Dishevelled phase separation promotes Wnt signalosome assembly and destruction complex disassembly

Abstract

The amplitude of Wnt/β-catenin signaling is precisely controlled by the assembly of the cell surface-localized Wnt receptor signalosome and the cytosolic β-catenin destruction complex. How these two distinct complexes are coordinately controlled remains largely unknown. Here, we demonstrated that the signalosome scaffold protein Dishevelled 2 (Dvl2) undergoes liquid-liquid phase separation (LLPS). Dvl2 LLPS is mediated by an intrinsically disordered region and facilitated by components of the signalosome, such as the receptor Fzd5. Assembly of the signalosome is initiated by rapid recruitment of Dvl2 to the membrane, followed by slow and dynamic recruitment of Axin1. Axin LLPS mediates assembly of the β-catenin destruction complex, and Dvl2 attenuates LLPS of Axin. Compared with the destruction complex, Axin partitions into the signalosome at a lower concentration and exhibits a higher mobility. Together, our results revealed that Dvl2 LLPS is crucial for controlling the assembly of the Wnt receptor signalosome and disruption of the phase-separated β-catenin destruction complex.

Graphical abstract

Figure S1.

Dvl2 undergoes LLPS. (A) Verification of Dvl2-mEGFP knock-in (KI) HEK293T cells by immunoblotting (left) and confocal imaging (middle). Two KI lines are shown, KI1 (KI in both alleles) and KI2 (KI in one allele). The TopFlash-luciferase reporter were transfected into WT and Dvl2-mEGFP KI cells (KI1 and KI2) and then treated with or without Wnt3a (right). (B) Verification of Dvl1/2/3 knock out (KO) HEK293T cells examined by immunoblotting. Two KO lines are shown as KO1 and KO2. (C) Levels of endogenous Dvl2 in WT HEK293T cells and Dvl2-EGFP expressing Dvl1/2/3 KO HEK293T cells. (D) Droplet formation by 2 µM (left) and 5 µM (right) Dvl2 without PEG8000. (E) FRAP analysis of Sulfo-Cyanine3 maleimide labeled Dvl2 droplets (red) induced without crowder. (F) The expression level of Dvl2 WT and mutant proteins detected by immunoblotting. (G) Statistical analysis of puncta number and puncta area of Dvl2 WT and mutant proteins in Dvl1/2/3 KO cells. (H) DIC images of 5 µM Dvl2 and its mutants at different NaCl concentrations after induction by 2.5% PEG8000. (I) Turbidity assay of Dvl2 mutants after induction of LLPS by 2.5% PEG8000 in different protein and salt concentrations. (J) LLPS of Dvl2 and its mutants at different NaCl concentrations after induction by 2.5% PEG8000 as assessed by DIC imaging. (K) Droplet formation by 0.5 µM purified Dvl2 or ΔDIX with 10% PEG8000. Statistical analyses were performed with the two-tailed unpaired t test. Quantitative data in G are shown as mean ± SEM (n = 3). ***, P < 0.001. Quantitative data in K are shown as mean ± SD (n = 3). ***, P < 0.001. Scale bars in A, 10 µm; in D, H and K, 2 µm; in E, 0.2 µm. Source data are available for this figure: SourceData FS1.

Figure 1.

Dvl2 undergoes LLPS. (A) GI-SIM images of Dvl2-mEGFP-KI cells with or without Wnt3a stimulation. The puncta number was counted from three independent cells. (B) FRAP showing the recovery of an area of Dvl2-mEGFP in Dvl2-mEGFP KI cells. (C) Confocal image showing Dvl2-EGFP puncta in Dvl1/2/3 KO HEK293T cells. The nuclei were counterstained by DAPI (blue). (D) FRAP showing the recovery of the fluorescent signal in Dvl2-EGFP puncta in Dvl1/2/3 KO HEK293T cells. (E) Formation of droplets by 0.2 µM (left) and 0.5 µM (right) Dvl2 protein with 10% PEG8000. (F) FRAP analysis of Sulfo-Cyanine3 maleimide labeled Dvl2 droplets (red) induced by 10% PEG8000. (G) Differential interference contrast (DIC) images showing the fusion of Dvl2 droplets. (H and I) Dvl2 LLPS induced by 2.5% PEG8000 in different protein and salt concentrations. DIC image analysis (H) and turbidity measurement (I) were shown. (J) Schematic diagram of Dvl2 and its mutants. (K) Confocal images of EGFP-tagged proteins in Dvl1/2/3 KO cells. Statistical analyses were performed with the two-tailed unpaired t test. Data in A and B are shown as mean ± SD (n = 3). ***, P < 0.001. Scale bars in A, 25 µm and insets in A, 5 µm; in C and K, 10 µm; in E, 2 µm; in F and G, 0.2 µm.

Figure 2.

Dvl2 LLPS is crucial for signalosome organization. (A) TIRF-SIM images of Dvl2-mEGFP-KI cells with or without Wnt3a stimulation. The puncta number was counted from three independent cells. The relative intensity of membrane-associated Dvl2-mEGFP puncta was normalized to the group without Wnt3a treatment. (B and C) The average intensity of membrane-associated Dvl2-mEGFP (B) and Axin1-tdTomato (C) puncta detected by TIRF-SIM. (D) The dissociation constant of RD (Wnt receptor/Dvl2 complex) and Dvl versus dissociation constant of RD and Axin. Scatter plot comparing dissociation constant of solutions that fit the experimental dynamic averaged Dvl2 intensity data. (E) Average intensity of Dvl puncta from the 106 fitted modeling solutions at different concentration of Axin. The ODE modeling was solved at the time interval of 20 min after Wnt3a treatment; red dash line marked the absolute total Axin level of 590 nM in the cell. (F) Formation of 0.5 µM Dvl2 droplets with or without 0.5 µM purified Fzd5 C-terminal fragment (522–585aa). Droplet formation was induced with 10% PEG8000. (G and H) Formation of droplets by mixtures containing 0.2 μM of the indicated components with 10% PEG8000. D: Dvl2; A: Axin1; F: Fzd5 C-terminal fragment; L: LRP6 C-terminal fragment; dRE: Dvl2(dRE). (I) Labeled Axin1, Dvl2, Fzd-C-terminal fragment, and unlabeled LRP6-C-terminal fragment were mixed (0.5 µM each) for LLPS assay (top row); labeled Axin1, Dvl2, LRP6-C-terminal fragment and unlabeled Fzd5-C-terminal fragment were mixed (0.5 µM each) for LLPS assay (bottom row). (J) Dvl1/2/3 KO HEK293T cells expressing Dvl2-EGFP or Dvl2(dRE)-EGFP were treated with Wnt3a stimulation. Quantification data showing the size and intensity of membrane-associated puncta detected by TIRF-SIM. The data were normalized to that in 0 min. (K) Axin1-mCherry was expressed alone or with untagged Dvl2-EGFP or Dvl2(dRE)-EGFP in Dvl1/2/3 KO HEK293T cells with Wnt3a stimulation. Quantification data showing the size of membrane-associated puncta detected by TIRF-SIM. The data corresponding to other time points were normalized to that in 0 min. Statistical analyses were performed with the two-tailed unpaired t test. Data in A, F, J and K are shown as mean ± SD (n = 3). ***, P < 0.001. Scale bars in A, 10 µm; in F, G, H and I, 2 µm; in J and K, 5 µm; and insets in F, G, and H, 0.5 µm.

Figure S2.

Visualizing the endogenous signalosome assembly process at the single-molecule level by TIRF. (A) Validation of Axin1-Flag-tdTomato KI HEK293T cells (Axin1-tdTomato KI cells) by immunoblotting. (B) Time-lapse TIRF-SIM images showing the membrane localization of Dvl2 puncta in Dvl2-mEGFP KI cells under Wnt3a stimulation. (C) Time-lapse TIRF-SIM images showing membrane localization of Axin1 puncta in Axin1-tdTomate KI cells under Wnt3a stimulation. (D) The binding interactions between the Wnt receptor (R), Dvl, and Axin for signalosome formation. RD, receptor-Dvl complex; ARD, Axin-receptor-Dvl complex. m represents the Dvl on the membrane, and n represents multiple Dvl in the system. (E) Average intensity of Dvl2-mEGFP puncta compared to the constrained and fitted well modeling solutions. (F) DIC images of 0.5 μM purified intracellular domain of Fzd5 (522–585aa) that were induced with 10% PEG8000. (G) Turbidity of purified Dvl2 (0.5 μM, 2 μM) with or without the same concentrations of intracellular domain of Fzd5 (0.5 μM, 2 μM) induced be 10% PEG8000. (H) Dvl2 WT and dRE have the similar expression in cells. The expression level of Dvl2 WT and dRE was detected with immunoblotting. Statistical analyses were performed with the two-tailed unpaired t test. Quantification data are shown as mean ± SD (n = 3). ***, P < 0.001. Scale bar in F, 2 µm. Source data are available for this figure: SourceData FS2.

Figure 3.

Dvl2 modulates the biophysical properties of Axin1 condensates. (A) Confocal images of Axin1-mCherry puncta when EGFP-tagged Dvl2 or Dvl2(dRE) were coexpressed in Dvl1/2/3 KO cells. The puncta size was normalized to that in the control group. (B) 0.5 µM Axin1 droplet formation induced by 10% PEG8000 after incubation with 0.5 µM dAPC2 and 0.5 µM Dvl2 or dRE. Quantification of droplet size and number are shown below. (C) TIRF images showing the membrane-associated Axin1-EGFP puncta in Axin1 KO cells with or without Wnt3a stimulation. (D) FRAP analysis of the membrane-associated Axin1-EGFP puncta detected by TIRF. (E) FRAP analysis of Alexa Fluor 488 NHS ester labeled Axin1 droplets incubated with or without the indicated components. The recovery of the fluorescent signal in Axin1 droplets was shown in the right. Statistical analyses were performed with the two-tailed unpaired t test. Data are shown as mean ± SD (n = 3). *, P < 0.05, **, P < 0.01, ***, P < 0.001. Scale bars in A and C, 10 µm; B, 2 µm; E, 0.2 µm.

Figure S3.

Dvl2 LLPS regulates the mobility of Axin1 and β-catenin. (A) Statistical analysis of the size and intensity of Axin1-mCherry puncta when Dvl2-EGFP or Dvl2(dRE)-EGFP was co-expressed in Dvl1/2/3 KO cells with Wnt stimulation. (B) FRAP analysis of Axin1-mCherry puncta in the cytoplasm when Dvl2-EGFP or Dvl2(dRE)-EGFP was co-expressed in Dvl1/2/3 KO cells. (C) Statistical analysis of overexpressed Axin1-mcherry puncta size in WT or Dvl1/2/3 KO HEK293T cells with or without Wnt3a stimulation for 30 min. Quantitative data are shown as mean ± SEM (n = 3). *, P < 0.05. (D) FRAP showing the recovery of the fluorescent signal in Axin1-mcherry puncta in WT or Dvl1/2/3 KO HEK293T cells with or without Wnt3a stimulation for 30 min. (E) FRAP analysis of Axin1 droplets labeled with Alexa Fluor 488 NHS ester with or without Dvl2 or the Dvl2(dRE) mutant. (F) Droplet formation of 0.5 μM Dvl2 and dRE induced by 10% PEG8000. (G) Turbidity assay of 2 μM Axin1 after incubation with indicated proteins. A: Axin1, P: dAPC2, D: Dvl2, dRE: Dvl2(dRE). (H) FRAP analysis of Axin1 puncta in Axin1-tdTomato KI cells localized in the cytoplasm or near the PM with or without Wnt3a stimulation. (I) FRAP analysis of Sulfo-Cyanine3 maleimide labeled Dvl droplets (red) with or without the indicated components. (J) FRAP analysis of β-cat-mCherry puncta when Axin1 and Dvl2-EGFP or Dvl2(dRE)-EGFP were co-expressed in Dvl1/2/3 KO cells. (K) FRAP analysis of Alexa Fluor 488 NHS ester labeled β-cat (green) recruited to destruction complex with or without Dvl2 or the Dvl2(dRE). Statistical analyses were performed with the two-tailed unpaired t test. Quantitative data are shown as mean ± SD (n = 3). *, P < 0.05, **, P < 0.01; ***, P < 0.001. Scale bars in B, F, and J, 2 µm; in E, I and K, 0.2 µm.

Figure 4.

Dvl2 LLPS disrupts the organization and function of the β-catenin destruction complex. (A) Alexa Fluor 488 NHS ester labeled 1 µM Axin1 protein (green) and other destruction complex components (1 µM) were mixed with or without Dvl2 or the dRE deletion and subjected to in vitro LLPS assay. The number and size of droplets were counted from three independent fields. (B) Confocal images of β-catenin-mCherry puncta when Axin1 was expressed with or without Dvl2-EGFP or Dvl2(dRE)-EGFP in Dvl1/2/3 KO cells. The number and size of puncta were counted from three independent cells. The puncta size was normalized to that in the control group. (C) Destruction complex components were mixed with or without Dvl2 and the Dvl2(dRE) for in vitro LLPS assay. In each assay, only one protein was labeled with Alexa Fluor 488 NHS ester. Fluorescence images (left) and the fluorescence intensity of the labeled protein recruited into the Axin1 droplets (right) are shown. 5C: Axin1, dAPC, β-catenin, CK1α, and GSK3β. (D) Dvl2 protein or the Dvl2(dRE) mutant (1 µM) were incubated with 1 µM Axin1, 50 nM GSK3β and 2.5 µM CK1α for in vitro phosphorylation. The band intensity of phosphorylated β-catenin was normalized to total β-catenin protein. 5C: Axin1, dAPC, β-catenin, CK1α and GSK3β. (E) Dvl1/2/3 KO cells were transfected with TopFlash-luciferase reporter and WT Dvl2-EGFP or its mutants, then treated with or without Wnt3a conditional medium for 12 h and harvested for determination of luciferase activity. The expression level of Dvl2 WT and mutant proteins in the reporter assay was detected with immunoblotting (bottom panel). (F) Working model of Dvl2 LLPS in assembly of the Wnt receptor signalosome and disruption of the β-catenin destruction complex. Statistical analyses were performed with the two-tailed unpaired t test. Data are shown in A, B, C, and E as mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Scale bars in B, 10 µm; A and C, 2 µm. Source data are available for this figure: SourceData F4.