LAUSR

Dear Editor, Dishevelled-2 (DVL2) is a central adaptor protein that propagates signals from upstream receptors to downstream effector β-catenin in canonical Wnt signaling. At high local concentration, DVL2 assembles into signalosomes through DIX domain-mediated homo-polymerization to provide a docking platform for the recruitment of signaling components, which is critical for Wnt signaling transduction and activation; however, the exact underlying mechanism remains inconclusive. Although the roles of phosphorylation and ubiquitination in DVL2-puncta formation have been discussed before, the impact of lysine acetylation of DVL2 on this process remains elusive. Here we investigated the effect of acetylation on DVL2-puncta formation. Exogenous DVL2-formed puncta that can aggregate into larger molecular signalosomes under treatment with Wnt3a or nicotinamide (NAM, SIRT deacetylases inhibitor) in Hela cells, suggesting the involvement of acetylation in DVL2 polymerization (Fig. 1a). Indeed, endogenous DVL2 was acetylated upon overexpression of Wnt3a in HeLa cells (Supplementary Fig. S1a). We further identified the acetyltransferase responsible for DVL2 acetylation. Among the acetyltransferases tested, only CREB-binding protein (CBP) that shuttles between cytoplasm and nucleus, exhibited acetylation induction activity toward DVL2 in cells (Supplementary Fig. S1b). Mass spectrometry was employed to identify K68 in DVL2 as a CBP-induced acetylation site, which was confirmed by loss of acetylation after site-directed mutation of lysine 68 to arginine (K68R) in DVL2 (Fig. 1b, c). Then a specific antibody targeting acetyl-K68 (aK68) of DVL2 (anti-DVL2 K68ac) was developed and verified (Supplementary Fig. S1c). Using this antibody, we confirmed Wnt3a-induced endogenous DVL2-K68 acetylation (DVL2-aK68) in HCT116 cells (Fig. 1d). However, DVL2aK68 was reduced in CBP MEF cells, which cannot be enhanced by Wnt3a stimulation, further confirming the CBP-dependent DVL2 acetylation (Fig. 1e). To explore the role of CBP acetylation in DVL2-puncta formation, DVL2 and CBP were co-expressed (Supplementary Fig. S1d, right panel), or was expressed separately (Supplementary Fig. S1d, left panel), in HeLa cells. CBP colocalized with DVL2 and significantly promoted formation of DVL2 super-molecular puncta in cells (Supplementary Fig. S1d). Moreover, upon Wnt3a stimulation, endogenous CBP shuttled into cytoplasm from nucleoplasm to colocalize with endogenous DVL2 in HeLa cells (Supplementary Fig. S1e). Overall, these results indicate that CBPdependent DVL2 acetylation has a role in DVL2-puncta formation. Next, we investigated the enzymes responsible for DVL2 deacetylation. Treatment of cells with NAM, but not TSA (Trichostatin A, inhibitor of HDACs), significantly enhanced the levels of DVL2-aK68 regardless of Wnt3a stimulation (Supplementary Fig. S2a). These data suggest that one or more SIRT family members are responsible for DVL2 deacetylation. We thus tested effects of various SIRT proteins (SIRT1-7) on DVL2 deacetylation (Supplementary Fig. S2b). Among four SIRTs (SIRT1,2,4,5) that can deacetylate DVL2 when overexpressed in cells, SIRT4 and SIRT5 were excluded due to their mitochondrial localization. SIRT2, but not SIRT1, showed deacetylase activity toward DVL2-aK68 peptides that could be inhibited by NAM in vitro (Supplementary Fig. S2c). Consistently, SIRT2 catalytic mutant (SIRT2-H150Y) lost deacetylase activity on DVL2 (Fig. 1f). In addition, depletion of SIRT2 increased levels of DVL2-aK68 and DVL2 homooligomerization that can be further enhanced by Wnt3a stimulation to result in much bigger size of DVL2 aggregates in HCT116 cells (Fig. 1g, h). By contrast, both the size and number of Wnt3ainduced DVL2 macromolecular puncta were reduced upon coexpression of SIRT2 (Fig. 1i). These results indicate that SIRT2 is responsible for DVL2 deacetylation and may attenuate DVL2 homo-polymerization via deacetylation on K68. Further, we analyzed the effect of DVL2-aK68 on puncta formation. Expression of DVL2-K68R mutant resulted in fewer puncta-containing cells that showed puncta with reduced size and number compared to cells expressing wild-type DVL2 (DVL2-WT) (Fig. 1j; Supplementary Fig. S3a). In contrast, other KR mutation did not significantly affect DVL2-puncta formation (Fig. 1j; Supplementary Fig. S3a). As K68 is right located on DIX domain of DVL2 and the DIX–DIX interaction is critical for DVL2 homo-polymerization (Supplementary Fig. S3b), we analyzed the impact of aK68 on DVL2–DIX homomeric interactions. Interaction between truncated Myc-tagged DIX domain and full-length Flag-DVL2 can be either interrupted by DVL2-K68R mutation (Fig. 1k), or enhanced by overexpression of CBP (Fig. 1l; Supplementary Fig. S3c). Crystal structural model of DIX–DIX interaction also showed that acetyl group added to K68 tightened their molecular bonding and stabilized the overall DIX–DIX conformation (Fig. 1m). Thus, aK68 in DIX domain may promote DVL2-puncta formation through directly strengthening DVL2 homomeric interactions conferred by DIX. Upon Wnt activation, cytoplasmic β-catenin is stabilized and translocated to nucleoplasm to promote Wnt target genes expression. Compared to DVL2-WT, DVL2-K68R significantly decreased cytosolic accumulation and nuclear translocation of β-catenin (Fig. 1n; Supplementary Fig. S4a). Furthermore, TOP/ Flash reporter assays showed that β-catenin transcriptional activity was strongly inhibited by DVL2-K68R, regardless of Wnt3a stimulation (Fig. 1o). Similarly, TOP/Flash activity induced by DVL2-WT, but not DVL2-K68R, was markedly enhanced by CBP with or without Wnt3a treatment (Supplementary Fig. S4b). These findings indicate that CBP-induced acetylation at DVL2-K68 is required for full activation of Wnt signaling by DVL2. Ectopic activation of Wnt/β-catenin signaling in ventral vegetal blastomeres can induce formation of secondary dorsal body axes in early Xenopus embryos. As expected, injection of DVL2-WT mRNAs into ventral side successfully induced secondary axes