Identification of DEGs
First, we used Limma package to screen a total of 1185, 371 DEGs from GSE55235 and GSE55457, respectively. Subsequently, a total of 200 overlapping DEGs were selected from the DEGs of the two datasets (Fig. 1A, Fig. 1B). Finally, by comparing the logFC, of the overlapping DEGs in the two datasets, we found that the regulation direction of all the overlapping DEGs in the two datasets were surprisingly consistent.
Enrichment analyses of overlapping DEGs and PPI network construction
The results of GO enrichment analysis showed that the main biological processes(BP) of overlapping DEGs were immune response−activating signal transduction, B cell activation, T cell activation, lymphocyte mediated immunity, positive regulation of lymphocyte activation (Fig. 1C). The KEGG pathway analysis results demonstrated that overlapping DEGs were particularly enriched in cytokine-cytokine receptor interaction, chemokine signaling pathway, Th17 cell differentiation, B cell receptor signaling pathway (Fig. 1D). With the help of STRING database and cytoscape software, the PPI network containing 100 nodes and 200 edges of overlapping DEGs was constructed (Fig. 2A). Furthermore, we obtained 9 hub nodes from the PPI network through the cytohubba plug-in (Fig. 2B).
Validation of hub nodes
In order to further verify the expression level of 9 hub nodes in synovium of other RA patients, we selected GSE77298 as the test dataset. The results showed that among the 9 genes, the expression level of CCL5, CCR5, CD27, CXCL9 and TLR8 in RA group were significantly higher than that in control group, but there was no significant difference in rest of the genes (Fig. 2C). We consider these 5 genes as the hub genes in RA.
Results of immune infiltration analysis
The CIBERSORT deconvolution method screened the synovial samples in GSE77298 with P<0.05, and finally obtained 21 credible samples (Fig. 3A). The difference of immune infiltrating cells in synovium between RA and healthy control was analyzed by violin graph. The results showed that the plasma cells and M0 macrophages were significantly increased in RA group. In addition, NK cells activated, Monocytes, Dendritic cells resting and Mast cells resting were significantly decreased in RA synovium (Fig. 3B). By analyzing the correlation between the infiltration scores of the above 6 cells and the expression level of each key gene, we found that CCL5, CCR5,CD27 and CXCL9 were positively correlated with plasma cells; CCL5, CD27 and CXCL9 were negatively correlated with NK cells activated; CCR5 and TLR8 were negatively correlated with Dendritic cells resting; CCL5, CCR5 and CXCL9 were positively correlated with Mast cells resting.