The Bcl9-beta-catenin interface as a potential therapeutic target for colorectal cancer

Aberrantly active Wnt signaling is the hallmark driver of colorectal cancer (CRC). Bcl9/9l proteins are components of the main transcriptional activation complex of canonical Wnt signaling and interact with beta-catenin via a highly conserved homology domain (HD2). Intestinal loss of Bcl9 affects wound healing and Lgr5+-cancer stem cell (CSC) maintenance, but does not perturb normal tissue homeostasis. The aim of the present thesis was to validate the Bcl9-β-catenin interaction as a potential target for modulating Wnt signaling and abolishing intestinal stemness traits in colorectal cancer. We investigated the role of the Bcl9-beta-catenin interaction in the mouse AOM-DSS chemical carcinogenesis CRC model. Subsequently, we validated our observations in an independent transgenic APC-Kras CRC model (truncated APC tumor suppressor and mutated Kras expressed in the intestine). The expression of Bcl9 proteins was abolished exclusively in the gastrointestinal epithelium using constitutive or inducible Cre-recombinase expressed under the control of the Villin promoter. To corroborate that the loss of Bcl9 phenotype was due to the abrogation of the Bcl9-beta-catenin interaction and not to a β-catenin-independent function of Bcl9, we made use of an additional Bcl9 mutant mouse strain in which only the HD2 of both Bcl9 proteins was deleted, selectively preventing the Bcl9-beta-catenin interaction. The inducible deletion of Bcl9 in established AOM-DSS tumors led to a loss in stem cell, Wnt and EMT gene expression signatures, recapitulating the phenotype that we previously described when Bcl9 was deleted constitutively. A 14d partial decrease of Bcl9 proved sufficient to suppress the expression of mesenchymal marker Vimentin in a cell autonomous fashion. Importantly, AOM-DSS tumors with Bcl9 proteins selectively lacking the HD2 beta-catenin interaction domain displayed the same phenotype as observed in tumors that entirely lacked Bcl9 proteins. Bcl9-knockout tumors induced in the Wnt active APC-Kras tumor model were compared with their wild-type counterparts. Wild-type APC-Kras mutant tumors displayed fewer mesenchymal traits when compared to AOM-DSS induced tumors. Global gene expression analysis by RNA sequencing confirmed a strong positive correlation between Bcl9/9l-dependent transcriptional changes in both independent tumor models. Genes differentially expressed between wild-type and Bcl9/9l ablated tumors were used to probe CRC databases for correlation with patient outcome and revealed a strong association with vastly improved disease-free and overall survival, independent of clinical and molecular parameters. We have shown that preventing the interaction of Bcl9 with beta-catenin in established tumors in Wnt-activated CRC models rapidly resulted in the loss of stemness and mesenchymal traits and concomitant induction of a more differentiated phenotype. Importantly, these loss-of-Bcl9 traits are prognostic for favorable outcome in CRC patients and we have not observed any overt anomalies upon intestinal ablation of Bcl9 in adult mice. Overall, our observations indicate that targeting the Bcl9-beta-catenin interaction in Wnt-activated tumors might be well tolerated and a generally applicable pharmacological strategy to target CSCs to potentially prevent tumor recurrence and improve progression free survival.