Fibroblasts Show Attenuated Wound Healing in the Setting of Dermal Melanoma
Oscar L. da Silva, Deshka S. Foster, MD, MA, Shamik Mascharak, BS, Alan T. Nguyen, Ashley L. Titan, MD, Ankit Salhotra, R Ellen Jones, MD, R Chase Ransom, BS, Jeffrey A. Norton, MD, Michael T. Longaker, MD, MBA.
Stanford University, Stanford, CA, USA.
PURPOSE: Fibroblasts are critical to wound healing providing deposition of scar connective tissue. In solid tumors, such as melanoma, cancer associated fibroblasts (CAFs) are activated in the stroma and support tumor proliferation via deposition of fibrotic desmoplasia. It has been known for many years that CAFs recapitulate wound healing signaling pathways. However, it has not yet been determined whether CAFs retain normal wound healing capabilities. Understanding such cellular behaviors has important implications towards designing therapeutics that effectively modulate these cells.
METHODS: For experimental assessment, endogenous melanoma mice (Braf;Pten;TyrCreER) were locally induced with activated tamoxifen and monitored for spontaneous tumor growth. When tumors reached 1cm in diameter, several wound-in-tumor experimental conditions were pursued in individual mice: Wound biopsy (with stenting of the wound edges to prevent rapid contraction, Fig. 1A), incisional wound (primary closure) or control (uninjured tumor tissue). On post-operative day (POD) 14 (Fig. 1B - biopsy model), mice were sacrificed and tissue was stained with picosirius red to characterize the connective tissue (Fig. 1C). Stented dermal wounds in C57B/6J mice were used for comparison and uninjured skin from healthy litter mates was used for control. Picosirius stained specimens were compared objectively using an automated connective tissue analysis methodology that uses machine learning technology.
RESULTS: Wound conditions for both wound-in-tumor and wound-alone models appeared grossly "healed" at POD14 (Fig. 1B). On automated analysis, t-SNE assessment of wound-in-tumor specimens from the endogenous melanoma model showed clustering alongside uninjured tumor specimens suggesting that even with wounding, extent of scar formation within tumors is similar to baseline tumor desmoplasia (Fig. 1C). This can be compared with non-tumor wound healing, which showed distinct clustering of dermal scar specimens away from uninjured skin (Fig. 1D). On isolated analysis of the melanoma specimens, connective tissue from both wound-in-tumor and uninjured tumor conditions appeared highly similar (Fig. 1E). These results suggest that melanoma CAFs are unable to enact normal wound healing activities.
CONCLUSION: In the context of dermal melanoma, fibroblasts do not recapitulate normal scar connective tissue deposition, suggesting terminal differentiation of these cells in the CAF context.
Figure Legend: A) Schematic of biopsy wound-in-tumor model, B) Images of biopsy model showing progressive wound closure over 14 days, C) t-SNE plot comparing picosirius red stained samples for the following conditions: non-tumor dermal wound healing, uninjured skin, uninjured tumor, biopsy wound-in-tumor model and incision wound-in-tumor model for an endogenous melanoma model, D) t-SNE analysis of dermal wound scar versus uninjured skin in isolation, E) t-SNE analysis of endogenous melanoma tumor model conditions in isolation. Clusters outlined with black dotted lines, examples of picosirius red stained specimens for labelled conditions in inset boxes, n=3 mice per condition per model, a minimum of 15 40X histological images assessed per condition per model.
Back to 2019 Abstracts