In this study, 10 hub genes related to the occurrence of IS were identified through WGCNA and LASSO regression analysis based on the merged datasets of GSE16561 and GSE58294. Values of the AUC for the 10 hub genes ranged from 0.841 to 0.974. ID3 had the highest interaction rate with other TF-genes in the TF-gene interaction network. ABCA1, transcription factors SP1 and hsa-miR-340 had a high interaction rate in the miRNA-TF regulatory network. HPA database analysis showed that ABCA1, SLC25A42, SLC22A4, ID3 and FAM102A had higher gene expression levels in the cerebral cortex than those in blood. We also predicted seven therapeutic drugs targeted with ABCA1 and SLC22A4, and four small molecule compounds targeted with ID3 and SLC22A4.
KEGG analysis showed that modular genes were mainly enriched in 13 pathways. Among them, the NF-κB signaling pathway, apoptosis, lipid and atherosclerosis pathways may be related to the pathogenesis and development of IS. As for the NF-κB signaling pathway, evidence showed that ischemic preconditioning could activate the NF-κB signaling pathway, and enhance the expression of NF-κB to protein in neurons after lethal ischemia-reperfusion (Lee et al. 2016). Study had shown that the NF-κB gene was the downstream target of the Notch signaling pathway (Schwarzer et al. 2012). The Notch signal partly participated in the activation and regulation of microglia after hypoxia through the interaction of the TLR4/MyD88/TRAF6/NF-κB signaling pathway (Yao et al. 2013). Fann et al reported that the NF-κB signaling pathway played a key role in the expression and activation of NLRP1 and NLRP3 inflammasomes in primary cortical neurons and brain tissue under ischemic conditions (Fann et al. 2018). For apoptosis pathway, current study manifested that the internal and external pathways for cerebral ischemia to trigger apoptosis were derived from the release of cytochrome C from mitochondria, the activation of caspase-3 and cell surface death receptors, which lead to the stimulation of caspase-8 (Broughton et al. 2009). Overexpression of the apoptotic protein bcl-2 can reduce cerebral infarction tissue in the middle cerebral artery occlusion model (Martinou et al. 1994). By preventing injury-induced down-regulation of the apoptotic protein bcl-2 expression, estradiol had a profound protective effect on cerebral ischemia caused by permanent middle cerebral artery occlusion (Dubal et al. 1999). The overexpression of a variety of pro-apoptotic and anti-apoptotic proteins in the penumbra after ischemic stroke may be used as potential therapeutic targets for stroke (Uzdensky 2019). For the lipid and atherosclerosis pathway, lipids accumulate on the arterial wall in the form of fatty deposits and cholesterol plaques. It was the main component of early atherosclerotic lesions (Kauw et al. 2018). Atherosclerosis was a chronic inflammatory disease characterized by the formation of lipid plaques on the blood vessel wall, which can increase the risk and recurrence of stroke (Chow et al. 2020).
Ten hub genes were identified in this study, which were ARG1, TPM2, LY96, ABCA1, SLC25A42, SLC22A4, ID3, FAM102A, CD79B and CD163. Previous experimental studies have revealed that 6 hub genes were related to the occurrence and development of stroke. Study have shown that STAT6/ARG1 regulated the efferocytosis of the microglia/macrophage phenotype, accelerated the resolution of inflammation, and improved the prognosis of stroke (Cai et al. 2019). Barr et al reported that LY96 and ARG1 had been identified as stroke expression profile genes, which may be biomarkers of stroke (Barr et al. 2010). As a major reverse cholesterol transporter, ABCA1 plays a key role in the formation of high-density lipoprotein (HDL) cholesterol in the brain. ABCA1/ApoE/HDL pathway was at least partially involved in the process of nerve recovery after stroke induced by gw3965 (Cui et al. 2017). In addition, ABCA1 rs2230806 GG was significantly associated with an increased risk of IS (Au et al. 2017). SLC22A4's rs273909 was significantly associated with IS (P=0.0123) (Yamase et al. 2015), which increased the susceptibility of IS. ID3, an Inhibitor Of DNA Binding 3, was involved in stress response, neural plasticity, neural circuits, and other processes (Avecilla et al. 2017). CD163 is a scavenger receptor expressed in the innate immune cell population. CD163 served as a potential biomarker for monocyte activation in patients with ischemic stroke (Greco et al. 2021).
Adamski et al showed that the transcriptome AUC containing 7 genes PLBD1, PYGL, BST1, DUSP1, FOS, VCAN and FCGR1A was 0.854 for identification of stroke by the high throughput next-generation qPCR verification (Adamski et al. 2014). In our research, we finally got 10 hub genes, 9 genes had AUC values higher than 0.854, the AUC of 7 genes had higher than 0.9. FAM102A had the highest diagnostic value with the AUC of 0.974, sensitivity of 0.919 and specificity of 0.936, respectively.
NetworkAnalyst database analysis showed that transcription factor SP1 encodes a zinc finger transcription factor, which was involved in many cellular processes such as cell differentiation, apoptosis, and immunity. It has been reported that Curcumin up-regulated Peroxiredoxin 6 through SP1 to induce ischemic oxidative damage after stroke in rats (Jia et al. 2017). H2O can be used as a signal molecule by neurons to quickly induce Sp1 translation through an ires-dependent translation pathway. Through feedforward transcription activation, with the late rise of Sp1, neurons can be protected from ischemic damage (Yeh et al. 2011). Yoo et al indicated that miR-340-5p regulated immune response by targeting ARG1 protein after acute ischemic stroke (Yoo et al. 2019).
HPA database analysis showed that the 10 hub genes in this study were expressed in the brain and blood. Studies have found a correlation between brain gene expression profiles and blood gene expression profiles, and blood gene expression profiles had the potential to evaluate various disease states in terms of diagnosis, mechanism and treatment (Tang et al. 2001, 2002). A brain tissue gene expression prediction model based on whole blood gene expression data proved that only whole blood expression profile data could accurately predict the gene expression of unsampled brain tissue Wenjian and Wei 2019). Iturria-medina Y et al showed that blood gene expression strongly reflected the severity of neuropathology, clinical deterioration and disease progression in vivo (Iturria-Medina et al. 2020). Therefore, we hypothesize that gene expression in the blood may reflect brain conditions in some extent. Among them, ABCA1, SLC25A42, SLC22A4, ID3 and FAM102A had higher gene expression levels in the cerebral cortex than in blood. Compared with the other five genes, we speculated that these five genes with high expression in brain tissue had more important value in the pathological mechanism of IS occurrence and development, and may be used as a target for drug therapy.
Next, we used DGIdb and cMap databases to predict 5 gene-targeted therapeutic drugs. DGIdb database analysis showed that there were 6 therapeutic drugs targeted with ABCA1, namely simvastatin, probucol, fenofibrate, cyclosporine, pravastatin and atorvastatin. Imatinib was a therapeutic drug targeted with SLC22A4. CMap database analysis showed that there were 3 small molecule compounds targeted with ID3, namely 2', 5'-dideoxyadenosine, DR-2313 and otenzepad. And dubinidine was a small molecule compound targeted with SLC22A4. Among the drugs targeted ABCA1, simvastatin, pravastatin and atorvastatin are all statins. According to previous studies, the use of statins for the treatment of stroke has long been widely accepted and used by related researchers, which has been promoted in the continuously updated guidelines. According to the 2013 AHA/ASA guidelines, statins were recommended to be applied in patients presumed to be atherosclerotic ischemic stroke or TIA to reduce the risk of stroke and cardiovascular events (Kernan et al. 2014). Statins can significantly reduce the infarct volume and neurological deficits in tMCAO model mice. At the same time, compared with atorvastatin, simvastatin and pravastatin were more significant in reducing infarct volume (Christophe et al. 2020). Statins seemed to reduce the risk of recurrence of ischemic stroke and other cardiovascular events in patients (Tramacere et al. 2019).
Probucol inhibited LPS-induced microglia activation and improved cerebral ischemic injury in mice through anti-neuro-inflammatory effects, and it may be a potential therapeutic drug for the treatment of ischemic patients (Jung et al. 2016). Fenofibrate could significantly reduce the volume of cerebral infarction, the activation of microglia and the infiltration of neutrophils in the ischemic area to play a protective effect on the brain, which may be related to the acute effect and preconditioning mechanism (Ouk et al. 2009). Fenofibrate played a preventive and protective effect on cerebral ischemia by reducing the infarct volume, and the related mechanism may be related to the genome regulation in the apoptosis pathway (Altintas et al. 2017). Cyclosporine A had potential neuroprotective properties in ischemic stroke, which interfered with the typical NF-κB signaling pathway, and the Akt pathway may also participate in the role of Cyclosporine A (Deng et al. 2020; Nighoghossian et al. 2016). Deng et al showed that DR2313 exerted neuroprotective effects through its powerful PARP inhibitory effect, and could easily penetrate the blood-brain barrier after peripheral to the medicine. It may be more useful than free radical scavengers in the treatment of acute stroke, which may be a promising drug for stroke treatment (Nakajima et al. 2005).