We aimed to explore molecular mechanism and drug candidates of vascular toxicities associated with melphalan after treating human retinal endothelial cells (RECs). GSE34381 microarray data was firstly downloaded and used to identify the differentially expressed genes (DEGs) in human REC treated with melphalan vs. untreated cells by limma package in R language. The transcription network was constructed based on TRANSFAC database and the top five transcription factors (TFs) were select with a measure of regulatory impact factor, followed by the construction of function modules. Gene ontology enrichment analyses were performed to explore the enriched functions. Connectivity Map analysis was conducted to predict the potential drugs overcoming the melphalan's actions on REC. Totally, 75 DEGs were identified, including 70 up-regulated and five down-regulated genes. Transcription network with 1311 nodes and 1875 edges was constructed and the top five TFs were CREM, MYC, FLI1, NF-κB1, and JUN. Functional modules indicated that NF-κB1 and MYC were the important nodes. The upregulated genes as well as the genes involved in the modules mainly participated in biological process of immune response, cell proliferation, and cell motion. Five small molecules were predicted to be potential drug candidates, including doxorubicin, fipexide, daunorubicin, tiabendazole, and GW-8510. Based on these results, we speculate that NF-κB1 and MYC might involve in the molecular mechanism of vascular toxicity induced by melphalan through regulating their target genes. Five small molecules might be drug candidates to overcome the melphalan-induced vascular toxicity via targeting to MYC and JUN.
Keywords: Connectivity Map analysis; Melphalan; Retinoblastoma; Transcription factor; Transcription network.