Background

Frizzled (FZD) proteins are G protein-coupled receptors (GPCRs). The extracellular N-terminus contains a cysteine-rich domain (CRD) through which FZDs bind WNT ligands. The intracellular C-terminus binds the PDZ domain of Dishevelled (Dvl) and interacts with G proteins. Up to now, 10 FZDs have been identified in human. These receptors mediate β-catenin pathway and various β-catenin-independent pathways depending on cellular context. WNT/β-catenin pathway is oncogenic and involved in almost every aspect of tumor development. In addition, β-catenin-independent WNT/PCP and WNT/Ca2+-PKC pathways are also implicated in tumor progression [1,2,3,4]. In recent years, some novel pathways such as WNT/Stat3 and WNT/Yap-Taz have been successively identified [5,6,7,8], indicating that FZD signalings are far more complicated and still incompletely comprehended.

Epithelial-mesenchymal transition (EMT) in tumor cells endows them with enhanced motility thereby increasing their metastatic potentiality. This process is characterized by reduced expression of epithelial-related factors and increased expression of mesenchymal-related factors. Mechanistically, EMT is driven by some transcription factors such as SNAI1/2 and ZEB1/2. Studies have shown that FZD signalings induce tumor cell EMT and metastasis. FZD4, FZD7 and FZD10 promote EMT through β-catenin pathway in prostate, liver and breast cancer [9,10,11]. FZD2 mediates WNT5-induced EMT dependent on Stat3 signaling in liver, lung, breast and colon cancer [5], and FZD8 mediates WNT11-induced EMT by interacting with TGF-β signaling in prostate cancer [12].

In gastric cancer, FZD4 and FZD7 similarly promote EMT and metastasis by activating β-catenin pathway [13, 14]. Genetic deletion of FZD7 was shown to inhibit the growth of gastric adenomas in vivo [15]. FZD2 and FZD8, through β-catenin pathway but not β-catenin-independent pathways, induce gastric cancer cell proliferation, migration, invasion and metastasis [16, 31]. Actually, WNT7A functions as an EMT inhibitor in lung cancer [32]. Moreover, interrogation of CCLE database showed that FZD5 is also associated with epithelial-like phenotype in lung cancer (data not shown).

As shown in our study, FZD5 is transcriptional inhibited by TEAD1-SNAI2 complex. Activation of tumor-suppressing Hippo pathway results in phosphorylation and degradation of Yap and Taz. When Yap and Taz are not phosphorylated, they translocate into the nucleus to bind transcription factor TEAD. Yap/Taz-TEAD signaling induces gastric cancer cell EMT, stemness, drug resistance and metastasis [33,41, 42], and promotes prostate cancer progression by interacting with NFκB [43]. In gastric cancer, ELF3 maintains epithelial-like phenotype, prevents EMT and is associated with longer survival, acting as a putative tumor suppressor. ELF3 also inhibits EMT in ovarian and breast cancer cells [44, 45]. Mechanistically, ELF3 represses activity and expression of EMT transcription factor ZEB1 [46, 47].

Activation of β-catenin pathway requires not only FZDs, but also their co-receptors LRP5/6 which is not involved in non-canonical β-catenin-independent pathway. A recent study reported that FZDs-LRP5/6 interaction can initiate β-catenin pathway in the absence of WNTs [48]. Surprisingly, another study revealed that FZDs-LRP5/6 interaction can block activation of FZD-mediated non-canonical pathway, further preventing tumor metastasis [49]. These findings, in combination with our observations, suggest that FZDs-LRP5/6 interaction may act as a tumor promoter by activating β-catenin pathway or a tumor suppressor by antagonizing β-catenin-independent pathway, depending on tumor or cell context.

Conclusions

In summary, our study for the first time uncovers that FZD5-ELF3 signaling prevents EMT and is associated with favorable prognosis in gastric cancer. Therefore, FZD5 and ELF3 function as putative tumor suppressors in this type of cancer. Antibodies against FZD5 including ipafricept and vantictumab have been shown effective on certain cancer types [50]. However, our study warns out that targeting FZD5 should be cautious and tumor-specific, since treatment with these antibodies may potentially promote metastasis in some other types of cancers including gastric cancer.