Identifying the regulatory network of microRNAs and mRNAs to clarify molecular mechanisms in stroke by bioinformatics analysis

To identify the regulatory network of microRNAs (miRNAs) and mRNAs to clarify molecular mechanisms in stroke. Four miRNA datasets and two mRNA datasets of stroke were downloaded from the GEO database. R-Studio was utilized to analyze differentially expressed miRNAs (DEmiRNAs) and mRNAs (DEmRNAs) in the blood of stroke and control patients. FunRich software was utilized to conduct GO and biological pathway analysis on DEmiRNAs, and to search for transcription factors (TFs) of DEmiRNAs. Subsequently, we used miRDB, miRTarBase, and TargetScan to identify DEmiRNAs target genes and intersected with DEmRNAs to �nd common target genes. The miRNA-mRNA regulatory network of common target genes was constructed by using the Cytoscape. The biological and functional roles of target genes in the regulatory network were predicted using GO and KEGG pathway analyses. 464 DEmiRNAs and 329 DEmRNAs were screened, respectively. The top ten TFs (SP1, SP4, EGR1, TCF3, NKX6-1, ZFP161, RREB1, MEF2A, NFIC, POU2F1) were visualized. 16747 target genes of DEmiRNAs were predicted. Target genes were intersected with DEmRNAs, 107 common target genes and 162 DEmiRNAs regulating these common genes were obtained, and then a regulatory network was constructed. Target genes of the regulatory network were primarily enriched in VEGF signaling pathway, lipid and atherosclerosis, T cell receptor signaling pathway. This study found that VEGF signaling pathway, lipid and atherosclerosis, T cell receptor signaling pathway are implicated in the biological process of stroke by constructing the regulatory network of miRNAs-mRNAs, which may have guide signi�cance for the pathogenesis and treatment of stroke.


Introduction
Stroke is one of the main causes of disability and death in the globe and exhibits high morbidity and recurrence rates.Stroke was rated second among the top ten main causes of disability-adjusted life years (DALYs) in 2019 for both people aged 50-74 and those aged 75 [1,2] .For every 10-year increase in age, the annual risk of stroke doubles [3] .Stroke exhibits high mortality and physical disability, and usually requires long-term rehabilitation after the acute phase, bringing a heavy burden on patients, families, and society as a whole [4] .It is critical to investigate stroke mechanisms, nd novel therapeutic interventions for stroke, provide more effective treatment for patients, and reduce the burden on patients.
MiRNAs are a ubiquitous class of small non-coding single-stranded RNAs composed of 21-23   nucleotides that play important roles in health and disease [5] .MiRNAs modulate gene expression at the post-transcriptional level by modulating target messenger RNAs (mRNAs), leading to changes in levels of target proteins.Single or multiple miRNAs can bind to the 3' non-translated region of a target mRNA, causing its degradation or inhibiting translation/transcription [6] .Thousands of downstream target genes may be regulated by a single miRNA, which could have an impact on whole gene network and protein synthesis [7] .There is evidence that miRNAs play crucial roles in neuropathological processes resulting from stroke.Therefore, miRNA levels after stroke can provide insight into the underlying pathogenesis of stroke [8] .
In this study, DEmiRNAs and DEmRNAs of stroke were screened and analyzed by analyzing multiple miRNA datasets and multiple mRNA datasets of stroke for the rst time, so as to construct a regulatory network of miRNAs-mRNAs, increase the understanding of the pathological mechanisms of stroke, and bring new perspective on clinical diagnosis and treatment of stroke.

Data preparation and processing
Data preparation and processing are shown in Fig. 1.The GEO database (https://www.ncbi.nlm.nih.gov/geo/) was used to download four stroke miRNA datasets (GSE60319, GSE86291, GSE95204, GSE117064) and two stroke mRNA datasets (GSE16561, GSE58294).The GSE60319 dataset included 117 patients with ischemic stroke and 82 healthy controls.The GSE86291 dataset included 7 patients with ischemic stroke and 4 controls.Next, the GSE95204 dataset comprised 6 ischemic stroke patients and 3 controls.The GSE117064 dataset comprised 173 patients with cerebrovascular disorders (CVD) and 1612 non-CVD controls.Furthermore, the GSE16561 dataset included 24 controls and 39 ischemic stroke patients.Finally, the GSE58294 dataset included 69 patients with stroke and 23 controls.These datasets are detailed in Table 1.

Differential Expression Analysis
The limma program of R-Studio was utilized to screen for differentially expressed miRNAs (DEmiRNAs) and mRNAs (DEmRNAs) across stroke and control groups, respectively.Sets the screening threshold of DEmiRNAs to adjust P value < 0.05 and |logFC| > 1.The screening threshold of DEmRNAs are adjust P value < 0.05 and |logFC| > 0.5.Pheatmap and ggplot2 packages were used to draw heatmaps and volcanic maps for DEmiRNAs and DEmRNAs respectively.

Gene Ontology (GO) and biological pathway analysis and transcription factors prediction of DEmiRNAs
FunRich software was used to conduct GO and biological pathway analysis on DEmiRNAs.GO describes the biological functions of target genes in terms of biological processes (BP), cellular components (CC), and molecular function (MF).Based on FunRich software, the transcription factors (TFs) were searched.
The condition of statistical signi cance was adjusted P value < 0.05.

Functional enrichment of target genes in regulatory network
After target genes were extracted from regulatory network, the org.Hs.eg.db and clusterPro ler package in R-Studio were utilized to carry on GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses for target genes of ceRNA network.GO describes the biological functions of target genes in terms of biological processes (BP), cellular components (CC), and molecular function (MF), while KEGG can explain the signaling pathways in which the differentially expressed target genes have been involved.The signi cant enrichment threshold was adjusted P < 0.05.

Differentially expressed miRNAs and mRNAs
The ow chart for this study is illustrated in Fig. 1.The results after batch effect elimination are presented using the principal component analysis (PCA) graph (Fig. 2).In present study, results showed that 464 DEmiRNAs and 329 DEmRNAs were obtained between the stroke and the control group, including 265 upregulated and 199 downregulated DEmiRNAs, and 195 upregulated and 134 downregulated DEmRNAs.
DEmiRNAs and DEmRNAs are displayed as volcano map and heatmap (Fig. 3), with red for upregulated genes and green for downregulated genes.

GO and biological pathway analysis and transcription factors prediction of DEmiRNAs
In order to better understand DEmiRNAs, we use FunRich software has carried on GO and biological pathway analyses.GO enrichment results were nucleotide, nucleoside and nucleic acid metabolism, apoptosis, regulation of nucleobase, regulation of cell growth, cell communication, signal transduction, transport, regulation of gene expression, fatty acid metabolism, regulation of cell cycle, epigenetic, regulation of translation (BP), cytoplasm, nucleus, golgi aparatus, lysosome, early endosome, endosome, actin cytoskeleton, axon, membrane, cytoplasmic vesicle (CC), and receptor signaling complex scaffold activity, ubiquitin-speci c protease activity, GTPase activity, transcription factor activity, transcription regulator activity, protein serine/threonine kinase activity, receptor binding, guanyl-nucleotide exchange factor activity, protein serine/threonine phosphatase activity, cytoskeletal protein binding (MF).Biological pathway enrichment results were proteoglycan syndecan-mediated signaling events, beta1 integrin cell surface interactions, glypican pathway, erbB receptor signaling network, TRAIL signaling pathway, integrin family cell surface interactions, VEGF and VEGFR signaling network, sphingosine 1-phosphate (S1P) pathway, PDGF receptor signaling network, plasma membrane estrogen receptor signaling (Fig. 4).TFs can bind to mRNA and perform post-transcriptional regulatory functions, both of which are crucial components of regulatory networks in organisms.The FunRich software was utilized to predict the upstream TFs of DEmiRNAs.The top ten TFs (SP1, SP4, EGR1, TCF3, NKX6-1, ZFP161, RREB1, MEF2A, NFIC, POU2F1) were visualized (Fig. 5).

Target genes prediction for DEmiRNAs and DEmiRNAs-DEmRNAs network construction
Three databases (TargetScan, miRTarBase and miRDB) were utilized to predict the target genes of DEmiRNAs, 16747 target genes were identi ed, and then the target genes were intersected with DEmRNAs, and 401 DEmiRNAs-DEmRNAs interactions pairs were obtained (Supplementary Table 1).

GO and KEGG enrichment Analysis
The functions of the DEmRNAs in DEmiRNAs-DEmRNAs network were investigated by GO and KEGG functional enrichment analyses.With regard to BP, the DEmRNAs were primarily involved in neutrophil mediated immunity, neutrophil activation, regulation of MAP kinase activity, neutrophil degranulation, positive regulation of membrane permeability, neutrophil activation involved in immune response, response to acid chemical, positive regulation of mitochondrial membrane permeability, positive regulation of T cell differentiation in thymus, secretory granule localization.The main CC involving the DEmRNAs in secretory granule membrane, membrane raft, lytic vacuole membrane, lysosomal membrane, azurophil granule membrane, vacuolar membrane, membrane region, membrane microdomain, primary lysosome, azurophil granule.In terms of MF, DEmRNAs were primarily enriched in phospholipid binding, phosphatidylserine binding.In terms of KEGG, the DEmRNAs were mostly enriched in kaposi sarcoma − associated herpesvirus infection, VEGF signaling pathway, T cell receptor signaling pathway, lipid and atherosclerosis, PD − L1 expression and PD − 1 checkpoint pathway in cancer.The GO and KEGG results are displayed as bar graphs and bubble diagrams respectively (Fig. 7).

Discussion
Acute ischemic stroke (AIS) has high morbidity, mortality, and disability rate, which brings heavy burden to economy and society, and is a serious public health problem in China [9,10] .Currently, intravenous thrombolysis is considered to be a satisfactory treatment for AIS.However, the narrow time window of thrombolytic therapy, high cost, and lack of speci city in ultra-early imaging characteristics of AIS, which often makes AIS patients unable to receive timely thrombolytic therapy [11] .Identi cation of blood biomarkers is essential for rapid diagnosis.Therefore, we identi ed potential stroke-related genes and constructed a DEmiRNAs-DEmRNAs regulatory network to further understand the pathogenesis of stroke and give novel insights for the prevention, diagnosis and therapy of stroke.
In present study, the differentially expressed genes of stroke were screened by analyzing multiple miRNA datasets and multiple mRNA datasets of stroke for the rst time, 464 DEmiRNAs and 329 DEmRNAs were obtained in the stroke group and the control group, including 265 upregulated and 199 downregulated DEmiRNAs, and 195 upregulated and 134 downregulated DEmRNAs.The top ten TFs (SP1, SP4, EGR1, TCF3, NKX6-1, ZFP161, RREB1, MEF2A, NFIC, POU2F1) were visualized.We obtained 401 DEmiRNAs-DEmRNAs interactions pairs through databases prediction, constructed a regulatory network of stroke containing 162 DEmiRNAs and 107 DEmRNAs, and analyzed the functional enrichment of DEmRNAs in the regulatory network.It was found that DEmRNAs of regulatory network are mainly concentrated in VEGF signaling pathway, lipid and atherosclerosis, T cell receptor signaling pathway.
Studies have shown that miRNAs expression can be regulated by TFs [12] .Therefore, we identi ed upstream TFs that may modulate DEmiRNAs.SP1, SP4 and EGR1 were the most signi cant transcription factors.SP1, along with SP2, SP3, and SP4, is the speci city protein (SP) family member and functions as a key transcription factor and multifunctional oxidative stress response protein [13] .After transient middle cerebral artery occlusion, 38 transcription factors were discovered to be differentially expressed in mice, among which SP1, SPi1, and Stat3 being the most substantially expressed [14] .It is speculated that SP1 is one of the principal genes involved in ischemic stroke [15] .SP1 protects neurons, endothelial cells and glial cells by promoting the transcription of a variety of antioxidant proteins and a variety of antioxidant enzymes, affecting ion transporters on the cytoplasmic side [16] .Early growth response 1 (EGR1) is involved in ischemia-reperfusion injury.EGR1 knockdown effectively reduces infarct size and apoptosis, and improves neurological function [17] .MEF2A, a nuclear element subtype found in myogenic cells, is upregulated in the hippocampus and cerebellum [18] .MEF2A transcriptional activity inhibition can promote neuronal apoptosis.Studies have found that knockdown of MEF2A can aggravate neurological damage and impair cognitive function in rats [19] .These results support the role of selected DEmiRNAs in the pathogenesis of stroke.
Based on functional annotation, VEGF signaling pathway, lipid and atherosclerosis and T cell receptor signaling pathway are signi cantly enriched pathway in stroke.Angiogenesis can increase collateral circulation, alleviate ischemic brain damage, and play a key role in neurological function recovery after cerebral infarction [20] .The vascular structures regeneration provides critical "support" for neurogenesis and synapse formation, promotes neural networks recovery, and plays a vital role in enhancing stroke patients' quality of life [21,22] .Angiogenesis is regulated by vascular endothelial growth factor (VEGF).VEGF is a pivotal cytokine associated with endothelial cell differentiation, proliferation, and migration.VEGF activates downstream signaling pathways and conveys angiogenic signals when it binds to its receptor [23] .Atherosclerosis (AS) is the pathological basis of several diseases, such as ischemic stroke, coronary heart disease and peripheral vascular diseases [24] .Cerebral AS is distinguished by increasing lipid accumulation, brous hyperplasia, and in ammatory cell in ltration, which may result in poor clinical consequences such as transient ischemic attack, stroke, and even death [25,26] .The mechanism of stroke induced by dyslipidemia includes inducing AS, endothelial dysfunction, thrombosis and bloodbrain barrier damage, reducing cerebral blood ow (CBF) and promoting neuronal apoptosis [27][28][29] .Hyperlipidemia can induce hypertension and cardiovascular disease, and can also interfere with CBF by reducing the synthesis and release of VEGF, thereby increasing the risk of stroke [30,31] .
In ammation and immunological response play key roles in the pathogenesis of ischemic stroke.After stroke, damaged nerve cells induce activation of glial cells, in ltration of peripheral immune cells, and release of in ammatory mediators [32] .T cell receptors are speci c receptors on the surface of T cells, which link T cells to antigen-presenting cells and play a crucial role in T cell immune response.Cerebral infarction rapidly activates the peripheral immune system, facilitating T or B lymphocytes, neutrophils and monocytes into the damaged region [33] .The study showed that mice lacking T and B cells had smaller infarcts than normal mice [34] .Selective inhibition of T cell receptors reduces infarct size and promotes long-term functional recovery [35] .Therefore, VEGF signaling pathway, lipid and atherosclerosis and T cell receptor signaling pathway are crucial in ischemic stroke.However, further experiments are needed to verify our study.

Conclusion
This study found that VEGF signaling pathway, lipid and atherosclerosis, T cell receptor signaling pathway are implicated in the biological process of stroke by constructing the regulatory network of miRNAs-mRNAs, which may have guide signi cance for the pathogenesis and treatment of stroke.The process of data preparation and processing.TFs: transcription factors; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes.

Figure 2 Principal
Figure 2

Figure 4 GO
Figure 4