Single-cell sequencing analysis of chronic subdural hematoma cell subpopulations and their potential therapeutic mechanisms

Background: Chronic subdural hematoma (CSDH) is a prevalent form of intracranial haemorrhage encountered in neurosurgical practice, and its incidence has notably risen in recent years. Currently, there is a lack of studies that have comprehensively classified the cells present in hematomas removed during surgery


Introduction
Chronic subdural hematoma (CSDH) is one of the most common types of intracranial haemorrhage in neurosurgical practice, especially in the elderly population (>65 years).The reported incidence of CSDH ranges from 1.72 to 20.6 cases/per 100,000 people per year, significantly increasing due to an ageing society (Feghali et al., 2020;Rauhala et al., 2019).Moreover, patients with CSDH have a high recurrence rate after surgery (approximately 10-20%), requiring reoperation, which imposes a substantial financial, physical, and emotional burden on the patient (Kolias et al., 2014).Although the clinical use of dexamethasone combined with atorvastatin has been shown to improve CSDH effectively, these results need further confirmation (Wang et al., 2020).
Existing theoretical hypotheses regarding the mechanisms of CSDH formation and development include the osmotic imbalance theory, the excessive fibrinolysis theory, the local inflammatory response due to blood exudation (pontine vein injury due to mild brain injury), the theory of persistent exudation due to impaired angiogenesis, and dysfunction of meningeal lymphatic drainage (Liu et al., 2020;Yu et al., 2009).However, no consensus has been reached.Not all patients with head trauma will develop CSDH.Early imaging examinations such as cranial CT cannot identify this group of patients.Therefore, by analysing the components in peripheral blood, it is possible to identify patients at an early stage and block the progression of CSDH with certain medications, which can help reduce the severity of the disease and improve prognosis.Different blood cells at the hematoma site in CSDH patients are critical for hematoma development, and inflammatory cells, including neutrophils, lymphocytes, monocytes, and eosinophils, have been observed at the hematoma site in patients with CSDH and are closely associated with local inflammatory responses (Hua et al., 2016;Stanisic et al., 2012;Tempaku et al., 2015).Chen et al. also demonstrated that the proportion of lymphocyte and neutrophil counts in hematoma tissues was significantly higher than that in the peripheral blood of healthy individuals (Chen et al., 2022).Therefore, the different ratios of cell occupancy at hematoma sites may have profound significance on the disease progression of CSDH.
Accordingly, to elucidate the effects of different percentages of hematoma cells on CSDH, this study will combine single-cell sequencing to analyse the cellular composition of the hematoma site and screen valuable cellular pathways to improve the prognosis of disease development in CSDH.

Patient information
Three patients with CSDH were admitted to Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, from November to December 2022.The diagnosis of CSDH was established by taking into account imaging data, clinical signs and symptoms, and a history of head trauma.
Diagnostic Criteria: Patients with intracranial blood pooling between the arachnoid and the dura mater on CT, chronic occupying hematoma compressing the ventricles and the brainstem, producing symptoms such as vomiting, impaired consciousness, headache, and increased intracranial pressure (Hamou et al., 2022).
(a) Age: ≥ 18 years; (b) The patient had not experienced severe trauma, surgical treatment of inflammatory diseases or fever within months; (c) Patients with known diagnoses of dementia, ischemic and hemorrhagic stroke, subarachnoid haemorrhage, hydrocephalus, brain tumours, other neurological disorders, and those receiving anticoagulant and antiplatelet therapy were excluded from the CSDH group.In addition, patients taking glucocorticoids or anti-inflammatory drugs within 3 months, pregnant women or women in labour, and patients with previous inflammatory or infectious diseases should also be excluded (Familiari et al., 2023).

Treatment
All patients were diagnosed with CSDH preoperatively by head CT scan, and surgical findings confirmed the diagnosis.All patients underwent a single-port surgery under general anaesthesia.The thickest part of the hematoma was selected for drilling based on the patient's preoperative imaging data, and intraoperatively, the hematoma cavity was flushed with large amounts of saline until the drainage fluid was clarified, and a drainage tube was placed in the hematoma cavity.Finally, the hematoma cavity was connected to a closed drainage system for continuous drainage for 3 days.The patient was instructed to get up, walk, and move around as early as possible after removing the drain.

10× genomics single cell sequencing
Haematoma blood and peripheral blood (2 mL each) from the three patients were collected and then mixed with cDNA reverse transcription reagents, 10× barcode gel beads and oil for sequencing package were added to the Chromium Chip B chambers, and GEM (Gel Beads-in emulsion) was formed by using the double-cross system of the microfluidic device.The created GEMs were transferred to PCR tubes for reverse transcription.Single cells are labelled using unique molecular identifiers (UMIs), which are random hexanucleotides that allow for more precise quantification of the initial amount of mRNA molecules in a single cell.After that, the cDNA was amplified by PCR, and then the amplified cDNA library was used for library preparation and highthroughput sequencing on the NovaSeq sequencing platform.The Seurat Package was used to correct data bias factors such as dead cells, doublet, and low-quality cells.

Data downgrading
The uniform manifold approximation and projection (UMAP) algorithm and principal component analysis (PCA) algorithm are used for data dimensionality reduction and information presentation.The data were screened for principal components based on the EIbowplot metrics, and the FindVariableFeatures function was used to identify genes in the dataset that exhibited high variability between cells (i.e., genes that were highly expressed in some cells but poorly expressed in others).Genes with high variability in cell-to-cell expression were categorised using the selection.Method "nfeatures= 2000" parameter, and the samples were divided into cell populations using the Graphcluster unsupervised clustering algorithm.Fisher's exact test was used to score the significance of the cell type to which the cells belonged based on the marker genes documented in the CellMarker database (http://biocc.hrbmu.edu.cn/CellMarker/).their target genes in the target cell population, the regulatory genes of each transcription factor in the cell and the intensity of its regulation (AUCell Score) were calculated.Through this analysis, we can determine the regulation of transcription factors by different cell populations and look for cell cluster-specific transcription factors.

Cell communication analysis
Cell communication analysis software based on CellPhoneDB (Version 2.0) systematically analyses intercellular communication molecules.Membrane, secreted, and peripheral proteins were annotated for different temporal clusters.Significant mean and Cell Communication significance (p value<0.05)were calculated based on the interaction and the normalised cell matrix achieved by Seurat Normalization.

QuSAGE analysis
Gene set enrichment analysis was performed using the R package QuSAGE to obtain each gene set's enrichment status and enrichment significance.The variance inflation factor (VIF) algorithm was used to perform GSEA-like enrichment analysis for gene sets.Enrichment analysis was performed for different gene sets of different clusters to compare the enrichment differences of different clusters.

Single-cell quality control
Haematoma blood and peripheral blood from the three patients' raw data from single-cell nuclear sequencing were analysed using the scCancer analysis software package, and a high number of dead cells was visible by mitochondrial gene occupancy (Fig. 1A).After removing dead cells, cell debris, multicellular droplets and empty droplets, the parameter interval for the number of genes detected in a single cell (nFeature_RNA) was set to 309-1903.A total of 40,676 cells were detected before data QC, and 33,681 cells were detected after QC (see supplementary material 1).Single cells that met the requirements for single-cell quality control were sequenced, and 2000 genes with a significant degree of dispersion were selected as target marker genes for subsequent screening of cells (red genes in Fig. 1B).Principal component analysis of the PCA function was performed (Fig. 1C).ElbowPlot ranked the importance of PCs based on variance scores (Fig. 1D).JackStrawPlot results showed (Fig. 1E) that the top 10 PCs had significant importance.

Single-cell sequencing for cell sorting
Through dimensionality reduction and clustering of principal component (PC) data, Fig. 2A illustrates the cellular distribution within a UMAP coordinate system.The expression levels of the entire gene set (Fig. 2B) and mRNAs (Fig. 2C) within cellular puncta are visualized in UMAP plots and represented as heatmaps, showing the distribution of expression intensity.Employing the UMAP clustering approach, cells from the PBMC and CSDH cohorts were stratified into 25 distinct clusters labelled from 0 to 24, as depicted in Fig. 2D.

Analysis of pathways of influence in blood cells of CSDH patients
To explore the differences in pathways between hematoma sites in CSDH patients and peripheral blood mononuclear cells (PBMCs), this study conducted analyses of gene ontology (GO) (Fig. 4A) and pathway enrichment (Fig. 4B) for genes that varied between CSDH and PBMCs.The results suggested that in CSDH, "neutrophil degranulation", "cytokine-mediated signalling pathway", "negative regulation of apoptotic process", "protein processing in endoplasmic reticulum", "antigen processing and presentation", and other pathways were significantly activated.Heatmaps depict the regulation of pathways across different cellular subpopulations, determined by the intensity of factors that influence these pathways (Fig. 4C).In CDSH clots, cDC2 cells are mainly associated with "Intestinal immune network for IgA production", "Antigen processing and presentation", "Haematopoietic cell lineage", "Th1 and Th2 cell differentiation", and "Th17 cell differentiation".M2 macrophages were mainly associated with "Ferroptosis", "Collecting duct acid secretion", "Lysosome", "Cholesterol metabolism", "PPAR signalling pathway", "Complement and coagulation cascades", and "Mineral absorption".Complement and coagulation cascades", and "Mineral absorption" pathway activation.

Pathways primarily affected by cDC2 in CSDH
To investigate the role of cDC2 enrichment in CSDH, this study first performed pathway analysis of differentially expressed genes in cDC2 cells (Fig. 5A), and the results suggested that in cDC2 cells, "Reactive Oxygen Species", "Oxidative phosphorylation", "MTORC1 signalling" and other signalling pathways were significantly regulated.QuSAGE analysis (Fig. 5B-C) showed that the "Reactive Oxygen Species" pathway was mainly associated with the upregulation of the LSP1, MGST1, GPX4, and FTL genes as well as the downregulation of the HMOX2, SOD2, and SOD1 genes.The "MTORC1 signalling" pathway was mainly associated with the upregulation of the ENO1, ACTR2, BHLHE40, and MTHFD2 genes and the downregulation of the ELOVL6, BTG2, TUBA4A, and CXCR4 genes.

Analysis of cellular communication signals between cDC2s and M2 macrophages
To investigate the influence of intercellular regulation in CSDH blood clots, all cells in CSDH were analysed by CellPhoneDB software, and the results suggested (Fig. 7A) that there was strong intercellular communication transmission between cDC2 cells and M2 macrophages.Although monocytes were also suggestive of more genes that could constitute a linkage, the number of monocyte cells was mainly in PBMCs, which accounted for a smaller number in CSDH and, therefore, was not considered a primary target for analysis.Then, the cell-to-cell gene regulatory relationships analysed by CellPhoneDB software were analysed to determine that when cDC2 acts as a ligand cell, it can interact with LRP1 in M2 macrophages by secreting PDGFB to achieve interaction with LRP1 in M2 macrophages, which in turn affects the changes in gene and protein expression downstream of macrophages.

Discussion
As the prevalence of CSDH increases in the ageing population, so does the socioeconomic burden of CSDH on older adults and the healthcare system (Rauhala et al., 2019;Yang and Huang, 2017).It is hypothesised that inflammation is a critical factor in the development of CSDH.Inflammatory cells, including neutrophils, lymphocytes, macrophages and eosinophils, play an essential role in this process (Hua et al., 2016;Stanisic et al., 2012;Tempaku et al., 2015).Fan et al. concluded that eosinophils are present at higher levels in CSDHs than in PBMCs and can mediate immunity and inflammation (Chen et al., 2022).However, the above studies only targeted a specific cell and did not comprehensively analyse the cell types at the hematoma site of CSDH to understand the characteristics of the composition of the cells at the hematoma site.
This study used 10× Genomics single-cell sequencing to detect mRNA changes in each cell in the sample.By carrying out single-cell sequencing, this study preliminarily determined that there are 17 major subpopulations of CSDH and PMBC cells: pDCs, CD8 T cells, CD4 T cells, MigDCs, cDC2s, cDC1s, plasma cells, neutrophils, naive B cells, NKs, memory B cells, M2 macrophages, CD8 Teffs, CD8 MAITs, CD4 Tregs, CD19 B cells, and monocytes.The present study found that cDC2s, M2 macrophages, and T lymphocytes were upregulated in CSDH and were the dominant fraction.
T cells can kill bacterial and viral infected cells or cytotoxins by releasing cytokines, which may exacerbate brain inflammation after cerebral infarction (Harrison et al., 2008).Cytokines and chemokines released by helper T cells may increase the expression of vascular adhesion molecules and attract other immune cells to the brain, leading to widespread apoptosis (Arumugam et al., 2005).In neurologic-like diseases, T lymphocytes have been shown to enter the brain and release cytokines/chemokines and superoxide, which can significantly contribute to neuronal damage (Brait et al., 2012).The results of this study also confirm the presence of inflammatory cell infiltration during the pathogenesis of CSDH.On the other hand, CSDH also belongs to the neurological class of disorders of the brain and is closely related to the inflammatory response.
Conventional dendritic cells (cDCs) are key initiators and modulators of adaptive immune responses (Trumpfheller et al., 2012).Maintenance of cDC homeostasis is essential for regulating immune responses, and its dysfunction can lead to infections and autoimmune diseases (Birnberg et al., 2008).Differentiation of cDCs begins in the bone marrow and differentiates into cDC1s and cDC2s in the periphery (Maraskovsky et al., 2000).In contrast, in the present study, we focused on the significant enrichment of cDC2s at the CSDH hematoma site, and therefore, we focused on the regulatory mechanism of cDC2s.cDC1s and cDC2s typically activate antigen processing and presentation pathways to stimulate CD4 and CD8 T-cell activation (Gutiérrez-Martínez et al., 2015).In addition, cDC2s have been shown to promote Th17, Th2, and Treg cell differentiation (Gao et al., 2013;Mayer et al., 2017;Persson et al., 2013).In contrast, when CSDH was analysed for pathway enrichment in the present study, it was found that Th2/Th17 differentiation and antigen presentation were significantly activated in the cDC2 cell population, which corresponded to the significant enrichment of cDC2 in CSDH.This suggests that T-cell activation and further brain tissue damage may be significantly related to the presence of cDC2s and that reducing the proliferation of cDC2s and their recruitment at focal sites may help alleviate disease progression in CSDH.
In addition, cDC2 in CSDH was analysed separately, and the results suggested that "reactive oxygen species" and the mTOR signalling pathway were significantly activated.ROS can promote the DC maturation process, substantially enhancing the secretion of large amounts of IL-1β and IL-6 for downstream T-cell activation (Qin et al., 2020).Elevated ROS is an important marker of oxidative stress (Schieber and Chandel, 2014), and activation of the oxidative stress pathway stimulates the activation of cDC2 cells and promotes their proinflammatory response, which is closely related to the activation of the inflammatory factor NF-κB pathway (Batal et al., 2014).Through the analysis of transcription factor-related genes, NF-κB, RELB and REL genes were significantly upregulated in cDC2 cells, which provided essential prerequisites for the proinflammatory and pro-T-cell activation functions of cDC2 cells.The activation of the mTOR pathway is essential for the maturation and activation of DCs, and the absence of mTOR results in the death of DCs and limits the activation response of DCs (Wang et al., 2013).The activation of ROS and mTOR signalling pathways is conducive to cDC2 activation, activation of the inflammatory response and further activation of T lymphocytes.
M2-type macrophages were detected only in the CSDH group, suggesting that M2-type macrophages may be significantly recruited to the hematoma site in CSDH patients.M2 macrophages are highly phagocytic and produce extracellular matrix components that promote angiogenesis and secretion of the chemokine IL-10 ( Bohlson et al., 2014;Fuentes et al., 2010).In addition to pathogen defence, removing apoptotic cells by M2 macrophages reduces the inflammatory response and promotes wound healing (Ferrante and Leibovich, 2012).Ferroptosis and lysosome pathway activities were significantly affected in M2-type macrophages in the present study.For example, the anti-ferroptosis-related genes HMOX1, FTL, FTH1, SLC7A11 and GPX4 were upregulated considerably in M2-type macrophages in CSDH, suggesting that M2-type macrophages at hematomas are more favourable for ferroptosis resistance.Here, we recommend that the recruitment of M2 macrophages at the hematoma site may be a form of self-protection carried out by the organism to phagocytose, for example, the carcasses of dead cells such as erythrocytes and to reduce the localised excessive inflammatory response through anti-inflammatory effects.The upregulation of anti-ferroptosis-related genes may be related to the phagocytosis of many erythrocytes by M2-type macrophages.Erythrocytes contain large amounts of ferroportin, and macrophages are prone to iron overload after phagocytosis (Recalcati and Cairo, 2021), inhibiting M2-type macrophage activity.
In addition, the lysosome pathway is significantly upregulated in M2type macrophages.Lysosomes are highly acidic enzyme-containing organelles capable of denaturing and hydrolysing a library of biomolecules, including proteins, lipids, sugars, and nucleotides.Two main mechanisms provide lysosomes with degradative substances: heterophagy (i.e., degradation of extracellular substances) and autophagy (i.e., degradation of intracellular substances).In xenophagy, cells take up extracellular contents through generalised processes such as receptormediated endocytosis and cytophagy or specific functions unique to macrophages (e.g., phagocytosis) (Sergin et al., 2015).Thus, the lysosome pathway is, to a certain extent, favourable for macrophage phagocytosis, e.g., phagocytosis of necrotic or aged erythrocytes.
The intercellular communication relationship in CSDHs was further analysed by CellPhoneDB software, and the results showed a close intercellular communication relationship between cDC2 cells and M2type macrophages.In contrast, as an exocytosed protein, IL1β appears to function as a signalling bridge between cDC2s and M2-type macrophages.As one of the hallmark proteins for dendritic cell maturation (Wu et al., 2017), it can bind to the IL1 receptor (IL1R) on the surface of M2-type macrophages to achieve functional regulation of M2-type macrophages, such as immune infiltration (Watari et al., 2014) and anti-inflammatory effects (Hagemann et al., 2008).Silencing IL1R in macrophages has been reported to induce polarisation toward the M1 phenotype of macrophages (Hagemann et al., 2008).In addition, in a study by Klaver et al., it was demonstrated that M2 phenotypic macrophages have higher levels of IL1R expression, which seems to be the body's way of controlling the excessive inflammatory response in response to the inflammatory signal IL1β through the polarization of M2 macrophages (Klaver et al., 2022).This then explains the significant enrichment of proinflammatory cDC2 cells and anti-inflammatory M2 macrophages at the same time at the site of the CSDH hematoma: on the one hand, due to local haemorrhage and the production of clots, the body's cells produce an inflammatory immune response; on the other hand, it is also intended to clear necrotic cells at the same time to prevent excessive inflammatory responses from causing further damage.This study demonstrates that the single-cell composition of intracranial hematomas can aid in early identification and potentially lead to the development of targeted drugs to intervene in this process, thereby reducing the severity of the disease and improving prognosis.However, this study still has significant shortcomings: 1, the clinical samples are small, so there may be variability; 2, the downstream pathway and the observed phenomena for further experimental clarification, such as at the cell or animal levels for validation.

Conclusions
In summary, the following conclusions were obtained in this study by single-cell sequencing analysis of blood and PBMCs from lesion sites of CSDH patients: 1, The presence of more cDC2 and M2 macrophages recruited from the lesion sites of patients with CSDH and the upregulation of the T-cell ratio may be a red flag for further brain damage; 2, ROS, a marker of oxidative stress, was significantly upregulated in cDC2 cells and may mediate the functioning of transcription proteins of inflammatory factors such as NF-κB, which in turn mediates T-cell activation; 3. cDC2 may regulate M2 macrophage immune infiltration and anti-inflammatory activity by secreting IL1β and binding to M2 macrophage IL1R protein.Therefore, reducing the recruitment of cDC2s at hematoma sites in CSDH patients and upregulating the percentage of M2 macrophages may be beneficial in suppressing local immune responses and reducing the risk of further immune cell attacks on brain tissue.

Fig. 1 .
Fig. 1.Single-cell quality control.(A) nFeature_RNA is the number of genes detected in each cell.nCount_RNA is the total number of mRNA molecules detected in the cell.Percent.mt is the ratio of mitochondrial gene expression to all gene expression in a cell.(B) The coefficient of variation dispersion of genes, red, indicates genes with high scattering and significant coefficients of variation in different kinds of cells.(C) PCA principal component analysis.(D) ElbowPlot scored the top PCs for variance and ranked them in order of importance.(E) JackStrawPlot analysed the first 6 PCs for significance.
Correlation of expression of single cell subpopulations with focal transcription factors based on SCENIC (Single Cell Regulatory Network Inference and Clustering) inference.Based on the transcription factor target database to analyse the expression of transcription factors and

Fig. 2 .
Fig. 2. Single-cell clustering based on PC principal components.(A) Distribution point positions of all cells in two-dimensional space after dimensionality reduction of PC principal components.(B) Expression intensity of nFeature RNA expression levels in cells at each site.(C) Expression intensity of mRNA expression levels in cells at each site.(D) Cells were clustered by UMAP dimensionality reduction clustering.

Fig. 3 .
Fig. 3. Cell subtype annotation and percentage analysis.(A) Distribution position of each cell subpopulation in the CSDH and PBMC groups in the UMAP clustering axis and results of individual cell subpopulation clustering analysis.(B-C) Pie chart of the percentage of cell subpopulations in each sample.(C) The percentage of each cell subpopulation in PBMCs and CSDHs.

Fig. 4 .
Fig. 4. Pathway analysis of differential gene regulation in CSDH and PBMC.(A) Results of GO enrichment analysis of differentially expressed genes.(B) Results of pathway enrichment analysis of differentially expressed genes.(C) The intensity of influence of each cell subpopulation in critical pathways of influence on CSDH is plotted on a heatmap, with red boxes representing the significant pathways of influence of cDC2 or M2 macrophages on CSDH.

Fig. 5 .
Fig. 5. Pathways analysing the main effects of cDC2 in CSDH.(A) Pathways mainly regulated by cDC2 cells in CSDH were analysed by Pathway.(B) Genes are involved in regulating the "Reactive Oxygen Species" pathway in cDC2 cells.(C) Genes regulate the "MTORC1 signalling" pathway in cDC2 cells.

Fig. 6 .
Fig. 6.Pathways analysing the main effects of cDC2 in CSDH.(A) Pathway analysis of the pathways mainly regulated by M2 macrophages in CSDH.(B) Genes are involved in regulating the "Ferroptosis" pathway in M2 macrophages.(C) Genes are involved in regulating the "Lysosome" pathway in M2 macrophages.

Fig. 7 .
Fig. 7. Analysis of cell communication signals between cDC2s and M2 macrophages.(A) CellPhoneDB analysis of the gene heatmap of cell-to-cell communication.(B) CellPhoneDB analysis of a bubble map of the strength of gene influences on communication between cDC2s and other cells.

Table 1
UMAP cell attribute table.

Table 2
Prediction of the binding relationship between transcription factors and the target gene PDGFB.
Q.Zhang et al.