Long-term self-renewing human epicardial cells generated from pluripotent stem cells under defined xeno-free conditions

Published online:


The epicardium contributes both multi-lineage descendants and paracrine factors to the heart during cardiogenesis and cardiac repair, underscoring its potential for use in cardiac regenerative medicine. Yet little is known about the cellular and molecular mechanisms that regulate human epicardial development and regeneration. Here, we show that the temporal modulation of canonical Wnt signalling is sufficient for epicardial induction from six different human pluripotent stem cell (hPSC) lines, including a WT1-2A-eGFP knock-in reporter line, under chemically defined, xeno-free conditions. We also show that treatment with transforming growth factor beta (TGF-β)-signalling inhibitors permitted long-term expansion of the hPSC-derived epicardial cells, resulting in more than 25 population doublings of WT1+ cells in homogenous monolayers. The hPSC-derived epicardial cells were similar to primary epicardial cells both in vitro and in vivo, as determined by morphological and functional assays, including RNA sequencing. Our findings have implications for the understanding of self-renewal mechanisms of the epicardium and for epicardial regeneration using cellular or small-molecule therapies.

The epicardium is the outermost mesothelium layer of the heart, and contributes both multi-lineage descendants and trophic signals to the myocardium and coronary vessels1,2. During cardiogenesis, epicardial cells undergo an epithelial-to-mesenchymal transition (EMT) and invade the myocardium to form epicardial-derived cells (EPDCs) including cardiac fibroblasts, coronary smooth muscle cells and at least a subset of endothelial cells3,4. In addition, mounting evidence points to the epicardium as having a paracrine role in modulating injury responses as well as being a source of progenitor cells in facilitating neovascularization and myocardial repair5,6. Particularly, epicardial cells were shown to contribute to the cardiomyocyte lineage, highlighting their cardiogenic potential in cardiac regeneration and repair7. Therefore, understanding the molecular mechanisms that control the specification and self-renewal of epicardial lineages from naive progenitor cells is of fundamental importance to elucidate the regulatory mechanisms underlying both human heart development and cardiovascular diseases.

The advent of human pluripotent stem cells (hPSCs) significantly advanced the study of early human embryonic development, disease modelling and cell-based therapy strategies8. Recently, there have been substantial advances in elucidating the regulatory mechanisms that control cardiomyocyte9,​10,​11,​12, endothelial cell13,​14,​15 and smooth muscle cell16,17 specification from hPSCs. More recently, epicardial cells have been generated from hPSCs via embryoid bodies (EBs) or monolayer differentiation with growth factor cocktails in processes that manipulate signalling pathways that regulate epicardial differentiation18,19. However, these hPSC models have not clearly elucidated developmental signalling mechanisms that are necessary and sufficient to specify epicardial cell formation from hPSCs. In addition, human epicardial cells spontaneously undergo an EMT in culture20, and little is known about signalling pathways that regulate epicardial cell self-renewal both in vivo and in vitro, limiting the ability to expand these hPSC-derived epicardial cells for clinical applications.

To address these issues, we generated self-renewing epicardial cells from hPSCs and investigated key signalling pathways during epicardial development and regeneration. We generated a WT1-2A-eGFP knock-in human embryonic stem cell (hESC) line and showed that temporal modulation of canonical Wnt signalling via small molecules is sufficient for epicardial induction from hPSCs under chemically defined, xeno-free conditions, and that a TGF-β-signalling-inhibitor treatment allows long-term self-renewal of hPSC-derived epicardial cells. In addition, these hPSC-derived epicardial cells retain many characteristics of primary epicardial cells, including formation of a polarized epithelial sheet, expression of the key epicardial genes WT1, TBX18 and ALDH1A2, and the ability to form smooth muscle cells and cardiac fibroblasts both in vitro and in vivo. RNA sequencing (RNA-seq) expression data and gene-set enrichment analysis (GSEA) confirm a similarity of hPSC-derived epicardial cells to primary human epicardial cells. These findings improve our understanding of differentiation and self-renewal mechanisms of the epicardium, and have implications for stimulating epicardial regeneration via small molecules or cell-based therapies.

Albumin-free conditions to generate cardiac progenitors

We previously demonstrated that temporal modulation of canonical Wnt signalling in RPMI basal medium (known as the GSK3 inhibitor, WNT inhibitor (GiWi) protocol) is sufficient to generate functional cardiomyocytes (CMs) from hPSCs21. We found that ISL1+NKX2.5+FLK-1+ cardiac progenitors are generated as intermediates during the GiWi protocol (Supplementary Fig. 1A). When re-passaged on gelatin-coated plates in LaSR basal medium14 on day 6, these hPSC-derived cardiac progenitors expressed progenitor markers including ISL1, NKX2.5 and FLK-1, as well as the proliferation marker Ki67 (Supplementary Fig. 1B). Molecular analysis of cardiac progenitor differentiation from hPSCs revealed dynamic changes in gene expression, with downregulation of the pluripotency markers OCT4 and NANOG, and induction of the primitive-streak-like gene T (refs 14,22) in the first 24 h after CHIR99021 (CHIR) addition (Supplementary Fig. 1C). Expression of the cardiac progenitor markers ISL1, NKX2-5 and FLK-1 was first detected between days 3 and 5, and was significantly upregulated at day 6 (Supplementary Fig. 1C).

Wnt/β-catenin signalling regulates epicardial specification

Pro-epicardium arises from ISL1+NKX2.5+ second-heart-field progenitors in vivo 1,23. To identify signalling mechanisms regulating cardiac progenitor specification to epicardial cells, we treated day 6 cardiac progenitors with different small molecules and protein modulators of developmental signalling pathways for 48 h (days 7 to 9) (Fig. 1a and Supplementary Table 1). hPSC-derived cardiac progenitors formed more than 85% WT1+ putative epicardial cells following CHIR treatment (Fig. 1b,c and Supplementary Movie 1), demonstrating that Wnt signalling induction between days 7 and 9 is sufficient to generate epicardial cells from cardiac progenitor cells in the absence of other exogenous signalling. In the absence of CHIR, robust beating sheets of cTnT+ cardiomyocytes were observed (Fig. 1b,c and Supplementary Movie 2), suggesting that the activation status of canonical Wnt signalling at day 7 toggles epicardial versus cardiomyocyte differentiation. Interestingly, untreated and bone morphogenetic protein 4 (BMP4)-, dorsomorphin (DM)- and retinoic acid (RA)-treated cells also yielded about 10% WT1+ cells that were distinct from cTnT+ cells (Fig. 1b,c and Supplementary Fig. 2A). Compared to the untreated condition, BMP4 did not significantly increase the WT1+ cell population up to 100 ng ml−1 when applied from days 7–9 (Supplementary Fig. 2B). To exclude the potential risk of using low-quality BMP4 protein, we tested the function of BMP4 used in our study by inducing brachyury expression according to a previous report24. As expected, 5 ng ml−1 BMP4 treatment for 2 days in mTeSR1 resulted in robust expression of brachyury (Supplementary Fig. 2C). In the presence of CHIR, BMP4 treatment did not generate cardiomyocytes, but instead yielded an unknown population at the expense of WT1+ cells. Inhibition of BMP4 signalling via DM resulted in a similar purity of WT1+ cells as CHIR treatment alone (Fig. 1b and Supplementary Fig. 2A), suggesting that BMP4 signalling is dispensable at this stage of epicardial development.

Figure 1: Wnt/β-catenin signalling directs the specification of WT1+ epicardial lineages from hPSC-derived cardiac progenitors.
Figure 1

a, A schematic of the protocol used to differentiate NKX2.5+ISL1+ hPSC-derived cardiac progenitors towards the epicardial lineage. b,c, H13 hESC-derived cultures differentiated as shown in a in the presence of the indicated molecular signalling regulators were subjected to flow cytometry analysis (b) and immunostaining analysis (c) for WT1 and cTnT at day 12. Scale bars, 100 μm. Data are represented as mean ± s.e.m. of five independent replicates. #P < 0.05, indicated treatment versus untreated condition. CHIR, CHIR99021; DM, dorsomorphin; PURM, purmorphamine; RA, retinoic acid.

Construction of WT1-2A-eGFP knock-in reporter hPSCs

WT1 is required for the development of epicardium25 and for the formation of cardiovascular progenitor cells26. To better monitor the epicardial cell differentiation process and purify hPSC-derived epicardial cells in vitro, we engineered the ES03 cell line via CRISPR/Cas9-catalysed homology-directed repair (HDR) and generated a homozygous WT1-2A-eGFP knock-in reporter cell line (Fig. 2a). Two 2-kilobase homologous arm sequences located just upstream and downstream of the WT1 stop codon were inserted into the Oct4-2A-eGFP donor plasmid27 and replaced the Oct4 homologous arms. We then introduced the 2A-eGFP sequence into the target sites by transfecting hESCs with the WT1-2A-eGFP donor plasmid and the Cas9/sgRNA plasmids. After puromycin (Puro) selection, PCR genotyping and sequencing showed that approximately 50% (21 out of 44) of the clones were targeted in one (heterozygous) and approximately 25% (12 out of 44) in both alleles (Fig. 2b), similar to a previous report28. The homozygous clones were then subjected to TAT-Cre recombinase treatment and the PGK-Puro cassette was excised from WT1-2A-eGFP (Fig. 2c). WT1-2A-eGFP-targeted hPSCs after Cre-mediated excision of the PGK-Puro cassette were subjected to CHIR treatment; eGFP was detected at day 10 and elevated at day 12 (Fig. 2d). Dual immunostaining with anti-WT1 and anti-GFP antibodies detected expression of eGFP in WT1+ cells (Fig. 2e), demonstrating the successful generation of a WT1 reporter cell line for potential cell tracking or purification.

Figure 2: Construction of the WT1-2A-eGFP knock-in ES03 hESC line using Cas9 nuclease.
Figure 2

a, A schematic diagram of the knock-in strategy at the stop codon of the WT1 locus. Vertical arrows indicate the sgRNA1 and sgRNA2 targeting sites. Red and blue horizontal arrows are PCR primers for assaying WT1-locus targeting and homozygosity, respectively. b, Representative PCR genotyping of hESC clones after puromycin selection is shown, and the expected PCR product for correctly targeted WT1 locus is ~3 kbp (red arrow) with an efficiency of 21 clones from a total of 44. A homozygosity assay was performed on the knock-in clones, and those without ~200 bp PCR products were homozygous (blue arrow). c, PCR genotyping of hESC clones after TAT-Cre-mediated excision of the PGK-Puro cassette. Clones with PCR products of ~1 kbp are PGK-Puro free, and those with ~3 kbp contain PGK-Puro. d, Live-cell flow analysis of GFP+ cells at days 0, 10 and 12 after initial CHIR treatment of WT1-2A-eGFP knock-in ES03. The numbers on the plots refer to the percentage of analysed cells in the indicated gated region. e, Phase-contrast images and corresponding eGFP fluorescent images of WT1-2A-eGFP hPSC-derived epicardial cells after excision of the PGK-Puro cassette. BF, bright field. Scale bars, 100 μm.

Chemically defined conditions to generate epicardial cells

We next optimized the concentration of CHIR and initial seeding density of cardiac progenitors at day 6 in LaSR basal medium, and found that 3 μM CHIR with an initial density of 0.06 million cells cm−2 yielded more than 95% WT1+ cells (Supplementary Fig. 3A–D), whereas the control (no CHIR) resulted in less than 10% WT1-2A-eGFP cells. However, LaSR basal medium, which contains bovine serum albumin, adds xenogenic components to the medium, which would not be desirable for the generation of epicardial cells that meet clinical requirements. In order to develop a xeno-free protocol, we systematically screened four commercially available basal media supplemented with 1 μg ml−1 human recombinant insulin and 100 μg ml−1 ascorbic acid (vitamin C), as these two factors were shown to improve the culture of cardiac cell lineages29,​30,​31. Use of the growth media DMEM, DMEM/F12 and RPMI resulted in the generation of more than 95% WT1+ putative epicardial cells from hPSC-derived cardiac progenitors (Supplementary Fig. 3E). To simplify the differentiation pipeline, we employed RPMI as the basal medium, referring to epicardial cell generation from hPSCs as the GiWiGi (GSK3 inhibitor, WNT inhibitor, GSK3 inhibitor) protocol.

Epicardial cell differentiation is β-catenin dependent

Selectivity is a concern when using chemical inhibitors of signalling pathways. Therefore, we tested other GSK3 inhibitors including BIO-acetoxime and CHIR98014 in the GiWiGi protocol, and found that 0.3 μM CHIR98014 and BIO-acetoxime generated WT1+ cells as effectively as 3 μM CHIR99021 (Supplementary Fig. 4A). In addition, we treated day 6 cardiac progenitors with Wnt3a of concentrations up to 500 ng ml−1, and found that Wnt3a significantly increased the WT1+ cell population compared with the no-Wnt3a control, although Wnt3a was less effective than small-molecule GSK3 inhibitors in generating WT1+ cells (Supplementary Fig. 4B). To further investigate the role of β-catenin in the GiWiGi epicardial differentiation, we employed an inducible pluripotent stem cell (iPSC) line (19-9-11 ischcat-1) expressing β-catenin shRNA under the control of a tet-regulated inducible promoter described in earlier work10. Upon doxycycline (dox) treatment, the shRNA efficiently downregulated β-catenin expression10. We also showed that the induction of NKX2.5+ISL1+ cardiac progenitors from hPSCs is β-catenin dependent10. In this study, we therefore focused on the examination of the stage-specific roles of β-catenin during the differentiation of epicardial cells from cardiac progenitors stimulated by GSK3 inhibition. We found that β-catenin knockdown at day 6 yielded significantly fewer WT1+ cells, instead generating robust beating sheets of cTnT+ cardiomyocytes at the expense of WT1+ cells (Supplementary Fig. 4C,D and Supplementary Movie 3). This finding is consistent with reports that Wnt/β-catenin inhibition is necessary for cardiomyocyte formation from cardiac progenitors both in vitro and in vivo 10,18,32,33, and further supports the notion that Wnt/β-catenin signalling regulates epicardial versus cardiomyocyte specification from cardiac progenitors. The inhibitory effects of β-catenin knockdown on WT1+ cell generation gradually diminished after day 6, with no inhibition after day 9 (Supplementary Fig. 4C,D and Supplementary Movies 3,4).

Molecular characterization of hPSC-derived epicardial cells

Pro-epicardial cells are marked by the expression of TBX18, WT1 and TCF2134,​35,​36. Molecular analysis of epicardial cell differentiation from hPSC-derived cardiac progenitors (Fig. 3a) revealed dynamic changes in gene expression, with upregulation of WT1 and TBX18, and undetectable changes in TNNT2 (Fig. 3b). This gene upregulation was consistent with the increased WT1-eGFP signals (Fig. 2d), and was also confirmed by western blot analysis of WT1, TBX18 and TCF21 expression (Fig. 3c and Supplementary Fig. 5A). Immunofluorescent analysis revealed expression of the pro-epicardial markers WT1, TBX18 and TCF21 (Fig. 3d,e). After passage at a low density, these cells adopted a cobblestone-like organization typical of cultured primary epicardium20,37 (Fig. 3e). In addition, the cells displayed intense β-catenin and ZO1 localization at sites of cell-to-cell contact. Taken together, these data confirm the epithelial nature of the cells. These post-passaged cells also expressed aldehyde dehydrogenase enzyme retinaldehyde dehydrogenase 2 (ALDH1A2) (Supplementary Fig. 5A–C), suggesting the ability to produce retinoic acids. Therefore, we refer to the day 12, pre-passaged WT1+ALDH1A2 cells as pro-epicardial cells (Pro-Epi) and the post-passaged WT1+ALDH1A2+ cells as epicardial cells (Epi). The GiWiGi protocol was also effective in other hPSC lines, including the hESC lines H9 and ES03, and the 19-9-7 iPSC line, generating more than 95% WT1+ cells (Supplementary Fig. 6A and Supplementary Table 2). Post-passaged epicardial cells retained the expression of WT1 and displayed strong β-catenin and ZO1 staining along the cell borders (Supplementary Fig. 6B).

Figure 3: Molecular analysis of hPSC-derived epicardial cells under chemically defined, albumin-free conditions.
Figure 3

a, A schematic of the optimized protocol for differentiation of hPSCs to epicardial cells in RPMI basal medium. Vc, vitamin C. b, H13 hESC-derived cardiac progenitors were differentiated as illustrated in a. Gene expression was assessed by quantitative RT-PCR. Data are represented as mean ± s.e.m. of three independent replicates. #P < 0.05, CHIR-treated condition versus untreated condition at indicated time; t-test. c, At different time points, WT1 and TBX18 expression was assessed by western blot. d, At day 12, immunostaining for TBX18 and TCF21 was performed. Scale bars, 50 μm. e, Representative phase-contrast microscopy and fluorescence immunostaining for WT1, ZO1 and β-catenin of day 12 pro-epicardium (Pro-Epi) and day 18 epicardium (Epi). Scale bars, 100 μm.

Long-term expansion of hPSC-derived WT1+ epicardial cells

Primary mouse epicardial cells have been cultured for more than 3 years34, but epicardial cells isolated from the adult human heart rapidly undergo EMT in culture20. Similar to primary human epicardial cells, hPSC-derived WT1+ epicardial cells only retained their morphology for approximately 2 weeks in culture. Although, in the short term, hPSC-derived WT1+ epicardial cells retained the epicardial cobblestone-like morphology, basic fibroblast growth factor (bFGF)- and TGF-β1-treated cells adopted a fibroid spindle or fusiform-shaped appearance typical of cultured fibroblasts and smooth muscle cells, respectively (Fig. 4a,b). The expression of calponin and smooth muscle myosin heavy chain (SMMHC) in TGF-β1 + bFGF-induced cultures further support their smooth-muscle-cell identity, and vimentin (VIM) and CD90 expression support their fibroblast identity (Supplementary Fig. 7). We also cultured WT1+TBX18+ epicardial cells in endothelial cell medium, but did not detect expression of the endothelial markers CD31 and VE-cadherin (Supplementary Fig. 7). The expression of WT1 and ZO1 significantly decreased in both bFGF- and TGF-β1-treated samples, indicating the transition from epithelial towards mesenchymal-like cells (Fig. 4b). In addition, CDH1 (E-cadherin) expression decreased whereas CDH2 (N-cadherin) and SNAIL2 expression increased in all treated samples compared with untreated controls, suggesting that bFGF and TGF-β1 induced an EMT in hPSC-derived epicardial cells (Fig. 4c,d).

Figure 4: hPSC-derived epicardial cells undergo an EMT in response to bFGF and TGF-β1 treatment, yielding epicardium-derived cells that display characteristics of fibroblasts and vascular smooth muscle cells.
Figure 4

a, A schematic of the protocols used for the EMT induction of H13 hESC-derived epicardial cells with 10 ng ml−1 bFGF and 5 ng ml−1 TGF-β1. b, Phase-contrast images displaying cell morphology, and fluorescence images taken at day 18, showing the presence of WT1, ZO1, α-SMA and TCF21. Scale bars, 100 μm. c, qPCR analysis of the EMT-related genes SNAIL2, CDH2 and CDH1. d, Immunostaining analysis of E-cadherin expression after the indicated bFGF and TGF-β1 treatments. Data are represented as mean ± s.e.m. of three independent replicates. #P < 0.05, untreated condition versus bFGF, TGF-β1 and bFGF + TGF-β1 conditions at the indicated time. Scale bars, 50 μm.

During long-term in vitro culture, hPSC-derived WT1+ epicardial cells spontaneously underwent an EMT and lost WT1 expression after several passages, even without exogenous TGF-β1 or bFGF treament (Fig. 5a). To identify signalling mechanisms regulating hPSC-derived epicardial cell self-renewal, we applied small molecules (Supplementary Table 1) that affect pathways that regulate cell proliferation, to study their effects on WT1+ cell self-renewal. TGF-β receptor 1 antibody (α-TGF-βI) and α-TGF-β pan-specific ­antibody (α-TGF-β Pan) partially increased the proliferation of epicardial cells and maintained their epithelial phenotype (Fig. 5a,b). A83-01, an inhibitor of TGF-β signalling, enabled expansion of hPSC-derived epicardial cells that retained polarized epithelial morphology and WT1 expression to a significantly greater extent than the untreated control (Fig. 5a,b). To reduce the likelihood that WT1+ cell self-renewal was an off-target effect of A83-01, we tested additional TGF-β inhibitors (RepSox, SB505124 and SB431542) and found that they also promoted the proliferation of hPSC-derived epicardial cells to a similar extent as A83-01 (Fig. 5b,c). Following A83-01 or SB431542 addition, hPSC-derived epicardial cells were capable of at least 25 population doublings, generating more than 10 million cells from a single hPSC-derived epicardial cell clone (Fig. 5c). After 48 days of expansion, the TGF-β-inhibitor-treated cells expressed significantly higher ­levels of WT1 and Ki67 than the untreated control cells (Fig. 5d). Epicardial cells generated from 19-9-11 iPSCs were also expandable after A83-01 treatment (Supplementary Fig. 8A), presenting a cobblestone morphology and expressing high levels of ALDH1A2, WT1 and TBX18 (Supplementary Fig. 8B–D). In addition, epicardial cells expanded for 48 days retained the potential to differentiate into SMMHC+calponin+ smooth muscle cells and VIM+CD90+ fibroblasts following TGF-β1 and bFGF treatment, respectively (Supplementary Fig. 8D). The day 50 epicardial cells retained normal karyotypes after long-term culture in medium containing A83-01 (Supplementary Fig. 9). These findings improve our understanding of self-renewal mechanisms of the epicardium and have implications for generating large quantities of hPSC-derived epicardial cells for research or cell-based therapy applications.

Figure 5: Long-term expansion of hPSC-derived epicardial cells. a,b, H13 hESC-derived day 18 epicardial cells were seeded at a density of 0.05 million cells per cm2 and treated with the indicated small molecules for 3 days (concentrations provided in Supplementary Table 1).
Figure 5

At day 4, representative phase-contrast microscopy and fluorescence immunostaining for WT1, ZO1 and α-SMA was performed (a) and the total cell numbers were assessed (b). Data are represented as mean ± s.e.m. of five independent replicates. #P < 0.05, treatment versus untreated condition; one-way ANOVA with Tukey’s honest significant difference (HSD) as post-hoc analysis. α-TGF-βI, TGF-β receptor 1 antibody; α-TGF-β Pan, TGF-β pan-specific antibody. PD, PD0325901; DM, dorsomorphin; VP, verteporfin; RA, retinoic acid; RO, RO4929097. Scale bars, 100 μm. c,d, H13 hESC-derived epicardial cells were passaged and counted every four days in the absence or presence of the indicated TGF-β inhibitors: 0.5 μM A83-01 or 2 μM SB431542. The population doublings were calculated (c), and day 48 cultures were subjected to flow-cytometry analysis of WT1 and Ki67 expression (d).

hPSC-derived epicardial cells were similar to primary cells

To further confirm the identity of hPSC-derived epicardial cells, RNA from four different hPSC-derived epicardial cell differentiations at two different expansion time points (days 12 and 48) and primary epicardial cells of four different donors were subjected to RNA-seq analysis. Hierarchical clustering analysis of RNA-seq expression data of hPSCs38,​39,​40, hPSC-derived endoderm (Endo)38, ectoderm (Ecto)39, mesoderm (Mes), CMs41, epicardial cells (Epi) and primary epicardial cells (Fig. 6a) showed that hPSC-derived epicardial cells were most closely related to primary epicardial cells and were distinct from all other cell populations together as a group. Day 12 hPSC-derived epicardial cells clustered closely with day 48 hPSC-derived epicardial cells. Next we explored the relationship between different cell types relevant for development (including hPSCs, mesoderm, cardiomyocytes and epicardial cells) using principal component analysis (PCA) on the gene expression data. hPSCs clustered relatively closer to mesoderm cells, from which epicardial cells and CMs divergently formed, in the 3D scores plot for the first three principal components (Fig. 6b). Importantly, hPSC-derived epicardial cells showed highest similarity with donor epicardial cells. We also performed GSEA to identify significantly enriched pathways (P < 0.05) in each cell type relative to hPSCs. Hierarchical clustering of the absolute value of normalized enrichment scores (NES) of these pathways confirmed the similarity between epicardial cells from hPSCs and those from donors (Supplementary Fig. 10A). As both epicardial cells and CMs are derived from mesoderm, we further compared the differences and similarities in the enriched pathways among these cell types. We observed that whereas 42 pathways were commonly enriched in all cell types, hPSC-derived epicardial cells shared 36 pathways with donor epicardial cells, and 22 with cardiomyocytes (Supplementary Fig. 10B and Supplementary Tables 3–5). Microarray data analysis has shown the enrichment of cell-adhesion and extracellular-matrix-organization genes in mouse primary epicardial cells42. Similarly, our hPSC-derived and donor epicardial cells also showed enrichment in extracellular-matrix-related pathways and keratinocyte (epithelial) differentiation, whereas CMs were enriched in heart-development- and heart-contraction-related pathways, as expected. Donor epicardial cells were highly enriched in ­endoplasmic-reticulum-related pathways compared with hPSC-derived epicardial cells.

Figure 6: hPSC-derived epicardial cells were similar to primary epicardial cells.
Figure 6

a, Hierarchical clustering analysis of RNA-seq expression data of hPSCs, hPSC-derived endoderm (Endo), ectoderm (Ecto), mesoderm (Mes), CMs, day 12 epicardial cells (19-9-7-Epi), day 48 epicardial cells (H9-Epi, ES03-Epi and 19-9-11-Epi) and primary epicardial cells (donors 9605, 9633, 9634 and 9635). b, 3D scores plot of the first three principal components (PCs) from the principal component analysis. The ellipses show the 95% confidence limit and each data point corresponds to different biological samples. Black arrows show the development transition from hPSCs to mesoderm, from which CMs and epicardial cells arise. c, Before transplantation to mouse heart, ES03-eGFP cells were differentiated as illustrated in Fig. 3a, cultured for five passages in A83-01-containing medium, and subjected to flow-cytometry analysis for WT1 and GFP expression. d,e, After 12 days, hearts were harvested, and representative haematoxylin and eosin (H&E) staining and dual immunostaining plots of smooth muscle actin (SMA) and human-specific mitochondria (Mito) (d) and SMA and GFP (e) of cross-sections of the hearts are shown. Arrows in the H&E images denote the corresponding sites in the immunofluorescence images. Scale bars, 50 μm.

Epicardial cells can undergo an EMT and give rise to cardiac fibroblasts and smooth muscle cells after transplantation into chicken embryos19 or infarcted mouse hearts20. To examine the ability of hPSC-derived epicardial cells to invade the myocardium and undergo an EMT in vivo, cardiac-fibroblast-derived extracellular matrix (CF-ECM) patches seeded with passage 5 eGFP-labelled hPSC-derived epicardial cells (Fig. 6c) were transferred to the heart surface (Supplementary Fig. 10C) of a mouse myocardial infarction model. eGFP+ cells were detected predominantly within the CF-ECM scaffold and in the epicardium beneath the scaffold before day 6, and scattered within the mid-myocardium after 12 days (Supplementary Fig. 10D,E), suggesting that epicardial cells invaded the myocardium. In addition, the hPSC-derived cells underwent an EMT and differentiated into αSMA+calponin+ smooth-muscle-like cells (Fig. 6d,e and Supplementary Fig. 10F,G) and VIM+ fibroblast-like cells (Supplementary Fig. 10F) in vivo. We did not observe any tumour formation 12 days after hPSC-derived epicardial cell transplantation into the mouse heart compared with the patch-only control (Supplementary Fig. 10H). These findings demonstrate that hPSC-derived epicardial cells can invade the myocardium and form EPDCs after infarction, underscoring their potential for cell-based therapeutic heart regeneration.


While differentiation of hPSCs to epicardial cells by applying complex mixtures of growth factors from different signalling pathways has been reported18,19, it is largely unknown which developmental signalling pathways are necessary and sufficient to specify epicardial cell fate and to control their self-renewal, limiting their large-scale production for clinical applications. This study reports the generation of a WT1-2A-eGFP knock-in stem cell line, and demonstrates efficient and robust generation of epicardial cells from multiple hPSC lines solely via stage-specific manipulation of Wnt/β-catenin signalling under chemically defined, albumin-free, xeno-free conditions. These hPSC-derived epicardial cells retain many characteristics of primary epicardial cells, including formation of an epithelial sheet, expression of the key epicardial proteins WT1, TBX18 and ALDH1A2, and the ability to generate fibroblast and vascular smooth muscle lineages both in vitro and in vivo. Their identity was further confirmed by RNA-seq expression data and GSEA at a global level. Using inducible knockdown hPSC lines, we showed that β-catenin is essential for epicardial cell induction from hPSC-derived cardiac progenitors under the GiWiGi protocol. Given the essential roles of β-catenin during cardiac progenitor induction from hPSCs10,43, we conclude that β-catenin is required at multiple stages of hPSC differentiation to epicardial cells via small-molecule modulation of canonical Wnt signalling. Differences between our study and previous reports18,19, including the initial starting cardiac progenitor cells and the exposure windows for signalling modulators, may account for their conclusion that BMP4 signalling, but not Wnt alone, is required for robust epicardial differentiation.

This study also demonstrates long-term self-renewal of hPSC-derived epicardial cells via TGF-β-inhibitor treatment in a chemically defined medium. For cell-based therapeutic applications, it is highly desirable to generate homogeneous committed progenitors that can expand in culture and differentiate into various tissue-specific cells of interest, avoiding the contamination of unwanted cell lineages, especially tumorigenic hPSCs44. We showed that TGF-β- inhibitor treatment is sufficient for the self-renewal of hPSC-derived epicardial cells, in contrast to primary mouse epicardial cells, which can self-renew in the absence of a TGF-β inhibitor34. It will also be interesting to test whether or not TGF-β-inhibitor treatment can promote the self-renewal of human primary epicardial cells in vitro. Recent work has demonstrated that epicardial cell lineages improved the performance of the scarred myocardium by preservation of cardiac function and attenuation of ventricular remodelling after transplantation into a myocardial infarction model45. More recently, it has been reported that the epicardium of the zebrafish heart is required for muscle regeneration and can itself regenerate following Sonic hedgehog (Shh) treatment46. Our results suggest that TGF-β inhibitors may impact heart regeneration following injection into the epicardium in vivo, similarly to strategies used to test the effect of TGF-β inhibitors on scar formation after glaucoma surgery in rabbits47.

In summary, our findings support a model (Fig. 7) of human epicardial development in which small-molecule-mediated exogenous modulation of Wnt/β-catenin signalling is sufficient for the specification of epicardial cells from hPSCs. This finding is consistent with the report that DKK1 and DKK2 double-null mice increase epicardial specification and display a hypercellular epicardium48. This completely defined, xeno-free epicardial differentiation platform is compatible with current good manufacturing practice (cGMP) standards and can be employed to efficiently derive self-renewing epicardial cell lineages from hPSCs, which can thereby provide insights into mechanisms of heart development, maturation and response to cardiac injury. Moreover, we show that hPSC-derived epicardial cells can invade the myocardium in an infarcted mouse model, suggesting potential applications in cell-based heart regeneration. Our results also point to TGF-β signalling as a regulator of epicardial cell self-renewal and differentiation, indicating the potential of TGF-β signalling modulators in heart regeneration.

Figure 7
Figure 7

A model highlighting the specification of hPSCs to epicardial lineages by stage-specific modulation of canonical WNT signalling and the long-term maintenance of hPSC-derived epicardial cells using TGF-β-signalling inhibitors.


Construction of donor plasmid and sgRNA

Human codon-optimized Streptococcus pyogenes wild-type Cas9 (pCas9-2A-eGFP) was obtained from Addgene (plasmid no. 44719) and chimeric guide RNA expression cassette was cloned into this Cas9-2A-eGFP plasmid with two BbsI restriction sites for rapid sgRNA cloning. Two sgRNAs targeting at or near WT1 stop codon (1: AACTCCAGCTGGCGCTTTGAGGG and 2: GGACACTGAACGGTCCCCGAGGG) were used. To generate the WT1-2A-eGFP donor plasmid, DNA fragments of about 2 kb in length were PCR amplified from the genomic DNA before and after the stop codon of WT1 and were cloned into the OCT4-2A-eGFP donor plasmid27 (Addgene no. 31938), replacing the OCT4 homologous arms.

Maintenance of hPSCs and TAT-Cre treatment of WT1 knock-in hPSCs

Transgene and vector-free hPSCs were maintained on Matrigel (Corning) or SyntheMax (BD Biosciences) coated plates in mTeSR1 or E8 medium (STEMCELL Technologies), according to previously published methods49. To remove the PGK-Puro cassette from the WT1-2A-eGFP cells, targeted homozygous clones were treated with 2 μM TAT Cre recombinase (Excellgen, EG-1001) for 6 h in E8 medium. After two days, cells were singularized with Accutase and seeded into a Matrigel-coated 96-well plate at a density of 100 to 150 cells per well. After two weeks, the cells were subjected to PCR genotyping.


hESCs were pre-treated with 10 μM ROCK inhibitor (Y27632) for 3 to 4 h before electroporation. Cells were digested by Accutase (Innovative Cell Technologies) at 37 °C for 8 min and 2.5–3 million single cells were electroporated with 3 μg gRNA1, 3 μg gRNA2 and 6 μg WT1-2A-eGFP donor plasmids in 200 μl cold PBS−/− using the Gene Pulser Xcell System (Bio-Rad) at 320 V, 200 μF and 1,000 Ω (time constant should be around 15 ms) in a 0.4 cm cuvette. Two electroporations were performed and 5–6 million cells were subsequently plated onto a Matrigel-coated 10 cm dish in 10 ml mTeSR1 with 10 μM Y27632. 24 h later, and every day afterwards, the medium was changed with fresh mTeSR1. Three days after electroporation, 1 μg ml−1 puromycin was added into the mTeSR1 for selection for about two weeks. Single cell clones were then picked into wells of a Matrigel-coated 96-well plate and subjected to PCR genotyping after 4 to 7 days.

Cardiac progenitor induction via modulation of canonical Wnt signalling

As described previously, hPSCs were singularized with Accutase (Innovative Cell Technologies) at 37 °C for 5 min once they achieved confluence and then seeded at 100,000–250,000 cells per cm2 in mTeSR1 or E8 supplemented with 5 μM ROCK inhibitor Y-27632 (Selleckchem) (day −3) for 24 h14,21. Cells were then cultured in mTeSR1 or E8, which was changed daily. At day 0, cells were treated with 6 μM CHIR99021 (Selleckchem) for 24 h in RPMI medium, followed by a change with RPMI medium at day 1. 2.5–5 μM IWP2 (Tocris) was added at day 3 and removed during the medium change at day 5. Cardiac progenitor cells can also be efficiently generated in RPMI/B27 medium using our previous GiWi protocol11.

Epicardial cell generation via activation of canonical Wnt signalling

At day 6, cardiac progenitor cells were singularized with Accutase at 37 °C for 5 min and then seeded onto a gelatin- or SyntheMax-coated cell culture plate at 20,000–80,000 cells per cm2 (or a 1:3–1:12 split) in LaSR basal medium (advanced DMEM/F12 with 100 μg ml−1 ascorbic acid) or RPMI/Vc/Ins medium (100 μg ml−1 ascorbic acid and 1 μg ml−1 human recombinant or bovine insulin (Sigma)) with 5 μM ROCK inhibitor Y-27632 for 24 h. The addition of 1% human recombinant albumin (Sigma) or fetal bovine serum (FBS) can improve cell attachment and survival, but they are not required. At day 7, cells were treated with 1–9 μM CHIR for 2 days in LaSR basal medium or RPMI/insulin/Vc medium. CHIR-containing medium was aspirated on day 9 and cells were cultured in LaSR basal medium or RPMI/insulin/Vc medium without CHIR99021 for 3–5 additional days, as explained previously14.

Long-term maintenance of hPSC-derived epicardial cells

To expand the epicardial cells, confluent cells on day 12 of differentiation or day 4 after expansion were split 1:3 to 1:9 at a density of 0.03 to 0.09 million cells per cm2 using Versene (Life Technologies) or Accutase and routinely passaged onto gelatin-coated plates in LaSR basal medium or RPMI/Vc/Ins medium and 0.5 μM A83-01 (Stemgent) or 2 μM SB431542 (Stemgent) with medium changed daily until the cells reached confluence. Overnight treatment of 5 μM Y27632 and 1% human recombinant albumin on single cells during passage was used to improve cell attachment and survival, but they were not required once cells attached. H13 hESC-derived day 50 epicardial cells were subjected for karotype analysis at WiCell Research Institute.

Single-cell passage and EMT induction

Confluent WT1+ cells were singularized with Accutase at 37 °C for 5 min and then seeded onto a gelatin-coated cell culture plate at a density of 10,000 cells cm−2 in LaSR basal medium supplemented with 5 μM Y-27632 for 24 h. After 24 h, the medium was changed to LaSR basal medium and cells were treated with TGF-β1 or bFGF (R&D Systems) as indicated. Medium was changed every 3 days until analysis.

Cardiac-fibroblast-derived extracellular matrix (CF-ECM) to transfer epicardial cells in vivo to the infarcted heart

Immunodeficient mice were purchased from Harlan Laboratories and all procedures were carried out in accordance with protocols approved by the Institutional Animal Care and Use Committee. As described previously50, myocardial infarction was induced 48 h before the transplantation. CF-ECM scaffolds were generated according to a previous report50. Briefly, cardiac fibroblasts isolated from porcine hearts were plated at a density of 0.1–0.25 million cells per cm2 in high-glucose DMEM supplemented with 10% FBS and cultured at 37 °C and 5% CO2 for 10–14 days. The combined cardiac fibroblasts and extracellular matrix were removed from the culture dish by incubation with 2 mM EDTA solution at 37 °C after a 10- to 14-day culture. The resulting cell sheet was treated with molecular grade water followed by 0.15% peracetic acid (PAA buffer) for 24–48 h at 4 °C with constant agitation to denude the cells. The resulting matrix was then rinsed with sterile water followed by PBS several times. CF-ECM scaffold was then seeded with 1–5 million ES03 eGFP hPSC-derived epicardial cells and incubated for 3 h before transfer to the epicardial surface of the MI area. After transplantation, the chest was closed. After 2, 6 and 12 days, the mouse hearts were harvested and excised for histology.

Immunostaining analysis

As explained in a previous study13, cells were fixed with 4% paraformaldehyde for 15 min at room temperature and then stained with primary and secondary antibodies (Supplementary Table 6) in PBS plus 0.4% Triton X-100 and 5% non-fat dry milk (Bio-Rad). Nuclei were stained with Gold Anti-fade Reagent with DAPI (Invitrogen). An epifluorescence microscope (Leica DM IRB) with a QImaging Retiga 4000R camera was used for imaging analysis.

Flow-cytometry analysis

As described in a previous report14, cells were dissociated into single cells with Accutase for 10 min and then fixed with 1% paraformaldehyde for 20 min at room temperature and stained with primary and secondary antibodies (Supplementary Table 6) in PBS plus 0.1% Triton X-100 and 0.5% BSA. Data were collected on a FACS Caliber flow cytometer (Beckton Dickinson) and analysed using FlowJo. FACS gating was based on the corresponding isotype antibody control.

Genomic DNA extraction and genomic PCR

Quick Extract DNA Extraction Solution (Epicentre Cat. no. QE09050) was used to rapidly extract genomic DNA from hESCs according to the manufacturer’s instructions. Genomic PCR was carried out using GoTaq Green Master Mix (Promega Cat. no. M7123). PCR primer sequences are provided in Supplementary Table 7.

RT-PCR and quantitative RT-PCR

As explained previously14, total RNA was prepared with the RNeasy mini kit (QIAGEN) and treated with DNase (QIAGEN). 1 μg RNA was reverse transcribed into cDNA via Oligo (dT) with Superscript III Reverse Transcriptase (Invitrogen). Real-time quantitative PCR was done in triplicate with iQSYBR Green SuperMix (Bio-Rad). GAPDH was used as an endogenous housekeeping control. PCR primer sequences are provided in Supplementary Table 7.

RNA sequencing and data analysis

Total RNA of day 12, 19-9-7 iPSC-derived epicardial cells and day 48, 19-9-11, ES03, H9 hPSC-derived epicardial cells were prepared with the Direct-zol RNA MiniPrep Plus kit (Zymo Research) according to the manufacturer’s instructions. Human primary epicardial RNAs from four different donors were provided by our collaborator (AstraZeneca, Sweden). Samples were performed in IIIumina HiSeq 2500 by the Biotechnology Center at the University of Wisconsin-Madison. The resulting sequence reads were mapped to the human genome (hg19) using HISAT51, and the RefSeq transcript levels (RPKMs) were quantified using the python script rpkmforgenes.py52. Hierarchical clustering of whole transcripts were then plotted using GENE-E. Fastq files of hPSCs38,​39,​40, hPSC-derived endoderm38, ectoderm39, mesoderm40 and CMs41 were downloaded from GEO or ArrayExpress. In particular, the mesoderm files were obtained from GEO with accession numbers GSM1112833, GSM915324 and GSM915325. Principal component analysis (PCA) (excluding 19-9-7 iPSC-derived day 12 epicardial cells) was performed using PLS Toolbox 8.1 (Eigenvector Technologies). The whole transcripts were preprocessed using the auto-scaling method (subtracting the mean from the variables and dividing by the standard deviation) to study the variance. Pathway enrichment analysis was performed using GSEA software53 (excluding 19-9-7 iPSC-derived day 12 epicardial cells). The gene expression data for each cell type were compared with that of hPSCs and the significantly enriched pathways (P < 0.05) were considered for further analysis. MATLAB 2013a (Mathworks) and Microsoft Excel (2013) were used to identify the unique and common pathways in different cell types. The absolute value of normalized enrichment score (NES) of the top 50 significantly enriched pathways for each cell type (ranked by the absolute NES) were further used for hierarchical clustering using GENE-E. To further investigate the similarity and differences in the number of enriched pathways in donor-derived epicardial cells, hPSC-derived epicardial cells and cardiomyocytes, the top 150 significantly enriched pathways for each cell type were selected.

Western-blot analysis

As described in a previous report43, cells were lysed in M-PER Mammalian Protein Extraction Reagent (Pierce) in the presence of Halt Protease and Phosphatase Inhibitor Cocktail (Pierce). Proteins were then separated by 10% Tris-Glycine SDS/PAGE (Invitrogen) under denaturing conditions and transferred to a nitrocellulose membrane. After blocking with 5% non-fat milk in TBST, the membrane was incubated with primary antibody (Supplementary Table 6) overnight at 4 °C. The membrane was then washed, incubated with an anti-mouse/rabbit peroxidase-conjugated secondary antibody for 1 h at room temperature or overnight at 4 °C, and developed by SuperSignal chemiluminescence (Pierce).

Statistical analysis

Data are presented as mean ± standard error of the mean (s.e.m.). Statistical significance was determined by Student’s t-test (two-tail) between two groups, and three or more groups were analysed by one-way analysis of variance (ANOVA). P < 0.05 was considered statistically significant.

Data availability

The final processed data and raw fastq files were submitted to Gene Expression Omnibus (GEO) with the accession number GSE84085. Source data for some of the figures in this study are available in figshare with the identifier (ref. 54). The authors declare that all other data supporting the findings of this study are available within the paper and its Supplementary Information.

Additional information

How to cite this article: Bao, X. et al. Long-term self-renewing human epicardial cells generated from pluripotent stem cells under defined xeno-free conditions. Nat. Biomed. Eng. 1, 0003 (2016).


  1. 1.

    , , , & Embryonic heart progenitors and cardiogenesis. Cold Spring Harb. Perspect. Med. 3, a013847 (2013).

  2. 2.

    & Development and function of the epicardium. Adv. Dev. Biol. 18, 333–357 (2007).

  3. 3.

    et al. Origin of coronary endothelial cells from epicardial mesothelium in avian embryos. Int. J. Dev. Biol. 46, 1005–1013 (2002).

  4. 4.

    et al. Thymosin β4 induces adult epicardial progenitor mobilization and neovascularization. Nature 445, 177–182 (2007).

  5. 5.

    et al. Adult mouse epicardium modulates myocardial injury by secreting paracrine factors. J. Clin. Invest. 121, 1894–1904 (2011).

  6. 6.

    et al. Retinoic acid production by endocardium and epicardium is an injury response essential for zebrafish heart regeneration. Dev. Cell 20, 397–404 (2011).

  7. 7.

    et al. Epicardial progenitors contribute to the cardiomyocyte lineage in the developing heart. Nature 454, 109–113 (2008).

  8. 8.

    & Differentiation of embryonic stem cells to clinically relevant populations: lessons from embryonic development. Cell 132, 661–680 (2008).

  9. 9.

    et al. Stage-specific optimization of activin/nodal and BMP signaling promotes cardiac differentiation of mouse and human pluripotent stem cell lines. Cell Stem Cell 8, 228–240 (2011).

  10. 10.

    et al. Robust cardiomyocyte differentiation from human pluripotent stem cells via temporal modulation of canonical Wnt signaling. Proc. Natl Acad. Sci. USA 109, E1848–E1857 (2012).

  11. 11.

    et al. Directed cardiomyocyte differentiation from human pluripotent stem cells by modulating Wnt/β-catenin signaling under fully defined conditions. Nat. Protoc. 8, 162–175 (2013).

  12. 12.

    et al. A small molecule that promotes cardiac differentiation of human pluripotent stem cells under defined, cytokine- and xeno-free conditions. Cell Rep. 2, 1448–1460 (2012).

  13. 13.

    et al. Chemically-defined albumin-free differentiation of human pluripotent stem cells to endothelial progenitor cells. Stem Cell Res. 15, 122–129 (2015).

  14. 14.

    et al. Efficient differentiation of human pluripotent stem cells to endothelial progenitors via small-molecule activation of WNT signaling. Stem Cell Rep. 3, 804–816 (2014).

  15. 15.

    et al. Manipulation of a VEGF-Notch signaling circuit drives formation of functional vascular endothelial progenitors from human pluripotent stem cells. Cell Res. 24, 820–841 (2014).

  16. 16.

    et al. Derivation of smooth muscle cells with neural crest origin from human induced pluripotent stem cells. Cells Tissues Organs 195, 5–14 (2012).

  17. 17.

    , , , & Generation of human vascular smooth muscle subtypes provides insight into embryological origin–dependent disease susceptibility. Nat. Biotechnol. 30, 165–173 (2012).

  18. 18.

    et al. Generation of the epicardial lineage from human pluripotent stem cells. Nat. Biotechnol. 32, 1026–1035 (2014).

  19. 19.

    et al. Robust derivation of epicardium and its differentiated smooth muscle cell progeny from human pluripotent stem cells. Development 142, 1528–1541 (2015).

  20. 20.

    et al. Epicardial cells of human adults can undergo an epithelial-to-mesenchymal transition and obtain characteristics of smooth muscle cells in vitro. Stem Cells 25, 271–278 (2007).

  21. 21.

    et al. Chemically defined, albumin-free human cardiomyocyte generation. Nat. Methods 12, 595–596 (2015).

  22. 22.

    et al. Directed induction of anterior and posterior primitive streak by Wnt from embryonic stem cells cultured in a chemically defined serum-free medium. FASEB J. 23, 114–22 (2009).

  23. 23.

    , , , & Nkx2-5- and Isl1-expressing cardiac progenitors contribute to proepicardium. Biochem. Biophys. Res. Commun. 375, 450–453 (2008).

  24. 24.

    et al. FGF2 sustains NANOG and switches the outcome of BMP4-induced human embryonic stem cell differentiation. Cell Stem Cell 8, 326–334 (2011).

  25. 25.

    , , , & YAC complementation shows a requirement for Wt1 in the development of epicardium, adrenal gland and throughout nephrogenesis. Development 126, 1845–1857 (1999).

  26. 26.

    et al. Wt1 is required for cardiovascular progenitor cell formation through transcriptional control of Snail and E-cadherin. Nat. Genet. 42, 89–93 (2010).

  27. 27.

    et al. Genetic engineering of human pluripotent cells using TALE nucleases. Nat. Biotechnol. 29, 731–734 (2011).

  28. 28.

    et al. Engineering human stem cell lines with inducible gene knockout using CRISPR/Cas9. Cell Stem Cell 17, 233–244 (2015).

  29. 29.

    et al. Insulin-like growth factor promotes engraftment, differentiation, and functional improvement after transfer of embryonic stem cells for myocardial restoration. Stem Cells 22, 1239–1245 (2004).

  30. 30.

    et al. Insulin-like growth factor promotes cardiac lineage induction in vitro by selective expansion of early mesoderm. Stem Cells 32, 1493–1502 (2014).

  31. 31.

    et al. Ascorbic acid enhances the cardiac differentiation of induced pluripotent stem cells through promoting the proliferation of cardiac progenitor cells. Cell Res. 22, 219–236 (2012).

  32. 32.

    et al. Biphasic role for Wnt/β-catenin signaling in cardiac specification in zebrafish and embryonic stem cells. Proc. Natl Acad. Sci. USA 104, 9685–9690 (2007).

  33. 33.

    et al. MesP1 drives vertebrate cardiovascular differentiation through Dkk-1-mediated blockade of Wnt-signalling. Nat. Cell Biol. 10, 338–345 (2008).

  34. 34.

    , , & Characterization of epicardial-derived cardiac interstitial cells: differentiation and mobilization of heart fibroblast progenitors. PLoS ONE 8, e53694 (2013).

  35. 35.

    et al. Experimental studies on the spatiotemporal expression of WT1 and RALDH2 in the embryonic avian heart: a model for the regulation of myocardial and valvuloseptal development by epicardially derived cells (EPDCs). Dev. Biol. 247, 307–326 (2002).

  36. 36.

    et al. A small molecule inhibitor of SRC family kinases promotes simple epithelial differentiation of human pluripotent stem cells. PLoS ONE 8, e60016 (2013).

  37. 37.

    , & Isolation and culture of mouse proepicardium using serum-free conditions. Methods 66, 365–369 (2014).

  38. 38.

    et al. In vitro generation of human pluripotent stem cell derived lung organoids. eLife 4, e05098 (2015).

  39. 39.

    et al. Transcriptional profiling of ectoderm specification to keratinocyte fate in human embryonic stem cells. PLoS ONE 10, e0122493 (2015).

  40. 40.

    et al. Differentiation of human pluripotent stem cells to cells similar to cord-blood endothelial colony-forming cells. Nat. Biotechnol. 32, 1151–1157 (2014).

  41. 41.

    et al. Inhibition of β-catenin signaling respecifies anterior-like endothelium into beating human cardiomyocytes. Development 142, 3198–3209 (2015).

  42. 42.

    et al. Revealing new mouse epicardial cell markers through transcriptomics. PLoS ONE 5, e11429 (2010).

  43. 43.

    , , & Interrogating canonical Wnt signaling pathway in human pluripotent stem cell fate decisions using CRISPR-Cas9. Cell. Mol. Bioeng. 9, 325–334 (2016).

  44. 44.

    , & Multipotent progenitor cells in regenerative cardiovascular medicine. Pediatr. Cardiol. 30, 690–698 (2009).

  45. 45.

    et al. Preservation of left ventricular function and attenuation of remodeling after transplantation of human epicardium-derived cells into the infarcted mouse heart. Circulation 116, 917–927 (2007).

  46. 46.

    , , & Epicardial regeneration is guided by cardiac outflow tract and Hedgehog signalling. Nature 522, 226–230 (2015).

  47. 47.

    , , , & SB-431542 inhibition of scar formation after filtration surgery and its potential mechanism. Invest. Ophthalmol. Vis. Sci. 50, 1698–1706 (2009).

  48. 48.

    , , & Dkk1 and Dkk2 regulate epicardial specification during mouse heart development. Int. J. Cardiol. 150, 186–192 (2011).

  49. 49.

    , & Directed endothelial progenitor differentiation from human pluripotent stem cells via Wnt activation under defined conditions. Methods Mol. Biol. 1481, 183–196 (2016).

  50. 50.

    et al. Cardiac fibroblast-derived 3D extracellular matrix seeded with mesenchymal stem cells as a novel device to transfer cells to the ischemic myocardium. Cardiovasc. Eng. Technol. 5, 119–131 (2014).

  51. 51.

    , & HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015).

  52. 52.

    , , & An abundance of ubiquitously expressed genes revealed by tissue transcriptome sequence data. PLoS Comput. Biol. 5, e1000598 (2009).

  53. 53.

    et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).

  54. 54.

    et al. Dataset for Long-term self-renewing human epicardial cells generated from pluripotent stem cells under defined xeno-free conditions. figshare (2016).

Download references


We thank D.A. Roenneburg and X. Wang for their technical support. We also thank members of the Palecek group for critical discussion of the manuscript. This work was supported by NIH grant EB007534 (S.P.P.), NSF grant 1547225 (S.P.P.), and a fellowship from the University of Wisconsin Stem Cell and Regenerative Medicine Center (X.B.).

Author information


  1. Department of Chemical & Biological Engineering, University of Wisconsin, Madison, Wisconsin 53706, USA

    • Xiaoping Bao
    • , Xiaojun Lian
    • , Tongcheng Qian
    • , Vijesh J. Bhute
    • , Tianxiao Han
    • , Mengxuan Shi
    •  & Sean P. Palecek
  2. Departments of Biomedical Engineering, Biology and Huck Institutes of the Life Sciences, The Pennsylvania State University, University Park, Pennsylvania 16802, USA

    • Xiaojun Lian
  3. Department of Medicine, University of Wisconsin, Madison, Wisconsin 53706, USA

    • Timothy A. Hacker
    •  & Eric G. Schmuck
  4. Department of Cardiovascular and Metabolic Diseases Innovative Medicine Unit, AstraZeneca, Mölndal, 43183, Sweden

    • Lauren Drowley
    • , Alleyn T. Plowright
    •  & Qing-Dong Wang
  5. Department of Molecular Cell Biology, Leiden University Medical Center, 2333 ZC Leiden, the Netherlands

    • Marie-Jose Goumans


  1. Search for Xiaoping Bao in:

  2. Search for Xiaojun Lian in:

  3. Search for Timothy A. Hacker in:

  4. Search for Eric G. Schmuck in:

  5. Search for Tongcheng Qian in:

  6. Search for Vijesh J. Bhute in:

  7. Search for Tianxiao Han in:

  8. Search for Mengxuan Shi in:

  9. Search for Lauren Drowley in:

  10. Search for Alleyn T. Plowright in:

  11. Search for Qing-Dong Wang in:

  12. Search for Marie-Jose Goumans in:

  13. Search for Sean P. Palecek in:


X.B. and S.P.P. designed this study and prepared the manuscript. X.B. undertook experimentation and data analysis. X.L. contributed to the study design and assisted in writing the manuscript. T.A.H. and E.G.S. designed and performed the in vivo study. T.H., V.J.B., T.Q. and M.S. assisted in differentiation experiments and data analysis. L.D., A.T.P., Q.-D.W. and M.-J.G. isolated and provided the human primary donor samples for RNA-seq. All authors reviewed and approved the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Sean P. Palecek.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    Supplementary figures and tables, and movie legends.


  1. 1.

    Supplementary Movie 1

    Non-contracting hESC-derived pro-epicardial cells at day 12.

  2. 2.

    Supplementary Movie 2

    Spontaneously contracting hESC-derived cardiomyocytes at day 12.

  3. 3.

    Supplementary Movie 3

    Spontaneously contracting hESC-derived cardiomyocytes at day 12.

  4. 4.

    Supplementary Movie 4

    Non-contracting iPSC-derived pro-epicardial cells at day 12.