Canonical Wnt signaling is antagonized by noncanonical Wnt5a in hepatocellular carcinoma cells.

BACKGROUND: beta-catenin mutations that constitutively activate the canonical Wnt signaling have been observed in a subset of hepatocellular carcinomas (HCCs). These mutations are associated with chromosomal stability, low histological grade, low tumor invasion and better patient survival. We hypothesized that canonical Wnt signaling is selectively activated in well-differentiated, but repressed in poorly differentiated HCCs. To this aim, we characterized differentiation status of HCC cell lines and compared their expression status of Wnt pathway genes, and explored their activity of canonical Wnt signaling.

RESULTS: We classified human HCC cell lines into "well-differentiated" and "poorly differentiated" subtypes, based on the expression of hepatocyte lineage, epithelial and mesenchymal markers. Poorly differentiated cell lines lost epithelial and hepatocyte lineage markers, and overexpressed mesenchymal markers. Also, they were highly motile and invasive. We compared the expression of 45 Wnt pathway genes between two subtypes. TCF1 and TCF4 factors, and LRP5 and LRP6 co-receptors were ubiquitously expressed. Likewise, six Frizzled receptors, and canonical Wnt3 ligand were expressed in both subtypes. In contrast, canonical ligand Wnt8b and noncanonical ligands Wnt4, Wnt5a, Wnt5b and Wnt7b were expressed selectively in well- and poorly differentiated cell lines, respectively. Canonical Wnt signaling activity, as tested by a TCF reporter assay was detected in 80% of well-differentiated, contrary to 14% of poorly differentiated cell lines. TCF activity generated by ectopic mutant beta-catenin was weak in poorly differentiated SNU449 cell line, suggesting a repressive mechanism. We tested Wnt5a as a candidate antagonist. It strongly inhibited canonical Wnt signaling that is activated by mutant beta-catenin in HCC cell lines.

CONCLUSION: Differential expression of Wnt ligands in HCC cells is associated with selective activation of canonical Wnt signaling in well-differentiated, and its repression in poorly differentiated cell lines. One potential mechanism of repression involved Wnt5a, acting as an antagonist of canonical Wnt signaling. Our observations support the hypothesis that Wnt pathway is selectively activated or repressed depending on differentiation status of HCC cells. We propose that canonical and noncanonical Wnt pathways have complementary roles in HCC, where the canonical signaling contributes to tumor initiation, and noncanonical signaling to tumor progression.