Extracellular Matrix–Mediated Maturation of Human Pluripotent Stem Cell–Derived Cardiac Monolayer Structure and Electrophysiological Function

Cryopreserved vials (liquid nitrogen) of iCell human cardiac myocytes were obtained from Cellular Dynamics International, Inc (Madison, WI). iCell cardiac myocytes are highly purified (>98%) hPSC-derived cells that are cryopreserved after 30 to 31 days of cardiac-directed differentiation. Cells were thawed and subsequently plated on bovine fibronectin-coated (20 μg/mL; Life Technologies, Thermo Fisher Scientific, Waltham, MA) glass coverslips, on bovine fibronectin-coated (20 μg/mL; Life Technologies) transparent polydimethylsiloxane (PDMS membranes, matrigel-coated (500 μg/mL; BD Biosciences, San Jose, CA) glass coverslips, or matrigel-coated (500 μg/mL; BD Biosciences) transparent PDMS membranes at a density of 125 000 cells per monolayer (1500 cardiomyocytes/mm²) in differentiation medium.^5,10 In all cases, a 100-μL droplet of cells was plated on each ECM. Differentiation medium (EB20), consisting of 80% DMEM/F12, 0.1 mmol/L nonessential amino acids, 1 mmol/L l-glutamine, 0.1 mmol/L β-mercaptoethanol, 20% FBS, and 10 μmol/L blebbistatin, was used to culture the cells to promote attachment. Each ECM was added as a 100-μL drop of solution in the center of each coverslip; this routinely produced a 1-cm diameter circle of ECM on which the cardiomyocytes were plated after a 30-minute incubation period and aspiration of liquid. PDMS silicone sheeting was obtained from SMI (Specialty Manufacturing, Inc, Saginaw, MI) with 40D (D, Durometer or ≈1000 kPa) hardness and cut to 18×18 mm coverslips. After 24 hours, the media was switched to RPMI (Life Technologies) supplemented with B27 (Life Technologies). The cells were subsequently cultured for 4 to 7 days at 37°C, in 5% CO₂ before phenotype analysis. The use of hPSCs and derived cardiomyocytes was approved by the Human Pluripotent Stem Cell Research Oversight (HPSCRO) Committee of the University of Michigan.

Optical action potentials (APs) in Figure 1 were recorded using FluoVolt membrane potential probe (F10488; Life Technologies). For calcium transient (CaT) and propagation experiments, monolayers were loaded with the intracellular Ca²⁺ indicator, rhod-2AM (10 μmol/L; Life Technologies) or fluo-4AM (5 μmol/L; Life Technologies) as indicated. After a 30-minute incubation time, the cells were washed in Hanks’ balanced salt solution (Life Technologies) for an additional 30 minutes before optical mapping recordings. All human cardiac monolayers displayed pacemaker activity, and the spontaneous APs or calcium waves were recorded using a charge-coupled device camera (200 fps, 80×80 pixels; Red-Shirt Little Joe, Scimeasure, Decatur, GA) with the appropriate emission filters and light-emitting diode illumination.²¹ Movies were filtered in both the time and the space domain, and CV was measured as described previously.^22,23

Details of Western blotting, flow cytometry, immunofluorescence, and image analysis can be found in the expanded Methods in the Data Supplement.

In Figure 1 (left), we provide the first demonstration that purified iCell cardiomyocytes cultured as monolayers on PDMS+matrigel achieve a high degree of electric maturity, with average AP propagation velocities as high as 55 cm/s. It is important to note, however, that while iCell cardiomyocytes are highly purified; one always encounters mixtures of different cardiomyocyte phenotypes, including atrial-like, ventricular-like, and pacemaker-like myocytes.^9,10,17 This is reflected in Figure 1A by the different optical AP configurations (right) in the monolayer. Pacemaker-like cells at the site of impulse initiation undergo slow diastolic depolarization at a steady rate until the threshold potential is reached and an AP is generated. More distally, ventricular-like APs have stable resting membrane potentials and respond to the propagating impulse with rapid upstrokes. In Figure 1B and 1C, we characterized AP propagation velocity and duration over electric pacing frequencies ranging from 0.7 to 2.5 Hz. Figure 1B shows CV restitution as one would expect with faster conduction at lower frequency (greater cycle length) of stimulation. Figure 1C demonstrates the AP duration restitution of mature hPSC-CM monolayers where AP duration gets shorter as pacing frequency increases. Thus, the biomatrix combination of matrigel+PDMS promotes functional electrophysiological maturation of hPSC-CMs in as little as a week.

To establish whether the softer ECM provided by PDMS+matrigel promoted maturation significantly more than stiffer matrices, we tested the combinations presented in Figure 2A by quantifying CaT propagation across the monolayer. First, the rate of spontaneous pacemaker activity was recorded for each of the 4 conditions. There was no difference in the spontaneous activation rate between the groups with the averages being the following: (i) fibronectin+glass=0.25±0.05 Hz, n=10; (ii) fibronectin+PDMS=0.22±0.03 Hz, n=4; (iii) matrigel+glass=0.24±0.09 Hz, n=14; and (iv) matrigel+PDMS=0.2±0.07 Hz, n=9 (mean±SEM; one-way ANOVA; P>0.9999). In Figure 2B, representative color CaT activation maps of wave propagation for each of the experimental conditions are shown. The fastest CV was observed in human cardiac monolayers cultured on matrigel+PDMS (Figure 2B, iv). The quantification of CV in each condition is shown in Figure 2C. The average CVs were as follows: fibronectin+glass=21.6±6.8 cm/s, n=10; fibronectin+PDMS=24.6±6 cm/s, n=4; matrigel+glass=22.0±4.0 cm/s, n=14; and matrigel+PDMS=43.6±7.0 cm/s, n=9 (mean±SEM; 1-way ANOVA; see figure and legend for details). Upper 95% confidence interval for the matrigel+PDMS group is 47.8 cm/s. Point stimulation of monolayers (15–20 V; 5-ms duration; 1 Hz) in each condition was performed in a separate group of experiments to determine differences in CV. The average CVs during 1 Hz pacing were as follows: (i) fibronectin+glass=24.2±1.8 cm/s, n=6; (ii) fibronectin+PDMS=28.1±1.5 cm/s, n=4; (iii) matrigel+glass=28.4±3.2 cm/s, n=5; and (iv) matrigel+PDMS=37.1±1.7 cm/s, n=6 (Figure 2D; mean±SEM). Thus, CV was faster during spontaneous pacemaker activations, as well as during 1 Hz electric pacing, in human cardiac monolayers cultured on matrigel-coated PDMS coverslips. In parallel experiments, iCell cardiomyocytes were plated on laminin or collagen purified from matrigel to test whether these individual components of matrigel would support confluent 2-dimensional (2D) monolayer maturation on PDMS. The results in Figure I in the Data Supplement show that monolayer confluence was not maintained for 7 days in culture on these purified components of matrigel. Optical mapping of AP propagation was continuous in each of the conditions tested (ie, laminin and collagen), suggesting that cellular reorganization was occurring rather than cell death. CV was fastest in monolayers plated on matrigel compared with those on collagen or laminin (Figure IC in the Data Supplement). Furthermore, cardiac troponin I (cTnI), the mature myofilament marker, was only induced when monolayers were plated on matrigel (Figure ID in the Data Supplement). Surface chemistry of cell culture surfaces can also impact on cellular phenotype,²⁴ and the surface chemistry of glass is distinct from PDMS. Therefore, additional experiments were performed where glass coverslips were silanized to make their surface chemistry comparable with PDMS. Results of optical mapping 2D monolayers cultured on silanized glass indicate that the differences in surface chemistry cannot account for the rapid CV that we observed on PDMS (Figure II in the Data Supplement).

APs are required for propagation of the electric signal that triggers the Ca²⁺-mediated excitation–contraction coupling.^10,13 Therefore, hPSC-CM monolayer APs were recorded and quantified by patch-clamp analysis in current-clamp mode. Properties of the AP, such as the dV/dt_max (V/s; Figure 3A and 3C), the maximum diastolic potential (MDP; Figure 3D), and the threshold potential (takeoff potential; Figure 3E), provide quantitative metrics of the degree of myocyte maturity.^9,11,13 We measured these functional parameters in hPSC-CMs cultured on the extremely rigid ECM condition of fibronectin+glass (black bars; Figure 3) and compared the results with hPSC-CMs cultured on the softest ECM condition of matrigel+PDMS (red bars; Figure 3). APs recorded from hPSC-CMs grown on matrigel+PDMS displayed significantly faster upstroke velocities (65.3±8.9 V/s, N=6 monolayers, n=37, versus 146.5±17.7 V/s, N=5 monolayers, n=24; Figure 3C), more hyperpolarized MDPs (−69.9±1.7 mV, N=6 monolayers, n=37, versus −77.5±0.6 mV, N=5 monolayers, n=24; Figure 3D), and more hyperpolarized takeoff potentials (−59.3±1.7 mV, N=6 monolayers, n=37, versus −70.5±1.2 mV, N=5 monolayers, n=24; Figure 3E), all indicative of cardiomyocyte maturation.²⁵

The faster dV/dt_max and faster impulse propagation may be attributed partially to the effect of matrigel+PDMS ECM to increase sodium current (I_Na) density. Indeed, single-cell patch-clamp analysis of I_Na revealed significantly elevated current density in cardiomyocytes cultured on PDMS+matrigel compared with cardiomyocytes from the same batch cultured on fibronectin on glass coverslips (Figure 3F and 3G; glass N=4, n=12, and PDMS N=4, n=19). Elevated I_Na density observed here is consistent with previous data, showing that I_Na density increases in the maturing heart.²⁶ Figure III in the Data Supplement shows the I_Na activation/inactivation profiles. The inactivation profile is left-shifted for cardiomyocytes cultured on PDMS+matrigel (V1/2=−79.0±2.45 mV, N=4, n=13, versus −88.0±1.55 mV, N=4, n=12; P=0.009), which suggests increased expression of cardiac sodium channel isoforms (ie, SCN5A and Na_V1.5) that is known to be larger in adult than immature embryonic cardiac myocytes.^26,27 Real-time–polymerase chain reaction analysis confirmed elevated SCN5A gene (Figure 3H) expression in iCell iPSC-CMs cultured on matrigel-coated PDMS compared with iPSC-CMs cultured on fibronectin-coated glass coverslips.

It has been reported by others that the MDP and spontaneous activity of hPSC-CMs are critically dependent on the density of the rapid component of the delayed rectifier potassium current (I_Kr) and that I_K1 density is extremely low if not absent in hPSC-CMs.²⁸ On the contrary, the hyperpolarized MDP of iCell cardiomyocytes cultured on PDMS+matrigel that we found here suggests that the I_K1 density is relatively high. Indeed, patch-clamp analysis revealed elevated I_K1 density in cardiomyocytes maintained on PDMS+matrigel biomatrix (Figure 4A and 4B). Western blotting shows expression of Kir2.1 exclusively in cardiomyocytes cultured on PDMS+matrigel compared with the same cells cultured on fibronectin-coated glass coverslips (Figure 4C). To further investigate whether PDMS+matrigel promotes the expression of other potassium currents, we tested the effects of E4031 (100 nmol/L), a selective blocker of I_Kr on the spontaneous beating rate and CaT duration 80 of hPSC-CM monolayers cultured on matrigel-coated PDMS compared with those cultured on standard rigid, plastic bottom 96-well dishes (also matrigel coated, Figure IV in the Data Supplement and Video I in the Data Supplement illustrate how this multiwell optical mapping platform was used). Small glass coverslips to fit into a 96-well format dish are not readily available, so here we used plastic as the rigid comparator for PDMS. Figure 4D and Figure IV in the Data Supplement each show that the electrophysiology of hPSC-CM monolayers cultured on PDMS, measured by the CaT duration, is less affected by I_Kr blockade, further supporting the evidence in Figure 4 of elevated inward rectifier potassium current density in hPSC-CMs cultured on matrigel-coated PDMS coverslips. For example, 100 nmol/L E4031 reduced spontaneous beating frequency by ≈50% (0.83±0.02 Hz pre-E4031 versus 0.45±0.05 Hz post-E4031; n=8) in monolayers plated on rigid plastic, whereas the same dose of E4031 modestly reduced beating frequency by ≈20% (1.02±0.04 Hz pre-E4031 versus 0.78±0.07 Hz post-E4031; n=5). This effect on beating frequency is likely because of the effect of E4031 to prolong the AP and CaT durations. Similarly, CaT duration 80 increased ≈3X after E4031 application in hPSC-CMs plated on plastic bottom dishes (605.9±24.3 ms pre-E4031 to 1873.8±278.9 ms post-E4031; n=8), but CaT duration 80 only increased by ≈1.34× in hPSC-CMs plated on PDMS bottom dishes (579.8±13.9 ms pre-E4031 versus 780.1±74.1 ms post-E4031; n=5; Figure IV in the Data Supplement).

After birth, myocytes in the heart switch from hyperplastic growth to hypertrophic growth, and this is part of the natural maturation process.^11,29,30 Therefore, another marker of maturation of hPSC-CMs is the transition from cardiomyocytes remaining in the cell cycle to myocyte terminal differentiation and hypertrophy. The hPSC-CMs cultured on pliable PDMS were significantly larger in size than cells plated on glass substrates (Figure 5C and 5D). This indicates induction of developmental hypertrophy in hPSC-CMs cultured on soft substrates. Binucleation, another marker of myocyte maturity,^30,31 was also apparent in hPSC-CMs cultured on matrigel+PDMS (Figure 6A and 6B; Figure VI in the Data Supplement). The cell cycle marker Ki67 was also used to determine the differentiation state of cardiomyocytes.³² In iCell cardiomyocytes, the percentage of binucleated cells that were also Ki67+ was lower in iPSC-CMs cultured on PDMS+matrigel (Figure 6B). Furthermore, using highly purified BJ-hPSC-CMs developed in our laboratory³³(Figure 5B), we quantified the effect of PDMS to reduce the number of cardiomyocytes remaining in the cell cycle (Figure 6C). One key myofilament marker of cardiomyocyte maturation is cTnI expression.^11,34 We took this parameter of maturation into consideration here by Western blot analysis of BJ-hPSC-CM expression of cTnI when purified monolayers were cultured on glass or PDMS coverslips coated with matrigel. Significantly more robust cTnI expression was detected relative to GAPDH in purified BJ-hPSC-CMs plated on PDMS coverslips compared with glass coverslips (Figure 6D), thus indicating a greater level of sarcomeric maturation and promotion of developmental isoform switching that is known to occur in the postnatal heart. Furthermore, cTnI sarcomeric incorporation was also apparent in hPSC-CM monolayers cultured on PDMS but not in immature monolayers cultured on Glass coverslips (Figure VC in the Data Supplement).

Integrins are transmembrane heterodimeric receptors essential for providing cell-ECM adhesion, cellular structural organization, and transduction of mechanical signals from the ECM into biochemical signals in cardiomyocytes.^35–37 β1 integrins are abundant in the adult heart and participate in the hypertophic response in rodent ventricular myoyctes.³⁸ Therefore, we hypothesized that integrin signaling underlies the maturation of hPSC-CM monolayers induced by the cell culture condition of plating them on matrigel-coated PDMS coverslips. First, real-time–polymerase chain reaction analysis showed that ITGB1 expression is significantly induced on PDMS coverslips compared with glass coverslips (Figure 7A). In addition, PTK2 gene expression is elevated in hPSC-CM monolayers cultured on PDMS coverslips (Figure 7B). The PTK2 gene encodes the focal adhesion kinase (FAK) intracellular molecule that is a primary mediator of integrin signaling.³⁹ Figure 7C shows the localization of β1 integrin receptors in monolayers; there was elevated expression in monolayers cultured on PDMS.

Next, we blocked integrin signaling by 2 different approaches: (1) neutralization of β1 integrin receptor activation using isoform-specific antibody, and (2) pharmacological blockade of FAK activation using FAK inhibitor-14. Importantly, lower dose of FAK inhibitor-14 did not reduce hPSC-CM monolayer confluence significantly (figure VII in the Data Supplement).Western blotting in Figure 7D indicates that FAK activation was prevented by Nab (neutralizing antibody) and also that cTnI protein expression was attenuated. Collectively this indicates that FAK activation via β1 integrin activation contributes to the expression of mature myofilament markers, such as cTnI. Furthermore, purified hPSC-CMs (iCell) cultured on PDMS in the presence of 10 μmol/L FAK inhibitor-14 failed to express cTnI and also expressed less β-MyHC (β-myosin heavy chain) (Figure 7E; Figure VIII in the Data Supplement). This suggests that FAK activation underlies expression of mature myofilament markers. In addition, we determined the role of FAK activation in the hypertrophic response of hPSC-CMs to the soft cell culture environment on PDMS (Figure 7F; Figure VII in the Data Supplement). FAK inhibitor-14 prevented hypertrophic growth of hPSC-CMs observed on PDMS coverslips. Interestingly, hPSC-CMs grown on glass coverslips were only affected at the higher dose of FAK inhibitor-14 (100 μmol/L). In both conditions (glass and PDMS), 100 μmol/L FAK inhibitor-14 reduced monolayer confluence and hPSC-CM expansion in culture (Figures VII and VIII in the Data Supplement). Finally, the effects of integrin receptor activation were next studied using a monoclonal antibody that stabilizes the dimerization of α₅β₁ integrins and promotes FAK activation (MAB1969). Figure 8 summarizes the Western blot analysis of hPSC-CM monolayers (iCell) treated with the activating α₅β₁ integrin monoclonal antibody or FAK inhibitor-14. FAK activation (phosphorylation) was induced by this specific antibody (Figure 8B) and so was cardiac sodium channel protein expression (Nav1.5; Figure 8C).