Dedifferentiated Schwann cell-derived TGF-β3 is essential for the neural system to promote wound healing

Abstract

Rationale: Wound healing is among the most complicated physiological processes and requires the synchronization of various cell types with distinct roles to re-establish the condition of the original skin. Patients affected by peripheral neuropathies often experience failure to heal. Loss of Schwann cells (SCs), a crucial population of peripheral nervous system cells in skin, may contribute to chronic wounds. However, the role of SCs in wound healing are poorly understood. Methods: The activity of SCs was investigated by using a cell atlas of the wound healing process, which was generated by integrating single-cell RNA sequencing (scRNA-seq) libraries covering different states of mouse back skin. The results of in silico analysis were validated by in vitro cell culture and in vivo mouse model. Selective inhibitors and conditional RNAi by virus transfection were utilized to investigate the role of SCs in wound healing. Findings from mouse experiments were further verified in scRNA-seq analysis of diabetic patients. Results: Our in silico analysis revealed the heterogeneous cellular components of skin and the dynamic interactions of neural crest derived cells (NCs) with other cell types. We found that SCs dedifferentiated at an early stage of wound repair with upregulated Wnt signaling. We also identified dedifferentiated SC (dSC) defect in diabetic wounds in both mouse and human. Wnt inhibition at the wound site repressed SC dedifferentiation, leading to defective repair. Furthermore, dSCs derived TGF-β3, which is context-dependent, promoted the migration of fibroblasts and keratinocytes. Moreover, TGF-β3 supplementation enhanced the healing of chronic wounds in diabetic mice with impaired SCs. Conclusion: Our study thus advances the understanding of the roles of neural-derived cells in skin regeneration and suggests a potential therapeutic strategy for wound healing disorders.

Keywords: Schwann cells; Single-cell transcriptome; TGF-β; skin; wound repair.

Figure 1

Single-cell analysis of mouse skin before and after injury. (A) T-distributed stochastic neighbor embedding (tSNE) plot for all samples with the major cell type populations denoted. FBIs, type 1 fibroblasts; FBIIs, type 2 fibroblasts; EBs, epidermal and basal cells; IMs, immune cells; VCs, vascular cells; NCs, neural crest derived cells; MIs, miscellaneous cells. (B) tSNE map with cells colored by time. (C) The expression levels of marker genes among cell types are shown in a violin plot. (D) The percentage of each cell type in UW skin and in WO skin at days 4, 8 and 14 (Wo4, Wo8, and Wo14). (E,F) Heatmap showing the major incoming or outgoing signals with the relative strength of each cell type in UW and WO skin. (G,H) River plot displaying the outgoing communication patterns of each cell type as well as the major signaling pathways in UW or WO skin. The thickness of the flow indicates the contribution of the cell type or signaling pathway to the corresponding pattern.

Figure 2

Identification of SCs in UW and WO skin. (A) UMAP analysis of NC transcriptomes with cells colored by time (left), Seurat cluster (middle) or cell type (right) identity. (B) Violin plots showing the expression levels of S100b and Plp1 in the subpopulations of NCs. (C) Pseudotime analyses trace putative SC differentiation trajectories with cells colored by State (left), time (middle) or pseudotime (right) identity. (D) Heatmap showing the dynamically changing genes of all the cells used for the Monocle analysis arranged based on their pseudotime values. (E) State identities and RNA velocity vectors projected onto pseudotime trajectories of SCs. Velocity vectors are displayed as arrows, with the predicted dedifferentiation paths highlighted in red. (F) Heatmap of the total number of putative interactions between cell types in UW skin and WO skin at Wo4, Wo8 and Wo14.

Figure 3

Characterization of SCs in normal and diabetic wounds. (A) Immunostaining of SCs labeled with S100b at D7. The higher-magnification image highlights the presence of Sox2⁺S100b⁺ dedifferentiated SCs (white arrowheads). Scale bars: 100 mm (left), 50 mm (right). (B) Immunostaining of Sox2 and Sox10 in regenerating skin at D7. Scale bars: 50 mm. (C) Representative images from immunohistochemical analysis of wounds from C57 control mice (C57) and mice with STZ-induced diabetes (C57-STZ) with Sox2 labeling at Wo7. Scale bar, 200 µm (up), 100 µm (down). (D) Quantification of Sox2⁺ cells in C57 or C57-STZ. (E) Immunostaining of Sox2 and S100b in skin sections from C57 and C57-STZ. Scale bar, 50 µm. (F) Quantification of Sox2⁺ cells among S100b⁺ cells in C57 or C57-STZ. (G) Immunostaining of Sox10 and S100b in wound slices from C57 and C57-STZ. Scale bar, 20 µm. (H) Quantification of Sox10⁺ S100b⁺ cells in C57 or C57-STZ. (I) Representative images of SOX10 immunostaining of diabetic skin in unwounded (diabetic non-wounded, DN) and wounded (diabetic ulcer, DU) regions and normal skin (NO). Scale bar, 50 µm. (J) Quantification of Sox10⁺ cells in unwounded and wounded regions of diabetic patients. The data are presented as the mean ± SEM, unpaired t test, *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

Inhibition of Wnt signaling at an early stage impacts wound healing and decreases SC dedifferentiation. (A) Heatmap showing the transcription factors with significantly changed expression among all the cells used for the Monocle analysis arranged by pseudotime values. (B) Immunostaining of Sox2 and Tcf4 in D7 wounds. Scale bar, 50 µm. (C) Heatmap showing the activity of the time-specific representative regulon at each time point. (D) Representative gross photographs of wounds treated with saline (Ctrl) or XAV939 (-Wnt) at D0 and D7. The inner diameter of the collar is 6 mm. (E) Quantification of individual wound areas at Wo7 in Ctrl and -Wnt animals relative to their initial sizes at D0. (F) Immunostaining of S100b and Sox2 in the skin of Ctrl and -Wnt wounds at D7. Scale bar, 100 µm. (G) Quantification of Sox2⁺ cells among S100b⁺ cells in Ctrl or -Wnt wounds. (H) Immunostaining of S100b and Sox10 in the skin of Ctrl and -Wnt wounds at D7. Scale bar, 20 µm. (I) Quantification of Sox10⁺ S100b⁺ cells in Ctrl or -Wnt wounds. (J) Representative images of skin sections with Masson staining from the denoted wounds at D14. Scale bar, 100 µm. The images of skin are representative of three independent experiments. The data are presented as the mean ± SEM, Unpaired t test, *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5

TGF-β inhibition in vivo at an early stage causes improper wound healing. (A) Heatmaps showing the pseudotime-dependent genes encoding ligands of all SCs across the pseudotime trajectory. (B) Immunostaining of TGF-β3 and Sox2 in the dermal region in D5 and D7 wounds. Scale bar, 50 µm. (C) Representative images of Sox10 and TGF-β3 immunostaining in the dermal region in D5 wounds. Scale bar, 20 µm. (D) Schematic (top lane) and gross images (bottom lane) of wounds treated with saline (Ctrl) or the TGF-β inhibitor SB431542 at D0 (-TGF-βe) and D7 (-TGF-βl). (E) Quantification of the relative wound size in the denoted wound types at D14. One-way ANOVA. (F) Hematoxylin and eosin (H&E) staining histology of Ctrl, -TGF-βe (-TGF-β early) and - TGF-βl (-TGF-β late) D7 wounds. Scale bar, 100 µm. (G) Quantification of the epidermal thickness of Ctrl, -TGF-βe and -TGF-βl wounds. One-way ANOVA. (H) Representative images of sections from Ctrl, -TGF-βe and -TGF-βl wounds immunostained for aSMA. The images are representative of three independent experiments. Scale bar, 100 µm. The data are presented as the mean ± SEM, *P <0.05, **P < 0.01, ***P < 0.001.

Figure 6

SCs promote the migration of fibroblasts and keratinocytes. (A) In vitro wound-healing assays of 3T3 cells cultured in mediums from 3T3, RSC96 or S16 cells. (B) Migration rates were evaluated according to the closure area at 18 h after lesion induction. (C) In vitro wound-healing assays of HaCaT cells cultured in mediums from HaCaT, RSC96 or S16 cells. (D) Quantification of wound closure at 18 h after scratching. (E) In vitro wound-healing assays of 3T3 cells cultured in medium from RSC96 or S16 cells after control or Tgfb3 RNAi (siCtrl or siTgfb3) treatment. (F) Quantification of wound closure of 3T3 cells under different conditions. (G) In vitro wound-healing assays of HaCaT cells cultured under the indicated conditions. (H) Quantification of wound closure in HaCaT cells under the indicated conditions. (I) Immunoblot analysis of the expression of aSMA, phosphorylated Smad2/3 (p-Smad2/3) and Smad2/3 in 3T3 or HaCaT cells after incubation with mediums from RSC96 or S16 cells for 24 h. The images presented are representative of three independent experiments. Scale bars: a, c, e, g, 200 mm. Unpaired t test. The data are presented as the mean ± SEM, *P<0.05, **P < 0.01, ***P < 0.001.

Figure 7

Analysis of the role of TGF-β3 in diabetic wounds. (A) Immunostaining of TGF-β3 in sections from C57 and C57-STZ wounds at D7. Scale bar, 100 mm. (B) Representative images of S100b and TGF-β3 immunostaining in the dermal region in D7 wounds. Scale bar, 50 µm. (C) Quantification of TGF-β3⁺ S100b⁺ cells in C57 or C57-STZ. (D) Immunostaining of TGF-β3 in intact skin from non-diabetic human (NO) and unwounded or wounded skin from diabetic patients (DN and DU). Scale bar, 100 mm. (E) Representative H&E staining of wounds from mice injected with control (LV-NC) or Tgfb3 RNAi (LV-shTgfb3) virus at D7. Scale bar, 1 mm. (F) Quantification of the wound width of LV-NC and LV-shTgfb3 wounds at D7. (G) Representative gross images of wounds from LV-NC or LV-shTgfb3 mice D0 and D14. (H) Quantification of individual wound areas at D14 in LV-NC and LV-shTgfb3 mice relative to their initial sizes at D0. (I) H&E staining of LV-NC and LV-shTgfb3 wounds at D14. Scale bar, 1 mm. (J,K) Quantification of the wound width (J) and epithelial thickness (K) of LV-NC and LV-shTgfb3 wounds at D14. (L) H&E staining of diabetic wounds with or without TGF-β3 injection (DM+T and DM wounds) at D7. Scale bar, 1 mm. (M,N) Quantification of the wound width (M) and epithelial thickness (N) of DM and DM+T wounds at D7. (O) Representative images of DM and DM+T wounds at D14. (P) Quantification of wound size for DM+T wounds versus DM wounds. (Q) Representative images of aSMA staining on sections from DM and DM+T wounds at D14. Scale bar, 100 µm. Unpaired t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 8

SCs facilitate wound healing by releasing TGF-β3. In response to injury, dermal SCs around the wound region undergo dedifferentiation with Wnt activation, contributing to wound healing by promoting migration of fibroblast and keratinocyte. In diabetic skin, dysfunctional SCs are failed to secrete TGF-β3, which could be a therapeutic strategy for chronic wound.