Application of multiple omics and network projection analyses to drug repositioning for pathogenic mosquito-borne viruses

Abstract

Pathogenic mosquito-borne viruses are a serious public health issue in tropical and subtropical regions and are increasingly becoming a problem in other climate zones. Drug repositioning is a rapid, pharmaco-economic approach that can be used to identify compounds that target these neglected tropical diseases. We have applied a computational drug repositioning method to five mosquito-borne viral infections: dengue virus (DENV), zika virus (ZIKV), West Nile virus (WNV), Japanese encephalitis virus (JEV) and Chikungunya virus (CHIV). We identified signature molecules and pathways for each virus infection based on omics analyses, and determined 77 drug candidates and 146 proteins for those diseases by using a filtering method. Based on the omics analyses, we analyzed the relationship among drugs, target proteins and the five viruses by projecting the signature molecules onto a human protein-protein interaction network. We have classified the drug candidates according to the degree of target proteins in the protein-protein interaction network for the five infectious diseases.

Figure 1

Schematic showing the process of drug repositioning against the five viruses (DENV, ZIKV, WNV, CHIV and JEV). Signature genes, signature proteins and interactions between human and viral proteins were selected from GEO datasets, review of the proteome literature and review of the human–viral PPI literature, respectively. To find drug candidates, a connectivity map (CMap) was generated for signature genes based on GEO data. STITCH was used for selecting drugs that interacted with the identified signature proteins from the proteome and human proteins in human–viral PPIs. Using our filtering method, we identified drug candidates for the five viral infections based on multiple omics analyses. Those disease-related proteins that control the onset and progression of the five viral infectious diseases were projected onto the protein–protein interaction network of the HPRD to identify drugs that are effective against the five diseases.

Figure 2

Signature genes and proteins by multiple omics analyses. Venn diagrams show the overlap of signature genes and proteins by multiple omics analyses for the five viruses: (a) DENV, (b) ZIKV, (c) WNV, (d) CHIV and (e) JEV. The number of signature genes obtained from transcriptomic data with significant differences is shown in the blue circle for infections. The number of signature proteins obtained from proteomics data is shown in the orange circle. The number of human proteins that interact with viral proteins in human–virus PPIs is shown in the green circle. (f) Venn diagram of common signature proteins, which represent the intersection between proteome and PPIs for the five viral infections. The total number of the unique proteins was 146 for the five viral infections.

Figure 3

Signature pathways by multiple omics analyses. Venn diagrams show the overlap of the signature pathways by GSEA analysis for (a) DENV, (b) ZIKV, (c) WNV, (d) CHIV and (e) JEV. The number of significant pathways of the transcriptome and proteome are shown in the blue and orange circles, respectively. (f) Venn diagram of the signature pathways, which are intersections derived from the transcriptomic and proteomic analyses for the five viral infections.

Figure 4

Drug candidates by multiple omics analyses. The numbers of drug candidates identified by the transcriptomic, proteomic and interactomic analyses are shown in blue, orange and green circles, respectively. The numbers of drug candidates that overlap between each analysis type are also shown: (a) DENV infection, (b) ZIKV infection, (c) WNV infection, (d) CHIV infection and (e) JEV infection. (f) The drug candidates of the intersection of the three omics analyses in five viral infections are shown. The numbers of drugs at the intersection of the three layers for DENV, ZIKV, WNV, CHIV and JEV are 58, 8, 31, 12 and 10, respectively. The total number of unique drugs was 97.

Figure 5

The network shows the relationship among drug candidates, signature proteins and the five viruses. The concentric circles show the network of diseases (purple circles), proteins (red and yellow circles) and drugs (green circles). The numbers of extended signature proteins (yellow) for DENV, ZIKV, WNV, CHIV and JEV are 262, 351, 352, 289 and 99, respectively. The degrees of the proteins calculated from the PPI network in the HPRD are divided into 0 to 10, 10 to 20, 20 to 30, 30 to 40, 40 to 50 and 50 or greater. The circularly distributed proteins with smaller radii indicate higher degrees. The numbers of common (red) and extended (yellow) proteins counting from the outer circle inwards are 32, 34, 40, 46, 48 and 57 for common proteins and 47, 54, 71, 94, 116 and 190 for extended proteins. The purple edge shows the connection between the disease and common proteins. The green edge shows the signature protein-drug candidate interactions.

Figure 6

Venn diagrams show the shared numbers of the common and extended signature proteins for the five viral infections. Common and extended signature protein numbers with all degrees are shown in (a) and those proteins with lower than and equal to 50 degrees are given in (b).

Figure 7

Two-dimensional matrix for drug-protein interactions between 77 filtered drug candidates and 15 shared proteins. Rows and columns represent 77 filtered drug candidates and 15 shared proteins, respectively. The proteins in the figure are arranged from the left in ascending order of the degree in the HPRD. Red squares indicate interactions between proteins and drug candidates found by STITCH.

Figure 8

(a) Minimum network with a low degree of the target proteins. The target proteins, which are shared proteins TNPO1, TCERG1, IKBKB, KPNB1 and XPO1 with 50 or less degrees, are shown as orange diamonds. Common (red ellipses) and extended (yellow diamonds) signature proteins in this network are connected to the five viral infections (purple round squares). Twenty-one added proteins are shown as light blue diamonds. (b) Minimum network with a high degree of the target proteins. The target proteins, which are shared proteins CASP3 and SNK2A1 with greater than 130 degrees, are shown as orange diamonds.