PW1⁺ cells give rise to cardiac adipocytes during development and myofibroblasts following injury

Abstract

Background

Pw1, a maternally imprinted gene, is expressed in various stem cell populations, underscoring its crucial roles in tissue development, maintenance, and regeneration. While our recent work has revealed its regulatory function in cardiac fibrosis following ischemic injury, whether PW1⁺ cells exhibit stem cell properties in the heart remains unclear.

Methods

We utilized genetic lineage tracing with Pw1^pCreER/m+;R26-tdT and Pw1^{p2A−CreER/m+};R26-tdT mouse models. The adipogenic and fibrogenic potential of PW1⁺ cells was assessed in vitro using sorted tdTomato⁺ cells subjected to differentiation assays. In vivo fate mapping was performed during postnatal development and after myocardial infarction (MI) induced by permanent coronary artery ligation. Single-cell RNA sequencing data from adult mouse hearts were reanalyzed to characterize Pw1 expression across cardiac cell types.

Results

PW1 was expressed in multiple cardiac cell types–such as mesenchymal cells, epicardial cells, and endothelial cells-during early postnatal development. Isolated PW1⁺ cells from infant hearts differentiated into adipocytes in vitro, and lineage tracing experiments confirmed their significant contribution to cardiac adipocyte formation throughout postnatal development. Single-cell transcriptomic analysis further revealed predominant Pw1 expression in adult cardiac fibroblasts. In vitro, adult PW1⁺ cells differentiated into myofibroblasts upon TGFb1 stimulation and showed upregulated expression of extracellular matrix genes and TGFb receptors. After MI, PW1⁺ cells preferentially expanded and contributed to the myofibroblast population within the infarcted region.

Conclusion

This study reveals that PW1⁺ cells serve as a progenitor population capable of generating cardiac adipocytes during postnatal development and contributing to myofibroblast formation after MI. These findings offer new insights into cardiac adipogenesis and fibrogenesis, providing a potential foundation for future therapies aimed at mitigating pathological adipose accumulation and fibrosis in the heart.

Supplementary Information

The online version contains supplementary material available at10.1186/s13287-026-05035-z.

Background

Various cardiac mesenchymal cell (MC) populations, such as PDGFRa⁺ fibroblasts, have been identified as stem cells contributing to postnatal heart development and responding to heart injuries. Specifically, Sox9⁺, PDGFRa⁺ and PDGFRb⁺ cardiac MCs have been shown to be sources of cardiac adipocytes (CAs) during postnatal heart development [1]. Additionally, these three MC populations also contribute to CAs in a mouse model of arrhythmogenic cardiomyopathy [2]. In healthy individuals, cardiac adipose tissues may play protective roles in regulating cardiovascular functions and energy partitioning. However, the ectopic deposition of cardiac adipose tissues turns them into a deleterious entity with both local and systemic implications for cardiovascular disease development [3]. Therefore, uncovering the cellular origins of CAs may have significant implication for the treatment of fat-associated cardiovascular diseases.

PW1, also known as paternally expressed gene 3 (PEG3), is a maternally imprinted gene specific to mammals. Although PW1 is expressed at low levels across most tissues, it is notably abundant in skeletal muscle, the central nervous system, and the pituitary gland [4–6]. PW1 has also been reported as a pan-stem cell marker in a wide array of adult tissues, including the intestine, blood, testis, central nervous system, bone, skeletal muscle, and skin [7–11], suggesting its critical role in the development, maintenance, and regeneration of specialized tissues. PW1 expression in the cardiovascular system has also been documented. For instance, PW1 marks a population of resident smooth muscle progenitor cells playing essential roles in pulmonary hypertension-associated vascular remodeling [10].

In recent studies using two distinct knockin mouse stains, Pw1^{CreER-2 A-eGFP} (Pw1^CreER hereafter) and Pw1^2A-CreER, it was shown that only the paternal allele of Pw1 is active in adult normal mouse hearts [12]. PW1 is expressed in a small proportion of cardiac MCs in ventricles of adult mice [12]. Interestingly, myocardial infarction (MI) induces a reduction in DNA methylation at the imprinting control region of the Pw1 gene, triggering a switch from monoallelic to biallelic expression of Pw1 in PDGFRa⁺ fibroblasts within the injured regions [12]. Elevated PW1 expression in activated fibroblasts (myofibroblasts) promotes extracellular matrix (ECM) production by enhancing purine biosynthesis. Accordingly, genetic deletion of Pw1 attenuates cardiac fibrosis and improves heart function post-MI [12]. These findings advance our understanding of the function of PW1 in MCs in regulating cardiac fibrosis post-MI. However, whether PW1⁺ MCs exhibit stem cell properties in postnatal heart remain unclear.

In this study, we demonstrate that PW1-expressing MCs give rise to CAs in postnatal heart development and contribute significantly to myofibroblasts within injured areas of adult hearts following MI. These findings suggest that PW1⁺ cells serve as a progenitor cell population involved in heart development and response to cardiac injury. Our findings provide valuable insights into the processes of adipogenesis and fibrogenesis within the heart, thereby informing the development of novel therapeutic strategies for cardiac diseases.

Materials and methods

Mice

All experiments were approved by the Institutional Animal Care and Use Committee of ShanghaiTech University and complied with ARRIVE guidelines 2.0. Animals were randomly assigned to different groups. Animal experiments were performed in both sexes, and similar results were obtained without sex bias. Samples were uniformly processed before treatment under controlled conditions, with all assessments performed using multiple independent replicates. Sample sizes were not predetermined statistically but were set based on field norms and the numbers needed to attain statistical significance. Investigators conducting surgery or evaluation of adipogenesis were blinded to the group assignments. Mice were housed in a 12-h light/dark cycle environment and maintained on a C57BL6/ICR background. The Pw1^CreER and Pw1^2A−CreER mouse lines were characterized in our previous study [12]. The R26-tdT mouse line was kindly provided by the Shanghai Model Organisms Center. Mice were euthanized by gradual exposure to carbon dioxide followed by cervical dislocation at the indicated time points for tissue collection.

Tamoxifen treatment

Tamoxifen was dissolved in corn oil to achieve a concentration of 20 mg/mL. The mice received intragastric administration of tamoxifen at a dose of 0.1–0.2 mg per gram of body weight.

MI surgery

Anesthesia was induced using 2% isoflurane to facilitate tracheal intubation, and the mice were maintained on a 37 °C heating pad to ensure stable body temperature. Once reaching the surgical anesthesia plane, isoflurane concentration was reduced to 1.5%. A vertical incision of approximately 1.5 cm was made along the sternum, followed by careful blunt dissection to expose the chest cavity while avoiding damage to blood vessels. The left anterior descending coronary artery was permanently ligated using an 8 − 0 suture. Following the procedure, incisions were closed with 6 − 0 sutures, and the mice were returned to their cages for recovery.

Immunofluorescence staining

Tissues were collected and fixed in ice-cold 4% paraformaldehyde for 1–2 h. Following three washes in PBS, the hearts were dehydrated in 30% sucrose overnight. The hearts were then embedded in OCT compound and stored at -80℃. Samples were sectioned at a thickness of 10 μm. For immunofluorescence staining, the heart sections were air-dried for 30 min and washed three times in PBS. The sections were then blocked with 5% donkey serum containing 0.2% Triton X-100 for 30 min at room temperature. Sections were incubated with primary antibodies overnight at 4 °C. The following day, sections were washed in PBS and incubated with Alexa fluor-conjugated secondary antibodies and DAPI for 1 h at room temperature. After three additional washes in PBS, sections were mounted and imaged using a confocal microscope (Nikon A1R, Leica TCS SP8 STED 3X, or Zeiss 710). The primary antibodies used included: anti-tdTomato (Rockland, 600-401-379), anti-PW1 (Invitrogen, PA5-99683), anti-GFP (Abcam, ab6662), anti-PDGFRa (R&D system, AF1062), anti-PDGFRb (eBioscience, 14-1402-82), anti-PLIN (Abcam, ab61682), anti-FSP1 (Dako, A5114), anti-aSMA (Sigma, A5228), anti-NG2 (Millipore, Ab5320), anti-CDH5 (R&D, AF1002), anti-CD31 (BD, 553370), anti-WT1 (Abcam, Ab89901), and anti-SM22 (Abcam, ab14106).

EdU administration and staining

Cell proliferation was assessed using a 5-ethynyl-2ʹ-deoxyuridine (EdU) assay. Five hours prior to sacrifice, mice received an intraperitoneal injection of EdU at a dose of 0.02 mg per gram of body weight. Tissue samples were subsequently fixed and sectioned. Following immunofluorescence staining, incorporated EdU was detected using a Click-iT reaction cocktail (Click-iT EdU Alexa Fluor 555 Imaging Kit, Thermo Fisher Scientific) following the manufacturer’s protocol.

Flow cytometric analysis

Heart tissues were finely minced and subjected to enzymatic digestion on a shaker at 37 °C using a lysis buffer containing 1 U/mL type 2 collagenase (Yeasen, 40508ES60), 0.05% trypsin (Invitrogen, 15090046), and 25 µg/mL DNase I (Roche, 4716728001). After 5 to 10 min of digestion, the supernatant was carefully aspirated and transferred to a fresh 15 mL conical centrifuge tube. The digestion process continued with the addition of fresh lysis buffer until complete tissue dissociation was achieved. Following centrifugation at 300 g for 5 min to pellet cardiomyocytes, the supernatant was carefully transferred to a new collection tube and centrifuged at 500 g for 5 min to pellet non-cardiomyocyte cells.

For quantification of the percentage of tdTomato⁺ myofibroblasts, these non-cardiomyocyte cells were resuspended in PBS and incubated with anti-PDGFRα-APC (Thermo Fisher Scientific, 17-1401-81) and anti-tdTomato (Rockland, 600-401-379) at 4 °C for 1 h. After three washes in PBS, the cells were incubated with a secondary antibody at room temperature for an additional hour. Following incubation, cells were washed and analyzed using a LSRFortessa flow cytometer (BD Biosciences). To isolate tdTomato⁺ cells from Pw1^pCreER/m+;R26-tdT mice, the non-cardiomyocyte cells were harvested and resuspended in 1 mL of PBS. The tdTomato⁺ cells were sorted using a FACSAria Fusion Cell Sorter (SORP ARIA Fusion).

Adipogenesis assay

Sorted cells were plated in 24-well plates and initially maintained in standard cell culture medium for 2 days to allow for stabilization. Following this period, adipogenic differentiation was initiated according to established protocols [2]. For this purpose, cells were exposed to an adipogenic differentiation medium consisting of a basal medium (Dulbecco’s Modified Eagle Medium with low glucose, supplemented with 10% Fetal Bovine Serum and 1% penicillin-streptomycin antibiotics) enriched with 5 µg/mL insulin, 1 µM dexamethasone, 0.5 mM isobutylmethylxanthine, and 1 µM rosiglitazone. Cells cultured exclusively in the basal medium served as the control group. The differentiation medium was replenished every 3 days. After 10 days of culture, cells were fixed with 4% paraformaldehyde for 1 h at room temperature and then subjected to Oil Red O staining and Perilipin immunostaining to evaluate adipocyte formation.

Oil Red O staining

Slides were initially immersed in a 60% isopropanol solution for 10 min. Subsequently, they were incubated in the freshly prepared Oil Red O working solution, which consisted of 0.5 g of Oil Red O powder dissolved in 60% isopropanol, for 15 min. To minimize background staining, the slides were briefly rinsed in 60% isopropanol for 20 s. Following this, the slides were washed in PBS for 5 min, stained with hematoxylin solution for 4 min, and then treated with a clarifying solution of 1% hydrochloric acid in 70% ethanol for 1 min. This step was followed by treatment with a 1% ammonia solution for 1 min. After thorough washing, the slides were mounted using preheated, dissolved glycerin gel. Images were captured using a stereomicroscope (Olympus CX23).

Myofibroblast genesis assay

Sorted tdTomato⁺ cells were cultured in 24-well plates with standard cell culture medium for an initial 2-day period. Following this, the cells were induced to differentiate into myofibroblasts by exposure to a differentiation medium, which was formulated by adding 10 ng/mL TGFb1 to the basal medium. Cells maintained in basal medium served as the control group. The differentiation medium was replenished every 2 days to ensure consistent differentiation conditions. After a total of 7 days in culture, the cells were collected for quantitative real-time PCR analysis or immunostaining.

RNA extraction and quantitative real-time PCR

Total RNA was extracted from hearts of wild-type mice at various developmental stages or from cells harvested following the myofibroblast genesis assay using Magzol reagent (Magen, R480102). The extracted RNA was stored at − 80 °C. RNA was used to synthesize complementary DNA (cDNA) using the HiScript II Q RT SuperMix (Vazyme, R223-01) according to the manufacturer’s protocol. Quantitative real-time PCR experiments were conducted using the SYBR-Green qPCR Master Mix (Vazyme, Q711-02) and analyzed on an AB QuantStudio 7 Real-Time PCR System (Life Technologies). The following primers were used: Gapdh: 5ʹ-TTGTCTCCTGCGACTTCAAC-3ʹ and 5ʹ-GTCATACCAGGAAATGAGCTTG-3ʹ; Pw1 set1 : 5ʹ-AGGATGCCGAGTCATACCAG-3ʹ and 5ʹ-TGGGTAGGCAGTTCTCTTGG-3ʹ; Pw1 set2 : 5ʹ-ACCACTCCGTTGGAGAGTTT-3ʹ and 5ʹ-CGTCGTCACTGTTCATGTCA-3ʹ; aSMA: 5ʹ-CTCCCTGGAGAAGAGCTACG-3ʹ and 5ʹ-CGCTGACTCCATCCCAATGA-3ʹ; Col1a1: 5ʹ-GAGAGGTGAACAAGGTCCCG-3ʹ and 5ʹ-AAACCTCTCTCGCCTCTTGC-3ʹ; Col1a2: 5ʹ-CCAAGGGTGCTACTGGACTC-3ʹ and 5ʹ-GCTCACCCTTGTTACCGGAT-3ʹ; Col3a1: 5ʹ-TGACTGTCCCACGTAAGCAC-3ʹ and 5ʹ-GAGGGCCATAGCTGAACTGA-3ʹ; Col4a1: 5ʹ-GGCCCTTCATTAGCAGGTGT-3ʹ and 5ʹ-GTGAGGACCAACCGTTAGGG-3ʹ; Col6a1: 5ʹ-GCCGGTGATGAAGGAAATGC-3ʹ and 5ʹ-GTCTCCCCTTGGTCCTCTCA-3ʹ; Fn1: 5ʹ-CGGTGGCTGTCAGTCAAAG-3ʹ and 5ʹ-AAACCTCGGCTTCCTCCAT AA-3ʹ; Postn: 5ʹ-CTGCTTCAGGGAGACACACC-3ʹ and 5ʹ-TCTGAAGTCAGTGTGGTGGC-3ʹ; Lama1: 5ʹ-GGACGGGAATTCCGTTAGGG-3ʹ and 5ʹ-CAGGTCCAAGGACTGCACTT-3ʹ; Tgfbr1: 5ʹ-TGCTCCAAACCACAGAGTAGGC-3ʹ and 5ʹ-CCCAGAACACTAAGCCCATTGC-3ʹ; Tgfbr2: 5ʹ-CCTACTCTGTCTGTGGATGACC-3ʹ and 5ʹ-GACATCCGTCTGCTTGAACGAC-3ʹ; Tgfbr3: 5ʹ-TCTCCGCTGAATGGCTGTGGTA-3ʹ and 5ʹ-CCGACTCCAAATCTTCGTAGCC-3ʹ; Adipoq: 5ʹ-AGATGGCACTCCTGGAGAGAAG-3ʹ and 5ʹ-ACATAAGCGGCTTCTCCAGGCT-3ʹ; Pparg: 5ʹ-GTACTGTCGGTTTCAGAAGTGCC-3ʹ and 5ʹ-ATCTCCGCCAACAGCTTCTCCT-3ʹ. All gene expression data were normalized to the endogenous Gapdh as standard control for each sample.

Single-cell transcriptome dataset preprocessing

For the single-cell RNA sequencing (scRNA-seq) re-analysis, we analyzed the publicly available dataset GSE198833 (GEO). Our analysis specifically focused on the quality-controlled sample GSM5957560, which comprises 6,261 cells encompassing all non-cardiomyocyte populations from a mouse heart. The initial quality-control filtering steps were performed as described in the original study [13]. Data preprocessing and single-cell transcriptome analysis were conducted using R (v4.2.3) along with the ‘Seurat’ package (v5.0.1).

First, the ‘CreateSeuratObject’ function was employed to generate a Seurat object from the Unique Molecular Identifier (UMI) count matrix. Minimum criteria for cell inclusion were set to a count of 3 cells with a corresponding minimum feature count of 50. The dataset was filtered based on the following criteria: a gene count per cell ranging from 200 to 6500, a mitochondrial gene percentage below 25%, and exclusion of any cells expressing the HBB gene. Subsequently, logarithmic normalization of the data was performed using the ‘NormalizeData’ function. All subsequent analyses were based on this preprocessed dataset.

Cell cluster identification and annotation

To identify variable features, we utilized the ‘FindVariableFeatures’ function in R, selecting 2000 variable features with the ‘vst’ method. Based on these features, principal component analysis (PCA) was performed using the ‘RunPCA’ function to achieve linear dimensionality reduction. To delineate distinct cell clusters, The ‘FindClusters’ function was employed with a resolution parameter set at 1. Subsequently, the Uniform Manifold Approximation and Projection (UMAP) algorithm were applied for nonlinear dimensionality reduction to further investigate the data structure. For identifying conserved cell markers, the ‘FindAllMarkers’ function in R was utilized. Additionally, the ‘FeaturePlot’ function was used to visualize genes of interest on UMAP plots, effectively showcasing their expression across different clusters.

Statistical analysis

All data were obtained from at least three independent experiments and are presented as mean ± standard error of the mean (SEM). Normality was assessed using the Shapiro-Wilk test; a p value ≥ 0.05 was considered consistent with a normal distribution. For comparisons between two groups, two-tailed unpaired Student’s t-tests were used. For comparisons among more than two groups with one independent variable, one-way ANOVA followed by Tukey’s post-hoc test was applied. Statistical significance was defined as p < 0.05. Where one-way ANOVA was used, a post-hoc power analysis was carried out with G*Power (v3.1.9.7, α = 0.05). All statistical analyses were performed in GraphPad Prism (v10.3.1). Manders’ correlation coefficients (M1 and M2) were calculated using 63× magnification to quantify the colocalization between signals. M1 represents the fraction of the green signal overlapping with the red signal, and M2 represents the fraction of the red signal overlapping with the green signal. Analyses were conducted using the Coloc2 plugin in Image J. Investigators performing the statistical analyses were blinded to group assignments throughout the study.

Results

PW1 expression in the infant mouse heart

CAs, which comprise both intramyocardial and subepicardial adipocytes, begin to emerge between postnatal weeks 3 to 4 (P3W to P4W) in mouse hearts [14]. To investigate the contribution of PW1⁺ cells to the formation of CAs in the postnatal heart, we first assessed the expression of PW1 in mouse hearts prior to P3W. Quantitative real-time PCR analysis demonstrated Pw1 expression in hearts at postnatal day 1 (P1D) and P2W, with a noted decrease in expression levels as the heart matured (Fig. 1a).

Fig. 1.

Expression of Pw1 in cardiac MCs during early postnatal development. a Relative Pw1 mRNA levels in mouse hearts at different developmental stages, measured using two independent primer sets. n = 3 independent biological replicates per group. Statistical significance was determined by one-way ANOVA followed by Tukey post hoc tests (primer set 1, F(2, 6) = 78.32; primer set 2, F(2, 6) = 58.21). b Schematic of experimental strategy to identify PW1-expressing cells in pup hearts. c, d Immunostaining for tdTomato with PDGFRb (c) or PDGFRa (d) in heart sections from Pw1^pCreER/m+; R26-tdT mice at P3W. e, f Immunostaining for tdTomato with PDGFRb (e) or PDGFRa (f) in heart sections from Pw1^{p2A−CreER/m+};R26-tdT mice at P3W. Yellow arrows in c and e mark tdTomato⁺PDGFRb⁺ cells; white arrows in d and f indicate tdTomato⁺PDGFRa⁺ cells. Scale bars, 100 μm

To trace the fate of PW1⁺ cells in the heart, we introduced previously established knockin mouse lines, Pw1^CreER and Pw1^2A−CreER [12]. Since Pw1 is a maternal imprinted gene, with only the paternal allele being active in organs from normal adult mice, we employed a breeding strategy where a male heterozygous Pw1^CreER/+ mouse was crossed with a female Rosa26-loxp-stop-loxp-tdTomato (R26-tdT) reporter mouse to produce Pw1^pCreER/m+;R26-tdT mice [12]. A similar approach was used to generate Pw1^{p2A−CreER/m+};R26-tdT mice [12].

We first assessed the fidelity of the Pw1^pCreER/m+;R26-tdT model in recapitulating endogenous PW1 expression using pulse-chase experiments. Tamoxifen was administered at P1D, and hearts were harvested at P3D (Fig. S1a). Co-staining for tdTomato and GFP, a surrogate for endogenous PW1, revealed strong tdTomato expression in GFP⁺ cells (Fig. S1b). Direct immunostaining for tdTomato and PW1 further confirmed that tdTomato marked the majority of PW1⁺ cells (87.29% ± 1.83%), and conversely, most tdTomato⁺ cells were positive for PW1 (70.44% ± 2.53%) (Fig. S1c, d). Similarly, in hearts of P2W Pw1^pCreER/m+;R26-tdT pups, most GFP⁺ cells were also tdTomato⁺ (Fig. S1e, f). Quantification of the direct PW1 and tdTomato co-staining showed that 85.97% ± 2.01% of PW1⁺ cells were tdTomato-labeled, while 69.28% ± 1.30% of tdTomato⁺ cells were positive for PW1 (Fig. S1g, h). No tdTomato signal was detected in tamoxifen-treated R26-tdT controls or in vehicle-treated Pw1^pCreER/m+;R26-tdT mice, confirming the absence of reporter leakage (Fig. S1i–n). Together, these data demonstrate efficient targeting of PW1⁺ cells by the Pw1^pCreER/m+;R26-tdT line during early postnatal stages.

Parallel pulse-chase experiments in Pw1^{p2A−CreER/m+};R26-tdT mice yielded comparable results. tdTomato labeled 86.77% ± 1.61% of PW1⁺ cells at P3D and 84.17% ± 2.07% at P2W (Fig. S2a–f). Conversely, 68.10% ± 1.78% of tdTomato⁺ cells at P3D and 68.91% ± 2.40% at P2W were positive for PW1 (Fig. S2a–f). No tdTomato signal was detected in in vehicle-treated Pw1^{p2A−CreER/m+};R26-tdT mice (Fig. S2g–i). These results demonstrate that Pw1^{p2A−CreER/m+};R26-tdT lines also effectively targets PW1⁺ cells in the infant mouse heart.

We next characterized the cell types targeted by these genetic lines in infant hearts. Mice received tamoxifen at P2W, and hearts were analyzed at P3W (Fig. 1b). In Pw1^pCreER/m+;R26-tdT hearts, tdTomato signal was clearly detected in a subset of PDGFRb⁺ and PDGFRa⁺ MCs (Fig. 1c, d). Although PDGFRb is expressed in some cardiac fibroblasts [1], it is conventionally considered a maker of vascular mural cells (including smooth muscle cells and pericytes). Co-staining for tdTomato and the mural-cell marker NG2 confirmed that tdTomato labels a subset of vascular mural cells (Fig. S3a). In addition, tdTomato also marked a small proportion of coronary vascular endothelial cells (cVECs) and endocardial ECs (Fig. S3a, b). Notably, a significant fraction of epicardial cells was also labeled by tdTomato (Fig. S3c). Together, these results show that in Pw1^pCreER/m+;R26-tdT hearts during early postnatal development, tdTomato targets cardiac MCs (fibroblasts and mural cells), epicardial cells, and a minor subset of ECs. Similar labeling patterns were observed in Pw1^{p2A−CreER/m+};R26-tdT hearts (Fig. 1e, f, and Fig. S4).

PW1⁺ cells contribute to CAs during postnatal development

To assess the adipogenic capacity of PW1⁺ cells, we isolated tdTomato⁺ cells from hearts of Pw1^pCreER/m+;R26-tdT mice at P3W (Fig. 2a) and cultured them in control or adipogenic differentiation medium. Oil Red O staining showed that tdTomato⁺ cells readily differentiated into adipocytes upon adipogenic stimulation, whereas tdTomato^– cells formed only sparse adipocytes under the same conditions (Fig. 2b, c). Immunostaining for the adipocyte marker Perilipin (PLIN) further confirmed that tdTomato⁺ cells robustly give rise to adipocytes in differentiation medium (Fig. 2d, e). Moreover, expression of the adipogenic key regulators Pparg and Adipoq was significantly upregulated in both tdTomato⁺ and tdTomato^–cells following adipogenic induction (Fig. 2f). These results demonstrate that PW1⁺ cardiac cells possess strong intrinsic adipogenic potential in vitro.

Fig. 2.

Adipogenic differentiation of PW1⁺ cells in vitro. a Isolation of tdTomato⁺ and tdTomato^– cells from Pw1^pCreER/m+;R26-tdT hearts. Tamoxifen was administered at P2W, and hearts were collected at P3W for FACS. b Oil Red O staining of cultured cells following adipogenic induction. Scale bars, 100 μm. c Quantification of the Oil Red O-positive area. n = 3 independent biological replicates per group. Difference was determined by one-way ANOVA followed by Tukey post hoc tests (F(3, 8) = 38.19). d Immunostaining for PLIN in cultured cells. Scale bars, 100 μm. Images in (b) and (d) are representative of three independent biological replicates. e Quantification of tdTomato⁺PLIN⁺ cells cultured under differentiation versus control conditions. n = 3 independent biological replicates per group; two-tailed Student’s t-test. f Fold changes expression of Pparg and Adipoq in cells cultured in differentiation medium versus control medium. n = 4 biological replicates per group; two-tailed Student’s t-test

To evaluate the contribution of PW1⁺ cells to CAs in vivo, tamoxifen was administered to Pw1^pCreER/m+;R26-tdT mice at P2W and hearts were collected at multiple developmental stages. PW1-derived (tdTomato⁺) cells displayed active proliferating at P3W but rarely proliferated by P8W, suggesting that PW1⁺ cells undergo rapid expansion during early postnatal stages (Fig. S5). Since CAs emerge between P3W and P4W [14], we assessed the contribution of PW1⁺ cells to CAs from P4W onward (Fig. 3a). Co-staining for tdTomato and PLIN at P4W revealed tdTomato-labelled adipocytes within both intramyocardial adipose tissues in the ventricles and subepicardial adipose tissues in the atrioventricular groove (Fig. 3b). Similar labeling patterns were also observed at P8W and P16W (Fig. 3c, d). Quantification showed that tdTomato labeled 73.15% ± 4.59%, 73.29% ± 4.30%, and 74.57% ± 2.42% of CAs at P4W, P8W, and P16W, respectively in Pw1^pCreER/m+;R26-tdT mice (Fig. 3e).

Fig. 3.

PW1⁺ cells give rise to CAs in postnatal heart. a Schematic of the genetic lineage tracing strategy used to track the fate of PW1⁺ cells during postnatal development in Pw1^pCreER/m+;R26-tdT mice. b, c, d Immunostaining for tdTomato and PLIN in heart sections from Pw1^pCreER/m+;R26-tdT mice at various developmental stages. Arrows indicate tdTomato⁺PLIN⁺ adipocytes. e Quantification of the percentage of CAs labeled by tdTomato. n = 6 biological replicates per group; both sexes included. Statistical significance was assessed by one-way ANOVA followed by Tukey post hoc tests (F(2, 15) = 0.04; achieved power = 0.81). Scale bars, 100 μm. LV, left ventricle; Atr, atrium; Epi, epicardium

Parallel lineage tracing experiments were performed in Pw1^p2A-CreER/m+;R26-tdT mice that received tamoxifen at P2W and were analyzed at P16W (Fig. S6a). Co-staining for tdTomato and PLIN demonstrated that 65.07% ± 2.65% of CAs were tdTomato⁺ (Fig. S6b–d). Together, these data demonstrate that PW1⁺ cells contribute substantially to the formation of CAs during postnatal heart development.

scRNA-seq reveals enriched Pw1 expression in fibroblasts of the adult mouse heart

Previously we showed that PW1 is expressed in PDGFRa⁺ and PDGFRb⁺ MCs in the normal adult mouse heart [12]. To further characterize its distribution across cardiac cell types, we re-analyzed a publicly available scRNA-seq dataset (GEO, GSE198833) comprising non-cardiomyocyte cells from adult mouse hearts [13]. Unsupervised clustering identified 22 distinct clusters (cluster 0–21) (Fig. S7a), which we annotated as nine major cell lineages using established marker genes: fibroblasts (Pdgfra, Col1a1, Col1a2), cVECs (Fabp4, Cdh5, Pecam), macrophages (Itgam, Cd68, Fcgr1), smooth muscle cells (Myh11, Tagln, Acta2), pericytes (Notch3, Cspg4, Pdgfrb), lymphatic ECs (Flt4, Lyve1, Prox1), schwan cells (Cnp, Kcna1, Plp1), endocardial ECs (Cdh11, Plvap, Npr3), and B cells (Cd79b, Ms4a1, Cd79a) (Fig. 4a–c, Table S1).

Fig. 4.

Single-cell transcriptomic analysis identifies enriched Pw1 expression in adult mouse cardiac fibroblasts. a UMAP visualization of major non-cardiomyocyte cell populations in adult (8-week-old) mouse hearts, characterized by unsupervised clustering. b Dot plot depicting expression levels of marker genes used to annotate each cell cluster. c UMAP feature plots showing the expression distribution of key lineage-specific markers across the identified populations. d, e Feature plot (d) and violin plot (e) illustrating Pw1 expression levels among the distinct cardiac cell populations

Pw1 expression was predominantly enriched in the fibroblast population (Fig. 4d, e), comprising clusters 0, 1, 3, 7, 8, 11, 14, and 21 (Table S1). These fibroblast subclusters consistently exhibited high Pw1 expression levels and a large proportion of Pw1-expressing cells (Fig. S7b, c). Low levels of Pw1 expression were also detected in other lineages, such as cVECs, endocardial ECs, and vascular mural cells (Fig. 4e).

PW1⁺ cells differentiate into myofibroblasts upon in vitro stimulation

It is well established that myofibroblasts in the injured hearts predominantly originate from tissue-resident fibroblasts [15–17]. Given that Pw1 is expressed in a subset of fibroblasts in the normal adult mouse heart, we asked whether PW1⁺ cells possess fibrogenic potential. To test this, we isolated tdTomato⁺ and tdTomato^– cells from the hearts of Pw1^pCreER/m+;R26-tdT mice that had been treated with tamoxifen 48 h before harvesting at P8W. These tdTomato⁺ cells were then stimulated with TGFb1 in culture. TGFb1 treatment induced strong expression of the myofibroblast marker SM22 in both tdTomato⁺ and tdTomato^– cells (Fig. 5a). Furthermore, the expression of several myofibroblast-associated genes, including those encoding αSMA and multiple ECM components, were significantly upregulated in TGFb1-treated tdTomato⁺ cells (Fig. 5b). These data indicate that PW1⁺ cells can differentiate into myofibroblasts upon appropriate stimulation in vitro, supporting their potential role in injury-induced cardiac fibrosis.

Fig. 5.

Induction of myofibroblast differentiation in PW1⁺ cells by TGFb1 stimulation. a Both tdTomato⁺ and tdTomato^– cells express the myofibroblast marker SM22 after 7 days of culture with TGFb1. Scale bars, 100 μm. Images are representative of three independent biological replicates. b Fold-change expression of aSMA and selected ECM genes in tdTomato⁺ cells treated with TGFb1 relative to untreated controls. n = 3 biological replicates, each with three technical replicates; two-tailed Student’s t-test. c, d Basal expression levels of ECM genes (c) and TGFb receptors (d) in tdTomato⁺ cells without TGFb1 treatment. n = 4 biological replicates, each with three technical replicates; two-tailed Student’s t-test

We next compared baseline ECM-related gene expression between tdTomato⁺ and tdTomato^– cells isolated from Pw1^pCreER/m+;R26-tdT hearts. tdTomato⁺ cells showed higher expression of several ECM genes, including Col1a1, Col1a2, Col4a1, and Col6a1 (Fig. 5c), suggesting that PW1⁺ cells have a greater intrinsic capacity for ECM production than their PW1– counterparts. We therefore analyzed the TGFb signaling pathway, which plays a central role in regulating ECM production in fibroblasts, and found that TGFb receptors–Tgfbr1, Tgfbr2, and Tgfbr3–were all expressed at higher levels in tdTomato⁺ cells (Fig. 5d). These results imply that PW1⁺ cells may exhibit increased sensitivity to TGFb and a heightened propensity to produce ECM or differentiate into myofibroblasts after injury.

PW1⁺ cells contribute to myofibroblasts in the adult heart after MI

To determine whether PW1⁺ cells contribute to myofibroblast formation after MI, we performed genetic lineage tracing using Pw1^pCreER/m+;R26-tdT and Pw1^p2A-CreER/m+;R26-tdT mice. We first evaluated the labeling efficiency in adult hearts and found that tdTomato marked 72.97% ± 2.78% of PW1⁺ cells in Pw1^pCreER/m+;R26-tdT and 75.11% ± 2.51% in Pw1^p2A-CreER/m+;R26-tdT mice (Fig. S8a–f). Conversely, a substantial proportion of tdTomato⁺ cells were PW1⁺ in each model (Pw1^pCreER/m+;R26-tdT: 66.84% ± 2.09%; Pw1^p2A-CreER/m+;R26-tdT: 65.90% ± 2.49%; Fig. S8c, f). No tdTomato signal was observed in tamoxifen-treated R26-tdT controls or in vehicle-treated Pw1^pCreER/m+;R26-tdT and Pw1^p2A-CreER/m+;R26-tdT mice, confirming the absence of reporter leakage (Fig. S8g–o).

Lineage tracing was then carried out in Pw1^pCreER/m+;R26-tdT mice. Tamoxifen was administered 7 days prior to MI, which was induced by permanent ligation of the left anterior descending coronary artery (Fig. 6a). Hearts were collected 3, 7, and 14 days post-MI (Fig. 6a). At 3 days, proliferating tdTomato⁺ cells were scarce in remote myocardium but readily detectable in injured regions (Fig. S9), indicating rapid proliferation of PW1⁺ cells after injury. By 7 and 14 days, whole-mount imaging showed increased tdTomato fluorescence in infarcted versus remote areas (Fig. 6b, d), and section staining confirmed robust tdTomato signal selectively in the infarct zone (Fig. 6c, e).

Fig. 6.

PW1⁺ cells contribute to myofibroblasts in adult hearts post-MI. a Schematic of the genetic lineage tracing strategy in Pw1^pCreER/m+;R26-tdT mice. Tamoxifen was administered 7 days before MI induction; hearts were analyzed 3, 7, and 14 days post-MI. b, d Whole-mount images of hearts at 7 (b) and 14 (d) days post-MI. Dashed lines outline infarcted regions. Scale bars, 5 mm. c, e Immunostaining for tdTomato on heart sections at 7 and 14 days post-MI. White scale bars, 5 mm; yellow scale bars, 200 μm. f, g Immunostaining for tdTomato and PDGFRa on heart sections at 7 and 14 days post-MI. Arrows indicate tdTomato⁺PDGFRa⁺ cells. Scale bars, 50 μm. h Quantification of the percentage of fibroblasts labeled by tdTomato. n = 6 biological replicates per group; both sexes included. Statistical significance was assessed by one-way ANOVA followed by Tukey post hoc tests (F(2, 15) = 243.5; achieved power = 0.81). i Flow cytometric analysis of the percentage of PW1⁺ cell-derived myofibroblasts in infarcted regions at 14 days post-MI. j Quantification of the percentage of PW1⁺ cell-derived myofibroblasts in infarcted tissues. n = 5 biological replicates; both sexes included. k, l Co-immunostaining for tdTomato with FSP1 (k) or aSMA (l) on heart sections at 14 days post-MI. Arrows indicate tdTomato⁺FSP1⁺ or tdTomato⁺aSMA⁺ cells in infarcted regions. Scale bars, 50 μm

Co-staining for tdTomato and the fibroblast marker PDGFRa revealed tdTomato⁺ fibroblasts within infarcted tissue (Fig. 6f, g). The proportion of PDGFRa⁺ fibroblasts that were tdTomato⁺ was significantly higher in infarcted regions (remote: 8.57% ± 1.07%; border: 44.63% ± 1.57%; Infarct: 53.97% ± 1.86%; Fig. 6h). Flow cytometry of the infarct tissues at 14 days showed that approximately 42% of PDGFRa⁺ cells were derived from PW1⁺ cells (Fig. 6i, j). Given that only about 6% of PDGFRa⁺ cells were tdTomato⁺ in normal adult Pw1^pCreER/m+;R26-tdT hearts [12], this rise to ~ 42% after MI indicates that PW1⁺ cells preferentially differentiate into myofibroblasts following injury. As PW1 is also expressed in a subset of ECs in the normal adult heart, some cVECs and endocardial ECs in infarcted tissues were also tdTomato⁺ (Fig. S10). Parallel tracing in Pw1^p2A-CreER/m+;R26-tdT mice showed a similar trend, with tdTomato⁺PDGFRa⁺ cells increasing from ~ 7% at baseline to ~ 33% post-MI (Fig. S11) [12].

Co-staining for tdTomato with other myofibroblast markers, FSP1 and aSMA, further confirmed the contribution of PW1⁺ cells to myofibroblasts in infarcted areas (Fig. 6k, l). Together, these results demonstrate that PW1⁺ cells play a key role in cardiac fibrogenesis after MI and may represent a potential therapeutic target for mitigating cardiac fibrosis in ischemic heart disease.

Discussion

Our previous work established that CAs originate from MCs (PDGFRa⁺ and/or PDGFRb⁺) during postnatal heart development [1]. Although both PDGFRa⁺ and PDGFRb⁺ MCs have been broadly implicated in adipogenesis, the specific progenitor subpopulations within this heterogeneous pool remained undefined. Here, we demonstrate that PW1⁺ cells–a distinct population encompassing subsets of both PDGFRa⁺ and PDGFRb⁺ MCs–contribute to the majority of CAs (> 75%, adjusted for tdTomato labeling efficiency) throughout postnatal development. This identifies PW1 as a marker of a committed adipogenic progenitor subset, refining the cellular resolution of adipogenic lineages beyond previously described mesenchymal populations and providing a potential target for understanding pathological cardiac adiposity.

Regarding fibrosis, our earlier study revealed that PW1 regulates ECM production in myofibroblasts after ischemic injury [12]. The present work extends this mechanistic insight by demonstrating that PW1⁺ cells in the normal adult heart serve as myofibroblast progenitors and represent a key cellular source of injury-induced myofibroblasts. This finding provides a cellular target for anti-fibrotic strategies aimed at modulating myofibroblast accumulation at its source.

Previous studies indicate that the vast majority (~ 97%) of intramyocardial adipocytes derive from cardiac MCs [1], and subepicardial adipocytes originate from the epicardium [18–20]. PW1 is highly expressed in neonatal and infant mouse hearts, and both Pw1^pCreER/m+;R26-tdT and Pw1^p2A-CreER/m+;R26-tdT lines efficiently label PDGFRa⁺ and PDGFRb⁺ MCs and epicardial cells at P3W. It is therefore consistent that tdTomato labels the majority of CAs in mice of these lines. In the postnatal heart, ECs do not contribute to intramyocardial or subepicardial adipocytes [14]. During embryonic development, endocardial ECs undergo endothelial‑to‑mesenchymal transition to generate mesenchymal progenitors that later migrate and differentiate into intramyocardial adipocytes [3, 21]. Thus, postnatal PW1⁺ ECs are unlikely to give rise to CAs.

Although PW1⁺ cells generate the majority of CAs during postnatal development and substantially contribute to myofibroblasts after MI, other progenitor populations also appear to be involved. In postnatal CA formation, lineage tracing in Pw1^pCreER/m+;R26-tdT mice showed that approximately 75% of CAs in P16W hearts were tdTomato-derived (Fig. 3e). Given that about 86% of PW1⁺ cells were labeled by tdTomato at P2W (Fig. S1h), a minor fraction of CAs likely originates from other progenitor pool(s). This is further supported by lineage tracing using Pw1^{p2A−CreER/m+};R26-tdT mice. Following MI, tdTomato-labeled cells in Pw1^pCreER/m+;R26-tdT mice accounted for only about 42% of PDGFRa⁺ myofibroblasts in injured regions (Fig. 6i, j), indicating a substantial contribution from other cellular sources. Consistent results were observed in Pw1^{p2A−CreER/m+};R26-tdT mice (Fig. S11h, i), confirming that multiple progenitor populations contribute to myofibroblast accumulation post-MI.

CAs are observed in various pathologies, such as arrhythmogenic right ventricular cardiomyopathy, myocardial ischemia, cardiac fibrosis, and dilated cardiomyopathy [22–30]. Ectopic CA deposition in these conditions may exert lipotoxic, prothrombotic, and pro-inflammatory effects, contributing to cardiovascular disfunction [3]. Understanding the cellular origins of CAs is therefore crucial for elucidating disease mechanisms and developing targeted therapies. While we have demonstrated the adipogenic capacity of PW1⁺ cells during postnatal heart development, it remains unclear whether PW1 is upregulated in pathological settings and whether these cells contribute to ectopic adipocyte deposition. Further studies should examine the role of PW1⁺ cells under different disease conditions.

The heterogeneity of both PW1⁺ and PW1⁻ populations–which include fibroblasts, pericytes, epicardial cells, ECs, and other cell types–poses a challenge for defining adipogenic or fibrogenic transcriptional signatures using bulk RNA‑seq. A more precise strategy would be to first identify the specific PW1⁺ mesenchymal subpopulation fated for adipogenesis, isolate it, and compare its transcriptome directly with corresponding PW1⁻ subpopulations. Similarly, resolving the heterogeneity within these populations is essential for accurately assessing fibrogenic priming. Future work aimed at characterizing the transcriptomic profiles of the specific subpopulations that give rise to CAs and myofibroblasts will be of great interest.

Persistent cardiac fibrosis disrupts tissue architecture, reduces compliance, and impairs mechanical and electrical function, often leading to heart failure [31]. Inhibiting fibrosis is therefore a major therapeutic goal in heart disease [32–34]. Previous studies have shown that genetic or immunotherapeutic ablation of myofibroblasts can reduce cardiac fibrosis and improve heart function after injury [35–38]. Given the significant contribution of PW1⁺ cells to post-MI myofibroblasts, it will be important to determine whether targeted ablation of PW1⁺ fibroblasts can mitigate fibrosis or enhance myocardial repair.

However, in vivo ablation of PW1⁺ cells presents considerable genetic challenges. PW1 is broadly expressed in multiple organs, including the heart, intestine, skeletal muscle, stomach, and lung [7, 12]. Systemic ablation would likely cause widespread tissue damage and lethality. More refined strategies–such as identifying more restricted markers for cardiac PW1⁺ subpopulations or employing dual Cre/Dre recombinase systems for subset-specific manipulation [39]—could enable targeted intervention without compromising viability. These approaches warrant further investigation to clarify the roles of PW1⁺ cells in adipogenesis, fibrosis, and other physiological or pathological processes.

Conclusions

In summary, our study has identified PW1⁺ cells as a progenitor population involved in the formation of CAs during postnatal heart development, and as significant contributors to cardiac fibrosis in response to MI. Elucidating the mechanisms that govern the differentiation of PW1⁺ cells into adipocytes and myofibroblasts could facilitate the development of novel therapeutic strategies targeting cardiac adiposity and fibrosis in pathological conditions.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MCs

Mesenchymal cells

CAs

Cardiac adipocytes

MI

Myocardial infarction

ECM

Extracellular matrix

cVECs

Coronary vascular endothelial cells

ECs

Endothelial cells

PLIN

Perilipin

scRNA-seq

Single-cell RNA sequencing

UMAP

Uniform Manifold Approximation and Projection

Author contributions

Y.W., S.K., C.L., J.T., and H.Z. conceived the study and designed the experiments. C.L., J.T., and H.Z. supervised the study and wrote the original draft. Y.W., S.K., L.Q., D.D, and T.F. conducted the experiments, performed data collection and analysis, and contributed to manuscript preparation and revision. W.Z. and J.M. provided technical support and assisted with experiments. All authors read and approved the final manuscript.

Funding

This work was sponsored by the National Key R&D Program of China (2023YFA1800700), National Natural Science Foundation of China (32322037, 32300683, 92268103, 32571300, 82570300), Shanghai Science and Technology Development Funds (22ZR1464900), Fundamental Research Funds for the Central Universities (22120240435), and Peak Disciplines (Type IV) of Institutions of Higher Learning in Shanghai.

Data availability

The scRNA-seq dataset used in this study is available in the GEO database under the accession number GSE198833. The complete Seurat-based analysis code is available on GitHub: https://github.com/qtian1817-lgtm/Peg3\_mice\_single-cell\_RNAseq-public-. All other data and materials generated or analyzed during this study are included in this published article and its supplementary files.

Declarations

Ethics approval and consent to participate

This research does not contain clinical experiments. All the animal experiments were performed at the ShanghaiTech University and approved under the project of “Heart development and regeneration” by the Institutional Animal Care and Use Committee of ShanghaiTech University. (Approval Number: 20200706003; Date of approval: July 6, 2020).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.