Single-cell transcriptome profiling reveals the heterogeneity of immune and non-immune cells after ICH.
In order to reveal the changes in immune cells during the progression of ICH, we collected 10×Genomics scRNA-seq datasets from fresh PHE samples, which were resected from brain tissue in the vicinity of hematoma during the evacuation of the hematoma. We constructed a multi-stage profile including Group1 (n = 3, 0–6 hours after ICH, G1), Group2 (n = 3, 6–24 hours after ICH, G2), and Group3 (n = 3, 24–48 hours after ICH, G3) (Fig. 1A). After quality control, a total of 3152–5823 cells were retained, with an average of 2639–4716 genes per cell and an average of 9535–24244 unique molecular identifiers (UMIs) per cell for the subsequent analysis. Following that, we clustered and visualized various cell types refering to their relative gene expression levels using uniform manifold approximation and projection (UMAP), an unsupervised nonlinear dimensionality reduction algorithm. Graph-based Louvain clustering algorithms were used to cluster all cells into subsets resulting in 19 clusters (Fig. 1B). Based on marker gene expression levels of microglia (AIF1, CSF1R, TMEM119, CX3CR1), clusters 5, 6, 7 and 13 were identified as microglial cells (Fig. 1C and 1E). Clusters 8 and 16 expressed genes (ALDH1L1, ATP1B2 and AQP4) specific to astrocytes (Fig. 1C and 1E). Within the infiltrating immune cell subsets, cluster 3 had neutrophil marker genes (CSF3R, S100A8, CXCR2 and FCGR3B) and cluster 9 had monocyte marker genes (CD300E and VCAN) (Fig. 1C and 1E). Cluster 11 expressed genes (CD3D, CD3E, NKG7, GZMA and GZMB) specific to cells of NK/T cells, while cluster 19 expressed genes (CD79A, CD79B and MS4A1) specific to B cells (Fig. 1C and 1E). Additionally, clusters 1, 2, 4, 10, 14, 15 and 17 expressed genes (MOG, SOX10, CNP and HAPLN2) specific to oligodendrocytes, and cluster 12 expressed genes (CSPG4 and PDGFRA) specific to neural progenitor cells (Fig. 1C and 1E). cluster 18 expressed endothelial cell marker genes (CLDN5, VWF, RGS5 and EGFL7). Therefore, we identified this cluster as endothelial cells (Fig. 1C and 1E). Furthermore, neutrophils, NK/T cells, and monocyte cells were present at all time points and increased to reach their highest level at G3 (Fig. 1D), a timepoint that appears crucial to investigating ICH-induced immune cell changes (14). Microglia and neutrophils, as the most important immune cell populations in the central nervous system and peripheral immune system, respectively, were essential in the pathophysiology of ICH (1, 3). Therefore, we next focused on analyzing the transcriptional profiles of these two cell types and investigated the crosstalk between central and peripheral immune cells during ICH progression. In addition, the number of B cells was too small for further bioinformatic analyses, so we excluded them from subsequent analyses.
Single-cell RNA sequencing reveals the complexity of microglia states during ICH progression.
As a result of cluster analysis of the 8353 microglia-like cells, 12 clusters were identified (Fig. 2A) characterized by differentially expressed genes (DEGs). We determined the enrichment of biological pathways in each cluster via gene set enrichment analysis (GSEA) (Fig. 2D). There was almost no overlap in the DEGs defining each cluster, supporting the unique nature of each microglia cluster (Fig. 2D, Additional file 1: Fig. S1). First, we identified annotated clusters of microglia phenotypically similar to those previously described in the human brain. Cluster-enriched sets of transcriptional regulators and transcription factors were observed in some clusters (1, 2, 4, 5, 6, 7, 8, 10, 11, and 12) but not in others (3 and 9) (Fig. 3A) (15). Besides, some selected cell-surface marker genes were not observed in cluster 9 (Fig. 3B). The absence of detectable unique cell-surface markers and on-off transcription factors among clusters 3 or 9 may represent homeostatic microglia, whereas the other clusters differed from them through the upregulation of specific genes. As an additional finding, we identified cluster 9 as an enriched cluster for homeostatic genes, owing to its high levels of P2RY12 and CX3CR1 expression (Additional file 1: Fig. S2A, Additional file 1: Fig. S3A) (11, 16–18). However, the higher expression of homeostatic markers was not found in cluster 3 (Additional file 1: Fig. S2A). Therefore, we annotated cluster 9 as homeostatic microglia (HM) cluster. HM was established as a comparative basis for evaluating DEGs from other microglia clusters (Table 1), referring to the approach in previous literature (15, 17, 18).
Initially, we identified annotated clusters of microglia phenotypically similar to those previously characterized in the human brain. Notably, Cluster Micro3 was characterized by the downregulation of homeostatic genes (Additional File 1: Fig. S3A), such as checkpoint genes (TMEM119 and CX3CR1) and purinergic receptors (P2RY12), and by the upregulation of encoding many metabolic genes (APOC1, VIM, LDHA, RPS2, RPS6, RPS10, RPS19, and RPL12), predominantly ribosomal subunits genes (Fig. 2C). In summary, cluster Micro3 reflects a degenerative phenotype of microglia, consistent with the responses of microglia to aging. Remarkably, Cluster Micro5 predominantly expressed genes characteristic of disease-associated microglia (DAMs), such as metabolic genes LPL and FABP5 (Fig. 2C) (19). DAM subtype is novel microglia associated with neurodegenerative diseases such as Alzheimer's (16). The pathway analysis of the Micro5 genes highlighted associations with "Alzheimer's disease" and "Huntington's disease" (Fig. 2D, Additional File 1: Fig. S1B). Hence, we annotated this cluster as "DAM-like" microglia. Cluster Micro6 and Micro10 were defined by genes and pathways involved in canonical inflammatory phenotype. GSEA indicated that these two clusters were enriched in Toll-like receptor (TLR) signaling, Nod-like receptor (NLR) signaling, and chemokine signaling pathway, suggesting inflammatory responses of downstream effectors to stimuli (Fig. 2D, Additional File 1: Fig. S1A). Cluster Micro11 was characterized by anti-inflammatory and repair-related genes (HTR7, PDLIM7, and LGALS3) and proinflammatory genes (KCNN4 and ITGB7) (20–22), indicating that this cluster was an intermediate state in the polarization of microglia (Fig. 2C). Cluster Micro12 was defined by expression of genes involved in DNA repair and cell cycle regulation, including MKI67, SKA1, E2F2 and E2F8 (Fig. 2C). Furthermore, Micro12 was enriched for pathways involved in DNA replication and the cell cycle (Fig. 1D, Additional File 1: Fig. S1B). Micro1 was defined by genes (XIST, VEGFA, KLF4) involved in microglial M1 polarization (Fig. 2C) (23–26). Moreover, we found that cluster Micro1 was between HM cluster (Micro9) and canonical inflammatory phenotype (Micro6), suggesting that this cluster could be the intermediate transition status from HM microglia to proinflammatory microglia.
Subsequently, we identified four microglial clusters—Micro2, Micro4, Micro7, and Micro8—that had not previously been characterized in human brain studies. These subclusters were distinguished by the enrichment for the pathway of DEGs relative to HM microglia. The significant DEGs of Micro7 were involved in the neurotrophin signaling pathway and Fc gamma receptor-mediated phagocytosis (Fig. 2D, Additional File 1: Fig. S1B). Micro4 showed gene enrichment for complement and coagulation cascades and the PPAR signaling pathway (Additional File 1: Fig. S1A), partially sharing a subset of DEGs with Micro7 (Fig. 2C). The activation of the PPAR signaling pathway has been demonstrated to attenuate proinflammatory responses and increase neurotrophic factors in patients with ICH. Accordingly, we annotated these two clusters as tissue repair phenotypes. Additionally, the pathway significantly enriched Micro8 in antigen processing and presentation, suggesting this subcluster of microglia is active in antigen processing and presentation for immune response (Additional File 1: Fig. S1A). Furthermore, the pathways enriched in Micro2 confirm the relative increase of genes involved in oxidative phosphorylation and glycolysis/gluconeogenesis and decreased chemokine and endocytosis genes (Additional File 1: Fig. S1A). This finding coincides with previous studies that persistent glycolysis exerts adverse effects on microglial functions: the activation of glycolytic metabolism impairs phagocytosis and chemotaxis of microglia (27, 28). Overall, while future research will likely refine our understanding of microglial subtypes, our study significantly advances the knowledge of microglial heterogeneity in human PHE tissue.
Microglia subclusters predominantly exhibit proinflammatory phenotypes after ICH
Previous studies of single-cell transcriptomics have shown diverse subclusters of microglia, which are considered to reflect their different functions. In this study, we employed scRNA-seq to investigate the biological pathways present in microglia within PHE tissue following ICH. We observed that common microglial marker genes such as AIF1, TREM2, and CSF1R were widely expressed across all microglial subclusters (Additional File 1: Fig. S3A). However, other specific marker genes of microglia (ITGAM, P2RY12, and CX3CR1) indicated differential expression across clusters (Additional File 1: Fig. S3A). Interestingly, the transcriptome of PHE tissue was almost dominated by proinflammatory pathways (Additional File 1: Figs. S3A-B). Our findings indicated a lack of significant activation of anti-inflammatory pathways within the first 48 hours post-ICH (Additional File 1: Fig. S3B). Proinflammatory genes such as CCL2, CCL4, and IL1B were among the most prevalently expressed cytokine and chemokine genes in ICH-associated microglia (Additional File 1: Fig. S3A-B). ICH microglial clusters 1, 3, 5, 6, 10, and 11 were characterized by high gene expression levels of HLA-DQA, HLA-DPB1, and HLA-DRA and low gene expression levels of P2RY12 and CX3CR1 (Additional File 1: Fig. S3A), suggesting their involvement in the primary immune response to ICH. Complement pathway-related genes (C3, C1QB, and C1QC) also maintained high levels of expression in all microglial cell clusters (Additional File 1: Fig. S3B). Overall, most microglial clusters displayed proinflammatory phenotype within 48 h of ICH, further corroborating the notion that a proinflammatory response is a key pathogenic mechanism in ICH.
Purinergic receptor P2RY12, the cell-surface proteins of microglia, play key roles in mediating neuroinflammatory responses (29). Our results indicated reduced expression of the P2RY12 gene in microglia clusters that had higher IL1B expression levels (Additional file 1: Fig. S2A). This finding was consistent with the previous studies that the expression of P2RY12 was gradually decreased accompanied by microglia activation following inflammatory stimulation (30). Further analysis was performed by comparing the differentially expressed genes in IL1B-expressing clusters (cluster 6 and cluster 10) with those in P2RY12-expressing clusters (cluster 7 and cluster 9). A significant difference in gene expression was found between IL1B-expressing microglia and P2RY12-expressing microglia, with 882 genes notably downregulated and 415 genes notably upregulated (adjusted P value < 0.05, log2 (fold change) > 1.5) (Additional file 1: Fig. S2B, Additional file 2: Tab. S2). A significant increase in chemokine and pro-inflammatory cytokines was observed in microglial cluster cells expressing IL1B (Additional file 1: Fig. S2C). (Additional file 1: Fig. S2C). Additionally, CX3CR1 expression was also higher in cluster cells expressing P2RY12 than in cluster cells expressing IL1B (Additional file 1: Fig. S2B). In addition, Gene Ontology term enrichment analysis indicated genes enriched for protein binding, extracellular exosome, focal adhesion, cytokine-mediated signaling pathway, and inflammatory response (Additional file 1: Fig. S2D). Furthermore, the Kyoto Encyclopedia of Genes and Genomes term enrichment analysis suggested genes enriched for apoptosis, NF − kappa B signaling pathway, IL − 17 signaling pathway, and Toll − like receptor signaling pathway (Additional file 1: Fig. S2E). According to DEGs analysis, pro-inflammatory microglia expressing IL1B are structurally and functionally different from those expressing P2RY12. According to these findings, as well as previous studies, PHE tissue removed from patients with ICH contains an immune pathogenic microenvironment that attracts and induces non-specific and specific immunity rapidly. Hence, we emphasized on the characterization of immune cells infiltrating in PHE tissues.
Microglia cluster-specific transcription factor regulatory networks.
To explore the regulatory networks of the microglia clusters in the dataset, we also applied SCENIC analysis to identify the top transcription factor-driven networks (regulons) controlling gene expression in each of these 12 microglia clusters (Fig. 4A, Additional File 1: Fig. S4A). Each microglia cluster was characterized by a specific set of regulons (Fig. 4B). This supports the theory that transcriptional regulation mechanisms are key determinants of the unique gene expression profiles observed in each microglia cluster. For example, Micro1 showed higher activity levels of POLR2A, NFKB2, GTF2B, and BCLAF1 (Fig. 4A, Additional File 1: Fig. S4A). Micro3 showed higher activity levels of SOX8, SOX10, IRF7, and STAT1 (Fig. 4A, Additional File 1: Fig. S4A); The activity of transcription factors, such as MAFB, SPI1, DDIT3, and XBP1, was higher in Micro11, while high activity levels of E2F1, TFDP1, and BRCA1 were associated with Micro12 (Fig. 4A, Additional File 1: Fig. S4A). Furthermore, our analysis revealed that MAFB, a regulon governed by transcription factors commonly linked with the anti-inflammatory polarization of human microglia, was prominently featured in the Micro11 cluster (Fig. 4B, Additional File 1: Fig. S4C). This is consistent with the finding that these cells experience a phenotypic polarization of microglia of M2. In this study, the NFKB1 regulon, associated with canonical inflammatory responses, was identified in Micro10 (Fig. 4B, Additional File 1: Fig. S4C). Conversely, Micro6 exhibited the RELB regulon, linked to non-canonical inflammatory responses (Fig. 4B, Additional File 1: Fig. S4C). In Micro9, the high specificity of FOXP2 (Additional File 1: Figs. S4C-D), a regulon unique to human microglia and crucial for brain development, was observed, aligning with previous research identifying this subtype as an HM cluster (31). The top three regulons in other microglia clusters also showed distinct variations (Fig. 4B, Additional File 1: Figs. S4B-C). These inferred transcription factor regulons provide insight into the diversity and difference within microglial clusters, suggesting novel potential regulatory targets for future research
The SPP1 signaling pathway was the fundamental bridge to self-communication among microglia subclusters
Along with the PHE progression, microglial subtypes also changed accordingly. The use of cell-cell communication networks between microglia subpopulations could contribute to a better characterization of microglia function. Interestingly, the interaction strength of the SPP1 pathway increased gradually with the progression of PHE (Fig. 5A, Additional File 1: Fig. S5). Moreover, the SPP1 pathway exhibited the strongest interaction strength, irrespective of incoming or outgoing signaling pathways (Fig. 5B). It indicates that the SPP1 signaling pathway could be responsible for self-communication between microglia subclusters. Regarding the incoming signaling, SPP1 was emitted by different microglial subclusters at different stages. At post-ICH in G1, G2, and G3, the strongest SPP1 signaling cell types were Micro6, Micro1, and Micro11, respectively (Fig. 5B). Regarding outgoing SPP1 signaling, the Micro6 subtype was also strongest at G1 after ICH. In ICH patients at G2 and G3, Micro11 showed strong SPP1 signals (Fig. 5B). Additionally, we visualized the crosstalk between each microglia subcluster in the SPP1 signaling pathway. We explored the specific receptor ligands and found that the SPP1-( ITGAV + ITGB1) ligand-receptor pair was the fundamental bridge of self-communication among microglia subclusters (Fig. 5C). Collectively, our findings demonstrated that the signaling pathway of SPP1 is the fundamental bridge mediating self communication between subclusters of microglia during the progression of ICH.
Time-dependent transcriptional heterogeneity of neutrophils in the human brain after ICH.
To analyze neutrophil transcriptional heterogeneity during ICH progression, we performed a multiplexed time series of scRNA-seq analyses combining transcriptomics. By comparing the gene expression patterns of all neutrophils at every time points, we depicted five different transcriptional cell clusters in PHE tissue after ICH, exhibiting a time-independent appearance. Due to the fact that the number of cells sampled at each time point (G1: 1312 cells; G2: 1132 cells; G3: 2109 cells) cannot reflect the true level of neutrophils in ICH PHE tissues (Fig. 6B), we calculated the proportion represented by each cluster at different time points (Fig. 6C). The majority of neutrophils at grade 1 (68.6%) segregated in cluster Neutro2 (CFD, CDKN2D, HSPA1B, S100A4, D100A6, S100A9, S100A12), a profile significantly reduced by half at later time points (< 30% at grade 2 and 3). At grade 2, Neutro3 cells (86.9% of total neutrophils; FOLR3, SLC8A1, AOAH, SYNE2) were predominant. Cluster Neutro5 (DPYD, KIFC3, BMP2K, ABHD5, SPDYA) was present at all time points and its levels significantly increased from 0.4% of cells at grade 1 to 10.4% at grade 3. Cluster Neutro1 also increased to reach its highest level at grade 3 (61.5% of all neutrophils; MCEMP1, TNFAIP3, IER3, TANK). Finally, cluster Neutro4 was characterized by highly specific expression of some transcripts (PLNA, PFN1, HMGA1), and represented 26.8%, 1.9% and 5.0% of all neutrophils at grade 1, 2 and 3 post-ICH, respectively (Fig. 6C, Additional file 1: Fig. S6A).
The unique gene signatures and the top 20 significantly DEGs of each neutrophil subcluster were delineated (Additional file 1: Fig. S7A). In addition, the Gene Set Variation Analysis (GSVA) was performed to functionally annotate the neutrophil subcluster (Fig. 6D). Neutro1 showed gene enrichment for steroid biosynthesis, ABC transporters, and PPAR signaling pathway, partially sharing a subset of significantly differentially expressed genes with Neutro5 (Fig. 6D). Neutro2 was characterized by the chemokine signaling pathway and antigen processing and presentation (Fig. 6D), indicating this subtype of cells is active in antigen processing and presentation for immune response. The significantly DEGs of Neutro3 were involved in the folate biosynthesis and the metabolism-related pathways including histidine metabolism, alpha-linolenic acid metabolism, and ascorbate metabolism (Fig. 6D). Neutro4 exhibited higher expression of genes associated with glycolysis and gluconeogenesis, and extracellular matrix (ECM)-receptor interaction (Fig. 6D). Neutrophils can carry preexisting matrix from nearby tissue to reestablish new ECM scaffold in the early stages of tissue repair, suggesting that these cells may represent a repair phenotype (32). Furthermore, Neutro5 was defined by the expression of genes involved in RNA polymerase and TCA cycle (Fig. 6D).
Neutrophil cluster-specific transcription factor regulatory networks.
We further employed SCENIC analysis to characterize the underlying molecular mechanisms driving the differentiation of different neutrophil phenotypes. A different transcription factor network was predicted to be expressed by neutrophils from different subtypes in this analysis. For instance, CEBPB and HIVEP2 regulons were upregulated in Neutro1 (Fig. 6E). Consistent with the findings that the CEBPB regulon is critical for the emergency granulopoietic response, this mechanism fulfills the increased demand for neutrophils during the innate immune response to inflammation (33). Additionally, there was a notable increase in the Neutro1 population, peaking at G3 (Fig. 6C). This suggests a substantial consumption of neutrophils at this stage, prompting the hematopoietic system to respond to the heightened demand through emergency granulopoiesis quickly. Neutro2 also upregulated networks driven by ZNF107, IRF1, and SPI1 (Fig. 6E). In the Neutro3 cluster, there was a noticeable predominance of the activity of regulons, particularly those linked to ZBTB16 and ELF1 (Fig. 6E). Some regulons showed preferential activity in Neutro4 (MECP2, SREBF1, HIF1A; Fig. 6E). Neutro5 showed highly active transcription factors related to cell proliferation (FOXO1, CEBPZ, YY1; Fig. 6E). By calculating the connection specificity index (CSI), we obtained a regulatory network composed of four modules and 34 regulons (Fig. 6F). Transcription factors of CEBPB/FOSL2/HIVEP2 in module 2 displayed the strongest activity in Neutro1 (Figs. 6F-G) and were related to emergency granulopoiesis.
Trajectory analysis of neutrophils in PHE tissue during ICH progression
In order to explore the dynamic transition of neutrophils from peripheral blood to PHE tissue, we constructed a pseudotime map of the neutrophil state trajectory using monocle2 (Figs. 7A, 7B, and 7D). Neutrophils are also known as short-lived cells that can mobilize rapidly from the bone marrow in response to tissue damage (34). Accordingly, we performed a 'bone marrow proximity score' of neutrophils based on the expression of a set of genes previously characterized in transcriptome analyses of neutrophils at different stages (35). Cluster Neutro2 (mainly in G1) had the highest BM proximity score (p < 0.0001 versus all other clusters) (Additional File 1: Fig. S6B).
In our analysis, the progression trajectory of neutrophils was established, starting with Neutro2 (Fig. 7C). This was followed by Neutro3, serving as a transitional state between Neutro2 and Neutro4, then moving through an intermediate infiltrating phase represented by Neutro1, and culminating in the terminally differentiated states of Neutro4 and Neutro5. Further examination of the single-cell transcriptomes of neutrophils along this trajectory identified 4043 significantly altered genes, classified into four distinct expression patterns (Figs. 7E-F): Module 1 included genes that showed increased expression levels along one trajectory. Pathway enrichment analysis suggested that these genes participated in the chemokine signaling pathway, NOD-like receptor signaling pathway, NF-kappa B signaling pathway, and TNF signaling pathway (Fig. 7G). Module 2 contained genes activated in the later stages of another trajectory, with enrichment in mitophagy and HIF-1 signaling pathways (Fig. 7G). Additionally, the genes in module 3 were associated with ribosome function, Fc gamma R-mediated phagocytosis, endocytosis, regulation of actin cytoskeleton, and leukocyte transendothelial migration (Additional File 1: Fig. S6C). Finally, module 4 genes, upregulated in the early stages, were associated with necroptosis, apoptosis, and cellular senescence (Additional File 1: Fig. S6C).
OPN-mediated microglia-monocyte interaction was essential for the crosstalk between central and peripheral immune cells
We previously discovered a mixed cluster (cluster 11) by immune cell analysis (Fig. 1E). Cluster 11 clearly showed CD3D, CD3E, NKG7, GAMA, and GAMB gene expression, indicating doublets of T cells and NK cells (Fig. 1E). Cluster NK/T cell was present at all time points and its levels gradually increased during ICH progression (Fig. 1D). We re-clustered cluster 11 doublet cells using major lineage gene expression markers and obtained four clusters (Fig. 8A). Cluster phenotypes were identified using gene expression levels (Fig. 8D). A CD8 + T cell cluster (CD8A, CD8B, LAG3; cluster 1 and cluster 2) was the main cluster found in PHE tissue (Figs. 8C-D). We also observed CD4 + T cells (CD4, CCR7, LEF1, SELL, IL2RA; cluster 4) and NK cells (FCER1G, NCR1, NCR3, CCL3, KLRC1, FCGR3A; cluster 3) (Fig. 8D). By inferring the paired ligand-receptor pairs based on CellChat analysis, we first depicted the overall connectivity patterns between peripheral immune cells and central immune cells in PHE. The number of cell-cell interactions between immune cells changed significantly in the context of the progression of ICH (Figs. 8E-F). However, the strength interaction involving microglia, monocytes, and CD8 + T cells is consistently higher in the progression of ICH (Fig. 8G). Intriguingly, monocytes exhibited more extensive communications with microglia than other immune cell types apart from CD8 + T cells and NK cells (Fig. 8G). We then extracted highly expressed interactions engaging microglia during ICH progression and uncovered underlying interactions with monocytes (Fig. 9A, Additional File 1: Fig. S8). Notably, we found interactions between SPP1, which encodes the pleiotropic cytokine OPN, and CD44, ITGAV, ITGA4, ITGA5, ITGA9, ITGB1, ITGB3, and ITGB5, which encode the OPN receptor, were prominent during microglia-monocyte interactions (Fig. 9A). Furthermore, the pair of OPN-CD44 stands out among all interaction pairs that mediate the crosstalk between microglia and monocytes and displays the highest score (Figs. 9A-B, Additional File 1: Fig. S8). Additionally, we found that the SPP1 gene was primarily expressed in the microglia rather than in other immune cells. In contrast, the CD44 gene was mainly expressed in the monocytes (Fig. 9C). This finding suggests that microglia-secreted OPN could regulate the immune environment of PHE by interacting with CD44 on monocytes. In summary, these findings indicate that OPN-mediated microglia-monocyte interaction is essential for communicating between central and peripheral immune cells.
IF
To verify our conclusion, we extrally collected a group of adjacent hematoma tissues from patients with ICH for immunofluorescence experiments. Iba-1 (surface marker of microglia), CD44 and OPN were labeled with different colors of fluorescence for immunofluorescence co staining. Afterwards, we used a Zeiss Imager Z2 confocal microscope to observe the stained tissue sections, such image was captured under a microscope: co localization existing between osteopontin and microglia (Iba-1), and highly spatially close to CD44 (Fig. 10). This also coincides with the conclusion drawn from our previous analysis: 1. Osteopontin is secreted by microglia; 2. Osteopontin as a mediator participates in the activation of microglia and CD44 cells, especially monocytes.