Simultaneous Yap and Sox9 repression eliminates Notch‐dependent hepatocyte‐driven cholangiocarcinogenesis

The most critical barrier for developing a globally effective regimen for Intrahepatic cholangiocarcinoma (ICC) is its intra‐ and extratumoral heterogeneity, which may be partly due to diverse cellular origin within the liver. Recent studies have suggested hepatocytes (HCs) as a cell source in a subset of ICCs, especially those associated with chronic liver injury due to non‐alcoholic steatohepatitis (NASH) or primary sclerosing cholangitis (PSC). Given that co‐expression of myrAKT with either NICD or YAP S127A in HCs using hydrodynamic tail vein injection leads to ICC, we initiated a comprehensive mechanistic analysis of ICC development in patients and in the preclinical mouse model. Based on IHC analysis of 108 CC patients, over 90% of CC samples exhibited high levels of nuclear SOX9 & YAP. Interestingly, we identified an enrichment of AKT activation in ICC (50%) compared to extrahepatic CC (17%), further supporting the clinical relevance of the myrAKT/NICD model for studying human ICC. We also identified significant upregulation of p‐AKT, SOX9 & YAP in HCs of patients with PSC and NASH, which are well‐known risk factors for ICC, as well as in murine models of cholestatic injury and NASH. Next, we assessed the function of Sox9 and Yap in ICC tumorigenesis. While activation of AKT/NICD led to lethal ICC, conditional deletion of either Yap or Sox9 in this model dramatically reduced cholangiocarcinogenesis. Yap deletion impaired HC‐to‐biliary epithelial cell (BEC) fate conversion and tumor cell proliferation, while Sox9 elimination only repressed proliferation but had no effect on HC‐to‐BEC reprogramming. Interestingly, following deletion of either Yap or Sox9 we observed a few AKT/NICD‐driven ICC tumors expressing either Sox9 or Yap but not both. This also occurred in a small subset of human CC tumors which were either Sox9+Yap− (4.6%) or Sox9‐Yap+ (3.7%), showing that deletion of Yap or Sox9 is not sufficient to completely abrogate Notch‐dependent ICC development. Co‐expression of AKT with Yap also led to development of HC‐derived ICC with some mixed hepatocellular carcinoma (HCC), showing that Yap is sufficient to induce ICC. However, co‐expression of AKT with Sox9 in HCs did not result in tumor formation. Remarkably, Sox9 deletion from AKT/YAP‐driven tumors causes a phenotype switch from ICC to aggressive HCC‐like tumors, suggesting context‐dependent pathologic roles for Sox9. Finally, we demonstrated that conditional deletion of both Yap & Sox9 completely blocked development of ICC tumors in the AKT/NICD model. Thus, we show that cholestasis or NASH in humans and mice induces HC‐to‐BEC reprogramming which may correlate with increased risk of ICC development. We also provide evidence for critical but distinct roles of Yap and Sox9 in ICC development and demonstrate the therapeutic potential of targeting both of these factors for treatment of subsets of ICC. Support or Funding Information NIH grants 1R01DK62277, 1R01DK100287, R01CA204586 and Endowed Chair for Experimental Pathology to S.P.M; 1P30DK120531‐01 to Pittsburgh Liver Research Center (PLRC); PLRC Pilot & Feasibility grant PF 2019‐05 to S.K.