Through the analysis of an 18-staged cardiac single-cell RNA sequencing (scRNA-seq) dataset in CD1 mice, previously published by our team and deposited in GEO with the accession number GSE193346 (Feng et al., 2022), we identified a new epicardial cell gene named Krt19 (Figures 1A,B). We found that Krt19 is highly expressed in epicardial cells at all analyzed stages, ranging from embryonic (E) day 9.5 to postnatal (P) day 9 (Figure 1E). Additionally, Krt19 is also expressed in atrial and ventricular cardiomyocyte (Atrial_CM, Ven_CM) at early developmental stages (mainly before E14.5) (Figures 1C,G). To better understand Krt19’s expression pattern, we compared it with the expression pattern of the well-known epicardial gene Wt1. We found that Wt1 is highly expressed in epicardial cells at all stages (Figures 1B,E) and in atrial and ventricular cardiomyocytes at early stages. However, Wt1 is also expressed in vascular endothelial cell (Vas_EC) and fibroblast (Fb) at all the analyzed stages, while Krt19 is barely detected in these 2 cell types at most of the analyzed stages (Figures 1F,H). Moreover, we analyzed Krt19 expression in two human fetal heart scRNA-seq datasets and found that it had high expression in epicardial cell and moderate expression in CM (Asp et al., 2019; Cui et al., 2019) (Figures 1I,J). In summary, we have identified a novel epicardial cell gene, Krt19, which is highly expressed in epicardial cells and exhibits a differential expression pattern compared to Wt1.

Next, to understand the lineage development of Krt19 positive cells, we lineage traced them by breeding Krt19-CreER mice with Rosa26-mTmG mice and administering tamoxifen at different time points. Initially, we treated the mice with tamoxifen at E9.5 and E10.5 and analyzed their hearts at E14.5 (Figure 2A). We observed strong eGFP signals on the outer surface of the heart, indicating efficient labeling of epicardial cells by the mice. Additionally, we observed eGFP-positive cells inside the chamber and septum, which could be derived from the labeled epicardial cells or non-epicardial cells expressing Krt19. We then moved the analysis to a later time window by treating the mice with tamoxifen at E13.5 and E14.5 and analyzing the hearts at E15.5 (Figure 2B). Again, we observed strong eGFP signals in epicardial cells, but also some eGFP-positive cells inside the chamber. In contrast, when we analyzed Wt1-CreER; Rosa26-mTmG mice at the same time period using the same dose of tamoxifen, we observed strong eGFP signals at the chamber surface and inside the chambers (Figure 2C), which could represent epicardial cells and vascular endothelial cells, respectively. These results are largely consistent with their expression pattern identified in the scRNA-seq analysis.

Subsequently, we conducted a long-term lineage tracing experiment by treating the mice with tamoxifen at E11.5 and E12.5 and analyzed their hearts at E17.5. We observed clear eGFP signals in epicardial cells and many eGFP-positive cells inside the chamber (Figure 3A). Finally, we treated the mice with tamoxifen at P2 and analyzed their hearts at P6 to understand their performance at the neonatal stage (Figure 3B). We found that all eGFP signals were on the outer surface, indicating that only epicardial cells were labeled by the Krt19-CreER; Rosa-mTmG mouse line during this time window. Finally, we quantified the percentages of eGFP-labeled cells within the outer layer of epicardial cells. We observed that about 3 percent of cells at the embryonic stage and 9 percent of cells at the neonatal stage were labeled (Figure 3C). However, please note that the labeling efficiency may vary under different doses of tamoxifen treatments.

In summary, the lineage tracing results suggest that the Krt19-CreER mouse line labels epicardial cells at the embryonic stage and displays specificity for epicardial cells at the neonatal stage. This indicates its potential utility in studying the function of epicardial cells and gene regulations within this cell type.

To investigate the function of epicardial cells, we crossed Krt19-CreER mice with Rosa26-DTA mice for ablation purposes. Tamoxifen was administered to pregnant dams at E13.5 and E14.5, and embryos were collected at E17.5 (Figure 4A). Strikingly, we observed that the ablated embryos, as well as their hearts, were noticeably smaller compared to the controls (Figures 4B,C). These results suggest that Krt19-positive cells are essential for embryo and heart development.

To gain further insights into the molecular defects in the ablated hearts, we conducted immunofluorescence analysis. Firstly, we analyzed the expression of the epicardial cell marker gene Aldh1a2. We observed strong fluorescence signals on the outer surface of the chambers in control hearts; however, in the ablation hearts, we found that the signal was largely eliminated (Figure 5A). This result indicates efficient epicardial cell ablation in the Krt19-CreER; Rosa-DTA mice. Furthermore, we stained the hearts with antibodies against PECAM1 for the endothelial cell lineage and MF20 for the cardiomyocyte lineage. Interestingly, we found that the ablated hearts have an obviously hyper-trabeculated myocardium compared to the control hearts (Figure 5A). These results suggest that the Krt19-CreER labeled epicardium is important for embryonic heart development, likely by regulating the growth of compact myocardium.

Lastly, we crossed Krt19-CreER mice with Rosa-DTA mice to investigate the function of the epicardium in neonatal mouse heart development. Newborn mice were treated with tamoxifen from P1 to P4, and their hearts were analyzed at P25. We observed that the ablated animal and its heart was noticeably smaller than its controls. Given that we have only successfully retrieved one ablated mouse after multiple breeding, these results can only imply that Krt19-positive epicardial cells are likely important for neonatal heart growth (Supplementary Figure S2). Next, we conducted similar experiments with Wt1-CreER; Rosa-DTA mice, treating them with tamoxifen at P1. However, we found that all the ablated mice died at P4. To better understand the causes of their death, we sacrificed them at P3 for detailed analysis (Figure 6). After inspecting multiple tissues, including the heart, liver, lung, kidney, and spleen from both control and ablated mice, we found that the ablated mice likely died due to kidney hemorrhage or spleen defects, which appeared wider and shorter than those in control mice (Figure 6). These results suggest that utilizing the Wt1-CreER mouse line to study epicardium function during neonatal heart development is challenging, while the Krt19-CreER mouse line emerges as a potential ideal candidate.

The animal experiments have been approved by the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC). The transgenic mice, including Krt19-CreERT2 (Strain #:026925) (Means et al., 2008), Wt1-CreERT2 (Strain #:010912) (Zhou et al., 2008), Rosa26-mTmG (Strain #:007676) (Muzumdar et al., 2007), and ROSA26-eGFP-DTA (Strain #:032087) (Ivanova et al., 2005) were ordered from the Jackson Laboratory.

To induce Cre activity, the pregnant mice were given the default dosage of 200 µg of tamoxifen per gram of body weight (200 μg/g) through oral gavage and the neonatal mice were given 10 μg/g of tamoxifen by direct injection into their stomach (Hortells et al., 2020). The pregnant and neonatal mice were euthanized using CO2 and decapitation-based methods, respectively. Following the standard procedure described previously, the mouse hearts were isolated and fixed at 4% paraformaldehyde for immunofluorescence staining.

The staining was performed following a standard procedure. Briefly, mouse hearts were fixed in 4% PFA overnight, embedded in OCT, and sectioned at 10 µm. After a brief wash in PBS to remove the OCT, samples were blocked for 1 h in blocking buffer (10% goat serum, 1% BSA, 0.1% Tween 20) and then incubated with primary antibodies in the primary antibody buffer (1% BSA in PBST) at 4°C overnight. On the second day, the samples were stained at room temperature with fluorophore-conjugated secondary antibodies in blocking buffer for 1 h. Finally, the samples were stained with DAPI, mounted with fluoromount-g, and imaged with a confocal microscope. The antibodies used in the study include anti-CD31 (BD, #550274), anti-Aldh1a2 (Sigma, #HPA010022), and MF20 (DSHB, #MF20). The outer layer of cells in Figure 3C was counted based on the DAPI staining signal.