Profibrotic macrophage-derived CXCL4 promotes pericyte-to-myofibroblast transition after spinal cord injury
Gang Li, Le Wang, Xiaoyu Wu, Xiaolin Zeng, Lingli Long, Wenwu Zhang, Jiewen Chen, Di Zhang, Xi Chen, YiLong Deng, XinZhi, Yong Wan, Xiang Li
TL;DR
This study shows that a protein called CXCL4, produced by macrophages, causes pericytes to turn into myofibroblasts, leading to spinal cord injury scarring, and blocking it improves recovery.
Contribution
The study identifies CXCL4 as a key driver of pericyte-to-myofibroblast transition in spinal cord injury and demonstrates its therapeutic potential.
Findings
CXCL4 is produced by Spp1+Fn1+ macrophages and promotes pericyte transition into myofibroblasts via CXCR3/PI3K/Akt signaling.
Blocking CXCL4 or PI3K reduces fibrotic scarring and improves axonal regeneration and motor recovery in SCI mice.
Abstract
Spinal cord injury (SCI) induces fibrotic scarring that impairs axonal regeneration. Pericytes contribute to scar formation via pericyte-to-myofibroblast transition (PMT), yet the mechanisms underlying PMT in SCI remain unclear. Although CXCL4, a pleiotropic chemokine, is implicated in various fibrotic disorders, its role in driving PMT post-SCI remains unexplored. To investigate whether CXCL4 drives PMT after SCI, elucidate its mechanisms, and assess its therapeutic potential. scRNA-seq characterized cell-type dynamics and profibrotic signals in injured mouse spinal cords. In vitro, primary pericytes were exposed to exogenous CXCL4 or co-cultured with Spp1+Fn1+ macrophages. PMT was evaluated by RT-qPCR, Western blot, immunofluorescence, and flow cytometry. PI3K/Akt inhibition or CXCR3 knockdown dissected signaling pathways. In vivo, intrathecal injections of a CXCL4-neutralizing…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSpinal Cord Injury Research · Neurogenesis and neuroplasticity mechanisms · Mesenchymal stem cell research
Introduction
1
Spinal cord injuries often result in incomplete recovery due to limited axonal regeneration [1]. This failure is driven by scar formation, persistent inflammation, inhibitory proteoglycans, and myelin debris [[2], [3], [4]]. Among these, scar formation presents a major physical and biochemical barrier. The scar comprises two principal components: a glial scar surrounding the lesion core and a dense fibrotic scar occupying the lesion center [5,6]. This response is initiated by injury-induced inflammation and progresses into organized scarring [7]. Although CNS scars have long been known to comprise multicellular structures consisting of reactive astrocytes and various non-neural fibrotic cell types [8], much of the research has mainly focused on the glial component.
Fibrosis is characterized by myofibroblast activation and excessive extracellular matrix (ECM) deposition. While myofibroblasts are essential for physiological tissue repair, such as wound healing, their persistence can lead to fibrosis [9,10]. Myofibroblasts, which synthesize the majority of the ECM, are considered the primary drivers of fibrotic disease and organ dysfunction.
Previously, reports have indicated that myofibroblasts can derive from multiple cell types, including epithelial cells and leukocytes [11,12]. Recent studies have shown that pericytes can be switched to myofibroblasts through a process of trans-differentiation known as pericyte-to-myofibroblast transition (PMT) [13,14]. As resident stromal cells lining the capillaries, pericytes serve as a significant source of myofibroblasts during fibrogenesis, contributing to the deposition of pathological matrix [[15], [16], [17]]. Additionally, multiple reports have highlighted the role of pericytes in the lung, retina, skeletal muscle, cancers, and Alzheimer's disease, providing strong evidence that these cells may be promising therapeutic targets [[18], [19], [20], [21], [22]].
Chemokine C-X-C motif ligand 4 (CXCL4), also known as platelet factor 4, exerts both anti-tumor effects by inhibiting angiogenesis and pro-tumor effects by modulating the immune micro-environment [23,24]. It is generally believed that CXCL4 is primarily stored in tumor cells and platelets [25]. CXCL4 is a multifunctional chemokine intricately involved in numerous biological processes, including the regulation of angiogenesis, immune responses, and pro-inflammation activity [24,26]. CXCL4 has been shown to be increased in inflammatory diseases, such as inflammatory bowel disease, atherosclerosis, and rheumatoid arthritis, as well as in fibrotic disorders, including chronic liver fibrosis and cystic fibrosis [[27], [28], [29], [30]]. Additionally, CXCL4 inhibits the expression of the transcription factor FLI1, which negatively regulates collagen synthesis [31]. Moreover, CXCL4 is associated with the induction of fibroblast differentiation, collagen synthesis and fibrosis development [32]. However, the role of CXCL4 and its relevance to fibrosis development following SCI remains largely unexplored.
To elucidate the contributors to fibrous scar formation, we leveraged scRNA-seq to uncover the cellular landscape in spinal cord injury and reported a previously unidentified pericyte subset characterized by Pdgfrb and Acta2. These cells possess the capacity to generate myofibroblasts, which can effectively reconstitute the entire fibrotic niche in vivo. Targeted inhibition of CXCL4 expression using antibodies can suppress the expression of the myofibroblast phenotype in pericytes, reduce fibrous scar formation after SCI in mice, promote axon regeneration, and improve long-term neurological function. These results indicate that CXCL4 may act as a key potential target for the treatment of fibrous scar excessive formation following SCI.
Methods and materials
2
scRNA-seq dataset
2.1
scRNA-seq data was downloaded from the GEO database (GSE162610) and clustered by the Seurat package (version 4.1.0). Highly variable genes (HVGs) were calculated using Seurat ‘FindVariableGenes’ function. Then we performed principal component analysis (PCA) using HVGs and significant top 20 principal components (PCs) were selected to perform t-SNE dimensionality reduction. We chose ‘Harmony’ to remove batch effects. t-SNE was used to visualize the single cells (total of 66,178 cells). Unbiased clustering generated 16 main clusters and was annotated to 10 known cell types according to canonical marker genes. Differential expression genes (DEGs, adjusted p < 0.05, |log2 fold change| >0.5) were identified by the ‘FindMarkers’ function in the Seurat package. Gene Set Variation Analysis (GSEA) was conducted using ClusterProfiler package in R (version 4.2.2) and Gene Set Variation Analysis (GSVA version 1.42.0). The gene sets of Gene Ontology (GO) pathways were obtained from MSigDB (https://www.gsea-msigdb.org/gsea/index.jsp). Pseudotime analysis was conducted using Monocle package (version 2.22.0). Cell–cell interactions were inferred by CellChat package (version 1.13).
Bulk RNA-seq and analyses
2.2
For bulk RNA-sequencing analysis of pericytes treated with CXCL4 or vehicle, total RNA was isolated using a Trizol reagent kit (Invitrogen) according to the manufacturer's protocol. Then, the enriched mRNA was fragmented into short fragments using fragmentation buffer and reverse transcribed into cDNA by SuperScript™ II Reverse Transcriptase (Invitrogen, cat. 1,896,649, USA). The resulting cDNA library was sequenced using Illumina NovaSeq6000 (LC-Bio Technology CO., Ltd., Hangzhou, China). To obtain high-quality clean reads, reads were further filtered by fastp (version 0.18.0). DESeq2 software was used to identify DEGs between two different groups (false discovery rate (FDR) < 0.05 and absolute fold change ≥2).
Animals
2.3
The animal experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University (Approval No: SYSU-IACUC-2022-003047). Adult female C57BL/6 J mice (aged 8–10 weeks, purchased from the Experimental Animal Center of Sun Yat-sen University, Guangzhou, China, animal license No. SCXK (Yue) 2021–0029) were used to establish a SCI animal model. The animals were housed in a temperature and humidity-controlled environment with a 12-h light–dark cycle and food/water available ad libitum.
Complete spinal cord crush model
2.4
All mice were specific pathogen-free grade and were divided into groups: sham group, 1-day post-injury (dpi) group, 3 dpi group, 7 dpi group, SCI + LY294002 group, and SCI + Anti-CXCL4 group. Briefly, mice were anesthetized by intraperitoneal injection of 1 % pentobarbital sodium (50 mg/kg, Baiwei Biology, Guangzhou, China). A laminectomy was performed to expose the dorsal portion of the spinal cord at thoracic level 10 (T10). The spinal cord was then fully crushed for 2 s with Dumont No.5 forceps (11,295–00, Fine Science Tools) without spacers and that had been filed to a width of 0.1 mm for the last 5 mm of the tips. By minimizing mechanical disruption of the dura, this injury model limits confounding factors such as invasion of dura-derived meningeal fibroblasts into the lesion [13].
Drug injection in mice
2.5
For intrathecal injection, spinal cord puncture was performed with a microinjection needle (1701, Hamilton, Switzerland) at the dorsal midline between the lumbar 5–6 intervertebral space [33]. Successful penetration of the needle into the intradural space was confirmed by the observation of a sudden tail flick response. Anti-CXCL4 neutralizing antibody (ab303494, USA) was intrathecally injected at a dose of 4.5 μg/10 μl, and LY294002 (HY-10108, MCE, USA), a specific PI3K antagonist, was dissolved in 10 % dimethyl sulfoxide (DMSO) and injected at a dose of 10 μg/5 μl [34]. Both injections were started 4 h post-injury and administered once daily until 7 dpi [35,36].
For CST tracing, animals were head-fixed in a stereotaxic apparatus (RWD Technology Corp., Ltd, Shenzhen, China) and burr holes were drilled over the sensorimotor cortex. Four microinjections (0.4 μl per injection) of virus (ScAAV-hSy-EGFP WPREs, titer 1-5E12 genome copies per mL, Brain VTA Co., Ltd., Wuhan, China) were targeted to layer V in the sensorimotor cortex (two injections/hemisphere; from bregma, AP ± 0.5 mm, LM 1 mm, DV 0.55 mm and AP ± 0.5 mm, LM -1 mm, DV 0.55 mm) at a rate of 0.1 μl/min with a microinjector. After injection, the needle was kept for an additional 5 min in place to allow virus diffusion and prevent backflow to the surface, and then slowly withdrawn [37]. For postoperative recovery, the animals were placed on a heating pad until fully awake before being returned to their cages. For experiments that terminated 18 weeks after the lesion model, mice were injected with virus 16 weeks after the SCI surgery.
Chromatin Immunoprecipitation Assay
2.6
According to the Magna ChIP™ protocol (Merck Millipore, Billerica, Mass), Spp1^+^Fn1^+^ macrophages were fixed with 1 % formaldehyde at room temperature for 10 min. Glycine was added to cells to quench unreacted formaldehyde for 5 min. Cell pellets in SDS Lysis Buffer containing 1X Protease Inhibitor Cocktail were sonicated at 4 °C to yield fragments from 100 to 750 bps. Immunoprecipitation was performed with anti-Mafb antibody (1:100, 20189-1-AP, Proteintech) and ChIP-grade Protein A/G Magnetic Beads (Merck Millipore). Protein-DNA crosslinks were reversed at 65 °C and DNA was purified for qPCR. The primers used were listed in Supplementary Table 1 (Table S1).
Small Interfering RNA-mediated Gene Silencing
2.7
Cxcl4-small interfering RNA (siRNA), Cxcr3-siRNA and Mafb-siRNA and a nontargeting RNA were synthesized by RiboBio. Cells were transfected using Lipofectamine 3000 (Thermo Fisher Scientific). The siRNA sequences are summarized in Supplementary Table 2 (Table S2).
CXCL4 Measurements
2.8
Supernatants were collected from cell culture and tissue lysis, then assayed for cytokine production. CXCL4 levels were assessed using NeoBiosence ELISA kits (Shenzhen, China) following the manufacturer's protocol. The data were acquired by using a Multiskan Sunrise microplate reader (TECAN, Austria).
Immunofluorescence staining
2.9
Animals were anesthetized with an intraperitoneal injection of pentobarbital sodium (50 mg/kg) and perfused with phosphate-buffered saline (PBS, BL601A, Biosharp, China) followed by 4 % paraformaldehyde (PFA, G1101, Servicebio, China). Spinal cord tissues were dissected and postfixed in 4 % PFA overnight at 4 °C. The tissues were dehydrated in a 30 % sucrose solution for 48 h and then embedded in OCT compound (BL557A, Biosharp, China). Sagittal sections (16 μm) of spinal cords were collected using a freezing microtome (NX70, Thermo Fisher, USA). Spinal cord sections were blocked in 5 % donkey serum in PBS containing 0.3 % Triton X-100 (T8200, Solarbio, China) for 1 h at room temperature and then incubated with primary antibodies in 1 % donkey serum containing 0.3 % Triton X-100 overnight at 4 °C. Then, secondary antibodies were used to incubate the sections for 1 h. 4′,6-Diamidino-2-phenylindole, dilactate (DAPI, P0126, Beyotime Biotechnology, China) was used for nuclear staining. Immunofluorescence images were acquired with an inflorescent microscope system (Axio Scope A1, Zeiss, Germany) and a confocal microscope (LSM 900, Zeiss, Germany). Images were created by Zen 3.3 software (Blue edition). Antibodies used are listed in Supplementary Table 3 (Table S3).
Masson's trichrome staining
2.10
In brief, the expression of collagen fibrils in the injured spinal cord was evaluated using a Masson trichrome stain reagent (Sigma–Aldrich). A pathological slice scanner (Leica, Germany) was used to collect images. Three areas of each sample were randomly selected.
Evaluation of functional recovery in SCI mouse
2.11
The recovery of spinal cord function in mice was evaluated by Basso Mouse Scale (BMS) score and footprint analysis. The mice were placed on a platform that allowed them to move freely. Two evaluators who were blinded to the experimental conditions recorded the hindlimb walking and limb movement and used the average to evaluate the BMS score. For footprint analysis, hindlimb locomotor evaluation was performed at 18 weeks. After applying red paint to the hind limbs, the mice walked on a 7.5 cm × 100 cm track. Stride length, stride width, and BMS score were analyzed to evaluate motor function.
Flow cytometry and cell isolation
2.12
Isolated spinal cord cells were prepared from 8-week-old C57BL/6 J mice. Tissue fragments were digested with collagenase (HY-E70005D, MCE, USA) and DNase (HY-108882, MCE, USA) at 37 °C for 1 h, followed by filtration through a 70um cell strainer (352,350, BD Biosciences).
For pericyte isolation, cells were incubated with PE anti-mouse CD140b (1:100, 12-1402-81, Invitrogen, USA) and APC anti-mouse CD13 (1:100, 164,005, BioLegend, USA) on ice for 30 min. To further discriminate pericyte subpopulations, cells were fixed and permeabilized, then stained with CoraLite® Plus 488-conjugated smooth muscle actin antibody (1:50, CL488-80008, Proteintech, China), allowing separation of Pdgfrβ^+^Acta2^+^ and Pdgfrβ^+^Acta2^-^ populations. The labeled cells were sorted using a fluorescence-activated cell sorter (FACS) (Beckman Coulter) (Fig. S1A) [38,39].
For macrophage isolation, cells were incubated with FITC anti-mouse CD45 (1:100, 103,107, BioLegend, USA) and APC/Cyanine7 anti-mouse F4/80 (1:100, 157,315, BioLegend, USA) on ice for 30 min. To further discriminate macrophage subpopulations, cells were fixed and permeabilized as above, followed by staining with CoraLite® Plus 488-conjugated osteopontin polyclonal antibody (1:50, CL488-22952, Proteintech, China) and Alexa Fluor® 647 mouse anti-Fibronectin (1:50, BD Biosciences, CA) (Fig. S1B). Sorted macrophages were used for downstream assays.
Real-time Quantitative PCR
2.13
Total RNA was extracted from cells and tissues using Trizol reagent (Invitrogen). Complementary DNA (cDNA) was synthesized from total RNA using a PrimeScript RT reagent kit (TaKaRa) according to the manufacturer's protocol. Quantitative real-time PCR was performed to amplify the cDNA on a Light Cycler 480 Real-time PCR system (Roche) using TB Green Premix Taq II (TaKaRa). All gene expression values were normalized to GAPDH and calculated using the 2^−ΔΔCt^ Method. Primers used are listed in Supplementary Table 4 (Table S4).
Western blot analysis
2.14
Total protein was extracted using radioimmunoprecipitation assay (RIPA) lysis buffer (Sigma Aldrich, USA) containing a protease inhibitor and phosphatase inhibitor cocktail (Roche). Protein concentrations were determined using a bicinchoninic acid (BCA) kit (Thermo Fisher, USA) and the samples were diluted to 2 mg/ml. Equal amounts of protein from each sample were separated by 10 % and 15 % SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (GE Amersham, USA). After blocking for 1 h in 5 % (m/v) milk, blots were incubated with primary antibody at 4 °C overnight, followed with secondary antibody for 1 h at room temperature (RT), and washed in Tris-buffered saline with 0.2 % Tween-20 in between steps. Protein bands were developed using ECL (Biosharp, China) and measured on ImageQuant Las4000mini (GE, Japan). Antibodies used are listed in Supplementary Table 5 (Table S5).
Chemicals and reagents
2.15
CXCL4 (250-39-20UG, mouse) was purchased from Peprotech. LY294002 (HY-10108) and MK2206 (HY-108232) were purchased from MCE.
Statistical analysis
2.16
All data were expressed as mean ± SD. The statistical analysis was performed using GraphPad Prism 8. The significant differences were determined by 2-tailed Student's t test (2-group comparisons) or one-way analysis of variance (ANOVA) with Tukey's post hoc test (multiple group comparisons) where appropriate. P < 0.05 was considered statistically significant.
Results
3
Pericyte-myofibroblast transition (PMT) occurs in response to spinal cord injury
3.1
To investigate the cellular constituents of the micro-environment in the lesion following SCI, we analyzed a publicly available scRNA-seq dataset from the spinal cord. A total of 66,178 cells were obtained from both uninjured tissues and tissues at 1, 3, and 7 days post-injury (dpi), resulting in a total of 15 distinct clusters when visualized on a tSNE plot (Fig. S2A). These clusters encompassed all major cell types present at the injury site, including microglia, macrophages, endothelial cells, astrocytes, OPCs, div-myeloid cells, pericytes, oligodendrocytes, lymphocytes, fibroblasts and neurons (Fig. 1A). Cell clustering and annotation were performed based on canonical marker genes, following established protocols [40] (Fig. 1B).Fig. 1. Pericyte-myofibroblast transition (PMT) occurs in response to spinal cord injury. A tSNE visualization plot showing cell types for the 66,178 cells from spinal cord tissue of uninjured and 1, 3, 7 dpi groups. B Dot plots showing the smoothed expression distribution of marker genes across the 11 identified cell types. C Dot plots showing pericyte activity scores across all clusters, as assessed by AUCell, Ucell, singscore, AddModuleScore, and Scoring algorithms. D tSNE visualization plot showing pericyte activity across all clusters. E Violin plots showing Pdgfrb and Acta2 expression across distinct cell clusters. F Dot plot showing the expression of myofibroblast genes across distinct cell clusters. G Violin plots showing Pdgfrb and Acta2 expression in pericytes between the sham and SCI groups. H Representative fluorescence images of PDGFRβ^+^ (green), α-SMA^+^ (red) and DAPI (blue) cells at the injured site of spinal cord (7 dpi). Scale bar: 200 μm (left), 20 μm (right). I Quantification of immunofluorescence data in panel H, n = 5. J Representative fluorescence images of PDGFRβ^+^ (green), Collagen I^+^ (red) and DAPI (blue) cells at the injured site of spinal cord (7 dpi). Scale bar: 200 μm (left), 20 μm (right). K Quantification of immunofluorescence data in panel J, n = 5. All data are presented as means ± SD. ∗∗∗p < 0.001 compared between groups by unpaired Student's t test.Fig. 1
To validate the accuracy of pericyte annotation based solely on Pdgfrb expression, which is also expressed by fibroblasts, we curated separate gene sets for pericytes and fibroblasts based on highly expressed marker genes reported in published studies [[40], [41], [42], [43], [44], [45], [46]], excluding shared markers such as Pdgfrb to reduce cross-lineage interference. Multiple scoring algorithms-AUCell, Ucell, SingScore, AddModuleScore, and Scoring-were then employed to assess pericyte and fibroblast activity across clusters. These algorithms provided an integrated approach to assess pericyte identity more robustly, circumventing the limitations of using Pdgfrb expression alone. All scoring methods consistently identified cluster 7 as exhibiting the highest pericyte signature, while other clusters showed lower levels of activity (Fig. 1C and Fig. S2B). This finding was further supported by tSNE visualization, which highlighted the highest pericyte activity within cluster 7 (Fig. 1D). In parallel, fibroblast signature activity was primarily detected in cluster 13, with minimal activity in cluster 7, further supporting the specificity of the pericyte identity (Fig. S2C–E).
Emerging evidence indicates that pericytes undergo pericyte-to-myofibroblast transition, contributing to fibrotic progression in diseases such as renal and cardiac fibrosis [47,48]. Building on these findings, we further investigated whether a similar transition occurs following SCI. Our analysis revealed that markers of myofibroblasts, including Acta2 and other associated genes, were significantly expressed in Pdgfrb^+^ pericytes (Fig. 1E–F), as defined by our scoring algorithms. Following SCI, Pdgfrb expression decreased in pericytes, while Acta2 exhibited a significant increase compared to the sham group (Fig. 1G). These findings suggest that Pdgfrb^+^ pericytes undergo a transition to myofibroblasts during SCI. To further validate these findings, immunofluorescence revealed a significant increase in the expression of myofibroblast markers in pericytes in the injured spinal cord at 7 dpi compared to the sham group (Fig. 1H–K). These results further indicate that PMT occurs in response to SCI, implicating their potential role in the fibrotic scar formation.
Pdgfrβ+Acta2+ pericytes are the primary contributors to PMT following SCI
3.2
To further elucidate the role of pericytes in contributing to PMT, we re-clustered Pdgfrb^+^ cells to explore the functional heterogeneity (Fig. 2A). Given that vascular smooth muscle cells (vSMC) also express Pdgfrb [45], we first assessed whether these clusters belonged to the pericyte lineage. Using curated gene sets, we performed pericyte and vSMC signature scoring across all clusters. The results confirmed that each cluster exhibited higher pericyte signature activity than vSMC signatures, supporting their classification as pericytes (Fig. S3). Notably, cluster 3 displayed slightly reduced pericyte activity compared to other pericyte clusters, raising the possibility of a transitional state.Fig. 2. Pdgfrβ^+^Acta2^+^ pericytes are the primary contributors to PMT following SCI. A tSNE visualization plot showing pericytes subsets, including cells from uninjured group and SCI groups (1, 3, 7 dpi groups). B Violin plots depicting the expression levels of pericyte genes (top) and myofibroblast genes (bottom) across distinct pericyte clusters. C-D Clustering of pericytes based on the expression of myofibroblast marker Acta2 (normalized expression level >0) reveals a population (cluster 3) with transcriptional wiring indicative of pericytes undergoing PMT. E Lineage trajectory analysis demonstrating the lineage relationship between pericytes and myofibroblasts. Pseudotime analysis sorts pericytes in right branches and myofibroblasts in left branches. F GO enrichment of the top 200 highly expressed genes in pericyte cluster 3, including three ontologies: biological process (BP), cellular component (CC), and molecular function (MF). G Flow cytometric analysis of pericyte cluster 3 and other clusters from spinal cord tissue at 7 dpi, n = 5. H Immunofluorescence analysis of PDGFRβ^+^ (green), α-SMA^+^ (red) and DAPI (blue) cells between cluster 3 and other clusters isolated from spinal cord tissue at 7 dpi. Scale bar: 100 μm. I Quantification of immunofluorescence data in panel H, n = 5. All data are presented as means ± SD. ∗∗∗p < 0.001 compared between groups by unpaired Student's t test.Fig. 2
To further characterize this subpopulation, we next examined gene expression profiles across all Pdgfrb^+^ clusters. Cluster 3 exhibited decreased expression of pericyte markers (Pdgfrb, Rgs5, and Mcam) with concomitantly increased expression of myofibroblast markers (Acta2, Myh11, and Vim) (Fig. 2B), consistent with a phenotypic shift toward a myofibroblast-like state. Of note, Acta2 was significantly overexpressed in cluster 3, where the PMT-associated PI3K/Akt signaling pathway was enriched (Fig. 2C-D) [49].
To further corroborate the notion that the PMT population is indeed derived from pericytes and represents an intermediate differentiation stage toward myofibroblasts, we performed a trajectory analysis, which sorts pericyte and myofibroblasts at distinct extremes of pseudotime space (Fig. 2E). Interestingly, the pericytes expressing Acta2 (cluster 3) were predominantly situated along the branch between pericytes and myofibroblasts with a stepwise increase in expression of myofibrotic markers during the pseudotime trajectory (Fig. 2E), supporting the notion that some pericytes may progressively acquire myofibroblast identity.
To determine the biological roles of the top 200 highly expressed genes in cluster 3, we performed GO term enrichment analysis across three categories: biological process (BP), cellular component (CC), and molecular function (MF). The enriched functional terms predominantly related to extracellular matrix remodeling, including processes such as extracellular matrix binding, laminin binding, and collagen-containing extracellular matrix (Fig. 2F). Flow cytometry results further confirmed that pericyte cluster 3 underwent trans-differentiation towards a myofibroblast phenotype (Fig. 2G), as supported by immunofluorescence staining (Fig. 2H–I). These findings further indicate that cluster 3 of pericytes (hereafter called myofibrotic pericytes) are the primary contributors to PMT following SCI.
Increase of CXCL4 after SCI promotes PMT
3.3
To further dissect the initiating factors driving PMT, bulk-seq analysis was performed on spinal cord tissues from uninjured and 3 dpi groups. The data revealed a significant up-regulation of Cxcl4 expression following SCI (Fig. 3A–B). Immunofluorescence staining of spinal cord tissue further confirmed this finding, visually showing elevated CXCL4 levels in the spinal cord tissue (Fig. 3C–D). For quantitative validation, RT-qPCR and Western blot analysis of spinal cord tissue further validated the increase in CXCL4 expression, peaking at 3 dpi before gradually declining (Fig. 3E–G). Similarly, enzyme-linked immunosorbent assay (ELISA) of spinal cord tissue demonstrated a consistent dynamic increase in CXCL4 levels (Fig. 3H), reinforcing its critical role in the response to SCI. Collectively, these findings suggest a significant elevation of CXCL4 following SCI.Fig. 3. Increase of CXCL4 after SCI promotes PMT. A-B Heatmap and violin plots showing Cxcl4 expression between uninjured group and 3 dpi groups. C Immunofluorescence staining of CXCL4^+^ (green), and DAPI (blue) cells in spinal cord tissue from uninjured and 1, 3, 7 dpi groups. Scale bar: 100 μm. D Quantification of immunofluorescence data in panel C, n = 5. E RT-qPCR analysis of expression of Cxcl4 in spinal cord tissue at the injured site at each time point, n = 3. F Western Blot analysis of CXCL4 in spinal cord tissue in different groups. G Quantification of Western blot data in panel F, n = 3. H ELISA analysis of expressions of CXCL4 at the injured site of spinal cord of each time point, n = 3. All data are presented as means ± SD. ∗∗p < 0.01, ∗∗∗p < 0.001 compared between groups by one-way ANOVA followed by Tukey's multiple comparisons tests.Fig. 3
CXCL4 promotes pericyte differentiation in vitro
3.4
Then, we tested whether CXCL4 acts as a driving force of PMT in vitro. Bulk-seq analysis revealed significant gene expression changes in pericytes treated with exogenous CXCL4, with 1445 genes up-regulated and 1449 genes down-regulated (Fig. 4A). Among the up-regulated genes, Acta2 and Col1a1 were notably increased (Fig. 4B). Furthermore, we exposed pericytes to increasing concentrations of CXCL4 [32]. Concurrent with the increased expression of myofibroblast markers Acta2, Col1a1 in pericytes, expression of the pericytes markers Pdgfrb, NG2 was decreased in a dose-dependent manner at mRNA and protein levels (Fig. 4C-E). Immunofluorescence staining corroborated that pericytes not only acquired immunophenotypic changes indicative of PMT but also cellular properties congruent with transition from a pericyte identity toward myofibroblasts fates (Fig. 4F-I). Taken together, these findings suggest that CXCL4 has the ability to convert pericytes into myofibroblasts in vitro.Fig. 4CXCL4 promotes pericyte differentiation in vitro. A Volcano plot showing differentially expressed genes (DEGs) in pericytes with or without CXCL4 treatment. Up-regulated genes are shown in red, and down-regulated genes in blue (fold change <0.6), with P < 0.05. B Heatmap analysis of DEGs in primary pericytes treated with or without CXCL4. C Pericytes were stimulated with different doses of CXCL4 for 24 h. RT-qPCR analysis of expression levels of Pdgfrb, NG2, Acta2, and Col1a1 between different groups, n = 3. D Pericytes were stimulated with different doses of CXCL4 for 48 h. Western Blot analysis of α-SMA, Collagen I, PDGFRβ, NG2 between different groups. E Quantification of Western Blot data in panel D, n = 3. F Immunofluorescence analysis of PDGFRβ^+^ (red), α-SMA^+^ (green) and DAPI (blue) cells between different groups. Pericytes were treated with CXCL4 at a concentration of 5 μg/ml for 48 h. Scale bar: 20 μm. G Quantification of immunofluorescence result in panel F, n = 5. H Immunofluorescence analysis of PDGFRβ^+^ (red), Collagen I^+^ (green) and DAPI (blue) cells between different groups, under the same treatment conditions as in panel F. Scale bar: 20 μm. I Quantification of immunofluorescence data in panel H, n = 5. All data are presented as means ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 compared between groups by one-way ANOVA followed by Tukey's multiple comparisons tests in panel C, E and by unpaired Student's t test in panel G, I.Fig. 4
CXCL4 blockage ameliorates fibrous scar and enhances functional recovery after SCI
3.5
To assess the functional significance of our findings, we conducted immunofluorescent staining at the injured site of the spinal cord with or without anti-CXCL4. We observed enhanced expression of α-SMA, Collagen I, Laminin, and Fibronectin following SCI, which was suppressed upon intrathecal injection of anti-CXCL4 in mice (Fig. 5A–D, and Fig. S4). These results indicate that anti-CXCL4 treatment not only reduced pericyte-to-myofibroblast transition but also led to a significant decrease in pericyte-derived fibrotic scarring. Masson's trichrome staining further corroborated the reduction in fibrosis observed following anti-CXCL4 treatment after SCI (Fig. 5E). Next, we investigated CST axon regeneration in mice 18 weeks after crush injury [37]. Mice received bilateral transduction with ScAAV9-hSy-EGFP-WPREs virus targeting layer V neurons in the sensorimotor cortex (Fig. 5F–G). These neurons project to spinal cord targets and provide excitatory input to the red nucleus and reticular formation, which descend axons into the spinal cord [37]. CST fibers reached the scar core but failed to go beyond the lesion in SCI animals. However, in mice treated with Anti-CXCL4, CST axons exhibited caudal extension, surpassing the lesion boundaries. Moreover, immunofluorescence analysis indicated this observation was in line with the distribution of 5-HT^+^ axons (Fig. 5H–I). Behavioral tests, including BMS motor function scores and footprint analysis, demonstrated that blocking CXCL4 improved the motor function of hindlimbs in SCI mice (Fig. 5J–L). These findings strongly indicate that targeting CXCL4 could attenuate PMT and fibrosis, thereby promoting tissue repair and locomotor functional recovery after SCI.Fig. 5CXCL4 blockage ameliorates fibrous scar and enhances functional recovery after SCI. A Representative fluorescence images of PDGFRβ^+^ (green), α-SMA^+^ (red) cells at the injured site of spinal cord (7 dpi). Scale bar: 200 μm (top), 20 μm (bottom). B Quantification of immunofluorescence data in panel A, n = 5. C Representative fluorescence images of PDGFRβ^+^ (green), Laminin^+^ (red) cells at the injured site of spinal cord (7 dpi). Scale bar: 200 μm (top), 20 μm (bottom). D Quantification of immunofluorescence data in panel C, n = 5. E Masson's trichrome staining (blue for collagen) at the injured site of spinal cord in different groups at 7dpi. F Lesion model and experimental timeline. Red represents the lesion. G Schematic depicting the strategy used to label the CST. H Serial sagittal sections of injured spinal cord demonstrating EGFP ^+^ CST axon (top) and 5-HT^+^ axon (bottom) growth through, around, and past the lesion site in different groups 18 weeks post-injury (wpi). Scale bar: 200 μm. I Cumulative percentage of EGFP ^+^ CST axon (top) and 5-HT^+^ axon (bottom) regeneration at 4 mm caudal to the injury in panel H, n = 5. J BMS scores in different groups. K-L Footprint analysis of different groups, red is forelimb, blue is hindlimb. Quantification of the footprint analysis after SCI, n = 8. All data are presented as means ± SD. ∗p < 0.05, ∗∗∗p < 0.001 compared between groups by one-way ANOVA followed by Tukey's multiple comparisons tests in panel B, D, J, and L, and by unpaired Student's t test in panel I.Fig. 5
CXCL4 promotes PMT through activation of the PI3K/Akt signaling pathway
3.6
To elucidate the underlying mechanism, we performed the functional enrichment analysis using bulk RNA sequencing of pericytes treated with or without CXCL4. GO (Gene Ontology) analysis indicated enrichment of extracellular matrix-related pathways in the pericytes. Notably, KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment analysis revealed significant enrichment of the PI3K/Akt signaling pathway. Given the established role of the PI3K/Akt pathway in scar formation, as evidenced by previous studies [49,50], we focused on exploring the function of this signaling cascade (Fig. 6A). To further analyze the changes in the PI3K/Akt pathway in CXCL4-treated pericytes, we utilized Gene Set Enrichment Analysis (GSEA), which demonstrated an increased enrichment score for PI3K/Akt signaling pathway modulation in response to CXCL4 treatment (Fig. 6B). Consistent with this, we also found that CXCL4 stimulation significantly enhanced the phosphorylation of PI3K (Fig. 6C–D). WB further demonstrated downregulation of PDGFRβ and NG2, accompanied by an upregulation of α-SMA and Collagen I, which were reversed by treatment with the PI3K inhibitor LY294002 or the Akt inhibitor MK2206 (Fig. 6E–F). Immunofluorescent staining confirmed these findings, showing increased expression of α-SMA and Collagen I in CXCL4-treated pericytes, which were rescued by inhibition of the PI3K/Akt pathway (Fig. 6G–H). Similar changes were observed in post-injury spinal cord tissue when mice were intrathecally injected with LY294002 (Fig. 6I-L). Collectively, these results suggest that CXCL4 promotes pericyte differentiation into myofibroblasts through activation of the PI3K/Akt pathway after SCI.Fig. 6CXCL4 promotes PMT through activation of the PI3K/Akt signaling pathway. A GO (left) and KEGG (right) enrichment of the top 200 highly expressed differential genes in primary pericytes treated with or without CXCL4. B GSEA of the enrichment of the PI3K-Akt pathway in bulk-seq data. C Western blot analysis of PI3K and p-PI3K levels in primary pericytes treated with CXCL4 at different time points. D Quantification of western blot data in panel C, n = 3. E Western blot analysis of PDGFRβ, NG2, α-SMA, and Collagen I in CXCL4-treated primary pericytes treated with LY294002 (PI3K, inhibitor), MK2206 (Akt, inhibitor). F Quantification of western blot data in panel E, n = 3. G Immunofluorescence analysis of PDGFRβ^+^ (green), α-SMA^+^ (red) cells (top) and PDGFRβ^+^ (green), Collagen I^+^ (red) cells (bottom) in different groups. Scale bar: 50 μm. H Quantification of immunofluorescence data in panel G, n = 5. I, K Immunofluorescence analysis of PDGFRβ^+^ (green), α-SMA^+^ (red) cells (panel I) and PDGFRβ^+^ (green), Collagen I^+^ (red) cells (panel K) at the injured site of spinal cord (7 dpi). Scale bar: 200 μm (top), 20 μm (bottom). J, L Quantification of immunofluorescence data in panel I, K, n = 5. All data are presented as means ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 compared between groups by one-way ANOVA followed by Tukey's multiple comparisons tests in panel D, F, H and by unpaired Student's t test in panel J, L.Fig. 6
Spp1+Fn1+ macrophage subset is the main source of CXCL4 following SCI
3.7
Next, we investigated the source of CXCL4 following SCI. Clustering and annotation revealed all major immune cell populations in the injured spinal cord (Fig. 7A). To identify profibrotic immune cell populations in an unbiased manner, we scored cells according to their expression of a set of profibrotic extracellular matrix (ECM) regulator genes, as defined by the matrisome project (Fig. 7B) [51]. By assigning these signatures, we found that ECM regulator expression was highest in Spp1^+^ macrophages (Fig. 7B). This subset, characterized by the expression of profibrotic genes (Spp1, Fn1), diverges from the classical M1 and M2 macrophage paradigms, thus being designated as Spp1^+^Fn1^+^ macrophages. The association of these markers with fibrosis underscores their potential role in mediating fibrotic responses following SCI. Interestingly, violin plots showed that Cxcl4 (Pf4) was predominantly expressed in macrophages following SCI, particularly in the Spp1^+^Fn1^+^ macrophage population (Fig. 7C and Fig. S5). Immunofluorescence staining confirmed the significant expression of Spp1^+^Fn1^+^ macrophages at 3 dpi (Fig. 7D-E). Additionally, flow cytometry and subsequently immunofluorescence indicated that Spp1^+^Fn1^+^ macrophage population exhibited markedly higher levels of CXCL4 expression, reinforcing the hypothesis that these macrophages are significant contributors to CXCL4 secretion (Fig. 7F-H). Furthermore, immunofluorescence staining of spinal cord tissue at 3 dpi showed a significant increase in CXCL4 expression within the Spp1^+^Fn1^+^ macrophage subpopulation (Fig. 7I-J). Collectively, these findings strongly suggest that Spp1^+^Fn1^+^ macrophages serve as the primary source of CXCL4 in response to SCI, thereby promoting the progression of fibrosis.Fig. 7. Spp1^+^Fn1^+^ macrophage subpopulation is the main source of CXCL4 following SCI. A tSNE visualization plot showing immune cells from spinal cord tissue of uninjured and 1, 3, and 7 dpi groups. B Featureplot showing ECM regulator score on the tSNE embedding shown in panel A. C Violin plots showing the expression of Cxcl4 (Pf4) across immune cells. D Immunofluorescence analysis of SPP1^+^ (white), FN1^+^ (red), CD68^+^ (green), and DAPI (blue) cells at the injured site of spinal cord (3 dpi). Scale bar: 20 μm. E Quantification of immunofluorescence data in panel D, n = 5. F The levels of CXCL4 were measured by flow cytometry in Spp1^+^Fn1^+^ macrophages and Spp1^+^Fn1^-^ macrophages. Representative histograms are shown. G Immunofluorescence analysis of FN1^+^ (red), CXCL4^+^ (green), and DAPI (blue) cells in Spp1^+^Fn1^+^ macrophages and Spp1^+^Fn1^-^ macrophages isolated from spinal cord tissue at 3 dpi. Scale bar: 20 μm. H Quantification of immunofluorescence data in panel G, n = 5. I Immunofluorescence analysis of SPP1^+^ (white), FN1^+^ (red), CXCL4^+^ (green) and DAPI (blue) cells at the injured site of spinal cord (3 dpi). Scale bar: 20 μm. J Quantification of immunofluorescence data in panel I, n = 5. All data are presented as means ± SD. ∗∗∗p < 0.001 compared between groups by unpaired Student's t test.Fig. 7
Spp1+Fn1+ macrophages promote PMT of myofibrotic pericytes via CXCL4/CXCR3 axis
3.8
As mentioned above, CXCL4 was likely derived from Spp1^+^Fn1^+^ macrophages following SCI. To further investigate the potential interaction between macrophages and pericytes. Clustering analysis was performed on subsets of Spp1^+^ macrophages and Pdgfrβ^+^ pericytes and visualized on a separate tSNE (Fig. 8A). Plotting clusters based on both their incoming and outgoing interaction strengths, confirmed that Pdgfrβ^+^Acta2^+^ pericytes exhibited a significant increase in receiving signals following SCI. Meanwhile, Spp1^+^Fn1^+^ macrophages displayed higher sending signals (Fig. 8B–C). Moreover, through the analysis of cell communication, we found that both the number and strength of interactions between Spp1^+^Fn1^+^ macrophages and Pdgfrβ^+^Acta2^+^ pericytes were prominent in SCI group (Fig. 8D-E). To further dissect these interactions, we calculated the information flow between macrophages and pericytes for each signaling pathway after injury. We found that the CXCL signaling pathway was one of the markedly enriched after SCI (Fig. 8F).Fig. 8. Spp1^+^Fn1^+^ macrophages promote PMT of myofibrotic pericytes via CXCL4/CXCR3 axis. A tSNE visualization plot showing interactions between subsets of Spp1^+^ macrophages and Pdgfrβ^+^ pericytes across uninjured group and SCI groups (1, 3, 7 dpi). B-C Scatter plots showing the signaling role analysis on the aggregated cell–cell communication network from all signaling pathways in normal (B) and injury (C) conditions. D-E Circle plot showing the differential interaction numbers and strengths between subsets of Spp1^+^ macrophages and Pdgfrβ^+^ pericytes following SCI. F Bar plots ranking the signaling pathway axes by overall information flow differences in the interaction networks between normal and injured samples. G Schematic diagram illustrating the establishment of Spp1^+^Fn1^+^ macrophages and Pdgfrβ^+^Acta2^+^ pericytes co-cultured system. The diagram was created using BioRender sofware (https://www.biorender.com). H Immunofluorescence analysis of PDGFRβ^+^ (green), CXCR3^+^ (red), and DAPI (blue) in pericytes. I-J RT-qPCR analysis showing the expressions of Acta2, Col1a1, Pdgfrb, and NG2 between different groups. All data are presented as means ± SD. ∗∗p < 0.01, ∗∗∗p < 0.001 compared between groups by one-way ANOVA followed by Tukey's multiple comparisons tests. ns, not significant.Fig. 8
Given that CXCR3 is the sole receptor for CXCL4 [52], we established a co-culture system of Spp1^+^Fn1^+^ macrophages and Pdgfrβ^+^Acta2^+^ pericytes (Fig. 8G), which were isolated from spinal cord at 3dpi by FACS (Fig. S1A–B). Immunofluorescence staining confirmed the co-expression of PDGFRβ and the receptor CXCR3 in pericytes, providing direct evidence of their interaction (Fig. 8H). Additionally, RT-qPCR results showed that the expression of Acta2, Col1A1 was significantly increased and the expression of Pdgfrb, NG2 was significantly decreased in pericytes co-cultured with Spp1^+^Fn1^+^ macrophages. These effects were reversed upon knockdown of either Cxcl4 or Cxcr3 expression with siRNA (Fig. 8I–J and Fig. S6). These results unequivocally establish the evidence that Spp1^+^Fn1^+^ macrophages promote PMT of myofibrotic pericytes via CXCL4/CXCR3 axis.
CXCL4 expression in Spp1+Fn1+macrophages is regulated by the transcriptional factor MAFB
3.9
To elucidate the regulatory mechanism of Cxcl4 expression, we investigated whether specific transcription factor (TFs) induces its transcription. Using the SCENIC pipeline [53], we identified several TFs that exhibited differential expression patterns between Spp1^+^Fn1^+^ macrophages and Spp1^+^Fn1^-^ macrophages (Fig. 9A). Based on the calculated Regulon Specificity Score (RSS), Mafb, Irf8, Bhlhe41, Bmyc, and Atf3 were significantly up regulated in the Spp1^+^Fn1^+^ macrophage population (Fig. 9B). According to the inferred TFs, reduction analysis further demonstrated the distinct distribution in Spp1^+^Fn1^+^ macrophages (Fig. 9C). Consistent with the transcriptomic data, MAFB protein expression was markedly elevated in FACS-sorted Spp1^+^Fn1^+^ macrophages at 3 dpi (Fig. 9D–E). However, despite the up-regulation of Bmyc and Atf3 at the RNA level, their protein expression levels did not show statistically significant differences between the two macrophage populations (Fig. 9E). Among the 5 TFs, MAFB was more likely to bind to the Cxcl4 promoter region (+1046 bp to +1057 bp) (Fig. 9F) using the JASPAR CORE database [54]. We next asked if MAFB binds to Cxcl4 promoter to mediate its transcription. ChIP-PCR assay revealed that MAFB could directly bind to the promoter region of Cxcl4 (Fig. 9G-H). Furthermore, Mafb knockdown significantly reduced Cxcl4 mRNA expression and CXCL4 protein levels in Spp1^+^Fn1^+^ macrophages (Fig. 9I-K and Fig. S6). Collectively, these data indicate that MAFB may modulate Cxcl4 transcription in Spp1^+^Fn1^+^ macrophages.Fig. 9CXCL4 expression in Spp1^+^Fn1^+^macrophages is regulated by the transcriptional factor MAFB. A Heatmap showing the differential expression of TFs between Spp1^+^Fn1^+^ macrophages and Spp1^+^Fn1^-^ macrophages isolated from spinal cord tissue at 3 dpi. B Top activities of TFs in Spp1^+^Fn1^+^ macrophages. RSS indicates the Regulon Specificity Score. C tSNE plot displaying the top 5 TFs in Spp1^+^Fn1^+^ macrophages. D Western blot analysis of the top 5 TFs in different groups. E Quantification of western blot data in panel D, n = 3. F Predicted binding sites of MAFB in the Cxcl4 promoter region. G-H ChIP-PCR assay confirmed that MAFB could directly bind to the promoter region of Cxcl4. I RT-qPCR analysis of Cxcl4 mRNA expression in pericyte with or without Mafb knockdown. J Western Blot analysis of CXCL4 protein expression in pericyte with or without Mafb knockdown. K Quantification of western blot data in panel J, n = 3. All data are presented as the means ± SD. ∗∗p < 0.01, ∗∗∗p < 0.001 compared between groups by unpaired Student's t test in panel E, H, and by one-way ANOVA followed Tukey's multiple comparisons tests in panel I, K. ns, not significant.Fig. 9
Discussion
4
Scar formation is critical for sealing damaged tissue and restoring integrity following central nervous system (CNS) injury [55,56]. However, it also poses a significant barrier to axonal regeneration and thus hinders functional recovery. Historically, astrocytes have been considered the primary contributors to glial scar formation, with reactive astrocytes long believed to inhibit axonal regeneration [57,58]. While astrocytic scars undoubtedly create physical and molecular barriers, emerging studies challenge their exclusive inhibitory role and suggest that astrocyte scar formation aids rather than prevents CNS axon regeneration [55,59,60]. Given this controversy, the role of fibrotic scars in impeding axonal regeneration has gained increasing attention. Despite this, the precise mechanisms underlying fibrotic scar formation after SCI remain inadequately characterized. Recent studies have shown that pericytes, which are highly plastic and multipotent cells [61], play crucial roles not only in vascular regulation but also in fibrosis. Thus, elucidating the origin and pathways involved in pericyte-driven fibrosis could offer new therapeutic avenues to improve outcomes following SCI.
Pericytes, residing in the neurovascular unit (NVU), are characterized by markers such as PDGFRβ, NG2, and CD146. However, the lack of exclusive markers for pericytes complicates precise identification. Recent lineage-tracing studies reveal that PDGFRβ, widely used to label pericytes, is also expressed in perivascular fibroblasts and vSMC [62] leading to potential overestimation of pericyte contributions to fibrosis. For instance, Pdgfrβ^+^Col1a1^+^ cells post-SCI may encompass both pericyte-derived myofibroblasts and activated fibroblasts [42,62], whose distinct roles require multi-marker dissection. Similarly, while Glast promoter-driven models implicate specific pericyte subsets in fibrosis [4,42,63], the broad expression of Glast in astrocytes, Bergmann glia, and neural stem cells [[64], [65], [66]] suggests that single-marker strategies may not fully exclude contamination from non-pericyte populations [45,66].
To address potential ambiguity in pericyte identification, we employed a multi-marker scoring approach incorporating AUCell, UCell, singscore, AddModuleScore, and Scoring to systematically evaluate pericyte, fibroblast, and vSMC signatures at the single-cell level. This strategy enabled robust discrimination of pericyte populations at single-cell resolution, even in the context of overlapping marker expression. All Pdgfrb^+^ clusters exhibited stronger pericyte signature activity relative to fibroblast and vSMC signatures, supporting their classification as pericyte-lineage cells. Notably, cluster 3 showed slightly reduced pericyte activity compared to other clusters, aligning with its phenotypic features of PMT. These findings reinforce the reliability of our cell identity annotations within the limits of transcriptomic resolution and highlight the utility of integrative scoring approaches in improving marker-based classification. Nevertheless, definitive lineage attribution will ultimately require genetic fate mapping and spatial transcriptomic validation. Future studies should also incorporate time-resolved sampling (e.g., 1, 3, and 7 dpi) to dynamically track pericyte subset transitions and clarify their causal roles in fibrosis.
Myofibroblasts are characterized by high expression of α-SMA (encoded by the Acta2 gene) and the production of ECM proteins, hallmark features of tissue fibrosis. Although resident fibroblasts were initially considered the primary source, it is now evident that myofibroblasts can originate from various precursors, including endothelial cells and pericytes [14,32,67]. Our in vivo data align with these findings, indicating a substantial presence of myofibroblasts expressing pericyte marker Pdgfrβ at the core of the spinal cord injury. Specifically, we identify a distinct subset of Pdgfrβ^+^Acta2^+^ pericytes undergoing PMT. This transition is pivotal for ECM deposition and fibrotic tissue development. These findings delineate a clear cellular lineage involved in PMT and underscore the significance of pericytes as primary contributors to the fibrotic landscape in SCI. Importantly, myofibroblasts play a critical role in scar tissue formation by producing ECM components, which contribute to barriers that hinder axonal regeneration. Our findings specifically highlight the significant contribution of PMT to the fibrotic process, contrasting with studies that have primarily considered resident fibroblasts as the main source. This enhanced understanding of pericyte contributions offers novel insights into the cellular dynamics of fibrosis and proposes new therapeutic targets for modulating PMT to improve functional recovery after SCI.
CXCL4 has been implicated in a range of fibrotic diseases due to its potent ability to stimulate fibroblast proliferation and extracellular matrix (ECM) production [32,68,69]. However, its role in SCI and subsequent fibrosis has remained unexplored. Our research unveils the novel involvement of CXCL4 in driving fibrosis post-SCI. Notably, elevated levels of CXCL4 were observed in the injured spinal cord tissue, correlating with increased fibrotic activity. In this study, we sought to elucidate whether CXCL4 functions as a pivotal mediator of the PMT and subsequent spinal fibrosis. Our findings demonstrate that exogenous CXCL4 robustly induces the profibrotic activation of pericytes in vitro. Moreover, inhibition of CXCL4 markedly mitigates spinal fibrosis in murine models, underscoring the therapeutic potential of targeting CXCL4 with monoclonal antibodies.
The advent of scRNA-seq has uncovered the remarkable heterogeneity of immune cells involved in fibrosis post-SCI, identifying macrophages as crucial players in fibrosis progression. Although many studies have focused on macrophage subsets involved in inflammatory response, the specific macrophage populations driving fibrotic scar formation have not been fully characterized. By diverging from the conventional M1 and M2 paradigm [70,71], we identified Spp1^+^Fn1^+^ macrophages as key profibrotic cells following SCI. This finding aligns with emerging studies identifying Spp1^+^ macrophages in fibrotic processes across various organs, including the lungs, kidneys, and heart [32,72,73]. While previous studies highlighted Spp1^+^ macrophages, we found that the Spp1^+^Fn1^+^ macrophage subset demonstrates a significantly higher ECM regulatory score, indicating a more pronounced role in fibrosis.
Strikingly, our analysis revealed that Spp1^+^Fn1^+^ macrophages exhibited significantly elevated expression of CXCL4, which is regulated by MAFB. Notably, CXCL4 expression was exclusive to this subset, reinforcing its unique and dominant role in driving fibrotic processes through enhanced interactions with pericytes. While additional transcription factors such as Bmyc and Atf3 were also enriched at the transcriptomic level, their protein expression did not show significant differences between macrophage subsets, indicating a potential divergence between RNA and protein readouts. Taken together, these findings suggest that Spp1^+^Fn1^+^ macrophages are the dominant drivers of fibrosis, challenging the notion that all Spp1^+^ macrophages equally contribute to this process.
Moreover, the mechanistic insights provided by the study elucidate the signaling pathways involved in this process. The activation of the CXCR3 receptor by CXCL4 triggers the PI3K/Akt pathway, which is known to regulate cellular processes such as proliferation and differentiation. Previous studies have reported that excessive deposition of extracellular matrix components such as fibronectin and laminin contributes to the formation of fibrotic scars, which serve as physical and biochemical barriers impeding axonal extension [63,[74], [75], [76]]. In line with this, our findings suggest that CXCL4-driven PMT enhances ECM accumulation and fibrotic scarring, thereby creating an inhibitory environment for axon regeneration. Conversely, blocking CXCL4 not only attenuates fibrosis but also renders the lesion site more permissive for axonal regrowth, which may explain the improved motor recovery observed in our model. The findings underscore the importance of understanding the molecular mechanisms underlying PMT, as they may offer novel therapeutic targets for enhancing recovery after SCI (Fig. 10).Fig. 10. The schematic illustrates the proposed mechanism through which CXCL4 mediates fibrosis following SCI. Following SCI, Spp1^+^Fn1^+^ macrophages at the injury site secrete CXCL4, promoting the transition of Pdgfrβ^+^Acta2^+^ pericytes into myofibroblasts. This cascade results in excessive extracellular matrix deposition and the formation of fibrotic scar tissue, ultimately impeding axonal regeneration.Fig. 10
This study has several limitations. Firstly, it focuses on the local immune microenvironment following SCI, without investigating the plasma levels of CXCL4. Future research should include Macrophage-specific Cxcl4^KO^ mice and Pericyte-specific Cxcr3^KO^ mice to further elucidate the mechanisms by which CXCL4 promotes fibrosis after SCI. Furthermore, the downstream molecular targets of this pathway require further investigation to fully understand how CXCL4 drives PMT and fibrosis, particularly to address concerns about the potential indirect effects of anti-CXCL4 antibodies on other cell types, and whether CXCL4 specifically improves neural function by acting on pericytes. In addition, existing research suggests that pericytes generally exert beneficial effects following CNS injury, although pericyte subtypes have not been well-characterized. For instance, pericytes derived from induced pluripotent stem cells were recently transplanted into a mouse stroke model, resulting in significant functional recovery. Proangiogenic treatments post-stroke, correlated with high pericyte numbers, demonstrated reduced blood–brain barrier leakage. Pericytes have also been implicated in neurodegenerative diseases. Future studies should aim to further characterize pericyte subtypes in these contexts to better delineate their dual roles in neurodegeneration and CNS injury recovery.
Together, this study establishes the critical role of CXCL4 derived from Spp1^+^Fn1^+^ macrophages in promoting the transition of pericytes to myofibroblasts via the CXCR3/PI3K/Akt signaling axis. Blocking CXCL4 using neutralizing antibody inhibits this transition and reduces the formation of dense scar tissue post-SCI. Additionally, targeting CXCL4 improves neuronal survival and promotes motor function recovery following acute SCI, highlighting its potential as a promising therapeutic target.
Authors’ contributions
YW and XL designed the study. GL, XZ and XL conducted the study. DZ, WZ and XC collected the data. GL, LW, JC, YW and XL analyzed the data. GL, LW, DZ, XC and LL performed data interpretation. GL and XL drafted the manuscript. YW and XL revised the manuscript content. All authors take responsibility for the integrity of data analysis. All authors approved the final version of the manuscript.
Ethical approval
All surgical interventions, treatments, and postoperative animal care procedures were approved by the Animal Care and Use Committee of Sun Yat-sen University and strictly followed the Guide for the Care and Use of Experimental Animals provided by the National Research Council (1996, USA).
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declaration of generative AI in scientific writing
No generative artificial intelligence (AI) or AI-assisted technologies were used in the preparation of this manuscript.
Funding
This work was supported by grants from the 10.13039/501100001809National Natural Science Foundation of China (Grant no: 81971151, 82102528, 82102583, and 82203677), the 10.13039/501100003453Natural Science Foundation of Guangdong, China (Grant no: 2024A1515012772, 2025A1515010512), Kelin New Star project of the First Affiliated Hospital of 10.13039/501100002402Sun Yat-sen University (Grant no: R08041), 10.13039/501100010256Guangzhou Science and Technology Plan Project (Grant no: 2024A04J9907, 2025A04J4016), Young Science and Technology Talent Support Program of Guangdong Precision Medicine Application Association (Grant no: YSTTGDPMAA202502), Youth S&T Talent Support Programme of GDSTA (Grant no: SKXRC2025179).
Declaration of competing interest
None of the authors has any potential conflict of interest.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Tanaka E.M.Ferretti P.Considering the evolution of regeneration in the central nervous system Nat Rev Neurosci 1020097137231976310410.1038/nrn 2707 · doi ↗ · pubmed ↗
- 2Cregg J.M.De Paul M.A.Filous A.R.Lang B.T.Tran A.Silver J.Functional regeneration beyond the glial scar Exp Neurol 25320141972072442428010.1016/j.expneurol.2013.12.024PMC 3951813 · doi ↗ · pubmed ↗
- 3Silver J.Schwab M.E.Popovich P.G.Central nervous system regenerative failure: role of oligodendrocytes, astrocytes, and microglia Cold Spring Harbor Perspect Biol 72014 a 02060210.1101/cshperspect.a 020602 PMC 435526725475091 · doi ↗ · pubmed ↗
- 4Dias D.O.Kim H.Holl D.Werne Solnestam B.Lundeberg J.Carlén M.Reducing Pericyte-Derived scarring promotes recovery after spinal cord injury Cell 173201810.1016/j.cell.2018.02.004PMC 587171929502968 · doi ↗ · pubmed ↗
- 5Bradbury E.J.Burnside E.R.Moving beyond the glial scar for spinal cord repair Nat Commun 10201938793146264010.1038/s 41467-019-11707-7PMC 6713740 · doi ↗ · pubmed ↗
- 6Zhao C.Rao J.S.Duan H.Hao P.Shang J.Fan Y.Chronic spinal cord injury repair by NT 3-chitosan only occurs after clearance of the lesion scar Signal Transduct Targeted Ther 7202218410.1038/s 41392-022-01010-1PMC 920379335710784 · doi ↗ · pubmed ↗
- 7Liddelow S.A.Barres B.A.Reactive astrocytes: production, function, and therapeutic potential Immunity 4620179579672863696210.1016/j.immuni.2017.06.006 · doi ↗ · pubmed ↗
- 8Camand E.Morel M.P.Faissner A.Sotelo C.Dusart I.Long-term changes in the molecular composition of the glial scar and progressive increase of serotoninergic fibre sprouting after hemisection of the mouse spinal cord Eur J Neurosci 202004116111761534158810.1111/j.1460-9568.2004.03558.x · doi ↗ · pubmed ↗