Coculture of Endothelial Cells with Human Pluripotent Stem Cell-Derived Cardiac Progenitors Reveals a Differentiation Stage-Specific Enhancement of Cardiomyocyte Maturation

Abstract

Cardiomyocytes (CMs) generated from human pluripotent stem cells (hPSCs) are immature in their structure and function, limiting their potential in disease modeling, drug screening, and cardiac cellular therapies. Prior studies have demonstrated that coculture of hPSC-derived CMs with other cardiac cell types, including endothelial cells (ECs), can accelerate CM maturation. To address whether the CM differentiation stage at which ECs are introduced affects CM maturation, we cocultured hPSC-derived ECs with hPSC-derived cardiac progenitor cells (CPCs) and CMs, and analyzed molecular and functional attributes of maturation. ECs had a more significant effect on acceleration of maturation when cocultured with CPCs than with CMs. EC coculture with CPCs increased CM size, expression of sarcomere and ion channel genes and proteins, the presence of intracellular membranous extensions, and chronotropic response compared to monoculture. Maturation was accelerated with increasing EC:CPC ratio. This study demonstrates that EC incorporation at the CPC stage of CM differentiation expedites CM maturation, leading to cells that may be better suited for in vitro and in vivo applications of hPSC-derived CMs.

1. Introduction

Cardiac failure affects 6.5 million people in the US annually and accounts for 9% of the total deaths._[1] Cell-based therapies represent a promising class of emerging treatments to combat this mortality. Heart attacks often result in heart failure, killing approximately one billion cardiomyocytes (CMs) via hypoxia._[2] Adult CMs are minimally proliferative, with only about 1% of the cells proliferating per year in an adult heart._[3] Current methods of treatment for heart failure are limited; left ventricular assist devices only provide temporary aid and cardiac transplants are constrained by a shortage of donor hearts._[4] The proliferation and differentiation potential of human pluripotent stem cell (hPSC)-derived cardiac cells and tissues provides a novel strategy to restore contractility to a failing heart._[5]

Recently, hPSC-derived CMs differentiation methods have been optimized such that a nearly pure population of CMs can be created in fully-defined conditions._[6] Scale-up of CM differentiations to produce over one billion CMs at >90% purity has been achieved using microcarriers in a 1L flask._[7] When hPSC-derived CMs were injected intramyocardially into major histocompatibility complex-matched or immune-suppressed nonhuman primate hearts, cardiac function after myocardial infarction was significantly enhanced._[8-10] Unfortunately, the implanted CMs caused transient arrhythmias and suffered from poor long-term survival._[8-10] The arrhythmias may result from intrinsic spontaneous electrical signals generated by immature hPSC-derived CMs,_[10] which more closely resemble fetal CMs than CMs found in adult hearts. For example, hPSC-derived CMs express higher levels of myosin light chain 2a (MLC2a), α-myosin heavy chain (MHC), and slow skeletal troponin I (ssTnI) and lower levels of myosin light chain 2v (MLC2v), β-MHC, and cardiac troponin I (cTnI) than adult ventricular CMs._[11] Additionally, expression of channels, regulators, and transporters such as CACNA1C, SCN5A, HCN4, KCNJ2, ATP2A2, RYR2, and GJA1 is generally lower in hPSC-derived CMs than in adult CMs._[12] Additionally, hPSC-derived CMs are much smaller and rounder than adult CMs, and lack aligned myofibrils, localized gap junctions, and organized sarcomeres. hPSC-derived CMs spontaneously contract, lack t-tubules, and exhibit slower Ca₊₂ conduction than adult CMs._[13-14] In adult CMs, gap junctions localize at the cell membrane at the end of the myofibrils, and contain the gap junctional protein Cx43, allowing the flow of ions between adjacent cells. Also, hPSC-derived CMs utilize metabolic pathways similar to those employed in the fetal heart, including glycolysis and glucose oxidation, rather than fatty acid β-oxidation._[15] These immature phenotypes limit the potential of hPSC-derived CMs in drug screening and clinical applications.

Several strategies have been shown to induce maturation in hPSC-derived CMs, but thus far no method has yet generated an hPSC-derived CM that fully mimics an adult CM._[5] With extended time in culture, hPSC-derived CMs gained more organized sarcomeres and more mature gene expression profiles and electrophysiology after 90 to 120 days._[16-17] Electrical stimulation of hPSC-derived CMs enhanced expression of ion channels, cTnI, and Cx43._[18-19] Mechanical stimulation induced expression of adult CM genes and proteins and accelerated Ca₂₊ handling._[20-22] Ronaldson-Bouchard et al. reported ultrastructural sarcomere organization, sarcomere spacing of 2.2 μm, and more mature gene expression profiles after 4 weeks of simultaneous and continuous mechanical and electrical conditioning, but the cells did not generate the same contractile forces as adult CMs._[23] Culturing CMs on micropatterned or soft substrates had a variety of impacts on maturation, including enhanced structural organization of sarcomeres and myofilaments, increased MLC2v and β-MHC expression, cell size, and contractility._[24-26] Incorporation of electrically conductive materials, through integration in cardiac spheroids or with the creation of cardiac films on polymer-covered plate, increased the expression of Cx43 and cellular alignment while decreasing the calcium transient time._[27-28] Similarly, biochemical cues such as hormone production or metabolic induction also can induce partial maturation._[29-30]

Heterotypic intercellular interactions also impact CM maturation. Coculture of mesenchymal stem cells with induced pluripotent stem cell (iPSC)-derived CMs increased CM contractility and sarcomere organization and alignment._[31] This effect was recapitulated with EC-derived exosomes containing proteins and microRNAs._[31] Similarly, fibroblasts have been shown to induce hPSC-derived CM maturation, including elevated cardiac troponin T (cTnT), Cx43, and α-actinin protein expression and contractility through both soluble factors and extracellular matrix protein production._[32-33]

To induce CM maturation and increase CM survival after implantation, hPSC-derived CMs have been cocultured with endothelial cells (ECs)._[34] ECs are abundantly found in the myocardium, with each CM in direct contact with at least one capillary._[35] Recently rat ECs and human umbilical vein ECs (HUVECs) were shown to induce structural and electrical maturation when cocultured hPSC-derived CMs; these effects were partially attributed to the transfer of microRNAs via gap junctions between the ECs and CMs._[36] ECs isolated from fat, aorta, and heart had similar effects on cocultured CMs. A similar study found that direct contact was required for ECs to enhance maturation in cocultured hPSC-derived CMs._[37] Giacomelli et al. cocultured purified hPSC-derived ECs and CMs, and found that ECs enhanced the CM chronotropic response to isoprenaline._[38]

In addition to these binary cocultures, cardiac tissues comprised of hPSC-derived CMs, ECs, and stromal cells have been constructed. Construction of cardiac patches containing mouse embryonic fibroblasts and hPSC-derived ECs and CMs led to tube-like structures and increased MLC2v expression in the CMs._[34] Another group implanted patches composed of these three cell types into rat hearts and found that the vascular-like structures integrated with host capillaries when the patches were implanted onto the surface of rat hearts._[39] The incorporation of fibroblasts and ECs enhanced transplanted CM survival compared to patches containing only CMs._[39] Another study found that culturing iPSC-derived CMs with human foreskin fibroblasts and HUVECs increased CM size._[40] Coculture with both primary cardiac fibroblasts and ECs increased hPSC-derived CM response to drugs such as isoprenaline, in comparison to monocultured CMs, EC and CM cocultures, and fibroblast and CM cocultures._[41] These microtissues displayed increased contractility, Ca₂₊ handling, and CM-specific gene expression. Interestingly, dermal fibroblasts and ECs did not induce CM maturation.

CM differentiation and the migration of ECs into the developing heart occur in concert. For example, an endocardial tube has formed next to the myocardial layer by the Hamburg Hamilton (HH) stage 8 of chick embryo development._[42] At stage HH13, endocardial-lined spaces have appeared within the myocardium. CMs begin spontaneously contracting at stage HH10 though vascularization of the myocardium does not occur until 6 days after the initiation of beating._[42] The proximity of ECs in the endocardium to the developing CMs in the myocardium may be crucial in providing the CPCs and CMs with intercellular interactions vital for CM maturation. In developmental biology, the endocardium has been shown to impact myocardial development through pathways such as FGF signaling, neuregulin, and NOTCH signaling which affect myocardial proliferation, trabeculation of the myocardium, and cardiac development gene expression._[43-46] Through these pathways and other yet unidentified mechanisms, ECs may be assisting CM maturation.

Here we addressed whether EC effects on hPSC-derived CM maturation is influenced by the state of development of the CMs at which the coculture is initiated. We investigated whether hPSC-derived ECs can induce CM maturation when the ECs are cocultured with hPSC-derived cardiac progenitor cells (CPCs) or with beating CMs. We found that culturing CPCs with ECs generated larger CMs with elevated cTnI, cTnT, and MLC2v protein expression, with greatest effects at a 3:1 EC:CPC ratio. In contrast, EC coculture had little effect on beating CMs. Additionally, electrically active intracellular membrane formation increased when CPCs were cocultured with ECs. Both CPCs and CMs exhibited greater chronotropic responses when cultured with ECs. Together these results suggest that ECs have differentiation stage-specific effects on maturation of hPSC-derived CMs.

2. Materials and methods

2.1. hPSC culture with CM and EC differentiation

hPSC lines were maintained in mTeSR1 (STEMCELL Technologies; Vancouver, Canada) on Matrigel-coated (BD Biosciences; Franklin Lakes, NJ) culture plates (Corning; Corning, NY) using versene (ThermoFisher; Waltham, MA) or ACCUTASE_™ (Innovative Cell Technology; San Diego, CA) for passaging. H9 human embryonic stem cells (hESCs) were differentiated to CPCs via the GiWi protocol._[47-49] In brief, on day 0 cells were treated with 8-12 μM CHIR99021 (Selleckchem; Houston, TX) for 24 hours in RPMI (ThermoFisher) medium supplemented with B27 minus insulin (ThermoFisher). On day 3 of the differentiation, 5 μM IWP2 (Tocris; Bristol, UK) was added to the medium for two days. On day 6, the cells were frozen for future use in the coculture experiments, with test wells maintained in culture in RPMI/B27 containing insulin (ThermoFisher) for at least 5 days to verify their CM differentiation potential.

For the EC differentiation, the protocol from Bao et al. was followed using the H1 hESC line and 19-9-11 iPSC line._[47,50] To initiate differentiation, undifferentiated cells were treated with 6 μM CHIR99021 for two days in LaSR medium (Advanced DMEM F12 (ThermoFisher) supplemented with 60 μg/mL ascorbic acid (Sigma; St. Louis, MO) and 2.5 mM Glutamax (ThermoFisher)). On day 5, CD34+CD31+ endothelial progenitor cells were purified using the EasySep FITC Positive Selection Kit (STEMCELL Technologies) for magnetic activated cell sorting (MACS) on CD34. Purified EC progenitors were plated on either 1 mg/mL gelatin (Sigma) or 10 μg/mL fibronectin (Sigma) coated, polystyrene, tissue culture plates and cultured in EGM-2 (Lonza; Basel, Switzerland) supplemented with 5 μM Y-27632 (Selleckchem). The ECs were then passaged using ACCUTASE_™ when near ~90% confluent, plated with Y-27632 in EGM-2, and maintained in EGM-2 medium.

2.2. Coculture of ECs, CPCs, and CMs

To initiate EC:CPC cocultures, cells were plated in the following ratios with 160 × 10₃ cells/cm₂ total added to each well of a 12-well plate: 1:3, 1:1, 3:1, 0:1. Cells were plated on Matrigel in EGM-2 with 5 μM Y-27632 and maintained in EGM-2 with the medium changed daily. Similarly EC:CM cocultures were initiated with 160 × 10₃ total cells at a 1:1 ratio. Upon thawing, the D6 CPCs were either seeded in EGM-2 medium with 5 μM Y-27932 with or without ECs or seeded in DMEM with 10% FBS and 5 μM Y-27932. The cells plated in DMEM with 10% FBS were then maintained in RPMI with B27 until D13-D18 (when beating was observed). At this point, the CMs were singularized with ACCUTASE_™ and plated with or without ECs in EGM-2 medium with 5 μM Y-27632, then maintained in culture for two weeks with daily medium changes. ACCUTASE_™ was used to singularize the cells for further analysis. For confocal microscopy, cells were replated onto Matrigel-coated glass plates (Cellvis; Mountain View, CA) or slides (LabTek; Grand Rapids, MI). For cell size analysis, the cells were replated onto gelatin-coated tissue culture plates at a density of 40 × 10₃ cell/cm₂.

2.3. Flow cytometry

After singularization with ACCUTASE_™, cells were fixed in 1% paraformaldehyde for 20 minutes. Primary antibodies were diluted in flow buffer (5% wt/vol BSA (ThermoFisher) in PBS containing 0.1% Triton X-100 (Sigma)) according to Supplementary Table 1 and added to fixed cells overnight at 4 °C. Secondary antibodies (1:1000 dilution in flow buffer) were incubated for 30 minutes at room temperature prior to analysis. Samples were run on a BD FACSCalibur flow cytometer. H9 hESCs were used to make CPCs and CMs and 19-9-11 and H1 hPSCs were used to make ECs for flow cytometry analysis. Statistical analysis was performed using one-way ANOVA with Tukey’s HSD post hoc analysis, Student’s t-test, Mann-Whitney test, or the Kruskal-Wallis test with Dunn’s post hoc analysis.

2.4. Immunochemistry and confocal microscopy

Cells were fixed in 4% paraformaldehyde for 15 minutes before incubation with primary antibody diluted (Supplementary Table 1) in a blocking buffer containing 5% wt/vol non-fat dry milk and 0.4% Triton X-100 (Biorad; Hercules, CA) in PBS at 4 °C overnight. Secondary antibodies were added (1:1000 dilution in blocking buffer) for at least 20 minutes at room temperature. Nuclei were stained with 0.4 μL/mL Hoechst in PBS for 5 minutes. For confocal imaging, gold antifade reagent with DAPI (ThermoFisher) was used to seal the slides. Images were taken either with an inverted Olympus IX70 microscope or with a Nikon A1R confocal microscope.

T-tubule analysis was performed by incubating live cells with 30 μM di-8-anepps (ThermoFisher) and 2.5 μL/mL of Pluronic F-127 (ThermoFisher) for 5 minutes at 4 °C. eGFP and di-8-anepps were excited with the 488nm laser on a Nikon A1R confocal microscope. eGFP emission was collected a 450/50 bandpass filter, and di-8-anepps emission with a 595/50 bandpass filter. Comparison of the immunochemistry data between cocultures and monoculture controls was performed using Student’s t-test.

2.5. FACS, RNA extraction, and qPCR

CPCs differentiated from H9-hTnnTZ-pGZ-D2, a hESC line which expresses GFP under control of the CM-specific TNNT2 promoter,_[51] facilitated separation of CMs from EC cocultures. After using ACCUTASE_™ to singularize the cells from the two week cultures, cells were suspended in versene containing 1% FBS (ThermoFisher) and 5 μM Y-27632 and kept on ice. 300nM DAPI (ThermoFisher) was added to the flow buffer with the samples at least five minutes before flow activated cell sorting (FACS) to eliminate dead cells during sort. After gating out the eGFP- cells, dead cells, and doublets, using the forward and side scatter area, eGFP+ cells were sorted directly into lysis buffer using a BD FACSAria cytometer. Immediately following the sort, total RNA was extracted using the RNEasy Micro kit (Qiagen; Venlo, Netherlands). Cells from three separate differentiations and cocultures were used to account for variation in the differentiated cells. The RNA was then converted to cDNA using the RT SuperScript III First-Strand kit (Invitrogen; Carlsbad, CA). qPCR was carried out with the PowerUP SYBR green master mix (ThermoFisher) and run for 45 cycles with annealing temperatures between 58-63 °C on an AriaMx Real-Time PCR System (see Supplementary Table 2 for further information of primer sequences). To normalize for the amount of cDNA loaded, averaged CT values of VIRMA and ZNF384 were used as two housekeeping genes. Relative gene expression was calculated using the ΔΔCt method, with normalization to the monoculture control for each differentiation. Significance was determined via the Kruskal-Wallis test with Dunn’s post hoc analysis between cocultures and controls.

2.6. Multielectrode array analysis

To obtain multielectrode array (MEA) recordings, monocultures and coculture were initiated on glass MEA plates coated with poly-D-lysine (Sigma) and Matrigel, with a total cell count of 160 × 10₃ cell/cm₂. After two weeks, initial recordings with the MEA were taken 11 for 100 seconds using the MEA2100 system (ALA Scientific; Farmingdale, NY) before the addition of 0.5 μL of 0.01 mM isoprenaline (Sigma). The cells were placed in the incubator for 15 minutes before an additional recording was taken. This was repeated with escalating isoprenaline doses (up to 500 μM) until recordings at all concentrations were made. Analysis of the MEA data was done with Student’s t-test.

3. Results

3.1. Coculture of hPSC-derived ECs and CPCs increased cardiac protein expression and cell size in the resulting CMs

During development ECs play crucial roles in the formation of cardiac tissue via many potential routes, including providing juxtacrine and paracrine factors such as Notch1, Fgf9, neuregulin-1, and nitric oxide._[43-46,52] ECs have also been shown to enhance maturation of hPSC-derived CMs in vitro._[36-38] The goal of this study was to determine whether hPSC-derived ECs affect maturation of hPSC-derived CMs when cocultured at different stages of differentiation, specifically comparing the effect of ECs on CPCs and early stage CMs. Since ECs are present at early stages of heart development, we hypothesized that hPSC-derived ECs would impact maturation phenotypes of hPSC-derived CPCs and CMs in vitro. To first test the impact of EC coculture on CPCs, hPSCs were differentiated to D6 CPCs via the GiWi protocol._[49] CPCs were frozen, with test wells maintained to ensure the capacity to differentiate to CMs. After culture for at least 4 additional days, these CPCs formed a population of spontaneously contracting cells with greater than 90% of cells expressing cTnT, demonstrating their high CM differentiation potential (Figure 1A). The CPCs expressed ISL1 and VEGR2, consistent with multipotent cardiac progenitors (Supplementary Figure 1A-B)._[53] Endothelial progenitor cells were differentiated from hPSCs using the protocol developed by Bao et al. and purified via magnetic activated cell sorting for CD34-expressing cells to obtain a progenitor population of at least 90% CD34+/CD31+ cells (Figure 1B)._[53] These progenitors were expanded in EGM-2 for at least five days, yielding nearly pure ECs that expressed membrane-localized CD31 and VE-cadherin along with cytoplasmic vWF but no longer expressed CD34 (Figure 1C-D). The ECs were then maintained for up to five passages in EGM-2. The two cell types were coplated at 160 × 10₃ cells/cm₂ on Matrigel-coated tissue culture plates with EC:CPC ratios of 1:3, 1:1, and 3:1, placing the two cell types in direct contact with each other to allow both juxtacrine and paracrine interactions (Figure 1E). A CPC monoculture also plated at 160 × 10₃ cells/cm₂ was used as a control to determine if the ECs enhance maturation of the resulting CMs. Brightfield images taken one day after plating the 1:1 EC:CPC coculture and the CPC monoculture control show cell attachment and formation of a confluent monolayers in both conditions (Supplementary Figure 1C). EGM-2 medium, changed daily, was used for both monoculture and coculture experiments. Other media, including RPMI/B27, caused the ECs to undergo an endothelial-to-mesenchymal transition to smooth muscle-like cells whereas the CMs survived and maintained their identity in EGM-2 medium (data not shown). After the two weeks in coculture, the percentage of VE-cadherin+ cells and cTnT+ cells in the culture was assessed via flow cytometry, and the final cell population approximately maintained the EC:CM cell ratios initially seeded, with the presence of small populations of other unidentified cell types (Figure 1F). Interestingly, 7.4 ± 1.0% VE-cadherin+ cells were present in the CPC monoculture control, suggesting that some endothelial progenitors or ECs exist in the CM differentiation. However, there were significantly fewer ECs in the monoculture control than the 24.7 ± 4.7% VE-cadherin-expressing cells in the 3:1 EC:CPC coculture (p<0.05).

Figure 1.

Cocultures of ECs and CPCs induced phenotypes in the resultant CMs that are associated with CM maturation. (A) H9 hESC-derived CPCs were differentiated to CMs via the GiWi protocol and the percentage of cells expressing cTnT was quantified by flow cytometry. (B) H1s and 19-9-11s were differentiated to EC progenitors by two days of CHIR99021 treatment followed by three days in culture, then purified by MACS for CD34 expression. CD34 and CD31 expression were determined by flow cytometry before (left) and after (right) MACS. (C-D) EC progenitors derived from 19-9-11 iPSCs were cultured in EGM-2 medium for five or more days, passaging when confluent. Before coculture, the cells were immunostained for (C) CD34 (green) and CD31 (red) or (D) vWF (green) and VE-cadherin (red) with DAPI (blue) to verify their identity prior to the coculture. (E) Schematic illustrating how hPSC-derived CPCs and ECs were cocultured for two weeks at a varying ratio with a constant 160k cells/well. (F) After two weeks, monoculture and coculture populations were analyzed for the percentage of cells expressing VE-cadherin and cTnT by flow cytometry. cTnT+ cells in the indicated monoculture and cocultures were analyzed via flow cytometry for (G) the percentage of cTnT+ cells expressing cTnI, (I) the median fluorescence intensity (MFI) of cTnT in cTnT+ cells, normalized to the monoculture, and (J) forward scatter of cTnT+ cells, normalized to the monoculture. (H) The cells in monoculture and cocultures were costained analyzed for expression of both MLC2a and MLC2v by flow cytometry and the percent of MLC2v+ cells were calculated from the total cells expressing either MLC2 isotype. For each differentiation, the MLC2v percentages were normalized to the monoculture control. Data represent mean ± SEM of three independent differentiations with at least two experimental replicates each. Comparison of percentage of cells expressing VE-cadherin, cTnT, MLC2a/v, and cTnI was performed using ANOVA with Tukey’s HSD post hoc analysis, N=3 (* p<0.05, ** p<0.01). cTnT MFI, forward scatter, and normalized %MLC2v values were normalized to the CPC monoculture in each differentiation, and comparisons were performed using the Kruskal-Wallis test with post hoc Dunn analysis, N≥8 (* p<0.05, ** p<0.005).

During heart development, CMs begin to express cTnI at the late fetal or early neonatal stage._[53] Induction of this protein has been elusive in hPSC-derived CMs and cTnI 13 expression has been suggested as a molecular benchmark for CM maturity._[54] To determine whether ECs induced cTnI expression in hPSC-derived CMs, we used flow cytometry to assess the co-expression of cTnI and cTnT+ cells after two weeks of coculture, and calculated the percent of cTnI+ cells within the cTnT+ cell population. With the addition of ECs, the percentage of cTnT+ CMs expressing cTnI+ increased from 18.4 ± 8.2% in the monoculture to 59.4 ± 4.1% and 64.7 ± 3.0% cells in the 1:1 and 3:1 EC:CM ratios (p<0.05) (Figure 1G). To verify the specificity of the cTnI antibody used in this study, three-month old CMs were immunostained to demonstrate expression and localization of cTnI to myofilaments. As shown in Supplementary Figure 1D-F, the cTnT and cTnI antibodies colocalized in a striated pattern throughout the CMs and the cTnI did not stain cTnT+ day 10 CMs nor undifferentiated hESCs.

Expression of MLC2v is induced during ventricular CM development._[11] The atrial MLC2 isoform, MLC2a, is expressed in embryonic ventricular CMs and is downregulated after induction of MLC2v._[56] At day 14, the monoculture CPC control contained 15.3 ± 4.4% MLC2v+ CMs while the 1:1 and 3:1 EC:CPC cultures contained 29.9 ± 5.5% and 28.9 ± 4.9% MLC2v+ cells respectively (Supplementary Figure 1G). When the MLC2v percentages were normalized to the monoculture control in each condition, the 1:1 and 3:1 EC:CPC cultures contained a greater percentage of MLC2v+ CM populations by 2.16 and 2.04-fold respectively (p<0.005) as shown in Figure 1H, indicating that EC coculture accelerated induction of MLC2v. While cTnT is expressed early in CM differentiation, the expression level increases throughout development and thus may be used as another marker of CM maturation._[57-59] To further investigate the influence of ECs on CM maturation, the degree of cTnT expression was measured in the cTnT+ CMs by the median fluorescence intensity (MFI) via flow cytometry. cTnT MFI was 1.9 ± 0.95-fold greater (p<0.001) in CMs from 3:1 EC:CPC samples compared to CMs in the monoculture control (Figure 1I). Additionally the size of CMs is known to increase during development._[16] Via assessment of forward scatter by flow cytometry, which increases with cell volume, the size of resulting CMs was seen to increase significantly after coculture with ECs (Figure 1I-J). Overall, the addition of ECs to CPC cultures increased CM size and induced expression of structural markers of maturation. The 3:1 EC:CM ratio resulted in the greatest percentage of cells expressing cTnI and MLC2v, the highest cTnT expression level, and the largest CMs.

3.2. CM-specific protein expression induced by coculture with ECs is dependent on the stage of CM development at which coculture is initiated

In EC cocultures, we used either early beating hPSC-derived CMs, obtained between days 13 and 18 of the differentiation, or pre-CM D6 CPCs as shown in Figure 2A. To generate CMs, hPSC-derived CPCs were maintained in RPMI/B27 medium for ~9 additional days. When coculture was initiated, the cells were plated at a density of 160 × 10₃ cells/cm₂ in EGM-2 medium. The cells were plated as a monoculture or in a 1:1 ratio of ECs to the CPCs or CMs. Cells were then maintained in monoculture or coculture for two weeks before analysis (Figure 2B). After two weeks the monocultures contained approximately a 1:2 EC:CM ratio whereas the 1:1 seeded cocultures contained approximately 2:1 EC:CM, perhaps as a result of greater proliferation of ECs than CMs. The EC:CPC coculture contained a significantly higher percentage of VE-cadherin+ cells and a lower percentage of cTnT+ cells compared to the monoculture control (Figure 2C; p<0.05), consistent with observations in Figure 1. Similarly, the EC:CM coculture also contained a significantly higher percentage of VE-cadherin+ cells compared to the monoculture control (p<0.005). Immunostaining for VE-cadherin and MLC2a after two weeks of coculture of the CPC, EC:CPC, CM, and EC:CM samples demonstrated the presence of ECs and CMs in the wells (Supplementary Figure 2A). In all conditions, the ECs did not display any notable organization but were evenly dispersed throughout the plate. The CPC and CM monocultures were comparable in the final EC percentage and organization, as were both EC coculture conditions and are therefore are suitable for comparison of the impact of the incorporation of the ECs into the cultures on the resulting CMs.

Figure 2.

Comparison of induction of CM maturation properties when hPSC-derived ECs were cocultured with hPSC-derived CPCs and CMs. (A) Schematic illustrating how hPSC- derived ECs were cocultured with either CPCs or CMs. (B) The timeline in which the EC cocultures were induced and maintained in EGM-2 for two weeks before analysis and compared to their time-matched monoculture controls. (C) After two weeks in culture, the indicated monoculture and coculture samples were analyzed for the percentage of cells expressing cTnT and VE-cadherin by flow cytometry. (D) After two weeks in culture the percentage of cTnT+ cells expressing cTnI was determined by flow cytometry. Representative flow cytometry histograms are shown for (E) the cTnT+ cells in the CPC monoculture (red) and the EC:CPC coculture (green) and the cTnT- cells in the EC:CPC (blue), and (F) the cTnT+ cells in the CM monoculture (red) and the EC:CM coculture (green) and the cTnT- cells in the EC:CPC (blue). The numbers indicate the percent of cTnI+ cells in the corresponding samples. (G) Cells in the cocultures were costained and analyzed for expression of both MLC2a and MLC2v by flow cytometry, and the percent of each were calculated based on the total number cells expressing either MLC2v or MLC2a. (H) The median fluorescence intensity of cTnT in cTnT+ cells was quantified by flow cytometry. Data represent mean ± SEM of three independent differentiations with at least two experimental replicates each. Comparisons of VE-cadherin, cTnT, and MLC2a/v expression between cocultures and monocultures were performed using Student t-test, N=3 (* p<0.05, ** p<0.01). Comparisons of percentage of cells expressing cTnI were performed using one-way ANOVA with Tukey’s HSD post hoc analysis, N=3 (* p<0.05, ** p<0.01). cTnT MFI values were normalized to the monoculture and statistical comparisons were performed using the two-way Mann-Whitney test, N≥8 (* p<0.05, ** p<0.005).

Next, we determined if the stages of CM differentiation (CPCs vs. beating CMs) impacts the ability of ECs to induce the expression of key proteins indicative of CM maturation. To compare induction of CM maturation via the switch in troponin I expression to the cardiac isoform, the percentage of cTnI+ CMs was examined after two weeks in coculture. A significantly greater percentage of cTnT+ cells in the EC:CPC coculture expressed cTnI (65.5 ± 6.4%) as compared to the monoculture control (23.5 ± 4.0%, p<0.05) (Figure 2D-F). The percentage of cTnI+ cells within the cTnT+ population was not significantly different between the CM monoculture and EC:CM conditions (67.6 ± 9.1% and 69.8 ± 8.5%). This demonstrates that ECs may accelerate the rate at which cTnI expression is initiated when coculture is initiated prior to CM commitment. To see if this acceleration of cTnI expression is representative of other markers of CM maturation, we also assessed the expression of MLC2v in the CMs in the CPC, EC:CPC, CM, and EC:CM conditions after two week in culture. Consistent with results shown in Figure 1, initiating EC coculture with the CPCs led to a significant increase in the percentage of CMs that were MLC2v+ compared to the monoculture control (p<0.005) (Figure 2G). However in the CM and CM:EC samples, MLC2v expression was not affected by EC addition. This supports the hypothesis that the CM differentiation stage at which the interaction with ECs is incorporated affects the ability of ECs to accelerate the induction of CM maturation. We next measured the expression level of cTnT in the cTnT+ cells to determine if EC coculture also increases cTnT expression when coculture is initiated with CMs. The MFI of cTnT demonstrated a significant increase in the cTnT expression in the CPC coculture compared to its control, but not in the CM coculture (Figure 2H, Supplementary Figure 2B). Overall, ECs accelerated production of structural markers of CM maturation when coculture was initiated while the cardiac cells were in their progenitor state but not with already beating CMs.

3.3. EC coculture with CPCs induced changes in CM morphology

Early hPSC-derived CMs are small, round cells._[60] In extended culture they grow and elongate over time, but they do not become as large or rod-shaped as adult CMs._[16-17] To assess whether EC coculture with CPCs or CMs affects the resultant CM cell size, we first measured forward scatter by flow cytometry. CMs from the EC:CPC coculture forward-scattered light significantly more than CMs differentiated in monoculture (Figure 3A), consistent with a larger cell volume. To directly measure CM size, cells were replated after two weeks in monoculture or coculture and immunostained for cTnT. Image J was used to quantify the cell area, perimeter, and circularity of cTnT+ cells (Supplementary Figure 3A). Consistent with the forward scatter results, CMs from the EC:CPC coculture had a 33 ± 13% greater cell area than CPCs differentiated in monoculture (Figure 3B; p<0.05). The cTnT+ cells in the EC:CPC cocultures displayed a 6.1 ± 3.1% greater circularity than the CMs from a monoculture control (p<0.05). This increase in circularity is inconsistent with increased structural maturation, which was interesting considering the ECs induced other structural markers of CM maturation. Both the EC:CPC and EC:CM cultures exhibited a greater perimeter than CMs in monoculture control of 10% (± 5% and 3% respectively) which could signify CM elongation. The raw values without normalization in each differentiation are provided in Supplementary Figure 3B-D.

Figure 3.

Cocultures of hPSC-derived ECs with CPCs, but not CMs, induced a more adult-like phenotype in the CMs after two weeks in culture. (A) The forward scatter of cTnT+ cells in the indicated monoculture and coculture samples was measured by flow cytometry after two weeks in culture. Data represent mean ± SEM of three independent differentiations with at least two experimental replicates each. Forward scatter values were normalized to the CPC monoculture in each differentiation and comparisons were performed using the two-way Mann-Whitney test, N≥8 (** p<0.005). (B) CM area, CM perimeter, and CM circularity of cTnT+ cells in the indicated monoculture and coculture samples were quantified by analyzing at least 40 cells per experimental replicate using ImageJ after replating two week samples onto gelatin-coated plates and culturing for 3 days. Values were normalized to the monoculture for each differentiation. Data represent mean ± SD of three independent differentiations with three experimental replicates each. Comparisons between corresponding cocultures and monocultures were performed using Student t-test, N=3 (* p<0.05). (C) H9 hESC-derived CPCs and CMs with or without hPSC-derived ECs were cultured for two weeks before replating on Matrigel-covered glass slides and immunostained for α-actinin (green) and phalloidin (red) with DAPI (blue) to compare sarcomere organization. (D) Blinded images of sarcomeres stained as described in panel (C) were ranked for their organization based on the following criteria: 1, no visible sarcomere structure; 2, some sarcomere organization; 3, H zones and Z lines visible in areas with some sarcomere organization; 4, near perfect sarcomeres with clear H zones, Z lines, and thick myofibrils. The percentage of CMs containing each rank in the indicated monoculture and coculture samples is plotted. (E) After replating the CPC, EC:CPC, CM, and CM:EC cultures after the two weeks in culture, cells were also stained for Cx43 (green), α-actinin (red), and DAPI (blue) to assess gap junction formation and localization. (F-G) hESCs expressing eGFP under the TNNT2 promoter were differentiated to CPCs and CMs and cultured in the indicated monocultures or with hPSC-derived ECs for two weeks. The cells were then replated on glass slides and were then stained with di-8-anepps to identify membranous extensions into the intracellular area of the CMs. (F) A high magnification image of an eGFP+ (green) CM is shown to illustrate the faintly striated di-8-anepps stain (red). (G) The percentage of the cell area, determined by eGFP, stained by di-8-anepps was quantified in ImageJ. Values were normalized to the monoculture for each differentiation (1.59% and 2.23% for the CPC and CM monocultures, respectively). Data represent mean ± SD of three independent differentiations for the CM and EC:CM data and two independent differentiations for the CPC and EC:CPC data. Cocultures were compared to monocultures using Student’s t-test (* p<0.05).

Additionally, adult CMs have organized, aligned sarcomeres and myofibrils in addition to gap junctions localized to the intercalated discs, which together coordinate the direction and timing of contractions between CMs. The organization and alignment of sarcomeres within the CMs were analyzed for the presence of H zones and Z lines, and sarcomere length was measured on the replated cells by α-actinin immunofluorescence and phalloidin staining for F-actin. No significant differences in the organization of the sarcomeres and myofibrils were seen (Figure 3C-D). To examine differences in sarcomere structure induced by EC coculture, at least 75 cells in each condition over three separate differentiations were analyzed for their organization (examples in Supplementary Figure 3E). In all conditions, CMs exhibited a wide variety of sarcomeric structures, demonstrating some of the CMs in all conditions had developed to contain organized sarcomeres and myofibrils. Also during heart development, sarcomere spacing increases to ~2.2 μm from 1.6 μm._[61] For every cell in which at least 11 Z lines in a row were visible, sarcomere length was measured. There was no significant change seen in the sarcomere length between the CPC, EC:CPC, CM, or CM:EC cultures. The average lengths ranged between 1.58 μm and 1.79 μm, indicating immature sarcomere spacing in all conditions (Supplementary Figure 3F). Finally, gap junction localization was visualized in CMs by immunostaining for Cx43 and α-actinin (Figure 3E). We did not see a difference in Cx43 localization between the CPC, EC:CPC, CM, and EC:CM cultures. Overall, culturing ECs with CPCs yielded larger CMs compared to the CPC culture without ECs, though no additional structural organization was observed and the coculture seemed to decrease the amount of elongation in the CMs. In contrast, ECs had very little effect when cocultured with beating CMs, leading to a small increase in CM perimeter but no significant change in area or circularity.

Along with organized sarcomeres and localized gap junctions, t-tubules are another structural feature of adult CMs. In rats, formation of t-tubules begins after birth and are fully developed after a month._[57] However, in hPSC-derived CMs, t-tubules are rarely seen, and this lack has been suggested to contribute to irregular Ca2+ handling in hPSC-derived CMs._[16,31] CMs and CPCs were generated from the H9-hTnnTZ-pGZ-D2, an hESC line that expresses eGFP under the TNNT2 promoter, and used in monoculture and coculture configurations with hPSC-derived ECs as illustrated in Figures 2A-B. Live cells were stained with di-8-anepps (Figure 3F), a lipophilic membrane dye that fluoresces upon changes in the membrane potential and can be used to identify the CM membrane and extensions of the membrane into the cell._[64] Over two separate differentiations, 43 eGFP+ cells in the CPC monoculture and 51 in the EC:CPC coculture were imaged. For the CM cultures, 71 eGFP+ cells were imaged in the monoculture and 78 in the EC-CM coculture over three differentiations. The percentage of the cell area stained with di-8-anepps was quantified for each CM analyzed (Supplementary Figure 3H). After normalizing to the monoculture control within each differentiation, we found that significantly more intracellular areas were stained by di-8-anepps in the EC:CPC coculture compared to the CPC monoculture (Figure 3G). Alternatively, the EC:CM coculture did not show any difference in di-8-anepps staining compared to the CM monoculture. The di-8-anepps staining demonstrates that EC coculture can induce the formation of membrane extensions in CMs only when the culture is instigated early in the differentiation. These extensions may accelerate t-tubule formation although bona fide t-tubules were not observed in any of the experimental conditions.

3.4. Upregulation of CM-specific genes was induced by the coculture of EC with CPCs

Genes encoding sarcomeric proteins, ion channels, and regulators of said channels are upregulated during CM differentiation and maturation._[29] By analyzing relative mRNA abundance between the CPC and EC:CPC cultures, we further addressed whether EC coculture induces CM maturation at the early stage of differentiation. We cultured CPCs or CMs derived from H9-hTnnTZ-pGZ-D2 hESCs with hPSC-derived ECs for two weeks. We then used FACS to collect eGFP+ CMs for RNA extraction, using at least two experimental replicates across three differentiations, and analyzed gene expression via qPCR in Figure 4 for the CPCs and Supplementary Figure 4 for the CMs. MYL2 expression was significantly upregulated in the EC:CPC coculture sample (p<0.01), consistent with the increase in MLC2v protein levels in the EC:CPC coculture shown in Figures 1H and 2G. A significant decrease (p<0.05) in TNNI1 (ssTnI) and an increase in TNNI3 (cTnI) in the EC:CPC coculture is also consistent with increased cTnI expression in this sample. Myosin heavy chain isoform switches from predominantly α-MHC (MYH6) to β-MHC (MYH7) during human ventricular CM development._[63] A decrease in MYH6 and a significant increase (p<0.01) in MYH7 provide additional evidence for accelerated CM maturation in the EC:CPC coculture compared to the CPC monoculture. Interestingly an increase in GJA1 (Cx43) expression was also observed in the coculture sample (p<0.05), even though no any changes in gap junction organization were observed (Figure 3F). Finally, CMs from the EC:CPC coculture exhibited higher expression of KCNJ2 and ATP2A2, genes encoding a potassium-gated channel and an intercellular ATP pump, respectively. Alternatively, the ECs only significantly increased KCNJ2 expression in the EC:CM coculture in comparison to the CM control as shown in Supplementary Figure 4. Summarizing this gene expression data, coculture of ECs with only CPCs resulted in upregulation of several genes involved in CM sarcomere formation and electromechanical coupling.

Figure 4.

CMs from EC:CPC cocultures express genes associated with CM maturation. hESCs expressing eGFP under the TNNT2 promoter were differentiated to CPCs, and maintained as CPC monocultures or cocultured with hPSC-derived ECs for two weeks. Then, eGFP+ cells were purified via FACS and expression of the indicated genes was quantified by qPCR. Values were normalized to the average of the expression of VIRMA and ZNF384, two housekeeping genes. Fold change values were calculated via the ΔΔCt method. Data represent mean ± SEM of at least six replicates taken from three independent differentiations with at least two qPCR experimental replicates each. The values were normalized to the CPC monoculture in each differentiation and statistical comparisons were performed using the two-way Mann-Whitney test, N≥6 (* p<0.05, # p<0.01).

3.5. ECs promoted a CM chronotropic response from a β-adrenoreceptor agonist

hPSC-derived CMs are less electromechanically mature than adult CMs, as seen by lower upstroke and conduction velocities, reduced excitation-contraction coupling, and a higher resting membrane potential._[13,16,64] These are in part due to their lower expression of specific ion channels and regulators along with the expression of different ion channel isoforms._[65] Also, hPSC-derived CMs express lower levels of β-adrenoreceptors than adult CMs._[66] Isoprenaline, a β-adrenoceptor agonist, generates a positive chronotropic response in CMs._[67] The extent of this response has been shown to be affected by CM maturation, with 10 day hPSC-derived CMs having smaller increase in beating rate when exposed to isoprenaline in comparison to 80 day CMs._[68] We initiated culture of CPC, EC:CPC, CM, and EC:CM on the MEAs. After two weeks, we measured changes in the time between each beat (RR interval) upon exposure to a range (0-500 nM) of isoprenaline concentrations. Example MEA recordings are shown for each sample with and without 500 nM isoprenaline in Supplementary Figure 5A-D. As shown in Figure 5A, there was no significant decrease in RR interval in EC:CPC and monoculture samples after isoprenaline treatment, although there was substantial variability between experimental repeats. Alternatively, the EC:CM coculture required only 50 nM isoprenaline to result in a statistically significant decrease in RR interval (p<0.01), compared to 500 nM for the CM monoculture (Figure 5B). Incorporation of ECs into only the CM cultures significantly decreased the RR interval, suggesting that ECs enhanced the positive chronotropic response of CMs when cocultured with beating CMs.

Figure 5.

EC coculture increased CM response to isoprenaline in CMs. Monocultures and cocultures were maintained on Matrigel-coated multielectrode arrays for two weeks. The RR interval was measured for (A) H9 hESC-derived CPC monocultures or cocultures with H1 hESC-derived ECs and (B) H9 hESC-derived CMs monocultures or cocultures with H1 hESC-derived ECs as a function of isoprenaline concentration. Recordings were first acquired for the control condition lacking isoprenaline, then increasing doses of isoprenaline were added for 15 min before each recording. Data for each electrode were normalized to the untreated control. Data represent mean ± SD of RR intervals for two biological replicates. Comparisons between cocultures and monocultures were performed using Student t-test, N=2 (* p<0.05).

4. Discussion

To determine the ability of ECs to impact CM maturation when coculture is initiated with cardiac progenitors, we seeded ECs and CPCs together in the ratios of 1:3, 1:1, and 3:1 along with a CPC monoculture. We observed an increase in the percentage of CMs that expressed MLC2v and cTnI, the expression of cTnT per cell and the size of CMs from the EC:CPC coculture compared to CMs from CPC monoculture. The extent of CM maturation depended upon the ratio of ECs to CPCs, with the greatest induction of maturation observed at a 3:1 EC:CPC ratio. Congruously, this ratio is similar to the ~2:1 ratio in which ECs and CMs are found in adult hearts._[69] CPC:EC coculture enhanced the expression of MYL2, TNNI1, MYH7, GJA1, KCNJ2, and ATP2A2, which indicates CM maturation. Alternatively, we did not see an induction of any of these genes except for KCNJ2 when the ECs coculture was initiated with beating CMs. We did not detect an increase Cx43 localization to longitudinal termini of the cells, even though the expression of the gene was upregulated. There was a slight, but not statistically significant, increase in the sarcomere organization when CPCs were cocultured with ECs.

Next, to determine if the induction of maturation when ECs were cocultured with CPCs was unique to the stage of the cells in the CM differentiation at which coculture was initiated, we compared the effects of coculturing the ECs with either young, beating CMs or with CPCs. When the coculture was initiated with beating CMs, we did not see a significant impact on cell size or expression of cTnT, cTnI, and MLC2v. This lack of cTnI or cTnT induction is consistent with the lack of TNNI3 or TNNT2 induction reported by Giacomelli et al. when they cocultured hPSC-derived CMs in spheroids with or without ECs._[38] Additionally the presence of membranous extensions, which might be t-tubule precursors, was enhanced by EC coculture only when the coculture was initiated with CPCs, but not CMs. However, ECs increased sensitivity of CMs to the β-adrenoreceptor agonist, isoprenaline, when coculture was initiated with only beating CMs. Similar enhancement in isoprenaline sensitivity was observed by Giacomelli et al. in EC:CM spheroids and by Ravenscroft et al. in hESC-derived CMs with coculture of both human primary cardiac fibroblasts and ECs._[38,41]

Based on our results, it is apparent that inducing EC coculture at the CPC stage resulted in greater acceleration of maturation than initiating EC coculture with beating CMs. This is consistent with a study performed by Ronaldson-Bouchard et al., where they found that the stage at which electrical stimulation is applied to hPSC-derived CMs impacted the extent of maturation achieved._[23] Electrical pacing of day 12 CMs induced greater sarcomere organization, upregulation of genes expressed in adult CMs, and enhanced electrophysiological maturation compared to pacing day 28 CMs. The induction of isoprenaline sensitivity in the EC:CM coculture may also point to the importance for EC-derived signals on the CMs between days 20 and 29 of differentiation.

It is possible that the tissue specificity of ECs may be vital to induce the hPSC-derived CM development or maturation. In Ravenscroft’s coculture system, human primary dermal fibroblasts and ECs were found not to enhance the response to drugs in comparison their cardiac-specific counterparts._[41] When Nolan et al. engrafted generic, stem cell-derived ECs into mouse kidneys and livers, they found that the ECs acquired the expression of markers specific to the tissue in which the cells were engrafted._[70] The hPSC-derived ECs used in this study arise from a mesoderm progenitor but lack tissue and vessel-type specificity, which may result from developmental immaturity._[71] It would be interesting to determine whether cardiac specificity is induced in these hPSC-derived ECs cocultured with hPSC-derived CPCs or CMs and if this specification has an impact on the observed CM maturation.

We observed the appearance of small populations of ECs in the CPC and CM monoculture controls. This may be attributed to the fact that ECs and CPCs arise from the same mesodermal population as CPCs, or the differentiation of ISL1+ CPCs to ECs._[72] Indeed, ECs have been found as a byproduct of small molecule-induced CM differentiation._[73-75] It is possible that our use of EGM-2, an EC growth medium, for our cocultures selectively expanded the EC population that arose during CM differentiation. Our data suggest that these “contaminating” ECs may in fact be beneficial to CM maturation.

Our results suggest the potential of ECs to influence CM development and maturation in a developmental stage-specific manner. For example, the endocardium influences myocardial development through pathways such as FGF signaling, neuregulin, and NOTCH signaling._[43-46] In addition to these paracrine signals, ECs have been shown to induce partial maturation via juxtacrine signaling through microRNA transfer via gap junctions._[36] It is yet unclear which of these pathways and other yet unidentified mechanisms, also including biomechanical signaling and ECM remodeling in addition to juxtacrine and paracrine signaling, may mediate CM maturation caused by EC coculture with CPCs. Additionally, incorporating EC:CPC interactions during CM manufacturing may accelerate production of more mature cells. The maturation phenotypes seen in this study included increased expression of some CM proteins and RNA along with some structural organization and sensitivity to isoprenaline, demonstrating that full maturation of the cocultured CMs has not been achieved. It is likely that other signals will be necessary to induce full maturation of hPSC-derived CMs, such as the incorporation of other cell types or the design of a novel ECM in which to culture the CPCs. Further research should investigate the mechanisms by which the cocultured ECs induce hPSC-derived CM maturation. Identification of the biochemical and biophysical interactions between the ECs and CPCs that induce CM maturation may facilitate efforts to manufacture mature CMs without EC coculture.

In conclusion, coculturing hPSC-derived ECs and CPCs accelerated acquisition of CM maturation properties, including gene and protein expression, cell size, and development of t-tubule-like membrane extensions. This induction of maturation required initiation of EC coculture at the CPC differentiation stage, before spontaneous contraction was initiated.

Abbreviations:

CPC

cardiac progenitor cell

CM

cardiomyocyte

cTnT

cardiac Troponin T

cTnI

cardiac Troponin I

EC

endothelial cell

FACS

flow activated cell sorting

hESC

human embryonic cell

hPSC

human pluripotent stem cell

iPSC

induced pluripotent cell

MACS

magnetic activated cell sorting

MFI

median fluorescent intensity

MHC

myosin heavy chain

MEA

multielectrode array

MLC2a

myosin light chain 2a

MLC2v

myosin light chain 2v

ssTnI

slow skeletal Troponin I