Preview

PDF, English Download (1MB) | Terms of use

Abstract

Primary liver cancer (PLC) comprises hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (iCCA), which may occur together as combined HCC-CCA (cHCC-CCA). For both tumor components, a single cell of origin is suggested. It is debated whether the cell of origin belongs to the stem/progenitor cell type or may represent a fully differentiated liver cell. It is believed that iCCA develops from biliary epithelial cells and that hepatocytes give rise to HCC. However, lineage tracing experiments in mice showed that iCCA may develop from hepatocytes while HCC can originate from ductular cells. These results suggest that differentiated liver epithelial cells can transdifferentiate based on their cellular plasticity. Mutation spectra of both hepatocellular carcinoma and cholangiocarcinoma have been published: However, it remains to be elucidated, which of these alterations are driving the tumor phenotype or determine its biological behavior. Determining the underlying mechanisms will have profound impact on our understanding of cancer biology and treatment options. In this project, an interspecies approach combining human, mouse, and in vitro data was performed to identify factors that affect the phenotype of liver cancer cells. Human primary liver cancer samples were selected by morphological analysis and subjected to genome-wide exomic and transcriptomic profiling. Next generation sequencing (NGS) data were integrated and candidate genes potentially affecting the phenotype of the tumor cells were identified. Between the two components of cHCC-CCA, 54 differentially expressed and/or differentially mutated genes were found. These were functionally validated by in vivo RNAi screening in mosaic mouse models of HCC (MYC-AKT1 in wildtype mice) and iCCA (KRASG12V in p19-deficient mice), which were generated by hydrodynamic tail vein injection of transposon vectors. Histological analysis of formalin-fixed paraffin-embedded individual tumor nodules followed by immunohistological assessment of hepatocellular (Hnf4α) and biliary (Sox9, Krt19) markers allowed for the identification of potential phenotype modulating genes. Thrombospondin 3 (THBS3) was identified as a phenotypic driver: THBS3 was mutated in the HCC compartment of human cHCC-CCA. Both its knockdown and expression of the synonymous mutation Thbs3R102Q in the iCCA mouse model resulted in a cHCC-CCA phenotype. Functional analyses were conducted in a primary murine isogenic iCCA cell line in vitro. The findings were confirmed in a human iCCA cell line. In both murine iCCA and HCC 8 models, Thbs3 wildtype maintains a cholangiocytic tumor phenotype (Sox9+, Krt19+, Hnf4α-) in vitro, while Thbs3R102Q expression promoted a hepatoid phenotype (Sox9-, Krt19-, Hnf4α+). Thbs3 knockdown largely phenocopied Thbs3R102Q expression in the iCCA cell line. Also, the results from functional assays suggested a tumor suppressor role of Thbs3 wildtype, while at the same time Thbs3 wildtype seems to be essential for the survival of iCCA cells. Furthermore, transcriptomic and ATAC sequencing analysis was performed on Thbs3 variant-expressing and knockdown cell lines in the iCCA model. NGS data integration and subsequent GSEA against the M2 mouse collection of the Molecular Signatures Database enabled the retrieval of four canonical pathways, including TGFβ signaling. Concluding, this study found that THBS3 knockdown alters TGFβ signaling at transcriptional level.