Identification of DEGs
Gene expression profile of GSE34608 and GSE83456 were downloaded from GEO database. The microarray data GSE34608 contains 18 control and 8 PTB patients. The GSE83456 data contains 61 control samples and 45 PTB samples. PCA plots of both datasets indicated the distinction expression of control and PTB samples (Figure 1. A and B). 2214 and 1025 DEGs were identified from GSE34608 and GSE83456 datasets, respectively (Figure 2A). Venn diagram demonstrated that, among the 180 shared DEGs, 51 genes were down-regulated and 129 genes were upregulated in both datasets (Figure 2A).
GO enrichment and KEGG pathway analyses
Candidate DEGs functional Gene Ontology (GO) and pathway enrichment analyses were performed with Metascape. The results showed that DEGs were significantly enriched in defense response to other organism, response to bacterium, myeloid leukocyte activation, cytokine production, positive regulation of defense responses, cytokine-mediated signaling pathway, interferon signaling, etc. (Figure 2B).
The subset of representative terms of gene function analysis were converted into a network layout in Metascape, as shown in Figure 2C. Based on gene function analysis, all the significant terms were hierarchically clustered into a tree based on Kappa-statistical similarities. Each term is represented by a circle node, where its size is proportional to the number of input genes fall into that term. The color represents its cluster identity (Figure 2C). Terms with a kappa score > 0.3 are linked by an edge. The statistically significant range of the node is marked by color range (Figure 2C).
PPI network construction and module analysis
The PPI network of 180 DEGs was constructed using the STRING online database, and further analyzed using app MCODE in Cytoscape software. Totally, seven modules were identified shown in Table 1. Module 1 from the PPI network complex containing 35 genes, indicating the core functional gene panel. GO analysis of 35 genes showed that their functions are related to defense response and cytokine related pathway (Figure 3A). PPI network of module 1 was redrawn by STRING (Figure 3B). The expression level of 35 genes in dataset GSE34608 were shown in Figure 3C. Genes CD27, CCR7, CD19, and CXCR3 were significantly down-regulated in PTB samples, where other genes were upregulated. This result was consistent with the gene expression levels in GSE83456 (Figure 3D). Furthermore, function analysis from STRING database were shown in Table 2. GO function was significantly related to defense response and immune system response (biological process function), chemokine binding and chemokine receptor activity (molecular function), and external side of plasma membrane and cell surface (cellular component). KEGG and Reactome pathways indicated that these 35 genes were involved in cytokine-cytokine receptor interaction, Toll-like receptor signaling pathway, immune system, and cytokine signaling in immune system. TIR domain, leucine rich repeat, and chemokine receptor family were the three important features revealed by PFAM and INTERPRO protein domains analysis (Table 2).
Hub genes analysis
A total of 14 genes were selected as hub genes based on criteria MCODE (scores > 10, degree > 20, neighborhood connectivity > 10) in Table 3. All the hub genes were belonging to the module 1. These hub genes were significantly associated with Toll-like receptors, interferon-induce proteins, and chemokine receptors (Table 3). Among them, two genes were upregulated, whereas as others were downregulated (Figure 3C). The expression levels were further validated in dataset GSE19439 (Figure 4). The expression levels were also significantly different between health and PTB patients, except gene CD19 and CXCR3 (Figure 4).
Gene expression level detection during PTB treatment
To figure out the expression level changes during PTB treatment process, dataset GSE31348 were used to evaluate the change level. GSE31348 contained the 27 PTB patients, including 135 samples from 5 time points. Heatmap showed that the expression level of genes related with the functions (Figure 5A). The expression level of CCR7, CD19, and CXCR3 were significantly increased, whereas the expression level of Interferon-induced proteins, Toll-like receptors were decreased during the treatment (Figure 5A). Among these 14 genes, the expression level of CXCR3 were significantly increased, and TLR2 and TLR5 were significantly decreased during the PTB treatment (Figure 5B). These three genes might have a potential to evaluate PTB as a gene panel.