2.1 Key biomarker screening
To investigate potential biomarkers in the peripheral blood of stroke patients, we analyzed the expression profiles of 74 stroke patients and 44 healthy controls, and corrected for batch effects. Using LASSO, we identified 102 characteristic genes (Figure1C;Supplementary file 8).The SVM-RFE algorithm identified 12,564 genes(Figure1D;Supplementary file 7), and we selected the top 1,000 genes as candidate biomarkers for stroke. We also conducted differential expression analysis using the “DEseq2” and “limma” packages with the criteria |Log2 Fold change|>1 and |adj.P.Val.|<0.05. Through Venn diagram analysis(Figure1E;Supplementary file 5), we found that KCNK12 was a unique key biomarker for stroke.
2.2 Functional enrichment analysis of common DEGs
To further investigate the gene functions and significate KEGG pathways in common DEGs (Figure 2A;Supplementary file 4,6), the metascape database was used. ”Hemostasis”, ”Neutrophil degranulation”……were significantly enriched. GO enrichment analysis showed notable pathways including “hemopoiesis”, ”response to bacterium”, ”response to hormone”, ”inflammatory response”. These results revealed neutrophils played significantly role in both stroke and COVID-19.
2.3 KCNK12 activated neuroactive ligand-receptor interaction
To determined the mechanisms of KCNK12 in stroke, GSEA was performed based on RNA-seq of 74 stroke samples. Then, according to KCNK12, 74 samples were divided into high expression and low expression groups (Figure 2C;Supplementary file 9). Those results suggest KCNK12 is significantly related to neuroactive ligand-receptor interaction(Figure2B,D;Supplementary file 10).
2.4 KCNK12 was mainly associated with neutrophil infiltration
To explore the relationship between KCNK12 and immune cells, ssGSEA was applied to see differences in 28 immune cells between high expression and low expression groups. Next, another algorithm, Cibersort, was used for further analysis. Results of both ssGSEA and Cibersort indicated KCNK12 was extremely related to neutrophil infiltration(Figure3A,B; Supplementary file 11).
2.5 Neuroactive ligand-receptor interaction promoted neutrophil infiltration
Based on ssGSEA algorithm, we calculated scores of neuroactive ligand-receptor interaction and neutrophil. Then, build the correlation between Neuroactive ligand-receptor interaction and neutrophil infiltration, It shows a significant correlation(Figure3C).
2.6 KCNK12 expression in tissues and cells
HPA database[20], [23] was used to show KCNK12 expression in human tissues, HPA and Monaco transcriptome data showed KCNK12 was predominantly expressed in brain(Figure4A,B). Then, single cell RNA sequencing data of hypoxic cerebral cortex and normal cerebral cortex was used to explore KCNK12 expression in cells. The expression clustering data showed that KCNK12 was highly consistent with Alox5a, which is a marker gene of neutrophils(Figure4C,D; Supplementary figure 1).
2.7 KCNK12 regulatory network
KCNK12 is regulated by couples of mechanisms, STRING database was used for PPI analysis. Gene expression is mainly regulated by transcription factors(TFs) and miRNAs. TFs regulate transcription by binding to promoter regions, while miRNAs regulate post-transcriptional gene expression. Cistrome DB database was applied to explore transcription factors(Figure5A). A gene regulatory network with TFs and miRNAs was established using the NetworkAnalyst[24] online tool(Figure5B,C).
2.8 Drug screening
KCNK12 and 30 hub genes (Supplementary figure 2) in enrichment were used to identified from DSigDB library in Enrichr to predict drugs. Top ten drugs(ranked by P values) was shown in Figure 6.( Supplementary file 12)