The Wound Microenvironment Reprograms Schwann Cells to Invasive Mesenchymal-like Cells to Drive Peripheral Nerve Regeneration

Abstract

Schwann cell dedifferentiation from a myelinating to a progenitor-like cell underlies the remarkable ability of peripheral nerves to regenerate following injury. However, the molecular identity of the differentiated and dedifferentiated states in vivo has been elusive. Here, we profiled Schwann cells acutely purified from intact nerves and from the wound and distal regions of severed nerves. Our analysis reveals novel facets of the dedifferentiation response, including acquisition of mesenchymal traits and a Myc module. Furthermore, wound and distal dedifferentiated Schwann cells constitute different populations, with wound cells displaying increased mesenchymal character induced by localized TGFβ signaling. TGFβ promotes invasion and crosstalks with Eph signaling via N-cadherin to drive collective migration of the Schwann cells across the wound. Consistently, Tgfbr2 deletion in Schwann cells resulted in misdirected and delayed reinnervation. Thus, the wound microenvironment is a key determinant of Schwann cell identity, and it promotes nerve repair through integration of multiple concerted signals. VIDEO ABSTRACT.

Keywords: Eph signaling; PNS regeneration; Schwann cell; TGFb signaling; dedifferentiation; plasticity.

Figure 1

RNA Sequencing of Differentiated and Progenitor-like SCs Identifies Novel Features of Dedifferentiation (A) Example of a regenerating sciatic nerve of tdTom^SC mice collected 6 days post-transection. A representative tiled fluorescent image of the nerve regions collected for FACS purification is shown. tdTomato⁺ SCs are in red, and DAPI-stained nuclei are in blue. Dotted lines demarcate the nerve bridge. (B) Representative FACS plots of the purification of tdTomato⁺ SCs from sciatic nerves of tdTom^SC mice (bottom). Control P0A-Cre⁻;tdTom^fl/fl cells (top) were used for gating. (C) RNA-seq differential gene expression analysis of SCs from pooled distal stump (dSCs) and intact nerve (iSCs). Genes regulated over 2.5× (adj. p < 0.05, light blue), myelination genes (red circles), SC markers (black circles), and dedifferentiation markers (dark blue circles) are highlighted. (D) Functional analysis of genes differentially regulated between dSCs and iSCs. Genes with dSC:iSC count ratios >2.5× (adj. p < 0.05) were selected. –log₁₀ of the enrichment p values for selected GO categories (BP: biological process, CC: cellular components) are plotted relative to Z scores of average dSC:iSC count ratios in each category. Circle size denotes the number of regulated genes. (E) GSEA enrichment plots of dSCs compared to iSCs for the “epithelial mesenchymal transition” and “Myc targets V1” MSigDB hallmarks. (F) Average log₂ FPKM expression of the ES core, PRC, and Myc modules from (Kim et al., 2010) for iSCs, pooled SCs isolated from the nerve bridge (bSCs), and dSCs, as well as differentiated cells of neural origin (Astrocytes, Neurons) and pluripotent ES cells (ES cells) (Zhang et al., 2014, Marks et al., 2012). FPKM have been median centered to allow comparison across datasets. (G) GSEA enrichment analysis of dSCs compared to iSCs (D–I) and bSCs compared to iSCs (B–I) for MSigDB gene sets related to pluripotency (lines 2–10), regulation by the Prc2 complex (11–15), regulation by Myc (16-18), three ES modules (lines 1, 11, and 16), and markers of neural crest identity (19–21). Numbers represent normalized enrichment score and colors represent the FDR q values ≤ 0.1 (red) or ≤ 0.25 (orange/blue). See also Figure S1.

Figure 2

Bridge and Distal SCs Have Distinct Molecular Signatures (A) Seven k-means clusters of dSC:iSC, bSC:iSC and bSC:dSC expression ratios. Colors represent the mean of log₂ expression ratios in each cluster and condition. Selected functional categories enriched in each cluster are shown. Cluster numbers from top to bottom: 2,6,3,4,7,1,5, as in Table S4. (B) GSEA enrichment analysis of preranked bSC:dSC ratios for MSigDB hallmarks. Gene sets with FDR q values <0.25 are plotted relative to normalized enrichment scores (NES). Categories with negative or positive NES are down- or upregulated, respectively, in bSCs. Circle size denotes the number of enriched genes in each category and circle colors represent FDR q values as indicated. (C) Representative EdU staining (green) of actively proliferating tdTomato⁺ SCs in the bridge and distal stump 6 days post-transection. (D) Quantification of EdU⁺/tdTomato⁺ SCs relative to total tdTomato⁺ SCs in the bridge and distal nerves at 4, 6, 8, and 10 days post-transection and in the intact nerve (mean ± SEM). n = 6 at day 6 and n = 3 at days 4, 8, and 10, ^∗∗p < 0.01 and ^∗∗∗p < 0.001, two-tailed paired Student’s t test. (E) Hierarchical clustering of log₂ expression ratios for genes belonging to the “epithelial mesenchymal transition” MSigDB hallmark. Genes significantly upregulated (adj. p < 0.05) in bSCs compared to iSCs are shown. Rows are dSC:iSC, bSC:iSC, and bSC:dSC log₂ ratios from top to bottom. (F–H) RNA-seq time course analysis of dSC and bSC gene expression changes after sciatic nerve transection. dSC:iSC (blue) or bSC:iSC (red) expression ratios are plotted relative to time after injury for the genes (F) upregulated in bSCs relative to dSCs from figure S2A, (G) downregulated in bSCs relative to dSCs from S2A, and (H) belonging to the EMT signature from (E). Boxes denote the interquartile range, and black strikes denote the median of ratios in the list. Open boxes represent dSC:iSC ratios for all genes as a reference. Data from cells isolated from single nerves are shown; n = 3–4 per time point. See also Figure S2.

Figure 3

TGFβ Signaling Is Increased in the Nerve Bridge (A) p-Smad3 expression in the proximal (top), bridge (middle), and distal (bottom) nerve 6 days post-transection. Paraffin sections were stained for p-Smad3 (DAB), the SC marker S100β (red), and DAPI (blue). White arrowheads indicate p-Smad3 colocalization with SC nuclei. (B) Quantification of p-Smad3-expressing SCs in each region (mean ± SEM). n = 9, ^∗∗∗p < 0.001, one-way ANOVA with Bonferroni correction. (C) GSEA enrichment plots of bSCs compared to dSCs and iSCs for the SC-specific transcriptional response to TGFβ (this study). TGFβ UP and TGFβ DN gene sets are up- or downregulated, respectively, by TGFβ. (D) RNA-seq time course analysis of dSC or bSC gene expression changes after sciatic nerve transection. dSC:iSC (distal:intact; blue) or bSC:iSC (bridge:intact; red) expression ratios are plotted relative to time after injury for the 30 TGFβ UP genes with the highest significant bSC:dSC ratios in the dataset in (C) (p < 0.05). Boxes denote the interquartile range, and black strikes denote the median of ratios in the list. Open boxes represent dSC:iSC expression ratios for all genes as a reference. Data from cells isolated from single nerves are shown; n = 3–4 per time point. See also Figure S3.

Figure 4

TGFβ Drives SC Invasion into the Bridge (A) Representative immunofluorescence staining for the SC marker S100β (green) in longitudinal sections of regenerating proximal stumps of Tgfbr2^ΔSC mice and P0A-Cre⁻ control littermates Tgfbr2^fl/fl 6 days post-transection. Nuclei are stained with DAPI. Dashed lines mark the border between the proximal stump and the nerve bridge. (B) Average length of the SC cords from the Tgfbr2^ΔSC and Tgfbr2^fl/fl cut nerves from (A) (mean ± SEM). n = 6 for Tgfbr2^fl/fl, n = 9 for Tgfbr2^ΔSC, ^∗∗p < 0.01, two-tailed paired Student’s t test. (C) Maximal migrated distance of the SC cords from Tgfbr2^ΔSC and Tgfbr2^fl/fl cut nerves from (A) (mean ± SEM). n = 6 for Tgfbr2^fl/fl, n = 9 for Tgfbr2^ΔSC, ^∗∗p < 0.01, two-tailed paired Student’s t test. (D) qRT-PCR analysis of the bridge-specific mesenchymal gene signature in bSCs and dSCs FACS-sorted from single nerves of tdTom;Tgfbr2^fl/fl and tdTom;Tgfbr2^ΔSC mice. Boxplots represent log₂ bSC:dSC expression ratios of 24 bridge-specific EMT genes. Pooled data for Tgfrb2^fl/fl (blue bar, n = 6) and Tgfbr2^ΔSC (purple bar, n = 5) mice are shown; p_Wilcoxon = 0.00024. The whiskers extend to the most extreme data point, which is no more than 1.5× the interquartile range from the box. (E) SC invasion through fibronectin-coated Boyden chambers in the absence or presence of TGFβ. Rat SCs (rSC, solid bars) and mouse SCs (mSC, hatched bars) were treated with vehicle (blue bar) or TGFβ (purple bar, 10 ng/mL) (mean ± SEM). n = 3, ^∗p < 0.05, ^∗∗∗p < 0.001, one-way ANOVA. (F) SC invasion through fibronectin-coated Boyden chambers upon inhibition of TGFR signaling. Rat SCs (rSC) were left untreated or treated with TGFR inhibitors LY2157299 (LY; purple hatched bar) or SD208 (SD; purple spotted bar) for 24 hr prior to the assay. Mouse SCs (mSC) were either wild-type (Tgfbr2^fl/fl, blue vertical lines) or knockout (Tgfbr2^−/−; purple vertical lines) (mean ± SEM). n = 3. (G) EdU incorporation in the bridges of tdTom;Tgfbr2^fl/fl (blue bar, n = 5) and tdTom;Tgfbr2^ΔSC (purple bar, n = 4) mice 6 days post-transection. EdU incorporation in tdTom;Tgfbr2^ΔSC is normalized to controls (mean ± SEM). (H) EdU incorporation in rat SCs treated with vehicle (Control, blue bar) or TGFβ (TGFβ, purple bar) for 16 hr in vitro (mean ± SEM). n = 3, p = 0.14, two-tailed paired Student’s t test. (I) Representative immunofluorescence staining for the axonal marker neurofilament (red) in longitudinal sections of regenerating proximal stumps of Tgfbr2^ΔSC mice and P0A-Cre⁻ control littermates (Tgfbr2^fl/fl) 6 days post-transection (Top). Dashed lines mark the border between the proximal stump and the bridge. The bottom panels show axonal tracings of images from the top panels obtained in NeuronJ. (J) Angles of axonal regrowth relative to the long axis of the nerves. Means ± SEM of the percentage of axons at angles >45° per group are shown. n = 6 for Tgfbr2^fl/fl and n = 13 for Tgfbr2^ΔSC, ^∗∗∗p < 0.001, two-tailed paired Student’s t test. (K) Analysis of distal stump reinnervation 6 weeks post-transection in Tgfbr2^fl/fl and Tgfbr2^ΔSC mice. Representative images of neurofilament staining in distal nerves are shown. (L) Quantification of neurofilament staining in the distal nerves of Tgfbr2^fl/fl and Tgfbr2^ΔSC mice shown in (K). Axon density was measured by fluorescence quantification in ImageJ. Data are normalized to Tgfbr2^fl/fl controls (blue bar) ± SEM n = 4 for each set, ^∗∗p < 0.01, two-tailed paired Student’s t test. See also Figure S4.

Figure 5

TGFβ Crosstalks with EphB2 to Modulate SC Sorting (A) Fluorescence images of rat SCs stained for S100β (green) plated on Fc control or ephrin-B2-Fc ligand in the presence or absence of TGFβ. (B) Quantification of cell sorting shown in (A) (mean ± SEM). n = 3, ^∗∗p < 0.01; ^∗∗∗p < 0.001, Fisher’s exact test. (C) Quantification of cell sorting as in (B) of rat SCs plated on Fc or ephrin-B2-Fc in the absence or presence of TGFR inhibitors LY2157299 (LY) or SD208 (SD). Cell sorting data is represented as mean ± SEM n = 3, ^∗∗∗p < 0.001, Fisher’s exact test. (D) Quantification of cell sorting as in (B) for wild-type (Tgfbr2^fl/fl) and knockout (Tgfbr2^−/−) mouse SCs plated on Fc or ephrin-B2-Fc (mean ± SEM). n = 3, ^∗p < 0.05; ^∗∗∗p < 0.001, Fisher’s exact test. (E) Representative fluorescence images of rat SCs treated with Scr siRNA or siRNA against Smad4 (siSmad4) and cultured on Fc or ephrin-B2-Fc recombinant proteins. (F) Quantification of cell sorting from (E) (mean ± SEM). n = 3, ^∗∗∗p < 0.001, Fisher’s exact test. See also Figure S5.

Figure 6

The TGFβ/EphB2 Crosstalk Is Mediated by N-Cadherin (A) Western blot analysis of p-EphB2 and total EphB2 levels in rat SCs seeded on Fc control or ephrin-B2-Fc ligands (eB2-Fc) in the presence of TGFβ or with ephrin-B2-Fc and TGFβ combined for 16 hr. β-actin is used as a loading control. (B) Western blot analysis of p-EphB2 and total EphB2 levels in rat SCs treated with Scr siRNA or siRNA against Smad4 (siSmad4) and seeded on Fc control or ephrin-B2-Fc (eB2-Fc) ligands for 16 hr. Smad4 indicates knock down efficiency and β-actin loading. (C) Densitometric quantification of western blots shown in (A) and (B) obtained in FiJi (mean ± SEM), n=3. (D) Representative confocal images of rat SCs treated with the indicated ligands for 16 hr and stained for S100β (green) and N-cadherin (red). (E) Quantification of N-cadherin fluorescence at cell-cell junctions in the conditions shown in (D). Bar graphs show pixel intensities normalized by cell number (mean ± SEM). n = 3, ^∗∗p < 0.01; ^∗∗∗p < 0.001, two-tailed paired Student’s t test. (F) Quantification of N-cadherin fluorescence at cell-cell junctions in SCs treated with Fc (−) or ephrin-B2-Fc (+) in the absence (Ctr) or presence of LY2157299 (LY) and SD208 (SD) TGFR inhibitors. Means ± SEM are shown; n = 3, ^∗∗∗p < 0.001, one-way ANOVA. (G) Quantification of N-cadherin fluorescence at cell-cell junctions in Scr and Smad4 knockdown SCs treated with Fc (−) or ephrin-B2-Fc (+). Means ± SEM are shown; n = 3, ^∗∗∗p < 0. 001, one-way ANOVA. (H) Western blot analysis of actin-bound and soluble N-cadherin in SCs cultured on Fc (−) or ephrin-B2-Fc (+) in the absence (Ctl) or presence of LY2157299 (LY) and SD208 (SD) TGFR inhibitors. β-actin is shown for loading. (I) Western blot analysis of total N-cadherin in Scr and Smad4-knockdown SCs plated on Fc and ephrin-B2-Fc in the presence or absence of TGFβ (TGFβ; eB2-TGFβ). Smad4 indicates knockdown efficiency and β-actin loading. (J) Quantitative RT-PCR analysis of Cdh2 levels in SCs treated with Fc, TGFβ, ephrin-B2-Fc, and TGFβ and ephrin-B2-Fc combined, as indicated. Fold changes relative to Fc-treated controls are shown (mean ± SEM). n = 6, ^∗∗p < 0.01, two-tailed paired Student’s t test. (K) qRT-PCR analysis of Cdh2 levels in SCs cultured in the absence (Control) or presence of LY2157299 (LY) and SD208 TGFR inhibitors, as in (J). Means ± SEM are shown. n = 8, ^∗∗∗p < 0.001, two-tailed paired Student’s t test. (L) Quantitative RT-PCR analysis of Cdh2 levels in SCs treated with Scr siRNA or siRNA against Smad4 (siSmad4). Means ± SEM are shown. n = 4, ^∗p < 0.05, two-tailed paired Student’s t test.

Figure 7

An Increase in N-Cadherin Levels Mediates TGFβ Effects on EphB2-Dependent Cell Sorting In Vitro and In Vivo (A) Quantification of cell sorting in Scr or partial Cdh2 knockdown in rat SC cultures treated with Fc, TGFβ, ephrin-B2, and TGFβ and ephrin-B2 combined, as indicated. Cell sorting data is represented as mean ± SEM. n = 3, ^∗∗p < 0.01; ^∗∗∗p < 0.001, Fisher’s exact test. (B) Quantification of cell sorting in rat SCs transduced with adenoviruses encoding GFP (Ad-Gfp) or Cdh2 (Ad-Cdh2) and treated with Fc, TGFβ, ephrin-B2-Fc, or TGFβ and ephrin-B2-Fc combined, as indicated (mean ± SEM). n = 3, ^∗∗p < 0.01; ^∗∗∗p < 0.001, Fisher’s exact test. (C) Quantification of cell sorting in rat SCs transduced with adenoviruses encoding GFP (Ad-Gfp) or Cdh2 (Ad-Cdh2) and treated with Fc or ephrin-B2-Fc in the absence or presence of LY2157299 (LY) and SD208 (SD) inhibitors, as indicated (mean ± SEM). n = 3, ^∗∗∗p < 0.001, Fisher’s exact test. (D) In vivo cell sorting in tdTom;Tgfbr2^fl/fl and tdTom;Tgfbr2^ΔSC in the bridge. Representative images of SC cords invading the bridge in tdTom;Tgfbr2^fl/fl and tdTom;Tgfbr2^ΔSC at 6 days post-transection. (E) Quantification of the in vivo cell sorting shown in panel D. Numbers of SCs in cords proximal to the migration front in tdTom;Tgfbr2^fl/fl and tdTom;Tgfbr2^ΔSC bridges are shown (mean ± SEM). n = 6 for each genotype, ^∗∗p < 0.01, Fisher’s exact test. (F) Boxplot of RNA-seq FPKM expression scores for Cdh2 in dSCs and bSCs. Colored dots represent single biological repeats. The whiskers extend to the most extreme data point, which is no more than 1.5× the interquartile range from the box. (G) qRT-PCR analysis of Cdh2 levels in bSCs and dSCs from single nerves 6 days post-transection from Tgfbr2^fl/fl (blue bars, n = 4–5) and Tgfbr2^ΔSC mice (purple bars, n = 3–5). The whiskers extend to the most extreme data point, which is no more than 1.5× the interquartile range from the box. (H) Representative fluorescence images of N-cadherin protein levels in longitudinal sections of nerve bridges from tdTom;Tgfbr2^fl/fl (top) and tdTom;Tgfbr2^ΔSC (bottom) mice 6 days post-transection. P marks the proximal side and the arrow marks the direction of SC invasion. See also Figure S7.