3: Multimodal Molecular Analysis Reveals Divergent Trajectories Of Wound Regeneration Versus Fibrosis

Abstract

Purpose: Scarring in the mouse dorsal dermis is mediated by pro-fibrotic, Engrailed-1 lineage-positive fibroblasts (EPFs). We recently showed that mechanotransduction blockade (YAP inhibition, using the drug verteporfin), results in complete wound regeneration, with full recovery of normal dermal appendages (hair follicles, glands), extracellular matrix (ECM) architecture, and tensile strength. This regenerative outcome following verteporfin treatment is mediated by Engrailed-1 lineage-negative fibroblasts (ENFs). The complex milieu of cell types and molecular signals involved in wound repair makes it difficult to study using any single data modality. Thus, we sought to use a holistic approach, incorporating multiple high-throughput, high-dimensional analyses, to define the divergent molecular events distinguishing typical scarring healing from verteporfin-induced wound regeneration. Methods: C57BL/6J mice underwent dorsal splinted excisional wounding per standard protocol. Wounds were treated with local injection of either verteporfin or vehicle control (PBS) on POD 0. We harvested unwounded skin and wounds at POD 2, 7, 14, and 30 (n=5 mice/timepoint and treatment) and subjected wound cells to three analyses: single-cell RNA-sequencing (scRNA-seq, using 10X Genomics Chromium); timsTOF, a recently-developed, high-throughput proteomic sequencing platform; and a novel machine learning algorithm for quantitatively comparing ECM ultrastructure. Results: Pseudotime analysis (Monocle3) of pooled scRNA-seq data revealed that fibroblasts followed two distinct transcriptional trajectories, one characterized by mechanical activation (En-1 lineage-positive, “fibrotic” trajectory) and the other characterized by developmental and regenerative pathways (En-1 lineage-negative; Rspo1, Dkk2/3, Trps1). Cross-platform data integration confirmed that fibroblasts in the fibrotic trajectory correlated with myofibroblast proteomic signatures (Col1a1/2, Fn1, etc.) and fibrotic/scar ECM features. In contrast, fibroblasts in the regenerative trajectory negatively correlated with myofibroblast markers and were associated with a “basket-weave” ECM pattern quantitatively indistinguishable from that of unwounded skin. Our integrated dataset suggested an important role for Wnt pathway proteins in ENF-mediated skin regeneration, so we compared POD 14 scars and regenerated wounds by multiplexed in situ hybridization (RNAScope) for Rspo1 (Wnt agonist), Trps1 (master hair follicle regulator), Ank1 (YAP target gene), and Dpp4 (EPF marker). Quantification of RNA granules across thousands of cells using a custom image analysis pipeline revealed that ENF-mediated healing (low Dpp4) in YAP-inhibited (low Ank1) wounds yielded regeneration of functional hair follicles through Wnt-mediated pathway activation (high Rpos1, Trps1). These data suggest that YAP inhibition unlocks wound regeneration via Wnt-active, En-1 lineage-negative fibroblasts. Conclusion: By studying regenerating (verteporfin-treated) versus scarring wounds across multiple healing timepoints and high-dimensional data modalities, we were able to profile fibrotic versus regenerative healing at unprecedented depth. Our integrated analysis revealed that dermal fibroblasts in these two wound settings exhibit distinct molecular trajectories defined by divergent transcriptomic, proteomic, and ultrastructural properties. Further, we found that wound regeneration in the context of verteporfin treatment is associated with suppression of mechanical signaling and activation of key Wnt pathway members including Trps1 (a gene with known hair follicle developmental roles). These results could have important implications for both the fundamental study of wound healing and potential anti-scarring therapeutic avenues