Alternatively activated neutrophils limit T cell–driven neuroinflammation

Abstract

Objective

Non-canonical myeloid cell populations are increasingly recognized as critical regulators of inflammation in neuroimmunological disease. Here, we investigate the role of alternatively activated neutrophils (aaN) in limiting encephalitogenic T cell responses during experimental autoimmune encephalomyelitis (EAE), a widely used model of multiple sclerosis.

Methods

Arginase-1–expressing aaN were identified and characterized within central nervous system (CNS) infiltrates during EAE using flow cytometry, single-cell RNA sequencing, and fluorescent in situ hybridization (RNAscope) combined with immunohistochemistry. The immunomodulatory properties of aaN were evaluated in vitro using CD4⁺ T cell suppression assays and in vivo by adoptive transfer of ex vivo–generated aaN during the preclinical phase following encephalitogenic T-cell injection.

Results

aaN were consistently detected within the CNS throughout EAE and spatially co-localized with encephalitogenic T cells. Transcriptomic profiling of aaN revealed enrichment of pathways associated with regulation of T cell activation and immune suppression. CNS-derived aaN potently inhibited CD4⁺ T cell proliferation in vitro. Therapeutic augmentation of this population, via adoptive transfer of ex vivo–generated aaN into mice following the injection of encephalitogenic T cells, delayed clinical EAE onset, markedly reduced the accumulation of pathogenic T cells within CNS lesions, and significantly enhanced neuronal survival. Mechanistically, ex vivo–generated aaN suppressed T cell responses through a contact-dependent, PD-L1–independent pathway, indicating a previously unrecognized mode of neutrophil-mediated immunoregulation.

Interpretation

These findings identify aaN as a previously underappreciated immunoregulatory population within the inflamed CNS that restrains pathogenic T cell responses and limits neuroinflammation during EAE. Collectively, our data support the therapeutic potential of strategies that augment aaN activity, including autologous aaN-based cell therapy or interventions that promote CNS homing, polarization, and persistence of endogenous aaN, as novel approaches for disease modification in multiple sclerosis.

Background

Multiple sclerosis (MS) is a multifocal, inflammatory demyelinating disorder of the central nervous system (CNS) and the leading cause of non-traumatic neurological disability among young adults in the Western Hemisphere [1]. Experimental autoimmune encephalomyelitis (EAE), a widely used murine model of MS, can be induced by adoptive transfer of myelin-specific, Th17-polarized CD4⁺ T cells into naïve recipients [2, 3]. Upon reactivation within the CNS, these encephalitogenic T cells recruit peripheral myeloid cells —particularly neutrophils and monocytes —that contribute to demyelination and axonal injury. Neutrophils have been implicated in blood–brain barrier disruption and amplification of neuroinflammation during early EAE lesion development [4–8], while monocyte-derived macrophages and dendritic cells exacerbate pathology via release of pro-inflammatory cytokines, reactive oxygen species, and matrix-degrading enzymes, as well as antigen presentation [9, 10].

Traditionally, monocyte derived cells and neutrophils have been portrayed as uniformly pro-inflammatory effector cells with limited phenotypic variability. However, advances in single-cell and spatial transcriptomic profiling reveal considerable heterogeneity and plasticity within these myeloid lineages [11–13]. For example, subsets of monocytes/macrophages, as well as macrophage-like cells such as microglia, can undergo alternative activation in response to specific environmental cues, acquiring an arginase-1 (Arg1)/mannose receptor (Mrc1/CD206) expressing phenotype associated with immunoregulatory and reparative properties [14]. There is growing evidence that alternatively activated macrophages and microglia can suppress destructive neuroinflammation and limit tissue damage in the nervous system, thereby promoting functional recovery [15–17]. These insights have stimulated efforts to develop therapeutic strategies that enhance the recruitment or induction of reparative myeloid populations, as well as ex vivo–engineered autologous myeloid cell therapies [18].

To date, most studies of alternatively activated myeloid cells in in EAE, MS, or spinal cord injury have focused on monocyte-derived cells and microglia [15, 19, 20]. In contrast, far less is known about the potential for neutrophils—typically regarded as short-lived and cytotoxic—to adopt immunoregulatory phenotypes or actively promote CNS repair. Recent evidence indicates that, under specific conditions, neutrophil lineage cells can also adopt non-canonical states [21–26]. For example, the granulocytic myeloid derived suppressor cells (G-MDSCs) and so-called “N2” neutrophils that accumulate in tumor microenvironments actively inhibit T cell responses [22, 27, 28]. In addition, we previously identified a distinct population of immature, alternatively activated neutrophils (aaN) that promote neuronal survival and axon regeneration following optic nerve and spinal cord injury, mediated in part through secretion of neurotrophic factors and modulation of the lesion microenvironment [29, 30]. Whether similar populations arise spontaneously during autoimmune demyelination and influence disease course is unknown.

Here, we report that Arg1-expressing aaN spontaneously emerge within EAE inflammatory lesions where they co-localize with pathogenic T cells and display potent immunosuppressive properties. Augmenting this population via adoptive transfer of ex vivo–generated aaN delayed clinical EAE onset, reduced the accumulation of encephalitogenic T cells within CNS infiltrates and enhanced neuron survival. Mechanistically, aaN suppressed myelin-reactive T cell proliferation in a contact-dependent manner, revealing a previously unrecognized immunomodulatory role for neutrophils in CNS autoimmunity.

Methods

Mice

Wild-type C57Bl/6 (strain #027) and congenic Ly5.1 (strain #564) mice were obtained from Charles River Laboratories (Wilmington, MA) at 8–12 weeks of age. Arg1^YFP reporter mice (strain #015857) and 2D2 T cell receptor (TCR) transgenic mice (strain #006912) were obtained from Jackson Laboratory (Bar Harbor, ME). All mice were housed under specific pathogen-free conditions in microisolator cages.

Induction and assessment of EAE

EAE was induced as previously described [31]. Briefly, donor mice were immunized with MOG_35–55 peptide (100 μg; MEVGWYRSP-FSRVVHLYRNGK; Biosynthesis) emulsified in CFA (Sigma, F5881), at four sites over the flanks. Inguinal, axial, and brachial lymph nodes and spleens were harvested 10–14 days later, homogenized, and passed through a 70 μm strainer to achieve a single-cell suspension. The dissociated cells were cultured with MOG_35–55 peptide (50 μg/mL; Biosynthesis) under Th17-polarizing conditions (8 ng/mL recombinant murine IL-23 (R&D), 10 ng/mL recombinant murine IL-1α (Peprotech), and 10 μg/mL anti–IFNγ monoclonal antibody (Bio X Cell)), in RPMI (Gibco, 21,870,084) supplemented with HEPES buffer (Gibco, 15,630,080), MEM non-essential amino acids (Gibco, 11,140,076), sodium pyruvate (Gibco, 1,360,070), β-mercaptoethanol (Gibco, 21,985,023), L-glutamine (Gibco, 25,030,081), Pen Strep (Gibco, 15,140,122), and 10% FBS (Gemini Bio-products, 100–106)). After 96 h of culture, CD4⁺ T cells were purified using L3T4 magnetic bead sorting (Miltenyi Biotec) and injected into naïve syngeneic recipients (5.0 × 10⁶ CD4⁺ T cells per mouse i.p.). EAE mice were intravenously injected on days 4 and 6 post- CD4⁺ T cell transfer with either purified bone-marrow neutrophils (5.0 × 10⁶ cells/injection), zymosan-modulated intraperitoneal neutrophils (5.0 × 10⁶ cells/injection), or PBS alone. Recipient mice were observed daily for signs of EAE by an examiner blinded to the experimental groups. Mice were scored using a 5-point scale, as previously reported [32]: 0 = no abnormality, 0.5 = distal limp tail, 1 = complete limp tail, 1.5 = difficulty righting from supine position, 2 = inability to right from a supine position, 2.5 = overt gait abnormality, 3 = difficulty elevating body while walking, 3.5 = unilateral hind limb paralysis, 4 = bilateral complete hind limb paralysis, 4.5 = moribund, 5 = death.

Flow cytometry

Cells were washed and resuspended in FACS buffer containing a fixable viability dye (eFluor506 or eFluor780; eBioscience), and Fc Block (anti-CD16/32, clone 2.4G2; 100 ng/mL), and incubated at 4 °C for 15 min. Cells were then washed with FACS buffer and incubated with fluorochrome-labeled monoclonal antibodies specific for individual surface markers at 4 °C for 25 min. Following a third wash, cells were fixed with 4% paraformaldehyde (PFA). For intracellular staining, the BD Biosciences’ Cytofix/Cytoperm™ Fixation/Permeabilization Kit was used according to the manufacturer’s instructions. Data were acquired on a FACS Symphony flow cytometer (BD Biosciences) or FACSMelody flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star).

The following flow cytometry antibodies were obtained from BD Biosciences: CD4 (GK1.5; BUV395), CD11b (M1/70; BV786), CD11c (HL3; BUV737), CD14 (rmC5-3; BV650), CD45 (30-F11; BUV805), CD62L (MEL-14; APC), CD84 (mCD84.7rMAb; BV711), CD279 (RMP1-30; BUV737), F4/80 (T45-2342; PE-CF594), and Ly6G (1A8; BUV563). The following antibodies were obtained from eBioscience: CD44 (IM7, APC) and Ki67 (SolA15; PE-Cy7). The following antibodies were obtained from Invitrogen: CD4 (RM4-5; APC), CD11b (M1/70; APC-Cy7), CD44 (IM7; PE-Cy7), CD45 (30-F11, eFluor 450), CD45.1 (A20; FITC), CD101 (Moushi101; PE-Cy7), and CD206 (MR6F3; Alexa Fluor 700). Arginase-1 (PE) was obtained from R&D systems.

RNAscope and immunohistochemistry

Spinal cords were harvested from mice following transcardial perfusion with 10 mL of 1 × PBS, followed by 4% paraformaldehyde (PFA) Thoracoumbar spinal cord segments, where inflammatory lesions are most prominent in EAE, were dissected for analysis. Tissues were post-fixed overnight in 4% PFA, washed in 1 × PBS, and cryoprotected in 30% sucrose at 4 °C. Tissues were then embedded in optimal cutting temperature (O.C.T.) compound (Fisher Scientific, Cat. No. 23–730-571), rapidly frozen on dry ice, and stored at − 80 °C until cryosectioning at − 20 °C on a Leica CM1950 cryostat. Sections (10–12 µm) were collected on Superfrost Plus slides for subsequent RNAscope and immunofluorescence analyses.

RNAscope Multiplex Fluorescent assays were performed according to the manufacturer’s protocol with minor modifications. Cryosections were warmed at 60 °C for 15 min, post-fixed in 4% PFA at room temperature for 5 min followed by antigen retrieval at 99 °C for 10 min. Sections were washed twice in distilled water, and dehydrated through graded ethanol for 5 min each. Slides were incubated with hydrogen peroxide at room temperature for 10 min, followed by dehydration and air-drying. After ethanol removal, sections were air-dried for 5 min and loaded into ACD slide holders. Protease III was applied, and sections were incubated at 40 °C for 30 min. Following washes with distilled water, sections were hybridized with the Cxcr2 probe for 2 h at 40 °C. Signal amplification was performed through three amplification steps at 40 °C, and fluorescent signals were developed using TSA Vivid dyes according to the manufacturer’s protocol.

For immunofluorescence co-detection, sections were incubated overnight at room temperature with an antibody cocktail prepared in ACD co-detection antibody diluent (Cat. No. 323160) containing αArg1 (1:100; R&D Systems, Cat. No. IC5868F), αCD3ε (1:100; Novus Biologicals, Cat. No. NB600-1441), and αLy6G (1:200; eBioscience, Cat. No. 16–9668-82). Corresponding secondary antibodies were applied the following day and the slides were incubated for 2 h at room temperature. Nuclei were counterstained with DAPI, and tissue autofluorescence was quenched using 0.5 × TrueBlack solution in 70% ethanol for 10 min. Coverslips were mounted using Fluoromount-G mounting medium.

Negative controls included RNAscope negative control probes (ACD) and sections processed without primary antibodies to assess nonspecific secondary antibody staining. In addition, fluorescence minus one (FMO) controls were used when appropriate to establish signal thresholds. Spinal cords from n = 6 mice were analyzed (biological replicates), and 1–2 sections per mouse were examined (technical replicates). Imaging and quantification were performed by investigators blinded to experimental group identity.

For retinal whole-mount staining and retinal ganglion cell (RGC) quantification, mice were transcardially perfused with 1 × PBS, eyes were enucleated, and tissues were fixed in PFA for 2 h. Retinas were dissected under a stereomicroscope, permeabilized with 3% Triton X-100, blocked with 5% normal donkey serum, and incubated with rabbit anti-mouse Brn3a antibody (Synaptic Systems, 411 003) at 4 °C for 72 h, followed by Alexa Fluor 594–conjugated donkey anti-rabbit secondary antibody (Invitrogen, A-21207) overnight. Whole mounts were washed, mounted on Superfrost Plus slides, and imaged using an Olympus IX71 inverted microscope with an Olympus DP72 camera. Brn3a⁺ RGCs were quantified using Imaris software across eight fields per retina, sampled from all quadrants at predefined distances from the optic nerve head.

Spinning disk confocal imaging

Tissue sections were imaged using a Yokogawa SoRa-W1 spinning disk confocal unit mounted on a Nikon Ti2 inverted microscope controlled with NIS-Elements. Images were acquired using a 60 × 1.4 NA Plan Apo oil-immersion objective and a Hamamatsu ORCA-Fusion BT sCMOS camera, yielding a pixel size of 108 nm and a z-step size of 150 nm. Laser lines at 405, 488, 577, 639, and 748 nm were used to sequentially excite the respective probes, with matched bandpass emission filters. Images shown in Fig. 2 were further processed using the 3D blind deconvolution algorithm with 25 iterations and low-noise parameters in the NIS Elements.

Fig. 2.

Alternatively activated neutrophils localize to inflammatory spinal cord lesions and are positioned adjacent to T cells. Confocal maximum-intensity projections (MIPs) of spinal cord sections obtained from 2 representative mice at clinical EAE onset. Sections were immunofluorescently labeled for Arg1 (green), Ly6G (magenta), CD3ε (orange), and DAPI (blue), and probed for Cxcr2 mRNA expression by RNAscope (white). Orange boxes in the schematics indicate the anatomical locations of the imaged regions; white boxes denote regions from which single optical sections are shown below. Single optical sections from the boxed regions are shown as merged images and corresponding single channels (DAPI, ARG1, CD3ε, Ly6G, and Cxcr2). Dashed outlines indicate alternatively activated neutrophils (aaN; CD3ε⁻Ly6G⁺/⁻Cxcr2⁺Arg1⁺; green), classically activated neutrophils (N; CD3ε⁻Ly6G⁺Cxcr2⁺Arg1⁻; magenta), and T cells (CD3ε⁺Ly6G⁻Cxcr2⁻; orange). Images were acquired at 40 × original magnification. A Single optical section from the MIP on the left side showing two Arg1⁺Ly6G⁺ aaN in close proximity to a CD3ε⁺ T cell; an Arg1⁻ conventional neutrophil (N) is also present. B Single optical section from the MIP on the right side highlighting an Arg1^high Ly6G^low Cxcr2⁺ aaN. Images are representative of 6 mice from 2 independent experiments; 9 sections were analyzed in total

Single-cell RNA-Seq analysis

Single-cell libraries were generated using the10X Chromium Next GEM Single Cell 5’ kit (PN-1000263). Approximately 18,000 live cells (targeting 10,000 cells) were loaded onto Chromium chip to create thousands of gel bead-in-emulsion (GEMs). Single-cell RNA sequencing data were analyzed using Seurat v4.4.0. Feature barcode matrices, generated by Cell Ranger from four blood and four spinal cord samples, were imported into a Seurat object (min.cells = 3). Quality control filtering was applied to remove low-quality cells using the following criteria: mitochondrial gene percentage < 10%, ribosomal gene percentage < 40%, nFeature_RNA > 200 and < 4000, and nCount_RNA < 10,000.

Datasets were normalized and integrated using the SCTransform workflow with glmGamPoi to account for sample-to-sample variation, followed by dimensional reduction using principal component analysis (PCA) with the top 20 principal components. Cells were clustered using graph-based clustering at a resolution of 0.1 and visualized using UMAP. Clusters were initially generated by unsupervised clustering. Neutrophil populations were subsequently identified based on expression of canonical granulocytic lineage markers, including Csf3r, Ly6g, and Cxcr2. These markers were selectively enriched in a subset of clusters within the master object and were not broadly expressed across other myeloid populations. Clusters enriched for these markers (clusters 0, 6, 8, and 12; Fig. S1) were annotated as neutrophils and extracted for downstream analysis. The neutrophil subset was then reclustered using 20 principal components at a resolution of 0.5, yielding several groups that were consolidated into three major neutrophil subclusters.

Cluster marker genes were identified using the Wilcoxon rank-sum test implemented in Seurat’s FindAllMarkers function. Marker genes were defined as those showing higher expression in one cluster relative to the others and meeting the criteria of adjusted p < 0.05 and log₂ fold change > 0.25. The genes Mpo and Mrc1 were also included in heatmap visualizations based on their biological relevance. Differential gene expression between spinal cord and blood neutrophils was determined using the FindMarkers function with a Wilcoxon test, with significance defined as adjusted p < 0.05 after multiple testing correction. Gene ontology enrichment analysis of cluster marker genes was performed using clusterProfiler.

Visualization of gene expression patterns was performed using ComplexHeatmap. Heatmaps display scaled expression values, with scaling performed across all tissues and clusters; however, only spinal cord and blood cells from Cluster 1 were shown in the final plots. Violin plots were used to display the distribution of Arg1 expression across individual cell barcodes. Dimplot visualizations split by tissue were generated using merged UMAP cluster assignments for color shading. Pie charts were used to visualize the relative distribution of cells from different disease stages within neutrophil subclusters. Because these plots summarize pooled single-cell counts from the integrated dataset, they are presented for descriptive purposes and were not subjected to statistical testing.

Intraperitoneal (i.p.) lavage and neutrophil isolation

Naïve C57BL/6 adult mice were injected i.p. with 500 μL of zymosan (2 μg/μL in 1XPBS) and euthanized 36–48 h later by CO₂ fixation. The peritoneal cavity was lavaged with 10 mL of sterile ice-cold PBS, which was recovered after a 5 min. dwelling time. Neutrophils were isolated from the lavage fluid using anti-Ly6G magnetic beads (MACS; Miltenyi Biotec) according to the manufacturer’s instructions. Cell purity (95–99%) was confirmed via flow cytometry.

Bone-marrow neutrophil isolation

Naïve C57BL/6 adult mice were euthanized by CO₂ fixation. Femurs, tibias, and humeri were dissected and cleared of surrounding muscle. Bone marrow cells were flushed out with MACS buffer (1XPBS supplemented with 0.5% BSA and 2 mM EDTA). Neutrophils were isolated from total bone marrow cells using anti-Ly6G magnetic beads (MACS; Miltenyi Biotec), according to the manufacturer’s instructions. Cell purity (95–99%) was routinely confirmed by flow cytometry.

CNS mononuclear cell isolation

Mice were anesthetized with isoflurane and transcardially perfused with 10 mL of 1 × PBS. Spinal cords were isolated by hydraulic extrusion through the spinal canal using 5 mL of FACS buffer. Brain and spinal cord parenchyma were mechanically dissociated in HBSS supplemented with collagenase A (1 mg/mL; Roche) and DNase I (1 mg/mL; MilliporeSigma) using an 18-gauge needle, followed by enzymatic digestion at 37 °C for 20 min. Mononuclear cells were enriched by centrifugation in 27% Percoll (GE Healthcare) for 5 min at 800 × g and 4 °C. The myelin and debris layer was removed, and the resulting cell pellet was collected for flow cytometric analysis (see Flow Cytometry).

Following dissection of the brain and spinal cord, meninges were carefully peeled from the skull cap and spinal column respectively. Brain and spinal cord meninges were processed separately and digested in HBSS containing collagenase A (1 mg/mL) and DNase I (1 mg/mL) at 37 °C for 20 min and passed through a 70-µm cell strainer. Meningeal cells were collected for flow cytometric analysis.

Preparation of CD4 T cell-depleted splenocytes

Spleens were mechanically dissociated through a 70-µm cell strainer. Red blood cells were lysed using ACK lysis buffer (Quality Biologicals), and the reaction was quenched with FACS buffer. CD4 cells were depleted from the dissociated cell preparation using MACS CD4⁺ magnetic beads (L3T4; Miltenyi Biotec) according to the manufacturer’s instructions.

T cell suppression assay

CD4⁺ T cells were isolated from pooled lymph nodes and spleens of naïve C57BL/6 mice by positive selection using anti-CD4⁺ magnetic beads (L3T4; MACS, Miltenyi Biotec) according to the manufacturer’s instructions. Purified CD4⁺ T cells were labeled with CFSE using the CellTrace™ CFSE Proliferation Kit (Invitrogen). Arg1⁺ neutrophils (CD45⁺ CD11b⁺ Ly6G⁺ YFP⁺) and Arg1⁻ neutrophils (CD45⁺ CD11b⁺ Ly6G⁺ YFP⁻) were FACS-sorted from CNS mononuclear cells isolated from Arg1^YFP reporter mice during EAE (Fig. S2A). The purity of CD4⁺ T cells and neutrophil populations was routinely confirmed by flow cytometry (95–99%).

Unless otherwise indicated, suppression assays were performed at a 1:1 T cell to neutrophil ratio. For polyclonal stimulation, CFSE-labeled CD4⁺ T cells (0.3 or 1.5 × 10⁵ per well) were cultured in 200 µl of media in 96-well round-bottom plates coated with anti-CD3ε monoclonal antibody (1 µg/mL; eBioscience, clone 145-2C11) and supplemented with soluble anti-CD28 monoclonal antibody (1 µg/mL; eBioscience, Cat. No. 16–0281-82). For antigen-specific stimulation, CFSE-labeled 2D2 CD4⁺ T cells (5.0 × 10⁵ cells per well) were co-cultured with CD4⁻ splenocytes (5.0 × 10⁵ cells per well) pulsed with MOG_35–55 peptide (50 µg/mL). In selected experiments, blocking monoclonal antibodies were added at 10 µg/mL including PD-L1 (Invitrogen, Cat. No. 16–5982-85), GPNMB (Invitrogen, Cat. No. PA5-47,301), Galectin-9 (Bio X Cell, Cat. No. BE0218), PD-L2 (Bio X Cell, Cat. No. BE0112), and ICAM-1 (Invitrogen, Cat. No. MA5407). Cells were cultured for 96 h at 37 °C in 5% CO₂, harvested and analyzed by flow cytometry.

Transwell assays were conducted using 24-well Transwell plates fitted with 5.0-µm pore size inserts (Corning Costar). CFSE-labeled CD4⁺ T cells (0.5 × 10⁶ cells/mL) were seeded in the lower chamber with plate-bound anti-CD3ε monoclonal antibody and soluble anti-CD28 monoclonal antibody, as described above, while neutrophils (0.5 × 10⁶ cells/mL) were placed in the upper insert.

Statistics

Comparisons between two groups were conducted using paired or unpaired two-tailed Student’s t tests, as appropriate. Comparisons involving multiple groups were analyzed by one-way or two-way ANOVA with correction for multiple comparisons, as specified in the figure legends. Disease severity curves were compared using unpaired two-tailed Student’s t tests with Welch’s correction at individual time points where indicated, as well as by analysis of the area under the curve (AUC) for individual mice. In addition, overall disease course was assessed using a mixed-effects model. Differences in disease incidence were analyzed using a log-rank (Mantel–Cox) test. Outliers were identified using ROUT analysis and excluded when appropriate. Statistical significance was defined as P < 0.05 (*) with P < 0.01 (**), P < 0.001 (***), and P < 0.0001 (****). Statistical methods employed for RNA-seq analysis are described in the preceding subsection.

Schematic design

All figure schematics were created with BioRender.com.

Study approval

All animal experiments were performed in accordance with an IACUC-approved protocol at The Ohio State University.

Results

Alternatively activated neutrophils accumulate in the inflamed CNS during EAE

To identify aaN within the CNS during EAE, CD45.1⁺ encephalitogenic Th17 cells were adoptively transferred into CD45.2 ⁺ Arg1^YFP reporter mice. On day 7 post-transfer (p.t.), spinal cord immune cells and peripheral blood mononuclear cells were harvested and analyzed by multiparameter flow cytometry. T-distributed Stochastic Neighbor Embedding (tSNE) of pooled CD45⁺CD11b⁺Ly6G⁺ cells revealed discrete clustering that clearly distinguished CNS-infiltrating neutrophils from circulating neutrophils (Fig. 1A). Each cluster contained cells derived from multiple animals, indicating that the observed populations were consistently present across mice. Peripheral blood neutrophils (red) exhibited a conventional mature phenotype– Ly6G^hiCXCR2^hi—and lacked expression of Arg1-YFP and other alternative activation markers (Fig. 1A, B). In contrast, spinal cord-infiltrating neutrophils segregated into two populations: one resembling conventional neutrophils (purple; Ly6G^hiCXCR2^intF4/80⁻Arg1-YFP⁻), and the other expressing features consistent with immaturity (blue; Ly6G^lowCXCR2^lowF4/80⁺) and alternative activation (Arg1-YFP⁺CD84⁺). Notably, the cells in the latter cluster aberrantly expressed CD14, a marker typically associated with monocytes and monocyte-derived cells. We have previously observed a similar CD14⁺ aaN subset with neuroprotective properties in the peritoneum and vitreal fluid following local injection of the fungal cell wall extract, zymosan [29]. Importantly, the relative abundance of aaN in the spinal cord inversely correlated with clinical EAE severity at the time of euthanasia (Fig. 1C), suggesting a potential immunoregulatory or neuroprotective role.

Fig. 1.

An endogenous population of alternatively activated immature neutrophils emerges in the CNS during EAE. Spinal cord immune cells and peripheral blood mononuclear cells were harvested on day 7 post-transfer (p.t.) of encephalitogenic CD4 ⁺ T cells. A-C Cells pooled from 3 mice were analyzed by multiparameter flow cytometry. A t-distributed stochastic neighbor embedding (tSNE) of pooled neutrophils, identified as viable Ly6G⁺ CD11b⁺ CD45⁺ cells. Spinal cord clusters were manually delineated based on their spatial segregation within the t-SNE map. B Representative histograms showing mean fluorescent intensity (MFI) of the indicated markers across the three identified neutrophil clusters. Gray histograms indicate fluorescence minus one (FMO) controls used to define gating thresholds. C Linear regression analysis depicting the correlation between the total number of Arg1⁺  neutrophils per spinal cord at peak EAE against the area under the clinical disease score curve (AUC) for individual mice. D-G Spinal cord immune cells and peripheral blood mononuclear cells were isolated from EAE mice at four stages (preclinical, onset, peak, and late) and analyzed by single-cell RNA sequencing (RNAseq). D UMAP visualization of blood-derived (left) and spinal cord-derived (right) neutrophils pooled across all time points. E Heatmap of selected differentially express genes (scaled log2) across the three neutrophil clusters, highlighting markers of neutrophil maturation and alternative activation. F Gene Ontology Biological Process (GO BP) enrichment analysis of Cluster 3 marker genes. G Pie charts showing the relative proportions of circulating (left) and spinal cord-infiltrating (right) neutrophils within each cluster across different stages. H Heatmap of differentially expressed genes comparing blood- and spinal cord- derived neutrophils within cluster 3. I Violin plots comparing Arg1 expression blood- versus spinal cord- derived neutrophils within cluster 3. Data are from 1 experiment representative of two independent experiments with n = 3–5 mice per time point per experiment

To further characterize the diversity of CNS-infiltrating neutrophils, single-cell RNA sequencing (scRNA-seq) was performed on CD45⁺ cells FACS-isolated from spinal cord mononuclear cell preparations of EAE mice. Cells identified as neutrophils (Fig. S1) were subset for further analysis. UMAP analysis revealed a distinct neutrophil cluster (Cluster 3; green) enriched for transcripts associated with alternative activation, namely Arg1, Mrc1, Cd14, Ccl5, and Tgfb1, while being relatively depleted of terminal maturation markers such as S100a8, S100a9, Cd101, Sell and Cxcr2 (Fig. 1D, E). This scRNA-seq–defined aaN cluster resembled the aaN subset identified by flow cytometry, and similarly exhibited elevated CD14 expression. In contrast, Cluster 2 (red) displayed a transcriptional program consistent with mature circulating neutrophils (Cxcr2, Sell, Csf3r, Cd101) and was enriched in blood samples, whereas Cluster 1 (blue) exhibited features of inflammatory tissue neutrophils (Mpo, S100a8, S100a9) and was enriched in the spinal cord. Although Cd101 has been proposed as a marker distinguishing immature from mature neutrophils, its expression can be heterogeneous in neutrophils that have entered inflamed tissues [33]. Accordingly, cluster identity was assigned based on the overall transcriptional program rather than Cd101 expression alone.

Gene Ontology Biological Process (GO BP) enrichment analysis of Cluster 3 marker genes revealed significant enrichment of pathways involved in regulation of T cell activation, regulation of immune effector and inflammatory responses, negative regulation of T cell activation, regulation of neuron apoptotic processes, wound healing, myelination, response to IL-4, and neuron projection regeneration (Fig. 1F). Cluster 3 aaN accounted for a substantial proportion of spinal cord–infiltrating neutrophils across the disease course (Fig. 1G), suggesting that they represent a prominent leukocyte subset during neuroinflammation. Arg1, Tgfb1, Cd274, Cd14, and Icam1 transcripts were significantly upregulated in spinal cord-derived neutrophils within Cluster 3 compared with their blood-derived counterparts (Fig. 1H, I), consistent with transcriptional reprogramming towards a tissue-adapted phenotype within the CNS microenvironment. An alternative possibility is that aaN or their precursors originate from adjacent bone marrow niches and access the CNS through recently described osseous channels connecting the skull or vertebral bone marrow to the meninges [34]. In support of this possibility, Ccrl2, a marker associated with bone marrow–derived neutrophil trafficking, was enriched in the aaN cluster (Fig. 1E, H). In contrast, transcripts encoding osteocalcin (Bglap), another marker linked to vertebral marrow–derived cells, were not detected in our dataset.

aaN localize to inflammatory EAE lesions and associate with T cells

To examine the spatial distribution of aaN within the CNS during EAE, we analyzed spinal cord sections from Arg1^YFP reporter adoptive transfer recipients by immunohistochemistry combined with fluorescent in situ hybridization (RNAscope). Rather than dispersing broadly, YFP⁺Ly6G⁺CXCR2⁺ neutrophils were selectively localized to inflammatory lesions (Fig. 2). Within these lesions, aaN were observed juxtaposed to infiltrating CD3ε⁺ T cells, raising the possibility of direct cellular crosstalk within the inflammatory niche. This spatial proximity suggests that aaN may not simply be bystanders to pathology but active participants in shaping T cell responses.

The aaN that accumulate in EAE lesions exert potent T cell–suppressive activity

Given the established capacity of tumor-associated aaN subsets, such as G-MDSC and N2 cells, to suppress antitumor immunity, we next investigated whether EAE lesion–localized aaN exhibit analogous immunoregulatory activity. CNS-infiltrating YFP⁺Ly6G⁺ neutrophils, FACS-purified from Arg1^YFP reporter mice at the onset of clinical EAE, markedly inhibited the proliferation of CD4⁺CD3⁺ T cells polyclonally activated with plate-bound αCD3 and soluble αCD28 antibodies, whereas conventional YFP⁻Ly6G⁺ neutrophils isolated from the same spinal cords and T-depleted splenocytes had no effect (Fig. 3). This robust, antigen presenting cell-independent suppression underscores the capacity of aaN to directly limit T cell expansion. There was no difference in the viability of stimulated T cells across the experimental groups (Fig. S2B). Collectively, these findings identify aaN as lesion-localized myeloid regulators capable of directly modulating pathogenic T cell responses during neuroinflammation.

Fig. 3.

Alternatively activated neutrophils isolated from EAE spinal cords suppress T cell proliferation ex vivo. CFSE-labeled naïve CD4⁺ T cells were stimulated with plate-bound anti-CD3 and soluble anti-CD28 and cultured either alone or with the indicated Ly6G⁺ neutrophil populations at a 1:1 neutrophil–to–T cell ratio for 4 days. T cell proliferation was assessed by CFSE dilution using flow cytometry. Representative flow cytometry plots (left) show CD44 versus CFSE expression among gated viable CD45⁺CD11b⁻CD4⁺ cells, illustrating activation and proliferation under each condition. CFSE dilution among CD44⁺ activated T cells was used to quantify proliferating cells. Quantification of the percentage of CFSE^low dividing CD4⁺ T cells is shown on the right. Splenic neutrophils (SPL) were isolated from naïve mice by Ly6G magnetic bead separation. Arg1⁺ and Arg1⁻ neutrophils were FACS-isolated from the spinal cords of Arg1^YFP mice with EAE and defined as CD45⁺CD11b⁺Ly6G⁺Arg1-YFP⁺ and CD45⁺CD11b⁺Ly6G⁺Arg1-YFP⁻ cells, respectively. Each data point represents an individual biological replicate (CNS neutrophil subsets pooled from 1–9 mice). Data are pooled from three independent experiments. Bars indicate mean ± SEM. Statistical significance was determined using an unpaired, two-tailed Student’s t test. ns, not significant. *p < 0.05; ****p < 0.0001

Zymosan-induced intraperitoneal Ly6G⁺ cells serve as a surrogate population of alternatively activated neutrophils

The limited yield of aaN from the inflamed CNS during EAE restricts their utility for adoptive transfer–based functional studies. To address this limitation, we applied our previously established protocol [30, 35] to generate a phenotypically comparable aaN population ex vivo. Intraperitoneal (i.p.) injection of zymosan, a fungal cell wall component, induces an alternatively activated, neuroprotective neutrophil phenotype within the peritoneal cavity. Forty-eight to seventy-two hours following zymosan administration, peritoneal neutrophils were harvested and enriched via Ly6G magnetic separation (Fig. 4A). Compared with naïve bone marrow neutrophils (BMN), these zymosan-modulated neutrophils (ZMN) exhibited upregulation of Arg-1, CD14, CD84, and F4/80, along with reduced Ly6G expression, closely resembling EAE lesion-localized aaN (Fig. 1B and 4B, C). This approach provides a scalable surrogate population suitable for downstream mechanistic and adoptive transfer studies.

Fig. 4.

Zymosan-modulated neutrophils exhibit an alternatively activated phenotype. A Schematic illustrating the generation and isolation of zymosan-modulated neutrophils (ZMN) from mice following intraperitoneal administration of zymosan. B, C ZMN, isolated from i.p. zymosan injected from Arg1^YFP reporter mice, and bone marrow neutrophils (BMN) isolated from naïve from Arg1^YFP reporter mice, were analyzed by flow cytometry. B t-distributed stochastic neighbor embedding (t-SNE) analysis of ZMN (blue) and BMN (green), gated as viable CD45⁺CD11b⁺Ly6G⁺ cells. C t-SNE feature plots (left) and corresponding histograms (right) depicting expression of the indicated markers across the neutrophil clusters. Gray histograms represent controls used to define gating thresholds: ZMN from a non-reporter mouse for Arg1/YFP and fluorescence minus one (FMO) controls for the remaining markers. Neutrophils were pooled from three mice per group in a single experiment, representative of two independent experiments

Adoptive transfer of ZMN delays EAE onset and attenuates acute disease severity

To determine whether augmenting the endogenous aaN pool modulates disease progression, 5 × 10⁶ ZMN were intravenously administered to EAE mice on days 4 and 6 p.t. of encephalitogenic Th17 cells. Control groups received either PBS or BMN. Mice treated with ZMN exhibited a significant delay in disease onset and reduced early clinical scores compared to either control group (Fig. 5A). On day 10 p.t., EAE incidence was 100% in PBS and BMN groups but only 50% in ZMN-treated mice, which did not reach full penetrance until day 12 (Fig. 5A, right panel). Although peak disease severity was ultimately comparable across groups (Fig. S1), likely reflecting the short lifespan of transferred neutrophils, ZMN-treated mice exhibited significantly reduced cumulative disability during the first 10 days, as quantified by area-under\-the-curve (AUC) analysis (Fig. 5A, middle panel). EAE is frequently accompanied by optic neuritis and loss of retinal ganglion cells (RGCs), which reflects inflammatory injury to the optic nerve and provides a quantitative measure of neuroaxonal damage in this model [36, 37]. Consistent with the reduced early disease severity, ZMN-treated mice exhibited greater preservation of RGCs at the experimental endpoint (Fig. 5B), indicating reduced neuronal injury despite similar clinical scores at that stage.

Fig. 5.

Adoptive transfer of zymosan-modulated neutrophils ameliorates EAE, enhances neuronal survival, and decreases encephalitogenic T cell accumulation. Clinical and immunological outcomes were assessed in C57BL/6 mice subjected to Th17-mediated adoptive transfer EAE. Mice received intravenous (i.v.) injections of PBS (black circles), 5 × 10⁶ bone marrow–derived neutrophils (BMN; green triangles), or 5 × 10⁶ zymosan-modulated neutrophils (ZMN; blue squares) on days 4 and 6 post-transfer (p.t.) of encephalitogenic CD4⁺ T cells. A Clinical disease measures including mean daily clinical scores over time (left), area under the clinical score curve (AUC) for individual mice (middle), and the percentage of mice exhibiting neurological deficits over time (right). Data are from one experiment (n = 4–7 mice per group), representative of three independent experiments. B Left: Representative retinal whole mounts stained for retinal ganglion cells (RGCs; using anti-RBPMS antibodies) from EAE mice that received either PBS, BMN, or ZMN at day 6 p.t. Scale bars represent 40 μm. Right: Quantification of average RGC density (cells per mm²). Data represent 6–8 retinas per group. C Representative flow cytometry histograms (left), absolute numbers of CD45.1⁺ donor-derived CD4⁺ T cells (middle), and frequencies (right) of CD45.1⁺ donor-derived cells among total CD4⁺ T cells (gated on viable CD45⁺CD11b⁻CD4⁺ cells) isolated from the spinal cord (top) and meninges (bottom) at day 20 p.t. from EAE mice treated with PBS, BMN, or ZMN. Data represent 7–10 mice per group, pooled from two independent experiments. Statistical significance was determined using a two-tailed Student’s t test unless otherwise indicated. Incidence curves in panel A were compared using a log-rank (Mantel–Cox) test. Data are presented as mean ± SEM. p < 0.05; *p < 0.01; **p < 0.001; ***p < 0.0001

The rapid therapeutic effect of ZMN treatment in EAE suggests an underlying immunosuppressive mechanism. Indeed, flow cytometric analysis of CNS tissues confirmed that ZMN-treated mice exhibited significantly reduced frequencies and absolute numbers of adoptively transferred CD45.1⁺ encephalitogenic T cells in both the meninges and spinal cord parenchyma (Fig. 5C), consistent with a dampening of autoimmune neuroinflammation.

Transferred ZMN are detectable in peripheral lymphoid and pulmonary tissues.

To track their trafficking, ZMN were generated from tdTomato reporter mice and injected intravenously into host mice on day 6 following the adoptive transfer of CD45.1⁺ encephalitogenic CD4⁺ Th17 cells. Four hours following their administration, tdTomato⁺ ZMN were readily detected in the spleen and lungs by flow cytometry (Fig. S4A) but were largely absent from CNS compartments, including the brain, spinal cord, and meninges. At this timepoint, donor-derived CD45.1⁺CD4⁺ T cells were present in the same peripheral tissues (Fig. S4B), suggesting that ZMN may exert immunoregulatory effects through modulation of T cell responses at extra-CNS sites. By 24 h post-administration, few tdTomato⁺ ZMN were detected in the CNS (data not shown), although transient accumulation at earlier intermediate time points cannot be excluded.

ZMN suppress CD4⁺ T cell proliferation via a contact-dependent mechanism in vitro

To test whether ZMN directly modulate CD4⁺ T cell responses, analogous to EAE lesion-resident aaN, we co-cultured ZMN with CFSE-labeled naïve 2D2 T cell receptor (TCR)-transgenic CD4⁺ T cells (specific for MOG_35–55) in the presence of peptide–pulsed, T cell–depleted splenocytes. Under these conditions, ZMN markedly suppressed 2D2 CD4⁺ T cell proliferation, as measured by CFSE dilution, when compared with cultures of the T cells alone or co-cultured with BMN (Fig. 6A). To determine whether antigen presenting cells were required for this immunosuppressive effect, we repeated the assay with polyclonal wildtype CD4⁺ T cells stimulated with plate-bound αCD3 and soluble αCD28. ZMN retained their suppressive activity (Fig. 6B, S5A), indicating a direct inhibitory effect on T cells. Importantly, this immunosuppressive capacity was temporally restricted, as when ZMN were introduced 48 h after T cell activation, they no longer inhibited proliferation (Fig. S5B). The viability of T cells was similar across all experimental conditions, indicating that aaN were not cytotoxic (Fig. S5D).

Fig. 6 .

Zymosan-modulated neutrophils suppress CD4⁺ T cell proliferation via a cell contact–dependent mechanism. A Antigen-specific suppression assays were performed by co-culturing CFSE-labeled MOG₃₅–₅₅–specific 2D2 CD4⁺ T cells with MOG₃₅–₅₅–pulsed, T cell–depleted splenocytes in the absence of neutrophils (black circles) or in the presence of bone marrow–derived neutrophils (BMN; green triangles) or zymosan-modulated neutrophils (ZMN; blue squares) for 4 days. T cell proliferation was assessed by CFSE dilution using flow cytometry. Representative flow cytometry plots showing CFSE dilution and CD44 expression in CD4⁺ T cells are shown, along with quantification of the percentage of CFSElow dividing cells among total CD4⁺ T cells (gated on viable CD45⁺CD11b⁻CD4⁺ cells). B CFSE-labeled naïve CD4⁺ T cells were stimulated with plate-bound anti-CD3 and soluble anti-CD28 and cultured alone or with BMN or ZMN for 4 days. C To assess contact dependence, CFSE-labeled naïve CD4⁺ T cells stimulated with anti-CD3 and anti-CD28 were co-cultured with the indicated neutrophil populations using transwell inserts to prevent direct cell–cell contact. Each data point represents an individual biological replicate. Statistical significance was determined using a two-tailed Student’s t test. Error bars represent mean ± SEM. p < 0.05; **p < 0.001

To distinguish between effects mediated by soluble factors versus direct cell–cell contact, we performed transwell experiments separating ZMN from αCD3/αCD28-stimulated CD4⁺ T cells. ZMN failed to suppress T cell proliferation when physical contact was prevented (Fig. 6C) and, similarly, conditioned media from ZMN cultures had no effect on T cell proliferation (Fig. S5C).

We next explored candidate immunoregulatory pathways. Flow cytometry revealed that ZMN expressed high surface levels of PD-L1 compared to BMN (Fig. 7A). Additionally, transcriptomic profiling indicated elevated expression of glycoprotein non-metastatic melanoma protein B (GPNMB), a checkpoint molecule associated with G-MDSC function (data not shown). However, blocking PD-L1 or GPNMB—individually or in combination—failed to reverse ZMN-mediated suppression of CD4⁺ T cell proliferation (Fig. 7B and Fig. S6). Similarly, supplementation of the co-cultures with L-arginine or an arginase-1 inhibitor, with or without PDL1 blockade, did not restore T cell proliferation (Fig. S6 and unpublished data), suggesting involvement of additional or redundant mechanisms.

Fig. 7.

Zymosan-modulated neutrophil–mediated immunosuppression is independent of PD-L1. A Representative histograms showing PD-L1 expression on naive bone marrow–derived neutrophils (BMN; green) and zymosan-modulated neutrophils (ZMN; blue), with fluorescence-minus-one (FMO) control shown as a gray dotted line. Neutrophils were gated as viable CD45⁺CD11b⁺Ly6G⁺ cells. B Representative CFSE dilution plots of naïve CD4⁺ T cells stimulated with anti-CD3 and anti-CD28 and cultured either alone or in co-culture with ZMN in the presence or absence of PD-L1–blocking antibodies. Quantification of the percentage of CFSE.^low dividing cells among total CD4⁺ T cells (gated on viable CD45⁺CD11b⁻CD4⁺ cells) is shown in the bottom panel. Data represent 2–3 biological replicates per group pooled from two independent experiments, with each biological replicate comprising 2–3 technical replicates. Statistical significance was determined using a two-tailed Student’s t test. Data are presented as mean ± SEM. p < 0.05

Discussion

Neutrophils have traditionally been regarded as short-lived cytotoxic cells that can contribute to neuropathology during autoimmune demyelinating disease by disrupting the blood–brain barrier and amplifying inflammation. Our findings broaden this paradigm by identifying a subset of immature aaN that arise spontaneously within inflammatory demyelinating lesions and directly suppress T cell proliferation. Single-cell profiling revealed that these aaN constitute a transcriptionally and phenotypically distinct population, divergent from conventional circulating neutrophils. They express Arg1, Cd14, and Mrc1, while lacking terminal maturation markers. Their selective presence in the CNS suggests that polarization occurs in response to local microenvironmental cues rather than peripheral signals. The colocalization of aaN with CD3⁺ T cells within EAE lesions, together with their potent, contact-dependent inhibition of T cell proliferation, positions them as active regulators of adaptive immunity rather than passive bystanders. We chose the Th17 adoptive transfer model to investigate the potential role of alternative neutrophil subsets in autoimmune demyelinating disease in part because Th17 cells are potent inducers of neutrophil mobilizing and chemoattractant factors. In future experiments we we plan to explore their roles in EAE induced by Th1 adoptive transfer and active immunization.

While immunosuppressive monocyte-derived and mesenchymal cells have shown therapeutic efficacy in graft-versus-host disease and systemic autoimmunity [38, 39], our results provide the first evidence that exogenously generated granulocytic aaN can attenuate CNS autoimmunity. Intravenous administration of zymosan-induced aaN (ZMN) delayed EAE onset, reduced encephalitogenic T cell infiltration, and preserved retinal neurons. These therapeutic benefits, though significant, were transient— likely reflecting the short lifespan of neutrophils and their dilution after systemic transfer. Consistent with this possibility, transferred ZMN were not detected among CNS mononuclear cells at the experimental endpoint by flow cytometry. Alternatively, aaN may exert their greatest effects during the acute phase of lesion development but have less impact during later stages of disease progression.

We previously showed that ZMN polarization requires dectin-1, a receptor for β1,3-glucan that mediates recognition of zymosan [35]. Whether endogenous aaN that emerge in Th17-mediated EAE are polarized by an intrinsic dectin-1 ligand, such as galectin-9, remains to be determined. This possibility aligns with prior findings that dectin-1 deficiency exacerbates actively induced EAE and increases CNS T-cell infiltration, with neutrophils representing the predominant dectin-1⁺ population [40]. The immunosuppressive mechanism of EAE-modulating ZMN differs from canonical G-MDSCs. Blockade of PD-L1, GPNMB, or arginase-1 failed to diminish their suppressive capacity, indicating reliance on distinct regulatory pathways. Dectin-1–activated bone marrow neutrophils have previously been reported to induce ROS-dependent apoptosis of CD4⁺ T cells [37]. In contrast, we found that dectin-1-activated peritoneal ZMN do not impair T-cell viability (Fig. S2B, S5D), underscoring the context-dependent specialization of immunoregulatory neutrophil subsets.

The anatomical site where transferred ZMN exert their immunosuppressive effects during EAE remains unclear. Donor ZMN were rarely detected in CNS cell preparations of EAE mice at either 4 or 24 h post-transfer, yet were readily observed in the spleen and lungs. Given evidence that myelin-reactive T cells undergo a critical phase of “education” within the lungs during the prodromal phase of EAE, prior to CNS entry [41], ZMN may act peripherally to limit encephalitogenic T cell activation and polarization. Alternatively, they could exert their effects remotely, through the release of extracellular vesicles or neutrophil extracellular traps (NETs) in the periphery, which subsequently traverse the blood–brain or blood-meningeal barrier to target infiltrating T cells in situ. Supporting this possibility, circulating NETs have been shown to modulate CNS localized cells in a model of post-stroke emotional disorders [42]. Finally, it remains possible that ZMN transiently accumulate within the CNS at time points not captured in our analysis, such that their peak presence was missed.

We previously showed that peritoneal ZMN exhibit intrinsic neuroprotective and axon-regenerative effects in models of traumatic optic nerve and spinal cord injury, partly via production of nerve growth factor (NGF) and insulin-like growth factor 1 (IGF-1) [30]. Zymosan stimulation can also drive neutrophils to express oncomodulin, a trophic factor with neuroprotective properties [40]. In the present study, however, the neuroprotection of RGC observed in ZMN-treated EAE mice likely occurred indirectly as a consequence of reduced optic nerve inflammation rather than through a direct neurotrophic effect. These findings suggest that therapeutic strategies designed to enhance endogenous aaN activity or to generate autologous aaN for adoptive cell therapy may hold particular promise for relapsing forms of multiple sclerosis, where suppression of inflammatory activity is a primary therapeutic goal. Future studies will determine whether repeated or delayed administration of ZMN can promote tissue repair during periods of disease remission or chronicity, which could potentially extend their therapeutic relevance to progressive stages of MS.

The relationship between ZMN and the endogenous aaN arising in Th17 adoptive transfer EAE also remains unresolved. Endogenous aaN appear to originate within the CNS, as we did not detect them in peripheral tissues. Similarly, Knier et al. reported that, in EAE induced by active immunization, Ly6G⁺ cells express Arg1⁺ only after CNS entry [41]. The specific CNS-derived factors driving aaN polarization remain unknown, as does the in vivo impact of endogenous aaN. Whether they limit inflammation in vivo or conversely whether their depletion will exacerbate disease, requires further study. Notably, in our scRNA-seq analysis, aaN emerged from within a transcriptionally defined neutrophil compartment characterized by expression of canonical granulocytic lineage genes such as Csf3r, Ly6g, Mpo, and Cxcr2, supporting the interpretation that aaN represent a specialized activation state within the neutrophil lineage during CNS inflammation.

Additional evidence supports an immunoregulatory role for neutrophils in EAE. Ioannou et al. identified CD3⁻CD19⁻CD11b^hiLy6G⁺ cells in spleen and spinal cord of mice actively immunized with MOG_35–55/CFA [42]. These neutrophils suppressed CD4⁺CD25⁻ T cell proliferation in vitro. Transfer of these cells reduced EAE clinical severity and dampened Th1/Th17 responses. Unlike ZMN, their activity was PD-L1–dependent and they lacked detectable arginase-1 expression, highlighting functional diversity among immunosuppressive neutrophil subsets that arise in inflammatory environments. Suppressive granulocyte populations described in other settings, including granulocytic myeloid-derived suppressor cells (G-MDSCs) that have been identified in tumor beds, share certain phenotypic and biological properties with aaN, such as the capacity to suppress T-cell responses and expression of markers including Arg1. However, accumulating evidence suggests that G-MDSCs largely represent pathologically activated or developmentally immature neutrophils rather than a distinct myeloid lineage [28, 43, 44]. We therefore interpret aaN as a distinct activation state within the broader spectrum of neutrophil heterogeneity rather than as a separate suppressor-cell lineage. Clarifying how ZMN, aaN, and other unconventional neutrophil subsets interrelate—and delineating their suppressive and reparative mechanisms—will be essential to harnessing therapeutic potential.

Our findings that exogenously generated aaN can attenuate autoimmune neuroinflammation broaden the scope of myeloid cell–based therapeutics beyond macrophages and mesenchymal stromal cells. Neutrophils are the most abundant circulating leukocyte in humans, rapidly mobilize to sites of injury, and are highly amenable to functional reprogramming—properties that position them as a potentially powerful yet underexplored platform for short-term, targeted immune modulation in CNS autoimmunity. Practical limitations, including their brief lifespan and dilution following systemic delivery, could be overcome through repeated dosing, localized administration, or bioengineering strategies to enhance persistence and potency. Moving toward clinical translation, we have established a protocol for generating human CD33⁺CD15⁺ aaN from bone marrow precursors, that are capable of stimulating robust neurite outgrowth [28]. Notably, CD33⁺CD15⁺ cells are enriched in the circulation of patients with active multiple sclerosis and potently suppress autologous CD4⁺ T-cell activation and proliferation [42]. These findings not only underscore the immunomodulatory and neurotrophic potential of human aaN but also lay a foundation for developing neutrophil-based interventions tailored to the inflammatory and neurodegenerative components of MS and related disorders. Harnessing the plasticity of neutrophils may open a new therapeutic frontier in which a patient’s own short-lived effector cells are reprogrammed to transiently undermine pathogenic immunity while simultaneously fostering neural repair.

Conclusions

In conclusion, our study identifies aaN as a previously underappreciated immunoregulatory population that arises within the inflamed CNS during autoimmune neuroinflammation and directly constrains encephalitogenic T cell responses. Using integrated single-cell, spatial, and functional approaches, we show that endogenous aaN are transcriptionally and phenotypically distinct from conventional neutrophils, co-localize with pathogenic T cells within neuroinflammatory lesions, and potently suppress CD4⁺ T cell proliferation. Therapeutic augmentation of this pathway using ex vivo–generated aaN attenuated early disease activity, reduced CNS T cell accumulation, and limited neuronal injury in EAE. Notably, ex vivo–generated aaN inhibited T cell proliferation through a contact-dependent mechanism independent of canonical immune checkpoint pathways. Collectively, these findings expand current paradigms of myeloid cell involvement in CNS autoimmunity and highlight distinct neutrophil subsets as a foundation for novel cell-based immunomodulatory strategies, with potential relevance for the treatment of multiple sclerosis and related neuroinflammatory disorders.

Abbreviations

aaN

Alternatively activated neutrophils

arg1

Arginase-1

BMN

Bone marrow neutrophils

CNS

Central nervous system

EAE

Experimental autoimmune encephalomyelitis

G-MDCS

Granulocytic myeloid derived suppressor cells

IGF-1

Insulin-like growth factor-1

i.p.

Intraperitoneal

MOG

Myelin oligodendrocyte glycoprotein

Mrc1

Mannose receptor

MS

Multiple Sclerosis

NETs

Neutrophil extracellular traps

NGF

Nerve growth factor

PFA

Paraformaldehyde

p.t.

Post-transfer

RGC

Retinal ganglion cell

RNA-seq

RNA sequencing

TCR

T cell receptor

ZMN

Zymosan-modulated neutrophils

Authors’ contributions

JA: Conceptualization, Data curation, Funding acquisition, Formal Analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. CB: Data curation, Formal Analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. AJ: Conceptualization, Data curation, Investigation, Methodology, Visualization. AG: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Visualization. HG: Data curation, Investigation, Methodology, Visualization, Writing – review & editing. SC: Data curation, Investigation, Methodology, Visualization. AMM: Data curation, Formal Analysis, Methodology, Visualization. JS: Investigation, Methodology. AD: Investigation, Methodology. LH: Methodology, Data curation, Formal Analysis, Writing – review & editing. TL: Investigation, Methodology. AW: Data curation, Formal Analysis, Visualization, Writing – review & editing. AS: Conceptualization, Investigation, Methodology, Project administration, Supervision. BS: Conceptualization, Funding acquisition, Project administration, Supervision, Formal Analysis, Methodology, Resources, Writing – original draft, Writing – review & editing.

Funding

The manuscript is supported by funding awarded to BMS from the National Eye Institute, NIH (R01 EY036099-01A1) and the Wings for Life Spinal Cord Research Foundation, and by a Career Transition Fellowship awarded to JA from the National Multiple Sclerosis Society (TA-2305–41248).

Data availability

The RNAseq data have been uploaded to the GEO repository and are available under the accession number GSE316626. All other datasets are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All animal studies were approved by the Institutional Animal Care and Use Committee of The Ohio State University. The study was conducted in accordance with the local legislation and institutional requirements.

Competing interests

Benjamin Segal, Andrew Sas, Andrew Jerome and Tom Lui are inventors on a patent (Publication No. US 2025/0236843 A1) for the use of myeloid cells in the treatment of neurological disorders.