SEARCH

SEARCH BY CITATION

Keywords:

  • Cardiac;
  • Embryonic;
  • Stem cells;
  • Coculture;
  • Endoderm

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Human embryonic stem cells (hESCs) can differentiate into cardiomyocytes, but the efficiency of this process is low. We routinely induce cardiomyocyte differentiation of the HES-2 cell line by coculture with a visceral endoderm-like cell line, END-2, in the presence of 20% fetal calf serum (FCS). In this study, we demonstrate a striking inverse relationship between cardiomyocyte differentiation and the concentration of FCS during HES-2-END-2 coculture. The number of beating areas in the cocultures was increased 24-fold in the absence of FCS compared with the presence of 20% FCS. An additional 40% increase in the number of beating areas was observed when ascorbic acid was added to serum-free cocultures. The increase in serum-free cocultures was accompanied by increased mRNA and protein expression of cardiac markers and of Isl1, a marker of cardiac progenitor cells. The number of beating areas increased up to 12 days after initiation of coculture of HES-2 with END-2 cells. However, the number of α-actinin–positive cardiomyocytes per beating area did not differ significantly between serum-free cocultures (503 ± 179; mean ± standard error of the mean) and 20% FCS cocultures (312 ± 227). The stimulating effect of serum-free coculture on cardiomyocyte differentiation was observed not only in HES-2 but also in the HES-3 and HES-4 cell lines. To produce sufficient cardiomyocytes for cell replacement therapy in the future, upscaling cardiomyocyte formation from hESCs is essential. The present data provide a step in this direction and represent an improved in vitro model, without interfering factors in serum, for testing other factors that might promote cardiomyocyte differentiation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Cardiomyocytes in the adult mammalian heart are essentially terminally differentiated and do not divide. Although a small percentage of the cells may be capable of proliferation [1], this is not sufficient for regeneration after myocardial injury. Conventional pharmacological therapy for patients with different stages of ischemic heart disease improves cardiac function, survival, and quality of life, but the ensuing failure is still the most life-threatening disease in Western society. Alternative therapies will be necessary to improve the clinical outcome of the increasing number of patients with ischemic heart disease. In recent years, cell replacement therapy has received considerable attention, intensified by the increasing number of potential cell sources for transplantation, which include skeletal myoblasts, adult cardiac stem cells, bone marrow stem cells, and embryonic stem cells (ESCs) [2].

ESCs can differentiate to all somatic cell types of the adult. Since the first description of the derivation of human ESCs (hESCs) from donor blastocysts [3, 4], we and others have reported their differentiation to cardiomyocytes in culture [58]. Recently, we demonstrated that coculture of hESCs with a visceral endoderm-like cell line (END-2), derived from mouse P19 embryonal carcinoma (EC) cells [9], resulted in the appearance of beating areas. Most (85%) of these hESC-derived cardiomyocytes had a ventricle-like phenotype based on morphological and electrophysiological parameters [5]. Others have reported the spontaneous differentiation of hESCs, cultured as aggregates or embryoid bodies [68] and enhancement of differentiation by the demethylating agent 5-aza-deoxycytidine [8]. Between 8% [6] and 70% [8] of the embryoid bodies showed beating areas in these studies, and 2%–70% of the beating areas consisted of cardiomyocytes. This wide variation in cardiomyocyte differentiation and the relative paucity of quantitative data make it difficult to compare these in vitro models.

In this study, we describe a striking enhancement of cardiomyocyte differentiation in serum-free hESC-END-2 coculture conditions compared with our previous standard coculture in 20% fetal calf serum (FCS). Quantification of the number of cardiomyocytes under these coculture conditions showed a significant increase in the yield of cardiomyocytes without genetic manipulation.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Cell Culture

END-2 cells and hESC lines hES2, hES3, and hES4 (passage numbers 41–84) were cultured as described previously [4, 5]. To initiate cocultures, END-2 cell cultures, treated for 3 hours with mitomycin C (10 μg/ml), replaced mouse embryonic fibroblasts (MEFs) as feeders for hESCs [5, 9]. As controls, hESCs were grown on MEFs for the same period under the same culture conditions. In standard cocultures, cells were grown in 12-well plates in Dulbecco's modified Eagle's medium (DMEM) containing L-glutamine, insulin-transferrin-selenium (ITS), nonessential amino acids, 90 μM β-mercaptoethanol, penicillin/streptomycin, and 20% FCS (Multicell, Wisent Inc., Saint-Jean-Baptiste de Rouville, Quebec, Canada, http://www.wisent.ca). Cocultures were then grown for up to 3 weeks and scored for the presence of areas of beating muscle from 5 days onward. To study the effect of FCS on cardiomyocyte differentiation, concentrations of FCS ranging from 0%–20% were compared with standard coculture conditions. To determine when the presence or absence of FCS might be critical for cardiomyocyte differentiation, hESC-END-2 cocultures were exposed to 20% FCS for the first 6 days and then 0% FCS for the last 6 days, or vice versa. In addition, instead of FCS, various concentrations of knockout serum replacement (KSR) were used during the cocultures. Finally, serum-free coculture experiments were carried out in the absence of insulin or ITS or in the presence of 10−4 M ascorbic acid (Sigma, St. Louis, http://www.sigmaaldrich.com).

Primary Human Adult and Fetal Cardiomyocytes

Primary tissue was obtained during cardiac surgery or following abortion after individual permission using standard informed consent procedures and approval of the ethics committee of the University Medical Center, Utrecht, Netherlands. Cardiomyocytes were isolated from fetal hearts (16–17 weeks of gestation) perfused by Langendorff's method and cultured on glass coverslips.

Western Blotting

Three wells of a 12-well plate containing 12-day hESC-END-2 cocultures, as well as 5-day cultures of human fetal hearts, were washed twice in phosphate-buffered saline (PBS) and collected in 500-μl RIPA-buffer. Protein concentrations were measured by bicinchoninic acid protein assay (Pierce, Rockford, IL, http://www.piercenet.com). Fifty micrograms of protein from human fetal hearts and 80 μg of hESC-END-2 cocultures were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Blots were incubated with antibodies against sarcomeric tropomyosin (monoclonal, 1:400; Sigma) and troponin T-C (goat polyclonal, 1:500; Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com). Proteins were visualized using electrogenerated chemiluminescence.

Immunohistochemistry

hESC-END-2 cocultures were grown in 12-well plates with 20% or 0% FCS on gelatin-coated coverslips. After 12 days, dissected beating areas or whole coverslips were fixed with 2.0% para-formaldehyde for 30 minutes at room temperature. Fixed beating areas were processed for immunohistochemistry, and 4-μm paraffin sections were made [10]. Endogenous peroxidase was blocked in 1.5% H2O2 in water, followed by antigen retrieval in citrate buffer. Sections were incubated with an antibody against Isl1 (mouse monoclonal 39.4 D5: 1:1000; Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww) and then a secondary goat–anti-mouse antibody (Powervision, ImmunoLogic, Duiven, Netherlands) visualized by 3,3′-diaminobenzidine (Sigma); sections were counter-stained with hematoxylin. For immunofluorescence, cells were permeabilized with 0.1% triton X-100 and stained overnight at 4°C with α-actinin (monoclonal, 1:800; Sigma) and α-Troma-1 (rat monoclonal, 1:10; Developmental Studies Hybridoma Bank) and used in combination with fluorescent-conjugated secondary antibodies (Jackson Immuno Research Laboratories, West Grove, PA, http://jackson.immuno.com). To visualize nuclei, cells were incubated with Topro-3 (1:1,000) in 0.002% Triton.

Quantification of α-Actinin–Positive Cells

Confocal images (Leica Systems) (× 10, 20, and 63 objectives) from two-dimensional projected Z-series at 10-μm intervals were made of α-actinin–positive cell cultures. Nuclei in these areas were counted. Care was taken to avoid counting the same cells in different planes. Only nuclei surrounded by α-actinin staining in a cardiomyocyte-like striated pattern were scored as positive. All counts were performed double-blind. For immunofluorescent staining of α-actinin on single cells, beating areas were dissected, followed by dissociation as described previously [5]. Cells were grown on gelatin-coated coverslips for 7 days.

Reverse Transcriptase–Polymerase Chain Reaction

hESC-END-2 cocultures with 20% or 0% FCS were washed in PBS and RNA from five wells pooled using Trizol (Sigma). Total RNA, 500 ng, was reverse transcribed (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for polymerase chain reaction (PCR) using Silverstar DNA polymerase (Eurogentec, SanDiego, http://usa.eurogentec.com). Primer sequences and PCR conditions for α-actinin, atrial natriuretic factor, MLC2a, phospholamban, and β-actin were described previously [5]. Primer sequences for Nkx2.5 were ggtggagctggagaagacaga (sense), cgacgccgaagttcacgaagt (antisense) (536bp); for GATA-4, accagcagcagcgaggagat (sense), gagagatgcagt-gtgctcgt (antisense) (512 bp); and for α-myosin heavy chain (MHC), ggggacagtggtaaaagcaa (sense), tccctgcgttccactatctt (antisense) (542 bp). PCR was performed at 55°C (annealing temperature) at 1.5 mM MgCl2 for 30 cycles. Products were analyzed on ethidium bromide–stained 1.5% agarose gel. β-Actin was used as RNA input control.

Real-Time Quantitative PCR

Real-time PCR was performed according to standard protocols on a MyIQ Real Time PCR detection system (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Briefly, 1 μg of total RNA was DNAse treated and transcribed to cDNA. Ten microliters of a 1/10 dilution of cDNA was then added to 12.5 μl of the 2 X SYBR green PCR master mix (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) and 500 μM of each primer. PCR was performed for α-actinin (sense primer: ctgctgctttg-gtgtcagag; antisense primer: ttcctatggggtcatccttg), Isl1 (sense primer: tgatgaagcaactccagcag; antisense primer: ggactggctac-catgctgtt), and acidic ribosomal phosphor-protein PO (HARP) (sense primer: caccattgaaatcctgagtgatgt; antisense primer: tgaccagcccaaaggagaag) as an internal control. PCR cycles for α-actinin, Isl1, and HARP were 3 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C, 30 seconds at 62.5°C, and 45 seconds at 72°C. The thermal denaturation protocol was run at the end of PCR to determine the number of products. Samples were run on a 2% agarose gel to confirm the correct size of the PCR products. All reactions were run in triplicate. As negative controls, PCR was performed on water and on RNA without reverse transcription. The cycle number at which the reaction crossed an arbitrarily placed threshold (Cτ) was determined for each gene. The relative amount of mRNA levels was determined by 2– Cτ. Relative gene expression was normalized to HARP expression.

Statistical Analysis

All data are presented as mean ± standard error of the mean, unless stated otherwise. Statistical significance of differences was calculated using a Student's t-test. Significance was accepted at the level of p < .05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Effect of Serum on Morphology and the Number of Beating Areas During Coculture

The results shown were consistent in all three hESC lines examined (HES-2, -3, and -4). Data presented are from HES-2 cells. To determine the effect of serum on the number of beating areas during coculture of hESCs with END-2 cells, the serum concentration was reduced to 10%, 5%, 2.5%, and 0% from the start of coculture on day 1 until the end at day 12 and compared with the number of beating areas in a 12-well plate in the standard 20% FCS conditions. As shown in Figure 1, examination of hESC morphology after 5 days in coculture with 20% FCS demonstrated three-dimensional structures with cells spreading out from them (Fig. 1A). After 12 days of coculture, this was more evident, and strings of differentiating hESCs were visible (Fig. 1B). In the absence of serum, the edges of the three-dimensional structures were clearer, and less outward spreading of cells was observed (Figs. 1C, 1D). hESCs cultured on MEF feeders for an additional 12 days in the presence or absence of serum resulted in fewer cells on day 5 (Fig. 1E) compared with hESCs on END-2 cells (Figs. 1A, 1C), but not in the formation of three-dimensional structures. After 12 days, hESCs had spread out but remained predominantly as a two-dimensional sheet (Fig. 1F).

thumbnail image

Figure Figure 1.. Morphology of hESCs during END-2 or MEF cocultures. Morphology of hESCs is shown after coculture with (A–D) END-2 or (E–F) MEF cells for (A, C, E) 5 days and (B, D, F) 12 days in the (A, B) presence or in the (C–F) absence of 20% fetal calf serum; magnifications ×5. Abbreviations: hESC, human embryonic stem cell; MEF, mouse embryonic fibroblast.

Download figure to PowerPoint

Besides these morphological differences, a significant increase in the number of beating areas was observed at lower concentrations of serum, with a 24-fold upregulation in its complete absence compared with cultures containing 20% FCS (Fig. 2A). On average, 1.35 ± 0.26 (n = 21) beating areas per plate were observed at day 12 in 20% FCS cocultures, whereas 32.7 ± 2.3 (n = 27) beating areas were observed in 0% FCS cocultures. Beating areas were normally observed from day 7 onward (occasionally as early as day 5 or 6), with an increase in the number of beating areas until day 12 under all culture conditions. From day 12 onward, additional beating areas appeared, but at a much lower rate (Fig. 2B).

thumbnail image

Figure Figure 2.. Effect of serum or KSR on the number of beating areas in hESC-END-2 cocultures. (A): Cocultures were initiated in 12-well plates in different concentrations of FCS, and beating areas were counted 12 days later or were counted from (B) days 8–18. (C): hESC-END-2 cocultures were performed in 0% FCS for the first 6 days and in 20% FCS for the next 6 days [0+20(d6)] and vice versa [20+0(d6)]. Beating areas were scored on day 12 and compared with 20% FCS and 0% FCS cocultures. The relative increase as fold-induction with respect to 20% FCS cocultures is shown. (D): Different concentration of KSR is added to hESC-END-2 cocultures, and beating areas are scored on day 12 and compared with 0% FCS cocultures. Each culture condition was tested in at least three independent experiments. *p < .05; **p < .01; ap < 10−12 compared with 20%; ###p < .001 compared with 20+0(d6). Abbreviations: FCS, fetal calf serum; hESC, human embryonic stem cell; KSR, knockout serum replacement.

Download figure to PowerPoint

To study whether the absence of serum was important throughout the 12-day coculture period, hESC-END-2 coculture was initiated in 0% FCS and then 20% FCS was added at day 6. Conversely, cocultures were also initiated in the presence of 20% FCS and changed to 0% FCS, at day 6. In cocultures starting in 0% FCS and changed at day 6 for 20% FCS, the number of beating areas decreased to 57% compared with cocultures maintained in 0% FCS continuously. However, in the cocultures in 20% FCS for the first 6 days, the number of beating areas decreased to only 2% compared with those in 0% FCS continuously (Fig. 2C).

An alternative to serum-free culture is the use of KSR. Various concentrations of KSR were added to hESC-END-2 cocultures. As shown in Figure 2D, a significant inverse relationship was found between the concentration of KSR in culture medium and the number of beating areas, just as in the FCS-supplemented medium. The elimination of insulin or ITS from the serum-free medium during coculture did not further affect the number of beating areas compared with serum-free medium alone (data not shown).

Expression of Cardiac Genes and Proteins in 20% and 0% hESC-END-2 Cocultures

To determine whether the increase in the number of beating areas resulted in a comparable increase in the expression of cardiac genes and proteins, reverse transcription (RT)–PCR and Western analysis were performed on hESC-END-2 cocultures in 0% and 20% FCS. A clear increase in the expression for all cardiac genes was observed by RT-PCR in the 0% FCS cocultures compared with those in 20% FCS (Fig. 3A). Nkx2.5, a homeobox-domain transcription factor, which plays an important role in early cardiac development, was slightly upregulated, whereas the cardiac zinc-finger transcription factor GATA-4 was not changed by 0% FCS compared with 20% FCS cocultures.

thumbnail image

Figure Figure 3.. Effect of serum concentration on the expression of cardiac genes and proteins in hESC-END-2 cocultures. (A): Reverse transcription–polymerase chain reaction on RNA from 12-day hESC-END-2 cocultures in 0% FCS or 20% FCS. (B): Real-time polymerase chain reaction for α-actinin in 0% FCS (n = 3) and 20% FCS (n = 2) hESC-END-2 cocultures using HARP mRNA levels as an internal control. (C): Western blot of protein extracts from 12-day hESC-END-2 cocultures in 0% FCS or 20% FCS and from HFCMs using antibodies against TM and Trop. Abbreviations: ANF, atrial natriuretic factor; FCS, fetal calf serum; hESC, human embryonic stem cell; HFCM, human fetal cardiomyocyte; MHC, myosin heavy chain; MLC, myosin light chain; P-Lamban, phospholamban; TM, tropomyosin; Trop, troponin T-C.

Download figure to PowerPoint

To confirm the results of the semiquantitative RT-PCR, mRNA levels for α-actinin in 0% and 20% FCS cocultures were accurately measured by real-time RT-PCR. PCR was performed in triplicate for each sample. As an internal control, HARP mRNA levels were determined. Standard deviations were less than 1% for all triplicate reactions. A 27-fold increase in α-actinin mRNA levels was observed in the 0% FCS cocultures compared with the 20% FCS cocultures (Fig. 3B), confirming the results of the RT-PCR.

Increased expression of cardiac structural proteins in 0% FCS cocultures was confirmed by Western blot analysis. In cocultures in 20% FCS, both tropomyosin and troponin T-C are not detectable or are at the detection limit of the assay, whereas in cocultures in 0% FCS, clear bands at 36 kDa for tropomyosin and 40 kDa for troponin T-C were observed. As expected, an even stronger band at the same molecular weight was observed in protein extracts from human fetal hearts (Fig. 3C).

Characterization of Beating Areas and the Presence of Cardiac Progenitor Cells

After 12 days, cocultures in 0% FCS were examined for the presence of beating areas and recorded on video (Fig. 4A). The same samples were then fixed and stained for α-actinin (Fig. 4B) and the films overlayed. All beating areas were also positive for α-actinin and displayed a characteristic cardiomyocyte-like striated pattern (Fig. 4C). No α-actinin–positive areas were detected that were not beating before fixation, indicating the high correlation between the number of beating areas and the number of α-actinin–positive areas. After dissection of beating areas and subsequent dissociation, cells were plated on gelatin-coated dishes, fixed, and stained for α-actinin. Between 5% and 20% of the cells were positive for α-actinin (Fig. 4D). Most of the other cells were positive for Troma-1, which recognizes intermediate cytokeratin 8 and is used as a marker for endoderm (Fig. 4E). By doublestaining immunofluorescence, it is clear that α-Troma-1–positive and α-actinin–positive cells do not colocalize (Fig. 4F)

thumbnail image

Figure Figure 4.. Relationship between beating areas with α-actinin staining and cardiomyocytes after dissociation. (A): Beating hESC-END-2 12-day cocultures from one well are recorded and then fixed and stained for α-actinin. (B): Identical areas are indicated by white dashed lines and are labeled a–e; ×5 magnification. (C): Magnification ×63 of white dashed box of (B). (D): Dissociated cell of beating areas stained for α-actinin (green) and Topro-3 (blue) (×40 magnification). (E): Dissociated and replated cells derived from beating areas stained for Troma-1 (green) and Topro-3 (blue) or α-actinin (red) (F).

Download figure to PowerPoint

To determine whether cardiac progenitor cells had formed during differentiation in the hESC-END-2 cocultures, as might be expected, we determined the expression of Isl1. By real-time PCR, a 2.5-fold increase in the expression of Isl1 was found in serum-free hESC-END-2 cocultures at day 12 compared with that of 20% FCS cocultures (Fig. 5A). By immunohistochemistry, we confirmed that nuclear Isl1 protein expression is present in tissue sections of 12-day beating areas (Figs. 5B–5D).

thumbnail image

Figure Figure 5.. Expression of Isl1 in hESC-END-2 cocultures. (A): Real-time polymerase chain reaction for Isl1 in 0% FCS (n = 2) and 20% FCS (n = 2) 12-day hESC-END-2 cocultures using HARP mRNA levels as an internal control; *p < .05. (B–D): Isl1 protein localization by immunohistochemistry in 4-μm sections of 12-day beating areas from serum-free hESC-END-2 cocultures; magnification (B, C) ×20 or (D) ×40. Abbreviations: FCS, fetal calf serum; hESC, human embryonic stem cell.

Download figure to PowerPoint

Number of Cardiomyocytes in Cocultures

To determine whether the increase in the number of beating areas and the increase in cardiac gene and protein expression was attributable to the increase of the actual number of cardiomyocytes, α-actinin–positive cells with striated sarcomeric patterns were counted by confocal Z-series. This was considered more informative than fluorescence-activated cell sorter (FACS) analysis, because cells showing striated α-actinin staining could be selectively included. Cells were counted in different optical planes (Fig. 6A). In cocultures in 20% FCS, the average number of cardiomyocytes per beating area was 312 ± 227 (n = 5). The number of cardiomyocytes per beating area in 0% FCS cocultures was 503 ± 179 (n = 15). However, this was not significantly different and reflects the wide variation in the number of cardiomyocytes per beating area (ranging from 1 to 2,500 cells) (Fig. 6B). Based on these numbers, the average number of cardiomyocytes in a 12-well coculture plate is therefore approximately 16,600 cells in 0% FCS cocultures and 450 cells in 20% FCS cocultures, representing a 39-fold increase in the total number of cardiomyocytes in 0% FCS cocultures (Table 1).

Table Table 1.. Number of cardiomyocytes in 0% and 20% FCS hESC-END-2 cocultures
  1. a

    Total number of cardiomyocytes from 12-well plate, 12-day hESC-END-2 cocultures. hESC input represents the estimated number of undifferentiated hESCs used per 12-well plate.

  2. b

    Abbreviations: BA, beating area; CM, cardiomyocytes; FCS, fetal calf serum; hESC, human embryonic stem cell.

 hESC-END-2 coculture
 20% FCS0% FCS
hESC input∼300,000∼300,000
BA/plate1.4 ± 0.333 ± 2.3
CM/BA312 ± 227503 ± 179
CM/plate∼45016,600
thumbnail image

Figure Figure 6.. Number of cardiomyocytes in 0% and 20% FCS hESC-END-2 cocultures. (A): BA of hESC-END-212-day coculture, stained for α-actinin (red) and Topro-3 (nucleus, blue) in different planes after confocal scanning (I and I'). Only nuclei surrounded by α-actinin are counted. Examples are given (white arrows); ×20 magnification. (B): Numbers of cardiomyocytes from 0% FCS and 20% FCS hESC-END-2 cocultures are counted and pooled from the different confocal planes. (C): Cocultures were initiated in 12-well plates in serum-free hESC-END-2 with or without AA (n = 6). BAs were scored on day 12; *p < .05. Abbreviations: AA, ascorbic acid; BA, beating area; FCS, fetal calf serum; hESC, human embryonic stem cell.

Download figure to PowerPoint

The serum-free hESC-END-2 coculture condition represents an improved culture model, without inhibitory factors from serum, for testing other factors for their effect on cardiomyocyte differentiation. Addition of 10−4 M ascorbic acid to serum-free hESC-END-2 cultures, for example, resulted in a further robust increase in the number of beating areas at day 12, 40% higher than in the serum-free cocultures alone (Fig. 6C).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

hESCs can differentiate to cardiomyocytes either spontaneously by growing them as aggregates or embryoid bodies in suspension [68] or by growing them in coculture with an endoderm-like cell line, END-2 [5]. The reported efficiency of spontaneous cardiomyocyte differentiation varies between 8% [6] and 70% [8] of the embryoid bodies contracting and reaches a maximum between days 16 and 30 of differentiation (growth of embryoid bodies in culture followed by plating on gelatin-coated dishes). The percentage of cardiomyocytes in dissected and dissociated beating areas has also been reported to vary widely, between 2% and 70% [7]. After Percoll gradient centrifugation, Xu et al. [8] obtained a cell fraction consisting of 70% sarcomeric MHC–positive cells, as determined by immunohistochemistry. Under our initial coculture conditions in the presence of 20% FCS, we observed that approximately 16% of the wells of hESCs from passages 41 through 84 contained beating areas. The variation in efficiencies reported for spontaneous differentiation has been substantial. In addition, methods for standard quantification of the number of cardiomyocytes have been lacking. This has made it difficult to compare the efficiencies of spontaneous versus induced cardiomyocyte differentiation directly.

In this study, we have described a method to increase in the number of beating areas by reducing FCS in the medium and have used a variety of assays to quantify this effect with consistent outcomes. After 12 days of coculture, the number of beating areas is 24-fold higher in the absence of FCS compared with 20% FCS. The total number of cardiomyocytes from a 12-well plate is approximately 16,600 cardiomyocytes for cocultures in 0% FCS, a 39-fold enrichment in the total production of cardiomyocytes per plate compared with serum-containing cultures. The effect of the absence of serum during cocultures was observed in all hESC lines examined (HES-2, -3, and -4), suggesting a generally applicable method for improved cardiomyocyte differentiation.

The permissive effect of serum-free culture conditions on differentiation for a variety of cell types in culture has been described. Skeletal myoblasts are induced to differentiate by the withdrawal of serum [11]. In addition, undifferentiated neuroblastoma cells form neurites in serum-free medium [12]. Recently, the inhibitory effect of serum on cardiomyocyte differentiation of mouse ESCs has been described [13]. In that study, replacement of 0.2% FCS/DMEM with serum replacement-2/DMEM, containing insulin, transferrin, and heat-treated bovine serum albumin, resulted in an approximately 4.5-fold increase in the percentage of embryoid bodies that were beating. In addition, the amount of cardiac MHC (cMHC) α/β, determined by chemiluminescence, was upregulated sixfold. When 0.2% FCS/DMEM was replaced with 10% FCS/DMEM after 2 days, no beating areas nor expression of cMHCα/β was observed. Most protocols for cardiomyocyte differentiation from mouse ESCs use 20% FCS/DMEM [14]. This suggests that cardiogenic stimulatory as well as inhibitory factors may present in serum. Depending on when serum is present and its concentration during culture, cardiac differentiation was either stimulated or inhibited. This is in agreement with our data on hESCs: The absence of serum promoted cardiomyocyte differentiation throughout the 12 days of hESC-END-2 coculture. However, serum-free differentiation conditions clearly had a greater effect on the number of beating areas during the first 6 days of coculture.

The increase in the number of beating areas in serum-free conditions suggested a greater efficiency in cardiomyocyte differentiation. The fact that the expression of cardiac genes and proteins and the number of striated α-actinin–positive cardiomyocytes was significantly increased largely excluded the explanation that the increase in the number of beating areas is only due to maturation in the organization of sarcomeric contractile units, although it cannot be excluded that increased cardiomyocyte maturation contributes to the effects observed [5, 9, 15].

Recently, Isl1, an LIM homeodomain transcription factor, was described as being important for cardiac development. Mice lacking Isl1 miss the outflow tract, right ventricle, and much of the atrial tissue. Isl1 expression marks a distinct subset of undifferentiated cardiac progenitor cells [16]. In this study we have shown that at day 12 in serum-free hESC-END-2 cocultures, expression of Isl1 mRNA increased 2.5-fold compared with serum-containing cultures. Also at the protein level, Isl1 could be detected in sections of day-12 beating areas. It is tempting to speculate that an increase in the number of cardiac progenitor cells present in serum-free hESC-END-2 cultures results in the increase in the number of beating cardiomyocytes. The observation that day-12 beating areas contain Isl1-positive cells suggests that a useful approach to a further improvement in cardiomyocyte differentiation efficiency might be to identify culture conditions that promote expansion of this Isl1-positive population. Of the components of the differentiation medium, insulin or insulin-like growth factors have been shown to have a positive effect on skeletal as well as cardiac differentiation [17, 18]. We therefore performed cocultures in serum-free medium without insulin or ITS. The average number of beating areas was not affected by the absence of insulin or ITS; if anything, an incidental increase on the number of beating areas was observed (data not shown).

Ascorbic acid has also been shown to enhance cardiomyocyte differentiation in mouse ESCs [19] and more recently in human bone marrow stem cells [20]. This is in accordance with the data here, in which an additional 40% increase in the number of beating areas was observed in serum-free hESC-END-2 cocultures in the presence of ascorbic acid.

Previously, we have shown that the visceral-endoderm–like cell line, END-2, induces mouse P19 EC, mouse ESCs, and hESCs to aggregate in coculture and give rise to cultures containing beating areas [9, 20, 21]. For mouse P19 EC cells, it has been established that direct contact between the two cell types is not necessary and that a diffusible factor, secreted by the END-2 cells, is responsible for the induction of cardiomyocyte formation [9]. Indian hedgehog, secreted by END-2 cells, was shown to be responsible for respecification of prospective neuroectodermal cell fate in mouse epiblast cells along the hematopoietic and endothelial lineages [22]. In this study we demonstrate that, in addition to the presence of cardiomyocytes, most of the differentiated hESCs are Troma-1–positive endodermal-like cells. This suggests that cardiomyocyte differentiation from hESCs by END-2 cells could be mediated either directly by END-2 cells or by hESC-derived endodermal cells. The END-2–derived factor or factors affecting cardiomyocyte differentiation are at present unknown. However, the fact that we never observed beating areas in hESCs cultured long-term on MEFs or on gelatin-coated dishes even in serum-free media suggests that differentiation-inducing activity is derived specifically from END-2 cells or is a result of hESC-END-2 interaction. This may mimic embryonic development in vivo in which functional anterior visceral endoderm has been shown to be essential for normal cardiac development [23].

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

We demonstrated an improved efficiency of cardiomyocyte differentiation from hESC-END-2 cocultures in serum-free medium. Serum-free hESC-END-2 coculture represents a more defined in vitro model for identifying the cardiomyocyte-inducing activity from END-2 cells and, in addition, a more straightforward experimental system for assessing potential cardiogenic factors such as bone morphogenic proteins, fibroblast growth factors, Wnts, and their inhibitors, apart from ascorbic acid as tested here, because there will be no interference from serum-derived modulatory factors.

After dissociation, between 5% and 20% of the cells were α-actinin–positive cardiomyocytes. This variation can be attributed to many different factors such as the size of the beating area, the number of cardiomyocytes per beating area, and the accessibility of the beating area. In addition, cell death and altered attachment during or after dissociation and time between plating and fixation of dissociated cells may play a role in determining the percentage of cardiomyocytes in the replated dissociated cells (higher proliferation rates of noncardiomyocytes reduce the proposition of cardiomyocytes present). Therefore, selection of cardiomyocytes by FACS using cell-surface markers or by genetic manipulation will further stimulate the use of hESC-derived cardiomyocytes for cell-replacement studies.

The higher number of hESC-derived cardiomyocytes in these cultures will not only provide us with a better in vitro model for understanding cardiac development in humans but will also facilitate upscale for transplantation studies to determine whether hESC-derived cardiomyocytes can survive and functionally integrate with host cardiomyocytes and improve cardiac function in animal models of heart failure. With respect to possible future clinical applications, it is of importance that cardiomyocyte differentiation is feasible in serum-free conditions and thus reduces the risk of cross transfer of animal pathogens. On that note, an alternative for serum, KSR, inhibited the number of beating areas, but upon withdrawal, the number of beating areas again increased (data not shown). This suggests that maintenance of undifferentiated hESCs in the presence of KSR (which would be favorable for future clinical applications), followed by serum-free differentiation cultures, would not affect cardiomyocyte differentiation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

This study was supported by ESI International (J.S. and J.K.) and the Interuniversity Cardiology Institute of the Netherlands. We thank S. van den Brink and R. Carvalho for experimental help.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References