LAUSR

BACKGROUND & AIMS Alterations of individual genes variably affect the development of hepatocellular carcinoma (HCC). Thus, we aimed to characterize the function of tumor-promoting genes in the context of gene regulatory networks (GRNs). METHODS Using data from The Cancer Genome Atlas, from the LIRI-JP (Liver Cancer - RIKEN, JP project), and from our transcriptomic, transfection and mouse transgenic experiments, we identify a GRN which functionally links LIN28B-dependent dedifferentiation with dysfunction of β-catenin (CTNNB1). We further generated and validated a quantitative mathematical model of the GRN using human cell lines and in vivo expression data. RESULTS We found that LIN28B and CTNNB1 form a GRN with SMARCA4, Let-7b (MIRLET7B), SOX9, TP53 and MYC. GRN functionality is detected in HCC and gastrointestinal cancers, but not in other cancer types. GRN status negatively correlates with HCC prognosis, and positively correlates with hyperproliferation, dedifferentiation and HGF/MET pathway activation, suggesting that it contributes to a transcriptomic profile typical of the proliferative class of HCC. The mathematical model predicts how the expression of GRN components changes when the expression of another GRN member varies or is inhibited by a pharmacological drug. The dynamics of GRN component expression reveal distinct cell states that can switch reversibly in normal conditions, and irreversibly in HCC. The mathematical model is available via a web-based tool which can evaluate the GRN status of HCC samples and predict the impact of therapeutic agents on the GRN. CONCLUSIONS We conclude that identification and modelling of the GRN provide insights into the prognosis of HCC and the mechanisms by which tumor-promoting genes impact on HCC development. LAY SUMMARY Hepatocellular carcinoma (HCC) is a heterogeneous disease driven by the concomitant deregulation of several genes functionally organized as networks. Here, we identified a gene regulatory network involved in a subset of HCCs. This subset is characterized by increased proliferation and poor prognosis. We developed a mathematical model which uncovers the dynamics of the network and allows us to predict the impact of a therapeutic agent, not only on its specific target but on all the genes belonging to the network.