Human pluripotent stem cell‐derived epicardial progenitors can differentiate to endocardial‐like endothelial cells

Abstract During heart development, epicardial progenitors contribute various cardiac lineages including smooth muscle cells, cardiac fibroblasts, and endothelial cells. However, their specific contribution to the human endothelium has not yet been resolved, at least in part due to the inability to expand and maintain human primary or pluripotent stem cell (hPSC)‐derived epicardial cells. Here we first generated CDH5‐2A‐eGFP knock‐in hPSC lines and differentiated them into self‐renewing WT1+ epicardial cells, which gave rise to endothelial cells upon VEGF treatment in vitro. In addition, we found that the percentage of endothelial cells correlated with WT1 expression in a WT1‐2A‐eGFP reporter line. The resulting endothelial cells displayed many endocardium‐like endothelial cell properties, including high expression levels of endocardial‐specific markers, nutrient transporters and well‐organized tight junctions. These findings suggest that human epicardial progenitors may have the capacity to form endocardial endothelium during development and have implications for heart regeneration and cardiac tissue engineering.


| I N T R O D U C T I O N
The epicardium is the outermost mesothelium layer of the heart that is essential for both heart development and cardiac remodeling. 1 During cardiogenesis, a subset of epicardial cells invades the underlying myocardium and contributes to various cardiac lineages, including cardiac fibroblasts and smooth muscle cells. 2 A recent report also suggests an epicardial origin for a subset of cardiomyocytes, 3 although the in vivo fate mapping studies used to draw this conclusion are susceptible to artifacts. More recently, human sinoatrial node cardiomyocytes have been derived from TBX181 progenitor cells that also contribute to epicardial cells, 4 but it remains unknown whether WT11/TBX181 epicardial cells can also contribute to cardiomyocyte populations. Similarly, the epicardial contribution to the developing cardiac endothelium remains controversial.
In vivo lineage tracing studies have shown that a subpopulation of coronary endothelial cells arise from the epicardium in the chicken, 5 while studies in mice failed to identify a significant epicardial contribution to endothelial cells via fate mapping using the well-known epicardial cell markers TBX18 and WT1. 3,6 Recently, Scleraxis (Scx) and Semaphorin 3D (Sema3D) were identified as markers of epicardial cells that contribute to both coronary vascular endothelium and cardiac endocardium. 7 Zhang et al. 8 identified natriuretic peptide receptor 3 (NPR3) as a specific endocardial marker and demonstrated their contribution of NPR3-expressing endocardial cells to coronary vessels. The expression of WT1 in developing human fetal hearts follows a pattern *These authors contributed equally to this study. starting at the epicardium and extending toward the lumen of the heart, and WT1 expression in endocardial cells nearly disappeared at week 20, suggesting WT11 epicardial cells as a potential cell origin of endocardial endothelial cells. 9 However, understanding of the developmental progression of human epicardial cells to endothelium and endocardium is still extremely limited, mainly due to ethical and logistical challenges of tracing cells in the developing human heart and the lack of an in vitro human model to study the epicardial-to-endothelial transition.
Over the past 3 years, multiple labs have developed robust protocols to generate epicardial-like cells from human pluripotent stem cells (hPSCs) by manipulating Wnt, bone morphogenetic protein and retinoic acid signaling pathways that are important for in vivo epicardium development. [10][11][12][13] While hPSC-derived epicardial cells from different protocols have the potential to differentiate into smooth muscle cells and cardiac fibroblasts both in vitro and in vivo, none of them have yet been shown to form endothelial cells so far. To develop an hPSC model to study epicardial cell differentiation to endothelial cells, we first generated CDH5-2A-eGFP knock-in hPSC lines and differentiated them into self-renewing WT11 epicardial cells which gave rise to endothelial cells upon VEGF treatment in vitro. We also showed that the purity of epicardial-derived endothelial cells is proportional to WT1 expression in a WT1-2A-eGFP reporter line. In addition, the resulting epicardialderived endothelial cells displayed many endocardium-like endothelial cell (EEC) properties, including high expression levels of specific endocardial markers, nutrient transporters and well-organized tight junctions.
These findings demonstrate that hPSC-derived epicardial cells have the capacity to differentiate to endocardial-like cells for potential applications in heart tissue engineering and suggest that human epicardial cells may contribute to endocardial cells during cardiac development.  Figure S1).

| VEGF signaling permits endothelial transition from hPSC-derived epicardial cells
We previously demonstrated that temporal modulation of canonical Wnt signaling was sufficient to generate self-renewing WT1 1 TBX181 epicardial cells from hPSCs. 10  Next, we tested different concentrations of VEGF in generating endothelial cells from epicardial cells, and found that 50 and 100 ng/ml VEGF significantly improved the epicardial-to-endothelial transition compared to the no-VEGF control (Supporting Information Figure S2a Figure S2c). This is consistent with previous avian studies demonstrating epicardial cells give rise to endothelial cells. [16][17][18]

| Endothelial cells from WT11 cells display endocardial properties
Previously, TGFb inhibition via A83-01 treatment was shown to promote proliferation of hPSC-derived endothelial cells. 19 To increase the  Figure   4a,b). In summary, A83-01 treatment increased the purity of endocardial-like endothelial cells from WT11 cells, and they displayed specific EEC markers. This is consistent with an in vivo report that epicardial cells can give rise to NFATc11 endocardial cells during mouse heart development. 7

| Characterization of hPSC-derived EECs
In order to further confirm the endocardial identity of these putative hPSC-derived EECs, we differentiated VE-cad-2A-eGFP knock-in cell lines into epicardial cells and then subjected these cells to sequential VEGF and A83-01 treatment as shown in Figure

| DISCUSSION
The endocardium is the innermost endothelial layer of the heart, serving as a BHB, an endothelial layer that regulates the ionic composition of the cardiac microenvironment via passive tight junctions and active transporter systems. The endocardium also modulates myocardium performance by releasing trophic factors in response to humoral and mechanical stimuli. [26][27][28] In addition to its signaling roles during heart development and regeneration, the endocardium has also been shown to contribute multi-lineage descendants to cardiac valves, septa, hematopoiesis, and coronary blood vessels. 29 The epicardial contribution to the endocardium in vivo in animal models is small likely due to the fact that endocardial tubes form before the epicardium during heart development, 7 which might also explain the low efficiency (2%) of epicardial-to-endocardial transition even with high VEGF in our hPSC model. Therefore, there may be other cardiac progenitors that contribute to the majority of the endocardium. Compared to the epicardium, mesodermal progenitor cells marked by Flk11 are better established as origins for endocardium in both chicken and mice embryos before primary heart field formation. [34][35][36][37] In addition, Isl11 second heart field progenitors were shown to give rise to the endocardium by re-expressing Flk1. 38,39 However, generation of human endocardium from either Flk11 or Isl11 progenitors have not yet been described. Based on cardiac developmental studies in animal models, we expect that multiple progenitor populations may contribute to endocardial formation in humans.  myocardium co-culture systems to study the paracrine or directcontact interactions as these cellular communications are crucial for normal heart development and function. Such a co-culture system might help to identify key regulators of cardiomyocyte maturation which would improve efforts to regenerate the damaged heart. However, to make these applications practical, it will necessary to increase the purity and yield of VE-cad1 cells from hPSC-derived epicardial cells and to develop strategies to expand hPSC-derived EECs. Enhancing WT1 expression levels in epicardial cells, use of a 3D culture system, addition of growth factors in addition to VEGF, and co-culture with physiologically relevant cardiac cells may improve EEC differentiation or self-renewal.
In summary, our results demonstrate that a subset of human endocardial endothelial cells arise from hPSC-derived WT11 epicardial cells in vitro (Supporting Information Figure S5), providing a more complete understanding of the epicardial progenitor populations that form the endothelium, and a method to produce human endocardial endothelial cells for both research and translational applications.

| Construction of donor plasmid and sgRNA
Dual Cas9 and sgRNA backbone was digested with BbsI restriction enzyme for rapid sgRNA cloning as previously described. 10 Two sgRNAs targeting near the CDH5 stop codon (1: TCAGCCAG-CATCTTAAACCTGGG and 2: TTTTTGGAGGCTGTGGTGCCTGG) with a G added at the beginning were used. To generate the CDH5-   Hierarchical clustering of whole transcripts was then plotted using GENE-E. PCA was performed using PLS Toolbox 8.1 (Eigenvector Technologies). The whole transcripts were preprocessed using an autoscaling method (subtracting the mean from the variables and dividing by the standard deviation) to study the variance. Pathway enrichment analysis was performed using GSEA software. 48 The gene expression data for each cell type were compared to expression data from undif-

CONFLICT OF INTERESTS
The authors declare no competing financial interests.

AUTHOR CONTRIBUTIONS
XB and SPP designed this study and prepared the manuscript. XB and VJB undertook experimentation and data analysis. TH, TQ, and XL contributed to study design and assisted in experiments and data analysis. All authors reviewed and approved the manuscript.