SEARCH

SEARCH BY CITATION

Keywords:

  • Human embryonic stem cells;
  • Differentiation;
  • Cardiomyocytes;
  • Gene expression

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Human embryonic stem cells (hESCs) can differentiate in vitro into spontaneously contracting cardiomyocytes (CMs). These cells may prove extremely useful for various applications in basic research, drug discovery, and regenerative medicine. To fully use the potential of the cells, they need to be extensively characterized, and the regulatory mechanisms that control hESC differentiation toward the cardiac lineage need to be better defined. In this study, we used microarrays to analyze, for the first time, the global gene expression profile of isolated hESC-derived CM clusters. By comparing the clusters with undifferentiated hESCs and using stringent selection criteria, we identified 530 upregulated and 40 downregulated genes in the contracting clusters. To further characterize the family of upregulated genes in the hESC-derived CM clusters, the genes were classified according to their Gene Ontology annotation. The results indicate that the hESC-derived CM clusters display high similarities, on a molecular level, to human heart tissue. Moreover, using the family of upregulated genes, we created protein interaction maps that revealed topological characteristics. We also searched for cellular pathways among the upregulated genes in the hESC-derived CM clusters and identified eight significantly upregulated pathways. Real-time quantitative polymerase chain reaction and immunohistochemical analysis confirmed the expression of a subset of the genes identified by the microarrays. Taken together, the results presented here provide a molecular signature of hESC-derived CM clusters and further our understanding of the biological processes that are active in these cells.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Authors contributions: J.S.: collection and assembly of data, data analysis and interpretation, manuscript writing; K.Å., K.D., H.V., C.A., D.S.: collection and assembly of data, final approval of manuscript; A.L.: data analysis and interpretation, final approval of manuscript; B.O.: data analysis and interpretation, manuscript writing; P.S.: conception and design, data analysis and interpretation, manuscript writing.

Human embryonic stem cells (hESCs) represent populations of undifferentiated cells with a seemingly unlimited ability to proliferate [1]. These cells can be maintained in vitro in their pluripotent state or they can be coaxed to differentiate along specific pathways to form a variety of specialized cell types, including cardiomyocytes (CMs) [2, [3], [4], [5]6]. Based on their fundamental properties, hESCs have the potential to be extremely useful for in vitro studies in basic research, drug discovery, and toxicological testing [7]. In addition, there are considerable expectations on the possible future usage of hESC derivatives for cell replacement therapies [8].

One of the major bottlenecks that researchers are presently struggling with is the insufficient yield and purity of the final cell preparations resulting from the various differentiation regimes. Importantly, in most cases, the industrial implementation of hESC-based technologies requires the generation of large quantities of specialized cells and thus improved protocols for hESC differentiation are urgently needed. One of the reasons for the relative scarcity of efficient differentiation protocols is the substantial gaps in our understanding of the molecular programs that govern early cellular differentiation. Regarding cardiogenesis, lessons from the developing embryo and in vitro experimentation using mouse and human ESCs have certainly been instrumental for providing important pieces of the puzzle of how to transform undifferentiated hESCs into CMs [7]. However, additional research is needed to further our understanding of these complex systems that subsequently can lead to more effective strategies for hESC manipulation and differentiation.

Besides improving the yield and purity of hESC-derived CMs, it is equally important to thoroughly characterize the phenotype of the differentiated cells. During recent years, several independent investigators have characterized hESC-derived CMs based on their ultrastructural, pharmacological, and electrophysiological properties [6, 9, [10], [11], [12]13]. In addition, gene and protein expression has been analyzed in hESC-derived CMs with specific focus on cardiac marker genes [9, 11, 14]. We and others have used microarray technology to study the gene expression profiles of hESCs that undergo cardiogenic induction [15, 16]. However, no global gene expression analysis has previously been reported using isolated spontaneously contracting hESC-derived CM clusters. Hence, in the present study, we generated beating CM clusters from hESCs and compared the gene expression profile of these clusters with the gene expression signature of undifferentiated hESCs. Using the significance analysis of microarrays (SAM) algorithm, we identified 530 genes that were specifically upregulated in the CM clusters and 40 genes that were downregulated. In addition, we grouped the family of upregulated genes based on their functional properties and identified possible protein interactions and cellular pathways that may be important for cardiogenic induction of hESCs but also for sustaining the CM phenotype. Taken together, these results provide valuable information about the molecular programs that are active in hESC-derived CM clusters.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Culture and Differentiation of hESCs

The hESC line SA002 (47, XX +13) (Cellartis AB, Göteborg, Sweden, http://www.cellartis.com/) was propagated on mitomycin-C-inactivated mouse embryonic fibroblast feeder layers using VitroHES medium (Vitrolife AB, Kungsbacka, Sweden, http://www.vitrolife.com/) supplemented with 4 ng/ml basic fibroblast growth factor as previously described [17]. The hESC line SA002 efficiently forms cells of the cardiac lineage, and the extra chromosome 13 does not appear to influence this process or the functionality of the CMs formed in an abnormal way [6, 16]. Undifferentiated hESCs were harvested at days 4–5 after passage for RNA extraction or differentiated to obtain hESC-derived CM clusters using cells in passage 23–41. To initiate differentiation of the hESCs, embryoid bodies (EBs) were formed, either in suspension cultures [6] or through forced aggregation [18]. After 6–7 days, the EBs were subsequently plated onto gelatine-coated culture dishes in knockout Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum, 1 mM GlutaMAX, 0.1 mM β-mercaptoethanol, 1% nonessential amino acids, and 1% penicillin-streptomycin solution (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com). The medium was changed every 2–3 days. Normally, within the first 4 days after plating, contracting areas could be observed in the outgrowth of the EBs. Spontaneously contracting clusters were identified by visual inspection using light microscopy and harvested by mechanical dissection. Specific care was taken only to harvest the beating areas with a minimum of surrounding noncontracting cells. For RNA extractions, CM clusters were pooled at various time points (< 22 days) after onset of contraction.

RNA Extraction and Microarray Experiments

Total RNA was extracted from undifferentiated hESCs and hESC-derived CM clusters using Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) according to the manufacturer's instructions. DNase treatment was performed on-column using Qiagen RNase-free DNase Kit (Qiagen).

Two separate sets of microarray experiments were conducted. In the first, one-cycle amplified RNA was used, whereas two-cycle amplified RNA was used in the second set of experiments due to the limited amount of available RNA. All subsequent calculations were conducted between samples within each experiment separately. For each experiment, the material consisted of one pooled sample of undifferentiated hESCs and two different biological replicates of pooled hESC-derived CM clusters.

The quality of the RNA and cRNA, labeled by in vitro transcription, was tested using an Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA, http://www.agilent.com). Fragmented cRNA was hybridized at 45°C for 16 hours to GeneChip Human, HGU 133 Plus 2.0 (Affymetrix, Santa Clara, CA, http://www.affymetrix.com). Each sample was hybridized to duplicate arrays.

Data Analysis

Identification of Differentially Expressed Genes.

Genes that were significantly upregulated or downregulated in the hESC-derived CM clusters compared with undifferentiated hESCs were identified using the SAM algorithm [19]. Briefly, the algorithm assigns a score to each gene based on differences in expression between conditions relative to the standard deviation of repeated measurements. The false discovery rate (FDR) is determined using permutations of the repeated measurements to estimate the percentage of genes identified by chance. The algorithm was applied to each of the data sets from the two experiments separately using FDR < 0.04. Subsequently, only the genes marked as significantly upregulated or downregulated in both data sets were considered as differentially expressed in hESC-derived CM clusters compared with undifferentiated hESCs.

Gene Ontology Analysis.

To further explore the biology of the significantly upregulated genes in hESC-derived CM clusters, Gene Ontology (GO) annotations were used to group the genes according to biological process, molecular function, and cellular component [20]. By comparing with a reference list, overrepresentation of annotations among sets of genes can be calculated as O/E, where O is the observed number of genes with a specific annotation and E is the expected number of genes with that annotation. E is calculated as E = R * I/R, where R is the number of reference genes and I is the number of genes of interest. All genes represented on the arrays were used as the reference list. Significantly overrepresented annotations (p < .01) among the upregulated genes from all three categories at level five in the GO annotation database were identified using the hypergeometric test.

Analysis of Protein Interaction Network.

To investigate the possible interactions among proteins from the significantly upregulated genes in hESC-derived CM clusters, the search tool STRING was used to mine for recurring instances of neighboring genes [21, 22]. STRING aims to collect, predict, and unify various types of protein-protein associations, including direct (physical) and indirect (functional) associations. Using the list of upregulated genes in hESC-derived CM clusters as input to STRING, 431 matches were made and potential interactions among products from these genes were investigated further. A protein interaction map was created from these gene products and compared with protein interaction maps from 10 different randomly generated sets of genes, all of equal size. For each protein, an interaction score was calculated as n*2/N, where n is the number of interactions (edges in the map) for the protein in question and N is the total number of input proteins. Furthermore, the number of hub proteins was determined. As described by Han et al., we designated a protein as a hub if it had ≥5 interactions with other proteins [23]. A distinction between “party hubs,” which interact with most of their partners simultaneously, and “date hubs,” which interact with their partners at different times or locations, has also been made [23]. Since our data sets do not include time series data, we identified only party hubs. As default STRING uses four different sources (genomic context, high-throughput experiments, coexpression, and previous knowledge) to derive protein interaction maps. However, we restricted our analysis to include only experimentally determined protein interactions, excluding, for example, text mining, to increase the validity of the results.

Pathway Analysis.

To identify significantly upregulated pathways in hESC-derived CM clusters, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database [24] was searched using WebGestalt [25]. Among the 530 significantly upregulated genes, 417 were identified as known genes by WebGestalt, and these were analyzed for participation in curated pathways. By comparing with a reference list of genes, significant (p < .01) overrepresentation of genes in a particular pathway can be calculated as O/E, as described above. Pathways containing upregulated genes were identified using the hypergeometric test, and of these only those with at least five upregulated genes were reported. For comparison, the 10 randomly generated sets of genes were also subjected to identical analysis. All genes on the array were used as reference list.

Real-Time Quantitative PCR

In separate experiments, hESCs were differentiated to CM as described above, and spontaneously beating clusters were harvested within 22 days after onset of contraction. Total RNA was extracted from undifferentiated hESCs and from pooled hESC-derived CM clusters using RNeasy Micro Kit (Qiagen), according to the manufacturer's instructions. DNase treatment was performed using an RNase-free DNase Kit (Qiagen). cDNA synthesis was performed using a High-Capacity cDNA reverse transcription kit (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com) according to the manufacturer's protocol. Assay on demand predesigned primers and probes for the genes POU5F1 (Oct-4), ACTC1, SPARC, FN1, TGFBI, and COL1A1 were used for polymerase chain reaction (PCR) (Applied BioSystems) in a Taq Man ABI 7500 sequencer (Applied BioSystems). Relative quantification for a given gene, expressed as relative mRNA levels compared with the control (hESCs), was calculated after normalization to CREBBP and using the ΔΔCT formula.

Immunohistochemistry

Clusters of hESC-derived CM were prepared and isolated as described above. Cryosections (18 μm) of the CM clusters were fixed in 4% paraformaldehyde and blocked with 1% bovine serum albumin/0.01% Triton X-100 in Dulbecco's phosphate-buffered saline (DPBS) (GIBCO, Grand Island, NY, http://www.invitrogen.com). The primary antibodies were diluted in blocking buffer and incubated with the sections overnight at +4°C. The sections were subsequently washed three times for 5 minutes with DPBS and incubated with diluted secondary antibodies for 2 hours at room temperature. Nuclei were visualized using blue-fluorescent 4′,6-diamindino-2-phenylindole (1:1000). After washing three times with DPBS, the slides were mounted in DakoCytomation Fluorescent Mounting Medium (DAKO, Glostrup, Denmark, http://www.dako.com) and imaged using a Nikon E400 fluorescent microscope (Melville, NY, http://www.nikonusa.com/). The primary antibodies were diluted as follows: rabbit monoclonal anti-cardiac troponin I (ab52862; Abcam, Cambridge, U.K., http://www.abcam.com) 1:250, mouse monoclonal anti-collagen I (ab6308; Abcam) 1:100, mouse monoclonal anti-fibronectin (ab6328; Abcam) 1:200, mouse monoclonal anti-ACTC1 (ab7799; Abcam) 1:200, rabbit polyclonal anti-SPARC (ab14174; Abcam) 1:100, and mouse monoclonal anti-troponin I (MAB3438; Chemicon, Temecula, CA, http://www.chemicon.com) 1:200. The secondary antibodies were diluted as follows: donkey anti-rabbit AlexaFluor594 1:500 and donkey anti-mouse AlexaFluor488 1:500 (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Identification of Differentially Expressed Genes in hESC-Derived CM Clusters

Using the SAM algorithm, and FDR < 0.04, we identified 530 genes that were significantly upregulated in hESC-derived CM clusters compared with undifferentiated hESCs (supplemental online Table 1). Table 1 shows 50 genes displaying the highest average fold change. In addition, we also identified a smaller group of 40 genes that were significantly downregulated in the hESC-derived CM clusters compared with undifferentiated hESCs (supplemental online Table 2). Among these are several genes that are associated with pluripotent hESCs, such as NANOG, TDGF1, POU5F1, LEFTY1, DPPA4, DNMT3B, and SOX2, demonstrating that the hESC-derived CM clusters represent a differentiated cell population. To explore if the upregulated genes were previously known to be overexpressed in human heart tissue, the tissue expression analysis in WebGestalt was used. Of the 530 genes in supplemental online Table 1, 417 were identified by WebGestalt and 331 (79%) were labeled as “expressed” in human heart tissue. One hundred of the genes (24%) are designated as significantly overexpressed in heart tissue (marked as category O in supplemental online Table 1 and in Table 1). Several of these genes have been used before to characterize hESC-derived CMs (e.g., MYH6, MYH7, PLN, TNNT2, CKM, MYL2, MB, NPPA, ACTN2, GATA4, and MEF2C) [2]. In addition, 25 genes (6%) were designated as significantly underexpressed in heart tissue (marked as category U in supplemental online Table 1 and in Table 1). Interestingly, some of these genes (e.g., TF, FGG, TTR, and SERPINA1) were recorded as significantly overexpressed in human liver by WebGestalt, suggesting the presence of endodermal derivatives in the hESC-derived CM clusters. Unexpectedly, and of particular note, is that the gene MYH6 in supplemental online Table 1 and in Table 1 was not identified by WebGestalt due to lack of a tissue expression record. For this reason, and to maintain consistency in Table 1 (and supplemental online Table 1), MYH6 is not marked as significantly overexpressed in human heart in either of these tables. However, MYH6 has been reported to be a cardiac-specific gene in other studies [26].

Table Table 1.. Upregulated genes in human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters
Thumbnail image of

Gene Ontology Analysis

To further explore the functional properties of the genes that were upregulated in the hESC-derived CM clusters (supplemental online Table 1), we used available GO annotations to group the genes according to biological process, molecular function, and cellular localization. The significantly (p < .01) overrepresented annotations at level five in the GO structure were determined and are shown in Figure 1. For “biological process” (Figure 1A) several annotations relate to CM function and differentiation (e.g., muscle contraction, development of mesoderm and muscle, and cellular differentiation). Furthermore, a large group of the genes are connected to transcriptional activities as shown in Figure 1B (“molecular function”), and functions related to muscle contractions are overrepresented (tropomyosin binding and calcium ion binding). In Figure 1C (“cellular component”), there is a striking predominance of gene annotations coupled to the cytoskeleton components and the contractile apparatus. Taken together, many of the overrepresented annotations are clearly associated with cardiac development and/or CM function, which further confirms the presence of cells of the cardiac lineage in the spontaneously contracting hESC-derived CM clusters.

thumbnail image

Figure Figure 1.. Overrepresented Gene Ontology annotations. Significantly (p < .01) overrepresented Gene Ontology annotations among the 530 upregulated genes in human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters are shown in the picture. (A): Biological process, (B): molecular function, and (C): cellular component. The x-axis shows the number of genes with a specific annotation.

Download figure to PowerPoint

Protein Interaction Maps Among Upregulated Genes in hESC-Derived CM Clusters

Protein interactions among the products of the upregulated genes in the hESC-derived CM clusters were explored, and STRING was used to create interaction maps based on previously reported protein interactions. Substantially more interactions were identified among the products of the upregulated genes in the hESC-derived CM clusters compared with randomly generated sets of proteins (supplemental online Fig. 1). The average interaction score (defined in Materials and Methods) was notably higher among gene products from the upregulated genes in hESC-derived CM clusters (2.34) compared with the average interaction score in the 10 randomly generated sets of proteins (0.90). Furthermore, the presence of hubs, defined as proteins with at least five interactions, was also determined among the group of upregulated genes in hESC-derived CM clusters. Only experimentally determined protein interactions were considered, and in total 15 hubs were identified (Fig. 2 and Table 2). For comparison, the 10 randomly generated sets of genes were also analyzed for the presence of hubs. On average, only 3.4 hubs were identified in the equally sized random data sets. Notably, the identified hubs in the hESC-derived CM clusters show a high connectivity to each other (Fig. 2). This contrasts with the observations made about the hubs in the randomly generated data sets that show lower connectivity, typically with several small nonconnected submaps (data not shown).

thumbnail image

Figure Figure 2.. Hub protein network in human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters. Proteins are identified as hubs if they have at least five experimentally determined protein interactions among the products of the upregulated genes.

Download figure to PowerPoint

Table Table 2.. Hub proteins in human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters
Thumbnail image of

Pathway Analysis in hESC-Derived CM Clusters

To explore the activation of pathways that could be involved in CM differentiation, or in sustaining the cardiac phenotype, the KEGG pathway database was searched using the genes upregulated in hESC-derived CM clusters (supplemental online Table 1) as input. Eight pathways, containing significantly many of these genes, were identified (Table 3). On average, 0.7 pathways were identified as upregulated using the 10 randomly generated sets of genes, indicating a substantial number of specific interactions between the upregulated genes in the hESC-derived CM clusters. Among the pathways identified are the focal adhesion pathway and the calcium signaling pathway that contained 23 and 16 genes, respectively. Both pathways are implicated in the contractile apparatus of CMs. Furthermore, the hedgehog signaling pathway, which is involved in embryo development and heart formation [27], also displayed many genes that were upregulated in the hESC-derived CM clusters.

Table Table 3.. Upregulated pathways in human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters
Thumbnail image of

Real-Time Quantitative PCR and Immunohistochemical Analysis of hESC-Derived CM Clusters

To corroborate the results from the microarray analysis, we selected a number of the markers identified as hubs (Table 2) for further validation. In separate experiments the relative mRNA levels of Oct-4 (POU5F1), ACTC1, SPARC, FN1, TGFBI, and COL1A1 were determined in the hESC-derived CM clusters using real-time qPCR. The results are shown in Figure 3 and confirm the expression profiles of these genes obtained in the microarray analysis.

thumbnail image

Figure Figure 3.. Real-time quantitative polymerase chain reaction (qPCR) analysis of gene expression. The relative mRNA levels were determined in undifferentiated human embryonic stem cells (hESCs) and in hESC-derived cardiomyocyte (CM) clusters using real-time qPCR. The mRNA levels were normalized to the expression of the housekeeping gene CREBBP in each sample. The data are expressed as relative mRNA level compared with the mRNA level in undifferentiated hESCs (= 1), and the bars represent the average value of duplicate measurements. (A):POU5F1, (B):ACTC1, (C):SPARC, (D):FN1, (E):TGFBI, and (F):COL1A1.

Download figure to PowerPoint

We also performed immunohistochemical analysis of hESC-derived CM clusters using antibodies directed against the hub proteins cardiac α-actin, Sparc, fibronectin, and collagen type I. To visualize the CM within each cluster, we used antibodies directed against cardiac troponin I. The results are shown in Figure 4 and demonstrate the presence of these proteins in the clusters, although their staining patterns differ slightly. The cardiac α-actin and cardiac troponin I stainings overlap well in the CM clusters, and Sparc appears to be localized mainly along the edges of the cardiac troponin I-positive areas. Fibronectin, on the other hand, is detected in the cardiac troponin I-negative regions, whereas collagen type I is present throughout the clusters, overlapping both with cardiac troponin I-positive and -negative cells.

thumbnail image

Figure Figure 4.. Immunohistochemical staining of human embryonic stem cell (hESC)-derived cardiomyocyte (CM) clusters. Immunohistochemical evaluation of cryosections of hESC-derived CM clusters was performed as described in Materials and Methods. Sections from at least three different CM clusters were analyzed for each antibody, and the results were consistent across the samples. The panels in the top row (A, E, I, M) show overlays of the pictures underneath. Individual stainings for cardiac troponin I are shown in (C, F, K, O). Immunolabelings of cardiac α-actin (B), Sparc (G), fibronectin (J), and collagen I (N) are also indicated separately. 4′,6-diamindino-2-phenylindole (DAPI) staining of nuclei is shown in (D, H, L, P). Scale bar = 50 μm.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Microarrays provide powerful tools to study large-scale transcriptional activity. Using various differentiation protocols, several groups have reported on the capacity of hESCs to differentiate into spontaneously contracting cells that resemble early-stage CMs [4, 18, 28, 29]. In the present study, we have, for the first time, investigated the global gene expression profile of isolated spontaneously contracting hESC-derived CM clusters. Cell models based on hESCs provide opportunities to study, in detail, the molecular mechanisms regulating the development of the human cardiac lineage [7]. Furthermore, the differentiated CMs represent promising tools for drug development, especially in safety assessment of new drug candidates [30].

Despite the substantial progress made by different investigators during recent years, the knowledge of the molecular signature of hESC-derived CMs and the factors that induce cardiogenesis during embryonic development remains limited. Using microarrays and stringent inclusion criteria, we identified differentially expressed genes in hESC-derived CM clusters in comparison with undifferentiated hESCs, and validated the results by real-time qPCR and immunohistochemical analysis. The family of upregulated genes in hESC-derived CM clusters encompasses several cardiac marker genes (e.g., MYH6, MYH7, TNNT2, and MYL7) as well as cardiac-related transcription factors (e.g., TBX5, MEF2C, GATA4, and ISL1) (Table 1 and supplemental online Table 1), confirming the cardiac phenotype of the cell population. Interestingly, our results overlap partially with data obtained from human fetal heart tissue and hESCs differentiated to CMs in an END-2 coculture system where the hESC line HES-2 was used [15]. Fifteen genes were reported as enriched in the hESC-derived CMs and in fetal heart tissue. Eight of these genes are also upregulated in our hESC-derived CM clusters (e.g., TNNT2, PLN, and MYL7). This suggests that there are some similarities between the CM cell populations obtained from hESCs that appear to be independent of differentiation protocol and cell line used.

A search using the WebGestalt toolbox demonstrated that approximately 1/4 of the genes identified had been previously shown to be overexpressed in human heart tissue. However, a small fraction of the upregulated genes were classified by WebGestalt as underexpressed in human heart tissue and some lacked previous human heart tissue record. These observations may be because some of these genes are expressed only at the early stages of cardiac development and not in adult heart tissue. Another plausible explanation is that some of these genes are expressed by noncardiac cell types that also may be present in the hESC-derived CM clusters. Notably, several of the genes not previously associated with CMs instead indicate the presence of endodermal derivatives (e.g., TF, FGG, TTR, SERPINA1, FGA, ALDH1A1, ORM1, FABP1, and ZBTB20), further supporting the importance of endoderm for the induction of CM differentiation [31].

Among the group of downregulated genes in the hESC-derived CM clusters, we observed markers for undifferentiated hESCs, such as OCT4, NANOG, TDGF1, SOX2, INDO, and DNMT3B, which have previously been shown to be downregulated during differentiation of hESCs to CMs [15]. In addition, a number of upregulated and downregulated genes not previously associated with hESCs, CMs, or cardiac development were identified. The possible biological significance of these genes in these specific processes remains to be determined.

To characterize the family of upregulated genes in hESC-derived CM clusters, we performed a GO analysis. Notably, several of the significantly overrepresented GO annotation terms shown in Figure 1 are associated with CM properties and functions. For example, approximately 10% of the upregulated genes were annotated as “calcium ion binding” in the molecular function category. In addition, the results in the biological process and cellular component categories also support the presence of cardiac lineage in the hESC-derived CM clusters. For instance, the upregulated genes are typically associated with processes such as “muscle contraction,” “cell differentiation,” and “development.” In addition, the cellular component annotations have a predominance of “cytoskeletal compartments” and “myofibrillar structures” typical for the contractile apparatus active in CMs.

The gene MYH6 is approximately 3,000-fold upregulated in the hESC-derived CM clusters (supplemental online Table 1). In line with our results, recent studies have demonstrated substantial and specific upregulation of MYH6 in hESC-derived CMs [26, 29]. Furthermore, the MYH6 promoter coupled to puromycin resistance has been used for transgenic selection and enrichment of hESC-derived CM-producing cultures in which approximately 90% of the cells were CMs [26]. This strategy would appear to be applicable also to the culture system and hESC line used in the present study.

Among the upregulated genes in the hESC-derived CM clusters is also phospholamban (PLN), which is involved in CM contraction [32]. Interestingly, from a developmental biology standpoint, a recent study reported on the importance of phospholamban in myocyte differentiation [33]. Phospholamban was constitutively expressed in both myoblasts and differentiated myotubes, and trafficking of phospholamban to the plasma membrane was observed. The authors suggested a possible dual role for phospholamban in myocyte differentiation. This hypothesis is supported by our results, where PLN is highly upregulated in the hESC-derived CM clusters.

Besides analyzing and interpreting the results from the expression profiling on single genes individually, we set out to define protein interaction maps among the upregulated genes in hESC-derived CM clusters. As shown in supplemental online Figure 1, the protein interactions are substantially more complex between proteins coded by genes that are upregulated in hESC-derived CM clusters than what is observed in randomly generated sets of proteins of equal size. For example, considerably more hub proteins were identified and all the hub proteins interacted directly, or indirectly, with each other, which resulted in a fully connected interaction network (Fig. 2). In the interaction networks obtained from the randomly generated sets of genes, we typically observed a couple of smaller subnetworks that lacked interaction with each other. Additional studies are required to determine the importance of the specific hub proteins identified here. Furthermore, their coding genes represent interesting targets for knockout and overexpression studies as they potentially serve as key genes in hESC-derived CMs.

In addition to identifying protein interaction networks among the differentially expressed genes, we also report on significant upregulation of a number of cellular pathways. Strikingly, 23 genes in the focal adhesion pathway are significantly upregulated in hESC-derived CM clusters. The focal adhesion pathway has been implicated in a diverse array of cellular processes, including tissue remodeling, cell migration, embryogenesis, growth factor signaling, cell cycle progression, and cell survival [34, 35]. In addition, the role of focal adhesions in mechanotransduction in CMs has recently been highlighted [36]. Interestingly, besides affecting the beat-to-beat regulation of cardiac performance, mechanotransduction also influences the proliferation, differentiation, growth, and survival of the cellular components that comprise the human myocardium. Furthermore, in neonatal rat ventricular myocytes it has been reported that focal adhesion kinase regulates the activation of the MEF2 and JNK/c-Jun pathways, which have important roles in the early activation of the hypertrophic genetic program by mechanical stress in CMs [37].

Notably, 16 genes in the calcium signaling pathway are also significantly upregulated in the hESC-derived CM clusters. This observation is not unexpected since Ca2+ is an important component for the initiation and regulation of cardiac contraction [38]. In addition, Ca2+ has been shown to be important already at the beginning of life to mediate the process of fertilization, and later on it regulates some of the cell cycle events during early development [39]. In this regard, the hedgehog signaling pathway, also known to be critical in a plethora of developmental processes [40], is significantly upregulated in the hESC-derived CM clusters. Specifically, Sonic hedgehog appears to be a critical signaling factor for cardiac development in P19 cells [41]. In addition, experimental mice lacking hedgehog signaling show a delay in the expression of Nkx2.5 [41].

An apparent limitation of the present study is that we isolated whole spontaneously contracting clusters of cells and used these for subsequent analysis. Hence, there may be a proportion of noncardiac cells present in the clusters, which could contribute to the molecular signature obtained. However, the advantage of profiling the whole clusters instead of isolated hESC-derived CMs is that this provides the possibility to also identify molecular events that induce and/or sustain CM differentiation. In this regard, the hESC-derived CM clusters displayed significant upregulation of INHBA (Activin A), BMP4, and TGFB2, which have been implicated in the induction of cardiogenesis in ESCs [42]. Our results provide a foundation for subsequent studies of specific factors and pathways that could induce and sustain CM differentiation from hESCs.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

In the present study, we performed global gene expression analysis of spontaneously contracting hESC-derived CM clusters. Using the SAM algorithm, we identified 530 upregulated and 40 downregulated genes in the hESC-derived CM clusters compared with undifferentiated hESCs. Moreover, the GO annotations for the upregulated genes showed a significant overrepresentation of annotations associated with CMs and cardiac development. We further analyzed the upregulated genes and created protein interaction maps between their gene products. Interestingly, 15 protein hubs were identified, and a distinctly different topology was observed in the protein interaction maps of the hESC-derived CM clusters compared with maps created using randomly generated sets of genes. Furthermore, eight significantly upregulated cellular pathways were identified in the hESC-derived CM clusters, and several of these pathways have previously been reported as important in cardiac development and CM function. Taken together, our results provide a comprehensive characterization of contracting hESC-derived CM clusters at the transcriptional level and reveal novel key genes and pathways of potential importance for inducing and sustaining CM differentiation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

The studies presented herein were supported by Cellartis AB (Göteborg, Sweden) and the Information Fusion Research Program (University of Skövde, Sweden) under grant 2003/0104 from the Knowledge Foundation. Cellartis AB is a member of the EU-funded projects HeartRepair (contract no. LSHM-CT-2005-018630) and InvitroHeart (contract no. LSHB-CT-2007-037636). We acknowledge the Microarray Resource Centre at Lund University for help with the microarray experiments.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosure of Potential Conflicts of Interest
  9. Acknowledgements
  10. References
  11. Supporting Information
FilenameFormatSizeDescription
SC-07-1033_Supplemental__Table_1.pdf40KSupplemental Table 1
SC-07-1033_Supplemental__Table_2.pdf9KSupplemental Table 2
SC-07-1033_Supplemental_Figure_1.pdf153KSupplemental Figure
SC-07-1033_Supplemental_Figure_Legend.pdf12KSupplemental Figure Legend

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.