Nucleation of the destruction complex on the centrosome accelerates degradation of β-catenin and regulates Wnt signal transmission

Significance Liquid–liquid phase separation (LLPS) governs a variety of mesoscale cellular processes. However, less is known about how cells utilize LLPS to drive cellular function. Here, we examined the destruction complex (DC), an organelle which controls Wnt signaling and whose components phase separate. Through a combination of advanced microscopy, CRISPR, computational modeling, and optogenetics, we find that the DC is nucleated by the centrosome and that this nucleation drives efficient signal transduction. Our work not only uncovers a biological function for LLPS but also highlights nucleation as a general method for controlling the function of intracellular condensates. Finally, our findings suggest a thermodynamic coupling between Wnt signal transduction and the cell cycle which could lead to insights into Wnt-driven cancers.

. Endogenously expressed β-catenin puncta are inversely correlated with CHIR-mediated Wnt pathway activation and β-catenin accumulation. A. Measurements of CRISPR cytoplasmic tdmRuby3-βcatenin in live 293Ts, data presented as mean fluorescent intensity fraction of t0 +/-s.e.m. (N = 30 cells per condition). B. Left: Representative images of tdmRuby3-β-catenin cells +CHIR for 24hrs. Arrows indicate puncta, asterisks indicate puncta absent. Right: Comparison of mean nuclear β-catenin fluorescence between +CHIR cells with and without visible β-catenin puncta. C. Sanger sequencing traces from genomic PCRs targeting 5' endogenous loci of CRISPR tdmRuby3 knock-ins. Red regions indicate tdmRuby3 insert. Representative images of indicated Destruction Complex (DC) components taken live, at various stages of the cell cycle. Montages follow the same cell through time. Scale = 10μm B. Representative fixed images of indicated DC components stained for endogenous GM130. Scale = 10μm. C. Representative images of fixed cells with varying Axin1 induction and co-stained for endogenous γ-tubulin. D. Representative images of fixed cells with varying APC induction and co-stained for endogenous γ-tubulin. E. Representative images of live cells bearing indicated proteins co-expressed with high Axin-1; Cumate-induced: APC, CRISPR-integrated: CK1α, GSK3β, Dox-induced: β-cat. Arrow indicates example of depleted centrosomal GSKβ observed under high Axin1 induction F. Upper: Representative images of live human induced-pluripotent stem cells (iPSCs) at various stages of the cell cycle. Lower: Images of fixed human iPSCs co-stained for endogenous γ-tubulin. Example lattice used to model the system with example nucleator region in black. After initial conditions are assigned, a model of diffusion operates on grid positions based on modified Cahn Hilliard equations. C. Quantification of the area of centrosomal droplets in comparison of total cell volume taken from CRISPR-tagged cells. Mean is represented by red line. D. Demonstration of the effects of nucleator size on system nucleation process. With a smaller centrosome, the droplet is more densely packed with enzymes whereas a larger centrosome results in droplet separation. E. Quantification of the effect of centrosome size on P4 β-catenin generation. F. Definition of nucleation efficiency as the ratio of the quotient of P4 β -catenin and β-catenin in a nucleated versus an unnucleated system. G. P4 b-catenin accumulation in log2 scan of kinase reaction rates. H. Nucleation efficiency of as a function of reaction rates and X (interaction parameter) of all clients and the cytoplasm. I. Quantification of in silico models of "opto"-β-catenin, "opto"-CK1, and "opto"-GSK. The graphs show increased gain from "opto"-GSK driven separation. Fig. S4. Opto-GSK3 suppresses β-catenin accumulation due to GSK3β inhibition or exogenous chemical induction. A. Representative images of live cells treated with CHIR or DMSO vehicle, with or without blue light stimulation. Scale = 10μm. B. Measurements of experiment shown in A. Data presented as mean +/-s.e.m. (N=30 cells per condition). C. Representative Western blots of lysates from 293Ts bearing Opto-GSK3 and treated with Wnt-3a, with or without blue light stimulation for the indicated time course. D. Representative images of cells bearing Opto-GSK3 fixed and stained for endogenous β-catenin after culture in the indicated conditions for 48 hrs. E. Violin plots of cells from D. F. Representative images of 293Ts bearing Opto-GSK3 and endogenously-expressed tdmRuby3-β-catenin treated with CHIR, with or without blue light stimulation for 24hrs. G. Measurements from experiment shown in F., lines represent fold-change from t0 means +/-s.e.m. for cells in each condition (Light ON N=67, Light OFF N=50 cells). H. Representative Western blots of lysates from 293Ts bearing Opto-GSK3 and treated with CHIR, with or without blue light stimulation for the indicated time course. I. Representative images of 293Ts bearing Opto-GSK3 and Dox-inducible -β-catenin-tdmRuby3 treated with Dox, with or without blue light stimulation. Scale = 10μm. J. Quantification of experiment in I: lines represent absolute means +/-s.e.m for cells in each condition (Light ON N=28, Light OFF N=27 cells).

Video S1. Cells with β-cat Puncta Resist β-cat Accumulation in response to CHIR. Left:
Imaging fields of live tdmRuby3-β-cat cells treated with DMSO control. Right: Imaging fields of tdmRuby3-β-cat cells treated with CHIR. Arrows indicate β-cat puncta.
Video S4. In-silico behavior of destruction components with a centrosomal region. In-silico model of phase separation behavior for every component involved in a hypothetical WNT pathway over 100 simulation time steps in the presence of a centrosome.
Video S5. In silico behavior of destruction components without a centrosomal region. Insilico model of phase separation behavior for every component involved in a hypothetical WNT pathway over 100 simulation time steps without the presence of a centrosome.

Video S6. Impact of interaction parameter χ on destruction complex component behavior.
In-silico model of the destruction components (CK1α, GSK3β, and β-catenin) at various interaction parameter values (χ) over 100 simulation time steps showing that increasing χ increases separation propensity.

Video S7. Activation of Opto-GSK3 Increases Centrosomal Condensate Partitioning.
Zoomed video of Opto-GSK3 cells stimulated with blue light throughout indicated timecourse.