The emergence of large-scale genomic, chemical and pharmacological data provides new opportunities for drug discovery and repositioning. In this work, we develop a computational pipeline, called DTINet, to predict novel drug–target interactions from a constructed heterogeneous network, which integrates diverse drug-related information. DTINet focuses on learning a low-dimensional vector representation of features, which accurately explains the topological properties of individual nodes in the heterogeneous network, and then makes prediction based on these representations via a vector space projection scheme. DTINet achieves substantial performance improvement over other state-of-the-art methods for drug–target interaction prediction. Moreover, we experimentally validate the novel interactions between three drugs and the cyclooxygenase proteins predicted by DTINet, and demonstrate the new potential applications of these identified cyclooxygenase inhibitors in preventing inflammatory diseases. These results indicate that DTINet can provide a practically useful tool for integrating heterogeneous information to predict new drug–target interactions and repurpose existing drugs.

ObjectiveIncreased de novo fatty acid (FA) synthesis and cholesterol biosynthesis have been independently described in many tumour types, including hepatocellular carcinoma (HCC).DesignWe investigated the functional contribution of fatty acid synthase (Fasn)-mediated de novo FA synthesis in a murine HCC model induced by loss of Pten and overexpression of c-Met (sgPten/c-Met) using liver-specific Fasn knockout mice. Expression arrays and lipidomic analysis were performed to characterise the global gene expression and lipid profiles, respectively, of sgPten/c-Met HCC from wild-type and Fasn knockout mice. Human HCC cell lines were used for in vitro studies.ResultsAblation of Fasn significantly delayed sgPten/c-Met-driven hepatocarcinogenesis in mice. However, eventually, HCC emerged in Fasn knockout mice. Comparative genomic and lipidomic analyses revealed the upregulation of genes involved in cholesterol biosynthesis, as well as decreased triglyceride levels and increased cholesterol esters, in HCC from these mice. Mechanistically, loss of Fasn promoted nuclear localisation and activation of sterol regulatory element binding protein 2 (Srebp2), which triggered cholesterogenesis. Blocking cholesterol synthesis via the dominant negative form of Srebp2 (dnSrebp2) completely prevented sgPten/c-Met-driven hepatocarcinogenesis in Fasn knockout mice. Similarly, silencing of FASN resulted in increased SREBP2 activation and hydroxy-3-methyl-glutaryl-CoA (HMG-CoA) reductase (HMGCR) expression in human HCC cell lines. Concomitant inhibition of FASN-mediated FA synthesis and HMGCR-driven cholesterol production was highly detrimental for HCC cell growth in culture.ConclusionOur study uncovers a novel functional crosstalk between aberrant lipogenesis and cholesterol biosynthesis pathways in hepatocarcinogenesis, whose concomitant inhibition might represent a therapeutic option for HCC.

MicroRNAs (miRNAs) are short non-coding RNAs that are involved in the epigenetic regulation of cellular processes. To identify more miRNAs which are involved in the macrophage inflammatory response to lipopolysaccharide (LPS) stimulation and dissect the mechanisms more clearly, microRNA profiling of LPS-treated RAW264.7 macrophage cells was performed by initial high-throughput array-based screen and further real-time RT-PCR validation; bioinformatics approaches were used to analyze the target genes of the differentially expressed miRNAs. Compared to the untreated control, two microRNAs (miR-146a and miR-155) with more than twofold higher expression and two microRNAs (miR-27a* and miR-532-5p) with twofold lower expression were detected by array-based screen, which can be validated by qRT-PCR, and more than 1,000 candidate target genes were detected by at least of one of four different algorithms (TargetScan, PicTar, miRDB, and microRNA.org); with gene ontology classification, we were able to correlate the upregulation and downregulation of miRNA to the differential expression of inflammation-related candidate target gene during LPS-induced inflammation. Our findings may provide the basic information for the precise roles of miRNAs in the macrophage inflammatory response to LPS stimulation in the future.

The emergence of large-scale genomic, chemical and pharmacological data provides new opportunities for drug discovery and repositioning. In this work, we develop a computational pipeline, called DTINet, to predict novel drug-target interactions from a constructed heterogeneous network, which integrates diverse drug-related information. DTINet focuses on learning a low-dimensional vector representation of features, which accurately explains the topological properties of individual nodes in the heterogeneous network, and then makes prediction based on these representations via a vector space projection scheme. DTINet achieves substantial performance improvement over other state-of-the-art methods for drug-target interaction prediction. Moreover, we experimentally validate the novel interactions between three drugs and the cyclooxygenase proteins predicted by DTINet, and demonstrate the new potential applications of these identified cyclooxygenase inhibitors in preventing inflammatory diseases. These results indicate that DTINet can provide a practically useful tool for integrating heterogeneous information to predict new drug-target interactions and repurpose existing drugs.