In Vitro Cardiac Tissue Models: Current Status and Future Prospects

Abstract

Cardiovascular disease is the leading cause of death worldwide. Achieving the next phase of potential treatment strategies and better prognostic tools will require a concerted effort from interdisciplinary fields. Biomaterials-based cardiac tissue models are revolutionizing the area of preclinical research and translational applications. The goal of in vitro cardiac tissue modeling is to create physiological functional models of the human myocardium, which is a difficult task due to the complex structure and function of the human heart. This review describes the advances made in area of in vitro cardiac models using biomaterials and bioinspired platforms. The field has progressed extensively in the past decade, and we envision its applications in the areas of drug screening, disease modeling, and precision medicine.

Graphical abstract

1. Introduction

Drug discovery and development is a challenging road, and current methods to evaluate drug safety and efficacy are costly and inefficient. The average time between drug discovery and commercialization is 10 - 15 years, with median costs over $5 billion [1]. During preclinical and clinical development, cardiotoxicity remains a major cause of failure, with high rates of post-approval withdrawal of medicines [2]. Furthermore, effective pre-clinical evaluation of drugs is essential for treating cardiovascular diseases affecting 17.5 million people worldwide and accounting for 31% of all global deaths in 2012 [3]. However, major barriers inhibit current research in human drug screening: experimental in vivo interventions have unacceptably high risks for humans enrolled in clinical trials, and non-human animal models fail to fully recapitulate human physiology. For example, the resting heart rate in mice is tenfold higher than in humans, while the mouse QT interval is one-fourth of a typical human [4]. Due to inter-species differences in ion channels, biological pathways, and pharmacokinetic properties, animal models do not faithfully predict human cardiotoxicity. Thus, human in vitro models of cardiac tissue that are predictive of human drug response would be a significant advancement for understanding, studying, and developing new drugs and strategies for treating cardiac diseases.

An ideal in vitro cardiac model should accurately recapitulate the physiological or pathological conditions of the human heart, including three-dimensional (3D) anisotropic tissue structure, orientation of the extracellular matrix (ECM) network, vascularization, and circulation (Figure 1). Traditional 2D in vitro systems, although informative [5, 6], cannot accurately mimic the complex 3D conditions due to their inability to recapitulate the dynamics of the biological and mechanical properties of the in vivo microenvironment [7]. The 3D models are characterized by establishment of adhesion complexes and tissue polarity, and by changes in cytoskeletal structure and cell volume that are significantly different from those found in cells cultured as monolayers. As a result, the translational results in 2D conditions are fundamentally different from those in 3D [8].

Figure 1.

Overview of in vitro cardiac tissue model. New in vitro biomaterial-based cardiac tissue models have the potential to be used for fundamental research and translational applications. In particular, the areas of drug discovery, disease modeling, and precision medicine could benefit immensely from these emerging technologies.

Human cardiovascular conditions in vitro can be achieved by developing engineered physiologically relevant 3D models, for instance by embedding cells in biomaterial matrices or microfabricated devices. For the purpose of in vitro modeling, biomaterials and microsystems not only serve as scaffolds for tissue formation, but also provide a highly-controllable microenvironment that incorporates key niche elements to enable precise regulation of cell fate and function [9-11]. Specifically, the complex tissue and organ architecture of the heart is maintained by extensive 3D ECM networks, including fibrous proteins (e.g. collagen, elastin), adhesive glycoproteins (e.g. laminin, fibronectin) and proteoglycans [12]. This ECM network, primarily in the form of perimysial collagen fibers, guides the anisotropic alignment of cardiomyocytes (CMs), mechanically confines the cells to connect each other, and contributes to stress-strain relationships for the heart [13]. Perimysial collagen fibers are comprised of bundles of twisted constituent fibrils (∼ 40 - 50 nm in diameter), forming fibers that range from ∼ 0.5 - 10 μm in diameter and ∼100-200 μm in spacing, allowing several CMs to fit in-between[14]. Furthermore, the perimysial collagen fibers are arranged parallel with the long axis of cardiac muscle and therefore are one of the most significant components of the myocardium that contributes to its non-linear passive stiffness in the direction of the cardiac muscle fibers [15]. The perimysial fibers interact with the CMs via various mechanotransduction pathways, and ultimately affect normal cardiac function. For example, the fibrillar collagen networks register sarcomere Z-line across the CM membrane, and thereby ensure equal stretching of contiguous cells and maintenance of the mechanical continuity between CMs [16]. Given the key role of ECM in heart development and mechanical functions, development of an in vitro cardiac model requires biomaterials, methods, and systems to host the cells, control the cell-cell and cell-ECM interactions, and regulate the cell fate and functions.

In this review, we focus on the important role of biomaterials and microsystems used for in vitro cardiac models. First, we briefly discuss the cell source used for cardiac tissue models, and emphasize human induced pluripotent stem cells (hiPSCs) as the most promising cell type for generation of human CMs. Then, we highlight key properties of different in vitro models, along with their advantages and limitations for applications such as drug cardiotoxicity screening and human heart disease modeling.

2. Cell Sources for Cardiac Tissue Models

In the adult human heart, CMs account for roughly 75% of the heart volume, although they represent only about 33% of the total cell number [17, 18]. Therefore, identifying the optimal source of beating CMs is the first step in the development of a functioning in vitro cardiac model. Early cardiac tissue models depended on either immortalized human cell lines or primary cells isolated from multiple species. The immortalized human ventricular AC cell line was developed using fusion of primary ventricular CMs with a SC-40 transformed fibroblast cell line [19]. Primary CMs isolated from embryonic chicken and neonatal mice and rats were the next most common cell sources for cardiac models [20-22], but increased awareness that animal cell-based models cannot truly recapitulate human physiology has led to the development of more sophisticated cells to build human-like tissue models.

The advancement of stem cell biology has spearheaded the development of in vitro cardiac models that employ differentiated pluripotent stem and progenitor cells [23]. Originally, mesenchymal stem cells (MSCs) were widely used for cardiac tissue models to investigate their beneficial effects on damaged cardiac tissues, either through transdifferentiation or paracrine signaling [24-26]. However, MSCs suboptimal capability for cardiac differentiation has limited the use of these cells in cardiac tissue models.

For better recapitulation of human physiology and pathology, in vitro cardiac models now focus on human pluripotent stem cells, including human embryonic stem cells (hESCs) and hiPSCs [27, 28] (Table 1). Contracting CMs were first generated from hiPSCs through co-culture with END2 mouse endoderm-like cells, a methodology restricted by its reliance on animal cells [29]. hiPSCs suspended in fetal bovine serum to make 3D aggregates of embryoid bodies was later used to generate CMs. Initial protocols produced contracting embryoid bodies with only 5%-15% efficiency [30], and subsequent optimization with timely addition of growth factors (such as Activin A, BMP4 and FGF) improved this efficiency to over 70% [31]. Nowadays, monolayer differentiation, which involves simple, serum-free, and scalable protocols, has largely replaced embryoid body formation [32, 33]. Meanwhile, Activin A and BMP4 have been replaced by small molecules CHIR99021 and IWP4, which leads to greater reliability and higher efficiency [34]. Recently, chemically defined method to replace Matrigel-coating with synthesized vitronectin peptide, and “B27” with L-ascorbic acid 2-phosphate and recombinant human albumin has been used to generate CMs at 85% purity, that can be enriched to 95% with sodium lactate [35]. Based on these advances, City of Hope scientists funded by California Institute of Regenerative Medicine (CIRM) are currently developing a bag-based bioreactor system for scalable and controllable production of Good Manufacturing Practices (GMP)-level hESC-CMs, which will remove a key barrier to developing regenerative medicine products, especially for cardiac repair requiring for high doses of human CMs [36].

Table 1.

Generation of cardiomyocytes from human pluripotent stem cells.

Methods	Media	Yield	Disadvantage
Feeder Layer	Serum-based media Mouse END-2 cells [29]	35%	Low yield Serum media Requirement of mouse feeder cells
Embryoid Bodies	Serum-based media [30]	5-15%	Low yield Serum media
	RPMI+B27 supplement ActivinA + BMP4 [31]	60%	Medium yield Requirement of EB formation Batch variability of growth factors Chemical undefined “B27”
	Bioreactor suspension culture [36] RPMI+B27 supplement Small molecules	90%	Chemical undefined “B27”
Monolayer	RPMI+B27 supplement ActivinA + BMP4 [33]	35%	Low yield Batch variability of growth factors Chemical undefined “B27”
	RPMI+B27 supplement Matrigel Sandwich ActivinA + BMP4 [32]	90%	Batch variability of Matrigel and growth factors Chemical undefined “B27”
	RPMI+B27 supplement Small molecules [34]	90%	Chemical undefined “B27”
	RPMI + human albumin L-ascorbic acid 2-phosphate (AA 2-P) Small molecules [35]	85%

Exciting advances in genome-editing methods by endonuclease (ZFN or TALEN) or palindromic repeat (CRISPR) are being introduced to engineer cardiac disease-associated gene mutations into hiPSC lines with the same genetic background, which will be instrumental for generating libraries of disease-specific CMs for drug testing and disease modeling [37, 38]. To work effectively in the area of patient-specific cells and disease models, a high degree of collaboration and coordination amongst academic laboratories and industry is required. To this end various institutions like CIRM, Cellular Dynamics International (CDI), Coriell Institute for Medical Research, Axiogenesis, and Stanford University are working cohesively to establish a bank of hiPSCs, which will ensure the development of standard operating procedures and practices in order to achieve efficiency, consistency, and high throughput [39, 40]. Making this hiPSC bank available to a broader base of researchers would strongly support a more thorough understanding of the nature of cardiovascular diseases, and the development of cures and stem cell therapies for said diseases.

One key area of research that needs to be addressed prior to full-scale use of iPSCs for cardiac drug screening and development is the maturity of the CMs. During heart development, cardiac muscle cells undergo a complex series of structural changes that ultimately result in their adult phenotype [41]. CM maturation in vivo is also regulated by diverse factors, including topographical, electrical, mechanical, biochemical, and cellular interaction cues. However, hiPSC-CMs in vitro retain a relatively immature phenotype and exhibit relatively small size, reduced electrical excitability, impaired excitation-contraction coupling, and incomplete adrenergic sensitivity [42]. This is one of the critical obstacles to the successful development of predictive drug and toxicology screens, as well as safe and efficient cardiac therapies. Currently, efforts focus on dissecting the external cues (e.g., chemical, physical, electrical), deciphering signaling pathways, and harnessing this information to accelerate the maturation process [43]. Hereby, engineering methods will play a crucial role to stimulate the in vitro processing of hiPSC-CMs maturation by providing relevant environmental motifs, such as anisotropic morphology, external electrical stimulation, mechanical loading, and extracellular matrices.

3. Cell Micropatterning for 2D CM Alignment

An optimal in vitro model would incorporate the aforementioned hiPSCs into an in vivo-like tissue structure while providing researchers with precise control over cell types, ECM composition, cell-cell interactions, and microenvironment geometry. In early studies on the effect of CM anisotropic morphology, cardiac cells were aligned on a thin collagen surface coating that was spread using a cell scraper and polymerized while slowly being poured within a slightly tilted dish [44]. Later, microabrasion was employed to create aligned CMs with anisotropic sarcomeric structure, by unidirectional abrading polyvinyl chloride (PVC) coverslips using lapping papers with different grit sizes [45].

More recently, microfabrication-based patterning techniques (Figure 2) have been used to establish in vitro culture models and investigate the fundamental physiological and pathological characteristics of CMs. Cell alignment can be controlled by surface topography [46-50] or by micromolding with microchannels fabricated from PDMS [51-53]. Microcontact printing ECM proteins created cell-adhesive areas of various shapes on cell-repelling surfaces, using, for instance, laminin onto polyacrylamide thin films [54, 55], fibronectin onto alginate [56], laminin onto PDMS [57], or repelling areas on adhesive surfaces using chitosan and hyaluronic acid onto PDMS and glass [58, 59]. Significant observations in calcium handling, action potentials, and conductional velocities were more similar to adult mouse myocardium in aligned CMs as compared to those grown in randomly oriented cultures [60, 61]. Monolayer of aligned neonatal rat CMs created by microcontact-printing method was found to undergo fibrosis after activation of TGF-β signaling pathway and reduce electrical conduction due to the mechanical interactions between myofibroblasts and CMs [62].

Figure 2.

Micropatterned 2D cardiac models. Topographical alignment of CMs with (A) microfabricated nanostructured surface [50]; (B) prestressed thermoplastic shrink film with tunable multi-scaled wrinkles [49]; and (C) microcontact-printed patterns of pattern CMs into (D) aligned stripes to mimic adult cardiac tissue structure [55] and (E) circular colonies for high-throughput screening [64]. (F) Using oxygen plasma to etch PEG surfaces under a PDMS stencil protection allows micropatterning hiPSCs and determining stem cell fate during cardiac differentiation [72].

Micropatterning hiPSC-derived CMs (hiPSC-CMs) by microcontact printing collagen onto polyacrylamide has been used to increase the maturity level of hiPSC-CMs with optimized culture media [63]. A similar microcontact printing approach with laminin was used to generate hESC-CM microarrays for functional analysis and drug screening, assessing the effects of treatment with H₂O₂ on CM viability and contractility [64].

Micropatterning techniques also enable precise control over the shape and size of cell colonies and are regularly used to generate uniform embryoid bodies (EBs) for studies of embryogenesis and cardiomyogenesis. Contained in poly(ethylene glycol) (PEG) hydrogel microwells, mouse ESCs formed homogeneous EBs of different sizes. The size of the EBs modulated differential expression of WNT5a and WNT11, leading to higher CM differentiation in large EBs, compared to higher endothelial differentiation in small EBs [65]. The size of 3D polyurethane microwells was also found to modulate cell-cell contact and canonical Wnt/β-catenin signaling in human ESCs, resulting in higher CM differentiation in larger wells [66, 67]. A more extensive study of microwells in silicone rubber sheets fabricated via laser cutting revealed that cell patterning resulted in homogeneous expression of pluripotent markers in hiPSCs and improved yield and reproducibility of cardiac differentiation [68]. Studies on the effects of patterning sizes on embryogenesis and cardiogenesis were also conducted by microcontact printing Matrigel to generate uniform EBs. Patterned EBs revealed that the ratio of Gata6 (endoderm-associated marker) to Pax6 (neural-associated marker) expression increased with decreasing colony size. Larger EBs with endoderm-biased (high Gata6/Pax6) gene expression at early stages exhibited higher mesoderm and cardiac induction[69]. This approach was further used for high-throughput analysis of cell fate determination and endogenous signaling pathway activation and differentiation bias [70].

Recently, researchers found that the geometric confinement from the micropatterned substrate was able to trigger self-organization of hESCs, which recapitulated spatial cell fate patterning during early embryonic development. In response to BMP4, colonies reproducibly differentiated to an outer trophectoderm-like ring, an inner ectodermal circle, and a ring of mesendoderm expressing primitive-streak markers in between [71]. Synergism of biochemical cues and geometric confinement on micropatterned hiPSCs can induce self-organizing lineage specification and creation of a 3D beating human cardiac microchamber, which resembles the developing primitive human heart. These in vitro cardiac microchambers were used to screen drugs likely to generate cardiac malformations during development. For example, applying thalidomide during the cardiac differentiation not only reduced differentiation efficiency, but also significantly damaged the formation of cardiac microchambers with smaller size, lower contractility, and decreased beat rates compared to the control [72].

Although micropatterning methods can confine colony geometry, regulate cell morphology and functions, and support high-throughput analysis, these 2D culture platforms lack the full architecture and functional properties of 3D human tissues and organs, and thus are of limited use for cardiac research. These 2D results have been seen as the first step towards engineering cardiac models, which can be used as templates for 3D tissue structure. Ongoing 2D research would focus on single-cell micropatterning and analysis, which can provide insight on cellular machinery, characterize the heterogeneity of cell population, and enable high-throughput screening for single-cell response to different environmental factors. Compared to CM alignment for mimicking heart muscle tissue, single CM micropatterning is extensively involved in exploring myofibrillogenesis and its relationship with extracellular factors. By microcontact printing ECM protein on the coverslip to shape single CM into the predesigned patterns, researchers found that not only cell shape was defined but also the cytoskeleton was under reorganization into the predicted architecture [73]. It was noticed that the spatial configuration of ECM played a key role in regulating the other three factors: cell shape [20], sarcomere orientation [74], and nuclear morphology [75].

4. Biomaterials Used to Generate 3D Cardiac Models

Engineering a 3D cardiac tissue with physiologically relevant microenvironment and cell morphology presents a significant challenge for in vitro cardiac modeling. Biomaterials have played a major role in creating 3D tissue models, since they not only support cell attachment and alignment, but also transmit load, provide physiologically relevant stiffness, and ideally can be degraded and replaced over time by cell-secreted ECM proteins. Several representative natural and synthetic biomaterials-based engineered heart tissue (EHT) systems are shown in Figure 3. We have classified theses as either hydrogel based or fibrous cardiac models, each is discussed in greater detail below.

Figure 3.

Biomaterial-based 3D cardiac models. (A) Engineered heart mini-tissues (millimeter scale) were made from fibrin and hiPSCs for implantation [78] and drug-screening purpose [79]. (B) Engineered heart micro-tissue (micron scale) made from collagen were used to model the dilated cardiomyopathy caused by titin mutation [82]. (C) Fibrin-based cardiac tissue patch was generated by soft lithography with controllable size and architecture and its drug response to isoproterenol [85]. (D) A biowires platform combining architectural and electrical cues generated a microenvironment conducive to the maturation of hiPSC-derived cardiac tissues [95]. (E) Electrospun nanofiber scaffolds were made for creating the continuous anisotropic cardiac tissue [89]. (F) Aligned nanofiber scaffolds made by rotary jet spinning promoted better sarcomere formation in CMs [91]. (G) High-defined filamentous scaffolds made by two-photon initiated polymerization were used to create an aligned hiPSC-CMs-based cardiac model for drug screening [92].

4.1 Natural hydrogel-based cardiac models

Hydrogels consisting of two naturally occurring proteins, collagen and fibrin, have been widely used to generate EHT. Matrigel was often used as a supplemental material to increase cell viability and attachment due to its various growth factors and matrix components. The first EHT consisted of a 3D scaffold of collagen I with embryonic chick CMs [76]. Later, they were further developed into ring structures with neonatal rat CMs [77], which could be stacked and implanted for successful improvement of the function of infarcted rat heart [78]. Currently, EHTs are primarily designed in a two-post configuration, allowing for characterization of contraction forces. Parallel EHT arrays consisting of a mixture of fibrinogen, Matrigel, thrombin, and neonatal rat CMs [79] or hESC-CMs [80], on a silicone post rack casted from Teflon molds, were used for preliminary drug screening. Proarrhythmic compounds chromanol and erythromycin was shown to affect EHT repolarization inhibition, and the cardiotoxic drug doxorubicin affected EHT force generation in a time- and dose-dependent manner [79]. Isoprenaline and carbachol were found to affect the spontaneous contractile rate. Repolarization was inhibited by E-4031 (3 nM IC₅₀), procainamide (100 μM IC₅₀), sertindole (10 nM IC₅₀), quinidine (1000 nM IC₅₀), and cisapride (30 nM IC₅₀). [80].

EHTs of collagen I and fibrinogen were also generated on microfabricated devices with micron-scale standing posts, and researchers found that the matrix composition affected the dynamic and static contractility of the cardiac tissues [81]. Using these micro-EHTs, researchers were able to model dilated cardiomyopathy caused by titin mutation, and demonstrated that 3D titin-mutant EHTs exhibited lower contraction forces compared to WT EHTs, such difference in contractile function was not possible to be detected by single-cell assays [82]. This configuration was further employed to generate EHTs based on a mixture of collagen I, Matrigel, and hESC-CMs for preclinical drug screening and gene transfer. The 610 nM IC₅₀ value generated for verapamil in these EHTs surpassed the 160 nM IC₅₀ for traditional iPSC cells in 2D culture, indicating better recapitulation of human physiology compared to a 2D culture system. However, an insufficient response to isoproterenol suggested cardiac tissue immaturity [83]. A multi-post configuration with collagen I and Matrigel was used to design and formulate cardiac microtissues using hiPSC-CMs, and researchers found that tissue structure and non-CM population played important roles in tissue integrity and maturation [84]. This multi-post platform was further applied to establish a tachycardiac model of arrhythmogenesis for in vitro patient-specific disease modeling.

The 3D cardiac tissue structure was also created using fibrin-based hydrogel matrix generated by soft lithography technique with controllable size and architecture; these EHTs demonstrated increased spontaneous beat rate and twitch amplitude upon exposure to isoproterenol, with an EC₅₀ of 95 nM falling within the reported 30-160 nM range for adult human ventricular tissue. CMs differentiated from ESCs and from cardiac progenitor cells (CPCs) were seeded into this engineered hydrogel to yield highly aligned CMs and robust intercellular coupling with rapid action potential conduction (22–25 cm/s) and significant contractile forces (up to 2 mN) [85, 86].

In all post-based in vitro cardiac microtissues, various natural biomaterials (e.g. collagen I, fibrin, Matrigel) served as an initial scaffold and ECM to support cell attachment, whereas the posts stabilized the developing tissue as the cells condensed and remodeled the scaffold, which had the effect of aligning in 3D the encapsulated CMs. The flexible PDMS posts additionally served as the sensor enabling the measurement of contraction force generated by the beating CMs. These contractile forces are a key output of EHTs and are coupled with CM electrophysiology and hypertrophy within EHTs; however, these forces are highly dependent on the biomaterial composition, making it difficult to compare drug responses among different EHTs developed by different research groups. A high degree of natural material variability is of major concern in efforts to establish standardized assays for drug screening with the requirements of consistency, reproducibility, and high-throughput capability. Such material variability will affect tissue formation and cellular responses, which will eventually lead to the variation of functional readout, such as contractile force measured by the posts.

4.2 Synthetic fibrous cardiac models

Synthetic biomaterials provide an attractive alternative to natural materials, as researchers can control the entire synthesis process as well as the materials' mechanical properties, topography, and structure. A number of synthetic polymers have also been used to create 3D cardiac scaffolds for either in vitro models or implantable patches to repair and regenerate the infarcted tissue. Key requirements for synthetic scaffolds are that they recapitulate the native 3D hierarchical fibrillar structure, possess biomimetic surface properties, and demonstrate mechanical integrity. The most frequently used synthetic polymers for cardiac tissue engineering are polyurethane, poly e-caprolactone (PCL), polylactic acid (PLA), polyglycolic acid (PGA), and their copolymers. One example of synthetic material-based cardiac constructs were generated with neonatal rat CMs and poly(glycerol sebacate) (PGS) and maintained in a bioreactor with simultaneous culture medium perfusion and electrical conditioning, which led to enhanced organization and functionality of engineered cardiac tissue [87].

Cell alignment can be obtained with electrospun nanofiber-based scaffolds, which provide flexible matrices and topographic properties offering support and guidance for the CMs. CMs organized into anisotropic cardiac tissue on aligned PCL/gelatin composite electrospun nanofibrous scaffolds to structurally mimic the oriented ECM in myocardium [88]. The orientation and density of electrospun polymethylglutarimide (PMGI) nanofibers defined the overall architecture of the cardiac tissue, which was optimized for best alignment with 30-50 fibers/mm and an average distance between fibers of under 30 μm [89]. An aligned fibrous mesh of electrospun polyester blend, poly(3–hydroxybutyrate-co-3-hydroxyvalerate) (PHBV), P(L-D,L)LA, and poly(glycerol sebacate) (PGS) was shown to enhance cardiomyogenic differentiation of human umbilical cord mesenchymal stem cells [90]. Similarly, rotary jet spinning was used to fabricate highly aligned nanofiber constructs from a blend of collagen, gelatin, and PCL polymer, which promotes better sarcomere formation in CMs [91].

Electrospun 3D scaffolds with aligned nanofibers using synthetic polymers successfully mimic the structure and orientation of native ECM in the myocardium and help CMs self-organize with anisotropic structure. However, the micron scale porosity of these scaffolds limits cell infiltration into the matrix and thereby the creation of a 3D tissue. As such, the scaffolds are 3D in nature, but the tissue is really a 2D structure similar to those created on micropatterned surfaces. To address this limitation, a highly defined scaffold structure was fabricated by two-photon initiated polymerization (TPIP) with unprecedented control over a wide range of matrix features. A human cardiac disease model was created by seeding hiPSC-CMs, with long QT syndrome type 3 (LQT3), on the TPIP-fabricated synthetic filamentous scaffolds. Tailoring the mechanical properties of the scaffolds modulated the contractility of residing hiPSC-CMs and, more importantly, recapitulated the abnormal contractility of long QT syndrome. Treatment with caffeine increased the spontaneous contractile rate and maximum contractile velocity and high doses of caffeine and nifedipine both caused cessation of beating. In contrast, treatment with E-4031 indicated irregular beating patterns, and propranolol induced significant uncoordinated beating, suggestive of cardiac arrhythmias, in a dose-dependent manner [92].

A collagen-based cardiac tissue model, termed ‘biowire’, combined architectural and electrical cues to generate a microenvironment conducive to maturation of hiPSC-derived cardiac tissues [93]. The hiPSC-CMs were seeded with collagen type I into a microfabricated well and subjected to electrical stimulation with a progressive increase in frequency. Biowires submitted to electrical stimulation had markedly increased myofibril ultrastructural organization, elevated conduction velocity, and improved both electrophysiological and Ca²⁺ handling properties compared to non-stimulated controls. These changes suggested enhanced CM maturation that depended on the stimulation rate. The biowire maturation represented an intermediate phenotype as CMs undergo maturation from the embryonic state, as evidenced by low membrane conductance. The use of electrical stimulation in conjunction with stretch as a mimic of cardiac load, concurrently or sequentially, might be required to induce terminal differentiation and maturation in hiPSC-CMs [94, 95].

These findings collectively suggest that 3D tissue engineered models with defined cellular microenvironments hold great promise for high-content drug screening and cardiotoxicity testing. The integration of biomaterials with existing iPSC-based disease models could better recapitulate disease pathology and may represent superior scalability and flexibility for creating large numbers of personalized models to meet diverse and urgent patient needs.

5. Microdevices for 3D Cardiac Models

Moving away from scaffold-based cardiac models, highly miniaturized and integrated microphysiological systems are currently being developed as “heart-on-a-chip” technology to provide more controlled 3D microenvironments, with enhanced multiple functionalities and increased throughput. Such microphysiological systems (Figure 4) combined with hiPSC technology are expected to not only better predict on toxicity and efficacy of potential drugs in human physiologically relevant conditions, but also provide a more in-depth understanding of human cardiac disease in complex and heterogeneous microenvironments.

Figure 4.

Microdevice-based 3D cardiac models. (A) PIPAAm-based ‘heart-on-chip’ microsystem [97] can measure the deformation of the elastomeric thin film to characterize the contractility of cardiac tissue derived from various cell types and assess the drug response to isoproterenol [98]. (B) A stacked-paper culture system containing CMs was used to mimic the pathological microenvironment occurring during cardiac ischemia [101]. (C) A microfluidic-based microphysiological system was designed to recapitulate a minimal organoid of the human myocardium with highly aligned tissue architecture and anisotropic beating behavior, allowing for accurate prediction of drug cardiotoxicity [102].

An engineered anisotropic ventricular myocardium was first developed by micropatterning neonatal rat CMs on poly(N-isopropylacrylamide) (PIPAAm)/PDMS-based thin elastomeric film, which can simultaneously measure the contractile function, quantify the electrical propagation, and evaluate cytoskeletal architecture in cardiac tissues during pharmacological interventions. A dose-dependent increase in spontaneous beat rate and stress was reported in response to epinephrine. [96]. This microsystem was further incorporated with fluidic control for drug washout, a heating element for temperature control, and embedded electrodes for electrical field stimulation [97]. This system was not only used to characterize the cardiac tissue derived from various cell types (primary neonatal mouse CMs, mouse iPSC-CMs, and human iPSC-CMs) [98], but also to model maladaptive cardiac hypertrophy [99] and patient-specific mitochondrial cardiomyopathy, specifically, the Barth Syndrome (BTHS) - a mitochondrial disorder caused by mutation of the gene encoding tafazzin (TAZ) [100]. To study the pathophysiology underlying BTHS the group generated hiPSC-CMs from two patients with BTHS and discovered metabolic, structural and functional abnormalities associated with TAZ mutation. This elegant study provided new insights into the pathogenesis of Barth syndrome, and pointed to a new treatment strategy for BTHS.

For improved modeling specific types of cardiac disease, unique platforms should be designed to mimic the pathological microenvironment occurring during the disease progression. A paper-based culture system was developed with multiple layers of paper-containing cells, suspended in hydrogels, stacked to form a layered 3D model of a cardiac tissue. Mass transport of oxygen and glucose into this 3D system was modulated to induce an ischemic environment in the bottom layers of the stack. This in vitro cardiac model mechanistically studied cellular motility and viability, and recapitulated the cellular interactions and gradients of molecules in the heart under ischemia. However, the cardiospheres in the stacked papers lacked the aligned structure to mimic the in vivo tissue structure. Moreover, this system currently makes it difficult to determine the concentration of small molecules (e.g., oxygen, glucose, or cytokines) in situ and to measure the contractility of CMs without complex optics [101].

To allow accurate prediction of drug cardiotoxicity, a microfluidic-based microphysiological system was designed to recapitulate a minimal organoid of the human myocardium [102]. Pharmacological studies on this system with verapamil (950 nM IC₅₀), isoproterenol (315 nM EC₅₀), metoprolol (244 μM IC₅₀), and E-4031 (392 nM IC₅₀) predicted a higher safety margin and had better concordance with tissue-scale values and clinical observations, compared to those in cellular-scale studies and large-scale animals. The human cardiac microphysiological system was proposed to complement animal models, and in the future may have the potential to replace animal studies, which often are expensive, unethical, and unable to accurately predict the drug's actual effect.

Since the discovery of Moore's law semiconductor industry has come a long way and the development of new microfabrication techniques has equipped the bioengineering community with tools, which can be employed for basic and translational applications. Microengineered in vitro models with multiple readouts have a great potential to better mimic the in vivo physiology and provide a deeper understanding of the physiological events that characterize cardiac development and function. These systems provide fine control over fluid flow creating microcirculation mimicking the in vivo transport; massive parallelization for high content readouts; miniaturization of large systems for convenient operation and reduction of reagent use leading to lower operational costs; and unprecedented control of system architecture and dimensions at the biological scale (nm to μm). We envision the use of microtechnologies coupled with hiPSC biology to revolutionize the areas of drug screening, disease modeling, and personalized medicine.

6. Perspective and Conclusions

The heart is a powerful, complex organ that has intrigued both artists and scientists for millennia. In vitro cardiac tissue models present great opportunities for regenerative medicine, drug screening, and disease modeling. The opportunities, however, coincide with enormous challenges due to the complexity of cardiac structure and function. A standardized, reproducible, and scalable process for differentiating hiPSCs to CMs is required for consistent cell quality. Recent developments in the cardiogenic differentiation open the possibility of obtaining such human CMs in the laboratory [34, 35].

The immaturity of hiPSC-CMs complicates the cells' adoption as a reliable readout for translational applications. Such immature embryonic or neonatal-like CMs cannot compare morphologically with large and stiff ventricular CMs in the adult human heart [103]. Thus, cardiac tissues constructed from hiPSC-CMs have significantly lower field potentials and contraction forces than adult ventricular tissue, so at this point cannot be considered an exact in vitro model of mature myocardium. It has been suggested that tissue-engineering methods would necessitate the maturation of hiPSC-CMs in a physiologically mimicked microenvironment [63, 86, 93]. This suggests that genetic and environmental factors interact and lead to CM maturation, though the mechanism and process is not fully understood.

We summarize the current in vitro cardiac tissue models, along with their advantages and limitations for applications, such as drug cardiotoxicity screening and human heart disease modeling (Table 2). An ideal in vitro cardiac tissue model should be physiologically relevant with multiple biological, mechanical, and electrical readouts, ensuring different functional endpoints for a particular application. Appropriate biomaterials used for the cardiac tissue models need to be chosen carefully according to the specific applications. For example, microsystems with conventional PDMS as a substrate result in drug stability problems and unpredictable device performance, due to its absorption and retention of highly hydrophobic compounds [104, 105]. Acceptance of these models will require automation, robustness, and easy integration into the workflow at pharmaceutical companies. Specifically for drug development and testing, the microfluidic-based system with standardized fabrication and process holds great promise on high-content screening with electrical and mechanical measurement and integration with multiple organs to achieve “human on a chip.”

Table 2.

Analysis of in vitro cardiac tissue models and the corresponding mechanical, electrophysiological, and biological outcomes.

Characteristics					Outcomes
Platform		Cell Type	Materials	Coating	Mechanical	Beat Rate (bpm)	Biological
Micropatterning	Surface topography	nrCM	PVC cover slips	Fibronectin [45]		*	+
		nrCM	PDMS	Laminin [46]		*	++
		nrCM	Polyurethane, polystyrene	[47]		15-50	+
		nmCM	PDMS	Fibronectin, laminin, collagen I [48]		*	++
	Microcontact printing	nrCM	Polyacrylamide	Laminin, Matrigel [54]	Young's modulus 5-35 kPa	*	++
		nrCM	Alginate	Fibronectin [56]	Young's modulus 57 kPa	60-240 (pacing)	++
		nrCM	PDMS, Stretch device	Collagen [57]		*	+
		nrCM	Polystyrene	Chitosan [58]		*	++
		nrCM	PDMS	Hyaluronic acid, fibronectin [59]		60-100	+
		nmCM	Glass, photoresist	[60]		*	++
		hESC-CM, hiPSC-CM	Polyacrylamide	Gelatin [63]	Contractile stress 0.2-0.5 mN/mm2	60-180 (pacing)	++
		hESC-CM	Polyacrylamide	Laminin [64]	Elastic modulus 15-35 kPa	50	++
Microfabrication	Microposts	hESC-CM, nrCM	Silicone post racks	Fibrin, Matrigel [80]	Contractile force 100-300 μN	300	++
		nrCM hiPSC-CM	PDMS	Collagen, fibrinogen [81, 82]	Contractile force 2-6 μN	33-60	++
		hESC-CM	PDMS posts	Collagen, Matrigel [83]	Contractile force 0.3 mN	70	++
		hESC-CM, nrCM	PDMS posts	Collagen, Matrigel [84]		*	++
		hESC-CM	PDMS posts	Fibrin [85]	Contractile force 2 mN	*	++
		hESC-CM	PDMS posts, patch	Fibrin, Matrigel [86]	Contractile force 3 mN	60-180 (pacing)	++
	Perfusion Bioreactor	nrCM	Poly(glycerol sebacate), channels	Laminin	Elastic modulus 34.55±1.26 kPa, Pore size 75-150 μm	180 (pacing)	++
	Biowires	hESC-CM	PDMS	Collagen [93, 94]	Conduction velocity 11-16 cm/s, Young's modulus 1-6 kPa	60-360 (pacing)	++
	Two-photon polymerization	hiPSC-CM, Long QT3 syndrome	Filamentous matrix	Fibronectin [92]	Maximal contraction velocity 15-25 μm/s	90	++
	S syst	nrCM, hESC-CM, hiPSC-CM, Barth syndrome	PDMS, PIPAAm	Fibronectin Gelatin [96, 97, 99,100]	Young's modulus 1.52 MPa, Systolic stress 15-20 kPa, Diastolic stress 8.0 kPa	120 (pacing)	++
		hiPSC-CM	PDMS	Fibronectin [102]	Average contraction velocity 3 μm/s	55 - 80	++

nrCM Neonatal rat CMs

nmCM Neonatal mouse CMs

hESC-CM Human embryonic stem cell-derived CMs

hiPSC-CM Human induced pluripotent stem cell-derived CMs

Beat rate Beat per minute (BPM);

^*Spontaneous contractions reported without beat rate

⁺Indications of CMs are limited to: cell alignment and elongation, morphological assessment, genetic assessment

⁺⁺Indications of CMs include: sarcomeres, functional gap junctions, appropriate responses to drug treatments, as well as indications from ‘+’

A more futuristic application is envisioned in the area of precision medicine, an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person [106, 107]. The promise of precision medicine for cancer therapeutics is already being realized with the recent introduction of several targeted therapies, some with companion diagnostic tests that identify patients most likely to benefit from treatment [108]. Moving forward, we hope to see physiologically functional in vitro cardiac models of individual- and disease-specific hiPSCs on chips, which can be termed as “patient on a chip”. This approach will help to diagnose and design better treatment strategy for individual patients. Success, however, will depend on how effectively and how efficiently engineering and biology can be integrated to create such systems.