Abstract

Background: Patients with primary sclerosing cholangitis (PSC) are at a 400-fold increased risk of cholangiocarcinoma (CCA), and are thus recommended to undergo annual screening with magnetic resonance cholangiopancreatography (MRCP). Unfortunately, MRCP findings are nonspecific for CCA in the setting of biliary fibrosis and screening may not confer a survival benefit. We have previously developed a new class of diagnostic tools that detect dysregulated protease activity in the tumor microenvironment. In this work, we sought to leverage one such tool, activatable zymography probes (AZPs), which enable visualization of tumor-associated protease dysregulation ex vivo and in vivo. Because proteases have been shown to be dysregulated in CCA, we set out to develop a protease activity-based diagnostic for PSC-associated CCA.

Methods: We induced CCA tumor formation in the Mdr2^-/- mouse model of biliary fibrosis via hydrodynamic injection of plasmids encoding for AKT and YAP1, and the Sleeping Beauty (SB) transposase (Fig. 1A). We then synthesized a panel of 26 AZPs and applied them to fresh frozen liver tissue sections from tumor-bearing and Mdr2^-/- mice (Fig. 1B). Finally, we administered PEGylated AZPs into tumor-bearing and Mdr2^-/- mice and performed IVIS imaging to assess tumor uptake in vivo (Fig. 1E).

Results: We found that 14 AZPs, most notably AZP6, exhibited significantly increased binding to CCA tumors (p_adj­ < 0.05, fold change > 1.5) relative to fibrotic bile ducts (Fig. 1C), which was abrogated by protease inhibitors (Fig. 1D, "INH"). We found that AZP6 bound significantly more strongly to cells with high expression of vimentin (p < 0.01) and low expression of cytokeratin 7 (p < 0.001), suggesting that AZP6 is cleaved by mesenchymal cells (not shown). To assess the generalizability of AZP6 across CCA models, we tested it on tissue sections from an FBXW7/AKT-mutant CCA model and found that it specifically labeled tumors in a pattern similar to that seen in the YAP1/AKT model (not shown). Finally, we found significantly increased fluorescence in explanted livers from mice with CCA after in vivo administration of AZP6 (p < 0.01) (Fig. 1F, G).

Conclusion: AZPs detect and localize dysregulated protease activity ex vivo and in vivo in mouse models of CCA, without false positives from benign biliary fibrosis. Protease-activated diagnostics like AZPs, when used in conjunction with existing screening tools, may enable earlier and more accurate detection of CCA in PSC.