Fig 2: PTEN regulates cilia disassembly and phosphorylation of DVL2.(a) Top: time-course scheme for cilia disassembly in hTERT-RPE1 cells transfected (Trx.) with siControl (siCTL) or siPTEN. Bottom: representative images of cells starved and after serum addition, stained for acetylated tubulin (cilia axoneme, green), pericentrin (centrioles, red), 4′,6-diamidino-2-phenylindole (DAPI, cell nuclei, blue); scale bar, 20 μm. (b) PTEN loss promotes cilia disassembly. The percentage of ciliated cells from a was quantified by automated imaging (Supplementary Fig. 4 and Methods). Graph shows mean with error bars (s.e.m.) from n=7 (n=4 for a 7-h time point), with ∼1,000 cells for each condition (***P<0.001 by a t-test). (c) DVL2 peptide sequence containing the serine-143 phosphorylation site (in red), recognized by the antibody. (d) PTEN knockdown enhances DVL2 phosphorylation on serine 143 during cilia formation. hTERT-RPE1 cells were transfected with siCTL or siPTEN and lysates were analysed at 24, 48 and 96 h post transfection. Cells were starved for the last 48 h before lysis. (e) Phospho-S143-DVL2 levels during cilia disassembly. The experiment was performed as in a with siCTL- or siPTEN-transfected hTERT-RPE1 cells. Cell lysates were analysed by immunoblotting using the indicated antibodies. Samples in d and e are representative from at least four independent experiments. (f) siRNA-resistant PTEN rescues accelerated cilia disassembly caused by PTEN knockdown. Cilia disassembly was evaluated in hTERT-RPE1 lines expressing siRNA-resistant Flag-tagged PTEN (3F-PTENr) or empty vector (pCAGIP), upon transfection with siCTL or siPTEN. Cilia disassembly was quantified as in b. Results are plotted as mean with error bars showing s.e.m.; n=4, **P<0.01 by a t-test. (g) Expression of DVL2 S143A rescues the effect of PTEN knockdown on cilia disassembly rates. Cilia disassembly was performed in hTERT-RPE1 stable cell lines expressing either 3F-DVL2 or its S143A mutant in the presence of siDVL2 (targeting its endogenous 3′-untranslated region) upon siPTEN or siCTL transfection. Cilia disassembly was quantified as in b and is plotted as percentage of ciliated cells (mean with error bars showing s.e.m. of n=4; **P=0.01 by a t-test) with an average of 170 cells analysed per condition in each experiment.

Fig 3: SiO2NPs target Dvl to the lysosome. Confocal microscopy images display the nucleus (indicated by DAPI staining), the localization of Dvl (indicated by Dvl2-GFP) and the lysosome (indicated by Lamp1-RFP) in (A) control cells, (B) cells incubated with SiO2NPs, (C) cells incubated with SiO2NPs and 3-MA, and (D) cells incubated with SiO2NPs and LY294002.

Fig 4: DVL2 is required for ETV4-mediated HCC metastasis and invasion.A In MHCC97H and HepG2.2.15 cells, ETV4 was stably knocked down, and DVL2 expression levels were detected by WB and quantitative real-time PCR (qRT–PCR) analysis. B In MHCC97H and HepG2.2.15 cells, the DVL2 expression level was detected by WB after transfection with the pcDNA3.1-ETV4 plasmid at different doses. C After the DVL2 construct was transfected into MHCC97H-LV-shETV4 and HepG2.2.15-LV-shETV4 cells, WB was performed to detect the rescue of the indicated protein expression. D WB assay showed that ETV4 regulate the protein expression levels of c-Myc and MMP7 by active β-catenin. E qPCR assays showed that ETV4 regulated c-MYC and MMP7 in transcriptional level. F In vitro, the migration and invasion capacity of MHCC97H and HepG2.2.15 cells were observed by knocking down ETV4 and then rescuing DVL2 expression. G Pearson’s correlation coefficient was conducted to evaluate the correlation between ETV4 and DVL2 mRNA level. The scatter plots showed the correlation of R = 0.4097 and P < 0.0001. H The positive relationship between the ETV4 and DVL2 expression was exhibited in 140 HBV-positive HCC patients. I The patients were divided into three groups as follows: group 1: high ETV4 + high DVL2, group 2: low ETV4 + low DVL2, and group 3: others. The combined prognosis analysis of ETV4 and DVL2 showed that group 1 had worse overall survival than the other two groups of HCC patients (P < 0.001).

Fig 5: PTEN interacts with Dishevelled-2.(a) A LUMIER screen (schematic) was employed to identify novel PTEN-interacting proteins using a collection of eighty 3Flag-tagged preys as indicated. Results of an average of two independent LUMIER screens are plotted as a heatmap of the Luminesence Intensity Ratio (LIR, color scale). Both DVL2 and DVL3 (red stars) were identified as PTEN interactors. DVL1v and TCF7v denote splice variants. (b) Interaction between endogenous PTEN and DVL2. Cell lysates were immunoprecipitated with PTEN or control (IgG) antibodies and blotted for DVL2 (top panel) or PTEN (bottom panel), as indicated. An aliquot of total cell lysate and molecular weight markers (M) were loaded on the left side of the gel. The separate panels to the left show darker exposures of total cell lysate samples from the same blots. Image is representative of five independent experiments. (c) PTEN functions in convergent extension (CE) morphogenetic movements. Xenopus embryos injected with control or PTEN morpholinos either alone or together with low doses (100 pg) of dominant-negative Xdd1 Dishevelled, were then scored for CE defects (dorsal elongation and neural tube closure defects) as illustrated on the left. The graph displays the mean with error bars representing s.e.m. for Xdd1 (n=4) with 100 and more embryos counted for every condition (***P<0.001 by a t-test). Note that interference with PTEN strongly synergizes with Xdd1 to induce CE defects. (d) Scanning electron micrographs of multiciliated cells in the epidermis of Xenopus embryos treated as in c. Note the enhanced cilia defects in the embryos co-injected with PTEN MO and low doses of Xdd1 (scale bar, 10 μm). Images are representative from three independent experiments. (e) Basal body (BB) docking defects in embryos co-injected with PTEN MO and Xdd1. BBs in the ectoderm of Xenopus embryos treated as in c were stained for gamma-tubulin (green) and counterstained for cortical F-actin to visualize the apical surface of the cell (Phalloidin, red). Apical–basal reconstruction of confocal images reveals enhanced BB docking defects when Xdd1 was co-injected with PTEN MO (scale bar, 4 μm). Images are representative from at least three biological replicates.