LB1021 Single-cell dynamics of fibroblast states during skin wound healing in mice

Abstract

Wound healing is a highly dynamic process over spatial and temporal scales. Key to wound healing outcomes are fibroblasts, whose precise functions in tissue repair are only partially understood. We constructed an atlas of mouse skin wound healing using several published single-cell RNA-seq (scRNA-seq) datasets. Using scRNA-seq data integration across wound healing time, we identified key cell types and states that are both shared across as well as unique to specific post-wounding timepoints. We reveal that fibroblasts dramatically alter after wounding as compared to unwounded skin fibroblasts, and also remain highly dynamic over time of healing. These transcriptional dynamics are not paralleled in other major wound-resident cell types. Furthermore, by “supervised clustering” and training-interpretable machine learning models we infer that these state dynamics are largely driven by selected gene categories, that are essential for key functions of fibroblasts. Finally, RNA velocity and pseudotime analysis suggest that fibroblast dynamics during small wound healing is an abridged version of large wound healing dynamics. Taken together, our findings reveal a highly dynamic transcriptomic landscape of fibroblasts during wound healing.