SEARCH

SEARCH BY CITATION

Keywords:

  • Human embryonic stem cells;
  • Differentiation;
  • Cardiomyocytes;
  • Transcriptional profiling;
  • Heart development

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Human embryonic stem cells (hESC), with their ability to differentiate into cardiomyocytes in culture, hold great potential for cell replacement therapies and provide an in vitro model of human heart development. A genomewide characterization of the molecular phenotype of hESC-derived cardiomyocytes is important for their envisioned applications. We have employed a lineage selection strategy to generate a pure population of cardiomyocytes (>99%) from transgenic hESC lines. Global gene expression profiling showed that these cardiomyocytes are distinct from pluripotent and differentiated hESC cultures. Pure cardiomyocytes displayed similarities with heart tissue, but in many aspects presented an individual transcriptome pattern. A subset of 1,311 cardiac-enriched transcripts was identified, which were significantly overpresented (p < .01) in the Gene Ontology (GO) categories of heart function and heart development. Focused analysis of the GO categories ion transport, sarcomere, and heart development uncovered a unique molecular signature of hESC cardiomyocytes. Pathway analysis revealed an extensive cardiac transcription factor network and novel peroxisome proliferator-activated receptor signaling components within the cardiac-enriched genes. Notably, approximately 80% of these genes were previously uncharacterized. We have evaluated the biological relevance of four candidates—Rbm24, Tcea3, Fhod3, and C15orf52—by in situ hybridization during early mouse development and report that all were prominently expressed in cardiac structures. Our results provide the fundamental basis for a comprehensive understanding of gene expression patterns of hESC cardiomyocytes and will greatly help define biological processes and signaling pathways involved in hESC cardiomyogenic differentiation and in human heart development. STEM CELLS 2009;27:2163–2174


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Human embryonic stem cells (hESCs) are a source for generating a virtually unlimited supply of cardiomyocytes for cell replacement therapies. They also serve as an in vitro model to investigate the molecular processes of embryonic development, lineage commitment, and cell differentiation.

Recently, microarray technology was employed to study the gene expression profiles of hESCs undergoing some extent of cardiomyogenic differentiation [1, 2], thereby providing new insights into molecular mechanisms underlying this process. However, these studies relied on manual enrichment of cardiomyocytes by mechanical dissection of contracting areas [1, 2], or limited enrichment by a Percoll gradient [3]. The molecular signature generated by these approaches represents that of a mixed cell population. Notably, Synnergren et al. [2] used a transformed hESC cell line in their study, which might further limit the general applicability of the published results.

To overcome prior limitations, we investigated highly enriched hESC-derived cardiomyocyte populations established by means of lineage selection [4]. A construct comprising the cardiomyocyte-restricted α-myosin heavy chain (α-MHC) promoter driving a neomycin-resistance cassette was introduced into hESC to establish stable transgenic cell lines. Combining an efficient differentiation protocol in suspension culture [4, 5] with antibiotic selection resulted in more than 99% cardiomyocyte purity as evidenced by immunocytology. In the present study we have, for the first time, investigated the global gene expression profile of these pure cardiomyocytes. Using fetal and adult heart tissue as a reference, we provide evidence that the “molecular signature” of hESC-derived cardiomyocytes most closely resembles that of human heart, but in many aspects displays properties of developmental-stage cardiomyocytes. We have identified novel cardiomyocyte-specific genes and characterized their expression pattern in the developing mouse embryo. New biological pathways potentially involved in early cardiac differentiation were revealed. Functional analysis confirmed that hESC-derived cardiomyocytes exhibit the expected physiological properties that were not significantly altered by our antibiotic-based enrichment. The study provides a framework of cardiomyocyte characterization on the transcriptome level, facilitating their envisioned application in safety pharmacology and regenerative medicine.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

hESC Culture, Differentiation, and Cardiomyocyte Selection

Generation and expansion of α-MHC-neo/pGK-hygro transgenic hESC lines from hES3 were performed as described [4]; transgenic lines displayed a normal karyotype (46 XX). hESC differentiation was conducted according to an established protocol [4]. At day 3, the culture was transferred to an orbital rotation device set at 45 rounds per minute [6]. Medium changes were performed every 3 days and selection was initiated by 400 μg/ml G418 (Invitrogen) addition at day 12. Antibiotic selection was maintained for 9 days.

Illumina Gene Chip Analysis

Nine experimental cell samples and six commercial RNA samples were used for expression profiling as described in supporting information Table 1 and Figure 1A. Microarray experiments were performed using Sentrix Human-8 Expression BeadChips (Illumina, San Diego, CA, //www.illumina.com), which analyzed 22,000 transcripts, according to manufacturer's instructions. In brief, total RNA (2 μg) from each sample of selected cardiomyocytes (CM), nonselected embryoid bodies (EB), hESC (ES), fetal heart (FH), and adult heart (AH) were converted to biotinylated cRNA using the Illumina RNA Amplification Kit. Each sample cRNA (750 ng) was hybridized to arrays at 58°C overnight following the Illumina Whole-Genome Gene Expression Protocol for BeadStation. Hybridized biotinylated cRNA was detected with streptavidin-Cy3 (Amersham Biosciences, Piscataway, NJ, http://www.gelifesciences.com). BeadChips were scanned with Illumina BeadArray Reader. Data was extracted, normalized, and background subtracted using Illumina BeadStudio software; transcript signals above 0.95 were accepted as presence of expression.

Identification of Differentially Expressed Genes and Clustering Analysis

The dataset extracted by BeadStudio were further analyzed using GeneSpring 7.3.1 (Agilent Technologies, Palo Alto, CA, //www.agilent.com). Genes differentially expressed were identified using Student's t test with a multiple testing correction (Benjamini and Hochberg false-discovery rate <0.05), and fold change ≥2. For example, to extract genes upregulated in CM, the CM data were filtered by the criteria “FDR <0.05, ≥2-fold” compared to ES data, and the resulting dataset was further filtered against EB data. Similarly, these criteria were applied to define cardiac-enriched genes in fetal and adult heart. Hierarchical clustering of differentially expressed genes was conducted. Note that multiple probe sets were present for some of the genes to distinguish gene isoforms. Detailed information can be obtained from the Illumina annotation file (//www.switchtoi.com/annotationprevfiles.ilmn). Extracted outliers were analyzed and interaction networks were generated using Pathway Studio (Ariadne Genomics, Rockville, MD, //www.ariadnegenomics.com).

Immunofluorescence Analysis

EBs or selected cardiomyocytes (spheres from suspension culture) were cryopreserved in 25% sucrose, snap-frozen in OCT medium (Jung), and sectioned to 7 μm using a cryotome. After fixation and permeabilization, sections were incubated with primary antibodies (supporting information Table 2). Primary antibodies specific to ATP1A1, ATP1B1, ATP2A2, NDUFA9, SCN5A, and COX17 were each used in combination with antibody to Nkx2.5 (Rabbit anti-Nkx2.5) for double staining. Primary antibodies specific to RYR2, CACNB1, KCNA5, TRPC3, and HCN4 were each used in combination with antibody to α-MHC (MF20, mouse anti-myosin heavy chain). Secondary antibodies were Alexa-fluor 488-conjuated goat antimouse (1:1,000) and Alexa-fluor 546-conjugated goat anti-rabbit (1:1,000; Molecular Probes Inc., Eugene, OR, //probes.invitrogen.com). Nuclei were stained with Hoechst 33342 (1:10,000; Molecular Probes).

MEA Electrophysiology

CM or nonselected EB were plated on microelectrode array (MEA) chips and extracellular recordings were made with the MEA1060 amplifier and MC-Rack software (Multi Channel Systems, Reutlingen, Germany, //www.multichannelsystems.com). The sampling rate was 5 kHz. A control recording was made before addition of compounds. Stock solutions of isoproterenol and astemizole (Sigma-Aldrich, St. Louis, MO, //www.sigmaaldrich.com) were prepared in water and dimethyl sulfoxide, respectively. Compounds were added to the MEA chamber in increasing concentrations and allowed to equilibrate for at least 5 minutes prior to recording. Interspike intervals and QT intervals were determined from the averaged signal over a 30-second window. The QT interval was measured from the onset of depolarization to the local maximum of the repolarization wave [7]. The corrected QT, QTc, was calculated using the Bazett's formula QTc = QT/(ISI)−1/2 where ISI denotes the interspike interval.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Global Gene Expression Profiling Reveals Similarity of Enriched Cardiomyocytes to Human Heart

We have previously described the efficient cardiomyogenic differentiation of hESCs in serum-free suspension culture followed by an antibiotic-based enrichment of essentially pure cardiomyocytes (>99%) from stable transgenic hESC lines (supporting information Fig. 1) [4]. Here, the Illumina microarray platform was applied to characterize the genomewide gene expression profile of these cardiomyocytes.

First, we profiled RNA samples generated from undifferentiated feeder-free hESC, 21-day old EBs, antibiotic-enriched CM, human FH tissue, and human AH tissue. Sample sources and experimental design are summarized in Figure 1A and supporting information Table 1. Three biological replicates either from independent experiments (ES, EB, CM) or from different commercial tissue sources (FH, AH) were used.

thumbnail image

Figure 1. Microarray analysis to identify cardiac-enriched genes. (A): Experimental design for RNA sample generation and microarray analysis. Human embryonic stem cells (hESC) were differentiated in suspension culture. At day 12 of differentiation, embryoid bodies (EB) were cultured for another 9 days with or without G418 addition. RNA samples were denoted as shown in the figure and described in Results. Three independent biological replicates of each sample were generated (refer to supporting information Table 1 for details). See Methods for details on Illumina 22k array hybridization. (B): Hierarchical clustering analysis. Cardiomyocytes (CMs) are clustered together with human fetal heart (FH) and adult heart (AH) tissue and distinguished from hESCs and EBs. Note the close clustering of independent biological replicates depicted in the dendrogram. (C): Microarray analysis identified 3,112 genes, depicted in the heat map, that exhibited a twofold or greater change in enriched cardiomyocytes compared to undifferentiated hESC and nonselected EB. Among these, 1,311 genes were upregulated and 1,801 were downregulated (refer to supporting information Tables 2 and 3 for full gene list). Hierarchical clustering of these differentially expressed genes was conducted using the GeneSpring software. The relative gene expression is color-coded: red indicates upregulation and blue indicates downregulation relative to hESC/EB. Since multiple probes are present on the microarray chip, some genes are represented several times. (D): The 1,311 cardiac-enriched genes extracted from CM were compared to FH (1,849 cardiac-enriched genes) and AH (2,198 cardiac enriched genes) by means of a Venn diagram. The number of genes showing shared or distinct expression among CM, FH, and AH is indicated. To evaluate the difference and similarity in biological properties of each subgroup in the CM, FH, and AH, Gene Ontology analysis was performed to identify significantly overrepresented biological processes in subgroups. A snapshot of categories comprising the largest number of genes are shown in boxes (for full list of annotations, see supporting information Table 6).

Download figure to PowerPoint

Next, we performed hierarchical clustering analysis by Pearson correlation. As shown in Figure 1B, the close clustering of independent replicates reflected high experimental reproducibility. Furthermore, the dendrogram showed that EB and ES are grouped together while the CM pattern is grouped with FH and AH. This suggests that the gene expression profile of enriched cardiomyocytes is more closely related to that of human heart tissue but clearly distinct from those of parental hESCs and parallel EB controls.

Differential Gene Expression Analysis: Identification of Cardiac Enriched Genes

Microarray analysis identified 3,112 genes that exhibited a twofold or greater change in enriched cardiomyocytes compared to undifferentiated hESC and nonselected EB. Among these, 1,311 genes were upregulated and 1,801 were downregulated (Fig. 1C, supporting information Table 3, supporting information Table 4). Among the downregulated genes, there were known markers associated with hESC pluripotency, such as NANOG, ZFP42 (REX1), POU5F1 (OCT4), TNFRSF8 (CD30), DPPA4, UTF1, DNMT3B, and LIN28; these transcripts were essentially undetectable in CM. There were also known lineage markers downregulated in CM when compared to unselected EB; examples include ectoderm markers TUBB2B (tubulin), NEFH (neurofilament, heavy polypeptide), INA (internexin), OTX2, FBN3 (fibrillin) and endoderm markers SOX17, CLDN6 (claudin 6), SST (somatostatin), and CXCR4. This finding suggests that nonselected EBs consisted of a mixed cell population comprising several differentiated lineages, which is consistent with our prior findings from immunocytology and qualitative reverse-transcription polymerase chain reaction (qRT-PCR) analysis [4, 5]. Apparently, these other lineages were depleted upon cardiomyocyte enrichment.

To explore whether the subset of 1,311 genes that were enriched in selected cardiomyocytes is representative of the molecular phenotype of the heart, we have introduced fetal and adult human heart tissue as a reference. We first extracted differentially expressed genes in FH and AH by also applying a twofold cutoff compared to ES and EB. This analysis resulted in 1,849 genes upregulated in FH and 2,198 genes upregulated in AH. Venn diagram analysis revealed genes that were commonly or specifically expressed among the three cardiac populations (Fig. 1D). Of the 1,311 genes upregulated in CM, 790 were found to be expressed in FH, 748 in AH, and 622 genes overlapped with both FH and AH (Fig. 1D). Among them are well-known markers of cardiomyocytes such as NPPA (ANF), NPPB, TBX5, MYL3 (MLC1v), MYL2 (MLC2v), MB (myoglobin), MYH6 (α-MHC), MYL7 (MLC2a), MYH7 (β-MHC), NKX2.5, ACTN2 (α-actinin), and TNNC1 (supporting information Table 5).

Gene Ontology Analysis: CM-Enriched Genes Belong to Known Categories of Heart Function and Development

To assess the potential function of CM-enriched genes, we grouped them into Gene Ontology categories (GO; //www.geneontology.org) using the Genespring software. By focusing on the 1,311 genes upregulated in CM, 951 were found to have known biological roles. These genes were classified according to GO annotations by the following categories: biological processes, molecular functions, and cellular components.

Biological processes that were identified with high significance (p < .01) are shown in Table 1A. The analysis pointed to processes involved in heart function, regulation, and development such as muscle contraction, circulation, striated muscle development, metabolism, and ion transport. This result strongly supports the expected enrichment of cardiomyocyte-specific transcripts in purified cardiomyocytes. Table 1B presents subcategories (p < .01) of molecular components, which are mainly involved in structural constituent of muscle, ion binding, and transport activity. These include actin-, calmodulin-, and calcium ion-binding.

Table 1. Gene ontology analysis of cardiomyocyte-enriched genes
inline image

Finally, significant (p < .01) subcategories of cellular components are listed in Table 1C. Interestingly, components identified in this category are mainly located in nonnuclear cellular compartments, such as the cytoskeleton, mitochondrion, and extracellular matrix. Other categories are predominantly related to sarcomeric structures such as the actin cytoskeleton, which comprises further subcategories, namely myofibril, sarcomere, myosin, striated muscle filaments, and A band and Z discs.

Besides highlighting processes and components specific to cardiomyocyte structure and function, the GO-based annotation collectively revealed that upregulated genes are predominantly involved in energy production and primary metabolism. This is in line with the known high capacity of ATP generation in cardiomyocytes.

To compare the CM, FH, and AH samples in the context of biological process, we further analyzed the subgroups in the Venn diagram (Fig. 1D) by GO annotation. The results showed that genes commonly upregulated in AH and FH but not in CM are involved in cell communication, signal transduction, and defense response (supporting information Table 6). These biological processes are generally overrepresented in RNA samples from many human tissues [8] and might thus not be specific to cardiac cell properties and heart morphogenesis. The bulk of genes upregulated in CM but not in AH or FH (395; Fig. 1D) were assigned to the process of development (supporting information Table 6B), including known players in heart formation such as BMP2, TGFB2, GATA5, and others (Fig. 3C). Notably, the 622 overlapping genes among the CM, FH, and AH samples overrepresent key biological processes known to be required for cardiomyocyte function, such as muscle contraction, metabolism, and ion transport (supporting information Table 6A).

Pathway Analysis Identified a Cardiac Transcription Factor Network and Peroxisome Proliferator- Activated Receptor Signaling Pathway in Cardiomyocytes

To identify signaling pathways involved in cardiac differentiation, we further explored the subset of 1,311 enriched genes using the software Pathway Studio (PS). PS converts datasets into networks displaying direct and indirect interactions of gene products based on published knowledge (Resnet Mammalian Database 5.0). By applying promoter binding relation parameters, an extensive cardiac transcription factor network was identified. As shown in Figure 2A, the DNA binding regulatory proteins, myocyte-specific enhancer factor-2 family (MEF2A, 2C, 2D) and members of the GATA family of zinc-finger transcription factors (GATA4, 5, 6) are present in this network. The cardiac specific transcription factor NKX2.5 interacts with MEF2C and MYCD (myocardin) in concert with TBX5 and GATA4 to synergistically activate expression of the atrial naturetic peptide gene (ANP), suggesting that this network is essential for cardiomyocyte differentiation and heart development.

thumbnail image

Figure 2. Transcription factor network and peroxisome proliferator-activated receptor (PPAR) signaling in cardiomyocytes. Genes coding for components of a cardiac transcription factor network (A) and PPAR pathways (depicted in their subcellular location, B) were found to be expressed in enriched cardiomyocytes. The interaction network and graphical view were generated by Pathway Studio (Ariadne Genomics) using the Ariadne Resnet Mammalian Database 5.0 and Kyoto Encyclopedia of Genes and Genomes database. A known functional interlink between PPAR signaling and the cardiac transcription factor network via PPARGC1A is highlighted in green. Graph legends were adapted from Pathway Studio (Ariadne Genomics). See supporting information Table 8 for a full description of genes.

Download figure to PowerPoint

We also searched the Kyoto Encyclopedia of Genes and Genomes database (http://www.kegg.com). Twenty-four pathways were identified when applying software-defined significance criteria (p < .05) including fatty acid metabolism, reductive carboxylate cycle, and signaling pathways involving peroxisome proliferator-activated receptor (PPAR), insulin, and MAPK signaling pathways (supporting information Table 7).

Notably, among the PPAR signaling-related genes, all the nuclear receptors (PPARA, PPARD, PPARG, PPARGC1A, RXRA, and RXRG) were expressed and many of their downstream targets were highly upregulated in CM (supporting information Table 8B). A Pathway Studio-generated interaction network of these components and their cellular locations are presented in Figure 2B. PPARs are a family of ligand-activated transcription factors. For example, PPARD is activated by the ligand prostacyclin I2 (PGI2) to modulate specific cellular functions such as embryo implantation, vascular endothelial barrier enhancement [9], and apoptosis [10]. One known functional interlink between PPAR signaling and the cardiac transcription factor network via PPARGC1A is highlighted in Figures 2A and B. Interestingly, we have recently demonstrated that prostaglandin I2 is a factor that can enhance hESC cardiomyogenic differentiation [11].

Enriched Cardiomyocytes Have a Specific Gene Expression Signature Related to but Distinct From Fetal and Adult Human Heart

The suitability of hESC cardiomyocytes for regenerative medicine and in vitro pharmacology depends on proper expression of genes involved in heart development and maturation, formation of sarcomeric filaments, ion transport, excitation-contraction coupling, and other essential properties. Aiming at an unbiased lineup, we have performed GO analysis on a pool of 3,082 “cardiac” genes, which are enriched in any one of the populations AH, FH, or CM, as indicated by the Venn diagram in Figure 1D.

We first focused on 63 genes classified in the GO cellular components categories related to sarcomeric structures (including subcategories of actin cytoskeleton, myosin, and sarcomere). As shown in Figure 3A (full list in supporting information Table 8C). Only a small group of genes showed relatively high expression levels in CM, including specific myosins such as MYH6 and MYL7, which we have previously confirmed by qRT-PCR [4]. Other genes identified in this category include MTSS1, DYNLT1, and KLHL2; none of these are yet known to play a role in cardiomyogenesis.

Some genes showed strikingly high expression levels, particularly in FH, such as SORBS2, SPTBN5, SPTA1, TLN2, and MYO1G. These genes are most likely to be transiently upregulated in fetal heart or have a specific role in heart development in vivo which might, in some aspects, not be emulated by hESC cardiomyogenic differentiation in vitro.

A large number of genes displayed the following pattern: high expression in AH and high-to-intermediate levels in FH, but relatively low level in CM. These included myosin genes MYL2, MYH7, MYL3, and MYH11, suggesting that they are upregulated upon cardiomyocyte maturation, which tallies with our prior qRT-PCR results [4]. Known cardiac troponin genes such as TNNI3, TNNC1, and TNNT2 also belong to this group. An equivalent expression pattern was shared by other genes (Fig. 3A), providing a large number of additional cardiac maturation markers.

thumbnail image

Figure 3. Hierarchical clustering of cardiac-enriched genes involved in functional categories of sarcomere (A), ion transport (B), and heart development (C) Genes that showed at least twofold overexpression in cardiomyocytes (CM), fetal heart (FH), or adult heart (AH) tissue in comparison to human embryonic stem cells (hESCs) or embryoid bodies (EBs) were extracted and grouped based on Gene Ontology annotation using GeneSpring software. Colors indicate the relative expression levels of each gene, with red indicating upregulation and blue indicating downregulation relative to hESC/EB. See supporting information Table 8 for a full list of genes. Boxes and genes highlighted in bold color are described in detail in Results and Discussion. (Additional genes highlighted in black are candidates being validated by immunochemical staining as shown in Figure 4.)

Download figure to PowerPoint

Similarly, of the 85 genes allocated in the GO category ion transport, the heat map in Figure 3B (full list in supporting information Table 8D) indicates that only a few of these genes showed comparable expression among all three samples. Notably, this group includes the voltage-gated potassium channel KCNH2 (HERG), known to play a role in the long-QT syndrome and drug-induced ventricular tachycardia, thus cross-validating the qRT-PCR analysis from our previous study [4].

We have identified relative high expression levels of some ion channels/subunits in hESC cardiomyocytes (Fig. 3B) that have not been emphasized in prior reports [10, 11]. One of these genes is CACNA1G coding for the T-type voltage gated calcium channel subunit 1G. While L-type channels are important in sustaining action potential, T-type channels are important in initiating them. Expression of the voltage-gated potassium channel KCNQ1 suggests the presence of the current IKs in CM, which is involved in shaping the diastolic phase of the action potential. High expression levels of ATP1A1/ATP1B4 (Na+/K+-ATPases) and ATP2B1 (Ca++ transporting ATPase) in CM further suggests that ion homeostasis in these cells is regulated by INaK and IpCa currents.

Notably, our analysis also identified a subset of genes that have previously been reported in hESC cardiomyocytes by expression analysis and / or electrophysiological assessment. These include CACNA1C, KCNA4 [12, 13], KCNH2 [4, 13], HCN1, HCN4 [12, 13], RYR2 [14], SLC8A3, SLC8A1, and ATP2A2 [15] PLN [2, 3] (Fig. 3B), thereby providing cross-validation of our findings. Taken together, the heat-map data in Figure 3B indicates similarities as well as key differences in ion transport in CM versus human heart tissue, many of which are novel to the field.

Next, we took a closer look at 25 genes annotated in the GO category heart development. Figure 3C illustrates a cluster of genes that showed highest expression in CM and FH, but much lower levels in AH. This includes NKX2.5, IRX4, MEF2A, TBX5, GATA4, and MEF2C. Apparently, these factors are not only involved in early stages of cardiomyocyte specification, but also play a role in maintaining the cardiac phenotype.

A differential pattern showing up-regulation at the EB stage, with highest expression in CM but substantially lower levels in FH and further reduction in AH, was observed for GATA5 TGF-β2 and BMP2, suggesting their transient involvement in early cardiac lineage determination. Hence, CM probably represents a less differentiated, earlier cardiomyocyte phenotype than FH. Conversely, a cluster of genes (Fig. 3C) showed relative high expression in AH and FH but markedly lower level in CM, such as CENTA2, NR4A1, FGF12, KLF2, and FBN1. These genes might be associated with a more mature cardiac phenotype.

Validation of Microarray Data by qRT-PCR and Immunocytochemistry

In order to verify the microarray data, qRT-PCR analysis was performed on a group of genes presented in supporting information Figure 2. The microarray expression pattern for representative PPAR pathway members such as PPARGC1A, RXRG, ACOX2, ACADL, and ACASB was confirmed by the qRT-PCR assay.

Immunocytochemistry was further performed to confirm the expressions of numerous ion channels identified by GO analysis in this study. Double staining specific to respective proteins was combined with cardiomyocyte markers α-MHC or NKX2.5 on sections of day 21 EB; noncardiomyocytes on the same section unequivocally prove staining specificity. As shown in Figure 4, ion channel-specific staining for ATP1A1, ATP1B1, ATP2A2, COX17, RYR2, and CACNB1 largely overlapped with those for α-MHC or Nkx2.5, thereby demarcating the cardiomyocyte clusters. NDUFA9, SCN5A, KCNA5, TRPC3, and HCN4 showed a more distributed expression, but signal intensity was much more prominent in cardiomyocyte clusters. In summary, both qRT-PCR and immunohistochemistry confirmed the microarray data and support our conclusion that these genes were upregulated in the selected cardiomyocytes.

thumbnail image

Figure 4. Ion channel expressions in cardiomyocytes derived from human embryonic stem cells (hESCs). Immunohistochemical staining of cryosections of hESC-derived beating embryoid bodies (EBs) was performed as described in Materials and Methods. Sections from at least three different cardiomyocyte clusters were analyzed for each antibody; results were consistent across the samples. Day 21 EBs showed overlapping staining of cardiomyocyte markers Nkx2.5 (green) or α-MHC (red). Staining specific to ion channel proteins ATP1A1, ATP1B1, ATP2A2, NDUFA9, SCN5A, and COX17 overlapped with that for NKX2.5 (right); RYR2, CACNB1, KCNA5, TRPC3, and HCN4 expression overlapped with that of α-MHC (left). Nuclei were stained with Hoechst 33342. Scale bars = 50 μm.

Download figure to PowerPoint

Functional Properties of Selected Cardiomyocytes

Enriched cardiomyocytes and cardiomyocyte clusters in nonselected EB controls were further characterized by assessing their electrophysiological properties using microelectrode arrays [7]. Contracting EB exhibited spontaneous action potentials as indicated by extracellular recordings of their field potentials (Fig. 5A). Addition of the β-adrenergic agonist isoproterenol, a chronotropic agent, increased the beating frequency in a dose-dependent manner as previously reported [16, 17]. Figure 5A shows the decrease in interspike interval, which reached around 60% of the baseline at 1 μM isoproterenol for both selected CM and EB controls. We also tested the effect of the antihistamine astemizole on the QT interval. This compound is a potent blocker of the HERG potassium channel and was withdrawn from the market due to its cardiotoxic profile [18, 19]. Figure 5B shows that increasing concentrations of astemizole prolonged the QT interval in an equivalent manner for both EB controls and CM. Taken together, these results show that the hESC-derived cardiomyocytes exhibited the physiological properties of primary cardiomyocytes, and these properties were not significantly altered by G418 selection.

thumbnail image

Figure 5. Electrophysiological characterization of human embryonic stem cell-derived cardiomyocytes (CMs) using microelectrode arrays (MEA). (A): Isoproterenol decreased the interspike interval for both selected CM and control embryoid bodies (EBs). Top two panels show representative traces from MEA recordings of selected and nonselected cells. Bottom graph illustrates the change in interspike interval with increasing concentration of isoproterenol (mean ± SEM; n = 3). Two-way analysis of variance (ANOVA) indicated no significant difference between the selected and nonselected populations but a significant effect of isoproterenol concentration. (B): Prolongation of the QTc interval with addition of the antihistamine astemizole. Top panels show representative traces for selected and nonselected EBs. Bottom graph plots the percent increase over baseline in the QTc interval with increasing concentrations of astemizole (mean ± SEM; n = 3). Two-way ANOVA indicated no significant difference between the selected and nonselected populations.

Download figure to PowerPoint

Discovery of Novel Genes in Heart Development

Aiming at discovering novel players orchestrating heart development, we have further applied qRT-PCR to validate the expression of some “candidate” genes. These are the genes found to be upregulated in cardiomyocytes in the microarray screen (supporting information Table 3), but for which little or no information is available regarding their function during heart development. Here, we have further analyzed four representative genes: RNA binding motif protein 24 (RBM24), transcription elongation factor A3 (TCEA3) [20], formin homology two domain containing 3 (FHOD3) [21]), and chromosome 15 open reading frame 52 (C15ORF52). qRT-PCR data on these four genes showed that their expression pattern in EB, CM, and FH were similar to that of well-characterized genes known to be associated with heart development, such as BVES and CXCL12 (supporting information Figure 2).

Subsequently, mouse orthologues were identified and whole-mount in situ hybridization on mouse embryos was performed. Expression in the developing heart was observed for all genes analyzed (Fig. 6). RBM24-specific staining was found as early as embryonic day 7 (E7); the pattern suggests expression in the region of early cardiac mesoderm. Strong, heart-specific expression proceeded on E8 in the cardiac crescent and was further detected in the whole heart until E10.5, the latest stage analyzed. In addition, expression in the myotome in somites was observed, which is in line with the expression in fetal skeletal muscle in human tissue (Data not shown). TCEA3 and FHOD3 were detected in the cardiac crescent and the heart anlagen from E8 onwards, whereas heart-specific expression of C15orf52 was not detected before E9.5, suggesting that the gene is a late marker, which tallies with the strong upregulation in adult human heart.

thumbnail image

Figure 6. In situ hybridization analysis of novel cardiac specific genes in mouse embryos. Mouse orthologues of the human genes Rbm24, Tcea3, Fhod3, and C15orf52 were identified and their expression at different stages of mouse embryonic development was detected. Expression in the heart was observed for all genes analyzed. Rbm24 specific staining was found as early as embryonic day 7 (E7); the pattern suggests expression in the region of early cardiac mesoderm development. Intense, heart specific staining for this gene proceeded on E8 in the cardiac crescent and was further detected in the whole heart until E10.5, the latest stage analyzed. In addition, expression was observed in the myotome in somites as well as in the developing brain. Tcea3 and Fhod3 expression was detected in the cardiac crescent from E8 onwards, whereas heart-specific expression of C15orf52 was not detected before E9.5. Additional expression domains were noted as follows: Tcea3, in the tail area from E8 onwards; the dorsal part of the most cranial (∼7 oldest) somites from E9.5 onwards, expanding to more caudal somites at E10.5; in the apical ectodermal ridge in the limbs of E10.5 embryos; Fhod3, intense stain along the midline of axial structures from E8.5 onwards; C15orf52, tail area.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Transcriptome analysis is a powerful tool to evaluate the molecular phenotype and developmental status of hESC-derived cell types. Comparing the gene expression pattern of somatic cells generated in tissue culture with their counterparts in developing and adult organs also interlinks in vitro and in vivo differentiation. Here, we have focused on cardiomyogenesis.

Compared to previous studies in the field, our approach overcomes the reported low and poorly defined enrichment of the analyzed cardiomyocyte populations achieved by manual dissection of beating areas [1, 2]. Instead, we have compared essentially pure cardiomyocytes to undifferentiated hESC and late-stage EB as well as fetal and adult human heart tissue, which might be defined as a gold standard for cardiomyocyte characterization. By comparing our list of 1,311 genes upregulated in CM (supporting information Table 3) to respective lists compiled by Beqqali et al. [1] and Synnergren et al. [2], we found 33 upregulated genes common among all three studies (supporting information Table 9). Of note, these 33 “overlapping” genes represent cardiac structural constituents, which are abundantly expressed in cardiomyocytes, such as myosin, actinin, and troponin components.

More recently, Cao et al. analyzed the transcriptome of a population consisting of 40%–45% cardiomyocytes (Percoll gradient enriched [3]). In contrast to our findings, no upregulation of known cardiac markers such as ACTC (α-actin) MYH6 (α-MHC), MYH7 (β-MHC), MYL2 (MLC-2v), MYL7 (MLC-2a), NPPA (ANF), and TNNT2 (cTnC) was observed in their Percoll purified population (CM) versus nonpurified control (EB; supporting information Table 9B). Hierarchical clustering showed that the enriched hESC cardiomyocyte population was more related to its origin, EB, than to samples isolated from fetal ventricular cells [3]. This perhaps reflects the limited enrichment of hESC cardiomyocytes (40%–45% purity) by means of a Percoll gradient. In addition, Cao et al. isolated RNA specifically from human ventricular cardiomyocytes for transcriptional comparison, in an attempt to prevent noncardiac cell types in the heart from contaminating gene expression data [3]. Indeed, in the adult mammalian heart, only about 30%–35% of the cell nuclei were found within cardiomyocytes [22]. However, our results showed that the gene expression pattern in pure hESC cardiomyocytes closely clustered with that from total fetal and adult human heart. Using RNA from whole heart including atrial and pacemaker tissue as opposed to ventricular cardiomyocytes only might better reflect the mixed cardiomyocyte phenotypes generated from hESC differentiation protocols [4, 5, 12, 13].

To date, only a few publications have combined gene expression analysis focused on about 10 ion channel-encoding genes with electrophysiological and pharmacological studies to understand the action potential pattern and underlying ion currents in hESC-CM. These studies identified expression of genes including the sodium channel SCN5A (current INa) and L-type calcium channel CACNA1C (ICa(L)) [12], voltage gated potassium channels KCNA4 and KCND3 (Ito1) [13], voltage-gated potassium channel KCNH2 (IKr) [4, 13], and hyperpolarization activated potassium channels HCN1 and HCN4 (If) [12, 13], as well as sarcoplasmic reticulum (SR) calcium release channels RYR2 and ITPR1 [14]. As outlined in the Results, we have confirmed these genes expression in the hESC-derived cardiomyocytes. Importantly, we have identified the expression of a number of ion channels in CM that have not been noted before. One of them, CACNA1G, codes for the T-type voltage gated calcium channel subunit 1G. In the adult heart expression of T-type channels is restricted to pacemaker cells (i.e., sinoatrial and atrioventricular nodes). However, the high expression level of T-type channels in CM suggests their role in initiating automatic depolarization activity, underplaying the well-documented spontaneous contraction of hESC cardiomyocytes. The relatively high expression level of CACNA1G, which we also observed in FH, agrees with prior reports on ventricular cardiomyocytes of the developing mouse heart [23] where T-type channels may contribute to the contraction automaticity of ventricular cells at that stage. An unexpected low expression of the SCN5A gene, a sodium channel that is abundant in heart muscle, was found in all samples tested. This is likely due to technical issues such as a suboptimal probe for this gene on the chip. We have validated the expression of this gene in differentiated EB by immunostaining (Fig. 4) .

One potential confounding factor of our approach is that the antibiotic-based selection process itself might somehow alter the gene expression pattern of the cardiomyocytes in culture. Although we can not fully rule out this possibility at present, our electrophysiological characterization of enriched versus untreated cardiomyocytes did not support this hypothesis. Addition of the chronotropic agent isoproterenol expectedly increased the beating frequencies of both cell samples. Similarly, addition of the hERG channel blocker astemizole prolonged the QT interval for both selected and nonselected populations. These results combined with our immunostaining data indicate that both selected and nonselected populations share the same cardiac markers and pharmacological properties.

We have discovered an extensive transcription factor network in cardiomyocytes and its linkage to the PPAR signaling pathway, which has not been extensively studied in cardiogenesis. PPARs belong to a family of ligand-activated transcription factors. Heterodimerization of PPARs (PPARA, PPARG, and PPARD) with retinoid X receptors (RXRs) is a prerequisite for DNA binding activity; recruitment of transcriptional cofactors (PPAR-binding proteins) is required to activate target genes. A wide spectrum of ligands for PPARs exists, including free fatty acids and their derivatives produced by the cyclooxygenase (COX) or lipoxygenase pathways. In vivo, PPARD is activated by the endogenous ligand prostaglandin I2 (PGI2), one of the major prostaglandins, to modulate specific cellular functions [24]. PGI2 has well-known vasodilating and antithrombogenic properties that confer cardioprotective effects. By functional analysis of a conditioned medium derived from the cardiomyocyte-inducing cell line END2 [25], we have recently identified PGI2 as a factor that enhanced hESC cardiomyogenic differentiation [11]. Recent studies suggest that RXR signaling regulates the differentiation of cardiomyocytes derived from embryonic stem cells in serum-free conditions [26], and stimulation of PPARA enhances cardiomyogenesis in ES cells using a pathway that involves ROS and NADPH oxidase activity [27]. Combined with our findings presented herein, these data point to the involvement of PPAR signaling in hESC cardiomyogenesis and that PGI2 possibly mediates its inductive effect through the activation of the PPAR pathway. Importantly, PPAR signaling might also play a role in human heart development and homeostasis in vivo.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The focused comparison of CM versus FH and AH in the GO categories myosin, ion transport, and heart development indicated that hESC-derived cardiomyocytes, at the transcriptional level, have some similarity to FH and AH, but in some aspects present an individual, specific transcriptome pattern. Based on this analysis, we have identified numerous genes whose relative expression level can serve to follow CM maturation in vitro. They include potentially valuable new markers to investigate the impact of novel culture platforms on the cardiomyocyte phenotype, such as culturing cells in engineered three-dimensional tissue followed by mechanical and electrical stimulation [28, 29]. Genes coding for ion channels that showed highest expression level in CM suggest the presence of still not described currents underlying the contraction automaticity in hESC cardiomyocytes and the specific action potential pattern observed in these cells. The extensive marker sets provided in this study, combined with recent sophisticated technologies for quantitative expression profiling of a large subset of marker genes in single cells [30], will pave the way to directly relate physiological characteristics and the gene expression pattern in single cardiomyocytes.

Finally, we have discovered numerous genes not known to be expressed during cardiomyocyte differentiation. To illustrate the relevance of our screen, four of these candidates were analyzed in more detail. Their predominant expression in the developing mouse heart clearly underscores the efficiency of our microarray approach and successful subsequent data mining. The functional analysis of these and other genes in relevant developmental models and in hESC cardiomyogenic differentiation is in progress and will shed new light on molecular processes leading to the formation of the heart.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

This study was supported by A-Star (Singapore). We thank Birgit Andree, Siew Tein Wang, Norris Ray Dunn, and Muhammed Yusuf Ali for their technical help with in situ experiments, and Tan Kar Tong for his contribution to generating immunohistochemistry and Western blot data. We thank Birgit Andree and Jeremy Crook for their critical reading of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information
  • 1
    Beqqali A, Kloots J, Ward-van Oostwaard D et al. Genome-wide transcriptional profiling of human embryonic stem cells differentiating to cardiomyocytes. Stem Cells 2006; 24: 19561967.
  • 2
    Synnergren J, Akesson K, Dahlenborg K et al. Molecular signature of cardiomyocyte clusters derived from human embryonic stem cells. Stem Cells 2008; 26: 18311840.
  • 3
    Cao F, Wagner RA, Wilson KD et al. Transcriptional and functional profiling of human embryonic stem cell-derived cardiomyocytes. Plos One 2008; 3: e3474.
  • 4
    Xu XQ, Zweigerdt R, Soo SY et al. Highly enriched cardiomyocytes from human embryonic stem cells. Cytotherapy 2008; 10: 376389.
  • 5
    Graichen R, Xu X, Braam SR et al. Enhanced cardiomyogenesis of human embryonic stem cells by a small molecular inhibitor of p38 MAPK. Differentiation 2008; 76: 357370.
  • 6
    Zweigerdt R, Burg M, Willbold E et al. Generation of confluent cardiomyocyte monolayers derived from embryonic stem cells in suspension: a cell source for new therapies and screening strategies. Cytotherapy 2003; 5: 399413.
  • 7
    Meyer T, Boven KH, Gunther E et al. Micro-electrode arrays in cardiac safety pharmacology: a novel tool to study QT interval prolongation. Drug Saf 2004; 27: 763772.
  • 8
    Shyamsundar R, Kim YH, Higgins JP et al. A DNA microarray survey of gene expression in normal human tissues. Genome Biol 2005; 6: R22.
  • 9
    Birukova AA, Zagranichnaya T, Fu P et al. Prostaglandins PGE(2) and PGI(2) promote endothelial barrier enhancement via PKA- and Epac1/Rap1-dependent Rac activation. Exp Cell Res 2007; 313: 25042520.
  • 10
    Cha YI, Solnica-Krezel L, DuBois RN. Fishing for prostanoids: deciphering the developmental functions of cyclooxygenase-derived prostaglandins. Dev Biol 2006; 289: 263272.
  • 11
    Xu XQ, Graichen R, Soo SY et al. Chemically defined medium supporting cardiomyocyte differentiation of human embryonic stem cells. Differentiation 2008; 76: 958970.
  • 12
    Satin J, Kehat I, Caspi O et al. Mechanism of spontaneous excitability in human embryonic stem cell derived cardiomyocytes. J Physiol 2004; 559: 479496.
  • 13
    Sartiani L, Bettiol E, Stillitano F et al. Developmental changes in cardiomyocytes differentiated from human embryonic stem cells: a molecular and electrophysiological approach. Stem Cells 2007; 25: 11361144.
  • 14
    Satin J, Itzhaki I, Rapoport S et al. Calcium handling in human embryonic stem cell-derived cardiomyocytes. Stem Cells 2008; 26: 19611972.
  • 15
    Binah O, Dolnikov K, Sadan O et al. Functional and developmental properties of human embryonic stem cells-derived cardiomyocytes. J Electrocardiol 2007; 40: S192S196.
  • 16
    Reppel M, Boettinger C, Hescheler J. Beta-adrenergic and muscarinic modulation of human embryonic stem cell-derived cardiomyocytes. Cell Physiol Biochem 2004; 14: 187196.
  • 17
    Xue T, Cho HC, Akar FG et al. Functional integration of electrically active cardiac derivatives from genetically engineered human embryonic stem cells with quiescent recipient ventricular cardiomyocytes: insights into the development of cell-based pacemakers. Circulation 2005; 111: 1120.
  • 18
    Paakkari I. Cardiotoxicity of new antihistamines and cisapride. Toxicol Lett 2002; 127: 279284.
  • 19
    Zhou Z, Vorperian VR, Gong Q et al. Block of HERG potassium channels by the antihistamine astemizole and its metabolites desmethylastemizole and norastemizole. J Cardiovasc Electrophysiol 1999; 10: 836843.
  • 20
    Labhart P, Morgan GT. Identification of novel genes encoding transcription elongation factor TFIIS (TCEA) in vertebrates: conservation of three distinct TFIIS isoforms in frog, mouse, and human. Genomics 1998; 52: 278288.
  • 21
    Kanaya H, Takeya R, Takeuchi K et al. Fhos2, a novel formin-related actin-organizing protein, probably associates with the nestin intermediate filament. Genes Cells 2005; 10: 665678.
  • 22
    Nag AC. Study of non-muscle cells of the adult mammalian heart: a fine structural analysis and distribution. Cytobios 1980; 28: 4161.
  • 23
    Cribbs LL, Martin BL, Schroder EA et al. Identification of the t-type calcium channel (Ca 23(v) 3.1d) in developing mouse heart. Circ Res 2001; 88: 403407.
  • 24
    Lim H, Dey SK. A novel pathway of prostacyclin signaling–hanging out with nuclear receptors. Endocrinology 2002; 143: 32073210.
  • 25
    Mummery CL, van Achterberg TA, van den Eijnden-van Raaij AJ et al. Visceral-endoderm-like cell lines induce differentiation of murine P19 embryonal carcinoma cells. Differentiation 1991; 46: 5160.
  • 26
    Honda M, Hamazaki TS, Komazaki S et al. RXR agonist enhances the differentiation of cardiomyocytes derived from embryonic stem cells in serum-free conditions. Biochem Biophys Res Commun 2005; 333: 13341340.
  • 27
    Sharifpanah F, Wartenberg M, Hannig M et al. Peroxisome proliferator-activated receptor alpha agonists enhance cardiomyogenesis of mouse ES cells by utilization of a reactive oxygen species-dependent mechanism. Stem Cells 2008; 26: 6471.
  • 28
    Naito H, Melnychenko I, Didie M et al. Optimizing engineered heart tissue for therapeutic applications as surrogate heart muscle. Circulation 2006; 114: I7278.
  • 29
    Caspi O, Lesman A, Basevitch Y et al. Tissue engineering of vascularized cardiac muscle from human embryonic stem cells. Circ Res 2007; 100: 263272.
  • 30
    Yan Chen JFZ. Microfluidic devices for high-throughput gene expression profiling of single hesc-derived neural stem cells. Totowa, NJ: Humana Press; 2008.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional supporting information available online.

FilenameFormatSizeDescription
STEM_166_sm_suppinfoandfigs.pdf1154KSupporting Information and Figures
STEM_166_sm_suppinfotable1.pdf19KSupporting Information Table 1
STEM_166_sm_suppinfotable2.pdf25KSupporting Information Table 2
STEM_166_sm_suppinfotable3.pdf364KSupporting Information Table 3
STEM_166_sm_suppinfotable4.pdf488KSupporting Information Table 4
STEM_166_sm_suppinfotable5.pdf70KSupporting Information Table 5
STEM_166_sm_suppinfotable6.pdf23KSupporting Information Table 6
STEM_166_sm_suppinfotable7.pdf15KSupporting Information Table 7
STEM_166_sm_suppinfotable8.pdf29KSupporting Information Table 8
STEM_166_sm_suppinfotable9.pdf25KSupporting Information Table 9

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.