BACKGROUND: Smooth muscle cells (SMCs) comprise the majority of cells in human atherosclerotic lesions and are thought to be a major source of cholesterol-overloaded foam cells in human and mouse atheromas. However, the transcriptomic profile, specific markers, and biologic itinerary of SMC foam cells relative to macrophage foam cells remain poorly defined. METHODS: Single-cell RNA sequencing (scRNA-seq) was performed on fresh coronary artery segments from heart transplant recipients with early- to intermediate-stage atherosclerosis. Gene expression in a putative SMC foam cell cluster was compared with cultured SMCs loaded with aggregated low-density lipoprotein (agLDL) or cholesterol-methyl-beta-cyclodextrin (Chol-MbetaCD). Candidate markers distinguishing SMC from macrophage foam cells were validated using additional publicly-available scRNA-seq datasets, Xenium spatial transcriptomics, and immunofluorescence microscopy of human coronary atheromas. Pathway analysis was performed using Gene Set Enrichment Analysis Hallmark gene sets. RESULTS: A distinct SMC foam cell cluster derived from fibromyocytes (‘lipomyocytes’) was identified using markers induced by in vitro cholesterol loading. agLDL loading reproduced the lipomyocyte transcriptional profile, whereas Chol-MbetaCD induced an inflammatory phenotype colocalizing with macrophages rather than lipomyocytes. Lipomyocytes highly expressed SERPINE1, encoding plasminogen activator inhibitor-1 (PAI-1), and CFH, encoding complement factor H, which were validated in human coronary lesions by spatial transcriptomics and immunofluorescence microscopy. Compared with macrophage foam cells, lipomyocytes demonstrated distinct pathway activation, including enrichment of extracellular matrix, coagulation and angiogenesis pathways. CONCLUSIONS: SMC foam cells, or lipomyocytes, represent a distinct foam cell phenotype with unique markers and biologic programs that differ from macrophage foam cells during atherosclerotic plaque development.
Target-Side Paraphrase Augmentation for Sign Language Translation with Large Language Models
arXiv:2605.31393v1 Announce Type: cross Abstract: Sign language translation (SLT) remains constrained by limited paired sign-video/text corpora and heavy-tailed target vocabularies. We study target-side augmentation in



