Microglial state transitions dominate the brain immune response in the subacute phase after cardiac arrest Authors: Lihong Dang, Ran Zhang, Jin Zhang, Peibang He, Xinyuan Yu, Ata Ur Rehman, Ángela del Águila, Vaibhav Jain, Jilin Tian, Emily Xu, Huaxin Sheng, Wei Yang Published: 2026 ABSTRACT Cardiac arrest (CA) is a life-threatening medical emergency, and most victims are elderly. Despite advances in resuscitation, post-CA morbidity and mortality remain high, and this is thought to result largely from brain injury in which neuroinflammation plays a key role. Yet, how individual immune cell populations contribute to the immune response in the post-CA brain is still poorly understood. Here, using single-cell RNA-sequencing (scRNA-seq), we revealed the first immune landscape of the young and aged brain on day 3 after CA. Our data demonstrate that transitions in microglial states constituted a dominant immune change in the post-CA brain. We identified 5 CA-associated microglial states that included 3 major clusters defined by pro-inflammatory signatures, a proliferative phenotype, and high Spp1 expression. These 3 states exhibited age-dependent differences: the inflammatory subset was markedly expanded in aged mice, whereas the proliferative and Spp1⁺ clusters were more prominent in young mice. Such divergent responses likely underpin age-related disparities in neurologic outcome after CA. Notably, Spp1+ microglia displayed unique spatiotemporal dynamics. These cells were robustly induced by CA, and were enriched in selective brain regions including the basal ganglia where they were associated with a microinfarct-like injury pattern. Lastly, we found that microglial depletion worsened functional outcome after CA, suggesting an overall protective role for microglia. Together, these findings provide novel insights into microglial heterogeneity and age-dependent immune responses in the brain after CA, and highlight the central role of microglia in shaping post-CA recovery.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12974-026-03874-4. Introduction Cardiac arrest (CA) is a life-threatening medical emergency, with most victims being 60 or older [1]. Although considerable advances in resuscitation have increased the number of CA patients who survive the initial arrest and are admitted to the intensive care unit, the rates of survival-to-discharge remain extremely poor. In fact, a mere 10.2% of patients are discharged from the hospital after an out-of-hospital CA (OHCA) event [2]. Further, OHCA has one of the highest rates of disability-adjusted life years (DALY), a measure of the overall burden of illness [3]. The high incidence of post-CA mortality and morbidity has been attributed primarily to brain injury. It has been estimated that approximately two-thirds of OHCA patients succumb to neurologic injury [4]. To date, besides timely resuscitation and therapeutic hypothermia, there is no effective treatment to prevent brain damage and improve neurologic recovery after CA. This dismal reality reflects our limited understanding of the pathophysiology of post-CA brain injury. One critical, yet poorly understood, aspect of this pathophysiology is neuroinflammation in the post-CA brain [5]. We and others have investigated the immune response in both peripheral organs and brain using experimental CA models [6–9]. In animals, various pro-inflammatory cytokines, including interleukin (IL)-1β, IL6, and tumor necrosis factor (TNF)-α, increase in the blood during the acute phase after CA, consistent with clinical observations [10]. Interestingly, our data also revealed a subsequent profound immunosuppressive phase following the initial inflammatory surge. This phase is characterized by massive loss of lymphocytes (lymphopenia) and impaired function of the remaining lymphocytes, [6-8] potentially rendering CA survivors prone to infections [11]. Critically, only a handful of experimental CA studies on the immune response has sought to understand the role of neuroinflammation in brain damage [7,9,12–18]. Data have shown infiltration of peripheral immune cells into the brain after CA [7-12]. We also know that microglia are activated after CA, as evidenced by increased Iba1 staining and morphologic changes [6,17]. However, the role of these immune changes in brain injury after CA remains largely unknown. This knowledge gap hinders our ability to develop treatments that target the immune response in the brain after CA. Thus, a critical next step is to uncover the immune landscape of the post-CA brain and then, leveraging such knowledge, begin to dissect the effect of specific immune cell populations in the brain on CA outcome. To this end, single-cell RNA sequencing (scRNA-seq) is well suited. Indeed, this technique has been increasingly used to dissect the immune cell populations and their potential functions in various brain disorders, including stroke [19,20]. Notably, two recent single cell transcriptomic studies analyzed hippocampus tissue up to 24 h after CA;[16-18] however, both studies included all brain cell types, rather than focusing on the immune cells. Moreover, data have shown that neuroinflammation is more prominent on day 3 after CA. [7-17] Therefore, in the current study, we performed the first scRNA-seq analysis of sorted immune cells from the brain during the subacute phase, i.e., on post-CA day 3. Given the critical impact of aging on CA outcome, both young and aged mice were used for the analysis. Our data identified notable differences in the immune response between the young and aged brains after CA. Notably, the post-CA brain contained relatively few infiltrating immune cells, but exhibited marked changes in microglial states. We further confirmed and analyzed a major CA-induced microglial subset, which was characterized by high expression of secreted phosphoprotein 1 (Spp1). Interestingly, these Spp1+ microglia were largely confined to selective brain regions: the hippocampus and basal ganglia, in particular. Finally, the net effects of microglial depletion on CA outcome suggested an overall protective role for microglia in CA. Methods Animals All animal procedures were approved by the Duke University Medical Center Animal Care and Use Committee, and were conducted in accordance with the United States Public Health Service’s Policy on Humane Care and Use of Laboratory Animals. The animals were maintained in a 14/10 light/dark cycle, and both water and food were provided ad libitum. Young (3–6 months old) C57Bl/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME), and aged (18–20 months old) male C57Bl/6 mice were supplied by the NIA Aged Rodent Colonies program. Tmem119-Ai6 mice in which microglia are specifically labeled with ZsGreen fluorescent protein were generated by crossbreeding Ai6 (RCL-ZsGreen; JAX #007906) reporter mice with Tmem119-2A-CreERT2 mice (JAX #031820).[19] Spp1GFPfl (JAX #038709) mice were used to track Spp1-expressing cells in the brain, as in this mouse line, the EGFP reporter is controlled by endogenous Spp1 promoter [21]. CA surgery A KCl-mediated CA mouse model was performed, as described previously [8]. Briefly, mice were orally intubated and mechanically ventilated with a mixture of 1.5%-1.7% isoflurane in 30% oxygen balanced with nitrogen using an automatic small animal ventilator (RoVent; Kent Scientific). Body and brain temperatures, as well as electrocardiogram (ECG), were continuously monitored throughout the procedure. When the body temperature reached 37.0 °C ± 0.2 °C, the KCl solution was administered via the jugular vein to induce CA. Immediately, ventilation and body heating were discontinued. During the CA period, the pericranial temperature was maintained at 38.5 °C ± 0.5 °C (for young mice) or 38.0 °C ± 0.5 °C (for aged mice) using a water-heated coil. CA was maintained for 8.5 min in young mice and 8 min in aged mice, after which cardiopulmonary resuscitation (CPR) was initiated with an epinephrine infusion, ventilation with 100% oxygen, and chest compressions. If return of spontaneous circulation (ROSC) was not achieved within 5 min, CPR was terminated, and the animal was excluded. Once spontaneous breathing stabilized, mice were placed in an incubator at 32 °C for 2 h. Sham-operated mice underwent the same surgical procedures without the experimental intervention. Functional outcomes Behavioral tests, as detailed previously, [22] were performed by an experimenter blinded to group assignments. Briefly, neurologic scoring was used to evaluate general functional deficits, with scores ranging from 9 points (normal function) to 0 points (severe injury). The open field test was used to measure spontaneous locomotor activity. Mouse activity in a chamber (50 cm x 50 cm x 50 cm) was recorded for 10 min and analyzed by Topscan software (CleverSys, USA). Grip strength of the 4 limbs of each mouse was measured using a grip strength meter (Ugo Basile, Italy). Mice were placed on a grid-shaped grasping tool, and their tails were grabbed and pulled straight backward. After the animal released the grasping tool, the preamplifier automatically recorded the maximum pulling force and displayed it on the screen. The pulling force was measured for each mouse 3 times with at least 10 min of recovery between measurements. The maximum of the 3 measurements was used for further analyses. Immunofluorescence and microscopy Mice were perfused with 30 mL of ice-cold saline followed by 30 mL of 4% paraformaldehyde (PFA). Brains were post-fixed in 4% PFA at 4 °C for 24 h and subsequently immersed in 30% sucrose solution at 4 °C until they sank. Coronal brain sections (35 μm thick) were prepared using a cryostat. Sections were washed and incubated in blocking buffer (1 x PBS containing 5% heat-inactivated goat serum, 2% BSA, and 0.5% Triton X-100) to permeabilize the tissue and block non-specific binding. The sections were then incubated with primary antibodies against Iba1 (Fujifilm, #019-19741), GFAP (DAKO, #Z0334), and MAP2 (Invitrogen, #PA1-16751) at 4 °C overnight, followed by 2-hour incubation with secondary antibodies at room temperature. Images were acquired on an Axio Imager upright microscope and a Leica Stellaris 8 FALCON confocal microscope. Image analysis was performed using either Imaris (Version 9.9.0, Oxford Instruments) or ImageJ software (NIH). For all imaging studies, the individual performing image acquisition and analysis was blinded to experimental groups. Assessment of microglial morphology Brain sections (35 μm) were collected from Spp1GFPfl mice on day 3 after CA and were stained for Iba1. For each animal, 2 brain sections were imaged using confocal z-stacks (0.5 μm step size) with a 40× oil-immersion objective. The acquired images were subjected to 3D reconstruction using Imaris. Reconstruction of microglial cell bodies and dendrites was performed using the Surfaces tool. The Surfaces module was used to quantify microglial morphologic parameters, including cell body area. Iba1+Spp1+ and Iba1+Spp1− cells were randomly selected from regions that showed increased Iba1 signal. A total of 40 cells per brain section were analyzed from each animal. BrdU labeling and immunostaining To assess microglial proliferation, mice were administered 5-Bromo-2’-deoxyuridine (BrdU, 40 mg/kg, intraperitoneally; Sigma-Aldrich) once daily for 5 consecutive days prior to CA. After CA, BrdU injections were continued daily for 2 days. For BrdU detection, 25-µm-thick brain sections were denatured in 2 N HCl (37 °C, 30 min). Sections were then incubated with BrdU (BioLegend, #317902) and Iba1 (Fujifilm, #019-19741) primary antibodies overnight at 4 °C, followed by fluorescent secondary antibody labeling and DAPI counterstaining. Images were captured using a fluorescence microscope. Fluoro-Jade C (FJC) staining FJC (Biosensis, #TR-160-FJC) staining was performed according to the manufacturer’s protocol. Briefly, coronal brain sections (25 μm) were mounted onto gelatin-coated slides and dried overnight at 38 °C. The slides were first immersed in a solution of 1% sodium hydroxide in 80% ethanol for 5 min, followed by sequential rinses in 70% ethanol and distilled water. To minimize background interference, sections were incubated in 0.06% potassium permanganate for 10 min. Subsequently, the sections were incubated with a 0.0001% FJC working solution (containing 0.1% acetic acid) for 10 min in the dark, with DAPI included for nuclear counterstaining. Cell quantification was performed in a defined subcortical region of the left hemisphere, where FJC-positive cells were manually counted using ImageJ by an investigator blinded to the experimental conditions. Flow cytometry Our standard protocols for flow cytometry analysis were followed [8,19,23]. Briefly, brains were harvested and gently dissociated in HBSS buffer using a Dounce homogenizer. The resulting suspension was filtered through 70-µm cell strainers and centrifuged at 450 × g for 6 min at 4 °C. The cell pellet was then resuspended in 35% Percoll (Cytiva) and centrifuged at 800 × g for 20 min at 4 °C (acceleration: 3; brake: 3). After washing, cells were resuspended in FACS buffer (RPMI-1640 containing 2% FBS and 2 mM EDTA). Flow cytometric staining was conducted as previously described [8,19,23]. Briefly, 30 µL of the antibody mixture was added to the cell suspension and incubated for 25 min at 4 °C in the dark. Subsequently, 1 µL of 7-AAD was added per sample to stain dead cells for 5 min. Cells were then washed and resuspended in 200 µL of FACS buffer for acquisition using an Aurora flow cytometer (Cytek). Data were analyzed using SpectroFlo (Cytek) and FlowJo (BD Biosciences) software. Antibodies used are listed in Table S1. Phagocytosis assay Phagocytic activity was assessed in vitro using the Phagocytosis Detection Kit (BioLegend, #421938) according to the manufacturer’s instructions. Briefly, phagocytosis detection beads in working solution were added to the enriched brain immune cell suspension and incubated at 37 °C for 1 h. Cells were then washed twice and resuspended in 200 µL of FACS buffer. Subsequently, cells were stained with anti-Tmem119–PE-Cy7 and 7-AAD before flow cytometry analysis (Cytek). ELISA Plasma SPP1 levels were measured using a DuoSet ELISA kit (R&D Systems, #DY441) following manufacturer’s instructions. Blood samples were centrifuged at 2,000 × g for 10 min at 4 °C to collect plasma. Assessment of cytokines and chemokines Cytokine and chemokine levels in the brain were assessed using a Proteome Profiler Mouse Cytokine Array (R&D, ARY006, Minneapolis, MN, USA), which detects 40 analytes [22]. Briefly, brain tissue was homogenized in RIPA lysis buffer supplemented with a protease inhibitor cocktail, followed by centrifugation at 14,000 × g for 15 min at 4 °C. Protein concentration was measured using the BCA method. Equal amounts of protein from three mice were pooled for each sample, and 300 µg of total protein per sample was used for analysis, following the manufacturer’s instructions. Membranes were imaged with an Amersham Imager 680 (GE), and signal intensities were quantified using ImageJ software (NIH). Quantitative PCR Total RNA was extracted from brain tissues using TRIzol reagent (Invitrogen, Gaithersburg, MD, USA), and cDNA was synthesized using the ProtoScript II First Strand cDNA Synthesis Kit (New England Biolabs, E6560), according to the manufacturer’s instructions. Quantitative PCR (qPCR) was performed using iTaq Universal SYBR Green Super Mix kit (Bio-Rad, #1725124). Relative fold changes in gene expression were calculated using the comparative CT (2 − ΔΔCt) method with GAPDH as the internal control. Primer sequences are listed in Table S2. Microglial depletion PLX3397 (MedChemExpress, HY-16749) was mixed with AIN-76 A rodent diet powder (Bio-Serv, F0761) at a dosage of 300 mg/kg per day. Mice (3–4 months old) were fed AIN-76 A chow with or without PLX3397 for 10 days before surgery. Single-cell RNA sequencing (scRNA-seq) analysis scRNA-seq was carried out at the Duke Molecular Genomics Core (MGC) following standardized protocols. The overall workflow for sample preparation, sequencing, and data analysis was largely consistent with our previous studies [19,20]. To prepare single-cell suspensions from brain tissue, we used a simplified protocol established in our laboratory [23]. Briefly, after removal of the meninges, brains were collected and mechanically homogenized in cold HBSS using a Dounce homogenizer. The homogenate was filtered and centrifuged, and the resulting cell pellet was resuspended in 35% Percoll (Cytiva). After centrifugation, the upper myeline layer was carefully removed, and the cell pellet was washed twice. Cells were then resuspended in FACS buffer. For each group, cells from 3 mice were pooled and then stained with CD45-FITC antibody and 7-AAD. All live CD45+cells were sorted on a Sony SH800S cell sorter. FACS-isolated cells were then delivered to the Duke MGC for library preparation and sequencing. Library preparation and sequencing were performed as previously described [19,20]. In brief, single-cell gel bead-in-emulsions (GEMs) were generated using the Chromium Controller (10x Genomics, Pleasanton, CA, USA), which encapsulate individual cells with barcoded gel beads and reverse transcription reagents. After reverse transcription within the GEMs, single-stranded cDNA was isolated using DynaBeads MyOne Silane Beads (Thermo Fisher Scientific), followed by amplification and purification. Sequencing libraries were then constructed using the Chromium Single Cell 3’ v3.1 Library Kit, following the manufacturer’s instructions. Final libraries were quantified and loaded onto an Illumina NovaSeq 6000 system (Illumina, San Diego, CA, USA) for high-throughput sequencing. Bioinformatics processing and data analysis Data analysis was performed as described previously [19,20]. Briefly, raw sequencing data were initially processed using the Cell Ranger pipeline v7.1.0 (10x Genomics) to generate the gene-barcode matrix. Subsequent quality control, normalization, and downstream analyses were performed in R (v4.3.2) using the Seurat package (v5.1.0). Cells with fewer than 250 detected genes, more than 5,000 genes, or > 20% mitochondrial gene content were excluded to remove potentially low-quality cells and doublets. Genes expressed in fewer than 5 cells were also filtered out. The remaining high-quality cells were then normalized and log-transformed using Seurat’s Normalize Data () function. Highly variable genes were identified using Find Variable Features (), and the data were scaled and subjected to principal component analysis (PCA). Dimensionality reduction was performed using Uniform Manifold Approximation and Projection (UMAP) based on the top 30 principal components. Cells were clustered using a graph-based clustering approach (Find Neighbors () and Find Clusters () functions), with the resolution parameter adjusted to identify distinct cell populations. To identify cluster-specific marker genes, differential gene expression analysis was conducted using the Wilcoxon rank-sum test (Find All Markers () function in Seurat). Genes with an adjusted p-value < 0.05 and |log₂ fold change| > 0.25 were considered significantly differentially expressed. For functional interpretation, gene ontology (GO) and pathway enrichment analyses were performed using tools such as cluster Profiler, fGSEA, and or Metascape, focusing on biological processes (BP) relevant to the identified cell clusters. The dataset has been deposited in NCBI Gene Expression Omnibus (GEO) database repository with the accession number GSE326906. Bulk RNA-seq Bulk RNA-sequencing was performed at the Duke MGC. Enriched brain immune cells were collected using our simplified protocol, as described above. Cells were resuspended in FACS buffer, followed by staining with anti-CD45, anti-Tmem119, and 7-AAD. Live CD45+Tmem119⁺ cells were sorted using a Sony SH800S sorter and collected by centrifugation. The cell pellet samples were then delivered to the Duke MGC for RNA preparation, library preparation, and sequencing, following established protocols. RNA-seq data were processed using the fastp toolkit to trim low-quality bases and sequencing adapters from the 3’ end of reads. Cleaned reads were then mapped to GRCh38 (version 106) using the STAR RNA-seq alignment tool, and reads aligning to a single genomic location were summarized across genes. For genes having an overlap of at least 10 reads, gene counts were normalized and differential expression analysis was carried out using the DESeq2 Bioconductor package in R. Consistent with the recommendation of the DESeq authors, independent filtering was applied prior to calculating adjusted p-values and moderated log2 fold-changes were derived using the ashr package. Gene set enrichment analysis (GSEA) was performed to identify GO terms and pathways associated with altered gene expression across experimental comparisons. Differentially expressed genes (DEGs) were further analyzed for GO enrichment to assess the biological processes underlying transcriptional changes between experimental groups. Statistical analysis All statistical analyses were performed using GraphPad Prism 10. Sample sizes were determined based on previous data. Statistical differences between groups were evaluated by unpaired Student’s t-test and Mann-Whitney U test (for neurologic scores). Data were presented as the median or mean ± SD. P < 0.05 was considered statistically significant. Results scRNA-seq analysis revealed marked changes in microglial states in the post-CA brain Using our murine KCl-induced CA model, we have demonstrated that CA/CPR results in neuronal death, kidney injury, and neurologic deficits, [9,24,25] effectively recapitulating clinical features observed in CA patients. In this study, we applied our standard 8.5-minute CA for young mice, but reduced CA duration to 8 min for aged mice, due to their higher mortality rates. We first examined microglial activation as a well-established indicator of neuroinflammation (Fig. S1). In our previous work, we showed that Tmem119-Ai6 reporter mice can be used to visualize microglial activation through an increase in ZsGreen fluorescence [19]. As expected, a robust increase in ZsGreen signal was observed in the hippocampus regions on day 3 after CA (Fig. S1A), confirming neuroinflammation in the post-CA brain. Notably, Iba1 immunostaining revealed significantly stronger signals in the hippocampus of aged brains compared to young brains after CA (Fig. S1B), suggesting an age-dependent difference in neuroinflammatory response. To gain a more in-depth understanding, we set out to perform a systematic and detailed analysis of the immune cells in the post-CA brain using scRNA-seq (Fig. 1A). Fig. 1Single-cell RNA sequencing (scRNA-seq) analysis of the immune cells in the brain of young and aged mice after cardiac arrest. A) Scheme of the experimental design for scRNA-seq analysis. Young and aged mice (n = 4/group) were subjected to cardiac arrest (CA) or sham surgery. Three days later, brains were collected for CD45+ cell sorting and scRNA-seq analysis. B) UMAP plot of aggregated data from all groups. C) Split view of UMAP plots of sham and CA groups. D) Dot plot showing the scaled expression of selected genes in each cluster. Signature genes are highlighted with a rectangle box. Dot size reflects the percentage of cells within the cluster expressing each gene, and color intensity depicts the expression level Single-cell RNA sequencing (scRNA-seq) analysis of the immune cells in the brain of young and aged mice after cardiac arrest. A) Scheme of the experimental design for scRNA-seq analysis. Young and aged mice (n = 4/group) were subjected to cardiac arrest (CA) or sham surgery. Three days later, brains were collected for CD45+ cell sorting and scRNA-seq analysis. B) UMAP plot of aggregated data from all groups. C) Split view of UMAP plots of sham and CA groups. D) Dot plot showing the scaled expression of selected genes in each cluster. Signature genes are highlighted with a rectangle box. Dot size reflects the percentage of cells within the cluster expressing each gene, and color intensity depicts the expression level We initially aimed to address 2 key questions regarding the immune response in the brain after CA: (1) the immune landscape in the post-CA brain and (2) its differences between young and aged brains. We focused on post-CA day 3, as this time point has been shown to be crucial for studying CA-induced immune changes in the brain [7]. On day 3 after CA or sham surgery, we enriched CD45+ cells from young and aged mouse brains by FACS, following our established protocol with all cell preparation steps carried out on ice or at 4 °C [19,26]. To minimize biological and experimental variations, cells from 4 mice were pooled for each sample. We sequenced a total of 9038 (Young_Sham), 10,794 (Young_CA), 7700 (Aged_Sham), and 6794 (Aged_CA) cells. Single-cell data from young and aged groups are comparable across major quality control metrics (Fig. S2). For data analysis, we excluded Ptprc (CD45)-negative cells and implemented various quality control steps, including filtering based on the percentage of mitochondrial genes and removing doublets. The aggregate data were first used for unsupervised clustering, which identified 15 transcriptionally distinct clusters, as visualized in the uniform manifold approximation and projection (UMAP) plot (Fig. 1B). The top 50 DEG markers of each cluster are listed in Table S3. Based on these marker genes, the clusters were annotated into 6 cell types. Of note, cluster 15 appeared to be a mixture of different cell types, with expression of some neuronal genes (e.g., Dlg2). Since cluster 15 cannot be clearly defined, we designated it as “Other”. As expected, the predominant cell population consisted of microglia with 9 different subsets (MG1-9; clusters 1–9). We also identified T cells (TC; cluster 10), B cells (BC; cluster 11), monocytes/macrophages (MM; cluster 12), neutrophils (Neut; cluster 13), and border-associated macrophages (BAM; cluster 14) [27]. The CA-induced proportional changes for each cluster are illustrated in the split UMAP plots (Fig. 1C). Selected marker genes for each cluster are depicted in Fig. 1D. Compared to other acute brain injuries, such as ischemic stroke and traumatic brain injury, there are considerably fewer infiltrating immune cells in the post-CA brain, due to milder cell death and less blood-brain barrier (BBB) damage [7]. Consistently, our scRNA-seq data revealed a small number of non-microglial immune cells in the post-CA brain. Of these cells, the only notable change was a percentage increase in monocytes/macrophages after CA (Fig. 2A). In contrast, 5 microglial clusters (MG5-9) showed an increase in percentages in the post-CA brain, indicating that these subsets represent CA-associated microglial states (Fig. 2A). Analysis of DEG markers for MG5-9 identified relatively specific signature genes for each cluster, except MG5 (Fig. 2B). Specifically, MG6 exhibited strong Apoe expression, whereas MG7 was characterized by high Spp1 expression and was thus referred to as the Spp1+ cluster. MG8 displayed upregulated expression of many chemokine and cytokine genes (e.g., Ccl2, Ccl12, and Tnf), indicative of a proinflammatory state. MG9 clearly represented a population of actively dividing microglia, as evidenced by highly-presented proliferation related genes (e.g., Mki67, Top2a, and Cdk1). Overall, changes in microglial states, rather than infiltration of peripheral immune cells, appear to constitute the dominant immune response in the brain on day 3 after CA. Fig. 2Differences in the immune response between young and aged brains after cardiac arrest revealed by scRNA-seq. A) Percent changes in individual cell clusters on day 3 after cardiac arrest (CA) vs. sham. The highlighted clusters (MG5-9) represent CA-associated microglial subsets. B) Heatmap showing the top 10 upregulated genes in CA-associated microglial clusters. C) Split view of UMAP plots of 4 groups. Major clusters exhibiting age-dependent differences after CA are highlighted, including MG7-9 and MM. D) Top 10 enriched GO BP terms of differentially expressed genes (DEGs) between sham and CA. The horizontal axis showed NES (normalized enrichment score). Color intensity represents p-value. The bubble size reflects the number of genes enriched in a certain GO term. E) Volcano plots showing DEGs from different comparisons. The gene selection criteria are Log2FC > 0.585 or < -0.585 and padj < 0.01. The 2 vertical dot lines represent fold changes of -1.5 and 1.5 Differences in the immune response between young and aged brains after cardiac arrest revealed by scRNA-seq. A) Percent changes in individual cell clusters on day 3 after cardiac arrest (CA) vs. sham. The highlighted clusters (MG5-9) represent CA-associated microglial subsets. B) Heatmap showing the top 10 upregulated genes in CA-associated microglial clusters. C) Split view of UMAP plots of 4 groups. Major clusters exhibiting age-dependent differences after CA are highlighted, including MG7-9 and MM. D) Top 10 enriched GO BP terms of differentially expressed genes (DEGs) between sham and CA. The horizontal axis showed NES (normalized enrichment score). Color intensity represents p-value. The bubble size reflects the number of genes enriched in a certain GO term. E) Volcano plots showing DEGs from different comparisons. The gene selection criteria are Log2FC > 0.585 or < -0.585 and padj < 0.01. The 2 vertical dot lines represent fold changes of -1.5 and 1.5 Differences in the immune landscape between young and aged brains after CA Next, we split the dataset to examine the effect of aging on immune cell changes in the brain. We found that although all clusters are present in both young and aged brains, there are notable differences in percentages and their changes (Fig. S3), especially for lymphocytes (TC and BC), neutrophils (Neut), monocytes/macrophages (MM), and microglial clusters MG7-9. The brain has long been considered immune-privileged, with immune cells rarely found in the parenchyma except for resident microglia. However, as the brain ages, immune cells begin to infiltrate the brain and become substantially present even in the absence of injury. Our scRNA-seq data confirmed that microglia were nearly the only immune cells in young sham brains, whereas aged sham brains had a noticeable number of other immune cells, including T cells, B cells, and neutrophils (Fig. S3A). After CA, these non-microglial immune cells did not show a percentage increase in either young or aged brains, based on our scRNA-seq analysis (Fig. S3B). However, monocytes/macrophages (MM) increased from 0.5% to 4.7% in the young brain after CA, indicating their infiltration. Interestingly, this increase was less evident in the aged brain (Fig. S3B). With respect to microglia, MG7-9 clusters were minimally present in control young brains, but markedly increased after CA (Fig. S3B). Thus, we focused our subsequent studies on these 3 microglial clusters. Of note, induction of the inflammatory cluster (MG8) was more pronounced in aged mice, whereas induction of the Spp1+ (MG7) and proliferative (MG9) clusters was more pronounced in young mice (Fig. 2C, S3B). Next, we performed pathway and DEG analyses (Fig. 2D, E and Table S4). Comparing sham with post-CA brains, pathways related to proliferation were more enriched in young brains, whereas type I interferon (IFN) pathways were more prevalent in aged brains, consistent with the major expansion of the inflammatory MG8 microglial subset in aged brains (Fig. 2D). Interestingly, Ccl5, a key chemokine for T cell recruitment, was the most upregulated gene in the aged vs. young brain under sham conditions (Fig. 2E). In young brains, the most CA-upregulated gene was Spp1, which aligned with the infiltration of monocytes and the increase in the Spp1+ microglial cluster (MG7). Notably, in aged brains, the top CA-upregulated genes were chemokines, including Cxcl10 and Ccl2, despite minimal immune cell infiltration (Fig. 2E). Consistent with microglial responses observed after brain disorders, homeostatic microglial genes (e.g., P2ry12, Cx3cr1, and Siglech) were downregulated after CA, reflecting microglial activation (Table S4). To validate the scRNA-seq findings related to cell composition, we performed flow cytometry analysis. On day 3 after CA, brains were collected from young and aged mice, and the major immune cell types were quantified (Fig. 3). Cells expressing high levels of CD45 (CD45High) were considered primarily infiltrating immune cells. Consistent with our scRNA-seq results, both total infiltrating immune cells (CD45High) and T cells (CD45HighCD11b−CD3+) were significantly more abundant in control aged brains compared to young brains. After CA, the number of CD45High cells increased in young brains, but remained largely unchanged in aged brains, suggesting less immune cell infiltration. Notably, NK cells were detected but with a small cell number, which may explain their absence in our scRNA-seq analysis. Overall, these results largely corroborated the scRNA-seq data. Fig. 3Flow cytometry analysis of the immune cells in young and aged brains after cardiac arrest. Young and aged mice were subjected to cardiac arrest (CA) or sham surgery. On day 3 after CA, brains were collected and analyzed by flow cytometry. A) Representative plots illustrate the difference in the CD45High population between young and aged brains. The gating strategy depicts the following immune cell populations: CD45High, NK cells (CD45HighCD11b+NK1.1+), neutrophils (CD45HighCD11b+Ly6G+), T cells (CD45HighCD11b−CD3+), and B cells (CD45HighCD11b−CD19+). B) Quantitative data shown in the bar graphs. The data are presented as mean ± SD. n = 4 or 5/group; *, p < 0.05 Flow cytometry analysis of the immune cells in young and aged brains after cardiac arrest. Young and aged mice were subjected to cardiac arrest (CA) or sham surgery. On day 3 after CA, brains were collected and analyzed by flow cytometry. A) Representative plots illustrate the difference in the CD45High population between young and aged brains. The gating strategy depicts the following immune cell populations: CD45High, NK cells (CD45HighCD11b+NK1.1+), neutrophils (CD45HighCD11b+Ly6G+), T cells (CD45HighCD11b−CD3+), and B cells (CD45HighCD11b−CD19+). B) Quantitative data shown in the bar graphs. The data are presented as mean ± SD. n = 4 or 5/group; *, p < 0.05 MG7 and MG9 microglial subsets In our scRNA-seq dataset, the most prominent change in microglial states after CA was a marked increase in Spp1+ microglia (MG7), especially in young brains where their proportion rose from 1.5% to 11.2% (~ 7.5-fold increase). Other signature genes in this cluster included Fn1, CD63, and Msr1 (Fig. 2B and Fig. 4A). A similar cluster was identified in our previous scRNA-seq study of the stroke brain;19 thus, this microglial state may represent a common response to brain ischemia. Notably, in this cluster, two highly upregulated genes CD63 and Msr1 are involved in the phagocytic process. Consistently, Metascope analysis of the top 50 DEGs from this cluster revealed significant enrichment in pathways related to the lysosome and phagosome (Fig. 4B). This functional role is also supported by our data below. Fig. 4MG7 and MG9 clusters. A) Feature plots showing 3 top highly expressed genes in MG7. B) Top enriched pathways in MG7. C) Feature plots showing 3 top highly expressed genes in MG9. D) Top 10 enriched GO BP terms in MG9. E, F) Representative immunofluorescence images of BrdU (green) and Iba1 (red) co-staining in the hippocampus (E) and basal ganglia (F) at 3 days post-CA. Arrowheads indicate colocalization of BrdU and Iba1 signals MG7 and MG9 clusters. A) Feature plots showing 3 top highly expressed genes in MG7. B) Top enriched pathways in MG7. C) Feature plots showing 3 top highly expressed genes in MG9. D) Top 10 enriched GO BP terms in MG9. E, F) Representative immunofluorescence images of BrdU (green) and Iba1 (red) co-staining in the hippocampus (E) and basal ganglia (F) at 3 days post-CA. Arrowheads indicate colocalization of BrdU and Iba1 signals MG9 constitutes a proliferative population of microglia. Microglia are known to proliferate locally in the brain after depletion or injury, which is crucial for replenishing lost microglia or clearing cellular debris. Our scRNA-seq data reliably identified the MG9 cluster, which showed many upregulated genes that are involved in the cell cycle, such as Mki67, Top2a, Cenpp, and Cdk1 (Fig. 2B and Fig. 4C). Consistently, the top enriched pathways of this cluster were predominantly related to cell division (Fig. 4D). Moreover, BrdU labeling analysis confirmed the presence of proliferating microglia enriched in selective regions, including the hippocampus and basal ganglia on day 3 after CA (Fig. 4E, F). Interestingly, this cluster was largely absent in the aged brain after either sham or CA surgery, suggesting a deficiency in the proliferative capacity of aged microglia. Of note, even in post-CA young mice, this cluster accounted for only 2.7% of the total microglial population, a much smaller fraction compared to its induction in the post-stroke brain [19]. MG8 microglia exhibited a pro-inflammatory state Major upregulated genes in the MG8 cluster included Cxcl10, Ccl2, Ccl2, Ccl4, Ccl12, and Tnf (Fig. 5A and Table S3), defining a pro-inflammatory state relative to other microglial clusters. Moreover, pathway analysis suggested that MG8 was involved in the type I IFN response (Fig. 5B). Importantly, this cluster appears largely age-related, because its frequency in the aged brain was already higher in the sham group (3.3% vs. 1.7%, aged vs. young); and after CA, its increase was more pronounced (14.5% vs. 5.2%, aged vs. young) (Fig. S3). In fact, MG8 was the most abundant microglial cluster in the aged brain after CA. Since this cluster may critically contribute to worse outcome in aged mice and scRNA-seq analysis only reflects transcriptomic levels, we further measured protein levels of immune mediators (e.g., cytokines and chemokines) in young and aged brains after sham or CA surgery, using a cytokine array. As shown in Fig. 5C, D, one obvious difference between groups was the elevated levels of nearly all immune mediators examined in the aged brains of both sham and CA mice. As expected, many of these mediators increased in the young brain after CA, including CCL2, CCL5, IL-1β, IL-1ra, and ICAM1. In contrast, while some immune mediators in the aged brain (e.g., CCL2, CXCL10, and CXCL13) increased, others, such as CXCL2, CXCL12, and G-CSF, decreased, but were still at levels comparable to or higher than those in the young brain. We further performed qPCR analysis and confirmed changes in several of these immune factors (Fig. 5E). Higher levels of chemokines in the aged brain seemed to contradict the observation of fewer infiltrating immune cells in aged vs. young brains after CA. We then examined the effects of AMD3100 in young and aged mice. AMD3100 is a CXCR4 antagonist widely used to promote bone marrow cell egress [28]. Our data indicated that immune cells in the blood increased similarly in both groups in response to AMD3100 treatment (Fig. S4). Thus, the observed age-related difference in the number of infiltrating immune cells in the post-CA brain may not be due to impaired immune cell mobilization in aged mice. Nonetheless, our data together demonstrated that the presence of pro-inflammatory MG8 microglia and the changes in these microglia after CA constitute a key age-related difference in the microglial response to CA, and that these cells likely play a major role in driving elevated cytokine and chemokine levels in the post-CA brain. Fig. 5MG8 cluster. A) Feature plots showing 4 top highly expressed genes in MG8. B) Top 10 enriched GO BP terms in MG8. C, D) Analysis of cytokines and chemokines in brain lysates. Young and aged mice were subjected to cardiac arrest (CA) or sham surgery. On day 3 after CA, hippocampus tissues were collected and analyzed using a cytokine array. C) Original membrane blots. Highlighted are cytokines and chemokines showing changes between groups. D) Heatmap analysis of signal intensities measured from the blots. E) Quantitative PCR analysis of selected cytokines and chemokines. The data are presented as mean ± SD. n = 4 or 5/group; *, p < 0.05 MG8 cluster. A) Feature plots showing 4 top highly expressed genes in MG8. B) Top 10 enriched GO BP terms in MG8. C, D) Analysis of cytokines and chemokines in brain lysates. Young and aged mice were subjected to cardiac arrest (CA) or sham surgery. On day 3 after CA, hippocampus tissues were collected and analyzed using a cytokine array. C) Original membrane blots. Highlighted are cytokines and chemokines showing changes between groups. D) Heatmap analysis of signal intensities measured from the blots. E) Quantitative PCR analysis of selected cytokines and chemokines. The data are presented as mean ± SD. n = 4 or 5/group; *, p < 0.05 Bulk RNA-seq analysis revealed attenuated transcriptomic responses in aged vs. young microglia after CA Given the central role of microglia in the immune response of the post-CA brain, we performed bulk RNA-seq on FACS-sorted microglia from young and aged brains (Fig. 6A). This approach complements scRNA-seq, as bulk RNA-seq is better suited to detect changes in overall gene expression levels. After validating Tmem119 antibody (Fig. S5A), we sorted CD45+Tmem119+ cells from young and aged brains on day 3 after sham or CA surgery for RNA-seq analysis. Each sample was pooled from 2 mice, with 2 samples sequenced per group. We first examined the expression of microglia lineage genes, confirming microglial enrichment (Fig. S5B). We also validated the upregulation of several aging-related genes in our sorted aged microglia (Fig. S5C). We then performed 3 comparisons (Aged_Sham vs. Young_Sham, Young_CA vs. Young_Sham, and Aged_CA vs. Aged_Sham) and identified DEGs relevant to each comparison (Fig. 6B-D; Table S5). Fig. 6RNA-seq analysis of microglial response in young and aged mice after CA. A) Experimental design. Cells were collected on day 3 after sham or CA surgery. B-D) DEG and pathway analysis for 3 different comparisons. Selected DEGs are highlighted with gene names in volcano plots RNA-seq analysis of microglial response in young and aged mice after CA. A) Experimental design. Cells were collected on day 3 after sham or CA surgery. B-D) DEG and pathway analysis for 3 different comparisons. Selected DEGs are highlighted with gene names in volcano plots A total of 110 DEGs were identified between Aged_Sham and Young_Sham. The top up-regulated DEGs included Cxcl13, S100a8, Ifitm3, Lgals3, and H2-Aa, and the most significantly enriched pathway was “Innate Immune Response”, indicating a baseline level of microglial activation in the aged brain even without injury (Fig. 6B). Comparing Young_CA to Young_Sham revealed a large number of up-regulated DEGs (822 genes), and the top 2 were Spp1 and Fn1 (Fig. 6C). This aligned well with our scRNA-seq data, which showed that the most increased microglial cluster in the young brain after CA was MG7, characterized by signature genes Spp1 and Fn1 (Fig. 2B). Moreover, the enriched pathways included those related to IFN response and cell cycle, consistent with an increase in MG8 and MG9 clusters. In contrast, after CA, aged microglia exhibited considerably fewer upregulated DEGs (315 genes), most of which (286 of the 315) were shared with young microglia (Figs. 6D and S5D). Importantly, the fold changes in DEGs were much less pronounced in aged brains than in young brains after CA. For example, 186 genes had a Log2(FC) > 5 in young microglia, but only 3 genes reached this threshold in aged microglia (Table S5). These results indicate that aged microglia mount a diminished response to CA/CPR compared to young microglia. Unique spatiotemporal dynamics of Spp1⁺ microglia in the post-CA brain Both scRNA-seq and bulk RNA-Seq data identified that upregulation of Spp1 expression was a prominent transcriptional change in microglia after CA. Moreover, the Spp1+ microglial subset (MG7) was markedly increased and represented a major microglial state in the post-CA brain. These findings suggest that Spp1-expressing microglia play a key role in brain injury or recovery after CA. Therefore, we next sought to confirm and further characterize this cell population. To this end, we leveraged an Spp1 reporter mouse line Spp1GFPfl in which EGFP is driven by the endogenous Spp1 promoter [21]. We subjected Spp1GFPfl mice to CA/CPR and collected brains for imaging analysis up to day 7. Because no EGFP signal was detected on brain sections at day 1 after CA, our analysis focused on post-CA days 3, 5, and 7. As shown in Fig. 7A, Spp1-expressing (EGFP+) cells were substantially induced on post-CA day 3, declined by day 5, but persisted in certain brain regions, most notably the basal ganglia, for at least 7 days after CA (Fig. S6A). Importantly, although CA induces global brain ischemia, these Spp1+ cells were not uniformly distributed. Instead, they appeared to be present primarily in brain regions that are considered vulnerable to ischemic injury, [29] such as the hippocampus and basal ganglia. Staining for Iba1 or GFAP revealed striking spatial patterns, especially in the basal ganglia region, including the caudate putamen (CPu) and globus pallidus (GP) nucleus (Fig. 7,S6). Clearly, this region responded differently to CA/CPR compared to other subcortical regions, characterized by a robust Spp1⁺ cell cluster and pronounced microgliosis and astrogliosis (Fig. 7,S6). Notably, the GP region showed markedly reduced MAP2 immunoreactivity, indicating neuronal damage (Fig. S7). Moreover, on day 7, many astrocytes in the basal ganglia were polarized toward the specific region that showed strong Spp1+ microglial activation on day 3 (Fig. 7B), and appeared to form a scar-like structure, similar to the glial scar observed after ischemic stroke [30]. Together, these features indicate a microinfarct-like pathology in the brain basal ganglia after CA. Fig. 7Unique spatiotemporal pattern of Spp1-expressing cells in the post-CA brain. Spp1GFPfl mice were subjected to cardiac arrest (CA). Brain sections were collected on post-CA days 3, 5, and 7, followed by immunohistochemistry analysis. EGFP+ (Spp1+) cells were predominantly present in the hippocampus and basal ganglia. A) Spp1 expression is selectively induced in a subset of microglia after CA. B) Astrogliosis around Spp1+ cells after CA. Representative images are shown from at least 2 mice per time point Unique spatiotemporal pattern of Spp1-expressing cells in the post-CA brain. Spp1GFPfl mice were subjected to cardiac arrest (CA). Brain sections were collected on post-CA days 3, 5, and 7, followed by immunohistochemistry analysis. EGFP+ (Spp1+) cells were predominantly present in the hippocampus and basal ganglia. A) Spp1 expression is selectively induced in a subset of microglia after CA. B) Astrogliosis around Spp1+ cells after CA. Representative images are shown from at least 2 mice per time point Next, we performed flow cytometry analysis and confirmed that the Spp1+ cells in the post-CA brain were predominantly microglia (Tmem119+; Fig. 8A, B). Further, as revealed by Iba1 staining (Fig. 7A), only a portion of activated microglia were EGFP+ (Spp1+). This prompted us to investigate the differences between Spp1⁺ and Spp1⁻ microglia. We thus performed morphologic analysis, which showed that Spp1+ vs. Spp1− microglia had a larger cell surface area (Fig. 8C, D), suggesting stronger activation. Consistent with this, in vitro assay demonstrated that Spp1+ microglia exhibited greater phagocytic activity (Fig. 8E). Lastly, as SPP1 is primarily a secreted protein, we measured its plasma levels and found a significant increase on day 3 after CA in young mice, with a less pronounced increase in aged mice (Fig. 8F), in line with scRNA-seq data. This finding raises the intriguing possibility that a portion of plasma SPP1 originates from activated microglia after CA. Indeed, plasma SPP1 has been proposed as a biomarker for brain injury [31,32]. Fig. 8Spp1+ microglia in the post-CA brain. A-E) Spp1GFPfl mice were subjected to cardiac arrest (CA). All analyses were performed on day 3 after CA. A, B) Flow cytometry analysis of the cell composition. EGFP+ cells were first gated for percentage analysis. Representative histograms were shown (n = 4). C, D) Morphologic analysis of microglia. C) Examples of reconstructed and rendered images. D) Quantification of cell surface areas of microglia in the basal ganglia regions. Representative microglial morphology from each group were shown. E) Phagocytosis. In vitro phagocytic activity was performed with fluorescence beads. F) Plasma SPP1 level. Young and aged mice were subjected to CA. On day 3 after CA, plasma samples were analyzed with ELISA. The data are presented as mean ± SD. *, p < 0.05. ****, p < 0.0001 Spp1+ microglia in the post-CA brain. A-E) Spp1GFPfl mice were subjected to cardiac arrest (CA). All analyses were performed on day 3 after CA. A, B) Flow cytometry analysis of the cell composition. EGFP+ cells were first gated for percentage analysis. Representative histograms were shown (n = 4). C, D) Morphologic analysis of microglia. C) Examples of reconstructed and rendered images. D) Quantification of cell surface areas of microglia in the basal ganglia regions. Representative microglial morphology from each group were shown. E) Phagocytosis. In vitro phagocytic activity was performed with fluorescence beads. F) Plasma SPP1 level. Young and aged mice were subjected to CA. On day 3 after CA, plasma samples were analyzed with ELISA. The data are presented as mean ± SD. *, p < 0.05. ****, p < 0.0001 Depletion of microglia worsened functional outcome after CA Our data above provided the first evidence that microglia considerably changed their states after CA, which is the primary immune response in the post-CA brain. However, research on the role of microglia in CA has been scarce. Even the net effect of microglia on CA outcome has not been explored. Thus, as the first step toward this aspect, we examined the effect of microglial depletion in our CA model (Fig. 9A). We found that PLX3397, a CSF1R inhibitor, effectively ablated microglia in 10 days (Fig. S8A). Moreover, mice with microglial depletion showed a baseline performance on grip strength and open field tests similar to control mice (Fig. S8B, C). We then subjected control and PLX3397-treated mice to CA. No significant difference was noted in body weight and time-to-ROSC between the 2 groups (Fig. S8D, E). Of note, 3 mice in the PLX3397 group and one mouse in the control group died on day 1 after CA (Fig. 9B). On day 2, functional outcome and the immune cells in the brain were assessed. Compared to control mice, PLX3397-treated mice exhibited almost no movement in the open filed test, reduced grip strength, and significantly worse neurologic scores (Fig. 9C-E). Consistently, PLX3397-treated mice exhibited significantly more degenerating neurons (Fig. 9F, S8F). Further, flow cytometry analysis confirmed effective depletion of microglia in PLX-3397-treated mice (Fig. S9). Unexpectedly, the number of infiltrating immune cells (CD45High), such as neutrophils, was lower in the brains of PLX3397-treated mice than in controls after CA (Fig. S9B). This finding suggested that the worse outcome was not due to an increase in infiltrating immune cells, but rather specifically related to microglial depletion. Fig. 9Depletion of microglia worsens functional outcome after CA. A) Experimental design. B) Survival rate. C-E) Functional outcome. On day 2 after CA, mice were subjected to behavioral tests, including open field (C), grip strength (D), and neurologic scoring (E). F) Representative images of Fluoro-Jade C (FJC) staining. Enlarged insets show FJC⁺ cells. The data are presented as mean ± SD or median (neurologic score). **, p < 0.01; ***, p < 0.001 Depletion of microglia worsens functional outcome after CA. A) Experimental design. B) Survival rate. C-E) Functional outcome. On day 2 after CA, mice were subjected to behavioral tests, including open field (C), grip strength (D), and neurologic scoring (E). F) Representative images of Fluoro-Jade C (FJC) staining. Enlarged insets show FJC⁺ cells. The data are presented as mean ± SD or median (neurologic score). **, p < 0.01; ***, p < 0.001 Discussion This is the first comprehensive study of the immune landscape in the post-CA young and aged brain. In line with previous reports, we found that compared to young brains, aged brains exhibited a chronic inflammatory state, as evidenced by greater lymphocyte presence, elevated cytokines and chemokines, and a higher proportion of activated microglial subpopulations. Interestingly, despite this pro-inflammatory environment, the post-CA aged brain had fewer infiltrating immune cells, and aged microglia displayed attenuated transcriptomic changes after CA and resuscitation, suggesting an impaired capacity to respond to an ischemic insult. Notably, after CA, the aged vs. young brain had fewer proliferative microglia but more pro-inflammatory microglia. These differences likely contribute to age-related disparities in neurologic outcome after CA. Further, we discovered that the distinct basal ganglia regions, including the CPu and GP nucleus, mounted a unique response to CA/CPR, characterized by clustering of Spp1⁺ cells and the accumulation of activated microglia and astrocytes, resembling a microinfarct injury pattern. Lastly, our functional outcome data suggest that microglia provided overall protective effects on the brain after CA. Our scRNA-seq data indicated a notable presence of lymphocytes, particularly T cells, in the aged brain, which is corroborated by our flow cytometry results. This phenomenon has been previously reported in several scRNA-seq studies in mice, and observed in elderly human brains as well [33–35]. For example, a recent study showed that aged mouse brains (18 months old) have 45 times more T cells than young mouse brains (3 months old) [34]. Notably, CD8+ T cells have been suggested to play a neuropathic role in the normal aging brain [35]. Interestingly, exercise can reverse this age-related progress, as demonstrated by a study in which 21-day exercise markedly reduced the presence of T cells in the aged brain. In our work, we also found that the young brain had only negligible T cells, even after CA, suggesting a less prominent role for T cells in the early phase after CA. Instead, monocytes appeared to be the primary infiltrating immune cells, increasing from 0.5% to 4.7% in the young brain post CA. Notably, this increase was not observed in the aged brain. Our data further indicate that the lack of monocyte infiltration into the aged brain is not due to a deficiency in chemokines. On the contrary, DEG analysis demonstrated that chemokines such as Cxcl10, Ccl3, and Ccl4 were among the top upregulated genes in the aged brain vs. the young brain. The cytokine array analysis further supported this finding, showing higher levels of chemokines in the aged brain. We also examined bone marrow cell egress using AMD3100 treatment, and found no significant difference between young and aged mice. These results led us to speculate that age-related defects in immune cell migration may be responsible for this differential response between young and aged mice. Consistent with our findings, Shi et al. reported that after ischemic stroke, fewer peripheral immune cells infiltrate the aged brain compared to the young brain [36]. Moreover, after ischemic stroke, young microglia exhibit more robust gene expression changes than aged microglia, aligning with our RNA-seq data. Of note, a lower fold change in gene expression in aged microglia does not necessarily indicate lower expression because baseline expression levels of many inflammatory genes are often already elevated in aged microglia, as shown in our data. Collectively, our findings highlight the age-related differences in immune cell infiltration and transcriptomic responses in the brain after CA. As only a small number of infiltrating immune cells infiltrate the post-CA brain, the major impact of CA on the brain’s immune response is reflected in the changes in microglial states, i.e., microglial clusters. In this context, mounting evidence has demonstrated the remarkable heterogeneity of microglia in healthy and diseased brains. For example, our recent scRNA-seq analysis documented 13 distinct microglial states in the stroke brain, of which 6 clusters are considered homeostatic states, while 7 clusters likely represented stroke-associated microglia (SAM) [19]. The prominent SAM clusters include Spp1+, proliferative state, chemokine-producing, and IFN responsive states. Interestingly, similar clusters were also identified and increased in the post-CA brain. Consistent with a previous study,37 no specific microglial population in the aged brain was identified; however, young and aged brains indeed exhibited different percentage changes in microglial clusters after CA. Previous studies have shown that microglial clusters with high expression of pro-inflammatory genes (e.g., Il1b, Ccl4, and Ccl2) and type I IFN-related genes (e.g., Ifitm3 and Oasl2) are more prevalent in the aged brain compared to the young brain [34,37]. In our scRNA-seq analysis, we also identified a prominent age-related microglial cluster (MG8), characterized by high levels of cytokines (e.g., Tnf), chemokines (Cxcl10, Ccl2, and Ccl4), and IFN-related genes (e.g., Isg15 and Stat2). This cluster was already more abundant in aged sham brains, and markedly increased after CA, compared to the young brain. Based on its pro-inflammatory expression profile, MG8 microglia likely contribute to brain damage after CA. Therefore, further research is warranted to elucidate their spatiotemporal distribution and functional roles after CA. Another notable change in microglia is that the percentage of the Spp1+ microglia cluster increased from 1.5% to 11.2% in the young brain, and from 1.5% to 6.2% in the aged brain after CA. A recent analysis of several published scRNA-seq datasets identified Spp1 as the most prominently upregulated gene in microglia from multiple disease models (e.g., glioblastoma multiforme and experimental autoimmune encephalomyelitis) [38]. Similarly, our previous scRNA-seq analysis identified Spp1 as the top upregulated gene in microglia after ischemic stroke [19]. These data together underscore the importance of Spp1+ microglia in the brain’s immune response under pathologic conditions. However, Spp1 has not yet been studied in experimental CA. Here, we provide the first evidence that Spp1 expression was markedly increased in a subset of microglia after CA, and these microglia emerged primarily on day 3 and largely declined by day 7 after CA. Of particular interest is the unique spatial distribution of these Spp1⁺ microglia in the post-CA brain (Fig. 7, S6). As expected from global brain ischemia induced by CA, Iba1 and GFAP staining revealed widespread activation of microglia and astrocytes in the brain, with some regions showing stronger responses. In contrast, Spp1⁺ microglia exhibited a highly localized distribution, predominantly in the hippocampus and basal ganglia. Both brain regions are known to be highly susceptible to ischemic insult, suggesting that local neuronal damage may trigger the transition of microglia into a Spp1+ phenotype. Indeed, the hippocampus has traditionally been the primary focus of research on CA-induced neuronal death. Equally important, however, both clinical and experimental evidence indicates that the basal ganglia also undergo substantial damage [39–41]. For example, one clinical study of 115 CA patients found that higher regional net water uptake in the caudate and putamen nucleus is strongly associated with poor outcome,41 highlighting the functional importance of the basal ganglia after CA. Indeed, increasing evidence suggests that damage to the basal ganglia critically contributes to motor and cognitive neurologic deficits in patients after CA [42-43] Consistently, our data revealed striking changes in the basal ganglia regions after CA, including clustering of Spp1⁺ cells, robust gliosis, and reduced MAP2 immunoreactivity. This pattern suggests the presence of focal neuronal lesions, which develop into microinfarct-like injury. To our knowledge, this provides the first experimental evidence that CA induces microinfarct-like injury in the brain, with Spp1⁺ cell clustering serving as a potential hallmark of such injury. But what role do Spp1⁺ microglia play in the formation, progression, or resolution of these microinfarct-like lesions? Spp1 encodes osteopontin, a primarily secreted protein involved in diverse biological processes such as wound healing and cell migration [44]. Interestingly, many studies in peripheral organs (e.g., kidney, [45] liver, [46] lung, [47] and heart [48]) have reported that the key function of Spp1+ macrophages involves scar formation and resolving damaged tissue. Thus, given their unique spatiotemporal dynamics in the post-CA brain, Spp1⁺ microglia likely perform similar roles. Moreover, our data indicate that Spp1+ microglia exhibit enhanced phagocytic activity, supporting their function in clearing cellular debris. Collectively, these dominant Spp1+ microglia may represent a reparative microglial subpopulation and warrant further research. Lastly, our functional outcome study after pharmacologic depletion of microglia, provided the first experimental evidence that microglia play a net protective role in CA outcome. This depletion approach has been widely used to investigate microglial function in brain disorders. Studies have demonstrated that microglial depletion reduces neuroinflammation and improves outcomes in intracranial hemorrhage and sepsis-associated encephalopathy [49,50]. However, other studies have shown that microglial depletion increases the production of inflammatory mediators, leukocyte infiltration, and infarct volumes after ischemic stroke in both young and aged mice [51,52]. We found that, in the absence of microglia, mice exhibited worse functional outcomes, greater neuronal damage, and fewer infiltrating immune cells in the brain after CA. These effects on CA outcomes are unlikely to result from systemic actions of PLX3397, as our data revealed no significant difference in neurologic function between control and PLX3397-treated mice before CA (Fig. S8) and a previous study has shown minimal effects of PLX3397on peripheral immune cells [53]. However, it should be noted that our data provide only initial insights into the overall role of microglia in CA. Further studies using more specific tools are needed to delineate the functions of CA- and age-associated microglial subsets. To this end, our scRNA-seq dataset can serve as an invaluable resource to guide and inform these investigations. In summary, our data demonstrate that a marked transition in microglial states constitutes a dominant immune response in the brain after CA. Targeting specific microglial subpopulations may offer a promising therapeutic strategy to modulate neuroinflammation, mitigate brain injury, promote recovery, and ultimately improve outcomes after CA. Supplementary Information Supplementary Material 1 Supplementary Material 1