Transcriptional Profiling and Deriving a Seven-Gene Signature That Discriminates Active and Latent Tuberculosis: An Integrative Bioinformatics Approach

Abstract

Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis (M.tb.). Our integrative analysis aims to identify the transcriptional profiling and gene expression signature that distinguish individuals with active TB (ATB) disease, and those with latent tuberculosis infection (LTBI). In the present study, we reanalyzed a microarray dataset (GSE37250) from GEO database and explored the data for differential gene expression analysis between those with ATB and LTBI derived from Malawi and South African cohorts. We used BRB array tool to distinguish DEGs (differentially expressed genes) between ATB and LTBI. Pathway enrichment analysis of DEGs was performed using DAVID bioinformatics tool. The protein-protein interaction (PPI) network of most upregulated genes was constructed using STRING analysis. We have identified 375 upregulated genes and 152 downregulated genes differentially expressed between ATB and LTBI samples commonly shared among Malawi and South African cohorts. The constructed PPI network was significantly enriched with 76 nodes connected to 151 edges. The enriched GO term/pathways were mainly related to expression of IFN stimulated genes, interleukin-1 production, and NOD-like receptor signaling pathway. Downregulated genes were significantly enriched in the Wnt signaling, B cell development, and B cell receptor signaling pathways. The short-listed DEGs were validated in a microarray data from an independent cohort (GSE19491). ROC curve analysis was done to assess the diagnostic accuracy of the gene signature in discrimination of active and latent tuberculosis. Thus, we have derived a seven-gene signature, which included five upregulated genes FCGR1B, ANKRD22, CARD17, IFITM3, TNFAIP6 and two downregulated genes FCGBP and KLF12, as a biomarker for discrimination of active and latent tuberculosis. The identified genes have a sensitivity of 80-100% and specificity of 80-95%. Area under the curve (AUC) value of the genes ranged from 0.84 to 1. This seven-gene signature has a high diagnostic accuracy in discrimination of active and latent tuberculosis.

Keywords: active TB; bioinformatics; biomarkers; differentially expressed genes; latent TB infection; tuberculosis.

Figure 1

(A) Volcano plot illustrating the identification of differentially expressed genes between ATB and LTBI in the Malawi and South African cohort (differentially expressed genes were distinguished with parameter −log10 p value in y axis and log2 fold change in x axis). Significant results were determined based on cut off range, p value < 0.01 and >1.5-fold change (B) Venn diagram demonstrating the intersection of differentially expressed overlapped or common genes, in Malawi and South African cohort derived from GSE37250 (Figure S1).

Figure 2

PPI network displaying the interaction of proteins coded by the upregulated genes derived from ATB vs. LTBI data. (A) Results of STRING analysis (p value < 1 × 10⁻¹⁶). PPI network with 76 nodes connected to 151 edges. (B) Closely connected subnetworks identified by MCODE analysis plugin of cytoscape. Two clusters enriched in top with cluster score above 3 are shown. (C) Functionally enriched edges identified from PPI network using clueGO/cluepedia of cytoscape software. Network connectivity among GO term and pathway determined based on the interaction of functional cluster, edges (kappa score > 0.4), and enriched terms/pathway with p value < 0.05. Functional groups are denoted in different color codes; the most enriched functional term is indicated in bold color. (D) Cytohubba (MCC method) analysis explored the most important hub nodes; nodes in red color indicate a high MCC score, and yellow color node represents a low MCC score.

Figure 3

Validation of DEG in an independent cohort using bioinformatics analysis. *** p < 0.001.

Figure 4

Receiver operating characteristic curve (ROC) analysis. ROC curves for upregulated genes FCGR1B, ANKRD22, CARD17, IFITM3, and TNFAIP6 and downregulated genes FCGBP and KLF12.