Collagens contain cryptic polypeptide modules that regulate major cell functions, such as cell proliferation or death. Collagen XVIII (C18) exists as three amino terminal end variants with specific amino terminal polypeptide modules. We investigated the function of the variant 3 of C18 (V3C18) containing a frizzled module (FZC18), which carries structural identity with the extracellular cysteine-rich domain of the frizzled receptors. We show that V3C18 is a cell surface heparan sulfate proteoglycan, its topology being mediated by the FZC18 module. V3C18 mRNA was expressed at low levels in 21 normal adult human tissues. Its expression was up-regulated in fibrogenesis and in small well-differentiated liver tumors, but decreased in advanced human liver cancers. Low FZC18 immunostaining in liver cancer nodules correlated with markers of high Wnt/β−catenin activity. V3C18 (Mr = 170 kD) was proteolytically processed into a cell surface FZC18-containing 50 kD glycoprotein precursor that bound Wnt3a in vitro through FZC18 and suppressed Wnt3a-induced stabilization of β−catenin. Ectopic expression of either FZC18 (35 kD) or its 50 kD precursor inhibited Wnt/β−catenin signaling in colorectal and liver cancer cell lines, thus downregulating major cell cycle checkpoint gatekeepers cyclin D1 and c-myc and reducing tumor cell growth. By contrast, full-length V3C18 was unable to inhibit Wnt signaling. In summary, we identified a cell-surface signaling pathway whereby FZC18 inhibits Wnt/β−catenin signaling. The signal, encrypted within cell-surface C18, is released by enzymatic processing as an active frizzled cysteine-rich domain (CRD) that reduces cancer cell growth. Thus, extracellular matrix controls Wnt signaling through a collagen-embedded CRD behaving as a cell-surface sensor of proteolysis, conveying feedback cues to control cancer cell fate.

The efficacy of anti-angiogenic therapies on cancer patients is limited by the emergence of drug resistance, urging the search for second-generation drugs. In this study, we screened an academic chemical library (DCM, University of Grenoble-Alpes) and identified a leader molecule, COB223, that inhibits endothelial cell migration and proliferation. It inhibits also Lewis lung carcinoma (LLC/2) cell proliferation whereas it does not affect fibroblast proliferation. The anti-angiogenic activity of COB223 was confirmed using several in vitro and in vivo assays. In a mouse LLC/2 tumor model, ip administration of doses as low as 4 mg/kg COB223 efficiently reduced the tumor growth rate. We observed that COB223 inhibits endothelial cell ERK1/2 phosphorylation induced by VEGF, FGF-2 or serum and that it acts downstream of PKC and upstream of Ras. This molecule represents a novel anti-angiogenic and anti-tumorigenic agent with an original mechanism of action that deserves further development as an anti-cancer drug.