henhaobio
Sign in

Abstract

BACKGROUNDSmooth muscle cells (SMCs) are reported to contribute the majority of cholesterol-overloaded foam cells in human and mouse atheromas. However, the transcriptome, specific markers, and biologic itinerary of SMC foam cells relative to macrophage foam cells have not been determined. METHODSTranscriptomic analysis by single cell RNA sequencing (scRNA-seq) was performed on fresh coronary segments from heart transplant recipients with early to intermediate stage atherosclerosis. The gene expression pattern of a putative cluster of SMC foam cells was compared to those of cultured SMCs loaded with either aggregated low density lipoprotein (agLDL) or cholesterol bound to methyl-{beta}-cyclodextrin (Chol-M{beta}CD). Candidate markers of SMC foam cells not expressed by macrophage foam cells were validated in ours and publicly available datasets, by spatial transcriptomics and by immunofluorescence microscopy of human atheromas. Pathway analysis was performed using Gene Set Enrichment Analysis Hallmark gene sets. RESULTSSMC foam cells derived from fibromyocytes were tentatively identified using a panel of markers upregulated with in vitro cholesterol loading of SMCs. agLDL loading reproduced the same transcriptional profile, whereas Chol-M{beta}CD did not reproduce any in vivo SMC state. Top genes highly represented in SMC foam cells included SERPINE1, encoding plasminogen activator inhibitor 1 (PAI-1) and CFH, complement factor H, which were validated in further human coronary scRNA-seq datasets, by Xenium spatial transcriptomics, and by immunofluorescence microscopy. Relative to macrophage foam cells, SMC foam cells exhibit a distinct biologic itinerary, including activation of extracellular matrix, coagulation and angiogenesis pathways. CONCLUSIONSSMC foam cells, which are derived from fibromyocytes ("lipomyocytes"), exhibit unique markers and biologic programs that differ markedly from macrophage foam cells in atherosclerotic plaque development. Further understanding of the role of lipomyocytes and their expression of CFH and PAI-1 expression in plaque biology may offer novel therapeutic options to reduce ischemic cardiovascular disease. Novelty and SignificanceO_ST_ABSWhat is Known?C_ST_ABSO_LISingle-cell transcriptomic studies have demonstrated vascular SMCs undergo extensive phenotypic modulation during atherosclerosis, giving rise to fibromyocytes and other intermediate cell states. C_LIO_LISMC-derived foam cells constitute a major proportion of plaque foam cells. C_LIO_LIPrevious studies have shown that SMC-derived foam cells differ from macrophage-derived foam cells in cholesterol handling and lipid metabolism, making their identification using conventional macrophage-associated foam-cell markers challenging, and highlighting the need for SMC foam cell-specific molecular markers. C_LI What New Information Does This Article ContributeO_LIBy integrating published transcriptomic datasets with bulk RNA-sequencing of agLDL-loaded human SMCs, we identify the transcriptional program of human SMC-derived foam cells and demonstrate that these cells are embedded within the fibromyocyte population rather than forming a distinct cell cluster. C_LIO_LIagLDL loading in vitro reproduces the in vivo SMC foam-cell phenotype, whereas loading with cholesterol bound to cyclodextrin induces an inflammatory state not found in any in vivo SMC state in human atherosclerosis. C_LIO_LICFH and SERPINE1 were identified as specific markers of SMC-derived foam cells and validated by immunostaining and Xenium spatial transcriptomics, providing robust markers for identification of these cells in human atherosclerotic plaques. C_LIO_LIThe top transcriptional programs activated in SMC foam cells differ completely from macrophage foam cells and include extracellular matrix remodeling, complement regulation, coagulation and angiogenesis. C_LI In this study, we mapped smooth muscle foam cells to the fibromyocyte cluster of arterial SMC subtypes in human coronary artery atheromas. In vitro loading of smooth muscle cells with aggregated low-density lipoprotein, the presumed physiologic substrate for foam cell formation during plaque development, induced a gene expression pattern that localized to the same fibromyocyte cluster. The proposed sequence of SMC foam cell formation is that fibromyocytes present in human pre-atherosclerotic intima generate the proteoglycans that bind lipoproteins and allow their uptake by fibromyocytes to generate cholesterol-loaded "lipomyocytes". Lipid loading of fibromyocytes drives expression of specific markers and transcriptional programs, but with an overall gene expression pattern that continues to resemble fibromyocytes. We demonstrate here that lipomyocytes express the specific markers CFH and SERPINE1, and exhibit activation of a completely different set of pathways relative to macrophage foam cells in the plaque. These findings provide evidence for the unique phenotype and specific markers of lipomyocytes relative to macrophage foam cells in human atherosclerosis, which will be critical in evaluating the response of these different types of foam cell to novel therapies.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 3 (1 Aug 2026) · CC BY 4.0

Smooth muscle foam cells in early-to-intermediate human coronary plaques map to fibromyocytes, express CFH and SERPINE1 unlike macrophage foam cells, and show matrix, coagulation and angiogenesis programs.

Smooth muscle cells (SMCs) are reported to supply most foam cells in human and mouse atheromas, but their transcriptome and markers had not been defined; earlier single-cell studies flagged 'foam-like' SMC clusters using lipid-metabolism gene panels. Loading cultured SMCs with aggregated LDL (agLDL) ties these cells to fibromyocytes, CFH and SERPINE1 are proposed as markers with spatial and protein checks, and the widely used cyclodextrin-cholesterol model is shown to match no in vivo SMC state.

Key results

  1. SuggestedThree separate SMC foam cell gene signatures score highest in the fibromyocyte cluster of coronary scRNA-seq. Macrophage foam cell signatures stay in macrophage clusters, with little crossover. On this basis, SMC foam cells sit inside the fibromyocyte population rather than forming their own cluster.

    After QC, 4,812 cells from 3 donors formed 15 clusters. These included SMC1-SMC4 (contractile), SMC5 (early fibromyocyte) and SMC6 (fibromyocyte), which agreed with MetaPlaq v2 labels transferred by scANVI. The literature-curated, Mosquera and Bashore SMC foam signatures peaked in fibromyocytes, with weaker signal in fibroblasts and pericytes. FBLN1+ fibroblasts were rare in the intima and BODIPY-negative. · Single-cell RNA Sequencing of Diverse Vascular…; Literature-guided Transcriptomic Mapping of SMC…

  2. SuggestedCFH (complement factor H) and SERPINE1 (encoding plasminogen activator inhibitor 1, PAI-1) are proposed SMC foam cell markers. Both are enriched in fibromyocytes and scarce in macrophages in two human coronary datasets. In ApoE-/- mice, both favour Cd45-negative (SMC-derived) over Cd45-positive macrophage foam cells.

    Chosen from fibromyocyte markers with pct.1 > 0.39 and pct.2 < 0.25; CFH also rose in agLDL-loaded SMCs. In IGVFDS2114XGWE (3 human coronary samples after excluding a calcified one), SERPINE1 was mainly in fibromyocytes and CFH in fibromyocytes and fibroblasts. In foam cells from ApoE-/- mouse aorta (GSM3215436), Serpine1 and Cfh were enriched in Cd45-negative cells, with minimal expression in Cd45-positive ones. · Identification of Candidate Markers of…; Figure legends

  3. SuggestedIn tissue, CFH and SERPINE1 transcripts sit in fibromyocyte-rich areas of human coronary lesions, and PAI-1 and CFH proteins sit in lipid-laden smooth muscle actin (SMA)-positive cells. Both are minimal in macrophage foam cell areas. Some CFH-positive fibromyocytes lack SERPINE1.

    Xenium on 4 coronary samples (5,001-gene pan-tissue panel plus 100 custom genes): both markers were in regions with high fibromyocyte scores and minimal where macrophage foam scores were high; THY1 showed only low enrichment. Immunofluorescence on 3 sections from 3 hearts: line scans showed PAI-1 and CFH overlapping SMA and BODIPY lipid stain, with little overlap with CD45. · Xenium Spatial Transcriptomic Processing and…; Spatial Transcriptomic Validation of SMC Foam…; Tissue-level validation of SERPINE1 and CFH…; Figure legends

  4. ShownLoading cultured human SMCs with agLDL produces a program that maps onto fibromyocytes. Loading with cholesterol bound to methyl-β-cyclodextrin (Chol-MβCD) instead induces an inflammatory program that maps to macrophage clusters, so the cyclodextrin model does not mimic in vivo SMC foam cells.

    Bulk RNA-seq, n = 4 per group, 72 h: 115 genes up with agLDL and 169 up with Chol-MβCD (log2FC > 1, adjusted P < 0.05). agLDL genes scored highest in fibromyocyte and fibroblast clusters and were enriched for extracellular matrix organization. Chol-MβCD genes scored highest in macrophage clusters, with cytokine, immune and cell cycle terms. BSA controls correlated best (Spearman) with fibromyocytes. · Comparison of bulk RNA-seq profiles with…; Transcriptomic Profiling of…

  5. ShownCompared with the most foam-like macrophage cluster, CFH-positive and SERPINE1-positive fibromyocytes are enriched for the Hallmark sets epithelial-mesenchymal transition, myogenesis, coagulation, angiogenesis and oxidative phosphorylation. The macrophages are enriched for inflammatory and interferon programs.

    Gene Set Enrichment Analysis (fgsea, MSigDB Hallmark, genes ranked by average log2 fold change, significance at FDR < 0.05) against Cluster 2 macrophages, the cluster with the strongest macrophage foam signature. The macrophages were enriched for TNFα signalling via NF-κB, interferon responses and IL6-JAK-STAT3. Curated SERPINE1 and CFH regulatory programs scored highest in early fibromyocytes and fibromyocytes. · Pathway Enrichment Analysis; Regulatory Pathways Associated with CFH and…; Figure legends

Methods, data and reproducibility

Coronary artery scRNA-seq (10x Chromium; three transplant recipients) was annotated into SMC states and scored with foam cell signatures, including signatures from bulk RNA-seq of human SMCs loaded with agLDL or Chol-MβCD. Candidate markers were checked in public human and mouse scRNA-seq, by Xenium and by immunofluorescence. Fibromyocytes were then compared with macrophages by Hallmark GSEA.

  • 10x Chromium 3' v3 scRNA-seq of noncalcified proximal-to-mid right coronary artery segments from heart transplant recipients with nonischemic cardiomyopathy · human · n = 3 donors; 4,812 cells after QC
  • Bulk RNA-seq of cultured human vascular SMCs treated for 72 h with agLDL (100 μg/mL), Chol-MβCD (10 μg/mL) or BSA control · human · n = 4 per group
  • MetaPlaq v2 vascular reference atlas (unpublished), VSMC subset, used for label transfer and signature scoring · human (carotid and coronary samples from patients) · n = 26,771 VSMCs
  • Public human left coronary artery scRNA-seq from the Quertermous group (IGVF) · human · n = 3 samples analysed (P4 excluded as heavily calcified) · IGVFDS2114XGWE
  • Public scRNA-seq of aortic foam cells from atherosclerotic ApoE-/- mice · mouse · GSM3215436
  • Xenium spatial transcriptomics of formalin-fixed, paraffin-embedded left anterior descending or right coronary arteries from transplant patients with cardiomyopathy (Stanford biorepository) · human · n = 4 samples from 4 explanted hearts
  • Immunofluorescence of coronary plaque sections for PAI-1 or CFH with SMA, CD45, BODIPY and DAPI · human · n = 3 sections from 3 explanted hearts
Methods & tools
10x Genomics Chromium, Cell Ranger, Seurat, MAST, Harmony, scArches (scVI/scANVI), Scanpy, DESeq2, fgsea (MSigDB Hallmark), Monocle 3, Xenium Analyzer, scRNAUtils
Reagents & environment
R 4.2.3; Seurat 4.3.0 (Methods) or 4.2.3 (Results) for scRNA-seq, Seurat 5.2.1 for Xenium; Xenium instrument software 4.0.1.4, onboard analysis 3.3.1.1; scArches scVI with 2 layers and n_latent=30.
Reproducibility gaps
The paper says the scRNA-seq data 'will be deposited' in GEO but gives no accession, and the printed GEO address is malformed. No code repository is given. The MetaPlaq v2 reference is unpublished. QC criteria and bulk read processing are left to the Supplemental Methods.
Limitations (5)
  • The scRNA-seq coronary samples came from transplant recipients whose cardiomyopathy (for example after viral myocarditis) could have altered intimal SMC expression. The authors argue that close alignment with the MetaPlaq atlas mitigates this. · stated by the authors
  • Lesions were noncalcified and early to intermediate stage. Whether SMC foam cells in advanced lesions share the same expression and pathways could not be ruled out. · stated by the authors
  • Xenium used formalin-fixed, paraffin-embedded tissue, where lipid is removed during processing, so no lipid stain on an adjacent section could confirm foam status in the spatial data. · stated by the authors
  • In the scRNA-seq data, foam status is inferred from gene signatures, not lipid content. The GSEA compares marker-positive fibromyocytes with macrophages, so lineage effects are not separated from lipid-loading effects; there is no comparison of foam with non-foam fibromyocytes. · noted in this summary
  • Sample sizes are small (3 scRNA-seq donors, 4 Xenium samples, 3 hearts for immunofluorescence) and marker validation is qualitative. The text is also inconsistent: Seurat 4.3.0 vs 4.2.3, and an SMC7 cluster in the Figure 1 legend is not mentioned in Results. · noted in this summary

Read this if you study SMC phenotypic modulation or foam cells in atherosclerosis and need markers to separate SMC-derived from macrophage foam cells, or if you are choosing an in vitro cholesterol-loading model for SMCs.

Citations
1

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Similarity Map
SelectedSame fieldCross-field
THISMosquera '23Ibikunle '25Sharysh '26Buono '23Bashore '24Gurung '25
68% semantic matchGenomics
José Verdezoto Mosquera, Gaelle E Auguste, Doris Wong+20 more·Cell Reports·Nov 1, 2023· 123 citations