Engineering cardiac microphysiological systems to model pathological extracellular matrix remodeling

Abstract

Many cardiovascular diseases are associated with pathological remodeling of the extracellular matrix (ECM) in the myocardium. ECM remodeling is a complex, multifactorial process that often contributes to declines in myocardial function and progression toward heart failure. However, the direct effects of the many forms of ECM remodeling on myocardial cell and tissue function remain elusive, in part because conventional model systems used to investigate these relationships lack robust experimental control over the ECM. To address these shortcomings, microphysiological systems are now being developed and implemented to establish direct relationships between distinct features in the ECM and myocardial function with unprecedented control and resolution in vitro. In this review, we will first highlight the most prominent characteristics of ECM remodeling in cardiovascular disease and describe how these features can be mimicked with synthetic and natural biomaterials that offer independent control over multiple ECM-related parameters, such as rigidity and composition. We will then detail innovative microfabrication techniques that enable precise regulation of cellular architecture in two and three dimensions. We will also describe new approaches for quantifying multiple aspects of myocardial function in vitro, such as contractility, action potential propagation, and metabolism. Together, these collective technologies implemented as cardiac microphysiological systems will continue to uncover important relationships between pathological ECM remodeling and myocardial cell and tissue function, leading to new fundamental insights into cardiovascular disease, improved human disease models, and novel therapeutic approaches.

INTRODUCTION

The healthy adult myocardium consists of layers of rhythmically contracting, synchronized cardiac myocytes that are elongated and aligned into fiber-like structures to maximize force generation parallel to the long axis of the tissue (Fig. 1A) (186). Cardiac myocytes are embedded in a compliant extracellular matrix (ECM) network that provides alignment cues, biochemical signals, and mechanical support and resistance to myocytes. Myocardial ECM contains mostly collagen fibers aligned in parallel to cardiac myocytes to provide structural stability and tensile strength. Cardiac myocytes anchor to collagen fibers through a basement membrane composed largely of the glycoproteins fibronectin and laminin (151, 186). The ECM network is synthesized and remodeled by cardiac fibroblasts, which are interspersed among cardiac myocytes and are mostly quiescent in the healthy myocardium (29).

Fig. 1.

Engineering cardiac microphysiological systems to define the functional impacts of pathological extracellular matrix remodeling. A: distinct features of native healthy and diseased/fibrotic myocardium, such as myocyte shape, tissue alignment, extracellular matrix (ECM) rigidity, and cell demographics, are used as design templates for engineering cardiac microphysiological systems. B: features of native healthy and diseased/fibrotic myocardium are replicated by combining appropriate biomaterials and microfabrication techniques. C: engineered cardiac cells and tissues are interrogated with functional assays to quantify contractility, electrophysiology, and metabolism as a function of their microenvironment. Collectively, these approaches implemented as cardiac microphysiological systems can identify the functional impact of ECM remodeling to streamline mechanistic studies and therapeutic development. Images in C were adapted from Ref. 119 (top left), Ref. 2 with permission from The Royal Society of Chemistry (bottom left), Ref. 117 (top middle), and Ref. 199 (bottom middle) with permission from Elsevier.

Many cardiac diseases are associated with extensive remodeling of both the cellular and extracellular components of the myocardium (50). For example, after a myocardial infarction, immune cells first remove necrotic cardiac myocytes. Subsequently, fibroblasts and their activated derivatives, myofibroblasts, migrate to the affected area and deposit scar tissue to stabilize the ventricle (181). The infarcted myocardium has more collagen and less fibronectin and laminin than the healthy myocardium (176) and is considerably stiffer (20, 48, 176), changing the biochemical environment and increasing the mechanical load on surviving myocytes. Additionally, the transitional tissue from infarcted to healthy myocardium, known as the infarct border zone, experiences a loss of myocyte alignment and remodeling of myocyte shape due to infiltration of scar tissue and fibroblasts/myofibroblasts (Fig. 1A) (115). These remodeling processes compromise the function of the ventricle and contribute to the progression toward heart failure. Similar patterns of ECM remodeling coupled with tissue dysfunction have been observed in many other pathological conditions, such as tachyarrythmia (171, 198), pathological hypertrophy (43, 73), and aortic valve disease (183), as well as aging (9). Thus, understanding the direct impact of distinct forms of ECM remodeling on the myocardium could lead to more effective treatment options for many forms of cardiac disease and dysfunction.

The effects of ECM remodeling on myocardial function remain elusive, which is in part due to the limitations of conventional model systems. For example, independently dictating the many diverse features of the ECM is extremely challenging, if not impossible, with animal models. Furthermore, complex microenvironmental features, such as tissue alignment and ECM elasticity, are not controllable with conventional cell culture substrates. To bridge the gap between in vivo and conventional in vitro models, new in vitro platforms, known as cardiac microphysiological systems, are now being developed and implemented. These are microfabricated platforms engineered to mimic many of the diverse microenvironmental cues observed in myocardial tissue in vivo while the level of user control and experimental accessibility provided by in vitro models is maintained (Fig. 1B). Additionally, these platforms are developed with integrated functional assays to enable well-designed biological studies, provide deeper mechanistic insights, and facilitate medium-throughput screening (Fig. 1C). In this review, we will describe present approaches for engineering and using cardiac microphysiological systems to interrogate the functional impact of pathological ECM remodeling on the myocardium. These platforms have unique potential for establishing relationships between the ECM and cardiac function with more granularity than previously possible, which will complement existing model systems, greatly enhance our understanding of cardiac disease, and pave the way for new treatment options.

TUNABLE BIOMATERIALS FOR ENGINEERING CARDIAC MICROPHYSIOLOGICAL SYSTEMS

Biomaterials can be used to replicate multiple features of the native ECM in vitro. Biomaterials are usually polymers and can be naturally derived or manmade. In this section, we will detail the characteristics, uses, advantages, and disadvantages of some of the most common natural and synthetic biomaterials used in cardiac microphysiological systems, which are also shown in Table 1.

Table 1.

Biomaterials implemented in cardiac microphysiological systems

	Advantages	Disadvantages	Examples with References
Natural
Mammalian ECM-derived hydrogels	• Tunable elastic modulus	• Difficult to decouple elastic modulus and ECM ligand	• Matrigel (52, 146, 158)
	• Naturally adhesive to cells and amenable to cellular remodeling	• Certain isolated proteins (fibronectin, etc.) are expensive	• Gelatin (116, 125, 156)
	• Compatible with 2-D and 3-D tissue engineering	• Limited range of elastic moduli	• Fibrin (24, 28, 196)
	• Physiological
Nonmammalian ECM-derived hydrogels	• Tunable elastic modulus	• Not physiological	• Chitosan (41)
	• Easier to decouple mechanical properties and ECM ligand	• Limited range of elastic moduli	• Alginate (1)• Silk (140, 175)
	• Some are naturally adhesive to cells and amenable to cellular remodeling
	• Relatively compatible with 2-D and 3-D tissue engineering
	• Relatively inexpensive
ECM from decellularized tissues	• Closely mimic the chemical and mechanical properties of native ECM	• Not tunable• Difficult to acquire• Highly heterogeneous and variable	• Decellularized rat and pig heart slices (25)• Solubilized decellularized ECM (59, 176)
Synthetic
Elastomers	• Easy to decouple elastic modulus and ECM ligand	• Not naturally adhesive to cells or amenable to cellular remodeling	• Polydimethylsiloxane (31, 54, 111, 134, 147)
	• Compatible with 2-D tissue engineering• Relatively inexpensive	• Not compatible with 3-D tissue engineering
	• Extensive range of elastic moduli
Hydrogels	• Easy to decouple elastic modulus and ECM ligand	• Not naturally adhesive to cells or amenable to cellular remodeling	• Polyacrylamide (11, 59, 69, 119, 176)
	• Compatible with 2-D tissue engineering• Relatively inexpensive	• Not compatible with 3D tissue engineering	• Polyethylene glycol (86, 87)

2-D, two-dimensional; 3-D, three-dimensional; ECM, extracellular matrix.

Mammalian ECM-Derived Hydrogels

For decades, cardiac myocytes have been cultured on glass or plastic surfaces coated with ECM proteins present in native myocardium, such as collagen, fibronectin, and laminin. These proteins are usually isolated from other ECM-rich tissues that are relatively easy to acquire, such as porcine skin, rat tails, blood plasma, and the placenta (37). Coating culture surfaces with isolated ECM molecules has been used to delineate the effects of distinct ECM ligands on cell phenotype and define the integrin types expressed and activated in cardiac myocytes in developmental, physiological, and pathological settings. For example, both neonatal and adult rat cardiac myocytes adhere to surfaces coated with basement membrane proteins such as laminin and collagen type IV. However, only neonatal rat cardiac myocytes adhere to interstitial ECM proteins, such as other types of collagen and fibronectin (27, 109), which is suggestive of developmentally regulated integrin expression patterns. Additionally, studies have shown that phenylephrine-induced hypertrophy occurs when cardiac myocytes are cultured on fibronectin or laminin, but not bovine serum albumin, implicating ECM composition and the subsequent activation of specific integrin receptors in hypertrophic responses (152).

Although coating ECM proteins onto glass or plastic surfaces can be used to interrogate the phenotypic effects of ECM composition on cardiac myocytes, this approach offers minimal control over other properties of the ECM, including rigidity. One approach for tuning the rigidity of many ECM molecules is to synthesize them as hydrogels. Hydrogels are hydrophilic, cross-linked polymer networks with significant water content and elastic moduli in the range of soft biological tissues (101). Because of their high water content, porosity, and biocompatibility, hydrogels synthesized from ECM molecules are relatively versatile. For example, they can be used as two-dimensional (2-D) culture substrates (116) or for encapsulating cells to engineer three-dimensional (3-D) tissues (180). Many ECM molecules are already commercialized and widely used as hydrogels for cell culture, such as Matrigel. Matrigel is a cell-derived, laminin-rich mixture of ECM proteins and growth factors that self-polymerize into a hydrogel at 37°C (94). Matrigel is commonly used for culturing stem cell-derived cardiac myocytes as either a monolayer coating on a surface or as a thicker hydrogel (52, 146, 158). However, because Matrigel is a heterogeneous mixture of ECM molecules, it can vary between batches and offers imprecise control of the ECM.

Collagen and its partially denatured derivative, gelatin, can also be cross-linked into hydrogels. Collagen type I is soluble at acidic pH and spontaneously cross-links into a thermostable hydrogel at neutral pH due to intermolecular bond formation between lysine and hydroxylysine residues (126). Gelatin requires cross-linking by chemical or enzymatic means, such as glutaraldehyde (14, 128) or microbial transglutaminase (130, 197), respectively, to form thermostable hydrogels. Likely because of the biomimetic composition and mechanical properties of gelatin hydrogels, neonatal rat cardiac myocytes naturally adhere to and maintain longer-term viability and contractility on these substrates compared with ECM-coated synthetic polymers (116). Gelatin and other ECM proteins, such as tropoelastin (8), can also be modified with methacrylate groups, which are cross-linked by light (125). Because methacrylated gelatin is photo-cross-linkable, users have both spatial and temporal control over hydrogel cross-linking while still preserving biocompatibility and cell adhesion (125, 156). Another protein widely used to fabricate hydrogels is fibrin, which is involved in blood clotting and is synthesized by combining fibrinogen and thrombin to form fibrin fibers (89). Although fibrin is non-native to the myocardium, fibrin hydrogels are still highly biocompatible and can be remodeled by cardiac cells, which have been shown to replace fibrin with their own collagen matrix (24). Fibrin can also be combined with collagen to fabricate hybrid hydrogels with tunable stiffness, which also increases the longevity of 3-D tissues because fibrin fibers are more robust than collagen (28).

Many properties of ECM-derived hydrogels, including stability, rigidity, optical clarity, and porosity, can be tuned by altering the concentration(s) of polymer and/or cross-linker. This feature can be used to controllably mimic both the chemical and mechanical features of healthy and diseased ECM environments in vitro. However, a major disadvantage of these hydrogels is that many of their physical properties are coupled (37). For example, gelatin hydrogels can be stiffened by increasing the concentration of gelatin, but this also increases ligand density and decreases porosity (99). These hydrogels also have a relatively limited range of elastic moduli and thus cannot always recapitulate the range of mechanical environments observed in vivo throughout development and disease.

Decellularized ECM Scaffolds

A more recent approach for defining the physiological and pathological effects of native myocardial ECM is decellularization. Decellularization entails using detergents to remove all cells from a tissue, leaving behind the intact ECM that can be used as a scaffold for tissue engineering. Entire intact hearts have been decellularized and repopulated with cardiac myocytes, with the long-term goal of organ regeneration (131, 188). This approach has also been adapted for in vitro studies by attaching thin slices of decellularized rat and pig hearts to coverslips, seeding the slices with neonatal rat cardiac myocytes, and measuring contractility and electrophysiology (25). The main advantage of decellularized tissue scaffolds is that they maintain the composition, architecture, and mechanical properties of native cardiac ECM, and therefore, they have high physiological relevance. Decellularized ECM from healthy and postinfarct rat ventricular myocardium has also been solubilized and incorporated into mechanically tunable polyacrylamide hydrogels to determine the effects of completely native ECM on mesenchymal stem cell differentiation toward cardiac lineages (59, 176). Despite its advantages related to physiological relevance, the major limitations of using ECM acquired from decellularized hearts are the limited availability of human samples and the relatively high amount of processing required.

Nonmammalian Natural Biomaterials

Macromolecules from nonmammalian organisms have also been isolated and used as biomaterials for in vitro cardiac models. For example, alginate and chitosan are polysaccharides isolated from algae and crustacean shells, respectively, that are widely used tissue engineering scaffolds due to their availability, low cost, and biocompatibility. Alginate and chitosan hydrogels are cross-linked with agents such as calcium chloride and glutaraldehyde, respectively (37, 101). The degree of cross-linking dictates their elastic modulus. However, because these materials are not found in native myocardium, they must be combined or functionalized with proteins or peptides that promote cardiac myocyte adhesion. Although this adds an additional fabrication step, this can be advantageous for gaining independent control over the chemical and mechanical properties of the scaffold. One approach for functionalization is to add ECM proteins directly to the hydrogel prepolymer solution. For example, chitosan has been functionalized with azidobenzoic acid and mixed with collagen to create photocrosslinkable collagen-chitosan hydrogels compatible with cardiac myocyte culture (41). Alternatively, alginate hydrogels have been functionalized by doping the hydrogel with streptavidin, which enables surface binding of biotinylated fibronectin and subsequent attachment of cardiac myocytes (1).

Silk protein has also been used as a cardiac tissue scaffold. Silk fibroin isolated from cocoons of Bombyx mori, one species of silkworm, has been mixed with collagen or ECM isolated from decellularized pig hearts to create fibrous scaffolds adherent for cardiac myocytes (175). Silk fibroin isolated from the silk glands of Antheraea myllita, another species of silkworm, contains more Arg-Gly-Asp groups than silk fibroin from B. mori, resembling the structure of fibronectin. Thus, cardiac myocytes adhere directly to coverslips coated with solutions of A. myllita silk (140), reducing the need for additional functionalization. In summary, macromolecules derived from nonmammalian organisms are a unique source of nontoxic biomaterials that can also be used to engineer tunable substrates for physiological and pathological in vitro models of the myocardium.

Synthetic Hydrogels

Synthetic biomaterials generally provide greater flexibility in terms of composition, fabrication, and processing compared with natural biomaterials. Similar to natural biomaterials, the mechanical properties of synthetic biomaterials are dictated by their molecular structure, including the degree of polymer cross-linking. In most cases, synthetic biomaterials are nonadhesive to cells and must be functionalized with peptides or proteins before cell culture. Hence, the elasticity and protein composition of synthetic biomaterial scaffolds can be tuned independently, which is advantageous for decoupling mechanical and chemical features.

Polyacrylamide hydrogels are a common synthetic hydrogel for culturing cardiac myocytes. These hydrogels are typically prepared by inducing free-radical polymerization of acrylamide and bis-acrylamide with ammonium persulfate and a catalyst. The elastic modulus of the resulting hydrogel is controlled by adjusting the ratio of acrylamide to bis-acrylamide (10, 48, 69, 81, 82, 118). To induce myocyte adhesion, ECM proteins can be directly mixed into the hydrogel prepolymer solution before cross-linking (59, 176) or attached to the surface of the hydrogel using sulfo-SANPAH and ultraviolet light followed by ECM protein coating (10, 47, 69). Polyacrylamide hydrogels can also be doped with streptavidin-acrylamide to enable surface attachment of biotinylated ECM proteins such as fibronectin (119) or laminin (11). Polyethylene glycol (PEG) hydrogels have also been used as mechanically tunable synthetic substrates for in vitro cardiac models. For example, PEG-diacrylate is an ultraviolet-curable modified version of PEG that can be coated with ECM molecules. These hydrogels have been successfully used to culture neonatal rat cardiac myocytes (87) and cardiosphere-derived cells (86). In summary, synthetic hydrogels offer independent control over ECM elasticity and protein composition, which is a useful feature for probing multiple extracellular variables in healthy and diseased myocardium.

Synthetic Elastomers

Another widely used synthetic biomaterial is polydimethylsiloxane (PDMS), an organic, silicon-based, elastomeric polymer that is optically clear, moldable, biocompatible, and easy to fabricate. The most common PDMS, Sylgard 184, is prepared by mixing a base and curing agent at a defined ratio, coating or molding this prepolymer solution onto different surfaces (such as spin coating onto coverslips) and temperature curing the substrate. The ratio of base to curing agent can be adjusted to define the elastic modulus of the final elastomer (189). Another approach for adjusting the elastic modulus of Sylgard 184 is to mix with different ratios of Sylgard 527 silicone dielectric gel, which is much softer than Sylgard 184 (111, 134). To promote cell adhesion, PDMS surfaces are oxidized in an ultraviolet-ozone cleaner or plasma treater to induce covalent binding to ECM proteins such as fibronectin (31, 54, 63), laminin (147), or Matrigel (40). Hence, PDMS offers independent control over ECM rigidity and protein composition. Because of its versatility and biocompatibility, PDMS is also widely used to fabricate cell culture chambers (76) and microfluidic devices (114) for cardiac microphysiological systems.

In summary, as shown in Table 1, mammalian ECM-derived biomaterials are advantageous for in vitro cardiac models because they closely match the ECM of native myocardium and are naturally adhesive to cells. However, decoupling their chemical, physical, and/or mechanical properties can be challenging. In contrast, nonmammalian natural biomaterials and synthetic biomaterials are a “blank slate” that offer greater independent control over chemical and mechanical properties at the expense of additional modifications required for myocyte adhesion, a characteristic similar to synthetic biomaterials. Furthermore, most synthetic biomaterials cannot be used to encapsulate cells for engineering 3-D tissues because they are not amenable to cellular remodeling and/or are toxic as a 3-D substrate. Therefore, researchers must weigh these tradeoffs and select biomaterials that best fit their research question when designing in vitro models to probe the effects of ECM remodeling.

ENGINEERING AND INTERROGATING 2-D CARDIAC MICROPHYSIOLOGICAL SYSTEMS

In the healthy myocardium, cardiac myocytes are uniaxially aligned and embedded in a compliant ECM network. Both of these microenvironmental features can be affected by ECM remodeling, but neither can be easily controlled with conventional in vitro approaches. Thus, to determine the direct effects of ECM remodeling on cardiac structure and function, microfabrication techniques have been developed to engineer customizable microenvironments for cardiac myocyte cells and tissues in vitro. In this section, we will detail approaches for engineering cardiac tissues in 2-D and interrogating their key functions, including contractility, as a function of their microenvironment.

Microcontact Printing

Patterned growth of cardiac tissue was originally achieved by culturing cardiac myocytes on photoresist-coated coverslips selectively exposed to ultraviolet light through a custom photomask (150). A similar photolithography approach is still used today to fabricate wafer templates for microcontact printing. Microcontact printing is a technique similar to rubber ink stamping that is used to define the 2-D architecture of cells and tissues in vitro (144), including cardiac myocytes (57, 136). The first step in microcontact printing is to fabricate a silicon wafer with a relief of the desired pattern. PDMS is then cured on the wafer and carefully removed to create “stamps” containing the desired pattern as raised features on the surface. These PDMS stamps are then coated with ECM proteins, such as fibronectin, and inverted onto substrates, such as coverslips spin coated with a layer of PDMS, to transfer the desired pattern from the stamp to the substrate. Cardiac myocytes seeded on these substrates will attach only to the protein-printed areas and, therefore, conform to the pattern.

Many cardiac diseases are associated with distinct changes in myocyte shape and/or tissue architecture, which can be recapitulated in vitro with microcontact printing. For example, the aspect ratio of healthy cardiac myocytes is 7:1, but this decreases in concentric hypertrophy and increases in eccentric hypertrophy (58, 68). To mimic these changes, neonatal rat cardiac myocytes have been cultured on 2,500-µm² rectangles of fibronectin microcontact printed on PDMS with length-to-width aspect ratios ranging from 1:1 to 7:1, which demonstrated that cell geometry directly impacts sarcomere alignment (31) and nuclear morphology (30).

To engineer confluent cardiac tissues with controlled alignment, cardiac myocytes have been cultured on PDMS-coated coverslips microcontact printed with parallel lines of fibronectin. Early studies engineered aligned neonatal rat ventricular tissues by patterning 10- or 20-µm-wide lines of high-concentration fibronectin, separated by 10- or 20-µm-wide regions of low-concentration fibronectin. In these studies, aligned tissues generated greater contractile stress, propagated action potentials more rapidly, and had distinct Ca²⁺ handling compared with unaligned tissues (53, 63, 143). More recently, 15-µm-wide lines of high-concentration fibronectin separated by 2-µm-wide spacing have been used to engineer confluent neonatal rat (2) and human-induced pluripotent stem cell (hiPSC)-derived cardiac tissues (187) for a variety of studies. More complex cardiac tissues with adjacent regions of distinct alignment have also been engineered to determine how sharp changes in tissue architecture affect conduction (35) and contractility (93). These tissues replicate some of the drastic and heterogeneous changes in tissue alignment that can occur in many pathological settings, such as the infarct border zone. To more closely mimic native cardiac microstructure, micropatterns that match the fiber orientation of mouse hearts have also been developed, acquired by high-resolution diffusion tensor magnetic resonance imaging (15). This approach could also be adopted to determine the functional impact of disrupted fiber alignment in disease. Thus, microcontact printing is a powerful technique for mimicking changes in cell and tissue architecture that often accompany pathological remodeling of the ECM in the myocardium.

To probe the joint effects of cell/tissue architecture and ECM elasticity, cardiac myocytes can be cultured on microcontact-printed polyacrylamide hydrogels with tunable rigidity. Polyacrylamide hydrogels can be microcontact printed by inverting stamps coated with biotinylated ECM proteins onto partially dehydrated, streptavidin-doped polyacrylamide hydrogels (119). Alternatively, polyacrylamide hydrogels can be polymerized on chemically treated glass coverslips microcontact printed with ECM proteins such as Matrigel (Fig. 2A) (121, 146). This transfers the patterned protein but avoids stamping directly on the hydrogel, which can introduce defects due to its low rigidity. Microcontact-printed polyacrylamide hydrogels have been used to demonstrate that sarcomere alignment and density are coregulated by both myocyte shape and ECM elasticity in single neonatal rat cardiac myocytes (119). A similar technique was used to engineer pairs of cardiac myocytes and show that electrical (117) and mechanical (118) cell-cell coupling is impacted by cell shape and/or ECM rigidity. Square aligned cardiac microtissues on custom ECM have also been engineered by microcontact printing 200 × 200-µm squares of arrayed lines of fibronectin or laminin on soft or stiff polyacrylamide gels. These substrates were seeded with primary neonatal rat ventricular myocytes, which also contain a small population of cardiac fibroblasts. Thus, these “micromyocardium” tissues consist of both cardiac myocytes and fibroblasts with ratios that are relatively tunable, adding another level of microenvironmental control to this platform (11). Tunable PDMS substrates can also be microcontact printed and used to control both tissue alignment and ECM elasticity for engineered cardiac tissues (111), although PDMS generally offers less fine control over elastic modulus in lower ranges compared with hydrogels. In summary, microcontact printing combined with tunable synthetic biomaterials can be used to independently define features such as cell/tissue architecture, ECM elasticity, and ECM composition to establish the direct intra- and intercellular effects of healthy and pathological microenvironments.

Fig. 2.

Dictating two-dimensional tissue architecture with substrate micropatterning. Customizable polydimethylsiloxane (PDMS) stamps (gray) are fabricated with photolithography and soft lithography. A: in one example of microcontact printing, stamps are coated with proteins (red) such as Matrigel and inverted onto glass coverslips (blue), transferring the pattern to the coverslip. These patterned coverslips are inverted onto polyacrylamide hydrogel prepolymer solution to transfer the extracellular matrix protein pattern to the hydrogel as it polymerizes. B: micromolding is performed by inverting PDMS stamps onto a hydrogel prepolymer solution (yellow) such as methacrylated tropoelastin (MeTro). The stamp remains in position during hydrogel polymerization, which is initiated by ultraviolet (UV) light in this example, to mold the pattern from the PDMS stamp into the hydrogel. Images were adapted from Ref. 146 with permission from the National Academy of Sciences (A) and from Ref. 8 with permission from John Wiley & Sons (B).

Surface Molding

Micromolding and nanomolding entail casting desired patterns, typically parallel lines with controlled height, ridge width, and groove width, onto the surface of a substrate. These topographical features also induce cell alignment and are similar to the fiber-like topography of native myocardial ECM (87). Microgrooved PDMS substrates have been molded on templates prepared by photolithography and soft lithography (145) or acoustic micromachining (193) and subsequently coated with fibronectin. Neonatal rat (193) and hiPSC-derived (145) cardiac myocytes cultured on microgrooved PDMS surfaces demonstrate substrate-mediated changes in calcium dynamics. Microgrooved PDMS stamps can also be inverted onto hydrogel prepolymer solutions to mold the surface of the hydrogel as it polymerizes (Fig. 2B). This approach has been used for gelatin (116), tropoelastin (8), and chitosan-collagen hydrogels (41), which are already adhesive to cardiac myocytes, and alginate, which requires additional functionalization with ECM proteins (1). Similar techniques can also be used to fabricate isolated microchannels of gelatin hydrogel embedded with cardiac myocytes (13).

Templates for nanomolding have been fabricated by using electron beam lithography to etch nanometer scale features into photoresist-coated silicon wafers. These wafers are then used to mold ultraviolet-curable poly(urethane acrylate) templates, which are subsequently inverted onto the surface of ultraviolet-curable PEG hydrogel prepolymer solution (88, 90). These PEG hydrogels can then be coated with fibronectin and seeded with cardiac myocytes. Nanoscale features have been shown to impact cell geometry, action potential conduction velocity, and connexin43 expression in cardiac myocytes, demonstrating the functional effects of topographical ECM remodeling (38, 87). In summary, micromolding and nanomolding can also be used to engineer substrates for probing the combined effects of tissue alignment and ECM mechanics on cardiac tissue function, with additional versatility for investigating the impact of 3-D topographical features.

Quantifying Contractility

Quantifying cardiac cell and tissue contractility is essential for understanding the direct functional effects of pathological ECM remodeling and evaluating the efficacy of possible therapeutic interventions. Contractility can be indirectly quantified by measuring sarcomere shortening (67), spontaneous beating frequency (165), or contraction velocity (76, 77). Atomic force microscopy (155) and manipulation of magnetic beads attached to the cell surface (194) have also been used to measure contractile forces generated by single cardiac myocytes, although these techniques are relatively invasive and low-throughput. Contractile forces can also be quantified by culturing cells on a bed of PDMS micropillars and tracking the displacement of each micropillar (Fig. 3A) (62, 133, 201). Micropillar length and width can be adjusted to tune the stiffness of the substrate and model conditions such as fibrosis (55, 179). This method has been used to measure forces generated by single cells (56, 62) and cell pairs (107). However, micropillar platforms are relatively difficult to fabricate, limiting their scalability.

Fig. 3.

Quantifying contractility in engineered two-dimensional cardiac myocytes and tissues. A: micropillar substrates can be used to quantify contractile forces generated by cardiac myocytes. In this example, Ca²⁺ transients were also measured in engineered tissues on micropillar substrates. hESC-CM, human embryonic stem cell-derived cardiomyocytes. FEM, finite-element modeling. B–D: traction force microscopy and microcontact-printed polyacrylamide hydrogels have been implemented to quantify the combined effects of extracellular matrix elasticity and cell/tissue architecture on peak systolic traction stress generated by single cardiac myocytes (B), cardiac myocyte pairs (C), and multicellular cardiac tissues (D). Scale bars = 10 μm. In B, white denotes α-actinin. In C, red denotes actin, green denotes β-catenin, and blue denotes nuclei. In D, red denotes α-actinin, green denotes actin, and blue denotes nuclei. E: muscular thin film assay entails culturing cardiac myocytes on precut polymer cantilevers (shown here, micromolded gelatin hydrogels), which are released from the substrate at the time of analysis. Stress is calculated based on cantilever deflection. Images were adapted from Ref. 62 with permission from the American Chemical Society (A), Ref. 119 (B), Ref. 118 with permission from the National Academy of Sciences (C), Ref. 11 with permission from the Royal Society of Chemistry (D), and Ref. 116 with permission from Elsevier (E).

Another approach for quantifying contractility in vitro is traction force microscopy (TFM). In this assay, strains, stresses, and forces generated by cells are determined based on the displacement of their substrate. Typically, this assay is performed by culturing cells on polyacrylamide hydrogels doped with fluorescent beads and coated or microcontact printed with ECM proteins (10, 148). TFM has also been performed on alternative substrate such as PDMS (160) or gelatin hydrogels (100). As the cardiac myocytes contract, they displace the hydrogel and embedded beads. By capturing and analyzing videos of bead motion, the displacement field generated by the contracting cells can be quantified noninvasively. Based on the displacement field and the elastic modulus of the hydrogel, stresses and forces can also be calculated (36), providing a quantitative readout of contractility. Importantly, TFM substrates are relatively easy to fabricate and thus this assay has moderate scalability.

TFM studies have shown that ECM elasticity regulates contractility in unpatterned embryonic quail (48), neonatal rat (22), and hiPSC-derived (69) cardiac myocytes. In general, these studies agree that displacement decreases and stress increases as the elastic modulus of the ECM increases. TFM has also been combined with microcontact printing to assess the contractility of cardiac myocytes as a function of their architecture and the elasticity of the ECM (146). For example, neonatal rat cardiac myocytes have been cultured on polyacrylamide gels tuned to match the elasticity of healthy and fibrotic myocardium and microcontact printed with rectangles of fibronectin of distinct aspect ratios (Fig. 3B). Interestingly, myocytes with a healthy aspect ratio (7:1) generated the most amount of contractile work on substrates with physiological elasticity (98), whereas myocytes with aspect ratios closer to those in concentric hypertrophy (2:1) generated the most amount of work on substrates with fibrotic rigidity (119). This suggests that changes in myocyte shape observed in concentric hypertrophy may be an adaptation to a stiffer fibrotic ECM and/or other forms of increased afterload observed in this disease. A similar technique was used to engineer “pairs” of neonatal rat cardiac myocytes and quantify stresses transmitted both extracellularly and intercellularly (Fig. 3C). With this system, a weakened relationship between cytoskeletal organization and force generation and defects in mechanical cell-cell coupling was identified on stiffer hydrogels compared with softer hydrogels (118), suggesting that remodeling of the ECM can directly affect cell-cell junction formation. To probe the combined effects of ECM rigidity, ECM ligand, and fibroblasts on cardiac tissue contractility, aligned MicroMyocardium with distinct myocyte-to-fibroblast ratios was engineered on polyacrylamide hydrogels microcontact printed with fibronectin or laminin (11) (Fig. 3D). On stiffer polyacrylamide hydrogels, tissues generated similar amounts of force but less work, as also demonstrated by other studies (139). Interestingly, contractile force was not significantly affected by ECM protein composition or cardiac myocyte to fibroblast ratio (11), suggesting that ECM rigidity dominates over changes in ECM ligand or fibroblast density. In summary, TFM combined with microcontact printing is powerful for delineating the effects of multiple forms of cellular and extracellular remodeling on cardiac myocyte contractility. However, scalability can be limited due to the amount of time required for substrate fabrication, reliance on high-speed, high-resolution imaging, and relatively intensive data processing.

Cardiac tissue contractility has also been quantified by measuring the deflection of cells grown on deformable membranes. For example, cardiac myocytes have been cultured on circular silicone membranes that deflect with each contraction. Tensile stress can be calculated from membrane deflection (60). A similar approach is to culture and monitor cardiac myocytes on PDMS microcantilevers (135). However, in both of these examples, the cardiac myocytes are not aligned, confounding force measurements. To overcome this, muscular thin films (MTFs) were established to provide a more accurate quantification of force by combining microcontact printing with deformable cantilevers (Fig. 3E). MTFs are fabricated by laser-engraving releasable, millimeter-scale cantilevers into microcontact-printed PDMS (2, 63, 187) or micromolded gelatin (116) or alginate (1) hydrogels selectively attached to glass coverslips at their base. Cantilevers are cut parallel to the micropatterning such that seeded cells form tissues aligned parallel to the length of the cantilever. After tissue formation, the user peels each cantilever and electrically paces the tissues, causing tissue contraction and cantilever deflection. Film movement is recorded and analyzed to calculate stresses generated by the tissue based on cantilever curvature (3, 53, 63). This platform has been successfully used to quantify contractile stresses in response to tissue architecture (53), genetic mutations (187), and drug exposure (2, 116), demonstrating its utility for a wide array of disease modeling studies. However, although MTFs can be fabricated from hydrogels, the range of elastic moduli compatible with this assay is relatively narrow compared with TFM. Thus, MTFs are well suited for determining how remodeling of select disease-relevant parameters, such as tissue architecture or genetic mutations, compromise contractility but have limitations related to probing ECM mechanics.

Quantifying Electrophysiology

Myofibril contraction is dependent on the release of Ca²⁺ from the sarcoplasmic reticulum, which is initiated by an action potential (21). Action potentials are rapidly propagated across myocyte cell-cell junctions by low-resistance gap junction channels to ensure that myocardial tissue contracts in synchrony (91). Thus, the morphology, dynamics, and propagation velocity of action potentials and Ca²⁺ transients are critical functional metrics because each can directly impact cell and tissue contractility. Historically, patch clamp has been one of the most common techniques to record changes in membrane voltage or current from single cardiac myocytes (19). However, although this technique is highly sensitive, it is low throughput and requires extensive technical expertise and specialized equipment. Another approach that is more scalable is to use voltage- and Ca²⁺-sensitive fluorescent probes to optically record the morphology and/or propagation of action potentials and Ca²⁺ transients in cardiac cells and tissues (83). Recently, these dyes have been implemented with optogenetic stimulation of cardiac myocytes in multiwell plates as an all-optical, contact-free approach to electrophysiology screening (92), further improving throughput and automation.

Several studies have investigated relationships between the ECM and electrophysiology. At the single-cell level, the elastic modulus of the ECM has been shown to alter the morphology of Ca²⁺ transients (82) and action potentials (26) in unpatterned neonatal rat cardiac myocytes, as measured by intracellular Ca²⁺ dyes and patch clamp, respectively. At the two-cell level, micropatterned neonatal rat cardiac myocytes and dual voltage clamp have demonstrated that myocyte shape and cell-cell junction morphology impacts connexin43 density and intercellular conductance (118). At the tissue level, high-speed imaging of transmembane voltage and/or intracellular Ca²⁺ dyes has shown that tissue alignment induced by microcontact printing (35, 53, 63), microtopography (42), or nanotopography (87) increases action potential duration and the anisotropy ratio of conduction. Furthermore, the elastic modulus and topography of the ECM have been shown to impact the anisotropy ratio of Ca²⁺ wave propagation velocity (142). Another technique for recording conduction velocity is to culture cardiac myocytes on microelectrode arrays, which are glass substrates with a grid of electrodes on the surface. The microelectrode array surface is compatible with select 2-D tissue engineering techniques such as microcontact printing (124, 172). Micromolded gelatin hydrogels have also been engineered on microelectrode arrays (97), which minimally interfere with electrical signals due to their high water content. Collectively, these studies indicate that properties of the ECM impact electrophysiological parameters in cardiac myocytes at the cell and tissue levels.

Metabolic Profiling

Cardiac myocytes have relatively high metabolic demands, as the human heart consumes ∼30 kg/day ATP (46). In healthy adult cardiac myocytes, ATP is produced primarily by the oxidative phosphorylation of products generated by the β-oxidation of fatty acids in the mitochondria. Consequently, cardiac myocytes are densely packed with mitochondria that align into parallel lanes packed between myofibrils to efficiently deliver the ATP that fuels sarcomere shortening (7, 78, 127). However, in many pathological settings, cardiac myocytes revert to a neonatal metabolic phenotype (61, 173, 177) that is characterized by increased reliance on glycolysis and glucose oxidative phosphorylation (39, 108, 185) and changes in the turnover of the mitochondrial network (5, 182). However, the mechanisms driving these metabolic changes, including any effects secondary to ECM remodeling, are poorly understood. In other cell types, mitochondria have been shown to be mechanosensitive (16, 32), suggesting that ECM remodeling could impact mitochondria structure and/or function in cardiac myocytes. Additionally, because of their proximity to myofibrils, mitochondria are mechanically compressed with each contraction (127) and are known to interact with other cytoskeletal proteins, such as microtubules (6, 65). Thus, the cytoskeleton could transduce signals from the ECM directly to the mitochondria to modulate metabolism based on the microenvironment. However, few studies have investigated the impact of ECM remodeling on mitochondria in cardiac myocytes specifically, as described below.

Several techniques exist for metabolically profiling cardiac myocytes in vitro. Intracellular ATP production can be quantified by lysing cells and using luciferin/luciferase-based assays (132, 141). However, it is not possible to differentiate between ATP generated by glycolytic or mitochondrial pathways with this approach. ATP can also be quantified using liquid chromatography-mass spectroscopy (33, 159), which can be combined with radioactive carbon-13 tagging to identify the specific metabolic pathways preferred by the cells (44, 137). Mitochondrial membrane potential can be assessed by imaging cells loaded with mitochondrial dyes, such as TMRM/TMRE (Fig. 4A) (45, 169, 170, 191) and rhodamine 123 (103, 113), although these dyes have relatively low sensitivity. Many imaging and plate reader-based assays also exist for quantifying oxidative stress and reactive oxygen species, a byproduct of oxidative phosphorylation (Fig. 4A) (64, 187, 191). Mitochondrial function can be more directly characterized in vitro by measuring oxygen consumption rates (OCRs). Historically, OCRs have been measured with a Clark electrode (45, 141), which measures ambient oxygen concentration in a liquid using a catalytic platinum surface. Another approach is to evaluate OCRs in mitochondria isolated from tissue samples or cells in suspension using an oxygraph, a closed chamber that contains an oxygen sensor (74, 75). However, both of these techniques are relatively low throughput and incompatible with cells adhered to culture surfaces, which is essential for establishing direct relationships between the microenvironment and metabolism in cardiac myocytes.

Fig. 4.

Metabolic profiling of cardiac myocytes and engineered two-dimensional cardiac tissues. A: cardiac myocytes isolated from adult rats subjected to sham surgery or aortocaval fistula (ACF) and tagged for mitochondrial membrane potential (TMRM) or reactive oxygen species (CM-DCF). B: microcontact-printed polydimethylsiloxane (PDMS) disks of distinct elasticities were fabricated in a multistep process (i–vi), transferred to the wells of modified microplates (vi), and seeded with neonatal rat ventricular myocytes (vii). These plates were inserted into an extracellular flux analyzer to quantify the effects of extracellular matrix elasticity and tissue architecture on oxygen consumption rates (OCRs). C and D: representative OCR profiles for isotropic (C) and aligned (D) tissues demonstrate distinct profiles due to extracellular matrix rigidity and tissue alignment. Images were adapted from Ref. 191 (A) and Ref. 111 (B–D).

A more recent technique that is compatible with conventional cell culture is extracellular flux analysis. This approach simultaneously measures changes in both oxygen concentration and pH to quantify OCRs and extracellular acidification rate for cultured cells or isolated mitochondria (4, 167). These experiments require cells or tissues to be cultured within a modified microplate, which can be easily coated with gelatin hydrogels or tunable PDMS to determine the effects of these biomaterials on cardiac myocyte metabolism (116). To establish how both ECM elasticity and tissue architecture regulate mitochondrial function, PDMS disks with distinct elastic moduli have been microcontact printed with fibronectin lines, transferred to the bottom of microplate wells, and seeded with neonatal rat ventricular myocytes (Fig. 4B). In this study, baseline and ATP production OCR values increased with ECM rigidity for both isotropic and aligned tissues, suggesting that mitochondria consume more oxygen in stiffer microenvironments (Fig. 4C). However, maximum respiration and spare respiratory OCR values were independent of ECM rigidity for isotropic tissues but were significantly higher in aligned tissues on the most rigid substrate (Fig. 4D) (111). These data suggest that ECM rigidity regulates baseline metabolic function, but both ECM rigidity and tissue architecture influence metabolic stress responses. In summary, cardiac microphysiological systems integrated with functional assays have provided many new insights into the effects of multiple forms of ECM remodeling on contractility, electrophysiology, and metabolism.

ENGINEERING AND INTERROGATING 3-D CARDIAC MICROPHYSIOLOGICAL SYSTEMS

In general, tissues engineered in 3-D replicate the architecture of native myocardium more closely than 2-D engineered tissues, and, therefore, they are potentially more accurate representations of healthy and diseased myocardium. The simplest approach to engineering 3-D tissues is to create spherical aggregates of cells grown in suspension (17, 18, 84) or adhered to a substrate (168). Previous work has shown that chick cardiac cells in these 3-D spheroids exhibit differences in morphology, contractile ability, and myofibril organization compared with cells in 2-D (168). Another approach is to fabricate cell sheets by culturing cells on dishes coated with a temperature-sensitive polymer and ECM hydrogel. After tissue maturation, the temperature is reduced to release cell sheets from the dish, which can be stacked to form 3-D tissues (66). Although spheroids and stacked cell sheets replicate the 3-D interactions of cells, they lack user control over key architectural features, such as ECM and myocyte alignment. 3-D cardiac tissues with aligned, elongated cardiac myocytes have been engineered by mixing cells into a fiber-forming ECM-derived hydrogel prepolymer solution and injecting this mixture into a mold with supportive structures, such as pillars. As the hydrogel polymerizes around the support structures, cells are encapsulated inside and gradually remodel the ECM, ultimately creating tension and forming a 3-D tissue with aligned fibers and cardiac myocytes (Fig. 5A). Thus, these 3-D tissues replicate both the alignment and dense cell-cell interactions of native myocardium. 3-D tissues in the shape of an elongated bundle have been fabricated on PDMS posts manufactured using soft lithography (28), laser-etched molds (123), silk sutures (190), or titanium wires (166). Larger square tissue constructs have been engineered by molding cell-matrix mixtures on arrays of PDMS pillars supported by nylon frames (23, 80, 199). Recently, this approach was implemented to engineer a 5 mm-wide “giga-patch” of iPSC-derived cardiac myocytes (163), an impressive demonstration of the scalability of this method.

Fig. 5.

Engineering and functionally assessing engineered three-dimensional cardiac tissues. A: aligned three-dimensional cardiac tissues are fabricated by filling a custom polydimethylsiloxane (PDMS) mold with a mixture of cells and an extracellular matrix hydrogel prepolymer solution, which gradually solidifies and compacts into a tissue supported by the PDMS pillars. B: the electrical and mechanical functionality of “iWire” engineered three-dimensional cardiac tissue constructs are quantified using patch clamp and optical recording of a flexible probe, respectively. Images were adapted from Refs. 157 (A) and Ref. 166 with permission from Elsevier (B).

A variety of ECM mixtures, including combinations of collagen, Matrigel, and fibrin (49, 196), have been used as scaffolds for 3-D tissue constructs. Thus, the composition and elasticity of the ECM is somewhat tunable in engineered 3-D tissues. However, ECM composition does have some constraints because it must be robust enough to maintain the structural integrity of the tissue. For example, whereas collagen hydrogels have high physiological relevance, collagen fibers are relatively rigid, and, therefore, bundles fabricated solely with collagen are susceptible to breakage. Fibrin is more extensible than collagen (106), and therefore, bundles with fibrin or a mixture of fibrin and collagen are typically more robust than pure collagen bundles. Additionally, supporting cell populations, such as cardiac fibroblasts or other stromal cell populations, can be easily mixed into hydrogel mixtures before polymerization, demonstrating another tunable aspect of engineered 3-D tissues (149, 154, 156). Because cardiac fibroblasts are a critical part of the fibrotic response, integrating these cells into 3-D tissues can be leveraged to mimic aspects of pathological ECM remodeling. Thus, engineering 3-D cardiac tissues on microfabricated molds enables control over some microenvironmental features, such as macroscale tissue alignment, but ECM composition and elasticity are somewhat limited to combinations that can maintain stable tissues. Furthermore, controlling microscale features in 3-D is not feasible with this approach.

The contractility of 3-D engineered cardiac tissue bundles is quantified by imaging movements at tissue attachment sites (28), attaching tissues to force sensors (23, 153), or measuring the deflection of a flexible probe (Fig. 5B) (161). Similar to 2-D tissues, electrophysiological function can be characterized using patch clamp (166) or optical mapping of cell membrane potentials or Ca²⁺ transients (28, 104). However, although approaches for engineering 3-D cardiac tissues and quantifying their function have been developed, these platforms have not been extensively used for modeling pathological changes in the ECM, in part because of the challenges associated with robustly controlling ECM composition and tissue architecture in 3-D compared with 2-D.

OUTLOOK

Largely because of parallel advances in microfabrication, materials science, imaging, and cellular reprogramming, cardiac microphysiological systems have become increasingly sophisticated and have already contributed many important new insights into cardiac disease processes. Here, we will highlight ongoing challenges and areas of growth that will continue to advance these platforms.

Renewable Sources of Human Cardiac Myocytes

Because adult cardiac myocytes are postmitotic and human heart biopsies are extremely rare, human stem cell-derived cardiac myocytes are the only practical source of human cardiac myocytes (122). hiPSC-derived cardiac myocytes are especially attractive because they are reprogrammed from adult somatic cells (typically skin fibroblasts) (34, 102), and, therefore, they can be used for patient-specific disease modeling and drug screening, especially when integrated into microphysiological systems (70, 76, 187). However, existing protocols for differentiating cardiac myocytes from stem cell populations generate cardiac myocytes at an embryonic or fetal state (192), with a weaker contractile phenotype (164). The relevance of these immature cardiac myocytes to adult cardiac myocytes is questionable. Thus, protocols for promoting the maturation of human stem cell-derived cardiac myocytes (110), which will likely include modulating the microenvironment (12, 195), are critically needed to enhance the relevance of cardiac microphysiological systems to model healthy and diseased human adult myocardium.

Fabrication Technologies

New fabrication technologies are presently under development to further increase experimental control, reduce cost, and improve throughput for cardiac microphysiological systems. As described above, 2-D engineered tissues poorly mimic the multilayered architecture of the myocardium, whereas existing approaches to engineer 3-D tissues offer limited spatial control over cell and ECM microarchitecture. To address these limitations, 3-D bioprinting systems are presently being explored for engineering cardiac tissues with enhanced control over cell and ECM placement in all three dimensions, including incorporation of vascular networks (95, 200). However, to date, most 3-D bioprinting systems are extrusion based, for which the “ink” of biomaterials, with or without cells, is extruded through a nozzle. For this reason, the inks have been limited primarily to purely scaffolding materials (178), cardiac spheroids (129), or relatively robust cell populations, such as fibroblasts (95), endothelial cells (79, 200), or myoblasts (71). It has yet to be shown whether suspensions of cardiac myocytes can be viably printed in extrusion-based bioprinting systems. Alternative approaches that avoid extrusion, such as digital light processing-based 3-D bioprinting (202), may be more supportive of cell viability. This technique projects patterns of light to selectively cross-link a solution of light-sensitive hydrogel prepolymer mixed with cells. Despite its present limitations, 3-D bioprinting has vast potential for engineering cardiac microphysiological systems with enhanced spatial control over multiple cell populations and the ECM, which will greatly improve our ability to model diverse forms of pathological remodeling in health and disease.

Sensing Technologies

Today, many of the functional assays integrated into microphysiological systems, especially for contractility, rely on imaging-based techniques, which inherently have limited scalability due to the requirement for microscopy. Thus, integrating sensors that provide direct functional readouts with high spatial and temporal resolution will increase throughput and accuracy. For example, force sensors were recently integrated into MTFs to provide electrical readouts of contractility without the use of a microscope (105). In terms of electrophysiology, microelectrode arrays can provide real-time measurements of extracellular field potentials in 2-D tissues, but with limited spatial resolution (174). Electrophysiological measurements from 3-D tissues still rely largely on voltage- or Ca²⁺-sensitive dyes. Recently, flexible sensor arrays have been developed to readout electrical parameters from whole hearts (51). A similar system could potentially be adapted for assessing electrophysiology in engineered heart tissues as well. Ultimately, integrating sensors that provide real-time, constant measurements of multiple functional outputs will greatly enhance disease modeling studies within cardiac microphysiological systems.

Disease Modeling

Microphysiological systems can be leveraged to model many diverse forms of acquired and inherited cardiac diseases. As described above, pathological hypertrophy, increases in afterload, and fibrosis can be modeled in 2-D by engineering tissues on stiffer biomaterials, which has been shown to impact contractility (11, 48, 119), cell-cell coupling (118), and metabolism (111). These same pathologies can be modeled with 3-D tissue systems using stiffer support posts, which induces myocyte enlargement, decreases force generation, and activates fetal gene programs (72). Sophisticated microfluidic systems can also be used to simulate the hypoxic microenvironment following a myocardial infarction (85).

Inherited cardiomyopathies can be modeled with microphysiological systems by integrating iPSC-derived cardiac myocytes acquired directly from afflicted patients. Importantly, microphysiological systems are critical for robustly extracting quantitative functional metrics from these cells, as demonstrated in a 2-D engineered tissue model of Barth Syndrome (187) and a 3-D engineered tissue model of titinopathies (70). In both of these examples, microphysiological systems were used to engineer patient iPSC-derived cardiac myocytes into aligned tissues and quantify contractile stresses in response to specific interventions. Because many inherited cardiomyopathies are also associated with ECM remodeling, such as fibrosis, utilizing microphysiological systems will lead to unique insight into how cell phenotype and ECM remodeling jointly impact cardiac tissue function in a diverse range of human diseases.

Integration of Supporting Cell Populations

Most cardiac microphysiological systems today consist primarily of cardiac myocytes. However, supporting cells in the myocardium, such as fibroblasts and endothelial cells, also play a role in disease progression. Thus, biological relevance of existing cardiac microphysiological systems can be improved by integrating these supporting cell populations. Existing studies have shown that mixing fibroblasts (138, 156) or endothelial cells (162) with cardiac myocytes improves tissue structure and function. However, randomly coculturing these cells does not mimic their relative positioning in native myocardium. To overcome this, microfluidic devices have been used to culture a layer of endothelial cells on a porous PDMS membrane suspended above a layer of cardiac myocytes. The endothelial cell layer was used to regulate drug transport to the underlying cardiac myocytes, which more closely recapitulates the relationship between these cells in vivo (112). Channels for vasculature have also been fabricated in hydrogels by using removable spacers (184) or 3-D bioprinting a dissolvable filament (96, 120), which can then be removed. The evacuated channels can then be seeded with endothelial cells, which form a lumen. More recently, tubes of endothelial cells have been directly bioprinted and subsequently coated with cardiac myocytes to generate endothelialized myocardium (200). Ongoing refinement of these sophisticated techniques for controllably integrating supporting cell populations will continue to improve the physiological relevance of cardiac microphysiological systems.

In summary, cardiac microphysiological systems are powerful new in vitro platforms for quantifying the functional impact of physiological and pathological ECM remodeling and many other disease-relevant variables. Although these platforms still have many fundamentally artificial characteristics, their key advantage is user control over many of the diverse extracellular cues observed in healthy and diseased myocardium. Additionally, these platforms provide real-time access to cell and tissue structure and physiology with relatively high spatial and temporal resolution. Thus, these platforms can provide new insights into biological processes that complement but do not replace existing in vivo and ex vivo models. Ongoing efforts to improve the maturation of human stem cell-derived cardiac myocytes, develop new fabrication and sensing technologies, and integrate supporting cell populations will continue to evolve these platforms for human-relevant disease modeling, leading to new therapeutic targets and treatment strategies for many diverse forms of cardiac dysfunction.

GRANTS

This project was supported by the USC Viterbi School of Engineering, USC Women in Science and Engineering, American Heart Association Scientist Development Grant 16SDG29950005 (to M. L. McCain), a Saban Research Institute Team Science Grant (to M. L. McCain), and the USC Graduate School (Provost Fellowship to N. R. Ariyasinghe and an Annenberg Fellowship to D. M. Lyra-Leite).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

N.R.A., D.M.L.-L., and M.L.M. prepared figures; N.R.A., D.M.L.-L., and M.L.M. drafted manuscript; N.R.A., D.M.L.-L., and M.L.M. edited and revised manuscript; N.R.A., D.M.L.-L., and M.L.M. approved final version of manuscript.