Single cell sequencing reveals endothelial plasticity with transient mesenchymal activation after myocardial infarction

The study investigates how endothelial cells (ECs) adapt to cardiac ischemia after myocardial infarction (MI), focusing on whether ECs undergo endothelial-to-mesenchymal transition (EndMT) or a partial, reversible mesenchymal activation. Given EC plasticity in development and disease and reports of EndMT in adult pathologies (tumors, atherosclerosis, fibrosis), the authors hypothesize that MI induces a transient mesenchymal activation in ECs, linked to metabolic reprogramming, rather than a stable mesenchymal fate. Using single-cell RNA sequencing and lineage tracing, they aim to map the temporal dynamics, metabolic signatures, and reversibility of EC phenotypic changes post-MI.

Background literature highlights EC heterogeneity and plasticity across development and disease, including EndMT in valve development and transitions to hematopoietic fate. In adult pathology, EndMT has been implicated in tumors, atherosclerosis, and cardiac fibrosis, though fibroblast lineage tracing suggests limited EC contribution to cardiac fibroblasts. EC phenotypic switching is known during angiogenesis (tip/stalk dynamics under VEGF/Notch control) with metabolic reprogramming (PFKFB3-driven glycolysis; FOXO1 coupling metabolism and growth). Single-cell studies in tumors revealed EC metabolic plasticity. Prior work suggested partial mesenchymal activation may facilitate angiogenesis, and clonally expanding ECs in ischemic tissue express mesenchymal markers. However, in vivo EC molecular signatures and plasticity in ischemia needed clarification.

• Study design and samples: Single-cell RNA sequencing (scRNA-seq) of non-cardiomyocyte fractions from murine hearts at homeostasis (day 0) and days 1, 3, 5, 7, 14, and 28 after MI induced by permanent LAD ligation. Initial analyses used a publicly available dataset; lineage tracing experiments generated new scRNA-seq datasets.
• Lineage tracing: Cdh5-CreERT2;mT/mG mice received tamoxifen to label ECs (GFP); hearts collected at multiple time points post-MI. Colla2-CreERT2;mT/mG mice were traced from days 3–7 post-MI and analyzed at day 28 to assess reversion to EC-like states.
• Tissue processing and scRNA-seq: Hearts digested with Miltenyi kits, dead cell removal, 10x Genomics Chromium Single Cell 3' v3 library prep, sequenced; CellRanger v3.0.1 used for alignment (custom mm10 reference including EGFP and tdTomato sequences), demultiplexing, counting.
• Bioinformatics: Seurat (v2.3.4) for QC (UMI/gene thresholds, mitochondrial content), normalization, clustering (t-SNE visualization), and cell cycle scoring. Endothelial cells identified by Cdh5 and Pecam1 expression. Monocle (v2.6.4) for pseudotime trajectory analysis using differentially expressed genes across time points. GO enrichment via Enrichr; gene set enrichment via clusterProfiler/MSigDB.
• EndMA definition and comparison: EndMA+ ECs defined as Pecam1+/Cdh5+ co-expressing mesenchymal markers (e.g., Serpine1, Fn1). Compared EndMA+ vs EndMA− ECs for gene expression, GO terms, and metabolic pathways (glycolysis, TCA, fatty acid signaling, PPP, glutamine metabolism).
• Ligand-receptor interactome: Weighted directed graphs of cell-cell communication constructed using curated ligand-receptor maps and STRING association scores with permutation testing for significance.
• In vivo pharmacology: TGF-β signaling inhibition with Galunisertib (LY2157299) administered by gavage to Confetti;Cdh5-CreERT2 mice to assess effect on EC clonal expansion 7 days post-MI.
• In vitro EndMT assays: HUVECs treated with TGF-β2 or TGF-β1+IL-1β±L-NAME; withdrawal experiments assessed reversibility of mesenchymal marker expression (TAGLN/SM22, CNN1), including conditions up to 10 days of treatment followed by medium switch.
• Metabolic assays: Seahorse XF96 extracellular flux for glycolysis; 13C-uniformly labeled glucose tracing with LC-MS/MS to quantify glycolytic flux changes upon TGF-β2 and after withdrawal.
• DNA methylation: Infinium Human-Methylation EPIC (850k) arrays analyzed with RnBeads and ADMIRE; assessed methylation in promoters/gene bodies and regulatory elements of mesenchymal genes across control, TGF-β2 treatment, and withdrawal.
• Validation: Immunostaining and FACS for mesenchymal markers in ECs at post-MI time points; bulk RNA-seq of Cdh5+ isolated cells at day 3 vs homeostasis.
• Statistics: Differential expression with Seurat’s bimodal likelihood test (Bonferroni correction); GO enrichment combined scores; qPCR analyzed by Kruskal–Wallis with Dunn’s post hoc or Student’s t-test with Welch’s correction; chi-squared for cell cycle ratio changes; permutation-based significance for ligand-receptor networks.

• Cellular composition and early responses: scRNA-seq identified 19 clusters including 4 EC clusters; EC and mesenchymal cell proportions declined during days 1–7 post-MI with immune cell influx. ECs at day 1 showed enriched GO terms for hypoxia response, inflammatory response, apoptosis, and angiogenesis, with increased expression of Bax, Trp53, Hif1a, Ldha, Il1b, Il6, and Tnf.
• Proliferation dynamics: EC proliferation increased peaking around day 3, with elevated S-phase and G2/M-phase gene signatures (e.g., Cks2, Tyms, Birc5, Cdc20, Ccna2, Cdca8). Cell cycle phase ratios changed significantly (χ2 with Yates correction p=0.0004). Proliferation returned to baseline by day 14.
• Transient mesenchymal activation (EndMA): Trajectory analysis revealed 5 EC states; ECs at days 1–7 predominantly occupied a state (state 4) enriched for extracellular matrix and mesenchymal genes (e.g., Col3a1, Fn1, Serpine1, Vim, S100a4, Mmp14) with reduced endothelial marker Cdh5. State 4 abundance increased between days 1–7 and returned to baseline by day 14, indicating transient EndMA. Bulk RNA-seq of Cdh5+ cells at day 3 confirmed mesenchymal marker induction.
• Metabolic reprogramming: EndMA+ ECs exhibited increased glycolysis gene expression and reduced fatty acid signaling and TCA cycle genes compared to EndMA− ECs, indicating a switch toward glycolytic metabolism. Time-course analysis showed early induction of glycolysis (day 1) and reduced fatty acid genes (e.g., Fabp4 at day 3). In vitro, TGF-β2 increased glycolytic flux (13C-glucose tracing) without altering glucose uptake; changes reversed upon withdrawal.
• Lineage tracing confirms reversibility: In Cdh5-CreERT2;mT/mG mice, GFP+ ECs transiently co-expressed mesenchymal markers (Col1a1, Col3a1, Serpine1) peaking at days 3–7, then returning toward baseline with few cells losing EC markers entirely. Re-clustering identified an EndMA-like GFP+ cluster enriched at day 7. Colla2-CreERT2;mT/mG tracing from days 3–7 showed GFP+ cells in EC clusters and IB4+ histology at day 28, indicating reversion to EC-like transcriptional/phenotypic state.
• Functional impact on expansion: TGF-β receptor inhibitor Galunisertib reduced EC clonal expansion in Confetti mice by 79 ± 19% at day 7 in the border zone versus infarcted controls (p=0.03), suggesting EndMA contributes to EC expansion during repair.
• Cell-cell communication: EndMA+ ECs showed increased outgoing ligand interactions, particularly involving extracellular matrix organization, PDGF binding, and collagen fibril organization, and targeted signaling to ECs lacking mesenchymal markers.
• Reversibility in vitro and epigenetics: HUVECs treated with TGF-β2 upregulated TAGLN (SM22) and CNN1 by day 3; marker expression returned to baseline within 7 days after withdrawal, whereas continuous stimulation sustained/increased expression. DNA methylation arrays revealed no promoter/gene body methylation changes in selected mesenchymal genes (e.g., COL4A1, TAGLN, CNN1) but identified reversible demethylation in regulatory regions upon TGF-β2 and re-methylation after withdrawal.

The findings support that myocardial infarction induces a transient mesenchymal activation (EndMA) of endothelial cells rather than a stable, full endothelial-to-mesenchymal transition. This activation peaks between days 3–7 post-injury and resolves by approximately day 14, aligning with a metabolic shift toward glycolysis and subsequent re-establishment of endothelial gene expression. Lineage tracing confirms that most ECs maintain some endothelial identity during EndMA and do not permanently convert into mesenchymal cells, addressing prior discrepancies between observations of EndMT markers and lack of long-term mesenchymal fate in lineage studies. Functionally, EndMA likely facilitates EC migration, proliferation, and clonal expansion necessary for vascular repair, as indicated by reduced clonal expansion under TGF-β inhibition. Enhanced ligand-receptor signaling from EndMA cells suggests roles in modulating the reparative microenvironment. The reversible nature and associated epigenetic changes at regulatory elements indicate a dynamic response to hypoxic and inflammatory cues rather than a fixed differentiation event.

This study delineates a time-resolved, reversible mesenchymal activation of endothelial cells after myocardial infarction, coupled with transient metabolic reprogramming toward glycolysis. Single-cell transcriptomics and endothelial lineage tracing demonstrate that ECs acquire mesenchymal features during the first week post-injury and revert to baseline by two weeks, without adopting a lasting mesenchymal fate. EndMA appears to promote EC expansion and tissue repair, as TGF-β pathway inhibition diminishes clonal expansion. The results reconcile conflicting views on EndMT in cardiac injury by framing it as a reversible continuum. Future work should define regulators that govern EndMA reversibility, assess its occurrence across other diseases and risk conditions, and determine how modulating EndMA impacts functional vascular regeneration and fibrosis.

• Sample size at the single-cell level was limited to one mouse per time point in several scRNA-seq analyses (both the public dataset used initially and the lineage tracing datasets), which may constrain generalizability and detection of rare cell states.
• EndMA identification relied on transcript-level markers; rare instances of complete EndMT might remain undetected despite lineage tracing.
• In vitro validation employed HUVECs, which may not fully recapitulate cardiac EC behavior in vivo.
• While DNA methylation analysis suggested reversible changes at regulatory regions, functional impacts of these epigenetic modifications were not fully resolved.
• The study cannot exclude indirect contributions of EndMA cells to fibrosis via secretion of extracellular matrix and inflammatory factors.