Abstract

Hepatoblastoma is a malignant disease of the liver. It accounts for about 1 % of all childhood cancers and is the most common malignant liver tumor in infancy. Hepatoblastoma is assumed to arise from immature liver progenitor cells by aberrant activation of genes important in the embryonic development. Based on its early manifestation it is generally assumed that hepatoblastoma displays a relatively normal genomic background. Whole-exome sequencing performed in our group identified hepatoblastoma as one of the genetically simplest tumors ever described, with recurrent mutations in beta-catenin (CTNNB1) and nuclear factor (erythroid-derived 2)-like 2 (NFE2L2). Based on this finding we performed targeted genotyping of a large cohort of primary hepatoblastomas, hepatoblastoma cell lines and transitional liver cell tumors and identified CTNNB1 and NFE2L2 to be mutated in 72.5 % and 9.8 % of cases, respectively. CTNNB1 is a key effector molecule of canonical WNT signaling pathway, a pathway that is essential in organogenesis and cellular processes such as cell proliferation, differentiation, survival and apoptosis. However, NFE2L2 is involved in the activation of the cellular antioxidant response to combat the harmful effects such as xenobiotics and oxidative stress. Interestingly, all NFE2L2 mutations were located in or adjacent to the DLG and ETGE motifs of the NFE2L2 protein that are needed to get recognized by the KEAP1/CUL3 complex for proteasomal degradation. Functional analysis showed that cells transfected with mutant NFE2L2 were insensitive to KEAP1-mediated downregulation of NFE2L2 signaling and that depletion of the NFE2L2 via siRNA downregulates the NAD(P)H dehydrogenase (quinine) 1 (NQO1), a target gene of NFE2L2, and inhibits proliferation. In the clinical setting, NQO1 overexpression in tumors was significantly associated with metastasis, vascular invasion, the adverse prognostic C2 gene signature as well as poor outcome. RNA sequencing in our group identified the ubiquitin-like with PHD and ring finger domains 1 (UHRF1), a protein known to preferentially bind to hemi-methylated DNA, to be highly overexpressed in hepatoblastoma tumors. UHRF1 is as a key regulator in the epigenetic crosstalk, by controlling DNA methylation and histone modification. Using immunoprecipitation, we were able to show that UHRF1 binds in concert with DNA methyltransferase 1 (DNMT1) and ubiquitin specific peptidase 7 (USP7) as a trimeric complex to promoter regions of tumor suppressor genes (TSG) relevant in hepatoblastoma, such as hedgehog interacting protein (HHIP), insulin-like growth factor binding protein 3 (IGFBP3), and secreted frizzled-related protein 1 (SFRP1). These genes are epigenetically silenced in hepatoblastoma, as evidenced by heavy DNA methylation and enrichment of the repressive H3K27me3 and H3K9me2 chromatin mark. Interestingly, knockdown of UHRF1 expression via RNA interference resulted in promoter demethylation, but no reactivation of TSG gene expression. Additionally, no effect on tumor cell proliferation was observed after UHRF1 knockdown. Chromatin immunoprecipitation experiments revealed a decrease of the repressive chromatin marks H3K27me3 and H3K9me2 after UHRF1 depletion, but neither a clear shift towards the active H3K4me2 chromatin mark nor enrichment of RNA Polymerase at the TSG loci was observed. Statistical analyses revealed that a high expression of UHRF1 was associated with advanced disease state and a worse overall survival. Taken together our study demonstrates that activation of WNT signaling in concert with activation of the NFE2L2-KEAP1 pathway might be the driving force in the development of liver cancers. Moreover, we defined aberrant NQO1 expression as a marker for adverse course of disease and poor outcome. In addition, we showed that an aberrant expression of the epigenetic regulator UHRF1 and its excessive binding on promoter regions results in methylation of TSGs. This may represent an important mechanism in the initial phases of embryonal tumorigenesis. However, UHRF1 depletion alone was not sufficient to re-induce TSG expression. Therefore, UHRF1 might be more useful as a biomarker for the prognosis of hepatoblastoma than a direct anti-cancer target for hepatoblastoma therapy.