Background
Stroke is a serious disease with high morbidity and mortality. Stroke is a leading cause of lifelong disability in adults worldwide, and ischaemic stroke accounts for more than 80%. With the increasing severity of social ageing, the acceleration of urbanization, the persistence of cardiovascular risk factors and the prevalence of unhealthy lifestyles, the burden of ischaemic stroke is rapidly increasing [
1]. Currently, reliable diagnostic methods for ischaemic stroke mainly rely on imaging methods, such as computed tomography (CT) [
2] and magnetic resonance imaging (MRI) [
3], which are time-consuming and laborious. Meanwhile, the traditional and effective treatment strategy is to carry out drug thrombolysis and interventional thrombolysis or thrombectomy as soon as possible after the occurrence of ischaemic stroke. These treatments not only require immediate treatment and intervention but also are significantly correlated with an increased risk of fatal bleeding, such as intracerebral haemorrhage and gastric bleeding [
4]. Therefore, the early diagnosis, prevention and treatment of ischaemic stroke are facing serious challenges. There is an urgent need to further explore potential reliable serum biomarkers significantly correlated with ischaemic stroke.
At present, a large number of studies have suggested that traditional risk factors including hypertension, hyperlipidaemia and hyperglycemia are significantly associated with several diseases, such as cancer [
5‐
7] and ischemic cardiovascular and cerebrovascular diseases [
8,
9]. However, in addition to these common cardiovascular risk factors, the role of inflammation or immune related mechanisms in ischemic cardiovascular and cerebrovascular diseases has received more and more attention. Krishnan et al. found that inflammatory cell infiltration can effectively stimulate and lead to a strong immune response, resulting in dysfunction in the immune microenvironment in the central nervous system and ultimately further leading to the deterioration of patients with cerebral ischaemia [
10]. Smith et al. also suggested that proinflammatory cytokines, especially interleukin-1 (IL-1), play a key role in the early inflammatory response after ischaemic stroke, and these inflammatory responses are associated with poorer clinical outcomes in patients with ischaemic stroke [
11]. In recent years, immunotherapy has become a novel method to treat cancer [
12] and cardiovascular disease [
13]. In addition, some studies have revealed that immune regulation can effectively delay the progression of ischaemic stroke, restore neurological function and improve the prognosis of patients, further emphasizing the importance of maintaining immune microenvironment homeostasis for protecting the central nervous system [
14,
15]. It has been demonstrated that specific inhibitors of IL-1β can delay the progression of atherosclerosis by inhibiting specific inflammatory pathways associated with atherosclerotic plaque formation [
16] and effectively reduce the risk of major cardiovascular adverse events and cardiovascular death [
17]. Meanwhile, IL-1 receptor antagonists have been found to be effective in reducing peripheral inflammation in acute ischaemic stroke, thereby improving clinical outcomes in these patients [
11]. Therefore, in addition to the current conventional treatment methods, immunoregulatory therapy is expected to be a practical alternative treatment method that is worthy of further in-depth research. In recent years, CIBERSORT, a widely used analysis tool, can use RNA-seq data or microarray data to investigate the infiltration pattern of immune cells and evaluate the proportion of 22 types of immune cells in samples [
18]. However, few studies investigated the infiltration pattern of immune cells and the identification of immune-related genes in the peripheral blood of ischaemic stroke patients. Therefore, evaluating the infiltration pattern of immune cells in the peripheral blood of ischaemic stroke patients could help further clarify the immune-related molecular mechanism involved in ischaemic stroke.
With the continuous promotion of gene chip technology, weighted gene coexpression network analysis (WGCNA), a powerful systematic biological method used to analyse network relationships and molecular mechanisms, is widely used to analyse massive amounts of gene expression profile data [
19]. WGCNA is often used to identify coexpressed gene modules and further explore the relationship between gene modules and interesting sample features [
20]. More recently, machine learning has significantly improved the predictive and accuracy value of key genes identified based on microarrays and next-generation sequencing data [
21]. The least absolute shrinkage and selection operator (LASSO) regression and support vector machine-recursive feature elimination (SVM-RFE) algorithm are the most widely used machine learning methods to identify key genes [
22]. However, few studies have combined WGCNA, LASSO and SVM-RFE to identify the key genes related to ischaemic stroke.
In the current research, the GSE22255 and GSE58294 datasets were used as the training set, and the GSE16561 dataset was used as the testing set; all datasets were downloaded from the Gene Expression Omnibus (GEO). By removing the interbatch differences between the GSE22255 and GSE58294 datasets, the 25% genes with the highest expression variance were selected for a WGCNA. The potential biological functions of the genes in several key modules that were significantly associated with ischaemic stroke were analysed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. The key genes significantly associated with ischaemic stroke were identified by a LASSO regression combined with SVM-RFE methods. Then, we explored the infiltration pattern of immune cells in peripheral blood from ischaemic stroke patients and further calculated the relationship between several key genes and 22 types of immune cells. Meanwhile, the expression of key genes and their diagnostic efficiency were further validated in the training set, testing set and validation samples.
Discussion
In the current research, GSE22255 combined with GSE58294 as training datasets were downloaded from the GEO database and analysed using a WGCNA. Then, three modules (pink, brown and cyan) were identified to be significantly associated with ischaemic stroke. Eight hub genes (ADM, ANXA3, CARD6, CPQ, SLC22A4, UBE2S, VIM and ZFP36) were revealed to be significantly correlated with ischaemic stroke by a LASSO logistic regression and SVM-RFE machine learning methods. The CIBERSORT results revealed decreased infiltration of naive CD4 T cells, CD8 T cells, resting mast cells and eosinophils and increased infiltration of neutrophils, activated memory CD4 T cells and M0 macrophages in the ischaemic stroke patients. The external validation combined with the RT‒qPCR analysis revealed that the expression levels of ADM, ANXA3, SLC22A4 and VIM were significantly increased in the patients with ischaemic stroke and that these key genes were positively correlated with M0 macrophages and neutrophils and negatively correlated with CD8 T cells. The ROC analyses based on the training set, validation set, and our clinical samples showed that the ADM, ANXA3, SLC22A4 and VIM genes remained highly effective in distinguishing the ischaemic stroke patients from the normal subjects. These results suggest that the ADM, ANXA3, SLC22A4 and VIM genes play a key role in the pathological process of ischaemic stroke.
Previous research has proven that the expression levels of adrenomedullin (
ADM) are significantly increased in ischaemic cortical neurons induced by ischaemic injury in patients with ischaemic cerebrovascular disease [
29]. Ischaemic cerebrovascular disease involves not only ischaemic brain cell injury but also arterial injury. Shinomiya et al. found that even in patients with fewer risk factors, the severity of atherosclerosis was significantly associated with elevated levels of mature ADM [
30]. Ishikawa et al. proved that the expression levels of ADM were significantly increased in patients with unstable coronary artery disease compared with those in patients suffering from stable coronary artery disease, and ADM may participate in the instability of atherosclerotic plaque in the form of autocrine or paracrine [
31]. Matthew et al. proved that ADM acts as an independent predictor of major adverse cardiovascular events (MACEs) in patients suffering from heart failure and acute myocardial infarction (AMI), and the quantification of the ADM levels may help improve the risk stratification of heart failure and myocardial infarction [
32]. In addition, a compelling study showed that elevated ADM levels were significantly associated with the severity of neurological damage, higher mortality, and poorer outcomes in patients with ischaemic stroke [
33].
Through a comprehensive search of the NCBI GENE database, we revealed that Annexin A3 (
ANXA3, also known as
ANX3; HGNC: 541, gene ID: 306, OMIM: 106,490) is located on chromosome 4q21.21 (exon count: 14), acts as a member of the annexin family, and plays a crucial role in regulating multiple biological processes, such as inflammatory responses, cell proliferation, apoptosis and tumorigenesis [
34]. Junker et al. [
35] and Kessler et al. [
36] reported that the expression levels of ANXA3 were significantly upregulated in the infarcted area after cerebral ischaemia injury in rats. Hua et al. proved that silencing the ANXA3 gene can promote the repair and healing of ischaemic myocardium by activating the PI3K/Akt signalling pathway in rats with AMI [
37]. Moreover, Min et al. found that miR-18b can protect cerebral ischaemia‒reperfusion injury by activating the PI3K/Akt signalling pathway by inhibiting the expression of ANXA3 [
38].
Solute carrier family 22 member 4 (
SLC22A4, also known as
OCTN1; DFNB60, gene ID: 5583, HGNC: 10,968, OMIM: 604,190) is located on chromosome 5q31.1 (exon count: 11) and encodes an organic cation transporter across the plasma membrane of epithelial cells. Previous research showed that the SLC22A4 variant, as an inflammation-related gene polymorphism involved in the innate immune response, is significantly correlated with an increased susceptibility to inflammatory bowel disease (IBD), Crohn's disease (CD) and ulcerative colitis (UC) by changing the transcription and function of organic cation transporters [
39‐
41]. Meanwhile, the genetic polymorphisms SLC22A4 rs2073838 and rs3792876 were reported to be significantly associated with rheumatoid arthritis (RA) in the Japanese population [
42] and Chinese population [
43]. Tokuhiro et al. also suggested that SLC22A4 was significantly overexpressed in the inflammatory joints of mice with collagen-induced arthritis, and runt-related transcription factor 1 (RUNX1) can affect the susceptibility to RA by regulating the expression of SLC22A4 [
44]. McCann et al. observed that inappropriate triggering of the inflammatory response can be effectively reduced by reducing the abnormal transport function of the SLC22A4 503F variant [
45]. In addition, Yamase et al. proved that the genetic polymorphisms of SLC22A4 rs273909 were significantly associated with ischaemic stroke in the Japanese population [
46].
Vimentin (
VIM, gene ID: 100,507,347, HGNC: 44,879, OMIM: 193,060) acts as a cytoskeletal intermediate silk protein, plays a crucial role in neuritogenesis and cholesterol transport, and functions as an organizer of several key proteins involved in subsequent biological processes, such as signal transduction, adhesion, migration, apoptosis, and differentiation [
47]. Kim et al. found that oxidized low density lipoprotein (ox-LDL) can induce the synthesis and secretion of VIM in macrophages, while extracellular VIM can induce macrophages to release inflammatory cytokines, such as tumour necrosis factor-α (TNF-α) and interleukin 6 (IL-6), which subsequently lead to atherosclerotic inflammation [
48]. He et al. proved that silencing the expression of miR-144 can significantly promote the expression of VIM and the formation of atherosclerotic plaques [
49]. Yao et al. suggested that inhibiting the expression and rearrangement of VIM can effectively reduce the migration of vascular smooth muscle cells induced by TNF-α, thereby alleviating the progression of atherosclerotic lesions [
50]. Gong et al. found that the serum VIM levels were higher in patients with coronary artery disease (CAD), and the VIM levels were positively correlated with the severity of CAD. In addition, these authors found that VIM can accelerate the occurrence and development of atherosclerotic lesions by inducing macrophages to secrete proinflammatory cytokines and adhesion molecules [
51]. Furthermore, Xiao et al. found that VIM can increase the instability of plaques, and an elevated level of VIM can significantly increase the risk of ischaemic stroke in patients with carotid plaques [
52].
Adverse innate immune responses are associated with several disease processes. Fernandez et al. provided the first systematic description of the morphology of immune cells during atherosclerosis, provided insight into which immune cells reside in plaques and described their different activation states, which opened the door to the study of atherosclerosis caused by the immune response [
53]. Monocyte subsets play a crucial role in the atherogenesis and inflammatory cascade of cardiovascular disease. Upregulated counts and monocyte activity are significantly related to clinical indices of chronic kidney disease (CKD) and atherosclerosis [
54]. T lymphocytes, which act as the most important type of immune cells, can be divided into CD4 and CD8 cell subsets according to their surface markers and functions. CD8 T cells play a dual role in atherosclerosis. Previous studies have suggested that CD8 T cells can secrete various inflammatory cytokines, which can aggravate the inflammatory response and increase the instability of atherosclerotic plaques [
55]. However, cytotoxic activity targeting antigen presenting cells and regulatory CD8 T cells could effectively inhibit the progression of atherosclerosis by alleviating the immune reaction [
55]. Other immune cell types, including neutrophils [
56] and master cells [
57], also play a key role in the occurrence and development of cardiovascular disease. Furthermore, Li et al. found that the proportion of M1 macrophages, gamma delta (γδ) T cells and neutrophils was significantly higher and that the proportion of eosinophils and resting dendritic cells was significantly lower in ischaemic stroke patients compared to those in healthy subjects. However, the immune infiltration pattern of ischaemic stroke has not been fully elucidated. Clarifying immune infiltration in ischaemic stroke and identifying the key genes related to immune cells could provide a new perspective for the prevention and treatment of ischaemic stroke.
To further evaluate the proportion and type of immune cell infiltration in ischaemic stroke, the CIBERSORT package in R was utilized to carry out a comprehensive assessment of 22 types of immune cell infiltration in ischaemic stroke patients. We noticed that there was a decreased infiltration of naive CD4 T cells, CD8 T cells, resting mast cells and eosinophils and an increased infiltration of neutrophils, M0 macrophages and activated memory CD4 T cells in ischaemic stroke patients. As previously mentioned, the inflammatory characteristics of circulating neutrophils were increased in the acute stage of ischaemic stroke, and activated neutrophils may promote the progression of ischaemic stroke by promoting systemic inflammation and destroying the blood‒brain barrier [
58]. Compared with the normal samples, the proportion of neutrophils in the ischaemic stroke samples was generally higher; neutrophils are involved in ischaemic injury after stroke and may be a promising target for ischaemic stroke therapies [
59]. In addition, CD8 T cells play a dual role in atherosclerosis, and our study showed that the proportion of neutrophils was higher while the proportion of CD8 T cells was lower in the ischaemic stroke patients compared with those in the control subjects. This finding implies that neutrophils can accelerate but CD8 T cells can inhibit the occurrence and progression of ischaemic stroke. However, whether the number of CD8 T cells and neutrophils in peripheral blood samples could reflect their infiltration pattern in the vascular wall remains unclear. In addition, the current research revealed the interaction of 22 types of infiltrated immune cells in ischaemic stroke. Neutrophils were negatively associated with CD8 T cells and eosinophils and positively associated with M0 macrophages. Moreover, these key genes including
ADM,
ANXA3,
SLC22A4 and
VIM were positively correlated with M0 macrophages and neutrophils and negatively correlated with CD8 T cells. However, a large number of studies have shown that immune checkpoint inhibitors targeting programmed cell death 1 (PD1), programmed cell death ligand 1 (PDL1) and cytotoxic T-lymphocyte associated protein 4 (CTLA4) can effectively improve the prognosis of many cancer patients, but it may lead to some vascular and cardiac toxicity such as atherosclerosis, ischaemic stroke or myocardial infarction and other adverse reactions [
7,
60,
61]. Therefore, further studies are needed to explore whether therapies targeting these genes such as
ADM,
ANXA3,
SLC22A4 and
VIM will bring some similar risks to patients with ischaemic stroke.
On the other hand, the gene enrichment analysis indicated that these key genes were mainly involved in inflammatory or immune-related signalling pathways, such as the NF-kappa B (NF-κB) signalling pathway, TNF signalling pathway, Toll-like receptor signalling pathway, NOD-like receptor signalling pathway and IL-17 signalling pathway. Previous studies have shown that the transcription factor NF-κB is a main regulator of genes involved in the inflammatory response [
60,
62], and NF-κB has been shown to play an important role in ADM-induced inflammation [
62]. The overexpression of NF-κB can participate in the rheumatoid arthritis-related inflammatory response by activating the SLC22A4 promoter [
63]. The inhibition of NF-κB can reduce the expression of VIM and affect the epithelial mesenchymal transformation and nerve infiltration in pancreatic cancer [
64]. In addition, Liu et al. suggested that the NF-κB signalling pathway plays a key role in the biological processes of cell proliferation, migration and apoptosis mediated by ANXA3 [
34]. These findings are consistent with our bioinformatics analysis and suggest that the NF-κB signalling pathway plays an important role in the biological processes mediated by these key genes, including
ADM,
ANXA3,
SLC22A4 and
VIM. However, the regulatory relationship among these key genes, the NF-κB signalling pathway and the mechanism of action in ischaemic stroke still need further experimental verification.
This research had several limitations. First, the RT‒qPCR analysis found that there was no significant difference in the expression levels of CARD6, CPQ, UBE2S and ZFP36 between our ischaemic stroke patients and normal subjects. The validation samples included in the current research were recruited from only a single centre with small sample sizes. Whether the expression levels of the above genes differ among individuals in different regions or races is unclear. Therefore, the results of this study need to be further tested in multicentre studies with larger samples. Second, whether the number of CD8 T cells and neutrophils in peripheral blood samples could reflect their infiltration in the vascular wall remains unclear. Third, more in vivo and in vitro studies are needed to clarify the underlying mechanism of these correlations among ADM, ANXA3, SLC22A4 and VIM and infiltrated immune cells in ischaemic stroke.