Saturday, June 10, 2023
BestWooCommerceThemeBuilttoBoostSales-728x90

Analysis of PPI networks of transcriptomic expression identifies hub genes associated with Newcastle disease virus persistent infection in bladder cancer – Scientific Reports


PPI network of DEGs from TCCSUPPi

To better understand the regulatory mechanisms employed by bladder cancer cells that enable the developmentĀ of NDV persistent infection, differentially expressed genes (DEGs) associated with persistent TCCSUPPi and EJ28Pi cell lines were used to construct protein–protein interactions (PPI) network(s) through STRING Interactome database22. All the DEGs associated with TCCSUPPi (63) and EJ28Pi (134) were analyzed separately for network interactions. As a result, 6 subnetworks that included a continent (subnetwork 1) and 5 islands (subnetwork 2–6) were identified in TCCSUPPi cells. Two subnetworks with highest scores were selected for further analysis. The subnetwork 1 contained 291 nodes, 309 edges and 12 seeds (Fig.Ā 1A) and subnetwork 2 had 34 nodes, 36 edges and 2 seeds (Fig.Ā 1B). TheĀ expressionĀ levels and degrees of connection between nodes were represented by colours and areas respectively. The hub nodes in the entire network were further analyzed and the top 14 hub nodes were selected and graphically presented (Fig.Ā 1C). Twelve (12) hub nodes out of 14 were identified to beĀ mostly from subnetwork 1 and two superfamily member of cadherin (CDH2 and CDH5) were clustered in subnetwork 2 (Fig.Ā 1B) with CDH2 upregulated and CDH5 downregulated.

Figure 1

First and second networks of TCCSUPPi cells. (A) First identified network of TCCSUPPi cells. TheĀ red and green colours represent the nodes expression, that are up- and down-regulated, respectively. The expression levels are represented by the shades of colour and the area of the nodes indicate the degrees in which the nodes areĀ connected to eachĀ other. Nodes with gene names are the top 4 nodes in the PPI network. (B) Second identified network in TCCSUPPi cells. Subnetwork 2 contains both up- and downregulated nodes that are affected in the pathways. Nodes in red and green colour are upregulated and downregulated in TCCSUPPi cells, respectively. (C) Hub nodes in the PPI network of TCCSUPPi cells. Top 14 hub nodes with their degree levels. Genes in blue colour are from subnetwork 1 and red from subnetwork 2. The PPI network figures A & B were generated using a multifunctional online software, Network Analyst (https://www.networkanalyst.ca)23,24.

Functional connections in the network

Connections between functions in the identified networks were further explored andĀ the related nodes were re-constructed (Fig.Ā 2). As illustrated, pathways of bladder cancer, malaria, mitophagy, p53 signaling, ECM-receptor interaction, TGF-beta signaling, phagosome, ribosome, focal adhesion and proteoglycans in cancer were significantly enriched (p < 0.05) inĀ the upregulated DEGs in the nodes connecting the PPI network (subnetwork 1) (Supplementary Table S1). Antigen processing and presentation, protein processing in endoplasmic reticulum, prion diseases, legionellosis, longevity regulating, complement and coagulation cascades, platelet activation and spliceosome pathways were significantly enriched (p < 0.05) in the downregulated DEGs with connected nodes in the PPI network (subnetwork 1) (Supplementary Table S2).

Figure 2
figure 2

Modules 0 and 1 in the TCCSUPPi PPI network. The modules in blue and red are module 0 and module 1, respectively. The degrees of the nodes that connect to others in the network are represented byĀ the areas of the nodes. The PPI network figure is generated using a multifunctional online software, Network Analyst (https://www.networkanalyst.ca)23,24.

To identify the nodes that are implicated in the aforementioned pathways, medullary analysis was carried out. A total of 9 functional clustered modules and related hub genes were discovered. Ā However, only modules with majority of the nodes were used for redesigning of the modular network. Modules 0 and 1 were observed to contain aĀ significant number of the nodes (p < 0.05) that contributed to the activation of pathways mentioned above. The top two significant modules were presented in different colours (Fig.Ā 2). The results demonstrate that module 0 (coloured blue) and module 1 (coloured red) are key players in in the PPI network of the TCCSUPPi cells, which means that the cluster of genes in modules 0 and 1 act together to promote the development of NDV persistent infection in TCCSUP bladder cancer cell line. The results further illustrate how the upregulated RPL8 and downregulated HSPA1A/HSPA4 are functionally connected (Fig.Ā 2).

Protein drug interactions in TCCSUPPi

To identify drug interactions between these connected nodes, we carried out protein-drug interaction analysis using the upregulated nodes that included RPL8 and THBS1 and downregulated nodes that included F2 and HSPA4. Two subnetworks were identified. Subnetwork 1 comprises of 104 nodes, 103 edges and 1 seed while subnetwork 2 has 4 nodes, 3 edges and 1Ā seed. Based on the analysis of subnetwork 1, several drugs were identified to be connected to the coagulation factor II, thrombin (F2) node. The top major drugs that are linked to F2 are lepirudin, bivalirudin, drotrecogin alfa, coagulation factor IX (recombinant), menadione, argatroban, and proflavine (Fig.Ā 3A). The remaining list of drugs can be found in the appendix. In subnetwork 2, ribosomal protein L8 (RPL8) that was upregulated is linked to alpha-hydroxy-beta-phenyl-propionic acid, anisomycin and puromycin drugs (Fig.Ā 3B). These results demonstrate that the identified drugs above can beĀ used to potentially suppress the upregulated and downregulated nodes, making it possible to prevent TCCSUP bladder cancer cells from acquiringĀ NDV persistent infection.

Figure 3
figure 3

Protein-drug interaction network. (A) The figure illustrates interactions between the downregulated node (F2) and several multiple drugs. (B) The figure shows interactions between the upregulated node (RPL8) and three drugs. The PPI network figure (A,B) were generated using a multifunctional online software, Network Analyst (https://www.networkanalyst.ca)23,24.

PPI network of DEGs from EJ28Pi

Subsequently, the 134 DEGs associated withĀ EJ28Pi cells were analysed for protein–protein network interactions. A network comprising of 14 subnetworks including one continent (subnetwork 1) and 13 islands (subnetwork 2–14)Ā was identified. The network with the highest scores were selected and analysed in order to provide an insight into the mechanisms associated with theĀ development of NDV persistent infectionĀ in EJ28 cells. Subnetwork 1 had 1161 nodes, 1662 edges and 57 seeds (Fig.Ā 4A), while subnetwork 2 had 21 nodes, 20 edges and 1 seed (Fig.Ā 4B). TheĀ expression of each node is represented by different colours while theĀ degree of connection between nodes are represented by theĀ area. The top 16 hub nodes from the entire network analysis were assessed for their distribution and 15 hub nodes out of this total were from subnetwork 1. Only NADH ubiquinone oxidoreductase core subunit S2 (NDUFS2) was from subnetwork 2 (Fig.Ā 4C).

Figure 4
figure 4

First and second networks of EJ28Pi cells. (A) First identified network of EJ28Pi cells. TheĀ red and green colours represent the nodes expression that are up- and down-regulated, respectively. The expression levels are represented byĀ colour and the area of the nodes indicate the degrees in which the nodes areĀ connected to eachĀ other. Nodes with gene names are the top 8 nodes in the PPI network. (B) Second identified network of EJ28Pi cells. Subnetwork 2 shows only the upregulated node that is affected in the pathway. The red colour node is is upregulated in EJ28Pi cells. (C) Hub nodes in the PPI network of EJ28Pi cells. The top 16 hub nodes with their degreeĀ of connection levels are shown. Genes in blue colour are from subnetwork 1 and red from subnetwork 2. The PPI network figuresĀ A & B are generated using a multifunctional online software, Network Analyst (https://www.networkanalyst.ca)23,24.

Next, functional connections within the constructed network were studied and theĀ related nodes were re-designed (Fig.Ā 5). As shown, pathways of renal cell carcinoma, viral carcinogenesis, proteoglycans in cancer, prostate cancer, insulin resistance, Ras signalling, circadian rhythm, neurotrophin signalling, cell cycle, aldosterone-regulated sodium reabsorption, FoxO signalling, microRNAs in cancer, Wnt signalling, influenza A, tight junction, viral myocarditis, Kaposi’s sarcoma-associated herpesvirus infection, PI3K-Akt signalling, epithelial cell signaling in Helicobacter pylori infection, adipocytokine signalling, epstein-Barr virus infection, adherens junction, bacterial invasion of epithelial cells, cAMP signalling, HTLV-I infection, Longevity regulating pathway, glucagon signaling pathway, leukocyte transendothelial migration, AMPK signaling pathway, pathways in cancer, natural killer cell mediated cytotoxicity and measles were significantly enriched (p < 0.05) in the nodes containing upregulated DEGs (subnetwork 1). The complete list of these pathways and their false discovery rates (FDRs) are listed in Supplementary Table S3.

Figure 5
figure 5

Modules 4, 5, 6, 7 and 8 of the EJ28Pi PPI network. The modules in red, blue, green, black and brown are for modules 4, 5, 6, 7, and 8, respectively. The degrees of connection between nodes in the network are represented by theĀ areas of the nodes. The PPI network figure was generated using a multifunctional online software, Network Analyst (https://www.networkanalyst.ca)23,24.

Meanwhile, pathways of Wnt signaling, proteoglycans in cancer, HTLV-I infection, cancer, breast cancer, cellular senescence, melanoma, glioma, MAPK signaling, transcriptional misregulation in cancer, focal adhesion, prostate cancer, endocrine resistance, Th17 cell differentiation, thyroid hormone signaling, thyroid cancer, FoxO signaling, bladder cancer, measles, oxytocin signaling, hippo signaling, amyotrophic lateral sclerosis (ALS), hepatitis C, Jak-STAT signaling, hepatitis B, endometrial cancer, basal cell carcinoma, axon guidance, mitophagy—animal, central carbon metabolism in cancer, Inflammatory bowel disease (IBD), non-small cell lung cancer, Kaposi’s sarcoma-associated herpesvirus infection, long-term potentiation, amphetamine addiction, PI3K-Akt signaling, p53 signaling, platinum, drug resistance, pancreatic cancer, chronic myeloid leukemia, regulation of actin cytoskeleton, colorectal cancer, Th1 and Th2 cell differentiation, small cell lung cancer, choline metabolism in cancer, neurotrophin signaling pathway, cell cycle, platelet activation and oocyte meiosis were significantly enriched in theĀ nodes containing the downregulated DEGsĀ (subnetwork 1). The identified pathways with their false discovery rates (FDRs) can be found in Supplementary Table S4.

The PPI network consisted of 41 modules but only modules with majority of the nodes were selected for redesigning of the modular network. Modules 4, 5, 6, 7, and 8 had the most significant number of nodes (p < 0.05) that contributed to the activation and enrichment of the aforementioned pathways. The modules are coloured red (module 4), blue (module 5), green (module 6), black (module 7) and brown (module 8) respectively (Fig.Ā 5). These five modules have been identified asĀ significantly (p < 0.05) acting together to contribute towards the development of NDV persistent infection in EJ28 bladder cancer cells; by virtue of theirĀ functionally connections via theĀ upregulation of EP300, IRS1, PTPN11, and RAC1and downregulation of TP53, SP1, CCND1 and XPO1 respectively.

Protein-drug interaction network analysis was performed. The upregulated nodes that include EP300, IRS1, PTPN11, and RAC1, as well as the downregulated nodes that includeTP53, SP1, CCND1 and XPO1 were mapped to theĀ DrugBank database for matching nodes to obtain specificĀ drug interaction information. Two subnetworks were discovered with subnetwork 1 containing 4 nodes, 3 edges and 1 seed and subnetwork 2 containing 3 nodes, 2 edges and 1seed. Dextromethorphan and guanosine-5′-diphosphate drugs were identified to effectively interact with theĀ upregulated rac family small GTPase 1 (RAC1) node (Fig.Ā 6A). In subnetwork 2, tumour protein P53 (TP53) that was downregulated is connected to acetylsalicylic acid, AZD, and 1-(9-ethyl-9H-carbazol-3-yl)-N-methylmethanamine drugs (Fig.Ā 6B), which basically means that these drugs can be potentiallyĀ used together with NDV to enhance the oncolytic activity of NDV against EJ28 bladder cancer cell lines and prevent the cells from acquiring persistent infection.

Figure 6
figure 6

Protein-drug interaction network. (A) Interactions between theĀ upregulated node (RAC1) and two drugs. (B) Interactions between upregulated node (TP53) and three drugs.

Validation of hub genes in Oncomine

To validate the expression profiles of the top hub genes that were identified in the protein–protein interaction networks, mRNA expression mining of these hub genes in publicly available Oncomine database (www.oncomine.org)25 that included the upregulated nodes RPL8 and THBS1 as well as the downregulated nodes F2 in TCCSUPPi cells was carried out. Likewise, the upregulated node RAC1 and downregulated node TP53 obtained from EJ28Pi cells were subsequently investigated. The results show that among the three hub genes (RPL8, THBS1 and F2) obtained from TCCSUPPi cells, RPL8 was significantly upregulated in bladder cancer cells as compared to normal bladder tissue (GSE3167; Fig.Ā 7 upper left; p = 6.36Eāˆ’5)26. In a study undertaken by Kim, Kim27, relative expression of THBS1 and F2 were slightly higher in bladder cancer than in the normal bladder tissue but the difference was not statistically significant (GSE13507; Fig.Ā 7 upper middle & upper right; p > 0.05), suggesting that RPL8, THBS1 and F2 genes potentiallyĀ play a vital role in bladder cancer progression and metastasis. These genes may also be the reason why cancer cells are able to maintain normal growth and development even after NDV infection as seen in the case of the established persistent TCCSUPPi cells.

Figure 7
figure 7

Expression of RPL8, THBS1, F2, TP53, and RAC1 mRNA levels obtained from TCCSUPPi and EJ28Pi. (A) Expression of RPL8, THBS1 and F2 mRNA levels obtained from TCCSUPPi in bladder cancer using Oncomine. (B) Expression of TP53 and RAC1 mRNA levels obtained from EJ28Pi in bladder cancer using Oncomine. Left plot representsĀ the expression in normal bladder tissue while the right plot represents the expression inĀ bladder cancer tissue.

TP53 and RAC1 mRNA expression levels in bladder cancer tissuesĀ were shown to be lower than thatĀ in the normal bladder tissue (GSE13507; Fig.Ā 7 lower left & lower right; p > 0.05). The data suggest that TP53 and RAC1 are downregulated in bladder cancer andĀ may likely contribute towards the development of NDV persistent infection in EJ28 cells; noting that TP53 was actuallyĀ found to be upregulated in the established persistent EJ28Pi cells.



Source link

Related Articles

Leave a Reply

Stay Connected

9FansLike
4FollowersFollow
0SubscribersSubscribe
- Advertisement -spot_img

Latest Articles

%d bloggers like this: