Single-cell and spatial transcriptomic atlas of pathological scars uncovers neuro-fibroblast crosstalk mechanisms across scar types

Pathological scars, including hypertrophic scars and keloids, arise from excessive production and deposition of connective tissue such as collagen during wound healing, posing significant clinical challenges. Understanding the cellular and molecular dynamics underlying these scars is critical for developing effective therapies. Here, we integrate single-cell and spatial transcriptomics to characterize the cellular heterogeneity and molecular signatures of key scar-associated cell types, including pericytes, fibroblasts, and endothelial cells. Notably, we identify ECM-activated fibroblasts as a prominent subtype associated with pathological scarring. Cell-cell communication analysis reveals significantly enhanced interactions between fibroblasts and neural cells in pathological scars, especially in hypertrophic scars—a finding that aligns with clinical features such as fibrosis and pruritus. Moreover, we identify subtype-specific ligand-receptor pairs, including LAMB2–ITGA6 in hypertrophic scar and NLGN2–NRXN1 in keloid, mediating these fibroblast–neural communications. To support future research, we developed CellCellMarker 3.0, an updated and curated database integrating experimentally validated and transcriptomics-derived scar marker genes, along with ScarGPT, an intelligent Q&A system optimized using the DeepSeek R1 large language model, integrated with CellCellMarker data and a large corpus of scar-related publications. To comprehensively characterize the cellular heterogeneity of pathologic scars, we analyzed single-cell RNA-sequencing (scRNA-seq) data comprising two biological replicates for each scar type—keloid, hypertrophic scar. After quality control, we obtained a total of 42,607 single cells, including 17,915 cells from hypertrophic scars, 11,594 cells from keloids, and 13,098 cells from normal scars. This integrated dataset forms the basis for exploring the cellular architecture and molecular pathways underlying distinct pathologic scar phenotypes. Unsupervised clustering identified 23 discrete cell populations. In addition, one spatial transcriptomics sample from hypertrophic scar and one from keloid were included. *************************************************************** Due to privacy and ethical considerations, raw sequencing data from human scar tissue samples are not publicly available ***************************************************************