Abstract Peripheral nerve injuries, particularly those affecting the sciatic nerve, often result in incomplete functional recovery due to the limited regenerative capacity of adult peripheral nerves. To elucidate the cellular and molecular mechanisms underlying nerve regeneration, we performed single−cell RNA sequencing (scRNA−seq) on rat sciatic nerve tissues at seven time points (Days 0, 1, 3, 5, 7, 10, and 14) following transection injury. Through unsupervised clustering, we identified four major cellular compartments—neurofibroblasts (NFs), glial cells (Glis), immune cells, and vascular cells—and delineated their dynamic trajectories during regeneration. Early responses were dominated by macrophage (Mac) and granulocyte infiltration (Day 1), followed by proliferative expansion of proliferating mesenchymal fibroblasts (NF5) and repair Schwann cells (Gli0) by Days 3–5. Vascular remodeling commenced from Day 7, while Glis progressively transitioned to mature myelinating states (Gli2/Gli5) by Day 14. Pseudotime analysis revealed subtype−specific reprogramming in both Macs and Glis, and cell–cell communication analysis uncovered key ligand–receptor interactions—particularly collagen and PTN signaling between Macs, NFs, and Glis. Bulk transcriptomic validation confirmed sustained and spatially distinct activation of the TGF− Inline graphic signaling pathway across cell types and anatomical locations. Comparative analysis with a sciatic nerve crush injury model revealed a stronger early immune response and delayed Gli recovery in transection injury, indicating a narrowed therapeutic window. Together, this work provides a time−resolved single−cell atlas of peripheral nerve regeneration, defines key regulatory circuits within the immune–NF–Gli axis, and identifies phase−specific therapeutic targets—such as early Mac heterogeneity, NF4−mediated matrix remodeling, and Schwann cell remyelination—for enhancing functional recovery following severe nerve injury. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-025-03514-3. Keywords: Sciatic nerve transection, Single−cell RNA−seq, Glial cell reprogramming, Macrophage heterogeneity, Neurofibroblast–glia interaction, TGF−β signaling, Nerve regeneration Introduction Peripheral nerve injuries, particularly those involving the sciatic nerve, pose significant clinical challenges with profound socioeconomic impacts. Globally, sciatic nerve trauma has an incidence of 7.7% in specific high−risk populations (e.g., acetabular fracture patients), among whom iatrogenic injuries account for 12.87% of postoperative complications [[62]1]. The primary etiologies include traumatic accidents [[63]1, [64]2], surgical interventions [[65]1], and metabolic disorders such as diabetic neuropathy [[66]3]. Despite advances in microsurgical techniques, functional recovery rates remain suboptimal, with only 50% of patients achieving clinically meaningful recovery, while persistent motor deficits and chronic neuropathic pain—strongly correlated with reduced quality of life in physical functioning and social participation domains [[67]4, [68]5]—continue to burden survivors. This incomplete recovery underscores the urgent need to dissect the multicellular dynamics that underlie nerve repair—particularly the role of macrophage (Mac) subpopulations and transcriptional reprogramming in axonal regeneration—as revealed by recent single−cell RNA sequencing (scRNA−seq) studies [[69]6]. The primary therapeutic strategies for sciatic nerve injury, such as nerve grafting and neurotrophic factor administration, remain limited by incomplete axonal regeneration and misdirected reinnervation [[70]7]. These limitations arise primarily due to the inability of traditional bulk RNA sequencing approaches to resolve the spatiotemporal coordination among Schwann cells, fibroblasts, immune cells, and vascular cells during regeneration. By averaging signals across heterogeneous cell populations, bulk RNA sequencing obscures critical cell type−specific responses [[71]8]. Schwann cells exhibit dynamic phenotypic transitions between dedifferentiated, pro−regenerative, and repair states [[72]9, [73]10], while Macs undergo polarization from pro−inflammatory to pro−repair subtypes [[74]11]—processes that bulk RNA sequencing fails to resolve. Emerging scRNA−seq technologies now enable unprecedented cellular−resolution dissection of neural repair mechanisms. In central nervous system injury models—such as ischemic stroke and spinal cord injury—scRNA−seq has revealed transitional cellular states and intercellular crosstalk networks that dictate regenerative outcomes [[75]12–[76]14]. More recently, in peripheral nerve injury models, scRNA−seq has been employed to unravel the heterogeneity and functional states of Schwann cells, fibroblasts, immune subsets, and vascular cells. For instance, distinct transcriptional alterations have been identified in both myelinating and non−myelinating Schwann cells under autoimmune conditions, providing insights into glial plasticity [[77]15]. Endothelial Plexin−D1 has been shown to play dual roles in peripheral nerve repair by not only guiding the directional growth of endothelial cells (ECs) but also regulating angiogenic patterning [[78]16]. In parallel, studies using Aire−deficient mouse models demonstrated that T cell–derived IFN− Inline graphic induces Mac TNF− Inline graphic expression, thereby driving Mac phenotype switching and amplifying inflammatory responses [[79]17]. Furthermore, in trauma−induced heterotopic ossification, scRNA−seq revealed a high degree of spatial colocalization between peripheral nerves and blood vessels [[80]18], implicating coordinated neurovascular remodeling. Collectively, these findings underscore the power of scRNA−seq to disclose cellular plasticity, spatial interactions, and regulatory networks that remain obscured in conventional bulk transcriptomic analysis. Critical knowledge gaps persist regarding the spatiotemporal regulation of cellular responses during sciatic nerve regeneration. Despite recent advances identifying transcription factors (e.g., Sox10 [[81]19] and Zeb2 [[82]20]) and signaling pathways (e.g., Neuregulin−1/ErbB [[83]21] and Wnt/ Inline graphic −catenin [[84]22]) as key regulators of Schwann cell development, the molecular mechanisms governing Schwann cell fate decisions—particularly the transition into repair−promoting phenotypes following nerve injury—remain incompletely understood.​ Immune cell dynamics, including the recruitment of bone marrow−derived Macs and neutrophils, exhibit temporal specificity—Macs peak at Day 3 post−injury, while neutrophils surge within 24 h—a pattern consistent with the time−dependent immune activation originally observed in the sciatic nerve crush injury model [[85]6]. Vascular ECs in the sciatic nerve demonstrate distinct subtypes, including epineurial, endoneurial, and lymphatic endothelial cells (lyECs)—each characterized by unique gene expression profiles. Marker genes such as Spock2, Rgcc, and Lrg1 have been validated in vivo for the identification of these subtypes, offering improved specificity over classical pan−endothelial markers like Pecam1 [[86]23]. To address the unresolved complexity of peripheral nerve regeneration, we conducted a longitudinal scRNA−seq study of rat sciatic nerves at multiple time points following transection injury. By mapping the dynamic transitions of major cellular compartments—including glial cells (Glis), neurofibroblasts (NFs), immune cells, and vascular populations—we systematically characterized the cellular choreography underlying nerve repair. Particular emphasis was placed on dissecting how Glis influence NFs behavior and how intercellular communication evolves across regenerative phases. Pseudotime trajectory analyses were used to reconstruct lineage transitions, notably the differentiation of monocytes (Mos) into functionally distinct Mac subsets and the reprogramming of Schwann cells from repair to remyelinating phenotypes. To further validate the cell type–specific signaling changes observed at the single−cell level, we performed bulk RNA−seq on anatomically defined nerve segments and demonstrated sustained, spatially distinct activation of the transforming growth factor beta (TGF− Inline graphic ) signaling pathway across the repair timeline. Moreover, by comparing cellular and molecular responses between transection and crush injury models, we uncovered both shared and injury−specific regenerative strategies. Together, our study presents a comprehensive, time−resolved cellular and transcriptomic atlas of sciatic nerve regeneration and identifies phase−specific signaling programs and cell–cell interactions—particularly involving TGF− Inline graphic and NF–Gli crosstalk—that may inform targeted therapeutic interventions for peripheral nerve repair. Methods Sciatic nerve transection model Adult female Sprague–Dawley rats (8–10 weeks old, 220–250 g) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All experimental procedures involving animals were approved by the Institutional Animal Care and Use Committee of PLA General Hospital (approval number: 2023‑x4‑19) and conducted in accordance with national and international guidelines, including those from the International Council for Laboratory Animal Science. All efforts were made to minimize animal suffering and reduce the number of animals used. Rats were anesthetized by intraperitoneal injection of sodium pentobarbital (30 mg/kg). Under sterile conditions, the right sciatic nerve was exposed via a mid‑thigh skin incision, freed from surrounding tissue, and sharply transected approximately 10 mm proximal to its trifurcation. The muscle fascia and skin were then closed in two layers with 4‑0 absorbable sutures. After recovery from anesthesia, animals were housed in standard plastic cages with a 12‑hour light/12‑hour dark cycle and free access to food and water. Five rats were used per time point. Reference sciatic nerve crush model (Zhao et al., 2022) The sciatic nerve crush model and corresponding data used in this study were originally established and generated by Zhao et al. (2022) [[87]6]. Although we did not perform the sciatic nerve crush procedure ourselves, here we briefly summarize their methodology for clarity. In their study, adult C57BL/6 mice (8–16 weeks old) were anesthetized and the right sciatic nerve exposed via a dorsal thigh incision. The nerve was crushed approximately 5 mm distal to the piriformis muscle using fine forceps (Dumont #55) for 15 s under stable pressure, producing axonal injury while preserving the epineurium. Sham controls underwent nerve exposure without crush. Postoperative analgesia and tissue collection procedures followed standard protocols as detailed by Zhao et al. (2022). Tissue collection and single−cell dissociation At each designated post‑injury time point (Day 1, 3, 5, 7, 10, and 14), five rats were re‑anesthetized and the original incision reopened (Additional file 1: Fig. [88]S1a). For Day 0 (immediately after transection), a ± 2 mm segment proximal and distal to the transection site was harvested. For Day 1–3 (before nerve‑bridge formation), a 2 mm segment was collected from each stump (proximal and distal to the transection site). For Day 5–14 (after nerve‑bridge formation), a ± 2 mm segment of the newly formed inter-stump nervebridge was harvested. Tissues from the five animals at each time point were pooled,immediately placed in ice-cold Hank’s Balanced Salt Solution (HBSS), and enzymaticallydigested in collagenase IV (1 mg/mL) and DNase I (50 U/mL) at 37 °C for 30 min. Theresulting cell suspension was filtered through a 40 μm cell strainer, centrifuged at 300 × gfor 5 min, and resuspended in phosphate−buffered saline (PBS) containing 0.04% bovineserum albumin (BSA) for subsequent scRNA−seq library preparation. scRNA−seq and preprocessing scRNA−seq was performed using the Chromium Single Cell Gene Expression Solution (10x Genomics), which enables the isolation and labeling of 500−10,000 individual cells. This technology is based on the GemCode microfluidics platform, where barcoded gel beads and single cells are encapsulated within oil droplets (Gel Bead−In − EMulsions, GEMs). Within each GEM, gel beads dissolve, and cells undergo lysis, releasing mRNA that is reverse−transcribed into barcoded cDNA. After breaking the emulsion, cDNA was amplified via PCR, followed by quality control to assess fragment size and yield. The amplified cDNA was then fragmented to 200–300 bp, end−repaired, A−tailed, and ligated with sequencing adapters before undergoing index PCR amplification. Library quality was validated before sequencing on the Illumina NovaSeq 6000 platform to obtain high−throughput single−cell gene expression data. Raw sequencing reads were processed using the Cell Ranger pipeline (10x Genomics), including read alignment to the rat genome (Rnor_6.0), barcode assignment, and unique molecular identifier (UMI) counting. Low−quality cells with high mitochondrial gene content (> 10%) or low total UMI counts (< 500) were removed. Doublet detection was performed using DoubletFinder, and doublets were excluded from downstream analysis. Tissue collection for bulk RNA−seq To spatially profile transcriptomic changes during nerve regeneration, we collected 2−mm segments from both the proximal and distal stumps of the transected sciatic nerve at designated time points (Days 1, 3, 5, 7, and 14), along with uninjured control samples. For each time point, tissues from 3 rats were pooled for each anatomical region. Samples were immediately snap−frozen in liquid nitrogen and stored at − 80 °C until RNA extraction. Bulk RNA−seq library preparation and sequencing Total RNA was extracted from tissue samples using a standard protocol, and RNA integrity was assessed using the Agilent 2100 Bioanalyzer. Samples with sufficient yield and RNA integrity were used for downstream bulk RNA−seq analysis. mRNA was enriched using oligo(dT) magnetic beads to capture polyadenylated transcripts or, alternatively, rRNA was depleted to retain a broader transcript profile. The enriched RNA was fragmented and reverse−transcribed to cDNA. Libraries were prepared using either standard or strand−specific protocols, the latter incorporating dUTP during second−strand synthesis to retain strand information [[89]24]. After end repair, A−tailing, adaptor ligation, size selection, and PCR amplification, libraries were sequenced on an Illumina platform with paired−end 150 bp reads. Clustering and cell type annotation Dimensionality reduction was performed using principal component analysis (PCA) on the top 3000 variable genes. The first 30 principal components were used for uniform manifold approximation and projection (UMAP) visualization. Unsupervised clustering was performed using the Seurat package [[90]25] (version 5.1.0) with the Louvain algorithm at a resolution of 0.8. Cell types were annotated based on the expression of established marker genes: NFs (Col1a1, Dcn, Col3a1), Glis (Mpz, S100b, Mag), immune cells (Aif1, Cd68, Cd3e), and vascular cells (Vtn, Esam, Plvap). Further subclustering within major cell populations was performed to identify distinct cellular subtypes. Bioinformatic analysis and statistics We employed Seurat for downstream analysis, following a structured pipeline to process and analyse the scRNA−seq data. The analysis was conducted in seven key steps: 1. Normalization: Raw gene expression counts were normalized on a per − cell basis using log−normalization (log1p transformation), in which the natural logarithm of 1 plus the counts per 10,000 was computed. This step ensures that expression levels are comparable across cells and suitable for downstream analyses. 2. Highly Variable Gene Selection and Batch Effect Correction: The top 3,000 highly variable genes were identified using the FindVariableFeatures function, capturing genes with the greatest variability across the dataset and likely reflecting biologically meaningful signals. To mitigate batch effects, integration anchors were first identified using the FindIntegrationAnchors function and then used to integrate the datasets with IntegrateData. 3. Data Scaling: Gene expression values were standardized across all cells using Z−score transformation via the ScaleData function. This step adjusts for differences in average gene expression levels, facilitating cross−cell comparisons and downstream dimensionality reduction. 4. PCA: PCA was performed on the scaled expression matrix of the highly variable genes to reduce dimensionality and capture the primary axes of variation in the dataset. 5. UMAP Visualization: UMAP was used to project the high−dimensional data into a two−dimensional space for visualization. The RunUMAP function in Seurat was executed using principal components 1 through 15 (dims = 1:15), as determined from the PCA on the subsetted data. 6. Clustering: Cell clustering was carried out using a graph−based approach with the Louvain algorithm, as implemented in the FindClusters function. A shared nearest neighbor (SNN) graph was first constructed using FindNeighbors, and clustering was performed with a resolution parameter of 0.3 to define discrete cell populations within the sciatic nerve dataset. 7. Differential Gene Expression (DGE) Analysis: To identify differentially expressed genes (DEGs) for each cluster, the Wilcoxon rank−sum test was applied using the FindMarkers function. The analysis was conducted on the log−normalized expression matrix, with min.pct = 0.25 set to include genes expressed in at least 25% of cells in either group. All other parameters were kept at their default values, including only.pos = TRUE, logfc.threshold = 0.1, and max.cells.per.ident = Inf. This strategy enabled robust identification of cluster−specific marker genes while maintaining sensitivity to subtle expression differences. Cell type annotation For unsupervised cell type annotation, we utilized the SingleR package [[91]26] (version 2.8.0) with the crush sciatic nerve injury single−cell dataset ([92]GSE198582 [[93]6]) as a reference. Cell type assignments were further manually validated by examining the DEGs for the presence of canonical marker genes for each cell type. Based on this analysis, we assigned metacells to various cell types, including NF, Gli, PC, SMC, Mac, Gran, DC, and T /NK/B cell. Subclustering and further resolution To achieve higher resolution of cell states, subclustering was performed on major cell types by first subsetting specific populations (e.g., NFs) from the integrated dataset. For each subset, data normalization and scaling were repeated using the SCTransform function to ensure consistency in preprocessing. Dimensionality reduction was carried out via PCA, and the appropriate number of principal components (PCs) was determined using ElbowPlot. The selected PCs (typically 1–10/15) were then used for constructing a SNN graph and performing graph−based clustering using the FindNeighbors and FindClusters functions, with an appropriate resolution to capture finer subpopulations. Low−dimensional embeddings were generated using UMAP based on the selected PCs. DEGs between subclusters were identified using the FindMarkers function with the parameter min.pct = 0.25, while other parameters remained at their default settings. DEGs were ranked by log2 fold−change values to identify representative marker genes. For visualization, z−scores were computed based on the average expression levels of each gene across subclusters, and the resulting matrix was visualized using heatmaps, enabling clear comparison of transcriptional profiles and aiding in the refinement of subpopulation annotations. DGE and enrichment analysis DGE analysis was performed using the FindMarkers function in Seurat. For identifying marker genes of general cell subtypes, genes were considered significant if they were both upregulated in the target cluster and had a p−value < 0.05. This dual criterion ensured that selected genes were not only statistically significant but also biologically relevant in distinguishing cell populations. This analysis enabled the classification of distinct cellular subtypes and facilitated the identification of key regulatory genes within each lineage. For the comparative analysis between crush injury and transection injury, a stricter threshold was applied to identify DEGs that distinguished the two injury models. After using FindMarkers, genes were considered significantly different if they met the criteria of fold−change > 1 and adjusted p−value < 0.01. These DEGs were subsequently subjected to Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis using clusterProfiler [[94]27] to identify biological pathways that were commonly activated in both injury types or uniquely enriched in either crush or transection injuries. This approach enabled a detailed characterization of both shared and distinct molecular mechanisms underlying nerve repair in different injury models. In addition to single−cell−based DGE analysis, we performed gene set enrichment analysis (GSEA) on bulk RNA−seq data to assess pathway activation dynamics at multiple anatomical locations and time points. Gene sets were referenced from the 2024 mouse version of the MSigDB database, and enrichment was calculated based on ranked gene expression profiles. Particular focus was placed on TGF− Inline graphic signaling–related pathways, whose activation patterns were compared across injury sites (e.g., proximal stump, distal stump) and different post−injury days. This bulk transcriptomic analysis confirmed the temporal regulation of TGF− Inline graphic signaling during nerve regeneration and provided orthogonal validation for the findings derived from single−cell data. Pseudotime and trajectory analysis To investigate dynamic cellular transitions during nerve repair, pseudotime trajectory analysis was performed using Monocle [[95]28] for Glis and slingshot [[96]29] for Mo to Mac differentiation. For Glis, Monocle was used to model the lineage trajectory from repair Schwann cells (Gli0) to myelinating Schwann cells (Gli5). After filtering out Day 0 samples, cells from Day 5 to Day 14 were selected for analysis. Cells were re−embedded based on highly variable genes, and a subset of genes expressed in at least 350 cells was retained. To define ordering genes for trajectory inference, we performed differential expression analysis across timepoints using differentialGeneTest, and genes with p < 1e^− 15 were selected. The top pseudotime−associated genes were then analyzed for expression trends using plot_pseudotime_heatmap, and k−means clustering (k = 3) was applied to identify co−expression modules. These clusters were interpreted as representing early−stage regulatory genes, intermediate−phase modulators, and late−stage effectors, corresponding to successive waves of gene activation during Gli reprogramming and remyelination. To explore the biological significance of these gene modules, we conducted GO enrichment analysis for each gene cluster using clusterProfiler [[97]27], revealing functional themes associated with glial plasticity, differentiation, and remyelination. For the immune compartment, slingshot was used to reconstruct lineage relationships between infiltrating Mos and differentiated Mac subtypes. The trajectory was inferred based on UMAP embeddings, with pseudotime values extracted along inferred lineages. For downstream analysis, cells with valid pseudotime values were filtered, and a generalized additive model (GAM) was fitted to gene expression using the tradeSeq framework. Genes exhibiting significant changes along the trajectory were identified using associationTest and startVsEndTest, highlighting transcriptional programs that govern Mo to Mac fate transitions. Top−ranked genes visualized with plotSmoothers were prioritized for further interpretation. Together, these analyses provided comprehensive insights into the temporal orchestration of glial reprogramming and immune differentiation during peripheral nerve regeneration. Cell−cell communication analysis Cell−cell communication analysis was performed to investigate intercellular signaling dynamics during sciatic nerve repair. The CellChat package [[98]30] was used to infer ligand−receptor interactions among different cell populations. Normalized single−cell RNA−seq data were used as input, and the analysis was conducted separately for different time points to track temporal changes in signaling activity. First, the expression of known ligand−receptor pairs was assessed across all major cell types, including NFs, Glis, immune cells (Macs, Mos, Grans, T cells, B cells, NK cells, and DCs), and vascular cells (PCs, ECs, SMCs, and lyECs). Communication networks were constructed based on the strength and specificity of ligand−receptor interactions. The computeCommunProb function was applied to calculate interaction probabilities, followed by computeCommunProbPathway to identify active signaling pathways. To visualize signaling patterns, network diagrams and heatmaps were generated to depict interactions between cell types. The netVisual_circle function was used to illustrate global intercellular communication, while netVisual_bubble provided insights into specific ligand−receptor pairs. Pathway−level analysis was conducted to identify key signaling cascades, with a focus on those implicated in nerve regeneration, including collagen−related signaling, PTN (Pleiotrophin) signaling between Glis and NFs, and Mac−mediated extracellular matrix remodeling. The strength and directionality of communication were assessed by examining changes in outgoing and incoming signaling patterns for specific cell populations over time. Statistical analyses were corrected for multiple comparisons using the Benjamini−Hochberg method to control the false discovery rate. Comparative analysis of crush injury and transection injury models To compare the cellular and molecular responses between crush injury and transection injury models, we conducted a detailed comparative analysis using scRNA−seq data. The scRNA−seq data from both injury models were integrated using the FindIntegrationAnchors function in Seurat, which identifies mutual nearest neighbors (anchors) between datasets to correct for batch effects and technical variations while preserving biological differences. Following anchor identification, the datasets were harmonized using the IntegrateData function, generating a combined dataset with minimized batch effects. The integrated dataset was log−normalized using the natural log1p normalization, and the 3,000 most variable genes were identified using the FindVariableFeatures function. Expression values were standardized across cells using Z−score transformation, and PCA was performed on the scaled variable gene matrix. Clustering was performed using the Louvain algorithm implemented in the FindClusters function, with a resolution setting of 0.7 to identify distinct cell populations. Cell type annotations were based on the classification derived from the crush injury dataset, which included various functional subtypes such as fibroblasts (e.g., proliferating, differentiating, and matrix−stabilizing subtypes), Schwann cells (e.g., proliferating, repairing, and myelinating subtypes), Macs (e.g., pro−inflammatory, pro−repair, and proliferating subtypes), Mo, Gran, T cells, B cells, ECs, and PCs. These annotations were manually validated by examining the expression of canonical marker genes for each cell type. DGE analysis was conducted using the FindMarkers function, with genes considered significantly differentially expressed if they exhibited a fold change > 1 and an adjusted p−value < 0.01. GO and KEGG pathway enrichment analysis were conducted using the clusterProfiler package, with pathways considered significantly enriched if they had an adjusted p−value < 0.05. This comprehensive comparative analysis provided insights into the distinct cellular and molecular mechanisms underlying nerve repair in crush injury and transection injury models, highlighting potential therapeutic targets for enhancing nerve regeneration in different injury contexts. Immunofluorescence staining of sciatic nerve sections At each designated post−injury time point (Day 0, 1, 3, 5, 7, 10, and 14), rats were deeply anesthetized with sodium pentobarbital (50 mg/kg, intraperitoneally) and euthanized by overdose. The sciatic nerve, including the proximal stump, distal stump, and the regenerating nerve bridge (if present), was carefully dissected under a stereomicroscope. Tissues were immediately immersed in 4% paraformaldehyde (PFA) at 4 °C for 24 h. Following fixation, samples were cryoprotected in 30% sucrose (w/v in PBS) at 4 °C until fully equilibrated (i.e., tissues sank). Cryoprotected tissues were embedded in optimal cutting temperature (OCT) compound, frozen, and sectioned longitudinally at a thickness of 10 μm using a cryostat (Leica CM1950). Sections were mounted onto glass slides and air−dried at room temperature. Immunofluorescence staining was performed in two separate rounds. In the first round, sections were permeabilized and blocked with 10% normal goat serum containing 0.3% Triton X−100 in PBS for 1 h at room temperature. Sections were then incubated overnight at 4 °C with primary antibodies against S100b (1:200, Abcam, ab52642), a marker for Schwann cells, and NF200 (1:300, Sigma–Aldrich, N0142), a marker for axons. After PBS washes, sections were incubated for 1 h at room temperature in the dark with goat anti–mouse Alexa Fluor 488 (1:1000, Invitrogen, A11029) and goat anti–rabbit Alexa Fluor 594 (1:1000, Invitrogen, A11012). In the second round, separate sections were stained with anti–TGF–β1 (1:200, Abmart, TA1027) to detect TGF–β1 expression. The same blocking and permeabilization protocol was used, and the sections were incubated overnight at 4 °C with the primary antibody, followed by goat anti–rabbit Alexa Fluor 594 (1:1000, Invitrogen, A11012) for secondary detection. In both staining procedures, nuclei were counterstained with DAPI and coverslipped using antifade mounting medium. Result Single−cell transcriptomics reveals dynamic cellular heterogeneity and intercellular communication in sciatic nerve repair We conducted scRNA−seq on rat sciatic nerve tissues following transection injury by analyzing 58,943 high−quality cells across seven time points (Day 0, 1, 3, 5, 7, 10, 14) (Additional file 1: Fig. [99]S1). Unsupervised clustering and UMAP dimensionality reduction revealed four major cellular compartments: NFs, Glis, immune cells (Mac, Mo, Gran, T/B/NK/DC cells), and vascular cells (PC, EC, SMC, lyEC) (Fig. [100]1 and Additional file 2: Fig. [101]S2a). Subclustering and annotation analysis revealed distinct cellular subtypes within each major compartment (Additional file 2: Fig. [102]S2 and Additional file 3: Fig. [103]S3a). In the NF lineage, we identified fibroblasts and mesenchymal cell populations, along with their respective subtypes (Additional file 2: Fig. [104]S2b). Glis exhibited functional heterogeneity, comprising proliferating, repairing, myelinating, and non−myelinating subtypes (Additional file 2: Fig. [105]S2c). Immune cell diversity was characterized by distinct Mac populations, including proliferating Macs, as well as plasmacytoid and mature/migrating dendritic cells (DCs) (Additional file 2: Fig. [106]S2f−k). Within the vascular compartment, we identified arterial pericytes (PCs), proliferating PCs and additional SMC−associated subtypes (Additional file 2: Fig. [107]S2d, e). These findings highlight the cellular complexity and dynamic responses underlying sciatic nerve repair. Fig. 1. [108]Fig. 1 [109]Open in a new tab Temporal Cell Type Heterogeneity Following Sciatic Nerve Transection. (a) UMAP visualization of integrated single–cell RNA sequencing (scRNA–seq) data across different post–injury time points. Each dot represents a single cell, colored by time point. (b) Left: UMAP showing clustering into four major cell types—neurofibroblasts (NFs), immune cells, glial cells (Glis), and vascular cells. Right: Cell type distribution across time points. (c) Bar plot of relative cell type proportions at each time point. (d) Dot plot of canonical marker genes across cell types. Dot size represents the proportion of cells expressing each gene (≥ 1 UMI); color indicates average expression level. (e) Heatmap of differentially expressed genes (DEGs) among primary cell types. Red: high expression; gray: low expression. (f) Volcano plots of DEGs for each major cell type. Top 10 DEGs are labeled; y–axis shows log₂ fold change (log₂FC). (g) Immunofluorescence staining for S100 (red, Schwann cells) and NF200 (green, axons) in uninjured and injured sciatic nerves (Days 3, 5, 7, 10, 14), showing spatiotemporal changes in Gli activity and axonal structure. Scale bar: 500 μm Dynamic changes in cellular composition were observed following sciatic nerve transection (Fig. [110]1a−c). At Day 0, NFs constituted the predominant population, accounting for 81.8% of the total cells, followed by vascular and immune cells, with Glis being the least abundant (Additional file 3: Fig. [111]S3c). NFs were characterized by the expression of key extracellular matrix−related genes such as Col1a1, Dcn, Col3a1, Col6a2, and Lum, which play a fundamental role in maintaining tissue integrity (Fig. [112]1d). Vascular cells were identified by markers including Vtn, Esam, Plvap, Acta2, and Des, while immune cells exhibited distinct signatures such as Aif1, Cd68, Cd3e, Cd3g, Inpp5d, and Adgre1. Glial cells, primarily Schwann cells, expressed Mpz, S100b, Mbp, and Mag, underscoring their role in nerve support and myelination. By Day 1 post−injury, immune cells expanded dramatically, exceeding 95% of the total cell population (Additional file 3: Fig. [113]S3c). This early inflammatory response was dominated by Macs and granulocytes (Grans), consistent with their essential roles in debris clearance and initiating the repair process [[114]31, [115]32]. Mac subsets were characterized by markers such as F13a1, Pf4, C1qc, C1qb, Ms4a7, and Folr2, whereas Grans displayed gene signatures including S100a8, S100a9, Il1r2, and Dgat2 (Additional file 5: Fig. [116]S5e). The marked immune infiltration suggested a rapid activation of innate immune mechanisms to facilitate the removal of myelin debris [[117]33]. By Day 5, NFs re−emerged in substantial numbers, marking a transition from the inflammatory phase to the regenerative phase. This increase in NFs coincided with a notable rise in Glis, which play a critical role in axonal regeneration and remyelination. At the same time, immune cell numbers began to decline, indicating a shift toward tissue remodeling and repair. By Day 7, vascular cells nearly doubled, primarily comprising PCs and ECs (Fig. [118]1c and Additional file 3: Fig. [119]S3c). This vascular expansion likely reflects an increase in angiogenesis and the establishment of a supportive microenvironment for regenerating axons. Given the observed strong cell–cell communication between Glis and vascular cells (Additional file 3: Fig. [120]S3b), their coordinated function may be essential in restoring nerve homeostasis and promoting functional recovery [[121]34]. The analysis of cellular communication during nerve repair reveals complex interactions between different cell types (Additional file 3: Fig. [122]S3b). Glis and vascular cells exhibit strong cell−cell communication both in normal and injured states, suggesting their pivotal roles in maintaining nerve homeostasis and promoting tissue repair. NFs and Glis show robust signaling interactions, with potential regulatory pathways that facilitate cellular differentiation and tissue regeneration [[123]35]. While immune cells play a crucial role in the early stages of injury, their cell−cell communication with other cell types is relatively weaker both in normal tissue and throughout the repair process [[124]30]. The intricate communication networks observed during the repair process highlight the need for coordinated signaling between multiple cell types to ensure efficient tissue regeneration and functional recovery. In summary, this study provides a comprehensive overview of the cellular landscape during sciatic nerve injury and repair, revealing dynamic shifts in cellular proportions and highlighting the complex cellular interactions that drive tissue regeneration. The identification of key cell subtype markers and their associated signaling pathways offers insights into the molecular mechanisms underlying nerve repair, which may inform future therapeutic strategies aimed at enhancing nerve regeneration. Temporal dynamics and subtype−specific remodeling of the immune landscape during sciatic nerve repair At Day 0, Macs constitute the predominant immune cell population, followed by T cells and DCs (Fig. [125]2a, b). Following injury, the immune response undergoes dynamic changes over time. During the early phase (Day 1–Day 3), the immune landscape is dominated by Macs and Grans, reflecting their crucial role in the immediate inflammatory response and debris clearance [[126]31, [127]36]. These early immune events are primarily driven by myeloid−derived cells. As the repair process progresses (Day 5–Day 14), the proportion of T cells and DCs increases, while Macs gradually decline, indicating a shift toward a lymphoid−and DC−driven immune environment. This transition highlights the shift from an inflammatory environment to an adaptive immune response and tissue remodeling [[128]37, [129]38]. Fig. 2. [130]Fig. 2 [131]Open in a new tab Clustering and Functional Profiling of Myeloid Immune Subtypes. (a) UMAP plot showing immune cell types across all time points: granulocytes (Grans), macrophages (Macs), monocytes (Mos), dendritic cells (DCs), T cells, B cells, and NK cells. (b) Bar plot of immune cell type proportions over time. (c–e) Unsupervised subclustering of myeloid cells. Left: UMAP of subtypes; Right: temporal distribution. (c) Gran0–Gran3. (d) Mac0–Mac4. (e) Mo0–Mo1. (f–h) Heatmaps of upregulated marker genes (Z–score, log scale) for each subtype. (f) Grans, (g) Macs, (h) Mos. (i) GO enrichment analysis of each Gran, Mac, and Mo subtype Myeloid−derived populations: Grans, Macs, and Mos Grans exhibit significant changes post−injury (Fig. [132]2c). At Day 0, the predominant Gran subsets are Gran0 and Gran1, with Gran0 increasing and Gran1 decreasing during the course of nerve repair. Additionally, two new Gran subtypes, Gran2 and Gran3, emerge post−injury. Marker gene analysis identifies Gran0 as expressing Riok3, Gadd45a, and S100a8, while Gran1 expresses Rps28 and Naca. Gran2 cells upregulate Ccl2 and Gpnmb, whereas Gran3 expresses Atp6v0d1, Psmb3, and Capg (Fig. [133]2f). Functional enrichment analysis reveals distinct roles among Gran subtypes during nerve repair (Fig. [134]2i). Gran0 is primarily associated with redox homeostasis and metabolic regulation, suggesting a role in detoxification and chemical balance. Gran1 is enriched in pathways related to translation and Toll−like receptor signaling, indicating its involvement in early innate immune responses. Gran2 shows signatures linked to vascular modulation and negative regulation of TGF− Inline graphic signaling, implicating it in immunoregulation and endothelial interactions. Gran3 is associated with sphingolipid metabolism, pointing to potential roles in membrane remodeling and inflammatory resolution. These findings highlight the functional heterogeneity of Grans and their dynamic contributions to different phases of peripheral nerve regeneration. During the early phase (Day 1), the Mac0 subtype dominates (Fig. [135]2d). These cells are characterized not only by their involvement in acute−phase responses and metabolic processes related to amino acid metabolism (Fig. [136]2i) but also by a unique marker profile that includes Cl1, Cxcl3, Ereg, Slc2a6, Slpi, Ass1, Vcan, Ccl24, and Nrg1 (Fig. [137]2g). By Day 3, there is a marked shift with Mac1 and Mac2 subtypes becoming more prominent. Mac1 cells—displaying markers such as Fxyd2, Hpse, Asgr2, Kctd4, Pdgfc, Akr1b8, Gdf15, Gsta1, Htr2b, and C6—likely represent a substantial fraction of blood−derived Macs, while Mac2 cells are defined by the expression of Dcn, Col3a1, and Col1a1 (Fig. [138]2g and Additional file 2: Fig. [139]S2f). These subtypes are involved in lipid homeostasis, cytokine regulation, and extracellular matrix remodeling (Fig. [140]2i). Their numbers decline by Day 5 and nearly vanish by Day 7, coinciding with the progressive increase in Mac3 and Mac4 populations. The Mac4 subset, identified by a distinct proliferative marker profile (Cep55, Mastl, Hist1h1b, Kif20b, Hmmr, Aspm, Kif4a, Sgo2, Pclaf, and Espl1), gradually increases in the later repair stages (Day 7, Day 10, and Day 14), underscoring its importance in cell proliferation and tissue regeneration (Fig. [141]2g). Enrichment analysis further support these findings (Fig. [142]2i). Specifically, Mac0 cells are enriched in pathways related to acute−phase response and both proteinogenic and non−proteinogenic amino acid metabolism. In contrast, Mac1 cells are associated with negative regulation of cytokine production, lipid homeostasis, and complement activation, among other immune regulatory processes. Mac2 cells are linked to responses to mechanical stimuli and collagen fibril organization. Meanwhile, Mac3 cells contribute to the regulation of hemopoiesis and non−membrane−bound organelle assembly, and Mac4 cells not only support the proliferative capacity by engaging in chromosome segregation and DNA repair−dependent chromatin remodeling but also modulate epigenetic regulation through pathways such as constitutive heterochromatin formation and negative regulation of the cGAS/STING signaling pathway. Mos, largely absent at Day 0, appear in significant numbers at the lesion site during the subsequent stages of repair (Fig. [143]2e, h). Two major Mo subsets are observed: Mo0 and Mo1. Mo0 cells are marked by Cdc42ep4, Polr2j, Dap, Polr2e, Cct3, Ccl2, Cd81, Card19, Rhoc, and Sh3kbp1, and their enrichment analysis shows a strong association with mRNA metabolism, nucleic acid catabolic processes, and T cell−mediated immunity, among other functions. In comparison, Mo1 cells—exhibiting markers such as Cops2, Dhx58, Psmg4, Isg15, Tsta3, Phc2, Pik3cd, Krt75, Gps1, and Polr3k—are enriched in pathways involved in RNA catabolism and the cellular response to increased oxygen levels (Fig. [144]2h, i). The stable presence and balanced proportions of these Mo subsets throughout the repair process highlight their crucial role in shaping the heterogeneity and functionality of the Mac populations. DCs DCs display dynamic changes in their subpopulations (Fig. [145]3a). At Day 0, the sciatic nerve primarily contains DC0 cells, which decrease during repair, while DC1 cells progressively increase. Subtype characterization reveals that DC0 cells, expressing Ccl17, Mfge8, and Clec4a1, are Mo−derived DCs (Fig. [146]3e and Additional file 2: Fig. [147]S2h). DC1 cells, marked by Mctp2, Slco4a1, and Siglech, predominantly represent plasmacytoid DCs, whereas DC3 cells (Gls2, Fscn1, and Lad1) correspond to mature/migrating DCs, and DC2 cells (Naaa, Slpi, and Xcr1) align with conventional DCs (Fig. [148]3e and Additional file 2: Fig. [149]S2h). Functional analysis indicates that DC0 cells are involved in biotic stimulus responses and Toll−like receptor signaling, DC1 cells participate in endoplasmic reticulum stress responses and nuclear receptor signaling, DC2 cells regulate innate immune responses and lipid absorption, and DC3 cells mediate actin filament organization and Toll−like receptor signaling (Fig. [150]3i). These findings suggest a coordinated DC response that facilitates antigen presentation and immune modulation during nerve repair. Fig. 3. [151]Fig. 3 [152]Open in a new tab Clustering and Functional Profiling of DC and Lymphoid Subtypes. (a–d) Unsupervised subclustering of lymphoid and DC subtypes. Left: UMAP; Right: temporal proportions. (a) Dendritic cells: DC0–DC3. (b) T cells: T0–T3. (c) B cells: B0–B2. (d) NK cells: NK0–NK1. (e–h) Heatmaps of upregulated marker genes (Z–score, log) for each subtype. (e) DCs, (f) T cells, (g) B cells, (h) NK cells. (i–j) GO enrichment analysis. (i) DC subtypes. (j) T, B, and NK cell subtypes Lymphoid−derived populations: T, B, and NK cells T cells exhibit distinct subtypes that vary throughout the repair process (Fig. [153]3b). At Day 0, the sciatic nerve primarily harbors T0 and T2 cells, with T0 gradually decreasing post−injury. In contrast, T1, T2, and T3 subpopulations expand during the repair phase. Marker gene analysis reveals that T0 cells are characterized by Calhm6 and S1pr1 expression, whereas T1 cells express Ccl5, Ccl3, and Ccl4, suggesting their involvement in immune activation and recruitment (Fig. [154]3f). T2 cells, marked by Btrc, Tnfrsf4, and Foxp3, may correspond to regulatory T cells contributing to immune modulation. T3 cells, enriched in genes related to chromosome segregation and mitotic activity, such as Ncapg, Cdca3, and Top2a, likely represent proliferative T cell subsets. Functional enrichment analysis further supports these findings (Fig. [155]3j), as T0 cells are associated with type II interferon production and ATP export, T1 cells with lymphocyte−mediated immunity and dopamine biosynthesis, T2 cells with cytokine regulation and mononuclear cell proliferation, and T3 cells with chromatin remodeling and amino acid metabolism. B cells, although present at low levels throughout the repair process (Fig. [156]3c), undergo subtype−specific changes. The primary subsets include B0, B1, and B2 cells. Marker gene analysis indicates that B0 cells express Cd40, Lbh, and Ms4a1, while B1 cells upregulate Gpr171, Grn, and Cd7 (Fig. [157]3g). Notably, B2 cells, characterized by Nipal1, Ctla4, and Tnfrsf17 expression, primarily correspond to plasma blasts (Fig. [158]3g and Additional file 2: Fig. [159]S2k). Functional enrichment analysis suggests that B0 cells participate in ribosome biogenesis and interleukin−2 regulation, B1 cells contribute to autophagosome assembly and lectin receptor signaling, and B2 cells engage in endoplasmic reticulum stress responses and protein localization (Fig. [160]3j). Natural killer (NK) cells are present at low levels in the sciatic nerve from Day 0 onward, with no significant changes in their proportions observed throughout the repair process (Fig. [161]3d). Two major NK cell subtypes are identified: NK0 and NK1. NK0 cells are enriched in Gata3, Cd27, and Gpr183, while NK1 cells express Batf, Il21r, and Gzma (Fig. [162]3h). Functional annotation suggests that NK0 cells contribute to T cell chemotaxis and calcium ion transport, whereas NK1 cells are involved in Gran chemotaxis and tissue disruption, implying a role in immune surveillance and cytotoxic activity (Fig. [163]3j). Collectively, these results demonstrate that immune cell populations undergo distinct temporal and subtype−specific changes during sciatic nerve repair. The early phase is dominated by Macs and Grans, facilitating debris clearance and acute inflammation. As repair progresses, T cells, DCs, and B cells expand, contributing to adaptive immunity and tissue remodeling. These findings provide insights into the immune landscape following nerve injury, with potential implications for therapeutic strategies targeting immune modulation in peripheral nerve repair. Mo−derived Mac diversification and functional trajectories shape the immune microenvironment during sciatic nerve repair Macs and Mos play critical roles in the repair process following sciatic nerve injury, orchestrating a dynamic response that evolves over time. At Day 0, Macs are predominantly of the Mac3 subtype, characterized by high expression of genes such as Slco2b1, Cd4, and Selenop (Fig. [164]2d, g). Following injury, the immune landscape undergoes significant shifts, with distinct Mac subtypes emerging at different stages. Pseudotime trajectory analysis reveals two distinct differentiation pathways for Mo−derived Macs, both originating from Mos (the root state) and progressing through intermediate stages (Fig. [165]4c, f). Trajectory 1 follows a sequential progression whereby Mos transition through early−stage Mac0, intermediate−stage Mac2, and ultimately differentiate into late−stage Mac1 and Mac3 subsets. In Trajectory 2, the cells share the early to intermediate stages (Mac0 Inline graphic Mac2) but diverge in the late phase, giving rise to a mix of Mac1, Mac3, and the proliferative Mac4 subpopulation. These trajectories are accompanied by dynamic transcriptional reprogramming. For instance, genes such as Apoe, Pltp, C1qa, C1qb, and C1qc peak in late−stage Mac of Trajectory 1 and then decline, indicating transient yet essential roles in lipid metabolism, cholesterol efflux, and complement system regulation (Fig. [166]4e). In Trajectory 2, the late−phase upregulation of proliferation−associated genes—including Tuba1b, Hmgb2l1, Tubb5, and Cst3—highlights the support for microtubule dynamics, chromatin remodeling, and lysosomal functions in promoting the expansion of the Mac4 subset. The fluctuating expression of additional markers such as Ctsz and Vim further underscores their roles in proteolysis and cytoskeletal reorganization during tissue repair (Fig. [167]4h). Fig. 4. [168]Fig. 4 [169]Open in a new tab Mo–to–Mac Trajectories and Cell–Cell Interactions. (a) UMAP clustering of Mos and Macs with annotated subtypes. (b) Chord diagram of CellChat–inferred collagen signaling between Macs and NFs at different time points. Edge thickness: interaction strength; color: sender cluster. (c, f) Slingshot trajectory analysis showing pseudotime for trajectories 1 and 2. (d, g) Density plots of Mac and Mo subtypes along each pseudotime trajectory. (e, h) Smoothed expression dynamics of representative genes along pseudotime (GAM fitting). Each dot represents a single cell, colored by its corresponding time point Finally, cell−cell communication analysis reveals that collagen−mediated signaling between Macs and NFs is significantly enhanced during the repair process (Fig. [170]4b). Although this interaction is minimal at Day 0 and absent at Day 1, it intensifies markedly from Day 3 onward. This observation reinforces the significance of collagen in extracellular matrix remodeling and nerve regeneration, as well as the critical interplay between Macs and NFs in facilitating tissue repair. The integration of marker gene analysis with functional and temporal data not only refines our understanding of Macs and Mos heterogeneity but also provides important insights for developing therapeutic strategies aimed at harnessing Mac−mediated regenerative processes. Temporal dynamics and functional specialization of vascular cell populations during sciatic nerve repair At Day 0, vascular cell populations were predominantly composed of smooth muscle cells (SMCs) (Fig. [171]5a, b). Following nerve injury, vascular cell dynamics underwent significant shifts across different time points. Notably, no vascular cell infiltration was observed on Day 1 post−injury. From Day 3 to Day 14, ECs remained the dominant vascular cell type, with PCs emerging on Day 5 and subsequently maintaining a stable proportion throughout the repair process. A substantial increase in vascular cell numbers was observed from Day 7, showing a rise that persisted in the later stages. In contrast, SMCs and lyECs remained relatively scarce throughout the repair timeline. Fig. 5. [172]Fig. 5 [173]Open in a new tab Vascular Cell Subtypes and Functional Characteristics. (a) UMAP clustering of vascular cells into four types: endothelial cells (ECs), lymphatic ECs (lyECs), pericytes (PCs), and smooth muscle cells (SMCs). (b) Bar plot of vascular cell type proportions over time. (c–f) Subclustering of vascular types. Left: UMAP; Right: subtype proportions over time. (c) EC0–EC2. (d) lyEC0–lyEC1. (e) PC0–PC3. (f) SMC0–SMC1. (g–j) Heatmaps of upregulated markers (Z–score, log) per subtype. (g) ECs, (h) lyECs, (i) PCs, (j) SMCs. (k–l) GO enrichment results. (k) EC and lyEC subtypes. (l) PC and SMC subtypes EC subtypes exhibited distinct temporal patterns and functional enrichments during nerve repair (Fig. [174]5c, g). At Day 0, ECs were rare, and no ECs were detected at Day 1 post−injury. From Day 3 to Day 10, EC0 and EC1 constituted the predominant endothelial populations contributing to repair. By Day 14, EC2 became the dominant subtype. GO analysis revealed functional specialization among these subpopulations (Fig. [175]5k). EC0 was enriched in pathways associated with epithelial tube morphogenesis, negative regulation of cell migration, and response to reactive oxygen species, highlighting its role in early−stage tissue remodeling. EC1 displayed enrichment in inflammatory response regulation, nitric oxide metabolism, and maintenance of blood vessel diameter, indicating its involvement in modulating vascular tone and oxidative stress response. EC2, emerging in the later stages, was linked to apoptotic signaling regulation, nucleic acid catabolism, and toll−like receptor signaling, suggesting a role in resolving inflammation and tissue remodeling. lyECs were scarcely present in the normal sciatic nerve, with lyEC1 constituting the primary population (Fig. [176]5d, h). Following injury, lyECs began to emerge at Day 5, predominantly consisting of lyEC0, and showed a notable numerical increase by Day 14, suggesting their involvement in the later phases of repair. Marker gene analysis revealed that lyEC0 expressed Plaur [[177]39], Ccl2 [[178]40], Nr4a3 [[179]41], Hmox1 [[180]42], and Serpine1 [[181]43]—genes involved in immune cell recruitment, oxidative stress regulation, and extracellular matrix remodeling (Fig. [182]5h, k). These features are consistent with the functional enrichment of lyEC0 in apoptotic signaling regulation, TGF− Inline graphic receptor signaling, and nucleic acid catabolic processes, indicating a role in immune modulation and tissue adaptation during early regeneration. In contrast, lyEC1, although present at baseline, became relatively less prominent post−injury but retained expression of genes such as Cytl1, Slco2a1, Smad4, and Rbp7. These genes are associated with immune signaling (e.g., Cytl1 [[183]44], Slco2a1 [[184]45]), transcriptional regulation (Smad4 [[185]46]), and metabolic function (Rbp7 [[186]47]), suggesting that lyEC1 may contribute to maintaining endothelial homeostasis and modulating repair−associated responses during the late stages of nerve regeneration (Fig. [187]5h, k). Both lyEC subtypes showed enrichment in TGF− Inline graphic receptor signaling pathways, reinforcing the pathway’s central importance across endothelial phenotypes. Moreover, cell–cell communication analysis revealed that lyECs maintained robust and persistent interactions with non−immune cell types, including vascular cells, Glis, and NFs, throughout the regenerative timeline (Additional file 3: Fig. [188]S3b). This suggests that lyECs may contribute to coordinating lymphatic remodeling, immune resolution, and stromal stabilization in the evolving nerve microenvironment. PCs exhibited a distinct temporal pattern, with minimal presence at Day 0 (Fig. [189]5e). PC populations emerged at Day 5 and followed a dynamic shift in subtypes over time. PC1 was identified as arterial pericytes, and PC3 as proliferative pericytes, each displaying distinct temporal and functional features (Additional file 2: Fig. [190]S2d). PC3, the proliferative subset characterized by marker genes such as Esco2, Cdca3, Kif2c, Top2a, and Nusap1, emerged as early as Day 5 and declined progressively, indicating an early role in proliferative expansion, chromosomal segregation, and stress adaptation (Fig. [191]5e, i, l). PC1, marked by Pln, Entpd1, Slit3, Col4a5, and Fmod, began to appear from Day 7 and maintained a stable proportion thereafter, consistent with its annotation as arterial pericytes and its role in structural vascular support and vascular tone regulation (Fig. [192]5e, i, l). Functionally, PC1 was enriched in muscle cell differentiation, cholesterol metabolism, and viral response pathways, suggesting involvement in both metabolic support and immune modulation (Fig. [193]5l). In the early repair phase (Day 5–Day 10), PC0 was the predominant subtype. It expressed RT1−A1, Il33, and Lum, and was enriched in extracellular matrix organization, regulation of blood circulation, and TGF− Inline graphic signaling, implicating it in early vascular stabilization and immune modulation (Fig. [194]5e, i, l). By Day 14, PC2 became the dominant population. PC2−specific markers included Adra2b, Ets2, and Tfpi2, with enrichment in apoptotic signaling, nuclear protein import, and pigmentation regulation, suggesting a transition toward late−stage tissue remodeling and homeostatic maintenance (Fig. [195]5e, i, l). Furthermore, intercellular communication analysis revealed that PCs maintained close interactions with vascular cells, NFs, and Glis during the repair process. Notably, analysis of NF subtype–PC subtype communication patterns revealed persistent and aberrant activation of the CD99 signaling pathway, with PCs as the sending and NFs as the receiving cells, implying a potential role in pathological remodeling or chronic repair signaling [[196]48] (Additional file 4: Fig. [197]S4c). SMCs were present in very low numbers throughout the repair process, appearing only at Day 7 post−injury (Fig. [198]5f, j). SMC0 and SMC1 were identified as the primary subtypes, each displaying distinct functional properties. SMC0 exhibited enrichment in extracellular matrix organization, muscle cell proliferation, nucleocytoplasmic transport, and calcium ion transmembrane transport, suggesting its involvement in vascular support and contractile function (Fig. [199]5l). SMC1 shared enrichment in extracellular matrix organization and TGF− Inline graphic signaling but also displayed enrichment in amino acid metabolic processes, potentially linking it to metabolic regulation within the vascular microenvironment during nerve repair. Overall, vascular cell dynamics during sciatic nerve repair revealed a highly coordinated response, with ECs playing a central role in early repair, followed by the recruitment of PCs and the late emergence of SMCs. The functional diversity of these subtypes suggests specialized roles in angiogenesis, inflammatory modulation, and extracellular matrix remodeling, highlighting their critical contributions to the regeneration process. Dynamic shifts in NF subpopulations reveal phase−specific roles in nerve regeneration Dynamic changes in NF subpopulations were observed throughout the sciatic nerve repair process, highlighting their diverse functional roles in different phases of regeneration (Fig. [200]6a−c). At Day 0, NF2 cells, identified by marker genes Crispld2, Sqle, Aldh1a1, Idi1, Ralgps2, Kcnk2, Myoc, Cttnbp2, Col9a1, and Col9a2, were the predominant population (Fig. [201]6d). These cells were classified as endoneurial mesenchymal cells and were primarily involved in extracellular matrix organization and TGF− Inline graphic receptor superfamily signaling pathways, supporting the structural integrity of the nerve and facilitating initial cellular responses post−injury (Fig. [202]6e and Additional file 2: Fig. [203]S2b). Fig. 6. [204]Fig. 6 [205]Open in a new tab Neurofibroblast Subtype Dynamics During Nerve Repair. (a) UMAP of NFs across time points. (b) Left: Subtype clustering of NF0–NF6. Right: Temporal distribution of each subtype. (c) Proportional changes in NF subtypes post–injury. (d) Heatmap of upregulated marker genes (Z–score, log). (e) GO enrichment of NF subtypes. (f) GO enrichment of glial subtypes (Gli0–Gli5) for comparison In the early inflammatory phase (Day 1 to Day 3), NF5 cells, characterized by Ncapg, Tpx2, Hmmr, Ect2, Kif4a, Cenpu, Cenpf, Sgo2, Pclaf, and Kif20b, exhibited significant expansion (Fig. [206]6c, d). These cells, categorized as proliferating mesenchymal cells, played a critical role in chromosome segregation and non−membrane−bounded organelle assembly (Fig. [207]6e and Additional file 2: Fig. [208]S2b) [[209]49, [210]50]. Their increased activity suggested a surge in cell proliferation, likely contributing to the rapid remodeling of the extracellular environment to accommodate immune cell infiltration and debris clearance [[211]51, [212]52]. By Day 5, NF0 cells, defined by Spon1, displayed a notable peak, surpassing their levels at Day 3 and Day 7 (Fig. [213]6c, d). NF0 cells, along with NF1 and NF3, were classified as differentiating mesenchymal cells (Additional file 2: Fig. [214]S2b). NF0 cells were specifically enriched in pathways related to extracellular matrix organization, TGF− Inline graphic receptor signaling and nuclear receptor−mediated signaling (Fig. [215]6e). The increased presence of NF0 at this stage indicated their crucial role in transitioning from the inflammatory to the regenerative phase, facilitating extracellular matrix remodeling and tissue stabilization. Simultaneously, NF4 cells, characterized by the expression of Lrg1 and Plvap—genes commonly implicated in vascular biology [[216]53, [217]54]—were enriched in collagen fibril organization and glycolytic pathways, indicating a role in matrix remodeling and metabolic adaptation essential for repair (Fig. [218]6c−e) [[219]55]. During the mid−to−late regenerative phase (Day 7 to Day 14), NF0 cells continued to increase gradually, whereas NF3 cells, marked by Sbsn, Cdkn2a, Bhlhe22, Apod, Rdh10, Slc16a11, A2m, Scn3b, Plcxd3, and Mrap2, showed a progressive decline (Fig. [220]6c, d). NF3 cells were involved in epithelial cell proliferation regulation and TGF− Inline graphic receptor signaling, suggesting their involvement in early repair mechanisms that diminished as regeneration progressed (Fig. [221]6e). These cells played a key role in the final stages of repair by contributing to the reconstruction of the extracellular matrix and supporting the structural and functional restoration of the nerve tissue [[222]56, [223]57]. Overall, the dynamic shifts in NF subpopulations underscore their essential contributions to different phases of nerve repair. The transition from NF5−driven proliferative responses to NF4−mediated metabolic transformation and NF0−associated extracellular matrix remodeling highlights the coordinated interplay of these NF subsets. Their involvement in key signaling pathways, particularly TGF− Inline graphic receptor superfamily signaling [[224]58], suggests that modulating these pathways could be a potential therapeutic strategy to enhance nerve regeneration. Dynamic heterogeneity and temporal fate transitions of Glis during sciatic nerve regeneration Dynamic changes in Gli populations were observed during the repair process following sciatic nerve transection (Fig. [225]7a−c). At Day 0, the predominant Gli subtype was Gli5, which represents myelinating Schwann cells (Additional file 2: Fig. [226]S2c). Following nerve injury, Glis underwent significant phenotypic transitions, with Gli0, identified as repair−associated Schwann cells, emerging as the dominant subtype during the early phase of regeneration (Fig. [227]7c and Additional file 2: Fig. [228]S2c). As the repair process progressed, Gli0 cells gradually declined, while Gli2 cells increased, suggesting a shift toward remyelination and structural recovery (Fig. [229]7c). Gli0, Gli1, and Gli4 contribute to nerve regeneration by engaging the TGF− Inline graphic receptor superfamily signaling pathway to modulate cellular responses, orchestrating extracellular matrix organization to reshape the tissue microenvironment, and regulating axon ensheathment to support the restoration of nerve function (Fig. [230]6f). Fig. 7. [231]Fig. 7 [232]Open in a new tab Glial Cell Subtype Differentiation and Signaling Activity. (a) UMAP of Glis across time points. (b) Left: Subtype clustering of Gli0–Gli5. Right: Distribution over time. (c) Changes in Gli subtype proportions post–injury. (d) Heatmap of upregulated markers (Z–score, log). (e) Pseudotime trajectories from Day 5–14 identifying two distinct differentiation paths. (f) Heatmaps of variable gene expression along each trajectory. (g) Density plots of cell distributions along trajectories. (h) Bubble plots of pathway activation over pseudotime. (i) CellChat chord diagram of PTN signaling from Gli to NF cells. (j) Bubble plot showing predicted Gli–to–NF signaling interactions by pathway Single−cell transcriptomic analysis identified distinct marker genes associated with different Gli subtypes (Fig. [233]7d). Gli5, characteristic of myelinating Schwann cells, expressed markers such as Ncmap, Sema5a, Mt1 and Kcna1. In contrast, Gli0 (repair Schwann cells) exhibited increased expression of genes such as Clcf1, Met, Artn, and Runx2, which are associated with regeneration and extracellular matrix reorganization (Fig. [234]6f and [235]7d). Gli3 (proliferating Schwann cells) exhibited high expression of Ube2c, Kif14, Kif4a and Plk1, indicating active cell cycle progression and proliferation (Fig. [236]6f and [237]7d). The transition toward remyelination was marked by the emergence of Gli2, which expressed genes such as Nefm, Cuedc2, Cldn19, and Mag, indicating a functional shift toward axon ensheathment and nerve fiber stabilization [[238]59, [239]60] (Fig. [240]6f and [241]7d). Functional enrichment analysis further elucidated the biological roles of different Gli subtypes during the repair process (Fig. [242]6f). Gli0 was associated with pathways involved in TGF− Inline graphic receptor signaling, neuron projection guidance, and glycoprotein metabolism, all of which are critical for early nerve repair. Gli1, which plays a role in extracellular matrix remodeling, was enriched for pathways related to TGF− Inline graphic signaling, peptide cross−linking, and interferon−mediated signaling. Gli2 exhibited functional enrichment in myelination−related processes, including amine transport and apical protein localization, consistent with its role in late−stage nerve repair. Gli3 was primarily associated with cell cycle regulation, with enrichment in chromosome segregation and organelle assembly. Gli4 was linked to extracellular matrix organization and nuclear receptor−mediated signaling, suggesting involvement in tissue remodeling. Finally, Gli5, as the mature myelinating Schwann cell population, was enriched for pathways regulating axon ensheathment, potassium ion transport, and sterol biosynthesis, all of which are crucial for maintaining functional nerve architecture. Pseudotime trajectory analysis revealed two major regenerative pathways governing Gli fate transitions (Fig. [243]7e−h and Additional file 7: Fig. [244]S7a−b). The first trajectory was characterized by increased expression of genes associated with myelination, proliferation, autophagy, and metabolic regulation, including Col3a1, Csrp2, Mbp, Mpz, posten, and Pmp22 (Additional file 7: Fig. [245]S7b). These genes were upregulated during later stages of repair, highlighting their role in structural restoration [[246]61–[247]63]. The second trajectory was primarily associated with growth factor secretion and extracellular matrix modulation, with early upregulation of genes such as Apod, Apoe, Col1a1, and Fn1 (Additional file 7: Fig. [248]S7b) [[249]64, [250]65]. The dynamic interplay between these two pathways underscores the complex cellular mechanisms governing Gli function during nerve regeneration. Additionally, cell−cell communication analysis indicated that Glis actively interacted with NFs during repair, particularly through enhanced PTN signaling (Fig. [251]7i and Additional file 7: Fig. [252]S7c), which is consistent with the known roles of PTN in promoting neurite outgrowth, axon regeneration, and cell–cell communication in the nervous system [[253]66]. This interaction was not only observed during nerve regeneration but has also been implicated in pathological conditions such as neurofibromatosis, where Gli−NF communication is dysregulated (Fig. [254]7j and Additional file 8: Fig. [255]S8) [[256]67]. Collectively, these findings provide a comprehensive understanding of Gli heterogeneity, their temporal dynamics, and their critical roles in coordinating nerve repair following injury. TGF− Inline graphic signaling exhibits sustained and spatially distinct activation during nerve repair To validate the cell−type−specific TGF− Inline graphic signaling activation observed in our single−cell transcriptomic analysis, we conducted a complementary bulk RNA−seq experiment. We collected 2−mm proximal and distal segments of the sciatic nerve at Day 1, 3, 5, 7, and 14 post−transection, along with uninjured control nerves, and performed transcriptome−wide differential expression analysis (Fig. [257]8a). Across all time points, proximal and distal segments displayed highly similar gene expression dynamics relative to uninjured controls. Fig. 8. [258]Fig. 8 [259]Open in a new tab Region – and Time–Resolved Bulk Transcriptomic Profiling of TGF– Inline graphic signaling after nerve transection. (a) Scatter plots comparing log₂FC in proximal vs. distal segments (relative to uninjured controls) at Days 1, 3, 5, 7, and 14. Red: genes upregulated in both; green: downregulated in both. Labeled genes are TGF– Inline graphic –related and significantly changed in both regions. (b) GO enrichment of genes from (a), categorized as shared, proximal–specific, or distal–specific upregulated genes. (c) GSEA heatmap of normalized enrichment scores (NES) for TGF– Inline graphic –related pathways across regions and time points. (d) Heatmap of key TGF– Inline graphic signaling components (ligands, receptors, mediators, targets) across regions and time points (Z–score, log). (e) Immunofluorescence staining of TGF– Inline graphic in uninjured and injured nerves (Days 3–14), showing spatial–temporal changes in protein expression. Scale bar: 500 μm Enrichment analysis of shared DEGs revealed consistent activation of the TGF− Inline graphic signaling pathway throughout the regeneration process (Fig. [260]8b). Notably, genes specifically upregulated in the proximal segment at Day 14 remained significantly enriched for TGF− Inline graphic signaling, whereas distal–specific genes did not, suggesting more sustained transcriptional activation in the proximal region. GSEA analysis further highlighted this spatial distinction: the distal nerve segment exhibited strong early activation of TGF− Inline graphic signaling at Day 1, which decreased over time, while the proximal segment maintained elevated activity across later time points (Fig. [261]8c). This pattern was partially reflected at the protein level. Immunofluorescence staining revealed robust TGF− Inline graphic upregulation in both proximal and distal segments post–injury, with expression persisting longer in the proximal region (Fig. [262]8e). However, transcriptomic heatmaps of canonical TGF− Inline graphic pathway components—including ligands (Tgfb1, Tgfb2), receptors (Tgfbr1, Tgfbr2), intracellular mediators (Smad2, Smad3, Smad4), and downstream targets (Serpine1, Ctgf, Fn1, Col1a1, Acta2)—revealed a noteworthy pattern: by Day 14, distal segments exhibited higher mRNA expression of many TGF− Inline graphic –related genes compared to the proximal side (Fig. [263]8d). This discrepancy between mRNA and protein levels likely reflects region – and phase–specific post–transcriptional regulation and feedback control mechanisms, suggesting that TGF− Inline graphic signaling persists transcriptionally in the distal stump even after protein–level activity has diminished. Together, these findings integrate protein and transcriptomic dynamics to reveal a complex spatiotemporal regulation of TGF− Inline graphic signaling, involving both early and sustained activation in different compartments. Moreover, enrichment of “negative regulation of TGF− Inline graphic signaling” at Day 1 in the distal stump and Day 7 in the proximal stump suggests location – and phase–specific feedback modulation. These bulk transcriptomic results are highly consistent with our single–cell data, which showed widespread TGF− Inline graphic pathway activity across major cell types including Glis, NFs, and immune cells, and align with previous reports highlighting the essential role of TGF− Inline graphic signaling in orchestrating inflammation, extracellular matrix remodeling, and cellular differentiation during peripheral nerve repair [[264]68, [265]69]. Divergent cellular and molecular repair programs in crush versus transection models of sciatic nerve injury The comparative analysis of crush and transection injuries revealed distinct cellular responses and molecular pathways involved in the repair process (Fig. [266]9). In the early stages following crush injury, Macs were the predominant immune cell type, maintaining a stable presence from Day 1 to Day 7 (Fig. [267]9b−e). In contrast, in transection injury, Mac presence was more pronounced in the early phases, with a higher proportion on Day 1 and Day 3 compared to crush injury. However, by Day 7, the proportion of Macs had significantly declined. Additionally, Gran were more abundant in the early stages of transection injury (Day 1 and Day 3) than in crush injury, but their presence diminished markedly by Day 7. Fig. 9. [268]Fig. 9 [269]Open in a new tab Comparative Analysis of Crush vs. Transection Injuries in Single–Cell Resolution. (a) UMAP of cells from sciatic nerve crush injury (SNCI) and transection injury (SNTI) at Days 0, 1, 3, 7. (b) Cell type annotation using SingleR with the Injured Sciatic Nerve Atlas (iSNAT) as reference. (c–d) UMAPs of SNCI (c) and SNTI (d) separately, colored by cell type. (e) Bar plot comparing cell type proportions between SNCI and SNTI at each time point. (f) Scatter plots of DEGs (log₂FC > ± 1, FDR < 0.05) in SNCI vs. SNTI at Days 1, 3, 7. Red: downregulated in both; green: upregulated in both. (g) GO enrichment of commonly upregulated genes (green in f). (h) GO enrichment of injury–specific genes (log₂FC > 1 in only SNCI or SNTI) Schwann cell dynamics also varied between the two injury models (Fig. [270]9b−e). In crush injury, Schwann cells progressively increased from Day 1 to Day 7, contributing to nerve regeneration. Conversely, in transection injury, Schwann cells were scarce in the early stages (Day 1 and Day 3), with a notable presence only emerging by Day 7. These findings suggest that Schwann cell recruitment and proliferation are delayed in transection injury, potentially affecting the overall repair process. Mesenchymal stem cells exhibited a distinct pattern in both models (Fig. [271]9b−e). In crush injury, they were continuously present throughout the repair timeline, suggesting a sustained role in tissue remodeling and repair. In contrast, in transection injury, mesenchymal stem cells were present at lower levels during the early stages (Day 1 and Day 3) but showed a substantial increase by Day 7, indicating a delayed yet significant involvement in the repair process. The enrichment analysis of shared upregulated genes between crush and transection injuries highlighted several key pathways (Fig. [272]9g). On Day 1, common pathways included leukocyte migration, positive regulation of cytosolic calcium ion concentration, interferon−mediated signaling, and toll−like receptor signaling. By Day 3, pathways related to chromosome segregation, multicellular homeostasis, and M phase regulation were enriched, reflecting active cell division and immune responses. On Day 7, pathways such as chromosome segregation and structural disruption in another organism remained prevalent, indicating ongoing cellular proliferation and inflammatory responses. In contrast, pathways specifically enriched in transection injury revealed distinct molecular mechanisms (Fig. [273]9h). On Day 1, pathways such as regulation of angiogenesis, extracellular matrix organization, inflammatory response modulation, and neuron apoptotic processes were significantly enriched, suggesting a robust early vascular and immune response. By Day 3, TGF− Inline graphic receptor signaling and potassium ion transport were notably active, highlighting their role in tissue remodeling. On Day 7, pathways involved in cytosolic calcium ion regulation, interleukin−1 response, and fatty acid biosynthesis were enriched, indicating a shift toward metabolic and immune modulation during later repair phases. These findings underscore the fundamental differences between crush and transection injuries in terms of immune cell recruitment, Schwann cell involvement, and mesenchymal stem cell dynamics. The enrichment analysis further elucidates the distinct molecular pathways governing nerve repair in each model, with transection injury demonstrating a more complex and delayed regenerative process. The differential activation of immune and metabolic pathways suggests potential therapeutic targets to enhance nerve regeneration in severe injury conditions. Discussion The present study integrates our single−cell transcriptomic findings into a comprehensive narrative of sciatic nerve repair post−transection, revealing a dynamic, multi−phasic process that diverges from traditional theories of nerve regeneration. In contrast to classical views that primarily emphasize Wallerian degeneration and subsequent axonal regrowth, our results demonstrate that multiple cell types participate in a temporally coordinated response. Notably, the immediate post−injury phase (Day 1–3) is marked by a significant influx of immune cells. During this critical window, distinct Mac subsets emerge—characterized by the expression of Slpi [[274]70] and Ass1 [[275]71]—alongside Gran identifiable by markers such as S100a8/9 [[276]72], Anxa1 [[277]73], Mmp8 [[278]74], and Pglyrp1 [[279]75]. These observations concur with earlier studies indicating that early immune activation is indispensable for efficient debris clearance via phagocytosis and proteolytic mechanisms. As repair commences, the inflammatory landscape gradually shifts. Immune cells, which dominate the early stages with over 95% of the cell population by Day 1, give way to a robust re−emergence of NFs and Glis. Detailed subclustering of NF populations reveals that the NFs, which are predominant in the uninjured nerve, begin to undergo significant reprogramming post−injury. By Day 5, a specialized NF subpopulation—NF4, distinguished by high Lrg1 [[280]54, [281]55] and Plvap [[282]53, [283]76] expression—appears. This NF4 subset is hypothesized to be a pivotal intermediary, potentially serving as a key transitional cell type that integrates signals from the early immune response and fosters the subsequent emergence of mature fibroblasts, Glis, and vascular cells. Such a role is consistent with the observation that NF4 marks the transformation of NFs from a proliferative state towards a more differentiated, mature phenotype as part of the evolving tissue repair process. In parallel, NF0 cells—marked by Spon1 and peaking at Day 5—play a distinct and essential role in remodeling the extracellular matrix and stabilizing the regenerating tissue microenvironment. These cells are classified as differentiating mesenchymal cells and are specifically enriched in pathways related to extracellular matrix organization, TGF− Inline graphic receptor signaling, and nuclear receptor−mediated transcriptional regulation. Unlike NF4, which is characterized by metabolic adaptation and vascular−associated gene expression, NF0 cells contribute to structural remodeling and act as mediators in the transition from inflammatory response to regenerative stabilization. Their gradual increase during the mid−to−late phase of repair (Day 7 to Day 14) further highlights their sustained involvement in tissue reorganization and suggests that NF0 fibroblasts may act as a scaffold−forming subpopulation that collaborates with other cell types to reestablish nerve architecture. Together, the complementary roles of NF0 and NF4 underscore the temporal coordination between ECM remodeling and metabolic reprogramming during sciatic nerve repair, revealing the functional heterogeneity within the NF lineage and emphasizing the potential of NF0 as a therapeutic target for enhancing matrix regeneration. Concurrently, Glis exhibit significant phenotypic transitions that underscore their essential role in restoring nerve function. In uninjured tissue, myelinating Schwann cells dominate; however, following injury, repair−associated subtypes rapidly emerge. The initial surge of Gli0 cells, characterized by the upregulation of regeneration−associated genes such as Clcf1 [[284]77] and Runx [[285]78, [286]79], facilitates early axonal guidance and extracellular matrix reorganization. As regeneration progresses, a shift occurs with the increase of Gli2 cells, which express markers including Mag [[287]80] and Cldn19 [[288]81], signifying a transition towards remyelination and stabilization of the axonal architecture. This sequential switch—from an initial repair−focused profile to one committed to remyelination—underscores a dynamic process in which Gli functionality is finely tuned to meet the evolving requirements of the regenerating nerve. Vascular remodeling also plays a critical role in the repair process. Although vascular cells are relatively sparse in the immediate aftermath of injury, ECs begin to accumulate from Day 3 onward, and by Day 7, there is a dramatic surge in PC numbers. The robust intercellular communication between Glis and vascular cells, coupled with the emergence of PC subsets involved in TGF− Inline graphic signaling, establishes a supportive microenvironment essential for angiogenesis. Such vascular expansion not only restores blood flow but also creates a niche that is vital for sustaining axonal regrowth and overall tissue homeostasis. Moreover, the trajectory analysis of Mo−derived Macs reveals distinct differentiation pathways that are intimately linked with the regenerative process. Early Mac subsets engaged in inflammatory clearance gradually give way to populations involved in tissue remodeling and extracellular matrix deposition. The interplay between these immune cells and NFs, mediated by enhanced collagen signaling, highlights an integrated network that regulates both inflammation resolution and tissue repair. To further validate the cell type–specific TGF− Inline graphic signaling activation observed in our scRNA−seq analysis, we performed bulk RNA−seq on anatomically defined sciatic nerve segments. Consistent with our single−cell findings, we observed sustained and spatially distinct activation of TGF− Inline graphic signaling pathways, particularly in the proximal stump. Enrichment analysis and GSEA revealed prolonged TGF− Inline graphic activity at Day 14 in proximal regions, whereas distal segments showed a more transient response. This spatial asymmetry may reflect differing regenerative capacities or mechanical environments across nerve segments and supports the notion that proximal Glis and NFs maintain a prolonged pro−regenerative state. These bulk transcriptomic results provide orthogonal validation of the single−cell data and underscore TGF− Inline graphic signaling as a key regulator orchestrating extracellular matrix remodeling, immune modulation, and Gli reprogramming during peripheral nerve repair. In summary, our time−resolved single−cell atlas of sciatic nerve transection injury reveals a coordinated, multi−phasic repair program that progresses through three principal biological phases: early immune activation, extracellular matrix remodeling, and Schwann cell−driven remyelination. Initially, the robust infiltration of specialized Macs and Grans not only facilitates debris clearance but also establishes a pro−regenerative cytokine environment. This is followed by a transitional phase marked by the emergence of NF4 fibroblasts and proliferative mesenchymal subsets, which remodel the extracellular matrix through TGF− Inline graphic and collagen−related signaling, setting the foundation for tissue repair. In the later phase, Glis exhibit dynamic fate transitions—from repair−associated Gli0 to myelinating Gli2 subtypes—underscoring their essential role in axonal ensheathment and functional restoration. Importantly, our cross−species integration with human neurofibroma data highlights conserved PTN signaling between NFs and Glis, implicating a broader relevance of the NF–Gli axis in both regenerative and pathological contexts. PTN exerts its effects by binding to receptors such as PTP1 and ALK, thereby moduating signaling pathways involved in neuronal migration, survival, and extracellular matrix remodeling [[289]82]. Its significant upregulation following peripheral nerve injury further supports its critical role in orchestrating a pro−regenerative microenvironment [[290]83]. Moreover, by comparing crush and transection injury models, we demonstrate that the delayed engagement of Schwann cells and mesenchymal subtypes in transection injury likely contributes to impaired regeneration, pointing to a narrower therapeutic window for intervention. Together, these findings provide not only a granular understanding of the cellular and molecular mechanisms orchestrating nerve repair, but also identify temporal checkpoints—such as early Mac heterogeneity, mid−phase NF4 activation, and late Schwann cell remyelination—as actionable targets for stage−specific therapeutic modulation. Notably, the sustained and region−specific activation of TGF− Inline graphic signaling revealed by both single−cell and bulk transcriptomic analyses highlights this pathway as a unifying axis that coordinates immune regulation, extracellular matrix remodeling, and Schwann cell fate transitions. This work lays a foundation for developing time−tuned regenerative therapies tailored to the specific needs of each phase of peripheral nerve repair. Conclusion In conclusion, our study establishes a comprehensive, time−resolved single−cell atlas of sciatic nerve transection injury, revealing three sequential and highly coordinated regenerative phases. In the acute phase (within 24 h), the lesion environment is dominated by infiltration of pro−inflammatory Macs and Grans, alongside the expansion of proliferative mesenchymal fibroblasts (NF5), which jointly initiate debris clearance and pro−regenerative cytokine signaling. During the intermediate phase (Days 3–7), activated fibroblasts (NF4/NF0) and repair Schwann cells (Gli0) drive extracellular matrix remodeling and reinitiate axonal guidance, largely via TGF− Inline graphic and collagen signaling pathways. In the late phase (Day 7 onward), Glis progressively transition into myelinating phenotypes (Gli2/5), vascular cells expand to re−establish perfusion, and Macs adopt resolution−associated profiles (Mac3/4), culminating in the reconstruction of nerve architecture by Day 14. Importantly, comparative analysis with crush injury revealed that transection triggers a more intense early immune response and delays Schwann cell remyelination, narrowing the therapeutic window for intervention. Notably, bulk RNA−seq of proximal and distal nerve segments confirmed the spatially sustained activation of TGF− Inline graphic signaling, reinforcing its central role in orchestrating the immune–stromal–glial axis during regeneration. These findings define temporal checkpoints and molecular targets—such as early Mac heterogeneity, mid−phase NF activation, and late Gli differentiation—that may guide the development of time−adapted regenerative strategies for peripheral nerve injury. Electronic supplementary material Below is the link to the electronic supplementary material. [291]Supplementary Material 1^ (3.5MB, docx) [292]Supplementary Material 2^ (1.4MB, docx) [293]Supplementary Material 3^ (1.7MB, docx) [294]Supplementary Material 4^ (2.6MB, docx) [295]Supplementary Material 5^ (2.9MB, docx) [296]Supplementary Material 6^ (2.6MB, docx) [297]Supplementary Material 7^ (2.1MB, docx) [298]Supplementary Material 8^ (2.8MB, docx) [299]Supplementary Material 9^ (2.4MB, docx) Acknowledgements