The origin of fat cells or adipocytes is a fundamental biological question with important ramifications for human health and disease [1]. Epicardial fat is associated with increased risk of cardiovascular diseases such as coronary atherosclerosis [2]. However, the origin of these epicardial fat cells remains largely unknown. Epicardial progenitors play a pivotal role in the developing heart, by secreting paracrine signals and by differentiating into fibroblasts, smooth muscle cells, and potentially endothelial cells and cardiomyocytes [3-6]. Some of these developmental programs are reactivated during postnatal heart injury [7, 8], and reactivated adult epicardial progenitors provide pro-angiogenic protection and undergo limited epithelial-to-mesenchymal transition in cardiac injury [7, 9]. Since epicardial fat is physically adjacent to epicardial cells, we hypothesized that some of these cells originate from the epicardium during development and in postnatal heart injury.

We first generated Wt1-CreER; Rosa26RFP/+ double transgenic mouse line. In the absence of tamoxifen, we found little detectable labeling of epicardial progenitors and their derivatives [7]. After tamoxifen induction, translocation of CreER into nucleus allowed removal of a loxP-flanked transcriptional stop cassette, thus leading to indelible RFP labeling of epicardial cells and their descendants, epicardium-derived cells (EPDCs, Supplementary information, Figure S1A). We injected tamoxifen at embryonic day 10.5 (E10. 5) to label epicardial progenitors and collected hearts during postnatal weeks 8 to 12 (P8w to P12w) to analyze their contribution to epicardial fat (Figure 1A). We found that fetal EPDCs contribute significantly to intramyocardial vessels in adult hearts (Supplementary information, Figure S1A). Interestingly, some EPDCs, particularly those in the atrioventricular groove, remained in an epicardial location and colocalized with fat tissue (Figure 1B). Co-immunostaining of RFP and adipocyte marker perilipin (PLIN) in tissue sections from four hearts (12 sections each) verified that among 3 560 PLIN+ fat cells counted, 23.4%±5.1% were RFP+ cells (Figure 1C). To further confirm this finding, we employed an ade-