Specific induction and long-term maintenance of high purity ventricular cardiomyocytes from human induced pluripotent stem cells

Currently, cardiomyocyte (CM) differentiation methods require a purification step after CM induction to ensure the high purity of the cell population. Here we show an improved human CM differentiation protocol with which high-purity ventricular-type CMs can be obtained and maintained without any CM purification process. We induced and collected a mesodermal cell population (platelet-derived growth factor receptor-α (PDGFRα)-positive cells) that can respond to CM differentiation cues, and then stimulated CM differentiation by means of Wnt inhibition. This method reproducibly generated CMs with purities above 95% in several human pluripotent stem cell lines. Furthermore, these CM populations were maintained in culture at such high purity without any further CM purification step for over 200 days. The majority of these CMs (>95%) exhibited a ventricular-like phenotype with a tendency to structural and electrophysiological maturation, including T-tubule-like structure formation and the ability to respond to QT prolongation drugs. This is a simple and valuable method to stably generate CM populations suitable for cardiac toxicology testing, disease modeling and regenerative medicine.


Introduction
The generation of cardiomyocytes (CMs) from human pluripotent stem cells (hPSCs), including human embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs), has been increasingly explored for various applications, such as cardiotoxicity screening, drug discovery, disease modeling, as well as regenerative medicine [1][2][3]. The generation of pure and stable human CM populations is a fundamental requirement to meet the wide CM demand. Similarly, developing methods for the long-term maintenance of high-purity, structurally and electrophysiologically mature ventricular cardiomyocytes is essential to improve cardiac toxicological and pharmacological studies. Although recent advances in CM differentiation protocols robustly and efficiently yield CMs, most available protocols still need some kind of purification step at the final stage of CM differentiation, such as the Percoll density gradient procedure [4,5], genetic manipulation using cardiacspecific promoters [6][7][8], cell sorting using mitochondrial dyes [9], microRNA-regulated fluorescence [10], or antibodies directed against cardiac cell surface markers such as signal-regulatory protein alpha (SIRPA) [11] and vascular cell adhesion molecule 1 (VCAM1) [11,12], or metabolic selection using glucose-depleted culture medium containing lactate [13]. Even after CM purification, the resulting purity can be reduced during long-term cultures that span several months [14]. This is likely because CM purification after terminal differentiation does not fully guarantee the abscence of other cell types. In addition, the purification process might cause some mechanical and physiological stress to the cells that may reduce the quality and viability of the purified CM population. Removal of non-suitable cells at the earlier stages of the CM differentiation process might be effective for the long-term maintenance of high-purity CM populations.
We have previously reported methods both for efficient CM differentiation and for the simultaneous differentiation of several cardiovascular cell types [12,15] based on a high-density hESC monolayer culture [5]. After mesoderm was induced by treating the cells with Activin A (ActA) for 24 hours, followed by bone morphogenetic protein 4 (BMP4) and basic fibroblast growth factor (bFGF) for 4 days, Wnt signal inhibitors such as Dickkopf-related protein 1 (Dkk1) were used to efficiently induce CM commitment [12,16,17]. Vascular endothelial growth factor (VEGF) was added instead of Wnt inhibitors for the simultaneous induction of endothelial cells (EC), pericytes and CMs [15]. In addition to these methods for CM differentiation, we previously developed a robust and efficient differentiation method for the induction of ECs from hiPSCs with extremely high efficiency (>99%). This method involves the purification of VEGF receptor-2 (VEGFR2)-positive (VEGFR2 + ) cells, which are responder cells to EC commitment signals, to eliminate non-responder cells at the early mesoderm stage (stimulation-elimination [SE] method) [18].
The highly specific induction of ECs by the SE method prompted us to examine and apply the SE approach to CM differentiation. In this study, we selected platelet-derived growth factor receptor-α (PDGFRα) as a marker for cells susceptible to CM differentiation cues. PDGFRα is highly expressed in paraxial mesoderm [19] and involved in cardiovascular tissue development in mouse embryo [20]. PDGFRα has often been reported as a marker for cardiac mesoderm and cardiovascular progenitors that give rise to CMs during both mouse and human PSC differentiation [21][22][23][24][25][26]. We recently reported that PDGFRα + cells derived from hiPSCs possess high CM differentiation potential [16] Thus, we induced and purified PDGFRα + mesoderm cells at differentiation day 5 (d5) and re-cultured them in medium containing Wnt inhibitors (XAV939 and IWP4) to prompt specific commitment to the CM lineage. With this approach, we succeeded in establishing a method for high-purity CM differentiation (>95%) without any CM purification steps, as well as ensuring the long-term maintenance (more than 200 days) of such high-purity CM population. These CMs showed a predominantly ventricular phenotype and a tendency towards structural and electrophysiological maturation during long-term maintenance. We further demonstrated the utility of these long term-cultured CMs for pharmacological studies.

Human iPSC lines
The 836B3 hiPSC line, established from an adult fibroblast cell line transfected with 6 factors (Oct3/4, Sox2, Klf4, L-Myc, LIN28, and Glis1) in episomal plasmid vectors [27], was used as a representative human iPSC line in all experiments unless stated otherwise. The 253G1 hiPSC line, established from skin fibroblasts following the retroviral transfection of 3-factors (Oct3/4, Sox2 and Klf4) [28], and the 1201C1 hiPSC line, established from an adult peripheral blood cell line transfected with 6 factors (Oct3/4, Sox2, Klf4, L-Myc, LIN28, and shRNA for TP53) in episomal plasmid vectors [29], were utilized to study the induction of CMs and the maintenance of CM purity. All hiPSC lines were originally established at and provided by the Center for iPS Cell Research and Application (CiRA), Kyoto University, Kyoto, Japan.

Long-term culture of CMs
On d21, CMs were detached using 0.25% Trypsin-EDTA for 10-15min at 37˚C, and reseeded on a culture dish at 2.0 × 10 5 cells/cm 2 with alpha minimum essential medium (αMEM; #11900-024, Thermo Fisher Scientific) supplemented with 2.2 g/L NaHCO 3 , 10% FBS and 5 × 10 −5 M 2-mercaptoethanol and containing Y-27632 (10 μM) and cultured for 2 days. Culture dishes were prepared by pre-coating them with Poly-L-lysine (5 μg/mL, Sigma P1399) at room temperature for 20 min. After thoroughly rinsing with PBS(-), dishes were coated with 0.5% gelatin (Sigma) or matrigel (Growth factor reduced, 1:60 dilution; Corning) overnight at 4˚C. After two days of cell incubation, culture medium was replaced by RPMI1640 with B27 supplement and refreshed every 2-3 days. Cells were passaged every 20 days. To assess CM proliferation, CMs were detached and collected after treatment with 0.25% trypsin/EDTA (#25200, Thermo Fisher Scientific), and then viable CMs were counted using a hemocytometer after 0.4% trypan blue staining (#15250061, Thermo Fisher Scientific). CMs on d21 were seeded at a 2.0 × 10 5 cells/cm 2 density (1.9 × 10 6 cells per well of a 6-well plate), and the accumulated cell number was calculated from counting viable CM numbers at each cell passage.

Quantitative Polymerase Chain Reaction (qPCR)
Total RNA was extracted from d21 and d91 CMs using the RNeasy Mini Kit (#74104, Quiagen, Hilden, Germany) according to the manufacturer's instructions. RPS18 was used to normalize gene expressions. Quantitative PCR was performed using the Power SYBR Green PCR Master mix (#4367659, Thermo Fisher Scientific) on the StepOnePlus system (Thermo Fisher Scientific) and with the Delta Delta Ct method. Forward and reverse primer sequences are shown in S1 Table.

Electrophysiology
We prepared CMs on d21 and d91. For single cell patch-clamp analysis, CMs were dissociated using 0.25% Trypsin in EDTA and re-seeded on fibronectin-coated coverslips in RPMI 1640 medium with B27 supplement and 10 μM Y27632. Patch clamp recordings were performed 4-6 days after plating. Coverslips were transferred to a patch clamp recording chamber, where spontaneous beating cells were subjected to whole-cell patch clamp at 35-37˚C for this study. Electrophysiological measurements were carried out using the Axopatch200B amplifier and the Digidata 1440A interface (Molecular Devices, CA). The physiological bathing solution contained 2.0 mM CaCl 2 , 1.0 mM MgCl 2 ・6H 2 O, 0.28 mM MgSO 4 , 5.0 mM KCl, 0.22 mM KH 2 PO 4 , 27.0 mM NaHCO 3 , 120 mM NaCl, 0.84 mM Na 2 HPO 4 , 5.55 mM D-Glucose; with pH adjusted to 7.4 by the addition of NaOH. Patch pipettes were pulled, fire-polished to final tip resistances of 2 to 4 mO, and then filled with a pipette solution containing 130 mM KOH, 130 mM L-Aspartic acid, 20 mM KCl, 1 mM MgCl 2 -6H 2 O, 5 mM NaCl, 10 mM EGTA, 5 mM Mg-ATP, 10 mM HEPES; pH was adjusted to 7.2 with KOH. Data was analyzed using the pClamp software (Molecular Devices). Action potential (AP) characterization was calculated from an average of 10 subsequent APs. The criteria to classify the APs into ventricular-like, atrial-like and nodal-like have been defined in a previous report [16,30]. Non-ventricular CMs (APD30-40 / APD70-80 ratio: < 1.5) were subsequently classified into nodal-like CMs (dV/dt max < 10 mV/ms) and atrial-like CMs (dV/dt max � 10 mV/ms). To further distinguish between early and late ventricular-like CMs (APD30-40 / APD70-80 ratio: � 1.5), we set a broader value at 30 mV/ms, which was the mean + 2SD of dV/dt max for d21 CMs.

Multi-electrode array experiment
For demonstrating the pharmacological utility of long term-cultured CMs, we recorded extracellular field potentials (FPs) in multielectrode arrays using the MED system (Alpha MED Scientific, Osaka, Japan) and evaluated cellular drug responses. The recording areas of multielectrode probes (MED64 probe: MED-P515A; Alpha Med Scientific, Osaka, Japan) were coated with 2 μL of 50 μg/mL fibronectin (BD) and incubated at 4˚C overnight. hiPSC-derived CMs were then dissociated using 0.25% Trypsin-EDTA for 10 mins at 37˚C, re-plated onto the multi-electrode probes at a density of 3.0 × 10 4 cells in 2 μL of αMEM+10% FBS containing 10 μM Y27632, and incubated for 3 to 5 h at 37˚C under a 5% CO2 atmosphere. Each singlewell MED64 probe was then filled with 1-2 mL of αMEM+10% FBS containing 10 μM Y27632, and cells cultured for 2-3 days. Medium was then completely replaced by RPMI1640 with B27 supplement and, after that, half of the medium was changed every 2 days. 3 days after the initial placement, spontaneous EFPs were recorded in each microelectrode.
We measured FPs according to a previous report, with some modifications [31]. In brief, samples were equilibrated for at least 30 min in a CO 2 incubator in 2 mL of fresh medium prior to the measurements. After equilibration, the MED probes were maintained at 37˚C with thermo-control systems and covered with a lid through which aeration (O 2 :CO 2 :N 2 = 20%: 5%: 75%) was administered. FPs from spontaneously beating samples were filtered with a vessel 1-1000 Hz band-pass filter using the MED64 System. FP duration (FPD) was defined as the interval from the first peak (depolarization) to the second peak (repolarization). After recording the basal state, 2 μL of dimethyl sulfoxide (DMSO; Wako) were added and FPs were recorded for 10 min. Then, E-4031 (Wako) or Chromanol 293B (Sigma-Aldrich) were cumulatively added to obtain the target concentrations, and FPs were recorded in the same manner as for DMSO administration. For E-4031, four drug concentrations (0.3 nM, 3 nM, 10 nM, 30 nM) were selected to evaluate dose-dependent effects, while for Chromanol 293B, three drug concentrations (3 μM, 10 μM, 30 μM) were selected. At each concentration, the FP was recorded for 10 min, and the last 30 beats recorded by the electrode with the most evident and easy-to-analyze peak among all 64 electrodes were extracted as a dataset for FPD and waveform analysis, according to a previous report [32]. The FPD values from the last 30 beats at each drug concentration were averaged.

Transmission Electron Microscopy (TEM)
hiPS-derived CMs were examined using electron microscopy with support from the Center for Anatomical, Pathological and Forensic Medical Research, Kyoto University Graduate School of Medicine, and a contract analysis with Hanaichi Ultrastructure Research Institute, Co., Ltd. (HURI), Japan. At Kyoto University, cells were fixed in 4% glutaraldehyde in PBS overnight at 4˚C and washed 3 times with PBS. Post-fixation was done using OsO4 (1% for 1 min, and 0.5% for 20 min at 4˚C) in PBS, dehydrated in ethanol and propylene oxide, and embedded in Luveak 812 (Nacalai Tesque, Japan). Ultrathin sections were then cut using an ultramicrotome (Leica, Heidelberg, Germany) and observed with TEM (H-7650; Hitachi, Japan). At HURI, cells were fixed in 2% glutaraldehyde in 0.1 M phosphate buffer at pH7.4, at 4˚C, overnight, and then washed 3 times with 0.1 M phosphate buffer at pH7.4. Post-fixation was done in 2% Osmium tetra-oxide/distilled water for 2 hours at 4˚C, dehydrated in ethanol and propylene oxide, and embedded in EPON812 (TAAB Laboratories Equipment, UK). Ultrathin sections were cut using an ultramicrotome (REICHERT-NISSEI, ULTRA CUTS, Germany) and observed with TEM (H-7600; Hitachi). Sarcomere length (distance between Zbands) were measured manually in randomly selected electron microscopy images.

Statistical analysis
The data was processed using Graph Pad Prism software (Ver5.0, GraphPad Software, San Diego, CA). Comparisons between two and three groups were made with Mann-Whitney tests or with one-way analysis of variance (ANOVA) followed by Tukey-Kramer test as post hoc, respectively. Values are shown as means ± SD. P values < 0.05 were considered significant.

Induction and long-term maintenance of high-purity hiPSC-derived CMs
Based on our previous 2-dimensional culture method for CM differentiation from hPSCs, we first induced mesoderm populations with Act-A, BMP4, and bFGF. We chose PDGFRα + cells at differentiation day 5 (d5) as a cell population that is responsive to CM differentiation cues. PDGFRα + cells were purified by MACS resulting in more than 95% purity, and then re-cultured with Wnt inhibitors to potently drive PDGFRα + cells towards the CM lineage (Fig 1).
In this protocol, we selected two small molecule Wnt inhibitors; XAV939 and IWP4, which have different mechanisms of action. XAV939 is a tankyrase inhibitor that stabilizes GSK3β, resulting in Wnt inhibition [33], while IWP4 inhibits porcupine, which in turn reduces Wnt ligand secretion and ligand-receptor binding affinity [34]. Simultaneous addition of XAV939 and IWP4 robustly induced cTnT + CMs from PDGFRα + cells on d21 with approximately 95% efficiency (cTnT + percentages: Control (without Wnt inhibitors); 7.1 ± 3.1%, XAV; 74.5 ± 6.5%, IWP; 72.2 ± 6.8%, XAV+IWP; 94.1 ± 2.0%) (Fig 2A and 2B). This dual Wnt inhibition led to an approximately 25 times higher yield in CM numbers, compared to the control (Control; 0.2 ± 0.1 × 10 5 cells/cm 2 , XAV; 3.0 ± 0.3 × 10 5 cells/cm 2 , IWP; 2.6 ± 0.5 × 10 5 cells/ cm 2 , XAV+IWP; 4.7 ± 0.3 × 10 5 cells/cm 2 ) (Fig 2C). To confirm whether this high CM purity can be continuously maintained during long-term culture, we re-plated CMs on d21 in αMEM medium with FBS and ROCK inhibitor (Y27632) for only the first two days to ensure CM attachment to the dish, followed by a switch to RPMI1640 medium with B27 supplement (RPMI1640+B27) for 18 days, and repeated this cycle every 20 days (Fig 2D). The percentage of cTnT + CMs was assessed by flow cytometry. The high-purity of the CM population on d21 (>95-98%) was maintained in this simple culture condition for approximately 2-3 months (Fig 2E and 2F). Notably, a high-purity CM population (> 95%) was still maintained even at d201 without any CM purification step (Fig 2G). We verified our protocol with two other cell lines, 253G1 and 1201C1 [28,29]. We also observed CM induction with more than 95% efficiency in these two lines and maintained that purity until d91 (Fig 2H and 2I). These data indicate that our protocol can robustly achieve CM differentiation with high efficiency as well as long-term maintenance of high-purity CMs in various hiPSC lines, regardless of their origin and derivation.
Undifferentiated hPSCs were plated 2 or 3 days before the induction of differentiation (-d2~3) on Matrigel-coated dishes in mouse embryonic fibroblast conditioned medium (MEF-CM). A matrigel overlayer was applied on -d1. After the induction of PDGFRα + cells with Activin-A followed by BMP4 and bFGF in RPMI1640 medium including B27 supplement without insulin (B27 minus insulin), PDGFRα + cells were purified using MACS and re-cultured on gelatin or matrigel-coated plates in FBS-containing medium with Wnt signal inhibitors (XAV939: 10 μM and IWP4: 5 μM) and a ROCK inhibitor (Y-27632) for 2 days. After that, medium was changed to a serum free condition (RPMI1640+B27) with a minimal dose of Wnt signal inhibitors (XAV939: 0.25 μM and IWP4: 0.125 μM). Following this protocol, highly pure CMs were induced and maintained in the long-term.

Proliferation capacity of hiPSC-derived CMs
To assess the proliferation capacity of induced human CMs, we continuously cultured CMs and counted their numbers from d21 to d101 with sequential passages every 20 days. A constant increase in the number of CMs was clearly observed until d81. A single CM on d21 was capable of generating approximately 5.6 ± 1.4 CMs following 60 days of culture (Fig 3A). We quantitatively examined the proliferation capacity of these CMs by Ki67 staining and flow cytometry (Fig 3B). Although Ki67 + proliferating CMs were approximately 30% on d21 (d21; 28.1 ± 4.0%), they declined to around 4% or less by d101 (d101; 3.6 ± 0.7%) (Fig 3C). Similar results were confirmed by immunostaining of Ki67 and cTnT (Fig 3D). The proliferative capacity of hiPSC-derived CMs was maintained for approximately 2 months after the appearance of CMs, and disappeared between d81 and d101 under this culture condition.

Chamber specification and structural maturation of hiPSC-derived CMs after long-term culture
Next, we evaluated the structural maturation and specification of CMs. We examined the atrial and ventricular specification of hiPSC-derived CMs. In human embryonic and adult hearts, myosin light chain-2a (MLC-2a) is expressed in both the atrium and the ventricle, whereas myosin light chain-2v (MLC-2v) is selectively expressed in the ventricle [35,36]. The expression of MLC-2a in immature MLC-2v + ventricular CMs gradually decreases as CMs mature [37]. In our experiments, MLC-2v expression was heterogeneous but positive in almost all CMs on d21 and d91 (Fig 4A and 4B). On d21, CMs with higher expression of MLC-2a than MLC-2v (A high+ V low+ ) were observed, but that higher MLC-2a expression had almost completely disappeared by d91 (d21; 67.0 ± 4.3%, d91; 0.6 ± 1.3%). On the contrary, cells expressing higher MLC-2v levels (A low+ V high + ) increased in number and dominated the ��� p < 0.001 by one-way ANOVA followed by Tukey-Kramer post hoc test (n = 4). (D) Scheme of CM long-term maintenance protocol. CMs were re-cultured under αMEM medium with 10% FBS and Y27632, for the initial 2 days, followed by a serum-free condition (RPMI1640+B27) for 18  population by d91, compared to d21 (d21; 32.5 ± 4.0%, d91; 99.4 ± 1.3%). MLC-2a + MLC-2v -(A + V -) atrial CMs were very rare (d21; 0.3 ± 0.6%, d91 A + V -; not detected). We further confirmed this ventricular specification at the mRNA expression level. Although Mlc2a mRNA expression was largely comparable between d21 and d91 CMs, Mlc2v expression was drastically increased in d91 CMs to approximately 37 times that in d21 CMs (Fig 4C). These results suggest that our differentiation protocol strongly and specifically induced ventricular-type CMs.
Next, we examined the structural maturation of CMs. Double immunofluorescent staining of cTnT and α-actinin clearly showed a tendency to morphological maturation (Fig 4D). Though sarcomere structures were formed in both d39 and d155 CMs, myofibers on d39 were not clearly aligned and cells showed a polygonal appearance, suggesting that CMs were still immature. On the other hand, some CMs on d155 exhibited a rod shape with clear myofiber  analysis also demonstrated the gradual maturation of CMs at the ultrastructural level. CMs on d21 contained myofibrils (MF) that lacked a clear alignment or an organized sarcomeric structure with scattered Z-body (Zb) patterns. Mitochondria exhibited round immature shapes ( Fig  4E). On the other hand, CMs on d91 showed a more mature ultrastructure (Fig 4F). Sarcomeres showed clearly aligned Z-disks and organized A-and I-bands with H-zones. CMs on d140 featured faint M-lines starting to form (Fig 4G). Mitochondria showed a tendency to increase in number and grow in size with a complex inner structure and location alongside sarcomeres. On d231, massive well aligned myofibrils with clear sarcomere formation and mitochondria alongside sarcomeres were observed (Fig 4H). In addition to a clear sarcomeric structure with Z-, I-, A-, H-, and M-bands and well established mitochondria with mature inner composition, transverse tubule (T-tubule)-like structures were found along the Z-bands, where T-tubules are located in CMs (Fig 4I). Measurements from electron microscopy images demonstrated that sarcomere length was increased during long-term culture from 1.37 ± 0.10 μm on d24 to 1.73 ± 0.19 μm on d91 and to 1.91 ± 0.15 μm on d231 (Fig 4K). These results indicate that the structural maturation of CMs gradually progresses during long-term culture.

Electrophysiological characterization and maturation of CMs after longterm culture
To examine the degree of electrophysiological maturation of CMs on d21 and d91, we examined action potentials (APs) using patch clamp analysis. APs were recorded from spontaneously beating single CMs on d21 and d91 (S2 and S3 Tables). We first categorized ventricularlike CMs and non-ventricular CMs according to their AP duration (APD) 30-40/APD70-80 ratio (� 1.5) as previously reported [30]. Then, according to the maximal rate of depolarization (dV/dt max), we further classified non-ventricular CMs into nodal-like and atrial-like, and ventricular-like CMs into early and late ventricular-like cells (Fig 5A). Briefly, non-ventricular CMs (APD30-40 / APD70-80 ratio: < 1.5) were classified into nodal-like CMs (dV/dt max < 10 mV/ms) and atrial-like CMs (dV/dt max � 10 mV/ms) [38,39]. To distinguish the early and late ventricular-like CMs, we set a broader value at 30 mV/ms, which was the mean plus two standard deviations of dV/dt max in CMs on d21 (Fig 5B). Ventricular-like CMs were further classified into early ventricular-like CMs (dV/dt max < 30 mV/ms) and late ventricular-like CMs (dV/dt max � 30 mV/ms). The overall distribution of CMs on d21 and d91 is displayed in Fig 5B. Consistent with the results of MLC-2a and MLC-2v immunostaining (Fig 4A and 4B), the majority of CMs were categorized as ventricular-like CMs (90% at d21, 96% at d91) (Fig 5C and 5D). Late ventricular-like CMs increased from 3% (d21) to 61% (d91), indicating that relative electrophysiological maturation as well as structural maturation occurred during long-term culture. Only one cell among the total 54 cells analyzed (on d21 and d91) was categorized as an atrial-like CM and few cells showed nodal-like CM features (7% and 4% on d21 and d91, respectively) (Fig 5B-5D). In addition to the dV/dt max as 1. �� p<0.01, � p<0.05, n = 6 with technical triplicates for each. (D) Double immunofluorescent staining of cTnT (green) and α-actinin (red) on d39 and d155. Right panels: higher magnification views of boxed areas in the left panels. Scale bars in the left panels: 10 μm. Scale bars in the right panels: 5 μm. characterization, late ventricular-like CMs on d91 had deeper depolarized maximum diastolic potentials (MDPs), and larger peak voltages and action potential amplitudes (APAs) compared to CMs on d21, indicating that ventricular-like CMs were predominantly induced and also underwent gradual electrophysiological maturation during long-term culture (Fig 5E, S2 and S3 Tables).
Finally, to demonstrate the utility of long term-cultured CMs for pharmacological studies, we examined the effects of selective ion channel inhibitors, the human Ether-a-go-go Related Gene (hERG) channel blocker, E4031, and the KCQN1 channel blocker, Chromanol-293B (Chromanol). We treated CMs with E4031 or Chromanol and examined their electrophysiological features by field potential duration (FPD) recorded using multi-electrode arrays (MEAs) [17]. E4031 and Chromanol have been reported to induce QT prolongation on the electrocardiogram by inhibiting the delayed-rectifying K + current (I Kr ) and the slow delayed rectifier current (I Ks ), respectively. Apparent dose-dependent FPD prolongations by E4031 and Chromanol, which correspond to QT prolongations on the electrocardiogram, were observed in CMs (Fig 5F and 5G). I kr has been detected in hESC-derived immature CMs whereas I Ks has been reported to show low expression levels in immature CMs and increased levels in mature CMs [30,40,41]. We compared the difference in the response to Chromanol among CMs on d31 and over d91 (Fig 5H). Although the average % change of FPD values in each dose was not apparently different, the deviation of these values tended to be smaller in late ventricular-like CMs over d91 (S4 Table). These results suggest that during maturation in long-term culture, CMs became more stable and homogenous in their electrophysiological features, implying potential utility for pharmacological studies.

Discussion
In the present study, we established an efficient method to obtain hiPSC-derived high-purity CM populations without the need for CM purification steps. Moreover, these populations maintain their high-purity (>95%) for over 200 days of culture (long-term culture). The majority of CMs were induced specifically into a ventricular phenotype and exhibited gradual structural and electrophysiological maturation. These high-purity CM populations with a relatively homogenous and matured ventricular phenotype could potentially have an advantage over immature CMs for pharmacological studies.
We previously reported the induction of pure ECs based on the SE method [18]. Adapting this approach to CM differentiation, we collected responder cells by purifying PDGFRα + cells on d5, and strongly committed these cells to the CM lineage. We did so by inhibiting canonical Wnt signaling with two Wnt inhibitors, XAV939 and IWP4 [17]. This CM induction approach, based on the SE method, yielded high-purity hiPSC-derived CM populations that can maintain their purity during long term maintenance without the need for CM purification steps (Fig 1). The removal of aggressive purification procedures prevents potential CM damage that could affect their purity and viability. In addition, our method allows for the long-term maintenance and expansion of pure CM populations across several hPSC lines with no gene modifications, such as the introduction of antibiotic-resistant genes, required. Since our differentiation method yields pure and viable high-quality CMs and is expandable to a broad range APs on d21 (n = 31 from three independent experiment), and d91 (n = 23 from three independent experiment). (D) Representative action potentials of early and late ventricular-like and nodal-like CMs. (E) Comparison of action potential parameters between early ventricular-like and late ventricular-like CMs on d21(n = 28 from three independent experiments) and d91 (n = 22 from three independent experiments). (n = 3). � p < 0.05, ��� p < 0.001 by one-way ANOVA followed by Tukey-Kramer post hoc test. (F) Administration of E4031 (Ikr blocker) prolonged FPD in a dose-dependent manner at 0-30 nM in CMs. (G) Administration of Chromanol (Iks blocker) prolonged FPD in a dose-dependent manner at 0-30 μM in CMs. (H) Homogeneous response to Chromanol in a time-dependent manner. d31 (n = 16), over d91 (n = 8).
https://doi.org/10.1371/journal.pone.0241287.g005 of hPSC lines (including disease lines), it could be very valuable for pharmacological testing. Furthermore, the elimination of cells that are not susceptible to target differentiation stimuli during the earlier differentiation stages successfully induced highly pure ECs [18] and CMs (this study), suggesting that the SE method approach could be useful to generate other hPSCderived cell populations with high purity and long-term maintenance capacity.
In our experiments, the proliferative capacity of CMs was gradually lost as CMs matured (Fig 3). Similarly, in human heart development, CMs almost completely lose their proliferative capacity after the first few postnatal months [42]. Although the relationship between the proliferative capacity of human embryonic and postnatal CMs and hiPSC-derived CMs remains uncertain, this decreased proliferative capacity of hiPSC-derived CMs throughout maturation may recapitulate postnatal heart development [43]. Another study reported that hiPSC-derived CM proliferation in culture and after transplantation into mouse hearts seemed to cease at around 3 months [44], supporting our results. The proliferative capacity of hiPSC-derived CMs seems to be restricted within this period.
There was one report about hiPSC-derived CMs structurally evaluated by TEM along 1 year of in vitro culture [37]. TEM demonstrated gradual sarcomeric formation with Z-disks and I-bands by 180 days and with H-zones, and M-lines by 1 year. However, T-tubules, which are indispensable in adult CMs for rapid excitation-contraction coupling, were never observed. In our study, T-tubule-like structures were observed on d231 (Fig 4I). We consider our highly pure ventricular CM culture may contribute to T-tubule formation. Recently, engineered cardiac tissues made from hiPSC-derived CMs and fibroblasts in fibrin hydrogels and subjected to stretch and auxotonic contractions for just four weeks showed structural maturation, including T-tubule formation [45]. In our experiments, we observed immature T-tubule-like structures along with Z-bands as early as d24 (Fig 4J). Based on this evidence, we consider that T-tubule formation may be initiated at an earlier stage of CM differentiation than was previously thought. Currently, the mechanisms involved in the generation and maturation of Ttubules in human CMs are largely unknown. Our method may provide a useful platform to manipulate and study CM maturation, including T-tubule formation.
As for the electrophysiological maturation of high-purity long-term cultured CMs, there was a prominent change in dV/dt max between CMs on d21 and d91 (Fig 5B). Although there was some heterogeneity among samples, some CMs on d91 showed more than 200 mV/ms. These values are still lower than those previously reported for human adult CMs (around 300-400 mV/ms) [46,47], but are comparable to those of human PSC-derived CMs exhibiting signs of in vitro maturation (>150 mV/ms) [48,49]. This tendency to electrophysiological maturation during long-term culture could be reflected in the formation of T-tubule-like structures like the ones observed in our electron microscopy analysis (Fig 4I). An increased expression and function of the I Ks channel is also an indicator of CM maturation [30,40,41]. The response of late ventricular-like CMs to Chromanol was relatively stable and reproducible, suggesting that I Ks functionally worked in CMs on d91. Additionally, the smaller standard deviations observed in d91 CMs suggest a less heterogeneous phenotype within the population in terms of electrophysiological maturation (Fig 5H).
Our differentiation protocol enabled us to obtain highly pure ventricular-like CMs with no specific selection steps. In previous reports, atrial specification of CMs derived from hESCs or hiPSCs has been mediated by retinoic acid or retinal aldehyde dehydrogenase 2 (which converts retinal aldehyde to retinoic acid) [50][51][52]. These in vitro studies were based on the mammalian heart development in vivo [53,54]. As to the specification of the ventricular lineage, a higher ratio of ActA to BMP4 signaling (5 ng/mL of BMP and 12 ng/mL of ActA) during the initial stage of hPSC differentiation was required to induce a CD235a + mesoderm population, which is considered a prerequisite for the generation of populations highly enriched in ventricular cardiomyocytes [50]. However, the percentage of CD235a + cells is not necessarily predictive of the subsequent percentage of ventricular cardiomyocytes. Some ventricular cardiomyocytes were also induced from an isolated CD235amesoderm population [50]. Some of our preliminary data showed that CD235a + cells were just a small subset of our responder population (PDGFRα + cells) (data not shown). In spite of this, ventricular-like cardiomyocytes were very specifically induced in our method, suggesting that the majority of CMs obtained in our method are in fact derived from a CD235apopulation. These contradictory conclusions may be due to the differences between these differentiation methods. For example, the initial differentiation condition in our protocol (100 ng/mL of ActA alone), is much different from that in the previous report. The methods for CM induction from the mesoderm stage onward are also largely different. Our method features a 2-D culture treated with Wnt inhibitors, being different from the cell aggregates with a Wnt inhibitor and VEGF utilized in [50]. The mechanisms and trajectory of the ventricular-specific induction achieved in our method are currently unclear.
In conclusion, we established a method for CM-specific induction and maintenance without CM purification steps which yielded high-purity ventricular CMs with a gradual tendency to structural and electrophysiological maturation. This method could offer a source of CMs for a number of applications, including cardiac toxicology testing and disease modeling as well as regenerative medicine.
Supporting information S1