Comprehensive analysis of immune cell infiltration and role of MSR1 expression in aneurysmal subarachnoid haemorrhage

Abstract Aneurysmal subarachnoid haemorrhage (aSAH), resulting from the rupture of intracranial aneurysms, can yield high mortality and disability. This study aimed to explore the immune infiltration of aneurysmal tissues and investigate a novel mechanism underlying aSAH. We downloaded datasets containing expression profiles of aneurysmal and normal arterial tissues from the online database. Then a comprehensive bioinformatic strategy was conducted to select the biomarkers of aneurysmal tissues. Two calculation algorithms were performed to identify the unique immune characteristics between aneurysmal tissues and normal arteries. Double immunofluorescence staining was used to investigate the role of pathway‐related proteins in the inflammatory process after aSAH. Six microarray datasets were integrated, and another RNA‐sequencing dataset was used as the validation dataset. Functional enrichment analysis of the differentially expressed genes indicated that immune‐related processes were closely related to the progression of aSAH. We then performed immune microenvironment infiltration analysis, and the results suggested macrophages were abnormally enriched in aneurysmal tissues. Core gene MSR1 was filtered through a comprehensive bioinformatic strategy. Our analysis suggested that MSR1 might be associated with macrophage activation and migration. Our study elucidated the impact of macrophage and MSR1 on aSAH progression. These findings were helpful in gaining insight into the immune heterogeneity of aneurysmal tissues and normal arteries, and in identifying patients who might benefit from immunotherapy.


| INTRODUCTION
Aneurysmal subarachnoid haemorrhage (aSAH) is a deadly cerebrovascular disease. [1][2][3] It is estimated that more than 25% of patients with aSAH died in the hospital, and more than 40% have persistent neurological deficits. 2 Moreover, patients who survive the initial haemorrhage are susceptible to cerebrovascular diseases and have mortality rates that exceed the general population. 3,4 For example, the recurrence of aSAH in survivors was estimated to be 15-times higher than that in the general population. Thus, it is critical to recognize the cellular and molecular differences between aneurysmal tissues from aSAH patients and normal intracranial arterial tissues, which could offer a theoretical basis for the high recurrence of cerebrovascular disease in patients with aSAH.
Inflammatory reactions are involved in the formation, progression and deterioration of aSAH. 5 Once the intracranial aneurysm (IA) ruptures, immune cells, haemoglobin and cytokines enter the central nervous system and mediate inflammatory responses. 6,7 Xing Wang and Dingke Wen contributed equally to this work as first authors.
The abnormal immune responses after aSAH have been proposed to be a critical component in the development of secondary brain injury. [8][9][10] At this point, researchers are inclined to believe that aSAH is rather a chronic disease that is caused by a massive systemic inflammatory response than a one-off event.
Among these immune reactions, evidence indicates that macrophages play a particular role in SAH-induced inflammation. 11,12 However, previous studies in the same field were conducted with small sample sizes, and their findings need to be confirmed in large cohorts and comprehensive bioinformatics analyses. To identify the potential biomarkers related to the pathogenesis of aSAH that might offer novel molecular features, we performed this study to clarify the cellular and molecular processes implicated in the development of aSAH, which might facilitate the timely diagnosis and subsequent treatment for this fatal disease. Our findings may also provide a theoretical basis for the high incidence of intracranial vascular disease after ASAH.

| Data acquisition and integration
We systematically searched Gene Expression Omnibus (GEO) database with the following keywords, 'subarachnoid haemorrhage', 'intracranial aneurysm', 'cerebral aneurysm' and so forth. Ultimately, seven available transcriptome data of tissue samples from aneurysmal walls and normal intracranial arteries were obtained; six were microarray datasets (GSE6551, GSE13353, GSE15629, GSE26969, GSE54083 and GSE75436). The detailed information of each dataset was presented in Table S1. Most studies collected ruptured IA tissues within 24 hours after initial bleeding. The integrated dataset contained 27 IA tissues from patients with aSAH and 38 normal arteries.
The R package 'sva' was applied to remove the batch effect. We also obtained one RNA-sequencing (RNA-seq) dataset (GSE122897) as the validation dataset, which contained 23 aneurysmal tissues from patients with aSAH and 16 normal arteries tissues. Additionally, five pairs of aneurysmal tissues from aSAH patients and superficial temporal arteries from patients who were free of vascular diseases during surgery were obtained from our hospital. This study was based on the ethical principles of the Helsinki Declaration and approved by the ethical committee of the West China Hospital of Sichuan University.

| Immunofluorescence assay
The detailed procedures have been described in the previous study. 13 Briefly, tissues were sliced into 5-μm-thick frozen sections. Next, after blocking for 1 hour with 5% bovine serum albumin, primary antibodies, including rabbit anti-MSR1 (ABclonal, A2401), mouse anti-IBA1 (Abcam, ab283319), rabbit anti-CD86 (ABclonal, A19026) were incubated with the samples overnight at 4 C. Then, the primary antibodies were washed three times by TBS. Secondary antibodies, including the goat anti-rabbit antibody (AS011, ABclonal) and the goat anti-mouse antibody (AS001, ABclonal) were placed on the slides and incubated for 1 hour at 25 C. The antibody dilution ratio was following the online instructions. The sections were subsequently stained with DAPI for nuclear labelling. Images were analysed by Image J 1.4 software.

| Enrichment analysis
Differentially expressed genes (DEGs) between two clusters were defined as those genes with adjusted p values <0.05 and absolute values of log2(foldchange) >0.50. DEGs were obtained from microarray datasets and RNA-seq dataset using the 'limma' and 'DESeq2' packages, respectively. The Gene Ontology (GO) pathway enrichment of DEGs was conducted with the DAVID database (https://david. ncifcrf.gov/). 14 A false discovery rate (FDR) value <0.05 was deemed statistically relevant. The procedures were achieved by 'clusterProfiler' package, and visualization was completed by the 'ggplot2' package.

| Gene set variation analysis
We downloaded the signature 'c7.all.v7.4.symbols_immune' from the Molecular Signatures Database (MSigDB). 15 Differential pathways between different groups were screened by the 'limma' package. The results were presented in the heatmap. Pathways with adjusted p values smaller than 0.05 were deemed as statistically relevant. The procedures were achieved using the R package 'GSVA'.  16 In addition, Xcell algorithm was applied to determine the proportion of infiltrating immune cell subsets in different tissues. This tool could provide independent enrichment scores across 64 immune and stroma cell types based on intrinsic algorithms. 17

| Additional analyses
The random forest approach was used to identify the most critical candidate genes associated with distinct characteristics. In brief, a set of concerned genes were employed as the input for the construction of prediction models. Preferred genes were chosen depending on the mean decrease Gini and mean decrease accuracy. To compare the predictive sensitivity of the selected genes, area under the curve (AUC) values obtained from the receiver operating characteristic (ROC) curves were calculated. The above analyses were done with the 'randomForest' and 'pROC' packages.
The Wilcoxon test compared the expression level of genes between aneurysmal tissues and normal intracranial arteries. We applied the Pearson correlation test to evaluate the linear relationship between different genes. The Spearman correlation analysis was constructed to determine the relationships between selected genes and various immune cells. The statistical analyses were carried out with the R software (version 4.0.5). All tests were conducted to be twosided, and a p-value smaller than 0.05 revealed a statistically significant difference.

| Functional enrichment analysis of DEGs
The cutoff criteria determined according to an adjusted p value smaller than 0.05 and an absolute value of log2(foldchange) larger than 0.5. Overall, after data merging and removal of batch effects, a total of 1649 genes were regarded as DEGs, among which 846 were upregulated and 803 genes were downregulated. We performed hierarchical clustering analysis to visualize the distinguishment ability of DEGs between aneurysmal tissues and control arteries ( Figure 1A). Volcano plot was used to exhibit a group of genes that were significantly highly and lowly expressed ( Figure 1B Figure 1F. Results of GSEA analysis suggested that the immune system and innate immune system were upregulated in aneurysmal tissues compared with normal arteries tissues ( Figure 1G-I).

| Immune profiles of aneurysmal and arteries tissues
Given that the above analysis revealed that immune-associated fea-

| Identification of key genes based on random forest model and Xcell immunescore
Random forest was conducted to filter the genes related to the above immune infiltration profiles. Two ranking methods were used to obtain the intersection ( Figure 3A). For the concerned genes, we compared their correlation coefficient with stromascore and immunescore derived from Xcell algorithm. As shown in Figure 3B, MSR1 was significantly and positively correlated with immune score but negatively correlated with stroma score. To verify the predictive power of the selected gene, ROC curves were developed. Then AUC values were obtained to assess the predictive consistency and accuracy of the selected gene ( Figure 3C,D).
Eventually, MSR1 was chosen as a candidate gene for further investigation. The 95% confidence interval for AUC was 0.876-1.000, and the predicted AUC was 0.942, revealing the stability of MSR1 in discriminating aneurysmal tissues from normal arteries. The expression profile of MSR1 in the integrated and validation datasets was shown in Figure 3E,F. MSR1 was highly expressed in aneurysmal tissues.

| Association of MSR1 and immune cell infiltration
Infiltration of 34 types of immune cells in the integrated dataset was analysed using the Xcell tool. The association between MSR1 and the infiltration landscape of immune cells was evaluated by the Spearman correlation test. Among all the immune cells, MSR1 expression was strongly correlated with macrophages (correlation coefficient: 0.652; p < 0.001; Figure 4A). The expression level of MSR1 was also significantly associated with macrophages group 1 (correlation factor: 0.474; p < 0.001) and group 2 (correlation factor: 0.449; p < 0.001).
Similar results were observed in the RNA-seq dataset ( Figure 4B-D).
The expression profiles of MSR1 and cell markers of macrophages were investigated by immunofluorescence double-staining. The results revealed that almost all MSR1-positive cells were also stained for IBA1 and CD86 ( Figure 4E-G). Gene set variation analysis indicated macrophage-related pathways activated along with the upregulation of MSR1 expression ( Figure 4H).

| Biofunction prediction of MSR1
To further discover the potential mechanisms of macrophage regulation by MSR1, DEGs between MSR1 low and high-expression groups were obtained in the combined dataset and the validation dataset ( Figure 5A,B). The intersection of DEGs was then obtained for further functional enrichment analysis ( Figure 5C). Considering the above results, GO analysis was performed to predict the biological functions of MSR1 related to macrophages ( Figure 5D). We also performed GO analysis using the intersection of DEGs ( Figure 5E). The results demonstrated that higher MSR1 expression was associated with macrophage activation and migration.
Additionally, various immunomodulators were downloaded from the TISIDB database. Pearson correlation test was then conducted to investigate the correlations between MSR1 and each immunomodulator. Immunomodulatory factors with correlation coefficients larger than 0.5 with MSR1 were CD80, CD86, CXCL16, HAVCR2, BTNL2, CCL18, IL2RA, CSF1R, TNFRSF14, CXCR4 and ICOS ( Figure 5F). These factors were highly expressed in aneurysmal tissues, and the results were validated by the RNA-seq dataset ( Figure 5G,H).

| DISCUSSION
This study indicated that macrophage-mediated inflammation played a critical role in the progression of IAs. Our findings thus provided a theoretical basis for developing drug therapies to inhibit macrophagemediated inflammatory responses. Furthermore, MSR1 was identified through extensive bioinformatic analysis and considered to be closely involved in the immune microenvironment. Since MSR1 might impact the progression of aSAH by regulating macrophages, novel nanomaterials could be expected to intervene in this process. [18][19][20] However, although we have discovered and validated the predictive value and potential biological function of MSR1 in aneurysmal tissues, future in vivo and in vitro research is still warranted to explore the upstream and downstream signalling pathways of MSR1.
It appears that IA formation, progression and rupture cannot be completely interpreted by a single mechanism. 21 However, it is widely  H) The expression profiles of a set of immunomodulators with correlation coefficients larger than 0.5 between aneurysmal tissues and healthy arteries in the integrated dataset and validation dataset. Statistical significance was tested by the Wilcoxon test (****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05). AT, aneurysmal tissues; DEGs, differentially expressed genes; GO, gene ontology; HA, healthy arteries acknowledged that inflammation has a role in the formation and development of IA. Several studies have demonstrated that inflammatory cells, particularly macrophages, infiltrated walls of IAs in humans. 22,23 It seems that elevated levels of inflammation in aneurysms are related to the destruction and rupture of the aneurysm wall.
In addition, recent evidence shows excessive neuroinflammation following aSAH results in second brain injury, leading to functional impairment and unfavourable outcomes. 24,25 Moreover, the administration of pharmacological agents with anti-inflammatory properties decreases the risk of developing SAH from ruptured IAs. Commonly used medications in clinical practice are statins and non-steroidal antiinflammatory drugs. 26 Also, recent studies have observed the favourable effects of tetrahedral DNA in suppressing neuroinflammation and in treating neurological diseases. [27][28][29][30][31] It is therefore important to thoroughly investigate the involve- Macrophages of monocytic lineage represent an essential element of the innate immune system in the nervous system. They play a crucial part in the initiation, regulation and termination of inflammation due to their involvement with antigen presentation and cytokine production. 32 As the majority of cellular components of leukocyte infiltration are associated with the risk of rupture. 33  and pro-inflammatory responses. CXCL16 is a scavenger receptor on the surface of macrophages. This gene has been recognized to be involved in the pathophysiological process of atherosclerosis. 44 Overexpression of CXC16 could also enhance apoptosis. 45 HAVCR2 is also involved in regulating the activation of macrophages. 46

| CONCLUSIONS
Until now, the underlying mechanisms of IA progression and rupture have not been elucidated in detail. Our analysis provided extended information on the immune landscape of aneurysm tissues after aSAH. We also identified MSR1 as one of the critical genes involved in inflammatory responses. Our study may help elucidate the impact of macrophage on IA progression and identify biomarkers and therapeutic targets. Further evaluation of the role of MSR1 in brain injury and complications after aSAH is necessary through the application of genetic deletion in animal models of SAH to provide informative treatment options for patients with aSAH.