Endothelial cells (ECs) form the lining of blood vessels and a specific barrier structure between surrounding tissue and blood. ECs respond to blood flow derived shear stress changes through release of vasodilating and vasoconstricting factors such as nitric oxide (NO) or endothelin to communicate with surrounding stromal cells, such as smooth muscle cells and fibroblasts, and to adjust vessel tone (Chatterjee and Fisher, 2014). ECs are known to differ among tissues and even within the same tissue, such as heart, ECs comprise a heterogenous cell population due to differing regional needs and vessel size (Zhang et al., 2005; Litviňuková et al., 2020). In MPS development, tissue-specific ECs should be preferred, and mechanical forces implemented in the modeling platform to replicate the vascular in vivo environment more closely.

ECs of the coronary vessels form vascular network throughout the heart myocardium and have a different origin compared to ECs of the outer endocardium (Mikawa and Fischman, 1992). Already from early stage in development, interaction between CMs and ECs is necessary for proper growth and development of heart. Communication between CMs and ECs is processed through gap junction proteins, such as connexin 43 (Cx43), that links electromechanical processes in the myocardium with the surrounding vasculature (Givvimani et al., 2011). The main growth factor that regulates ECs is VEGF that binds to and activates VEGF receptor 2 in ECs (Olsson et al., 2006). VEGF mediated signaling induces angiogenesis, the growth of new blood vessels from the pre-existing ones, e.g., in ischemic areas or in a growing heart (Olsson et al., 2006).

In cardiac MPS, the benefit in integrating ECs can be detected with increased secretion of angiogenic growth factors and improved vessel forming capacity. Narmoneva et al. (2004) were the first ones to show the importance of EC-CM interactions in 3D environment in vitro. Their results demonstrated increased survival and spatial reorganization of CMs in the co-culture mediated partly via Cx43 gap junctions. The results of this study suggested an extended role of cardiac endothelium not only in the delivery of blood and oxygen but also in the formation and maintenance of structure in myocardium (Narmoneva et al., 2004). Sekine et al. (2008) showed that cardiac cell sheets with ECs secrete significantly more VEGF, bFGF, and hepatocyte growth factors (HGF) in comparison to the cardiac sheets without ECs. They also studied co-culture of fibroblasts and ECs and did not detect increased secretion of any of the angiogenic factors suggesting a specific interaction between CMs and ECs (Sekine et al., 2008). Endothelial-cardiomyocyte co-cultures have been established also in MPS where ECs are typically seeded to coat microfluidic channels in order to mimic blood flow (Ellis et al., 2017). In MPS, ECs experience perfusion derived shear stress, align according to the fluidic flow (Ellis et al., 2017). Importantly, they can be used to establish intra- and extra-vascular interactions with the surrounding cells and ECM with increased in vivo vascular resemblance compared to static cultures.

Beside ECs, fibroblasts are the other major cell type present throughout the myocardium and interacting with CMs (Brutsaert, 2003). In the heart, fibroblasts are the primary cell type responsible for synthesis and degradation of ECM, and in providing mechanical support for the contracting CMs (Baxter et al., 2008). Additionally, cardiac fibroblasts play a crucial role in wound healing and repair, produce growth factors and other signaling molecules that directly regulate CM function (Brutsaert, 2003; Baxter et al., 2008; Ieda et al., 2009; Howard and Baudino, 2014).

In MPS design, developmental stage of the used fibroblasts should be considered since, e.g., embryonic fibroblasts have been shown to promote proliferation of CMs whereas adult fibroblasts promote CM hypertrophy, i.e., increased size of CM (Ieda et al., 2009). After cardiac tissue injury, due to, e.g., myocardial infarction, fibroblasts are the main cell type contributing to fibrotic scar formation through ECM production. Fibroblasts from the infarcted region present phenotype with increased proliferative and collagen synthesis capacity (Squires et al., 2005). In development of MPS disease models, fibroblasts should be considered as an essential cell type for mimicking the in vivo situation of cardiac injury.

Although cardiac fibroblasts are considered as non-excitable cells, they have been shown to modulate electrical properties of CMs. Kofron et al. (2017) showed that cardiac fibroblast activation altered AP and Ca²⁺ transients of CMs and lead to proarrhythmic activity of CMs (Kofron et al., 2017). Through gap junction protein Cx43 and Cx45 connections, fibroblasts and CMs are able to communicate with one another electrically. Furthermore, by fibroblast coupling, fibroblasts may form bridges and allow CMs in different cardiac locations to communicate with one another (Kamkin et al., 2005).

Several studies have demonstrated that fibroblasts improve hiPSC-CM viability and structural and functional maturity (Baudino et al., 2008; Tiburcy et al., 2017; Giacomelli et al., 2020) in 3D environment that more closely mimics the native myocardium (Beauchamp et al., 2020; de Lange et al., 2021). hiPSC-CMs in 3D microtissues with cardiac fibroblasts showed improved sarcomeric structures with T-tubules, enhanced contractility, and mitochondrial respiration and were electrophysiologically more mature compared to microtissues lacking fibroblasts (Giacomelli et al., 2020). In addition to fibroblasts, human mesenchymal stem/stromal cells (MSCs) have been shown to improve hiPSC-CM structural maturation, contractility and electrophysiological development when co-cultured in vitro or co-transplanted in vivo (Yoshida et al., 2018). Bone marrow-derived MSCs secreted soluble factors, including VEGF, bFGF, stromal cell-derived factor 1, granulocyte-macrophage colony-stimulating factor growth factors and exosomes, enhanced hiPSC-CM mitochondrial energetics, performance and reduced reactive oxygen species production by CMs under stress (Yoshida et al., 2018).

Beside improving CM maturation, fibroblast have been shown to support formation of vascular network in 2D (Vuorenpää et al., 2014) and in 3D cardiac in vitro models (Caspi et al., 2007). Similar contribution to angiogenesis has been detected with MSCs when compared to EC monocultures in vascularization studies (Sato et al., 2011; Sarkanen et al., 2012; Jeon et al., 2014; Mykuliak et al., 2022). In the next chapter, we will review the current methods to vascularize cardiac in vitro models.

It is estimated that humans have approximately 100,000 km of blood vessels networking throughout the body thus making vascular endothelium the largest endocrine organ (Talman and Kivelä, 2018). Blood vessels support heart with oxygen, nutrients and signaling molecules and interact with the surrounding tissue to maintain homeostasis. The continuous activity of heart creates a large demand for oxygen and nutrient delivery to cardiac tissue and, therefore, one capillary resides next to every cardiomyocyte in normal adult myocardium (Brutsaert, 2003). Function as selective barrier and as an interface between blood and the surrounding tissue is critical in several (patho-)physiological processes (Talman and Kivelä, 2018; Browne et al., 2021).

In EHTs, hydrogel mass transport properties can be adjusted to replace vascular functions, but only to a certain limit. The mesh size of hydrogels is typically in the nanometer scale and pore size in the micrometer scale (Li and Mooney, 2016). Both affect how molecules are allowed to move inside these water filled regions in the scaffolds (Karvinen et al., 2019). In addition, the solute molecules may be bound by electrical charges on the polymer chains (Amsden, 1998). Diffusion properties and the need for vascularization becomes important when the diffusion distances (Soto et al., 2016) and cell densities increase (Figueiredo et al., 2018).

Vascularization has been one of the main targets in development of MPS and can be considered in two level; vascular network formation inside individual tissue model and circulation between tissue models to establish connected MPS. Both means of vascularization are essential in mimicking tissue level functions, and in increasing physiological relevance and accuracy of MPS. Vascularization inside tissue model (1) enables to overcome the growth limit (∼200 µm) inside a tissue model dependent on diffusion of oxygen and nutrients (Browne et al., 2021); (2) improves target cell differentiation and maturation (Caspi et al., 2007; Vuorenpää et al., 2014; Vuorenpää et al., 2017) and (3) enables 3D structure, mechanical support and growth factor related stimulation for other cell types (Sarkanen et al., 2012; Vuorenpää et al., 2014). Connective circulation between tissue models is important in delivering nutrients, oxygen, signaling molecules and test substances such as drugs between tissue models. It enables (1) long term viability of the cells; (2) metabolic activities; (3) tissue-tissue signaling; (4) implementation of mechanical forces such as shear stress and interstitial flow, and, importantly, (5) environmental control.

There are several methods to vascularize MPS (Figure 3B). ECs can be cultured on designed patterns or engineered interfaces using biofabrication techniques such as sacrificial molds (Paek et al., 2019; Kim et al., 2021) or tunnels made with laser-based cavitation molding (Enrico et al., 2022). Additionally, ECs are used to coat or endothelialize surfaces (Paek et al., 2019; Zhang et al., 2020) or induced to self-assemble into microvasculature (Kim et al., 2013; Jeon et al., 2014; Mykuliak et al., 2022). Ability to form perfusable vascular structures with intact vessel wall and open lumen increases the physiological relevance of the vasculature and enables, therefore, more complex applications such as immune cell extravasation studies (Kim et al., 2012; Gopalakrishnan et al., 2015; Wu et al., 2015; Marzagalli et al., 2022). In establishing perfusable vascular model in reproducible manner, EC source and passage are shown to be critical. We and others have shown (Liu Y. et al., 2020; Mykuliak et al., 2022) that vasculogenesis, de novo production of blood vessels, is critically impaired when EC passaging exceeds 7–8. Moreover, vasculogenesis process leading to perfusable microvasculature is not dependent only on the culture time but also on the cell source. Human umbilical vein endothelial cells are still considered as a gold standard in vascularization due their robustness, easy availability andfor their ability to adapt in different tissue engineering applications (Vuorenpää et al., 2014; Liu Y. et al., 2020; Calejo et al., 2020). However, iPSCs provide an option to receive not only personalized or patient-specific ECs but also ECs with tissue specific characteristics for MPS development (Browne et al., 2021).

Although EC monocultures form vascular structures with lumen in vitro, supporting pericytes significantly contribute to angiogenesis and improve vascular maturation (Sarkanen et al., 2012; Jeon et al., 2014; Liu Y. et al., 2020; Mykuliak et al., 2022). The importance of pericytes in physiological blood vessel formation has been shown in several cardiac applications (Caspi et al., 2007; Vuorenpää et al., 2014; Vuorenpää et al., 2017) and, more recently, in pathological conditions such as cardiac ischemia (O’Farrell et al., 2017) and atherosclerosis (Zhang et al., 2020). Different types of supporting cells such as fibroblasts, smooth muscle cells and MSCs have been used in co-culture with ECs in MPS (Jeon et al., 2014; Fernandez et al., 2016; Mykuliak et al., 2022) and in cardiac in vitro models (Caspi et al., 2007; Vuorenpää et al., 2014; Vuorenpää et al., 2017). Caspi et al. (2007) were the first ones to show the generation of vascularized 3D human cardiac in vitro model based on co-culture of hESC-CM, EC and fibroblasts. Importantly, they showed that vascularization resulted in upregulation of early and late markers of CM differentiation and maturation (Caspi et al., 2007). Also in regenerative medicine, the benefit in pre-vascularizing cardiac tissues instead of transplanting single CMs has been widely acknowledged (Sekine et al., 2008; Stevens et al., 2009; Masumoto et al., 2016; Ishigami et al., 2018).

Earlier, we reported 2D cardiac construct with ECs, fibroblasts and hiPSC-derived CMs showing aligned vascular network with CMs and an increased expression of cardiac structural proteins compared to CM monoculture (Vuorenpää et al., 2017). Also an elegant study by King et al. (2022) highlights the role of non-CMs as fundamental regulators of iPSC-CM electrophysiology by demonstrating that regulation of beating rate and Ca²⁺ transients were lost when cells were physically separated. In heart-on-a-chip, they were able to show establishment of perfusable lumen with intra-vascular red blood cell transport and pulsatile red blood cell velocity according to CM contractile cycle (King et al., 2022). For MPS development in general, similar physiological vascular flow utilizing CM contractile function as pump mechanism may be beneficial.

Tissues are mostly consisted of other cell types than neuronal cells. Still, their important functions and behavior patterns are controlled via nerve stimulation (Park et al., 2020). In addition to CM, fibroblasts and endothelial cells, also network of neuronal cells is important part of the human heart (Figure 3A) (Zaglia and Mongillo, 2017). The heart is a densely innervated organ. The high number of neurons in the heart has been recognized, but extent of the innervation density has been underestimated (Zaglia and Mongillo, 2017). In general, the innervation has crucial role in tissue development as well as in their functional control, modulation, and regeneration and therefore would be important part to consider including in the MPSs. Despite its importance, innervation for the functionality of the tissue has been generally overlooked in many non-neuronal applications (Das et al., 2020). Only a few cell types have been studied in vitro with integrated innervation, such as mesenchymal stem cells, muscle cells, retinal pigment epithelial cells, human adipose stem cell derived corneal keratocytes, and heart cells (Morimoto et al., 2013; Silva et al., 2017; Yeste et al., 2017; Mörö et al., 2022).

Cardiac nervous system (NS) innervate myocardium and cardiac conduction system (Habecker et al., 2016). It is comprised of sympathetic, parasympathetic and sensory neurons, and is regulated by adrenergic and cholinergic signaling via neuromodulator molecules norepinephrine and acetylcholine (Elia and Fossati, 2023). All cardiac neurons are originally developed from neural crest cells (Tampakakis and Mahmoud, 2021) and each of these neuronal cell populations has certain phenotype and distribution which is crucial for their functions (Habecker et al., 2016). In humans, the cell bodies of the cardiac preganglionic neurons are located in grey matter of the thoracic spinal cord. Only few postganglionic neuron cell bodies have been found in the myocardium (Coote and Chauhan, 2016).

The cardiac functions are regulated by sympathetic and parasympathetic neurons. During development, sympathetic neurons and CMs undergo co-maturation process meaning that signals from cardiac tissue regulate properties such as growth and patterning of the sympathetic neurons whereas maturation of the CMs is influenced by signaling of the nerves (Habecker et al., 2016). In addition, sympathetic neurons play important role in postnatal maturation of the CMs (Tampakakis and Mahmoud, 2021). The sympathetic neurons that predominantly innervate heart are located in stellate ganglion adjacent to the spinal cord segments T1-T4. When they enter through epicardium to the heart they extend and branch out to the myocardium (Odnoshivkina and Petrov, 2021). The sympathetic nervous system has important role in maintaining homeostasis in healthy and diseased state of the heart. The main role of sympathetic neurons is to increase heart rate, cardiac contractility and conduction velocity as well as regulate blood pressure. Parasympathetic neurons act in an opposite manner, they decrease heart rate, contractility and conduction velocity (Das et al., 2020). Sensory nerves are responsible for pain perception and starting protective cardiovascular response in ischemia (Kimura et al., 2012).

Unlike neuromuscular junctions, synaptic junctions between neurons and cardiac tissue are not that well understood. It has been shown that protein complexes are involved in cellular interactions in case of sympathetic neurons and CMs or in exocytosis. These protein complexes include post-synaptic β-adenoreceptors in CM cell membrane (Shcherbakova et al., 2007). However, it is known that crosstalk between the sympathetic neurons and CMs occur via contact of varicosities, clusters of extensions filled with neurotransmitters located in the nerve endings, and CMs (Odnoshivkina and Petrov, 2021).The dysregulation and abnormalities in cardiac innervation cause impairment in the cardiac functionality and cardiac nerve fiber abnormalities, and are thus linked to the life threatening arrhythmias, congestive heart failure and myocardial infarction (Bairey Merz et al., 2015; Elia and Fossati, 2023). In diseased state, change in oxidative stress and expression of biomolecules (growth factors and cytokines) in heart and vasculature leads to changes in phenotype and morphology in cardiac nerves (Habecker et al., 2016).

Examples of the previously reported methods to innervate cardiomyocytes are collected to the Table 1. Major of the applications are still conducted in 2D platforms, either as simple co-culture of mixed cells (Winbo et al., 2020) or in microfluidic devices which support axon-mediated innervation (Takeuchi et al., 2011; Takeuchi et al., 2013; Oiwa et al., 2016; Takayama and Kida, 2016; Sakai et al., 2017; Häkli et al., 2022a; Bernardin et al., 2022). There are still only a few reports describing cardiac innervation performed in 3D environment (Soucy et al., 2020).

The significant benefit of the microfluidic devices for innervation applications is the restriction of the cell somas from the target cells or tissues with compartmentalized structures and microtunnels (Figure 3B). These structures allow assessment of the axon-mediated innervation and mimics in vivo conditions of central nervous system and peripheral nervous system better (Neto et al., 2016; Park et al., 2020; Pelkonen et al., 2020; Ristola et al., 2021; Tong et al., 2021; Häkli et al., 2022a). In addition, in vitro neuronal network models have shown to share many similarities with in vivo networks. This aspect increases the relevance of the model for preclinical tests and drug screening.

Currently, the importance of innervation for the functionality and development of other tissue structures has still mostly been shown in co-cultures with only one cell type, or in simple multicellular cultures, not yet in more complex MPS systems. While studies combining CMs and neuronal cells are already concentrating on investigating physiological conditions with hiPSC-CMs, still mostly animal derived neuronal cells are used. However, completely human based models are needed for neuro-cardiac research for better translation of human physiology. Current trend goes towards more clinically relevant cell types, better in vivo conditions mimicking in vitro models and MPS and towards 3D environment in the future.

2D cell culture methods do not allow the cells to be in natural contact with other cells and the ECM. For better biomimicking, cell cultures adding a spatial dimension is necessary. Different 3D culturing approaches have shown success with hiPSC-CMs. It has been previously demonstrated already with 2D island cultures that dimensionality affects hiPSC-CM maturation (Werley et al., 2017). In comparison to a planar culture, a 3D culture allows even more of cell-cell and cell-matrix contact. To this date many hiPSC-CM 3D cell cultures have been developed (Andrysiak et al., 2021).

Substrate stiffness and topography can greatly affect hiPSC-CM maturation. Soft substrates close to the physiological range of myocardium are often reported to promote structural and functional maturation in comparison to rigid surfaces (Körner et al., 2021; Pioner et al., 2022). However, the concept of the substrate stiffness is complex and cell response greatly depends on spatial dimension (e.g., 2D vs. 3D) and topography, such as aligned grooves, and the choice of material. To add complexity, myocardium is anisotrophic and stiffness values vary greatly between longitudinal and cross sectional parameters, which makes mimicking of the physical properties of the native matrix challenging (Querceto et al., 2022). The effects of substrate stiffness and different materials on hiPSCs have been summarized in Table 2.

In 2D configuration hiPSCs have been cultured on substrates ranging from soft hydrogel (0.56–300 kPa) and PDMS (5–30 kPa) to rigid glass (25 GPa) as shown in Table 3. Rigid (>1 GPa) surfaces were found to cause unphysiological contractions for hiPSC-CMs meanwhile soft (22 kPa) hydrogel substrate allowed normal contractility (Huethorst et al., 2022). In a single cell study, where hiPSCs were compared between stiffer (112 kPa) 2D surfaces and softer (9.83 kPa) 3D configuration, cells exhibited increased Cx-43 density, cell membrane stiffness and CT amplitude in the latter (Kit-Anan et al., 2021).

In 3D, hiPSC-CM have shown stronger maturation characteristics in 9 kPA and 2 kPA gelatin hydrogel in comparison to 16 kPa hydrogel (Lee et al., 2017). Martewicz et al. reported that RhoA/ROCK mechanotransduction pathway is involved in hiPSC-CMs sensing the stiffness of the material (Martewicz et al., 2017). Ribeiro et al. reported that the material stiffness affects hiPSC-CM sarcomere dynamics up to the physiological levels whereas stiffness values beyond that did not seem to have an increasing effect (Ribeiro et al., 2020).

Inhomogeneities within the hydrogel scaffolds after preparation can cause variations in the stiffness each cell is experiencing (Soto et al., 2016). The stiffness also changes over time as the cells remodel their environment as is shown with microrheological measurements (Hafner et al., 2020; Koivisto et al., 2022). It is important to acknowledge that the cells in one culture are sensing varying stiffnesses. Measuring the stiffness of the whole bulk scaffold gives useful insight into the mechanical properties, while characterizing the cells’ actual microenvironment does give more realistic insight to the situation.

Passive stretch in 3D EHTs is done by fixing the constructs between posts. The distance of the posts is either kept the same throughout the study or increased in stepwise manner mimicking the dynamic changes in the myocardium during development. While stretch creates tension in the cell structures, it also orientates the cells in the direction of stretch (Ruan et al., 2016; Lui et al., 2021). Leonard et al. tested different increases of afterload on hiPSC-CM-based EHT (Leonard et al., 2018). Afterload is the systolic load on the left ventricle during a contraction (Kligfield, 1998). From afterload forces between 0.09 up to 9.2 μN/μm the mid-range was discovered to most improve the maturation of hiPSC-CM in terms of increased sarcomere length, cardiomyocyte area and elongation. Progressive afterload was found to increase several key markers of maturation including a shift from fetal to adult ventricular myosin heavy chain isoforms. Interestingly, markers of pathological hypertrophy and fibrosis were upregulated at the highest afterload condition (Leonard et al., 2018). The effect of mechanical stimulation on hiPSC maturation is summarized in Table 3.

In active mechanical stimulation, cells are stimulated externally with changing forces rather than causing the stimulus, for example, by the contracting cells attached to fixed anchors. In comparison to passive mechanical stimulation, significantly less studies are done on the active mechanical stimulation of hiPSC-CM. Song et al. studied the effect of 1 Hz cyclic stretching to generate a 10% strain for 48 h on hiPSC-CM in 2D PDMS platform. Cyclic stretching increased the expression of genes related to nuclear mechanotransduction. Actin alignment and sarcomere length was higher in the direction of the stretch. Beating rate was similar in both groups (Song et al., 2022). In contrast, in our previous 2D study with cyclic stretching (0.5 Hz pre-stretch with 8% elongation following 0.8 Hz stretch with 8% elongation) of hiPSC-CM on PDMS substrate showed orientation of cells perpendicular to the direction of stretch (Kreutzer et al., 2020). Different magnitudes of 1 Hz cyclic stretching were compared by Dou et al. using building 2D PDMS membrane as the substrate for hiPSC-CM (Dou et al., 2022). The contractile stress of the cells increased with increased strain (0%, 5%, 10%, 15% and 20%) up to 15% of strain, where it plateaued. Strain of 20% did not increase contractile stress. Mechanical stress overall increased intercellular alignment and sarcomere orientation to radial direction (circumferential) and MYH7 protein expression.

Shear stress for mimicking the early stage of heart tube development has been studied for hiPSC-CM in an MPS system by Kolanowski et al. (Kolanowski et al., 2020). Pulsatile fluid flow generated in the MPS system increased several indicators of maturation such as the beating intensity, sarcomere length, mitochondrial density, and the elongation of the mitochondrial network.

Electrical stimulation for maturation of hiPSC-CM has been more widely studied than active mechanical stimulation. The reason for this could be that electrical stimulation is relatively easy to implement in comparison to active mechanical stimulation where the accurate anchoring of cells could pose additional challenges. This chapter encompasses also those studies made on hiPSC-CM that include mechanical stimulation with electrical stimulation but have not compared those conditions to mechanically non-stimulated controls. Therefore, they focus solely on studying the effect of electrical stimulation. The effect of electrical stimulation on hiPSC maturation is summarized in Table 4.

By varying the pulse frequency of electrical stimulation, different “training regimes” can be applied for striated muscles. The state of maturity of hiPSC-CM has also shown to play a role when responding to electrical stimulation. Ronaldson-Bouchard et al. hypothesized that electrical stimulation should be initiated to early hiPSC-CM, which still have high developmental plasticity, because responsiveness to physical stimuli declines as differentiation progresses (Ronaldson-Bouchard et al., 2018). In addition, a gradual ramp-up of the frequency of electrical stimulation each day pushes hiPSC-CM to adapt on the workload (Ronaldson-Bouchard et al., 2018; Ronaldson-Bouchard et al., 2019). They compared constant 2 Hz monophasic (4.5 V/cm) stimulation for 3 weeks, with “high intensity training” parameters where the frequency increases 0.33 Hz each day up to 6 Hz by day 12 (Ronaldson-Bouchard et al., 2019). Because of the pillar configuration, passive mechanical stimulation was triggered by contractions of the tissue structures. However, the conditions were not compared to non-static conditions. The stage of maturation with the early-stage high intensity-trained hiPSC-CM was at similar levels as achieved with previous biowire configuration (Nunes et al., 2013). Only these hiPSC-CM displayed orderly signal propagation and anisotropic gap junctions and exhibited positive force-frequency relationship (Ronaldson-Bouchard et al., 2018).

The voltage amplitude is another major variant that needs to be balanced case-to-case depending on for instance electrode materials, distance between electrodes and the pulse width (Björninen et al., 2018). Pretorius et al. compared low voltage (15 V) and high voltage (22 V) stimulation on hiPSC-CM with the commonly used C-Pace (IonOptix) stimulator (Pretorius et al., 2022). Cell constructs were manufactured into triple-layered cardiomyocyte patches and 2 ms voltage pulses at 2 Hz was applied for the patches under static stretching. Patches with static stretching and with no stretching were used as controls. The stretched patches with 22 V stimulation showed greatest expression and alignment of contractile proteins. Stretching and electrical stimulation seemed to cause greater development in Z-lines and gap junctions. Sarcomeres were significantly longer in any of the stimulated group compared to the non-stretched control.

Ruan et al. compared static stretch and that added with electrical stimulation. They found that while static stretch of hiPSC-CM increased in contractility, tensile stiffness, construct alignment, and cell size, additional electrical stimulation increased force production. They suspected that electrical stimulation promoted maturation of excitation-contraction coupling as expression of RYR2 and SERCA2 increased (Ruan et al., 2016).

Active electromechanical stimulations (mechanical stimulation is active), where both the effects of mechanical stimulus and electrical stimulus are compared together or separately, are still few. Lu and coworkers applied progressive stretch on EHTs and compared with those subjected to static stress (Lu et al., 2021). Electrical stimulation parameters were kept constant. Posts, where EHTs were anchored were moved further from the other post daily and electrical stimulation was applied immediately after. The highest rate of stretch resulted in greatest force development. The role of electrical stimulation remained unclear as it was not compared to a control without electrical stimulation.

The limitless options of cell culture conditions and parameters in electromechanical stimulation makes the area particularly challenging to navigate. The stimulus, that ought to be mimicking physiological conditions, can also have a negative effect on maturation markers. Such effect is demonstrated by Kroll et al. who tested elecromechanical stimulus on hiPSC-CM up to 7 days of stimulation (Kroll et al., 2017). After each stretching phase of cyclic stretch of 5%, electrical stimulation was applied. Both stimulation types were also separately tested in separate groups. The electromechanical stimulus aimed to mimic the filling of the ventricles and an isovolumetric contraction. The cells had been synchronously beating for 3 days before the start of the stimulus. Neither of the stimulation regimes altered structural phenotype of the cells. Surprisingly sarcomere length was the shortest in all stimulated groups whereas non-stimulated control was nearly on the level of the optimal range in human heart tissue ranging in 1,9–2.2 μm (Hamdani et al., 2007). The greatest effect on myofibrillogenesis was observed with the EM stimulation which also significantly reduced the transmembrane calcium current. The authors speculated that a gradual increase in stimulation would be required to alleviate the stress response of the cells (Kroll et al., 2017).

The measurements of CM contraction cycle quantify the CM model functionality and acute responses to environment changes. In short, the CM contraction cycle begins with an AP, rapid depolarization of the membrane, which causes an inflow of Ca²⁺ ions. Due to the depolarization and calcium-induced calcium release, more Ca²⁺ is released from sarcoplasmic reticulum. The binding of Ca²⁺ to troponin C initiates cellular contraction. Thus, the contraction cycle can be roughly characterized on three levels, as illustrated in the Figure 5 analogue: measurements of electrophysiological properties, measurements of intracellular calcium, and measurements of contractility and biomechanics. The rapid processes of contraction can be illustrated as a series of waterfalls, each influencing the flow in subsequent waterfall. On the other hand, structural defects may not manifest themselves as disturbances on the electrical level. The waterfalls require measurements with high sampling rates in situ, whereas the (bio-)chemical changes in the culture are slower, and can thus be measured in the calm lakes, accumulations of several contractions.

An advantage of heart MPS is the capability to include multiple sensors and electrodes for continuous measurement and monitoring of multiple measurement modalities, even concurrently. However, the system itself, and the tissues being measured present challenges to the traditional in vitro measurement techniques, with aspects such as long-term measurements and the inclusion of medium perfusion. The inclusion of vascular-like perfusion in MPS studies is an important aspect of experiment design due to its physiological analogues and its use in pharmacological studies. Further, it enables the long-term quantification of functionality and improves the model quality, such as Ca²⁺ transient amplitude increase and time-to-peak duration decrease (King et al., 2022). Here, the following sections and Table 5 briefly review these methods, their uses cases and discuss possible prospective technologies in MPS taking the previously mentioned aspects into consideration.

Our review suggests that there are emerging technologies for mechanical and electrical measurements, which are likely to be adopted in the future. The efficacy of the most novel methods have yet to be evaluated in hiPSC-CM disease models and drug studies. The reviewed studies show the potential of combining different measurement modalities, which may be fruitful for perfusion-enabled heart-on-chip studies as the field matures.

In a proper MPS, the environmental conditions should mimic the corresponding conditions in vivo. In essence, temperature and pH should be kept within physiological values and, importantly as well, partial oxygen pressure (pO₂) should be regulated within tissue-specific limits (Wilson, 2008). In addition, MPS systems modelling cardiac ischemia reperfusion should provide means to generate ischemic events on demand. For the pO₂ regulation and control purposes alone, MPS systems should contain means for real time pO₂ monitoring. Moreover, as described earlier the monitoring of pO₂ and other chemical and biochemical parameters provides important functional data about the metabolic and systemic state of the system. The following two subchapters and Table 6 gives a brief survey on the most important chemical analytes for cardiac MPS and available measurement technologies.

Computational models of hiPSC-CM ion dynamics and electrophysiology have been developed since 2012, and their pedigree has diverged to five unique branches. They do, however, share a basis that is common to all biophysics-based models of AP in cardiomyocytes: Hodgkin and Huxley (Hodgkin and Huxley, 1952) formalism to describe ion currents and fluxes. The shared and unique characteristics of the five pedigree branches of hiPSC-CM models are described in detail in Table 7.

In both low and high throughput in vitro studies of hiPSC-CMs, the central readouts were initially the AP and the CT. Indeed, analysis of the (rather) standardized morphological biomarkers of these two variables provides a good characterization of the electrophysiological phenotype, while also alluding towards the bigger picture of the excitation-contraction coupling phenotype. Building on that tradition, during the past few years, contractile force and its dynamics have gained an equally important role as a measurand in hiPSC-CM studies.

The developmental steps of computational hiPSC-CM models have paralleled the progress in experimental procedures. Initially, the emphasis was in modeling of electrophysiology and intracellular ion dynamics (see Table 7 for details). Recently, hiPSC-CM models have been extended to include mathematical description of contractile cellular mechanics. Actually, pioneering research was published by Frotscher and colleagues already in 2015 and 2016 (Frotscher et al., 2015; Frotscher et al., 2016). As the authors’ focus was to develop an electromechanically coupled model for EHT, they used a purkinje fibre CM model (McAllister et al., 1975) to recapitulate the immature phenotype of hiPSC-CM and coupled that cellular electrophysiology model with the contractility component (Hunter et al., 1998). We have recently published the first hiPSC-CM-specific electromechanical models (Forouzandehmehr et al., 2021; Forouzandehmehr et al., 2022).

As discussed in chapter 3, although there is continuous development of differentiation methods, the hiPSC-CMs phenotype differs from the native human counterparts, having some foetal features. In this context, computational modelling offers essential translational tools. Indeed, those capabilities have already been demonstrated in multiple independent studies, in which in vitro hiPSC-CM-based findings were translated to the human context (Gong and Sobie, 2018; Koivumäki et al., 2018; Tveito et al., 2018; Jæger et al., 2020). Accessibility and applicability of such tools would greatly benefit from transparent comparisons of the computational models, as was done recently for hiPSC-CM models by (Paci et al., 2021).

Capabilities of the so-called in silico drug trials in early-phase cardiotoxicity detection have already been demonstrated in that human ventricular CM models to have higher accuracy than ex vivo animal models in predicting clinical risk of drug-induced arrhythmias (Passini et al., 2017). The physiological robustness and specificity of computational hiPSC-CM models is approaching the needed level, so that they serve a corresponding role in high-throughput analysis of drug candidates.

The technical concepts for utilization of hiPSC-CM-based in vitro system stretch from single CMs, CM clusters, and 2D monolayers to organoids, and all the way to EHT. While the more environmentally complex end of the spectrum replicates to some extent in situ cell-cell communications, in 2D monolayers and CM clusters even the basic electrophysiological and biomechanical cell–cell interactions have unique properties. To recapitulate the conduction of electrical and mechanical activation in such an MPS system, the classical homogenization/continuum-based approaches of modelling tissue/organ function are likely to fall short. Instead, approaches that explicitly represent the extra- and intracellular spaces as well as cell–cell connections will be essential to capture the heterogeneities (for an extensive review, see (Tveito et al., 2021)).

To our knowledge, multiphysics models that would put together all the building blocks of MPS in integrative simulations have not yet been published. First steps towards using a classical continuum-based fully coupled electromechanical model for MPS were taken by (Frotscher et al., 2015; Frotscher et al., 2016). More recently, the same group extended their tissue model to include also fibroblasts in addition hiPSC-CMs, thus enabling mechanistic analysis of their contribution to contractility in MPS (Jung and Staat, 2019). Expanding the models to include vasculature and innervation is an interesting but not trivial step. Also, as preliminary studies of replicating environmental control and physical stimuli of MPS in simulations have been teased in conference presentation, those building blocks are bound to see light of the day in the not-too-distant future.