A phospho-switch controls RNF43-mediated degradation of Wnt receptors to suppress tumorigenesis

Abstract

Frequent mutation of the tumour suppressor RNF43 is observed in many cancers, particularly colon malignancies. RNF43, an E3 ubiquitin ligase, negatively regulates Wnt signalling by inducing degradation of the Wnt receptor Frizzled. In this study, we discover that RNF43 activity requires phosphorylation at a triplet of conserved serines. This phospho-regulation of RNF43 is required for zebrafish development and growth of mouse intestinal organoids. Cancer-associated mutations that abrogate RNF43 phosphorylation cooperate with active Ras to promote tumorigenesis by abolishing the inhibitory function of RNF43 in Wnt signalling while maintaining its inhibitory function in p53 signalling. Our data suggest that RNF43 mutations cooperate with KRAS mutations to promote multi-step tumorigenesis via the Wnt-Ras-p53 axis in human colon cancers. Lastly, phosphomimetic substitutions of the serine trio restored the tumour suppressive activity of extracellular oncogenic mutants. Therefore, harnessing phospho-regulation of RNF43 might be a potential therapeutic strategy for tumours with RNF43 mutations.

Post‐translational Wnt receptor regulation: Is the fog slowly clearing?

“…The phosphorylations of conserved serines by CK1 function as a phospho‐switch and confer the ability to downregulate Fzd by ubiquitination. [ 59,60 ] These findings provide novel insights into the regulation of Wnt receptor proteins on the plasma membrane that complement the mechanisms already known, although further details are yet to be established.…”

“…Only the extracellular PA domain, intracellular RING finger domain, and cytoplasmic serine‐rich region harboring the phospho‐switch exhibit moderate conservation, while the other parts show a low level of similarity (Figure 1). [ 59,60 ] Notwithstanding the low similarity across whole proteins, essential regions required for the downregulation of Fzd, such as the RING finger domain for Fzd ubiquitination and the phospho‐switch for functional activation, are well conserved. Therefore, it is expected that the combined inactivation of RNF43 and ZNRF3 is necessary to cause tumorigenesis, due to the compensatory nature of the two ubiquitin ligases.…”

“…Interestingly, the co‐occurrence of mutations in RNF43 and ZNRF3 genes is not frequently observed in tumor samples. [ 59 ] These results suggest that RNF43 mutations might not be simple LOF mutations, but function as dominant‐negative (DN) mutations, although the expression of ZNRF3 is sometimes suppressed by epigenetic modifications (e.g., hypermethylation). [ 66 ] Furthermore, it has been reported that the contribution of ZNRF3 mutations to tumorigenesis is much lower than that of RNF43 mutations.…”

“…Instead, the interaction of RNF43‐Dvl and the interaction of Dvl‐Fzd appeared to be mutually exclusive (Figure 2B) and the interaction of RNF43 with Fzd occurred independently of RNF43‐Dvl binding. [ 55,59 ] Moreover, the exact Dvl‐interacting site on RNF43 and the RNF43‐binding domain on Dvl are still controversial. [ 55,73,74 ] We believe that additional studies will resolve this issue in the near future.…”

“…Even an oncogenic DN mutant of RNF43 harboring mutations in the extracellular domain (R127P) that lead to accumulation in the ER, was shown to regain activity upon introduction of phospho‐mimetic substitutions in those conserved serines. [ 59 ] As the R127P mutant RNF43 shows dominant ER retention even after the introduction of phospho‐mimetic mutations, there is a possibility that the ER‐retaining RNF43 also shows ubiquitination activity toward Fzd. However, we do not know yet where RNF43 is normally phosphorylated and activated by CK1 during intracellular membrane trafficking from the ER to the cell surface (Figure 2C).…”