Growth arrest specific-6 and angiotoxin receptor-like signaling drive oral regenerative wound repair

Abstract

Rapid and scarless wound repair is a hallmark of the oral mucosa, yet the cellular and molecular mechanisms that enable this regeneration remain unclear. By comparing populations of murine oral mucosal fibroblasts (OMFs) and facial skin fibroblasts (FSFs), we have identified mechanisms that facilitate regeneration over fibrosis. We found that OMFs used growth arrest specific-6 (GAS6)-angiotoxin receptor-like (AXL) signaling to suppress fibrosis-related mechanosignaling through focal adhesion kinase (FAK) in vitro. Inhibition or knockdown of AXL in the murine oral mucosa resulted in fibrotic wounds and increased activation of FAK. Stimulation of AXL by exogenous GAS6 in the murine facial skin yielded wounds that healed regeneratively as assessed by collagen deposition and organization. Rare human oral scars that resulted from repetitive injury showed decreased expression of GAS6 and AXL and increased FAK. Activating AXL by exogenous GAS6 in repetitively injured mouse oral tissue resulted in better wound healing outcomes and reduced scarring. Altogether, we show that AXL signaling is necessary for murine regenerative wound healing in the oral mucosa and sufficient to limit facial skin fibrosis.

Fig. 1.. Single-cell RNA sequencing reveals distinct mouse fibroblast subpopulations after oral and facial injury.

(A) Schematic depicting the timeline of wounding before harvest for single-cell RNA sequencing. Figure was created using Biorender.com. (B) Histology of oral and dermal tissue for UW, POD1, POD2, POD4, and POD7; dashed black and white lines indicate the wound bed. Scale bars, 250 μm (top) and 50 μm (bottom). (C) Immunofluorescent staining of collagen I (Col1; top) and collagen III (Col3; bottom, both in red) of facial and oral POD4 wounds (dashed lines mark the wounded dermis). Scale bar, 25 μm (n = 10). Quantification for staining (right); facial and oral percentages of positive staining for Col1 (left) and Col3 (right) were compared separately using unpaired t test (*P < 0.05). (D) Dendrogram of bulk RNA sequencing of oral and facial UW and POD7 fibroblasts. (E) UMAP (uniform manifold approximation and projection) plots visualizing all cells captured in both tissues across all time points; the fibroblasts (outlined) were selected on the basis of differential gene expression and isolated in silico to generate (F) a UMAP plot showing the nine types of fibroblasts identified (left), with their key genes and functions noted in the corresponding table (right) (n = 10 mice per POD). (G) Ratio plots showing the ratio of each fibroblast cluster at each time point in facial and oral wounds. Fibroblast populations are colored according to (F).

Fig. 2.. AXL signaling is up-regulated in the oral mucosal fibroblasts upon wound repair.

(A) Heatmap showing the differential number of interactions between each cell type, with red representing an increased number of interactions in the OM (oral mucosa) compared with FS (facial skin) and blue indicating a lower number of interactions. Each row and column represent a cell type, labeled by color, with the populations with the highest increase in OM signaling outlined in black. (B) Heatmap of the Gas6-Axl signaling network in facial versus oral nonimmune cells. Each row and column refer to a cell type, with cells pseudocolored according to the bar (left). Each colored box represents the cell type interacting with the cell type indicated by the labels on the left. Bars on top refer to the strength of interaction as calculated by CellChat (C) A violin plot of Axl expression in fibroblasts within the scRNA-seq dataset is shown, split into FSFs (black) and OMFs (red) (n = 10 mice per POD). (D) RNAscope in situ hybridization for Axl transcripts (red) and IF (immunofluorescence) staining for COL1A1 (green) and DAPI (blue) for facial (left) and oral (right) samples at all time points; dashed white lines indicate the wound bed. Scale bar, 25 μm, (n = 12). (E) Quantification of FACS experiment showing the percentages of AXL⁺ and pAXL⁺ fibroblasts in the wound bed of facial (black) and oral (red) tissue at each time point. At each time point, facial AXL⁺ fibroblasts were compared with oral AXL⁺ fibroblasts using a one-way ANOVA test and Tukey post hoc corrections (*P < 0.05).

Fig. 3.. Spatial analysis of facial and oral wounds by Visium and CODEX analysis.

(A) Experimental outline of spatial analysis Visium spatial plots of (B) UW oral, (C) UW facial, (D) POD4 oral wounds, and (E) POD4 facial wounds by UMAP (left) and scRNA-seq annotations (right). (F) Spatial plots of Axl and Ptk2 (Fak) expression in facial (top) and oral wounds (bottom). (G) Differential interaction maps in facial UW versus oral UW (top) and facial POD4 wounds versus oral POD4 wounds (right) (n = 2). (H) UMAP of CODEX-defined clusters. (I) Bar graph quantifying fibroblast 2 and fibroblast 4 proportions in facial and oral wounds. (J) Histograms of Axl expression in CODEX-defined fibroblast clusters (n = 3) (*P < 0.05).

Fig. 4.. FAK expression after mechanical stimulation suppresses AXL expression in vitro.

(A) Facial and oral fibroblasts were isolated and plated on a 3D hydrogel stretching device. Gels were either unstretched, stretched, or treated with inhibitor during stretching. IF staining was performed on unstretched (top), stretched (middle), and stretched plus FAKi (bottom) gels plated with FSFs (left) and OMFs (right). Gels were stained for FAK (green), GAS6 (red), and AXL (yellow), along with DAPI (blue). Scale bars, 100 μm (n = 3 gels per group). (B) Quantification of FAK, GAS6, and AXL staining from in vitro stretching experiments in (A). All groups were compared using one-way ANOVA with Tukey’s post hoc corrections (*P < 0.05). (C) IF staining was performed on unstretched (top), stretched (middle), and stretched plus AXLi (bottom) gels plated with FSFs (left) and OMFs (right). Gels were stained for FAK (green), GAS6 (red), and AXL (yellow) along with DAPI (blue). Scale bars, 100 μm (n = 3 gels per group). (D) Quantification of FAK, GAS6, and AXL staining in (C). White arrows show double-positive AXL⁺GAS6⁺ fibroblast cells. All groups were compared using one-way ANOVA with Tukey’s post hoc corrections (*P < 0.05).

Fig. 5.. AXL signaling up-regulation can revert facial wounds to heal like oral wounds.

(A) Schematic of the experiment workflow for AXLi and exogenous GAS6 experiments and (B) schematic depicting the predicted outcome of both +GAS6 protein and +AXLi on oral and skin wounds (d.-# = # days before wounding). Schematic made using Biorender.com. (C) H&E staining (right) of UW, POD4 wounded control, POD4 plus GAS6 protein, and POD4 plus AXLi facial and oral tissues, with dermal thickness quantified. All groups were compared using one-way ANOVA with Tukey’s post hoc corrections (*P < 0.05) (right). Scale bars, 250 μm (n = 3 mice per group). (D) IF staining of the controls, +GAS6 protein, and +AXLi tissues. Merged FAK (green), AXL (red), and DAPI (blue) images (top) and FAK (middle) and AXL (bottom) channels. Scale bars, 250 μm (n = 3 mice per group). (E) Quantification of immunofluorescence in (D), AXL staining (left) and FAK staining (right). All groups were compared using one-way ANOVA with Tukey’s post hoc corrections (*P < 0.05, **P < 0.01, and ***P < 0.001).

Fig. 6.. Human and mouse oral scars lose Gas6 and Axl expression.

(A) Schematic depicting the timeline for obtaining human samples. Schematic made using Biorender.com. (B) H&E staining of healthy (left) and reinjured (right) facial and oral tissues. Scale bar, 250 μm (n = 5 samples per group) (C) IF staining for DAPI (blue), COL1 (red), FAK (white), and AXL (green) in human facial (left) and oral (right) reinjured tissue. Quantification for staining (right); facial and oral percentages of positive staining for FAK (left) and AXL (right) were compared separately using unpaired t test (*** P < 0.001). White boxes delineate insets for each tissue. Scale bars, 250 μm (left) and 50 μm (right) (n = 5 samples per group). (D) Schematic showing the timeline of mouse repetitive injury model and predicted outcomes of untreated versus GAS6-treated injuries. (E) H&E staining of POD14 and POD30 facial (top) and oral (bottom) reinjured tissue versus reinjured tissue plus GAS6 protein. Scale bars, 150 μm (n = 3 mice per group). (F) IF staining as in (C), with POD14 plus GAS6 samples from facial (top) and oral tissue (bottom). Quantification for the percentage of positive staining of FAK (left) and AXL (right); groups were compared using one-way ANOVA with Tukey’s post hoc corrections (***P < 0.001). Scale bars, 150 μm (n = 3 mice per group).