Revamped transposon-induced hepatocellular carcinoma model reveals deep similarities of cancer-driver landscapes across mice and humans

Genome-scale cancer-driver discovery is limited by statistical uncertainty in driver calling and by experimental constraints on mutations available for selection. We developed 3×fOnc3—a three-frame Sleeping Beauty (SB) mutagenesis system designed to broaden functional perturbations and reduce integration biases of traditional SB screens—and tested it in vivo in mouse liver. Within 12 months, 60% of mice developed heterogeneous hepatocellular tumors. Tumor-selected integrations were interpreted against an empirical reference from SB-based therapeutic gene delivery to hepatocytes, enabling correction for gene targetability. Background-aware filtering prioritized 315 driver candidates, including 243 protein-coding genes. Tumors were multiclonal, and identical integration-marked lineages expanded differently across lesions, supporting context-dependent clonal expansion. SB hits within young L1 subfamily members were selectively enriched in tumors. Human orthologues of candidate drivers were enriched for Cancer Gene Census genes and protein-altering pan-cancer mutations. Together, 3×fOnc3 reveals a background-corrected driver landscape linking driver selection to clonal tumor evolution and cross-species cancer relevance.