OUP user menu

Comparative study of human-induced pluripotent stem cells derived from bone marrow cells, hair keratinocytes, and skin fibroblasts

Katrin Streckfuss-Bömeke, Frieder Wolf, Azadeh Azizian, Michael Stauske, Malte Tiburcy, Stefan Wagner, Daniela Hübscher, Ralf Dressel, Simin Chen, Jörg Jende, Gerald Wulf, Verena Lorenz, Michael P. Schön, Lars S. Maier, Wolfram H. Zimmermann, Gerd Hasenfuss, Kaomei Guan
DOI: http://dx.doi.org/10.1093/eurheartj/ehs203 2618-2629 First published online: 12 July 2012

Abstract

Aims Induced pluripotent stem cells (iPSCs) provide a unique opportunity for the generation of patient-specific cells for use in disease modelling, drug screening, and regenerative medicine. The aim of this study was to compare human-induced pluripotent stem cells (hiPSCs) derived from different somatic cell sources regarding their generation efficiency and cardiac differentiation potential, and functionalities of cardiomyocytes.

Methods and results We generated hiPSCs from hair keratinocytes, bone marrow mesenchymal stem cells (MSCs), and skin fibroblasts by using two different virus systems. We show that MSCs and fibroblasts are more easily reprogrammed than keratinocytes. This corresponds to higher methylation levels of minimal promoter regions of the OCT4 and NANOG genes in keratinocytes than in MSCs and fibroblasts. The success rate and reprogramming efficiency was significantly higher by using the STEMCCA system than the OSNL system. All analysed hiPSCs are pluripotent and show phenotypical characteristics similar to human embryonic stem cells. We studied the cardiac differentiation efficiency of generated hiPSC lines (n = 24) and found that MSC-derived hiPSCs exhibited a significantly higher efficiency to spontaneously differentiate into beating cardiomyocytes when compared with keratinocyte-, and fibroblast-derived hiPSCs. There was no significant difference in the functionalities of the cardiomyocytes derived from hiPSCs with different origins, showing the presence of pacemaker-, atrial-, ventricular- and Purkinje-like cardiomyocytes, and exhibiting rhythmic Ca2+ transients and Ca2+ sparks in hiPSC-derived cardiomyocytes. Furthermore, spontaneously and synchronously beating and force-developing engineered heart tissues were generated.

Conclusions Human-induced pluripotent stem cells can be reprogrammed from all three somatic cell types, but with different efficiency. All analysed iPSCs can differentiate into cardiomyocytes, and the functionalities of cardiomyocytes derived from different cell origins are similar. However, MSC-derived hiPSCs revealed a higher cardiac differentiation efficiency than keratinocyte- and fibroblast-derived hiPSCs.

  • Hair keratinocytes
  • Mesenchymal stem cells
  • Skin fibroblasts
  • Human-induced pluripotent stem cells
  • Cardiomyocytes

Introduction

Induced pluripotent stem cells (iPSCs) are generated from somatic cells by overexpression of a set of pluripotency-specific transcriptional factors.1,2 Given the robustness of the approach, human-induced pluripotent stem cells (hiPSCs) can be derived and used for understanding of the pathophysiology of monogenic and complex diseases and for screening drugs. This may facilitate the development of novel patient-specific therapeutic strategies.

In the mouse system, many cell types have been successfully reprogrammed into iPSCs, including embryonic fibroblasts, adult dermal fibroblasts, B cells, gastric epithelial cells, hepatocytes, β-cells, bone marrow cells, and neural stem cells.38 However, recent studies demonstrate that pluripotent cell lines of different origins differ in growth characteristics, developmental potential, transcriptional activity, and epigenetic regulation. Upon derivation, iPSCs generally acquire new properties, but they often also retain a ‘footprint’ of their tissue of origin.911 Dependent on the application, this variance could have significant effects on the outcome, for example, in efficient differentiation and functional properties of cells required for future drug screens, or in understanding disease mechanisms.

Up to now, reprogramming of human somatic cells was mainly focused on using foetal, neonatal, or adult fibroblasts.1,2,12,13 The use of human dermal fibroblasts and foreskin or hair keratinocytes for generating hiPSCs shows great promise in clinical application because they are the most accessible cell sources.12,14,15 However, many aspects before the application of hiPSCs in disease modelling and in regenerative medicine have to be addressed, for example, which donor cell source would be the best for the generation of hiPSCs, how to efficiently differentiate these cells into certain cell types needed, and how to use these cells.

Recent reports suggest that the successful generation of hiPSCs may be easier to achieve from actively dividing cells than from slow/non-dividing cells.16 The adult bone marrow contains mesenchymal stem cells (MSCs), which are multipotent and proliferative in vitro and can contribute to the regeneration of mesenchymal tissues such as bone, cartilage, muscle, ligament, tendon, adipose, and stroma.17

Accordingly, in the present study, we directly compared three different cell sources: hair keratinocytes, bone marrow MSCs, and skin fibroblasts regarding their reprogramming efficiency, and the DNA methylation status of OCT4, NANOG, and SOX2 promoters. We induced these cells using two different lentiviral systems into hiPSCs, and compared their cardiac differentiation potential and functionalities of differentiated cardiomyocytes. Our results demonstrate that hiPSCs can be generated from different somatic cells with different efficiency. Although the cardiac differentiation efficiency varied among different cell lines, the functionalities of cardiomyocytes derived from different hiPSC lines are comparable.

Methods

Isolation and cultivation of hair keratinocytes, bone marrow mesenchymal stem cells, and skin fibroblasts

The study was approved by the ethics committee of Universitätsmedizin Göttingen. The keratinocyte culture was established from plucked hairs, whereas the fibroblast culture was derived from skin punch biopsies of the donors. Human MSCs were isolated from bone marrow aspirate and cultured as previously described.17

Generation and characterization of human-induced pluripotent stem cells

Two different lentiviral systems were used and compared: the OSNL system containing four reprogramming factors (OCT4, SOX2, NANOG, and LIN28) in four separate lentiviral vectors,1 and the STEMCCA system, which is a humanized excisable lentivirus system containing all four reprogramming factors OCT4, SOX2, KLF4, and c-MYC in a single ‘stem cell cassette’ (pHAGE2-EF1aFull-hOct4-F2A-hKlf4-IRES-hSox2-P2A-hcMyc-W-loxP).18

Reprogramming was typically initiated on cells at passage 3 (p3) or p4. The protocols used for reprogramming are illustrated in Figure 1A and B. After mechanical expansion for three to four passages, the generated hiPSCs were propagated by using collagenase IV (200 U/mL, Worthington) and cultured on mitomycin C-inactivated mouse embryonic fibroblasts (MEFs) in human embryonic stem cell (hESC) medium composed of DMEM/F12 (Invitrogen) containing 20% Knockout serum replacement (Invitrogen), l-glutamine (2 mmol/L, Invitrogen), 1× non-essential amino acids (NEAA, Invitrogen), β-mercaptoethanol (50 µmol/L, Serva), and basic fibroblast growth factor (bFGF, 10 ng/mL, TEBU). For characterization of the generated hiPSCs, RT–PCR analysis, immunocytochemical staining, bisulfite sequencing assays, and teratoma formation was carried out using standard protocols. Human ESC line HES319 was used as a control. The use of human ESCs are proved by the Robert Koch Institute (Az 1710-79-1-4-28).

Figure 1

Generation and characteristics of human-induced pluripotent stem cells derived from keratinocytes (Kera), mesenchymal stem cells and fibroblasts. (A and B) The scheme of the reprogramming procedures of keratinocytes, mesenchymal stem cells, and fibroblasts. (C) The reprogramming efficiency for the generation of Kera-, mesenchymal stem cell-, and fibroblast-human-induced pluripotent stem cells. The reprogramming efficiency was evaluated as total iPSC colonies obtained up to Day 42 from 104 starting cells. *P≤ 0.05; **P≤ 0.01; ***P≤ 0.001. (D) The morphology of Kera-, mesenchymal stem cell-, and fibroblast-human-induced pluripotent stem cells as well as their parental keratinocytes, mesenchymal stem cells, and fibroblasts. Human-induced pluripotent stem cells show similar morphology as human embryonic stem cells and are positive for alkaline phosphatase. Human embryonic stem cells were used as a positive control. Scale bar, 100 µm.

In vitro differentiation

For embryoid body (EB) formation, hiPSCs, and HES3 cells were dispersed with collagenase IV into small clumps and cultured in suspension in hESC medium for 1 day. To initiate differentiation, the formed EBs were then cultured in suspension for another 7 days in differentiation medium, consisting of Iscove's modified Dullbecco's medium (Invitrogen), 20% foetal calf serum, l-glutamine (2 mmol/L), 1× NEAA, and α-monothioglycerol 3-mercapto-1,2-propandiol (450 µmol/L, Sigma). Embryoid bodies at Day 8 were then plated on 0.1% gelatin-coated dishes and cultured in differentiation medium for another 30 days. The percentage of EBs containing beating cardiomyocytes at Day 33 was used as a measure of the cardiac differentiation efficiency of generated hiPSCs. Seven keratinocyte-derived, nine MSC-derived, and eight fibroblast-derived iPSC lines from 12 donors were investigated at early (p8–12) and late passages (p16–20). HES3cells at p15–20 were used as a control. At least three independent experiments (n = 24–48 EBs per experiment) per cell line were performed.

Electrophysiology and measurement of Ca2+ transients

For electrophysiological analysis, the beating clusters were mechanically isolated at Day 8 + 25 and digested with collagenase type 2 (300 U/mL, Worthington) into single cells, which were further cultured and used 2–3 days later.20 The membrane potential of single cardiomyocytes was measured with a ruptured-patch whole cell current clamp.20,21 In some experiments, isoproterenol (100 nmol/L, Sigma) was added to the bath solution during recording.

Intracellular calcium signals were recorded after incubating cells with 10 µmol/L fluo-4 acetoxymethylester (Mobitec) on a confocal microscope (Zeiss LSM5 PASCAL).20 In some experiments, caffeine (10 mmol/L, Sigma) was added to the bath solution during recording.

Generation of circular EHTs

Engineered heart tissues (EHTs) were generated based on protocols previously described for neonatal rat cardiomyocytes.22 For the generation of EHTs, beating clusters (n = 100 per EHT) were collected from Day 15 to 20 after EB plating. Circular EHTs (10-day old) were subjected to isometric force measurements followed by immunofluorescence staining.

Statistical analysis

To compare the reprogramming efficiency of three cell types, cardiac differentiation efficiency of generated hiPSCs, as well as DNA methylation statuses of the cells, data that passed tests for normality were analysed with the use of Student's t-test. Two-sided P-values of <0.05 were considered statistically significant. The statistical software used for analysis was Sigma-Stat (version 3.5). All data are reported as means ± SEM.

The detailed procedures are described in Supplementary material online, Methods.

Results

Optimization of culture conditions for hair keratinocytes, bone marrow mesenchymal stem cells, and skin fibroblasts

To establish hair keratinocyte cultures, the enzyme-digested hair follicles with a visible outer root sheath were cultured on MEFs, and keratinocyte colonies (Figure 1D) are visible in the culture within 8–10 days. Normally, the cells were first passaged at Day 12–14 when the colonies around hair follicles reached 1 cm in diameter. From 40 hair follicles, ∼1–2 × 105 keratinocytes were obtained after 2–3-week cultivation, which were used for transduction experiments. With our protocol, keratinocyte cultures were generated with an efficiency of 96% (Supplementary material online, Table S1). However, >80% of the keratinocyte cultures stop proliferating after three passages.

Mesenchymal stem cells were isolated from mononuclear cells by their property to adhere to tissue culture dishes and they showed fibroblast-like morphology (Figure 1D). About 2–5 × 106 of MSCs were developed in culture at 9 days (Supplementary material online, Table S1), which were used for transduction or further expansion. All analysed MSCs were expanded for at least five passages. From 27.8% of donors, we could culture MSCs up to the 10th passage.

Dermal fibroblast cultures were established from the skin punch biopsy (3.5–4 mm) of the donors (Supplementary material online, Table S1). Fibroblasts grew out from the tissue pieces within 7–14 days. When fibroblasts spotted around skin pieces reached 1 cm in diameter, they were passaged. About 2–5 × 105 fibroblasts (Figure 1D) were obtained 2 weeks after the initial culture and were ready for transduction experiments. About 80% of the fibroblast cultures were expanded for at least five passages.

We tested the expression levels of pluripotency genes in all three cell types (Figure 2A–C). OCT4 and SOX2 were slightly expressed, whereas NANOG was absent in all three cell types. LIN28 was not expressed in keratinocytes and fibroblasts (Figure 2A and C), but very slightly in some MSCs (Figure 2B).

Figure 2

The expression of pluripotent genes and DNA methylation analysis. (A–C) RT–PCR analysis of endogenous expression of OCT4, NANOG, LIN28, and SOX2 in keratinocytes (A), mesenchymal stem cells (B), and fibroblasts (C) as well as human-induced pluripotent stem cells derived. (D) Bisulfite sequencing analysis of DNA methylation of OCT4 and NANOG promoter regions. Colour codes indicate 0% (yellow), over 50% (green) to 100% (blue) methylation. The y-axis shows individual CpGs analysed. The x-axis shows cell types. HES3 was used as a positive control and MEFs as a negative control.

To verify the potential epigenetic mechanisms mediating gene expression, we analysed the DNA methylation status of the OCT4, NANOG, and SOX2 promoters in all three cell types (Figure 2D, Supplementary material online, Figure S1). Whereas the mini-promoter region of the SOX2 gene was greatly demethylated in all three cell types (Supplementary material online, Figure S1), the mini-promoter regions of the OCT4 and NANOG genes were highly methylated (Figure 2D). It is noteworthy that the methylation status of the mini-promoter regions of the OCT4 and NANOG gene was significantly (P < 0.05) higher in keratinocytes than in MSCs and fibroblasts.

Induction of pluripotent stem cells

To investigate which virus system is more efficient for the generation of iPSCs, we transduced keratinocytes, MSCs, and fibroblasts using two different virus systems: the OSNL system and the STEMCCA system (Supplementary material online, Table S2). The number of hiPSC lines generated with three cell types and two virus systems are summarized in Supplementary material online, Table S2. Our data showed that success rates and the reprogramming efficiency were higher for all three cell types with the STEMCCA system in comparison with the OSNL system (Supplementary material online, Table S2, Figure 2C). Furthermore, our data showed that the reprogramming efficiency was significantly higher for both MSCs and fibroblasts compared with keratinocytes (Figure 1C). Furthermore, the reprogramming efficiency was higher in MSCs than in fibroblasts (Figure 1C). Importantly, we are able to generate hiPSCs from all three cell types derived from the people aged 70 years and over.

In addition, we found that fresh isolated cells at passages <3 were more efficiently reprogrammed than cells at higher passages or frozen/thawed cells. Because keratinocytes are very sensitive to high calcium concentration, they do not expand, but die in hESC medium. After transduction, changing to hESC medium too early results in the death of the cells before the reprogramming occurs. Therefore, slowly increasing hESC medium is necessary for successful reprogramming (Figure 1A). We also found that plating dermal fibroblasts post-transduction on MEFs at low density was crucial to avoid the overgrowth of fast-growing non-iPSCs (Figure 1B).

All generated iPSC lines showing typical hESC morphology were positive for alkaline phosphatase (Figure 1D). Each of the iPSC lines has been maintained for >20 passages without any obvious phenotypic changes.

Proof of pluripotency

To prove the pluripotency of the generated hiPSCs, we first analysed mRNA expression of the endogenous OCT4, NANOG, SOX2, and LIN28. We found significantly higher levels of the tested genes in hiPSCs than those in parental cells (Figure 2A–C). Accordingly, immunocytochemical staining of human stem cell markers, including OCT4, NANOG, SOX2, LIN28, SSEA4, and TRA-1-60, was similar in all the analysed hiPSCs and hESCs (Figure 3A). We also analysed the DNA methylation status of the OCT4, NANOG, and SOX2 promoters in generated hiPSCs (Figure 2D, Supplementary material online, Figure S1). We randomly selected four parental cultures and three to four hiPSC lines for each cell type. We found that the analysed hiPSC lines exhibited a hypomethylated pattern similar to hESCs, and the OCT4 and NANOG promoters were significantly (P < 0.001) demethylated in all analysed hiPSCs compared with their parental cells (Figure 2D). No significant differences in the DNA methylation status of SOX2 were observed in all analysed hiPSCs compared with their parental cells (Supplementary material online, Figure S1).

Figure 3

Characterization of human-induced pluripotent stem cells. (A) Immunocytochemical analyses show that human-induced pluripotent stem cells derived from keratinocytes (Kera1-iPS1, Kera3-iPS1), mesenchymal stem cells (MSC2-iPS1), and fibroblasts (FB1-iPS1) are positive for OCT4, SOX2, NANOG, LIN28, SSEA4, and Tra-1-60. Nuclei were stained with 4',6-Diamidin-2-phenylindol (DAPI). (B) Teratoma formation of MSC2-iPS1 cells in SCID-beige mice. Shown are epithelium with intestinal differentiation (endoderm) striated muscles, cartilage (mesoderm), and neural tissues (ectoderm). Scale bars, 100 µm.

Next, we injected iPSCs of nine cell lines (Kera1-iPS1, Kera2-iPS1, Kera3-iPS1, MSC1-iPS1, MSC2-iPS1, MSC3-iPS1, FB1-iPS1, FB2-iPS1, and FB3-iPS1) derived from three cell types at p16 subcutaneously in the SCID-beige mice (n = 4 per cell line). Teratoma formation was observed from all nine lines. The tumours contained abundant differentiation of advanced derivatives of three embryonic germ layers (Figure 3B, Supplementary material online, Figure S1).

Finally, to determine the in vitro differentiation potential of generated hiPSCs, we examined the expression of a panel of genes and proteins during EB differentiation (Supplementary material online, Figure S1 and S2; Figure 4). Tissue-specific genes were expressed in a developmentally controlled and comparable manner during the differentiation of both keratinocyte- and MSC-derived hiPSCs (Supplementary material online, Figure S1and S2). We found the expression of the endodermal marker α-fetoprotein (AFP), the mesodermal marker smooth muscle-α-actin (SM-α-actin), and the ectodermal marker βIII-tubulin in EB outgrowths using immunostaining (Figure 4).

Figure 4

In vitro differentiation of Kera1-iPS1, MSC2-iPS1, and FB1-iPS1 cells into derivatives of three germ layers. Immunostaining of EB outgrowths at Day 8 + 25 revealed the expression of endodermal (α-fetoprotein), mesodermal (smooth muscle-α-actin), and ectodermal (βIII-tubulin) marker proteins. Nuclei are stained with DAPI. Scale bars, 50 µm.

Cardiac differentiation in vitro

To compare the cardiac differentiation efficiency of hiPSCs derived from the three different cell sources and functionalities of differentiated cardiomyocytes, we included a great number of hiPSC lines (n = 24) from 12 different donors at early (p8–12) and late passages (p16–20) derived from the three different cell types using the two different virus systems. We found that all analysed iPSCs can differentiate into cardiomyocytes with efficiency ranging from 3 to 42%. First beating clusters (Supplementary material online, Video S1) were observed at Day 8 + 4 to 8 + 7 for most of analysed iPSC lines similar to HES3. In Supplementary material online, Table S3, we summarized the maximal cardiac differentiation efficiency of all analysed hiPSC lines. We did not observe significant differences between hiPSCs at early and late passages. However, we found remarkable line-to-line variability, which is mainly associated with their cell origins. The overall cardiac differentiation efficiency is significantly higher in MSC-iPSC lines compared with Kera- or FB-iPSC lines at both early and late passages (Figure 5A). In addition, we found variations of hiPSC lines from the same donor in some cases (Nr.3 and Nr.5, Supplementary material online, Table S3), whereas in other cases (Nr.6, 9, 10, 11, and 12; Supplementary material online, Table S3) we did not observe significant variation of hiPSC lines from the same donor. Furthermore, we included patient-matched Kera- and FB-iPSC lines from two donors (Nr.7 and Nr.8, Supplementary material online, Table S3) and found that the cardiac differentiation efficiency of FB-iPSC lines is higher than that of the Kera-iPSC line from the same patient.

Figure 5

Cardiac differentiation of human-induced pluripotent stem cells and EHT generation. (A) The overall cardiac differentiation efficiency of human-induced pluripotent stem cells with different cell origins. Six Kera-, eight mesenchymal stem cell, and five fibroblast-human-induced pluripotent stem cells lines at p8–12 as well as six Kera-, seven mesenchymal stem cell, and seven fibroblast-human-induced pluripotent stem cell lines at p16–20 were analysed and data were summarized. *P≤ 0.05; **P≤ 0.05; ***P≤ 0.001. (BF) Immunostaining using antibodies against cardiac troponin T (B), myosin heavy chain (C), α-actinin (D and F), and connexin 43 (E and F). Scale bars, 25 µm. (GI) Sarcomeric structures of cardiomyocytes in EHTs shown by staining with actin (G and I) and α-actinin (H and I). Scale bars, 50 µm. (J) Representative force measurements. (K) Stimulation of EHTs with isoproterenol. Nuclei were stained with DAPI.

We next compared the functionalities of cardiomyocytes differentiated from MSC-iPSCs, Kera-iPSCs, and FB-iPSCs. We found no difference regarding sarcomeric striations in cardiomyocytes derived from all three origins following staining with cardiac troponin T (Figure 5B), sarcomeric myosin heavy chain (Figure 5C), and α-actinin (Figure 5D and F). The expression of gap junction protein connexin 43 at cell-to-cell contacts in cardiac clusters (Figure 5E and F) indicated cell-to-cell communication.

Electrophysiological studies showed that cardiomyocytes derived from iPSCs exhibited spontaneous action potentials (APs, Figure 6A). We examined the shape and properties of APs (AP upstroke velocity, dV/dtmax; AP duration, APD80; maximum diastolic potential, MDP; AP amplitude, APA; and spontaneous AP frequency, SAF) from single beating cardiomyocytes. All four major AP types characteristic for pacemaker-, ventricle-, atrial-, and Purkinje-like cells were found in cardiomyocytes derived from MSC-iPSCs, Kera-iPSCs, and FB-iPSCs (Table 1, Figure 6A). We did not observe significant differences of AP parameters among cardiomyocytes derived from hiPSCs with different cell origins. In addition, we found that the stimulation of cardiomyocytes with β-adrenergic agonist isoproterenol (100 nmol/L) resulted in the significant increase of the AP frequency (P < 0.05; Figure 6B), demonstrating that β-adrenergic receptors are present in iPSC-derived cardiomyocytes and stimulation of these receptors produces a positive chronotropic response.

View this table:
Table 1

Characteristics of action potentials

APA (mV)APD80 (ms)Max. dV/dt (Vs−1)MDP (mV)SAF (Hz)
MSC-iPSC derived
 Pacemaker (n = 7)41.6 ± 3.8312.5 ± 33.54.0 ± 0.4−26.4 ± 3.10.35 ± 0.10
 Ventricle (n = 13)88.4 ± 3.1426.7 ± 24.910.6 ± 2.5−57.9 ± 2.50.36 ± 0.04
 Atrial (n = 4)85.1 ± 5.0225.7 ± 26.911.9 ± 4.1−55.8 ± 2.90.47 ± 0.07
 Purkinje (n = 5)86.9 ± 8.9357.5 ± 33.619.5 ± 6.8−56.3 ± 5.90.45 ± 0.11
kera-iPSC derived
 Pacemaker (n = 4)43.5 ± 5.7342.5 ± 58.93.5 ± 0.3−27.6 ± 2.30.43 ± 0.09
 Ventricle (n = 12)100.8 ± 4.7459.8 ± 27.913.4 ± 3.8−60.6 ± 2.40.44 ± 0.03
 Atrial (n = 4)94.6 ± 5.2267.9 ± 37.915.2 ± 5.3−60.6 ± 2.50.45 ± 0.07
 Purkinje (n = 4)96.0 ± 4.3448.8 ± 115.618.7 ± 6.8−58.3 ± 3.00.52 ± 0.03
FB-iPSC derived
 Pacemaker (n = 6)35.8 ± 6.3297.9 ± 13.33.5 ± 0.2−28.0 ± 4.30.53 ± 0.07
 Ventricle (n = 11)98.5 ± 1.9461.0 ± 43.88.4 ± 1.3−60.7 ± 1.40.59 ± 0.05
 Atrial (n = 4)83.9 ± 5.6276.8 ± 10.39.2 ± 2.0−50.3 ± 1.80.53 ± 0.16
 Purkinje (n = 4)74.9 ± 9.1405.7 ± 107.118.4 ± 12.7−51.9 ± 3.30.62 ± 0.09
  • Data are mean ± SEM. n, the cell number.

Figure 6

Electrophyiological analysis of iPSC-derived cardiomyocytes. (A) Exemplary original traces of action potentials. Distinct action potential morphologies represented pacemaker-, ventricle-, atrial-, and purkinje-like cardiomyocytes. (B) Application of isoproterenol resulted in a significantly increased action potential frequency.

Finally, we assessed the spontaneous intracellular Ca2+ fluctuations in iPSC-derived cardiomyocytes using confocal microscopy. We found that calcium increased homogeneously throughout the cell, pointing to a fine regulated Ca2+ release from intracellular Ca2+ stores, most likely the sarcoplasmic reticulum (SR; Supplementary material online, Video S2). Using the line-scan mode, spontaneous rhythmic Ca2+ transients were found in cardiomyocytes derived from hiPSCs with different cell origins (Figure 7A, Supplementary material online, Figure S3). Moreover, Ca2+ sparks were observed (Figure 7E, Supplementary material online, Figure S3). Again here, we did not observe significant differences regarding spontaneous Ca2+ transient amplitudes and frequencies as well as Ca2+ spark frequencies and amplitudes among cardiomyocytes derived from hiPSCs with different cell origins. In addition, the application of caffeine (10 mmol/L) resulted in a pronounced synchronized increase in cytosolic Ca2+ confirming the presence of ryanodine receptors and SR (Figure 7C). The amplitudes of caffeine-induced Ca2+ transients were significantly larger compared with the spontaneous Ca2+ transient amplitudes (Figure 7D). The fractional release was ∼58% of the total SR Ca2+ content.

Figure 7

Measurements of cytosolic Ca2+ in human-induced pluripotent stem cell-derived cardiomyocytes. (A) Typical line scans of spontaneous Ca2+ transients. The corresponding F/F0 values are displayed below the line scans. (B) Mean data for spontaneous Ca2+ transient decay. The time from Ca2+ peak to 50 or 20% of Ca2+ transient amplitude, respectively (RT50 or RT80) are shown. (C) Caffeine-induced Ca2+ transients. (D) Mean data for the amplitudes of spontaneous Ca2+ transients and caffeine transients. (E) Ca2+ sparks.

Generation of EHTs

To test if EHT can be generated from iPSC-derived cardiomyocytes, we mechanically isolated and digested beating cardiac clusters (Supplementary material online, Video S3) into single cells and re-aggregated them in a hydrogel matrix.22 Spontaneously and synchronously beating EHTs were observed after a 10-day culture period (Supplementary material online, Video S4). We observed that EHTs generated from iPSC-derived cardiomyocytes composed of compact muscle bundles with a high degree of sarcomeric organization and alignment (Figure 5G–I), suggesting advanced maturation of iPSC-derived cardiomyocytes within EHT. Representative force measurements showed a positive Frank–Starling mechanism and positive inotropic response to increasing calcium concentrations (50 µN at 0.2 mmol/L to 240 µN at 2.4 mmol/L calcium) (Figure 5J). Moreover, EHTs responded to stimulation with isoproterenol (1 µmol/L) with an increase in the contractile force and the relaxation velocity (Figure 5K).

Discussion

In the present study, we directly compared three somatic cell sources and two virus systems for the generation of hiPSCs and the functionalities of cardiomyocytes derived from hiPSCs with different cell origins. Our results show that all three cell types can be reprogrammed into hiPSCs, but MSCs and fibroblasts are more easily reprogrammed than keratinocytes. The success rate and reprogramming efficiency was significantly higher by using the STEMCCA system than the OSNL system. All analysed hiPSC lines are pluripotent, show phenotypical characteristics similar to hESCs, and can differentiate into cardiomyocytes with an efficiency ranging from 3 to 42%. However, MSC-iPSCs exhibit significantly higher cardiac differentiation efficiency than Kera- and FB-iPSCs. We did not observe differences in the cardiac differentiation efficiency when we compared the early-passage hiPSCs with the late-passage hiPSCs. There is no significant difference in the functionalities of cardiomyocytes derived from hiPSCs with different origins. Furthermore, we showed the generation of spontaneously and synchronously beating and force-generating EHTs from iPSC-derived beating EBs.

We found that the primary culture of MSCs and fibroblasts have a high proliferation capacity at early passages. Mesenchymal stem cells in a sufficient number for reprogramming can be obtained from 10 mL of bone marrow aspirate already 7–9 days after bone marrow aspiration, whereas sufficient fibroblasts can be obtained 2 weeks after skin biopsy. However, compared with MSCs and fibroblasts, it takes longer to get enough keratinocytes for reprogramming experiments. In addition, we found that the OCT4 and NANOG promoters in keratinocytes are significantly higher methylated than those in MSCs and fibroblasts. These may explain the lower reprogramming efficiency of keratinocytes. Although keratinocyte cultures are not so easy to be cultured and reprogrammed as MSCs and fibroblasts, plucked hairs as the cell source for reprogramming still offer significant advantages over skin biopsies and bone marrow aspirate. Hairs can be easily obtained, whereas a skin biopsy requires a professional surgeon, and bone marrow aspirations needs clinical specialists. Therefore, generating hair keratinocyte cultures allow us to generate hiPSCs from many patients who cannot undergo a skin biopsy, or from those who decline a skin surgeon. Notably, we generated hiPSCs from hair keratinocytes of donors aged over 70 years. In addition, the fact that keratinocytes are sensitive to calcium is an advantage for reprogramming because all emerged colonies in hESC medium are true iPSC colonies without contamination of unreprogrammed cells. In contrast, skin fibroblasts grow so fast that the generation of FB-iPSCs requires mechanical expansion for several passages in order to get rid of fibroblast contamination.

Recently, Kotton's laboratory reported the effective generation of iPSCs by using the STEMCCA system.18,23 Our results showed that both keratinocytes and MSCs were reprogrammed with a higher efficiency by using the STEMCCA system than the OSNL system. Another advantage of the STEMCCA system is allowing the generation of iPSCs free of transgenes by the application of Cre, which may be used for autologous cell replacement procedures in the future.

We demonstrate that our analysed hiPSCs with different cell origins fulfil the criteria defining fully reprogrammed hiPSCs and can differentiate into functional cardiomyocytes similar to HES3, which was reported as one of the best hESC lines for cardiac differentiation.19 Recent studies demonstrate that iPSCs of different origins often retain a ‘footprint’ of their tissue of origin.911 However, as iPSCs are maintained in culture for a long time period, they more closely resemble ESCs, and the ‘epigenetic memory’ of the donor cells is gradually lost in iPSCs upon extended culturing.9 We compared 24 hiPSC lines at passages of >16 with those at early passages (p8–12) for cardiac differentiation. It is noteworthy that the overall cardiac differentiation efficiency is significantly higher with MSC-iPSCs than Kera-, and FB-iPSCs, although inter-individual variability exists. Furthermore, we did not observe significant differences between early and late passages. Comparison of patient-matched iPS cell lines could avoid a confounding variable caused by inter-individual variability. Given the difficulties to obtain different tissues from the same patient, we do not have a big collection of patient-matched hiPSC lines. Nevertheless, we have patient-matched Kera- and FB-iPSC lines from two patients. We found that the cardiac differentiation efficiency of FB-hiPSCs from the two patients is higher than that of Kera-hiPSCs from the same patient. These data indicate that the observed differences in cardiac differentiation efficiency are mainly dependent on the different cell origins.

Our data are in line with a recent report demonstrating that iPSCs derived from mouse ventricular myocytes exhibited a higher cardiac differentiation efficiency than genetically matched ESCs and iPSCs derived from tail-tip fibroblasts, and the differentiation efficiency was not changed significantly after extensive passaging of the cells indicating the relative stability of the memory retained by iPSCs.24 The higher cardiac differentiation potential observed with MSC-iPSCs than Kera-iPSCs and FB-iPSCs may relate to the mesodermal origin of MSCs, which may facility cardiac differentiation than the ectodermal origin of keratinocytes and fibroblasts. To gain insight into the molecular mechanisms by which MSC-iPSCs differentiate into cardiomyocytes with higher differentiation efficiency, a transcriptional and epigenetic analysis based on genome-wide transcriptional profiling and DNA methylation level profiling in iPSCs from fibroblasts, MSCs, and hair keratinocytes is necessary to be completed in the future. Although MSCs are not the easy accessible cell source, their higher reprogramming efficiency and higher cardiac differentiation potential still make them interesting for cardiovascular research.

The contractile properties of the iPSC-derived cardiomyocytes were similar to those derived from hESCs, and did not differ significantly among hiPSCs with different cell origins. Our data showed that the iPSC-derived cardiomyocytes have the complex functional properties of native cardiomyocytes, including a positive response to β-adrenergic stimulation and an intact calcium cycling. Critically important for cardiac function, we also showed that hiPSC-derived cardiomyocytes expressed sarcomeric and gap junction proteins. Action potential analysis demonstrates the presence of pacemaker-, ventricle-, atrial, and purkinje-like cardiomyocytes, similar to those found in cardiomyocytes derived from hESCs and other iPSCs previously reported.25,26 Taken together, our findings demonstrate that cardiomyocytes generated from hiPSCs with different cell origins are structurally and functionally comparable.

It is known that cardiomyocytes in 2D culture remain embryonic based on their size, structural organization, and electric properties.20,27 It has been suggested that the 3D culture may promote the maturation of cardiomyocytes.28 We showed here that EHTs can be generated from hiPSC-derived cardiomyocytes, and that the generated EHTs were characterized by organotypic properties such as spontaneous coherent beating, Frank–Starling mechanism, and positive inotropic response to calcium and isoproterenol. Future studies are needed, however, to address the feasibility of EHT to advance the maturation of iPSC-derived cardiomyocytes.

In summary, the present study shows that hiPSCs can be generated from three different somatic cell types with different advantages and disadvantages. All analysed iPSCs can differentiate into cardiomyocytes with efficiency ranging from 3 to 42%. Functional cardiomyocytes with structurally and functionally comparable properties to those obtained from ESCs can be generated from hiPSCs with different cell origins. Furthermore, circular EHTs can be generated. These findings indicate that hiPSCs hold great promise in investigating cardiac diseases and drug screening, and as potential autologous cell sources for myocardial regeneration.

Funding

This work was supported by the Anschubsfinanzierung from Universitätsmedizin Göttingen (K.G.), by Deutsches Zentrum für Herzkreislaufforschung (DZHK), by Bundesministerium für Bildung und Forschung grants 01GS0837 (K.G.), 01GN0822 (K.G. and G.H.), and 01GN0957 (K.G., W.Z., G.H.), and by DFG grants GU595/2–1 (K.G.) and SFB1002, TP A04 (K.G.).

Conflict of interest: none declared.

Acknowledgements

We thank Anke Cierpka, Yvonne Hintz, Thomas Sowa, Sandra Georgi, and Yvonne Wiegeräfe for excellent technical assistance.

Footnotes

  • These authors contributed equally.

References

View Abstract