MMP-9 affects dural cell composition and granulocyte accumulation during neuroinflammation

Introduction: Matrix metalloproteinases MMP-2 and MMP-9 modulate inflammatory processes at the blood–brain barrier (BBB), but their presence and function in the dura mater remain unclear. We therefore investigated their contribution to immune regulation at this site during neuroinflammation.

Methods: We analyzed dura mater from naïve and experimental autoimmune encephalomyelitis (EAE) mice using immunofluorescence, gelatin zymography, flow cytometry, and single-nucleus RNA sequencing (snRNA-seq). Comparisons were performed between wild-type (WT) and Mmp9-/- mice to define gelatinase expression, cellular sources, transcriptional alterations, and immune cell dynamics.

Results: Pro- and activated-MMP-2 and MMP-9 were detectable in naïve dura, with selective upregulation of MMP-9 during EAE. Mmp9-/- mice demonstrated delayed EAE onset but more severe disease at peak. SnRNA-seq revealed that Mmp9 deficiency altered endothelial transcriptional programs, especially in venous endothelial cells, and promoted expansion of granulocyte populations with mature neutrophil signatures. Flow cytometry and immunofluorescence confirmed increased Ly6C⁺Ly6G⁺ neutrophils in Mmp9-/- dura during EAE.

Discussion: Our findings indicate that MMP-9 regulates immune cell composition and activation within the dura during neuroinflammation. While Mmp9 deficiency delays disease onset, it enhances neutrophil accumulation and inflammatory responses at this CNS border, suggesting a previously unrecognized, potentially anti-inflammatory role for MMP-9 in the dura.

Introduction

The gelatinases, matrix-metalloproteinases (MMP)-2 and MMP-9, are expressed by immune and stromal cells (1–4) and were traditionally considered to function in extracellular matrix (ECM) remodeling during development and disease (5). However, increasing data suggest a major role in vivo in the fine tuning of inflammatory processes (6–8). We have previously demonstrated gelatinase expression by both immune cells and astrocytes during neuroinflammation (2, 9, 10). Astrocyte-derived gelatinases modulate cell-cell interactions and chemokine signaling at the blood-brain barrier (BBB) (8) particularly at the parenchymal/astroglial border of postcapillary venules, and are essential for T cell infiltration into the parenchyma of the central nervous system (CNS) during experimental autoimmune encephalomyelitis (EAE) (9). Whether MMP-2 and MMP-9 are also present and contribute to neuroinflammation in the dura layer of the meninges, which forms a border tissue of the CNS, remains unknown.

The multi-layered meninges have recently gained substantial interest due to the controversial discussion of their involvement in modulating CNS homeostasis and inflammation (11–13). The meninges ‘encase’ the brain parenchyma, forming an interface for neuroimmune interactions similar to that of BBB at the level of smaller vessels. The dura, the outermost layer of meninges, has been suggested to be an immune relevant tissue, containing not only lymphatic vessels but also dural-associated lymphoid tissues (14–16). Notably, in homeostatic conditions the dura contains a rich repertoire of immune cells, including B cell populations at different developmental stages, contributing to CNS immune surveillance and immune defense (17–19). Mechanisms controlling the composition of dura leukocytes remain poorly understood (20) and a potential contribution of gelatinases has not yet been studied.

Here, we demonstrate that both MMP-2 and MMP-9 are expressed and activated in the naïve WT murine dura, with specific upregulation of pro- and activated-MMP-9 during EAE. To identify potential gelatinase functions in the dura, single-nuclei transcriptomes were generated from dural cells of naïve and EAE wild-type C57BL/6 mice (WT), EAE and MMP-9 deficient (EAE Mmp9^-/-) mice. We identified cell clusters annotated as dural fibroblasts and osteoblasts as the primary cellular sources of Mmp2, while Mmp9 was predominantly expressed by granulocytes and osteoclasts. Focusing on Mmp9 due to its selective upregulation in EAE, a dural venous endothelial subtype was identified in Mmp9^-/- dura which displayed the greatest transcriptional alterations associated with enhanced inflammatory responses during EAE. Furthermore, a granulocyte subtype with mature neutrophil features was observed at higher proportions in the dura of Mmp9^-/- mice during EAE. The accumulation of the overall neutrophil population in these conditions was confirmed by flow cytometry and immunofluorescence staining. MMP-9, thus, may modulate immune responses within the dura during EAE, potentially by regulating the inflammatory milieu and myeloid cell populations.

Materials and methods

Animal subjects

Animal housing and ethical approval: All animal experiments were performed in accordance with institutional and national guidelines for animal welfare and approved by the Landesamt für Verbraucherschutz (LAVE) (Nordrhein-Westfalen, Deutschland; permit number AZ2024-42). Mice were housed under standard laboratory conditions (12 h light/dark cycle, temperature 20–24°C, relative humidity 45–65%) with free access to food and water. Animals were kept in groups in individually ventilated cages with appropriate bedding and environmental enrichment. All efforts were made to minimize animal suffering, and experiments were conducted in line with the 3Rs principle.

Euthanasia: No anesthesia was used in this study. For sacrifice, animals were euthanized by controlled exposure to carbon dioxide (CO₂) in accordance with the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals (2020). CO₂ was introduced into the chamber at a displacement rate of 30–70% of the chamber volume per minute. Cervical dislocation was not applied in order to preserve the brain tissue for subsequent analyses.

Gelatin zymography

For gelatin zymography, whole dura preparations of PBS perfused mice or small pieces of tissue isolated from new-born mice (40 mg) were meshed using 2.8 mm ceramic beads (Precellys) in 200 µl lysis buffer (cOmpleteTM EDTA-free proteinase Inhibitor cocktail (Roche), 0.1% Triton X-100 in PBS). 120 µl were recovered for gelatin prepurification by incubation with gelatin- Sepharose-4B for 20 min on a rotating wheel (21). Total purified protein was taken up in sample buffer (2x) and run on a 10% polyacrylamide gel containing 1 mg/ml gelatin under non-reducing conditions. After the electrophoresis, the gels were washed with H₂O containing 2.5% Triton X-100 (v/v) for 30 min followed by 2.5% Triton X-100 in H₂O for another 30 min to renature the gelatinases. Subsequently, the gels were placed in developing buffer (TBS, 5 mM CaCl₂, and 0.02% NP-40) overnight at 37°C. The next day, gels were fixed in acetic acid:ethanol:H₂O (10:50:40) for 30 min at RT and washed for 30 min in acetic acid:methanol:H₂O (10:50:40). Gels were then stained with Coomassie Brilliant Blue for 1h at RT, followed by washing of the gels acid:methanol:H₂O (10:50:40) for 120 min or until cleaved gelatin bands became visible. Areas of cleaved gelatin showed as clear bands that correspond to the activated MMP subunits; recombinant mouse MMP-2 and MMP-9 were used as molecular weight controls.

Immunofluorescence staining

Mice were euthanized, in case of EAE at the peak of disease severity; skull and attached dura were collected and fixed overnight in 2% paraformaldehyde (PFA) at 4°C. The next day samples were washed with PBS and the dura was manually peeled from the skull using a binocular.

For immunofluorescent staining whole mount dura were blocked for 2-3h in 1% BSA solution containing 1% Triton-X in a humidified chamber at room temperature. Subsequently, primary antibodies were applied overnight at 4°C. Primary antibodies included rat anti-mouse CD45.2 (30G12), Rabbit anti-mouse laminin 111 (22), rat anti-mouse CD4 (RM4-5) Alexa Fluor647 (BD Bioscience), Isolectin GS-B4 Alexa Fluor 647 (23), FITC rat anti-mouse Ly6G (24) and rat anti-mouse PECAM-1 clone: 2H8 (25). After washing thrice secondary antibodies including Alexa Fluor488 donkey anti-rat IgG (Invitrogen), Alexa Fluor594 donkey anti-rabbit IgG (Invitrogen) and Alexa Fluor647 goat anti-hamster IgG (Abcam) were applied together with DAPI (4′,6-diamidino-2-phenylindole) (1 μg/ml) to visualize nuclei. Sections were examined with LSM800 AxioVert microscope.

Image processing and quantification

For vessel density quantification, whole-mount dura tile images were analyzed using Imaris 10.0.1. Images were first stitched using ZEN Blue software, then converted from.czi to.ims format using the Imaris File Converter. The resulting.ims files were opened in Imaris for 3D analysis. Surface rendering was applied to segment a region of interest (ROI). The PECAM-1 (CD31) channel was selected for analysis. Surface smoothing and background-subtracted thresholding were used to refine the segmentation. Surfaces were classified, and the total PECAM-1-positive surface area within the ROI was measured. This total area was then normalized to the size of the ROI. The normalized vessel density data were plotted and statistically analyzed using GraphPad Prism.

Flow cytometric characterization of dural neutrophils in EAE dura

Dura mater was carefully peeled from naïve or EAE-induced C57BL/6 or Mmp9^-/- mice and immediately transferred to ice-cold PBS. Tissue was enzymatically dissociated in Collagenase type I and DNase I (37°C, 45 min, gentle agitation), then mechanically triturated and passed through a 70 µm cell strainer to obtain single-cell suspensions. Cells were washed and resuspended in FACS buffer (PBS, 2% FCS). For surface staining, cells were incubated for 30 min at 4°C with fluorochrome-conjugated antibodies (CD45-PerCP (BD Bioscience), CD45-FITC (BD Bioscience), CD4-PE (BD Bioscience), CD8-BV605 (BD Bioscience), B220-BV421 (BD Bioscience), CD11b-FITC (BD Bioscience), CD11c-APC (eBioscience), Ly6C-PE (BioLegend), Ly6G-APC (BioLegend). After staining, cells were washed, filtered, and analyzed on a BD FACSCelesta flow cytometer. Neutrophils were quantified using a standard gating strategy (FSC/SSC to exclude debris, singlet discrimination, life/dead and leukocyte (CD45) gate, and Ly6G⁴/Ly6C⁴ subsets). All steps were performed on ice or at 4°C unless noted otherwise.

Mice, EAE induction, and dura isolation for snRNA-seq

EAE was induced in C57BL/6 mice from Charles River Laboratories (Cologne, Germany) and MMP-9 (Mmp9^-/-) deficient mice bred on a C57BL/6 background (8). Female mice aged 8–12 weeks were immunized with MOG peptide (MOG_35-55, 112.5 µg per mouse, Rudolf Volkmer, Charite Berlin, Germany) and Mycobacterium tuberculosis H37RA extract (5 mg ml^-1, BD) emulsified in complete Freund’s adjuvant (150 µl per mouse). On days 0 and 2, pertussis toxin (200 ng per mouse) was injected intravenously. Mice were monitored daily, and disease severity was assigned as stage 1 (flaccid tail), stage 2 (hindlimb weakness), stage 3 (severe hindlimb weakness), stage 4 (hind quarter paralysis), and stage 5 (forelimb weakness). EAE incidence in WT and Mmp9^-/- mice was comparable. Only the mice that developed EAE were included in the experiment and in the statistical analyses. For snRNA-seq, the dura tissue was collected from naïve WT mice and from EAE samples at 14 days after immunization, when WT and Mmp9^-/- mice had similar disease scores (Figure 1C). Under terminal anesthesia, the mice were intracardially perfused with PBS, and the dura was peeled off from the dissected skull caps using fine forceps under a binocular. Dura preparations were kept at 4°C in PBS before nuclei extraction for snRNA-seq.

Figure 1

Figure 1. MMP-2 and MMP-9 are expressed and activated in the murine dura. (A) Whole-mount immunofluorescence of dura mater from naïve and peak EAE WT mice stained for pan laminin (Pan LM, basement membranes, magenta), CD45 (immune cells, green), and CD4 (CD4⁺ T cells, white). Scale bar = 250 µm. (B) Gelatin zymography of dura extracts from naïve, EAE onset, and peak EAE WT mice (two biological replicates per condition) showing pro- and activated forms of MMP-2 and MMP-9. (C) Clinical EAE scores in WT and Mmp9^-/- mice; data shown are means ± SEM from 12 WT and 6 Mmp9^-/- mice and is one example of 3 separate experiments using a total of 24 WT and 23 Mmp9^-/- mice. Statistical analysis was performed using the Mann-Whitney U test, **p < 0.01; ***p < 0.001. (D) Gelatin zymography of brain extracts from EAE WT and EAE Mmp9^-/- mice.

Nuclei preparation for snRNA-seq

Nuclei extraction was performed according to the Milteny Biotec nuclei extraction protocol (Catalog no. 130-128-024). Four freshly collected duras from each experimental condition were pooled. Hence, the single nuclei data are derived from multiple biological replicates that were pooled and processed together. Pooled samples were transferred to lysis buffer (RNA inhibitor (0.2 U/µl) in nuclei extraction buffer) and single nuclei suspensions were generated using a gentleMACS dissociator with gentleMACS program 4C_nuclei_1 at 4°C. To minimize the debris contamination, the resulting single nuclei preparations were passed through a 70 µm as well as 30 µm cell strainers and a resuspension buffer (PBS with 0.1% ultrapure BSA and 0.2 U/µl RNase inhibitor) was used between the filtration steps. Subsequently, individual single nuclei preparations were stained with trypan blue and manually inspected in a Fuchs-Rosenthal chamber for determining the total number of the intact nuclei. A total of 10,000 nuclei per condition were used for downstream procedures including single nuclei libraries construction and sequencing.

Generation of single nuclei libraries and sequencing

Single nuclei preparations were loaded onto the Chromium Single Cell Controller utilizing the Chromium Single Cell 3’ Library & Gel Bead Kit with v3.1 chemistry (10X Genomics). Quantification was performed with precision using the TapeStation High Sensitivity D1000 kit (Agilent Technologies) to ascertain the average fragment size and library concentration. Subsequently, the libraries underwent normalization, dilution, and were sequenced on a local Nextseq 2000 using the P3–100 cycle kit with a 28-10-10–90 read setup. Processing of sequencing data was performed with the cellranger pipeline v7.0.1 (10X Genomics) in accordance with the manufacturer’s instructions. Briefly, raw bcl files were analyzed using the cellranger mkfastq pipeline and subsequent read alignments and transcript counting were done individually for each sample with standard parameters.

Data analysis of single-nuclei RNA-sequencing

CellBender was employed for individual datasets to remove background noise (26) prior to downstream bioinformatic analysis using the R-package Seurat v4.4.1 (27) in datasets of EAE Mmp9^-/-, EAE Mmp2^-/- and WT samples, as well as a naïve murine dura were integrated after SCTransform (28). Low-quality cells (min.cell = 3) and cell doublets (scDblFinder.class = singlet) were excluded by filtering cells with few genes (<2% percentile), high number of genes (>97% percentile), high mitochondrial percentages (>5%), high percentages of erythrocytes (>2%), and high ribosomal percentages (>5%). The datasets were merged and normalized using SCTransform. The total remaining cell number for further analysis was 24219. To integrate the datasets and correct for batch effects, cells were aligned between different conditions with Harmony (29). Clusters were identified by the “FindNeighbors”and “FindClusters” (with resolution of 0.5) functions in Seurat v. 4.4.1 and Uniform Manifold Approximation and Projection (UMAP) was performed with Harmony embeddings to visualize the results. “FindMarker” function (Wilcoxon Rank Sum test) in Seurat v. 4.4.1 was applied to determine marker genes for each cell cluster. Four cell clusters were excluded for downstream analyses: a cluster expressing Tph1, thus likely representing pinealocytes; a cluster expressing Aqp4, thus likely representing astrocytes; a cluster expressing Ttr, thus likely representing epithelial-like cells, and a cluster expressing mainly ribosomal and mitochondrial RNA, thus representing low quality cells. Re-clustering was performed for the subset dataset (“FindClusters”; with a resolution of 0.37). To visualize individual cell transcriptomes, Uniform Manifold Approximation and Projection (UMAP) was performed with Harmony embeddings. We manually inspected the top DE genes of each cluster in a literature search cell cluster annotation. To determine DE genes between EAE conditions, scaled data were split by condition (WT, Mmp2^-/- and Mmp9^-/-) and the “FindMarker” function in Seurat v.4.4.1 (Wilcoxon Rank Sum test) was applied. DotPlots and FeaturePlots were created using the internal visualization functions of Seurat v.4.4.1. Volcano plots of DE genes were created using ggplot2 v.3.5.1 (30), and for GO enrichment analysis, the package ClusterProfiler v.4.12.0 (31) was used. Pseudotime analysis was performed on the vascular endothelial cell clusters subset from the combined dataset using monocle3 (32) with default settings.

Results

Pro- and activated MMP-2 and MMP-9 are expressed in the dura of naïve and EAE mice

We first assessed whether gelatinases are expressed and activated in the dura of the meninges. Dura samples were isolated from naïve and peak EAE WT mice and whole-mounted immunofluorescently stained with pan laminin (Pan LM) antibody to label all basement membranes, anti-CD45 to mark immune cells and anti-CD4 to identify CD4⁺ T cells (Figure 1A). In separate stainings, immunofluorescence staining for CD206 and F4/80 was performed to identify resident and total macrophages, respectively (Supplementary Figure 1). In naïve dura, the majority of CD45⁺ immune cells exhibited an elongated morphology typical of resident macrophages, consistent with the high number and morphology of CD206⁺ resident macrophages (Supplementary Figures 1A, B), and with virtually no CD4⁺ T cells present. By contrast, at peak EAE CD4⁺ T cells were detectable throughout the dura but remained low in number compared with total CD45⁺ (Figure 1A) and CD206⁺ cells (Supplementary Figure 1C). Gelatin zymography of dura extracts confirmed the expression of pro-forms of MMP-2 and MMP-9 even in non-diseased tissue (Figure 1B). During EAE, both pro- and activated forms of MMP-9, and to a lesser extent MMP-2, were more abundant (Figure 1B), as previously reported for EAE brain tissues (9, 33). Thus, while both MMP-2 and MMP-9 are active in the dura under homeostatic conditions, activated MMP-9 is preferentially increased in the presence of abundant immune cell infiltration during EAE (onset and peak). The expression and activation pattern of gelatinases in the inflamed dura, thus, is similar to that of postcapillary venules at the border to the brain parenchyma (8, 9, 33). Given the selective upregulation of pro- and activated-MMP-9 in EAE dura, we focused subsequent analyses on Mmp9^-/- dura to investigate its cellular origin/s and potential role/s in meningeal neuroinflammation.

We first analyzed the EAE disease course in Mmp9^-/- mice, revealing delayed onset of symptoms compared to WT controls (Figure 1C), implicating MMP-9 in early disease processes. However, these mice ultimately developed more severe disease at later stages. To assess potential compensatory mechanisms, we performed gelatin zymography on brain extracts from WT and Mmp9^-/- adult mice (Figure 1D), confirming the absence of MMP-9 and no compensatory upregulation or activation of MMP-2 in the Mmp9^-/- samples. These data indicate that the altered EAE kinetics in Mmp9^-/- mice are not due to increased MMP-2 activity.

SnRNA-seq reveals cellular heterogeneity of murine dura

We next tested which cells in the dura express the gelatinases at the transcriptional level. As MMP-9 was specifically upregulated in EAE, snRNA-seq was performed on all nuclei of freshly dissected dura from WT and Mmp9^-/- mice at day 14 after EAE induction, when they exhibited similar disease scores (Figure 1C). Since Mmp9^-/- mice exhibited delayed onset of EAE symptoms compared to WT, but ultimately developed more severe disease, analysis at EAE day 14 captured peak disease in WT mice, while representing a potentially earlier disease phase (onset period) in Mmp9^-/- mice. Only WT and Mmp9^-/- mice with similar disease scores were analysed by snRNA-seq. We obtained single-nuclei transcriptomes (henceforth referred to as cells) of dural cells from EAE WT and EAE Mmp9^-/- mice and integrated a naïve WT murine dural dataset as control. Cell clusters transcriptionally identified as pinealocytes (34), contaminating astrocytes (35), choroid plexus epithelial cells (36), and low-quality cells were excluded (Supplementary Table 1) (see Methods). We thereby retained and analytically integrated a total of 20,804 high quality single-nuclei transcriptomes from naïve WT (n = 3,096 nuclei), EAE WT (n = 4,208 nuclei) and EAE Mmp9^-/- (n = 8,236 nuclei) conditions.

After unsupervised clustering using all the cells, we identified 21 transcriptionally distinct cell clusters that were present across all conditions (Figure 2A, Supplementary Figure 2A). Among non-immune populations, two clusters corresponded to dural fibroblasts (DuFibro_1: Col1a1, Runx2; DuFibro_2: Scara5, Epha3), and two border fibroblasts (BordFibro_1 and BordFibro_2; Slc4a10, Slc47a1), as well as a cluster of arachnoid fibroblasts (arachFibro; Alcam, Aldh1a2, Ptgds) sharing transcriptional signatures with previously described meningeal fibroblast-like cells (12) (Supplementary Figure 2B). Within BordFibro_1, we detected a subpopulation expressing arachnoid barrier cell markers (Cdh1, Cldn11, Klf5, Dpp4), consistent with an arachnoid barrier phenotype (12, 37). One cluster transcriptionally resembled osteoblasts (Ibsp, Bglap, Sparc) (12). Endothelial cell populations were represented by five clusters expressing Pecam1 (Supplementary Figure 2B). One cell cluster with Flt4 and Prox1 expression was annotated as lymphatic endothelial cell (LymEndo) (11), while the remaining clusters were assigned to vascular endothelial phenotypes along an arterial-to-venous transcriptional continuum: arterial (art_Endo; Arl15, Cdh13, Efnb2), arterial-capillary (art_capEndo; Flt1, Hmcn1, Insr), venous-capillary (vein_capEndo), and venous (vein_Endo; Il1r1, Lcn2, Pla2g4a) (Supplementary Figure 2C, D), as described previously (38). Notably, Slco2a1, and F8 were enriched in vein_capEndo, consistent with dural sinus endothelial identity, whereas vein_Endo expressed leptomeningeal endothelial cell markers (Tjp1, Bsg, Slc2a1) (39) (Supplementary Figure 2E).

Figure 2

Figure 2. Identification of Mmp9 expression in naïve and EAE murine dura. (A) Merged Uniform Manifold Approximation and Projection (UMAP) plot displaying 21 cell clusters of the combined dataset from dura of naïve WT, EAE WT and EAE Mmp9^-/- mice. (B) A feature plot depicting the expression of Mmp9 in the combined dataset. (C) Feature plots depicting the expression of Mmp9 in the respective conditions. (D) A directional bar plot illustrating the (log₂) fold changes in relative abundance of distinct immune cell clusters between EAE Mmp9^-/- vs. EAE WT comparisons. (E) Volcano plot illustrating the differentially expressed genes between Granulo_1 and Granulo_2 (Granulo_2a and 2b); Log₂ fold change >0.5 and adjusted p value <0.05 considered to be differentially expressed. (F) Selected enriched GO terms in Granulo_1 vs. Granulo_2.

Additional non-immune populations included pericytes (Peri; Rgs5, Pdgfrb), smooth muscle cells (SMC; Myh11, Acta2), and myelinating (m_SC; Mpz) and non-myelinating Schwann cells (nm_SC; Cdh19). Immune cell clusters comprised myeloid cells (Myeloid; Lyz2, Mrc1), T cells (Tc; Cd3e, Il7r), two granulocyte subsets (Granulo1 and Granulo2; Hp, S100a9), mast cells (Mast_cells; Cpa3), and osteoclasts (Osteoclasts; Ctsk). This dataset, thus, reconstructs the expected cellular landscape of the murine dura under naïve and inflammatory conditions (Supplementary Figure 2B).

Identifying cellular sources of MMP-9 in murine dura

We next characterized the cellular sources of gelatinase gene expression. As expected, Mmp9 transcripts were detected in the combined dataset, while no expression was observed in Mmp9^-/- transcriptomes, confirming effective genetic ablation (Figures 2B, C; Supplementary Figure 2F). In non-diseased dura snRNA-seq data, Mmp9 expression was primarily restricted to immune cell clusters, including Granulo1, Granulo2, and osteoclasts (Figures 2B, C). This expression pattern is consistent with previous findings in other tissues (9, 18, 40, 41), indicating that Mmp9 is expressed by specific myeloid cell populations under homeostatic conditions.

Mmp9 deficiency reshapes meningeal cell composition

Gelatinases facilitate lymphocyte penetration of the parenchymal border of the BBB and play various roles, including activation or inactivation of chemokine gradients and selective cleavage of cell-cell junction molecules and receptors for BM components (8). However, the functional relevance of gelatinases in the meninges is poorly understood. We therefore assessed whether Mmp9 affects the composition of meningeal cell populations during EAE. Comparative analysis of Mmp9^-/- and WT EAE mice revealed the most pronounced alterations in the composition of dural cell populations. Clusters associated with barrier localization and function, including bordFibro1 and arachFibro, were relatively enriched in Mmp9^-/- EAE samples, whereas clusters with predicted dural fibroblast identity were reduced (Supplementary Figure 2A). This was accompanied by an increased abundance of endothelial cell clusters annotated as venous and venous capillary endothelial cells, along with a reduced representation of T cell populations (Figure 2D, Supplementary Figure 2A). Notably, changes were also detected in cells that do not express Mmp9, consistent with the secreted nature of MMPs and their varied effects on the microenvironment. Fold differences (log₂) revealed the most prominent changes between EAE Mmp9^-/- and EAE WT samples (Figure 2D) within the granulocyte compartment, characterized by an increase in the Granulo1 cluster and a concomitant reduction in the Granulo2 cluster (Figure 2D). Comparisons with WT naïve confirmed that the Granulo1 and Granulo2 undergo the most substantial alterations upon EAE induction (Supplementary Figure 3). This shift likely reflects a phenotypic reprogramming of granulocytes in the absence of MMP-9 during neuroinflammation. To better understand the identity and functional differences between these two granulocyte populations, we performed differential gene expression analysis followed by GO term enrichment (Figures 2E, F). While both clusters share a myeloid signature, Granulo1 cells were characterized by the upregulation of genes associated with regulatory and homeostatic processes, such as Prkg1, Ptprj, and Foxp1. By contrast, Granulo2 cells, which were enriched in Mmp9^-/- EAE dura, showed a transcriptional profile consistent with an activated, pro-inflammatory neutrophil phenotype rather than a distinct cell type. GO terms enriched in this cluster included defense response to bacterium, cell chemotaxis, reactive oxygen species (ROS), metabolic processes, antimicrobial humoral responses, and phagocytosis. These pathways were driven by upregulated expression of classical neutrophil activation markers such as Il1b, S100a8, S100a9, Lcn2, Cyba, and Ncf4, among others (Figures 2E, F). These results suggest that Mmp9 deficiency in EAE promotes a functional shift in dura-associated granulocytes toward a more inflammatory phenotype rather than the appearance of a different cell type.

Angiogenic and migratory phenotype induced in dura venous endothelial cells by Mmp9 deficiency

We next tested how Mmp9 deficiency affected the transcriptome (as opposed to the relative abundance) of individual cell clusters in EAE. In the comparison between EAE Mmp9^-/- and EAE WT mice, differentially expressed genes were detected across most cell clusters, with the highest number observed in the vein_Endo cluster (Up:28; Down:43) (Figures 3A, B). GO term enrichment analysis identified that upregulated genes in EAE Mmp9^-/- mice, including Adamts9, Hif1a, Lcn2, Hdac9, Rhoj, and Myh9, were associated with endothelial cell migration. Additionally, Adamts9, Hif1a, Hdac9, Rhoj, and Tcf4 were linked to angiogenesis (Figure 3C). Venous endothelial cells were, thus, among the most responsive cell types to Mmp9 deficiency during EAE. However, immunofluorescence staining for PECAM-1 in the dura of Mmp9^-/- vs. WT mice did not show significant changes in vessel density in either naïve or EAE conditions (Figures 3D, E), suggesting that potential vascular changes are subtle or confined to specific vessel subsets.

Figure 3

Figure 3. Mmp9 deficiency triggered transcriptional changes in murine dura during EAE. (A) Stacked bar plots showing the number of differentially expressed genes in individual cell clusters in EAE Mmp9^-/- vs. EAE WT comparison. (B) Volcano plots depicting differentially expressed genes in vein_Endo in EAE Mmp9^-/- vs. EAE WT. (C) Selected GO terms mediated by upregulated DEGs in EAE Mmp9^-/- vs. EAE WT. (D) Representative images of whole-mount dura immunofluorescently stained for PECAM-1, showing blood vessels. Scale bars = 100 µm. (E) PECAM-1⁺ vessel area in dura was analyzed using Imaris and normalized to naïve WT. Each dot represents an individual mouse. N = 6 – 11; data are presented as means ± SD. (F) Volcano plots depicting differentially expressed genes for Granulo1 in EAE Mmp9^-/- vs. EAE WT. (G) Volcano plots depicting differentially expressed genes for Granulo2 in EAE Mmp9^-/- vs. EAE WT.

Analysis of the Granulo1 and Granulo2 populations in WT EAE versus Mmp9^-/- EAE revealed that most differentially expressed genes were downregulated in the Mmp9^-/- Granulo1 and Granulo2 populations (Figures 3F, G). The only gene upregulated under Mmp9^-/- EAE conditions was Rora in the Granulo1 cluster, a transcription factor normally associated with T lymphocytes but which has implicated in diverse biological processes (42). However, GO term enrichment analysis did not identify any significantly enriched biological pathways in either Granulo1 or Granulo2 in either WT and Mmp9^-/- under EAE conditions.

Mature granulocytes accumulate in the dura of Mmp9^-/- mice during EAE

We next performed higher-resolution sub-clustering to characterize immune cell responses to EAE in the absence of MMP-9. The 2,736 Ptprc⁺ nuclei from the combined dataset segregated into distinct leukocyte subclusters based on known marker genes (Figure 4A, Supplementary Figure 4A). Lymphoid lineage clusters comprised T cells (Gata3, Cd3e, Cd3g) and B cells (Cd79a, Ms4a1). Among myeloid lineage clusters (43), two populations were characterized by macrophage-associated transcripts and corresponded to BAM_1 and BAM_2 (Lyz2). Both clusters exhibited a border-associated phenotype (Mrc1, Pf4, Lyve1) (18). BAM_2 (H2-Eb1/H2-Aa/H2-Ab1, Cd74, Ctss) displayed a pronounced MHC-II–mediated antigen-presentation program, whereas BAM_1 (F13a1, Slc9a9, Mctp1) exhibited a transcriptional profile linked to tissue homeostasis, including extracellular matrix stabilization and endosomal trafficking, consistent with a homeostatic, repair-oriented macrophage state (Supplementary Figure 4B). Additional clusters expressed monocyte-associated markers (Vcan, Ccr2) and three distinct granulocyte clusters were identified (Granulo_1, Granulo_2a, Granulo_2b; Lyz, Hp, S100a9). Granulo_2a cells were absent in naïve WT dura (Figure 4B) and displayed higher expression of Csf3r, Il1b, and Ccl6, indicative of mature neutrophils (44), whereas Granulo_2b displayed higher expression of Chil3, Camp, Lcn2 and Ltf indicative of bone marrow–associated neutrophils (44) (Figure 4C). GO term analysis of marker genes suggested that Granulo_2a exhibit immunoregulatory features (Supplementary Figure 4C). In Mmp9^-/- vs. WT EAE, the largest relative increase was observed in Granulo_2a (+4.83%), followed by monocytes (+4.14%), BAM_2 (+3.02%), BAM_1 (+2.29%), and Granulo_2b (+0.64%). By contrast, B cells, Granulo_1, mast cells, and T cells were reduced in relative abundance compared to WT (Figure 4D). The most prominent shift induced by Mmp9 deficiency was a redistribution from Granulo_1 to Granulo_2 populations. Additional subclusters corresponded to dendritic cells (Cd209a, Clec9a) and mast cells (Tpsb2, Cpa3).

Figure 4

Figure 4. Immune cell responses in murine dura to gelatinase deficiencies during EAE. (A) UMAP plot of immune cell clusters from dura of naïve and EAE WT (naïve WT, n = 285 nuclei; EAE WT, n = 623 nuclei) and EAE Mmp9 deficient (EAE Mmp9^-/-, n = 1114 nuclei) mice. (B) Bar plot depicting the relative abundance of individual immune cell clusters in the respective conditions. (C) Dot plot of maturity related genes in granulocyte populations. Dot size depicts percentage of cells within a cell cluster expressing the selected genes, and color defines the average expression level of the selected genes. (D) A directional bar plot illustrating the (log₂) fold changes in relative abundance of distinct immune cell clusters between EAE Mmp9^-/- vs. EAE WT comparisons. Bars oriented upwards indicating increased relative abundance in the dura of EAE Mmp9^-/- mice, whereas the downwards-oriented bars indicating increased relative abundance in the dura of EAE WT mice. (E) Representative gating strategy for identification of Ly6C⁺Ly6G⁺ neutrophils in the dura by flow cytometry. (F) Quantification of leukocyte subtypes in the dura of naïve and EAE induced WT and Mmp9^-/- mice. Each data point represents a pool of three dura samples. Statistical analysis was performed using two-way ANOVA, ***p < 0.001. (G) Representative images of whole-mount dura immunofluorescence staining for Ly-6G to mark neutrophils and PECAM-1 or isolectin B4 to mark blood vessels. Scale bars = 1 mm (in lower magnifications) and 200 µm (in higher magnifications).

To validate these transcriptomic findings, we performed flow cytometry of immune cell populations in the dura of naïve and EAE-induced Mmp9^-/- and WT mice (Figures 4E, F). No differences in immune cell composition were detected in naïve animals, indicating the absence of baseline alterations. Consistent with the snRNA-seq results, Ly6C⁺Ly6G⁺ neutrophils were more abundant in the dura of Mmp9^-/- EAE mice compared to WT EAE controls (Figure 4F). Immunofluorescence staining for Ly6G together with isolectin and PECAM-1 revealed focal Ly6G⁺ cell accumulations in the dura of both genotypes; these foci appeared more frequent and larger in Mmp9^-/- mice (Figure 4G). Collectively, these data indicate that Mmp9 deficiency during EAE promotes expansion of myeloid cell populations, particularly mature neutrophils, and alters the composition of granulocytic infiltrates within the dura.

Discussion

In this study, we identify MMP-9 as a selective and dynamically regulated gelatinase in the murine dura during neuroinflammation. We show that both MMP-2 and MMP-9 are present under homeostatic conditions, but only MMP-9 is markedly upregulated during EAE. Mmp9^-/- mice exhibited a delayed onset of clinical symptoms, but developed more severe disease at peak, a phenotype accompanied by pronounced alterations in dural immune cell composition and endothelial transcriptional states. Single-nucleus RNA-seq revealed that Mmp9 deficiency led to a redistribution of granulocyte subsets toward a mature, pro-inflammatory neutrophil phenotype, as well as transcriptional activation of venous endothelial cells toward migratory and angiogenic programs. Flow cytometry and immunohistology confirmed increased neutrophil accumulation in the dura of Mmp9^-/- mice. These findings suggest that, in the dura, MMP-9 may exert context-dependent, potentially anti-inflammatory functions, which contrast with its more barrier-modulating pro-inflammatory role at the BBB (9, 33).

Venous endothelial cells were among the most transcriptionally responsive cell types to Mmp9 deficiency in EAE, displaying upregulation of genes associated with migration (Adamts9, Hif1a, Hdac9, Rhoj, Myh9) and angiogenesis (Adamts9, Hif1a, Tcf4). Such signatures are consistent with previous reports implicating MMP-9 in vascular remodeling and angiogenic processes through ECM remodeling and growth factor release (2, 6). However, despite these transcriptional changes, we did not observe overt differences in vessel density between genotypes, suggesting that endothelial adaptations may be subtle, temporally restricted, or confined to specific vascular segments. The altered endothelial state may also reflect secondary responses to changes in immune cell composition or cytokine milieu, as Mmp9 deficiency has been linked to altered systemic cytokine and chemokine levels during EAE (45).Thus, vascular phenotypes in Mmp9^-/- dura may arise from a combination of direct MMP-9-dependent effects and indirect effects mediated by the inflammatory microenvironment.

The most striking immune alteration in Mmp9^-/- dura was a shift from Granulo1 to Granulo2 populations, with Granulo_2a cells exhibiting a mature neutrophil transcriptome (Csf3r, Il1b, Ccl6) and enriched pro-inflammatory pathways, including defense responses to bacteria, ROS metabolism, and chemotaxis. These changes were paralleled by a significant increase in Ly6C⁺Ly6G⁺ neutrophils by flow cytometry. Such findings align with earlier work showing that MMP-9 modulates neutrophil recruitment and activation, in part through proteolytic processing of chemokines such as CXCL8/IL-8, CXCL1, and CXCL2, thereby shaping chemokine gradients (1, 46, 47). MMP-9-mediated chemokine cleavage can have both activating and inactivating effects, depending on context, suggesting that its absence could prolong or intensify neutrophil-attracting signals in the dura (9, 48).

The significance of these changes may extend beyond local effects. MMP-9-expressing granulocytes are abundant in the periphery, and systemic alterations in cytokine and chemokine levels have been reported in Mmp9^-/- mice during EAE (45). This, together with the primary granulocytic source of MMP-9 in our dataset, suggests that peripherally derived MMP-9 contributes to establishing an effective and temporally regulated immune response. In support of this possibility, WT mice reconstituted with MMP-2/MMP-9 double knockout bone marrow display predominantly peripheral immune defects (9, 10). In the dura, absence of MMP-9 may thus impair resolution of neutrophil recruitment signals, leading to sustained accumulation of mature neutrophils.

The dual-phase EAE phenotype observed in Mmp9^-/- mice, delayed onset but more severe peak, may be directly related to the temporal and qualitative changes in granulocyte recruitment we observe. Early in disease, the absence of MMP-9 could delay immune cell penetration or alter chemokine gradient formation, slowing initial disease progression. However, as inflammation advances, unmodulated chemotactic cues and endothelial activation in the dura may favor excessive neutrophil accumulation, amplifying local and possibly systemic inflammation, thereby exacerbating disease severity at peak. This model reconciles the seemingly paradoxical early protective and later deleterious effects of Mmp9 deficiency, highlighting a context-dependent role for MMP-9 in neuroinflammation that differs between CNS border compartments.

Beyond MMP-2 and MMP-9, other MMP family members also contribute to CNS barrier regulation and inflammation. Notably, MMP-14 (MT1-MMP) has recently been shown to mediate cerebrospinal fluid barrier dysfunction and immune cell trafficking (49). Together, these findings highlight the broader roles of MMPs in brain pathophysiology, including inflammation, vascular remodeling, and tumor progression, and emphasize the need for context-specific understanding when targeting MMPs therapeutically.

Our findings underscore the importance of the dura as a distinct immunological niche during CNS inflammation and reveal that MMP-9, beyond its established role at the BBB, modulates the composition and activation state of dural immune and endothelial populations. The apparent anti-inflammatory function of MMP-9 in the dura raises caution for therapeutic strategies aiming to broadly inhibit MMP-9 in neuroinflammatory conditions. Future work should dissect the temporal dynamics of MMP-9 activity in distinct CNS border tissues and clarify its mechanistic interplay with chemokine networks and endothelial signaling, ideally integrating longitudinal single-cell and functional approaches in both murine and human systems.

Data availability statement

The data presented in the study are deposited in the European Nucleotide Archive (ENA), accession numbers PRJEB101900 and PRJEB102008.

Ethics statement

The animal study was approved by Landesamt für Verbraucherschutz und Ernährung Nordrhein-Westfalen. The study was conducted in accordance with the local legislation and institutional requirements.

Author contributions

S-HS: Formal Analysis, Investigation, Writing – original draft. HG: Formal Analysis, Investigation, Writing – original draft, Writing – review & editing. MB: Investigation, Methodology, Writing – review & editing. Y-LT: Formal Analysis, Investigation, Writing – review & editing. I-NL: Formal Analysis, Methodology, Writing – review & editing. AB:. Data curation, Formal analysis, Writing - review & editing. TD: Formal Analysis, Investigation, Writing – review & editing. LS: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing. GH: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. This work was supported by funding from the German Research Foundation to LS and GH (SFB/TR128/B03). LS has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No. 101054805). GH was supported by grants from the Deutsche Forschungsgemeinschaft (DFG) (ME4050/12-1, ME4050/13-1).

Acknowledgments

The authors thank Laura Schnell for help with the gel zymographies, and Devika Thulichery and Sai-Kiran Samawar for help with immunofluorescence stainings. We thank Julia Tietz for excellent technical assistance.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2025.1713166/full#supplementary-material

References

1. Opdenakker G, den Steen PEV, and Damme JV. Gelatinase B: a tuner and amplifier of immune functions. Trends Immunol. (2001) 22:571–9. doi: 10.1016/S1471-4906(01)02023-3

PubMed Abstract | Crossref Full Text | Google Scholar

2. Vandooren J, Van den Steen PE, and Opdenakker G. Biochemistry and molecular biology of gelatinase B or matrix metalloproteinase-9 (MMP-9): The next decade. Crit Rev Biochem Mol Biol. (2013) 48:222–72. doi: 10.3109/10409238.2013.770819

PubMed Abstract | Crossref Full Text | Google Scholar

3. Bates AL, Pickup MW, Hallett MA, Dozier EA, Thomas S, and Fingleton B. Stromal matrix metalloproteinase 2 regulates collagen expression and promotes the outgrowth of experimental metastases. J Pathol. (2015) 235:773–83. doi: 10.1002/path.4493

PubMed Abstract | Crossref Full Text | Google Scholar

4. Song J, Zhang X, Buscher K, Wang Y, Wang H, Di Russo J, et al. Endothelial basement membrane laminin 511 contributes to endothelial junctional tightness and thereby inhibits leukocyte transmigration. Cell Rep. (2017) 18:1256–69. doi: 10.1016/j.celrep.2016.12.092

PubMed Abstract | Crossref Full Text | Google Scholar

5. Nagase H and Woessner JF. Matrix metalloproteinases*. J Biol Chem. (1999) 274:21491–4. doi: 10.1074/jbc.274.31.21491

PubMed Abstract | Crossref Full Text | Google Scholar

6. Rodríguez D, Morrison CJ, and Overall CM. Matrix metalloproteinases: What do they not do? New substrates and biological roles identified by murine models and proteomics. Biochim Biophys Acta (BBA) - Mol Cell Res. (2010) 1803:39–54. doi: 10.1016/j.bbamcr.2009.09.015

PubMed Abstract | Crossref Full Text | Google Scholar

7. Hannocks M-J, Zhang X, Gerwien H, ChashChina A, Burmeister M, Korpos E, et al. The gelatinases, MMP-2 and MMP-9, as fine tuners of neuroinflammatory processes. Matrix Biol. (2019) 75–76:102–13. doi: 10.1016/j.matbio.2017.11.007

PubMed Abstract | Crossref Full Text | Google Scholar

8. Burmeister M, Fraunenstein A, Kahms M, Arends L, Gerwien H, Deshpande T, et al. Secretomics reveals gelatinase substrates at the blood-brain barrier that are implicated in astroglial barrier function. Sci Adv. (2023) 9:eadg0686. doi: 10.1126/sciadv.adg0686

PubMed Abstract | Crossref Full Text | Google Scholar

9. Song J, Wu C, Korpos E, Zhang X, Agrawal SM, Wang Y, et al. Focal MMP-2 and MMP-9 activity at the blood-brain barrier promotes chemokine-induced leukocyte migration. Cell Rep. (2015) 10:1040–54. doi: 10.1016/j.celrep.2015.01.037

PubMed Abstract | Crossref Full Text | Google Scholar

10. Gerwien H, Hermann S, Zhang X, Korpos E, Song J, Kopka K, et al. Imaging matrix metalloproteinase activity in multiple sclerosis as a specific marker of leukocyte penetration of the blood-brain barrier. Sci Transl Med. (2016) 8:364ra152. doi: 10.1126/scitranslmed.aaf8020

PubMed Abstract | Crossref Full Text | Google Scholar

11. Rustenhoven J and Kipnis J. Brain borders at the central stage of neuroimmunology. Nature. (2022) 612:417–29. doi: 10.1038/s41586-022-05474-7

PubMed Abstract | Crossref Full Text | Google Scholar

12. Pietilä R, Del Gaudio F, He L, Vázquez-Liébanas E, Vanlandewijck M, Muhl L, et al. Molecular anatomy of adult mouse leptomeninges. Neuron. (2023) 111:3745–3764.e7. doi: 10.1016/j.neuron.2023.09.002

PubMed Abstract | Crossref Full Text | Google Scholar

13. Betsholtz C, Engelhardt B, Koh GY, McDonald DM, Proulx ST, and Siegenthaler J. Advances and controversies in meningeal biology. Nat Neurosci. (2024) 27:2056–72. doi: 10.1038/s41593-024-01701-8

PubMed Abstract | Crossref Full Text | Google Scholar

14. Aspelund A, Antila S, Proulx ST, Karlsen TV, Karaman S, Detmar M, et al. A dural lymphatic vascular system that drains brain interstitial fluid and macromolecules. J Exp Med. (2015) 212:991–9. doi: 10.1084/jem.20142290

PubMed Abstract | Crossref Full Text | Google Scholar

15. Louveau A, Herz J, Alme MN, Salvador AF, Dong MQ, Viar KE, et al. CNS lymphatic drainage and neuroinflammation are regulated by meningeal lymphatic vasculature. Nat Neurosci. (2018) 21:1380–91. doi: 10.1038/s41593-018-0227-9

PubMed Abstract | Crossref Full Text | Google Scholar

16. Fitzpatrick Z, Ghabdan Zanluqui N, Rosenblum JS, Tuong ZK, Lee CYC, Chandrashekhar V, et al. Venous-plexus-associated lymphoid hubs support meningeal humoral immunity. Nature. (2024) 628:612–9. doi: 10.1038/s41586-024-07202-9

PubMed Abstract | Crossref Full Text | Google Scholar

17. Brioschi S, Wang W, Peng V, Wang M, Shchukina I, Greenberg ZJ, et al. Heterogeneity of meningeal B cells reveals a lymphopoietic niche at the CNS borders. Science. (2021) 373:eabf9277. doi: 10.1126/science.abf9277

PubMed Abstract | Crossref Full Text | Google Scholar

18. Schafflick D, Wolbert J, Heming M, Thomas C, Hartlehnert M, Börsch A-L, et al. Single-cell profiling of CNS border compartment leukocytes reveals that B cells and their progenitors reside in non-diseased meninges. Nat Neurosci. (2021) 24:1225–34. doi: 10.1038/s41593-021-00880-y

PubMed Abstract | Crossref Full Text | Google Scholar

19. Wang Y, Chen D, Xu D, Huang C, Xing R, He D, et al. Early developing B cells undergo negative selection by central nervous system-specific antigens in the meninges. Immunity. (2021) 54:2784–2794.e6. doi: 10.1016/j.immuni.2021.09.016

PubMed Abstract | Crossref Full Text | Google Scholar

20. Hartlehnert M, Börsch A-L, Li X, Burmeister M, Gerwien H, Schafflick D, et al. Bcl6 controls meningeal Th17–B cell interaction in murine neuroinflammation. Proc Natl Acad Sci. (2021) 118:e2023174118. doi: 10.1073/pnas.2023174118

PubMed Abstract | Crossref Full Text | Google Scholar

21. Masure S, Proost P, Van Damme J, and Opdenakker G. Purification and identification of 91-kDa neutrophil gelatinase. Eur J Biochem. (1991) 198:391–8. doi: 10.1111/j.1432-1033.1991.tb16027.x

PubMed Abstract | Crossref Full Text | Google Scholar

22. Sorokin LM, Pausch F, Frieser M, Kröger S, Ohage E, and Deutzmann R. Developmental regulation of the laminin α5 chain suggests a role in epithelial and endothelial cell maturation. Dev Biol. (1997) 189:285–300. doi: 10.1006/dbio.1997.8668

PubMed Abstract | Crossref Full Text | Google Scholar

23. Pukrop T, Dehghani F, Chuang H-N, Lohaus R, Bayanga K, Heermann S, et al. Microglia promote colonization of brain tissue by breast cancer cells in a Wnt-dependent way. Glia. (2010) 58:1477–89. doi: 10.1002/glia.21022

PubMed Abstract | Crossref Full Text | Google Scholar

24. Daley JM, Thomay AA, Connolly MD, Reichner JS, and Albina JE. Use of Ly6G-specific monoclonal antibody to deplete neutrophils in mice. J Leukocyte Biol. (2008) 83:64–70. doi: 10.1189/jlb.0407247

PubMed Abstract | Crossref Full Text | Google Scholar

25. Pasciuto E, Burton OT, Roca CP, Lagou V, Rajan WD, Theys T, et al. Microglia require CD4 T cells to complete the fetal-to-adult transition. Cell. (2020) 182:625–640.e24. doi: 10.1016/j.cell.2020.06.026

PubMed Abstract | Crossref Full Text | Google Scholar

26. Fleming SJ, Chaffin MD, Arduini A, Akkad A-D, Banks E, Marioni JC, et al. Unsupervised removal of systematic background noise from droplet-based single-cell experiments using CellBender. Nat Methods. (2023) 20:1323–35. doi: 10.1038/s41592-023-01943-7

PubMed Abstract | Crossref Full Text | Google Scholar

27. Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. (2024) 42:293–304. doi: 10.1038/s41587-023-01767-y

PubMed Abstract | Crossref Full Text | Google Scholar

28. Hafemeister C and Satija R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. Genome Biol. (2019) 20:296. doi: 10.1186/s13059-019-1874-1

PubMed Abstract | Crossref Full Text | Google Scholar

29. Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. (2019) 16:1289–96. doi: 10.1038/s41592-019-0619-0

PubMed Abstract | Crossref Full Text | Google Scholar

30. Wickham H. ggplot2: elegant graphics for data analysis. New York: Springer-Verlag (2016).

Google Scholar

31. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb). (2021) 2:100141. doi: 10.1016/j.xinn.2021.100141

PubMed Abstract | Crossref Full Text | Google Scholar

32. Cao J, Spielmann M, Qiu X, Huang X, Ibrahim DM, Hill AJ, et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature. (2019) 566:496–502. doi: 10.1038/s41586-019-0969-x

PubMed Abstract | Crossref Full Text | Google Scholar

33. Agrawal S, Anderson P, Durbeej M, van Rooijen N, Ivars F, Opdenakker G, et al. Dystroglycan is selectively cleaved at the parenchymal basement membrane at sites of leukocyte extravasation in experimental autoimmune encephalomyelitis. J Exp Med. (2006) 203:1007–19. doi: 10.1084/jem.20051342

PubMed Abstract | Crossref Full Text | Google Scholar

34. Mays JC, Kelly MC, Coon SL, Holtzclaw L, Rath MF, Kelley MW, et al. Single-cell RNA sequencing of the mammalian pineal gland identifies two pinealocyte subtypes and cell type-specific daily patterns of gene expression. PloS One. (2018) 13:e0205883. doi: 10.1371/journal.pone.0205883

PubMed Abstract | Crossref Full Text | Google Scholar

35. Batiuk MY, Martirosyan A, Wahis J, de Vin F, Marneffe C, Kusserow C, et al. Identification of region-specific astrocyte subtypes at single cell resolution. Nat Commun. (2020) 11:1220. doi: 10.1038/s41467-019-14198-8

PubMed Abstract | Crossref Full Text | Google Scholar

36. Dani N, Herbst RH, McCabe C, Green GS, Kaiser K, Head JP, et al. A cellular and spatial map of the choroid plexus across brain ventricles and ages. Cell. (2021) 184:3056–3074.e21. doi: 10.1016/j.cell.2021.04.003

PubMed Abstract | Crossref Full Text | Google Scholar

37. Smyth LCD, Xu D, Okar SV, Dykstra T, Rustenhoven J, Papadopoulos Z, et al. Identification of direct connections between the dura and the brain. Nature. (2024) 627:165–73. doi: 10.1038/s41586-023-06993-7

PubMed Abstract | Crossref Full Text | Google Scholar

38. Kalucka J, de Rooij LPMH, Goveia J, Rohlenova K, Dumas SJ, Meta E, et al. Single-cell transcriptome atlas of murine endothelial cells. Cell. (2020) 180:764–779.e20. doi: 10.1016/j.cell.2020.01.015

PubMed Abstract | Crossref Full Text | Google Scholar

39. Kim Y-C, Ahn JH, Jin H, Yang MJ, Hong SP, Yoon J-H, et al. Immaturity of immune cells around the dural venous sinuses contributes to viral meningoencephalitis in neonates. Sci Immunol. (2023) 8:eadg6155. doi: 10.1126/sciimmunol.adg6155

PubMed Abstract | Crossref Full Text | Google Scholar

40. Baryawno N, Przybylski D, Kowalczyk MS, Kfoury Y, Severe N, Gustafsson K, et al. A cellular taxonomy of the bone marrow stroma in homeostasis and leukemia. Cell. (2019) 177:1915–1932.e16. doi: 10.1016/j.cell.2019.04.040

PubMed Abstract | Crossref Full Text | Google Scholar

41. Tsukasaki M, Huynh NC-N, Okamoto K, Muro R, Terashima A, Kurikawa Y, et al. Stepwise cell fate decision pathways during osteoclastogenesis at single-cell resolution. Nat Metab. (2020) 2:1382–90. doi: 10.1038/s42255-020-00318-y

PubMed Abstract | Crossref Full Text | Google Scholar

42. Hsu AY, Wang T, Syahirah R, Liu S, Li K, Zhang W, et al. Rora regulates neutrophil migration and activation in zebrafish. Front Immunol. (2022) 13:756034. doi: 10.3389/fimmu.2022.756034

PubMed Abstract | Crossref Full Text | Google Scholar

43. Jordão MJC, Sankowski R, Brendecke SM, Locatelli G, Tai Y-H, TL T, et al. Single-cell profiling identifies myeloid cell subsets with distinct fates during neuroinflammation. Science. (2019) 363:eaat7554. doi: 10.1126/science.aat7554

PubMed Abstract | Crossref Full Text | Google Scholar

44. Grieshaber-Bouyer R, Radtke FA, Cunin P, Stifano G, Levescot A, Vijaykumar B, et al. The neutrotime transcriptional signature defines a single continuum of neutrophils across biological compartments. Nat Commun. (2021) 12:2856. doi: 10.1038/s41467-021-22973-9

PubMed Abstract | Crossref Full Text | Google Scholar

45. Ugarte-Berzal E, Berghmans N, Boon L, Martens E, Vandooren J, Cauwe B, et al. Gelatinase B/matrix metalloproteinase-9 is a phase-specific effector molecule, independent from Fas, in experimental autoimmune encephalomyelitis. PloS One. (2018) 13:e0197944. doi: 10.1371/journal.pone.0197944

PubMed Abstract | Crossref Full Text | Google Scholar

46. Van den Steen PE, Van Aelst I, Hvidberg V, Piccard H, Fiten P, Jacobsen C, et al. The hemopexin and O-glycosylated domains tune gelatinase B/MMP-9 bioavailability via inhibition and binding to cargo receptors. J Biol Chem. (2006) 281:18626–37. doi: 10.1074/jbc.M512308200

PubMed Abstract | Crossref Full Text | Google Scholar

47. Dufour A, Bellac CL, Eckhard U, Solis N, Klein T, Kappelhoff R, et al. C-terminal truncation of IFN-γ inhibits proinflammatory macrophage responses and is deficient in autoimmune disease. Nat Commun. (2018) 9:2416. doi: 10.1038/s41467-018-04717-4

PubMed Abstract | Crossref Full Text | Google Scholar

48. Van Den Steen PE, Wuyts A, Husson SJ, Proost P, Van Damme J, and Opdenakker G. Gelatinase B/MMP-9 and neutrophil collagenase/MMP-8 process the chemokines human GCP-2/CXCL6, ENA-78/CXCL5 and mouse GCP-2/LIX and modulate their physiological activities. Eur J Biochem. (2003) 270:3739–49. doi: 10.1046/j.1432-1033.2003.03760.x

PubMed Abstract | Crossref Full Text | Google Scholar

49. Zhao L, Jin S, Wang S, Zhang Z, Wang X, Chen Z, et al. Tertiary lymphoid structures in diseases: immune mechanisms and therapeutic advances. Signal Transduct Target Ther. (2024) 9:225. doi: 10.1038/s41392-024-01947-5

PubMed Abstract | Crossref Full Text | Google Scholar