Abstract Many neurological diseases remain difficult to treat, necessitating further elucidation of their pathogenesis. Conditional inactivation of Pdgfra in Nestin-expressing cells leads to the depletion of platelet-derived growth factor receptor-alpha^+ (PDGFRα^+) oligodendroglial lineage cells responsible for myelination, resulting in forebrain hypomyelination and severe, progressive neurological deficits in neonatal mice. The present study examined the cerebral cortex of these mice to better understand the mechanisms underlying such progressive neurological deficits, that are often observed in refractory neurological diseases. Histological and single-cell RNA sequencing analyses showed that, following activation of meningeal border-associated macrophages (BAMs), PDGFRα^+ fibroblasts that escaped gene inactivation were extensively recruited from the meninges into the hypomyelinated subpial cerebral cortex. Transcriptional reprogramming suggested that these fibroblasts originated from the pial fibroblast lineage and adopted a myofibroblast-like transcriptional phenotype. The recruited fibroblasts established stable cell–cell interactions with activated brain macrophages, including BAMs and microglia, accompanied by signaling pathways associated with chronic, tissue-damaging inflammation. Subsequently, inflammatory cortical lesions emerged, characterized by glial activation, angiogenesis, and neuronal oxidative stress. Treatment with a PDGFRα-neutralizing antibody significantly reduced fibroblast recruitment and mitigated glial activation and angiogenesis. These findings suggest that meningeal BAMs and pial fibroblasts are key contributors to the formation of tissue-damaging subpial cortical lesions. The interactions between brain macrophages and pial fibroblasts may contribute to the mechanisms underlying chronic and progressive neurological deficits and represent potential therapeutic targets for refractory neurological diseases. Supplementary Information The online version contains supplementary material available at 10.1186/s40478-025-02063-3. Keywords: Angiogenesis, Border-associated macrophage, Fibrosis, Myelination failure, PDGFRα, Perivascular fibroblast Introduction The diverse resident cells within the central nervous system (CNS) coordinate extensive tissue remodeling during normal development and under various pathological conditions. For example, regulating oligodendrocyte progenitor cell (OPC)-derived hypoxia-inducible factor (HIF) expression controls both the induction and suppression of robust angiogenesis, as well as subsequent myelination in the neonatal mouse brain [[48]22, [49]45, [50]125]. A diverse range of CNS resident cells, including border-associated macrophages (BAMs), microglia, oligodendroglial lineage cells, blood vessel-associated cells, and CNS fibroblasts, purportedly contribute to the pathogenic mechanisms of multiple sclerosis (MS) [[51]53, [52]59, [53]76, [54]93, [55]131]. MS is the most common chronic neurodegenerative and neuroinflammatory disease of the CNS, characterized by the loss of myelin, oligodendrocytes, and axons [[56]12, [57]32]. Most patients exhibit the relapsing-remitting form of MS (RRMS), which ultimately leads to the progressive form of MS (PMS), a stage predominantly refractory to disease-modifying drugs, including those approved for RRMS [[58]14, [59]30, [60]41, [61]90]. Similar to MS, myelin damage is involved in the development and progression of Alzheimer`s disease, although the specific mechanisms remain poorly understood [[62]5, [63]17, [64]87]. Moreover, persistent and chronic neuroinflammation involving diverse cell types may contribute to the initiation and progression of neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease, and amyotrophic lateral sclerosis [[65]6, [66]60, [67]81]. Hence, a deeper understanding of the intricate cellular interactions within the CNS is imperative for developing effective therapeutic approaches for neurological disorders. Platelet-derived growth factors (PDGFs) and their receptors (PDGFRs) are the primary growth factors for mesenchymal cells, with neurotrophic properties and high expression in the CNS [[68]48, [69]98, [70]121]. Accordingly, their homeostatic and pathological roles in nervous tissues have been extensively studied [[71]36, [72]54, [73]126]. PDGFRα is highly expressed in OPCs and CNS fibroblasts, and the PDGF–A/PDGFRα signaling axis contributes to myelination by regulating OPCs in both the developing and adult brain [[74]16, [75]35, [76]61, [77]84, [78]110, [79]128]. Transient global inactivation of Pdgfra by tamoxifen in CAGG-CreER; Pdgfra^flox/flox mice suppresses the recruitment of PDGFRα^+/COL1^+ perivascular fibroblasts (pvFs) in the murine stroke brain, suggesting the involvement of PDGFRα in hemorrhagic infarction [[80]82]. In these mice, PDGFRα^+ OPCs recover after transient depletion due to their potent plasticity; neither progressive neurological symptoms nor neuroinflammatory lesions are observed [[81]16, [82]82]. In contrast, conditional inactivation of Pdgfra in mice harboring a gene encoding a persistently active form of Cre recombinase targeting Nestin-expressing (Nestin^+) cells (Nestin-nlsCre) (N-PRα-KO mice) suppresses the maturation of Olig2^+ cells into myelinating Sox10^+ cells during embryogenesis, leading to Olig 2^+ cell depletion and diffuse myelination failure in the neonatal forebrain [[83]44]. N-PRα-KO mice also exhibit axonal damage and aberrant angiogenesis, accompanied by severe progressive neurological deficits, including gait disturbance and death by approximately postnatal day 17 (P17). However, the underlying mechanisms mediating these neurological deficits remain unexplored, as there are, to our knowledge, no reports of such severe neurological disabilities resulting from experimentally induced myelination failure. The role of pvFs in normal brain development and various neurological diseases has become a research focus. pvFs secrete the extracellular matrix (ECM) components and numerous cytokines and growth factors with beneficial and detrimental roles in pathological processes, including stroke, neurodegenerative diseases, and neuroinflammatory diseases [[84]24, [85]61, [86]72, [87]82, [88]102, [89]115]. Although multiple origins of fibrosis-inducing cells have been proposed, accumulating evidence suggests that the primary source is pvFs derived from leptomeningeal fibroblasts (menFs) [[90]2, [91]8, [92]24, [93]38, [94]56, [95]61, [96]120, [97]123]. A recent study reported six distinct transcriptional identities of brain and leptomeningeal fibroblasts in normal mouse brains, including parenchymal pvFs (BFB1a) and pia and neighboring perivascular fibroblasts (BFB1b) [[98]91]. CNS fibroblasts and BAMs are closely localized and develop concurrently from meninges into parenchymal perivascular regions in the postnatal mouse brain, where they exhibit intimate cellular interactions [[99]3, [100]37, [101]61, [102]75, [103]95, [104]110, [105]128]. Similarly, fibroblasts and macrophages are often distributed in close proximity and establish a close interdependent relationship that mediates homeostatic and pathological tissue remodeling in peripheral organs [[106]1, [107]9, [108]57, [109]77, [110]108, [111]127]. However, such cellular interactions have not yet been characterized in nervous tissue. Therefore, further investigation is required to elucidate the behavior and functional roles of each PDGFRα^+ CNS fibroblast subtype and their interaction with other CNS resident cells, especially brain macrophages (BAMs and microglia), in the context of brain development and neurological disease. Although the involvement of various CNS resident cells in inflammation and myelin damage have been proposed in the pathogenesis of neurological diseases, the precise mechanisms remain incompletely understood. In this study, we aimed to elucidate the mechanisms underlying progressive neurological deficits in neurological diseases. Specifically, we evaluated the mechanisms by which developmental myelination failure leads to diffuse cortical lesions occur in the brains of N-PRα-KO mice. The study findings revealed chronic and persistent inflammatory mechanisms in the subpial cerebral cortex of N-PRα-KO mice, where numerous PDGFRα^+ pial fibroblasts, sharing transcriptional features with the BFB1b subtype and having escaped gene inactivation, were recruited from the meninges and transitioned into parenchymal pvFs resembling the BFB1a subtype [[112]91], playing a key role in cortical lesion formation. These fibroblasts were at least partly recruited in response to prior activation of meningeal BAMs. CNS fibroblasts and brain macrophages, including BAMs and microglia, established stable cell–cell interactions that mediated tissue-damaging, chronic, and persistent inflammatory lesions. These cortical lesions were characterized by neuronal damage, glial activation, angiogenesis, and fibrosis; hallmarks of CNS lesions in chronic neurological diseases. Thus, this study highlights a mechanism by which activated BAMs and pial fibroblasts are recruited and serve as a potent contributors to tissue-damaging inflammatory lesions in the subpial cortex. The findings may inform therapeutic strategies for chronic and progressive neurological diseases, through a better understanding of cellular interactions involved in their pathogenesis. Methods Animal experiments The Institutional Animal Care and Use Committee at the University of Toyama (University of Toyama, Toyama, Japan) approved all animal procedures. All mice were housed at 25 °C, under a 12/12-h light/dark cycle, with unrestricted access to pellet chow and water. Pdgfra conditional inactivation mice (N-PRα-KO mice) Descriptive analyses were performed on N-PRα-KO mice, as previously described [[113]44, [114]107]. The genotype of these transgenic mice was confirmed via polymerase chain reaction (PCR) using genomic DNA extracted from tail biopsies, as previously described [[115]16, [116]62, [117]122]. Briefly, the mouse tail was lysed using SNET buffer (20 mM Tris, 5 mM EDTA, 400 mM NaCl, 0.3% SDS; Nacalai Tesque, Kyoto, Japan) supplemented with 10 mg/mL proteinase K (MilliporeSigma, Burlington, MA, USA). PCR was performed using a C1000 Thermal Cycler (Bio-Rad, Hercules, CA, USA) and Ampdirect Plus reagent (Nacalai Tesque). The primers used were as follows: Pdgfra^flox forward: 5ʹ-ATGCCAAACTCTGCCTGATTGA-3ʹ and reverse: 5ʹ-CTCACGGAACCCCCACAAC-3ʹ; Nestin-nlsCre forward: 5ʹ-GTACTTTCTGTGACTGTCAGCTATCGCTTTGTAAAAC-3ʹ and reverse: 5ʹ-CAGCACCAGTGTGGAGCTGCACAC-3ʹ. WT mice were used as controls. RNA isolation and quantitative real-time PCR Mice were deeply anesthetized via intraperitoneal medetomidine (0.75 mg/kg body weight, NIPPON ZENYAKU KOGYO, Fukushima, Japan), midazolam (4 mg/kg body weight, SANDOZ, Holzkirchen, Germany), and butorphanol (5 mg/kg body weight, Meiji Seika Pharma, Tokyo, Japan) injection. Mice were then transcardially perfused with ice-cold 0.01 M phosphate-buffered saline (PBS, pH 7.4, Nacalai Tesque). Excised tissue samples were frozen at -80 °C and thawed in QIAzol Lysis Reagent (Qiagen, Hilden, Germany) after crushing. Cell samples were immediately lysed using QIAzol Lysis Reagent. Total RNA was isolated using the miRNeasy Mini Kit (Qiagen), according to the manufacturer’s instructions. Purified RNA was reverse transcribed into first-strand cDNA using the PrimeScript™ II 1st-strand cDNA Synthesis Kit (Takara Bio, Shiga, Japan). The expression level of each gene was determined using the C1000™ Thermal Cycler (Bio-Rad) and the Thermal Cycler Dice^® Real-Time System III (Takara Bio). β-actin (Actb) was used as the reference gene. Primer sequences are available from the Takara Bio Inc. website ([118]http://www.takara-bio.co.jp). Measurement of PDGF ligand concentration in the whole brain hemisphere PDGF ligands were measured, as previously described [[119]16, [120]62, [121]122]. Brain hemispheres excised from deeply anesthetized mice were frozen at -80 °C and lysed in T-PER™ Tissue Protein Extraction Reagent (Thermo Fisher Scientific, Waltham, MA, USA) after crushing. To detect PDGF ligands, the appropriate ELISA kits were used (PDGF-AA, Mouse PDGF-AA ELISA Kit for Serum, Plasma, and Cell Culture Supernatants, RayBiotech, Norcross, GA, USA; PDGF-BB, Mouse PDGF BB ELISA Kit, Abcam, Cambridge, United Kingdom; PDGF-AB, Mouse/Rat PDGF-AB Quantikine ELISA Kit, R&D Systems, Minneapolis, MN, USA; PDGF-CC, Mouse platelet-derived growth factor C ELISA Kit, MyBioSource, San Diego, CA, USA), according to the manufacturers’ instructions. Immunostaining The mice were deeply anesthetized and perfused with PBS, followed by transcardial perfusion with 1% paraformaldehyde (PFA; Nacalai Tesque) for tissue fixation. The excised brains were immediately submerged in 1% PFA at 4 °C for frozen section preparation. The brains were then cryoprotected in 30% sucrose (FUJIFILM Wako Chemicals, Osaka, Japan) at 4 °C for over 36 h, followed by embedding in Tissue-Tek^® O.C.T.™ compound (Sakura Finetek, Tokyo, Japan) and freezing on dry ice. Coronal brain sections, 20-µm thick and corresponding to 0.5–1.1 mm relative to the bregma, were prepared using a cryostat and mounted on glass slides (Frontier FRC-05, Matsunami Glass, Osaka, Japan), as previously described [[122]82]. For paraffin section preparation, the excised brains were immediately fixed with 4% PFA in PBS, dehydrated through a graded ethanol series, and embedded in paraffin. Coronal brain sections of 5-µm thickness were prepared for immunostaining. Immunofluorescence was performed, as previously described [[123]16, [124]117, [125]118]. Antigen retrieval was performed using target retrieval solution pH 9 (Nichirei, Tokyo, Japan) at 98 °C for 25 min, following the manufacturer’s protocol, before immunostaining. Non-specific staining was prevented with Blocking One Histo (Nacalai Tesque) for 30 min at room temperature. Additionally, tissue sections were treated with the Histofine Mouse Stain Kit (Nichirei) for 1 h before incubation with mouse-raised primary antibodies. Sections were incubated overnight at 4 °C with the following primary antibodies, diluted in 0.03% Triton X-100 (Nacalai Tesque)/PBS containing 10% normal goat serum (Vector Laboratories, Burlingame, CA, USA) or 10% normal donkey serum (ImmunoBioScience, Mukilteo, WA, USA): goat polyclonal anti-PDGFRα (1:100, R&D Systems), goat polyclonal anti-PDGFRβ (1:100, R&D Systems), rabbit monoclonal anti-PDGFRβ (1:100; Abcam), rat monoclonal anti-CD31 (1:100; Dianova, Geneva, Switzerland), hamster monoclonal anti-CD31 (1:100; MilliporeSigma), rabbit polyclonal anti-NG2 (1:250; MilliporeSigma), rabbit polyclonal anti-Collagen type I (1:250; Abcam), rabbit monoclonal anti-ALDH1A1 (1:50; Abcam), rabbit-polyclonal anti-ALDH1A2 (1:100; MilliporeSigma), rabbit monoclonal anti-S100 alpha6/PRA antibody (1:100; Abcam), rabbit polyclonal anti-phospho-SMAD2/SMAD3 (Thr8) (1:100; Invitrogen; Thermo Fisher Scientific), mouse monoclonal anti-CXCL12/SDF-1 (1:500; R&D Systems), mouse monoclonal anti-Angiopoietin-2 (1:100; R&D Systems), rabbit polyclonal anti-MBP (1:200, Abcam), mouse monoclonal anti-NeuN (1:100; MilliporeSigma), rabbit polyclonal anti-4HNE (1:300, Abcam), rabbit polyclonal anti-GFAP (1:500; Dako, Agilent, Santa Clara, CA, USA), mouse monoclonal anti-GFAP (1:500; MilliporeSigma), goat polyclonal anti-Iba1 (1:300; Abcam), Rabbit polyclonal anti-ZO-1 (1:500; Invitrogen, Thermo Fisher Scientific), rabbit polyclonal anti-Claudin-5 (1:100; Invitrogen, Thermo Fisher Scientific), rat monoclonal anti-PLVAP (1:200; Abcam), rat monoclonal anti-CD206 (1:100; Bio-Rad), Rabbit polyclonal anti-LYVE-1 (1:100; Abcam), rat monoclonal anti-LYVE-1(1:100; eBioscience, San Diego, CA, USA), rabbit monoclonal anti-CCL5 (1:50; Invitrogen; Thermo Fisher Scientific), rabbit monoclonal anti-M-CSF (1:100; Abcam), and rabbit monoclonal anti-YAP(D8H1X)XP (1:100; Cell Signaling Technology, Danvers, MA, USA). Additionally, Alexa-Fluor488-, Alexa-Fluor568-, or Alexa-Fluor633-conjugated secondary antibodies (Invitrogen, Thermo Fisher Scientific) were used at 1:500 dilutions. Nuclei were counterstained with Hoechst 33,258 (Nacalai Tesque). Immunofluorescence images were acquired using the BIOREVO BZ-9000 microscope (Keyence, Osaka, Japan), TCS SP5 confocal system (Leica Microsystems, Wetzler, Germany), and the LSM780 confocal system (Carl Zeiss, Oberkochen, Germany). Image processing was performed using Photoshop software version 22 (Adobe, San Jose, CA, USA). Morphometrical analyses pvFs were double immunolabeled for PDGFRα and COL1. Images of the coronally-cut parietal cerebral cortex were captured using a BIOREVO BZ-9000 microscope. Measurement of the pvF area was conducted as described (Fig. [126]S1h). The pvF area was quantified using BZ-II Analyzer software (Keyence), as previously described [[127]50, [128]122], with some modifications. Single PDGFRα^+ cells located beneath the meninges were counted using images captured at high magnification (63× objective lens TCS SP5 confocal system). The number of penetrating blood vessels in the parietal region of coronal brain sections was quantified per unit length of the brain surface, as previously described [[129]44]. Brain vasculature was visualized using CD31— a common marker for endothelial cells. Neuronal oxidative stress was quantified by measuring 4-HNE immunopositivity in NeuN-positive neurons. Mean values of 4-HNE intensity was calculated from ten randomly selected NeuN-positive cells within the same cortex area (175 × 175 μm) were quantified using the ZEISS Efficient Navigation system (Carl Zeiss AG). To assess astroglial activation, the GFAP^+ area was measured in a defined 300 × 300 μm region of the middle cortex using BZ-II Analyzer software. Iba1^+ cells in the cortex were counted from a high-power field images acquired with a 40× objective lens (Keyence). For quantifying mean IgG intensity, a 100 × 100 μm area of the middle cortex was analyzed using the ZEISS Efficient Navigation system. The number of CD206^+ cells localized just below the meninges of the parietal area was counted using images from a high-power field (40× objective, TCS SP5). PDGFRα^+ cell sorting and primary culture Mice were deeply anesthetized and perfused with PBS. The cortex was excised from the brain and dissociated using a Neural Tissue Dissociation Kit-Postnatal Neurons (Miltenyi Biotech, Bergisch Gladbach, Germany). PDGFRα^+ cells were collected via magnetic cell sorting (MACS) using CD140a (PDGFRα) MicroBeads (Miltenyi Biotech), according to the manufacturer’s protocol. PDGFRα^+ cells were centrifuged and the cell pellet was resuspended in CS-C complete medium Kit R (KAC, Kyoto, Japan) containing 20% fetal bovine serum (FBS, MilliporeSigma) and 1% antibiotic–antimycotic mixed stock solution (100×; Nacalai Tesque). The cells were seeded in a 12-well plate (Corning, Corning, NY, USA). The complete culture medium was changed every 3–4 d. Preparation of skin fibroblasts and primary culture Skin fibroblasts and cell cultures were prepared, as previously described [[130]50, [131]116, [132]122]. Briefly, the skin was harvested from P1 to P3 WT pups deeply anesthetized and perfused with PBS. Following treatment with 0.05% collagenase (Worthington Biochemical, Lakewood, NJ, USA), dissociated skFs were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Nacalai Tesque), containing 10% FBS. PDGF ligand stimulation of cultured cells Stimulation of PDGF ligands was conducted as previously described [[133]116, [134]122]. Briefly, 1 × 10^4 PDGFRα^+ cells were seeded into a 24-well plate (Corning). The cells were subjected to starvation with DMEM containing 0.5% FBS for 24 h. For stimulation, the starvation medium was replaced with a medium containing 10 ng/mL PDGF-BB (PeproTech, Rocky Hill, NJ, USA) or 10 ng/mL PDGF-AB (PeproTech) and cultured for 24 h. As a control, cells were continuously cultured for 24 h with a starvation medium only. The cells were lysed in QIAzol Lysis Reagent (QIAGEN) to prepare samples for real-time PCR. Identically treated skin fibroblasts were used as a control. PDGFRα-neutralizing antibody administration Antibody-based neutralization of PDGFRα signaling in the brain was achieved by administering N-PRα-KO mice with anti-PDGFRα mAb (APA5) diluted in PBS (Nacalai Tesque) at 3.3 µg/µL; 3 µL/mouse was injected via the anterior fontanel into the subarachnoid cavity on P1 and P7 [[135]16]. For vehicle treatment, N-PRα-KO mice were identically treated with PBS or the InVivoMab rat IgG2a isotype control (BioCell, Irvine, CA, USA) at the same concentration. scRNA-seq The cerebral cortex with the meninges of P10–14 mice was dissociated, and MACS was performed using the CD140a antibody. The cells were collected in PBS containing 0.05% bovine serum albumin. Cell suspensions were prepared as a mixture of cells from mice of the same genotype. Following the manufacturer’s protocol, single-cell libraries were prepared using the BD Rhapsody Single-Cell Analysis System (BD Biosciences, San Diego, CA, USA) and the BD Rhapsody™ WTA Reagent Kit (BD Biosciences). The quantity and quality of the libraries were confirmed using the Qubit dsDNA High Sensitivity Assay Kit (Invitrogen, Thermo Fisher Scientific). The libraries were subsequently sequenced on an Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA), aiming for a sequencing depth of 20,000 reads per cell. Unsupervised clustering of cells and UMAP visualization Reads were obtained from 11,438 cells from three WT mice and 8,336 cells from three N-PRα-KO mice. Data from both genotypes were merged, and the combined data were analyzed on the SeqGeq bioinformatics platform (BD Biosciences). The applied techniques adhered to the recommended bioinformatics protocols developed by the scientific community and refined by experts in single-cell data exploration at BD Biosciences. Data files were concatenated and normalized to counts per 10,000 reads, ensuring consistent analysis across all datasets. PhenoGraph was suitable for accurately extracting fibroblast populations due to its high-resolution cluster classification. Therefore, clustering of total WT and N-PRα-KO cells was first performed using the PhenoGraph package (version 3.0) implemented in R. To correct batch effects, the mutual nearest neighbors (MNN) method was applied, as implemented in the R package BatchLR (version 1.0.9), utilizing the fast MNN function and default parameters [[136]42]. K-means filtering (k = 30) was conducted, and the 32 clusters identified by PhenoGraph were overlaid onto the UMAP using the R package UMAP (version 3.1). Each cluster, identified by PhenoGraph, was manually annotated with known brain cell-type markers (Note S1, Fig. [137]S4). Subsequently, identified fibroblast clusters were subdivided using the Seurat package (version 3.8.1) with algorithms from PhenoGraph. The clusters were overlaid on the UMAP. To identify the cells derived from each genotype, namely WT and N-PRα-KO, AutoGateCategorial (version 2.6) was used. Pseudotime analysis Monocle (version 4.2) was used to conduct single-cell trajectory analysis, focusing on pial fibroblasts to define transitional states during differentiation. Dot and violin plots were generated using the R packages Seurat and ggplot2, respectively. Pathway enrichment analysis The WEB-based GEne SeT AnaLysis Toolkit 2019 (WebGestalt, [138]www.webgestalt.org) was used to conduct an over-representation analysis using Biological Process gene ontology (GO) [[139]67]. The displayed categories were derived from WebGestalt’s “Weighted set cover” option, which presents the GO categories that minimize redundancy. For Seurat-identified Clusters I, VI, and VII, all upregulated genes with q-values < 0.05 relative to all cells in the remaining clusters were selected. The Wilcoxon rank sum test was applied to each q-value, and genes with a false discovery rate (FDR) < 0.05 were selected. For states i−iii, and v identified by Monocle, upregulated genes with a fold-change > 1.5 and q-values < 0.05 relative to the state iv populations were selected. For state iv identified by Monocle, upregulated genes were determined and compared with all cells in other states. The Benjamini–Hochberg algorithm was applied to each q-value, and genes with an FDR < 0.05 were selected. Statistical analysis A two-way analysis of variance was conducted for imaging analysis, followed by the Newman–Keuls test for multiple comparisons. The differences between genotypes or various treatments at the same time point were evaluated using the Student’s t-test for real-time PCR, ELISA, and anti-neutralization administration data. Graphs were created with GraphPad Prism 9 (GraphPad Software, Inc., La Jolla, CA, USA), and quantified data were presented as mean ± standard error of the mean (SEM). For clusters and states obtained in scRNA-seq, the Kruskal–Wallis test was conducted, followed by the Wilcoxon test. Violin plots were generated using the R packages Seurat and ggplot2. Statistical significance was set at p < 0.05. Results Pdgfra inactivation in N-PRα-KO mice had potential compensatory effects on other PDGF family genes and proteins The degree of Pdgfra inactivation and potential compensatory changes in the expression of other PDGF family genes and proteins was evaluated in N-PRα-KO mice. Significantly reduced, albeit not entirely absent, Pdgfra expression was observed in N-PRα-KO mice compared to age-matched control wild-type (WT) mice at P5, P10, and P15 (Fig. [140]1a, Fig. [141]S1a). Meanwhile, Pdgfrb mRNA expression increased in the N-PRα-KO cerebrum (Fig. [142]S1b). ELISA results revealed decreased cerebral PDGF-AA concentrations at P5 and P10 in the N-PRα-KO mice compared with those in the controls. Conversely, PDGF-AB levels were elevated at P5, while PDGF-BB levels were elevated at P5 and P10. PDGF-CC expression was not affected at any age in either group (Fig. [143]S1c−f). Considering that PDGF-A and PDGF-C bind PDGFRα and PDGF-B binds PDGFRα and PDGFRβ, the PDGF ligand availability appeared to correspond to cognate receptor expression. Fig. 1. [144]Fig. 1 [145]Open in a new tab PDGFRα^+/NG2^−/COL1^+ perivascular fibroblasts increased in the cerebral cortex of N-PRα-KO mice with depleted OPCs. a Transgenic and mutated alleles in N-PRα-KO mice, harboring a Nestin promoter/enhancer-driven Cre recombinase, possessing a nuclear localization signal (nls) and Pdgfra flanked by loxP sequences. b Immunofluorescence for PDGFRα and CD31 in WT and N-PRα-KO cortex at P15. Arrowheads: perivascular PDGFRα^+ cells exclusively localized near the meninges in WT mice. c Immunofluorescence of PDGFRα and NG2 in N-PRα-KO cortex at P10. d Immunofluorescence of PDGFRα, NG2, and CD31 in N-PRα-KO cortex at P10. Cyan arrowheads: PDGFRα⁺ cells, green arrowheads: NG2⁺ pericytes. e Immunofluorescence of PDGFRα, COL1, and CD31 in N-PRα-KO cortex at P15. Nuclei are counterstained with Hoechst. Scale bars, 250 μm (b and c), 20 μm (insets of b), 5 μm (d), and 25 μm (e). See also Fig. [146]S1a−f The cerebral cortex of N-PRα-KO mice contained abundant PDGFRα^+ perivascular fibroblasts and depleted OPCs Endogenous brain Pdgfra expression is not limited to OPCs, which are targeted by Nestin-nlsCre in N-PRα-KO mice, but is also expressed in CNS fibroblasts [[147]61, [148]91], which are not targeted by Nestin-nlsCre. Therefore, the loss of PDGFRα in OPCs may influence other PDGFRα-positive cells. This could contribute to hypomyelinated cerebral cortex remodeling, including changes in penetrating blood vessels and neuronal cell damage [[149]44]. Indeed, changes in the brain bulk concentrations of PDGF ligands were observed in N-PRα-KO mice, further prompting this hypothesis (Fig. [150]S1c−f). In the coronally-cut parietal cerebral cortex, PDGFRα^+ OPCs with typical ramified morphologies were evenly distributed throughout the cerebral cortex of WT mice but were depleted in N-PRα-KO mice at P15 (Fig. [151]1b) [[152]44]. In contrast, PDGFRα + cells in the meninges were normally distributed along the parietal surface and within the longitudinal fissure of N-PRα-KO mice at P15 (Fig. [153]1b). PDGFRα^+ cells were also observed along blood vessels in WT mice and were present in large penetrating vessels near the meningeal surface, as previously reported [[154]61, [155]91, [156]110]. In contrast, N-PRα-KO mice exhibited perivascular PDGFRα^+ cells deep in the brain parenchyma, associated with large-caliber vessels and along capillaries. Overall, prominent hyperplasia of perivascular PDGFRα^+ cells was observed (Fig. [157]1b). Thus, the loss of PDGFRα^+ OPCs in N-PRα-KO mice was paralleled by an increase in perivascular PDGFRα^+ cells in capillaries that do not normally harbor perivascular PDGFRα^+ cells. This partially accounted for the reduced Pdgfra mRNA expression observed in the N-PRα-KO mouse brain (Fig. [158]S1a), and raised questions about the identity of the ectopic pericapillary PDGFRα^+ cells. Three possible origins and identities were proposed: OPCs (resistant to Nestin-nlsCre-mediated Pdgfra targeting), pericytes, and fibroblasts [[159]16, [160]82, [161]110]. NG2 is a marker common to pericytes and OPCs [[162]84, [163]86]; hence, the relationship between NG2^+ and PDGFRα^+ cells was evaluated in perivascular areas. At P10, PDGFRα^+ cells were distributed along penetrating blood vessels accompanied by NG2^+ mural cells in N-PRα-KO mice. However, PDGFRα^+ cells were less abundant in the deeper cortical layers compared to NG2^+ cells (Fig. [164]1c). Higher magnification imaging revealed that PDGFRα^+ cells were located peri-endothelially, distinct from NG2^+ pericytes in the cerebral cortex (Fig. [165]1d). The absence of NG2 expression suggests that the peri-capillary PDGFRα^+ cells in N-PRα-KO mice were neither pericytes nor OPCs. Instead, the PDGFRα^+ cells co-expressed the canonical fibroblast marker collagen type I (COL1; Fig. [166]1e) and were present in leptomeninges of both N-PRα-KO and WT mice (Fig. [167]2a). The perivascular abundance of PDGFRα^+/COL1^+ cells was markedly higher in N-PRα-KO than in WT mice at P10 (Fig. [168]2a, Fig. [169]S1g). Hence, the PDGFRα^+/NG2^−/COL1^+ cells were predicted to be pvFs or menFs [[170]8, [171]20, [172]24, [173]61, [174]82, [175]91, [176]102, [177]110]. Fig. 2. [178]Fig. 2 [179]Open in a new tab PDGFRα^+ fibroblasts with pial fibroblast-like immunophenotype express numerous cytokines and angiogenetic factors in the cerebral cortex of N-PRα-KO mice. a Immunofluorescence of PDGFRα and COL1 in WT and N-PRα-KO cortex at P10. Arrowheads: pvFs in the WT cortex limited in COL1^+ large-sized blood vessels. (b−g) Immunofluorescence of PDGFRα, b NG2, c CD31, d RALDH1, e RALDH2, f S100A6, and g PDGFRβ in N-PRα-KO cortex at P10. h pvF area in WT (WT) and N-PRα-KO (KO) cortex; n = 6−8, one area randomly selected from the left and right cerebral hemispheres. i indFs in WT (WT) and N-PRα-KO (KO) cortex; n = 6, two areas randomly selected per mouse. (j−n) mRNA expression levels of jAldh1a1, kAldh1a2, lCxcl12, mTgfb1, and nAngpt2 in WT (WT) and N-PRα-KO (KO) brains; n = 5−6 per genotype; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT at the same time point. ^##p < 0.01, ^####p < 0.0001 versus P5 in the same genotype. All values represent means ± SEM. Nuclei were counterstained with Hoechst. Scale bars, 25 μm (b, d, and e), 250 μm (a), 20 μm (c), and 50 μm (f and g). o Immunofluorescence of CXCL12, PDGFRα, and CD31 in N-PRα-KO cortex at P15. Cyan arrowheads: PDGFRα^+/ CXCL12^+/CD31^− cells, Green arrowheads: PDGFRα^−/ CXCL12^−/CD31^+ cells. p Immunofluorescence of pSmad2/3, PDGFRα, and CD31 in N-PRα-KO cortex at P10. q Immunofluorescence of angiopoietin 2 and PDGFRα in N-PRα-KO cortex at P15. Arrowheads: PDGFRα^+/ angiopoietin2^+ cells. Nuclei counterstained with Hoechst. Scale bars, 25 μm (o and q) and 50 μm (p). (r−v) mRNA expression levels of rPdgfra, sPdgfrb, tCxcl12, uTgfb1, and vAngpt2 of cultured skin fibroblasts (skFs) and PDGFRα^+ brain fibroblasts (praFs) cultured with in vitro supplemented PDGF-AB (+ AB) or PDGF-BB (+ BB), or without PDGF ligands (−). *p < 0.05, **p < 0.01 versus control without PDGF treatment within the same fibroblast. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001, ^####p < 0.0001 versus skFs of the same culture condition; n = 3 per cultured condition. All values represent means ± SEM. See also Fig. [180]S1g−j, S2, and S3 Increased PDGFRα^+ cells in N-PRα-KO mice may correspond to fibrosis-inducing fibroblasts The distinct distribution of PDGFRα^+/NG2^−/COL1^+ cells at P10 and P15 suggested that pvFs may be recruited from menFs and migrate into deeper cortical regions along blood vessels (Figs. [181]1b and c and [182]2a, Fig. [183]S1g). To verify this, perivascular PDGFRα^+ cell recruitment was compared between N-PRα-KO and WT mice by examining the pvF area (Fig. [184]S1h). While pvF areas were comparable at P5, the pvF area increased in N-PRα-KO mice with age from P7 to P15 compared with WT mice (Fig. [185]2h). Individual PDGFRα^+ cells were abundant in the N-PRα-KO mouse brain. Their cytoplasm contained multiple spiny cellular processes contacting the vascular wall and meninges (Fig. [186]2b, c, Fig. [187]S1i, j, 2a). They were distinguishable from OPCs based on their NG2 – status and unique morphology (Fig. [188]2b, Fig. [189]S1i). These features were consistent with fibrosis-inducing cells previously described as pvFs that have detached from the vasculature [[190]31, [191]61, [192]102]. Their abundance significantly increased with age in N-PRα-KO mice while remaining absent in WT mice at P10 and P15 (Fig. [193]2i). RALDH1 and S100A6 are pial fibroblast markers [[194]20, [195]61, [196]99]. In WT mice, RALDH1, S100A6, and RALDH2—also enriched in pial fibroblasts—were detected in leptomeninges and weakly detected in the parenchymal perivascular area at P10 (Fig. [197]S2b−d). PDGFRα was primarily expressed by OPCs and leptomeninges. In contrast, PDGFRβ was distributed in leptomeninges and parenchymal perivascular areas, consistent with the PDGFRβ expression in pvFs and pericytes (Fig. [198]S2e). In N-PRα-KO mice, immunoreactivity against RALDH1, RALDH2, S100A6, and PDGFRβ was detected in leptomeninges as well as in pvFs and individually distributed PDGFRα^+/NG2^− cells, suggesting characteristics similar to pial fibroblasts (Fig. [199]2d−g; Fig. [200]S2b−i) [[201]20, [202]61, [203]91]. Moreover, the increases in pvF and individually distributed fibroblasts (indFs) were paralleled by increased expression of Aldh1a1 and Aldh1a2, which encode RALDH1 and RALDH2, respectively, in N-PRα-KO mice compared to that in WT at P10 and P15. However, the levels at P5 were comparable between the groups (Fig. [204]2j, k). Fibrosis is the pathological deposition of collagen-rich ECM produced primarily by fibroblasts. Fibrosis with perivascular ECM deposition has been reported in the CNS and peripheral organs [[205]23, [206]66, [207]69, [208]78]. Hence, the increased presence of pvFs and indFs in N-PRα-KO mice, cells with a close topological relationship and similar immunophenotypes to pial fibroblasts, supports their identity as fibrosis-inducing fibroblasts. PDGFRα^+ fibroblasts expressed cytokines, including angiogenic factors in the cerebral cortex of N-PRα-KO mice The involvement of increased fibroblasts in aberrant vascular formation previously observed in the cerebral cortex of N-PRα-KO mice was investigated [[209]44]. The accumulation of fibroblasts in the cerebral cortex of N-PRα-KO mice was accompanied by significantly upregulated expression of Cxcl12 from P5 to P15, and Tgfb1 and Angpt2 at P10 and P15, compared with age-matched WT mice (Fig. [210]2l−n; Fig. S3a−k). In contrast, the expression of Hif1a and Rtn4—whose encoded products contribute to myelination and angiogenesis in neonatal mouse brains—was similar between the two genotypes (Fig. [211]S3e, k) [[212]88, [213]112, [214]125]. Immunoreactivity against CXCL12, phospho-Smad2/3 (pSmad2/3)—an active downstream signal transducer of TGF-β receptors—and angiopoietin 2 was predominantly observed in PDGFRα^+ menFs, pvFs, and indFs in the cerebral cortex of N-PRα-KO mice at P10 and P15 (Fig. [215]2o−q). In contrast, CXCL12 and angiopoietin 2 were undetectable around parenchymal blood vessels, and pSmad2/3 was limited to a few menFs in WT mice (Fig. [216]S3l−n). PDGFRα^+ brain fibroblasts (praFs)—assumed to be menFs, pvFs, and indFs—were harvested from the cerebral cortex of P13 N-PRα-KO mice and cultured in vitro with or without PDGF-AB or -BB to mimic the increase in those factors observed in the brain (Fig. [217]S1d, e). Identically treated skin fibroblasts (skFs) from WT mice were used as controls. Pdgfra levels were similar between skFs and praFs under all culture conditions (Fig. [218]2r). Meanwhile, Cxcl12 and Tgfb1 expression was comparable between skFs and praFs in untreated conditions (Fig. [219]2s−v; Fig. [220]S3o−t). Compared to skFs, Cxcl12 was upregulated in praFs cultured with PDGF-BB, but not PDGF-AB; Tgfb1 in praFs was more highly expressed in praFs after PDGF-AB and PDGF-BB treatment (Fig. [221]2t, u). Pdgfrb and Angpt2 were more highly expressed in praFs compared with skFs under all culture conditions (Fig. [222]2s, v). Pdgfrb was significantly induced in skFs and praFs by PDGF-BB treatment, and Angpt2 was induced in praFs, not skFs, by PDGF-AB and -BB treatment. These in vivo and in vitro data suggest that menFs, pvFs, and indFs possess a unique capacity to express angiogenic and fibrogenic factors that may contribute to the aberrant vascular formation observed in the cerebral cortex of N-PRα-KO mice. These upregulated factors are pleiotropic and can promote inflammatory tissue reactions, cellular damage, and tissue remodeling, suggesting the role of menFs, pvFs, and indFs in the development of cortical lesions in N-PRα-KO mice. Transcriptional profiling suggests a role for Pial fibroblasts in the cortical lesions of N-PRα-KO mice Single-cell RNA sequencing (scRNA-seq) analyses were conducted to characterize fibroblasts involved in the formation of cerebral cortical lesions. Using PhenoGraph for unbiased single-cell clustering analyses, 32 cell clusters were segregated in a uniform manifold approximation and projection (UMAP) display (Fig. [223]S4a). The cell type for each cell cluster was determined by employing previously reported representative cell type-specific transcripts (Fig. [224]S4b−v, Note S1). The majority of the cell clusters contained cells from both control WT and N-PRα-KO mice, but closely mirrored the effects of conditional gene targeting, with WT cells clearly predominating in the OPC clusters where Pdgfra was inactivated, whereas no such clear population bias was observed in the fibroblast clusters where Pdgfra was not inactivated (Fig. [225]S4b, c). The cells from fibroblast clusters 11, 13, 18, 24, 27, and 30 were re-clustered into subclusters I–IX through non-hierarchical analyses using Seurat (Fig. [226]3a). Clusters I and VI were dominated by N-PRα-KO cells, while the others contained a similar number of cells from N-PRα-KO and WT (Fig. [227]3b−d). Fig. 3. [228]Fig. 3 [229]Open in a new tab Implication of fibroblasts with gene expression profiles of pial fibroblasts in the cortical lesion of N-PRα-KO mice. (a−c) UMAP projection of nine clusters segregated by Seurat; a genotypes overlaid, b WT, and c N-PRα-KO. d Nine cell clusters identified by non-hierarchical analysis, Seurat. e Heatmap of the top 10 genes for each cell cluster, such as Col4a1, Col15a1, and Fbln2. Asterisks indicate the top 10 genes overlapping with those in Cluster VI. Purple: minimal, black intermediate, and yellow high expression. f Heatmap of the compartment-specific genes in meningeal fibroblasts (pia, arachnoid, and dura matter). Expression of each gene is color-coded on the logFC of the dataset: blue denotes minimal, white intermediate, and red high expression. g Pseudo-time plots of Clusters I, VI, and VII in (A) that have pial gene expression profile; sub-divided into five states: i−v. h Five states of fibroblasts identified by pseudotime analysis, Monocle. (i, j) Pathway enrichment analyses of i Clusters I, VI, and VII, and j cellular states i−v analyzed according to gene ontology (GO) terms. The upregulated genes in each state were evaluated compared to state iv. The upregulated genes in state iv were detected by comparison with all other states. See also Fig. [230]S4−[231]S6 and Note [232]S1 In the heatmap of the top 10 differentially expressed genes (DEGs), the genes in Cluster I were also highly expressed in Clusters VI and VII; three genes in Cluster I overlapped with those in Cluster VI, indicating a close lineage relationship among these three clusters (Fig. [233]3e). Consistently, the heatmap analysis revealed patterns corresponding to pial fibroblasts in Clusters I, VI, and VII, while Clusters II, IV, V, and VIII exhibited arachnoid fibroblast-like patterns; Cluster III displayed a dura fibroblast-like expression profile (Fig. [234]3f) [[235]20, [236]91, [237]99]. Cluster IX was excluded from analysis as its marker genes suggested the presence of microglial cells (Fig. [238]3e). CNS fibrosis has been suggested to be mediated by parenchymal pvFs with close transcriptional similarity to pial fibroblasts, including Col15a1, Lama1, Fbln2, Co4a1, and Lpl expression [[239]8, [240]20, [241]61, [242]91]. Consistently, the same genes were specifically enriched in the pial fibroblast Clusters I, VI, and VII (Fig. [243]3e, f). Considering that the fibrosis-inducing pvFs and indFs in the cerebral cortex of N-PRα-KO mice exhibited close spatial relationships and immunophenotypes with pial fibroblasts (Figs. [244]1 and [245]2, Fig. [246]S1, 2), subsequent analyses focused on Clusters I, VI, and VII. Within Clusters I−VIII, Gene Ontology (GO) enrichment analysis revealed that DEGs in Clusters I and VI were associated with fibrosis pathways, including blood vessel development, cellular locomotion, cellular adhesion, and extracellular structure organization (Fig. [247]3i). These transcripts were upregulated in N-PRα-KO cells compared with WT cells in Clusters I and VI (Fig. [248]S5a−c). GO enrichment analysis also revealed cell cycle-associated transcripts in Cluster VII, containing similar cell abundances from N-PRα-KO and WT mice and demonstrated similar enrichment of GO term-related gene induction (Fig. [249]3d, i, Fig. [250]S5d). These findings suggest that fibroblasts in Clusters I and VI are fibrosis-inducing cells, and those in Cluster VII are actively proliferating cells with a similar transcriptional profile to WT fibroblasts in the cerebral cortex of N-PRα-KO mice. The subsequent pseudo-time analysis using Monocle [[251]11] segregated Clusters I, VI, and VII into five cellular states, including specific N-PRα-KO cellular states (Fig. [252]3g, h, Fig. [253]S6a). State iv, comprising both genotypes, shifted to the N-PRα-KO-specific states i−iii and v as pseudo-time progressed from left to right. GO enrichment analysis revealed that state iv was enriched in genes associated with cell division and the cell cycle, similar to Cluster VII (Fig. [254]3j, Fig. [255]S6f, g); other N-PRα-KO-specific states were enriched in pathways associated with fibrosis and inflammatory cellular response, similar to Clusters I and VI (Fig. [256]3j, Fig. [257]S6b−e, h, i). A recent study reported six distinct identities of brain and leptomeningeal fibroblast transcriptomes in normal mouse brains, i.e., BFB1−6; BFB1 includes parenchymal pvFs (BFB1a) and pia and neighboring perivascular fibroblasts (BFB1b) [[258]91]. Hence, pial fibroblasts in the current study were compared with BFB1a or BFB1b. A heat map of genes enriched in BFB1a or BFB1b was created, focusing on pial fibroblasts of states i−v; N-PRα-KO and WT fibroblasts within state iv were analyzed separately (Fig. [259]4a, b). Compared with WT fibroblasts in state iv, N-PRα-KO fibroblasts in states i−v were more enriched with BFB1a marker genes (Fig. [260]4a), but not BFB1b marker genes (Fig. [261]4b). Fig. 4. [262]Fig. 4 [263]Open in a new tab Partial transcriptional transition from BFB1b to BFB1a subtype in pial fibroblasts, and precedent BAM activation in the cerebral cortex of N-PRα-KO mice. (a, b) Heat map of the representative cell type-specific genes of a BFB1a cells and b BFB1b cells in the cellular states obtained by Monocle. Blue denotes minimal, green to yellow intermediate, and red high expression. (c−t) Violin plots of cCol4a1, dItih5, eCol15a1, fCol4a2, gEdn3, hAldh1a1, iSpp1, jLpl, kCol12a1, lEce1, mCcdc80, nFam180a, oSerpine2, pMgp, qClec3b, rNgfr, sTnxb, and tCpxm2 in the five cellular states. Cyan: cells from WT mice, red: cells from N-PRα-KO mice. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT cells of state iv. Red asterisks: upregulation, blue: downregulation. u Immunofluorescence of CD206 in the cortex of WT and N-PRα-KO mice at P1, P7, and P10. Nuclei counterstained with Hoechst. Scale bars, 75 μm. (v, w) CD206⁺ cells of v meninges and w perivascular area in WT (WT) and N-PRα-KO (KO) cortex; n = 12, four areas randomly selected per mouse for meninges; n = 6−8, two areas randomly selected per mouse for perivascular areas. (x, y) Immunofluorescence of x CD206 and LYVE1, y LYVE1 and CCL5 in N-PRα-KO cortex at P7 (KO P7). Nuclei counterstained with Hoechst. Scale bars, 25 μm (x and y). (z-ad) mRNA expression levels of zCcl5, aaCcr5, abTnf, acPtgs1, and adPtgs2 in WT (WT) and N-PRα-KO (KO) brains; n = 5−6 in each genotype. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT at the same time point. All values represent means ± SEM Violin plots were used to compare gene expression profiles between N-PRα-KO fibroblasts and WT fibroblasts (Fig. [264]4c−t). Expression profiles were similar between N-PRα-KO and WT fibroblasts within states iv, excluding the upregulated three BFB1a genes (Fig. [265]4f, i, l) and one BFB1b gene (Fig. [266]4p) in the former. However, compared to WT fibroblasts in state iv, seven BFB1a genes were upregulated in states i and ii, and eight BFB1a genes were upregulated in states iii and v (Fig. [267]4c−n). Among BFB1b genes, Mgp expression was elevated in N-PRα-KO fibroblasts across states i−v (Fig. [268]4o−t). Accordingly, the transcriptional profile in state iv most closely resembled the BFB1b subtype, while the profiles in states i-iii and v resembled the BFB1a subtype. WT fibroblasts were detected only in state iv among pial fibroblasts, enriched in cell cycle-related genes, and exhibited similar expression as the BFB1b subtype. The histological data showed PDGFRα^+ fibroblasts, primarily localized in the meninges of WT mice. Actively proliferating meningeal fibroblasts have been previously identified, and a portion postnatally develop into parenchymal pvFs in the normal neonatal mouse brain; however, the identity of these postnatally developed pvFs and fibrosis-inducing fibroblasts in CNS largely remains to be explored [[269]20, [270]56, [271]61, [272]91]. The current pseudo-time and histological analyses showed that the recruitment of actively proliferating pial fibroblasts of the BFB1b subtype into parenchymal pvFs of the BFB1a subtype was largely accelerated, suggesting that these recruited cells are involved in tissue remodeling as profibrotic fibroblasts in the cerebral cortex of N-PRα-KO mice. BAM activation and upregulated inflammatory cytokines preceded pvF recruitment in N-PRα-KO mice Similar to normal development, the increased recruitment of menFs to the perivascular region of the brain parenchyma was accompanied by meningeal BAMs in N-PRα-KO mice [[273]56, [274]75]. In WT mice, a small number of CD206^+ cells were distributed in the meninges from P1 to P10 and the perivascular areas of large-sized penetrating blood vessels at P10 (Fig. [275]4u). In comparison, N-PRα-KO mice contained a greater number of CD206^+ cells both in the meninges from P1 to P10 and along the penetrating blood vessels from P7 to P10, where they also occurred deeper in the brain parenchyma (Fig. [276]4u). This increase in CD206^+ cells was significant in the meninges from P1 to P7, as well as along with penetrating blood vessels at P10 (Fig. [277]4v, w). Co-labeling of CD206^+ cells with LYVE1 confirmed that the cells were BAMs rather than microglia or monocytes (Fig. [278]4x) [[279]27, [280]119]. CCL5—an inflammatory cytokine—was identified in LYVE1^+ BAMs in the meninges of N-PRα-KO mice (Fig. [281]4y). Inflammatory cytokine-related genes were also induced in the N-PRα-KO mouse cerebral cortex shortly after birth in conjunction with BAM activation (Fig. [282]4z−ad). The expression of Ccl5 increased at P5, while its receptor Ccr5 increased at P5 and P10 in N-PRα-KO mice compared with age-matched WT mice (Fig. [283]4z, aa). Tnf expression was significantly higher at P5 and P15, while Ptgs1 expression was higher at P5 and P10 in N-PRα-KO mice than in WT mice (Fig. [284]4ab, ac). Ptgs2 expression exhibited an increasing trend at all time points in N-PRα-KO mice, albeit no individual time point reached statistical significance (Fig. [285]4ad). The recruitment of PDGFRα^+ fibroblasts increased from P7 (Fig. [286]2h, i). Thus, BAM activation preceded PDGFRα^+ fibroblast recruitment; these two closely-related phenomena suggest a functional correlation between BAMs and brain fibroblasts in the meninges and brain parenchyma of N-PRα-KO mice; a correlation that is explored in the following section. Transcriptional profiling suggests the importance of BAM and pvF in forming cortical lesions in N-PRα KO mice The macrophage−fibroblast interaction, based on CSF1/PDGF-B-centered reciprocal signaling, has been implicated in tissue remodeling under physiological and pathological conditions [[287]1, [288]9, [289]57, [290]77, [291]108, [292]127]. Pdgfb was enriched in the BAM, microglia, endothelial cell, and mural cell clusters, although expression levels were similar between the two genotypes within each cell cluster (Fig. [293]5a, b, Fig. [294]S7a−c). Meanwhile, expression of Pdgfra and Pdgfrb was respectively increased in N-PRα-KO fibroblasts in states i, iii, iv, and v, and states i−iii and v, compared with WT fibroblasts in state iv; the highest expression of Pdgfra and Pdgfrb was observed in state i (Fig. [295]5c−e). The CSF1–CSF1R axis mediates key signals for macrophage and microglia survival, proliferation, and migration [[296]13, [297]25]. Csf1 was enriched in the fibroblast, microglia, and endothelial cell clusters (Fig. [298]5f). Csf1 was significantly induced in pial fibroblast Clusters I, VI, and VII, compared with non-pial clusters comprising arachnoid and dura fibroblast Clusters II−V and VIII. This induction was especially enriched in N-PRα-KO fibroblasts in cluster I (Fig. [299]5g, Fig. [300]S7d). Csf1 was the most abundant in state i (Fig. [301]5h), with many N-PRα-KO fibroblasts overlapping between Cluster I and state i (Fig. [302]3h). Csf1r was selectively enriched in four microglia and BAM clusters, with higher expression in N-PRα-KO cells than those in WT cells (Fig. [303]5i, j). Correspondingly, CSF1 immunoreactivity was significantly increased in PDGFRα^+ menFs, pvFs, and indFs in N-PRα-KO mice than in WT mice (Fig. [304]5k, l). Fig. 5. [305]Fig. 5 [306]Open in a new tab Transcriptional profiling defines cell–cell interactions between brain macrophages and fibroblasts in N-PRα-KO mice. a Violin plots of Pdgfb in 32 cell clusters obtained by PhenoGraph. b Violin plots of Pdgfb in endothelial cell, microglia, mural cell, and BAM clusters within the 32 cell clusters obtained by PhenoGraph, comparing WT and N-PRα-KO derived cells. (c, d) Violin plots of cPdgfra and dPdgfrb in pial fibroblast states i−v obtained by Monocle. e Gene expression encoding major growth factor receptors in pial fibroblast states i−v obtained by Monocle. f Violin plots of Csf1 in 32 cell clusters obtained by PhenoGraph. gCsf1 in eight fibroblast clusters obtained by Seurat, comparing WT and N-PRα-KO derived cells. h Violin plots of Csf1 in five pial fibroblast states obtained by Monocle. i Violin plots of Csf1r in 32 cell clusters obtained by PhenoGraph. j Violin plots of Csf1r in microglia, mural cell, and BAM clusters within the 32 cell clusters obtained by PhenoGraph, comparing WT and N-PRα-KO derived cells. (k, l) Immunofluorescence of PDGFRα (red) and CSF1 (green) in the cortex of k WT and l N-PRα-KO mice at P15. Nuclei counterstained with Hoechst. Scale bar, 10 μm. m Violin plots of Tgfb1 in 32 cell clusters obtained by PhenoGraph. n Violin plots of Tgfb1 in microglia and BAM clusters within the 32 cell clusters obtained by PhenoGraph, comparing WT and N-PRα-KO derived cells. (o−t) Violin plots of oTgfbr2, pTgfbr3, qActa2, rMyl9, sTagln, and tTns1 in pial fibroblast states i−v obtained by Monocle. EC, endothelial cell; MG, microglia; Fb, fibroblast. Cyan: cells from WT mice, red: cells from N-PRα-KO mice in (b−e, g, h, j, n−t). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT cells of each cluster in (b, j, n), versus WT cells of state iv in (c, d, h, o−t). Red asterisks: upregulation, blue: high expression; and size represents the percentage of cells expressing each marker in (e, g). See also Fig. [307]S7 Hence, the cell–cell interactions between pial fibroblasts and CNS macrophages, including BAMs and microglia, were suggested to contribute to inflammatory tissue remodeling in the cerebral cortex of N-PRα-KO mice through the mutual exchange of CSF1/PDGF signaling. TGF-β signaling, acting directly or synergistically with other growth factors, mediates pleiotropic effects, including the fibroblast−myofibroblast transition, ECM synthesis, angiogenesis, and cell death. TGF-β strongly enhances PDGF-induced fibroblast migration [[308]19, [309]92, [310]96]. Moreover, TGF-β in brain macrophages is involved in inflammatory neurological diseases [[311]24, [312]55, [313]85, [314]113]. pSmad2/3 was primarily detected in PDGFRα^+ fibroblasts in the meninges and large penetrating vessels of N-PRα-KO mice (Fig. [315]2p). Tgfb1 was enriched in microglia and BAM clusters, with higher expression in N-PRα-KO than in WT cells in microglia from Cluster 12 and in BAMs from Cluster 25 (Fig. [316]5m, n). Tgfbr2 and Tgfbr3 expression was higher in states of N-PRα-KO fibroblasts than in state iv of WT fibroblasts (Fig. [317]5o, p, Fig. [318]S7e), indicating augmented TGF-β signaling in N-PRα-KO fibroblasts. Accordingly, marker genes for smooth muscle differentiation (Acta2, Myl9, Tagln, and Tns1) were induced in all N-PRα-KO fibroblast states, excluding Tagln in states iii and v, compared with WT fibroblasts (Fig. [319]5q−t). TGF-β1 contributes to fibrosis in the cerebral cortex of N-PRα-KO mice as the fibroblast transition into heterogeneous, persistent myofibroblasts, a hallmark of tissue fibrosis [[320]101, [321]124]. CXCL12 was predominantly localized in PDGFRα^+ pvFs in N-PRα-KO mice (Fig. [322]2o). Moreover, Cxcl12 was enriched in fibroblast and endothelial cell clusters (Fig. [323]6a), with significantly higher expression in pial fibroblast Clusters I, VI, and VII compared with that in non-pial clusters, with the highest abundance in the N-PRα-KO fibroblasts of Cluster VI (Fig. [324]6b, Fig. [325]S7f). Cxcl12 was induced in N-PRα-KO fibroblasts in states iii and v compared with WT fibroblasts (Fig. [326]6c); many N-PRα-KO fibroblasts overlapped between Cluster VI and state iii (Fig. [327]3h). Cxcr4, the CXCL12 receptor, was enriched in astroglial cell and endothelial cell clusters (Fig. [328]6d). The CXCL12–CXCR4 axis mediates astroglial proliferation [[329]7] and is a key signal axis in angiogenesis [[330]97, [331]130]. These results suggest that pvFs likely promote angiogenesis and astrogliosis, consistent with GO enrichment analyses revealing enriched blood vessel development-related gene enrichment in Clusters I and VI (Fig. [332]3i). Meanwhile, CXCL12, a surrogate marker for meningeal inflammation in PMS [[333]73]. Similar to Cxcl12, many genes encoding other surrogate PMS markers were significantly upregulated in N-PRα-KO compared to WT fibroblasts (Note S2, Figure S8), further highlighting the role of pvFs in forming cortical lesions, similar to those induced in PMS as one of the most representative pathological changes [[334]21, [335]52]. Fig. 6. [336]Fig. 6 [337]Open in a new tab Transcriptional profiling discloses the molecules related to cell–cell interactions between brain macrophages and fibroblasts in N-PRα-KO mice. a Violin plots of Cxcl12 in 32 cell clusters obtained by PhenoGraph. b Dot plots of Csf1 in eight fibroblast clusters obtained by Seurat, comparing WT and N-PRα-KO derived cells. Gray: minimal expression, blue: high expression; size represents the percentage of cells expressing each marker. c Violin plots of Cxcl12 in five pial fibroblast states obtained by Monocle. d Violin plots of Cxcr4 in 32 cell clusters obtained by PhenoGraph. e Violin plots of Tnf in 32 cell clusters obtained by PhenoGraph. f Violin plots of Tnf in microglia and BAM clusters within the 32 cell clusters obtained by PhenoGraph, comparing WT and N-PRα-KO derived cells. g Violin plots of Cyba in 32 cell clusters obtained by PhenoGraph. h Violin plots of Cyba in microglia and BAM clusters within the 32 cell clusters obtained by PhenoGraph, comparing WT and N-PRα-KO derived cells. i Violin plots of Tnfrsf1a in pial fibroblast states i−v obtained by Monocle. (j, k) Violin plots of Itga1 in eight fibroblast clusters obtained by j Seurat and in five pial fibroblast states obtained by k Monocle. (l, m) Violin plots of Cdh11 in eight fibroblast clusters obtained by l Seurat and in five pial fibroblast states obtained by m Monocle. (n, o) Violin plots of Yap1 in eight fibroblast clusters obtained by n Seurat and in five pial fibroblast states obtained by o Monocle. (p, q) Violin plots of Wwtr1 in eight fibroblast clusters obtained by p Seurat and in five pial fibroblast states obtained by q Monocle. r Violin plots of Gas1 in five pial fibroblast states obtained by Monocle. EC, endothelial cell; MG, microglia; Fb, fibroblast. The cyan numerical denotes cells from WT mice, and the red numerical denotes cells from N-PRα-KO mice in (b, c, f, h, i, k, m, o, q, and r). *p < 0.05; **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT cells of each cluster (f and h), versus WT cells of state (c, i, k, m, o, q and r), versus the entire cell population excluding clusters I, VI and VII (b, j, l, n, and p). Red asterisks: upregulation, blue asterisks: downregulation. (s−u) Immunofluorescence of YAP1 (green) and PDGFRα (red) in the cortex of WT (s) and N-PRα-KO (t, u) mice at P15. Nuclei were counterstained with Hoechst. Scale bar, 10 μm (s, t) and 5 μm (u). White arrowheads: PDGFRα^+ fibroblasts with positive nuclear staining for YAP1 (t, u). See also Fig. [338]S7, 8 and Note S2 Axon damage and neuronal oxidative stress were detected in N-PRα-KO mice (Fig. [339]7d) [[340]44]. TNFα and p22phox, a subunit of the NADPH oxidase, are implicated in neurotoxicity and oxidative damage in different neurological diseases [[341]15, [342]33, [343]58, [344]131]. Tnf and Cyba (encoding 22phox) were enriched in all four cell clusters of microglia and BAMs, but not Nox1 (encoding NADPH oxidase) (Fig. [345]6e, g, Fig. S7g). Tnf was upregulated in the microglia of Cluster 12, and Cyba was upregulated in both cell types from the corresponding four cell clusters in N-PRα-KO compared with WT cells (Fig. [346]6f, h). Therefore, activated microglial cells and BAMs by pial fibroblast-derived CSF1 may contribute to neuronal cell damage in N-PRα-KO mice. Additionally, Tnfrst1a (encoding TNF receptor) was more abundantly expressed in N-PRα-KO fibroblast states than in state iv of WT fibroblasts (Fig. [347]6i). TNFα promotes fibroblast-to-myofibroblast differentiation [[348]51, [349]103]. Concordantly, the pvFs of N-PRα-KO mice were distributed near BAMs with enriched Tnf expression. TNFα may also work in concert with TGFβ to promote fibroblast−myofibroblast transition. Fig. 7. [350]Fig. 7 [351]Open in a new tab PDGFRα neutralization suppresses pvF recruitment and ameliorates histological alterations, excluding BAM activation, in the cortex of N-PRα-KO mice. (a−d) Immunofluorescence of a MBP, b Iba1, c GFAP, d NeuN, and 4HNE in the cortex of WT and N-PRα-KO mice at P15. Nuclei counterstained with Hoechst. Scale bars, 250 μm (a), 25 μm (b), 100 μm (c), 10 μm (d). (e−g) e Penetrating blood vessels, f Iba1⁺ cells, and g GFAP⁺ areas in the cortex of WT (WT) and N-PRα-KO (KO) mice; n = 8−10, two areas randomly selected from each mouse (e, g); n = 12, four fields randomly selected per mouse f. h mRNA expression levels of Gfap in WT (WT) and N-PRα-KO (KO) mouse brain. n = 5–6/genotype. i 4-HNE expression levels in NeuN⁺ neurons in the cortex of WT(WT) and N-PRα-KO (KO) mice at P15; n = 30−40, ten NeuN^+ neurons randomly selected per mouse. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT at the same time point. All values represent means ± SEM. (j−n) Immunofluorescence of j PDGFRα, k CD31, l Iba1, m GFAP, and n CD206 in the cortex of PBS-treated (KO P15 + PBS) and PDGFRα-neutralizing antibody-treated N-PRα-KO (KO P15 + Neu) mice at P15. Nuclei counterstained with Hoechst. Scale bars, 250 μm (j and m), 75 μm (k and n), and 25 μm (l). (o−t) o pvF area, p penetrating blood vessels, q Iba1^+ cells, r GFAP^+ areas, and CD206^+ cells in s meninges and t parenchymal perivascular area of N-PRα-KO mice treated with PDGFRα-neutralizing antibody (Neu) (black hatched), and age-matched control N-PRα-KO mice treated with vehicle (PBS) (white hatched) or control IgG (IgG2) (gray hatched); n = 4−7 mice/genotype (o, p, r), n = 8−10, two areas randomly selected per mouse (q); n = 12, four areas randomly selected per mouse (s); n = 6−8, two areas randomly selected per mouse (t). *p < 0.05, **p < 0.01, ****p < 0.0001 versus age-matched vehicle treated controls. ^#p < 0.05, ^##p < 0.01 versus non-immune IgG treated controls. All values represent means ± SEM. See also Fig. [352]S9 Cell–cell contact and related signals are crucial for the functional interactions between mononuclear phagocytes and fibroblasts [[353]127]. Integrin subunit alpha 1 (ITG1) supports BAM migration to parenchymal perivascular tissues with pvFs [[354]56, [355]75]. Cadherin-11 mediates cellular adhesion and paracrine TGF-β signaling, establishing a stable profibrotic niche [[356]68, [357]127]. Itga1 expression was significantly upregulated in each pial fibroblast Cluster (I, VI, and VII) compared with that in non-pial clusters. This effect was significantly higher in N-PRα-KO fibroblasts in states i−iv than in WT fibroblasts (Fig. [358]6j, k). Cdh11 was also enriched in fibroblast clusters (Fig. [359]S7h), the dura fibroblast Cluster III, and pial fibroblast Clusters I, VI, and VII, compared with non-pial clusters (Fig. [360]6l). Cdh11 was upregulated in N-PRα-KO compared with WT fibroblasts, most abundantly in state i (Fig. [361]6m). Therefore, in the profibrotic niche of N-PRα-KO mice, fibroblast-derived ITG1 and cadherin-11 may encourage stable cell–cell interactions, inducing reciprocal exchange of paracrine signaling between BAMs and fibroblasts. YAP1 and TAZ, downstream of the Hippo pathway, are induced in fibroblasts via macrophage contact and are crucial for organ fibrosis [[362]26, [363]46, [364]111]. Yap1 was significantly expressed in pial fibroblast Clusters I and VII, and Wwtr1 (encoding TAZ) in pial fibroblast Clusters I, VI, and VII compared with those in non-pial clusters. Yap1, but not Wwtr1, was higher in N-PRα-KO fibroblasts in state i than that in WT fibroblasts (Fig. [365]6n−q, Fig. S7i, j). N-PRα-KO fibroblasts were frequently accompanied by a low cell density signal for Gas 1, but not by a high cell density signal for Serpine1, suggesting that Yap1 and Wwtr1 were induced via the Hippo pathway (Fig. [366]6r, Fig. [367]S7k) [[368]111]. Yap1 and Wwtr1 induction was often associated with Csf1 enrichment in N-PRα-KO fibroblasts from Cluster I and state i (Fig. [369]5g, h). In contrast, the induction of Csf1 in astrocytes was not accompanied by Yap1 or Wwrt1 induction (Fig. [370]S7l−n). Therefore, YAP1 may specifically induce Csf1 in pial fibroblasts, but not in astrocytes, in N-PRα-KO mice [[371]40, [372]129]. This Yap1 induction was likely accompanied by YAP1 activation in N-PRα-KO mice, as YAP1 immunoreactivity in PDGFRα^+ menFs and pvFs was observed in both nuclei and cytoplasm of N-PRα-KO mice. In contrast, in WT mice, YAP1 immunoreactivity in these cells was limited to the cytoplasm (Fig. [373]6s−u, Fig. [374]S7o−q). Therefore, these findings suggest that YAP1 and TAZ play key roles in fibroblast-mediated tissue remodeling, including CNS fibrosis, in N-PRα-KO mice. More specifically, YAP1 and TAZ may contribute to the fibroblast-to-myofibroblast transition [[375]26, [376]46, [377]74]. Taken together, the cell signaling pathways shown here suggested the mechanisms by which the close cell–cell interactions are established between pial fibroblasts and CNS macrophages and mediate tissue-damaging chronic and persistent inflammatory changes in the subpial cerebral cortex. Hypomyelinated cerebral cortex exhibits inflammatory changes in N-PRα-KO mice Immunofluorescence showed that myelin basic protein (MBP)^+ myelin sheaths had developed in WT mice, and were diffusely hypoplastic in PRα-KO mice at P15 in the parietal cerebral cortex (Fig. [378]7a). Compared with WT mice, penetrating blood vessels emanating from the meninges of the parietal surface of the cerebral cortex increased in N-PRα-KO mice at P7, P10, and P15; however, their density was comparable at P5 (Fig. [379]7e). The density of Iba1^+ microglia was similar at P5 and P7 but increased significantly at P10 and P15 in N-PRα-KO mice compared with that in age-matched WT mice (Fig. [380]7b, f). N-PRα-KO mice exhibited more extensive GFAP^+ cell distribution at P15, a larger GFAP^+ area at P7, P10, and P15, and higher Gfap expression at P10 and P15 than age-matched WT mice. This suggested that more GFAP^+-activated astrocytes were present in N-PRα-KO mice than in WT mice (Fig. [381]7c, g, h). Oxidative stress was observed in the cerebral cortical neurons of N-PRα-KO mice, as evidenced by the increased neuronal immunostaining for 4-hydroxy-2-nonenal (4-HNE) compared with WT mice at P15 in cortical lesions [[382]33, [383]43] (Fig. [384]7d, i). Leakage of macromolecules through the blood–brain barrier (BBB), measured by IgG extravasation, was similar in N-PRα-KO and WT mice at P7 and P10 but significantly higher in N-PRα-KO mice at P15 (Fig. [385]S9a). Nevertheless, the expression of ZO-1 and claudin 5—tight junction proteins of the BBB [[386]94]—and PLVAP—a marker of vascular fenestration [[387]39]—was similar between the two genotypes at P15 (Fig. [388]S9b−f). Therefore, increased transcellular transport may be associated with BBB leakage, as suggested in neurodegenerative diseases [[389]4, [390]80]. These data suggest that hypomyelinated cortical lesions were related to neuroinflammatory lesion-like changes, including aberrant vascular formation, glial activation, oxidative neuronal cell damage, and BBB leakage in the cortex of N-PRα-KO mice [[391]33, [392]43, [393]47, [394]70, [395]100]. PDGFRα-neutralizing antibodies were injected into the intrathecal spaces of N-PRα-KO mice to assess the role of PDGFRα^+ pvFs in cortical lesion formation. The pvF areas were significantly smaller in treated mice at P7 and P15 than those in age-matched controls, including those treated with PBS from P7 to P15 or with control IgG2 at P15 (Fig. [396]7j, o). This demonstrated the involvement of PDGFRα in pvF recruitment. The density of penetrating blood vessels also decreased significantly at P7, P10, and P15 in PDGFRα-neutralizing antibody-treated N-PRα-KO mice compared with that in age-matched controls (Fig. [397]7k, p). The PDGFRα-neutralizing antibody also substantially reduced the Iba1^+ cell proportion and GFAP^+ area at P15, whereas the number of BAMs in the meninges and the perivascular regions were unaffected (Fig. [398]7l−n, q−t). The PDGFRα neutralization study suggested that the recruited fibroblasts, at least partly dependent on PDGFRα signaling, contributed to cerebral cortical lesion formation in N-PRα-KO mice. Meanwhile, BAM activation was not affected by PDGFRα-neutralizing antibodies and, therefore, did not seem to be downstream of PDGFRα signaling or pvF recruitment (Figs. [399]4u−w and [400]7n, s and t). Discussion Well-coordinated regulation of different CNS resident cells directs the developmental process sequences in the early neonatal rodent brain [[401]22, [402]45, [403]88, [404]112, [405]125]. In the present study, the mechanisms involving numerous resident cells that mediate neuroinflammatory and degenerative responses in the cerebral cortex of neonatal N-PRα-KO mice, as a secondary consequence of developmental myelination failure, were elucidated. Meningeal BAM activation occurred first, followed by enhanced recruitment of pial fibroblasts as fibrosis-inducing pvFs and indFs deep in the brain parenchyma. BAMs and pial fibroblasts establish stable two-cell interactions through reciprocal signaling centered on the CSF1/PDGF-B signal exchange [[406]1, [407]127]. The inflammatory mechanisms associated with this two-cell interaction were suggested to sustain active inflammation in the cerebral cortex, leading to fibrosis, angiogenesis, glial activation, and neuronal damage in N-PRα-KO mice. Importantly, intrathecal administration of PDGFRα neutralizing antibody significantly reduced pial fibroblast recruitment and alleviated cortical inflammatory changes, supporting the contributory role of fibroblast-macrophage interactions in disease pathogenesis. Collectively, these results highlight brain macrophages and pial fibroblasts as potent mediators of cortical lesion formation. It has been suggested that parenchymal pvFs originate from meningeal fibroblasts during development and are considered a primary source of fibrosis-inducing fibroblasts under pathological conditions [[408]8, [409]20, [410]24, [411]56, [412]61]. Additionally, parenchymal pvFs (BFB1a) and pia and neighboring pvFs (BFB1b) form an independent cell population (BFB1) among brain and leptomeningeal fibroblasts [[413]91]. However, the differentiation process and the functional relevance of BFB1a and BFB1b fibroblasts remain underexplored. In the present study, pseudo-time analyses showed that actively proliferating pial fibroblasts had transcriptional profiles that overlapped with that of normal pial fibroblasts from WT mice, but transformed into myofibroblast-like fibrosis-inducing cells with different transcriptional profiles in N-PRα-KO mice accompanied by a partial transcriptional transition from the BFB1b to BFB1a subclass. Other scRNA-seq studies have similarly reported the dynamic processes of fibrosis-inducing cells to gain multilineage-like states not detected in normal resident cells during lung, heart, and skin fibrotic processes [[414]28, [415]106, [416]114]. Thus, in conjunction with morphological studies, the current scRNA-seq study showed that the actively proliferating normal pial fibroblasts represent a potentially important origin of fibrosis-inducing cells with myofibroblast-like transcriptional features in the cerebral cortex. This may be further supported by previous reports suggesting a role for actively proliferating resident fibroblasts in experimentally induced fibrosis of the spinal cord [[417]24, [418]102]. Cell–cell interactions between PDGF^+/CSF1R^+ macrophages and PDGFRα^+/CSF1^+ fibroblasts contribute to various homeostatic and pathological conditions [[419]1, [420]9, [421]57, [422]77, [423]108, [424]127]. Inhibiting this interaction may offer a therapeutic target for organ fibrosis, as CSF1R inhibition suppresses experimentally induced lung fibrosis [[425]9, [426]57, [427]77]. Perivascular regions have been considered important profibrotic niches since the discovery of pvFs as major fibrosis-inducing cells in the CNS [[428]24, [429]102]. In the N-PRα-KO mice, the profibrotic perivascular niche was characterized by enhanced interactions between BAMs and COL1^+/PDGFRα^+ fibroblasts, driven by CSF1/PDGF-B reciprocal signaling and paracrine factors, including TGFβ and TNFα. Moreover, the environment-sensing Hippo pathway genes, including Yap1 and Wwtr1, were upregulated in N-PRα-KO fibroblasts; their encoded proteins contribute to fibrosis by mediating fibroblast–to–myofibroblast differentiation [[430]26, [431]46, [432]74]. These findings collectively suggest that the cell–cell interactions in the perivascular profibrotic niche mediate inflammatory lesions in the cerebral cortex accompanied by glial activation, angiogenesis, and neuronal toxicities in N-PRα-KO mice. Thus, the cell interactions and accompanying signal network revealed in this study represent a potential therapeutic target for treating chronic and progressive neurological diseases, the pathogenesis of which has been suggested to involve intricate interactions of CNS resident cells involved in inflammation, fibrosis, and myelin damage [[433]6, [434]23, [435]29, [436]53, [437]60, [438]72, [439]81, [440]115]. Tissue-damaging fibrosis is caused by chronic stimulation, such as viral hepatitis, or a failure to replace damaged cells, which prolongs the inflammatory response [[441]10, [442]18, [443]79]. Accordingly, the progressive cortical lesions and neurological symptoms observed in N-PRα-KO mice may reflect mechanisms that prolong inflammation and mediate fibrotic tissue damage. Cdh11 induction in N-PRα-KO fibroblasts may contribute to destructive fibrosis in the CNS, as cadherin-11-mediated adhesion between macrophages and myofibroblasts is a key process that converts acute, beneficial repair into destructive, progressive fibrosis via TGFβ activation [[444]68]. Separately, scar formation has been observed in an in vitro co-culture model of macrophages and fibroblasts after repetitive or prolonged activation of macrophages [[445]1]. In MS, demyelination and death of iron-rich oligodendrocytes release excess iron, activating microglial cells and macrophages toward a pro-inflammatory phenotype [[446]104]. In N-PRα-KO mice, BAM activation preceded other pathological changes in the cerebral cortex. Therefore, tissue-damaging fibrotic responses may have been induced by persistent BAM activation via inflammatory stimuli, including iron ions derived from the Olig2^+ lineage, because this lineage underwent increased postnatal apoptotic cell death in N-PRα-KO mice [[447]44], although this remains hypothetical that needs to be clarified. Microglia/macrophage activity occurs early and intensifies during MS progression [[448]49, [449]105]. Diffuse demyelinating subpial gray matter lesions are a major pathological change potentially responsible for progressive neuronal deficits in PMS [[450]21, [451]52]. Moreover, diffusion of pro-inflammatory cytokines from the meningeal tissues mediates subpial cortical damage; however, the underlying mechanisms remain largely unclear [[452]55, [453]71, [454]109]. Meanwhile, BAM activation first occurred in the meninges of N-PRα-KO mice, followed by the inflammatory cerebral cortical lesions in N-PRα-KO mice, with similar key pathological characteristics including expanded ECM-rich perivascular niche populated by fibrosis-induing PDGFRα^+ cells, and overlapping proinflammatory cytokine expressions as reported in cortical MS lesions (Note S2, Fig. [455]S8) [[456]53, [457]64, [458]89, [459]131]. Thus, cell–cell interactions among CNS resident cells, centered on activated and recruited meningeal BAMs and pial fibroblasts recruited from meninges to brain parenchyma, may be involved in the pathogenesis of neurological diseases including cortical lesions in MS. Despite ongoing clinical trials, the etiopathogenesis of PMS remains poorly understood, partly due to the absence of an appropriate animal model [[460]65]. Myelination failure due to disrupted immature oligodendroglial lineage differentiation has been proposed as a pathogenic event in PMS [[461]12, [462]34, [463]63, [464]83]. These potential etiological and pathological commonalities are suggestive of shared mechanisms with N-PRα-KO mice, at least partially. Therefore, the disclosed pathogenetic mechanisms underlying cortical lesions in N-PRα-KO mice may provide an improved therapeutic approach for intractable PMS. However, a clear limitation of the N-PRα-KO mouse model is that, unlike PMS, an adult-onset disease, it represents developmental hypomyelination. Therefore, the development of animal models that more closely resemble PMS is warranted. A limitation of this study was that behavioral and intervention experiments were difficult due to the early lethality of N-PRα-KO mice. Additionally, the mechanism underlying BAM activation in these mice remains speculative. In summary, this study characterized the tissue remodeling associated with myelination failure in the cerebral cortex of neonatal N-PRα-KO mice. As a consequence of OPC depletion and myelination failure, activated resident cells, such as BAMs and pial fibroblasts, may mediate cortical damage and neuronal stress. The regulatory mechanisms of resident fibroblasts advance our understanding of the pathological fibrotic processes in the CNS. Furthermore, the identified mechanisms of cortical tissue remodeling may provide useful therapeutic targets for treating chronic neurological disease, including PMS. Electronic supplementary material Below is the link to the electronic supplementary material. [465]Supplementary Material 1^ (7.4MB, docx) Acknowledgements