Integrated regulatory-metabolic network model reveals critical mechanism and potential targets for Hepatocellular Carcinoma

Abstract Background HCC (Hepatocellular carcinoma), the predominant form of liver cancer, has long been the top three leading cause of death in cancer worldwide. Although researchers have spent lot of effort to identify molecular targets available for treatment, the high tumor heterogeneity makes it difficult to develop effective therapy options and the drug response remains low. Under this circumstance, the precise stratification strategies are more than required. However, previous researches generally focused on single biological level, such as genome, transcriptome, or proteome, and are not able to discover effective therapeutic targets, so the systematic study of both regulation and metabolism of HCC is needed. Methods In this paper, we use two different algorithms to reconstruct regulatory networks for both HCC and normal liver cells, then integrate them with corresponding metabolic models in order to discover TFs (transcriptional factors) affecting tumorigenesis. Furthermore, a machine learning algorithm is utilized to classify HCC samples, differentially expressed genes, altered metabolic reactions and biological pathways are identified in lowest overall survival (OS) rate sub-type compared to others. Results We classify TCGA-LIHC samples into three sub-types with significantly different OS rate, and this stratification strategy is validated in another independent dataset LIRI-JP. Then, we identify 5 key TFs affecting cancer cell growth and CREB3L3 is believed to be associated with poor prognosis. The comprehensive metabolic analysis on personalized metabolic models highlight 18 metabolic genes essential for tumorigenesis in all three sub-types of patients, besides, ACADSB and CMPK1 are highly possible to be strongly correlated with lower OS. Conclusions Among 20 metabolic genes identified through metabolic analysis, 15 of them have already been targeted by approved drugs according to DrugBank. In addition, miRNAs targeting key TFs and genes are also involved in well-known cancer related pathways. The multi-scale regulatory-metabolic model reveals the critical mechanism of HCC cell proliferation and suggests potential targets.