Deleting the β-catenin degradation domain in mouse hepatocytes drives hepatocellular carcinoma or hepatoblastoma-like tumor growth

Abstract

One-third of hepatocellular carcinomas (HCCs) harbor mutations activating the β-catenin pathway, predominantly via mutations in the CTNNB1 gene itself. Mouse models of Apc loss-of-function are widely used to mimic β-catenin-dependent tumorigenesis. Given the low prevalence of APC mutations in human HCCs, we aimed to generate liver tumors through CTNNB1 exon 3 deletion (βcatΔex3). We then compared βcatΔex3 liver tumors with liver tumors generated via frameshift in exon 15 of Apc (Apcfs-ex15). We used hepatocyte-specific and inducible mouse models generated through either a Cre-Lox or a CRISPR/Cas9 approach using adeno-associated virus vectors. Tumors generated by the Cre-Lox models were phenotypically analyzed using immunohistochemistry and were selected for transcriptomic analysis by RNA-sequencing (RNAseq). Mouse RNAseq data were compared to human RNAseq data (8 normal tissues, 48 HCCs, 9 hepatoblastomas) in an integrative analysis. Tumors generated via CRISPR were analyzed using DNA sequencing and immuno-histochemistry. Mice with CTNNB1 exon 3 deletion in hepatocytes developed liver tumors indistinguishable from Apcfs-ex15 liver tumors. Both Apcfs-ex15 and βcatΔex3 mouse models induced growth of phenotypically distinct tumors (differentiated or undifferentiated). Integrative analysis of human and mouse tumors showed that differentiated mouse tumors cluster with well-differentiated human CTNNB1-mutated tumors. Conversely, undifferentiated mouse tumors cluster with human mesenchymal hepatoblastomas and harbor activated YAP signaling. Apcfs-ex15 and βcatΔex3 mouse models both induce growth of tumors that are transcriptionally similar to either well-differentiated and β-catenin-activated human HCCs or mesenchymal hepatoblastomas. New and easy-to-use transgenic mouse models of primary liver cancers have been generated, with mutations in the gene encoding beta-catenin, which are frequent in both adult and pediatric primary liver cancers. The mice develop both types of cancer, constituting a strong preclinical model.