key: cord-0015201-pzy73t7d
authors: Golforoush, Pelin; Schneider, Michael D.
title: Intensive care for human hearts in pluripotent stem cell models
date: 2020-03-06
journal: NPJ Regen Med
DOI: 10.1038/s41536-020-0090-7
sha: 2f2db7a474867db40feadc7c25e1f55255643d81
doc_id: 15201
cord_uid: pzy73t7d

Successful drug discovery is ultimately contingent on the availability of workable, relevant, predictive model systems. Conversely, for cardiac muscle, the lack of human preclinical models to inform target validation and compound development has likely contributed to the perennial problem of clinical trial failures, despite encouraging non-human results. By contrast, human cardiomyocytes produced from pluripotent stem cell models have recently been applied to safety pharmacology, phenotypic screening, target validation and high-throughput assays, facilitating cardiac drug discovery. Here, we review the impact of human pluripotent stem cell models in cardiac drug discovery, discussing the range of applications, readouts, and disease models employed, along with the challenges and prospects to advance this fruitful mode of research further.

efficacy [10] [11] [12] [13] [14] [15] . But, there is a disturbing paucity even of new cardiac drug candidates put forward into early phase development, i.e., the number of New Molecular Entity (NME) applications seeking to initiate human trials 3 . In short, both ends of the cardiac drug pipeline are dry.

What are the barriers to trying more often, more effectively? Which, if any, might be remediable? Among the myriad obstacles most cited are: the drug development costs to bring a new agent to market; regulatory uncertainties; discrepancies in philanthropy and research grant support; the lack of appealing biological targets; commercial viability; the length, size, and complexity of trials required; risks of reliance on surrogate endpoints; and, of course, poor return on investment, given the frequency and cost of clinical failures 3, 16 . The net result is, consequently, few novel "first-in-class" therapies. Expert think tank recommendations have focused on the trials eco-system (reducing operating costs, streamlining trial design, embracing adaptive design, funding early-phase development more assiduously) but, appropriately, also call for strengthening "novel scientific methods to further define the pathophysiology" 3 .

One key driver of failure in human cardiac trials is, likely, the recurring lack of systematic human preclinical data for target validation and compound development, i.e., a gap in the information available to de-risk the proposition before ever entering human trials. Model organisms, ranging from the traditional to bespoke genetic lines, are instructive, to be sure, but have, in the aggregate, failed thus far to show sufficient predictive power for efficient translation of targets and drugs to benefit human health 17 . The abysmal track record for cardiac drug development speaks for itself. Drawing again on oncology as an instructive comparison, cancer drugs enter human trials having first, at a very early stage, been vetted in dozens to hundreds of well-characterized human cancer cell lines, available as turnkey resources in laboratories worldwide [17] [18] [19] . Highlighted by the National Cancer Institute's pioneering 60 human tumor cell line screen 20 and by later, larger initiatives including the Cancer Cell Line Encyclopedia 21, 22 , the role played by human cancer cells in cancer drug discovery is central, essential, and incontrovertible. In contrast, human heart muscle has never been available for equivalent proof-of-concept studies, other than sporadicallythrough biopsies and explanted hearts-never the routinized, scalable, renewable resource required for library screening, systematic compound development, and pull-through to translation. Apart from this want of starting material, even short-term expansion of adult human cardiomyocytes in primary culture is thwarted by the terminally differentiated myocytes' characteristic state of growth arrest.

By contrast, human cardiac muscle cells made in limitless quantities from human pluripotent stem cells (hPSC-CMs) now provide unprecedented access to "heart disease in a dish," with encouraging potential to accelerate the present tepid pace of cardiac drug discovery [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] (Fig. 2 ; Table 1 ). Cultured from the inner cell mass of the blastocyst, human embryonic stem cells (hESCs) are the native "gold standard" for human pluripotency 33 , have all the requisite properties including efficient cardiomyocyte creation 34 , and have even progressed into primate trials of cardiomyocyte grafting 35 . Human induced pluripotent stem cells (hiPSCs) are made instead from adult somatic cells reprogrammed with ESC transcription factors 36 , augmented or replaced by chemical reprogramming. Compared to hESCs, hiPSCs have similarly well-proven cardiogenic potential 34 but pose fewer ethical or religious concerns, and make possible the interrogation of patient-specific genetic variants 24, 37 , in addition to modeling the pandemic forms of acquired heart disease that have greatest public health significance. Here, except if otherwise noted, we refer to hPSCs, for ESCs and iPSCs collectively.

Historically, the unmet scientific need to consider hPSC-CMs in the context of drug development begins with safety pharmacology: namely, the failure of conventional animal models to predict drug toxicities 38 . In the USA, roughly one in seven approved compounds is later withdrawn from clinical use, 28% of these for adverse cardiac events, including potential lethal arrhythmias, muscle cell death, and heart failure 39 . Using microelectrode arrays (MEA), pioneering studies by Hoffman La-Roche 40 , GlaxoSmithKline 41 and Johnson and Johnson 42 profiled 10 to 30 compounds, for their pro-arrhythmic effects in hPSC-CMs. All three surveys concluded that the relevant pharmacology was captured in these human cardiomyocyte assays, with obvious inherent advantages over non-cardiac cells that are engineered to express a single human cardiac ion channel such as hERG (substrate for the ventricular arrhythmia Torsade de Pointes [TdP]). The hPSC-CMs enabled very high accuracy over the relevant range of concentrations, compared to standard lower throughput, higher cost ex vivo methods (ventricular wedge, Purkinje fiber, Langendorff). Indeed, hPSC-CMs were superior to the routine preclinical models for detecting some key parameters of risk, such as effects on repolarization akin to human QTc prolongation 41 .

Building on these encouraging findings, more systematic use of hPSC-CMs has been promoted, together with in silico modeling and other prediction tools, by the Comprehensive In Vitro Proarrhthymia Assessment (CiPA) initiative 25, 32, 43 . Explicitly, hPSC-CMs are viewed as "more complete 'biological integrators' that detect not only complex effects of drugs on multiple cardiac currents, but also modulatory effects on currents elicited through signaling pathways, channel-associated subunits, altered intracellular calcium handling, additional transporters and exchangers, and potential changes in channel densities in myocytes" 25 . The potential utility of hPSC-CMs to predict drug-induced proarrhythmic effects was demonstrated most conclusively in a blinded, international, 10-site study of 28 drugs, using two commercially available lines and diverse electrophysiological platforms including MEAs and voltage-sensing potentiometric dyes 32 . The test compounds were first categorized by degree of known clinical risk for TdP, then were analyzed in blinded fashion for the prevalence of drug-induced repolarization abnormalities and arrhythmia-like events. These data, from 15 cell type × platform combinations, were then used to a construct a predictive model of TdP risk. Significant predictors in the hPSC-CMs were: arrhythmia-like event at any tested concentration, maximum prolongation at any tested concentration, and estimated prolongation at the clinical concentration of drug. These parameters in turn were fed into a composite model of TdP risk, with dichotomous outcomes (low, versus high or intermediate). The receiver operating characteristic (ROC) area under the curve (AUC) was 0.87, regardless of the cell line used or any local differences in culture method. In short, blinded studies have made it clear that measurements in hPSC-CMs are a highly reliable preclinical classifier of clinical TdP risk. Given such evidence, hiPSC-CMs have gained acceptance by industry and regulatory authorities as predictive of drug safety in humans, at least with respect to arrhythmic risk 25, 32, 44, 45 .

Two further initiatives with hPSC-CMs for safety pharmacology are the Japan iPS Cardiac Safety Assessment (JiCSA) 46, 47 and the CRACK IT InPulse Challenge [48] [49] [50] . Key lessons from JiCSA, which likewise is focused on arrhythmic risk, have included fidelity of the relationship between MEA-measured field potential duration and interspike interval in hPSC-CMs to the QT-RR relation deduced from clinical electrocardiograms in the Framingham Heart Study 46 . Human-relevant characteristics included repolarization delay at slow beating rates, reverse use-dependency, categorical analyses as potential indices of risk, and a threshold field potential duration that predicts early afterdepolarizations (EADs) and triggered activity, useful as a potential marker of risk for the onset of human Torsade de pointes 46 . A subsequent large-scale validation study of 60 compounds' torsadogenic risk markedly expanded the conclusions available from CiPA 32 , though not yet in blinded fashion. Findings were highly concordant in iCell and Cor.4U hPSC-CMs 47 , despite the lines' differences in ion channel and membrane transporter expression 51 .

The InPulse academic-industry consortium, by contrast, has emphasized developing a robust in vitro platform to monitor cardiac contractility, with cells that are phenotypically mature [48] [49] [50] . (See the Challenges and Prospects section for a detailed discussion of this concern.) This project's discoveries include a synthetic polymer that promotes the maturity of hPSC-CMs (a copolymer of isobornyl methacrylate and tert-butylamino-ethyl methacrylate), identified using combinatorial subtrate material microarrays 48 , novel open-source tools for quantifying cardiomyocyte contraction 49 , and proof that driving contractile work in hPSC-CMs induces mitochondrial biogenesis and recapitulates the normal post-natal shift to fatty acid oxidation 50 .

Other cardinal features of potential cardiotoxicity-notably, myocyte loss and ensuing heart failure-are likewise amenable to profiling in hPSC-CMs, with the cardiotoxicity of anti-cancer drugs being an especially robust area of investigation 26, [52] [53] [54] . Readily applicable assays include biochemical readouts (defective ATP production, cardiac troponin release), microscopy (loss of surface membrane integrity, apoptosis), and video imaging to assess Shown, by year, are the cumulative US Food and Drug Administration NME approvals for cancer chemotherapy (consistently, 20-25% of the total approvals each year), compared with the paucity of new drugs targeting cardiac muscle 110 . See text for details.

P. Golforoush and M.D. Schneider contractility (edge detection or deformation maps). As was done for arrhythmic risk, these quantitative parameters of toxicity can be integrated as a composite cardiac safety index 26, 54 . Interestingly, interpatient variations in heart tissue transcriptomics were largely recapitulated in patient-specific hPSC-CMs, including the NRF2-PPARGC1A pathway controlling oxidative stress, and correlated well with the patient-derived cells' functional difference in cardiotoxic responses 31 . In a related study, patient-specific hPSC-CMs reproduced the patients' respective vulnerability to trastuzumab 55 . More recently, a network-level comparison of cardiotoxicity in hPSC-CMs used RNA sequencing to distinguish differences in the pathways engaged by anthracyclines versus tyrosine kinase inhibitors 56 . Here, the principal finding was the categorically different transcriptomic signatures evoked by these anti-cancer agents in human cardiomyocytes. Specifically, whereas doxorubicin (DOX) induces pathways that initiate DNA damage, tyrosine kinase inhibitors disrupt mitochondrial energetics even at nonlethal concentrations, downregulating oxidative phosphorylation and upregulating glycolysis 56 , metabolic reprogramming that is a common feature of hypertrophied and failing hearts 57 . In agreement with DNA damage as the main target for DOX, genome editing to delete topoisomerase-II beta (TOP2B) markedly reduced the vulnerability of hPSC-CMs to DOX-induced DNA double strand breaks and cell death 53 .

As a potential step change beyond just safety assessment and risk prediction, might hPSC-CMs also be useful as human preclinical proof of efficacy, to guide and inform experimental therapeutics? In one early study of this kind, investigators sought to mitigate the known cardiotoxicity of DOX, a mainstay of cancer chemotherapy, as mentioned, targeting DOX selectively to breast cancer using liposomes conjugated with antibody against human epidermal growth factor receptor 2 (HER2) 58 . By this means, cardiotoxicity was virtually abolished in hPSC-CMs 58 , and a Phase 1 clinical doseescalation study confirmed the improvements expected on the basis of the pioneering human preclinical results 59 . No cardiac adverse events occurred in patients receiving this form of DOX alone, and the toxicity of combination therapy along with trastuzumab also was reduced 59 . Thus, human proof-of-principle was substantiated in hPSC-CMs (in this instance, cardiac safety of the targeted DOX), in advance of progression into human trials. Analogously, the cardiotoxicity of trastuzumab was found to be associated with defective energy metabolism in hPSC-CMs and was ameliorated by treatment with activators of AMP-activated protein kinase 55 , suggesting the potential for cardioprotection in this context by a current approved drug. A further success in translational relevance is the progress made using hPSC-CMs for drug repurposing, in particular thus far where based on the cells' fidelity to clinical phenotypes in certain hereditary heart disorders. This progress is perhaps most notable for personalized treatment of channelopathies such as long QT syndrome due to mutations in KCNH2 that disrupt intracellular trafficking of the hERG potassium channel Kv11.1 (LQT2). Lumacaftor (the protein chaperone VX-809) was evaluated in patient-specific hPSC-CMs, and shown to shorten the cell culture equivalent of QTc, as measured with multi-electrode arrays. This benefit was achieved solely in patients with Class 2 mutations 2 Enhancing cardiac drug discovery in hPSC-CMs. Diverse cardiac disorders have been successfully modeled in hPSC-CMs, beyond merely the patient-specific mutations for which this technology was first used, with broad applicability now demonstrated for the widespread, acquired forms of human heart disease. Concurrently, the readouts relevant to cardiac drug development have expanded beyond the arrhythmias first studied, to encompass the full spectrum of molecular and functional cardiomyocyte phenotypes including mechanical performance, energetics, myocyte formation, and myocyte loss. The impact on drug development has been manifested initially through more predictive safety pharmacology (including the improved profiling of non-cardiac drugs) and through human preclinical studies of approved agents, toward novel applications. In the development of novel agents, hPSC-CMs can augment not only target-based approaches, as platforms for validation by gene silencing and the investigation of new chemical series, but also as a human substrate for mechanistically agnostic, phenotype-driven screens. Diverse approaches promote cardiomyocyte maturation and fidelity to the adult human heart itself, which remains an acknowledged limitation of these models.

(which affect channel trafficking) but not in patients with Class 1 mutations (which affect channel synthesis) 60 . A clinically approved drug, Orkambi, combines Lumacaftor and Ivacaftor (a CFTR potentiator), rescues the analogous defect in patients with a homozygous CFTR-F508del mutation, and was taken forward into two patients with Class 2 mutations of LQT2, shortening QTc in both, as had been hypothesized 61 . Concomitantly, however, this landmark first-in-human study also acknowledged several "differences between the cellular model and clinical reality," which provide instructive caveats, including the magnitude of rescue achieved (much greater in hPSC-CMs than in the clinic) and expression of hERG (higher in hPSC-CMs than in native adult CMs) 61 .

What about cardiac drug development, more broadly than just reformulation or repurposing? Miniaturization of phenotypic assays to a 384-well format makes it possible to implement high-throughput chemical or genetic screens (HTS) for target validation and drug discovery, more rooted in human cardiac biology than has been possible heretofore. In an early example of tool-building toward high-throughput studies, hPSC-CMs were implemented to model myocarditis due to coxsackievirus B3, using CVB3-luciferase as an easy bioluminescent readout of virus proliferation 62 . Antiviral drugs including interferon-β1 and ribavirin were shown to suppress virus production. In addition, mechanistic insights were captured in this human cardiomyocyte milieu: by microarray profiling, interferon-β1 was shown to activate a network of downstream anti-viral genes including EIF2AK2, encoding protein kinase R, an inhibitor of viral mRNA translation 62 .

For cardiac muscle hypertrophy, one readily assayable readout is the induction of brain natriuretic peptide, a highly dynamic protein with especially strong diagnostic and prognostic significance in the clinical setting 63 . Its induction is triggered in hPSC-CMs by endothelin-1, much as in non-human models of pathological hypertrophy, and its detection by ELISA or highcontent imaging can be multiplexed with other parameters such as increased cardiomyocyte size 64 . In an early pilot phenotypic screen, hPSC-CMs were provoked at a saturating concentration of endothelin-1, and were treated in quadruplicate at 10 concentrations with candidate inhibitors including the calcium channel blocker verapamil, a PI3K-mTOR inhibitor, BEZ-235, and a broadspectrum histone deacetylase inhibitor 64 . A design feature worth noting was the use of serum-free fatty acid-supplemented media, to accelerate cardiomyocyte maturation 64 . More fundamentally, these experiments demonstrated that inhibitor profiling in this human platform was robust and workable at scale. 

Many logical targets have failed in clinical trials for heart disease, perhaps especially those aiming to enhance cardiac muscle cell survival after myocardial infarction [10] [11] [12] [13] [14] [15] . While specific shortcomings in trial design or implementation are sometimes culpable, what these failed trials have in common uniformly is that none was based on proof of efficacy in human preclinical studies, before proceeding into the clinic. The use of hPSC-CMs toward drug discovery for cardioprotection was championed in our recent study creating novel inhibitors of the stress-activated kinase MAP4K4 (mitogen-activated protein kinase kinase kinase kinase-4) 65 , an upstream member of the MAPK superfamily with connections to Jun N-terminal kinase 66, 67 and NFkB 68 , as well as to several non-canonical effectors as substrates 69, 70 . The scientific case for MAP4K4 as a druggable target in heart muscle cell death began with human tissue characterization, finding that myocardial MAP4K4 was activated in end-stage heart failure regardless of cause (dilated, hypertrophic, ischemic, and anthracycline cardiomyopathy). MAP4K4 was likewise activated in a range of disease models in adult mouse myocardium and cultured rat cardiomyocytes, including ischemia/reperfusion injury and H 2 O 2 as a surrogate oxidative stress. The causal role of MAP4K4 suggested by these observations was then corroborated using transgenic over-expression in mice, plus gain-of-function mutations, dominant-negative mutations, and gene silencing in rodent cardiomyocytes. Yet, even collectively, these methods-typical of the toolkit for academic target validation-leave altogether unanswered the question of whether MAP4K4 is a mechanistically sound therapeutic target in human heart muscle cell death. Consequently, using hPSC-CMs as a human platform for target validation and proof-of-concept studies, the requirement for MAP4K4 in human cardiac cell death was affirmed by gene silencing, giving uniquely direct impetus to a small-molecule discovery program 65 . Throughout, several high-throughput platforms were utilized, including automated microscopy (DRAQ7 uptake, TUNEL staining for apoptosis), ATP generation, and cardiac troponin release. Protection was conferred in three wholly independent human cardiomyocyte lines, suggesting not only the reliability of hPSC-CMs as a model but also the unvarying dependence on MAP4K4 in the tested forms of cardiac cell death. Beyond these key readouts of viability, protective effects of inhibiting MAP4K4 were also proven under sublethal stress, using the Seahorse extracellular flux (XF) method to study mitochondrial function and FLIPR assays to measure calcium cycling. Cardiomyocyte viability and function (auxotonic force) were even preserved in human 3D engineered heart tissue 65 , a model with further maturity of structure and physiological properties 71 . Importantly, the pathway and compounds developed in hPSC-CMs were substantiated further by proof-of-concept studies in mice, with the MAP4K4 inhibitor reducing infarct size by more than 55% in blinded studies, even given an hour after injury 65 . These MAP4K4 studies demonstrate the pivotal role played by hPSC-CMs in validating a suspected target by gene silencing, then building a small-molecule program upon efficacy proven in this human platform.

Will treatments devised in human models be more likely to succeed than prior ones, upon eventual testing in the clinic? This overall hypothesis-the crux of using hPSC-CMs as a model in drug discovery-will require a decade or more to resolve empirically, as exemplars of this class progress into human trials. However, it is tantalizing to apply the "retrospectoscope": examining the outcome, in human cardiomyocytes, of manipulating pathways whose success or failure is already known in reducing infarct size. Indeed, from this perspective, the potential predictive value of studies in hPSC-CMs is suggested by finding that human cardiac muscle cell death can be suppressed experimentally by β-adrenergic blockade (as found in METOCARD-CNIC 72,73 ) but not by inhibiting p38 MAPK (as was ultimately true in SOLSTICE 14 ) .

By contrast to more reductionist but biased target-based approaches, phenotype-driven screens, discussed here and in the following sections, require no a priori assumption about the drug's target, can pursue wholly novel or unanticipated mechanisms of action, engage the cell type-specific signaling and transcriptional context, and encompass diverse endogenous readouts of the disease 74 -including cardiomyocyte survival, but also myocyte function and creation.

In a mechanistically agnostic version of the MAP4K4 cell death studies detailed above, an unbiased chemical biology screen of nearly 50,000 small molecules was performed to identify and validate compounds that protect hPSC-CMs from H 2 O 2 75 . Cardiomyocyte protection was demonstrated using ATP content as the primary end-point and cellular impedance as the secondary readout, a measure of monolayer integrity and contractile function. From this screen, in 1536-well format, using a 35% improvement in viability as the criterion, 220 hits were identified of which half were confirmed upon retesting. A novel compound, designated cardioprotectant-312, was found to protect hPSC-CMs from the oxidative stress of H 2 O 2 , triggering upregulation of the essential endogenous anti-oxidant, heme oxygenase-1 75 . Of note, no protection was seen in rat H9c2 cells, embryonic heart-derived myoblasts that are commonly used in toxicology research; conversely, none of the authors' earlier lead compounds identified in H9c2 cells 76 was protective in human cardiomyocytes. Together, these vivid reciprocal disparities highlight the importance of implementing human preclinical cardiac models, in lieu of basing human trials solely on results obtained in rodent cardiomyocytes alone.

The power of hPSC-CMs for drug discovery to alleviate heart failure was demonstrated in two studies using patient-specific iPSCs to model diabetic cardiomyopathy 57, 77 . To mimic the diabetic-like environment, normal hPSCs were cultured in a maturation medium to promote the substrate utilization of adult ventricular myocytes (fatty acid β-oxidation), then were subjected to glucose excess, plus endothelin-1 and cortisol 77 , mimicking the systemic environment. The combined effect was marked induction of brain natriuretic peptide (BNP), other molecular markers of cardiac hypertrophy, and myocyte enlargement itself. Functional abnormalities included less frequent Ca 2+ transients, decreased beat amplitude, and increased beat irregularity. Lipid accumulation and peroxidation were other associated findings. The cardiomyopathic phenotype was captured, too, in cardiomyocytes derived from diabetic patient-specific iPSCs, even in the absence of the diabetic milieu 77 .

This model was then used as a phenotypic drug screening platform, determining the success of a therapeutic compound as monitored by BNP production, nuclear area, and sarcomere organization (α-actinin staining). From a library of 480 compounds, 47 were identified that improved all three of these disease parameters 77 . These compounds encompassed diverse modes of action, including regulators of Ca 2+ homeostasis (fluspirilene, thapsigargin, and the calmodulin inhibitor W7), Na + and K + channel blockers, phosphodiesterase inhibitors, and multiple protein kinase inhibitors (H89, K252a, SB202190 ). From this screen, in turn, compounds were found that rescued the reduction of Ca 2+ transients in cardiomyocytes subjected to the diabetic milieu and improved the phenotype of cardiomyocytes from diabetic P. Golforoush and M.D. Schneider patients. Analogous improvement of diabetic phenotypes in hPSC-CMs were elicited by empagliflozin, an inhibitor of sodium-glucose co-transporters that are upregulated in diabetes, possibly explaining the unexpectedly improved cardiovascular mortality in trials of this compound for glycemic control 78 . Together, these findings support the utility of hPSC-CMs as models not merely for simple monogenic disorders, on the one hand, and for wild-type cardiomyocytes' responses to lethal stress, on the other, but even for diabetic cardiomyopathy, a highly complex polygenic disease.

Ultimately, myocardial infarction can be viewed as a "myocytedeficiency disease" whose phenotype is determined not just by the extent of myocyte death but also by the lack of functionally significant restorative growth. Indeed, the plausible clinical benefits of hPSC-CMs very clearly include therapeutic grafting, as cell therapy 35, 79, 80 , Recently, though, advances in understanding the fundamental biology of cardiac growth arrest have pointed to greater plasticity that was formerly evocable, with significant potential for restarting the cardiac cell cycle therapeutically, at least in model organisms 81, 82 . Might induced proliferation as a route to heart repair also be amenable to exploration or triage in hPSC-CMs? As one starting point, functional screening of more than 10,000 hPSC-CM organoids was undertaken to optimize diverse aspects of the culture milieu, resulting in enhanced maturation and recapitulation of the adult heart's notorious resistance to cell cycling, driven by a shift to fatty acid oxidation 83 . Conversely, this successful model of implementing cardiac cell cycle arrest in hPSC-CMs was then subjected to a functional screen of 105 small molecules, resulting in the identification of novel cell cycle activators, working through the mevalonate pathway 84 . Thus, notwithstanding the potential immaturity of pluripotent cellderived myocytes with respect to cell cycle control, a post-mitotic phenotype could be imposed experimentally, and means to override it discovered.

The research advances reviewed here highlight the many benefits of hPSC-CMs as a transformative human platform for cardiac drug discovery 74 -accessible, scalable, faithful by a large number of clinically relevant parameters, amenable to genetic engineering, amenable to tissue engineering, able to capture patient variations, predictive of clinical success at least retrospectively, and predictive prospectively at least of success in whole-animal studies ( Fig. 2 ; Table 1 ). One must, of course, temper optimism with caution. For the moment, it remains a supposition that new drugs developed by this route will achieve clinical success more reliably than drugs lacking human preclinical proof of effect. But, this hypothesis is testable-even if needing years of accumulated experience to do-while drawing ample credence from the established precedents and practices in cancer chemotherapy.

How might the predictive power of hPSC-CMs be augmented or ensured? Despite the predictive power shown even with routine 2D models, it is clear that existing lines-or, more accurately, their current embodiment in tissue culture-do not suffice to model all possible phenotypes of concern. The many acknowledged shortcomings, which have been mitigated to date only partially, include morphology (lack of sarcomere organization, T-tubules, and the normal mitochondrial density), molecular profile (weak expression of maturation-associated genes and splicing isoforms), metabolism (glycolysis, not fatty acid oxidation), contractility (lower maximum contractile force), and electrophysiology (lesser action potential upstroke velocity and amplitude) [48] [49] [50] [51] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] . For instance, by comparison to adult human hearts, commercially available hPSC-CMs were uniformly deficient in the expression of KCNJ2, with higher than normal expression of HCN4 and large line-to-line variations in the other ion channels and membrane transporters assayed (CACNA1C, KCNH2, KCNQ1, SLC8A1, ATP1A1, ATP2A2) 51 . As a consequence, some pharmacological and pathobiological responses can be deficient or anomalous. In some reports, hPSC-CMs do not mirror the TdP risk of drugs with late sodium current effects, like ranolazine 32 , and hiPSC-CMs were less sensitive to hypoxia/reoxygenation than to other death signals 65, 96, 97 . Such disparities must be taken into account, whether inherent shortcomings or idiosyncratic.

Efforts to enhance (further) the stem cell-derived myocytes' predictive value center on manipulating chamber and cell subtype specificity on the one hand, and on improving structural and functional maturity on the other. Purely pharmacological efforts at enhancing maturation in routine 2D culture include fatty acids 85, 86 , thyroid hormone 87 , and inhibition of mTOR 88 . In another approach, transduction of the defectively expressed gene KCNJ2 has been applied to rescue the channel levels and promote aspects of fidelity directly 98 . More generally, however, the procedures of most proven value to enhance the maturity of hPSC-CMs include the use of 3D human engineered heart tissue (EHT), mechanical or electrical conditioning, and heart-on-chip technologies, advances discussed at length elsewhere [89] [90] [91] [92] [93] [94] [95] . The tissue engineering solutions to create more heart-like phenotypes in hPSC-CMs range in complexity from micropatterned 2D substrates to scaffolds, organoids, microfluidics, 3D bioprinting, and even the construction of hollow spheres. Apart from just geometry, key elements of these tissue engineering strategies notably include cyclic electrical or mechanical stimulation. Incorporation of other cell types can promote maturity or function, as well as the microvascularization required for oxygen delivery at larger scale than mere diffusion can confer. Indeed, 3D spheroids composed of hPSC-CMs plus hPDSC-derived endothelial cells showed progressive changes in gene expression typical of postnatal development 99 . Analogously, 3D culture of hPSC-CMs as engineered heart tissue in concert with cardiac fibroblasts, combined with long-term electrical stimulation, enables the development of physiological responses that are absent from the cardiomyocytes cultured routinely 93, 100 . The reported adultlike properties included a positive force-frequency relationship, postrest potentiation of force, and inotropic beta-adrenergic responses to isoproterenol and dobutamine, as well as other compounds and pathways tested 93, 100 .

Although these collective efforts should be viewed with enthusiasm, progress toward standardization is confounded by the diversity of available hPSC lines, stem cell and differentiation media, physical substrata, timing, purification methods, presence or absence of serum, cell density (syncytial sheets, versus sporadic single cells), the heterogenous mixtures of cell types (atrial, ventricular, pacemaker cells, non-cardiomyocytes), and even lotto-lot variation in ostensibly standardized cells. However, though procedures exist for the selective production of ventricular myocytes versus atrial myocytes or conduction system cells 93,101-104 , even these simple advances are not yet exploited uniformly. Ultimately, it is plausible that the drivers of standardization will include the emergence of consensus best practice solutions, but also adherence within the scientific community to procedures validated by multi-site initiatives such as CiPA, JiCSA, and InPulse, functioning as exemplars.

A complementary approach-obvious once the issue is raisedis also to improve the breadth of human cardiomyocytes surveyed, including but not limited to demographic features like gender and ethnic background. To illustrate, genetic determinants of susceptibility include a polymorphism in ALDH2 that predominates in East Asians and renders the carriers' myocytes much more vulnerable to ischemic heart damage 105 . Analogously, screening a single commercial line failed to capture the known cardiotoxicity of rosiglitazone, an anti-diabetic drug that enhances PPARγ activity but can exacerbate heart failure 106 . Indeed, marked inter-patient variation was later found in the response of hPSC-CMs to rosiglitazone, including adverse effects on reactive oxygen and nitrogen species, associated with divergent transcriptomic signatures that relate to NRF2-mediated oxidative stress 31 . A third demographic axis, aging, may be resistant to capture in hPSCderived models, given the rejuvenation signature imparted by reprogramming to a primitive, pluripotent state; this aspect might be addressable using directly induced cardiomyocytes, instead 107,108 , for which a precedent is the success using forward programming to model age-related neurodegeneration 109 . Much like trials in the real world, clinical trials in a dish may need to take patient recruitment into account-along with their choices of compound, regimen, and readout-in developing robust new counter-measures to combat human heart disease.

Received: 29 October 2019; Accepted: 6 February 2020;

Global, regional, and national disability-adjusted life-years (DALYs) for 359 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study

Global, regional, and national agesex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study

Cardiovascular drug development: is it dead or just hibernating?

Cardiovascular drug discovery: a perspective from a research-based pharmaceutical company

Angiotensin receptor neprilysin inhibition compared with enalapril on the risk of clinical progression in surviving patients with heart failure

Ivabradine for patients with stable coronary artery disease and left-ventricular systolic dysfunction (BEAU-TIFUL): a randomised, double-blind, placebo-controlled trial

Effects on outcomes of heart rate reduction by ivabradine in patients with congestive heart failure: is there an influence of beta-blocker dose?: findings from the SHIFT (Systolic Heart failure treatment with the I(f) inhibitor ivabradine Trial) study

Cardiovascular function and treatment in beta-thalassemia major: a consensus statement from the

Tafamidis in transthyretin amyloid cardiomyopathy: effects on transthyretin stabilization and clinical outcomes

Cardioprotection: chances and challenges of its translation to the clinic

Targeting myocardial reperfusion Injury-the search continues

Effect of cyclosporine on reperfusion injury in acute myocardial infarction

Inhibition of delta-protein kinase C by delcasertib as an adjunct to primary percutaneous coronary intervention for acute anterior STsegment elevation myocardial infarction: results of the PROTECTION AMI Randomized Controlled Trial

Losmapimod, a novel p38 mitogen-activated protein kinase inhibitor, in non-ST-segment elevation myocardial infarction: a randomised phase 2 trial

Cyclosporine A in reperfused myocardial infarction: the multicenter, controlled, open-label CYCLE trial

The productivity crisis in pharmaceutical R&D

Cell line-based platforms to evaluate the therapeutic efficacy of candidate anticancer agents

The clinical relevance of cancer cell lines

From mice to men and back: an assessment of preclinical model systems for the study of lung cancers

The NCI60 human tumour cell line anticancer drug screen

Correlating chemical sensitivity and basal gene expression reveals mechanism of action

Next-generation characterization of the cancer cell line encyclopedia

Human stem cells for modeling heart disease and for drug discovery

Induced pluripotent stem cells: the new patient?

The evolving roles of human iPSC-derived cardiomyocytes in drug safety and discovery

High-throughput screening of tyrosine kinase inhibitor cardiotoxicity with human induced pluripotent stem cells

Human induced pluripotent stem cell-derived cardiomyocytes recapitulate the predilection of breast cancer patients to doxorubicininduced cardiotoxicity

Drug screening using a library of human induced pluripotent stem cell-derived cardiomyocytes reveals disease-specific patterns of cardiotoxicity

Identification of a Titin-derived HLA-A1-presented peptide as a cross-reactive target for engineered MAGE A3-directed T cells

Modeling treatment response for Lamin A/C related dilated cardiomyopathy in human induced pluripotent stem cells

Transcriptome profiling of patient-specific human iPSCcardiomyocytes predicts individual drug safety and efficacy responses in vitro

International multisite study of human-induced pluripotent stem cell-derived cardiomyocytes for drug proarrhythmic potential assessment

Human-mouse chimerism validates human stem cell pluripotency

Stage-specific optimization of activin/nodal and BMP signaling promotes cardiac differentiation of mouse and human pluripotent stem cell lines

Human embryonic stem cell-derived cardiomyocytes restore function in infarcted hearts of non-human primates

Induction of pluripotent stem cells from adult human fibroblasts by defined factors

Induced pluripotent stem cells: at the heart of cardiovascular precision medicine

Cardiomyocytes from human pluripotent stem cells: from laboratory curiosity to industrial biomedical platform

Cardionomics: a new integrative approach for screening cardiotoxicity of drug candidates

Estimating the risk of drug-induced proarrhythmia using human induced pluripotent stem cell-derived cardiomyocytes

Comparison of electrophysiological data from human-induced pluripotent stem cell-derived cardiomyocytes to functional preclinical safety assays

Repolarization reserve determines drug responses in human pluripotent stem cell derived cardiomyocytes

Evolution of strategies to improve preclinical cardiac safety testing

Integrating cardiomyocytes from human pluripotent stem cells in safety pharmacology: has the time come?

Human induced pluripotent stem cell (hiPSC)-derived cells to assess drug cardiotoxicity: opportunities and problems

Proarrhythmia risk prediction using human induced pluripotent stem cell-derived cardiomyocytes

Development of torsadogenic risk assessment using human induced pluripotent stem cellderived cardiomyocytes: Japan iPS Cardiac Safety Assessment (JiCSA) update

A defined synthetic substrate for serum-free culture of human stem cell derived cardiomyocytes with improved functional maturity identified using combinatorial materials microarrays

MUSCLEMOTION: a versatile open software tool to quantify cardiomyocyte and cardiac muscle contraction in vitro and in vivo

Contractile work contributes to maturation of energy metabolism in hiPSC-derived cardiomyocytes

Systematic expression analysis of genes related to generation of action potentials in human iPS cell-derived cardiomyocytes

Identification of the molecular basis of doxorubicin-induced cardiotoxicity

Modeling doxorubicin-induced cardiotoxicity in human pluripotent stem cell derived-cardiomyocytes

Use of human induced pluripotent stem cell-derived cardiomyocytes to assess drug cardiotoxicity

Human-induced pluripotent stem cell model of trastuzumabinduced cardiac dysfunction in patients with breast cancer

Adaptation of human iPSC-derived cardiomyocytes to tyrosine kinase inhibitors reduces acute cardiotoxicity via metabolic reprogramming

Loss of metabolic flexibility in the failing heart

HER2-targeted liposomal doxorubicin displays enhanced anti-tumorigenic effects without associated cardiotoxicity

Safety and pharmacokinetics of MM-302, a HER2-targeted antibody-liposomal doxorubicin conjugate, in patients with advanced HER2-positive breast cancer: a phase 1 dose-escalation study

Identification of a targeted and testable antiarrhythmic therapy for long-QT syndrome type 2 using a patient-specific cellular model

From patient-specific induced pluripotent stem cells to clinical translation in long QT syndrome Type 2

Human induced pluripotent stem cell-derived cardiomyocytes as an in vitro model for coxsackievirus B3-induced myocarditis and antiviral drug screening platform

Heart failure association of the European Society of Cardiology practical guidance on the use of natriuretic peptide concentrations

Phenotypic screening with human iPS cell-derived cardiomyocytes: HTS-compatible assays for interrogating cardiac hypertrophy

MAP4K4 inhibition promotes survival of human stem cellderived cardiomyocytes and reduces infarct size in vivo

MAP4K4 activation mediates motor neuron degeneration in amyotrophic lateral sclerosis

The Ste20 family kinases MAP4K4, MINK1 and TNIK, converge to regulate stress induced JNK signaling in neurons

Endothelial protein kinase MAP4K4 promotes vascular inflammation and atherosclerosis

HGK/MAP4K4 deficiency induces TRAF2 stabilization and Th17 differentiation leading to insulin resistance

MAP4K family kinases act in parallel to MST1/2 to activate LATS1/ 2 in the Hippo pathway

Engineering cardiac muscle tissue: A maturating field of research

Effect of early metoprolol on infarct size in ST-segment-elevation myocardial infarction patients undergoing primary percutaneous coronary intervention: the Effect of Metoprolol in Cardioprotection During an Acute Myocardial Infarction (METOCARD-CNIC) trial

Long-term benefit of early pre-reperfusion metoprolol administration in patients with acute myocardial infarction: results from the METOCARD-CNIC trial (Effect of Metoprolol in Cardioprotection During an Acute Myocardial Infarction)

The convergence of stem cell technologies and phenotypic drug discovery

Discovery of novel small-molecule inducers of heme oxygenase-1 that protect human iPSC-derived cardiomyocytes from oxidative stress

Cardioprotective Inhibitors of Reperfusion Injury

Disease modeling and phenotypic drug screening for diabetic cardiomyopathy using human induced pluripotent stem cells

Empagliflozin ammeliorates high glucose induced-cardiac dysfuntion in human iPSC-derived cardiomyocytes

Human embryonic-stem-cell-derived cardiomyocytes regenerate non-human primate hearts

Harnessing cell pluripotency for cardiovascular regenerative medicine

Mending a faltering heart

Stimulating cardiogenesis as a treatment for heart failure

Functional screening in human cardiac organoids reveals a metabolic mechanism for cardiomyocyte cell cycle arrest

Drug screening in human PSC-cardiac organoids identifies proproliferative compounds acting via the mevalonate pathway

Studying arrhythmogenic right ventricular dysplasia with patientspecific iPSCs

Fatty acid-treated induced pluripotent stem cell-derived human cardiomyocytes exhibit adult cardiomyocyte-like energy metabolism phenotypes

Thyroid and glucocorticoid hormones promote functional Ttubule development in human-induced pluripotent stem cell-derived cardiomyocytes

Inhibition of mTOR signaling enhances maturation of cardiomyocytes derived from human induced pluripotent stem cells via p53-induced quiescence

Human iPSC-derived cardiomyocytes cultured in 3D engineered heart tissue show physiological upstroke velocity and sodium current density

Defined engineered human myocardium with advanced maturation for applications in heart failure modeling and repair

Modeling the mitochondrial cardiomyopathy of Barth syndrome with induced pluripotent stem cell and heart-on-chip technologies

Advanced maturation of human cardiac tissue grown from pluripotent stem cells

A platform for generation of chamber-specific cardiac tissues and disease modeling

A tissue-engineered scale model of the heart ventricle

Maturing human pluripotent stem cell-derived cardiomyocytes in human engineered cardiac tissues

In vitro models of ischemia-reperfusion injury

Modelling ischemia-reperfusion injury (IRI) in vitro using metabolically matured induced pluripotent stem cell-derived cardiomyocytes

Overexpression of KCNJ2 in induced pluripotent stem cell-derived cardiomyocytes for the assessment of QT-prolonging drugs

Three-dimensional cardiac microtissues composed of cardiomyocytes and endothelial cells co-differentiated from human pluripotent stem cells

Engineered cardiac tissues generated in the Biowire II: a platform for human-based drug discovery

A simple, cost-effective but highly efficient system for deriving ventricular cardiomyocytes from human pluripotent stem cells

Human pluripotent stem cell-derived atrial and ventricular cardiomyocytes develop from distinct mesoderm populations

Atrial-like cardiomyocytes from human pluripotent stem cells are a robust preclinical model for assessing atrial-selective pharmacology

Canonical Wnt signaling promotes pacemaker cell specification of cardiac mesodermal cells derived from mouse and human embryonic stem cells

Characterization of the molecular mechanisms underlying increased ischemic damage in the aldehyde dehydrogenase 2 genetic polymorphism using a human induced pluripotent stem cell model system

Structural and functional screening in human inducedpluripotent stem cell-derived cardiomyocytes accurately identifies cardiotoxicity of multiple drug types

Direct reprogramming of fibroblasts into functional cardiomyocytes by defined factors

Cardiac reprogramming factors synergistically activate genome-wide cardiogenic stage-specific enhancers

Aging in a dish: iPSC-derived and directly induced neurons for studying brain aging and age-related neurodegenerative diseases

US Food and Drug Administration. Novel Drug Approvals for 2019

High-throughput imaging of cardiac microtissues for the assessment of cardiac contraction during drug discovery

This work of the authors is supported by Wellcome Trust Seeding Drug Discovery awards (WT10638, WT205256) and Apollo Therapeutics.

P.G. and M.D.S. wrote and edited this review.

M.D.S. declares patent applications relating to MAP4K4 as a druggable target in human cardiac muscle cell death (UK Patent Applications 1716867.5, WO/2019/ 073253; 1819839.0; 1905472. 5; 1905476.6) . P.G. has no conflict of interest to disclose.

Correspondence and requests for materials should be addressed to M.D.S.Reprints and permission information is available at http://www.nature.com/ reprintsPublisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.