Modeling Arrhythmia in a Dish: An Open View from Human-Engineered Heart Constructs

1. Introduction

The heart is the first functional organ in the human embryo, garnering extensive interest in its development during embryogenesis and lifelong electrophysiological-mechanical coupling despite heart disease or aging. Various animal models have been utilized to explore the mysteries behind heart development and related diseases [1]. Still, these models pose challenges in elucidating mechanisms due to notable disparities in ion channels, membrane potential, and cardiac contraction frequency between animals and humans [2]. Additionally, primary adult human cardiomyocytes and engineered cells designed with specific ion channels for pharmacological testing inadequately represent the holistic efficacy of cardiac medications, primarily due to lacking intricate three-dimensional (3D) structures and functional characteristics of the myocardium [3,4]. The advent of human pluripotent stem cells (hPSC), including embryonic stem cells (hESC) and induced pluripotent stem cells (hiPSC), offers immense potential for advancing in vitro drug screening [5], developmental research, and regenerative medicine. Leveraging patient-derived hiPSC with specific genotypes and gene editing techniques facilitates the development of personalized human disease models to understand individual genotypes better and evaluate functional recovery. These in vitro models can address the intrinsic limitations of animal models, such as ethical concerns [6] and interspecies variability [7], and contribute to a better understanding of cardiac development and disease, ultimately tackling cardiovascular disease, the leading cause of global mortality [8].

The continuous enhancement of differentiation protocols for directing hPSCs into various types of human cardiac cells, such as cardiomyocytes(CM) [9,10], cardiac fibroblasts (FB) [11,12], and endothelial cells (ED) [13,14], has significantly advanced the development of in vitro human cardiac models. With the availability of human cardiac cells, the primary focus of human heart engineering has transitioned to fabricating cardiac tissue. Currently, two dominant in vitro heart models effectively replicate the heart’s cellular composition and physiological function: self-assembled cardiac spheroids and directionally assembled cardiac tissue. These models are collectively referred to as human-engineered heart constructs (hEHC) [15,16]. hEHC was selected as an overarching term over alternatives like cardiac organoid, cardiac microtissues, engineered heart tissue, or cardiopatches, as both self-assembled and directionally assembled approaches strive to utilize engineering technologies to orchestrate cardiac cells in forming synergistically tissue that partially mimics the heart’s structure and function.

The self-assembled hEHC shows increased structural heterogeneity under various stimuli and provides significant advantages in delineating early cardiogenesis and the multicellular interactions associated with congenital heart disease. Zhen Ma et al. pioneered the generation of 3D hEHC featuring a central CM core and outer FB perimeter and predominantly comprising cell-free microchambers [17,‍18]. Drakhlis et al. advanced hEHC complexity by embedding hPSC-derived embryoid bodies in Matrigel to form a myocardium including cavities, as well as an inner core expressing anterior foregut endoderm markers SOX2, and an outer layer expressing posterior foregut endoderm markers like HNF4α and PDX1 [19]. In contrast, the multicellular hEHC fabricated mainly through the directed assembly of cardiac cells, aimed to replicate certain electromechanical properties of adult myocardium in vitro. Selecting the most pertinent hEHC system—encompassing manufacturing strategies, genetic/non-genetic modeling factors, and functional assessment parameters—is paramount for cardiac-related investigations. Hence, this review elaborates on the evolutionary trajectory of hEHC over the past two decades, aiming to consolidate methodologies to effectively enhance the complexity and maturity of hEHC. Additionally, the representative application of hEHC in modeling arrhythmia, a prevalent phenotype associated with heart disease and drug-induced cardiotoxicity was highlighted, to ultimately advocate for a modular, multidimensional approach to constructing hEHC-based cardiac disease models.

2. Striving for Excellence: For More Mature and Complex hEHC

The heart is a highly sophisticated organ that continuously pumps blood to deliver oxygen and nutrients via the circulatory system. This complexity poses challenges for in vitro models attempting to replicate the intricate architecture and electromechanical coupling of the myocardium. In 1960, Harary’s team discovered that rat heart cells could self-aggregate to form beating fiber-like heart tissue in vitro [20]. This self-assembly property led to the initial cultivation of rat heart cells capable of producing defined, beating masses known as “mini-hearts” in the 1970s [21]. Concurrently, embryonic chick heart cells were cultured on an oriented substrate to form aligned muscle cell cores enveloped by an outer sheath of FB-like cells, also called cardiac muscle strands or bundles [22], thus laying the groundwork for heart tissue engineering. By the 1990s, Thomas’s group developed a 3D collagen-based heart muscle model system [23], formally establishing the notion of engineered heart tissue in 2000 [24]. Undeniably, CM are the protagonists in executing the heart’s function, comprising approximately 70% of the myocardial volume but only 20–30% of the total cellular composition [25,26]. The remaining 70% consists of interstitial cells such as FB, ED, epicardial cells (EP), and immune cells, which play indispensable roles in maintaining myocardial structure and functionality [27,28].

The initial impetus for introducing additional cells into hEHC was to improve their angiogenic capabilities in defective myocardial areas. In 2007, Oren et al. pioneered vascularized hEHC by introducing hESC-derived cardiomyocytes (hESC-CM), endothelial cells (hESC-ED), and embryonic FB into the poly L-lactic acid/polylactic glycolic acid scaffolds [29]. The integration of hESC-EC significantly enhanced vascularization and CM proliferation, while FB bolstered both the ED survival and proliferation capacity within hEHC [29]. In parallel, adding ED and FB not only augmented the passive mechanical force of the hEHC but also enhanced its capacity to generate myocardium and microvessels post-implantation in rodent hearts [30]. In 2013, Donghui et al. first investigated the impact of these additional cells on hEHC’s electromechanical characteristics, revealing that higher purities of CM (SIRPA⁺ cells >90%) increased conduction velocities (up to 25.1 cm/s), although individual CM contractile force decrease [31]. To optimize hEHC design, Nimalan et al. developed hEHC (including directly assembled microwire and self-assembling aggregates) of varying ratios of CM-to-FB (100%, 75%, 50%, and 25% of NKX2.5-GFP⁺ CM, with the remainder comprising CD90⁺ FBs) and discovered that a 75% CM ratio resulted in more stable architecture and improved sarcomere alignment. They noted that the absence of FB and FB-mediated tissue remodeling led to structural instability in hEHC, while an excess FB triggered fibrosis and spatial heterogeneity [32]. Additionally, hEHC subjected to uniaxial mechanical stresses exhibited mature myocardium gene expression and a conduction velocity comparable to that of a healthy human heart, contrasting with self-assembling hEHC [32] (Table 1).

To assess the influence of diverse interstitial cell types on remodeling and cellular dynamics within hEHC, Meredith et al. constructed hEHC incorporating four interstitial cell populations. They found that the interstitial cells with optimal extracellular matrix (ECM) remodeling capabilities, such as dermal FB, significantly enhance CM maturation and alignment in hEHC [33]. Tiburcy et al. developed a serum-free method for hEHC construction [34,35], confirming that a CM to FB ratio of 0.7:0.3 could maximize contractile forces. They further highlighted that adding TGF-β1 during the early consolidation phases of hEHC substantially improves functional outcomes via FB-mediated ECM remodeling. Introducing hESC-derived epicardial cells (hESC-EP) to hEHC also promoted structural compaction and functional maturation [36], attributed to their embryonic FB-like characteristics and elevated fibronectin synthesis [36]. Notably, epicardial-derived FB can serve as a reservoir for cyclic guanosine monophosphate (GMP), boosting cyclic adenosine monophosphate (AMP) levels in CM via connexin 43-mediated gap junctions, thereby promoting CM maturation in tri-cellular hEHC [37,38]. ED within hEHC secrete specialized ECM components, like laminin-521, bolstering hEHC contractility through paracrine PDGFRβ signaling [39] (Table 1). The quantifiable presence of defined non-cardiomyocytes(non-CM), particularly EP and FB, beneficial for advancing the hEHC structural and functional maturation [37,38,40,41]. Beyond integrating non-CM into hEHC to enhance overall maturation, there exist more straightforward strategies targeting CM to finally manifest the traits of adult myocardium in hEHC. Extending culturing time from one to three weeks [42], or utilizing the fatty-acid-enriched media [43], can effectively boost CM connectivity and mitochondrial function, achieving peak contractility of 30 mN/mm², which is comparable to the 25–44 mN/mm² range observed in adult human myocardium [42]. Moreover, electrical stimulation [44,45] or mechanical stretching [46,47] can directly “train” the CM in hEHC to further mature. Noteworthy is the progressive increase of the electrical stimulation frequency from 2 Hz to 6 Hz, which enables CM to exhibit physiologic sarcomere length (2.2 μm) and large, well-organized T-tubules [44]. Nonetheless, achieving adult-level CM maturation in hEHC presents considerable challenges [48], particularly the efforts to enhance the conduction velocity of electrical signaling to align with the 46.4 cm/s found in adult myocardium [49]. Additionally, contemporary hEHC lacks the intricate complexity inherent to the native heart, including the endocardial, epicardial, immune, and nervous systems [50]. Thus, a deeper understanding of the interactions between CM and non-CM, along with their modulation, will be pivotal in developing hEHC with enhanced CM maturity and a cellular composition and organizational structure closer to the adult heart.

Table 1. Characterization of representative hEHC.

3. Spin Silk from Cocoons: Arrhythmia in hEHC-Based Cardiac Disease Modeling

Cardiac arrhythmia is an irregular heart rate caused by various diseases such as myocardial infarction, diabetic cardiomyopathy, and congenital heart disease, potentially leading to heart failure [51]. Additionally, drug toxicity and conditions such as myocarditis can also cause life-threatening ventricular arrhythmias [52,53]. These conditions are all characterized by disorganized electromechanical coupling from fibrotic scar, particularly in the ventricular myocardium, which may result in sudden death in severe cases. Due to their significant risk and unpredictability, the diagnosis and pathogenesis of arrhythmias have consistently been research hotspots [54,55]. Early afterdepolarization (EAD) is crucial in the onset of Long-QT syndrome (LQTS) and can lead to life-threatening Torsade de Pointes (TdP) arrhythmia. When assessing the effects of pharmacological agents or gene mutations related to LQTS on the EAD frequency, hEHC showed similar action potential characteristics and reverse use-dependence to adult left ventricular myocardium, unlike guinea pigs or rabbit models [56]. Nonetheless, the diminished coupling of CM in hEHC, due to relatively lower cellular density, alongside increased calcium influx from the immature hPSC-CM, predisposes them to a higher susceptibility to arrhythmic triggers than LV [56,57]. This susceptibility difference was also found between atrial CM-derived hEHC and ventricular CM-derived hEHC, where only atrial-hEHC showed spontaneous single or multiple spiral-wave reentrant loops. These electrophysiological reentrants can be rescued to normal rhythm by anti-arrhythmic drugs [58]. Notably, Masahide et al. first visualized the occurrence of reentrants characterized by featuring a meandering spiral wave center, exemplifying polymorphic ventricular tachycardia in hEHC. They proved that the accentuation of the 3D heterogeneity mediated by non-CM within hEHC is a prerequisite for the manifestation of TdP-like arrhythmias [59] (Table 2).

Arrhythmogenic cardiomyopathy patient-derived hiPSC-FB carrying the PKP2 mutation, but not the corresponding hiPSC-CM, significantly impaired the capacity of hEHC to capture high-frequency stimuli, thereby promoting an arrhythmic phenotype [37]. Furthermore, hEHC fabricated from LQTS patient-derived hPSC-CM (LQTS-CM) or FB (LQTS-FB) exhibited a prolonged action potential duration compared to normal hEHC, but both were shorter than that in hEHC fabricated using both LQTS-CMs and LQTS-FBs, indicating that the synergistic interaction between disease-type FB and CM contributes to a more realistic representation of LQTS characteristics [60]. The electrical coupling between FB and CM plays a vital role in cardiac excitability and arrhythmogenesis in animal models [61,62], which undoubtedly provides solid theoretical insights to the further clarification of the non-CM composition of hEHC, particularly the highly heterogeneous FB and their interactions with CM during disease progression. The reticular structures formed by EC in self-assembled hEHC generated anisotropic stress gradients, resulting in the formation of multiple chambers within hEHC. Utilizing a micro-physiological system that simultaneously evaluates oxygen consumption, extracellular field potentials, and contraction of these multicellular hEHC, Mohammad et al. found that mitochondrial calcium oscillations triggered by MCU genetic deletions can induce arrhythmias [63] (Table 2). Additionally, optogenetic-based chronic intermittent tachypacing [64] and cryoinjury [65] also can serve as reproducible non-genetic arrhythmic triggers. When weakened the goal of mimicking tissue-level arrhythmias such as reentry, the self-assembled, micro hEHC (approximately 400 μm diameter) exhibits greater batch stability for large-scale fabrication and assessment of action potential properties via optical mapping in response to various drugs [66], but of course, this miniaturization compromises CM maturation and electromechanical coupling heterogeneity across spatial dimensions, as illustrated by Kenneth et al. using a dual-camera to characterize the varying calcium signals on the upper and lower surfaces of directionally assembled arrhythmic hEHC [67]. The choice between prioritizing the number of assessment subjects using self-assembled hEHC versus the quality of assessment outcomes using directionally assembled hEHC to predict potential arrhythmic risk from genetic or non-genetic factors is like two sides of the same coin.

Table 2. Characterization of representative hEHC-based arrhythmia models.

4. Where to Go Next: Accurate Modeling and Prediction

To develop an effective in vitro cardiac model for adaptation research and practical large-scale applications, four essential features need to be considered: (1) Using human-derived cellular components that are consistently sourced and produced under strict quality control; (2) Incorporating a 3D structure resembling adult myocardium, which contains various cell types suitable for investigating cell interaction; (3) Including multiple measurable parameters related to clinical cardiac functionality; (4) Allowing integration with automated, high-throughput, and standardized testing. When developing hEHC-based arrhythmia models, prioritizing the enhancement of CM maturity and hEHC complexity is crucial. Adult CM do not exhibit spontaneous excitation and have dyad structures that swiftly react to membrane potential changes, thereby facilitating excitation-contraction coupling, indicating the higher presence of regulatory proteins in matured CM that maintain their stable resting membrane potential and support high levels of calcium homeostasis [68]. Consequently, advancing hEHC maturation minimizes false positives in the drug toxicity assessment [43] and diminishes the occurrence of irregular arrhythmia phenotypes in disease models [69]. Furthermore, the ion channel isoforms present in fetal-like CM mask the arrhythmic phenotype associated with mutations in adult myocardium isoforms, such as the SCN5A gene, which encodes the α-subunit of the cardiac sodium channel Nav1.5, mutations in its adult exon 6 can trigger severe tachycardia and conduction disorders [70]. The multicellular composition that reflects the hEHC complexity also necessitates the consideration of multi-system interactions. For instance, integrating atria-like, atrioventricular canal-like, and ventricles-like hEHC can effectively replicate the heart’s unidirectional conduction and “fast-slow-fast” activation pattern, inclusive of the associated atrioventricular conduction block [71]. Furthermore, the emergent concept of electroimmunology [51] underscores the essential role of immune cells in modulating myocardial electrophysiology in both physiological and pathological conditions. Macrophages, for example, can expedite CM repolarization through connexin-43 coupling, thereby enhancing electrical conduction within the atrioventricular node [72]. Additionally, exploring the influence of vagal and sympathetic nerve interactions on cardiac electrophysiologic signals [73,74] based on multicellular and multi-system hEHC represents a promising and intriguing area of research. In light of the intricate system delineated above, additional priority is to dissect the disease into interconnected genetic and non-genetic factors for acute simulations (Figure 1), as heart diseases can stem from genetic and non-genetic interactions. Genetic factors can include mutations in functional genes and changes in non-coding regions. For cardiomyopathy, gene mutations in the cellular cytoskeleton, metabolism, and ion channels can lead to structural and functional abnormalities.

Figure 1. Myocardium arrhythmias induced by fibrotic scarring are common events in various heart diseases.

Multiple pathogenic factors, such as atherosclerosis, hyperglycemia, infections, and genetic alterations, prompt myocardial fibrosis via cardiac FB. The resultant dense fibrotic scar, lacking the electrophysiological characteristics akin to those of oriented myocardium, obstructs the consistent electrical signal propagation, causing localized conduction block or reentry. Concurrently, the FB at the margins of the fibrotic scar can connect CM via synaptic or gap junctions, resulting in CM asynchronous activation, thereby heightening the susceptibility to arrhythmias. This arrhythmia phenotype can be replicated in hPSC-derived-hEHC. To benchmark the pathogenic factors causing arrhythmia, both unifactorial and multifactorial modeling approaches in hEHC, including drug cardiotoxicity, elevated glucose/lipid cultures, virus/bacterial infections, aging factor, and genetic modifications related to ion channel, mitochondrial, cytoskeleton, reveal varying degrees of electrophysiological abnormalities, including conduction blocks and reentry circuits, which, in turn, provides a valuable humanized platform for studying arrhythmogenesis and potential therapeutic targets.

Somatic cell reprogramming and gene editing techniques allow for generating cell lines with specific genotypes as patients, thereby further providing cardiac cells with targeted gene alterations. Meanwhile, non-genetic factors contributing to cardiac anomalies can include ischemia, metabolite induction, nutrient shortages, drugs, infections, and inflammation (Figure 1). hEHC-based heart disease models can integrate non-genetic factors through various culture conditions. A combined approach of genetic and non-genetic factors enables the development of multifactorial heart disease models, simulating significant secondary hit theories. In cases of drug withdrawal due to toxic effects exacerbating the arrhythmia phenotype, early detection of potential toxicity before preclinical trials can improve drug development efficiency and reduce cardiac-related mortality risk. Relying only on morphology, contraction amplitude, and molecular biology assessments may not be sufficient for detecting arrhythmic patterns in hEHC-based heart disease models. Therefore, emphasis should be placed on quantitatively correlating and forecasting electrical conduction characteristics of hEHC and related arrhythmia risk (Figure 2).

Figure 2. Accelerated preclinical drug development through multidimensional efficacy evaluation of compounds for arrhythmia risk reduction in hEHC.

Overall, the advancement of in vitro human cardiac models should focus on functional attributes dominated by various cardiac cells, especially advanced adult myocardium-level functions, rather than solely on CM fate determinations. Quantitatively understanding and forecasting electrical conduction properties alongside the arrhythmia risk in hEHC can be a significant focal point in the field of cardiac modeling.

The drug development process can be categorized into three primary periods: drug candidate discovery, preclinical drug development utilizing diverse models, and clinical trials involving varied patient demographics. Physiological or pathological models derived from hEHC can mitigate the limitations of cell models in functional characteristics observed at the myocardial level, while also tackling the species-specific differences between animals and humans. This hEHC approach facilitates the efficient screening of compound libraries and the identification of potential cardiotoxicity.

Author Contributions: S.W., P.Y., J.N.S. and D.Z. wrote the manuscript and organized the figure. All authors have read and agreed the published version of the manuscript.

Funding: D.Z. was funded by the National Key R&D Program of China (2021YFA1101902) and the National Natural Science Foundation of China (32171107).

Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.

Data Availability Statement: Not applicable.

Acknowledgments: Some of the drawings in Figure 1 and 2 were created in https://BioRender.com.

Conflicts of Interest: The authors declare that they have no known competing financial interests.