitulated the phenotypes described in the transgenic mouse model bearing the same mutation {hypercontractility, impaired relaxation, increased Ca 2+ sensitivity, and pro-arrhythmic decreased action potential (AP) duration [77]}. The reduced systolic Ca 2+ release upon caffeine addition is likely to be due to calcium buffering by the calcium-sensitive mutant cTnT because treatment with calcium sensitizer, EMD57033, precipitates AP instability and arrhythmias in both models.
Smith et al. generated isogenic sets of E99K-ACTC1 variants, either by correcting or inserting the mutation in patient-derived hPSCs [78]. Notably, inserting the E99K-ACTC1 mutation in healthy Trends in Molecular Medicine lines did not induce the pathological phenotypes observed in patients with the same genotype (3.6-fold higher contraction force in EHTs, increased Ca 2+ sensitivity, and double the prevalence of arrhythmogenic events). Genomic correction of the diseased line restored the abnormalities back to baseline levels, which renders this mutation as 'necessary but not sufficient' to cause HCM. These results are in line with studies on a previous transgenic mouse model [79,80] which also reported enlarged atria and increased interstitial fibrosis. However, it is likely that the noncardiomyocyte cell types and full embryonic development that are absent from the hPSC-CM model play an important role in developing HCM phenotypes. Nevertheless, isogenic sets allowed a deeper understanding of the genetic causation of the disease by suggesting that background (epi)genetics acts as an important modulator of the pathogenic phenotype.
Engineering the same mutation in several hPSC lines and analyzing disease phenotypes facilitates direct evaluation of its penetrance. This approach was explored by CRISPR/Cas9-mediated engineering of the R453C-β-MHC change in three independent hPSC lines, generating 11 different genetic variants in total (entailing heterozygous, homozygous, homozygous plus α-MHC frameshift, and knockout genotypes) [81]. Extensive molecular and functional evaluation of hPSC-CMs showed recapitulation of the main hallmarks of HCM in all three isogenic sets, with differences in magnitude that were correlated with the mutant:WT allele expression ratio in the heterozygous lines. In the most diseased line (~60% mutant:WT allele expression), hypertrophy (51% increase in cell volume), sarcomeric disarray, increased mitochondrial respiration, arrhythmias, and hypocontractility (16.3-81.3% lower force) were reported. The mutation load was associated with the level of phenotypic perturbation, and the energy-depletion phenotype likely hampers the energetically demanding SERCA2a pump, resulting in higher levels of intracellular Ca 2+ , leading in turn to increased risk of arrhythmia. This human model corroborated energy depletion as an underlying response to HCM mutations [4,22].
Unveiling HCM Complexity: Hypercontractility versus Hypocontractility Contraction changes reported in R453C-βMHC hPSC-CMs are opposite to the response previously reported in transgenic mouse models, which showed 80% higher force in homozygous mutant cardiac myosins [82] and~50% increase in maximal force in heterologous mouse cells [83] (Table 1). However, the hPSC-CM data are consistent with isometric tension analysis from explanted human HCM tissue [29] and E848-βMHC hPSC-CMs [84]. This discrepancy is probably due to species differences between mouse and human hearts, namely pertaining to the ventricular predominance of α versus β-MHC expression [51,52].
Accordingly, a recent review compiled all the data from human skinned muscle strips, isolated cardiomyocytes, and myofibrils obtained from frozen HCM hearts bearing mutations in the MYBPC3, MYH7, TPM1, TNNT2, or TNNI3 sarcomeric genes. Across the 15 studies reviewed, spanning tens of gene-mutation combinations, maximal force developed by cardiomyocytes averaged~40% lower than control hearts in almost all cases [85].
Arguably, hypo-versus hypercontractile effects are mutation-specific, accounting for the highly complex genetic causation of HCM, which often shows overlapping features with other cardiomyopathy forms such as dilated cardiomyopathy, thwarting understanding of disease progression [86].
A hypocontractile phenotype in HCM can be clinically explained by the concentric nature of the hypertrophy: thickening of the ventricular walls without increase in heart size leads to a smaller LV end-diastolic volume. This causes an apparent hypercontraction [resulting in similar or higher LV ejection fraction, (LVEF)] but not when normalized to cardiomyocyte or myofibril density Trends in Molecular Medicine [87,88]. Interestingly, echo-based strain imaging has consistently reported hypocontractile function in HCM patients [89].
These findings question the reliability of over-simplistic assays of surrogate measurements of contractility solely based on the interaction of two myofilament proteins. They also question the physiological relevance of at least some murine models which are known to have striking species differences relative to humans [52]. In others, such as mouse and human muscle fibers bearing the F764L-MYBPC3 HCM mutation, a hypercontractile phenotype was consistently reported in both [90]. Notwithstanding, whole-organism in vivo models do not simply reflect situations in vitro. Importantly, it is conceivable that variations in contractile force in HCM change with disease course, in other words it may be hyperdynamic early (compensatory phase) but hypocontractile later (end-stage).

Pharmacological Intervention Strategies for Treating HCM
Disease models that recapitulate HCM phenotypes offer a powerful platform for screening pharmacological rescue strategies. However, the varied phenotypes and molecular mechanisms associated with different HCM-associated mutations prevent a consensus on the most efficient treatment strategies. Thus, pharmacological interventions with diverse modes of actions can be effective in HCM rescue (Figure 3). Treatments include β-blockers, L-type Ca 2+ channel blockers, antiarrhythmic drugs, calcium desensitizers, and metabolic and contractility modulators [91,92].
β-Blockers were first used to treat symptomatic HCM in the 1960s [93] because, by modulating heart rate, ventricular contractility, and stiffness, they can improve ventricular relaxation in HCM patients [94,95]. A range of β-blockers have been used, including propranolol and metipranolol, Purified actomyosin Mouse Cell line heterologous expression

Trends in Molecular Medicine
but it remains unclear why some are more effective than others [96]. Patient specific hPSC-CMs have demonstrated their efficacy: metoprolol significantly decreased beating irregularity and arrhythmias in an R442G-β-MHC model [61].
Elevated intracellular Ca 2+ and dysfunctional Ca 2+ cycling are commonly reported to be central to the pathogenesis of HCM, and inhibition of Ca 2+ entry by L-type Ca 2+ channel blockers has been investigated for the treatment of HCM [97]. In a hPSC-CM model of R663H-β-MHC mutation, diltiazem abolished calcium-handling abnormalities and arrhythmias, whereas verapamil also prevented myocyte hypertrophy, fully alleviating the HCM phenotype [60]. Another Ca 2+ blocker, nifedipine, reduced arrhythmia in R453C-β-MHC hPSC-CMs [81].
Intracellular Ca 2+ levels can be targeted indirectly through antiarrhythmic drugs that inhibit Na + influx, such as mexiletine and ranolazine [98]. These reduce intracellular Ca 2+ by targeting the Na + /Ca 2+ exchanger and promoting Ca 2+ efflux to restore intracellular Na + levels. Ranolazine halved the number of arrhythmias in R453C-β-MHC hPSC-CMs [81] and reduced hypertrophic brain natriuretic peptide (BNP) signaling in E99K-ACTC1 hPSC-CMs [78]. Hypertrophic signaling was reduced further in combination with dantrolene, a drug that blocks SR Ca 2+ release through inhibiting  cardiomyocytes that can be targeted by drugs (in red). β-Blockers prevent adrenergic signaling, leading to reduced heart rate and improved ventricular relaxation. Drugs targeting LTCC, RYR2, and NCX reduce the intracellular calcium build-up that underlies arrhythmogenic events associated with HCM. Metabolic modulators such as perhexiline and trimetazidine inhibit fatty acid oxidation, thereby promoting the use of glucose as an energy substrate and fostering improved energy efficiency by reversing the energy depletion phenotype. Blebbistatin and epigallocatechin-3-gallate (EGCG) alter myofilament calcium sensitivity. Mevacamten reduces contractility by decreasing the ATPase activity of sarcomeric myosin heavy chain. Omecamtiv mecarbil increases actin-myosin binding, which results in enhanced cardiomyocyte contraction. Abbreviations: β-AR, β-adrenoreceptor; FA, fatty acid; LTCC, L-type calcium channel; NCX, sodium/calcium exchange pump; SR, sarcoplasmic reticulum; RYR2, ryanodine receptor.

Trends in Molecular Medicine
the ryanodine receptor (RYR2) [78]. However, Phase II clinical trials (clinicaltrialsregister.eu 2011-004507-20) have not shown efficacy of ranolazine to improve functional capacity in HCM [99]. This is possibly due to other organ-level features such as myocardial fibrosis that may overrule the cellular effects or ranolazine, or to the high variability in late sodium current expression in HCM cardiomyocytes [100].
An alternative therapeutic strategy for reducing intracellular Ca 2+ is to target myofilament Ca 2+ sensitivity with desensitizing drugs. The myosin inhibitor blebbistatin alters myofilament sensitivity via an inhibitory effect on actomyosin cross-bridge formation [101], and 3 μM treatment prevented pro-arrhythmic AP triangulation in I79N-TnT hiPSC-CMs [77]. Epigallocatechin-3-gallate (EGCG) is a calcium desensitizer that acts through the formation of a ternary complex with cTnC and cTnI [102]. EGCG treatment completely reversed abnormal Ca 2+ sensitivity analyzed by an in vitro motility assay of seven HCM-related mutations in cTnT [103].
Energy depletion also has therapeutic potential in HCM. Metabolic modulators that shift the substrate preference from fatty acids towards glucose are being increasingly investigated [103] because they increase ATP production with the same oxygen consumption (thus improving myocardial efficiency). Perhexiline promotes the use of carbohydrates as the substrate for myocardial energy by inhibiting carnitine-palmitoyltransferase [104], increasing exercise capacity in Phase II trials for HCM patients (ClinicalTrials.gov NCT00500552) [105]. However, other metabolic modulators such as trimetazidine, a direct inhibitor of fatty acid β-oxidation, have shown limited effectiveness in recent HCM clinical trials (ClinicalTrials.gov NCT01696370) [106], although its use for preventive therapy is now being investigated [107].
Because disease models have shown changes in contraction force, modulators of contractility are under investigation. The allosteric modulator of cardiac myosin, mevacamten, reduces contractility by decreasing the ATPase activity of cardiac MHC in a mouse model of HCM [101], also showing effectiveness in reversing the hypercontactility seen in truncated MYBPC hPSC-CM mutants [108]. Phase III clinical trials are now ongoing to test mevacamten efficacy in adults with symptomatic obstructive HCM (ClinicalTrials.gov NCT03470545), with completion expected in June 2020 [97]. Conversely, omecamtiv mecarbil (OM) is being tested to treat hypocontractility in patients by augmenting the speed of ATP hydrolysis, thus increasing myosin head binding to actin, which results in an enhanced force-producing state [109]. Interestingly, the effects of OM are dependent on intracellular Ca 2+ levels (increasing contractility at low Ca 2+ concentrations, and decreasing it at higher concentrations [110]). Despite displaying limited effectiveness in R453C-β-MHC hPSC-CMs [81] (possibly due to the inherent immaturity of these cells), OM has shown encouraging clinical results, progressing to Phase III trials (ClinicalTrials.gov NCT02929329) that are due for completion in January 2021 [111].
Altogether, data from earlier models revealed that HCM is governed by multiple pathways, leading to progression of new drugs into clinical trials, although some are showing disappointing efficacy. HCM disease models based on isogenic hPSC-CMs offer a highthroughput system for investigating the relative efficacy of a range of combinatorial pharmacological interventions in diverse HCM genetic backgrounds. Data from such screening projects may inform more effective and mutation-specific use of therapeutic drugs. These studies highlight the importance of greater mechanistic understanding to facilitate therapeutics for HCM.

Clinician's Corner
HCM is a prevalent and complex cardiomyopathy that often leads to heart failure when cardiac function is unable to meet the metabolic requirements of the body [1]. HCM is defined by thickening of the left ventricle (LV) that is associated with preserved to increased ejection fraction [1]. Diagnosis of HCM relies on detection of increased LV wall thickness by several imaging technologies as well as by genetic screening [1,97].
There is no current effective treatment for HCM, and heart transplantation is the only long-term solution [97].
Previous HCM models have highlighted hallmarks of disease: hypertrophy, perturbations in calcium handling (arrhythmias), contractility, and energy imbalance [3]. Gene-edited human pluripotent stem cell-derived cardiomyocytes enable more refined analysis of the genetic causation of HCM and multiparametric investigation of disease mechanisms.
Developing better models of HCM will enable deeper understanding of HCM molecular mechanisms leading to more efficient treatment options.

Concluding Remarks
HCM is an 'umbrella' term used to describe a tough-to-treat and very complex disease associated with an intricate genetic causation. It exhibits a heterogeneous set of clinical manifestations with a wide spectrum of molecular mechanisms.
Earlier disease models have greatly contributed to understanding HCM progression, and have proved to be useful as a collective to advance treatment. However, they have often failed to faithfully phenocopy HCM in vitro owing to its intricate and multifarious nature. Unlimited production of hPSC-CMs has facilitated deeper mechanistic studies, complementing approaches from previous models. Genome-edited isogenic cell models have enabled detailed investigation of the genetic causation of HCM, and their further study is expected to unearth new gene modifiers, mechanisms, and therapeutic opportunities (see Outstanding Questions).
Importantly, transcriptomic data are available and are expected to highlight new molecular targets for therapy. Two reports have performed RNA sequencing analysis from nongenetically profiled patient-derived tissue explants, identifying hundreds of differentially expressed genes relative to unrelated healthy controls [112,113]. Data on transcriptomic analysis of isogenic sets of HCM hPSC-CM lines have recently been published [81,114] providing genetically precise comparisons for identifying new molecular mechanisms of disease.
The unsatisfactory efficacy of drug management of HCM [96] is likely due to the use of suboptimal preclinical testing carried out in physiologically irrelevant animals [56] and/or by lack of consideration of the complex genotype-phenotype relationship characteristic of HCM. The development of refined models of HCM will further elucidate this complex disease by recapitulating human adult conditions of pathology. Although HCM is primarily a condition of the cardiomyocytes, incorporating multiple cardiac cell lines in in vitro models will improve in vivo pathophysiological relevance. Optimizing technical constraints associated with human tissue explant analysis and solving the relative immaturity of hPSC-CMs will also foster the generation of more sophisticated HCM models that could accurately predict drug responses to treat this multifaceted disease.

Outstanding Questions
Refinement of disease models: the heart comprises cardiomyocytes, cardiac fibroblasts, smooth muscle cells, and endothelial cells, with additional neurohormonal control. Therefore, what cell types should be used to recapitulate HCM in vitro, and in what proportions, while managing the balance between practical use and over-complexity? Are there protocols to efficiently derive all these cell types from hPSCs bearing HCM mutations? Will these cells integrate in functional tissues?
How can the whole-tissue/organ interstitial and perivascular fibrosis characteristic of HCM be reproduced in vitro?
What is the best strategy to improve maturation of hPSC-derived cardiac cell types while maintaining compatibility with high-throughput phenotypic assays?
Less explored mechanisms of disease: what is the contribution of autophagy to HCM progression? What is the molecular mechanism linking telomere shortening to HCM progression?
Is heart tissue polyploidy associated with the compensatory response to HCM?
Is there an underlying/unifying phenotype of HCM among patients with different sarcomeric mutations?
What are the most relevant gene modifiers/candidates in HCM that aggravate/attenuate phenotypes caused by primary structural mutations?
Can various gene-editing tools be efficiently used to engineer additional variants of sarcomeric mutant hPSCs to investigate the role of less studied mutations in HCM progression so as to clarify the involvement of secondary gene modifiers?
Treatment of HCM: can HCM be treated with single drugs directed at individual phenotypes? Are combinational therapies and lifestyle interventions more promising approaches?
Will drugs targeting novel/secondary disease mechanisms be more efficient in treating HCM?

Trends in Molecular Medicine
Trends in Molecular Medicine, Month 2019, Vol. xx, No. xx 13