When comparing PSC-CMs to the developmental CM stages, there exhibits apparent morphological similarities to early fetal CM stages. PSC-CMs and both stages of fetal CMs are morphologically round and single nucleated (Kim et al., 1992; Veerman et al., 2015; Feric and Radisic, 2016). Postnatally, CMs switch to an elongated morphology and some become binucleated (Anversa et al., 1980). Mature adult CMs are elongated and rod-shaped, with about thirty percent of the cells being binucleated (Olivetti et al., 1996; Mollova et al., 2013). The surface area of PSC-CMs and both stages of fetal CMs ranges from about 1,000–1,300 μm² compared to adult CM surface area of about 10,000–14,000 μm² (Li et al., 1996; Ribeiro et al., 2015). As a major transitional period, neonatal CMs increase in surface area with values that are much greater than fetal and PSC-CMs but lower than adult (Anversa et al., 1980). Surface area increases are the result of CMs switching from hyperplastic growth, which involves CM proliferation, to hypertrophic growth, which involves increase in CM size during the neonatal development stages (Bernardo et al., 2010). Neonatal CMs are more aligned than either stage of fetal CMs (Anversa et al., 1980; Feric and Radisic, 2016). Adult CMs are anisotropic, which facilitates electrical conduction and contractility efficiency, compared to the random alignment of PSC-CMs and early fetal CMs (Gerdes et al., 1992; Feric and Radisic, 2016).

There exists substantial differences in the contractile machinery. PSC-CM sarcomeres, the basic contractile unit of muscular tissue, are less organized than late stage fetal CMs (Mummery et al., 2003; Feric and Radisic, 2016). Adult CM sarcomeres are highly aligned and uniformly distributed along cell edges and contain high densities of aligned myofibrillar structures (Maillet et al., 2013; Feric and Radisic, 2016). In comparison, PSC-CMs and early fetal CMs are randomly aligned and distributed, found mostly in the perinuclear region, and contain less dense myofibrillar structures (Chacko, 1976; Mummery et al., 2003; Feric and Radisic, 2016). Late fetal CMs are well organized and aligned while neonatal CMs have even more alignment and a substantial increase in fiber density (Chacko, 1976).

Striations are hallmarks of mature sarcomeres formed from overlapping and non-overlapping myofilaments. In adult CMs, all the prominent myofilament regions can be seen including the I bands, A bands, M-line, and H-zone, as well as the Z-disks, which denote the sarcomere ends (Gregorio and Antin, 2000). In contrast, PSC-CM sarcomere striations are mostly indistinguishable. PSC-CMs usually only have Z disks and sometimes I bands since the immature sarcomeres are not organized to provide the different myofilament regions (Figure 1B; Snir et al., 2003). Only one study has been able to produce a small percentage of IPS-CMs with all myofilament regions after 360 days of culturing, but contractile protein expression levels still remained much lower than adult CMs (Kamakura et al., 2013). Early fetal CMs contain only early forming Z disks but late fetal CMs have completely developed Z disks as well as I bands, A bands, and H zones (Figure 1A). M-lines develop during the neonatal period (Chacko, 1976).

Although PSC-CMs have disorganized sarcomeres, the majority of the contractile proteins found in mature CMs can be detected in PSC-CMs, but at much lower levels of expression or present in different isoforms (Veerman et al., 2015). These isoforms change naturally during CM development. The myosin heavy chain structural protein isoform, beta-myosin heavy chain (β-MHC), is prevalent in mature adult ventricular CMs while alpha-myosin heavy chain (α-MHC) is much lower (Reiser et al., 2001). In contrast, both stages of fetal CMs contain both isoforms but α-MHC is more predominant (Razeghi et al., 2001; Xu et al., 2009). After birth, the ratio of β-MHC to α-MHC increases in neonatal CMs to lead to the adult form (Siedner et al., 2003). PSC-CMs have varying amounts of both MHC isoforms due to the heterogeneous culture of various cardiac cell types (Xu et al., 2009; Ivashchenko et al., 2013; Lundy et al., 2013). The myosin light chain (MLC) isoforms change during development. Both stages of fetal CMs contain both isoforms, MLC2a and MLC2v, however, in adult ventricular CMs, only MLC2v can be found (Hailstones et al., 1992; Macera et al., 1992). MLC isoform prevalence switch occurs during the neonatal stage (Chuva de Sousa Lopes et al., 2006). Similar to the MHC isoforms, PSC-CMs contain both isoforms of MLC (Zhang et al., 2013). PSC-CMs likewise exhibit low levels of expression of troponin isoforms, the protein involved in regulating myosin binding to actin, and alpha-actin isoforms, which contributes to the actin myofilaments, compared to adult CMs (Khan et al., 2015; Veerman et al., 2015; Li et al., 2016). Fetal CMs express mostly cTnT1 and cTnT2 which then switch to higher amounts of cTnT3 and cTnT4 during the neonatal stage and is abundant in adult CMs (Siedner et al., 2003). Troponin I isoform changes from slow skeletal troponin I, predominant in fetal and neonatal CMs, to cardiac troponin I found mostly in adult CMs (Siedner et al., 2003). Fetal and neonatal CMs contain mostly skeletal alpha-actin which changes to cardiac alpha-actin within adult CMs (Schwartz et al., 1992).

Due to these differences in protein isoforms and organization, PSC-CMs are unable to contract efficiently, which limits their overall functionality without any further improvements in their maturity. In three-dimensional culture, the active contraction force of PSC-CMs ranges from about 0.1 to 0.5 mN/mm² while adult CMs generate two orders of magnitude larger contractile forces of 10–50 mN/mm² (Mulieri et al., 1992; Holubarsch et al., 1998; Schaaf et al., 2011; Tulloch et al., 2011). Fetal CMs contract with 0.4 mN/mm² forces, which is within the range of PSC-CMs (Ribeiro et al., 2015). Neonatal CMs dramatically increase their contraction strength as the contractility machinery matures, which causes a 2 or 3 fold increase in force compared to the fetal stage (Wiegerinck et al., 2009).

PSC-CMs are known to spontaneously contract asynchronously, due to their immature electrical coupling, while adult CMs are only excited when provided an electrical stimulus and contract in a synchronous manner (Keung et al., 2014). In comparison, early fetal CMs contract very asynchronously and spontaneously during the early stages of heart tube formation. As development progresses, late fetal CMs are more synchronized. Synchronous beating continues in neonatal stages on to adulthood with greater electrical coupling (Liau et al., 2012). Normally, ventricular adult CMs have distinct action potential profiles with four phases. Phase 0 includes sodium and calcium channels, which depolarize the cell. In phase 1, outward potassium channels cause partial cell membrane repolarization. Phase 2 contains a plateau where calcium moves inward and potassium travels outward. This calcium influx leads to excitation-contraction coupling where the cell converts the electrical signal into cross-bridge cycle mediated contractions. Phase 3 ensues immediately after with outward potassium movement bringing the resting membrane potential (RMP) back down to −85 mV. Phase 4 involves the inward rectifying potassium channel (KCNJ2), which maintains the RMP (van den Heuvel et al., 2014).

Similar to their structural machinery, ion channels change throughout cardiomyocyte development, leading to varying action potential profiles and electrophysiological maturity. Unlike adult CMs, PSC-CMs lack a majority of KCNJ2, which causes unstable RMP ranges of −50 to −60 mV (Sartiani et al., 2007; Amin et al., 2010; Ma et al., 2011). Early fetal CMs have minute amounts of KCNJ2, which increases 10 fold in the late fetal CMs and progressively increases into the adult stage (Liu et al., 2016). In correlation with this progressive increase, the RMP of CM stages becomes progressively more negative from −40 mV in early fetal stages to −85 mV in adult stages (Gennser and Nilsson, 1970; Mummery et al., 2003). PSC-CMs have pacemaker current channels, which are absent or expressed in low amounts in adult ventricular CMs (Sartiani et al., 2007; Verkerk et al., 2007; Lieu et al., 2013). Similarly, early fetal CMs express high levels of pacemaker channels, which decrease during the late fetal and neonatal stages (Liu et al., 2016). Pacemaker channels along with unstable RMPs are responsible for the spontaneous contractility of PSC-CMs and early fetal CMs (Veerman et al., 2015). T-type calcium channels, that are prominent within cardiac conducting cells, are not present within adult ventricular CMs but are detected in varying amounts within PSC-CMs most likely due to their heterogeneous differentiation (Bkaily et al., 2005; Ono and Iijima, 2010). During development, early fetal CMs express low T-type calcium channel amounts which decreases in late fetal and is basically absent by the neonatal stage of CM development (van den Heuvel et al., 2014). In contrast, L-type calcium channels increase progressively from the early fetal stage into the adult stage. PSC-CMs generally have lower amounts of L-type calcium channels within the fetal range (Liu et al., 2016).

Underdeveloped cell-cell connections between PSC-CMs similarly impact their electrophysiological function (Figure 1D). Cardiac specific cell-cell connections, called intercalated discs (ICDs), contain adheren junctions, desmosomes, gap junctions, and sodium channels, which allow rapid transmission of electrical signals and contractile forces within CMs (Noorman et al., 2009; Wang et al., 2012). Connexin-43, N-cadherin, SCN5A are respectively the most prevalent gap junction, adheren junction proteins, and sodium channels, and are found within adult, fetal, neonatal, and PSC-CMs at similar amounts (Chen et al., 1994; Noorman et al., 2009; Lundy et al., 2013; Liu et al., 2016). However, spatiotemporal distribution of these structures changes during development (Figure 1C). Adult CMs have them densely distributed around mature ICDs (Angst et al., 1997; Jansen et al., 2010; Feric and Radisic, 2016). Early fetal CMs and PSC-CMs contain these proteins sporadically distributed along the cell membrane with no distinguishable intercalated discs present, which greatly contributes to their immature asynchronous beating behavior (Vreeker et al., 2014; Feric and Radisic, 2016). Late fetal CMs have some localization of N-cadherin and sodium channels to their premature ICD but sporadic connexin-43 distribution. Sodium channels and N-cadherins are co-localized with each other on ICDs at 5 months postnatal development while connexin-43 is not completely localized to ICDs until 7 years (Vreeker et al., 2014). Neonatal ICDs have a less folded morphology with a smaller width compared to adult ICDs (Wilson et al., 2014).

The immature electrophysiology of fetal CMs and PSC-CMs directly affects several electrical efficiency parameters. The depolarization velocity of fetal and PSC-CMs ranges from a 6 to 50 fold decreased amount compared to adult CMs (Mummery et al., 2003; Koncz et al., 2011; Doss et al., 2012; Ribeiro et al., 2015; Feric and Radisic, 2016). The capacitance, which is the ionic charge that is necessary to fluctuate the potential of the cell membrane, is generally lower in PSC-CMs and fetal CMs compared to adult and neonatal CMs (Drouin et al., 1995; Zhu et al., 2009; Polak and Fijorek, 2012; Sheng et al., 2012; Feric and Radisic, 2016). The velocity of the action potential propagation through CMs, known as conduction velocity, is lower in early PSC-CMs than fetal CMs (Gennser and Nilsson, 1970; Taggart et al., 2000; Lee et al., 2012; Veerman et al., 2015). Conduction velocity naturally increases during CM development (Gennser and Nilsson, 1970; Yang et al., 2014).

Structures involved in calcium handling likewise mature throughout development (Figure 2C). Normally in adult CMs, action potentials propagate down transverse tubules, which are deep protrusions of the sarcolemma that permit efficient action potential propagation. These action potentials cause activation of L-type calcium channels along the transverse tubules, which leads to calcium induced calcium release from the sarcoplasmic reticulum (SR) by ryanodine receptors (RyR) that are co-localized with the L-type calcium channels. The sarcoplasmic reticulum stores significant amounts of calcium through the use of calcium handling buffer proteins called calsequestrin (CASQ2). This intracellular calcium release leads to excitation-contraction coupling as the myofilaments undergo cross-bridge cycles to contract. After the contractions are complete, intracellular calcium is taken back up into the sarcoplasmic reticulum by a specialized structure called the sarcoplasmic endoplasmic reticulum calcium ion ATPase (SERCA). Calcium is transported out of the cell by the sodium calcium exchanger (NCX) (van den Heuvel et al., 2014).

Conversely, PSC-CMs have underdeveloped calcium handling protein structures (Figure 2D). They do not have transverse tubules, which significantly effects the efficiency of the calcium handling and excitation-contraction coupling (Snir et al., 2003; Satin et al., 2008). PSC-CMs have underdeveloped SR with generally low amounts of SERCA, RyR, and CASQ2 (Dolnikov et al., 2006; Binah et al., 2007; Satin et al., 2008; Lee et al., 2011; Synnergren et al., 2011; van den Heuvel et al., 2014). The calcium handling maturity of PSC-CMs are similar to early fetal CMs. Transverse tubules are absent in early fetal CMs and present as only small indentations of the sarcolemma in late fetal CMs (Ziman et al., 2010). During the neonatal stage, transverse tubules become fully developed by about 2–3 weeks (Liau et al., 2012). In terms of calcium kinetics, since fetal CMs and PSC-CMs have lower amounts of CASQ2, RyR, SERCA, and L-type calcium channels, the immature calcium handling relies more heavily on sarcolemma calcium inflow most likely from T-type calcium channels and the NCX as well as small amounts of calcium release from the SR by IP3 regulated calcium channels (van den Heuvel et al., 2014). Similar to other structures, the maturation of calcium handling occurs during the neonatal stage and therefore increases with development (Ziman et al., 2010).

Dissimilarities in energy production are prevalent between native CMs and PSC-CMs (Figure 2B). Adult CMs have a mature network of ovular-shaped mitochondria that are approximately 35% of the cellular volume and are aligned in the direction of the sarcomeres to provide ample ATP for contraction (Kim et al., 1994; García-Pérez et al., 2008; Piquereau et al., 2010, 2013). PSC-CMs have immature mitochondria that are disorganized around the perinuclear region, present in smaller amounts and sizes (St. John et al., 2005; Robertson et al., 2013; Keung et al., 2014). Cristae, which are folds that give mitochondria a significant surface area for efficient cellular respiration, are absent within PSC-CMs but densely packed within adult CMs (Porter et al., 2011; Feric and Radisic, 2016). In adult CMs, the majority of energy production occurs by oxidative metabolism while PSC-CMs use mostly glycolysis, which is less efficient (Lopaschuk and Jaswal, 2010; Rana et al., 2012; Keung et al., 2014).

From a developmental standpoint, maturation of the metabolism occurs after birth when cardiomyocytes are exposed to higher demands of energy as well as oxygen and fatty acids (Lopaschuk and Jaswal, 2010; Keung et al., 2014). Fetal CMs contain small, cristae absent, rounded mitochondria that are not localized adjacent to sarcomeres and rely on glycolysis for energy production (Porter et al., 2011). After birth, the metabolic stress causes neonatal CMs within 7 days to switch to mostly oxidative metabolism with an increase in mitochondria size and amount, development of cristae, and a more ovular-shaped mitochondria morphology (Figure 2A; Yatscoff et al., 2008; Kreipke et al., 2016). Besides the role in energy production, metabolism regulates cellular processes. Therefore, the switch to a high metabolic stress may be one of the driving factors of the significant maturation that occurs during the neonatal stage (Yanes et al., 2010; Folmes et al., 2011).

The CM developmental stages show drastic differences in their ability to proliferate. Early fetal CMs are highly proliferative but start to decrease rapidly into the late fetal CM stage. Postnatally, neonatal CMs have a window of proliferative ability that lasts for about 7 days before their cell cycle machinery is silenced permanently leading to non-proliferative adult CMs (Ikenishi et al., 2012). PSC-CMs overall are mostly capable of proliferating (Feric and Radisic, 2016). Surprisingly, neonatal mouse hearts can regenerate following a partial tissue removal but this ability diminishes after the first 7 days post birth (Porrello et al., 2011). Studies have shown that oxygen exposure reduces neonatal CM proliferative ability and therefore the change to a mostly aerobic environment post birth could be partially responsible for this shift (Puente et al., 2014). Other researchers hypothesize that metabolic stress similarly plays a key role in this proliferation change (Yutzey, 2014). Overall, however, the permanent cell cycle arrest seems to be necessary for CM maturation and thus heart function (Takeuchi, 2014). As evidence, rare cases of mitogenic cardiomyopathy in human infants have been reported where the neonatal CMs have atypical proliferative ability but an immature phenotype that results in lethality due to cardiac dysfunction (Chang et al., 2010; Shenje et al., 2014).

In summary, a large divide in native CM development vs. PSC-CM development is apparent (Figure 3). Human PSC-CMs within three-dimensional culture systems have ultimately failed to obtain all characteristics of late fetal CM maturity, even after improving their maturity through natural engineering approaches. Most achievements in PSC-CM maturity occur at a single cell level while syncytial units of uniformly more mature PSC-CMs have not been developed well in three-dimensional engineered tissues. As evidence, unlike late fetal CMs, no human PSC-CMs have been observed to have early indentations of developing transverse tubules (Satin et al., 2008; Ziman et al., 2010). The majority of sarcomeres within late PSC-CM tissue constructs likewise do not have almost completely organized sarcomeres with Z-disks, A bands, I bands, and H zones (Chacko, 1976; Snir et al., 2003). KCNJ2 expression in PSC-CMs also still remains mostly lower than late fetal CMs (Amin et al., 2010; Liu et al., 2016). In addition, when native CMs are isolated and placed within in vitro culture they have been observed to go backwards in maturity (Nguyen et al., 2015). When more mature native CMs are cultured in vitro, they can mature more than PSC-CMs, such as forming fully developed transverse tubules (Bian et al., 2014a). Therefore, more emphasis needs to be placed on understanding the natural fetal stages of CM development. Once researchers understand how to more accurately recapitulate the fetal stages of CM development, further progress in PSC-CM maturity can be made toward the neonatal stage and thus eventually to the adult mature phenotype.

Mechanical stimuli affect all developmental stages of the heart from early embryogenesis to post-natal maturation (Takahashi et al., 2013). As a whole organ, the heart's main function is a mechanical pump that consistently responds to mechanical stimuli throughout its entire lifetime (Jacot et al., 2010; Takahashi et al., 2013). The influence of mechanical stimuli on the heart became apparent over 100 years ago with the establishment of the Frank-Starling Law, which indicates that as the heart tissue stretches from filling of blood, the heart's contractile force also increases (Konhilas et al., 2002). Studies on zebrafish embryos have shown that a lack of proper mechanical load can lead to abnormal heart development, which provides further evidence for the importance of mechanical stimuli during development (Hove et al., 2003).

One important concept of the heart involves the mechanoelectric feedback (MEF), where there exists an interplay between mechanical input and electrical output. MEF regulates the beating rate as well as excitation-contraction coupling within cardiomyocytes (Kohl and Ravens, 2003). This feedback mechanism involves the many mechanosensitive ion channels that respond to mechanical stimuli (Hu and Sachs, 1997; Takahashi et al., 2013). Studies on embryonic chick heart development have shown that hemodynamic loads directly influence maturation of the heart's conduction system (Figure 4; Reckova et al., 2003). Mechanical stretch leads to an increase in connexin-43, and thus an increase in the overall conduction velocity (Sheehy et al., 2012). Therefore, mechanical stimuli have a direct influence on the electrophysiological maturation of CMs.

Cardiomyocytes experience three main types of constant mechanical stimuli that influence development: stretch during heart filling, hemodynamic loads from blood flow, and active self-induced cytoskeletal contractions during heart ejection (Taber, 2001). As the developmental stages of CMs progresses, more mechanical stimuli play an influential role. Early fetal CMs begin to contract spontaneously and asynchronously in the early heart tube before coordinated blood flow and the sarcomeres completely develop (Lindsey et al., 2014; Geach et al., 2015). Although the reasoning behind this early spontaneous beating is still being investigated, studies have shown that the early contractions cause hemodynamic forces to develop, which in turn impact subsequent heart development (Granados-Riveron and Brook, 2012). A study on frog heart development showed that early CM contractions drive sarcomerogenesis (Geach et al., 2015). Therefore, the contractions may act as a feedback loop to continue early fetal CM development within the heart tube to the next developmental stage.

The cross-bridge cycles of the early fetal CM's cytoskeleton are known to drive morphogenesis of the cardiac looping phase by causing the heart tube to bend (Manning and McLachlan, 1990; Taber, 2001). In contrast, extrinsic mechanical stimuli, such as an increase in hemodynamic force, are what cause the cardiac loop to twist and rotate (Wyczalkowski et al., 2012). During the late fetal CM stage of heart chamber formation involving trabeculation and compaction of the myocardium, strain begins to play a larger role (Figure 4). A strain gradient, which varies across the myocardial wall with higher amounts of strain in the inner trabeculated layer, may influence late fetal CM maturation (Damon et al., 2009). The outer compact layer contains late fetal CMs that have a greater mitotic activity than late fetal CMs found within the trabeculated layer. Therefore, an increase in strain causes a reduction in proliferation rate and thus may drive CM developmental maturity (Sedmera and Thompson, 2011). Other studies on the impact of hemodynamic forces have shown that an increase in hemodynamic blood flow causes ventricular CMs to become more elongated and rod-shaped (Auman et al., 2007). Compaction increases during the late cardiac looping and chamber formation stages of heart development due to the change in hemodynamic flow from laminar to vortexed flow (Figure 4; Samsa et al., 2013).

As the demand for more blood increases throughout development, an increase in hemodynamic loading on cardiomyocytes occurs. CMs become more mature to meet these higher workload demands, especially during the neonatal to adult stages (Figure 4; Taber, 2001; Zhu et al., 2014). As a result, cardiac output, which has only been measured during the four chamber heart stage, increases exponentially throughout development with values of 120 mL/min at 15–20 weeks gestation, 1,700 mL/min around the neonatal stage, and around 5,000 mL/min in adulthood (Figure 4; De Smedt et al., 1987; Matthews et al., 2012). Human heart rate varies greatly during development. From the early heart tube stage to the four chambered heart stage, heart rate progressively increases to about 170 bpm, which then declines from the late fetal stage into the neonatal stage to a value around 130 bpm (Figure 4; Doubilet and Benson, 1995; Hornberger and Sahn, 2007; Lindsey et al., 2014). From the neonatal to adult stage, heart rate continues to decrease until reaching about 60 bpm (Nunes et al., 2013b).

Besides mechanosensitive ion channels, several other protein structures are involved in mechanotransduction pathways of cardiomyocytes. These protein structures include the cytoskeleton, focal adhesions, integrins, rho kinase, integrin-linked kinase, and extracellular signal-related kinase (Takahashi et al., 2013). Mechanical stimuli causes rho/rock activation which regulates several cellular processes, such as hypertrophy, proliferation, differentiation, and maturation (Loirand et al., 2006). Thus, mechanical stimuli are crucial to the overall natural developmental program of CMs.

Electrical conduction is also a vital part of cardiac functionality. Studies on zebrafish hearts have shown that impaired cardiac conduction leads to abnormal cardiac remodeling and development (Chi et al., 2010). Development of the cardiac electrical conduction system has been readily studied in chick model organisms. During the early tubular heart stage, a slow and linear electrical conduction is established. Therefore, CMs undergo a constant electrical stimuli from the early onset of CM specification. As the heart tube undergoes the cardiac looping phase, the conduction velocity increases but the conduction is still in the blood flow direction. As the myocardial trabeculations form, temporary conduction pathways develop in the anterior and posterior. These temporary conduction pathways are present only before the development of the His-Purkinje fiber network, which denotes a mature conduction system. The trabeculations are also involved in the early conduction system of the heart since they can electrically conduct well. As the four chambers of the heart form, the direction of conduction from base to apex shifts to a mature apex to base path. This shift occurs when the His-Purkinje fibers, composed of specialized cardiac conducting cells, have fully formed underneath the endothelial layer, which allows coordinated contractions from the apex that push blood out of the ventricles (Sedmera et al., 2004). Therefore, the late fetal CMs experience more coordinated electrical impulses that further drive their structural and functional maturity. The functional maturity of the electrical conduction system is complete by the end of embryonic cardiac development. However, further maturation of CM electrophysiology occurs after birth. As transverse tubules develop during the neonatal stage, the excitation-contraction coupling of CMs mature, which allow action potentials to reach deep within the CMs (Ziman et al., 2010).

Electrical stimuli directly impact the electrophysiological function of CMs. Gap junctions, which regulate the propagation of electrical signals between neighboring CMs are critical to the electrical conduction (Angst et al., 1997). Several studies using in vivo connexin protein knockouts have shown to lead to CM electrical conduction defects, such as arrhythmias (Britz-Cunningham et al., 1995; Simon et al., 1998). The ion channels are also critical structures. Mutations in ion channels can lead to pathological disease states, such as mutations in voltage gated potassium channels, which cause ventricular tachycardia (Xiong et al., 2015). As CMs undergo more coordinated and efficient electrical stimuli with the development of the mature cardiac conduction system, the CMs respond to their environment by upregulating electrophysiological structures. Studies on electrical stimuli of CMs have shown an increase in cell size, alignment, connexin-43 expression, ion channel expression, and sarcomere organization (Baumgartner et al., 2014). Overall, efficient propagation of electrical signals is a crucial aspect of CM natural developmental program and functionality.

The extracellular matrix (ECM) plays a dynamic role in the natural development of each CM stage as it provides structural support, modulates many cellular processes, and is constantly being remodeled in response to various stimuli (Lockhart et al., 2011). Since the ECM helps withstand the continuous mechanical stimuli associated with the never ending cardiac cycles throughout an organisms lifetime, it is a crucial component of the heart (Mishra et al., 2013). Deviations from the normal ECM can cause several cardiac formation abnormalities as shown in studies on in vivo mouse heart development (Lockhart et al., 2011). Congestive heart failure occurs due to adverse ECM remodeling following a myocardial infarction and therefore undesirable changes in the ECM are linked to several disease states (Konstam et al., 2011). The ECM is also associated with several mechanotransduction pathways through integrin-ECM connections that drive cellular processes (Corda et al., 2000). In animals that can regenerate their myocardium, such as newts, the ECM deposition in a trabecular network fashion is a critical first step of cardiac regeneration (Piatkowski et al., 2013).

During cardiac development, the ECM drastically changes its composition and organization to lead to the formation of the complex four chambered heart. The most significant changes occur during the neonatal stage when the ECM is rapidly remodeled to resemble the adult myocardium composition within 2–3 weeks post birth. From the neonatal and adult stages, the amount of ECM deposition increases (Sullivan and Black, 2013). The most functionally relevant cardiac ECM proteins are collagen type I and III for structural support, basement membrane collagen type IV for facilitating cellular alignment, fibronectin which connects integrins to other ECM proteins, and elastin to provide elastic properties (Hanson et al., 2013). Collagen proteins increase in structural organization and amount as the heart develops to meet the higher work load demands (Gershlak et al., 2013). No distinct structural collagen fibers are found during the early fetal stage. However, as development progresses, collagen fiber networks can be found in late fetal CMs, which increase organization and density even more during the neonatal stage (Hanson et al., 2013). Fibronectin is essential to heart development. If absent within mouse embryos, inadequate heart tube formation occurs (George et al., 1993). The role of fibronectin in early cardiac morphogenesis has been known for almost 30 years. If fibronectin is blocked, cardiac precursor cells cannot migrate appropriately (Linask and Lash, 1988). Similar to collagen, the organization of fibronectin increases significantly during the neonatal stage but the amount of fibronectin is higher in both fetal stages (Sheehy et al., 2012; Hanson et al., 2013). Basement membrane proteins, such as collagen type IV, laminin, glycoproteins and proteoglycans likewise can be found in all stages but increases in organization and density during the neonatal stage and surrounds CMs of the adult stage (Hanson et al., 2013; Yang et al., 2015). Studies have shown that laminin organization facilitates sarcomere organization throughout the neonatal stage of CMs (Yang et al., 2015). Elastin remains mostly constant during the stages of CM development (Hanson et al., 2013).

Other proteins are found throughout the ECM and impact CM development. Hyaluronan, a GAG protein, is known to facilitate proliferation and is involved in cardiac morphogenesis (Baldwin et al., 1994; Toole, 2001). Perlecan and versican, which are proteoglycans, are involved in cardiac morphogenesis since disruption of these proteoglycans causes cardiac formation abnormalities (Henderson and Copp, 1998; Costell et al., 2002; Lockhart et al., 2011). Matrix metalloproteases regulate ECM remodeling and modulate CM contraction and calcium handling (Mishra et al., 2013).

Besides the overall ECM composition, stiffness is a critical factor that influences CM behavior. ECM stiffness has many effects on CMs, such as contractility, calcium handling, and cytoskeletal structure (Takahashi et al., 2013). A study on chick heart development demonstrated that stiffness differences along the heart tube facilitated the heart looping process (Zamir et al., 2003). During the neonatal CM stage after birth, stiffness of the heart wall is known to significantly increase between 2 and 3 fold and may be due to the increase in the neonatal ECM organization (Jacot et al., 2010; Gershlak et al., 2013). Understanding how the ECM interacts with CMs in vivo will enable researchers the ability to naturally engineer heart tissue that more closely mimics physiological conditions for enhanced maturity.

Non-cardiomyocytes readily interact with cardiomyocytes and influence cardiac development through a variety of different mechanisms. From a developmental perspective, non-cardiomyocytes are required for the specification of the cardiac lineage. Early on during development, the anterior endoderm induces the specification of the cardiac mesoderm, by FGF8 signaling, which differentiate into all myocardial cell types (Schultheiss et al., 1995; Alsan and Schultheiss, 2002). Several types of cells reside within the ventricular heart wall including cardiomyocytes, cardiac fibroblasts, vascular smooth muscle cells, and endothelial cells. CMs directly interact with endothelial cells derived from the endocardium, as well as cardiac fibroblasts derived from the epicardium (Brutsaert, 2003; Sullivan and Black, 2013). The only stage of CMs that has minimal interactions with non-CMs are the early fetal CMs that reside within the heart tube and early looping stage.

The importance of in vitro mechanical and electrical stimulation on CM maturation have been known for over a decade (Zimmermann et al., 2000; Radisic et al., 2004). Many studies have developed bioreactor systems that subject CMs seeded in a three-dimensional matrix to create engineered heart tissue (EHT) to these stimuli. However, many systems do not employ a natural engineering program since they use uninspired, non-physiological values. Therefore, this review discusses some recent approaches that are natural developmentally inspired to improve overall CM maturation. Table 1 summarizes this information. Some recent thorough reviews on mechanical and electrical stimulation of CMs can be found elsewhere (Rangarajan et al., 2014; Cao et al., 2015; Stoppel et al., 2016).

The studies that were discussed in the electrical and mechanical stimulation sections utilized more basic ECM compositions for their scaffolds. Although these ECM components provided a 3D microenvironment for the CMs, the ECM of these EHTs do not effectively recapitulate many aspects of the natural cardiac ECM that are known to drive the natural developmental program of CMs. As shown in studies that modulated the topographical alignment, stiffness, and/or the overall material composition of the scaffolds being used, the ECM actively influences CM maturation in vitro. Table 2 summarizes this section.

Due to the essential role of non-cardiomyocytes during normal physiological development and function of the heart wall, various researchers have investigated the use of non-cardiomyocytes within in vitro co-cultures to drive cardiomyocyte maturation. To create a more robust, physiologically-relevant cardiac construct, other cell types are necessary, such as cardiac fibroblasts and endothelial cells. As a result, researchers are moving away from CM-only EHTs and are beginning to investigate cardiomyocytes with other cell types.

Many studies with neonatal or fetal CMs co-cultured with fibroblasts and/or endothelial cells functionally improved and matured the CMs (Narmoneva et al., 2004; Radisic et al., 2008; Garzoni et al., 2009; Iyer et al., 2012; Vuorenpää et al., 2014). One particular interesting study involved co-culturing embryonic quail EPDCs with neonatal mouse CMs in vitro. When the EPDCs were directly connected to the CMs, the proliferation, connexin-43 expression, N-cadherin expression, cellular alignment, and contractility all significantly increased for CMs (Weeke-Klimp et al., 2010).

Kim et al. was one of the first groups to conduct co-culture experiments of non-cardiomyocytes with human PSC-CMs. This group cultured human ESC-CM spheroids with smaller clusters of embryonic non-cardiomyocyte cells, which contained the majority of the heart cell types. These interactions led to functional improvements in ion channel profiles and electrophysiology with an increase in the action potential amplitude, and more negative resting membrane potential (Kim et al., 2009). Another co-culture system involved the use of an endodermal cell line directly connected to mouse and human pluripotent stem cells, which led to the differentiation of the cells into cardiac progenitors that expressed the cardiac specific genes of NKX2.5 and Isl1 (Uosaki et al., 2012).

Several studies have cultured PSC-CMs with fibroblast cells. Thavandiran et al. generated EHTs composed of human ESC-CMs and human ESC-CFs in a collagen/matrigel construct that were also electrically stimulated. They demonstrated that the optimal CF to CM ratio was 3:1, which most significantly increased the CM maturity. This optimized co-culture ratio resulted in increases in cardiac specific genes, such as MHC and MLC isoforms, increased conduction velocity, enhanced overall tissue integrity, greater sarcomere organization, better cellular alignment, and promotion of beneficial collagen matrix remodeling (Thavandiran et al., 2013). Improvements in differentiation efficiency can be achieved by CM co-cultures. One experiment involved mouse ESCs in a collagen gel indirectly co-cultured with CFs that were placed into a well insert that sat above the mouse ESCs. This 3D indirect co-culture system led to greater CM differentiation efficiency and improved sarcomere organization, and connexin-43, alpha-actinin, and cardiac troponin expression (Ou et al., 2011).

The combination of more than one cell type with PSC-CMs further improves the EHTs maturity. Human IPS cells were differentiated into endothelial cells, cardiomyocytes, and vascular smooth muscle progenitors and seeded within a collagen/matrigel 3D construct to assess the multi-culture systems functional maturity. The combination of all three cell types improved the EHT's contractile force, cellular alignment, and sarcomere organization in vitro. When implanted in vivo on infarcted rat hearts, vasculature was present within the EHTs that originated from the host as well as the construct itself and some functional heart improvements occurred (Masumoto et al., 2016). Roberts et al. engineered large scale vascularized EHTs containing human ESC-CMs co-cultured with human dermal fibroblasts in a high density collagen matrix that contained an engineered microvascular network seeded with human umbilical vein endothelial cells. Developing EHTs with patent microvascular networks and ample CM-CM connections for proper electrophysiological function has proven to be difficult. Low density collagen scaffolds allow for CMs to electrically couple but do not provide enough structural support to the microvascular network, which become occluded by CM contractions. In contrast, high density collagen scaffolds provide structural support for the vasculature but the CMs are unable to properly electrically couple. In this study, the addition of the fibroblasts increased cellular alignment, provided the microvascular network with structural support for patency up to 2 weeks, and allowed CM electrical coupling within a high density collagen matrix (Roberts et al., 2016). These results illustrate the significance of naturally engineering EHTs with non-cardiomyocytes that provide synergistic benefits to the functional maturity.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.