Multiple Sclerosis (MS) is an autoimmune disease of the central nervous system (CNS) characterized by immune-mediated demyelination, neuroinflammation, and progressive neurodegeneration [1–4]. While adaptive immune cells such as T and B lymphocytes are well established as key drivers of MS and its animal model, Experimental Autoimmune Encephalomyelitis (EAE), the roles of innate immune cells remain less defined [3, 5, 6]. Over twenty FDA-approved disease-modifying therapies reduce relapse rates in relapsing-remitting MS; however, none halt long-term progression, underscoring the need to identify pathogenic and protective immune subsets through high-resolution profiling [7].

The Janus Kinase/Signal Transducer and Activator of Transcription (JAK/STAT) pathway, activated by more than 70 cytokines, is central to immune regulation [8–11]. Aberrant STAT activation has been reported in MS, Alzheimer’s disease (AD), and Parkinson’s disease (PD) [8, 12–16]. Suppressors Of Cytokine Signaling (SOCS) proteins serve as negative feedback regulators of JAK/STAT signaling [9, 17–20], and dysregulated Socs3 expression in MS correlates with enhanced STAT3 activation [21, 22]. Our prior work showed that pharmacologic JAK/STAT inhibition ameliorates EAE by suppressing pathogenic T-cell differentiation, dampening myeloid activation, and limiting leukocyte infiltration into the CNS [8, 23].

EAE manifests as distinct clinical subtypes: classical EAE (cEAE), with ascending paralysis and spinal cord (SC) inflammation, and brain-targeted EAE (btEAE), marked by ataxia, tremors, and cerebellar (CB) inflammation[6, 24–28]. Previously, we showed that mice lacking Socs3 in myeloid cells (Socs3^ΔLysM) develop a severe mixed phenotype with both cEAE and btEAE. The btEAE phenotype depends on cerebellar neutrophil infiltration and activation [24–26]. These findings suggest that myeloid SOCS3 restrains region-specific CNS inflammation and neutrophil-driven pathology.

Neutrophils are increasingly recognized as important mediators of neuroinflammation across several CNS disorders, including MS, Neuromyelitis Optica Spectrum Disorders, AD, PD, and stroke [29–38]. In MS, neutrophils display hyperactivation characterized by increased degranulation, reactive oxygen species (ROS) production, and formation of neutrophil extracellular traps (NETs) [39–41]. Neutrophils are among the earliest immune cells to infiltrate the CNS, where they disrupt the blood–brain barrier (BBB), promote demyelination, and amplify inflammatory responses [42–46]. Conversely, neutrophils can exert protective roles, such as suppressing pathogenic B-cell responses [47] and promoting axon regeneration [48–51]. While single-cell RNA sequencing (scRNA-seq) has revealed neutrophil heterogeneity in MS and cEAE [45–48, 52, 53], the diversity and function of these cells in btEAE remain poorly understood, revealing a critical knowledge gap relevant to disease mechanisms.

In this study, we generated mice with neutrophil-specific Socs3 deletion (Socs3^ΔLy6G) to define the role of neutrophil-intrinsic JAK/STAT signaling in CNS autoimmunity. Socs3^ΔLy6G mice developed both severe btEAE and cEAE, with the btEAE phenotype driven by Socs3-deficient neutrophils. Using single-cell transcriptomics, we identified distinct neutrophil subsets, including inflammatory clusters enriched in Socs3^ΔLy6G mice during EAE. Importantly, Saa3 emerged as a candidate effector and potential therapeutic target. Together, these findings link dysregulated neutrophil JAK/STAT signaling to brain-targeted neuroinflammation and reveal a previously unappreciated mechanism of neutrophil-mediated pathology in MS.

Mice. Transgenic mice with the Socs3 locus flanked with flox sequences (Socs3^fl/fl) [54], the generous gift of Dr. Warren Alexander (Walter and Eliza Hall Institute of Medical Research; Victoria, Australia), were bred at UAB. Socs3^ΔLysM mice were generated as previously described [25]. Mice with Socs3 deletion exclusively in neutrophils (Socs3^ΔLy6G) were generated by serial breeding of Socs3^fl/fl mice with Ly6gCre^Tdtomato/+ mice (“Catchup mice”) [55], a generous gift from Professor Matthias Gunzer, Institute for Experimental Immunology and Imaging, University Hospital, University Duisburg-Essen, Essen, Germany. The highly neutrophil-specific locus for Ly6g was genetically modified to drive expression of both Cre-recombinase and tdTomato. Ly6G^+/− mice serve as controls. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of UAB.

Human Plasma Samples. Subjects with MS and healthy controls (HC) were recruited from the UAB Comprehensive MS Center. Data were collected in a double-blinded manner until all data collection was completed. These studies were conducted in compliance with the Helsinki Declaration and approved by the Institutional Review Board at UAB. All participants provided informed consent. EDTA plasma was collected in BD Vacutainer EDTA tubes and stored at − 80°C [56]. Information on the MS patient population is summarized in Supplementary Table 1.

Experimental Autoimmune Encephalomyelitis (EAE) Induction and Assessment. EAE was induced in Socs3^fl/fl, Ly6G^+/−, Socs3^ΔLysM and Socs3^ΔLy6G mice by s.c. injection of MOG_35–55 emulsified in CFA (Hooke Laboratories) along with i.p. injection of 100 ng Pertussis Toxin (Hooke Laboratories) on Days 0 and 1, as previously described [25, 26, 57]. Mice experience cEAE, btEAE, and/or a mixed phenotype after EAE induction. cEAE was scored as follows: 0, no disease; 1, decreased tail tone; 2, hind limb weakness or partial paralysis; 3, complete hind limb paralysis; 4, front and hind limb paralysis; and 5, moribund state. Assessment of btEAE was as follows: 0, no disease; 1, hunched appearance, slight head tilt; 2, ataxia, scruffy coat; 3, severe head tilt, slight axial rotation, staggered walking; 4, severe axial rotation, spinning; and 5, moribund. For mixed phenotypes, cEAE and btEAE were scored separately [26]. Mice were sacrificed on days 12–14 at the peak of EAE for histology, flow cytometry and scRNA-Seq analysis. The CB and SC were isolated as previously described [26].

Assessment of Demyelination and NET Formation. Mice were anesthetized and intracardially perfused with PBS, followed by 4% paraformaldehyde. CB tissues were fixed with 4% PFA at 4°C overnight and then dehydrated with 30% sucrose. Cryoprotected CB tissues were embedded in an Optimal Cutting Temperature compound and cryosectioned to produce 40 µm slices in the sagittal plane from the center. Sections were stained using the Black Gold II Myelin Staining Kit (Millipore Sigma, AG105) [26, 58]. Images of stained sections were acquired with the Keyence Microscope BZ-X800. The total arbor vitae (white matter) area and the myelinated area (area of black-gold staining) of the CB were measured using the “Hybrid Cell Count” module provided by Keyence Microscope. “% Cerebellar Myelination Area” was defined as the myelinated area divided by the total arbor vitae area. For assessment of NET formation, slides were stained for citrullinated histone 3 (citH3), a biomarker of NET formation, the neutrophil marker Ly6G, and DAPI.

Antibodies and Cytokines. For flow cytometry experiments, antibodies (Abs) directed against murine CD11b (M1/70), CD45 (30-F11), Ly6C (HK1.4), Ly6G (1A8), CXCR2 (SA044G4), CD62L (MEL-14), CD63 (NVG-2), CD3 (17A2), CD4 (GK1.5), CD8 (53 − 6.7), CD44 (IM7), IFN-γ (XMG1.2) and IL-17A (TC11-18H10.1) were from BioLegend (San Diego, CA). The LIVE/DEAD® Fixable Aqua Stain kit (L34957) was from Thermo Fisher Scientific (Waltham, MA).

Flow Cytometry. For surface protein detection, cells were incubated with Fc Block (2.4G2) for 15 min. and washed, followed by incubation with viability dye and the indicated Abs, as previously described [26]. CM-H2DCFDA, a general oxidative stress indicator, was used to detect total ROS production [59, 60]. For intracellular cytokine staining, cells were stimulated with PMA (25 ng/ml) and ionomycin (1 µg/ml) in the presence of GolgiStop (BD Biosciences, San Jose, CA) for 4 h and permeabilized using the BD Fixation/Permeabilization Kit (BD Biosciences, San Jose, CA), as previously described [61].

For analysis of cells from the EAE experiments, mice were sacrificed and whole-body perfusion was performed. Mononuclear cells were isolated from the CB and SC using a 30%/70% Percoll gradient. Cell phenotypes were determined based on surface and intracellular staining patterns analyzed by flow cytometry, as previously described [23, 25, 26]. All flow cytometry data were analyzed using FlowJo software (version 10.8.1; TreeStar, Ashland, OR).

Cytokine/Chemokine Analysis. The supernatant of CB and SC tissue homogenates was used to determine cytokine and chemokine levels as measured by the Cytokine/Chemokine Multiplex ELISA assay (Millipore, St. Louis, MO) as previously described [62]. Analyte concentrations were normalized by protein concentration.

Sorting of Neutrophils. Mononuclear cells were isolated from the CB and SC of Ly6G^+/− or Socs3^ΔLy6G EAE mice as previously described [24, 63]. CD45⁺CD11b⁺Ly6G⁺Ly6C^low neutrophils were isolated by flow cytometry.

Quantitative RT-PCR. 500–1000 ng of RNA from sorted neutrophils was used as a template for cDNA synthesis. qRT-PCR was performed using TaqMan primers purchased from Thermo Fisher Scientific. The resulting data were analyzed using the comparative cycle threshold method to calculate relative RNA quantities [57].

ELISA. The murine SAA3 ELISA kit (EZMSAA3-12K; EMD Millipore, Burlington, MA determined the concentrations of SAA3 in murine plasma and supernatants from CB and SC tissue homogenates. The human SAA1 ELISA kit (DY3019-05; R & D Systems, Minneapolis, MN) detected human SAA1 in plasma samples from MS patients and HC. The Nu.Q^® Discovery H3.1 ELISA kit (1001-01-03; Volition, Carlsbad, CA) quantified nucleosomes in plasma samples from MS patients and HC, as well as in murine plasma and supernatants of CB and SC tissue homogenates.

Cell Isolation and Single-cell RNA Sequencing (scRNA-Seq). Due to the scarcity of CD45⁺CD11b⁺ cells in the CB of naïve mice (both Ly6G^+/− and Socs3^ΔLy6G) and Ly6G^+/− mice with cEAE, we could not perform scRNA-Seq on cells from the CB of these mice. Mononuclear immune cells were isolated from the SC of naïve Ly6G^+/− or Socs3^ΔLy6G mice, from the SC and CB at the peak of EAE disease at day 12 for Socs3^ΔLy6G mice, or from the SC at the peak of EAE disease at day 14 for Ly6G^+/− mice following anesthetization, intracardial PBS perfusion and a 30%/70% Percoll gradient separation [26]. Sorted live CD45⁺CD11b⁺ cells from the SC of Ly6G^+/− mice with EAE (n = 3), from the SC and CB of Socs3^ΔLy6G mice with EAE (n = 3), and from the combined SC of naïve Ly6G^+/− (n = 3), and naïve Socs3^ΔLy6G (n = 4) mice were subjected to scRNA-Seq. One biological sample consisted of samples pooled from three mice with TotalSeq™-B Hashtag antibodies from BioLegend [64]. All sorted CD45⁺CD11b⁺ cell libraries were prepared using the 10X Genomics Chromium Single Cell 3′ Reagent Kit V3.1 and sequenced on Illumina NextSeq 500 as previously described [45, 47, 48, 65]. Raw base call files were demultiplexed into FASTQ files. Sequencing files were processed and mapped to mm10, and count matrices were extracted using the Cell Ranger Single Cell Software (v 7.1.0) [45, 66, 67].

scRNA-Seq Analysis. The count matrices in the h5 file format generated from Cell Ranger were imported into the Partek Flow (Partek Inc) pipeline [66, 68]. Single-cell quality control was performed by applying an inclusion filter on counts per cell (500-15000) and detected genes per cell (250–5000). Cells with greater than 10% mitochondrial gene expression were excluded to eliminate apoptotic or dying cells [69].

The dataset was also applied by setting the noise reduction filter to exclude features where the value ≤ 0 in at least 99.9% of cells. The filtered dataset was normalized and scaled with the SCTransform workflow. Principal Component Analysis (PCA) was performed on the SCTransform-scaled data. The PCA data node was chosen to perform graph-based clustering based on the Louvain algorithm, with the number of PCA set to 20. The data was visualized using 3D Uniform Manifold Approximation and Projection (UMAP) dimensional reduction with the first 20 principal components. Cell annotations for each cluster were determined using the top differentially expressed genes (DEGs) in computed biomarkers and canonical markers following the classification workflow in Partek Flow [66, 70].

CD45⁺CD11b⁺ cells from four conditions are as follows: combined SC of naïve Ly6G^+/− (356 cells) and naïve Socs3^ΔLy6G mice (795 cells); SC of Ly6G^+/− EAE mice (4,106 cells); SC of Socs3^ΔLy6G EAE mice (11,394 cells); and CB of Socs3^ΔLy6G EAE mice (10,639 cells). The neutrophil clusters were subsetted from all of the cell clusters. Neutrophils from four different conditions are as follows: combined SC of naïve Ly6G^+/− mice (114 cells) and naïve Socs3^ΔLy6G mice (263 cells); SC of Ly6G^+/− EAE mice (2,102 cells); SC of Socs3^ΔLy6G EAE mice (6,458 cells); and CB of Socs3^ΔLy6G EAE (6,914 cells). DEGs between different samples were determined by the Hurdle model on log2-normalized counts. The dot plots and violin plots were generated with sc.pl.dotplot and sc.pl.violin functions in Scanpy (1.9.1) package [71] using the annotated h5ad files exported from Partek workflow [66].

Pathway Enrichment Analysis. GSEA (Gene Set Enrichment Analysis): DEG analysis between individual neutrophils versus other neutrophil clusters was performed with Gene Specific Analysis (GSA) test in Partek workflow [66]. The exported DEG list was ranked by -log(P) and converted to an RNK file, which was uploaded to GSEA software (Version 4.3.2, BROAD Institute) to run GSEAPreRanked by choosing a hallmark gene sets database [72]. The pathway analysis results were plotted in terms of normalized enrichment score (NES) and false discovery rate (FDR) using the ggplot2 (Version 3.4.0) package in RStudio.

NicheNet Analysis. The annotated h5ad files were read into R using the anndata package. The Seurat object [73, 74] was created for NicheNet analysis. The expression data of interacting cells was extracted from the Seurat object of integrated data. Neu1, Neu2, Neu3, Neu4 and Neu5 clusters were defined as the sender cell populations and macrophages or microglia were defined as the receiver cell populations. One comparison of interest was Socs3^ΔLy6G CB versus Ly6G^+/− SC: the condition of interest was set to Socs3^ΔLy6G CB and the reference condition was set to Ly6G^+/− SC. NicheNet analysis was performed according to the published workflow utilizing published ligand-target, ligand-receptor network and weighted integrated networks [75]. The selected differentially expressed ligand or receptor was visualized in violin plots.

Statistics. Significant differences between the two groups were analyzed by Student’s t-test distribution. One-way ANOVA was used to compare differences between more than two samples, and the Mann-Whitney rank sum test was used for EAE scores. p-values less than 0.05 were considered statistically significant. All error bars represent the standard error of the mean (SEM). Statistical analyses were performed with GraphPad Prism 9 (GraphPad Software, La Jolla, CA).

Data Set Availability. ScRNA-Seq data will be available online. The single-cell data have been deposited in the GEO under the accession number GSE304332. Raw files supporting our findings are available from the corresponding authors upon reasonable request.

Mice with Targeted Deletion of Socs3 in Neutrophils Exhibit Brain-Targeted EAE. Our previous work established that Socs3^ΔLysM mice develop severe btEAE [26], but Socs3 deletion in both neutrophils and macrophages precluded identifying the responsible cell type. To assess whether Socs3 deficiency in neutrophils induces brain-targeted neuroinflammation, mice with neutrophil-specific Socs3 deletion (Socs3^ΔLy6G) were generated. Effective gene deletion was confirmed by genotyping and reduced Socs3 mRNA levels (data not shown). Ly6G^+/− mice, used as controls, showed classical EAE scores similar to Socs3^fl/fl mice (Fig. 1A) and did not develop brain-targeted EAE (btEAE), consistent with previous findings [26]. Socs3^ΔLy6G mice developed both severe btEAE and cEAE (Fig. 1B–C), with scores and survival curves closely matching those of Socs3^ΔLysM mice (Fig. 1C–D). Antibody-mediated neutrophil depletion reduced btEAE severity (Fig. 1E) but did not affect cEAE (data not shown), indicating that neutrophils are required for the brain-targeted disease phenotype. Neutrophil-specific Socs3 deletion is sufficient to induce the mixed btEAE/cEAE phenotype previously observed in myeloid-specific knockout mice, establishing neutrophils as essential mediators of btEAE.

Cerebellar Demyelination and Immune Cell Infiltration in Socs3^ΔLy6G Mice.

Due to the neurological manifestations of btEAE, cerebellar pathology and immune cell infiltration were evaluated in Socs3^ΔLy6G mice. Black Gold staining demonstrated severe cerebellar demyelination at the peak of disease, with the myelinated area reduced to approximately 27% compared to 85% in controls (Fig. 2A–B). Flow cytometry confirmed increased total immune cell infiltration in the CB (Fig. 2C). There was an expansion of CD45^hiCD11b⁺ myeloid cells and a reduction in CD45^loCD11b⁺ microglia (Fig. 2D). Multiplex cytokine analysis of cerebellar homogenates revealed elevated G-CSF, IL-1α, IL-1β, CXCL2, TNF-α, GM-CSF, and CCL2 (Fig. 2E). These patterns indicate a pronounced pro-inflammatory environment. Socs3^ΔLy6G mice exhibit pathology characterized by demyelination and a substantial inflammatory infiltrate, demonstrating that neutrophil-specific Socs3 loss establishes a pathogenic immune environment in the cerebellum.

The activation state and functional properties of cerebellar neutrophils were investigated in Socs3^ΔLy6G mice. Flow cytometry showed CD11b upregulation and CD62L downregulation, while CXCR4 and CXCR2 levels stayed unchanged (Fig. 3A). Both the frequency and absolute number of cerebellar neutrophils increased compared to controls, which exhibited few neutrophils in the cerebellum (Fig. 3B–C). Socs3^ΔLy6G neutrophils showed increased ROS production (Fig. 3D) and higher mRNA expression of inflammatory mediators (Fig. 3E). In inflammatory and autoimmune diseases, neutrophils release NETs, exacerbating tissue damage during sustained inflammation [76–80]. Immunofluorescence for citrullinated histone H3 demonstrated clear NET formation in cerebellar lesions (Fig. 3F), with increased nucleosome release in plasma (Fig. 3G) and cerebellar tissue (Fig. 3H) of Socs3^ΔLy6G mice. Socs3 deficiency induces a hyperactivated, pro-inflammatory neutrophil phenotype characterized by NET formation, indicating that these cells are key effectors of CNS inflammation and demyelination.

Given that Socs3^ΔLy6G mice exhibited cEAE symptoms, spinal cord (SC) pathology was also assessed. Flow cytometry indicated increased total infiltrates (Supplementary Fig. 1A), with a rise in CD11b⁺ myeloid cells and fewer microglia (Supplementary Fig. 1B). Socs3^ΔLy6G neutrophil frequency and numbers increased in the SC compared to controls (Supplementary Fig. 1C–D). Also, nucleosome levels were higher in both Socs3^ΔLy6G and control EAE SCs compared to naïve mice (Supplementary Fig. 1E). Neutrophil activation and NET formation were observed in both SC and CB. This suggests that Socs3-deficient neutrophils contribute to CNS inflammation in multiple anatomical regions.

Th1 CD4⁺ T-cells Predominate in Socs3^ΔLy6G Mice.

To determine whether neutrophil-driven inflammation influences adaptive immunity, T-cell subsets were profiled in the CB and SC. Socs3^ΔLy6G mice exhibited increased total CD4⁺ T-cell numbers in the CB (Supplementary Fig. 2A). There was a significant elevation in IFN-γ–producing T helper 1 (Th1) cells (Supplementary Fig. 2B). A similar Th1 bias was observed in the SC (Supplementary Fig. 2C–D). Neutrophil-specific Socs3 deletion promotes a shift of CD4⁺ T-cells toward the Th1 phenotype, establishing a link between neutrophil activation and adaptive immune polarization in CNS autoimmunity.

scRNA-Seq Reveals Neutrophil Heterogeneity and Expansion of Inflammatory Subsets in Socs3^ΔLy6G Mice.

To study transcriptional heterogeneity in neutrophils, we performed scRNA-Seq on CNS-infiltrating cells from Socs3^ΔLy6G and control EAE mice. Cell clusters were annotated with top differentially expressed gene markers and canonical markers for neutrophils, macrophages, microglia, dendritic cells, T-cells and B-cells (Supplementary Fig. 3A). Unsupervised clustering found five neutrophil subsets (Neu1–Neu5) based on gene markers (Supplementary Fig. 3A–B).

Neu1 was a quiescent group enriched in naïve SC, while Neu2–Neu4 had inflammatory signatures (Fig. 4A–D). Neu2 expressed Saa3, Il1b, Id2, Chil3, Csf3r, Hist1h2bc, and Klhl6. Neu3 expressed Ccl3 and Ccl4. Neu4 showed an interferon and chemokine profile including Saa3, Il1b, Cxcl2, Cxcl3, Cd14, Ifit1, Isg20, and Isg15 (Fig. 4C–D and Supplementary Fig. 4A–B). The top 20 marker genes from each Neu cluster are shown in Supplementary Fig. 4A. In Socs3^ΔLy6G mice, Neu2 and Neu4 clusters expanded in both CB and SC, while Neu1 was nearly absent (Fig. 4A–B). The Neu5 cluster was detected at very low percentages in all four conditions (1–2%) (Fig. 4B). Gene set enrichment showed IFNα/γ response, IL-6–JAK–STAT3, TNFα–NF-κB, and general inflammatory pathways across Neu2-5, with Neu4 most enriched for these signatures (Fig. 4E). Collectively, transcriptomic analysis reveals inflammatory neutrophil subsets expanded by Socs3 loss, with increased STAT3- and NF-κB-driven gene programs likely driving CNS pathology.

Transcriptomic Analysis Identifies Saa3 as One of the Highly Upregulated Genes in Socs3^ΔLy6G Mice.

DEG analysis revealed the top 20 upregulated genes in the four conditions (Fig. 5A). Saa3, Id2, Ly6a, Ifi207, Ifi204, Fcgr4, Ctss, Ifitm1 and Isg15 were increased in the SC of Ly6G^+/− mice compared to the SC of naïve mice (Fig. 5A–B). Saa3, Id2, Chil3, Ly6a, Prnp, Entpd3, Cxcl2, Mmp19, Ifi207, Ifi204, Egr1, Acadl and Cd14 expression was elevated in the SC of Socs3^ΔLy6G mice compared to Ly6G^+/− mice (Fig. 5A–C). Ly6a, Entpd3, Klhl6, Ifi207 and Ifi204 were increased in the CB compared to the SC of Socs3^ΔLy6G mice (Fig. 5A–B). Strikingly, Saa3 was markedly upregulated in neutrophils from the SC and CB of Socs3^ΔLy6G mice compared to Ly6G^+/− (Fig. 5A–B). Cd14 and Cd274 expression was elevated in neutrophils from the SC and CB of Socs3^ΔLy6G mice (Fig. 5C).

Differential gene expression analysis identified Saa3 as one of the most highly induced genes in Socs3^ΔLy6G mice (Fig. 5A–C). Saa3, with Id2, Chil3, Cxcl2, Mmp19, Ifi204, and Cd14, was upregulated in both SC and CB. CB neutrophils had even higher Saa3 expression. Saa3 transcripts were especially high in the Neu2–Neu4 clusters (Fig. 4C and Supplementary Fig. 6). GSEA confirmed robust activation of inflammatory cytokine and interferon pathways associated with Saa3. Saa3 serves as a marker of Socs3-deficient inflammatory neutrophils and may mediate amplification of neuroinflammation.

To investigate intercellular communication underlying CNS inflammation, ligand–receptor analysis was conducted between neutrophils and macrophage/microglia [75]. Upregulation of Ccl4 in Neu2/Neu4 and Ccr1/Ccr5 in macrophages/microglia predicted enhanced myeloid recruitment (Supplementary Fig. 5A–C). Saa3 also increased in neutrophils and macrophages, while Tlr4 and Tlr2 rose in macrophages (Supplementary Fig. 5D–E, 5G). Saa3 can activate TLR2/TLR4 to boost cytokine production [81–84]. These results suggest neutrophil-derived Saa3 activates macrophages and initiates a self-reinforcing inflammatory circuit. Saa3 induces a strong up-regulation of Saa3 transcripts, indicating the self-amplifying potential of Saa3 [85]. Socs3-deficient neutrophils promote pro-inflammatory crosstalk with macrophages via Saa3–TLR2/TLR4 signaling, establishing a feed-forward loop that sustains CNS inflammation.

Saa3 Expression is Markedly Elevated in Socs3^ΔLy6G Mice.

Transcriptomic analysis confirmed substantial Saa3 upregulation in the SC and CB of Socs3^ΔLy6G mice relative to controls (Fig. 6A). Saa3 was most highly expressed in the Neu2, Neu3, Neu4, and Neu5 clusters, with the highest expression in the Neu4 cluster (Supplementary Fig. 6 and Fig. 4C–D), and was also expressed by macrophages and microglia (Supplementary Fig. 6). RT–PCR confirmed elevated Saa3 transcripts in neutrophils isolated from both CB and SC of Socs3^ΔLy6G mice (Fig. 6B–C). At the protein level, Saa3 expression was increased in CB and SC homogenates and plasma from both Socs3^ΔLy6G and Ly6G^+/− mice with EAE compared to their corresponding naïve mice (Fig. 6D–F). The systemic Saa3 rise shows that neutrophil-specific Socs3 deficiency induces both local and peripheral inflammation. Saa3 upregulation reflects widespread inflammatory activation following Socs3 deletion and may serve as a biomarker of disease activity.

To assess translational relevance, we analyzed serum from MS patients. Plasma SAA1, the human ortholog of mouse Saa3, was higher than in healthy controls (Fig. 7A). Circulating nucleosome levels, which mark NET formation, were also higher (Fig. 7B). These findings match reports of neutrophil hyperactivation in MS [39, 41, 86]. Increased SAA1 and NET markers in MS patients mirror the Socs3^ΔLy6G mouse phenotype. This links neutrophil-driven Saa3/SAA1 signaling to human CNS autoimmunity and points to therapeutic and diagnostic potential.

Neutrophils are increasingly recognized as active regulators of neuroinflammation rather than passive bystanders. Here, we demonstrate that loss of Socs3 in neutrophils alone (Socs3^ΔLy6G mice) drives a severe btEAE, which features cerebellar neutrophil infiltration, amplified inflammatory signaling, elevated ROS and NET formation, and extensive demyelination. Notably, depleting neutrophils completely prevented disease, establishing them as primary drivers of the brain-directed phenotype.

Single-cell transcriptomic analysis identified marked neutrophil heterogeneity within the CNS during EAE. We found five transcriptionally distinct clusters (Neu1–5) across naïve and diseased mice and observed expansion of the Neu2 and Neu4 populations in Socs3^ΔLy6G mice. The Neu4 cluster, which increased in both the SC and CB, showed enrichment for JAK/STAT and NF-κB signaling as well as interferon responses (IFNα, IFNγ). This transcriptional profile resembles interferon-stimulated neutrophil subsets previously observed in infection, autoimmunity, and chronic inflammation [87–91], pointing to a conserved inflammatory program engaged by Socs3 deficiency.

NETs contribute to the pathogenesis of MS [79]. NETs disrupt the BBB, promote leukocyte infiltration, and exacerbate demyelination [53, 92]. Consistent with this, cerebellar neutrophils from Socs3^ΔLy6G mice released abundant NETs, accompanied by elevated expression of interferon-stimulated genes and chemokines such as Cxcl2 (Fig. 5A–B), both of which promote NET formation [93, 94]. We also identified elevated circulating nucleosomes in the plasma of MS patients compared with healthy controls (Fig. 7B), further supporting the translational relevance of NETs as biomarkers and potential therapeutic targets.

A notable finding showed strong Saa3 activation in neutrophils from Socs3^ΔLy6G mice (Fig. 5A–B, 6A), with expression also seen in microglia and macrophages (Supplemental Fig. 6). Saa3 levels increased in SC and CB neutrophils from Socs3^ΔLy6G mice compared with controls, indicating a role in brain targeting. Since Saa3 signals through TLR2 and TLR4, both of which are upregulated in Socs3^ΔLy6G macrophages (Supplementary Fig. 5E, 5G), our data support a feed-forward inflammatory loop between neutrophils and myeloid cells. Previous studies found that SAA-deficient mice develop delayed or milder EAE [95], highlighting the pathogenic role of this pathway. Similarly, plasma SAA1 levels were higher in MS patients than in controls (Fig. 7A), linking the murine Saa3 axis to human disease.

Mechanistically, SOCS3 negatively regulates both NF-κB [96, 97], and JAK/STAT3 signaling [20, 22, 98], which are known drivers of Saa3 expression [99–103]. Therefore, loss of Socs3 removes these inhibitory controls, leading to persistent activation of these signaling pathways, upregulated Saa3 production, and amplification of inflammatory signals. SAA proteins, whose production is increased as a result, can cross and impair the intact BBB. Thus, the Socs3-deficient state promotes a mechanistic cascade from enhanced signaling and Saa3 production to disruption of the BBB and CNS pathology in MS.

Collectively, our findings designate neutrophils as active effectors of autoimmune neuroinflammation. Socs3 loss triggers neutrophil activation, ROS and NET release, and Saa3 upregulation, establishing a proinflammatory circuit that drives cerebellar demyelination. By resolving neutrophil heterogeneity and defining transcriptional programs in EAE, this study highlights the Saa3/SAA1 axis as both a biomarker and a potential therapeutic target for modulating CNS autoimmunity.